Handbook of Residual Stress and Steel Deformation

2002 ASM International. All Rights Reserved.
Handbook of Residual Stress and Deformation of Steel (#06700G)
Handbook of Residual Stress

and
Deformation of Steel
Edited by
G. Totten
M. Howes
T. Inoue
Materials Park, Ohio 44073-0002

www.asminternational.org

Copyright 2002
by
ASM International
All rights reserved
No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any
means, electronic, mechanical, photocopying, recording, or otherwise, without the written permission of the
copyright owner.
First printing, March 2002
Great care is taken in the compilation and production of this book, but it should be made clear that NO
WARRANTIES, EXPRESS OR IMPLIED, INCLUDING, WITHOUT LIMITATION, WARRANTIES OF
MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, ARE GIVEN IN CONNECTION
WITH THIS PUBLICATION. Although this information is believed to be accurate by ASM, ASM cannot
guarantee that favorable results will be obtained from the use of this publication alone. This publication is
intended for use by persons having technical skill, at their sole discretion and risk. Since the conditions of
product or material use are outside of ASMs control, ASM assumes no liability or obligation in connection
with any use of this information. No claim of any kind, whether as to products or information in this
publication, and whether or not based on negligence, shall be greater in amount than the purchase price of this
product or publication in respect of which damages are claimed. THE REMEDY HEREBY PROVIDED
SHALL BE THE EXCLUSIVE AND SOLE REMEDY OF BUYER, AND IN NO EVENT SHALL EITHER
PARTY BE LIABLE FOR SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES WHETHER OR
NOT CAUSED BY OR RESULTING FROM THE NEGLIGENCE OF SUCH PARTY. As with any material,
evaluation of the material under end-use conditions prior to specification is essential. Therefore, specific
testing under actual conditions is recommended.
Nothing contained in this book shall be construed as a grant of any right of manufacture, sale, use, or
reproduction, in connection with any method, process, apparatus, product, composition, or system, whether or
not covered by letters patent, copyright, or trademark, and nothing contained in this book shall be construed as
a defense against any alleged infringement of letters patent, copyright, or trademark, or as a defense against
liability for such infringement.
Comments, criticisms, and suggestions are invited, and should be forwarded to ASM International.
Prepared under the direction of the ASM International Technical Book Committee (20002001), Sunniva R.
Collins, Chair
ASM International staff who worked on this project included Veronica Flint, Manager of Book Acquisitions;
Bonnie Sanders, Manager of Production; Carol Terman, Production Project Manager; and Scott Henry,
Assistant Director of Reference Publications.
Library of Congress Cataloging-in-Publication Data
Handbook of residual stress and deformation of steel/[edited by] G. Totten, M. Howes, T. Inoue.
p. cm.
Includes bibliographical references and index.
1. SteelFatigueHandbooks, manuals, etc. 2. Residual stressesHandbooks, manuals, etc. I. Totten,
George E. II. Howes, Maurice A.H. III. Inoue, Tatsuo, 1939TA473 .H3215 2001
620.176dc21
2001040042
ISBN: 0-87170-729-2
SAN: 204-7586
ASM International
Materials Park, OH 44073-0002
Printed in the United States of America
Multiple copy reprints of individual articles are available from Technical Department, ASM International.

Contributors
I. Alexandru
Faculty of Materials
Technical University of Iasi
Romania
T. Inoue
Department of Energy Conversion Science
Graduate School of Energy Science
Kyoto University, Japan
H. Bhadeshia
Department of Materials Science and
Metallurgy
University of Cambridge, UK
D.-Y. Ju
Saitama Institute of Technology
Saitama, Japan
V. Bulancea
Faculty of Materials
Technical University of Iasi
Romania
V.V. Dobrivecher
Ukraine National Academy of Science
Institute of Engineering Thermophysics
Kiev, Ukraine
T. Ericsson
Linkopings Tekniska Hogskola IKP
Linkoping, Sweden
F.D. Fischer
Vorstand des Institutes fur Mechanik
Montanuniversitat Loeben
Loeben, Germany
A.V. Fomin
Leading Research Scientist
Institute Russian Academy of Sciences
Moscow, Russia
K. Funatani
Nihon Parkerizing Co. Ltd.
Nagoya, Japan
Bo Gong
Department of Metallurgy & Materials Science
University of Toronto, Canada
J. Grum
Faculty of Mechanical Engineering
University of Ljubljana, Slovenia
A.Y. Hassan
Director/Dean, School of Mechanical
Engineering
Universiti Sains Malaysia, Malaysia
K. Heess
Karl Heess GmbH
Lampertheim, Germany
F.T. Hoffmann
ITW Bremen
Bremen, Germany
R. Hoffmann
ITW Bremen
Bremen, Germany
N.I. Kobasko
Kiev, Ukraine
Z. Kolozsvary
S.C. Plasmaterm S.A.
Tg-Mures, Romania
A.I. Kovalev
Surface Phenomena Research Group
CNIICHERMET
Moscow, Russia
J. Kritzler
Metal Improvement Company, Inc.
Unna, Germany
K.-H. Lang
Institut fuf Werkstoffkunde 1
Universitat Karlsruhe (TH), Germany
R.W. Lewis
Department of Mechanical Engineering
University College of Swansea, UK
D. Lohe
J. Lu
Universite de Technologie de Troyes
Trayes Cedex, France
T. Lubben
ITW Bremen
Bremen, Germany
M.V. Medvedev
Research Scientist
Moscow, Russia
V.P. Mishina
CNIICHERMET
Moscow, Russia
V.S. Morganyuk
Kiev, Ukraine
iii
M. Narazaki
Utsunomiya University
Tochigi, Japan
J. Pan
School of Materials Science and Engineering
Shanghai Jiao Tong University
Shanghai, P.R. China
P. Ramakrishnan
Department of Metallurgical Engineering and
Materials Science
Indian Institute of Technology
Bombay, India
I.A. Razumovsky
Leading Research Scientist
Moscow, Russia
T. Reti
Banki Donat Polytechnic
Budapest, Hungary
C. Ruud
Pennsylvania State University
University Park, PA USA
G. Schleinzer
Vorstand des Institutes fur Mechanik
Montanuniversitat Leoben
Leoben, Germany
B. Scholtes
Institut fur Werkstofftechnik
Universitat Kassel, Germany
K.N. Seetharamu
School of Mechanical Engineering
Universiti Sains Malaysia, Malayasia
G.E. Totten
G.E. Totten & Associates Inc.
Stony Point, NY USA
O.Vohringer
D.L. Wainstein
CNIICHERMET
Moscow, Russia
H.W. Walton
Consultant
Forest City, NC USA
Z. Wang
Department of Metallurgy & Materials Science
University of Toronto, Canada

G.M. Webster
Union Carbide Corporation
Tarrytown, NY USA
W. Wubbenhorst
Unna, Germany
V.V. Zabilsky
Physical Technical Institute, Ural Branch of
RAS
Ijevsk, Russia
iv
W. Zinn
Institut fur Werkstofftechnik
Universitat Kassel, Germany

Contents
Preface ........................................................................... vi
Residual Stress During Hardening Processes
Effect of Materials and Processing
Residual Stresses in Carburized, Carbonitrided,

and Case-Hardened Components ........................................ 189
T. Reti
Material Factors .................................................................. 3

H.K.D.H. Bhadeshia
Residual Stresses in Nitriding............................................... 209

Z. Kolozsvary
Prestress Engineering of Structural Material:

A Global Design Approach to the Residual Stress Problem ..........11
J. Lu
Induction Hardening.......................................................... 220

J. Grum
Residual Stresses and Fatigue Behavior.....................................27

D. Lohe, K.-H. Lang, and O. Vohringer
Hardening by Reheating and Quenching .................................. 248

M. Narazaki, G.E. Totten, and G.M. Webster
Stability of Residual Stresses .................................................54

D. Lohe and O. Vohringer
Metallo-Thermo-MechanicsApplication to Quenching ................ 296

T. Inoue
Control of Residual Stress Formation and Steel Deformation
during Rapid Heating and Cooling...................................... 312
N.I. Kobasko, V.S. Morganyuk and V.V. Dobrivecher
Effect of Residual Stress on Hydrogen Embrittlement

and Stress Corrosion Cracking.............................................70
A.I. Kovalev, V.P. Mishina, D.L. Wainstein, and V.V. Zabilsky
Effect of Cryogenic Cooling on Residual Stresses, Structure,

and Substructure ........................................................... 331
Ioan Alexandru and Vasile Bulancea
Measurement and Prediction of Residual Stress and Distortion

Deflection Methods to Estimate Residual Stress ...........................89
H. Walton
Inducing Compressive Stresses through Controlled Shot Peening .... 345

J. Kritzler and W. Wubbenhorst
Measurement of Residual Stresses ...........................................99

C. Ruud
Residual Stress Formation During Manufacturing Processes

Residual Stress Formation during Casting ................................ 361
R.W. Lewis, K.N. Seetharamu and A.Y. Hassan
Stress Determination in Coatings........................................... 118

J.Albert Sue and Gary S. Schajer
Residual Stress Formation during Casting:

Continuous and Centrifugal Casting Processes........................ 372
D.-Y. Ju
Methods for Determination of Inhomogeneous Residual

Stress Fields ................................................................ 125
I.A. Razumovsky, M.V. Medvedev, and A.V. Fomin
Residual Stress Formation in the Shaping of Materials
Residual Stress Formation Processes during Welding and Joining.... 391

W. Zinn and B. Scholtes
Residual Stress in the Forming of Materials.............................. 141

Z. Wang and B. Gong
Residual Stresses in Powder-Metal Processing........................... 397

P. Ramakrishnan
Residual Stress Formation and Distortion of Rail Steel................. 424
F.D. Fischer and G. Schleinzer
The Effect of Final Shaping Prior to Heat Treatment ................... 150

T. Ericsson
Residual Stresses during Gear Manufacture .............................. 437

K. Funatani
Factors Affecting Final Part Shaping ...................................... 159

P. Jiansheng
Metric Conversion Guide .................................................. 459

Effects of Process Equipment Design...................................... 183
F.T. Hoffmann, T. Lubben, R. Hoffmann, and K. Hee
Index ........................................................................... 465

v

Preface
Control of steel deformation is one of the most
common concerns within the metals processing
industry. Numerous surveys have been conducted by various organizations in recent years
to assess the critical needs of the industry. In
nearly every survey that has been conducted, distortion is either the greatest or second greatest
concern among the steel heat treating community. Steel distortion control will exhibit tremendous effects on the profitability of the commercial enterprise. Therefore, it is not surprising that
the ability to understand the overall distortion
process and to be able to design solutions to this
problem typically rank very high on these same
surveys.
In view of the enormous visibility and importance of steel deformation problems, the editors
decided to put together an engineering handbook
on steel deformation. To address this subject
properly, contributing factors to overall steel deformation problems, including material effects,
machining, heating and cooling, must be examined.
This handbook contains 27 articles, divided
into five sections: Effect of Materials and Processing, Measurement and Prediction of Residual Stress and Distortion, Residual Stress Formation in the Shaping of Materials, Residual
Stress During Hardening Processes, and Residual Stress Formation During Manufacturing Processes.
There are five articles in the section Effect of
Materials and Processing. Material Factors
discusses the effects of various material properties such as thermal properties and the interactions of residual stresses on the transformation
products formed and steel deformation during
fabrication. Transformation plasticity is discussed in some detail along with the use of modeling to better understand these processes.
Prestress Engineering of Structural Material
provides a global design approach to understanding the effects of residual stress generated during
surface engineering manufacturing processes
such as PVD and CVD on the material properties
obtained. Some of the topics discussed in this
chapter include developments in the measurements of residual stresses, advanced mechanical
surface treatments, and modeling of fatigue behavior taking residual stresses into consideration.
The effect of residual stresses on fatigue behavior is discussed in detail in the next article.
Examples of topics discussed include stability of
residual stresses, some aspects of fatigue in
steels, influence of residual stresses on cyclic deformation behavior, influence of residual stresses
on crack initiation and propagation, and effect of
residual stresses on S-N curves; an overview of
modeling of the effect of residual stresses on fatigue behavior is provided.
The next article provides an overview of the
stability and relaxation behavior of macro and
micro residual stresses in steel due to thermal
and mechanical treatments. This discussion includes relaxation of residual stresses by annealing, residual stress relaxation by uniaxial deformation, and relaxation by cyclic deformation.
Hydrogen embrittlement of metals, as well as
other types of brittle fracture, result from nucleation and development of micro-cracks caused
by internal stresses. The last article in this section provides an overview of the effect of residual stress on hydrogen embrittlement and stress
corrosion cracking (SCC) of steel. This discussion includes the effect of hydrogen on structure
and transformation of steel, types of hydrogen
embrittlement, delayed fracture in steel, crack
initiation and growth, SCC of low alloy steels,
crack initiation and growth mechanism of SCC
processes, methods of estimating sensitivity to
SCC, effect of alloying elements on resistance
to SCC, and the role of structure and thermal
processing in SCC.
In the section Measurement and Prediction of
Residual Stress and Distortion, the first article
describes a number of simple, inexpensive deflection (dissection) methods used to estimate residual stress of various types of components. The
methods include Almen strip; Navy C-Ring;
plate or bar slitting and deflection; tube slitting
and opening; and bending of bars, H-beams, and
channels.
The next article provides an overview of residual stress measurement methods. Topics include residual stresses arising from various
manufacturing processes, measurement methods
including strain measuring technique, post-stress
relaxation measurement, sectioning and material
removal methods. In addition, strain measurement methods such as x-ray and neutron diffraction, ultrasonic, birefringent and laser, optical
gages, brittle coatings, Barkhausen noise, and
chemical coatings are discussed. Semidestructive methods such as blind hole drilling and ring
coring are discussed.
Measurement of residual stresses in coatings
and thin films is important because their influence on mechanical and physical properties affect component service performance. Stress Devi
termination in Coatings provides a guide for

measuring residual macrostress in coatings, Specific topics include origin of residual stresses in
coatings and residual stress measurement methods including the deflection method, x-ray diffraction, and hole-drilling. A comparison of
these methods is provided.
The last article in this section provides a detailed review of methods used to measure and
subsequent data analysis of inhomogeneous residual stress fields. This discussion includes residual stress as an inverse problem of experimental mechanics, indicator crack method of
measuring residual stress, arbitrary cut-out indicator method, and experimental methods and
equipment including photoelastic coating
method, and optical interferometry. Although
this is a relatively rigorous numerical discussion,
practical examples also are provided.
Residual Stress Formation in the Shaping of
Materials contains four articles. The first article
covers residual stress in the steel forming processes. The steel forming processes included are
cold forming such as wire drawing, and hot
forming such as extrusion, rolling, and forging.
The effects of residual stresses involved in these
processes are reviewed, and specific topics include residual stress in cold metal forming such
as bending of sheet, drawing of wire, rod, and
tube, and residual stresses in deep drawn cup,
sunk tubes, and radial forging products.
The effect of final shaping prior to heat treatment on residual stress formation is discussed in
the next article. The effects of shaping processes
including grinding, milling, turning, shot peening, and straightening on residual stress are discussed. Also discussed is distortion after final
part shaping and experimental and computational studies of these processes.
The next article provides a practical overview
of the factors affecting residual stress and distortion during final part shaping. Included are
discussions of influence of component shape on
heat treatment distortion, the effect of crosssection size and asymmetry, effect of heat treating procedure and machining process on final
component shape, effect of sequence of heat
treating and machining, influence of machining
allowance and stress relieving procedure, influence of residual stresses caused by cutting, methods of manufacturing blanks and effect of original structure, hot-rolled steels or forgings and
effect of banded segregation and carbide segregation, influence of heat treating methods, the
effect of heating including the rules of heating,

quenching and system design, tempering, and

equipment and racking.
A more focused, but practical, discussion on
the effect of process equipment design on distortion follows. Subjects that are covered include
distortion generating process equipment, methods that may be used to minimize equipmentrelated distortion, quench system design, and
press quenching.
Residual Stress During Hardening Processes
contains eight articles. The first article provides
a detailed discussion on the residual stresses in
carburized, carbonitrided, and case-hardened
components. Topics include process considerations for carburized and carbonitrided components, transformations and stress evolution in
carburized and case-hardened components, effect of heat treating operations on residual stress
distribution, relationship between residual
stresses and properties of carburized parts and
modeling and prediction of residual stress field.
The article on residual stresses in steel nitriding includes a discussion of nitrided layer structure as a function of nitriding process, residual
stresses in nitrided layers, influence of residual
stresses on fatigue behavior of nitrided steel
components, and modeling and prediction of residual stresses in nitrided steel components.
The article on residual stress formation in induction hardening processes include an overview of the induction hardening process and
steels used for this process, magnetic flux concentrators, conditions in induction heating and
quenching of machine parts, residual stress surface profiles after induction surface hardening,
stress profiles in the machine part in the loaded
state, workpiece distortion in induction surface
hardening, induction surface hardening of gear
wheels, fatigue strength of materials, and residual stresses after induction surface hardening and
finish grinding.
The next article provides an overview of residual stresses and distortion resulting from reheating and quenching. Topics include phase
transformation during heat treating including
steel transformations, TTT and CCT diagrams,
metallurgical crystal structure, estimation of volumetric change due to steel transformation upon
quenching, cooling of steel with and without
metallurgical transformation, tempering, basic
distortion mechanism, relief of residual stresses,
material movement due to thermal gradients during heating and cooling, material, component
and process effects, retained austenite, quench
severity and uniformity and process design effects on distortion, quench distortion and cracking, quenchant selection, measurement and evaluation of quenching power, estimation of heat
transfer coefficient, wetting behavior and nonuniform quenching, surface conditions, and
quench process modeling and simulation of residual stress and distortion after quenching.
A detailed approach to modeling and simulation of residual stress and distortion applied to
quench processing follows. This discussion is
based on a metallo-thermo-mechanics approach,
and topics discussed include an overview of
metallo-thermo-mechanics and numerical simulation methodology with practical examples.

In the article on the control of residual stress
formation and steel deformation during rapid
heating and cooling, a particular emphasis is on
intensive quenching. This is the first detailed,
article-length discussion of this old, but littleknown technology in the western world. Topics
include mathematical model for calculation of
thermal and stress-strain state, computation of
stress-strain state, possibility of predicting hardening cracks, predicting the deformation of bearing rings during hardening, thermal stresses
formed in carburized steel products due to excessive cooling rates, generalization of computational and experimental results for heating and
cooling of parts with different geometries and
thermal and physical fundamentals of processing
of high-strength materials.
An often contradictory subject is the cryogenic processing of steels, and the detailed overview of the effect of cryogenic cooling on residual stress is presented here. Specific topics
include role of residual stresses within martensitic transformation at cryogenic temperatures,
evaluation of residual stresses after cryogenic
cooling, influence of cryogenic cooling on residual stresses and dimensional stability of steels,
and influence of cryogenic cooling on the structure and substructure of steels.
The practical use of controlled shot peening
to induce compressive residual stresses is described in detail next. This discussion includes a
historic overview of shot peening, elementary
processes of shot peening, workpiece and material process parameters, process monitoring,
process optimization, x-ray diffraction, and industrial examples.
In Residual Stress Formation During Manufacturing Processes, the first article includes an
extensive discussion of residual stress and deformation problems arising from the casting process, and modeling of residual stress formation
during casting. Discussion includes finite element analysis of heat flow during casting, formulation of the elasto-viscoplastic stress model,
and deformation of a solidifying material.
The next article describes residual stress formation during the casting process, and it includes continuous and centrifugal casting. Topics discussed include inelastic behavior and
unified constitutive theory of metallic material
in solidification, analytical method of the
thermal-mechanical problem for the casting process, residual stress formation during semicontinuous casting, residual stress formation during
centrifugal casting, and residual stress formation
during strip casting by the twin-roll method.
The origin and assessment of residual stresses
during welding or brazing is discussed next.
Welding residual stresses are discussed including residual stresses due to shrinkage, quenching, and phase transformations. Characteristic
residual stress distributions in brazed components is also discussed.
The article Residual Stresses in Powder
Metal Processing is divided into two parts. The
vii
first part describes manufacturing of ferrous P/

M parts including powder characteristics, compaction in rigid dies, isostatic compaction, sintering, heat treatment of P/M parts, hot pressing,
roll compaction, powder forging, metal injection
molding, spray forming, warm compaction, and
rapid prototyping. The second part discusses residual stresses in P/M processing including
powder production, compaction of metal powders, sintering of metal powders, pressure sintering and hot isostatic pressing, heat treatment
of P/M parts, and microstructural development
and properties.
Residual Stress Formation and Distortion of
Rail Steel covers the cooling process including
the cooling boundary conditions and heat transfer, residual stress state analysis, weight and frictionthe rail end problem, experimental results;
roller straightening including residual stresses in
unused roller-straightened rails, behavior of rail
steel under plastic deformation, simulation of
roller straightening; and rails in service including residual stresses due to welding and residual
stress formation in rolling contact.
The last article provides a detailed description
of residual stress formation during hypoid gear
manufacture. It includes an overview of residual
stress formation in carburized and hardened
work, profiles and peak magnitudes of residual
stresses, measurement methods including the
Sachs hole-drilling method, x-ray and neutron
diffraction, influence of steel properties on residual stresses, influence of carburizing process parameters on residual stress formation, benefits of
residual stresses on fatigue strength, and the effects of hardness, case depth, intergranular oxidation, influence of shot peening, change of residual stresses during fatigue, and distortion of
carburized and hardened steels.
The preparation of a text of this scope was a
tremendous task. The editors are deeply indebted
to many colleagues for their patience, support,
and assistance; without them this text would not
have been possible. Special thanks go to the
ASM staff who often labor in the background
but who are vital members of the team. Particularly, thanks go to Veronica Flint and Carol Terman of ASM International for their help and encouragement.
Very special thanks go to our families for their
seemingly unending support. Without their understanding and encouragement, this project
would never have been completed.
George E. Totten, Ph.D., FASM
Editor
G.E. Totten & Associates Inc.
Stony Point, NY USA
Prof. Maurice A.H. Howes, Ph.D. (Retired)
Editor
Worcestershire, England
Prof. Tatsuo Inoue, Ph.D., FASM
Editor
Department of Energy Conversion Science
Faculty of Energy Science
Kyoto University
Kyoto, Japan
Handbook of Residual Stress and Deformation of Steel

G. Totten, M. Howes, T. Inoue, editors, p3-10
DOI: 10.1361/hrsd2002p003
Copyright 2002 ASM International

All rights reserved.
Material Factors
H.K.D.H. Bhadeshia, University of Cambridge
RESIDUAL STRESSES are a consequence of

interactions among time, temperature, deformation, and microstructure (Fig. 1). Material or
material-related characteristics that influence the
development of residual stress include thermal
conductivity, heat capacity, thermal expansivity,
elastic modulus and Poissons ratio, plasticity,
thermodynamics and kinetics of transformations,
mechanisms of transformations, and transformation plasticity.
Many general statements can be made about
the role of material factors in the evolution of
residual stress. Spatial variations in temperature
give rise to nonuniform thermal strains, the effect of which becomes exaggerated when the
material is elastically stiff and has a high yield
strength. A large thermal conductivity helps reduce residual stress by reducing temperature gradients (Ref 2). The dissipation or absorption of
heat depends not only on the external environment of the component but also on internally
generated heatfor example, during adiabatic
deformation or due to the latent heat of transformation. Similarly, the plastic strain distribution
in the component depends both on the constitutive properties and on how the shape deformations due to phase transformations compensate
for the development of stress.
The fundamental material properties are, of
course, temperature dependent. Table 1 illustrates how several key properties might vary
with temperature (Ref 3). Some of these properties, which can to some extent be estimated
quantitatively, are discussed in detail in the sec-
tions that follow; others such as elastic modulus

and thermal conductivity still have to be measured for individual alloys.
Heat Capacity
The dominant contribution to specific heat capacity comes from lattice vibrations (phonons),
since the majority of free electrons are prevented
from participation in heat absorption by the Pauli
exclusion principle. However, for iron and its
alloys, a further important contribution comes
from magnetic changes. The net specific heat capacity can therefore be factorized into three components:
T C
CP{T } C LV
TD
CeT C Pl{T }
where C LV{TD /T} is the Debye specific heat function and TD is the Debye temperature. The funcTable 1
Physical properties that affect the development of residual stress in steels

Temperature, C (F)
Property
Phase(a)
Elastic modulus, GPa
c
P
b
Poisson ratio
c
P
b
c
P
b
Thermal expansivity, K1
Time,
temperature
1. Thermal stress
6. Latent
heat
Stress,
strain
4. Heat of
deformation
5. Stressinduced
transformation
2. Temperaturedependent phase
transformations
3. Transformation
strain
Microstructure
Fig. 1
The coupling of temperature, stress, and microstructure. Source: Ref 1
(Eq 1)
tion C1 corrects C LV{TD /T} to a specific heat at

constant pressure. Ce is the electronic specific
heat coefficient, and C lP is the component of the
specific heat capacity due to magnetism. Figure
2 illustrates the data for ferrite and austenite in
pure iron. Whereas it is well known that ferrite
undergoes a paramagnetic to ferromagnetic transition on cooling below 1042.15 K, the magnetic
properties of austenite are seen from Fig. 2 to be
of some consequence in determining the heat capacity. There are two coexisting electron states
of austenite, one of which is ferromagnetic with
a Curie temperature of 1800 K and the other of
which is antiferromagnetic with a Neel temperature of 55 to 80 K (Ref 4). The balance between
these states changes with temperature, giving
rise to corresponding changes in heat capacity.
The data in Fig. 2 are for pure iron, but there
is now sufficient understanding of the components of heat capacity to enable similar estimates
for iron alloys, using internationally available
computer programs and thermodynamic data-
Thermal conductivity, W/m K
c
P
b
Specific heat capacity, 106 J/m3 K
c
P
b
Yield strength, MPa
c
P
b
0 (32)
200
210
210
200
0.291
0.280
0.280
0.280
15.0
49.0
49.0
43.1
4.15
3.78
3.78
3.76
190
360
440
1600
300 (570)
175
193
193
185
600 (1110)
150
165
165
168
0.309
0.296
0.296
0.296
2.1 105
1.4 105
1.4 105
1.3 105
18.0
41.7
41.7
36.7
4.40
4.46
4.46
4.45
110
230
330
1480
(a) , P, b, and represent allotriomorphic ferrite, pearlite, bainite, and martensite, respectively. Source: Ref 3
0.327
0.310
0.310
0.310
21.7
34.3
34.3
30.1
4.67
5.09
5.09
5.07
30
140
140
1260
800 (1470)
124
120
120
...
0.345
0.325
0.325
...
25.1
27.0
27.0
...
4.90
5.74
5.74
...
20
30
30
...
4 / Effect of Materials and Processing

bases (Ref 6). After all, changes in fundamental
thermodynamic quantities such as enthalpy and
entropy are derived from heat capacity data. It is
surprising that this capability has not yet been
exploited in any calculation of residual stress,
even though the methodology is widely available.
Expansion Coefficient and Density

Table 1 shows that the expansion coefficient
of austenite is larger than that of ferrite; this
might be considered surprising given the lower
density of ferrite. However, the behavior is again
c0
c1
V cm{T } (1 y)V m
{T } y V m
{T }
30
Heat capacity, CP, J mol 1K1
V cm0 6.695(1 2.043 105T 1.52 108T 2)

V cm1 7.216(1 2.043 105T 1.52 108T 2)
40
20
V am{T }
7.061(1 2.043 105T 1.52 108T 2)
Austenite
10
0
75
Ferrite
50
25
0
a reflection of the two coexisting electronic

states of austenite (c0 and c1), each with a thermal expansion coefficient that is identical to that
of ferrite. The c0 component has the lower molar
volume and is the antiferromagnetic form,
whereas the denser c1 form is ferromagnetic. The
relative proportion of atoms in the c0 and c1
states changes with temperature, so that the apparent expansion coefficient of austenite as a
whole, as detected experimentally, is much
larger than that of ferrite (Fig. 3).
The molar volumes (in cm3 /mol) of c0, c1, c,
and over the temperature range of 300 to 1775
K are:
300
600
900
1200
1500 1800
Temperature, K
where y is the fraction of atoms of austenite in

the c1 state, the details of which can be found
elsewhere (Ref 4, 5).
These data are for pure iron, but thermodynamic data can be used to assess how the expansion coefficients would change with alloying,
since there are quite sophisticated treatments of
the effect of solute elements on the magnetic and
other components of the free energies of iron.
Note that the two electronic states picture of
austenite is a simplification of the real scenario,
but first-principles calculations (Ref 7), which
Specific heat capacities of ferrite and austenite in

pure iron, as a function of temperature. The thin
lines represent the combined contributions of the phonons
and electrons, whereas the thicker lines also include the
magnetic terms. The dashed vertical lines represent the Curie, c, and c d transitions. d-ferrite is simply an
alternative historical name for high temperature . Source:
Ref 5
deal with higher levels of complexity, are not yet

applicable to practical alloys.
Plastic Deformation
The familiar mechanisms of plastic deformation are slip, mechanical twinning, and creep.
Phase transformations also cause permanent deformation (Ref 811). In steels, austenite can decompose into a large variety of microstructures
that are distinguished by the atomic mechanism
of transformation (Fig. 4). In a displacive transformation, the change in crystal structure is
achieved by a deformation of the parent structure. A reconstructive transformation is one in
which the change in structure is achieved by a
flow of matter, which occurs in such a way that
strains are minimized.
All the transformations cause changes in
shape (Fig. 5a), which for reconstructive transformations simply reflects the change in density.
For displacive transformations, the shape change
is an invariant-plane strain (IPS), that is, a combination of a shear on the invariant plane and a
dilatation normal to that plane. The strain energy
associated with a constrained IPS is minimized
when the product phase has a thin-plate shape.
This is why Widmanstatten ferrite, bainite, acicular ferrite, and martensite in steels grow in the
form of plates. The distinguishing features of a
variety of deformation modes are compared in
Table 2, and Table 3 describes the shape deformations.
The permanent strain caused by any transformation is called transformation plasticity. A
Fig. 2
Displacive
Reconstructive
Invariant-plane strain shape

deformation with large shear
component. No iron or
substitutional solute diffusion.
Thin plate shape
Diffusion of all atoms during

nucleation and growth.
Sluggish below about 850 K
8.0
7.8
Allotriomorphic
ferrite
Volume, cm3 mol1
7.6
V1
7.4
Widmansttten ferrite
Carbon diffusion during
paraequilibrium nucleation
and growth
Idiomorphic
ferrite
Bainite and acicular

ferrite
7.2
Massive ferrite
7.0
No change in
bulk composition
Carbon diffusion during

paraequilibrium nucleation.
No diffusion during growth
6.8
V0
6.6
300
600
900
1200
1500 1800
Temperature, K
Fig. 3
Molar volumes of the various forms of iron.

Source: Ref 5
Fig. 4
Pearlite
Martensite
Cooperative growth
of ferrite and
cementite
Diffusionless
nucleation and
growth
Transformation products of austenite. Source: Ref 12
Material Factors / 5
phase change in a stress-free material is usually
triggered by heat treatment, when the parent
phase passes through an equilibrium transformation temperature. Alternatively, the application of a stress in isothermal conditions can trigger transformation in circumstances where it
would not otherwise occur. Unusual effects can
occur when stress and temperature work together. The transformation may occur at remarkably low stresses or at very small deviations
from the equilibrium temperature. This is why
even minute stresses can greatly influence the
development of microstructure, and vice versa.
It is not surprising that transformation plasticity
can be obtained at stresses that are much smaller
than the conventional yield stress of the parent
phase.
Transformations, Residual Stresses,

and Related Phenomena
The strains due to phase transformations can
alter the state of residual stress or strain. It is well
known that the martensitic transformation of the
carburized surface of a steel component puts the
surface under compression. It is argued that this
is because of the expansion at the surface due to
formation of the lower-density martensite from
austenite.
Phase transformation can also compensate for

stress. Greenwood and Johnson (Ref 13, 14)
showed that when a phase change is accompanied by a change in volume, the tensile strain
expected when transformation occurs under the
influence of a tensile stress r is given by:
e
5 DV r
6 V rY
(Eq 2)
where rY is the yield stress of the weaker phase

and DV/V is the transformation volume strain.
The role of shear strains associated with transformation has been emphasized in later work by
Magee and Paxton (Ref 15, 16), and subsequently by Fischer (Ref 17), Leblond et al. (Ref
1822), Olson (Ref 23), and Bhadeshia et al.
(Ref 24). Not only does transformation affect
stress, but the latter modifies the development of
microstructure. The microstructure tends to be
more organized when transformation occurs in a
stresss parent phase, because the stress favors
the formation of certain orientations relative to
others. This is illustrated schematically in Fig.
5(b) to (d). These aspects will now be discussed
in more detail, because transformation plasticity
can radically alter the state of residual stress.
Table 2
Reconstructive
Displacive
Single
crystal
Causes permanent change in shape

Invariant-plane strain shape changewith a large shear component
Changes crystallographic orientation
Changes lattice type
Can lead to a density change
Table 3
Allotriomorphic ferrite
Idiomorphic ferrite
Pearlite
Widmanstatten ferrite
Bainite
Acicular ferrite
Martensite
Cementite plates
Mechanical twins ()
Annealing twins (c)
Slip
deformation
Mechanical
twinning
Displacive
transformation
Reconstructive
transformation
Yes
Yes
No
No
No
Yes
Yes
Yes
No
No
Yes
Yes
Yes
Yes
Yes
Yes
No
Yes
Yes
Yes
Shape change due to transformation
Transformation
(a)
Displacive transformations can be regarded as

modes of plastic deformation. Just as a combination of a plane and a direction constitutes a
deformation system for slip or twinning, the
habit plane and displacement vector of the
invariant-plane strain accompanying displacive
transformation completely describe the deformation system responsible for transformation
plasticity. The displacement vector describes the
sense of the macroscopic displacements accompanying transformation and, along with the habit
plane indices, also contains information about
the magnitude of the shear component and dilatational component of the displacements. Typical data for the deformation systems associated
with transformations are listed in Table 4. Note
that reconstructive transformations involve only
a volume change together with diffusional mass
flow, so it is not appropriate to regard them as
deformation systems in the present context.
Given the cubic crystal structure, and the fact
that habit planes tend to be irrational, there will
in general be 24 of these systems per austenite
grain, and they may operate simultaneously to
varying extents. Of course, unlike ordinary slip,
Characteristics of different modes of deformation
Characteristic
Deformation System
Shape change (a)
s(b)
d(b)
Morphology
Volume change
Volume change
Volume change
Invariant-plane strain
Invariant-plane strain?
0.00
0.00
0.00
0.36
0.22
0.22
0.24
0.21?
1/2
0.00
0.02
0.02
0.03
0.03
0.03
0.03
0.03
0.16?
0.00
0.00
Irregular
Equiaxed, faceted
Spherical colonies
Thin plates
Thin plates
Thin plates
Thin plates
Thin plates
Thin plates
Faceted
(a) An invariant-plane strain here implies a large shear component as well as a dilatational strain normal to the habit plane. (b) s and d refer to the
shear and dilatational strains, respectively. The values stated are approximate and will vary slightly as a function of lattice parameters and the details
of crystallography.
(b)
(c)
(d)
Shape changes accompanying unconstrained

transformations. Note that the horizontal scale
bars are all the same length. (a) The two kinds of shape
changes that occur when a single crystal of austenite transforms to a single crystal of ferrite, as a function of the mechanism of transformation. (b) Polycrystalline sample of austenite. (c) Polycrystalline sample of austenite that has
partially transformed by a displacive transformation mechanism into a random set of ferrite plates. (d) Polycrystalline
sample of austenite that has partially transformed by a displacive transformation mechanism into an organized set of
ferrite plates.
Fig. 5
Table 4
Deformation systems associated with transformations
Phase
Habit plane indices
Displacement vector
(0.363 0.854 0.373)
[0.195 0.607 0.771]
0.185
Bainite
(0.325 0.778 0.537)
[0.159 0.510 0.845]
0.27
Widmanstatten ferrite
(0.506 0.452 0.735)
[0.867 0.414 0.277]
0.36
Martensite
Note: Typical habit plane and displacement directions for low-alloy steels. The indices all refer to the austenite phase. Note that the indices stated are
approximate, since the habit plane and displacement direction are usually irrational. The displacement vector does not quite lie in the habit plane
because the dilatational strain is directed normal to the habit plane. The magnitude of the displacement is given by m, which is the total displacement
including the shear and the dilatational components.

the different deformation systems within an austenite grain cannot intersect, except in special
circumstances where intervariant transformations are possible, as is the case with some
shape-memory alloys. It follows that the ordiN
A (applied tensile stress)
max
Resolution of the applied stress, rA. The normal

stress, rN, and the shear stress, s, both act on the
habit plane. The vector d is the direction along which lie
the shear displacements of the shape deformation. smax is
the maximum shear stress on the habit plane, but s is given
by resolving smax along d. Note that d differs slightly from
the displacement vector of the IPS, which includes a dilatational component in addition to the shear.
Fig. 6
Table 5 Typical values of the mechanical

driving force coefficients
DG/r, J/(mol MPa)
Nature of stress
Uniaxial tension
Uniaxial compression
Elastic crack tip (a)
0.86
0.58
1.42
Tensile
+
Compression
0
Mechanical Driving Force

The interaction of an applied elastic stress
with a phase change can occur in two ways:
1. The stress can alter the driving force for the
transformation.
2. The stress can change the appearance of the
microstructure by favoring the formation of
those variants which best comply with the applied stress.
For reconstructive transformations, only the
hydrostatic component of stress can interact with

Hydrostatic
compression
Stress (below austenite yield strength)
Indication of how the transformation-start temperature (for Widmanstatten ferrite, bainite, acicular ferrite, or martensite) should vary as a function of the
nature and magnitude of an applied stress whose magnitude is less than that of the yield stress.
Fig. 7
Table 6 Sensitivity of transformation-start

temperatures in steels to applied stress
Phase
Nature of stress
Martensite
Bainite
Eutectoid
Martensite
Pressure
Pressure
Pressure
Tensile
Source: Ref 32
Sensitivity, K/MPa
0.06
0.09
0.011
0.06
1000
Free energy, J mol1
Change in bainite-start temperature
(a) The stress state for the crack tip is multiaxial, but the coefficient is
calculated by expressing the stress in terms of the von Mises equivalent
tensile stress. Source: Ref 32
nary notion of work hardening does not apply.

Work hardening nevertheless manifests itself via
a different mechanism, in which the stability of
the austenite increases as it becomes ever more
finely divided.
The Taylor/von Mises criterion (Ref 25, 26)
states that in any given crystal, a minimum of
five independent slip systems is necessary to
produce an arbitrary shape change. A crystal in
a polycrystalline aggregate has to accommodate
the arbitrary deformations of neighboring grains.
Therefore, a polycrystalline material is brittle
unless each grain contains at least five independent slip systems. Similar logic can be applied
to the crystallographic variants of a phase generated by displacive transformation. The habit
plane is predicted theoretically (Ref 27, 28) and
found experimentally (Ref 29) to have irrational
indices. This means that there exist, in principle,
24 possible variants of the habit plane per grain
of austenite (that is, 24 independent deformation
systems). Given this large number of transformation variants available per grain, the Taylor
criterion leads to the conclusion that transformation plasticity can cause, or accommodate,
any externally imposed, arbitrary shape
changeassuming that a sufficient quantity of
parent phase is available. It follows that polycrystalline samples can remain intact at grain
boundaries when transformation plasticity is the
sole mode of deformation.
the volume change. The corresponding interaction with displacive transformations is much
larger because of the shear component of the
IPS.
For displacive transformations, the influence
of stress on the transformation can be expressed
as a mechanical driving force (DGmech), which is
the work done by the external stress in producing
the macroscopic shape deformation (Ref 30, 31):
DGmech rNd ss
(Eq 3)
where rN is the normal stress on the habit plane

and s is the component of the shear stress on the
habit plane that is parallel to the direction along
which the shear displacements of the shape deformation occur (Fig. 6). The strains d and s are
the dilatational and shear components, respectively, of the shape deformation. Some typical
values of the mechanical driving force terms are
given in Table 5. Given a free choice of some
12 to 24 crystallographic variants of the transformation product in each grain of austenite, the
work done by the shear stress is always expected
to be positive, whereas that due to the dilatational component depends on the sign of rN. For
steels, this latter component is relatively small.
Any observed consequences of stress must therefore reflect the dominant role of the shear component unless the stress is purely hydrostatic.
Since the shear stress remains positive irrespective of whether the sample is pulled in tension or uniaxially compressed, and since the
shear component of the shape change is large, a
uniaxial stress will always cause a temperature
increase for displacive transformations in steels.
Hydrostatic stress, on the other hand, has no deviatoric components and consequently interacts
only with the dilatational component of the
shape change. Thus, hydrostatic compression is
expected and found to lead to a decrease in the
transformation temperature (Fig. 7); some data
(Ref 32) on the sensitivity of the transformation
temperature to applied stress are presented in Table 6.
Limits to Stress-Assisted
Transformation
Chemical
0 Mechanical
1000
Total
2000
300
400
500
600
700
800
900
Temperature, C
Typical magnitudes of the chemical and mechanical driving forces for stress-affected transformation. The mechanical driving force is estimated for an
applied stress that is equal to the yield stress of austenite.
Since this yield stress becomes small at high temperatures,
the contribution of the mechanical driving force also decreases. Therefore, transformation becomes impossible as
the temperature exceeds about 700 C (1290 F).
Fig. 8
At temperatures close to that at which the

equilibrium transformation occurs, an applied
stress can assist reaction when the chemical driving force is insufficient to achieve the change on
its own. There must exist a point, however, when
the applied stress simply cannot provide enough
mechanical driving force to complement the
chemical term to give a driving force large
enough to induce transformation. After all, the
magnitude of the stress that can be applied is
limited by the yield point of the parent phase.
Thus, there are limits to what can be achieved
by the application of stress as a stimulus to transformation (Fig. 8).
Residual stresses are often introduced unintentionally during fabricationfor example,

during welding or heat treatment. A few elegant
experiments illustrate how phase transformations interact with the buildup of residual stress.
Using bainitic, martensitic, and stable austenitic steels, Jones and Alberry (Ref 33, 34) demostrated that transformation plasticity during
the cooling of a uniaxially constrained sample
from the austenite phase field acts to relieve the
buildup of thermal stress as the sample cools. By
contrast, the nontransforming austenitic steel exhibited a continuous increase in residual stress
with decreasing temperature, as might be expected from the thermal contraction of a constrained sample.
When the steels were transformed to bainite
or martensite, the transformation strain compen-
500
AISI 316
9CrMo
2CrMo
Austenite YS
400
Stress, MPa
300
200
100
0
100
200
400
600
800
1000 1200 1400
Temperature, C
(a)
Stress
Plastic
strain in
austenite
Transformation
finished
Stress due to
thermal
contraction
of austenite
Stress due to
thermal contraction
of ferrite
Transformation
begins
sated for any thermal contraction strains that

arose during cooling. Significant residual
stresses were therefore found to build up only
after transformation was completed and the specimens approached ambient temperature (Fig. 9).
The experiments contain other revealing features. The thermal expansion coefficient of austenite (1.8 106 /K) is much larger than that
of ferrite (1.18 106 /K), and yet the slope of
the line prior to transformation is smaller when
compared with that after transformation is complete (Fig. 9). This is because the austenite yields
to accommodate the thermal contraction, which
is possible because the yield strength of the austenite is reduced at elevated temperatures. Ferrite
is strong at low temperatures, so the slope of the
stress/temperature curve (after transformation is
complete) is steeper and consistent with the magnitude of thermal contraction strains.
Interpretation of experimental data of the kind
illustrated in Fig. 9 is difficult in the region of
the stress/temperature curve where transformation occurs. The popular view that the volume
change due to transformation is the major component of transformation plasticity is probably
incorrect for displacive transformations such as
bainite or martensite. The shape change due to
transformation has a shear component that is
much larger than the dilatational term (Table 3).
Admittedly, this shear component should, on average, cancel out in a fine-grained polycrystalline sample containing plates in many orientations (Fig. 5c). However, the very nature of the
stress effect is to favor the formation of selected
variants, in which case the shear component rapidly begins to dominate the transformation plasticity (Fig. 5d).
Bulk transformation strain, %
Transformation under Constraint:

Residual Stress
3
75 MPa compressive stress
2
1
0
2
100
200
300
400
500
600
Transformation time, s
(b)
Temperature
(a) Plot of residual stress versus temperature for

a martensitic (9CrMo), bainitic (2CrMo), and austenitic steel (AISI 316). Adapted from Ref 33, 34. (b) Interpretation of the Jones and Alberry experiments. The thermal
expansion coefficient of austenite is much larger than that
of ferrite.
Fig. 9
Anisotropic Strain and

Transformation Plasticity
When an unstressed polycrystalline sample of
austenite is transformed to plates of ferrite, the
shear caused as each randomly oriented plate
forms is canceled on a macroscopic scale; only
the volume expansion is observed experimentally. However, if the plates do not form at randomfor example, when certain variants are favored because they comply better with the
external stressthe shear strains are no longer
canceled out. Transformation will then lead to
highly anisotropic strains, as illustrated in Fig.
11. Naturally, any anisotropy will be greatest for
displacive rather than reconstructive transformations, given that the former involve large
shear strains.
Modeling Anisotropic
Transformation Strains
1
The residual stress at ambient temperature is

larger when the austenite finishes transformation
at a high temperature. This is because thermal
contraction strains can no longer be compensated by transformation plasticity once the austenite has decomposed. Low transformation temperatures help minimize residual stresses.
High-strength welding alloys used for making
submarine hulls therefore have transformation
temperatures of less than about 250 C (480 F).
Figure 10 illustrates one kind of distortion
found in welds, measured in terms of the angle
h through which the unconstrained plates rotate
as they cool. Table 7 shows how the distortion
depends on the temperature at which the majority of the transformation is completed, for two
manual metal arc welds deposited with a 60 Vjoint preparation in a multipass fabrication involving about 11 layers, with two beads per layer
to complete the joint. The distortion is clearly
larger for the case where the transformation is
exhausted at the higher temperature.
Development of anisotropic transformation

strain when bainite forms under the influence
of a constant, elastic applied compressive stress. Note that
the shear strain associated with the formation of one plate
is about 26%, with a volume change of about 3%. The
potential for anisotropy is therefore much greater than illustrated here.
Fig. 11
Consider a distribution of bainite variants

along all radial directions in a circle with the
compression axis as its diameter (Ref 35, 36).
The circle is divided into 18 equal segments (i
1 18), each segment representing a particular orientation of bainite habit plane. The
choice of 18 segments is convenient and arbitrary. The compression axis of the sample is
taken to be the z direction, the x and y directions
Table 7 Chemical composition, calculated transformation temperature range (DT ), and

measured distortion (h) for two manual metal arc, multipass weld deposits
Composition, wt%
C
Fig. 10
Distortion caused by welding two plates that

were originally flat
0.06
0.06
Si
Mn
Ni
Mo
Cr
DT, C(F)
0.5
0.3
0.9
1.6
...
1.7
...
0.4
...
0.35
802400 (1476750)
422350 (792660)
14.5
8
Source: H.K.D.H. Bhadeshia and L.-E. Svensson, unpublished data, 1994

being radially orientated; the unit vectors x, y,
and z define the orthonormal basis X of the sample, giving a corresponding reciprocal basis X *.
The shear and dilatational components of the IPS
accompanying the growth of bainite are approximately s 0.22 and d 0.03. Thus, the 3
3 deformation matrix describing the shape deformation is given by:
P
1 fse1 p1 f dp1 p1
fse2 p1 f dp2 p1
fse3 p1 f dp3 p1
fse1 p2 f dp1 p2
1 fse2 p2 f dp2 p2
fse3 p2 f dp3 p2
e1
p1
P I fs e2 ( p1 p2 p3) f d p2 ( p1 p2 p3)
e3
p3
where I is a 3 3 identity matrix. A further

reduction of notation is achieved using the
MacKenzie and Bowles notation (Ref 27):
(XPiX ) I fis[X;ei]( pi ;X *)
fid[X;pi]( pi; X *)
(Eq 4)
where the subscript i identifies a particular segment of interest and X and X *, respectively, represent the real and reciprocal bases of the coordinate system in which the deformation is
described. The notation due to MacKenzie and
Bowles (Ref 27) is discussed in detail in Ref 35.
The components of the shear direction and the
dilatation direction are given by:
[X;ei] fi[cos(hi) 0 sin(hi)]
[X; pi] fi[sin(hi) 0 cos(hi)]
120
180
210
(a)
Ui
330
where is the angle between the shear direction

and the direction of the shear component of the
applied stress as resolved onto the habit plane.
To facilitate a two-dimensional analysis, the
value of i is taken to be zero. A positive value
of Ui adds to the chemical driving force (DG ac
G c G a) for transformation; a negative
value thus opposes transformation. Using these
values of interaction energies, the model can be
modified so that the segments transform in an
order of decreasing Ui. There is, however, a further complication. The effect of stress should be
largest when the interaction energy is large compared with the chemical driving force. To allow
for this, the volume fraction fi of each segment
90
60
150
30
180
210
300
270
r
[s sin 2hI cos i d(1 cos 2hi)]
2
120
30
(X Pi X ) [1 0 0]
i1
where (x 1) and (z 1) give the strains

along the x and z directions. These are assumed
to be equal to radial and longitudinal strains eR
and eL, respectively.
It is expected that those segments that comply
best with the applied stress transform most rapidly, whereas the others do so at a lower rate, or
not at all. This can to some extent be incorporated into the model by calculating the energy
change Ui as the stress interacts with the shape
deformation of a particular variant (i). Patel and
Cohens method (Ref 30) gives:
60
150
240
(X Pi X ) [0 0 1]
i1
18
[X; x]
where p is the unit normal to the habit plane and

e is the unit direction along which the shear occurs. This can be written more succinctly as:
90
18
[X; z]
whereas a unit vector along x changes to
fse1 p3 f dp1 p3
fse2 p3 f dp2 p3
1 fse3 p3 f dp3 p3
where hi represents the orientation of the habit

plane of variant i and fi is the volume fraction of
bainite located in segment i.
A unit vector along the z direction changes to
a new vector z given by:
330
240
(b)
300
270
Transformation behavior inherent in the model (Ref 36). The dots illustrate the area fraction of each segment,
which in the calculations is scaled according to the value of Ui DG c. (a) Zero stress. All segments have
equal area fraction, and the order in which they transform is irrelevant. (b) Applied stress of 40 MPa and DG c 400 J/
mol. The area fractions of the segments are no longer equal. The segments in which the distance of the dot from the origin
is largest transform first.
Fig. 12
can be scaled according to the value of Ui

DG ac.
Note that for the model calculations, the transformation occurs with the most favored variants
growing first (Fig. 12). The model thus exaggerates the effect of stress, since in reality, for the
sort of stress levels considered experimentally,
no variant is likely to be entirely suppressed. In
addition, the grains in a polycrystalline sample
are randomly oriented, so that perfect compliance with the applied stress is impossible. Nevertheless, the trends revealed by the model are
expected to be correct.
The experimental data that need explaining,
and their interpretation in terms of the model are
summarized in Fig. 13 and may be stated as follows:
Without any stress, in a random polycrystal-
line sample, the transformation strains are

isotropic. This is easily understood since the
shear components of randomly oriented
plates tend to cancel out (Fig. 13a).
The application of the small stress at a high
transformation temperature (that is, a small
chemical driving force) causes the development of anisotropic strains, the transverse
strain first being negative and then positive
(Fig. 13b). The same effect is observed for a
large stress and low temperature (that is, a
large driving force). The model explains this
effect when it is assumed that the favored variants form first, but that the stress is not large
enough to suppress the eventual formation of
other variants. The signs of eL and eR are always opposite for the favored variants, but
are identical for the rest of the variants.
Therefore, the transverse strain is initially
negative but then becomes positive as transformation progresses. The low-stress/hightemperature situation is equivalent to the
high-stress/low-temperature case because in
both of these circumstances, variants that are
not favored cannot be suppressed. In the former case the stress is too small for suppression, whereas in the latter case the chemical
driving force is too large to permit suppression.
When a large stress is applied at a high temperature, the favored variants dominate.
Therefore, the strains are always of opposite
sign (Fig. 13c).
The model is thus capable of qualitatively explaining all the essential features of the formation of bainite under the influence of a small tensile stress. A uniaxial compressive stress (as
used in the experiments described below) simply
causes a reversal of the signs of the longitudinal
and transverse stresses; there is also a minor effect from the unfavorable interaction between
the compressive stress and the dilatational component of the IPS shape deformation.
The most interesting conclusion to emerge
from comparison of the model with experimental
data is that transformation under the influence of
a mild stress occurs sequentially. Variants that
comply with the applied stress grow first, fol-
lowed by those that do not. This also carries the
implication that the interaction of the stress is
with the growth process (that is, the IPS shape
deformation) rather than the strain field of the
nucleus, which is likely to be different. It is
worth noting that there are similar results for
martensite: most favored variants grow first in
the sequence of transformation under stress (Ref
15, 24).
Summary
Many of the thermal properties of steelsfor
example, heat capacity, thermal expansion coefficients, and latent heats of transformation
are remarkably well understood. Indeed, commercially available thermodynamic databases
and programs can be used to estimate these
quantities as a function of temperature and
chemical composition. This capability has not
been exploited in the analysis of residual
stresses, even though phase diagram calculations
using the same software are now routine in industry and academia.
Other properties, such as elastic modulus, are
not yet calculable in the same manner. It may be

the case that they are insensitive to alloying, but
that remains to be demonstrated in the context
of residual stress analysis.
There is little doubt that transformations in
steel play a major role in the development of
residual stresses. For reconstructive transformations (for example, pearlite), it is the difference
in density between the parent and product phases
that contributes to transformation plasticity. The
plasticity can be much larger for displacive
transformations (Widmanstatten ferrite, bainite,
martensite) because of the large shear component of the shape deformation when these transformation products form. These are quite sophisticated effects which, with few exceptions,
are not incorporated in most residual stress analyses.
3.
4.
5.
6.
7.
8.
REFERENCES
9.
1. T. Inoue and Z. Wang, Mater. Sci. Technol.,
Vol 1, 1985, p 845850
2. P.W. Fuerschbach, in The Metal Science of
Joining, M.J. Cieslak, J.H. Perepezko, S.
10.
11.
12.
L
T
+
L, T
Strain
Strain
13.
14.
Transformation time
15.
16.
Equivalent time
(a)
17.
L
18.
Strain
Strain
L
0
19.
20.
21.
Equivalent time
Transformation time
(b)
22.
L
23.
+
Strain
Strain
0
Transformation time
24.
0
T
25.
Equivalent time
26.
(c)
27.
Schematic of the reported variations (Ref 24) in longitudinal and radial strains during the isothermal formation
of bainite under the influence of a tensile load, presented alongside predictions (Ref 36) from the crystallographic/thermodynamic model. The stresses are all intended to be below the austenite yield strength, and the data in this
case refer to uniaxial tension. (a) Zero stress, any temperature. (b) Small stress, low temperature. (c) Small stress, high
temperature; or large stress, low temperature.
Fig. 13
28.
Kang, and M.E. Glicksman, Ed., Minerals,

Metals and Materials Society, 1992, p 21
30
R. Schroder, Mater. Sci. Technol., Vol 1,
1985, p 754764
L. Kaufman, E.V. Clougherty, and R.J.
Weiss, Acta Metall., Vol 11, 1963, p 323
335
L. Kaufman, in Energetics in Metallurgical
Phenomenon, Vol III, W.M. Mueller, Ed.,
Gordon and Breach, 1967, p 5584
K. Hack, Ed., The SGTE Casebook: Thermodynamics at Work, Institute of Materials,
1996
D.G. Pettifor and A.H. Cottrell, Ed., Electron Theory in Alloy Design, Institute of
Materials, 1992
J.W. Christian, in Decomposition of Austenite by Diffusional Processes, V.F. Zackay
and H.I. Aaronson, Ed., Interscience, 1962,
p 371386
H.M. Clark and C.M. Wayman, Phase
Transformations, American Society for
Metals, 1970, p 59114
J.D. Watson and P.G. McDougall, Acta Metall., Vol 21, 1973, p 961973
J.W. Christian, Physical Properties of Martensite and Bainite, Special Report 93, Iron
and Steel Institute, 1965, p 119
H.K.D.H. Bhadeshia, Bainite in Steels, 2nd
ed., Institute of Materials, 2001, p 1453
G.W. Greenwood and R.H. Johnson, Proc.
Roy. Soc., Vol 238, 1965, p 403422
R.H. Johnson and G.W. Greenwood, Nature, Vol 195, 1962, p 138139
C.L. Magee, Ph.D. thesis, Carnegie Mellon
University, 1966
C.L. Magee and H.W. Paxton, Trans. Met.
Soc. AIME, Vol 242, 1968, p 17411749
F.D. Fischer, Acta Metall. Mater., Vol 38,
1990, p 15351546
J.B. Leblond, G. Mottet, and J.C. Devaux,
J. Mech. Phys. Solids, Vol 34, 1986, p 395
409, 411432
J.B. Leblond, J. Devaux, and J.C. Devaux,
Int. J. Plasticity, Vol 5, 1989, p 551572
J.B. Leblond, Int. J. Plasticity, Vol 5, 1989,
p 573591
J.B. Leblond and J. Devaux, Residual
Stresses, Elseviers, 1989, p 16
J.B. Leblond, Internal Report CSS/L/NT/
90/4022, FRAMASOFT, 1990, p 112
G.B. Olson, Deformation, Processing and
Structure, American Society for Metals,
1982, p 391424
H.K.D.H. Bhadeshia, S.A. David, J.M. Vitek, and R.W. Reed, Mater. Sci. Technol.,
Vol 7, 1991, p 686698
R. Von Mises, Z. Angew. Math. Mech., Vol
8, 1928, p 161
G.I. Taylor, J. Inst. Met., Vol 62, 1928, p
307
J.K. MacKenzie and J.S. Bowles, Acta Metall., Vol 2, 1954, p 138147
M.S. Wechsler, D.S. Lieberman, and T.A.
Read, Trans. Amer. Inst. Min. Metall. Eng.,
Vol 197, 1953, p 15031515

29. A.B. Greninger and A.R. Troiano, Trans.
Amer. Inst. Min. Metall. Eng., Vol 140,
1940, p 307336
30. J.R. Patel and M. Cohen, Acta Metall., Vol
1, 1953, p 531538
31. L. Delaey and H. Warlimont, in Shape
Memory Effects in Alloys, J. Perkins, Ed.,
TMS-AIME, Plenum Press, 1975, p 89114
32. S. Denis, E. Gautier, A. Simon, and G.

Beck, Mater. Sci. Technol., Vol 1, 1985, p
805814
33. W.K.C. Jones and P.J. Alberry, Ferritic
Steels for Fast Reactor Steam Generators,
British Nuclear Engineering Society, 1977,
p 14
34. W.K.C. Jones and P.J. Alberry, Residual
Stresses in Welded Constructions, Welding

Institute, 1977, Paper 2
35. H.K.D.H. Bhadeshia, Worked Examples in
the Geometry of Crystals, Institute of Metals, 1987
36. A. Matsuzaki, H.K.D.H. Bhadeshia, and H.
Harada, Acta Metall. Mater., Vol 42, 1994,
p 10811090

DOI: 10.1361/hrsd2002p011

Prestress Engineering of Structural

Material: A Global Design Approach
to the Residual Stress Problem
J. Lu, LASMIS, Universite de Technologie de Troyes, France
ALL MANUFACTURING PROCESSES introduce residual stress into mechanical parts,

which influences its fatigue behavior, fracture
strength, and even its corrosion resistance. Few
metalworking methods exist that do not produce
new stresses. The role of residual stress is, therefore, very important when designing mechanical
parts. Over the last few years, an increasing number of studies have been carried out to understand the effects of residual stress on mechanical
performance. This article attempts to present a
global approach to including residual stress in
expected fatigue life calculations, and the possibility of introducing it into mechanical engineering design offices. The definitions and origins of residual stress according to production
methods are first presented. Then, shown are the
problems involved in correctly adapting these
modeling techniques for use in design offices
and the industrial consequences of taking residual stress into account on quality assurance control procedures. This article deals mainly with
residual stress measurement techniques and the
overall necessity to combine destructive (incremental hole-drilling method) and nondestructive
(x-ray and neutron diffraction) methods in order
to precisely evaluate the residual stress distribution. Some new results concerning optical
methods are also discussed. Shown, too, are the
beneficial and harmful effects of residual stress
on the resistance of structures or industrial components, depending on whether they are tensile
or compressive. The different models used to
predict residual stress induced by different types
of processing are then presented. The last section
shows the effect of residual stress on fatigue behavior. A model based on the finite element
method (FEM) for predicting the relaxation of
residual stresses is presented. Finally, prediction
of fatigue life, taking residual stress into account,
using FEM is discussed. A new approach to concurrent engineering applied to the design of mechanical components with residual and applied
stress consideration is presented.
Residual stress is usually defined as the stress

that remains in mechanical parts that are not subjected to any outside stresses. Residual stress exists in practically all rigid parts, whether metallic
or not (wood, polymer, glass, ceramic, etc). It is
the result of the metallurgical and mechanical
history of each point in the part and the part as
a whole during its manufacture.
In the case of structural materials, surface engineering can lead indirectly to the innovation of
genuinely new materials based on conventional
materials. The frustrating slowness of some of
these developments reflects not only difficulties
in scaling up laboratory techniques, but also general conservation in the engineering industries
and a reluctance to change established habits.
One of the factors that contributes to slowing the
pace is that the current approach to surface engineering directly correlates performance to processing parameters; designers are left without the
tools needed to optimize and take into account
modified or prestressed surfaces for a particular
application, and the processors are left without
appropriate goals for surface properties.
The recent European Network of Surface and
Prestress Engineering and Design (ENSPED)
project led by the University of Technology of
Troyes and funded by the European Union is one
of the contributions currently being made to the
development of such tools. The aim is to develop
a project with a hard core of about 18 partners
representing a balanced selection of Europes
major industries, such as SNECMA; European
Aeronautic, Defense and Space (EADS); ABB;
Siemens; Volvo; Fiat; British Aerospace; Robert
Bosch; Hydro Aluminum; Wartisla; and so on.
It is based on building a bridge between the
fields of surface modification and prestress processing in materials engineering and computeraided design in mechanical engineering in order
to find an appropriate way of approaching this
interdisciplinary area. The members of the project will focus on the field of prestress processing. The main goal of the prestress engineering
approach is the optimization of residual stress for
an objective mechanical behavior of the material

and structure. Under this approach, the residual
stress must not be considered as a parameter that
only depends on the material processing conditions, but must also be considered as a parameter
that can be optimized.
There is an increasing interest in how the state
of residual stress affects the mechanical properties of a material and its structure. The failure of
a structure or a mechanical component is not
only due to external loads. Residual stress is an
important parameter in this respect. All manufacturing processes, for example, introduce a
new state of residual stress. This can have a positive effect, such as increasing the fatigue limit in
the case of surface compressive stress, or it can
have a negative effect, such as decreasing the
stress corrosion behavior of a material in the case
of tensile residual stress.
Basic and applied research in the field of residual stress has been stepped up in the last few
years. Residual stress is taken into account in
advanced design in the aerospace, automotive,
and nuclear industries. Even the microelectronics industry is starting to take residual stress into
account for the dimensional stability of electronic packaging.
The introduction of advanced materials has
also contributed to the development of knowledge in the field of residual stress. In fact, many
new materials are multimaterials, for example,
metal-matrix composites, plasma-sprayed coating, physical vapor deposition (PVD), and chemical vapor deposition (CVD) coatings, which
contain residual stress as a result of the thermal
and mechanical incompatibilities of the different
phases of the material or structure.
Figure 1 shows the different fields of research
in which residual stress is taken into account and
its relevance for industrial applications. Three
main fields must be developed for a global approach of prestress engineering: measurement
techniques for the quality control and processing
analysis, processing parameter optimization and
processing modeling, and modern design tool for

the life cycle simulation with residual stress consideration.
Origins of Residual Stress

In general, macroscopic residual stress can be
induced due to:
Nonhomogeneous plastic flow under the ac-
tion of external treatment (shot peening, autofretting, roller burnishing, hammer peening,
shock laser treatment)
Nonhomogeneous plastic deformation during
nonuniform heating or cooling (ordinary
quenching, molding of plastics)
Structural deformation from metalworking
(heat treatment)
Heterogeneity of a chemical or crystallographic order (nitriding or case hardening)
Various surface treatments (enameling,
nickel plating, chrome plating, PVD and
CVD coating)
Differences in expansion coefficients and mechanical incompatibility of the different components of composites (composites with a
metallic and organic matrix, ceramic coatings)
Table 1 shows the different origins of residual

stress for metalworking operations usually carried out in the industry. To produce an industrial
part, one or several of the techniques listed in the
table can be used. To calculate the residual stress
existing in a part, the source of the stress must
be identified first.
Residual Stress Measurement Techniques
in Global Approach and Quality Assurance.
Over the last few decades, various quantitative
and qualitative techniques have been developed
(Ref 1). These techniques are used for the processing optimization and quality control of material. In general, a distinction must be made between destructive and nondestructive methods.
The first series of methods is based on destroying the state of equilibrium in the mechanical component. The residual stress is then evaluated from its relaxation. However, it is only
possible to measure the consequences of stress
relaxation and not the relaxation itself (displacement, fracture, strain). In most cases, the change
in strain is selected as the parameter to be studied. The following procedure is used:
1. Creation of a new stress state by machining
or layer removal
2. Detection of the local change in stress by
measuring the strain or displacement
3. Calculation of the residual stress as a function
of the strain measured using the elastic theory
(analytical approach or numerical calculations such as FEM)
During the recent years, the incremental holedrilling method is extensively used. It is sensitive to the first kind of residual stress, that is, the
macroscopic residual stress. The principle of this
technique is simple. It involves monitoring the
change in strain when a hole is drilled in a component with residual stress. These strain measurements can be related to the original residual
stress distribution in the analyzed sample at the
hole location. The relationship between the
strain and the residual stress can be calculated
with the calibration coefficients Ain and Bin. The
general approach used to determine the Ain and
Bin FEM is detailed in Ref 2. Recently, the high
sensitivity moire interferometer and incremental
hole-drilling method for residual stress measurement has been developed. The theoretical development of a combined method is introduced in
Ref 3 and 4. The relationship between the threedimensional surface displacements produced by
introducing a blind hole and the corresponding
residual stress is established by employing the
existing theoretical solution containing a set of
undetermined coefficients. The coefficients are
calibrated by the three-dimensional finite ele-
Quality control
Manufacturing
Measurement
technique
Prestress process
development and
modeling
Take residual stress into

account in integrated design
Relaxation of
residual stress
Dimensional
stability
Fig. 1
Effect of residual
stress on behaviors
New design
Mechanical
design
Main research fields and industrial application fields in which residual stress is taken into account
ment method. The in-plane surface displacements Ux and Uy and out-of-plane surface displacement Uz produced by the relaxation of
residual stresses are obtained using moire interferometry and Twyman-Green interferometry,
respectively. Figure 2 shows three-dimensional
displacement as a function of the drilling depth.
The main advantage of this technique is the possibility of studying the residual stress with an inplane stress gradient (Ref 5). It is also possible
to study composite materials (Ref 6). In this
case, the through thickness residual stress distribution was evaluated ply by ply in a carbon fiber/epoxy composite. Figure 3 shows an example of residual stress determined by using this
technique.
Recently, this method was used (Ref 7) to
study the Plastic Ball Grid Array (PBGA) package, which is a cost-effective surface-mounting
package with a high-density interconnection,
low profiles, and light weight. It is currently used
in many electronic products, including portable
telecommunication and computing products. A
typical structure in a PBGA package consists of
four layers: a plastic molding compound, a silicon chip, a chip-attach adhesive layer, and an
organic chip carrier. Due to the coefficient of
thermal expansion mismatch between the silicon
chip, the plastic compound, and the organic chip
carrier, considerable residual stresses are developed in the package during the assembly process. The process-induced residual stress can
play a significant role in the reliability of electronic components and packages. Since a PBGA
package is small and the surface layer is made
of a plastic material, it has proved very difficult
to use other existing methods of residual stress
measurement. In this research work, a practical
method has been developed to determine residual stress for electronic packaging. In this
method, blind holes are drilled into the specimens, and relationships are established between
the released surface displacement and the corresponding residual stress by introducing a set
of calibration coefficients. A multilayer three-dimensional FEM is established to determine the
relevant calibration coefficients. The surface displacements are measured accurately in a small
region around the hole. For a practical PBGA
package, the tensile residual stress is determined
in both the plastic molding compound and the
glass/epoxy laminate chip carrier. The method
is accurate, simple, convenient, and practical.
More applications in the field of electronic products are anticipated.
The x-ray diffraction and neutron diffraction
methods are based on the measurement of lattice
strains by studying variations in the lattice spacing of the polycrystalline material. The first
method measures the residual strain on the surface of the material, and the second measures the
residual strain within a volume of the sample.
Diffraction techniques can be used to study all
three types. The peak shift method is sensitive
to the first two, whereas line broadening is sensitive to the second and third types. These techniques have been used to measure the first type
Prestress Engineering of Structural Material: A Global Design Approach to the Residual Stress Problem / 13
of residual stress in different phases of advanced
materials, such as metal-matrix composites (in
matrix and in reinforcement) (Ref 8, 9).
Residual stress can be incorporated into the
design of mechanical components. Although this
leads to a better knowledge of the fatigue life of
parts and reduces the safety coefficient at the design stage, it also poses a host of new problems
on a quality assurance level. All statistical controls are only applied today to a few critical components in the aeronautical and nuclear industries; this practice could easily become
Fig. 2
widespread. Rapid ways of checking the residual

stress must therefore be developed. The methods
used industrially (x-ray diffraction and the incremental hole method) will not be sufficient in the
future. Other nondestructive testing (NDT) techniques (ultrasound, magnetic methods, acoustic
emission) are presently being developed. But as
they currently stand, these techniques use physical parameters that depend not only on the residual stress present in the parts, but also on microstructural changes. In the near future, NDT
techniques will be applied at the same time as
Three-dimensional displacements measured for different hole depth by a combination of optical methods and
the incremental hole-drilling method in a shot peened sample. Champ, field
Table 1
the reference techniques. Figure 4 proposes a residual stress inspection plan. Other new techniques, such as neutron diffraction and optical
method, will be introduced in industry. A detailed analysis of different techniques of measurement is out of the scope of this article.
Modeling of Process
The experimental techniques developed previously can contribute to the development of a
residual stress prediction model. The results calculated by the model make it easier to take the
residual stress into account during the mechanical design. The residual stress induced by the
thermal processing is extensively treated in the
other articles of this Handbook. Only some examples concerning the mechanical surface treatments and the comparison between the numerical simulation and experimental validation are
presented briefly.
Two kinds of models can be mentioned: analytical models and numerical models. In this
case, mechanical surface treatment models were
developed for shot peening (Ref 10) and cold
rolling (Ref 11), while several finite element
codes were used or developed for welding,
grinding, heat treatment (quenching), and thermal cutting (Ref 12, 13). The results show good
correlation between the prediction of the model
and the experimental results. But it can also be
seen that three-dimensional calculations are necessary to obtain good results in all directions. If
a two-dimensional calculation is used, the residual stress evaluation correlates well in one direction only. So, in the future, three-dimensional
calculations will be of greater significance for
real case modeling (Ref 13).
In the case of shot peening, a three-dimensional finite element dynamic model was developed to obtain a better description of the shotpeening process and to introduce this approach
into the component design (Ref 14). With the
Main origins of residual stress resulting from different manufacturing processes
Process
Casting
Shot peening, hammer peening, roller burnishing, laser
shock treatment, bending, rolling, chasing, forging,
straightening, extrusion
Grinding, turning, milling, drilling, boring
Mechanical
Thermal
Structural
No
Heterogeneous plastic deformation
between the core and surface of the
part
Plastic deformation due to the removal
of chips
No
No
Temperature gradient during cooling

No
Phase transformation
Depends on the material
Temperature gradient due to heating

during machining
Temperature gradient
Phase transformation during machining if

the temperature is sufficiently high
No
Change of volume due to a phase change
No
Thermal incompatibility
Welding
Brazing
Electroplating
Shrinkage
Mechanical incompatibility
Thermal incompatibility
Thermal spraying (plasma, laser, HVOF)
Mechanical incompatibility,
microcracking
Thermal incompatibility, temperature

gradient
New chemical component with volume

modification
Microstructural change (HAZ)
New phase at interface
Composition of plating depending on bath
used
Change of phase in plating
Quenching without a phase transformation

Surface quenching with a phase change (induction, EB,
laser, plasma, classical methods)
Case hardening, nitriding
PVD, CVD
Composite
EB, electron beam; HVOF, high-velocity oxygen fuel; HAZ, heat-affected zone
Change of phase
No

help of three-dimensional modeling, the enormous influence of shot interaction is verified by
simulating simultaneous impacts. This simulation is very similar to the industrial process. Results of the residual stress obtained by the simulation are closer to the experimental results, if a
three-dimensional FEM is used (Fig. 5). The
higher the coverage rate, the lower the intensity
of the stress. This study shows how FEM can be
used to model shot peening and determine the
associated residual stress field.
The results obtained for both a single impact
and several impacts follow the general distribution of measured residual stress fields. This type
of analysis seems promising for studying the repercussions of a large number of parameters.
However, these methods are only efficient when
a maximum number of experimentally measured
parameters can be introduced. They also need to
be improved if they are to take most phenomena
into account. The proposed models could be
completed by analyzing the influence of the type
of contact, the direction of the shot during the
impact, behavior of the material, and so on. It
could then be possible to introduce the residual
stress field into metal parts in order to study their
behavior, using numerical simulation, for several
types of mechanical and fatigue life tests.
70
60
Average per lamina fibers

Average in two laminas of fibers
Average per lamina of epoxy

Average in two laminas of epoxy
50
40
Residual stress, MPa
30
20
10
0
112.5
225
337.5
450
562.5
675
787.5
900
10
20
Effect of Residual Stress on the

Mechanical Strength of Materials
Generality. When a part is subjected to a field
of elastic residual stresses characterized by a tensor rR, on which is superposed a field of service
stresses defined by the tensor rS, the real stress
to which the part is subjected is characterized by
the tensor rR rS (Fig. 6). If the residual
stresses are added to the service stresses (residual tensile stress, for example), the part is locally
overloaded due to residual stress. If, on the contrary, an appropriate finishing operation (shot
peening or roller burnishing, for example) is
used to introduce residual compressive stress,
the part is relieved of some of the load locally
and the mechanical performance of the materials
is increased as a result.
Figure 7 shows the properties of materials that
are influenced by residual stress. In the subsequent section, several quantitative examples of
the effect of residual stress are given.
Influence on the Fatigue Strength (Initiation Crack Phase). Residual stress plays an extremely important role with respect to the fatigue
strength of materials. It can be considered to be
a mean or static stress superimposed on the cyclic stress. As the mean stress rm increases, the
fatigue strength decreases. This is demonstrated
in the Haigh and Goodman diagrams.
30
y
40
50
Hole depth, m
Fig. 3
Residual stress distribution in the half-thickness of a carbon fiber/epoxy composite
S
R + S
I M
Mechanical components
manufacturing lines
Continuous control using

NDE techniques (US,
magnetic, AE)
Fig. 6
Superposing of residual stress and service stress
100
Good
quality
50
Defect
Assembly
of parts
Statistical analysis
of results
0
Measurement of the residual
stress distribution through
the thickness of the part
using conventional techniques
(x-ray, hole drilling)
0.10
0.20
0.30
Fatigue
0.40
50
100
Corrosion
under tensile
stress
150
200
Residual
stress
250
300
Breaking
srr L = 0.3 mm
szz L = 0.3 mm
Friction
Tensile
strength
350
400
Improvement of
process
Residual stress inspection plan for the purposes
of quality assurance. NDE, nondestructive evaluation; US, ultrasonic; AE, acoustic emission
Fig. 4
450
Depth, mm
Modeling of the residual stress distribution induced by shot peening using the 3-D finite element method
Fig. 5
Dimensional
stability
Fig. 7
Interfacial
bond strength
of coating
Effect of residual stress on the performance of

materials
leads to a tangential residual stress equal to or
slightly greater than the longitudinal stress. The
thickness of the material subjected to residual
compressive stress is in the same order of magnitude as the layer transformed during treatment
(Ref 15). Fatigue tests were carried out by the
French Technical Center for Mechanical Industry (CETIM) on 36 mm diameter XC42 steel cy-
Quenching treatment, after induction heating,

introduces very high residual compressive stress
into the hardened layer, which results from the
increase in volume of the martensitic structure
with respect to the ferrito-pearlitic structure (this
applied to the treatment of annealed steel, for
example). In induction-quenched cylindrical
bars, the residual stress on the surface usually
Table 2
Effect of quenching conditions and residual stress on fatigue strength
Type and depth of treatment

at 45 HRC, mm (in.)
Surface
hardness, HRC
Residual stress stabilized

at the fatigue limit, MPa (ksi)
Fatigue limit
after 5 106
cycles, MPa (ksi)
rm
ra
A, induction 2.7 (0.11)
5556
596 (87)
584 (85)
B, induction 4.2 (0.17)
5556
623 (90)
610 (88)
C, induction 4.7 (0.19)

D, water quenched after through
heating without stress-relieving
annealing 3.5 (0.14)
5459
6061
670 (97)
780 (113)
660 (96)
750 (109)
Longitudinal
stress
Transverse
stress
128 (19)
243 (35)
273 (40)
341 (49)
655 (95)
863 (125)
777 (113)
468 (68)
571 (83)
583 (85)
676 (98)
603 (87)
1132 (164)
1156 (168)
Source: Ref 16
Sample type
2.10, MPa
State
Crack
initiation site
BX, MPa
As welded
207
40
Post weld heat treated
207
37
Shot peened
392
519
Weld toe
1
Multipass
Fig. 8
Effect of residual stress on the fatigue strength of E690 welded joints. rRX, residual stress perpendicular to the
fillet (rRX); rRI, principal maximum residual stress evaluated at a depth of 0.1 mm below the surface. Source:
Ref 17
r rR rf*
106
ZAT D.T.
(40-75 mm)
ZAT B.S.
(40 mm)
Nontreated metal
da/dN, m/s
107
108
Rs = 0.1
E36Z
109
10
20
30
40
50
60
70
80
K, MPa m
Fig. 9
Ref 18
lindrical bars, quenched after induction heating,

and subjected to repeated bending stress (Ref
16). The results obtained are presented in Table
2. It can be seen that the higher the residual compressive stress, the greater the fatigue strength.
The resulting gain in fatigue strength produced
by the residual stress can be as much as 50% of
the fatigue strength of the base material treated.
Figure 8 shows the effect of residual stress on
the fatigue strength of welded HLE (E690) steel
joints (Ref 17). Three cases are show: as-welded
(residual tensile stress), stress relieved (no residual stress), and shot peened (residual compressive stress). A marked increase in the fatigue
strength was observed in the case of shot peening.
Influence on Fatigue Failure (Propagation
Phase) and Brittle Fracture (Ref 13, 18, 19).
In the case of welded assemblies, the presence
of welding defects at the weld toe and the geometric profile of the latter generally lead to a
limited period of crack initiation. The cracking
phase must be considered by taking into account
the residual stress field induced by the welding
operation.
The decisive influence of the residual stress
field on the crack propagation speed has been
demonstrated (Ref 18). Figure 9 shows the results of cracking as a function of the residual
stress. Relieving residual stress by heat treatment
changes the crack propagation speed considerably when the stress is high.
In the case of a brittle fracture, cleavage starts
in a grain when the local stress reaches a critical
value of rf* and it generally propagates without
difficulty in the adjacent grains by producing a
brittle fracture. The tensile residual stress rR, in
addition to the applied stress r, initiates this type
of failure for low loads, such that:
Effect of a residual stress-relieving treatment on the cracking speed in the HAZ (butt-welded assembly of an
E36Z steel). ZAT D.T., HAZ of the heat treated sample; ZAT D.T., HAZ of the as-welded sample. Source:
Once cracking has been initiated, the applied

stress alone can be enough to allow propagation
to continue at a high speed. Failure is therefore
very sudden. Residual stresses that facilitate the
initiation of brittle fracture by cleavage are therefore very dangerous for steels under load at low
temperature. This is why the stress relieving of
welded joints is also recommended.
Grain slips come up against inclusions and
create concentrated stresses at their interface that
lead to fracture of either the interface or the inclusion. Cavities then appear for a critical initiation stress and grow by plastic deformation of
the matrix until their coalescence leads to ductile
fracture at least on a microscopic level. The
speed at which the cavity grows is not only proportional to the plastic deformation speed but
also to the degree of triaxial state of the stresses
and to the ratio of the mean stress to the ultimate
stress. Coalescence is a plastic instability phenomenon that no doubt occurs for a critical cavity size. Tensile residual stress not only facilitates the initiation of cavities but, by increasing
the mean stress, also accelerates growth. These
two effects combine to decrease the critical elon-

gation of ductile fracture. However, this is only
important if the ductility is already very low in
the absence of residual stresses, since plastic deformation can eliminate them.
Effect on Stress Corrosion (Ref 20, 21).
Stress corrosion is a mechanical and chemical
cracking phenomenon that can lead to failure under the combined effect of tensile stress and a
corrosive environment. Cracking is generally
transcrystalline and can appear on all types of
materials, such as aluminum alloys, steels, copper, titanium and magnesium. The introduction
of residual compressive stress can considerably
increase the fatigue life of parts subjected to
stress corrosion. Tests carried out on magnesium
test specimens placed under stress in a salt solution gave the following results:
used, can remain in the coatings and in the substrates. They are of several types: microstresses
in the grain, produced during cooling, and
macrostresses affecting the entire coating.
Macrostresses are created not only by cooling
but also by the difference in temperature between the substrate, the sprayed layer, and the
outside surface. The differential contraction thus
produced between the various materials, due to
the difference in physical and mechanical properties, determines the stresses in the coating and
the coating-substrate interface. These stresses
therefore influence the mechanical and thermomechanical behavior of the coated parts.
In order to appreciate the quality of a coating,
three types of damage to parts in service can be
considered:
Ground test specimen: failure after two min-
The coating deteriorates rapidly.

The properties of the substrate are modified
utes
Shot peened test specimen: no cracking after
12 days under the same conditions
The tests conducted by W.H. Friske show that
the fatigue life is 1000 times greater for a shotpeened 304 grade stainless steel part than it is
for a non-shot-peened part (Ref 20). Tests carried out by CETIM on Z6CN18.9 stainless steel
produced similar results (Ref 21).
Effect on Adhesion of Coatings (Ref 22
25). Most coatings are produced for a specific
reason, particularly to improve the corrosion and
wear resistance of the base material, or to provide a thermal barrier for use at high temperature. But this is only achieved if the coating adheres to the substrate correctly. Adhesion
therefore indicates correct preparation of the surfaces to be coated and the quality of the coating
operation. The last few years have seen the appearance of plasma-spraying techniques, both at
atmospheric pressure and at reduced pressure.
These processes offer a high degree of flexibility
for coatings in critical areas. However, high residual stress, inherent to the coating method
R
Kc
7
6
300
200
4
3
100
2
1
0
Kc, MPa m
1
2
TT
STT
100
Influence of heat treatment on the residual

stress and the toughness of the interface: case
of plasma-sprayed coatings at atmospheric pressure. TT,
heat treated; STT, as-sprayed
Fig. 10
by the coating.
The damage is common to both materials. It
is located at the interface and jeopardizes both

the adhesion and the fatigue life.
C. Richard et al. have shown that decreasing
the residual stress by thermal treatment of the
coating considerably improves adhesion at the
interface (Ref 23). Figure 10 illustrates the effect
of residual stress. It can be seen that the apparent
toughness of the interface is improved by 100%
when heat treatment is applied. There is a high
level of residual tensile stress in the test specimen without heat treatment. When the level of
residual tensile stress increases, the true toughness of the coating decreases. An increase in the
residual compressive stress produces the opposite effect.
Influence of Residual Stress on the Tensile
Strength, Friction, Wear and Dimensional
Stability. The effect of residual stress on the tensile strength is obvious, particularly in structures
made of composite materials or when the prestressed layer is very thick compared with the
thickness of the parts. In composites, residual
stress is produced as a result of the thermal and
mechanical incompatibility of the reinforcements and matrix. This can influence the macroscopic properties of composites under tensile
or compressive stress (Ref 26).
Little research has been carried out on the effect of residual stress on friction and wear properties. Their role is often masked by other parameters. The increase of hardness during treatment
and changes in the toughness and adhesion of
antiwear coatings due to residual stress can considerably affect the resistance to friction. Up until the present, this effect has been integrated into
the global parameter of adhesion. In the future,
work will be carried out to try to determine the
real effect of residual stress.
The problem of dimensional stability has been
known for a long time. When a part is machined
that contains residual stress produced by heat
treatment or welding, the shape of the part can
change after operation due to the relaxation of
residual stress. This is why stress-relieving treatments are frequently used to avoid this type of
defect. Reference 27 gives a very methodical approach to defining the criteria and processes relating to relieving stress in welded structures.
The same type of reflection can be applied to
other types of parts.
Taking Residual Stress into Account

when Calculating the Expected
Fatigue Life
In the previous section, the different effects of
residual stress on the mechanical strength of
structures and materials were mentioned. Although the ability to quantitatively estimate the
fatigue life taking residual stress into account is
just beginning, it is still too early to extend these
predictions to other types of stress that are far
more complex and involve physical and chemical phenomena. Statistics show that failures of a
purely mechanical origin are mainly due to fatigue. It is for the reasons indicated previously
that this article only addresses problems concerning the prediction of fatigue life. Two articles (H. P. Lieurade and A. Pellissier-Tanon) in
Ref 13 deal with the question of predicting the
effect of residual stress on crack propagation
phase. Although, they concern welded structures, the concepts developed in these two references can be applied to other types of structures. By limiting the approach to prediction of
the fatigue life to the fatigue cracking initiation
stage, the problem of predicting the fatigue life
of mechanical components subjected to a high
cycle fatigue can be analyzed.
Calculating the Effect of Residual Stress on
the Fatigue Strength. Based on the experimental results mentioned previously, it would seem
that a linear relationship of the Goodman type
can be used to take residual stress into account:
ra rD
rD
(rm rR)
Rm
(Eq 1)
where ra is the amplitude of admissible stress,

rm is the mean fatigue stress, rD is the purely
reverse tensile fatigue limit, Rm is the true rupture strength, and rR is the residual stress measured in the direction of the applied service
stress. The numerous studies mentioned in Ref
28 show that the effect of residual stress is
greater when the properties of the materials are
high.
If we try to represent the development of ra
according to the residual stress rR by an equation
of the following type:
ra rD rR
(Eq 2)
the experimental results generally show that

increases with the strength of the material; for
example, in the case of machining stresses in an
XC38 grade steel, Syren et al. found:
0 in the annealed state
0.27 when quenched and tempered
0.4 when quenched
Unfortunately, these results are in contradiction with an equation of the Goodman type. In
Eq 2, the coefficient is none other than what
is usually called the endurance ratio:
rD
Rm
This parameter decreases as the rupture strength

of steels increases.
This apparent contradiction is probably explained by the fact that the residual stress relaxation phenomenon has not been taken into account. The value of the residual stress rR to be
introduced into equation of type 1 or 2 mentioned previously, must correspond to the stabilized fatigue stress, or the coefficient of influence
will include the relaxation process. The article
by D. Lohe and O. Voehinger in this Handbook
presents a large number of results concerning the
relaxation of residual stress under mechanical
and thermal loading. References 29 and 30 provide the information on the relaxation mechanism of residual fatigue stress as a function of
800
A
B Repeated
C bending tests
D
Repeated tension test
700
a, MPa
600
500
400
300
200
100
0
0
1000
m, MPa
2000
Use of Haigh diagrams to take longitudinal residual stress into account (XC42 steel
quenched after induction heating). A through D represent
results of repeated bending tests; X, repeated tension test.
Table 2 presents fatigue test results that correspond to A to
D.
Fig. 11
a, MPa
A
B
Repeated
C
bending tests
D
Repeated tension test
400
200
100
100
200
300
400
Pmax, MPa
Use of the Dang Van criterion to take residual
stress into account (XC42 grade steel quenched
after induction heating). A through D represent repeated
bending test; X, repeated tension test.
Fig. 12
salt f (A, B Pmhp, C Palt )
where Pmhp is the mean hydrostatic pressure, salt

is the amplitude of octahedral shearing or amplitude of the maximum shearing, and Palt is the
amplitude of hydrostatic pressure.
An example can be given as follows:
E
salt A B PD
mhp C Palt
300
200 100
the cyclic behavior of the materials. However,

Syrens results show that relaxation is much
greater when the mechanical properties are
lower.
When these experimental results are used with
the residual stress measured after carrying out a
fatigue test, and therefore stabilized, it is sometimes possible to use an equation such as Eq 2.
In the case of the fatigue bending test on cylindrical XC42 steel bars quenched after induction
heating (Table 2), the fatigue test results for the
different treatments correspond perfectly to the
Haigh diagram, provided any possible influence
of transverse residual stress on the fatigue stress
is ignored (Fig. 11).
It is not possible, however, to extend these
results to all materials and to the different manufacturing processes that introduce residual stress.
Also, preliminary tests are needed to validate the
methodology. The use of residual stress in calculations based on endurance diagrams of the
Haigh or Goodman type usually only allows for
an estimation of the increase in fatigue strength
as a function of the residual stress. Secondly, this
approach only allows for the combination of uniaxial stresses. Yet the residual stresses produced
by the various manufacturing methods used to
make the part are always multiaxial. The stresses
on the surface are biaxial while those inside the
part are triaxial. Depending on the area in which
the fatigue crack is initiated (on or below the
surface), the biaxial or triaxial stresses need to
be included when calculating the fatigue life.
This raises the problem of choosing a multiaxial
fatigue stress criterion. A simplified approach
based on an endurance diagram can therefore
only be an approximation.
The test carried out by the CETIM (Ref 16)
shows that the traditional Mises and Tresca criteria can only be used in the presence of higher
mean or residual stress. In this case, it is preferable to use criteria (Ref 31) that include the
amplitude of octahedral shearing (socta) or the
maximum shearing (sa) and maximum hydrostatic pressure (Pmax), as indicated here:
(Eq 3)
where A, B, C, D, and E are material constants.

If Dsalt is taken on the maximum shearing
plane, D E l and B C, the result is the
Dang Van criterion (Ref 32).
If Dsalt is taken on the octahedral shearing
plane, when D E 1 and B C, the Crossland criterion results (Ref 33); when D E
1 and C 0, the Sines criterion (Ref 34); and
when D E 1 and B C, the Kakuno criterion (Ref 35).
This type of development can be continued to
invent new criteria, but it leads to complications because of the increasing number of parameters that need to be determined. Even with a
linear relationship of the Dang Van type, two
Wohler curves have to be determined to obtain
at least the two points needed to produce the diagram. If additional constants are added, the test
plane will be even greater, which means that the
criterion cannot be used in industry. As a result,
the criterion to be used must be simplified as
much as possible. This case deals with radial
loading problems (r1 K1 r2 K2 r3), and a
relationship of Eq 3 is sufficient. To simplify
matters further, the Crossland or Dang Van criterion can be used. In the case of combined and
out-of-phase loading, new criteria have been developed to take the out-of-phase effect into account (Ref 3639).
But as yet, these criteria have not been validated in a study in which combined and out-ofphase residual stresses have been taken into account. When the fatigue stress is complex, it is
also very difficult to calculate the expected residual stress relaxation.
When fatigue cracks are initiated on the surface, the stresses to be taken into account are
biaxial; this gives the following for the Crossland or Dang Van criterion:
socta
sa
Pmax
2
3
r 1a r 2a r r
2
1a 2a
r1a
2
1
(r1a r2a r1m r2m r1R r2R)
3
where r1a, r2a represent the amplitude of the

main reversed fatigue stresses (r1a r2a); r1m,
r2m represent the average value of the main fatigue stresses; and r1R, r2R are the residual stress
values measured in the two main directions (stabilized values).
To use the multiaxial fatigue criteria, the reference curve for the material being considered is
needed, just as it is when using the Goodman or
Haigh diagram. Reference 16 shows that the use
of Crossland or Dang Van criteria takes the increase in the bending fatigue strength into account perfectly as a function of the residual
stress introduced by the various treatment conditions (Fig. 12).
When the multiaxial aspect is brought into the
picture, the method that consists in introducing
residual stress into the calculation in the same
way as a mean stress, therefore, seems to give
satisfaction. The whole problem lies in defining
the residual stresses to be included in the calculation.
Taking residual stress into account is essential
for correct prediction of the fatigue limit. Figure
13 shows the important role played by compressive residual stress. If it is not taken into account,
the fatigue strength is underestimated (Fig. 13a).
If the residual stress measured or calculated is
used without taking relaxation of the residual
stress into account, the fatigue strength is over-

estimated (Fig. 13b). The correct method consists in calculating the fatigue strength after taking relaxation into account (Fig. 13c).
In order to correctly evaluate the effect of residual stress, various problems must be solved:
Measuring methods must be available to de-
a, MPa
a, MPa
(a)
a, MPa
Pmax
Fig. 13
termine residual stress in the critical zone. A

large range of measuring methods currently
exists and it is possible to take measurements
(Ref 1) in most of the cases studied, particularly as a result of development of the x-ray
method and the incremental hole method.
When a complete measurement profile is
Relaxation
des R
Pmax
Initial
(b)
Pmax
(c)
Illustration of the different methods used to take residual stress into account
a, MPa (Surface)
Xian
CETIM
300
200
100
100
200
300
400
500
Pmax, MPa (Surface)
Fig. 14
Fatigue results for a 45SCD6 steel treated under different conditions in a Dang Van diagram, taking the residual
stress on the surface into account. Xian, results obtained by Xian Jiaotong University; CETIM, results obtained
by CETIM
a, MPa
available of the residual stress in the surface

layer in which the fatigue crack is initiated,
the profile sometimes has high stress gradients. It is then difficult to know what stress
value to choosethe surface stress or the
stress peak that is often slightly below the surface (in the case of shot peening, for example).
To make a correct calculation, it would be
necessary to use calculation methods that take
the stress gradient into account and make the calculation not only for a single point, but for a
sufficient thickness of the material (thickness of
critical layer) for it to be representative of the
basic volume in which the fatigue damage process occurred (Ref 31). Figures 14 and 15 give
an example of processing of our results (details
can be found in Ref 29) in the case of the fatigue
of a shot-peened spring steel. Figure 14 shows
the fatigue results on a Dang Van diagram, taking both the residual stress and its relaxation into
account. A fairly good correlation can be observed. This indicates that a multiaxial fatigue
criterion taking the hydrostatic pressure into account can be used to predict the fatigue strength
in the presence of residual stress. Since, in this
case, the crack initiation zones are below the surface, calculations were made for different critical
layer depths. Figure 15 shows the results obtained for a critical layer depth of 100 lm. Better
alignment of the experimental points was observed. This example illustrates the possibility
of improving the calculation precision by using
the critical layer thickness approach. It is particularly relevant in the case of notched parts.
It has long been known that residual stresses
are not stable when they are subjected to fatigue
loading. To calculate the expected fatigue life,
precise information is therefore needed in order
to introduce stable residual stress values into the
calculation presented previously, that is, the
stress values that are really likely to be present
in the part during the best part of its lifetime.
The stress must therefore be measured on a
part already under cyclic loading or relaxation of
the residual stress estimated according to expe-
600
500
350
With
Without
400
A Consideration
600 MPa
200
a, MPa
300
Xian
CETIM
300
550 MPa
2 106
5 106
100
250
300
200
100
100
200
300
400
500
Fatigue results for a 45SCD6 steel treated under different conditions in a Dang Van diagram, taking the residual
stress into account using the critical layer approach. Xian, results obtained by Xian Jiaotong University;
CETIM, results obtained by CETIM
50
100
150
200
Pmax, MPa
Pmax, MPa
Fig. 15
107
240
Prediction of the fatigue life using the Dang

Van criterion for a shot-peened 35NCD16
steel, without taking residual stress into account
Fig. 16
rience or modeling. In Ref 29 and 30, a complete
model was presented using FEM to determine
the stabilized residual stress after fatigue loading. This estimation of the residual stress can
then be used to calculate the fatigue life of a part,
taking residual stress into account. Despite the
initial definition of the Dang Van criterion,
which proposes that it only be used in cases of
fatigue strength with an unlimited number of cycles, an attempt was made to extend this criterion
to include a limited fatigue life with a very large
number of cycles (more than 2 106) (Ref 40).
Figure 16 shows an example of the fatigue life
estimated by calculation. First, a fatigue strength
diagram of the Dang Van type was defined according to the fatigue life obtained from a series
of fatigue life contours. Then the residual stress
relaxation was calculated using FEM (Ref 29).
Finally, the stabilized residual stress was introduced into the diagram. In this example, in the
case of a loading of 550 MPa, the point corresponding to loading is inside the limit of the
fatigue life at 107 cycles. No failure occurs. For
a loading of 600 MPa, the point corresponding
to loading including the residual stress is between the line corresponding to 5 106 cycles
and that of 2 106 cycles. Failure therefore
alt
2 Hardening effect
3 Roughness increase
1
3
Pmax
Fig. 17
Illustration of the effect of the surface finish and

strain hardening on the fatigue strength
Role of hardening
and residual stress
Fatigue life increasing, %
20
Role of
residual stress
10
0
Role of superficial hardening
10
20
30
40
50
HRC
Effect of the resistance of the base metal on the
increase in the fatigue strength after shotpeening treatment, distinguishing between the effect of
strain hardening and that of residual stress
Fig. 18
A new roughness that changes the local stress
concentration
Additional strain hardening of the surface due
to plasticizing
A new metallurgical structure of the surface
layers
Figure 17 shows the effect of the surface finish
and strain hardening on the fatigue strength of
materials. It can be seen that an increase in the
roughness decreases the safety area, and strain
hardening increases the safety area, provided it
does not damage the material.
In the case of thermal or thermochemical surface treatments (induction quenching, case hardening, etc.), for example, it is necessary to take
the new fatigue strength of the treated layer into
account in the calculation.
The problem is more complex in the case of
residual stress introduced by plastic deformation
(pre-straining, machining, shot peening, roller
burnishing), since it is more difficult to distinguish between the influence of residual stresses
and residual microstresses present in the grains
of the deformed material and that of strain hardening of the material.
Evans (Ref 41) made this distinction in the
case of shot peening; to do so, he carried out
three types of fatigue tests on materials with
various mechanical properties:
Fatigue tests on a non-shot-peened material
Fatigue tests on a shot-peened material
Fatigue tests on a shot-peened material, but
with a mean test stress rm, which compensated for the surface residual stress. In this
type of test, the effect of the macroscopic residual stress is cancelled out and the fatigue
strength obtained only depends on the increase in the mechanical properties of the
plastically deformed material and the residual
microstresses distributed throughout the material. The results obtained are presented in
Fig. 18.
40
30
occurs between 2 and 5 million cycles. The tests

give an average fatigue life for the previously
mentioned loading in the order of 3.5 million
cycles. This example shows that, if the cyclic
properties of the material are correctly known, it
is possible to predict the fatigue strength of the
material in the presence of residual stress.
However, it should not be forgotten that other
factors must also be taken into account in calculating the fatigue lifethe introduction of residual stress is often accompanied by other
changes in parameters that have an influence on
the fatigue strength. In particular, these include:
It can be observed that, for materials with low

resistance, the increase in the fatigue strength is
mainly due to surface strain hardening. On the
other hand, for highly resistant materials, it is
mainly the influence of the residual stress that
governs the fatigue strength. When materials
have low elastic limits, the stresses introduced
by shot peening relax much more easily than
they do when the elastic limit is high. This test
only shows a general tendency and does not ex-
actly show what the author is trying to demonstrate, since in the two types of tests for the same
material, the ratio of R(rmin /rmax) is modified.
As has been shown, the residual stress relaxation
changes when the R ratio is varied (Ref 29). The
real contribution of each factor will therefore be
different from that indicated in Fig. 18.
It can thus be seen that taking residual stress
into account in the calculation requires a serious
examination of the different parameters involved. When reliable results are needed, fatigue
tests will no doubt need to be carried out on the
part or the structure concerned. However, modeling enables the variation in the different parameters to be rapidly simulated in order to find
an optimal solution.
Partial Summary. The previous results show
that it is now possible to take residual stress into
account in calculations designed to predict the
fatigue life using a global approach. This must
take the relaxation of residual fatigue stress into
account, as well as the other effects (strain hardening, hardness) introduced by the manufacturing method used. A multiaxial fatigue criterion
that can integrate both the problem of residual
stress and the effect of the stress gradient applied
to a zone in the presence of stress concentration
has been developed, that is, the Crossland or
Dang Van criterion. It is used for a stabilized
state of residual stress, averaged out for a basic
volume of damage (thickness of critical layer),
and applied within a network of contours that
represents the fatigue life. In the future, tests will
be carried out to validate this type of criterion in
the case of combined stresses on notched parts
in the presence of residual stress.
Incorporating the Notion of

Residual Stress into the Design
Office
Incorporating the notion of residual stress into
the design office must be gradual and can be divided up into several phases.
Today, very few industrial sectors consider
the residual stress parameter directly. In technical specifications, requirements are included that
are often closely related to residual stress without actually naming it. An Almen intensity must
be guaranteed in the case of shot peening, for
example, a roller-burnishing load, a machining
procedure or a minimum treated thickness in the
case of thermal or thermochemical treatment,
and a maximum dimensioning tolerance in the
case of a machined or welded part.
In the first phase of incorporation, a semiquantitative notion can be used to evaluate the
increase in performance in terms of fatigue life
or fatigue strength. A few examples can be presented. Table 3 gives an example of the effectiveness of shot peening in increasing the fatigue
life of different types of mechanical parts, and
Fig. 19 shows the beneficial role played by roller
burnishing on the fatigue strength of spherical
graphite cast iron crankshafts. Figure 20 shows
of residual stresses is realized in the first series of fatigue cycles.
The cyclic behaviors of material are very im-
Table 3 Increase in the fatigue life of various mechanical components as a result of shot
peening
Type of part
Type of stress
Spindles
Shafts
Gear box
Crankshafts
Aircraft coupling rods
Driving rods
Cam springs
Helical springs
Torque rods
Universal joint shaft
Gear wheel
Tank chain
Weld
Valve
Rocker arm
Increase in the fatigue life (%)
Reverse bending
Torsional
Fatigue life tests in service
Fatigue life tests in service
Tensile compression
Tensile compression
Dynamic stress
Fatigue life in service
Dynamic stress
Reverse bending
Fatigue life tests
Fatigue life tests
Fatigue life tests
Fatigue life tests
Fatigue life tests
4001900
700
80
3000, but highly variable
105
45
100340
3500
140600
350
130
1100
200
700
320
250
200
Alternated
stress
150
0
100
0
50
Repeated
stress
0
70
S-
G
70
S-
G
70
S-
G
40
S-
G
40
70
S-
S-
llin
g
en
ng
in
ro
en
io
ce
rfa
su
ct
di
at
re
tri
pe
fe
ot
du
in
sh
ni
tt
ou
ni
ni
te
io
ith
By using the cyclic behaviors of material, a

simplified method to calculate residual stress relaxation has been proposed in the first section.
Secondly, a method to predict fatigue life by taking the stabilized residual stresses into account
is presented. This design tool is based on the
FEM. It has been applied to shot-peened
35NCD16 grade steel. The different fatigue parameters often used in material research are studied. On the other hand, an experimental investigation about this material had been done by
Bignonnet (Ref 42). The results of the study
show that this design tool on fatigue developed
by the Laboratory of Mechanical Systems Engineering (LASMIS) (Ref 43) is able to take into
account different loading parameters.
The residual stresses, however, can be relaxed
by the deliberate application of thermal or mechanical energy. They will be especially relaxed
when the structure is subject to cyclic loading.
The relaxation phenomenon depends on a complex interaction of a number of factors, such as
the applied stress amplitude, the number of cycles of loading, the state of initial residual
stresses, and the nature, origin, and mechanical
properties of the material. In this article, only
relaxation during cyclic loading and the influence of the stable residual stresses on the fatigue
life are studied.
In only a few cases are the residual stresses
systematically analyzed using measurement of
the residual stress state during and after fatigue
testing. This is usually a difficult, time-consuming task. Several fatigue tests under tensile and
torsion loading with different stress amplitudes
have been done.
A numerical method for the prediction of the
residual stress distribution during and after fatigue has been developed. Finite element software is used for incorporating cyclic plasticity
into the calculation. A simplified method to calculate stabilized residual stresses was proposed.
stresses relax with the increase of the number

of cycles up to a stabilized state of cyclic
properties.
For a cyclic hardening material, the relaxation
Maximum admissible stress, MPa
Example of Integrated Design Tool
For a cyclic softening material, the residual
According to a mechanical approach, we can

make the following general predictions (Ref 30):
Fig. 19
Effectiveness of roller burnishing in increasing the fatigue strength
Thermal treatment by electron beam, R = 0.1

TT by CO2 laser, R = 0.1
TT by CO2 laser, R = 1
TT by laser shock, R = 0.1
Treatment or process
a horizontal comparison of gains to be expected

in terms of fatigue strength from various surface
treatments. The results presented here are not at
all exhaustive and are taken from a limited bibliography. However, this figure should not be
taken as a reference, since the geometry of the
test specimens differs for each type of treatment.
In certain cases, this parameter can have an important effect on the gain achieved. Each industrial sector must carry out this type of comparison for the treatments and materials used in order
to help engineers design their products more effectively.
The second phase consists of predicting the
fatigue life using the notions developed in the
previous paragraph.
The third phase is the development of integrated tools for taking the residual stress into
account. The following paragraph presents an
example of such a design tool on fatigue for
three-dimensional components.
ST by nitriding R = 0.1
TT by induction R = 0.1
TS carburizing R = 0.1
ST by surface rolling R = 0.1
200%
ST by surface rolling R = 0.1
ST by shot peening R = 0.1
0
Fig. 20
25
50
Increase, %
75
Beneficial effect of various surface treatments on the fatigue strength (maximum gain reported in the literature).
TT, thermal treatment; ST, surface treatment
portant in the prediction of residual stress relaxation. They make it possible to calculate the residual stress state for a corresponded number of
cycles.
Then, a method to predict fatigue life while
taking these stabilized residual stresses into account is presented. The fatigue criterion under
multiaxial stresses is used. After the relaxation,
the residual stresses can be combined directly
with the cyclic loading. They change the value
of the mean stresses and influence the fatigue
life.
As a result, a design tool on fatigue for threedimensional components, FATIGUE3D, has
been developed in the laboratory of LASMIS
Calculation of
residual stresses
Outside solicitation,
cyclic loading
(Ref 43). It can take the residual stresses into

account when calculating the fatigue life of a
structure. It is based on the FEM. With this tool,
an iso-colored image of distribution of fatigue
life or safety factor can be obtained.
These methods mentioned above are applied
to shot-peened 35NCD16 grade steel. The different fatigue parameters often used in material
research are studied. A comparison with the experimental results has been carried out (Ref 43).
Basic Steps for the Calculation. Figure 21
shows the elementary steps of the calculation of
fatigue life while taking into account the residual
stresses.
First, the initial state of the residual stresses
Introduction of initial field

of residual stresses
Measurement of
residual stresses
Prediction of
residual stress
Cyclic behaviors
of material
Fatigue, 3D
Mechanical properties
of material
Distribution of fatigue life,

safety factor, and
admissible residual stress
Fig. 21
Basic steps of calculation while taking into account the residual stresses
N1
Example and Results

N1
Fig. 22
Principle of calculating fatigue life
s = OM/OM
M
N-cycles
M
Fig. 23
can be obtained by the measurement method or

by the simulation of surface treatment. Then, this
initial field of residual stresses can be introduced
by FEM using ABAQUS (Ref 44, 45). Furthermore, a stabilized field of residual stresses can
be predicted after an elastic-plastic calculation,
in which the cyclic behaviors of material are
used. This stabilized field of residual stresses after relaxation must be taken into account when
calculating the fatigue life. At last, with the help
of a design tool on fatigue, FATIGUE3D, the
distribution of fatigue life and that of the safety
factor on fatigue can be obtained.
Principle and Basic Function. According to
the criteria of fatigue under multiaxial loading,
the fatigue life changes along with two parameters. One of them is maximum static pressure;
another is equivalent octahedral stress. Therefore, two stress-number of cycles (S-N) curves
are needed under simple cases (for example,
bending or torsion). With these two curves, a line
can be determined for a certain number of cycles
(life) in the stress plane P s (Fig. 22).
A different fatigue life is represented by a different line. If there is a complex stress state, it
gives a point in the plane P s. Respectively,
the line to which this point is nearest represents
the fatigue life of that stress state. In this way,
the fatigue life for any complex stress state and
for any part of a component can be predicted. A
distribution of the fatigue life is helpful for the
designer.
On the other hand, the safety factor on fatigue
is another interesting parameter for the designer.
FATIGUE3D provides a distribution of the
safety factor. In Fig. 23, N-cycles are the designed life and M is the point given by the stress
state. The safety factor on fatigue is defined as S
OM/OM.
P
Principle of calculating the safety factor on fatigue
Fig. 24
Three-dimensional mesh
Material and Geometry. The shot-peening

process is a widely used technique because it can
produce a field of residual stress on the surface
of a part. These stresses are compressive and will
improve the properties on fatigue. As an example, a shot-peened part made of 35NCD16 grade
steel (0.35% carbon, 3.7% chromium, 0.3% molybdenum and having tensile properties of: a
yield strength of 1000 MPa, an ultimate tensile
strength of 1100 MPa, and an elongation of
17.5%) is studied. Many results of fatigue experiments are available regarding tension compression and torsion loading (Ref 44).
The geometry of the part is defined from the
real geometry so that the testing results can be
compared. It is similar to the fatigue test sample,
only the numerical values (radius, length) vary
for tension compression or torsion loading. The
difficulty is to define a thin mesh near the surface
(0.20.4 mm, or 0.0080.016 in.) where the residual stresses are introduced while maintaining
a normal mesh in other zones. Figure 24 shows
the three-dimensional mesh of the part.
Introduction of the Initial Field of Residual

100
100
Axial stresses
under tensile loading
100
Measurement
Calculated
200
Tangential stresses
under tensile loading
0
Residual stresses, MPa
300
400
500
Measurement
Calculated
100
200
300
400
500
600
600
0.05
0.10
0.15
0.20
0.25
0.30
0.05
0.10
0.20
0.25
0.30
0.4
0.5
0.6
100
100
Axial stresses
under torsion loading
100
0
0
0.15
Depth, mm
Depth, mm
Measurement
Calculated
200
300
400
500
600
Tangential stresses
under torsion loading
Measurement
Calculated
100
200
300
400
500
600
700
0
0.1
0.2
0.3
0.4
0.5
0.6
0.1
Depth, mm
Fig. 25
0.3
Depth, mm
Initial distribution of the residual stresses
Stress. The initial residual stress distribution is

calculated from the residual stresses measured
by the x-ray diffraction method in the upper
layer where initial stresses are introduced by surface treatment. In depth, the stresses are calculated for the structural equilibrium. Figure 25
shows the comparison of initial residual stresses
between the calculation and the experiment.
Prediction of the Residual Stress Relaxation. In order to predict the relaxation of the residual stresses, a simplified method proposed by
Lu et al. has been used (Ref 30). It supposes that
the relaxation of the residual stresses depends on
elastic-plastic properties of material, because the

plastic deformation is the main cause of the relaxation. So, the cyclic behaviors of material are
very important for the prediction of the residual
stresses. However, they can be measured by experiment.
Figure 26 shows the cyclic behaviors of
35NCD16 grade steel, a cyclic softening material.
When the stabilized residual stresses are calculated, the behavior corresponding to NR /2 is
used. After an elastic-plastic calculation with
FEM, the relaxation of the residual stresses un-
Rec(MPa) = 920 55,14 log N

1 cycle
2
3
15
50
150
/2, MPa
1000
900
NR/2
800
700
600
35 NCD 16
(Rm (UTS) = 110 MPa)
0.5
0.75
N,
cycles
1
2
3
15
50
150
NR/2
Re(Y.S.),
MPa
940
920
910
870
830
800
660
t/2, %
Fig. 26
0.2
The cyclic behavior of 35NCD16 grade steel. Re, yield stress; R ec, cyclic yield stress as a function of number
of cycles; Det, axial plastic strain range during cyclic loading
der the cyclic loading can be obtained. Part of

the results and their comparison with the test values have been shown in Fig. 27. In the case of
the traction-compression loading, two loading
levels were calculated. The same level for torsion loading with two different cycle numbers
was also analyzed. Figure 28 shows the stabilized residual stresses for different traction-compression load levels that are used in the prediction of the fatigue life. Unfortunately, it is
impossible to compare with the experiment results.
Prediction of the Fatigue Life. For predicting the fatigue life under a complex stress state,
it is necessary to have two S-N curves. One is
for simple traction or traction compression; another is for alternated torsion. These curves are
basic data of a material, and they can be obtained
by experimental method.
Using the program FATIGUE3D, a design
tool on fatigue developed in the LASMIS laboratory, a distribution of fatigue life and a distribution of safety factor for a structure can be obtained. In this example, the fatigue life under
traction-compression cyclic loading and torsion
loading has been calculated. The stabilized residual stresses play a role as static load. Figure
29 shows the results of fatigue life and their comparison with the experimental results. It is very
clear that the proposed method is available.
Prediction of Admissible Residual Stress.
In this approach, a calculation method has been
developed that can predict the admissible residual stress for a given fatigue life. This tool can
100
100
103 cycles, traction compression 550 MPa
100
102 cycles, traction compression 700 MPa
Axial measurement
Axial calculated
Tangential measurement
Tangential calculated
200
300
400
Axial measurement
Axial calculated
100
200
300
400
500
500
0
0.10
0.05
0.15
0.20
0.25
0.30
0.05
0.10
0.15
Depth, mm
105 cycles, torsion 380 MPa
100
200
300
400
Axial measurement
Axial calculated
0
100
200
300
400
500
600
500
0
0.1
0.2
0.3
0.4
0.5
0.1
Fig. 27
life. In other zones, tensile residual stress is admissible.

The fatigue life can then be used to deduce a
stabilized residual stress. An example of defining
the quenching conditions according to the residual stress field obtained before relaxation is analyzed.
An inverse technique designed to obtain the
quenching conditions is then developed (Ref 46,
47). In this technique, it is necessary to simulate
quenching using the FEM. To simulate quench-
100
Tangential residual stress, MPa
100
Traction-compression
550 MPa
600 MPa
620 MPa
650 MPa
700 MPa
200
300
Traction-compression
550 MPa
600 MPa
620 MPa
650 MPa
700 MPa
0
100
200
300
400
500
0.05
0.10
0.15
Depth, mm
Fig. 28
0.3
0.4
0.5
The relaxation results of the residual stresses
be used during the design phase to evaluate

whether or not residual compressive stress needs
to be introduced. The computer code also indicates the level of the residual stress and the zone
in which it needs to be introduced. In this way,
the treatment condition during the mechanical
design phase can be defined. Figure 30 shows a
map of the prestress zone indicated for a notch
sample. It can be seen that in the high stress concentration area, residual compressive stress must
be introduced to obtain a predetermined fatigue
0.2
Depth, mm
Depth, mm
Axial residual stresses, MPa
0.30
102 cycles, torsion 380 MPa

100
Axial measurement
Axial calculated
100
0.25
200
100
0.20
Depth, mm
0.20
0.25
0.30
0.05
0.10
0.15
Depth, mm
The stabilized residual stresses under different traction-compression loading cases
0.20
0.25
0.30
ing under different conditions, the physical and

mechanical data of the material need to be
known. Generally speaking, these parameters
depend on the temperature, but the surface heattransfer coefficient depends not only on the temperature, but also the sample geometry, the quenchant, and the quenching temperature. In order
to solve the problem, it is considered that there
is no change in any of the quenching parameters
or the sample geometry. Only the quenching
temperature changes. Quenching at different
temperatures was carried out. This gives the variation of the surface heat-transfer coefficient as a
function of the temperature for each quenching
temperature. It should be mentioned that numerical methods alone are not sufficient to obtain
the heat-transfer coefficient. The temperature
must be measured during quenching. Since the
heat-transfer-coefficient curve is known, the parameters of these curves as a function of the
quenching temperature can be defined. Once this
coefficient is known, quenching can be modeled
at different temperatures. The quenching temperature in the case of water quenching varies
between 20 and 80 C (68 and 176 F). The lower
the quenching temperature, the higher the level
of the residual stress field introduced. Initially,
the residual stress field obtained was found to be
lower than at 20 C (68 F) and higher than at
750
Without residual stresses
Experimental
Numerical
700
650
600
550
500

Experimental
Numerical
420
400
380
360
340
320
450
105
Fig. 29
440
Torsion load, MPa
Traction-compressionn load, MPa
106
107
Fatigue life N cycles
300
105
106
107
Fatigue life N cycles
Summary
Results of fatigue life
S11
Value
2.24E+02
4.24E+01
+1.39E+02
+3.20E+02
+5.01E+02
+6.82E+02
1
3
2
Fig. 30
Prediction of admissible residual stress (or the prestressed area) for a notched sample under bending loading,
using the finite element method
Mechanical component
development
Introduction of a surface
and prestress treatment
Optimization of geometry
according to the choice
of manufacturing process
No
Stress analysis by
a finite element method
Adjustment of a
finite element model
Is fatigue life
in accordance
with
specifications?
Yes
Stop
Experimental checking of
stress: rapid prototyping
No
Fig. 31
Are results
coherent?
Yes
80 C (176 F). After this condition has been

verified, the quenching temperature can be determined by varying the quenching temperature
during simulation. After each simulation, the
calculated residual stress field is compared to
that obtained from the fatigue life. A limit condition can be defined to improve the precision of
the results. When comparing the residual stress
fields, the hydrostatic pressure was used at each
point.
Fatigue life
estimation
Various connections between residual stress-integrated design and other sectors that use the concurrent engineering approach
Residual stress plays a very important role

with respect to the different properties of materials. The gain obtained from the presence of residual stress can be enormous. This article attempts to show the effects of residual stress
through the example of fatigue strength. Here, it
has been shown that it is now possible to predict
the fatigue strength of materials, taking residual
stress into account. The results of this study
show that it is possible to predict the residual
stress relaxation and fatigue life, with consideration of the influence of residual stress by the
FEM. It has been found that the calculated results of fatigue life at surface agree very well
with experimental results.
Although the author is not in a position to provide the same type of calculation tools for other
properties, such as corrosion resistance and the
adhesion of coatings, it is now reasonable to expect that the notion of residual stress will be
gradually introduced into the design stage of mechanical parts. Numerical modeling of the behavior beforehand saves a considerable amount
of time because of the reduction in the number
of experimental tests required. These tests are
often very long and costly, but they have proved
to be indispensable. The problem of taking residual stress into account at the design stage will
become more and more critical with the development of new materials (multimaterials, etc.)
and new treatments (combined treatments, etc.)
With the development of different experimental and numerical techniques, it is now possible
to introduce residual stress into the design office
for the integrated design of mechanical components, thus offering a new concurrent engineering approach applied to the design of mechanical
components taking residual and applied stress
into consideration. Figure 31 shows the different
connections between residual stress-integrated
design and other sectors that use the concurrent
engineering approach. A mechanical component
designer can simulate dynamic characteristics,
material processing, and product life.
Consideration of residual stress is becoming
increasingly important for two reasons: the introduction of multimaterials that induce residual
stress, and the need for the designer to reduce
the weight of components in order to remain
competitive. Basic research has brought a better
understanding of the phenomena relating to residual stress.
The main aim is to develop an integrated qual-
ity control tool. For industrial applications, future developments are necessary:
Measurement techniques
grant Brite-EuRam, BRRT-CT985090, ENSPED project), and the National Science Foundation of China (two bases project) is acknowledged.
Improvement of ultrasonic and magnetic
measurement methods
Integration of the portable optical method for
strain measurement in destructive techniques
Processing and materials
Development for the industrialization of new
prestress processes: ultrasonic shot peening,

laser shock, surface nanocrystallization, and
so on.
Optimization of residual stress in advanced
materials: functional gradient coating system,
for example, plasma spray coating, PVD,
CVD, or diamond-like carbon (DLC) coating;
metal-matrix composite; electronic packaging; intermetallic materials
Modeling
Development of advanced tools for the pre-
diction of initial residual stress in various processes: forging, casting, machining, welding,
plasma spray coating, composite processing,
electronic packaging manufacturing, mechanical surface treatment
Integration of the residual stress modeling
method into a global life design (3-D) and the
creation of optimization codes for the design
of mechanical components using the concurrent engineering approach
With the recent progress in the field of research and development, it is now possible to
introduce a new area of research: prestress engineering of structural materials (PESM or
PRESMA). In the near future, the mechanical
engineer will be able to use these tools in a prestressed concrete approach to civil engineering.
The ENSPED program in Europe is an excellent
example of the cooperation of industrial and research partners. Using this new approach, the fatigue behavior and mechanical behavior of new
materials can be considerably improved, increasing two- or threefold in some cases. In fact, many
failures occur in the area close to the surface.
Improving the mechanical behavior is often only
necessary on the surface (fatigue, fretting fatigue, wear, stress corrosion). As a result, this
type of approach can be expected to develop
very quickly in order to introduce adequate prestressing treatment in the area indicated by the
advanced design tool.
ACKNOWLEDGMENTS
The author is grateful to all the people at LASMIS working in the field of residual stress for
their help during the preparation of this paper:
Prof. J.L. Chaboche, Dr. D. Retraint, Dr. E. Rouhaud, Dr. X.L. Gong, Dr. L. Couturier, Dr. R.
Akrache, Dr. Z. Wu, Dr. S. Rasouli Yazdi, Mr.
A. Milley, Mr. D. Deslaf, Mr. B. Guelorget, Mr.
A. Voinier, Mr. F. Belahcene, Mr. K. Gong. Financial support from the European Union (under
12.
REFERENCES
1. J. Lu, Ed., Handbook of Measurement of Residual Stresses, Society for Experimental
Mechanics, USA, 1996
2. J. Lu, A. Niku-Lari, and J.F. Flavenot, Latest Development of Residual Stresses Measurement by the Hole Drilling Method, Materiaux et Techniques, Dec 1985, p 709718
3. Z. Wu, J. Lu, and B. Han, Study of Residual
Stress Distribution by a Combined Method
of Moire Interferometry and Incremental
Hole Drilling. Part I: Theory, J. Appl.
Mech., ASME, Sept 1998
4. Z. Wu, J. Lu, and B. Han, Study of Residual
Stress Distribution by a Combined Method
of Moire Interferometry and Incremental
Hole Drilling. Part II: Implementation of
Applied Mechanics, J. Appl. Mech., ASME,
Sept 1998
5. Z. Wu and J. Lu, Residual Stresses by Moire
Interferometry and Incremental Hole Drilling, Experimental Mechanics, Advances in
Design, Testing and Analysis, Vol 2, I.M.
Allison, Ed., Conf. ICEM 98, 2428 Aug
1998, (Oxford), p 13191324
6. Z. Wu, X.K. Niu, J. Lu, and P.G. Ifju, Study
of Process-Induced Stress in Orthotropic
Composite LaminateCarbon/Epoxy [O2 /
902]2s, Proceedings of the SEM Spring
Conference on Experimental and Applied
Mechanics, 13 June 1998, (Houston,
Texas), p 179181
7. Z. Wu, J. Lu, and Y. Guo, A New Method
of Residual Stress Measurement on Electronic Packaging, Experimental Mechanics,
Advances in Design, Testing and Analysis,
Vol 2, I.M. Allison, Ed., Conf. ICEM 98,
2428 Aug 1998, (Oxford), p 987991
8. J. Lu, J.F. Flavenot, and S. Thery, Study on
the Effect of the Finishing Treatment on the
Residual Stress Gradient in SiC Reinforced
Aluminum Metal Matrix Composite, J.
Compos. Technol. Res., ASTM, Vol 12 (No.
4), 1990, p 232238
9. J. Li, J. Lu, M. Perrin, M. Ceretti, and A.
Lodini, Study on the Residual Stress in
Cold-Rolled 7075 Al-SiC Whisker Reinforced Composites by X-ray and Neutron
Diffraction, J. Compos. Technol. Res., July
1995, p 194198
10. J. Lu, P. Peyre, C. Omam Nonga, A. Benamar, and J.F. Flavenot, Mechanical Surface
Treatments, Current Trends, and Future
Prospects, Surface Modification Technologies VIII, T. S. Sudarshan and M. Jeandin,
Ed., TMS, Sept 1994, p 589601
11. A. Benamar, J. Lu, J.F. Flavenot, P. Barbarin, G. Chalant, and G. Inglebert, Modeling Residual Stresses by Cold Rolling an
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
Austenitic Stainless Steel, Residual Stress,

V. Hauk, H.P. Hougardy, E. Macherauch,
and H.D. Tietz, Ed., DGM, Verlag, 1992, p
891900
J.B. Roelens, F. Maltrud, and J. Lu, Determination of Residual Stresses in Submerged
Arc Multi-Pass Welds by Means of Numerical Simulation and Comparison with Experimental Measurements, Welding in the
World, Vol 33 (No. 3), 1994, p 152159
Proceedings of the National Conference on
the Residual Stress and Design Office, CETIM, (Senlis), 1991
D. Deslaef, E. Rouhaud, and J. Lu, Finite
Element Modeling of Shot-Peening Processes, Proceedings of the SEM Spring Conference on Experimental and Applied Mechanics, 13 June 1998, (Houston, Texas),
p 419422
A. Chabenat, and J.F. Flavenot, Measurement of Residual Stress on the Quenched
Axles after Heating by Induction, Traitement thermique/Revue de metallurgie,
(Heat treatment/Journal of Metallurgy), No.
124, 1978, p 233239 (in French)
N. Skalli, and J.F. Flavenot, Prediction of
Fatigue Life by Simulation with the Residual Stress Considerations, Conference proceedings, Spring Conference, SF2M
(French Society for Metal and Materials),
2223 May 1984, p 98117 (in French)
H.P. Lieurade, P. Castelucci, J.F. Flavenot,
and J. Lu, Efficiency of Improvement Techniques on the Fatigue Strength as a Function
of the Type of Welded Joint, Welding in the
World, Vol 31 (No. 4), 1991, p 268271 (in
French)
H.P. Lieurade, C. Maillard-Salin, and M.
Truchon, Fatigue Cracking of Welded Joints
of High Strength Steel, Proceedings of
IABSE, Colloquium, Lausanne, 1982, p
137144 (in French)
D. Francois, Role of the Residual Stress on
the Fracture, Conference Proceedings, Residual Stress in Welding Structure, CETIM,
SENLIS, Dec 1987, p 8797 (in French)
W.H. Friske, Shot Peening to Prevent the
Corrosion of Austenitic Stainless Steels,
Rockwell International Cooperation, 1975
A. Niku-Lan`, M. Menier, and G. Beranger,
Effect of the Surface State on the Stress
Corrosion Behaviour of a Stainless Steel,
CETIM Informations, No. 78, Dec 1982
C. Richard, J. Lu, J.F. Flavenot, and G. Beranger, Study of Residual Stress in the
Plasma Sprayed NiCrAlY Coating and
Characterization of Interface Toughness,
Memoires et Etudes Scientifiques, Revue de
metallurgie, May 1991, p 295306 (in
French)
C.S. Richard, G. Beranger, J. Lu, and J.F.
Flavenot, The Influence of Heat Treatments
on the Adhesion of Plasma Sprayed NiCrAlY Coatings, Surf. Coat. Technol., Vol
82, 1996, p 99109

24. C.S. Richard, G. Beranger, J. Lu, J.F. Flavenot, and T. Gregoire, Four-Point Bending
Tests of Thermally Produced WC-Co Coating, Surface and Coatings Technology, (No.
78), 1996, p 284294
25. C.S. Richard, J. Lu, G. Beranger, and F. Decomps, Study of Cr2O3 Coating Materials,
Part II: Adhesion to a Cast Iron Substrate,
J. Therm. Spray Technol., ASM International, Dec 1995, Vol 5 (No. 4), p 347352
26. R.J. Arsenault, Tensile and Compressive
Properties of Metal Matrix Composites,
Metal Matrix Composites: Mechanisms and
Properties, R.K. Evertt and R.J. Arsenault,
Ed., p 133167
27. P. Balladon, Stress Relieve After Welding,
Criterion and Process, Conference Proceedings, Residual Stress in Welding Structure,
CETIM, Senlis, Dec 1987, p 4569 (in
French)
28. R. Lematre, J.L. Lebrun, and J. Maeder,
Residual Stress and Fatigue, Materiaux et
Techniques, Sept 1982 (in French)
29. J. Lu, J.F. Flavenot, A. Turbat, and D. Francois, Modeling of the Relaxation of Residual Stress under Fatigue Loading, Conference proceedings, Residual Stress at Design
Office, CETIM, Senlis, Nov 1991, p 7198
(in French)
30. J. Lu, J.F. Flavenot, and A. Turbat, Prediction of the Residual Stress Relaxation During Fatigue, Mechanical Relaxation of Residual Stresses, ASTM STP993, 1998, p
7590
31. J. Lu, J.F. Flavenot, A. Diboine, S. Lasserre,
C. Froustey, M. Bennebach, and T. PalinLuc, Development of a General Multiaxial
Fatigue Criterion for High Cycles of Fatigue
Behavior Prediction, Multiaxial Fatigue
and Design, ESIS 21, A. Pineau, G. Cail-
32.
33.
34.
35.
36.
37.
38.
39.
40.
letaud, and T.C. Lindley, Ed., Mechanical

Engineering Publications, London, 1996, p
477487
K. Dang Van, On the Fatigue of Metals, Science Technique Armement, Memorial de
lartillerie francaise, 3ie`me Fascicule, 1973,
p 647722 (in French)
B. Crossland, Effect of Large Hydrostatic
Pressure on the Torsional Fatigue Strength
of an Alloy Steel, Int. Conf. on Fatigue of
Metals, IME/ASME, (London), 1956, p
138149
G. Sines, Fatigue Criteria under Combined
Stress or Strains, Transactions ASME, J.
Eng. Mater. Technol., Vol 13, April 1981,
p 8290
H. Kakuno and Y. Kawada, A New Criterion of Fatigue Strength of a Round Bar
Subjected to Combined Static and Repeated
Bending and Torsion, Fatigue of Engineering Materials and Structures, Vol 2, 1979,
p 229236
I.V. Papadopoulos, Cyclic Fatigue of
Metal, a New Approach, Ph.D. dissertation, Ecole Nationale des Ponts et Chausse`es, France, Dec 1987 (in French)
A. Galtier and J. Seguret, Multiaxial Fatigue
Criterion, Revue Francaise de Mecanique,
No. 4, 1990, p 291299 (in French)
A. Deperrois, On the Fatigue Limit Calculation, Ph.D. dissertation, Ecole Polytechnique, France, June 1991 (in French)
R. Akrache and J. Lu, 3D Calculation of
High Cycle Fatigue Life under Out-of-Phase
Multiaxial Loading, Fatigue and Fracture
of Engineering Materials and Structures,
Vol 22, July 1999, p 527534
J. Lu and J.F. Flavenot, Prediction of the
Residual Stress Relaxation During Fatigue
Loading and Taking the Residual Stress in
41.
42.
43.
44.
45.
46.
47.
Multiaxial Fatigue Criteria into Account,

Proceedings of 2nd International Conference on Residual Stress, (Nancy, France),
Elsevier Applied Science, Beck et al., Ed.,
Nov 1988, p 784790
W.P. Evans, R.E. Ricklefs, and J.F. Millian,
Local Atomic Arrangements Studied by Xray Diffraction, J.B. Cohen and J.E. Hillard,
Ed., Proc. of a Symposium, (Chicago), Feb
1965, Chap. 11
A. Bignonnet, Fatigue Behavior of a Shot
Peened 35NCD16 Grade Steel, Evolution of
Residual Stress Due to the Shot Peening as
a Function of the Type of the Loading, Proceedings of Ninth Conference on Shot Peening, CETIM, Senlis, France, Nov 1985, p
107116 (in French)
FATIGUE3D, User Manual, UTT-LASMIS, Jan 2000
ABAQUS Theory Manual, Hibitt, Karlsson,
and Sorensen, Inc., 1996
A. Milley, E. Rouhaud, and J. Lu, Prediction
of Residual Stress Relaxation in Three-Dimensional Structures under Tensile and
Torsional Loadings, Fifth International
Conference on Residual Stress, Vol 1, T. Ericsson, Ed., (Linkoping), 1618 June 1997,
p 424429
S. Rasouli Yazdi and J. Lu, Simulation of
Quenching and Fatigue Relaxation of Residual Stresses in Aluminium Parts, Fifth International Conference on Residual Stress,
Vol 1, T. Ericsson, Ed., (Linkoping), 1618
June 1997, p 490495
S. Rasouli Yazdi and J. Lu, Inverse Method
for Determining the Quenching Conditions
Using a Residual Stress Field, Proceedings
of the Third Int. Conf. on Quenching Control and Distorsion, ASM, (Prague), March
1999

DOI: 10.1361/hrsd2002p027

Residual Stresses and Fatigue Behavior

D. Lohe, K.-H. Lang, O. Vohringer
University of Karlsruhe, Germany
COMPONENTS MAY FAIL DURING SERVICE in very different ways. Important aspects
are:
Onset of plastic deformation or exceeding of
allowable plastic deformation
Crack initiation or exceeding of allowable
crack length
Unstable crack propagation that results in
fracture of the component
Instability such as buckling, which may result
in elastic or plastic collapse and, hence, catastrophic failure

In practice, failure of a component is almost
always determined by the interaction of several
external parameters, such as unexpected high
static and/or cyclic loads, friction, thermal energy, oxidation, corrosion, and so forth. On the
other hand, the component itself may contribute
to premature failure, if there is confusion of the
component material, wrong heat treatment, surface decarburization, and so forth.
The residual stress state in a component may
be one of the most important parameters influencing its service behavior, particularly regarding high-strength material states. Residual
stresses may reduce yield or elastic collapse
loads and may promote corrosion cracking (Ref
1, 2). With respect to fatigue, which is the topic
of this article, residual stresses may alter the cyclic deformation behavior, promote or retard
crack initiation, accelerate or decelerate crack
propagation, and may be beneficial or detrimental to finite fatigue life and the endurance limit.
It is important to realize that in a component locally very high amounts of residual stresses may
exist. For example, in a spring steel up to 2000
MPa compressive residual stresses may be produced by a combined warm and stress peening
process (Ref 3).
In practice, no component is free of micro residual stresses. Almost all components have
macro residual stresses to a certain extent. Only
a small number of components exist where
macro residual stresses are negligible, for example after a suitable stress relaxation heat treatment. The reasons for residual stress formation
in components are treated in other articles in this
Handbook. The most important ones are the various manufacturing processes. The fundamental
mechanisms of residual stress formation are inhomogeneous plastic deformation caused by mechanical, thermomechanical, or thermal attack
and the development of constraint between different constituents of a component such as
phases or coatings during the formation of the
constituents (e.g., during the coating process)
and/or during cooling from a higher process
temperature (e.g., the sintering temperature).
Manufacturing processes not only result in the
formation of definite macro residual stress states,
but may also effect other changes of the component state close to its surface:
Formation of a specific topography
Work-hardening or (in very hard material
states) work-softening processes and hence

alteration of the micro residual stress state
Phase transformation
Crack initiation
All these processes may change the fatigue behavior of a given component. It is hardly possible to completely separate these influences, even
though this is very important for the understanding of the existing relationships.
Residual Stress
Stability of Residual Stresses. For the assessment of the influence of residual stresses on
fatigue behavior, the stability of the residual
stress state is of utmost importance. The reader
is referred to the article Stability of Residual
Stresses in this Handbook, which deals with the
relaxation of residual stresses due to monotonic
or cycling loading or due to thermal energy. It
is shown there that the micro residual stress state
is more stable against mechanically and thermally induced relaxation than the macro residual
stress state. On the other hand, it is also proved
that during fatigue the micro residual stress state
of a given steel may be changed by cyclic hardening and/or cyclic softening processes, which
are closely related to cyclic plastic deformations.
Hence, during fatigue loading of a component
with locally varying macro as well as micro residual stress states, complex interactions of the
macro residual stress state with the cyclic loading stresses and the micro residual stress state
with work-hardening and/or work-softening

processes occur. In both cases, the amount of
cyclic plastic deformation is the most important
parameter. Since during fatigue loadings that result in technically relevant lifetimes or in infinite
life, the cyclic plastic deformations decrease
with increasing hardness, it is generally expected
that the influence of macro residual stresses is
lower in low-strength steels than in high-strength
steels.
Comparison between Loading Stresses and
Residual Stresses. Since the alteration of the
micro residual stress state caused by manufacturing processes changes the local hardness
within a component and hence the local strength,
it is convenient to comprise the corresponding
influence on the fatigue behavior of the component as a change of the local fatigue resistance
of the steel used for the component.
On the other hand, stable macro residual
stresses are static stresses and may be regarded
as locally varying mean stresses. However, it is
important to realize that there are definite differences between loading mean stresses and residual stresses:
Loading mean stresses exist as a consequence
of acting external forces, moments, pressure

and/or internal pressure, and eventually temperature gradients. Residual stresses exist as
a consequence of inhomogeneous plastic deformation and/or as a consequence of constraints between different constituents of a
component.
Distribution and sign of loading mean
stresses depend on the external loading, the
geometry of the component, the elastic-plastic deformation behavior of the material (if
maximum loading stresses exceed the yield
strength or the cyclic yield strength), and on
thermophysical material properties (if temperature gradients exist). Distribution and
sign of residual stresses depend on the
amount and extension of inhomogeneous
plastic deformation in relation to the whole
component volume, on the interaction of mechanically and/or thermally driven deformation during manufacturing processes, on the
strength of the material state, and/or on the
volume fraction, the elastic as well as the
Stress amplitude (a), MPa
thermophysical properties of the constituents

of the component that impose constraint on
each other.
Loading mean stresses are in equilibrium
with external forces and moments. Residual
stresses are in equilibrium with themselves
regarding balance of forces and moments
with relation to any sectional area and axis,
respectively. The latter is also true for thermally induced loading stresses.
Loading mean stresses may be changed and
finally disappear, if external loadings and/or
Nf
Ni
Ns
Number of cycles, N
Fig. 1
Wohler curves for slip band formation (Ns), microcrack initiation (Ni), and failure (Nf). Source:
Plastic strain amplitude (a,p), %
Ref 5
0.4
0.3
0.2
0.1
0
1
10
102
103
104
105
Some Aspects of Fatigue of Steels
Plastic strain amplitude versus number of cycles

of normalized AISI 4140 steel. Source: Ref 68
The fatigue behavior of metallic materials is

characterized by different processes. Generally,
it contains four successive stages A, B, C, and
D:
0.15
Inhomogeneous
softening
Homogeneous
hardening
0.10
0.05
01
10
102
Number of cycles, N
Cyclic deformation curve and development of

inhomogeneous strain distribution during cyclic
softening of normalized AISI 4140 steel. Source: Ref 68
Fig. 3
The interaction of the residual stress state with

the cyclic loading stresses determines the fatigue
behavior of components. The following sections
of this article discuss the consequences of this
interaction on cyclic deformation behavior, on
crack initiation and crack propagation, as well as
on the fatigue life and on the endurance limit.
Throughout this article, no special account is
given for multiaxial loading or residual stress
states and equivalent stresses are not used. Instead, this article considers simple loading states
and those components of the residual stress
statewhich is of course always multiaxial
that are relevant for the interaction of the cyclic
loading and the residual stress state under consideration. With regard to notched specimens
in which a multiaxial stress state always exists
upon mechanical loadingthe treatment is
based almost entirely on nominal stresses and
stress amplitudes. Hence, the fatigue strengths
and the cyclic loadings corresponding to finite
fatigue lives given in this article are nominal
stress amplitudes. First of all, the section Some
Aspects of Fatigue of Steels introduces some
basic definitions and relations of fatigue. This is
done here rather briefly. The phenomenon and
the processes of the fatigue of steels are presented in general and in detail in Volume 19 of
the ASM Handbook (Ref 4).
106
Number of cycles, N
Fig. 2
if existing temperature gradients are changed

and finally removed. Residual stresses may be
changed and eventually disappear by thermally and/or mechanically induced relaxation.
Loading mean stresses are not influenced by
cyclic plastic deformation in stress-controlled
loading, unless stress gradients become very
highfor example in the root of sharp
notches, where mean stress redistribution
may occur. Residual stresses always relax, if
the cyclic loading exceeds certain threshold
values (see the article Stability of Residual
Stresses in this Handbook).
Stage A: Cyclic deformation with work-hard-
ening or/and work-softening effects in the total volume (homogeneous loading such as
push-pull of smooth specimens) or in the
highest loaded regions (inhomogeneous loading such as bending, torsion, or any loading
of notched components) and development of
persistent slip bands at the surface
Stage B: Microcrack initiation and propagation. Normal evolution of a macrocrack in the
surface region
Stage C: Stable macrocrack propagation connected with changes in the state of the material in the crack tip plastic zone
Stage D: Unstable crack propagation and fail-
ure
An example of the consequences of these fatigue
stages is shown in Fig. 1 in an S-N diagram
(Wohler diagram) for the plain carbon steel SAE
1020 (German grade C 20), which presents experimental results for persistent slip band formation (ra /log Ns curve), microcrack initiation
(ra /log Ni curve), and failure (ra /log Nf curve)
(Ref 5).
Cyclic Deformation Behavior. Investigations on the cyclic deformation behavior in stage
A using computerized servohydraulic test systems enable the determination of the complete
stress/total-strain response at distinct numbers
of cycles that yields to hysteresis loops. In a
stress-controlled test, the plastic strain amplitude
ea,p plotted versus log N results in a cyclic deformation curve that corresponds to the microstructural changes in the material during cyclic
loading. Decreasing (increasing) ea,p values with
an increasing number of cycles N are typical for
a cyclic work-hardening (work-softening) behavior.
Characteristic cyclic deformation curves of
the low-alloy steel AISI 4140 (German grade 42
CrMo 4) in a normalized state are shown in Fig.
2. During stress-controlled push-pull tests with
R rmin/rmax 1 and ra RP0.2, the plastic
strain amplitudes systematically change with the
stress amplitudes applied. In the initial stage of
cyclic loading, a quasi-linear elastic behavior is
found. This stage is followed by a period of increasing ea,p values due to cyclically induced
work-softening effects that are connected with
inhomogeneous deformation processes caused
by fatigue Luders bands. This can be proved with
photoelastic investigations. A typical result obtained for this steel is presented in Fig. 3. Plastically deformed areas appertaining to the individual stages of inhomogeneous cyclic
deformation were registered photographically
and marked as hatched areas on the specimens
that are allied to the cyclic deformation curve.
Similar macroscopic softening effects occur in
all normalized states of plain carbon steels and
low-alloy steels with carbon contents below 0.6
wt% (Ref 68).
The quasi-linear elastic period shortens and
the extent of cyclic work softening grows simultaneously with increasing stress amplitude
ra. After reaching a maximum of softening, the
materials behavior is mainly characterized by
work hardening, and the ea,p values decrease
with increasing number of cycles. At the end of
the specimens life, the plastic strain amplitude
increases again, however fictitiously as a consequence of changes in the compliance of the
specimens due to crack propagation. The plot of
ra versus ea,t at N Nf /2, the so-called cyclic
stress-strain curve, of this steel is presented in
Fig. 4 together with the monotonic-stress/totalstrain curve at nearly the same strain rate e . In the
elastic-plastic range, the cyclic curve is lower
than the monotonic curve due to the work softening. The cyclic stress-strain curves of plain
Residual Stresses and Fatigue Behavior / 29
Monotonic:
0.4 101s1
400
300
Cyclic:
0.3 101s1
N = Nf/2
200
100
0
0
0.1
0.3
0.2
Total strain (t) or total strain amplitude (a,t), %
Monotonic stress-strain curve and cyclic stressstrain curve at Nf /2 of normalized AISI 4140 steel
at nearly the same strain rates. Source: Ref 68
Fig. 4
0.8 wt% C
0.45
400
0.22
300
0.01
200
100
0
0.1
0.2
0.3
810
750
0.2
0.1
700
600
500
Fig. 6
68
a = 850 MPa
0.3
10
102
103
104
Number of cycles, N
105
(one) 101s1
800
Cyclic:
0.6 101s1
N = Nf/2
600
400
200
0
10
Total strain (t) or total strain amplitude (a,t), %

Monotonic stress-strain curve and cyclic stressstrain curve at Nf /2 of quenched-and-tempered
AISI 4140 steel at nearly the same strain rates. Source: Ref
68
Fig. 7
Fig. 5 Cyclic stress-strain curves of plain carbon steels

with the indicated carbon contents (wt%) in normalized conditions. Ref 810
0.4
Monotonic:
1000
0.4
0.5
Stress () or stress amplitude (a), MPa
500
ments is characterized by a continuous increase

of the plastic strain amplitude that lasts up to
failure. The cyclic softening sets in after an incubation number of cycles Ni, which depend on
the applied loading amplitude. Figure 6 shows
the cyclic deformation curves ea,p versus log N
at different stress amplitudes for AISI 4140 tempered at 570 C. As a consequence of this cyclic
deformation behavior, the cyclic stress-strain
curves of quenched-and-tempered steels proceed
below the monotonic curves (see Fig. 7). In this
material state, the cyclic softening is caused by
an inhomogeneous deformation behavior. The
plastic deformation develops localized in distinct
zones and remains localized in these zones up to
failure. Figure 8 shows results from photoelastic
investigation carried out on quenched-and-tempered AISI 4140. In the plastically deformed
zones, the cyclic softening is caused by a rearrangement of the dislocations to energetically
more favorable dislocation structures such as cell
structures. As a consequence, the stored strain
energy and the hardness decrease. Within these
plastically deformed zones, microcrack initiation
and propagation occur (Ref 68).
Crack Initiation. The plastic strains generated by a cyclical loading lead to the formation
of microcracks in the material after a certain
0.3
0.2
a = 750 MPa
Inhomogeneous
softening
0.1
0
10
106
Plastic strain amplitude versus number of cycles

of quenched-and-tempered AISI 4140 steel. Ref
number of cycles. These fatigue cracks usually

develop at the surface. Only in special cases
for example, very heterogeneous materials such
as case-hardened steelor at special loading
conditionssuch as Hertzian pressurefatigue
cracks may initiate below the surface. The first
microscopic cracks can appear in a very early
stage of the cyclic loading. Depending on the
loading conditions and the material, microcracks
can already be detected after 0.1 to 10% of the
lifetime and number of cycles to failure, respectively.
In ductile materials, changes of the surface
structure precede fatigue crack initiation. Fatigue
slip bands (persistent slip bands, PSBs), which
often are connected to extrusions and intrusions,
arise from irreversible gliding. These persistent
slip bands are often places in which first fatigue
cracks initiate. The crack initiation at persistent
slip bands often is typically for lower to middle
loading amplitudes. At higher amplitudes, cracks
may also initiate at grain boundaries. Crack initiation and the first phase of crack propagation
in ductile materials are controlled by shear
stresses. Within the first grain layers, the cracks
grow at an angle of 45 to the load axis. This
phase is called stage I crack growth. After reaching a certain length, the cracks become the socalled stage II cracks. Further propagation of theses cracks is controlled by normal stresses. Stage
II cracks grow approximately rectangular to the
load axis. In brittle materials, crack initiation is
often observed at inclusions and other defects.
Crack Propagation. To describe the stress
and strain field in front of a crack tip the loading
details and the geometry of a cracked specimen
or component must be considered. If the plastic
zone in front of a crack tip is small in comparison
with the dimensions of the specimens or components, the linear elastic fracture mechanics are
applicable, and the local loading conditions in
front of the crack tip can be described by the
stress intensity factor, K. Therefore, Paris,
Gomez, and Anderson (Ref 11) considered the
stress intensity factor as a driving force for the
propagation of a crack and suggested a crack
growth equation in the form:
da
C (dK)m
dN
Stress () or stress amplitude (a), MPa
carbon steels with 0.01, 0.22, 0.45, and 0.80 wt%

C (German grade Ck 01, Ck 22, Ck 45, and C
80) in normalized conditions, determined from
cyclic deformation curves at N Nf /2, are
drawn up in Fig. 5. With increasing carbon content and pearlite content, respectively, an increasing stress amplitude at ea,p constant occurs due to a phase-hardening mechanism (Ref
810).
The cyclic deformation behavior of quenchedand-tempered steels in stress-controlled experi-
102
103
104
Number of cycles, N
Cyclic deformation curve and development of
inhomogeneous strain distribution during cyclic
softening of quenched-and-tempered AISI 4140 steel.
Source: Ref 68
Fig. 8
(Eq 1)
where da/dN is the crack growth rate, DK the

range of the mode I stress intensity factor, and
C and m are constants (Ref 12). With this equation, which is the well-known Paris law, it is
possible to describe the crack propagation in the
midrange of the crack growth rate. Figure 9
shows the qualitative features of the crack
growth behavior in a double logarithmic plot of
the growth rate versus the range of the stress
intensity factor for experiments with constant
stress range Dr rmax rmin. Distinct curves
are observed for different mean stresses rm or
values of R rmin /rmax, respectively. The
lower parts of the curves exhibit a threshold
value of DK below which crack growth is essentially arrested. Typically, the crack growth
Crack growth rate, log da/dN
R1
R2
R3
Stress intensity factor range, log K
Fig. 9
Crack propagation rate versus stress intensity factor range for stress ratio values R1 R2 R3
rate in this region is below 108 mm/cycle. The

upper parts of the curves represent the crack
growth behavior as the Kmax values approach the
critical stress intensity factor KIc, which marks
the onset of unstable crack propagation. Then the
crack growth rate increases boundlessly.
The Paris law does not represent the complete
crack growth curves shown in Fig. 9. The upper
part of the curves depends on KIc, and the lower
part on the threshold value DKth. Therefore, several complementations were suggested. An equation that incorporates all of the features of crack
growth curves has been used by NASA (Ref 13).
It has the form:
Quantitatively, the following relationships are

valid:
ea,p ef N a
f
and
ea,e rf /E N b
f
(Eq 6)
with ef the fatigue ductility coefficient, rf the

fatigue strength coefficient, the fatigue ductility exponent, b the fatigue strength exponent,
and E the Youngs modulus. Rearrangement of
both equations yields to:
ea,p N fa constant
da
C (1 R)m DKn (DK DKth)p
dN
[(1 R) KIc DK]q
(Eq 5)
(Eq 7)
and
(Eq 2)
ea,e E N b
ra N b
constant
f
f
log a
log f
log Rm
log Rf
log Ns
Strain amplitude (a)
Fig. 10
log Ne log Nf
Stress Wohler curve (S-N curve)
f/ E
E N
N
Number of cycles to failure, Nf
Fig. 11
Total strain Wohler curve
This equation is able to account for the threshold

effect, the R ratio effect, and the final instability
effect. By suitable choice of the exponents, this
equation can accept a number of different variants of crack propagation laws. For m p q
0, for example, the Paris law is obtained.
The concepts of linear elastic fracture mechanics and the use of the stress intensity factor
are limited to cases in which small-scale yielding
develops at the crack tip. Moreover, the applicability of Eq 1 and 2 is restricted to sufficiently
long macroscopic cracks. Short cracks can grow
at smaller ranges of the stress intensity factor as
DKth. At comparable DK values, they can exhibit
clearly higher crack growth rates than long
cracks (Ref 14).
Lifetime Behavior. The lifetime behavior of
steels is commonly described in an S-N diagram
(Wohler diagram) (Ref 15), where the loading
amplitude is plotted versus the number of cycles
to failure in double logarithmic scaling. For
stress-controlled fatigue tests, the S-N curve is
frequently described by three straight lines (see
Fig. 10). In the range of the quasi-static strength
Rm (Nf Ns) and the fatigue strength Rf(N
N0f ), the endurable stress amplitude is constant
and given by the ultimate strength Rm and the
fatigue strength Rf, respectively. In the range of
finite life (Ns Nf Ne), the Basquin relation
(Ref 16):
Fatigue strength, Rf
ra rf N b
f
often describes the lifetime behavior satisfactorily. rf is the fatigue strength coefficient and b
the fatigue strength exponent.
In the case of total-strain-controlled fatigue
tests the ea,t values in the total strain Wohler
curve (Fig. 11) can be separated into the elastic
and plastic parts:
Re
Goodman
0
Rf
Gerber
ea,t ea,e ea,p

Re
Mean stress m
Fig. 12
Endurance limit versus mean stress
(Eq 3)
Rm
(Eq 4)
From experience it is known that both parts extend exponentially from the number of cycles to
failure Nf, which means that the plots log ea,e
versus log Nf and log ea,p versus log Nf are linear.
(Eq 8)
which are the well-known relationships from

Coffin-Manson (Ref 17, 18) (Eq 7) and Basquin
(Eq 8).
Tensile mean stresses promote crack initiation
and accelerate crack propagation. Therefore, the
fatigue strength Rf decreases with increasing
mean stresses. This correlation is normally represented in a Smith (Ref 19) or in a Haigh (Ref
20) diagram. As an example of such fatigue
strength diagrams, Fig. 12 shows a Haigh diagram. The fatigue strength Rf is plotted versus
the mean stress rm. The Goodman approximation (Ref 21):
Rf R0f (1 rm /Rm)
(Eq 9)
which represents a straight line between the fatigue strength R0f at rm 0 and the tensile
strength Rm, is frequently used to estimate fatigue strength. For practical use, the allowable
stresses are restricted by the demand that macroscopically no plastic deformation should occur. This restriction leads to a straight line connecting the yield strength Re at both axes.
Through this, the allowable stress amplitudes lie
in the hatched area in Fig. 12. The slope of the
Goodman straight line is often also described as
mean stress sensitivity M R0f rm /Rm. The
Goodman approximation normally leads to conservative estimations. Often, the Gerber parabola
(Ref 22):
Rf R0f [1 (rm /Rm)2]
(Eq 10)
is used as a nonconservative estimation of the

fatigue strength. Mostly, experimental results
range between the Goodman and the Gerber approximations.
In the range of infinite lifetime, the influence
of mean stresses can be described by damage
parameters. Two frequently used damage parameters are the one proposed by Smith, Watson,
and Topper (Ref 23):
PSWT (ra rm) ea,t E
(Eq 11)
and the one proposed by Ostergren (Ref 24):

(Eq 12)
ra and rm are the imposed stress amplitude and

mean stress, respectively, ea,t and ea,p are the resulting total strain amplitude and plastic strain
amplitude, respectively, and E is the Youngs
modulus. Normally, for ea,t and ea,p the values at
half of the lifetime are used.
Influence of Residual Stresses on

the Cyclic Deformation Behavior
0.4
a in MPa
Unpeened
Shot peened
0.3
380
0.15
10
102
103
104
105
106
Number of cycles, N
0.4
a in MPa
450
Unpeened
Shot peened
0.3
550
600
0.05
400
375
350
350
400
275
0
1
10
(b)
102
103
104
Number of cycles, N
1.0
900
0.8
105
102
103
104
Number of cycles, N
10
a in MPa
Unpeened
800
Shot peened
750
700
0.2
500
400
550
600
(c)
10
102
103
104
Number of cycles, N
105
a in MPa
Unrolled
0.03
Deep rolled
600
650
550
600
0.02
500
0.01
550
500
0
102
103
104
Number of cycles, N
10
(b)
105
106
Cyclic deformation curves for stress-controlled

push-pull loading of (a) normalized and (b) nitrocarburized specimens of the SAE 1045 steel in unpeened
and deep-rolled conditions. Source: Ref 27
Fig. 14
0.10
Untreated
0.08
Shot peened
0.06
Deep rolled
0.04
0.02
0
106
Cyclic deformation curves for stress controlled

push-pull loading of (a) normalized, (b)
quenched-and-tempered (730 C/2 h), and (c) quenchedand-tempered (570 C/2 h) smooth specimens of the steel
AISI 4140 in unpeened and shot peened conditions.
Source: Ref 25, 26
Fig. 13
106
0.04
1
1
105
106
0.6
0.4
500
450
1
450
550
0.2
0.1
a in MPa
Unrolled
Deep rolled
510
0.10
(a)
(a)
600
325
300
250
0.1
0.20
350
0.2
Characteristic Examples. The influence of

macro and micro residual stresses on the cyclic
deformation behavior can be studied very well
after mechanical surface treatments, for example
by shot peening or deep rolling. Characteristic

cyclic deformation curves for stress-controlled
push-pull loading of different heat treated
smooth specimens of the steel AISI 4140 (German grade 42 CrMo 4) are compared in Fig. 13
in unpeened and in shot peened conditions with
compressive residual stresses at the surface (Ref
25, 26). In the normalized state (Fig. 13a), the
onset of cyclic deformation is different in both
conditions, since the shot peened specimens with
surface compressive residual stresses rrs
POST (ra rm) ea,p E
10
100
103
Number of cycles, N
104
105
Cyclic deformation curves of different mechanically surface treated materials for stress-controlled push-pull loading in the low-cycle fatigue (LCF) regime of the plain carbon SAE 1045 steel. Stress amplitude
ra 350 MPa, Almen intensity I 0.120 mm A, rolling
pressure p 150 bar; surface residual stresses: shot peening rrs 500 MPa, deep rolling, approximately 600/
350 MPa. Source: Ref 28, 29
Fig. 15
400 MPa show cyclic softening from the first

cycle and higher plastic strain amplitudes during
the first cycles for stress amplitudes ra between
250 and 350 MPa. After a certain number of cycles, the opposite tendency can be detected and
the plastic strain amplitudes of the shot peened
conditions are smaller than those of the unpeened
material states. However, for the same ra values
the plastic strain amplitudes of both conditions
approach another at relatively high numbers of
cycles. Corresponding results for a quenchedand-tempered (730 C/2 h) AISI 4140 steel are
presented in Fig. 13(b). In the unpeened condition, the characteristic cyclic deformation behavior of quenched-and-tempered steels occurs with
a quasi-elastic incubation period, which is followed by cyclic softening until crack initiation.
After shot peening that generates surface compressive residual stresses rrs 410 MPa, the
onset of cyclic softening is shifted to smaller
numbers of cycles. Furthermore, it is interesting
to note that for identical stress amplitudes and
comparable numbers of cycles, the higher plastic
strain amplitudes are always measured for the
shot peened specimens. Figure 13(c) shows a
compilation of cyclic deformation curves for another quenched-and-tempered (570 C/2 h) AISI
4140 steel with a higher strength compared with
the steel condition in Fig. 13(b). In this case, the
shot peened condition that has surface compressive residual stresses rrs 530 MPa is characterized for all investigated ra values by small
measurable plastic strain amplitudes during the
first cycle that diminish or disappear first of all
with an increasing number of cycles. After a subsequent regime of quasi-elastic behavior, cyclic
softening, which yields to lower plastic strain
amplitudes and larger numbers of cycles to failure in comparison with the unpeened conditions,
is dominant.
The effect of deep rolling on the behavior of
the cyclic deformation curves is presented in Fig.
14(a) for the normalized plain carbon steel SAE
1045 (German grade Ck 45) (Ref 27). At comparable stress amplitudes essentially smaller
plastic strain amplitudes are measured in the
deep-rolled condition. In this materials state, the
cyclic softening effects are extremely restricted,
but continue during the whole lifetime until
crack initiation. As a consequence of the smaller
plastic strain amplitudes at identical stress amplitudes, higher numbers of cycles to failure are
observed for the deep-rolled condition compared
with the normalized ones. Figure 14(b) shows
the influence of deep rolling in the case of a nitrocarburized SAE 1045 steel. Due to the different microstructures in the near-surface areas of
the specimens, the same stress amplitudes yield
plastic strain amplitudes that are five times
smaller than those of the normalized state. The
deep-rolling process modifies the cyclic deformation curves of the nitrocarburized specimens
in a manner similar to the normalized specimens.
However, the reduction of the plastic strain amplitudes is much less pronounced.
Figure 15 shows plastic strain amplitudes as a
function of the number of cycles plotted for
Cyclic stress-strain curves of deep-rolled AISI

304 (rolling pressure 150 bar, surface residual
stresses r 350 MPa) using compact as well as hollow
specimens with different percentages of strain-hardened
layers compared with an untreated state and a shot peened
condition (rrs 450 MPa). Source: Ref 28, 29
damage is clearly correlated with plastic strain

amplitude, the benefit of mechanical surface
treatments becomes obvious. The influence of
mechanical surface treatments on cyclic plasticity can be summarized in cyclic stress-strain
curves, which correlate stress amplitudes and
plastic strain amplitudes for, for example, half
the number of cycles to failure (see the section
Cyclic Deformation Behavior). Figure 16, as
an example, shows data for an austenitic AISI
304 steel in shot peened as well as deep-rolled
conditions compared with untreated specimens
(Ref 28, 29). Effects of mechanical surface treatments are more distinct with a higher ratio between the area of affected surface layer and the
cross section of the specimen. In the case presented here, hollow thin-walled specimens were
prepared from compact specimens for analysis.
As shown in Fig. 16, cyclic yield strength considerably increases compared with compact
specimens, if only the fatigue behavior of the
surface layers of mechanically surface treated
components is investigated.
If residual stresses are present in the root of
notched specimen, marked changes always occur
in the initial parts of the notch root cyclic deformation curves (ea,p /log N curves) and in the
notch root cyclic mean strain curves (em /log N
curves) that depend on the sign and the magnitude of the residual stresses. Characteristic examples are presented in Fig. 17 for normalized
True plastic strain amplitude (a,p), %
True plastic strain amplitude (a,p), %

push-pull loading with the stress amplitudes indicated for a normalized SAE 1045 steel (Ref
28, 29). Results for annealed as well as for shot
peened or deep-rolled conditions are shown. One
can clearly see that both mechanical surface
treatments considerably diminish the plastic
strain amplitudes. Due to the thicker affected
surface layer in the case mentioned first, the effect for deep-rolled states is more pronounced
than for shot peened states. Because fatigue
700
600
500
Deep rolled (1 mm wall thickness)

Deep rolled (0.5 mm wall thickness)
Deep rolled (0.3 mm wall thickness)
400
300
200
Untreated
Deep rolled
Shot peened
100
0
0
0.1
0.2
0.3
0.4
0.5
Fig. 16
rs
l
0.4
n,a in MPa
rs = 15 MPa
rs = 490 MPa
0.3
220
0.2
240
200
220
175
0.1
150
0
1
10
102
170 150
190
103
104
105
106
0.06
n,a = 200 MPa
0.04
rs = 710 MPa
460
0.02
320
0
1
10
102
103
104
105
106
Number of cycles, N
(a)
Number of cycles, N
0.06
0.30
n,a in MPa
220
190
170
150
0.15
0.00
Plastic mean strain (m), %
(a)
n,a = 200 MPa
0.04
0.02
460
rs = 710 MPa
0.2
1
0.15
1
(b)
10
102
103
104
105
106
(b)
10
102
103
104
105
106
Number of cycles, N
Number of cycles, N
True plastic strain amplitude (a) and plastic

mean strain versus number of cycles (b) for
stress-controlled push-pull loading of normalized notched
specimens (notch factor Kt 3.0) of the SAE 1045 steel in
unpeened and shot peened conditions. Source: Ref 30, 31
Fig. 17
320
True plastic strain amplitude (a) and plastic

mean strain versus number of cycles (b) for
stress-controlled push-pull loading of quenched-and-tempered (400 C/2 h) notched specimens (notch factor Kt
3.0) of the SAE 1045 steel in cut-milled and shot peened
conditions with different surface residual stresses. Source:
Ref 30, 31
Fig. 18
specimens and in Fig. 18 for quenched-and-tempered specimens of SAE 1045 with a notch factor Kt 3.0 that were investigated in push-pull
tests (Ref 30, 31). Figure 17(a) compares the cyclic deformation curves of material states with
very low residual stresses (produced by milling
and subsequent annealing) with rather high residual stresses (produced by shot peening). In
both conditions plastic deformation occurs in the
first cycle at nominal stress amplitudes rn,a
above 150 MPa. The annealed state yields continuous cyclic softening for all investigated rn,a
values. Contrarily, the initial plastic strain amplitudes of the shot peened condition that are
much less dependent on the stress amplitude
compared to the milled and annealed state are
reduced during the first cycles. This is a result
of residual stress relaxation that is the more pronounced the higher the nominal stress amplitude
is (see the following section Evaluation of Experimental Results). The consequences are
smaller effective stress amplitudes and smaller
ea,p values in the following cycles. Hence, the
observed cyclic work hardening is fictitious.
However, at higher numbers of cycles, cyclic
softening occurs with lower ea,p values than in
the unpeened condition.
On the other hand, as can be seen from Fig.
17(b), the shot peening notch root residual
stresses have an influence on the initial parts of
the mean strain curves that were determined for
different nominal stress amplitudes (Ref 30, 31).
As a consequence of the cyclic deformation induced relaxation of the compressive residual
stresses in the notch, the initial parts of the em /
log N curves show negative plastic mean strains,
where the magnitudes of em increase with increasing nominal stress amplitude. The minima
of em are shifted to lower numbers of cycles with
increasing rn,a values. However, from the minima of em at approximately 5% of the number of
cycles to failure, the em /log N curves show increasing mean strains for all nominal stress amplitudes. Simultaneously, a decrease of the tensile compliance in the hysteresis loops that
intensifies with an increasing number of cycles
is observed. By that, the compressive compliance is approximately constant. This finding
proves that this increase of the mean strains with
the number of cycles is caused by an increase of
the number of microcracks and/or the growth of
microcracks. These sections are indicated by
dashed lines in the em /log N curves of Fig. 17(b).
The true plastic strain amplitudes in the cyclic
deformation curves of the quenched-and-tempered (400 C/2 h) specimens in Fig. 18(a) are
influenced by the magnitude of the notch root
residual stresses, but not by their sign (Ref 30,
31). All investigated conditions in Fig. 18(a)
show for the initial cycles fictitious cyclic-workhardening effects due to the relaxation of residual stresses rrs of the surface layers (rrs 460
MPa and 320 MPa produced by upcut milling
as well as 710 MPa by shot peening). Up to
N 20 cycles, increasing magnitudes of the
compressive residual stresses cause higher negative mean strains. However, positive milling re-
0
n,a = 220 MPa
200
n,a = 150 MPa
400
0.20
250
a in MPa
0.16
125
0.12
150
0.08
350
0.04
200
100
0
1
102
10
(a)
3
10
102
103
104
105
106
Number of cycles, N
(a)
105
106
125
250
9
350
15
a in MPa
102
10
400
450
103
104
105
106
Number of cycles, N
Plastic strain amplitude (a) and plastic mean

strain versus number of cycles (b) for axial
stress-controlled cyclic loading with mean stress rm 300
MPa for specimens of the normalized SAE 1045 steel.
Source: Ref 32
Fig. 20
n,a = 200 MPa
rs
300
100
300
600
900
0
(b)
10
102
103
104
105
106
Number of cycles, N
Residual stress relaxation during stress-controlled push-pull loading of notched specimens

(notch factor Kt 3.0) for (a) normalized SAE 1045 in shot
peened conditions and (b) quenched-and-tempered (400
C/2 h) SAE 1045 in cut-milled and shot peened conditions.
Source: Ref 30, 31
Fig. 19
Ratio of cycles N/Nf, %
A Untreated
B Deep rolled
80
60
40
20
0
C Shot peened
Macrocracks
(>750 m)
Short cracks
(200-750 m)
Microcracks
(>200 m)
No damage
marks
Influence of mechanical surface treatments on

the damage evolution of the push-pull loaded
austenitic steel AISI 304. ra 320 MPa; R 1; A, Nf
3859; B, Nf 4445; C, Nf 20,265. Source: Ref 28
Fig. 21
plastic strain of notched specimens (see Fig.

18b).
A study of the residual stress effects on the
cyclic deformation behavior seems to be possible by their simulation by applied mean stresses,
which are homogeneously distributed over the
cross section of the specimen. An example is
presented in Fig. 20 for smooth, normalized SAE
1045 steel specimens that were stress-controlled
loaded in push-pull tests with a constant mean
stress of rm 300 MPa (Ref 32). The cyclic
deformation curves at stress amplitudes ra
125 and 150 MPa in Fig. 20(a) show first of all
a quasi-elastic incubation interval, followed by
considerable cyclic softening within some cycles. Subsequently, within some further cycles,
cyclic work hardening occurs to such an extent
that extremely low plastic strain amplitudes result. For ra 200 MPa, the compressive stress
peak | rm ra |, which is higher than the yield
strength, induces plastic deformation in the first
cycle. This procedure is followed by a rapid cyclic work hardening in such a manner that at cycles above 100 extremely low ea,p values are observed. As a consequence of the constant
compressive mean stress, the specimens shorten
for all investigated stress amplitudes, as proved
by Fig. 20(b). For ra 200 MPa, the mean
strains increase considerably and continuously.
However, for ra 200 MPa the ea,p values saturate above 20 cycles.
200
3
(b)
1
104
150
100
21
600
103
Number of cycles, N
Residual stress (l ), MPa
stricted mean free path of the mobile dislocations

in the work-hardened surface layers.
In hardened as well as in quenched-and-tempered at low temperature conditions, the changes
in the cyclic deformation behavior result not
only from surface hardening or softening effects,
but also from the more stable residual stresses as
can be seen in Fig. 19(b) for notch root residual
stresses of the quenched and tempered (400 C/
2 h) SAE 1045 steel under the loading conditions of Fig. 18(a) (Ref 30, 31). The consequences are a considerable influence of the
residual stresses on the cyclic behavior, especially with regard to the development of mean
rs
sidual stresses yield to positive mean strains. The

increases in the plastic mean strains that appear
for the numbers of cycles above the minimum of
em are caused also by microcracks as described
previously.
Evaluation of Experimental Results. In relatively soft material states, as for example in normalized as well as in quenched-and-tempered at
high-temperature conditions, the consequences
of mechanical surface treatments by shot peening or deep rolling on the cyclic deformation behavior are mainly caused by near-surface micro
residual stresses, that is, work hardening of the
surface layers, because the macro residual
stresses are relaxed very soon by cyclic plastic
deformation (see the article Stability of Residual Stresses in this Handbook). This relaxation
of macro residual stresses is shown in Fig. 19(a)
for shot peening residual stresses in the notch
root of a normalized SAE 1045 steel (Ref 30,
31) under the loading conditions of Fig. 17(a).
The dislocation structures in the ferrite after the
mechanical surface treatment are not stable and
change during cyclic loading in energetically
more favorable arrangements. The formation of
typical fatigue-induced dislocation structures is
combined with cyclic softening effects as presented in Fig. 13(a), 14(a), and 17(a). The small
plastic strain amplitudes of the shot peened or
deep-rolled conditions and the resulting increase in fatigue life are caused by the re-

the Crack Initiation
Characteristic Examples. Crack initiation
occurs as a consequence of microstructural
changes in metallic materials during cyclic loading. Different mechanisms are responsible for
their formation (e.g., Ref 33). If it is accepted
that for given materials states at comparable load
amplitudes increasing amounts of plastic strain
amplitudes lead to decreasing numbers of cycles
to crack initiation Ni, it follows that residual
stresses may extend, shorten, or leave unchanged
the number of cycles to crack initiation. However, experimental investigations concerning the
influence of macro and micro residual stresses
on crack initiation are scarce. This is due to the
difficulties connected with the observation of the
formation and the propagation of small cracks.
A recently published investigation gives a report on the influence of mechanical surface treatments on crack initiation and crack propagation
in push-pull loading of steels (Ref 28). In untreated materials, crack initiation normally takes
place at positions of high localized slip, for example, at extrusions and intrusions that are connected with persistent slip bands. However, as
shown in Fig. 21 in shot peened and deep-rolled
conditions of the austenitic steel AISI 304, crack
formation occurs later than in the untreated state
due to the consequence of numerous obstacles
for slip (dislocations, grain and twin boundaries)
in the work-hardened surface layer that impede
localized slip. In these surface-work-hardened

conditions, no persistent slip band is observed at
all. Furthermore, crack propagation is slower
than in the untreated state due to the effect of
microstructure and compressive residual
stresses. Similar results are found in Ref 28 for
the normalized steel SAE 1045, in Ref 34 for the
plain carbon steel SAE 1080, and in Ref 35 and
36 for a high-strength spring steel AISI 6150
(German grade 50 CrV 4).
However, as reported in Ref 1, the crack initiation time of shot peened specimens is sometimes shorter than that of unpeened specimens
despite increased lifetimes. One example is
given in Fig. 22 for the quenched-and-tempered
steel SAE 1045 in the case of bending fatigue
tests in seawater (Ref 37). With the exception of
high stress amplitudes, the cracks are formed
earlier in shot peened specimens than in ground
ones. This finding is attributed to an enhanced
crack initiation at micronotches resulting from
shot peening, which is obviously supported by
corrosion pittings in the case of seawater environment. However, the number of cycles to failure of the shot peened conditions are higher than
that of ground conditions.
Residual stresses may have a remarkable influence on the location of crack initiation. In the
case of bending fatigue, for example, cracks may
start below the surface, if the magnitude of the
800
compressive residual stresses near the surface is

high enough. Characteristic results are presented
in Fig. 23 for shot peened bending specimens in
a hardened state of the steel SAE 1045 with the
bending fatigue strength Rf 960 MPa after
shot peening with the shot size d 0.3 mm and
Rf 1050 MPa with d 0.6 mm, respectively
(Ref 38, 39). For stress amplitudes ra Rf, crack
initiation was observed directly at the surface.
However for ra Rf, subsurface cracks occurred
as shown in Fig. 24. If the centers of the rosettes
on the scanning electron micrographs (SEMs)
are considered to be the crack initiation points,
it is obvious that the crack initiation depth increases with decreasing stress amplitude (Ref
39). At stress amplitudes that result in a fatigue
life Nf above 107 cycles, crack initiation is located approximately 0.3 mm below the surface.
These findings stimulated the development of a
concept of the local fatigue strength (see the paragraphs on experimental results in the section
Influence of Residual Stresses on S-N curves
in this article).
Evaluation of Experimental Results. The definition of the number of cycles to crack initiation
Ni and particularly the establishment of the affiliated crack length is of decisive meaning for
the evaluation of the influence of different parameters on Ni from experimental results from
(a)
Fracture
Crack initiation
600
200 m
(b)
400
200 m
200
Shot peened
(c)
Ground
0
104
105
106
Number of cycles, N
107
Stress amplitude versus number of cycles to

crack initiation and to failure of a quenchedand-tempered SAE 1045 steel under bending fatigue loading in seawater. Source: Ref 37
200 m
(d)
200 m
200 m
Distance from surface, mm
0.4
200 m
0.2
Site of
macrocrack
initiation
However, decreasing Ni values are observed,

if in the surface area:
The roughness is increased, for example, by
shot peening
Tool marks or material overlaps from shot
peening or deep rolling exist
Tensile macro residual stresses occur that fa-
vor interface cracks between matrix and hard

second phases.
In this context, an investigation of the fatigue
crack initiation in double-edge notch specimens
of SAE 1080 steel with a fine-grained, spheroidized microstructure shows considerable effects
of the macro residual stresses (Ref 34). In a
press-fitted condition, large compressive residual macro stresses occur that remained relatively
stable throughout the fatigue life and thus greatly
increased the numbers of cycles to crack initiation Ni. With reference to the Basquin-relationship, Ni is estimated from the following relationship:
r (r
2ra
a
1/b;
r )
rs
(Eq 13)
where Ni is defined as the number of cycles at a

crack length of a 0.1 mm, rrs is the macro
residual stress, ra is the applied stress amplitude,
and ri 2315 MPa and bi 0.197 are material constants.
200 m
Shot size 0.3 mm

Shot size 0.6 mm
1000
1200

the Crack Propagation
(g)
1400
Crack initiation sites versus stress amplitude of

a hardened SAE 1045 steel in different shot
peened conditions. Source: Ref 38, 39
Fig. 23
stresses
Is smoothed by polishing or deep rolling
(f)
0.3
800
face treatment
Bears additional compressive macro residual
(e)
Becomes work hardened by mechanical sur-
Ni
Fig. 22
0.1
literature. If the corresponding crack length

amounts to only some grain diameters, then Ni
is determined by shear-stress-controlled processes in stage I of crack propagation. However,
if crack lengths of several hundred microns appear (stage II of crack propagation), Ni will be
determined by normal-stress-controlled processes. In the last case, stable macro residual
stresses may have a considerable influence on
the Ni values.
The experimental results achieved until now
for crack formation directly at the surface with
an appropriate consideration of the Ni definition
show increasing numbers of cycles to crack initiation, if the surface area:
Scanning electron micrographs of fractured

specimens of hardened and shot peened SAE
1045 at different stress amplitudes. (a) ra 1300 MPa. (b)
ra 1250 MPa. (c) ra 1200 MPa. (d) ra 1100 MPa.
(e) ra 1000 MPa. (f) ra 950 MPa. (g) ra 900 MPa.
Source: Ref 39
Fig. 24
Characteristic Examples. During cyclic

loading, the lifetime of components is determined significantly by the stage of crack propagation, especially in the case of existing macro
residual stresses. Crack propagation as a consequence of fatigue loading without mean stresses
can be described, with the exception of the crack
initiation and the near-fracture stage, by the Paris
equation (see Eq 1):

da
C(DK)m
dN
(Eq 14)
where DK is the stress intensity range and C and

m are constants. Mean stresses are taken into account by the Foreman equation:
da
C(DK)m
dN
(1 R)Kc DK
200
rs
200
Crack propagation rate (da/dN), mm/cycle
(Eq 15)
where R is the stress ratio and Kc is the stress

intensity factor for plane stress. Both relationships are valid for crack propagation in materials
states without macro residual stresses. However,
Residual stress (y ), MPa
these descriptions imply that a propagating crack

itself is surrounded by a typical residual stress
field as shown in Fig. 25(a) for a high-strength
structural steel of the European grade S690QL1
(Ref 40, 41). A crack was produced by cyclic
loading up to a stress intensity range of DKI
47.4 MPa m. The distribution of the macro re-
104
As received
105
106
Autofrettaged
0.2
0.4
0.6
0.8
1.0
Ratio a/W
Crack propagation rate da/dN versus ratio a/

W in an autofrettaged and an untreated tube of
the SAE 4337 steel. Source: Ref 43
Fig. 27
400
12
8
4
0
4
8
Distance from crack tip, mm
(a)
12
10
150 bar
Crack propagation rate
(da/dN), mm/cycle
rs
Residual stress (y ), MPa
200
200
8
4
0
4
8
Distance from crack tip, mm
Residual stress component rrsy versus distance

from crack tip of the steel S690QL1 after a
mode I base load of DK 47.4 MPa m (a) and an overload of DK 94.8 MPa m (b). Source: Ref 40, 41
102
=2
103
104
=3
105
8
4
75 bar
225 bar
4
2
300 bar
0
1
2
Crack length (a), mm
Fatigue crack propagation rates da/dN in annealed and with different rolling pressures
deep-rolled steel AISI 304. Stress amplitude ra 320 MPa,
surface residual stresses rrs 200 MPa (75 bar), 350
MPa (150 bar), 400 MPa (225 bar), and 300 MPa (300
bar). Source: Ref 29
Fig. 28
Fig. 25
12
103
104
Stress relieved
As welded
105
106
10
20
30 40
60 80
Stress intensity range (K), MPa m
Crack length (aol), mm
Crack propagation rate da/dN versus crack

length a0l for overloads with k 2 and 3.
Source: Ref 40, 41
Fig. 26
Untreated
12
12
(b)
106
400
103
Crack propagation rate da/dN versus stress intensity range DK in the heat-affected zone of
EN S355 steel in an as-welded condition and after a stressrelief heat treatment. Source: Ref 44
Fig. 29
sidual stress component rrs

y , which acts perpendicular to the crack flanks, shows maximum
compressive residual stresses of approximately
350 MPa at the crack tip. The alteration of this
distribution after application of 20 overload cycles with an overload ratio k 2 (DKI 94.8
MPa m) is given in Fig. 25(b). In front of the
crack tip, a larger maximum value and a larger
area with compressive residual stresses compared with Fig. 25(a) are developed. The influence of different overload cycles on the crack
propagation rate is shown in Fig. 26 for k 2
and 3. For both overload ratios, a delayed retardation of crack propagation occurs that is more
pronounced for k 3 than for k 2 due to the
effect of overload-induced compressive residual
stresses. Thus, by sufficiently high overloads,
crack arrest can be produced (Ref 42).
If a crack propagates into a macro residual
stress field, the crack propagation behavior can
be considerably influenced by magnitude and
distribution of the residual stresses. This is
shown in Fig. 27, in which crack propagation
rates within an autofrettaged tube of SAE 4337
steel are compared with those in the untreated
material (Ref 43). As a consequence of the autofrettaging process, a triaxial residual stress
state is created with very high compressive residual stresses near the inner surface of the tube.
In agreement with Eq 14, loading with a constant
stress intensity range results in a constant crack
propagation rate in residual-stress-free tubes.
However, if the crack propagates through the
compressive residual stress field at the inner surface of the autofrettaged tube, considerably
smaller crack propagation rates occur.
Crack velocity is influenced by work hardening as well as by residual stresses in mechanically treated surface layers. This is demonstrated
in Fig. 28 for differently surface-rolled AISI 304
stainless steel. The crack propagation rate da/
dN, which was determined by analyzing striations on cracked surfaces, increases with increasing rolling pressure and is considerably
diminished compared with untreated specimens
(Ref 29).
Finally, the influence of welding residual
stresses at the steel European grade EN S355 is
compared with the behavior of the same steel in
a stress-relieved condition in Fig. 29 in a da/dN
versus DK diagram (Ref 44). The crack propagation rate in the heat-affected zone is considerably lower in the as-welded state than in the
stress-relieved state due to the distribution of
welding compressive residual stresses.
Evaluation of Experimental Results. For
practical purposes, it is very important to know
that crack propagation through residual stress
fields can be modeled quantitatively by introducing an effective stress intensity range DKeff.
This can be seen in Fig. 30, which shows crack
propagation rates da/dN in welded and unwelded specimens of SAE 1019 steel (Ref 45).
Exactly at the welding seam, tensile residual
stresses of about 340 MPa exist that disappear
in a distance of 10 mm. Figure 30 presents crack
propagation rates in the unwelded base material
103
0.1 0 0.5 1.0
104
0.2
R = 0.4
2.0
105
106
10
20
30
40 50 60
(a)
103
R=0
0.5
104
105
106
1.0
2.0

6
10
20
30
40 50 60
(Eq 16)
If the influence of the residual stress contribution

on Kop is known, DKeff can be estimated and used
to predict the fatigue crack growth. Thus, the
double logarithmical plot of the crack propagation rate da/dN versus the effective stress intensity range DKeff in Fig. 30(c) shows that all measured data points form a narrow scatter band. In
welded specimens, the (da/dN)/DK relations are
completely changed (e.g., Fig. 30a and b). Obviously, for small DK values the crack propagation behavior is entirely controlled by the
welding residual stress state, which leads to
identical crack propagation rates irrespective of
the R ratio. However, outside the volume areas
Nominal stress amplitude (n,a), MPa
DKeff Kmax Kop
(b)
103
104
105
106
10
20
30
300
Upcut milling
rs = 234 MPa
200
Downcut milling
rs = 242 MPa
kt = 2.5
=5
0
104
105
106
107
Alternating bending S-N curves of notched

specimens of normalized plain carbon SAE
1045 steel after annealing, downcut milling and upcut milling. Source: Ref 4749
Fig. 31
103
104
105
106
10
20
30
40 50 60
Effective stress intensity range (Keff), MPa m
(d)
Crack propagation rate da/dN versus stress intensity range DK in the base material (a) and (c)
and versus effective stress intensity range DKeff (b) and (d)
in the welded zone of the SAE 1019 steel. Source: Ref 45
Fig. 30
300
200
=1
= 0.4
= 0.1
kt = 1.7, = 2
kt = 2.5, = 5
100
kt = 4.4, = 15
kt = 2.5, = 2
Smooth
Notched
0
750 500 250
250
500
750
rs
Surface residual stress (l ), MPa

Alternating bending fatigue strength of milled
smooth and notched specimens of normalized
plain carbon SAE 1045 steel versus surface residual stress.
Source: Ref 4752
Fig. 32
with high tensile residual stresses, individual

curves for each R value used are observed. For
R (pulsating compression), first increasing and then decreasing crack propagation rates
are observed indicating the influence of the tensile residual stress field. However, also in this
case, the crack propagation behavior can be described by a single, relatively narrow scatter
band if the influence of the welding residual
stresses on the DK values appearing at the crack
tip is taken into account (see Fig. 30d). In this
way, the crack propagation behavior in macro
residual stress fields can be quantitatively described if crack propagation data of macro residual-stress-free materials are available and the residual stress distribution is known.

S-N Curves
Characteristic Examples
Low-Strength Steel. Figure 31 shows S-N
curves for alternating bending of normalized
SAE 1045 steel (German grade Ck 45) (Ref 47
49). The notched specimens had a stress-concentration factor kt 2.5. The stress gradient at
the notch root dr/dz related to the maximum
stress r (normalized stress gradient g):
g
Annealed
rs = 0 MPa
100
40 50 60
Effective stress intensity range (Keff), MPa m
(c)
as a function of DK for different stress ratios R.

As expected, with increasing R values, higher
crack propagation rates are observed. It is important to note that a mode I crack can only grow
during that portion of loading cycle where the
crack is open. This portion is influenced by the
R value itself and the near-crack-tip residual
stress distribution. The effective stress intensity
range can be determined quantitatively as the
difference between the maximum stress intensity
and the stress intensity where the crack opens:
Bending fatigue strength (Rf), MPa
1 dr
r dz
z0
(Eq 17)
was 5 mm1. All data are nominal stress amplitudes and are valid for a failure probability of
50%. The bending fatigue strength was evaluated at an ultimate number of cycles Nu 107.
By downcut milling and upcut milling, surface
residual stresses of 242 and 234 MPa, respectively, were generated. The corresponding S-N
curves are almost identical. A third batch of
specimens was annealed 2 h at 700 C after
downcut milling. The annealing results in a reduction of bending fatigue life and bending fatigue strength (Ref 4749).
The alternating bending fatigue strengths of
milled smooth and notched specimens with different geometries are plotted in Fig. 32 as a function of the surface residual stresses (Ref 4652).
Again, all data are given for a failure probability
of 50%, and the bending fatigue strengths are
nominal stress amplitudes at Nu 107. With
increasing stress-concentration factor and decreasing stress gradient, the bending fatigue
strength decreases. The influence of the stressconcentration factor is clearly visible from a
comparison of the specimens with the same
value g 2 mm1, but different values kt
1.7 and 2.5, respectively. On the other hand, the
increase of g from 2 to 5 mm1 at specimens
with kt 2.5 results in a significant increase of
bending fatigue strength. It is also interesting to
note that specimens with kt 4.4, g 15 have
a somewhat higher strength than specimens with
300
280
260
Deep rolled
240
220
200
180
104
105
106
107
320
280
rs
260
240
220
200
180
104
1000
Deep rolled
300
(a)
2 20 m, 30 m/s
500
2 6 m, 15 m/s
0
2 3 m, 15 m/s
500
105
106
107
0.1
0.2
(b)
600
500
400
rs = 221 MPa
300
200
100
0
104
19
kt = 1.7
=2
602
105
106
107

specimens made from quenched-and-tempered (60 C/2 h) plain carbon SAE 1045 steel after different grinding processes. Source: Ref 4749
Fig. 34
Depth distribution of the residual stress in

notched specimens made from quenched-andtempered (600 C/2 h) plain carbon SAE 1045 steel by different grinding processes with the indicated two steps of
final feed and cutting speed. Source: Ref 4749
Fig. 35
S-N curves of specimens made from normalized plain carbon SAE 1015 steel in the asheat-treated state and after an additional deep rolling for
(a) push-pull loading and (b) rotating bending. After Ref 53
Fig. 33
with rather small and with compressive surface

residual stresses yield identical fatigue behavior.
It should be noted that the surface layer bearing
compressive residual stresses is rather small.
Contrarily, the generation of tensile residual
stresses causes a significant decrease of bending
fatigue strength and a rather small decrease of
finite fatigue life, which obviously vanishes at
high stress amplitudes.
The alternating bending fatigue strength of
ground smooth and notched specimen with different geometries made of the same material
state is plotted in Fig. 36 as a function of the
surface residual stresses (Ref 4752). Similar to
Fig. 32, the influences of the stress-concentration
factor and the normalized stress gradient are
clearly discernible. However, contrary to the
normalized steel there is distinct reduction of the
bending fatigue strength with increasing tensile
residual stress, this being more pronounced in
the comparison of smooth specimens with
notched specimens. In the range of compressive
residual stresses covered, the influence of residual stress on bending fatigue strength is rather
small.
The S-N curves of smooth specimens in the
ground state and after an additional shot peening
are compared in Fig. 37(a) (Ref 4749). There
is a distinct increase of the bending fatigue
strength by shot peening, but a rather small in-
400
= 0.4
Smooth
300
= 0.1
kt = 1.7
Notched
=2
200
100
0
500
=5
kt = 2.5
250
=2
0
250
500
750
rs
Alternating bending fatigue strength of ground

smooth and notched specimens made from
quenched-and-tempered plain carbon SAE 1045 steel versus surface residual stress. Source: Ref 4752
Fig. 36
320
machining procedures. After correction of the

data points given in Fig. 32 to the same hardness,
it turns out that the bending fatigue strength is
hardly changed or slightly diminished at most,
if the residual stresses change from compressive
to tensile (Ref 49).
S-N curves for push-pull loading of smooth
specimens made from normalized SAE 1015
steel (German grade Ck 15) in the as-heat-treated
state and after an additional deep rolling are
shown in Fig. 33(a) (Ref 53). Again, the fatigue
behavior in the range of finite fatigue life and the
fatigue strength do not differ much. Figure 33(b)
shows a plot of S-N curves for rotating bending
of the same material states. Now, by deep rolling
finite fatigue life is increased by one order of
magnitude or more, and the bending fatigue
strength is increased significantly.
Medium-Strength Steel. Figure 34 shows SN curves that were determined in alternating
bending on notched specimens of quenched-andtempered (600 C/2 h) SAE 1045 steel (Ref 47
49). All data are valid for a failure probability
of 50% and Nu 107. Three batches of specimens were manufactured by grinding. The grinding parameters (final feed, cutting speed) and the
resulting depth distributions of residual stresses
are given in Fig. 35 (Ref 4749). Specimens
600
500
Shot peened
400
300
Ground
200
100
0
104
105
106
107
(a)
kt 2.5, g 2 mm1. However, with regard

to the residual stress state, there is no significant
influence on the bending fatigue strength, even
though the range of residual stresses covered
comes to more than 1000 MPa regarding specimens with kt 4.4, g 15 mm1. Careful
inspection of the hardness of the specimens
tested shows that a positive slope of the lines in
Fig. 32 is not related to the changing (macro)
residual stress state, but to different hardness of
the specimen and, hence, differences in the micro residual stress state produced by different
600
500
Shot peened
400
Milled
300
200
Ground
100
0
104
(b)
105
106
107
Alternating bending S-N curves of specimens

made from quenched-and-tempered (600 C/2
h) plain carbon SAE 1045 steel. (a) Smooth specimens after
grinding and after additional shot peening. (b) Notched
specimens after milling, grinding, and grinding with additional shot peening. Source: Ref 4749
Fig. 37
400
Smooth
= 0.4
300
grinding. Again, in the range of finite fatigue life,

the influence of the different manufacturing processes is almost negligible. The relative increase
of the bending fatigue strength by shot peening
is more pronounced compared with smooth
specimens. It is interesting to note that milled
specimens have a higher bending fatigue
strength than ground ones, even though they
have lower compressive residual stresses (159
200
Notched
kt = 1.7
=2
100
0
750 500 250
250
500
750
rs
Maximum residual stress (l ), MPa

Alternating bending fatigue strength of
quenched-and-tempered (600 C/2 h) plain
carbon SAE 1045 steel versus surface residual stress evaluated from Fig. 36 and 37
Fig. 38
fluence on finite fatigue life. Figure 37(b) compares S-N curves of notched specimens which
were milled, ground, and shot peened after
800
600
400
Notched
Smooth
200
1000
2000
3000
4000
Deep-rolling force, N
5000
Bending fatigue strength smooth and notched

specimens made from quenched-and-tempered SAE 5135 steel versus deep-rolling force. After Ref
57
Alternating bending S-N curves of smooth

specimens made from blank-hardened AISI
5115 steel in the as-blank-hardened and with different conditions shot peened states including one with a subsequently electropolished surface. 1, as-blank-hardened; 2,
shot velocity v 23 m/s, coverage c 100%, mean diameter of the shot d 0.6 mm; 3, v 53 m/s, c 100%,
d 0.3 mm; 4, v 53 m/s, c 100%, d 0.6 mm; 5,
v 81 m/s, c 600%, d 0.6 mm; 6, v 53 m/s, c
100%, d 0.6 mm, 100 lm surface layer electrolytically removed. After Ref 5456
Fig. 39
Fig. 41
600
rs = 400 MPa
500
400
300
200
923
100
0
104
kt = 1.7
=2
105
106
107

specimens made from quenched plain carbon
SAE 1045 steel after different grinding processes. Source:
Ref 4749
Fig. 42
1000
200
0
200
rs
rs
400
3
4
5
400
600
800
1000
0
0.1
0.2
0.3
0.4
0.5
Depth distribution of the residual stress in specimens made from blank-hardened AISI 5115
steel in the as-blank-hardened (1) and in different conditions of the shot peened state (3), (4), and (5) corresponding
to Fig. 39. After Ref 5456
Fig. 40
2 15 m, 30 m/s
500
2 9 m, 15 m/s
0
2 3 m, 15 m/s
500
0.1
0.2
Depth distribution of the residual stresses in

notched specimens made from quenched plain
carbon SAE 1045 steel and ground with the two steps of
final feed and cutting speed indicated. Source: Ref 4749
Fig. 43
MPa) at the surface than the latter ones (221

MPa).
Figure 38 shows plots of the alternating bending fatigue strengths evaluated from Fig. 37 as a
function of the surface residual stresses. The arrows mark the shift in bending fatigue strengths
and surface residual stresses produced by shot
peening. Additionally, data points of ground
specimens with negligible or tensile residual
stresses shown in Fig. 36 are included. In the
case of notched specimens, all data points lie on
a common line with the slope 0.154 except for
ground specimens with compressive residual
stresses at the surface. Regarding smooth specimens, the influence of tensile residual stresses
on the bending fatigue strength is much more
pronounced than the influence of compressive
residual stresses.
Figure 39 shows S-N curves for alternating
bending of smooth specimens (g 1 mm1) of
blank-hardened AISI 5115 steel (German grade
16 MnCr 5) determined in the unpeened and
various shot peened states including one with
electrolytically removed surface layer (Ref 54
56). The corresponding depth distributions of residual stresses are given in Fig. 40. From the
comparison of both figures, it becomes clear that
the surface residual stress is not a suitable parameter for the assessment of the influence of the
various treatments on the fatigue behavior.
The influence of the deep-rolling force on the
rotating bending fatigue strength of smooth and
notched specimens made from quenched-andtempered SAE 5135 steel (German grade 37 CrS
4) is shown in Fig. 41 (Ref 57). In both cases, a
maximum of the fatigue strength occurs at a certain force. However, the increase of the fatigue
strength of notched specimens (kt 2) by deep
rolling is much more pronounced than that of
smooth specimens. In the end, the optimal bending fatigue strength of notched specimens
which is a nominal stress amplitudeis higher
than the bending fatigue strength of smooth
specimens.
High-Strength Steels. S-N curves for alternating bending of notched specimens of
quenched SAE 1045 steel in differently ground
conditions are compared in Fig. 42 (Ref 4749).
The grinding parameters (final feed, cutting
speed) and the resulting depth distributions of
residual stresses are given in Fig. 43. Similar to
ground quenched-and-tempered specimens, the
fatigue behavior of specimens with very small or
compressive residual stresses (which have a very
small penetration depth) hardly differ. Tensile
residual stresses, however, not only cause a
strong reduction of bending fatigue strength, but
also of finite fatigue life.
In Fig. 44, the bending fatigue strength that
was evaluated from Fig. 42 and corresponding
data from tests on smooth specimens are plotted
as a function of the surface residual stresses. The
negative influence of tensile residual stresses on
the bending fatigue strength of smooth specimens is even more pronounced compared to
notched specimens. The influence of compressive residual stresses generated by grinding is
800
Smooth
= 0.4
600
400
200
0
1000
Notched
kt = 1.7
=2
500
500
fatigue life, which comes up to one and a half

orders of magnitude at high stress amplitudes.
The S-N curves of notched specimens of the
same steel state after grinding and after additional shot peening with shot of different hardness are shown in Fig. 45(b). Compared to
smooth specimens, shot peening produces a
much stronger increase of the bending fatigue
strength. Again, there is also a remarkable in-
much smaller than the influence of tensile

stresses.
In Fig. 45(a), the S-N curves of smooth specimens in the ground state and after additional
shot peening are compared (Ref 4649). Similar
to quenched-and-tempered specimens (see Fig.
37), shot peening produces a significant increase
of the bending fatigue strength. Contrary to the
results of the medium-strength steel, however,
there is also a very pronounced increase of finite
1000
800
Smooth
= 0.4
600
400
Evaluation of Experimental Results
Ground
Shot peened
Milled
200
Notched
kt = 1.7
=2
0
1000
500
500
1000
rs
1500
rs
Fig. 44 Alternating bending fatigue strength of ground

smooth and notched specimens of quenched
plain carbon SAE 1045 steel versus surface residual stress.
Source: Ref 4749
Alternating bending fatigue strength of smooth

and notched specimens made from quenched
plain carbon SAE 1045 steel with different surface conditions versus surface residual stress
Fig. 46
Ratio Rf/Rh, MPa/m
900
Shot peened
800
700
Ground
600
10
4
1
500
400
104
200
400
600
800
Surface hardness, HV 5
105
106
107
(a)
Changes of alternating bending fatigue strength

DRf relative to changes of surface roughness
height DR h versus surface hardness HV 5 for different heat
treated SAE 1045 and AISI 5115 steels. After Ref 58, 59
Fig. 47
1000
900
0
5458 HRC
800
700
600
500
400
104
4650 HRC Shot peened

Milled
Ground
105
106
107
(b)
Alternating bending S-N curves of specimens
made from quenched plain carbon SAE 1045
steel. (a) Smooth specimens after grinding and after additional shot peening. (b) Notched specimens after grinding,
milling, and grinding with additional shot peening with
shot of the indicated hardness. Source: Ref 4649
Fig. 45
rs
SAE 1045
AISI 5115
40
100
1000
crease of finite fatigue life. Additionally, the SN curve of milled specimen is included. Finite
fatigue life and bending fatigue strength of these
specimens are lower compared to shot-peened
ones, even though they contain very high surface
compressive residual stresses of 1200 MPa.
In Fig. 46, the bending fatigue strength data
already plotted in Fig. 44 are complemented by
data evaluated from Fig. 45. The arrows mark
the shift in bending fatigue strengths and surface
residual stresses produced by shot peening.
Similar to the discussion of Fig. 39 and 40, it
becomes obvious that the magnitude of surface
residual stress is not a suitable parameter for the
assessment of the influence of shot-peening-induced residual stresses on the fatigue strength.
n,a = 360 MPa
200
325
400
280
200
600
185
0
10
102 103 104 105 106 107

Number of cycles, N
Surface residual stress in notched (kt 1.7)

upcut milled specimens made from normalized plain carbon SAE 1045 steel during alternating bending fatigue at different nominal loading amplitudes versus
number of cycles. Source: Ref 49
Fig. 48
Low-Strength Steels. Regarding Fig. 31 to

33, different machining processes or process parameters were applied to the specimens investigated producing different surface topographies.
In Fig. 47, the ratio DRf /DRh, which gives the
change of the alternating bending fatigue
strength of differently treated SAE 1045 and
AISI 5115 steels per lm increase of roughness
height is plotted as a function of the hardness at
the surface (Ref 58, 59). As is well known from
many other investigations, the susceptibility of
the bending fatigue strength to roughness increases with increasing hardness. For the steel
states investigated, a value of 2 MPa/lm can
serve as a guide. Since the roughness height of
the specimens did not vary by more than 2 lm,
the results plotted in Fig. 31 and 32 are hardly
influenced by surface topography. It should be
kept in mind, however, that a large increase of
surface roughness will significantly decrease the
fatigue strength even in low-strength material
states.
In low-strength steels, cyclic plasticity appears if the cyclic loading approaches the fatigue
strength. Extensive cyclic plasticity occurs in the
range of finite fatigue life, being combined with
cyclic softening and/or hardening processes (see
the section Some Aspects of Fatigue of Steels
in this article). Cyclic plasticity results in macro
residual stress relaxation, which is more rapid
the higher the cyclic loading is (see the article
Stability of Residual Stresses in this Handbook). Figure 48 (Ref 49) shows the relaxation
of the surface residual stresses during alternating
bending fatigue in notched upcut milled specimen made from normalized SAE 1045 steel
starting at an initial value of 590 MPa. The
bending fatigue strength amounts to 190 MPa
(see Fig. 32). During loading with an amplitude
of 185 MPa just below the fatigue limit, the
change of the macro residual stresses is almost
negligible. During all other fatigue loadings, extensive residual stress relaxation occurs. Therefore, almost no influence of macro residual
stresses is found in Fig. 31 and 32 (Ref 4752).
The amount of cyclic plasticity, which occurs
during cyclic loading resulting in a given finite

fatigue life or which corresponds to the fatigue
limit, decreases with increasing strength of the
steel. This results in decreasing residual stress
relaxation rates and finally in stable macro residual stresses. On the other hand, in a low-strength
steel, local work hardening may occur during
manufacturing processes such as milling, turning, or grinding. In other words, the micro residual stress state is changed, mainly by the increase
of the dislocation density. As discussed in the
article Stability of Residual Stresses in this
Handbook, the resistance of the micro residual
stress state against cyclic relaxation is much
higher compared to that of the macro residual
stress state. So one may assume that macro residual stresses that exist in a part of a component
that has undergone work hardening by manufacturing processes are more stable than expected
from the basic strength of the material. Looking
at Fig. 32, this relationship may be responsible
for the small reduction of the bending fatigue
strength by tensile residual stresses that occurs
in some of the specimen series investigated.
A much more important consequence of the
change of the micro residual stress state by local
work hardening is the direct alteration of the local fatigue behavior. Increasing strength reduces
the local cyclic plastic deformation at a given
external loading, thus increasing the number of
cycles to crack initiation and decreasing the

propagation rate of short cracks (Ref 60, 61).
Therefore, as is evident from Fig. 31, a residual
stress relief heat treatment that eliminates the
machining-induced work hardening reduces the
finite fatigue life and the fatigue limit of normalized SAE 1045 steel. At rather high loading
amplitudes, this influence vanishes because then
the cyclic deformation behavior is almost entirely governed by cyclic softening and hardening processes during the fatigue loading (see the
section Cyclic Deformation Behavior in this
article) irrespective of initial variations of the local work-hardening state.
A smaller influence of the micro residual
stress state appears in Fig. 33(a) and a somewhat
stronger one in Fig. 33(b). Here, it becomes clear
that even in a rather soft material state such as
normalized SAE 1015 steel, an improvement of
the fatigue behavior can be obtained from local
work hardening. In this case, the depth of the
surface layer of the smooth specimens influenced
by deep rolling is relatively large compared to
the milled specimens, which are the basis of Fig.
31 and 32. Nevertheless, push-pull loading (Fig.
33a) without stress gradient reveals a small influence of deep rolling, because of two major
effects. First, the crack initiation site may be
shifted below the work-hardened zone. Then, the
small increase of the fatigue strength mirrors the
benefit of omitting surface effects (such as spec-
600
Smooth
= 0.4
n,a = 200 MPa
rs
Ratio Rf/HV 5
Rh = 3 m
1
Rh = 15 m
0
0
200
400
600
800
1000
Surface hardness, HV 5
400
Notched
kt = 1.7, = 2
n,a = 150 MPa
200
kt n,a = 255 MPa
0
0
Changes of alternating bending fatigue strength

DRf relative to changes of surface hardness
DHV 5 versus surface hardness HV 5 for different surface
roughnesses heights Rh. After Ref 58, 59
10
Fig. 49
600
=2
kt = 2.5
=5
1.5
kt = 1.7
400
200
200
0
0
200
400
106 107
10
102
103
104
105
106
Number of cycles, N
600
rs
Fatigue notch factor kf of specimens made from

quenched-and-tempered (600 C/2 h) plain
carbon SAE 1045 steel versus surface residual stress.
Source: Ref 4749
Fig. 50
105
Notched
n,a = 350 MPa
kt = 1.7
=2
kt n,a = 595 MPa
=2
1.0
400
104
Smooth
= 0.4
n,a = 400 MPa
rs
2.0
103
(a)
Fatigue notch factor, kf
2.5
102
Number of cycles, N
(b)
Surface residual stress in smooth and notched
specimens made from quenched-and-tempered (600 C/2 h) plain carbon SAE 1045 steel versus
number of cycles (a) at a lifetime of approximately 2 106
cycles and (b) at a lifetime of approximately 105 cycles.
Source: Ref 4749
Fig. 51
imen roughness, corrosion, oxidation) on the

crack initiation. Secondly, if crack initiation
occurs at the surface, the influence of the workhardened zone may also be rather small, becauseat a given loadinga stress redistribution may occur from the softer core region of the
specimen to the work-hardened zone. This effect
is combined with cyclic plasticity in the core region and, hence, with rather high loading amplitudes. The first effect predominates in the range
of the fatigue limit, where cyclic plasticity is
small, and the second one in the range of finite
fatigue life. On the other hand, during rotating
bending (Fig. 33b) with a distinct stress gradient,
crack initiation always occurs at the surface. If
the work-hardened zone is thick enough, there is
no stress redistribution from the core region
which is then purely elastically loadedto the
work-hardened zone. Again, at very high loading
amplitudes, the cyclic softening/hardening processes produced by the cyclic loading itself push
the influence of the initial work-hardening state
into background.
In Fig. 49, the ratio DRf /DHV, which gives
the increase of alternating bending fatigue
strength per unit hardness, is plotted as a function of the hardness itself (Ref 58, 59). The diagram shows that the sensitivity of the fatigue
strength to the hardness decreases with increasing hardness. Taking into account that the resistance of the macro residual stress state against
cyclic relaxation is low in low-strength steels, it
becomes clear from Fig. 49 thatbesides the
change of the surface topographythe alteration of the micro residual stress state is the governing parameter for the assessment of the influence of machining on the fatigue behavior of
components made from low-strength steel.
Medium-Strength Steel. A detrimental effect
of tensile grinding residual stresses appears in
Fig. 34 and 36 on alternating bending fatigue
strength, being more pronounced for smooth
than for notched specimens. Therefore, the fatigue notch factor decreases with increasing residual stress, as shown in Fig. 50. There is relaxation of surface residual stresses during
bending fatigue loadings of smooth and notched
specimens, which results in lifetimes of approximately 2 million cycles, as shown by Fig. 51(a).
The stress relaxation in the notched specimens
is significantly more pronounced than in smooth
ones. An elastic estimation of the stress in the
notch root r* kt rn,a yields a much higher
value than the stress amplitude in the smooth
specimens. Even though r* is only an upper
bound for the true stress at the notch root, it becomes clear that it is the higher local cyclic loading in notched specimens that causes the stronger
cyclic stress relaxation and, hence, the smaller
residual stress sensitivity of the bending fatigue
strength compared to smooth specimens. Regarding the relatively high loadings indicated in
Fig. 51(b), stress relaxation in notched specimen
is almost complete at the end of fatigue life, and
therefore only a small influence of initial macro
residual stresses on fatigue life at high loadings
is expected. This is shown in Fig. 34.

According to Fig. 34 and 36, the beneficial
effect of compressive grinding residual stresses
on bending fatigue strength is small regarding
smooth and almost negligible regarding notched
specimens. Again, the smaller effect in notched
specimens is due to a stronger residual stress relaxation. However, during cyclic loading corresponding to the fatigue limit, a considerable portion of the initial residual stresses is retained to
the ultimate number of cycles even in notched
specimens. Therefore, and in view of the small
penetration depth of the compressive residual
stresses (
20 lm, see Fig. 35), one would assume that the crack initiation site is shifted below the surface. However, this could not be
proved unambiguously in either smooth or
notched specimens (Ref 49). Therefore, it is
thought that very thin surface zones bearing
compressive residual stresses do not retard the
initiation and early growth of microcracks.
Regarding smooth specimens in Fig. 38, a
smaller effect of compressive residual stresses
produced by grinding or shot peening appears
compared to tensile grinding residual stresses,
even though crack initiation occurred always at
the surface. The small influence of compressive
rs
n,a = 400 MPa
500
N 107
Before loading
1000
1500
0
0.1
0.2
0.3
0.4
0.5

Residual stress of smooth ground and additional shot peened specimens made from
quenched-and-tempered (600 C/2 h) plain carbon SAE
1045 steel before loading and after 107 cycles versus distance from surface. Source: Ref 4649
Fig. 52
rs
0
n,a = 460 MPa
N 103
n,a = 300 MPa
500
N 107
Before loading
1000
1500
0
0.1
0.2
0.3
0.4
0.5

Residual stress of notched ground and additional shot peened specimens made from
1045 steel before and after loading at different nominal
stress amplitudes versus distance from surface. Source: Ref
4649
Fig. 53
grinding residual stresses was already attributed

to their small penetration depth. Contrarily, as
evident from Fig. 40, compressive residual
stresses generated by shot peening have a rather
large penetration depth. However, since the
bending fatigue strength of shot peened specimens (414 MPa) is much higher than that of
specimens with tensile grinding residual stresses
(195 MPa), also the residual stress relaxation is
more pronounced during corresponding cyclic
loadings. As shown in Fig. 52, residual stresses
relax significantly during fatigue loading, which
does not cause failure within 107 cycles. Furthermore, the surface roughness of the shot
peened specimens (Rh 35.6 lm) was much
higher than that of ground specimens (Rh 6.5
lm) (Ref 4649). If the bending fatigue strength
of shot peened specimens is corrected for the
high surface roughness by means of Fig. 47 and
is plotted in Fig. 38 at the surface residual stress
remaining after 107 cycles (see Fig. 52), one gets
a data point that lies significantly above the extrapolated relationship between bending fatigue
strength and tensile residual stresses.
The relationship between bending fatigue
strength and surface residual stresses of notched
specimens (Fig. 38) is described by a straight
line except for specimens with compressive
grinding residual stresses as already discussed.
In all cases, crack initiation occurred at the surface of the specimens that failed. Since the extrapolated straight lines for smooth and notched
specimens cross at a tensile surface residual
stress of approximately 750 MPa the fatigue
notch factor approaches a unity with increasing
tensile residual stress (see the paragraphs on medium-strength steel in the section Haigh Diagram in this article). Again, different residual
stress relaxation plays an important role. As
shown in Fig. 51(a) (Ref 49), and discussed previously, the residual stress relaxation in a smooth
specimen loaded just above the fatigue limit,
which failed after 2 million cycles, is significantly smaller than in a notched specimen that
failed after approximately the same number of
cycles. Figure 53 proves that residual stress relaxation is even more significant up to a distance
of 100 lm from the surface during fatigue loading of shot peened notched specimens, which did
not fail up to 107 cycles. During fatigue loading
resulting in a fatigue life of approximately 104
cycles, residual stress relaxation is almost complete. Therefore, no influence of macro residual
stresses appears in the low-cycle fatigue range
(see Fig. 37b).
From Fig. 51(a) and 53, it follows that at the
fatigue limit, the residual stress relaxation in shot
peened notched specimens with compressive
tensile residual stresses is more pronounced than
in ground notched specimens with tensile residual stresses. For this comparison, in Fig 51(a)
only the stress relaxation up to the appearance
of a crack (arrow at 5 105 cycles) may be
regarded, because the final relaxation is influenced by crack propagation and does not exist
in Fig. 53. Furthermore, the surface roughness
of shot peened notched specimens (Rh 18.0
lm) was larger than that of ground specimens

(Rh 5.5 lm). Therefore, the question arises
why the data points of shot peened specimens do
not fall below the solid line drawn in Fig. 38.
Careful inspection of microhardness distribution
and the width of x-ray interference lines as a
function of the distance from the surface show
thatcompared to grindingshot peening produces a more intense work hardening which, in
addition, penetrates into a larger depth. The same
holds for smooth specimens. The work hardening reduces cyclic plasticity and retards crack
initiation. Hence, it becomes clear that the bending fatigue strength of medium strength steel is
strongly influenced by both the micro and the
macro residual stress states and their stability
against cyclic loading. However, in the low-cycle fatigue (LCF) range, the macro residual stress
state is unstable and the influence of the (more
stable) micro residual stress state vanishes in
view of the strong cyclic-softening processes occurring during this loading condition (see the
section Cyclic Deformation Behavior in this
article). Therefore, neither an influence of the
macro residual stress state nor an influence of
the micro residual stress state appears in Fig. 37
at high loading amplitudes.
Figure 40 shows that rather different depth
distributions of compressive residual stresses
can be produced in smooth specimens (g 1
mm1) made from a medium-strength steel by
shot peening with different peening parameters
(see also the article Stability of Residual
Stresses in this Handbook). To understand the
influence of the surface roughness, the work
hardening and the macro residual stress state
produced by shot peening on the resulting bending fatigue behavior it is necessary to discuss
crack initiation and crack propagation separately, and this is done in the section Propagating and Nonpropagating Cracks in this article.
Some relationships, however, should be pointed
out here. The highest bending fatigue strength
(curve 6 in Fig. 39) was determined for specimens shot peened according to curve 4 in Fig.
40. Then, a 100 lm thick surface zone was electrolytically removed, shifting the maximum
compressive residual stresses to the surface and
reducing the surface roughness. The smooth surface (Rh 3 lm) and the work hardening of the
surface zone oppose crack initiation. The relatively high compressive residual stresses at the
surface hinder microcrack propagation. This influence outweighs the larger maximum and the
greater penetration depth of the compressive residual stresses according to curve 5. In this shot
peening state, a higher surface roughness (Rh
43.8 lm) and a lower work hardening at the surface result in crack initiation at a lower stress
amplitude, and the lower surface compressive residual stresses offer less resistance to early microcrack propagation. On the other hand, in the
range of finite fatigue life, there is almost no advantage of the electropolished state 6 compared
to the initial shot peened state 4 and a clear disadvantage compared to state 5. Here, the higher
number of cycles to crack initiation (state 6) is

almost compensated (state 4) or overbalanced
(state 5) by a slower rate of crack propagation
due to the larger penetration depth andin the
case of state 5the higher amount of the compressive residual stresses. Since the surface
roughness, the surface work-hardening state and
the surface residual stress state (see Fig. 40) of
the material states 2 to 5 do not differ much, they
have similar resistance to crack initiation at the
surface. Hence, the different fatigue strength appearing in Fig. 39 is determined by the appertaining resistance to microcrack propagation,
which, in turn, depends on the depth distribution
of the compressive residual stresses. This is discussed in more detail in the section Propagating
and Nonpropagating Cracks in this article.
A very interesting example of the interaction
of changing surface roughness as well as micro
and macro residual stress state in a mediumstrength steel are given in Fig. 41. The strong
increase of the rotating bending fatigue strength
of notched specimens by deep rolling is clearly
attributed to the generation of micro- and macrostresses with a rather large penetration depth. It
is interesting to note that the resistance to microcrack initiation is hardly influenced by the deeprolling treatment (Ref 57), and the increasing
bending fatigue strength is almost entirely combined with an increasing resistance to micro-
Fatigue notch factor, kf
2.5
2.0
kt = 1.7
=2
1.5
1.0
500
500
1000
rs

Fatigue notch factor kf of specimens made from
quenched plain carbon SAE 1045 steel versus
surface residual stress. Source: Ref 4749
Fig. 54
rs
0
Before loading
n,a = 750 MPa
500
N 107
1000
1500
0
0.1
0.2
0.3
0.4
0.5
Residual stress of smooth ground and additional shot peened specimens made from
quenched plain carbon SAE 1045 steel before loading and
after 107 cycles versus distance from surface. Source: Ref
4649
Fig. 55
crack propagation, resulting in crack arrest at

stress amplitudes increasing with the deep-rolling force. At relatively high forces, however, the
bending fatigue strength decreases, because then
the impairment of the surface zone by the strong
increase of the surface roughness and microcracking as well as an unfavorable residual stress
state at the surface enhances the formation of
rather long initial cracks that only arrest at reduced stress amplitudes. The bending fatigue
strength of smooth specimens increases only
weakly with increasing rolling force. Regarding
the maximum value, an increase of only 15% is
achieved, compared to 120% in the case of
notched specimens. Since the stress gradient is
rather small in smooth specimens, there is some
speculation (Ref 56, 57) that the crack initiation
site is shifted below the surface by deep rolling.
Then, the small increase of the fatigue strength
mirrors the benefit of omitting surface effects
(such as specimen roughness, corrosion, oxidation) and the effect of a decreasing local stress
amplitude on the crack initiation. This mechanism is discussed in detail in the section Haigh
Diagram in this article. At high rolling forces,
however, the crack initiation occurs at the surface again, where it is enhanced by a high roughness, an increased roughness sensitivity due to
work hardening, and an unfavorable residual
stress state. Therefore, the bending fatigue
strength decreases, and there is no improvement
by deep rolling at all. In the end, the maximum
bending fatigue strength of the notched specimens (remember, evaluated as nominal stress
amplitude) is higher than the maximum bending
fatigue strength of the smooth specimens. These
results show that the gradients of the micro and
the macro residual stresses in comparison to the
gradient of the loading stresses are of great importance, and this will become even more evident in the next section.
High-Strength Steels. The S-N curves of
quenched SAE 1045 steel in Fig. 42 prove that
there is a detrimental effect of tensile grinding
residual stresses on bending fatigue strength and,
contrary to the quenched-and-tempered state (see
Fig. 34) also on finite fatigue life. This is because
there is only a weak relaxation of the initial residual stresses even at rather high stress amplitudes. Similar to quenched-and-tempered specimens, compressive grinding residual stresses
with very small penetration depth (
20 lm, see
Fig. 43) have almost no effect on bending fatigue
strength and finite fatigue life. Therefore and
since in the quenched state the compressive residual stresses are retained to the largest part during fatigue loading in the range of the fatigue
limit, one may expect that crack initiation occurs
below the surface. Analogous to the quenchedand-tempered state, this could not be proved definitely (Ref 49). Since in the quenched state a
grinding-induced work softening occurs close to
the surface and an influence of the surface roughness always exists, it is very probable that crack
initiation occurs at the surface. Then, one must
conclude that compressive residual stresses with
a penetration depth of only 20 lm do not retard
or stop the propagation of microcracks existing

after crack initiation.
The extrapolated bending fatigue strength versus surface residual stresses curves in Fig. 44
cross in the range of tensile residual stresses at
1125 MPa. This means that the fatigue notch
factor becomes unity. On the other hand, at negligible or compressive residual stresses, the fatigue notch factor approaches the stress-concentration factor, as shown in Fig. 54. Qualitatively
similar to the quenched-and-tempered state, the
influence of tensile residual stresses is stronger
than that of compressive residual stresses and the
residual stress sensitivity of notched specimens
is smaller than that of smooth ones. However,
this cannot be explained by different residual
stress relaxation so far. Careful x-ray analyses
prove that in notched specimens loaded with a
stress amplitude of 450 MPa and in smooth specimens loaded with 750 MPa (corresponding to
the fatigue limit of each) the relaxation of compressive residual stress is small and of equal
amount. This is not astonishing, since an elastic
estimation of the stress amplitude in the notch
root (r* kt r 765 MPa) yields almost the
same value as the stress amplitude of the smooth
specimen. No corresponding measurements for
smooth and notched specimens with tensile residual stresses were performed. However, since
the fatigue limit is reduced by tensile residual
stresses, it is clear that their relaxation is negligible.
A comparison of Fig. 45(a) with Fig. 37(a)
shows that the shot peening treatments performed increases the bending fatigue strength of
smooth specimens in the quenched state less
than in the quenched-and-tempered state. In
view of the different residual stress states and
their different stabilities in both heat treating
states, this finding is rather astonishing. In Fig.
55, the depth distribution of residual stresses in
shot peened quenched specimens before fatigue
loading and after 107 cycles at the fatigue limit
is shown. In the range of specimen scatter, the
results are identical. Contrary to quenched-andtempered specimen (see Fig. 52), the shot peening residual stresses are much higher and much
more stable in quenched specimens. However,
in the former, crack initiation occurs at the surface. Hence, the shot-peening-induced changes
of the micro and the macro residual stress state
have a strong influence on the measured bending
fatigue strength. In smooth quenched specimens,
cyclic loading at the fatigue limit initiates cracks
below the surface. Therefore, the influence of the
shot-peening-induced changes of the residual
stress state have a rather small influence on the
measured bending fatigue strength. This is
treated more detailed in the section Modeling
the Influence of Residual Stresses on Fatigue Behavior. Contrarily, in the range of finite fatigue
life, crack initiation in both states occurs at the
surface, and there is a strong influence of shot
peening regarding quenched specimens and a
very small one regarding quenched-and-tempered specimens. This finding is directly correlated with very different residual stress relaxa-
tion. While surface residual stress relaxation is

definitely complete in quenched-and-tempered
specimens after 50,000 cycles at a stress amplitude of 570 MPa, there is hardly any change of
the surface residual stresses in the quenched
specimens after the same number of cycles even
at 1200 MPa (Ref 49), although there may be
some residual stress relaxation below the surface.
Regarding notched specimens, a different picture appears (compare Fig. 45b with Fig. 37b).
Now, shot peening of quenched specimens pro-
800
Smooth
= 0.4
600
400
Ground
Shot peened
Milled
200
Notched
kt = 1.7
=2
0
1500 1000 500
0
500
1000
rs
Maximum residual stress (max ), MPa
Alternating bending fatigue strength of smooth

and notched specimens made from quenched
plain carbon SAE 1045 steel with different surface conditions versus surface residual stress
Fig. 56
Before loading
500
rs
n,a = 640 MPa
N 107
1000
1500
0
0.1
0.2
0.3
0.4
0.5
(a)
after
107 cycles. In this case, a significant relaxation and redistribution of the residual
stresses occur.
Modeling the Influence of Residual

Stresses on Fatigue Behavior
Haigh Diagram
Basic Relationships. One obvious way to account for the influence of (macro) residual
stresses on the fatigue behavior is to treat them
as local mean stresses. In doing so, one has to
realize that there are several important differences between (loading) mean stresses and residual stresses, as discussed in the section Comparison between Loading Stresses and Residual
Stresses in this article. Figure 58 shows a Haigh
diagram for smooth and notched specimens
made from a medium-strength steel (Ref 1, 62).
The Goodman approximation (see Eq 9) is used
to account for the influence of residual stresses
on the fatigue strength. If the amount of the minimum stress or the maximum stress in smooth
specimens does not exceed the critical stress amplitude ra,crit, which is a function of the cyclic
yield strength (see the article Stability of Residual Stresses in this Handbook), the residual
stresses do not relax, and the line AB gives the
influence of the residual stress on the fatigue
strength. Then, all combinations of residual
stress and stress amplitude inside the shaded area
result in neither residual stress relaxation nor fatigue failure. However, if the amount of the minimum stress or the maximum stress exceeds the
critical stress amplitude, it is assumed that the
residual stresses relax to the value given by the
points A and B, respectively, and the fatigue
strength remains constant at the value given by
these points. In the case of notched specimens,
the cyclic yield strength and the notch fatigue
strength (both in terms of nominal stress amplitudes) are less than the respective values of
smooth specimens. However, the ultimate tensile
strength of notched specimens is larger than that
of smooth ones in such a material state because
Before loading
500
rs
duces a strong increase of both fatigue limit and

finite fatigue life. Milling, which may generate
rather high compressive residual stresses, is less
effective. In all cases, crack initiation was observed at the surface. However, as evident from
Fig. 46, the surface residual stress state is not an
appropriate parameter to assess bending fatigue
life in the presence of compressive residual
stresses. In Fig. 56, the bending fatigue strength
is plotted as a function of the maximum residual
stresses. Again, the arrows mark the shift of the
residual stresses and bending fatigue strength
obtained from shot peening. Now, the data of
ground notched specimens with tensile or negligible residual stresses and of notched specimens peened with shot of different hardness are
described satisfactorily by one common line
covering a residual stress range of 2350 MPa.
The data points of ground and those of milled
specimens bearing compressive residual stresses
fall below this line, which has the slope 0.214.
In Fig. 56, it becomes even more clear than in
Fig. 46 that the improvement of the bending fatigue strength of smooth specimens by shot
peening is very limited. As a consequence, the
notch effect on bending fatigue strength is completely eliminated at high compressive residual
stresses.
It is interesting to look at the stability of very
high compressive residual stresses. As already
shown in Fig. 55, fatigue loading of smooth
specimens in the range of the fatigue limit does
not cause any significant change of the residual
stress state. As shown in Fig. 57(a), the same
holds for notched specimens peened with shot of
relatively low hardness, which have a very similar depth distribution of the residual stresses.
Elastic calculation of the minimum stress 70 lm
below the notch root (at the maximum of the
compressive stresses) yields the value 2035
MPa. This stress value may occur more than 107
times without effecting any change of the residual stress state. As shown in Fig. 57(b), notched
specimens peened with shot of higher hardness
bear even higher compressive residual stresses.
In this case, elastic calculation of the minimum
stress in a depth of 100 lm yields the initial
value 2475 MPa and the value 2130 MPa
n,a = 760 MPa
1000
N 107
1500
0
0.1
0.2
0.3
0.4
0.5
(b)
Residual stress of notched ground and additional shot peened specimens made from
quenched plain carbon SAE 1045 steel before loading and
after 107 cycles versus distance from surface. (a) Shot
peened with shot of a hardness of 46 to 50 HRC. (b) Shot
peened with shot of a hardness of 54 to 58 HRC. Source:
Ref 4649
Fig. 57
Fig. 58
Haigh diagram. Bending fatigue strength Rf of smooth and notched specimens made from a medium-strength
steel versus residual stress. After Ref 1, 62

of the triaxial stress state in the interior of the
notched specimens. Now, the Goodman relationship holds between points C and D, and residual
stress relaxation occurs outside the lightly
shaded area. From these relationships, it is expected that the residual stress sensitivity of
notched specimens is less than that of smooth
specimens.
Application to a High-Strength Steel. Figure 59 shows Haigh diagrams for smooth and
notched specimens made from a high-strength
steel, which are adopted to Fig. 44, 46, and 56.
In Fig. 59(a), the data points give the correlation
(a)
(b)
Haigh diagrams. Bending fatigue strength Rf of
quenched plain carbon SAE 1045 steel with different surface conditions. (a) Versus surface residual stress. (b) Versus
maximum residual stress
Fig. 59
between the residual stress and the nominal

stress amplitude at the surface. In Fig. 59(b), the
data points give the correlation between the maximum residual stress and the nominal stress amplitude at the locus of the maximum residual
stress. In the range of tensile residual stresses,
this is always the surface. The arrows connecting
some data points illustrate the residual stress relaxation that occurs during bending fatigue loading. The ultimate tensile strength Rm 1910
MPa of smooth specimens was taken from Ref
49. The ultimate tensile strength Rm,no of notched
specimens is unknown, but the value 2000 MPa,
which results from an extrapolation of the data
given in Fig. 44, is reasonable for this highstrength material state and the stress-concentration factor 1.7 (Ref 63, 64). The cyclic yield
stresses of smooth and notched specimens are
also unknown. However, using the results shown
in Fig. 55 (smooth specimens) and Fig. 57(a)
(notched specimens), a border line for residual
stress relaxation can be obtained. As obvious
from the figures and discussed in the previous
section, the cyclic loading indicated there causes
no residual stress relaxation. The corresponding
data points are given in Fig. 59(b) as open square
(smooth specimens) and open circle without arrow (notched specimens). Assuming that residual stress relaxation starts at a limiting amount
of the minimum or maximum stress irrespective
of the fractions of the mean stress and the stress
amplitude, one gets the shaded areas in which
the residual stresses are stable. These areas are
transferred to Fig. 59(a), too. The two open circles connected with an arrow illustrate the
change of the amount of the surface residual
stress (Fig. 59a) and of the minimum stress (Fig.
59b) during the cyclic loading of notched specimens indicated in Fig. 57(b). In milled specimens, there is also some residual stress relaxation as indicated by the triangles connected with
an arrow. Obviously, the amounts of the minimum stresses are reduced to a value corresponding to the border line, confirming this estimation.
Extrapolation of this line to zero residual stress
yields the values 1785 MPa (smooth specimens)
and 1660 MPa (notched specimens), respectively, which are higher than the corresponding
yield strengths (Ref 49).
Even though there are some uncertainties in
establishing the relationships in Fig. 59, the following conclusions can be drawn for a highstrength steel with tensile residual stresses:
Residual stress relaxation has no influence re-
Haigh diagram. Bending fatigue strength Rf of

1045 steel with different surface conditions versus surface
residual stress
Fig. 60
garding the effect of tensile residual stresses

on the fatigue strength.
The influence of tensile residual stresses on
the bending fatigue strength corresponds satisfactorily to the Goodman approximation.
This means that the residual stress sensitivity
and the mean stress sensitivity of the bending
fatigue strength is almost the same.
In this formalism, the different residual stress
sensitivity of smooth and notched specimens,
which causes a reduction of the fatigue notch
factor with increasing residual stress (see Fig.
54), is related to the different stress states in
smooth and notched specimens via the different ultimate tensile strengths.
In the range of compressive residual stresses,
the relationships are much more difficult. The
fatigue strength of smooth specimens is much
less than expected by the Haigh diagrams, regardless of whether the existing surface or the
maximum residual stresses are concerned. In the
case of shot peened specimens, this finding is
clearly related to subsurface crack initiation.
Therefore, neither the stress state at the surface
(Fig. 59a) nor the stress state in the depth of the
maximum compressive stresses (Fig. 59b) is directly relevant for the fatigue strength. Regarding notched specimens, the fatigue strength of
shot peened states is much higher than expected
by the Haigh diagram, if it is correlated with the
surface residual stresses (Fig. 59a). This simply
means that crack initiation must occur at the surface, if the specimens are loaded with amplitudes
in the range of the fatigue limit. On the other
hand, the stress amplitude in the depth of the
maximum of the compressive stresses of shot
peened specimens is somewhat lower than expected from the Haigh diagram (Fig. 59b). The
same holds for the data points of milled specimens, which have maximum compressive residual stresses at the surface. These findings can
only be understood if crack initiation and crack
propagation or crack arrest are treated separately.
This is done in the sections Concept of Local
Fatigue Strength and Propagating and Nonpropagating Cracks.
Application to a Medium-Strength Steel. In
medium-strength steels, residual stress relaxation definitively influences the relationships in
the Haigh diagram. Since the manufacturing process, which generates the macro residual
stresses, almost always changes the micro residual stresses and, hence, the local cyclic yield
strength, there are no unique borderlines for the
onset of residual stress relaxation as assumed in
Fig. 58. Nevertheless, Fig. 60 tries to bring the
results shown in Fig. 36 and 38 into a Haigh
diagram.
Regarding smooth specimens, the ultimate
tensile strength was taken from Ref 49. In
ground specimens with tensile residual stresses,
stress relaxation (marked by an arrow) brings the
corresponding data point rather close to the
Goodman line. The border lines for the onset of
residual stress relaxation are constructed using
this data point. However, these border lines are
not valid for shot peened specimens, in which
the residual stresses relax to a much smaller extent than expected as shown by the lower arrow.
Obviously, the resistance to residual stress relaxation is enhanced by shot peening, which results
in a stronger work hardening and, hence,
stronger increase of micro residual stresses than
grinding. Work hardening, however, should also
increase the fatigue strength at zero residual
stress and, hence, should produce deviations
from the solid line established with the aid of the
data points of ground specimens. However, as
already discussed in the paragraphs on medium-
SWT parameter, MPa
strength steel in the section Evaluation of Experimental Results, there is a detrimental effect
of a very high surface roughness. The second
arrow illustrates the correction, which is based
on Fig. 47, for this effect. Compressive residual
stresses produced by grinding show some relaxation, although their initial value is inside of the
border lines. This finding is consistent with the
consideration given so far, since work hardening
is almost negligible in this specimen. Moreover,
the data point corresponding to the relaxed compressive grinding residual stresses still falls below the Goodman line. This finding was already
attributed to the low penetration depth of the
grinding residual stresses.
The residual stress sensitivity of notched specimens with kt 1.7 is significantly smaller than
that of smooth specimens, even though the consideration is based on the relaxed residual
stresses. Therefore, the different residual stress
sensitivity of the fatigue strength of smooth and
notched specimens, respectively, is not solely
based on different residual stress relaxation, and
the third conclusion given previously for a highstrength steel is also valid for a medium-strength
steel to some extent. From an extrapolation of
the Goodman line plotted, an ultimate tensile
strength of 1133 MPa is obtained, which is significantly higher than the ultimate tensile
strength of smooth specimens. Again, this finding results from the triaxial stress state in the
interior of the notched specimens and is consistent with data measured on other mediumstrength steels (Ref 63, 64). Regarding material
states with compressive residual stresses, the
data point of milled specimens comes rather
close to the Goodman line. No measurements
were done concerning the stability of this residual stress state. Since the data point falls inside
of the border lines, the residual stress state
should be stable. Contrarily, there is some relaxation of the compressive residual stresses in
ground specimens even though their initial value
falls inside of the border lines, too. Again, this
is a consequence of different micro residual
stress states. Similar to smooth specimens, the
data points fall below the Goodman line as a
result of the small penetration depth of the com-
1000
800
SAE 1045
600
400
AISI 304
300
200
100
103
Untreated
Shot peened
Deep rolled
105
106
104
107
Smith-Watson-Topper parameter versus number of cycles to failure for specimens made

from normalized SAE 1045 and AISI 304 steel in different
surface conditions. After Ref 28
Fig. 61
pressive grinding residual stresses. The same

holds for notched specimens with kt 2.5, g
5. The data points of shot peened specimens fall
above the expected relationship as a consequence of pronounced work hardening, which
increases both fatigue strength and resistance of
the compressive residual stresses against relaxation.
From the relationships shown in Fig. 60, the
following conclusions can be drawn for a medium-strength steel:
Residual stress relaxation plays a significant
role for the influence of residual stresses on

fatigue behavior.
Residual stress relaxation depends on the
work hardening produced by the various
manufacturing processes. Not only the
amount, but also the penetration depth of
work hardening is important. As a consequence, even for one initial material state
there are no unique borderlines for the onset
of residual stress relaxation.
If comparable work-hardening states (or micro residual stress states) are concerned, the
influence of relaxed residual stresses on the
bending fatigue strength corresponds satisfactorily to the Goodman approximation. This
means that the relaxed residual stress sensitivity and the mean stress sensitivity of the
bending fatigue strength is almost the same.
In this formalism, the different initial residual
stress sensitivity of smooth and notched specimens, which causes a reduction of the fatigue
notch factor with increasing residual stress, is
related to both the different residual stress relaxation and the different stress states in
smooth and notched specimens via the different ultimate tensile strengths.
Damage Parameters
As discussed in the section Lifetime Behavior, there are some so-called damage parameters
that correlate fatigue life with loading parameters. The well-known Manson-Coffin relationship (Eq 5) is not applicable to material states
with residual stresses, because it does not account for mean stresses. On the other hand, the
Ostergren parameter (Eq 12) and the Smith-Watson-Topper parameter (Eq 11) include the mean
stress via the maximum stress. With the limitations already outlined in the previous section,
one could try to use these parameters to assess
the effect of residual stresses on fatigue life.
However, the application of the Ostergren parameter requires the knowledge of the local
plastic strain amplitude, which in turn needs sophisticated modeling of the local cyclic elasticplastic deformation behavior. This seems not to
be a useful approach. However, the Smith-Watson-Topper parameter (SWT parameter) is easier
to evaluate and was used in Ref 28 to account
for surface treated steels (normalized SAE 1045
and AISI 304). Figure 61 shows characteristic
plots of the SWT parameter versus the number
of cycles of stress-controlled tests on both steels
in shot peened as well as deep-rolled conditions

in comparison with the untreated states. For each
steel, a unique curve is obtained for the three
material conditions. Hence, concerning these
steels the quantities that are included in the SWT
parameterthe maximum stress and the total
strain amplitude (see Eq 11)form a reliable
basis for the assessment of the influence of residual stresses on the fatigue life.
Concept of Local Fatigue Strength

Basic Relationships. As shown in the previous section, in the paragraphs on applying the
Haigh diagram to high-strength and mediumstrength steels, the use of the Haigh diagram for
the assessment of the effect of residual stresses
on the fatigue behavior has serious limitations.
Regarding, for example, smooth specimens
made from a high-strength steel, shot peening
produces much less improvement in fatigue
strength than expected from the residual stress
state. This finding is combined with subsurface
crack initiation. Regarding medium-strength
steel, difficulties arise because not only the
(macro) residual stress state is changed by manufacturing processes, but also the micro residual
stress state. The latter influences the fatigue
strength at zero mean stress and the resistance of
the residual stress state against relaxation. Therefore, it is appropriate to look for the local fatigue
strength, which depends on the local micro residual stress state, the local macro residual stress
state, and (as far as crack initiation at the surface
is concerned) on the topography of the surface
of the component. Strictly speaking, the local resistance against fatigue crack initiation is concerned, as shown later.
The concept of the locally effective fatigue
strength, which has its origin in Ref 38 and 39,
enables quantitative predictions of the effect of
the depth distributions of residual stresses on the
locus of crack initiation as well as on the fatigue
strength. The basic assumption of the concept is
that a crack can only be initiated at or below the
surface if the local loading stress exceeds the local fatigue strength. Especially in the case of
relatively hard materials, for example, hardened
steels, this concept yields to a good estimation
of the corresponding properties. For that purpose, it is necessary to have a good knowledge
of the depth distributions of the fatigue strength
in the residual-stress-free condition R0f (z) as well
as of the macro residual stress rrs(z) and the residual stress sensitivity m(z). The locally effective fatigue strength Rf(z) as a function of the
distance x from the surface is calculated by the
relationship:
Rf(z) R0f (z) m(z) rrs(z)
(Eq 18)
where the residual stress sensitivity m(z) approaches the mean stress sensitivity M of the
Goodman relationship (e.g., Eq 9) if the residual
stresses are stable. Then, the residual stress sensitivity is determined approximately by m(z)

R0f (z)/Rm(z) (Ref 39). However, if residual stress
relaxation occurs, the residual stress sensitivity
m is smaller than the mean stress sensitivity M,
if the initial residual stress distribution is used
(Ref 39, 62). However, if the relaxed residual
stress distribution is used, the residual stress sen-
sitivity m again approaches the mean stress sensitivity M as shown in the paragraphs applying
the Haigh diagram to medium-strength steels in
the previous section. The depth distributions of
the tensile strength Rm(z) and of R0f (z) can be
estimated from appropriate correlations with
Notched
Residual stress, rs
Residual stress, rs
Smooth
0
(a)
(b)
N f = 104
a and Rf,loc
a and Rf,loc
N f = 104
Nf =
Local fatigue strength

Loading stress
0
Nf =
0
(c)
(d)
Distribution of residual stress (a) and (b) and distribution of loading stress and local fatigue strength for loadings
in the range of finite and infinite lifetime (c) and (d) for smooth (a) and (c) and notched (b) and (d) specimens
made from a high-strength steel
Fig. 62
Propagating crack
Not propagating
crack
d = 0.6 mm
1000
Rf
800
Loading
stresses
1200
d = 0.3 mm
a and Rf,loc, MPa
a and Rf,loc, MPa
1200
1000
Rf
800
600
600
0
0.1
0.2
0.3
0.4
400
rs
400
800
1200
0.4
0.1
0.2
0.3
0.4
400
800
1200
0
(c)
0.1
0.2
0.3
(b)
400
l, MPa
rs
l, MPa
(a)
0.1
0.2
0.3
0.4
(d)
Distribution of the local fatigue strength and the loading stress for different stress amplitudes with marking of
the locus of crack initiation (a) and (b) and distribution of residual stress (c) and (d) for specimens made from
a quenched SAE 1045 steel after shot peening with shot with a mean diameter of 0.6 mm (a) and (c) and 0.3 mm (b) and
(d), respectively. After Ref 38, 39
Fig. 63
measured depth distributions of the hardness

(see e.g., Ref 65).
Application to High-Strength Steels. In a
high-strength steel, manufacturing processes
may generate very large (macro) residual
stresses, as shown by several examples in the
paragraphs on high-strength steel in the section
Characteristic Examples. Contrarily, the
changes of the micro residual stress state are
rather small. Therefore, the local fatigue strength
is almost entirely determined by the initial material state and the depth of the (macro) residual
stresses. Figure 62(a) and (b) show depth distributions of compressive residual stresses in
smooth and notched bending specimens produced by shot peening. In Fig. 62(c) and (d), the
resulting depth distributions of the fatigue
strength for infinite life and for one arbitrarily
chosen finite life are plotted together with the
corresponding depth distributions of the loading
stress. It becomes clear from this figure that the
improvement of the fatigue strength for infinite
life of smooth specimens (low stress gradient)
by shot peening is small because crack initiation
occurs below the surface in a depth with vanishing residual stresses. Contrarily, with increasing
stress amplitude, the crack initiation site is
shifted toward the surface, and the shot-peeninginduced stresses influence crack initiation and
early fatigue crack growth. Hence, there is a
large influence of shot peening on the low-cycle
fatigue strength and a small influence on the
high-cycle fatigue strength. This is shown in Fig.
45(a). Contrarily, crack initiation in notched
specimens (high stress gradient) occurs always
at the surface and both low-cycle and high-cycle
fatigue strength are improved, as shown in Fig.
45(b).
The assumption of the local fatigue concept
that fatigue life determining cracks initiate only
in places where the loading stress exceeds the
local fatigue strength is verified by several experimental results (e.g., Ref 38, 39, 62). Figure
63 presents for two series of specimens of the
aforementioned investigations from Ref 38 and
39 on hardened and shot peened SAE 1045 steel
graphs of the depth distribution of the local fatigue strength (see Fig. 63a and b). Rf,loc can be
calculated from the compressive residual stress
distributions (Fig. 63c and d) under the assumption of uniform hardness. The closed circles
mark the initiation sites of damage-relevant
cracks for each load stress distribution, which
could be taken from SEM of the fracture surface
(see Fig. 24). As can be seen, the actual crack
initiation positions agree with the expectation after the concept of the local fatigue strength. For
high stress amplitudes (1150 MPa), the cracks
that determine fatigue life can initiate at the surface. For lower stress amplitudes (1150 MPa),
the damage behavior is changed. Of course,
stress amplitudes between 1000 and 1100 MPa
are able to initiate cracks at the surface (open
circles in Fig. 58a and b). However, these stress
amplitudes are too small to propagate the cracks
into greater depth because of the high compres-

sive residual stresses existing there. This aspect
is treated in more detail in the paragraphs on
high-strength steel in the section Propagating
and Nonpropagating Cracks. At stress amplitudes below 1000 MPa, cracks are generated
only beneath the surface as expected by the plotted straight lines of the loading stress in comparison with the courses of the local fatigue
strength (Fig. 62a and b). The fatigue strength Rf
is determined by the maximum loading stress
amplitude, which in any cross section of the
specimen exceeds the local fatigue strength and
initiates a crack that is able to propagate. The
corresponding distributions are plotted in Fig.
63(a) and (b) as dashed straight lines.
Figure 64 applies the concept of local fatigue
strength to shot peened smooth and notched
specimens according to Fig. 55 to 57. In notched
specimens peened with shot of low (Fig. 64a)
and high (Fig. 64b) hardness and loaded in the
range of the fatigue strength, crack initiation is
predicted at the notch root, as actually observed.
However, the surface loading stress amplitude is
much higher than the surface fatigue strength.
The same result is obtained from the Haigh diagram in Fig. 59(a). Therefore, one has to assume that loading stress amplitudes below the
fatigue limit can initiate a crack at the surface
and that the fatigue strength is not determined

by the maximum loading stress amplitude that
does not initiate a crack. This shows the limitations of the concept of the local fatigue strength
and is treated in more detail in the paragraphs on
high-strength steel in the section Propagating
and Nonpropagating Cracks.
Figure 64(c) shows that crack initiation in
smooth specimens is expected below the surface
where the local fatigue strength is hardly increased by compressive residual stresses. This is
verified by metallographical investigation (Ref
46, 49). This is the reason for the small improvement of the bending fatigue strength by shot
peening shown in Fig. 45, 46, 56, and 59. However, from Fig. 45(a) it becomes clear that shot
peening significantly improves finite fatigue life.
This finding cannot be understood on the basis
of the local fatigue strength alone and is treated
in the paragraphs on high-strength steel in the
section Propagating and Nonpropagating
Cracks.
In the case of notched specimens finished by
milling (see Fig. 45b, 46, and 56), the concept
of local fatigue strength predicts crack initiation
below the surface, because the gradient of the
residual stresses is steeper than the gradient of
the loading stresses. Hence, the increase of the
a and Rf,loc, MPa
1500
Local fatigue
strength
1000
Propagating and Nonpropagating

Cracks
Loading stress
500
0
0
0.2
0.4

(a)
0.2
0.4
0.2
0.4
(b)
(c)
Distribution of the local fatigue strength and the loading stress with marking of the locus of crack initiation
for notched (a) and (b) and smooth (c) specimens made from a quenched SAE 1045 steel after shot peening
with a shot of a hardness of 46 to 50 HRC (a) and 54 to 58 HRC (b) and (c), respectively. Source: Ref 49
Fig. 64
a and Rf,loc, MPa
1500
1000
Local fatigue
strength
500
Loading stress
0
0
0.2
0.4

(a)
0.2
0.4

(b)
Distribution of the local fatigue strength and the loading stress with marking of the locus of crack initiation
for notched (a) and smooth (b) specimens made from a quenched-and-tempered (600 C/2 h) SAE 1045 steel
after shot peening with a shot of a hardness of 46 to 50 HRC
Fig. 65
bending fatigue strength is smaller than expected

in view of the high surface compressive residual
stresses (see the Haigh diagram in Fig. 64a), but
larger than in shot peened smooth specimens
with subsurface crack initiation. This is a consequence of the higher loading stress gradient in
notched specimen compared to smooth specimens. However, crack initiation below the surface of these specimens was not proved (Ref 49).
Application to Medium-Strength Steels.
The application of the concept of the local fatigue strength to medium-strength steels is somewhat more difficult compared to high-strength
steels because residual stress relaxation becomes
important. Therefore, a knowledge of the initial
depth distribution of the residual stresses and of
the residual stress sensitivity m or of the relaxed
residual stress distribution is necessary. In Fig.
65, the concept is applied to shot peened smooth
and notched specimens made from quenchedand-tempered SAE 1045 steel. The initial and
relaxed residual stress distributions are given in
Fig. 52 and 53. Again, in notched specimens
(Fig. 65a) crack initiation at the surface is predicted as actually observed. Regarding smooth
specimens, crack initiation below the surface is
expected. Contrarily, crack initiation at the surface is found in metallographical investigation.
In both cases, the difference between local loading stress and local fatigue strength is small
throughout the depth range under consideration.
Therefore, factors such as the high roughness of
the smooth specimens may well account for the
deviation of the expected locus of crack initiation.
General Remarks. It is well known that in

notched specimens or components that are
loaded in the range of the fatigue strength, cracks
may initiate in the root of sharp notches and may
arrest in a certain depth, where the driving force
for crack propagation falls below its threshold
value because of the steep drop of the loading
stress. In most cases, such sharp notches are not
relevant for components (Ref 57, 66). However,
in the presence of high residual stresses, the interaction of loading stresses and residual stresses
may produce strong gradients of the driving
force for crack propagation making crack arrest
possible even in notches with low stress-concentration factor. On the other hand, in smooth specimens with large compressive residual stresses
at and below the surface, crack initiation may
occur below the surface because the gradient of
loading stresses is small. Then, the question
arises whether or not the crack can propagate
toward the surface where its propagation is hindered by the residual stress field. Both situations
are illustrated by the Haigh diagrams in Fig. 59.
In the case of shot peened smooth specimens,
the fatigue strength falls below the values expected by the Haigh diagrams, because crack initiation occurs below the surface, where no ben-

eficial effect of the compressive residual stresses
is effective. Contrarily, the combination of surface loading stress and surface residual stress occurring at the notch root of specimens with kt
1.7 (Fig. 59a) falls significantly above the corresponding Goodman line. As already men-
Keff, MPa m
(Eq 19)
DKeff kt rn,a p aY Kmax

4
Range of Kth,eff
0
0
0.05
0.1
0.15
0.2
0.25
0.3

Range of effective stress intensity factor DKeff
with and without consideration of the residual
stresses versus distance from surface for ground notched
specimens made from a quenched SAE 1045 steel at cyclic
bending loading (kt 1.7, g 2)
Fig. 66
(Eq 20)
for smooth (kt 1) and notched (kt 1) specimens without residual stresses. rn,a is the nominal stress amplitude. If residual stresses are
present, the maximum stress is increased or decreased, depending on the sign and the amount
of the residual stresses, leading to:
DKeff (kt rn,a
rrs) p aY Kmax
(Eq 21)
if Kmin 0 or:
DKeff kt Drn p aY
60
Keff, MPa m
DKeff Kmax Kop DKth,eff
(see Eq 16). The following concerns alternating

bending fatigue (R 1). In a rough approximation, it is assumed that Kop equals zero, resulting in:
With residual stresses
12
tioned, this means that cracks initiate at the notch

root, but are arrested below the surface.
A crack is arrested when the driving force for
its propagation falls below the threshold value:
Initial residual
stresses
40
Relaxed
residual
stresses
20
Range of Kth,eff
0
0
0.1
0.2
0.3
0.4
0.5
0.6

without and with consideration of the initial
and the relaxed, respectively, residual stresses versus distance from surface for shot peened (shot of a hardness of
54 to 58 HRC) notched specimens made from a quenched
SAE 1045 steel at cyclic bending loading (kt 1.7, g
2)
Fig. 67
(Eq 22)
if Kmin 0.
Application to a High-Strength Steel. In a
high-strength SAE 1045 steel, literature data for
the threshold value Kth,eff fall in the range from
2.2 to 3.2 MPa m (Ref 67). As a first example,
ground notched specimens (kt 1.7, g 2)
with tensile residual stresses (see Fig. 43, curve
with open circles) are treated. Concerning the
half-elliptical cracks that initiate at the notch
root, a geometry factor Y 1 is assumed. The
interaction of this residual stress state with the
loading stress amplitude ra 230 MPawhich
corresponds to the fatigue strengthresults in
the depth distribution of DKeff shown in Fig. 66.
In a depth of 5 lm only, DKeff exceeds the
threshold value. Therefore, a crack that initiates
at the notch root will always propagate into the
interior of the specimen, and the fatigue limit
40
30
Relaxed residual
stresses
Keff, MPa m
Keff, MPa m
15
Initial
residual
stresses
Range of Kth,eff
20
10
Initial residual
stresses
Relaxed
residual
stresses
10
Range of Kth,eff
Crack initiation
0
0
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.2
0.4
0.6
0.8
1.0
1.2

without and with consideration of the initial
and the relaxed, respectively, residual stresses versus distance from surface for shot peened (shot of a hardness of
4650 HRC) notched specimens made from a quenched
SAE 1045 steel at cyclic bending loading (kt 1.7, g 2)
Fig. 68
Range of the effective stress intensity factor

DKeff for subsurface crack initiation without and
with consideration of the initial and the relaxed, respectively, residual stresses versus distance from surface for
ground and additional shot peened smooth specimens
made from a quenched SAE 1045 steel at cyclic bending
loading (g 0.4)
Fig. 69
corresponds to the maximum stress amplitude

that does not initiate a crack. This is in full agreement with the Haigh diagram in Fig. 59(a). It is
interesting to note that DKeff has a local maximum in a depth of 80 lm. Up to a depth of 150
lm, the residual stress field increases significantly DKeff as compared with the loading of a
specimen without residual stresses (dotted line).
A completely different situation prevails regarding shot peened notched specimens (kt
1.7, g 2) with the initial and the relaxed residual stress distributions shown in Fig. 57(b).
As shown in paragraphs on high-strength steel
in the sections Haigh Diagram and Concept
of Local Fatigue Strength, a crack will initiate
at the notch root if a specimen is loaded in the
range of the fatigue strength. Hence, if the crack
has the length a, the loading stress and the residual stress in the distance a from the surface determines if the crack will continue propagation
or not. In Fig. 67, DKeff is plotted versus the
distance from the surface. If the initial residual
stress distribution is concerned, DKeff becomes
zero at a depth of 40 lm, and higher stress amplitudes than the measured bending fatigue
strength would be necessary to propagate a crack
into the interior of the specimen. However, if the
relaxed residual stress distribution is concerned,
DKeff approaches the threshold range between 50
and 200 lm below the surface, and cracks may
or may not propagate. Hence, the fatigue limit
corresponds to the boundary between propagating or nonpropagating cracks that initiate at the
notch root. It is interesting to note that at depths
greater than 0.3 mm, DKeff becomes much larger
than in Fig. 66 as a consequence of the high loading amplitude. This means that the strong increase of finite fatigue life by shot peening is
entirely based on very small crack propagation
rates at small distances from surface.
Regarding shot peened notched specimens
with the residual stress distributions shown in
Fig. 57(a), almost no residual stress relaxation
occurs during loading in the range of the fatigue
limit. Therefore, the depth distributions of DKeff
corresponding to the initial and the relaxed residual stress distributions shown in Fig. 68 are
almost the same. At a depth of 40 lm, DKeff
becomes zero and exceeds the threshold value
only at a depth of 170 lm. Therefore, one would
expect that cracks could still be arrested at stress
amplitudes higher than the measured bending fatigue strength. This shows the limitations of the
simple estimation given here.
Figure 64(c) shows that crack initiation in shot
peened smooth specimens with the residual
stress distribution plotted in Fig. 55 occurs below the surface at a depth of 500 lm. For the
following rough estimation, it is assumed that
the crack has a circular shape (see Fig. 24), that
it propagates with the same rate toward the surface as well as into the interior of the specimen,
and that the residual stresses in depths greater
than 500 lm are zero. If the loading stress in the
depth of the crack front and a geometry factor Y
0.63 (Ref 68) are used for the calculation of
DKeff, one finds the relationship shown in Fig.
8
With residual
stresses
Keff, MPa m
2
Range of Kth,eff
0
0
0.05
0.1
0.15
0.2
0.25
0.3

DKeff with and without consideration of the residual stresses versus distance from surface for ground
notched specimens made from a quenched-and-tempered
(600 C/2 h) SAE 1045 steel at cyclic bending loading (kt
1.7, g 2)
Fig. 70
Keff, MPa m
20
Without residual
stresses
15
Initial residual
stresses
Relaxed
residual
stresses
10
Range of Kth,eff
5
0
0
0.1
0.2
0.3
0.4

DKeff without and with consideration of the initial and the relaxed, respectively, residual stresses versus
distance from surface for ground and additional shot
peened notched specimens made from a quenched-andtempered (600 C/2 h) SAE 1045 steel at cyclic bending
loading (kt 1.7, g 2)
Fig. 71
Relaxed residual
stresses
Without residual
stresses
Keff, MPa m
16
12
8
Initial residual
stresses
4
Range of Kth,eff
0
0
0.1
0.2
0.3
0.4
0.5
of the effective stress intensity factor

Fig. 72 Rangewithout
and with consideration of the iniDKeff
tial and the relaxed, respectively, residual stresses versus
distance from surface for ground and additional shot
peened smooth specimens made from a quenched-andtempered (600 C/2 h) SAE 1045 steel at cyclic bending
loading (g 0.4)
between 300 and 200 lm below the surface and

finally crack arrest. However, in this stage of
crack propagation the crack extension will be
large compared to the depth of the compressive
residual stresses. Therefore, loading stress redistribution will occur, as will relaxation of the
compressive residual stress field in the process
zone ahead of the crack front. The crack will
continue to propagate toward the surface causing
final failure. Then, the fatigue limit is correlated
with the maximum stress amplitude that does not
cause crack initiation below the surface. Therefore, the improvement of the fatigue strength of
smooth specimens by shot peening is limited, as
shown in Fig. 45(a). However, Fig. 69 illustrates
that the overall value of DKeff and, hence, the
driving force for crack propagation is rather
small (compare, for example, Fig. 69 with Fig.
66 in view of the very different loading stress
amplitudes). Therefore, already in the transition
from infinite to finite fatigue lifewhere crack
initiation still occurs below the surface (see Fig.
63)there is a strong influence of shot peening
on fatigue life, as shown also in Fig. 45(a). The
influence of shot peening on finite fatigue life
increases with increasing stress amplitude, because the crack initiation site is shifted toward
the surface, as discussed in the paragraphs on
high-strength steel in the section Concept of
Local Fatigue Strength.
Application to a Medium-Strength Steel. In
a medium-strength SAE 1045 steel, the threshold
value Kth,eff ranges from 2.3 to 3.3 MPa m (Ref
67). In the following, ground notched specimens
(kt 1.7, g 2) with tensile residual stresses
(see Fig. 35, curve with open circles), are treated.
The interaction of this initial residual stress state
with the loading stress amplitude ra 143
MPawhich corresponds to the fatigue
strengthresults in the depth distribution of
DKeff shown in Fig. 70. At a depth of 15 lm,
DKeff exceeds the range of threshold values.
Therefore, a crack that initiates at the notch root
will always propagate into the interior of the
specimen, and the fatigue limit corresponds to
the maximum stress amplitude that does not ini-
69. It becomes obvious that a crack, once initiated, can propagate. Regarding crack propagation toward the surface, the simple model predicts very slow crack growth in the range
1200
R m = 1650 MPa
Final failure
1000
Transition range
1400 MPa
No final
failure
1150 MPa
800
600
400
0
0.2
0.4
0.6
0.8
1.0
1.2
Length of arrested crack, mm

Influence of the nominal bending stress amplitude on the length of arrested cracks in deeprolled notched specimens (kt 2) made from SAE 5135
steel with the ultimate tensile strengths indicated. After Ref
66
Fig. 73
tiate a crack. Again, similar to the quenched state

(see Fig. 66), DKeff has a local maximum in a
depth of 70 lm, and up to a depth of 150 lm the
residual stress field increases significantly DKeff
as compared with the loading of a specimen
without residual stresses (dotted line).
The corresponding relationships for shot
peened notched specimens with the residual
stress distributions shown in Fig. 53 are given in
Fig. 71. The loading stress amplitude ra 310
MPa which corresponds to the bending fatigue
strength, is considered. From Fig. 60 it becomes
clear that crack initiation at the notch root occurs
under such loading conditions because the corresponding data point (open square) falls significantly above the Goodman relationship. On the
basis of the initial residual stress distribution,
there is no driving force for the propagation of
such a crack up to a distance of 170 lm from
the notch root. However, if the distribution of
the relaxed residual stresses is considered, DKeff
comes close to the threshold in depths ranging
from 30 to 200 lm. Therefore, the fatigue limit
corresponds to a loading condition where cracks
that initiate at the notch root may propagate or
not.
In the case of smooth shot peened specimens
with the residual stress distributions shown in
Fig. 52, it is difficult to determine on the basis
of the concept of the local fatigue strength
whether crack initiation occurs below or at the
surface, if the loading approaches the fatigue
limit (see paragraph on medium-strength steel in
the section Concept of Local Fatigue Strength
and Fig. 65b). Actually, crack initiation at the
surface is observed that is favored by the rather
high roughness. Figure 72 gives the depth distribution of DKeff for the loading stress amplitude
ra 414 MPa, which corresponds to the bending fatigue limit. Regarding the initial residual
stress state, the longest nonpropagating crack
would have a length of 250 lm. Due to stress
relaxation, this value is reduced by 50%. Still,
rather long cracks have to be formed during the
initiation stage to be able to propagate into the
interior of the specimen.
The difference between the minimum loading
stress amplitude for crack initiation at the root
of notched specimen and the maximum loading
stress amplitude at which crack arrest is observed may be very large in deep-rolled specimens, which have rather small or even tensile
residual stresses at the notch root, but large compressive residual stresses down to a relatively
great depth below the surface. Then, the length
of an arrested crack will strongly depend not
only on the depth distribution of the residual
stresses, but also on the loading stress amplitude.
This is proved by Fig. 73 concerning notched
specimens (kt 2) made from differently heat
treated SAE 5135 steel with the ultimate tensile
strength indicated (Ref 66). The diagram gives
the length of arrested cracks as a function of the
nominal stress amplitude. The higher the
strength of the material state, the less the crack
length increases with increasing stress amplitude
because the amount and the stability of the com-

pressive residual stresses produced by deep rolling increase with increasing hardness. From this
figure it becomes clear that the large increase of
the notch fatigue strength visible in Fig. 41 is
almost entirely based on an increase of the resistance against crack propagation produced by
deep rolling.
In view of the relationships presented so far,
it is interesting to look again at Fig. 39 and 40.
The lowest bending fatigue limit was determined
with ground specimens having low tensile residual stresses at the surface (curve 1 in both figures). An increase of 324 MPa (curve 6 in Fig.
39) was achieved with specimens that were shot
peened according to curve 4 in Fig. 40 and from
which a surface layer of 100 lm thickness was
removed electrolytically, resulting in a shift of

the maximum compressive residual stresses into
the surface. Taking the slope of the straight line
for smooth specimens, g 0.4, in the range of
(initial) tensile residual stresses in Fig. 36 as a
rough approximation, one would expect a difference of both specimen series of 316 MPa
caused by the different residual stresses. On the
other hand, on the basis of the Goodman line in
Fig. 60, one would expect a much higher difference of 429 MPa. This shows that stress relaxation occurs in this medium-strength steel state,
as expected. Anyway, the bending fatigue limit
of both specimen series is determined by crack
initiation at the surface. The specimens corresponding to curve 5 in both figures have a lower
30
n,a = 640 MPa
n,a = 600 MPa
Keff, MPa m
Initial residual
stresses
20
Condition 6
n,a = 640 MPa
Condition 5
n,a = 600 MPa
Assumed
relaxed
residual
stresses
10
Range of Kth,eff
0
0
0.1
0.2
0.3
0.4
0.5
0.6
Range of the effective stress intensity factor DKeff without and with consideration of the initial and the relaxed,
respectively, residual stresses versus distance from surface for (corresponding to Fig. 39) shot peened smooth
specimens made from a blank-hardened AISI 5115 steel at cyclic bending loading
Fig. 74
rs
ls
Smooth
Fatigue strength, Rf
kt
Rh
fatigue limit, but higher finite fatigue lives than

the specimens corresponding to curve 6. Applying the same procedure with the aid of Fig. 36
as above to the surface residual stress state, an
increase of the bending fatigue limit of 220 MPa
compared to ground specimens is expected. Actually, a difference of 279 MPa is determined.
This finding means that the bending fatigue limit
is not determined by crack initiation at the surface, but by the largest amplitude at which crack
arrest below the surface occurs. These relationships become more clear in Fig. 74. There, DKeff
for surface cracks developing in conditions 5 and
6 at nominal stress amplitudes corresponding to
the bending fatigue strength are plotted as a
function of the distance from the surface. Regarding the initial residual stress distribution,
cracks smaller than 320 lm (condition 5) or 100
lm (condition 6) would not be able to propagate.
Actually, stress relaxation occurs, the extent of
which, however, is not given in Ref 54 to 56. If
it is assumed that the residual stresses relax by
50%, one gets the curves for the relaxed residual
stress state in the figure. Now, one can see that
a crack once initiated in condition 6 will continue propagation. Hence, the bending fatigue
strength of this condition is determined by the
resistance against crack initiation, as stated
above. Contrarily, in condition 5 the bending fatigue strength corresponds to the maximum
nominal stress amplitude at which crack arrest
below the surface occurs.
From these considerations, it becomes clear
that the depth distribution of relaxed residual
stresses must be taken into account to gain a reliable assessment of the influence of residual
stresses on the fatigue behavior of mediumstrength steel. Frequently, these data are not
known. Then, the use of initial tensile residual
stresses for the dimensioning of components is
a conservative procedure, because their detrimental effect on the fatigue behavior of a component will be reduced by residual stress relaxation. However, the beneficial effect of
compressive residual stresses will be overestimated by using the initial values, and this results
in a nonconservative dimensioning. Therefore, it
is important to estimate the remaining residual
stresses conservatively, for example with the article Stability of Residual Stresses in this
Handbook.
rs
micro
rs
ls
Summary and Recommendations

kt
Rh
Notched
Smooth
rs
micro
ls
Notched
Smooth
kt
Rh
Notched
Some important conclusions that can be

drawn from this article are illustrated by Fig. 75,
in which the fatigue strength Rf is plotted as a
function of the macro residual stress. The following parameters are regarded:
The material strength; a low-strength steel
Residual stress ,rs

Influence of the macro residual stress and some other parameters on the fatigue strength of smooth and notched
specimens made from a low-strength steel (lower band), a medium-strength steel (middle band), and a highstrength steel (upper band)
Fig. 75
(for example, a normalized steel, lower band

in the figure), a medium-strength steel (for
example, a quenched-and-tempered (at a medium temperature) steel, middle band in the
figure), and a high-strength steel (a quenched
or quenched-and-tempered (at a low temperature) steel, upper band in the figure) are concerned
The depth distribution of the macro residual
stress rrs, characterized by its sign, magnitude, and gradient grs
The depth distribution of the micro residual
stress rrs
micro
The notch factor kt (smooth and mildly
notched specimens are regarded in the figure)
The gradient of the loading stress gls
The surface topography, characterized by the
roughness height Rh
Low-Strength Steel
In a low-strength steel, there will be no or very
little influence of the macro residual stress, because it is relaxed more or less completely if the
cyclic loading approaches the fatigue strength
(see Fig. 32 and 48). A change of the micro residual stress state by work hardening may significantly increase Rf since the resistance against
cyclic plastic deformation and hence, crack initiation increases (see Fig. 31). Then, also the resistance against macro residual stress relaxation
is raised resulting in a certain sensitivity of the
work-hardened zone to macro residual stress.
This may be detrimental or beneficial (see Fig.
33b) for Rf depending on the sign and the magnitude of the macro residual stress. The influence
of the surface topography is rather small in a
low-strength steel (see Fig. 47). The fatigue
notch factor kf is significantly smaller than the
notch factor kt because cyclic plastic deformation and stress redistribution occur in the notch
root (see Fig. 32). With increasing loading stress
gradient at a given kt, the fatigue strength increases due to the decrease of the highly stressed
volume of the component or specimen (compare
kt 2.5, g 2 with kt 2.5, g 5 in Fig.
32).
Medium-Strength Steel
In a medium-strength steel, there is a significant influence of the macro residual stress on Rf
since only a small part of rrs relaxes during cyclic loading in the range of the fatigue limit (see
Fig. 36, 38, 51a, and 52). However, in the lowcycle fatigue range, relaxation becomes more
complete with increasing amplitude, and the influence of the macro residual stress vanishes (see
Fig. 34, 37, 39, 51b). Tensile residual stresses
are always detrimental to Rf. Therefore, in the
presence of large tensile residual stresses a medium-strength steel may have equal or even
lower fatigue strength than a low-strength steel
(compare Fig. 32 with 36). If relaxed tensile residual stresses are concerned, the residual stress
sensitivity m of the fatigue strength approaches
the mean stress sensitivity M (see Fig. 60). In
material states of similar hardness loaded in the
range of the fatigue limit, compressive residual
stresses relax stronger than tensile ones simply
because of the different corresponding stress am-
plitudes. However, the micro residual stress state

may be changed significantly by the processes
that generate the macro residual stress. Hence,
in a medium-strength steel the local resistance to
residual stress relaxation may be quite different
even if the same initial material state is concerned, and unique borderlines for the onset of
residual stress relaxation do not exist (see Fig.
60). A significant benefit from compressive residual stresses is only obtained if their penetration depth is sufficiently high and/or their gradient is sufficiently low as compared to the
gradient of the loading stresses. Therefore, compressive residual stresses produced by grinding
do hardly influence the fatigue strength of
smooth and notched specimens (see Fig. 36).
Contrarily, compressive residual stresses generated by deep rolling or shot peening increase the
notch fatigue strength (see Fig. 38, 39, and 41).
The fatigue strength of smooth specimens is less
improved, if subsurface crack initiation occurs.
Then, the fatigue strength of notched specimens
may even be higher than corresponding values
of smooth specimens (see Fig. 41). This means
that the fatigue notch factor kf becomes less than
unity. At vanishing residual stresses, kf comes
rather close to kt if the loading stress gradient is
small, but is significantly less at higher gls values. With increasing tensile residual stresses, the
fatigue notch factor is reduced again, as expected
on basis of the Goodman relationship for smooth
and notched specimens (see Fig. 58 and 60).
There is a considerable influence of the surface
roughness on the fatigue strength, as shown in
Fig. 47.
High-Strength Steel
In high-strength steels, stress relaxation during cyclic loading in the range of the fatigue
strength only occurs in notched specimens bearing very high compressive residual stresses.
Then, the resulting fatigue strength is also high,
and during corresponding cyclic loading very
high magnitudes of the minimum stress occur,
which leads to some residual stress relaxation
(see Fig. 57b). Contrarily, in the range of high
tensile residual stresses and cyclic loadings that
lead to infinite life or to technically relevant lifetimes the occurring maximum stresses are much
lower and no residual stress relaxation is observed even in the range of low-cycle fatigue.
Consequently, the residual stress sensitivity and
the mean stress sensitivity of Rf are identical (see
Fig. 59), and the fatigue strength is strongly reduced with increasing tensile residual stress.
This is also true for the finite fatigue life (see
Fig. 42). In the range of compressive residual
stresses, complex relationships exist. A strong
effect of rrs will only occur if cracks are initiated
at the surface. However, in thick smooth specimens or components cyclically loaded in the
range of the fatigue strength, the loading stress
gradient is usually lower than the gradient of the
local fatigue strength, which depends on the
depth distribution of the residual stresses. Consequently, cracks are initiated below the surface
(see Fig. 62a and c, 63, and 64c) and the improvement of the fatigue strength even by high
compressive residual stresses is limited (see Fig.
45a, 46, and 56). With increasing stress amplitude and hence loading stress gradient, the crack
initiation site is shifted to the surface. Therefore,
the finite fatigue life is much more improved
than the fatigue strength (see Fig. 62a and c, 63,
and 45d). In notched specimens, the loading
stress gradient is larger than in smooth ones. If
high compressive residual stresses with a sufficient depth and/or a small gradient exist, crack
initiation will occur in the notch root resulting
in a strong improvement of the fatigue strength
(see Fig. 64a and b, 45b, 46, and 56). However,
if the penetration depth of the compressive residual stresses is low and therefore the gradient
of the local fatigue strength highfor example,
after grindingthe improvement of the notch
fatigue strength will be small (see Fig. 44 and
59). Frequently, after shot peening or deep rolling, maximum compressive residual stresses occur below the notch root. Then, the fatigue
strength may not be determined by the maximum
cyclic loading, which does not result in crack
initiation, but by the maximum cyclic loading at
which crack arrest below the surface is possible
(see Fig. 67, 68, and 73). As a consequence of
all these relationships, the fatigue notch factor
may vary strongly in the presence of compressive residual stresses regarding one-notch geometry and may take values ranging from less
than unity to the notch factor kt, as sketched in
Fig. 75. In fact, regarding smooth and notched
specimens, Fig. 44 and 54 prove that kf of ground
specimens with compressive residual stresses
produced by grinding approaches kt. From Fig.
56 it can be deduced that kf of shot peened specimens approaches unity. At vanishing residual
stresses, kf comes close to kt as expected in a
high-strength steel. With increasing tensile residual stress, the fatigue notch factor is reduced
and finally approaches unity (see Fig. 44 and 54).
Again, this finding is in correspondence with the
Goodman relationship for smooth and notched
specimens (see Fig. 58 and 59). The surface
roughness has principally a large influence on
the fatigue strength of high-strength steel, as
shown in Fig. 47. On the other hand, in practice
the roughness height of hard steel is rather low
even after mechanical surface treatments such as
shot peening.
Recommendations
From all of these relationships, some recommendations may be deduced. In medium- and
high-strength steels tensile macro residual
stresses must strictly be avoided since they always promote crack initiation and crack propagation and are detrimental to the fatigue strength
andat least in higher strength steelto finite
fatigue life. In a low-strength steel, the influence
of tensile macro residual stresses is usually small
or negligible, and the change of the micro residual stress state by work hardening is much more
important. In most cases, work hardening will be

beneficial for fatigue strength andif the micro
residual stress state is sufficiently stablefor finite fatigue life. However, processes must be
avoided in which strong work hardening is combined with the generation of tensile residual
stresses. Then, residual stress relaxation during
cyclic loading will be incomplete, and the detrimental influence of tensile residual stresses on
the fatigue strength will appear.
The beneficial influence of compressive
macro residual stresses ranges from almost nil to
the full extent expected after the mean stress sensitivity of the material state under consideration
and depends on the wholeness of the parameter
considered in Fig. 75. In a low-strength steel, it
is recommended to use processes in which the
generation of compressive residual stresses is
combined with strong work hardening. Then,
there will be on one hand the beneficial influence
of work hardening on the reduction of cyclic
plasticity and the increase of the materials resistance against crack initiation. On the other hand,
the resistance against residual stress relaxation
will also increase, resulting in some beneficial
influence of the compressive macro residual
stresses on crack initiation and crack propagation. One example is deep rolling (see Fig. 33).
Another example would be shot peening. However, shot peening of a low-strength steel will
produce a rather rough surface, and this will
eventually counterbalance a large part of the
beneficial influence on the fatigue behavior.
In medium-strength steels it is also recommended to utilize processes that produce both
work hardening and compressive macro residual
stresses. Again, deep rolling (see Fig. 41) or shot
peening (see Fig. 37 and 39) can serve as examples. Now, for a significant effect it is necessary to adjust the penetration depth of the
macro residual stresses and the width of the
work-hardened zone to the depth distribution of
the loading stress. Clearly, this is impossible in
push-pull loading of smooth specimens or components (gls 0) and very difficult in any cyclic
loading of thick smooth specimens or components, where gls is rather small. Then, some improvement of the fatigue behavior will be obtained because crack initiation at the surface is
retarded or crack initiation is shifted below the
surface (see Fig. 37a and Fig. 41, smooth specimens). However, in notched or rather thin
smooth specimens or components with higher gls
it is possible and recommended to produce a
depth distribution of the macro and micro residual stresses and, hence, of the local fatigue
strength that ensures that crack initiation can
only occur at the surface. Then, a strong improvement of the fatigue strength and finite fatigue life is possible by one of two mechanisms.
On one hand, the resistance of the steel against
crack initiation and growth of very short cracks
may strongly be increased by a smooth surface,
strong work hardening, and maximum compressive residual stresses at the surface. Examples
are given in Fig. 37(b) (shot peened notched
specimens) and Fig. 39 (condition 6, shot peened
smooth specimens). In these cases, the fatigue
strength is determined by the maximum loading

stress amplitude that does not initiate a crack at
the notch root. On the other hand, by high compressive stresses below the surface, the resistance against propagation of cracks with lengths
up to the distance of the residual stress maximum
from the surface may strongly be increased.
Again, strong work hardening, is essential to prevent the compressive residual stresses as much
as possible from relaxation. Examples are given
in Fig. 39 and 40 (condition 5, shot peened
smooth specimens) and Fig. 41 and 73 (deeprolled notched specimens). In these cases, the
fatigue strength is determined by the maximum
loading stress amplitude at which crack arrest
below the surface is obtained.
In high-strength steels, residual stress relaxation is negligible except for extreme conditions
that were discussed above (see Fig. 57b). Apart
from this item, principally the same relationships
as just discussed are valid. Again, it is difficult
to achieve a large improvement of Rf in thick
smooth specimens or components, since crack
initiation will occur below the surface (see Fig.
56, 63, and 64c). However, a significant improvement of finite fatigue life in the low-cycle
fatigue range may be obtained, since crack propagation toward the surface is hindered by the
high compressive residual stresses (see Fig. 69)
and/or the crack initiation site is shifted to the
surface as a result of the increasing loading stress
gradient (see Fig. 63). In thin smooth or in
notched specimens and components it is recommended to produce high compressive residual
stresses with a large penetration depth by processes such as shot peening, deep rolling, and
others described in this Handbook. Then, crack
initiation and early crack growth is forced to occur at the surface, and enormous increases of the
fatigue strength are obtained by one of the two
mechanisms described previously. However, by
application of the aforementioned processes to a
high-strength steel, frequently rather small surface compressive residual stresses are obtained
and maximum values of |rrs| occur below the
surface. Then, the second mechanism (crack arrest below the surface) is solely effective (see
Fig. 67 and 68), and the fatigue strength may be
much higher than expected by the surface residual stress and loading state plotted in a Haigh
diagram (see Fig. 59a).
Regarding surface roughness, numerous investigations exist. The interaction of an increased roughness with tensile residual stresses
will always be detrimental to the fatigue strength
and to finite fatigue life, especially in highstrength steels. In the presence of compressive
residual stresses, different influences exist. If the
fatigue strength is entirely and the finite fatigue
life mainly determined by crack initiation at the
surface, there will be decisive influence of the
surface roughness. A much smaller influence of
surface roughness, however, may exist if the fatigue strength is determined by crack initiation
or crack arrest below the surface or if the finite
fatigue life is mainly influenced by crack propagation below the surface. If compressive resid-
ual stresses are utilized in the dimensioning of

components, it is of utmost importance to account for their stability. The reader is referred to
the article Stability of Residual Stresses in this
Handbook.
REFERENCES
1. B. Scholtes, Residual Stresses Generated by
Mechanical Surface Treatments, DGM Informationsgesellschaft, Oberursel, 1990, in
German
2. V. Hauk, Ed., Structural and Residual
Stress Analysis by Nondestructive Methods:
Evaluation, Application, Assessment, Elsevier Science B.V., Amsterdam, 1997
3. H. Ishigami, M. Matsui, Y. Jin, and K.
Ando, Proc. ICRS-6, IOM Communications
Ltd., London, 2000, p 667673
4. Fatigue and Fracture, Vol 19, ASM Handbook, ASM International, 1996
5. E. Macherauch, Laboratory Course on Material Science and Engineering, ViewegVerlag Braunschweig, Germany, 1992, in
German
6. D. Eifler, Dr.-Ing. Thesis, University Karlsruhe (TH), 1981
7. D. Eifler and E. Macherauch, Defects, Fracture and Fatigue, G.C. Sih and J.W. Prowan, Ed., Martinus Nijhoff, The Hague,
1983, p 171182
8. D. Eifler and E. Macherauch, Int. J. Fatigue,
Vol 12 (No. 3), 1990, p 165174
9. M. Becker, Dr.-Ing. Thesis, University
Karlsruhe (TH), 1987
10. M. Becker, D. Eifler, and E. Macherauch,
Mater.wiss. Werkst.tech., Vol 24, 1993, p
5764
11. P.C. Paris, M.P. Gomez, and W.E. Anderson, The Trend in Engineering, Vol 13,
1961, p 914
12. P.C. Paris and F. Erdogan, J. Basic Eng.
(Trans. ASME ), Vol 85, 1963, p 528534
13. NASA/FLAGRO: Fatigue Crack Growth
Computer Program, National Aeronautics
& Space Administration, p 5
14. R.C. McClung, K.S. Chan, S.J. Hudak, Jr.,
and D.L. Davidson, Behavior of Small Fatigue Cracks, Fatigue and Fracture, Vol 19,
ASM Handbook, ASM International, 1996,
p 153158
15. A. Wohler, Z. Bauwesen, Vol XX, 1870, p
73106
16. O.H. Basquin, Proc. ASTM, Vol 10, 1910,
p 625630
17. L.F. Coffin, Trans. ASME, Vol 76, 1954, p
931950
18. S.S. Manson, NACA Report 1170, Lewis
Flight Propulsion Laboratory, 1954
19. J.H. Smith, J. Iron Steel Inst., Vol 82, 1910,
p 246318
20. B.P. Haigh, No. 85, Report Brit. Association, 1915, p 163170
21. J. Goodman, Mechanics Applied to Engineering, Longmans-Green, London, 1899
22. H. Gerber, Z. Bayr. Architekt. Ing.-Ver., Vol
6, 1874, p 101110

23. K.N. Smith, P. Watson, and T.H. Topper, J.
Mater., Vol 5, 1970, p 767778
24. W.J. Ostergren, J. Test. Eval., Vol 4 (No.
5), 1976, p 327339
25. A. Ebenau, Dr.-Ing. Thesis, University
26. A. Ebenau, D. Eifler, O. Voehringer, and E.
Macherauch, Proc. ICSP 4, K. Iida, Ed., The
Japan Society of Precision Engineering, Tokyo, 1990, p 327336
27. F. Hoffmann, Dr.-Ing. Thesis, University Bremen, 1987; Fortschritts-Berichte VDI Reihe 5
Nr. 123, VDI-Verlag, Dusseldorf, 1987
28. I. Altenberger, Dr.-Ing. Thesis, University
GH Kassel, 2000
29. I. Altenberger, B. Scholtes, U. Martin, and
H. Oettel, Mater. Sci. Eng. A, Vol 264,
1999, p 116
30. G. Kuhn, Dr.-Ing. Thesis, University Karlsruhe (TH), 1991
31. G. Kuhn, J.E. Hoffmann, D. Eifler, B.
Scholtes, and E. Macherauch, ICRS 3, H.
Fujiwara, T. Abe, and K. Tanaka, Ed., Elsevier Applied Science, London, 1991, p
12941301
32. A. Glaser, D. Eifler, and E. Macherauch,
Mater.wiss. Werkst.tech., Vol 22, 1991, p
266274
33. H. Mughrabi, Fatigue Behavior of Metallic
Materials, D. Munz, Ed., DGM-Informationsgesellschaft Verlag, Oberursel, 1985, p
738, in German
34. J.D. Almer, J.B. Cohen, and B. Moran, Mater. Sci. Eng. A, Vol 284, 2000, p 268279
35. H. Berns and L. Weber, Residual Stresses
in Science and Technology, E. Macherauch
and V. Hauk, Ed., DGM-Informationsgesellschaft Verlag, Oberursel, 1987, p 751
758
36. H. Berns and L. Weber, Shot Peening, ICSP
3, H. Wohlfahrt, R. Kopp, and O. Voehringer, Ed., DGM-Informationsgesellschaft
Verlag, Oberursel, 1987, p 647654
37. R. Herzog, Dr.-Ing. Thesis, University
Braunschweig, 1997
38. E. Macherauch and H. Wohlfahrt, Fatigue
Behavior of Metallic Materials, D. Munz,
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
Ed., DGM-Informationsgesellschaft Verlag,

Oberursel, 1985, p 237283, in German
P. Starker, E. Macherauch, and H. Wohlfahrt, Fatigue Eng. Mater. Struct., Vol 1,
1987, p 319327
S. Jaegg, Dr.-Ing. Thesis, University GH
Kassel, 1999
S. Jaegg and B. Scholtes, ICRS 5, Linkoping, Sweden, 1997, p 10781083
E. Welsch, D. Eifler, B. Scholtes, and E.
Macherauch, Proc. Sixth European Conf. on
Fracture (ECF 6), V. Elst and A. Bakker,
Ed., Eng. Mater. Adv. Services Ltd
(EMAS), Amsterdam 1986, p 13031320
A. Stacey and G.A. Webster, Materials Research Society Proc., Vol 22, 1984, p 215
H.P. Lieurade, Advances in Surface Treatments, A. Niku-Lari, Ed., Pergamon Press,
Oxford, 1987, p 455482
K.J. Kang, J.H. Song, and Y.Y. Earmme,
Fatigue Fract. Eng. Mater. Struct., Vol 12,
1989, p 363376
J.E. Hoffmann, D. Loehe, and E. Macherauch, Proc. ICSP3, H. Wohlfahrt, R. Kopp,
and O. Voehringer, Ed., Garmisch-Partenkirchen, DGM Verlag, Oberursel, Germany,
1987, p 631638
J.E. Hoffmann, D. Loehe, and E. Macherauch, Residual Stresses in Science and
Technology, Proc. ICRS 1, E. Macherauch
and V. Hauck, Ed., Vol 1, DGM Verlag, Oberursel, Germany, 1987, p 801808
J.E. Hoffmann, D. Eifler, and E. Macherauch, Residual Stresses: Origination, Measurement, Assessment, E. Macherauch and
V. Hauck, Ed., Vol 2, DGM Verlag, Oberursel, Germany, 1983, p 287300, in German
J.E. Hoffmann, Dr.-Ing. Thesis, University
B. Syren, H. Wohlfahrt, and E. Macherauch,
Arch. Eisenhuttenwes., Vol 46, 1975, p
735739
B. Syren, H. Wohlfahrt, and E. Macherauch,
Proc. Second Int. Conf. Mech. Behavior of
Materials (ICM2) (Boston), 1976, American Society for Metals, p 807811
52. B. Syren, Dr.-Ing. Thesis, University Karlsruhe (TH), 1975

53. H. Traiser and K.-H. Kloos, Z. Werkstofftechn., Vol 16, 1985, p 135143
54. R. Schreiber, H. Wohlfahrt, and E. Macherauch, Arch. Eisenhuttenwes., Vol 49,
1978, p 207210
55. R. Schreiber, H. Wohlfahrt, and E. Macherauch, Arch. Eisenhuttenwes., Vol 48,
1977, p 653657
56. H. Wohlfahrt, Mechanical Surface Treatments, H. Wohlfahrt and P. Krull, Ed., Wiley-VCH, Weinheim, Germany, 2000, p
5584, in German
57. K.-H. Kloos, B. Fuchsbauer, and J. Adelmann, J. Fatigue, Vol 9, 1987, p 3542
58. P. Starker and E. Macherauch, Z. Werkstofftechn., Vol 14, 1983, p 109115
59. P. Starker, Dr.-Ing. Thesis, University
60. A. Navarro and E.R. de los Rios, Philos.
Mag. A, Vol 57 (No. 1), 1988, p 1536
61. K.J. Miller and E.R. de los Rios, The Behavior of Short Fatigue Cracks, Mechanical
Engineering Publications Limited, London,
UK, 1986
62. B. Scholtes, Structural and Residual Stress
Analysis by Nondestructive Methods: Evaluation, Application, Assessment, V. Hauk,
Ed., Elsevier Science B.V., Amsterdam,
1997
63. W. Backfisch and E. Macherauch, Arch. Eisenhuttenwes., Vol 50, 1979, p 167171
64. K. Wellinger and D. Dietmann, Calculation
of Strength, A. Kroener Verlag, Stuttgart,
1969, in German
65. B. Winderlich, Mater.wiss. Werkst.tech.,
Vol 21, 1990, p 378389
66. K.H. Kloos, Notches and Structural Durability, H. Nowack, Ed., Deutscher Verband
fur Materialforschung und-pruefung e.V.,
1989, p 740, in German
67. J. E. Hoffmann and D. Loehe: to be published
68. T. Fett and D. Munz, Stress Intensity Factors and Weight Functions, Computational
Mechanics Publications, 1997

DOI: 10.1361/hrsd2002p054

Stability of Residual Stresses

D. Lohe and O. Vohringer, Institut fur Werkstoffkunde I, Universitat Karlsruhe, Germany
RESIDUAL STRESSES are generated in

structural components during manufacturing
processes such as forging, machining, heat treating, shot peening, and many others. These
stresses are always a consequence of inhomogeneously distributed dimensional changes due
to inhomogeneous plastic deformations, thermochemical treatments, and/or phase transformations. They can be either beneficial or detrimental to component behavior in service,
depending on the materials state as well as the
sign, magnitude, and stability of the residual
stresses, mechanical loading, and environmental
conditions (Ref 115). Compressive macro residual stresses in the surface region of materials
with medium and high hardnesses increase the
fatigue life and the fatigue limit at cyclic loading
compared to materials states that are free of residual stresses. This improvement is caused by
an increased resistance against crack initiation
and, to a certain extent, against crack propagation if the residual stresses are sufficiently stable
in the areas of highest loading (in most cases, the
component surface areas). Moreover, compressive macro residual stresses can increase the resistance of certain materials against corrosion fatigue and stress corrosion cracking. They can
also improve the wear resistance. Therefore, in
many cases compressive macro residual stresses
are intentionally generated in near-surface regions by controlled heat treatments or by posttreatments such as shot peening or deep rolling.
The stability or relaxation behavior of these
residual stresses at purely thermal or mechanical
loadings as well as superimposed thermal and
mechanical loadings is thus of decisive importance for the service behavior of components and
hence of great interest from a scientific as well
as a practical point of view (Ref 1618). If residual stresses are relaxed by annealing or mechanical treatment, they naturally have little if
any influence on subsequent component failure.
Residual stresses can be reduced or completely relaxed by the application of mechanical
and/or thermal energy. The elastic residual
strains, ee, that are associated with the residual
stresses via Hookes law can be converted into
micro plastic strains, ep, by suitable deformation
processes. For example, this transformation can
be achieved by dislocation slip, dislocation
creep, grain-boundary sliding, and/or diffusion
creep (Ref 1617). If these processes occur to

only a limited extent or not at all, relaxation of
residual stresses is also conceivable by crack formation and propagation. Relaxation of residual
stresses in the real case occurs by complex interaction of a large number of factors. It depends
not only on the residual stress state itself but also
on the material state, loading condition, geometry, and environment of the component under
consideration.
The best known and most important techniques for inducing residual stress relaxation are
annealing (tempering), uniaxial deformation
(drawing, stretching), and cyclic deformation
(Ref 1618). Relaxation can also be caused by
thermal cycling, quenching, neutron bombardment, the effect of alternating magnetic fields (in
the use of ferromagnetics), vibration, and partial
damage. As discussed in other articles in this
Handbook, residual stresses can be differentiated
by their technological origins (i.e., whether they
were produced during forming, machining, heat
treatment, joining, coating, or casting).
Before the effect of applied thermal or mechanical energy on the residual stress state can
be assessed, the latter must be determined quantitatively. In the case of mechanical methods, the
necessarily large number of measurements involve considerable expenditure of both time and
money. Most investigations have thus employed
x-ray methodsfor example, the sin2w method
(Ref 19) with CrK radiation on the {211}-interference plane of the ferrite, as well as
interference-profile analyses (e.g., Ref 20, 21).
In this article, current knowledge about residual stress relaxation of steels is presented in a
condensed and, as far as possible, systematic
form. The abundance of available experimental
data nevertheless necessitates a limited scope.
Thus, the findings of residual stress relaxation
by annealing, tensile loading, compressive loading, and cyclic loading will be taken as examples
to demonstrate universally valid rules. Shot
peening residual stresses are particularly suitable
for such investigations due to their sign and their
relatively large absolute value. Possible formulas
that can be used to quantify residual stress relaxation are discussed, and the underlying microstructural as well as micromechanical mechanisms are considered.
Relaxation of Residual Stresses by

Annealing
If a pure metal is annealed for several hours
at a temperature of
Ta 0.5 Tm [K]
(Eq 1)
where Tm is the melting or solidus temperature,

and then cooled slowly to room temperature, almost total relaxation of the residual stresses arising from forming, machining, heat treatment, or
joining operations can be achieved. The necessary annealing time depends essentially on the
workpiece dimensions and the material state.
Since an annealing temperature of 0.5 Tm corresponds to the recrystallization temperature of
ferritic steels, complete relaxation of macro residual stresses can be expected. In the case of
pearlitic steels, macro residual stresses are completely eliminated during the pearlite-austenite
transformation above the A1 temperature. Micro
residual stresses are considerably reduced but
not entirely removed, since lattice defects (particularly dislocations) and, in the case of heterogeneous materials, the different expansion coefficients of the various phases are always
responsible for some micro residual stresses. Residual stress relaxation by annealing is brought
about by so-called thermally activated processes
for which the annealing temperature and the annealing time are interchangeable within certain
limits. In order to achieve comparable residual
stress relaxation at a lower annealing temperature, the annealing time must be increased correspondingly (Ref 16, 17).
Thermal residual stress relaxation is fundamentally affected by the residual stress state itself and by the material state. This is convincingly demonstrated by the findings presented in
Fig. l and 2 (Ref 16, 17), which show the effect
of 1 h anneals on surface macro residual stresses
of a variety of origins in a variety of steels investigated using x-ray methods. Figure 1 is a bar
chart showing the macro residual stresses present
in the original state and after the normal industrial stress relieving for hardened components of
1 h at 200 C (390 F). The materials deformed
in tension have undergone negligible residual
stress relaxation. The greater the original compressive residual stresses, the greater the residual
Stability of Residual Stresses / 55

stress relaxation produced by annealing of
surface-machined specimens: 11% for a plain
carbon steel with 0.45 wt% Carbon, SAE 1045
(Fig. 1C); 17% for SAE 1045 (Fig. 1D); and
19% for the rolling bearing steel AISI 52100
(Fig. 1E). The most pronounced relative residual
stress relaxation, however, is observed in the
hardened materials: 25% for plain carbon steel
with 0.22 wt% C, SAE 1023 (Fig. 1F); and 37%
for SAE 1045 (Fig. 1G).
Figure 2 shows the relative residual stresses
rrs(Ta)/rrs(293 K) for the same steels as a function of the homologous annealing temperature,
Ta /Tm. It can be seen that a 1 h anneal at 0.5 Tm
(about 600 C, or 1110 F) results in complete
relaxation of the macro residual stresses in every
case. Characteristic, material-specific stresstemperature curves are obtained. Clearly, residual stresses produced by hardening are relaxed
at lower temperatures, machining residual
stresses at medium temperatures, and deformation residual stresses at higher temperatures. In
order to achieve the same degree of residual
stress relaxation, (e.g., 50%) in 1 h in hardened
and in deformed SAE 1045, a temperature difference of 150 C (270 F) is necessary. Furthermore, according to Fig. 1, larger initial residual stresses result in a shift of the rrs versus
T curve to lower annealing temperatures in the
case of hardened or machined materials.
Micro residual stresses and their relaxation behavior can be determined directly from the halfwidth values (HW) of the x-ray interference
lines, which are a measure of the microstructural
work-hardening state as well as the dislocation
density. The change of the HW values are approximately proportional to the changes of the
micro residual stresses (DHW Drrs
micro).
Again, the micro residual stresses in steels are
more effectively relaxed at a higher temperature
and, like macro residual stresses (see Fig. 2),
hardening micro residual stresses are relaxed at

lower temperatures, such as those produced by
machining; likewise, machining micro residual
stresses are relaxed at lower temperatures, such
as those produced by deformation in tension. In
comparable steel states, micro residual stresses
are relaxed only after a longer period or at a
higher temperature than macro residual stresses
(Ref 16, 17). For example, in order to achieve
50% micro residual stress relaxation in tensiledeformed SAE 1045 annealed for 1 h, the annealing temperature must be 100 C (180 F)
higher than that needed to achieve the same degree of macro residual stress relaxation.
Data pertaining to the thermal relaxation of
micro residual stresses for hardened steels are
available by analyzing x-ray interference lines of
different steels in (Ref 2224). The mean lattice
distortions e21 / 2, which are proportional to the
micro residual stresses according to
2 1/2
rrs
micro E110 e
(Eq 2)
E110 is the Youngs modulus of iron in the 110

direction, increase with growing carbon content
mainly due to the increasing dislocation density
and, secondarily, to the increasing number of
solute carbon atoms in octahedral sites of the
tetragonal martensite lattice. Usually, it can be
assumed that with increasing carbon content the
temperature for the onset of residual stress relaxation is lowered and the recovery rate increases.
However, it must be remembered that with the
annealing-induced formation of carbides on dislocations, which leads to their pinning, the relaxation procedure slows.
The relaxation behavior of residual stresses
depends for a similar surface treatment, as for
example by shot peening, as shown in Fig. 3 for
the plain carbon steel SAE 1045 with 0.45 wt%
C, from the heat treatment before shot peening
400
Ck22
F
Ck45
A
Ck22
B
Ck45
C
Ck45
D
Ck45
G
and thus, from the hardness of the material (Ref

25). The macro residual stresses (left-hand side
of Fig. 3) and the micro residual stresses (righthand side; characterized by the half-width ratio
DHW/DHW0), yield with the hardness increasing
from the normalized to the quenched and tempered and to the hardened condition, greater relaxation rates. Figure 3 also illustrates the delay
of the relaxation of the micro residual stresses,
which was mentioned above in comparison with
the macro residual stresses. Further experimental
investigations of the macro and micro residual
stress relaxation of metallic materialstempering steel AISI 4140 (Ref 2628), titanium alloy
Ti-6Al-4V (Ref 29), and several CuZn alloys and
AlMg alloys (Ref 30, 31)lead to similar results.
Materials with surface residual stresses produced by grinding, milling, deep rolling, or shot
peening frequently show different residual stress
relaxation behavior when comparing the surface
layer with subsurface areas. In most cases, this
behavior is connected with work-hardening gradients, where an increasing work hardening due
to increasing dislocation densities causes growing mean strains and greater half-width values.
Characteristic examples of depth distributions
of the macro residual stresses and the half-width
values in SAE 1045, which were produced after
different heat treatments and electrolytical removal of thin material layers, are shown Fig. 4
and 5 (Ref 25). Normalized SAE 1045 samples
in the ground, milled, or shot-peened condition
result in typical depth distributions that depend
on the heat treatment (Fig. 4). In this case, the
relaxation behavior of the macro residual
stresses (left-hand side), and of the half-widths
(right-hand side), is more pronounced at the surface than in deeper layers. After selected annealing treatments, this behavior holds to completely relaxed shot peening residual stresses in
the near-surface layers in contrast to the interior,
where remaining residual stresses exist (see Fig.
4, bottom row, left, after annealing 30 min at 550
C, or 1020 F).
The relaxation behavior of normalized SAE
1045 in the shot-peened condition is reproduced
100Cr6
E
Ta, C
400
Normalized
+5% deformed
in tension
Normalized
+ ground
800
Untempered
1200
Fig. 1
200 C/1 h tempered
1.0
rs(Ta)/rs(293 K)
rs, MPa
Hardened
Hardened +
shot peened
Macro residual stress (rrs) of different steels before and after annealing for 1 h at 200 C (390 F). Ck 45 SAE
1045; Ck 22 SAE 1023; 100 Cr 6 AISI 52100
100
200
300
400
C
B
0.6
0.2
0
0.15 0.20
600
ta = 1 h
0.8
0.4
500
E
F
G
0.25 0.30 0.35
0.40 0.45 0.50
Ta/Tm, K
Macro residual stress ratio versus homologous
annealing temperature of different steels (see
Fig. 1) after annealing for 1 h
Fig. 2

again in Fig. 5 (top row) in comparison with
shot-peened conditions of quenched and tempered as well as hardened states of SAE 1045.
The shot peening residual stresses in the hardened state (bottom row) are reduced by annealing of 1 h at 300 C (570 F) almost 50% at the
surface and in the depth of the maximum of compressive residual stresses. The corresponding
HW values in the area close to the surface are
smaller than those measured in deeper regions.
This finding is well known from relatively hard
materials, such as hardened steels (500 HV)
with relatively high dislocation densities, where
the shot peening process causes decreasing HW
values due to work-softening effects (Ref 32).
The higher the hardness, the more effective the
work softening. In these material states, decreases in HW values are caused by rearrangements of the dislocations with high density in
low-energy structures. However, in the
quenched and tempered condition, and in the
normalized condition, enlarged relaxation effects are observed in the surface layer due to a
dislocation-induced higher driving force and the
relaxation rate decreases with increasing distance from the surface. These findings must be
taken into account for the valuation of residual

stress relaxation of corresponding components.
can be linearized with the help of a Zener-WertAvrami function (Ref 33) to:
m
rrs/rrs
0 exp [(Ata) ]
Thermally Activated Relaxation

Processes
In order to analyze the thermally activated
processes responsible for residual stress relaxation, the influence of annealing time, ta, and of
annealing temperature, Ta, must be known. If the
time and temperature ranges are sufficiently restricted and if rrs(Ta,ta)/rrs(293 K) constant
(abbreviated in the following to rrs/rrs
0 ), the relationship
ta t0 exp (DH/kTa)
(Eq 3)
should hold between annealing time and temperature. In Eq 3, t0 is a time constant, k is the
Boltzmann constant, and DH is the activation enthalpy for the relaxation process in the ta versus
Ta range under consideration. Data in the form
of rrs versus 1og ta plots as in Fig. 3 are not
suited to a direct evaluation using Eq 3.
The effect of time on residual stress relaxation
where m is a numerical term dependent on the

dominant relaxation mechanism and A is a function dependent on material and temperature according to
A B exp (DH/kTa)
Ta, C
Ta, C
200
250
300
350
400
450
500
550
200
250
300
0.5
350
400
450
500
550
0.5
Normalized
0
1.0
150
150
1.0
250
250
0.5
350
350
450
1.0
Quenched and tempered
0.5
HW/HW0
rs, (Ta , t a )/rs(293 K)
1.0
450
1.0
150
150
250
250
0.5
0.5
350
350
450
0
0
10
100
1000
450
Hardened
0
0
10
100
1000
Anealing time, ta, min

Macro residual stress ratio versus annealing time, and ratio of the changes of the half-width values DHW/DHW0
versus annealing time, at different annealing temperatures at the surface of shot-peened SAE 1045 samples in
normalized, quenched and tempered, and hardened states
Fig. 3
(Eq 5)
where B is a constant. It follows from Eq 4 that

rs
log ln (rrs
0 /r ) m log ta m log A
(Eq 6)
with the result that for a constant annealing temperature Ta, a plot of log ln(r0rs/rrs) versus log ta
gives a straight line (see Fig. 6).
Corresponding annealing temperatures and
times for rrs/r0rs constant can be derived from
such plots, as schematically illustrated in Fig. 6
in the log ta versus l/kTa plot. The values for
rrs/rrs0 constant lie on a straight line. Based
on Eq 5 and 6, this line results in
log ta constant
1.0
(Eq 4)
DH
ln 10 kTa
(Eq 7)
which can be rearranged to give Eq 3. The slope

of the log ta versus l/kTa plot in Fig. 6 yields to
the activation enthalpy for residual stress relaxation.
An experimental verification of Eq 6 is presented in Fig. 7, with results for shot-peened turbine blades made from a 12% Cr steel that was
annealed for residual stress relaxation in the temperature range from 150 to 600 C (300 to 1110
F) up to 104 h (Ref 34, 35). The measuring data
are fitted very well by straight lines with a slope
of approximately m 0.10. The evaluation of
these lines by Eq 7 leads to the activation enthalpy DH 3.5 eV. This finding will be discussed together with the following results.
A complete quantitative description of the
thermal relaxation of residual stresses is of practical importance. This was achieved by an extensive study on the steel AISI 4140, which was
quenched, tempered 2 h at 450 C (840 F), and
then shot-peened. The shot peening was carried
out with cast steel shot S170 (44 to 48 HRC), at
a pressure of 1.6 bar and a coverage of 98% (Ref
26, 27). The absolute values of the macro residual stresses are plotted in Fig. 8(a) versus the
logarithm of the annealing time for different annealing temperatures. They decrease with increasing time and temperature. Because the conventional way for determining the Avrami
quantities DH, m, and B in Eq 4 and 5 requires
strong extrapolations to very high and low annealing times, a new iterative method was used
that allows determination of these parameters by
a nonlinear minimization of the residual sum of
squares (Ref 26, 27). To achieve this, all measurements in the evaluation are taken into account with the same weight. The algorithm

yields DH 3.3 eV, m 0.122, and B 1.22
1021 min 1 for the surface values of the macro
residual stresses. The curves in Fig. 8(a) were
calculated using these constants in Eq 4 and 5.
They describe the time and temperature dependence of the relaxation process very well.
The alterations of surface half-widths by annealing at temperatures between 250 and 450 C
(480 and 840 F) for different annealing times
are shown in Fig. 8(b) for the same steel. The
reductions of the HW values are similar to the
relaxation of the macro residual stresses presented in Fig. 8(a). Again, the Avrami approach
is used to describe the relaxation of half-widths,
applying the new iterative method to the differences between the half-widths after annealing
and the value HW 1.65 of a normalized specimen related to their starting values instead of
the ratio rrs(Ta,ta)/rrs
0 in Eq 4. The curves in Fig.
8(b) were calculated by means of the material
properties DH 2.48 eV, m 0.116, and B
1.09 1013 min 1 and agree well with the measured values. The dependence of the relaxation
of mean strains e21 / 2 and micro residual
stresses rrs
micro, respectively, on the annealing
time was measured in the temperature range between 250 and 450 C (480 and 840 F). The
data can also be modeled by the Avrami approach using e2(Ta, ta)1/2/e201/2 instead of
rrs(Ta, ta)/rrs0 in Eq 4. The new algorithm was

used to determine the material parameters DH
2.64 eV, m 0.096, and B 5.32 1012
min1, which allow description of the measured
values as shown in Fig. 8(c).
Comparison of these relaxation data in Ref 26
and 27 shows that the relaxation of macro residual stresses at the surface of the shot-peened state
is a bit faster than the relaxation of half-widths
and mean strains; this corresponds with the fact
that the constant B differs by nearly nine orders
of magnitude (see Table 1). The reason for the
differences in the relaxation rate is that for the
relaxation of macro residual stresses, dislocation
movement is sufficient. For a distinct relaxation
of micro residual stresses, however, additional
dislocation annihilation is necessary. The exponents m show no significant alterations, and the
activation enthalpies approach the values of the
activation enthalpy of self-diffusion of iron: DHS
2.8 eV. Additional data for the same steel in
a normalized and another quenched and tempered state in Table 1 (with exception of the
hardened state) confirm this statement (Ref 28).
Accordingly, volume diffusion-controlled dislocation creep in the residual stress field that is
dominated by climbing of edge dislocations
should be the rate-controlling process for the relaxation of the shot peening residual stresses of
3.0
400
Ground
350 C/22 min
2.5
200
2.0
0
0
2.4
350 C/22 min

100
2.2
HW, 2
rs, MPa
Milled
2.0
200
0.1
0.2
0.1
0.2
0
Shot peened
3.0
400 C/30 min

500 C/30 min
200
550 C/30 min
2.5
400
2.0
0
0.1
0.2
0.3
0.1
0.2
0.3
Fig. 4
Macro residual stress and half-width of the x-ray interference line versus distance from surface before and after
annealing of ground, milled, and shot-peened SAE 1045 samples in a normalized state
AISI 4140 in normalized as well as quenched

and tempered states.
In order to obtain some information about the
relaxation behavior of macro residual stresses in
subsurface layers, the values of the residual
stresses after the annealing processes of AISI
4140 mentioned above were measured for different distances from the surface (Ref 26, 27).
The evaluated parameters of the Avrami approach vary only at the surface itself from those
values measured below the surface. There, the
evaluated data show no significant tendency and
amount to the mean values DH 2.99 eV, m
0.172, and B 6.09 1017 min1. As
shown in Fig. 9, the dependence of the macro
residual stresses after different annealing times
at 450 C (840 F) on the distance from surface
can be described quantitatively using the surface
material properties and the mean values for all
subsurface layers. The agreement between measured values and modeled curves is very good.
During heating to sufficiently high annealing
temperatures, a distinct relaxation of macro residual stresses occurs. An attempt was made to
model this behavior by extending the Avrami approach to nonisothermal stress relaxation (Ref
27). The transient relaxation of the macro residual stresses was calculated for specimens that
were immersed up to 90 s in a salt bath at 450
C (840 F). The temperature at the specimen
surfaces developed according to the T(t) curve
shown in Fig. 10. For calculation of the relaxation of residual stresses, the real T(t) relationship
was partitioned into a staircase curve with small
equidistant steps and isothermal sections. Calculation of the relaxation was determined by a
numerical integration after the so-called stresstransient method (Ref 27) under the application
of the Avrami quantities DH, m, and B (Table 1)
which are valid for the surface state. The magnitudes of the macro residual stresses measured
after interruption of the heating and cooling
down to room temperature are marked with triangles. It becomes evident that the stresstransient method describes the measured values
in an excellent manner. Figure 10 also shows the
isothermally calculated relationship between
heating time and absolute values of residual
stresses. While there are distinct differences during the first stage of heating time of approximately 20 s, the curves calculated isothermally
and by the stress-transient method closely approach at long annealing times.
A similar discussion of thermally activated
processes was carried out for other steels (Ref
16, 17, 28, 30, 31) and for AlMg alloys and
CuZn alloys (Ref 30, 31) with surface residual
stresses originating from a variety of processes.
In the case of these steels, the activation enthalpy
DH depends on the state of the material and lies
in the range of 1.1 to 2.6 eV. DH is lowest for
relaxation of residual stresses due to hardening
and highest for those due to deformation in soft
annealed states. That proves unequivocally that
residual stress relaxation in steels can occur by
several processes.
Characteristic structural changes occur during

the deformation, machining, or hardening of
steels. Typically, an increase in the dislocation
density and a change in the dislocation arrangement are observed. In the case of hardening, the
concentration of solute interstitial atoms differs
from the equilibrium value, and this, together
with the presence of dislocations, has a decisive
influence on the residual stress fields. If a single
heat treatment is carried out in the temperature
range corresponding to recovery (T 0.5 Tm),
the dislocations adopt arrangements of lower energy by elementary processes such as glide and
cross slip by screw dislocations and glide and
climb by edge dislocations. In the case of hardened material, diffusion of carbon atoms dependent on the annealing temperature and accompanied by the formation of characteristic
carbides is superimposed on these processes.
The rate-determining process, with the exception
of the early stages in the annealing of hardened
steels (Ref 36), is clearly the thermally activated
climb of edge dislocations (Ref 17). If diffusion
of matrix atoms occurs along the edge dislocations to the dislocation core, the activation enthalpy should be DH 0.5 DHS, where DHS is
the activation enthalpy of self-diffusion. If volume diffusion predominates, the DH value for
climb is determined by DHS. In the real case,
both processes occur simultaneously but to different degrees. The dislocation density and arrangement are of considerable importance. In the
case of randomly distributed dislocations or tangles of extremely high density, qt for example,
in hardened steels where qt 1012 cm2 (Ref
23, 24)residual stress relaxation is expected
to involve dislocation-core diffusion-controlled
climb by edge dislocations. Predominantly volume diffusion will determine recovery if the dislocation configurations are relatively stable and
consist of plane arrangements, cell walls, or subgrain or low-angle grain boundaries. This recovery process probably occurred in the investigated
shot-peened quenched and tempered steels and
normalized steels as well as nonferrous alloys.
In view of the activation energies, the residual
stress relaxation in the hardened steels can probably be classified between these two extremes
and occurred by two recovery mechanisms in
competition with each other.
Up to now, discussion has centered on residual stress relaxation at temperatures Ta 0.5
Tm that is, those temperatures brought about
by typical recovery processes. In this case, mechanical parameters such as hardness and yield
strength are not significantly altered. During a
recrystallization anneal at Ta 0.5 Tm, the dislocation density rapidly takes very small values
as a result of the growth of new grains. This leads
to complete removal of macro residual stresses
and to small micro residual stresses, but is associated with pronounced changes in mechanical
properties. If extensive residual stress reduction
is required in a component without significant

change in yield strength or tensile strength, the
annealing temperature and time must be chosen
to correspond with the recovery stage and not
with recrystallization.
Resistance to Residual Stress Relaxation

Residual stress relaxation by heat treatment is
fundamentally impossible if in a predominantly
uniaxial residual stress state rrs is smaller than
the creep yield strength. This resistance for the
onset of plastic creep deformation, designated in
the following by Rce, is characterized by the
creep strain limit at vanishingly small plastic deformation. As shown schematically in Fig. 11,
Rce decreases with increasing temperature and
load time. In contrast to relevant times regarding
the creep condition of a high-temperature component at service, only very short times are necessary for residual stress relaxation by creep processes (dislocation creep, grain-boundary glide,
or diffusion creep). With increasing T and/or t,
Rce approaches a localized residual stress peak
of magnitude rrs. For T Tti, rrs equals Rce
associated with the localized onset of creep deformation. Further increases in temperature or
time result in an increasing and measurable microplastic creep strain. As illustrated in Fig. 11,
residual stress relaxation begins at higher temperatures (Ti) the smaller the load time (ti) or the
residual stress (rrs) and the greater the creep resistance (Rce) of the material. Changes in Rce can
be achieved by deliberate alterations in the state
0
Shot peened
400 C/30 min
400
500 C/30 min

550 C/30 min
800
Ta,4
Normalized
Ta,3
6
300 C/60 min
400
HW, 2
Shot peened
rs, MPa
Ta,2
rs
log ln (0/rs)
Normalized
Ta,1 < Ta,2 < Ta,3 < Ta,4
800
Ta,1
log ta
400
5
4
Shot peened
800
Hardened
0
Fig. 5
log ta
0.2
Hardened
0.4 0
300 C/60 min

0.2
Ha/ln10
3
0.4
Macro residual stress and half-width versus distance from surface before and after annealing of shot-peened
SAE 1045 samples in normalized, quenched and tempered, and hardened states
1/kTa
Fig. 6
Schematic of conventional determination of

Avrami approach parameters
Loading stresses acting in opposition to the residual stresses in a given region of the material
delay the onset of microplastic deformation in
this region. Since the distribution of residual
stresses in a component is always inhomogeneous, there will be other regions in which superposition of rrs and rls in the same sense promotes plastic deformation. Therefore, it depends
on the superposition of rls and rrs at any point
in the component if a loading stress retards (|rls
rrs|max |rrs|max) or enhances (|rls rrs|max
|rrs|max) residual stress relaxation.
At this stage it is interesting to explain the
result shown in Fig. 1 and 2, where residual
stress relaxation in a deformed sample of SAE
1045 is delayed in comparison to that in SAE
1023 despite the compressive residual stresses
having double the magnitude. Due to its higher
carbon content, SAE 1045 contains a greater number of ferrite-cementite phase boundaries than
SAE 1023. These phase boundaries represent stable obstacles to dislocation slip. Apparently, the
restricted possibilities for recovery-driven movement of dislocations to arrangements of lower en-
(rrs1 r2rs)2 (r2rs r3rs)2 (r3rs r1rs)2
(Eq 9)
on the basis of the von Mises hypothesis. The

equivalent residual stress is thus dependent on
the differences between the principal residual
stress components. These in their turn are proportional to the shear stresses acting on dislocations in the slip systems. Residual stress relaxation therefore does not occur for
rrs
eq Rce
ep C(T)rntm
but does occur for
can be established for constant loading stress rls

(Ref 16, 17). C is a quantity depending on the
temperature and the condition of the material;
the exponents have the values n 1 and 0 m
1; and the creep strain increases with applied
stress and with time. Residual stress relaxation
600
rrs
eq Rce
0.4
10
500
250 C
400
300 C
300
350 C
200
400 C
100
The case rrs

eq Rce obtains only briefly, since the
immediate onset of creep deformation attempts
to restore the condition rrs
eq Rce.
10
102
ta, min
103
104
(a)
Effect of the Magnitude of Residual

Stresses on Relaxation Behavior
3.2
Ta = 250 C
The data in Fig. l and 2 clearly show that the

residual stress relaxation in machined or hard-
102
300 C
2.8 350 C
400 C
450 C
2.4
103
104
600 C
90
525 C
0
10
80
410 C
60
375 C
340 C
320 C
50
250 C
20
40
30
0.8
200 C
10
150 C
102
10
ta, min
103
104
(b)
1.4
200
Ta = 250 C
< 2 >1/2 in 103
70
rs
0.4
50 C
Ta = 4
Residual stress relaxation, %
log ln (0/rs)
Ta = 450 C
0
10
Annealing time, h
1
(Eq 10)
1.2
300 C
350 C
400 C
1.0
250
450 C
200
rs
, MPa
micro
(Eq 8)
rrs
eq
ened steels occurs more quickly or at lower temperatures, the greater the magnitude of the residual stresses themselves. This is a consequence of
temperature, time, and stress-dependent processes similar to those observed in so-called primary microcreep. Empirical relationships of the
form
|rs|, MPa
|rls rrs| Rce
ergy outweigh the increasing driving force of the

larger residual stress. The delayed residual stress
relaxation in deformed SAE 1045 is thus explained by an increase in the creep resistance Rce.
In the case of a multiaxial residual stress state,
the residual stress rrs employed in the arguments
just given must be replaced by an equivalent residual stress rrs
eq. If the principal components are
rs
rs
rrs
1 , r2 , and r3 , it can be formulated to
HW, 2
of the material. All thermally stable obstacles

that have an additional work-hardening effect
(Ref 17) shift the onset of residual stress relaxation and the entire rrs versus T curve to higher
temperatures.
Residual stress relaxation is also affected by
the superposition on a localized residual stress
rrs of an applied loading stress rls in the same
direction. Increasing rls values shift the onset of
residual stress relaxation to shorter times and/or
lower temperatures according to the relation
0.8
150
0.6
10
1.2
102
10
ta, min
103
104
(c)
1.6
1
Fig. 7
Influence of annealing time and temperature on

the (a) absolute values of macro residual stress,
(b) half-width, and (c) mean strain and micro residual stress
on the surface and their description by the Avrami approach for shot-peened AISI 4140 in a quenched and tempered condition
Fig. 8
2
log ta
Plot of log ln versus log ta for a shot-peened 12% Cr steel
(Eq 11)
However, macro residual stresses are inhomogeneously distributed over the cross section, and
for the residual stress relaxation in a localized
area the following relationship holds true:
rrs
0
E
e t e p e e 0
(Eq 13)
Thus, the Norton law and Eq 13
rs
constant ep
r
E
e p C*(T) (rrs)n e e
rs
(rrs
0 r )
ep
E
r rs
E
(Eq 14)
(Eq 12)
Regarding real values of the residua1 stresses,

plastic strains of several tenths of a percent maximum are produced by complete residual stress
relaxation (rrs 0); that is, the deformation is
in the microcreep range.
Equations 10 and 11 form the basis for quantitative estimates of the residual stress relaxation.
Multiaxial and inhomogeneous residual stress
states are neglected or excluded. Although Eq 10
cannot be substituted directly in Eq 11 on account of the variable value of the stress (r
rrs), it can be seen qualitatively from a combi-
give a proportionality between the elastic and

plastic strain rates, and between the residual relaxation rate and the actual residual stress to the
power of the Norton exponent n, respectively.
The experimental data shown in Fig. 12 reveal
strain rates that present a strong dependency on
temperature and residual stress values (Ref 26).
Furthermore, the data reveal strain rates that are
typical for creep processes. This finding supports
the conclusion drawn above that diffusioncontrolled dislocation creep in the residual stress
field should be the rate-controlling process for
the relaxation of residual stresses in steels.
650
200
500
T( t )
200
Ta = 450 C
rs, MPa
rs, MPa
ta = 6000 min
60 min
400
6 min
600
Measured
Calculated
400
Tsalt = 450 C
300
Surface data
450
Transient relaxation
350
T, C
550
250
0.1
0.2
0.3
0.4
Measured and calculated macro residual stress

at Ta 450 C (840 F) and different annealing
times versus distance from the surface of shot-peened AISI
4140 in a quenched and tempered condition (450 C, or
840 F, for 2 h)
Fig. 9
rs+ ls
t2, 2
20
40
60
80
100

650 C (1200 F)
450 C (840 F)
Hardened
T1
T2
Temperature, T
Fig. 11
Absolute values of macro residual stress after

short-time immersion in a salt bath at 450 C
(840 F) versus time. Comparison is made with curves calculated with the stress-transient method using the occurring
T(t) dependence and with modeling for isothermal annealing.
Fig. 10
Measuring property
DH, eV
B, min1
Macro residual stress
3.30
0.080
1.40 1018

Half-width
Mean strain (micro rs)
t2 >> t1
2
1
Creep yield strength versus temperature at different creep times and strain rates
t, s
Table 1 Material properties DH, m, and B of the Avrami approach determined for different
heat treatment conditions of the shot-peened steel AISI 4140
Normalized
T 1
100
105
106
107
Heat treatment
t1, 1
rs
200
Isothermal relaxation
0 min
800
In certain cases in practice, uniaxial deformation is often employed in addition to stressfree annealing to relieve residual stresses. In the
case of forming, for example, the residual
stresses can be reduced by a second forming
stage using a smaller reduction in the crosssectional area. This can be achieved by redrawing, restretching, rerolling, repressing, and
straightening (Ref 16, 17). However, these techniques can be used only on simply shaped components with a uniform cross section. In the case
of welded seams, a uniaxial load is applied to
reduce or redistribute macro residual stresses.
When a critica1 value of the applied loading
stress is exceeded, directed dislocation movement converts the elastic strain associated with
the macro residual stress into micro plastic
strain.
Several typical examples will serve to illustrate the relaxation of residua1 stresses due to
joining and shot peening by uniaxial deformation. In an evaluation of the effect on the strength
of macro residual stresses set up during welding,
it is important to know the stability of these
stresses on loading the weld. Figure 13 illustrates
the considerable reduction in macro residual
stresses accompanying tensile loading of joints
Creep yield strength, Rce
r
E
Residual Stress Relaxation by

Uniaxial Deformation
Strain rate p, s1
et constant ep ee ep
nation of the two expressions that increasing residual stress values lead to a more effective residual stress relaxation. As a result of the greater
driving force, shorter times and/or lower temperatures are necessary. This is in agreement
with the experimental results presented.
The influence of the magnitude of the residual
stresses can also be illustrated by describing the
residual stress relaxation with the Norton approach, which is known from high-temperature
creep. The total strain rate as the sum of elastic
and plastic strain rate must vanish according to
>>
cannot, however, be compared directly to a creep

test. It is much more like a stress relaxation experiment. In the latter case, the total strain remains constant while elastic strain is converted
into plastic strain. With a homogeneous stress
distribution over the cross section of a specimen,
the following expression would be valid:
3.38
3.29
2.48
2.64
2.33
0.080
0.122
0.116
0.096
0.110
1.90 1021
1.22 1021
1.09 1013
5.32 1012
1.20 1016
108
T = 450 C
109
400 C
350 C
300 C
1010
1011
60
100
200
400
250 C
600
1000
rs, MPa
Plastic strain rate versus mean residual stress of
shot-peened AISI 4140 in a quenched and tempered condition (450 C, or 840 F, for 2 h). Determined
from the data of Fig. 8(a)
Fig. 12

produced by electron beam welding and tungsten
inert gas (TIG) welding in a maraging steel (Ref
37). In the electron beam welded specimen, it
begins at about 70% of the yield strength Re of
the unannealed weld. Residual stress relaxation
is not complete. In the case of the TIG weld,
noticeable residual stress relaxation first occurs
above Re. By the time the tensile strength Rm is
reached, residual stress relaxation is virtually
complete.
Shot peening residual stress states under tensile or compressive loading are unstable, and relaxation occurs when critical loading stresses are
exceeded. Corresponding results of surface residual stresses of the quenched and tempered
steel AISI 4140 are presented in Fig. 14 as a
function of loading stress and in Fig. 15(a) as a
function of total deformation (Ref 38). At tensile
loading, relaxation of the peening-induced compressive residual stresses in the surface of 540
MPa (80 ksi) sets in at 95% of the yield
strength of the unpeened material (Re(t) Rp0.01(t)
1180 MPa, or 170 ksi). In compression, relaxation starts at 45% of the yield strength of the
unpeened material (Re(c) Rp0.01(c) 1300 MPa,
or 190 ksi). In the quenched and tempered condition of this material exists a typical strengthdifferential effect of Re(c) Re(t) 120 MPa (20
ksi) (Ref 38). The relaxation rate of macro resid-
200
Re
ual stresses directly after the onset of relaxation

occurs much more rapidly during tension than
during compression. After a total strain et
1% (rls 1250 MPa, or 180 ksi), the
surface residual stresses rrs are completely removed (see Fig. 15a). If et 1%, the sign of
rrs changes. In the case of tensile loading, the
residual stress relaxation, however, is incomplete. Thus, the relaxation behavior is different
and anisotropic during tensile and compressive
loading. Combined with the relaxation of macro
residual stresses, only a relatively small decrease
of the half-width values, which are a measure of
the micro residual stresses, is observed at the surface. Figure 15(b) illustrates this behavior as a
function of total strain.
The effect of different strengths on the relaxation behavior of shot peening residual stresses
at the surface of AISI 4140 due to uniaxial deformation is illustrated in Fig. 16. Results of Fig.
14 for specimens quenched and tempered at 450
C (840 F) are compared with results for normalized specimens and for specimens quenched
and tempered at 650 C (1200 F) (Ref 39). The
courses of rrs versus rls are similar. However,
the absolute values of the critical loading stresses
for the onset of residual stress relaxation in tension |rlscrit(t)|, and in compression, |rlscrit(c)|, increase with increasing yield strength of this steel
in different heat treatment conditions.
Figure 17 represents the influence of loading
stresses rls on the shot peening residual stresses
Rm
rrs at the surface of quenched and tempered AISI

4140 for different testing temperatures (Ref 40).
The rrs versus rls course at 25 C (75 F) is the
same as in Fig. 14. At higher temperatures (250
and 400 C, or 480 and 750 F), the initial macro
residual stresses are reduced by heating and
waiting for temperature compensation due to
thermal residual stress relaxation. The values of
the critical loading stresses for the onset of residual stress relaxation during tension and compression as well as the relaxation rate decrease
clearly with increasing temperature.
The relaxation of macro residual stresses by
uniaxial deformation begins at relatively small
loads or plastic strains. It may be complete or
partial, both the degree and rate of residual stress
relaxation depending on the type and state of the
material as well as on the nature of the applied
load.

The deformation behavior and relaxation of
macro residual stresses in materials can be
roughly approximated in terms of a cylindrical
rod with longitudinal residual stressesnamely,
constant compressive residual stresses (rrs
s ) at
the surface and constant tensile residual stresses
(rrsc ) in the core (Fig. 18). The surface is characterized by the yield strength Re,s and the core
by Re,c. Plastic deformation commences in the
core at the critical tensile loading stress:
rlscrit(t) Re(t),c rrs
c
(Eq 15)
Plastic deformation under compressive load first

occurs in the surface region due to the superposition of loading and residual stresses at the critical compressive loading stress:
200
200
400
400
800
Re Rm
1200
0
rs, MPa
rs, MPa
ls, MPa
Welding residual stress versus tensile loading
stress of a maraging steel. (a) Transverse residual stresses in the welding seam center of an electron beam
welded joint. (b) Transverse residual stresses 3 mm (0.12
in.) from the welding seam center of a TIG welded joint
Fig. 13
200
|rlscrit(c)| |Re(c),s| |rrs

s|
400
Residual stress relaxation does not occur during

tensile loading for
600
6
rls rlscrit(t) Re(t),c rrs

c
4
t, %
and during compressive loading for

|rls| |rlscrit(c)| |Re(c),s| |rrs
s|
15
HW, min
rs, MPa
0
200
600
1600
15
30
45
400
800
800
1600
60
6
ls, MPa
Macro residual stress at the surface versus loading stress under tension and compression, respectively, of shot-peened AISI 4140 in a quenched and
tempered condition (450 C, or 840 F, for 2 h)
Fig. 14
t, %
Influence of total deformation on (a) the macro

residual stress and (b) the half-width change at
the surface of shot-peened AISI 4140 in a quenched and
tempered condition (450 C, or 840 F, for 2 h)
Fig. 15
(Eq 17)
30
200
(Eq 16)
(Eq 18)
The properties rlscrit(t) and rlscrit(c) thus have the

meaning of resistances against residual stress relaxation by uniaxial deformation.
The critical compressive loading stress,
rlscrit(c), can be determined experimentally from
corresponding loading and unloading tests. The
yield strength of the surface, Re(c),s, can be estimated from Eq 16 with measurements of the surface residual stresses and the knowledge of the
initial surface residual stress by this relationship:
ls
|Re(c),s| |rcrit(c)
| |rrs
s|
(Eq 19)
In the case of a multiaxial residual stress state,

modified yield strengths of the surface occur.
Consideration of an isotropic biaxial surface re-
req |Re(c),s|
2
ls
rs
(rlscrit(c))2 (rrs
s ) rcrit(c) rs
(Eq 20)
where req is the equivalent loading stress. This

relationship leads to somewhat smaller yield
strengths, as with a uniaxial estimation (Ref 29,
41). The application of Eq 20 for the shot-peened
AISI 4140 in different conditions results in the
Re(c),s values summarized in Table 2. The corresponding critical loading stresses for initiation of
macro residual stress relaxation rlscrit(c), the yield
strengths of the unpeened conditions Re(c), the
ratio Re(c),s /Re(c), and the half-widths ratio of surface and core are also indicated. For the normalized condition the relation
|Re(c),s| |Re(c)|
is valid. For the quenched and tempered conditions, however, the yield strengths show with
|Re(c),s| |Re(c)|
a reverse behavior. The shot peening treatments

thus generate in comparison with the core a
work-hardened surface state in the normalized
condition and a work-softened surface state in
the quenched and tempered conditions. The final
observations seem to be in contradiction with the
work-hardening behavior obtained on the basis
of the measurements of the half-width values of
surface and core. The last column in Table 2
presents the ratio HWs /HWc with values between

1.05 and 1.58. Therefore, in all cases a peeninginduced microstructural work-hardening effect
appears. The smallest effect occurs at the 450 C
(840 F) quenched and tempered condition. The
other heat-treated states show a considerably
larger microstructural work-hardening effect.
These apparent contradictions become clear
with a more exact consideration of the conditions
of the local deformation of the surface layers
during shot peening and the subsequent uniaxial
compression test. Besides a biaxial plastic
stretch-forming of the surface area with maximum flow stresses and work hardening direct at
the surface, the peening-induced deformations
include additional Hertzian stresses with maximum effects below the surface (Ref 42). Now,
at the determination of the surface yield strength
Re(c),s under homogeneous compressive loading,
the direct surface layer is deformed exactly inverse to the deformation direction during shot
peening. In this case, the Bauschinger effect relieves the deformations in the inverse direction
and reduces the resistance of the work-hardened
surface layers due to back stresses and an anisotropic mobility of dislocations. It is known that
the Bauschinger effect appears especially distinctive in quenched and tempered steels (Ref
43). Therefore, it is not surprising that the shot
peening treatments of quenched and tempered
AISI 4140 lead to values of Re(c),s /Re(c) 1. The
surface yield strengths are controlled not only by
the peening-induced surface work-hardening,
which is characterized by the half-width ratio
HWs /HWc, but also by the Bauschinger effect.
Obviously, the condition Re(c),s /Re(c) 1 is valid
if the peening-induced work-hardening has a
stronger effect than the Bauschinger effect. If the
Bauschinger effect dominates, however, the ratio

of the yield strengths is Re(c),s /Re(c) 1.
Modeling of the Relaxation Behavior

A quantitative description of the relaxation
behavior of multiaxial macro residual stresses
with known distribution over the specimen cross
section under uniaxial deformation is possible
with the aid of finite-element modeling (Ref 28,
44). Isoparametric rectangular elements with
plane strain behavior were applied for this purpose. The corresponding depth courses of the
macro residual stresses as well as the initial parts
of the stress-strain curves under tensile and compressive loading must be known and must be input with certain assumptions in a discrete form.
Figure 19 shows corresponding results of tests
at 25 and 400 C (75 and 750 F) for AISI 4140
in a quenched and tempered condition. There is
a good agreement between experimental results
and finite-element modeling. An important finding is that besides the pure mechanical and thermal relaxation of macro residual stresses during
heating and waiting for temperature compensation, no additional relaxation effects occur under
200
0
rs, MPa
sidual stress state that is generated due to vertical

shot peening results on the basis of the von
Mises hypothesis for the onset of residual stress
relaxation at compressive loading to:
T = 25 C
T = 250 C
T = 400 C
200
400
600
1600
800
800
1600
ls, MPa
,
200
,
,
Macro residual stress at the surface versus loading stress under tension and compression, respectively, at different deformation temperatures for shotpeened AISI 4140 in a quenched and tempered condition
(450 C, or 840 F, for 2 h)
Fig. 17
Quenched and
tempered 450 C
Quenched and
tempered 650 C
Normalized
rs, MPa
Surface
Core
rs
rs
200
Surface
Cross
section
area
400
srs
600
1600
1200
800
400
400
800
1200
1600
ls, MPa
Macro residual stress at the surface versus loading stress under tension and compression, respectively, of shotpeened AISI 4140 in different heat treatment conditions: normalized; quenched and tempered at 450 C (840
F), for 2 h (qt 450); and quenched and tempered at 650 C (1200 F) for 2 h (qt 650)
Fig. 16
Simplified distribution of longitudinal macro

residual stress for residual stress relaxation due
to uniaxial loading
Fig. 18
The fatigue strength of steels can be greatly

influenced by macro and micro residual stresses.
Since in certain cases they can be considered as
locally variable mean stresses, they can lead to
substantia1 increases in fatigue strength (cf. Ref
16, 17). The interaction of macro residual
stresses with the cyclic deformation mechanism
during fatigue in the microcrack-free stage is of
particular importance. Coupled with this is the
stability or the cyclic relaxation of residual
stresses. Residual stress relaxation occurs when
a critical value of the applied loading stress amplitude is exceeded and a cyclic directed dislocation movement converts the elastic strain associated with the macro residual stress into
micro plastic strain.
Numerous investigators have shown that the
effect of macro residual stresses decreases with
increasing stress amplitude and growing number
of cycles as a result of residua1 stress relaxation
(cf. Ref 16, 17). This is illustrated in Fig. 20 for
quenched and tempered AISI 4140 in a shotpeened condition that has undergone a push-pull
fatigue test (Ref 28, 46). Figure 20(a) presents
cyclic deformation curves with a plot of the plastic strain amplitudes, ea,p, versus the number of
cycles, N, during stress-controlled cyclic loading
for different stress amplitudes, ra. Initially, the
specimens show macroscopically a quasi-elastic
cyclic deformation behavior at 400, 500, and 600
MPa (60, 70, and 90 ksi). At loading with ra
700 MPa (100 ksi), small plastic strain amplitudes occur from the beginning of the test. After
a distinct number of cycles for incubation, which
decreases with increasing stress amplitude, cyclic work-softening occurs at all amplitudes.
This augmentation of plastic strain amplitude
continues until specimen failure. The maximum
plastic strain amplitude observed just before failure increases with increasing stress amplitude.
The alterations of the shot peening-induced
macro residual stresses at the surface rrs with the
number of cycles are shown in Fig. 20(b). The
residual stress values are strongly reduced in the
first cycle. Afterward, a linear dependence of the
Micro residual stresses behave in a complex

fashion during the relaxation of macro residual
stresses by uniaxial deformation. So far, only
isolated experimental data based on x-ray profile
analysis are available.
When a material is formed or machined, both
directed and inhomogeneous micro residual
stresses are set up. Those of the first type are the
result of back stresses due to dislocation pileups
at grain or phase boundaries and elastic strained
second phases of heterogeneous materials. If
these dislocations move in the reverse direction
during localized plastic deformation, the back
stresses and hence the directed micro residual
stresses are initially reduced. Further deformation causes renewed buildup of back stresses in
the opposing direction associated with dislocation multiplication and hardening, which must
once again increase the micro residual stresses.
Dislocation arrangements in a random distribution or in tangles or cells give rise to inhomogeneous micro residual stresses. Microplastic
deformation can lead to the rearrangement of
dislocations into arrangements of lower energy
and thus bring about micro residual stress relaxation. If new dislocations are produced, a renewed buildup of micro residual stresses is superimposed on the relaxation process.
A reduction in the micro residual stresses in
hardened steels is observed both on deformation
and on machining (Ref 16, 17, 32, 45). Two superimposed effects can operate here. After hardening, a very high density of dislocations is presenteither randomly distributed or in tangles.
Microplastic deformation brings about a rearrangement of these dislocations into configurations with lower distortion energy and therefore
a reduction in the micro residual stresses. On the
other hand, solute carbon atoms may be induced
to jump into the energetically more favorable octahedral sites in the martensite lattice under the
influence of the stress field of the moving dislocations. This causes a reduction in the tetragonality and hence the lattice distortion due to
solute carbon atoms (Ref 45).
rrs A(ra) m(ra) log N
(Eq 21)
200
T = 25 C
0
rs, MPa
Behavior of Micro Residual Stresses
residual stresses on the logarithm of N occurs

according to a logarithmic creep law:
200
400
600
Calculated
Measured
T = 400 C
200
0
rs, MPa
Residual Stress Relaxation by Cyclic

Deformation
200
400
Calculated
Measured
600
1600
800
800
1600
ls, MPa
Measured and calculated macro residual stress
at the surface versus loading stress under tension and compression, respectively, at different deformation temperatures for shot-peened AISI 4140 in a quenched
and tempered condition (450 C, or 840 F, for 2 h)
Fig. 19
a = 700 MPa
a,p in 103
quasi-static uniaxial loading at elevated temperatures.
2
1
(a)
600 500
400
0
0
700
rlscrit(c)
Temperature
Re(c)
F
MPa
ksi
MPa
ksi
MPa
ksi
Re(c),s /Re(c)
25
250
400
75
480
750
275
300
340
40
45
49
345
300
275
50
45
40
480
496
516
70
72
75
1.39
1.65
1.87
1.58
1.33
1.28

650 C (1200 F)
20
450 C (840 F)
25
250
400
70
75
480
750
340
600
400
360
49
85
60
52
805
1300
950
830
115
190
140
120
685
953
758
606
100
138
110
88
0.85
0.73
0.79
0.73
1.56
1.05
1.05
1.05
Normalized
500
a = 400 MPa
400
Re(c),s
C
Heat treatment
200
Table 2 Critical compressive loading stress rlscrit(c), yield strength Re(c) of unpeened conditions, surface yield strength Re(c),s, yield strength ratio Re(c),s /Re(c), and ratio of half-width values
HWs /HWc of shot-peened AISI 4140 in different heat treatment conditions
rs, MPa
600
HWs /HWc
600
10
(b)
102
104
Number of cycles, N
106
Plastic strain amplitude and longitudinal macro

residual stress versus number of cycles during
strain-controlled push-pull tests with different stress amplitudes of shot-peened AISI 4140 in a quenched and tempered condition (600 C, or 1110 F, for 2 h)
Fig. 20

tionally peened conditions and warm-peened
conditionsthe changes in the macro residual
stresses, rrs
l,s, measured at the side of the specimens primarily loaded in compression as a function of the logarithm of the number of cycles.
For both peening variants, the residual stresses
at the surface relax more rapidly with increasing
stress amplitude during the first cycle as well as
during further cycling. The linear correlation between residual stresses and the logarithm of the
number of cycles N according to Eq 20 can be
recognized for wide intervals of N 1. In the
conventionally peened samples, however, the
900
800
200
700
600
300
l,s
500
400
400
300
500
600
700
10
102
103
104
Number of cycles, N
105
106
107
(a)
0
a,s, MPa
100
1000
Tpeen = 290 C
900
200
800
700
rs , MPa
600
500
400
400
500
400
600
Top longitudinal
Bottom longitudinal
Top transverse
Bottom transverse
200
0
10
1000
100
l,s
|rs|, MPa
600
a,s, MPa
Tpeen = 20 C
300
800
macro residual stresses relax faster than in the

warm-peened condition. Pronounced differences
can be seenespecially in the first cycle, where
the residual stresses in the warm-peened samples
show a higher stability than those in conventionally peened samples. For N 1, the relaxation
rate of the variant peened at room temperature is
also higher compared to the variant peened at
290 C (555 F).
The resulting surface values of the half-widths
HWs, related to their initial values HWs,0, are
given in Fig. 23 as a function of the logarithm
of the number of cycles. The half-widths of the
rs , MPa
due to the so-called cyclic creep, where the properties A and m can be determined for each stress
amplitude from the experimental data in the linear sections of the curves in Fig. 20(b). In the
case of the stress amplitude ra 400 MPa (60
ksi), after the first cycle no further residual stress
relaxation is observed with increasing number of
cycles N. Obviously, this ra value is smaller than
the materials resistance against cyclic residual
stress relaxation ra,crit at the surface. For the two
highest stress amplitudes, ra 600 and 700
MPa (90 and 100 ksi), at the end of fatigue life
greater residual stress relaxation rates are observed than predicted by the rrs /log N law. The
beginning of this phase can be correlated with
the onset of cyclic work softening in the ea,p versus N curves in Fig. 20(a). Hence, increasing
plastic strain amplitudes at constant stress amplitude may cause increasing relaxation rates of
residual stresses.
For shot-peened AISI 4140 in a quenched and
tempered state and subjected to alternating bending at the stress amplitude ra,s 700 MPa (100
ksi), Fig. 21 presents the course of the absolute
values of the macro residual stresses at the surface of flat specimens versus the number of cycles in the longitudinal and transverse directions
(Ref 40). While the longitudinal residual stresses
at the top side (which was loaded in compressive
direction in the first half-cycle) relax from 600
MPa (90 ksi) to about 480 MPa (70 ksi),
the longitudinal residual stress at the bottom side
shows no considerable relaxation. In the second
half-cycle, this trend is reversed and the residual
stresses measured become similar. In the course
of further cyclic loading up to 105 cycles, the
residual stresses decrease linearly with the logarithm of the number of cycles according to Eq
21. After crack initiation at 105 Ni 4 105,
further reductions in the magnitudes of the residual stresses are measured. The residual
stresses in the transverse direction always relax
more slowly than in the longitudinal direction.
For initial stress amplitudes at the surface between 300 and 1000 MPa (45 and 145 ksi), Fig.
22 summarizes for the same steelin conven-
102
104
Number of cycles, N
700
106
10
102
103
104
Number of cycles, N
105
106
107
(b)
Absolute values of macro residual stress at the
surface versus number of cycles during alternating bending tests of shot-peened AISI 4140 in a
quenched and tempered condition (450 C, or 840 F, for
2 h) for a surface stress amplitude of 700 MPa (100 ksi)
Fig. 21
Macro residual stress at the surface in longitudinal direction versus number of cycles during alternating
bending tests of quenched and tempered AISI 4140 (450 C, or 840 F, for 2 h) at different surface stress
amplitudes. (a) Conventionally shot-peened condition (Tpeen 20 C, or 70 F). (b) Warm-peened condition (Tpeen 290
C or 555 F)
Fig. 22

conventionally peened samples relax for ra,s
700 MPa (100 ksi). With increasing stress amplitude, an increasing relaxation is observed.
The reduction of the half-widths and, hence, of
the micro residual stresses is much more pronounced for the samples peened at room temperature compared to the other peening variant.
For the warm-peened states, the measured values after cycling are often a bit higher than the
initial values. Only for ra,s 900 MPa (130
ksi) and N 104 can a significant decrease be
seen.
For the same samples peened at 20 or 290 C
(70 or 555 F), the stability of the residual

stresses was investigated in detail (Ref 47, 48).
The depth distribution of the residual stresses
and the half-widths in the region close to the
surface was determined at different numbers of
cycles at a surface stress amplitude 1000 MPa
(145 ksi). The curves for the conventionally
peened samples in Fig. 24(a) show that the compressive residual stresses in the region close to
the surface are strongly reduced with increasing
number of cycles. During the first cycle, the absolute value of the residual stress of the surface
decreases from about 600 MPa (90 ksi) to 350
1.0
0.9
a,s, MPa
HWs/HWs,0
300
400
0.8
500
600
700
0.7
800
900
1000
0.6
1
10
102
103
104
105
106
107
Number of cycles, N
(a)
1.0
MPa (50 ksi). After N 1000, the |rrs| value is

150 MPa (20 ksi). The depth where the residual
stresses change their sign is not significantly affected. As shown in Fig. 24(a), a shot peening
treatment at 290 C (555 F) considerably delays
the relaxation of residual stresses. During the
first cycle, the initial absolute value of the residual stress at the surface, |ra,s| 660 MPa (95
ksi), is reduced only by about 140 MPa (20 ksi).
After N 1000, |ra,s| 380 MPa (55 ksi) is
measured. The affiliated depth distributions of
the half-widths of the conventionally peened
condition in Fig. 25(a) verify that with increasing number of cycles, reductions of the halfwidths and consequently of the micro residual
stresses occur due to larger rearrangements of
mobile dislocations in the surface layers. In opposition to this finding, under the same loading
conditions the warm-peened condition shows
practically no relaxation of the micro residual
stresses. Obviously, the dislocation arrangements in the warm-peened condition under cyclic loading are considerably more stable than in
the conventionally peened condition. This is due
to dynamic and static strain aging effects that
occur during and after warm peening (Ref 47,
48). Dynamic strain aging results in a more diffuse and stable dislocation structure caused by
pinning of dislocations by solute carbon atoms
and formation of extremely fine carbides.
The effect of stabilized residual stresses on fatigue behavior is illustrated in Fig. 26, with S-N
curves at alternating bending of the conventionally peened and two warm-peened conditions in
comparison with the ground condition. The S-N
curves for a failure probability of P 50% were
determined in each case from approximately 30
tests after the arcsin P procedure (Ref 49). Shot
peening at room temperature increases the fatigue strength Rfb for about 90 MPa (13 ksi) from
440 MPa (65 ksi) to 530 MPa (80 ksi) compared
to the ground condition. Compared to that, peening at 200 and 290 C (390 and 555 F) increases
the fatigue strength to 590 and 640 MPa (85 and
95 ksi), respectively.
a,s, MPa
0.9
400
The cyclic relaxation of macro residual

stresses can be divided into four phases (Ref 16,
17, 50):
HWs/HWs,0
500
600
0.8
700
Quasi-static relaxation during the first half of
800
the first cycle, which is caused by quasi-static

deformation processes
Quasi-static relaxation during the second half
of the first cycle, which is influenced by the
load reversal
Cyclic relaxation at 1 N Ni (Ni number of cycles to crack initiation) due to cyclic
deformation processes, which usually leads to
a linear reduction of the residual stresses as a
function of the logarithm of the number of
cycles according to Eq 21 due to cyclic creep
Cyclic relaxation at Ni N Nf (Nf number of cycles to failure) for sufficiently high
stress amplitudes at the surface, which is
900
0.7
1000
0.6
1
(b)
10
102
103
104
105
106
107
Number of cycles, N
Ratio of half-width at the surface versus number of cycles during alternating bending tests of quenched and
tempered AISI 4140 (450 C, or 840 F, for 2 h) at different surface stress amplitudes. (a) Conventionally shotpeened condition (Tpeen 20 C, or 70 F). (b) Warm-peened condition (Tpeen 290 C, or 555 F)
Fig. 23

enforced in the plastic zones at the tips of
initiated surface cracks.
Assuming a predominantly uniaxial macro residual stress state, it is a fundamental principle
that relaxation of residual stresses at the surface
will commence within the first cycle, if the stress
amplitude ra reaches the critical loading stress
rs,crit. It should be noted that in the case of pushpull tests the critical loading stress rs,crit is identical with the property |rlscrit(c)| (see the section
Residual Stress Relaxation by Uniaxial Deformation in this article). If ra rs,crit, the macro
residual stresses should remain stable in the first
cycle. For N 1, relaxation of macro residual
stresses can begin gradually if the critical loading stress amplitude ra,crit is exceeded (Ref 16,
17, 40, 41, 47, 48, 50):
ra ra,crit
Thus, the property ra,crit has the meaning of a

resistance against relaxation of macro residual
stresses. Equation 16 is equally valid for the
critical cycling loading stress ra,crit if Re(c),s is
replaced by the corresponding cyclic yield
strength for surface R cycl
e,s . That means for a predominantly uniaxial residual stress state, relax-
200
ation of macro residual stresses starts together

with cyclic plastic deformation in the direct surface if this condition is fulfilled:
rs
a ra,crit Rcycl
e,s m |rs |
(Eq 22)
The property m has the meaning of a residual

stress sensitivity factor. In this consideration, it
is assumed that m 1. The macro residual
stresses are stable if the condition is fulfilled:
rs
ra ra,crit Rcycl
e,s |rs |
(Eq 23)
For cyclic bending tests under the conditions illustrated in Fig. 27, the critical loading stress
rs,crit can be found by plotting the residual stress
values after the first cycle from Fig. 22 versus
the loading stress on the surface. This is given
in Fig. 28 for conventionally peened and warmpeened samples. The data points at N 1 show
that the critical loading stress at which quasistatic residual stress relaxation begins is much
higher for the warm-peened condition (rs,crit
500 MPa, or 70 ksi) than for the conventionally
peened variant (rs,crit 310 MPa, or 45 ksi) due
to the above-mentioned strain aging effects. The
residual stress relaxation during cyclic loading
Tpeen = 20 C
rs, MPa
200
N=0
400
N=1
N = 10
N = 102
N = 103
600
0.05
0.10
0.15
0.20
0.25
0.30
0.35 0.40

(a)
200
Tpeen = 290 C
a,s = 1000 MPa
rs, MPa
200
N=1
N = 10
400
N = 102
600
N = 103
0.05
0.10
0.15
0.20
0.25
0.30
0.35 0.40

(b)
Macro residual stress in longitudinal direction versus distance from surface for quenched and tempered AISI
4140 (450 C, or 840 F, for 2 h) after different numbers of cycles at the surface stress amplitude of 1000 MPa
(145 ksi). (a) Tpeen 20 C (70 F). (b) Tpeen 290 C (555 F)
Fig. 24
Knowledge of the critical loading stress rs,crit

and the initial residual stress value rrs
s at the surface enables the estimation of the compressive
yield strength of the peened surface region Re(c),s
with Eq 20 if the property |rlscrit(c)| is substituted
by rs,crit according to:
(Eq 24)
Applying Eq 24 for the shot-peened variants investigated, clear differences of Re(c),s can be recognized (see Table 3) (Ref 48). For both variants, however, the resulting compressive yield
strength at the surface is smaller than that of the
core region, which is qualitatively in agreement
with the results in Table 2. This is also due to
the Bauschinger effect. However, the
Bauschinger-induced work softening is much
smaller for the warm-peened samples than for
the samples peened at room temperature, due to
the dynamic and static strain aging effects that
occur during and after warm peening.
By using the critical loading amplitude ra,crit
as well as the residual stresses at the surface
rrss,N1 after N 1, and the assumption that the
residual stress state at the surface is still almost
axisymmetric, the cyclic yield strength at the
surface Rcycl
e,s can be calculated similarly to Eq 24
by:
2
rs
2
rs
Rcycl
e,s (ra,crit) (rs,N1) rs,N1 ra,crit
(Eq 25)
N=0
800
Evaluation of the Relaxation Behavior
|Re(c),s|
2
rs
(rs,crit)2 (rrs
s ) rs rs,crit
a,s = 1000 MPa
800
due to cyclic creep is characterized by a linear

reduction of residual stresses with increasing
logarithm of the number of cycles for 1 N
104 according to Eq 21. Therefore, the residual
stress values at the surface after 104 cycles from
Fig. 22 can be taken as a measure of the degree
of cyclic residual stress relaxation. These values
are shown also in Fig. 28 for both shot-peened
variants versus the absolute value of the applied
loading stress amplitude at the surface ra,s. The
linear fit of these points intersects the line fitted
through the residual stresses measured at N 1
at the critical loading amplitude ra,crit, which is
a measure for the onset of cyclic residual stress
relaxation. This procedure leads to ra,crit 514
MPa (74 ksi) for the conventionally peened condition and ra,crit 714 MPa (103 ksi) for the
warm-peened condition (see Fig. 28).
These values are listed in Table 4 (Ref 48). It

can be seen that Rcycl
e,s for the conventionally
peened samples is much smaller than for the
warm-peened state. Moreover, the ratio of the
calculated cyclic yield strength at the surface and
that for the core region Rcycl
1090 MPa, or
e
160 ksi) found in push-pull tests (Ref 28) indicates that the cyclic work softening that is typical
for quenched and tempered steels does not apcycl
pear in the warm-peened variant Rcycl
e,s /Re
1.07). The increase of Rcycl
at
the
warm-peened
e,s
Behavior of Micro Residual Stresses

Residual stress relaxation when N l can be
attributed to cyclic plastic deformation. In this
case, the dislocation arrangements present in the
as-received materials are rearranged to configu-
Summary
The stability of the mechanically treated surface state of components during the application
of thermal and/or mechanical energy is desirable
800
P = 50%
700
Tpeen = 290 C
Tpeen = 200 C
600
Tpeen = 20 C
500
Ground
400
104
105
106
Nf
640
590
530
Rfb, MPa
rations characteristic of the fatigued condition at

the given stress amplitude. Especially for steels
with high dislocation densities (cold-worked as
well as hardened or quenched and tempered conditions), dislocation rearrangements are frequently associated with work-softening processes in the microcrack-free stage of fatigue.
Rearrangement of dislocations from configurations typical of the machined or heat-treated
condition to that of the fatigued state is fundamental to the relaxation of macro residual
stresses. Residual elastic strains associated with
the macro residual stresses are once again converted into microplastic strains. If fatigue softening processes are involved, relaxation of micro
residual stresses can also be expected. In the case
of machined and heat-treated materials, this
manifests itself in decreasing hardness and
sharper x-ray interference lines (Ref 16, 17). Differences in the residual stress relaxation behavior
of cold-worked, machined, and heat-treated materials under cyclic load can be attributed to different dislocation arrangements and densities.
The closer the similarity to the dislocation configuration characteristic of fatigue, the less the
extent of dislocation rearrangement and hence of
residual stress relaxation.
a,s, MPa
samples is assumed also to be the result of a very

strong pinning of dislocations by clouds of carbon atoms and extremely fine carbides that exist
due to strain aging effects. This pinning is so
strong that even at highest loading amplitudes,
ra,s, the pinned dislocations cannot move. Therefore, dislocations that were newly generated during cyclic loading cause the macro residual stress
relaxation. The increase of the half-widths for
these variants during alternating bending tests
(see Fig. 23) is a further hint for this assumption.
With these relationships, the effects of the different peening treatments on fatigue strength can
be interpreted. Compared to the conventionally
peened condition, the significantly higher fatigue
strength of the warm-peened variants cannot be
explained only by the slightly higher compressive residual stresses. Moreover, their higher stability seems to be the main reason for this behavior.
440
107
108
S-N curves (failure probability, P 50%) for

quenched and tempered AISI 4140 (450 C, or
840 F, for 2 h). Samples shot-peened at different temperatures are compared with the ground condition.
Fig. 26
Cross section
area
Surface
4.0
Tpeen = 20 C
a,s = 1000 MPa
N=0
N=1
N = 10
3.5
srs
HW, 2
N = 102
Core
N = 103
rs
rs
c
3.0
2.5
Surface
2.0
0.05
0.10
0.15
0.20
0.25
0.30
0.35 0.40
Simplified distribution of longitudinal macro

residual stress for residual stress relaxation due
Fig. 27
to bending
(a)
4.0
Tpeen = 290 C
a,s = 1000 MPa
100
N = 1 N = 104
Tpeen = 20 C
N = 102
200
Tpeen = 290 C
N = 103
300
l,s
3.0
rs , MPa
N=1
N = 10
3.5
HW, 2
N=0
2.5
400
500
0.05
0.10
0.15
0.20
0.25
0.30
0.35 0.40

(b)
Half-width versus distance from surface of a quenched and tempered AISI 4140 (450 C, or 840 F, for 2 h)
after different numbers of cycles at a surface stress amplitude of 1000 MPa (145 ksi). (a) Tpeen 20 C (70 F).
(b) Tpeen 290 C (555 F)
Fig. 25
s,crit
600
700
200
2.0
a,crit
s,crit
a,crit
400
600
800
|s| or a,s, MPa
1000
Macro residual stress at the surface in longitudinal direction versus absolute value of surface
stress or surface stress amplitude of quenched and tempered AISI 4140 (450 C, or 840 F, for 2 h) after alternating
bending to N 1 and N 104 in conventionally shotpeened and warm-peened conditions
Fig. 28

Table 3 Quasi-static surface yield strength in compression Re(c),s of quenched and tempered
AISI 4140 in conventionally peened and warm-peened conditions
Shot peening
temperature
C
20
290
rrs
s
rs,crit
Re(c),s
F
MPa
ksi
MPa
ksi
MPa
ksi
Re(c),s /Re(c)
70
555
310
500
45
75
600
660
90
95
801
1008
116
146
0.60
0.78
Table 4 Cyclic surface yield strength Rcycl

e,s of quenched and tempered AISI 4140 in
conventionally peened and warm-peened conditions
Shot peening
temperature
C
20
290
rs,rs N1
ra,crit
cycl
Re,s
F
MPa
ksi
MPa
ksi
MPa
ksi
cycl
Rcycl
e,s /Re
70
555
514
714
74
103
520
620
75
90
895
1156
130
168
0.82
1.07
in order to preserve improvements in mechanical

properties. In this connection, residual stress relaxation behavior is the most important aspect to
be considered. Macro and micro residual stresses
can be reduced or completely relaxed either by
heat treatment or under unidirectional or cyclic
mechanical loading.
Elastic residual strains related to macro residual stresses can be converted into microplastic
strains by suitable deformation processesfor
example, by dislocation slip or creep. The onset
of residual stress relaxation is determined by the
critical applied loading stress, rcrit, with the
meaning of a resistance against relaxation of residual stresses, which is accompanied by the beginning of plastic deformation. In the case of
thermal treatment (rcrit 0), unidirectional deformation (rcrit rlscrit(c) and rs,crit, respectively),
as well as cyclic loading (rcrit ra,crit), residual
stress relaxation occurs when under the assumption of a predominantly uniaxial stress state this
relationship is valid:
|rls| |rcrit| |Ri| |rrs|
(Eq 26)
The characteristic materials resistances are Ri

Rce for thermal treatment, Ri Re(c),s and Re(t),s,
respectively, for unidirectional loading, and Ri
Rcycl
e,s for cyclic loading. Stability of the residual stress state exists, and residual stress relaxation does not occur, if this condition is fulfilled:
|rls| |rcrit| |Ri| |rrs|
sufficient stable obstacles to dislocation movement, which increases rcrit at the loads given. For
T 0.4 Tm (Tm melting temperature in K)
these could, for example, be grain and phase
boundaries, finely dispersed incoherent particles,
coarse secondary phases, and, as long as no diffusion can occur, solute alloying atoms and certain dislocation arrangements. The onset and extent of residual stress relaxation is influenced in
a complex manner by the combined effects of
heat treatment, unidirectional and cyclic stress,
and multiaxial applied and/or residual stress
states. The kinetic of residual stress relaxation is
essentially determined by the difference |rls|
|rrs| |Ri|. The greater the applied loading stress
isthat is, the higher the temperature, the
longer the time, and the greater the magnitude
and amplitude of the applied stress, plastic deformation, or number of cyclesthe greater the
increase in the rate of residual stress relaxation.
The rate often increases as the absolute value of
the residual stress increases. The fewer the stable
obstacles to dislocation slip, the faster the rate.
(Eq 27)
The applied loading stress r may be zero,

have a value that does not vary with time (e.g.,
thermal residual stress relaxation), have a steadily increasing value (e.g., unidirectional deformation), or have a value varying periodically
with time (e.g., cyclic deformation). Since, according to Eq 26, the onset of residual stress relaxation depends on the resistance of the material
rcrit as well as on the applied loading stress r,
all the parameters characterizing the state of the
material as well as temperature and time or frequency are of importance. The onset of residual
stress relaxation is delayed by the presence of
REFERENCES
1. E. Macherauch and V. Hauk, Ed., Proc. Int.
Conf. Residual Stresses 1, DGM-Informationsgesellschaft, Oberursel,1987
2. B. Beck, S. Denis, and A. Simon, Ed., Proc.
Int. Conf. Residual Stresses 2, Elsevier Applied Science, London, 1989
3. H. Fujiwara, T. Abe, and T. Tanaka, Ed.,
Proc. Int. Conf. Residual Stresses 3, Elsevier Applied Science, London, 1992
4. M. James, Ed., Proc. Int. Conf. Residual
Stresses 4, Society for Experimental Mechanics, 1994
5. T. Ericsson, M. Oden, and A. Andersson,
Ed., Proc. Int. Conf. Residual Stresses 5, Institute of Technology, Linkopings Universitet, 1998
6. A. Niku-Lari, Ed., Proc. Int. Conf. Shot
Peening 1, Pergamon Press, Oxford, 1982
7. H.O. Fuchs, Ed., Proc. Int. Conf. Shot Peen-
ing 2, American Shot Peening Society, 1984

8. H. Wohlfahrt, R. Kopp, and O. Vohringer,
Ed., Proc. Int. Conf. Shot Peening 3, DGMInformationsgesellschaft, Oberursel, 1987
9. K. Iida, Ed., Proc. Int. Conf. Shot Peening
4, Japan Society for Precision Engineering,
Tokyo, 1990
10. D. Kirk, Ed., Proc. Int. Conf. Shot Peening
5, Oxford, 1993
11. J. Champaigne, Ed., Proc. Int. Conf. Shot
Peening 6, San Francisco, 1996
12. A. Nakonieczny, Ed., Proc. Int. Conf. Shot
Peening 7, Institute of Precision Mechanics,
Warsaw, 1999
13. E. Kula and V. Weiss, Ed., Residual Stress
and Stress Relaxation, Plenum Publishing
Corp., 1982
14. A. Niku-Lari, Ed., Advances in Surface
Treatments, Vol 4, International Guidebook
on Residual Stresses, Pergamon Press, Oxford, 1987
15. V. Hauk, Ed., Structural and Residual
Stress Analysis by Nondestructive Methods:
Evaluation, Application, Assessment, Elsevier Science B.V., 1997
16. O. Vohringer, in Residual Stresses: GenerationMeasurementValuation, Vol 1, E.
Macherauch and V. Hauk, Ed., DGMInformationsgesellschaft, Oberursel, 1983,
p 4983
17. O. Vohringer, in Advances in Surface Treatments, Vol 4, International Guidebook on
Residual Stresses, A. Niku-Lari, Ed., Pergamon Press, Oxford, 1987, p 367396
18. M.R. James, in Advances in Surface Treatments, Vol 4, International Guidebook on
Residual Stresses, A. Niku-Lari, Ed., Pergamon Press, Oxford, 1987, p 349365
19. E. Macherauch and P. Muller, Z. Angew.
Phys., Vol 7, 1961, p 305312
20. B.E. Warren, Prog. Met. Phys., Vol 8, 1961,
p 147202
21. B.E. Delhez, Th.H. Keijser, and E.J. Mittemeijer, Fresenius Z. Anal. Chem., Vol 312,
1982, p 116
22. O. Vohringer and E. Macherauch, Hart.Tech. Mitt., Vol 32, 1977, p 153168
23. B. Hoffmann, O. Vohringer, and E. Macherauch, Mater. Sci. Eng. A, Vol 234236,
1997, p 707710
24. B. Hoffmann and O. Vohringer, in Residual
Stresses and Distortion due to Thermal Effect, D. Aurich, K.-H. Kloos, G. Lange, and
E. Macherauch, Ed., DFG-Research-Report,
Wiley-VCH Verlag, Weinheim, 1999, p
160174
25. J. Hoffmann, B. Scholtes, O. Vohringer, and
E. Macherauch, in Proc. Int. Conf. Shot
Peening 3, H. Wohlfahrt, R. Kopp, and O.
Vohringer,
Ed.,
DGM-Informationsgesellschaft, Oberursel, 1987, p 239246
26. V. Schulze, F. Burgahn, O. Vohringer, and
E. Macherauch, Materialwiss. Werkstofftech., Vol 24, 1993, p 258267
27. V. Schulze, O. Vohringer, and E. Macher-
28.
29.
30.
31.
32.
33.
34.
auch, in Proc. Int. Conf. Shot Peening 5, D.

Kirk, Ed., Oxford, 1993, p 264273
H. Holzapfel, V. Schulze, and O. Vohringer,
in Shot Peening: Compilation of Conferences, J. Flavenot and B. Scholtes, Ed., Centre Technique des Industries Mecaniques
(CETIM), 1995, p 1530
T. Hirsch, O. Vohringer, and E. Macherauch, Hart.-Tech. Mitt., Vol 38, 1983, p
229232
J. Hoffmann, Dr.-Ing. thesis, University
J. Hoffmann, B. Scholtes, O. Vohringer, and
E. Macherauch, in Proc. Int. Conf. Residual
Stresses 1, E. Macherauch and V. Hauk,
Ed., DGM-Informationsgesellschaft, Oberursel, 1987, p 695702
O. Vohringer, in Proc. Int. Conf. Shot Peening 3, H. Wohlfahrt, R. Kopp, and O. Vohringer, Ed., DGM-Informationsgesellschaft,
Oberursel, 1987, p 185204
M.W. Fine, Introduction to Phase Transformations in Condensed Systems, Macmillan,
1965
M. Roth, in Proc. Int. Conf. Residual
Stresses 1, E. Macherauch and V. Hauk,
35.
36.
37.
38.
39.
40.
41.
Ed., DGM-Informationsgesellschaft, Oberursel, 1987, p 435440

M. Roth, Z. Werkstofftech., Vol 18, 1987, p
225228
E. Hanke, Fortsch.-Ber. VDI-Z., Vol 5 (No.
9), 1969
H. Hickel, H. Wohlfahrt, and E. Macherauch, DVS-Bericht 26, Strahltech. VI, 1973,
p 5356
H. Hanagarth, O. Vohringer, and E. Macherauch, in Proc. Int. Conf. Shot Peening 4,
K. Iida, Ed., Japan Society for Precision Engineering, Tokyo, 1990, p 327336
B. Eigenmann, V. Schulze, and O. Vohringer, in Proc. Int. Conf. Residual Stresses 4,
M. James, Ed., Society for Experimental
Mechanics, 1994, p 579588
H. Holzapfel, V. Schulze, O. Vohringer, and
E. Macherauch, in Proc. Int. Conf. Residual
Stresses 5, T. Ericsson, M. Oden, and A.
Andersson, Ed., Institute of Technology,
Linkopings Universitet, 1998, p 115120
H. Holzapfel, V. Schulze, O. Vohringer, and
Peening 6, J. Champaigne, Ed., San Francisco, 1996, p 385396
42. H. Wohlfahrt, in Residual Stress and Stress

Relaxation, E. Kula and V. Weiss, Ed., Plenum Publishing Corp., 1982, p 19151921
43. B. Scholtes and O. Vohringer, Hart.-Tech.
Mitt., Vol 41, 1986, p 347354
44. H. Holzapfel, Dr.-Ing. thesis, University
45. B. Hoffmann, O. Vohringer, and E. Macherauch, Mater. Sci. Eng. A, Vol 319321,
2001, p 299303
46. V. Schulze, K.-H. Lang, O. Vohringer, and
Peening 6, J. Champaigne, Ed., San Francisco, 1996, p 413423
47. A. Wick, V. Schulze, and O. Vohringer, in
Proc. Int. Conf. Shot Peening 7, A. Nakonieczny, Ed., Institute of Precision Mechanics, Warsaw, 1999, p 102109
48. A. Wick, V. Schulze, and O. Vohringer,
Mater. Sci. Eng. A, Vol 293, 2000, p 191
197
49. D. Dengel, Z. Werkstofftech., Vol 8, 1975,
p 253261
50. H. Holzapfel, V. Schulze, O. Vohringer, and
E. Macherauch, Mater. Sci. Eng. A, Vol 248,
1998, p 918

DOI: 10.1361/hrsd2002p070

Effect of Residual Stress on Hydrogen

Embrittlement and Stress Corrosion
Cracking
A.I. Kovalev, D.L. Wainstein, V.P. Mishina, Surface Phenomena Researches Group, Moscow
V.V. Zabilsky, Physical Technical Institute, Russia
METALLURGISTS AND METALS SCIENTISTS traditionally consider hydrogen an undesirable impurity in metals and alloys, associating its presence with the origination of
floccules, porosity, and corrosion. Due to specific behavior of the element in steel, it makes
worse technological and exploitation properties
of the material. Hydrogen embrittlement (HE),
as well as other types of brittle fracture, of metals
result from nucleation and development of microcracks caused by internal stresses. Peculiarities of HE are involved with steel physical properties and mechanisms of its interaction with
hydrogen. Considerable hydrogen solubility difference depending on external factors, a high diffusion mobility, and the capability to form molecules on interfaces that induces a high internal
hydrostatic pressure all promote various types of
brittleness.
Hydrogen in Steel
Sources of Hydrogen Penetration in Steel.
Water-containing raw charge materials and fur-
nace gases involved in the open-hearth (Martins) process are the main sources of hydrogen
penetration in steel (Ref 1). Industrial cast usually contains H2 of not less than 1.5 to 2.0 cm3 /
100 g. Use of vacuum evaporation in a ladle or
at casting is the only way to decrease the content
of hydrogen in steel down to a concentration not
affecting the mechanical properties, of the steel.
Penetration of hydrogen into a carbonic and
alloyed steel was observed in a hydrogen atmosphere at a high pressure (Ref 2, 3). Metal welding is one of the most dangerous technological
processes inducing penetration of hydrogen in
steel. During the process, hydrogen, contained in
electrode coatings as organic compounds and
water molecule, is dissociating, then being in the
atomic state it dissolves in a molten metal and
partly diffuses into neighboring zones of the basic material. Various electrochemical processes,
etching in acids, and interaction with liquid
agents containing hydro-sulfide also promote to
steel saturation with hydrogen. Therefore, hydrogen is an inevitable impurity in steel pro-
T, C
Solubility of hydrogen, cm3/100 g Fe
102
900 800 700
600
500
K K0 p1 / 2 exp(D
H / RT)
lg H 1376/T 0.665
400
(Eq 1)
where K is constant of solubility, K0 is constant

of solubility in standard conditions, R is gas constant, T is temperature, and DH is endothermal
heat of solution. Those of hydrogen in pure iron,
calculated by W. Geller and T.H. Sun (Ref 4)
and then experimentally refined (Ref 57), account for 27230 J/ 0.5 mol H2 (6500 cal/0.5 mol
H2). Solubility of hydrogen in iron at various
temperatures and atmospheric pressures (Ref 5
8) is shown in Fig. 1.
From recent data, solubility of hydrogen in a
metal also depends on its purity, microstructure,
grain size, and a character of distribution of dislocations (Ref 915). The solubility curve equation for -Fe at H pressure of 0.1 MPa (0.015
ksi) is:
(Eq 2)
where H is hydrogen concentration, at.%. Temperature dependence of hydrogen solubility in cFe can be evaluated from (Ref 3):
lg H 1411/T 0.468
(Eq 3)
That in liquid iron at a pressure of approximately 0.1 MPa (0.015 ksi)

lg H 1820/T 0.112
101
0.4
Fig. 1
1700 1400 1100
duced during most modern metallurgical methods.

Hydrogen Solubility and Diffusion in Steel.
Hydrogen solubility in steel depends on temperature and pressure to suit the following equation:
0.6
0.8
1.0
1.2
1000/T, K1
1.4
1.6
1.8
Hydrogen solubility in iron in dependence on temperature at atmospheric pressure. Data source: Ref 25
(Eq 4)
A large quantity of H2, measurable with tens

cm3 per 100 g of metal, can be brought into a
steel from a gas phase only at high pressures and
temperatures (Ref 16). At standard conditions, hydrogen gradually escapes from steel, its solubility
in steel then being negligible. At temperatures
higher than 200 C (392 F) a real solid solution
Effect of Residual Stress on Hydrogen Embrittlement and Stress Corrosion Cracking / 71

of hydrogen in iron becomes unstable (Ref 17).
Thermodynamic data of H diffusion in iron at
high temperatures are well reproducible in various works (Ref 1820) that cannot be told about
the data obtained below 200 C (392 F). Diffusion mobility of hydrogen in iron at low temperatures varies in a wide range and depends on
imperfection of the crystal lattice and the content
of impurities. For example, Ref 2124 there
were obtained values of the diffusion coefficient
at 25 C (77 F) varying from 109 up to
105cm2 /s. When studying diffusion in pure
iron by intensity of hydrogen emission out of
massive samples, it was found that its diffusion
mobility is sharply depressed at lower than 200
C (392 F) (Fig. 2). Above 200 C (392 F) the
diffusion coefficient is the following:
D 1 4 103 exp (3200/ kT) cm2 /s
(Eq 5)
Below 200 C (392 F) that is:

D 0.12 exp (7820/ kT) cm2 /s
(Eq 6)
Just as impurities forming interstitial solid solutions, hydrogen interacts with various crystal
defects, creating nonuniform solid solutions.
Presence of microdeformation near the crystal
imperfections increases hydrogen solubility
there. Therefore, the solubility of hydrogen in
steel rises with an increase in density of defects.
When diffusing inside the crystal, hydrogen atoms are retained by the lattice imperfections, and
its effective diffusion coefficient essentially decreases. This results in sharply diminishing diffusion mobility of hydrogen in iron at low temperatures. Defects retaining hydrogen are known
as traps. Its concentration controls an excessive
hydrogen solubility, compared with the lattice
one (i.e., in an ideal lattice). Moreover, the number of places capturing hydrogen is inversely
proportional to the value of the effective diffusion coefficient.
The role of traps can be played by edge dislocations, boundaries of mosaic blocks or grains,
single pores uniformly distributed in crystals, inclusions of a second phase, and atoms of alloy-
ing elements. The energy of hydrogen interaction with different structural components varies
in a narrow interval (0.10.4 eV per atom). In
particular, the one with dislocations in iron was
evaluated as about 1.6 1020 J (0.1 eV per
atom) (Ref 25).
The Role of Stresses. Hydrogen atoms interact
with any anisotropic strain field, similar to its
interaction with dislocations. If concentration of
hydrogen atoms is insignificant, then distribution
of the dissolved ones in a strain field of a crystal
lattice can be found using:
C C0 exp (U/kT)
(Eq 7)
where C is a concentration of hydrogen in a certain point of the field, C0 is an average concentration, k is the Boltzmann constant, T is the kelvin temperature, and U is the interaction energy
of hydrogen atoms with the strain field in the
place of a defined concentration, C.
Accounting for only hydrostatic components
of the field, the interaction energy, U, makes:
U p dV
(Eq 8)
where p is an average hydrostatic pressure, that

is, p 13(rxx ryy rzz), and dV is the lattice
volume alteration under the influence of hydrogen.
From these equations it follows that at a constant temperature the following condition is
valid:
ln (C/Co) A 13(rxx ryy rzz)
(Eq 9)
where A kT/dV.
According to (Eq 9), hydrogen in the crystal
lattice is redistributed under stress influence.
These deductions were confirmed with experiments (Ref 26, 27). It is settled that the effective
diffusion coefficient of hydrogen essentially decreases, and its solubility increases under loading the metal in a macro-elastic interval (below
yield strength). Thus, stresses promote nucleation of new reversible hydrogen traps, disappearing after elimination of an external stress.
Similar sites arise near the crack tip, in a several
micron size zone of maximal three-axis stresses,

where hydrogen dissolved in the metal lattice
diffuses and its concentration in steel at equilibrium conditions attains maximal values as:
C rmax C0 exp (1.9 103 ry)
Therefore, in steel having the yield strength ry

of approximately 1725 MPa (250 ksi) the equilibrium concentration of hydrogen in the zone of
maximal tensile stresses exceeds its content in
volume by 26 times (Ref 28) (Fig. 3).
At absence of a crack or any other concentrator, the stresses are localized in front of a dislocation pile-up, where considerable enrichment
with hydrogen also can be observed. Both factors, concentration of tensile stresses and dissolved hydrogen, initiate nucleation of cracks
and promote its propagation.
Hydrogen Effect on Structure and Phase
Transformations in Steel. The basic consequence of hydrogen presence in iron and steels
at solidification is the development of porosity,
due to a gas-eutectic transformation (Ref 29)
similar to the well-investigated eutectic one. Total amount of pores is proportional to those of
dissolved hydrogen and is inversely proportional
to the pressure of crystallization. Hydrogen in
steel stabilizes austenite and decreases the critical point A3, the temperature of martensite transformation, and the critical rate of cooling for
quenching (Ref 29). This is the cause of martensite formation even in a carbon-free iron at standard rates of cooling. Unlike a carbonic martensite, at low temperatures of tempering there
arises a hydrogen-free martensite and a gaseous
hydrogen escaping from a sample or accumulating in micropores, creating high pressure and microcracks like floccules (Ref 29).
States of Hydrogen in Steel. Hydrogen absorbed by metal can be present in various states:
An interstitial solid solution
A segregating one on imperfections of the
crystalline structure
Adsorbing one on the surface of micropores
and particles of second phases
T, C
107
800
400
200
100
25
D, cm2/s
107
107
107
107
0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6
1000/T, K1
Fig. 2
Diffusions of hydrogen in pure annealing iron.

Source: Ref 1825
H
H2
HH
H2
H2
H
HH
HH
H
HH
H
H
H2
H2
H2
H
H
Fig. 3
(Eq 10)
Influence of hydrogen on crack propagation. Source Ref 28
H
H
H

In a molecular form, accumulating in micro-
pores
Forming hydrides
Interacting with alloying elements and second
phases
When hydrogen solutes in transition metals,
its s-electrons transfer to a vacant d-shell of a
metal with the lower energy level. The free proton of hydrogen is partly screened by electrons,
which as electronic gas can be involved in its
moving in the metal lattice (Ref 28). Hydrogen
reduces a shear modulus of -iron; at 100 K temperature, this is approximately an 8% decrease
per 1% of hydrogen atoms (Ref 30). At a room
temperature and standard atmosphere pressure
the solubility of hydrogen in steel is insignificant; its supersaturated solid solution is unstable
and decomposes over time. One part of hydrogen
dissolves into the atmosphere and another is
caught by crystalline lattice imperfections. It is
necessary to distinguish two types of defects active in relation to hydrogenpores (or collectors) and traps. Collectors represent volumes
where hydrogen molization is observed and a
considerable hydrostatic pressure develops.
Traps represent defects of smaller sizes than collectors, such as vacancies, substitution and interstitial atoms, edge dislocations, its pile-ups
and thresholds, high-angle boundaries, and fields
of a volume tension (i.e., near the crack tip).
There is a principle difference between traps and
collectors. Transition of hydrogen from a crystal
lattice into traps is reversible, while transition
into collectors is accompanied by formation of
H2 molecule and is nonreversible. The considered features depend on a steel structure.
Ferrite-Pearlite Steels. Nonmetallic inclusions, such as manganese sulfides, are traps of
hydrogen in these soft steels. The maximum solubility of hydrogen in such steel depends not
only on the content but also on the shape of manganese-sulfide particles. The rolled-out prolonged ones increase hydrogen solubility and decrease its effective diffusion coefficient in
direction perpendicular to the planes enriched by

the inclusions.
Globularization of the sulfides decreases hydrogen solubility and increases its effective diffusion coefficient. Sulfide-matrix boundaries are
the sites of a preferential accumulation of dissolved hydrogen, essentially weakening the cohesive forces and promoting disruption of the
interface continuity.
Martensitic Steels. High interior residual
stresses inducing an elastic bend of the martensite crystals and localizing in places of its junction with boundaries of primary austenitic grains
are characteristic for these steels having the
structure of an untempered martensite. In addition, martensitic steels feature a high density of
crystalline structure defects, which are boundaries of grains and twins, single dislocations. All
these elements of structure affect solubility and
diffusion of hydrogen in steel. Dispersed cementite particles observed after tempering also
act as traps for hydrogen. In maraging steels such
a role is executed by intermetallic-matrix interfaces and elastic deformation fields around the
inclusions.
Types of Hydrogen Embrittlement. In general, hydrogen affects metals in two ways:
Physical action at low temperatures when
chemical reactions are not observed
Physical-chemical effect when a chemical in-
teraction of hydrogen with various phases and

components of an alloy on the surface and in
the bulk is observed.
The attempt to enumerate all the phenomena
tied with harmful influence of hydrogen in metals was made in 1972 (Ref 31). Reference 28
specifies the following results.
Hydrogen corrosion: Develops in carbonic
Fig. 4
Microfloccule in ferrite-pearlite steel. Courtesy of

A.I. Kovalev
and low-carbon steels at long-term exposure

in a high-pressure hydrogen at high temperatures. The cause of this brittleness is hydrogen-carbon interaction accompanied by steel
surface decarbonization and methane formation (Ref 32).
Primary gas porosity: Develops due to hydrogen precipitation in a molecular form in a
smelt or in the front of crystallization (Ref 33)
Secondary porosity: Develops at decomposition of a supersaturated solid solution with
formation of small submicroscopic pores
filled by hydrogen (Ref 33, 34)
Decreasing plasticity under tension: Observed at low-rate deformation of steel containing hydrogen (Ref 33, 34)
Delayed fracture: Steel cracking under constant loading below the yield limit
Floccules or disruption cracks: Defects
mainly found in large forged pieces. Floccules are the result of gas hydrogen precipitation in pores and micro flaws during steel
crystallization and cooling of the slab or the
forged piece (Ref 35, 36).
Blisters or bubbles: Develop at electrolysis,
metal exposing in a water vapor or in corrosion agents containing H2S (Ref 33, 35)
Corrosion cracking: For the most part is tied
with precipitation of an atomic hydrogen during corrosion reactions, its adsorption on the
surface and dissolution in metal with development of hydrogen embrittlement
Main Aspects of the Hydrogen Embrittlement Theory. Definite kinds of HE involve two
basic aspectsthe mechanism of hydrogen
transfer to a fracture place, and 2-H influence on
the steel fracture toughness.
Usually hydrogen is transported by moving
dislocations or diffuses (itself) through the crystal lattice under inhomogeneous fields of elastic
stresses (Ref 3740). Being in steel, hydrogen
decreases the fracture work by influencing mechanisms of plastic deformation and reducing the
energy of disruption of interatomic bonds near
the crack tip.
In ferrite-pearlite steels, HE is accompanied
by nonreversible structural damages in the form
of micropores containing molecular hydrogen
under high pressure. Those conditions, however,
are not enough for steel fracture that occurs due
to hydrogen migration together with moving dislocations into the zone of deformation near the
tip of a developing crack. Fracture of soft steels
is always accompanied by a considerable plastic
deformation essentially affected by hydrogen
(Ref 37) facilitating the beginning of plastic deformation of ferrite due to adsorption effects.
However, hydrogen increases the hardening coefficient at higher deformation degrees (Ref 41).
As a result, it localizes the plastic deformation
in the zone of fracture, essentially reducing the
plasticity of steel and diminishing the total fracture energy.
In steels with martensitic structure, hydrogen
migrates through the field of elastic deformations into the zone of three-axial stresses before
the front of a crack (Ref 42, 43). When the content of external and dissolved hydrogen reaches
a critical value, the crack immediately propagates by the mechanism of disruption of interatomic bonds (Ref 44). In this case, the basic
mechanism of HE is tied with breaking weakened bonds in grain boundaries and decreasing
the plastic deformation energy in local sites of
ductile fracture.
Forms of HE of steels are numerous, but only
in some cases (floccules, delayed fracture, and
stress corrosion) the role of macro and micro
stresses influencing hydrogen behavior is clearly
defined.
Floccules
Floccules are interior cracks (flaws) observed
in steel: in the central zone of deformed preforms and, as an exception, in some sites of
molded wares. In polished (not etched) cross
macro-grinds these defects look like hairline
cracks (Fig. 4) opened along a plastic flow direction of forged pieces and rolled products.
Only when considerable axial stresses are present in a pre-form, the orientation of floccules al-

ters and can be perpendicular to the axis. By
macroscopic investigation of fracture surfaces of
steel affected by floccules, the light brilliant
spots of an oval shape are observed. Fractography research of those sites, using an electronic
microscope, allows the detection of the micro
quasi-chips involving a slightly expressed riverpattern picture, pores, and plastically opened
hairline cracks (Ref 44). These defects can be
found in all steels, but mostly they are observed
in chromic, chrome-nickel, chrome-nickel-tungsten, and chrome-manganese steels. Floccules
originate in forged pieces or rolled stock of large
section some distance apart on the surface under
accelerated cooling at temperatures below 150 to
200 C (302 to 392 F), sometimes at 300 C
(572 F), but more often at room temperature.
One of the main reasons of floccules formation
is a dissolved hydrogen that does not have
enough time to escape large forged pieces at
temperatures below 200 C (392 F) and diffuses
into neighboring pores, creating molecules and
considerable gas pressures (Ref 45, 46). However, hydrogen present in metal is a necessary,
but not the only, condition for origination of the
defects under consideration. Internal tensile
stresses developing under plastic deformation,
non-uniform cross section cooling of steel wares,
and structural transformations stimulate formation of floccules (Ref 4749). Decrease of rupture resistance in sites enriched by nonmetallic
inclusions or in liquation zones promotes this
kind of brittleness. The most favorable conditions occur when the flocculation temperatures
coincide with an interval of development of martensite transformation. Dissolving in austenite in
a higher quantity than in martensite, hydrogen
during martensite transformation moves from a
solid solution into pores. Resultant high hydrostatic pressures summarize with stresses caused
by volume alteration under c phase transformation. For this reason, a quenched-in-air
chrome-nickel steel is very susceptible to flocculation. This is the result of total action of the
molecular hydrogen pressure in pores and the
field of tensile stresses existing in the metal (Ref
50). Floccules are developing during the time,
and the process quite often does not terminate
when cooling finishing and continues for many
hours at room temperature.
At high contents of hydrogen and temperatures approximately 400 to 600 C (752 to 1112
F), formation of H2 molecules in pores results
in intergranular brittle failure of steel, without
participation of external (relative to the pores)
stresses. For fracture at lower hydrogen concentrations and temperatures below 200 C (392 F),
the combined effect of tensile stresses and gas
pressure in pores is necessary. In that case, a
brittle transgranular failure develops.
Processes of formation of floccules are nonreversible. An extremely slow cooling after deforming a preform made of steel susceptible to
flocculation prevents its development. The same
effect was observed at a prolonged isothermal
annealing at heightened or high temperatures.
However, heating in an atmosphere containing
even 12 to 15% hydrogen can again make steel

susceptible to flocculation. That tendency of
steel can develop at welding or cementation
when wares are quenched directly after treatment
heating. Floccules are not observed in a number
of superalloyed steels such as ferritic, semiferritic, and austenitic. High-speed steels are not
susceptible to flocculation, in spite of the fact
that very high internal stresses at accelerated
cooling develop there. Influence of alloying elements on steel susceptibility to flocculation is
tied with action on austenite transformation under cooling, anisotropy of properties, and a quantity of generated hard inclusions. At Ac3 temperature decreasing and Ac3-Ar3 hysteresis
increasing the diffusion of hydrogen and its escaping off steel pre-forms becomes slower. This
explains a high susceptibility to flocculation of
a chrome-nickel steel after its alloying with molybdenum or tungsten. Manganese and silicon,
promoting anisotropy of mechanical characteristics of a rolled stock, favor its susceptibility to
flocculation. Increasing concentration of carbon
in a hydrogen-containing steel promotes formation of methane and floccules.
Preventing Flocculation in Steel. The main
condition of preventing floccules in steel is decreasing its supersaturation with hydrogen and
elimination of H2 molecule formation at temperatures below 200 to 300 C. Also, it is necessary to minimize the anisotropy of mechanical
properties of deformed pre-forms. Melting and
casting of steel in a vacuum is the basic and the
most effective way to eliminate of floccules.
Transportation to forging and rolling of ingots
in a hot state, being in common practice, is undesirable for a steel susceptible to flocculation,
because a high quantity of hydrogen remains.
Special heat treatment schemes preventing flocculation are available for large pre-forms made
of steels of various classes.
Delayed Fracture in Steels

Delayed fracture, or the real hydrogen embrittlement, is an evolving-in-time process of
cracking (fracture) materials exposed to hydrogen and static stresses. This type of hydrogen
embrittlement is the greatest practical problem,
especially for high-strength steels, as far as the
initiation and the growth of crack with hydrogen
present are observed at stresses much less than
the yield strength of steel. The crack growth up
to the critical value with hydrogen present begins
accordingly at the threshold value of the stress
intensity coefficient, K1th, which has a much
lower value than the fracture toughness in the air
or inert medium, K1C. Thus, the danger of the
delayed fracture process exists in that the fracture of constructions occurs under loads that are
entirely undangerous under normal operating
conditions (i.e., in the absence of hydrogen). Another important point is that the process of delayed fracture develops most intensively at temperatures close to room temperature, at which the
overwhelming majority of materials and products are used.
The main factors determining the kinetic features of delayed fracture are the chemical composition and the structure of steel, a hydrogen
content, the presence of stress concentrators, the
test temperature, the amount and the distribution
of impurities and nonmetallic inclusions, and the
presence of inherent stresses (Ref 29). In this
case the delayed fracture can develop both under
the effect of external hydrogeneous media and
inherent hydrogen having entered the metal as
the result of melting, welding, etching, anticorrosive electroplating, and other production
operations. An example is in the formation of
so-called cold cracks in the heat-affected zone
while welding alloyed steels (Ref 51). The increased hydrogen content in the near-weld zone
is the result of hydrogen diffusion from the melt
where it enters due to dissociation of moisture
in electrode coating (welding flux) (Ref 52, 53).
The effect of the steel-strength level on the
resistance to delayed fracture has been supported
by many studies. It has been shown that at the
low-strength level (up to 750 MPa, or 110 ksi)
the influence of hydrogenation is comparatively
moderate. At strength more than 1000 MPa (145
ksi), the resistance to delayed fracture reduces so
that the value of the threshold stress, rth, decreases to 200 MPa (29 ksi) (Ref 53). It has
been found also that as to the high-strength
steels, especially in the presence of stress concentrators, delayed fracture can develop even
with the low-hydrogen content not exceeding approximately 10 4 at.% (1.1 cm3 per 100 g of
metal) (Ref 1). Numerous examples of delayed
fracture of products fabricated from highstrength alloyed steels (shafts made of Fe-0.45C2Ni-Cr-Mo-V steel, axles made of Fe-0.2C-2Cr4Ni steel in the carburized state, etc.) are
presented in a monograph (Ref 54). The presence
of internal hydrogen in association with high residual stresses with a value reaching 1000 MPa,
(145 ksi) is the most commonly occurring cause
for delayed fracture in the case under consideration.
A great deal of publications in which various
methods of hydrogenation are presented are devoted to the study of delayed fracture of steels.
A main body of experimental data has been obtained with the use of the cathode polarization
method. Therefore, in analysis of the influence
of chemical composition and structure, it is necessary to take into consideration that, because of
electrolytic hydrogenation of large-size samples,
the occurrence of nonuniform distribution of hydrogen over the section can give rise to the largescale effect (Ref 55). Hydrogenation from the
gas phase under conditions of elevated temperatures (above 473 K) may cause irreversible damages in the form of bulgings and pores appearing
because of high pressure during the formation of
inner molecular hydrogen or methane. Additionally, in this case the probable surface decarburization can also have an effect on the initiation
and growth kinetics of cracks (Ref 56). From a
practical standpoint, on the contrary, these im-

portant experiments make it possible to simulate
service conditions for the purpose of developing
adequate measures to eliminate delayed fracture.
Main Criteria of Delayed Fracture of
Steels. In general, steel delayed fracture takes
1
1
Stress,
th
Time for failure,

Curves of long-term strength of hydrogenated
steel (scheme). 1, time before failure s; 2, incubation period si, r1, upper critical stress; rth, lower criticastress. Source: Ref 35, 56, 57
Fig. 5
Stress, , MPa
900
700
3
4
500
300
1
100
0
1
20
10
30
40
50
Time to failure, , h
60
100
Dependences of long-term strength of iron (1,

1), steel Fe-0.2C (2, 2), steel Fe-0.45C (3, 3)
and steel Fe-0.8C (4, 4). 14, unhydrogenated state (testings in the air); 14 testings in the process of hydrogenation. Source: Ref 59
Velocity of crack propagation, dl/dt
Fig. 6
IV
III
II
K1c
K1t
Stress intensity factor, K1
Fig. 7
Schematic illustration of kinetic diagram of

cracking of hydrogenated steel. Source: Ref 57
three stagescrack initiation; its slow growth

up to the critical value; and the stage of the unstable, accelerated crack growth. It is common
practice to characterize the sensitivity of steels to
delayed fracture depending on their chemical
composition and structure by so-called long-term
strength dependencies (Fig. 5), which usually are
represented in coordinates: applied stress r, logarithm s, where s is the time up to fracture. Dependencies mentioned are typical for both smooth
samples and samples with the stress concentrator.
The preliminary hydrogenation in constructing
dependencies, s f (r), can be performed by any
of the known techniquescathode polarization,
for example. These dependencies take the analogous character on tests performed just in hydrogeneous media (e.g., on tests in gaseous hydrogen
medium, hydrogen sulfide medium, or in the process of the electrolytic hydrogenation) (Ref 35,
56, 57).
In an analysis of dependencies s f (r), the
following characteristics are used as the main parameters:
The upper critical stress, r1, which corre-
sponds to the value of the tensile strength of

hydrogenated samples
The lower critical (threshold) stress, rth (i.e.,
the stress lower which the delayed fracture
process does not develop)
The time before the crack initiation, si (an incubation period)
The time up to final failure, s, which characterizes the prolonged strength of steel (Ref
35). Thus, a dangerous area of stresses
where the delayed fracture process develops
is an interval between the upper and lower
critical stresses. In this case the value of the
lower critical (threshold) stress, rth, represents a crucial important characteristic,
namely, the prolonged strength limit of steel
under the specific conditions of hydrogen embrittlement.
It is apparent that as the hydrogen content in
steel increases, all of the parameters mentioned
are changed. This especially affects the value of
the threshold stress, rth; the duration of the incubation period, si; and the time up to fracture,
s. From the data given in (Ref 58), for example,
it follows that as the hydrogen content in highstrength steel (1600 MPa, or 232 ksi) increases,
the duration of the incubation period successively decreases from several tens of hours to
several seconds. The time up to fracture in the
limiting case does not exceed 0.1 h. The duration
of the incubation period, si, is influenced by the
preholding under load including holding under
stresses r rth. Reduction of si, observed in
this case is due to the formation of sites with the
increased hydrogen concentration, which then
serve as points of the crack initiation (Ref 35).
Curves of the long-term strength of iron, and
steels Fe-0.2C, Fe-0.45C, and Fe-0.8C are given
in Fig. 6 (Ref 59). Samples have been polished
mechanically and annealed in vacuum. Delayed
fracture tests have been carried out while holding
the samples under load in the conditions of con-
tinuous cathode saturation with hydrogen in the

26% aqueous solution H2SO4 (the current density i 10 mA/sm2). Smooth thin plates with
the cross section of 0.8 3 mm (0.03 0.12
in.) that ensures the through saturation of samples with hydrogen have been used as samples.
From the results obtained it is evident that the
higher the hydrogen content in steel is, the more
tangible the decrease of the long-term strength,
and it does not practically take place in the case
of armor-iron and low-carbon steel Fe-0.2C. The
value of the threshold stress for steel Fe-0.45C
is of 150 to 170 MPa (20 to 25 ksi) that is several
times lower than the yield strength of this steel.
The lifetime of samples with the stress concentrator under load is mainly determined by the
time before the crack initiation (i.e., the duration
of the incubation period) (Ref 55). In the study
of crack growth it is worthwhile to construct socalled kinetic diagrams of cracking (i.e., dependencies of the crack growth rate, v, on the stress
intensity factor, K1. To construct kinetic diagrams of cracking, fatigue precracked samples
are used. The character of kinetic diagrams of
cracking is shown schematically in Fig. 7. Some
characteristic sections can be distinguished in
the diagram. At section I, which corresponds to
the threshold value of the stress-intensity factor,
K1th, the crack does not practically grow or
grows very slowly. Therefore, it is assumed that
the value of K1th characterizes the condition of
crack nonpropagation in metal. In section II, the
crack growth rate increases rapidly, whereupon
in region III, the crack growth rate over a wide
range, K1, is retained practically unchanged. Unlike the stage of accelerated, unstable fracture
(section IV), the stage of the slow, stable crack
growth can embrace the greatest part of longevity of a product. It is apparent that the delayed
fracture development at this stage is controlled
by the rate of transferring hydrogen to the region
of the crack top. At section IV, the crack growth
rate increases rapidly again, and at K1 K1C
(plane-strain fracture toughness) the final fracture occurs. The characteristic most sensitive to
the hydrogen content is the value of the threshold
coefficient of stress intensity, K1th. According to
data given in Ref 60, an increase of diffusely
moving hydrogen content in high-strength steel
Fe-15Cr-5Ni-2Cu-Ti from 0.14 104 to 2
104 at.% causes the value of K1th to reduce by
77%.
Besides the main characteristics just discussed, a number of other characteristics also
have been suggested for defining the sensitivity
to delayed fracture. For example, dependencies
of the initial coefficient of stress intensity, K10,
the time up to fracture, s, type are suggested to
be used (Ref 60). In this case long-term strength
curves can be described by the logarithmic relations of the K10 a ln s b type bounded
by the value K1C at the top and the threshold
value of the stress intensity coefficient K1th at the
bottom. In work (Ref 61) it was suggested to
construct the long-term strength curves in coordinates r 1/Zs, but not in conventional coordinates r lg s. In this case dependencies of

the long-term strength s f (r) are transformed
into straight lines convenient for further treatment according to the equation r rth
c(1/ s), where c is a constant. And finally, the
material sensitivity to delayed fracture may be
estimated by its sensitivity to the deformation
rate, namely, by the degree of their plasticity decrease under static loading at the rate of e
104 106 sec1. Such tests, for example,
are in common practice among Japanese researchers in making an estimate of the sensitivity
to the delayed fracture of high-strength maraging
steels (Ref 6264).
Mechanism of Crack Initiation and Its
Growth. It is the opinion of the authors who
studied a phenomenon of delayed fracture, that
in a qualitative sense the mechanism of the crack
initiation when the inner hydrogen embrittlement is as follows. For a crack to be initiated,
the critical concentration of hydrogen C* must
be created in local sites. The hydrogen redistribution in the solid solution is initiated by the
stress field produced by outer load. With the uniform distribution of tensile stresses, the value of
the chemical potential of hydrogen, l0, is the
same in the whole iron-hydrogen system; therefore, the hydrogen redistribution does not take
place. With the nonuniform distribution of
stresses, for example, because of the presence of
a concentrator, the pattern is essentially changed.
At the top of the concentrator (in the region of
maximum tensile stresses), the hydrogen chemical potential, l1, becomes less than that one in
the rest of the regions of the system, l2. The
appearance of the chemical potential gradient (l1
l2) brings about the diffusion of hydrogen into
the tensile stress region that with time gives rise
to the critical concentration of hydrogen C* and
the crack initiation. The critical concentration of
hydrogen C* is directly proportional to the
square root of the time of diffusion (C* C0
qt, where q is the coefficient of proportionality)
and must be so much higher than the level of
applied stresses is lower (Ref 61).
It is apparent that the increase of the hydrogen
content causes a decrease of the duration of the
incubation period, si, which is necessary for creating the critical concentration of hydrogen.
Lowering of the temperature gives the reverse
result (i.e., an increase of si caused by a decrease
of the hydrogen diffusion rate). Since to create
the critical concentration of hydrogen it is essential to localize tensile stresses, the crack initiation depends on the type of samples used. In
samples with no concentrator, the crack initiation occurs, as a rule, in the axial part of a sample
(i.e.. in the zone of maximum three-axis stresses
where the hydrogen concentration is also maximum). In samples with the sharp stress concentrator, the crack initiation takes place adjacent to
its top; in the case of the less sharp and deep
notches, the failure begins at a certain distance
from the notch top (i.e., under the surface) (Ref
35). At the moderate temperatures, hydrogen
gathers mainly on phase boundaries, microdiscontinuities, grain boundaries, piling-up dislocations, and other defects. Therefore, sites of or-
igin of fracture can be both the interphase

boundaries between ferrite and cementite and
nonmetallic inclusions as wellfor example,
sulfides, which are the most efficient traps for
hydrogen (Ref 56).
The crack growth kinetics upon the internal
hydrogen embrittlement also is controlled by the
hydrogen diffusion rate into the region of maximum tensile stresses ahead of the crack top. It
is suggested that this region is on the boundary
of the plastically deformed zone or at the much
less distance equal to the double opening of the
crack, d. As soon as the hydrogen concentration
becomes critical in the weakest element of a
structure, fracture of this element occurs, following which the crack formed merges into the main
one (Ref 65). The crack propagation is under
progress due to the multiple recurrence of this
process. Therefore, the crack growth can be considered as the process consisting of a sequence
of incubation periods, followed by the restricted
advance of the crack. The process of the crack
stepwise growth is displayed evidently under the
testing conditions at lowered temperatures when
the process of hydrogen diffusion is retarded
(Ref 35). The occurrence of the incubation period of the crack start and the stepwise advance
of separate sections of the crack front is evident
from the results of the registration of the acoustic
emission, which have been obtained from testing
of low-alloy steel with 0.4% C content (Ref 66).
In addition, the registration of the acoustic emission is indicative of the realization of more complicated mechanisms of the crack growth. For
example, when performing work (Ref 67) for
testing the high-strength maraging steel, a
growth of a great quantity of microcracks, 30 to
50 mkm in size, which appeared at 1 to 3 mks
intervals, was revealed.
Contrary to the internal hydrogen embrittlement, in the external hydrogen embrittlement the
crack growth kinetics is controlled mostly by the
following surface phenomena: adsorption, dissociation, and chemosorption of hydrogen on the
metal surface. The following experimentally observed effects serve as a guide for this statement:
The crack growth in the gaseous hydrogen
medium begins practically instantaneously

after applying an external load with no
marked incubation period.
The activation energy of the process at the
stage of the stable crack growth (region III,
Fig. 7) differs noticeably from the activation
energy of the hydrogen diffusion in a solid
solution and is comparable to the activation
energy of surface processes.
When gaseous hydrogen is doped with impurities, which activate (H2S) or suppress
(SO2), the process of the surface interaction
of metal with hydrogen, the crack growth rate
is also increased or reduced, respectively (Ref
57). The location of the failure zone ahead of
the growing crack front depends also on the
type of hydrogen source. In case of the external hydrogen embrittlement, the failure
zone, as a rule, does not reach the point of the
maximum tensile stresses (i.e., it is located

much nearer to the crack top than it is at the
internal hydrogen embrittlement) (Ref 65).
The extent of the embrittlement in the medium
of gaseous hydrogen depends both on the pressure and the chemical composition of a medium,
in part, of the steam content and oxygen, as well.
In the case of high-strength steels a dramatic effect of the hydrogen pressure on the crack
growth rate has been revealed at pressures not
exceeding 1.3 103 MPa (Ref 68). The presence of steams in the hydrogen atmosphere has
a quite profound initiating effect on the delayed
fracture development. Oxygen has the opposite
effect of efficacious retarding of the crack
growth up to the critical value. On data given in
Ref 69, for example, a mere 0.6% of O2 is sufficient for the complete stop of the crack development in the medium with humidified hydrogen
(steel with the yield strength of 1600 MPa, or
232 ksi). The crack stopped by oxygen begins to
grow again only after the complete removal of
oxygen from the medium of tests. It is suggested
that oxygen forms oxides that prevent the hydrogen from penetration into the crack top, and after
the removal of oxygen the reverse process occurs
(i.e., oxides are restored by hydrogen) (Ref 70).
The temperature factor should be set off from
other factors that affect the character of delayed
fracture. According to the data in hand, the maximal sensitivity of steels to delayed fracture takes
place at temperatures close to, room temperature
(Ref 29). But in a number of cases this regularity
is not obeyed. In Ref 71, the temperature dependence of the delayed fracture of the high-strength
steel Fe-15Cr-5Ni-2Cu-Ti containing 5 104
at.% of hydrogen was studied. The threshold
value of the stress intensity coefficient for this
steel is K1th 46 MPa m1 / 2. At 20 C (68 F)
the crack growth occurs at the rate of 2 107
m/s. At 100 C (212 F) the crack growth rate
up to the critical value increases by more than
300 times and reaches 6.6 105 m/s. In this
case the time up to fracture (s) reduces from 500
to 600 min at 20 C (68 F) to s 5 to 6 min
100 C (212 F). At the temperature of 180 C
(356 F) the development of delayed fracture
stops completely. The calculation of the activation energy of the crack growth Q has shown
good agreement between the obtained value
(64.8 kJ/mol) and the activation energy of the
hydrogen embrittlement (60.9 kJ/mol).
Chemical Composition and Structure Influence. The resistance of steels to delayed fracture depends on the carbon content, alloying, and
impurity elements. From the data given in Fig.
6, it is evident that a buildup of the carbon content is attended by the significant reduction in
the resistance to delayed fracture. This effect derives from the fact that with increased carbon
content the strength properties of steel are improved, the level of residual internal microstresses is raised, and the amount and volume of
the carbide phase are increased. The interphase
surface area (i.e., the number of places, or traps,
for the accumulation of hydrogen and the crack
initiation) is correspondingly increased (Ref 29).

The influence of alloying elements is also
quite profound. The effect of noncarbide-formative alloying elements (Mn, Al, Si, Co, Ni) is
mainly connected with their influence on the
properties of matrixthat is, with the change of
its resistance to the viscous, brittle, and grainboundary fracture. These elements (but Ni) decrease the long-term strength of steels subjected
to quenching followed by high tempering (Table
1). Carbide-formative elements (Cr, Mo, Ti, Nb,
V) have an influence mainly through the change
of the amount of the carbide and carbide-nitride
phase, its distribution, and morphology. The
positive influence of molybdenum, niobium, and
vanadium, which bring about to an abrupt increase of the time up to fracture, s, is most pronounced (Table 1). In the case of noncarbideformative elements there is, as a rule, a direct
relationship between the resistance to the hydrogen embrittlement and standard characteristics
of brittle fracturefor example, the critical temperature of brittleness, T50. In the case of carbide-formative elements such an unambiguous
relationship usually is not observed (Ref 72).
As to impurity elements, phosphor and sulfur
have the most negative influence (Table 1). The
negative influence of sulfur is connected with the
formation of manganese sulfides MnS, which,
just as indicated, are the most effective traps for
hydrogen. Manganese sulfides assist in initiating
cracks at more low critical concentration of hydrogen C* in the region of maximal tensile
stresses as evidenced by the results of the reg-
istration of hydrogen in the fracture zone (Ref

73). In this case, a decrease of sulfur to a degree
less than 0.005 wt% does not protect against the
development of the hydrogen embrittlement.
The form of nonmetallic inclusions also has a
dramatic influence, especially in steels with the
ferrite-pearlite structure. Inclusions of the spherical form are considered to be the safest ones.
The negative role of sulfur can be connected in
that this element has a retarding effect on the
recombination of hydrogen atoms on the steel
surface and so brings about an increase of the
amount of absorbed hydrogen (Ref 74).
Phosphorus and its analogs (Sb, Sn) are the
most dangerous on formation of segregation at
grain boundariesthat is, in the process of thermal treatment, which results in the development
of the temper embrittlement (Ref 7577). In the
absence of segregations the steels fracture in hydrogeneous media connected with the intensive
plastic deformation in the crack top. Fracture
mainly occurs transgranularly (i.e., along a
grain). In this case, in fractograms of the crack
growth up to the critical value, either a viscous
cup fracture or a quasi-spall takes place. During
the formation of segregation another mechanism,
the brittle fracture along grain boundaries, is realized with a consequent abrupt decrease of K1th.
In this case, an intensive reduction of K1th takes
place already at initial stages of the segregation
formation when temper embrittlement has hardly
shown up. It is considered that one of the reasons
Table 1 Influence of alloying elements and impurities on transition temperature, T50; work
of propagation of a ductile crack, A; plasticity loss after hydrogenation, Fw and resistance
to hydrogen embrittlement, s (chromium-molybdenum steel, 0.2% C)
Element (limits of
alloying, mass, %)
Dependence of fracture characteristics on

element concentration (per 0.1% of each)
Optimal content of
the element, mass, %
T50, C
A, MJ/m2
Fw%
s, h
for such a strong influence of phosphorus, antimony, and tin segregation is a buildup of hydrogen concentration on the grain boundaries as a
consequence of the chemical interaction of hydrogen with these impurities (Ref 75).
The steel structure has a determining effect on
the resistance to delayed fracture. The least resistance is typical for the martensitic structure,
especially at low tempering temperatures. An increase of the tempering temperature to 420 C
(788 F) causes the value of rth to increase by
several times. The decrease of the volume fraction of martensite to 20 to 30% means to that
even in the hardened state delayed fracture does
not develop (Fig. 8) (Ref 78). In Ref 54 attention
is drawn to the increased sensitivity of steel Fe0.4C-Cr-Si with the lower bainite structure to delayed fracture, although this structure is characterized by high-impact toughness and adequate
plasticity. Unlike the martensitic structures,
whose failure process develops mainly along
boundaries of former austenitic grains, the bainite structure fails along the grain body (by mechanism of quasi-spall).
On evidence given in Ref 74 the resistance to
the hydrogen embrittlement of low-alloyed pipe
steels increases in the direction from normalization to the controlled rolling, but the best results are ensured by the thermal improving
(quenching high tempering) and quenching
from the intercritical temperature interval. A
comparison shows that the thermal improving
and quenching from the intercritical interval ensure about the sixfold enhancement of the longterm strength as compared with the normalization. Water quenching at 775 C (1425 F) when
the structure contains approximately 30% of
bainite affords the best combination of properties. The marked enhancement of the resistance
to delayed fracture is provided by the high-temperature thermomechanical processing of the
C (0.2 to 0.8)
0.20.3
20 to 60
0.1 to 0.2
32
Solid solution
Si (0.2 to 1.8)
0.40.7
4 to 6(a)
5 to 8(b)
3(a)
7 to 16(b)
4 to 10
1 to 1.5
9
0.12
0.05
0.05
0
0
0.17
3
2
2
7(a) to 1.3(b)
1.2
4
5
15
23
6.5
5.7
7.5
0.04
0.01
0
2.9
1.5
8
80
1
260
0
0
0
36(a) to 10(b)
12
10
250
250
400
0.04
0
35
70
2.4
1.0
2.5
2.5
0.04
0
80
360
500
32
1.2
Ni (0 to 3.0)
Co (0 to 3.0)
Al (0.03 to 0.7
0.51
0.5
0.25
Carbide-forming
Cr (0.5 to 3.0)
Mo (0 to 1.0)
0.4 to 0.5
Ti (0 to 0.15)
Nb (0 to 0.20)
V (0 to 0.4)
0.05
0.02 to 0.06
0.1
1(a)
2 to 5(a)
15 to 30(a)
3 to 5(b)
60(a) to 14(b)
0(a) to 40(b)
20(a) to 15(b)
Modifiers
Rare-earth (Ce)(0 to 0.5)
AIN, VN, NbN (0 to 0.2)
0.1 to 0.3
0.2
10
40
Impurities
S (0.015 to 0.045)
P (0.004 to 0.026)
Sb (0.0006 to 0.027)
Sn (0.0005 to 0.03)
Cu (0 to 1.5)
11.5
1400
5
1200
Stress , MPa
Mn (0.6 to 2.5)
1600
1000
800
600
2
200
1
0.01
0.015
0.01
0
0.5
0
230
180
180
1
(a) Alloying for optimal content. (b) Alloying over optimal content. Source: Ref 143
2
3
Time to failure , h
Influence of martensite content and tempering

temperature on delayed failure of quenched carbon steel from 0.35% C. 1, 100% of martensite, 0.023% of
S, without tempering; 2, the same, tempering at 200 C; 3,
the same, tempering at 420 C; 4, 7080% of martensite,
0.025% of S, tempering at 200 C; 5, 7080% of martensite, 0.005% of S, tempering 200 C; 6, 2030% of martensite, 0.026% of S, without tempering. Source: Ref 78
Fig. 8

high-strength steel Fe-0.3C-2Ni-Cr-Mn-Si (Ref
79).
Some Specific Cases of Delayed Fracture.
In the scientific literature, including the educational one, it was a widespread opinion that delayed fracture of steels can develop involving no
hydrogenthat is, under the effect of only structure factors (Ref 28, 54, 80). In particular, the
question about the nature of the enhanced sensitivity to delayed fracture of steels in the socalled newly quenched state is long discussed.
Cases of quenched products cracking, which are
met in practice of thermal processing (Ref 52,
54), have been associated with the development
of delayed fracture of untempered martensite.
Initiation and growth of crack provoking delayed
fracture occurs mostly over the boundaries of
primary austenitic grains. The short-term tempering or resting of quenched samples at room
temperature enhances resistance to delayed fracture; therefore, its development calls for further
hydrogenation. Within the span of more than 50
years of investigations, alternate hypotheses for
explaining this phenomenon have been proposed. The structure hypothesis based on the
prevailing role of residual microstresses inherent
in the untempered martensitic structure has received the most acceptance (Ref 79, 80). The
role of internal hydrogen, which is always present in steel as an inevitable impurity, has not
been discussed practically in the literature.
The phenomenon of delayed fracture of
quenched steels has been studied in detail in Ref
8183. Steels Fe-0.3C-Cr-Mn-Si and Fe-0.4CCr-Mn quenched in oil at 860 C (1580 F) have
been chosen as objects of investigation. The pro-
cess of initiation and growth of a crack when

testing on a delayed fracture in a freshly hardened state is quite the same as that in the grain
boundaries. The typical initiating crack found in
the stress concentrator top is shown in Fig. 9.
The work has been methodically performed in
the following way. First, regularities of the initiation and crack growth have been studied thoroughly by methods of mechanical tests in conjunction with the acoustic emission. In
particular, the effect of the deformation-rate inhibition of the crack growth has been established, durations of the incubation period, si, under different conditions of loading have been
determined and so on. Second, the simulation of
the newly quenched state has been performed.
For this purpose, after approximately 70 h exposure at room temperature (to eliminate the sensitivity to delayed fracture) quenched samples
have been hydrogenated by the special procedure.
Subsequent tests have shown that all the characteristics of delayed fracture of quenched
steelsthe threshold stress, rth, the duration of
the incubation period, si, and so on, as well as
the mechanism of fracture, are analogous with
the case of delayed fracture caused by hydrogen.
Therefore, it is beyond the reason to consider
that in the case of the untempered martensite
structure a certain different type of delayed fracture that is not connected with the presence of
hydrogen in steel takes place. The fact that the
kinetics of relaxation of interior microstresses in
the process of resting at room temperature does
not correspond to the kinetics of the change of
the threshold stress, rth, bears witness to this
also. Judging by the change of the interior fric-
80
3
, %
60
40
1
20
2
0
2500
3
B, MPa
2000
1500
2
1
1000
500
340
380
420
460
500
540
T, C
Nucleating grain boundary crack revealed at the
top of the stress concentrator on testings of
quenched steel Fe-0.3C-Cr-Mn-Si for delayed failure.
1100. Source: Ref 81
Fig. 9
Influence of testing media on delayed failure of

steel Fe-18Ni-9Co-5Mo-Ti. 1,1 testings in the
air, u 4045%; 2.2, the same, u 6570%; 3,3,
vacuum testings, P 1.3 Pa. Source: Ref 86
Fig. 10
tion, the microstress relaxation process completes in a time of no more than 10 to 20 h since
quenching. In this state a material still possesses
the sensitivity to delayed fracture. In the process
of holding samples at room temperature, the evolution (desorption) of hydrogen from them into
the atmosphere has been found to occur. The
sensitivity to delayed fracture is just concurrently with termination of hydrogen desorption
(i.e., about 70 h after quenching has been
elapsed). According to the data of the laser spectroscopy, the amount of desorbent hydrogen over
the period of holding comes to approximately 2
106 at.% (Ref 84).
The source of diffusive-mobile hydrogen in
the structure of the quenched steel is inner hydrogen, whose amount at conventional process
of melting makes up no less than 2 to 3 104
at.%. The lattice solubility of hydrogen increases
from approximately 107 at 20 C (68 F) to 4.4
104 at.% at 860 C, (i.e., by about 3 orders
of magnitude), as a result of heating to the austenization temperature. A reverse decrease of
solubility in the process of the fast quenching
results in a part of hydrogen being kept for some
time in the oversaturated state in the -solid solution and has the increased diffusive mobility
and the property of localizing in the region of
maximal tensile stresses. High interior microstresses inherent of the quenched steel structure
make the process of the crack initiation and the
crack growth easier but are not the factors that
control kinetic parameters of delayed fracture.
Another interesting case is delayed fracture of
high-strength maraging steels, which develops
over the certain temperature interval of aging
400460 C (Ref 85). The development of delayed fracture in these steels requires no special
hydrogenation. Therefore, within a long period
of time this phenomenon was also connected
only with the peculiarities of the structure state
forming at indicated aging temperatures. Furthermore, this problem is urgent because aging
in the indicated interval provides for obtaining
the maximum elastic limit of steel. However,
owing to the low resistance to delayed fracture,
this mode of aging usually is not practiced. An
analysis of data in hand has shown that the key
question in explaining the phenomenon of delayed fracture of maraging steels is a question on
the role of the testing environment (air, vacuum)
as well as the role of titanium (Ref 85).
The investigations have been carried out with
steels Fe-16Ni-10W-Mo-Ti and Fe-18Ni-9Co5Mo-Ti (Ref 86, 87). The tests have been performed under static loading at the reduced rate
of e 2.8 105 s1. Dependencies of the mechanical properties of steel Fe-18Ni-9Co-5MoTi on the aging temperature when testing in the
air and in vacuum are presented in Fig. 10. A
crevasse in curves of the change of mechanical
properties on tests in the air (curves 1,1) corresponds to the aging temperature of 420 to 460
C. Plastic properties of steels in this range come
practically to the zero level. An effect of delayed
fracture for steel Fe-16Ni-Mo-10W-Ti is observed within a more narrow temperature inter-

val at 420 C. On testing in vacuum (residual
pressure of 1.3 Pa) the indexes of mechanical
properties abruptly increase (curves 3,3). An effect of the delayed fracture elimination is due to
the change of the crack-initiation mechanism.
On testing in the air the brittle grain-boundary
fracture occurs, on vacuum testing the viscous
inter-granular fracture takes place (Fig. 11). In
the study of the titanium influence it has been
found that the influence of the testing environment takes place only with a definite content of
titanium in steel exceeding 0.5%. When titanium
content is 0.5%, mechanical properties tests in
the air and vacuum tests do not practically differ
from one another (i.e., delayed fracture does not
develop).
It has been found also that resistance to delayed fracture depends on moisture of the environmental atmosphere (the laboratory air),
namely, on a season in the course of which tests
are carried out. The figure shows that an increase
of relative humidity (from 4045% to 6570%)
Fig. 11
results in the deterioration of plasticity and the

reduction of the tensile strength by 30% (curves
1,1, 2,2). Thus, tests have been carried out in
the air-dried chamber, for which purpose the special procedure has been devised. On testing in
the air-dried atmosphere (u 0.04%) plastic
and strength properties of steel Fe-18Ni-9Co5Mo-Ti come to such a degree that corresponds
tests in vacuum (i.e., the effect of delayed fracture disappears). This means that the structure
state formed on aging within the interval of 400
to 460 C possesses the increased sensitivity to
the presence of water vapors in the air environmentthat is, delayed fracture of high-strength
maraging steels is caused by the effect of atmospheric moisture adsorbed on a surface.
Investigations of grain boundaries of fractured
samples by x-ray electron spectroscopy have
shown that on the grain boundaries there are hydrides, first of all, titanium hydrides, TiHx, apart
from apparent components (segregations of alloying elements, oxides, hydroxides, and inter-
Mechanism of failure of steel Fe-16Ni-10W-Mo-Ti at crack initiation stage (aging at temperature of 420 C).
(a) Testings in the air. (b) Vacuum testings (P 1.3 Pa). 1050. Courtesy of A.I. Kovalev
H3O
102
H2O
10-20
Oxide-hydroxide layer
P
Grain 1
~10
H3O
Ni3Ti
Hads
Grain 2
Grain 1
Oxygenless zone
Grain boundary
(a)
Fig. 12
(b)
Grain 2
Ti segregation
Ni3Ti
H2O
Air
Habs
TiHx
Grain boundary
Mechanism of the crack growth on delayed failure of maraging steel. (a) Stage before crack initiation. (b) Stage
of crack growth. Source: Ref 87
metallics). Since the process of hydride linking

of the Ti-H type in the air is practically excluded
(titanium, by virtue of its enhanced activity, in
its interaction with oxygen and moisture in the
air forms first of all the hardly permeable oxide
phase TiO2), the conclusion has been drawn that
titanium hydrides are formed on inner interfaces
(grain boundaries). In other words, the TiHx formation occurs in the zone of tensile stresses
ahead of the crack top as diffusive-mobile hydrogen enters there.
According to the mechanism suggested (Ref
86), the process of grain boundary embrittlement
includes the following successive stages: adsorption of atmospheric moisture at the metal surface, oxidation of the most electronegative steel
components (iron atoms), transport of H3O
ions in the film of adsorbed water toward the top
(mouth) of the growing crack, formation of
atomic hydrogen H0ads on the freshly formed surface by cathodic reduction of H3O ions on
nickel atoms, transition of atomic hydrogen
H0ads into the absorbed state and subsequent diffusion of hydrogen to the tension-stressed zone
ahead of the crack top, and embrittlement of
grain boundaries due to both the formation of
titanium hydrides and segregations of hydrogen
solutes ahead of the crack top. Thus, the final
stage of the process is the formation of diffusivemobile hydrogen, which causes embrittlement
owing to both the formation of titanium hydrides, TiHx, and the direct action (decohesion).
This is the main difference of the studied type of
brittleness from the delayed failure of the
quenched steel, which is caused by internal diffusive-mobile hydrogen formed in the process of
steel heat treatment.
The mechanism of the crack growth, which
illustrates the sequence of stages under consideration is represented schematically in Fig. 12
(Ref 87). The formation of titanium hydrides in
the crystal lattice is attended with the considerable volume effect, which is about 3 times higher
than for the martensite transformation in steel.
Besides, hydrides favor the crack growth owing
to the reduced strength of the hydride-matrix interface (Ref 28). Atomic hydrogen H0ads formed
in the crystal lattice is ionized producing ions
H and H, which quantitative ratio, on data
given in Ref 88, makes up 1:9, approximately.
The rate of diffusion of ionized hydrogen is 1 to
2 orders of magnitude higher than that of atomic
hydrogen. According to data obtained by the
method of the secondary ion mass-spectrometry
(Ref 88, 89) more mobile ions H, which serve
as main carriers of diffusive-mobile hydrogen in
metal, are responsible for the formation of the
hydride phase and the process of embrittlement
of the grain boundaries. Ion H possesses the
lesser mobility in the crystal lattice, and its contribution to the embrittlement is not so considerable essential.
In Ref 90 and 91, delayed failure of high-alloy
steel Fe-24Cr-7Ni-3Mo-Al-Ti containing no
martensitic structure has been studied. This steel
displays the enhanced sensitivity to delayed failure in the so-called newly-smelted state (i.e.,

under the effect of metallurgical hydrogen determining the role of which in the development
of delayed failure is of great practical importance). The studied steel structure is embrittled
d-ferrite with austenite c-phase interlayers uniformly distributed in it. Unlike the cases discussed previously, the initiation and the crack
growth in the structure under study occur in accordance with a transgranular mechanism, with
defects of the cast structure (pores) of 20 mkm
in size being the source of fracture. As evident
from the registration of the acoustic emission,
the crack grows according to the mechanism of
the discrete (stepwise) growth of separate sites
of the crack front followed by the stop at viscous
sites (austenitic inter-layers). According to observations, the resistance to delayed fracture depends on the period of the natural recreation after
smelting. Within the span of 4 years the metallurgical hydrogen content drops from approximately 18 104 to 6 104 at.%, and it is
a limiting one at which the development of delayed fracture is still possible.
The object under study is of interest as far as
it allows the different mechanisms of the embrittlement of grain boundaries or their volume
to be realized. The thermal treatment has been
carried out under operating conditions including
the high-temperature preheating to 1250 C
(2282 F) for obtaining the one-phase structure
of d-ferrite and postaging (annealing) at different
temperatures. The hydrogen content in thermally
treated samples does not exceed 1.4 104
at.%. Annealing at 680 C (1256 F) makes it
possible to obtain the brittle fracture along the
grain boundaries, which, on data of the auger
spectroscopy, arises from the segregation of sulfur and phosphor impurities. After the 850 C
(1562 F) annealing on the grain boundaries of
ferrite, the abundant precipitation of embrittling
r-phase responsible for the quasi-brittle grain
boundary fracture takes place. Aging at 440 C
(824 F) makes it possible to obtain the brittle
intergranular fracture in accordance with the
transgranular mechanism.
The tests performed on embrittled samples
have once again favored the view that structure
factors bringing about the embrittlement of the
grain boundaries or their volume cannot in themselves result in the development of delayed fracture. As is obvious from this data obtained from
steels with the different structures, the development of delayed fracture is only possible with
the participation of diffusive-mobile hydrogen
independently of the source of its origination,
namely on quenching, smelting, from atmospheric moisture, and so on.
Stress-Corrosion Cracking of LowAlloy Steels

Stress-corrosion cracking (SCC) is a specific
type of delayed fracture that develops under the
concurrent action of tensile stresses and the corrosion medium. The given type of fracture occurs in chemical, gas-oil producing, metallurgic,
shipbuilding, and other branches of industry, and

accounts for approximately 30% of material
damage caused by corrosion. The most dangerous media that can cause SCC under the certain
conditions have been established for most steel
classes used. The greatest body of information
accumulated to date refers to high-alloy steels,
in part, to corrosion-resistant steels with austenitic structure. Stress-corrosion cracking of lowalloy steels has been studied least, although such
steels, especially in the high-strength state, have
an enhanced tendency to corrosion-mechanical
fracture. Low-alloy steel cracking can occur under the action of various actuating media including solutions of acids, nitrates, alkali, and sulphur-hydrogeneous media as well. The low
resistance of high-strength steels to SCC makes
itself evident even in such, at first sight, undangerous media as pure water, moist air, and water
vapors (Ref 92, 93).
One of the characteristic peculiarities of SCC
is a pronounced selectivity of a material relative
to a corrosion medium. Therefore, the results of
one metal-medium system investigation cannot be extended to another system. Besides, the
noncorrodive media, in common conception of
this term, may appear to be the most aggressive
ones, because the corrosion cracking process is
actively developing in them under the effect of
tensile stresses. Another interesting peculiarity
lies in a usual insensitivity of high-pure metals,
including iron, to SCC in any environment. The
susceptibility of iron to cracking under the effect
of stresses appears only in the presence of introduction impurities, namely, carbon and nitrogen.
The response to the cold plastic deformation is
also intriguing. Unlike austenitic corrosion-resistant steels, the cold plastic deformation of carbon steels makes them relatively insensitive to
SCC. Annealing of the cold-deformed 0.15% Ccontaining steel at 600 C (1112 F) restores this
sensitivity to the level of nondeformed steel (Ref
70).
Peculiarities of Crack Initiation and
Growth Mechanism under SCC. The mechanism of initiation and growth of cracks under
SCC depends on the steel structure, the value of
applied stresses, and the testing medium. Fracture along the primary austenitic grain boundaries is the most dangerous from the standpoint
of loss in corrosion-mechanical strength. Carbon
steels, as well as many of the low-alloy steels,
fail in alkaline, nitrate, phosphate, carbonate solutions, and also in aqueous acid solutions by
such a mechanism (Ref 75). Depending which
reaction controls the cracking processelectrochemical, anodic, or cathodictwo main SCC
mechanisms are differentiated: the active-part
corrosion (APC) or the hydrogen-induced corrosion cracking (HISCC). Kinetics of the anodetype cracking is determined largely by the conditions of passivation (i.e., by the peculiarities of
the protection corrosion films formation and by
their properties). This is reflected in the fact that
the region of danger of the given type of SCC
corresponds to the so-called active-passive region of potentials in the polarization curves. The
region of cracking by the HISCC mechanism has

no clearly defined boundaries and extends from
the potential of free corrosion to the potential of
cathodic protection (Ref 94).
The mechanism of part cracking presupposes
the accelerated metal dissolution, which is considered to be connected with the local disruption
of passivity at the crack top i.e., with the protective film breakdown and occurrence of the newly
formed (juvenile) surface. The mechanism of
HISCC involves the following main reactions:
discharge of hydrogen H3O ions from water
solution onto the steel surface, the transport of
hydrogen to the region of the elevated stress concentration ahead of the crack top, and the crack
propagation caused by the local hydrogen embrittlement. The concrete mechanism of the hydrogen embrittlement depends on the structure
and properties of steel, the loading conditions
and the ability of a medium to hydrogenation
(Ref 94).
Whatever the mechanism of corrosive-mechanical fracture is, the three stages of the process are distinguished: incipient defects (microcracks) formation, microcracks growth, and fast
fracture as soon as the main crack reaches the
critical size. Both metallurgical defects of the
surface and defects of corrosion origin may present as incipient defects (Ref 93). In the SCC
theory, particular attention is given to the processes proceeding in the crack top, which are
much different from the processes proceeding on
the surface. In particular, an important role is
given to the newly formed (juvenile) surface,
which appears either under the effect of the external loading or in the process of APC. All the
particular SCC reactions, namely, APC, cathode
precipitation of hydrogen, its absorption, and so
on, are sharply speeded up on such a surface. The
newly formed surface possesses a more negative
potential compared with both the previous surface and the passivated surface of crack edges.
It is suggested that in the corrosion crack there
is a local galvanopair of the newly formed surface-crack edges surface type that contributes to
the accelerated selective crack propagation (Ref
95).
Criteria and Methods of Estimating Sensitivity to SCC. A great body of different criteria,
namely, mechanical, electrochemical, physical,
and so on, has been suggested to estimate the
sensitivity of steels to SCC (Ref 92). All the
known methods of corrosion-mechanical testings can be divided into three main groups:
Testings under constant external load (P
constant)
Testings under constant strain (e constant)
Testings at the constant load rate (e, v constant).
Samples without a notch, with a notch, and with
the precreated fatigue crack can be used in each
of these above-stated methods.
The service conditions are best reproduced by
corrosion-mechanical testings under constant external load. In this case, as the crack initiates and
grows, the stress rises in the rest part of a sample
that contributes to the development of fracture.

By analogy with delayed fracture (see the section
Delayed Fracture in Steels), the sensitivity to
SCC is characterized by the time till a crack initiation, si, durability till complete failure, s, and
the value of the threshold stress, rth. Unlike delayed fracture, it often occurs that dependencies,
s f (r), are not brought out onto the plateau,
when testings for corrosion cracking are performed (i.e., there is no threshold stress). The
conventional threshold stress, rth, which should
be determined in this case, is determined at the
specified (base) duration of testings. The duration of the incubation period si during which the
crack initiation occurs takes usually 80 to 90%
of the whole time prior to fracture (s) (Ref 96).
Testings under constant (fixed) strain are usually carried out according to the scheme of threeor four-point bending of samples fixed in the
rigid attachment. On such testings the external
load is reduced gradually because of the crack
formation and the residual bending strain accumulation. Testing under constant strain render
well the cracking processes in the material structure that develop under the action of residual
stresses.
Tests at constant load rate are mainly performed by the slow strain rate testing procedure
(SSRT). In this case the resistance of a material
to cracking is characterized by the value of tensile strength in the corrosion medium, characteristics of plasticity, W, d, the critical fracture
stress, r*, and the area below the strain curve, r
e, characterizing the fracture work. With the
plasticity characteristics taken into accounts, the
relative (as compared to the air) sensitivity of a
material to corrosion cracking is determined.
Used as criteria in this case are the so-called coefficients of medium effect of the (W0 W1)/
W0 type, where W0 and W1 are the relative narrowing on testings in the air and in the corrosion
medium, respectively. The strain rate e with the
use of the SSRT method is usually selected
within the range of 104 to 106 sec1.
When using samples with a crack, a load, the
load line displacement, the crack length, and the
crack opening displacement (COD) are registered as per the SSRT method. For analyzing the
test results, the criteria of fracture mechanics are
used. The threshold (critical) stress-intensity factor, KISCC, which makes it possible to estimate
quantitatively the steel resistance to cracking in
the corrosion medium is taken as the main criterion. The value of KISCC corresponds to the beginning of the crack growth before the critical
value (Ref 97). It is apparent that if K1 KISCC
or r rth, the fracture of the construction element inevitably takes place. When the criterion
KISCC is used, the sensitivity of material to the
action of the corrosion medium is characterized
by the ratio KISCC /KIC, where KIC is the stressintensity factor when testing in the air (planestrain fracture toughness) (Ref 98100). In all
the schemes of testings the maximum length of
a crack resulting in fracture, the crack growth
rate, the number of cracks per unit surface, etc.,
can be used as characteristics of a material (Ref
101, 102).
The most reliable values of KISCC are obtained

on the basis of standard kinetic cracking resistance diagrams analogous to that represented in
Fig. 3. The plateau-figurative regions in kinetic
diagrams are connected with the branching and
blunting of a crack. If relative resistance of materials to the action of corrosion media is to be
determined quickly, the SSRT method has an advantage over the constant load testing method for
determining KISCC, although it is considered the
basic one. The fact of the SSRT method high
efficiency is evidenced by the data of many scientists. The console samples made of Fe-0.3C2Ni-Cr-Mn-Si steel, for example, have been
tested in the NaCl 3.5% solution (Ref 103). The
values of KISCC obtained at a constant rate of load
and on standard testing (under constant load)
have turned out practically to be equal to 11.69
and 1196 MPa m1 / 2, respectively. To choose the
proper load rate is of great importance in this
case. At the high load rate the corrosion process
has no time to produce changes at the crack top;
at low rate, on the contrary, the passivation process develops at the crack top, thus preventing
from corrosion cracking propagation (Ref 96).
The linear mechanics of fracture are unsuited
for high-ductility steels. In this case, resistance
to crack growth is characterized by the value of
the J-integral, in particular, its threshold value,
JISCC. A tendency to corrosion-mechanical
cracking is estimated also by electrochemical parameters, namely, by the critical value of the hydrogen index of the medium pH, the electrode
potential, E and the density of anodic current at
which the cracking process begins. When testing
the carbon steel in the 3% solution of NaCl,
strong jumps of the electrode potential (300
mV) and the anodic current at the moment of a
crack propagation have been detected (Ref 104).
This effect is connected with revealing the juvenile surface, which has an enhanced electrochemical activity and it can also be taken as one
of criteria for the sensitivity of steels to corrosion-mechanical fracture. The important information on kinetics of SCC, especially at initial
stages of the process, can be obtained from
acoustic emission measurements (Ref 105107).
Effect of Alloying Elements on Resistance
to SCC. The effect of most alloying elements on
steel resistance to SCC depends on their concentration, the presence of other alloying elements,
impurities, the corrosion medium content, the
value of electrode potential, and other factors
(Ref 108). For high-alloy austenitic steels there
is a certain relationship between the effect of an
element and its position in the periodic system.
The analogous relationship is not observed for
low-alloy steels (Ref 109). The theory of alloying that makes it possible to predict the resistance of low-alloy steels to SCC in different media also has not been developed. Hypotheses
suggested provide, as a rule, an explanation for
the role of alloying elements by only one of
some factors, namely, effect on strength of interatomic bonds, the rate of forming protective
films, the effect on the carbon distribution in
steel, and so on (Ref 108).
Information on the effect of alloying elements

on mechanisms of the crack growth in different
corrosion media is necessary to develop successfully new grades of steels with an enhanced
resistance to corrosion cracking. For lack of this
information, the empirical data are usually used
which are often contradictory. The findings on
the individual and combined effects of 27 alloying elements and impurities on the resistance to
SCC of iron and low-alloy steels, including highstrength steels, have been generalized in Ref
108. The effect of elements depending on the
medium (aqueous solutions of chlorides, nitrates, carbonates, alkali, hydrogen sulfide, etc.),
the temperature, the electrode potential, and the
pH value have been considered. The following
elements have been analyzed: N, Al, B, V, H,
Ge, Ca, O, Co, Si, La, Mn, Cu, Mo, As, Ni, Nb,
Sn, Pb, S, Sb, Ti, C, P, Cr, Ce, Zn. It has been
shown that the majority of these elements, especially in the case of combined alloying, ambiguously affects resistance to SCC. The same
element can have both a positive and negative
effect depending on the corrosion medium. Silicon, for example, enhances steel resistance to
cracking in nitrate solutions but reduces it in alkali ones. However, the previously listed elements can be divided arbitrarily into four main
groups: those that enhance the resistance to SCC
(Ti, Nb, Mo, Al, Co, Cu, Pb, B, Ce, Ca, La),
those that reduce it (N, O, H, S, P, V, Sb, Cr,
Mn, Ge), elements of the alternating effect (C,
Si, Ni), and neutral ones (Sn, As, Zn).
The characteristic properties of distribution of
elements and impurities in the steel structure are
of great importance. Segregations of sulphur and
phosphor impurities, for example, change the
electrochemical characteristics of the grain
boundaries and initiate fracture by the local parth
corrosion mechanism (Ref 108). Steels produced
by vacuum melting possess the enhanced resistance to SCC in so far as they contain 2 to 3
times as few detrimental impurities as standard
steels do (Ref 110). It has been noted that the
effect of alloying elements can be judged by the
mechanism of fracture. Elements with a positive
effect assist in realizing the transgranular mechanism of the crack growth. Those that have a
negative effect cause cracking along the grain
boundaries. The effect of the chemical composition may manifest itself indirectly, namely,
through the change of the structure and mechanical properties of the steel. Therefore, the thermal processing can drastically change degree
and even the direction of some alloying elements
effect.
Role of Structure and Thermal Processing
in the SCC Process. The basic parameters of the
SCC, namely, the threshold stress, rth, corrosion
crack-toughness, KISCC, the crack growth rate,
and so on, depend to a great extent on the following structure factors: phase composition,
austenite grain size, state of grain boundaries,
and nature and amount of nonmetallic impurities
(Ref 111). Figure 13 shows that the crack-tough-

ness reaches its peak values KISCC in hightempered steels. In the region of high-strength
states realized at low tempering, the parameter
KISCC takes quite low values not exceeding 20
MPa m1 / 2.
As for steels of mean and low strength, the
structures of fine-grain ferrite, martensite, and
bainite possess the best properties. The spheroidized structure with uniformly distributed finegrain carbides possesses somewhat worse properties, and the lamellated pearlite structure
possesses the worst ones (Ref 112). There is also
evidence that compound structures, for example,
consisting of fine-grain ferrite and bainite
wherein the intergranular ductile mechanism of
failure is realized (Ref 113, 114) show the highest resistance to SCC. For the steel Fe-0.3C-MnSi-Ni with the compound structure of martensite
lower bainite, the value KISCC in the NaCl
3.5% solutions is 21 MPa m1 / 2. In the case of
the martensitic structure this index is reduced by
half (i.e., steel with this structure possesses the
much lower resistance to SCC) (Ref 115).
The positive role of residual austenite is observed frequently. In the scope of the SCC model
by the hydrogen embrittlement mechanism this
derives from the fact that the saturability of austenite with hydrogen is essentially higher, and
the hydrogen diffusion factor is three to four orders of magnitude lower than that in the martensite structure. Cracks propagating in martensitic
slow down in the region bordering hydrogen-resistant austenite. However, with the insufficient
stability of residual austenite its negative effect
is a possibility. In this case, as residual austenite
is turned to martensite under deformation at the
crack top, the resistance to SCC reduces (Ref
116).
The use of high-temperature thermo-mechanical processing which raises both the parameters
100
1
2
3
4
5
6
7
KISCC, MPa m1/2
80
60
40
20
1000
1400
B, MPa
1800
Strength level effect of steels Fe-0.3C-2Ni-CrMn-Si (1,2,3), Fe-0.4C-Cr-Ni-Mn (4,5), and Fe0.3C-Cr-Mn-Si (6,7) on value of threshold stress-intensity
factor KISCC in corrosion medium. 1,4,6,8, tests in water;
2,5,7, tests in H2S solution; 3, tests in gaseous H2S. Source:
Ref 93
Fig. 13
KISCC and KIC in the case of high-strength combined-alloy steels (when testing in the NaCl
3.5% solution), holds promise (Ref 117).
Quenching from the intercritical temperature interval can also be profitable. The formation of
the ferritic-martensitic structure as the result of
quenching from the intercritical temperature interval (740 C, or 1364 F) makes it possible to
enhance vastly the resistance of steel Fe-0.5CCr-Mo to the hydrogen SCC in salt water (Ref
118). The two-stage austenitization (high-temperature conventional one) forms a toothed
structure of grain boundaries, which also inhibits
the grain boundary failure by enhancing the resistance to SCC.
The effect of grain size on the resistance of
steels to SCC is then ambiguous. In many works
it was pointed out that the finer the austenitic
grain, the higher resistance to SCC is observed.
For example, on the data given in Ref 119 and
120 the time before failure in the NaCl 3.5%
solution increases by 2 to 3 orders of magnitude
with decreasing the austenite grain in the steel
Fe-0.4C-Cr-Ni-Mo from 172 to 12 mkm in size.
However, there is other evidence. For example,
overheating in the process of the austenitization
of the high-strength steel Fe-0.45C-2Ni-Cr-MoV favors a great increase of KISCC when testing
in distilled water (Ref 121). The high-temperature austenitization is most profitable for cast
steels in so far as they are less prone to the grain
growth as compared to hot-strained steels (Ref
122). Unlike quenching, the overheating in normalizing, as a rule, has a negative effect, especially in the case of low-alloy steels containing
manganese and vanadium (Ref 123).
The problem on the effect of initial structure
is of great importance. It is found that, compared
with hot-strained steels, cast steels have higher
values of the corrosion cracking resistance, although standard mechanical properties of cast
steels are much lower (Ref 122). Studied in Ref
124 is the effect of different methods of melting
(an electric furnace, an arc-vacuum furnace, a
vacuum induction furnace, electroslag remelt)
and their combinations on the corrosion mechanical properties of maraging steel. It has been
shown that combined methods of melting make
it possible to increase considerably (by 25%)
KISCC when testing in the NaCl 3% solution. The
analogous data have been obtained in work (Ref
110). Commercial Fe-0.4C-2Ni-Cr-Mo and Fe0.4C-2Ni-Cr-Mo-Si steels that have been melted
in the air and vacuum were tested in the chloride
medium at K 0.9KI. Vacuum-melted steels
possessed the enhanced SCC resistance, whereas
in air-melted steels, cracks developed practically
at once.
In corrosion-mechanical steel failure one of
the key problems is the problem of nonmetallic
impurities. In this case, both morphological
characteristics of impurities (the quantity, form,
sizes, and distribution) and their chemical activity in the given corrosion medium are of importance. The impurities of sulphides, silicates, and
oxides as well, which play the main role in the
process of the crack initiation, are the most ac-
tive (Ref 111). The local electrochemical corrosion processes developing in the corrosive medium around the impurities, resulting in
inevitable formation of aggressive medium-filled
microcavities. The role of MnS impurities in the
course of corrosion-mechanical failure of the Fe0.2C-Cr-Ni-Mo type reactor steel intended for
the production of high-pressure containers was
studied in (Ref 125). The suggested mechanism
of initiation and growth of cracks includes, as
the first stage, the appearance of a slot between
an impurity and a matrix owing to dissolving
MnS in the electrolyte. Atomic hydrogen is adsorbed at crack lateral surfaces, following which
it diffuses into the metal toward the crack top
and causes local cracking.
In acid media sulfide impurities dissolve and
yield H2S that greatly reduces the resistance to
SCC (Ref 126). Dissolution products of some
impurities enhance the adjacent metal corrosion
that also promotes the cracking process. The enhancement of cracking in dissolving nonmetallic
impurities is supported by the data given in Ref
127 devoted to the study of SCC of the Fe-0.2CMn-Ni-Mo steel in the medium intended for
cooling atomic reactors. As the medium has been
contaminated by dissolution products, the MnS
samples have broken down due to the development of the corrosion cracking process, whereas
in the medium free of the corrosion products, this
process has not occurred. At the same time, as
shown in Ref 128, with cathode deposits present,
the SCC process of pipe steels of Fe-0.1C-2MnSi-V-Nb and Fe-0.1C-2Mn-Si-V, and so on type,
which have been chosen from zones of main
pipeline failure cannot be explained in the context of the suggested model of the sulphide impurity dissolution (Ref 128). On authors data,
the negative effect of sulphides is connected
mainly with their role as concentrators of internal stresses (Ref 129). Nonetheless, in spite of
distinctions between interpretations, it is believed that the enhancement of the corrosion-mechanical steel strength can be attained by reducing the nonmetallic impurity content that is
achieved in part by rational deoxidizing and by
modifying the steels by the addition of rare-earth
and alkaline-earth elements.
Stress Corrosion Cracking of Steels for
Main Gas-Pipe Lines. In many countries,
namely, the United States, Canada, Australia,
and Russia, there are recorded emergency failures of subterranean buried gas-pipe lines because of pipe steels SCC (Ref 130, 131). The
summary data displaying the time of trouble-free
operation of main gas-pipe lines used by three
major Russian companies are given in Fig. 14.
Pipes of different diameter (1020, 1220, 1420
mm) and different chemical composition have
been analyzed. It has been found out that pipes
put into service from 1967 to 1980 have the resistance to SCC much higher than those which
have been laid recently in a period from 1981 to
1989. The time of safe service of the former is
from 9 to 23 years, that of the latter is only from
2 to 14 years. It is significant that all the shortlived pipes have been fabricated from steels of

the Fe-0.1C-2Mn-Si-V-Nb and Fe-0.1C-2MnSi-V type produced in France, Italy, Germany,
and Japan. Also, it has been found that in the
most cases (more than 60% of emergencies) the
failure happens in places with the increased
ground humidity. Cracks are revealed most often
in the lower part of pipes. The mean rate of crack
growth is about 1.15 mm/year (0.05 in./year).
The critical length of cracks that cause failure is
in the range of 140 to 150 mm (5.55.9 in.) (Ref
129).
According to current concepts, main gas-pipe
line steels are susceptible to two types of SCC:
APC and HISCC. The first (classical) type of
SCC is developed at medium acidity of pH
9.5 (high pH SCC); the second one is developed
at pH 6.5 (low pH SCC). In both cases crack
groups (colonies) orientated along the axis of the
pipe arise (Ref 94). More high density of cracks
in the colony is inherent in cracking by the APC
mechanism. Furthermore, it has been found that
in the overwhelming majority of cases, SCC at
a high pH is observed within the limits of regions
in extent of 20 km (12.4 miles) from the compressor stationthat is, where the temperature
and the pressure of gas have their top values (Ref
132134). As to cracking by the HISCC mechanism, the analogous regularity is not fulfilled,
but in this case, the well-defined connection with
the weld location becomes evident. A major part
of failures at low pH begins in centers located at
a distance of 200 to 250 mm (810 in.) from the
weld that is connected with the increased level
of processing residual stresses. The failure at
high pH develops mainly in the intercrystalline
manner (i.e., along the grain boundaries), at low
pH, it develops in the transcrystalline manner
(Ref 94).
Period of safe service, years
30
25
20
15
C
T
Y
T
10
T
5
0
1966
1971
1976
1981
1986
1991
Year of putting into operation

Period of gas main pipes service till first emergency failure depending on year of puttingthem into operation and manufacturing country. Closedcircle data points, steels Fe-0.1C-2Mn-Si-V-Nb and
Fe-0.1C-2Mn-Si-V made in Germany, Italy, France, and Japan; open circle data points, steels Fe-0.17C-Mn-Si, Fe0.15C-2Mn-Si made in Russia; rhomb, steels Fe-0.14C2Mn-Si-N-V, Fe-0.17C-2Mn-Si-N-V made in Russia. Size
of small circle corresponds to 1020 mm pipe diameter;
middle one, to 1220 mm pipe diameter; and large one, to
1420 mm pipe diameter. Y, T, and C denote companies
Ultransgas (Y), Tyumentransgas (T), and Severgasprom (C),
respectively. Source: Ref 129
Fig. 14
The composition of the corrosion medium

forming under the peeled-off insulation is the
main factor on which the realization of the given
SCC mechanism depends. High pH media are
concentrated carbonate-bicarbonate earth solutions with temperatures up to 75 C (135 F).
These solutions are formed under conditions of
the cathode protection that initiates an increase
of pH (acidifying) of the corrosion medium on
the pipe surface. The crack grows as a result of
the rapid electrochemical metal corrosion at its
top where the main processes stimulating the
crack growth, namely, the stress localization, the
plastic strain, emerging of slip strips on the
newly formed surface, are grouped. Low pH
SCC develops in deleted earth electrolyte media
containing dissolved CO2. It is suggested (Ref
130, 133137) that hydrogen arises in microregions filled with the electrolyte (i.e., in slots, pittings, etc.). As a result of corrosion product hydrolysis, the electrolyte in these microregions is
acidified to a level sufficient to discharging ions
of hydrogen, which then diffuses into the region
of tensile stresses ahead of the crack top. As in
the classic model (high-pH SCC), the plastic
strain at the crack top, which benefits the metal
hydrogenation, is required as the obligatory condition (Ref 135, 138). In this case, as seen from
the direct measurements (Ref 94), the zone of
the increased hydrogen concentration extends
for approximately 10 mm (0.4 in.) beneath the
crack surface.
Pipe surface metallurgical defects arising
from rolling may present as nucleous cracks.
These defects, for example, arise from surface
microcracks generating at the stage of continuous steel casting. Cracks initiate on the continuously casted ingots over definite temperature
intervals, mainly over the interval from 700 to
1000 C (12921832 F) (webbed and cross
cracks), and at temperatures close to the solidus
one (spider-shaped cracks). In the given temperature intervals, low-alloy steels are subject to
the strong embrittlement connected with the
change of the grain boundary chemical composition. The present state of the problem concerning high-temperature steel brittleness and the
mechanism of the surface crack initiation in continuous casting has been analyzed thoroughly
(Ref 139, 140).
Practically all the steels applied in the presentday gas-pipe lines, whatever their chemical composition, structure, and strength level, are prone
to SCC to some extent. That is why the prediction of a steels resistance to SCC takes on great
significance. The results of testing a number of
pipe steels for SCC have been correlated with
their mechanical properties determined in the air
(Ref 141). In addition to the standard mechanical
properties, the mechanical properties of samples
under bending with the preliminary-created fatigue crack have been determined. Any correlation between the resistance to SCC and the standard mechanical characteristics has not been
detected. At the same time, according to the
analysis made, a distinct relation is observed between the threshold stress, rth, in the corrosion
medium and characteristics of the mechanical

properties of samples with a crack while performing bending tests in the air. The resistance
to SCC in the studied aggressive media enhances
with an increasing of such characteristics as failure work, yield strength, and ultimate strength.
On this basis, the conclusion has been drawn regarding the necessity of introducing an additional testing of pipe steels in manufacturing
plants, namely, tests of samples with a crack to
predict the resistance of different types of steels
to SCC.
Effect of Alloying and Impurity

Elements on Hydrogen
Embrittlement Resistance of Steel
Hydrogen-resistant engineering steels should
have a number of the following basic characteristics:
High resistance to a brittle fracture for pre-
vention of development of an origin crack
Low content of nonmetallic inclusions and
phases suitable as hydrogen traps
Low level of internal stresses for elimination
of an opportunity of crack nucleation
Presence of surface protecting films prevent-
ing penetration of hydrogen in bulk

To ensure the first requirement, the structure and
alloying of steel should provide:
Hardenability of steel to gain a homogeneous
structure of martensite or its mixture with

lower bainite
Decreasing the ductile-brittle transition temperature, T50, and growth of an energy for a
ductile crack propagation, ac
Low-grain-boundaries
concentration of
harmful impurities
Influence of alloying elements and impurities
on these basic characteristics has been set (Ref
51). The corresponding data, also involving the
optimal content of the basic and impurity elements providing a high hydrogen resistance of
steel, are shown in Table 1. In this case, tendency
of steel to hydrogen embrittlement was estimated from the relative reduction of its plasticity
under tension:
FW
W0 WH
100%
W0
before (W0) and after (WH) hydrogenation of

plain samples during 5 h, and also from reduction of the average time for failure, st, of the ringnotched ones (h 1 mm, r 0.25 mm) under
approximately 0.6 ry loading.
Carbide-forming elements affect the quantity,
distribution, and morphology of carbonitrides.
The carbides are spheroidized in such a manner
that interfaces of the particles with a matrix are
not collectors for hydrogen. These elements in a
solid solution depress the hydrogen thermodynamic activity; as a result, the occluding ability
of steel is essentially diminished (Ref 142).

Those additions in steel have dissimilar influence on the resistance to different kinds of fracture. If its content does not exceed an optimal
value (Table 1), resistance to all kinds of brittleness increases (Ref 143). At higher contents, the
resistance of steel to hydrogen is increased, but
it is reduced to other kinds of brittle fracture.
Carbide-non-forming elements affect steel resistance to a ductile, brittle, and intergranular
fracture. Addition of 0.1 % Si, Mn, Co or Al in
a solid solution of an improved engineering steel
reduces its fracture resistance; the transition temperature, T50 rising by 5; and the impact strength
at decreasing by 0.07MJ/m2. Nickel is an exception; 0.1% of it makes T50 lower by 4 to 10.
Similarly, these elements influence hydrogen resistance of steel. Thus, manganese, aluminum,
silicon, and cobalt decrease the time for failure,
st, and increase the loss of plasticity, Fw, under
hydrogenation. Nickel increases steel resistance
to hydrogen.
Thus, when these elements are added in a solid
solution, the direct dependence between the hydrogen and brittleness resistance characteristics,
T50, ac, is observed.
Impurities such as sulfur, phosphorus, antimony, tin, and other elements have great influence on the resistance of engineering steels to
hydrogen embrittlement. Their action is manifested via weakening of grain-boundary binding
(temper-embrittlement processes) and via nonmetallic inclusions formation. According to existing opinions (Ref 144), loaded grain junctions
can be zones for the origination of three-axial
stress state. These grain junctions are energetically preferable for the accumulation of hydrogen atoms having a low chemical potential
value. It is assumed that hydrogen decreases the
cohesive forces between iron atoms at grain
boundaries (Ref 145). Segregation of impurities
(phosphorus, antimony, tin, etc.) at boundaries
of original austenite grains results in recurring
temper embrittlement, which further weakens intergranular cohesive forces. These considerations have been proved by direct experiments
(Ref 146). The results indicate that hydrogen
does not interact with impurities or segregations
at grain boundaries (i.e., the grain-boundary embrittlement actions of impurities and of hydrogen
mutually supplement each other).
Published data indicate that steel improving
with impurities and gases cleanliness enables it
to boost its fracture resistance (Ref 147). The use
of metallurgical refining and remelting processes
greatly increases hydrogen-embrittlement resistance due to the removal of nonmetallic inclusions from the steel, especially sulfides and
stringer oxide inclusions (Ref 148).
Improving the quality of heat-hardenable engineering steels (unalloyed with molybdenum
and tungsten) enables them to boost by 2 to 3
times their resistance to brittle and ductile fracture.
Sulfur is present in steel mainly in the form of
manganese sulfides. Nonmetallic inclusions of
this type are very ductile and, hence, during rolling are rolled out in the form of elongated
streaks. Sulfide streaks have S:Fe:Mn ratio of

2:1:1, which comply with the formula (FeMn)S;
their thickness is 10 to 15 nm, and they are
mainly located at the boundaries of original austenite grains (Ref 149). The sulfide-matrix interfaces are sites for the accumulation of diffusionmobile hydrogen atoms, which at subsequent
stages cause hydrogen blistering in low-alloy
ferritic-pearlitic steels due to hydrogen molecules formation from its ions and greater (than
the yield strength) planar pressure. Under loading conditions the hydrogen also will accumulate
at the sulfide-matrix boundaries and other sites
where there is a high stress level and will cause
sulfide stress cracking, which is characteristic
both to mild and alloy high-strength steels (Ref
150, 151). Shape modification of sulfide inclusions from plate-like to globular lowers the sensitivity of steels to hydrogen cracking, since the
globules are less favorable to hydrogen accumulation than plates are. Calcium and rare-earth
metals are suitable for shape modification of inclusions (Ref 152).
Thus, the nonmetallic inclusions, especially
sulfides, enhance accumulation of hydrogen in
steels and increase the susceptibility of steels to
hydrogen embrittlement. Therefore, in order to
lower the sensitivity of steels to hydrogen embrittlement, it is necessary to decrease the sulfur
content in them and to modify the shape and type
of sulfide inclusions.
Sulfur, therefore, is a harmful impurity in
steels, lowering their resistance to fracture and
hydrogen embrittlement. In mild steels with ferritic-pearlitic structure, the sulfur, forming manganese sulfides of plate-like shape (stringers),
enhances hydrogen blistering, which is manifested even at 0.001% S. Sulfide (hydrogen)
stress cracking of such steels is sharply reduced
with the decrease of sulfur content below
0.007%. An effective way for prevention of blistering and hydrogen stress cracking is modification of engineering steels with rare-earth elements (0.10.3%), especially with cerium
spheroidizing sulfide inclusions.
Phosphorus, Antimony, and Tin. The adverse
effect of these elements on steel properties is
well studied. Impurities enriching grain boundaries during tempering cause reversible temper
brittleness. This phenomenon has a considerable
effect on the level of steel resistance to fracture.
Phosphorus in engineering steels greatly lowers brittle and ductile fracture. Phosphorus facilitates steel hydrogen embrittlement by enhancing temper brittleness development and
establishing chemical and structural heterogeneity.
Quantitative relationships have been found for
the embrittling effect of phosphorus, tin, and antimony on steel: Each 0.01% of the impurities
increases brittle fracture temperature, T50, by 20
to 23 C; (3641 F), decreases crack-development energy, Ap, by 10 J/cm2; and increases
plasticity loss value during hydrogenation by
8%.
Reference 153 describes a study on the influence of phosphorus content on the resistance of
heat-hardenable steel (0.4% C, 1.0% Cr, 0.37%

Mo, Fe-bal) to hydrogen-sulfide cracking, and it
has been shown that increasing its content from
0.009 to 0.06% lowers the threshold stress from
518 MPa (75 ksi) to 140 MPa (20 ksi). Such an
effect of phosphorus is attributed, first, to the fact
that it activates the hydrogen absorption process
and, second, that it changes the shape and distribution of carbides. Phosphorus or manganese
segregation can be formed at a pearlite band in
low-alloy steels (Ref 154). Such formation
causes structural changes, an increase of steel
occlusive capacity by 2 times, and a decrease of
the time for failure under load in a hydrogenating
medium by 2 times.
The adverse effect of phosphorus on steel resistance to brittle fracture is comparable with the
favorable effect of nickel and molybdenum:
Each 0.01% lowering of phosphorus content in
steel is equivalent, in its influence on the T50
level, to steel alloying with 1% nickel or 0.1%
of molybdenum.
Antimony and Tin. The resistance of engineering steels to brittle fracture depends, to a
large extent, on its cleanliness with regard to antimony and tin impurities. These impurities are
brought into steel in considerable quantities
(0.010.03%) during melting when using scrap.
Both elements increase the cold-shortness
threshold and lower the impact strength level
(Ref 155). The degree of its influence may vary
with the steel alloying system to a great extent.
In low-alloy steels with ferritic-pearlitic structure, the addition of antimony, which is a more
electro-positive noble element than iron, improves the corrosion resistance in hydrochloric
and sulfuric acids, assumed to be due to a film
containing antimony forming on the steel surface. The surface layer inhibits penetration of hydrogen into steel (Ref 156).
To a greater extent, antimony and tin lower
the steel resistance to brittle fracture: Each additional 0.01% Sb or Sn raises the T50 by an average of 18 C (32 F). Their adverse effect in
this respect is comparable with the favorable influence of nickel and molybdenum.
It is possible to assume that at an antimony
content above 0.3%, the effect of lowering steel
resistance to brittle and ductile fracture will prevail over the effect of lowering the quantity of
absorbed hydrogen. Hence, additional alloying
with approximately 0.20 to 0.30% antimony is
advantageous for low-alloy steels to be used in
hydrogen-sulfide media. The associated reduction in the level of toughness properties is insignificant as compared with the improvement in
the resistance to hydrogen embrittlement by 2
times.
With respect to these considerations the following conclusions can be drawn:
Antimony and tin impurities lower the resistance of structural steels to brittle and ductile
fracture: Each 0.01% increase in antimony or
tin raises T50 of heat-hardenable steels by 18
C (32 F), and each 0.01% increase in antimony raises T50 of low-alloy steels by 1 to 5
C (1.89 F).

The embrittling effect of antimony and tin is
due to the enhancement of recurring temper

brittleness process: Addition of 0.0015% Sb
or Sn increases the T50 of heat-hardenable
steels by 100 C (180 F).
Antimony additions (up to 0.30.5%) in lowalloy ferritic-pearlitic steels retard electrochemical reactions on a steel surface, reducing in the amount of absorbed hydrogen.
Consequently, these additions increase (by
more than 2 times) the resistance of steel to
SCC.
REFERENCES
1. H. Epstein, J. Chipman, and N.J. Grant,
Met. Trans., Vol 9 (No. 4), 1957, p 597
2. N.N. Kolgatin, Zavodskaya Laboratoriya,
Vol 9, 1957, p 1098 (in Russian)
3. T.C. Poulter and L. Uffelman, Physics, Vol
3, 1932, p 147
4. W. Geller and T.H. Sun, Arch. Eisenhuttenwes, Vol 21 (No. 11/12), 1950, p 423
5. E. Martin, Arch. Eisenhuttenwes, Vol 3
(No. 6), 19291930, p 403
6. A. Sieverts et al., Z. Phys. Chem., Vol 183
(No. 1), 1938, p 19
7. L. Luckemeyer-Hasse and H. Schenk,
Arch. Eisenhuttenwes, Vol 5 (No. 6),
19321933, p 209
8. M.J. Armbruster, Am. Chem. Soc., Vol 65
(No. 11), 1943, p 1043
9. W. Eichenbauer, H. Kunzig, and A. Pebler,
Z. Metallkd., Vol 49, 1958, p 220225
10. M.Z. Hill and E.W. Johnson, Trans. AIME,
Vol 221, 1961, p 622629
11. Th. Heumann, D. Primas, and Z. Naturforschung, Vol A-212, 1966, p 260269
12. V.I. Saly, R.A. Riabov, and P.V. Geld, Fiz.
Met. Metalloved., Vol 35 (No. 1), 1973, p
119122 (in Russian)
13. H. Schenk and K.W. Lang, Arch. Eisenhuttenwes, Vol 37, 1966, p 739744
14. J.R.G. Da Silva, S.W. Stafford, and R.B.
McLellan, J. Less-Common Met., Vol 49,
1976, p 407420
15. J.R.G. Da Silva and R.B. McLellan, J.
Less-Common Met., Vol 50, 1976, p 15
16. C.A. Zapffe, Metals Handbook, American
Society for Metals, 1948, p 1208
17. J. Hewitt, Hydrogen in Steel, London
Chapman, Ed., 1962
18. T.N. Stross and F.C. Tomkins, Chem.
Abstr., Vol 50, 1956, p 230
19. W. Eichenauer et al., Z. Metallkd., Vol 49
(No. 5), 1958, p 220
20. E.W. Johnson and M.L. Hill, J. Met., Vol
11 (No. 9), 1959, p 572
21. A.E. Schuetz and W.D. Robertson, Corrosion, Vol 13 (No.7), 1957, p 437
22. R.C. Frank, J. Appl. Phys., Vol 29 (No. 6),
1958, p 892
23. M. Smialowski, Neue Hutte, Vol 10, 1957,
p 621
24. J. Hobson, J. Iron Steel Inst., Vol 189 (No.
4), 1958, p 315
25. H.I. Read, Hydrogen Embrittlement in

Met. Finish., Reinhold, N.Y., 1961 p 213
26. F. Kazinczy, Jernkontorets Ann., Vol 1,
1955, p 885
27. C.A. Zapffe, J. Iron Steel Inst., Vol 156,
1947, p 123
28. B.A. Kolachev, Hydrogen Embrittlement
of Metals, Metallurgia, Moscow, 1985, p
217 (in Russian)
29. V.I. Shapovalov, Hydrogen Influence on
the Structure and Properties of Iron-Carbon Alloys, Metallurgia, Moscow, 1982, p
232 (in Russian)
30. E. Lunarska, A. Zielinski, and M. Smialowski, Acta Metall., Vol 25 (No. 1),
1977, p 161171
31. J.P. Hirth and H.H. Johnson, Corrosion,
Vol 32 (No. 1), 1976, p 326
32. U.I. Archakov, Hydrogen Resistance of
Steel, Metallurgia, Moscow, 1979, p 152
(in Russian)
33. G.V. Karpenko and R.I. Kripyakevich, Hydrogen Influence on Structure and Properties of Steels, Metallurgia, Moscow,
1962, p 198 (in Russian)
34. L.S. Moroz and B.B. Chechulin, Hydrogen
Brittleness of Metals, Metallurgia, Moscow, 1967, p 255 (in Russian)
35. Hydrogen in Metals, Proc. 2nd International Congress, 1977 (Paris), Pergamon
Press, Oxford, 1978
36. C.J. McMahon and V. Vitek, Acta Metall.,Vol 27 (No. 4), 1979, p 507513
37. M. Cornet and S. Talbot-Besnard, Metall.
Sci., Vol 7, 1978, p 335
38. J.P. Hirth, Metall. Trans. A, Vol 11 (No.
6), 1980, p 861890
39. G.M. Pressouyre, Metall. Trans. A, Vol 10
(No. 10), 1979, p 15711973
40. A.G. Kumnick and H.H. Johnson, Acta
Metall.,Vol 28 (No. 1), 1980, p 3339
41. C.D. Beachem, Metall. Trans., Vol 3,
1972, p 437
42. A. Troiano, Met. Prog., Vol 71, 1960, p
112
43. R. Oriani and P. Josephic: Acta Metall.,
Vol 22, 1974, p 1065
44. L. Engel and H. Klingel, Rasterelectronenmikroskopishe Untersuchungen von Metallschaden,
Karl
Hanser
Verlag,
Munchen-Wien, 1982
45. J.M. Hodge, M.A. Orehoski, and J.E.
Steiner, Trans. AIME, Vol 230 (No. 5),
1964, p 1182
46. J. Mussati and A. Reggiori, Met. Prog.,
Vol 53 (No. 7), 1936, p 51
47. P. Bardenheuer, Stahl u. Eisen, Vol 45,
1925, p 1782
48. W. Eilender and H.Z. Kiessel, VDI, 1932,
Vol 76, p 729
49. E.A. Klausting, Metallurgist, Vol 5, 1937,
p 21
50. E. Houdremont, Handbuch der Sonderstahlkunde, Verlag Springer, Berlin, 1934
51. V.V. Zabilsky et al., Fiz. Met. Metalloved.,
Vol 83 (No. 2), 1997, p 160175 (in Russian)
52. Y.M. Potak, Brittle Fractures of Steel and

Steel Parts, Oborongiz, Moscow, 1955 (in
Russian)
53. K. Oniki, Kinzoku, Vol 46 (No. 4), 1976,
p 1518
54. M.A. Balter, Fractography as Tool for Diagnosis of Fracted Parts, Maschinostroenie, Moscow, 1987 (in Russian)
55. Y. Nakatani, K. Utsubo, and K. Yamada,
Trans. Jpn. Soc. Mech. Eng. A, Vol 64 (No.
626), 1998, p 25612566
56. M.M. Shved, Change of Operating Properties of Iron and Steel under Hydrogen
Action, Naukova dumka, Kiev, 1985 (in
Russian)
57. A.Ye. Andreykiv, V.V. Panacjuk, and V.S.
Charin, Fiz. Met. Metalloved., (No. 3),
58. E.A. Steigerwald, Trans. AIME, Vol 215,
1959, p 1048
59. L.M. Bilyi and M.M. Shved, Fiz. Met. Metalloved., (No. 6), 1978, p 106107 (in
Russian)
60. I.S. Yablonsky, Fiz. Khim. Mekh. Mater.,
(No. 6), 1979, p 4756 (in Russian)
61. F.F. Azhogin, A.V. Sacharov, and S.S. Ivanov, Fiz. Khim. Mekh. Mater., (No. 3),
62. K. Soeno, Trans. Iron Steel Inst. Jpn., Vol
21 (No. 7), 1981, p 469476
63. K. Soeno and K. Taguchi, Trans. Iron Steel
Inst. Jpn., Vol 21 (No. 9), 1981, p 618
623
64. K. Soeno and M. Tsuhiya, Trans. Iron
Steel Inst. Jpn., Vol 22 (No. 11), 1982, p
848853
65. V.A. Maritchev, Fiz. Khim. Mekh. Mater.,
(No. 5), 1981, p 2429 (in Russian)
66. S. Yamasaki and T. Takahashi, J. Iron
Steel Inst. Jpn., Vol 83 (No. 7), 1997, p
460465
67. O. Tamura and M. Takemoto, J. Iron Steel
Inst. Jpn., Vol 83 (No. 8), 1997, p 532
537
68. C.L. Briant and S.K. Banerji, Ed., Treatise
on Materials Science and Technology,Vol
25, Embrittlement of Engineering Alloys,
Academic Press, 1983
69. G.G. Hancock and H.H. Johnson, Trans.
Met. Soc. AIME, Vol 236, 1966, p 513
70. H. Liebowitz, Ed., Fracture, Vol 3, Mir,
Moscow, 1971 (in Russian)
71. V.S. Savchenko and K.A. Yutshenko, Fiz.
Khim. Mekh. Mater., (No. 1), 1982, p 54
57 (in Russian)
72. V.N. Zikeev, MiTOM, (No. 5), 1982, p 8
23 (in Russian)
73. G.G. Irzhov, S.A. Golovanenko, and T.K.
Sergeeva, Fiz. Khim. Mekh. Mater., (No.
3), 1982, p 8993 (in Russian)
74. S.A. Golovanenko and T.K. Sergeeva,
Steel, (No. 7), 1984, p 7378 (in Russian)
75. L.M. Utevsky, Ye.E. Glikman, and G.S.
Kark, Reversible Temper Brittleness of
Steel and Iron Alloys, Metallurgia, Moscow, 1987 (in Russian)

76. A.I. Kovalev, G.V. Tsherbedinsky, Modern Methods of Investigations of Metal and
Alloy Surface, Metallurgia, Moscow,
1989 (in Russian)
77. V.V. Zabilsky, Metal. Termoobrabotka,
(No. 1), 1987, p 2432 (in Russian)
78. L.I. Gladshtein and N.I. Kelberin, Zavodskaya Laboratoriya, (No. 1), 1995, p 38
41 (in Russian)
79. A.Yu. Kazanskaya, M.A. Smirnov, and
V.V. Zabilsky, Fiz. Met. Metalloved., (No.
8), 1990, p 201203 (in Russian)
80. M.L. Bernshtein and V.A. Zaymovsky,
Mechanical Properties of Metals, Metallurgia, Moscow, 1979 (in Russian)
81. V.V. Zabilsky, O.A. Bartenev, V.V. Velitchko, and S.M. Polonskaja, Fiz. Met.
Metalloved., (No. 4), 1986, p 793800 (in
Russian)
82. V.V. Velitchko, G.M. Micheev, V.V. Zabilsky, and D.I. Maleev, Fiz. Khim. Mekh.
Mater., Vol 27 (No. 1), 1991, p 112114
(in Russian)
83. V.V. Zabilsky and V.V. Velitchko, Metal.
Termoobrabotka, (No. 4), 1993, p 915 (in
Russian)
84. V.V. Velitchko, V.V. Zabilsky, and G.M.
Micheev, Fiz. Met. Metalloved., (No. 1),
1995, p 130140 (in Russian)
85. I.S. Toydorova and V.V. Zabilsky, Fiz.
Met. Metalloved., (No. 7), 1991, p 511
(in Russian)
86. V.V. Zabilsky, V.V. Velitchko, and S.G.
Iliyina, Fiz. Met. Metalloved., Vol 80 (No.
12), 1995, p 108118 (in Russian)
87. V.V. Zabilsky, V.V. Velitchko, F.Z. Gilmutdinov, V.G. Makletzov, and Yu.P. Vorobiov, Fiz. Met. Metalloved., Vol 83 (No.
2), 1997, p 160175 (in Russian)
88. A.V. Vaynman, R.K. Melechov, and O.D.
Smiyan, Hydrogen Embrittlement of Elements of High Pressure Kettles, Naukova
dumka, Kiev, 1990 (in Russian)
89. I.K. Pohodnya, Electron and Ion Spectroscopy of Solids (No. 2), L. Fiermans, J.
Vennik, and W. Dekeyser, Plenum Press,
90. V.V. Zabilsky and V.V. Velitchko, Fiz.
Khim. Mekh. Mater., Vol 25 (No. 3), 1989,
p 1721 (in Russian)
91. V.V. Zabilsky, V.V. Velitchko, and A.Yu.
Kazanskaja, Fiz. Met. Metalloved., (No. 3),
1990, p 143149 (in Russian)
92. V.I. Voronenko, Zashch. Met., Vol 33 (No.
2), 1997, p 132143 (in Russian)
93. O.N. Romaniv and G.N. Nikifortchin, Mechanics of Corrosion Failure of Structural
Alloys, Naukova dumka, Kiev, 1986
94. N.I. Volgina and T.K. Sergeeva, Materialovedenije, (No. 12), 1998, p 2130 (in
Russian)
95. L.N. Petrov, V.N. Tischenko, and A.S.
Tchistov, Fiz. Khim. Mekh. Mater., (No.
4), 1979, p 2025 (in Russian)
96. V.I. Nikitin, Fiz. Khim. Mekh. Mater., Vol
25 (No. 1), 1989, p 31 (in Russian)
97. F.R. Ford, Mech. Behav. Mater. Proc. 3rd
Int. Conf., 1979 (Cambridge U.K.), Pergamon Press, Toronto, 1979, (No. 2), p 431
98. O.N. Romaniv, G.N. Nikifortchin, and

A.Z. Student, Fiz. Khim. Mekh. Mater.,
99. V.P. Voronin, A.Yu. Schulte, and Yu.V.
Tomkin, Fiz. Khim. Mekh. Mater., Vol 19
(No. 5), 1983, p 114 (in Russian)
100. A. Mazur and B. Pawlowski, Corros. Sci.,
Vol 26 (No. 1), 1986, p 7
101. T.J. Kaneko, J. Iron Steel Inst. Jpn., Vol
72 (No. 13), 1986 p 1472
102. J. Kuniya, I. Masaoka, and R. Sasaki, J.
Pressure Vessel Technol., Vol 107 (No. 4),
1985, p 430
103. W. Diton and P. Huiying, Proc. Int. Congr.
Met. Corros., 1984 (Toronto), (No. 3),
1984, p 573577
104. L.N. Petrovm, Fiz. Khim. Mekh. Mater.,
(No. 4), 1984, p 1621
105. W.W. Gerberich, R.N. Jones, M.A. Friesel,
and A. Nozue, Mater. Sci. and Eng. A, Vol
103 (No. 1), 1988, p 185
106. A. Nozue, T. Kishi, and R. Horiuchi, J.
Iron Steel Inst. Jpn., Vol 67 (No. 10),
1981, p 1787
107. G.B. Muravin, L.M. Lezvinscaja, N.O.
Makarova, and G.S. Pavlovskaja, Defektoskopiya, (No. 2), 1990, p 1828
108. B.I. Voronenko, Zashch. Met., Vol 33 (No.
5), 1997, p 472488 (in Russian)
109. B.I. Voronenko, Metal. Termoobrabotka,
(No. 9), 1990, p 58 (in Russian)
110. C.S. Carter, Corrosion, Vol 25 (No. 10),
1969, p 423
111. B.I. Voronenko, Zashch. Met., Vol 33 (No.
6), 1997, p 573589 (in Russian)
112. P. McIntyre and T.M. Banks, Eur. Offshore Steels Res. Seminar, VIP 19/1, 1978
(Cambridge), Abington Publishing, 1980
113. D. Huerta and K.E. Heusler, Werkst. Korros., Vol 45 (No. 9), 1994, p 489
114. O.N. Romaniv, G.N. Nikifortchin, A.Z.
Student, Fiz. Khim. Mekh. Mater., (No. 2),
115. B. Pawlowski, Werkst. Korros., Vol 37
(No. 8), 1986, p 448
116. R.O. Ritchie, M.H.C. Cenedo, and V.F.
Zackay, Metall. Trans. A, Vol 9 (No. 1),
1978, p 3540
117. M. Turdy, V. Suchanek, L. Hyspecka, and
K. Mazanec, Adv. Res. Strength and Fract.
Mater./4th Int. Conf. Fract. 1977 (Waterloo, NY), 1978, p 255
118. D. Zhou, W. Cao, and H. Guo, Heat Treat.
Met., (No. 1), 1992, p 8
119. Y. Hirose, K. Tanaka, and K. Okabayashi,
22nd Congr. Mater. Res., 1978 (Kyoto, Japan), Proc. Kyoto, 1979, p 76
120. Y. Hirose and K. Tanaka, Met. Corros./8th
Int. Congr./7th Congr. Eur. Fed. Corros.
(111th event), 1981 (Mainz, Germany),
Frankfurt/M., European Federation of
Corrosion, 1981, (No. 1), p 553
121. O.N. Romaniv, G.N. Nikifortchin, and
A.Z. Student, Fiz. Khim. Mekh. Mater.,
122. W. Cao and X. Lu, Mater. Sci. and Eng.
A, Vol 110 (No. 151), 1989, p 165
123. W. Radeker and B.N. Mishra, Werkst. Korros., Vol 21 (No. 9), 1970, p 692
124. F.W. Fraser and E.A. Metzbower, Fractogr. and Mater. Sci. Symp., 1979 (Williamsburg, VA), Philadelphia, 1981, p 51
125. K. Rippstein and H. Kaesche, Corros. Sci.,
Vol 29 (No. 5), 1989, p 517
126. Z. Szklarska-Smialowska and E. Lunarska,
Werkst. Korros., Vol 32 (No. 11), 1981, p
478
127. H. Hanninen, Corrosion, Vol 41 (No. 7),
1990, p 563
128. A.G. Gareev, I.G. Abdullin, and G.I. Abdullina, Gazovaja Promischlennost, (No.
11), 1993, p 29 (in Russian)
129. A.F. Matvienko, A.V. Baldin, A.P. Grigorjev, V.A. Kanaikin, V.G. Rjibalko,
V.V. Sagaradze, and Yu.I. Philippov, Fiz.
p 139146 (in Russian)
130. R.N. Parkins, Corrosion, (No. 50), 1994,
p 384
131. T.K. Sergeeva, E.P. Turkovskaja, N.P. Michailov, and A.I. Tchistjakov, State of
Problem of Stress-Corrosion in SNG
Countries and Abroad, IRZ RAO, Gazprom, Moscow, 1998 (in Russian)
132. R.N. Parkins, C.S. Dell, and R.R. Fessler,
Corros. Sci., Vol 24 (No. 4), 1984, p 343
374
133. Public Inquiry Concerning Stress Corrosion Cracking on Canadian Oil and Gas
Pipelines, Report NEB, 1996, No. 11, MH295
134. M. Urednicek, S. Lambert, and O. Vosikovsky, Monitoring and Control Proc. Int.
Conf. on Pipeline Reliability (Calgary,
Canada), 1992, N VII-2
135. T.K. Sergeeva, Proc. Int. Symp. on Pipe
Line Safety and Aging, CNIICHERMET,
1995 (Moscow), Vol 1, p 139164
136. O.M. Ivantzov, Stroitelstvo Truboprovodov, No. 6, 1993, p 1016 (in Russian)
137. B.G. Mazel, Stroitelstvo Truboprovodov,
No. 7, 1993, p 3639 (in Russian)
138. D. Rudy, Prev. of Pipeline Corrosion
Conf., 1994 (Houston, TX), p 1721
139. V.V. Zabilsky and V.I. Lebedev, Tchernaja Metallurgija. Bjulleten NTI, No. 2,
140. R.M. Nikonova and V.V. Zabilsky, Fiz.
Metal. Metalloved., Vol 76 (No. 8), 1993,
141. A.F. Matvienko, A.V. Baldin, A.P. Grigorjev, V.A. Kanaikin, V.G. Rjibalko,
V.V. Sagaradze, and Yu.I. Philippov, Fiz.
142. G. Libowitz, The Solid-State Chemistry of
Binary Metal Hydrides, Vol 6, 1965, p 11
143. V.N. Zikeev, Metal. Termoobrabotka, Vol
5, 1982, p 1823 (in Russian)
144. V.I. Shapovalov, Hydrogen Influence on
the Structure and Properties of Iron-Carbon Alloys, Metallurgia, Moscow, 1982, p
232 (in Russian)
145. R. Oriani, Acta Metall., Vol 24, 1977, p
979988

146. M. Vornet, Mem. Sci. Rev. Met., Vol 74
(No. 5), 1977, p 307316
147. A.P. Guliaev, Clean Steel, Metallurgia,
Moscow, 1975, p 184 (in Russian)
148. V.G. Starchak, A.B. Kuslitsky, U.A.
Kliachko, and I.A. Tamarina, Metal. Termoobrabotka, No. 4, 1977, p 1113 (in
Russian)
149. A. Joshi, Corrosion, Vol 34 (No. 2), 1978,
p 4752
150. J. Hirth, Corrosion, Vol 32 (No. 1), 1976,

p 3437
151. Z. Szklarska-Smilowaska, Werkst. Korros., Vol 32 (No. 11), 1982, p 478485
152. G.M. Itskovich, Deoxidation of Steel and
Modification of Nonmetallic Inclusions,
Metallurgia, Moscow, 1981, p 296 (in
Russian)
153. E. Snape, Corrosion, Vol 24 (No. 9), 1968,
p 261282
154. H. Gondo and M. Iino, Nippon Steel Corporation Studies on Materials for Steel
Pipes for Service in Wet Sour Gas Medium, Technical Report B2 (No. 11), Nippon Steel Corporation, 1979, p 71
155. E. Stephenson, Metall. Trans. A, Vol 11
(No. 3), 1980, p 517524
156. S. Shichi, M. Takeshi, and D. Utaka, Japanese Application kl 12A41 (C23Fg/02),
No. 54, 1979

DOI: 10.1361/hrsd2002p089

Deflection Methods to Estimate

Residual Stress
H.W. Walton, Consultant
NOT EVERYONE has sophisticated instrumentation such as x-ray or neutron diffraction

readily available for the determination of residual stresses. Nor do they have the time or inclination to use these procedures for checking production parts. Mathematical derivations such as
the one for plate by the Treuting and Read
method are also very time consuming (the
method is described at the end of this article).
To determine if residual stresses from prior
processing are the cause of uncontrollable distortions during processing of thin section parts,
it is a simple matter to slit an as-received piece
and note the amount of deflection resulting from
relaxation of residual stresses, for example, the
net opening of the saw cut. If no deflection occurs, then it can be safely assumed that the problem lies with the processing parameters and not
with the material (e.g., high feeds and speeds or
high chucking pressures during machining, or
surface overheating during grinding resulting
from insufficient wheel dressing).
These simple techniques, sometimes known
as dissection, are old but still very useful. However, because slitting a component is a destructive method, there are limitations, particularly if
the component in question is large.
This article provides a low-cost, easy method
of determining if residual stresses are the cause
of component distortion during manufacture.
Residual Stress
To understand the type of stress that is being
measured, a distinction first needs to be made
between microstresses and macrostresses extending over large volumes. Microstresses resulting from changes occurring at the atomic
level are equilibrated over very small volumes
and cannot be measured by use of deflection
methods. Macrostresses, on the other hand, are
primarily the result of forming operations or
thermal imbalances during casting, welding, heat
treating, and so on.
If the stress induced by nonuniform cooling
exceeds the yield strength of the material, plastic
deformation occurs. On cooling to a uniform am-
bient temperature, the deformed volume being

restrained by the surrounding material is now
placed under significant elastic stress. This
would be considered macrostress and is the
stress most likely to cause deformation during
subsequent processing.
Consider the simple example of bending a bar
to form a permanent set. The surface material at
the elbow or concave surface undergoes permanent plastic deformation while the inside diameter is undergoing compressive plastic deformation. When the applied force is removed, the
underlying elastically stretched material (at the
elbow) is now prevented from relaxing completely. The result is that the plastically deformed surface layers of the elbow are placed
under elastic compression by the underlying
elastically stretched material. The opposite is
true for the inside diameter. It follows that if the
bar is placed in service in this condition, corrosion and development of cracks are more likely
to occur at the surface of the inside diameter.
In general terms, if a surface undergoes permanent plastic deformation by cold rolling, shot
peening, machining, thermal processing, and so
on, which in turn results in the underlying material being elastically stretched or compressed,
the material plastically deformed by tensile
forces is left in elastic compression. Material
plastically deformed by compressive forces is in
elastic tension.
The influence of nonuniform residual stress is
sometimes dramatically illustrated by fishtailing of large hot-rolled steel plate. As the plate
cools after rolling, transformational stresses are
developed and, if an extensive central lamination
is present due to inadequate discard or casting
problems, these stresses can cause the plate to
separate into two halves (Fig. 1).
The stress distributions present before and after the occurrence of a fishtail are shown in the
sketch. In the as-rolled plate, tensile stresses
(tv stress) on the surface are trying to contract
the surface while compressive stresses (tv
stress) are trying to expand the plate core. How
this stress distribution develops is illustrated in
Fig. 2. After fishtailing, the two halves of the
plate attain a new equilibrium stress distribution.
While air cooling after hot rolling, the plate

surface will be cooling faster than the core material, and thermal stresses developed at this
stage will be relaxed by the plastic deformation
of the low strength austenite. At lower temperatures, transformation to ferrite and pearlite begins at the surface (Fig. 3), causing complex interactions and a variety of stress conditions.
Because of the relatively low yield strength of
the various phases and the role of transformation
induced plasticity, the final stresses are low.
However, over a large surface area even low residual stresses can have a profound effect, particularly in the presence of large defects such as
a lamination.
If not fixtured in a vertical position, severe
distortion of quenched flat steel sections may occur. This is due to differences in the cooling rate
and transformation behavior between the top and
bottom surface.
In a carburized product, because of the compositional changes causing transformation to
progress from the inside out and the considerably
higher-strength case, the final residual stress profile is reversed and a beneficial high-compressive
surface stress is developed (Ref 1). Figure 4
shows the development of these residual stresses
during the heat treatment of a carburized component.
It is very important to avoid heating small,
isolated areas on the surface of hardened steel.
Fig. 1
stress
Schematic representation of fishtailing in plate

steels. tv, tensile stress; tv, compressive
90 / Measurement and Prediction of Residual Stress and Distortion

Examples of where this may occur are:
Components ground with inadequate cooling
fluid or faulty equipment (vibrating wheel

heads)
Using a gas flame to enlarge a hardened component for assembly on a shaft
Electrical arcing from inadequately earthed
drive motors
Accidental arcing during magnetization for
magnetic particle inspection
The thermally damaged area is restrained from
expanding by the surrounding material and becomes plastically deformed under compression.
Consequently, on cooling to ambient temperature, the surface is left in a tensile condition, and
cracking ensues either immediately or after the
component is placed in service (more worrying).
Heating at temperature above the original tempering temperature may also lead to a contraction.
Shot peening appears to be an anomaly. At
first sight, the impact of the shot particles seems
to cause plastic deformation in compression. It
should follow that the surface is in tension following the peening treatment. However, the ad-
ditive influence of the multiple small areas of

plastic extension of the surface must be considered. The overall surface undergoes a plastic extension, thus putting the underlying layers in tension. Consequently, on completion of the
shot-peening operation, the surface is restrained
in a beneficial compressive state.
The maximum level of residual stress that can
be generated is approximately equal to the elastic
limit or yield strength of the material measured
under the same stress condition and for the same
cross section. The existence of triaxial stresses
in heavy section components has a major influence on the ease of plastic flow (plane-strain conditions). If surrounding material inhibits plastic
flow by significantly limiting shear stresses, the
levels of residual tensile stress may become
high. Catastrophic fracture occurs when the tensile stress level exceeds the cohesive strength of
the steel.
Manufacturing Implications of
Residual Stresses
There are many examples that show how relaxation or redistribution of residual stresses
contributes to high processing costs. One example is the need to use multiple grinding passes
to avoid distortion in thin section components.
A frequent problem often wrongly attributed to
the heat treatment department is heat treatment
distortion from prior residual stresses in the raw
material.
Premature failure due to harmful residual surface tensile stress in the manufactured product is
to be avoided at all costs.
On the other hand, the presence of residual
stresses may be advantageous, as in the example
of compressive stresses from case hardening and
shot peening.
Using residual stresses to intentionally deform
a component can be illustrated by the use of shot
peening to shape the aircraft fuselage, by using
a peen hammer to straighten a long shaft, or by
simply hot spotting to bend a simple bar of steel.
For example, a small area on one side of a
steel bar is heated rapidly to a temperature, T, of
approximately 600 C (1100 F) (Ref 2). If free
to expand, the heated disc-shaped region increases in length, l, by:
l T
14 600
Cooler outside
surface
8.4 103 lm/mm
Hotter core
Transformation
Consider a slice through half the thickness of a plate.
At the onset of the cooling, both the surface and
the core are austenitic and stress free.
Transformed
Not transformed
The surface will transform first, acccompanied by
an expansion. Restraint by the underlying material
will place the transformed element in compression "C".
Although the adjoining austenitic element "stretches"
and plastically yields, it will be in tension "T".
CT
where is the coefficient of thermal expansion

in lm/mm K. However, the surrounding cooler
material restrains expansion. Because of the relatively low yield strength at 600 C (YS600) of
approximately 30 MPa (4.5 ksi), the heated steel
deforms plastically under compression, increasing the thickness of the heated disc. On cooling,
the disc contracts the same amount and, assuming no stress relaxation occurred, a tensile stress,
r, is now developed in the disc:
r (E l) YS600
(200 103 8.4 103) 30
1650 MPa (240 ksi)
Schematic representation of transformational

expansion in the absence of any restraining effect
by the surrounding material
The already extended underlying element then

undergoes transformation and expands further. This
induces tensile stresses in the material on either side,
but due to the restraining influence of surrounding
material, compressive stresses are induced in the
newly transformed steel.
TCT
Fig. 2
TCT
The transformation front proceeds deeper into the

section, leaving behind a surface that is in tensile.
The magnitude of stress is determined by the relative
yield strength of the phases. Because of the TRIP
effect (transformation induced plasticity), the plasticity
of the transforming steel is considerably higher.
The development of residual stresses during cooling of steels
where E is the modulus of elasticity (MPa). The

stress in the disc cannot be greater than the yield
strength of about 500 MPa (73 ksi). Therefore,
either the disc stretches plastically, or the bar
bends elastically to balance the stresses (or
both).
The manufacture of seamless high-carbonbearing steel tube involves hot rolling, sizing,
straightening, and spheroidize annealing. Each
stage of manufacture introduces various degrees
of residual stress. If residual stress levels are
higher than normal, difficulties may be experienced during subsequent machining and heat
treatment of the rings. As seen from the previous
calculation, the higher the yield strength of the
material, the greater the potential for high levels
of residual stress. As thermal and mechanical
treatment of steel increases the strength of a material, potential problems with residual stress
also increase. For example, a small area of thermal damage on the surface of a hardened high-
Deflection Methods to Estimate Residual Stress / 91

carbon steel (e.g., electric arc discharge) may result in a level of triaxial residual stress close to
the cohesive strength of the material, possibly
leading to immediate or delayed cracking.
Although the stress may be multidirectional,
hoop stress that arises at several stages of manufacture is the primary cause of many of the
problems. When a length of tube is parted off
and slit in a longitudinal direction, any hoop
stress present tends to open the slit (Fig. 5).
Opening of the slit indicates compressive stress
in the inside diameter (ID) of the tube and tensile
stress on the outside diameter (OD). On rare occasions, the slit may close, indicating the reverse
condition. The stress is not always uniform along
the length of the tube. Stress measurements have,
on occasions, indicated a cyclic variation corresponding to differences in cooling of areas in
contact with the cooling bed cross ties.
Shot peening intensity is monitored by using
a method developed by J.O. Almen of General
Motors Company. In this method, 75 mm long
by 18.75 mm wide (3.00 by 0.75 in.) strips of
1070 spring steel are quenched and tempered to
a deep-blue oxide finish (blue tempered) for a
hardness of 45 to 50 Rc and exposed on one side
to the same shot intensity as the component undergoing treatment (Ref 3). Three different
thicknesses are used: N, 0.79 mm (0.031 in.);
A, 1.30 mm (0.051in.); and C, 2.38 mm
(0.0938 in.) to allow for differences in degree
and shot peening required and materials being
treated. For example, peen forming of an aluminum alloy requires considerably less peen intensity than a carburized gear tooth. During treatment, the Almen strip assumes a concave shape.
The lift height is proportional to the level of
compressive stress developed in the upper sur-
face during peening. Peening is continued until

the material has reached saturation; that is, when
the lift height increases by no significant amount
and the compressive stress in the surface layers
corresponds closely to the elastic limit of the
steel strip (Fig. 6).
Consider the residual stress distribution
through the Almen strip. Deflection of the strip
occurs in order to relieve some of the high surface compressive stress. The peened surface area
endeavors to expand by bowing in order to nullify the compressive stress. The magnitude of
bowing is limited by the restraint of the remainder of the material, and equilibrium is reached
when the remaining compressive stress is in balance with the elastic compressive stress developed in the lower surface. The depth and intensity of the compressive layer is proportional to
the lift height at saturation. The Almen strips are
primarily used to ensure that the process is in
control (shot condition, uniformity in application, impeller operation, etc.). A photograph of
three typical strips is shown in Fig. 7.
Metal Improvement Company, Inc., in collaboration with ENSAM, a French advanced engineering school, has developed a software program called PeenStress (Metal Improvement
Co., Inc., Paramus, NJ) that is used to assist in
shot peening callouts (Ref 4). The user selects a
material from a library of about 80 materials,
then selects a shot size and shot intensity and
inputs some basic geometry considerations. Figure 8 is a curve generated on a chromium-silicon
spring wire shot peened with a hardened, 0.023
in. shot to a 10 A intensity. A 10 A intensity
equates to a 0.25 mm (0.010 in.) arc height on
the A-strip.
One of the older standard tests for evaluating
Ar3
Ar1
Austenite
quench distortion is the Navy C-ring test (Ref 5

7). The amount of distortion (deflection) of a
quenched test piece is measured by the change
in the gap width (Fig. 9).
Methods for Measuring Residual

Stresses from Deflection Data
Sectioning to allow relaxation of residual
stress in actual components where stresses are
thought to be present can be performed by several methods, from simple saw cutting to the
more sophisticated compliance method (Ref 8
10). In the latter method, the residual stress profile is calculated from the strains caused by introducing a cut of progressively increasing depth
into the component. The strains are measured by
using suitably positioned strain gages cemented
to the surface adjacent to the cut. The cutting is
performed by using various techniques, however, electrical discharge machining is the preferred method. Usually, two computer-based approaches are used to analyze the data. These are
the forward and inverse solution. The forward
solution derives the measurable strains (compliance functions) that develop from introducing a
successively deeper slot into a part containing an
arbitrary known stress distribution. The inverse
solution develops the original residual stress distribution that best matches the actually measured
strains.
Similar methods based on drilling small holes
in the stressed material have been around for
some years. Strain relaxation is measured using
strain gages or photoelastic coatings. Strain
gages have been used to measure strain relaxation in bevel gears following successive removal
of layers by electrochemical machining (Ref 11).
Saw cutting or slitting is an easy quality control test that gives a global overview of the state
of residual bulk hoop stress in rings. Interestingly, such a technique was used recently to validate the measurements of residual stress levels
in railway wheels using electromagneticacoustic transducers (Ref 12).
Ferrite
Pearlite
Temperature
Core
Bainite
Surface
Martensite
Time, T
Schematic representation of the relative transformation at the surface and in the core of the mild steel plate. At
time, T, the core is still austenitic while the surface has already transformed to a ferrite-pearlite structure. Ar3
and Ar1, upper and lower transformation temperatures, respectively, on heating (refroidissant) a hypoeutectoid steel
Fig. 3
Mathematical Derivations for

Simple Cases Based on the Saw-Cut
Methods
Definitions of the symbols used in the following derivations are given in Table 1.
To develop the mathematical algorithms for
interpreting deflection measurements after slitting a simple shape (plate, round bar, or tube),
simple beam theory is used (Fig. 10).
The basic formula for the state of affairs at
any point (x) along a beam is (Ref 13):
M
r
E

I
c
R
(Eq 1)
where M is the bending moment to which the

beam is subjected at x. Bending moment M is
%C
Carbon profile through case. As the carbon level

decreases, the transformation temperature during
quenching increases (Ms). Although the cooling rate
is less below the surface, transformation occurs from
the inside out.
Consider a section through the carburized case. At

the onset of quench, both the surface and the core
are austenitic and stress free.
Start of transformation
Even though the cooling rate is lower than at the
surface, the low carbon core will begin to transform,
starting first below the case and progressing inward.
The adjoining austenitic case element will plastically
expand outward to accommodate the resulting tensile
stress.
equal to load multiplied by distance (lbf in., or

kgf m). If M is not obvious from the first principles for a given case, it may be obtained from
the data given subsequently. I is the moment of
inertia of the section of the beam at x, usually in
inches. Moments of inertia are treated in textbooks of elementary mechanics. The moment of
inertia in question is that about the neutral axis
which, in most beam problems, will be a line
through the center of gravity of the section. If
the I of the section is not already known, it is
usually easily calculated by applying one of the
following formulae.
For a rectangle:
bt3
12
Schematic representation of transformational

expansion in the absence of any restraining effect
by the surrounding material
C CC T T
+tv
Following the progressive transformation of the core,

the innermost element of the case transforms. The
resulting expansion of the already extended case
element creates tensile stresses in both remaining
austenitic case and adjoining core.
Sections made of rectangles, such as rectangular

tubes, H-beams, and channels, can be calculated
by subtracting the I of the empty areas.
For a circular cross section:
The continued transformation of the progressively

higher carbon case and the resulting expansion of the
already plastically stretched elements result in high
compressive stresses. Consequently, the underlying
transformed core is pulled into tension.
Stress profile through case. Compressive stress

near the surface of the case can reach levels of
300/500 N/mm2 and is one of the primary reasons for
carburizing.
tv
Fig. 4
(Eq 2)
The development of compressive stress in the case of a quenched carburized steel. Source: Ref 1
pr 4
4
(Eq 3)
The moment of inertia for hollow tubes can

be calculated by subtraction, where: r is stress
in the material. c is distance from the neutral
axis. In symmetrical sections this is the distance
from the midpoint or center of gravity of the section. E is Youngs modulus. (In thick cross sections and/or high-strength steels, this should be
corrected for Poisson contraction where m is
Poissons ratio [0.3 for steel].)
E
E
1 m2
(Eq 4)
R is the radius of curvature of the beam when it

bends under load.
To calculate the stress in a simple beam:
r
Mc
I
(Eq 5)
For a rectangular beam at the surface where the

stress is greatest:
Fig. 5
Schematic of the residual stress distribution in rings manufactured from tube before and after slitting. ID, inside
diam; OD, outside diam
bt3
12
(Eq 6)
t
2
(Eq 7)
so
rmax 6
M
bt2
(Eq 8)
EI
R
Symbols used in the derivations and formulas

Units
This explains why a beam that is twice as thick

is four times as strong.
The distribution of residual stress in actual
components is unlikely to be linear. However,
for the subsequent approximate analysis, the
beam stress is assumed to vary linearly through
the section.
In sheet and bar, as the sheet fishtails or is cut
in a central planar direction, the bending moment
created by the residual stress is released (Fig.
11). The bending moment may be expressed as:
M
Table 1
(Eq 9)
Symbol
Description
M
I
Bending moment
Moment of inertia
r
c
Stress in material
Distance from neutral axis
E
R
L
r
t
d
c
Youngs modulus
Radius of curvature of beam or displaced section
Length of curved beam
Radius of round bar cross section
Thickness of beam/plate/tube wall
Measured deflection
Poissons ratio
SI
English
Notes
NM
m4
lbf in.
in.4
Pa
m
psi
in.
Pa
m
m
m
m
m
Dimensionless
psi
in.
in.
in.
in.
in.
...
...
I bt3/12 for rectangle (where b and
t are the dimensions of the
cross section) I pr 4/4
for circle
...
In a symmetrical section this
is the distance from the midpoint or center of gravity
...
...
...
...
...
...
0.3 for steel
given that:
c
t
4
(Eq 10)
30 0.015 in.
where t is plate thickness, and assuming the distribution of the residual stress that resulted in the
fishtail varies linearly over the half-thickness of
the sheet, the maximum longitudinal stress at the
surface is given by:
r
Mc
I
(Eq 11)
0.0310.001 in.
Measuring dial
0.0510.001 in.
N strip
Peening nozzle
A strip
C strip
Almen strips
0.09380.001 in.
0.7450.750 in.
Shot stream
4-6 in.
Almen test strips
Substituting Eq 9 and 10 in Eq 11 gives:

r
Et
4R
Hardened
ball supports
10/32
screws
(Eq 12)
Arc height
3.0 in.
If the deflection, d, is small compared to the radius of curvature, R, R may be expressed in terms
of the deflection, d, and the length of the curved
surface, L, by:
R
L2
2d
3.0 in.
Holding
fixture
0.75 in.
Strip removed. Residual
stresses induced arching
Fig. 6
Use of Almen strips to monitor the degree of shot peening. Source: Ref 3
Fig. 7
Almen strips used to monitor different levels of shot peening intensity
(Eq 13)
This is derived as in Fig. 12.

Referring to Fig. 12, the two radii O1 and O2
are drawn as shown, and the tangents 1,3 and 2,4
are drawn to these radii. A chord is drawn between the points of tangency 1 and 2 and the line
2,3 is drawn perpendicular to the chord at 2. The
two isosceles triangles RLk and mdn are similar
and therefore:
R
m
L
d
(Eq 14)
or
R
Lm
d
(Eq 15)
For small angles, L is the arc 1,2 and m is 12 the

arc 1, 2, or
Arc height
Strip mounted for
height measurement
1
L
2
(Eq 16)
therefore (as in Eq 13):

R
L2
2d
(Eq 17)
where d is the deflection of the strip (Fig. 11a).

According to Eq 12:
Et
r
4R
(Eq 18)
it follows:
r
A simpler technique to measure the circumferential hoop stress in a thin-walled tube is to

slit the tube longitudinally with a carborundum
cutoff wheel (using plenty of coolant to avoid
heating of the surfaces) and measuring the
change in diameter of the tube (Fig. 13b):
Etd
2L2
r Et
1
0
1
D1
Using a carborundum wheel, a ring, either
parted from the tube or machined, is cut axially through the section.
If unknown, the saw curf width (width of saw
(Eq 22)
0.600
0
where D0 is initial diameter and D1 is diameter

after slitting.
Alternatively, the net opening displacement at
the saw cut is measured by subtracting the width
of the saw blade, and the residual hoop stress is
measured using the formula:
(Eq 19)
A similar derivation can be obtained for round

bar (Fig. 11b):
following method is often used to identify the

possible cause:
1
1
r Et
x
D0
D0
p
1.000
2.900
B
1.900
(Eq 23)
E
5.000
(a)
1.65 Erd
L2
(Eq 20)
For thin-walled tubing (Fig. 13), the longitudinal

stress is given by:
r
Etd
L2
(Eq 21)
Equation 21 is the same as Eq 19 except that

t/2 is used in the former case because only half
the sheet thickness deflects.
The method of sectioning a tube to obtain the
previously mentioned approximate measurement
is shown in Fig. 13(a).
where x is the net opening displacement.

A summary of the derivations is given in Table 2.
A Practical Example. The formulas in Table
2 are used extensively for measuring residual
hoop stresses in seamless tube and machined
bearing rings. These stresses may arise from tube
straightening without a stress-relief anneal or
nonuniform rapid cooling following the spheroidized anneal. In the past, contact with the support bars on cooling beds has been recognized
as being a potential cause of cyclic nonuniform
stress along the length of a tube.
Where 52100 bearing steel rings are being machined on multispindle machines and the operator is having difficulty in maintaining size, the
0.25 in.
0.50 in.
2.50 in.
0.5 in.
1.45 in.
(b)
2 in.
1/4 in.
1 1/4 in.
Residual stress , ksi
3/4 in.
2000:5:9
Peen stress
ENSAMMIC
Residual stress distribution

8
Depth, mils
5/8 in.
d = 5.198 mils
h = 0.294 mils
10 A
S 230 shot
V = 124 ft/s
HS/4320 CB/SP
100
(c)
Examples of C-ring test specimens used for
quench distortion studies. (a) Source: Ref 5. (b)
Source: Ref 6. (c) Source: Ref 7
Fig. 9
Dir. Z
x
s (ksi)
102
m (ksi)
203
Pm (in.)
0.001
P0 (in.)
0.008
B
d
A
l
200
W
M = Wx
Max. M = Wl at B
Fig. 8
Calculated residual stress distribution in a shot peened chromium-silicon spring wire. d, depth; h, height. Source:
Peen Stress, Metal Improvement Company, Inc.
Fig. 10
Simple beam theorydefinition of bending

moment. W, weight in lbs or kg

cut) is determined by measuring a shallow cut
adjacent to the final cut.
The gap width is measured using internal calipers. The net opening (x) equals the gap
width minus the saw curf (width).
On occasions the gap may decrease, indicating a
compressive hoop stress. In extreme situations,
the blade may become nipped and may even disintegrate. Ensure proper protection when using
this method.
For greater accuracy in large cross sections,
the value for E can be corrected for Poissons
ratio by:
E
E
28.57 106 psi
1 m2
these may lead to quenched-in nonuniform

stress. During subsequent grinding operations,
the stress distribution undergoes a redistribution
and may, in extreme cases, lead to the rings distorting. Although heat treating receives the
blame, many similar problems may, in fact, be
attributed to the incoming raw material.
Validation of Net Opening

Calculation
Considerable work has been accomplished in
recent years in the area of heat treatment modeling and, in particular, of software capable of
Elastic constants for several materials are

given in Table 3 (Ref 16). (These must only be
used as a guide. For more rigorous treatment of
elastic and plastic distortion, actual values
should be obtained for each material and condition under consideration.)
From experience in dealing with distortion
problems in manufacturing that are attributable
to residual stress in tube stock, it was determined
that the following criteria could be used as a
guide to the acceptability of the calculated stress
levels.
Residual stress
Acceptability
5000 psi
500010,000 psi
10,000 psi
Acceptable
Borderline
Unacceptable
The criteria used to determine what residual

stress is acceptable must also be based on the
cross section of the ring: the more rigid the design, the higher the stress level tolerated. However, this only applies to holding tolerance during machining. During subsequent heat
treatment, the magnitude of residual stress may
still be sufficient to cause distortion during heating. As the temperature increases, the yield
strength of the steel decreases, allowing relaxation by plastic deformation. This is manifested
in the parts being out-of-round exiting the furnace.
In very thin sections, some of the stress may
be from the machining operation (e.g., feeds and
speeds too high).
In order to lessen the effects of residual
stresses during heat treatment, pre-heating at a
subcritical temperature has been found to be useful. A portion of the stress is allowed to relax
without causing a gross shape change.
Examples of extreme levels of residual stress
in tube material are usually seen when it was
impossible to machine an acceptable part. Stress
relieving one particular batch of tubes (new supplier) resulted in them moving so far out-ofround that they no longer fit in the multispindle
machine collets.
In an attempt to avoid out-of-round rings, die
quenching techniques are often used. However,
predicting distortion in heat-treated products. An

example is DANTE (Deformation Control Technology, Inc., Cleveland, OH), developed by a
collaborative team comprising Ford Motor Company, General Motors Company, The Torrington
Company, Eaton Corporation, IITRI, Colorado
School of Mines, The Department of Energy National Labs, Deformation Control Technology,
Inc., and under the auspices of the National Center for Manufacturing Science. In order to accurately predict the thermal and allotropic stressinduced size changes and distortion in
heat-treated products (including carburized), it is
necessary to incorporate many factors into the
finite element-based model, including the transformational characteristics of the steel, the ele-
2d
Residual stress
distribution
(a)
2d + 2 r
(b)
Fig. 11
Determination of longitudinal residual stress by the deflection method. (a) Rolled sheet. (b) Drawn bar. Source:
Ref 14
t
d
L
(a)
D1
D0
x
(b)
Determination of residual stresses in thinwalled tube by deflection methods. (a) Longitudinal stress. (b) Circumferential stress. t, thickness; d, deflection; D0, initial diam; D1, diam after slitting; x, net
opening displacement. Source: Ref 14
Fig. 13
Fig. 12
Diagrammatic representation of the deflection

formula derivation. Source: Ref 15

Using DANTE, it has been possible to subject
the net opening deflection method to a more rigorous analysis.
An example of the simulation of the opening
of a slit ring in the presence of an imposed hoop
stress is shown in Fig. 14 and 15.
Two-dimensional modeling was used, and
plane-strain conditions were assumed. The rings
were assumed to be 152 mm (6 in.) OD and man-
vated temperature properties and behavior of the

individual and mixed phases, geometry of the
part, quenching factors, heat-transfer coefficients, and so on. Several papers have been published on the development and application of the
model (Ref 1721). Commercialization of the
software package is being handled by Deformation Control Technology, Inc., Cleveland,
Ohio.
Table 2 List of formulas used for calculating approximate levels of residual stress in simple
geometries by deflection methods
Method
Formulas
Longitudinal stress in plate (Fig. 11a)
Etd
r 2
2L
Longitudinal stress in solid bar (Fig. 11b) where r is the radius of the bar
1.65 Erd
L2
Longitudinal stress in thin-walled tube (Fig. 13a)
Etd
L2
Circumferential (hoop) stress in tube from change in diameter where D0 and D1

are the initial and final diameters, respectively (Fig. 13b)
r Et
Circumferential stress in tube from net opening displacement x
r Et
Table 3
1
0
1
D1
1
1
x
D0
D0
p
Example: Determining Biaxial

Residual-Stress State
Elastic constants for some of the common metals

Modulus elasticity
Shear modulus
Materials
GPa
106 psi
GPa
106 psi
Poissons ratio
Aluminum alloys
Copper
Steel (plain carbon and low alloy)
Stainless steel
Titanium
Tungsten
72
110
200
193
117
400
10.5
16.0
29.0
28.0
17.0
58.0
28
41
76
66
45
157
4.0
6.0
11.0
9.5
6.5
22.8
0.31
0.33
0.33
0.28
0.31
0.27
Source: Ref 16
Fig. 14
ufactured from annealed 52100 bearing steel

with two different cross sections. A residual
stress profile was superimposed on the rings by
creating a 93 C (200 F) linear temperature gradient across the cross section from ID to OD.
For the ring in Fig. 14, simulating axial slitting
of the ring resulted in a net opening of 2.9 mm
(0.114 in.) with an accompanying redistribution
of the residual stress.
Using Eq 23, the average hoop stress in the
ring prior to slitting was calculated to be 172
MPa (25 ksi). In thinner cross section rings, the
net opening will be larger for the same level of
residual stress, as shown in Fig. 15. For a cross
section of 9.5 mm (0.375 in.) and a similar residual stress condition, the net opening is calculated to be 6.3 mm (0.247 in.). Equation 23
gives a calculated hoop stress of 179 MPa (26
ksi). These calculations used a Youngs modulus
value of 29 106 psi and Poissons ratio of
0.33.
It is interesting to note that one of the many
techniques used to exercise DANTE was based
on the carburizing of a modified Almen strip
(Ref 22, 23). Excellent agreement between the
predicted and measured transverse deflections
were obtained.
Simulation of the effects of residual stress on the net opening displacement of a cut 0.75 in. thick ring
Treuting and Read developed a method for determining the biaxial residual-stress state on the
surface of a thin sheet (Ref 24). The method assumes the metal behaves in an elastically homogeneous manner and that the stress varies, not
in the plane of the sheet, but only through the
thickness. To apply the method, the sheet specimen is cemented into a flat parallel surface, and
Fig. 15
Simulation of the effects of residual stress on the net opening displacement of a cut 0.75 in. thick ring
the thickness is reduced a certain amount by

careful polishing and etching. The sheet specimen is then released from the surface and measurements are made of the longitudinal radius of
curvature Rx., the transverse radius of curvature
Ry, and the thickness, t. Figure 16 illustrates the
orientation of the principle stresses and the curvature of the sheet. The measure values of radius
of curvature are expressed in terms of two parameters, Px and Py.
Px
1
m
Rx
Ry
Py
1
m
Ry
Rx
rx
(t
t)2
ry
(t
E
6(1 m2)
dPx
4(t0 t)Px 2
dt
t0
Px dt
E
6(1 m2)
t)2
dPy
4(t0 t)Py 2
dt
t0
Py dt
Values of dP/dt are obtained from the slope of

the curves of P versus t, and the integrals are
evaluated by determining the area under the P
versus t curve over the appropriate limits.
REFERENCES
Measurements of Rx and Ry are made for different amounts of metal removal, and Px and Py are
plotted against the sheet thickness, t. The residual stresses in the x and y directions of the sheet
are determined for any value of t by the following equations.
1. G. Parrish and G.S. Harper, Production Gas

Carburizing, Pergamon Press, 1985
2. F.W. Jones, JISI, May 1969, p 556562
3. Shot Peening Applications, 7th Ed., Metal
Improvement Company, Inc., 1995
z
z
y
y
x
x
t0
(a)
Fig. 16
Rx
Ry
(b)
(a) Coordinate system for measuring biaxial stress in thin sheet. (b) Curvature produced by removing material
from top surface
4. D. Breuer, Metal Improvement Company,

private communication, 1999
5. Heat Treating, Cleaning and Finishing, Vol
2, Metals Handbook, 8th ed., ASM International, 1964, p 41
6. Tenaxal, Ucon Quench AThe Fast Safe
Way to Quench Steel Alloys, Product Information Bulletin, Tenaxal, Inc., Milwaukee, WI, revised 1972
7. H.J. French, The Quenching of Steels,
American Society for Steel Treating, 1930,
p 133
8. B. Prime, Residual Stress Measurement by
Successive Extension of a Slot: The Crack
Compliance Method, Los Alamos National
Laboratory, Publication LA-UR- 983857,
1998
9. W. Gremaud, I. Cheng, M. Finnie, and B.
Prime, The Compliance Method for Measurement of Near Surface Residual
StressesAnalytical Background, J. Eng.
Mater. Technol., Vol 116, p 550555
10. I. Cheng, M. Finnie, M. Gremaud, and B.
Prime, Measurement of Near Surface Residual Stresses Using Electric Wire Machining,
J. Eng. Mater. Technol., Vol 116, p 17
11. Kovac, Residual Stress Measurements in
Bevel Gear after Different Production
Phases, J. Mater. Eng. Perform., Vol 3 (No.
1), Feb 1994, p 6164
12. E. Shramm, J. Szelazek, and A.V. Clark,
DynamometerInduced Residual Stress
in Railroad Wheels: Ultrasonic and Saw Cut
Measurements, National Institute of Standards and Technology Publication NISTIR
5043, Report Number 30, March 1995
13. J.E. Gordon, The New Science of Strong
Materials, Penguin Books, 1968, p 258
261
14. G.E. Dieter, Mechanical Metallurgy, 1st
ed., McGraw-Hill Book Company, New
York, 1961
15. R.L. Anderson and E.G. Fahlman, A

Method for Measuring Internal Stress in
Brass Tubes, J. Inst. Met., Vol 32, 1924, p
367383
16. G.E. Totten and M.H. Howes, Steel Heat
Treatment Handbook, Marcel Dekker Inc.,
New York, 1997, p 256261
17. D. Shick et al., Development of a Carburizing and Quenching Simulation Tool: Determination of Heat Transfer Boundary Conditions in Salt, Second International Conf.
on Quenching and Control of Distortion
Proc., ASM International, 1996
18. D. Brammann et al., Development of a Carburizing and Quenching Simulation Tool: A
Material Model for Low Carbon Steels Undergoing Phase Transformations, Second
International Conference on Quenching and

Control of Distortion Proceedings, ASM
International, 1996
19. C. Anderson et al., Development of a Carburizing and Quenching Simulation Tool:
Numerical Simulation of Rings and Gears,
Second International Conference on
Quenching and Control of Distortion Proceedings, ASM International, 1996
20. B.L. Ferguson, A.M. Freborg, and G.J. Petrus, A Software Tool to Simulate Quenching of Alloy Steels, Heat Treating Progress,
to be published
21. M.T. Lusk, Y.K. Lee, H.J. Jou, W.E. Elliott,
and G.M. Ludtka, An Internal State Variable
Model for the Low Temperature Tempering
of Low Alloy Steels, First International

Conference on Thermal Process Modeling
and Computer Simulation, Jiatong University, Shanghai, March 2000
22. V.C. Prantil, M.L. Callabresi, G.S. Ramaswamy, and J.F. Lathrop, Simulating Distortion and Residual Stresses in Carburized
Thin Strips, ASME J. Eng. Mater. Technol.,
to be published
23. M. Henriksen, D.B. Larson, and C.J. Van
Tyne, On the Analysis of Distortion and Residual Stress in Carburized Steels, ASME J.
Eng. Mater. Technol., Vol 114, 1992, p
362367
24. R.G. Treuting and W.T. Read, J. Appl.
Phys., Vol 22, 1951, p 130134

DOI: 10.1361/hrsd2002p099

Measurement of Residual Stresses

C. Ruud, Pennsylvania State University
THIS ARTICLE PROVIDES an insight into

the principles, practices, and limitations of residual-stress measurement procedures for steel. It is
not meant to provide sufficient detail for the performance of the various methods described, but
references are cited where such procedural details may be found (e.g., Ref 15 or for general
inquiries, www.residualstress.com on the Internet).
There have been many methods and techniques proposed for the measurement of residual
stress, but only a few may be applied in practice
on components ranging from small to very large,
such as bridges and aircraft. These few include
x-ray diffraction (XRD) and blind hole drilling
with electrical resistance strain gages. For some
of the methods described, the component in
which residual stresses are to be measured must
be brought to the measuring instrument, but for
others the measurement devices are portable and
may be brought to the component (Fig. 1). However, in some cases it may be feasible to remove
a section from the component and bring that section to the residual-stress measuring device.
Great caution must be observed in this sectioning
because it will change the stress field by relieving and/or inducing stresses.
have resulted in disagreement between residualstress measurements made by the mechanical

stress relaxation type methods discussed in the
section Introduction to Destructive Procedures
and the XRD methods discussed in the section
Nondestructive Techniques in this article. The
disagreement is nearly always because the volume of the component in which the stress is measured is not the same in the two methods, and
thus a different portion of the stress gradient is
measured by each. Ruud et al. (Ref 6) showed
that the hole-drilling method resulted in measured stresses approaching that of the stresses
measured at the surface by XRD when the holedrilling results were extrapolated to the surface.
Furthermore, because of the steep gradients, unless the stress field caused by the process is well
understood, stress measurements must be performed at many locations in the manufactured
solid in order to establish the magnitude and distribution of the stress field of interest. Many researchers in residual-stress techniques have fo-
Any manufacturing process that changes the

shape of a solid, or where severe temperature
gradients exist during the process, causes residual stress. By their very nature, processes that
change the shape of a solid cause nonuniform
plastic deformation in the solid, which leads to
residual stress. These processes include forging,
rolling, drawing, machining, and so forth. Also,
processes that produce high thermal gradients in
a solid often lead to residual stress. These processes include quenching, casting, welding, and
so forth. Furthermore, processes that induce localized phase changes produce residual stress.
These processes include martensitic hardening.
The residual stresses caused by manufacturing
processes usually show very steep residual stress
to distance gradients as shown in Fig. 2. Many
of the other articles in this Handbook discuss and
describe in detail the causes and mechanisms of
residual-stresses development in manufacturing
processes. The steep gradients typical of residual
stresses induced by manufacturing processes
) dxd > 200mmMPa )

y
650
MPa
Welding
d y
~ 200
dx
3 mm
650
MPa
Machining
d y
~ 3000
dx
y
0
0.1 mm
y
650
Measurement of residual stresses on oil-drilling

platform component weldments using a portable
x-ray diffraction instrument
MPa
Drawing
d y
~ 600
dx
Drawn
cup
0
Fig. 1
Fig. 2
3 mm
x
x
Residual-stress magnitudes and distributions typical of a 650 MPa yield strength steel

cused on enhancing the accuracy of
residual-stress measurement and have ignored
the fact that from a practical standpoint, tens to
hundreds of stress measurements are needed to
define the stress field of interest. Thus, many of
the residual-stress measurement techniques require too much time to perform and are thus impractical. This includes some of the techniques
developed for XRD, strain gaging, and other
methods. Measurement times on the order of a
second are available with XRD and some other
methods, and automated stress mapping has been
performed with such techniques (Ref 7). Another
concern in the measurement of residual stresses
in manufactured components is that the area or
volume over which the stresses are resolved
must often be on the order of 1 mm (0.04 in.) or
less. This is because the residual-stress gradients
are usually quite steep, and measurement resolution larger than this tends to average the
stresses to such an extent that high stresses are
not detected.
Some characteristics of steel that can cause
error in residual-stress measurement by the various methods described in this article include
phase composition, plastic strain, grain size,
crystallographic texture, and others. For example, in a mixed ferrite/austenitic structure the residual stresses in the ferrite are invariably different than those in the austenite (see Table 1).
Also, Wimpory et al. (Ref 8) described the influence of varying amounts of cementite in a ferrite
matrix. The possibility of one or more of these
microstructural characteristics causing error in
the residual-stress measurements performed by
the methods described in this article should be
assessed by an expert in the method selected, and
details of the errors and their causes are not discussed in detail in this article.
The subsequent sections of this article discuss
and describe:
The need for measurementwhat problem is
the engineer or metallurgist trying to solve by

obtaining information about the residualstress field?
The nature of the residual-stress fields in
steelsexamples of the magnitudes and distributions.
The strain basics for residual-stress measurementselastic strain measured, not stresses.
Table 1 Sample of residual-stress readings
from a 316 stainless steel pipe weldment
Distance from
the weld
fusion line
Stress
(a) in
austenite
Stress
(a) in
ferrite
in.
mm
ksi
MPa
ksi
MPa
0.04
0.07
0.11
0.15
1
1.8
2.8
3.8
21
19
16
17
145
131
110
17
46
67
62
65
317
462
427
448
Note: The ferrite places tensile stresses on the lattice of the austenite,
while the austenite tends to compress the ferrite; therefore, the more
compressive the stress in the ferrite, the less compressive the stress in
the austenite. (a) The precision of these measurements was 3.0 ksi
(21 MPa).
The destructive procedures of residual-stress
measurementthese procedures are all

based on sectioning or removal of material to
cause a redistribution of the residual stress,
which is measured as a strain change.
The semidestructive methods of residualstress measurementthese methods are
based on the same principle as the destructive
methods or on the perturbation of the residual-stress field by other means.
The nondestructive methods of residual-stress
measurementthese methods do not permanently disturb the residual-stress field, but
directly measure the atomic lattice strain
caused by the stress or measure some physical
property perturbed by the lattice strain.
Need for Residual-Stress

Measurements
Before the engineer or metallurgist commits
to measuring residual stresses in some component or workpiece, he or she must be sure that
the reason for the measurement is clearly understood. The major reasons that residual stresses
are of concern are:
Failures that are suspected as being caused by
fatigue, stress corrosion, corrosion fatigue, or

hydrogen embrittlement
Assessment for the continued serviceability
of a component, for example, life assessment;
this is usually focused on a concern for inservice failure
Distortion occurring during processing of a
component
Distortion of components during storage or in
service
It is extremely important that the investigator
understand the mechanism for the inducement of
the residual-stress field of concern. As implied
by the other articles in this Handbook, most
cases of suspected harmful residual-stress fields
are induced by manufacturing processing or repair procedures, although sometimes abusive
service conditions or an accident may have
caused them (Ref 9). When manufacturing processes or sometimes repair procedures are
judged the most likely source of the residual
stresses, it is often possible to predict the magnitude and distribution of the residual stresses.
Such information can be obtained through consulting the literature or the application of computer modeling (Ref 1018 or the Internet at
www.residualstress.com). A preconceived
model of the residual-stress field will aid the investigation in determining the best method to
measure residual stress and the location and
number of measurements that need to be made.
Nevertheless, sometimes the cause of the residual-stress field is not evident, and the investigator is compelled to perform measurements as
a means to determine the cause. In such cases,
measurement methods and location must be selected without the aid of a priori knowledge of
the stress field, and it is prudent to consult the
literature and experts in the field of residualstress measurement and manufacturing processes.
Nature of Residual Stresses

Residual stresses are the inevitable consequence of thermomechanical processing of steel.
The resulting stress fields usually are nonuniform and show high stress gradients. For example, Fig. 2 illustrates the residual-stress magnitudes and distributions typical of a steel with
a 650 MPa yield strength. Because of the high
stress gradients, tens to hundreds of residualstress measurements with resolution on the order
of 1 mm may be required to identify precisely
the maximum stress and its location.
The characteristically nonuniform, high stress
gradient nature of residual stresses require that
either the induced stress field is well understood
and predictable, or many residual-stress measurements must be performed on one or more
components in order to reveal the nature of the
stress fields. Often, scientists attempt to gain an
understanding of a residual-stress field by making a few measurements on one or two components, and from this often arise erroneous conclusions regarding the nature of the stress field.
A few measurements may be useful if the scientist or engineer knows the distribution of the
stresses a priori. However, this is seldom the
case, and tens to hundreds of measurements are
required on a single component or many samples
to really understand the residual stresses induced
by a given manufacturing process. This means
that the measurement method must be as rapid
and labor-efficient as possible. Some of the new
semidestructive hole-drilling procedures and
XRD and ultrasonic (for special cases) instrumentation meet this criteria. Stress mapping is
offered with some XRD instruments to map
stresses over the surface of a component (Ref
19). Also, because the stress gradients are often
very high, the measurement method must be able
to resolve the stresses in dimensions on the order
of 1 mm or less. Here, some of the hole-drilling
procedures and recently developed XRD instruments can offer the best resolution.
Finally, the component in which residual
stresses are to be determined is often too large
to be brought to a laboratory, and removing sections is not a rational solution. Note that sectioning often disturbs the existing stress field to the
extent that it is not possible to reconstruct the
original stress field from the residual stress measured in pieces removed from the original whole.
Hole-drilling, XRD, and ultrasonic instrumentation are available as portable devices that may
be brought to the component in the field (Ref 7,
9, 20, 21).
Stress Measurement
A number of procedures and methods have
been applied to determine the residual stresses
Measurement of Residual Stresses / 101

extant in a metallic component, usually as a result of manufacturing processing. However,
stress is never the quantity measured because
stress is a quantity that is applied to a metal and
can only be measured in the process of its application. What is invariably measured to determine residual stress is elastic straineither the
elastic strain resulting directly from the existing
residual stress in the metal or the elastic strain
change resulting from relief of some portion, or
all, of the existing residual stress. The stress that
is causing, or has caused, the strain is then calculated using the applicable elastic constants for
the metal.
The methods described in the section Nondestructive Procedures in this article measure,
directly or indirectly, the strain response of the
metal to the residual stress in situ, while the
methods described in the sections Destructive
Measurement Procedures and Semidestructive
Procedures measure the strain change caused in
relieving some or all of the residual stress in the
metal.
Destructive Measurement
Procedures
The first concern in selecting a destructive residual-stress measurement procedure is whether
it is reasonable to destroy one or more components or samples in order to determine the residual stresses. Usually this implies that one or a
few of the components are a small portion of the
total number produced. Also coupled to this decision is whether the one or more components in
which the residual stresses are to be measured
are representative of all the others. In other
words, how great is the expected variation of the
residual-stress field from part to part?
As mentioned at the beginning of this article,
the need for the residual-stress information about
a component must be clearly understood. This
includes whether the triaxial stress field must be
established or if the uniaxial or biaxial condition
of the stress field along specific directions is sufficient. Examples of these three situations are
discussed in this section. However, in any case,
the fact that residual stresses are usually not uniform in any direction and show high stress gradients must be kept in mind when stress-measurement criteria are selected.
Destructive methods of residual-stress measurement are fundamentally stress-relaxation
procedures; that is, the information is obtained
by relaxing the residual stress in some finite-volume element of the component and measuring
the resulting strain change. The strain change is
then used, along with applicable assumptions
about the nature of the stress field, to reconstruct
the original stress field. Assumptions about the
nature of the stress field include the magnitudes
and gradients in the stress field and whether it is
sufficient to assume that the gradients are one-,
two-, or three-dimensional. In particular, the gradients that exist will dictate the size of the element that is to be isolated and made stress-free;
that is, the higher the stress gradient is, the

smaller the finite element must be in the direction of that gradient. It must be emphasized that
the larger the element and the higher the stress
gradient, the less quantitative and more qualitative are the measurement results. Electrical-resistance strain-gage technologies are emphasized
as the dominant method of strain measurement
due to their economic, procedural, and precision
advantages over other methods. However, modern XRD equipment when available has all of
these advantages as well and can be used to measure the stresses existing before and after sectioning.
A generic destructive stress-relief procedure is
described first, along with the issues generally
involved in each procedural step. This section on
destructive methods concludes with some discussion of qualitative chemical methods of residual-stress measurement.
Generic Destructive Procedure

Once the decision is made to measure the residual stresses destructively, the following steps
are usually applied in a typical stress-relief technique for residual-stress measurement.
Stress-Field Conditions. The engineering
problem for which the residual-stress information is needed must be analyzed. This need is
often generated by failures of the component in
service or by anticipated failures due to problems
with similar components. Distortion of a product
in storage or during manufacturing can also be a
concern. The shape of the componentthat is,
cylinder, disk, plate, and so forth, or some irregular shapemust be considered. This consideration, along with the process or processes by
which residual stresses were introduced, must be
analyzed. The justification for assumptions regarding the condition of the residual-stress field
can be established from these considerations.
This may lead to simplifying assumptions about
the stress-field condition such as axial symmetry
for a cylinder in which the dominant residualstress field is caused by quenching during heat
treat processing or about stress uniformity in the
surface plane of a plate where stress gradient
with depth is the major concern.
These assumptions and considerations lead to
the methodology, that is, equations, to be used
for computational reconstruction of the stress
fields from the measured strains.
Strain-Measurement Technique. With the
stress-reconstruction approach established, the
method of strain measurement and consequently
the number and/or spatial frequency of measurements can be determined. The strain-measurement technique selected will greatly affect
the resolution of the stress measurement because
of the spatial precision inherent in the technique.
There are a number of techniques that have
been used to measure the strain induced by the
relief of stresses due to sectioning or material
removal in destructive residual-stress measurement. These include mechanical gages, often dial
gages, employed with specially made jigs and
fixtures, reflected light schemes, photoelastic

coatings, and electrical-resistance strain gages.
However, since the 1960s, the use of the latter
has become dominant due to the variety, availability, and precision of these gages. They are
available as uniaxial, biaxial, and rosette gages
of many sizes. The section Strain-Measurement
Methods in this article provides some detail regarding these methods. Also, since the late
1970s, extensive use of XRD has been applied
to provide rapid and numerous stress measurements on sectioned components to gain information regarding the internal stress field (Ref 2,
2225).
Preparation for Strain Measurement. With
the strain-measurement technique selected, the
measurement location must be established and
the component and/or element prepared for the
measurement by, for example, attaching strain
gages. A prestress-relief reading must now be
made before stress relaxation and isolation of the
element is initiated.
Isolation of Gaged Element. With the measurement technique in place, material removal to
isolate the gaged volume must be performed.
The technique for material removal, or sectioning, must be carefully considered because mechanical chip-removal processes such as lathe
turning, milling, sawing, grinding, and so forth
introduce surface residual stresses that can be as
great as the yield strength of the strain-hardened
material and several thousandths of an inch (tens
of micrometers) in depth. The section Sectioning and Material-Removal Methods in this article discusses methods to isolate the gaged element.
Post-Stress Relaxation Measurement. After
the residual stresses have been relaxed and thus
the elements isolated, strain measurements are
repeated, the final reading is subtracted from the
initial to obtain the strain change resultant from
the residual-stress relaxation. The resultant
quantities are then used in the residual-stress reconstruction equations to obtain the original
stress state of the component. The stress-reconstruction equations were selected as a result of
the assumptions made about the stress-field conditions described previously and in more detail
in the following sections.
Stress-Field Condition Assumptions

Engineers and research scientists have approached the measurement of residual stresses
using destructive methods with the aid of assumptions about the stress-field conditions, including that the stresses in only one axis are of
interest in order to simplify the measurement and
reconstruction of the stress field (Table 2). These
have included certain uniaxial, biaxial, and triaxial stress-field assumptions.
Uniaxial Conditions. A procedure applicable
only to the measurement of residual stresses in
rods, cylinders, and tubesthat is, components
with axial symmetrywas reported by Heyn
(Ref 26). In this work, it was assumed that the
stresses were axially symmetric and that only the

longitudinal stresses were of interest. Thus, the
change in length of the components (cylinders,
rods, and tubes) was measured after removal of
an axially symmetric layer from the outside radius, or it was bored out of an inside radius. The
length of the component was measured after
each layer-removal process and entered into
various equations described in their paper. This
procedure is called the Bauer-Heyn method (Ref
26) and is applicable only to measuring longitudinal stresses in axially symmetric components. It does not measure radial or circumferential stresses.
A second procedure assuming a uniaxial stress
field, or that only the stresses in one axis are of
interest, was proposed by Stablein (Ref 27). Here
the component was a bar with a rectangular cross
section, and the residual stresses acting along its
length and varying through its thickness (smallest dimension) were measured. The material
from one face of the bar, one of the two faces
with the largest area, was removed by milling.
Equations used to reconstruct the original stress
field are described in Ref 27. The depth of the
removed layer must be significantly greater than
the depth of plastic deformation caused by milling (see the section Sectioning and MaterialRemoval Methods) and sufficient to cause a
measurable bend in the material. The length of
the opposite face of the bar from where the material was removed is measured before and after
layer removal to determine the effect of the removal of the stressed layer. Presently, this is usually done using electrical-resistance strain gages
(see Strain-Measurement Methods), but was
performed in the past by sensitive mechanicalgaging techniques. This measurement can also
be accomplished by measuring the bend in the
bar with suitable mechanical gages and fixtures,
for example, a cantilever-beam approach. This
procedure is applicable only to components of
rectangular cross section where the stresses parallel to the length are to be measured as they vary
through the thickness.
Biaxial Conditions. A procedure applicable
to axially symmetric components is the Mesnager-Sachs boring-out technique (Ref 28, 29). The
Table 2
technique is applicable to cylindrical components with an axially symmetric distribution of

stresses. Here the change in length and diameter
of the component are measured as material is
removed by axially boring-out material from the
inside to produce a hollow cylinder. Presently,
biaxial electrical-resistance strain gages (see the
section Strain-Measurement Methods) are
usually attached to the outside of the component
to measure the dimensional changes in the axial
and tangential directions. The strain change results are entered into equations described in Ref
28 and 29.
A procedure applicable to measuring biaxial
residual stresses homogeneous over the planar
surface of a flat metal plate or sheet was proposed by Trenting and Read (Ref 30). It was
based on removing uniformly thin layers of the
metal on one side of the sheet or plate and measuring the changes in curvature as the layers
were removed. It was assumed that the stresses
were constant over the plane of the sheet or plate
and varied only through the thickness. Electricalresistance strain gages (see Strain-Measurement Methods) or mechanical gaging may be
used to measure the change in curvature.
Another procedure for measuring biaxial
stresses homogeneous over the planar surface of
a metal plate was developed by Gunnert (Ref
31). This procedure assumes that a biaxial stress
condition was uniform throughout the depth of
a circular groove that was milled around an elemental volume of material to render it stress
free. Thus, the strain change on only one surface
was measured. The mechanical-gaging technique involved measuring the distance between
each of four sets of shallow holes drilled in the
element before the groove was milled into the
surface using a core drill. The distance between
each set of holes was measured before and after
the groove was produced and provided the information necessary to calculate original, assumed biaxial residual-stress condition parallel
to the surface of the plate. Theoretically, only
three sets of holes are required to measure the
biaxial stresses, but Gunnert used a fourth set to
improve the accuracy. A mechanical gage,
termed an extensometer, was used to measure the

distance between each set of holes. It should be
noted that this technique could be used to measure the gradient in the biaxial stress condition
by pausing in the milling operation at selected
depths and measuring the distance between the
holes at each groove depth. Also, as with many
of the older techniques that originally applied
mechanical strain-gage apparatus, electrical-resistance strain gages or modern XRD techniques
could be used. A procedure involving the drilling
of a blind hole and electrical-resistance strain
gage is somewhat similar to Gunnerts original
technique and is described in the section on
semidestructive methods.
A more accurate procedure was later used to
measure residual stresses in pipe weldments.
Here the component (pipe) was divided into a
network (grid) of squares, and biaxial electrical
strain gages were placed in the center of each
grid square on the outside diameter of the pipe.
The pipe was then sectioned as shown in Fig. 3
into elements assumed to be stress free, and the
strain induced by the stress relief was read from
the gages. In placing the gages only on the outside diameter, the biaxial stress field was assumed to be uniform with depth; however, had
gages been placed on the inside and outside diameter a more complete measurement of the
stress field could have been obtained, albeit assuming a linear variation in the residual stress
from the outside to the inside surface.
A variation on the Gunnert procedure described previously was later published (Ref 1,
32). This variation was also used for application
to plates and implied that the triaxial stress field
could be measured by the technique. It assumed
that there was a homogeneous residual biaxial
stress field that varied with depth through the
plate thickness. Four holes were drilled in a
square pattern through the plate thickness. The
distances between all of the holes was then measured at selected hole depths. Next, a circular
groove was milled in steps of, for example, 2
mm around the drilled holes using a core drill as
in Gunnerts original procedure (Ref 31). The
distance between the holes at the various depths
Summary of destructive residual-stress measurement procedures
Component shape
Rods, cylinders, tubes

Rectangular cross-section bar
Plate, sheet
Cylinder, plate
Plate, weldment
Stress-field-condition assumptions
Stress direction measured
Method
Uniaxial stresses, axial symmetry

Biaxial stresses, axial symmetry
Uniaxial stresses varying through thickness
Homogeneous planar, biaxial stresses
varying through thickness
uniform through thickness
varying through thickness
Planar biaxial stresses varying through
thickness
Planar biaxial stresses varying through
thickness
Triaxial
Longitudinal
Longitudinal, radial
Longitudinal
Biaxial in the plane of the
component
component
component
component
component
All
Bauer-Heyn
Mesnager-Sachs
Stablein
Treuting and Read
Various
Triaxial
All
All
Moore and Evans

Johanssen
Gunnert
Gunnert
Rosenthal and
Norton
Moore and Evans
Chen
Section and reference
Triaxial Conditions (Ref 26)

Biaxial Conditions (Ref 28, 29)
Triaxial Conditions (Ref 27)
Strain-Measurement Techniques (Ref
30)
31)
Strain-Measurement Techniques (Ref 1,
32)
33)
Strain-Measurement Techniques (Ref 2)
Preparation for Strain Measurements
(Ref 34)
Strain-Measurement Techniques (Ref 2)
25)

was then measured at each core depth until the
core was milled completely through the plate.
The cored plug was then assumed to be completely free of residual stress. The measured distances were then used to reconstruct the original
biaxial stress condition of the plug at each cored
depth. The mechanical measuring gage used in
this technique was similar to that used in Gunnerts first technique (Ref 31).
Another approach to measuring residual
stresses that has a broader application with respect to the shape of the component and the
stress-field distribution was proposed by Rosenthal and Norton (Ref 33). It is applicable to
plates and plate-shaped weldments.
The procedure involved cutting two narrow
blocks having the full thickness of the plate, each
with its long axis parallel to one of the assumed
biaxial principal, residual-stress directions in the
surface of the plate (see the near side of Fig. 4).
Thus, the long axes of the blocks are perpendicular to each other and parallel to the face, the
largest area surface of the plate. The smallest
dimension of the block should be several times
smaller than the plate thickness and its largest
dimension at least twice the thickness. The block
then can be further sectioned in order to determine the biaxial stress variation through the
thickness of the component. This proceeds by
first cutting the block at the location representing
the midthickness of the component plate, then
removing thin slices parallel to the original surface and from the bisected block as shown in the
upper part of Fig. 4.
The change in strain of the blocks is measured
using shallow holes or dimples in the original
surfaces of the component. These gage points are
located along the long axis of the block on the
original faces of the plate (see Fig. 4). The distance between these gage marks is measured before and after removal of the blocks from the
plate and after each sectioning of the block (Ref
33).
This procedure assumes a constant biaxial
stress field over the length of the blocks, which
is not the case in welded plates in the direction
transverse to the weld. The far side of the block
in Fig. 4 shows a sectioning procedure that
would reveal the stresses parallel to the weld
along a gradient transverse to the weld. The far
side of the block in Fig. 4 shows a sectioning
procedure that would reveal the stresses parallel
to the weld along a gradient transverse to the
weld. Also, electrical-resistance strain gages
could be used instead of measuring the distance
between shallow holes or dimples.
Another approach to a constant biaxial stress
field in a flat plate, varying only through thickness, was described by Moore and Evans (Ref
2). They relied on XRD for the measurement of
the strains from which the stress was calculated,
and the procedure consisted of removing layers
0
40
10 20 20 20
50
from the surface of the plate and measuring the

strains existing at each layer. The process assumed that the stress perpendicular to the surface
was zero. It should be noted that in order for this
procedure to be valid, the areas measured by
XRD would have to be free of plastic deformation caused by layer removal (see Sectioning
and Material-Removal Methods).
Triaxial Conditions. In reality, in most components in which residual stresses have been induced, usually due to manufacturing processes,
the stress field is triaxial and varies from point
to point (element to element) in all three directions. Thus, a number of destructive procedures
and stress-field condition assumptions have been
applied in order to measure the three-dimensional residual-stress-field condition existing in
most components of practical engineering interest. Two of these are described in this section.
Chen (Ref 34) revised Rosenthal and Nortons
approach to deriving the triaxial residual-stress
condition (Ref 33) as follows. The typical
method of residual-stress measurement is by mechanically removing part of a body and measuring the change of stress in the rest of the body.
The method of Rosenthal and Norton instead
deals only with a small element that has been cut
free from a plate. The sectioning procedure consists of removal of a narrow block from a plate
with gages attached, followed by splitting the
block in half with gages attached on the top and
bottom surface of the block, and then successive
slicing of both halves from the midsection to the
outer surface, as shown in Fig. 4.
When the half-block is sliced to a thickness of
0.1 in., gages are removed and stresses are measured (at least two points) on the surface by
XRD. The basic assumptions are:
Partial stress relief occurs in the direction of
90
Initial
sectioning
270
Weld
the long axis of the block and a total stress

relief occurs in the direction transverse to the
long axis.
The small amount of stress relaxed in the remainder of the block follows a linear law
through the thickness when a thin slice of
metal is removed.
Variation of transverse stress along the axis
of the weld is small in the middle portion of
the plate weldment.
45
10
Saw cut
180
Fine cutter
(i.e., green abrasive saw 0.5 mm thick)
10
Further
sectioning
5
10
Final
sectioning
Gage
Weld
10
10
10
Rosette gage
Residual-stress measurement of a girth-welded pipe by strain gaging and sectioning. Note that strain gages
shown in the final sectioning should be placed on the pipe prior to initial sectioning; and for a more complete
analysis several of the layer sections detailed in the final sectioning step should be strain gaged and sectioned. Dimensions
given in mm
Fig. 3
Welded steel plate, the near side of which shows

the two narrow blocks suggested in Rosenthal
and Nortons procedure (Ref 33). The far side shows several
blocks sectioned to reveal the stresses parallel to the weld
with a gradient transverse to the weld.
Fig. 4

The determination of residual stresses may be
divided into the following steps:
1. Determination of e1 and e,
t which represent
the amount of strain relaxed in longitudinal and
transverse directions by cutting one longitudinal
and one transverse block free from the plate.
This was done by determining the strain relief
between two indentations on the top and bottom
surface of each block using a mechanical gage
(see Fig. 4) and subtracting the strain-gage readings on the blocks from the initial reading from
the plate. Note that this measurement can be performed using electrical-resistance strain gages or
XRD (Ref 7, 22, 23).
2. Determination of strain relaxed on the top
and bottom surfaces by splitting each block in
half and then successively slicing the blocks.
3. Determination of the strain relieved (using
at least two points) on the top and bottom surface
using sensitive mechanical gages to measure the
distance between indentations and using XRD
after the thickness of the top and bottom halves
of the blocks had been reduced to 0.1 in. Note
that this restriction was necessary because they
are using conventional scanning x-ray instrumentation. With modern x-ray instruments, measurements can be made on any size of specimen
(Ref 7, 19, 20, 2224).
4. An enhancement of the Rosenthal and Norton procedure (Ref 33) was suggested by Chen
(Ref 34) where the determination of the residual
stress remaining in the top and bottom slices was
measured using XRD. Here the lattice strain was
measured in the remaining slices in at least two
places near the gage points using XRD, and the
absolute residual stress, S*t and S*1 remaining
was calculated as described in the section Stress
Measurement in this article. These XRD-determined stresses measured for each block were averaged and designated S*
t or S*
1 for that face. The
original stress S present before slicing was determined by:
St S*t Eet
(Eq 1)
S1 S* Ee1
The total strain e may be obtained by dividing

St or S1 by the modulus of elasticity, E.
5. Computation of stress relieved by cutting
the blocks from the plate was accomplished using the values of e and e obtained in steps 1
and 3, the amount of longitudinal and transverse
stress (St and S)
1 relaxed by cutting the blocks
from the plate:
S1
E
mE
(e1 me)
(et me)
t
1
1 m2
1 m2
St
E
mE
(et me)
(e1 me)
1
t
1 m2
1 m2
(Eq 2)
where m is Poissons ratio. Equation 2 is valid

for the case where the length of the block is at
least twice the thickness of the block. In case of
shorter blocks:
S1
E
E
(e1 me)
t
1 m2
1 m2
me
t
m 2 b2
e1
1 b
E
E
St
(et me)
1
1 m2
1 m2
me1
m 2 b2
et
1 b
(Eq 3)
where b is the correction factor for a shorter

block. When plotted for the top and bottom faces
of the block and joined by a straight line, these
values give the stress S relaxed by cutting the
block from the plate.
6. Computation of stress (S) relieved by splitting and slicing the blocks was done using:
2 d (1 ) 2(e e )
(S)
a tE
1 det
et et
2 dl
0.5 (1 l)
0
3(1 )
(5 4)et (1 )eb0
0
(Eq 4)
eb eb0
2 dl
0.5 (1 l)
a
(5 4)eb0 (1 )et 0
(Eq 5)
where l is the thickness of the block; is the

fraction of the total thickness removed; e,
t e
b are
the relaxed strain measured on top and bottom
surface for position ; and et 0, eb0 are the relaxed strain on top and bottom surface when
splitting the block in half.
7. The total stresses relaxed across the thickness of the block in the longitudinal and transverse directions were obtained by:
S1 S1 S1
(Eq 6)
St St St
8. Determination of shearing stresses in the

longitudinal and transverse directions were determined by:
sxy syx
Sw S v
2
(Eq 8)
Sy
sxy
syz
Y 0
y
x
z
(Eq 9)
Sz
sxz
syz
Z 0
z
x
y
(Eq 10)
Differentiation of Eq 8, 9, and 10 with respect

to x, y, z, respectively, yields:
2Sx
2sxy
2sxz
0
x2
xy
xz
(Eq 11)
2Sy
2sxy
2syz
0
y2
xy
yz
(Eq 12)
2Sz
2sxz
2syz
0
z2
xz
yz
(Eq 13)
Subtracting the summation of Eq 11 and 12 from

Eq 13, results in:
2Sz
2Sx
2Sy
2sxy
2
2
2 2
z
x
y
xy
1 deb
(S)
(1 ) 2(eb eb0)
a bE
2 d
3(1 )
Sx
sxy
sxz
X 0
x
y
z
(Eq 7)
where s is the shear stress and Sw and Sv are the

stresses measured in the direction w and v making angles 45 and 45, with the longitudinal axis. However, if the biaxial stress condition
is assumed, only the strain relief in one 45 direction need be determined since they would be
equal. This requires that either a block at a 45
angle to the longitudinal block be removed, or
that electric residual-strain-gage rosettes be
used, or XRD measuring at the 45 angle.
9. Determination of stress in the thickness direction was determined by the equilibrium equations:
(Eq 14)
Variation of Sx along the x-axis is small in the

middle portion of the plate; therefore, the first
term in Eq 14 can be neglected, and Eq 14 can
be approximated using:
2Sz
2
2
(Sy1 Sy0)
sx2y2
z2
Dy21
Dx2Dy2
(Eq 15)
where Sy0 is the value of Sy at the axis, Sy1 is the

value of Sy1 at a distance Dy1 on either side of
the x axis, and sx2y2 is the value of sxy at a distance Dx2 and Dy2 from the y axis and x axis,
respectively.
Therefore, the stress in the thickness direction
Sz can be obtained through double integration
with the boundary condition that Sz vanishes at
both the top and bottom surfaces. In this way,
the entire triaxial residual-stress state was determined. Note that this procedure assumes that the
residual stresses are uniform along the length of
each block.
A variation on Rosenthal and Nortons
method (Ref 33) using electrical-resistance
strain gages or a nondestructive technique such
as XRD is as follows. The largest face of the
blocks described by the thickness of the component (plate) and the longest dimension of the
blocks is divided into a two-dimensional grid of
elements. An electrical-resistant strain gage is
placed on each element, or a nondestructive measurement such as XRD is performed, and the
block is sectioned along the grid lines to produce
elements that are assumed to be stress free. Note
that if a nondestructive technique such as XRD
is used, the plastically deformed surface created
by removing the block from the original component must be removed. This is best done using
electropolishing (see the section Sectioning and

Material-Removal Methods). If electropolishing of these cut faces is done to remove the plastic deformation and resultant residual stress induced by a mechanical cutting procedure, and
XRD is applied, then the blocks need not be sectioned. The measured XRD stress will provide
the absolute residual-stress-field condition in the
block, andcoupled with the strain relieved by
the original removal of the block from the
platethe entire triaxial residual-stress condition of the plate can be obtained.
This variation of Rosenthal and Nortons
method (Ref 33) provides the information necessary to derive the biaxial stress condition of
each block, which can in turn be used to derive
the triaxial condition of the original plate. The
stresses on each of the measured faces of the
blocks must be measured in three directions to
provide the information necessary to obtain the
principal stresses in the block faces in each element. Note that the strain change caused by the
sectioning of the blocks must be added to the
strains measured in each element.
This procedure can be applied to a weldment
with a single weld through its center as described
by Rosenthal and Norton (Ref 33), or to a more
elaborate stress field, for example, where two orthogonal welds existed in the component (plate).
With a single weld, only the block intersecting
weld needs to be sectioned into elements because
the residual-stress field in the block parallel to
the weld is likely to be constant along the direction parallel to the weld.
Moore and Evans (Ref 2) proposed mathematical procedures for the reconstruction of the
original three-dimensional residual-stress fields
in cylindrical and flat plate components, and
Constantinescu and Ballard (Ref 3) proposed a
modification of Moore and Evans work. They
proposed using XRD as the measurement technique and presented stress-reconstruction equations for the following conditions:
Solid cylinder bar, rotationally symmetric
are then used to calculate the original stresses

rh(r) and rz(r), as well as the radial stress rr(r).
The theory of elasticity provides nine partial
differential equations: the three equations of
equilibrium and the six equations of compatibility. Unique solutions are possible, depending on
boundary conditions, and for the case considered
they give the following working formulas:
rr(r1)
r1
rhm(r)
dr
r
rz(r1) rzm (r1) 2
metric stresses
Hollow cylinder bar, rotationally symmetric
stresses
Flat plate, biaxial stresses
The Moore and Evans procedures were summarized in the Society of Automotive Engineers
Handbook (Ref 4) as described in the following
paragraphs.
Solid Cylinder Bar, Rotationally Symmetric
Stresses. It is presumed that the residual-stress
distribution has both rotational and longitudinal
symmetry, except near the ends where measurements are avoided. Stresses are therefore functions of the radius, r, and do not depend on the
angle, h, measured around the cylinder, nor on
the distance z taken parallel to the axis. By repeatedly removing thin concentric shells, the
stresses on the exposed surface in depth can be
obtained. The circumferential and longitudinal
measured stresses rhm(r) and rzm(r), respectively,
rzm(r)
dr
r
r1
rh(r1) rhm(r1) rr(r1)
rr(r1, h)
(Eq 17)
r1
g0(r)
2r1
2 g1(r)cos h dr
r
r
rh(r1, h) rhm(r1, h)
g0(r)
2r1
2 Re[g1(r)eih] dr
r
r
rh(r1, h) rhm(r1, h)
(Eq 19)
r1
g0(r)
6r1
2 Re[g1(r)eih] dr
r
r
(Eq 20)
(Eq 25)
2h0(r)
8r1h1
2 cos h dr
r
r
(Eq 26)
2r1
g1(r) sin hdr
r2
(Eq 27)
srh(r1, h)
r1
2r1
Im [g1(r)eih]dr
r2
(Eq 21)
(Eq 22)
where Re and Im indicate real and imaginary,

respectively, of whatever follows.
The measured stresses rhm (r1, h) and rhz (r1,
h) are related to the g and h functions by a Fourier series relation:
rhm (r1, h)
gn(r)enih
rzm (r1, h)
hn(r)enih
r1
srh (r1, h)
r1
If stresses are measured at four points around

the cylinder 90 apart, and, regardless of symmetry, the g and/or h functions from Eq 23 become:
h 0
rm0 (r) g0(r) g1(r) g1(r)
h 90
rm1 (r) g0(r) ig(r) ig1(r)
h 270 rm3 (r) g0(r) ig1(r) ig1(r)

(Eq 28)
in which the summation was expanded using

values of n 1, 0, and 1. This is a reasonable approximation, providing the stresses do
not drastically vary with h.
Adding the equations cancels the g1 and g1
terms, giving:
g0(r) 14[rhm (r) rhm (r)
r1
2h0(r)
8r1
2 Re[h1(r)eih] dr
r
r
R
r1
g0(r)
6r1
2 g1(r) cos h dr
r
r
h 180 rm2 (r) g0(r) g1(r) g1(r)
r1
(Eq 24)
(Eq 18)
rz(r1, h) rzm(r1, h)
rz(r1, h) rzm(r1, h)
where r is the original radius and r1 is the radius

at depth of interest.
Solid Cylinder Bar, without Rotationally Symmetric Stresses. Stresses are again assumed to be
independent of z but allowed to vary in the circumferential h direction. Complex variable
methods give the following general solutions:
stresses
Solid cylinder bar, without rotationally sym-
(Eq 16)
rr(r1, h)
rhm (r) rhm (r)]

2
(Eq 29)
Subtracting the equations in pairs and then adding the results cancels the g0 and g1 terms, giving:
g1(r) 14{[rhm (r) rhm (r)]
0
i[rhm (r) rhm (r)]

1
(Eq 30)
Equations 29 and 30 can be used in the general

solutions, Eq 19 to 22. However, if the stresses
are symmetric with respect to a plane through
the axis, then rhm (r) rhm (r) and Eq 30 becomes:
1
(Eq 23)
If the stresses are symmetric with respect to a

plane through the axis of the cylinder, the general solutions simplify because the g1(r) and
h1(r) functions will be real. This is a reasonable
approximation when the residual stresses are induced by heat treatment or straightening and the
equations become:
g1(r) 14[(rhm (r) rhm (r))]

0
(Eq 31)
Equations 29 and 31 are then used in Eq 24

to 27, along with similar formulas for the h function, determined from Eq 23 in the same way. It
is again interesting to note that Eq 16 to 18 are
included in the general solution of Eq 19 to 21,
when the stresses are rotationally symmetric.
The shear stress of Eq 22 or 27 exists only as a
result of asymmetry of the h stresses. When sym-

metry is obtained, the g1(r) terms are zero, as is
srh(r1, h).
Hollow Cylinder Bar, Rotationally Symmetric
Stresses. If R1 is the inside radius of the cylinder,
similar analysis generalizes the previous equations to the following final formulas:
1 r r
R21
2
1
r1
r2
rhm(r)
dr
R12
r
rz(r1) rzm(r1) 2
(Eq 32)
(Eq 33)
r21 R21
2
2 rr(r1)
1 R1
(Eq 34)
The above equations include those of the solid

cylinder when R1 is zero.
Flat Plate, Biaxial Stresses. It is assumed that
the residual stresses in a flat plate of uniform
thickness depend only on the distance from one
of the flat surfaces of the plate, except, of course,
near the edges. It is also assumed that the principal stresses are rx and ry, lying in the plane of
the flat surfaces, and that the stress normal to the
flat surfaces, rz, is zero at all points sufficiently
distant from the edges.
From the authors assumptions and conditions
of equilibrium, the true stresses rx(z1) at depth
z1 can be expressed in terms of the measured
stress rxm (z1) by the relation:
6z1
z1
z1
rxm(z)
dz
z
rxm(z)
dz
z2
(Eq 35)
where H is the original thickness of the plate and

z1 is the distance from lower surface to uncovered depth of interest.
A similar expression holds for the y direction.
Equation 35 holds, even if rx and ry are not prin-
in.
I
II
III
I
4 ft
2 ft
1 in.
A 2 by 2 by 4 ft (0.6 by 0.6 by 1.2 m) weldment

showing the layers proposed by Johanssen (Ref
25) where the thickness (T) of the layers are 0.5 in. (12.5
mm) and rI ry, rII rx, and rIII rz.
Fig. 5
z1
z1
sxym(z)
z
sxym(z)
dz
z2
(Eq 36)
sxym (z) is determined from:
sin2 2sxym(z1) sin cos
r1
rx(z1) rxm(z1) 2
ram (z1) rxm (z1) cos2 rym(z1)
r2
rzm(r)
dr
R12
r
rh(r1) rhm(r1)
1/ 2
sxy(z1) sxym(z1) 2
dz 6z1
rr(r1)
cipal stresses, but in this case a shear stress also

exists, expressed by:
(Eq 37)
where is the acute angle that the measured

stress ram (z1) makes with the x axis. When measurements are taken 45 apart, sxym (z1) becomes:
rx
2
y2
ry
sxym(z1) r45(z1)
12[rxm(z1) rym(z1)]
Johanssen showed that the change in stresses

resulting from the removal of material can be
determined by Dr(x) and m(x). These were to be
measured at a number of positions xi, i 1,,
N, on the lower side of the plate. Drx(x) and m(x)
are the difference between the stress and deformation measured prior to and following the removal of material. Changes in the internal stress
conditions are thus calculated directly and need
not be calculated as accumulated stress changes
resulting from several layers of material being
removed.
By introducing Aireys stress function u(x, y)
that is defined as:
(Eq 38)
meaning that three x-ray stress measurements are

required after each layer removed.
Johanssen proposed a procedure (Ref 25) for
the determination of the three-dimensional residual-stress field in thick plate (plate weldments)
components using XRD techniques to measure
the strains on the surfaces of the plate and plate
sections, and upon removal of layers of surfaces.
The procedure includes the measurement of the
biaxial stress field existing on the top surface of
the component (see Fig. 5), assuming that the
stress perpendicular to the surface is zero. Material is removed from this surface by, for example, milling and electropolishing, or electropolishing alone (see Sectioning and
Material-Removal Methods) and the biaxial
stresses remeasured at the new depth. Each time
material is removed, the forces that the removed
layer exerted on the remaining component must
be accounted for, and the subsequent measurements corrected for this change in the stress field.
Johanssen based his method on the following assumptions:
When a layer of material is removed, the re-
sulting changes in the stress condition will be

linear elastic; that is, Hookes law is applicable.
The residual-stress distribution is constant in
the z-direction, except at the surface, and rz
is a principal stress in the z-direction (see Fig.
5).
Upon material removal, it is assumed that the
strain ez remains unchanged. Together with
the previous assumption, this implies that the
change in stresses can be treated as a plane
problem.
It is assumed that the stresses are symmetrical
with respect to the y-z plane. This assumption
is, however, not necessary. The procedure can
be developed to include asymmetrical stress
states.
Johanssens justification for his procedure to
measure the three-dimensional stress field in the
weldment shown in Fig. 5 is as follows (Ref 25).
2
2
and sxy
x2
xy
(Eq 39)
Equilibrium and compatibility are automatically satisfied if is a solution to the biharmonic

differential equation:
2 2u 0
(Eq 40)
The homogeneous boundary conditions are:

ry(x, 0) 0
(Eq 41)
y 0
sxy(x, 0) 0
(Eq 42)
sxy(a, y) 0
(Eq 43)
x a
rx(a, y) 0
(Eq 44)
and at y 0 there should be agreement with

experimental measurements; that is:
rx(x, 0) Drx(x)
(Eq 45)
Pickel (Ref 35) described a method for analytical solution of problems with this type of
boundary condition, using infinite, related series;
however, in this instance an approximate method
is used with trial solutions that are both more
convenient and numerically more stable.
r
The force T is divided into Tx(x) and Ty(x) (see
Fig. 6) and as a solution to Eq 45 to 47, u is
taken as u ux uy where ux is the solution
for the loading case where only Tx(x) differs
from zero (and therefore the boundary condition
ry(x, b) 0 must be applied), and uy is similarly
the loading case where only Ty(x) is not equal to
zero. This ensures that the trial solutions will be
sufficiently general and not involve any limitar
tions on T (x).
The solutions for ux and uy are then adjusted
so that they satisfy Eq 43 and as many of the
boundary conditions as possible. By direct substitution, it is easily verified that the trial solutions in the following satisfy all the boundary
conditions except rx(a, y) 0; however, this

condition can be filled and the measurement can
be fit using an approximate method, such as the
least-squares method. If this can be performed
sufficiently accurately, these trial solutions are
close to the exact solution, since it has been
shown that the solution is uniquely determined
by the stated boundary conditions.
For the loading case Tx(x):
y3
y2
Cnx
C0x
2
6
2
n1 n
(1 nb coth nb)ny sinh ny]cos n
(Eq 46)
is valid where n np/a and 2a and 2b are the

width and thickness of the plate after removing
material.
Equation 39 gives:
exy
Cnx[(nb sinh ny ny cosh ny)

(1 nb coth nb)ny sinh ny]cos nx
sxxy
2x
xy
Cnx[(nb(coth
n1
bb 1)(2 cosh ny ny sinh ny) nb

(sin hny ny cosh ny)]coshnx
(Eq 47)
vx
y
(Eq 52)
one obtains:
(Eq 48)
v(1 v)
y2
Cx
C0xy
E
2
vx(x, y)
Cnx
n1
1v
E
(Eq 49)
Cnx nby sinh ny 2b
n1
cosh ny
v
nby sinh ny
1v
Since plane strain has been assumed xz can be

calculated from:
(1 nb coth n y cosh ny
xz m(rx ry)
1
v
sinh ny
n
1v
(Eq 50)
where m is the Poissons ratio. Further, the strain

exy can be obtained from:
exy
2x
Cxy C0x
y2
(sinh ny ny cosh ny)]sinh nx
[nb(sinh ny ny cosh ny)
rxx
2x
x2
n1
[2nby sinh ny (1 nb coth nb)
x(x, y) Cx
rxy
1 m2
m
ry
rx
E
1 m
(Eq 51)
where E is the elastic modulus.

Since the condition for compatibility is satisfied, the displacement in the y-direction is
uniquely determined, excepting rigid body displacements, from the relationship:
cos x
sinh ny y cosh ny
(Eq 53)
A similar treatment is used for the loading case

r
T (x). In that case, the solution is given by:
y3
y2
C0y
6
2
y(x, y) Cy
Cny
n2
n1
[(1 nb coth nb)(sinh ny ny

cosh ny) 2n by sinh ny]cos nx
(Eq 54)
from which are obtained:

yx
2y
Cy C0y
y2
n1
Cny[2nb cosh ny 2n by sinh ny

(1 nb coth nb) (sinh ny
Removed layer
ny cosh ny)]cos nx
x
x(x)
Ty (x)
yy
2y
x2
(Eq 55)
Cny
n1
[(1 nb coth nb)(sinh ny ny
T (x )
cosh ny) 2nby sinh ny]cos nx

Internal forces

xy
2
Tx(x)
syxy
(Eq 56)
n1
Cny[nb sinh ny n2by cosh ny)

x
x(x)
(1 nb coth nb) ny
sinh ny]sin nx
(Eq 57)
Finally, the displacement is given by:

x(x)corr
v(1 v)
y2
Ey
C0yy
E
2
vy(x, y)
1 v
E
Cny b sinh ny
n1
nby cosh ny
(x)
x(x)
v
1 v
(b sinh ny nby cosh ny)
(1 coth nb) y sinh ny

x(x)
Residual-stress distributions, forces, and distortion of a plate before and after layer removal. Top: residual-stress
distribution in the x-direction in the center of the plate in the x-z plane; center: same as top after removal of a
layer with the forces Ti(j) caused by the residual stresses tending to distort the plate; bottom: same as center with the
distortion displacement shown (Ref 25)
Fig. 6
2
v
cosh ny
n
1 v
y sinh ny cos nx
(Eq 58)

By letting the stresses and displacements be
equal to the measured values, the following relations are obtained:
rxx(xi, 0) ryx(x2, 0) Drx(xi)
i 1, 2,
(Eq 59)
i 1, 2,
(Eq 60)
and satisfying the boundary condition stated in

Eq 44 at a suitable number of positions:
rxx(a, yi) rxy(a, yi) 0
r(z1) rm(z1) 2
m (xi, 0) m (xi, 0) m(xi)

x
using the exact equations of the previous sections. The method is summarized for two of the
previous cases as follows.
Flat Plate (see Flat Plate, Biaxial Stress). A
generalized solution is written:
i 1, 2,
(Eq 61)
Equations 53 to 61 form an overdetermined system, from which the unknown coefficients Cnx
and Cny can be determined using, for example,
the method of least squares. The desired correction of the stresses measured at the surface after
removing material can finally be expressed as:
rxcorr rxx(x, b) rxy(x, b)
(Eq 62)
z1
dz 6z1
z1
rm(z)
z
rm(z)
dz
z2
(Eq 64)
The subscripts x, y, or xy have been dropped,

because the form of Eq 35 and that of Eq 36 are
exactly the same. r(z1) represents the true stress
in any direction at depth z1 before a layer was
removed, and rm(z1) represents the measured
value at that depth.
The correction in stress c(z1) at z1 is the difference between the true and measured value,
given by:
H
c(z1) r(z1) rm(z1) 2
and
dz 6z1
z1
z1
m[rxcorr
b)
ryy(x,
b)]
(Eq 63)
Sikarskie (Ref 36) proposed a stress-reconstruction procedure (series method) when thin
layers were removed from the surface of a component. He described procedures applicable to
flat plates and solid cylinders. The procedure
works well for shallow depths (a few percent of
specimen diameter or thickness), or in instances
where the stress gradient over the total depth removed does not change too rapidly and is of essentially one sign. The practicality of this
method depends on the fit of the measured
stresses in depth by a Taylors series referred to
the surface values of stress and successive derivatives at the surface. When the method is applicable, very convenient relations are obtained,
which describe the stress correction in terms of
the influencing factors, for example, layer depth,
stress magnitude, stress gradient, and specimen
size. Judgment is necessary, however, in using
the series approximation that does not arise when
(Eq 65)
z1
Fig. 7
Stresses in flat plate after layer removal
H z1
H
H z1
H
r1
where, again, the subscripts r, h, and z have been

dropped because the form is the same.
When rr(r1) is desired:
r(r1) rr(r1)
rm(r) rhm(r)
(Eq 71)
When rz(r1) is desired,

r(r1) rz(r1)
rm(r1) rzm(r1)
c(r1) r(r1) rm(r1)
(Eq 72)
(Eq 66)
(Eq 67)
where Dz1 H z1. This correction is seen to

be approximately proportional to the magnitude
of the surface stress and thickness of the removed layer (Fig. 7). It is inversely proportional
to the specimen thickness.
By solving for Dz1, the question of proper
slice is given by:
1 Hc(z1)
4 rm(H)
(Eq 68)
Thus, for example, if the measured stress is to

be in error by less than 5%, c(z1)/rm(H)
0.05 and the appropriate slice depth is:
(Eq 73)
Again, expanding the integrand of Eq 70 in a

Taylors series and integrating term by term, a
final form for the correction is obtained:
R r1
1
R
2
c(r1) k rm(R)
R r1
R
[rm(R) Rrm(R)]
Dz1
H
(Eq 70)
rh(r1) is calculated from rhm(r1) using Eq 18.

The correction term in stress, c(r1), is written
as before:
where rm(H), rm(H), are true surface stress and

successive derivatives with respect to z at the
surface.
For shallow depths, only the first terms of the
series can be used and:
c(z1) 4rm(H)
rm(r)
dr
r
k 2
1
[2rm(H) Hrm(H)
3
2H 2rm(H)]
r(r1) rm(r1) k
rm(r) rzm(r)
[rm(H) 2Hrm(H)]
Dz1
Removed layer
The integrands are then expanded in a Taylors series referred to the surface values, after
which the integration is performed term by term.
The final form for the correction is:
H z1
c(z1) 4rm(H)
H
For a plate 0.4 in. (10 mm) thick, for example,

the slice depth is 0.005 in. (0.13 mm).
If the stress gradient is high, then the next term
in the correction series should be included and a
quadratic in Dz1 solved. This requires an estimate of rm(H) based on experience.
Solid Cylinder (see Solid Cylinder Bar, Rotationally Symmetric Stresses). A generalized
solution from Eq 16 and 17 is written:
k 1
rzcorr
ryx(x,
(Eq 69)
rm(r1) 0
rm(z)
z
rm(z)
dz
z2
1
(0.05)H
4
Dz1
1
6
[2rm(R) 2Rrm(R) R2rm(R)]
R r1
(Eq 74)
where rm(R), rm(R), and so forth, are the surface

stress and successive derivatives with respect to
z at the surface.
Insight into the factors that influence the correction holds exactly as previously discussed, as
do the limitations of the method.
Ruud et al. (Ref 2224) applied a modification of Johanssen method to measure the triaxial
stress condition of thick plate 214Cr-1Mo plate
weldments. Ruud et al. actually measured the
strains in all directions and calculated the
stresses, but did not correct for layer removal due

to the complex nature of the stress field. Ruud et
al. also measured the residual-stress condition of
expanded tubing including 304 stainless steel
tubing (Ref 37) but focusing on the residual
stresses on the inside surface of the heat-exchanger tube components.
Residual stresses parallel

to the milling direction, ksi
10
20
30
0
Distance from the workpiece surface, mils

Residual stresses at the surface and near the surface due to milling a medium-carbon steel work-
Fig. 8
piece
40
20
Residual stresses, ksi
0
20
40
60
80
100
120
10
12
14
Sectioning and Material-Removal

Methods
As discussed in the previous sections in this
article, many procedures require that the component (sample or part) be sectioned and/or
some material be removed from it to measure the
residual stresses. This is especially true for the
measurement of internal residual-stress fields
where the component nearly always has to be
sectioned to reveal the internal stress field. The
two exceptions to the necessity of sectioning and
material removal, neutron diffraction and ultrasonic methods, are described in the section
Nondestructive Procedures in this article.
Sectioning or material removal is required by a
particular residual-stress measurement procedure
(see the section Stress-Field Condition Assumptions) or method (see the section StrainMeasurement Methods). Mechanical chipremoval processes are usually applied because
of their economy and speed. All chip-removal
processes, including lathe turning, drilling, milling, sawing, grinding, and so forth, introduce
surface residual stresses that can be as high as
the yield strength of the strain-hardened metal
and several thousandths of an inch (tens of microns) in depth (Ref 3843). Figures 8 to 10
show the residual stresses in steels caused by
various machining processes. Furthermore, some
steels are especially prone to strain hardening
for example, austenitic stainless steelsand extra care must be used with these materials when
selecting a material-removal technique. Figure
11 shows plots of the residual stresses in 304
austenitic stainless steel caused by various grinding methods. Note that these plots are only a
sample and may not be typical. If the size of the
element in which the strain change is measured
is smaller, or thinner in the case of surface-depth
gradients, than about 0.1 in. (23 mm), then a
chemical or electrochemical material-removal
technique must be used to remove the surface
residual stresses caused by mechanical chip re-
Residual stresses in a 440 C (825 F) stainless

steel workpiece induced by facing
Fig. 9
100
80
20
0
20
40
60
80
Strain-Measurement Methods
Abusively
machine
ground
Handheld grinder
60
As discussed in the section Stress Measurement, all residual-stress-determination methods

measure elastic strain, not stress, and the residual
stress is calculated from the strain values. Several methods for the measurement of strain have
been applied in residual-stress studies and have
been mentioned previously. These methods include:
40
Handheld
flapper
wheel
20
0
100
120
20
140
10
12
14
Fig. 10
Residual stresses in an alloy steel workpiece

induced by turning
Samplings of residual-stress distributions induced in 304 stainless steel workpieces by

common grinding procedures
Fig. 11
moval. These techniques may be used solely or

in conjunction with, and after, the chip-removal
method. It should be noted that material-removal
techniques such as electrodischarge milling
(EDM) induce residual stresses (Ref 44) as do
chip-removal methods. Other methods, such as
laser, flame, or plasma cutting that cause heating
of the element, must be applied with caution because they may reduce the stress field by annealing before it can be measured.
Thus, the only methods for material removal
from a component surface that do not induce residual stresses are electrolytic or chemical polishing. Electropolishing is described in some detail in Ref 45, and guidelines are provided for
application to various ferrous alloys. In electropolishing, the electrolyte and operating conditions depend on the alloy being polished, as
shown in Ref 4 and 45. Electropolishing combined with XRD is used extensively to reveal
residual-stress gradients on machined, ground,
and hardened surfaces. However, application of
these techniques requires that the subsurface
stresses be corrected for the removal of prior surface layers (Ref 4).
Another concern when reducing components
to a more convenient smaller size to place them
on or in a measurement device is that the stresses
of interest are not changed by the sectioning.
Generally, plates should be cut to a length and
width of at least three times the thickness to
avoid end effects. Cylinders, both thin-walled
and solid, should be a minimum of three diameters in length.
Where the manufacturing process affects the
entire thickness of a component, such as heat
treatment or forging, it might not be advisable
to section without means of measuring the stress
change extensively over the entire component
before sectioning. In other words, selection of
the stress-measurement procedure and methods
should avoid sectioning unless techniques to
measure the effects of sectioning are applied before sectioning is initiated. On the other hand,
when the processes that have induced the residual stress produce only shallow stress fields, then
the three-times rule suggested previously is applicable.
Mechanical gages
Electrical-resistance gages
Optical gages
Birefringent methods
Diffraction methods (x-ray and neutron)
Ultrasonic methods
Magnetic methods

Diffraction, ultrasonic, and magnetic methods
are discussed in the section Nondestructive Procedures in this article.
Mechanical Gages. The application of mechanical gages such as those described by Heyn
(Ref 26), Stablein (Ref 27), Gunnert (Ref 31),
and others generally preceded the availability of
electrical-resistance strain gages and are not discussed due to their general lack of precision,
poor spatial resolution, and inefficiency.
Electrical-Resistance Strain Gages. Most
bonded electrical-resistance strain gages are
made from either metallic wire or foil materials.
There are also the recently developed semiconductor gages. A variety of sizes, shapes, and configurations are available including single-element gages and rosettes with two, three, or four
elements. Electrical-resistance strain gages are
available in sizes as small as about 1 mm and
thus provide a resolution of strain measurement
on that order. Information on electrical strain
gages is available in numerous sources including
Ref 46 and in reviews by Crites (Ref 47) and
Masubuchi (Ref 14) as well as suppliers (e.g.,
Measurements Group Inc., Raleigh, NC).
Changes in temperature tend to cause an apparent strain; therefore, some type of temperature compensation is needed. Frequently, a
dummy gage, which is not subjected to the
strain, is exposed to the same temperature as the
actual gage to provide a basis for comparison. A
temperature-compensated gage can also be used.
Gages must be bonded securely to the specimen. Various types of cements have been developed. Sometimes gages must be protected
from metal chips produced during machining as
well as from the oil or water. A number of systems have been devised for protecting gages under various conditions.
Brittle coatings are a simple inexpensive
strain gage that will only provide qualitative indications of residual stress. A brittle lacquer is
applied to the area where the stresses are to be
measured by material removal. After the lacquer
has cured (dried), a change in the stress field is
induced, and if the change is sufficient strain will
be produced in the lacquer, causing it to crack.
If the material removal is in the form of a hole
drilled in the lacquer, radial cracks indicate a tensile residual stress in the plane of the component
surface and circular cracks indicate compressive
stress.
Optical Gages. Light reflective methods can
be used to magnify the movement of a reflective
surface in a well-fixtured component that is held
securely in place during material removal to
change the stress field and therefore induce strain
change. This method can also be used if the component can be removed and replaced precisely in
a fixture and the position of the reflected light
measured before and after removal and replacement during which a change in the residualstress field was induced.
Laser Methods. Other techniques applying laser light have been proposed as well. These have
included shearography (Ref 48), interferometry
(Ref 49), speckle-correlation interferometry (Ref
50), and others. Vikram and others (Ref 50) suggested that a small volume of the material could
be stress relieved by heating via a laser to induce
a change in the stress field and the strain change
measured by an optical technique to reveal the
residual stresses existing in the volume before
heating. However, it must be recognized that
heating a volume of steel sufficiently to change
the residual-stress field will result in tensile residual stresses in the heated volume (Ref 51),
and this would likely be detrimental to the component in which stresses were being measured.
Birefringent Methods. Under the action of
stresses, transparent materials become doubly refracting (birefringent), and if a beam of polarized
light is passed through a model (under stress)
made of such a material, a colored picture is obtained from which the stress distribution can be
determined. This technique is called the photoelastic technique (Ref 46). The actual component
in which the stresses are to be measured is coated
with a photoelastic polymer. When residualstress changes are induced in the component,
strain changes are caused and transmitted to the
polymer coating, which then becomes birefringent. This can be observed and measured using a reflection polariscope (Ref 14). Instructions
for analyzing fringe patterns in this application
(nearly the same as those obtained in ordinary
photoelasticity) are provided by the manufacturer.
The photoelastic coating may be applied by
brushing a liquid polymer on the surface of the
specimen and polymerizing it by applying heat.
Alternatively, a prefabricated flat or contoured
sheet of polymer may be bonded to the part at
room temperature (Ref 52). The maximum strain
that can be measured ranges between 3 and 50%,
depending on the type of polymer used; the
strain sensitivity usually decreases with the increase in the maximum measurable strain.
Chemical Methods
A number of qualitative methods to detect residual stresses that may lead to stress-corrosion
hydrogen-induced cracking in steel have been
applied to specimens representing components
to be manufactured. Magnesium chloride solutions have been applied extensively to the study
of stress corrosion in austenitic stainless steels,
including some recent work by Bouzina et al.
(Ref 53).
Masabuchi and Martin (Ref 54) studied the
susceptibility of SAE 4340 steel weldments to
hydrogen-induced stress cracking. The test procedure was to immerse the weldment specimens
in a 4% H2SO4 aqueous solution charged with
H2 and to which two drops of a 5% P solution
of CS2 was dissolved. A direct current (dc) was
applied between a specimen and a lead anode to
provide a current density of 0.35 to 0.8 A/in.2
The crack patterns that developed were related
to the surface tensile residual-stress distribution
in each specimen.
Stress-corrosion cracking induced by residual
stresses in carbon and low-alloy high-strength
steels has been investigated by several researchers (Ref 5456). One procedure consisted of immersing the specimens in a boiling aqueous solution of 60% Ca(NO3)2 and 4% NH4NO3 for 31
h. The crack patterns that developed were related
to the surface tensile residual-stress distributions
in the specimen.
A number of standard practices for testing the
susceptibility of metals to stress-corrosion cracking have been published by ASTM, including:
ASTM G 38 Standard Recommended Prac
tice for Making and Using C-Ray Stress Corrosion Test Specimens, 1979
ASTM G 58 Standard Practice for the Preparation of Stress Corrosion Test Specimens
for Weldments, 1978
ASTM G 39 Standard Practice for Preparation and Use of Bent-Beam Stress Corrosion
Test Specimens, 1979
ASTM G 30 Standard Recommended Practice for Making and Using U-Bend Stress
Corrosion Test Specimens, 1979
ASTM STP 425 Stress Corrosion Testing,
1967
However, these tests for the most part do not

reveal the residual stress, but rather the susceptibility of the metal to cracking under known
stresses in the specified corrosion medium.
Semidestructive Procedures
Nondestructive methods of residual-stress
measurement are characterized as methods that
in no way affect the serviceability or reduce the
mechanical strength or other properties of the
component in which stresses are measured. Between the nondestructive and destructive methods, which have a severe effect on the serviceability, strength, and properties, are the
semidestructive methods. These are methods
that have a small to negligible effect on the components in which stresses are measured, or methods in which the component may be repaired after the measurement.
The methods that are considered semidestructive are those that require small holes to be
drilled or rings to be trepanned in the component
or indentations to be made in the surface. The
first two methods provide quantitative data and
the third only qualitative data.
Blind Hole Drilling and Ring Coring

The hole-drilling method was proposed in the
1930s (Ref 57) and is based on measurement of
the change in surface strain caused by stresses
relieved by machining a shallow hole in the testpiece. The principle is that removal of stressed
material results in the surrounding material readjusting its stress state to attain equilibrium.
The method has been standardized in ASTM E
837 (Ref 58). The ring-core method (Ref 59) is
also based on the strain caused by disturbing the
stress field, but in this case a relatively stressfree island of material is isolated by making a

shallow ring around a strain gage. This method
is also called trepanning. These two methods are
the least destructive mechanical stress-relief
techniques and are relatively simple and economical. They, as do nearly all stress-relief techniques, rely on electrical-resistance strain gages
to measure the strain change due to metal removal. Rosettes of strain gages are available especially for hole drilling. The size of the rosettes
has been progressively reduced over the last few
decades and are now available in sizes less than
about 10 mm from a number of manufacturers.
As with most residual-stress techniques, hole
drilling and ring coring have been applied
mostly to steels. Most applications have been
done on flat plate or cylindrically round parts
(Ref 5, 60, 61). Stresses can be determined at
various depths into the surface of the material,
down to a depth equal to the diameter of the hole
or core (Ref 62, 63). Kelsey (Ref 64), however,
observed that stresses with depth cannot be measured accurately to greater than half the hole diameter. The thickness of the layers in which
measurements are resolved is about 10 to 20%
of the hole or core diameter.
The equipment necessary to perform the measurement is reasonably inexpensive, portable,
and can be used in a manufacturing shop environment. However, experienced technologists
are needed to take the readingsfrom selecting
the area in which stresses are to be measured to
preparing the surface, applying the strain gages,
and reading and interpreting the data. Due to the
possibility of residual stresses being induced by
the hole-drilling or coring technique, prior calibration of the application is recommended in all
cases, with the possible exception of certain applications where holes are produced by abrasive
jet machining (AJM) (Ref 65).
Rendler and Vigness (Ref 66) developed calibration constants for cold-rolled steel that they
proposed as generally applicable to all metals
provided that the elastic constants were known.
However, they seem to have overlooked variations in the strain-hardening coefficients and the
accompanying residual stress, which exist between alloys and even between tempers of the
same alloy. Dini et al. (Ref 67) showed that direct experimental determination of the necessary
constants for any isotropic material with known
elastic constants can be eliminated by using data
available for cold-rolled steel and calculating
these constants using a formula presented. Despite the success that some researchers have
claimed in circumventing the development of
calibration constants, experimental calibration is
strongly recommended. This is best done by applying strain gages and drilling the holes in testpieces prior to stressing them known amounts
(Ref 65, 66). The AJM technique should be applied to any material with high propensity to
work-hardening during machining, for example,
austenitic stainless steel (Ref 68).
The following are general limitations and/or
concerns of hole drilling and ring coring:
Areas of high stress gradients should be
avoided because the stress gradient must be
assumed to be constant across the hole or ring

diameter.
Areas where stresses are greater than onethird the yield strength of the material are
likely to produce erroneous results due to local plastic yielding during metal removal.
The thickness of the part or specimen must
be at least four times the hole or core diameter.
Strain hardening of the steel in the vicinity of
the hole may result during metal removal,
which can cause tens of ksi (10 ksi 69
MPa) error.
Heating may result during the metal removal.
Holes or cores must be spaced at least eight
times their diameter apart.
The area in which stresses are to be measured
must be accessible to a rather bulky drilling
or coring alignment device.
Preparation of the surface for strain-gage adherence may induce residual stresses that introduce substantial error to the subsequent
measurement (Ref 69).
In conclusion, the drilling and ring-coring

methods are nearly nondestructive variations of
the destructive mechanical stress-relief techniques and require only rather simple equipment
and instrumentation. The state-of-the-art is relatively well developed compared to many nondestructive methods, some of which require considerable research and development work before
they will ever be suitable to general application
in terms of alloys and stress-field conditions.
Technological advancements in hole drilling and
ring coring have largely been due to advancements in the more general area of mechanical
stress-relief methods and research in new metalremoval techniques for metal fabrication.
Indentation Methods
Engineers and scientists have proposed the
use of indentors, such as those used to perform
hardness measurements, as a means to measure
or detect surface-residual stresses since the
1930s. Kokubo in 1932 reported that stresses applied under bending load changed the apparent
Vickers hardness values in carbon-steel rolled
sheets, both as-rolled and annealed (Ref 70). He
showed that tensile stresses tended to decrease
the apparent hardness and compressive stresses
tended to increase the hardness. The stresses applied in tension and compression were sufficient
to cause 0.3% strain.
In the 1950s, Sines and Carlson (Ref 71) proposed a method that required various amounts of
external loads to be applied to the component in
which residual stresses were to be measured
while hardness measurements were made. The
loads were made to cause both tensile and compressive applied stresses. The quality, that is,
whether the residual stress was compressive or
tensile, was then revealed by comparing the effect of the applied stress, and whether it was tensile or compressive, on the hardness measurement. At about the same time Pomey et al. (Ref
72) proposed that residual stresses could be measured by pressing a ball-shaped penetrator into
the component in which residual stresses were
to be measured and establish the relationship between the pressing load while it was progressively increased, and the electrical resistance at
the interface between the penetration and the
component. He maintained that a smaller decrease in electrical resistance indicated that portions of material under the ball were plastically
yielding and that the corresponding load on the
ball could be related to the existing residual
stress.
Later, Chiang et al. (Ref 73) provided a critique of several existing indentation analyses and
proposed an interpretation of indentations exhibiting hemispherical plasticity. Nevertheless, the
applications illustrated in this paper focused on
brittle materials and not metals.
There have been numerous papers published
proposing various approaches to interpreting the
indentation loads and shapes so as to estimate
the residual-stress field on the surface and nearsurface regions of materials. However, indentation methods have not earned the degree of confidence of XRD or hole-drilling methods for
general applications and thus are rarely applied.
Spot Annealing
Another method that has been proposed to
measure residual stresses in metal surfaces is to
reduce the residual stresses in a small volume by
annealing the metal in the volume. It has been
proposed that this annealing be performed by intense laser light (Ref 50). This technique was
envisioned to be similar to residual-stress relief
by removal of the material as accomplished in
the hole-drilling techniques. However, as Cullity
discussed in Ref 51, such localized heating
would induce high surface residual tensile
stresses in the heat-affected region.
Nondestructive Procedures
The methods described in the section StrainMeasurement Methods all measure the change
in some dimension (strain) of the component
produced by the removal of a finite volume of
stressed metal from that component. Thus, they
measure the strain induced by removing material
so as to perturb the residual-stress field. On the
other hand, nondestructive procedures measure
a dimension in the crystal lattice of the metal or
some physical parameter affected by the crystal
lattice dimension. Whenever a mechanical force,
resulting in stress that is less than the yield
strength, is placed on a solid metal component,
that component distorts and strains elastically.
That elastic strain results in a change in the
atomic lattice dimension, and this dimension, or
change, is measured by the nondestructive
stress-measurement procedures. For example,
the diffraction methodsx-ray and neutron
measure an actual crystal dimension, and this

dimension can be related to the magnitude and
direction of the stress that the metal is subject to,
whether that stress is residual or applied. Subsequently, in this section the following methods
of stress measurement are described:
X-ray diffraction
Neutron diffraction
Ultrasonic velocity
Magnetic Barkhausen noise
X-Ray Diffraction
X-ray diffraction techniques exploit the fact
that when a metal is under stress, applied or residual, the resulting elastic strains cause the
atomic planes in the metallic crystal structure to
change their spacings. X-ray diffraction can directly measure this interplanar atomic spacing;
Ns
X-ray source
S1
Film
Np1
Np2
S2
Specimen
surface
O
2
One-angle arrangement for the single-exposure technique. N, specimen normal; b, angle

that the incident beam makes with Ns, Np1, and Np2 the
normals to the diffracting planes 1 and 2, respectively; w1
and w2, angles between Ns, Np1, and Np2, respectively; g,
the angles between the incident beam and the diffracting
plane normals; Ro, camera radius; O, point of incidence of
the x-ray beam of the specimen; 1 and 2, diffracting planes
at various attitudes to the specimen surface; S1 and S2, measured parameters representing the distance from a reference point of known distance from the incident beam and
the diffracted x-ray beam position. S1 and S2, directly related to the Bragg angles, h1 and h2.
Fig. 12
Miniature x-ray diffractometer for the one-angle technique arrangement of XRD stress measurement. This device incorporates a Ruud-Barrett Position-Sensitive Scintillation Detector and is capable of being
inserted in a 4 in. (100 mm) inside diameter and measuring
residual stress (Ref 22).
Fig. 13
from this quantity, the total stress on the metal

can then be obtained.
Since metals are composed of atoms arranged
in a regular three-dimensional array to form a
crystal, most metal components of practical concern consist of many tiny crystallites (grains),
randomly oriented with respect to their crystalline arrangement and fused together to make a
bulk solid. When such a polycrystalline metal is
placed under stress, elastic strains are produced
in the crystal lattice of the individual crystallites.
In other words, an externally applied stress or
one residual within the material, when below the
yield strength of the material, is taken up by interatomic strain. X-ray diffraction techniques
can actually measure the interatomic spacings
that are indicative of the elastic strain in the specimen. Stress values are obtained from these elastic strains in the crystals by knowing the elastic
constants of the material and assuming that stress
is proportional to strain, a reasonable assumption
for most metals and alloys of practical concern.
Reference 74 describes XRD and instrumentation in detail.
There are three basic techniques for measuring
stresses, based on XRD. They are the doubleexposure or two-angle technique (DET), the single-exposure or one-angle technique (SET), and
the sin-square-psi or multiangle technique. The
angle of exposure referred to is that between the
incident x-ray beam and the specimen surface
normal. It should be noted that in any XRD
stress-measurement technique, x-ray peaks in the
far back-reflection range, that is, peaks with
Bragg angles (h) of near 90, are much preferred
because they show the greatest effect with a
given amount of applied or residual stress. This
is illustrated by:
r (2h1 2h2)
cot h1
E
1
p
2
1 m sin2w 180
(Eq 75)
where h1 is the Bragg angle of the planes diffracting at w1 and h2 is the Bragg angle of the
planes diffracting at w2. As h1 increases, its cotangent decreases; therefore, a larger difference
(2h1 2hw) would result from a given ru to
maintain an equality.
For a residual-stress measurement, the diffracting angle, h, of interatomic planes of at least
two different psi (w) angles with respect to the
surface normal must be measured (Fig. 12).
These planes are crystallographically equivalent
(same Miller indices, hkl ) and in the unstressed
state of the metal would have the same interatomic (d) spacing for the planes labeled 1 and
2 in Fig. 12 (Ref 51, 75, 76). In a stressed material, however, the two or more orientations of
diffracting planes are selected so that they are at
different angles to the surface; thus their normals
are at different psi (w) angles to the surface normal. Then, depending on the angle of these
planes to the stress vector, their interplanar
atomic spacing is increased or decreased by
varying amounts.
The most common sources of errors and misapplications in stress measurements by x-rays
are related to stress constant selection, focusing
geometry, diffracted peak location, cold-working crystallography, texture, grain size, microstructure, and surface condition. The source, significance, and correction techniques for these
errors will not be elaborated upon here; details
may be found in Ref 77.
A point of interest in the error sources concerns cold working and microstresses. As stated
previously, microstresses are usually considered
to be those manifested by strain variation across
single metallic grains. This strain variation is detected in XRD as broadening of the x-ray peak
a distinctly different phenomenon from the peak
shift caused by residual stresses. However, microstrain variation can be measured simultaneously with stress. This microstrain phenomenon has been proposed as a means of judging
cold work, dislocation density, and fatigue damage (Ref 78).
Despite the facts that x-rays provide stress
readings only to a depth of less than 0.001 in.
(0.025 mm) and that the error sources listed previously must be considered, the noncontact XRD
method is presently the only time-proven, generally applicable, truly nondestructive method
for measuring residual stresses. Its reliability has
been proved and documented by thousands of
engineers and scientists since the 1960s, beginning with Bolstad et al.s classic work at Boeing
using x-ray film cameras (Ref 79). This documentation includes measurement of stresses in
the Brooklyn Bridge (Ref 20) and tempering
evaluation of carburized steels. The Society of
Automotive Engineers considers the method of
sufficient practical importance to have printed
three handbook supplements on the subject (Ref
4), and another is under revision.
Even so, this nondestructive technology has
been largely restricted to the laboratory because
of the general lack of knowledge regarding the
state-of-the-art instruments and the limitations of
the more widely known and available conventional scanning XRD equipment. Instrumentation for bringing this technology into the field
and manufacturing area has advanced rapidly
since the 1980s, especially toward increased
portability, compactness, and speed of operation.
As shown in Fig. 13 and 14, instrumentation
has been developed and is commercially available for stress measurement in situ on the inside
diameter of 10 mm (4 in.) diam pipe, and soon
for 5 mm (2 in.) diam steel tube (Ref 22, 23).
Position sensitive x-ray detectors have been
largely responsible for these improvements to
both laboratory-based and field-deployable residual-stress-measuring instruments (Ref 9, 19,
2123, 8082). Also, with the speed of data collection being less than 0.1 s with conventional
x-ray tube sources, XRD stress measurement can
be performed on moving components (Ref 11).
Nevertheless, many engineers have been frustrated in applying XRD to residual-stress measurement. This has been largely due to crystallographers inexperienced in residual-stress

measurement, attempting to apply conventional
scanning x-ray diffractometers and techniques to
residual-stress measurement. For example, in
conventional XRD analysis and crystallography,
sharp resolution of the diffracted spectra is very
beneficial. However, in XRD stress measurement the need to measure psi (w) angles that are
not zero defocuses the beam, and attempts to refocus leads to significant error in the stresses
read (Ref 83). In XRD stress measurement, the
ability to repeatedly measure the position of a
defocused diffracted x-ray peak is more important than sharp resolution (Ref 84). Thus, it is
recommended in most cases that XRD residualstress measurement be performed by trained
technologists using x-ray instrumentation specifically designed and built for stress measurement, not conventional scanning diffractometers.
Software packages specifically for residualstress measurement used with conventional
scanning diffractometers do not in most cases
eliminate the mechanical and focusing problems
of applying these instruments to residual-stress
measurement. It is necessary to mount the component (or specimen) in which stresses are to be
measured on the conventional scanning diffractometer, which usually requires sectioning of the
component and which complicates and adds error to the measurement procedure.
Neutron Diffraction
Neutron diffraction (ND) is capable of measuring the elastic strains induced by residual
stresses throughout the volume of relatively
thick steel components with a spatial resolution
as small as 1 mm3. Such capabilities provide for
residual-stress measurement inside components
without sectioning or layer removal. Principal
ND methods, as with XRD methods, measure the
spacing between crystallographic planes in a
component, and this spacing is affected by residual and applied stress. The spacing between a
selected set of crystallographic planes (u) is related to the angle of incidence and diffraction of
21/ 4 in. (5.7 cm)
the neutron radiation (h), which are equal, and

the wavelength of the monochromatic radiation
(k) by Braggs Law:
k 2d sin h
(Eq 76)
The elastic strain (e) induced by the residual

stress perpendicular to the diffracting crystallographic plane then is related to d by:
e
d do
sin h0
do
sin h sin h0
(Eq 77)
where do is the distance between the unstressed

crystallographic planes. If the orientation of
principal stresses are known in the component,
the stress in any principal direction may be calculated by:
rA
E
(1 m)
m
(eA eB eC)
(1 2m)
(Eq 78)
where rA, rB, and rC are the principal stresses,

eA, eB, and eC are the strains measured in the
corresponding principal directions, and E and m
are the Youngs modulus and Poissons ratio,
respectively. If the principal stress directions are
unknown, strains in at least six directions must
be measured to determine the residual stresses
acting on the volume of material in which strains
are being measured.
Limitations and Applications. For residualstress measurements in most alloys, especially
steels and cast irons, the unstressed spacing (do)
between crystallographic planes at the exact
point of strain measurement is not known and
not easily measured. This means that do or ho in
Eq 77 cannot be precisely established, and this
leads to various degrees of error in the accuracy
and precision of ND residual-stress measurements. This condition is aggravated by the fact
that the elemental composition, and thus do, vary
considerably within a component and markedly
51/ 16 in. (12.9 cm)

41/ 8 in. (10.5 cm)
within the phase (e.g., martensite, austenite, ferrite) of the alloy at various locations. Additional
limitations are that the component must be
brought to a nuclear reactor, each strain measurement requires over an hour, a single stress
determination in one small volume of the component requires at least three strain measurements, and the measurements are very costly.
Nevertheless, the ND methods have been applied
to residual-stress measurements in steel weldments (Ref 85), cylindrical steel forgings, plastically deformed steel plate (Ref 86), steel rocket
case forgings (Ref 87), and many other types of
components.
Ultrasonic Velocity
The principle underlying the measurement of
stress and thus elastic strain by ultrasonic (acoustic) techniques is the phenomenon of an approximately linear change in ultrasound velocity with
applied stress. In addition, it has been shown that
under certain narrow conditions residual stress
can be measured by exploiting this phenomenon.
Stress is measured by inducing a sound wave in
the frequency of several megahertz into the
metal specimen and measuring the time of flight
or some other velocity-related parameter. Since
many other characteristics of metals besides
stress-induced elastic strain affect velocity, these
must be sorted out; however, neither the technology nor the fundamental knowledge for such
sorting is usually available. The great interest in
ultrasonic techniques for residual-stress measurement stems from their promise for three-dimensional nondestructive measurements within
the material.
Principle. A number of velocity-related phenomena have been used in various methods to
measure stress effects by ultrasound. All utilize
the deviation of the reaction of the metal from
the linearity of Hookes law of elasticity, r
Me, where r is stress, e is strain, and M is elastic
modulus. This has been referred to as the anharmonic property of the solid and may be represented by a power series r Me Ce2 De3
, where C is the third-order anharmonic
constant, D is the fourth, and so on. Most research done for stress measurement has used expressions in which terms past the third-order
constant, C, are dropped. Of the several anharmonic property effects that may be used to measure stress, the following are probably the most
exploited:
Velocity dependence on the elastic modulus
Dispersion of frequency amplitudes in sur-
face waves
11/ 4 in.
(3.3 cm)
Birefringence of orthogonally polarized shear
waves
Harmonic generation in surface waves

1.6 in. (12.9 cm)
Fig. 14
Engineering drawing of the miniature XRD stress measurement head showing major dimensions
A very simplified form of the anharmonic

stress strain law has been written as r Me
Ce2 and rewritten as r e(M Ce). The term
in parentheses is approximately related to the velocity of sound as qV 2 M Ce, where q is
the density of the medium and V is the velocity

of sound. This may be approximately rewritten
in terms of velocity dependence on strain as:
V
q 1 2M e
M
(Eq 79)
Then, to solve for strain, e 2(V Mq

2M)/C (Ref 88).
A simple view of the dependency of ultrasonic
velocity on the elastic modulus and density may
be shown by rewriting the equation pV 2 M
Ce in terms of V, differentiating, and dividing
by V to yield an expression for DV/V. The result
will readily show that a fractional change in elastic modulus or density would affect the velocity.
The density of metal, for which the Poisson ratio
is near 0.3, obviously changes as a compressive
or tensile stress is placed on the specimen, and
it is reasonable that the speed of sound would
then change.
Limitations and Applications. Ultrasonic
technology offers a number of types of wave
modes in which to probe metals; these include
bulk waves, such as longitudinal and shear, and
surface waves usually confined to Rayleigh type.
Each mode offers many unique parameters for
extracting information. As has been discussed,
the primary effect of stress-induced strain on ultrasonic propagation in metals is on velocity.
This may be detected in a number of ways, including measurements of wave velocity, shear
wave birefringence, and dispersion. However,
there are other characteristics of metals that affect the ultrasonic velocity to the same degree as
stress. These include crystallographic texture,
microstresses, multiple phases, coherent precipitates, composition gradients, and dislocation
density and distribution.
Crecraft (Ref 89) discussed velocity effects,
manifested as texture-induced birefringence, and
the marked change seen with ultrasonic frequency. He also reported birefringence due to
cold-work in nickel-steel specimens, but did not
attempt to separate the cold-work effects in
terms of texture, dislocation density, and so
forth. In the early 1950s, Bradfield and Pursey
(Ref 90) and Pursey and Cox (Ref 91) reported
showing the influence of small degrees of texture
on ultrasonically measured elasticity in polycrystalline bars. They showed how the true isotropic elastic constants can be determined using
measurements of both longitudinal and shear
wave speeds along several directions. They presented stereographic charts that illustrated the relationship between elastic behavior of cubic
crystals and results of x-ray texture determinations.
McGonagle and Yun (Ref 92) noted the coldwork effects in a paper comparing XRD results
with Rayleigh wave velocity measurements. Boland et al. (Ref 93) also recognized that other
material properties can affect ultrasonic velocity
and recommended that methods be developed to
distinguish stress-induced velocity changes from
those from other sources.
James and Buck (Ref 94) pointed out that
since the third-order elastic constants for most
structural materials are not readily available

from the literature, ultrasonic stress measurement must be calibrated relative to the particular
material being investigated. In the same paper
they discounted the possible effect of mobile dislocations on the sound velocity in structural engineering metals with high yield strengths due
to the short dislocation loop lengths prevalent.
However, they did mention that crystallographic
preferred orientation (texture) during deformation or fatigue is capable of severely modifying
the elastic constants on which the sound velocity
depends.
Papadakis (Ref 95) noted marked velocity
changes for ultrasonic waves in various steel microstructures, and Moro et al. (Ref 96) measured
the effect of microstructural changes caused by
tempering on the ultrasonic velocity in low-alloy
steel. Tittman and Thompson (Ref 97) evaluated
the near-surface hardness of case-hardened steel
with Rayleigh waves; because hardness in this
case is a combination of composition, microstress, and macrostress, the velocity change was
due to a combined effect.
The temperature sensitivity of ultrasonic
stress measurements has also been cited as an
important source of error. Salama et al. (Ref 98,
99) proposed that this dependence be used as a
means to measure stress, but also noted the
marked effect of dislocations and did not address
a methodology of separating the stress from the
dislocation effect.
Much of the work cited above is concerned
with attempts to measure the effects of a variety
of material properties on the changes in ultrasonic velocity. However, there apparently is no
comprehensive study that demonstrates the capability of quantitatively separating stress effects
on ultrasonic propagation from other variables
found in structural metals, such as dislocation
density or crystallographic texture. Furthermore,
most of the studies cited observed velocity
changes in bulk waves. Velocity measurements
on these waves must be measured through the
thickness of a component, and, as most metallurgists recognize, obtaining uniform properties
through thicknesses greater than a few millimeters, especially in steels, is difficult. The subtle
property variations to which ultrasound velocity
is sensitive and the inherent lack of homogeneity
in engineering metals present additional serious
problems for through-thickness stress measurements.
In spite of the microstructural variations in
manufactured steel products, success in the application of ultrasonic methods to residual-stress
measurement has been achieved in specific
cases. One is in the measurement of hoop
stresses in railroad wheels (Ref 100). Here
changes or variations of the residual stress in the
hoop direction is of concern while that in the
radial or axial direction can often be assumed to
be constant or negligible. Some techniques then
for the measurement of the residual hoop stresses
has relied on normalizing the hoop velocity
against the axial (Ref 101). Also, European railroads have monitored ultrasonic velocity along
the wheel rims during use and attributed changes

to residual-stress changes (Ref 102). Schramm
in his paper mentioned a number of approaches
for the application of ultrasound to the measurement of residual stresses in railroad wheels, and
these examples may find application in the measurement of residual stress in other axially symmetric shapes (Ref 100). Ultrasonic residualstress measurements have also been applied to
rails as reported by Egle and Bray (Ref 103) and
Bray and Leon-Salamanca (Ref 104).
Barkhausen Noise Analysis

The Barkhausen noise analysis technique
(BNA) is concerned with measuring the number
and magnitude of abrupt magnetic reorientations
made by expansion and contraction of the magnetic domains in a ferromagnetic metal. These
reorientations are observed as pulses somewhat
random in amplitude, duration, and temporal
separation and therefore are roughly described as
noise.
Applications. A few applications of BNA to
ferromagnetic metallic components have been
made, Gardner (Ref 105) mentions a number of
applications that include helicopter rotor blade
spans, autofrettaged gun tubes, gas turbine engine components, and rolling-element antifriction bearing components. In these examples, the
change in residual stresses caused by known service histories was measured.
Chait (Ref 106) qualitatively measured the residual-stress condition of a high-hardness laminar composite steel weldment and compared
some of the BNA data with XRD stress readings.
Sundstrom and Torronen (Ref 107) applied their
BNA method to a number of microstructural
measurements, including evaluation of grainsize measurement for low-carbon ferritic and ferritic-pearlitic steels, evaluation of anisotropy in
deep drawing and textured steels for electrical
applications, measurement of the degree of aging
in rimmed carbon steels, and pearlite morphology in steel wires. They have also measured iron
loss in magnetic material used for transformers
and have proposed using BNA for residual-stress
measurements, pointing out that quantitative results can be obtained if the material and its fabrication history are known and calibration is possible.
Most studies and applications of BNA to
stress measurement have focused on the uniaxial
stress state. However, Sundstrom and Torronen
(Ref 107) implied that the instrumentation they
used could simultaneously measure stress in two
directions to give biaxial stress conditions for
magnetic inspection of roller-bearing components, including BNA for monitoring residualstress change.
Summary. The BNA method certainly has
been demonstrated to be sensitive to the stress
condition in ferromagnetic materials (Ref 108).
Nevertheless, its possibilities for application are
limited by the condition that the material must
be ferromagnetic, the narrow total range of stress
sensitivity (i.e., 40 ksi, or 5.6 MPa), and the

shallow depth of measurement. The latter condition might be relieved by using magnetomechanical mechanical acoustic emission (MAE)
(Ref 109), an ultrasound analog to BNA. However, the sensitivity of either of these techniques
to other characteristics of metallic components
and the consequent need for calibration with a
nearly identical specimen severely restrict the
general applicability of BNA and MAE. Many
misapplications have been made that have severely damaged the reputation of the BNA methods (Ref 107, 110). Such restrictions can be removed only if the basic phenomena responsible
for the effect of microstructural properties on
BNA and MAE are understood and quantified in
terms of the signal.
Barkhausen noise analysis is not recommended where variations in elemental composition, phase composition, grain size, strain hardening, crystallographic texture grain shape, grain
orientation, carbide size and distribution, and
other microstructural characteristics accompany
variations in residual stress. A recent evaluation
of BNA by Allison and Hendricks (Ref 111)
confirms the uncertainty of BNA residual-stress
measurements.
5.
6.
7.
8.
9.
Summary and Recommendations

Only the destructive stress-relief, semidestructive hole-drilling, or nondestructive XRD
methods of residual-stress measurement are generally reliable over a broad range of steel alloys
displaying residual-stress fields induced by the
various manufacturing processes. Measurement
of residual stresses can be very expensive and
time consuming, and it is often worthwhile to
consult experts in the field before deciding on a
measurement method. Before an engineer or scientist who is inexperienced in residual-stress
measurement selects a method and attempts to
measure stresses he/she should consult someone
experienced in residual-stress measurement and
analysis.
10.
11.
12.
13.
REFERENCES
1. R. Gunnert, Method for Measuring TriAxial Residual Stresses, Document No.
X-184057-OE, Commission X of the International Institute of Welding, 1957;
Weld. Res. Abroad, Vol 4 (No. 10), 1958,
p 1725
2. M.G. Moore and W.P. Evans, Mathematical Corrections in Removal Layers in XRay Diffraction Residual Stress Analysis,
SAE Trans., Vol 66, 1958, p 340345
3. A. Constantinescu and P. Ballard, On the
Reconstruction Formulae of Subsurface
Residual Stresses after Matter Removal,
Fifth Int. Conf. Residual Stresses, Vol 2,
Institute of Technology, Linkopings University, Sweden, 1997, p 703708
4. Residual Stress Measurement by X-Ray
Diffraction, SAE J784a, Handbook Sup-
14.
15.
16.
plement, Society of Automotive Engineers, 1971

A.J. Bush and F.J. Kromer, Residual
Stresses in a Shaft after Weld Repair and
Subsequent Stress Relief, SESA Spring
Meeting Paper No. A-16, Society for Exp.
Stress Analysis, 1979
C.O. Ruud, P.S. DiMascio, and J.J. Yavelak, Comparison of Three Residual Stress
Measurement Methods on a Mild Steel
Bar, Exp. Mech., Vol 25 (No. 4), Sept
1985, p 338343
J.A. Pineault and M.E. Brauss, In-Situ
Measurement of Residual and Applied
Stresses in Pressure Vessels and Pipeline
Using X-Ray Diffraction Techniques, Determining Material Characterization, Residual Stress and Integrity with NDE,
PUP-Vol 276, NDE-Vol 12, American Society of Mechanical Engineers, 1994
R. Wimpory, G.M. Swallow, and P. Lukas,
Neutron Diffraction Residual Stress Measurements in Carbon Steels, Fifth Int.
Conf. Residual Stresses, Vol 2, Institute of
Technology, Linkopings University, Sweden, 1997, p 676681
M.E. Brauss and J.A. Pineault, Residual
Strain Measurement of Steel Structures,
NDE for the Energy Industry, NDE-Vol
13, D.E. Bray, Ed., American Society of
Mechanical Engineers, 1995
C.O. Ruud and P.S. DiMascio, A Prediction of Residual Stress in Heavy Plate Butt
Weldments, J. Mater. Eng. Sys., Vol 31
(No. 1), Jan 1981, p 6265
C.O. Ruud and M.E. Jacobs, Residual
Stresses Induced by Slitting Copper Alloy
Strip, NDC of Materials VI, Plenum Press,
1994, p 413424
E.F. Rybicki and R.B. Stonesifer, Computation of Residual Stresses Due to Multipass Welds in Piping Systems, ASME
78-PVP-104, American Society of Mechanical Engineers, 1978
M. Ehlers, H. Muller, and D. Loke, Simulation of Stresses and Residual Stresses
Due to Immersion Cooling of Tempering
Steel, Fifth Int. Conf. Residual Stresses,
Vol 2, Institute of Technology, Linkopings
University, Sweden, 1997, p 400405
K. Masubuchi, Analysis of Welded Structures: Residual Stresses, Distortion, and
Their Consequences, 1st ed., Pergamon
Press, 1980
M. Bijak-Zochowski, P. Marek, and M.
Tracz, On Subsurface Distributions of Residual Stress Created by Elasto-Plastic
Rolling Contact, Fifth Int. Conf. Residual
Stresses, Vol 2, Institute of Technology,
Linkopings University, Sweden, 1997, p
430445
D. Green and S. Bate, Calculation of Residual Stresses Using Simplified Weld
Bead Modelling Technique, Fifth Int.
17. H. Michaud, F. Mechi, and T. Foulquies,

Three Dimensional Representation of the
Residual Stress Distribution in Steel Wires
or Bars, The Fifth Int. Conf. Residual
534538
18. K.W. Mahin, W.S. Winters, T. Holden,
and J. Root, Measurement and Prediction
of Residual Elastic Strain Distributions in
Stationary and Traveling Gas Tungsten
Arc Welds, Practical Applications of Residual Stress Technology, ASM International, 1991, p 103109
19. J. Pineault and M. Brauss, Automated
Stress MappingA New Tool for the
Characterization of Residual Stress and
Stress Gradients, Proc. Fourth Int. Conf.
Residual Stresses, Society for Experimental Mechanics, 1994, p 4044
20. M. Brauss, J. Pineault, S. Teodoropol, M.
Belassel, R. Mayrbaurl, and C. Sheridan,
Deadload Stress Measurement on Brooklyn Bridge Wrought Iron Eye Bars and
Truss Sections Using X-Ray Diffraction
Techniques, ICB-9751, Proc. 14th Annual International Bridge Conf. and Exhib., Engineering Society of Western PA,
1997, p 457464
21. M.G. Carfaguo, F.S. Noorai, M.E. Brauss,
and J.A. Pineault, X-Ray Diffraction
Measurement of Stresses in Post-Tensioning Tendons, IABSE Symposium (San
Francisco, CA) 1995, Extending the Life
Span of Structures, Vol 71/1, International
Association for Bridge and Struct. Eng.,
ETH Honggerberg, CH-8093, Zurich,
Switzerland, 1995, p 201206
22. C.O. Ruud, P.S. DiMascio, and D.J.
Snoha, A Miniature Instrument for Residual Stress Measurement, Adv. X-Ray Anal.,
Vol 27, Plenum Press, 1984, p 273283
23. C.O. Ruud, R.N. Pangborn, P.S. DiMascio,
and P.J. Snoha, Residual Stress Measurement on Thick Plate Low-Alloy Steel Narrow Gap Weldments by X-Ray Diffraction, ASME 84-PVP-128, American
Society of Mechanical Engineers, 1984
24. C.O. Ruud, J.A. Josef, and D.J. Snoha, Residual Stress Characterization of ThickPlate Weldments Using X-Ray Diffraction,
Weld. Res. J., Supplement, March, 1993, p
875915
25. P. Johanssen, Determination of Residual
Stresses in Thick Welded Plated Utilizing
X-Rays and Layer Removal Techniques,
Part 1: Theory for Reconstruction of Original Stress State, Internal Report, Dept. of
Materials and Solid Mechanics, The Royal
Institute of Technology, Stockholm, Sweden, April 1978
26. E. Heyn, Internal Strains in Cold Wrought
Metals, and Some Troubled Caused
Thereby, J. Inst. Met., Vol 12, 1914, p 1
37
27. E. Stablein, Stress Measurements on Billets Quenched from One Side, Stahl Eisen,
Vol 52, 1932, p 1517

28. M. Mesnager, Methods de Determination
dees Tensions Existant dans un Cylindre
Circulative, C.R. Hebed. Seances Acad.
Sci., 1926, p 169
29. G. Sacks, Evidence of Residual Stresses in
Rods and Tubes, Feitsch. Metallkde., Vol
19, 1927, p 352357
30. R.G. Treuting and W.T. Read, Jr., J. Appl.
Phys., Vol 22 (No. 2), Feb 1951, p 130
134
31. R. Gunnert, Residual Welding Stresses,
Method of Measuring Residual Stresses
and Its Application to a Study of Residual
Welding Stresses, Alquist E. Wicksell,
Stockholm, 1955
32. R. Gunnert, Method for Measuring Residual Stresses in the Interior of a Material,
Document No. X-16257, Commission X
of the International Institute of Welding,
1957; Weld. Res. Abroad, Vol 6, 1960, p
1024
33. D. Rosenthal and J.T. Norton, A Method
for Measuring Triaxial Residual Stresses
in Plates, Weld. J., Vol 24 (No. 5), Research Supplement, 1945, p 295s307s
34. R. J.-S. Chen, Determination of Residual
Stresses in a Thick Weldment, M.S. Thesis, Department of Engineering Science
and Mechanics, The Pennsylvania State
University, Nov 1983
35. G. Pickel, Application of the Fourier
Method to the Solution of Certain Boundary Problems in the Theory of Elasticity,
J. Appl. Mech., 1944, p 176182
36. D.L. Sikarskie, On a Series Form of Correction to Stresses Measured Using X-Ray
Diffraction, AIME Trans., Vol 39, 1967, p
577580
37. A.R. McIlree, C.O. Ruud, and M.E. Jacobs, The Residual Stresses in Stress Corrosion Performance of Roller Expanded Inconel 600 Steam Generator Tubing, Proc.
Int. Conf. Expanded and Rolled Joint
Technology, Canadian Nucl. Society,
1993, p 139148
38. Y.W. Park, P.H. Cohen, and C.O. Ruud,
The Development of a Model for Plastic
Deformation in Machined Surface, Mater.
Manuf. Proc., Vol 8 (No. 5), 1994
39. Y.W. Park, P.H. Cohen, and C.O. Ruud,
Sensitivity of Shear Process on Metal Cutting to the Development of Residual
Stress, Rev. Prog. NDE, Vol 14, 1995, p
11831188
40. Y.C. Shin, S.J. Oh, and C.O. Ruud, Investigation of Residual Stresses of Machined
Surfaces by an X-Ray Diffraction Technique, NDC of Materials IV, Plenum Press,
1992, p 408418
41. J.E. Hoffman, D. Lohe, and E. Macherauch, Influence of Machining Residual
Stresses on the Bending Fatigue Behaviors
of Notched Specimens of Ck45 in Different Heat Treating States, Residual Stresses
in Science and Technology, Vol 2, DGM
Informationsgesellschaft Verlag, Germany, 1987, p 801814
42. E. Brinksmier, Residual Stressed in Hard

Metal Cutting, Residual Stresses in Science and Technology, Vol 2, DGM Informationsgesellschaft Verlag, Germany,
1987, p 839846
43. E. Schreiber and H. Schlicht, Residual
Stresses after Turning of Hardened Components, Residual Stresses in Science and
Technology, Vol 2, 853860, DGM Informationsgesellschaft Verlag, Germany,
1987, p 853860
44. F. Ghanem, H. Dishom, C. Braham, R. Fatallak, and J. Leider, An Engineering Approach to the Residual Stresses Due to
Electric-Discharge Machining Process,
Proc. Fifth Int. Conf. on Residual Stresses,
Vol 1, Ericson, Oden, and Anderson, Ed.,
ICRS-5, Institute of Technology, Linkopings University, Sweden, 1977
45. Electropolishing, Surface Cleaning, Finishing and Coating, Vol 5, 9th ed., Metals
Handbook, American Society for Metals,
1982, p 303309
46. Hetenyi, Ed., Handbook of Experimental
Stress Analysis, John Wiley & Sons, 1950
47. N.A. Crites, Your Guide to Todays Strain
Gages, Prod. Eng., Vol 33 (No. 4), 17 Feb
1962, p 6981; Equipment and ApplicationTodays Strain Gages, Prod. Eng.,
Vol 33 (No. 6), 19 March 1962, p 8593
48. Y.Y. Hung, Shearography: A New Optical
Method for Strain Measurement and Nondestructive Testing, Opt. Eng., Vol 21 (No.
3), 1982, p 391394
49. K. Li, Interferometric Strain/Slope Rosette
Technique for Measuring Displacements/
Slopes/Strains/ Residual Stresses, Proc.
of the 1997 NSF Design and Manufacturing Grantees Conf., University of Washington (Seattle, WA), National Science
Foundation, Arlington, VA, 1997, p 571
572
50. C.S. Vikram, M.J. Pedensky, C. Feng, and
D. Englehaupt, Residual Stress Analysis
by Local Laser Heating and Speckle-Correlation Interferometry, Exp. Tech., Nov/
Dec 1996, p 2730
51. B.D. Cullity, Elements of X-Ray Diffraction, 2nd ed., Addison-Wesley, 1978, p
469472
52. C.O. Ruud, Residual and Applied Stress
Analysis of an Alloy 600 Row 1 U-Bend,
NP-5282, R.P 53033, Electric Power Research Institute, 1987, p 211
53. A. Bouzina, C. Braham, and J. Ledion,
Evaluation and Prediction of Real Stress
State Stress Corrosion Cracking Specimens, Fifth Int. Conf. Residual Stresses,
Vol 2, Institute of Technology, Linkopings
University, Sweden, 1997, p 10601065
54. K. Masubuchi and D.C. Martin, Investigation of Residual Stresses by Use of Hydrogen Cracking, Weld. J., Vol 40 (No.
12), Research Supplement, 1961, p 553s
556s
55. C.E. McKinsey, Effect of Low-Temperature Stress Relieving on Stress Corrosion
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
Cracking, Weld. J., Vol 33 (No. 4), 1954,

Research Supplement, p 161s166s
W. Radeker, A New Method for Proving
the Existence of Internal Stress Caused by
Welding, Schneissen Schneider, Vol 10
(No. 9), 1958, p 351358 (see also abstract
by Claussen, Weld. J., Vol 38 (No. 4)
1959, p 300
J. Mathar, Determination of Initial Stresses
by Measuring Deformation around Drilled
Holes, Arch. Eisenhuttenwes., Vol 6, 1934,
p 277281; Trans. ASME, Vol 56 (No. 4),
1934, p 249254
Determining Residual Stresses by the
Hole Drilling Strain-Gage Method, E
837, Annual Book of ASTM Standards,
ASTM, 1983
H. Wolf and D.C. Sauer, New Experimental Technique to Determine Residual
Stresses in Large Turbine-Generator Components, presented at the Amer. Power
Conf. (Chicago, IL), 1 May 1974
H. Wolf and W. Bohn, Origin, Measurement and Assessment of Residual Stresses
in Large Forging for Turbines and Generators, Arch. Eisenhuttenwes., Vol 42 (No.
7), 1971, p 509511 (in German)
H. Wolf, E. Stucker, and H. Nowack, Investigations of Residual Stresses in the
Turbine and Generator Industry, Arch. Eisenhuttenwes., Vol 48 (No. 3), March
1977, p 173178
J. Lu and J.F. Flavenot, Applications of the
Incremental Hole-Drilling Method for
Measurement of Residual Stress Distributions, Experimental Techniques, Presented
at the SEM Spring Conference (Portland,
OR), Society of Exp. Mechanics, 510
June 1988
M.T. Flaman and B.H. Manning, Determination of Residual Stress Variation with
Depth by the Hole-Drilling Method, Exp.
Mech., Vol 25, 1985, p 205207
R.A. Kelsey, Measuring Non-Uniform Residual Stresses by the Hole Drilling
Method, Proc. Soc. Exp. Stress Anal., Vol
14 (No. 1), 1956, p 181184
A.J. Bush and F.J. Kromer, Simplification
of the Hole-Drilling Method of Residual
Stress Measurement, ISA Trans., Vol 12
(No. 3), 1973, p 249259
N.J. Rendler and I. Vigness, Hole-Drilling
Strain-Gage Method of Measuring Residual Stress, Exp. Mech., Vol 6 (No. 12) Dec
1966, p 577586
J.W. Dini, G.A. Beneditti, and H.R. Johnson, Residual Stresses in Thick ElectroDeposits of a Nickel-Cobalt Alloy, Exp.
Mech., Feb 1976, p 5660
F. Witt, F. Lee, and W. Rider, A Comparison of Residual Stress Measurements Using Blind-Hole, Abrasive-Jet and Trepanning Methods, Exp. Tech., Vol 7, Feb
1983, p 4145
P.S. Prevey, Residual Stress Distribution
Produced by Strain Gage Surface Prepa-
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
rations, 1986 SEM Conference on Experimental Mechanics, 1986

S. Kokubo, Changes in Hardness of a Plate
Caused by Bending, Sci. Rep. Tohoku Imperial Univ., Jpn., Series 1, Vol 21, 1932,
p 256267
G. Sines and R. Carlson, Hardness Measurements for the Determination of Residual Stresses, ASTM Bull., No. 180, Feb
1952, p 3537
J. Pomey, F. Goratel, and L. Abel, Determination des Cartraintes Residuelles dans
les Pieces Einentees, Publication Scientifiques et Techniques du Ministere de lair,
1950, p 263
S.S. Chiang, D.B. Marshall, and A.G.
Evans, The Response of Solids to Elastic/
Plastic Indentations I: Stresses and Residual Stresses, J. Appl. Phys., Vol 53 (No.
1), 1982, p 291311
C.O. Ruud, X-Ray Analysis and Advances
in Portable Field Instrumentation, J. Met.,
Vol 31 (No. 6), June 1979, p 1015
H.P. Klug and L.E. Alexander, X-Ray Diffraction Procedures, 2nd ed., John Wiley
& Sons, 1974, p 768770
C.S. Barrett and T.B. Massalski, Structure
of Metals, 3rd ed., McGraw-Hill, 1966, p
474476
C.O. Ruud and G.D. Farmer, Residual
Stress Measurement by X-Rays: Errors,
Limitations, and Applications, Nondestructive Evaluation of Materials, J.J.
Burke and V. Weiss, Ed., Plenum Press,
1979, p 101116
R.N. Pangborn, S. Weissman, and I.R.
Kramer, Dislocation Distribution and Prediction of Fatigue Damage, Metall. Trans.,
Vol 12A (No. 1), 1981, p 109120
D.A. Bolstad, R.A. Davis, W.E. Quist, and
E.C. Roberts, Measuring Stress in Steel
Parts by X-Ray Diffraction, Met. Prog.,
1963, p 8892
D.S. Kurtz, P.R. Moran, K.J. Kozaczek,
and M. Brauss, Apparatus for Rapid Psi
Squared Stress Measurement and Its Applications to Titanium Alloy Jet Engine
Fan Blades, Fifth Int. Conf. Residual
744749
B.B. He and K.L. Smith, Strain and Stress
Measurements with Two-Dimensional Detector, Adv. X-Ray Anal., Vol 41, 1977
M.R. James and J.B. Cohen, The Application of a Position-Sensitive X-Ray Detector to the Measurements of Residual
Stresses, Adv. X-Ray Anal., Vol 19, Gould,
Barnett, Newkirk, and Ruud, Ed., 1976, p
695708
H. Zantopulou and C.F. Jatczak, Systematic Errors in Residual Stress Measurement
Due to Specimen Geometry, Adv. X-Ray
Anal., Vol 14, Plenum Press, 1970
C.O. Ruud, D.J. Snoha, and D.P. Ivkovich,
Experimental Methods for Determination
of Precision and Estimation of Accuracy in
85.
86.
87.
88.
89.
90.
91.
92.
93.
94.
95.
96.
97.
98.
99.
XRD Residual Stress Measurement, Adv.

X-Ray Anal., Vol 30, 1987, p 511522
J.H. Root, C.E. Coleman, J.W. Bowden,
and M. Hayashi, Residual Stresses in Steel
and Zirconium Weldments, J. Press. Vessel Technol., Vol 119, 1997, p 137141
M. Hayaski, S. Ohkido, N. Minakawa, and
Y. Murii, Residual Stress Distribution
Measurement in Plastically Bent Carbon
Steel by Neutron Diffraction, Fifth Int.
J.H. Root, R.R. Hosbaus, and T.M.
Holden, Neutron Diffraction Measurements of Residual Stresses Near a Pin Hole
in a Solid-Fuel Booster Rocket Casing,
Practical Applications of Residual Stress
Technology, C.O. Ruud, Ed., ASM International, 1991, p 8393
G.A. Alers, Ultrasonic MethodsOverview, Proc. of a Workshop on Nondestructive Evaluation of Residual Stress,
NTIAC-762, 1975, p 155161
D.I. Crecraft, Ultrasonic Measurement of
Stresses, Ultrasonics, 1968, p 117121
G. Bradfield and H. Pursey, Philos. Mag.,
Vol 44, 1953, p 295
H. Pursey and H.L. Cox, Philos. Mag., Vol
45, 1954, p 295
W.J. McGonagle and S.S. Yun, Measurement of Surface-Residual Stress by Nondestructive Methods, Proc. Fifth Int. Conf.
Nondestructive Testing (Montreal, Canada), May 2125 1967, p 159164
A.J. Boland et al., Development of Ultrasonic Tonography for Residual Stress
Mapping, EPRI RP 5042, Final Report,
Electric Power Research Institute, 1980
M.R. James and O. Buck, Quantitative
Nondestructive Measurement of Residual
Stresses, Crit. Rev. Solid State Mater. Sci.,
Aug 1980, p 61105
E.P. Papadakis, Ultrasonic Attenuation
and Velocity in Three Transformation
Products of Steel, J. Appl. Phys., Vol 35
(No. 5), 1964, p 14741482
A. Moro, C. Farina, and F. Rossi, Measurement of Ultrasonic Wave Velocity of
Steel for Various Structures and Degrees
of Cold-Working, NDT Int., Aug 1980, p
169175
B.R. Tittman and R.B. Thompson, Measurement of Physical Property Gradients
with Elastic Surface Wave Dispersion,
Proc. Ninth Symp. on NDE, Southwest Research Institute, 1980, p 2028
K. Salama and C.K. Ling, The Effect of
Stress on The Temperature Dependence of
Ultrasound Velocity, J. Appl. Phys., Vol
51 (No. 3), March 1980, p 15051509
K. Salama and G.A. Alers, Third-Order
Elastic Constants of Copper at Low Temperature, Phys. Rev., Vol 161 (No. 3), Sept
1967, p 673680
100. R.E. Schramm, J. Szelazek, and A. Van

Clark, Jr., Ultrasonic Measurement of Residual Stress in the Rims of Inductively
Heated Railroad Wheels, Mater. Eval.,
Aug 1996, p 929933
101. H. Fukuaka, H. Tada, K. Hirakawa, H.
Sakawoto, and Y. Toyo, Acoustoelastic
Measurements of Residual Stresses in the
Rim of Railroad Wheels, Wave Propagation in Inhomogenous Media and Ultrasonic Nondestructive Evaluation, Vol 6,
G.C. Johnson, Ed., ASME, 1984, p 185
193
102. E.R. Schneider, R. Harzer, D. Bruche, and
H. Frotscher, Reliability Assurance of
Railroad Wheels by Ultrasonic Stress
Analysis, Proc. Third European Conf. on
Residual Stress Analysis (Frankfurt, Germany), 46 Nov 1992
103. D.M. Egle and D.E. Bray, Ultrasonic Measurement of Longitudinal Rail Stresses,
Mater. Eval., Vol 378 (No. 4), March
1979, p 4146
104. D.E. Bray and T. Leon-Salamanca, ZeroForce Travel-Time Parameters of Ultrasonic Head-Waves in Railroad Rail, Mater. Eval., Vol 43 (No. 7), June 1985, p
854858
105. C.G. Gardner, Barkhausen Noise Analysis,
Proc. Workshop on Nondestructive Evaluation of Residual Stress, NTIAC-762,
Nondestructive Testing Information Analysis Center, 1975, p 211217
106. R. Chait, Residual Stress Pattern in a High
Hardness Laminar Composite Steel Weldment, Proc. Workshop on Nondestructive
Evaluation of Residual Stress, NTIAC-76
2, Nondestructive Testing Information
Analysis Center, 1975, p 237245
107. O. Sundstrom and K. Torronen, The Use
of Barkhausen Noise Analysis in Nondestructive Testing, Mater. Eval., Feb 1979,
p 5156
108. J.R. Barton, F.N. Kusenberger, R.E. Beissner, and G.A. Matskanin, Advanced
Quantitative Magnetic Nondestructive
Evaluation Methods, Theory and Experiment, Nondestructive Evaluation of Materials, J.J. Burke and V. Weiss, Ed., Plenum
Press, 1979, p 461463
109. K. Ono, M. Shibata, and M.M. Kwan, Determination of Residual Stress by Magnetomechanical Acoustic Emission, UCLA
Department of Materials, ONR Tech. Rep.
No. 8001, Conference on Residual Stress
for Designers and Engineers, ASM International, April 1980
110. R.R. King and V.D. Smith, Residual
Stress Measurements in Structural Steels,
Final Report SWRI Project 154600, Contract DOT-FH-119133, Federal Highway
Administration, April 1981
111. H.D. Allison and R.W. Hendricks, Correlation of Barkhausen Noise Signal and XRay Residual Stress Determinations in
Grinding-Burned 52100 Steel, Fifth Int.

DOI: 10.1361/hrsd2002p118

Stress Determination in Coatings*

COATINGS AND THIN FILMS can be produced by a large variety of deposition techniques. Typical processes are physical vapor
deposition (PVD), chemical vapor deposition
(CVD), electroplating, electroless deposition,
anodizing, thermal growth, and thermal spraying. Since the early 1980s, considerable progress
has been made in improving deposition processes for a wide range of high-technology applications. Consequently, many new ceramic
coatings and films have been introduced in various industries. Typical examples are metal-oxide
semiconductors for microelectronics; titanium
nitride, titanium carbide, aluminum oxide, and
silicon nitride for machining tools; and thermal
sprayed tungsten carbide/cobalt, M-chromiumaluminum-yttrium (where M stands for iron, cobalt, or nickel), and yttrium oxide/partially stabilized zirconia coatings for aerospace
applications.
Residual stresses, which are internal and
therefore locked in, are contained in materials
that are produced by nearly every mechanical,
chemical, and thermal process, either alone or in
combination. As a result, most coatings are in a
state of internal stress, including metallics and
ceramics. The stress can be either compressive
or tensile. It is generally recognized that compressive stresses in coatings are more favorable
than tensile stresses because they increase resistance to fatigue failure. However, extremely high
compressive stresses may cause either coating
separation from the base metal or intracoating
spallation. Generally, if a tensile stress causes
strain that exceeds the elastic limit of the coating,
then it will cause cracking in the coating perpendicular to the direction of the stress. Therefore,
understanding the formation of residual stress in
the coating is important to prevent the coating
from peeling or cracking during service. Furthermore, residual stresses have significant influences on the mechanical and physical properties
of the coatings, particularly electrical resistivity,
optical reflectance, fatigue, and corrosion.
There are three types of residual stresses:
Macrostresses, which are nearly homoge-
neous over macroscopic areas of the material

Microstresses, which are nearly homogeneous over microscopic areas, such as one
grain or subgrain
Inhomogeneous microstresses, which are in-
homogeneous even on a microscopic level

Residual macrostresses are the ones of most interest in engineering practice because they can
substantially affect component service performance. Both residual and inhomogeneous microstresses are of more interest in materials science.
This article intends to provide a useful guide
for measuring residual macrostress on a coating.
The most commonly used measurement methods
are mechanical deflection, x-ray diffraction, and
hole-drilling strain gage. After a discussion on
the origins of residual stress, the fundamental
principles, as well as examples of practical measurements, are described for each method.
Origins of Residual Stress

Residual macrostress in a coating combines
the intrinsic stress and the thermal stress acting
in the coating plane parallel to the coating/substrate interface:
rt ri rth
(Eq 1)
where rt is the total macrostress, and ri and rth

are intrinsic stress and thermal stress, respectively. Intrinsic stress results from the growth
processes, depending primarily on deposition
parameters, whereas thermal stress arises from a
mismatch in coefficients of thermal expansion
between the coating and the substrate.
Many phenomenological models have been
proposed to explain the occurrence of intrinsic
stresses by correlating them with a variety of
coating microstructure and process features. To
varying degrees, the intrinsic stress of a coating
is associated with these deposition conditions
and coating features:
Incorporation of residual gas atoms in the
coating
Grain size, microvoid, and dislocation density
in the coating
Combined effect of surface tension and
growth process at grain boundaries
Deposition temperature relative to the melt-
ing temperature of the coating material
Annealing and shrinkage of disordered ma-
terial buried behind the advancing surface of

a growing coating
Although many studies have described the intrinsic stresses, information on the corresponding structural details is limited. It seems unlikely
that one can formulate a generalized model of
intrinsic stress for various coating materials and
deposition processes.
Any coating that is prepared at elevated temperatures (T2) and then cooled to room temperature (stress measurement temperature, T1) will be
thermally stressed because of the difference in
the coefficients of thermal expansion between
the coating and the substrate. Assuming no deformation of the substrate, the magnitude of the
thermal stress in the coating is:
rth (c s)(T2 T1)Ec /(1 mc)
(Eq 2)
where c and s are the thermal expansion coefficients for the coating and the substrate, respectively, and Ec and mc are the Youngs modulus and Poissons ratio of the coating,
respectively. A coating deposited at an elevated
temperature exhibits compressive stress if s
c, but tensile stress if s c. In the case of s
c, the substrate shrinks more than the coating
does during cooling from the deposition temperature and compresses the coating to maintain
dimensional compatibility.
In some cases, thermal stress is the primary
residual stress of the coating. For example, a titanium nitride coating can be deposited on a cemented carbide substrate (tungsten carbide-1
wt% Ta-10 wt% Co) via a CVD process at 1000
C (1830 F). With the values of s 5
106 /K, c 9.54 106 /K, ETiN 411 GPa
(60 106 psi), and mTiN 0.24, the stress in
the coating, as calculated from Eq 2, is 2.39 GPa
(0.35 106 psi) in tension at 25 C (77 F).
Energetic particle bombardment during coat-
Deflection Method
ing growth
Lattice misfit between the substrate and the
growing coating
The deflection method is the most widely used

technique for determining the residual stress in
*This article is adapted from Stress Determination for Coatings by J. Albert Sue and Gary S. Schajer in Surface Engineering, Volume 5, ASM Handbook, ASM International, 1994, p 647
653.
Stress Determination in Coatings / 119

a coating. In terms of basic principles, it involves
measuring the amount of bending in a strip that
is due to the deposition of the coating. A formula
for calculating the residual stress in an electrodeposited coating was derived first by Stoney
(Ref 1) in 1909. Subsequent researchers have derived more complex formulas to improve the accuracy of the stress evaluation. Comparative
studies on those stress-evaluation formulas are
reviewed and discussed elsewhere (Ref 24).
Nearly all formulas are variants of Stoneys formula.
Consider a coating deposited on one side of a
strip substrate. Both coating and substrate are assumed to be homogeneous. A mismatched force
at the coating/substrate interface results in residual stress in the coating, which bends the strip
either upward (concave) or downward (convex),
depending on whether the stress is tensile or
compressive. For overall force and moment
equilibrium of the coating/substrate composite,
it can be shown that the residual stress in the
coating is (Ref 5):
rc (Es d 2s )/[6(1 ms)Rdc](ds dc)
rc {Es d s2/[6(1 ms)Rdc]}

{(1 CH3)/(1 H)}
(Eq 4)
where
C [Ec(1 ms)]/[Es(1 mc)]
H ds /dc
R L2 /8f L2 /2d
(Eq 5)
where E is Youngs modulus; m is Poissons ratio;

d is thickness, with the subscript c denoting coating
and s denoting substrate; R is the radius of curvature of the bent strip, L is the length of the strip, f
is the deflection from the free end of the strip, and
d is the deflection at the center of the strip
(Fig. 1).
(Eq 3)
or, in general:
Although Eq 3 is used most often in practice

and represents a generalized Stoneys formula
for a planar state of stress, it tends to overestimate the value of measured stress. However, it
does not require knowledge of the elastic properties of the coating. Equation 4 provides a much
better approximation than Eq 3, but it does require knowledge of elastic properties. Equation
4 differs from Eq 3 in terms of a correction factor
[(1 CH3)/(1 H)]. Table 1 compares
stresses as a function of ds /dc, calculated from
both Eq 3 and 4. In the calculation, it was assumed that Ec 400 GPa (60 106 psi), Es
200 GPa (30 106 psi), and mc ms. It is clear
that both equations are in good agreement when
ds /dc 50. By choosing a large value of ds /dc,
the error can be minimized if Eq 3 is used.
Cantilever Beams. To measure the deflection
of cantilever beams, various techniques have
been developed (Ref 7), including optical, capacitance, mechanical, electromechanical, interferometric, and electromechanical or magnetic
restoration. Figure 1(a) shows the setup for the
deflection measurement when an optical system
is used. In this example, a titanium nitride coating was deposited on a quartz beam via a PVD
process, and the average residual stress in the
coating was determined by measuring the
amount of deflection at the free end, f, of the bent
beam, according to the relation:
rc [4Es d s2 f ]/[3(1 rs)L2 dc]
(Eq 6)
With the values of Es 71.7 GPa (10.4 106

psi), ms 0.16, ds 3 mm (0.12 in.), dc 3
106 m (1.2 104 in.), L 50 mm (2
in.), and f 2 106 m (8.0 105 in.), the
stress in the coating is determined to be 273 MPa
(40 ksi) in compression.
Disks. The average stress in the coating on a
disk substrate can be determined from the
amount of deflection, d, at the center of the disk
caused by the deposition of a coating on one
side. This value can be measured optically by
interferometry or microstylus profilometry. Figure 1(b) shows a typical interferometry apparatus
setup. Either technique measures deflection at
the same position, across a diameter of the disk,
both before and after coating deposition. The
stress in the coating is then calculated:
rc {Es d s2[(Ddx Ddy)/2]}/[3(1 ms)r2dc]
Fig. 1
Stress measurement techniques. (a) Bending of

cantilever beam. (b) Disk deflection. Source:
(Eq 7)
Ref 6
Table 1 Comparison of average residual

stress calculated using Eq 3 and 4
Calculated stress
Eq 3
Eq 4
ds /dc
MPa
ksi
MPa
ksi
200
100
50
20
10
1143
571
286
114
57
165.7
82.8
41.5
16.5
8.3
1137
565
280
109
52
164.9
81.9
40.6
15.8
7.5
Typical patterns of interference fringes. (a) rx

ry. (b) rx ry, where both components are in
compression or tension. (c) rx ry, where one component
is in compression and the other is in tension. Source:
Ref 8
Fig. 2
where Ddx and Ddy are deflection changes measured before and after coating deposition, and r
is the radius of the disk. Whether the stress is
tensile or compressive, it is determined by the
curvature of the disk before and after coating
deposition, as determined by a depth microscope. When the change in curvature is upward
(concave), viewed from the coating side, the
stress in the coating is tensile. The opposite
change in curvature indicates a compressive
stress in the coating.
Typical interference fringe patterns are shown
in Fig. 2. Figures 2(a) and (b) correspond to

equal and unequal principal stresses of the same
sign, whereas Fig. 2(c) corresponds to a biaxial
stress state with principal stresses of opposite
signs. Using Fig. 2(a) as an example, the measured deflection, d, is equal to the wavelength of
monochromatic light multiplied by the number
of light fringes (d 632.8 nm 17).
Figure 3 shows typical microstylus traces on
the titanium nitride coated surface of an AISI
304 stainless steel disk, both before and after
coating deposition. The curvature of the coated
surface was downward (convex), viewed from
the coated side. With the values of Es 193
GPa (28 106 psi), ms 0.28, ds 4.74 mm
(0.186 in.), r 11.94 mm (0.4700 in.), dc 20
lm (800 lin.), and the deflection Dd 6.3
106 m (2.5 104 in.), the stress in the titanium nitride coating is 4.44 GPa (0.644 106
psi) in compression. The measurement errors
from both the optical and the microstylus trace
setups are within one-half light band and 0.1
lm (4 lin.), respectively.
Practical Considerations. First, either strip
or disk-shape substrate specimens should be parallel within 0.02 mm (0.001 in.) over their length
or diameter. Specimen edges should be free from
visible flaws and chips. In addition, the surfaces
on which the deflection measurements (interferometer or profilometer) will be performed should
be ground and lapped to a finish better than 0.08
lm (3 lin.) Ra.
Second, the dimensions of a strip of length L,
width w, and substrate thickness ds should obey
L 10w 10 ds. The substrate thickness of a
strip or disk depends on the coating thickness to
be deposited. The thickness ratio of the substrate
to the coating, ds /dc, should be greater than 50
to ensure the accuracy of the stress calculation
from Eq 3.
Third, substrate specimens should be stress relieved before coating deposition. They should be
placed between two stainless steel surface plates
under at least an 8.8 kPa (1.3 psi) normal load,
at an annealing temperature, in a vacuum furnace
for at least 1 h. The annealing temperature is

dependent on the substrate material.
Fourth, for coatings with an inherently smooth
surface, such as those produced by PVD and
CVD, the deflection can be determined either by
interference fringe or microstylus profilometer
measurements on the coated surface. For a coating with a relatively rough surface, such as those
produced by thermal spraying and electroplating, the measurements can be made on the surface that is opposite the coated surface.
Significance and Use. The deflection measurement method is recommended for determining the average stress in the cross section of a
coating with a thickness ranging from several
hundred angstroms to several hundred micrometers. Typically, the elastic constants of a thin
coating are much different from those of a bulk
material. Equation 3 provides a means for stress
measurement on a thin coating without any
knowledge of its elastic constants. The measurement normally applies only to a test sample. The
disk deflection method is particularly useful for
the direct inspection of silicon wafers used in
solar cells or integrated circuits.
X-Ray Diffraction Method

Basic Principles. Stress measurement, using
the x-ray diffraction method, is based on the
change in the interplanar spacing (strain) close
to the surface of the specimen material. The details of the theory and interpretation of residual
stress measurements are well described in the article X-Ray Diffraction Residual Stress Techniques in Volume 10 of the ASM Handbook, as
well as in Ref 9 to 11. Consider an isotropic
material with a lattice parameter d hkl
for un0
stressed material in the sample plane normal.
The strain in a direction inclined by an angle y
to the surface normal of the coating and the stress
acting in the surface plane of the coating at an
angle u with the principal axis of the specimen
are related by:
hkl
hkl
hkl
ehkl
uw (d uw d 0 )/d 0
[(1 m hkl)/E hkl]

(rx cos2 u sxy sin2 u ry sin2 u rz)
sin2 w [(1 m hkl)/E hkl]rz (m hkl/E hkl)
(rx ry rz) [(1 m hkl)/E hkl]
(sxz cos u syz sin u) sin w)
2
Typical microstylus trace on titanium nitridecoated AISI 304 stainless steel disk before and
after coating deposition
Fig. 3
coating in various stress states can be determined

using Eq 8.
Biaxial Stress. At a free plane, the out-ofplane stress components rz, sxz, and syz are all
zero, at a free surface. Because the penetration
depth of x-rays is very small, the resulting measurements refer specifically to near-surface material. Plane stress conditions, therefore, often
apply to x-ray measurements, and Eq 8 is simplified to:
hkl
hkl hkl
hkl
hkl
ehkl
uw (d uw d 0 /d 0 [(1 m )/E ]
ru sin2 w (m hkl/E hkl)(rx ry) (Eq 9)
where ru rx cos2 u ry sin2 u is the macrostress in the coating parallel to its surface at
an angle u with the principal axis of the sample.
For a biaxial stress state, rx ry ru, and
at w 0:
hkl
hkl
hkl
hkl
(d hkl
u,w0 d 0 )/d 0 2m ru/E
(Eq 10)
From Eq 9 and 10:

hkl
hkl
hkl
hkl
2
d hkl
uw d u,w0 rud 0 [(1 m )/E ] sin w
(Eq 11)
In practice, high-angle diffraction peaks of an

(hkl) reflection are obtained from w to w at
a given angle u. Lorentz polarization, absorption, and background corrections are applied to
the diffraction peak profile. The peak positions
are determined by profile fitting or other methods
and are subsequently converted to interplanar
spacing d hkl
uw for stress analysis.
2
In the linear plot of d hkl
uw versus sin w, the
intercept is I d hkl
,
and
the
slope
is M
u,w0
hkl
hkl
rud hkl
]. The stress of the coating
0 [(1 m )/E
can then be determined by:
ru M/{d 0hkl[(1 m hkl)/E hkl]}
(Eq 12)
Triaxial Stress State without Shear Stress. For

a material in a three-dimensional (triaxial) stress
state without shear stress, but with the stress
component rz having a finite value within the xray penetration volume, Eq 8 becomes (Ref 9,
11, 12):
hkl
hkl
hkl
ehkl
uw (d uw d 0 )/d 0
[(1 m hkl)/E hkl](rx cos2 u ry sin2 u rz)

(Eq 8)
where h, k, and l are the indices of the Bragg

hkl
reflection; ehkl
uw and d uw are the strain and interplanar spacing of (hkl) in the direction of (u, w),
respectively; m hkl and Ehkl are Poissons ratio and
Youngs modulus in (hkl) in the coating; rx, ry,
and rz are normal stresses; sxy, syz, and sxz are
shear stresses; rx is the normal stress acting in
the x direction on a plane perpendicular to the x
axis; and syz is the shear stress on a plane normal
to the y axis (the first subscript) in the z direction
(the second subscript). Residual stresses of the
(rx ry rz)
(Eq 13)
Two data sets, u 0 and u 90, are needed

to obtain rx, ry, and rz. The slopes and intercepts
2
of a linear function of d hkl
uw versus sin w at u
0 and u 90 are given by:
hkl
hkl
Mu0 d hkl
0 [(1 m )/E ](rx rz)
Mu90 d 0hkl[(1 m hkl)/E hkl](ry rz)

hkl
hkl
hkl
hkl
I d hkl
0 {[(1 m )/E ]rz (m /E )
(rx ry rz)}
(Eq 14)

Stresses rx, ry, and rz can be determined from
the sum of the slopes and the intercept in Eq 14.
Triaxial Stress State with Shear Stress. A
coating with a three-dimensional (triaxial) stress
state, including shear stresses, is fully described
by Eq 8. The shear stresses, sxz and syz, have a
2
sin2 w dependence. The d hkl
uw versus sin w distribution is no longer linear and has two branches
of an ellipse for w 0 and w 0. This effect
is termed w splitting, which is an indication of
the presence of shear stress.
To obtain these stress-tensor components,
three data sets (u 0, u 45, and u 90)
are obtained for both w and w. The average
strain a1 and the deviation a2 from the strains of
w splitting are determined to be (Ref 9, 11,
12):
hkl
a1 (euw
ehkl
uw)/2
hkl
hkl
[(d uw
d uw
)/2d 0hkl] 1
[(1 m hkl)/E hkl]

(rx cos2 u sxy sin2 u ry sin2 u rz)
(rx ry rz)
(Eq 15)
hkl
a2 (euw
ehkl
uw)/2
hkl
hkl
(d uw
d uw
)/2d 0hkl
[(1 m )/E ](sxz cosu syz sinu) sin w

(Eq 16)
hkl
hkl
The stress-tensor components can be calculated from the slopes of linear plots of a1 versus
sin2 w and a2 versus sin2 w. For a1 versus sin2
w, (rx rz) is obtained at u 0, (ry rz)
at u 90, and sxy at u 45, whereas rz is
evaluated from the intercept if d hkl
0 is known.
Similarly, sxz and syz are obtained when u 0
and u 90, respectively, from the slope of a2
versus sin2 w.
Stress Measurement. Modern diffractometers are fully automated and equipped with computer software for performing numerically intensive analyses. Diffractometers are capable of
measuring residual stresses efficiently and economically. Typically, the measurement can be
completed in several hours. The following examples illustrate some stress measurements using Cu K radiation on cathodic arc PVD titanium nitride coatings on a substrate with various
stress states.
Biaxial. An approximately 10 lm (400 lin.)
thick coating of highly (111) oriented titanium
nitride was deposited on AISI 304 stainless steel
at 500 C (930 F). The x-ray diffraction sin2 w
technique was applied to determine the residual
stress in the (333)/(511) reflection of the coating.
Figure 4 shows the linear distribution of d hkl
uw versus sin2 w, indicating typical biaxial stress in the
coating. As shown, the slope M 1.1933
103 nm (4.698 1011 in.) and the intercept
I d hkl
uw0 0.082123 nm (0.003233 lin.).
Youngs modulus and Poissons ratio for titanium nitride in (333)/(511) are 364 GPa (52.8
106 psi) and 0.245, respectively (Ref 13). The
calculated residual stress for titanium coating is
4248 MPa (615 ksi) in compression. For
highly (111) oriented titanium nitride film, (511)
contribution is negligible. This is addressed in
Ref 13. The (422) reflection has 2h 130,
which is not desirable.
Triaxial without Shear Stress. To exemplify a
triaxial stress distribution, without shear stresses,
triaxial stress analysis was applied to a titanium
nitride coating deposited on Inconel 718 sub2
strate at 550 C (1020 F). The d hkl
uw versus sin
w distributions were obtained from (333)/(511)
for u 0 and u 90 (Fig. 5). The slopes
Mu0 and Mu90 are 0.96782 and 1.01415
103 nm (3.8103 and 3.9927 1011
in.), respectively, and the intercept I is 0.082083
nm (0.003231 lin.). The strain-free interplanar
spacing for (333)/(511) is 0.08160 nm
(0.003213 lin.). Based on Eq 14, the stress tensor (rij) from this analysis, in units of MPa, is:
rx sxy sxz
rij syx ry syz
szx szy rz
2653 151
0
0
0
2819 150
0
0
0
815 140
The result shows that the planar stresses are

equal biaxial within experimental error and that
the stress perpendicular to the coating surface is
in tension.
Triaxial with Shear Stress. A triaxial stress
distribution, including shear stresses sxy, syz, and
sxz, was studied in a cathodic arc PVD titanium
nitride coating on AM-355 stainless steel. The
2
d hkl
uw versus sin w distributions were obtained
from (333)/(511) at u 0, u 45, and u
90 using Cu K radiation. Figure 6 shows the
2
typical ellipse distribution of d hkl
uw versus sin w
at u 0. Based on analysis discussed at the
beginning of this section, the stress tensor in the
titanium nitride coating, in units of MPa, is:
rij
2699 148
1 181
110 32
1 181
2776 148 34 34
110 32
34 34 833 148
Practical Considerations. First, a W-diffractometer is preferable for conducting stress measurements. The X- and W-diffractometers are defined on the basis of the sample axis for w tilt
perpendicular or parallel to the plane of the incident and detected x-ray beam, respectively. A
W-diffractometer gives symmetric irradiated areas in w and w tilted angles at a given (hkl)
reflection line and a greater range of sin2 w (from
0 to 0.95 for a W-diffractometer and 0 to 0.5 for
an X-diffractometer).
Second, the diffractometer should be mechanically aligned and calibrated using a stress-free
standard sample (NIST SRM 660 lanthanum
hexaboride powder or SRM 640 silicon powder)
to obtain the peak position of the (hkl) reflection
line within 0.01 at Bragg angles 2h in the
range of w 0 and w 0.
Third, an appropriate x-ray wavelength should
be selected to achieve the desired sampling volume (penetration depth) for a particular set of
(hkl) planes. The penetration depth, which is defined as the distance from the surface to the depth
with 63% or 1/e of the intensity of the reflection
line, is calculated for X- and W-diffractometers,
respectively:
sX (sin2 h sin2 w)/[2l(sinh cosw)]
sW (sinh cosw)/2l
Fig. 4
flection
hkl
Linear distribution of d uw
as a function of sin2 w
of titanium nitride coating from (333)/(511) re-
hkl
Linear distribution of d uw
as a function of sin2 w
of titanium nitride coating from (333)/(511) reflection at u 0 and u 90
Fig. 5
where l is the linear absorption coefficient,

which can be obtained from Ref 14 for various
materials and radiations.
Fourth, to achieve high accuracy in residual
stress, the measurement should only be carried
out on an (hkl) reflection line with Bragg angles
2h 130 and with sufficient intensity for peakposition determination.

Fifth, an appropriately sized collimator should
be selected. It should have an irradiated area
large enough to ensure that a statistically relevant number of grains or subgrains in the coating
are included in the measurement. The angular
resolution is essential in stress measurement. If
the spot mode of x-ray beam cannot sample a
sufficient number of grains, one should use another method.
Sixth, appropriate methods (Ref 9, 10, 15
17), such as modified Lorentzian, Gaussian,
Cauchy, Pearson VII, parabola, center of gravity,
gravity line, and cross correlation should be applied to determine the peak positions of the reflection line (hkl).
Seventh, a sufficient number of measurements
in the w and w directions should be conducted to obtain an accurately linear distribution
2
or w-splitting of d hkl
uw versus sin w. The significance of this distribution should be verified by
repeating measurements at different u angles. If
a nonlinear relation other than w-splitting is ob2
tained in the d hkl
uw versus sin w distribution, then
the coating being measured is inhomogeneous
and, therefore, the x-ray diffraction sin2 w
method is no longer applicable.
Eighth, the x-ray elastic constants of the coathkl
hkl
ing for an (hkl) plane, S hkl
and
1 m /E
hkl
hkl
S hkl
/2
(1
l
)/E
,
can
be
calculated
from
2
single-crystal compliance, according to an appropriate model (Ref 11, 12), or measured experimentally in uniaxial tension or bending tests
with a series of loads (Ref 13, 18, 19). X-ray
elastic constants and applied loads (ra) obey
these equations:
practice. For biaxial stress analysis, the lattice

spacing measured at w 0, d hkl
uw0, can be substituted for d hkl
0 . In this case, the contribution to
the total error is less than 0.1%. For a triaxial
stress analysis, the difference between d hkl

uw0
and d hkl
0 is included in the calculation. Consequently, a small error in d hkl
0 can lead to a relatively large error in the measured stress. To obtain acceptable stress results, d hkl
0 must be within
0.01% of its true value. The stress-free interplanar spacing d hkl
0 in the strain-free direction w*
is given by (Ref 20):
hkl
sin2 w* [S hkl
1 /(S 2 /2)]
{1 [(ry rz)/(rx rz)]

hkl
[3 (S hkl
2 /2S 1 )rz]/(rx rz)}; u 0
sin2 w* [S 1hkl/(S 2hkl/2)]

{1 [(ry rz)/(rx rz)]}
when |rx ry 2rz| 2|rz|
d hkl
0
hkl
d uw0
/[1 S hkl
1 (rx ry rz)
(S hkl
2 /2)rz]
Significance and Use. The advantage of the

x-ray diffraction method is its capability for
measuring residual stress directly and nondestructively on a product component. Portable diffractometers are commercially available and can
be used for on-site measurement. The method
can apply to a coating with a thickness ranging
from 0.5 to 350 lm (2 105 to 1.4 102
in.). However, it is difficult to measure residual
stresses in extremely thin coatings and, in some
cases, highly textured coatings. Furthermore, xray diffraction is inapplicable to amorphous
coatings, and a large scatter in stress measurement is often encountered in coatings with large
grain size.
hkl
hkl
S hkl
2 /2 (1 m )/E
hkl
2
(1/d hkl
0 )[/ra(d uw/sin w)]
Hole-Drilling Method
S 1hkl m hkl/E hkl [1/(2d 0hkl)](d hkl

uw0/2ra)
(a) Typical three-element strain-gage rosette. (b)
In-plane strain components caused by release of
residual stress through introduction of a hole. Source:
Ref 21
Fig. 7
Ninth, a stress-free interplanar spacing, d hkl

0 ,
for a coating may not be readily available in
Fig. 6
Basic Principles. The hole-drilling method

for measuring residual stresses involves drilling
a shallow hole in the test specimen to a depth
approximately equal to the hole diameter. Typ-
hkl
Typical ellipse distribution of d uw
as a function of sin2 w of titanium nitride coating from (333)/(511) reflection at varying angles. (a) u 0. (b) u 45. (c) u 90

ical hole diameters range from 0.8 to 5.0 mm
(0.0300.200 in.). The creation of the hole redistributes the stresses in the material surrounding the hole. A specially designed three-element
strain-gage rosette, such as that shown in Fig.
7(a), measures the associated partial strain relief.
The in-plane residual stresses that originally existed at the hole location can then be calculated
from the measured strain reliefs using the
method described in ASTM E 83792 (Ref 21).
The ASTM standard also gives details of practical drilling procedures.
The partial strain relief measured by one of
the three strain gages in the rosette in Fig. 7(b)
is related to the principal in-plane residual
stresses by:
er (rmax rmin)A
(rmax rmin)B cos
2
(Eq 17)
where rmax and rmin are maximum and minimum

principal residual stresses, and is the angle
from the gage axis to the maximum principal
stress direction. A and B are calibration constants, the values of which depend on the specimen material properties, the rosette geometry,
the hole diameter, and the hole depth. ASTM E
837-92 tabulates the calibration constants for the
standard rosette pattern shown in Fig. 7.
Equation 17 can be inverted to determine the
principal residual stresses from the measured
strain reliefs. The result is:
rmax, rmin [(e3 e1)/4A]
{[(e3 e1)2 (e3 e1 2e2)2]1 / 2 /4B}
b 12 arctan [(e3 e1 2e2)/(e3 e1)]
(Eq 18)
Table 2 Numerical values of dimensionless

calibration coefficients a and b
D0 /D
0.30
0.31
0.32
0.33
0.34
0.35
0.36
0.37
0.38
0.39
0.40
0.41
0.42
0.43
0.44
0.45
0.46
0.47
0.48
0.49
0.50
Through-thethickness hole(a)
a
b
0.089
0.095
0.101
0.108
0.114
0.121
0.128
0.135
0.143
0.150
0.158
0.166
0.174
0.183
0.191
0.200
0.209
0.218
0.228
0.237
0.247
0.278
0.295
0.312
0.329
0.347
0.364
0.382
0.400
0.418
0.436
0.454
0.472
0.490
0.508
0.526
0.544
0.562
0.579
0.596
0.613
0.629
Blind hole(b),
depth 0.4D
a
0.111
0.118
0.126
0.134
0.142
0.150
0.158
0.166
0.174
0.182
0.190
0.199
0.208
0.217
0.226
0.236
0.246
0.255
0.265
0.275
0.285
0.288
0.305
0.322
0.340
0.358
0.376
0.394
0.412
0.430
0.448
0.466
0.484
0.503
0.521
0.540
0.558
0.576
0.594
0.612
0.630
0.648
(a) In a thin sheet. (b) In a thick material. Source: ASTM E 83792
where b is the angle measured clockwise from

the location of gage 1 to the direction of rmax.
The above equations are valid for a homogeneous isotropic material that is wide, when compared with the hole diameter, and thick, when
compared with the hole depth. The equations
also apply to a through-the-thickness hole in a
material in the form of a thin sheet, provided that
the sheet thickness is uniform near the hole. The
associated calibration constants have slightly
different values than those of the thick-material
case. Table 2 (from ASTM E 837-92) lists the
dimensionless calibration coefficients for both
the thin-sheet and thick-material cases. A and B
can be determined from the dimensionless coefficients using:
A (1 m)a/2E
B b/2E
(Eq 19)
Residual stress measurements in coated materials create an additional complication because

the combination of coating and substrate is no
longer a homogeneous material. The calibration
coefficients provided in ASTM E 837-92 no
longer accurately apply. However, the above two
equations are still valid for coated materials, if
the coating has a uniform thickness. If the substrate is less than several times greater than the
hole depth, then it is also necessary for the substrate thickness to be uniform.
The calibration coefficients A and B for hole
drilling in a coated material differ from the standard values given in ASTM E 837-92. The actual
values depend on the elastic properties of the
coating and the substrate, the coating thickness,
and the hole diameter and depth. These coefficients can be determined by either experimental
calibrations (Ref 22) using known externally applied stresses or finite-element calculations (Ref
23).
Approximate values of A and B for thick coatings can be estimated from the values given in
ASTM E 837-92. For this purpose, a thick
coating is one that is at least 0.25 times the mean
radius of the strain-gage rosette. For the smallest
commercially available hole-drilling rosette, the
mean radius is about 1.25 mm (0.050 in.). Therefore, the minimum acceptable coating thickness
is about 0.3 mm (0.012 in.).
An approximate estimation of A and B for a
coated material is based on the observation that
the hole-drilling method is most sensitive to the
stresses closest to the specimen surface. Almost
all of the measured strain relief is due to the
stresses in the material within a depth of about
0.25 times the mean radius of the hole-drilling
rosette. Thus, a specimen coated to at least this
depth is likely to behave similarly to a homogeneous thick specimen consisting only of coating material. Thus, the A and B calibration coefficients for a thick coating are approximately
equal to the ASTM tabulated values for a homogeneous material with the elastic properties
of the coating.
The A and B coefficients for coatings that are
thinner than 0.25 times the mean radius of the
hole-drilling rosette will deviate significantly

from the ASTM tabulated values. The coefficients must be determined on an individual basis,
either by experimental or computational means.
The use of the hole-drilling method with such
thin coatings is not generally recommended
because the sensitivity of the resulting strain
measurements is rather low. As a result, small
absolute errors in the strain measurements can
cause large relative errors in the computed residual stresses.
Stress Measurement. The following example
illustrates a hole-drilling measurement on a detonation-gun type of tungsten carbide-cobalt
(WC-Co) coating deposited on an AISI 1018
steel substrate. The coating was approximately
0.75 mm (0.03 in.) thick and had a macroscopically homogeneous structure. The elastic properties of the coating were Ec 172 GPa (25
106 psi) and mc 0.3.
A 062-RE hole-drilling strain-gage rosette
(5.13 mm, or 0.202 in., strain-gage mean diameter) was attached to the coated specimen. A 2.44
mm (0.096 in.) diam hole was cut in the WC-Co
coating by abrasive-jet drilling using 27 lm
(1080 lin.) alumina particles. Drilling proceeded
in four approximately equal depth increments,
up to a final depth of 0.356 mm (0.014 in.). The
strain measurements listed in Table 3 were made
after each hole-depth increment. Using the A and
B calibration coefficients from ASTM E 837-92,
adjusted for the elastic properties of the coating
material, the principal residual stresses in the
coating were found to be 260 MPa (38 ksi)
and 286 MPa (41 ksi), respectively. As
might be anticipated, the residual stresses in the
coating are approximately isotropic.
Practical considerations for the use of the
hole-drilling method are:
A high-speed drilling technique using carbide
drills is recommended for producing a hole

in a ductile coating. Abrasive-jet drilling is
recommended for a brittle, hard coating (Ref
24).
The use of specially made hole-drilling
strain-gage rosettes is essential. The application of the strain-gage should follow the procedure recommended by the manufacturer.
A smooth coating surface less than 0.41 lm
Ra (16 lin. Ra) is desirable for secure straingage adhesion. An abrading or grinding process that does not induce significant residual
surface stress should be used for surface preparation.
The selection of an appropriately sized strain
gage should be based on coating thickness, as
well as on the depth and diameter of the hole
to be drilled. The diameter of the drilled hole,
D0, should be related to the diameter of the
gage circle, D, where 0.3 (D/D0) 0.5.
A depth microscope with a resolution better
than 12.7 lm (0.0005 in.) should be used to
measure the depth of the drilled hole at each
depth increment.
The center of the drilled hole should coincide
with the center of the strain-gage circle within

Table 3
Hole-drilling residual stress measurements on a detonation gun WC-Co coating

Average stress
Strain, le
Depth
rx
ry
sxy
rmax
rmin
smzx
mm
in.
e1
e2
e3
MPa
ksi
MPa
ksi
MPa
ksi
MPa
ksi
MPa
ksi
MPa
ksi
0.00
0.10
0.20
0.28
0.36
0.000
0.004
0.008
0.011
0.014
0
56
116
163
210
0
52
107
155
200
0
52
109
168
218
271
39.3
275
39.9
13
1.9
260
37.7
286
41.5
13
1.9
41
Note: Rosette type, EA-XX-062RE; Youngs modulus, 172.0 GPa (25 106 psi); Poissons ratio, 0.3; hole diameter, 2.440 mm (0.0961 in.); and rosette mean diameter, 5.13 mm (0.202 in.)
0.015 D0. A measurement microscope

should be used to align the drill holder or
abrasive-jet nozzle with the center of the rosette.
Precautions should be taken to ensure that the
walls of the drilled hole are square to the coating surface on which the rosette is cemented.
It is important to protect the strain gage from
abrasive-particle erosion or mechanical damage during the drilling operation.
Values for the Youngs modulus and Poissons ratio of the coating should be independently measured in order to determine the residual stress from strain relaxations.
Significance and Use. The hole-drilling
strain-gage method is a semidestructive technique for measuring residual stress on a coating
with a thickness of at least 0.1 mm (0.004 in.).
The method, which is quite versatile, can apply
to test samples as well as to actual components
with complex geometries. Furthermore, it can be
used for on-site measurements.
Method Comparison
The mechanical-deflection method is capable
of measuring the average stress throughout the
coating thickness, but requires the stress to be
uniform over large distances in the in-plane directions. In contrast, the x-ray diffraction and
hole-drilling methods can make a much more localized measurement in-plane, but they have a
significantly more limited depth capability. A
good agreement in stress measurements between
the deflection and x-ray diffraction methods has
been demonstrated (Ref 13). With the extrapo-
lation of blind-hole measurements to the near

surface, the stress measurement is in good agreement with that measured by x-ray diffraction
(Ref 25). A user can select the most suitable
method based on economics, environment, coating microstructures, and the geometry of the
component to be measured.
REFERENCES
1. G.G. Stoney, Proc. R. Soc. (London) A, Vol
82, 1909, p 172
2. C.N. Kouyumdjev, Surf. Technol., Vol 26,
1985, p 35
3. C.N. Kouyumdjev, Surf. Technol., Vol 26,
1985, p 45
4. G. Sotirova and S. Armyanov, Surf. Coat.
Technol., Vol 28, 1986, p 33
5. M. Ohring, in The Materials Science of Thin
Films, Academic Press, 1991, p 461
6. R.W. Hoffman, in Physics of Thin Films,
Vol 3, 1966, p 211
7. D.S. Campbell, Handbook of Thin Film
Technology, L.I. Maissel and R. Glang, Ed.,
McGraw-Hill, 1970, p 1221
8. R.E. Cuthrell, D.M. Mattox, C.R. Peeples,
P.L. Dreike, and K.P. Lamppa, J. Vac. Sci.
Technol., Vol 6A (No. 5), 1988, p 2914
9. V.M. Hauk, Adv. X-Ray Anal., Vol 27,
1984, p 81
10. Residual Stress Measurement by X-Ray Diffraction, J784a, 2nd ed., SAE Handbook,
Society of Automotive Engineers, 1971
11. I.C. Noyan and J.B. Cohen, Residual Stress
Measurement by Diffraction and Interpretation, B. Ilschner and N.J. Grant, Ed.,
Springer, 1987
12. H. Dolle, J. Appl. Crystallogr., Vol 12,

1979, p 489
13. J.A. Sue, Surf. Coat. Technol., Vol 54/55,
1992, p 154
14. C.H. Macgillavry and G.D. Rieck, Ed., International Tables for X-Ray Crystallography, The Kynoch Press, Birmingham, 1962
15. W. Parrish, T.C. Huang, and G.L. Ayers,
Trans. Am. Cryst. Assoc., Vol 12, 1976, p
55
16. A. Brown and J.W. Edmonde, Adv. X-Ray
Anal., Vol 23, 1980, p 361
17. A. Brown and S. Linde, Adv. X-Ray Anal.,
Vol 30, 1987, p 343
18. V. Hauk, International Conference on Residual Stresses, ICRS2, G. Beck, S. Denis,
and A. Simon, Ed., Elsevier, London, 1989,
p 292
19. H. Behnken and V. Hauk, International
Conference on Residual Stresses, ICRS2, G.
Beck, S. Denis, and A. Simon, Ed., Elsevier,
London, 1989, p 341
20. V.M. Hauk, Adv. X-Ray Anal., Vol 27,
1984, p 101
21. Determining Residual Stresses by the
Hole-Drilling Strain-Gage Method, E 83792, Annual Book of ASTM Standards,
ASTM
22. Measurements of Residual Stresses by
Hole-Drilling Strain Gage Method,
TN503-4, Measurements Group, Wendell,
NC, 1993
23. G.S. Schajer, J. Eng. Mater. Technol.
(Trans. ASME), Vol 103 (No. 2), 1981, p
157
24. M.T. Flaman and J.A. Herring, Exp. Tech.,
Vol 9 (No. 8), 1985, p 30
25. C.O. Ruud, P.S. DiMascio and J.J. Yavelak,
Exp. Mech., Vol 25 (No. 4), 1985, p 338

DOI: 10.1361/hrsd2002p125

Methods for Determination of

Inhomogeneous Residual Stress Fields
I.A. Razumovsky, M.V. Medvedev, and A.V. Fomin, Mechanical Engineering Research Institute of the Russian Academy of
Sciences, Moscow
AMONG NUMEROUS PROBLEMS encountered by mechanics of nonrigid solids, analysis of residual stress presents one of the most
challenging tasks because it involves the analysis of physical-mechanical processes and structural changes in the material under various mechanical and heat impacts. Although methods of
analyzing residual stress have been under development since the 1930s, this field still presents
quite a number of problems to be solved. Development of experimental techniques providing
for large amounts of strain field data suitable for
later mathematical processing gave impetus both
to the beginning of new and the development of
already available approaches to residual stress
analysis, this fact significantly expanding the
range of solvable applied problems. The objective of this chapter is to present results from new
developments in methodology and applied techniques for solving problems in the field of residual stress analysis.
Residual stresses can be subdivided provisionally into two classes: macrostresses (stresses of
the first type) and microstresses (stresses of the
second type). Microstresses are stresses that actually exist directly in the crystalline grain and
experience changes within the crystalline aggregate (grain) due to crystal inhomogeneity, different alignment of crystallographic planes, and
some other factors. Macrostresses can be regarded as averaged microstresses on some finite
base. It must be noted that to determine macrostresses by microstress averaging is possible
only if microstress distributions comply with
consistent patterns of statistics.
Most methods for determining residual stress
aim at obtaining values of macrostresses. This is
because the majority of modern approaches to
stress-state evaluation are based on continual
models of material whose deformation and destruction are described on the basis of continua
mechanics. It might be well also to point out
considerable difficulties in experimentally determining the microstress due to minuteness of
crystal grains characterized by arbitrary irregular
form and generally being inhomogeneous.
At the same time, for example, the x-ray study
of residual macrostress (the so-called sin2 w
method) is based on measuring changes in distances between atomic planes of metal grain lattice (i.e., microstrains).
With regard to problems considered in domains of strength evaluation, reliability, and optimal structure design, different methods and respective experimental techniques have been
developed to solve typical problems of residual
stress analysis for most practical cases.
Efficiency of one or another method is determined by the studied object geometry, type of
stress state, nature of residual stress distribution,
and materials properties. Methodology of research, including distinctive features of experimental techniques and facilities are given in
Ref 8.
Classical methods of residual stress research
used nowadays, both destructive and nondestructive, provide no opportunity to reliably determine residual stresses in terms of their high
gradients when a pronounced change of stress
occurs on the gage length in the order of 1 mm
(0.04 in.). At the same time, determining residual
stress on this gage length is necessary when analyzing strength, integrity, and durability of
structures with welded joints, welded deposits,
and other inhomogeneities.
Special methods that provide for analysis of
high-gradient (up to breakage) fields of residual
stress in parts made of homogeneous and bimetallic materials have been developed to analyze
such problems.
This chapter is dedicated to methodology developments that accompany experimental determination of significantly inhomogeneous fields
of residual stress. It presents a systematic outline
of modern methods for determining residual
stress fields characterized by significant inhomogeneity, including theoretical foundations of
respective methods as well as aspects of their
application. These methods are based on mathematical processing of extensive experimental
data obtained when notches of various types are
made in the part under investigation. Here, the
procedure for obtaining stresses from experimental data is based on the fact that residual
stress determination is posed as an inverse problem from the mechanics of nonrigid solids.
Also presented are problems encountered in

obtaining experimental data necessary for determining residual stress, including requirements to
the nature and scope of acquisition procedures
as well as descriptions of special equipment used
in solving practical tasks. Examples of applying
these techniques to the study of high-gradient
residual stress fields are given.
Study of Residual Stress as Inverse

Problem of Experimental Mechanics
The source of residual stress appearance lies
in the initial strain, e 0ij, that fails to meet the equations of strain compatibility. This strain is connected by the Hookes law to the elastic strain,
eij e 0ij, emerging in the body (where eij are full
strains corresponding to the general solution of
the problem with external loads missing). Residual stress is adjusted for equilibrium inside the
body after eliminating the impacts that caused it.
In the theoretical study of residual stress, either
the initial incompatible strain, e 0ij, or the sequence of changes in force and temperature conditions of loading that leads to initial strains,
e 0ij, are assumed to be known.
The approaches to the study of residual stress
presented in this chapter belong to destructive
methods. Yet, in contrast to the known investigation techniques that involve notching, these
methods do not assume any a priori correlation
between distribution of residual stress to be
found, on the one hand, and the stress-strain state
of the investigated part after notching, on the
other, the latter being determined experimentally.
Following are experimental design methods of
determining the continuous distribution of residual stress in flat parts of arbitrary shape by measured parameters of stress-strain state close to
notches made. Such parameters can be represented by stress-strain tensor components or
those of displacement vector, as well as by some
of their functionals (e.g., the first invariant of
stress tensor) that retain the attribute of linear
dependence on residual stress to be found.

The problem of determining residual stress
from measured parameters of stress state in the
vicinity of a created notch belongs, by the manner of posing, to the class of inverse problems.
The observed picture of the stressed state is a
response of the residual stress that is relieved
along the notch line and must be restituted by
this response. Inverse problems are distinguished by being ill defined in most natural (with
regard to practice) spaces (C and L2), which incorporate initial data and target distributions of
residual stress (Ref 1). These problems are ill
defined because close distributions of observed
stressed-state parameters (close within measurement error limits) can be matched against significantly different distributions of residual
stress along the notch line.
Consistent procedures for determining residual stress can be based on regulating algorithms
for various integral-solving equations. Consider
two variants of obtaining consistent derivation
procedures with differently posed problems.
In the first variant, the notch is gradually expanded and measurements are made for a certain
parameter of stress state at some specific locus
that follows the shift of a notch tip. Thus, the
distribution of the observed parameter is determined along the notch line as the part geometry
changes. The corresponding inverse problem is
reduced to the Volterra integral equation.
In the second variant, the created notch corresponds to its final length, and the respective
overall effect along some line in the vicinity of
the notch is measured. The corresponding inverse problem is reduced to the Fredholm integral equation.
Method of Volterra Integral Equations. As
the notch is expanded, residual stress on its
edges is relieved, this stress playing the role of
edge load for a new section of part boundary
(Fig. 1a). Due to the superposition principle, the
relationship between the required residual stress,
ry (n), and the measured parameter of stress state
along the notching line, l(x), (Fig. 1b) may be
represented as the Volterra integral equation of
the first type:
x
B(x,n)r (n)dn l(x)

0
0nx L
(Eq 1)
where L is the maximum length of the notch. The

operator kernel, B(x,n), corresponds to values
l(x) at point x from d-impact at point n, while
the domain of its definition is a triangle (i.e., at
n x, B(x,n) 0 is obtained.
Apart from considering Eq 1, it is possible to
suggest its transformation on the basis of an additional conditionsmoothness of the changed
function l(x). Then, by way of differentiating,
Eq 1 is reduced to the Volterra equation of the
second type:
B(x,x)ry(x)
B(x,n)r
x
y(n)dn l(x)
(Eq 2)
When applying the method of the Volterra integral equation, the following important fact
should be considered. While preparing the experiment, account is taken of its conditions and
measuring tool sensitivity in choosing the value
of a step in expanding the notch and taking measurements. In this case, the discrete character of
obtained information is in no way connected to
the nature of measured distributions. From this
factor develops the task of subjecting the initial
data to preliminary processing and that of creating some continuous and differentiable analogues from discrete data and within error limits
comparable to experiment errors. Particular patterns for processing initial data can be designed
in various ways, including regularization methods.
Method of Fredholm Integral Equation of
the First Type. A line running close to the notch
tip is selected (Fig. 2). Due to the linearity of the
problem under consideration, the relationship
between the required residual stress ry(n) along
the notch with length L and the values of the
measured parameter l(n) on this line is:
L
H(s,n)ry(n)dn l(s) s S,n L
tional experiment to select measurement lines

and points with high information resolution.
As for Eq 2, its solution is a correctly solvable
problem, and the operator of this equation is a
regularizing operator for Eq 1. When using Eq
2, the accent in solving is on finding an efficient
algorithm for numerical differentiation of discrete experimental data.
Values l(x) and l(s) are known from the experiment in the presence of additive random disturbance whose value is dependent both on the
experimental method used and the specific conditions of measurements. The accuracy of measurements cannot be increased as far as it is provided by the numerical methods used to
determine kernels of integral operators. They are
constructed by multiple solving of boundary
problems for the notched part studied. As a rule,
errors of initial data are always significantly
y()
(Eq 3)
The kernel of integral operator H(s,n) is determined in the region S L and corresponds
to measured parameters l(n) on line S from d
impacts at points n L. From the point of methodology, reducing the problem to solving the
Fredholm equation of the first type is very important. First, conducting an experiment is relatively simple compared with the approach that
uses the Volterra equation because in this case,
all experimental data are obtained from the single length notch. Second, when solving Eq 3 numerically, creating a matrix analogue of an integral equation would require (under equal
discrete values) significantly fewer boundary
problems to be solved in comparison with the
first approach, which allows for the changing
boundary of the notch tip area.
Solutions of equations of the first type (Eq 1,
3) are unstable toward variations of initial data
(right side) and call for regularizing methods.
The indicated distinctions in experiment arrangement determine specific features in the
mathematical structure of these equations. Thus,
the Volterra equation (Eq 1), all other factors
being the same, is conditioned with regard to ry
to a greater degree (i.e., has wider information
possibilities than Eq 3). According to the regularization principle (Ref 1), information possibilities of considered computational/experimental procedures can be evaluated by the
maximally admissible degree of smoothness in
solving respective integral equations. Information capacities are largely determined by
smoothness of integral equation kernel. The
higher the kernel smoothness, the higher is the
smoothing (averaging) effect of the integral operator, and consequently, the less distribution ry
manifests itself in peculiarities of measured distributions l(s). Accordingly, the considered approaches provide for a preliminary computa-
x,
(a)
x
(x)
x
l
(b)
Design diagram of Volterra integral equation
method. (a) Residual stress relieved at notching.
(b) Measured parameters of stress state at the notchfollowing point
Fig. 1
(S)
y(x)
+
Fig. 2
Design diagram for method of Fredholm integral

equation
Methods for Determination of Inhomogeneous Residual Stress Fields / 127

higher than those of the problem operator, and
in the event of considered problems, we regard
the operator to be specific with precision.
Following the principle of regularization, it is
necessary to introduce a priori information about
functional properties of residual stress distribution when solving integral equations of the first
type (Eq 1, 3). Usually, there is no quantitative
information about distributions searched for, yet,
from the physical viewpoint, it is natural to regard these distributions as sufficiently smooth
functions. Regularization methods based on
qualitative data about smoothness of searched
distributions allow suppression of oscillating
components, whatever their nature. Various regularizing algorithms and principles of their construction are given in Ref 1 to 3.
One of the most universal methods in solving
ill-conditioned problems based on variation principle is the method of smoothing functional
(Tikhonovs method). This functional has the
following form (arguments are left out): T a[ry]
q2L 2(Hry,l) X2[ry], where q2L 2 is a residual in L2; H is the integral operator in Eq 3 (B is
for Eq 1); 0, regularization parameter; and
X is the stabilizing functional whose value is a
measure of function smoothness. In the considered problem that searches for smooth distributions, when selecting X, it is expedient to apply
restriction both to the norm of function ry itself
and to that of its first derivative.
According to the variation principle, the regularizing solution is represented by function r ay ,
supplying the minimum to the functional and determined from Eulers equation for this functional:
(H *H
L)r ay
H *l
where H is an H-dependent operator, while the

operator L depends on the selected functional X.
Accordingly, the functional takes the form:
X2[ry] qryl22 p
dn
dry
l2
where q and p are nonnegative continuous

weight functions. Then:
Lry qry
d
dry
p
dn
dn
Regularization parameter can be selected from

the conditions:
qL 2(Hry,l) d
where d is the specified error of initial data. The

constraint for the norm of derivative in the functional X is an important thing when searching
for required distributions. In this case, uniform
convergence of regularized solutions, ray (n), to
the actual one, ry(n), is ensured at decreasing
dthat is, max|ray (n) ry(n) 0| at d 0
and corresponding decrease of ( can tend to
zero not faster than d) (Ref 1). It should be noted
that solving Eq 1 and 3 by reducing them to Eulers equation provides for transition to a respective self-conjugate operator (H *H or B*B), this
resulting in the loss of Volterra property. There
are ways of regularization to retain this property
(Ref 4), but they are more complex and effort
consuming. To construct matrix analogues for
operators of Eq 1 to 3, grid quantizing is selected
for the area, including a created notch where the
grid becomes dense. Kernels of integral operators B(x,n), Bx(x,n), and H(s,n) are calculated by
solving a set of boundary value problems, with
d impacts being reproduced by applying a uniform load ry(n*) const in the region of notch
edge formed by one or several grid steps, where
n* is the center of such region. Elements of kernels B(x,n) and H(x,n) are parameters of the
stress-strain state analogous to measured parameters l(x) and l(s), respectively. The kernel
Bx(x,n) is constructed by B(x,n) values with the
aid of respective difference formulae.
Solutions of boundary-value problems for
two-dimensional elasticity theory can be obtained easily on the basis of software complexes
ANSYS, NACTRAN, and the like, which provide high accuracy of results for complex geometric parts with high-stress gradients.
Calculation of residual stress on the basis of
solved integral Eq 1 and 3 is implemented with
the smoothing functional method grounded on
programs developed by the authors.
Evaluation of Method Accuracy. As an example of residual stress restitution in flat parts,
consider results of the numerical experiment
based on reducing the problem to the Fredholm
integral equation of the first type (Eq 3). Such
an experiment aims at demonstrating efficiency
of proposed approaches because Eq 3, all other
conditions being the same, is less conditioned
compared with the Volterra equation of the first
type (Eq 1). A square plate with the side 2l0 (Fig.
3), where a thin notch with the length l0 was cut
along the axis n, was chosen. Residual stress distribution was specified on edges of this notch,
and distribution of the stress state parameter l(s)
on the line s parallel to the axis y was found by
solving a boundary value problem of elasticity
theory (direct problem).
Values of tangential stress sny(s) were calculated as l(s) with a bearing on photoelastic coatings. These values were used as exact initial data
into which random disturbances were introduced
with a normal law of distribution, the level of
disturbances comprising about 5% from the
maximum values sny(s) over the notch length.
The disturbance data were used in residual stress
restitution (inverse problem).
Solution of direct and inverse problems as
well as construction of the matrix analogue for
the integral operator of Eq 3 was made on the
grid crowded in the vicinity of 0.2l0 notch tip.
Figures 3 and 4 present restitution results for
two variants in distribution of residual stress
with different stress gradients.
Solution of direct and inverse problems as
well as construction of the matrix analogue for
y
s
1.0
l0
0.2
0.6
s/l0
(a)
y
12
8
4
0.6
0.2
x/l0
(b)
Problem 1. Determining residual stress by Fredholm integral equation (computation experiment). (a) Initial data. (b) Stress along the notch line: full
lines, exact values; dotted line, restituted stress for two variants of random disturbance in initial data
Fig. 3
1.0
0.2
0.6
s/l0
(a)
y
12
8
4
0.2
0.6
x/l0
(b)
Problem 2. Determining residual stress by Fredholm integral equation (computation experiment). (a) Initial data. (b) Stress along the notch line: full
lines, exact values; dotted line, restituted stress for two variants of random disturbance in initial data
Fig. 4

the integral operator of Eq 3 was made on the
grid crowded in the vicinity of 0.2l0 notch tip.
The value of error introduced into initial data
corresponds to the most real level of random oscillating irregular component in distribution of
sny(s) parameter. The fact is that the distribution
sny(s) over the line s, every point of which belongs to the regular zone of the considered region, is an analytical function and determined by
solving a problem of elasticity theory. Thus,
when processing the experimental data preliminarily, it is necessary to use them for constructing the smoothest analogue of this distribution.
All coarse oscillating errors of initial data of the
order of 10 to 20% will be eliminated when constructing such analogue. The level of oscillating
error in initial data of the order of 3 to 5% is
difficult to control, thus making the main disturbance value responsible for the looseness of
the searched solution. As to the systematic error
in the form of smooth component of initial data
shift, its influence on the accuracy in solving the
inverse problem is less than that of the oscillating component. The presented results illustrate
the effectiveness of the regularizing algorithm
(2.4) when interpreting the distribution sny(s)
containing random disturbance. The maximum
deviation of restituted values from exact values
does not surpass 10 to 12% relative to maximum
values over the notch length, this being quite a
satisfactory result at 5% error level of experiment, a figure that corresponds to the actual accuracy of the photoelastic coating method. Numerical analysis demonstrated that the
regularization parameter in the scheme (2.4) can
be chosen out of the condition mindray /d
(i.e., it is possible to confine oneself to choosing
a quasi-optimal value while solving the problem). Attempts to solve Eq 3 without regularization methods are totally invalid. The disturbance level of obtained solution conditioned by
irregular random components in initial data is so
high that the obtained solution has nothing in
common with the actual one.
The considered approaches to solving the
problem of residual stress continuous distributions demonstrate that recurring to methods for
solving ill-conditioned problems, together with
numerical ones for solving boundary-value problems, allows for efficient computational/experimental schemes of investigation, which are free
from simplifying assumptions about the nature
of searched distributions. An important peculiarity of such approaches lies in the combined use
of numerical and experimental methods as the
unified system for investigation where the main
role is played by the computational experiment.
The results presented can be extended to problems of determining spatial distributions of residual stress, difficulty in solving such tasks being caused by a large body of information to be
processed, and by two-dimensional integral
equations.
Some Peculiarities in Applying the Method.
Practical application of the approach developed
for studies of residual stress includes three principle methodological aspects:
The process of creating notchesindicators
of residual stressin the part under investigation

Methods for obtaining the initial information
l(x) or l(s) for subsequent calculation of residual stress
Selection of procedure to solve equations that
link the residual stresses to be found with
those experimentally discoveredl(x) and
l(s)
Processes and methods used in indicator-crack
studies of residual stress and presented in the
section Evaluation of Method Accuracy can be
applied as techniques of workpiece notching and
methods of obtaining source information for
subsequent residual stress calculation.
Calculation of residual stress on the basis of
solved integral Eq 1 and 3 is implemented with
the smoothing functional method grounded on
programs developed by the authors of this work.
other, can be established after analytically solving the problem of SIFs in the inner crack (Fig.
5), whose edges experience arbitrarily distributed loads ry ry(x) and sxy(x) (Ref 6, 7):
x
[ r (n)n/(x n)]dn
K (x) 2/(px) [ s (n)n/(x n)]dn
KI(x) 2/(px)
II
xy
It should be observed that the equations in Eq

4 are Volterra integral equations of the first type
with the Abel kernel, this providing for the possibility of their analytical conversion. After applying the Abel operator
d
dx
x n
2
[]dn
to both the right and left sides of the equation,

the following is obtained:
Indicator Crack Method
d/dx
ry(x) 2(px)1/2d/dx
Method of crack as indicator of residual stress

can be interpreted as a particular case of the general approach considered in the article Measurement of Residual Stresses in this Handbook, which regards the determining of residual
stress as an inverse problem of experimental mechanics. In this case, the notch, whose edges feature relieved residual stress, is represented by a
mathematical cut (crack), while the parameters
of stress state l(x) are represented by distribution
of stress-intensity factors for normal breakoff
and transverse shear, KI(x) and KII(x). Such statement for some boundary-value problems reduces
the main equation of the problem to the Volterra
integral equation of the first type, with Abels
kernel, this providing for their closed analytical
solution. As a result, the consistent calculation
procedure for computing residual stress from experimental data becomes quite simple. At the
same time, the experimental investigation procedure is more effort consuming since, in order
to obtain the function of stress-state parameter,
values KI and KII must be determined at each step
of experimental study.
Study of Residual Stress in Inner Regions
of Plates. This method is intended for determining locally inhomogeneous fields of residual
stress in flat parts.
The general outline of the method proposed
in Ref 5 that makes it possible to investigate residual stress in inner regions of plates can be
summarized as follows. A crack is built up stepwise along the line n [0,x], which must be
determined for residual stress (Fig. 5). At each
step in the vicinity of the crack tip, the stress
field must be determined, which, in turn, is used
to calculate values of stress-intensity factors
(SIF) for standard breakoff and transverse shear
KI and KII. Connection between residual stresses
rx(x) and sxy(x) on the one hand, and relationships KI(x) and KII(x) obtained experimentally on
continuously increasing the crack length, on the
(Eq 4)
sxy(x) 2(px)1/2
[K (n)n/(x n)]dn
[KI(n)n/(x n)]dn
II
(Eq 5)
These expressions are main-calculated relationships for the built-up crack procedure used
to determine residual stresses in the inner regions
of homogeneous plates. From Eq 5 follows the
correctly set task of calculating residual stress on
the basis of experimental data, this fact, in turn,
ensuring consistency of obtained results against
measurement errors. On the other hand, applying
this method requires more effort-consuming experiments as compared with the general method
considered in the article Measurement of Residual Stresses in this Handbook.
It should be noted that in contrast to the methods of successive grinding, the method of builtup crack is applicable for the study of two-dimensional fields of residual stress. Moreover,
since the indicator crack is a more intense stress
concentrator in comparison with the circular
hole (let alone the methods that require the part
to be cut completely), this provides for its higher
susceptibility.
y
(x )
x
(x)
xy
x,
Fig. 5
Internal crack in infinity plate

Steps at crack buildup can be quite small, thus
ensuring the possibility of determining residual
stresses in regions of higher gradients. For example, in the investigation described in the section Residual Stress Investigation Examples,
steps for crack buildup amounted to the value of
about 1.0 mm (0.04 in.). If applied, the method
of hole drilling would call for 0.1 mm (0.04 in.)
bores to determine stress for 1.0 mm (0.04 in.)
row pitch, this being practically unfeasible
(Ref 8).
At the same time, the possibility of applying
the correlation in Eq 5 to calculate residual
stresses from experimentally found dependencies KI(x) and KII(x) is restricted to determining
stress in the inner regions of flat parts.
Study of Residual Stress Close to Edges of
Piecewise-Homogeneous Plates. The method
just described, which is based on Eq 5, is applied
only to determine residual stress in the inner regions of homogeneous parts. Yet a number of
important practical tasks is connected with the
necessity to investigate residual stress in the vicinity of borders. In the first place, here belongs
the study of residual stress in bimetallic materials used in power installation shells with corrosion-proof surfacing. During the heating process,
the difference in thermal-physical and mechanical properties of base and cladding layers of a
bimetal causes plastic compressive strains,
which, when parts are cooled down to room temperature, lead to residual stresses. These stresses
cannot be relieved by subsequent heat treatment.
Another important practical task is the analysis
of residual stress in members of structures made
of heteromodulus materials. Along with the process manufacturing of a structure member or a
two-layer material, another important factor re-
y(x, 0) = (x, y)
E1,v1
E1,v2
y(x, 0)
sponsible for emergency of residual stress is represented by the difference in elasticity moduli of
materials with two-layer composition. In this
connection, the research (Ref 9) obtained analytical relationships of Eq 5 type in the event of
edge crack in a bimetallic band (Fig. 6). These
relationships made it possible to determine stress
on crack edges ry(x) and sxy(x) (residual stress)
on the basis of known relations KI(x) and KII(x),
which correspond to behavior on crack edges x
[0,a] (E1, E2 and l1, l2 represent Youngs
moduli and Poissons ratios for materials of the
first and second layers, respectively). When deriving respective relationships, the solution involving calculation of SIFs for an edge crack in
the band whose edges experience arbitrary loads
ry ry(x) and sxy sxy(x), obtained in
Ref 10, was used.
If the tip of the indicator crack is in the first
layer, the expressions for calculating residual
stress are of the following form (Ref 9):
KI paW1(a) KII paW2(a)
[W1(u)m1(x,u)]du
sxy(x) (d/dx)
[sW2(s)/(x 2 s 2)]ds
[W2(u)m2(x,u)]du
(Eq 8)
Expressions for calculating m1(x,s) and m2(x,s)

can be found in the subsection Calculation formulae for determining residual stress.
For an essentially important case, E1 E2, l1
l2, and h/H 0 (edge crack in the homogeneous plate), the following expressions are deduced from Eq 7 and 8:
m(x,u)
Border crack in bimetallic strip under arbitrary

loading on the crack bands
x [0,a] (Eq 7)
ry(x) (d/dx)
+xy(x, 0)
[sW1(s)/(x 2 s 2)]ds
[sW1(s)/(x 2 s 2)]ds
Fig. 6
ry(x) (d/dx)
(Eq 6)
[W1(u)m(x,u)]du
u(1 ax)e axn(au)d

p/2
n(au) 2/p
(1 au sin u)e au sin udu

(Eq 9)
y(x) =
a/2
a
3
2
+(x) = 25
Fig. 7
cessing
+
(x) =
y(x)
y
Schematic of the test problem for estimation of

procedure accuracy for experimental data pro-
The study described in Ref 9 also obtained

relationships for determining residual stress in
the case that h a H. Being cumbersome,
they are not cited here. At the same time, there
is no practical need to apply the solution for
cases when the crack tip is in the second layer
of bielastic band. Preferable is the investigation
procedure based on the processing of two experiments during which the indicator crack is, at
first, built up in one layer, then at some distance
from the first crack in the second layer (or in
analogous specimen).
Evaluation of Method Accuracy. Based on

experimentally obtained relationships KI(x) and
KII(x), the correctly formulated task for residual
stress calculation provides consistency of respective procedures. Various approaches can be
used to solve the problem.
A simple and reliable way to process experimentally obtained relationships KI(x) and KII(x)
on the basis of Eq 5 and 7 to 10 is represented
by the following procedure:
1. Experiment results obtained as KIi(xi) and
KIIi(xi) values (where xi is the length of the
indicator crack corresponding to i-step) were
smoothed on the basis of spline approximation.
2. Integrals are computed by the trapezium
methodfor example, on the basis of the
Romberg algorithm.
3. Obtained relations are differentiatedfor example, using the modified method of Ridder.
When solving problems for residual stress
analysis in cases described by Eq 5 and 10, it is
possible to realize this procedure of experimental
data processing by applying standard program
complexes.
Following is an example of solving the test
task for a piecewise-linear stress field of residual
stress close to the border of a homogeneous half
plane, the example being considered in the context of evaluating the accuracy in the mentioned
procedure of data processing (Fig. 7).
In this case, the problem of determining the
KI(x) function can be represented as a sum from
the solution of three boundary problems:
Problem A ry(x)
Problem B ry(x)
Problem C ry(x)
ry(x)
r
2rx/a
0
5r
(0 x
(0 x
(0 x
(a/2
a)
a)
a/2)
x a)
KI(x) relationships were plotted on the basis

of available analytical solutions for problems A,
B, and C (Ref 11, 12). To simulate experiment
errors with the aid of a random-number generator, random values K I*(xi) KI(xi) DKI(xi)
were introduced in i-points uniformly distributed
in the range [0,a]. Two cases of uniformly partitioned section [0,a] were considered in doing
so: I 20 and I 40. Range of scatter in values
of SIF DKI(xi) comprised 10% of the maximum
value KI, this being in agreement with experimental error while determining the factor on the
basis of modern techniques (the section Experimental Methods and Equipment provides details). Residual stress was calculated in accordance with this procedure. The results are shown
in Fig. 8.
These data reveal high accuracy of calculation
for residual stress ry(x) and consistency of results. Even in the vicinity of point x a/2
that is, the zone featuring an abrupt change of
stress (an unlikely event in practical tasks)the
difference between exact values and those computed by Eq 9 does not exceed 10%. Outside this
zone, the value of the mentioned error does not

exceed 5% of the maximal stress (and 3% from
the value of stress leap at point x a/2) in both
considered cases of the partitioned section 0
x a.
Some Peculiarities in Applying the
Method. Practical application of the developed
approach to investigation of residual stress comprises three principal methodological aspects:
Technique for creating the indicator crack,
which would preclude extra stresses in the

workpiece and coating
Methods for determining SIFs at combined
loading on the basis of results processed for
stress fields, strains, or displacements in the
zone of crack tip
Selection of procedure for solving equations
that connect residual stresses to be found with
experimentally discovered relationships KI(x)
and KII(x).
A rectangular cutout made by a 0.2 to 0.5 mm
(0.0080.02 in.) milling cutter can be used as an
indicator crack. If the cutout length is much
greater than its width, h (h/a 0.03), then, on
the basis of methods described in the section
Experimental Methods and Equipment, it is
possible to determine quite accurately those SIFs
that could develop should a corresponding crack
be created. In this case, the source information
about the stress-strain state is obtained in the region that excludes the vicinity of cut-out extremity r 3h (r, distance to the cut out tip). It
might be practical to use a narrow rectangular
cutout instead of an indicator crack also because
at high levels of residual stress, local secondary
plastic strains can develop due to high-stress
concentration in the vicinity of the crack tip. If
narrow indicator-cutouts are used, the likelihood
of developing secondary plastic strains is reduced. It is worth noting that the use of SIFdetermination methods presented in the section
Experimental Methods and Equipment rules
out the impact of local plastic strains in the in-
dicator-crack tip zone on the determining accuracy. On the other hand, presence of local plastic
strain to some extent influences the stress-strain
state of the object surfaces relieved on object cutting. It is determined experimentally (on the basis of strain field results) whether local plastic
strains are present or absent. Each specific case
calls for assessment of their impact on residual
stress to be found with due consideration to their
actual level and peculiarities of the problem
solved. Still, it should be pointed out that development of secondary plastic strains is not limited
to the method of built-up crack. This phenomenon may well attend any destructive method
for example, in the widely used method of circular hole drilling when, due to stress
concentration in the contour zone, there may develop stresses that exceed the yield point of the
material under investigation.
The problem of determining SIF from processed results of polarization-optical measurements at combined loads is that there is a pronounced experimental error (Ref 8, 14, 15, et al.)
in the crack tip zone (Fig. 9), where the stressed
state is assumed, from the standpoint of fracture
mechanics, to be described by Vestergards asymptotic relationships:
rij(x,y)
2pr
KN FN (h)
k1
(i.j x,y; N I, II, III)
(Eq 10)
This calls for applying data obtained at some

distance from the crack tip while determining
SIF, and this fact, in turn, requires that not only
asymptotic elements be considered, but also regular ones of the full solution for the respective
problem of elasticity theory.
For the reliable determination of the SIFs KI
and KII under combined loading, it is necessary
to use the methodology, based on mathematical
treatment of a substantial volume of the experi-
mental information (Ref 1618 et al.). Therefore, for the measurement of strains and displacements fields in the area of the crack
indicator, the optical-geometrical methods
should be used in the first place, as well as the
methods of holographic interferometry and electronic digital speckle interferometry, and also the
method of photoelastic coatings (Ref 8). In considering the holographic methods, the scheme of
the receiving of real-time hologram (Ref 19) can
be used for the receiving of the interferencion
pictures.
The modern methods for SIF determination
under combined loading require only the determination of the function coefficients. This function describes the field of stresses in the crack
zone on the basis of the experimental information. The function of stresses in the form of the
series can be used for the analytical representation of the field of stresses in the crack zone. The
terms of the series are the own functions (Williams functions) of the solution of elastic theory
problem for the regions with the wedge cutouts:
U(r,h)
r n/21f (h)
n1
(Eq 11)
where r,h are the polar coordinates, f (h)

an{sin[(s 1)h] (s 1)/(s 1) sin [(s
1)h] bn{cos[(s 1)h] cos [(s 1)h]}; s
n/2; and an,bn are the coefficients for the symmetric and antisymmetric components of the
function U(r,h) expansion. Notice that KI
a12p, KII b12p.
The finding coefficients an, bn, (n 1,2,3
N) are defined on the basis of the mathematical
treatment of the measurement data of the band
order smax (the method of photoelastic coatings
is using) or the normal displacements, w (the holography method) in i points (i N ).
Using the method of photoelastic coatings: the
expressions for the stresses have the form:
2smax r1 r2 (rh rr)2 4s r2h
rr rh
2
sr h
r n/21(an F1n bn F2n)

n1 2
r n/21(an F3n bn F4n)

n1 2
(Eq 12)
where:
F1n (s
(s
F2n (s
(s
F3n (s
(s
F4n (s
(s
1)
1)
1)
1)
1)
1)
1)
1)
sin[(s 1)h]
sin[(s 1)h]
cos[(s 1)h]
cos[(s 1)h]
cos[(s 1)h]
cos[(s 1)h]
sin[(s 1)h]
sin[(s 1)h]
(Eq 13)
and, using the method of holographic interferometry:

N
lt
r n/212n
E n1
{an sin[(s 1)h] bn cos[(s 1)h]
Fig. 8
Results of calculation of the test problem. I, number of steps
(Eq 14)
G1, 1
G2, 2
r
Fig. 9
Crack with a tip on the boundary of heterogeneous materials
where E is the materials elastic modulus, l is

Poissons coefficient, and t is the thickness of
the part.
Since the local geometry of the crack-tip zone
does not correspond to the mathematical slitting,
the place of crack-tip position (r 0) in certain
measure is conditional. In connection with that
in Ref 16 and 17, the specification of the cracktip position, corresponding to the experimentally
obtained fields of stresses or displacements in the
crack zone, was provided. Therefore, the corrections on the crack-tip state (dx,dy) are the finding
parameters also as an,bn. That specification also
is important because the errors in the determination of the crack-tip state can appreciably influence the finding values of SIFs.
If the photoelastic coating method is used, the
obtained system of equations is nonlinear, which
requires the application of corresponding methods for their solution (for example, the NewtonRaffsons method).
The summary approach is provided to the
creation of the analytical solution of the elastic
theory task, corresponding to the initial experimental information. In difference of the other approaches to the solution of the determination of
the stress-intensity factors task, this method is
true in view of the uniqueness of the elastic theory task solution.
In Ref 18, on the basis of the numerical tests,
the possibility of an application of the given approach to the determination of the SIFs under
combined loading in the case of loading of crack
banks with the nonhomogeneous stresses changing with the high gradients is discussed. The task
about the piecewise-continuous broken field of
normal and shear stresses was considered test
one; the stresses act on the banks of a crack in
an infinite plate. The appeared stress-strain state
was defined on the basis of the precise analytical
solution of the corresponding task of elastic theory (Ref 7). The values KI and KII were determined using the approach with the displacement
field. The results of the investigation show that
the relative difference between the precise values
of the SIFs and values, based on the mentioned
approach, consist of less than 5%. It happens
even in the cases in which the point-of-stresses
rupture lies in close distance from the crack tip
D/a 0.05 (D is a point of slitting and a is the
crack length).
These results justify the recommendation of

the mentioned approach for the determination of
the SIFs caused by creation of a crack indicator
in the investigated parts.
Calculating residual stress from experimentally obtained dependencies KI(x) and KII(x) on
the basis or relationships in Eq 5 to 9 does not
call for specially developed techniques due to
correctness of the problem solved. Both direct
numerical methods and standard software complexes can be used for computation. Preference
should be given to those procedures that provide
for spline approximation of KI(x) and KII(x)
dependencies at the first stage of experimental
data processing.
Peculiarities in Studying Piecewise-Homogeneous Materials. Equations 6 to 8, used
to calculate residual stress from experimentally
obtained dependencies KI(x) and KII(x), correspond to the case in which an edge indicator
crack is located in a layer of bielastic bands.
Hence, it follows that in studies of residual
stress, an indicator-crack tip should be at some
distance from the boundary between heterogeneous materials.
If the crack tip is on the juncture of heterogeneous materials (Fig. 9), then the respective
boundary value problem of elasticity theory has
an asymptotic solution, different from that for
the case of a crack located inhomogeneous milieu.
The characteristic equation to determine the
degree of k stress singularity, in this case obtained by Zak and Williams, is of the following
type (Ref 20):
2(k2 k1)(k1 1) cos kp 2k 21 2k1k2 2k1
k2 1 4k1(k2 k1)(k 1)2 0
(Eq 15)
where k1 (G1 /G2 1)/4(1 v1) and k2

k(1 v1)/(1 v2); k G1 /G2; G1,G2 are shear
modules of materials.
The k value to be found is the sole real root
of Eq 15 in the range (1, 0). Thus, stresses in
the crack tip have the peculiarity of the r k type,
with k being different from k0 0.5, which
corresponds to the case of crack located in the
homogeneous material.
The following table presents k values calculated on the basis of Eq 10 for l1 l2 0.3
and different ratios G1 /G2:
G1 /G2
0.2
0.6
1.0
2.0
4.0
6.0
10.0
the juncture D. It is evident that at k* k at D

0; k* k0 0.5 with growing D.
With those considerations, the following procedure for study of residual stress in piecewise
homogeneous materials (bielastic band) can be
accepted:
1. Two analogous specimens are used in the research. Residual stresses in each layer are determined on separate specimens. If residual
stress along the boundary of heterogeneous
materials does not change, the study can be
made using only one specimen, with the first
and second indicator cracks located at the distance that rules out their mutual influence.
2. Dependencies KI(x) and KII(x) in the range 0
x h are determined experimentally for
each layer. In this case, the SIF-determination
methods developed for study of cracks in the
homogeneous material are applied.
3. Experimentally obtained dependencies KI(x)
and KII(x) for the first and second layers are
used to calculate residual stresses ry(x) and
sxy(x) in the range 0 x h with the aid of
Eq 17 to 20 in the following section.
4. Values of residual stress on the boundary between heterogeneous materials are determined by extrapolating the found dependencies ry(x) and sxy(x) to a corresponding locus
of the boundary.
The specified method excludes the necessity
to determine SIF for the case when the crack tip
is located either on or close to the boundary between heterogeneous materials.
Calculation formulae for determining residual stress in the piecewise-homogeneous
(bielastic) band are as follows. If the tip of the
indicator crack is in the first layer (a h), the
expressions for calculating residual stress are of
the following form:
KI paW1(a) KII paW2(a)
As the crack tip nears the boundary of heterogeneous materials, the singularity degree of k*
stress changes depending on elastic properties of
materials and the distance from the crack tip to
[sW1(s)/(x2 s2)]ds
[W1(u)m1(x,u)]du
x [0,a]
(Eq 17)
where
m1(x,u)
u[A1()(xch x sh x)
A2()(2ch x sh x)
M(u)ch ax]da
k
0.366
0.450
0.5
0.575
0.654
0.699
0.754
ry(x) (d/dx)
(Eq 16)
Aj () Aj (h, h1, u, k, l1, l2) k G1/G2

4
Ai fji qj
i1
( j 1,4)
f11 2(1 l1)C2 S21 f12 (1 2l1)S2 C21

f13 k[S11 2(1 l2)C1]
f14 k[(1 2l1)S1 C11]
f21 C21 (1 2l1)S2
f22 S21 2(1 l1)C2
f23 k[(1 2l1)S1 C11

f24 k[2(1 2l2)C1 S11]

f31 S2 C21 f32 S21 f33 S1 C11
f34 S11 f43 f32 f42 S2 C21
f43 f34 f44 S1 C11
q1 Y1 MS2
q2 Y2 MC2
q3 Y3 MC2
q4 Y4 MS2
Y1 [2(1 l1)I0(u) Y4]e ah

Y2 [2(1 l1)I0(u) Y2]e ah
Yq (c h)I0(u) sI1(u)
M(u) I0(u) L0(u) u[I1(u) L1(u)]
Cj chuhj Cj1 uhjCj
Sj shuhj Sj1 uhjSj
Cj 1,2 h2 h; h1 H h; q 3,4
at g 3, C 1
at g 4, C 2
Ij ( j 0,1) modified Bessel functions of the first

type
Lj ( j 1,0) modified Struve functions
(Eq 18)
sxy(x) (d/dx)
[sW2(s)/(x2 s2)]ds
[W (u)m (x,u)]du
m (x,u) au[M(au)ch ax
On the basis of the analytical relations between the contour residual stresses and the
stresses (or strains, displacements) arising in the
specimen, and which can be measured on the
cut-out contour, the residual stresses can be determined. The previously mentioned stages of
the proposed approach were put to the base of
all the rupture methods of the residual stresses
measured. However, a main feature of the cutout indicator method consists in the suggestion
about arbitrary distribution of the residual
stresses on the contour, on which the cut is carried out.
General Relations. To define the residual
stresses by use of the cut-out indicator method,
it is necessary to receive the contour residual
stresses by some characteristics of stress-strain
state, which were obtained with the help of the
experimental methods in some area adjoining the
contour of the cutting, without its inclusion. That
means the task inverse to the first general boundary problem of the elastic theory is solved.
As a basic relation, the Muskhelishvili equation of the elastic theory can be used in the view:
0
u(r) x(r)/x(r)u(r) w(r) f1 if2 (Eq 21)
A2()(axch ax sh ax)
A1()ax sh ax]da
(Eq 19)
where M(u) is determined on the basis of the

relationships in Eq 7, while the formulas for Eq
16 and 17 are:
( j 1,2,3,4)
d1 Y 1* MC2 d2 Y 2* MS2
d3 Y *3 MS2
d4 Y *4 MC2
Y *1 Y *4 Y0 Y *2 Y *3 Y0
Y *4 Y0/2(1 l1) Y3
Y0 2(1 l1)I0(u)eah
Y3 [hI0(u) uI1(u)]eah
f1 if2 i
t0
(Xn iYn)dS
(Eq 22)
The expansion into a Fourier series under the

mapping onto the circle of plane specimen,
bounded by given contour c (Ref 7) can be used
for the solution of the first general boundary
problem of the elastic theory:
A1 fji dj
i1
on c-loaded contour. Here u and w are the stress

functions; x(r) is the function, which conformally maps the contour c on the circle; f1 if2
are the functions of loading on the contour and
(Eq 20)
Arbitrary Cut-Out Indicator

Method
The cut-out indicator method is based on an
application of the boundary equation of elastic
theory. The main idea of this experimental-analytical method is that the cut of a plane specimen
is made along the fixed contour, out of which
arises an additional stress-strain state. This state
appears because of stresses releasing along the
contour of the cutout. This (additional) field of
stresses, strains, and displacements can be measured on the basis of the known interferencialoptical methods (method of optically sensitive
coating, methods of holographic, speckle-interferometry, method of electronic digital speckleinterferometry, and so on, Ref 8). The discrete
values of strains can be obtained by the other
experimental methods (for example, with the
help of a small-base gages measurement).
u(f)
0 ak f k
w(f)
k
0 af
k
(Eq 23)
The functions of conformal mapping and

loading can be presented in the following form:
x(r)/x(r)
bkrk
(Eq 24)
f1 f2
Akrk
(Eq 25)
The possible loadings are limited by ones, which

can be adequately offered by expansion into a
Fourier series.
After the transformation, the system of linear
equations is obtained:
[C]a A
(Eq 26)
where [C] is the matrix of coefficients and a and

are the vectors of expansion coefficients u, w,
A
and f1 if2.
It should be noted that the different functions
of the conformal mapping permit the use of the
given relations for an arbitrary cutout (e.g., the

infinite plate with arbitrary cutout and arbitrary
cutout from the plate with any shape, including
the circular hole and the mathematical cutcrack).
Experimental Data Treatment. The determination methodology of the internal stresses
using the method of round and nonround holes
and cutouts requires the experimental definition
of functions u and w of Kolosov-Muskhelishvilis potentials on the basis of relations for the
stresses:
XX YY 2(u(z) u(z))
YY XX 2iXY 2(zu(z) w(z))
(Eq 27)
or for the displacements:

2l(u iv) xu(z) zu(z) w(z)
(Eq 28)
The given functions are searched in the form

of corresponding Fourier series, and the task of
their determination is reduced to the finding of
coefficients under the expansion into series.
As initial information, the displacements data
u and v in the plane of plate, the normal displacements data (w rx ry XX YY), the
maximum shear stresses data (smax) can be used.
If two or more components of stresses, strains,
and displacements near the cut-out indicator are
known, the treatment and optimization of the
data are maximally simplified. If the normal displacements are known, with the help of them and
one of Kolosov-Muskhelishvilis relations, the
function u can be obtained. To find the function
w, it is necessary to use the data of the other
components of displacements.
If the experimental determination of all three
components of the displacements and their numerical interpolation on cutting contour is possible, then the determination of the contour residual stresses is possible using the relations of
displacements and loading functions on the contour. In this case the nonlinear (including plastic)
residual stresses can be defined because the
given relations are valid for the nonlinear area as
well (Ref 7).
Evaluation of Method Accuracy. The correctness of the residual stresses calculation on
the basis of the experimentally determined fields
of displacements, strains, and/or stresses is provided with the use of received analytical relations, which connect the experimental data with
the searching values on the cut-out contour. It
can be provided also by the possibility of corresponding calculation procedures optimization
and the possibility of treating the substantial
masses of the experimental information. The different approaches and procedures can be used for
the numerical calculations. It should be noted
that the given approach is more simple than the
general method described in previous sections,
and the method of growing cracks, which allow
for simple optimization of the calculating procedures used.
In Ref 13, on the basis of numerical tests, the
precision evaluations of the residual stresses de-

termination using the experimental data, obtained by method of holography and optically
sensitive coatings (the fields of the normal and
tangential displacements, also the difference of
stresses), were fulfilled. The task about piecewise-continuous broken field of the residual
stresses in infinite plate was considered as a test.
The round hole is an indicator of stresses. The
arising stress-strain state is defined by means of
the precise analytical solution of the corresponding task of an elastic theory (Ref 8). The numerical tests were carried out for the three types
of experimental information, characterizing the
stress-strain state in the area of circular cutout:
sums and differences of stresses, values of normal w and radial u displacements, and values of
normal w and tangential v displacements were
used. The results confirm the effectiveness of the
hole-indicator method for the practical application for the investigations of appreciably nonhomogeneous fields of the residual stresses in
plane elements of constructions.
Features of Practical Application of the
Method. Technology of the creation of cut-out
indicators is similar to usual technology, applied
for the investigations of the residual stresses by
the small circular hole method (Ref 8).
Experimental information about the fields of
stresses or displacements in the area of holes
(cut-outs) indicators is presented in the following section.
Experimental Methods and

Equipment
It is possible to apply different stress-strainstate parameters determined experimentally in
studying stress fields, deformation, and displacement as the source data to examine locally inhomogeneous fields of residual stress. This section presents some problems in obtaining
experimental data necessary for determining residual stress on the basis of methods and respective computation procedures whose fundamentals are given in preceding sections.
Determining inhomogeneous fields of residual
stress within the considered framework requires
significant amounts of data be obtained with regard to strain fields or displacements in local sur-
Fig. 10
Outline of photoelastic coating
face zones of investigated objects that develop

after a notch is introduced (exception in this case
is represented by the Volterra integral equation
method described previously). These circumstances prevent one from applying methods of
measuring strain state at individual spots, among
them the small-base tensiometry (with the use of
small gages) widely used in the study of residual
stress. In view of the requirements to the scope
and accuracy of experimental data needed to analyze residual stress by considered methods, it is
possible to conclude that coherent optical methods (holography interferometry, speckle interferometry, etc.), as well as the photoelastic coating method (Ref 8), are most efficient for
obtaining experimental data.
Photoelastic Coating Method. In terms of
equipment employed, the photoelastic coating
method (Ref 8, 21) is one of the simplest and
most reliable ways to study strain fields on the
structure surface under static loads and to investigate stationary interim processes. Once developed, this method has been actively pursued to
study residual stress using various techniques
that employ workpiece cutting, and in particular,
in combination with the circular hole drilling
(Ref 8, 2226, et al.).
Still, this method has several peculiarities, and
it is subject to associated errors that sometimes
can significantly influence the validity of results.
Respective problems are considered in studies
(Ref 22, 24) where one can find ways to reduce
errors or take them into account while applying
the photoelastic coating method. Accuracy of
strain-stress measurement on the surface of photoelastic coating objects is known to depend on
several factors: errors from measurements of optical propagation difference, effect of coating
hardness, and uneven thickness of coating and
effects of irregular strain distribution over the
coating thickness. It was pointed out in some
papers that when measurements in stress concentration zones are taken, the most significant
among these factors that determine experimental
error are represented by the influence of deformation irregularity over the coating thickness
due to high gradients of strain measured on the
surface of the studied workpiece. M. Akhmetzyanov (Ref 24) suggests that the respective adjustment of optical measurement results for the
coating requires that mean deformations be determined separately by coating thickness and
second order equations be solved numerically in
partial derivatives for the region studied. F.
Zandman (Ref 21) recommends repeat experiments with reduced thickness of photoelastic
coating, this, on the one hand, significantly extending the range of conducted experiments, yet,
on the other, being practically unfeasible while
studying residual stress. In keeping with the
aforesaid, the following method is proposed for
choosing the thickness of photoelastic coating,
optimal with regard to obtaining the maximum
value of optical effect with an acceptable value
of result error.
Selecting the Thickness of Photoelastic Coating in the Study of Holes and Notches. The value
of the optical effect in the photoelastic coating

is proportional to its thickness (i.e., as the coating thickness grows, the error in optical measurement decreases). On the other hand, the
growth in the thickness of photoelastic coating
leads to the increase of errors related to the influence of coating hardness and to that of strain
being distributed unevenly over the coating
thickness.
A simulation problem in the study (Ref 21)
called for computations whose results allow for
the quantitative assessment of aforementioned
errors while examining within high gradients of
stress, which are always present when methods
described in the preceding sections are applied.
Such assessment makes it possible to establish
the optimal thickness of the optically sensitive
coating from the required measurement accuracy
with regard to the level of the measured optical
effect.
The simulation problem is outlined in Fig. 10.
It is represented by a two-layer plate with a circular hole, the plate being subjected to the uniform all-directional tension load far enough from
the hole. A thin layer with thickness h corresponds to the photoelastic coating, while a
thicker layer corresponds to the steel specimen.
The change in parameter h/c leads to the change
of strain gradient (e1 e2)/r at constant thickness of coating. Consideration is given to the
parameter change in the range 0 h/c 20.
Calculations were made for the following mechanical properties of materials: ratio of elasticity moduli E1 /E2 0.015, values of Poissons
factors l1 0.36; l2 0.3 (for epoxy photoelastic material and steel). The obtained results
allow evaluation of the method error that
emerges in zones of high gradients of stress and
is caused by uneven distribution of strain over
the coating thickness, this evaluation being made
by comparing difference values of main strains
on the studied surface (e1 e2)z0 and mean
strains over the coating thickness (e1 e2)*
(1/h) h0(e1 e2)(z)dz. Main results of calculation are given in Fig. 11 and 12.
These results make it possible to lay down domains of relationship for parameters h/c and r/
c where the value of considered error, dmax
[(e1 e2)* (e1 e2)]/(e1 e2), does not
exceed the specified value.
On the basis of obtained results and in view
of known quantitatively evaluated errors conditioned by other previously mentioned factors, the
following general conclusions can be drawn and
recommendations made as to the application of
the elastic coating method when taking measurements in zones of high gradients of stress for
steel workpieces studied. If relationships of mechanical properties for materials of studied parts
are different from specified ones, it is necessary
to obtain estimated values analogous to those
implemented.
The results also can be used to estimate measurement error on the notch contour at non-axisymmetrical stress distribution as well as for
noncircular holes, notches, and fillets. To do this,
the value of the maximum gradient of stress in

the measurement zone must be estimated on the
basis of available solutions, boundary layer
problems of elasticity theory with similar geometry, and types of effective load, or those
given directly from experiment data. The value
for the radius of equivalent hole c and parameter h/c then can be determined.
It follows from Fig. 11 and 12 that coatings
with thickness less than 0.3 of hole radius (h/c
0.3) or fillets provide for high accuracy of
measurement. Calculation results demonstrate
that in this case, the error dmax does not surpass
5 to 6% on the hole contour and comprises less
than 3% at distance 0.15 R from the contour.
When optical measurement results are extrapolated to the hole contourthis usually done
while processing data from the photoelastic
method to rule out fringe effect in the optically
sensitive material (Ref 21, 24)the mentioned
error will not exceed 2%.
When residual stress is investigated by the

standard method of hole drilling, the minimal
values of hole radii are commonly assumed to
be approximately 3 mm (0.12 in.). Corresponding evaluations show that in this case 1 mm (0.04
in.) coatings should be used to provide for the
necessary level of optical effect at sufficiently
small deformations. When the arbitrary notch
method described in the section Arbitrary CutOut Indicator Method is applied, the radius of
the hole (indicator of residual stress) is, as a rule,
at least 7 to 10 mm (0.28 to 0.4 in.), this enabling
2 to 3 mm (0.08 to 0.12 in.) coatings to be used
for measurements.
Should the relationships of mechanical properties for the studied object material and photoelastic coating differ significantly from accepted
relationships, respective simulation problems
have to be calculated to evaluate errors. To this
end, it is possible to apply standard software
13
12
11
(e1 e2)T (1 l)ra/r/E2
10
max, %
8
7
6
(e1 e2)0 (1 l)2rc 2 /Er 2
5
4
0
0.2
0.3
0.4
0.6 0.8 1.0
6 7 8 9 10
h/c
where (e1 e2)0 is the difference of main deformations, which was found after solving the
problem analytically.
Gradients of main deformation difference are
determined from the following expression:
Relationship of maximum error dmax on hole contour to the parameter h/c
d(e1 e2)0 /dr (1 l)4rc 2 /Er 3
max = 3%
1.0
(r/c 1)
4%
5%
6%
8%
11%
0
10
(h/c)
Fig. 12
(Eq 29)
where (e1 e2)T is the difference of main deformations in the crack zone. For the circular
hole with radius c in the unlimited size plate under infinite all-directional tension r at r c:
Fig. 11
r1 r2 2KI /8pr
For the inner crack with length 2a in the unlimited size plate being stretched and under infinite
tension stress r, KI rpa. With regard to the
photoelasticity law for the photoelastic coating
method, the following equation is obtained (Ref
1):
14
complexes that allow computation of stressstrain state for bielastic parts using the leastsquare method.
Selecting the Thickness of Photoelastic Coating while Determining the SIF. Results given in
Fig. 11 and 12 can be used to determine the optimal thickness of coating when the indicatorcrack method is applied. This is possible because
the recommended technique for determining SIF
from measurements made on the coating excludes from consideration the zone closest to the
crack tip and makes it possible to reliably determine KI and KII from processed results of measurement in the range r/a 0.05/0.1.
Stress distribution in the zone of crack tip,
which corresponds to an asymptotic solution,
takes the following form at h p/2:
Domains of relationship for parameters r/c and h/c where value dmax does not exceed the specified value
(Eq 30)
By comparing right sides of Eq 29 and 30, it

is possible to determine the radius of hole c at
which there are equal gradients of main deformation difference on the hole contour and on the
boundary of crack vicinity used as domain of
source data when determining SIF (r/a 0.05/
0.1). Corresponding assessments demonstrate
that if the vicinity of the crack tip is used for SIF
determining (as shown in the following sections), the value r/a 0.05, and at r/a 0.07,
c 0.2a. Hence, it follows that at crack length
a 15 mm (a 0.59 in.) it is acceptable to
use 1 mm (0.04 in.) coatings if the domain for
obtaining source information r/a 0.07 is
adopted.
It should be noted that this result is obtained
with allowance made for stress fields in the crack
tip vicinity being represented asymptotically;
therefore, values of stress gradients in the crack
zone actually are overestimated.
Optic Interferometry Methods. Considering all types of notches discussed in the three
previous main sections, parameters of stress-

strain state to be determined in the vicinity of
notches (residual stress indicators) can be represented by components of surface displacement
field in the object studied, these components appearing after a notch is introduced. The most effective ways to register local changes in the surface form are coherent-optical interference
methods. Compared with the photoelastic coating method (Ref 27), these have certain advantages since they provide for measurements directly on the surface under study. It should be
noted, however, that the equipment used in this
case is much more complex compared with that
applied with the photoelastic coating method.
The same applies to the experiment technique,
including special requirements to notches-indicators in preparing which any displacements of
the workpiece as the rigid ensemble should be
ruled out.
Different approaches aimed at studying residual stress in cut parts by coherence optics have
been well developed so far. Various schemes are
used to obtain displacement fields after a notch
is made: the real-time method, double exposure,
and sandwich-hologram method (Ref 2934).
Most of these works, as well as some publications in digests (Ref 25, 26), are dedicated to the
study of residual stress employing the hole drilling method; the paper (Ref 28) considers layerwise grinding while the work (Ref 19) considers
the method of successive crack buildup. Study
of residual stress that employs coherent optics is
systematically presented in monographs (Ref 35,
36).
The study in Ref 34 is quite distinctive among
these publications. This work presents a method

that, apart from resolving components of normal
displacement, makes it possible to reliably determine and use tangential components of displacements for residual stress evaluation, this
fact significantly expanding research potential.
With regard to peculiarities of investigating
locally inhomogeneous fields of residual stress,
the most promising measurement tool seems to
be the electronic digital speckle interferometry
that combines the known benefits of coherent optical methods (contact-free nature, high-sensitivity, lack of preliminary procedures on the studied
object) with efficacy of direct computer presentation, storage, and processing of data (Ref 8,
37). In this context, the design and performance
capabilities of the special interferometer developed for research of residual stress is discussed
subsequently. The interferometer provides for
separate recording of all three components of the
displacement vector (Ref 38). It is worth mentioning that a special small-size portable unit was
developed from this design to enable research
both inside and outside the laboratory.
A plane laser beam passes through a glass
prism (1) (Fig. 13). In the classical Michelson
interferometer, a deflected wave front is used to
measure the normal component of the displacement vector W with the aid of a semi-transparent
mirror (2) that developed two beams out of the
plane wave, these beams lighting both the studied area on the object (3) and the diffusely reflecting base surface (4). Images of resulting
speckle structures are focused with the lens (5)
onto the charge-coupled device (CCD) (6)
5
4
CCD
10
placed in the imaging plane (c), with a digitized

signal from the latter being registered in the form
of computer files.
On passing through holography elements, the
beam is transformed into three plane waves, two
of which lighten the studied surface by means of
mirrors (8) and (9) at 45 angles to the normal
in the plane of axis x. The semitransparent mirror
(10) transforms the central beam in two waves,
which, when reflected from mirrors (11) and
(12), illuminate the surface under study at 45
angles to the normal in the plane of axis y. To
measure displacements in the object after its
notching, well-known optical designs of speckle
interferometers sensitive to components can be
used in the plane (U and V) and away from the
plane (W). Such complex experimental installation actually combines three interferometers,
with optical designs for determining U and V
being completely identical yet turned by 90
round the normal to the object. Each of the three
components in the spatial vector of displacement
is determined in arbitrary order and independently from the two remaining ones. Figure 14
shows a plane fragment of the optical design for
determining components W and U. To measure
the normal component of the displacement vector W, there is a classical Michelson interferometer with the semi transparent mirror (1) that
makes two beams out of the plane wave A, these
beams illuminating along the normal both the
studied region on the object (2) and the diffusely
reflecting base surface (3). The tangential component U is determined using the optical design
with two inclined beams B and B (with beam A
closed) placed in the same plane with the normal.
Images of the total speckle structures formed are
focused with the lens (4) onto the CCD matrix
(5) placed in the imaging plane (c), with digitized signals from the latter being registered in
the form of computer files. Deducing image intensity distributions (speckle structures) recorded before and after the object deformation
and corresponding to each type of interferometer
finally results in the system of interference
fringes
12
y
1
7
13
11
Optical design for the special speckle interferometer. 1, glass prism; 2 and 10, semi-transparent mirror; 3,
object; 4, reflecting base surface; 5, lens; 6, charge-coupled device; 7, diffracting screen; 8, 9, 11, and 12,
mirrors; 13, rotated parallel-sided plate
Fig. 13
Fig. 14
Fragment of optical design for the special

speckle-interferometer
Fig. 15
Interference patterns in the zone of hole-indicator. (a) Normal displacement w. (b,c) Tangential displacements u and v
whose orders are connected to values of displacement components by linear relations:

x
B(x, )
1
A
B
0.5
A D
C
D
0.5
Layout of tracking points in the vicinity of

notch tip; behavior of the Volterra integral
equation kernel at these points
Fig. 16
Fig. 17
Fringe pattern in the photoelastic coating (figures represent fringe order)
w kNi/2 u kNj/2 cos

w kNk/2 cos
(Eq 31)
where k is the wavelength of the emission used;

Nl, l i,j,k are orders of interference fringes on
corresponding speckle interferograms, that is,
geometric loci of equal displacement points, normal to the surface (if displacement is determined
from the plane) and tangential (inside the plane
that contains direction vectors of lighting
beams).
The speckle interferometer can be applied successfully to study locally inhomogeneous fields
of residual stress, as a means of measuring the
strain response after a notch is introduced. High
sensitivity of speckle interferometry enables one
to measure reliably even with small diameter
holes (23 mm, or 0.080.12 in.).
To demonstrate interferometer capabilities,
Fig. 15 displays typical speckle interference patterns that characterize displacements in the zone
of the circular hole indicator with a 2 mm diameter, obtained when studying residual stress
in the weld zone.
The method allows for values of displacement
both on the notch (hole) contour and those at
some arbitrary range from it as input data while
solving the inverse problem, with no distortion
in scale commonly inherent to the cognate
method of holographic interferometry.
Electron-digital speckle interferometry makes
it possible to accumulate data as computer files
for comparing different states attendant to the
notch-crack length expansion and to the increase
of indicator hole size if necessary. It is worth
noting that compared with holography interferometry, electron-digital speckle interferometry
is about two times more sensitive to tangential
components of displacement. Noteworthy, also,
is the fact that when electron speckle interferometry is applied, displacement fields are registered directly in digital form, convenient for
further mathematical processing of results from
experimental procedures considered in previous
sections.
Residual Stress Investigations

Examples
Study of Residual Stress in the Flat Specimen Made of Hull Plate Steel Using the Volterra Integral Equation. Following is an example of determining residual stress in practice,
on the basis of the general approach presented
in the section Study of Residual Stress as Inverse Problem of Experimental Mechanics. A
flat specimen cut in a circular direction out of
the cylinder-shaped shell of a water-moderated
reactor pressure vessel was examined. The base
material of the pressure vessel was alloy steel.
The shell inner surface was faced with corrosionpreventive cladding 9 mm thick.
The calculation procedure is based on applying the Volterra integral equations (Eq 1 and 2).
Regular points A, B, C, and D close to angle
points and over the symmetry axis (Fig. 16) were
selected as typical measurement points when applying these equations.
Oriented to measurements made with the elastic coating method, the value smax was used on
the symmetry axis as the value to be measured,
and the value sxy in the vicinity of angle points.
Values B(x,e) obtained close to the notch tip are
actually independent from the notch length; that
is, B(x,e) B(x e), while derivatives with
respect to x and e coincide; that is, Bx(x,e)
Be(x,e). Conditionality of the system of linear
equations (Eq 3) depends on the kernel structure
determined by a measurement point chosen. To
obtain a better conditionality, it is rational to
choose a kernel with prevailing elements of the
main diagonal that corresponds to the closer vicinity of the notch tip (e.g., point A or B) (see
Fig. 16).
The method of elastic coating was applied in
measurement taking. The 1.8 mm coating of optically sensitive material based on epoxy rosin
(Ref 8, 24) was glued over the specimen. The
notch on the coated specimen was gradually expanded by 1 mm steps with the 0.5 mm diskmilling cutter. Optical measurements of fringe
orders m and isocline parameters u at typical
points (Fig. 17) that track the notch tip growth

were made with the reflection polariscope Photolastic Inc. Figure 17 gives the picture of the
typical fringe pattern in the notch tip zone for l
7 mm.
Figure 18 presents the dependency sA
max(x) obtained from processed data of optical measurements as well as the residual stress profile ry(x)
calculated using Eq 3.
Study of Residual Stress in Bimetal Using
the Crack-Indicator Method. As an example
of applying the indicator crack method in practice, results from the study of residual stress in
a flat specimen taken from bimetallic cylindrical
MPa
y
x
7h
100
max(x)
y(x)
max, y, MPa
200
x/h
casings of power installations are presented here.

The specimen was cut out of a full-fledged
template from the material used in the reactor
casing of a nuclear power plant. The cladding
(stainless steel 08Cr18Ni10Ti) was applied to
the base material of the vessel (low alloy steel
10CrNi1Mo) by explosion welding. Thickness
of the cladding layer was 7.5 mm. The manufacturing process for large-scale members from
such material envisages butt welding of bimetallic elements. The weld zone was studied for
residual stress (Fig. 19).
During this study, methods and procedures
(way of crack creating, method of determining
SIFs, and that of calculating residual stress) described in the section Indicator Crack Method
were used.
To determine values of SIFs in indicator
cracks, the photoelastic coating method was
used. Figure 20 gives typical fringe patterns that
emerge in the zones of edge indicator cracks,
these patterns being obtained when investigating
the first specimen.
Experiment results demonstrate that at all
measuring points, KII(xi) values make up less
than 7% of KI(xi) values. Hence, it follows that
r y(x) s xy(x).
100
Determining residual stress in the hull plate still
specimen with corrosion-preventive facing.
Data points indicate optical measurements in the photoelastic coating taken at notch-tracking points, with the
notch being expanded stepwise, through 1 m steps
Fig. 18
42
2
42
Fig. 20
Photograph of fringe patterns in zone of crackindicator tip (crack length is 12 mm)
1
50
40
(a)
30
r (x)
K I , MPa
20
10
0
6 8
10
12
100
200
300
400
y(x)
500
x , MPa
Fig. 19
Drawing of (a) weld zone and (b) photograph

of specimen surface
Results of residual stress study in the weld zone

of bimetallic material from nuclear power plant
reactor casing
Fig. 21
Principal results of the study are given in Fig.

21 as plots of relations KI(x) and residual stress
ry(x) for bimetals under consideration.
REFERENCES
1. A.N. Tikhonov and V.Y. Arsenin, Solutions
to Ill-Posed Problems, V.H. Winston &
Sons, Washington, D.C., 1977
2. J.V. Beck, B. Blackwell, and Ch.R. St.
Clair, Jr., Ill-Posed Problems of Heat, Wiley
Interscience, 1985
3. C.L. Lawson and R.J. Hanson, Solving
Least Squares Problems, Prentice-Hall, Englewood Cliffs, NJ, 1974
4. V.O. Sergeyev, Regularizing the Volterra
Equations of the First Type, Proceedings of
USSR Academy of Sciences, Vol 282 (No.
3), 1971, p 531534 (in Russian)
5. A. Vajdanatan and A. Finny, Determination
of Residual Stresses from Stress Intensity
Factors Measurements, Trans. ASME, Ser.
C, 1971, p 131135
6. G.R. Irvin, Fracture, Handbuch der Physik,
Springer (Berlin), Vol 6, 1958, p 551590
7. N.I. Muskhelishvili, Some Main Problems
of Mathematical Theory of Elasticity,
Nauka, Moscow, 1966 (in Russian)
8. A.S. Kobayashi, Ed., Handbook on Experimental Mechanics, Vol 2, Prentice-Hall
Inc., Englewood Cliffs, NJ, 1988
9. V.D. Kuliev and I.A. Razumovskii, Determination of Residual Stresses in Bimetalls,
Sov. Phys. Dokl. RAN (Proceedings of the
Russian Academy of Sciences), Vol 315
(No. 13), American Institute of Physics,
Nov 1990, p 985987
10. I.F. Obraztcov, V.D. Kuliev, and I.A. Razumovskii, Fracture of Bimetallic Materials
with an Edge Crack, Sov. Phys. Dokl. RAN
(Proceedings of the Russian Academy of
Sciences), Vol 308 (No. 13), American Institute of Physics, Sept 1989, p 859861
11. Y. Murakami, Ed., Stress Intensity Factors
Handbook (in 2 volumes), Pergamon Press,
1987
12. V.V. Panasyuk, Ed., Fracture Mechanics
and Strength of Material Handbook, Vol 2,
Kiev, Naukova Dumka, 1988, p 619 (in
Russian)
13. M.V. Medvedev and I.A. Razumovsky,
Methods for Studying Locally Nonuniform
Fields of Residual Stresses by Cutting the
Parts, J. Mach. Manuf. Reliab. (Russia), No.
5, 1998, p 7076
14. G.W. Smith, Use of Photoelasticity in Fracture Mechanics, Experimental Evaluation of
Stress Intensity Factors, Mech. Fracture,
Vol 7, G. Sih, Ed., The Haughe, 1981, p
163189
15. H.P. Rossmanith and R. Chona, Survey of
Recent Development in Evaluation Stress
Intensity Factors from Isochromatic Crack
Tip Patterns, Proc. Int. Conf. Fracture
Mech. ICEM (Cannes), 1981, p 348354
16. R.J. Sanford and J.W. Dally, A General
17.
18.
19.
20.
21.
22.
23.
Method for Determining Mixed-Mode

Stress Intensity Factors from Isochromatic
Fringe Patterns, Eng. Fract. Mech., Vol 11,
1978, p 621633
N.A. Makhutov and I.A. Razumovsky, Determination of Stress Intensity Factors for
Three-Dimensional Cracked Parts by Use of
Polarization Optic Methods/Testing Equipment for Experimental Investigations of
Mechanical Properties of Material and
Structures, Proc. Second Intern. Conf.
(Moscow), 1989, Vol 2, IMEKO, p 513
I.A. Razumovsky and M.V. Medvedev, Procedure of SIF Determination from the Patterns of Normal Displacement, Proc. Intern.
Society for Optical Eng., International Society for Optical Engineering, Vol 2791,
1995, p 129133
N.A. Makhutov, V.J. Kulijev, I.A. Razumovsky, and N.S. Cherpakova, Experimental Analytical Method for Investigating of
Local-Inhomogeneous Residual Stresses
with the Use of Edge Crack as Indicator,
Proc. Intern. Society for Optical Eng., International Society for Optical Engineering,
Vol 2791, 1995, p 115118
A. Zak and M. Williams, Crack Point Stress
Singularities at Bimaterial Interface, J.
Appl. Mech. (Trans. ASME), Vol 30 (No. 1),
1963, p 142143
F. Zandman, S. Redner, and J.W. Dally,
Photoelastic Coatings, SESA monograph,
No. 3, p 373
F. Zandman, Photoelastic Coatings Technique for Determination Stresses Distribution in Welded Structures, Weld. J., Vol 39
(No. 5), 1960, p 2329
M. Nisida and H. Tacabayashi, Thickness
24.
25.
26.
27.
28.
29.
30.
Effects in Hole Method and Application of

Method to Residual Stress Measurement,
Handbook on Experimental Mechanics, Vol
2, A.S. Kobayashi, Ed., Prentice-Hall, Inc.,
1988
A.Y. Alexandrov and M.Kh. Akhmetzyanov, Polarization-Optical Methods for
Mechanics of Strained Solids, Nauka, Moscow, 1973, p 572 (in Russian)
Residual Stress and Methods of Control,
Proc. of All-Union Symposium (Moscow),
Institute of Mechanics, USSR Academy of
Sciences, 1982, p 278 (in Russian)
Residual Technological Stress, Proc. of the
Second All-Union Symposium (Moscow),
Institute of Mechanics, USSR Academy of
Sciences, 1985, p 240 (in Russian)
I.A. Razumovsky, Photoelastic-Coating
Technique for Study of Zones with Large
Stress Gradients, Sov. Mach. Sci., Academy
of Sciences of the USSR, No. 2, 1984, p 80
84
V.G. Seleznev, A.N. Archipov, and T.V.
Ibragimov, Holography Interferometry in
Residual-Stress Determination, Ind. Lab.
(Russia), translated from Russian original,
(c/b Consultants Bureau New York and
London), Vol 42 (No. 6), 1976, p 976984
B.S. Kasatkin, A.B. Kudrin, and L.M. Lobanov, Experimental Methods for Deformation and Stresses Research, Kiev, Naukova Dumka, 1988, p 359 (in Russian)
L.M. Lobanov, B.S. Kasatkin, V.A. Pivtorak, and S.G. Andrrushenko, Determination
of Residual Stresses by Holographic Interferometry Using One Hologram, Sov. Phys.
Dokl. RAN (Proceedings of the Russian
Academy of Sciences), Vol 271 (No. 13),
31.
32.
33.
34.
35.
36.
37.
38.
Yule, American Institute of Physics, 1983,

p 555556
A.A. Rassokha, Technique Residual Stress
Investigation by Holographic and Speckle
Interferometry, Probl. Prochn., 163, 1983,
D. Nelson and J. McCrickerd, ResidualStress Determination through Combined
Use of Holographic Interferometry and
Blind-Hole Drilling, Exp. Mech., 26, 1986,
p 371378
L. Wang and J. Ke., The Measurement of
Residual Stress by Sandwich Holographic
Interferometry, Opt. Lasers Eng., No. 9,
1988, p 111119
V.S. Pisarev, V.P. Shepinov, and A.Yu. Shikanov, Reflectance Hologram Application
of Interferometers to Determine Residual
Stress by Probing Hole Method, J. Theoretical Phys., Vol 66 (No. 1), 1966, p 99
113 (in Russian)
Y.I. Ostrovskii, V.P. Shchepinov, and Yu.I.
Yakovlev, Holographic Interferometry in
Experimental Mechanics, Springer Series in
Optical Sciences, Vol 60, Springer-Verlag
(Berlin), Heilberg, 1991, p 248
V.P. Shchepinov and V.S. Pisarev, Strain
and Stress Analysis by Holographic and
Speckle Interferometry, Chichester: John
Wiley & Sons, 1996, p 496
R. Jones and C. Wykes, Holographic and
Speckle Interferometry, Cambridge University Press, 1984
A.A. Apalkov, I.N. Odintzev, and I.A. Razumovsky, Application of Speckle Interferometry for Residual Stress Measurement,
Ind. Lab. (Russia), translated from Russian
original, (c/b Consultants Bureau New
York and London), Vol 66, 2000

DOI: 10.1361/hrsd2002p141

Residual Stress in the Forming

of Materials
Z. Wang and B. Gong, University of Toronto, Canada
ALL METAL PRODUCTS, except casting

components, are subjected at some point in their
manufacturing process to at least one metalworking or metal forming operation. Such operations can be very diverse, but all have the same
primary objective: to produce a desired shape
change. Several different operations may often
be necessary; for example, a specific steel chosen
to make tubes for manufacturing golf clubs is
first hot forged, rolled, extruded into tubular
shape, and then cold drawn with specific dies
into final shape, apart from all the subsidiary
treatment.
Metal forming can be further divided into hot
forming and cold forming. This article addresses
cold forming and its resultant residual stresses.
Cold forming (working) is defined as a forming
process or plastic deformation operation carried
out at temperatures below the recrystallization
temperature of the workpiece material, but more
often it is simply referred to as the forming process at room or ambient temperature. The most
commonly used cold forming operations (drawing, extrusion, rolling, and forging, deep drawing, and pressing) are briefly discussed below.
Drawing. Large quantities of metal or alloy
products such as rods, tubes, and wires with various special sections are finished by the cold
drawing process.
Bars. Large bars, up to 150 mm (6 in.) in diameter or sometimes more, are frequently given
a light sizing pass, reducing the diameter by
about 1.5 mm (116 in.) to improve surface finish
and dimensional tolerance (Ref 1). Light reductions are also made in the production of brightdrawing angles, channels, and strips. Many of
the smaller-sized round bars, however, are drawn
with much greater reductions of area, even up to
50% per pass. The general configuration of dies
for rod and wire drawing is shown schematically
in Fig. 1. A theoretical approach can be used to
predict the maximum reduction of area possible
in a single pass. In drawing round bars, for instance, the limiting reduction would be as high
as about 63% (Ref 1).
Wires are usually drawn through a large number of dies before reaching the final size, and
may be reannealed several times as drawing proceeds. For wire drawing of high-carbon steel
(0.5 to 0.8 wt% C), for example, six or more

dies are needed. Power-driven capstans are provided between each die and the one ahead of it.
Final drawing speed often reaches 7.5 m/s
(1,500 ft/min). Dies are often made of tungsten
carbide. The reduction at each die in a range of
5 to 30% is determined by the size of the current die and also the preceding one, and can be
altered by changing one of the dies. The deformation in wire drawing is not uniform, causing
nonuniform hardness distribution along the radius direction. This phenomenon is particularly
severe in the last couple of dies. The higher the
die angle, the more severe the phenomenon (Ref
2). The total reduction after sequential passes
could be more than 90%, allowing the tensile
strength of the final wires to be doubled compared with that of the as-received rods (Ref 3).
Tubes. Most tubes are made by hot extrusion
to hollows, followed by one or multiple drawing
passes at room temperature. Several major cold
drawing processes are used for the production of
tubes, as shown diagrammatically in Fig. 2. In
the first three processes (Fig. 2a-c), the major
part of deformation is the reduction of wall
thickness. But it is also possible for the diameter
to be reduced, accompanied usually by a small
increase in wall thickness, by sinking the tube
without internal support (Fig. 2d). The heaviest
reduction in tube size may be reached with a
moving mandrel, because the friction at the inner
surface can also carry a portion of the drawing
load. Most tube drawing with a mandrel or plug
involves some reduction in internal diameter, if
only an initial clearance in the bore is allowed.
Extrusion. In this process a cylinder or billet
of metal is forced through an orifice by means
of a ram so that the elongated and extruded metal
has a transverse shape that coincides with the die
orifice. Cold extrusions are usually with the final
forming processes.
Low- and medium-carbon steels, low-alloy
steels, and stainless steels are the common materials that can be cold extruded. In cold extrusion, a punch applies pressure to the slug or preform, causing the work metal to flow in the
required direction (Fig. 3). The relative motion
between punch and die is obtained by attaching
either one (almost always the die) to the stationary bed and the other to the reciprocating ram.
The pressure can be applied rapidly by a sharp
blow, as in a crank press or header (impact extrusion), or more slowly by a squeezing action,
as in a hydraulic press.
Rolling is a process in which the thickness of
the material is reduced by passing between a pair
of revolving rollers, which are generally cylindrical in shape for producing flat products such
as sheets or strip. Their surfaces can also be
grooved or textured in order to change the profile
as well as the emboss pattern of the products.
Specifically, cold rolling is carried out for special purposes such as the production of good surface finish or special mechanical properties. According to statistics, more materials are rolled for
further manufacturing applications than are
treated by all other processes.
Forging, Deep Drawing, and Pressing.
Forging compresses metal between a hammer
and an anvil or in a pair of dies. Large-size materials such as castings and ingots are usually
forged at high temperatures at which the yield
strength is reduced as far as possible. Cold forging, often combined with cold extrusion, is mainly
used for smaller-size stock. Deep drawing is an
extension of pressing in that the metal blank is
given a substantial third dimension after flowing
through a die (Fig. 4). Simple pressing is carried
out by loading a blank between a punch and a
Fig. 1
Schematic of a converging die for rod and wire

drawing
142 / Residual Stress Formation in the Shaping of Materials

die so as to indent the blank and give the product
a measure of rigidity. Can ends in food and beverage containers are the most widespread examples.
Advantages. Overall, metal forming processes such as extrusion, drawing, rolling, and
forging offer many advantages compared with
other production engineering methods. The de-
formation is carried out at constant volume with

no loss of material in the form of chips. In the
case of cold forming, a more favorable grain
flow and dislocation structure are achieved,
which could enhance mechanical strength and
fatigue strength significantly. In particular, forming can improve the dimensional accuracy and
surface finish of metal.
Residual Stress in Metal Forming

When a metal is being plastically deformed,
internal stresses frequently are created due primarily to deformation incompatibilitythat is,
the different strain levels experienced in different locations at the same time. This difference in
the actual strain level in different locations may
be caused by multiple reasons, including (1) a
difference in strength between different but coexistent phases in the material, (2) different actual strains accommodated at different locations
due to die/mold shape or constraints from the
gripping force on the testpiece or workpiece, and
(3) a possible temperature gradient in different
locations. These internal stresses may remain in
materials after the deformation/forming processes in the form of residual stresses.
The most typical and simple example of the
effect of residual stresses on materials behavior
is the Bauschinger effect, a behavior reflecting
Fig. 4
Fig. 2
Tube elongation by drawing with internal support by (a) mandrel, (b) plug, (c) floating plug, and (d) without
internal support by sinking
Generation of residual stress in sheet bending. (a)

Geometry of a bent sheet. (b) Stress-strain relationship. (c) Distribution of residual stress through thickness
Fig. 5
Fig. 3
Schematic of deep drawing
Schematic of forward extrusion
Residual Stress in the Forming of Materials / 143

the effect of reversing the direction of straining
in a testpiece. If a metal specimen is plastically
deformed first in one directione.g., in tensionthe yield stress measured in the reversed
loadingi.e., in compressionis generally
lower than that measured in tension. Similarly,
an initial plastic compression produces a tensile
yield stress lower than the original compression
yield stress. Such an effect is due, in part, to the
internal stressesknown as residual stresses
developed during the initial deformation. In general, residual stresses can be defined as the selfequilibrating internal stresses existing in a free
body that has no external forces or constraints
on its boundary (Ref 4). As mentioned above,
these stresses arise from the elastic response of
the material to an inhomogeneous distribution of
nonelastic strains.
Residual stresses are usually subdivided into
macrostresses and microstresses, according to the
distances over which they show significant variation. The former vary continuously or smoothly
over distances of the order of the dimensions of
the body, and the latter show marked variations
in pattern and magnitude over distances of the
order of the dimensions of microstructural elements such as grains and particles. Both types
result from inhomogeneous deformation caused
by external forces. Note that in this article, residual stresses produced in cold forming processes are mainly referred to as macrostresses.
It is generally realized that the shape of the
deformation zone, which is the region undergoing plastic flow in the material, exerts a strong
influence upon the magnitude and distribution of
residual stresses. Deformation zone geometry for
forming processes such as drawing, extrusion,
and rolling can be characterized by a single parameter, D, defined as the ratio of the mean thickness or diameter, h, of the work metal to the
contact length between tool and work metal:
D
h
L
Bending with superimposed tension. With sufficient tension, the neutral axis moves out of the sheet so that the
strain is tensile across the entire section (a). With the stress-strain curve shown in (b), the stress distribution in
(c) results. After removal of the moment, elastic unloading leaves very minor residual stresses, as shown in (d). Source:
Ref 5
Fig. 6
(Eq 1)
For drawing or extrusion, according to Fig. 1,

the contact length, L, is equal to (h0 h1)/(2sin)
and the mean thickness or diameter, h, is equal
to (h0 h1)/2. Thus,
D [(h0 h1) sin]/(h0 h1)
(Eq 2)
For axisymmetrical drawing, where h is the diameter, the reduction is given by r (d 20

d 2f )/d 20. Hence,
D
sin (1 (1 r)1/2)2
r
(Eq 3)
Similarly, one can demonstrate that for flat rolling,

D
(2 r)(h0 /rR)1/2
r
(Eq 4)
In the formulas for drawing and extrusion, D increases with decreasing reduction and increasing
Fig. 7
Residual stresses in lightly drawn steel rods. Source: Ref 5

die angle; in the case of rolling, it increases with
ratio of the strip thickness to the roller radius
(Ref 5).
Nearly all cold-formed products contain residual stresses. In most cases, these residual stresses
are undesirable because they often lower the
elastic limit of the material and may cause warpage during subsequent machining operations.
The usual pattern of residual stresses with the
surface in tension are particularly undesirable.
Once a product with such a residual stress state
is put in service under a certain external load, the
nominal applied stress would then be superimposed onto these residual stresses of tension
sense on the surface and in the subsurface. Hence,
the real stress on the surface and in the subsurface would become higher than the nominal applied stress leading to a significant reduction in
fatigue life of the product. Furthermore, even if
without such a superposition of stresses, surface
residual stresses alone would increase the susceptibility to stress corrosion.
In practice, these residual stresses can be eliminated by heat treatment such as annealing or
stress relieving. However, these treatments are
costly and time consuming because of the extra
operation; more importantly, heat treatment may
considerably soften the work-hardened materials. From the point of view of forming, on the
other hand, it is possible to reduce the residual
stress or even to produce useful residual stress
by optimizing forming parameters to control the
deformation zone. Furthermore, for some processes, an additional deformation can also reduce residual stresses significantly. This will be
discussed in later sections on wire drawing.
Nevertheless, it is of utmost importance to understand the distribution and magnitude of residual stresses in cold forming products. There are
usually two major approaches to the study of residual stresses: direct experimental measurements and theoretical simulations. Experimental
approaches such as hole-drilling, x-ray, and neutron diffraction methods usually give reliable
and even straightforward results. However, hole
drilling requires destructive testing, and the latter
two methods are expensive and time consuming.
In recent years, the finite-element method (FEM)
has received considerable attention and has
proved a powerful tool for the analysis of metal
forming processes, including residual stress
analysis. In the following section, the generation
of residual stress due to inhomogeneous deformation is illustrated by a few simple examples.
Residual stress distributions studied either experimentally or theoretically in workpieces under various cold forming processes are presented
and discussed in detail.
can be readily demonstrated by the example of

the plastic bending of a piece of sheet. Figure
5(a) represents a sheet bent in the manner shown,
where the subject sheet is assumed to be composed of many contact thin layers stacked along
the thickness direction. The stress is a maximum

in tension at the upper surface of the sheet, layer
5. It decreases to zero at the neutral layer 0 and
increases to a maximum in compression at the
lower surface. If the stress does not exceed the
Fig. 8
Influence of area reduction on residual stresses after drawing with one die. Source: Ref 6
Fig. 9
Residual stresses after drawing with different area reduction in the second die. Source: Ref 6
Residual Stresses in Specific Cold

Forming Processes
Bending of Sheet
The development of a system of internal stress
as the result of inhomogeneous plastic straining

yield strength of the materials at any point, removal of the applied bending force will permit
the sheet to return to its original shape and thus
no internal stress can remain. If, however, the
maximum stress exceeds that for yielding, then
the layers will extend or be compressed by an
amount that depends on the extent to which the
yield stress is exceeded. This amount is greatest
at the surfaces and decreases in the layers nearer
the neutral axis.
While the bending load is still applied, the
stress distribution across the thickness of the
sheet will be as represented by Fig. 5(b), with
layers 2 to 5 being on the plastic portion and
layer 1 on the elastic portion of the stress-strain
curve in tension. If the bending moment is now
released, the beam will spring back toward its
original flat profile, but not entirely, since this
would require the complete reversal of both elastic and plastic strain. As a matter of fact, the
amount of springback will be very different
along the thickness direction of the sheet because
of different plastic deformation. Suppose layer 5
is isolated from layer 4; in this case, the unloading process of the layer will follow the dashed
line of point 5 in Fig. 5(b). Considering the interaction of layer 5 with layer 4, the unloading
path of point 5 is actually along the solid line,
since layer 4 has less plastic deformation or displacement than layer 5 and thus tends to resist
the plastic extension of layer 5. As a result, layer
5 feels compression; the compression stress level
would be DeE, where De is the residual strain
shown in Fig. 5(b) and E is the Youngs modulus
of the material. For layer 4, the situation is a bit
complicated as layer 3 will resist the extension
of layer 4, but layer 5 promotes the extension.
The combined result will produce less compression residual strain. It is predictable that at a certain layer, these two influencing factors will become balanced, resulting in zero residual stress
(strain) state, as shown by layer 3 in Fig. 5(b).
Likewise, layers below the zero residual stress
layer, such as layers 2 and 1, could feel extension.
The overall pattern of the residual stress distribution is shown in Fig. 5(c). Two important
points should be noted here: First, for equilibrium, the effect of tension must be balanced by
that of compression; second, the surface that
originally plastically extended finally remains in
compression, while that which was plastically
compressed is left in tension.
This analysis can also be applied to the development of residual macrostresses in workpieces inhomogeneously deformed/formed by
other processing methods. For instance, rolling
sheet with insufficient camber produces material
with thin edges, so that, since lateral spread is
relatively difficult, the edges must be elongated
in the direction of rolling by a greater amount
than the central fibers of the sheet. The longitudinal residual stress should therefore be compressive near the edges, with the central fibers in
balancing tension. A somewhat similar residual
stress distributioncompressive at the surface
and tension in the central regionsmay be ex-
pected in cold-drawn rods due to the greater

elongation in the surface layers when compared
with those near the center. In reality, the problem
is much more complex than what is presented
above. Residual stresses are rarely unidirectional, and the inhomogeneity of the deformation
usually produces a bi- or triaxial stress system.
Springback in the case of sheet bending would
not only create severe residual stresses but also
cause problems in tool design. Fortunately, this
kind of problem often can be avoided. Taking
stretch forming as an example, the tooling does
not apply a pure bending moment as assumed
above. Rather, tension is applied simultaneously
with bending. With increasing tensile forces, the
neutral plane shifts toward the inside of the bend,
and in most operations this tension is sufficient
to move the neutral plane completely out of the
sheet so that the entire cross section yields in
tension. For such a case, the strain and stress
distributions are shown in Fig. 6, leaving very
minor residual stresses.
Drawing of Wire, Rod, and Tube

The nature of residual stresses introduced by
drawing strongly depends upon the shape of the
deformation zone (D parameter), as discussed
previously. With D-values equal to 1 or lower
(i.e., with high reduction and small die angle),
the flow pattern of the materials or the deformation is relatively uniform and the residual
stresses created are minor. In general, as D is
increased to values above unity, the surface of
the product wire is left in residual tension (axial
Fig. 10
direction stress) and the center in compression.

The magnitude of the residual stresses usually
increases with increasing D. However, the typical pattern of residual stresses may be completely different if D is very largethat is, if the
reduction is so small that the actual deformation
zone does not penetrate to a considerable depth
of the wire interior. In that case, the surface can
be left in residual compression, as shown in Fig.
7 for lightly drawn steel rods (Ref 5).
The residual stress in cold-drawn steel rods
and wires can now be well described by using
FEM verified by experimental measurements
(Ref 6, 7). Specifically, residual stresses in single-pass drawing have been found to be not only
dependent on the reduction in area and die angle,
but also affected by a light deformation through
an additional die immediately after the preceding
one. As an example, Fig. 8 shows the influence
of the value of area reduction (eA), after drawing
with one single die, on the four components of
the residual stressnamely, axial, radial, tangential, and equivalent stresses. Parameters and conditions for this analysis are as follows:
Material
Die angle
Final rod diameter
Area reduction
CK15 steel
2 25
d1 15mm
eA
Take eA 18% as an example. Under this

condition, the axial residual stress in the core
section is in the compression condition, while
residual stress in tension can be seen in the sur-
Residual stresses after drawing, with three different distances between the two dies. Source: Ref 6

face region. A lower reduction (eA 10%) results in a higher residual stress on the surface,
but a lower one in the core. Along the radial
direction, all residual stresses are in the compression condition, being highest at the core and
close to zero at the surface. Figure 9 shows the
influence of the area reduction in a second die
on the residual stress distribution. The area reduction in the first die is 10%, the distance between the two dies is 5 mm (0.02 in.), and the
die angle is 2 25. The small deformation of
only eA2 0.6 to 1% in the second die results
in an effective reduction in stress levels, especially for axial and equivalent residual stresses.
Figure 10 demonstrates the influence of distance between the main die and the second die
(5, 13, and 21 mm, or 0.2, 0.5, and 0.8 in., respectively) on the residual stresses for an initial
wire diameter of 16 mm (0.6 in.). The area reductions in the first and the second dies are both
constant (eA1 10% and 1%, respectively). The
stress distributions for the two inter-die distances, 13 and 21 mm (0.5 and 0.8 in.), are very
similar. However, the equivalent residual stress
for the 5 mm (0.2 in.) distance is much lower.
This is because the plastic deformation zones in
these two dies are connected due to the small
distance between the two dies, and thus the inhomogeneous deformation pattern produced by
the primary die is not only affected but also controlled completely by the secondary die (Ref 6).
The residual stress pattern in multistage colddrawn steel wire is quite similar to that of singledie drawing. Figure 11 shows the distribution of
longitudinal residual stresses in 1H18N9 wire, a
high-chromium, high-nickel alloy steel, from
center to surface (Ref 8). Chemical composition
of the steel is similar to that for 300 series stainless steels according to the SAE specification
(0.12% C, 9.0% Ni, 2.0% Mn, 0.3% Mo, 18.0%
Cr, 0.3% W, 0.15% V). The 1H18N9 steel wire
was drawn through six or eight passes from 2.0
to 1.0 mm (0.08 to 0.04 in.), with a total reduction in area of 75%. The tensile strength after
drawing is 1540 MPa (220 ksi), and the yield
stress is about 1380 MPa (200 ksi). Residual
stresses, measured using the electrolytic etching
method, were very high (close to the yield stress)
both at the wire surface and in the interior. To
relieve these high residual stresses, two destressing operations were tried: multiroll straightening
and additional light deformation drawing
(3.8%). Figure 11 also shows the influence of
these two methods. It is apparent that additional
light drawing can significantly reduce residual
stresses.
The pattern of residual stresses in cold-drawn
tubes, on the other hand, seems sensitive only to
the area reduction. Figure 12 shows axial and
tangential residual stresses at the outer surface
of tubes before and after cold drawing (Ref 9).
The tubing was made from ferritic steel (DIN St
35) with an initial outer diameter of 28 mm (1.1
in.) and a wall thickness of about 3.5 to 4 mm

(0.14 to 0.16 in.). The tool used for drawing consisted of a die and a mandrel (see Fig. 2a). The
reduction in area is represented by a nature
strain, defined as u ln(A0 /A1), where A0 and
A1 are the initial and final cross-sectional areas,
respectively. Residual stresses were measured
using the x-ray technique and calculated from
the {200} and {211} reflections. Before deformation, the residual stress state at the outer surface is compressive and of rotational symmetry.
As a result of the plastic processing during cold
drawing (u 0.23), both the axial and tangential residual stresses shift toward the tensile direction. With further increasing in values of u,
they are reduced in both directions. When u
0.34, even compressive stresses, although small
in magnitude, occur.
Extrusion and Rolling

Similar to the case of the drawing process, residual stresses produced by extrusion and rolling
processes are strongly dependent on deformation
zone shape. Figure 13 compares residual stresses
created by open-die extrusion without ejection
and by the single-die drawing process, and
shows very similar patterns and magnitudes of
residual stresses upon the same amount of forming ratio. If the extrusion is rapid enough, temperature gradients caused by inhomogeneous deformation may cause an added effect upon
subsequent cooling. Nevertheless, this is not the
major cause of residual stresses (Ref 5). The ef-
Influence of destressing in multiroll straightening and additional drawing (3.8%) on the

distribution of longitudinal residual stresses in 1H18N9
steel wire. Deformation, 75% (six passes). Source: Ref 8
Fig. 11
Fig. 12
Residual stresses at the outer surface of the

tubes. Source: Ref 9
Fig. 13
Comparison between residual stresses after open-die extrusion without ejection and after drawing with one
die. Source: Ref 6

fect of die angle, reduction in area, and deformation zone parameter D on surface residual
stresses in extruded steel rods is shown in Fig.
14. The numbers marked in the figure are the
axial residual stress values on the surface in kg/
mm2, and the solid lines represent the constant
D-lines. Apparently, higher die angles and lower
extrusion rates produce higher residual stresses.
The residual stress pattern within the kernel of
the cold extruded rods, on the other hand, seems
dependent only on area reduction (Ref 9).
Residual stresses at the surface of rolled strip
are shown in Fig. 15 (Ref 5). The residual
stresses are normalized by the yield strength in
the figure, and the abscissa is D2. With the increase of D, residual stress goes up and could
even reach the yield stress of the material. High
residual stresses, coupled with damage at the
centerline, can result in spontaneous splitting or
alligatoring of the work material as the material leaves the deformation zone (Ref 5). This
type of failure is most likely in early breakdown
rolling of ingots because the section thickness is
usually large relative to the roll diameter, while
the reduction per pass is low. Furthermore, the
ingot structure will not have benefited from prior
deformation.
Residual Stresses in Cold-Formed Steel

Members
Steel elements (members) produced by cold
forming are a valuable alternative to hot-rolled
sections. They offer weight savings because of
their wall thinness, as well as a greater variety
of shapes. The cold forming processes commonly used for steel members are rolling and
press braking. The latter is a procedure in which
sheet metal is bent to a desired angle. Rolling,
used mainly for large-section series because it is
continuous, increases both yield strength and ultimate strength. Press braking is economical for
a limited series of section, the length of which
is restricted by the length of the press machine
itself.
As a typical example, the distribution of residual stresses in steel members formed by press
braking and rolling methods, respectively, are
shown in Fig. 16(a) and (b) (Ref 10). Tensile
strength for the rolled steel member is 480 MPa
(70 ksi) and for the press-braked member is 350
MPa (50 ksi). At the top of Fig. 16 are geometric
locations for residual stress measurement, which
was performed using electric discharge machining (EDM). It should be noted that a measured
negative residual strain corresponds to a tension
residual stress, and a measured positive residual
strain corresponds to a compression residual

stress. The figure provides following information:
1. Compression residual stresses (positive residual strain) were found on the inside surface
of the sections, and tension residual stresses
(negative residual strain) were found on the
outside surface.
2. The magnitudes of the surface residual
stresses of the sections ranged between ap-
Fig. 17
Residual stresses in a sunk tube of mild steel.

Source: Ref 11
Fig. 18
Residual stresses at the middle height of a cup

wall. Source: Ref 11
Fig. 14 Effect of die angle and reduction on residual

stresses in extruded steel rods. Numbers are the
axial residual stress on the surface in kg/mm2, and the constant D-lines are calculated from Eq 4. A0 and A1 are the
initial and final cross-sectional areas, respectively. Source:
Ref 5
Residual stresses at the surface of rolled strip.

The residual stresses are normalized by the
yield strength, and the abscissa is D2. Source: Ref 5
Fig. 15
Fig. 16
Residual strain. (a) Press braking. (b) Rolling.

Source: Ref 10

proximately 25 and 70% of the yield stress
of the material.
3. The magnitudes of the residual stresses on the
flat portions of the section were found to be
approximately uniform along the section perimeter.
4. The magnitudes of the residual stresses on the
corner regions were higher than those on the
flat portionsup to 30% of the yield stress
of the material.
5. At the same location, the magnitudes of the
residual stresses on the inside and outside
surfaces of the flat portions of the section
were quite close.
The fact that residual stress in the corner regions was higher than in the flat portions is understandable, since more and nonuniform cold
work was introduced in the corner regions. Thus,
the yield stress at corners increased, which in
turn increased residual stress.
The residual stress pattern and magnitudes in
cold-formed steel members are quite different
from those in hot-rolled shapes, where the maximum residual stress is about 30% of the yield
stress of the material and the residual stresses are
assumed to be uniformly distributed throughout
the plate thickness. Therefore, the direct application of the design standard for flexural buckling strength of cold-formed steel columns used
Table 1
Original tube
Die
Lubricant
Drawing speed
in the AISI specification (1986) would not be

appropriate, because this standard was based on
the residual stresses measured in hot-rolled steel
shapes (Ref 10).
Residual Stresses in Deep-Drawn Cups,

Sunk Tubes, and Radial Forging
Products
Condition
0.15% C steel; outside diam, 30 mm

(1.2 in.), wall thickness, 1 mm (0.04 in.)
Die hold diam, 24 mm; half-die angle,
20; die profile radius at the exit,
20 mm (0.08 in.)
Machine oil mixed with MoS2
1 m/min (3.3 ft/min)
Fig. 20
Residual stress distribution on the outer surface

of the workpiece. Source: Ref 12
Fig. 21
Residual stress distribution on the inner surface

of the workpiece. Source: Ref 12
Fig. 22
Residual stress distribution with respect to

thickness. Source: Ref 12
Deep drawing conditions
Parameter
Value
Die throat diam

Punch diam
Blank diam
Blank holding force
Die profile radius
Punch profile radius
Blank thickness
Maximum punch load
Table 3
ditions
Schematic of the dies used in a radial forging machine. Source: Ref 12
Tube sinking conditions
Element/parameter
Table 2
Fig. 19
66.4 mm (2.6 in.)

64.0 mm (2.5 in.)
120 mm (4.7 in.)
5 tons
3 mm (0.12 in.)
3 mm (0.12 in.)
1 mm (0.04 in.)
9 tons
Forging geometry and process con-
Geometry
Workpiece geometry
Outside radius
Inside radius
Mandrel radius
Outside radius of forged tube
Reduction in area
Die configuration
Length of die land
Die inlet angle
Process conditions
Friction
Friction
Contact stiffness
Value
33 mm (1.3 in.)
7.94 mm (0.31 in.)
7.94 mm (0.31 in.)
30 mm (1.2 in.)
18.3%
12.7 mm (0.5 in.)
6
Rigid Coulomb friction
0.05
4.0 106 N mm2
The residual stresses in deep-drawn cups and

sunk tubes are related to season cracking phenomena that occur in cylindrical products. The
cracks usually appear in the longitudinal direction. Therefore, the cause of cracking must be
circumferential residual stress or hoop stress in
the tube walls. The residual stresses in these
kinds of products are considered to be caused by
the axial unbending of the wall at the die exit
(see Fig. 4). To demonstrate the relation, Fig. 17
shows the residual stress distribution through the
wall thickness of a sunk, low-carbon steel
(0.15% C) tube (Ref 11). The tube sinking conditions are listed in Table 1. The yield stress of
the sunk tube is 570 MPa (80 ksi). The circles
and squares in the figure are measured residual
stresses obtained using the Sachs method. The
solid and dashed lines are from numerical analysis, and r*
1 and r*
t represent residual stresses
in the longitudinal and circumferential directions, respectively. The experimental and theoretical methods give a very close result. The residual stress pattern and magnitude are very
similar in the longitudinal and circumferential
directions. At the inner surface, the residual
stress in compression is rather severe, while at
the outer surface it is relatively lower in tension.
The residual stress distribution in a colddrawn cup is similar to that in the sunk tube case.
Figure 18 plots the residual stress distribution at
the middle height of a cold-drawn cup of mild
steel (0.23% C); the corresponding deep drawing
conditions are given in Table 2. The notations in
the figure have the same meaning as those in Fig.
17 (Ref 11).
Radial forging is a cost-effective and materialsaving forming process for reducing the cross
sections of rods, tubes, and shafts. The advan-

tages of radial forging are smooth surface finish,
considerable material or weight saving, preferred
fiber structure, minimum notch effect, and increased material strength. Components processed by radial forging generally have higher
residual stresses and thus undergo some deflections after forging. The residual stress distribution in high-strength metal is of great importance
since its presence, if above a particular level, can
shorten the service life of critical components
used under severe conditions.
A hollow shaft is an example of a common
machine element manufactured by radial forging. A radial forging die is shown schematically
in Fig. 19. The residual stress distribution in a
steel tube (MIL-S-1195) produced by cold forging has been simulated using a three-dimensional FEM model (Ref 12). The forging geometry and parameters are listed in Table 3.
Figures 20 and 21, respectively, show the stress
distributions at the outer and the inner surfaces
along the axial direction, from which it can be
seen that the outer surface was dominated by tensile residual stresses with maximum value at the
rear part of the workpiece. The axial stresses experienced a minimum (48 MPa, or 7 ksi) at
the front part of the workpiece and increased

along the axial direction to the maximum stress
(300 MPa, or 45 ksi) at the rear part of the workpiece. However, the inner-surface stresses consisted mostly of compressive residual stresses,
with only small tensile stresses at the front part
of the workpiece. Figure 22 shows the residual
stresses as a function of workpiece thickness.
The axial stresses varied greatly between the inner and outer surfaces, but the radial and hoop
stresses did not change significantly with thickness. The maximum tensile stress occurred at the
outer surface of the workpiece along the axial
direction.
REFERENCES
1. G.W. Rowe, Principles of Industrial Metalworking Processes, Edward Arnold Ltd.,
London, 1977
2. E.P. Riley-Gledhill, Recent Development in
Wire Making, Steel Times, Vol 214 (No. 1),
1986, p 12, 14
3. B. Gong and Z. Wang, Investigation of
Cold-Drawing ProcessMechanical Prop-
4.
5.
6.
7.
8.
9.
10.
11.
12.
erty Relationship of High C Steel Wires,

unpublished technical report, University of
Toronto
T. Mura, Micro-Mechanics of Defects in
Solids, Martinus Nihoff, The Hague, 1982
W.F. Hosford and R.M. Caddell, Metal
Forming, Prentice Hall, 1993
J. Gerhardt and A.E. Tekkaya, Advanced
Technology of Plasticity, Vol II, SpringerVerlag, 1987, p 841
J. Gerhardt and A.E. Tekkaya, in Residual
Stress in Science and Technology, E. Macherauch and V. Hauk, Ed., DGM Informationsgesellschaft, 1987, p 875
A. Skolyszewski, J. Luksza, and M. Packo,
J. Mater. Process. Technol., Vol 60, 1996,
p 155160
C. Genzel, W. Wreimers, R. Malek, and K.
Pohlandt, Mater. Sci. Eng., Vol 205A, 1996,
p 7990
C.C. Weng and T. Pekoz, J. Struct. Eng.,
Vol 116, 1990, p 16111625
K. Saito and Y. Shimahashi, in Metal Forming Plasticity Symposium, H. Lippman, Ed.,
Springer-Verlag, 1979
D.Y. Jang and J.H. Liou, J. Mater. Process.
Technol., Vol 74, 1998, p 7482

DOI: 10.1361/hrsd2002p150

The Effect of Final Shaping

Prior to Heat Treatment
T. Ericsson, Linkoping University, Sweden
Final Shaping Processes

Grinding is a chip-forming machining
method. Many similarities when it comes to residual stress creation exist between grinding,
milling, turning, and planing (Ref 1). The creation of residual stresses by grinding has been
treated in many studies, (Ref 17). The cutting
energy used in the machining process results in

temperature increase and plastic deformation.
Three factors that can cause residual stresses can
be distinguished:
Plastic deformation involving a smearing of
the material in the plane of the surface tends

to give compressive residual stresses.
Temperature increase that momentarily
causes an expansion, which is constrained by
the bulk material. The resulting thermal
stresses may exceed the yield stress at the actual temperature and, therefore, the surface
material will be upset. During the subsequent
cooling, tensile residual stresses are created.
If the workpiece is made of hardenable steel,
martensite may form due to the rapid heating
and cooling, causing compressive residual
stresses.
The total resulting residual stress depends on
the balance between these three factors (Fig. 1).
The residual stress is given as a function of
grinding power, but it could equally well have
been described as machining power. The higher
the grinding power, the more dominant the tensile residual stress contribution rthermal due to
factor 2 will be. For the highest powers, martensitic hardening may occur, shifting the residual
stress in the compressive direction.
The residual stress creation during grinding
has been modeled mathematically (Ref 4, 7). Important parameters are the cutting speed; the
depth of cut and the feed rate, which results in
cutting forces; the mechanical properties of the
workpiece; and the heat conductivity of the
grinding wheel and the workpiece. The importance of the latter is well illustrated when comparing the residual stresses after grinding with
an alumina wheel and cubic boron nitride (CBN)
wheel, respectively. The much better heat conductivity in the CBN wheel causes less temperature increase and, hence, more compressive residual stresses (Ref 4).
Examples of residual stress depth profiles are
given in Fig. 2 for a 0.45% C steel in two heat
treatment conditions and in Fig. 3 for the steel
AISI 1055. Typically, the affected depth is relatively shallow, usually considerably less than
100 lm. The magnitude depends on the material

and grinding conditions. In principle it should be
less than the yield strength of the affected material. For AISI 5100, values between 600 MPa
(87 ksi) in tension and compression have been
reported (Ref 9).
Milling. In contrast to grinding, there is a
well-defined cutting geometry in milling. Similar to grinding, a particular tool is not continuously in action. Down-cut milling, up-cut milling, and face-milling are distinguishable (Fig. 4).
In all cases the residual stresses are due to the
balance between the three factors mentioned previously. The mechanical process usually is divided into chip formation and smearing of the
workpiece surface. Chip formation (i.e., metal
separation) tends to give tensile residual stresses
while smearing (i.e., plastic deformation) of the
surface tends to give compressive ones. Most of
the generated heat is stored in the chip and is
removed from the workpiece. Factor 2, the tem-
therm
Residual stress
FINAL SHAPING OF PARTS, including

operations such as grinding, cutting, milling,
turning, drilling, and processes such as shot
peening and blasting, involves local plastic deformation. This means that they invariably are
accompanied by residual stress formation. The
residual stress can be compressive or tensile and
at times even have large shear components. The
primary concern regarding residual stresses has
been the effect on the mechanical properties
in particular, fatigue, contact fatigue, wear, and
stress-corrosion cracking (SCC). However, the
presence of residual stresses also means the presence of elastic deformation, although usually on
a very small scale. If this deformation is included
in the final shape, distortion may result if the
residual stresses are released by, for instance,
heat treatment. The degree of distortion depends
on the magnitude and geometric extension of the
residual stresses. A highly stressed but thin surface layer is less disturbing than the presence of
a large zone with smaller residual stresses.
Therefore, it is also important to consider residual stresses from preceding operations such as
straightening. This article first discusses the
mechanism for residual stress formation due to
the most important final shaping operations and
typical examples of residual stress distributions.
This is an area that is fairly well understood and
for which mathematical models for calculating
the residual stress are becoming available. Next,
the effect on shape of relaxing the residual
stresses, a less-studied topic, is discussed. Only
for applications with extreme requirements of
shape precision has this been done. Examples
based on published work about bearing rings and
turbine blades are presented. Recent experimental studies of straightening of bars also are reviewed.
0
res
mech
Grinding power
Contributions to the residual stress after grinding
as a function of the grinding power. rtherm is the
thermoelastic part, rmech is the mechanical part, and rres is
the total residual stress. In region a, only elastic deformation occurs; in region b, thermoplastic deformation; in region c, thermoplastic plus thermomechanical deformation;
and in region d, rehardening occurs. Source: Ref 2
Fig. 1
The Effect of Final Shaping Prior to Heat Treatment / 151

400
Residual stress, N/mm2
200
200
Ck 45 Surface ground with CBN
as-quenched 700 HV10
To the grinding
Parallel
direction
Transverse
400
600
Quenched and tempered 420 HV10
1000
To the grinding
direction
Parallel
Transverse
800
25
50
75
100
125
150
Distance from surface, m
Fig. 2
Residual stress distributions in the longitudinal and transverse directions after CBN grinding of steel SAE 1045
(Ck 45) in two heat treatment conditions: as-quenched and quenched and tempered. Source: Ref 5
perature effect, is favored in down-cut milling

and face milling with inclined cutting axis and
factor 1, the smearing effect, in up-cut milling
and face milling with perpendicular axis (Ref 1).
The depth of the affected layer is usually thicker
than in grinding. Typical examples are shown in
Fig. 5(a) and (b) for down-cut milling and upcut milling of recrystallized 0.45% C steel.
Stress grinding direction, MPa
500
Neutron
X-ray
400
300
200
100
0
100
100
100
200
300
Depth, m
400
500
Stress cross direction, MPa
500
Neutron
X-ray
400
300
200
100
0
100
100
100
200
300
Depth, m
400
500
The residual stress distributions in the longitudinal and transverse directions after grinding of
steel AISI 1055 using grinding wheel speed 40 m/s and
depth 3.8 lm. Both neutron and x-ray diffraction were
used. Source: Ref 8
Fig. 3
Turning. A difference between milling and

turning is that the tool is in continuous cut in
turning. In turning, tensile stresses often are
found near the surface, and compressive stresses
appear deeper into the material (Fig. 6). The
stress distribution is affected by friction in the
cutting process, which tends to cause tensile
stresses in the surface near layer and by plastic
deformation, which causes compressive stresses
in the deeper layers. Figure 7 shows the residual
stress distribution after face grooving (turning) a
cylindrical bar of the nickel-base alloy IN718
(Ref 10). The affected depth after turning is
larger than after grinding; up to about 200 lm
can be found after turning. Computer models
have been developed for turning (Ref 11).
Shot Peening. The primary objective of shot
peening is not to shape the material but to
strengthen it. In shot peening, shots made of
steel, ceramics, glass, or other materials are
thrown on the steel surface by an air jet or centrifugal forces. It results in a plastic deformation
of the surface, which gives rise to residual compressive stresses. Depending on the size (diameter), hardness, density, and speed of the shots;
the hardness of the workpiece; and the coverage
of the surface by shots, the magnitude and depth
distribution of the residual stress will vary. The
process is well understood, and fairly accurate
methods are used to calculate the residual stress
by FEA (finite-element analysis) (Ref 1215).
However, it is more instructive to give a qualitative explanation based on Fig. 8 (Ref 16).
The residual stress field is caused by two parallel processes. One process is the formation of
a Hertzian pressure due to the vertical forces
connected with the impacts of spherical shot
balls. As indicated on the left side of Fig. 8, the
resulting shear stress has a maximum at a certain
depth. If the pressure is high enough, plastic
elongation will occur and, therefore, compressive residual stresses will be produced. The other
process is a direct plastic deformation of the surface as indicated on the right side of Fig. 8. Thus,
plastic elongation of the surface layers occurs.
The former effect is more pronounced for a hard
workpiece material, and the latter for a soft one.
A few examples of residual stress profiles are
shown in Fig. 9(a) and (b). It is obvious that the
depth of the affected layer increases with the radius and speed of the shot balls. Depths approximately equal to the ball diameter have been reported in steel. The magnitude of the stresses is
less than the yield stress. Work hardening can
cause the residual stresses to exceed the yield
stress of the unpeened material. The largest compressive residual stresses can be achieved by
strain peening when the workpiece is strained in
tension during the peening.
Shot peening also causes shape changes,
which can be considerable for thin sections. This
effect is used in a manufacturing process known
as peen forming in the aeronautical industry. It
also is used to monitor the shot-peening intensity
by help of so-called Almen strips. The bending
of standardized steel strips then is measured.
Pressure rolling is another mechanical
strengthening process with some similarities to
shot peening. It usually is applied to workpieces
with cylindrical symmetry. A roll is pressed
against the workpiece to obtain high-compressive residual stresses. Very deep affected zones
can be obtained, more than 1 mm (0.04 in.) (Fig.
10).
Straightening of bars, tubes, and plates often
is carried out to meet the requirements of
straightness tolerances. The methods for
straightening are described in Ref 17 and 18. Examples include stretching, pressing, rolling,
peening, heating, or other deformation methods.
These procedures involve plastic deformation in
local areas of the part and bring about the generation of residual stresses and sometimes work
hardening of the steel. Figure 11 illustrates how
after bending partially into the plastic regime,
residual stresses are generated due to elastic recovery. During subsequent machining and/or
heat treatment, the residual stresses can become
released and cause distortions. Normally, the
plastic deformation goes fairly deep into the material. The most common straightening methods
depend on bending and usually are carried out
in several small steps, either by pressing or rolling. Stretching of bars and shapes is less frequent
because it is difficult to get the same productivity
as for bending (Ref 19). However, stretching is
attractive because it introduces less residual
stresses than bending does. For strips, a combination of bending and stretching frequently is
applied.
Distortion after Final Shaping

After this survey of residual stress effects
caused by important final shaping operations,
distortion, which can arise during subsequent
heat treatment if the residual stresses relax completely or partially, should be considered. The

thickness of the surface layer with residual stress
and the magnitude of the stresses are then decisive. To estimate if any noticeable effect is to be
expected, Stoneys formula (Ref 20) can be used.
This formula permits calculation of the bending

of a strip due to a surface layer with a coating
with residual stress. It is in analogy with the Almen strips:
Face-milling
Residual stress
Down-cut milling
Cutting axis
perpendicular
Up-cut milling
Cutting axis
perpendicular
6 1 m h
h2sub
film
1
R
where r is the stress in the surface layer, E is

the elastic constant, m is the Poisson constant,
hsub is the thickness of the substrate, hfilm is the
thickness of the surface layer, and R is the radius
of curvature of the strip. For the present application, it is of interest to calculate the bending
out d of a strip of length 2l (Fig. 12). It is similar
to the bending of an Almen strip used to measure
the shot-peening intensity. For small d it holds
that:
d 2R 2
+
r
r
Distance from surface
Fig. 4
Characteristic residual stress distributions in steels due to different milling processes. Longitudinal (solid line)
and transverse (dashed line) residual stresses are shown. Source: Ref 1
400
400
200
Longitudinal
0
Transverse
200
400
0
Fig. 5
SAE1045, normalized
Up-cut milling
SAE1045, normalized
0.02
0.04
0.06
Distance from the surface, mm
Down-cut milling
200
Longitudinal
0
Transverse
200
400
0
0.08
0.02
0.04
0.06
0.08
Residual stress distributions in longitudinal and transverse directions after up-cut milling (left) and down-cut
milling (right). Source: Ref 1
800
Residual stress, N/mm2
f = 0.18 mm/r
f = 0.36 mm/r
f = 0.71 mm/r
400
400
0
0.1
0.2
0.3
0.1
r
0.2
0.3
0.1
0.2
0.3
Fig. 6
(Eq 2)
Inserting Eq 2 into Eq 1:
r
(Eq 1)
Residual stresses in the parallel (solid line) and transverse (dashed line) directions caused by turning of SAE
1045 for different cutting feeds. The cutting speed is 90 m/s and no cooling. Source: Ref 1
3(1 m)
E
hsub
r hfilm
(Eq 3)
The important quantity when comparing the effect of different manufacturing processes, is the
product r hfilm. The stress, r, cannot exceed the
yield stress of the workpiece material. In principle, it should be considered that work hardening in the affected layer could increase the stress
value. It is also obvious that the thickness of the
workpiece has a strong effect because d depends
on the inverse square of the substrate thickness,
hsub.
According to this analysis, grinding should be
the least risky operation to cause distortion and
straightening, shot peening, and pressure rolling
the most risky, with milling and turning in-between. Shot peening, however, rarely is followed
by heat treatment because the beneficial effect
against fatigue or SCC may be lost due to residual stress relaxation.
Some careful studies about distortion due to
prior final shaping are reviewed next.
Experimental and computational studies of
distortion of some components have been performed.
Bearing rings are an example of a product that
is extremely sensitive to distortion and for which
distortion-related costs can be very high. It is the
background to some very careful studies about
the origin of distortions (Ref 2123).
Reference 21 points out that one source of distortion is the presence of uneven residual stresses
that have been generated and stored in the material during the manufacturing route. Although
the rings are perfectly round after turning or cold
rolling, they will become distorted during heating to the austenitizing temperature due to the
release of the residual stresses. This is exemplified in Fig. 13, which shows the residual stress
variation around a ring made of SAE 52100 and
the yield stress at different temperatures. At
room temperature no plastic deformation due to
the residual stresses is expected, but at 600 C
(1112 F) it will occur at certain positions, causing out-of-roundness.
In another example, rings were manufactured
from cold-rolled, straightened, and peeled tubes.

The cut-off rings were distorted and the measurement of ovality of a series of parted-off rings
showed that the ovality followed the helix movement of the manufacturing operation of the tube
(Fig. 14). In a further study, seven tube-manufacturing routes were investigated regarding
their ovality of rings after soft annealing and after hardening. Residual stresses were measured
at the outer surface of each ring at three different
locations spaced 120 apart. Figure 15 shows

depth profiles of residual stresses in the best case
and the worst case. It is evident that the residual
stress is much smaller in the best case, and that
the deviation in residual stress between the three
locations in the worst case is considerable. The
depth of residual-stressed layer is around 400 lm
in this case. When the standard deviation of ovality after hardening is plotted versus the range of
measured residual stress, a clear correlation is

found (Fig. 16). It is also pointed out by the authors that there is a clear difference between
cold-rolled and hot-rolled tubes, probably due to
the different hardnesses, which could be regarded as the potential for storing uneven residual stress.
In a further study, an attempt was made to
discriminate between the influence of the manu0
1000
200
Residual stress N/mm2
800
Radial
Tangential
600
400
200
400
600
800
1000
0
200
1200
0
400
0.5
20
40
60
80
100
120
160
140
Residual stresses in the radial (transverse) and tangential (longitudinal) directions caused by face grooving of IN
718. The cutting speed is 1200 m/s and the feed, 0.5 mm/revolution. Source: Ref 10
Hertzian pressure as a consequence

of the vertical forces connected with
the impact of shot balls
Stretching of a surface layer as

a consequence of surface hammering
Residual stress N/mm2
Distance from the surface, m
200
400
600
800
1000
1200
0
a
po
Elastic
deformation
of surface
Unpeened
Shot velocity = 23 m/s
Coverage = 100%
Shot diameter d = 0.6 mm
0.1
0.2
0.3
0.4
0.5
Influence of shot-peening parameters on residual

stress distributions. (a) Shot diam 0.3 mm and 0.6
mm. (b) Shot velocity 23 m/s and 81 m/s. Source: Ref 16
Fig. 9
Plastically
extended
layer
y
H
0.47 a
200
Residual stress
Residual stress
Residual stress, N/mm
pl
Tangential
400
Axial
600
37CrS4
800
0
Fig. 8
0.2
0.3
0.4
0.1
200
800
Fig. 7

Coverage = 100%
400
600
1000

Effect is possible with a

soft shot and a hard workpiece
Effect is marked with a hard

shot and a hard workpiece
Schematic illustration of the formation of residual stresses as a consequence of two different processes in shot
peening: Hertzian pressure and direct stretching of the surface. Source: Ref 16
0.2
0.4
0.6
0.8
1.0
Residual stress distributions in tangential (rolling direction) and axial directions in a pressure-rolled round bar with 20.7 mm diam made of SAE
5132 (37CrS4) steel. The x-axis shows the distance from
the surface in mm. Source: Ref 9
Fig. 10

facturing route and the heat treatment. An intermediate normalizing treatment was carried out,
and the rings were measured after soft annealing,
normalizing, and hardening. Figure 17 shows the
shares of the out-of-roundness as influenced by
the different steps. For both, the average and the
maximum out-of-roundness the share for the
hardening process itself is relatively small. The
authors comment that normalizing or tempering
just below Acm gives a good possibility to demonstrate the effect of residual stresses present in
the components before hardening. It should also
be pointed out that the out-of-roundness measured after soft annealing probably is not due to
machining residual stresses, but to straightening
or cold rolling before the actual machining,
which can introduce much deeper residual stress
layers. Part of these stresses were mechanically
released when material was removed. This part
is further analyzed in the following section.
Modeling and Measurement of the Effect of
Bending. The effect of the residual stresses generated during plastic deformation of tubes before
machining and heat treatment to bearing rings
Fig. 13
Influence of the temperature on the yield strength on the occurrence of plastic deformations during heating
of rings from hot-rolled tubes of SAE 52100, soft annealed condition. Source: Ref 21
Reference line
250 rings measured in total
DH
10
Compression
Tension
1
B
C
10
9
8
7
6
5
4
Ss
Sy
Ring No.
3
2
The principal stress distribution in a rectangular bar during loading to plastic yielding to the
depth, DH (curve A); the elastic recovery (curve B); and the
resulting residual stress (curve C)
Fig. 11
Fig. 14
Changes in the orientation of the maximum ovality in relation to a reference line of rings parted-off from a
cold-rolled tube of SAE 52100. Source: Ref 21
300
300
2l
hsub
200
100
0
120
0
240
100
200
200
Fig. 12
The bending of a strip due to a thin surface

layer with the residual stress r
Fig. 15
Worst case
hfilm
Best case
400
200
240
100
120
0
0
100
200
200
400
Tangential residual stress distributions under the surface in soft machined rings from different manufacturing
routes. Source: Ref 21

has been studied by computer modeling and experiments (Ref 23). The heat treatment was soft
annealing because it was undesirable to add
quenching stresses. All experiments and simulations were performed for the bearing steel SAE
52100. Three-point bending was performed on
490 mm (19 in.) long tubes with 81 mm (3.2 in.)
outer and 65 mm (2.5 in.) inner diameter. The
tubes were placed on two fixed supports of 1300
mm (51 in.) radius and bent with a punch of the
same radius. Rings of 77 mm (3 in.) outer and
66 mm (2.6 in.) inner diameter and 15 mm (0.6
in.) height were produced by turning. The rings
were soft annealed under controlled heating to
820 C (1508 F) and slowly cooled. The total
cycle time was 20 h.
The computer simulation was carried out using the finite element modeling (FEM) code
ABAQUS in combination with a heat treatment
simulation code DistSIMR. To model the bending, an elastic-plastic model with isotropic hardening and von Mises flow criterion was used.
The material removal was divided into four
steps. First was the cutting of a tube section to
be turned; the second and third steps were the
turning of the inside and outside of the tube, respectively; and the fourth, cutting of the ring by
removing all the material not belonging to the

ring. In the simulation of the soft annealing, no
phase transformations needed to be included.
Most of the plastic deformation during heat treatment takes place at high temperature, and a creep
model was used to describe this deformation.
The simulation of the turning occurs in four
steps. First, a short section is cut from the tube
and distortion is calculated from the elastic
spring back. Next, the inside is turned, and it is
assumed to produce a circular shape that is later
distorted when the elastic stresses are equilibrated. After that the outside is turned and the
springback is limited. Finally, the ring is cut
from the tube. This step releases substantial constraints from the ring and the shape becomes
strongly distorted. During the turning, one end
of the ring is constrained, simulating chucking.
There, the calculated distortion is smaller than at
the other end. In the course of the subsequent
soft annealing, the remaining residual stresses
are relieved and some additional out-of-roundness distortion occurs.
In Fig. 18 calculated distortions are shown for
the outer ring profile at three positions after turning from bent tube and after soft annealing. The
positions correspond to 5, 10, and 16 mm (0.2,
0.4, and 0.6 in.) from the midpoint of the bent

tube. As expected, the section nearest the center
has the largest out-of-roundness. In Fig. 19 the
measured out-of-roundness distortion expressed
as the peak-to-valley radius variation is compared for five rings with the calculated distortion. The agreement is good. It should also be
pointed out that the distortion of a ring made
from an undeformed tube was very small. Two
conclusions can be drawn from these results.
First, substantial out-of-roundness can be generated from the internal stresses. Second, the distortion caused by the soft annealing is relatively
small compared with the distortion due to the
mechanical release of residual stresses. Although
this study treats bending of a straight tube, the
general conclusions are valid for straightening of
a bent tube.
An effect of the condition of the heat treatment equipment was also noticed in the first
study about bearing rings cited previously (Ref
21). A furnace characterized as the best in
class gave only about half the ovality as a standard furnace. Important factors are uneven heating and/or poor support. Preheating rings prior
to the introduction into the austenitizing furnace
has been reported to be beneficial if considerable
Standard deviation ovality (after hardening), m
70
Cold rolled
peeled, straightened
Calculated outer profile at top
Cold rolled, ground

SAE 52100
60
After turning
Cold rolled,
straightened
Cold rolled
peeled
50
50 m
After turning and

soft annealing
Calculated outer profile at center

After turning
Calculated outer profile at bottom
50 m
After turning
After turning and

soft annealing
50 m
After turning and

soft annealing
Cold rolled,
as-rolled
Hot rolled,
peeled
40
Hot rolled,
as-rolled
30
50
150
200
100
Range of measured residual stress, MPa
Fig. 18
Correlations between the residual stress ranges

in tubes and the standard deviation of the ovality of rings after hardening. Source: Ref 21
Calculated accumulated distortion of the outer ring profile at three positions. The distortion is given after
turning from bent tube and after soft annealing. Source: Ref 23
Fig. 16
SAE 52100, rings turned from bent tube

Outer radius, peak/valley
SAE 52100, rings turned from bent tube

Inner radius, peak/valley
Stress relaxation
during normalizing
40.0%
Maximum
(total: 126 m)
Soft
machining
46.0%
Stress relaxation
during normalizing
26.0%
150
200
Out-of-roundness, m
Average
(total: 50 m)
Out-of-roundness, m
200
Simulation
100
50
Hardening
14.0%
Soft
machining
47.0%
Hardening
27.0%
Origin of the out-of-roundness of rings after

hardening of rings from cold-rolled tubes in
SAE 52100. Source: Ref 21
Fig. 17
Simulation
100
50
Experiment
150
Before soft
annealing
Experiment
After soft
annealing
Experiment
Experiment
Before soft
annealing
After soft
annealing
Measured and calculated peak and valley out-of-roundness distortion of bearing rings made from tubes deformed in three-point bending. The distortion is given after turning and after soft annealing. a) Outer radius.
b) Inner radius. Source: Ref 23
Fig. 19

Table 1 Residual stresses and residual stress differences in cross sections of bars 80 80
mm of SAE 4142
Ballring cooling bed
Trailer table
Treatment
Residual stress,
MPa (ksi)
Maximum difference,
MPa (ksi)
Residual stress,
MPa (ksi)
Maximum difference,
MPa (ksi)
Hot rolled
Hot rolled roll straight
Hot rolled press straight
45 (6.5)
32 (4.6)
68 (9.9)
28 (4)
50 (7.3)
102 (14.8)
49 (7.1)
41 (5.9)
84 (12.2)
43 (6.2)
81 (11.7)
123 (17.8)
Cooling after hot rolling has been carried out on a ballring bed or a trailer table and subsequent straightening in a press or parallel roll machine.
Accuracy of measurements, 30 MPa.
Source: Ref 26
residual stresses exist in the machined parts to

be heat treated (Ref 24). The preheat not only
allows the relaxation of residual stresses without
gross plastic deformation but also minimizes
thermal shocking as the parts are subsequently
loaded into the high-temperature furnace.
Shot Peening of Gas Turbine Components.
Shot peening normally is not a process that is
followed by heat treatment, but shot-peened
Magnetoelastisk parameter
140
120
100
80
60
40
Over before
Over after
Under before
Under after
20
0
Left end
Center
5
Right end
Position along the length of the flat bar

The Barkhausen noise parameter measured before (open symbols) and after (filled symbols)
parallel roll straightening of flat bars, 215 22 5000
mm, of the tool steel UHB Chipper. Measurement points
are equally spaced over the length of the bar. Circles from
the top surface and the squares from the bottom surface.
Source: Ref 27
Out-of-roundness, mm
Fig. 20
0.5
0.4
Ring thickness 2 mm
1
2
3
0.3
0.2
0.1
0
1: radius
40 mm
2: radius
34 mm
3: radius
28 mm
Distortion of cut-out rings of the high-speed

steel SS44 after crossed-axis-roll straightened
round bars with 81 mm diam. The filled and open symbols
are for bars that showed little and large ovality, respectively, before straightening. The thickness of the rings is 2
mm. Source: Ref 27
Fig. 21
components may be required to operate at high

temperature. In a recent work (Ref 25) on shotpeened turbine blades made of Inconel 718, the
potential magnitude of mechanical distortion,
which could result from thermal stress relaxation
of compressive layers induced by shot peening,
is calculated using FEM. It is stated that any
shot-peened thin section component could be
subject to distortion, and that the worst is expected when the section thickness is of the order
of the depth of the compressive layer induced by
shot peening. Uneven relaxation such as full relaxation on only one side of a blade is extra damaging. This can arise if there are large temperature differences between the two sides. Two
cases were modeled: uniform relaxation from
both sides of the blade, and relaxation on the
concave side only. As expected, the latter case
created the maximum distortion. Based on experiments it was concluded that thermal relaxation progresses in two stages: a primary, very
rapid stage, and a secondary, more gradual stage.
The rate and amount of relaxation are directly
correlated with the degree of cold work caused
by the shot peening. A high degree of cold work
gave a more rapid and complete relaxation of the
residual stress.
Straightening. Few reports about measurements of residual stresses due to straightening
exist in the literature. Such measurement should
not only give the residual stress values on the
surface but should include depth profiles to be
really meaningful.
Residual stresses have been measured (Ref
26) before and after straightening of 6 m (20 ft)
long bars with a quadratic cross section, 80
80 mm (3 3 in.) and 110 110 mm (4 4
in.). Examples of the results are shown in Table
1 for the steel SAE 4142 (42CrMo4). After hot
rolling, bars were cooled on two types of cooling
bedsballring bed and trailer table, respectively. The former gave out-of-straightness between 20 and 30 mm (0.8 and 1.2 in.) and the
latter, between 40 and 60 mm (1.6 and 2.4 in.).
The bars were either press or roll straightened.
The residual stress was measured by x-ray diffraction in five points in cross sections, near the
corners and in the center of the cross sections.
In Table 1, typical residual stress values are
shown together with the total variation between
residual stress values in a cross section. Notice
that these stresses are not in the longitudinal direction but in the plane of the cross section. The
residual stress values are in the range 30 to 80

MPa (4 to 12 ksi) and the variations 30 to 130
MPa (4 to 19 ksi). The measurement uncertainty
was less than 30 MPa (4 ksi). The variations are
bigger after the straightening. The report does
not tell exactly where the measurement points
are situated, but they are probably a few millimeters from an edge. The yield stress was not
given, but it is normally about 400 MPa (58 ksi)
in the as-hot-rolled condition.
In a large collaborative research project carried out by the Swedish steel industry (Ref 27),
measurements of straightness, distortion, and
FEM of straightening have been carried out for
different steels and equipment.
In one series of studies the residual stresses in
flat bars before and after parallel roll straightening were measured on the top and bottom surfaces along the bar. The Barkhausen noise technique was applied using Stress Scan 500
equipment. The measurements only give approximate stress values, the higher the value, the
higher the residual stress in the tensile direction
will be. An example is shown in Fig. 20 for rectangular bars 215 22 5000 mm (8.5 0.9
197 in.) made of a tool steel UHB Chipper.
The residual stress varies along the length of the
bar, being mainly compressive on the top and
bottom surfaces before straightening, and tensile
on the top surface after straightening. At the two
ends of the bar the residual stress does not
change so much due to straightening and is
slightly compressive. Not shown here is that the
residual stress varied across the width of the bar.
In a further study on rectangular bars 63 4.7
400 mm (2.5 0.19 15.7 in.) made of the
same steel, the bars were carefully ground in
steps from one side and the bow measured after
each step. A conclusion was that most of the bow
was due to operations before straightening.
For round bars with 81 mm (3.2 in.) diameter
made of the high-speed steel SS44 the out-ofroundness of rings with different diameters
turned and cut-off from the bar was measured
(Fig. 21). The bars to be turned were 70 mm (2.8
in.) long and cut from long straightened bars.
These bars had been rotary straightened in a
crossed-axis-roll machine. The figure presents
data for several bars with large and small out-ofroundness before straightening. It is obvious that
the cut-out rings with the largest diameter had
the largest out-of-roundness. This indicates that
the residual stresses have been largest near the
surface of the bars. The change in average diameter (not shown here) was quite small, around
0.1 mm (0.004 in.), indicating that the out-ofroundness was due primarily to unsymmetric residual stresses in the hoop direction created during straightening.
The distortions so measured on the rectangular and the round bars after machining (grinding
and turning, respectively) should be better understood. This is important for being able to minimize them in order to increase the custom satisfaction of delivered bars. A step was taken in
this direction by carrying out an FEM simulation

100
80
60
Height, mm
40
8.
20
0
20
40
0.004
0.012
0.02
9.
60
80
100
200
Fig. 22
200
400
600
Axial position, mm
800
1000
1200
10.
The calculated plastic deformation in the depth of flat bars with thickness 30 mm and length 2500 mm. The
legend shows the effective plastic strain. Source: Ref 27
11.
functions at elevated temperature. The same

should hold for a subsequent heat treatment.
12.
of parallel-roll straightening using DYNA2D.

The model includes four upper and four lower
rolls and comprises 2501 elements and 1600
nodes. The flat bar has the dimension 30 2500
mm (1.2 98 in.). The width is not considered
in the two-dimensional model. Figure 22 illustrates how the plastic deformation grows during
the straightening and reaches 2% effective strain
at the surface and 0.4% in the center after passing
the last rolls. The residual stresses (not shown
here) can also be calculated. The deep plasticized
zone indicates that the residual stresses can extend deep into the material. A main conclusion
is that straightening can be simulated and the
effect of material operation parameters studied.
Conclusions
A survey has been presented about the generation of residual stresses due to machining
such as grinding, milling, and turning, and surface deformation processes such as shot peening,
pressure rolling, and straightening. The text illustrates by typical examples of residual stress
profiles. The so-called Stoneys formula is used
to show that due to the thin affected layers after
machining only small distortions are expected
after relieving the residual stresses when a machined component has a dimension (thickness)
much larger than a few hundred microns. Shot
peening and, in particular, straightening should
give more distortion when relieving the residual
stresses due to larger affected depths. These conclusions are supported by careful studies on
manufacturing of bearing rings and straightening
of bars. It is also pointed out that preceeding heat
treatments and straightening operations could
give much more substantial distortions due to the
presence of deep and irregular residual stress
fields. A modeling study of stress relaxation of
shot-peened material demonstrates that severe
distortions can occur when a relatively thin part
REFERENCES
13.
1. B. Scholtes, Residual Stresses Introduced
by Machining, Advances in Surface Treatments, Vol 4, A. Niku-Lari, Ed., Pergamon
Press, Oxford, U.K., 1987, p 5971
2. E. Brinksmeier, Process Near Control of the
Residual Stress State of Ground Workpieces
with Magnetic Methods, Residual Stresses,
V. Hauk, H. P. Hougardy, E. Macherauch,
and H.-D. Tietz, Ed., DGM Informationsgesellschaft Verlag, Oberursel, Germany,
1993, p 863872
3. J. Grum and P. Zerovnik, Residual Stresses
in Steels after Different Heat Treatments
and Grinding, Fifth International Conference of Residual Stresses, Linkoping University, Linkoping, 1997, p 250255
4. E. Brinksmeier, A Model for the Development of Residual Stresses in Grinding, Advances in Surface Treatments, Vol 5, A.
Niku-Lari, Ed., Pergamon Press, Oxford,
U.K., 1987, p 173189
5. R. Herzog, A. Sollich, and H. Wohlfahrt,
Residual Stresses of Heat Treated Steels
with Different Hardness after Grinding with
Cubic Boron Nitride (CBN), International
Conference on Residual Stresses, ICRS-2,
G. Beck, S. Denis, and A. Simon, Ed., Elsevier Applied Science, London, 1989, p
740746
6. A. Sollich and H. Wohlfahrt, Residual
Stresses in Quenched and Tempered Steels
after CBN-Grinding with and without Prestress, Advances in Surface Treatments, Vol
5, A. Niku-Lari, Ed., Pergamon Press, Oxford, U.K., 1987, p 919926
7. P. Belardi, E. Capello, and Q. Semeraro,
14.
15.
16.
17.
18.
19.
20.
21.
Surface Integrity in Cylindrical Grinding of

39NiCrMo3 Steel, Sixth International Conference on Residual Stresses, ICRS-6, IOM
Communications, London, 2000, p 217
224
D.F. McCormack, W.B. Rowe, X. Chen, A.
Bouzina, M.E. Fitzpatrick, and L. Edwards,
Characterising the Onset of Tensile Residual Stresses in Ground Components, Sixth
International Conference on Residual
Stresses, ICRS-6, IOM Communications,
London, 2000, p 225232
B. Scholtes, Residual Stresses in Mechanically Surface Deformed Materials, DGM
Informationsgesellschaft Verlag, Oberursel,
Germany, 1991 (in German)
C. Schlauer, Ongoing Ph.D. thesis work,
Department of Mechanical Engineering,
Linkoping University, Sweden, 2001
L. Olovsson, L. Nilsson, and K. Simonsson,
An ALE Formulation for the Solution of
Two-Dimensional Metal Cutting Problems,
Comput. Struct., Vol 72, 1999, p 497
507
D. Deslaeff, Numerical Modelling of Shot
Peening: A Three-Dimensional and Dynamic Approach, Ph.D. thesis, Universite
de Technologie de Troyes, 2000 (in French)
S.A. Mcguid, G. Shagal, and J.C. Stranart,
Finite Element Modeling of Shot Peening
Residual Stresses, J. Mater. Process. Technol., Vol 9293, 1999, p 401404
K.I. Mori and K. Osada, Application of Dynamic Viscoplastic Finite Element Method
to Shot Peening Process, Trans. NAMRI/
SME, Vol XXII, 1994, p 2934
K. Schiffner and C. Droste gen Helling,
Simulation of Residual Stresses by Shot
Peening, Computers and Structures, Vol 72,
1999, p 329340
H. Wohlfahrt, Shot Peening and Residual
Stresses, Residual Stress and Stress Relaxation, Plenum Press, New York, 1982
Straightening of Bars, Shapes, and Long
Parts, Forming and Forging, Vol 14, Metals
Handbook, 9th ed., American Society for
Metals, 1988, p 680689
Straightening of Tubing, rev. by P. Delori,
Forming and Forging, Vol 14, Metals
Handbook, 9th ed., American Society for
Metals, 1988, p 690693
G. Horn, Modern Section Straightening
Technology, Proceedings of Straightening
of Long Products, 6 Nov 1991
G.G. Stoney, The Tension of Metallic Films
Deposited by Electrolysis, Proc. R. Soc.,
Vol A82, 1909, p 172175
J. Volkmuth, U. Sjoblom, J. Slycke, and A.
Thuvander, Effect of Uneven Residual
Stresses on Dimensional Changes and Variations of Through Hardening Bearing Steel
Rings, Proc. 20th ASM Heat Treating Society Conference, 912 Oct 2000, ASM International, 2001

22. J. Volkmuth, F. Hengerer, and Th. Lund,
Influence of Casting Process and Casting
Cross Section on the Out-of Roundness,
Harterei-Technische Mitteilungen, Vol 6,
1995, p 352358 (in German)
23. A. Thuvander, Out-of-Roundness Distortion of Bearing Rings due to Internal
Stresses from Tube Bending, Submitted to
Mater. Sci. Technol., 2000
24. H.W. Walton, Understanding and Control-
ling Distortion in Large Bearing Rings

Some Practical Aspects, Proc. 2nd Int.
Conf. on Quenching and the Control of Distortion, G.E. Totten, M.A.H. Howes, S.
Sjostrom, and K. Funatani, Ed., ASM International, 1996, p 143147
25. P.S. Prevey, D.J. Hornbach, and P.W. Mason, Residual Stress Relaxation and Distortion in Surface Enhanced Gas Turbine Engine Components, Proc. 17th Heat Treating
Society Conf., ASM International, 1997, p

312
26. E. Doege and F. Weber, The Effect of Production Conditions on Residual Stresses in
Bars, Stahl und Eisen, Dusseldorf, Germany, Vol 111, 1991, p 85 88 (in German)
27. J. Bergstrom, The Effect of Straightening on
Material Properties, Jernkontorets forskning, TO 4037, Jernkontoret, Stockholm,
1996 (in Swedish)

DOI: 10.1361/hrsd2002p159

Factors Affecting Final Part Shaping

J. Pan, Shanghai Jiao Tong University, P.R. China
THE FACTORS THAT AFFECT the final

shape of components are quite complicated.
Many resultant factors from each step of the design and manufacture processes, besides heat
treatment and alignment processes, have various
influences on the final shape of the components.
Factors and the relationships between heat treatment and distortion of components are summarized briefly in Fig. 1. The complexity results not
only from the large number of factors but also
from complicated interactions among them. For
example, with the contents of the alloying elements in a steel increasing, the heat conductivity
of the steel decreases, and the difference in temperature increases in the component during heating and cooling processes, resulting in larger distortion. On the other hand, the hardenability of
the steel increases with the increase of the contents of alloying elements. The proper cooling
rate can be employed in its quenching process in
order to decrease the distortion of a component
made of the steel. Another example is an eccentric cylinder shown in Fig. 2. The distortion tendency of the component is large due to its thin
wall and asymmetry. If the component is made
of 45 steel and quenched in a salt solution, the
tendency of distortion further increases because
of the high cooling rate. The external diameter
of the component is obviously expanded after
quenching. The distortion of the component is
very small if the component is made of 40Cr
steel and marquenched in nitrate (Table 1).
When a screw plate is molded by a rolling
process, the shape of its tooth face is protruded
due to the influence of plastic-flow resistance of
the metal from which the screw plate is made.
The shape of the tooth face is concave after
quenching if a certain method is employed to
decrease the cooling rate of the back side of the
screw plate during quenching. For example,
when the fixture shown in Fig. 3 was used during
quenching, the tooth faces of the two screw
plates were concave. If a steel plate with appropriate thickness determined by experiments is
placed on the back sides of the screw plates to
make the cooling rate of the back sides lower
than that of the tooth faces, the cooling rate of
the back of the screw plate will be relative to the
cooling rate of the tooth faces.
For a component, the asymmetry is another
factor that increases the tendency of distortion.
The distortion of a component made of AISI

1045 steel, shown in Fig. 4, was heated at 830
C (1530 F), quenched in water for 3 s and then
quenched in oil. The external with a diameter of
57 mm (2 in.) (D1) expanded by 0.17 to 0.27
mm (0.0070.011 in.) and the internal diameter
(d1) expanded by 0.37 mm (0.015 in.) (Ref 1).
The distortion of D1 and d1 obviously could be
reduced after quenching if the end of the component was put on a small sleeve with an external
diameter of 80 mm (3 in.), internal diameter of
57 mm (2 in.), and height of 30 mm (1 in.).
It can be seen from these examples that the
influences of various factors affecting the final
shape of the component sometimes superpose,
and sometimes offset, one another. Thus, it is
necessary to know every factor comprehensively
when the reasons that cause the distortion are
analyzed and the measures of controlling distortion are studied. But, it is true that under various
conditions, a factor plays a different role in the
distortion of a component; it can play an important role under one condition and an unimportant
role under another condition.
Influence of Shape of Component

on Heat Treatment Distortion
Quenching Distortion of a Component
with Simple Shape. Besides the volume change
caused by the phase transformation during heat
treatment, the shape of a component will deform
with the functions of thermal stress and structure
stress. The distortions of simple components
caused by these stresses are shown in Fig. 5. The
functions of thermal stress and structure stress
are contrary to each other, as shown in Table 2.
The quenching distortion of a certain component
is the result of the common effects of thermal
stress and structure stress. Generally speaking,
the distortion of the component with simple
shape, uniform thickness of the wall, and good
symmetry is relatively small and regular. The
more complex and worse the symmetry of the
shape of a component is, and the larger the difference in thickness of the component, the larger
the distortion of the component will be.
The Cylindrical Component. The length and
diameter of the component with axial symmetry
after quenching will have a little change. The
distortion relates to the ratio of length to diameter (L/D), size, quenching condition, heating
temperature, the composition of the steel, and so
on. Thus, the quantity and direction of the distortion under various conditions are quite different. Some examples are listed in Table 3.
For a medium-carbon cylindrical component,
the length will increase while the diameter will
decrease after being quenched. The distortion
will increase with the heightening of the quenching temperature (Fig. 6). The distortion is also
sensitive to the ratio of L/D (Fig. 7). If the medium-carbon steel cylinder has not been fully
hardened, the distortion will decrease with the
increasing of the diameter (Fig. 8). The distortion of 1045 steel after being heated to 840 C
(1540 F) and quenched in nitrate bath is less
than that in water (Table 6).
Carbon content greatly influences the distortion of cylindrical carbon steel components (Fig.
9). The distortion of medium-carbon steel is
greatest, while the distortions of both low-carbon
and high-carbon steels are relatively small. For
T12 steel after being quenched from 780 C
(1430 F) and 1015 steel after being carburized
and then quenched in water from 850 C (1560
F), the length decreased and the diameter increased.
If the cooling condition is different between
the two sides of the cylindrical component during the quenching process, there will be bending
deflection. The face with a higher cooling rate
will protrude in a longitudinal direction due to
thermal stress, but will be concave because of
transformational stress. The direction of the
bending is dependent on the relative amount of
thermal stress and transformational stress.
Asymmetry. The asymmetrical axle component will experience bending deflection caused
by uneven cooling of the two sides during
quenching. If there were only thermal stress, the
component would protrude to the side with the
faster cooling rate when quenched. But, because
of transformation stress, the component protruded to the side with the slower cooling rate
when quenched. The amount of final distortion
depends on the relative amount of these two
stresses:
The axle with keyslot (Fig. 10) is the most
popular asymmetrical component. The influ-

ences of keyslot on quenching distortion are
shown in Table 4.
The axle with cross slot experienced bending
deflection after being quenched. The distor-
tion of a small axle made of AISI 52100 steel

(Fig. 11) was more than 0.75 mm (0.03 in.)
after being heated to 840 C (1540 F),
quenched in nitrate bath for 5 to 10 min, then
cooled in air. The 1045 steel valve stem had

such bending deflection as shown in Fig. 12
after being heated to 820 C (1500 F),
quenched in water for 2 s, and then quenched
The shape of the component

The design of
the component
The rationality of qualification and performance requirement
The technological
process
The metallurgical quality of steel
Forging
The manufacturing
methods of the blank
Rolling
Casting
Weldment
Cutting
The influences of pre-heat

treating and original structure
The selection of heat

treatment methods
The factors that
have influences
on the final shape
of the component
Equipment and medium

Charging form
Heating
Heating rules
Control of heating
Equipment
Heat treatment
Composition
Cooling (quenching)
Quenching medium
Temperature
The design of fixture
Agitation
Cooling methods
Alignment during quenching
Tempering and aging
The factors of chemical heat treatment which cause distortion

Alignment and
destressing
The stability of the

component after being
heat treated
Fig. 1
Factors affecting the final shape of a component
The stability of microstructure
The residual stresses
Factors Affecting Final Part Shaping / 161

in oil. The shrinkage of the 4 mm (0.16 in.)
slot was 0.145 mm (0.006 in.).
The cylinder and screw with ringlike slot
shrank in the longitudinal direction, as shown
in Table 5.
For components with a deep slot in the end
side, quenching distortion of the slot is influenced by the quenching direction. As shown
in Fig. 13, with the slot downward, width b
increases after quenching (Fig. 13a); with the
slot upward, width b decreases (Fig. 13b). In
the latter cases, embedding a steel block with
the width of b, the distortion is effectively
controlled (Fig. 13c).
Influence of Shape on the

Distortion of Hollow Cylinders
Cylinder with Even Wall Thickness. The
distortions of symmetric hollow cylinders with
uniform wall thickness caused by thermal stress
and transformational stress are shown in Fig.
5(d). The actual distortion of the component is
the combined action of those two inner stresses.
Example 1 is a long jacket made of 1045 steel.
Its size is 70 50 70 mm (25 2
2 in.). The distortion after quenching is shown
in Fig. 14. The internal radius and external radius
swell while the height increases, as shown in Table 6.
Example 2 is a hollow cylinder made of 52100
steel. The distortion after being heated at 830 C
(1530 F), oil quenched, and tempered at 160 C
(320 F) for 2 h is shown in Fig. 15. The deformation of the external radius is 0.02 to 0.04
mm (0.00080.002 in.) while the deformation
of the internal radius is 0.04 to 0.08 mm
(0.0020.003 in.). The deformations of both
ends are larger than that of the middle.
The distortion of the symmetric hollow cylinder with even wall thickness after being heat
treated relates to diameter, wall thickness,
height, and so on. Figure 16 shows the relationship between wall thickness and the deformation
of hollow cylinders made of 1045 and 5140
steels, respectively.
The principal laws of the distortion of a hollow cylinder are:
35
2
48
4
25
Fig. 2
Distortion of an eccentric cylinder bushing
Table 1
Distortion of 1045 steel eccentric cylinder bushing
Quenching technology
800 C (1470 F); quenching in water

for 1.5 s; quenching in oil
810 C (1490 F); quenching in aqueous
alkali with 180 C (356 F)
Hardness after quench, HRC
External radius, 48
Internal radius, 36
49 to 53
0.15 to 0.25
0.30 to 0.40
42 to 43
0.08 to 0.11
0.10 to 0.15
Source: Ref 1
Table 2
Comparison of distortions caused by thermal stress and structure stress
Distortion caused by thermal stress
Distortion caused by structure stress
Contracting along the largest-sized direction

Elongating along the shortest-sized direction
Surface protruding
Smoothing of the edges and corners
Exradius expanding, inradius contracting
The end inner hole tending to horn type
Elongating along the largest-sized direction

Contracting along the shortest-sized direction
Surface concavity
Sharpening the edges and corners
External radius contracting, internal radius expanding
The end inner hole tending to close up
80(D)
39(d)
3
The internal radius and external radius of a
component made of 1045 steel swell and the

height increases.
If the component is fully hardened, the
amount of swelling of the internal radius and
external radius increases with the increase of
wall thickness when the wall thickness is
small. When the wall thickness is larger than
10 mm (0.4 in.). However, the amount of
swelling of the internal radius and external
radius decreases with the increase of the wall
thickness if the component is not fully hardened.
For the component with wall thickness less
than 12 mm (0.5 in.), the amount of swelling
of internal radius is approximately 0.7 to
0.9% of the swelling quantity of external radius when being quenched in nitrate bath. The
2
67
1
30
39(d1)
57(D1)
Fig. 3
Jig and fixture method for quench of screwplate.

(1) screwplate, (2) ironplate, (3) jig
Fig. 4
Distortion of a stop ring made of AISI 1045 steel.

Source: Ref 1

amount of swelling of the internal radius is
less when being double-liquid (salt and oil)
quenched, approximately 0.5 to 0.6% of that
of the external radius. The deformation is
much less when being marquenched, normally less than 0.15%, and tends to be reduced.
After being tempered at medium temperature,
internal and external radii of the component
decrease, but the deformations are much less
than that during quenching.
When medium-carbon low alloy steel 5140 is
quenched in oil or marquenched, the internal
radius shrinks if the wall thickness is smaller
than 10 mm (d 10 mm) (0.4 in., d 0.4
in.), while the internal radius, external radius,
and height increase if the wall thickness is
larger than 10 mm (d 10 mm) (0.4 in., d
Initial state
0.4 in.). The deformation is approximately

0.1%, which is much less than that of 1045
steel after being quenched in water. The distortion of the internal radius of the hollow
cylinder made of 1045 increases with the increase of d. When the thickness of wall
reaches a certain value, the amount of swelling will decrease with the increase of d.
The distortion patterns of high-carbon steel
1080, 10100, and 10120 are generally the
same. For quenching a hollow cylinder made
of high-carbon steel, the swelling deformations of internal radius and external radius are
less than that of medium-carbon steel. The
higher the carbon content, the smaller the
swelling deformation will be. When the wall
thickness is larger than the thickness of the
hardening layer, the larger the size is, the
smaller the swelling deformation of internal

and external radii will be, and even the internal and external radii shrink.
When high-carbon low-alloy steel 52100 is
heated to 830 C (1530 F) and quenched in
oil, its internal and external radii shrink, while
the height increases.
The inner hole of low-carbon steel 1015
shrinks after being carburized and quenched.
The larger the internal radius, the larger the
shrinking deformation will be. The shrinking
deformation increases with the increase of
height. The deeper the carburized layer is, the
larger the shrinking deformation will be. The
external radius shrinks when the wall thickness is smaller than 10 mm (0.4 in.) and
swells when the wall thickness is larger than
10 mm (0.4 in.).
L
L
Thermal
stress
Structural
stress
(a)
Fig. 5
Table 3
(b)
(c)
(d)
(e)
Distortion of some typical simple components
Examples of a smooth cylinder with varying sizes after being quenched

DD, mm (in.) (a)
Steel No.
1045
4140
0.65% C, 0.85% Mn
1.52% C, 0.37% Mn
1.2% C
52100
Carburized 1015
Heat treating process
Top
Medium
Bottom
DL, mm (in.) (b)
Water quenching from 820 C (1508 F)

Nitrate bath quenching from 840 C (1544 F)
Oil quenching from 830 C (1526 F)
Nitrate bath quenching from 830 C (1526 F)
0.11 (0.0043)
0.0075 (0.0003)
0.013 (0.0005)
0.007 (0.00028)
0.016 (0.0006)
0.022 (0.00087)
0.033 (0.0013)
...
0.045 (0.0018)
0.020 (0.00079)
0.025 (0.00098)
...
...
0.005 (0.0002)
0
0.004 (0.00016)
0.01 (0.00039)
0.018 (0.0007)
0.020 (0.00079)
0.050 (0.0020)
0.048 (0.0019)
0.052 (0.002)
0.025 (0.00098)
0.030 (0.0012)
0.041 (0.0016)
0.038 (0.0015)
0.005 (0.0002)
...
0.008 (0.0003)
0.01 (0.00039)
0.023 (0.0009)
...
...
...
...
...
...
...
...
0.535 (0.0211)
0.097 (0.0038)
0.135 (0.0053)
0.104 (0.0041)
0.206 (0.008)
0.025 (0.00098)
0.110 (0.0043)
0.078 (0.003)
0.015 (0.0006)
0.1 (0.0039)
0.19 (0.0075)
0.010 (0.0004)
0.105 (0.0041)
Ellipses, no experimental data. (a) D 25 mm (1 in.). (b) L 125 mm (5 in.), Source: Ref 1

0.6
L
D
0.5
0.4
When low-carbon low-alloy steel 5120 is car-
The internal radius of high-alloy tool steel
burized and quenched in oil, its internal radius shrinks, while the external radius and
height have no obvious change.
Cr12Mo swells after being quenched. The

thicker the wall is, the larger the swelling deformation will be. Normally, the external ra-
, %
0.3
%
Deformation in length
Deformation in diameter
Sample size: 25, L/D = 5 ~ 6
0.43
0.2
0.1
+
0.206
0.208
0.10
0.00084
0.1
0.2
700
750
800
850
900
Quenching temperature, C
950
0
0.000014
Effect of quenching temperature on the distortion

of medium-carbon steel cylinder
Fig. 6
0.024
0.012
0.15% C
0.45% C
1.0% C
1.2% C
0.45
L
D
0.4
Fig. 9
0.35
Effect of content of carbon on the distortion of carbon steel cylindrical components
0.3
, %
0.25
0.2
0.15
0.1
0.05
Fig. 10
0
0.05
3
L/D, mm
Table 4
Effect of length to diameter (L/D) on the distortion of medium-carbon steel cylinder
Fig. 7
Distortion of asymmetrical axle with keyslot
Quenching distortion of asymmetrical axles with keyslots
Type of steel and

quenching method
0.7
Middle carbon steel: water
oil
0.6
Convex deformation toward the side with keyslot

Medium-carbon steel
marquenched in nitrate
bath
High-carbon steel
marquenched in nitrate
bath
L/D > 4.5

0.5
0.4
L,%
Direction of bend deformation
Deformation of keyslot
Medium-carbon steel
quenched in water
High-carbon steel quenched
in water
0.3
The bend with convex or concave deformation is limited.

The deformation of axes results from other factors.
Width of keyslot
dwindles
Width of keyslot
dwindles or expands
slightly
Various alloying steels

quenched in oil
0.2
0.1
0.0
Fig. 8
Convex deformation toward the side opposite keyslot
Width of keyslot
expansion
Various alloying steels

quenched in nitrate bath
0
20
40
60
D, mm
80
Effect of diameter of medium-carbon steel cylinder on its distortion
Convex deformation toward the side opposite to keyslot
Width of keyslot
expansion

6
3.2
20
20
10
220
355
Fig. 11
Bend distortion after marquenching of 52100 steel axle with keyslot. Source: Ref 2
20
12
Bend direction after quenching
Fig. 12
Distortion of valve stem after quench. Source: Ref 1

A
Table 5 Comparisons of distortion of a smooth cylinder and distortion of a screw with

ringlike slot after quenching
Distortion after quenching, mm (in.)
Shape and size of workpiece
DD
Steel Size before quenching, mm (in.)
DL
1050
U28 119
(U1.102 4.685)
1050
U28 119
(U1.102 4.685)
0.03
(0.0012)
5140
U22 120
(U0.866 4.724)
0.02 to 0.03
0.63 to 0.69
(0.0008 to 0.0012) (0.0248 to 0.0272)
U22 120
(U0.866 4.724)
0.08
(0.0032)
5140
0.27 to0.32
(0.0106 to 0.0126)
0.18 to 0.15
(0.0071 to 0.0059)
d(50)
D(70)
0.13 to 0.18
(0.0051 to 0.0071)
Fig. 14
Source: Ref 3
Distortion of a long cylindrical jacket made of

1045 steel after quenching. Source: Ref 1
25
60
50
+b
Fig. 13
Distortion tendency of axle with keyslot in the end side
Fig. 15
Ref 1
Distortion of hollow cylinder made of 52100

steel after quenching and tempering. Source:

Table 6
Distortion of a long cylindrical jacket made of 1045 steel after quenching

A-A
B-B
DD, mm (in.)
Dd, mm (in.)
DD, mm (in.)
Dd, mm (in.)
DD, mm (in.)
Dd, mm (in.)
820 C (1508 F),

quenched in water
780 C (1436 F),
quenched in water
0.36 (0.014)
0.44 (0.017)
0.22 (0.0087)
0.35 (0.0138)
0.30 (0.012)
0.36 (0.014)
0.29 (0.011)
0.39 (0.015)
0.24 (0.009)
0.35 (0.0138)
0.37 (0.015)
0.43 (0.017)
Source: Ref 1
0.4
0.4
a(D)
b(d)
0.3
0.3
0.2
D, mm
0.2
0.1
1
0.0
0.2
10
Concave dies normally have complex shape,

and the size of each part is very different. Thus,
the quench distortion of the cavity of concave
dies is more complex than that of rings. It is the
25
0.2
30
10
25
15
20
, mm
30
0.08
0.06
0.04
0.04
3
0.02
0.02
4
, mm
D, mm
0.1
15
20
, mm
0.06
0.00
0.00
0.02
0.02
8
C(D)
0
10
0.04
0.04
0.06
0.1
0.0
0.1
Influence of Asymmetry
Quench Distortion of the Cavity of

Concave Dies
When the cooling rates at the two sides of an

asymmetric component differ, there will be
bending deflection after the component has been
quenched. The side with a high cooling rate will
be concave when thermal stress acts on it, while
the side with a low cooling rate will be concave
when structure stress acts on it. Figure 20 shows
several examples of asymmetric components.
The bending deflection is dependent on the
steels, cooling method, and cross section size, as
shown in Table 7. The greater the size of the
component is, the greater the tendency of bending deflection. Figure 21 shows examples of the
asymmetry of the parts shown in Fig. 20. The
quenching distortion of an asymmetric alloying
steel thin structural gear heated by high frequency induction is shown in Fig. 22. When the
frame of a height ruler made of 9CrSi steel is
heated to 860 C (1580 F) and quenched in oil
in the longitudinal direction, the notch size decreases because of the notch direction during immersion; the outer side is cooled quickly, while
the inner side is cooled slowly (Fig. 23).
C-C
Heat treatment process
D, mm
dius shrinks. The shrinking deformation decreases with the increase of wall thickness.
When the external radius is smaller than the
height, the deformation of height is larger
than that of diameter. Otherwise, the deformation of height is smaller than that of diameter.
Influence of Keyslot on the Distortion of a
Hollow Cylinder after Quenching. The
change of keyslot width is the same as internal
radius, observed with the inside radius in the tubular example just discussed. The keyslot becomes wide if the internal radius swells. On the
other hand, the keyslot becomes narrow if the
internal radius shrinks (Fig. 17). The smaller the
wall thickness is, the larger the distortion of
keyslot will be.
Uneven Distortion of Internal Radius. For
hollow cylinders with keyslot, the distortion of
internal radius is uneven. The shape of the inner
hole after being quenched becomes elliptical.
Eccentric Hollow Cylinder. Because the
wall thickness of an eccentric hollow cylinder is
quite different in various directions, the deformation is different after being quenched, as
shown in Fig. 18.
Component with Different Cross Sections.
The deformations of internal and external radii
at both ends of the component are different because of the difference in wall thickness. Meanwhile, the adjoining parts restrict each other, thus
the deformation is as shown in Fig. 19.
15
20
, mm
25
d(d)
0.06
30
10
15
20
D, mm
25
30
Fig. 16
Effect of wall thickness on the distortion of hollow cylinder after quenching, U70 35 mm (U2.756 1.378
in.). 1: 5140, heated at 840 C (1544 F), quenched in oil; 2: 5140, heated at 840 C (1544 F), quenched in
oil, tempered at 410 C (770 F); 3: 5140, heated at 840 C (1544 F), quenched in 140 C (284 F) nitrate bath; 4: 5140,
heated at 840 C (1544 F), quenched in 140 C (284 F) nitrate bath tempered at 410 C (770 F); 5: 1045, heated at 840
C (1544 F), quenched in 5% NaCl water; 6: 1045, heated at 780 C (1436 F), quenched in 5% NaCl water; 7: 1045,
heated at 840 C (1544 F), quenched in 150 C (302 F) nitrate bath. Data Source: Ref 1
result of the combined effect of thermal stress,

transformation stress, and volume variation
caused by phase transformation that affect the
shape and size of the dies after quenching. For
example, the die with a thin wall or thin edge
exhibits less temperature difference and less
thermal stress during the quenching process, and
is fully hardened easily, leading to greater transformational stress so that the cavity expands. As
the wall of the die increases, the thermal stress
will become the major cause of cavity shrinkage.
If the thickness of the wall is not uniform, the

thinner position is easy to expand and the thicker
is easy to shrink (Fig. 24).
Influences of Treating Procedure

and Machining Process on the Final
Shape of Component
Coordinating of Cold-Working and HotWorking. The coordinating of machining and

heat treating has obvious influence on the
quenching distortion of a component that is easy
to deform. The elastic chuck made of 1080 steel
shown in Fig. 25 is easy to deform when quenching because of its bad rigidity. The distortion
will be reduced if position A is not cut before
being quenched but is cut after quenching.
Influence of the Sequence of Heat Treating
and Machining. Example 1: Ball Screw. The
length of long and thin ball screw made of AISI
52100 steel (Fig. 26) extends after being
quenched. The thinner the component is or the
deeper the hardening layer is, the larger the elongation will be. When screw rollaway nest is machined before the component being quenched,
the distance of screw should be shrunk. When
the component is quenched, its length will be
increased to reach the length required.
For the ball screw with lead less than 10 mm
(d 10 mm) (0.4 in., d 0.4 in.), such method
as quenching the smooth pole and then the machining screw can be employed. It is useful to
get the exact final shape.
Example 2: High-Speed Gear Wheel. The
high-speed gear wheel shown in Fig. 27 is made
of alloy carburized steel. The rotating speed is
4000 rpm. Thin wall structure and size precision
are required to reduce the centrifugal force. If
the spoke is first machined and then carburized
and quenched, it is easy to have obvious heat
treating distortion. The following technique,
however, can guarantee the precision of the final
shape and size of the gear wheel and meet the
requirement of a high-speed gear wheel:
1. Forge
2. Anneal
3. Machine (the machining allowance at both
ends is 3 mm (0.12 in.); machining the external circle to the required size, the machining allowance of internal circle is 3 mm
(0.12 in.), hobbing
4. Carburize
5. Temper at high temperature

6. Machine (cut the carburized layers at both
ends to the required sizes; cut the carburized
layer of the internal circle)
7. Quench and temper at low temperature
8. Machine the spoke to the required size; fine
machine inner hole
9. Stabilize (150 C, or 300 F)
10. Machine key slot; grind inner hole, gear
grind
next process. If the deflection cannot meet the

requirement, the component has to be aligned
first and then heated to 650 C (1200 F) to relieve stress.
If these two stress relieving times are replaced
by one stress relieving, or the stress relieving
temperature is up to 620 C (1150 F), the deformation of the component after nitriding is
larger than 0.05 mm (0.002 in.).
Influence of Machining Allowance

and Stress-Relieving Procedure
Influence of Residual Stress Caused

by Cutting on Heat Treating
Distortion
The main axle of boring lathe is made of

38CrMoAl steel. The external radius is 105 mm
(4 in.) and the length is 2010 mm (79 in.). The
required hardness after being quenched and tempered is 24 to 28 HRC; after nitriding, the surface hardness is more than 900 HV, and the
depth of the nitriding layer is 0.4 to 0.5 mm
(0.01570.0197 in.). The machining procedure
follows:
1. Rough machine; the machining allowance
of external circle is 3 to 4 mm (0.120.16
in.)
2. Quench and temper at high temperatures
3. Align
4. Stress relieve (650 C, or 1200 F, 15 h)
5. Fine machine; the machining allowance of
external circle is 0.25 mm (0.01 in.)
6. Stress relieve (650 C, or 1200 F, 10 h)
7. Grind; the machining allowance of external
circle is 0.06 mm (0.002 in.)
8. Nitride
9. Half-fine grind; the machining allowance of
external circle is 0.02 to 0.03 mm (0.0008
0.0012 in.)
10. Artificial age (450 C, or 840 F, 6 h)
11. Fine grind to the required size
The deflection should be checked after each
stress relieving to meet the requirement of the
The residual stress caused by the cutting process causes distortion in the subsequent heating
process due to stress relaxation.
Residual stress will be high and the distortion
will increase if the front rake angle and inner
angle of the tool edge are not up to the standard
or the cutting load (depth of cut and feeding) is
too large. Long and thin components are especially sensitive to these factors. Improper clamping of the component also can cause more residual stress.
For example, there is a gear wheel made of
18Cr2Ni4W steel. Its external radius is 240 mm
(9 in.), internal radius is 224 mm (8.8 in.), modulus is 4 mm. The ovality is more than 4 mm
(0.16 in.) after the component is nitrided at 500
C (930 F) for 40 h. It has been found that the
four-jaw chuck used for clamping during heating
caused greater residual stress. The following
measures should be employed to keep the ovality
at less than 0.04 mm (0.002 in.):
Use plastic clamping apparatus, bring pres-
sure to bear on circle evenly
Increase cutting frequency and decrease cut-
ting depth
Perform stress relieving annealing during cut-
ting process
Larger expansion
2
35
48
Height, h = 25
(a)
Fig. 17
(b)
Distortion of a hollow cylinder with inner keyslot, after quenching. (a) The keyslot becomes wider with the
increase of internal radius. (b) The keyslot becomes narrower with the decrease of the internal radius.
Fig. 18
in water
Distortion of inner hole of an eccentric hollow

cylinder made of 1045 steel, after quenching
Methods of Manufacturing Blanks

and the Influences of Their Original
Structures
Castings. Influences of the original structures
of castings are as follows.
Segregation in the Castings. The segregation
results in many zones with different composition
in the same casting, which causes the physical
properties of these zones, such as heat conductivity, specific heat, elastic modulus, yield limit,
melting point, and phase transformation temperature to be inconsistent. The thermal stress
and transformation stress in the workpiece will
increase. Segregation is particularly troublesome

for sand casting. If the blank is obtained by metallic mold casting, this disadvantage can be improved to some extent. The segregation in the
blank, however, becomes comparatively much
less in the case of pressure casting or centrifugal
casting.
Grain Size. Grain coarsening in sand casting
is also a problem, resulting in increased distortion during subsequent heat treatment. However,
much finer grains can be obtained by the methods of metallic mold casting, pressure casting,
or centrifugal casting.
The Morphology of Eutectic Structures. The
morphology of eutectic structures is very sensi-
Bend distortion of long asymmetrical parts with

small cross section. The lower sides were
cooled quickly.
Fig. 19
Table 7
With the increase in asymmetry, the bend distortion increases.
Fig. 20
Distortion of a hollow cylinder with varying wall thickness
Direction of bend distortion caused by asymmetry under different conditions
Condition
Bending direction
The side that

cooled slowly
The side that

cooled quickly
Carbon steel, quenched in water or salt water

Bad hardenability of steel or component with large
cross-section area; the hardened layer is thin
Fig. 21
Asymmetry of the parts in Fig. 20
Fig. 22
Distortion of an asymmetric thin structural steel

gear heated by high-frequency induction, after
Carburizing of low-carbon alloying steel

Medium-carbon alloying steel or alloying tool steel,
quenched in oil or nitrate
Good hardenability of steel or component with
small cross-section area; it could be hardened
completely using small cooling rate
Carburizing of low-carbon steel (1015, 1020, etc.)
The side that

cooled slowly
The side that

cooled quickly
quenching

tive to the cooling rate during solidification. The
heating and cooling rates within the casting are
uneven because of the existence of the coarse
dendritic grains. Thus, the larger the casting, the
greater the residual stress will be. On the other
hand, residual stress will increase if the workpiece shape is favorable for the occurrence of
inhomogeneous shrinkage. For example, the
concave surface of the sliding guide often appears in cast-iron lathes.
Influences of Methods of Castings. Blanks
made by pressure casting or centrifugal casting
are expected to possess sound structure, reduced
segregation, and fine primary grains, all of which
are beneficial to reducing heat treatment distortion.
Stress Relieving Treatment and Pre-Heat
Treatment. Stress relieving, which is performed
to stabilize the shape and size of the casting, can
eliminate residual stress within the casting. The
tendency of the distortion in casting can be reduced by the pre-heat treatment. For example,
intercrystalline segregation can be improved by
diffusion annealing; the primary coarse grains
can be refined by annealing or normalization.
However, improvement of banded segregation
and eutectic morphology cannot be reduced by
pre-heat treatment.
Hot-Rolled Steels or Forgings. Influences of
the original structures of hot-rolled steels or
forgings are as follows.
Zone segregation in the steel ingots cannot be
eliminated completely by rolling or forging,
though the shape of the segregated zone possibly
can be changed. For example, square-shape segregation often appears in the cross section of hotrolled steel. Therefore, heat treatment distortion
will be intensified because of this segregation.
Segregation of Carbide. The morphology of
ledeburite in the ingots of high-carbon and high-
alloy steel can be changed by hot rolling or hot

forging. However, carbide segregation cannot be
eliminated completely. The carbide morphology
varies with the amount of the deformation
caused by hot rolling or forging and the methods
of forging. Therefore, it can be divided into different carbide-segregated grades according to the
morphology of carbide.
Carbide segregation not only causes inhomogeneous carbide distribution, but also leads to
the increase of the alloy elements and carbon
contents in the austenite near the carbide-rich
zones. This leads to inhomogeneity of heat conduction, and martensitic transformation becomes
asynchronous, increasing the tendency of heat
treatment distortion.
Deformation Caused by Banded Segregation
of Carbide. The banded segregation of carbide
is often observed in the rolling and forging of
high-carbon and high-alloy steel. After quenching, large volumetric expansion parallel to the
bands may occur in the workpiece; less expansion, or even contractions, occur perpendicular
to the bands (as shown in Fig. 28). The effects
of direction of the banded carbide on quenching
Fig. 24
distortion of dies in various shapes are shown in

Fig. 28.
Influence of Forging. The segregation of carbide in hot-rolled tool steel with ledeburite structure is found to be beltlike distribution, which
has a significant effect on the quenching distortion. The amount of expansion along the bands
tends to increase, but the increment of the swelling perpendicular to bands is relatively smaller.
Banded segregation can be eliminated by forging
because the grade of the banded carbide can be
reduced greatly by upsetting and drawing alternately and repeatedly, which is favorable to reducing the quenching distortion.
Beltlike structures of ferrite and pearlite in the
hypoeutectoid steels cause nonuniform distortion because of the differences between the
quenching distortion along the bands and that in
the perpendicular direction.
Cold-Rolled Steel Sheet Stampings. The deformation along different directions caused by
quenching is inconsistent because of the fiber
structures existing in cold-rolled steel sheet. The
cold-rolled 1015 steel sheet, for example, elongates 0.2% in the longitudinal direction, yet con-
Examples showing distortion of cavity
Cut after quenching

30
Quenching direction
30
Fig. 25
70
Elastic chuck made of 1080 steel
Fig. 23
Distortion of keyslot of height ruler made of

9CrSi, after quenching in oil
Fig. 26
Schematic of a ball screw made of AISI E52100 steel. S, screw distance

tracts 0.05% in the transverse direction after
quenching at 860 C (1580 F). Having been
quenched at 80 C (176 F) in hot oil (hardening
layer is about 0.25 mm, or 0.01 in.) after carbon
nitriding at 860 C (1580 F), the deformation of
the forging-formed 1015 steel ferrule is shown
in Fig. 29. The figure shows that the ellipticity
approaches 0.005 to 0.10 mm (0.00020.0039

in.) because of the deformations caused by
stretch forming. The fibrous orientation is
changed along the arc and thus results in different deformations in the different diametrical directions, which leads to an evident ellipse. But
on the same conditions, the ellipticity is below
Shape before carburizing
220
Fig. 27
Schematic of a high-speed heavy-loaded wheel
0.04 mm (0.0016 in.) if the ferrule is made of

1015 steel sheet with fine equiaxed grains.
Influence of Heat Treatment on the

Final Shape of Component
The influences of pre-heat treating and original structure on the heat treatment distortion are
shown in Fig. 30. The prenormalizing or quenching and tempering can refine the crystal grain of
the original structure. Thus, uniform microstructure can be obtained to reduce the distortion of
the component when quenching. Especially under rapid heating conditions of such processes as
high-frequency quenching, it is necessary to heat
the component to a higher temperature if there
is a large block of ferrite in the original structure.
But the component can be quenched to obtain
the suitable hardness at a lower temperature if it
is prenormalized or quenched and tempered
properly. This is useful to reduce the distortion
of the component after high-frequency quenching.
If there are beltlike ferrite and pearlite in the
original structure, the distortion of those parallel
to the beltlike structure is quite different from
those perpendicular to it. For example, if a sample prepared by 0.13% C steel is quenched from
860 C (1580 F), the elongation parallel to the
lamellar structure is 0.17%, while the contractibility along the direction vertical to the lamellar
structure is only 0.015%. However, the size variations for both directions are similar (0.04%
and 0.025%, respectively) after being normalized.
If quenching and high-temperature tempering
are employed as pre-heat treating, the final distortion of the semi-finishing product can be reduced. Moreover, the scattering of distortion can
be limited to a relatively small scope. When the
semi-finishing gear wheel is quenched and tempered before high-frequency quenching, the
structure can be refined and uniformed so as to
be transformed to uniform austenite quickly
when induction heating. This process is useful
in reducing distortion.
The specific volume of globular pearlite is
larger than that of lamellar pearlite. The volume
1.75
25
Fig. 28
Influence of banded segregation of carbide on quenching distortion of workpieces made of high-carbon and
high-alloy tool steel. Arrows indicate the direction of carbide segregations.
28.5
Fig. 29
Distortion of a forging-formed 1015 steel ferule

change of uniform and fine globular pearlite after
quenching is the smallest among various annealing microstructures of tool steel. The spheroidizing grades of high-carbon alloy tool steel
such as 9Mn2V, CrWMn, and 52100 exhibit
large influence on the thermal straightening
property. When the original structure is uniform
fine globular pearlite, the component is straightened easily when the component has been
quenched to 200 to 260 C (390500 F). Dimensional stability is quite good; that is, the
shape of the component is not changed easily
after the component has been aligned and cooled
to room temperature. If components such as the
long slideway and screw, for which alignment is
necessary, have not been spheroidized and annealed properly, their thermal alignment properties will be very poor.
The cutability of the original structures influences the residual stress caused by machining
and thus affects the distortion of the component
indirectly after being heat treated. For the highcarbon tool steels, the cutability of globular
pearlite is the best. For medium- and low-carbon
steels, however, cutability becomes poorer after
being spheroidize annealed. As for medium- and
low-carbon steels, as well as medium- and lowcarbon low-alloy steels, the cutability is the best
if the original structures are fine lamellar pearlite

and free ferrite that are distributed uniformly
along the crystal boundary of austenite. The cutability will worsen if there is globular cementite
or free ferrite in block form or Widmanstatten
structure or granular bainite in the original structures.
Influence of Heat Treating Method to Tendency of Distortion. For many components,
there are many different methods of heat treatment that meet the requirements for application.
However, the distortions caused by these methods are quite different, as shown in Table 8.
Thus, the selection of a heat treating method is
an important factor affecting the final shape of
the heat treated component. The general rules
are:
The lower the heating temperature, the
smaller the distortion will be. For example,

component distortion is minimal if the temperature can be controlled to 200 to 300 C
(390570 F) when an ion implantation
method is used. But the component distortion
is obvious if kept at a high temperature such
as 1000 C (1830 F) for several hours when
boronizing or metallic cementing to reduce
distortion.
Distortion after the component is quenched
The difference of specific volume

between the original structure and
quenching structure
The influences on the

uniformity of structure
Pre-heat treating and

original microstructure
The influences on the

austenitizing temperature
locally or surface quenched is less than that

of being immersion quenched.
Normally, among various surface-quenching
methods, the higher the heating rate and the
more shallow the depth of hardening is, the
smaller the distortion will be. The distortion
is the least if self cooling after rapid heating
is used. The comparison of several surface
quenching methods is shown in Fig. 31.
For chemical heat treatment, when the types
of penetrating elements are similar, the lower
the temperature or the shorter the time, the
smaller the distortion will be. For example,
distortion of a nitrocarburized component is
less than that of a carburized component. The
distortion of a ferrite nitrocarburized (560 C,
or 1040 F, 24 h) component is smaller than
that of austenite nitrocarburized components
(600700 C, or 11101290 F, 24 h). The
distortion of a ferritic nitrocarburized component is less than that of ferritic nitriding.
Normally, the distortion of an isothermal nitrided component is less than that of twostage nitriding.
Influence of Heating on the Final

Shape of the Component
Influence of Heating Medium. The influence of heating medium on the distortion of a
heat treated component is reflected mainly by
two factors, heating rate and heating uniformity.
Generally, the distortion is less when the component is heated at a low heating rate, while the
distortion increases when the component is
heated at a high heating rate. Distortion is also
less when the component is heated in the medium with good heat transfer uniformity. The
sensitivities of these two factors relate to specific
production conditions. The distortion of a large
hob made of high-speed steel after being heated
The influences on cutting

ability and residual stress
The influences on hot

alignment ability
Fig. 30
Heat treat method
Influence of pre-heat-treating and original microstructure on the final shape of component
Power frequency
quenching
Mid frequency
induction
hardening
Flame quenching
Ultra-high
frequency hardening
Electrical heating
and quenching
High frequency
hardening
Laser quenching
Electron beam
hardening
Big
Fig. 31
Table 8 Deformation tendency of various

heat treat methods
Distortion comparison of various surface quenching methods
Small
Metallic cementation
Boriding
Solid quenching
Carburizing quenching
Nitrocarburizing
Chemical vapor deposition
Local quenching
Induction heating-surface quenching
Flame heating-surface quenching
Nitriding
Nitrocarburizing of austenite
Electron beam heating-quenching
Physical vapor deposition
Nitrocarburizing of ferrite
(Tufftride, TruTec, Springfield, OH)
Nitrocarburizing
Short-time nitriding
Laser heat treatment
Ion implantation
Ion implantation enhanced vapor
phase deposition
Carburizing at low temperature
Deformation
tendency
Very big
Big
Medium
Small
Very small
Slight
Temperature of the
component going
into the furnace
The influences of
heating rules
Fig. 32
The control of
heating rate
Holding temperature
Keeping warm
Holding time
Pre-heating
Pre-heating temperature
Heating temperature
and keeping
warm time
Pre-heating time
Influences of heating rules on distortion
Table 9
Influences of heating media on heat treatment distortion
Medium
Heating rate
Heating uniformity
Gas, various protective

gases or air
Slow
Liquid
Fastest
Depends mainly on the structure of the

heating furnace; depends on the
radiation uniformity at high
temperature and on gas-flowing
uniformity at low temperature
Different faces can be heated evenly.
Melting metal
Fastest
Melting salt bath
Faster
Fluidized bed
Fast
Vacuum
Slowest
The heat transfer coefficients of the

side, bottom, and top are quite
different. The heating uniformity is
bad.
Depends on radiation of heating. The
uniformity is bad when a large
number of components are loaded.
Comments
It is useful to reduce the distortion caused

by the weight of the component with
the effect of buoyancy.
It is useful to reduce the distortion caused
by the weight of the component with
the effect of buoyancy.
The buoyancy is approximately 40% of
the above, so it also can decrease the
distortion caused by the weight of the
component. For an internal electrode
salt bath furnace, the long and thin
component is easily deformed with the
effect of electromagnetism stirring.
2
3
Unreasonable
Fig. 33
Unreasonable
Reasonable
Influence of loading pattern on heating uniformity. (1) Workpiece. (2) Heating elements. (3) Refractory materials
in a vacuum furnace is less than when heated in

salt bath because the heating rate of the hob in
the vacuum furnace is lower. But for the highspeed steel drills with a small diameter being
loaded in batches, the distortion after being
heated in a vacuum furnace is greater than that
in salt bath. The influences of heating medium
on quenching distortion are shown in Table 9.
Influence of Rules of Heating. For the components that deform easily, distortion may be reduced if the heating rate, especially in the phase
transformation temperature range, is reduced.
For example, the distortion of high-speed steel
or high-alloy tool steel components may be reduced when put into the furnace with lower temperature, kept at 800850 C (14701560 F),
and then heated to the quenching temperature. It
is better for these steel components to be preheated at 500 C (930 F) in an air furnace and
800850 C (14701560 F) in a salt bath, and
then pre-heated at high temperature in a different
salt bath. Normally, the pre-heating time is double the heating time.
The influences of heating rate and time on the
heat treatment distortion are complex. Generally,
heating at a relatively low temperature and shortening the heating time properly not only reduces
the distortion caused by the weight of the component when heated, but also reduces the distortion caused by the temperature difference during
the subsequent quenching processes. But for
some alloy tool steels, acceptable amounts of retained austenite can be obtained if the component is quenched at a specified temperature. This
process is called micro distortion quenching.
If the quenching temperature is lower than that
temperature, the volume of the component will
increase. If this is not done, shrinkage will occur.
The quenching temperature may be decreased
when sufficient hardness is obtained by the following methods.
Changing the Original Structure of the Component. The lower-limit temperature may be
used for quenching and tempering hypoeutectoid
steel because of the relative ease of austenite nucleation. This is observed during the high-frequency induction heating process. However, the
component should be heated at a higher temperature to ensure complete austenitization if
block-free ferrite is present in the original microstructure.
In the original microstructures of eutectoid
and hypereutectoid tool steels, the finer and better distributed the globular pearlite is, the lower
the austenitizing temperature will be. For exTable 10 Influences of increasing quenching temperature on the distortion of mold
hole
Type of steel
Distortion of mold hole
Medium-carbon
High-carbon
Low-carbon (carburizing)
Medium-carbon low-alloy
High-carbon low-alloy
High-alloy
Expand
Contract
Expand
Expand
Expand
Contract

ample, the austenitizing temperature of 52100 is
850 C (1560 F) if the microstructure is the normal globular pearlite (the average diameter of
cementite is about 1 lm). The austenitizing temperature may be decreased to 810 C (1490 F)
if the component has been pre-heat-treated to
form super-fine carbide (the average diameter of
cementite is smaller than 0.3 lm).
Changing the Chemical Composition of the
Surface Layer. The component that has been nitrided in advance can be quenched at 700 C
(1290 F) to harden the surface layer. The component can be quenched at 600 to 650 C (1110
1200 F) to obtain a high hardness when it is
being nitrocarburized to form austenite with 1.0
to 2.8% N in the nitrocarburized layer. The distortions of the component with a thin wall after
being treated by these methods are minimal.
The influence of heating on distortion is
shown in Fig. 32. For the mold with a cavity
inside, however, the relationship between distortion and heating temperature is comparatively
complex, as shown in Table 10.
The distortion caused by weight of the part
will increase if the aging period is prolonged unnecessarily. The heating period influences dissolution of carbon and alloy elements in austenite and therefore affects the transformation
stresses.
Heating Equipment and Loading Pattern.
The influences of heating equipment and loading
pattern on the distortion of the component are
described here.
Heating Uniformity. Nonuniform heating will
increase the temperature difference, resulting in
large thermal stresses inside the component.
Generally, uniform heating is easy in a continuous furnace, while uniform heating in a periodic furnace is relatively poor, especially if the
batch is large.
The components being heat treated in a passing-type furnace are relatively small and therefore are easily heated evenly. The uniformity of
heating is fairly poor because the charging basket in a periodic furnace is larger and can load
more components, especially when loaded concentratedly.
For furnaces with working temperatures
above 600 C (1110 F), radiation is the main
heat transfer process. Therefore, the location of
the heating elements relative to the load exhibits
a large influence on the uniformity of heating.
Heat radiation conditions throughout the furnace
may vary greatly. For example, the heat radiation
is intensive near the furnace door and, thus, the
temperature is lower here. In this case, the power
distribution of the heating furnace also influences the uniformity of furnace temperature.
For a combustion-heated furnace, direct flame
impingement onto the component should be
avoided. Radiant tube may be used to improve
heating uniformity.
For furnaces with working temperatures lower
than 600 C (1110 F), such as those used for
solution treatment and agency of aluminum-alloys, convection is the main heat transfer process. The uniformity of air flow is critical with
respect to the uniformity of heating. It is necessary to have a circulating fan with enough power,
an air duct, and a conducting plate to guarantee
uniformity of air flow around the component
during heating.
Racking. The loading pattern (racking) will
promote the uniformity of heating, as shown in
Fig. 33 and 34. The loading pattern that best reduces distortion caused by the weight of the
component should be used. Random loading
leads to poor heating uniformity. For long and
thin components, vertical loading should be used
in a pit furnace. Meanwhile, movement of the
heated parts should be smooth and steady when
removing the component from the furnace to reduce distortion caused by swaying and colliding.
Charging basket and fixtures should have
enough rigidity and hot strength so that their distortion, caused by their repeated operation, does
not distort the component. The selection of the
supporting point also has a great influence on the
distortion of the component, as shown in Fig. 34.
Smooth running of the machine driving system
for uploading and downloading. The machine
driving system for transmitting the component
into or out of the furnace must start and brake
steadily to avoid moving up and down and colliding.
Influence of Cooling on Distortion

The cooling process has the greatest influence
on the final distortion of the component. There
are many complicated factors that affect the distortion, such as the type and the design of the
cooling equipment, the quenchant, temperature
and agitation of the quenchant, quenching
method, cooling or immersing time, and the
pressure in the cooling process.
Type and Design of the Cooling Equipment
The influences of cooling equipment on the
quenching distortion are as follows.
Cooling Equipment. Types of cooling equipment and influencing factors include:
Quenching tank for immersion: volume and
size of quenching tank, fluid flow and agitation condition of quenchant, and fluid maintenance
Gas quenching equipment: cooling gas and
pressure, flow rate of gas, design of dynamics
of flow field, and heat transfer
Gas jet and fog jet cooling equipment: structure of nozzle, site and arrangement pattern
of the nozzles, water pressure and flux, air
pressure and flux, and adjusting and controlling system for pressure and flux
Quenching tanks, including water-based polymer quenching tanks, oil-quenching tanks, stepquenching tanks, and isothermal-quenching
tanks, are the most popularly used quenching
equipment.
The volume of the quenching tank directly affects the heat dissipation. If the tank is too small,
the quenchant temperature will vary considerably, leading to nonuniform temperature distri-
bution during the quenching process, which will

influence quench distortion.
In a normal quenching tank, quenchant flows
from the upper overflow vent, heat exchanger,
and is then returned to the tank by the pump. The
tank design influences the uniformity of the
cooling of the component and thus influences the
quenching distortion.
The volume of the quenching tank shown in
Fig. 35(a) is comparatively small. Cold quenchant is put into the tank from the right bottom
and flush through the component unevenly. This
brings about the nonuniform cooling of the component and increased distortion. Cold quenchant
flows into the tank from the holes of the pipe
located at the bottom of the quenching tank
shown in Fig. 35(b). The quenching uniformity
has thus been improved. There is a quenching
ring in the tank, as shown in Fig. 35(c). The liquid quenchant is sprayed on to the component
from the holes on the liquid sprayer using a highpressure pump. This kind of tank is suitable for
large cylinder components. The cooling of the
component is uniform and component distortion
can be reduced. Several stirrers are arranged at
the right side of the tank shown in Fig. 35(d).
The quenchant flows evenly from the bottom by
means of the baffle. The design of the tanks
shown in Fig. 35(b), (c), and (d) is useful in reducing the distortion caused by the nonuniform
flow of the quenchant and the unidirectional flow
through the component.
Gas-Quenching Equipment. High-pressure
gas quenching has increased very quickly in recent years due to its cleanliness (no pollution)
and minimal distortion. The cooling rate of highpressure gas quenching is less than that of oil,
and it is not necessary to move the component
when using high-pressure gas quenching in a
vacuum furnace. These factors are beneficial in
reducing quenching distortion. However, under
some circumstances, the distortion of vacuum
quenching is more than that of oil quenching or
salt-bath quenching. The distortion of vacuum
quenching is related to cooling uniformity, while
cooling uniformity is related to the hydrokinetic
feature of equipment and the loading of components. For example, in some high-pressure
gas-quenching equipment, gas has been input
and output from the top and the bottom. Two
kinds of methods have been employed. Gas has
been input from the top and output from the bottom; gas has been input from the bottom and
output from the top. Nozzles can move right and
left to improve the cooling uniformity. Nevertheless, cooling of the upper-side and the lowerside of the component is not the same. For the
component with tabular shape, the distortion is
less if it is placed in the furnace as shown in Fig.
36(a), while there is bending deformation if the
component is placed as shown in Fig. 36(b).
When several plates are piled in the furnace (Fig.
36(c)), the flow velocity in every interval increases due to the effect of a narrow slot, which
can cause the uneven cooling of the component
and the final distortion. In the designing of some
gas-quenching furnaces, nozzles are positioned
evenly in the chambers. This improves the
Incorrect
Correct
Incorrect
Incorrect
Correct
Incorrect
Correct
Unreasonable
Reasonable
Unreasonable
Fig. 34
Reasonable
Reasonable
Influence of a supporting point on the distortion of a component
(a)
(c)
(b)
Fast cooling
2
1
5
C
(d)
P
(e)
1. cylindric tank
2. mixer
3. flow-guiding pipe at side
4. flow-guiding pipe at bottom
5. thermocouples
Fig. 35
Structure of quenching tank on the cooling uniformity. C, heat exchanger; P, pump. (f) Source: Ref 6
(f)
1. slipping slot connected with furnace
2. high pressure nozzle
3. quick cooling mesh belt
4. slow cooling mesh belt
5. liquid surface

quenching distortion of large blocks, while it can
result in the uneven cooling of cutting tools that
are quenched in the loading basket (Fig. 36(d)).
Moreover, the gas distribution along longitudinal direction is also uneven and, thus, long and
thin components tend to deform when they are
batch loaded (Fig. 36(e)).
Jet-Hardening and Fog-Quenching Equipment. Factors in jet-hardening and fog-quenching equipment that have influence on the
quenching distortion of a component are as follows.
Design of Nozzles. The sizes of jet angle and
jet stream, the uniformity of jet intensity, the
quality of atomization, and so on all can influence the cooling uniformity of the component.
The positions of nozzles and their relative positions to the component are also important factors that can influence cooling uniformity.
Ring injectors can be used in the quenching
of axle parts. The cooling uniformity not only
depends on the distribution uniformity of jet
stream, but it also is influenced to a great extent
by the racking position of the central lines of the
axle and ring. If the central line of the axle deviates from the central lines of the ring, bending
will occur, as shown in Fig. 37(a) and (b). An
ideal jet-hardening system (Fig. 37(c)), should
meet the following three requirements:
air flux change at different positions. This design can meet the requirement for quenching
of a large batch of small parts.
Computer simulation may be used to optimize
the jet-hardening and fog-quenching process and
equipment design.
Influence of Quenchant
Cooling characteristics of various quenchants
and their influence on quench distortion are dis-
(a)
(b)
cussed in other chapters of this book. The cooling intensity of the quenchant is not only related
to the physical and chemical properties of the
quenchant, but also related to some outside factors, as shown in Table 11. The comparison of
several major quenchants is shown in Table 12.
Generally, the greater the cooling intensity,
the deeper the hardened layer, the greater the distortion will be. However, there is no exact relationship between these variables. This is due to
the influence of the fastest cooling rate, the relative degree of the cooling rates within pearlite
(c)
The axle is vertically hung.

There is an accurate locating system to ensure
that the central line of the axle coincides with

that of the ring.
Component can rotate smoothly in the cooling process.
The pressure and flux of water have intensive
influence on the cooling rate of jet hardening,
while the cooling rate of fog quenching is sensitive to the proportion of air to water. Various
cooling rates can bring about various distortions.
The Influence of the Control System. In jet
hardening, the Grossman Quench Severity factor
may be as high as 6.0 if water pressure is 6 atm,
while the cooling rate under 1 atm is similar to
that of brine solution. The cooling rate of fog
quenching varies, depending on the proportions
of air and water, between the cooling rate water
and air quenching. In this case, the cooling abilities of jet hardening and fog quenching may be
adjusted on a wide range. If quenching equipment is provided with a control system with regulators of water pressure, water flux, air pressure,
and air flux, it is possible to obtain various cooling intensities at different temperature ranges to
reduce distortion optimally.
There are two design schemes for adjustable
jet-hardening and fog-quenching equipment:
Degression of pressure
(d)
(e)
Fig. 36
Effect of high-pressure gas quenching equipment on the cooling uniformity
Fig. 37
Effect of site of spray nozzle on the distortion of axle component after quenching
The relative positions of components and
nozzles are fixed, while such parameters as

water pressure, water flux, air pressure, and
air flux change with time. This design can be
employed to the quenching of large components with regular shapes.
Components are placed on the conveyor using a continuous cooling process, and thus
water pressure, water flux, air pressure, and
(a)
(b)
(c)

Table 11
Factors that influence the cooling intensity of liquid quenchants
Factor
Function(s)
Physical and chemical properties of quenchants

Heat of vaporization
The larger the heat vaporization is, the stronger the cooling intensity. The heat of
vaporization of water at 100 C (212 F) is 2260 kJ/kg while the heat of vaporization
of mineral oil is about 210 kJ/kg. Therefore, the cooling intensity of water is much
more powerful than that of oil.
Vapor pressure
The lower the vapor pressure is, the more difficult the gasification, and therefore the
lower the cooling intensity is.
Boiling point
Lower boiling point brings about the increasing of cooling rate in range of martensite
transformation.
Surface tension
The smaller the surface tension is, the less the stability of vapor film is, and the stronger
the cooling intensity.
Specific heat
The higher the specific heat is, the stronger the cooling intensity.
Viscosity
The smaller the viscosity is, the stronger the cooling intensity will be.
The separation behavior of solute
Vapor film can burst when the inorganic (NaCl, NaOH, Na2CO3, etc.) dissolved in
water-based quenchant is separated at the surface of hot component, which can
at the surface of hot
increase the cooling intensity of quenchant.
component
Inverse-solvent power
Polymer quenchant has the power of inverse-solvent. Polymer film can be formed
around hot component so as to separate water from the component, and thus decrease
the cooling intensity of quenchant. The cooling intensity can be adjusted by changing
the concentrating of polymer.
Outside factors
The temperature of quenchant
The cooling intensity of water solution drops intensively with the increasing of the
temperature and the temperature corresponding to the fastest cooling rate obviously
moves to lower. However, the cooling rate of oil increases slightly with the increase
of temperature due to the decreasing of the viscosity of oil.
Agitation can drop down the stability of vapor film and thus increase the cooling
intensity of quenchant. The higher the flow rate is, the greater the influence. The
uniformity of fluid flow around component has great influence on distortion.
The higher the gas pressure is, the bigger the heat of vaporization, and therefore the
faster the cooling rate.
For spray hardening, the higher the pressure is, the more powerful the cooling intensity.
When component is cooled in cycling gas, its cooling intensity increases with the
increase of gas pressure.
The cooling rate of coarse surface is faster than that of smooth surface. The salt or
borax sticking to component surface can promote the burst of vapor film.
Circulation and agitation
Gas pressure above liquid

The pressure of quenchant
The surface condition of

component
Table 12
Comparison of several commonly used quenchants in cooling rate and distortion
Quenchant
Cooling rate
transformation period and martensite transformation period, or strictly speaking, the cooling
curve. There is a strong relationship between surface heat transfer coefficient, surface-to-arc temperature gradient, and distortion. For example,
the cooling intensity of 10% NaOH solution or
10% NaCl solution is much higher than that of
water and, therefore, the hardened layer is also
much deeper. However, the distortion of the
component being quenched by 10% NaOH water
solution or 10% NaCl water solution is just
slightly less than that of water because the cooling rates of those quenchants within martensite
transformation period are similar. The depths of
hardened layers after being quenched in 50%
NaOH water solution is greater than that of water
but the distortion is less. The reason is that the
viscosity of NaOH water solution with high concentration is large and its cooling intensity
within martensite transformation period is only
1
6 of the cooling intensity of water. The maximum cooling rate and characteristic temperature
of high-speed quench oil with additives are obviously higher than that of straight unaccelerated
oil, while its cooling rate within a low-temperature range is slower than that of ordinary mechanical oil due to the greater viscosity. Therefore, although the depth of hardened layer
quenched by high-speed quench oil is deeper
than that of unaccelerated quench oil, the distortion is similar.
The distortions of a C-ring test specimen made
of 5140 steel (Fig. 38) quenched in various quenchants are shown in Table 13.
Influence of Quenching Methods
Distortion
Large
10% NaCl water solution
Large
10% Na2CO3 water solution
Large
50% NaOH water solution
Large
Water
Large
Polymer quenchants
Medium
25% NaNO3 20% KNO3 20% NaNO2 35% H2O
Medium
Accelerated quenching oil
Small
Alkaline salt bath
Very small
Nitrate bath
Very small
Ordinary mechanical oil
Small
Air
Very small
Cooling in One Medium. The factors that

influence component distortion are quenchant
selection, quenchant temperature, agitation rate,
and uniformity of agitation. The influence of
quenchants on the distortion of cavity of coldforging dies is shown in Table 14.
The selection of quenchant should account for
the following two opposing factors, depth of
hardened layer and quenching distortion. According to the dimension of the workpiece and
037
10% NaOH water solution
Table 13
Quenchant
Water
10% NaCl
1% NaOH
50% NaOH
75% NaOH
Source: Ref 1
The influences of quenchants on the distortions of C-style sample

Temperature,
C (F)
Cycling rate,
m/s (ft/s)
Hardness after
quenching, HRC
Gap opening,
mm (in.)
20 (68)
20 (68)
20 (68)
20 (68)
15 (59)
0.5 (1.6)
0.5 (1.6)
0.5 (1.6)
0.5 (1.6)
0.5 (1.6)
5660
5760
5759
5860
5557
0.39 (0.015)
0.18 (0.007)
0.22 (0.009)
0.12 (0.005)
0.05 (0.002)
12
051
Fig. 38
C-style samples

Table 14
Influence of quenchants on the distortion trend of cavity of cold forging dies

Types of steel
Quenching
medium
Mediumcarbon
steel
Highcarbon
steel
Carburized
low-carbon
steel
Carburized
low-alloy
steel
Medium-carbon
low-alloy
steel
Low-alloy
tool steel
Mediumalloy
steel
Highalloy
steel
Saline water
Oil
Alkaline salt
Nitrate bath
Air
Expand
Shrink
Expand
Expand
Shrink
Shrink
(b)
Expand
Shrink
Shrink
Shrink
Shrink
Shrink
Shrink
Shrink
Shrink
Expand
Expand
Expand
(a)
Expand
(b)
Expand
(a)
(b)
(a)
(b)
(a) For alloy tool steel, hot oil can increase the expanding trend, while cold oil can increase the shrinking trend. (b) Too much water in alkaline salt
or nitrate bath, decreasing the bath temperature, and short keeping time will result in the expanding trend of cavity. Source: Ref 4
Table 15 Influence of various stirring degrees on the cooling intensity of quenchants

(Grossman factor, in.1)
Agitation degrees
No agitation
Mild agitation
Moderate agitation
Good agitation
Strong agitation
Violent agitation
Air
Oil
Water
Saltwater
Nitrate
0.008
0.2
0.250.30
0.30.35
0.350.40
0.400.50
0.500.80
0.801.00
0.91.1
1.01.1
1.21.3
1.41.5
1.62.0
4.0
2.0
2.2
45
0.5
2.25
Source: Ref 5
Table 16
Factors that influence distortion of a component during marquenching
Factor
Function(s)
Cooling rate above isothermal temperature

Isothermal temperature
above martensite point
below martensite point
Isothermal time or keeping time
Affecting thermal stress obviously

Affecting structural stress obviously
Affecting the quantity of residual austenite
Having certain influence on structural stress
Affecting the quantity of retained austenite
Intensively affecting structural stress
Affecting the quantity of retained austenite
Cooling rate below isothermal temperature
Table 17
Factors that affect the distortion of spray hardening
Factor
Function
Water pressure
Water flow
Injector or nozzle design
Spray hole or nozzle position
Component shape and relative movement against nozzles
Table 18
Intensively affects cooling intensity

Affects cooling intensity
Affects cooling uniformity and rate
Affects cooling uniformity
Affects cooling uniformity
Distortion tendency of gears after carburizing and quenching
Gear parameter variation
Addendum circle
diameter
Bore diameter
Surface warpage and

taper
Tooth profile
Helix angle variation
Distortion tendency of gears after carburizing and quenching
Addendum circle diameter of plate gear generally expands.

Addendum circle diameter of gear shaft generally shrinks.
Bore diameter generally shrinks. The amount of shrinkage increases with hardenability and case
depth.
Inner diameter of gears with anticarburizing bore expands slightly.
For gears with nonuniform cross section, the bore shrinks more significantly in the thinner part,
resulting in conical deformation of the bore.
For disk gear with large outer diameter/height ratio, its end surface warps easily and gear rim
tapers off easily.
End surface of bevel gear warps.
Tooth thickness increases after carburizing and quenching, particularly as approaching tooth tip
or both end faces.
Helix angle of helical gear after carburizing and quenching becomes smaller.
hardening ability of the steel, quenchant with a

relatively lower cooling intensity should be selected to reduce distortion, provided the hardness
after quenching is guaranteed.
Temperature of Quenchant. The temperature
of aqueous quenchant can affect its cooling character intensively. Increasing the temperature of
the quenchant will increase the stability of vapor
film and reduce the cooling rate in the range of
pearlite transformation, while only decreasing
slightly the cooling rate in the martensite transformation range. Therefore, increasing the temperature of the aqueous quenchant is not a good
method to control component distortion. Although the boiling point of a quenching oil is
typically relatively high, increasing the temperature of quenching oil usually will not obviously
affect the stability of vapor film but can decrease
the viscosity of oil. In this case, thermal stress,
transformation stress, and retained austenite can
be controlled by adjusting the temperature of the
quenching oil. Thus, regulating the temperature
of quenching oil is an effective method for controlling distortion.
Agitation Rate. It can be seen from Table 15
that agitation has a great influence on the cooling
intensity of the quenchant and thus affects the
depth of the hardened layer and quenching distortion. Some good quenching tanks have an adjustable circulation rate, and it is possible to
change the cooling rate rationally. For example,
the following technical method can reduce the
quenching distortion without influencing the
depth of the hardened layer:
1. Moderate agitation rate should be used when
the temperature of the component is above
650 C (1200 F).
2. Strong agitation should be used when the
temperature of the component is between 450
and 650 C (840 and 1200 F).
3. Agitation slows when the temperature of the
component is below 400 C (750 F).
4. Stop agitation at the final period of cooling.
Uniformity of Circulation. Nonuniform stirring will create the difference in flow rate between the two sides of the component and thus
cause the nonuniform cooling of the component,
resulting in an increased amount of distortion of
the component.
Two-Step Quenching. The component is
quenched in quenchant with strong cooling intensity during pearlite transformation period and then
quenched in medium with weak cooling intensity
during martensite period. An example of a twostep quenching is first, the component is quenched
in salt water and then oil. The factors that affect
the distortion of a component are type of quenchant and cooling time in those two media.
Pre-cooling quenching is a commonly used
method to reduce distortion. It includes delayed
quenching, isothermal precooling quenching,
quenching for decreasing temperature after carburizing, and so on. The factors that influence
pre-cooling quenching on the distortion of a
component are:

Time: Quenching distortion often can be re-
before being placed into the quenchant. Both

isothermal temperature and keeping time influence the distortion of the component.
Temperature and holding time of quenching
for decreasing temperature after carburizing:
It can effectively reduce the distortion of a
carburized component if the component is
kept in a carburizing furnace for a certain period of time when the temperature of the furnace is dropped to approximately 840 to 870
C (15401600 F) (depending on type of
steel).
Marquenching is an effective way to reduce
quenching distortion. In this process, the component first is quenched into an isothermal bath.
After its temperature becomes uniform, the com-
Temperature, C
duced if the component is kept in the air or

protective gas for a certain period when moving from the heating furnace into the quenchant. The longer the keeping time is, the less
the distortion will be. However, the keeping
time can be too long so as not to affect the
hardness and the depth of the hardened layer
of the component.
Temperature and time of isothermal precooling: It is useful to decrease the temperature
difference between the inside and outside of
the component and thus reduce distortion. To
be sure the component is isothermal, it should
be kept in a fluid bed or salt bath or an isothermal furnace for a certain period of time
Ms
2
1
Time, lg
Fig. 39
Marquenching and isothermal quenching
ponent is cooled to room temperature. There are

two types of marquenching. In the first type, the
component is quenched into a bath in which the
temperature is above the martensite point, and in
the other, the quenching bath temperature is below the martensite point. Curve 1 in Fig. 39 indicates marquenching below the martensite
point, while curve 2 indicates broken hardening
above the martensite point. The factors that affect the distortion of a component are shown in
Table 16.
Isothermal Quenching. The cooling curve of
isothermal quenching is shown as curve 3 in Fig.
39. The microstructures after isothermal quenching are bainite with a certain quantity of retained
austenite. In this case, there is only a small variation in specific volume, which results in a small
amount of the distortion of the component. The
factors that influence isothermal quenching distortion are isothermal temperature and time.
Spray Hardening. A significant cooling rate
can be achieved when high-pressure water is
sprayed onto the surface of the component, making spray hardening suitable for large-sized components. The cooling intensity of spray hardening can be changed in a wide range by adjusting
the pressure and flow of water. With the help of
computer simulation of transient temperature
field and phase transformation in the spray hardening process, nozzles can be positioned and
pressure and flow of water can be regulated so
that a better hardening effect can be obtained.
This method is generating increased interest. The
factors that affect the distortion of spray hardening are shown in Table 17.
Influence of Tempering, Deep

Freezing Process, and Stabilization
Before tempering
After tempering
Influence of Tempering. There exist some

factors that will cause a volume change in
quenched steel during its tempering process. In
the tempering process, while the carbide is separating from the martensite, the carbon concentration of the -phase decreases, causing contraction. Meanwhile, the transferring of the
retained austenite causes expansion. Normally,
the former takes greater effect than the latter, so
the final volume changing tends to be decreased
(compare with quenched state). However, the
temper process should not be expected to make
up the distortions caused by annealing. Initially,
the tempering temperature is insufficient, which
can totally reverse the volume change caused by
the quenching process. More importantly, the
plastic distortion caused by inner stresses during
the quenching process cannot be changed by
tempering. The distortion incurred during temper is far less than that in quenching. Therefore,
the final distortion after the quench-temper process is decided primarily by the distortion in the
quenching process.
Factors that influence the distortion in the
temper process are:
Temperature: At higher temperatures, where
more obvious contractions occur for carbon
The higher nitrogen content of

the layer, the larger deformation.
The thicker nitrided layer,
the larger deformation.
The volume of nitrided

layer expands.
The more nitride formation

elements the steel contains, the
larger deformation.
The higher nitriding temperature
and longer nitriding time, the
larger deformation.
The higher nitriding potential,
the larger deformation.
The worse uniformity of nitriding
layer, the larger deformation.
Internal stress
before nitriding.
Effect of residual stress caused

by machining.
Effect of the prestress relieving
Distortion of
nitrided workpiece.
Effect of loading method

Creep caused by the
weight of workpiece.
Effect of fixture design

The thinner wall thickness, the
larger deformation due to the
volume expansion of nitrided layer.
The greater difference of wall
thickness, the larger deformation
during nitriding.
Shape factor
The worse shape symmetry, the
larger deformation during nitriding.
The worse rigidity, the larger
deformation due to creep.
Fig. 41
Distortion of tube-type parts after carburizing and quenching
Effect along the direction

of gravity.
Effect on the gravity
of workpiece.
Effect of loading tools

on heat treatment
distortion
Influence of the Deep Freezing Process.

When quenched steel is cryogenically treated,
the retained austenite will transform to martensite. Thus, the size of the workpiece will have
only a little expansion, and the size stability of
the workpiece will be raised.
Influence of Stabilization. Temper makes
the microstructure of steel stable and reduces the
residual stress (or make the stress state stable),
which can improve dimensional stability. In
most situations, the dimensional stability can
be obtained only through a temper process.
Some precision tools tempered at a low temperature, however, need to be heated several
times, for an extended period, and kept at a
temperature slightly below the temper temperature. Low-temperature aging can reduce
the residual stress while stabilizing the retained austenite to increase dimensional stability of precision tools.
Factors Affecting Chemical Heat

Treatment Distortion
Effect on the heating

uniformity.
Factors Affecting the Distortion of a Carburized Workpiece. Besides shape, quenching

methods, quenchant selection, and hardenability,
other factors relevant to the carburizing process
also can influence the final size and shape of the
workpiece significantly:
Effect on the cooling

uniformity.
Surface carbon content and carbon concen-
Effect on the homogeneity

of chemical heat treatment
case.
Effect on the stability of

loading/unloading.
Fig. 42
Effect on the weights

of other workpieces in
the same furnace.
steel and low-alloy steel, the volume can

shrink back to what it was before the quenching process when tempering at a temperature
above 500 to 600 C (9301110 F). For the
high-alloy tool steel, however, after the temper process at the temperature where the second hardening occurred, the amount of the
volume contraction will be reduced due to the
transferring of the retained austenite.
Time: The volume contraction is mostly being processed in the initial period of the temper process and will begin stabilizing after 1
to 2 h.
Pressure: When the workpiece is being
pressed during the temper process, its shape
can be fixed due to the function of its elastic
stress being relaxed. The typical example is
the tempering of a piston ring. The method
involves keeping the notch opening by iron
mass during the temper process. Under the
same principle, the workpieces can be flattened after the temper process, during which
they are pressed (Fig. 40).
Effects of loading tools on heat treatment distortion
tration distribution of carburized layer: Surface layer with high-carbon content has a
lower martensite-transition point, greater
amount of retained austenite, and larger volumetric increase during martensite transformation. Transformation stress of carburized
parts varies with case carbon content and carbon concentration distribution in the carburized layer.
Case depth of carburization: The deeper the
case depth of carburization, the greater the

transformational stress after quenching, and
the greater deformation of the parts
Control accuracy of carbon potential: The
more accurate the control of carbon potential
is, the more consistent the deformation of carburized parts will be.
Carburizing temperature and time: Deformation caused by creep increases as carburizing temperature and time increase, and the influence of temperature is more significant.
Therefore, for carburizing parts with thin wall
or shallow case depth, the carburizing temperature should be lowered.
(a)
Fig. 43
Quenching methods after carburizing: Dis-
Heterogeneous case depth or local carburiz-
tortion of carburized parts through direct

quenching is usually less than that through
reheating and requenching because of less
times of heating and cooling in the former
process.
Core hardness of carburized parts: Core
hardness reflects the degree of hardening.
With the increase of core hardness, the deformation changes from being caused by thermal stress to that being caused by transformational stress.
ing: Heterogeneous case depth or local carburizing will result in nonsymmetrical distribution of structural stress and, therefore, lead
to distortion.
(b)
(c)
(d)
Comparisons of the distortion of slender shaft resulting from gravity. (a) Large distortion. (b) Medium distortion.
(c) Little distortion. (d) Illustration of universal joint hanger
16
These factors, along with the size and shape of

the parts, result in the distortion of carburized
and quenched parts. For example, distortion patterns of tube-type parts after carburizing and
quenching are described as follows.
Distortion of low-carbon steel or low-carbon
alloy-steel-tube-type parts with a wall of 5 to 15
mm (0.200.60 in.) thickness after carburizing
and quenching is shown in Fig. 41. The bore
tends to shrink. Larger bore diameter and deeper
case depth will result in greater shrinkage of the
bore. The height and outer diameter of tube-type
parts are generally reduced slightly (their reduction is smaller than that of the bore). When the
bore has key grooves, carburizing and quenching
will result in the width reduction of the key
grooves.
Distortion tendency of gears after carburizing
and subsequent quenching is illustrated in Table
18.
Factors Affecting the Distortion of Nitrided Workpiece. Nitriding is a method of heat
treatment that causes the least distortion. In industry, nitriding is used on many important precise machine components, which have rigid requirements of deformation. Factors influencing
the distortion of nitrided parts are described subsequently.
Effects of Auxiliary Equipment and

Loading Methods on the Final
Shape of Components
100
12
Effects of loading tools and arrangement patterns on the final heat treatment distortion are
20
90
(a)
(b)
(c)
Fig. 44
Supporting pattern of shaft gear in furnace
(d)
Do
Di
3 120
Incorrect
Fig. 45
Loading methods of a gear
Correct

shown in Fig. 42. Charging batcher, supporting
frame, and hanging tools determine the direction
and magnitude of thermal stress and affect the
transportation stability during uploading and unloading. Moreover, many factors, such as heating uniformity, flow resistance of furnace gas
during chemical heat treatment, and flow state of
quenchant during cooling, are related to the loading patterns. Therefore, the design of loading
tools has evident influence on the heat treatment
distortion, which is discussed in the following
paragraphs.
Effect of Loading Tools on Distortion
Caused by Gravity. Creep caused by gravity
during heating is one of the significant causes of
heat treatment distortion, and loading tools have
decisive effect on it. The effects of furnace tray,
supporting frame, and hanger on heat treatment
distortion are shown schematically in Fig. 42.
The distortion caused by gravity can be eliminated if the hanger is used to keep the slender
workpieces in a vertical state. In Fig. 43, deformation of (a), (b), and (c) decrease successively.
In Fig. 43(d), a universal joint hanger is designed
to ensure the workpiece is suspended vertically.
The shaft gear illustrated in Fig. 44 is carburized and then quenched in trays. Different types
of supporting patterns can affect distortion significantly. Placing the shaft gear in the horizontal
position on the base plate, as shown in Fig. 44(b)
and (c), will cause large deformation due to gravity. In Fig. 44(d), the stem of 16 mm is plugged
in the grid of fixture, surrounded with three
height-adjustable supporting bolts to ensure it is
in the vertical position; the distortion will be reduced greatly.
For normal disk-shaped parts with through
hole in the center (Fig. 45), suspension from the
center hole can reduce deformation. However, if
there is only a small difference between the outer
diameter (Do) and the inner diameter (Di), suspension from the center hole will cause large deformation (it may become elliptic). In this case,
other options should be used, such as hanging
from around the outer circle of the part, or just
placing it in the horizontal position on a flat, stiff
base plate (Fig. 46).
A disk gear for mass production should be
placed in the horizontal position, so as to make
full use of furnace chamber space and improve
productivity.
Influence on Heating and Cooling Uniformity. If the workpieces are placed disorderly in
baskets, the heating and cooling uniformity will
be greatly affected. For example, during carburizing and quenching automobile gears in a sealing chamber furnace or continuous furnace, special baskets are designed for distributing the
parts evenly and allowing appropriate space between them (Fig. 47).
The four sides of the basket can prevent the
heater from irradiating the parts directly, and act
as diversion of circulated furnace gas and quenchant during cooling, which are in favor of improving heating and cooling uniformity.
The bottom of the basket should have good
stiffness to make it more stable, which can also
80
30
1
30
67
2
57
Fig. 48
Auxiliary sleeve used for reducing quenching distortion. (1) workpiece, (2) auxiliary sleeve
(a)
Fig. 46
(b)
(c)
Loading method of ring-shaped parts with a large hole in the center (a) circumferentially hanging, (b) centrally
hanging, and (c) horizontally placed
4
5
2
1
Fig. 47
Schematic layout of basket and parts. (1) pedestal, (2) basket, (3) cramp bar, (4) locating pedestal, (5) parts
NH3
3
1
2
4
A-A
Fig. 49
Schematic of the unimproved layout of bolt nitriding. (1) furnace chamber, (2) ammonia inlet, (3) parts, (4)
heater
be designed as a dismountable locating pedestal

if necessary. The surface of the locating pedestal
and parts should be machined to be smooth.
For some easily deformed parts with walls of
heterogeneous thickness, properly designed auxiliary fixtures can improve heating and cooling
uniformity, and therefore, reduce heat treatment
distortion. As mentioned previously, while
quenching of 1045 steel stop ring, a sleeve underneath during quenching can significantly reduce distortion (Fig. 48).
Effect of Loading Methods on the Uniformity of Chemical Heat Treatment. During
chemical heat treatment, loading tools and loading methods will affect case homogeneity, which
is also an important cause of heat treatment distortion. Well-designed baskets and properly distributed workpieces can make furnace gas flow
evenly, and therefore improve case homogeneity, as shown in Fig. 48. For another example,
in unimproved furnaces, as shown in Fig. 49, the
bolts used in a plastic machine are nitrided in a
pit furnace. Gas flow around the bolt is rather
heterogeneous because of the small size of the
furnace chamber, unevenly distributed interface,
and one side of the bolt being closer to the furnace wall than the other. As a result, bending
deformation of nitriding bolts is beyond the tolerance limit. In the improved design, as shown
in Fig. 50, the following measurements are
adopted:
Enlarging chamber volume
Installing centrifugal impeller on furnace lid
Adding a diversion tank in furnace chamber
3
6
These improvements greatly improve atmosphere and heating uniformity, and therefore, the
distortion is controlled within the tolerance limit.
Function of Forced Distortion Devices in
Heat Treatment. Some devices employed in
quenching can control the workpiece distortion
by use of internal stress emerged in heat treatment, though they exert no external force on the
workpiece. For example, in Fig. 13, a workpiece
with a groove in the end face was quenched with
the groove upward, and groove width shrank evidently. If a smooth steel block is plugged in the
groove, which is heated and quenched with the
parts, and then taken out after subsequent tempering when stress relaxed, distortion can be
greatly reduced.
In actual production processes, to avoid bending distortion, quench press machines are used
to quench parts such as gears, friction plates, and
even butterfly springs formed during quenching.
Moreover, for mass production of disk gears,
quench press machines are also applied to the
quenching treatment.
The disk gear shown in Fig. 51 was quenched
in a pulsating quench press machine. At the same
time, end face warpage and bore distortion were
corrected. The compressive forces were (Ref 4):
Ring pressure at position B: 11,740 N (2,639
lbf)
Fig. 50
Schematic of the improved layout of bolt nitriding. (1) guide flow tank, (2) furnace chamber, (3) guide flow
ring, (4) annular ammonia inlet, (5) furnace lid, (6) fan
Pressure at position A: 16,003 N (3,598 lbf)

Expansion pressure at position C: 3,910 N
(879 lbf)

All of the properties and quality of the quench
press machine, the design and manufacture quality of the compression die, the magnitude of
compressive force, and so on, will contribute to
the distortion of the workpiece quenched by a
press machine, as shown in Table 19.
32
REFERENCES
18
110
220
Fig. 51
Table 19
Disk gear quenched in pulsating quench press machine. (A) pressure from outer hold-down ring, (B) pressure
from inner hold-down ring, (C) pressure from expansion heap, (D) pressure from bottom die ring
Factors affecting workpiece distortion quenched by press machine
Factor
Unsuitable fit between conical expansion rod and cone face of central die
Untidy work surface of central die or gear bore surface
Inadequate dimension of central die and limiting ring or inappropriate dimension of limit
ring
Insufficient pressure of central die
Excessive pressure of central die or oversized inner diameter of limiting ring
Insufficient pressure of inner hold-down ring
Excessive pressure of outer hold-down ring
Excessive taper of lower mold surface
Insufficient pressure of outer hold-down ring
Excessive pressure of inner hold-down ring
Inadequate taper of lower mold surface
Source: Ref 4
Influence
Elliptical bore
Elliptical bore
Elliptical bore
Elliptical bore
Nonsmoothness of inner end surface
Nonsmoothness of outer end surface
1. Z. Peiyu, The Control of Heat Treamtment

Distortions of Common Machine Parts, Mechanical Industry Publisher, Beijing, China,
1990
2. X. Dawei, F. Zhijun, and X. Huizhi, The Heat
Treatments of Fine Long Parts, Mechanical
Industry Publisher, Beijing, China, 1991
3. Y. Lunian, The Control of Distortion in Steel
Heat Treatments, Mechanical Industry Publisher, Beijing, China, 1987
4. Heat Treatment Handbook, 2nd ed., Vol 2,
Heat Treatment Institute of China Mechanical Engineering Society, Beijing, China,
1991
5. Heat Treating, Cleaning and Finishing, Vol
2, Metals Handbook, 8th ed., American Society for Metals, 1964, p 18
6. G.E. Totten and M.A. Howes, Steel Heat
Treatment Handbook, Marcel Dekker Inc.,
New York, 1997

DOI: 10.1361/hrsd2002p183

Effects of Process Equipment Design

F.T. Hoffmann and T. Lubben, IWT Foundation Institute for Material Science, Germany
R. Hoffmann, IVA Industrieofen, Germany
K. Heess, Karl Heess GmbH, Germany
IN MANY CASES, fabrication of parts ends

with a heat treatment process to set the required
properties. Unwanted effects that may appear after the heat treatment process are dimensional
alterations and distortion as a result of a processinduced inner stress state of the parts. The origin
of the stresses may be in each part of the manufacturing process of the parts (i.e., construction,
production process of the material, manufacturing processes, and heat treatment processes) between and at the end of the production process.
Further on stresses generated in different stages
of the production process can additively or compensatively superpose during the manufacturing
process. At the last heat treatment will be the
production process that will release the residual
stresses as distortion. Distortion can increase to
such an extent that the parts can no longer be
used, or that at least reoperation processes will
be necessary. These additional necessary processes will be costly because they have to be
done in the high-hardness state after finishing the
heat treatment processes. Accordingly, more or
less material has to be charged so that the distortion, which is to be expected, can be adjusted
by, for example, partial grinding operations.
The exact knowledge of the overmeasures is
of high importance because too small additional
measures will lead to defective goods. On the
other side, too high overmeasures will increase
the costs. The results are higher grinding costs
and an increase of the heat treatment costs. For
instance, carburizing time has to be increased to
set a higher carburizing depth. Another problem
is that distortion will differ from part to part and
that the scattering of the values is based on multifarious reasons.
Distortion as a result of the manufacturing
process is not a new problem. It is documented
in literature broadly and has been the subject of
research work in various technical fields. However, the state of the art is very inhomogeneous
and conclusions cannot be made, or only with
unacceptable uncertainties.
A new starting point to minimize distortion is
the awareness that distortion is not only the problem of the material or a single step in the manu-
facturing process but a property of the whole

manufacturing system. Thus, it has to be seen as
a system property of the whole manufacturing
process, including:
Construction: Construction prearranges the
distortion potential. This means the liability

to changes in shape and form (e.g., by
changes in cross sections materials selection).
Steel producing: Melting and solidification
and the following deformation processes appoint the local inhomogeneity of the chemical
composition, which result in a local different
transformation behavior. This leads to different microstructures and so to a inhomogeneous distortion behavior, in extreme cases,
to crack formation.
Forming: The state of microstructures and residual stresses in a part is influenced by the
state of the material (chemical composition,
inhomogeneity of the microstructure, segregations, etc.) before forming and by the deformation parameters. They influence the following machining operations and the
subsequently appearing distortion and dimensional alterations:
Machining: Machining (turning, milling,
drilling) adds specific distortion potentials to
the distortion system by generating mostly inhomogeneous residual stress states.
Heat treatment: Heat treatment implies a specific distortion potential. As a finishing manufacturing process, it releases the total distortion potential, accumulated as residual
stresses, inhomogeneities of material, and microstructure up to this moment, as dimensional alterations and distortion. The origins
of distortion are quite complex and connected
with the total manufacturing process and
should not be attributed solely to the heat
treatment process.
Therefore, to minimize distortion, the whole production process has to be observed. In the following chapters, the heat treatment process, and
especially the influence of process equipment,
are discussed
Identification of Distortion
Generating Process Equipment
Distortion can be increased by different steps
of heat treatment processes. In the majority of
cases, quenching steps are brought into connection with the generating of distortion because
quenching is the last step of the process. However, a high amount of distortion will be generated before quenching or cooling is started. To
avoid or to minimize distortion, it is important
that the parts being heat treated are heated and
cooled uniformly so that a homogeneous temperature distribution results over all the cross
sections. On the other hand, if cooling and heating are as homogeneous as possible, but the
component is very inhomogeneous (chemical
composition, microstructure, residual stress state
before heat treatment) or the shapes of the parts
show very different transverse sections, bores,
holes, and so on, distortion will not be avoided
without using special methods such as press
quenching. Therefore, if there are problems dealing with distortion and there is the presumption,
that distortion could have been released by
equipment, it should be checked simultaneously
if material or shape could also be responsible for
the failure.
With respect to the process and the equipment,
a general remark is that heating and cooling must
be done as uniformly as possible.
General Possibilities to Influence

Distortion Generation by Process
Equipment
Distortion is generated by elastic and plastic
deformations during the production process of
the part. Origins may be an inhomogeneous microstructure of the material due to its metallurgical history or nonuniform or unbalanced conditions during heating or cooling processes
during the manufacturing process.
The former case cannot be influenced by normal process equipment. Beyond this, these in-

homogeneities are unidentified if the quality
control has not been done sufficiently.
Influences of the Heating System

The insulation of the furnace has to be uniform. Cold spots may lead to an undershooting
of the dew point, which will result in the formation of moisture. Also, the insulation material
has to be fitted to the atmosphere to avoid side
reactions between atmosphere and insulation
material. Otherwise, the local homogeneity of
the treatment may be influenced negatively,
which may lead to an inhomogeneous impact of
the atmosphere to the parts (e.g., followed by
inhomogeneous carbon content in the case of
carburizing or oxidizing in protective atmospheres). In turn, this will influence the transformation behavior and may increase distortion
during the quenching or cooling step of the process.
Heat Irradiating Areas in Ratio to the Size
of the Batch. If parts are located too closely to
heating devices, local differences in temperature
may occur. Different temperatures will influence
locally the carburizing or nitriding behavior in
the case of thermo-chemical processes or in the
transformation behavior if temperature differences result in a different amount of solved alloying elements. These inhomogeneities promote distortion. Therefore, the ideal distance of
heating devices from the batch should be as far
away from the parts that approximately a 15
radiation angle still can be realized. However, it
should not be closer to the parts than 150 mm (6
in.).
If this rule of design is fulfilled, the usable
space will be significantly bigger than the workspace, and bigger parts than allowed for ideal
charging can be used. In respect of ideal heating
conditions for a uniform heat treating a furnace
should not be overloaded. The ideal size of a
charge will be smaller than the potentially usable
space. Thus, an unfavorable charging is preprogrammed, also in respect of the operating efficiency of the process.
A furnace, if not constructed as a tube, has six
sides. Normally, only two sides are stoked. The
temperature homogeneity can be enhanced by
heating more than two sides. However, this will
not be possible for all areas; for example, specific problems will occur at walls with feeding
units. Also, temperature inhomogeneities can be
caused by metallic parts forming a temperature
dip. In addition, each flange worsens the temperature homogeneity. Inner doors are thinner,
in consequence of their function. It has to be considered that the heat loss in areas with doors is
higher than in the better-insulated surrounding
areas.
Number and surface area of the heating devices influence the temperature homogeneity by
the necessary surface temperature to reach the
desired batch temperature. The surface temperature of the heating devices should not exceed that
of the batch temperature by more than 30 K. The
distance to the batch is especially important at

point-shaped heat sources.
Gas Atmosphere. Heat transfer into the parts
is induced by enforced convection. Gas velocity
at the surface of the parts, homogeneity of the
gas velocity at the surface and in the inner regions of the batch, and differences of the velocity
in the inside and outside of the batch will influence the heat transfer and, therewith, the temperature homogeneity. Theoretically, more uniform conditions should be reachable by baffles.
Because of the high temperatures, however, only
simple-shaped constructions can be used. Optimal rules of design cannot be applied.
Thermocouples. The location of thermocouples must allow for correct temperature measurement; the tip of the thermocouple has to be
placed near the workpieces in the furnace. It has
to be exposed to the same heating environment
as these pieces. The uniformity of the temperature in the furnace must be checked before installing control thermocouples.
Special Problems at Thermochemical
Treatments. The revolution of the process gases
serves only for equalization of the atmosphere.
It is not strong enough to manage a good temperature equalization. The limitation of the degree of revolution is given by the strength of the
fan material. Metals have a low strength at process temperature, and carbon fibre composite
(CFC) materials are not usable in conjunction
with oxygen-containing atmospheres. Inside a
furnace it has to be distinguished between lowand high-temperature areas. The limit will be at
approximately 700 to 800 oC (12901470 F).
Above this temperature, radiation will be the
main heat transfer mechanism, and this is the main
reason that revolution will not enhance the temperature uniformity.
Hot and cool spots must be avoided. Hot spots
will result in a higher dissociation of the process
gas; at cold spots, soot can be formed. Local different gas compositions may lead to differences
in element profiles, and soot on the parts will
hinder the carbon transition. Both effects influence the transformation behavior of the material
during quenching and, therewith, distortion behavior. A further influencing parameter on atmosphere homogeneity is the disposition of the
gas-inlet points. Atmosphere uniformity must be
reached before the atmosphere comes into contact with the parts to be heat treated.
As mentioned previously, for the uniformity
of the carburized layer, temperature and atmosphere homogeneity are responsible. However,
the choice of treatment temperature also may be
of high influence (e.g., if parts with differences
in cross section should be carburized or carbonitrided to small carburizing depths). High temperatures in connection with those geometries
can result in high temperature differences. As a
result, overcarburized spots and spots with toolow carburizing depths will occur. In these cases,
low carburizing temperatures offer higher uniformity of carburizing depth, but longer treatment times are necessary. Additionally, large
sections of the parts stay ferritic during carburizing and will not be hardenedonly the sur-
face near cross sections that will pick up carbon

transform to austenite.
Special Problems of Pit-Type Furnaces. Atmosphere homogeneity in a furnace is a prerequisite to avoid unequal case conditions during
thermochemical treatments and so prerequisite
to avoid distortion. To reach a good homogeneity the flow field must be optimized. In the
case of pit-type furnaces, this can be done by
guide cylinders. If the length to diameter ratio of
the furnace is small (1:1 or 2:1), there is no need
for such a measure. But, at higher ratios, guide
cylinders are necessary to receive a uniform atmosphere distribution. Further improvement at
higher temperatures (e.g., carburizing) is achievable by usage of gas lances. At low temperatures
(e.g., nitriding), stronger fans can be used.
Special Problems of Vacuum Furnaces.
Temperature homogeneity is predominantly determined by the disposition and layout of the
heating elements. The number of the heating devices is substantial for the equability but also for
the costs. Thus, depending on the size and construction of the furnace, a minimum number is
necessary. Charging with large distances between the parts in too-large furnaces will be
good for homogeneous heating but worse with
regard to the costs.
Influences of the Quenching System

Quenching is the most dangerous process step
of heat treatment to generate distortion. Thus,
nonuniform agitation/quenching or nonuniform
circulation of quenchant around a part results in
inhomogeneous cooling rates that create shape
distortion. The extent of distortion also increases
with the intensity of cooling. As discussed previously, however, uneven hardening, with the
formation of soft spots, increases distortion.
Similarly, uneven case depths in case-hardening
steels have the same impact.
As a general rule, cooling should be done as
slowly and as homogeneously as possible. When
quenching in fluids, there should be a good revolution of the quenching media also within the
batch. Influences by charging should be considered as well, for example, the hardenability of
the parts, their form and size, and their distances
from each other (batch density). The composition of the charge will influence distortion
greatly, too.
Forces by the self weight of the parts can deform the parts if they are not held up sufficiently.
In dependence of the shape of the parts, weight
forces are dependent on the manner of charging.
Long, slim parts, therefore, should be hung. If
parts have to be laid down, they must be supported sufficiently.
Grids. Metallic grids will not stay even,
which also can influence distortion by weight
forces (gravity). In consequence of uneven grids,
the charge will not stay in a plane. When distortion of a grid increases, it can be turned. Dense
charging will cause the same behavior as the behavior of a large compact part. Therefore, the
inner areas of the batch will be cooled down
more slowly than the outer regions.
Effects of Process Equipment Design / 185

The best possibility to adapt quenching to the
requirements of the parts or the batch is given
by gas quenching. By gas type, pressure, flow
rates, and other influencing parameters, the
quenching rate can be adapted to the parts. However, there are a lot of additional parameters that
influence quenching behavior and therewith distortion, that are not fully recognized at each process. Some of these influences are:
Heat radiation emitting and absorbing areas

Inflow of heat-emitting areas
Cooling of the quenching gas
Pressure as function of quenching time
Power of the fan (flow velocity of the gas)
Flow field in the charging area
Local turbulences
The design of the gas flow is essential for the

homogeneity; impinging jets with high velocity
rmax
rmin
Roundness chart of a bearing ring (184 200

25 mm diam, or 7.2 7.9 1 in. diam, SAE
52100) after free quenching in a high-speed quenching oil.
The difference between maximum and minimum radius
(roundness), r, is 0.436 mm (0.017 in.).
Fig. 1
Roundness chart of a bearing ring (184 200

25 mm diam, or 7.2 7.9 1 in. diam, SAE
52100) after quenching in a quenching press. The difference between maximum and minimum radius (roundness)
is 0.057 mm (0.002 in.).
Fig. 2
results in a high heat flow, tangential inflow in

a low heat transfer. The homogeneity of the heat
flow increases with increasing flow rate.
Gas quenching is the only quenching method
that allows adaption of the quenching intensity
to the special needs of single parts or a batch.
Also, different quenching rates at one part can
be realized (e.g., by quenching in a nozzle field).
In this way, distortion, or at least the scatterband
of distortion, can be minimized.
Distortion Minimizing by
Quenching in a Hardening Press
In actual industrial practice, the heat treatment
process frequently is carried out with large
charges. In such operations, many workpieces
are simultaneously heated, then simultaneously
cooled. Such multiple-workpiece treatment procedures have the obvious advantage of substantially increasing output rate; however, the cooling curves of each individual workpiece are
different. Thus, in order to ensure that the required minimum effective cooling rate is
achieved, it is necessary, in such cases, to orient
the procedure on the basis of the slowest-cooling
section of the processed charge.
A corresponding selection of quenching conditions (involving, for example, the use of a
high-efficiency quenching oil) leads, in such
cases, to the cooling of all other portions of the
charge at an excessively high rate. Therefore, in
such a situation, it currently is impracticable to
establish an optimal charge-processing procedure. In cases in which the workpieces undergoing treatment are distortion sensitive (such as,
thin-walled roller-bearing rings), extreme deformation may occur (Fig. 1).
In the current state of industrial technology,
workpieces of such sensitive types can be treated
successfully only through the use of expensive
and time-consuming grinding operations, which
produce workpieces with the required end dimensions. However, some workpieces may be in
such bad shape that they cannot even be subjected to post-production treatment, and in such
cases, the benefits of any previously applied
treatment procedures are lost. One potential solution to this problem is the use of hardening
fixtures.
An essential feature of such an installation is
that each workpiece is subjected to an individual
cooling regimen, while the deformation is maintained at a very low level by applying external
forces. A comparison between Fig. 1 and Fig. 2
shows the accuracy (roundness) generated
through the use of this procedure and clearly indicates the high level of effectiveness of the
hardening fixture installation in procedures of
the type in question.
The primary goal of the fixture-hardening procedure is the enhancement of end-product quality, through achieving a uniform hardness distribution for the processed workpieces, minimal
level of deformation, as well as a substantial reduction in the need for the use of supplemental
finishing procedures, such as grinding.
Fixture hardening is characterized by the fact

that it is carried out through the use of an integrated hardening press unit, into which the
heated manufactured workpiece, either singly or
in batches, is inserted and then subjected to a
precisely regulated, optimized cooling regimen,
while simultaneously being subjected to a specific applied force. For example, in the case of
the quenching of round workpieces, a radial
force is applied to prevent the workpiece from
undergoing any significant changes in shape during the structural-transformation process.
Changes in workpiece dimension, which are
common products of structural-transformation
procedures, in the context of fixture hardening
(which involves the use of applied external
forces) may, within this context, be converted to
variations in the geometric relations (for instance, in the case of ring-shaped workpieces, to
variations in wall thickness). The dimensions of
freely hardened workpieces differ markedly
from those of fixture hardened; in the latter case,
it is possible to adjust workpiece dimensions in
such a manner that the fixture-hardened workpieces uniformly exhibit the required dimensions. However, whatever quenching method is
selected, it is vital to bear in mind the close relationship obtained between the hardening
operation itself and previous production steps.
In the case of rotationally symmetrical, ringshaped workpieces, as a general rule, a mandrel
is employed, which aids in the centering of the
workpiece, as well as the maintaining of the
roundness of the inner ring surface during the
quenching procedure.
Therefore, at the beginning of and/or at some
point during the quenching procedure, an external force is applied to the workpiece. This applied external force leads to the plastic deformation of the workpiece, and by this means,
effectively prevents the workpiece from undergoing the loss of roundness that can be generated
by nonuniform cooling rates of its various crosssectional regions and released stress.
The selection of the specific mandrel system
to be employed in a given operation is determined by such factors as the material of which
the workpiece is composed, its geometrical characteristics, and the hardening process used.
Thus, in cases where the workpiece undergoing
treatment exhibits shrinkage at the conclusion of
the procedure, a solid mandrel can be used (Fig.
3). During the hardening process, the workpieces
shrink on the mandrel and then must be pressed
off. Solid mandrels are simple and inexpensive
to manufacture and yield extremely good shape
results when employed in operations of the type
described.
In order to obtain the higher precision values
required in geared workpieces or cogs, the use
of smooth-surface mandrels can be effectively
supplemented with cog-cut mandrels. In such
cases, the workpiece undergoing treatment can
be positioned in the cog mandrel in either the
tooth flank or the root of the gear tooth (Fig. 3).
Special cogged mandrels also ensure conformity
to dimensional specifications in free-cut cogs or
gears, which, with the use of conventional

smooth-surface or expanding mandrels, can be
partially lost since the free-cut cog or gear regions do not mesh precisely with the mandrel
employed.
Another problem typically encountered in the
hardening of fully hardenable workpieces is a
substantial volume increase as a result of the formation of a martensitic structure throughout the
entire workpiece, which in turn, leads to an undesirable expansion. As a result, during the first
part of the quenching process, the workpiece
shrinks on the mandrel, and when the Ms point
(the temperature at which martensite starts to
form from austenite on cooling) is attained, the
workpiece dissociates from the mandrel again.

As a result, deformation can appear again during
the generation of martensite. Plastic deformation
of the workpiece during the martensite generation cannot be prevented through the use of a
solid mandrel.
In order to eliminate the difficulty just described, an expanding mandrel system is commonly employed, which allows a slight variation
of the diameter (Fig. 3). Through the expansion
of the segments of the mandrel over a cone, the
workpiece, while undergoing martensitic structure generation, has its inner diameter pressed
out round as a result of plastic deformation.
Smooth mandrel
Fixed mandrel
Expanding mandrel
Toothed mandrel
Basic fixing
Fig. 3
Fig. 4
Flank fixing
Digest of actual mandrel system at press quenching
Synchronizing cone
Fixed mandrel receptacle with
retention and mandrel
ejection device
Ring gear
Expanded mandrel receptacle
with separated retention
Synchronizing ring
Receptacle on conical fixed
mandrel with retention, conical
outer retracing, and
mandrel ejection device
Sliding sleeve
Fixed mandrel takeup,
with retention, mandrel
ejection device, and optional
external pressing
Examples of fixtures for hardening of ring-shaped bodies
However, the expanding mandrel is substantially

more expensive to produce than a standard type
fixed mandrel; if the expanding mandrel is not
produced to conform with precise dimensional
tolerance limits, there will be a substantial adverse effect upon the dimensional precision of
the end product. In addition, the diameter variability of expanding mandrels is very limited,
because if mandrel-segment radii and workpiece
radii were different, they would touch tangentially only. Through the applied force, the workpiece maintains a polygon-shaped inner diameter, determined by the number of segments (as a
general rule, from 5 to 10 segments).
In accordance with the specific geometry of
the workpiece to be treated, fixture components
in addition to the mandrel may be employed,
such as mounting and support rings, down holders, or devices for applying force to the outer
diameter. These components usually are used
only in combination with solid mandrels since
the simultaneous application of force to the inner
and outer diameters at the same time can result
in undefined dimensional variations. Pressure to
the outer surface is frequently applied to workpieces, which, owing to their constituent materials and/or geometries, undergo a significant increase of their internal diameter in the course of
the structural-transformation procedure.
Particularly in the case of thin-walled rings,
often it is not possible to obtain the required levels of both shape and dimension stability in conventional fixture-hardening systems without the
application of pressure to the outer surface of the
workpiece undergoing treatment. Figure 4 shows
schematic representations of fixtures for use in
the hardening of various types of ring-form
workpieces. The final decision regarding the use
of a fixture-hardening system is conditioned primarily by the degree of precision that is required
of a given end-product workpiece.
As noted, both the material of which the workpiece is composed and its specific geometric
configuration are important factors in this context. Other factors also have an essential influence on the accuracy of the hardened workpieces. Thus, for any given workpiece, it is
critical to establish precise values for amount of
compressive force to be applied, cooling time,
and handling time.
Programmable control units, operating in conjunction with personal computers, make possible
the convenient storage of workpiece- and material-specific treatment parameters in a database.
In such an automated system, any significant
variations from production charges can easily be
evaluated and effectively compensated.
In addition, in automated systems of this type,
it is possible to control procedures by comparing
actual values with desired values, which is of
vital importance from the standpoint of quality
control. The treatment of individual workpieces
at fixture-equipped processing stations also
makes it possible to establish workpiece- and
material-specific data records, which would not
be possible in a system in which the simultaneous treatment of an entire batch was carried out.

DOI: 10.1361/hrsd2002p209

Residual Stresses in Nitriding

Z. Kolozsvary, S.C. Plasmaterm S.A., Tg-Mures, Romania
A NUMBER OF technological processes are

based on the diffusion of nitrogen in steel or of
nitrogen combined with other elements, including carbon, oxygen, and sulfur. The basic difference between nitriding and carburizing is that,
in carburizing, the specific properties of the case
against the core are obtained by a controlled carbon-rich martensite in the diffusion layer. No
difference occurs in the structural nature of the
case against the core, and all the property modifications may well be predicted considering the
changes in carbon content. (Carbide-forming
elements modify only the hardenability of the
case and its stability.)
Nitriding has a completely different mechanism. Due to the very low solubility of nitrogen
in iron, the nitrogen diffused in the surface layer
will form nitride precipitates with iron (mainly
c-Fe4N) and with the alloying elements of the
steel, according to their chemical affinity to nitrogen. The mechanism of hardening is different
Fig. 1
Coarse, needle-shape nitride precipitates in the

diffusion layer of low-carbon steel
from that of carburizing, and is practically based

on a precipitation process. The greater the occurrence of uniform, finely dispersed precipitates
of stable nitrides (of aluminum, titanium, chromium, niobium, etc.), the higher the hardness in
the diffusion layer.
Because of the high nitriding potential in the
nitriding atmosphere, however, a compound
layer will almost inevitably form on the steel
substrate. This compound layer consists of c(Fe4N) and e-(Fe23N)-type nitrides on the surface, rarely as a monophase layer. Modern technologies offer close process control, resulting in
a more or less monophase layer, or even bright
nitriding (without a compound layer), but in industrial practice duplex compound layers generally are characteristic. This is especially true
for gaseous treatments, less so for plasma processing. Simultaneous action of carbon and oxygen results in mostly compact, hexagonal etype, crystals with varying amounts of nitrogen
and carbon, or nitrogen, carbon, and oxygen.
A few characteristics are common to all nitriding/nitrocarburizing procedures. First, the
process temperature is between 450 and 570 C
(840 and 1060 F), which means no austenitic
transformation occurs and the structure remains
basically ferritic or, more exactly, in the ferritic
domain. Due to the lack of phase transformation,
no specific residual stresses appear.
Second, this lack of phase transformation
means that the structure of the steel before nitriding strongly affects both the mechanical parameters of the treated parts and the stress-strain
state after nitriding. Normalizing, annealing, and
quenching and tempering will influence the substrate properties, as well as the post-treatment
residual stress level and resulting deformation.
Best results are generally obtained for quenchedand-tempered structures prior to nitriding, presuming the tempering temperature is higher than
that of nitriding or nitrocarburizing. In this case,
practically any influence of carbide precipitation,
spheroidization, or carbide coagulation can be
neglected.
Third, the low treating temperature also modifies the thermodynamics and kinetics of layer
formation. The activity of the steel surface plays
a determining role in the whole process; thus,
machining and finishing operations may influence the stress-strain state of the surface layer.
The compound layer is mainly responsible for

the tribological properties of the nitrided/nitrocarburized engineering components, whereas the
substrate with its diffusion layer provides loadbearing capacity and improved fatigue properties. Structure, composition, and morphology of
the compound layer cover a rather large spectrum as a function of process parameters, and
some discrepancies may be found in the literature concerning the properties of compound layers depending on their composition. The effect
of oxygen in the layer and on layer formation is
still being debated (Ref 15).
To outline the problem of residual stresses in
nitrided/nitrocarburized layers, a short summary
of their structure and properties follows.
Structure of Nitrided Layers

Gaseous Nitriding in NH3-N2-H2 Mixtures.
Basically, nitrogen diffusion in steel results in a
low-nitrogen ferrite according to the Lehrer
phase diagram (Ref 6, 7) and precipitated nitrides of the alloying elements, especially chromium, aluminum, molybdenum, and titanium.
The most desirable structure consists of fine, uniformly dispersed nitride precipitates in the ferritic matrix. However, due to the low solubility
of nitrogen in iron, even with close control of
process parameters, some Fe4N occurs, mostly
in the form of coarse nitride needles. Usually, a
brittle c-(Fe4N) compound layer also appears on
the surface, which is later removed by various
chemical or mechanical processes. These structures are obtained at low temperatures of 510 to
530 C (950 to 990 F) with a subsequent slow
cooling. Thus, residual stresses primarily originate due to the modification of chemical composition and crystal structure by precipitation in
the diffusion layer (Fig. 1).
Nitrocarburizing (Ferritic). The combined
effect of nitrogen, carbon, and oxygen results in
a mainly e-type or e c compound layer of 5
to 25 lm on the steel substrate, supported by a
diffusion layer. Contrary to gaseous nitriding,
the main purpose of nitrocarburizing is to improve the tribological properties, and it can also
be applied to unalloyed or low-alloyed steels. A
certain amount of porosity in the compound
layer is generally present due to the mechanism
210 / Residual Stress During Hardening Processes

of layer growth. The hardness of the diffusion
layer is given by the nitride precipitation phenomena in the matrix, just as for gaseous nitriding. However, because of the generally higher
treating temperature (540 to 570 C, or 1005 to
Microstructure of compound and diffusion layer

in plain carbon steel. Gaseous nitrocarburizing,
570 C (1060 F), 2 h
Fig. 2
Microstructure of diffusion layer in low-alloy

structural steel. Gaseous nitrocarburizing, 570 C
(1060 F), 2 h
Fig. 3
1060 F), the increase in hardness is less prominent even for special alloy steels, as the precipitates are coarser and less uniformly distributed
(Fig. 2, 3).
The interdependence of tribological/mechanical parameters and chemical composition of the
compound layer is not completely elucidated.
The nitrogen/carbon ratio in the e-type layer has
been extensively investigated (Ref 6, 8, 9), but
it remains difficult to provide accurate parameters for industrial processes. The presence of oxygen is also of utmost importance, increasing the
layer growth rate due to increased surface activity (Ref 5, 10, 11).
The problem of residual stresses is more sophisticated in this case, because consideration
must be given not only to stresses in the compound layer or in the substrate, but also to some
particular phenomena that appear in the interface
between the compound layer and the substrate.
Another important factor is the dissolution of
carbide particles in the e compound layer from
the pearlite grains of the substrate (Fig. 4).
Salt bath nitrocarburizing results basically in
the same structure discussed above, but due to
the dissolution phenomenon in the bath an equilibrium in the layer growth process will occur
one that is dependent not only on the bath active
composition but also on its contamination, especially with iron (Ref 12). Usually extremely
fine salt inclusions also can be found in the compound layers outmost film, as it may be observed in studying the fracture toughness of
compound layers (Ref 13). This dissolution process modifies not only the stress state in the
Fig. 4
layer, but first of all that of the interface between

the compound layer and substrate (Fig. 5).
Plasma Nitriding and Plasma Nitrocarburizing. Plasma-assisted processes involve cathode sputtering, which significantly affects both
the interface (at the beginning of the process) and
layer formation. Very accurate and close control
of the process parameters in plasma nitriding and
nitrocarburizing is possible, so different structures and properties can be obtained. When a low
treating temperature (up to 510 C, or 950 F) is
maintained a thin but very hard layer will appear
with extremely fine, dispersed nitride precipitates in the steel substrate. In this case, the compound layer is either very thin c-type or may be
avoided altogether. As in gaseous nitriding, such
a process may only be applied for alloy steels
(Fig. 6).
For improved tribological properties, the compound layer plays the key role. Different data are
available in the literature on the tribological
properties of compound layers of different structures and composition (Ref 10, 11, 14, 15). There
appears to be good agreement among the specialists that plasma-nitrided alloy steels will
show the best properties with a relatively thin,
but uniform monophase c compound layer on a
hard substrate.
On the other hand, some researchers claim that
the ductility and fatigue wear resistance of enitrocarbide compound layers are better than
those of c. Usually the controversy finds its
roots in the fact that it is very difficult to control
the porosity of the e layers, and this porosity has
a negative effect on service properties. The most
Dissolution of carbide particles in the compound layer from the pearlite grains of the substrate. Transmission
electron micrograph (TEM)
Residual Stresses in Nitriding / 211

favorable nitrogen/carbon ratio in the compound
layer is not completely known, and the reproducibility of the case composition is rather problematic, even with up-to-date process control. A
certain amount of carbon may also be found in
plasma-nitrided duplex compound layers (c
e) due to the reverse diffusion of carbon from
the substrate into the compound layer. This retrodiffusion may be very important on the interface,
where the e nitrides absorb important quantities
of carbon, causing even the dissolution of carbide lamellas from the pearlite grains in a ferritic/pearlitic substrate (see Fig. 4). Certainly, as
plasma nitrocarburizing becomes a common
practice, it is possible to obtain a monophase,
compact e compound layer with an optimized
nitrogen/carbon ratio (Fig. 7 ac) (Ref 8, 16).
The residual stresses in the compound layer
and in the substrate are strongly connected to the
nature of compound layers. Whenever a thick,
more or less monophase e-type compound layer
is desired, the higher treating temperatures result
in less significant substrate hardness and in a
lower residual stress level. The influence of a
compound layer on fatigue properties may not
be neglected, however, as it reduces the stress
concentrators in the surface.
Slightly different phases appear when stainless steels are plasma nitrided (Ref 7, 13, 17
21). The problem in this case is to obtain a simultaneous increase in tribological properties,
without significant loss in corrosion behavior.
Passivated or oxide surface films offer only a
partial solution.
stresses as compared to those in carburized layers. Due to the complex mechanism of hardening in nitriding (the presence of a compound
layer), investigations are more difficult and the
mathematical modeling must consider numerous
factors.
Most of the data in the literature are connected
to the residual stresses in the diffusion layer, and
generally relate to gaseous nitriding. Surprisingly few data can be found on residual stresses
in nitrocarburizing, especially for compound
layers. From the experimental data on tribological and fatigue properties it is obvious that not
only the compound and the diffusion layer play
Fig. 5
Compound Layer
Several theories related to the origin of residual stresses in nitrided layers associate these
stresses mostly with the diffusion layers (Ref
2225). The main causes of the residual stresses
are considered to be:
Changes in chemical composition in the dif-
fusion zone
Phase transformation in the diffusion zone,
and precipitation phenomena
Volume changes caused by the phase trans-
formation and the growth of precipitates
Thermal effects due to different coefficient of
expansion during layer formation

The residual stresses connected to the compound layer formation are rarely considered;
however, their effect on the fatigue limit of
treated parts may be significant, especially if the
film thickness of the layer is comparable to the
total wall thickness (engineering components for
precision mechanical applications). These residual stresses originate from:
Volume changes during the formation of dif-
ferent phases in the compound layer
Internal stresses caused by molecular nitro-
gen formation in the porosity
Residual Stresses in Nitrided Layers

Residual stresses and their computer modeling
have been extensively studied for case hardening. However, a rather limited number of papers
deal with the problem of residual stresses in nitriding and related technologies. No doubt one
of the reasons for this is the lesser effect on distortions and generally lower level of these
an important role in properties of treated parts,

but that specific phenomena on the interface between the compound layer and substrate are also
critical.
Fig. 6
Microstructure of plasma-nitrided low-alloy steel

(18MnCr10). 560 C (1040 F), 8 h
Modified interface between the compound layer and substrate as result of carbide dissolution. TEM
Dependent on process parameters, the porosity

is generally associated with the high nitrogen potential for the e-type compound layer. This porosity usually is concentrated at the outer zone of
the compound layer, sometimes causing tensile
residual stresses to appear there (Ref 14, 26, 27).
Investigations have shown that compressive residual stresses exist in the compound layer during layer growth, and these stresses increase during cooling as a result of the volume misfit
generated by different thermal expansion coefficients of the compound layer and the diffusion
zone (Ref 28). These observations apply particularly to c structures but are also true for compact
e structures.
The internal stresses in the compound layer
are of the second and third type (Ref 3, 29). Considering the definition of residual stresses given
by Macherauch et al. (Ref 29), residual stresses
of the third type are particularly important on the
interface, where they occur due to high dislocation density at grain-boundary inhomogeneities
and to formation of the new crystal structure (e
or c).
Very few measurements of residual stresses in
compound layers are known, partly because of
technical difficultieseven by x-ray diffraction
(XRD) methodsand partly because the diffusion layer seemed to play a more important role
in the residual stress state of the surface. In situ
measurements of residual stresses have been performed for a c-Fe4N-type compound layer built
on a steel surface during gas nitriding (Ref 28).

Most of the data related to residual stresses in
compound layers result from experiments in gaseous nitriding or plasma nitriding (Ref 27, 30
32). The combined effect of nitrogen and carbon,
and of nitrogen, carbon, and oxygen, has been
extensively investigated (Ref 9, 10, 3335), but
these studies have focused primarily on their effect on wear and fatigue behavior and less on the
residual stress state. When considering the residual stresses in compound layers, the layer structure cannot be neglected. Structures of c or etype (or, more exactly, pure c or mixtures of
c-e with different compositions) were investigated, with the aim of finding the most favorable
effect on fatigue behavior. Combined wear and
fatigue resistance is of critical importance for the
mechanical industry (Ref 7, 15, 17, 3638). In
terms of fatigue behavior, a particular aspect
may be connected to the penetration of the compound layer into the substrate: Electron-microscopic examination of nitrocarburized surfaces
has shown that a significant removal of surface
defects resulted from the finishing process, thus
decreasing the stress concentrators that can initiate the first microcracks in long-cycle fatigue
phenomena (Ref 34).
Some discrepancies among the experimental
results, including their interpretation and concluTable 1
sions, are also found in the literature. Frequently,

the role of residual stresses in compound layers
is considered of lesser importance, while other
studies emphasize that the compressive stresses
in the surface will practically disappear with removal of the compound layer (Ref 31, 39).
As discussed in the article Material Factors
in this Handbook, significant differences appear
in the structure and morphology of nitrided (nitrocarburized) layers, a function of the technological process. These differences are reflected
in the stress-strain state of the layers. However,
it is very difficult to obtain comprehensive and
reliable data on residual stresses in compound
layers, and it seems even more problematic to
measure the residual stress state at the interface
between the compound layer and the substrate.
Most experiments have been performed using
the XRD method for measuring the residual
stresses, but this presumes chemical or mechanical removal of subsequent films of the compound layer, certainly modifying its stress state
as well. In situ measurements of residual stresses
in compound layers have been performed on gasnitrided samples (Ref 28).
Data in the literature show a wide range of
values for residual stresses, but it is generally
agreed that tensile stresses appear in e compound
layers. According to one study, tensile stresses
Residual stress distribution in the e and c compound layers

Residual stress(a), MPa (ksi)
e-Fe23N
Distance from surface, mm (in.)
0 (0)
0.005 (0.0002)
0.01 (0.0004)
0.015 (0.0006)
0.02 (0.0008)
0.03 (0.001)
54 (8)
492 (71)
594 (86)
590 (86)
c-Fe4N
48 (7)
264 (38)
432 (63)
587 (85)
327 (47)
624 (90)
721 (104)
705 (102)
393 (57)
528 (77)
764 (111)
724 (105)
640 (93)
(a) The residual stress values given for the same layer structure were measured for layer thicknesses of 22 and 25 lm on 31CrMoV9 steel treated at
570 C (1060 F) for 16 and 32 h, respectively. Source: Ref 27
Fig. 7
up to 200 MPa (30 ksi) were found on Armco

iron, whereas another found approximately 150
MPa (20 ksi) tensile or compressive stresses on
C45 steel. Rozendaal et al. (Ref 40) measured
tensile stresses up to 800 MPa (115 ksi) in the
outer zone of e layer, which decreased toward
the interface and changed to compressive stress.
In c compound layers, the same authors measured compressive stresses between 60 and
600 MPa (10 and 90 ksi). Oettel and Ehrentraut (Ref 26) studied the residual stresses in
compound layers of different nitrided alloy
steels. According to their data, compressive
stresses between 350 and 1500 MPa (50
and 220 ksi) are typical for c compound layers, whereas tensile stresses of 100 to 370 MPa
(15 to 55 ksi) characterize the basically e structures. However, compact e (pore-free) layers
may show compressive stress values up to 580
to 800 MPa (85 to 115 ksi) at the interface with the diffusion layer. In general, thinner
layers presented higher stress values, which may
be explained primarily by their more compact
nature and also by a lesser effect of the sophisticated influence of the relaxation process during
the long treating cycle.
Investigations performed by Spies (Ref 27)
with Cr-Mo, Cr-Mn, and Cr-Mo-V steels have
shown stress values of 500 to 600 MPa
(70 to 90 ksi) in e-Fe(23)N compound layers of 20 to 25 lm and 620 to 720 MPa (90 to
105 ksi) for c. Increasing the treating time from
16 to 32 h, a compound layer thickness of 20 to
30 lm caused no significant decrease in residual
stress level. The most significant data are shown
in Table 1.
All the available data indicate higher compressive stresses in compound layers of pure c
than in e or c e layers. According to Spies
(Ref 27), maximum residual stresses may attain
700 to 750 MPa (100 to 110 ksi) in c-
Plasma nitrocarburized EN-9 (070M55) steel. (a) Microstructure. (b) GDOS carbon and nitrogen profile. (c) X-ray diffraction pattern at the surface (showing almost pure ephase). 570 C (1060 F), 3 h, 4 mbar. Source: Ref 16

Fe4N compound layers, whereas values of 550
to 600 MPa (80 to 90 ksi) were measured in eFe23N layers. In situ measurements in gaseous
nitriding performed by Hoffmann et al. (Ref 28)
resulted in a slightly lower compression stress
level of about 450 MPa (65 ksi). The carbon
content of the substrate, especially the carbon
level in the compound layer, has a considerable
effect on residual stresses; increased carbon content results in an increased residual stress level
(Ref 8, 28) (Fig. 8). However, the importance of
carbon in the compound layer is more emphasized in e-Fe23(C,N) structures, due to the increased solubility of carbon in the compact
hexagonal e-Fe23N crystals (Ref 8). The apparently lower residual stress values in e-type
compound layers probably also can be explained
by the ductility of this structure.
These observations pertain to gas nitriding,
where the compound layer consists almost exclusively of Fe4N nitrides. Very few data have
been published on residual stresses in nitrocarburizing, where the compound layer consists
mainly of e-carbonitrides with some oxygen
(Ref 6, 8, 10).
Investigations have shown that compressive
stresses appear in the compound layer due to volume changes during e-crystal formation as a result of dissolved carbon from the matrix. This
phenomenon is clearly shown in Fig. 9 (Ref 9).
Due to the solubility of carbon in e-nitrides (carbonitrides), the carbide particles of the pearlite
grains in the interface are dissolved in e-crystals;
thus, the compound layer is rooting in the substrate at the boundaries of pearlite grains. This
phenomenon causes compressive residual
stresses not only in the compound layer, but also
in the substrate in the interface region. It may
also be partly responsible for increased fatigue
resistance of nitrocarburized samples and for the
significant increase in surface fatigue resistance
when dynamic loading is combined with sliding
wear. According to the literature, the incorporation of carbon and/or oxygen in the layer does
not significantly affect the residual stresses in the
usual composition ranges (Ref 10, 33). This may
well be explained by the almost similar interstitial influence on lattice parameters of these elements.
Mathematical modeling of residual stresses in
Residual stresses in c-nitride during cooling

from the nitriding temperature to room temperature. Source: Ref 28
compound layers has been little studied. Computer modeling of compound layer formation,
which is of crucial importance for the mathematical formulation of stress patterns, was initiated in a few investigations (Ref 32, 41), but
the results fail to answer a number of basic questions. Both the kinetics of surface reactions in
nitriding and the thermodynamics of Fe-N-C
have not yet been satisfactorily elucidated. The
very good model proposed by Sun and Bell (Ref
42) for the diffusion of nitrogen and the formation of nitride precipitates in the matrix also
gives only a few indications for the compound
layers.
Diffusion Layer
As was discussed earlier, a number of studies
present different aspects of the residual stresses
in nitrided layers (Ref 14, 2224, 26, 28, 31, 32,
37, 4349). The mechanism of nitriding causes
practically no differences in the phenomena in
the diffusion layer function of the applied technology. Due to its low solubility in the ferrite
matrix, nitrogen will form Fe4N-type nitride needles or precipitates in the diffusion layer substrate. The presence of alloying elements will
modify the composition of the nitrides, resulting
in nitrides and carbonitrides of the alloying elements (Ref 44, 45). The treating temperature
strongly influences both the structure and composition of the nitrides, but basically the kinetics
of diffusion layer formation will remain the same
for different industrial processes. Thus, the main
causes of the residual stresses are connected to
volume changes in the layer due to:
Formation of iron nitrides in the compound
layer
Dissolved nitrogen in the ferrite matrix of the
layer
Precipitation of different nitrides in the dif-
fusion layer
Decarburization in the diffusion layer
Differences in thermal contraction between
the layer and the substrate during cooling

Thermal volume misfits between the matrix
and precipitates, generating thermal residual

stresses during cooling from the process temperature. (These volume misfits originate
from differences in the linear thermal expansion coefficients of the precipitates and matrix.)
Thus, it seems logical that the residual stress
level increases with increased nitrogen content,
respectively with increased number and volume
of the nitride precipitates. This was clearly
shown in the experimental work of several investigators (Ref 26, 28, 32, 40, 44, 45, 50). It is
also important to emphasize that the residual
stresses may attain a maximum value limited by
the thermal-induced creep phenomena. Higher
values should cause plastic deformations and
thus a reduction of the stresses (Ref 45). Maximum stress levels are shown in Table 2. The
maximum compressive stresses are determined
primarily by the nitrogen concentration and less
by the structural phase itself. Due to the maximum of 0.1% nitrogen solubility in ferrite, the
stress level cannot be higher than 250 MPa
(35 ksi). In contrast to the development of re-
Fig. 8
Fig. 9
Dissolution of carbides from pearlite grains of the subsurface region in the e-type compound layer. TEM

sidual stresses in carburizing, there is no significant influence of other factors associated with
cooling (i.e., the changing solubility of nitrogen
in iron, or other thermal stresses associated with
quenching).
Both the material composition and the process
parameters covering the treating temperature and
the nitriding potential influence the residual
stress state of the nitrided steel. As might be expected, the steel composition influences not only
the precipitates, but also the behavior of nitrogen
in the diffusion layer. The effect of carbon content must also be considered. Specimens containing a large amount of strong nitride-forming
elementssuch as chromium, aluminum, and ti-
Table 2 Influence of nitride-forming elements on residual stress

Residual stress
(a)
Element
Ferrite, 0.1N
Fe4N
AlN
CrN
Cr2N
Dissolving Fe3C
Residual stress
(b)
MPa
ksi
MPa
ksi
820
440
210
120
65
30
2700
2600
1600
1700
1600
1250(c)
390
375
230
245
230
180(c)
taniumform a high-hardness case and high residual stresses in the layer. A direct correlation
seems to exist between the microhardness of the
diffusion layer and the residual stress level (Ref
49). Also, the higher the carbon content of the
steel, the higher the compressive residual stress
level (Ref 28). This is illustrated in Fig. 10 (Ref
45).
The process temperature has an important effect on the residual stresses, by its influence both
on the diffusion rate and on differences in precipitate formation during cooling. In situ measurements of residual stresses during the nitriding process revealed important data on not only
stress relaxation by plastic deformation during
nitriding, but also the influence of process parameters on induction of residual stresses in both
the compound and the diffusion layers (Ref 28).
The cooling process can be considered the most
important step in generating residual stresses in
the compound layer, whereas stresses are created
in the diffusion zone mostly during nitriding and
only a very slight change occurs in cooling. Figure 11 shows the general influence of treating
(a) % Me mass. (b) % N mass. (c) % C mass. Source: Ref 45
Fig. 12
Influence of treating time on residual stresses.

Source: Ref 45
Increase of residual stresses as a function of the

concentration of nitride-forming elements in
steel. Source: Ref 45
K-oxygen
K-no oxygen
5
Reaction rate, k 107
0
0
Fig. 11
Decrease of residual stress level with increasing nitriding temperature. Source: Ref 45

Fatigue Behavior of Nitrided Steel
Components
Residual Stresses and Bending Fatigue
Resistance
Fig. 10
temperature on stress pattern (Ref 45).

Nitriding time also influences the residual
stress profile. A long nitriding cycle produces a
deeper compressive stress distribution, but with
lower maximum values in the near-surface region (Ref 37). The simultaneous stress generation and relaxation process can explain this, with
the dominance of the relaxation in long treating
time. The influence of treating time on the residual stress pattern can be clearly seen in Fig. 12
(Ref 45).
In Fig. 11 and 12, notice the specific decrease
in residual stresses in the near-surface region,
which is caused first of all by the redistribution
of carbon in the diffusion layer. In chromiumcontaining steels, an increase in carbon content
in front of the diffusion layer is observed, together with decarburization in the surface and
precipitation of cementite on the grain boundaries (Ref 24).
In gaseous processes, oxygen in the treating
atmosphere also has a remarkable effect on the
kinetics of layer formation. Oxygen causes a surface activation, which has a positive effect on
both nitrogen diffusion and uniformity in the
surface layer. It has been proven that in the presence of a controlled oxidation, the surface layer
contains no soft spots, which are frequently characteristic of passive surfaces resulting from machining operations (especially deep drawing,
lapping, honing, etc.) (Ref 10, 35). This effect is
shown in Fig. 13. Though no specific data are
known on the influence of oxygen on residual
stresses, it may well be assumed that enhanced
nitrogen diffusion increases the residual stress
level.
10
20
30
40
50
Undissociated ammonia, R, %
60
Effect of oxygen addition on the reaction rate

(k) in gaseous nitrocarburizing, as a result of
surface activation
Fig. 13
Nitrided surface layers (especially those that

have been nitrocarburized) have a clearly positive influence on both the bending fatigue and
surface fatigue behavior of steel components.
However, this positive effect is the result of a
rather complex mechanism. The surface of the
components usually presents a complex structure
after machining, with the characteristic Beilby
layer. The inner zone of this layer, consisting
mostly of deformed and partially damaged
grains and crystals, is the source of microcrack
initiation. These microcracks are extremely deleterious when friction combined with local
hertzian loading appears (gears, bearings, etc.).
The recrystallization associated with the dissolution phenomena in the surface film eliminate
these stress concentration points, thus reducing
the crack initiation hazard (Ref 38).
On the other hand, the considerable compressive stresses in the near-surface region of the
compound layer further reduce the crack initiation in this zone. Certainly, this effect is impor-

tant first of all in long-cycle fatigue behavior.
The compressive residual stresses in the diffusion layer are of primary importance in reducing
the subsurface crack initiation, as well as the
crack propagation rate. However, remember that

the compound layer also has a combined effect
with the diffusion layer. This can be illustrated
by comparing the fracture toughness of different
nitrided and nitrocarburized layers, using the reverse impending method (Ref 51). Figure 14
shows clear evidence of these differences, but
unfortunately no residual stress measurements
were performed on the test samples.
The influence of nitriding and carbonitriding
on bending fatigue properties was investigated
for different steels and different process parameters (Ref 36, 5254). The fatigue resistance of
both unnotched and notched samples depends on
the value and distribution of the residual stresses
in the diffusion zone. This is practically correlated with the hardness and its distribution in the
layer. For the same hardness values, a deeper
nitride case will increase the residual stresses in
the diffusion zone, thus causing an increase in
long-cycle fatigue strength (Ref 54). According
to Spies et al. (Ref 52, 54), an e-type compound
layer results in higher residual stresses in the diffusion layer and in a higher fatigue limit (Tables
3 and 4).
There is no doubt that all the data from the
technical literature outline the efforts for a more
fundamental, mathematical approach to the
problem of residual stresses in nitrided steels and
their influence on fatigue behavior. Though nitridingwith all its industrial versionsis a
well-established and widely applied technology,
the mathematical description of layer growth and
the formation of local strength and macrostresses
is far from that of carburizing. The explanation
may be connected to the more sophisticated kinetics and thermodynamics of the process, but
also to the practical fact that the prediction of
residual stresses in nitrided components is less
critical than for carburized components.
Residual Stresses in Contact Fatigue

Behavior
Fig. 14
TEMs showing typical crack pattern of surface layer. (a) Gaseous nitrocarburizing. (b) Salt-bath nitrocarburizing. (c) Plasma nitriding. (d) Plasma nitrocarburizing
Table 3
Influence of steel grade on residual stresses and fatigue life of structural steels
Compound layer
Steel
Type
Thickness, lm
Case hardness, HV0.3
rE(a), MPa (ksi)
rAD(b), MPa (ksi)
Wohler line exponent(c), k
45MnV3BY
c
e(c)
c
e(c)
c
e(c)
e(c)
810
2022
810
2224
10
2225
25
535550
135 (20)
185 (27)
300 (45)
350 (50)
255 (35)
315 (45)
260 (40)
499 (72)
528 (77)
574 (83)
593 (86)
608 (88)
670 (95)
615 (90)
15.8
19.9
19.3
29
11.2
15.8
15.3
20MnCr5N
20MnCr5QT
660680
640650
640650
600
(a) rE, residual stress; depth below surface, 50 lm. (b) rAD, fatigue, when a nominal mean stress or residual stress is applied (long life alternating
strength); bending fatigue test, P 50%. (c) The Wohler line exponent was calculated for long-cycle fatigue. Source: Ref 52
There is a considerable difference between the

bending or torsion fatigue resistance and the surface fatigue behavior of treated steels. Failure of
components exposed to contact fatigue occurs in
different forms, such as pitting, exfoliation, or
surface crack formation (which later can cause
pitting or exfoliation). Surface fatigue also has a
specific form when mechanical loading combines with a corrosive environment, as in many
chemical industry applications. This fretting corrosion also must be considered on nitrided samples.
Surface fatigue as the result of combined rolling and sliding load usually starts as microcracks
at small angles to the surface, in the immediate
subsurface region due to shear stresses. The
greater the sliding loading effect, the closer to
the surface are these microcracks. Contact fatigue behavior of nitrided/nitrocarburized components depends on both the compound and the
diffusion layers. The compound layer has a positive influence on sliding wear properties of the
components, whereas the diffusion layer is the
support for hertzian, normal cyclic loads. This
composed structure leads to a significant difference in the resistance of nitrided and, especially,
nitrocarburized steels compared to carburized

steels, because not only is the layer thickness
smaller, but the hardness gradient in the layer
presents different shapes for the two different
surface layers. Due to these differences, nitrocarburized surfaces show better resistance to
sliding wear, but lower normal loading capacity
may be expected. Compressive residual stresses
in the diffusion layer also improve contact fatigue resistance (Ref 37, 38, 41). According to
Li et al. (Ref 37), the short-time plasma nitriding
has a significant effect on the fretting fatigue resistance of En19 steel, increasing it by more than
250%. Other sources state that there are no practical differences between plasma-nitrided and
gaseous-nitrided samples exposed to contact fatigue, presuming the same type of layer structure
was obtained (Ref 31, 39).
Modeling and Prediction of

Residual Stresses
Modeling of the nitriding and nitrocarburizing
process is much less developed than modeling
of carburizing. Much experimental data exist
concerning the structure and properties of ni-
trided/nitrocarburized layers, but only a few attempts have been made to develop a mathematical description of the layer-growth mechanism
and the formation and distribution of residual
stresses. Probable explanations for this situation
may be connected to the complexity of the parameters to affect the process itself, and to the
less critical effect on deformations and service
behavior. No practical attempts have been made
to mathematically model the compound layer
formation and residual stress distribution. In
contrast, a few remarkable studies may be found
on modeling the formation and stress-state of the
diffusion layer. A model elaborated by Buchhagen and Bell (Ref 32) for the simulation of
residual stress development in the diffusion layer
of low-alloy plasma-nitrided steels considers the
thermally controlled creep process, with the residual macrostress being the driving force for
stress relaxation.
According to this model, the thermally induced stress was the temperature-dependent volume part of the constituents and their volumes
per metal atom. These volumes were considered
separately for heating to the process temperature
and for cooling. Due to the missing quantitative
data on stress relaxation and on the redistribution

of carbon in the diffusion layer, some assumptions have to be made using measured residual
stress profiles and carbon profiles from analysis
by glow discharge spectroscopy. The plasma-nitriding simulation program (PLASMA) developed by Sun and Bell (Ref 42) calculates the
profiles of nitrogen concentration, volume fraction of the nitrides, and hardness as a function
of distance from the surface. The modified program activates the stress simulation unit after
every 10 min of the simulated nitriding time. The
time-constant for the relaxation must be provided experimentally.
Extensive work on computer simulation of the
formation of diffusion layers in nitriding chromium-alloyed steels was carried out by Schreiber et al. (Ref 23, 43, 50). The model developed
assumes that the nitrogen in the diffusion layer
is dissolved partly in the ferritic matrix and
partly in nitride precipitates. The nitrogen diffusion generally takes place with a constant surface nitrogen concentration; at the beginning of
the process the compound layer is formed and
the diffusion of the dissolved nitrogen will be
determined by the equilibrium concentration on
the phase boundary between the ferrite and the
iron nitrides (the compound layer and the substrate) (Ref 23). The total solubility thus is much
higher than in pure iron, because the strain
caused by the nitride precipitates leads to elastic
distortions of the matrix around the precipitated
particles. This excess nitrogen is determined by
the stress state caused by the misfitting nitrides
and the maximum solubility of the strain-free lattice, C a0,N, and can be described by:
r V
R T
C aN C a0,N exp
r
4G
eM
4
1 GK
3
C a0,N A exp
Fig. 15
Calculated nitrogen concentrations for iron and FeCr alloys. 550 C (1020 F) 10 h. Source: Ref 50
Table 4
(Eq 1)
(Eq 2)
R T
(Eq 3)
where A is a constant, Q is activation energy, R

is a gas constant, T is temperature, V is the molar
volume of nitrogen in iron, G is shear modulus,
K is the compressibility of nitride precipitates, e
is volume misfit, and M is the molar fraction of
the precipitates.
Structure of nitrided layer and fatigue limit (K 1)

Hardness,
HV0.3
rE (c)
Ncd (b)
Stress-intensity
factor
rAD (d)
Steel (a)
Case
Core
mm
in.
MPa
ksi
MPa
ksi
Wohler line exponent (e), k
Ratio of increase (f)
MPa
ksi
20MnCr5N
20MnCr5QT
630
600
600
590
470
190
260
250
250
250
0.42
0.24
0.44
0.56
0.60
0.017
0.009
0.017
0.022
0.023
365
360
430
235
140
53
52
60
35
20
490
640
670
745
760
70
90
100
108
110
14.1
13.0
9.3
9.2
18.5
1.55
1.22
1.28
1.49
1.60
15.7
16.7
18.9
21.9
23.7
2.3
2.4
2.7
3.2
3.4
C45QT
(a) Neds nitrided case depth. (b) Pretreatment: N, normalized; QT, quenched and tempered. (c) rE, residual stress; depth below surface, 50 lm. (d) rAD, fatigue, when a nominal mean stress or residual stress is applied (long
life alternating strength); rotating bending test: P 50%; diameter, 7 mm. (e) For high-cycle fatigue: P 50%; (f) rAD nitrided/rAD unnitrided for P 50%. Source: Ref 52

The fraction of the precipitated nitrides, W(t),
can be described by the following Avrami equation:
W(t) (1 e(t/s) )
n
(Eq 4)
Assuming homogeneous nucleation and spherical particles, the precipitation rate constant is
given by:
s
cP c0
r2e
cm c0 2 DMe
(Eq 5)
where DMe is the diffusion coefficient of the alloying element; re is the equilibrium radius of
precipitates; and cm, cP, c0 are the alloying element concentration in the matrix, in the precipitate, and at the phase boundary of the precipitate
respectively.
Assuming that the z-axis is perpendicular to
the surface, the diffusion and precipitation process can be described using the following onedimensional inhomogeneous diffusion equation:
dW
k dt
dC
d2C
DN
dt
dz2
(Eq 6)
where DN is the diffusion coefficient of nitrogen

and ki represents the nitrogen fraction in the precipitate of the alloying element i. The diffusion
equation can be solved numerically. The boundary condition is given by the sum of nitrogen
a
concentration (C aN) in strained and (C0,N
) in
strain-free lattice of ferrite. The calculated nitrogen concentrations for pure iron and FeCr alloys
are shown in Fig. 15 (Ref 50). With the calculated volume fraction of precipitation as a function of the distance from the surface, the strain
distribution due to volume change is given.
The total strain increment consists of three
components:
vol
detx deel
deplx
x dex
(Eq 7)
The elastic strain component is given by

Hookes law:
rel
x (z)
E
eel
x (z)
1 m
(Eq 8)
The strain due to volume change is given by the

volume misfit between nitride and matrix (DV/
V) and the volume fraction of precipitated alloying element CME:
evol
i (CME)i 3 V i
1 DV
(Eq 9)
For plastic strains, thermally induced creep is

taken into account and described with a Norton
equation:
depl
x A
Fig. 16
Measured residual stresses in an FeAlC alloy with 0.6 wt% Al and 0.15 wt% C. 590 C (1095 F) Source:
Ref 50
G dt
r
where A and n are constants.

Using the mechanical equilibrium condition
for an endless plate with thickness d,
r (z)dz 0
el
x
Fig. 17
Calculated residual stresses for an FeAlC alloy with 0.6 wt% Al and 0.15 wt% C, 590 C (1095 F). Source:
Ref 50
(Eq 10)
(Eq 11)
the macrostress can be calculated. The measured

and simulated stress profile data show a very
good agreement.
Figure 16 shows the macrostress distribution
in the diffusion layer of an FeAlC alloy nitrided
at 590 C (1095 F) for 4 and 36 h, respectively
(Ref 50). The stress was measured by x-ray diffraction combined with chemical removal of the
layer. The sample nitrided for 4 h presents a
compressive stress maximum of 380 MPa (55
ksi) near the surface. After 36 h of nitriding, the
compressive stress maximum decreases to 200
MPa (30 ksi) and shifts to a distance of about
550 lm from the surface. The maximum of compressive stresses near the surface disappears
completely during prolonged nitriding.
Figure 17 presents the simulation of macrostress distribution (Ref 50). The position, the
value of the maximum, and the extension of the
stress almost agree with the measured values.
The calculated decrease of stress near the surface
is lower than the measured one. This effect is

connected to the stress relaxation due to surface
decarburization.
REFERENCES
1. C. Kim, Fracture of Case-Hardened Steels
and Residual Stress Effects, Case Hardening Steels: Microstructural and Residual
Stress Effects, D.E. Diesburg, Ed., TMSAIME, 1986
2. E. Macherauch and O. Vohringer, Residual
Stresses after Quenching, Theory and Technology of Quenching, B. Liscic, H.M. Tensi,
and W. Luty, Ed., Springer Verlag, 1992, p
117181
3. H.J. Eckstein, Technology of Heat Treatment of Steel, VEB Deutcher Verlag fur
Grundstoffindustrie, Leipzig, 1977, p 74
93 (in German)
4. H. Mallener, Dimensional and Shape
Changes in Case Hardening, HTM 45, Vol
1, Rudolph Haufe Verlag, 1990, p 6672
5. C. Ruset, V. Stoica, S. Janosi, Z. Kolozsvary, A. Bloyce, and T. Bell, The Influence
of Oxygen Contamination on Plasma Nitrided and Nitrocarburised Layers, Proceedings of the 11th Congress of the IFHTSE,
4th ASM Heat Treatment and Surface Engineering Conference in Europe (Florence,
Italy), Vol 1, Associazione Italiana di Metallurgia, 1998, p 271279
6. W. Vogel, Influence of Atmosphere-Composition in Nitrocarburizing on the Structure
of the Compound Layer, Daimler-Benz
AG, Stuttgart, unpublished report
7. Y. Sun, T. Bell, Z. Kolozsvary, and J. Flis,
Response of Austenitic Stainless Steel to
Low Temperature Plasma Nitriding, Heat
Treat. Met., Vol 1, 1999, p 916
8. S. Pietzsch, S. Bohmer, and H.-J. Spies, Investigations of the Adjustment of Nitrogen
and Carbon in the e- Phase, Proceedings of
the Second International Conference of
Carburizing and Nitriding with Atmospheres, (Cleveland, OH), J. Grosch, J.
Morral, and M. Schneider, Ed., ASM International, 1995, p 295300
9. Z. Kolozsvary, Influence of Surface Conditions on the Formation and Properties of Nitrocarburized Layers, Neue Hutte, Vol 31
(No.1), 1986, p 121128 (in German)
10. H.-J. Spies, P. Schaaf, and F. Vogt, Influence of Oxygen Addition during Gas Nitriding on the Structure of the Nitrided Layers,
Mat. Wiss. Werkstofftech., Vol 29, 1998, p
588594
11. A. Molinari et al., Wear Behavior of Diffusion and Compound Layers in Nitrided
Steels, Surf. Eng., Vol 14 (No. 6), 1998, p
489495
12. Z. Kolozsvary, Observations Concerning
the Salt Bath Nitriding and Salt Bath Sulphuriting of Machine Parts, HTM, O.
Schaaber, W. Stuhlmann, Ed., Rudolf Haufe
Verlag, 1969, p 286291
13. Z. Kolozsvary, A New Method for Testing
Fracture Toughness of Nitrided Compound
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
Layers, Proceedings of the 2nd International Congress on Heat Treatment of Materials of IFHT, 1st National Conference on
Metallurgical Coatings of AIV (Florence, Italy), Associazione Italiana di Metallurgia,
1982, p 973982
H.-J. Spies and H. Cramer, Influence of the
Structure of Nitridred Layers on Their
Toughness, Mat. Wiss. Werkstofftech., Vol
24, 1993, p 320326 (in German)
Sz. Farkas, E. Filep, and Z. Kolozsvary,
Plasma Nitriding Improves Service Behaviour of Textile Machine Components, J.
Heat Treat., Vol 1 (No. 4), 1980, p 296
311
C. Ruset, A. Bloyce, and T. Bell, Recent
Investigation into the Mechanisms of the
Plasma Nitrocarburizing Process, presented at National Conference on Heat
Treatment, 2325 April 1997, Tg. Mures,
Romania
S. Janosi, Z. Kolozsvary, C. Ruset, T. Bell,
and Y. Sun, Consideration on Plasma Nitriding of the 30 C 130 Martensitic Stainless
Steel, presented at National Conference on
Heat Treatment, 2325 April 1997,
Tg.Mures, Romania
S. Janosi, Z. Kolozsvary, V. Sandor, and C.
Ruset, Consideration on Plasma Nitriding
Austenitic and Martensitic Stainless Steel,
presented at Seminar on Surface Engineering, Polish Academy of Sciences, Warsaw,
1997, unpublished
M.F. Danke and F.G. Worzala, The Structure and Properties of Ion Nitrided 410
Stainless Steel, Mater. Sci. Forum, Vol
102104, 1992, p 259270
W.L. Wang, Y.T. Chou, and S. Lee, Chemical Stresses Induced by Grain-Boundary
Diffusion, Metall. Mater. Trans., Vol 29A,
1998, p 21212125
M. Samandi, B.A. Shedden, T. Bell, G.A.
Collins, R. Hutchins, and J. Tendys, Significance of Nitrogen Mass Transfer Mechanism on the Nitriding Behavior of Austenitic Stainless Steel, J. Vac. Sci. Technol. B,
Vol 12 (No. 2), 1994, p 935939
W. Darbinjan, H. Oettel, and G. Schreiber,
Comparison of Mechanical and X-ray
Methods for Measurement of Residual
Stresses on Nitrided Steels, Residual
Stresses: European Conference on Residual
Stresses, Frankfurt, 1992, p 565574
G. Schreiber and H. Oettel, Modeling of
Precipitation and Relaxation Process: Diffusion Layers, Nitrieren und Nitrocarburieren, TU Bergakademie Freiberg, p 167
175 (in German)
H.-J. Spies, H. Berns, A. Ludwig, K. Bambauer, and U. Brusky, Hot Hardness and
Residual Stresses in Nitrided Steels, TU
Bergakademie Freiberg, unpublished (in
German)
M. Oden, C. Persson, and T. Ericsson, Measurement and Calculation of Internal
Stresses in CMC and MMC, Residual
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
Stresses, Informationgesellschaft Verlag,

1993, p 10011010
H. Oettel and B. Ehrentraut, Macroscopic
Residual Stresses in Compound Layers of
Nitrided Steels, HTM 40, Rudolph Haufe
Verlag, Munich, Germany, 1985, p 183
187 (in German)
H.-J. Spies, Residual Stresses in Compound Layers (Measurement with Cr-KRadiation), University of Freiburg, Germany, unpublished report (in German)
F.T. Hoffman, U. Kreft, T. Hirsch, and P.
Mayr, In situ Measurement of Residual
Stresses during the Nitriding Process, Proceedings of the Second International Conference of Carburizing and Nitriding with
Atmospheres (Cleveland, OH), ASM International, 1995, p 289293
E. Macherauch, H. Wohlfahrt, and U. Wolfstieg, Practical Definition of Residual
Stresses, HTM 28, Vol 3, Rudolf Haufe
Verlag, Freiburg, Germany, 1973, p 201
211 (in German)
E.J. Mittemeijer, The Relation between Residual Macro- and Micro-stresses and Mechanical Properties of Case-Hardened
Steels, Case Hardening Steels: Microstructural and Residual Stress Effects, TMSAIME, 1986
J.M. Cowling and J.W. Martin, Effect of Internal Residual Stresses on the Fatigue Behavior of Nitrided En 41B Steel, Heat Treatment 79, The Metals Society, London,
1979, p 178181
P. Buchhagen and T. Bell, Simulation of the
Residual Stress Development in the Diffusion Layer of Low Alloy Plasma Nitrided
Steels, Computational Mater. Sci., Vol 7,
1996, p 228234
H.-J. Spies and F. Vogt, Gaseous Nitriding
of High Alloy Steels, HTM 52, Vol 6, Rudolf
Haufe Verlag, Freiburg, Germany, 1997, p
342349 (in German)
Z. Kolozsvary, Observation Concerning the
Characteristic of Soft Nitriding in Controlled Atmosphere, Casting and Forging,
No. 293, 1972, p 4346
Z. Kolozsvary, A. Barabassy, and V. Sandor, Effect of Surface Activation on the Formation of Carbonitride Compound Layers
and Their Properties, Proceedings of XVIIth
International Conference of Heat Treatment
(Barcelona), Asociacion Tecnica de Tratamentos Termicos, 1978, p 112
H.-J. Spies, M. Scharf, and N.D. Tan, Influence of Nitriding on the Durability of Steels,
HTM 46, Vol 5, Rudolf Haufe Verlag, Freiburg, Germany, 1991, p 288293 (in German)
C.-X. Li, Y. Sun, and T. Bell, Consideration
of Fretting Fatigue Properties of Plasma Nitrided En19 Steel, Surf. Eng., Vol 15 (No.
2), 1999, p 149153
Z. Kolozsvary, Contribution to the Study of
Surface Fatigue in Sliding Wear, Wear, No.
25, 1973, p 215224
D.H. Jack and S. Winnik, Effect of Process
40.
41.
42.
43.
44.
Variables on the Fatigue Resistance of Nitrided Steel, Heat Treatment 79, The Metals Society, London, 1979, p 169177
H.C.F. Rozendaal, P.F. Colijn, and E.J. Mittemeijer, Morphology, Composition and
Residual Stresses of Compound Layers of
Nitrocarburized Iron and Steels, Proceedings Heat Treatment 1984, The Metals Society, London, 1984, p 31.131.6
G. Schreiber and H. Oettel, Modeling of the
Precipitation and Relaxation Processes: Diffusion Layers in Nitriding and Nitrocarburizing, Nitrieren und Nitrocarburieren, Weimar, 1996, p 167175 (in German)
Y. Sun and T. Bell, Proceedings of 9th International Congress on Heat Treatment
and Surface Engineering, 2628 Sept 1994
(Nice, France), F. Moreaux, Ed., PYC Edition, Ivry-Sur-Seine Cedex, France, 1994, p
385394
G. Schreiber, H. Oettel, W. Darbinjan, and
R. Wiedemann, Residual Stresses in the Precipitation Zone of Nitride Layers, Residual
Stresses, European Conference on Residual
Stresses, Frankfurt, 1992, p 965974
H. Oettel, G. Schreiber, Residual Stresses in
45.
46.
47.
48.
49.
Nitriding of Steels, 6th Konferenz Metallkundl. Probleme d. Werkstoffentw., TU Bergakademie Freiberg, 1989, p 194199 (in
German)
H. Oettel and G. Schreiber, Formation of
Residual Stresses in the Diffusion Layer,
Nitrieren und Nitrocarburieren, AWT Tagungsband, 1991, p 139151 (in German)
J. Zurn, C. Razim, and J. Gosch, The Effect
of Residual Stress in Case Hardening, Heat
Treatment and Surface Engineering: Proc.
6th International Congress on Heat Treatment of Materials, 1988, p 279283
W. Daves and F.D. Fischer, Numerical
Simulation of the Development of Residual
Stresses during Nitriding, Residual Stresses,
Informationgesellschaft Verlag, 1993, p
957964
B. Larisch, H.-J. Spies, and K. Hock, Bright
Nitriding of Cr-Mo-Steels in Plasma and
Gas, Proceedings of the Second International Conference of Carburizing and Nitriding with Atmospheres, (Cleveland, OH),
ASM International, 1995, p 213217
N.L. Loh and L.W. Siew, Residual Stress
50.
51.
52.
53.
54.
Profiles of Plasma Nitrided Steels, Surf.

Eng., Vol 15 (No. 2), 1999, p 137142
H. Oettel and G. Schreiber, Calculation of
Macro-stresses in Nitrided Steels, 4th Conference on Residual Stresses, Baltimore,
1994, p 12851291
Z. Kolozsvary, S. Janosi, V. Sandor, A.
Rusu, and T. Reti, A New Approach in
Characterizing the Mechanical Properties of
Surface Layers, presented at 18th ASM
Heat Treating Society Conference and
Expo, Rosemont, IL, 1315 October 1998
H.-J. Spies, Fatigue Behaviour of Nitrided
Steels, Steel Re., Vol 64 (No. 8/9), 1993
H.-J. Spies, K.H. Kloos, J. Adelmann, T.
Kaminsky, and P. Trubitz, Modeling of
Bending Fatigue resistance of Samplers Resembling Engineering Components with
Standard Wohler-bands, Mat. Wiss. Werkstofftech., Vol 27, 1996, p 6071 (in German)
H. Spies, N.D. Tan, and F. Vogt, Influence
of Morphology of Nitrided Layers on the
Fracture Toughness of Notched Samples,
HTM 48, Vol 3, Rudolf Haufe Verlage,
Freiburg, Germany, 1993, p153161 (in
German)

DOI: 10.1361/hrsd2002p220

Induction Hardening
J. Grum, University of Ljubljana, Ljubljana, Slovenia
INDUCTION SURFACE HARDENING offers a number of advantages over other heat

treatment methods. The most important advantages are short heat treatment times, good repeatability concerning the hardened-layer quality, small or negligible subsequent distortion,
and a minimum subsequent product surface oxidation. Induction hardening offers good possibilities for automation and can be incorporated
into a manufacturing cell (Ref 1, 2).
Induction surface hardening is applicable to
axisymmetric or near-axisymmetric steel or cast
iron machine parts that are produced in substantial volumes. There are two basic techniques for
induction hardening of machine parts: single
shot and scanning. The former employs selective heating and quenching to harden a specific area or areas of the machine part in one
operation. The latter is usually applied to progressively harden long, continuous sections such
as shafts and spindles. In this instance, the scanning inductor traverses the length of the section,
heating only a relatively small area at any given
time, and is followed closely by the quench arrangement, which is often an integral part of the
inductor (Ref 1, 3).
A basic characteristic of induction hardening
is that heat is generated in the workpiece due to
the skin effect. If a ferromagnetic material is surrounded by a load coil or inductor through which
high-frequency alternating current passes, flow
of eddy currents is induced in the surface layer
of the workpiece due to the alternating magnetic
field. The flow of these currents heats the workpiece. The mode of heating may be influenced
by the workpiece arrangement in the inductor as
well as by electric parameters. Heating is very
fast and very reliable. Consequently, induction
heating, that is, induction hardening, has become
successfully established in automated production systems. Induction heating is suited to
small, medium-size, and large, as well as extremely large, machine parts since, by varying
energy input, heating to the required temperature
and the specified case depth may be ensured.
Flame hardening is suited mostly for individual hardening of parts with large dimensions and
of comparatively uncomplicated shapes. A characteristic of flame hardening is that a gas mixture
is burning at the blowpipe outlet and thus indirectly heating the part surface. The machines, de-
vices, and accessories used in flame hardening

are comparatively simple and inexpensive, but
the process requires an operator with much experience in blowpipe positioning if a uniform
case depth without softer areas within the hardened layer is to be obtained (Ref 4, 5).
Recent hardening processes related to surface
hardening are laser hardening (Ref 6, 7) and
electron beam hardening (Ref 8). In both cases,
heat is generated in the surface layer of the workpiece due to the interaction of the beam and the
material. From the heat treatment point of view,
laser can be considered a versatile and flexible
high-intensity heat source that can operate in air.
It is capable of undertaking a range of processes,
essentially simultaneously, since the laser beam
can be directed through air by metal mirrors and
switched and shared among a number of workstations. Manipulative techniques using mirrors
allow the beam to be directed to the areas not
accessible by other techniques, for example, the
bores of tubes (Ref 6, 7).
Deficiencies of laser hardening are a very lowenergy efficiency in transformation of electric
energy into heat energy and a comparatively
high cost of investment. Laser hardening has
lately become successfully established for the
following specific applications only:
Hardening of small products of intricate
shapes, which are hard to adapt to induction

heating
Hardening of small internal surfaces such as
small boreholes for which inductors small
enough cannot be produced
Hardening of parts with exacting shapes, with
which admissible distortion is exceeded in induction hardening
From the technology viewpoint, these surface
hardening processes are very much alike since
they all have to ensure adequate energy input and
the case depth required. In the same manner, regardless of the hardening process applied, in the
same steel the same microstructural changes,
very similar microhardness variations, and similar variations of residual stresses within the hardened surface layer may be achieved.
In induction hardening, the surface of suitable
materialsusually plain-carbon or low-alloy
steels or cast ironsis austenitized and then
quenched to produce a hard martensitic surface
layer, which is usually tempered in a subsequent

operation. Case depths normally range between
0.5 and 5 mm. Case hardnesses amount typically
to around 700 HV after hardening and range between 600 and 650 HV after tempering at 180 to
220 C (Ref 1).
Normally the hardening process also introduces compressive stresses into the surface layers, leading to an improvement in fatigue properties. Hardened partsalways ground because
of their high hardnessrequire a minimum level
of final grinding. This can only be achieved by
a minimum oversize of a machine part after hardening, thus shortening the final grinding time and
reducing costs of the final grinding to a minimum. With an automated manufacturing cell,
one should be very careful when selecting individual machining processes as well as machining
conditions related to them. Ensuring the required
internal stresses in a workpiece during individual
machining processes should be a basic criterion
of such a selection. In those cases when the internal stresses in the workpiece during the machining process exceed the yield stress, the
operation results in workpiece distortion and residual internal stresses. The workpiece distortion, in turn, results in more aggressive removal
of the material by grinding as well as a longer
grinding time, higher machining costs, and a
less-controlled residual stress condition. The
workpiece distortion may be reduced by subsequent straightening, that is, by material plasticizing, which, however, requires an additional
technological operation, including appropriate
machines. This solution is thus suited only to
exceptional cases when a particular machining
process produces the workpiece distortion regardless of the machining conditions. In such
cases, the sole solution seems to be a change of
shape and product dimensions so that material
plasticizing during the machining process can be
prevented.
It is characteristic of induction hardening that
machine parts show comparatively high compressive stresses due to a lower density of the
martensitic surface layer. The compressive
stresses in the surface layer act as a prestress on
the most strongly loaded surface layer, which increases the load capacity of the machine part and
prevents crack formation or propagation at the
surface. The machine parts treated in this way
Induction Hardening / 221

are suited for the most exacting thermomechanical loads since their susceptibility to material
fatigue is considerably lower. Consequently,
much longer operation life of the parts can be
expected.
Residual internal stresses, that is, the so-called
residual stresses, are the stresses present in a material or a workpiece/machine part when there is
no external force and/or external moment acting
upon it. The residual stresses in metallic machine
parts have attracted the attention of technicians
and engineers only after manufacturing processes improved to the level at which the accuracy of manufacture exceeded the size of deformation, that is, distortion, of a workpiece/
machine part.
Thus it was in the 1850s that the effect of internal stresses on plastic strain, for example, of
machine parts was already known. It was then
that experts introduced measurement of individual dimensions of machine parts. For a given
type of machining process, they connected the
influence of the selected machining conditions
with the size of deviations of dimensions. This
was also the beginning of an expert approach to
the selection of the most suitable machining conditions based on the criterion of minimum deviations of dimensions, that is, minimum machine part distortion. Still, at the time of
publication, measurement of individual machine
part dimensions is a very practical, uncomplicated, and reliable method of machine part quality assessment.
The surface and surface-layer conditions of
the most exacting machine parts, however, are
monitored increasingly by means of the socalled surface integrity. This is a scientific discipline providing an integral assessment of the
surface and surface layers and defined at the beginning of the 1960s. For high-quality machine
parts and products subjected to heavy thermomechanical loads, different levels of description
of the surface integrity were defined. A basic
level of the surface-integrity description includes
measurement of roughness and the analysis of
microstructure and microhardness in the thin
104
Strip
Strip heating,
heating
contour
contour hardening,
hardening
single-shot hardening
Power (P), kW
103
Preheating
102
Tempering
Scan hardening
10
1
102
101
10
102
103
Frequency (f ), kHz
Fig. 1
Typical power/frequency regions of induction

heat treatment applications. Source: Ref 13
surface layer resulting from the machining process under given machining conditions. The second level of the surface-integrity description includes studies of residual stresses in the surface
layer and of mechanical properties of the given
material. The third level of the surface-integrity
description includes tests making clear the behavior of the given part during machining. More
detailed information on the levels of surface integrity description may be obtained in Ref 9 and
10.
The previous paragraphs indicate that the
knowledge of residual stresses in induction hardening alone does not permit a reliable assessment
of the influence of the sum of the residual
stresses present on the behavior of a machine
part in an assembly under given thermomechanical loads.
Characteristics of Induction
Hardening
Induction heat treatment is a segment of the
much larger technical field of induction heating,
which combines many other industrial processes
using the phenomenon of heating by induction
(Ref 1114).
Induction heating is often one of the most effective heat treatment processes available for a
variety of applications including surface hardening, through-hardening tempering and stress
relief, annealing and normalizing, grain refinement, precipitation hardening or aging, and sintering of powdered metals.
In most of these applications, induction heating is used to selectively heat only the portion
of the workpiece that requires treatment (Ref 1,
15, 16). This usually means that the process can
be accomplished in a relatively short time and
with high efficiency because energy is applied to
the workpiece only where it is needed.
One of the main features of induction heating
compared with conventional heating procedures
is that heat is generated in the workpiece itself.
In conventional heating procedures the heat input achieved is only 5 to 200 kJ/m2 s energy,
whereas in induction heating this energy input is
300 MJ/m2 s. In induction heating, heat penetrates the workpiece with the aid of highfrequency alternating current; the choice of frequency depends on heating requirements.
Induction heating power supplies are frequency
changers that convert the available utility-line
frequency power to the desired single-phase
power at the frequency required by the induction
heating process. They are often referred to as
converters, inverters, or oscillators, but they are
generally a combination of these. The converter
portion of the power supply converts the linefrequency alternating-current input to direct current, and the inverter or oscillator portion
changes the direct current to single-phase alternating current of the required heating frequency.
Many different power supply types and models are available to meet the heating requirements of a nearly endless variety of induction
heating applications (Ref 13). The specific application will dictate the frequency, power level
(Fig. 1), and other inductor parameters such as
coil voltage, current, and power factor (cos u) or
Q factor.
In induction surface hardening, it is necessary
to carefully plan individual characteristic phases
of manufacture of machine parts, which are:
Design of a machine part from a blank to its
final shape with additions for final grinding
A technique and conditions of surface induc-
tion heating
A technique and a quenching process

Tempering the surface layer after hardening
Final machining of the machine part by grind-
ing
The selection of a blank is related to various
steel semi-finished products or various primary
forming processes such as casting, working, and
so forth. Various shapes of blanks do not influence the selection of the conditions of induction
heating, but they result in different variations of
residual stresses, which have to be considered
together with quenching residual stresses.
Because of the character of the forming process itself, particularly the prescribed thermomechanical conditions of forming the blank, the
expected stress state is obtained. Further forming
of blanks by cutting also produces changes in the
stress state depending on the machining process
applied and the machining conditions. Very often with very complex workpieces, the thermomechanical conditions can cause a certain distortion of workpiece dimensions in addition to
residual stresses. No studies treating a relationship between the stress state in the workpiece
after primary forming and that after secondary
forming could be traced. The primary reason is
that the primary forming, that is, blank shaping,
is usually carried out in one company and the
secondary forming, that is, manufacture of a final
product, in another. It has also been proved that
with suitable heating conditions in different heat
treatment processes, certain heat effects characteristic of a certain annealing process that affect
the range of residual stresses can be obtained.
The difference between induction heating and
furnace heating is related to considerable
changes of heating times and rates up to the hardening temperature and, consequently, to heating
costs. Induction hardening is most often used for
surface hardening of machine components and
has the following advantages over other procedures (Ref 1, 5, 11):
Heating times are relatively short.
Heating procedure is not strictly governed by
hardening temperature. All that matters is that

the heating process does not end at a temperature low enough for transformation into
austenite to occur. The maximum temperature
of heating is limited by the solidus-line temperature, since the process is to be carried out
while the material is in a solid condition.
With a short heating time, there is no danger
that at higher austenitization temperatures the

austenite grains would grow, which also
means that there is no danger of formation of
coarse and brittle martensite .
The quenching procedure is easy to perform,
contributing to short surface hardening times.
In progressive induction hardening, a spray
coil is located right beneath the inductor, directly quenching the heated surface. In single-
103
Skin thickness (), 103 cm
f = 104
102
f = 105
f = 5 105
10
f = 106
f = 5 106
1
102
103
104
Power density (Q ), W/cm2

Reference depth of skin effect as a function of
power density and selected generator frequency
for ferromagnetic steel. Source: Ref 1
Fig. 2
Power
density
curves
100
20
3 kHz
450 kHz
0.5
10
Heating
time
curves
10 kHz
5
Surface power density (Q), MW/mm2
3 kHz
50
Heating time (t H ), s
200
10
0.2
10 kHz
0.1
0.2
0.5
450 kHz
5
10
2
20
Hardened depth (z ), mm
Influence of high-frequency generator on selection of power density and heating time with
given thickness of surface induction hardened layer.
Source: Ref 1
Fig. 3
shot hardening, the inductor is designed so

that it heats as well as quenches. The coil
around the workpiece functions as an inductor in the heating phase, and after the austenitization temperature has been reached the
current is interrupted and the coil starts functioning as a quenching spray.
Induction hardening is a short procedure that
does not require any additional protection
against oxidation. Thus, compared with other
similar procedures such as cementation, it
does not require much subsequent machining.
Because of the nature of the procedure, the
workpieces are less susceptible to undesirable
deformation after induction hardening, especially if they are symmetrical. The volume
changes in the workpieces after hardening the
surface layer can be very accurately predicted
or estimated. The volume changes after induction hardening of thin layers are so small
that quite often the function of the machine
component is not affected.
Especially in induction hardening of thin layers and workpieces with low mass, it is possible to achieve the desired critical cooling
rate by self-cooling in air alone, that is, by
heat conduction from the heated surface layer
into the remaining cold part of the workpiece.
With thicker layers and workpieces of greater
mass, it is necessary to use such quenching
agents that move the actual cooling rates
close to the critical cooling rate. These requirements can be met with the right selection
of quenching oils or polymer water solutions.
Practical experience has shown that polymer
water solutions are very suitable for quenching induction-heated surfaces, since with the
right choice of concentration of the polymer
water solution optimal quenching can be ensured.
The induction hardening procedure enables
the engineer, by simply adapting the shape of
the induction coil, to ensure a desired shape
of the hardened profile of the surface layer.
Likewise, the engineer can surface harden
only that part of the surface (local hardening)
on which a certain increased level of hardness
and wear resistance are wanted. One of the
main advantages of induction hardening is
that it makes it possible to harden a surface
layer only on certain places, at defined penetration depth and shape.
Induction hardening can be fully automated
and is especially suitable for large series of
workpieces.
Induction hardening always leaves compressive residual stresses in the surface layer,
which make machine components more resistant to dynamic loads. Compressive residual
stresses in the surface layer after induction
hardening prevent the occurrence of cracks in
dynamically loaded components and prevent
the growth of existing cracks on the workpiece surface if these are present due to hardening or hardening and grinding.
Induction hardening is appropriate for smallsized workpieces, since by well-chosen tech-
nology of heating and cooling or quenching

one can ensure a hardened surface layer and
a refined core. Thus one can create the required wear resistance of the machine component on a certain location as well as the
required load-bearing capacity of the component with only a slight loss in toughness of
the core.
In induction heating there are a number of parameters influencing the heating process with the
given heat treatment. This article discusses the
influence of individual parameters on surface induction hardening. In surface induction hardening the heating conditions should be treated as a
system consisting of a high-frequency generator,
an induction loop, and the workpiece, which has
to be defined by the characteristic system parameters.
The course of surface induction hardening depends on:
The type of steel or cast iron
Electrical and magnetic properties of the ma-
terial
The effect of the primary material microstruc
ture on the properties of the hardened microstructure

The time/temperature relationship in induction hardening and possible tempering
The selection of power density and current
frequency
The workpiece shape and the shape of the
hardened surface layer
A correct execution of the coupling between
the induction loop and the workpiece
Selection of appropriate heating and hardening processes
Selection of the induction loop with or without magnetic-flux concentrator
Time/Temperature Dependence in
Induction Heating
The time variation of temperature in induction
heating of the thin surface layer depends on the
type and shape of the induction coil used,
alternating-current frequency f (Hz), and power
density Q (W/cm2). Power density is defined by
the selected power of the high-frequency (HF)
generator and the surface layer of the workpiece.
Surface heating depends on the coupling between the induction coil and the workpiece. Figure 2 shows the influence of the selected power
density and frequency on the reference depth of
the skin effect in a ferromagnetic material. A
higher power density results in a greater reference depth of the skin effect and a greater depth
of the heated layer with the same maximum temperature obtained at the workpiece surface. (Ref
1). Figure 3 shows the interdependence between
the heating parametersthat is, power density
and generator frequencyas a function of the
specified depth of hardened layer and the heating
time required for single-shot techniques of surface induction hardening (Ref 1). Although the
data supplied by the diagram provide basic in-

formation, they make the selection of an optimal
surface induction condition easier. With the
scanning technique of surface induction hardening, however, instead of time, the speed of
workpiece movement ensuring the depth of
hardened layer required should be defined. Generally, it may be stated that longer heating times
are required with smaller power densities and
vice versa. Also, for the same depth of hardened
layer, longer heating times are required with
lower current frequencies. With regard to the
depth of hardened layer selected between 0.5 and
10.0 mm, generator frequencies of 450, 10, and
3 kHz can be selected in the single-shot surface
hardening technique, in which case appropriate
power densities between 2 and 50 MW/m2 are
obtained. With lower high frequencies such as
10 kHz, the same depth of surface hardened
layer, that is, 2.0 mm, can be ensured only in the
cases when the power density is changed to 50
MW/m2. The lowest generator frequency, that
is, 3 MW/m2, shown in Fig. 3, cannot ensure the
depth of hardened layer smaller than 2.5 mm.
Immediately after tempering an intensive inverse
heat flow is also expected.
With the surface heat treatment processes, the
influence of the selected heating and quenching
conditions on the depth of hardened layer and
the size of transition zone between the hardened
microstructure and the unhardened layer is often
studied. One simple and practical procedure to
control surface heat treatment is measurement of
the time variation of temperature from the beginning to the end of the heating process and
then also from the beginning to the end of the
quenching process. The heat process can be
changed by changing the power density and the
generator frequency, whereas the quenching process can be changed by selecting different
quenching agents and quenching processes. Figure 4 shows the time variation of temperature at
the surface during heating and quenching as well
as in the core, which is due to heat conduction
into the remaining cold workpiece material during heating as well as during quenching (Ref 17).
The total time variation of temperature during

both heating and quenching is called the description of the surface heat treatment process by thermal cycles, which may be divided into those dependent on heating and those dependent on
quenching. The same figure also shows thermal
inversion in workpiece quenching when the core
temperature during quenching is higher than the
workpiece surface temperature.
Denis and coauthors (Ref 18) presented different physical models to describe different processes occurring in the material during surface
induction hardening. The authors analytically
treated studies of a cylindrical specimen 16 mm
in diameter and 48 mm in length and made of
hypoeutectoid carbon steel with 0.43% C. The
theoretical study was then verified by appropriate experimental methods. The experimental verification was to confirm the size of the expected
error of individual characteristic quantities, for
example, variation of temperature/time cycles,
variation of through-thickness microhardness of
the cylindrical specimen, and the throughthickness variation of residual stresses.
Figure 5 shows the measured and calculated
temperature cycles for the surface, the core, and
in a radius r of 7 mm at a depth of 1 mm in the
cylindrical specimen (Ref 18). A comparison of
the temperature cycles shows that in surface induction hardening a thermal flow of 3 MW/mm2
in heating and that of 5.8 MW/mm2 in quenching were selected. Under such heating conditions, a maximum temperature of nearly 1000 C
was attained while heating above 800 C was
somewhat slowed down. The data in the diagram
show that the time required for heating the specimen from the ambient temperature to that of 800
C is equal to the time required for heating from
the latter to the maximum temperature obtained
at the surface, that is, 1.6 s. A temperature cycle
at the surface takes 3.2 to 3.3 s. The temperature
differences between the surface and the core in
a given moment are the greatest during the heating process, DTmax 600 C. During the
quenching process, however, they can reach up
to 360 C. Temperature gradients change much
more in heating than in quenching. In material
heating, there is also a great difference in yield
stress of the material, which can produce plastic
deformation of the material. Another very important finding of the authors (Ref 18) is that the
theoretical model is appropriate since the results
obtained were confirmed by the standard experimental methods such as temperature measurement with thermocouples, diamond pyramid
hardness test, and measurement of residual
stresses with x-ray diffraction. The difference
between the measured and the calculated temperature cycles is very small. It occurs mostly in
heating and reaches up to 60 C at maximum not
taking into account the losses due to eddy currents.
Figure 6 shows two temperature cycles. One
of them takes account of the influence of the enthalpy of phase transformations in the physical
model of surface induction heating and quenching for the surface and the core, and the other
does not. A comparatively small influence of the
enthalpy of phase transformations on the thermal
cycles, however, confirms the small influence on
the changes of the stress state of the cylindrical
specimens during the process of surface induction hardening (Ref 18).
Bruckner and coauthors (Ref 19) determined
the through-thickness temperature variation for
a thin cylindrical plate with a diameter of 120
mm and made of steel Ck45 for hardening and
tempering and for surface hardening with the finite element method. Figure 7, which shows the
temperature changes from the surface toward the
core, also shows that after 14 s a maximum temperature of 870 C is reached at the surface. It is
higher than the temperature of magnetic transformation, which is around 750 C. The through-
1000
900
Power off
Temperature (T ), C
800
Cooling time
Center
Surface
z = 1 mm
Temperature (T ), C
Heating time
Calculated
Measured
Surface
700
600
500
Center
400
300
Surface
200
100
Center
0
Time (t ), s
10
Time (t ), s
Fig. 4
Ref 17
Variation of temperature cycles of surface and

core in surface induction hardening. Source:
Fig. 5
Comparison of variations of calculated and measured temperatures cycles for cylindrical specimens 16 mm in
diameter. Source: Ref 18

thickness temperature variation confirms a slow
change of temperature in the surface layer and
ensures, after quenching, a slow transition from
compressive to tensile residual stresses in the
thin surface layer. This slow change of temperature in the transition zone should be obtained by
an optimal heating condition to attain the desired
variation of the stress state in the surface layer
and to enhance fatigue strength.
Figures 8 and 9 show the variation of temperature from the surface to the core in a round
plate with 30 mm in diameter. Figure 8 shows
the through-thickness temperature variations
from the surface toward the core in a cylindrical
specimen after different heating times (Ref 19).

In the diagram the austenitizing temperature,
TA3 800 C, and the highest temperature obtained at the surface, TA 900 C, are indicated.
With a final heating time t of 9 s and a volume
power density Q of 2.4 103 MW/m3, the highest temperature difference DTsc max between the
core (C) and the surface (S), that is, around 450
C, was obtained. The high temperature obtained
increases slightly in the core due to the cooling
process in the surface, which can have a beneficial influence on residual stress relief. Residual
stress relief can be ensured by heat transfer from
the central part of the specimen to the surface
1000
With enthalpy
Without enthalpy
Surface
Temperature (T ), C
800
600
Center
400
200
10
Time (t ), s
Fig. 6
Variations of temperature cycles of surface and core considering enthalpy of phase transformations and without
it. Source: Ref 18
immediately after quenching. In any case, it is

desired that after quenching the temperature TT
due to the inverse thermal flow does not increase
above 180 to 220 C after core cooling by heat
transfer toward the cold surface. This is the only
way to ensure tempering of tetragonal martensite
and a possible change of the residual austenite
immediately after hardening.
Figure 9 shows the variation of temperature
from the surface toward the core with various
volume power densities Q 0.4 103, 1.2
103, and 2.4 103 MW/m3 (Ref 19). With
smaller volume power densities also the maximum temperature differences between the surface and the core become smaller and the austenitizing times longer. Because of higher
temperatures attained in the core, one should be
aware of the fact that the inverse heat flow may
be so high as to produce even a change in microstructure and thus, consequently, reduction of
hardness in the surface layer.
Figure 10 shows the influence of specimen
mass on the heating time with the same volume
power density and the same maximum final heating temperature of the surface Tmax, 900 C (Ref
19). The maximum temperature in the core
should also be mentioned. With the longest diameter, for example, 120 mm, it equals the ambient temperature, whereas with the specimen diameter of 60 mm it raises to 150 C and with
that of 30 mm to 450 C. The data on the variation of temperature from the surface toward the
core permit a conclusion that for the specimen
with the greatest mass and the diameter of 120
mm additional induction heating of the surface
should be ensured to allow tempering of the surface layer.
Melander (Ref 20) and the author quoted in
Ref 21 first treated surface induction hardening
of low-alloy steel with 0.4% C, 0.7% Mn, and
1.1% Cr for quenching and tempering as well as
1000
Ck45 N
120
1000
1000
9s
800
800
TAC
800
TAC
t H =14 s
400
Tmax
Temperature (T ), C
600
7s
600
5s
400
3s
200
200
TAC
Q1
Temperature (T ), C
Temperature (T ), C
Q2
600
Q3
400
Q, W/m3
200
10 s
t =1s
6s
0
2s
0
20
40
60
10
15
Radius (r ), mm
tH,s
0.4 109
57.4
1.2 109
18.8
2.4 109
9.0
10
15
Radius (r ), mm
Radius (r ), mm
Calculated time variation of through-thickness
temperature in surface induction heating by the
finite-element method. Source: Ref 19
Fig. 7
Calculated time variation of temperature through

specimen cross section in induction heating with
volume power density Q of 2.4 103 MW/m3 and different heating times. Source: Ref 19
Fig. 8
Calculated variation of temperature through

specimen cross section in induction heating up
to hardening temperature with different volume power
densities. Source: Ref 19
Fig. 9

surface hardening. For an analysis, a representative size of machine parts, for example, a cylindrical specimen 40 mm in diameter, was chosen. Induction-heating conditions were selected
so that the temperature of the diameter TA1 was
exceeded to a depth of 5.0 mm. This means that
a change of microstructure and hardness was expected even to the depth of 5.0 mm, where only
partial austenitization was obtained. Such
induction-heating conditions were chosen that
the surface layer was subjected to heating for up
to 35 s. Time variations of temperature at the
surface of the cylindrical specimen and in its
subsurface in the depths of 2.0, 4.0, and 10.0 mm
are shown in Fig. 11. Time/temperature diagrams differ from the previous figures since a
distinctive deviation occurs in heating the specimen material when the temperature of magnetic
1000
t =9s
t = 12 s
t = 14 s
800
TAC
Temperature (T ), C
600
400
200
Material: Ck45
Q = 2.4 109 W/m3
0
15
30
45
60
75
Radius (r ), mm
Calculated variation of temperature through
specimen cross section in induction heating of
specimens of different masses up to given hardening temperature. Source: Ref 19
Fig. 10
1000
Surface
Depth 2.0 mm
Depth 4.0 mm
Depth 10.0 mm
Temperature (T ), C
800
process as well as the quenching process, various

thermokinetic changes of microstructure affecting volume changes, development of internal
stresses, and distortion of machine parts occur;
therefore, knowledge of changes of material
properties such as yield stress of individual
phases which in the given moment compose the
microstructure, permeability, and electric conductivity, as a function of temperature is indispensable.
The entire process of evaluation, that is, modeling, of surface induction hardening is shown
in Fig. 12. Modeling of the induction heating and
quenching processes includes following essential numerical calculations (Ref 23):
domain TA2 is reached and exceeded. The course

of surface heating indicates that the transformation temperature TA2 was obtained in 10 s. In
spite of the same power density, further heating
of the surface up to 850 C was very slow due
to nonmagnetic character of the surface layer of
steel and took another 28 s. Which models of
induction heating are more suitable than others
is difficult to assess. It turned out, however, that
the heating process suggested more reliably ensured a homogeneous, fine austenitic microstructure giving, after quenching, the finest martensite
with the highest possible hardness of the given
steel. Because of the presence of alloying elements and high rates of cooling of the surface
layer, in addition to fine martensite up to 3%
residual austenite also appears. Based on the
time/temperature variation of heating, it can be
evaluated that the depth of hardened layer ranges
between 2.0 and 4.0 mm. From the time/temperature variation in the depth of 10 mm it can
be assumed that the specimen was heated
through the entire volume, that is, to the very
core of the specimen. Because of a different time
variation of temperature of the fourth temperature cycle, it may be concluded that the maximum temperature obtained at the depth of 10
mm is lower than the magnetic transformation
temperature of the given steel. Because of strong
overheating of the cylindrical specimen toward
its center, lower temperature gradients occur, resulting in a reduction of thermal stresses during
the heating process. Because of a comparatively
high temperature in the core, temperature gradients between the surface and the core are generally lower, producing a decrease of axial internal stresses generated during quenching and
also a decrease of axial residual stresses.
Langeot and Delalean (Ref 23) treated numerical simulation of the induction heating process for materials and machine parts from industrial practice. A general method of modeling
is based on knowledge of the magnetic field surrounding the induction loop with the inserted
specimen, that is, workpiece. Knowledge of the
temperature field in the workpiece, namely, permits an analysis of momentary conditions in the
material during working, for example, strain,
stress state during induction heating and quenching, and residual stresses. During the heating
1. Determination of a relationship between the

density of induced currents and the temperature field created in the specimen material (Joule heating effect)
2. Influence of a temperature on electrical and
magnetic properties of the material
3. Determination of the influence of the induced currents or the magnetic field in the
material on transformation of microstructural phases in the material
4. Determination of the influence of individual
microstructural phases in the workpiece material on electromagnetic properties of the
material
5. Description of mechanical characteristics of
the material as a function of temperature
6. Description of the relationship between heat
energy and mechanical strains
7. Description of volume changes due to phase
transformations
8. Description of transformation plasticity in
phase changes during heating and quenching
9. Influence of latent heat of phase transformations on individual characteristics of
surface-induction hardening
10. Description of phase transformations as a
function of thermokinetic processes during
surface-induction heating as well as quenching
11. Influence of strain and distortion of machine
parts on magnetic and electrical properties
of the material
600
1
Magnetic field
analysis
Eddy current
400
Temperature
2
11
200
3
10
5
9
20
40
60
80
Phase
transformation
Time (t ), s
Fig. 11
Ref 20
Time/temperature cycle during single-shot surface induction heating and quenching. Source:
Fig. 12
12
Material
Mechanical properties
Internal stresses
Residual stresses
Strain
Langeots method of mathematical modeling of surface induction heating and quenching. Source: Ref 23

12. Influence of the magnetic field in the material on the development and magnitude of
strain and distortion
The authors of Ref 23 focused primarily on
the treatment of electric energy transforming into
heat energy, making possible heating of the
workpiece material. Heat energy in the workpiece material is represented by temperature
fields in the workpiece near the induction loop.
Unfortunately, the authors deal with the issues
primarily from the viewpoints of electromagnetism and quality assessment of the numerical calculation procedure proposed for individual thermomechanical characteristics of the workpiece
material during heat treatment. The thermomechanical characteristics are plotted graphically as
temperature fields obtained during induction
heating, and as graphical charts of microstructural changes and sizes, namely, shapes, of the
hardened surface layer. A comparison of the numerical results obtained in the thermomechanical state during and after surface induction heating with those obtained by other authors is very
difficult since standard presentations of the results of thermomechanical characteristics such
as time variation of temperature and time variation of the stress state as well as the size and
variation of residual stresses after hardening are
missing.
Inoue and coauthors (Ref 24) treated numerical simulation of single-frequency and dualfrequency surface induction hardening of gear
wheels. Machine parts such as a gear wheel
should show appropriate surface hardness and
through-depth hardness profile of teeth as well
as toughness so that tooth surfaces are highly
wear resistant and the root of tooth is tough
enough to withstand dynamic loads as well as
unexpected mechanical loads. Gear wheels are,
therefore, most often case hardened, but they can
also be nitrided and surface induction hardened.
Dual-frequency induction hardening is related to
the selection of an appropriate preheating procedure with a current having a mean frequency
of 3 to 10 kHz and very fast further heating with
a high frequency of 100 to 250 kHz, which
should be followed by suitable spray quenching.
Practical results confirmed that such a surface
induction hardening process permits, in surface
hardening, the best properties of machine parts
in terms of hardness magnitude and variation in
the hardened layer, high compressive residual
stresses in the surface layer with a minimum distortion of machine parts, and a sufficiently tough
core to be obtained.
The authors had already reported on simulation of dual-frequency induction hardening
based on a metallothermomechanical model (Ref
24). With this model, the influence of the magnetic field on heat generation in the workpiece
due to eddy currents in relation to analyses of
mechanical fields and microstructural changes
was analyzed. Such an approach was needed
since Joule heat is generated in the workpiece
material during induction heating due to eddy
currents, which depend on the current supplied
to the coil, and the temperature dependence of
permeability of the workpiece material should be
taken into account as well. The numerical calculations were related to the formation of the
magnetic field in the workpiece material; thus, it
was necessary to elaborate a comprehensive
electrothermomechanical model to relate electrical quantities to material heating and then, at the
second stage, to relate temperature to the mechanical calculations required. The authors,
therefore, suggest the establishment of a complete relationship between the magnetic and thermomechanical fields in the workpiece material.
The thermomechanical analysis takes as a basis
the temperature field formed due to the action of
the magnetic field and relates to the temperature
dependence of magnetic properties. The heat
Magnetic field
Temperaturedependent
electric conductivity
and magnetic
permeability
Heat generation
due to
induction heating
Thermal stress
Stress
Strain
Temperature
Mechanical work
Latent
heat
Temperaturedependent
phase transformation
Transformation
stress
Stress-induced
transformation
Microstructure
Fig. 13
Relationship between magnetic field and thermomechanical state of workpiece material in numerical modeling of surface induction hardening. Source: Ref 24
generated in the workpiece material by a temperature field permits establishment of a numerical relation between the formation and development of the microstructure, internal residual
stresses, and distortion. In Fig. 13, the authors
show the principle of relationship between the
magnetic field generated in the workpiece material and the changes of thermomechanical
properties of the material (microstructure, yield
stress) for the determination of the stress-strain
state of the workpiece material (Ref 24).
A technique of numerical calculation consists
of two separated analyses shown in Fig. 14the
metallothermomechanical analysis and the electromagnetic analysis to establish heat generation
during heating and calculate thermomechanical
fields for the characteristics selected (Ref 24). A
heat increase in the workpiece was simulated by
the electromagnetic analysis of the field for each
step, taking into account momentary temperatures obtained in the material, its permeability,
and conductivity. These data were then transferred to another group of analyses based on a
description of the state of microstructure for the
determination of stress-strain state of the machine part. The whole procedure was iterated
starting with calculations of the ambient temperature and ending with a maximum temperature of induction heating, which was followed
by quenching. Magnetic permeability generally
depends on magnetic field intensity as well as on
a momentary temperature of the material of the
machine part at the spot considered. In the calculation, a constant permeability of the workpiece material is assumed in the first approximation. In heating, for each iteration change
permeability and electric conductivity are taken
as a function of temperature.
Figures 15(a) and (b) show two different ways
of induction heating of the gear wheels, dual frequency and single frequency (Ref 24). In both
cases the heating process ends up in almost the
same time, around 2.8 s. In dual-frequency heating, heating takes place in two phases, first with
a power of 293 kW and the lower frequency, 3
kHz and then with a power of 339 kW and the
higher frequency, 150 kHz. In the first phase,
which takes 1.8 s, the edge of the gear wheel
warms up to a temperature of 740 C, and after
a pause of 0.9 s follows heating in the second
phase, which takes 1.18 s, up to a final temperature of 858 C, which is followed by quenching.
The entire heating cycle including the pause in
between takes 2.88 s.
In single-frequency heating, heating with a
power of 165 kW and frequency of 25 kHz takes
place without the pause and takes 2.8 s. It is followed by quenching. The maximum temperature
obtained at the surface of the gear wheel at the
end of heating was much higher than in dualfrequency hardening and amounted to 970 C.
temperature cycles in dual-frequency induction
heating of gear wheels (Ref 24). The measured
and calculated temperatures refer to the tip of
tooth, the plane of division of the gear wheel,
and the root of tooth. The temperature cycle of
dual-frequency heating including the pause in-

of the gear wheel and the smallest differences at
the root of tooth.
A comparison of the temperature cycles obtained in single-frequency induction heating
shows important temperature differences among
individual measuring points occurring with the
calculated temperature cycles as well as with the
measured ones. The temperature differences became smaller as soon as the temperature of magnetic transformation of the material was exceeded. Slowed heating of gear wheels above the
temperature of magnetic transformation ensures
longer heating times up to the hardening temperature, that is, the temperature of homogeneous austenitization. Such a heating procedure
results in lower micro- and macrostresses during
the heating process, which is highly important,
since materials show a considerably lower yield
stress at elevated temperatures. In dualfrequency induction heating of gear wheels, similarly as in single-frequency induction heating, it
was found that there were low-temperature gradients through the entire depth of tooth; yet there
were obvious differences in the obtained maximum temperature from which the gear wheels
were quenched. With regard to distortion of the
teeth of the gear wheel, thermal conditions during the quenching process are important. If the
calculated temperatures and the temperatures
measured at the same measuring point in a given
moment are compared, it will be found that the
temperature differences in dual-frequency induction hardening are very small. The temperature
differences among different measuring points
are at their minimum as well. Hence it follows
that the quenching conditions were very favorable, and a very uniform thickness of the hard-
ened layer and a uniform microstructural state

with minimum distortion of the teeth as well as
of the gear wheels could be expected.
Figure 18 shows the temperature distribution
in the half-cut of a gear-wheel tooth in dualfrequency induction heating. Figure 19, however, shows the temperature distribution in the
same cut of the gear-wheel tooth after singlefrequency induction heating (Ref 24). With dualfrequency induction heating it can be found that:
A maximum temperature of 937 C is ob-
tained at the tip of the gear-wheel tooth.
At the end of heating, a temperature higher
than the hardening temperature required was

obtained from the tip to the root of tooth.
The temperature is considerably lower than
the hardening temperature only a few millimeters below the root radium, for example, at
a depth of 6 mm around 150 C.
In single-frequency heating it can be found
that:
A maximum temperature of 1042 C is obtained at the tip of tooth.
A temperature higher than the quenching
temperature is obtained along the entire gearwheel tooth, except at the root of tooth.
A temperature below the hardening temperature is obtained from the root of tooth downward, and 6 mm below the root of tooth it
still amounts to 300 C.
A comparison of both heating modes shows that
heating of the tooth is much more uniform in
dual-frequency heating, and the variation of the
stress distribution during quenching is much
Temperature (T ), C
dicates that low-frequency heating is performed

up to the temperature of magnetic transformation
and high-frequency heating up to the hardening
temperature. It is essential for this dualfrequency induction-hardening procedure that
after hardening tempered martensite is obtained
in the hardened layer and gear-wheel distortion
is smaller than in single-frequency induction
heating.
temperature cycles in single-frequency induction
heating of gear wheels (Ref 24). Because of progressive and uniform induction heating of the
surface layer of the gear wheel after a longer
single-frequency induction heating not including
the pause, a temperature Ts somewhat higher
than in dual-frequency heating was obtained at
the surface of the gear wheel, namely, 1120 C.
The surface temperature was much higher, although Curie temperature was exceeded and the
root of tooth of the gear wheel was heated to a
temperature higher than 800 C. Another very
important finding, resulting from the time variation of temperature, is very similar temperature
variations at different measuring spots at the
tooth of the gear wheel regardless of whether it
is the tip or root of tooth. It is also very important
that the differences between the calculated temperatures and the ones measured at the same
measuring point are small in heating and a little
bit greater in quenching. This indicates that the
model elaborated is suitable, particularly for
heating and the measuring equipment for temperature measurement at individual measuring
points. The greatest temperature differences between the measured and the calculated temperatures in quenching are found at the tip of tooth
Magnetic field
anaylsis
Eddy current
150 kHz
339 kW
858 C
740 C
3 kHz
293 kW
1.8
Heat generation
2.7 2.88
Time (t ), s
(a)
Electromagnetic
properties
Temperature
microstructure
stress/strain
Temperature
Temperature cycle
Heat distribution
Temperature distribution
Volume fraction of martensite
Distortion after hardening
Temperature (T ), C
970 C
Thermomechanical
analysis
25 kHz
165 kW
2.8
Permeability
Conductivity
Fig. 14
Flow chart of the numerical calculation. Source: Ref 24
(b)
Fig. 15
Time (t ), s
Dual-frequency (a) and single-frequency (b) induction heating of gear wheel. Source: Ref 24

more favorable due to a lower temperature,
which ensures lower residual stresses after hardening.
Figure 20 shows a volume-fraction distribution of martensite calculated with the finiteelement method and using a relatively coarse
grid for calculation of the conditions existing in
dual-frequency heating of the gear-wheel tooth
(Ref 24). Figure 21 shows a volume-fraction distribution of martensite in single-frequency heating. The temperature distribution in heating
shows an exceptionally good relationship with
the result of quenching, that is, with the volumefraction distribution of martensite through the
gear-wheel tooth (Ref 24). The results obtained
confirm that the mathematical model used was
correctly chosen and the procedure of numerical
calculation of the temperature field as well as of
the volume-fraction distribution of martensite
suitable. Simulation of the heating process can
be monitored in time by a momentary temperature distribution in the gear-wheel tooth, which
can point out possible weaknesses in heating as
well as quenching. The simulation can be enhanced by a denser grida greater number of
calculated values, for example, of temperature,
distribution of microstructural phases, and so
forth, at the same area of the gear-wheel tooth.
The paper treats a simulation with a very coarse
grid, which provides a less accurate image of the
heating process and the final result of quenching,
that is, of reliability of the assessment of volume
distribution of martensite after quenching, respectively.
transverse to the half-cut of the gear-wheel tooth
in dual-frequency induction heating of the gear
wheel, and Fig. 23 shows the same in singlefrequency induction heating (Ref 24). In the parallel figure, a gray scale indicating the time variations of power density per volume unit in
individual fields of the gear-wheel tooth or the
amount of heat generated in the gear-wheel tooth
is shown. The greatest amount of heat generated
at the end of heating (legend) was 96,500 MW/
cm3 with the dual frequency and 81,320 MW/-
C 103
Max 0.9372
20 mm
Temperature distributions in the tooth gear at t

2.88 s induction heating (dual frequency).
Source: Ref 24
Fig. 18
C 104
Max 0.1042
1200
20 mm
Simulated Measured
Temperature (T ), C
1000
Tip
PCD
800
t = 2.88 s
0.9600
0.8800
0.8000
0.7200
0.6400
0.5600
0.4800
0.4000
0.3200
0.2400
0.1600
0.0800
Min 0.0399
0.1100
0.1000
0.0900
0.0800
0.0700
0.0600
0.0500
0.0400
0.0300
0.0200
0.0100
0.0000
Min 0.0030
Bottom
Temperature distributions in the tooth gear at t

2.8 s induction heating (single frequency).
Source: Ref 24
Fig. 19
600
400
Max 0.9970
1.0000
0.9000
0.8000
0.7000
0.6000
0.5000
0.4000
0.3000
0.2000
0.1000
0.0000
Min 0.0000
200
0
Time (t ), s
Fig. 16
Measured and calculated time variations of temperature in dual-frequency induction heating of teeth followed
by quenching. PCD, pitch circle diameter. Source: Ref 24
1200
Simulated volume fractions of martensite in the

tooth gear after induction surface hardening
(dual frequency). Source: Ref 24
Fig. 20
Simulated Measured
Temperature (T ), C
1000
Tip
PCD
800
Bottom
Max 0.9970
1.0000
0.9000
0.8000
0.7000
0.6000
0.5000
0.4000
0.3000
0.2000
0.1000
0.0000
Min 0.0000
600
400
200
0
Time (t ), s
Fig. 17
Measured and calculated time variations of temperature in single-frequency induction heating of teeth followed by quenching. PCD, pitch circle diameter. Source: Ref 24
Simulated volume fractions of martensite in the

tooth gear after induction surface hardening
(single frequency). Source: Ref 24
Fig. 21

cm3 with the single frequency. The value of the
entire analysis of thermal conditions increases if
they are monitored in frequent time intervals and
with as fine a grid as possible for their calculation.
The theoretically calculated values of the heat
distribution in the gear-wheel tooth can be verified only by relevant measurements carried out
at thermally treated gear wheels. Thus, a micro-
[W/m3]
structure can be analyzed and hardness measured. The authors decided to measure hardness
along the tip, the middle, and the root of tooth.
Because of different tooth widths at different locations, namely, radii, a greater number of measurements with the same distances between the
measuring points were made. Figure 24 shows
the hardness variation at the tip, the pitch circle
diameter (PCD), and the root of gear-wheel tooth
after dual-frequency induction hardening (Ref
24). Along the entire tooth surface from the tip
to the root of tooth the same hardness, 750 HV,
1011
Max 0.0965
t = 2.88 s
800
Hardness, HV
20 mm
0.1100
0.1000
0.9000
0.8000
0.7000
0.6000
0.5000
0.4000
0.3000
0.2000
0.1000
0.0000
0.100
600
400
Tip
PCD
Root
200
0
1010
Max 0.8132
0.9000
0.8250
0.7500
0.6750
0.6000
0.5250
0.4500
0.3750
0.3000
0.2250
0.1500
0.0750
0.0000
Min 0.0000
Generated heat distributions in the tooth gear
at t 2.8 s induction heating (single frequency). Source: Ref 24
Tip
PCD
Root
400
200
1.0
Tip
PCD
Root
0.8
0.6
0.4
0.2
0
Simulated volume fraction of martensite across

the tooth gear after induction surface hardening (dual frequency). PCD, pitch circle diameter. Source:
Ref 24
Fig. 26
Hardness, HV
800
Distance from surface (z ), mm
Volume fraction of martensite (Vf )
Fig. 23
[W/mm3]
600
Fig. 25 Measured hardness profiles across the tooth

gear after induction surface hardening (single
frequency). PCD, pitch circle diameter. Source: Ref 24
Volume fraction of martensite (Vf )
Generated heat distributions in the tooth gear

at t 2.88 s induction heating (dual frequency). Source: Ref 24
20 mm
Fig. 22
t = 2.8 s
Min 0.0000
1.0
0.8
0.6
0.4
Tip
PCD
0.2
0

Measured hardness profiles across the tooth
gear after induction surface hardening (dual
frequency). PCD, pitch circle diameter. Source: Ref 24
Fig. 24
Simulated volume fraction of martensite across

the tooth gear after induction surface hardening (single frequency). PCD, pitch circle diameter. Source:
Ref 24
Fig. 27
was attained. It decreased slowly towards the

tooth core. The hardness of the tooth core after
hardening was 220 HV. The data on the hardness
measured confirm that in dual-frequency heating
the gear-wheel tooth is hardened only to a small
depth, that is, approximately 0.4 to 0.6 mm at
the tip, up to 1.5 mm in the middle, and approximately 1.0 mm at the root.
In single-frequency induction heating, hardening layers are much thicker. Thus, the depth
of hardened layer at the tip of tooth was 1.1 mm
and the greatest depth in the root of tooth. Figure
25 shows the variation of hardness at the same
measuring points at the gear-wheel tooth after
single-frequency induction hardening (Ref 24).
Along the entire length of the tooth surface from
the tip to the root of tooth the same hardness,
800 HV, was attained. At the tip and the root of
tooth it changed gradually with the depth. The
results obtained show that in single-frequency
induction hardening the thickness of the hardened layer increases with the increase in distance
from the tip of tooth.
Conditions are somewhat different in dualfrequency hardening in which the greatest depth
of hardened layer was obtained in the middle of
the tooth, and hardness decreased toward the tip
and the root of tooth. It follows from the hardness variation transverse to the gear-wheel tooth
that the core of the tooth was through-hardened
and a uniform and very high hardness, even up
to 800 HV, was obtained through the entire tooth
width.
Figure 26 shows calculated line changing of
the volume fraction of martensite transverse to
the tooth after dual-frequency induction hardening at the characteristic measuring points. Figure 27 also shows calculated line changing of the
volume fraction of martensite transverse to the
tooth, but this time after single-frequency induction hardening (Ref 24). The efficiency of the
numerical calculation of the volume distribution
of martensite can be assessed by the microhardness measured at the characteristic points. In the
case given, major deviations occurred between
the calculation of volume distribution of martensite and the hardness measured at the given measuring points only in dual-frequency induction
hardening, whereas in single-frequency induction hardening the deviation was much smaller.
Fujio and coauthors (Ref 25) monitored distortion of gear-wheel teeth and the variation of
residual stresses after induction hardening of
gear wheels. The authors treated the conditions
in induction hardening and quenching of gear
wheels made of carbon hardening and tempering
steel S35 experimentally and theoretically. The
experiments of induction hardening were carried
out at gear wheels with an outer diameter of 196
mm and width of 150 mm and the tooth of 22.9
mm in height. The inductive loop surrounding
the gear wheel had nine windings with an inner
diameter of 226 mm and a width equal to the
gear-wheel width, 150 mm. The output power of
the high-frequency generator was 400 kVA and
the frequency 10 kHz. The gear wheels were induction heated to a maximum temperature of 860

C, which was reached in 187 s. During heating,
temperature was measured at four measuring
points, that is, at the tip of tooth 1, at the pitch
circle at the tooth surface 2, in the tooth core 3,
and at a depth of 25 mm from the tip of tooth 4.
Figure 28 shows the gear-wheel tooth with the
marked measuring points for temperature measurement with thermocouples (Ref 25). The time
variation of the temperature measured in heating
the gear wheel at the same measuring points is
shown in Fig. 29(a) and (b) (Ref 25). It follows
from both figures that the magnetic transformation occurred above 700 C. In the calculation,
heat fluxes of the temperature range below the

Curie temperature, q01 75 W/cm2 and that
above the Curie temperature, q02 10 W/
cm2 were taken into account. A comparison of
the calculated and of the measured time variations of temperature in heating shows that there
are certain differences. The maximum temperature differences exceed even 100 C, which is,
according to the authors, still acceptable. These
data on the time variation can, therefore, be used
efficiently in quality assessment of induction
hardening of gear wheels in terms of microstructure, microhardness as well as residual stresses,
and/or distortion of the gear-wheel tooth. The
investigation conducted thus gave useful information on the tooth condition after induction
hardening. The following figures show isotherms transverse to the gear-wheel tooth calculated for the induction heating process.
Figure 30 shows the heating process after 20,
60, and 187 s when it ends (Ref 25). It is very
important that:
The gear-wheel tooth is heated only by 80 C
25 mm
3
Fig. 28
Tooth gear and position at given measuring

points for thermocouple fitting. Source: Ref 25
Temperature (T ), C
1000
(from 780 C after 60 s to 860 C after 187 s

of induction heating) in a comparatively long
heating time, that is, 127 s, when the Curie
temperature of magnetic transformation of
steel was reached.
Isotherms are running almost in the direction
of gear-wheel tooth radius.
The entire gear wheel obtains martensitepearlite microstructure due to the temperature
fields in it.
Figure 31(a) shows the time variation of the
temperature measured at the individual measuring points of the gear-wheel tooth during
quenching with a pressurized water jet (Ref 25).
Data on the temperature variation show that the
cooling temperatures were very similar at mea-
suring points 1 and 2, somewhat lower at measuring point 3, and the lowest at measuring point
4, where after 40 s of quenching it still amounted
to around 350 C.
Figure 31(b) shows the calculated time variation of temperature at the same measuring
points of the gear-wheel tooth during quenching
taking into account selected heat transfer coefficients 0 of 5000 and 10,000 kcal/m2Ch (Ref
25). The calculated time variations of temperature differ from the measured ones only at those
measuring points with higher cooling rates. The
time variation of temperature refers only to the
selected measuring points; therefore, assessment
of the stress state during the quenching process
and of distortion of gear-wheel teeth or of a tooth
is very difficult.
The authors focused on an analysis of the conditions during the quenching process using a calculated distribution of temperature shown as isotherms (0 5000 kcal/m2Ch) at the half-cut
of the gear-wheel tooth. Figure 32 shows the
variation of isotherms during the quenching process after 0.1, 1.0, and finally 10 s of cooling
(Ref 25). In quenching it is very important that
the temperature gradients are high enough to prevent plastification, that is, distortion, of the gear
wheel during quenching. Such quenching conditions including high-temperature gradients result in decrease of favorable compressive residual stresses after hardening. It follows that a
sufficiently high cooling rate has to be ensured
at the tooth surface as well as to the required
depth of hardening. That means that temperature
gradients in the gear-wheel tooth have to be attained so that gear-wheel teeth are hardened with
the minimum cooling rate required, thus ensuring the required martensite transformation in it.
A sequence of images showing isotherms in very
short time intervals from the beginning of
800
1
2
600
3
4
400
200
0
80
120
Time (t ), s
40
860 C
C
780 C
C
End of
induction heating
600 C
C
840 C
C
160 187200
500 C
C
(a)
Temperature (T ), C
820 C
C
700 C
C
1000
400 C
C
800
1
800 C
C
600
3
4
400
q
q
200
01
02
= 75 W/cm2
300 C
C
= 10 W/cm2
600 C
C
780 C
C
Curie point TA = 721 C

2
40
80
120
Time (t ), s
760 C
C
160 187200
200
200 C
C
500 C
C
740 C
C
(b)
Temperature profiles of tooth gear during induction surface heating process. (a) Measured.
(b) Calculated. Source: Ref 25
20 s
Fig. 29
Fig. 30
60 s
187 s
Temperature distributions during induction surface heating process. Source: Ref 25

quenching allows assessment of the size of distortion of the tooth and residual stresses to be
expected after quenching.
Figure 33(a) shows the measured volumefraction distribution of martensite and Fig. 33(b)
the calculated volume-fraction distribution of
martensite when different values of heat transfer
coefficient 5000, 7500, and 10,000 kcal/
m2Chare taken into account (Ref 25). The
main difficulty encountered in the calculation is
how to determine the heat transfer coefficient in
order to obtain a description of real quenching
conditions. It should be considered that quenching is very intensive since it is carried out under
a pressurized water jet; the gear wheel has been
induction heated whereas the core is cold.
negligible influence on the results of heat treatment. The quenching system for induction hardening is defined by eight parameters (Ref 26):
heat time/scan rate, power level, power frequency, part position/rotation, quench flow,
quench temperature, quench time, quench concentration.
Important simultaneous changes of one or
more of these parameters can produce unwanted
effects on the workpiece, which in extreme cases
result in an unsuitable microstructure, deviations
in the depth of hardened profile, unsuitable hardness variation (too low hardness, soft spots), and
exceeding distortion of the machine element.
Quenching Systems for Induction

Hardening
In practice, there are machine parts of different shapes and size, requiring different depth of
hardened layer. In these cases, the type of material chosen and its through-hardenability
should be considered.
Thus, with alloyed steels having good
through-hardenability and/or machine parts with
a comparatively thin hardened surface layer, the
martensite microstructure can be obtained without the application of a quenchant. In such cases,
heat sinks from the surface into the cold workpiece core so that the critical self-cooling rate
obtained at the surface is higher than the critical
cooling rate.
500 C
C
100 C
1000
C
600 C
C
850 C
C
Control of the quenching process of the gear

wheel from high hardening temperatures ensures
the martensite microstructure. The cooling rate
should be high enough to prevent formation of
unwanted softer microstructures such as pearlitic
or bainitic; therefore, it is very important that in
a development of new systems of induction hardening, quenching systems are designed properly
(Ref 22, 26).
Process parameters should be accurately controlled to ensure permanent and reproducible results after hardening of machine parts. It is also
indispensable to define individual parameters
and determine their permissible deviation in
operation, presuming that the deviations have a
700 C
C
200 C
C
800 C
C
300 C
C
750 C
C
400 C
750 C
C
500 C
C
Temperature (T ), C
1000
800
0.1 s
600
Stop of
spouting water
under a pressure 3
400
200
0
10
Fig. 32
1s
10 s
Temperature distributions during cooling process. Source: Ref 25
4
2
20
Time (t ), s
30
Martensite
volumetric
fraction, %
40
90-100
50-90
0.1-50
0-0.1
(a)
1000
Temperature (T ), C
0 = 5000 kcal/(m2 Ch)

0 = 10,000 kcal/(m2 Ch)
800
4
600
3
400
200
1
0
0
10
20
30
40
Time (t ), s
0 = 5000
(b)
(a)
Temperature profiles of tooth gear at given
measuring points during cooling process. (a)
Calculated. (b) Measured. Source: Ref 25
Fig. 31
Fig. 33
0 = 7500
0 = 10,000 kcal/m2 Ch
(b)
Martensite distribution of hardened gear. (a) Measured. (b) Calculated at the maximum cooling rate. Source:
Ref 25

Two common quenching methods using a
quenchant are spray quenching and immersion
quenching. Particularly popular spray quenching
techniques offering different possibilities are:
60
Ratio (inlet:outlet)
1:1
Flow (q ), gpm
50
1:1.7
40
Spray with progressive scan heating (scan
hardening)
1:3
30
Spray after heating in position (single-shot
hardening)
20
Spray quenching out of location after heating
1:5
10
0
10
20
40
30
60
50
Pressure (p), psi
Fig. 34
Effect of inlet:outlet ratio on quench flow versus pressure. Source: Ref 26
Hole size (d), in.
3/16
Flow (q ), gpm
4
5/32
3
1/8
2
1.45
3/32
1/16
0.5
0
10
20
30
Fig. 35
40
50
60
70
80
Pressure (p), psi
15
Effect of hole size on quench flow versus pressure. Source: Ref 26
1000
Curve 1:10%
800
Temperature (T ), C
Curve 3:20%
600
Curve 2:15%
400
Curve 4:25%
200
20
40
60
80
100
120
140
Cooling rate (v), C/s
Fig. 36
Polymer additive ratio affects cooling rate.

Source: Ref 26
Spray quenching is carried out immediately

after heating with a short pause. It is used with
machine parts made of steels with good hardenability and with a comparatively thin hardened
layer. Immersion quenching is carried out after
the inductor has been disconnected from the
high-frequency generator. Gradual hardening is
very efficient and prevents the influence of inverse heat from the core toward the surface so
that no reverse heat flow and no heating of the
already cooled surface occur. This problem is
characteristic of large parts of which the workpiece surfaces should be hardened selectively.
In the past, straight oils and water-soluble oils
were used for quenching after surface induction
heating. Straight oils and water-soluble oils produce mild quenching effects, which reduce difficulties due to distortion and/or crack initiation.
Straight oils require immersion quenching to
minimize the risk of oil ignition whereas watersoluble oils are more suitable for spray quenching. Other quenchants used are water, polymeric
water solutions of different concentration, water
soluble oils, salt water, and so forth.
Polymeric water solutions are inflammable
quenchants. They are prepared in various concentrations to be suitably adapted to various
cooling rates. For spray quenching various flow
rates of quenchants can be selected. They depend
on the size of inlet openings and the number of
outlet nozzle boreholes in one, two, or more
rows at the quenching ring. The fluid-flow rate
can also be regulated by pressure. Technical literature provides data on shaping of quenching
nozzles as well as on the selection of the quenchant flow rate required. Figure 34 shows the
dependence of the flow rate on the quenchant
pressure and the size and number of outlet boreholes (Ref 26). It is very important that a sufficient flow rate of the quenchant is chosen to ensure sufficient heat removal from the workpiece
surface layer during quenching.
The flow rate is controlled also by the number
of boreholes for the flow and spraying of the
quenchant, respectively. Figure 35 shows a dependence of the quenchant flow rate on the borehole diameter of the quenching nozzle, and the
pressure required (Ref 26). An appropriate arrangement of the boreholes in the ring permits a
uniform heat removal after spray and proper and
uniform cooling. The borehole cross sections
should amount to only 15% of the total ring surface available and of the inductor in the singleshot type application, respectively. Rotation velocities of the workpieces in surface induction
heating are relatively high and can be selected
between 800 and 1000 rotations/min, whereas
the rotation velocity of the workpieces during

quenching is considerably lower, namely, 40 to
60 rotations/min. Very high rotation velocities
of the workpieces are required when a uniform
and reproducible depth of hardened layer are to
be ensured and particularly with relatively short
heating times.
Figure 36 shows dependence of changes of
cooling rates on the momentary temperature at
the workpiece surface for four different concentrations of polymeric water solutions (Ref 26).
The selection of an appropriate concentration of
the polymeric water solution for the selected
steel and the depth of hardened layer required
should ensure minimum distortion of the workpiece, which is the final purpose of any heat
treatment. Thus, the required heat removal from
the heated surface layer of the workpiece, considering the thickness of the heated layer, should
be ensured by an appropriate quenching system.
Time Variation of Stresses and

Residual Stresses in a Material
With certain heat treatments, the required microstructural changes and an appropriate magnitude and variation of hardness are obtained. It
is also required that distortion of a machine part
is as small as possible so that the final size of a
machine part can be obtained with minimum precision machining. Consequently, it is very important that distortion be kept under control.
That means that an appropriate technology of
machining and appropriate heat treatment processes should be selected, and internal stresses
lower than the yield stress ensured at any moment and at any location of a machine part during heating and/or cooling. Analytical methods
give an insight into heat treatment conditions if
the time variation of internal stresses is monitored and the dependence between the cooling
time and the specimen temperature at each point
is known. With regard to the specimen temperature determined in this way, at each point on the
specimen yield stress can be determined as well.
Consequently, with different heat treatment conditions, different values of physical quantities
can be selected. They are reflected in the
changed conditions in the material, which makes
it possible to study distortion of the machine part
during cooling and determination of the magnitude of residual stresses.
Figure 37 shows the time variation of axial
stresses at the surface and in individual depths,
that is, from 2.0 to 4.0 mm under the surface and
in the middle of the cylindrical specimen (Ref
21). It follows from the time variation of temperature in the diagram in Fig. 26 that the austenitizing time was shorter than 15 s, which
means that after this time quenching of the surface followed (Ref 21). In surface heating, tensile stresses occurred at the surface due to thermal extension of the surface layer, but when the
temperature of transformation of the microstructure from pearlitic to austenitic was exceeded,

additional compressive stresses occurred at the
surface.
The compressive axial stresses transform into
tensile stresses in the zone of undercooled austenite. In the transformation of undercooled austenite into martensite, the compressive stresses
increase with the increased martensite fraction.
In the core, the opposite sign of the stress was
obtained, namely, the tensile stress. At the end
of quenching, the compressive residual stresses
were obtained in the surface hardened layer
amounting to around 1600 N/mm2 and the
tensile stresses in the core amounting to around
870 N/mm2.
1000
Figure 38 shows a distribution of the individual components of the residual stresses, that is,
the axial component rz, the tangential rT, and
the radial rr (Ref 21). The axial and the tangential components of the residual stresses were
compressive at the surface and amounted to
1600 N/mm2. Then they gradually reduced so
that at a depth of 5.0 mm, the residual stresses
changed their sign to become tensile stresses. A
maximum value of the residual stresses was up
to 1000 N/mm2. The radial stress at the surface
was 0, but it gradually increased to a depth of
5.0 mm to attain 500 N/mm2.
Denis et al. (Ref 18) calculated and measured
the distribution of residual stresses through the
entire radius of a cylindrical specimen of 16 mm
in diameter and 48 mm in length. The specimen
was made of carbon steel with 0.45% C and of
pearlitic-ferritic composition.
500
60
1000
50
Pearlite
Surface
Depth 2.0 mm
Depth 4.0 mm
Center
1500
2000
20
40
60
80
Time (t ), s
Volume fraction (Vf), %
Internal stress (), N/mm2
500
Axial stress distribution at various depths below the surface during single-shot induction
surface hardening. Source: Ref 21
Fig. 37
High-carbon
martensite
Ferrite
40
Low-carbon
martensite
30
20
10
Austenite
Depth below the surface (z), mm

1000
Calculated distribution of microstructures
across the cylinder radius at the end of cooling.
Source: Ref 18
Fig. 39
0
1000
500
Residual stress RS, N/mm2
Residual stress (RS), N/mm2
500
1000
z
T
r
1500
2000
200
0
200
20
40
60
600
T
1000
1400
0
600
80
Calculated
Measured
Depth below the surface (z ), mm
Depth below the surface (z), mm

Residual stress distribution after single-shot induction surface hardening. rz, axial component; rT, tangential component; rr, radial component.
Source: Ref 21
Fig. 38
Calculated and measured residual stress profile

of particular components after induction surface hardening. rz, axial component; rT, tangential component; rr, radial component. Source: Ref 18
Fig. 40
The cylindrical specimen was surface induction heated to a temperature of 980 C and then
quenched in salt water. Figure 39 shows the calculated distribution of individual microstructural
phases from the surface to the center of the cylindrical specimen at the end of cooling (Ref 18).
The initial microstructure was preserved to a radius r of 5.5 mm. In the radii between 5.5 and
6.6 mm, the pearlitic-ferritic microstructure and
low-carbon martensite appeared. With the radii
exceeding 6.6 mm, a fine martensitic microstructure with around 6.0% residual austenite was obtained due to intensive cooling. With the selected
conditions of induction heating and fast cooling
of the specimen, no homogeneous martensite
was formed in the surface hardened layer. Figure
40 shows the calculated and measured variations
of individual components of residual stresses
(Ref 18). The calculated variations of residual
stresses represent high compressive stresses at
the surface, that is, the axial component of residual stresses rz amounting to 803 N/mm2 and
the tangential one rT amounting to 588 N/
mm2. On the contrary, the tensile residual
stresses were calculated after hardening in the
core, that is, with the preserved pearlitic-ferritic
microstructure. Thus the axial component of residual stresses rz calculated for the core
amounted to 370 N/mm2 and the tangential
component rT to 62 N/mm2. The diagram in
Fig. 40 indicates that the maximum tensile
stresses were attained in the transition zone between the hardened and the unhardened layer
(Ref 18). At a greater depth, very low stress gradients occurred and in the opposite direction,
that is, in the thin surface layer to a depth of 2.5
mm, very high gradients of residual stresses occurred. The variations of residual stresses were
determined experimentally by x-ray diffraction.
The great changes of the gradient of the measured residual stresses in the thin surface layer
can also be confirmed by the measurements
made. The results of the measured and calculated
variation of residual stresses in the surface hardened layer agree sufficiently with small local deviations. In order to determine the local deviations of the variation and of residual stresses,
numerous calculations were made with varying
physical parameters of the material and taking
account of different process parameters, too.
A particular problem with steels having
ferritic-pearlitic and pearlitic-ferritic microstructure is that heating of short duration does not
ensure complete homogenization of austenite.
Figure 41 shows the influence of inhomogeneity of austenite on the level of residual
stresses, which is particularly noticeable in the
martensite zone (Ref 18). The initial inhomogeneous austenitic microstructure resulted in the
appearance of the martensitic transformation
with a small fraction of residual austenite. With
regard to the volume fraction of martensite and
residual austenite, plastification of the material
occurred, which produced internal stresses and
the variation of residual stresses, particularly in
the thin surface hardened layer.

Figure 42 shows the calculated distribution of
residual stresses due to heating of the specimen
with heating rates of 200 and 800 C/s to a temperature of 1050 C followed by cooling with a
cooling rate of 1500 C/s (Ref 18). With the high
heating rate v, 800 C/s, a very steep transition
of residual stresses from the compressive zone
to the tensile one was obtained, which resulted
in a decrease of fatigue strength. With the considerably lower heating rate, 200 C/s, it turned
out that the variation of residual stresses in the
thin surface layer was essentially more favorable
than that with the high heating rate. The variation
of the tangential and axial components of residual stresses permits the following findings:
With the lower heating rate, the residual
stresses at the surface are lower by 100 to 200

N/mm2.
The stress gradient for the tangential and axial
components rt and rz is very small in the
subsurface, that is, from a depth of 0.7 to 4.5
mm.
The transition from compressive to tensile residual stresses does not occur earlier at a
depth of 2.6 mm due to a small stress gradient.
residual stresses with the favorable rate of surface induction heating VH1, 200 C/s, which
gives the maximum temperature obtained at the
surface Tmax, 1050 C (Ref 18). This was followed by quenching with two cooling rates,
1500 and 300 C/s.
The variation of the tangential and axial components of residual stresses permits the following findings:
The radial component of residual stresses is 0 at

the surface. In the subsurface it is tensile. Thus
it amounts to around 50 N/mm2 with the higher
heating rate, that is, 800 C/s, and to 400 N/mm2
with the lower one, that is, 200 C/s.
Figure 43 shows simulation of the variation of
sidual stresses are obtained at the surface, that

is, the axial component of 800 N/mm2 and
the tangential component of 1050 N/mm2.
With a low gradient, the axial and tangential
components of the stresses vary; they change
their sign to the tensile zone only in the depth
between 2.6 and 2.8 mm.
With the lower cooling rate, considerably
lower compressive residual stresses are obtained at the surface, that is, an axial component of 275 N/mm2 and a tangential
component of 390 N/mm2.
A comparison of the axial and the tangential
components of residual stresses indicates that
because of the considerably reduced cooling
rate the latter is lower in the thin surface layer
even by a factor of 3.
The gradients of residual stresses with the
higher or lower cooling rate are favorable
since a slow decrease of compressive to tensile residual stresses results in a minor susceptibility of a machine part to fatigue under
dynamic loads.
With the higher cooling rate, compressive re-
1000
2
Residual stress RS
, , N/mm
600
r
200
0
-200
-600
T
-1000
-1400
Heterogeneous austenite
Homogeneous austenite
6
3
4
5
Calculated residual stress profile of particular components after induction surface hardening for heterogeneous
and homogeneous austenite at austenitizing temperature. rz, axial component; rT, tangential component; rr,
radial component. Source: Ref 18
Fig. 41
1200
z
Residual stress RS, N/mm2
800
400
400
800
1200
Tmax = 1050 C
Heating rate 200 C/s
Cooling rate1500 C/s
3
4
5
Depth below the surface (z ) , mm
Simulated residual stress profile of particular components at maximum surface temperature (Tmax 1050 C)
with various heating rates vH1 200 C/s and vH2 800 C/s) and at given cooling rate vc of 1500 C/s. rz,
axial component; rT, tangential component; rr, radial component. Source: Ref 18
Fig. 42
Bruckner et al. (Ref 19) discussed their investigations conducted on residual stresses due to
surface induction hardening as well as the time
variation of internal stresses. They applied the
finite-element method to the calculation of the
time variation of individual components of internal stresses rz, rr, and rt. Figure 44 shows the
time variation of all these components after heating times of 1, 4, 6, and 9 s in the cylindrical
specimen with 30 mm in diameter, which was
heated with a heat flow Q3 of 2.4 106 kW/m3
(Ref 19).
Radial stresses rr vary with time. They are
always 0 N/mm2 at the surface. Then they gradually increase toward the center of the cylindrical specimen. After the heating time t of 4 s they
reach the highest value in the core, 210 N/mm2,
and after the heating time of 9 s they reach the
value of 100 N/mm2. The radial stresses during
heating and cooling may be critical if they essentially exceed the yield stress, which may produce lamination of the surface from the core.
Axial stresses rz vary with time as well. They
are the highest after the heating time t of 4 s, 500
N/mm2 in the core and around 130 N/mm2 at
the surface. The time variation of temperature
through the radius of the cylindrical specimen
indicates that these stresses occur at around 500
C at the surface and at around 150 C in the
core. It can be concluded that because of high
axial internal stresses in the core and considerably lower ones at the surface, they rarely produce distortion of a machine part or even its fail-

ure. It is interesting that after a heating time of
1 s the tensile stress in the core ra is around 250
N/mm2 and at the same time a little higher, that
is, 280 N/mm2, at the surface. The transition
from a compressive stress to a tensile stress
shifts toward greater depths with time.
Tangential internal stresses vary with time,

similar to axial and radial ones, but the absolute
values are somewhat lower than that of axial
stresses. This means that tangential stresses in
the core are tensile and the highest after the heating time of 4 s (rTc 200 N/mm2). At the sur-
1500
Residual stress RS , N/mm2
1000
r
500
500
1000
1500
Tmax = 1050 C
Cooling rate 1500 C/s
Cooling rate 300 C/s
T
0
3
4
6
5
Depth below the surface (z ) , mm
Simulated residual stress profile of particular components at maximum surface temperature (Tmax 1050 C)
with heating rate of 200 C/s, and at various cooling rates (vC1 200 C/s and vC2 800 C/s). rz, axial
component; rT, tangential component; rr, radial component. Source: Ref 18
Fig. 43
4s
200
500
r , N/mm2
Material: Ck 45
Q = 2.4 109 W/m3
4s
400
6s
1s
Surface
100
9s
face the highest compressive stresses occur after

a heating time of 1 s (rTs 270 N/mm2). The
investigation showed that values of tangential
and radial stresses at individual heating times are
very similar in the core, but the differences between axial and tangential stresses there are
greater, that is, by a factor 2 in the axial direction.
Taking into account the magnitude of time
variation of tangential internal stresses during
the heating process, internal stresses should
probably also be monitored during the quenching process. The diagram in Fig. 45 shows the
time variation of tangential internal stresses using the finite-element method in the selected moments, that is, t 9 s (at the beginning of cooling), t 9.1 s, t 10.5 s, t 12.5 s, t 18
s, and t 53 s (Ref 19). At the beginning of the
quenching process, there are very low compressive tangential internal stresses rT, 50 N/
mm2, at the surface. Then they gradually change
to become of a tensile character and amount to
260 N/mm2 after the cooling time of 10.5 s. As
cooling continues, after a total time of 18 s and
a quenching time of 9 s, compressive stresses rT
at the surface amount to 320 N/mm2, and then
slightly increase, during cooling, up to 390 N/
mm2. The variation of tangential residual
stresses after quenching is very favorable since
compressive stresses change slowly from the
surface to a depth of 6.0 mm, where the sign for
the stresses changes and the maximum tensile
residual stresses is reached in a length of 8.5 mm.
Melander (Ref 20) proceeded from a magnetic
field and a temperature field to a phase composition at the given temperature and, consequently, internal stresses and residual stresses
formed due to heat treatment, respectively. The
entire procedure of calculating the characteristics
mentioned is based on suppositions that:
Surface
0
6s
300
5
10
Radius (r ), mm
10.5 s
Material: Ck 45
Q = 2.4 109 W/m3
200
4s
Internal stress (T ), N/mm2
Surface
100
1s
100
9s
0
100
100
200
200
10
5
Radius (r ), mm
15
300
100
12.5 s
9.1 s
0
9.0 s
100
200
300
18 s
400
0
10
5
Radius (r ), mm
53 s
0
15
Internal stress time distributions in particular components during surface induction heating process. rz, axial
component; rT, tangential component; rr, radial component. Source: Ref 19
Surface
6s
300
Fig. 44
300
15
200
9s
200
T , N/mm2
Internal stress (z ), N/mm2
1s
10
15
Radius (r ), mm
Fig. 45
Tangential internal stress component distribution during quenching process. Source: Ref 19

The workpiece is of infinite length and has a
Electromagnetic properties: magnetization
Regardless of the measuring point, the tem-
cylindrical, axially symmetric shape.

Heat conductivity does not depend on the
workpiece length.
The magnetic field varies.
For the calculation, equations for:
Magnetic field
Heat conduction
Phase transformations
Stresses
Input data to the physical model and computer
program consist of:
Basic mesh of nodes and their start temperature
curve (B-H)
Curie temperature TA2, surface magnetic field
strength H0(T), and frequency f
Surface heat flux q(T)
Material properties: thermal conductivity
kk(T), specific heat c(T), density q, heating
diagram (ITh diagram), and cooling diagram
(ITc diagram), latent heat L
perature difference between the measured

temperature and the calculated temperature is
exceptionally small.
At the individual measuring points, the differences between the calculated temperature
and the measured temperature are greater in
the heating cycles.
The greatest difference between the measured
temperature and the calculated temperature in
the heating cycle is found at the measuring
point at the surface, that is, with the radius r
of 18.5 mm.
1000
r = 18.5 mm
800
Temperature (T ), C
r = 14.5
r =0
600
400
200
0
101
102
10
Time (t ), s
Measured temperature cycles (dashed lines)

and calculated temperature cycles (solid lines)
at different radii of cylindrical specimens. Source: Ref 20
Fig. 46
Residual stresses (RS ), N/mm2
800
400
The computer program for solving the temperature field and phase content consists of heating and cooling phases:
1. During heating, the eddy current loss is responsible for the temperature field rise. The
heat flux through the surface is due to the
surrounding air. The phase transformation
into austenite is calculated using a heating diagram (ITh diagram). When the steel part has
been heated for a given time or to a desired
temperature, heating is switched off.
2. Water quenching can be preceded by air cooling for some seconds or vice versa or other
quenchants can be used. The material that
transformed into austenite during heating is
now transformed into ferrite, pearlite, bainite,
or martensite depending on the cooling diagram (ITc diagram). The material untransformed during heating is assumed to remain
untransformed during quenching. Throughout the calculations, material parameters are
chosen from the input curves at the temperature at hand.
3. Temperatures and phase content as a function
of time in every point in the body are the
results obtained. These values together with
mechanical data are used as input data to the
stress program. Stresses are calculated by the
finite element method.
temperature cycles in the middle of the cylindrical specimen with 40 mm diameter and 120 mm
length (Ref 20). In the middle of its length, that
is, at a distance of 60 mm, two thermocouples
are mounted in the radial direction. They measure the temperature during heating and cooling.
The first thermocouple is located at a radius r of
11.5 mm and the second at a radius of 18.5 mm.
In the figure the temperatures measured are indicated by a dotted line and the calculated ones
at the same measuring points by a solid line. The
differences obtained at the given time t are surprisingly small, namely:
Figure 47 shows the variations of individual

components of residual stresses along the radius
r of 20 mm (Ref 20). The authors calculations
show that:
Axial residual stresses are compressive at the
surface (rz 850 N/mm2), the transition

to the tensile zone is found 6.0 mm below the
surface, and the maximum axial tensile stress
rz, 400 N/mm2, occurs at a depth of 10.0 mm.
Tangential residual stresses vary in a very
similar way and have similar values (rT
880 N/mm2) in the surface layer, whereas
tensile residual stresses in the central part rT
amount to 200 N/mm2.
There are no radial residual stresses rr at the
surface, but in the central part they slowly
increase up to 200 N/mm2.
The authors also experimentally verified the
magnitude of axial residual stresses at three measuring points in the thin surface layer and calculated the mean value. The measurement of axial residual stresses was performed with x-ray
diffraction. Table 1 shows measured residual
stress values at various measuring points (Ref
20). The average of three measurements at three
measuring points gave 852 N/mm2 at the surface and 1377 N/mm2 in the subsurface at a
radius of 19.91 mm. The deviations between the
measured and the calculated values are comparatively great. The authors attribute them to measurement errors as well as to the errors in calculation due to inappropriately selected values
of physical parameters.
Fujio et al. (Ref 25) in their third report on
induction hardening focused on studies of distortion of gear wheel teeth and of residual
stresses in gear wheels. The authors measured
the outer diameter and the root diameter of the
gear wheel across two opposite teeth and the root
400
z
T
r
800
1200
0
5.0
Table 1
Residual stress (rz), N/mm2
10.0
15.0
20.0
Radius (r ), mm
Calculated residual stress in particular components after surface induction surface hardening. rz, axial component; rT, tangential component; rr,
radial component. Source: Ref 20
Fig. 47
Measured residual stress values by x-ray diffraction at various measuring points
Radius (r), mm
Point 1
Point 2
Point 3
Mean value
20.03
20.00
19.96
19.59
19.91
858
1019
1443
1109
1315
779
1278
1250
1204
1423
920
1156
1292
1206
1394
852
1151
1328
1173
1377
Source: Ref 20
Change of tip diameter (DT ), mm
0.3
0.2
0.1
Gear No. 2
0
0.3
0.2
0.1
Gear No. 3
0
0.3
0.2
0.1
0
Gear No. 4
114
316
57918
20
22
Tooth number
1124
1326
Experimental
Theoretical
Fig. 48
Gear No. 2
0.2
Change of whole depth (WD ), mm
Change of tip gear wheel diameter after

quenching. Source: Ref 25
0.1
Gear No. 3
0.2
0.1
Gear No. 4
0.2
0.1
0
10
15
20
25
Tooth number
Experimental
Theoretical
Fig. 49
Change of tip whole depth of gear wheel after

parts of the teeth with a micrometer before and

after quenching. The same dimensions were also
calculated theoretically, taking into account the
volume changes due to phase transformations,
and serve as a basis of determination of deviations.
Figure 48 shows a change of the outer diameter of the gear-wheel tooth of gear wheels 2, 3,
and 4 (Ref 25). The gear wheel has 26 teeth,
which means that the deviation was measured
between the first and the twenty-sixth tooth, 3
and 16, and so forth, so that 13 total measurements were made. The results of the measurements are shown as points in the diagram. The
dotted line represents the theoretically calculated
increase in diameter due to heat treatment.
Figure 49 shows a change of tooth height of
gear wheels 2, 3, and 4, as in the previous case
(Ref 25). The dotted lines in the individual diagrams represent the theoretical deviations of
height of the individual teeth by 0.18 mm and
the deviations found by measurement, which are
mostly greater than the theoretical values. The
greatest deviation can be found with gear wheel
4. It even exceeds 0.2 mm.
Figure 50 shows changes of the root diameter.
Calculations show that after heat treatment the
root diameter reduces by around 0.02 mm (Ref
25). The measured values of the same diameter,
however, remained unchanged for gear wheels 2
and 3 and decreased by 0.1 mm at maximum for
gear wheel 4.
A comparison of the outer tooth diameter and
the tooth root diameter, taking account also of
the tooth height, shows that distortion of individual gear wheels is considerably more complicated and cannot be described by the selected
measurement methods; therefore, also changes
of tooth profiles along tooth height were measured. Figure 51 shows tooth profile error curves
after induction surface heating and after quench-
ing (Ref 25). Prior to quenching, the left tooth

surface was marked with the letter a and the
right tooth surface with the letter b. Measurements of the height profile are shown for two
teeth, that is, those marked 1 and 14, of gear
wheels 2 and 3. Considering that the tooth height
is 10 mm and the deviations are plotted in millimeters, their absolute value can be evaluated.
It ranges between 9 and 100 lm at each tooth
surface concerned.
Residual stresses due to surface induction
hardening of the gear wheels were calculated
along the tooth surface during the heating process as well as the quenching process. Figure
52(a) shows distributions of internal stresses
during the heating process after 20, 60, 100, and
187s (Ref 25). The profile variation of internal
stresses indicates that:
In the initial heating phase, the internal
stresses in the root of tooth are compressive

and reach up to 700 N/mm2; with further
heating they change into tensile internal
stresses ranging from 200 to 300 N/mm2.
They are considerably lower in the zone
reaching from the root to the pitch circle; up
to a heating time of 60 s, they are tensile and
reach up to 700 N/mm2; with further heating
they gradually change to compressive stresses
attaining 100 N/mm2.
They are obviously very low and insignificant
in the upper part of the gear wheel tooth, that
is, from the pitch-circle diameter to the tip of
tooth; therefore, they are not plotted. Deformations are for this very reason the greatest
at the tip of gear-wheel tooth.
Figure 52(b) shows changes of internal
stresses during the heating process and the cooling and quenching processes (Ref 25). It is characteristic of the quenching process that considerably higher internal stresses occur during the
Tooth tip
Gear No. 2
0.1
Gear No. 3
Change of root diameter (DR ), mm
0
0.1
Gear No. 2
0.2
0.1
0
0.1
Gear No. 3
Tooth root
0.2
(a)
(b) Left
Tooth No. 1
Right (b)
(a)
(a)
(b) Left
Tooth No. 1
(a)
(b) Left
Tooth No. 14
Right (b)
(a)
(a)
(b) Left
Tooth No. 14
Right (b)
(a)
0.1
0
Gear No. 4
0.1
0.2
1,214,15
4,517,18
7,820,21
10,1123,24
13,1426,1
Tooth number
Experimental
Theoretical
Fig. 50
Change of root gear wheel diameter after

Fig. 51
Right (b)
Tooth profile error curves after induction surface heating (a) and after quenching (b). Source: Ref 25
(a)
The magnitude of residual stresses was measured with x-ray diffraction and strain gages.
Figure 53 shows the results of measurement of
stresses in the pitch circle, that is, at the middle
of the tooth surface for both teeth and in the root
of tooth where critical residual stresses occur
(Ref 25). Using x-ray diffraction, the authors
found that:
The highest residual stresses occur in the root;
at the surface they amount to around 540

N/mm2, then increase to 750 N/mm2 at a
depth of 30 lm to reach 870 N/mm2 at a
depth of 60 lm.
20 s
60 s
100 s
187 s
(a)
1s
0.1 s
Stress
0
(b)
10 s
300 s
daN/mm2
50
Principal stress during induction surface heating and quenching process. (a) Heating process. (b) Quenching process. Source: Ref 25
Fig. 52
A general conclusion can be drawn that the experimental as well as theoretical results agree
very well and provide a lot of useful information
especially to a technologist in manufacture of
gear wheels.
In Ref 24, Inoue et al. treated the variation of
tangential residual stresses in the root of tooth
after single-frequency and dual-frequency surface induction hardening. Figure 54(a) shows the
variation of the tangential residual stresses in the
tooth root where usually the highest residual
stresses in gear wheels are found (Ref 24). The
calculated residual stresses and the measured
ones after dual-frequency hardening refer to the
root width of gear-wheel tooth. The residual
stresses are measured with the x-ray diffraction.
Individual measurements are marked by points.
The calculated residual stresses at the root surface of tooth show the same values as the measured stresses, but the variations of the two differ
strongly, particularly at a depth greater than 0.5
mm. The calculated variation of stresses in the
root cross section of tooth shows a linear dependence to a depth of 2.0 mm. At the surface, the
calculated tangential residual stresses amount to
650 N/mm2, and, with a small stress gradient,
they change so that at a depth of 1.2 mm there
is a transition to tensile stresses, which at a depth
of 1.9 mm attain around 400 N/mm2. The calculated variation of tangential residual stresses
is ideal since the stresses are compressive and
show a sufficiently high value at the surface,
slowly decreasing with depth. Such a variation
ensures high fatigue strength of the material, and
the latter ensures, in turn, a longer life of a machine part.
Figure 54(b) shows the variation of tangential
residual stresses after single-frequency induction
surface hardening (Ref 24). Resulting from a better and a more uniform heating of the central part
of the gear wheel across the tooth height, a
martensite-bainite microstructure forms after
quenching, and very low compressive residual
stresses are obtained through root depth of tooth.
The tangential compressive residual stresses are
very low and amount to around 200 N/mm2.
Because of a very small stress gradient, they
change to give, at a depth of 0.25 mm, the transition of the compressive residual stresses into
the tensile ones reaching 100 N/mm2. Such a
variation of the residual stresses is unfavorable
because it does not provide sufficient compressive residual stresses in the surface layer, which
produces a decrease in fatigue strength of gearwheel teeth. Because of a symmetric variation of
the residual stresses at both sides of the tooth,
that is, the left and the right tooth surfaces, and
a symmetric load state of the gear-wheel tooth
in operation, the tooth may often be subjected to
a considerable tensile load, which results in damage to the tooth root or even a tooth failure.
Figure 55 shows shape distortion after singlefrequency and dual-frequency induction surface
hardening in the half-cut of the gear-wheel tooth
(Ref 24). Figure 55(a) shows simulation of distortion by showing first the initial shape of the
tooth half-cut prior to hardening (dotted line) and
the modified shape after dual-frequency induction surface hardening (continuous line). Figure
55(b) shows simulation of distortion in the same
way as Fig. 55(a), but for the conventional
single-frequency induction surface hardening.
In the figure the calculated actual size of distortion is shown to be 50 lm with an equivalent
geometrical scale of 2.0 mm. A comparison of
Residual stress (RS), da N/mm2
and 1 s, the internal stresses are of the tensile

character along the entire tooth height; they
are the highest in the root of tooth and gradually decrease toward the tooth tip.
Between the quenching times of 1 and 10 s,
the sign of the internal stresses changes, the
latter gradually becomes compressive and
ranges between 100 and 150 N/mm2 in the
upper part of the tooth and up to 600 N/mm2
in the root of tooth.
At the end of quenching, that is, after the
cooling time of 300 s, the internal stresses
change considerably only in the root of tooth
and attain up to 1500 N/mm2.
middle of the tooth surface, that is, at the

pitch circle; at the surface they amount to
around 90 N/mm2, then gradually increase
with a greater depth to reach 400 N/mm2
at a 60 lm depth.
100
80
Root point
60
40
Pitch point
20
0
30
60
Distance from tooth profile (z ), m
90
Residual stress measurements on root and pitch

point below the tooth surface. Source: Ref 25
Fig. 53
800
Residual stresses (RS), N/mm2
In the initial quenching phase, that is, in 0.1
The residual stresses are a bit lower in the
400
0
400
Calculated
Measured
800
1200
2.5
0.5
1.0
1.5
2.0
3.0
(a)
quenching process and are more important in the

root of tooth. The figures show the magnitude
and variation of internal stresses after the cooling
times of 0.1, 1, 10, and 300 s, when the gear
wheel is finally cool. The series of graphical representations of the internal stresses at the tooth
surface indicates that:
800
400
0
400
800
Calculated
Measured
0
2.5
0.5
1.0
1.5
2.0
3.0
(b)
Measured and calculated residual stress distributions below the surface for dual (a) and single (b) frequency. Source: Ref 24
Fig. 54

both cases of hardening from the viewpoint of
tooth distortion clearly shows that distortion in
single-frequency surface induction hardening is
twice that obtained after dual-frequency hardening.
This difference in tooth distortion can be attributed to the fact that in single-frequency surface induction hardening, the martensite transformation occurs in the whole tooth.
Consequently, the volume change of the tooth is
greater and so is distortion. In dual-frequency
surface induction hardening only the tooth contour gets hardened. Consequently, the volume
change of the tooth is smaller and so is distortion; that is, it amounts to only half of that obtained after single-frequency hardening.
In Ref 28, Fujita et al. dealt with the development of an experimental procedure for assessment of surface durability of two hardened materials in a pair. Since the 1980s, steels have been
used most frequently for surface hardening in
manufacture of various gear wheels since this
hardening process is faster and very economical.
In order to clarify surface durability of surface
hardened gear wheels, investigations were conducted on surface defects, rolling fatigue life,
and determination of the optimal thickness of a
hardened layer under sliding/rolling conditions
5 102 mm
(a)
of gear-wheel teeth. For testing of surface durability of surface induction hardened machine
parts, specimens with 60.3 mm in diameter and
made of steel S45C with 0.45% C were selected.
The steel was hardened from an austenitizing
temperature of 850 C for 1 h and then quenched
in water by high tempering at 650 C for 2 h.
Finally, the surface layer was induction hardened
under different conditions so that the depths of
hardened layer were 1.75, 2.95, and 5.00 mm.
After surface hardening, tempering at a temperature of 150 C was carried out for 2 h. Figure 56
shows the hardness distributions of induction
surface hardened roller specimens IA, IB, and IC
(Ref 28). The surface hardnesses of roller specimens are designated HVIA, HVIB, and HVIC, respectively. The hardness of induction surface
hardened layer is about 750 HV, and the core
hardness is 200 HV in any specimen. The case
depth is greatest in specimen IA and followed
by specimens IB and IC. The variations of hardness under different conditions of induction surface hardening show almost the same hardness
at the surface and a very similar throughthickness variation of hardness of the hardened
layer.
Figure 57 shows the residual stress distributions of induction hardened rollers. The variation
of the calculated tangential residual stresses
agrees with the expectations based on the variation of hardness and the hardened-layer thickness (Ref 28). The influence of the variation of
residual stresses on surface durability depends
on the load applied to the machine part concerned. With a continuous compressive load,
plastification of the subsurface can occur; therefore, it is necessary to ensure a sufficient thickness of the hardened layer and a sufficient
strength of the subsurface layer. The sufficient
thickness of the hardened layer can be achieved
by selecting appropriate conditions of induction
Table 2
Case depths of roller specimens
Specimen
mark
Effective case
depth zef, mm
Total case
depth zT, mm
1.32
2.20
3.02
1.75
2.95
5.00
IA
IB
IC
Source: Ref 28
200
5 102 mm
100
(b)
800
IA
IB
IC
600
0
Residual stress (T ), N/mm2
Distortion of tooth gear after induction hardening with dual (a) and single (b) frequency.
Source: Ref 24
Fig. 55
Vickers hardness, HV
surface hardening and the sufficient strength of

the subsurface by selecting appropriate steel or
a steel to which a hardened-and-tempered core
has been added. Machine parts are mostly subjected to combined compressive and transverse
loads that produce surface wear and subsurface
plastification. In the third case, that is, when the
side of a gear wheel is subjected to an alternating
bending load, the magnitude of compressive residual stresses in the surface layer is important
too. An appropriate heat treatment ensuring
compressive residual stresses in the surface can
result in an increase in fatigue strength and, consequently, a longer life of a machine part. Thus
it can be concluded that the knowledge of residual stresses is most important in connection with
those machine parts that are subjected to alternating or pulsating loads, that is, dynamic loads.
In this way, compressive residual and applied
stresses are ensured at any moment so that no
cracks can appear or propagate. The whole procedure proposed by the authors is, with the residual stresses known beforehand, connected
with a search of the optimal thickness of the surface layer in consideration of the radius of curvature of the sliding/rolling cylindrical specimen.
Table 2 (Ref 28) shows the effective and total
case depths of each roller specimen determined
from hardness and residual stress distributions
after various induction surface hardening con-
446 HV
100
200
300
400
500
Heat treatment
conditions
600
700
IA Water quenching at fixed position

IB Coil: 65 10
IC Tempering 2 h 150
Material: S45C
800
400
1.75
2.95
900
5.0
1000
200
0
Fig. 56
1.0
2.0
3.0
4.0
5.0
Distance from roller surface (z ), mm
Hardness distribution of induction surface
hardened roller specimens. Source: Ref 28
Distance from roller surface (z ), mm
Fig. 57
Tangential component of residual stress distributions of induction surface hardened roller specimens. Source:
Ref 28
Induction surface hardened depth (z ), mm
ditions. Figure 58 shows the optimal induction

surface hardened depth for surface durability calculated by the proposed method for various relative radii of curvature, assuming that there are
no mass effects in heat treatments of induction
surface hardening and no size effects for surface
10
700 HV at subsurface hardened layer
210 HV at core
8
6
Total case depth
4
2
Effective case depth
0
0
10
20
30
40
Relative radius of curvature (R ), mm
Fig. 58
50
Optimal induction surface hardened depth for

surface durability. Source: Ref 28
Hardness
Martensite
800
700
600
500
4
400
300
200
Microhardness, HV0.5
Martensite fraction (Vfm ), %
10
durability even though the radius of roller varies

(Ref 28). The effective case depths were obtained by interpolating or extrapolating the calculated total case depths considering the relation
of Table 2. It is clear that as the relative radius
of curvature increases, the optimum inductionsurface-hardened depth increases and the optimal total case depth increases almost rectilinearly up to the relative radius of curvature of 40
mm.
Beck and Simon (Ref 29) treated theoretical
fundamentals of calculation of stresses due to
different heat treatment processes. Among others
they dealt also with surface induction hardening.
The calculations of temperature fields, and finally stress fields, in machine parts during their
heating and quenching are made in accordance
with a physical model and the finite-element
method. Numerical calculations and possibilities
of simulation of surface induction hardening require very extensive and lengthy calculations,
but give a very good insight in the heat treatment
process as a whole. The results obtained until
now in the field of the simulation of surface induction hardening are accurate enough so that
the time variation of the stress state in a machine
part can be predicted efficiently, which permits
optimization of heating and/or quenching technology.
In 1984 Melander (Ref 27) was the first to
show some results of his numerical calculations
and verify them experimentally. Figures 59 and
60 show theoretical and experimental results
concerning the martensite fraction, and the variations of microhardness and residual stresses in
the induction surface hardened layer (Ref 27). In
Fig. 59, the continuous line indicates the calculated variation of the martensite fraction and the
dotted line the variation of microhardness in the
thin surface hardened layer (Ref 27). The alloyed

chrome-molybdenum heat treatment steel AISI
4142 shows very good agreement between the
calculated martensite fraction and the microhardness measured in the surface layer.
Figure 60 shows the through-thickness variation of calculated axial residual stresses (solid
line) and the through-thickness variation of the
measured axial residual stresses (dotted line) of
a surface hardened layer. Although parameters
of surface induction hardening are not known, it
can be confirmed that with regard to the martensite fraction and the microhardness attained, the
variation of the calculated axial residual stresses
is within expectations. The variation of the measured axial residual stresses in the thin surface
hardened layer to a depth of 0.75 mm is surprising. There, the deviation between the calculated
and measured values of the residual stresses is
exceptionally strong. The variation of axial residual stresses depends on the variation of the
martensite fraction in the surface hardened layer
and the increase in the specific volume of the
surface layer with the martensitic microstructure.
The measured variation of hardness indicates
that a probable variation of axial residual stresses
is closer to the one calculated.
The first group of diagrams (Fig. 61) shows
the time variations of through-thickness temperature in single-shot induction heating with the
high-frequency generator with powers of 40, 60,
100, and 180 kW and a medium-frequency current (Ref 30). Characteristic temperature transformations TA1 and TA3 for equilibrium heating
are plotted in the diagram. Induction heating is
a very fast process; therefore, the temperature
transformations mentioned shift to higher temperatures. In order to obtain homogeneous austenite in the surface layer, it is necessary to heat
100
0
0
1
2
3
Measured hardness (dotted line) and calculated martensite distribution (dashed line) after
induction surface hardening. Source: Ref 27
Fig. 59
400
200
0
200
400
600
800
Calculated
Measured
1000
1200
0
1
2
3
tH, s
Heating time
tH, s
Heating time
tH, s
tH, s
Ac3
Ac7
25
7
15
20
4.5
20
7
30
2.5
15
10
10
Fig. 61
7
15
4.5
(a)
Measured axial stress distribution (dotted line)
and calculated (solid line) after induction surface hardening. Source: Ref 27
Fig. 60
Heating time
Heating time
Temperature (T ), C
Residual stress (zRS), N/mm2
600
1150
1100
1050
1000
950
900
850
800
750
700
650
600
550
500
450
400
350
300
250
200
150
100
50
16
24
4.5
32
0
(b)
4.5
2.5
16 24 32 0
8
16
(c)
24
32
16
24
32
(d)
Time/temperature variations in single-shot heating at various HF generation powers. (a) P1 40 kW. (b) P2
60 kW. (c) P3 100 kW. (d) P4 180 kW. Source: Ref 30

the surface layer up to the hatched temperature
range. The results shown in Fig. 61(a) to (d)
make it possible to draw the following conclusions:
The heating curves differ strongly.
The temperature does not reach the hatched
temperature range in 30 s only with the power

of 40 kW.
The hatched temperature range is reached in
a shorter heating time with all other powers
higher than 60 kW.
More intensive heating of the surface layer
occurs (shown in a more steep hardness curve
in the hatched temperature range) with all
powers higher than 60 kW.
The first question raised by an engineer would
be how to evaluate the through-thickness variations of hardness and residual stresses with reference to the time variation of temperature. The
second question would be how to select
induction-heating parameters to obtain optimal
properties of the surface hardened layer with a
minimum energy input. An answer is that one
must take into account the initial steel microstructure and choose an appropriate quenchant
to ensure, in the surface layer, a cooling rate
equal to or a bit higher than the critical cooling
rate required.
The second group of diagrams (Fig. 62) shows
the time and through-thickness variations of
temperature of the specimen at heating with different powers and different velocities of workpiece movement, that is, v1 140 mm/min, v2
220 mm/min, v3 370 mm/min, and v4
680 mm/min in scan hardening (Ref 30). In the
first case, that is, with v1, the hardened layer is
obtained with the powers of 74 and 50 kW. In
the second case, that is, with a higher velocity of
movement v2, the hatched temperature range is

reached only in heating with the powers of 78
and 59 kW. In the third case, that is, with v3, the
hatched temperature range is reached only in
heating with the power of 112 kW, whereas in
the fourth case, that is, with the highest velocity
v4, this temperature range is not reached in spite
of a very high power, namely, 151 kW.
In his contribution Kegel (Ref 31) treated the
conditions in surface induction heating and
quenching of flat steel with a thickness of 3 mm.
In 1943 Seulen and Voss published a study on
surface induction hardening in shot hardening
and scan hardening (Ref 30). Figure 63 shows
the time variation of temperature in surface induction heating with a power density Q of 6.5
kW/cm2, a high frequency f, that is, 250 kHz,
and the selected heat transfer coefficient of
40,000 kcal/m2 Ch after 0.02, 0.04, 0.05, 0.08
(dotted line), and 0.8 s (continuous line) (Ref
31). The other group of curves (continuous lines)
indicates the time and through-thickness variations of temperature due to quenching of the flat
specimen with a water jet. A starting point for a
study of the conditions of cooling with a water
jet is the time and through-thickness dependence
of temperature at the end of 0.1 s heating, when
a maximum temperature, that is, 1050 C, was
reached at the specimen surface. The time/temperature curves of through-thickness cooling of
the specimen indicate that:
The surface temperature changes quickly due
to the influence of the quenchant and heat removal using a cold quenchant.
The temperature toward the specimen core increases due to the cold core and the heat conducted from the surface to the colder core material; thus it increases at a depth of 2 mm
from the initial 130 C even up to 340 C.
Figure 64 shows the variation of rate of induction heating of the surface layer (dotted line)
(Ref 31). At the beginning of heating, that is,
after 0.02 s, the heating rate at the surface
amounts to 12,500 C/s and at the end of heating, that is, after 0.1 s, decreases to 6800 C/s.
The cooling rate at the surface and in the subsurface is around 3700 C/s after 0.15 s and then
gradually decreases so that after 1.0 s it amounts
to only 200 C/s. It is very important that in the
depths greater than 0.8 mm heating of the subsurface occurs. It reaches a maximum rate, that
is, around 2300 C/s, just before surface cooling
with the quenchant (water). After 0.1 s of heating
and additional 0.05 s of cooling, cooling of the
surface layer to a depth of 1.2 mm and then heating in greater depths are obtained. Heating in the
greater depths occurs due to heat convection
from the thin surface layer toward the cooler core
material.
One method of induction surface hardening
appropriate for large gear modules is known as
tooth gap hardening, a progressive hardening
method. In this case, the coil is placed so that it
ensures uniform gap between the coil and the
flanks of two adjacent teeth. The tooth gap hardening method is very demanding and requires a
lot of experience and knowledge to achieve the
desirable properties of the gear. This method is
also known as contour hardening. It is an ideal
method for heat treatment of gears as it increases
the hardness on the surface of the tooth while
only slightly decreasing the load-bearing capacity in the root of the tooth. Gears heat treated in
this way exhibit very good behavior in operation
as compressive residual stresses are present in
the root of the tooth. Gears with induction hardened flanks, given that the dimensioning is carefully carried out, can achieve highest fatigue
1300
1200
74
1100
1000
56
900
Temperature (T ), C
78
112
Power kW
35
TAC
59
800
700
TAC
92
600
151
500
80
140
400
60
39
300
128
92
81
60
40
200
40
100
0
(a)
Fig. 62
12
16
0
(b)
8
12 16 0
4
8
12
(c)
16
115
12
16
(d)
Time/temperature variations in scan hardening at various traveling speeds and various HF generation power. (a) v1 140 mm/min. (b) v2 220 mm/min. (c) v3 370
mm/min. (d) v4 680 mm/min. Source: Ref 30

strength. To verify the results of induction surface hardening, it is necessary to take up certain
measures for controlling the quality of the hardened layer. For this purpose, hardness and microhardness measurements supported by microstructure analysis are commonly used. A
disadvantage of this procedure is also that, due
to the method of heating and quenching (nonuniformly overheated left and right tooth flank),
one might get slightly higher dimensional deviations than in the case of simultaneous hardening
of both flanks of the same tooth (Ref 11). Figure
65 shows the deviation in the dimensions of the
tooth after induction surface hardening in heating with a coil that encircles the gear tooth. Measurements of gear teeth and gear gaps after induction surface hardening show an increased
volume in the tooth root and thus a smaller gap

and increased volume at the tip of the tooth (increased gear diameter). These volume changes
result in a slightly smaller width of the tooth
above the pitch circle and slightly increased
tooth width below the pitch circle of the gear.
These dimensional deviations are relatively
small and negligible in the case of gears with
small diameter or small module, but might become more important in gears with larger modules and greater tooth width (Ref 32, 33).
Surface hardening of gears is one of the most
frequent applications of induction surface hardening. The first part of this article discusses different methods of induction heating of gears;
these methods depend not only on gear size, but
also on the properties expected of gears after heat
treatment. Gears are axisymmetric machine elements; therefore, volume changes due to the
transformation into martensite microstructure in-
1200
1000
Profile before
hardening
Temperature (T ), C
0.1 s
Profile after
hardening
800
0.08
0.2
600
0.3
0.06
0.4
400
0.5
0.7
200
1.0
0.02
0
0
0.04
1
2
3
Distortion of individual tooth shape after induction surface hardening caused by volume
changes. Source: Ref 32
Fig. 65
Time/temperature profiles in surface induction

heating of steel plate with 3 mm thickness at
given heating conditions. Source: Ref 31
Fig. 63
Stress Profiles in Machine Parts in

the Loaded State
12,500
0.1
Cooling rate (v ), C/s
6000
4000
2000
1.0 0.5
0.3
0
2000
Compression
0.02 s
0.04
Tension
10,000
8000
0.2
0.15
4000
0.12
6000
0.7
8000
0
Fig. 64
Ref 31
1
2
Surface induction heating and cooling rate in

thin surface layer at given conditions. Source:
duced by heat treatment are to be expected. Also,

the teeth themselves are symmetrical, which
means that careful application of uniform heating of particular gear teeth and uniform quenching, regardless of the method applied, should not
cause distortion after heat treatment. In induction
surface hardening, compressive residual stresses
are created in the surface, which in addition to
hardness and wear resistance of the surface, creates a considerably high fatigue strength and resistance to bending loads.
Figure 66 shows a part of a gear in cross section with residual stress distribution in the surface hardened layer of the gear tooth. Also in
this case, one obtains a residual stress distribution typical of induction hardening, that is, compressive residual stresses in the hardened layer
with a martensite microstructure, followed by
tensile residual stresses, and at greater depth
again compressive residual stresses. It is known
that the size and distribution of residual stresses
can be influenced by power density. A number
of authors have done research into the effects of
power density or energy input used in heating
with the purpose of determining optimal induction hardening conditions leading to good mechanical properties of gears. Lower power density at the same current frequency requires
longer heating times at the same depth of the
hardened layer and results in higher compressive
residual stresses with a moderate transition of
residual stresses from the compressive into the
tensile region.
In induction hardening of gears, it is necessary
to ensure the most uniform depth of the hardened
layer to achieve a symmetric distribution of residual stresses in the gear tooth cross section.
Fig. 66
Ref 32
Residual stress distribution in the induction surface hardened layer of the gear tooth. Source:
Heat treatment engineers have to be very careful in choosing the conditions of induction surface hardening to benefit from the distribution of
residual stresses achieved in dynamically loaded
parts. In industrial practice, induction surface
hardening should satisfy the requirement of fatigue resistance of machine components. This
worsening of the properties of the machine part
is mainly attributed to tensile residual stresses in
the hardened layer and undesirable hardness distribution in the transition zone from the hardened
into the unhardened part of the subsurface. These
effects are quite natural and are in the first phase
a result of very rapid local heating of the thin
surface layer while in the second phase this is
accompanied by forced quenching, which ensures a critical cooling rate and the occurrence
of martensite microstructure. Both phases in the
induction surface hardening can increase the risk
of fatigue especially if the latter is assessed only
from the point of view of surface hardness. For
successful estimation of the quality of the hardened layer, it is thus recommended one selects
the optimal synergistic effects between the input
electric energy and the interdependence between
the induction coil and the workpiece surface,

connected with the occurrence of eddy currents
in the workpiece surface layer, which leads to
heating. Due to complex synergistic effects in
induction heating or hardening, it is necessary to
carefully study each influence on the properties
of the hardened surface layer (Ref 11, 34). Fatigue strength in machine components that have
been induction surface hardened is increased if
the summed up load tensions and residual
stresses in the surface layer are compressive. In
order to ensure the highest fatigue strength of a
component, it is necessary to provide:
In dynamically loaded components, the sur-
face is prone to fatigue occurrence, so the sur-
Heated above hardening temperaturequenched

Heated below hardening temperaturequenched
Core not heated
(a)
Residual stress
+RS
Tension
Martensite
RS
Compression
Tempered
structure
Residual
stress
(b)
+
External load stress
Tension
Compression
(c)
Tension
Resultant stress
Compression
Resultant
stress
(d)
Stress profile in a round bar in the loaded state
where residual stresses after induction surface
hardening and loading stresses add up. See text for details.
Source: Ref 34
Fig. 67
face must have the highest compressive

stresses.
If the total sum of stresses, namely, load tension plus residual stresses on the surface, are
always compressive, then there is no chance
for the occurrence of cracks and crack
growth.
To ensure good behavior of the surface and
the hardened surface layer in the loaded condition, it is necessary to induce a suitable
prestressing in the surface layer. This can be
achieved by a carefully selected heat treatment method that would create the highest
compressive residual stresses on the surface
and desirable profile of the latter in the hardened subsurface layer.
Induction surface hardening offers great opportunities to ensure considerable compressive stresses in the machine component surface and likewise to ensure restrained
transformation of compressive surface
stresses in the subsurface layer into tensile
residual stresses.
The endurance of machine components subjected to bending and torsion loads can be successfully increased by ensuring sufficiently high
compressive residual stresses. To create a sufficient amount of compressive residual stresses
that would have a favorable distribution is what
any manufacturer wishes to achieve since only
in this way is it possible to increase the reliability
of components in operation. An early failure of
a component in operation might cause catastrophic damage on a machine and thus a loss of
profit. A decisive role in the occurrence of residual stresses is played by the synergistic effects
among the heat treatment method, the kind of
material, and shape of the workpiece. For this
reason, heat treatment has to be treated from the
point of view of heating, overheating, and cooling/quenching and from the point of view of created internal stresses at a certain point during the
treatment. During heat treatment, internal
stresses are created by the temperature differences and phase transformations between the
core and the surface, which is a result of the volume differences between the core and the surface. The created volume differences between
the core and the surface then give rise to internal
stresses. During the process of heating and cooling, internal stresses may produce the following
effects:
In the case when in each moment lower in-
ternal stresses than that of the yield point are

ensured, one can expect higher residual
stresses induced by the heat treatment in the
workpiece, but these would not cause distortions, cracks, or failure.
In the case when during the heat treatment at
a certain moment internal stresses exceed the
yield point, distortions and usually lower residual stresses in the workpiece result.
In an extreme case of very detrimental conditions during the heat treatment when internal stresses are higher than the tensile
strength of the material, the workpiece would
crack, and larger distortions and usually also

high residual stresses would occur.
Numerous changes that take place in the hardened surface layer of the workpiece are always
a result of the heating and quenching conditions.
Therefore, it is necessary to study the events taking place in the workpiece directly after the hardening temperature is reached. In the workpiece
heated to the hardening temperature one can distinguish three zones, shown in Fig. 67(a): the
first where the outer layer is heated to the hardening temperature, the second zone layer heated
below the hardening temperature between the
temperatures TA1 and TA3 for rapid heating, and
finally the third zone layer, which remains cold
during the entire heat treatment process.
Heating to the hardening temperature at a certain depth is followed by quenching. Quenching
results in the occurrence of compressive residual
stresses, as shown in Fig. 67(b), when the familiar transformations in the hardened layer take
place. The second layer does not suffer the same
distortions as the surface layer, though the heating there has been sufficient to improve the properties of the material. In the second layer, hardening is incomplete, which in comparison with
the first layer results in lower hardness and
strength of material. In cases when a machine
component that has been surface induction hardened in this way is subjected to an external load,
as shown in Fig. 67(c), then additional tensile
residual stresses can be noted in the first layer
and the second layer (Fig. 67d).
Since fatigue of a machine component is a
very delicate problem connected with accumulated tensile stresses in the second zone, there is
a danger that, due to the size and shape and location of the hardened trace, the effects of fatigue are transferred to the surface. When a
loaded layer is exposed to tensile external stress
up to the surface, then the surface becomes very
sensitive to the occurrence of cracks. This tendency to crack is further increased if defects are
present in the surface of the workpiece material.
A critical state in the second zone of the hardened surface layer also occurs on locally hardened workpieces where the surface was overheated or heated nonuniformly.
Fatigue Strength of Materials

In professional literature, data on fatigue
strength of materials are usually presented for a
prescribed shape and size of specimens that have
been adjusted to the testing device. Specimens
for fatigue strength are usually cylindrical with
a smaller diameter in the middle part and a
rounded transition into the larger diameter part.
The latter is then usually clamped for testing.
Modes of loading the specimens vary but are
usually either torsion, bending, and/or tension/
compression. The highest fatigue strength is displayed by a specimen subjected to bending loads
(Ref 11, 32). For other modes of loading, the
relation with bending fatigue strength rwb is ex-
Surface hardening applied to specimens with
a smooth shape (curve 1) and specimens with

a slot (curve 2)
Quenching and tempering applied to smooth
specimens (curve 4) and specimens with a
slot (curve 6)
The cementation steel is a chromium nickel steel
with 0.15% C where the specimens were
smoothly shaped (curve 3) and slotted (curve 5).
The achieved hardness on the surface of the
hardened specimens was 56 to 59 HRC and on
the cemented specimens 58 to 59 HRC. In order
to test the effects of the slot on fatigue strength
on all the specimens, whether they were
quenched and tempered, hardened, or cemented,
a slot of equal size and shape was made. It was
made in the middle of the cylindrical specimen
to an equal depth of 0.4 mm. The depth of the
surface hardening and cementation was the
same1.5 mm. The test results showed that
there are significant differences in terms of different heat treatment methods and that the highest fatigue strength was found in surface hardened specimens.
A comparison of the results of fatigue testing
has shown that:
In surface hardened specimens with a smooth
cylindrical shape and with a slot, the differ-
70
ence in the achieved fatigue strength is minimal. This can be attributed to a very desirable
distribution and size of compressive residual
stresses throughout the hardened layer. Since
the depth of the slot reaches only one quarter
of the hardened surface layer, the size of compressive residual stresses at the slot is still
very high, so that the weakening due to the
slot and stress concentration along the slot
does not cause any essential drop in fatigue
strength.
In quenched-and-tempered specimens with
no earlier prestressing of the surface layer ensured, the fatigue strength was considerably
lower than in surface hardened specimens.
Also remarkably lower was the fatigue
strength of quenched-and-tempered specimens with a slot. The results in the graph
show that the difference in fatigue strength in
slotted surface hardened specimens and slotted quenched-and-tempered specimens is 5 to
1.
Smooth cemented specimens displayed 25%
lower fatigue strength than the surface hardened specimens of the same shape, whereas
the cemented specimens with a slot displayed
50% lower fatigue strength than the same surface hardened specimens.
Tooth-root bending stress (b), N/mm2
pressed empirically, that is, fatigue strength in

torsion sw 0.58 rwb, and fatigue strength in
tension/compression rwz 0.70 rwb.
Figure 68 presents Wohler curves for different
modes of dynamically loaded specimens made
from different steels that had been heat treated
in different ways. From among six curves, four
represent specimens made from heat treatable
steel and two for cementation steel specimens
(Ref 11). The heat treatable steel is Cr-Mo-Ni
steel with 0.37% Ni, and the specimens were
heat treated in two different ways:
1400
1200
1000
800
400
300
200
40
Surface-hardened (notched)
(a)
Carburized (smooth)
Tooth-root bending stress (b), N/mm2
50
wb, daN/mm2
104
105
106
107
Number of load cycles, N
60
Quenched and tempered (smooth)
30
Carburized (notched)
20
10
0
Residual Stresses after Induction

Surface Hardening and Finish
Grinding
600
103
Surface-hardened (smooth)
Confidence limit
according to DIN 3990
Quench and tempered (notched)
Fatigue strength of surface hardened and carburized specimens. Curves: 1, surface hardened (smooth); 2, surface hardened (notched); 3, carburized (smooth); 4, quenched and tempered (smooth); 5,
carburized (notched); 6, quench and tempered (notched).
Source: Ref 11
Fig. 68
1400
1200
1000
Confidence limit
according to DIN 3990
800
600
Shot peened
400
300
200
103
Number of load cycles, N 106
The question arises as to the fatigue strength

in those specimens where the slot reaches deeper
than the hardened layer. A regular problem occurring in these cases is crack occurrence and
crack propagation starting from the slot. Due to
the shape of the specimen and the slot, stresses
start concentrating at these places, depending of
course on the kind and size of external loads.
Here it should not be forgotten that along the slot
there are no compressive residual stresses, and
that the size of tensile stresses along the slot
plays a decisive role in crack occurrence.
Figure 69 shows bending stress in the tooth
root subjected to dynamical load versus the number of oscillations (Ref 32). Figure 69(a) shows
the bending stress for the case of induction surface hardening of adjacent flanks of two teeth
with a coil reaching into the tooth gap. In the
process the tooth flank as well as the tooth root
are hardened. This kind of heat treatment of
gears from steels for induction surface hardening
gives a fatigue bending strength in the range
from 320 to 490 N/mm2. Figure 69(b) shows the
same relationship for a case when the induction
coil encircles an individual gear tooth. In this
process, the tooth flank is hardened and the microstructure and hardness in the tooth root are
preserved. A result of this method of hardening
is that the fatigue strength is drastically lowered
to values ranging from 200 to 300 N/mm2 for
the entire range of steels suitable for induction
surface hardening. This is a considerable drop in
fatigue strength of the material in the root of the
tooth (Ref 34, 35).
104
105
106
107
Number of load cycles, N

(b)
Bending fatigue strength of gear tooth at tooth
gap hardening (a) and flank hardening (b) for
various steels. Dashed lines denote confidence limit according to DIN 3990. Source: Ref 32
Fig. 69
For manufacturing crankshafts, a heat treatable 4140 AISI steel was used. This steel is very
appropriate for statically and dynamically loaded
parts of car engines and machines, especially because of its high hardness achieved after hardening (57 HRC). The steel is characterized by
good hardenability and is thus suitable for manufacturing machine parts with large cross sections
in which after refinement a very high strength
can be obtained. After tempering the steel does
not show a tendency to brittleness, and therefore
no special heat treatment procedures are required. This steel is also suitable for surface
hardening (flame hardening, induction surface
hardening) and displays a very good resistance
to wear. However, special attention has to be
paid during product design, and great care
should be given to the design of radius and transition areas to prevent notch effects under dynamical loads. The steel is adapted for the use
in a wide range of temperatures and preserves
high toughness even at low temperatures (Ref
17, 36).
Figure 70 shows residual stress distribution
after induction surface hardening in the central
bearing location (A) and on the extreme left-side
bearing location (C). For both bearing locations

residual stresses were measured on two specimens. The distribution of residual stresses on the
bearing location (A) is, as expected, very similar
on both specimens, the highest compressive residual stresses ranging between 1020 and
1060 N/mm2 at a depth around 250 lm, then
slowly dropping to a depth of 3.5 mm having the
size of around 800 N/mm2, then experiencing
a sharp fall to a depth around 5.5 mm.
The residual stresses distribution after induction surface hardening on the bearing location
(C) is very similar to that in the central bearing
location (A), only its absolute values are slightly
lower and a distinct fall in the residual stresses
can be noted as early as around the depth of 3.5
mm, reaching its minimum value already at a
depth around 5.5 mm.
It can be seen that the residual stress distribution is just as favorable as in the central location except that its absolute values are slightly
lower. The authors believe that the difference in
the residual stress distribution can be related to
the period of overheating on the austenitization
temperature, which resulted in a thinner layer in
austenitization and thus also a thinner hardened
surface layer.
0
Bearing location A;
Specimens A1, A2
200
400
A1
600
C2
800
Bearing location C;
Specimens C1, C2
1000
1200
A2
C1
1
2
3
4
5
6
Residual stress profiles after induction hardening on the specimens A1 and A2 of the mean
bearing location in the middle of the crankshaft and on the
specimens C1 and C2 on the extreme left-side bearing location. Source: Ref 37
Fig. 70
0
Bearing location A:
specimens A2: Normal
grinding conditions
200
400
Bearing location C:
specimens C2: Gentle
grinding conditions
600
800
1000
1200
50
100
150
200
250
300
Depth below the surface (z ), m
Fig. 71
Variation of the residual stresses after induction

surface hardening and grinding. Source: Ref 37
Under different machining conditions of

grinding, different temperature cycles were obtained on the surface and in the depth of the heataffected zone, which has effected microstructural changes and changes in the microhardness
and residual stresses (Ref 3641). In the grinding process, the changes of temperature in the
surface layer of the workpiece are very important. Temperature measurements provide data on
temperature cycles, which, because of the relative movement of the workpiece and the grinder,
indicate the variation of temperature both in
heating and cooling of the given spot at the
workpiece. The maximum temperatures
achieved at the surface and in the surface layer
are also very important. A distinction can be
made among four characteristic cases of temperature cycles:
The maximum temperatures at the surface
and in the surface layer are higher than the

fusion temperature of the workpiece material,
depending on the temperature cycles. Such
conditions may occur due to very sharp grinding conditions or due to the selection of an
inappropriate grinder with regard to the workpiece material. Both influences contribute to
blunting of abrasive grains, which produces a
larger contact zone between the abrasive
grain and the workpiece, which, in turn, produces a stronger overheating of the workpiece. In such cases burns will occur at the
workpiece surface and, consequently, softened spots that do not ensure the working life
required of a component, that is a machine
part.
Thus, on the surface a maximum temperature
higher than the melting temperature of the
workpiece material was obtained. The depth
of the remelted layer is only a few microns
and makes a very fine ledeburite microstructure containing fine cementite spread in residual austenite. The newly formed microstructure has a slightly lower hardness than
martensite. The residual stresses in the thin
surface layer will be tensile due to plastic deformation of the surface layer in grinding
caused by tensile forces in the contact zone
of the workpiece material and the tensile
stresses induced by the occurrence of residual
austenite should be added to this.
The maximum temperature in the contact
zone is lower than the temperature required
for the beginning of fusion of the given material and higher than the austenitizing temperature. The lower temperature, that is the
austenitizing temperature, will shift because
of a high rate of heating of the workpiece toward higher temperatures as known from
transformation diagrams. In such cases, the
surface layer is expected to heat up to such a
temperature as to produce a change in the
crystal structure so that an austenitic microstructure, through which the accompanying
diffusion processes are going on, will form.
Provided the previous microstructure of the
surface layer was martensite-cementite-
carbide, after grinding a finer martensite microstructure with a higher carbon content obtained at the expense of cementite-carbide
phases and with a possibility of a lower residual austenite content may be expected in
the thin surface layer. The modified content
of the cementite-carbide phase depends on the
heating conditions, whereas the content of the
residual austenite depends on the cooling
conditions. Such a microstructure is still a
typically quench microstructure, only in the
very thin surface layer between 50 and 200
lm first austenization and then requenching
occurred. The thickness of the requenched
layer depends on the grinder, the grinding
conditions, and the state of the grinder (sharp
or used grains). Such a microstructure is still
adequate and ensures the working life of the
machine part. The only risk is too high a content of the residual austenite formed or various damages such as cracks, grooves, and
galling to the surface and the surface layer.
The maximum temperature in the contact
zone is lower than the temperature required
for the beginning of austenitization and
higher than the lower temperature, which is
limited by the temperature of steel tempering
and amounts to approximately 200 C. In this
case, the grinding conditions are very mild so
that with the selection of the right kind of
grinder neither major nor important changes
are expected either at the surface or in the
surface layer. In the surface layer only martensite tempering may occur if it was not performed already during the heat treatment of
the workpiece. Such grinding conditions are
very welcome in the cases that the distortion
of the workpiece is relatively small, and the
additions for final fine grinding of the surface
are quite small.
The volume changes that may occur in the surface layer after steel grinding are thus:
Formation of the residual austenite contrib-
utes to the formation of tensile residual

stresses due to grinding.
The lower content of the cementite-carbide
phase produces formation of ledeburite containing the residual austenite, which contributes to the formation of tensile residual
stresses due to grinding.
The tempering effects in the martensite structure also contribute to the formation of tensile
residual stresses due to grinding.
Material plastification due to grinding can
produce compressive or tensile residual
stresses depending on the type of grinding
and the grinding conditions.
Figure 71 shows residual stress changes resulting from induction hardening and grinding
for main bearing spots A and C (Ref 37). The
residual stresses were measured after grinding
only to a depth of 200 lm. The variation of the
residual stresses indicates that the favorable
compressive residual stresses obtained with induction hardening were reduced only slightly so

that in the thin surface layer, which is very critical, very high compressive residual stresses remained. Outer loads never exceeded the high
compressive stresses, which means that there is
no risk of crack initiation in strongly dynamically loaded universal-joint shafts.
Figure 72 shows residual stress changes resulting from induction hardening and grinding
for main bearing spots F and G (Ref 37). The
same holds true for these two bearing spots,
which means that the selected induction hardening and grinding conditions were favorable.
Although grinding produced tensile stresses due
to and characteristic mainly of grinding, the latter were almost negligible with regard to the total
of the preceding residual stresses.
Summary
Induction surface hardening of machine components and especially gears is a very complex
process involving a whole range of possible heat
treatment methods, which are all reflected in either good or bad serviceability of machine components. The heat treatment engineer has to be
aware of the different effects of particular design
shapes of induction coils, to be familiar with
electromagnetic phenomena and eddy currents,
and to have some experience in the right choice
of energy inputs necessary for heating. The energy needed for heating can be provided by
changing the generator power as well as by
changing the frequency of the current. In progressive hardening, to achieve a suitable energy
input, the workpiece feed rate or the rate at which
the coil is moved has to be adjusted, whereas in
single-shot hardening suitable energy input is
achieved by adjusting the heating time with a
high-frequency current. An essential advantage
of induction surface hardening is that it is possible to achieve a sufficient repeatability of the
hardened layer thickness on the workpiece as
well as a desirable or even prescribed hardened
layer profile, ensuring sufficient hardness and favorable distribution of residual stresses in the
hardened layer. A variety of steels and a whole
0
200
400
Bearing location G
Abusive conditions
600
800
Bearing location F
Normal grinding conditions
1000
1200
Fig. 72
50
100
150 200
250
Depth below the surface (z ), m
300
Variation of the residual stresses after induction

surface hardening and grinding. Source: Ref 37
range of induction-hardening methods provide

the possibilities for very accurate planning of the
size and distribution of residual stresses. This is
of growing importance since manufacturers are
more and more often required to produce machine components, which, among other surface
properties, will have to have quite specific residual stress distribution along the depth of the
hardened layer. It has become a proven fact that
high compressive stresses ensure high fatigue
strength of machine components and reduce the
danger of the occurrence and growth of cracks
on the surface of components. As far as induction surface hardening is concerned, it is also
quite important to choose the right quenching
medium and quenching method. For this reason,
engineers have to direct their attention not only
to the method of heating and possible overheating of the surface layer, but also to the methods
of quenching and the right choice of the medium
for quenching.
REFERENCES
1. P.A. Hassell and N.V. Ross, Induction Heat
Treating of Steel, Heat Treating, Vol 4,
ASM Handbook, 1991, p 164202
2. S. Lampman, Introduction to Surface Hardening of Steels, Heat Treating, Vol 4, ASM
Handbook, 1991, p 259267
3. K.E. Thelning, Steel and Its Heat Treating,
Chapter 6.6, Induction Hardening, Bofors
Handbook, Butterworth, London and Boston, 1975, p 432451
4. T. Ruglic, Flame Hardening, Heat Treating,
Vol 4, ASM Handbook, 1991, p 268285
5. K.E. Thelning, Steel and Its Heat Treatment,
Chapter 6.7, Flame Hardening, Bofors
Handbook, Butterworth, London and Boston, 1975, p 451465
6. K. Sridhar and A.S Khanna, Laser Surface
Heat Treatment, Chapter 3, Lasers in Surface Engineering, N.B. Dahotre, Ed., Surface Engineering Series, Volume 1, ASM
International, 1998, p 6970
7. O.A. Sandven, Laser Surface Hardening,
Heat Treating, Vol 4, ASM Handbook,
1991, p 286296
8. S. Schiller, S. Panzer, and B. Furchheim,
Electron Beam Surface Hardening, Heat
Treating, Vol 4, ASM Handbook, 1991, p
297311
9. M. Field and J.F. Kahles, Review of Surface
Integrity of Machined Components, Ann.
CIRP, Vol 20 (No. 1), 1970, p 107108
10. M. Field, J.F. Kahles, and J.T. Cammet, Review of Measuring Method for Surface Integrity, Ann. CIRP, Vol 21 (No. 2), 1971, p
219237
11. M.G. Lozinski, Industrial Applications of
Induction Heating, Pergamon Press, Oxford, 1969
12. V.I. Rudnev, R.L. Cook, D.L. Loveless, and
M.R. Black, Introduction Heat Treatment,
Basic Principles, Computation, Coil Construction and Design Considerations, Chap-
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
ter 11A, Steel Heat Treatment Handbook,

G.E. Totten and M.A.H. Howes, Ed., Marcel Dekker, 1997, p 765767
D.L. Loveless, R.L. Cook, and V.I. Rudnev,
Introduction Heat Treatment, Modern
Power Supplies, Load Matching, Process
Control, and Monitoring, Chapter 11B Steel
Heat Treatment Handbook, G.E. Totten and
M.A.H. Howes, Ed., Marcel Dekker, 1997,
p 873874
Electromagnetic Induction and Electric
Conduction in Industry, Chapter 9, Heat
Treatments by Induction, Centre Francais de
IElectricite, 1997
G. Benkowsky, Induktionserwarmung,
VEB Verlag Technik, Berlin, 1985
W. Brunst, Die Inductive Warmebehandlung, Springer-Verlag, Berlin, 1957
S. Zinn and S.L. Semiatin, Elements of Induction Heating (Design, Control and Applications), Electric Power Research Institute and ASM International, 1988
S. Denis, M. Zandona, A. Mey, and S.A.
Boufoussi, Calculation of Internal Stresses
During Surface Heat Treatment of Steels,
presented at the European Conference on
Residual Stresses (Frankfurt, Germany),
1992, Residual Stresses, V. Hauk, H. P.
Hougardy, E. Macherauch, and H.-D. Tietz,
DEM Informationsgesselschaft mbH, Oberursel, 1993, p 10111020
U. Bruckner, W. Schuler, and H. Walter,
Untersuchungen zum Eigenspannungszustand beim inductiven Randschnitharten,
Eigenspannungen: Entstehung-MessungBewertung, Band 1, E. Macherauch and V.
Hank, Ed., Deutsche Gesselschaft fur Metallkunde E.V., Oberursel, 1983, p 293308
M. Melander, Computer Calculations of Residual Stresses due to Induction Hardening,
Eigenspannungen: Entstehung-MessungBewertung, Band 1, E. Macherauch and V.
Hank, Ed., Deutsche Gesselschaft fur Metallkunde E.V., Oberursel, 1983, p 309328
A.J. Fletcher, Thermal Stress and Strain
Generation in Heat Treatment, Chapter 9.2,
Induction Hardening, Elsevier Applied Science, London and New York, 1989, p 182
187
G.H. Ledl, Why Quenching Is Important in
Induction Heating, Met. Prog., Dec 1967, p
200204
O. Longeot and C. Delalean, Simulation numerique des procedes de traitment par induction, Trait. Therm., No. 280, 1995, p
3347
T. Inoue, H. Inoue, F. Ikuda, and T. Horino,
Simulation of Dual Frequency Induction
Hardening Process of a Gear Wheel, Proc.
Third International Conf. Quenching and
Control of Distortion, G.E. Totten, B. Lisicic, and H.M. Tensi, Ed., ASM International, 2426 March 1999, p 243250
H. Fujio, T. Aida, Y. Masumoto, and T.
Tsuruki, Paper No. 16914, Distortion and
Residual Stresses of Gears Caused by Hard-
26.
27.
28.
29.
30.
31.
ening, Third Report, Induction Hardening of

Gears Made of 535 C Carbon Steel, Bull.
JSME, Vol 22 (No. 169), July 1979, p
10011008
D.J. Williams, Quench Systems for Induction Hardening, Reprint from Metal Heat
Treat., July/August 1995
M. Melander, Proc. International Symposium on the Calculation of Internal Stresses
in Heat Treatment of Metallic Materials, E.
Attebo and T. Ericsson, Linkoping, Sweden,
1984, p 399419
K. Fujita, A. Yoshida, and K. Nakase, Surface Durability of Induction Hardened
0.45% Carbon Steel and Its Optimum Case
Depth, Bull. JSME, Vol 22 (No. 169), July,
1979, p 9941000.
G. Beck and A. Simon, Prediction of Residual Stresses, Advances in Surface Treatments: Technology, Application, Effects,
Vol 4, International Guidebook on Residual
Stresses, A. Niku Lari, Ed., Institute for Industrial Technology Transfer and Pergamon
Press, 1987, p 303326
G. Seulen and H. Voss, Oberflachenhartung
mit Induktionserhitzung bei mittleren Frequenzen, Stahl Eisen, Vol 63 (No. 51),
1943, p 929935
K. Kegel, Die Praxis der Induktiven War-
32.
33.
34.
35.
36.
37.
mebehandlung, Werkstatt Betr., Vol 85 (No.

5), 1952, p 183188
G. Parrish, D.W. Ingham, and M. Chaney,
The Submerged Induction Hardening of
Gears, Part 2: Properties Imparted and Problems Solved, Heat Treat. Met., Vol 25, No.
2, 1998, p 4350
G. Parrish, D.W. Ingham, and M. Chaney,
The Submerged Induction Hardening of
Gears, Part 1: Processing and Application
Aspects, Heat Treat. Met., Vol 25 (No. 1),
1998, p 18
S.L. Semiatin and D.E. Stutz, Induction
Heat Treatment of Steel, ASM International,
1987
P.K. Braisch, The Influence of Tempering
and Surface Conditions on the Fatigue Behaviour of Surface Induction Hardened
Parts, Mater. Sci. Forum, Vol 102104,
1992, p 319334
J. Grum and D. Ferlan, Residual Internal
Stresses after Induction Hardening and
Grinding, 17th ASM Heat Treating Society
Conf. Proc., Including the First International Induction Heat Treating Symposium,
D.L. Milan, D.A. Poteet Jr., G.D. Pfaffmann, V. Rudnev, A. Muehlbauer, and
W.A. Albert, Ed., ASM International, 1998,
p 629639
J. Grum, How to Select Induction Surface
38.
39.
40.
41.
Hardening and Finish Grinding Conditions

in Order to Ensure High Compressive Residual Stresses on Machine Part Surface,
The 12th International Federation of Heat
Treatment and Surface Engineering Congress, Congress Proceedings, Vol 1, The Institution of Engineers, Australia, Melbourne, Oct 2000, p 133138
E. Brinksmeier, A Model for the Development of Residual Stresses in Grinding, Advances in Surface Treatments: Technology,
Application, Effects, Vol 5, International
Guidebook on Residual Stresses, A. Niku
Lari, Ed., Institute for Industrial Technology
Transfer and Pergamon Press, 1987, p 173
189
R. Snoeys, M. Maris, and J. Peters, Thermally Induced Damage in Grinding, KeyNote Papers, Ann. CIRP, Vol 27 (No. 2),
1978, p 571581
P.G. Althaus, Residual Stresses in Internal
Grinding, Ind. Diamond Rev., No. 3, 1985,
p 124127
M. Moris and R. Snoeys, Heat Affected
Zone in Grinding Operations, Proc. 14th International Machine Tool Design and Research Conf., F. Koenigsberger and S.A.
Tobias, Ed., Manchester, 1973, p 569
669

DOI: 10.1361/hrsd2002p248

Hardening by Reheating and Quenching

M. Narazaki, Utsunomiya University
G.E. Totten and G.M. Webster, G.E. Totten and Associates Inc.
OF THE VARIOUS STEEL PROCESSING

METHODS, heat treating has the greatest overall impact on control of residual stress and on
dimensional control. Among the various processes involved in heat treating, quenching is
one of the most important. One author has estimated that approximately 20% of the problems
in heat treating relate to heating and as much as
80% of the problems relate to cooling (Ref 1).
This article provides an overview of the effects
of various material- and quenchant-related parameters on residual stress, distortion control,
and cracking. The subjects that are discussed include:
Measurement and evaluation of quenching
power
Effect of process variables on cooling behavior and heat transfer
Effect of cooling characteristics on residual
stress and distortion
Methods of minimizing distortion
Tempering
Prediction of residual stress and distortion
Phase Transformation during Heat

Treating
Phase transformations during heat treating

Metallurgical sources of stress and distortion
Steel Transformation
during reheating and quenching

Effect of materials and quench process design
on distortion
Quenchant selection
Properties such as hardness, strength, ductility, and toughness are dependent on the microstructures that are present in steel. The first step
800
A3
700
A1
Pearlite start
Temperature, C
600
Pearlite end
Bainite
Bainite start
end
500
Bainite end
400
300
Martensite
200
100
0
101
102
103
104
105
Seconds
1
4
Time
Fig. 1
8
15
Minutes
30
60
1
4 68
Hours
16 24
Time-temperature-transformation (TTT) diagram of an unalloyed steel containing 0.45% C. Austenite temperature 880 C. Courtesy of Verlag Stahlessen mbH Dusseldorf
in the transformation process is to heat steel to

its austenitizing temperature. Steel is then cooled
rapidly to avoid the formation of pearlite, which
is a relatively soft transformation product, to
maximize formation of martensite, a relatively
hard transformation product, to achieve the desired as-quenched hardness.
The most common transformation products
that may be formed in quench-hardenable steels
from austenite are in order of formation with decreasing cooling rate: martensite, bainite, pearlite, ferrite, and cementite. The formation of these
products and the proportions of each are dependent on the time and temperature cooling history
of the particular alloy and elemental composition
of that alloy. The transformation products
formed are typically illustrated with the use of
transformation diagrams that show the temperature-time dependence of the microstructure formation process for the alloy being studied. Two
of the most commonly used transformation diagrams are time-temperature-transformation
(TTT) and continuous-cooling-transformation
(CCT) diagrams.
TTT diagrams, also called isothermal transformation (IT) diagrams are developed by heating small samples of steel to the temperature
where austenite transformation structure is completely formed, that is, austenitizing temperature,
then rapidly cooling to a temperature intermediate between the austenitizing and the martensite start (Ms) temperature, and then holding for
a fixed period of time until the transformation is
complete, at which point the transformation
products are determined. This is done repeatedly
until a TTT diagram is constructed such as that
shown for an unalloyed steel (AISI 1045) in Fig.
1. TTT diagrams can only be read along the isotherms.
CCT Diagrams. Alternatively, samples of a
given steel may be continuously cooled at different specified rates, and the proportion of transformation products formed after cooling to various temperatures intermediate between the
austenitizing temperature and the Ms are determined to construct a CCT diagram such as the
one shown for an unalloyed carbon steel (AISI
1045) in Fig. 2. CCT curves provide data on the
temperatures for each phase transformation, the
amount of transformation product obtained for a
Hardening by Reheating and Quenching / 249

given cooling rate with time, and the cooling rate
necessary to obtain martensite. The critical cooling rate is dictated by the time required to avoid
formation of pearlite for the particular steel being quenched. As a general rule, a quenchant
must produce a cooling rate equivalent to, or faster than, that indicated by the nose of the pearlite transformation curve to maximize martensite
transformation product formation. CCT diagrams can only be read along the curves of different cooling rates.
Caution: Although it is becoming increasingly common to see cooling curves (temperature-time profiles) for different cooling media
(quenchants) such as oil, water, air, and others,
superimposed on either TTT or CCT diagrams,
this is not a rigorously correct practice and various errors are introduced into such analysis due
to the inherently different kinetics of cooling
used to obtain the TTT or CCT diagrams versus
the quenchants being represented. A continuous
cooling curve can only be superimposed on a
CCT, but not on a TTT diagram.
Metallurgical Crystal Structure. When steel
is slowly cooled, it undergoes a crystal structure
(size) change as it transforms from a less densely
packed austenite (face-centered cubic, or fcc) to
a more densely packed body-centered cubic
(bcc) structure of ferrite. At faster cooling rates,
the formation of ferrite is suppressed and martensite, which is an even less densely packed
body-centered tetragonal (bct) structure than
austenite, is formed. Illustrations of these crystal
structures are provided in Fig. 3. This results in
a volumetric expansion at the Ms temperature as
shown in Fig. 4.
Figure 5 shows that the crystal lattice of austenite expands with increasing carbon content
(Ref 2). Typically, when a carbide-ferrite mixture is converted to martensite, the resulting expansion due to increasing carbon content is approximately 0.002 in./in. at 0.25% C and 0.007
in./in. at 1.2% C (Ref 2). The fractional increase
in size when austenite is converted to martensite
is approximately 0.014 in./in. for eutectoid compositions. This illustrates the effect of carbon
structure and steel transformation on residual
stresses and distortion, which leads to dimensional changes.
Estimation of Volumetric Change. In the
previous discussion, it is shown that there are
various microstructures possible upon quenching of steel and that the potential microstructural
transformations that are possible for a given steel
are shown by their CCT or TTT diagrams. Furthermore, dimensional changes are possible, depending on the carbon content and microstructural transformation product formed. Table 1
summarizes the atomic volumes of different microstructural components as a function of carbon
content (Ref 3). Table 2 provides an estimate of
volumetric changes as a function of carbon content for different metallurgical transformations
(Ref 4).
Thelning reported that volumetric expansion
that occurs as a result of quenching could be estimated from (Ref 5):
and C equals the weight percentage of carbon

dissolved in austenite and martensite.
Berns reported that if the value of (DV/V) is
known or can be computed, then internal stresses
that are developed in a part due to temperature
differences (DT) arising from either one-dimensional heating or cooling could be estimated
from (Ref 6):
r E e E 13 (DV/V) E DT
where E (modulus of elasticity) 2 105 N/

mm2 and (coefficient of thermal expansion)
1.2 105. Relative volume changes due to
phase transformation are shown in Fig. 6 (Ref
6).
Figure 7 shows that stresses, such as an increase in hydrostatic pressure, accelerate phase
transformations (Ref 7). This will occur whether
the stresses are tensile or compressive and will
result with accelerated austenite decomposition
and increasing Ms temperature. The strain of this
process is often estimated as being equal to the
volumetric expansion divided by 3 (Ref 8).
Transformation plasticity is a process whereby a
stress will affect linear strain. This is shown in
DV/V 100 (100 VC Va) 1.68C

Va (4.64 2.21C)
where (DV/V) 100 equals the percentage

change in volume, VC equals the volume percentage of undissolved cementite, (100 VC
Va) equals the volume percentage of martensite,
Va equals the volume percentage of austenite,
(a)
800
A3
700
A1
Temperature, C
600
Pearlite start
Pearlite end
500
Bainite start
400
(b)
Bainite end
Martensite
300
200
100
0
0.7
102
10 60
103
104
105
Seconds
1
4
Time
Fig. 2
8
15
Minutes
30
60
1
4 68
Hours
16 24
Continuous-cooling-transformation diagram of an unalloyed steel containing 0.45% C. Austenite temperature

880 C. Courtesy of Verlag Stahlessen mbH Dusseldorf
(c)
Fig. 3
Crystal structures. (a) Austenite, fcc. (b) Ferrite,

bcc. (c) Martensite, bct

Fig. 8 where the effect of applied stress on martensitic transformation strain is shown (Ref 8).
Generally, the Ms temperature is increased by
tensile stress and decreased by hydrostatic pressure. Figure 9 shows that stress exhibits very
Temperature, F
1.4
200
400 600 800 1000 1200 1400 1600
1.2
Austenite
1.0
Linear expansion, %
High-temperature
transformation
Table 1 Atomic volume of different microstructure constituents of ferrous alloys
0.8
Very slow
cooling
3
Apparent atomic volume, A
Phase
Ferrite
Cementite
Ferrite carbides
Pearlite
Austenite
Martensite
0.6
Rapid
quenching
Pearlite
0.4
Martensite
forms
large effects on the start and stop times for pearlitic transformation (Ref 7).
Cooling and Steel Metallurgical Transformation. The following paragraphs discuss cooling of steel both with and without metallurgical
transformation.
Cooling without Transformation (Ref 8, 9). If
steel is cooled sufficiently fast, cooling will not
be accompanied by microstructural transformation changes. Under these conditions, the surface
of the component cooled much more quickly
than the core at first, as illustrated in Fig. 10 (Ref
10). At this point, the specific volume in the core
is greater than that of the surface, and the reduction in volume at the surface (due to lower temperature) is resisted by the greater volume in the
core, resulting in the surface being in tension and
the core in compression.
11.789
12.769
11.786 0.163 C(a)
11.916
11.401 0.329 C(a)
11.789 0.370 C(a)
Austenite
(a) C, % C
0.2
Ms
Table 2 Volumetric changes with different

steel transformations
0
0
100 200 300 400 500 600 700 800 900

Temperature, C
Fig. 4
Steel expansion and contraction upon heating

and cooling
Steel transformation
Volumetric change (a)
Pearlite austenite
Austenite martensite
Austenite acicular lower bainite
Austenite feathered upper bainite
4.64
4.64
4.64
4.64
2.21
0.53
1.43
2.21
C
C
C
C
(a) C, % C
1.080
Austenite, A
3.61
3.59
a
0.132
1.060
3.57
c/a
3.55
3.04
c
2.96
1.020
2.88
0
0
a
0.4
0.8
1.2
1.000
1.6
Specific volume, cm3/g
Martensite, A
1.040
c/a
0.130
Martensite
Martensite
Retained
austenite
Tempered
martensite
0.128
Ferrite + carbide
0.126
Austenite
0.124
0.122
Carbon, wt%
Carbon content versus lattice parameters of (retained) austenite and martensite at room temperature. a at the top of the graph is the lattice parameter
of fcc austenite. a and c in the lower half of the graph are
the two lattice parameters of tetragonal martensite. The ratio of c/a for martensite as a function of carbon content is
also given.
Fig. 5
0.120
0
0.4
0.8
1.2
1.6
2.0
Carbon content, %
Fig. 6
200 C
Specific volume (DV/V) of carbon steels relative

to room temperature. Tempered martensite,
At the point of maximum temperature difference between the surface and the core (point T),
the core cools (shrinks) more quickly than the
surface, leading to an elastic dimensional reduction of the surface until a point of stress conversion is obtained, at which point the surface is in
compression relative to the core. After cooling
has been completed, the residual stress distribution, between the surface and the core shown
at the bottom right in Fig. 10 (Ref 10), will be
obtained. If the surface stresses exceed the hot
yield strength of the material, it will plastically
deform, resulting in thermally induced dimensional changes.
Cooling with Transformation (Ref 8, 9). When
steels that may undergo transformational
changes are quenched, the possibility of the formation of both thermal and transformational
stresses must be considered. Figure 11 illustrates
three different examples of this process (Ref 9).
Example 1, illustrated in Fig. 11(a), occurs
when phase transformation of both the surface
and the core occurs before the thermal stresses
change sign. Above the Ms transformation temperature, the stresses that are formed are thermal.
Upon further cooling, the stresses in the core exceed the yield strength and plastic deformation
(elongation) occurs. Subsequent martensitic
transformation at the core provides a substantial
stress component due to volumetric increases
from martensitic phase transformation. This
causes the core to be in compression and the surface to be in tension simultaneously.
Example 2 (Fig. 11b) illustrates the case that
begins after the thermal stresses change sign.
The transformation-induced volume increase of
the surface layer adds to the compressive stresses
at the surface. Since the stresses are balanced,
there is a corresponding increase in the tensile
stresses in the core.
Example 3 is a case where, although the transformation of the core starts later, it finishes before the surface (Fig. 11c). During cooling, the
sign of the stresses changes three times. There
are important consequences whether the core
transforms before or after the stress reversal because thermal stresses may be counteracted, and
tensile surface residual stresses may result (Ref
8).
This is illustrated in Fig. 12(a) where it is
shown that if the steel transformation occurs before the thermal stress maximum, the ferrite/
pearlite structure of a cylindrical test specimen
will be distorted into a barrel-shape (Ref 6). If
the transformation occurs after the thermal stress
maximum, the austenite is pressed into a barrel
shape followed by a volumetric increase due to
martensitic transformation as shown in Fig.
12(b). This will result in high tensile residual
stresses on the surface. If the steel transformation
occurs simultaneously with the maximum thermal stresses, as shown in Fig. 12(c), transformation in the core will occur prior to surface
transformation, and a barrel shape will appear
with high compressive stresses on the surface. If
the surface is transformed prior to the core, the
transformational stresses will decrease, or pos-
Tempering
The tempering process involves heating hardened steel to some temperature below the eutectoid temperature in order to decrease hardness
and increase toughness. In general, tempering is
divided into four stages, which are summarized
in Table 4 (Ref 10). These include:
Tempering of martensite structure
Transformation of retained austenite to mar-
tensite
Tempering of the decomposition products of
martensite and at temperatures 900 F (480
C)
Decomposition of retained austenite to martensite (Ref 11)
Figure 13 illustrates the effect of microstructural
variation during tempering on the volume
changes occurring during the tempering of hardened steel (Ref 10).

Figure 14 (Ref 11) illustrates the effect of dimensional variation and retained austenite content of a bearing steel (100 Cr 6) as a function
of tempering temperature. In addition to those
effects discussed previously, tempering may also
lead to dimensional variation due to relaxation
of residual stresses and plastic deformation,
which is due to the temperature dependence of
yield strength (Ref 10).
Metallurgical Sources of Stress and

Distortion during Reheating and
Quenching
occur in the plastic zone and are permanent.

Upon heating, the stresses are gradually relieved
by changes in the shape of the part due to plastic
flow. This is a continuous process, and as the
temperature of the part is increased, the material
yield stress decreases as shown in Fig. 16 (Ref
13). It is a function not only of temperature but
also of time, since the material will creep under
lower applied stresses. It is apparent that the
stresses can never be reduced to zero, because
the material will always possess some level of
yield strength below which residual stresses cannot be reduced.
Material Movement. When parts are heated
during heat treatment, a thermal gradient exists
across the cross section of the component. If a
Basic Distortion Mechanism. Shape and

volume changes during heating and cooling can
be attributed to three fundamental causes (Ref
12):
3
285 MPa
Residual stresses that cause shape change
when they exceed material yield strength.

This will occur on heating when the strength
properties decline.
Stresses caused by differential expansion due
to thermal gradients. These stresses will increase with the thermal gradient and will
cause plastic deformation as the yield
strength is exceeded.
Volume changes due to transformational
phase change. These volume changes will be
contained as residual stress systems until the
yield strength is exceeded.
Strain (L/Lo), %
sibly even reverse the thermal stresses. If this

occurs, a spool shape will be formed.
These data show that the position of cooling
curves for both the surface and the core for the
quenching process must be considered with the
appropriate TTT curve for the steel of interest
and that there are numerous mixtures of thermal
and transformational stresses possible. Table 3
provides a summary of these processes (Ref 6).
In some steels, those with higher carbon content and alloy steels, the martensite finish (Mf)
temperature is below 32 F (0 C), which means
that it is likely that at the conclusion of heat treating there will be as much as 5 to 15% of austenite
remaining (Ref 11). The amount of retained austenite will exhibit significant effects on the magnitude of compressive stresses formed and ultimately on dimensional changes to be expected.
Some of the factors affecting retained austenite
formation include: chemical composition (which
dictates the Ms temperature), quenching temperature, quenching cooling rates, austenitizing
temperature, grain size, and tempering.
Relief of Residual Stress. If a part has lockedin residual stresses, these stresses can be relieved
by heating the part until the locked-in stresses
exceed the strength of the material. A typical
stress-strain curve obtained from a tension test
is shown in Fig. 15 (Ref 12). Initial changes in
shape are elastic, but under increased stress they
Fig. 8
2
209 MPa
107 MPa
1
18 MPa
100
200
300
Temperature (), C
Dilatometer curves for a 0.6% C
(60NCD11) for different applied stresses
steel
200
1000
900
Tim e ( t ) , s
800
1 atm
Temperature, C
700
F.P. =(12
12kbar)
kbar
F.P. =(23
23kbar)
kbar
600
500
B =(11atm)
atm
tF
400
300
200
100
0
1
B =(12
12kbar)
kbar
Mss =(11atm)
atm
B =(23
23kbar)
kbar
12kbar)
kbar
Mss =(12
tD
0
Mss =(23
23kbar)
kbar
20
40
60
80
Tensile stress (), MPa

10
102
103
104
105
Time, s
Fig. 7
100
Effect of hydrostatic pressure on the transformation kinetics of 50CV4 steel. B, bainite; F.P., ferrite-pearlite
transformation; Ms, martensitic start temperature. Source: Ref 7
Effect of tensile stress on pearlite transformation

starting and ending times. Isothermal transformation at 673 C, eutectoid steel. The tD and tF times are
transformation starting and ending times, respectively.
Fig. 9

component design. Poor component design promotes distortion and cracking by accentuating
nonuniform and nonsymmetrical heat transfer
during quenching. Component design characteristics that are common to distortion and cracking
problems include (Ref 19, 20):
There are various factors that can affect distortion and growth of steel heat treating. These
include: component design, steel grade and condition, machining, component support and loading, surface condition, heating and atmosphere
control, retained austenite, and the quenching
process (Ref 18).
Parts that are long (L) with thin (d) cross sec-
tions. Long and thin parts are defined as:

greater than L 5d for water quenching, L
8d for oil quenching, and L 10d for
austempering where L is the length of the
Component Design
One of the overwhelming causes of steel
cracking and unacceptable distortion control is
1000
Center
Temperature, C
section is heated so that a portion of the component becomes hotter than the surrounding material, the hotter material expands and occupies
a greater volume than the adjacent material and
will thus be exposed to applied stresses that will
cause a shape change when the material strength
is exceeded. These movements can be related to
heating rate and section thickness of the component.
Volume Changes during Phase Transformation. When a steel part is heated, it transforms to austenite with an accompanying reduction in volume as shown in Fig. 4 (Ref 14).
When it is quenched, the structure transforms
from austenite to martensite and its volume increases. If these volume changes cause stresses
that are constrained within the strength of material, a residual stress system is created. If the
stresses cannot be contained, then material
movement will occur, which will cause cracking
under extreme conditions. The expansion is related to the composition of the steel. Fig. 17
shows the relative volume increase of two steels
as a function of austenitizing temperature and
specimen dimensions.
While all of these phenomena are well-known
physical changes, the situation is made more
complex when all three events occur simultaneously. In addition, other events such as heating
rate, quenching, and inconsistent material composition further complicate the process.
Surface
0
500
10
100
1000
Residual stress
Time, s
Tension
Tension
B
Center
B
Surface
0
Surface
Effect of Materials and Quench

Process Design on Distortion
Compression
Fig. 10
Development of thermal stress on cooling in steel specimen. T, time instant of maximum temperature difference; O, time instant of stress reversal; curve A, stress variation at the surface under elastic condition. B and
C are actual thermal stress variations at the surface and the core under elastic-plastic conditions.
Temperature
TA
Core
TA
Core
Surface
Surface
Surface
Ms
Core
TA
Ms
Ms
RT
+
Stress, z
Quenchant selection and quenching conditions are critically important parameters in

quench system design. For example, one study
compared the distortion obtained with quenching
of a 0.4% medium-carbon plain steel bar of 200
mm diameter by 500 mm long in water or oil
from 680 C (Ref 15, 16). The results, shown in
Fig. 18(a) and (c), show that essentially equivalent variation in diameter and length with both
cooling processes were obtained that were due
to thermal strains within the steel. Interestingly,
the well-known diameter variations at the end of
the bar, known as the end effect, were observed; this is attributed to heat extraction from
both the sides and ends of the bar.
When the bars of the same type of steel and
the same dimensions were heated to 850 C to
austenitize the steel and then quenched in water
or oil, the results shown in Fig. 18(b) and (d),
respectively, were obtained (Ref 15, 16). Considerably greater dimensional variation and
lengthening of the bar (for the oil quench) were
obtained due to both thermal and transformational strains with the steel.
Thuvander and Melander modeled the dimensional changes of a 70 mm steel (0.15% C, 1%
Mn, 0.75% Cr, 0.85% Ni) cube after austenitizing and then quenching in water and oil (Ref 15,
17). The results of this work (Fig. 19) show that
the edges and faces shrink (becoming concave),
and the effect is greater when quenched in water
than when quenched in oil.
Compression
Surface
Center
tu
tu
(a)
(b)
Surface
tu
Time
Time
Time
(c)
Core
Thermal stress
Total stress
Fig. 11
Comparison of thermal and transformational stresses for three different quenching conditions. See text for
details. tu, time instant of stress reversal

parts and d is the thickness or diameter. Long
and thin parts may be quenched using a support mechanism such as that shown in Fig.
20.
Parts that possess large cross-sectional area
(A) and are thin (t), are defined as: A 50t.
of alignment. A general rule for solving such

quench distortion problems is that the short side
is the hot side, which means that the inside of
the bowed metal was quenched more slowly than
the opposite side (Ref 20).
Parts that exceed these dimensions often must be

straightened or press quenched to maintain dimensional stability (Ref 20). If possible, materials with sufficient hardenability should be oil
or salt quenched. A schematic of a press quenching system is shown in Fig. 21.
Design symmetry is also an important variable
to minimize distortion. For example, the unsymmetrical gear design shown in Fig. 22(a) typically may undergo distortion as shown in Fig.
22(b) (Ref 19). (The load on a gear tooth increases as the 4.3 power of the taper, Ref 19.)
Solution to the gear design problem shown in
Fig. 23 is to provide greater symmetry as shown
in Fig. 23. If this is not possible, press quenching
or tooth-by-tooth induction hardening may be
the only solutions (Ref 19, 20).
Another common design problem is with parts
having holes, deep keyways, and grooves. One
illustration of this problem is hardening of a
shaft over a lubrication cross hole as shown in
Fig. 24 (Ref 19). Preferred alternative designs
are also shown in Fig. 24. If a radial cross hole
is mandatory, the use of a carburized steel with
oil quenching would be preferred.
Distortion that was encountered when quenching a notched part, such as a shaft with a milled
slot, is shown in Fig. 25 (Ref 20). In this case,
nonuniform heat transfer results. The metal
within the notch is affected by the shrinkage of
the metal around it due to slower cooling within
the slot caused by vaporization of the quenchant.
Therefore, upon cooling the metal on the side
with the shaft is too short, pulling the shaft out
Steel Grade and Condition

Although quench cracking is most often
caused by nonuniform heating and cooling, material problems may be encountered. Some typical material problems include (Ref 19):
Table 3
Fig. 12)
The compositional tolerances should be
checked to ensure that the alloy is within

specification.
However, some alloys are particularly problematic. For example, some steel grades must
be water quenched when the alloy composition is on the low side of the specification
limit. Conversely, if the alloy composition is
on the high side, cracking is more common.
Steel grades that exhibit this problem include:
1040, 1045, 1536, 1541, 1137, 1141, and
1144. As a rule, steels with carbon contents
Size change and residual stress caused by heat treatment of prismatic parts (see
(DT)2
(DV/V)3
Dimensional
dependency
(De)
Change of microstructure
Small
e1 e11 e111
Thermal stress
Big
e1 e11 e111
0
Thermal and
transformation stress
Transformation R K
after thermal stress
maximum
Transformation R
before K after thermal
stress maximum
Transformation K
before R after thermal
stress maximum
Transformation R K
before thermal stress
maximum
(rE)
Small
Big
0
0
e1 e11 e111
e1 e11 e111
00
0
0
Big
0
e1 e11 e111
Big
e1 e11 e111
Small
Big
0
0
e1 e11 e111
e1 e11 e111
0
0
00
Example
Tempering, precip.
Hardening,
isotropic material
Quenching of
austenitic steels
In air
Full hardness
penetration
In water
Medium hardness
penetration
Shallow hardness
penetration (shell
hardening)
Normalizing,
quenching
without hardening
R surface; K core; DT (TK TR); DV/V relative volume change; Convexity De ek em; em and ek dimensional change in middle
of the plane, respectively, at the edge; rE residual stress at the surface (remaining stress)
Table 4 Metallurgical reactions occurring at various temperature ranges and related physical changes of steel during tempering
Transformation for 2
Temperature
range
K
C
F
Metallurgical reaction
0200
200300
230350
350700
32392
392572
446662
6621292
Precipitation of e carbide. Loss of tetragonality

Decomposition of retained austenite
e carbides decompose to cementite
Precipitation of alloy carbides. Grain coarsening
1
2
3
4
Expansion/contraction
Contraction
Expansion
Contraction
Expansion
Tmax
B
M
log t, s
1 Tmax
Barrel
Fig. 12
UR
UK
UR
Tmax
UK
UK
Tmax
UR
UR
UK
Decomposition of
martensite to
ferrite and cementite
Increase in volume
Temperature , C
Stage
Retained austenite
to martensite
Retained austenite
to bainite
Carbide
precipitation
Tmax
0
Spool
Size change due to thermal changes and phase
transformation. K, core; R, surface
100
200
300
400
500
600
700
800
Tempering temperature, C
Fig. 13
Effect of microstructural constitutional variation on volume changes during tempering
900

and hardenability greater than that of 1037 are
difficult to water quench (Ref 19).
Some steel grades with high manganese are
prone to microsegregation of manganese and
gross segregation of chromium are prone to
cracking. These include: 1340, 1345, 1536,
1541, 4140, and 4150. If possible, it is often
a good choice to replace the 4100 series with
the 8600 series of steels (Ref 21).
Dirty steels, those containing greater than
0.05% S, are more prone to cracking such as
1141 and 1144. The reasons for this include:
greater alloy segregation in dirty steels lead
to alloy-rich and alloy-lean regions, there are
0.4
Dimensional alteration
0
0.4
0.8
1.2
Austenitizing temperature
800 C
850 C
910 C
1.6
typically more surface seams that act as stress

risers with dirty steels, and steels with higher
sulfur levels are often manufactured to coarse
grain practice for improved machinability,
which also imparts greater brittleness and
propensity for cracking.
Decarburization of up to 0.0025 in. per 116 in.
diameter may be present.
It is well known that cracking propensity increases with carbon content. Therefore, carbon
content of steel is one of the determining factors
for quenchant selection. Table 5 summarizes
some steel mean carbon content concentration
limits for water, brine, or caustic quenching (Ref
21).
It is well known that regions containing high
concentrations of coarse carbide microstructure
as a result of improper forging may become the
initiation point for subsequent quench cracking,
particularly with parts of complex shape (Ref
22). It is important to provide a sufficient forging
reduction ratio to allow the carbide formation to
become fine and uniform (Ref 23).
Since part manufacture, such as gear production, often requires machining, the condition of
the steel that is going to be machined is critically
important. Some workers have recommended
that normalized and subcritical annealed steel is
the ideal condition (Ref 18). Subcritical annealing is performed to relieve stresses incurred during normalization without softening or homogenizing the steel. The subcritical annealing
process reduces the carbon content and alloy carbide content in the austenite allowing the production of more lath martensite in the microstructure, which provides higher fracture
toughness and higher impact toughness (Ref 22).
Machining
Material removal during machining can result
in high residual stress levels and ultimately unacceptable distortion (Ref 18). When excessive
machining stresses are imparted, the process
may require modification to include a rough machining and stress relieving followed by fine machining. This subject is covered in the article in
chapter 5 and is not discussed in detail here.
Component Support and Loading

Toshioka (Ref 16) examined the bending behavior of a piston rod that is heated vertically
while suspended from the top of a furnace. After
austenitizing, the piston rods were then oil
quenched while still hanging vertically and the
16
Temperature, F
200
800
8
4
0
RT
180
220
240
Tempering temperature, C
Dimensional variation and retained austenite
content of 100Cr6 steel as a function of tempering temperature
Fig. 14
400
600
800
1000
1200
50
Low-alloy steel
600
40
Creep-resistant
austenitic steel
400
30
20
200
10
Carbon-managanese
Carbon-manganese steel
steel
0
100
200
300
400
500
600
0.2% proof stress, tsi
12
0.2% proof stress, MPa
Retained austenite
20
700
Temperature, C
Fig. 16
P
Stress, =
Ao
Elastic
Variation of yield strength with temperature for three generic classes of steel
Plastic
90MnV8
Ultimate
tensile
strength
760 C
800 C
145CrV6
870 C
830 C
Yield
stress
Uniform
elongation
Necking
Volume increase, %
0.6
Fracture
0.4
0.2
0
Offset
Fig. 15
50 50 10
L Lo
Strain, =
Lo
Various features of a typical stress-strain curve

obtained from a tension test
50 50 20
50 50 50
50 50 250
Specimen dimension, mm
Fig. 17
Volume increase of steels 90MnV8 and 15CrV6 as a function of austenitizing temperature and specimen
dimensions

amount of distortion obtained, shown in Fig. 26,
is widely variable from nearly zero to as much
as 20 mm (Ref 16). Many parts may sag and
creep under their own weight when heat treated;
this is an important cause of distortion. An example of a type of component that is susceptible
to such distortion is a ring gear, the dimension
limits by which ring gears are classified are provided in Fig. 27 (Ref 18). (A general dimensional classification of various distortion-sensitive shapes is provided in Fig. 28, Ref 18.)
Proper support when heating is required to minimize out-of-flatness and ovality problems,
which may result in long grinding times, excessive stock removal, high scrap losses, and loss
of case depth (Ref 18). To achieve adequate distortion control, custom supports or press quenching may be required.
Pinion shafts, as defined in Fig. 28, are susceptible to banding along their length if they are
improperly loaded into the furnace as shown in
Fig. 29 (Ref 18). When this occurs, the pinion
shafts must then be straightened, which will add
to production cost.
Surface Condition
Quench cracking may be due to various steelrelated problems that are only observable after
the quench, but the root cause is not the quenching process itself. Many of these have been reviewed earlier and include: prior steel structure,
stress risers from prior machining, laps and
seams, alloy inclusion defects, grinding cracks,

chemical segregation (bonding), and alloy depletion (Ref 24). In this section, three problems
related to surface conditions that may contribute
to poor distortion control and cracking are discussed: tight scale formation, decarburization,
and the formation of surface seams or nonmetallic stringers.
Tight scale problems are encountered with
forgings hardened from direct-fired gas furnaces
with high-pressure burners (Ref 20, 21). The effect of tight scale on the quenching properties of
two steels, 1095 carbon steel and 188 stainless
steel, is shown in Fig. 30 (Ref 14). These cooling
curves were obtained by still quenching into fast
oil. A scale of not more than 0.08 mm (0.003
in.) increased the cooling rate of 1095 steel compared with the rate obtained on a specimen without scale. However, a heavy scale (0.13 mm, or
0.005 in., deep) retarded the cooling rate. A very
light scale (0.013 mm, or 0.0005 in., deep) also
increased the cooling rate of the 188 steel over
that obtained with the specimen without scale.
In practice, the formation of tight scale will
vary in depth over the surface of the part resulting in thermal gradients due to differences in
cooling rates. This problem may yield soft spots
and uncontrolled distortion and is particularly a
problem with nickel-containing steels. Surface
oxide formation can be minimized by the use of
an appropriately protective atmosphere.
The second surface-related condition is decarburization, which may lead to increased distortion or cracking (Ref 23). At a given depth
within the decarburized layer, the part does not

harden as completely as it would at the same
point below the surface if there were no decarburization. This leads to nonuniform hardness,
which may contribute to increased distortion and
cracking because (Ref 20):
The decarburized surface transforms at a
higher temperature than the core (the Ms temperature decreases with carbon content). This
leads to high residual tensile stresses at the
decarburized surface or a condition of unbalanced stresses and distortion.
Since the surface is decarburized, it will exhibit lower hardenability than the core. This
will cause the upper transformation products
to form early, nucleating additional undesirable products in the core. The decarburized
side will be softer than the side that did not
Oil
quench
A
0.1 mm
A
+0.1
Diameter change, mm
Diameter change, mm
Water
quench
0
Change in length: 0.17 mm
0.1
250
500
+0.1
Distance from end of bar

first into quench, mm
250
500

Change across
midsection (A-A)
Dimensional changes in a 70 mm steel (0.15%

C, 1% Mn, 0.75% Cr, 0.85% Ni) bar after austenitizing and then quenching in water or oil
Fig. 19
(b)
(a)
680C/water quench.
No transformation.
Thermal strain only.
850C/quench.
Thermal and
transformation strain.
+0.1
Diameter change, mm
Diameter change, mm
Outer-face
change
WQ change in length: +0.17 mm

0.1
0.3 mm
0
Change in length: 0.15 mm
0.1
250
500

(c)
+0.1
OQ change in length: +0.8 mm
0
WQ change in length: 2.45 mm
0.1
250
500

(d)
Dimensional variation of a medium-carbon (0.4%) steel bar (200 mm diam by 500 mm) after the indicated
heat treatments. These bars were quenched vertically with one end down (marked 0 in the figure). (a) and (c)
show no transformation, thermal strain only after water quenching from 680 C. (b) and (d) show thermal and transformation strains after quenching from 850 C. OQ, oil quench; WQ, water quench
Fig. 18
Fig. 20
Die quench system. Courtesy of Gleason Tooling Products Group

undergo decarburizing, which is harder. The
greater amount of martensite leads to distortion.
The solution to this problem is to restore carbon
into the furnace atmosphere or machine off the
decarburized layer.
The third surface-related condition that may
lead to cracking or material weakening is the formation of surface seams or nonmetallic inclusions, which may occur in hot-rolled or coldfinished material. The presence of these defects
prevents the hot steel from welding to itself during forging, for example, creating a stress riser.
To prevent this problem with hot-rolled bars,
stock should be removed before heat treatment.
Recommendations made earlier by Kern are provided in Table 6 (Ref 21). Although not a published standard, Kern has reported that a seam or
nonmetallic depth of 0.001 in./0.13 in. diameter

maximum is usually acceptable for cold-finished
bars (Ref 21). If the seam depth is excessive, it
is recommended that the bars be magniflux inspected prior to heat treatment.
Before heat treatment
Heating and Atmosphere Control

An important source of steel distortion and
cracking is nonuniform heating and not using the
appropriate protective atmosphere. For example,
if steel is heated in a direct-gas-fired furnace with
high moisture content the load being heated may
adsorb hydrogen leading to hydrogen embrittlement and subsequent cracking that would not normally occur with a dry atmosphere (Ref 4, 19).
Localized overheating is particularly a potential problem for inductively heated parts (Ref 4,
After heat treatment
Fig. 25
Distortion often encountered when quenching

a notch
Table 5 Suggested carbon content limits

for water, brine, and caustic quenching
Hardening method/shapes
Fig. 21
Schematic of a press quench system. Courtesy of Gleason Tooling Products Group
Carbon, % max
Furnace hardening
General usage
Simple shapes
Very simple shapes, for example, bars
0.30
0.35
0.40
Induction hardening
Simple shapes
Complex shapes
0.50
0.33
100 diam
80 diam
140 diam
385
1775
2260
5
Shallow round
bottom groove
The problem: gear

tapers or warps
Fig. 22
Bending, mm
Gear before heat treating
4
3
2
1
0
Schematic of a gear that is difficult to harden

without the distortion shown
Distortion of JIS SCM 440 (0.4 C, 1.05 Cr, 0.22

Mo) steel piston rods after oil quenching while
vertically suspended from 850 C and tempering at 600 C
Fig. 26
Harden
this
length
WT
ID
A satisfactory design if
hub must be offset
L
Usual design
(poor)
Design solutions to the distortion problem

shown in Fig. 22
OD
Design for hardening

over holes
Solution: the ideal design
Fig. 23
Better
Design solutions to the quench-cracking problem often encountered in shaft hardening over
a cross hole
Fig. 24
Dimensions of a ring gear shape. Shape limitation: length/wall thickness 1.5; ID/OD
0.4. Minimum wall thickness (WT) is defined by WT 2.25
module 0.4 5mod L OD3
Fig. 27

25). Subsequent quenching of the part leads to
quench cracks at sharp corners and areas with
sudden changes in cross-sectional area (stress
risers). Cracking is caused by increases of residual stresses at the stress risers during quenching.
The solution to the problem is to increase the
heating speed by increasing the power density of
the inductor. The temperature difference across
the heated zone is decreased by continuous heating or scanning of several pistons together on a
single bar (Ref 25).
For heat treating problems related to furnace
design and operation, it is usually suggested that
(Ref 19):
The vestibules of atmosphere-hardening fur-
naces should be loaded and unloaded with

purging. Load transfer for belt and shaker
hearth furnaces should only occur with thorough purging to minimize atmosphere contamination.
Hardening furnaces typically contain excessive loads prior to quenching. If the steel at
quenching temperature is greater than 20% of
the distance from discharge to charge door, it
is too much. Either the production rate can be
increased or some of the burners can be
turned off.
Retained Austenite
Dimensional changes may occur slowly or
quickly and are due to the volume composition
of the transformation products formed upon
quenching. One of the most importantwith respect to residual stress variation, distortion, and
crackingis the formation and transformation
of retained austenite. For example, the data in
Table 7 illustrate the slow conversion of retained
austenite to martensite, which was still occurring
days after the original quenching process for the
two steels shown (Ref 15, 16). This is particularly a problem when dimensional control and
stability are the primary goals of heat treatment.
Therefore, microstructure determination is an essential component of any distortion-control process.
Quenching Process
ommendations for quench media selection for

use with various steel alloys is provided by standards such as AMS 2770. Some additional general comments regarding quenchant selection include (Ref 4, 20):
Most machined parts made from alloy steels
Other than component design, the quenching

process itself is one of the most frequently encountered problems in heat treating. When designing a quenching process, it is important to
consider quenchant selection and quench severity and uniformity.
Quenchant Selection and Severity. A more
detailed discussion on quenchant selection and
the ranges of quench severity possible with these
media are provided in a subsequent section.
However, here the general issue of quench severity is addressed. Quench severity is defined
as: ability of a quenching medium to extract
heat from a hot steel workpiece, expressed in
terms of the Grossmann number (H) (Ref 26).
A typical range of Grossmann H values for commonly used quench media are provided in Table
8, and Fig. 31 provides a correlation between the
H value and the ability to harden steel as indicated by the Jominy distance (J distance) (Ref
20). Although Table 8 is useful to obtain a relative measure of the quench severity offered by
different quench media, it is difficult to apply in
practice because the actual flow rates for moderate, good, strong, and violent agitation
is unknown.
Alternatively, the measurement of actual cooling rates or heat fluxes provided by a specific
quenching medium does provide a quantitative
meaning to the quench severity provided. Some
illustrative values are provided in Table 9 (Ref
27).
Typically, the greater the quench severity, the
greater the propensity of a given quenching medium to cause increased distortion or cracking.
This usually is the result of increased thermal
stress not transformational stress. Specific rec-
are oil quenched to minimize distortion.
Most small parts, or finish-ground larger parts
are free quenched. Larger gears, typically

those larger than 205 mm (8 in.) are fixture
(die) quenched to control distortion. Smaller
gears and parts such as bushings are typically
plug quenched on a splined plug typically
constructed from carburized 8620 steel.
Although a reduction of quench severity leads
to reduced distortion, it may also be accompanied by undesirable microstructures such as
the formation of upper bainite (quenched
pearlite) with carburized parts.
Quench speed may be reduced by quenching
in hot (150205 C, or 300400 F) oil.
When hot oil quenching is used for carburized
steels, lower bainite, which exhibits properties similar to martensite, is formed.
Excellent distortion is typically obtained with
austempering, quenching into a medium just
above the Ms temperature. The formation of
retained austenite is a significant problem
with austempering processes. Retained austenite is most pronounced where manganese
and nickel are major components. The best
steels for austempering are plain carbon,
chromium, and molybdenum alloy steels (Ref
20).
Aqueous polymer quenchants may often be
used to replace quench oils, but quench severity is still of primary importance.
Gas or air quenching will provide the least
distortion and may be used if the steel has
sufficient hardenability to provide the desired
properties.
1.00
WT
Thin-walled tube
(annulus)
0.80
Boundary A
Ratio ID/OD
ID
Example
Outside
diameter: 200 mm
Length: 1500 mm
Typical distortions
and growths
OD
0.60
Out-of-straightness:
1000-2000 m
Ring
OD shrinkage at
center: 300-500 m
0.40
Tube
Wheel
0.20
Fig. 28
Increase in length:
1000-1500 m
Pinion
Classification of shapes
3
4
5
6
Ratio length/wall thickness, L/WT
Retort
wall
Fig. 29
Center
pole
Retort
wall
Typical pinion shaft distortion due to furnace

loading

Low-hardenability steels are quenched in
brine or vigorously agitated oil. However,

even with a severe quench, undesirable microstructures such as ferrite, pearlite, or bainite can form.
Quenchant Uniformity. Quench nonuniformity is perhaps the greatest contributor to
quench cracking. Quench nonuniformity can
arise from nonuniform flow fields around the
part surface during the quench or nonuniform
wetting of the surface (Ref 1, 20, 2830). Both
lead to nonuniform heat transfer during quenching. Nonuniform quenching creates large thermal gradients between the core and the surface
of the part. These two contributing factors, agitation and surface wetting, are discussed here.
Poor agitation design is a major source of
quench nonuniformity. The purpose of the agitation system is not only to take hot fluid away
from the surface and to the heat exchanger, but
it is also to provide uniform heat removal over
the entire cooling surface of all of the parts
throughout the load being quenched. The batch
quench system in Fig. 32 illustrates a system
where axial (vertical) quenchant flow occurs
throughout a load of round bars lying horizontally in a basket (Ref 19). In this case, the bottom
surfaces of the bars experience greater agitation
than the top surfaces. Cracks form on the upper
surfaces because of the nonuniform heat loss.
Agitation produces greater heat loss at the bottom, creating a large thermal gradient between
the top and the bottom surfaces.
If a submerged spray manifold is used to facilitate more uniform heat removal, the following design guidelines are recommended:
The total surface of the part should experi-
ence uniform quenchant impingement.
The largest holes possible (2.3 mm, or 0.09
in., minimum) should be used.
The manifold face should be at least 13 mm
(0.5 in.) from the surface of the parts being

quenched.
Repeated removal of hot quenchant and vapor
should be possible.
Excessive distortion was also obtained with an
agitation system shown in Fig. 33 when the
quenchant flow was either in the same direction
relative to the direction of part immersion or in
the opposite direction (Ref 31). The solution to
this problem was to minimize the quenchant
flow to that required for adequate heat transfer
during the quench and to provide agitation by
mechanically moving the part up and down in
the quenchant.
Even though agitation is a critically important
contributor to the performance of industrial
quenching practice, relatively little is known
about the quality and quantity of fluid flow encountered by the parts being quenched. Recently, agitation in various commercial quenching tanks has been studied by computational
fluid dynamics (CFD), and in no case was optimal and uniform flow present without subsequent modification of the tank (Ref 28). Thus,
identifying sources of nonuniform fluid flow
during quenching continues to be an important
tool for optimizing distortion control and minimizing quench cracking.
The second source of nonuniform thermal gradients during quenching is related to interfacial
wetting kinematics, which is of particular interest with vaporizable liquid quenchants including, water, oil, and aqueous polymer solutions.
Most liquid vaporizable quenchants exhibit boiling temperatures between 100 and 300 C at atmospheric pressure. When parts are quenched in
these fluids, surface wetting is usually time dependent, which influences the cooling process
and the achievable hardness.
Leidenfrost described the wetting process in
the 1750s (Ref 32). The Leidenfrost Temperature is defined as the surface temperature where
the vapor film is ruptured and the surface wetted
by the liquid. Literature describes temperature
values for this event for water at atmospheric
pressure between 150 and 300 C (Ref 3336).
1000
1400
1200
600
1000
No scale
800
400
600
400
200
Medium scale
0.08 mm (0.003 in.)
0
(a)
800
Temperature, C
Temperature, C
Heavy scale
0.13 mm (0.005 in.)
18-8 strainless
1B-B
strainless steel
steel
1600
Temperature, F
1095
800
10
20
30
Time, s
1400
1200
No scale
600
1000
800
400
600
400
200
Light scale
0.013 mm (0.0005 in.)
200
40
50
1600
Temperature, F
1000
10
20
30
Time, s
Table 6 Minimum recommended material

removal from hot-rolled steel products to
prevent surface seam and nonmetallic
stringer problems during heat treatment
Minimum material removal
per side (a)
200
40
50
(b)
Centerline cooling curves showing the effect of scale on the cooling curves of steels quenched in fast oil
without agitation. (a) 1095 steel, oil temperature 50 C (125 F). (b) 18-8 stainless steel, oil temperature
25 C (75 F); test specimens were 13 mm diam by 64 mm long (0.5 by 2.5 in.)
Fig. 30
For nonsteady-state cooling, the surface temperature is not equal to the Leidenfrost Temperature when the vapor blanket (or film boiling)
collapses and wetting begins by nucleate boiling
because of the influence of lateral heat conduction (relative to the surface) (Ref 37). This is due
to the simultaneous presence of various heattransfer conditions during vapor blanket coolingor film boiling (FB), nucleate boiling
(NB), and convective heat transfer (CONV) with
significantly varying heat-transfer coefficients
FB (100250 W/m2 K), NB (1020 kW/
m2 K), and CONV (700 W/m2 K). Figure
34 shows the different cooling phases on a metal
surface during immersion cooling (Fig. 34a)
with the so-called wetting front w (separating the
film-boiling phase and the nucleate-boiling
phase) and the change of the heat-transfer coefficients, , along the surface coordinate, z, (mantle line). In most cases, during immersion cooling (Fig. 34a) the wetting front ascends the
cooling surface with a significant velocity, v,
whereas during film cooling (Fig. 34b) the wetting front descends in the fluid direction (Ref 35,
38).
An example of wetting heated cylindrical and
prismatic probes that are submerged in water is
shown in Fig. 35(a) and (b), respectively (Ref
35, 39). Because of the different wetting phases
on the metal surface (and the enormous differences of their values of FB, NB, and CONV), the time-dependent temperature distribution within the metal probes will also be
influenced by the velocity of the wetting front as
well as geometry.
Figure 36 illustrates different types of wetting
behavior under different conditions (Ref 40). By
changing a quenching parameter, for example
the chemical composition of the fluid, the period
of wetting (tW) can be reduced over one magnitude.
To quantitatively define the change of the wetting behavior, for example, to determine the
cooling process or to develop or analyze quenching fluids, the measurement of the electrical conductance between the submerged sample and a
counterelectrode is helpful (Ref 35). During the
film-boiling phase, the hot metal is largely insulated by the vapor film surrounding the metal
and conductance between the metal and the
counterelectrode is low. When the vapor blanket
Condition
Nonresulfurized
Resulfurized
Turned on centers
Centerless turned
or ground
3% of diameter
2.6%
3.8 % of diameter
3.4%
(a) Based on bars purchased to special straightness, i.e., 3.3 mm in 152

cm (0.13 in. in 5 ft) maximum
259 / Hardening by Reheating and Quenching

(film boiling) ruptures on the metal surface, localized wetting begins and conductance increases. The increase in conductance of the wetted metal is proportional to the amount of the
metal surface wetted by the quenchant. When the
metal surface is completely wetted, conductance
is at its highest value.
Figure 37(a) shows a normal electrical conductance, G, increase corresponding to the percentage of the wetted surface. Three other possibilities of wetting are also shown. Figure 37(b)
shows a rapid rewetting process (or explosive
wetting) similar to that shown in Fig. 35. Figure
Quench severity in terms of H value
x
5.0
2.0
H value
1.5
1.0
0.70
0.50
0.35
0.20
6
Fig. 31
12
15
18
J distance (1/16 in.)
21
24
Quench severity in terms of Grossmann (H)

values. J, Jominy distance. Source: Ref 20
37(c) illustrates rapid wetting followed by insulation by bubbles adhering to the metal surface, and Fig. 37(d) illustrates rapid wetting with
repeated new formation of film boiling, a process
that occurs with many aqueous polymer quenchant solutions. In all four diagrams, the temperature, TC, is shown that is measured in the
center of the probes. The TS point characterizes
the temperature and the time when wetting begins. This shows that temperature measurements
in the center of probes provide poor information
about the real quenching process that is insufficient to adequately characterize the hardening
process. This is also illustrated in Fig. 38 (Ref
30).
The influence of the physical properties of the
quenchant and the probe on the wetting process,
and the heat-transfer coefficients (FB, NB, and
CONV) and therefore on the quenching results is
very strong. The wetting kinematics are identified as the starting time, tS, of wetting, the finishing time, tf, of wetting, and the difference between tf and tS is the wetting time tW. The effect
of variation of these parameters on the quenching process is summarized in Table 10.
The influence of a wetting process occurring
over a long time (non-Newtonian wetting) on the
temperature distribution in quenched parts is
shown in Fig. 39 (Ref 23), where the temperatures measured near the surface of a submerged
cylindrical probe at different distances z from its
lower end are shown. The wetting front requires
about 18 s to arrive at a height of 80 mm. If there
is an explosionlike wetting of the probe (Newtonian wetting), the five temperature slopes are
congruent. If the temperature was measured (as

usual) in the center of the probe, the large differences in wetting behavior would not be observed.
The temperature distribution within the probe
during quenching (indicated by isotherms), having a non-Newtonian wetting is shown in Fig.
40(a), where there is a great difference relative
to that of the Newtonian wetting (Fig. 40b). In
the case of Fig. 40(b), the main grad T is radial,
whereas in the case of Fig. 40(a) it is axial. Of
course the hardness distribution in the probes
must be extremely different.
Another major source of nonuniform quenching is foaming and contamination. Contaminants
include sludge, carbon, and other insolubles. It
includes water in oil and oil in water and aqueous
polymer quenchants. Foaming and contamination leads to soft spotting, increased distortion,
and possibly cracking.
Quench Distortion and Cracking. In the
previous discussion, two quenchant-related phenomenon have been discussed: quench cracking
and quench distortion. Although quench cracking can be eliminated, quench distortion cannot.
Instead, the issue is distortion control, not elimination. Both quenching-related distortion control and quench cracking are discussed in this
section.
Quench Distortion. One form of distortion
that may occur upon quenching is defined as
shape distortion, which is dimensional variation
that occurs when the parts are hot and the stress
Quench cracks
Table 7 Dimensional variation in hardened high-carbon steel with time at ambient temperature
Change in length, % 103
Steel type
Tempering
temperature, C
Hardness,
HRC
After
7 days
After
30 days
After
90 days
After
365 days
None
120
205
260
None
120
205
260
66
65
63
61.5
64
65
62
60
9.0
0.2
0.0
0.0
1.0
0.3
0.0
0.0
18.0
0.6
0.2
0.2
4.2
0.5
0.1
0.1
27.0
1.1
0.3
0.3
8.2
0.7
0.1
0.1
40.0
1.9
0.7
0.3
11.0
0.6
0.1
0.1
1.1% C tool steel 790 C quench
1% C/Cr 840 C quench
Table 8 Typical quenching conditions and

Grossmann H values
Quenching medium
Poor (slow) oil quenchno agitation

Good oil quenchmoderate agitation
Very good oil quenchgood agitation
Strong oil quenchviolent agitation
Poor water quenchno agitation
Very good water quenchstrong agitation
Brine quenchno agitation
Brine quench violent agitation
Ideal quench
Grossmann H value
0.20
0.35
0.50
0.70
1.00
1.50
2.00
5.00
It is possible with high-pressure impingement to achieve H values greater

than 5.00.
Tray
Quenchant flow
Fig. 32
Harmful effects of vertical quenchant flow

through the load
Table 9 Comparison of typical heat transfer rates

Quench medium
Still air
Nitrogen (1 bar)
Salt bath or fluidized bed
Nitrogen (10 bar)
Helium (10 bar)
Helium (20 bar)
Still oil
Hydrogen (20 bar)
Circulated oil
Hydrogen (40 bar)
Circulated water
Quenchant
flow
Heat-transfer rate, W/m2 K
5080
100150
350500
400500
550600
9001000
10001500
12501350
18002200
21002300
30003500
Quenchant
flow
Fig. 33
Effect of quenchant flow direction on distortion

of their own weight leads to changes such as
bending, warpage, and twisting (Ref 1). Shape
distortion is attributable to component support
and loading as discussed previously.
A second form of distortion is size distortion,
which includes dimensional changes observable
as elongation, shrinkage, thickening, and thinning. Size distortion is caused by the volumetric
variation that accompanies each of the transformational phases formed upon quenching (Ref 1).
Size distortion can be further classified as: onedimensional, two-dimensional, and three-dimensional. One-dimensional distortion is exemplified by the change in length when rod or wire is
quenched and is primarily caused by transformational distortion. Two-dimensional distortion
is also primarily caused by transformational distortion, and it is exemplified by size variation in
two dimensions such as would occur with sheet
or plate stock upon quenching. With bulk material that can undergo dimensional variation in
three dimensions, the problem is due to both
thermal and transformational distortion and is
much more complicated (Ref 1). Owaku (Ref 1)
has shown that one of the primary contributors
to quench distortion is uneven (nonuniform)
cooling.
Quench Cracking. Stresses that lead to cracking are thermal, nonuniform cooling from TA and
Ms and transformational stress formed between
Ms and Mf. Nonuniform thermal stresses occur
when nonuniform surface cooling occurs. In this
case, there will be a deferred contraction in the
slower cooling areas creating a pull stress re-
sulting in pull cracking (Ref 41). This is

shown in Fig. 41(a) (Ref 41), where rapid, but
nonuniform surface cooling to point A has occurred at which time it affects adjacent point B,
which is cooling more slowly, causing a deferred
contraction at point B, both of which occur
while the martensitic transformation occurs at
the Ms point. As a result, point A, which cooled
rapidly, undergoes compression, and point B,
which cooled more slowly, undergoes pulling.
This means that pull stress is a tensile stress, and
if this stress becomes large enough, cracking will
occur and it will occur in areas where cooling is
delayed.
When there is nonuniform cooling within the
part between the Ms and Mf, there will be a
stretching or elongation in areas where the cooling is slow, which will act as a push stress
leading to push cracking (Ref 41). This is
shown in Fig. 41(b) (Ref 41) where point A,
which cooled rapidly, stretches point B (which
cooled more slowly) to point B below the Ms
temperature, which is a state that will undergo
expansion due to martensitic transformation. As
a result, the rapidly cooled point A (exterior) will
undergo pulling and the slowly cooled point B
(interior) will undergo compression. This results
in interior point B causing pushing and cracking
at point A, which will occur in areas where
quenching occurs rapidly.
Therefore, pull cracking, which occurs with
nonuniform surface cooling between TA and Ms,
and push cracking, which occurs with nonuniform cooling within the part between Ms and Mf,
Steam
Film boiling
Nucleate boiling
Fluid
Convective
heat transfer
Heat transfer
coefficient
(a)
Film of liquid
Convection
boiling
Sputtering
Mist cooling
Fluid drops
(b)
Fig. 34
vorable if parts are immersed into the quenchant so the perimeter of the stress riser (PSC)
simultaneously touches the liquid along the
entire length.
In the case where the perimeter of the stress
riser touches the quenchant at the moment of
immersion slowly in individual regions,
quench cracking will be reduced substantially.
In addition to providing a uniform quench by

optimizing agitation and surface cooling uniformity, the success of the quenching process is often dependent on the selection of a particular
quenchant. However, there are typically many
options to select from, and most quenching media are capable of providing a range of cooling
rates. This section provides a brief overview of
the various quench media that are available. Unfortunately, there is no uniform standard for reporting cooling rates available for a broad range
of quenchant media. Therefore, various published comparisons are provided, and they will
be sufficient for the reader to determine the relative quench severity offered by these quenchants.
Air
Gas
Wetting front w
Conditions for quench cracking are most fa-
Quenchants
z
Wetting front w
are the opposite of each other, although the

cracking event takes place for both between the
Ms and Mf. Figure 42 shows both push and pull
cracking (Ref 41).
For cracking to occur, the presence of a stress
riser (points of stress concentration) is typically
necessary. Two types of stress risers may be
found. One is a geometric notch that includes
cutting tool marks, sharp corner angles, and areas with rapid section thickness changes. The
other type of stress riser commonly found is a
notch in the material that may include intergranular effects, carbide segregation, and aggregates
of impurities (Ref 41).
Although it would be desirable to eliminate
all stress risers to facilitate quenching without
any potential for cracking, this simply is not possible in most cases. Instead, quenching methods
that minimize potential cracking often must be
developed. Polyakov (Ref 42) has provided immersion recommendations for parts with complex shapes where cracking potential is minimum and maximum (Fig. 43). The Polyakov
rules for quenching such parts include (Ref 42):
Wetting behavior and change of heat-transfer coefficient along the surface. (a) Immersion cooling. (b) Film
cooling. Source: Ref 14, 26
Perhaps the oldest, most common, and certainly least expensive quenching medium is air.
Air, being a gas, cools the heated part by a filmboiling cooling mechanism. (See Fig. 34 for a
schematic of different cooling mechanisms.) The
relative quench severity offered by still air relative to other quench media is shown in Table 9.
As with other quenching media, the heattransfer rates of air cooling are dependent on the

flow rate of air past the cooling metal surface.
Figure 44 shows that the cooling rate, as measured by an instrumented 20 mm (0.8 in.) diam
solid silver sphere, is significantly greater for
cooling in compressed air (1 kg/cm2) relative to
still air quenching (Ref 43). However, although
air is inexpensive and commonly available, air
quenching provides insufficient quench severity
to adequately harden many steels.
Gas Quenching
Process of transition between the three cooling phases: film boiling (FB), nucleate boiling (NB), and convective
cooling (CONV) during immersion cooling of CrNi-steel specimen with a cylindrical geometry 25 mm diam
by 100 mm (a) wetting process of a cylindrical CrNi-steel specimen being quenched from 850 C into water at 30 C, an
agitation rate of 0.3 m/s curves and (b) prismatic geometry (15 by 15 by 45 mm) in water of 60 C without forced
convection; immersion temperature of 860 C. Source: Ref 14, 27
Fig. 35
Vacuum furnace heat treating technology,

which includes gas quenching, is the fastest
growing technology in heat treating. Gas
quenching is typically performed by pressurizing
a vacuum furnace after heating is completed. A
broad range of quench severities, up to those
achievable with water, are possible with some
gas quenching media, as shown in Table 5 and
in Fig. 45 (Ref 44).
There are three main factors that affect the severity of gas quenching media: gas, flow velocity, and gas pressure. The most common gases
used in gas quenching, in order of increasing
quench severity, are argon, nitrogen, helium, and
hydrogen (Ref 45). The high cost of argon precludes its use except for applications that demand chemical inertness. Hydrogen exhibits the
best heat-transfer properties, as shown in Fig. 46
(Ref 45). Gas-quenching applications based on
gas mixtures such as nitrogen/helium have also
been studied (Ref 46). Gas blending allows the
user to calibrate a particular system more cost
effectively to the desired heat-transfer coefficient
than can a single gas system.
Quench severity is also dependent on gas pressure and flow rate, as shown in Fig. 47 (Ref 47).
To avoid undesirable hot spots and corresponding thermal gradients, as shown in Fig. 48, an
optimization of flow direction and flow rate is
required to provide optimal distortion control
and to prevent cracking (Ref 48).
Molten Metal
One of the more common metals used in molten form as a quenchant is lead. Lead exhibits a
melting point of 327 C (621 F) and is typically
used at 343 to 927 C (6501700 F). Below 343
C (650 F), the lead is too mushy (Ref 49).
Molten lead is used for patenting of steel wire
and for austempering. Due to the toxicity and
disposal problems with lead, it is seldom used in
the thermal processing of steel. However, because lead possesses a high thermal conductivity
and no film-boiling stage, it provides relatively
rapid cooling rates in a high-temperature range
not easily achievable with other quenching media.
Salt-Bath Quenching
Fig. 36
agitation.
Transition from film boiling (FB) to nucleate boiling (NB) during immersion of cylindrical silver specimen (15
mm diam by 45 mm) quenched from 850 C into a 10% aqueous polymer quenchant solution at 25 C without
Two quenching processes that require high

temperatures are austempering and martempering (marquenching). Molten-salt baths have re-
Highly alloyed steels, such as those shown in
Table 11, can be isothermally annealed.
Scaling, cracking, and distortion of high-
speed tool steels are minimized.
The risk of cracking during martensite for-
mation products is enhanced.

Most quenching salts are either binary or ternary mixtures of potassium nitrate (KNO3), sodium nitrite (NaNO2), and sodium nitrate
(NaNO3). The minimum quenching temperature
depends on the melting point of the salt mixture.
The ratio of the salt mixture may also affect the
viscosity of the medium and therefore affect
cooling rates. Quenching temperatures may
Conductance, G
Temperature, Tc
Conductance, G
mation, for example, spring wire, is reduced.
The formation of high-temperature transfor-
Tc
ts
Temperature, Tc
placed molten lead in most heat treating processes and are usually the medium of choice for
high-temperature quenching. The advantages of
salt quenching include:
Tc
range from 140 to 600 C (2851110 F). However, melting points as low as 85 C (175 F) are
attainable by the addition of up to 10% water.
Salt-baths are susceptible to potential explosive
degradation at temperatures greater than 600 C
(1110 F).
The selection of a martempering or austempering process depends on the steel being
quenched. Salt-bath temperatures in the range of
195 to 350 C (385660 F) exhibit only a minimal effect on the Grossmann H value (Ref 50).
However, like all quenchants, heat-extraction capabilities depend on the agitation rate, as shown
in Fig. 49 (Ref 50).
The quench severity of a molten-salt bath may
be substantially affected by the presence of water
as shown in Fig. 50 (Ref 51). Variations in both
agitation rate and water addition can profoundly
affect the depth of hardening of AISI 4140 steel,
as shown in Fig. 51 (Ref 51). Therefore, control
of water content is important to achieve the maximum benefit of the increase in quench severity.
However, automated monitoring equipment to
ts = tf
tf
Time, t
Table 10 Effect of quenchant and material

properties on quenching performance
Time, t
(b)
(a)
Conductance, G
Tc
Temperature, Tc
Temperature, Tc
Conductance, G
Effect on
property variation
Tc
ts
ts = tf
Time, t
tf
Fluid property
ts, tf
Dtw
Fluid property
Type of quenchant
Addition of additives
Increasing agitation (v)
Increasing bath temperature (Tb)
I, D
I, D
D
I
I, D
I, D
D
I
I, D
I, D
I
D
Metal property
Increasing thermal diffusivity (a)
Increasing cross-section size
Increasing surface roughness
Increasing surface oxidation
I
I
D
D
I
I
D
D
I
D
I
D
Time, t
(c)
ts, starting time; tf, finishing time; tw, wetting time; , heat-transfer coefficient; I, increasing; D, decreasing
(d)
Temperature decrease, TC, and increasing electrical conductance, G (proportional to the wetted surface) during
quenching in different quenchants. (a) Slow wetting. (b) Rapid (explosive-type) wetting. (c) Rapid wetting
followed by isolation of bubbles adhering to the metal surface. (d) Rapid wetting with repeated formation of film boiling
after this process. See also Fig. 35 and 36. Source: Ref 14, 28
Fig. 37
1000
1000
750
Temperature, C
Temperature, C
750
800
3
2
1
Water
500
Polymer
250
z = 80 mm
Wetting
front
600
Temperature, C
1000
500
3
250
60
R
z
40
400
20
0
200
1-3
0
0
0
10
20
30
40
Time, s
(a)
10
15
20
Time, s
10
15
20
25
30
35
Quenching time, s
(b)
Comparison of cooling curves measured at different positions in a cylindrical CrNi-steel probe (25 by 100
mm) during slow wetting (water) and sudden wetting (aqueous polymer solution) at (a) center (indicated by
C) and (b) close to the probe surface at three indicated heights (1,2, and 3)
Fig. 38
Temperature drop on the surface of a cylindrical steel probe (25 mm diam by 100 mm) at
various distances z from the lower end. Source: Ref 39
Fig. 39

control water content to within reasonably close
tolerances is currently available.
700 C
600
500
700 C
600
400
800 C
600
400
200 C
200 C
200 C
t =5s
t = 10 s
t = 15 s
(a)
825 C
725 C
700 C
250 C
600
200
500
150 C
750
800
775 C
200 C
775 C
t =5s
t = 10 s
t = 11 s
t = 15 s
(b)
Fig. 40
Time-dependent temperature distribution during cooling of cylindrical steel probes (25 mm diam by 100 mm)
in the case of slow non-Newtonian wetting and explosive Newtonian wetting
Pull stress
TA
Push stress
TA
B
Disturbance
A
B
Interior
M1
B
Fast
Exterior
B
M1
B
Slow
Quench Oils
Conventional (nonaccelerated) oils or cold
Stretch
Fluidized-bed quenching processes may be

used as safer and more ecologically acceptable
alternatives to molten-metal or molten-salt
quenching systems. A fluidized bed is generated
by blowing a gas (e.g., nitrogen, argon, carbon
dioxide, helium, or hydrogen) through a solid
support. Examples of solid supports include aluminum oxide, iron, copper, and molecular sieves
(Ref 52). Generally, aluminum oxide provides
the best heat-transfer capacity, thermal stability,
and environmental compatibility. Particle size
also contributes substantially to cooling properties. In general, cooling rates increase as alumina
grain size increases. The lowest limit of particle
size is 40 mesh (Ref 50). Fluidized-bed temperature can also affect heat-transfer rates. As
expected, cooling rates decrease with increasing
bed temperature, as shown in Fig. 52 (Ref 53).
Traditionally, the most common gases for fluidized beds were nitrogen and air. However,
other gases, such as hydrogen and helium, provide even greater potential for wider ranges of
heat-transfer rates. The heat-transfer flow rates
of hydrogen, helium, and nitrogen as a function
of flow rate is shown in Fig. 53 (Ref 53). Reynoldson (Ref 54) has shown that it is possible to
obtain cooling rates between those achievable by
a mineral oil and by a low-pressure gas quenching system using air-fluidized beds.
Most quench oils are derived from petroleum

oil basestocks, and the hydrocarbon fractions are
classified as paraffinic, naphthenic, and aromatic. The ratio of these hydrocarbon components in the petroleum basestock used to formulate a quench oil determines the viscosity,
viscosity-temperature, surface wetting, sludging,
staining, and fire-resistant properties. Quench
oils are selected on the basis of their ability to
mediate heat transfer during quenching. Quench
oils are usually classified on the basis of their
quenching speed and application temperature as:
A1
Elongation
Elongation
A1
Fluidized Beds
oils
Temperature
C
Pull cracking
Push cracking
Interior A
Stretch
Disturbance
Pull
Exterior B
Uneven surface cooling

(a)
Fig. 41
C
Ms
Ms
Fast
A
Temperature
Push
Uneven body cooling

(b)
Mechanisms that explain the generation of (a) pull stress and pull cracking and (b) push stress and push
cracking
Accelerated oils
Marquenching oils or hot oils
Although there are other classifications such as

fast, superfast, and water emulsifiable, the above
three classifications are the most commonly encountered.
Conventional quench oils are typically mineral oils, which may contain antioxidants to reduce the rates of oxidative and thermal degradation. Most of these oils have viscosities of up
to 100 to 110 SUS (Saybolt Universal Seconds)
at 40 C (100 F), although some have viscosities
of up to 200 SUS at 40 C (100 F). Conventional
quenching oils do not contain additives to increase cooling rate.

Accelerated quenching oils are formulated
from basestocks used for conventional quenching oils and are used at comparable temperatures.
In addition to antioxidants, accelerated quench
oils contain additives that enhance the wetting
process during the quench, which results in reduced-film boiling as shown in Fig. 54. These
oils are used with crack-sensitive, lower-hardenability steels.
Martempering oils are used at temperatures
between 95 and 230 C (200 and 450 F). They
are usually formulated using solvent-refined oils
with a very high paraffinic fraction to optimize
oxidative and thermal stability, which is necessary in view of their high use temperatures. Stability is also enhanced by the addition of antioxidants. Because martempering oils are used at
relatively high temperatures, a protective, nonoxidizing environment is often employed that allows them to be used much closer to the flash
point than otherwise possible under open-air
conditions. Nonaccelerated and accelerated martempering oils are available.
The quench severity of accelerated and nonaccelerated quench oils is relatively insensitive
to bath temperature as shown in Fig. 55 (Ref 55).
However, agitation does significantly impact the
performance of a quench oil as illustrated in Fig.
56 (Ref 54).
Aqueous Polymer Quenchants
Water
Water is a relatively inexpensive and available
quenchant that has been used for many years.
The quenching properties of water are shown in
Fig. 60, which shows (Ref 50):
quenching, higher cooling rates may increase the

potential for distortion and cracking (Ref 50).
The cooling power of brine solutions is not
significantly affected by small variations in bath
temperature. Although a brine can be used near
the boiling temperature of water, maximum
cooling rates are obtained at about 20 C (70 F).
Cracks are
not formed
Cracks are
formed
PSC
PSC
Cooling times increase with increasing water
temperature.
The maximum cooling rate decreases with increasing water temperature.
The temperature where the maximum cooling
rate occurs decreases with increasing bath
temperature.
Although the quenching properties of water
are highly variable and very sensitive to thermal
gradients in the water surrounding the cooling
metal, particularly at areas of varying cross-section size and other stress risers, what is not seen
in Fig. 60 is the fact that the film-boiling process
is unstable and highly variable and that water
does not rewet steel in a uniform manner during
the quench. These variations produce high surface thermal gradients and often increasing distortion and cracking. It is interesting to note that
the addition of 5 to 7% of an aqueous polymer
quenchant will provide more uniform wetting
properties without a significant decrease in cooling rates, which will result in great reductions in
cracking and distortion (Ref 50).
A
B
PSC
AA BB
Rotate
B
AA
View A
PSC
A
PSC
PSC
Immersion of cylindrical parts in a quenchant.
The dashed horizontal lines indicate the level
of the liquid; the vertical arrows indicate the direction of
immersion of the part. PSC indicates the perimeter of the
stress concentrator.
Fig. 43
Brine and Caustic Solutions

The term brine refers to aqueous solutions
containing various concentrations of salts such
as sodium chloride. The cooling rates provided
by brine are higher than that provided by water
at the same temperature. Although the presence
of salts will decrease the propensity for film boiling and soft spots often accompanying water
Compressed air
800
Temperature, C
Aqueous polymer quenchants are aqueous solutions of a water-soluble polymer, the most
common examples of which include: polyalkyleneglycol, polyvinylpyrrolidone, polyethyloxazoline, polysodium acrylate, and an additive to
provide corrosion protection. Quenchant performance is due to polymer selection and solution
viscosity. These quenchants are being used increasingly to replace conventional and accelerated oils because of their reduced cost, fire
safety, and relatively low environmental impact.
Typically, quench severity decreases with increasing quenchant concentration, increasing
bath temperature, and decreasing agitation rates.
This is shown in Fig. 57 for a polyalkylene glycol quenchant (Ref 56). As shown in this figure,
it is possible to obtain quench severities analo-
gous to a slow oil to those faster than water by

adjusting these variables. Of course, this also
means that to minimize stresses from nonuniform quenching it is necessary to optimize agitation and control bath temperature and polymer
concentration during use. The effect of bath temperature and agitation on the ability to through
harden steel is shown, for one polymer quenchant, in Fig. 58 and 59, respectively.
600
Still air
400
(a)
200
10
20
30
40
Cooling rate, C/s
Fig. 42
Comparison of cooling capacity of still and

compressed air. Solid line, still air; dashed line,
compressed air
Fig. 44
(b)
Two forms of quench cracking: (a) pull cracking and (b) push cracking

fer coefficient. The Grossmann number (H) is
defined as:
The most common alternative to NaCl brine

solutions is aqueous solutions of NaOH (caustic). The addition of NaOH produces slower
cooling rates than NaCl brines in the martensitic
transformation temperatures for many steels
(350 C, or 600 F) which would be expected
to reduce cracking propensity as shown in Fig.
61 (Ref 50). Interestingly, the maximum cooling
rate occurs at about 10% salt concentration for
both NaCl and NaOH.
H h/(2k)
where k is the conductivity of the metal. Thus,

the H value is equal to the interfacial heat-transfer coefficient divided by twice the thermal conductivity of the metal.
When the H value is multiplied by the dimensions of the body, their product corresponds to
the well-known dimensionless Biot number (Bi)
(Ref 57):
Measurement and Evaluation of

Power of Quenching Media
Bi HD hD/(2k) hR/k
When designing a quenching process, it is important to consider quenchant selection, quench

severity, and quench uniformity. Quench severity is defined as: the ability of a quenching medium to extract heat from a hot steel workpiece.
Heat removal from workpieces during quenching can be quantitatively described in terms of
the interfacial heat-transfer coefficient. The heattransfer coefficient is defined as:
where D and R (D/2) are the diameter and the

radius of the cylindrical or spherical workpiece
or, correspondingly, the thickness and halfthickness of the plate.
The Biot number compares the relative magnitudes of surface heat transfer and internal conduction resistance to heat transfer. A very low
value of the Biot number means that internalconduction resistance is negligible compared
with surface heat-transfer resistance. This in turn
implies that the temperature will be nearly uniform throughout the workpiece. Therefore, the
magnitude of the Biot number shows the nonuniformity of temperature distribution throughout the body of the workpiece during quenching
(Ref 57).
The most significant deficiency of the H value
is the incorrect assumption that cooling occurs
at a constant rate during quenching. Even with
the well-known limitation, the H value has been
widely accepted by heat treaters. The Grossmann
H values for commonly used quenching media
are shown in Table 8.
h Q/[A(Tp T1)]
where h is the heat-transfer coefficient, Q is the

heat flow from the workpiece to the quenchant,
A is the surface area of the workpiece, Tp is the
workpiece surface temperature, and Tl is the
quenchant temperature.
Grossmann Number
In steel quenching, the Grossmann number
(Ref 26) is used more widely than the heat-trans-
Water
Cooling-Curve Analysis
The cooling-curve tests with some metallic
probes have been used as the most useful means
of testing the cooling power of liquid quenchants
such as oils, water, and polymer quenchants. The
cooling curves are usually obtained by quenching a test specimen instrumented with thermocouples and measuring the temperature as a
function of time at specific points within the
specimen. The testpieces used are cylinders,
plates, and spheres; the most common is a cylinder with a length three to four times its diameter and with a thermocouple located at its geometric center. There are several standard
cylindrical specimens (standard probes).
From the time-temperature curve of the standard probe, one can calculate the temperature/
cooling rate curve, the heat-transfer coefficient
curve, and the heat flux curve. Several characteristic parameters that can be obtained from
these curves are the maximum cooling rate, the
cooling rate at 300 C (572 F), the characteristic
temperature (the minimum temperature of film
boiling), the temperature at the maximum cooling rate, cooling time from 800 to 400 C, and
so forth. In order to harden steel, the maximum
cooling rate and the temperature must be sufficiently high to minimize ferrite and pearlite
transformation. The cooling rate at 300 C (572
F) can be used to indicate the probability of distortion and cracking because it is near the martensite temperature of most carbon and low-alloy
steels.
Cooling-Curve Test Methods

The most common method in use throughout
the world to evaluate the cooling properties of
quenching oils is cooling-curve analysis. Cooling-curve analysis provides a cooling time versus temperature pathway, which is directly pro-
Fluids
Polymer
Oil
Salt
Nitrogen
Helium
40 bar
20 bar
Gases
10
6
1000
30
20
10
Single component
Nitrogen
Helium
6 bar
3
1
Temperature, C
Load
10 bar
5
1200
Hydrogen
6 bar
3
2
1000
600
6 bar
Argon
400
3
2
2000
800
Helium
200 Hydrogen
Hydrogen
3000
4000
Heat-transfer coefficient, W/m2 K
Fig. 45
Nitrogen
Comparison of the heat-transfer coefficients achievable with different gas quenching media for bulk-loading
and single-component quenching
12 18 24 30 36 42 48 54 60
Time, min
Fig. 46
Comparison of cooling properties of common

gases
Cross-section
size
Steel type
mm
in.
Hardness,
HRC
Martempering
Austempering
3565
3565
63
3555
3555
Low-alloy steel 13 0.5

Alloy steel
100 4.0
Cast iron
13 0.5
100
100
80
Cooling time, min
Cooling time, min
80
60
40
Diameter, in.
(by 3 diameters long)
3
3000 rpm
2
2000 rpm
1000 rpm
400
400 F
F (205 C)
C)
0
0
10
Time, min
3
Diameter, in.
Table 11 Martempering (marquenching)

versus austempering for various steels
or another preferred temperature (Ref 58). The

temperature inside the probe and the cooling
times are recorded at selected time intervals to
establish a cooling curve (temperature-time
curve). The cooling rate is derived from the temperature-time curve.
JIS K 2242. The cylindrical silver probe (Fig.
63) was adopted and set in 1965 for the JIS
method (Japanese industrial standard K 2242)
(Ref 60), because silver does not exhibit phase
transformations or surface-oxidation effects on
the cooling curve and it also exhibits a high thermal sensitivity. However, the considerably different thermal conductivity from that of steel, the
difficulty of preparing the delicate surface thermocouple assemblies, and the poor mechanical
strength have been cited as disadvantages. Furthermore, the JIS silver probe is not appropriate
for water and polymer quenchants, because water or polymer quenching of the JIS probe significantly decreases the lifetime of the silver-alumel thermocouple. Therefore, a modified silver
probe (Fig. 64) is used for polymer quenchants
3000 rpm
2000 rpm
1000 rpm
1
500
500 F
F (260 C)
C)
0
0
10
Time, min
3
Diameter, in.
portional to physical properties, such as hardness

of quenched steel parts. The results obtained by
this test may be used as a guide in selecting heat
treating oil or comparing the quench severities
of different heat treating oils, new or used.
ASTM D 6200 and ISO 9950. Cooling-curve
analysis of quenching oils according to ASTM
D 6200 (Ref 58) and ISO 9950 (Ref 59) is conducted using the Inconel 600 alloy probe (Fig.
62). The probe is heated to 850 C (1562 F) in
an electric furnace and then immersed in at least
700 mL of the quenching oil, typically at 40 C,
3000 rpm
2000 rpm
2
1000 rpm
1
600
600 F
F (315 C)
C)
0
0
10
Time, min
60
Fig. 49
Effect of agitation on the quench severity of

molten salt
40
20
20
50
0
4000
8000
12000
16000
Volumetric flow rate, ACFM
20
40
60
Cooling rate calculated

from temperature curve
measured in the center of a
25 mm diam steel specimen
80
Chamber pressure, PSIG

(b)
(a)
(a) Effect of volumetric flow rate at constant density on cooling time. (b) Effect of chamber pressure at constant
volumetric rate on cooling time
Areas of low
heat transfer
Cold spots formed
360
cooling
Maximum cooling rate, K/s
45
Fig. 47
1 vol% water
added
40
Without water
addition
35
30
(a)
(b)
Effect of gas flow pattern on quench uniformity. (a) Gas-injection nozzles with flow perpendicular to the
workpiece. (b) Gas-injection nozzles with flow horizontal to the workpiece. (c) Combination of horizontal
and vertical flow
Fig. 48
(c)
0.3
0.6
Agitation rate, m/s

Maximum cooling rate of a hot-salt bath (Degussa AS-140) at 200 C (390 F) as a function
of agitation and percentage of water added
Fig. 50

(Ref 61). It is the same shape as the JIS probe,
and it is equipped with a metal-sheathed Chromel-Alumel (CA) thermocouple (K type) (1.0
mm outer diameter, 0.2 mm wire diameter) at
these geometrical centers instead of the silveralumel thermocouple of the JIS probe.
Techniques for Estimation of HeatTransfer Coefficient

To evaluate the cooling power of quenchants
more qualitatively, the surface heat-transfer co-
55
55
With water
addition
Without water
addition
5
50
50
16 mm
efficient and/or the surface heat flux during

quenching are estimated from the measured
cooling-curve data of standard probes. The surface heat-transfer coefficient or the surface heat
flux is necessary as the surface-boundary condition for computer simulation of the quenching
process.
Some practical techniques for estimation of
the surface heat-transfer coefficient and surface
heat flux have been proposed. The lumped-heatcapacity method (Ref 62) is a classical and simple method for the determination of the surface
heat-transfer coefficient and the surface heat flux
during quenching. This method provides accurate heat-transfer coefficients from quenching
data obtained from silver probes that have high
thermal conductivity (Ref 6365). On the other
hand, inverse methods are more suitable for estimating the heat-transfer coefficients from
quenching data obtained using the Inconel 600
alloy probe, because of its low thermal conductivity (Ref 6365).

The lumped-heat-capacity method can be
used if an assumption of uniform probe temperature during cooling can be justified (Ref 63
65). If the probe temperature is uniform, the heat
loss from the probe Q is equal to the decrease in
the internal energy of the probe:
Q hA(Tp T1) cqV(dTp /dt)
where h is the heat-transfer coefficient on the

probe surface, A is the surface area of the probe,
Tp is the probe temperature, Tl is the quenchant
temperature, c is the specific heat of the probe
material, q is the specific density of the probe
material, V is the volume of the probe, and t is
time. So, dTp /dt is the cooling rate of the probe.
If the quenchant temperature around the probe
Tl is uniform, the next relationship is derived
from Eq 1:
45
2
40
40
35
35
1
30
30
3/ 4 R 1/ 2 R 1/ 4 R
1/ 4 R 1/ 2 R 3/ 4 R
H2
Fig. 51 Influence of agitation rate and water addition

on hardness distribution in 50 by 200 mm (2
by 8 in.) AISI 4140 steel bars quenched in a hot-salt bath
(Degussa AS-140) at 200 C (390 F). Agitation rates and
water additions: 1, 0 m/s, 0 vol%; 2, 0.3 m/s, 0 vol%; 3,
0.6 m/s, 0 vol%; 4, 0.6 m/s, 1 vol%; 5, 0.6 m/s, 2 vol%
(Eq 2)
where q is the heat flux on the probe surface.

Here, probe weight qV is constant, but c and
A change with temperature. If the change of the
surface area of the probe A is ignored, only the
temperature dependence of the specific heat c
(Eq 2) need be considered, and then:
He
200
q h(Tp T1) c(Tp)(qV/A)(dTp /dt)

100
10
12
14
16
Fluidizing flow
Minimum fluidizing flow
Fig. 53
(Eq 3)
where c(Tp) is the specific heat as a function of

probe temperature Tp.
Thus, the heat flux q and/or the heat-transfer
coefficient h can be directly calculated from the
cooling rate dTp /dt (Ref 64, 65). Figure 65
shows heat-transfer coefficients calculated with
the lumped-method and cooling-curve data of
the JIS silver probe quenched in different quench
media.
N2
50 mm diam
q h(Tp T1) (cqV/A)(dTp /dt)
300
Heat transfer (600 C), Btu/h ft2 F
47 HRC
Surface hardness, HRC
Surface hardness, HRC
4 mm
45
(Eq 1)
Effect of fluidized gas on heat transfer
1600
1400
1200
5
Cooling rate
800 500 C
Temperature, F
Cooling rate, C/s
3
2
Cooling rate
800 300 C
100
Fused salt
500 F
F
Martempering oil
500 F
F
1000
Molten
metal
500 F
F
800
600
1
0
100
Conventional
quench oil
150 F
F
200
300
400
500
Fluidized bed temperature, C

Effect of fluidized-bed temperature on cooling
rate between 800500 C (1470930 F) and
800300 C (1470570 F) measured with 75 by 100 mm
(3 by 4 in.) 90MoCrV8 steel sample; temperatures measured at the center of the section
Accelerated
quench oil
150 F
F
400
Fig. 52
200
0.7
1.0
1.5
10
15
20
30
40
Time, s
Fig. 54
Comparison of cooling-curve performance of accelerated versus a nonaccelerated quench oil
60

The numerical inverse methods require
techniques that decrease instabilities due to excessive sensitivity to measurement errors and
noise. The accuracy of the solutions obtained by
the numerical inverse methods may be questionable because of various conditions in the numerical calculation including: time step, spatial
step size, magnitude of noise, error in measured
data, and so forth. Furthermore, the accuracy is
not achieved for all cooling stages in quenching,
unless these conditions are appropriately set for
each stage.
One of the numerical inverse methods is the
solution of the inverse heat conduction problem
(Ref 66, 67). Another method is the iteration
method, which utilizes both the lumped-heat-capacity method and a two-dimensional inverse
method with the least residual procedure (Ref
64, 65). The calculation procedure is:
1. The initial values of heat-transfer coefficients

are approximately calculated by a lumpedheat-capacity method from the cooling-curve
data of the probe.
2. The total temperature region of the cooling
process is divided into about 20 subregions
and the initial heat-transfer coefficients are
defined as stepwise continuous functions in
each region.
3. The cooling curve is calculated by a two-dimensional finite difference method using
these coefficients as the surface-boundary
condition.
4. The residual sum of squares between the
measured temperatures Texp and the corresponding calculated temperatures Tcal is
given as R (Texp Tcal)2.
5. The cooling curve is calculated again after
changing the heat-transfer coefficients.
6. The iteration of 3, 4, and 5 is performed until
Tcrit {R/n}1 / 2 is satisfied.
16
Strong agitation
Moderate agitation
No agitation
15
Cooling rate at 550 C, C/s
14
where Tcrit is a critical value of the temperature

difference between the measured temperatures
and the corresponding calculated temperatures
and n is the number of data Texp in the subregion.
An iterative technique to minimize R is used.
The same procedure is repeated in each subregion in the whole cooling process.
Figure 66 (Ref 65) illustrates an example of
heat-transfer coefficients calculated with the inverse method and cooling-curve data obtained
using an ISO Inconel 600 alloy probe quenched
in various quenchants. Heat-transfer coefficients
that were calculated from the cooling-curve data
of the JIS probe are also shown in Fig. 66. The
lumped-heat-capacity method program LUMPPROB (Ref 63) was used for the JIS silver
probe, and the inverse method program InvProbe-2D (Ref 64, 65) was used for the ISO
13
12
11
10
9
8
7
6
5
Fig. 55
8
12
16
20
24
28
Effect of agitation of a quench oil on the ability

to through harden steel
probe. The values of the heat-transfer coefficients in the convection-cooling stage (the
lower-temperature region) are almost the same
for each quenchant, despite the differences of the
sizes and the materials of these probes. However,
the values in the higher-temperature region are
very different for each probe. This is caused by
the difference of the thermal properties between
the JIS probe and the ISO probe.
Wetting Behavior and Nonuniform

Quenching
It is well known that the wetting process on
the surface of hot metal affects cooling characteristics during water, polymer, and oil quenching (Ref 40). If the initial temperature of hot
metal is sufficiently high, a vapor film is formed
on the surface immediately upon immersion. After vapor blanket cooling stage (film boiling), a
collapse of the vapor blanket (i.e., wetting) occurs progressively or explosively. The progressive wetting during quenching could increase
nonuniform cooling because the heat-transfer
coefficient on the nonwetted surface is significantly lower than that on the wetted surface.
Hence, the wetting behavior is a major factor in
cooling, and it affects quench distortion of steel
parts.
Cooling-curve analysis provides relatively little information about the wetting process. The
temperature in the center of the probe at the transition point from lower to higher cooling velocity only describes the wetting of the nearest surface point and gives no information about the
wetting process on the entire surface of the
probe. The following techniques are employed
for measuring the wetting process.
140
900
0.4
Predicted hardness for a quench oil (25 mm, or 1 in. bar)
0.6
135
800
0.4
700
Temperature, C
600
15.7 HRC, 43 C (110 F), FPM = 0
31.5 HRC, 43 C (110 F), FPM = 100
16.6 HRC, 88 C (190 F), FPM = 0
32.1 HRC, 88 C(190 F), FPM = 100
500
400
Bath temperature, F
130
0.6
125
0.8
120
0.6
115
0.8
1.0
110
300
Flow rate = 0
200
1.0
Flow rate = 100
100
15
100
0
1E 2
1E 1
1E + 0
1E + 1
1E + 2
1E + 3
1E + 4
1E + 5
1E + 6
1E + 7
Time, s
Fig. 56
0.8
105
Relative insensitivity of the ability of a quench oil to harden steel with respect to varying bath temperature.
FPM, flow rate
Circ rate = 50
Circ rate = 75
Circ rate = 100
17
19
21
23
Polymer concentration, %
25
Effect of variation of quenchant concentration,

agitation rate, and bath temperature on the
Grossmann H value for one polyalkylene glycol quenchant. Circ rate in ft/min. Note: these curves vary greatly
for each polymer quenchant; therefore, these data are for
general illustrative purposes only.
Fig. 57

Measurement of Wetting Kinematics
Direct Observation with Video or Photographs. For transparent quenchants, the surface
wetting process may be observed by video or
photographically. Images captured on videotape
or sequential photographs provide detailed information about the wetting process. Figure 67
(Ref 68) shows examples of the wetting behavior
and change of wetting distance Lf with time, during water, polymer, and oil quenching of the
stainless steel disk specimen (30 mm diameter
by 10 mm thick). A progressive collapse of the
vapor blanket from the sharp edges was observed
in water and oil quenching, while an instantaneous collapse occurred in polymer quenching.
Wetting length Lf versus time t was estimated
from the video images. The t-Lf curves provide
wetting velocity, time when wetting starts (ts),
and time when wetting is finished (tf).
Temperature Measurement. The current
method for examining the wetting process is to
measure cooling curves by inserting thermocouples in near-surface positions along the specimen
surface. Figure 68 (Ref 68) shows cooling curves
during water, polymer, and oil quenching of the
stainless steel disk specimen (30 mm diameter
by 10 mm thick) without agitation. Temperatures
were detected by metal-sheathed thermocouples
attached to the specimen. Figure 69 shows the
positions of metal-sheathed thermocouples for
the measuring cooling curves. Thermocouples
were inserted tightly into holes drilled to the geometrical center A and positions B, C, and D near
the specimen surface. Basic information can be
derived from the near-surface cooling curves. A
period of slow cooling and a subsequent period
of fast cooling are equivalent to a period of nonwetting cooling stage (vapor blanket stage) and
nucleate-boiling stage. The transition point indicates the transition from film boiling to nucleate boiling at the thermocouple location. The onset and conclusion of wetting, and thus the
velocity of wetting front, can be calculated by
extrapolating the local transition times to the
starting and finishing positions of the wetting.
Conductance
Measurement.
Another
method for determining the wetting process is to
measure the electrical conductance between the
specimen and an electrode immersed in the
quenchant (Ref 69) (Fig. 70). When a vapor
blanket forms around the entire specimen surface, the conductance between the specimen and
the electrode is low. When the vapor blanket collapses on the specimen surface, the conductance
increases as shown in Fig. 71. The increase in
the conductance is approximately proportional to
the wetted portion of the specimen surface. The
conductance is at its highest value when the entire sample surface is completely wetted. Conductance-time curves provide the time when
wetting starts (ts) and the time when wetting is
finished (tf).
Effect of Process Variables on

Cooling Behavior and Heat Transfer
The quenching of steel requires a wide variation and sufficient reproducibility of heat transfer in order to achieve the required cooling rate,
which is dependent on the hardenability of the
steel and the section size of the workpiece.
Therefore, it is important to recognize the influence of the variables during quenching.
Heat transfer during quenching is affected by
many factors, such as the quenchant, bath temperature, agitation, and the dimensions and
shape of the workpiece being quenched. Only a
few of these factors can be realistically varied,
including bath temperature, agitation rate, and
the quantity and racking arrangement of the
workpieces.
Effect of Quenchant Selection

The selection of quenchant is the most basic
factor that affects cooling behavior. When selecting a quenchant, it is important to consider
quench severity and quench uniformity.
The cooling curves in Fig. 72(a) were obtained using a 10 mm diam silver specimen
1000
1000
30 C
C
800
40 C
C
Temperature, C
Temperature, C
800
600
50 C
C
400
30 C
C
200
Fig. 58
50 C
C
600
40 C
C
400
750
0 m/s
500
0.3 m/s
0.6 m/s
250
200
20 C
C
C
20 C
Temperature (TC), C
1000
quenched in various quenchants without agitation. Different cooling curves were obtained for
each quenchant. A lumped-heat-capacity method
was used to estimate heat-transfer coefficients
from these cooling-curve data. These coefficients are shown in Fig. 72(b).
In general, the cooling curve during water,
polymer, and oil quenching has three stages
(stage A: vapor blanket cooling (film-boiling)
stage, stage B: nucleate-boiling stage, and stage
C: convection-cooling stage). Heat-transfer coefficients in the high-temperature region (vapor
blanket stage) are relatively low. Cooling curves
of workpieces quenched into nonvolatile quenchants such as molten salt, liquid metal, and gas,
do not exhibit vapor blanket cooling or nucleate
boiling.
The wetting behavior in the quenching of hot
metal is affected by quenchant properties such
as the type of quenchants, quenchant temperature, and so forth. As shown in Fig. 67, in water
and oil quenching, the vapor blanket was formed
around the entire surface of the specimen immediately after immersion and then ruptured
progressively from the sharp edges to the center
of the flat surface of the disk specimen. The progressive collapse of the vapor film begins immediately after immersion in water. The progressive collapse starts after the formation of a
stable vapor blanket stage in oil quenching. The
time of duration varies according to the oils used.
The starting time of the wetting during polymer
quenching depends on the type of polymer used,
polymer concentration, and liquid temperature.
Wetting starts immediately upon immersion for
some aqueous polymer solutions and after the
formation of a stable vapor blanket stage for
other polymer quenchants.
The wetting velocity (the propagation velocity
of wetting front) during water or oil quenching
is significantly slower than during polymer
quenching. Such a slow wetting velocity could
cause a nonuniform surface cooling in the specimen, because the heat-transfer coefficient on the
nonwetting surface is significantly lower than
that on the wetting surface. Figure 68(a) shows
the cooling curves at representative positions
0
10
Time, s
15
20
Effect of bath temperature on the ability to through harden steel
50
100
150
Cooling rate, C/s
200
Fig. 59
15
30
Time (t ), s
45
60
Effect of agitation rate on the ability to through

harden steel
In the vapor blanket stage during water, polymer, and oil quenching, heat is transferred across
1000
1800
Temperature, C
800
600
NaOH
14
NaCl
2.8
2.5
12
2.1
10
1.8
Fig. 61
10
15
20
Salt concentration, %
1.4
25
Effect of salt concentration on the heat-transfer

coefficients of NaCl and NaOH brine solutions
1600
1400
1200
1000
800
400
16
Temperature, F
Water temperature:
40 C
50 C
60 C
70 C
80 C
90 C
fer depends on the temperature difference between the quenchant and the surface of the
workpiece. It also depends on quenchant viscosity, which governs the convection flow of the
quenchant around the workpiece.
The effect of water temperature on cooling
curves of a 12.5 mm diam by 60 mm Inconel
600 probe is shown in Fig. 60. By decreasing
Heat transfer coefficient (),

(Btu/ft2 h F) 103
Effect of Quenchant Temperature
the surface through the vapor blanket by conduction, radiation, and convection. When the
surface temperature of the workpiece decreases,
the thickness of the vapor blanket is reduced until the liquid contacts the hot surface of the workpiece. This is the start of the wetting process. By
direct contact with the surface, the vapor film
begins to collapse, dramatically increasing the
heat flux and the cooling rate. By increasing liquid temperature, the energy required for evaporation is reduced and the thickness of the vapor
film increases. As a result, both the cooling rate
and the heat-transfer rate decrease. In addition,
by stabilizing the vapor blanket, the transition
temperature from lower to higher cooling rate
(the minimum temperature required for film
boiling) decreases.
Once wetting has begun, heat transfer by nucleate boiling controls the cooling rate. This
stage is also dependent on quenchant temperature. In the convection-cooling stage, heat trans-
Heat transfer coefficient (),

mW/m2 K
and the temperature differences between these

positions during water quenching. Cooling immediately upon immersion is very rapid on the
edges (position D), and the difference of the surface temperature, TB TD, is very large (about
500 C at maximum).
Instantaneous collapse of the vapor blanket
occurs in polymer quenching. Wetting velocity
after the start of the breakdown of the vapor film
is very high. Such wetting behavior could reduce
the nonuniformity of surface cooling of a specimen. Figures 68(b) and (c) show that the differences of the surface temperature (TB TC and
TB TD) during polymer quenching is smaller
than that in water quenching (Fig. 68a) or oil
quenching (Fig. 68d).
Termination
12.5
9.5
600
400
200
Support
tube
200
0
10
20
30
Time, s
40
50
60
1.5
(a)
1000
100
150
Cooling rate, F/s

50
250
200
30
1800
1000
800
Water temperature:
40 C
50 C
60 C
70 C
80 C
90 C
200
20
40
60
80
100
Cooling rate C/s
120
140
200
r 0.75
600
400
30
400
1.5 mm
sheathed
thermocouple
160
1200
Temperature, F
Temperature, C
1400
600
30
1600
800
Probe
200
160
(b)
Fig. 60
Effect of water bath temperature on heat removal using a 12.5 mm diam by 60 mm Inconel 600 probe with
a flow velocity of 0.25 m/s (50 ft/min)
ISO Inconel 600 alloy probe. International

standard, ISO 9950. r, radius; , diameter. Dimensions given in mm. Source: Ref 59
Fig. 62
Insulating
tube
Supporting
rod
Silver
wire
Silver
pipe
30
Heatresistant
insulator
10
M6 1
5
Almel
wire
3
15
Silver
rod body
15 0.1
Insulation
tube
10 0.1
Japanese industrial standard silver probe. JIS K

2242-1997. , diameter. M6 1, metric fine
screw thread. Dimensions given in mm. Source: Ref 63
Fig. 63
water temperature, the duration of film boiling

decreases, which is indicated by the accelerated
transition from slower cooling to faster cooling.
In addition, cooling rates of the three different
cooling stages are increased.
Cooling curves during oil quenching are relatively insensitive to bath temperature as shown
in Fig. 55. However, the temperature of an aqueous polymer solution does significantly impact
on the cooling performance as shown in Fig. 58.
Cooling curves of workpieces quenched by
nonvolatile quenchants such as molten salt, liquid metal, and gas do not possess a vapor blanket
and nucleate-boiling stage. Workpieces are
cooled by radiation and convection of the quenchant. In this case, the heat-transfer coefficient
depends on the temperature difference between
the quenchant and the surface of the workpiece.
Effect of Agitation
In addition to quenchant temperature, vaporfilm stability is greatly affected by quenchant agitation. A high flow velocity increases the heat
transfer by convection and reduces the thickness
of the vapor film. Furthermore, agitation of the
liquid reduces the stability of the vapor/liquid
interface by disturbing the liquid flow adjacent
to the vapor film. Therefore, agitation reduces
the duration of film boiling. Heat transfer in the
convection-cooling stage directly depends on agitation. The effect of agitation on the heat transfer by nucleate boiling is not significant because
nucleate boiling agitates the liquid adjacent to
the boiling surface by itself (self-agitation effect).
Figure 73 shows the effect of selected flow
velocities (00.4 m/s) on cooling curves of a
cylindrical silver probe (10 mm diameter by 30
mm, Fig. 64) quenched in 80 C oil. In the vapor

blanket stage and convection stage, the temperature of the specimen decreases slowly and the
cooling rates in these stages accelerate by the
agitation. As the vapor film collapses, the temperature drops rapidly because of the transition
from film boiling to nucleate boiling. This transition temperature from lower to higher cooling
rates is almost unaffected by flow velocity.
Figure 59 shows the effect of selected flow
velocities (0, 0.3, and 0.6 m/s) on cooling curves
in polymer quenching. In this case, the liquid
flow around the workpiece reduces the stability
of the vapor blanket. Therefore, the transition
from lower to higher cooling rates may be
strongly influenced by the liquid flow velocity,
and the agitation reduces the duration of film
boiling.
The effect of agitation on the wetting process
on a stainless steel disk (30 mm diameter by 10
mm thick) during water quenching is shown in
Fig. 74 (Ref 68). Figure 75 shows the effect of
agitation on the cooling curves of this stainless
steel disk at the representative positions (shown
in Fig. 69) and the temperature differences between these positions during water quenching.
These results show that the agitation of water
increases the wetting velocity and hence shortens
the finishing time of wetting tf. However, the
starting time of wetting ts in water quenching
does not depend on the flow velocity because the
wetting at the sharp edges always starts immediately after immersion (edge effect). As a result,
the degree of nonuniformity of the surface cooling (see the values of TB TD and TC TD)
during water quenching significantly is decreased by the agitation.
Figures 76 and 77 (Ref 68) illustrate the effect
of agitation on the wetting process and the cool-
Supporting
rod
105
M6
5
5
5
15
Silver
rod body
Water, 30 C
10% polymer quenchant (PAG), 30 C
10% brine, 30 C
Oil (JIS1-2), 80 C
Oil (JIS2-1), 120 C
Molten salt, 230 C
105
Water (30 C): JIS probe and LUMPPROB

Water (30 C): ISO probe and InvProbe-2D
10% PAG (30 C): JIS probe and LUMPPROB
10% PAG (30 C): ISO probe and InvProbe-2D
Oil: JIS1-2 (80 C): JIS probe and LUMPPROB
Oil: JIS1-2 (80 C): ISO probe and InvProbe-2D
5 104
Heat-transfer coefficient (h ), W/m2 K
1.0 sheathed
thermocouple
106
104
103
2 104
104
5000
2000
1000
500
200
15
100
0
102
61
Modified silver probe. M6, metric coarse screw

thread. Dimensions given in mm. Source: Ref
400
600
800
1000
Surface temperature, C
10
Fig. 64
200
Heat-transfer coefficients calculated with the

lumped method and the cooling-curve data of
the JIS silver probe quenched in representative quenchants
Fig. 65
200
400
600
800
Surface temperature (T ), C
1000
Heat-transfer coefficients calculated with the

lumped method and the cooling-curve data of
the JIS silver probe or with the inverse method and the
cooling-curve data of the ISO Inconel alloy probe. Source:
Ref 65
Fig. 66
30 C still water
TA TB
TB TC
TB TD
TB (surface: B)
600
TC (surface: C)
400
TD (edge: D)
200
10
800
600
400
A
B
C
D
15
7
7
1
200
800
0
30
20
TA (center)
TD (edge: D)
200
10
400
A
B
C
D
15
7
7
1
200
0
30
20
Time (t ), s
TA TB
TB TC
TB TD
TA (center)
TB (surface: B)
600
TC (surface: C)
TD (edge: D)
400
200
10
20
30
A
B
C
D
40
800
600
400
15
7
7
1
50
200
80 C still oil (JIS 1-2)

T A TB
TA (center)
T B TC
TB (surface: B)
T B TD
TC (surface: C)
800
Temperature (T ), C
60 C still 25% PAG
Temperature difference (T ), C
(b)
800
Temperature (T ), C
600
TC (surface: C)
400
Time (t ), s
0
60
600
TD (edge: D)
400
A
B
C
D
15
7
7
1
800
600
400
200
200
Time (t ), s
Fig. 68
800
TA TB
TB TC
TB TD
TB (surface: B)
600
(a)
(c)
30 C still 10% PAG
20
10
Time (t ), s
(d)
Cooling curves and variation of temperature difference during (a) water, (b) and (c) polymer, and (d) oil quenching of stainless steel disk
0
30
Temperature (T ), C
TA (center)
Temperature (T ), C
800
Change of wetting length with time on the surface of stainless steel disk specimen during water, polymer, and
oil quenching. Quenchants: city water, PAG polymer aqueous solutions, and quenching oils. Source: Ref 68
Fig. 67
ing characteristics during polymer quenching

where agitation of the polymer quenchant hardly
affects the wetting velocity during quenching.
Therefore, the degree of nonuniformity of the
surface cooling during polymer quenching is not
affected by the agitation. However, the starting
time of wetting ts in still polymer quenching varies widely because the poor reproducibility of
the formation and collapse of the vapor blanket
without agitation. However, agitation of the
polymer quenchant often prevents the formation
of the vapor blanket and achieves the reproducible rapid cooling immediately after immersion.
Figures 78 and 79 (Ref 68) show the effect of
agitation on the wetting process and the cooling
characteristics during oil quenching. Agitation
of the quenching oil hardly affects the starting
time of wetting ts, but increases the wetting velocity and shortens the finishing time of wetting
tf. As a result, the degree of nonuniformity of the
surface cooling during oil quenching is somewhat decreased by agitation. However, the effect
of agitation on the cooling characteristics is insignificant, because the vapor blanket in oil
quenching is stable in the high-temperature region where the temperature is greater than the
transition temperature with or without agitation.

Furthermore, the transition temperature in oil
quenching hardly depends on the flow velocity
and the wetting at the sharp edges starts after
vapor blanket cooling.
Effect of Surface Conditions

In actual heat treating, the surface conditions
of the parts vary due to oxidation, contamination, coating with other material, surface roughening, and so forth. Therefore, the effect of surface conditions on the cooling characteristics in
quenching is very important.
Surface Oxidation. As discussed in the section Effect of Materials and Quench Process
Design on Distortion in this article, oxide scale
affects cooling during oil quenching of steel
parts, as shown in Fig. 30 (Ref 14). In general,
light scale (thin oxide layers) increases the cooling rate in oil quenching, but heavy scale (thick
oxide layers) retards the cooling rate. This phenomenon often occurs in water or polymer
quenching.
10
1.5
Figures 80 and 81 (Ref 70, 71) illustrate the

effect of surface oxidation of silver, nickel, stainless steel, and pure iron cylinders with hemispherical ends (without edge effect) on cooling
curves during quenching in still water. These results show:
Oxidation of a probe surface exhibits little ef-
fect on the cooling characteristics during vapor blanket cooling.

Surface oxidation increases the lower-limit
temperature of the vapor blanket stage. This
tendency is more remarkable in the case of
lower water temperature.
The presence of a porous oxide layer on the
surface results in a remarkable rise of the vapor blanket, the transition temperature from
lower to higher cooling rate (the minimum
temperature required for film boiling) because of the low thermal conductivity and
large roughness of the oxide layer.
A similar result was reported for spray cooling
of a hot steel plate (Fig. 82) (Ref 72). It has been
confirmed that this cooling acceleration effect
with thin oxide layer is caused either by suppression of the vapor blanket formation or by
acceleration of the collapse (destabilization of
vapor blanket) by the existence of the oxide
layer.
1.6 sheathed
thermocouple
Surface texture and roughness exhibit little
Specimen
effect on the cooling characteristics at the vapor blanket stage.

Surface roughness causes the rise of the
lower-limit temperature of the vapor blanket
stage.
30
B
7
14
D
3
Unstable cooling phenomena due to surface

oxidation were observed for water quenching of
steel (Fig. 83). Unstable changes of the cooling
rate during cooling were obtained due to flaking
of the oxide layer (Ref 70, 71).
Coating Effects. The presence of a low thermal conductivity layer such as oxides exhibits a
remarkable influence on the cooling characteristics of hot metals, as described previously.
Coatings of clay, glass, salt, and heat-resistant
paint exhibit a similar effect. Figure 84 shows
that a clay coating less than about 100 lm thick
results in a cooling acceleration effect in water
quenching of S45C steel cylinders. Such a phenomenon was observed at the full range of water
temperatures (Ref 73). This effect is caused either by suppression of the vapor blanket formation or by acceleration of the collapse by the existence of the clay coating layer (Ref 71, 73).
Kikuchi et al. (Ref 74) explained this phenomenon by considering the surface-temperature decrease during intermittent liquid-solid contact.
Figure 85 illustrates a similar result for a clay
coating on a silver probe (Ref 71, 73). In this
case, the thicker the clay coating layer, the lower
the cooling acceleration effect, because of the
insulation effect of the clay layer.
Surface Texture and Roughness. The effect
of surface texture and roughness on the cooling
characteristics in quenching is very important.
Narazaki et al. confirmed the effect of the surface
texture on the cooling curves of metallic specimen during quenching in still water (Ref 75).
The results showed:
A similar result (Ref 72) has been reported for

spray cooling of a hot copper plate (Fig. 86). It
is presumed that these effects are caused by the
destabilization of the vapor blanket, which results from microscopic liquid-solid contacts on
the rough surface.
Various effects of surface texture on cooling
characteristics have been reported (Ref 71, 75,
76). Figure 87 shows the effect of surface finish
on the cooling curves of a S45C steel specimen
during polymer quenching (Ref 71, 76). In this
case, the surface texture finished by lapping stabilizes film boiling. Sandblasted surfaces often
cause a similar effect. These phenomena may be
due to the stabilization of vapor film in liquidsolid contacts by recovering of vapor film, which
is promoted by many homogeneous bubble nucleation cavities on the lapped sandblasted surfaces.
Steel disk specimen for measuring cooling

curves. Dimensions in mm. Specimen: steel
disk (30 mm diam by 10 mm thick). Thermocouple: metallic sheathed CA thermocouple (1.6 mm outer diameter).
Measuring position: geometrical center (A) and 1.5 mm beneath surface (B, C, and D). Source: Ref 68
Fig. 69
Data scanning system:

Temperature T
Conductance G
Fluid
Unwetted
Wetted
Sample
surface
(a) Cooling curve of center of sample and variation of wetted surface area as a function of
time during cooling of cylindrical chromium-nickel steel
sample (15 mm diam by 45 mm long). (b) Photographs
show the wetting behavior at times A, B, and C. TS and tS,
respectively, are temperature and time of starting wetting;
Tf and tf, respectively, are temperature and time of finishing
wetting. Source: Ref 69
Fig. 71
Backplate
electrode
Fig. 70 Measuring principle for determination of percentage of sample surface wetted during cooling of immersion-cooled samples. Source: Ref 69
Effects of Geometry
Edge Effect. The shape of steel parts is a very
important factor for water and aqueous polymer
quenchants, but it is not as important for oils (see
Fig. 8890) (Ref 71). The initial rupture of the

curve tests to avoid the influence of surface oxidation and phase transformation. The effect of
specimen material on cooling characteristics in
quenching has been reported (Ref 71, 76). Figure
91 illustrates surface heat-transfer coefficients
obtained during oil quenching of cylindrical silver, nickel, and stainless steel specimens. The
results show that the specimen material does not
affect the heat-transfer coefficients in the vapor
blanket stage and during the convection-cooling
stage. However, the minimum temperature of the
vapor blanket stage is affected by the specimen
material. Similar results are confirmed in the
case of water quenching of cylindrical silver,
nickel, and stainless steel specimens with hemispherical ends (without edge effect) (Ref 71,
76). This phenomenon arises from instability of
the vapor blanket on the surface of lower thermal
10 30 mm silver probe
80 C quench oil (JIS 12)
800
600
500
400
300
200
100
0
Fig. 73
Temperature (T ), C
80 C oil JIS 1-2
600
120 C oil JIS 2-1
230 C molten salt

30 C water
30 C, 10% polymer
quenchant
30 C 10% brine
0
Fig. 72
10
30 C 10% brine
230 C
molten salt
30 C water
104
120 C oil
JIS 2-1
103
30 C, 10%
polymer
102
200
80 C oil JIS 1-2
400
600
800
Time (t ), s
(a)
Effect of agitation on cooling curves in oil quenching. Probe: cylindrical silver probe (10 mm diam by 30 mm
long). Quenchant: 80 C quench oil (JIS 1-2 type)
800
200
Time (t ), s
105
400
Still
0.1 m/s
0.2 m/s
0.3 m/s
0.4 m/s
700
Material Effects
Heat-resistant materials such as silver, gold,
platinum, stainless steel, nickel, Inconel alloy,
and so forth are used as specimens for cooling-
conductivity of specimen. If the thermal conductivity of the specimen material is lower, destabilization and premature collapse of the vapor
film occurs at a higher temperature because of
the increase of the surface temperature drop by
the intermittent liquid-metal contact during the
vapor blanket stage.
Steel chemical composition determines the
thermodynamic parameters of the material,
transformation behavior of austenite, and surface
oxidation. Therefore, it affects cooling behavior
during quenching. Latent heat of transformation
from austenite into ferrite/pearlite bainite, or
martensite also influences the cooling rate. The
transformation temperature and the amount of latent heat depend on the chemical composition of
the steel and the cooling rate. Thermodynamic
properties of the material also have an effect on
cooling behavior.
900
Temperature (T ), C
vapor blanket usually starts at sharp edges of

steel parts, and wetting behavior is strongly influenced by the geometry of the edges (edge effect) (Ref 35, 40, 77). Figures 88(a) and 89(a)
show that the edge shape of the ends of cylindrical specimens is very important in still water
and polymer quenching. This arises from the instability and the premature collapse of the vapor
blanket at the edges of both ends. However,
cooling curves during oil quenching are not sensitive to the end or edge shapes of cylindrical
specimens as shown in Fig. 90. This arises from
the stability of the vapor blanket around the
specimen during oil quenching. However, the effect of edge shape in water and polymer quenching can be eliminated by agitation as shown in
Fig. 88(b) and 89(b), because agitation destabilizes the vapor blanket and eliminates the vapor
blanket cooling stage independent of the edge
shape.
Effect of Size. Cooling becomes proportionately slower when the thickness of a workpiece
is increased (mass effect). Therefore, rapid cooling of the core of a workpiece with large cross
sections is impossible with any quenching media. In addition, cooling rates of the core are limited by the thermal diffusion in the workpiece.
Size of a workpiece also affects the stability
of vapor film on the surface of the workpiece and
the wetting behavior. The velocity of wetting is
reduced, and the time interval of wetting increases with increased section size. Vapor flow
rate, vapor thickness, and quenchant temperature
around the workpiece increases with the increasing workpiece size. As a result, surface heat
transfer of a workpiece is influenced by the
workpiece size.
(b)
Cooling curves of a silver probe and heat-transfer coefficients in quenching in representative quenchants. (a) Cooling curves of a 10 mm diam by 30 mm long silver probe.
(b) Heat-transfer coefficients estimated with lumped-heat-capacity method
Effect of Cooling Characteristics on

Residual Stress and Distortion from
Quenching
Steel quenching requires a wide variation in
cooling rates to achieve the required hardness
and strength, which is dependent on the hardenability of the steel and section size of the workpiece. At the same time, residual stresses, distortion, and crack formation must be minimized.
However, these are often contradictory objectives. For example, although increasing cooling
rates increases hardness, it often increases the
potential for distortion, stress, and cracking.
Similarly, distortion and stress during quenching
are affected by many factors, such as the quenchant, the bath temperature, and agitation. The
dimensions, shape, and material of the workpiece also influence the distortion, stress, and
cracking.
Effect of Quenchant Selection

The selection of quenchant is the most basic
factor affecting the cooling characteristics of
workpieces. Therefore, it is the basic factor to be
considered for stress and distortion control during quenching. The selection of a particular
quenchant depends on the quench severity desired. For example, water, brine, or lower concentrations of aqueous polymer solutions are
used for plain carbon steels. Accelerated oils are
used for lower-alloy steels. Conventional oils or
higher concentrations of polymers are used for
higher-alloy steels. Molten salts or liquid metals
are often used for martempering (marquenching)
and austempering.
The dimensions and shape of the workpiece
being quenched should also be considered in selecting a quenchant. Generally speaking, the
thicker the workpiece, the more severe the quenchant. However, severe quenching often increases stress and distortion of the quenched
workpiece. For steel parts with thick and thin

cross sections, the selection of a quenchant is
more difficult. Many such shapes increase the
nonuniformity of cooling and therefore increase
the potential for stress, cracking, and distortion.
Wetting during quenching in a volatile quenchant, such as water, oil, or aqueous polymer
solutions, results in nonuniform (uneven) cooling of the workpieces producing high surface
thermal gradients and often increasing distortion
and stress. Many aqueous polymer quenchants
will provide more uniform wetting properties,
which will result in substantial reductions in
cracking and distortion (Ref 68).
Figures 92 and 93 show the distortion and residual stress of 30 mm diam, and 10 mm thick
carbon steel disk specimens quenched in various
quenchants without agitation. Different stress
distributions and distortions were obtained for
each quenchant. This is a result of the difference
of the cooling power of each quenchant, which
dominates the cooling path on the CCT curve
and therefore the internal distribution of martensite and ferrite/pearlite. The wetting process on
the surface of the specimen during water, polymer, and oil quenching also affects the stress and
distortion. After vapor blanket cooling, a collapse of the vapor blanket (i.e., wetting) occurs
progressively during water quenching. This results in a nonuniform cooling of a steel specimen
and increases stress and distortion. However, if
the vapor blanket collapses simultaneously or
explosively, as in polymer quenching, the simultaneous collapse provides uniform quenching, which is effective for reducing stress and
distortion. The results shown in Fig. 92 and 93
illustrate the effectiveness of uniform quenching
using a polymer quenchant.
Molten-salt and liquid-metal quenching also
provide uniform quenching, decrease cooling
rate, and nonuniformity of the temperature of a
steel part because of the high quenchant temperature. These characteristics are effective for
the reduction of stress, distortion, and cracking.
The cooling rate of a workpiece being

quenched shows an inverse relationship with respect to the increasing thickness of the workpiece by mass effect. In addition, the cooling
rates of the core are limited by the thermal diffusion in the workpiece. Therefore, stress and
distortion during quenching are affected by the
dimensions, shape, and material of the workpiece being quenched.
Figure 94 (Ref 78) shows the axial stress development during water quenching of AISI 1045
steel cylinders. The 10 mm (0.4 in.) diam cylinder starts to transform to martensite at the surface, and the transformation front moves gradually inward, resulting in a typical tensile stress
at the surface. The large-diameter cylinders first
transform to ferrite/pearlite at intermediate radii
and then to martensite at the surface. This causes
two stress minima, seen in the dashed curves in
Fig. 94(a) to (c). The final residual stress is compressive at the surface and tensile in the core.
The relationship of stress to specimen diameter
and quenching medium is summarized in Fig. 95
(Ref 78, 79). The difference between oil and water quenching decreases with increasing diameter.
Effect of Agitation
As discussed in the section Effect of Materials and Quench Process Design on Distortion
in this article, quench nonuniformity can arise
from nonuniform flow fields around the part surface during the quench or nonuniform wetting of
the surface. In addition, poor agitation design is
a major source of quench nonuniformity. The
purpose of the agitation is not only to increase
cooling power of quenchant, but also to provide
uniform cooling in order to suppress the excessive distortion and stress of quenched steel parts.
Nonuniformity of the surface cooling during
water quenching is decreased significantly by the
30
Still
0.3 m/s
0.7 m/s
Starting (t s ) and finishing (t f )

time of wetting, s
Wetting length (Lf ), mm
Workpiece Size Effects
20
10
Lf
t s : Starting time of wetting

t f : Finishing time of wetting
tf
1
ts
0
0
0
Time (t ), s
(a)
Fig. 74
0.3
0.7
Flow velocity (v ), m/s

(b)
Effect of (a) agitation and (b) flow velocity on wetting process on the surface of stainless steel disk specimen (30 mm diam by 10 mm thick) during water quenching.
Quenchant: 30 C city water. , diameter. Source: Ref 68
10
Time (t ), s
20
600
TB (surface: B)
400
400
TC (surface: C)
TD (edge: D)
200
200
10
Time (t ), s
20
800
TA T B
TB T C
TB T D
Temperature (T ), C
TA (center)
600
Lf
600
TB (surface: B)
400
400
TC (surface: C)
TD (edge: D)
200
200
10
Time (t ), s
20

time of wetting, s
TA (center)
800
TA T B
TB T C
TB T D
TA (center)
600
800
T A TB
T B TC
T B TD
600
TB (surface: B)
400
400
TC (surface: C)
TD (edge: D)
200
200
10
Time (t ), s
20
(b)

800
tf
TA (center)
600
T A TB
T B TC
T B TD
800
600
TB (surface: B)
TC (surface: C)
400
400
TD (edge: D)
200
200
ts
20
Time (t ), s
(a)
Temperature (T ), C
10
15 400
7
7
1 200
10
Time (t ), s
800
600
600
(a)
Still
Still
0.3 m/s
0.7 m/s
20
(b)
800
200
30
(a)
800
400
TC (surface: C)
TD (edge: D)
A
B
C
D
200
400
15
7
7 200
1
600
800
TD (edge: D) A
B
C
D
400
TB (surface: B)
TA TB
TB TC
TB TD
0.3
0.7
10
Time (t ), s
20
TC (surface: C)
TA (center)
800
Temperature (T ), C
600
Figures 100 and 101 show the effect of agitation of quenchants on the residual stresses on
the side surface of 20 mm diam, 60 mm long
carbon steel bars quenched in water and a polymer quenchant. Figure 100 shows that nonuniform surface cooling in still water quenching results in uneven residual stress distribution on the
surface of the steel bar. Agitation of water results
in even stress distributions except near the ends.
In addition, the submerged and open spray cooling result in high compression stresses. Figure
101 shows the effect of agitation on stress distribution after polymer quenching. Agitation of
polymer quenchant results in even stress distribution and high compression stresses except
near the ends. However, still polymer quenching
can also make even and high compression stress
Temperature (T ), C
Temperature (T ), C
TB (surface: B)
800
TA TB
TB TC
TB TD 600
of the vapor blanket collapse on the surface.

However, agitation of polymer quenchant does
not decrease the quench distortion (Fig. 97) because the instantaneous and explosive collapse
of the vapor blanket on the specimen surface occurs with and without the agitation.
Figures 98 and 99 show the effect of agitation
methods of quenchants on the quench distortion
of 20 mm in diam, 60 mm long carbon steel bars
quenched in water and a polymer quenchant.
Figure 98(a) shows that nonuniform surface
cooling in still water quenching results in uneven
diameter of the steel bar. The increases of diameters near the ends of bars that were observed
are attributable to heat extraction from the edges
of the bar by edge effect. The upward flow of
water decreases end effect, because the agitation
reduces the nonuniformity of the surface cooling
of the steel bar. The lateral submerged and open
spray decrease the diameter and increase the
length of steel bars, because the lateral flow
causes the fast cooling of the side surface and
the thermal shrinkage of diameter, which also
results in the elongation in length of the steel bar
(Fig. 98b). On the other hand, agitation of polymer quenchant hardly affects the quench distortion (Fig. 99) because the instantaneous and explosive collapse of the vapor blanket on the
specimen surface definitely occurs with and
without the agitation.
Temperature (T ), C
TA (center)
800
agitation (as shown in Fig. 75). Therefore, stress

and distortion are largely affected by agitation.
However, polymer quenchant agitation hardly
affects the degree of nonuniformity of the surface cooling (as shown in Fig. 77) because of the
explosive collapse of vapor blanket.
Figures 96 and 97 show the effect of agitation
of quenchants on the profile of the flat surface of
the steel disk after quenching. Figure 96 shows
that nonuniform surface cooling in still water
quenching results in concaving of the steel disk
specimen and the agitation of water significantly
decreases the quench distortion in water quenching. This occurs because agitation reduces the
nonuniformity of the surface cooling of the steel
disk since agitation accelerates the propagation
(c)
(b)
(c)
Effect of agitation on cooling curves and temperature differences of stainless steel disk specimen during water quenching. Quenchant: 30 C city water. Flow velocity: (a) still, (b) 0.3 m/s, (c) 0.7 m/s. Source:
Ref 68
Effect of (a) agitation and (b) flow velocity on

wetting process on the surface of stainless steel
disk specimen (30 mm diam by 10 mm thick) during polymer quenching. Quenchant: 30 C 10% PAG polymer
aqueous solution
Effect of agitation on cooling curves and temperature differences of stainless steel disk specimen during polymer quenching. Quenchant: 30 C 10%
PAG polymer aqueous solution. Flow velocity: (a) still, (b)
0.3 m/s, (c) 0.7 m/s
Fig. 75
Fig. 76
Fig. 77
A
B
C
D
200
400
200
15
7
7
1
10
Time (t ), s
20
600
800
TB (surface: B)
TC (surface: C)
TD (edge: D)
400
600
400
200
200
10
Time (t ), s
20
400
600
T A TB
T B TC
T B TD
TA (center)
800
Temperature (T ), C
800
TB (surface: B)
TC (surface: C)
TD (edge: D)
600
TA T B
TB T C
TB T D
TA (center)
800
Temperature (T ), C
because the uniform cooling occurs with and

without the agitation.
Table 12 shows the effect of agitation on the
frequency of quench cracking in water and polymer quenching of steel disks, the geometry and
dimensions of which are shown in Fig. 102. The
specimen in this case (30 mm in diameter by 10
mm thick) did not contain any notches; however,
it contained an eccentrically located 10 mm diam
hole. This specimen, used in the work of Owaku
(Ref 1), was adopted by the Quench Cracking
Working Group of the Japan Heat Treatment Society. Steel materials used were Japanese standard S45C, SK4, and SCM435. These results
show that agitation of water largely suppresses
the occurrence of quench cracking. On the other
hand, quench cracking susceptibility to agitation
of polymer quenchant is not clear because there
is no crack on the polymer-quenched specimen
with or without agitation.
(b)
(a)
Fig. 79
Effect of agitation on cooling curves and temperature differences of stainless steel disk specimen during oil
quenching. Quenchant: 80 C quenching oil (JIS 1-2 oil). Flow velocity: (a) still, (b) 0.3 m/s
Effect of Surface Condition
30
800
Still
0.1 m/s
0.2 m/s
0.3 m/s
20
Stainless steel
cylinder
Nickel
cylinder
Temperature (T ), C
Effect of Surface Texture and Roughness.

As already described in the section Effect of
Process Variables on Cooling Behavior and Heat
Transfer in this article, the surface texture and
roughness affect the cooling characteristics of
the steel workpiece in quenching. Therefore,
these factors affect the stress and distortion of
600
400
Silver
cylinder
3
2
200
2
1
3
0
0
2
1
3
20
10
30
40
50
60
70
Time (t ), s
Effect of surface oxidization on cooling curves of cylindrical silver, nickel, and stainless steel specimens in
water quenching. Quenchant: 30 C still water. Specimen shape: cylinder with hemispherical ends (10 mm
diam by 30 mm long). Heating conditions: 1, 5 min up to 800 C in argon; 2, 5 min up to 800 C in air; 3, 8 h at 800 C
in air. Source: Ref 70, 71
Fig. 80
10
Lf
0
Time (t ), s
(a)
800
50 C
tf
Temperature (T ), C

time of wetting, s

2
1
ts
600
Water temperature = 95 C
5 min
in argon
400
10 min
in air
200
5 min
in argon
10 min
in air
0
0
0.1
0.2
0.3
0
0
(b)
10
20
30
40
50
60
70
80
90
100
Time (t ), s
Effect of (a) agitation and (b) flow velocity on
wetting process on the surface of stainless steel
disk specimen (30 mm diam by 10 mm thick) during oil
quenching. Quenchant: 80 C quenching oil (JIS 1-2 oil)
Fig. 78
Effect of surface oxidization on cooling curves of cylindrical pure iron specimens in water quenching. Quenchant: 50 and 95 C still water. Specimen shape: cylinder with hemispherical ends (10 mm diam by 30 mm
long). Source: Ref 70, 71
Fig. 81
Heat-transfer coefficient, W/m2 K 103

the workpiece quenched. In addition, the surface
texture and roughness are very important factors
for quench cracking because the microscopic geometry and roughness of the surface affect the
tendency to crack. Table 13 shows an example
of such a case (Ref 80). The results are:
20
5 m
10 m
15
50 m
10
Surface roughness increases the tendency for
Water flux = 0.6 m3/m2 min
0
100
300
200
400
500
600
700
800
Fig. 82
Effect of surface scale on heat-transfer coefficient during spray cooling of a hot steel plate. Source: Ref 72
1000
Flaking of scale
Temperature (T ), C
800
Fall of scale flake
600
Fall of scale flake
400
Flaking of scale
Heavy scale
200
No oxide scale
0
0
10
Time (t ), s
Fig. 83
Unstable cooling phenomena due to surface oxidization during water quenching of steel specimen. Quenchant: 30 C still water. Specimen: JIS S45C cylinder (10 mm diam by 30 mm long)
quench cracking of steel if surface roughness

(maximum height of irregularities Ry, or 10
points height of irregularities Rz) is larger
than approximately 1 lm.
Surface texture made by lapping tends to
cause a higher occurrence of quench cracking
than by grinding or emery polishing if each
roughness is almost similar.
This phenomenon is caused mainly by the stress
concentration at the surface of the steel workpieces. Geometric shapes on the surface such as
polishing marks, lapping marks, grinding marks,
cutting tool marks, micronotches, and so forth
act as a stress riser and a trigger for inducing
quench cracking.
Effect of Oxide or Coating Layer. The presence of a thin layer such as oxide scale and clay
coating, for example, causes a cooling acceleration effect by suppression of vapor blanket formation or by acceleration of the collapse by the
existence of a thin coating layer (as mentioned
in the section Effect of Process Variables on
Cooling Behavior and Heat Transfer). In addition, uniform cooling is caused by the existence
of a thin layer. Therefore, quench cracking is
suppressed by a thin scale or coating layer. Tables 14 and 15 show that the existence of oxide
scale or clay coating largely suppresses the occurrence of quench cracking. However, a heavy
oxide scale of the steel workpiece often causes
unstable cooling and decarbonization.
Minimizing Distortion
Distortion can be controlled by appropriate selection of component design, steel grade, quenchant, agitation, and quenching method.
800
1000
Without coating
800
600
Temperature (T ), C
Temperature, C
Without coating
600
Clay coating
15
5 m
mthick
thick
400
15 m
40 m
200
25 m
80 m
Clay coating thickness
100 m
400
10
Time, s
Fig. 84
200 m
25 m
50 m
0
0
0
0
150 m
10 m
200
Effect of clay coating on cooling curves of steel specimen in water quenching. Quenchant: 30 C still water.
Specimen: JIS S45C steel cylinder (10 mm diam by 30 mm long)
2
Time (t ), s
Effect of clay coating on cooling curves of silver specimen in water quenching. Quenchant:
30 C still water. Specimen: silver cylinder (10 mm diam
by 30 mm long)
Fig. 85

Component Design
Heat-transfer coefficient, W/m2 K 103
20
Water flux = 0.6 m3/m2min
15
13.2
9.5
7.5
5.5
1.8
1.0
10
One of the causes of unacceptable distortion

and cracking of steel parts is component design.
Poor component design promotes distortion and
cracking by accentuating nonuniform and nonsymmetrical heat transfer during quenching.
Component designs that minimize distortion and
cracking include the following.
Design Symmetry. It is important to provide
greater symmetry. One of techniques for design
symmetry is to add dummy holes, key grooves,
or other shapes to steel parts.
Balance of Cross-Sectional Area. The difference between a large cross-sectional area and
a thin one can be decreased by:
60
120
1000
0
100
Fig. 86
Surface roughness
Rz,
Paper
m
No.
300
200
400
500
Adding dummy holes to large cross-sectional
Effect of surface roughness on heat-transfer coefficient during spray cooling of a hot copper plate. Source:
Ref 72
areas
Changing from blind holes to through-type
holes
Changing from thick solid shapes to thin hol-
low shapes
Dividing a complicated shape into sectional
components
Emery No. 800
Emery No. 320
Temperature (T ), C
800
Avoiding Sharp Corners and Edges. Distortion and cracking are encountered when quenching a part with sharp corners and edges, which
increase cooling nonuniformity and act as stress
risers. Therefore, it is effective to round corners
and edges or to employ a tapered shape.
Emery No. 240

Lapping C No. 800
600
Lapping C No. 320

400
Steel Grade Selection

200
Although quench distortion and cracking are

most often due to nonuniform cooling, material
selection can be an important factor. Some attention should be paid to material selection:
0
0
4
5
Time (t ), s
10
The compositional tolerances should be
1000
800
800
Temperature (T ), C
1000
Temperature (T ), C
Fig. 87
Effect of surface finish on the cooling curves of a S45C steel specimen during polymer quenching. Quenchant:
30 C 15% polymer (PAG ) quenchant. Specimen: JIS S45C steel cylinder (10 mm diam by 30 mm long)
600
400
R0
0.2 m/s
600
JIS
200
C1
200
R5
R3
1.0 m/s
R3
R0
JIS
C1
R5
400
R3
C1
JIS
Cracking propensity increases as the Ms temperature and the carbon equivalent increase.
Quench cracks were prevalent at carbon equivalent values above 0.525, as shown in Fig. 103.
R5
R0
Selection of Quenchant and Agitation
0
0
(a)
10
Time (t ), s
15
20
8 10 12
Time (t ), s
14 16 18
20
(b)
Effect of specimen shape on the cooling curves of cylindrical silver specimen cylinder (10 mm diam by 30
mm long) during quenching in 30 C water. Specimen shape: JIS, JIS silver probe type; R0, with flat ends and
sharp edges; R3, with rounded edges of 3 mm radius; R5, with hemispherical ends of 5 mm radius; C1, with chamfered
edges of 1 mm by 45
Fig. 88
checked to ensure that the alloy is within

specification.
It is often better to choose a lower carbon
content, because the higher carbon content often causes the higher susceptibility for distortion and cracking (as mentioned in the section Effect of Materials and Quench Process
Design on Distortion).
If possible, it is better to choose a combination of a high-alloy steel and a very slow
cooling rate. As a matter of course, the selection of high-alloy steels markedly raises the
material cost.
Quenchants must be selected to provide cooling rates capable of producing acceptable microstructure in the section thickness of interest.
However, it is not desirable to use quenchants
with excessively high heat removal rates. Typically, the greater the quench severity, the greater
Quenching Methods
Part design, material selection, quenchant selection, and so forth. are the most important factors to suppress the quench distortion and cracking of steel parts. In addition, several methods
for minimizing distortion and eliminating cracking are employed, for example, interrupted
quenching, press quenching, plug quenching,

and so forth.
Interrupted quenching refers to the rapid
cooling of the steel parts from the austenitizing
temperature to a temperature where it is held for
a specified time, followed by slow cooling.
There are several different types of interrupted
quenching: austempering, marquenching (martempering), time quenching, and so forth. The
temperature at which the quenching is interrupted, the length of time the steel is held at temperature, and the cooling rate can vary, depending on the type of steel and workpiece thickness.
Marquenching (martempering) is a term that
describes an interrupted quenching from the austenitizing temperature of steels. The purpose is
to delay the cooling just above the martensitic
transformation temperature until temperature
equalization is achieved throughout the steel
part. This will minimize the distortion, cracking,
and residual stress. The term martempering is
somewhat misleading, and the process is better
described as marquenching. Figures 104(a) and
(b) (Ref 81) show the significant difference between conventional quenching and marquenching. The marquenching of steel consists of:
1. Quenching from the austenitizing temperature into a hot quenching medium (hot oil,
molten salt, molten metal, or a fluidized particle bed) at a temperature usually above the
martensite range (Ms point)
2. Submerging in the quenching medium until
the temperature throughout the steel workpiece is substantially uniform
3. Cooling (usually in air) at a moderate rate to
prevent large differences in temperature between the outside and the core of the workpiece
The formation of martensite occurs fairly uniformly throughout the workpiece during the
cooling to room temperature, thereby avoiding
excessive amounts of residual stress. Straightening or forming is also easily accomplished
upon removal from the marquenching bath while

the part is still hot. Marquenching can be accomplished in a variety of baths including hot oil,
molten salt, molten metal, or a fluidized particle
bed. Cooling from the marquenching bath to
room temperature is usually conducted in still
air. Deeper-hardening steels are susceptible to
cracking while martensite forms if the cooling
rate is too rapid. Marquenched parts are tempered in the same manner as conventionally
quenched parts. The time lapse before tempering
is not as critical because the stress is greatly reduced.
The advantage of marquenching lies in the reduced thermal gradient between surface and core
as the part is quenched to the isothermal temperature and then is air cooled to room temperature. Residual stresses developed during marquenching are lower than those developed
during conventional quenching because the
800
Temperature (T ), C
the propensity to cause increased distortion or

cracking. Although a reduction of quench severity leads to reduced distortion, it may also be
accompanied by undesirable microstructures.
Therefore, it is difficult to select an optimal
quenchant and agitation. Cooling power of quenchant should be as low as possible while maintaining a sufficiently high cooling rate to ensure
the required microstructure, hardness, and
strength in critical sections of the steel parts.
Specific recommendations for quench media selection for use with various steel alloys and general comments regarding quenchant selection
were already shown in the section Effect of Materials and Quench Process Design on Distortion.
Nonuniformity of cooling is perhaps the greatest contributor to quench distortion and cracking.
Nonuniform cooling can arise from nonuniform
flow fields around the part surface during
quenching or nonuniform wetting of the surface.
Nonuniform quenching creates large thermal
gradients between the core and the surface of a
steel part, or among the surfaces of the parts.
These contributing factors, agitation, and surface
wetting should be considered to select a quenchant and an agitation method.
Good designs of agitation, racking, and support fixturing are major factors of uniform
quenching. The purpose of the agitation system
is not only to take hot fluid away from the surface of steel parts, but it is also to provide uniform heat removal over the entire cooling surface of all of the parts throughout the load being
quenched.
600
R3
R5
JIS
R0
400
C1
200
0
0
6
8
Time (t ), s
10
12
Effect of specimen shape on the cooling curves

of cylindrical silver specimen cylinder (10 mm
diam by 30 mm long) during quenching in 80 C still oil.
Specimen shape: JIS, JIS silver probe type; R0, with flat ends
and sharp edges; R3, with rounded edges of 3 mm radius;
R5, with hemispherical ends of 5 mm radius; C1, with
chamfered edges of 1 mm by 45
Fig. 90
Still
600
C1
400
R5
JIS
200
1.0 m/s
0.2 m/s
800
Temperature (T ), C
Temperature (T ), C
800
600
R0
C1
R5
R3
JIS
R0
R0
0
R3
0
0
(a)
C1
R5
JIS
400
200
R3
Heat-transfer coefficient (h), W/m2 K
104
8 10 12
Time (t ), s
14 16 18
20
(b)
8 10 12
Time (t ), s
Stainless steel
cylinder
Nickel cylinder
Silver cylinder
2000
1000
500
200
100
50
14 16 18
20
Effect of specimen shape on the cooling curves of cylindrical silver specimen cylinder (10 mm diam by 30
mm long) during quenching in 30 C 15% polymer (PAG) quenchant. Specimen shape: JIS, JIS silver probe
type; R0, with flat ends and sharp edges; R3, with rounded edges of 3 mm radius; R5, with hemispherical ends of 5 mm
radius; C1, with chamfered edges of 1 mm by 45
Fig. 89
5000
200
400
600
Tl = TW Tl, (K)
800
Effect of specimen material on the heat-transfer

coefficients during oil quenching. Quenchant:
80 C still quenching oil (JIS 1-1 type). Specimen: cylindrical silver, nickel, and stainless steel specimen with hemispherical ends (10 mm diam by 30 mm long)
Fig. 91
Change of thickness (t/2), mm
greatest thermal variations occur while the steel

is in the relatively plastic austenitic condition
and because the final transformation and thermal
changes occur throughout the part at approximately the same time. Marquenching also reduces susceptibility to cracking.
Modified marquenching differs from standard
marquenching only in that the temperature of the
quenching bath is below the Ms point (Fig. 104c)
(Ref 81). The lower temperature increases the
severity of quenching. This is important for
steels of lower hardenability that require faster
cooling in order to harden to sufficient depth.
Therefore, modified marquenching is applicable
to a greater range of steel compositions than is
standard marquenching.
Austempering consists of rapidly cooling the
steel part from the austenitizing temperature to
a temperature above that of martensite formation, holding at a constant temperature to allow
isothermal transformation, followed by air cooling. The steel parts must be cooled fast enough
so that no transformation of austenite occurs during cooling and then held at bath temperature
long enough to ensure complete transformation
of austenite to bainite. Molten-salt baths are usually the most practical for austempering applications. Parts can usually be produced with less
dimensional change by austempering than by
conventional quenching and tempering. In addition, austempering can decrease the likelihood
of cracking and can improve ductility, notch
toughness, and wear resistance. However, austempering is applicable to a limited steel and
parts size. Important consideration for the selection of steel parts for austempering are:
The location for the nose of the TTT curve
and the speed of the quenching
The time required for complete transforma-
tion of austenite to bainite at the austempering

temperature
The location of the Ms point
Maximum thickness of section that can be
austempered to a fully bainitic structure
Carbon steels of lower carbon content will be
restricted to a lesser thickness. For 1080 steel,
the maximum section thickness is about 5 mm
(0.2 in.). Low-alloy steels are usually restricted
to about 10 mm (38 in.) or thinner sections, while
more hardenable steels can be austempered in
sections up to 25 mm (1 in.) thick. Nevertheless,
sections of carbon steel thicker than 5 mm (0.2
in.) are regularly austempered in production
when some pearlite is permissible in the microstructure (Ref 82).
Time quenching is an interrupted quenching
method and is used when the cooling rate of the
part being quenched needs to be abruptly
changed during the cooling cycle. The usual
practice is to lower the temperature of the part
by quenching in a medium with high heat removal characteristics (for example, water) until
the part has cooled below the nose of the TTT
0.05
0.025
0
0.025
0.05
10
15
(Left side)
10
Radius horizontal (l s ), mm
15
(Right side)
Water (30 C, still)

10% PAG polymer solution (30 C), still)
Oil (80 C, still)
t = 10 mm
ds
ls
o
o
2r = 30 mm dia.
ls
Distance from the center

of thickness ds, mm
(a)
5
2.5
0
2.5
5
0.1 0.05 0 0.05 0.1
Change of radius (r), mm
(b)
Fig. 92
distortion
Effect of quenchants on quench distortion of JIS S45C carbon steel disk quenched in still quenchants. Specimen
dimensions were 30 mm diam by 10 mm thick. (a) Distribution of axial distortion. (b) Distribution of radial
curve and then to transfer the part to a second

medium (for example, oil, air, an inert gas), so
that it cools more slowly through the martensiteformation range. Time quenching is most often
used to minimize distortion, cracking, and dimensional changes.
Restraint quenching refers to the quenching
under restraint of distortion of steel, for example,
quenching by using restraint fixtures, press
quenching, cold-die quenching, plug quenching,
and so forth. There are many round, flat, or cylindrical parts that distort to an unacceptable degree by conventional immersion quenching. Under such conditions, it is necessary to resort to
special techniques. However, the equipment for
those special techniques are expensive and production rates are slow. Consequently, the resulting cost of heat treatment is relatively high.
Therefore, use of these techniques should be
considered only when minimal distortion is mandatory.
Restraint fixturing is costly and is used primarily for highly specialized applications. Representative examples are the quenching of rocket
and missile casings or other large components
with thin wall sections.
Press Quenching and Plug Quenching. Probably the most widely used special technique is
press quenching. To minimize distortion caused
by the quenching cycle, press and plug quenching dies must be made to provide the proper
quenchant flow and hold critical dimensions of
the part being quenched. In quenching, the die
or plug contacts the heated part, and the pressure
of the press restrains the part mechanically. This
occurs before quenching begins, while the part
is hot and plastic. The machine and dies then
force the quenching medium into contact with
the part in a controlled manner.
Cold-Die Quenching. Thin disks, long thin
rods, and other delicate parts that distort excessively by conventional immersion quenching can
often be quenched between cold dies with no
distortion. Cold-die quenching is limited to parts
with a large surface area and small mass, such
as washers, rods of small diameter, thin blades,
and so forth. Large, thin thrust washers have to
be flat after quench hardening, but considerable
distortions are developed as a result of blanking
and machining stresses. To ensure the required
flatness, the washers are squeezed between a pair
of water-cooled dies immediately after they
leave the furnace.
Other quenching techniques that can be
used to minimize cracking and distortion are discussed here. Selective quenching is used when it
is desirable for certain areas of a part to be relatively unaffected by the quenching medium. This
can be accomplished by insulating an area to be
more slowly cooled so the quenchant contacts
only those areas of the part that are to be rapidly
cooled. The clay-coating method is employed for
selective quenching, for example, water quenching of Japanese swords. The distortion and microstructure of Japanese swords is controlled by
the thickness distribution of the clay layer. A
thick clay layer (0.1 mm) suppresses the cool-

ing rate by its insulating effect, and thin clay
layer increases the cooling rate during water
quenching by its cooling accelerating effect (Fig.
84 and 85). Selective quenching is often effective to suppress excessive distortion or cracking.
Spray quenching involves directing high-pressure streams of liquid onto the surfaces of steel
parts. Spray quenching is often useful for minimizing distortion and cracking, because it can
realize a uniform quenching by selecting optimal
spray conditions and can be used where higher
cooling rates are desired. The cooling rate can
be faster because the quenchant droplets formed
by the high-intensity spray impact the part surface and remove heat very effectively.
Fog quenching utilizes a fine fog or mist of
liquid droplets in a gas carrier as the cooling
agent. Although similar to spray quenching, fog
quenching produces lower cooling rates because
of the relatively low liquid content of the stream.
Intensive quenching (Ref 83) involves forcedconvection heat exchange both during high-temperature cooling and also during the transformation of austenite into martensite. Intensive
quenching promotes temperature uniformity
during cooling through the cross section of the
part. Cooling intensity should be sufficient to
promote maximum surface compressive stresses.
This occurs when the Biot number is 18. The
second criterion of intensive quenching is that
the intensive cooling be stopped at the moment
that maximum surface compressive stresses are
formed.
Tempering
Steel parts are often tempered by reheating after quench hardening to obtain specific values of
mechanical properties. Tempering of steel in-
Cylinder diameter, in.
Effect of quenchants on residual stresses on side surface of JIS S45C carbon steel disk quenched in still
quenchants. Specimen dimensions were 30 mm diam by 10 mm thick. (a) Circumferential stress on end
surface. (b) Radial stress on end surface
Fig. 93
800
0.4
1.2
2.0
4.0 115
Water quench, core

85
600
400
60
200
30
Oil quench, surface
Axial stress distribution during water quenching for various AISI 1045 steel cylinders diameter D, at selected
times (in seconds) after the start of quenching from 850 C (1560 F) in 20 C (70 C) water. The final microstructure of the 10 mm (0.4 in.) diam cylinder is completely martensite, while the others have a ferritic-pearlitic core.
Source: 78, 79
30
200
60
400
Water quench, surface
600
85
800
115
1000
Fig. 94
10
30
50
Cylinder diameter, mm
100
Axial residual stress, ksi
Axial residual stress, MPa
Oil quench, core
145
Dependence of axial residual stresses on cylinder diameter. Same steel as in Fig. 94. The
core is martensite for 10 mm (0.4 in.) diameter, but is
ferrite-pearlite for larger diameters. Source: Ref 78, 79
Fig. 95
100 m
(a)
(b)
Fig. 96
creases ductility and toughness of quench-hardened steel and also relieves quench stresses and
ensures dimensional stability.
As mentioned in the section Phase Transformation during Heat Treating, the tempering
process is divided into four stages:
100 m
100 m
(c)
Effect of agitation of water on quench distortion of JIS S45C carbon steel disk. Specimen dimensions were 30
mm diam by 10 mm thick. Quenchant was 30 C city water. Flow velocity: (a) still water, (b) 0.3 m/s, and
(c) 0.7 m/s
100 m
100 m
(a)
100 m
(b)
(c)
Effect of agitation of polymer quenchant on quench distortion of JIS S45C carbon steel disk. Specimen dimensions were 30 mm diam by 10 mm thick. Quenchant was 30 C 10% polymer (PAG) quenchant. Flow
velocity: (a) still polymer solution, (b) 0.3 m/s, and (c) 0.7 m/s
Fig. 97
1. Tempering of martensite structure

2. Transformation of retained austenite to martensite
3. Tempering of the decomposition products of
martensite
4. The decomposition of retained austenite to
martensite.
These microstructural variations during tempering result in volume changes during the tempering of hardened steel (Ref 84). Figure 105 shows
an example of dimensional changes of tool steel
plates by tempering with different soak times.
Tempering up to 200 C (390 F) is accompanied
by a slight contraction in all directions of the
plate. At a higher tempering temperature, there
is an increase in the dimensions, with a maximum increase at 300 C (570 F), after which
dimensions again decrease. The increased vol-
Water, 30 C
Still
Upward (0.3 m/s)
Upward (0.7 m/s)
Submerged spray
Open spray
0.05
0.05
0
0.4
0.2
0.2
20
40
Rate of diameter change, %
Change of diameter, mm
0.1
Open spray
(lateral)
Submerged
spary (lateral)
Upward
(0.7 m/s)
Upward
(0.3 m/s)
Still
60
0.05
Distance from lower end, mm
0.15
0.20
0.25
0.30
0.35
Change of length, mm
(a)
Fig. 98
0.10
(b)
Effect of agitation methods on distortion in water quenching of JIS S45C steel rod (20 mm diam by 60 mm long). Quenchant was 30 city water. Agitation methods were
still, 0.3 m/s upward flow, 0.7 m/s upward flow, and lateral submerge in immersion quenching, and lateral open spray quenching in air. (a) Change of diameter. (b) Change
of length
0.05
10% polymer quenchant, 30 C

Still
Upward (0.3 m/s)
Upward (0.7 m/s)
Open spray (lateral)
0.05
0
0.4
0.2
0.2
20
40
Rate of diameter change, %
Change of diameter, mm
0.1
Submerged
spray (lateral)
Upward
(0.7 m/s)
Upward
(0.3 m/s)
Still
60

(a)
Fig. 99
0.05
0.10
0.15
0.20
0.25
0.30
Change of length, mm
(b)
Effect of agitation methods on distortion in polymer quenching of JIS S45C steel rod (20 mm diam by 60 mm long). Quenchant was 30 C 10% polymer (PAG) quenchant.
Agitation methods were still, 0.3 m/s upward flow, and 0.7 m/s upward flow in immersion quenching. (a) Change of diameter. (b) Change of length

ume at 300 C (570 F) is attributed to the transformation of retained austenite to bainite. At 400
C (750 F), the dimensions revert to values
closer to the original values, prior to quenching
and tempering.
In addition to dimensional change by microstructural variation, tempering may also lead to
dimensional variation due to relaxation of residual stresses and plastic distortion caused by the
temperature dependence of yield strength (Ref
85). Figure 105 shows the distortion of round
steel bars (200 mm in diameter and 500 mm in
length) by quenching and by stress relieving in
tempering. A medium-carbon steel bar (Fig.
105a to c) and a hardenable steel bar (Fig. 105d
to f) are used in this experiment. Figures 105a
and d are the results of quenching from 650 C
without phase transformation. The distortion in
each case is almost the same, regardless of the
different quenchants and the different chemical
compositions. These convex distortions are

caused by nonuniform thermal contraction and
resultant thermal stress during cooling. Figures
105(b) and (e) are the results of quenching from
850 C with phase transformation. The distortion
in Fig. 105(e) (hardenable steel) shows a convex
configuration, but the distortion in Fig. 105(b)
(medium-carbon steel of poor hardenability)
shows a configuration that combines convex and
concave distortion. In addition, water quenching
(WQ) has a greater effect on distortion than oil
quenching (OQ). Figures 105(c) and (f) show the
configurations after tempering. These results
show that tempering after quenching results in
not only volumetric changes but also in convex
distortions. Such distortions seem to be related
to relieving residual stresses by tempering.
Figures 106 and 107 (Ref 86) show an example of stress relief by tempering. A solid cylinder 40 mm in diameter and 100 mm in length
60
4
10
10
A
20
B
600
400
200
0
200
400
600
800
1000
0
Still water
0.3 m/s upward
0.7 m/s upward
10
Spray (submerged) lateral

Spray (open) lateral
20
30
40
50
60
(a)
600
400
200
0
200
400
800
1000
0
Still water
0.3 m/s upward
0.7 m/s upward
10
The computer prediction of thermomechanical

behavior in quenching is very useful for the determination of a quenching condition, because it
is possible to predict the microstructure, hardness, distortion, and stress of steel parts after
quenching. In order to predict thermomechanical
behavior, residual stresses, and distortion in
quenching of steel parts, a simulation method
based on metallothermomechanical theory (Ref
87, 88) and finite-element analysis is often applied. The metallothermomechanical theory is
coupled with temperature, phase transformation,
and stress/strain fields. In quenching, the fields
of metallic structures and stress/strain as well as
temperature affect each other. The triangular diagram shown in Fig. 108 shows such coupling
effect (Ref 88). In carburizing-quenching process, the effect of carbon content on the three
fields is also represented by the dashed lines in
the figure. A series of theoretical models with
consideration of the effect of carbon diffusion
and distribution is introduced.
An example of introducing the governing

equations in the framework of thermomechanical behavior for describing temperature and
stress/strain fields incorporating metallic structures in the quenching process is summarized in
this section (Ref 8891).
Mixture Rule. When a material point undergoing heat treatment is assumed to be composed
of a multiphase structure, an assumption is made
that a material parameter is described by the
mixture law:
N
Spray (submerged) lateral

Spray (open) lateral
20
30
40
50
60

(b)
Effect of agitation methods on residual stress after water quenching of JIS S45C steel rod (20 mm diam by
60 mm long). Quenchant was 30 C city water. Agitation methods were still, 0.3 m/s upward flow, 0.7
m/s upward flow and lateral submerge in immersion quenching, and lateral open spray quenching in air. (a) Axial stress
on surface. (b) Tangential stress on surface
Fig. 100
Prediction of Distortion and

Residual Stress after Quenching
Governing Equations
600
was examined for analyses and experiments of

tempering performed after water quenching. Calculated residual stress distributions after water
quenching are shown in Fig. 106; open and solid
circles correspond to measured stresses on the
surface of the cylinder by x-ray diffraction. Residual stress distributions after tempering at 400
C are shown in Fig. 107(a) and (b) for typical
elapsed times of 2 and 50 h with measured values on the surface. These results show that the
stresses in all directions decrease with elapsed
time in tempering.
vI nI and I1
nI 1
I1
(Eq 4)
where nI denotes the volume fraction for the Ith

phase.
Heat-Conduction and Diffusion Equations.
The temperature field is governed by a special
heat-conduction equation coupling stress work
and latent heat due to phase transformation during quenching:

T
qCT
k
rij e ijp
xi
xi
q l n 0
60
4
10
10
(Eq 5)
I I I
where k and lI denote the coefficient of heat conduction and the latent heat produced by the progressive Ith constituent.
The boundary conditions of heat transfer on
the outer surface are assumed to be:
20
B
600
400
Still polymer, 10%

0.3 m/s upward, 10%
Spray (open) 10%

0.7 m/s upward, 10%
200
0
T
ni h(T Tw)
xi
(Eq 6)
where h is a function dependent on temperature,

h and Tw denote the heat-transfer coefficient and
the temperature of coolant on heat-transfer
boundary with unit normal ni, respectively.
Carbon content during carburizing is solved
by the diffusion equation:
200
400
600
800
1000
0
10
20
30
40
50
60
D C
C
xi
xi
(Eq 7)
(a)
where C is content in the position xi direction, D

is the diffusion constant determined by the
boundary condition being specified by the reaction across the surface layer:
600
400
200
Still polymer, 10%

0.3 m/s upward, 10%
Spray (open) 10%

0.7 m/s upward, 10%
0
200
400
600
800
1000
0
10
20
30
40
50
60

(b)
Effect of agitation methods on residual in polymer quenching of JIS S45C steel rod (20 mm diam by 60 mm
long). Quenchant was 30 C 10% polymer (PAG) quenchant. Agitation methods were still, 0.3 m/s upward
flow, and 0.7 m/s upward flow in immersion quenching. (a) Axial stress on surface. (b) Tangential stress on surface
Fig. 101
C
ni hc(Ce Cs)
xi
(Eq 8)
where hc and Ce are the surface reaction rate coefficient and the known content of the external
environment, respectively.
Constitutive Equation. Total strain rate e ij is
assumed to be divided into elastic, plastic, and
thermal strain rates and those by structural dilatation due to phase transformation and transformation such that:
e ij e eij e pij e Tij e m
tp
ij e
ij
(Eq 9)
Here, the elastic strain:

eeij
1 m
m
rij (rkk)dij
E
E
(Eq 10)
Table 12 Effect of agitation on quench cracking in water and polymer quenching of steel
disks shown in Fig. 100
Steel materials are Japanese standard S45C, SK4, and SCM435.
10
Frequency of occurrence of quench cracking, %

Quenchants and its agitation
City water (30 C)

Still (nonagitated)
0.3 m/s upward
S45C
0.45%C-0.67%Mn
SK4
0.98%C0.77%Mn
SCM435
0.35%C-0.76%Mn1.06%Cr-0.20%Mo
100
70
100 (flat surface)

30 (flat surface)
70 (hole surface)
0 (flat surface)
100 (hole surface)
10 (flat surface)
90 (hole surface)
100
100
0.7 m/s upward
5 m/s open spray
10% polymer quenchant (30 C, PAG)

Still (nonagitated)
0%
0.3 m/s upward
0%
0.7 m/s upward
0%
7
10
30
60
0
Disk specimen for quench-cracking test.
Specimen dimensions were 30 mm diam by
10 mm thick; specimen contains an eccentrically located
10 mm diam hole.
Fig. 102

with Youngs modulus E and Poissons ratio m.
In the simulation of carburizing-quenching
process, (1 m)/E and m/E are expressed as the
functions of carbon content and volume fraction
of the structure:
1 m
l1
m
E
In the simulation of carburizing-quenching process, is expressed as the function of carbon

content and volume fraction of the structure:
(C C1)
mI2
EI 2
(Eq 12)
N
(Eq 11)
eTij (T T0)dij
(Eq 13)
Table 13 Effect of surface texture and roughness on quench cracking in water quenching
of steel disks shown in Fig. 100
Steel materials are Japanese standard S45C and SK4. Quenchant: water (30 C) agitated by upward nozzle jet (1 m/s)
S45C
0.45%C-0.67%Mn
Surface finishing
and roughness
Grinding (WA No. 46J)

Ry 1.0 lm, Rz 0.75 lm
Emery polishing (No. 800)
Ry 0.15 lm, Rz 0.13 lm
Ry 1.1 lm, Rz 1.0 lm
Lapping (WA No. 3000)
Ry 1.1 lm, Rz 0.88 lm
Lapping (C No. 800)
Lapping (C No. 500)
I {(C2 C)I 1
I1 C2 C1
n
The thermal strain is the function of initial temperature of material T0 and thermal expansion
coefficient :
nI
(C2 C)
C2 C1
1 mI 1
1 mI2
(C C1)
EI 1
EI 2
I (C2 C) EII11
I1 C2 C1
SK4
0.98%C-0.77% Mn
Frequency of
occurrence of
quench cracking, %
Surface finishing
and roughness
Frequency of
occurrence of
quench cracking, %
Grinding (WA No. 46J)

Ry 0.90 lm, Rz 0.75 lm
Ry 0.15 lm, Rz 0.13 lm
20 (flat surface)
80 (hole surface)
0 (flat surface)
100 (hole surface)
20

0 (flat surface)
100 (hole surface)
20
80
Lapping (C No. 500)

100 (flat surface)

0 (hole surface)
30
0
0
60
Ry, maximum height of irregularities; Rz, ten points height of irregularities
(C C1)I 2}
(Eq 14)
The plastic strain rate is reduced to the form by

employing temperature-dependent materials parameters:
e pij k
F
rij
(Eq 15)
F rkl F T
k G
rkl
T
N
F
F
nI
C
C
I1 nI
1
F
F
F
epmn j rmn rmn
G
(Eq 16)
(Eq 17)
with a temperature-dependent yield function:

F F(T,C,rij ,ep,nI,j)
(Eq 18)
where j is the hardening parameter.

Strain rates due to structural dilatation and
transformation plasticity depend on the Ith constituent:
Table 14 Effect of surface oxidation on quench cracking in water quenching of steel disks
shown in Fig. 100
Steel materials are Japanese standard S45C and SK4. Quenchant: water (30 C) agitated by upward nozzle jet (1 m/s)
S45C
0.45%C-0.67%Mn
Surface condition
(heating condition)
SK4
0.98%C-0.77%Mn
Frequency of
occurrence of
quench cracking, %
No-scale (heated in argon)
Surface condition
30
Light scale (heated in air, 3 min)
Heavy scale (heated in air, 20 min)
No-scale
(heated in argon)
Light scale
(heated in air, 3 min)
Heavy scale
(heated in air, 20 min)
Frequency of
occurrence of
quench cracking, %
20 (flat surface)
80 (hole surface)
80 (flat surface)
20 (hole surface)
0
Table 15 Effect of clay coating on quench cracking in water quenching of steel disks shown
in Fig. 100
Steel materials are Japanese standard S45C, SK4, and SCM435.
Frequency of occurrence of quench cracking, %
Surface condition (coating condition)
No coating (bare)
Clay coating (0.3 mm thick)
S45C
0.45%C-0.67%Mn
SK4
0.98%C-0.77%Mn
SCM435
0.35%C-0.76%Mn1.06%Cr-0.20%Mo
100
10
100
0
100
0
Fig. 103
steels
Effect of MS temperature and carbon equivalent on the quench cracking of selected
Fig. 104
Comparison of cooling curves as workpieces pass into and through martensite transformation range for a conventional quenching and tempering process and for interrupted
quenching processes. (a) Conventional quenching and tempering. (b) Marquenching. (c) Modified marquenching
Deformation of medium-carbon and hardenable steel bars by quenching from below and above transformation temperature and by stress relieving. lC, change of length.
(a) and (d) Quenched from 650 C. (b) and (e) Quenched from 850 C. (c) and (f) Tempered at 680 C (a) to (c) JIS S38C steel (0.38%C). (d) to (f) JIS SNCM 439 steel
(0.39C-1.80Ni-0.80Cr-0.20Mo).
Fig. 105
150
150
150
T = 400 C
t =2h
100
T = 400 C
t =
100
50
50

Fig. 106
10
15
Radius (r ), mm
50
50
z
100
100
150
50
z
100
50
Stress (), kg/mm2
Stress (), kg/mm2
Stress (), kg/mm2
100
150
20
Stress distribution in cylinder after quenching
10
15
Radius (r ), mm
20
(a)
Fig. 107
150
(b)
Stress distribution in cylinder during tempering
10
15
Radius (r ), mm
20

N
e m
ij
bI n I dij and
I1
e tp
ij
3
2
KI h(nI )n I sij
I1
(Eq 19)
where b and KI stand for the dilatation due to

structural change and parameter due to transformation plasticity depending on the Ith constituent.
Kinetics of the Quenching Process. In the
case of quenching, two kinds of phase transformation are anticipated: one is governed by the
diffusionless or martensite mechanism. From
thermodynamic consideration, the formula for
this type of reaction from austenite is often assumed to be governed by the modified Magees
rule (Ref 92) as:
nM 1 exp{w1T w2(C C0)
w31rij
w32 J 21/2
w4
(Eq 20)
where nM is the volume fraction of martensite

that is the function of carbon content, temperature, and stress. Here J2 is the second invariant
of deviatoric stress. w1, w2, w3, w4 are all coefficients obtained from experiments.
The other type of phase transformation is controlled by diffusion mechanism, and the volume
fraction of developing phase such as pearlite
may be expressed by modifying the JohnsonMehl relation (Ref 93) as:
np 1 exp
t
0
ft(T) fs(rij ) fc(C)(t s)3ds
(Eq 21)
In this equation, ft(T), fs(rij), and fc(C) are the

function of temperature T, stress rij, and carbon
content C, respectively. Since the TTT diagram
under the applied mean stress rm in logarithmic
scale deviates from the one without stress, which
is represented by the function f(T), the kinetic
equation of diffusion type is often applied to the
variations of pearlite or ferrite structure in
quenching processes. Identification of the function f(T) can be made possible by the use of some
experimental data of the structure change.
Transformation plasticity is known as the
phenomenon of accelerated plastic deformation
caused by a low level of applied stress during

the phase transformation. In addition to the coupling effects between stress and phase transformation, this phenomenon is also expected to influence the stress and/or strain distribution
during quenching. Following Desalos et al. (Ref
94), the transformation plastic strain rate e tp
ij is
determined by:
N1
e tp
ij
(Eq 22)
Based on the metallothermomechanical theory

(Ref 88, 89), the formulated finite-element equation system considering the coupling between increment of nodal displacement {Du} and temperature {T} as well as volume fraction of
structure nI can be expressed as:
[P]{T } [H]{T} {Q(nI ), rij }
(Eq 23)
and
[K(ui)]{Dui} {DF(T,nI)}
(Eq 24)
Here, matrices [P], [H], and [K] represent the

matrices of heat capacity, heat conduction, and
stiffness, respectively, and the vectors {Q} and
{DF} are heat flux and increments of thermal
load. These equations are strong nonlinear equations and are derived by the use of the expression
of stress increment vector as:
{}dT
(Eq 26)
and
([De]1{r}) b {1}
{b}
I
nI
(Eq 27)
I and bI are the coefficients of thermal expansion and dilatation due to Ith phase transformation, respectively. [De] denotes the elastic matrix
of materials.
In order to treat unsteady coupled equations
depending on time, a step-by-step time-integration method and Newton-Raphson method
are introduced in numerical calculation, while an
incremental method is used for deformation and
stress analysis. Because heat-transfer coefficient
depends on variation of the temperature on the
boundary of heat transfer, one also uses the nonuniform time step to calculate temperature,
phase transformation, and deformation fields
(Ref 89).
Examples of Simulation Results

Simulation of Quenching Processes of a
Carbon Steel Disk and Cylinder (Ref 95). A
finite-element method code HEARTS was used
for simulating the quenching processes of steel
disk and cylinder (Ref 95). Models of a 0.45%
I1
30
{b}dn
I
I1
3d e c
{s}
2r
I1
I
I I
Coupling Algorithm in Simulation by

Finite-Element Analysis
{dr} [D] {de}

([De]1{r})
T
n {1} T n dT{1}
I ij
Here, KI is the transformation plasticity coefficient for martensitic or pearlitic transformation,

which may be identified by the experimental results dependent on dilatation-temperature diagrams.
{}

2 K h(n )n s and
h(nI) 2(1 nI)
N1
where e c and {s} are equivalent creep strain and

deviatoric stress vector. [D] denotes an elasticplastic matrix based on Mises-type yield function. Here, the functions depending on temperature and phase transformation and b can be
written as:
3
1
KI h(nI )dnI {s}
2
S0
r 2
dT
T
I1
r 2
dnI {s}
nI
(Eq 25)
10
Thermal stress
Temperature
Stress/strain
Heat generation due to work
Carbon content
Latent
heat
Stress-induced
transformation
Metallic structure
Temperature-dependent
phase transformation
Fig. 108
Elements: 136
Nodes: 162
Transformation stress and

transformation plasticity
Coupling effects of temperature, stress/strain, and metallic structure in heat treatment
Fig. 109
Finite-difference mesh for steel disk specimen 30 mm diam by 10 mm thick)

5 104
2 104
Heat-transfer coefficient (h), W/m2 K
C steel disk were employed to simulate the

quenching process. The diameter and thickness
of the disk are 30 and 10 mm, respectively. Then,
the finite-element model shown in Fig. 109 pertained to an axisymmetrical problem.
Figure 110 shows the heat-transfer coefficients used for the simulation. One of these coefficients was estimated using the lumped-heatcapacity method program LUMPPROB (Ref
63) and the cooling-curve data of the JIS silver
probe. Another coefficient was estimated using
the inverse method program InvProbe-2D (Ref
64, 65) and the cooling-curve data of the ISO
Inconel 600 alloy probe. These coefficients were
modified by a trial-and-error method repeating
the calculation of cooling curve and the modification of the surface-boundary condition.
Figure 111 shows the calculated volume fraction of martensite after water quenching of S45C
steel disk. The calculated volume fraction of
martensite depends largely on the accuracy of
the heat-transfer coefficients. Figure 112 illustrates the distribution of the calculated and the
measured distortion of steel disk These results
show that the modified heat-transfer coefficient
gave the most accurate simulation of the distortion after water quenching. Figure 113 shows the
distribution of residual stress on the flat end surface of S45C steel disk after water quenching.
The residual stresses measured by x-ray diffraction are compared with calculated results in Fig.
113. The calculated stresses of simulation have
good agreement to measured results as shown in
this figure. However, the modified heat-transfer
coefficient gave the better result of the simulation than the other.
Finite-difference mesh for steel cylinder specimen (1/4 model, 20 mm in diameter by 60 mm)
is shown in Fig. 114. Figure 115 illustrates the
distribution of residual stress on the side surface
of the steel cylinder after polymer quenching.
The calculated residual stresses have good agreement to measured results. Figures 116(a) and (b)
are simulated shape changes of S45C steel cylinder during water and polymer quenching, respectively. These results clarify that nonuniform
cooling in water quenching causes larger distortion near the end than in polymer quenching.
Prediction of Warpage of Steel Shafts with
Keyway (Ref 96). The finite-element method
system DEFORM-HT (Ref 97) was used for
simulating quenching of a JIS S45C (0.45% C)
steel shaft with keyway. A cylinder, 10 mm in
diameter and 100 mm long, was with a keyway,
2.5 mm deep and 4 mm wide. A tetrahedral mesh
containing approximately 33,000 elements was
applied to one-half of the specimen; symmetry
conditions were applied to the center surface.
The results were evaluated against the experimental data. The simulated quench distortion
history was compared against its real counterpart. Excellent resemblance was found between
them, as Fig. 118 is compared with Fig. 117. In
both reality and simulation, the shaft bent toward
the keyway side in the beginning 1.7 s and bent
to the other direction after 1.7 s. This phenomenon was believed to be caused by the combi-
hside
104
5000
2000
1000
hend
hend
500
200
100
50
JIS probe + LUMPPROB

ISO probe + InvProbe-2D
Modified: Even
hside
20
10
0
(hside = hend)
200
400
600
800
Even
Temperature (T ), C
Heat-transfer coefficients used for the simulation. These coefficients were estimated by using the lumpedheat-capacity method program LUMPPROB and the cooling-curve data of the JIS silver probe, or by using
the inverse method program InvProbe-2D and the cooling-curve data of the ISO Inconel 600 alloy probe, and modified
by a trial-and-error method by repeating the calculation of cooling curve and the modification of the surface boundary
condition.
Fig. 110
Fig. 111
Calculated volume fraction of martensite after water quenching of S45C steel disk used for surface boundary
conditions in simulation of 1/4 F M model of disk specimen shown in Fig. 109
Radius (horizontal)(ls), mm
Change of thickness (t ), mm
(Left side)
0.075
15
10
23 mm diam cylindrical model of SAE 1055

steel under scanning induction hardening as
shown in Fig. 124. Figure 125 shows measured
hardness and simulated volume fraction of martensite for scanning velocity v 10 mm/s. The
simulated effective case depth (defined as the region with 50% of martensite fraction) almost
corresponds to the measured effective case depth
(defined as the region over 500 HV).
Figure 126 shows the simulated residual stress
distribution along the surface of the cylinder
with the measured values by x-ray diffraction.
Both data of simulation and experiment for scanning velocity v 10 mm/s give higher values
than for v 5 mm/s.
cross section of the ring at final step. Experimental results of distortion are plotted as solid
marks in the figure. Changes in the distortion
with time with and without transformation plasticity are shown in Fig. 123. The most remarkable difference between the two lies in the mode
of distortion. Transformation plasticity gives
convex shape on all surfaces, but concave without the effect. Experimental data agree well with
simulated distortion with the effect of transformation plasticity in both carburized and normal
quenching.
Computer Simulation of Residual Stresses/
Distortion in Scanning Induction Hardening
(Ref 99). Ikuta et al. (Ref 99) carried out a coupled metallothermomechanical analysis using
heat treatment simulation code HEARTS for a
nation of thermal stress, deformation, and transformation-induced volume changes.

Prediction of Distortion during Carburized-Quenching of Chromium-Molybdenum Steel Ring (Ref 98). Yamanaka et al. (Ref
98) carried out a coupled metallothermomechanical analysis using heat treatment simulation
code HEARTS for a ring-shaped model (see
Fig. 119 and 120) of chromium-molybdenum
steel under the carburized-quenching process
shown in Fig. 121. Diffusion equation for carbon
content was solved and coupled with metallothermomechanical analysis. The results were
compared with the experimental data to verify
and evaluate the influence of transformation
plasticity on simulation results.
Figure 122 shows the simulated distortion of
5
Summary
(Right side)
10
15
This article provides a detailed explanation of

the contributing factors affecting residual stress
and dimensional control during quenching.
These factors include phase transformations during heat treating, metallurgical sources of stress
0.05
0.025
0
0.025
0.05
0.075
0.1
20
Measured
Horizontal
Vertical
15
10
10
Radius (vertical)(ls), mm
(Upper)
15
(Lower)
Calculated heat-transfer coefficients

Modified (even on all surfaces)
(a)
30
Distance from the center

of thickness (
t), mm
ds
ls
10 mm
0
2.5
5
0.1
Fig. 114
400
200
200
Calculated and measured distortion after water quenching of S45C steel disk. (a) Distribution of axial distortion. (b) Distribution of radial distortion
400
0
200
400
600
800
1000
15
Measured
Calculated
Heat-transfer coefficients
10
Finite difference mesh for steel cylinder specimen (20 mm diam by 60mm long)
0
200
400
600
800
10
1000
15
15
Radius, mm
(a)
r
Elements: 300
Nodes: 341
0.1
Change of radius (
r ), mm
(b)
Fig. 113
60
2.5
30 mm diam
Fig. 112
10
10
15
Radius, mm
(b)
Calculated and measured residual stress distribution after water quenching of S45C steel disk. (a) Radial stress distribution on end surface. (b) Tangential stress distribution
on end surface

400
Measured
Calculated
Heat-transfer coefficients
200
0
200
400
600
800
1000
10
20
30
40
50
60

(a)
400
200
0
200
400
600
800
1000
10
20
30
40
50
60

(b)
Calculated and measured residual stress distribution after polymer quenching of S45C
steel cylinder. (a) Axial stress distribution on side surface.
(b) Tangential stress distribution on side surface
Fig. 115
Fig. 117
Boiling behavior and distortion during quenching in still city water at 30 C. Keyway is on left side of
specimen. Source: Ref 97
After quenching
Before quenching
t=2s
t=4s
t=6s
t=8s
Before heating
o
(a)
0.5 s
0.8 s
1.0 s
1.2 s
1.7 s
2.0 s
2.4 s
3.3 s
4.0 s
(b)
Simulated shape change history of a S45C steel cylinder (1/4 shape of cylinder, 20 mm diam by 60mm
long) during water quenching (a) and polymer quenching (b). Quenchants were 30 C city water and 30 C
10% polymer (PAG) quenchant without agitation. Distortion is enlarged by 100 times.
Fig. 116
0.3 s
Fig. 118
Simulated quench distortion history during

water quenching. Source: Ref 97

and distortion during reheating and quenching,
effect of materials and quench process design on
distortion, quenchant selection, measurement
and evaluation of quenching power, effect of
process variables (bath temperature, agitation,
and others) on cooling behavior and heat transfer, effect of cooling characteristics on residual
stress and distortion, methods of minimizing distortion, and tempering. Although quenching is a
complex process, and although it exhibits one of
the largest contributing effects to overall residual
stress generation and dimensional control problems, an understanding of these effects allows
the engineer to bring and keep the heat treating
process under control.
REFERENCES
1. S. Owaku, Quench Distortion of Steel Parts,
Netsu Shori (J. Jpn. Soc. Heat Treat.), Vol
32 (No. 4), 1992, p 198202
2. S. Mocarski, Carburizing and Its ControlI.
Basic Considerations, Ind. Heat., Vol 41
(No. 5), 1974, p 5870
3. A. Bavaro, Heat Treatments and Deformation, Trait. Therm., Vol 240, 1990, p 3741
4. F. Legat, Why Does Steel Crack During
5.
6.
7.
8.
9.
10.
(a)
75
11.
25
12.
10
13.
(b)
Fig. 119
With transformation plasticity

Without transformation plasticity
Initial
Experimental
Dimension of ring-shaped specimen of

chromium-molybdenum steel
14.
Simulated distortion after quenching with

measured data. Distortion is enlarged by 100
times. Central axis is the left side. (a) Carburized quenching. (b) Normal quenching. Source: Ref 98
Fig. 122
15.
Quenching, Kovine, Zlitine, Technologie,

Vol 32 (No. 34), 1998, p 273276
K.E. Thelning, Steel and Its Heat Treatment, Butterworths, London, 1985
H. Berns, Distortion and Crack Formation
by Heat Treatment of Tools, Radex
Rundsch., Vol 1, 1989, p. 4057
G. Beck, Internal Stresses Associated with
Heat Treatment of Alloy, Mem. Etud. Sci.
Rev. Metall., June 1985, p 269282
T. Ericsson and B. Hildenwall, Thermal and
Transformational Stresses, 28th Sagamore
Army Materials Research Conf., 1981, p
1938
P. Mayr, Dimensional Alteration of Parts
Due to Heat Treatment, Residual Stresses in
Science and Technology, Vol 1, GarmischPartenkirschen, 1986, DGM Informationsgesellschaft mbH, Federal Republic of Germany, 1987, p 5777
C. Jensen, Dimensional Changes During
Heat Treatment of Hot Work Steel, Ind.
Heat., March 1993, p 3338
R.W. Edmonson, Dimensional Changes in
Steel During Heat Treatment, Met. Treat.,
Vol 20 (No. 6), 1969, p 319
G.E. Totten and M.A.H. Howes, Distortion
of Heat Treated Components, chapter 5,
Steel Heat Treatment Handbook, G.E. Totten and M.A.H. Howes, Ed., 1997, Marcel
Dekker, p 292
D.A. Canonico, Stress-Relief in Heat Treating of Steel, Heat Treating, Vol 4, ASM
Handbook, 1991, ASM International, p 33
34
C.E. Bates, G.E. Totten, and R.L. Brennan,
Quenching of Steel, Heat Treating, Vol 4,
ASM Handbook, 1991, ASM International,
p 67120
W.T. Cook, Review of Selected Steel-Related Factors Controlling Distortion in
1
2
Time
10 s
60 s
2400 s
Fig. 120
Mesh division near a edge of ring-shaped

model
930 C
4h
850 C
1h
1 h 30 min

(a)
Cp
1.2%
Fig. 121
0.85%
OQ
(130 C)
Heat cycle of carburized-quenching process

(b)
Comparison of simulated results depending on the effect of transformation plasticity in course of quenching.
Distortion is enlarged by 100 times. Central axis is the left side. (a) Carburized quenching. (b) Normal
Fig. 123

Heat-Treatable Steels, Heat Treat. Met., Vol
26 (No. 2), 1999, p 2736
16. Y. Toshioka, Heat Treatment Deformation
of Steel Products, Mater. Sci. Tech., Vol 1,
Oct 1985, p 883892
17. A. Thuvander and A. Melander, Calculation
of Distortion During Quenching of a Low
Carbon Steel, First ASM Heat Treatment
and Surface Engineering Conference in Europe, Part 2, Trans. Tech. Publications,
1992, p 767782
18. P.C. Clarke, Close-Tolerance Heat Treat-
19.
20.
21.
22.
ment of Gears, Heat Treat. Met., Vol 25

(No. 3), 1998, p 6164
R.F. Kern, Thinking Through to Successful
Heat Treatment, Met. Eng. Q., Vol 11 (No.
1), 1971, p 14
R. Kern, Distortion and Cracking, II: Distortion from Quenching, Heat Treat., March
1985, p 4145
R. Kern, Distortion and Cracking, III: How
to Control Cracking, Heat Treat., April
1985, p 3842
H.-J. Chen and Z.W. Jiang, Microstructure
23.
24.
Finite-element mesh
Cooling zone
Heating zone
25.
r
D
A
26.
z
54
110
27.
54
55
70
300
100
70
440
Fig. 124
Finite-element mesh and boundary condition of scanning induction quenching system. Source: Ref 99
28.
100
Martensite fraction (M ), %
90
80
29.
70
60
30.
50
40
31.
30
20
10
0
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
9.0
10.0
32.
Depth (d ), mm
(a)
100
33.
90
Hardness (H ), HV
80
70
34.
60
50
40
35.
30
20
10
0
10
Depth (d ), mm
(b)
Fig. 125
Simulated volume fraction of martensite (a) and measured hardness (b) for scanning velocity v 10 mm/s
36.
Improvement
and
Low-Temperature
Quenching of Dies Made of Cr12-Type
Steel, Jinshu Rechuli, No. 8, 1992, p 3941
H.-H. Shao, Analysis of the Causes of
Cracking of a 12% Cr Steel Cold Die During
Heat Treatment, Jinshu Rechuli, No. 11,
1995, p 43
R.R. Blackwood, L.M. Jarvis, D.G. Hoffman, and G.E. Totten, Conditions Leading
to Quench Cracking Other Than Severity of
Quench, Heat Treating Including the Liu
Dai Memorial Symposium, Proc. 18th
Conf., R.A. Wallis and H.W. Walton, Ed.,
1998, ASM International, p 575585
X. Cheng and S. He, Analysis of Quenching
Cracks in Machine-Tool Pistons under Supersonic Frequency Induction Hardening,
Heat Treat. Met. (China), No. 4, 1991, p
5152
J.R. Davis, ASM Materials Engineering
Dictionary, ASM International, 1992, p 407
P.F. Stratton, N. Saxena, and R. Jain, Requirements for Gas Quenching Systems,
Heat Treat. Met., Vol 24 (No. 3), 1997, p
6063
H.M. Tensi, G.E. Totten, and G.M. Webster, Proposal to Monitor Agitation of Production Quench Tanks, Heat Treating: Including the 1997 International Induction
Heat Treating SymposiumProceed. 17th
Conf., D.L. Milam, D.A. Poteet, G.D. Pfaffmann, V. Rudnev, A. Muehlbauer, and
W.B. Albert, Ed., ASM International, 1997,
p 423441
G.-F. Long, Control of Quenching Distortion in Gears, Jinshu Rechuli, No. 5, 1994,
p 3940
H.M. Tensi, A. Stich, and G.E. Totten, Fundamentals of Quenching, Metal Heat Treat.,
March/April 1995, p 2028
R.T. Von Bergen, The Effects of Quenchant
Media Selection on the Distortion of Engineered Steel Parts, Quenching and Distortion Control, G.E. Totten, Ed., ASM International, 1992, p 275282
G.J. Leidenfrost, De Aqua Communis Nonnullis Tractus, 1756, translated from original; C. Waves, Int. J. Mass Transfer, Vol 9,
1966, p 11531166
A. Yamanouchi, Effect of Core Spray Cooling in Transient State after Loss of Cooling
Accident, J. Nucl. Sci. Technol., Vol 5,
1968, p 547558
R.B. Duffly and D.T.C. Porthouse, The
Physics of Rewetting in Water Reactor Engineering Core Cooling, Nucl. Eng. Des.,
Vol 31, 1973, p 234245
Th. Kunzel, Einfluss der Wiederbenetzung
auf die Allotrope Modifikationsanderung
tauchgekuhlter Metallkorper, Dr. Ing. Dissertation, Faculty for Mechanical Engineering of the Technical University of Munich,
1986, 138 pages
D. Hein, Modellvorstellung zur Wiederbenetzung durch Fluten, Doctoral Thesis
Technical University of Hannover, 1980,
182 pages

37. R. Ladish, Untersuchung der Minimalen
Filmsiedetemperaturen auf Keramischen
und Metallischen Leitern, Report of Kernforschungsstelle Karlsruhe, KfK-2970,
1980, 96 pages
38. P. Stitzelberger-Jakob, Hartevorherbestimmung mit Hilfe des Benetzungsablaufes
beim Tauchkuhlen von Stahlen, Dr. Ing.
Dissertation, Faculty for Mechanical Engineering of the Technical University Munich,
1991, 160 pages
39. H.M. Tensi and K. Lainer, Wiederbenetzung und Warmeubergang beim Tauchkuhlen in Hochleistungsolen, Hart.-Tech.
Mitt., Vol 52, 1997, p 298303
40. H.M. Tensi, Wetting Kinematics, Theory
and Technology of Quenching, B. Liscic,
H.M. Tensi, and W. Luty, Ed., SpringerVerlag, Berlin, 1991, 484 pages
41. S. Owaku, Criterion of Quench Cracking:
Its Sources and Prevention, Netsu Shori (J.
Jpn. Soc. Heat Treat.), Vol 30 (No. 2), 1990,
p 6367
42. A.A. Polyakov, Quenching Properties of
Parts Having Stress Concentrators, Met. Sci.
Heat Treat., Vol 37 (No. 78), 1995, p 324
325
43. A. Rose, Cooling Power of Quenching
Agents for Steel, Arch. Eisenhuttenwes.,
Vol 13, 1940, p 345354
44. P.F. Stratton and D. Ho, Individual Component Gas-Quenching in Progress in Heat
Treatment and Surface Engineering, Proc.
5th ASM Heat Treatment and Surface Engineering Conf., E.J. Mittemeijer and J.
Grosch, Ed., 79 June 2000 (Gothenburg,
Sweden), ASM International, p 367375
45. T.W. Ruffle and E.R. Byrnes, Quenching in
Vacuum Furnaces, Heat Treat. Met., Vol 4,
1979, p 8187
46. B. Lhote and O. Delcourte, Gas Quenching
with Helium in Vacuum Furnaces, Quenching and Carburizing, Third International
Seminar, International Federation for Heat
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
Treating and Surface Engineering, Melbourne, Australia, 1991, p 104127

S.R. Heubner, High Circulation, Low Pressure: Simple, Cost-Effective Approach,
Heat Treat., Sept 1985, p 2829
W.R. Jones, High Velocity Gas Flow Seen
as Key to Rapid Quench, Heat Treat., Sept
1985, p 3435
H.E. Boyer and P.R. Cary, Quenching and
Control of Distortion, ASM International,
1988, p 7273
G.E. Totten, C.E. Bates, and N.A. Clinton,
Other Quenching Processes, chapter 8,
Handbook of Quenchants and Quenching
Technology, ASM International, 1993, p
291338
B. Liscic, State of the Art in Quenching,
Quenching and Carburizing, Third International Seminar, International Federation
for Heat Treating and Surface Engineering,
Melbourne, Australia, 1991, p 127
M.A. Delano and J. Van Den Sype, Fluid
Bed Quenching of Steels: Applications Are
Widening, Heat Treat., Dec 1988, p 3437
P. Summer, Quenching in Fluidized Beds,
Heat Treat. Met., No. 2, 1986, p 3944
R. Reynoldson, Use of Fluidized Beds for
Quenching in the Heat Treatment Field,
Quenching and Carburizing, Third International Seminar, International Federation for Heat Treating and Surface Engineering, Melbourne, Australia, 1991, p 85
105
G.E. Totten, Y.H. Sun, G.M. Webster, C.E.
Bates, and L.M. Jarvis, Computerized
Steel Hardness Predictions Based on Cooling Curve Analysis, Paper No. CM98
203, Society of Manufacturing Engineers,
1998
G.E. Totten, C.E. Bates, and N.A. Clinton,
Polymer Quenchants, chapter 5, Handbook
of Quenchants and Quenching Technology,
J.P. Holman, Heat Transfer, 1976, Mc-
300
0
100
100
200
300
400
500
600
700
800
61.
62.
63.
64.
65.
66.
0
100
200
300
400
500
600
700
800
50
100
150
200
250
300
350
900
1000
400
Distance (z ), mm
(a)
60.
z calculated
calculated
z measured
measured
200
Residual stresses (z,), MPa
Residual stresses (z,), MPa
100
Fig. 126
59.
300
z calculated
calculated
z measured
measured
200
900
1000
58.
Graw-Hill Kogakusha, Ltd., Tokyo, p 114

115
Standard Test Method for Determination of
Cooling Characteristics of Quench Oils by
Cooling Curve Analysis, Standard D6200
97, ASTM
Industrial Quenching OilDetermination
of Cooling CharacteristicsNickel-Alloy
Probe Test Method, ISO 9950, International Standards Organization, 1995(E)
Heat Treating Oils, JIS K 22421997,
Japanese Standards Association, Tokyo, Japan
M. Narazaki, M. Kogawara, A. Shirayori, S.
Fuchizawa, and G.E. Totten, Experimental
and Numerical Analysis of Cooling Curves
During Quenching of Small Probes, Heat
Treating: Proc. 20th Conf., K. Funatani and
G.E. Totten, Ed., 912 Oct 2000, ASM International, p 423441
J.P. Holman, Heat Transfer, McGraw-Hill
Kogakusha, Ltd., Tokyo, 1976, p 97
M. Narazaki, M. Kogawara, A. Shirayori,
and S. Fuchizawa, Analysis of Quenching
Processes Using Lumped-Heat-Capacity
Method, Sixth Int. Seminar, International
Federation for Heat Treating and Surface
Engineering, 1997 (Kyongju, Korea), p
428435
M. Narazaki, H. Hiratsuka, A. Shirayori,
and S. Fuchizawa, Examination of Methods
for Obtaining Heat Transfer Coefficients by
Quenching Small Probes, Proc. Asian Conf.
Heat Treatment of Materials, 1315 May
1998 (Beijing), p 269274
M. Narazaki, K. Kogawara, A. Shirayori,
and S. Fuchizawa, Accuracy of Evaluation
Methods for Heat Transfer Coefficients,
Heat Treating Including the Liu Dai Memorial Symposium, Proc. of the 18th Conf.
R.A. Wallis and H.W. Walton, Ed., ASM
J.V. Beck and A.M. Osman, Analysis of
Quenching and Heat Treating Processes Using Inverse Heat Transfer Method, Proc.
50
100
150
200
250
Distance (z ), mm
(b)
Residual stresses along the surface of SAE1055 steel cylindrical specimen. Scanning velocity v (a) 10 mm/s and (b) 5 mm/s
300
350
400
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
First Int. Conf. Quenching and Distortion

Control, 2225 Sept 1992 (Chicago, IL), p
147153
B. Hernandez-Morales, J.K. Brimacombe,
E.B. Hawbolt, and S.M. Gupta Determination of Quench Heat-Transfer Coefficients
Using Inverse Techniques, Proc. First Int.
Conf. Quenching and Distortion, Control,
2225 Sept 1992 (Chicago, IL), p 155164
M. Narazaki, M. Kogawara, A. Shirayoria,
and S. Fuchizawa, Influence of Wetting Behaviour on Cooling Characteristics During
Quenching of Hot Metal, Proc. Third Int.
Conf. Quenching and Control of Distortion,
2426 March 1999 (Prague), p 112120
H.M. Tensi, Determination of Quenching
Power of Various Fluids, Theory and Technology of Quenching, B. Liscic, H.M. Tensi,
and W. Luty, Ed., Springer-Verlag, Berlin,
1991, p 208219
M. Narazaki, S. Fuchizawa, M. Kogawara,
and M. Inaba, Effects of Surface Oxidation
on Cooling Characteristics During Quenching of Heated Metals in Subcooled Water,
Tetsu-to-Hagane (J. Iron Steel Inst. Jpn.),
Vol 79, 1993, p 583589
M. Narazaki, A. Asada, and K. Fukuhara,
Recent Research on Cooling Power of Liquid Quenchants in Japan, Proc. Second Int.
Conf. Quenching and Distortion Control,
47 Nov 1996 (Cleveland, OH), p 3746
T. Tamari and H. Yoshida, Effect of Surface
conditions on Heat Transfer Coefficient
During Spray Cooling on a Hot Stationary
Plate, CAMP-ISIJ (Report of ISIJ Meeting),
Vol 8, 1995, p 444
M. Narazaki, S. Fuchizawa, and N. Takeda,
Effect of Surface Coatings on Cooling Characteristics During Water Quenching of Hot
Metals, Netsu Shori (J. Jpn. Soc. Heat
Treat.), Vol 28, 1988, p 279285
Y. Kikuchi, T. Hori, H. Yanagawa, and I.
Michiyoshi, The Effect of Thin Layer on
Heat Transfer Characteristics During
Quenching of Hot Metals in saturated Water, Int. Trans. ISIJ, Vol 26, 1986, p 576
581
M. Narazaki, S. Fuchizawa, and M. Inaba,
Effects of Surface Texture on Cooling
Curve During Quenching of Heated Metals
in Subcooled Water, Tetsu-to-Hagane (J.
Iron Steel Inst. Jpn.), Vol 76, 1990, p 902
909
M. Narazaki, Effects of Work Properties on
Cooling Characteristics During Quenching,
35, 1995, p 221226
77. M. Narazaki, S. Fuchizawa, and M. Usuba,

Effects of Specimen Geometry on Characteristic Temperature During Quenching of
Heated Metals in Subcooled Water, Tetsuto-Hagane (J. Iron Steel Inst. Jpn.), Vol 75,
1989, p 634641
78. T. Ericsson, Principles of Heat Treating of
Steels, Heat Treating, Vol 4, ASM Handbook, ASM International, 1991, p 16
79. H.J. Yu, U. Wolfstieg, and E. Macherauch,
Zum durch messereinfluss auf die Eigenspannnngen in ol-und Wasserabgeschreckten Stahlzylindern, Arch. Eisenhuttenwes.,
Vol 51, 1980, p 195
80. M. Narazaki, S. Fuchizawa, and M. Kogawara, Effect of Surface Roughness and Surface Texture on Quench Cracking of Steel,
33, 1993, p 5663
81. H. Webster and W.J. Laird, Jr., Martempering of Steel, Heat Treating, Vol 4, ASM
Handbook, ASM International, 1991, p 137
82. J.R. Keough, W.J. Laird, Jr., and A.D. Godding, Austempering of Steel, Heat Treating,
Vol 4, ASM Handbook, ASM International,
1991, p 152163
83. N.I. Kobasko, M.A. Arpnov, G.E. Totten,
and A.V. Sverdlin, Method and Equipment
for Implementation of Intensive Quenching,
Heat Treating Including the Liu Dai Memorial Symposium, Proc. of the 18th Conf.,
R.A. Wallis and H.W. Walton, Ed., ASM
84. M. Wisti and M. Hingwe, Tempering of
Steel, Heat Treating, Vol 4, ASM Handbook, ASM International, 1991, p 121136
85. Y. Toshioka, Heat Treatment Deformation
of Steel Products, Mater. Sci. Technol., Vol
1, 1985, p 883892
86. T. Inoue, K. Haraguchi, and S. Kimura,
Analysis of Stresses Due to Quenching and
Tempering of Steel, Trans. ISIJ, Vol 18,
1978, p 1115
87. T. Inoue and B. Raniecki, Determination of
Thermal-Hardening Stress in Steels by Use
of Thermoplasticity, J. Mech. Phys. Solids,
Vol 26, 1978, p 187212
88. T. Inoue, D.Y. Ju, and Y. Arimoto, MetalloThermo-Mechanical Simulation of Quenching ProcessTheory and Implementation
of Computer Code HEARTS, Proc. First
Int. Conf. Quenching and Distortion Control (Chicago), 1992, p 205212
89. D.Y. Ju, M. Sahashi, T. Omori, and T. Inoue, Residual Stresses and Distortion of a
Ring in Quenching-Tempering Process
Based on Metallo-Thermo-Mechanics,
90.
91.
92.
93.
94.
95.
96.
97.
98.
99.
Proc. Second Int. Conf. Quenching and Distortion Control (Cleveland), 1996, p 249
257
M. Narazaki and D.Y. Ju, Simulation of
Distortion During Quenching of Steel Effect
of Heat Transfer in Quenching, Heat Treating Including the Liu Dai Memorial Symposium, Proc. 18th Conf., R.A. Wallis and
H.W. Walton, Ed., 1998, ASM International, p 629638
D.Y. Ju, M. Narazaki, H. Kamisugi, and H.
Hirano, Computer Predictions and Experimental Verification of Residual Stresses and
Distortion in Carburizing-Quenching of
Steel, Proc. First Int. Conf. Thermal Process Modeling and Computer Simulation
(Shanghai), 2000, p 165172
C.L. Magee, Nucleation of Martensite,
American Society for Metals, 1968
W.A. Johnson and R.F. Mehl, Trans. AIME,
Vol 135, 1939, p 416458
Y. Desalos et al., Deformations et Contrains lers du Traitement Thermique de
Pieces en Acier, Report 902, IRSID, St.German-en-Laye, 1982
D.Y. Ju and M. Narazaki, Simulation and
Experimental Verification of Residual
Stresses and Distortion During Quenching
of Steel, Heat Treating: Proc. 20th Conf.,
K. Funatani and G.E. Totten, Ed., ASM International, 2000, p 441447
D. Huang, K. Arimoto, K. Lee, and M. Narazaki, Prediction of Quenching Distortion
on Steel Shaft with Keyway by Computer,
Heat Treating: Proc. 20th Conf., K. Funatani and G.E. Totten, Ed., ASM International, 2000, p 708712
K. Arimoto, G. Li, A. Arvind, and W.T. Wu,
The Modeling of Heat Treating Process,
Heat Treating Including the Liu Dai Memorial Symposium, Proc. 18th Conf., R.A.
Wallis and H.W. Walton, Ed., ASM International, 1998, p 2330
S. Yamanaka, T. Sakanoue, T. Yoshii, T.
Kozuka, and T. Inoue, Influence of Transformation Plasticity on the Distortion of
Carburized Quenching Process of Cr-Mo
Steel, Heat Treating Including the Liu Dai
Memorial Symposium, Proc. of the 18th
Conf., R.A. Wallis and H.W. Walton, Ed.,
F. Ikuta, K. Arimoto, and T. Inoue, Computer Simulation of Residual Stresses/Distortion and Structural Change in the Course
of Scanning Induction Hardening, Proc.
Second Int. Conf. Quenching and Distortion
Control (Cleveland), 1996, p 259266

DOI: 10.1361/hrsd2002p296

Metallo-Thermo-Mechanics
Application to Quenching
T. Inoue, Kyoto University, Japan
COMPUTER SIMULATION is indispensable

for research, development, and design in various
fields of industry. In particular, popularization of
general-purpose programs based on the finiteelement method enables engineers to execute
highly reliable analysis and design. However,
many open problems still remain to which the
general-purpose programs cannot be applied.
Heat treatment process associated with phase
transformation is one of the examples, and the
development of a program specialized for heat
treatment simulation is required from an engineering point of view.
Owing to the fact that metallic structures, temperature, and stress/strain are strongly coupled
in the process incorporating phase transformation, metallo-thermo-mechanics for taking the
coupling effect into account was proposed and
applied to some problems by the authors (Ref 1
7).
In this article, the summary of the metallothermo-mechanics is briefly introduced; governing equations identifying the fields of metallic
structure, temperature, and stress/distortion are
stated; and the correlation between the problems
of phase transformation and the theory is briefly
discussed. In the following sections, the developing strategy, methodology, and structure of
the developed system HEARTS (HEAt tReaTment Simulation system) based on the theory
for simulating the engineering process incorporating phase transformations are stated (Ref 8
10).
Some examples of simulated results are presented: infinitely long cylinder quenched by water (Ref 8), induction hardened and carburized
quenched rings (Ref 11, 12), induction hardened
gear wheel (Ref 13), and Japanese sword during
quenching (Ref 1416). The validity of the theory and the system are evaluated by comparing
the simulated results with the experimental data.
phase changes, such as solid-solid transformation in quenching, tempering, and other heat
treatments of steels, and also solid-liquid transformation in welding, casting, continuous casting, and so on. In such cases, three kinds of parameters are dominant: phases of metallic
structures, temperature field, and elasticor in
many cases, inelasticstress and strain. A variation in any of the three parameters will have an
effect on the others, and the interaction is properly termed a metallo-thermo-mechanical coupling (Ref 17).
Figure 1 depicts a schematic diagram for these
parameters and the effect of their interaction (Ref
16). When the temperature distribution in a material (and in some cases the heating and cooling
rate) is uniform, thermal stress:
Is caused in the solid and the temperature-
dependent phase transformation
Leads to alterations in metallic structure (i.e.,
austenite-pearlite and austenite-martensite

transformation in the solid phase and the
melting or solidification in solid-liquid transition); local dilation due to phase changes in
the solid brings about transformation stress
Interrupts the stress or strain field of the body.
In addition, transformation plasticity effect is
sometimes considered to be important (Ref 17
21).
Metallo-Thermo-Mechanics
Materials often behave in a complicated manner during engineering processes incorporating
Fig. 1
Metallo-thermo-mechanical coupling
Contrary to these effects, which are taken into

account in ordinary stress analyses, arrows
pointing the opposite direction indicate the coupling modes. Part of the mechanical work done
by existing stress in the material converts into
heat generation, thus disturbing the temperature.
The acceleration of phase transformation by
stress or strain, called stress- (or strain-)induced
transformation (Ref 2228), has been discussed
and is sometimes applied to improve the mechanical properties of metallic materials. Arrow
corresponds to the latent heat generation due to
phase transformation, which affects the temperature distribution.
When distribution of chemical composition
occurs inside the material, say, in the case of
carburized quenching, the effect of diffused carbon content also changes the property of such
fields, as indicated by the dotted lines in the figure.
Since mechanical and thermophysical properties appearing in the analysis of such fields depend on the material itself, the necessity of a
database for the materials is strongly recognized,
and some trials are ongoing.
The aim of this section is to formulate the fundamental relationships governing these three interacting parameters relevant to some problems
of heat treatment of steels, taking into account
the effects of coupling, and to formulate some
Metallo-Thermo-Mechanics-Application to Quenching / 297
Kinetics of Phase Transformation

A material undergoing structural change due
to phase transformation is assumed to consist of
a mixture of N constituents (Ref 29), such as
pearlite, austenite, bainite, martensite, and liquid
phase, as schematically shown in Fig. 2. Liquid
phase is to be considered in welding, casting, or
so on, but this is not necessary in the usual case
of heat treatment. Denoting the volume fraction
of the Ith constituent as nI, the mechanical and
physical properties v of the material are assumed
to be expressed by the mixture law as a linear
combination of the properties vI of the constituents:
N
v
vI nI
I1
(Eq 1)
where
N
nI 1
(Eq 2)
I1
In the practice of heat treatment, mostly four

kinds of volume fraction are chosen: pearlite nP,
austenite nA, bainite nB, and martensite nM in the
case of quenching.
Diffusion-Type Transformation. Many rate
equations for diffusion-type transformation have
been proposed to depict pearlitic reaction from
austenite, including the form by Johnson and
Mehl (Ref 30):
nP 1 exp(Ve)
(Eq 3)
t0
where Ve
f (T) (t s) ds, are employed
as a basis. Here, Ve is the extended volume at
time, t (0 s t).
The existing stress, or sometimes strain, is
known to accelerate the reaction, as experimentally observed by many authors (Ref 2228).
3
Figures 3 (Ref 26) and 4 (Ref 25) show examples

of pearlitic reaction accelerated by applied stress.
So, the volume fraction of pearlite may be expressed as:
by use of the function w(rij) representing the effect of stress in the form:
nP 1 exp(Ve)
J2
Ve
f (T, r ) (t s)3ds
ij
(Eq 4)
by modifying the Johnson-Mehl relation (Eq 3).

For the function f (T, rij), one uses:
f (T, rij) exp( prm) f (T)
(Eq 5)
since
the
time-temperature-transformation
(TTT) diagram under the applied mean stress rm,
with some extent p, on a logarithmic scale deviates from one without stress and since Eq. 5
expresses the normal TTT diagram by applying
the function f (T).
The same type of kinetic equation (Eq 4)
might be adopted for the reaction of austenite
from pearlite during heating process, when necessary.
Diffusionless Transformation. The other
type of phase transformation is the diffusionless
one, or martensite reaction. In this reaction, volume fraction of martensite nM is just the function
of temperature, but not of the history as seen in
diffusion-type reaction, and the equation proposed by Magee (Ref 31).
w(rij) Arm BJ 21/2

1
sij sij
2
Heat Conduction Equation and

Diffusion Equation of Carbon
The local energy balance or the first law of
continuum thermodynamics is usually given in
terms of internal energy U [G Tg
1/qrij eeij] as (Ref 32):
rij e eij qi 0
qU
xi
1.00
nM 1 exp[u(T Ms)]
(Eq 6)
is widely used with the material parameter u,

which means that martensite generates as the
temperature changes from martensite start temperature Ms.
Introducing thermodynamic considerations on
the effect of stress based on the experimental evidence, the equation is modified to:
nM 1 exp[u(T Ms) w(rij)]
(Eq 7)
0.75
0.25
0.00
10
Fig. 4
104
Growth of pearlite depending on stress
0.8
Fracture of martensite, M
103
1.0
150
100
50
Tension
0.6
0.4
Compression
0.2
Concept of mixture. n1 to nN (Nend) denote the

volume fractions of constituents.
102
Time (t ), s
Start time
Finish time
Fig. 2
, MPa
,0
, 72
, 144
, 226
, 316
, 412
0.50
250
Time (t), s
(Eq 9)
with Gibbs free energy G, entropy g per unit

mass with density q, stress rij, and elastic strain
eeij. Applying the expressions for the specific heat
c (T g/T) and enthalpy H (G Tg),
200
(Eq 8)
The first term denotes the effect of mean stress

rm (rkk /3), and the second is that by deviatoric stress sij. The parameters A and B can be
identified if one has the data of martensite transformation affected by stress as shown in Fig. 5
(Ref 26).
Fracture of pearlite, P
fundamental equations for the finite-element

method.
Fig. 3
20
40
60
Stress (), MPa
80
100
Start and finish time of pearlite reaction depending on stress
0
0
500
1000
Stress (), MPa
Fig. 5
Martensite fraction depending on stress
1500

Eq 9 is reduced to the general form of heat conduction equation:
eeij
H
qcT T
r ij rij e iji q
T
eiij
H
e iij q
j q
lI nI
j
I1
2T
xi x i
(Eq 10)
Here, Fouriers heat conduction law:

qi k
T
xi
(Eq 11)
with heat conductivity k is adopted, and the latent heat lI when producing the Ith constituent:
H
lI
nI
derive the constitutive equation. Readers are requested to refer to monographs on thermodynamics (Ref 32) for detailed discussion, but here
the authors simply utilize some results for the
application.
The Gibbs free energy G in the thermodynamic state is assumed to be the function of
stress rij, temperature T, and other internal variables related to inelastic deformation such as inelastic strain eiij, hardening parameter j, and
back stress ij as well as the volume fraction nI
of each constituent or phase:
G G(rij , T, eiji , ij, j, nI)
C
C
D
t
xixi
D exp[a(C b)]
nI GI (rij, T, eiji , ij, j)

I1
I4(T T0)dij I1dij
(Eq 17)
GI(rij, T, eiji , ij, j)
2I3
1 mI
EI
Now the restriction of the second law of thermodynamics, called Clausius-Duhem inequality
is adopted, indicating that the internal dissipation is nonnegative, or that the inequality (Ref
32):
2I2
mI
EI
gT)
r ij eije rij e iji
q(G
when, EI, mI and I are Youngs modulus, Poissons ratio, and thermal expansion coefficient of
the constituent, respectively, and 3bI is volumetric dilation due to phase change in this case.
Then, one has:
GIe(rij, T) GiI(T, eiji , ij, j)
(Eq 18)
1
T
qi
0
T xi
(Eq 19)
holds in every process. Applying Eq 16 to Eq 19

one obtains:
(Eq 20)
When one focuses simply on the first term, the

relation:
eeij
Ge
rij
I4 I
I1 bI
e
eIij
G
G
r ij q g
rij
T
1
T
T qi
0
T xi
(Eq 21)
is to be satisfied, otherwise the inequality is not

satisfied for the change of stress rij.
Substituting Eq 18 into Eq 21, one obtains:
(Eq 15)
Elastic Strain. In the case of a complicated

coupling of mechanical field with temperature
and phase change, continuum thermodynamics
provides a tool for fundamental consideration to
(Eq 25)
Since the first two terms of Eq 25 mean the

Hookes law, the third is thermal strain, and the
fourth corresponds to isotropic strain of the Ith
constituent provided that the parameters are constant, one can put:
eije q
nI
I1
GeI (rij, T)
rij
(Eq 27)
with the Kronechers delta dij. Finally, elastic

stress-strain relation of the mixture takes:
eeij
1 m
m
rij rkkdij
E
E
(T T0)dij bdij
with global material parameters:

E
(Eq 22)
Expansion of Geij(rij, T) around the natural state

rij 0 and T T0 (T0 is reference temperature),
(Eq 26)
1 mI
mI
rij
rkkdij
EI
EI
I(T T0)dij bIdij
1
N
nI
EI
nImI
EI
I1
e ij e eij e iij
(Eq 24)
It is already confirmed in the framework of infinitesimal strain theory that the Gibbs free energy is divided into recoverable and inelastic
parts:
Stress/strain state in the process incorporating

phase transformation is very complicated because inhomogeneous temperature and the gradient vary with time, thus inducing thermal
stress and phase transformation stress. Such
stress often exceeds the yield stress, and elasticplastic stress analysis considering the temperature-dependent material parameters is needed.
Strain and the rate (or the increment) are generally assumed to be divided into recoverable
elastic part e eij and irrecoverable inelastic one
e iji .
nIeIije
I1
eije
(Eq 23)
where uI0, uI1, uI2, uI3 and uI4 are the polynomial
function of stress invariants and temperature and
fI (T T0) is the function of temperature rise.
Then, Eq 22 reads:
eeIij 2I3rij 2I2rkkdij
eeij q
Stress-Strain Constitutive Relation
T0)rkk fI(T T0)
G(rij, T, eiji , ij, j, nI)
(Eq 13)
(Eq 14)
I2(rkk)2 I3rklrkl I4(T
with
with diffusivity D as the form, for example:
GeI (rkl, T) q[I0 I1rkk
When the same hypothesis of mixture law for

each phase is again applied to G, then:
(Eq 12)
is adopted.
The third term on the left hand side of Eq 10
denotes the heat generated by inelastic dissipation, which is significant when compared with
the elastic work represented by the second term,
and the fourth term arises from the latent heat by
phase change. Equation 10 is easily reduced to
the original equation of heat conduction, provided these terms are neglected.
As for the case of carburized quenching, one
needs the distribution of carbon content diffused
from the surface of the body. This process is generally performed before quenching, so the coupling with stress and structural change is not to
be considered. Then, the normal type of diffusion equation for carbon content C takes:
(Eq 16)
based on the theory of isotropic function, leads

to:
I1
N
I
I1 EI
(Eq 28)

N
nI I
I1
nIbI
I1
(Eq 29)
Yield Function and Plastic Strain Rate. Subsequent yielding as well as initial yielding of material with plastic strain epij, hardening parameter
j at temperature T under multiaxial stresses is
controlled by a criterion using the yield function
expressed in stress space as:
F
F(rij, epij,
j, T, nI) 0
(Eq 30)
with existing plastic strain epij, hardening parameter j, to be discussed later, temperature T, and
volume fraction of phases nI.
The yield function F necessitates a simple
mathematical structure and is adequately capable
to fit the experimentally observed yield condition. One of the most widely used yield functions
in engineering practice is the von Mises type,
taking the form:
F
1
1
sijsij r 2(j, T, nI)
2
3
(Eq 31)
where sij is the deviatoric stress such that:

sij rij
1
dijrkk
3
(Eq 32)
and r(j,
T, nI) stands for the flow stress of the

material with hardened state j at temperature T,
also depending on temperature T.
The other type of yield function widely used
for ductile metals is the Tresca type:
F (r1 r3) r(j,
T, nI)
(Eq 33)
200
150
Torsional stress (), MPa
100
50
where r1 and r3 denote the maximum and minimum principal stresses.

The hardening parameter j is identified depending on loading history, and the evolution is
expressed either by plastic work as:
j rije ijp
(Eq 34)
or by equivalent plastic strain as:

j e p
e p
3 e e
2
p p
ij ij
(Eq 35)
where epij is plastic (or inelastic) strain. The parameter defined by Eq 34 is sometimes called the
work-hardening parameter, while the one by Eq
35 is the strain-hardening parameter.
How the subsequent yield surface develops
during the course of successive plastic loading
is called the hardening rule. Experiments determining the surface show that the yield surface
expands its diameter accompanying the shift of
center position in stress space (see Fig. 6, Ref
33). For simplicity, however, two characteristic
types of hardening rule are used.
Isotropic-Hardening Hypothesis. The yield
surface is assumed to expand isotropically in
size, keeping the center by this hypothesis,
which gives the yield function in the form:
F F(rij, j, T, nI) f (rij)
r(j,
T, nI)
(Eq 36)
This means that plastic strain epij does not affect

the shape of the yield surface. The solid line in
Fig. 7(a) illustrates the subsequent yield surface
represented in p plane, while the circle indicated
by a broken line is the initial surface. If the initial
yield point is denoted by Y on the stress-strain
curve in Fig. 7(b), unloaded stress AB from point
A is the same as the compressive yield stress BC
and is larger than initial yield stress. The concept
of this hypothesis is too simple to express the
generalized Bauschinger effect, but nevertheless
it is widely used for many kinds of analyses
when the direction of loading trajectory changes
slightly. Original meaning of Bauschinger effect
is such that yield stress with prior plastic strain
0
50
200
0
1
(a)
50
100
150
200
250
Tensile stress (), MPa
Fig. 6
How can one identify the back stress ij in Eq

37? Two typical models can be used.
The Prager model (Ref 34) asserts that the
subsequent yield surface moves parallel to plastic strain rate e pij and is expressed in the form:
ij Cepij
(Eq 38)
with a parameter C. Since the plastic strain occurs perpendicular to the normal of the yield surface as is discussed later, the direction of ij by
this model is parallel to the normal surface (see
Fig. 9).
The Ziegler model (Ref 35) to determine the
back stress rate is expressed as:
ij l(r
ij ij)
(Eq 39)
which follows that the center moves in the direction connecting the stress point and the back
stress as seen in Fig. 9.
Now, the normality of the plastic strain rate
vector induced at the regular point of the yield
surface leads to the flow rule for plastic strain
rate:
e pij K
F
rij
(Eq 40)
A
0
Experimentally determined yield surfaces
A
Y
T C
20
150
400
450
0
(Eq 37)
3
Y
150
F F(rij ij, T, nI)
100
(say, in compression) is lower than initial yield

stress in the opposite direction (in tension). The
concept is extended to the general case in multidimensional stress space, and the generalized
Bauschinger effect is stated as the change in
yield stress in the different direction from that in
prestrained direction.
Kinematic-Hardening Hypothesis. The model
assumes that the yield surface with unchanging
shape moves in stress space due to plastic deformation as illustrated in Fig. 8(a), and stressstrain diagram is given under the condition YY
AC in Fig. 8(b). Qualitative representation of
Bauschinger effect is available by this model
since OY BC.
Provided the center of the yield surface in
stress space, called back stress, is denoted by ij,
the mathematical description of the kinematichardening hypothesis is now derived by substituting rij ij instead of rij as the initial yield
function in Eq 36 as:
(b)
0
1
(a)
2
(b)
300
Fig. 7
curve
Representation of yield hypothesis of isotropichardening type. (a) Yield surface. (b) Stress-strain
Fig. 8
curve
Representation of yield hypothesis of kinematichardening type. (a) Yield surface. (b) Stress-strain

since the normal direction of the function F in
stress space is represented by F/rij. The parameter K is a function that depends on stress,
stress rate, and strain and its history. Since the
equation implies that the plastic strain rates derived from a potential F, the theory is known as
the plastic potential theory.
The next problem is how to identify the parameter K in Eq 40. During plastic loading, the
stress components always satisfy the yield condition F 0 at any moment, which implies F
0. Then, one has the relation called the Pragers consistency relation:
F
F
F
F
r ij p e ijp
j
rij
eij
j
N
F
nI 0
nI
I1
(Eq 41)
Substituting the expression in Eq 40 into Eq 41

as well as the definition of hardening parameters
in Eq 34 and 35, the parameter K is easily determined as:
1 F F T
(ekl dklT)
S0 rij T
(Eq 47)
where
S0
2GG
F F
rmn rmn
(Eq 48)
(Eq 42)
1 F F
T
S0 r T
{r}T {rx ry rz sxy syz szx}

(Eq 50)
0 0}
0
0
0
ij
F F
nI
nI
rij
1.0
Axial stress, z (MPa)

50
0
30
50
1.5
ij
(Prager)
ij
2.5
(Ziegler)
ij
1/2
F
j
(Eq 45)
for the strain-hardening parameter in Eq 35.

Rate Form of the Stress-Strain Relation.
The elastic strain rate e eij is easily derived from
Eq 28:
e eij
1
m
r ij
dijr kk
2G
1 m
dijT dijb
(Eq 46)
where G, m, and stand for shear modulus, Poissons ratio, and linear expansion coefficient, respectively, in the global forms in the sense of
Eq. 29. Substituting Eq 44 and 47 into Eq 46,
Fig. 9
Equivalent stress,
mn
F
rmn
800
0
169 MPa
137
92
55
0
1
F F

p
rmn emn
G
2 F
600
0.5
ij ij
for the work-hardening parameter in Eq 34, and:
3 r
400
(a)
ij
(Eq 44)
Direction of back stress evolution
Axial strain (z), %
200
Temperature (T ), C
1
F
F
F
p
rmn
emn
j
rmn
G
1
2
0 12
0 0 12
(Eq 52)
0
0
0
(Eq 43)
is termed as the hardening function and

Here, G
takes the form:
0
0
0
F r kl F
e pij G
rkl
T
I1
sym
(Eq 49)
{e}T {ex ey ez cxy cyz cyx}
and finally:
1 m
1 2m
m
1 m
1 2m 1 2m
m
m
1 m
[De] 2G
1 2m 1 2m 1 2m
and (see Eq 53).

As an example of yield functions, consider
{r}
[Dep] ({e} {}T)
{}T { 0
with
For further application of the constitutive equation in the finite-element method, it is useful to
present the relation in the matrix form (Ref 36):
(51)
0.5
F
nI
nI
I1
[Dep] [De] [Dp]
where the components in matrices of stress and

strain in engineering notation and coefficient of
linear expansion are, respectively:
F r mn F T
KG
rmn
T
N
m
1 F F
r ij 2G dikdjl
dijdkl
1 2m
S0 rij rkl
Here, the matrix [Dep] is given as (Ref 36):
Axial dilation (Ez), %
F
T
T
one obtains the expression for the total strain

rate:
0.5
1
41
82
137
164
1.5
d
H =
d
p
2
200
400
600
800
Temperature (T ), C
0
0
Fig. 10
Equivalent plastic strain, p

Stress-plastic strain curve and definition of
hardening coefficient
(b)
Experimental data of the effect of applied stress
on induced strain. (a) Pearlitic transformation.
(b) Martensitic transformation
Fig. 11
2G
[D ]
S0
p
F
r
sym
F F
rx ry
r
F
F F
rx rz
F F
ry rz
r
1 F F
2 rx sxy
1 F F
2 ry sxy
1 F F
2 rz sxy
1 F
4 sxy
1 F F
2 rx syz
1 F F
2 ry syz
1 F F
2 rz syz
1 F F
4 sxy syz
1 F
4 syz
1 F F
2 rx szx
1 F F
2 ry szx
1 F F
2 rz szx
1 F F
4 sxy szx
1 F F
4 syz szx
F
the von Mises condition with isotropic hardening:

F
1
1
sijsij r(
ep, nI)2
2
3
(Eq 54)
1 F
4 szx
1
2
smnsmn
3
G
dr
(Eq 55)
Introducing Eq 55 into Eq 43 with T 0 leads

to the so-called Prandtl-Reuss flow rule:
(Eq 53)
e pij
3 e p
sij
2 r
(Eq 56)
The hardening coefficient H is defined as the

tangent modulus of the (equivalent) stress(equivalent) plastic strain diagram in Fig. 10.
When employing H , Eq 56 becomes:
Here, temperature dependence on flow stress r

is ignored for simplicity.
Bearing in mind that F/rij sij, the hardening function in Eq 45 is easily expressed as:
e pij
3 r
sij
2 rH

(Eq 57)
The matrix [Dp] in Eq 53 is simplified in this

case (Ref 37) to the form:
2G
S0
0.2
Transformation plastic strain (tp), %
4 r 2r
p
3 e
[Dp]
z
s2x
sxsy
sxsz
sxsxy
sxsyz
sxszx
s2z
2
szsxy sxy
szsyz sxysyz s2yz
szszx sxyszx syzszx s2zx
(Eq 58)
with
S0
0.1
1
3
2
1 r
r 1
3
3G e p
0.2
7
8
Experimental
25
25
sym
s2y
sysz
sysxy
sysyz
syszx
0.1
0.3
50
2
H
r 1
3
3G
(Eq 59)
Creep Strain. When the quenching time is
50
Stress (), MPa

(a)
(a)
0.5
Transformation plastic strain (tp), %
1/2
3 r de
Pre/postprocessor
2
18
10
0.4
3
19
Intermediate
file
Intermediate
file
14
13
0.3
7
21
11
0.2
12
1
17
5
Interface
15
4
0.1
Experimental
0
16
Post-treatment
file
20
25
50
75
100
Stress (), MPa

(b)
Transformation plastic strain versus applied
stress diagram. (a) Pearlitic transformation. (b)
Martensitic transformation
Control/analytical
condition file
List image
data file
Solver
Final
results file
(b)
Fig. 12
Node/element
file
Fig. 13
Finite elements available in the system. (a)

Two-dimensional. (b) Three-dimensional
Fig. 14
Material
database
Structure of HEARTS
e iij
e pij
e cij
(Eq 60)
e pij
Here, the plastic strain rate is given in Eq 43,

and the creep strain rate e cij is given in the similar
form:
e cij Kc
G
rij
(Eq 61)
is normally expressed as a linear function of applied stress, and the rate is expressed by:
e tp
ij
3
2
KIh(nI)sij
I1
h(nI) 2(1 nI)
(Eq 66)
for N kinds of transformation in the form of

GreenwoodJohnson (Ref 38). This component
of strain is also superimposed to inelastic strain
rate, if necessary.
0.8
Experimental
Calculated
Carbon content (C), %
long in the case of machine component with

large dimension, effect of time on inelasticity is
not negligible, especially at elevated temperature. This effect in metals and alloys is due to
creep in most cases. Such time-dependent (or
rate dependent) inelasticity, sometimes termed
viscoplasticity, has been developed to formulate
sophisticated constitutive equations, under plasticity-creep interaction condition.
The most simple and practically used constitutive model is the so-called superposed type of
plasticity and creep, while the unified model
does not separate the inelastic strain into two
parts. In the superposed model, the inelastic
strain rate e iij in Eq 15 is composed of plastic and
creep strain rates:
by use of creep function, equivalent to the yield

function:
0.6
0.4
0.2
Method of Numerical Simulation

A series of equations governing the heat treatment process described previously are to be numerically solved, since the fundamental equations, Eq 4, 7, 10, and 15 are highly nonlinear.
There are some techniques for numerical calculation, but the finite-element method (FEM)
would be most promising for engineering practice.
Based on the theory available to the heat treatment simulation, the authors developed some
programs for simple case of two-dimensional
and axisymmetric problems in their laboratory
to simulate the quenching process of a solid and
thick-walled cylinders, and gear wheels (Ref 39
44).
In 1992, modification of the program was carried out to develop the computer-aided engineering (CAE) system HEARTS (HEAt tReaTment Simulation code) by the cooperation with
CRC Research Institute Company, Japan, to deliver a commercial code, and it has been used by
many users.
After the development of the HEARTS, some
codes were independently released: DEFORM-
0
0
1.0
2.0
3.0
4.0
Distance from center (r ), mm
G G(rij, ecij, jc, T, nI)
(Eq 62)
Fig. 15
Distribution of diffused carbon

900
with creep-hardening parameter j .

Consistency relation held during creep loading:
(Eq 63)
leads to the final form for creep constitutive

equation:
G
r
r kl
kl
G
G
T
T
G
n I rij
I1 nI
With
transformation
plasticity
400
200
Without
transformation
plasticity
0
0
1.0
3.0
2.0
von Mises or Tresca type is also employed when

the creep function is similar to the yield function.
Transformation Plastic Strain. Large inelastic deformation is induced under relatively low
stress applied during phase transformation (Ref
1721). As is observed in some alloys, such an
inelastic strain is sometimes very large, thus
called transformation superplastic strain. The
strain induced during quenching is, however,
mostly small since the operating time is relatively short; this phenomenon is simply called
transformation plasticity.
Figure 11(a) (Ref 21) and (b) (Ref 19) show
examples of experimental data of the effect of
applied stress on the induced strain during pearlite and martensite transformations, respectively.
The transformation plastic strain is plotted
against applied stress in Fig. 12 (Ref 21), which
Fig. 16
600
300
0
300
4.0
5.0
600
900
6.0
Time (t ), s
(Eq 64)
Fig. 18
2
3
1
Residual stresses distribution
Temperature variation on the surface

900
Volume fraction of constituents, A, M, P
c
e cij G
600
1.0
Axial residual stresses (z), MPa
800
Temperature (T ), C
G rij G e ijc G T
G
rij
eijc
rT
N
G
0
I1 nI
Residual stresses (r, , z), MPa
Pearlite
0.8
0.6
0.4
Martensite
0.2
Austenite
0
Fig. 17
1
2
3
Distribution of volume fraction of martensite,

pearlite, and retained austenite
Experimental
600
300
0
300
600
900
Fig. 19
2
3
1
Effect of transformation plasticity on axial residual stresses
Temperature (T ), C
HT by SFTC Company, United States; SYSWELD by FRAMATOME Company, France;

GRANTAS by Komatsu Company, Japan (see
Ref 45). By introducing the optional code available to calculate phase change calculation into
ABAQUS, some trials are also made.
The strategy and the structure of the HEARTS
are briefly introduced in this section.
1200
1000
800
600
400
200
0
S45C out-top
S45C out-mid
S45C in-mid
SUS304 out-top
SUS304 out-mid
SUS304 in-mid
Volume fraction of austenite

and martensite (A, M), %
(a)
10 20
Time (t ), s
50 100
250
The finite elements available for the

HEARTS system are two- and three-dimensional
type as illustrated in Fig. 13. The two-dimensional element in Fig. 13(a) is converted to a
triangular element by overlapping three nodes,
for example, No. 1, 2, and 5, and other threedimensional type elements are easily generated
by joining the three nodes located on the corner
in Fig. 13(b). The continuum elements are used
in the analyses of heat conduction, stress/strain,
and diffusion. The two-dimensional element in
Fig. 13(a) is also applied to axisymmetrical
problems. For stress/strain analysis, plane
stress/strain problems and generalized plane
strain problems are also available. This procedure is proposed for developing a program for
practical use that can analyze very large systems
and add some different types of elements and
kinds of functions without difficulty.
In the case of the coupled analysis of timedependent metallic structure, temperature, and
stress/strain, the step-by-step method with very
short time step is employed to avoid the increase
of necessary memory due to the nonsymmetric
matrix and also to reduce execution time.
The heat-conduction analysis is carried out by
considering the latent heat generation calculated
by microstructural evolution analysis followed
100
80
Martensite
60
40
Austenite
20
0
10 20
Time (t ), s
50 100
250
10 20
Time (t ), s
50 100
250
Radial displacement
(Ur), mm
(b)
0.4
0.3
0.2
0.1
0.0
0.1
(c)
Axial displacement
(Uz), mm
0.25
0.20
0.15
0.10
0.05
0.00
0.05
10 20
Time (t ), s
50 100
Fig. 22
Mode of distortion of cross section (central axis on the left)
Fig. 23
Distributions of tangential and axial residual stresses on the top surface
250
(d)
Variation of temperature, structures, and distortions, at typical locations of a ring for carbon
steel and SUS304
Tangential plastic strain (), %
Fig. 20
0.6
0.4
S45C
0.2
SNC815
0
SCM440
0.2
0.4
36
Fig. 21
SUS304
38
40
42
44
46
Radius (r ), mm
48
50
Tangential plastic strain distribution for four

steels
Fig. 24
by the stress/strain analysis depending on the

structural distribution based on the temperature
distribution determined. When the stress/strain
analysis ends, calculation in the next step is carried out. In the case when the coupling effect is
negligible, independent analysis of three fields is
also possible. The time integration in the heat
conduction and diffusion analyses employ the hmethod, which is reduced to the trapezoid and
Euler backward methods in the special cases.
Nonlinear equations in each analysis are made
by the Newtons iteration method.
CAE Environment. As shown in Fig. 14, the
system HEARTS was developed under the prerequisite condition to be used in the CAE environment combined with a proper pre/post-processor. The system is designed for exchanging
data between solver and pre/post-processor,
such as PATRAN and I-DEAS.
Data necessary for the simulation are created
by a pre/post-processor. These data are output
in the format of intermediate file and then converted into control/analysis condition data file
and nodal point/element data file by an interface program. In addition, input data regarding
the characteristics of the elements are made by
users in the form of characteristics data file.
The results of calculation by the solver are output in such forms as post-processing file, list
image file, and restart file. The contents of
the post-processing file are converted to the form
of the intermediate file through the interface. The
resulting data contained in the intermediate file
can be displayed in various forms by using the
pre/post-processor according to the instructions
of the users.
Distributions of tangential and axial residual stresses on the outer face
Illustrative Examples of Simulation
Time-dependent distribution of diffused carbon content
5000
4000
3
3000
2
1
2000
4
1000
0
Fig. 26
200
400
600
800
Temperature (T ), C
1000
Temperature-dependent heat transfer coefficient on each face (Ref 46)
0.4
Radial displacement (ur), mm
4
Radial displacement (ur), mm
Heat transfer coefficient (ht), W/(m2 C)
Fig. 25
Some examples of simulated results on temperature, structural change, and stress/strain as

well as carbon content during the heat treatment
process of engineering components in two- (or,
axisymmetric) and three-dimensional shape and
a Japanese sword are illustrated by use of the
system HEARTS combined with an available
pre/post-processor, and the validity of the system is evaluated by comparing with experimental data.
With TP Without TP
Outer
Inner
0.3
0.2
0.1
0
0.1
1
(a)
Fig. 27
10
102
Time (t ), s
103
104
0.4
0.3
0.2
0.1
0
0.1
1
(b)
10
102
Time (t ), s
103
104
Variation of radial distortion. TP, transformation plasticity. (a) Carburized quenching. (b) Normal quenching

Carburized Quenching of Cylinder
The first example is the carburized water
quenching of a long cylinder of a case-hardening
steel containing molybdenum (Ref 8). The cylinder of 8 mm diameter is treated. Since the cylinder is assumed to be long enough and no temperature gradient is imposed in axial direction,
the calculations are made for a series of elements
arrayed in the radial direction, and the generalized plane-strain condition for the stress/strain
analysis is such that the resultant surface force
acting in axial direction disappears, and that axial displacement of all nodes at any radial point
is the same, or the axial cross section is kept inplane.
Before quenching, the cylinder is kept in
methane gas at 925 C for carburization for 9000
s and diffused for 9000 s in vacuum. Calculation
of diffusion mode of carbon based on the diffusion equation (Eq 13) is made prior to quenching, and the result is shown by solid line in Fig.
15. Circles in the figure denote the experimentally measured data by x-ray diffraction, which
gives good agreement with the simulation.
The cylinder is heated to 800 C and quenched
in water at 20 C. Calculated temperature variation on the surface is presented in Fig. 16. Here,
the solid line is the data when considering the
effect of transformation plasticity formulated by
Eq 66, while the broken line is without the effect.
At the beginning of cooling from 800 C, no difference is seen between both lines, but from the
pearlite start temperature at 400 C, where the
latent heat generation raises the temperature, the
effect of transformation plasticity appears. Figure 17 represents distribution of volume fraction
of metallic structures after quenching such as
martensite, pearlite, and retained austenite.
Calculated residual stress distribution in axial,
tangential, and radial components are plotted by
curves in Fig. 18 with the comparison of measured axial residual stress indicated by circles
near the surface. The effect of transformation
plasticity on the residual stress is also depicted
in Fig. 19.
t=8s
M, %
100
80
70
60
t = 30 s
50
40
30
t = 600 s
20
10
0
t = 2400 s
With TP
Fig. 28
Without TP
Successive change of cross section with martensite fraction (central axis on left). TP, transformation plasticity
(a)
(b)
Iinitial
Experimental
Induction Hardening and Carburized

Quenching of Ring
A ring-shaped body is widely used as a typical
example of fundamental quenching tests, since
it is also considered as a model of a gear wheel
with boss hole. From this viewpoint, two kinds
of simulations are carried out: one is induction
hardening (Ref 11), and the other is carburized
quenching (Ref 12), discussed in next section.
A ring with the dimension of 100 and 75 mm
in outer and inner diameter, respectively, and 24
mm in height has the upper half modeled as the
axisymmetric problem. Induction heating is
made on the outer surface with expected hardening depth of 2 mm during 4.0 s heating period,
followed by cooling (Ref 14). Materials employed are carbon steel (JIS-S45C), Cr-Mo steel
(SCM440), and Ni-Cr steel (SNC815) with hardenability and stainless steel (SUS304) without
hardenability for comparison.
90
t = 15 s
Fig. 29
Comparison of distorted cross section with experiment. (a) Carburized quenching. (b) Normal quenching
Fig. 30
Comparison of residual stresses with experiment. (a) Carburized quenching. (b) Normal quenching
Fig. 31
Additional magnetic field to metallo-thermomechanical coupling
The heat generation during induction heating

is assumed to occur in the area of 2.0 mm depth
from the outer surface. The heat-generation rate
is determined by trial and error so that the maximum temperature at the surface reaches the
measured value. All surfaces are regarded as adiabatic boundary during heating process, while in
cooling process the outer surface is defined as
the convection boundary with uniform and constant convection coefficient 20,000 W/(m2 K).
To know the effect of hardenability, or phase
transformation, on some parameters, calculation
of temperature, structures and distortion at typical locations of the ring are compared for typical
two steels of plain carbon steel and SUS304 in
Fig. 20, and the difference in plastic strain distributions for four steels is depicted in Fig. 21.
Such effect is characterized for distortion for
four steels as seen in Fig. 22 with triangles indicating measured data; volumetric dilation due
to martensitic transformation is seen to give the
expansion of radius (see Fig. 22a, b, and c),
while contraction in diameter occurs for SUS304
(Fig. 22d). Figure 23 depicts the distributions of
tangential and radial residual stresses on the top
surface where a wavy pattern of stress is observed in SUS304 steel. Tangential and axial
stresses on the outer surface are shown in Fig.
24.
Carburized Quenching of Ring

To investigate the more detailed behavior of
the influence of transformation plasticity on carburized quenching, a ring of chromium-molyb-
Fig. 32
denum steel (SCM420) with 25 and 75 mm inner

and outer diameter, respectively, with 10 mm in
height is employed for the simulation (Ref 12).
Carburizing and diffusion were performed in
the atmosphere of controlled butane gas keeping
carbon potential Cp at 1.2% for 4 h and 0.85%
for 1 h at 930 C, respectively, followed by intermediate period for 1 h and steady temperature
stage of 850 C for 30 min with Cp 0.85%.
Consequently, quenching was carried out with a
hot oil quenchant of 130 C and specimen was
kept in oil bath for 10 min. Flat surface of the
specimen was held horizontally in both carburizing and quenching.
Figure 25 represents the calculated time-dependent distribution of diffused carbon content
with the experimental data by means of electron
probe microanalysis (EPMA) from flat surface
to 2.5 mm in depth. Carbon seems to diffuse beneath the surface with carburizing time, and the
gradient becomes lower with diffusion time. Experimental data measured by EPMA after the
process of 23,400 s show good agreement with
calculated results.
Different values of heat transfer coefficient
(Fig. 26) are applied to each surface according
to the experimental data by Bodin and Segerberg
(Ref 46). After 10 min elapsed, air cooling was
executed until complete cooling with steady heat
transfer coefficient of 200 W/(m2 C).
The simulated variation of radial distortion
with respect to time at the middle point on outer
and inner faces of quenched ring in both conditions with and without consideration of transformation plasticity is depicted in Fig. 27(a) for carburized and normal quenching, respectively.
Finite-element discretization of a gear and induction coil
1200
Simulated Measured
Tip
PCD
Bottom
Temperature (T ), C
1000
Tip
PCD
800
Fig. 34
Generated heat pattern. t (a) 0.0 s. (b) 1.8 s. (c) 2.7 s. (d) 2.88 s
tt=13,500 s
600
400
9,000 s
200
0
Fig. 33
3
Time (t ), s
Simulated and measured temperature variation
Fig. 35
Temperature pattern. t (a) 1.8 s. (b) 2.88 s
Fig. 36
In the case of carburized quenching, radial displacement both at outer and inner faces is moderate when considering the effect in comparison
with that without transformation plasticity. In the
case of normal quenching, on the other hand, the
effect on radial displacement is smaller, as
shown in Fig. 27(b).
Variations of the simulated cross-sectional
distortion with and without the effect are shown
in Fig. 28 with the volume fraction of martensite.
Outer sides of ring in different conditions at ap-
Pattern of martensite fraction
Volume fraction of martensite, m
1.0
0.8
Tip
PCD
PCD
0.6
0.4
0.2
0
9,000 s
Tip
0
Fig. 37
Root
4
3
2
1
Distance from surface (d ), mm
Distribution of martensite fraction
proximately 8 s are similarly distorted to the bottom due to the difference of heat transfer coefficient among the surfaces. Since martensite
transformation starts around at 15 s from the corner for normal quenching (see Fig. 28b), the concave shape of cross section takes place, while a
similar shape is seen in the final stage for carburized quenching without transformation plasticity effect. The effect appears at 30 s in the case
of carburized quenching to give a convex shape
being kept until end of cooling at 2400 s.
Figure 29 illustrates the simulated distortion
of cross section at final step under both conditions. Solid lines in the figure represent the estimated pattern of distortion from initial shape
shown in thin dotted line when considering the
effect. Experimental results of distortion are
plotted in solid marks in the figure to compare
with simulated patterns. The most remarkable
difference between surfaces with transformation
plasticity and those without in carburized
quenching lies in the mode of distortion as depicted in Fig. 28 and 29. The transformation
plasticity gives a convex shape to all surfaces,
but surfaces without transformation plasticity are
concave. Experimental data agree well with simulated distortion with the effect in both carburized and normal quenching.
Residual stress distribution for carburized and
normal quenched ring are represented in Fig. 30.
It is concluded that the transformation plastic
strain contributes remarkable effects on distortion of ring-shaped models especially in the case
of carburized quenching.
Dual-Frequency Induction Hardening

of a Gear Wheel
Fig. 38
Distortion of a tooth. (a) Dual frequency. (b)

Single frequency
Fig. 39
Heat transfer coefficient depending on clay

thickness
Dual-frequency induction hardening is conducted for suitable appropriate preheating period

by medium frequency (310 kHz) and a rapid
subsequent heating by high frequency (100250
Fig. 40
Pattern of pasted clay
Fig. 41
Finite-element meshes. (a) Global view. (b) Near the tip

kHz) followed by spray quenching. This process
makes it possible to achieve contour hardening,
and the advantage of the process is to provide
high hardness and residual compressive stresses
on the surface compared to small distortion with
enough toughness in the core part. The pattern
of frequency used in the present experiments is
3 kHz for 1.8 s in preheating and 150 kHz for
0.18 s during subsequent heating (Ref 13).
Fig. 42
Effect of clay thickness on martensite distribution
Fig. 43
Successive mode of distortion with temperature variation
Combining two kinds of coupling analyses,

metallo-thermo-mechanical and electromagnetic, to evaluate heat generation during heating
and to introduce the result into the calculation of
metallo-thermo-mechanical fields is performed
by the HEARTS and MAGNA, for electromagnetic field analysis (see Fig. 31). The magnetic
permeability generally depends on the magnetic
field as well as on temperature. One assumes,
Fig. 44
however, in the first approximation of practical

calculation, that permeability is constant in the
workpiece during heating since the magnetic
field is almost steady, but that the magnetic permeability and conductivity are functions of temperature.
The workpiece treated throughout the study is
a spur gear of 0.45% C steel (JIS-S45C). The
dimensions are 126 mm outside diameter, 120
Successive mode of fraction of martensite (a) and pearlite (b)

mm in the pitch circle diameter (PCD) (or module 3), 66 mm in boss hole diameter, and 20 mm
in tooth width with 40 teeth. A one-turn coil with
an inner diameter of 130 mm and a cross-sectional area of 22 by 35 mm2 is equipped outside
of the gear wheel. Figure 32(a) shows finite-element division of a gear and coil. The workpiece
is enlarged in Fig. 32(b).
The simulated and experimental temperature
variations at three typical locations such as the
tooth tip, PCD, and the bottom in dual-frequency
induction hardening are depicted in Fig. 33. The
simulated distribution of heat generation at the
beginning of heating t 0.0 s and the end of
heating at t 1.8 s during preheating are shown
in Fig. 34(a) and (b), respectively. Figures (c)
and (d) represent the distributions at the beginning of heating t 2.7 s and the end of heating
t 2.88 s in the subsequent heating process.
The simulated temperature variation at the
tooth bottom is delayed, as seen in Fig. 33, but
the maximum temperature is higher compared to
the measured one in preheating. In the subsequent heating stage, the temperature variations
agree well with measured data except at the tooth
bottom saturating to around 800 C. One of the
reasons for this disagreement may come from
rough mesh division in depth direction.
Fig. 45
Successive mode of longitudinal stress
Simulated results of temperature distribution

at the end of the preheating and the subsequent
heating in the dual-frequency process are shown
in Fig. 35(a) and (b), respectively. Volume fraction of martensite after quenching is also presented in Fig. 36, and Fig. 37 shows the distribution of martensite fraction, the pattern of
which is similar to measured values of Vickers
hardness.
Figures 38(a) and (b) represent the comparison of simulated distortion for induction hardening using dual frequency and conventional,
single frequency. Actual distortion of 5 lm is
expressed by 2.0 mm in the geometry scale.
When comparing the two cases, the distortion of
conventional induction hardening in the tooth
width direction is about two times as much as
that of dual-frequency induction hardening. The
typical characteristic of dual-frequency induction hardening that less distortion and deformation occur due to hardening only in the contour
zone is confirmed by these results.
Quenching of Japanese Sword

Quenching of a Japanese sword is one of the
typical examples of quenching processes. The
sword master controls the initial heating temperature and that of water as well as the heat
transfer coefficient by pasting a kind of clay on
the surface. The blade pattern with martensite
structure, called hamon, and distortion or bent
shape, termed sori, are evaluated from artistic
viewpoint.
Most Japanese swords with some exceptions
are made of steel called tamahagane, or traditional Japanese steel, specially prepared by the
tatara system using iron sand, not normal iron
ore. The hammering process is repeated to obtain
a block, where about 15 folds, called orikaeshi,
are repeated to get laminated materials bonded
by the mechanism of mechanical alloying.
Before quenching, mixed clay with powders
of charcoal and whetstone, yakibatsuchi, is
pasted on the surface of the sword to control the
heat transfer coefficient depending on the thickness (see Fig. 39). The thickness on the back is
about 1 mm and 0.1 mm on the blade, as shown
in Fig. 40. Then, the sword is heated up to about
800 C in the furnace and quenched in water of
40 C. During this process, the tip is severely
cooled to be the blade made of martensite structure, while the other part is transformed into
pearlite and troostite. Very complicated deformation, or sori, is observed during the process
due to the time delay of cooling and phase transformation, and the stress changes to induce the
residual stresses.
Three-dimensional finite-element mesh division of the sword is represented in Fig. 41, which
shows the sword cut in half in the width direction
for symmetry. This model is supposed to consist
of two regions to which different material data
are applied and also two parts of the surface with
heat transfer coefficients of different values.
To investigate the effect of the clay thickness
on the martensite formation corresponding to the
width of blade, quenching simulations are made
with different patterns of pasting the clay. Figure
42 shows the volume fractions of martensite after quenching under different conditions. When
the sword is quenched without any clay, martensite hardly appears except for on the blade (Fig.
42a). However, almost the entire region becomes
martensite when thin clay is pasted on the entire
surface as shown in Fig. 42b. If the clay is pasted
thin on the blade side, and thick on other part,
an ideal pattern of martensite is obtained, where
martensitic hardened blade and ductile, pearlitic
main body are realized (Fig. 42c). The border of
the martensite structure in white color is termed
hamon.
of the sword with time variation from the beginning of the quenching, and the mode of enlarged
deformation is also depicted in the figure. The
thin part of the blade shrinks by thermal contraction due to severe cooling, which leads to
downward bending, termed as gyaku-sori, or reverse bending. However, when martensitic transformation starts to occur in that part, normal
bending, or sori, to the upper direction is observed due to the volumetric dilation by martensite formation. Gyaku-sori again appears, be-
3.
4.
5.
6.
7.
8.
Distribution of residual stress along typical
three portions. (a) Hasaki, edge (b) Shinogi,
ridge (c) Mune, back surface
Fig. 46
cause of the pearlitic transformation in the back

part.
In the successive stages, the back side shrinks
gradually because of thermal contraction, and finally normal bending can be obtained owing to
the difference of the coefficient of dilation by
martensite and pearlite transformation, as shown
in Fig. 44. The change in the longitudinal stress
distribution on the surface and the inside of the
sword is represented in Fig. 45. The calculated
residual stresses in the longitudinal direction on
typical three locations of the sword being equivalent to the stress distribution at t 30 s are
depicted in Fig. 46, and the data are compared
with experimental data denoted by solid circles
by x-ray diffraction technique.
9.
10.
11.
REFERENCES
1. T. Inoue, S. Nagaki, T. Kishino, and M.
Monkawa, Description of Transformation
Kinetics, Heat Conduction and Elastic-Plastic Stresses in the Course of Quenching and
Tempering of Some Steels, Ing.-Archiv.,
Vol 50 (No. 5), 1981, p 315332
2. T. Inoue and Z.G. Wang, Coupling Phenomena between Stress, Temperature and
Metallic Structures in the Process with
Phase Transformation, Proc. of the Inter-
12.
13.
national Symposium on Calculation of Internal Stresses in Heat Treatment of Metallic Materials, Linkoping Univ. Press, 1984,
p 298310
T. Inoue, Metallo-Thermo-Mechanical CouplingApplication to the Analysis of
Quenching, Welding and Continuous Casting Processes, Berg Huttenmann. Monatsh.,
Vol 132 (No. 3), 1987, p 6371
T. Inoue, Inelastic Constitutive Relationships and Applications to Some Thermomechanical Processes Involving Phase
Transformation, Thermal Stresses V, R.B.
Hetnarski, Ed., North Holland, 1988
T. Inoue, Metallo-Thermo-Mechanics
Application to Phase Transformation Incorporated Processes, Trans. JWRI, Special Issue on Theoretical Prediction in Joining and
Welding, Vol 25 (No. 2), 1996, p 6987
T. Inoue, Residual Stress and Distortion
Metallo-Thermo-Mechanics Simulation of
Engineering Processes Incorporating Phase
Transformation, Mathematical Modeling of
Weld Phenomena, H. Cerjak, Ed., IOM
Communications Ltd., Vol 4, 1998, p 547
575
T. Inoue and B. Raniecki, Determination of
Thermal-Hardening Stresses in Steels by
Use of Thermoplasticity Theory, J. Mech.
Phys. Solids, Vol 26 (No. 3), 1978, p 187
212
T. Inoue, K. Arimoto, and D.Y. Ju, MetalloThermo-Mechanical Simulation of Quenching ProcessTheory and Implementation
of Computer Code HEARTS, Proc. First
Int. Conf. Quenching and Control of Distortion (Chicago, IL), ASM International,
2225 Sept 1992, p 205212
T. Inoue, K. Arimoto, and D.Y. Ju, Theory
and Implementation of Heat Treatment
Simulation Program HEARTS, Proc.
Eighth Int. Congress on Heat Treatment of
Materials (Kyoto), 1720 Nov 1992, Japan
Heat Treating Society, p 569572
T. Inoue and K. Arimoto, Development and
Implementation
of
CAE
System
HEARTS for Heat Treatment Simulation
Based on Metallo-Thermo-Mechanics, J.
Mater. Eng. Perform., Vol 6 (No. 1), 1997,
p 5160
T. Inoue, H. Inoue, T. Uehara, F. Ikuta, K.
Arimoto, and T. Igari, Simulation and Experimental Verification of Induction Hardening Process for Some Kinds of Steel,
Proc. Second Int. Conf. Quenching and
Control of Distortion (Cleveland), ASM International, 47 Nov 1996, p 5562
S. Yamanaka, T. Sakanoue, S. Yosii, T. Kozuka, and T. Inoue, Influence of Transformation Plasticity on the Distortion of Carburized Quenching Process of Cr-Mo Steel
Ring, Proc. 18th Conf. Heat Treating, 12
15 Oct 1998, ASM International, p 657
664
T. Inoue, H. Inoue, F. Ikuta, and T. Horino,
Simulation of Dual Frequency Induction
Hardening Process of a Gear Wheel,
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
Proc. Third Int. Conf. Quenching and Control of Distortion (Prague), ASM International, 1999, p 243250
T. Inoue, T. Uehara, H. Ikuta, and I. Miyata,
The Japanese SwordHeat Treatment Process Simulation Incorporating Phase Transformation, Proc. Int. Conf. Materials and
Mechanics 97 (Tokyo), Japanese Society of
Mechanical Engineering, 1997, p 137142
T. Inoue, The Japanese SwordThe Material, Manufacturing and Computer Simulation of Quenching Process, Mater. Sci.
Res. Int., Vol 13 (No. 4), 1997, p 193203
T. Inoue, The Japanese Sword in Comparison of Others, Proc. Eighth Int. Conf. Mechanical Behaviour of Materials (Victoria,
Canada), 1999, p 458463
F.G. Rammerstorfer, D.F. Fischer, W. Mitter, K.J. Bathe, and M.O. Snyder, On
Thermo-Elastic-Plastic Analysis of Heat
Treatment Processes Including Creep and
Phase Changes, Comp. Struct., Vol 13,
1981, p 771779
J.B. Leblond, G. Mottet, J. Devaux, and J.C.
Devaux, Mathematical Models of Anisotropic Phase Transformation of Steels and
Predicted Plastic Behaviour, Mater. Sci.
Technol., Vol 1, 1985, p 815822
G. Besserdich, B. Schoaltes, H. Muller, and
E. Macherauch, Consequence of Transformation Plasticity on the Development of
Residual Stresses and Distortion during
Martensitic Hardening of ASE 4140 Steel
Cylinder, Steel Research, Vol 65 (No. 1),
1994, p 4146
P.H. Shipway and H.K.D.H. Bhadesia, The
Effect of Small Stress on the Kinetics of the
Bainite Transformation, Mater. Sci. Eng.,
Vol A201, 1995, p 143149
T. Inoue, Z.G. Wang, and K. Miyao, Thermal and Phase Transformation Stresses in a
Carburized Quenched Gear Wheel, Proc.
32nd Japan Cong. Materials Research, Society Materials Science, Japan, 1989, p 21
26
E. Gautier, A. Simon, and G. Beck, Deformation of Eutectoid Steel during Pearlitic
Transformation under Tensile Stress, Proc.
Fifth Int. Conf. Strength of Metals and Alloys (ICSMAS), P. Haasen, V. Gerold, and
G. Kostorz, Ed., 1980, Vol 2, p 867873
S. Denis, A. Simon, and G. Beck, Estimation of the Effect of Stress/Phase Transformation Interaction when Calculating Internal Stress during Martensitic Quenching of
Steel, Trans. ISIJ, Vol 22, 1982, p 504513
M. Fujita and M. Suzuki, The Effect of High
Pressure on the Isothermal Transformation
in High Purity Fe-C Alloys and Commercial
Steels, Trans. ISIJ, Vol 14, 1974, p 4453
S. Bhattacharyya and G.L. Kehl, Isothermal
Transformation of Austenite under Externally Applied Tensile Stress, Trans. ASM,
Vol 47, 1955, p 351379
H. Onodera, H. Gotoh, and I. Tamura, Effect
of Volume Change on Martensitic Transformation Induced by Tensile or Compressive Stress in Polycrystalline Iron Alloys,
27.
28.
29.
30.
31.
32.
33.
34.
Proc. First JIM Int. Symposium on New Aspects of Martensitic Transformation, Japan
Inst. Metals, 1976, p 327332
J.P. Patel and M. Cohen, Criterion for the
Action of Applied Stress in the Martensitic
Transformation, Acta Metall., Vol 1, 1953,
p 531538
S.V. Radcliffe and M. Schatz, The Effect of
High Pressure on the Martensitic Reaction
in Iron-Carbon Alloys, Acta Metall., Vol 10,
1962, p 201207
R.M. Bowen, Theory of Mixture, Continuum Physics, Vol 3, A.C. Eringen, Ed., Academic Press, 1976, p 2129
W.A. Johnson and R.F. Mehl, Reaction Kinetics in Processes of Nucleation and
Growth, Trans. AIME, Vol 135, 1939, p
416458
C.L. Magee, Chapter 3, The Nucleation of
Martensite, American Society for Metals,
1968
G.A. Maugin, The Thermodynamics of Plasticity and Fracture, Cambridge University
Press, 1992
T. Inoue and K. Tanaka, Subsequent Yield
Conditions of Metal under Cyclic Loading
at Elevated Temperature, Ing. Archiv., Vol
44, 1975, p 5362
W. Prager, A New Method of Analyzing
Stress and Strains in Work-Hardening Plastic Solids, J. Appl. Mech., Vol 23, 1956, p
493496
35. H. Ziegler, A Modification of Pragers

Hardening Rule, Quart. Appl. Math., Vol
17, 1959, p 5565
36. T. Inoue and K. Tanaka, An Elastic-Plastic
Stress Analysis of Quenching when Considering a Transformation, Int. J. Mech. Sci.,
Vol 17, 1975, p 361367
37. Y. Yamada, N. Yoshimura, and T. Sakurai,
Plastic Stress-Strain Matrix and Its Application for the Solution of Elastic-Plastic
Problems by the Finite Element Method,
Int. J. Mech. Sci., Vol 10, 1968, p 343
354
38. G.W. Greenwood and R.H. Johnson, The
Deformation of Metals under Small Stresses
During Phase Transformation, Proc. R. Soc.
(London) A, Vol 283A, 1965, p 403422
39. T. Inoue and Z.G. Wang, Coupling between
Stresses, Temperature and Metallic Structural During Processes Involving Phase
Transformation, Mater. Sci. Technol., Vol
1, 1985, p 845850
40. T. Yamaguchi, Z.G. Wang, and T. Inoue,
Analysis of Temperature, Metallic Structure
and Stress During Carburized Quenching of
a Gear with Consideration of Carbon Content, J. Soc. Mater. Sci. Jpn., Vol 33, 1984,
p 14701476 (in Japanese)
41. T. Inoue, T. Yamaguchi, and Z.G. Wang,
Stresses and Phase Transformations Occurring in Quenching of Carburized Steel Gear
42.
43.
44.
45.
46.
Wheel, Mater. Sci. Technol., Vol 1, 1985, p

872876
K. Miyao, Z.G. Wang, and T. Inoue, Analysis of Temperature, Stress and Metallic
Structures in Carburized-Quenching Gear
Considering Transformation Plasticity, J.
Soc. Mater. Sci. Jpn., Vol 35, 1986, p 1352
1357 (in Japanese)
T. Inoue, Z.G. Wang, and K. Miyao, Simulation of Quenching Process of Carburized
Steel Gear under Metallo-Thermo-Mechanical Coupling, Proc. IUTAM Symposium on
Thermomechanical Coupling in Solids
(Paris), 1986, p 257262
Z.G. Wang and T. Inoue, Analysis of Temperature and Elastic-Viscoplastic Stress during Continuous Casting Process, Proc. International Conference on Computational
Mechanics (Tokyo), Vol VIII, Japan Soc.
Computational Mechanics, 1986, p 103
108
G. Besserdich, M. Ehlers, Th. Lubben, A.
Majorek, G. Schmitt, and D. Wiedmann,
Weniger Verzung beim Harten durch Computersimulation, Hart.-Tech. Mitt., Vol 54,
1999, p 201208
J. Bodin and S. Segerberg, Further Investigations of Quenching Conditions and Heat
Transfer in Hardening Steel Components,
Proc. Ninth Int. Congress on Heat Treatment and Surface Engineering (Nice), Ivrysur-Seine, 2228 Sept 1994

DOI: 10.1361/hrsd2002p312

Control of Residual Stress Formation and

Steel Deformation during Rapid
Heating and Cooling
N.I. Kobasko, V.S. Morganyuk, and V.V. Dobrivecher, Ukraine National Academy of Science, Kiev
THE ADVANTAGES OF INTENSIVE

HEATING AND COOLING when quenching
alloy and high-alloy steels are discussed in this
article. Alloy steels are quenched in mineral oils
or high-concentration aqueous polymer solutions,
which can replace the oil. Mineral oils retard the
cooling process in the area of the transformation
of austenite into martensite. Proceeding from numerous experimental data, it has been considered
a firmly established fact that intensive cooling in
the field of martensite transformations results in
the formation of quench cracks and, afterward,
extensive deformation of parts after their
quenching.
In the late 1990s, many papers appeared based
on experimental and calculating data that prove
that starting with definite intensity of cooling,
the heat-transfer intensification results in avoiding the formation of quench cracks and reduction
of the deformation of quenched parts. This allows a change in the technology of alloy steel
quenching in principle, because instead of expensive and fire-dangerous oils one can use water (Ref 1, 2).
This article discusses the mathematical model
and gives the analysis of results of computer
simulation and physical grounds for the results
obtained. This article considers in detail two
steel-quenching methods: intensive and uniform
cooling of alloy steel parts during the entire cooling interval and the influence of the growth of
intensity of cooling on distortion and crack formation of quenched parts is studied.
Two-step quenching is investigated. It has
been shown that one can reach minimum distortion at the first stage, which can be preserved
during intensive cooling at the second interval
of the transformation of austenite into martensite.
On the basis of the Pisarenko-Lebedev criterion, it has been shown that one can prevent the
formation of quench cracks during intensive
quenching at both the first and second stages of
cooling (Ref 3). The criterion of Pisarenko-Lebedev establishes the conditions of the quench-
crack formation. If the strength factor is greater

than one, quench cracks will not form. If it is
less than or equal to one, quench cracks will
form.
It is emphasized that the reduction of distortion and prevention of the formation of quench
cracks are due to the formation of hard martensite layer on the entire surface, which fixes the
initial sizes of the part and forms high residual
compressive stress on the surface quenched
parts. Examples of the application of intensive
steel-quenching methods for alloy steel parts are
given in practice.
Mathematical Model for the

Calculation of Thermal and StressStrain State
The coupled equation of nonstationary thermal conductivity, as known, is given as:
cq
T
div(k grad T) rij e ijp qI lI n I 0
s
with corresponding boundary conditions for film

boiling:
T
f
(T Ts)|rR 0
r
k
T
bm
(T Ts)m|rR 0
r
k
T(r, sf) u(r)
Finally, at the area of convection heat transfer,

the boundary conditions are analogous to those
for film boiling, that is:
T
cn
(T Tc) 0
r
k
T(r, snb) w(r)
The condition of transfer from nucleate boiling

to one-phase convection is given by qnb qconv.
These equations determine boundary conditions of the third kind and describe conditions of
the change of heat-transfer regimes. However, in
practice, one often uses boundary conditions of
the first kind, which can be obtained from experiments. In this case, the main differential
equation of heat conductivity must be added
with the aforementioned conditions, that is:
T(ri, s)|riRi u(Ri, s)
where i is, in this case, the coordinate index, and

u(Ri, s) is a value of temperature at different
points of the surface, which is being cooled.
Plasticity theory equations are presented in
detail in Ref 4 and 5 and have the following
form:
and initial conditions:

T(r, 0) T0
The transition from film boiling to nucleate

boiling is made when the following is fulfilled:
qcr2 f(Treg Ts)
where qcr2 0.2 qcr1.

At the stage of the nucleate boiling, the
boundary conditions appear as:
e ij e pij e eij e Tij e ijm e tp

ij
with relevant initial and boundary conditions,

where T is temperature, rij is stress, e ij is total
strain rate, e eij is elastic strain rate, e pij is plastic
strain rate, e Tij is thermal strain rate, e m
ij is strain
rate for structural dilation due to phase transformations, e tp
ij is strain rate for structural dilation
due to transformation plasticity, k is thermal
conductivity, and c is specific heat. These equations are described in detail in the article
Control of Residual Stress Formation and Steel Deformation during Rapid Heating and Cooling / 313
Metallo-Thermo-Mechanics: Application to
Quenching in this Handbook.
Crack formation probability, %
100
80
Computation of Stress-Strain State
60
40
20
0
200
400
600
800
Cooling rate, K/s

Effect of cooling rate in the martensite range on
the crack formation in cylinder-shaped samples
of 6 mm diameter, of steel 40X (DIN 41 Cr4), which is
equivalent to AISI 5135
Fig. 1
Input
Discretion of computation region

into finite elements
Output listing of initial data
Time cycle start
Iterative solution of nonlinear problem

of heat conduction
Construction and output of isometric fields

and the material structural composition
On the basis of numerous experimental studies, it has been established that with the cooling
rate increase within the martensite transformation range, the probability of quench-crack formation grows to a maximum value and then lowers to zero (Ref 1) (Fig. 1).
Substantiation of regularity is of great practical and scientific importance. To that end, the
calculation was performed for the thermal and
stress-strain states of steel parts on the basis of
finite-element methods (FEM) (Ref 6). For
mathematical simulation of part quenching, a
mathematical model was used that included a
nonlinear equation of nonstationary heat conduction, equations of the theory of thermal, elastic, and plastic flow with kinematic strengthening under the appropriate boundary conditions of
the first and third kinds (Ref 5).
The calculation process is divided into time
steps. At each time step, a heat-conductivity
problem is solved. The resulting temperature
field determines mechanical characteristics and
temperature load for the thermal plasticity problem. The initial conditions for thermal conductivity and thermal plasticity problems are taken
from results of solving the corresponding problems at previous steps. Besides, at each time step
the calculation results are compared with the
thermokinetic diagram of the supercooled austenite transformations, and new thermal, physical, and mechanical characteristics for the next
step are chosen depending on the structural components (Ref 7). In every separate region of the
diagram a specific transformation law is established, if necessary. The calculations are made
Iterative solving the problem of

thermoelastoplasticity
800
600
Construction and output of fields of

relocations, strains, and stresses
400
i
, MPa
Evaluation of possibility of destruction

and calculation of safety factor
200
Construction and output of the

field of safety factor
r
0
e
Time cycle end
200
400
0
Output
Fig. 2
Chart describing the algorithm of the software

package
0.005 0.01 0.015 0.02 0.025 0.03 0.035

Radius of the specimen, R, m
Fig. 3
stress
Stresses at the end of the specimen heating. rr,

radial stress; rh, circumferential stress; rz, axial
until the part is cooled. These regularities are

described in detail in Ref 8. Joint consideration
of these issues (the selection of calculation methods, scheme of the phase-transformation diagram, boundary conditions, and physical and
mechanical properties of the material) is of great
practical and scientific interest.
The principal block diagram of the computing
complex developed at the Institute for Strength
Problems with the participation of Engineering
Thermophysics Institute, both part of the National Academy of Sciences of Ukraine, is
shown in Fig. 2. The calculation results are: temperature field, material phase composition, migration of points in the volume which is calculated, components of stress and strain tensors,
intensities of stress and strain, and the field of
the safety factor, that is, a relation of stresses at
which the material is destroyed. These values are
presented in the form of tables and isometric
lines that allow observation of the kinetics of
phase changes in the process of heating and cooling (Ref 810). Using the potentialities of the
computing complex, the current and residual
stresses were determined, depending on the cooling capacity of quenchants. Calculations were
also fulfilled for quenching parts of complex
configuration.
With the purpose of testing these computation
methods, a problem has been solved allowing
comparison of the computational data with experimental ones. A specimen for this purpose
was cylinder-shaped, made out of steel 45
(0.44% C), diameter of 65 mm, length of 130
mm. Quenching was made by induction, and
cooling was under intensive water shower. The
thermal plasticity problem at heating was solved
based on the assumption that the temperature
distribution is the same at time heat ends and
cooling starts and is known (in Fig. 6, this is a
curve at time 0 s). For this purpose, the properties from handbooks for this grade of steel
have been used. Figure 3 shows results of solving the heating problem.
The cooling conditions have been set in the
form of boundary conditions of the first kind
(Fig. 4, curve 1). The initial conditions are determined by results of solving the heating problem. The temperature and time dependencies of
density q and linear expansion coefficient have
been calculated with the use of Yuriev formulas
(Ref 11) and thermokinetic diagram (Fig. 4). The
other dependencies have been built on the basis
of experimental data for dependencies of material properties on the temperatures with corrections on the basis of continuous cooling transformation (CCT) diagrams. Figure 5 shows
isochronous curves of dependencies of mechanical properties on temperature at time, s 1 s.
Results of solutions of problems of thermal
conductivity and thermal plasticity for cooling
process give the full picture of functions of temperature field, structure, and stress-strain state
for part quenching. Figure 6 shows the change
in temperature field for cooling of the specimen.
The comparison of cooling data and CCT diagram established that the depth of the layer con-
1100
25%
0%
1000
3
100%
75%
50%
900
ments have been made in the search for intensively cooling quenchants, such as aqueous solutions of salts with special admixtures
preventing surface corrosion of quenched parts.
Optimal concentrations of aqueous polymer solutions have been found, which exhibit the inverse solubility. Such solutions provide intensive and uniform cooling of parts to be
quenched.
Circular residual stresses on the surface of cylindrical bodies versus generalized Biot number
are presented in Fig. 9. Intensive quenching
methods suppose that after quenching on the sur-
400
200
0
200
400
700
0%
600
600
1300
10%
0.2 s
800
400
0s
30%
500
50%
1100
1s
70%
400
90%
2s
900
102
10
103
104
, s
T, K
1
300
101
200
z, MPa
T, K
800
axial stresses to become negative and remain

compressive until cooling was complete. Figure
8 shows calculated residual stresses compared
with experimental data. One can consider that
the results of calculations agree well with experimental data.
The most important results used for developing the intensive methods of quenching are discussed next. It has been stated that with the
quench-intensity increase, the residual stresses
grow at first, then lower, with further increase in
Biot number and change over to compression
stresses. Intensive quenching can be reached
through the application of jet cooling, directed
streams of water, and aqueous solutions providing high critical heat flux densities. In the aforementioned conditions, the boiling film is reduced
to the minimum or eliminated. The basic processes are nucleate boiling and one-phase convection heat transfer. In this case, it is important
to know conditions of the transfer from nucleate
boiling to one-phase convection heat transfer.
On the basis of this approach, a lot of experi-
e, MPa
taining more than 50% martensite was 5 mm; the

total depth of the hard layer was 7.5 mm. Figure
4 shows curves (1) of the change in temperature
on the surface of a specimen, (2) at a depth of 5
mm, and (3) at a depth of 7.5 mm.
Figure 7 shows circumferential stresses at
characteristic points versus time. At the beginning of cooling, the stresses on the surface were
tensile because of the thermal reduction of the
material. From 0.15 s on the surface martensite
transformation started, and an abrupt extension
of the steel resulted, causing circumferential and
3s
0
200
700
400
Continuous cooling transformation diagram of

cooling austenite transformation for steel 45
Fig. 4
4s
600
500
300
0
1000
0
0
iment.
T, 106/K
0.2
400
400
3
200
400
4
2
200
300

2
200
400
300
Residual stress ( ), MPa
, MPa
0
0
200
1
400
600
500
700
900
Residual stresses in the specimen. Solid lines relate to calculations; dotted lines relate to exper-
Fig. 8
0.3
y
1100
T, K
Isochronous curves exhibiting dependencies of
mechanical properties on temperature. Dashed
lines relate to handbook data; solid lines relate to data used
in experiments. ru, ultimate strength; ry, yield strength; T,
coefficient of linear expansion
0.005 0.01 0.015 0.02 0.025 0.03 0.035

R, m
Temperature field at the time of cooling of the

specimen
Fig. 6
u
600
200
Radius of the specimen, R, m
0.4
800
u, y, MPa
0.005 0.01 0.015 0.02 0.025 0.03 0.035
0.5
r, MPa
400
1200
800
102
101
10
102
103
, s
Fig. 5
Circumferential stresses for cooling of the specimen. 1, on the surface; 2, at 5 mm depth; 3, at

7.5 mm depth; 4, at the axis
Fig. 7
200
100
0
100
200
300
400
Fig. 9
2.5
5
7.5
Biot number
10
12.5
Circular residual stresses versus generalized Biot

number
face residual compressive stresses always appear
(when Biv 4.5). Working within the range
where Biv 4.5, one can avoid cracking. Note
that between the generalized Biot number and
usual Biot number there is a link, which follows:
Biv
S
K
k V
where Biv is generalized Biot number, is heattransfer coefficient, k is heat conductivity, K is

Kondratiev form coefficient, S is surface, and V
is volume.
Some results of calculation of current surface
stresses of parts to be quenched are shown in Fig.
10. During quenching, great compressive
stresses appear on the surface, and then they disappear. Maximum compressive surface stresses
correspond to the optimal depth of hard layer.
The technology to fix these stresses has been developed. Figure 10 shows current surface
stresses and inside the part at slow and intensive
cooling. As one can see from Fig. 10, the higher
the cooling intensity, the greater the value of
maximum compression stresses on the surface of
hard parts.
Possibility of Predicting Hardening

Cracks
It is well known that intensification of quenching processes improves the hardening characteristics and strength of steel articles, but in this
way it is possible for quench cracks to form. In
order to avoid damage to articles during hardening, in practice one often uses oil cooling together with alloy steels. Small stresses arise during oil cooling, and alloying promotes sufficient
tempering capacity under these conditions.
However, economically it is more suitable to use

simple carbon steels and water or an aqueous
solution as cooling agent. It is necessary to
achieve cooling methods that would provide sufficient strength and the absence of quench
cracks. For example, this is possible with water
quenching under pressure and in aqueous solutions of polymers, and so forth. Before starting
the manufacture of industrial equipment for realizing such procedures, it is useful to carry out
numerical analysis of different cooling schedules
and to select the most effective of them. This
analysis is possible on the basis of calculating
the thermal and stress-strain state during hardening.
Now consider developing a procedure of
strength calculation for engineering components
during hardening, taking account of phenomena
accompanying steel structural transformations
(Ref 7). On the basis of building a mathematical
model for the hardening process, a suggestion is
made about the dependence of steel thermal,
physical, and mechanical properties on temperature and cooling time, since temperature and
time govern the structural state of steel during
cooling. A complete set of equations for calculating the thermal stress state during hardening
includes a nonlinear thermal conductivity equation and equations for plastic flow theory; these
are given in Ref 8 to 10 and 12.
All computational works have been made as
follows:
1. Computation of thermal and stress-strain
state during heating to the austenite formation
temperature.
2. Final results of heating to the austenite formation temperature (distortions and changes
in sizes of parts) were initial data for cooling
process.
1200
1200
T0
T0
TS
800
800
TS
400
400
0
S
Fig. 10
(Eq 1)
where v rf /rc, rf is ultimate tensile strength,

rc is ultimate compressive strength, r1 is greatest principal stress, and ri is stress intensity. For
an ideal plastic material v 1, and for an ideal
elastic material v 0.
The use of the general criterion (Eq 1) applied
to hardening requires the existence of experimental dependencies of ultimate tensile strength
and ultimate compressive strength on temperature and time rf rf(T,s) and rc rc(T,s) in
accordance with the thermokinetic diagram of
supercooled austenite transformation. Equation
1, confirmed by numerous experiments of different workers, is one of the most widely used
criteria in strength calculations for machine and
structural components since its use requires the
minimum experimental data (only rf and rc).
In order to prove the selected mathematical
model, the hardening process of an inner bearing
ring 7515/02 made of steel ShKh15 (AISI
52100) (Fig. 11) with water cooling was studied.
The points on the surface indicated in Fig. 11
show places where the boundary conditions of
the first kind have been set. Thermocouples were
102
10
31
400
800
101
vri (1 v)r1 rf
Temperature, (T ), K
Stress ( ), MPa
Ms
The results of computations after final heating

and cooling to the temperature of the environment are considered in this article.
Considering the dependence of physical and
mechanical properties on temperature and time
and in accordance with plastic flow theory with
kinematic strengthening, temperature fields and
stress distribution have been calculated.
Solution of the thermal conductivity and thermoplasticity problems gives a complete picture
of the change in temperature field. The thermokinetic diagram for the transformation of supercooled austenite (Ref 7) allows one to follow the
kinetics of phase transformation and to determine the depth of hard layer. By evaluating the
stress-strain state of an article during hardening
using some failure criterion, it is possible to assess the probability of hardening crack formation.
A generalized Pisarenko-Lebedev criterion
(Ref 3) was selected as a failure criterion during
hardening:
11
12
13
75
6
5
4
3
2
1
14
15
8
7
16
1
10
Time (t ), s
Current stresses versus time
102 102
101
1
Time (t ), s
87
10
Fig. 11
Drawing of bearing ring 7515/02. Thermocouple location is indicated by dots.

melted to the surface of the ring, which allowed
boundary conditions of the first kind to be fixed
at many points. The temperature values at intermediary points between thermocouples were
found through the extrapolation of temperature
values obtained through experiment. The experimental and extrapolated values of temperatures
are given in Table 1.
However, methods of calculating temperature
fields and stresses in relation to intensive steelquenching methods with regard to their specifics
need to be developed. As known, the bearing
industry generally uses slowly cooling oils as
quenchants. These oils are very expensive and
can cause fire, but reduce distortion and prevent
the formation of quench cracks. However, there
is an alternative way: intensive cooling, which
must gain acceptance in the bearing industry.
Numerous experiments confirm this idea. The al-
ternative method has great prospects, since water

and aqueous solutions of polymers of optimal
concentration exhibit inverse solubility, are inexpensive, do not cause or support fire, and allow the required values of Biot numbers to be
reached, so that high compression stresses always form on the surface to be quenched. In turn,
the compression stresses on the surface of
quenched parts make their service life longer.
Therefore, the control and determination of the
residual stresses on quenched metal surfaces is
of great practical value.
(a)
(d)
(b)
A record of changes in temperature at different points on the surface and the moment of failure was made for hardening an actual model. Ten
specimens were tested, and all of them were
heated to 1100 K, then cooled in water at 300 K.
As a result of this quenching, cracks of the upper
flange occurred in the radial direction (Fig. 12).
Average time from the start of cooling to the
moment of failure was 4.3. s.
Experimental data (see Table 1) were used as
boundary conditions in calculating the first-kind
thermal conductivity problem. Since it is known
that for the selected brand of steel through-hardening is specified over the entire cross section of
the bearing ringthat is, only martensitic transformation occursthen the dependence of material characteristics on time may be ignored and
only the dependence of thermal, physical, and
mechanical properties of steel ShKh15 on temperature is used in the calculation. In view of the
absence of an experimental temperature dependence of ultimate compressive strength for steel
ShKh15 during hardening, some simplification
was used. It was assumed that in the austenitic
condition and partly in the region of martensite
transformation (510480 K) the material is ideally plastic (v 1). Below 440 K, when more
than 50% of martensite forms in supercooled
austenite, the material becomes ideally brittle (v
0). In the range of 480 to 440 K it was assumed that the dependence of value v on temperature is linear. Consequently:
1 with T 480 K
v (T 440)/40 with 480 K T 440 K
0 with T 440 K
(Eq 2)
(e)
In the course of the calculation at each time

step, in addition to solving the thermal conductivity and thermoplasticity problems, the cooling
path was compared with the thermokinetic diagram in order to determine the structural composition of the material, and a field for the save
factor was plotted:
k(xi,s)
(c)
rf
with re 0, re vri (1 v)r1
re
with re 0
(f)
Structural composition and shape of the bearing ring profile during hardening. (a) Initial
(cold and heated) states. (b) After 0.8 s. (c) 1.6 s. (d) 2.2 s.
(e) 3.2 s. (f) 4 s. The unshaded area is austenite, single
hatching indicates the area up to 50% martensite, and cross
hatching indicates the area of failure
Fig. 13
Fig. 12
Table 1
Bearing ring failure in the case of water

quenching. Sizes of ring are given in Fig. 11.
In addition to Fig. 13, it should be noted that

in the case of very intensive quenching, a thin
martensite layer is formed simultaneously on the
Results of thermometry during water quenching for a bearing ring 7515/02

Temperature for thermocouple No., K
s, s
0.1
0.2
0.3
0.5
1.0
2.0
5.0
10.0
10
11
12
13
14
15
16
1023
773
623
498
413
393
378
348
1048
856
714
573
446
399
380
350
1098
1008
874
699
508
415
385
354
1100
1010
819
644
497
422
390
365
1100
1000
773
598
488
418
395
375
1120
993
848
673
503
428
393
373
1100
1026
898
723
513
413
388
361
1098
1020
897
727
515
415
383
350
1090
1025
902
748
518
417
384
352
1088
1038
925
769
543
428
389
363
1080
1044
937
794
619
447
393
369
1090
1051
951
791
700
469
397
376
1100
1053
953
775
674
664
394
371
1097
1050
949
723
385
445
387
355
1074
1022
898
694
513
413
382
352
1055
880
740
585
455
401
380
350
entire surface independently of the thickness of
the bearing ring, which is the main cause of the
reduction of bearing ring distortion. Thus, very
slow and very intensive cooling reduced the distortion of bearing rings. The advantages of intensive cooling are as follows: the increase in the
hardenability of steel parts and improvement of
mechanical properties of the material.
Thus, as a result of this calculation a complete
picture was obtained for the change in thermal,
structural, stress-strain, and strength states of the
bearing ring during hardening.
In order to build up the shape of the profile
from the original cross section (cold condition),
values of calculated displacements for points on
the contour were plotted for capacity at a magnification of 30 times. Figure 13(a) shows the
profile of the bearing ring cut section in the cold
condition and heated to the austenitization temperature (1130 K); Fig. 13(b) shows the bearing
ring profile after the start of cooling. This instant
corresponds to the set of phase transformation in
the coldest part of the bearing ring (lower
flange). In the time interval of 0 to 0.8 s, there
is only thermal contraction of supercooled austenite. Nonuniformity of temperature field connected with more rapid cooling of thin parts of
the bearing ring leads to very large thermal
stresses exceeding even the yield point and to a
change in cross-section configuration, in particular to the occurrence of bearing ring tapering at
the inner surface. Mainly tension stresses act
over the entire surface of the bearing ring. Starting from 0.8 s in the most highly cooled parts of
the bearing ring, martensite forms which, having
a greater specific volume, leads to material expansion. This involves an increase in diameter
of the lower part of bearing ring, whereas the

central and upper parts are essentially austenite,
which continues to contract (Fig. 13c). Martensite formation in the most highly cooled layers
at first causes a reduction in tensile stresses and
then a transition to compressive stresses. As long
as cooling continues, martensite transformation
proceeds within the cross section, which increases the diameter of both the lower and central part of the bearing ring. This is the reason
for the change in taper to an opposite sign (Fig.
13d). There is some concavity of the inner surface adjacent to the hottest part of the bearing
ring, which is explained by further contraction
of austenite predominant in the area. In the most
highly cooled areas, martensite formation is essentially finished, and therefore with a reduction
in temperature martensite starts to contract,
whereas in the inner layers there is expansion
due to rapid formation of martensite. This leads
to a reduction in compressive stresses in the cold
parts of the bearing ring and to the appearance
of tensile stresses.
The bearing ring profile at 3.2 s is shown in
Fig. 13(e) when the martensite transformation
covers the whole area. With further cooling, the
400 K
100
100
100
440
0
420
200
400
100 MPa
400
300
(a)
(b)
200
200
1.5
100
, MPa
100
0
200
100
100
1.5
Predicting the Deformation of

Bearing Rings during Hardening
100
0
200
1.5
100
200 MPa
300
(c)
1.5
(d)
, s
Nature of change in principal stress rl, stress
intensity rI, equivalent stress re, and ultimate
tensile strength rf with passage of time
Fig. 14
diameter of the lower part increases slightly

since the main mass of austenite has been converted to martensite. In the inner layers of the
upper part of the bearing ring there is rapid formation of martensite, which leads to a site that
predominantly tends to contract thermally. This
leads to a marked increase in circumferential tensile stresses. At 4 s in the area denoted in Fig.
13(f) by cross hatching circumferential tensile
stresses reach values exceeding the ultimate
strength. Typical changes in values of rl, ri, re,
and rf with the elapse of time in the center of
this region are shown in Fig. 14. In the time interval of 0 to 1.8 s according to Eq 2, the equivalent stress re coincides with stress intensity of
ri, and starting from 3.2 s the value of re coincides with the principal stress rl, which is the
circumferential stress (according to Eq 1, the latter stress at 4 s leads to material failure, that is,
the cross-hatched area in Fig. 13f).
Fields for isotherms, axial and circumferential
stresses, and also the reserve strength factor
(only the most critical zones are shown in Fig.
13d) at 4 s are shown in Fig. 15. At this point
the calculation was stopped since the reason, location, and instant of failure have been revealed.
By comparing experimental and calculated
data it is possible to note that cracks form in the
bearing ring due to tensile circumferential
stresses since they spread into planes of meridianal cross sections. The location and time of
crack development coincide satisfactorily. It
should be noted that chipping at the upper surface of the flange in the region of a crack (see
Fig. 12) indicates that the crack propagates from
within (from the area shown in Fig. 13f), branching at the surface.
Thus, the method proposed can be used to predict hardening cracks. This is very important for
intensive steel-quenching methods. As known,
in the case of intensive steel-quenching methods,
quenched compression stresses appear on the
surface of parts. Therefore, during intensive
quenching, quench cracks can appear inside the
part, which are invisible but can result in disruptions. The above-mentioned forecast is especially important for quenching of bearing rings,
which are hardened thoroughly through intensive cooling. On the other hand, in the work process quenchants change their quenching properties. In this condition it is important to predict
the formation of quench cracks by criterion of
Pisarenko-Lebedev discussed previously.
Fields of isolines for temperature (a), axial (b),

and circumferential (c) stresses, and also the
reserve strength factor (d) at the instant for the start of failure
of a bearing ring 7515/02
Fig. 15
Machining allowances should be made with

consideration of the real deformation of components during heat treatment. In plant practice,
the change in the dimensions of components during hardening is evaluated on the basis of experimental-statistical data. Theoretical computation of the change in dimensions makes it
possible to predict machining allowances and to
ascertain basic factors having the most signifi-

the basis of the computation of its temperature
and stress-strain states during heating and cooling carried out in conformity with the method
cited in Ref 3 and 6 (Fig. 18, 19). After the bearing ring had been immersed in oil for 30 s, its
temperature in the most cooled section had been
lowered to 510 K; this corresponded to the outset
of the austenite-to-martensite transformation
(curve 3 in Fig. 19). From the start of cooling to
this moment, the specific volume of the steel decreased, and the generatrix was displaced to the
right owing to the temperature drop. The position of the generatrix after 39 s of immersion in
the oil corresponded to curve 4 (Fig. 19). At this
moment, the basic mass of the austenite had already been transformed into martensite in the
lower section of the bearing ring, as a result of
which the specific volume of the steel, and, correspondingly, even the diameter of the lower
section of the bearing ring increased. In the upper portion of the bearing ring, however, the
martensitic transformation had just begun. According to computations, the final position of the
generatrix corresponded to curve 5 (Fig. 19). The
transformation of austenite into martensite in the
upper portion of the bearing ring effected a
change in the slope of the generatrix.
The mean-statistical values of the displacements of the generatrix of the inner surface in
the case of broad and narrow bearing ring ends,
which were determined experimentally, are
given by the points in Fig. 19. It is apparent that
the computed and experimental displacement
values are very close: the maximum deviation of
the computed from the experimental data was 3
lm (1%). This suggests the correct selection of
the mathematical model of the process. In addition to the comparison of numerical and experimental data relating to distortions, the comparison has been also made for values of residual
stresses in cylinder-shaped samples treated by
induction heating and cooled in water. Reference
2500
100
2000
50
E
1500
Ultimate
strength ,u
Yield strength ,y
1000
50
100
500
0
T, 106/K
E(102), E (101), u, y, MPa
cant effect on the deformation of components.

Determination of the character of the effect of
these factors on current and residual strains is a
very critical and complex problem. The method
used for these investigations is inadequately developed.
Below are some results of the computation of
the strain state during the hardening of an internal bearing ring made of steel ShKh15 (AISI
52100). The heating temperature of the bearing
ring for quenching was 1130 K, while cooling
was carried out in MZM-16 quenching oil with
a temperature of 330 K. Reference data on the
thermal and physical properties of steel ShKh15
as a function of temperature and boundary conditions of the third kind were used to solve the
heat-conductivity problem. The dependence of
the heat-transfer coefficient in the MZM-16 oil
on the surface temperature of the specimen is
shown in Fig. 16 (Ref 13).
Heat-transfer coefficients presented in Fig. 16
have been determined through solving inverse
problems using selection of electric models (Ref
13), which is one of many possible methods of
solving inverse problems (regularization method
and others).
It is known that the stress-strain state of a solid
subjected to hardening is dependent to a significant degree not only on the temperature gradient
in this solid, but also on the variation of the specific volume of structural components during
phase transformations, as well as on the kinetics
of the variation in mechanical properties, for example, the tendency to exhibit superplasticity
(Ref 14). The temperature dependence of the mechanical properties of steel ShKh15 with allowance for the effect of superplasticity, which
arises during cooling at the time of the most vigorous formation of martensite, was used to compute the thermoelastoplastic stress-strain state
(Fig. 17a). The change in the specific volume
during the transformation of austenite into martensite was accounted for by the temperature dependence of the coefficient of linear expansion
(Fig. 17b). Values of the physical and mechanical properties of steel ShKh15 were taken from
Ref 15.
The complete picture of the shape variation of
a bearing ring during hardening was obtained on
400
600
800
T, K
(a)
1000 1200
400
600
800
1000 1200
T, K
(b)
Variation in strength properties, elastic modulus E, hardening modulus E (a), and coefficient of linear expansion T of steel ShKh15 (AISI 52100) with allowance for increase in specific volume as austenite transforms
into martensite (b), as functions of temperature
Fig. 17
260
140
120
420
460
220
100
180
160
6000
380
, W/m2 K
5000
140
180
4000
100
80
3000
340
200
2000
60
220
1000
0
1000
500
T sf,K
Fig. 16
men
60
1500
Dependence of heat-transfer coefficient in

MZM-16 oil on surface temperature of speci-
(a)
240
260
(b)
280
300
20
(c)
(d)
Lines of equal displacements (indicated near curves in lm) in section of bearing ring. (a) and (b) Current radial
displacements after 30 and 39 s, respectively, of immersion in oil. (c) and (d) Residual radial and axial
displacements, respectively
Fig. 18
5 presents this comparison, which shows that
there is good agreement between such numerical
calculations and experimental data.
For mathematical modeling of the quenching
process, different intensities of cooling within
the martensite range were set by giving different
values of coefficients of heat transfer , selected
with regard to water flow rate, sprayer, mixing,
vibration, and so forth. There are formulas allowing determination of heat-transfer coefficients (Ref 14) for all these cases.
It appears that at very intensive cooling within
the martensite range, the conicity of the rings
being quenched is the least, while at moderate
intensity of cooling it is higher (Table 2). Bearing rings are cooled in two steps. The first step
is mineral-oil cooling until the minimum conicity is achieved. At the second step, different
cooling rates determined by the technology are
applied. A hard shell is formed over the entire
surface almost simultaneously at the maximum
cooling rate, and therefore the conicity is at the
minimum, since this shell hampers further deformation of the part through phase changes inside the rings.
Maximum compression and tensile residual
stresses observed over the ring cross section are
given in Table 2. While quenching in circulating
water and in a shower, the residual stresses are
low since the condition BiV 4.5 is satisfied
(Fig. 9). The decrease in conicity at intensive
shower cooling can be explained by simultaneous formation, over the whole surface, of the
martensite layer that fixes the ring initial form.
Under moderate cooling rates, a thin cross section of the ring is cooled at first in which martensite transformations begin that cause displacements and coning formation. By adjusting
the intensity of cooling, one can control the
value of residual stresses and reduce the distortion to the minimum.

This is true for both one-step and two-step
quenching. In the first case, the martensite shell
is formed uniformly on the whole surface at the
time of immersion of the part into the quenchant.
In the second case, a more uniform martensite
shell is formed at the second stage of two-step
cooling.
Figure 20 shows that at slow cooling (a) the
phase transformation starts in the thin section of
the part and due to change in specific volume of
martensite results in the increase in distortion. In
the case of intensive cooling Fig. 20(b), a thin
shell of the martensite phase forms uniformly on
the whole surface, it fixes the shape of the part,
and thus the distortion is lower.
Thus, the method of numerical modeling of
the hardening process can be used to predict
hardening distortions and to designate machining allowances for bearing rings, as well as to
define the laws governing deformation as a func-
tion of the methods adopted for the heating and

cooling of rings of different steels with different
sectional configurations, diameter, and so forth.
The above-stated problem is especially important for the bearing industry because in the
case of big distortions of bearing rings, workers
have to perform a lot of turning and polishing.
Upon removal of the thin surface layer from the
surface of the quenched bearing rings, residual
stresses in the bearing rings are redistributed,
which can have negative results; that is, it can
reduce the service life of finished bearing rings.
Moreover, the surface layer of bearing rings
is the hardest. The removal of this surface layer
also reduces the quality of a bearing, to say nothing of the extremely high costs for the turning
treatment of bearing rings. Therefore, the idea
regarding the reduction of distortions of bearing
rings by the intensification of quench cooling is
very appealing.
It should be also added that instead of mineral
oils, at the first stage of cooling one can use
Table 2 Residual stresses and bearing ring conicity depending on the cooling conditions
within the martensite range at the second stage of cooling
Stresses, MPa
Method of cooling
In air
In oil
In circulating water
Under intensive shower
r22
r33
rI
Conicity
90130
110130
75200
70130
130165
150180
180230
100180
70
180
240
170
0.044
0.045
0.051
0.005
35
30
Height of ring, Z, mm
25
20
15
10
0.1
0.2
0.3
0.4
0.5
Displacement, U, mm
Displacements of generatrix of inner surface of
bearing ring during hardening (lines denote
computed data, and points experimental data). 1, after turning on lathe; 2, after heating to quenching temperature; 3
and 4, after 30 and 39 s, respectively, of immersion in oil;
5, after quenching
Fig. 19
(a)
Fig. 20
(b)
Phases distribution at slow and intensive cooling. 1, austenite; 2 and 5, martensite
190
1
700
Carburized layer
225
203.8
88.2
Tsurface
3
500
2
300
100
342.4
(a)
4
, min
(b)
Fig. 21
(a) Sizes of the carburized bushing. (b) Surface temperature in the carburized bushing during oil quenching
at 137 C
aqueous polymer solutions exhibiting properties

of inverse solubility. At the second cooling
stage, rings are simultaneously washed and intensively cooled in accordance with results presented in Table 2. Thus, two-stage quenching of
bearing rings with the use of intensive cooling
at the second stage results in the reduction of
distortion. It is very important to solve the problem of the optimization of steel part heat treatment processes for two-stage quenching.
Study of Thermal Stresses Formed

in Carburized Steel Products due to
High-Rate Quenching
Quenching of a carburized bushing with twostep cooling was studied as a thermoplasticity
problem using FEM. Instantaneous and residual
stresses were analyzed in the component at various cooling rates. It was shown that an increase
0
100
0 MPa
0 MPa
200
100
0
200
100
0
100
(a)
(b)
Distribution of axial (a) and circumferential (b)

stresses in the carburized bushing at the end of
the first cooling stage. Deformation is enlarged by 25.
Fig. 22
of cooling rate through the phase-transformation

range improves the distribution and lowers the
level of residual stresses in the as-quenched
components (Ref 16).
The carburized bushing (Fig. 21) made of
steel 14KhGSN2MA was heated in an electric
furnace to 850 C and cooled initially for 20 s in
air and then in hot MS-20 oil at an optimal temperature of 100 to 140 C. On reaching a mean
temperature of 220 C in the carburized part, the
first-stage cooling process was interrupted and
the part was transferred to another cooling medium for cold treatment. Use of the optimal MS20 oil temperature led to better bushing roundness (Ref 16).
The second stage of multistage quenching
(when the carburized layer undergoes martensite
transformations) can be carried out at a slow
cooling rate (in air) or using higher-rate cooling,
for example, in a circulating fluid. It was noted
in Ref 16 that high-rate cooling through the martensitic range raises mechanical properties and
reduces the risk of quenching cracks. For this
reason, it was of practical and theoretical interest
to reveal the effect of second-stage high-rate
cooling on the strain and stress patterns in the
hardened part.
The simulation study of the quenching process
was based on a set of equations describing the
distribution of temperatures, stresses, and
strains, including a nonlinear unsteady-state
heat-conduction equation and plastic flow equations, as well as initial and boundary conditions.
It is assumed in the CCT diagram of supercooled austenite that all thermal and mechanical
characteristics of material are functions of temperature and time. The iterative approach to solving such heat-conduction and high-temperature
plasticity problems on the basis of the FEM was
reported elsewhere (Ref 9, 15).
Solution of heat-conduction and thermoplasticity problems gives a good picture of the variations of temperature, stress, and strain distributions in the product during quenching. The
variations of the temperature field in combination with a known CCT diagram of supercooled
austenite can be used to follow the kinetics of
phase transformations and to describe the
quenched layer depth (Ref 15).
To test the computer package, the authors

solved a problem for which theoretical results
could be compared to experimental data (Ref
17). In addition, FEM was used to solve the nonlinear unsteady-state heat-conduction problem
and thermoplasticity problem, both of which
have an analytical solution. The comparison
demonstrated a good agreement of analytical and
numeric data (Ref 6).
To analyze the first cooling stage, the authors
selected first-kind boundary conditions that were
determined experimentally (see Fig. 21), and for
the second cooling stage they prescribed thirdkind boundary conditions.
During cooling in MS-20 oil, temperature gradient over the bushing section was insignificant.
This was due to high thermal conductivity of
steel. The upper and lower bushing flanges cool
down slightly faster than the wall, and, in addition, the coefficient of thermal contraction for
carburized steel (a) is higher than for noncarburized. These two factors result in appreciable distortions. During the first cooling stage, the bushing is barrel-shaped. The distribution of axial
and circumferential stresses at the end of stage
one is illustrated in Fig. 22.
In the second stage, the carburized layer begins to undergo martensitic transformations.
During further cooling, the carburized layer expands because of the greater specific volume of
martensite compared to austenite, and the bushing crown sags in.
The difference of the maximum and the minimum radii of the bushing inner diameter of cylinder generatrix, Dr, as a function of secondstage heat-transfer coefficient is shown in Table
3. Increase in the second-stage cooling rate reduces residual strain, which is in agreement with
earlier research (Ref 15).
The minimum value of Dr versus maximum
cooling rate at the second stage can be interpreted as follows. During high-rate cooling, the
martensitic transformations are completed almost simultaneously over the entire carburized
layer surface. The compliance of the surface material in noncarburized steel also drops sharply
due to reduced temperature. Thus the surface
layers of the part form a rigid skin that resists
any further deformation of the bushing which
could result from cooling of the core section.
The difference Dr can be diminished by increasing heat transfer at the first cooling stage.
An increase of the first-stage cooling rate will
raise the temperature gradient and enhance bushing deformation. Since deformation at stage two
is opposite to stage one, the increase in the first-
Table 3 Difference of the radii with respect

to heat-transfer coefficient at the second
stage
Heat transfer coefficient
, W/(m2 K)
30
1000
5000
Difference of radius
Dr, m
0.165
0.162
0.155
Generalization of Computational
and Experimental Results for
Heating and Cooling of Bodies of
Various Configurations
For bodies of a simple shape (plate, ball, cylinder, parallelepiped), solving the heating problem for the medium with constant temperature
(boundary condition of third kind) one can state:
Stage I
=0
ln(Tc T1)
=R
180
ln(Tc T2)
1
Fig. 24
180
2
350 MPa
50
100
550
850
950
50
1000
50
400
1050
350
350
350
100
900
1050
1150
1200
1250
1300
1350
1400
1450
TC T
TC TO
300
250
200
150
100
300
1050
400
1400
50
1000
950
850
550
150
200
1350
250
1300
1250
150
1200
1150
100
1050
50
900
100
0
(a)
Fig. 23
(Eq 3)
l1,i l2,i l3,i ln,i
FoV (s/R2V) is the Fourier number, where the

determining size of the part is taken as a generalized size.
According to the above inequality, each following term of Eq 3, when FoV grows, will be
negligibly small in comparison with the previous
term, and the sum of all roots will have very little
difference from the value of the first term. Therefore, beginning from a definite value of Fourier
number Fo1, the use of the first term of Eq 3,
can be restricted, that is:
TC T
TC TO
An,iUln,i Rii
i1
x
2
RV
FoV
R2i
when FoV Fo1
2
exp ln,i
(Eq 4)
Beginning from this value of Fo1, the time function of (TC T) is described by a simple exponent.
Applying the logarithm to Eq 4, one obtains:
200
1100
2
RV
FoV
R2i
where TC is the temperature of medium, TO is

the initial temperature of the body, and T is the
temperature of the body at a point (x, y, z) and
at time s. An,i (An,1, An,2, An,3), which are initial
thermal amplitudes depending on the initial temperature distribution and shape of the part.
U[ln,i (xi /Ri)] is a function regarding the change
in temperature on coordinates (x x1, y x2,
z x3); R1, R2, and R3 are the sizes of the part;
RV is a generalized size of the part equal to the
part volume (V) divided by the surface area (S);
that is, RV V/S (for infinite plate RV R, for
infinite cylinder RV 12R, for ball RV 13R);
ln,i are roots of characteristic equations, ordered
as:
1n
1450
100
350
50
1150
1150
An,iUln,i Rii
n1 i1
2
exp ln,i
Function ln(TC T) versus time in the cooling

process
0 MPa
1100
Stage II
ln[Tc T(,)]
stage deformation can reduce residual strain. As

can be seen in Fig. 22, the first cooling stage
results in low bushing stresses. This suggests
that high-rate cooling at this stage does not lead
to material failure.
Note that the distribution of residual stresses
in the part is the same for second-stage heattransfer coefficient of 5000 and 300 W/(m2 K)
as for 1000 W/(m2 K). Figure 23 shows
that the highest compressive stresses occur in the
carburized layer. A comparison of results obtained at various second-stage heat-transfer coefficients demonstrates that high-rate heat transfer at this cooling stage does not increase
residual tensile stresses and even slightly reduces them. A similar favorable distribution of
residual stresses during high-rate cooling was
observed in multistage quenching of ShKh15
steel bearing rings (Ref 17).
Thus it is possible to improve the efficiency
of carburized bushing cooling process (compared to oil quenching) by raising the cooling
rate without risk of a quenching crack. Similar
computations can be done for heat treatment of
rolled steel and tubing. More information about
intensive quenching is given in Ref 14, 18,
and 19.
(b)
Distribution of axial (a) and circumferential (b) residual stresses in the carburized bushing at the end of the
second cooling stage at 1000 W/m2 K. Deformation is enlarged by 25.
TC T
TC TO
i1
1n A1,iU ln,i
xi
RV
ln,i
Ri
Ri
2
V
(Eq 5)
Thus, the time function of 1n(TC T) on the

graph is a straight line (see Fig. 24). In the case
of lengthy cooling, the temperature at all points
will be the same and equal to TC (stationary
state).
Therefore, the entire cooling process can be
divided into three stages. The first stage of irregular conditions is characterized by the fact that
the biggest role is played by the initial temperature distribution. Any irregularity in the initial
distribution is reflected in the temperature distribution at next moments of time. The dependence
between (TC T) and s is described by Eq 4.
The second stage is called regular conditions.

The dependence between (TC T) and s is described by simple exponent (Fig. 24). The temperature distribution inside the body is described
by function U and does not depend on the initial
temperature distribution, because the value of
A1,i is a factor; that is, it determines the scale and
not the essence of the phenomenon. The third
stage corresponds to the stationary state (FoV
), when the temperature in all points of the

body equals the temperature of environment.
Figure 24 shows graphs of 1n(TC T) as a
function of s for the surface and core of the body.
As is seen from Fig. 24, at the stage of regular
conditions these graphs are in the form of
straight lines. If the temperature at the initial
time is equal in all the points and equal to TO,
then the curves must originate from one point.
Since the surface layers are cooled more
quickly than central ones, the curve 1n(TC T)
f(s) at the first stage for the central layer has
the convexity facing the ordinate and for the surface layers facing the abscissa (Fig. 24). The
above-mentioned analysis is correct for bodies
of any shape.
It has been shown (Ref 20) that any heating
(cooling) problem for bodies of any shape can
be reduced to heating (cooling) problem for a
body of simple shape (plate, cylinder, ball) using
the criterion of approximate similarity.
The tangent of the angle of the line inclination
(for the stage of regular conditions) will be:
tg(180 w) tgw
1n(TC T1) 1n(TC T2)
m const
s2 s1
The constant m is the rate of the time function

of the logarithm of the superfluous temperature,
that is:
[1n(TC T)]
m
s
(Eq 6)
It is the same for every point of the body and

also for volume average temperature T and is
l1,t
i1
RV
Ri
a
R2V
(Eq 7)
Therefore, the value of m is determined by thermal and physical factors, size, and shape of the
body. On the basis of Eq 6 and boundary condition of the third kind, for body of any shape
one can state the following equation, which is
true at the stage of regular conditions of cooling
(heating):
ccV
dT
S(TC Tp) ccV(TC T)m

ds
(Eq 8)
S TC TP
a
W 2 Kn
ccV TC T
kRV
RV
(Eq 9)
0.8
2
Kn BiVW
RVW
k
mR2V
l1,i
i1
RV
Ri
0.6
0.2
0
Fig. 25
8
12
Biot number, BiV
16
Universal approximate function Kn f(BiV). 1,

for plate; 2, for ball; 3, for cylinder
2.095 3.867Bi
kBiV
is a Kondratiev number, W (TC TP)/(TC

T) is a parametrical criterion characterizing
the unevenness of the temperature field, since it
is equal to the superfluous temperature of the
part surface divided by the volume-averaged superfluous temperature. If the temperature distribution in the body is even (BiV 0), then W
1. The greater the temperature unevenness, the
lesser W. When W 0, the unevenness of the
temperature distribution is the greatest (BiV
, and TP TC). Thus, the Kondratiev number

characterizes not only the unevenness of the temperature field, but also the intensity of interaction
of the part surface with the environment.
The theory of regular conditions is based
mainly on Eq 6; that is, the ratio of the local
cooling (heating) rate and superfluous temperature is a constant value:
Kn WBiV
0.4
kBiV
Therefore, the Kn criterion is determined by the

shape of the part and characteristic numbers,
l1,1, l1,2, l1,3, and therefore also by Biot criterion, since the characteristic numbers are functions of Biot criteria.
It has been proved that curves Kn f(BiV)
for geometrically absolutely different bodiesa
ball, parallelepiped, cylinder, and so onare so
close to each other that all of them can be replaced by one averaged curve (see Fig. 25). The
aforementioned formulas are true for the steel
1nh
or
where:
With regard to Eq 9, the important function of

the theory of regular conditions is determined:
2.095 3.867Bi
heating. They can be used for calculation of the

cooling process. In this case, one should use dimensionless formula: H (T1 TC)/(TO
TC) instead of H (TC T1)/(TC TO). The
results of theoretical and experimental investigations for determination of the heating and
cooling time for bodies of various configuration
have been generalized by author of Ref 10 and
presented in the form of Eq 10 convenient for
calculations.
Theory of the regular heat conditions describes the process of cooling for bodies of different shapes. The cooling time of such bodies
can be evaluated on the basis of the dimensionless dependence:
FoVKn
From this it follows that:
dT
1
m const
ds TC T
1.0
Kondratiev number, Kn
called heating factor or cooling factor. Equation

5 is followed by equality:
1n
TO TC K
T TC aKn
(Eq 10)
where t is cooling time, FoV is generalized Fourier number, BiV is generalized Biot number, k
1, 2, 3corresponding to bodies of plate,
cylindrical, and spherical shape, K is Kondratiev
form factor, Kn is Kondratiev number, a is thermal diffusivity coefficient, TO is initial temperature of heated part, TC is temperature of the medium, and T is current temperature.
According to patented technology, the process
of intensive cooling should be performed in such
a way that the martensite transformations begin
uniformly at the entire surface simultaneously,
forming high compressive stresses which, as was
noted, achieve the maximum at a definite moment of time. This moment of time can be determined by Eq 10. For bodies with different
configurations, temperatures are found at points
the furthest from the surface that correspond to
the maximum compressive stresses at the surface. Using these data and Eq 10, one can easily
calculate time of achieving maximum compressive stresses without making expensive and
complicated calculations. The proposed approach is also correct for bodies of complex configurations.
The results of the above-stated generalization
have served as a base for the creation of new
technology of steel part heat treatment. The essence of the new technology is that alloy steel
parts are quenched in conditions of intensive
heat transfer until the moment when the maximum compressive stresses on the surface of the
part to be cooled are reached, and then intensive
cooling is interrupted and the part is kept at constant temperature of martensite start until full
transformation of the overcooled austenite in
central layers of the part (Ref 21). The moment
when maximum compressive stresses on the surface to be quenched are reached is determined
by generalized Eq 10. There is a database for the
determination of Kondratiev form coefficient K
and Kondratiev number Kn, which are present in
Eq 10. The values of Kondratiev number Kn for
simple bodies can be calculated by formulas presented in Table 4. The Kondratiev form coefficient characterizes the shape and size of parts to
be quenched, and Kn characterizes the cooling
capacity of the quenchant. There is an average
thermal conductivity a in the formula, which
characterizes thermal and physical properties of
the material. Thermal conductivity a of the material is a linear function of temperature. For this
reason, one can use the average value of the thermal conductivity and it will not add big errors
to calculations.
Knowing the temperature in the core of the
part when the compressive stresses are optimal
on the surface, one can calculate the time of intensive cooling by the generalized Eq 10. The
generalized data given previously can be used
for the computerization of industrial processes.
Link between Usual and Generalized Biot
Criteria. In analytical solutions, a usual Biot
number generally is used, which is, as known,
dimensionless (Ref 22):
Bi
R
k
where is the heat-transfer coefficient, c is the

heat conductivity of the material, and R is the
radius of a cylinder, ball, or half thickness of a
plate.
It is noted in Ref 22 that the characteristic size
can be also determined as the volume of the body
divided by its surface area, that is:
RV
V
S
where V is the volume of the body (m3) and S is

the surface area (m2).
In the theory of regular heat conditions (Ref

23) the characteristic size is determined as value
L that includes three parameters S, V, and K, that
is:
L
Kn WBiV
This universal relation remains true for bodies of

arbitrary shape and when the range of the generalized Biot number BiV is from 0 to
, the
Kondratiev number Kn changes from 0 to 1, and
the criterion of temperature field nonsmoothness
W changes from 1 down to 0 (see Table 5). The
physical sense of the criterion of temperature
field nonsmoothness W follows from the formula:
Table 4 Functions for the analytical calculation of Kondratiev form factor

Kondratiev form
factor, K
Body shape
4R2
p2
Infinite plate, thickness of 2R
R2
5.783
Infinite cylinder
R2
p2
Ball
Finite plate, dimensions of sizes:
L1, L2, L3
Cube
Finite cylinder, height: Z
1
1
1
1
p 2 2 2
L1
L2
L3
L2
3p2
1
5.783
p2
2
R2
Z
(Eq 11)
where Tsf is the average temperature on the surface of the body, Tm is the temperature of quenchant, and TV is the volume-average temperature.
On the other side, the criterion of temperature
field nonsmoothness W for bodies of arbitrary
shape can be expressed through the generalized
Biot number (BiV) as follows (Ref 22, 23):
1
(Bi 2V 1.437BiV 1)0.5
(Eq 12)
The analysis of Eq 11 and 12 shows that when

BiV 0, then W 1 and Tsf TV that is, the
temperature field on the section of a body to be
cooled is even. If BiV
, then W 0 and Tsf
Tm, that is, the temperature of the body to be
cooled becomes equal to the temperature of environment at the time of its immersing into the
quenchant.
Since the above-stated dimensionless number
is very important in practice, let us find the link
between the generalized Biot number [Bi (/
k)R] and generalized Biot number [BiV (/
k)K(S/V)].
For an infinite plate it is as follows:
BiV
S
4R2 1
4
K
2 Bi
k V
k p2 R
p
since K (4R2 /p2), where in this case R is the

half thickness of the plate.
R2
S
2
and
5.783
V
R
For a ball:
BiV
This approach allowed establishment of the

universal link between the Kondratiev number
Kn, criterion of temperature field nonsmoothness
W, and generalized Biot number BiV:
R2
2
2

Bi
k 5.783 R
5.783

S
L K
k k V
T sf Tm
W
TV Tm
BiV
since
S
K
V
where K is the Kondratiev form coefficient and

in this case the generalized Biot criterion is:
BiV
For an infinite cylinder:
R2
4pR2
3
2 Bi
k p2 4/3pR2
p
Since for the ball:

K
R2
V
3
and
p2
S
R
Thus, there are the following links for bodies that

are plate-shaped, cylinder-shaped, and ballshaped correspondingly:
BiV 0.405Bi BiV 0.346Bi BiV 0.304Bi
Discussion of Results
The studies made have shown that the values
of the current and residual stresses can be controlled by many methods. Making intensive and
uniform quenching, one can reach high compression stresses on the surface (see Fig. 9). It is
very important that selecting heat-transfer coefficient, one can reach even null residual stresses
on the surface, if necessary. Thus, increasing the
heat-transfer coefficient beginning from BiV
4.5, one can obtain high residual compression
stresses on the surface. These residual compression stresses are compensated by tensile stresses
in the core. It should be noted that residual compression stresses increase the service life of the
steel parts quenched.
It should be noted that intensive cooling has
the effect of not only the formation of high compression stresses on the surface of parts
quenched, but also the simultaneous reduction of
distortion of quenched parts. The reduction of
distortions and prevention of the formation of the
quench cracks are due to the formation of hard
martensite layer on all the surface, which fixes
the initial sizes of parts and forms high compressive residual stresses on the surface of
quenched parts. In addition, intensive heat transfer in the area of martensite transformations results in the improvement of mechanical properties of the material and makes the service life of
the parts longer. The generalization of the abovementioned is given in Ref 24.
It is especially important to discuss the process of two-stage quenching. This article considers the two-stage cooling process. During the
first stage, one arranges slow cooling in oils or
aqueous polymer solutions until the time of minimum distortions is reached, and then, at the second stage, one arranges intensive cooling and
washing of the parts. Intensive cooling at the second stage fixes the attained minimum distortion

Table 5
Links between generalized Biot number (BiV), criterion of temperature field nonsmoothness (W), and Kondratiev number (Kn)
BiV
Kn
BiV
Kn
0.00
0.01
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
0.24
0.26
0.28
0.30
0.32
0.34
0.36
0.38
0.40
0.42
0.44
0.46
0.48
0.50
0.52
0.54
0.56
0.58
0.60
0.62
0.64
0.66
0.68
0.70
0.72
0.74
0.76
0.78
0.80
0.82
0.84
0.86
0.88
0.90
0.92
0.94
0.96
0.98
1.00
1.05
1.10
1.15
1.20
1.25
1.30
1.35
1.40
1.45
1.50
1.55
1.60
1.65
1.70
1.75
1.80
1.85
1.90
1.95
2.00
2.05
2.10
2.15
2.20
2.25
2.30
2.35
1.00000
0.99284
0.98574
0.97171
0.95791
0.94434
0.93101
0.91792
0.90507
0.89246
0.88009
0.86796
0.85607
0.84441
0.83298
0.82178
0.81081
0.80007
0.78954
0.77923
0.76913
0.75923
0.74954
0.74005
0.73076
0.72166
0.71274
0.70401
0.69545
0.68708
0.67887
0.67082
0.66294
0.65522
0.64766
0.16109
0.15981
0.15855
0.15732
0.15609
0.15489
0.15371
0.15254
0.15140
0.15026
0.14915
0.14805
0.14590
0.14382
0.14179
0.54484
0.53940
0.52622
0.51362
0.50157
0.49003
0.47898
0.46839
0.45823
0.44848
0.43911
0.43011
0.42146
0.41312
0.40510
0.39737
0.38992
0.38273
0.37580
0.36910
0.36263
0.35637
0.35032
0.34447
0.33880
0.33331
0.32799
0.32283
0.31783
0.00000
0.00993
0.01971
0.03887
0.05747
0.07555
0.09310
0.11015
0.12671
0.14279
0.15842
0.17359
0.18833
0.20266
0.21657
0.23010
0.24324
0.25602
0.26844
0.28052
0.29227
0.30369
0.31481
0.32562
0.33615
0.34640
0.35637
0.36608
0.37555
0.38476
0.39374
0.40249
0.41103
0.41934
0.42745
0.87796
0.87898
0.87998
0.88096
0.88193
0.88289
0.88383
0.88475
0.88566
0.88656
0.88744
0.88831
0.89001
0.89165
0.89325
0.53395
0.53940
0.55253
0.56498
0.57680
0.58804
0.59873
0.60891
0.61861
0.62787
0.63672
0.64517
0.65326
0.66100
0.66842
0.67553
0.68236
0.68892
0.69523
0.70129
0.70712
0.71274
0.71816
0.72338
0.72841
0.73328
0.73797
0.74251
0.74690
2.40
2.45
2.50
2.55
2.60
2.65
2.70
2.75
2.80
2.85
2.90
2.95
3.00
3.05
3.10
3.15
3.20
3.25
3.30
3.35
3.40
3.45
3.50
3.55
3.60
3.65
3.70
3.75
3.80
3.85
3.90
3.95
4.00
4.05
4.10
4.15
4.20
4.25
4.30
4.35
4.40
4.45
4.50
4.55
4.60
4.65
4.70
4.75
4.80
4.85
4.90
4.95
5.00
5.05
5.10
5.15
5.20
5.25
5.30
5.35
5.40
5.45
5.50
5.55
5.60
5.65
5.70
5.75
5.80
5.85
5.90
5.95
6.00
6.10
6.20
6.30
6.40
6.50
6.60
0.31298
0.30827
0.30369
0.29925
0.29493
0.29074
0.28665
0.28268
0.27882
0.27505
0.27139
0.26782
0.26434
0.26095
0.25764
0.25442
0.25127
0.24820
0.24520
0.24228
0.23942
0.23662
0.23389
0.23122
0.22862
0.22606
0.22357
0.22113
0.21874
0.21640
0.21411
0.21186
0.20967
0.20751
0.20540
0.20334
0.20131
0.19932
0.19738
0.19547
0.19359
0.19175
0.18995
0.18817
0.18644
0.18473
0.18305
0.18140
0.17979
0.17820
0.17664
0.17510
0.17359
0.17211
0.17065
0.16922
0.16781
0.16642
0.16506
0.16371
0.16239
0.16109
0.15981
0.15855
0.15732
0.15609
0.15489
0.15371
0.15254
0.15140
0.15026
0.14915
0.14805
0.14590
0.14382
0.14179
0.13981
0.13789
0.13603
0.75115
0.75525
0.75923
0.76309
0.76682
0.77045
0.77396
0.77737
0.78069
0.78390
0.78703
0.79007
0.79302
0.79590
0.79870
0.80142
0.80407
0.80665
0.80917
0.81162
0.81402
0.81635
0.81863
0.82085
0.82302
0.82513
0.82720
0.82922
0.83120
0.83313
0.83501
0.83686
0.83866
0.84043
0.84216
0.84385
0.84551
0.84713
0.84872
0.85027
0.85180
0.85329
0.85476
0.85619
0.85760
0.85898
0.86034
0.86167
0.86297
0.86426
0.86551
0.86675
0.86796
0.86915
0.87032
0.87147
0.87260
0.87371
0.87480
0.87587
0.87692
0.87796
0.87898
0.87998
0.88096
0.88193
0.88289
0.88383
0.88475
0.88566
0.88656
0.88744
0.88831
0.89001
0.89165
0.89325
0.89480
0.89631
0.89778
BiV
6.70
6.80
6.90
7.00
7.20
7.40
7.60
7.80
8.00
8.50
9.00
9.50
10.0
10.5
11.0
11.5
12.0
12.5
13.0
13.5
14.0
14.5
15.0
15.5
16.0
16.5
17.0
17.5
18.0
18.5
19.0
19.5
20.0
21.0
22.0
23.0
24.0
25.0
26.0
27.0
28.0
29.0
30.0
35.0
40.0
45.0
50.0
60.0
70.0
80.0
100.0
Kn
0.13421
0.13244
0.13072
0.12904
0.12580
0.12273
0.11980
0.11700
0.11434
0.10817
0.10263
0.09764
0.09310
0.08897
0.08519
0.08171
0.07851
0.07555
0.07280
0.07025
0.06787
0.06564
0.06356
0.06160
0.05976
0.05803
0.05639
0.05485
0.05339
0.05200
0.05068
0.04943
0.04824
0.04602
0.04400
0.04214
0.04044
0.03887
0.03741
0.03607
0.03481
0.03364
0.03255
0.02799
0.02456
0.02187
0.01971
0.01647
0.01414
0.01239
0.00993
0.00000
0.89920
0.90059
0.90194
0.90325
0.90578
0.90817
0.91045
0.91262
0.91468
0.91945
0.92371
0.92754
0.93101
0.93416
0.93704
0.93967
0.94210
0.94434
0.94641
0.94833
0.95012
0.95179
0.95336
0.95482
0.95620
0.95749
0.95871
0.95986
0.96095
0.96198
0.96296
0.96389
0.96478
0.96642
0.96792
0.96929
0.97055
0.97171
0.97278
0.97377
0.97470
0.97556
0.97636
0.97970
0.98221
0.98417
0.98574
0.98810
0.98979
0.99106
0.99284
1.00000
of parts to be quenched, since the formation of

martensite shell is observed on the entire surface,
which fixes the shape of the part to be quenched.
In addition to the method considered for the
first stage, one can also arrange very intensive
cooling until nucleate boiling finishes or maximum compression stresses are reached on the
surface to be quenched. The latest technology is
being patented in the Ukraine and is the subject
of a separate discussion.
It should also be noted that the technology developed is furnished with good mathematical
support and corresponding software, allowing
simulation of complicated technological processes and selection of optimal ways of quench
cooling.
The software package developed during 1975
through 1982 won good reputation with regard
to the agreement between results of computations and results of experiments. The good
agreement between results of computations and
results of experiments is explained by physically
grounded selection of all parameters included in
the mathematical models and describing the
quenching process.
Thus, intensive steel-quenching methods have
great prospects; therefore the approach and calculation methods developed can be used for the
implementation of intensive steel-quenching
methods in practice. In practice it is important to
select proper boundary conditions. In practice
cooling properties of quenchants are determined
by tests of specimens, and on this basis heattransfer coefficients are determined. Unfortunately, one cannot always use these results when
the shapes and sizes of steel parts are changed,
that is, at the transition from specimen to real
parts (Ref 14).
The authors have suggested another approach
to the selection of boundary conditions based on
the conception of self-regulated thermal process.
The new approach lies in that it is possible to
forecast the character of changes in the temperature of parts that are to be quenched, in the process of nucleate boiling, and also to determine
the time of this process (Ref 1, 14, 25).
Previous sections of this article discuss ways
to reduce distortion and eliminate cracking during intensive heating and cooling of parts to be
quenched. This is highly important for technological processes, since the intensive heating followed by intensive cooling results in the improvement of mechanical properties of steels and
improvement of the material plasticity.
The main regularities of intensive steel heating are presented in extensive work (Ref 26), and
they can be briefly described:
When the heating rate increases, the critical
steel points Ac1 and Ac3 grow continuously

proportionally to V 1/3
H , and the growth rate for
steel is significantly higher than for the iron.
The higher the heating rate the lower the size
of the austenite grain. By further quenching
the austenite of such structure, one can reach
fine-grain martensite or high-dispersion products of intermediary dissociation.
In the intensive heating the austenite grain is
less susceptible to the growth when the temperature increases and during tempering at
such temperature.
The problem of obtaining the fine grain at
heating has been usually solved by the selection
of steel alloying elements. As a rule, natural finegrain steels are those deoxidized by aluminum.
Adding aluminum, which forms aluminum nitride (AlN) and does not bind the carbon, does
not reduce the steel-strengthening capacity at
quenching. Heat treatment of steel deoxidized by
aluminum affects the dispersion values, distribution, and the amount of AlN, and through this,
it affects the susceptibility of the austenite grain
to growth. The temperature above which heating
results in the growth of austenite grain is called
the rough-condition temperature (Ref 26). The
rough-condition temperature for steel with AlN

depends on temperature of its dissolution. Heating to 1000 C (1830 F) considerably reduces
the amount of AlN, and heating to 1150 C (2100
F) results in its full dissolution. Maintaining the
temperature below that of full dissolution for a
long time causes the coagulation of isolations
resulting in the growth of austenite grain. There
exists a critical amount of AlN isolation; when
this amount is exceeded, the effect of stopping
is reduced.
The presence of AlN particles affects mainly
the rough-condition temperature for austenite
and in no other way is a factor contributing to
getting a fine grain of the austenite at phase
transformation c. The optimal amount of
aluminum added to steel, in the case of deoxidizing not reducing the rough-condition temperature below 975 C (1790 F), should not exceed 0.03 to 0.04% (Ref 27, 28).
Making steel alloys with carbide-forming elements allows an increase in the rough-condition
temperature when carbides formed have sufficiently high dissolution temperature in the austenite. Compounds sufficiently resistant to dissolution are vanadium carbides (VC, in
particular, V4C2) and vanadium nitride (VN). Its
effect on the susceptibility of steel to austenite
grain growth is considered in Ref 26 and 29 to
31. The temperature of full dissolution for VC
in 30X2 steel is 1000 C (1832 F); for vanadium
nitride it is approximately 1100 C (2010 F).
The addition of vanadium allows slowing of austenite grain growth of the steel in the case of
heating to 980 C (1800 F) (Ref 31). Besides,
its critical amount is about 0.06%. The effect of
zirconium on the susceptibility to austenite grain
growth has been studied in Ref 32 to 36.
The effect of slowing down due to adding titanium and zirconium to steel is connected with
producing carbides, nitrides, and carbonitrides.
The critical amounts of titanium and zirconium,
which have great effect on the growth of austenite grains, can be varied from 0.03 to 0.08%,
depending on the dispersion value of producing
phases containing titanium in steels with various
chemical composition. Adding titanium and zirconium to steel allows increase of the rough-condition temperature to 1150 to 1200 C (2100
2200 F) (Ref 26).
In practice, great interest is paid to steels alloyed by niobium (Ref 33, 3639), which is connected with not only great resistance to overheating in comparison with steel alloyed with
titanium, zirconium, or aluminum (Ref 30), but
also with their good strengthening as a result of
dispersion hardening and their high resistance to
tempering (Ref 31).
Since the carbides and nitrides of niobium dissolve worse in austenite than vanadium carbides
and nitrides, steels with niobium have higher
rough-condition temperatures. Their high resistance to overheating is due to the fact that niobium, in contrast to vanadium, slows down the
process of recrystallization and increases its energy of activation by two times (Ref 40). The
critical amount of niobium in steel is approximately 0.02%.

Thus, one of the advantages of microalloying
is the possibility of preserving fine-grain austenite in cases when steel, for some reason (great
weight of parts, danger of distortion, and so on),
undergoes slow heating, and the strict requirements for the preservation of the shape or geometry of the part do not allow use of other
methods of influence on the austenite structure.
The other advantage of this method is that it allows regulation of the initial sizes of the austenite grain with the use of the selection of the complex of alloy elements of the compound, which
can serve as barriers for the growth of austenite
grain in steel heating at a wide range of temperatures.
The determinative role of austenite grain size
on the formation of steel properties in intensive
thermal hardening is well known. One of the
most systematic studies has been described in
Ref 41 and 42. This study considers the role of
the kinetic factor in the formation of the austenite grain. The experimental data received by this
approach are very convenient for the selection of
optimal temperature-time parameters of steel
austenite formation in the case of intensive heat
treatment, since they describe the dependence of
the size of austenite grain on the temperature and
heating rate.
Figure 26 shows medium diameter of the austenite grain in steels 40 (Ref 42) and
40Kh2NGSM versus temperature and heating
rate. Each of these steels exhibits different susceptibility of the austenite grain to the growth
while the heating rate increases. In the case of
the same percentage of carbon, they are also
characterized by different susceptibility to austenite grain growth, which is connected with not
only the degree of alloying, but also initial structure states before heating. The initial sizes of
austenite grains for both carbon steel 40 (Ref 42)
and
complex-alloy
construction
steel
40Kh2NGSM are the same in the case of ferritepearlite structure when heating to c phase
transformation temperatures. However, as long
as the heating temperature increases, the difference in the susceptibility to austenite grain
growth is observed in these steels, and it is
greater when the heating rate is lower, which is
connected with not only kinetic characteristics
of the austenite formation process for steels with
different alloy composition in the case of intensive heating, but also with barrier effect of barely
soluble carbides with the movement of the leading edge of the austenite grain in the alloy steel.
Figure 27 shows various techniques of steel
quenching, at which it is possible to prevent
quench-crack formation (Ref 43); intensive heating has the effect of finer austenite grain, which
improves the mechanical properties of the material. Intensive cooling in the martensite range
is connected with additional strengthening of the
material, which can be explained as follows.
Martensite plates form in supercooled austenite,
which is in an extremely high compressive stress
state. The martensite has a larger specific volume
than the austenite. As a result, the austenite undergoes extensive plastic deformation due to

0.03
0.06
0.08
2
1
2
0.04
3
0.01
Size (d), mm
0.02
Size (d), mm
Size (d), mm
0.06
0.02
0.04
2
3
0.02
4
0
800
900
0
750
1000
Temperature (T ), C
(a)
850
950
1050
0
750
1250
1150
Temperature (T ), C
900
1000
1100
1200
Temperature (T ), C
(b)
(c)
Size of austenite grain versus temperature and heating rate. (a) Steel 40, the initial state is condition after burning (Ref 42). 1, 2, 3, and 4 are heating rates 0.03, 8, 200, and
1000 /s correspondingly. (b) Steel 40Kh2NgSM, the initial state is condition after burning at 650 C (1200 F) for 3 h and distortion of 50%. 1, without distortion, heating
rate of 2 /s; 2, 3, and 4 are 2, 25, 200 /s correspondingly. (c) Steel 35Kh2NgSM, the initial state is condition after intensive tempering and distortion of 50%, 1, 2, 3, and 4 are at
heating rates of 0.01, 2, 5, and 200 /s correspondingly.
Fig. 26
Slow (or fast or

intensive) + slow
Slow + intensive
Fast + fast
Intensive + intensive
A1
Temperature
Austenite
temperature
range
Ms
Martensite
temperature
range
Mf
Quench cracking
prevented
Susceptible
to quench cracking
Fig. 27
Scheme of several steel quench methods and their effect on the steels susceptibility to quench cracking
95
55
0.3
R60
15
8.5
45
1:15
Punch made of the molybdenum high-speed
steel R6M5 AISI M20. Punch length, 126 mm;
maximum diameter, 15.3 mm
Fig. 28
Fig. 29
0.3
Die made out of high-speed steel quenched intensively
microhammering by the martensite (this process can be compared to low-temperature thermomechanical processing, which also strengthens the material being processed) (Ref 44, 45).
A high dislocation density is produced, and,
because of the very rapid cooling, the dislocations are uniformly frozen in the quenched surface layer. This is what is believed to produce
the additional strengthening or superstrengthening. Very rapid quenching (intensive quenching)
in the martensite range and the formation of very
high surface compressive stresses are prerequisites. The superstrengthening effect produced by
intensive quenching has been used to enhance
the performance of punches and dies made of
high-speed steel (Fig. 28, 29) and of small dies
made of bearing steel.
The R6M5 punches are used at high loads in
automatic forming machines. Conventionally
quenched tools have relatively short lives, and
productivity suffers when machines are shut
down for tool changes. Punch life increased 2.5
times when tools were intensively quenched in
water solutions of chlorides. Special additives
were used to prevent corrosion. Performance
data for the punches are given in Table 6.
Dies for growing artificial diamonds must survive repeated cycles of high pressure and high
temperature. Intensive quenching of high-speed
steel dies increased their service life by 1.5
times. These dies are intended for growing artificial diamonds and work with heavy cyclical
temperature and pressure loads. Despite this, the
service life of these dies under such loads became greater by 1.5 to 2 times.
It should be noted that these dies were
quenched in water solutions of CaCl2 with special admixtures preventing the corrosion on the
surface. The intensification took place due to the
destruction of the steam film. Instead of aqueous
solutions of salts, one can use aqueous solutions
Table 6 Tool life of punches in automatic
forming machine (machine capacity: 175
strokes/min)
Durability(a), No. of strokes
Test
No.
1
2
3
4
5
6
7
Existing
technology
Intensive
technology
Increase in
tool life(b)
6,460
6,670
3,200
4,000
6,620
2,890
2,340
15,600
16,500
5,300
12,075
8,110
10,500
7,300
2.4
2.5
1.65
3.0
1.2
3.6
3.1
(a) Number of strokes until a tool wore out. (b) Increase in tool life is
due to intensive quenching
of polymers of optimal concentration exhibiting

inverse solubility (Ref 19). During cooling in
aqueous solutions of polymers exhibiting inverse solubility, a polymer film forms on the surface of parts, and thus the duration of the fullfilm boiling is reduced to the minimum. The
mechanism of the heat transfer in these conditions has been studied insufficiently and requires
the further experimental studies. In still another
application (Fig. 30), the service life of dies
made of ShKh15 (AISI 52100) bearing steel
doubled when the tools were intensively
quenched.
Intensive heating and cooling have been
widely used in practice. Some benefits of this
technology are given in Ref 42 and 45 to 55 and
in Table 7. This technique, on the whole, produces good results at numerical experiments and
design of technological conditions.
15
50
50
Fig. 30
Table 7
Die made out of AISI 52100 steel quenched

intensively
Thermal and Physical Fundamentals

of Making High-Strength Materials
In 1967, it was established that a high cooling
rate within the martensite range results in not
only the creation of high compressive surface
stresses, but also the additional strengthening of
material (Ref 56). These investigations are discussed in more detail in Ref 14, 18, 57, and 58.
A rapid induction heating and intensive cooling
of steel parts where the depth of the quenched
surface layer is controlled, which increases the
service life, is described in (Ref 2). Steel
quenched using this method generally has low
depth of hardened layer and fine grain with arrested growth of austenite grains at high temperatures. Due to limited hardenability, compressive stresses appear on the surface of such
steel parts and the fine grain provides high
strength. In addition to providing an increase in
the service life of such heat treated steel parts,
there is an opportunity to replace relatively expensive high-alloy steels with less expensive
lower-alloy steels and replace fire and environmentally dangerous quench oils with water and
water-based quenching solutions. However, the
depth of hardness in steel parts hardened using
this method is controlled by the chemical composition of the steel parts being hardened.
Steel quenching where the depth of the hardened surface layer is controlled in accordance
with this method is made in water jets. The service life of such heat treated steel parts where
the depth of the hardened surface layer is controlled generally increases when compared to oil
quenching. However, it is often necessary to select or create an appropriate steel alloy for use
in steel parts having different configurations and
sizes to obtain the effect of high surface compressive stresses.
In addition, with this quenching method no
criteria exist for calculating the rate of water flow
for steel parts having different configurations
and/or sizes. Thus, a relatively high water flow
rate is normally chosen for all steel parts that is
not always justified, results in unnecessary energy expenses, and makes the industrial process
more complicated than necessary. While the
high service life of steel parts where the depth
of the hardened surface layer is controlled is considered an advantage for certain steel grades,
other steel grades can also achieve the effect of
increased strength (as compared with steel
quenching methods known before) and high residual compressive surface stresses if the heat
treating parameters are properly controlled. The
heat treating method described previously is en-
tirely dependent on the composition of the steel

alloys available. As a practical matter, it may be
difficult to obtain steel alloys having a suitable
composition. Accordingly, in practice, the hardening method should be adapted to those steel
alloys that are available.
Another known steel-quenching method is described in Ref 59. This steel-quenching method
involves shell hardening, which results in uniform quenching of the entire surface to a certain
depth until reaching high hardness using intensive jet cooling. In this method, examples of the
application of medium carbon 1045 steel are
given. One advantage of this method is the opportunity to increase the service life of steel parts
using standard carbon steels, rather than alloy
steels where the depth of the hardened surface
layer is controlled by the composition of the
steel. As discussed in publications (Ref 60), no
consideration in this method is given to the parameters necessary to optimize the depth of the
hardened surface layer, and the following correlation that the depth of the hardened surface
layer should be changed for steel parts having
different configurations and sizes should be
added:
Dd
constant
D
(Eq 13)
where Dd is the optimal hardened depth, and D

is the cross-sectional thickness. This correlation
was developed in Ref 60 and is considered to
provide a foundation for the quenching apparatus and method for hardening steel parts. In addition, this method does not have any criteria
allowing the calculation of the optimal cooling
solution quench flow, and the technological process is not optimized.
Another steel-quenching method is described
in Ref 61. In this method, alloy steel parts are
quenched in such a manner that a hard surface
layer of a given depth and an arbitrarily hard
matrix are obtained. For given steel grades,
ranges for hardening regimes are found by experimentation to increase the service life of such
steel parts. One example of this method involves
an alloy steel specimen containing 0.65 to 0.85%
C, 0.23 to 0.32% Si, 0.4 to 0.9% Mn, approximately 2% Ni, 0.5 to 1.5% Cr, and 0.1to 0.2%
Mo that is heated to 800 to 850 C and spray
quenched with water fed under a 0.4 to 0.6 MPa
pressure for 0.2 to 0.8 s. The steel specimen is
then isothermally heated at 150 to 250 C for 10
to 50 min. This method considers only high-carbon alloy steels. In this method, the depth of the
hard surface layer is not optimal for steel parts
Production applications of intensive-quenched limited-hardenability steels
Applications
Gears, modulus, m 5 to 8 mm
Large-modulus gears, m 10 to 14 mm
Truck leaf springs
Rings and races of bearings thicker than 12 mm
Former steel and process
18 KhGT
12 KhN3A
60 C2KhG
ShKh15CG and 20Kh2N4A
New steel
and process
58 (55PP)
ShKh4
45C
ShKh4
Advantages
No carburizing; steel and part costs decrease; durability increases

No carburizing, durability increases by 2; steel cost decreases by 1.5
Weight decreases by 1520%; durability increases by 3
No sudden brittle fracture in service; durability increases by 2; high production rate

having different configurations and/or sizes and,
because of this, steel strengthening is not consistently achieved in all parts. In addition, this
method does not consider the optimization of the
quenchant solution circulation rate.
Reference 62 describes a method of intensive
quenching steel parts using intensive jet cooling.
The author points out that the method of intensive cooling is applied to superficial hardening
of small parts (shafts, axles, pins, etc.) made of
alloy steels. The author underlines that under the
condition of very intensive cooling, the deformation reduction is observed.
Other authors (Ref 63) came to the conclusion
that it is possible to provide a method for producing high-strength steel by heat treatment
within a short time. Steel having the composition
(wt%) 0.60 to 0.70% C, 0.35% Si, 0.60 to 0.80%
Mn, 12.50 to 13.50% Cr and the rest Fe with
inevitable impurities is heated to a quenching
temperature by induction heating, and this
heated steel is rapidly cooled (Ref 63). The authors have considered only induction heating and
rapid cooling for alloy steel having 0.60 to
0.70% C.
The intensive steel-quenching method described in Ukraine Patent (Ref 21) deals with
heating, cooling until the appearance of maximum compressive surface stresses, and then isothermal heating (tempering). This method is
based on cooling in the range of 0.8 Kn 1,
where Kn is the Kondratiev number, until reaching maximum compressive surface stresses, then
isothermally heating at martensite start (Ms) temperature until the complete transformation of the
overcooled austenite of the matrix occurs, and
then tempering. At the time of the end of intensive cooling, the optimal depth of the hard layer
is automatically reached, which meets the condition in Eq 13.
The time of reaching maximum compressive
stresses on the surface can be calculated by Eq
12. In most cases, the time of the end of intensive
cooling can be tied with the end of nucleate boiling. In this case the quenching process can be
controlled, since the noises connected with the
nucleate boiling are controlled with special devices (Ref 64). More detailed information on
regularities of nonstationary nucleate boiling for
steel part quenching is given in Ref 65 and 66.
These works consider the so-called self-regulated thermal process, which is based on the fact
that the surface temperature for the body to be
quenched during nucleate boiling is supported at
the level of the quenchant boiling temperature
and differs very little from this value.
The regularities of the nonstationary nucleate
boiling, which have been discovered, have been
used as a foundation of the new intensive steelquenching method, which provides the additional hardening (superhardening) of the material and reduces the deformation; it guarantees
the absence of quench cracks during intensive
quenching (Ref 49). The intensive cooling in accordance with the specified method should be
terminated at the time of the end of the self-regulated thermal process, and then, after the tem-
perature field becomes equal at the cross sections

of steel parts, intensive cooling should be continued to room temperature or below (Ref 49).
That is why it is very important to elaborate a
quenching apparatus for hardening a multitude
of alloy steel parts used in, for example, metallurgy, machine construction, bearing and tool industry, quenching of carburized parts, and parts
heated by induction, salt bath, the usual oven
heating, and vacuum furnaces. At the end of the
quench, some steel parts will benefit from isothermal cooling in the air (self-tempering). This
process method results in additional strengthening of steel parts, and maximum compressive
surface stresses are achieved, resulting in increased service life of the steel parts. Relatively
expensive alloy and high-alloy steels can be replaced with less expensive low-alloy or standard
carbon steels and instead of an oil-based quenching material, water or a water-based solution is
used (Ref 49).
The developed intensive methods of steel part
heating and cooling have great prospects, since
because of them it is possible to save materials
and finances as well as increase the labor productivity considerably. These technologies are
also ecologically clean.
Summary
It has been established that one can control
the value of residual stresses on the surface

of quenched parts.
It has been shown that one can forecast the
time and place of the formation of quench
cracks on the basis of criterion of PisarenkoLebedev.
It has been established that for quenching of
wedge-shaped bearing rings there is a definite
time at which the distortion becomes zero. It
is suggested that this state can be fixed by the
intensification of heat transfer at the second
stage within the range of martensite transformations.
The intensive and uniform cooling within the
range of martensite transformations reduces
the distortion of quenched parts in connection
with the formation of a thin martensite layer
on the entire surface of a quenched part.
Two-step quenching of carburized parts,
which lies in the moderate cooling at the first
stage and intensive cooling at the second
stage within the range of martensite transformations, also results in the reduction of distortion of quenched parts.
A new approach has been developed for the
reduction of distortion and elimination of
crack formation by intensification of heat
transfer at the phase transformations.
Basic results of the calculations of intensive
cooling have been proved by numerous experiments made in recent years.
The intensive heating of steel parts has the
effect of finer austenite grains and transition
to higher temperatures of critical points on the
iron-carbon diagram.
In many cases, intensive heating to the aus-
tenite temperature is equivalent to slow

growth of austenite grains, which is reached
due to optimal addition of alloys such as titanium, zirconium, and vanadium.
Intensive heating and intensive cooling improve mechanical properties of the material
and reduce the distortion of quenched parts,
which is unexpected and explained by the formation of the uniform martensite shell on the
entire surface of the part.
The history of the development of intensive
heat steel treatment methods has been discussed, and the recent achievements in this
field, which have been patented in many
countries, have been considered.
REFERENCES
1. N.I. Kobasko, Basics of Intensive Quenching, Adv. Mater. Proc./Heat Treat. Prog.,
Part I, Vol 148 (No. 3), Sept 1995, p 42W
42Y; Part II, Vol 150 (No. 2), Aug 1996, p
40CC40EE; Part III, Vol 153 (No. 2), Feb
1998, p 36FF36HH
2. K.Z. Shepelyakovskii and F.V. Bezmenov,
New Induction Hardening Technology, Adv.
Mater. Proc., Oct 1998, p 225227
3. G.S. Pisarenko and A.A. Lebedev, Deformation and Strength for Complex Stress
State, Naukova Dumka, Kiev, 1976 (in Russian)
4. T. Inoue and K. Arimoto, Development and
Implementation of CAE System Hearts
for Heat Treatment Simulation Based on
Metallo-Thermo-Mechanics, J. Mater. Eng.
Perform., Vol 6 (No. 1), Feb 1997, p 5160
5. V.S. Morganyuk, Method for Calculating
the Thermal and Stress-Strain State During
Hardening, Prob. Prochn., No. 6, 1982, p
8085
6. O.C. Zienkiewiez, The Finite Element
Method in Engineering Science, Mir, Moscow, 1975, 542 pages
7. A.A. Popov and L.E. Popova, Isothermal
and Thermokinetic Transformation Diagrams for Supercooled Austenite, Mashgiz,
Moscow, 1961, 430 pages (in Russian)
8. T. Reti, L. Horvath, and I. Felde, A Comparative Study of Methods Used for the Prediction of Nonisothermal Austenite Decomposition, J. Mater. Eng. Perform., Vol 6
(No. 4) Aug 1997, p 433441
9. N.I. Kobasko, V.S. Morganyuk, and Yu.B.
Gnuchiy, Study of Machine Part Manufacturing Processes in Machine Industries,
Znanie, Kiev, 1979, 18 pages (in Russian)
10. N.I. Kobasko and V.S. Morganyuk, Investigation of Thermal and Stress State in the
Process of Heat Treatment, Znanie, Kiev,
1981, 18 pages (in Russian)
11. S.F. Yuriev, Specific Volumes of Phases in
Martensite Transformation of Austenite,
Metallurgizdat, Moscow, 1950, 45 pages
12. V.S. Morganyuk, N.I. Kobasko, and V.K.
Kharchenko, Strength Problems, No. 6,
13. N.I. Kobasko, L.A. Kozdoba, and A.F.
Moshnyanskii, Solution of Inverse Nonlinear Problems of Non-Steady Thermal Conductivity on Electric Models, Teplofiz. Teplotekh., No. 31, 1976, p 256
14. N.I. Kobasko, Steel Quenching in Liquid
Media under Pressure, Naukova Dumka,
Kiev, 1980, 206 pages (in Russian)
15. V.S. Morganyuk, N.I. Kobasko, and A.N.
Kulakov, Predicting the Deformation of
Bearing Rings During Hardening, Metalloved. Term. Obrab. Metal. (MiTOM), No.
9, 1982, p 2224 (in Russian)
16. N.I. Kobasko, V.S. Morganyuk, and L.V.
Lushchik, Study of Thermal Stresses
Formed in Steel Products Due to High-Rate
Quenching, Sov. Mater. Sci. Rev., No. 1,
1987, p 2732
17. G.F. Golovin, Residual Stresses, Strength
and Deformation During High-Frequency
Surface Quenching, Mashinostroenie, Leningrad, 1973 (in Russian)
18. N.I. Kobasko, Intensive Steel Quenching
Methods, Handbook: Theory and Technology of Quenching, B. Liscic, H.M. Tensi,
and W. Luty, Ed., Springer-Verlag, Berlin,
1992, p 367389
19. G.E. Totten and M.A.H. Howes, Steel Heat
Treatment Handbook, Marcel Dekker,
1997, 1192 pages
20. G.A. Temkin, Analytical Theory of NonStationary Heat and Mass Transfer in the
Process of Drying and Inverse Problems of
Analytical Theory of Drying, Naukai Tekhnika, 1964
21. N.I. Kobasko, No. 4448, Ukrainian Patent,
Bulletin 6-I, 1994
22. V.A. Lykov, Heat Conductivity Theory,
Vysshaya Shkola, Moscow, 1967
23. G.M. Kondratiev, Heat Measurements,
Mashgiz, Moscow, 1957, 244 pages (in
Russian)
24. M.A. Aronov, N.I. Kobasko, J.F. Wallace,
D. Schwam, and J.A. Powell, Practical Application of Intensive Quenching Technology for Steel Parts, Ind. Heat., April 1999,
p 5963
Methods on the Way of Automation, Proc.
First Int. Automotive Heat Treating Conf.,
R. Colas, K. Funatani, and C.A. Stickels,
Ed., ASM International, 1998
26. V.N. Gridnev, Yu.Ya. Meshkov, S.P. Oshkaderov, and N.F. Chernenko, Technological Fundamentals of Electrical Heat Treatment of Steel, Naukova Dumka, Kiev, 1977,
204 pages
27. L. Bezdek, Castice AlN a mezni rozmer austenitickeho rzna, Kovove Mater., Vol 5 (No.
4), 1967, p 335344 (in Russian)
28. C. Dasarathy and R.C. Hudd, An Example
of Secondary Recrystallization Induced by
Aluminum Nitride Precipitation, Acta Metall., Vol 15 (No. 10), 1967, p 16651671
29. V.I. Psarev and Yu.I. Makovichuk, Coagulation of Dispersion Phases at Tempering of
Steel Quenched, Izv. Vuzov, Ferrous Met-
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
allurgy, Vol 7, 1971, p 127132 (in Russian)

K.Z. Shepelyakovskii, A.N. Marshalkin,
and V.A. Kanygin, Influence of Quantity
and Dispersion Capacity of Hardly Soluble
Particles upon the Size of Austenite Grain
for the Steel with 0.6% C, MiTOM, Vol 8,
1973, p 913
M. Hetmanczyk, B. Rauszer, B. Trzcionka,
and H. Woznica, Influence of Vanadium
Microadmixture upon Kinetics of the
Growth of Austenite Grain for Steel 18G2A,
Wiad. Hutn., Vol 28 (No. 78), 1972, p
258277
S.V. Zagulyaeva and M.I. Vinograd,
Growth of the Austenite Grain in Construction Steel with Titanium, MiTOM, Vol 10,
1972, p 4045
T. Matsuoka, Kinetics of Secondary Recrystallization in Silicon Iron, Trans. Iron
Steel Inst., Jpn., Vol 7 (No. 5), 1967, p 238
243
R. Mitsche and A. Legat, Size of the Austenite Grain of Steel during Heat Treatment
in the Field Where the Temperature
Changes, Radex Rundschau, No. 34, 1967,
p 695698 (in German)
V. Orsovy, L. Hyspecka, and K. Mazanec,
On the Issue of Recrystallization during the
Heat and Mechanical Treatment at High
Temperature, Mem. Sci. Rev., Met., Vol 69
(No. 4), p 243248 (in French)
H. Weiss, A. Gittins, G.G. Brown, and W.J.
Tegart, Recrystallization of a Niobium-Titanium Steel in the Austenite Range, J. Iron
Steel Inst., Vol 211 (No. 10), 1973, p 703
709
L.M. Panfilova, V.M. Farber, M.I. Goldshtein, and B.Z. Belenkiy, Dissolution and
Isolation of Special Carbides and Nitrides in
Construction Steels, Structural and Phase
Transformation During Heating of Steels
and Alloys, No. 1, 1969, p 8085 (in Russian)
T.L. Capeletti, L.A. Jackman, and W.J.
Childs, Recrystallization following HotWorking of a High-Strength Low-Alloy
(HSLA) Steel and a 304 Stainless Steel at
the Temperature of Deformation, Metall.
Trans., Vol 3 (No. 4), 1972, p 789796
G.A. Wilber, J.R. Bell, J.H. Bucher, and
W.J. Childs, The Determination of Rapid
Recrystallization Rates of Austenite at the
Temperatures of Hot Deformation, Trans.
Met. Soc. AIME, Vol 242 (No. 11), 1968, p
23052308
M. Lamberigts, T. Greday, and L. Habraken, Recrystallization of Austenite Deformed during Hardening of Metal Constructions, Mem. Sci. Rev. Met., Vol 70 (No.
6), 1973, p 457465
I.N. Kidin, Physical Fundamentals of Electrical Heat Treatment of Metals and Alloys,
Metallurgiya, Moscow, 1969, 378 pages
K.Z. Shepelyakovskii, Strengthening of Machine Parts by Surface Hardening with In-
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
duction Heating, Mashinostroenie, Moscow, 1972, 288 pages

M. Daming, Intense Quenching Method for
Preventing Quench Cracking, Proc. Seventh
Int. Congress on Heat Treatment and Technology of Surface Coating, Vol 2, 1114
Dec 1990, Vneshtorgizdat, Moscow, p 62
71
N.I. Kobasko, Basics of Intensive Quenching. Part IV, Adv. Mater. Proc./Heat Treat.
Prog., No. 12, 1999, p H31H33
N.I. Kobasko, Generalization of Results of
Computations and Natural Experiments at
Steel Parts Quenching, Proc. First Int. Conf.
Thermal Process Modeling and Computer
Simulation, 2830 March 2000, Shanghai
Jiaotong University, Shanghai, China
K.Z. Shepelyakovskii and B. Ushakov, Induction Surface Hardening-Progressive
Technology of XX and XXI Centuries,
Proc. Seventh Int. Congress on Heat Treatment of Materials, 1114 Dec, Vol 2,
Vneshtorgizdat, Moscow, 1990, p 38
K.Z. Shepelyakovskii and R.I. Entin, New
Method of Surface Hardening of Wheels
with the Middle Module, Vestn. Mashinostr., No. 1, 1958, p 53
N.I. Kobasko, No. 4005, Ukrainian Patent,
Bulletin 6-I, 1994
N.I. Kobasko, Alloy Steel Quenching
Method, No. 27059, Ukrainian Patent, Application No. 95020715, 17 Feb 1995
G.E. Totten, N.I. Kobasko, B. Liscic, and
S.W. Han, Advances in Polymer Quenching
Technology, First Int. Automotive Heat
Treating Conf., R. Colas, K. Funatani, and
C.A. Stickels, Ed., ASM International,
1998, p 3744
G.E. Totten, N.I. Kobasko, and M.A.
Aronov, Intensive Quenching Practices for
Automotive Part ProductionAn Overview, First Int. Automotive Heat Treating
Conf., R. Colas, K. Funatani, and C.A.
Stickels, Ed., ASM International, 1998
J.S. Pan, J.F. Gu, M.J. Hu, and X. Chen,
Computer Simulation of Complex Components During Quenching Process and Its Application in Industry, Proc. Seventh Int.
Seminar of IFHT Heat Treatment and Surface Engineering of Light Alloys, 1517
Sept 1999 (Budapest, Hungary), Hungarian
Scientific Society of Mechanical Engineering (GTE), p 217227
X.L. Chen and L. Meekisho, Computer
Simulation of Temperature and Thermal
Stress Fields During Quenching Process,
Proc. Second Int. Conf. Quenching and the
Control of Distortion, ASM International,
1996, p 241
L. Meekisho and X. Chen, Modeling of
Austenite and Martensite Phase Transformation in Salt-Bath Heating and Oil
Quenching Processes, Proc. Seventh ANSYS
International Conference & Exhibition, Vol
3, 1996, p 221
A. Majorek, B. Scholtes, H. Mueller, and E.

Macherauch, The Influence of Heat Transfer
on the Development of Stresses, Residual
Stresses, and Distortions in Martensitically
Hardened SAE 1045 and SAE 4140, Proc.
First Int. Conf. Quenching and Control of
Distortion, 2225 Sept 1992 (Chicago),
ASM International, p 171179
56. N.I. Kobasko, Effect of Structural and Thermal Stresses on Crack Formation at Steel
Quenching, All-Union Conference on Increasing Output and Economical Efficiency
of Heating Furnaces, Dnepropetrovsk, Dec
1967, Abstracts of papers, p 2627
57. R.F. Ganiev, N.I. Kobasko, V.V. Kulik, et
al., Vibration Phenomena in Multiphase
Media and Their Application in Technology, Tekhnika, Kiev, 1980, 144 pages
58. R.F. Ganiev, N.I. Kobasko, and K.V. Fro-
59.
60.
61.
62.
63.
lov, On Principally New Ways of Increasing

Metal Part Service Life, Doklady Akademii
Nauk USSR, Vol 194 (No. 6), 1987, p 1364
1473
R.F. Kern, Intense Quenching, Heat Treat.,
No. 9, 1986, p 1923
N.I. Kobasko and V.S. Morganyuk, Study of
Thermal and Stress-Strained State at Heat
Treatment of Power Plant Industry Parts,
Znanie Ukr. SSR, Kiev, 1983, 16 pages
T. Naito, Method of Steel Quenching, No.
59170039, Japanese Patent, Application
6148514, 16 Aug 1984
Sh. Owaku, Intensive Cooling, Kindzoku
Met. Technol., Vol 57 (No. 3), 1987, p 48
49
K. Kawasaki, H. Koga, H. Yokota, and T.
Nishikawa, Production of High Strength
Steel, No. 11293336 A, Japanese Patent,

26 Oct 1999, Filed: 14 April 1998
64. S.G. Povsten, N.I. Kobasko, A.A. Moskalenko, and A.A. Tyltin, On Application of
Noise Control in New Methods of Steel
Quenching, Proc. Seventh Int. Congress on
Heat Treatment of Materials, 1114 Dec
1990, Vol 3, Vneshtorgizdat, Moscow, p
5357
Methods on the Way of Automation, Proc.
First Int. Automotive Heat Treating Conf.,
1315 July 1998 (Puerto Vallarta, Mexico),
66. N.I. Kobasko, Self-Regulated Thermal Process at the Hardening of Steel Parts, Promteplotekhnika, Vol 20 (No. 5), 1998, p 10
14

DOI: 10.1361/hrsd2002p331

Effect of Cryogenic Cooling of Residual

Stresses, Structure, and Substructure
Ioan Alexandru and Vasile Bulancea, Technical University of Iasi, Romania
CRYOGENIC COOLING represents a useful

and efficient means that, if correctly applied
within the secondary heat treatment technologic
flow, may cause marked increases in the service
properties of steel parts and tools. The main purpose of cryogenic cooling is the continuation of
the transformation to martensite of as much undercooled retained austenite as possible. After
applying the cryogenic cooling to some steels
with martensite finish (Mf) critical temperature
below 273 K, such as plain-carbon steel with
more than 0.6% C as well as alloy and high-alloy
steels, important improvements were noticed in
some of the service properties, including durability, wear resistance, fatigue life, hardness, and
so on. To ascribe these marked improvements
exclusively to the increase of martensite and the
corresponding decrease of retained austenite
would be rather simplistic and nonscientific.
Recent research carried out at the Gh. Asachi Technical University of Jassy, Romania
(Ref 13) and the Technologic Institute of Louisiana has emphasized that besides the continuation of the transformation of austenite to martensite, cryogenic cooling also provides:
A favorable respreading of alloying elements
between martensite and the carbide phases
The precipitation of a large amount of alloy-
ing carbide elements of very small size (under

1 lm)
A favorable redistribution of residual stresses
As shown in Fig. 1, the transformation of austenite to martensite takes place at cryogenic temperatures both during cooling and during subsequent heating back to the ambient temperature.
The larger the cooling rate (curve B), the more
intense the transformation of austenite to martensite, both during isothermal transformation
and during heating back to ambient temperature,
due to the thermal stresses induced during cooling.
Cryogenic cooling proved to be effective only
in those cases where the retained amount of austenite is larger than 10%, at room temperature,
after classical (ordinary) quenching. Moreover,
cryogenic cooling has been especially effective

for some parts (gear wheels, ball bearings, elements of injection pumps) and tools (for plastic
deformation, for measuring and verifying) that
require high values for certain service properties
such as hardness, wear resistance, fatigue life,
and durability, as well as dimensional and shape
stability. For most of the alloy and plain-carbon
steels, cryogenic cooling must be immediately
applied after classical quenching, while for highspeed steels (HSS) it may be applied also after
a previous tempering, particularly in the case of
complex shape tools.
The typical thermal range for cryogenic cooling lies between 203 and 153 K. It should be
noted that cryogenic cooling cannot represent a
final heat treatment since subsequent tempering
heat treatment is absolutely necessary to achieve
thermal stress relieving, fine carbide precipitation, and more favorable respreading of alloying
elements between martensite and austenite.
The possible heat treatment variants both with
and without cryogenic cooling applied to highspeed steels are shown in Fig. 2, where the heating temperatures range between 1493 and 1563
K, the specific cryogenic temperatures are 203,
153, and 77 K, and the tempering temperatures
are from 813 to 853. The variants designated by
B and E are classical for high-speed steels and
for plain-carbon and low-alloy steels, respectively. The variants A, C, and F are not practically applicable, being important only for research purposes, while the variants E and D are
also applicable for plain-carbon and alloy steels,
after a suitable adjustment of the temperature.
Finally, the variants B, G, H, I, K, and M are
applied mostly to HSS.
Residual Stresses Induced by

Cryogenic Cooling and Their
Measuring Methods
Role of Internal Stresses within
Martensitic Transformation at
Cryogenic Temperatures
The accurate determination of residual
stresses within the cryogenically cooled metallic
materials constitutes a major factor for the

proper design, building, and operating of the machine parts and tools since these stresses strongly
influence the aforementioned service properties.
The residual stresses are those stresses that persist in a solid body as a consequence of the treatment and mechanical manufacturing in the absence of any external or internal loads (forces,
momenta, or accelerations). These stresses make
up a force system that in normal conditions
reaches local and global balance, which is thus
hardly detectable, unless this balance is altered.
According to the volume size at which they
manifest themselves, three types of residual
stresses are noticeable: first-order stress (macroscopic), second-order stresses (microscopic),
and third-order stresses (ultramicroscopic), as illustrated in Fig. 3. Quantitatively, the cumulated
residual stress in any point can be expressed as:
rR r1rIIrII
(Eq 1)
The residual stress state is typically determined

by the first-, second-, and third-order stresses
overlapping, as shown in Fig. 4. In a biphase
material ( b) the residual stress state is also
the result of overlapping stresses, but Fig. 5
shows an even more complex stress distribution
compared to Fig. 4. The residual stresses induced
by cryogenic quenching heat treatments are especially of the second order, being the effect of
both shape and volume accommodation of martensite within the austenite matrix. Consequently, as compared to classical quenching, it
follows that cryogenic cooling causes a decrease
of structural residual stresses, since it reduces the
amount of retained austenite.
In order to trigger the martensitic transformation it is necessary not only that martensite
becomes more stable than austenite, but also that
a certain critical value is reached for the difference between their corresponding free-energy
levels. This critical free-energy difference requires austenite to be considerably undercooled,
below its thermodynamic equilibrium temperature. Thus, for an austenite undercooling of 200
K, the above difference reaches 350 cal/at. gr.
Etot Esurf Estrain Echem
(Eq 2)
where Esurf is the surface energy at the austenite/

martensite interface, Estrain is the coherent deformation energy caused by the misfit between the
two lattices (that accompanies an elastic deformation), and Echem is the free-energy difference
that occurs at the formation of the martensite
volume (corresponding to a plastic deformation).
Calculations have emphasized that martensite
develops as a result of heterogeneous nucleation
and that this nucleation is enabled by a favorable
reaction between two stress fields: one caused by
the Bain distortion and the other clustered
around one or more dislocations from the undercooled austenite. Therefore, the dislocations
are the preferred loci for martensite nucleation,
which occurs between the critical martensite
start (Ms) and Mf temperatures.
Considering the stress state in a polycrystalline body, it is obvious that thermal variation is
accompanied by the occurrence of an internal
stress field (Ref 9). Such a stress field can be
generated by several sources such as: the temperature gradient on the cross section, the anisotropy of mechanical properties, the different
state of atoms on the grain boundary and inside
the grains, the nonhomogeneity of chemical
composition, the structural imperfections, the
different space orientation of crystals, the differ-
ent specific volumes, and thermal expansion coefficients of austenite and martensite below the
critical Ms temperature.
All of the theories given in literature agree that
martensitic transformation occurs in certain thermodynamic conditions by means of the nucleation and instant growth of martensite nuclei.
These theories, however, have different points of
view regarding the role of the thermal agitation
of atoms or of the stress fields within the martensitic transformation (thermal theoryG.V.
Kurdjumov, G. Fisher, et al. [Ref 1013]; stress
theoryA. Guleav, S. Crussard, et al. [Ref 4,
14, 15]; mixed theory, Bain distortion assisted
by dislocationsK.E. Easterling, A.R. Tholen
et al. [Ref 16]). Consequently, it is important to
settle whether it is the thermal agitation (oscillations) or the stress field that plays the major
role within the martensitic transformation in the
cryogenic field.
As previously mentioned, martensitic transformation also proceeds when the thermal oscillations are prevented in the highest degree, such
14531563
Temperature, K
Besides the large free-energy difference, the

martensitic transformation also requires an additional energy source to be available in austenite in order to overcome the energy barrier that
separates the parent and product phases. Such an
additional energy source for the initiation of the
martensitic transformation may be supplied by
two sources: nonuniform thermal gradients and
mechanical stresses.
Since in steels with carbon amounts higher
than 0.6% the martensitic transformation also
occurs in the proximity of 0 K, the first of the
additional energy sources must be excluded.
Therefore, only the second one becomes available.
Easterling and Tholen (Ref 8) have determined the total energy exchange, associated with
the martensitic transformation, under the form:
803843
Air
203
153
77
C1
C2
C3
F1
F2
G1
G2
F3
D3
H
203
203
G3
203
D1
D2
203
Oil
Time, h
Fig. 2
as in the proximity of 0 K. Moreover, as illustrated in Fig. 6, which refers to an alloy steel

with Ms 220 K cooled to 4 K, the lower the
cooling rate the larger the amount of transformed
martensite.
A more comprehensive illustration of the effects of stress on martensitic transformation is
obtained by comparing the transformation kinetics for two identical parts, with and without residual stresses, respectively. Such an example is
given in Fig. 7, for an AISI H12 steel quenched
from 1423 K, with Ms lower than 273 K. Thus,
the transformation curve 1 was obtained for an
ordinary polycrystalline specimen (with firstand second-order residual stresses), while curve
2 corresponds to a single-crystal powder specimen (with negligible stresses). It is noticeable
that martensitic transformation occurred only
within the polycrystalline specimen (curve 1).
Therefore, it is the stresses and not the thermal
agitation that represent the driving force for martensitic transformation at cryogenic temperatures.
203
203
Possible heat treatment variants for high-speed steels. Source: Ref 5
40
Martensite, %
B
30
A
20
10
B
0
313
R
273
233
193
153
113
73
Temperature, K
Fig. 1 Transformation of austenite to martensite at cryogenic temperatures. Curve A, low cooling rate;
curve B, high cooling rate. Source: Ref 4
First-order stresses
(macrostresses)
Second-order stresses
Third-order stresses
(microstresses and microstrains)
Fig. 3
Different observation levels of residual stresses. Source: Ref 6
Effect of Cryogenic Cooling of Residual Stresses, Structure, and Substructure / 333

Evaluation of Residual Stresses after
Cryogenic Cooling
Based on these observations, it appears that

the lower the temperature within the cryogenic
field the weaker the residual stresses that actuate
the martensitic transformation, since martensitic
transformation develops with more and more
difficulty as the temperature drops closer to Mf.
There are a multitude of methods for determining residual stresses, yet x-ray diffractometry
is the only one that allows an estimation of the
The formation of crystallites and subgrains
III
II
R
with dimensions within the range of 102 to

101 lm
The elastic distortion of relatively large crystals (1 lm)
The formation of stacking faults
III
mG
II
Evaluation of First-Order Residual

Stresses. In face-centered cubic (fcc) materials,
residual-stress-induced faults cause both displacements and asymmetries of the diffraction
maxima distribution; this effect is less intense in
body-centered cubic (bcc) materials, while in
hexagonal close-packed (hcp) materials, only a
symmetrical enlargement of the above distribution is observed. In order to evaluate the residual
stress that has caused a certain strain, the interplanar spacings must be determined both for the
undeformed and the deformed material (dohkl and
dhkl, respectively), and the elastic constants
(Youngs modulus, or E, and Poissons ratio, or
t) must be known.
According to Braggs law, a relationship exists between the interplanar spacing (d) and the
diffraction angle (h) for the same wavelength
(k). If the spacing is expressed as a function of
the diffraction angle and the relationship is differentiated, it follows that:
III
Fig. 4
Retained stress, R
First-, second-, and third-order residual stresses overlapping in a single-phase material. Source: Ref 6
II
Dd/d [cot h]Dh
<Il>
<Il>
<>
<>
r E(Dd/d) E[cot h]Dh
80
Martensite, %
Martensite, %
1
20
10
40
r
20
2
250
200
150
100
50
0
313
Temperature, K
Martensitic transformation curves of an alloy
steel 0.85% C, 14.8% Ni, 2.1% Mn, cooled
down to 4 K. Curve 1, slow cooling; curve 2, fast cooling.
Source: Ref 4
Fig. 6
293
0
300
60
273
233
193
153
113
E
p
[cot h0]
2(1 t)
180
(2h)
(2h)
K
(sin2w)
(sin2w)
(Eq 5)
73
Temperature, K
Martensitic transformation in AISI H12 steel,
quenched from 1423 K. Curve 1, ordinary polycrystalline specimen; curve 2, single-crystal powder.
Source: Ref 4
Fig. 7
(Eq 4)
where Dh is the angular displacement, caused by

stress on the diffractogram, of the diffraction line
corresponding to the h angle, as compared to the
same line obtained with a standard unstressed
specimen.
Surface stresses in the direction OX can be
determined with the relationship:
First-, second-, and third-order residual stresses overlapping in a biphase material. Source: Ref 7
30
(Eq 3)
where the left terms (Dd/d) represents the strain.

Therefore, the stress (r), expressed by the
Hookes law, becomes:
Ill
Fig. 5
dynamics of stresses up to atomic group level.

The quantitative determination of residual
stresses is based on Hookes law, which connects stresses to strains, the latter being, in fact,
the very parameters that are actually measured.
Considering that every type of stress induces a
certain structural change, this change will be accompanied by an alteration of the diffraction pattern, namely an enlargement of the diffraction
lines. This enlargement can be determined by:
where w is, according to Fig. 8(a), the angle between the normal of the reflecting planes and the
normal at the surface of the specimen, and h0 is
the diffraction angle for the unstressed specimen.
Equation 5 can also be written under the form:

r K B
(Eq 6)
where
K
B
E cot h0
p
2(1 t) 180
(2h)
(sin2w)
(Eq 7)
The value of the coefficient K depends on the

specimen material and on the characteristic
wavelength corresponding to h0. For several materials, the calculated K values are given in
Ref 17.
The value of the coefficient B from Eq 7 is
determined, according to Fig. 8(b), as the slope
of the straight line 2hw f(sin2 w). In fact, the
values of the h angle are measure for o group of
planes (with the same Miller indices), and a plot
similar to Fig. 8(b) is drawn from which the
slope B is determined.
The group of planes is chosen to maximize the
measuring accuracy of h. Table 1 lists both the
coefficient K and other parameters requested for
the evaluation of the first-order residual stresses
for several technical metallic materials.
Evaluation of Second-Order Residual
Stresses. Second-order residual stresses are located at microscopic scale at the level of mosaic
blocks or crystalline grains and cause an enlargement of the diffraction maxima. Since the enlargement of the diffraction line can be correlated with the average grain size, provided it is
less than 104 cm, it is expected that both heat
and mechanical treatments cause crystallite
breaking. If such a process occurs, then the enlargement of the diffraction line should be proportional to the ratio k/cos h in accordance with
the relationship:
ehkl
0.94k
b1 cos h
(Eq 8)
in which ehkl is the average size of the crystallite

(mosaic block) along the normal direction on the
(hkl) plane, k is the wavelength of the applied xray, b1 is the physical width of the diffraction
line on the diffractogram, and h is the corresponding diffraction angle for the analyzed (hkl)
line.
However, experimental results have emphasized that the above enlargements are proportional to tan h and are independent of k. This
disagreement could be explained if larger average crystallite sizes are taken into consideration
(about 104 cm) in order to avoid the enlargement of the diffraction lines and the deformation
of each crystallite due to the presence of secondorder stresses.
Assuming the existence of second-order residual stresses within the lattice of a stressed mosaic
block, then certain close-packed planes should
exist with altered interplanar spacing, under the
form dhkl a Da.
The nonhomogeneity degree of the interplanar
spacing, designated by g(gDd/d), can be determined by three methods known as direct
method, approximation method, and harmonic
analysis method.
The direct method allows evaluation of second-order residual stresses (rII) by means of the
nonhomogeneity degree of the interplanar spacing expressed as:
g B 2 B 20 /4R tan h
(Eq 9)
where B0 is the experimental width of the diffraction line of the standard specimen, B is the
experimental width of the diffraction line of the
analyzed specimen, is the photometer parameter, namely the ratio between the paper speed
and the counter speed, R is the radius of the cylindrical chamber, and h is the diffraction angle
corresponding to the diffraction line under study.
The approximation method considers that
both the second-order internal microstrains and
the breaking of mosaic blocks with dimension
Dhkl 10 lm cause a supplementary enlargement of the diffraction lines. When the physical
width b1, corresponding to dimensions Dhkl
0.1 lm, is influenced only by the microstresses
or only by the mosaic blocks, then one of the
following conditions must be true: b2 /b1 tan
h2 /tan h1 or b2 /b1 cos h1 /cos h2, respectively,
where h1 and h2 are angles and b1 and b2 the
widths corresponding to the diffraction lines
originating from the same reflecting plane but
obviously with different reflection orders, that is,
the planes (hkl) and (nh, nk, nl), respectively.
When the actual width of the diffraction line
is influenced both by the small size of mosaic
blocks and the presence of second-order residual

microstresses, then the influences may take a
Gauss (G), a Cauchy (C), or a Lorentz (L) form.
For any combination under the form G-G, C-C,
L-L, G-C, G-L, C-L, the physical width of the
diffraction line becomes:
b
mn
M(x)N(x)dx
where M and N are any of the distribution functions associated with the widths m and n (m is
the physical width of the line caused by the influence of mosaic blocks; n is the physical width
affected by second-order residual microstress).
Considering a combination of the type L-C,
the size of the mosaic blocks may be expressed
according to Eq 8, and the nonhomogeneity degree of the interplanar spacing becomes g Dd/
d n2 /(4 tan h2), where h2 is the diffraction
angle corresponding to the large indices line (nh
nk nl). The determination of n2 is graphically
performed according to the constructive diagram
in Fig. 9, by means of the curves N f (b2 /b1).
Second-order residual stresses are evaluated
according to Hookes law with the relationship:
rII g E (in MPa)
=0
d1
d2
= 90
dmin
(a)
Fig. 8
Ref 5
d1
d2
d3
dmax
1
2
3
d4
(b)
2
3
sin2
(a) Variation of interplanar spacing, d, and the w angle according to the orientation of the reflecting closepacked planes with respect to the stress direction. (b) Dependence of 2hw diffraction angle on sin2 w. Source:
(Eq 11)
where E is the longitudinal electricity modulus

of the material under study.
The harmonic analysis method consists of
both the decomposition of the diffraction line
profile into Fourier series and the consideration
of the width influenced by microstrains. When
the mosaic blocks are large, the enlargement of
the diffraction line may represent a direct microstress evaluation. If besides the microstresses
small mosaic blocks also exist, then the Fourier
coefficient An corresponding to the physical
width of the diffraction line, b, will be:
g
An AD
n An
(Eq 12)
g
where AD
n and An are the Fourier coefficients revealing the influence of the presence of mosaic
blocks and the second-order microstresses, respectively, on the diffraction-line enlargement.
By the logarithmation of Eq 12, under the condition Agn exp(2n212Z n2), it follows that:
lnAn lnAnD 2p212Z 2n
(Eq 10)
(Eq 13)
Knowing An for several lines that differ from

one another by the interference order, the relationship ln An f (H2 K 2 L2) can be obtained for n 1, 2, , which leads to the fact
that the slope of both obtained straight lines
equals Z 2n and their intersection with the coordinate axes is ln AD
n . In the above relationship
Z 2n represents the average component of the deformation vector on the direction a3, from the
space a1a2a3 (Ref 19).
If the indices (nh nk nl) designate a line of the
type (001), (002), (003), , then the relationship
ln An f (l 20) represents a straight line, as shown
in Fig. 10. It should be noted that the dependen-

M 8p2U 2d sin2 h/k2
cies in Fig. 10 have the tendency to curve toward

low or toward high values of n, revealing the fact
that the shape of the diffraction line undergoes a
more intense influence of the mosaic blocks or
the second-order microstrains, respectively.
Evaluation of Third-Order Residual
Stresses. Heat and/or mechanical treatments can
cause a third stressing mode consisting of the
displacement of a restricted group of atoms from
their ideal positions. The change of the diffraction image due to the presence of third-order
stresses seems to be analogous to that determined by the thermal agitation of atoms within
crystals. For this mode of motion, a temperature
factor, e2M, can be introduced as:
U 2d
3U , U
2
where U is the average of the square displacements of atoms on a given direction and U 2d is
the average of the square dynamic displacements
of atoms from their ideal positions at the temperature T, determined as:
2
U 4p hmh uT 4T
9h2T
2
d
(Eq 14)
where u(/T) is the Debye function and is the

Debye temperature.
The first type of imperfection, caused by ther2
1.0
M1 = m110/110
1
0.6
(111)
101
0.4
L = 60
AL
M1, N2
0.8
0.2
(333)
N2 = n220/220
1.3
1.5
1.7
1.9
2.1
2.3
102
2.5
10
2/1
Dependence of the distribution functions M(x)
and N(x) on the ratio of the physical widths of
diffraction lines (110) and (220), recorded for high-speed
steel by means of Mo K characteristic x-ray radiation (constructive diagram). Source: Ref 18
Table 1
20
30
(Eq 15)
50
Family of linear dependencies [ln An (nh nk

nl ); (h2 k2 l 2); n 1, 2, 3, ] for a brass
with 30% Zn that underwent three deformation degrees,
corresponding to the values of L, determined by means of
other methods. Source: Ref 19
Fig. 10
The evaluation of the microstrain U 23, in

solid solutions, is accomplished by eliminating
the influence of thermal oscillations, since a
standard pure annealed steel specimen is used.
Using recorded diffraction spectra, the intensities of the diffraction peaks can be measured
both for the analyzed and the standard specimens, Isp and Ist, respectively. It follows that:
ln(Ist /Isp) 2(M M k)
(Eq 16)
Values of the parameters requested for the evaluation of the first-order residual stresses in several technical metallic materials
Sample material
(the matrix)
Lattice
type
Lattice
parameter, 10 mm
Youngs
modulus (E), GPa
Poissons
ratio m
Characteristic
x-ray radiation
Miller indices of the

diffraction planes hkl
Diffraction
angle, 2h,
Coefficient
K(a) (Eq 7)
Cr K
Co K
Cr Kb
Mn K
Cr K
Co K
Co K
Co K
Cr Kb
Co K
Cu K
Co K
Co Kb
Co K
Co K
Co K
Co K
Co K
(211)
(310)
(311)
(311)
(222)
(420)
(331)
(333)
(311)
(400)
(420)
(400)
(331)
(321)
(400)
(331)
(400)
(121)
156.08
161.35
149.6
154.8
156.7
162.1
148.7
164.0
146.5
163.5
144.7
150.0
145.0
155.5
151.0
146.0
15.84
162.5
30.33
23.51
36.26
29.87
9.40
7.18
12.78
6.41
25.00
12.04
26.42
15.62
18.38
8.30
15.08
17.83
16.55
58.84
Co K
Co K
Cr Kb
Cu K
Cr K
Co K
Co K
Cr K
Co K
Co K
(114)
(222)
(311)
(420)
(222)
(331)
(420)
(211)
(310)
(531)
154.2
142.2
157.7
155.6
152.1
145.1
156.4
153.0
157.5
154.1
17.51
26.17
27.88
29.53
13.11
16.60
11.03
Fe (ferrite-martensite)
bcc
2.8664
20.621.6
0.280.3
cFe (austenitic)
fcc
3.656
19.22
0.28
Brass or aluminum
fcc
4.049
6.90
0.345
Copper
fcc
3.6153
12.73
0.364
Brass 4-6
bcc
fcc
9.045.45
0.380.27
Brass 7-3
fcc
9.04
0.38
Copper-nickel
Tungsten carbide
fcc
fcc
hcp
13.00
52.42
0.333
0.22
Titanium
11.35
0.321
Nickel
hcp
bcc
fcc
3.595
a 2.91
b 2.84
c 0.976
2.9504
3.5238
20.80
0.31
Silver
fcc
4.0856
8.11
0.367
Chromium
bcc
2.8845
Diamond
5.4282
Silicon
40
h2 + k2 + l 2
Fig. 9
8p
sin 2h
U 23
3
k2
L = 90
(222)
L = 120
0
1.1
mal oscillations of atoms within the crystals and/

or by the displacement of atoms from their ideal
positions, has dynamic character, having been
designated as dynamic deformation, characterized by U 2d. Another kind of imperfection is
caused by the static displacements of atoms, Us,
from their ideal position (due to the difference
between the atomic sizes or to the instability induced by the heat and/or mechanical treatments).
According to Hookes law, the above displacements (considered microstrains) are associated with the residual microstresses, also called
third-order residual stresses. Both the third-order
microstrains, U3, and the dynamic displacements, Ud, lead to an intensity decrease of the
diffracted beams, by means of a factor designated by e2k and analogous to the thermal displacement, where:
(a) K [E/2(1 m)] cot h0 p/180. Source: Ref 17

Thus, the microstrain U 23 becomes:
U 23
4p2 2kk2
3a2 sin2h
fect of the diffraction lines, caused by the simultaneous existence of both compression and
tension in either the crystalline lattice or in mosaic blocks.
As mentioned in the section Evaluation of
Residual Stresses after Cryogenic Cooling, the
temperature drop within the cryogenic range is
expected to reduce the residual stresses. This effect has been emphasized in the research performed on high-speed steel SEW 320 S 3.3.2
(Fe-3Mo-2V-0.98C, wt%) (Ref 18), heat treated
according to the variants A to G shown in Fig.
2. Figure 11 shows the variation of second-order
residual stresses as a function of the cooling temperature in the case of S 3.3.2 high-speed steel.
It is obvious that second-order residual stresses
decrease from about 300 to almost 100 MPa,
while the temperature drops deeper within the
cryogenic field. Therefore, cryogenic cooling reduces these residual stresses to less than one-half
of their initial value. In fact, cryogenic cooling
has the same effect on the level of residual
stresses as does tempering when applied after
classical quenching.
The number of tempering treatments significantly reduces the magnitude of residual stresses
(compare, for instance, the variant B to the variants E or A), and if the effect of cryogenic cooling is also added, then the steel is almost completely stress relieved. The final stress value of
100 MPa is close to the usual value of the elastic
stresses within the equilibrium crystalline lattice.
The experimental values of the residual
stresses, evaluated in the steel that has been cryogenically cooled at 153 and 77 K, are not significantly different; therefore a cooling performed at a temperature close to the industrial
heat treatment temperature (203 K) appears to be
sufficient. Thus, after a treatment performed at
203 K the stress decreased from 1.33 (in the
case of variant D1) to 1.62 (variant G1). These
decreases range from 1.40 (variant D2) to
1.82 (variant G2) after a cooling performed at
153 K and from 2.03 (variant D3) to 2.47
(variant G3) after cooling to 77 K. It is obvious
that the heat treatment G variants, due to their
(Eq 17)
where M and M from Eq 16 are thermal coefficients previously determined by means of the
evaluation of Debye temperature.
A cryogenically cooled steel would certainly
contain third- and second-order residual stresses,
and first-order stresses might also exist. In this
case, for the evaluation of third-order residual
stresses, a rather large number of diffraction
lines will be chosen, corresponding to the values
of the sin2 h/k2 parameter that are less than 0.36
2 in such a way that second-order residual
A
stresses are neglected.
Influence of Cryogenic Cooling on

Residual Stresses and Dimensional
Stability of Steels
As emphasized in the previous section of this
article, heat and/or mechanical treatments cause
anomalies within the fine structure of metallic
materials that are diffractometrically noticeable
by both an enlargement and a displacement of
the maxima of the diffraction lines. These
anomalies, which are directly correlated with
the substructure elements, are usually located either on the crystallite level or within the slip
planes that occur in the crystallite structure.
However, not all of the substructure elements are
pragmatically interesting from both technical
and scientific points of view. In practice, mostly
the density of dislocations, the size of the mosaic
groups, and the second-order strains (stresses)
are of particular interest.
Influence of Cryogenic Cooling on the

Residual Stresses
The quantitative evaluation of second-order
residual stresses is based on the enlargement ef-
Second-order stress, MPa
350
A
C1
E
250
C2
F1
B
F3
D2
150
G1
50
313
C3
F2
D1
D3
G2
G3
273
233
193
153
113
73
Temperature, K
Fig. 11
Dynamics of second-order residual stresses as a function of the cooling temperature, for the heat treatment
experimental variants from Fig. 2 applied to SEW 320 S 3.3.2 steel. Source: Ref 20
multiple tempering, have always caused a more

significant decrease with 18 to 23% of the stress
state, as compared to the D variants.
Since the stress-temperature dependence
shown in Fig. 11 is linear, the crystalline lattice
of the cryogenically treated steel undergoes elastic relaxation with the decrease in cooling temperature. In fact, it appears that the actual values
of the second-order residual stresses are a little
higher at 77 K than at the ambient temperature
(where the measurements have been performed
because in situ measurements were impossible)
because during the heating back up to the ambient temperature the martensite amount slightly
increases, as shown in Fig. 7.
Influence of Cryogenic Cooling on the

Dimensional Stability of Steels
The achievement and use of both high-precision mechanical equipment and tools for measuring and verifying require obtaining parts and
tools with geometries that have extremely accurate tolerance and that must be preserved unchanged in time. Therefore, both parts and tools
require dimensional and shape stability; that is,
their structure must be as stable and as stress
relieved as possible.
The most common heat treatment applied to
the majority of parts and tools consists of
quenching and one or several subsequent tempering treatments. The resulting structures contain tempered martensite and, according to the
tempering conditions, a lower or higher amount
of retained austenite. However, both these
phases are metastable and, when held within a
thermal range close to ambient temperature, they
slowly evolve toward more stable conditions.
This natural evolution is accompanied by contraction and expansions that are the main causes
of dimensional and shape variations.
Cryogenic cooling affects dimensional stability by acting on those deformation sources that
have the largest contribution on the dimensional
variations, namely the continuation of the transformation of the retained austenite to martensite.
When compared with classical quenching, performed to ambient temperature, cryogenic cooling has more effect on the steels with larger
amounts of carbon or alloying elements. Within
the negative temperature range that represents
only 17 to 18 of the total temperature drop during
ordinary quenching, just a small amount of austenite transforms to martensite. The largest
amount of martensite was formed after ordinary
quenching.
After extensive research, G. Murry (Ref 21)
introduced several heat treatment schemes in order to ensure a fair dimensional stability, according to two steel categories.
Steels for which Retained Austenite Decomposes at 473 to 673 K. In this case, the
retained austenite can be reduced by cryogenic
cooling, by tempering at 523 K, or by a combi-

nation of these treatments. The following procedures can be applied:
Procedure A: Austenitizing and quenching to
ambient temperature, immediate cooling to
163 K, maintaining this temperature, and then
slow heating back to room temperature followed by a stress-relieving tempering at 423
to 473 K
Procedure B: Austenitizing and quenching to
163 K, maintaining this temperature, slow
heating back to room temperature, and tempering at 523 to 573 K, which achieves the
total decomposition of retained austenite.
Procedure C: Austenitizing and quenching to
ambient temperature and tempering at 523 to
573 K with the same effects as above
Procedure A obtains the maximum values for
hardness and wear resistance in steels, because
the very low retained austenite amount, which
has been observed within the microstructure, can
be neglected as long as the treated parts work at
ambient temperature or at temperatures lower
than 473 to 523 K. Procedures B and C cause
the complete disappearance of both retained aus-
tenite and residual stresses. Consequently, a high

dimensional stability is achieved; however, both
hardness and wear resistance are reduced. It
should be noted that Procedure C is more economical and productive compared with the other
two.
Steels for which Retained Austenite Decomposes at 823 to 973 K. This category includes alloy steels with elements that form stable
carbides such as tungsten, molybdenum, and vanadium. They belong to the class of high-speed
steels for which high amounts of retained austenite (1860% Aret) are obtained after classical
quenching. It should be noted that this retained
austenite is highly stable from a thermal point of
view. The cryogenic treatment applied to this
category of steels has a high efficiency. It is usually accompanied by one or two tempers meant
to achieve both the prescribed hardness and the
improvement of steels properties. The following
procedures are applicable:
Procedure A: Austenitizing and quenching to
heating back to room temperature followed
by tempering at 823 to 848 K, cooling to ambient temperature, immediate cryogenic cooling to 163 K, maintaining this temperature,
and subsequent slow heating back to room
temperature with final tempering to 823 K
Procedure B: Austenitizing and quenching to
heating back to room temperature, tempering
to 873 to 923 K, cooling to room temperature,
cryogenic cooling to 163 K, maintaining this
temperature, and subsequent slow heating
back to room temperature with a final tempering to 823 K
Procedure C: Austenitizing and quenching to
ambient temperature, tempering to 873 to 923
K, cooling to room temperature followed by
immediate cooling to 163 K, maintaining this
Expannsion l/l, m/mm
+20
A
+10
1%
1% C1.5%
C-1.5% Cr
Cr steel
steel
10
B
20
400
800
1260
1600
2000
Time, h
Dimensional variation of an alloy steel, with
1% C and 1.5% Cr, during a long period at 348
K, of a classically (A) and a cryogenically (B) heat treated
specimen. (The Mikus diagram). Source: Ref 21
Fig. 12
Table 2 Applied heat treatment variants and the resulting Vickers hardness, in the case of
several low-alloy steels
Steel
Heat treatment variant
Austenitizing
Quenching
Cryogenic cooling
Tempering
Vickers hardness
A
B
C
A
B
C
D
1098 K for 30 min

1098 K for 30 min
1098 K for 30 min
1123 K for 15 min
1123 K for 15 min
1123 K for 15 min
1123 K for 15 min
Oil
Oil
Oil
Oil
Oil
Oil
Oil
Yes
Yes
No
Yes
Yes
No
No
448 K for 1 h
523 K for 1 h
523 K for 1 h
448 K for 1 h
523 K for 1 h
523 K for 1.5 h
448 K for 1 h
861
712
969
826
715
696
802
02 SAE
L2 AISI
Source: Ref 21
Table 3 Applied heat treatment, with the aim of improving dimensional stability, in the
case of several supercarburized high-speed steels
Austenitizing
1523 K for 10 min

1523 K for 10 min
1523 K for 10 min
1523 K for 10 min
Source: Ref 21
Quenching
Cryogenic cooling
Tempering, K
Cryogenic cooling
Tempering, K
Vickers hardness
Oil
Oil
Oil
Oil
Yes
Yes
No
No
823
873
873
873
Yes
Yes
Yes
No
823
823
823
823
888
897
914
905
temperature, slow heating back to room temperature, and tempering to 823 K

Procedure D: Austenitizing and quenching to
ambient temperature, tempering at 873 to 923
K, and subsequent tempering to 823 K
Procedure A can leave a very small amount of
retained austenite that does not exist during Procedures B, C, and D due to their high tempering
temperatures. These tempers are performed at
873 to 923 K, which is the maximum temperature range applicable when fair hardness values
are required. The service temperature for this
category of steels can reach 673 to 723 K without
causing dimensional stability problems, providing two successive tempers have been applied.
The cryogenic treatment followed by tempering
represents an additional guarantee for obtaining
high dimensional stability.
In order to prove the effectiveness of cryogenic treatment on dimensional and shape stability, Murry (Ref 22) analyzed several steels to
which both ordinary and cryogenic treatments
were applied. The applied treatments and the resulting Vickers hardnesses are reviewed in Table
2 for low-alloy steels and in Table 3 for supercarburized high-speed steels. After a dilatometric study performed on an AISI L2 low-alloy
steel part heat treated according to the variants
listed in Table 2, the following conclusions can
be reached:
Dimensional stability increases with increasing tempering temperature applied after
quenching.
The large amount of retained austenite, detected after the classical treatment D, significantly decreases after the cryogenic treatment
A; however, it does not disappear completely
unless tempering is applied at 523 K as in
variants B and C. Thus, the above heat treatments can be classified according to their influence on the dimensional stability: D, the
lowest stability; A and C, average stability;
B, the highest stability.
It should be noted that treatment B induces a
marked increase of both hardness and wear
resistance over treatment C. The same aspects
are also revealed in the case of the supercarburized high-speed steels under study.
In general, the results offered by Murry have
proved that by using cryogenic cooling it is possible to obtain a high dimensional stability with
out decreasing the hardness and wear resistance of steels.
Figure 12 shows the dimensional evaluation,
during a long period at 348 K, of two specimens
made from alloy steel (1% C, 1.5% Cr)one
classically treated (A) and the other cryogenically treated (B). In this way, the transformations
occurring in both austenite and martensite can
be analyzed and understood, especially during
the service regime of the parts.
Specimen A, subjected to a classical treatment, continuously expands in time due to the
presence of a high-retained austenite amount at
ambient temperature. This expansion is caused
by the decomposition of the retained austenite

that is accompanied by an important volume increase. Specimen B underwent a cryogenic heat
treatment immediately applied after quenching
and consequently contains an extremely low
amount of retained austenite. During the long period at 348 K, the quenching martensite undergoes the first stage of tempering, namely the decrease of the tetragonality degree that causes a
continuous contraction. This contraction continues to occur after 10 tempers at 348 K, since
there is still tetragonal martensite that did not
transform to cubic martensite. However, if specimen b was subjected to the same tempering (10
h at 348 K) as specimen a, tempered martensite
Hardness, HV5
1000
900
860
820
780
200
400
600
800
1000
Relative shrinkage, m/m
Fig. 13
would occur and any contraction would no

longer be observed.
In conclusion, two phenomena were found responsible for the variation of dimensional stability of quenched steel parts: the transformation
to martensite of the retained austenite left after
quenching, which is accompanied by an increase
of volume, and the evolution of metastable tetragonal quenching martensite toward more stable cubic tempered martensite, which is accompanied by a contraction. In order to avoid both
the subsequent expansions and contractions, the
most effective procedure would be to eliminate
both the retained austenite by means of cryogenic cooling and the tetragonal martensite, respectively, using low tempering. However, particular care is to be taken when choosing both
the temperature and the maintenance period for
tempering since the contraction, occurring during the formation of tempered martensite in either oil or cryogenically quenched steels, is always accompanied by a sensible decrease of
hardness. Such an effect is shown in Fig. 13,
which shows the Vickers hardness decreasing as
the relative shrinkage increases during the tempering heat treatment applied to a full martensitic
steel part.
Table 4 lists several variants of heat treatment,
including cryogenic cooling, applied for the di-
mensional stabilization over time of AISI D2

steel.
A high-speed steel, grade AISI M1, was either
classically or cryogenically heat treated (Ref 10)
according to the same variants shown in Fig. 2
for high-speed steels (SEW 320) S 3.3.2. The
dimensional variation of heat treated AISI M1
steel, after five different storage periods (i.e., 10,
20, 50, 80, and 110 days) proved to be proportional to the retained austenite amount (Aret) and
therefore also to the residual stress decrease, as
listed in Table 5.
The dimensional variations as a function of
the storage periods described for AISI M1 steel
heat treated according to several variants (illustrated in Fig. 2) are shown in Fig. 14. After about
three months of storage, the relative dimensional
variations of the cryogenically treated specimens
are extremely lowless than 0.1 lm/mm, in
particular, 0.074 lm/mm for variant D3 and
0.085 lm/mm in the case of variant G1, as
shown in Fig. 14.
The usual procedure used for analyzing the
dimensional stability of different heat treated
steel grades is based on the so-called French
specimens. The typical geometry of a French
specimen is shown in Fig. 15. In order to point
out the superiority of the heat treatments applied
below room temperature compared with classi-
Dependence of hardness on the relative shrinkage observed during tempering. Source: Ref 5
B
M1
Expansion (l / l ), m/mm
Table 4 Heat treatment variants applied for

the dimensional stabilization in time of AISI
D2 steel
Dimensional
variation after
3 months
1. Austenitizing 1283 K, water

quenching
2. Cooling at 77 K
3. Stress relieving for 1 h at 593 K,
oil quenching (63 HRC)
0.012 lm/mm
1. Austenitizing 1283 K, water

quenching
2. Cooling at 77 K
oil quenching (63 HRC)
4. Resuming three times operations 2
and 3
air cooling (59.5 HRC)
0.002 lm/mm
5
4
3
D1
2
1
0
1
10
20
50
Time, days
Dimensional variation in time, of AISI M1 steel, heat treated according to several variants. Source: Ref 23
Table 5 Variation of the h typical dimension of French specimens, made from AISI M1 heat
treated steel
Heat treatment
variant (Fig. 2)
B
D1
D2
D3
G1
G2
G3
D3
2
3
Fig. 14
80
D2
G1
110
G2
G3
Retained austenite
(Aret), %
(Dh/h)0(a),
lm/mm
(Dh/h)1(a),
lm/mm
(Dh/h)2(a),
lm/mm
(Dh/h)3(a),
lm/mm
(Dh/h)4(a),
lm/mm
(Dh/h)5(a),
lm/mm
51 (Dh/h)i(b),
lm/mm
14.6
7.5
6.8
3.9
5.1
4.6
3.82
52
77
81
90
81
109
120
3.9
1.9
1.05
0.27
0.36
0.25
1.7
1.1
0.54
0.25
0
0.26
0.18
0
0.76
0.28
0.28
0.25
0.12
0.2
0.18
0.52
0
0.16
0.11
0.1
0
0.08
0.38
0.16
0.1
0.074
0.085
0.16
0.082
6.66
2.56
0.836
0.164
0.555
0.75
1.878
(a) The relative expansions are measured: during the heat treatment [(Dh/h)0]; after 10 days [(Dh/h)1]; after 20 days [(Dh/h)2]; after 50 days [(Dh/
h)3]; after 80 days [(Dh/h)4]; and after 110 days [(Dh/h)5]. (b) Cumulative relative expansion, 110 days after the heat treatment
6 mm thick
Schematic of a French specimen, with standard
dimensions: D 50 mm, d 25 mm, h 6
mm. Source: Ref 5
Fig. 15
Temperature, K
cal treatments, the French specimens made from

UNS K19667 ball-bearing steel (Fe-1.1Mn0.55Si-1.6Cr-0.98C) were heat treated according
to the specific variants illustrated in Fig. 16.
The typical dimensions of the heat treated
French specimensnamely, D, d, and hhave
been measured after 20 days and after 30 days
of storage at room temperature as well as after a
short exposure at 373 K (in boiling water). The
results have been designated by the subscripts 2,
3, and 4, respectively, while subscript 1 corresponds to the dimensions of heat treated specimens. Therefore, the dimensional differences
have been designated as D21 (D2 D1); (d2
d1); (h2 h1), D3 1 (D3 D1); (d3 d1); (h3 h1),
D41 (D4 D1); (d4 d1); (h4 h1) and are illustrated in Fig. 17(a), (b), and (c), corresponding to the 20 days storage, the 30 days storage,
and the 373 K exposure, respectively.
The data in Fig. 17 show that the narrowest
spreading of the dimensional differences occurs
in the case of specific heat treatment variant V,
which comprises cooling to 203 K followed by
tempering to 503 K. The largest variations are
noticed in the case of typical dimension h and
occurred in the French specimens tempered at
373 and 423 K (heat treatment variants III and
IV, respectively) both after 20 days of storage
(Fig. 17a) and after 30 days of storage (Fig. 17b).
The other variants are less significant, with the
lowest values shown in Fig. 17(c), corresponding to a short 373 K heating.
In the case of the heat treatment variation II,
which includes no tempering after the cooling
performed at 203 K, it is noticeable that residual
stresses largely influence the dimensional stability due to the stress-relieving phenomenon. As
the tempering temperatureapplied after
quenching at 203 Kreaches the upper admissible limit, the dimensional variation increases
during heat treatment. However, the dimensional
stability improves in time due to the stress relieving that occurs in the ball-bearing heat
treated steel. From this viewpoint, the heat treatment variants designated by IV and V led to the
best results in the case of UNS K19667 ballbearing steel.
If these phenomena are considered with respect to residual stresses, which represent the
only driving force for direct martensitic transformation, since the transformation becomes
more and more difficult as the temperature decreases, it follows that the stresses also diminish
with decreasing cooling temperature. A slight in-
1073
873
673
473
293
Influence of Cryogenic Cooling on

the Structure and Substructure of
Steels
The structure, or in other words the phase
composition, represents an important feature of
the quality of metallic materials. The phase
amounts and mostly the phase ratio are responsible for the values of several important properties of metallic materials, such as hardness,
toughness, dimensional and thermal stability,
durability, fatigue life, wear resistance, and so
on. Thus, it is well known that retained austenite
deteriorates the service properties of the tools
and machine parts manufactured from quenched
steel, due to its specific characteristics and
mostly to its high instability.
As previously pointed out, applying one or
more tempers might not always be the most effective way to reduce the amount of retained austenite, because of the marked decrease of hardness, mechanical strength, and wear resistance
that partially cancels the positive effect of retained austenite transformation. For this reason,
a more effective method could be cryogenic
cooling. However, because of the rather low
amount of retained austenite (less than 15%) left
by classical quenching in the microstructure of
plain-carbon and alloy steels it appeared that cryogenic cooling would not cause marked additional microstructure improvements compared to
ordinary quenching. However, the structure of
cryogenically cooled metallic materials proved
to be much more uniform and dense. In addition,
cryogenic cooling has induced the occurrence of
very fine carbides, with dimensions less than 1
1123
473
373
203
Fig. 16
crease is noticed, however, during the heating

from cryogenic temperature back to ambient
temperature due to the small amount of retained
austenite that partially transforms to martensite.
These results and discussion have proved that
cryogenic cooling followed by low tempering (in
order to maintain a high hardness value) represents a safe and effective procedure for the increase of dimensional and shape stability of precision parts and tools. Since cryogenic cooling
achieves both an increase of quenched martensite and a considerable quantitative decrease of
residual stresses, all international research carried out to date has emphasized its opportunity
and advantages as an effective, practical heat
treatment.
II
203
III
Time, min
503
423
203
IV
203
V
Specific heat treatment variants applied to French specimens made from UNS K19667 steel. Source: Ref 5
lm, which occupy the microvoids and contribute

to the increase of both density and coherence
within the metallic material. The microcarbides
that have been identified and counted by quantitative electronic microscopy are mostly responsible for the increase of wear resistance as well
as other properties of cryogenically treated
steels.
Unlike surface treatments, the structural
change induced by cryogenic cooling is uniform
and takes place within the entire volume. Consequently, a cryogenically treated tool will undergo no sudden changes of structure until it is
completely worn out, no matter which mechanical manufacturing operations (such as finishing,
sharpening, and so on) it subsequently undergoes. Therefore, the cryogenic treatment causes
a permanent irreversible molecular change (Ref
24).
Two micrographs (Fig. 18) emphasize the microstructural effects caused by cryogenic cooling
on the phase structure of classically quenched
SEW 320 S 3.3.2 steel. Figure 18(a) is a micrograph of S 3.3.2 steel after a classical heat treatment (oil hardening followed by 2 to 3 tempers)
corresponding to the variant B in Fig. 2. It is a
usual microstructure comprising primary and
secondary carbides embedded into a matrix composed by tempered martensite and retained austenite representing 10% of the phase amount.
The microstructure in Fig. 18(b) is the same
steel, but it underwent a cryogenic cooling and
a subsequent stress-relieving tempering, in accordance with the variant D3, illustrated in Fig. 2.
Besides stress relieving, the tempering heat
treatment enables both microcarbide precipitation and alloying element respreading. For this
reason, the observed structure has become more
uniform, revealing a regular distribution of fine
microcarbides with sizes less than 1 lm within
a matrix formed mostly of tempered austenite,
since the retained austenite amount is about
1.1%, detectable only by x-ray diffraction. In addition, the lower both cooling and maintaining
temperature within the cryogenic range (20377
K), the more uniform and refined the structure,
compared to Fig. 18(a).
For those tools made of high-speed steel that
exhibit complex configurations as well as abrupt
passes of the cross section, cryogenic cooling
can be applied between two tempers, according
to the heat treatment variant designated by G in
Fig. 2. The final tempering has a special part
consisting on one hand of the distribution of alloying elements among matrix and carbides and
on the other hand of the carbide refinement that
mostly improves the service properties of highspeed steels.
By twice repeating the cryogenic cooling-tempering sequence, according to the variants designated by H and I in Fig. 2, as well as by increasing the number of tempers, according to the
variants K and M, the precipitated carbides during the first tempering would coagulate, thus
causing a structural nonuniformity, as depicted
in Fig. 19. In fact, among the heat treatment variants for high-speed steels shown in Fig. 2, the

30
30
30
25
25
25
20
20
20
15
15
15
10
3-1, m
2-1, m
10
0
5
10
15
20
25
30
II
III
373
IV
V
473
Temperature, K
Fig. 17
Fig. 18
0
10
15
15
20
20
II
III
373
IV
V
473
25
573
Temperature, K
10
30
30
II
III
373
IV
V
473
573
Temperature, K
Variation tendency of the differences of typical dimensions of French specimens as a function of the tempering within the specific heat treatment variants (Fig. 16) applied
to UNS K19667 steel. (a) After 20 days storage. (b) After 30 days storage. (c) After a short exposure at 373 K. Source: Ref 5
Microstructure of SEW 320 S 3.3.2 rapid steel. (a) Classically heat treated. (b) Cryogenically heat treated.
Micron bars 10 lm. Source: Ref 20
Microstructure of T1 heat treated high-speed

steel after a double cryogenic cooling tempering sequence according to the variant H from Fig. 2. Micron bars 10 lm. Source: Ref 25
Fig. 19
0
5
25
573
10
4-1, m
variants designated by D, and under certain circumstances also the variants G, proved to have
the most homogeneous and uniform microstructures.
Similar microstructural aspects have been observed after the cryogenic cooling of plastic deformation tools manufactured from chromium
steels, plain-carbon tool steels, carburized or carbonitrided steels, or from ball-bearing steels. For
these steels, only the variants designated by D,
I, and K in Fig. 2 are applicable, for which cryogenic cooling is a continuation of ordinary
quenching.
The influence of cryogenic cooling on the
amount of retained austenite, after several heat
treatment variants (Fig. 2) applied to some tool
steel grades, such as SEW 320 S 3.3.2, AISI T1,
M1, M2, D2 is described in Table 6. Tempering
at 813 to 853 K, after cryogenic cooling, causes
martensite stress relieving as well as a slight increase of the martensite to the detriment of aus-
tenite. Thus, the retained austenite amount can

reach 3.9% for M1 steel and 1.1% for S 3.3.2
steel.
As previously pointed out, for those tools and
machine parts that exhibit abrupt cross-section
passes, the most suitable proved to be the heat
treatment variant G, in order to avoid the occurrence of first-order stresses that could induce
cracking. However, in the case of S 3.3.2 and
M1 steels the variant G did not cause spectacular
decreases of the retained austenite, since this
heat treatment is especially meant to reduce the
general stress state. Besides martensite and retained austenite, the structure of high-alloy steels
also contains a carbide phase. The carbide precipitates are particularly influenced by cryogenic
cooling regarding their development, quantity,
dimensions, and distribution.
A diffractometric study (Ref 23) performed on
several high-speed steels as well as on the carbide powders that precipitate after the electrolytic decomposition of the metallic matrix (the
carbide extract) obtained after different cryogenic heat treatment variants reveals an increase
of the amount of carbide in alloying elements.
At the same time, it was noticed that the sum of
the carbide-forming elements that take part in
carbide formation decreases after cryogenic
cooling. This means that the martensitic matrix
grows rich in alloying elements after cryogenic
cooling, and this might explain the observed improvements of hardness and cutting performances at the tools made from cryogenically
cooled high-speed steels.
Chemical analysis of the carbide extracts, obtained from several heat treated high-speed
steels, revealed much different structures and redistribution of alloying elements among the matrix and the carbide phase, at the variants that
include cryogenic cooling compared with the
classic heat treatment variant. This fact is emphasized in Fig. 20, which shows the repartition
of the most important carbide-forming alloying
elements (Cr, Mo, W, and V) among the matrix

and the carbide phase as a function of the cryogenic cooling temperature in the case of M2
steel. The curves illustrated in Fig. 20 delineate
the two domains among which the carbide-forming alloying elements are distributed: the matrix
(the upper field) and the carbide phase (the lower
field, under the curves). From Fig. 20 it is obvious that the lower the cryogenic cooling temperature the richer the matrix, in carbide-forming
elements, and therefore the higher the hardness
and the cutting characteristics of high-speed
steel.
Other researchers (Ref 21, 23) have emphasized that the increase in the iron amount within
the carbides might be an argument for the formation of the (MFe)6C-type complex carbide
with high iron content and eventually for the de-
composition of (MFe)2C-type complex carbide

with low iron content (where M is metal). This
fact is expressed by the increase of the ratio of
those carbides that contain high iron amounts,
and this increase leads to a quantitative augmentation of the carbide phase within the steel.
The qualitative and quantitative determination
of carbides has been accomplished using diffractometric analyses performed on carbide powders, extracted by electrolytic decomposition of
the high-speed steel matrix.
The diagram illustrated in Fig. 21 presents a
decrease of the total chromium carbide amount,
as much as 30%. Chromium can form carbides
of the Cr23C6-type that are primary, coarse, with
cubic lattice and very stable. Their amount decreases with the cryogenic cooling temperature.
Table 6 Influence of the heat treatment variant on the retained austenite in several highalloy tool steels
Percentage of retained austenite (Aret), after:
Heat treatment
variant
Quenching
AISI T1 steel
B
G
H
I
K
M
AISI M2 steel
B
D
H
G
SEW 320 S 3.3.2 steel
B
D
G
AISI M1 steel
B
D
G
AISI D2 steel
E
D
I
Cryogenic
cooling to 203 K
First
tempering
Cryogenic cooling
to 203 K
Second
tempering
Cryogenic
cooling to 203 K
Third
tempering
10
1.65
3.6
27
27
27
27
27
27
12
12
24
23
24
9
9
24
19
18
7
16
9
10
0
5
16
21
21
21
21
7.5
17
2.25
17
17
13
13
11
3.64
3.6
29
29
29
8.23.3(a)
23
6.251.1(a)
23
8.33.5(a)
17
5.21.9(a)
10
20.1
20.1
20.1
7.25.4(a)
16.7
73.9(a)
16.7
8.54.8(a)
15
5.13.8(a)
14.6
3.9
3.3
10.1
10.1
00.1
6
6
7.2
4.2
4.2
(a) Values determined after cryogenic cooling to 77 K.
12.44
40
11
Carbides contents, %
Alloying elements sum, %
12
Matrix
10
9
8
7
6
273
Carbides
223 203 173 153 123

Temperature, K
30
Cr
W
Mo
20
V
77 73
Fig. 20 Dynamics of the repartition of the most important carbide-forming alloying elements among
matrix and carbides as a function of the cryogenic cooling
temperature, in the case of M2 steel, heat treated according
to the variants D (open circles) and G (X data points).
Source: Ref 20
10
273
223 203 173 153 123

Temperature, K
73
Variation tendency of the amount of chromium, tungsten, vanadium, and molybdenum

carbides as a function of the cryogenic cooling temperature, in the case of SEW 320 S 3.3.2 steel. Source: Ref 20
Fig. 21
The other carbides, such as trigonal Cr7C3 and

orthorhombic Cr3C2, are slightly increasing,
which compensates for the decrease of the
Cr23C6 amount. All chromium carbides have a
rather low iron content (Ref 20).
In the case of S 3.3.2 steel, the decrease in the
cryogenic cooling temperature also determines a
reduction in the total amount of vanadium carbides (VC and V2C), which usually undergo secondary precipitation due to the decrease in the
number of tempers. It can be seen in Fig. 21 that
the decrease of cryogenic cooling temperature
induces an increase of tungsten carbide. Tungsten carbide can be cubic WC, hexagonal W2C,
and complex cubic Fe3W3C. Disregarding the
evolution of the amount of WC and W2C carbides, cryogenic cooling determines an increase
in the amount of Fe3W3C carbide with high iron
content. With regard to hexagonal Mo2C carbide, Fig. 21 shows that after cryogenic cooling
at 77 K its amount becomes at least double compared with S 3.3.2 steel heat treated at room temperature.
Besides the increase in both martensite and
carbide, the improvement of cutting characteristics of the tools made from high-alloy steel,
after cryogenic cooling, can be also ascribed to
the changes produced in the size and distribution
of the carbide phase.
The values of the qualitative and quantitative
parameters of the carbide phase are listed in Table 7 for some high-alloy steels. The measurements were performed by means of a Quantimet
720 quantitative microscope equipped with a
computer interface. The determined parameters
are: A, the total surface occupied by the carbide
phase (%); I, the total sum of carbide dimensions
on surface unit (mm/mm2); Im, the average size
of carbides (mm); Nc, the total average number
(density) of carbides on square millimeter
(mm2), and the carbide density is distributed
according to the dimensional ranges: 0 to 0.5
lm, 0.5 to 1 lm, 1 to 2 lm, 2 to 4 lm, 4 to 8
lm, 8 to 16 lm, 16 to 32 lm, and above 32 lm.
Table 7 reveals the net superiority of cryogenic
heat treatment variants compared with the classical treatment (B), with respect to both the size
and distribution of carbides. Thus, the carbide
surface (A) is larger for the cryogenic variants
(except M1 steel for which three cryogenic variants, D2, G2, and G3 induce smaller surfaces)
compared with the classical heat treatment variant.
Considering that the average size of carbides
(Im) has the same or lower magnitude order and
that the total sum of carbides average dimensions
on surface unit (I) is always larger, it follows that
cryogenic cooling should cause an increase in
the average number of carbides observed on
square millimeter (Nc) compared with the classical heat treatment variant (B). This assumption
is confirmed in Fig. 22, particularly in the case
of smaller-dimensional carbides.
It is obvious from these results that cryogenic
cooling causes the occurrence of a large number
of small-size carbides, coherent with the matrix,
and that this could explain the improvement of

the cutting characteristics of the tools made from
high-alloy steel. In addition, due to cryogenic
cooling a refinement of the carbide phase occurs
by means of the increase of carbide density with
sizes less than 1 lm. For instance, this increase
reached 151% (for variant G1) in the case of S
3.3.2 steel and 146% (variant D2) in the case of
M1 steel.
The improvement of the quality of high-precision tools made from AISI D2 steel (for die
forging, thread rolling, and so on) can also probably be ascribed tobesides the increase of
small-size carbide densitythe precipitation of
fine g carbides instead of e carbides (Ref 26).

This fine carbide contributes to the increase of
hardness within the martensite matrix that finally
causes an improvement of both the durability
and wear resistance of the tool. Another essential
element for determining the quality of a steel is
the substructure that can be defined as a detail of
the structure, observable within the crystalline
grains.
The study, control, and evaluation by x-ray
diffraction of the substructure parameters (dislocation density, second-order stresses, average
size of the mosaic blocks, texture, short-range
0.14
30
D2
F3
D3
G3
D1
G1
20
15
0.12
F2
107, m
103 carbides, mm2
25
G2
B
10
0.10
C2
E
A
0.08
5
C1
D1
G1
0.06
313
0
0
0.5
2
4
8
16
Dimensional range, m
32
64
Variation tendency of the average number of

carbides on square millimeter (Nc) as a function of the dimensional range, in the case of M1 steel.
Source: Ref 20
Fig. 22
Table 7
C3
F1
273
D3
order, etc.) also enable, among other prognoses,

the optimization of heat treatment procedures.
For the evaluation of the substructure elements,
a special part is played by the real physical width
of the diffraction line (b). As pointed out by Eq
10, the real physical width b of the diffraction
line depends both on the mosaic blocks and on
the second-order residual stresses.
The boundaries of the mosaic blocks, with
the dimensions determined using Eq 14, are
formed by dislocation walls. At the beginning of
the transformations, these walls are grouped in
such a way as to create bidimensional dislocation
networks (monolayer). As the transformation or
the plastic deformation develops further, the
boundaries of the mosaic blocks acquire an aspect of dense network with a rather large thickness. In this way, the dislocation density increases up to 1012 cm2.
If both the real size erhkl of the mosaic blocks
and the second-order residual stresses are
known, the dislocation density can be determined according to the relationship introduced
by Williamson and Smallman (Ref 19):
D2
G3
G2
233
193
153
Temperature, K
113
73
Variation tendency of the average size of the

mosaic blocks as a function of the final cooling
temperature in the case of SEW 320 S 3.3.2 steel subjected
to several heat treatment variants. Souce: Ref 3
Fig. 23
q 3n/(erhkl)2
(Eq 18)
where n is the number of dislocations located on

the boundary of the mosaic blocks.
Considering a chaotic distribution of dislocations along the mosaic blocks boundary (n 1),
it follows that q 3/(erhkl)2. In order to emphasize the influence of the cryogenic cooling on
two substructure elements (namely, dislocation
Qualitative and quantitative parameters of the carbide phase in some high-alloy tool steels
Carbide number distributed according to the
dimensional range, mm2
AISI T1 steel
B
D
G
H
AISI M2 steel
B
D
G
H
SEW 320 S 3.3.2 steel
B
D1
D2
D3
G1
G2
G3
AISI M1 steel
B
D1
D2
D3
G1
G2
G3
AISI D2 steel
A
E
D
I
A, %
l, mm/mm2
lm, mm
Nc, mm2
00.5
0.51
12
24
4.48
8.10
7.52
9.39
55.12
71.26
74.48
83.35
0.0013
0.0010
0.0009
0.0010
49,215
72,520
83,619
85,601
8,003
12,032
17,100
16,822
29,004
41,639
47,036
46,719
6,609
13,358
14,280
15,945
4,418
4,806
4,717
5,361
2.9
5.14
4.90
5.19
30.16
35.59
46.28
60.86
0.0009
0.0007
0.0009
0.0011
31,358
40,898
56,466
83,530
15,136
19,372
29,322
58,101
8,274
11,556
13,548
11,545
5,707
6,858
9,692
10,247
2.35
3.07
3.08
2.59
4.41
3.35
2.98
20.13
26.87
29.35
21.97
39.02
27.07
23.07
0.0009
0.0009
0.0008
0.0009
0.0008
0.0009
0.0009
22,080
30,885
36,600
23,350
48,935
28,901
24,585
11,450
15,541
20,456
11,457
24,871
14,174
11,718
5,554
6,985
7,852
4,951
11,472
6,407
5,593
8.10
9.70
8.21
7.34
9.32
7.39
7.31
52.13
59.14
68.08
58.16
53.29
46.97
61.14
0.0022
0.0018
0.0012
0.0012
0.0013
0.0012
0.0013
23,797
32,773
55,071
47,068
42,032
38,506
45,448
13,014
19,870
28,537
25,042
18,032
13,860
21,776
6.02
4.88
5.72
5.97
11.45
10.25
9.98
11.41
0.0081
0.0072
0.0068
0.0068
1,405
1,424
1,470
1,678
58
66
72
81
816
1632
32
924
575
367
545
238
110
101
188
19
0
18
21
0
0
0
0
2,002
2,943
3,627
3,359
199
138
218
238
40
31
59
13
0
0
0
0
0
0
0
0
3,387
5,465
5,664
4,544
8,370
5,122
4,401
1,253
2,177
2,038
1,852
3,298
2,395
2,220
407
692
575
935
860
746
632
29
25
14
11
64
57
21
0
0
0
0
0
0
0
0
0
0
0
0
0
0
4,365
6,746
14,231
11,439
11,429
12,026
12,189
3,487
3,498
8,706
6,614
6,992
7,049
8,006
1,499
1,335
2,381
2,302
3,659
4,140
2,384
985
735
907
1,278
1,417
1,178
810
417
543
296
389
471
246
261
28
46
13
0
32
0
14
0
0
0
0
0
0
0
110
126
170
203
270
274
292
283
378
401
400
454
293
284
250
317
202
193
185
244
83
72
89
85
9
7
10
80
48

density and average size of the mosaic blocks)
two high-speed steels, S 3.3.2 and M1, were subjected to the heat treatment variants illustrated in
Fig. 2.
For S 3.3.2 steel, Fig. 23 shows that the average size of the mosaic blocks increases with
1016 , m2
G1
G2
D1
C1
A
E
G3
D2
D3
C2
F1
C3
F2
F3
1
313
273
233
193
153
Temperature, K
113
73
Variation tendency of the dislocation density

as a function of the final cooling temperature
in the case of SEW 320 S 3.3.2 steel subjected to the same
heat treatment variants as in Fig. 23. Source: Ref 3
Fig. 24
Expansion l /l, m/mm
12
10
D1
8
6
G1
D2
the decrease in the cryogenic cooling temperature, as in the case of the variants F, C, D, and
G. Figure 23 also shows that the largest values
of the average sizes of the mosaic blocks have
been obtained in the case of the variants E and
F, which include tempering heat treatments after
ordinary quenching. The increase in the number
of tempers causes the decrease in the mosaic
block size due to the martensite refinement. Actually, the blocks are broken and their crystalline
lattice becomes stress relieved, as illustrated in
Fig. 23 for the variants B, D, and G, causing the
smallest sizes of the mosaic blocks. The deep
cooling of steel enables the coagulation of the
mosaic blocks, which therefore increase in size.
The linear aspect of the variation shown in Fig.
23 proves that the coagulation phenomenon is of
elastic type.
The last aspect of the cryogenic cooling effects on the substructure of the S 3.3.2 heat
treated steel is emphasized in Fig. 24, which
shows the variation tendency of the dislocation
density as a function of the final cooling temperature. After the first temper, a decrease of the
dislocation density is noticed, and this decrease
becomes more important after the subsequent
tempering due to the breaking of the mosaic
blocks. After cryogenic cooling, a marked decrease is noticed in the dislocation density,
which proves that the crystalline lattice becomes
more relaxed and the stress field diminishes, thus
coming closer to the normal state. Since a linear
variation is also noticeable in this figure, it is
obvious that the dislocation density decrease can
be related to the elastic phenomena produced
within the crystalline lattice.
G2
G3
0
D3
2
0
100
120 140
160
180
200 220 240

Dynamics of the dimensional stability as a
function of the stress state, in the case of SEW
320 S 3.3.2 steel, heat treated according to the variants D
(solid line) and G (dashed line). Source: Ref 20
Fig. 25
PropertiesResidual Stresses
Structure and Substructure
Correlation for Cryogenically
Cooled Steels
The increase in the transformation degree of
retained austenite to martensite causes the de-
crease of second-order stresses that become favorably redistributed among the two phases.
Moreover, the structure becomes harder and
more uniform and finally durability increases.
The influence of cryogenic cooling on the formation, the amount, and the distribution of the
carbide phase is first revealed by the emergence
of the fine carbides, which usually have smaller
sizes compared with those occurring after classical heat treatments. The large number of extremely fine carbides, which has been experimentally determined after the cryogenic
treatment, represents an argument for the conclusions discussed previously. The increase in
both the surface occupied by the carbides and
the carbide density as well as the large number
of fine carbides measured after the cryogenic
cooling lead to the development of a hard, uniform, homogeneous structure that induces a
marked improvement of the properties of heat
treated steels.
As previously mentioned, cryogenic cooling
contributes as well to the decrease in the stress
state that largely assists the improvement in
properties. Thus, in the case of the tools and parts
made from S 3.3.2 steel, the decrease in the
stress state leads to a fair dimensional stability,
as shown in Fig. 25, where three heat treatment
variants (B, D, and G in Fig. 2) are considered.
Another effect of decreasing the stress state due
to the cryogenic cooling is shown in Fig. 26,
which shows the variation of the amount of retained austenite with the value of the secondorder stresses, for the same heat treatment variants. It is obvious that the lower the value of the
stress state the more reduced the amount of retained austenite.
It is well known that durability, which is the
basic criterion for the effectiveness assessment
of any heat treatment, is conditioned by the connection between the strength intrinsic characteristic of the material and the actual service conditions of the tools. Besides hardness and wear
resistance, the most important characteristic that
determines the durability is the structure quality.
100
G3
8
6
D1
G1
D2
G2
90
80
G2
D3
100
D3
D2
G3
G1
G2
D2
G1
70
60
D1
S 332 steel
D1
M1 steel
G3
0
0
D3
100
120
50
40
140
160
180
200
220
Variation tendency of the retained austenite

amount as a function of the second-order residual stresses, in the case of SEW 320 S 3.3.2 steel. Source:
Ref 20
G3
G2
D2
80
D1
70
G1
G1
D2
D3
G2
D1
M1 steel
60
50
SS3.3.2
332 steel
steel
60 80 100 120 140 160 180 200 220 240

Fig. 26
G
D3 3
90
Durability (T), min
B
Durability (T), min
Retained austenite, %
10
Fig. 27 Dependence of durability on the relaxation of

the elastic stress field in the case of two highspeed steels subjected to intensive service regimes (cutting
speed: 32.67 m/min; cutting advance: 0.25 mm/rotation;
cutting depth: 2.5 mm). Source: Ref 20
40
3
5
6
7
8
Alloying elements sum, %
10
Dependence of durability on the total amount

of alloying elements dissolved in martensite in
the case of two high-speed steels. Source: Ref 20
Fig. 28

Thus, the phase composition (namely, the retained austenite amount), the phase distribution,
the uniformity and refinement degrees of the
phases, the distribution of the alloying elements
among the martensite matrix and the carbides,
the repartition of the carbides on dimensional
ranges, the average size of the mosaic blocks,
the dislocation density, and the second-order
stress state represent the influencing elements of
the durability of cutting tools made from highalloy tool steels. For instance, a low stress state,
namely a relaxation of the elastic stress field, involves an increase in the martensitic transformation degree that causes an augmentation of the
durability of cutting tools made from several
high-speed steels, such as S 3.3.2 and M1, according to Fig. 27.
On the other hand, Fig. 28 shows the influence
of the matrix alloying degree on the durability
of cutting tools. The figure illustrates an increase
in durability of the tools made from two highspeed steels with the increase of the total amount
of the alloying elements dissolved in the martensite matrix.
In conclusion, the discussion in this article has
brought evidence that the residual stresses play
a very important part both during and after the
cryogenic cooling. These stresses constitute the
main cause for the continuation of the transformation of the retained austenite to martensite,
within the cryogenic field. Surprising but true is
the fact that after cryogenic cooling the value of
the residual stresses becomes even lower compared with classical quenching, either in water
or in oil, to the ambient temperature.
4.
5.
6.
7.
8.
9.
10.
11.
REFERENCES
1. I. Alexandru, Gh. Ailincai, and C. Baciu,
Influence de traitements thermique a basse
temperature sur la duree de vie des aciers a`
outils a` coupe rapide tre`s allies, Rev. Metall., Vol 6, 1990, p 383388
2. V. Bulancea, I. Alexandru, and D. Bulancea,
Carburic Phase in Cryogenically Treated
Tools Steels, MATEHN 94, First Int. Conf.
Materials and Manufacturing Technology,
1820 May 1994 (Cluj-Napoca, Romania),
p 377380
3. V. Bulancea, I. Alexandru, and D. Bulancea,
The Substructure Elements Modification by
12.
13.
14.
15.
16.
Cryogenically Treated Fe-3W-3Mo-2V-C

Highly Alloyed Steel Tools, Bull. Polytech.
Inst. Jassy, Romania, Vol XLII (No. XLVI),
1996, p 167174; ASM International Second Int. Congress Mater. Sci. Eng., (Jassy,
Romania), 1997
A.P. Guleaev, Heat Treatment of Steel,
MASHGIZ, Moscow, 1960, 438 pages
V. Bulancea, I. Alexandru, D. Bulancea, and
D.G. Galusca, Advanced Heat Treatment
Technologies. Cryogenic Cooling of Steels,
Cermi, Jassy, Romania, 1998, 196 pages
N. Skalli and J.F. Flavenot, Prise en compte
des contraintes residuelles dans un calcul
previsionnel de tenue en fatigue, CETIMInformations, No. 90, 1985, p 3547
V. Hauk, Micro Residual Stresses, Evaluation by X-Rays, Results, Assessment, Proc.
of the Ninth Int. Conf. on Experimental Mechanics, 2024 Aug 1990 (Copenhagen,
Denmark), Technical University of Denmark, Vol 3, p 12271236
K.E. Easterling and A.R. Tholen, The Nucleation of Martensite in Steel, Acta Metall.,
Vol 28 (No. 4), 1976, p 333342
I. Alexandru, Gh. Ailincai, C. Picos, A.
Dima, and V. Bulancea, Aspects of the Influence of Cryogenic Thermal Treatments
on the Redistribution of Alloying Constituents between the Matrix and Carbides, Bull.
Polytech. Inst. Jassy, Romania, Vol XXIX,
1983, p 712
G.V. Kurdjumov, Non-Diffusion (martensitic) Transformation in Alloys, J. Technol.
Phys., Vol XVIII (No. 8), 1948 (in Russian)
G.V. Kurdjumov and O.P. Maksimova, Low
Temperature Martensitic Transition, Metallurgical and Metal Physics Problems, Tome
2, Metalurgizdat, 1951 (in Russian)
G. Fisher and D. Turnbull, J. Appl. Phys.,
Vol 19, 1948
G. Fisher and D. Turnbull, Austenite-Martensite Transition Kinetics, J. Met., Vol 19,
1949
S. Crussard, Comptes Rendues de
lAcademie des Sciences, Vol 24, 1955
F.L. Lokshyne, Dynamical Theory of Martensitic Transformation, Bull. of Polytech. of
Novotcherkask, Vol 71 (No. 85), 1957 (in
Russian)
K.E. Easterling and A.R. Tholen, On the
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
Growth of Martensite in Steel, Acta Metall.,

G.B., Vol 28 (No. 9), 1980, p 12291234
I. Alexandru, V. Bulancea, D. Pop, F. Pop,
and A. Alexandru, Phase Analyses of Metallic Alloys, Cermi, Jassy, Romania, 1998,
233 pages
V. Bulancea, I. Alexandru, and D. Bulancea,
The Stress State Modification by Cryogenically Treated Fe-3W-3Mo-2V-C High Alloyed Steel Tools, Bull. Polytech. Inst.
Jassy, Romania, Vol XLII (No. XLVI),
1996) p 161166; ASM Int. Second Int.
Congress Materials Science and Engineering, 1997 (Jassy, Romania)
C. Gheorghies, The Fine Structure Control
with X Radiation, Tehnica, Bucharest, Romania, 1990, 272 pages
V. Bulancea, The Influence of the Cryogenic Treatments on the Aspects Connected
with the Substructure of the Highly Alloyed
Steels and the Functional Characteristics of
the Cutting Tools, Ph.D. thesis, Gh. Asachi Technical University of Jassy, Jassy,
Romania, 1992
G. Murry, La stabilite dimensionnelle des
aciers apres traitements thermique. Aspects
metallurgiques, Trait. Therm., France, No.
128, Oct 1978, p 4350
G. Murry, Decomposition of Retained Austenite, Trait. Therm., No. 81, France, 1974,
p 4042
V. Bulancea, I. Alexandru, and D. Bulancea,
The Dimensional Stability of the Frozen
HSS Cutting Tools, Bull. Polytech. Inst.
Jassy, Romania, Vol XL (No. XLIV), p
499502; ASM Int. First Int. Congress Materials Science and Engineering, 1517
Nov 1994 (Jassy, Romania)
P. Paulin, Frozen Gears, Gear Technol.,
MarchApril 1993, p 2629
I. Alexandru, Contributions on the Influence of Thermal Treatments at Temperatures below 273 K of High-Alloyed Steels
on their Cutting Properties, Ph.D. thesis,
Gh. Asachi Polytech. Institute of Jassy,
Jassy, Romania, 1980
F. Meng, K. Tagashira, R. Azuma, and H.
Sohma, Role of Eta-Carbide Precipitations
in the Wear Resistance Improvements of Fe12Cr-Mo-V-1.4C Tool Steel by Cryogenic
Treatment, ISIJ Int., Vol 34 (No. 2), 1994,
p 205210

DOI: 10.1361/hrsd2002p345

Inducing Compressive Stresses through

Controlled Shot Peening
J. Kritzler and W. Wubbenhorst, Metal Improvement Company, Inc., Germany
WITH MOST VEHICLESwhether for land

or air transportthe level of stress increases
during the production cycle. The increased starting weights of planes, the higher load on the axles of trucks, and the mounting of stronger and
heavier engines in cars can be named as examples. In other cases, extreme forms of reduced
weight construction must be attempted, for example, in aircraft construction or when unexpected damage occurs during trials. If the increased stress, as in the case of most
components, is oscillating, the fatigue resistance
must be increased. This is possibleapart from
the trivial solution of increasing cross sections
with their disadvantage of increasing weight
by measures relating to the materials, to the construction, or to the manufacturing process.
The use of a so-called better material does
less than generally expected, especially if one
considers that increases of the tensile strength of
the material is subject to strict limits due to
manufacturing as well as cost considerations.
An improvement of the construction in detail
is in many cases the most elegant solution; however, especially in mass production such as car
manufacture or high-cost light engineering, for
example, aircraft construction, such possibilities
are often already exhausted. Notches, and sharp
notches at that, are frequently needed for functional reasons if one considers the threading of
bolts or the radii of slots for fittings keys.
In such cases and lately in increasing volumes,
shot peening is used (Fig. 1). Apart from im-
Fig. 1
Shot peening process. Courtesy of Metal Improvement Company, Inc.
proving fatigue strength, shot peening is also

used:
In cases of corrosive fatigue, stress corrosion,
hydrogen embrittlement, friction corrosion,

and wear by cavitation
In surface texturing
To close off porosity
To test adherence of layers as well as peen
forming
Historic Development of Shot

Peening
Sandblasting as an example for cleaning castings can be regarded as a predecessor of shot
peening. Literature on this dates from the 1800s
(Ref 1, 2). Shot peening was introduced by Zimmerli (Ref 3, 4) and Almen (Ref 59) in their
respective companies Associated Spring Company and General Motors prior to 1930. In 1934,
patents were granted for the shot peening machines (Ref 10), and during the years 1935 to
1945 Almen carried out the fundamental work
(Ref 1114), for example, development of the
Almen strip.
Shot peening was also investigated in Germany prior to World War II at the Technical Universities of Darmstadt and Braunschweig (Ref
1518). Opel used shot peened valve springs
from 1935. In about 1940 it was clear that shot
peening increases the fatigue strength primarily
by improved residual compressive stress and not
by increasing the hardness of the material surface.
World War II precipitated the industrial application on a large scale in the United States
(Ref 1921). An American Society for Metals
conference about peening was held in 1944 (Ref
22, 23). One gets the impression that in contrast,
the industrial application in Germany was lagging behind. All the same, a report (Ref 24) appeared about the optimizing peening depending
on the type of machine, impact time, and shot
size.
In 1946, the measurement of the residual compressive stress of peened torsion springs was de-
scribed for the first time (Ref 25). In 1949, a

report appeared on stress peening (Ref 26) in
which leaf springs were peened under prestress
to achieve a particularly high level of fatigue
strength. The Society for Automotive Engineers
(SAE) published its Manual on Shot Peening in
1952 (Ref 27). During the years 1951 to 1954,
there were two conferences in the Soviet Union
in which broad attention was dedicated to shot
peening (Ref 2838).
Since 1954 (Ref 39) peen forming has been
used for the manufacture of complex curved surfaces. The first norm, the MIL specification
MIL-S-13165 (Ref 40), came into existence in
1956 and still applies today in a developed form
(Ref 41, 42). Soon after, practically all aircraft
(Ref 4348) and automobile companies, and so
forth, joined with their own regulations and
norms. The norms named in the references represent merely a selection of the total available
(Ref 4958).
A general industrial application of shot peening in the Federal Republic of Germany can be
recorded from approximately 1955, and in this
context especially the automobile industry
should be named in which all springs (valve and
suspension) are peened since this time. A report
by W. Schutz (Ref 59), stemming from the same
period, reports about shot peening for improving
the vibratory fatigue properties of components
for tracked and road vehicles. The same author
later covered the special requirements of drive
technology (Ref 60).
Elementary Shot Peening Processes

The elementary procedure in shot peening is
an elastic-plastic impact process between the
peening material and the workpiece. In this process, a transformation of the kinetic energy to
work aimed at the elastic and plastic deformation
of the work material depending on the mass of
the peening material and the shot velocity takes
place. The residual compressive stresses in the
wake of shot peening are a result of the locally
occurring plastic deformation on the surface of
the workpiece and are caused by the impact of

pact of the peening material passes, this leads to
residual compressive stresses at the surface and
internal tensile stresses.
Hertzs Compression. In the process of the
Hertzs compression, the maximal plastic deformations occur below the place of impact of the
spherical peening material and the flat workpiece
(Fig. 4b). Here, a residual compressive state is
formed that has its maximal values below the
workpiece surface. Both elementary processes
occur always, as already stated, side by side and
are influenced by both the workpiece material
properties and the parameter of shot peening.
The characterization and the key data of a
compressive stress process caused by shot peening can be seen in Fig. 5 (taken from Ref 63).
Detailed examples relating to this are shown in
Ref 64.
the individual peening material particles. The

distribution of the residual compressive stresses
varies widely due to the variety of the working
and peening materials and possesses significant
differences (see Fig. 2, 3).
The deformation processes or the generation
of residual compressive stresses may be fundamentally explained according to Wohlfahrt (Ref
61, 62) in terms of two elementary processes:
plastic stretching of the immediate workpiece
surface and Hertzs compression.
Plastic Stretching of the Immediate Workpiece Surface. In Fig. 4(a) the stretching process of the area near the surface and its plastification is recognizable, declining with increasing
distance to the edge layer. The only plastically
deformed edge layer is impaired by the only
elastically deformed layers below. After the im-
AL 2014 T6
(Z 400/0.020"A/100%)
AZ 31
(MI 330/0.016"A/100%)
Workpiece Material and Process

Parameters
100
Workpiece material and process parameters

exert different effects on the residual compressive stresses generated by shot peening (Ref 65).
The key parameters are shown in Fig. 7.
Changing the parameters shifts the distribution of the generated residual compressive
stresses into the respective direction shown. It is
important to note that nearly all changes of the
parameters lead on one hand to changes in the
residual compressive stresses on the surface and
on the other hand influence to a much greater
extent the depth of the maximal residual stress
below the surface.
The hardness of the shot peening material influences the extent and the depth of the compressive stresses. For this reason, the peening
SF-Cu F20
(Z850/0.012"A/200%)
(Z8500/0.012"A/100%)
200
300
400
50 CrV 4
(MI 780/0.010"C/100%)
500
600
700
0.0
0.2
0.1
0.3
0.4
0.5
0.6
Depth, mm
Fig. 2
The advantages of the residual stress caused

by shot peening in a workpiece are shown in
Fig. 6.
The fatigue failure in most components is
caused by tensile stresses on the surface. These
tensile stresses probably arise when the components are newly installed. As each bending cycle
or rotation puts the workload on the component
and removes it, the component becomes increasingly susceptible to fatigue cracking until finally
it cracks or a crack from stress corrosion becomes manifest, and the component fails. Including shot peening in the original design of the
critical components prolongs its life span under
fatigue.
Figure 6(a) shows a component that has not
been shot peened and has tensile stresses on the
surface. The second component (Fig. 6b) has
been shot peened and has compressive stresses
on the surface. Figure 6(c) is a theoretical diagram of the shot peened component under load.
Typical stress profiles of soft metallic material. Courtesy of Metal Improvement Company, Inc.
15 CrNi 6
(MI 230H/0.014"A/200%)
r(z)
500
x(z) = y(z)
max
r(z)
z(z)
ES(z)
ES(z)
1000
18 CrNiMo 5
(MI 230H/0.020"A/200%)
1500
0.00
0.05
0.10
0.15
0.20
0.25
(a)
Depth, mm
Fig. 3
Typical stress profiles of hard metallic material.

Courtesy of Metal Improvement Company, Inc.
Fig. 4
z
(b)
Residual stress formation. (a) Stretching surface layer. (b) Hertzian pressure. Courtesy of Metal Improvement
Company, Inc.
Inducing Compressive Stresses through Controlled Shot Peening / 347
SS
TS(MAX)
100
0
2
CS(MAX)
4
6
8
12
Example of residual stress profile created by shot

peening. Ss, surface stress, which is the stress
measured at the surface. CS(max), maximum compressive
stress, which is the maximum value of the compressive
stress is induced, which normally is highest just below the
surface. d, depth of the compressive stress is the point at
which the compressive stress crosses over the neutral axis
and becomes tensile. TS(max), maximum tensile stress, which
is the maximum value of the tensile stress induced. The
offsetting tensile stress in the core of the material balances
the surface layer of compressive stress so that the part remains in equilibrium; TS(max) must not be allowed to become
large enough to create early internal failures. Courtesy of
Fig. 5
Tension
Compression
Tension
Compression
HS
HM
Tension
Compression
stress
Residual
stress
v, d, HS, HM
Shot peening parameters versus residual stress

distribution. v, velocity of shot, m/s; d, shot size,
mm; C, coverage, %; HS, hardness of shot, HRC, Hm, hardness of material, HRC. Courtesy of Metal Improvement
Company, Inc.
Fig. 7
Depth, in.
0
0.004
0.008
0.012
0
50
46 HRC shot
500
100
1000
61 HRC shot
0
Tension
stress
Compression
stress
Tension
stress
v , d , C, HS
HM
0.1
0.2
Depth, mm
150
Residual stress, ksi
50
0.3
Peening 1045 steel at 48 HRC with 330 shot.

Fig. 8
Peened surface
Surface metal
(a) Unloaded beam
Depth, in.
(b) Shot peened beam

0
0
Compression
Residual
stress
M
Applied
load
Resultant
stress
Peened surface
(c) Shot peened beam under load
Tension
0.008
0.012
0
46 HRC shot
500
150
61 HRC shot
1500
Fig. 6
Fig. 9
0
50
100
1000
200
250
0
Comparison of stress distributions. (a) Typical stress distribution in surface of metal beam, unloaded by exhibiting
residual tensile stress from normal fabricating procedures. (b) Same beam after shot peening, still without external load. Surface stress is now compressive. (c) Beam, when subjected to design loading, still shows some residual
compressive stress at surface. Courtesy of Metal Improvement Company, Inc.
0.004
0.1
+50
+100
()
Compression
Depth below surface, %
(+)
Tension
terial (see Fig. 10). Coverage at less than 100%

is only admissible in the case of peen forming.
Despite the fact that compressive stress layers
are formedin an incomplete and uneven distributionwith shot peening at less than 100%
coverage the danger of a potential crack in an
unpeened area remains. Especially in conjunction with high-tensile gear-wheel materials, it is
Compression
Ultimate tensile strength, %
shown in Fig. 8 and 9 confirm the advantages of

using hard peening material on high-tensile steel.
Similar reports about the results of shot peening
using specially hard peening materials are contained in Ref 66 and 67.
A change in shot size influences the distribution of residual compressive stresses and the
roughness.
In the case of high-tensile working materials,
an increase in the shot size leads to a substantial
increase in the size of the compressive layer. The
surface tension remains nearly identical; however, the maximum residual compressive stress,
as well as the positioning of the maximum in
depth, increases.
In soft working materials, such as aluminum, the depth of the compressive layer and the
roughness are influenced. A larger shot size always leavesgiven the same intensitya
smoother surface. Reference 68 contains a report
about peening with two differently sized peening
materials, known as dual shot peening.
The shot velocity is probably the most important parameter, as it can be easily varied by the
air pressure and the revolutions per minute of
peening wheel. As is the case with the shot size,
the depth of the maximum compressive layer is
primarily influenced and the surface tension is
influenced to a lesser extent. Extremely high shot
velocities may, however, lead to overpeening
effects, that is, damage to the surfaces. Such
effects coincide with reduced surface stresses.
Depending on the material hardness and consolidation condition of the workpiece, such effects
may lead, for example, to surface ruptures.
A proper shot peening process must be carried
out in such a manner that the processed surface
displays a coverage of at least 100%. The original state after processing should no longer be
visible. The entire shot peened area must display
spherical indentations caused by the peening ma-
material should be at least as hard or harder than

the working material subjected to the process,
unless the surface gradient is a decisive factor.
For a large number of ferrous and nonferrous
working materials, this can be achieved with a
peening material of the normal hardness of 45 to
52 HRC (Rockwell C).
The increasing use of high-tensile steel (50
HRC and above) demonstrates that the use of
especially hard peening material (5562 HRC)
is required. The compressive stress profiles
0.2
0.3
Depth, mm
Peening 1045 steel at 62 HRC with 330 shot.

Visual examination using a ten-power magnifying glass. Courtesy of Metal Improvement

Company, Inc.
Fig. 10
proved to be advantageous to apply degrees of

coverage at 200% and more, as it results in an
increase of the subsurface residual stress.
The shot impact angle and the resulting intensity or consolidation in the proximity of the edge
layer is an additional important parameter for
shot peening. The optimal peening angle for jet
shot peening emerged at 85. At this angle, the
maximum power transference occurs without the
following peening material being impeded by
ricocheting peening material. It can be seen in
Fig. 11 at 30 that peening tangentially reduces
the possible power transfer by 50%. The shot
impact angle becomes important in the peening
of boreholes and workpieces that have roundings
and notches of different sizes in need of processing.
0.010" A
0.0087" A
85
45
Fig. 11
Prestress level, MPa

400
Resulting
intensity
Impingement angle versus intensity. Courtesy

of Metal Improvement Company, Inc.
Nozzle
Workpiece
Tension
0.005" A
30
Impingment
angle
600
0.007" A
1350
700
200
0
0
60
200
Shot peened condition
400
600
Process Monitoring
Compression
800
Reciprocating
shot deflector
Stress peened condition
1000
M
1200
1400
0.2
0.4
0.6
0.8
1.0
1.2
Depth, mm
Fig. 13
(a)
Stress peening of 50 CrV 4. Courtesy of Metal

Improvement Company, Inc.
Shaded area
indicates
peened surface
Reciprocating
and rotating
peening lance
(a) Acceptable shapes
(b) Unacceptable "deformed" shapes
(b)
(b) Unacceptable "broken" shapes
Shot peening of bores. (a) Internal shot deflector. (b) Lance method. Courtesy of Metal Improvement Company, Inc.
Fig. 12
If the depth of a borehole exceeds the bore

diameter of the part, an external processing of
the borehole wall cannot be properly shot
peened. A peening lance with deflecting nozzle
or the internal deflection method must be applied
under controlled conditions.
Boreholes with diameters up to 2.4 mm are
peened in production by applying the internal
deflection method (Fig. 12a). Using the deflecting nozzle (Fig. 12b), the internal surfaces of
long pipes, oil drilling gear, helicopter rotor
blades, hydraulic cylinders, and turbine blades
are shot peened.
Figures 2 and 3 show the residual stress profiles of different working materials depending on
the special shot peening parameters. The figures
clarify the potential variation in breadth in forming residual compressive stresses. Zinn listed the
compressive stresses from peening with respect
to welded aluminum alloys (Ref 69).
For workpieces that are load bearing in one
direction only and that require a durability in excess of that which can be reached by conventional shot peening, the peening is applied under
prestressed condition. The workpiece is peened
in a prestressed or expanded state. Stress peening
generates a much higher residual compressive
stress than is possible by conventional shot peening. The residual stress values generated under
prestressed conditions can be as high as the yield
strength of the working material. Residual
stresses resulting from stress peening are shown
in Fig. 13.
Fig. 14
Peening media. Courtesy of Metal Improvement Company, Inc.
There is no nondestructive testing method to

determine the proper shot peening of a component. To obtain optimal shot peening results it is
necessary to optimize the shot peening parameters applied and ensure that all parameters are
controlled during the peening process. For an effective control of the shot peening process the
parameters must be addressed in the following
order:
1.
2.
3.
4.
Shot (media) control

Intensity control
Coverage control
Equipment control
The most important shot peening parameter relates to the microstructure, chemistry, shape,
size, material, and hardness of the peening material shot. All these variables are described in
Ref 42 as well as by the condition of the peening
material.
The peening material must be uniform in size
and fundamentally spherical in shape without
sharp edges or broken particles. Broken or sharpedged peening material can be potentially damaging to the surface of the component. Examples
of this are shown in Fig. 14 to 16. During the
automatic shot peening process, the peening material must be sieved and sorted continuously
(Ref 42).
Cast steel shot is available in two hardness

ranges, namely 45 to 55 HRC and 55 to 65 HRC.
For steel components exceeding 200 ksi (1375
MPa) in tensile strength hard shot with a hardness range 55 to 65 HRC or ceramic shot with
comparable hardness at 57 to 63 HRC should be
used (Ref 42).
For nonferritic materials, shot peening is carried out with stainless steel shot or using ceramic
or glass beads. The decontamination of nonferritic components after shot peening with ferritic
shot by using glass is a common practice to exclude cathodic corrosion.
The shot peening intensity is controlled by the
Almen strip. This is a strip of cold-rolled
spring steelSAE 1070square-edged and
uniformly hardened and tempered to 44 to 50
HRC. The Almen strip is fixed on a specified
holding fixture. During the peening process, the
Almen strip will be exposed to the peening material from one side. The plastic deformation on
the strip surface facilitates a bending of the strip.
The height of the curvature on being measured
in a standard Almen gage is referred to as the arc
height and is a dimension for measuring shot
peening intensity.
Fig. 15
Fig. 16
The Almen block(s) with the strip can be

mounted on an actual component and inserted at
the critical angles of the part to be shot peened
(Fig. 17).
Frequently, there are no actual components
available for cost reasons. A dummy representing the surfaces to be shot peened is used instead.
The mounted Almen strips have to be placed in
the same positions as they would be on the actual
component. Also the distance between the actual
surface and the reference surface incorporating
the Almen strip to the nozzle must be identical
(Fig. 18).
There are three different types of Almen
strips, namely, N, A, and C. These Almen strips
are selected after determining the proper depth
of compression required on the part to be shot
peened and the energy level necessary to obtain
that depth of compression. The details of the
measurement referenced by the specification
AMS-S-13165 (Ref 42) are shown in Fig. 19.
Another important shot peening parameter is
the coverage. The effect of controlled shot peening is considered achieved if the entire component surface is uniformly dimpled or obliterated.
The visual inspection of large areas of steel parts,

fillets, cavities, grooves, or internal bores with a
tenfold magnifying glass to determine a full coverage is extremely difficult, if not impossible.
There are two methods to execute the second
parameter of coverage control (Ref 42):
Visual inspection using a magnifying glass
with a tenfold magnification
Inspection using an approved liquid tracer
system.
Since 1976 the Dyescan fluorescent tracer liquids have been available as they are used in the
Peenscan process, an approved liquid tracer system.
The working principle of this process is similar to the dye penetration method to determine
cracks.
Prior to shot peening, an approved fluorescent
tracer liquid is brushed, sprayed, or dipped on
the component, and it is allowed to dry. This
forms a fluorescent elastic coating that is subsequently removed at a proportional rate to the
shot peening coverage. The fluorescent tracer
system reacts to deviations in shot peening intensity, incomplete coverage, and an excessively
low angle of impact. By inspection under ultraviolet (UV) light the Peenscan process provides
a practical method of determining the peening
coverage in terms of quantity and the uniformity
Shot peening media. (a) Unacceptable. (b) Acceptable. Courtesy of Metal Improvement Company, Inc.
(a) Surface damage caused by poor shot control. (b) Acceptable surface. Courtesy of Metal Improvement
Company, Inc.
Fig. 17
Almen fixture (strapped turbine disk). Courtesy

of Metal Improvement Company, Inc.
Fig. 18
Almen fixture (dummy). Courtesy of Metal Improvement Company, Inc.

of tracer removal. This process has been established to be clearly superior for determining the
peening coverage to visual inspection using a
tenfold magnifying glass. The process is demonstrated in Fig. 20 and 21.
Only automatic shot peening plants facilitate
the repeatability of the shot peening parameter
and are therefore mandatory.
These shot peening plants offer the possibility
to monitor, control, and record all the important
parameters for accurately verifying the shot
peening process. Such characteristic peening parameters are the air pressure in every nozzle of
the peening wheel, material flow of every nozzle
or speed of every peening wheel, revolution or
forward speed of the workpiece, oscillation of
the nozzles, distance of the nozzles, peening time
of every nozzle, and the total processing time.
In the set intervals, the process parameters are
compared with the maximum and minimum
threshold values. On exceeding or falling below
the threshold values, the machine stops itself and
indicates the deviation to the operator and the
quality assurance department. The mistake must
be eliminated prior to the machine resuming the
working process precisely from the point of stop-
page. Figure 22 shows a sketch of a computercontrolled shot peening plant. A typical operational chart of a computer-controlled MIC shot
peening plant is shown in Fig. 23. Every working
step is completed by the printout as proof and
documentation. An interruption of the shot peening process and its causes are recorded on the
printout (ABORT in Fig. 24). The computer
printout provides the data which document that
every component was treated with the same shot
peening parameters as the sample component.
The shot peening plants are equipped with important additional functions, such as peening material sorting by size as well as by shape in order
to eliminate irregular or inadmissible particles.
Process Optimization
Optimization of the effects of shot peening is
achieved by a careful selection of the composition, hardness, and size of the peening material;
the transferred energy at impact (which controls
the depth of the compressive layer); and the percentage of coverage, in addition to component
material and geometry. The many possible com-
30 0.015 in.
0.031
0.001 in.
0.051
0.001 in.
0.0938
0.001 in.
N strip
Peening nozzle
A strip
C strip
Almen strips
Shot stream
46 in.
Measuring dial
0.745
0.750 in.
Hardened
ball
supports
Arc
height
3.0 in.
Holding
fixture
(a)
Arc height
0.75 in.
Strip removed. Residual
stresses induce arching
(b)
Peenstress Software
Peenstress (Ref 70) is a computer program
that is capable of modeling the residual compressive stresses that are introduced by controlled shot peening in most of the engineering
metals, mild and high-strength steels, stainless
steels, carburized and carbonitrided steels, aluminum alloys, titanium alloys, and nickel alloys
(Fig. 25).
The Peenstress program will print out a stress
curve that shows both the magnitude of the residual stress distribution and the depth profile
beneath the surface, for the material and the parameters of shot peening that were selected (Fig.
26). One can use the curve to read the magnitude
of the compressive stress at the surface and at
depth and use that magnitude in load calculations. One can see very graphically what would
be the resultant residual tensile stress on the opposite unpeened surface or in the core of a part
that is peened all over (Fig. 27) and then have
the opportunity to adjust the peening parameters,
if necessary. Most significantly, this can now be
done on a computer screen without having to
resort to very expensive and time consuming xray diffraction (XRD) measurements or even
laboratory and field testing of many variables.
Testing is still recommended, but can be reduced
to a few selected data points.
C.A.S.E. Process
Almen test strip
10/32
screws
binations of these factors make the correct

choice a difficult one.
Strip mounted for

height measurement
(c)
Fig. 19
The Almen strip system. Courtesy of Metal Improvement Company, Inc.
Fig. 20
Shot peening coverage. (a) Partial coverage. (b) Full coverage. Courtesy of Metal Improvement Company, Inc.
Frequently, when the bending fatigue life of

gears has been increased by shot peening the
tooth roots, surface fatigue or pitting of the contact faces became the next mode of failure. There
has been a need, particularly for automotive and
aircraft transmissions, to extend the surface fatigue life beyond that provided by shot peening
alone and smoothing the surface after peening
has been found to be very effective (Ref 67).
Unfortunately, the processes to accomplish this
have been time consuming and therefore costly.
The C.A.S.E. process (Ref 71) can reduce the
relative finishing time and costs, after shot peening, and provide a surface which, in addition to
having the compressive stresses from the shot
peening, can have Rsk value of as little as 1.1.
Very significant increases in surface contact fatigue (pitting) life have been observed.
C.A.S.E. is an acronym for chemically assisted surface engineering and covers a twostage process. The C.A.S.E. Process requires that
the gears are properly shot peened, followed by
chemically assisted superfinishing.
Very different surface profiles can be described using nearly the identical Ra values (Fig.
28). The aim of C.A.S.E., however, is to smooth
the spherical edges and to create a profile as
shown in Fig. 29. The value corresponding to
this profile is:

Rsk skewness
More details are provided in Ref 72 and 73.

The values in Table 1 show clearly that a surface processed with C.A.S.E. is not sufficiently
described by the Ra value. The meaningfulness
lies in Rsk, which has a value clearly below 1,
which is the minimum value striven for.
The stress profiles in Fig. 30 show the tensile
stresses and the residual peening stresses as well
as the C.A.S.E. influence.
X-Ray Diffraction Reading
Fig. 21
The Peenscan system. (a) Coated unpeened. (b) Peened 15 s partial coverage. (c) Peened 60 s full coverage.
(d) Peened 60 s, improper nozzle angle in cavity. Courtesy of Metal Improvement Company, Inc.
A significant breakthrough in the controls of

the shot peening process was recently introduced
to industry. This new method entitled Milam,
utilizes XRD readings of residual stress profiles,
which are much more economical and are available in days rather than weeks. The new method
permits the use of XRD profiles as quality-control measurements that can be recorded for statistical process control data. The practical Milam
system has wide applications: for instance, lowvolume aircraft components, high-volume automotive transmission gears, and even in-the-field
shot peening of chemical processing vessels. It
can also be used to predetermine with high accuracy the shot peening parameters for process
development.
The Milam system actually offers three alternative applications. The first employs coupons
that are essentially the same size as a standard
Almen strip, 0.75 by 3 in. (19 by 75 mm), but
Display
Display
Machine
terminal
Air
pressure
Master
terminal
Manual input
Manual
input
Production
data
Data
Product
data storage
Data
Shot
flow
New
procedures
Data
Input
Master
computer
Data
Instruction
Part
motion
Instruction
Nozzle
motion
Printer
Data
Data
Peening
machine
comput.
Procedures
from storage
Status
Data
collection
system
Instruction
Limit switches
and misc.
devices
Fig. 22
Computer-controlled shot peening machine.

Fig. 23
Status
Status
Software path flow diagram. Courtesy of Metal Improvement Company, Inc.
Bulk shot
handling and
conditioning
system
Customer
Part number
:
:
Specification
operation
: AMS-S-13165
:
Mic procedure
Shot size
Intensity
Coverage
Saturation time
100% cover time
:
:
:
:
:
:
Nozz. 01
Fig. 24
Table 1
21-8592
Ml-170R
6-10A
100%
6 minutes
6 minutes
Air press TMR

Shot flow TMR
Turntable TMR
Speed change
Max shot flow
Turntable RPM
5
15
5
5
25
16.0
Sec
Sec
Sec
Sec
Lbs
RPB
Mode
Cycle count
Cycle time
Horz
NA
6:00:00
Oscillation program
Pos In min Pos
Home 1 30.0 32.0 11 19.5
start 2 24.0 4.0 12 19.0
3 23.5 8.0 13 18.5
4 23.0 12.0 14 18.0
5 22.5 14.0 15 17.5
6 22.0 15.0 16 17.0
7 21.5 16.0 17 12.0
8 21.0 17.0 18 12.0
9 20.5 18.0 19 12.0
10 20.0 19.0 TA 12.0
Nozzle program
Nozz. Active
1 2:00:00 A
2 2:00:00 A
3 2:00:00 A
4 2:00:00 A
5 4:00:00 A
6 4:00:00 A
7 4:00:00 A
8 6:00:00 A
9 6:00:00 A
10 6:00:00 A
Hi limit
Air Shot
Psi lb M
60 18.0
Nozz. 02 Nozz. 03 Nozz. 04 Nozz. 05 Nozz. 06 Nozz. 07 Nozz. 08 Nozz. 09 Nozz. 10 Turntable
speed
Air Shot Air Shot Air Shot Air Shot Air Shot Air Shot Air Shot Air Shot Air Shot
Psi lb M Psi lb M Psi lb M Psi lb M Psi lb M Psi lb M Psi lb M Psi lb M Psi lb M
RPM
60 18.0 60 18.0 60 18.0 60 18.0 60 18.0 60 18.0 40 16.0 40 16.0 40 16.0
11.0
Data
55
16.0
55 16.0 55 16.0 55 16.0 55 16.0 55 16.0 55 16.0 35 14.0 35 14.0 35 14.0
Lo limit
55
14.0
SN001
50
14.0
SN002
Abort 1:14:10 low shot flow nozz. 02 at lb min
SN002
56
16.0
Remaing
cycles
0
Oscill.
position
In min
2.0
Inches
6.0
50 14.0 50 14.0 50 14.0 50 14.0 50 14.0 50 14.0 30 12.0 30 12.0 30 12.0
9.0
2.0
56 16.0 55 15.0 55 14.0 56 15.0 51 14.0 52 15.0 35 13.0 34 13.0 34 14.0
10.0
6.0
9.5
5.0
56 15.0 57 17.0 56 15.0 57 16.0 55 16.0 54 15.0 36 14.0 34 14.0 34 14.0
Tue 5:19:47 PM
Feb 16, 1999
Oscill.
speed
10.0
Typical documentation by computer printout. Courtesy of Metal Improvement Company, Inc.
Surface characteristics (17 CrNiMo 6)

Grounded and
shot peened
Grounded
Grounded and shot

peened and C.A.S.E.
Parameter
Axial
Tangential
Axial
Tangential
Axial
Tangential
Rz
Ra
Rt
Rsk
0.36
0.04
0.59
0.83
1.52
0.21
1.96
0.23
0.68
0.10
0.87
0.19
0.69
0.12
0.85
0.18
0.57
0.07
0.85
1.22
0.66
0.07
0.96
1.35
Source: Metal Improvement Company, Inc.
Material Library
Material: Al/7010 * T7651/cysp
Review library
Monotonic
472
Cyclic
425
Elastic limit
(MPa)
Ultim. tensile str.
(MPa) : 560
Elongation
(%)
Endurance limit
(MPa) :
Cancel material
Young's modulus
(MPa) : 71000
Print data sheet
Poisson's ratio
k factor
n factor
Modify mater.
Run
: 0,300
(MPa)
Research mat.
New material
: 10,00
Monotonic
607,60
0,0272
Remarks : ENSAM/Prof. Ericsson/20 cycles/1992
Cyclic
641,20
0,0446
Help
Quit
ENSAMMIC
Material library. The selected material is aluminum alloy 7010 T 76. Screen printout shows the monotonic
strength coefficient k and the strain-hardening coefficient, n, as well as the cyclic strengths coefficient k and
the strain-hardening coefficient, n, which define the true monotonic stress strain curve and the cyclic stress strain curve,
respectively. Also shown are the ultimate and yield strengths, elongation, Youngs modulus, and Poissons ratio, for this
material. Courtesy of Metal Improvement Company, Inc.
Fig. 25
In min
20.0
21.0
22.0
23.0
24.0
25.0
25.0
25.0
25.0
that are 0.25 in. (6.5 mm) thick. The coupons are
designed to fit on a standard Almen block and
can therefore be used on an existing Almen fixture. In this way, the Milam coupons are placed
to represent the critical areas of the part to be
peened. The Milam coupons are made from the
same material (or even at the same temperature
at the customers option) as the parts and heat
treated along with the parts. They are shot
peened after the Almen strip verification and just
before the parts are processed and periodically
thereafter on request of the customer.
The second is an alternative to the Milam coupons but still part of the Milam system. MIC and
Lambda can handle small parts directly as long
as they are within the dimension of 2 by 2 by
1.5 in. (50 by 50 by 38 mm).
With the third alternative, larger parts can be
cut up so that the critical areas for XRD reading
are available, but Lambda must do the cutting.
After shot peening, the test parts are then sent
to the XRD laboratory; because of their standard
size, the coupons or the small parts can be etched
and x-rayed at successive depth levels for residual stress readings in a very expeditious manner.
This is a quick, economical method for measuring induced compressive stresses in components.
Shot Peening Applications

As has already been explained, components
manufactured from all kinds of materials containing metals, ranging from magnesium to steel,
can be utilized for host peening and, thus, for
creating a compressive residual stress in the
workpiece surface. In the case of materials with
a relatively low strength, slight surface faults
that could be the initial cause of fatigue cracks
are in turn obturated by the plastic extension and

0.2
0.1
Depth, mm
F 15 A
S 400 Shot
V = 29 m/s
Al/7010 * T7651/C
100
1999:2:18
Peenstress
ENSAM-MIC
d = 0,092 mm
h = 0,005 mm
Dir. z
200
s, MPa
300
310
m, MPa
332
Pm, mm
0.005
PO, mm
0.210
400
Residual stress profile. Peenstress software calculates and displays the stress profile and the calculated shot
velocity. The separate panel indicates the surface stress s, the maximum stress m, the minimum depth of
the maximum stress Pm, and the depth of compression PO. Courtesy of Metal Improvement Company, Inc.
Fig. 26
1999:2:18
Peenstress
ENSAM-MIC
0.1
0.2
100
200
100
AL/7010 * T7651/C
200
300
300
400
400
enclosed by the residual stress. In the case of

materials that are of high strength, such as may
be utilized for the manufacture of gears, for example, a higher compressive residual stress can
be achieved, which will, among other things, increase the fatigue strength.
The increase in the size of fatigue cracks depends on the stress factors at the peak of the
crack. An increase in the size of any crack can
be prevented by the stress (Ref 74) based on the
following factors:
As a rule, a fatigue crack will not spread un-
less the tensile stresses increase.
The crack opening stretch does not occur as
long as a compressive stress remains.

Stress profile in a thin plate. When the selected
structure is a thin plate, the stress profile shows
tensile stress compensating for the compressive stress below the surface, which is induced by the shot peening. The
structure is a 0.125 in. (3 mm) thick plate, peened on both
sides with equal parameters. If the peening intensity is too
high, excessive tensile stress may be generated internally
that will have a negative effect on fatigue. Courtesy of
Figure 31 shows clearly how this will operate.

Figure 31(a) shows in a simplified form the
distribution of stress under applied load. Figure
31(b) shows the distribution of the stress of the
shot peened component. Figure 31(c) shows the
distribution of stress resulting from the addition
of load stress and residual stress. Figure 31(d)
shows the distribution of stress after a crack has
spread around the area subject to tensile stress.
The broken lines denote the original distribution
of the stress as shown in Fig. 31(c). Figure 31(e)
shows a well-advanced stage in the formation of
a crack as far as point (I). In the surrounding area
there is a change in the stress. Figure 31(f) shows
that the crack no longer extends within the area
subject to residual compressive stress.
Controlled shot peening in the manufacture of
gears (Fig. 32) is to be found in the automobile
industry, in commercial vehicles, in marine
gears, in the manufacture of drive shafts, and
extends even to the manufacture of small gears
in hand machine tools and, at the other extreme,
extends to its use in the manufacture of large
gears used in marine engines, as well as for mining equipment. The radiuses at the base of the
tooth are normally those parts of the component
that are subject to most stress and should, therefore, be peened (Fig. 33 and 34) (Ref 75). In
practice, it has also been found, though, that the
peening results in concaving on the flanks that
will retain oil and thus provide better lubrication.
Wear, noise, pitting, and seizing may also be reduced, and the running temperature will drop.
Gears with a low dimension tolerance or with a
high level of surface quality can be lapped or
honed after shot peening, provided that during
the fine finishing care is taken not to remove
more than 10% of the residual compressive
stress layer.
Compressive stresses induced by controlled
shot peening may have a lot of benefits, for example, in the chemical industry applications by
Fig. 27
Negative skewness
Positive skewness
Ra = 2.4 m
Negative
Ra = 2.5 m
Ra = 2.4 m
Comparison of Ra values for different profiles

in C.A.S.E. processing. Courtesy of Metal Improvement Company, Inc.
Fig. 28
Positive
Fig. 29
Skewness of the amplitude distribution in C.A.S.E. processing. Courtesy of Metal Improvement Company, Inc.

increasing of the resistance against stress-corrosion cracking. This mechanism eliminates surface tensile stress. Stress-corrosion cracking cannot occur in an area of compressive stress.
Figures 35 and 36 show the measured residual
stress of a welded StE 460 (P 460 N) steel plate.
This material is used for manufacturing railway
tank cars for ammonia. The residual stresses
were measured to document the suitability of
shot peening. Stresses were measured in the untreated semifinished material on the untreated
welding seams as well as on the shot peened areas. During 1989 and 1990, about 60 railway
tank cars used for transporting ammonia were
restored, showing no sign of stress corrosion

even after ten years of operation.
Shot peening is also successfully applied to
chemical process vessels made of austenitic
steel. The results of the laboratory tests for stress
corrosion are shown in Table 2. In these tests,
samples were exposed to a 42% MgCl2 solution
at 145 C.
In the absence of stress-corrosion cracks, the
investigations into the shot peened specimens in
line with exposure were discontinued at 1000 h.
The respective production costs of a 5000 L tank
container in different materials are compared in
Fig. 37.
200
0
Residual stress (ES), MPa
200
400
600
Lasershot Peening
800
1000
ES
Ground
Ground and shot peened
Ground and shot peened and C.A.S.E.
1200
1400
1600
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Depth, mm
Fig. 30
Stress profile (17 CrNiMo 6). Courtesy of Metal Improvement Company, Inc.
II
(a)
(c)
(e)
I
II
60
(b)
(d)
(f)
Crack arrest by compressive self stress. (a) Load stress. (b) Self stress. (c) Sum of (a) and (b). (d) Crack starts.
(e) Crack progressed to edge of former compressive zone. (f) Crack arrested. Courtesy of Metal Improvement
Company, Inc.
Fig. 31
On the basis of the corrosion test results and

the production cost considerations, tanks and
other constituent parts of the plant could be manufactured from a less corrosion-resistant and
therefore less expensive material. Overall costs
can be reduced by 50% after allowing for the
cost of shot peening.
Also the resistance against corrosion fatigue
will be improved by shot peening, caused by the
beneficial effect of the compressive stress. Fretting fatigue can occur when a rotating component is press fitted onto a shaft. A slight relative
movement of the mating surfaces caused by vibration or shifting loads will shear the high
points from the top of the peaks in the mating
surfaces. The resulting debris will oxidize and
produce the typical rusty powder characteristics
of fretted steel. This very hard debris tends to
build up in one area, abrading the surface and
creating stress risers that can propagate into fatigue cracks.
In industry, the piston rod bushes of large diesel engines, joint faces of machine components,
bearing pins, and threads are shot peened.
Lasershot peening, also a surface treatment,

employs laser-induced shocks to create deep
compressive stresses beyond a 0.040 in. (1 mm)
depth with magnitudes comparable to that produced by shot peening. Lasershot peening has
proven to significantly enhance the damage tolerance of components and has generated sufficiently impressive results to move it from a laboratory-demonstration phase into a significant
industrial process technology.
Until now this evolution has been slowed
down by the absence of a laser system meeting
the power requirements for economical processing. Commercial lasers available at present are
limited to power outputs of 10 to 100 J, with
repetition rates of around 0.25 Hz and a new design in development with 1 Hz capability. The
estimated time required to process an area of 1
by 3 in. (25 by 75 mm) could take anything from
1 to 3 h using one of these machines.
Robust laser systems have been developed
that for the first time push the average power
output to 200 J at 10 Hz. It enables the processing of a 3 in.2 (19 cm2) area in minutes instead
of hours.
With the invention of the laser, it was rapidly
recognized that the intense shocks required for
peening could be achieved by means of tamped
plasmas that were generated at metal surfaces by
means of high-energy density (200 J/cm2) lasers with pulse length in the tens of nanoseconds
range. Initial studies on laser shock processing
of materials were carried out by the Battelle Institute (Columbus, OH/USA) from about 1968
to 1981 (Ref 76, 77). Excellent recent work has
also been reported in Ref 78.
Laser intensities of 100 to 300 J/cm2 with a
pulse duration of about 30 ns can generate shock
pressure of 104 to 105 atm when absorbed on a
metal surface. A thin layer of black paint on the

surface provides an excellent absorber that produces a plasma burst that is inertially confined
with a surface layer (tamp) such as water (Fig.
38). These shocks impart compressive stresses
to depths greater than 0.040 in. (1 mm).
Special techniques for controlling the temporal and spatial shape of the pulse are used to
prevent the high-intensity laser from breaking
down the water column or generating stimulated
processes that reflect the laser energy before
reaching the paint surface. With appropriate care
given to the setup, impressive results can be
achieved from Lasershot peening.
In order to use the high throughput capability

of the laser, it is important to have manipulation
hardware to rapidly and precisely move surfaces
to be treated through the rapidly pulsing laser
beam. The authors have designed a five-axis,
computer-controlled robotics system with motion synchronized to the laser pulses.
As an example of the laser process, Fig. 39
shows the residual stress induced in Inconel by
Lasershot peening and contrasts it with typical
results achieved by shot peening. Laser-generated shock can be tailored to develop compressive stresses deeper into the material with mag-
nitudes similar to those generated by shot

peening. Additionally, corrosion takes more time
to penetrate the compressive layer induced by
Lasershot peening.
Deep compressive stress is important for critical areas of components such as turbine blades
because it prevents the foreign object damage
caused by debris being sucked up into a turbine
engine from becoming a crack initiation site.
Most metals can be successfully Lasershot
peened to very deep depths of compression (Ref
7982).
REFERENCES
Compressive stress, ksi

0
25
50
75
100
125
150
175
200
0.0015
0.05
0.10
0.0045
Depth, mm
Depth, in.
0.0030
0.15
0.0060
0.0075
0.02
0.0080
0
500
1000
Compressive stress, MPa
Fig. 32
Stress profile of carburized gear tooth root, ground, then shot peened with special hardness shot. Courtesy of
6000
Shot peened
Peak cyclic load, nonreversing, lb
Mean
1000
4000
None shot peened
Mean
3000
500 bevel gear
Spiral
3,625 D.P.
Material: 8620H bar stock
Heat treatment: carburized and hardened, 61 HRC
Load transmitted: 135 hp, 5000 rpm
Shot peening specifications:
Shot size: 170H
Intensity: 0.010-0.014 A
Coverage: 200%
2000
1000
500
0
104
105
106
107
Cycles to failure
Fig. 33
Increase in fatigue resistance of spiral bevel gear. Courtesy of Metal Improvement Company, Inc.
Peak cyclic load, nonreversing, N
5000
1. W.E. Newton, Sand Blast and Its Adaption

to Industrial Purposes, J. Soc. Arts, Vol 23,
1875, p 257260
2. F.C. Brooksbank, Sand Blast Apparatus for
Cleaning Castings, Iron Age, Vol 57, 1896,
p 640642
3. F.P. Zimmerli, Shot Blasting and Its Effects
on Fatigue Life, Surface Treatment of Metals, American Society for Metals, 1941, p
261278
4. F.P. Zimmerli, How Shot Blasting Increases
Fatigue Life, Mach. Des., Vol 12, Nov
1940, p 62
5. J.O. Almen, Peened Surfaces Improve Endurance of Machine Parts, Met. Prog., Vol
43 (No. 2), 1943, p 209315
43 (No. 5), p 737740
43 (No. 8), p 254261
43 (No. 9)
9. J.O. Almen, Shot Blasting to Increase Fatigue Resistance, SAE Trans., Vol 51 (No.
7), 1943, p 248268
10. Minich, Abrasive Wheel Throwing Machine, U.S. Patent 2077636, 1934
11. R.L. Mattson and J.O. Almen, Effect of
Shot Blasting on the Mechanical Properties
of Steel, Final Report, OSRD, 3274, 4825,
6647, (NA-115), Office of Scientific Research and Development, Washington D.C.,
1945
12. R.L. Mattson, H.E. Fonda, and J.O. Almen,
Final Report on Effect of Shot Blasting on
the Mechanical Properties of Steel, (NA115) (OD-177), National Defense Research
Committee of the Office of Scientific Research and Development, War Metallurgy
Division, 1944
13. R.L. Mattson and J.O. Almen, Effect of
Shot Blasting on the Physical Properties of
Steel, Final Report, part 3, OSRD 6647,
National Defense Research Committee of
the Office of Scientific Research and Development, War Metallurgy Division, 1946
14. J.O. Almen, Fatigue of Metals as Influenced
by Design and Internal Stresses, Surface
15.
16.
17.
18.
Stressing of Metals, H.F. Moore, W.M.

Murray, J.O Almen, O.J. Horger, and P.R.
Kosting, Ed., 48 Feb 1946, p 3384
berblick uber die UntersuchunW. Bautz, U
gen uber die Dauerhaltbarkeit von Federn.
Interner Untersuchungsbericht, 28.3, 1938
(in German)
H. Wiegand, Oberflache und Dauerfestigkeit, Berlin-Spandau, BMW-Flugmotorenbau, 1940, p 83 (in German)
O. Foppl, Oberflachendrucken zum Zweck
der Steigerung der Dauerhaltbarkeit mit
Hilfe des Strahlkugelgeblases, Mitteilungen
des Wohler-Instituts, Braunschweig, 1939,
p 4459 (in German)
R. Z. von Manteuffel, Dauerhaltbarkeit von
Kraftfahrzeugfedern und Moglichkeiten zu
Table 2
19.
20.
21.
22.
ihrer Beeinflussung, Dtsch Kraftfahrt., No.

49, 1941 (in German)
H.F. Moore, Effect of Surface Stressing
Metals on Endurance under Repeated Loadings, Surface Stressing of Metals, H.F.
Moore, W.M. Murray, J.O Almen, O.J. Horger, and P.R. Kosting, Ed., 48 Feb 1946,
p 118
W.M. Murray, Measurement of Surface
Stresses, Surface Stressing of Metals, H.F.
Moore, W.M. Murray, J.O Almen, O.J. Horger, and P.R. Kosting, Ed., 48 Feb 1946,
p 1932
O.J. Horger, Mechanical and Metallurgical
Advantages of Shot Peening, Iron Age, 29
March/5 April 1945
M. Olley, Influence of Shot Peening on
Shot peened area

Grounded area
60
1
135
320
380
Stress data of welded StE 460 (P 460 N). Stress:
location 1, 484 MPa; location 2, 515 MPa;
location 2.1, 171 MPa, 1 mm subsurface; location 5,
20 MPa; location 6, 389 MPa. Courtesy of Metal Improvement Company, Inc.
Fig. 35
Results of stress-corrosion cracking tests

Load, h
Material and shot peening parameter
At 50%
At 70%
At 80%
At 90%
X 6 CrNiTi 1810 (1.4541)

Not shot peened
SSCW 0.8, 5 A, 100%
SSCW 0.8, 12 A, 100%
SSCW 0.8, 20 A, 100%
11
1000
3.2
1000
X 6 CrNiTi 17122 (1.4571)

Not shot peened
SSCW 0.8, 5 A, 100%
SSCW 0.8, 12 A, 100%
SSCW 08, 20 A, 100%
17
11.3
1000
1000
7.5
1000
1000
X 2 CrNiMoN 225-3 (1.4462)

Not shot peened
SSCW 0.8, 5 A, 100%
SSCW 0.8, 12 A, 100%
SSCW 0.8, 20 A, 100%
5.3
3.3
1000
1000
1000
1000
1.3
2
5.3
Specimen tensile tested in 42% MgCl2 medium at 145 C and at a load of 50 to 90% of the 0.2% yield point
Shot peened area

Repaired area
7
9
Grounded area
320
Stress data of repair welded StE 460 (P 460 N).
Stress: location 7, 357 MPa; location 7.1,
78 MPa, 1 mm subsurface; location 9, 434 MPa. Courtesy of Metal Improvement Company, Inc.
Fig. 36
90,000
80,000
100
700
600
80
500
70
Not peened
Computed stress, MPa
Computed stress, ksi
Peened
Cost of production, Euro
70,000
90
60,000
50,000
40,000
30,000
20,000
60
400
10,000
0
50
105
1.4541
106
Cycles to failure
Fig. 34
Fatigue chart of carburized gears. Courtesy of Metal Improvement Company, Inc.
107
1.4571
1.4539
1.4462
Material
Fig. 37
Cost comparison of 5000 L vessel. Courtesy of


Mechanical Properties, Symposium on
Shot Peening, American Society for Metals,
1944
23. F.P. Zimmerli, Shot Peening of Coil and
Leaf Springs, Symposium on Shot Peening, American Society for Metals, 1944
24. K. Walz, Federfragen- insbesondere aus
dem Gebiet der Zeitfestigkeit bei sinusformiger und schlagartiger Beanspruchung,
Broschure der Mauser-Werke, Oberndorf,
1943
25. H.O. Fuchs and R.L. Mattson, Measurement
of Residual Stresses in Torsion Bar Springs,
Proc. Soc. Exp. Stress Anal., Vol 4 (No. 1),
1946, p 64
26. J.C. Straub and D. May, Stress Peening,

Iron Age, 21 April 1949, p 6670
27. Manual on Shot Peening, Sp-84, Society of
Automotive Engineers, 1952
28. S.W. Serensen, Zusammenhange zwischen
der Schwingfestigkeit, der Ober- flachenverfestigung und der konstruktiven Gestaltung, Hohere Gestaltfestigkeit durch Oberflachenverfestigung, Vortrage auf der
Tagung der Wnitomasch, Moscow, April
1954, p 1339
29. N.P. Stschapow, Oberflachenverfestigung
im Eisenbahnwesen, Hohere Gestaltfestigkeit durch Oberflachenverfestigung, Vortrage auf der Tagung der Wnitomasch, Moscow, April 1954, p 4055
Laser beam
Pressure
Tamping material
Protective overlay
for light absorption
Part to be
processed
Shock waves
Plasma
burst
causes
pressure
Fig. 38
0+
Pressure
wave
Residual
compressive
stress
Lasershot process. Courtesy of Metal Improvement Company, Inc.
Depth, mm
0
0.25
0.5
0.75
Shot peened
0.010 A
40
200
400
60
80
600
100
Lasershot peened
by MIC-LLNL
800
120
140
1000
160
1200
180
0
0.01
0.02
0.03
0.04
Depth, in.
Fig. 39
Stress profile of Inconel 718. Courtesy of Metal Improvement Company, Inc.
1400
0.05
20
200
1.25
0
30. B.G. Gurewitsch and S.F. Jurjew, Einflu

der Restspannungen auf die Erhohung der
Schwingfestigkeit nach der chemischen und
thermischen Behandlung, Hohere Gestaltfestigkeit durch Oberflachenverfestigung,
Vortrage auf der Tagung der Wnitomasch,
Moscow, April 1954, p 5678
31. B.F. Balaschow, Festigkeitserhohung durch
Nitrieren von Maschinenteilen, Hohere
Gestaltfestigkeit durch Oberflachenverfestigung, Vortrage auf der Tagung der Wnitomasch, Moscow, April 1954, p 7999
32. I.W. Kudrawzew, Einflu der Restspannungen auf die Schwingfestigkeit geschweiter
Werkstucke Hohere Gestaltfestigkeit durch
Oberflachenverfestigung, Vortrage auf der
1954, p 100118
33. M.M.
Kobrin,
Festigkeit
der
Preverbindungen mit oberflachengewalzten Sitzflachen Hohere Gestaltfestigkeit
durch Oberflachenverfestigung, Vortrage
auf der Tagung der Wnitomasch, Moscow,
April 1954, p 119130
34. A.W. Rabtschenkow, Oberflachenbehandlung zur Erhohung der Schwingfestigkeit
von Stahl unter Korrosionswirkungen Hohere Gestaltfestigkeit durch Oberflachenverfestigung, Vortrage auf der Tagung der
Wnitomasch, Moscow, April 1954, p 131
152
35. I.S. Koslowski, Festigkeitserhohung durch
einsatzgehartete Zahnrader im Kraftfahrzeugbau Hohere Gestaltfestigkeit durch Oberflachenverfestigung, Vortrage auf der Tagung der Wnitomasch, Moscow, April
1954, p 153169
36. W.T. Tschirikow, Erhohung der Festigkeit
und Lebensdauer einsatzgeharteter Kraftfahrzeugzahnrader, Hohere Gestaltfestigkeit
durch Oberflachenverfestigung, Vortrage
auf der Tagung der Wnitomasch, Moscow,
April 1954, p 170180
37. N.A. Karasjew, Erhohung der Dauerfestigkeit von Maschinenteilen durch Kugelstrahlen, Hohere Gestaltfestigkeit durch Oberflachenverfestigung, Vortrage auf der
1954, p 181194
38. D.S. Listgarten, Einflu von Oberflachengute und Warmbehandlung auf die
Schwingfestigkeit von Federstahl, Hohere
Gestaltfestigkeit durch Oberflachenverfestigung, Vortrage auf der Tagung der Wnitomasch, Moscow, April 1954, p 195217
39. Shot Peen Forming, Steel, Metalwork.
Week., 7 July 1958
40. Shot Peening of Metal Parts, Military
Specification No. MIL-S-13165, 1956
41. Shot Peening of Metal Parts, Military
Specification MIL-S-13165 B Amendment2, 25 June 1979
42. Shot Peening of Metal Part, AMS-S13165
43. Werkstofftechnische Verfahr en. Techn. Betriebshandbuch der DLH Nr. TBH-90-K,
15. April 1965
44. Spezifikation fur Kugelstrahlen von Metal-
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
len, MBB-UFE-Entwurf Nr. 18/70, Aug

1970
A 300 B-Landing GearTechnical Information for Implementing Main Parts Repair
Procedures, Doc. No. 16171, Messier-Hispano Product Support
Shot Peening of Steel and Titanium Alloys, Process Specification No. MA-57,
Rev. B, Northrop Corp. Norair Div., 28 July
1967
Materials and Processes Requirements for
Air Force Systems, Military Standard No.
MIL-STD-1587 (USAF), 31 July 1976
Shot Peening, Report No. 201003,
WU 813, Boeing, 15 June 1969
Shot Peening, Computer Monitored,
AMS-S-2432
Shot Peening, AMS-2430
Peening Media, AMS-2431
Cut Wire Shot, SAE J 441, Society of Automotive Engineers
Test Strip, Holder and Gage for Shot Peening, SAE J 442, Society of Automotive Engineers
Cast Shot and Grit Size Specifications for
Shot Peening and Cleaning, SAE J 444,
Society of Automotive Engineers
Metallic Shot and Grit Mechanical Testing, SAE J 445, Society of Automotive Engineers
Cast Steel Shot, SAE J 827, Society of
Automotive Engineers
Specification for Low-Carbon Cast Steel
Shot, SAE J 2175, Society of Automotive
Engineers
Procedures for Using Standard Shot Peening Test Strip, SAE J 443, Society of Automotive Engineers
W. Schutz, Betriebsfestigkeitsversuche mit
Spurstangenhebeln und Lenkhebeln zum
Typ Buffel LBF-Bericht Nr. 961/9621/
1058, Darmstadt, Nov 1960
W. Schutz, Impact/Anwendungsbericht Kugelstrahlverfahren, Kugelstrahlen in der Antriebstechnik, Metal Improvement Company, Inc., Unna/ Deutschland
61. H. Wohlfarth, Ein Modell zur Vorhersage

kugelstrahlbedingter
Eigenspannungszustande. In Macherau, Hauck: EigenspannungenEntstehung, Messung, Bewertung,
DGM Informationsgesellschaft Verlag, Oberursel, 1983, Band 2
62. H. Wohlfahrt, Residual Stress and Stress
Relaxation, E. Kula and V. Weiss, Vol 28,
Sagamore Army Mat. Res. Conf., Plenum
Press, Paris, London, 1982, p 71
63. Shot Peening Applications, Engineering
Manual, 8th ed., Metal Improvement Company
64. H.O. Fuchs, Shot Peening Stress Profiles,
Metal Improvement Company
65. B. Scholtes, Eigenspannungen in mechanisch rundschichtverformten WirkstoffzustandenUrsachen, Ermittlung und Bewertung.
DGM
Informationsgesellschaft,
Oberursel 1980
66. D.P. Townsend and E.V. Zaretsky, Effect
of Shot Peening on Surface Fatigue Life of
Carburized and Hardened AISI 9310 Spur
Gears, NASA Technical Paper 2047, National Aeronautics and Space Administration, 1982
67. D.P. Townsend, Effect in Surface Fatigue
Life of Hardened Gears by High-Intensity
Shot Peening, NASA Technical Memorandum 105678, National Aeronautics and
Space Administration
68. A. Ahmad and E.D. Crouch, Dual Shot
Peening to Maximize Beneficial Residual
Stresses in Carburized Steels, ASM International/Reprint 8904006
69. W. Zinn, Untersuchungen zum Dauerschwingverhalten
von
Stumpfschweiverbindungen aus den naturharten
Aluminiumlegierungen AlMg3 Deutscher
Verlag fur Schweitechnik, Dusseldorf,
1988
70. Y. Le Guernic, Impact/Review of Shot
Peening Technology, Peenstress Software
Selects Shot Peening Parameters, Metal
Improvement Company, Inc., France
71. Impact/Review of Shot Peening Technol-
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
ogy, The CASE for Superfinishing, Metal

Improvement Company, Inc.
L. Mummery, Surface Texture Analysis,
Hommelwerke GmbH, VS-Schwenningen/
Germany Handbook
Hommel Surface Roughness, Terminology
and Parameters, Hommelwerke GmbH, VSSchwenningen/Germany
H.O. Fuchs, Regional Tensile Stress as a
Measure of the Fatigue Strength of Notched
Parts, Mechanical Behavior of Materials,
Vol II, p 478448
J.B. Seabrook and D.W. Dudley, Results
of a Fifteen Year Program of Flexural Fatigue Testing of Gear Teeth, American Society of Mechanical Engineers No. 63-WA199
B.P. Fairand and B.A. Wilcox, J. Appl.
Phys., Vol 43, 1972, p 3893
A.H. Clauer, B.P. Fairand, and J. Holbrook,
J. Appl. Phys., Vol 50, 1979, p 1497
P. Peyre and R. Fabbro, Optical Quantum
Electron., Vol 27, 1995, p 12131229
P. Prevey, D. Hombach, and P.I. Mason,
Thermal Residual Stress Relaxation and
Distortion in Surface Enhanced Gas Turbine
Engine Components, Proc. ASM/TMS
Materials Week (Indianapolis, IN), 1518
Sept 1997, ASM International
S.R. Mannava, W.D. Cowie, and A.E.
McDaniel, The Effects of Laser Shock
Peening (LSP) on Airfoil FOD and High
Cycle Fatigue, USAF Structural Integrity
Program Conference, United States Air
Force, Dec 1996
C. Brent Dane, J. Wintemute, B. Bhachu,
and L. Hackel, Diffraction Limited High
Average Power Phaselocking of Four 30J
Beams from Discrete ND:Glass Zigzag Amplifiers, Postdeadline paper CPD27, CLEO
97, 22 May 1997
J.R. Harrison and L.A. Hackel, New Laser
Technology Enables Lasershot Peening to
be a Commercially Affordable Process,
SAE 981963, Society of Automotive Engineers, 1998

DOI: 10.1361/hrsd2002p361

Residual Stress Formation during Casting

R.W. Lewis, University College of Swansea, Wales
K.N. Seetharamu, Z.A. Zainal, and A.Y. Hassan, University of Science Malaysia
MOST FOUNDRIES have stories about castings that flew into pieces with a bang when being
machined, or even when simply standing on the
floor. It is easy to dismiss such stories, but they
should be viewed as warnings. They warn that,
in certain conditions, castings can have such high
stresses locked inside that they are dangerous and
unfit for serviceeven though they look perfect
(Ref 1). More than $50 million is attributed to thermal stress-generated casting defects. Such defects
can be essentially eliminated through the application of computer predictions.
If a casting were to be cooled at a uniform rate
and with a uniform constraint acting at all points
over its surface, then it would reach room temperature perfectly in proportionperhaps a little
large, or a little small, but not distorted. In practice, of course, the casting generally is somewhat
large, or somewhat small, and not quite accurate
in shape. Occasionally, it may be very seriously
distorted. Again, in an ideal world, if the constraint by the mold were either zero or infinite,
in both cases the casting would be of predictable
size and correct shape.
Even if a casting were subjected to no mold

constraint at all, it would certainly suffer internally generated constraints as a result of uneven
cooling. The famous example of this effect is the
mixed-section casting shown in Fig. 1(a). If a
failure occurs, it always happens in the thicker
section. The explanation of this behavior requires careful reasoning. First, the thin section
solidifies and cools. Its contraction along its
length is easily accommodated by the heavier
section, which simply contracts under the compressive load since it is hot, and therefore plastic,
if not actually still molten. Later, however, when
the thin section has practically finished contracting, the heavier section starts to contract. It is
now unable to compress the thin section, which
has become rigid and strong. Thus, the heavy
section goes into tension. Depending on its temperature it will either stretch plastically, hot tear,
or cold crack.
The example shown in Fig. 1(b) is another
common failure mode. The internal walls of the
castings remain hot longest, even though the
casting may have been designed with even wall
or
(a)
or
(b)
Fig. 1
(a) Thick/thin-section casting showing tensile stress in the thick section. (b) Even-walled casting showing internal
tensile stresses
sections. This is, of course, simply the result of

the internal sections being surrounded by other
hot sections. The reasoning is therefore the same
as for the thick/thin-section casting in Fig. 1(a).
The internal walls suffer tension at a late stage
of cooling. This tension may be retained as a
residual stress in the finished casting, or may
lead to catastrophic failure by tearing or cracking. The same reasoning applies to the case of a
single-component heavy-section casting such as
an ingot, billet, or slabespecially when cast in
steel, which has poor thermal conductivity. The
inner parts of the casting solidify and contract
last, putting them into tension. Because of the
low yield point of the hot metal, extensive plastic
yielding occurs.
There is a further type of internal constraint
that appears to be universal in castings of all
sizes and shapes. It is rarely recognized, but was
investigated by Weiner and Boley (Ref 2) in a
theoretical study of a simple slab casting. They
assumed elastic-plastic behavior of the solid, and
that the yield point of the solid was zero at the
melting point (not quite true, but a reasonable
working approximation) and increased as the
casting cooled. They found that plastic flow of
the solid occurs at the very beginning of solidification. The stress history of a given particle
was found to be as follows. On freezing, the particles are subject to tension, and since the yield
is initially zero, its behavior is at first plastic. As
it cools, the tensile stress on it increases and remains equal to the yield stress corresponding to
its temperature, until such time as the rate of increase of stress upon it is less than the rate of
increase of its yield stress. It then starts to behave
elastically. Soon after unloading begins, the
stress on the particle decreases rapidly, becoming compressive, and finally reaching the yield
stress in the opposite direction. Its behavior remains plastic thereafter.
To sum up their findings, in a solidifying material there will be various deformation regimes:
(1) a plastic zone in tension at the solidification
front, since the strength of the solid is low; (2)
a central region where the stresses are in the elastic range; and (3) a zone at the surface of the
casting where there is a plastic flow in compression. The overall scheme is illustrated in Fig. 2.
362 / Residual Stress Formation During Manufacturing Processes
Elastic
compression
Plastic
compression
Elastic
tension
Plastic
tension
(Time)0.5
Liquid
Distance
Plastic
Mold
Elastic
Plastic
Compression
Liquid
+
solid
Tension
Tensile
stress
0
Compressive
stress
0.2
0.4
0.6
0.8
1.0
Distance through solid
Fig. 2
Elastic-plastic regimes in a simple slab casting
Residual tensile stress, MPa
90
80
Gray cast iron
70
60
50
40
Al alloy
30
20
10
0
Approximate solidification time

0
Fig. 3
2
4
8
16
32
Time in mold before stripping
64
Residual stresses in aluminum alloy and gray

iron castings as a function of stripping time
The model probably would be improved, but not

fundamentally altered, by using creep flow behavior instead of the elastic-plastic flow behavior
with yield stress a function of temperature.
Richmond and Tien (Ref 3, 4) criticize this
model on the grounds that it does not account
for the friction at the casting/mold interface.
When Richmond and Tien included this, they
found that the casting/mold interface is no
longer in compression but in tension. This is almost certainly true for large castings in metal
molds, such as steel ingots in cast iron ingot
molds, where the pressure between the mold and
casting is high, the friction is high, and the mold
is rigid. These authors explain the occurrence of
surface cracks in steel ingots in this way. How-
ever, Weiner and Boley (Ref 2) are likely to be

more nearly correct for smaller castings in sand
molds. Here, the interfacial pressure will be less,
and the surface more accommodating, reducing
the restraint due to friction. Thus, their analysis
probably remains the most appropriate for medium-size shaped castings.
As in cases where the casting is constrained
by the mold, removing the casting from the mold
at an early stage would be expected to be normally
beneficial in reducing residual stress. Figure 3
shows the result for iron and a high-strength aluminum alloy (Ref 5). All the observations made
thus far appear to be explainable assuming that
the cause of the development of residual stress
is the interaction of different members of the
casting cooling at different rates.
Prediction of Residual Stresses in

Castings
The strain DL/L due to differential contraction
is determined by the temperature difference, DT,
and the coefficient of thermal expansion of the
alloy, . Since e DL/L DT, r eE,
results in r EDT. The stress therefore depends on the temperature difference between
members. It is also worth noting that the stress
is independent of casting length L.
Further influence of the geometry of the threebar frame casting was found by Von Steiger (Ref
6), who measured the increase of the stress in the
center bar of gray iron castings by increasing the
rigidity of the end cross-members. He found that
a center bar of more than twice the diameter of

the outer bars would suffer a residual stress of
more than 200 MPa (30 ksi), which was sufficient to fracture the bar during cooling.
The stress analysis of castings poses several
problems not seen in more traditional problems
in mechanics. Residual stress formation during
casting is a consequence of various regions of a
geometrically complicated component cooling at
different rates. The stress response is the result
of coupled thermal, microthermal, and stress histories. Stress predictions are strongly influenced
by the thermal and microstructural histories. The
accuracy of thermal and microstructural predictions is a primary factor in the accuracy of residual stress predictions.
Figure 4 schematically represents the types of
analyses required to completely describe a casting process (Ref 7). An overall architecture of a
comprehensive solidification modeling system is
shown in Fig. 5, which depicts the various modules available in the current state-of-the-art solidification simulation of casting processes, the
information available from each module, and the
interconnection between each module (Ref 8).
The primary and most obvious phenomenon
controlling casting is the transfer of heat from
the cooling metal to the mold. The early models
of casting cooling were straightforward heat conduction analyses. However, the mechanics of
fluid flow are important for both mold-filling effects and physics-based models of interdendritic
porosity formation. Buoyancy effects after the
mold is full exert varying degrees of influence
during the cooling cycle, depending on the thickness of the casting being produced. In addition,
analysis of the flow of various chemical species
is very important for crystal growth, and many
thermofluid models today incorporate species
flow.
The heat-transfer processes occurring are
complex, the cooling rates employed range from
105 to 1010 K/s, and the corresponding solidification systems extend from several meters to a
few micrometers. These various cooling rates
produce different microstructures and hence a
variety of thermomechanical properties. Yu et al.
(Ref 9) related the occurrence of casting defects
to cooling rates. During the solidification of binary and multicomponent alloys, the concentrations vary locally from the original mixture as
material may be preferentially incorporated or
rejected at the solidification front. The material
between the solidus and the liquidus temperatures is partly solid and partly liquid (resembling
a porous medium) and is referred to as the
mushy zone. Lewis et al. (Ref 10) have given
an account of several aspects of modeling of heat
transfer, fluid flow, and thermodynamics in castings.
Solidification kinetics, including phase selection, nucleation, and growth, are now being investigated in several laboratories. Incorporation
of these principles into the more traditional thermofluid models promises to enable quantitative
microstructural predictions in the near future
(Ref 9), and predictions of engineering proper-
Residual Stress Formation During Casting / 363

ties such as tensile strength and elongation will
be possible before long. These predictions will
enable product-design engineers to evaluate the
effects of nonuniform properties and defects on
the life-cycle performance of components. Finally, the coupling of mechanical analysis with
thermal analysis enables the prediction of residual stresses and distortions in castings.
Another important problem is the material response to the thermal cycle, as material properties are temperature dependent (Ref 11). For
steels and other materials, dramatic changes in
mechanical properties can occur at temperatures
above 600 C (1110 F). Phase transformations,
with their attendant volume changes, also play
an important role in the development of residual
stresses.
Thermomechanical models were first applied
to modeling of continuous casting and ingot
casting. Elastoplastic (Ref 1214) and elastoviscoplastic (Ref 1517) behaviors have been
considered with thermo-dependent material properties. The effect of different kinds of constitutive equations has been discussed (Ref 18), and
elastoviscoplastic behavior seems to be more appropriate to model the development of strain and
stress in casting and to take into account the different states of metal (liquid, semisolid, solid).
Such models were recently applied to the castings of complex-shape parts (Ref 19, 20).
The focus is currently on the validation of
models: Do they correctly describe reality, and
what must be the accuracy of materials data for
models to be predictive? The great influence of
thermal properties of alloys on calculated solidification time have been pointed out (Ref 21).
Coupling between thermal and mechanical behavior magnifies the complexity of this problem
since not only thermal and mechanical properties
must be accurately known, but the way the coupling is modeled also is of importance.
Chandra (Ref 22) presents the basic concepts
for a comprehensive finite-element-based computer simulation method for the prediction of hot
tears, hot cracks, residual stresses, and distortions in precision castings using a sequential
thermomechanical analysis approach. The existing capabilities of several industry standard
commercial and research finite-element codes
are also reviewed. Drezet et al. (Ref 23) predict
the deformation and temperature field evolution
during direct chill (DC) and electromagnetic
(EM) casting of aluminum alloy slabs using a
transient thermomechanical model based on viscoplastic law. The model, validated on the basis
of the measurements presented earlier in this article, enables prediction of the influence of casting parameters on butt curl and swell, rolling
faces pulling, and residual stress state for DC and
EM cast ingots.
Stefanescu (Ref 24) has made an attempt to
predict features such as microsegregation, microstructure length scale, fraction of phases,
structural transitions, hardness, microhardness,
and tensile properties. Distribution of residual
porosity in castings is calculated based on a uniform solidification assumption for the particular
case of aluminum-rich Al-Cu castings (Ref 25).

An analytical criterion, identifying conditions
under which there will be no porosity formation,
is established. Hiebler et al. (Ref 26) showed that
the strength properties of in situ solidified steel
near the solidus temperature can be readily determined and hence may allow development of
a stress criterion in order to further improve the
assurance of inner soundness even under the
most critical conditions. However, Yamanaka et
al. (Ref 27) found that internal cracking occurs
and extends when the total amount of strain applied between zero strength temperature (ZST)
and zero ductility temperature (ZDT) exceeds
the critical strain, independent of deformation
mode, whether continuous or intermittent.
It is interesting to note that Zhang et al. (Ref
Heat-flow
analysis
Macroporosity (isolated hot spots)

Microporosity (empirical criteria; e.g., Niyama)
Primary dendrite arm spacing (empirical criteria)
Secondary dendrite arm spacing (empirical criteria)
Fluid-flow
analysis
Better representation of the thermal

field during and after mold filling
Darcy flow for interdendritic
porosity models
Solidification
kinetics
Grain size of castings via nucleation

and growth laws
Secondary dendrite arm spacing (coarsening relations)
Solidification path
Species flow
analysis
Prediction of macrosegregation
Prediction of microsegregation
Enhanced representation of the solidification
kinetics, fluid flow, and heat transfer
Thermomechanical
analysis
Fig. 4
28) have presented a fast-acting simultaneous

filling and solidification model based on reduction of Navier-Stokes equations to a transient
Bernoulli equation and potential equation along
with enthalpy method with a one-dimensional
heat conduction model. Ruiz and Khandia (Ref
29) have used LS-DYNA 3D both for the filling
simulation and to carry out a coupled thermal
and stress analysis of casting during solidification, predicting cooling rates, residual stresses,
and as-cast shapes. Magnin et al. (Ref 30) determined the elastoviscoplastic constitutive equation using an optimization model, which calculates the best rheological parameters to fit
experimental stress curves. This law is then introduced in a finite-element model of billet DC
casting using the commercial software package
Physics-based solidification shrinkage

Hot tearing from yield criteria
Prediction of air gap with modification
of heat transfer
Prediction of residual stresses and
casting dimensional control
Types of analysis available for solidification modeling, with their benefits
Initial design
Material
properties
(metal, mold)
Quick analysis
(solidification time,
modulus, etc.)
Rules (empirical
charts, etc.)
Fig. 5
Special boundary
conditions (interface,
heat-transfer
coefficients, etc.)
Rigging design
Solidification
simulation (temperature, solid fraction,
velocity, etc.)
Stress analysis
(shrinkage
stresses
distortion)
Microstructural
evolution (grain
size, dendrite
arm spacing, etc.)
Defect prediction
(macro- and
microporosity,
misrun, etc.)
Mushy zone
fluid flow
(macrosegregation)
Typical architecture of a comprehensive solidification modeling system

MARC. A good idea of product quality and performance can often be obtained at the initial
stage of the product development cycle through
prediction of filling, solidification, stresses, and
shrinkage (Ref 31). Examples of simulations
conducted using an experimental die, an automotive housing, and a wheel in an aluminum
alloy are given.
Finite Element Analysis

An experimental on-site study of the influence
of operating parameters on the quality of castings
produced in a working steel mill would be very
costly and time consuming. It is far more practical to simulate the entire complex process numerically. Any acceptable mathematical model
should be capable of accommodating variations
in the operating parameters as they arise in practice so that their effects on the final product can
be predicted. Another requirement placed on the
model is that it be suitable for designing molds
so that the optimum mold type may be used for
given casting conditions and steel qualities.
Since the entire process is dependent on heat
flow from the ingot to the mold and subsequently
to the surroundings, it is logical to determine the
temperature field at various stages of the process.
From this temperature field, initial strains can be
calculated and the resulting thermal stresses determined.
Since the appearance of the pioneering paper
by Sarjant and Slack (Ref 32), a number of authors have published studies on the development
of mathematical models for calculation of the
solidification process in steel ingots. Although
the solidification of an ingot is a three-dimensional process, it appears to be possible to obtain
an adequate picture of the process using a twodimensional model. It has been demonstrated
(Ref 33) that for the ingot dimensions used in
actual practice, a true picture of the solidification
process can be obtained by calculation in a horizontal section at midheight through the ingot.
The thermal model used here uses the Galerkin method (Ref 34) and eight-noded quadratic
isoparametric elements and possesses the novel
feature of using elements with time-varying conductivity to model the heat transfer in the air-gap
that forms between ingot and mold. Calculated
temperature fields may be used to evaluate the
loading stresses, and in this way the thermal
stress development may be determined using either an elastic formulation (Ref 35) or an elastoplastic formulation (Ref 36). The elastoplastic
formulations have the disadvantage of being unable to model any time-dependent creep effects.
The mathematical stress model in this article
embodies a general solution procedure for determining the development of thermal stresses in
an elastoviscoplastic multiphase body and is capable of accounting for time-dependent properties. The constitutive model used is of the type
proposed by Perzyna (Ref 37) and first implemented numerically by Zienkiewicz and Cormeau (Ref 38). In this model the behavior of the
material is elastic if a certain function of the

yield condition has a value of less than zero,
while time-dependent viscoplastic flow occurs
when the value of this function is positive. The
model can also be used to give the purely plastic
solution, in which time and viscous effects do
not have real meaning but are used merely as
means for attaining steady-state condition.
The temperature field and the development of
the thermal stress in a section at the midheight
of an ingot casting are calculated and shown.
Finite-Element Formulation of the

Heat-Flow Problem (Ref 39)
The variation of temperature T with time in a
two-dimensional region X bounded by a curve
C, for both ingot and the mold, is given by:
qc
T

T

T
kx
ky
t
x
x
y
y
(Eq 1)
q12
Ni Ti(t)
i1
KT CT F
Ni Nj
Ni Nj
Ky
x x
y y
e X
dX
where T3 is the temperature of the outer wall of

the mold and T is the ambient temperature.
The heat-transfer coefficient h12 changes when
the air gap between the ingot and the mold
forms. To take this phenomenon into account,
h12 is regarded as a function of position on the
ingot perimeter and of time. Oeters et al. (Ref
40) have observed that the calculations are in
good agreement with the actually measured heatflow densities in the gap if the breadth of the gap
is taken to be of the order of 0.5 mm (0.02 in.).
Thus, in the present model, the heat transfer
across the gap is modeled as a conduction mode,
with the conductivity being a function of time
and position on the perimeter. This approach allows the effect of the lubrication used on the
inner wall of the mold (Ref 40) to be modeled
by using a suitable value of the thermal conductivity in the air-gap region. The heat-transfer coefficient between the outer wall of the mold and
the surrounding air takes into account heat trans-
(Eq 5)
in which typical matrix elements are
(Eq 2)
(Eq 3)
(Eq 4)
where Ti is the time-dependent value at node i.

The substitution of Eq 4 into Eq 1 and the application of the Galerkin method (Ref 34) results in
Kij
where T1 is the surface temperature of the ingot

and T2 is the temperature of the inner wall of the
mold.
(2) Between the outer wall of the mold and
the surroundings:
q3 h3 (T3 T)
where qc is the thermal capacity, and kx, and ky

are the thermal conductivities in the x and y directions, respectively.
The boundary conditions for this partial differential equation are the expressions for the heat
fluxes between the ingot and mold and between
the mold and the surroundings. These boundaries
occur in two regions:
(1) Between the ingot and inner wall of the
mold:
k12
h12(T1 T2)
(T1 T2)
d12
fer by radiation and by free convection. The variation of density and specific heat with temperature is handled as discussed by Comini et al. (Ref
41), and any phase change that occurs is treated
by the method discussed by Morgan et al. (Ref
42). The thermal conductivities of the ingot and
mold are also allowed to vary with temperature.
The temperature fields in the ingot and the mold
are computed simultaneously using a finiteelement formulation. The region of interest is divided into a number of eight-noded isoparametric elements Xe, with boundary Ce with
quadratic shape functions Ni associated with
each node i. In the isoparametric elements, the
shape functions are used to transform the coordinates (Ref 34), and this enables better representation of any curved boundaries that may be
present in the problem. The unknown temperature T is approximated in the solution domain at
any time by:
e C hNi Nj dC
e
Cij
e X
Fi
e C
qcNi Nj dX
Ni hTdC
(Eq 6)
(Eq 7)
(Eq 8)
where the summations are carried out over each

element e.
The coupled system of ordinary differential
equations represented by Eq 5 is solved by a finite-difference two-stage Crank-Nicolson predictor-corrector method, and in this way the
complete thermal history of the region can be
determined. The change in temperature in a certain time can then be used to calculate incremental initial node strains dei via
DTi
dei DTi
0
(Eq 9)
where is the temperature-dependent coefficient

of expansion. An elastoviscoplastic stress model
is then used to calculate the stress distribution
resulting from the application of this initial
strain.

Finite-Element Formulation of the
Elastoviscoplastic Stress Model (Ref 39)
In the elastoviscoplastic stress model, it is assumed that the total strain is the sum of elastic
and plastic components, together with any initial
strains:
e ee evp ei
(Eq 10)
Only elastic strains are produced initially by the

application of a load, and the elastic strain rate
is linearly related to the total stress rate by the
matrix of elastic constants D (Ref 43) according
to:
r Dee
(Eq 11)
where the overdot denotes differentiation with

respect to time. The viscoplastic strain occurs
only if the stress levels exceed some previously
defined yield stress. This yield stress is given by
the yield function:
e Bd
(Eq 12)
Therefore, when F 0, purely elastic behavior

takes place, but F 0 represents the onset of
plastic deformation. The viscoplastic strain is
given by:
evp
e vp f (r, evp)
t
BTrdX
(Eq 19)
Combining Eq 10, 11, and 19 then leads to:
BTD(e e vp e i)dX 0
(Eq 20)
which, on using Eq 17 to substitute for e, becomes:
BTDeidX
BTDevpdX
(Eq 23)
The solution procedure for Eq 22 is given by

Zienkiewicz and Cormeau (Ref 38). In this
method, a simple Euler iterative procedure is
used which can become unstable if critical time
steps are too large. The selection of critical time
steps has been studied by Cormeau (Ref 44), and
his stability limits are followed here.
(Eq 18)
As the constitutive relation for viscoplastic problems has been specified in time rate form in (Eq
11), it is convenient to rewrite Eq 18 as
Examples
BTrdX 0
BTDBddX
0
(Eq 21)
That is,
(Eq 13)
Ksd R 0
To define completely a strain-rate law, it is

assumed that, in common with plasticity, the directions of straining are given by gradients of a
plastic potential Q:
Q
r
(Eq 17)
where B is the standard matrix derived from the

derivatives of the shape functions (Ref 43). The
virtual work principle in this case enables the
equilibrium equation to be written as:
F (r, evp) 0
e vp w(F)
where d is the column vector of nodal values di.

With the displacements at all points given by Eq
16, the strains at any point can be determined
from the relationship:
(Eq 22)
where KS is the overall stiffness matrix and R

represents the total loading rate, which turns out
to be
Because of the complexities and practical difficulties of experiments on ingot castings, only
a limited number of experimental investigations
are reported in sufficient detail in the literature
to make possible a comparison with the results
of a mathematical model. Oeters et al. (Ref 40)
have reported extensive results of the investigation on a 6 ton ingot casting, and this is the example chosen for the present investigation.
At the midheight of the 6 ton ingot, the cross
section is in the form of a square with rounded
corners and a side length of 600 mm (23.6 in.).
Along two opposite sides of the square the surrounding mold has a thickness of 150 mm (6 in.),
while this thickness becomes 166 mm (6.5 in.)
along the other sides. The variation of the properties with temperature is assumed to take the
form reported by Williams et al. (Ref 11). The
finite-element mesh used in the analysis consists
of 52 elements (illustrated in Fig. 6) and 185
nodes. The nodes are located at the element corners and at the midpoints of the element sides.
The time of air-gap formation at various locations on the ingot perimeter follows roughly a
parabolic law proceeding from the corner to the
middle of the face (Fig. 7). Initially, the model
(Eq 14)
8
16
24
32
46
52
23
31
45
51
21
30
40
29
28 36
44
50
43
49
42
48
41
47
Viscoplastic flow laws can then be written as:

w(F )
Mold
0 if F 0
w(F) if F 0
Combining Eq 10, 11, and 13 results in the

complete constitutive relation:
e D1r w(F)
Q
e i
r
Air
gap
6
4
15
14
12
22
13
27
3
(Eq 15)
11
(Eq 16)
35
39
26
2
10
18
17
34
38
25
Ni di NTd
i1
19
Ingot
The finite-element stress analysis is performed

only on the ingot, which is discretized by using
the same elements and shape functions as were
used in the thermal analysis. If the vector u denotes the displacement field, then this may be
approximated over in terms of the nodal displacements di by:
u u
20
Fig. 6
Finite-element mesh for ingot, air gap, and mesh
33
37

1000
Center of the side
800
Quarter of the side
Temperature, C
is run with intimate contact everywhere, and

then the gap opens from the corner toward the
center as time proceeds. In the gap, the heattransfer coefficients given by Oeters et al (Ref
40) are utilized as functions of time in the form
shown in Fig. 8. Since the gap width is fixed at
0.5 mm (0.02 in.), this variation is supplied to
the model as a variation in thermal conductivity
of the air gap. At the start, the time steps need
to be very small, on the order of seconds, and
they gradually increase up to a value of 2 min in
the later stages of the simulation run.
The calculated temperature of the inner wall
of the mold for various locations is plotted
against time in Fig. 9. For the sake of comparison, the results of Oeters et al. (Ref 40) are also
shown. As the exact thermal properties used by
Oeters et al. are not known, it can be concluded
that a reasonable agreement has been obtained.
Figure 10 compares surface temperatures on the
ingot as functions of time and location; the predicted temperatures are again slightly lower than
those of Oeters et al. Figure 11 shows the solidification fronts determined at various time intervals; these results are in good agreement with
the values determined by Weingert (Ref 45).
The temperature field in the ingot is used for
evaluating the thermal loads employed in the
600
Corner
400
Experimental
Present results
200
20
10
30
40
Fig. 9
80
90
60
70
80
90
Experimental
Present results
Temperature, C
Time, min
70
1200
1100
20
60
Temperature curve for inner wall of mold
Center of the side
40
30
50
Time, min
1000
Quarter of the side

900
10
Corner
0
150
800
300
20
10
30
40
Distance from corner
Heat transfer coefficients, cal/cm2min C
Fig. 7
Time for air-gap formation
Fig. 10
Average value of ingot surface temperature
viscoplastic stress analysis. A plane-strain stress

analysis was performed using a Von Mises yield
criterion (Ref 43). The model is used to give a
purely plastic solution at steady-state conditions
at each stage. The rate of viscoplastic straining
is given by Eq 14, in which the flow function
used is:
3.0
Center of face
2.4
1.8
1.2
Quarter of face
18
30
0.6
74
80
0
20
40
60
80
Time, min
Fig. 8
F
Fo
(Eq 24)
52
Corner
0
50
Time, min
Average heat-transfer coefficients in the air gap
Fig. 11
Solidification fronts for one quadrant of the ingot
where Fo is some reference value of stress and c

is a fluidity coefficient. The mechanical properties vary as a function of temperature and are
taken from Williams et al (Ref 11). A Poissons
ratio of 0.33 is used.

Figures 12 and 13 show the computed stress
field in the ingot at two different times. The region near the ingot surface is subjected to tension
initially and, as expected, this gradually changes
to compression with the passage of time. Prediction of such stress concentration regions would
be particularly useful in the case of complex
shapes to avoid possible failures.
Next, examples of the calculation of residual
stresses for pure aluminum under axially symmetric and realistic boundary conditions are
given. The effects of the melt pressure and cooling conditions on the residual stresses are also
reported.
Molten pure aluminum at a uniform temperature equal to or above the melting temperature is
assumed to start at time t 0 when part of the
boundary of the body is cooled down in an axially symmetric way to a temperature equal to or
below the solidification temperature. Figure 14
shows on the r-z plane, at time t, the two-dimensional model sections. The interface velocity and
location are treated as primary variables of the
heat-transfer analysis, and the isostatic stress
condition at the front is utilized as an initial condition in the stress analysis. The freezing interface condition as an initial rather than a bound-
Fig. 14
Tension
Compression
Fig. 12
One quadrant of the principal stress field after

4 min
Model section for the solidification problem
ary condition at the time of solidification of a

material print was first discussed by Richmond
(Ref 46) and implemented in one-dimensional
solidification problems by Richmond et al. (Ref
4, 47) and in two-dimensional plane stress applications by Zabaras et al (Ref 48).
The thermal properties of pure aluminum are
given in Table 1, and the temperature-dependent
mechanical properties reported by Heinlein et al.
(Ref 47) are given in Table 2. A viscoplastic constitutive model is needed to prescribe the inelastic deformation, which should include important
effects such as rate sensitivity, strain hardening,
and recovery in a rather wide range of temperatures up to the melting point of the solidifying
pure metal. In the simulation reported here, a
hyperbolic-sine constitutive law is used to prescribe the inelastic deformation. The details of
the formulations for thermal and thermomechanical analysis are given by Zabaras et al. (Ref 49).
The examples that follow will consider solidification of a cylinder initially filled with liquid
aluminum at melting temperature. Part of the
boundary of the cylinder will be assumed to be
cooled down as follows:
T0 (t) Ta (Tm Ta) eQt
Tension
Compression
Fig. 13
Principal stress field after 40 min
(Eq 25)
where Ta is the final steady-state temperature, Tm

is the melting temperature, and Q is a cooling
rate parameter. In the following examples, Ta
500 C (930 F), Q 0.2 s1, and Tm 660
C (1220 F). Four-noded elements were used for
the temperature analysis, while eight-noded
Table 1
Thermal properties of aluminum
Property
Value
Thermal conductivity in solid, kS

Thermal conductivity in liquid, kL
Heat capacity in solid, CS
Heat capacity in liquid, CL
Latent heat, L
Density, q
Initial temperature, Tin
Melting temperature, Tm
0.0548 kcal/msC
0.0548 kcal/msC
0.2526 kcal/kgC
0.2526 kcal/kgC
94.44 kcal/kg
2650 kg/m3
660 C (1220 F)
660 C (1220 F)
Source: Ref 46
Table 2
num
Mechanical properties of alumi-
Coefficients of constitutive law, eN 3/2 AeC/(T273)

[sin h Br]
n]/r s
A
0.382 1012 /s
B
0.037/MPa
C
18,849 K
n
3.84
Thermal expansion coefficients, (T)n(a), m/m C 106
At 25 C
23.19
At 300 C
27.86
At 400 C
30.23
At 660 C
38.355
Poissons ratio (m)
0.37
Youngs modulus
E(T) F GT, where F 6.93
104 and G 43.7152 MPa / C.
(a) n: The variation of (T) is assumed to be piecewise linear within the
temperature intervals 25 to 300 C, 300 to 400 C, and 400 to 660 C.
Source: Ref 46

quadrilateral elements were employed in the deformation part of the problem.
In the first example, it is assumed that the cylindrical body is insulated at the top and bottom
and that the rest of the surface (r R) is cooled
with the temperature history depicted in Eq 24.
Plane-strain conditions with traction-free outer
surface were assumed. The geometrical parameters were assumed as h 0.04 m and R
0.018 m. The front position and the temperature
history at various locations are shown in Fig. 15
and 16, respectively. These results were compared with a one-dimensional deforming finiteelement implementation of the problem (Ref 50)
and were found to coincide to within plotting
accuracy. Figure 17 shows the residual stress
distribution with the plane-strain assumptions on
the plane z 0.002113 m at the end of solidification, t 8.32 s. The stress history at the
location r 0.017714 m and z 0.002113 m
near the bottom and the surface r R is also
shown in Fig. 18. As shown, the hoop stress at
the outer surface r R is compressive, while
near the center of the cylinder all the stresses are
tensile. Generally, the residual stress in the axial
direction is tensile, but it becomes slightly com-
pressive in the region close to the surface r R

near the end of solidification. In this example, as
expected, the stresses were found not to significantly vary in the axial direction.
In the second example, the geometry and the
temperature boundary conditions are the same as
before. It is assumed that the bottom is fixed in
the axial direction and the traction-free conditions are applied to the top surface and to the
outer surface, r R. It was observed that the
traction conditions at the top surface affect primarily the axial stress while leaving the radial
and hoop stress almost the same for both freetop and plane-strain conditions (Ref 51). Large
tensile residual axial stresses appear in the region
close to the center of the cylinder. Figure 19
shows the residual stress variation in the axial
direction at time t 8.3202 s, and the stress
history at location r 0.0177145 m and z
0.0010566 m (close to the bottom and r R) is
given in Fig. 20. The hoop and axial stress histories are almost the same, an assumption used
by Heinlein et al. (Ref 47) to simplify a threedimensional problem to a unidirectional one.
To demonstrate the effect of melt pressure, a

longer cylinder is considered (h 0.4 m), and
other conditions are kept the same as those in the
first example. In this case, the pressure at the
bottom of the cylinder is about 0.01 MPa (0.0014
ksi). The displacements of three points at r R
are given in Fig. 21, where it is shown that the
surface r R near the bottom of the cylinder
first expands before it starts contracting. Generally, the residual stress distribution is very close
to that discussed in the first example; but early
on, the differences of the displacements at r
R and the stress distribution between this and the
first example can be observed. Also, the hoop
stress was found to have higher values early on
(Ref 51). Therefore, as is expected, one can conclude that the melt pressure has a significant effect on the stress early in the process and can
play an important role in the location and time
of formation of air-gaps in the solid shell/mold
interface. For cases with larger Q (high cooling
rate), it was observed that the solidification process proceeds faster and that the calculated
stresses obtained at the end of solidification are
larger than those in the first example (Ref 51).
Stress, MPa
Stress, MPa
Radial front position, m
0.01
2
0
0
Fig. 15
6
Time, s
0.01
r, m
Radial stress
Axial stress
Hoop stress
r = 0.017714 m
z = 0.0021133 m
0.02
10
Fig. 17
Front position for solidification of a cylindrical

body uniformly cooled at the surface at r R
Residual stress distribution in the radial direction at t 8.32 s using a plane-strain assump-
tion
Fig. 18
Thermal stress history with a plane-strain assumption
660
640
Radial stress
Axial stress
Hoop stress
r = 0.002918 m
r = 0.002918 m
r = 0.002918 m
Stress, MPa
620
600
580
6
Time, s
2
1
r = 0.017324 m
r = 0.018000 m
r = 0.014731 m
r = 0.009685 m
r = 0.006980 m
560
r = 0.017324 m
1
8
10
r = 0.017324 m
2
0
Temperature history at various locations for solidification of a cylindrical body uniformly

cooled at the surface at r R
10
Time, s
680
Temperature, C
Radial stress
Axial stress
Hoop stress
z = 0.0021133 m
0.02
Fig. 16
Fig. 19
0.01
0.02
Z, m
0.03
0.04
Residual stress variation in the axial direction at time t 8.3202 s with a traction-free top surface

1
Stress, MPa
1
Radial stress
Axial stress
Hoop stress
r = 0.0177145 m
z = 0.0010566 m
2
3
0
10
Time, s
Fig. 20
Thermal stress history with a traction-free top

surface
In the final example, a cylinder with R h

0.018 m was cooled with the condition of Eq
24 at both the bottom surface and the surface r
R. The pattern of solidification is shown in
Fig. 14. It is assumed that the outer surface, r
R, and the top surface are traction-free, while the
bottom is fixed in the axial direction. Principal
residual stresses in the r-z plane near the end of
solidification are plotted in Fig. 22. Large tensile
residual stresses appeared at the top around the
center region, the stresses were small in the area
close to the surface r R, and the residual
stresses were compressive at the bottom close to
the center region. The effect of casting size and
thermal boundary conditions on residual stresses
and deformation has been investigated by Sathya
Prasad (Ref 52).
0.02
0
Displacement, mm
0.02
0.04
0.06
Radial displacement z = 0.00 m
0.08

0.10
Axial displacement z = 0.00 m, z = 0.40 m

Axial displacement z = 0.20 m
0.12
0
10
Time, s
Fig. 21
Radial and axial displacements at the surface r R with plane-strain assumption and large melt pressures
0.0011
Tension
Compression
1.5 MPa
z, m
0.0010
0.009
0
0
0.009
r, m
Fig. 22
Principal stresses in the r-z plane at time t 7.3294 s for the solidification problem
0.018
Special Considerations
The deformation of a solidifying material is
very different from that of a standard fixed body.
A solidifying body develops residual (initial)
stresses immediately after solidification and is
never in a state of zero stresses (stress free).
Thermal stress problems carry with them difficulties not normally found in the analysis of either thermal or stress problems. The coupling between the temperature and stress fields works in
both directions. It is possible to imagine cases
where the basic boundary conditions for the thermal analysis are affected by the deformations, as
in the formation of the air gap that controls the
heat flow to the mold from the cooling ingot.
Fully coupled analyses are slightly more complicated because they require deciding when to
update the effect of one process to the other. Decultieux et al. (Ref 53) apply a three-dimensional
finite-element coupled thermomechanical model
to the solidification of the hollow cylinder. An
experiment has been developed to measure heattransfer coefficients and air-gap width in permanent mold casting of aluminum-silicon alloys.
Comparison between experimental and calculated temperature and air-gap width shows the
validity of the coupled approach.
The thermal strain fields computed from the
temperature fields introduce complications. The
incompressible nature of plastic deformation
creates a constraint at each gauss point in an element. When the number of constraints arising
from incompressibility exceeds the numbers of
degrees of freedom, locking is said to occur, as
there are no possible solutions for this case. The
solution to this may be the reduced order of the
integration for the hydraulic components of the
stress, which may lead to, most notably, a failure
to satisfy the patch test (Ref 54). A related effect
comes from the order of the thermal strain fields
within the element. If nodal temperatures are interpolated to give gauss point values, and these
are used to determine thermal strains, then the
thermal strain field has the same order as the
displacement field in the element. The total
strains, which are computed from the partial derivatives of the displacements, are one order
lower. This incompatibility can also lead to locking problems. The need to avoid incompatible
strain fields has been known for some time (Ref
55) and extremely large errors can result if it is
not observed (Ref 56). Using linear elements in
the thermal analysis and quadratic elements in
the stress analysis is the most common strategy
(Ref 49).
Plane-strain conditions are most commonly
used in the stress analysis. This condition implies that the longitudinal heat flow and longitudinal displacement are zero. Longitudinal interactions in the stress may not always be
satisfied even though longitudinal heat flow is
probably never very large. To compensate for
this problem, extension to a generalized plane
strain has been used (Ref 54) where plane sections must remain plane but may rotate or trans-

late with respect to one another, in which case
net longitudinal stress on any cross section now
becomes zero.
One of the main purposes of solidification
modeling is to identify casting designs that are
likely to cause defects due to porosity formed
when shrinkage occurs during phase change.
This porosity, which frequently takes the form
of interdendritic voids, affects the ultimate tensile strength of the casting and hence is to be
avoided. It can be eliminated only when the liquid metal from the riser is able to travel to the
hot spot location to compensate for the solidification. Experiments on cast steels have shown
that when the temperature gradient at the end of
the solidification falls to below 10 K/cm, porosity is observed even in well-degassed material.
On the basis of solidification simulation, the solidification time can be found and the temperature gradients at that instant can be determined.
Based on the contour of 10 K/cm thermal gradient, it is possible to determine the location and
size of the shrinkage defect. However, it may be
noted that the value of 10 K/cm is valid for a
three-dimensional casting, and a different limit
may have to be taken for two-dimensional simulations. Sathya Prasad (Ref 52) has predicted
the size and location of shrinkage cavities for Lshaped and T-shaped castings with and without
chills. Residual stress determination must also
make sure that no shrinkage cavity exists in the
castings. If it does, its presence must be accounted for, as there may be stress concentration
around such a cavity.
In the case of continuous casting of steel, compared with ingot casting, several additional difficulties are encountered when predicting the residual stresses. High stresses develop in the
solidifying shell as a result of a number of forces
acting externally and internally on the strand.
The values of heat extraction and solidification
are relatively slow for static casting (solidification time measured in hours), whereas solidification time for continuous casting is in minutes.
A static casting meets with a single cooling environment over its solidification period, while a
continuous cast section encounters different environmentsmold, sprays, pinch rolls, radiationbefore complete solidification takes place.
Thus, continuous castings change rapidly from
one zone to another, resulting in higher thermal
stresses than in ingot castings. A continuous cast
section also is stressed by pinch rolls, bending
and straightening operations during solidification, mold oscillation, and misalignment of mold
and roller cages. Ferrostatic pressure can also
produce sufficient stress to cause bulging across
the wide faces of large slabs.
ACKNOWLEDGEMENTS
The material presented here is collected from
several papers apart from the work of the authors, who gratefully acknowledge the efforts of
these fellow researchers. The authors especially
thank their students Habib, Paragasam, Parthi-
ban, Revathy, Rushyendran, Sanjay, Vyasa, and

Yun Hung for their generous help in putting together this article.
REFERENCES
1. J. Campbell, Castings, Butterworth Heinemann, 1991
2. J.H. Weiner and B.A. Boley, Elasto Plastic
Thermal Stresses in a Solidifying Body, J.
Mech. Phys. Solids, Vol 11, 1963, p 145
154
3. O. Richmond and R.H. Tien, J. Mech. Phys.
Solids, Vol 19, 1971, p 273284
4. R.H. Tien and O. Richmond, Theory of
Maximum Tensile Stresses in the Solidifying Shell of a Contained Rectangular Casting, J. Appl. Mech. (Trans. ASME), Vol 49,
1982, p 481486
5. R.D. Dodd, Ph.D. thesis, University of Birmingham, UK, 1950
6. R. Von Steiger, Stahl Eisen, Vol 33, 1913,
p 1442
7. T. Overfelt, The Manufacturing of Solidification Modeling, J. Met., Vol 44 (No. 6),
1992, p 1720
8. G. Upadhya and A.J. Paul, Solidification
Modeling: A Phenomenological Review,
AFS Trans., 1994, p 6980
9. K.O. Yu, J.J Nichols, and M. Robinson, Finite Element Thermal Modeling of Casting
Micro Structures and Defects, J. Met., Vol
44 (No. 6), 1992, p 2125
10. R.W Lewis, K.N. Seetharamu, and B. Prasad, Modeling of Heat Transfer, Fluid Flow
and Thermodynamics in Castings, Modeling
in Welding, Hot Powder Forming and Casting, L. Karlsson, Ed., ASM International,
1997, p 161273
11. J.R. Williams, R.W. Lewis, and K. Morgan,
An Elasto-Viscoplastic Thermal Stress
Model with Application to the Continuous
Casting Metals, Int. J. Num. Meth. Eng., Vol
14, 1979, p 19
12. A. Grill, J.K. Brimacombe, and F. Weinberg, Mathematical Analysis of Stresses in
Continuous Casting of Steel, Ironmaking
Steelmaking, Vol 1, 1976, p 3847
13. J.E. Kelly, K.P. Michalek, T.G. OConnor,
B.G. Thomas, and J.A. Dantzig, Initial Development of Thermal and Stresses Fields
in Continuous Cast Steel Billets, Metall.
Trans., Vol 19A, 1988, p 15892602
14. J. Guan and P.R. Sahm, Numerical Investigations of Thermal Stresses in Real 3D
Structural Casting, Gisserei, Vol 8, 1992, p
318322
15. H.G. Fjaer and A. Mo, ALSPEN, A Mathematical Model for Thermal Stresses in Direct Chill Casting of Aluminum Billets, Metall. Trans., Vol 21B, 1990, p 10491061
16. T. Inoue and D.Y. Ju, Simulation of Solidification and Viscoplastic Stresses during
Vertical Semi-Continuous Direct Chill
Casting of Aluminum Alloy, Int. J. Plast.,
Vol 8, 1992
17. B.G. Thomas, I.V. Samarasekara, and J.K.

Brimacombe, Mathematical Model of the
Thermal Processing of Steel Ingots, Metall.
Trans., Vol 19B, 1988, p 119147
18. P.F. Kowlowski, B.G. Thomas, J.A. Azzi,
and H. Wang, Simple Constitutive Equations for Steel at High Temperatures, Metall. Trans., Vol 23A, 1992, p 903918
19. A. Shapiro, W. Stein, and P. Raboin, Casting Process Modeling Using ProCAST and
CAST2D, in Modeling of Casting, Welding
and Advanced Solidification Processes, Vol
VI, T.S. Piwonka, V. Voller, and L. Katgerman, Ed., TMS, 1993, p 493500
20. M. Bellet, M. Menai, and F. Bay, Finite Element Modeling of the Cooling Phase in
Casting Processes, in Modeling of Casting,
Welding and Advanced Solidification Processes, Vol VI, T.S. Piwonka, V. Voller, and
L. Katgerman, Ed., TMS, 1993, p 561568
21. T. Overfelt, Sensitivity of a Steel Plate Solidification Model to Uncertainties in Thermophysical Properties, in Modeling of Casting, Welding and Advanced Solidification
Processes, Vol VI, T.S. Piwonka, V. Voller,
and L. Katgerman, Ed., TMS, 1993, p 663
670
22. V. Chandra, Computer Predictions of Hot
Tears, Hot Cracks, Residual Stresses and
Distortions in Precision Castings: Basic
Concepts and Approach, Proc. 124th TMS
Annual Meeting, 1995, p 11071117
23. J.M. Drezet, M. Roppaz, and Y. Krahenbuhl, Thermomechanical Effects during Direct Chill and Electromagnetic Castings of
Aluminum Alloys, Part II: Numerical Solution, Proc. 124th TMS Annual Meeting,
1995, p 941950
24. D.M Stefanescu, Methodologies for Modeling of Solidification, Microstructure and
Stress Capabilities, ISIJ Int., Vol 35 (No. 6),
1995, p 637650
25. A.V. Kuznetsov and K. Vafai, Development
and Investigation of Three Phase Model of
the Mushy-Zone for Analysis of Porosity
Formation in Solidifying Castings, Int. Heat
Mass Transfer, Vol 38 (No. 14), 1995, p
25572567
26. H. Hiebler, J. Zirngast, and C. Bernhard, Inner Crack Formation in Continuous Castings: Stress or Strain Criterion, Proc. 77th
Steelmaking Conf., Vol 77, 1994, p 405
416
27. A. Yamanaka, K. Nakajima, and K. Okamuza, Critical Strain for Internal Crack Formation in Continuous Casting, Ironmaking
Steelmaking, Vol 22 (No. 6), 1995, p 508
512
28. Y.F. Zhang, H.P. Wang, and W.K Liu, Fast
Acting Simulation of Simultaneous Filling
and Solidification, in Transport Phenomenon in Solidification, ASME, HTDV 284,
1994, p 215216
29. D.J. Ruiz and Y. Khandia, Filling and Solidification with Coupled Heat Transfer and
Stress Analysis, Proc. 7th Conf. Modeling
30.
31.
32.
33.
34.
35.
36.
37.
38.
of Casting, Welding and Advanced Solidification Processes, TMS, 1995, p 9911006

B. Magnin, L. Katgerman, and B. Hannart,
Physical and Numerical Modeling of Thermal Stress Generation during DC Casting of
Aluminum Alloys, Proc. 7th Conf. Modeling of Casting, Welding and Advanced Solidification Processes, TMS, 1995, p 303
310
J.F. Hetu, D.M Gao, K.K Kabanemi, S. Bergeron, K.T. Nguyen, and C.A. Loong, Numerical Modeling of Casting Processes,
Adv. Perf. Mater., Vol 5 (No. 122), 1998, p
6582
R.J. Sarjant and M.R. Slack, International
Temperature Distribution on the Cooling
and Reheating of Steel Ingots, J. Iron Steel
Ins., 1954, p 428444
R. Sevrin, Mathematical Study of the Thermal Evolution of Steel Ingots, in Mathematical Models in Metallurgical Process
Developments, Iron and Steel Institute, London, 1970, p 147157
R.W. Lewis, K. Morgan, H.R. Thomas, and
K.N. Seetharamu, The Finite Element Methods in Heat Transfer Analysis, John Wiley
& Sons, 1996
R.W. Lewis and B.R. Bass, The Determination of Temperature and Stresses in Cooling Bodies by Finite Elements, J. Heat
Transfer (Trans. ASME),Vol 98, 1976, p
478484
R.W. Lewis, K. Morgan, and R.H. Gallagher, Finite Element Analysis of Solidification and Welding Processes, PVP-PB026, ASME, 1977, p 6780
P. Perzyna, Fundamental Problems in
Visco-Plasticity, Adv. Appl. Mech., Vol 9,
1966, p 243377
O.C. Zienkiewicz and I.C. Cormeau, ViscoPlasticity, Plasticity and Creep in Elastic
Solids: A Unified Numerical Solution, Int.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
J. Num. Meth. Eng., Vol 8, 1974, p 821

845
K. Morgan, R.W. Lewis, and K.N. Seetharamu, Modeling Heat Flow and Thermal
Stress in Ingot Casting, Simulation, 1981, p
5563
F. Oeters, K. Ruttiger and H.J. Selenz, Heat
Transfer in Ingot Pouring, Information
Symp. Casting and Solidification of Steel,
Luxemborg, Vol 1, 1977, p 126167
G. Comini, S. del Guidice, R.W. Lewis, and
O.C. Zienkiewicz, Finite Element Solution
of Non-Linear Heat Conduction Problems
with Special Reference to Phase Change,
Int. J. Num. Meth. Eng., Vol 9, 1974, p 109
127
K. Morgan, R.W. Lewis, and O.C. Zienkiewicz, An Improved Algorithm for Heat
Conduction Problems with Phase Change,
Int. J. Num. Meth. Eng., Vol 12, 1978, p
11911195
O.C. Zienkiewicz, The Finite Element
Method, McGraw-Hill, London, 1977
I.C. Cormeau, Numerical Stability in QuasiStatic Elasto-Viscoplasticity, Int. J. Num.
Meth. Eng., Vol 9, 1975, p 109127
H. Weingert, Untersuchungen Uberder
Temperatur-Und-Erstarrungsablanf,
Schwerer Blocke Und Brammen, Ph.D.
dissertation, Clausthal, West Germany,
1968
O. Richmond, Models of Stresses and Deformations in Solidifying Bodies, Modeling
of Casting and Welding Processes, AIME,
1982, p 215222
M. Heinlein, S. Mukherjee, and O. Richmond, A Boundary Element Method Analysis of Temperature Fields and Stresses during Solidification, Ironmaking Steelmaking,
Vol 59, 1986, p 5981
N. Zabaras, Y. Ruan, and O. Richmond,
Front Tracking Thermomechanical Model
for Hypoelastic-Viscoplatic Behavior in a
49.
50.
51.
52.
53.
54.
55.
56.
Solidifying Body, Comput. Meth. Appl.

Mech. Eng., Vol 81, 1990, p 333364
N. Zabaras, Y. Ruan, and O. Richmond, On
the Calculations of Deformations and
Stresses during Axially Symmetric Solidification, J. Appl. Mech., (Trans. ASME), Vol
58, 1991, p 865871
N. Zabaras and Y. Ruan, A Deforming Finite Element Analysis of Inverse Stefan
Problems, Int. J. Num. Meth. Eng., Vol 28,
1989, p 295313
N. Zabaras and O. Richmond, Analysis and
Finite Element Approximations of Deformation and Thermal Stresses in Solidifying
Bodies, in Computer Modeling and Simulation of Manufacturing Processes, B.
Singh and Y.T. Im, Ed., ASME, 1990
B. Sathya Prasad, Solidification Simulations of Castings Using Finite Element
Method, Ph.D. thesis, Indian Institute
Technology, Madras, India, 1999
F. Decultieux, M. Menai, F. Bay, C. Levaillant, P. Schmidt, I.L. Svensson, and M.
Bellet, Thermomechanical Modeling in
Casting with Experimental Validation, in
Modeling in Welding, Hot Powder Forming
and Casting, L. Karlsson, Ed., ASM International, 1997, p 291313
A.S. Oddy and L.E. Lindgren, Mechanical
Modeling and Residual Stresses, in Modeling in Welding, Hot Powder Forming and
Casting, L. Karlsson, Ed., ASM International, 1997, p 3159
B.A.B. Anderson, Thermal Stresses in a
Submerged-Arc Welding Joint Considering
Phase Transformations, J. Eng. Mat. Technol., (Trans. ASME), Vol 100, 1978, p 356
362
A.S. Oddy, J.M.J. McDill, and J.A. Goldak,
Consistent Strain Fields in 3D Finite Element Analysis of Welds, J. Pressure Vessel
Technol., (Trans. ASME), Vol 112 (No. 3),
1990, p 309311

DOI: 10.1361/hrsd2002p372

Residual Stress Formation during Casting:

Continuous and Centrifugal Casting
Processes
D.-Y. Ju, Saitama Institute of Technology, Saitama, Japan
DETERMINATION
OF
RESIDUAL
STRESS in manufacturing processes incorporating solidification, such as casting, welding, and
so on, is important for evaluating the mechanical
properties and strength of materials and to optimize manufacturing conditions (Ref 14). However, residual-stress formation depends not only
on the thermomechanical behavior and processing effect due to the variation of microstructure
and macrostructure during manufacturing of materials, but also on the interaction of heat conduction, change of phase transformation due to
solidification, and stress/strain. A unified constitutive model (Ref 5, 6) that describes variation
of the mechanical behavior from viscous fluid to
solid and the mixture domain in the material due
to solidification is necessary. This article presents some developments in thermomechanical
theory and numerical analysis method incorporating solidification of material to simulate the
residual-stress formation during casting. A unified inelastic constitutive relationship capable of
describing both elastic-viscoplastic solids and
viscous fluids to simulate the casting process is
proposed and verified by experimental and numerical results. A proposal based on the finiteelement method to combine temperature and
stress fields as well as deformation during solidification is also presented. The mechanism of
residual-stress formation during continuous and
centrifugal casting can be represented using
simulation. The thermomechanical modeling
was also verified by comparison with the experimental data, such as the measured residual
stress and variation of casting temperature.
Inelastic Behavior and Unified

Constitutive Theory of Metallic
Material in Solidification
When dealing with the high-temperature processes incorporating solidification, it is important to know the mechanical behavior of material
nearly up to the melting temperature, and unified

constitutive models are desirable for describing
characteristics of both inelastic solid and viscous
fluid (Ref 710). Studies of constitutive relationships at high temperature have been carried
out by many researchers (Ref 1114). In particular, many constitutive models for creep and
plastic behavior under some loading paths below
700 C have been developed in the 1970s to
1990s (Ref 1522). However, it is necessary to
verify whether these models are applicable to
mechanical behavior at high temperature close
to the melting point. Therefore, the authors first
carried out tension, creep, and cyclic loading
tests for type 304 austenitic stainless steel from
room temperature to melting temperature, and
strain rates depending on flow stress and strainhardening behavior were investigated.
Some typical inelastic constitutive equations
based on unified constitutive theory, such as the
types of Perzyna and Bodner as well as the superposition model, are used to describe hightemperature behavior and are compared with the
test results. The material parameters in the equations were identified by the data of tension tests
to simulate the creep and cyclic stress/strain behavior for verification of the models.
Experimental Procedure
Axial load and displacement (or strain) must
be controlled in tension and creep tests, and the
controlled value should be changeable independently of the time interval. A SUS 304 steel with
melting point of 1399 C was tested. The chemical compositions of material and the mechanical
properties at room temperature are shown in Tables 1 and 2, respectively. The specimen was
shaped into a solid cylinder with two lugs as
shown in Fig. 1.
Four temperature levels, that is750, 900,
1000, and 1200 Cwere chosen for the experiments since the authors were interested in the
material behavior at high-temperature condi-
tions. The material behaviors at lower temperatures were also examined to cover a wide range
of temperatures. The temperature variation along
the gage length was controlled within 2 C.
The heating rate of the specimen was nearly 8
C/min, and the test was started after the temperature was held for 60 min. The tension tests
were made under strain-controlled condition
with the strain rates of 1.0, 0.1, 0.01, and
0.001%/s. In the creep tests, several stress levels
below the yield stress were employed.
Table 1
Chemical composition of SUS 304
Element
Composition, wt%
Carbon
Silicon
Manganese
Phosphorus
Sulfur
Nickel
Chromium
0.08
0.59
1.00
0.032
0.005
8.26
18.30
Table 2 Room-temperature mechanical

properties of SUS 304
Property
Value
Yield stress (ry), MPa

Tensile strength (rs), MPa
Elongation (), %
274.4
607.6
62.0
10
M20X2.5
GL 50
33
104
33
170
Fig. 1
Shape and dimensions (in mm) of the specimen
Residual Stress Formation during Casting / 373
e Iij Kw(F)
200
750 C
150
900 C
100
1000 C
50
1200 C
0
104
T = 1200 C
150
103
102
0.1
Strain rate (), %/s
10
(a)
T = 1000 C
= 1.0%/s
50
0.01%/s
0.001%/s
900 C
10
1000 C
0
104
103
102
0.1
Strain rate (), %/s
100
0.01%/s
7.5 MPa
900 C
750 C
3
40.0 MPa
100.0 MPa
80.0 MPa
25.0 MPa
0
0
Time (t ), h
Fig. 5
0.001%/s
T = 750 C
= 1.0%/s
150
0.001%/s
100
0.01%/s
0.001%/s
0.01
1.0
250
200
150
100
Creep curves at several temperature levels
0
1.0
1.5
2.0
2.5
3.0
3.5
Strain (), %
Stress-strain curves at several temperature levels
900 C
n = 5.20
103
104
105
0
0.5
1200 C
n = 2.50
1000 C
n = 3.00
750 C
n = 6.25
50
50
0
0.0
Strain rate (), %/s
50
102
300
0.2 % proof stress (0.2), MPa
Stress (), MPa
= 1.0%/s
Stress (), MPa
150
Fig. 2
12.5 MPa
10
(a) Strain-rate dependence of flow stress at e

2.0%. (b) Strain-hardening coefficient
Fig. 3
T = 900 C
200
3.0 MPa
(b)
0
200
50.0 MPa
1200 C
100
5.0 MPa
150
1200 C
1000 C
25.0 MPa
Creep strain (c), %
50
(Eq 1)
750 C
15
F
rij
where K denotes a viscosity constant of the material. w(F) is a function of the static yield func-
= 1.0%/s
0.01%/s
0.001%/s
100
200
Stress (), MPa
In order to open complex relation of stress and

strain under engineering environments of loading and temperature, there has been substantial
development of the so-called unified theories
of viscoplasticity in which all aspects of inelastic
d/d, MPa/%
Stress (), MPa
200
Inelastic Constitutive Models
behavior are intended to be represented by the

same variable. Here, three types of constitutive
models that have some possibility of describing
the experimental behavior obtained above are
considered. Generally, the total strain rate e ij is
expressed as the sum of the elastic strain e eij and
the inelastic strain rate e Iij , that is, e ij e ije
e Iij , where the elastic strain rate is given by
Hookes law. The models for inelastic strain are
Perzynas model based on the excess stress theory, superposition model, and Bodners model.
Perzynas Model. The excess stress theory
assumes that the inelastic strain is produced as:
Creep strain (c), %
The experimental data on material behavior at

high temperature, which involve the conventional tension tests at various strain rates, creep,
and cyclic straining under several stress levels,
are discussed in this section. Figure 2 indicates
a series of stress-strain curves under monotonic
tension under several strain rates at 750, 900,
1000, and 1200 C. The variation of strain rate
has a large influence on the yield and flow
stresses, especially at high temperatures above
1000 C. The relation between strain and flow
stress at e 2.0% at various temperatures is
depicted in Fig. 3(a), and Fig. 3(b) represents the
effect of strain rate on hardening coefficient in
the same conditions. Figure 4 gives the results
of the yield stress depending on temperature.
In creep tests, the stress level was chosen to
be lower than proof stress r0.2. The creep curves
are shown in Fig. 5, which shows that the creep
strain at 750 C is very small, while the creep
strain over 1000 C is to present enough larger
value even when the stress level is lower, so the
flow mechanism of fluid rather than solid seems

to govern the deformation behavior at those temperature levels. Figure 6 shows the relation between stress and creep strain in a logarithmic
scale. The number n in Fig. 6 represents an exponent in the Norton law e Ar n.
Flow stress (0.02), MPa
Experimental Results and Discussions
Fig. 4
250
500
750
1000
Temperature (T ), C
1250
Relation between yield stress and temperature

under different strain rates
10
50 100
500
Stress (), MPa
Fig. 6
Relation between stress and steady-state creep

strain rates

tion F(rik , eIij , j, T) in stress space involving
inelastic strain eIij , hardening parameter j, and
temperature T. As a proposal of viscoplastic constitutive theory, Perzyna has proposed four kinds
of functions for w(F), among which a special
case w(F) F is employed to give the viscoplastic strain rate e vp
ijI :
ij e
3k
2
e vp
ij
1/2
2 s s
3
kl kl
ry
sij
(3J2)1/2
(Eq 2)
where J2 is the second invariant of deviatoric

stress sij , and ry denotes the yield stress. In order
to deal with viscous fluid, the authors chose k
13l with the viscosity l. When the value of
parameter n tends to unity, Eq 2 is reduced to a
uniaxial stress-strain relation under monotonic
tension and is then expressed as:
e I k(|r| ry)
eIij eijp eijc
Viscosity (), GPa s
(Eq 4)
The superposition model has various possibilities, but only two forms are employed here.
First, a simple superposition model (model A)
is considered. The plastic strain and creep strain
are simply added to represent the inelastic response of the material. The plastic strain rate
epij is given as:
skl skl
sij
(4/9)ry2dry/d e p
(Eq 5)
Here, yield stress ry is identified as a function

of equivalent plastic strain ep as:
1200
ry r0{1 l exp(bep)}
(Eq 6)
800
where r0, l, and b are material parameters. A

classical creep equation of Nortons law is used
to describe the creep strain rate e cij :
400
e cij mA 1/mr (nm)/m(ec)(m1)/mSij

0
300
600
900
1200
Temperature (T ), C
Fig. 7
Variation of temperature-dependent viscosity
200
Stress (), MPa
(Eq 3)
from which the parameter k, or viscosity l is to

be identified.
The superposition model assumes that the
inelastic strain eIij can be decomposed into the
time-independent plastic component epij and the
pure time-dependent (creep) strain ecij such that:
e Iij
Experiment
Superposition
Bodner
150
= 1.0%/s
100
rr
2
r2ydry/d e p
mA1/m(ec)(m1)/mrn/m
50
0
200
Experiment
Perzyna
Perzyna + creep
= 1.0%/s
0.01%/s
100
(Eq 8)
Another superposition model (model B) is

chosen that is a combination of the Perzynas
viscoplastic model and the Nortons creep equation (Eq 7). When Eq 8 is adopted to be the relationship for the creep process, the inelastic
model under uniaxial stress is then expressed as:
0.01%/s
150
(Eq 7)
where r represents the von Mises equivalent

stress and ec indicates the equivalent creep
strain, while A, n, and m are material constants.
Therefore, the constitutive form under the uniaxial stress can be written as:
e I
0.001%/s
Stress (), MPa
r
|r|
e I c(|r| ry)
r
Ar n
|r|
(Eq 9)
Bodners Model. A form of Bodners model

is employed as:
2 3J J
e ijI D0 exp
1 Z2
sij
1/2
2
(Eq 10)
50
0.001%/s
0
Fig. 8
2
3
Strain (), %
Stress-strain curves under different strain rates at

900 C
On the basis of incremental isotropy of the

flow, the scalar hardening variable Z ZI ZD
in the model is taken to consist of the parameters
of isotropy Z, and directional components ZD.
The simple form of the model under the uniaxial
stress can be derived:
e I
|r|
2
1 Z2
D0 exp
3
2 r2
(Eq 11)
where parameter D0 in Eq 10 and 11 is a material

constant.
Simulated Results by Some Constitutive

Models
Some results of simulated stress/strain behavior using the constitutive models described in the
previous section are presented. The parameters
of the constitutive models are identified from the
conventional experimental data stated in the section Experimental Procedure. The temperature-dependent viscosity l in the Perzynas
model is plotted in Fig. 7, and the parameter n
of the superposition models A and B is chosen
to be of the same value. Using these parameters,
experimental stress-strain curves under compression with results simulated by the three models
are shown in Fig. 8. The calculated creep strain
by superposition model B is given in Fig. 8,
which is compared with the experimental data.
In order to explain the work-hardening behavior
by these models, Fig. 9 shows two examples of
stress/strain behavior under the cyclic path at
900 and 1100 C, which are compared with the
experimental data.
From Fig. 8, the effect of the work hardening
is revealed to be very small when the temperature is above 900 C. The creep curves simulated
by superposition model B and experimental data
show almost no difference, and the model reflects the mechanical behavior under the cyclic
strain path at high temperature. Because the
model can be regenerated as a constitutive relationship for viscous fluids, one can say that the
Perzynas model B superposed by creep can be
used to describe the mechanical behavior of the
material near melting temperature. Among these
models, the Perzynas model may also be found
to be most suitable to simulate ultrahightemperature behavior including liquid state since
it is reducible to the model of the Newtonian
fluid above melting temperature.
Numerical Analysis Method of

Thermomechanical Problems in
Casting
One must often resort to a numerical analysis
in order to obtain quantitative solutions of nonlinear equations in continuum mechanics. However, regardless of the initial assumptions and
the methods to formulate such a problem in evaluating the results by numerical methods, the continuum is, in fact, approximated by a discrete
model in the solution process. One such approach, based on the idea of approximating continuous fields, is referred to as the finite-element
method. Its simplicity and generality is very attractive, and it is applied to a wide range of nonlinear problems.

The thermomechanical problem incorporating
solidification processes is a complicated frame
of coupling equations, which are not only related
to solid mechanics, fluid mechanics, or heatconduction problems, but also involve inelastic
constitutive relationships. The finite-element
method is a powerful tool for finding the numerical solution to simulate such thermomechanical behavior in casting. This section discusses numerical analysis technology, when the
finite-element method is applied to the thermomechanical problem, and what kind of algorithm
is employed to solve nonlinear equations in practice. Special attention is placed on the phase
transformation in solidification processes in the
numerical analysis, and the effects of latent heat
and dilation due to phase transformation are considered.
Governing Equation of
Thermomechanical Fields
From the continuum thermomechanics viewpoint, solidification is a complicated phenomenon characterized by the interaction between
temperature and stresses as well as phase transformation from liquid to solid, and the state in
solidifying material termed the mushy zone,
which is a mixture of solid and liquid phases. In
the case of some kinds of casting processes discussed later in this article, molten metal is
poured directly into a mold in addition to the
existing material. Figure 10 shows boundary
movement with increasing volume as well as interface movement during solidification. Here, an
initial domain V 0 at time t including liquid and
solid is considered, and C0 is an initial boundary
for the body. DV and DC denote the increments
of volume and boundary during time period Dt.
Hereafter any material parameter v is assumed
to be expressed by the mixture law using the
parameters of solid vs and liquid vl such that:
v vsns vlnl
qu dV
i
d
qui dV
dt
V
d
dt
q2 u u
1
i i
V 0 rij nj dC
e dV
hi)ni dC
qlsn s qf
DV qbi dV
(Eq 14)
T
ni h(T Tw)
xi
(Eq 19)
T
ni g(T 4 T 4w)
xi
(Eq 20)
where Tw is the temperature of the coolant, and

h and g are the heat-transfer coefficient and
Stefan-Boltzmann coefficient, respectively.
Kinetics of Phase Transformation in

Solidification
(rij uj
q(c bi ui)dV
qu rij ,j qbi
qe qc
To identify the volume fraction of the solid

phase ns during solidification (Eq 18), a phase
diagram for the alloying system is employed,
and the volume fraction of the solid phase ns is
assumed to be determined by the Scheil equation:
(Eq 15)
hi
rij e ij
xi
ns 1 1(1k0) and n1 1 ns
Here, is a dimensional function that is dependent on temperature:
(Eq 17)
100
900 C
= 0.001%/s
0.5
1.0
100
1.0
(b)
Hysteresis loop of stress at (a) and (b) 900 C and (c) 1100 C
1100 C
= 0.01%/s
Experiment
Perzyna
Perzyna + creep
Bodner
Superposition
50
0.5
0.0
0.5
1.0
100
1.0
Strain (), %
Strain (), %
(Eq 22)
50
50
0.0
(T Ts)/ms
(T Ts)/ms (T1 T )/m1
where ml and ms are the gradients of the liquidus

and solidus temperature Tl and Ts with respect to
50
0.5
(Eq 21)
(Eq 16)
Substituting the Fourier law h k(grad T)

into Eq 17, one can obtain a heat-conduction
50
(Eq 18)
where c, ls, and f are the specific heat, latent heat,

and heat supply, respectively.
The boundary condition of heat transfer and
thermal radiation can be written as:
(Eq 13)
where e and hi are the internal energy and the

heat flux, and c is the heat supply carried by the
additional volume of growing body discussed
later.
Applying Gausss theorem to Eq 14 and 15,
the local forms of the equation of motion and the
energy equation are obtained:
Stress (), MPa
Stress (), MPa
100
(a)
d d
i dV
DV q u
where bi is body force. The total energy balance

in the combined region with the surface normal
ni is represented by:
50
Fig. 9

T
qcT
k
rij eijvp
xi
xi
where q, q0 represent the density of each part.

Then the global momentum balance in the combined domain can be represented by:
900 C
= 0.01%/s
100
1.0
0 0
i dV
V0 q u
equation coupled with heat generation due to

mechanical work rij eijvp and latent heat ls as:
Stress (), MPa
100
(Eq 12)
with the volume fraction of solid phase ns.

When the initial velocity and stress in the domain V 0 are denoted as u0i , and r0ij , and the components in DV are udi and rdij , the linear momentum in the combined body V V 0 DV with
C C0 DC having the velocity ui is reduced
to:
0.5
0.0
Strain (), %
(c)
0.5
1.0

the alloying element on the phase diagram, respectively, and k0 is a distribution coefficient
representing the segregation effect. When the
segregation effect is assumed to be neglected, as
in this article, Eq 22 can be reduced to the simple
form:
ns
must also be used. Since the viscoplastic strain

increment of the element at the kth time step may
be written by Taylors expansion:
{Devp}k {evp}kDt hHk{Dr}kDt
placement rate ui different from vi. Then, the

material-time derivative of any time-dependent
scalar, vector, or tensor function w(xi,t) is reduced to:
(Eq 25)
w

where h [0,1] and Hk ({e }/{r})k. Here,

the stress increment can be represented by:
vp
(T1 T)/m1
(T Ts)/ms (T1 T)/m1
(Eq 23)
{Dr}k [De]{Dee}k [De]({De}k
which is called the lever rule.
{Devp}k {Dem}k {DeT}k)
Formulation of Numerical Analysis

This section considers formulations of finiteelement method in order to simulate the steadyand unsteady-state problems in the casting process.
Formulation of Finite-Element Method for
Unsteady-State Casting. By using Gausss theorem and a weighted residual method on the
heat-conduction equation (Eq 18) and boundary
condition (Eq 20 and 21), one can obtain a system of finite-element formulation in the weak
form of Galerkin type:
{Qh} {Qf} {Qm}

{Qn} {Qr}
qcVi
with the elastic stress-strain matrix [D ]. By substituting the stress increment in the form of Eq
25 into Eq 26, the elastic-viscoplastic stressstrain matrix [Dvp] leads to:
(Eq 24)
where [P], [Hk], [Hh], and [Hf] are the matrices

related to capacity, conductivity, heat transfer,
and heat radiation, respectively. {Qh}, {Qf},
{Qm}, {Qg}, and {Qr} are the vectors of heat
transfer, heat radiation, latent heat, and heat supply due to stress power by the increasing molten
metal. Due to the time dependence in the finiteelement equation (Eq 24), time discretization is
needed, and a differential scheme based on the
Crank-Nicholson method is employed.
With regard to the nonlinear solution of the
thermal stress problem, an incremental approach
T

T
k
rij evp
ij
xi
xi
xi
qlsVi
ns
0
xi
(Eq 30)
The temperature in each element is expressed by:

[Dvp]k ([1] [De]hHkDt)1
(Eq 27)
T [N]{T}e
Using the variational principle, the finiteelement equation for stress analysis finally can
be obtained as:
[K]{Du} {DFf} {DFvp}
{DFm} {DFT}
([Hk] [Hh] [Hf]){T}

[P]{T}
(Eq 29)
When the term of (w)/t in Eq 29 vanishes for

steady state, heat conduction Eq 18 is now reduced to:
(Eq 26)
e
w
w
vi
t
xi
(Eq 31)
in which [N] is a trial function matrix. The finite

element formula of heat conduction equation is
obtained by allying the Galerkin method to Eq
30 with the boundary conditions (Eq 19, 20);
thus, an integral form is obtained:
(Eq 28)
where [K] is the stiffness matrix, and {DFf},

{DFvp}, {DFm}, and {DFT} are the incremental
vectors of load included by body force, effect on
inelastic deformation and thermal expansion,
and phase transformation, respectively.
Formulation of Finite-Element Method for
Steady-State Continuous Casting. In the
continuous-casting process, the material flows in
space, so that the fields of temperature and
stress/strain can be generally described in a spatially fixed Eulerian coordinate system xi instead
of a Lagrangian coordinate Xi being moved with
the material with velocity Vi. In the case of casting, the material velocity Vi is assumed to be a
sum of the casting speed vi and the small dis-
[N]T qcVi
T
ns
rij evp
dV
ij qlsVi
xi
xi
k[G] [G]{T}dV
T
T
C[N] qdC
(Eq 32)
where
[G] []T[N]
(Eq 33)
and []T is a differential matrix. The thermal

flow q can be calculated by Eq 19 or 20. From
Eq 32, the formation for finite element analysis
is obtained:
([P] [H]){T} {Qs} {Qh} {Qq}
V i = Vi+ V i
Vi
i
[P]
Vi+
e VV q]T xi [N]{m}dV
Vi+
i +
Fig. 10
Model considering growth of domain and solidified mushy zone
(Eq 34)
The first term of the right-hand side of Eq 34

represents the latent heat generation due to solidification, and the second and third terms correspond to the heat flows through boundaries Sh
and Sq, respectively. Here, there are two terms
dependent on the casting speed, which are generally formed by:
i
i+
(Eq 35)
{Qs}
e V qls i ns
x
[N]{m}dV
(Eq 36)
When the vector of displacement rate {u} and

strain rate {e} in a finite element are interpolated
by the nodal displacement vector {u} in the
forms:

}e and {e} [B]{u}e
[u} [N]{u
(Eq 37)
with the strain-displacement matrix [B] dependent on coordinates, the system of stiffness equation in incremental form is derived from Eq 37
by applying the principle of virtual work as:
[K]{u} {L b} {L T}
{L m} {L v}
(Eq 38)
The terms in the above equation are summed up

with the element matrices generally shown as:
[K]
e V [B]T[De][B]dV
(Eq 39)
{L b}
T [b]
e V [N]
xi [N]{m}dV
(Eq 40)
{L T}
e V [B]T([ ]grad[N]){m}dV
(Eq 41)
{L m}
e V [B]T[De][b](ns xi [N]{m}dV
(Eq 42)
Here, the temperature and the volume fraction of

solid phase at the central point of the element are
used in calculation of these matrixes. Total matrix [K] denotes a total stiffness matrix, and [De]
represents the elastic stress-strain matrix, while
{r} and {V} are the vectors of deviatoric stress
and casting speed. Other vectors {}, {b}, and
scalar w are:

[De]1{r}
T
(Eq 43)

[b] [b]
[De]1{r}
ns
(Eq 44)
[]
[]
x
1
ry
1
2l
(3J2)1/2
(Eq 45)
Algorithm of Numerical Analysis

In the process incorporating phase transformation, the coupling effect between the fields of
temperature, microstructure of material, and
stress/strain are essentially to be taken into account in the analysis. As stated in the previous
section, however, the numerical analysis of
steady-state temperature distribution coupled
with solidifying phase change is made independently followed by stress analysis based on obtained temperature field. Here, and assumption
is made on the velocity field in both analyses that
the casting velocity vi is large enough compared
with the displacement rate ui in the case of uniform casting, which follows that the velocity {v}
in Eq 35 to 42 is to be replaced by the uniform
casting speed {v}.
Liquidus and solidus lines are determined by
solving the finite element Eq 24 by neglecting
the first term {Qs} on the right-hand side repre-
senting latent heat generation in the first approximation. By use of the lever rule, the volume
fraction of solid ns is identified, and the term
{Qs} is evaluated. The same iteration procedure
was repeated to obtain the converged distributions of temperature and phase. The obtained result is applied to determine the displacement rate
vector {u} in Eq 37, and displacement ui, strain
eij, and stress rij can be evaluated by integrating
the rate e ij , and r ij . Moreover, Eq 24 and 38
derived in the previous section are nonlinear
with respect to the temperature {T} and displacement rate {u}, respectively, and special attention is necessary in solving the equations.
Equation 34 has only one term for latent heat
generation determined by volume fraction of
solid ns, thus, it employs the simple method of
iteration.
As for stress analysis, on the other hand, the
nodal forces on the right-hand side of Eq 38 are
the nonlinear functions of temperature T and
structural change ns being dependent on the rate
{u}. Furthermore, in the region of molten state,
a large strain rate is produced in spite of a small
variation of nodal forces. A modified NewtonRaphson scheme, or initial method, was adopted
to solve the nonlinear equation.
Residual Stress Formation during

Semicontinuous Casting
Vertical semicontinuous direct-chill casting
process is one of the most efficient methods of
producing ingots of aluminum alloys and other
metals. It is beneficial for optimizing the operating conditions to simulate the thermomechanical field in the solidifying ingot. Many reports
have been published concerning such finiteelement analyses of the temperature distribution
incorporating solidification, but a few papers
treat the induced stress/strain field: simulations
of thermal stress in continuous-casting slab were
made using an elastic-plastic constitutive model,
for example, Grill, Brimacombe, and Weisenberg (Ref 23) and Sorimachi and Brimacombe
(Ref 8), and viscoplastic stresses were simulated
based on the solidification analysis by Williams,
Lewis, and Morgan (Ref 1). However, in their
studies, the influence of casting speed was neglected, so that the numerical simulation along
with the variation of casting conditions cannot
be realized. In order to solve this problem, Ju
and Inoue (Ref 2426) proposed a numerical
simulation method by Eulerian coordinate, and
application to the continuous-casting process of
steel slab was carried out.
The aim of this section is to apply the coupled
method of temperature and stresses incorporating solidification to semicontinuous direct-chill
casting of aluminum alloy. When the bottom
block plate is located at the upper position and
the length of the growing ingot is small, the temperature, liquid/solid interface, and stresses in
the ingot vary with time, both in the sense of
space and of material. However, when the ingot
becomes long enough, the physical field in the
upper part is regarded to be time-independent or

steady in the spatial coordinate fixed to the system. In the first part of this section a steady-state
heat-conduction equation with heat generation
due to solidification is formulated in a spatial
coordinate system when considering the material
flow. A numerical calculation for the temperature in the solidifying ingot as well as the simulation of the location of liquid/solid interface is
carried out by finite-element analysis.
Most metallic materials at low temperature
may be treated as an elastic-plastic solid. However, if they are heated beyond the melting point,
the materials can be regarded as viscous fluid,
and they behave in a time-dependent inelastic
manner at high temperature close to the melting
temperature. Therefore, a unified constitutive
model needs to be established to describe the
elastoplastic and viscoplastic behavior of the solidified part of the ingot as well as the viscous
property of the liquid state. Taking into account
the effects of such phenomena, a modification of
Perzynas constitutive model as in other sections
of this article is presented in the second part of
the section, and some experimental results of the
viscosity appearing in the model are presented
for the aluminum-zinc type alloy employed. Using the model, elastic-viscoplastic stresses are
calculated in the ingot to find the residual stress
distribution and are verified by the measured
data from the hole-drilling strain-gage technique.
Finally, results of numerical simulation are
presented on the influence of operating conditions on temperature and stresses, such as ingot
size, casting speed, initial temperature, and so
on, to give fundamental data for optimizing the
operating condition.
Finite-Element Model and Casting

Conditions
The theory and the procedure developed as
discussed previously are now applied to the
simulation of the vertical semicontinuous directchill casting process shown schematically in Fig.
11. The material treated is an aluminum-zinctype alloy with 5.6% Zn and 2.5% Mg. A quadrilateral finite-element mesh pattern of 600 elements with 1941 nodes illustrated in Fig. 12 is
employed for both analyses of temperature and
stress fields.
The boundary condition for heat conduction is
assumed in such a way that the temperature of
the meniscus of molten metal is prescribed to be
w0, and that heat is insulated along the central
line and the bottom of the ingot as well as the
surface contacted with the refractory. The cylinder facing to the mold is regarded as the
boundary Sq on which heat flux is given, and the
other part of the surface Sh is given by a heattransfer boundary due to cooling of water. Figure
13 depicts the measured heat flux q absorbed by
the mold, and heat-transfer coefficient h depending on flow rate of water is shown in Fig. 14.
Other data used for temperature calculation incorporated with solidification are shown in Table

3. Characteristic results of calculated temperature and residual stresses for an ingot 1 m in
length with the diameter of 240 mm are compared with experimental data to verify the
method. Simulations in other cases of different
Molten metal
Refractory
operating conditions such as casting velocity,

size of the ingot, and cooling rate are also made.
Results of Simulation
Figure 15 indicates a birds-eye view and an
isothermal representation of calculated temperature distribution. The lines denoted by Tl and Ts
in the figure are the liquidus and solidus tem-
Mold
Water flow
Billet
Heat flux (q), cal/mm2 s C
Cooling
water
0.8
0.6
0.4
0.2
0
0
Bottom
block
Withdrawal
Fig. 11
Schematic of semicontinuous vertical directchill casting
Refractory
Mold
Heat-transfer coefficient (h), kcal/m2 h C
Fig. 13
40
80
120
Distance from top of mold (d ), mm

Variation of heat flux from top of the mold
1.6
Wl = 145 l/min
115
85
1.2
0.8
0.4
0
100
200
300 1000
Distance from top of mold (d ), mm
Fig. 14
Table 3
Distribution of heat-transfer coefficient depending on discharge of cooling water
Material properties of aluminum 7075
Heat conductivity, cal/(mm C)

Density, g/mm3
Specific heat, cal/(g C)
Latent heat due to solidification, cal/g
Casting speed, mm/min
Liquidus temperature, C
Solidus temperature, C
Gradient of liquidus line, C/%
Gradient of solidus line, C/%
Youngs modulus, MPa
Poissons ratio
Viscosity, MPa s
Initial yield stress, MPa
Hardening coefficient, MPa
Fig. 12
Finite-element mesh pattern for semicontinuous vertical direct-chill casting
perature, respectively. Figure 16 shows the data

replotted to give the temperature change at the
center and on the surface, while the circles represent the measured temperature by thermocouples. The fact that the calculated temperature on
the surface coincides well with the measured
data may indicate the validity of the simulation
method. Change of the volume fraction ns at four
characteristic points along the distance from meniscus is represented in Fig. 17, which may give
the information on progressing mode of solidification and thickness of solidified shell.
For the simulated results of temperature and
solidification mode presented above, stress analysis was carried out for the entire area of the
ingot, including the molten metal, by use of the
finite-element equation (Eq 28) based on the
elastic-viscoplastic constitution model (Eq 2).
The displacement mode is depicted in Fig. 18.
Contour of radial, tangential, and axial stress distributions rr, rh, and rz are shown in Fig. 19.
The contours of stress distribution are represented in Fig. 20.
Examples of the calculated stress distribution
by elastic-viscoplastic constitutive model and
the one by time-independent elastic-plastic
model are shown in Fig. 21. When compared
with each other, the viscoplastic stress analysis
gives smaller results, at least on the surface, than
elastic-plastic stresses. The stresses are found to
generate at the location where the solidification
starts (see Fig. 19 and 20), and the radial distribution becomes steady going downward owing
to the flatter temperature distribution. In order to
examine this effect, the stress distributions at
several locations are given in Fig. 22, in which
the distribution at the very end of the ingot (Fig.
22d) can be regarded to be residual stresses.
Open-circle data points in the figure indicate the
experimentally measured residual stresses by a
hole-drilling strain-gage method (Ref 27). The
fact that the experimental data coincide well with
the analytical results suggests the validity of the
simulation procedure based on the viscoplasticity developed here. As seen from Fig. 22, the
shear stresses are insignificant in the area of mol-
Thermal expansion coefficient, 1/C

Dilation due to solidification
k1 0.0425; ks 0.0125 0.0583T

q 2.8 103
c1 0.015 0.00173T; cs 0.083 0.00267T
ls 93.16
V 80.0
T1 638
Ts 600
m1 3.69
ms 9.09
E1 500.0; Es 75000.0 76.3T
m 0.33
l 3700.0 (T 346 C)
l 0.007178T 2 21.1698T 10160.7 (T 346 C)
ry0 2.0 (T 346 C)
ry0 150.0 0.428T(T 346 C)
H 350.0 0.13333T (T 150 C)
H 330.0 1.683 (T 150) (150 T 346 C)
H 0.132 (T 346 C)
1 33 106; s 21.8 106 0.2T 107
b 7.5%

ten state, and the normal stresses rr, and rz (see
Fig. 19 and 21) are regarded to be hydrostatic
stresses, which means that the constitutive equation employed here reveals modified deformation of the liquid.
fect of temperature and stress distributions on the

casting speed, size of the ingot, and cooling condition. Figures 23 and 24 represent examples of
temperature profiles for different casting speed
and the radius of ingots with various cooling
rates. The effect of discharge of cooling water
on thickness of the mushy zone at the center is
summarized in Fig. 25. The effects of casting
speed and ingot radius on the stress rh are represented in Fig. 26 and 27, respectively, and the
relation between stresses and casting speed is
Simulations for Other Operating

Conditions
Let us apply the procedure to cases with other
operating conditions. Focus is placed on the ef-
Temperature (T ), C
TI
Ts
1000
600
800
400
600
200
40
s (r
diu
Ra
200
120
om
e fr
anc
Dist
80
), m
m
Fig. 15
),
s (z
400
mm
cu
enis
Calculated temperature profile. V 80 mm/min
800
Experimental
Calculated
r = 120 mm
1.0
Volume fraction of solid, s
Temperature (T ), C
600
400
Center
200
Surface
0.8
0.6
0.4
r = 0 mm
r = 380 mm
r = 76 mm
r = 120 mm
0.2
0
0
0
100
200
300
80
400
Temperature variations at the center and the

surface of the ingot. V 80 mm/min
Displacement (Ur), mm
Fig. 16
100
120
140
Fig. 17
Volume fraction of solid along the distance

from meniscus. V 80 mm/min
1000
0.2
800
0.0
600
0.2
400
Ra
40
diu
200
80
s (r
120
), m
Fig. 18
160
Distance from meniscus (z ),mm
Distance from meniscus (z ), mm
Displacement mode. V 80 mm/min
nce
a
Dist
from
r (z
te
cen
), m
summarized in Fig. 28. The simulated stresses

varying with the discharge of cooling water is
also plotted in Fig. 29. The calculated results
shown above seem to simulate the characteristics
of temperature and stresses depending on operating conditions. If the data of such simulation
is accumulated at different operating conditions
as shown in Fig. 23 to 27, it would be possible
to optimize the design of the system.
Residual Stress Formation during

Centrifugal Casting
From the viewpoint of the solidification microstructure, materials present various solidifying types: for example, solidification of pure media and eutectic material should be a smooth
interface between liquid and solid (Ref 28, 29);
however, solidification of the alloy has no obvious interface, but has been a mixture domain
of liquid and solid. Furthermore, as stated in the
section Numerical Analysis Method of Thermomechanical Problems in Centrifugal Casting, solidification with growing domain is a
practical process, and it has many important effects in some practical engineering problems
such as centrifugal casting, welding, and so on.
It is necessary to consider the thermomechanical
process accompanied by heat supply due to increment or input of the material as time goes on.
Because moving the total domain as well as the
boundary will cause an energy variation in the
total system, this problem has general significance. However, because it is difficult to consider the entire phenomenon, the beginning of
the study only considers a static problem involving a moving boundary due to increment of domain, and this section discusses an unsteadystate solidification process. In order to describe
the inelastic behavior using a unified mathematical model, two aspects are emphasized. One is
that the process is associated with a nonlinear
problem with a moving external boundary depending on time and location, and the other is
characterized by the fact that it incorporates
phase transformation such as solidification based
on the mixture theory.
For the numerical analysis of the solidification
and growth of domain, a finite-element scheme
proposed in the section on numerical analysis is
used. Here, heat input due to successively increasing liquid metal was introduced into the
heat-conduction equation by adding the newly
developed elements in the finite-element
scheme. Moreover, latent heat generation due to
solidification is also taken into account by varying fraction of solidified volume, which can describe the state of the mushy zone. Simultaneously, a method of stress analysis based on the
obtained temperature distribution is developed,
where an elastic-viscoplastic constitutive relationship capable of describing the viscoplastic
solid as well as viscous fluid is employed. Here,
in order to simulate stress and deformation, one
assumes that the velocity along direction Vr is
small and has the same stress state on contacted

boundary. Examples of numerical calculation of
temperature with solidification and stress are
presented for some practical cases of centrifugalcasting processes, and the validity of calculated
temperature field under several operating conditions is discussed in comparison with the experimental results.
Stress (r ), MPa
1000
800
100
600
50
400
0.0
40
80
200
Stress ( ), MPa
), m
mm
e fro
anc
Dist
0.0
1000
800
100
0.0
100
0
40
80
600
400
us (
200
e fro
anc
Dist
0.0
isc
men
z ),
mm
Stress (z ), MPa
1000
800
100
0.0
100
0
40
80
600
400
cus
200
e fro
anc
Dist
0.0
Fig. 19
s (z
cu
enis
is
men
(z ),
mm
Contour of radial (rr), tangential (rh), and axial (rz) axial stress distributions. V 80 mm/min
1000
24
48
72
96 120
Radius (r ), mm
Fig. 20
24
96 120
24
48
MPa
10
.2
MPa
MPa
54.2
Pa
72
Radius (r ), mm
40.5
Pa
43.1 M
Pa
0M
7
5.
2.76 MPa
48
12
6.0
M
167.2
800
23.7 MPa
MPa
32.1 MPa
98.7 M
1.02 MPa
600
49.5 MPa
70.1 MPa
400
Pa
13.6 MPa
Distance from meniscus (z), mm
200
72
96 120
Radius (r ), mm
Isostress contours. Radial (rr), tangential (rh), and axial (rz) axial stress distributions. V 80 mm/min
Analysis Models and Casting

Conditions
Centrifugal casting is one of the most efficient
processes for producing pipes, tubes, cylinder
sleeves, and so on. It is well known that the technique utilizes the centrifugal force generated by
a rotating cylindrical mold to fit the molten metal
on the mold wall and to form the desired shape,
as shown in Fig. 30. Many problems still remain
in the process regarding thermal and mechanical
behaviors, for example, progressive domains of
molten and solidified metals, moving interface
between both parts, and development of stresses
along the solidified field. Centrifugal-casting
methods have been developed to make many machine parts and structures. However, there are no
numerical analyses of the casting process, because it is difficult to solve temperature field,
solidification, and residual stress field due to the
growing domain in the casting process. In order
to simulate the process, the authors applied the
above-mentioned analytical procedure to the
casting practice. In this section, centrifugal casting that produces rolled products and pipes are
calculated.
An axisymmetrical model as shown in Fig.
31(a) is employed for finite-element analysis for
the first model (I), where both ends of mold are
fixed during casting process and the casting roll
is long (L 5 m) with a large diameter (internal
diameter of mold do 1270 mm). Moreover, in
order to evaluate the distribution of temperature
and stress fields in the end corner of a pipe, the
authors chose another model (II) with finite
length (L 1 m) as shown in Fig. 31(b).
In model I, the material used for the pipe is a
nickel-chromium alloy, and the metal mold is
made of a carbon steel. The rotating speed and
the initial mold temperature is x 440 rpm and
T0 100 C, respectively. The initial temperature of molten metal was Ti 1330 C with
liquidus temperature Tl 1230 C and solidus
temperature Ts 1140 C. Since the heat is
transferred from the mold surface to air, the heattransfer coefficient is taken as h 2.78 103
cal/(mm2 s C).
The materials of the metal mold and pipe of
model II are both 7075 aluminum alloy, as used
in the process described in the section Residual
Stress Formation During Semicontinuous Casting. The initial mold temperature and molten
metal temperature are chosen as T0 650 C
and Ti 680 C, and liquidus and solidus temperature are Tl 638 C and Ts 600 C. The
heat-transfer coefficient is chosen as h 5.56
103 cal/(mm2 s C). The rotating speed
is kept to x 540 rpm, and the time period of
0
6M
Pa
1000
6
7.
5M
Pa
Pa
1
6
Pa
124 M
MPa
12
.8
27.8 MPa
v = 80
mm/min
640 C
1
1
MPa
52
.5
MPa
76.2
164 M
Pa
600
79.5 M
Pa
400
800
24
48
72
96 120
24
48
72
v = 95
mm/min
640 C
640 C
250
82
0
Radius (r ), mm
Fig. 21
v = 70
mm/min
Mold refractory
20 MPa
200
28.1 MPa
0.0
85
96 120
110
Radius (r ), mm
500
Calculated stresses (rh) by (a) elastic-viscoplastic and (b) elastic-plastic models. V 60 mm/min
120 0
120 0
120
Radius (r ), mm
Isothermal lines for different casting speeds
v = 70
mm/min
0.0
v = 60
mm/min
200
Stress, (r, , z), MPa
100
0.0
100
200
100
0.0
100
24
48
72
98
120
200
24
Radius (r ), mm
100
0.0
640
640
110
110
72
98
500
120
150 0
250 0
300
Radius (r ), mm
Fig. 24
100
0.0
r
z
100
24
48
72
98
120
200
Radius (r ), mm
Fig. 22
660 C
Isothermal lines for different ingot radii
200
100
(c)
48
(b)
200
660 C
400
Radius (r ), mm
(a)
200
640
110
Thickness of mushy zone (ts), mm
200
660 C
v = 35
mm/min
Mold refractory
Fig. 23
rz
0
24
48
72
98
120
38
v = 80 mm/min
36
34
32
30
80
Radius (r ), mm
100
120
140
150
Discharge of cooling water (wt), l/min
(d)
Stress distribution in several sections of ingot. (a) z 121 mm. (b) z 400 mm. (c) z 850 mm. (d) z
1000 mm
Fig. 25
water
Relation between thickness of mushy zone at

the center depending on discharge of cooling

outer surfaces, where solid lines represent the
results of coupled calculation that considered the
heat supply of stress power, while dashed lines
are the uncoupled results without effect of stress
power.
The simulated temperature is compared with
experimental results represented by open circles
measured by radiation thermometer. Figures 33
and 34 depict the distributions of density and
volume fraction of solid in radial direction as the
parameter of time. From the results, it can be
seen that the cooling rate on the inner surface
tends to be slow because of the latent heat generation due to solidification, while the rate becomes larger after the solidification was completed. The distribution stress associated with
Results of Simulation for Model I. As the

result of temperature, lines in Fig. 32 show the
temperature variations with time on the inner and
v = 60 mm/min
1000
24
72
Fig. 26
v = 80 mm/min
24
48
2.76 MPa
23.7 MPa
7
5.0
12
MP
6.0
a
MP
a
98.7 M
Pa
MP
83
.2
72
Pa
32.1 MPa
Pa
.0 M
167.2 M
12
6
31.6 MPa
169.1 M
96 120
Radius (r ), mm
3.51 MPa
MPa
Pa
MPa
99.5
5
48
12
8
800
27.6 MPa
164 M
Pa
600
2.5
76.2 M
Pa
MPa
400
23.4 MPa
200
v = 75 mm/min
30.1 MPa
Stress (z, ), in solid shell, MPa
Results of Numerical Analysis
96 120
w l = 112 L/min
48
72
96 120
Radius (r ), mm
w l = 142 L/min
45
w l = 245 L/min
11.0 MPa
10.9 MPa
9.9 MPa
2
55.
.7
57
500
35
z
30
25
55
85
70
Casting speed (v), mm/min
100
Relation between stresses at the center of ingot

and casting speed
40
20
0
70
500
r = 120 mm
w l = 115 mm
40
Fig. 28
Stress distributions (rh) for different casting speeds
50
60
24
Radius (r ), mm
development of solidification is represented in

Fig. 35. Compared with the data of volume fraction of solid shown in Fig. 34, stress in liquid
state, in fact, vanishes, while drastic variation of
the stresses is observed in the mushy zone between liquid end and solid front. Simulated results of the residual stresses are plotted in Fig.
36 when the cast region and mold cooled down
to room temperature at t
.
Results for Model II. Figure 37 shows the
temperature distribution at different time steps.
Figures 38 and 39 represent the variation of density and volume fraction of solid. It follows from
the results that the variation of temperature, density, and volume fraction of solidus in the corner
of model II is increased very rapidly due to cool-
Stress in solid shell, MPa
the pouring molten metal is 3 s. Mechanical parameters of the alloy as shown previously also
are employed.
The same thermal radiation coefficient g
7.028 107 cal/(mm2 s K4) is used for two
models, and the radiation condition is situated
on the inner surface of pipe. In model I, heat
insulation condition is assumed on both sides of
the model, while it is also imposed on the outer
surface of the mold in model II.
110
130
90
Discharge of cooling water (wt), t/min
Relation between stresses at the center of ingot

and discharge of cooling water. V 80 mm/
min. z 112 mm
Fig. 29
500
Ladle
Molten pool
35.4
150
Radius (r ), mm
129.0
Metal mold
35.7
82.5
46.7
1000
Fig. 27
113.0
126.0
80.7
10
4.0
10
1.7
.5
86
1000
1000
0
250
Radius (r ), mm
150
20
300
Bearing
Radius (r ), mm
Stress distributions (rz) dependent on effect of ingot diameter and cooling water rate. V 80 mm/min
Fig. 30
Schematic of centrifugal casting

and fluid velocities in both liquid phase and solid
phase. On the other hand, due to the symmetry
to the central line, a half part of the model shown
in Fig. 42(b) is treated for the analysis.
Calculated Results of Material A
Strip continuous casting by twin-roll method

is a promising technology. Not only can the casting method save energy, but it can also reduce
production costs in the manufacturing of material. However, it is difficult to control the quality
of the strip because of the existence of the deformation of the strip itself, due to thermal expansion or thermal stress. Here, there are two key
points: first, if the solidification is completed before the liquid reaches the minimum clearance
point between the rolls, then the strip will incur
a fixed gap. Therefore, one key point is controlling solidification. Another key point is that the
viscoplastic deformation incorporating material
flow has to be considered in this thermomechanical process.
Continuous Casting by Twin-Roll

Method
The twin-roll continuous-casting system is
schematically illustrated in Fig. 42(a). In this
process, molten metal is sent between two rolls
rotating in opposite directions with some angular
velocity. The level of the molten metal is always
kept constant by overflowing the excess molten
metal from the nozzle. As soon as the molten
material is poured into the rolls, solidification
takes place on the roll surface, which is cooled
by circulating water inside the roll. Therefore,
the problem then is to find this steady solidification profile and the distribution of temperature
The procedure developed above is now applied to the simulation of the thin-slab casting
process under some operating conditions. The
results are summarized as follows. Figure 43
represents the finite-element discretization of the
whole region of the strip and roll. The approximate velocity field based on the incompressibility condition and boundary conditions of velocity due to rotation of roll is shown in Fig. 44.
The roll surface as well as the boundary in contact with the roll and strip is assumed to belong
to heat-transfer boundary and the surface of the
strip to the heat-radiation boundary.
In order to verify the numerical analysis
method proposed in the section on numerical
analysis, two kinds of casting material are taken
into consideration: one kind of material (material
A) is a low-melting-point alloy (tin-bismuth alloy), and the other (material B) is SUS 304 steel
mentioned in the section Inelastic Behavior and
Unified Constitutive Theory of Metallic Material
in Solidification. The casting speed is set at Vc
77.2, 64.5, and 51.1 mm/s. The thickness of
the slab is 3 mm. The initial temperature of the
molten metal is T0 190.0 and 185.0 C, respectively. The material coefficients of tinbismuth alloy are used for the temperature calculation and stress analysis. The initial
temperature of rolls is chosen as Tm 10 C.
The liquidus temperature and solidus temperature of tin-bismuth alloy are Tl 150 C and Ts
138 C.
When the casting process of SUS 304 is considered, the thickness of the slab is 1 mm, and
two casting speeds are set: Vc 400 and 600
mm/s. The properties of material can be found
Simulated results of steady temperature distribution in both the strip and roll are shown in
Fig. 45(a) for the casting speed of 77.2 mm/s
with T0 190 C on the meniscus, and Fig.
45(b) depicts the volume fraction of solid phase,
in which the line indicated by ns 1.0 shows
the solid front. Measured temperature distribu-
1200
Temperature (T ), C
Analytical Models and Parameters
Residual Stress Formation During

Strip Continuous Casting by TwinRoll Method
in the section Inelastic Behavior and Unified

Constitutive Theory of Metallic Material in Solidification.
Measured
Uncoupled method
Coupled method
800
400
1000
Fig. 32
2000
3000
Time(t ), s
4000
Temperature distribution (model I)
2010
1050
103
Density, (), g/mm3
ing from both ends. Figure 40 shows the development of equivalent stress in the growing domain under the solidification process. Residual
stresses at the end of cooling are shown in Fig.
41.
480
8.0
7.5
t = 210 s
7.0
500
550
600
635
Radius (r ), mm
Fig. 33
500
2010
850 mm
r = 635
Additional
element
r = 108
114 mm
1050
Additional
element
Fig. 31
Metal
mold
(a)
Density distribution (model I)
5.0
(b)
Finite-element division. (a) Model I. (b) Model II
510
1.0
t = 210 s
0
500
550
600
635
Radius (r ), mm
Fig. 34
Distribution of volume fraction of solid (model I)

ure 53 shows that temperature distribution on the
slab surface at the roll outlet presents violent
fluctuations due to latent heat generation by solidification domain. When the casting speed Vc
is raised, the temperature fluctuation toward a
small range is presented on the slab surface.
The distribution of the horizontal stress rx
from meniscus to the junction point at the roll
outlet is represented in Fig. 54. From these figures, the high-level stress field by rapid cooling
is found to occur near the shell surface, and the
equivalent stress distribution on the slab surface
at the roll outlet also presents violent fluctuations
due to the temperature fluctuation. The equivalent stress analysis results based on several constitutive equations are plotted in Fig. 55. In the
section of slab at the roll outlet, calculated equivalent stress by three types of constitutive equations is approximated to the same distribution.
However, when casting speed is lower, the effect
of creep strain is evident.
Calculated Results of Material B

Simulated results of steady temperature field
in both the strip and roll are shown in Fig. 51(a)
and (b) for the casting speed of 400 and 600 mm/
s. The fraction of solid phase in solidified region
of the strip with liquidus and solidus temperature
Tl 1460 C and Ts 1399 C is depicted in
Fig. 52. Temperature distribution along the central line is plotted in Fig. 53. In the early stage
of rapid cooling by the roll, the solidified shell
is seen to grow gradually toward the central part.
The change in casting speed is known to affect
the distribution of temperature and shell thickness. As the casting speed becomes faster, the
position of liquid line moves downstream. Fig-
in some casting processes, which include semicontinuous casting, centrifugal casting, and continuous casting by twin-roll method. A method
to simulate solidification and temperature as well
as stress distribution in casting is formulated,
and the implementation by finite-element cal-
100
Mold
0
50
r
e
z
100
150
200
500
Concluding Remarks
600
550
650 700
750 800 850
Radius (r ), mm
Fig. 36
Temperature ( ), C
This article provides a general discussion on

the framework of thermomechanical theory incorporating solidification, and a series of practical analytical schemes based on the finiteelement method are proposed. These schemes
are applied to simulate thermomechanical fields
Casted
region
50
tion by thermocouples along the central line and

surface of the strip are plotted by data points in
Fig. 46 and 47, while the solid lines indicate the
calculated temperature results. It may be suggested from the figure that the simulations give
fairly good agreement with the experiment.
The stress distribution is shown in Fig. 48, and
the results at several locations are given in Fig.
49. The distribution at the very end of the strip
can be seen to grow up again, which may become the origin of residual stresses. Figure 50
gives the comparison of the measured data of the
thickness with the calculated results.
700
400
0
100
t = 0.5 s
200
Di
t = 450 s
Residual stress distribution (model I)
Solid
front
r
e
z
t = 300 s
Liquid
end
Temperature ( ), C
st
a
300
pl nce
an fr
e om
(
400
), ce
m nt
m ra
l
700
108
112
Radius (r ), mm
400
t = 2.0 s
0
100
200
l
ra
nt
ce
m m
fro ), m
ce (z
an e
st lan
Di p
t = 210 s
300
400
20
Liquid
front
112
104
400
0
100
t = 4.0 s
200
20
500
Casted
region
700
300
400
Mold
600
l
ra
nt
ce
m m
fro ), m
ce (z
an e
st plan
t = 30 s
800
112
Radius (r ), mm
Fig. 35
108
Radius (r ), mm
700
Di
t = 90 s
Temperature ( ), C
t = 120 s
Stress distribution (model I)
108
104
Radius (r ), mm
Fig. 37
Temperature distribution (model II)
200
st
an 300
of ce
m fro
ol m
d, b 400
m ott
m om
108
112
Radius (r ), mm
2.8
2.6
t = 2.0 s
2.4
100
200
Di
st
400
Di
Variation of density (model II)
112
108
1.0
104
Radius (r ), mm
0
400
st
an
pl ce
an fr
e om
(
), ce
m nt
m ra
200
0.1
0
100
Di
104
108
Radius (r ), mm 112
400
t = 1.0 s
300
1.0
200
0.5
100
104
108
Radius (r ), mm 112
t = 2.5 s
300
Di
st
a
200
pl nce
an fr
e om
(
), ce
m nt 100
m ra
l
108
112
104
Radius (r ), mm
Fig. 39
0.3
0.2
t = 2.0 s
300
st
a
pl nce 200
an fr
e om
(
), ce
m nt 100
m ra
l
104
112
Radius (r ), mm
Fig. 38
108
Radius (r ), mm
an
300
of ce
m fro
ol m
d, b 400
m ott
m om
Di
112
300
t = 1.0 s
2.4
100
1.0
400
t = 0.5 s
an
pl ce
an fr
e om
(
), ce
m nt
m ra
2.6
t = 1.0 s
st
200
a
pl nce
an fr
e om
(
), ce
m nt 100
m ra
l
Di
st
112
Radius (r ), mm
2.8
300
Di
Variation of volume fraction of solid phase

(model II)
400
300
t = 1.5 s
10.0
an
pl ce
an fr
e om
(
), ce
m nt
m ra
l
108
400
200
5.0
100
104
Di
st
Di 200
st
an
300
of ce
m fro
ol m
d, b
400
m ott
m om
Equivalent stress (), MPa
t = 0.5 s
2.4
100
1.0
4. D.Y. Ju and T. Inoue, Metallo-Mechanical

Simulation of Centrifugal Casting Process
of Multi-Layer Roll, J. Mater. Sci. Res. Int.,
Vol 2 (No. 1), 1996, p 1825
5. H. Matsui, D.Y. Ju, and T. Inoue, Inelastic
Behavior and Unified Constitutive Equations of SUS 304 at High Temperature, J.
Soc. Mater. Sci. (JSMS), Vol 41 (No. 466),
1992, p 11531159
6. D.Y. Ju, T. Inoue, and H. Matsui, ViscoPlastic Behaviour of SUS 304 Stainless
Steel at Ultra-High Temperature, Advances
in Engineering Plasticity and Its Application, D. Lee, Ed., 1993, p 521528
7. P. Perzyna, Thermodynamic Theory of Viscoplasticity, Advance of Applied Mechanics,
Vol 11, Academic Press, 1971, p 313354
8. K. Sorimachi and J.K. Brimacombe, Improvements in Mathematical Modeling of
Stresses in Continuous Casting of Steel,
Iron Steel., Vol 4, 1977, p 240245
9. K.S. Chan, S.R. Bodner, K.P. Walker, and
U.S. Lindholm, A Survey of Unified Constitutive Theories, Proc. Second Symp. Nonlinear Constitutive Relations for High Temperature Applications (Cleveland, OH),
1984, p 108112
10. P. Perzyna, The Constitutive Equations for
Rate Sensitive Plastic Materials, Arch.
Mech. Stos., Vol 20, 1968, p 499512
11. S.R. Bodner and Y. Partom, Constitutive
Equations for Elastic-Viscoplastic Strain
Hardening Materials, J. Appl. Mech., Trans.
ASME, Vol 42, 1975, p 385389
2.6
Density ( = 103), g/mm3
2.8
REFERENCES
1. J.R. Williams, R.W. Lewis, and K. Morgan,
An Elastic-Viscoplastic Thermal Stress
Model with Applications to the Continuous
Casting of Metals, Int. J. Num. Meth. Eng.,
Vol 14, 1979, p 19
2. T. Inoue, Metallo-Thermo-Mechanical CouplingAnalysis of Quenching, Welding
and Continuous Casting Processes, Berg.
Hutten. Monatsh., Vol 132, 1987, p 6375
3. T. Inoue and D.Y. Ju, Thermo-Mechanical
Simulation of Some Types of Steady Continuous Casting Processes, Advances in
Continuum Mechanics, O. Bruller, V.
Mannl, and J. Najar, Ed., Springer-Verlag,
1991, p 389406
posed in the section on numerical analysis are

verified by comparing the calculated results
with the experimental data for temperature on
the inner surface. The relaxation of temperature variation along time dependence of increment of molten metal and the occurrence
of the latent heat in solidifying domain is displayed.
may describe the stress and deformation in

the entire region of the solidifying process including liquid and solid states.
In the simulation of stress field, the development of stresses with a following increment of molten metal is presented. On the
other hand, when the effects of centrifugal
force are considered, the residual stresses in
the casting process would occur with the
compressive stress.
The theory and method of simulation pro-
The unified inelastic constitutive equation
culation is presented in this article as an example

of the application of theory and procedure developed in the proceeding. A modification of
Perzynas viscoplastic constitutive relationship
is proposed, which reflects actual liquid-solid
phase transformation during the casting process.
The results of simulation are also verified by
comparing with experimental data, and application of the technique to other operating conditions is carried out to obtain the fundamental
data of optimal design of the system.
A new algorithm for solving the solidification
problem with growing domain has been presented as an application of the theory in the section on numerical analysis. From simulated results of the centrifugal-casting process, the
following conclusions can be drawn:
108
Radius (r ), mm 112
Fig. 40
Equivalent stress at initial stage (model II)

Molten steel
Steel roll
Steel shaft
400
Stress (r), MPa
102
Free boundary
300
st
an
pl ce
an fr
e om
(
), ce
m nt
m ra
1.0
200
1.0
ym
r
Cooling water
Di
108
Radius (r ), mm 112
100
104
Copper roll sleeve

0
Steel pinch roll
400
102
Roll
300
bo
200
0
2.0
100
108
Vo
Free boundary
Cast strip
Di
104
Radius (r ), mm 112
(a)
Fig. 42
(b)
Twin-roll casting system (a) and model for simulation (b)
400
102
300
l
2.0
st
an
pl ce
an fr
e om
(
), ce
m nt
m ra
200
0
2.0
100
Di
104
108
Radius (r ), mm
Fig. 41
112
Residual stresses at the end of cooling (model II)
20
Distance from meniscus (y), mm
Stress (z), MPa
2.0
st
an
pl ce
an fr
e om
(
), ce
m nt
m ra
Stress (), MPa
Strip
20
40
Detail of region A
60
80
100
50
Fig. 43
40
30
10
20
Distance from center (x), mm
Finite-element division for continuous casting

by twin-roll method
Fig. 44
Velocity field. Vb 77.2 mm/s
20
1.1
14
18
21
25
28
40
60
142
111
95
80
190 C
179
170
167
100
50
25
178
169
162
140
20
13
17
20
25
40
60
108
50
0
80
Fig. 46
0.15
60
1.0
25
40
80
100
80
20
0.15
40
0.83
60
1.0
25
80
50
Distance from center (x ), mm
25
Distribution of temperature (a) and volume fraction of solid (b) (material A). To 190 C
200
Calculation
150
Experiment
V = 77.2 mm/s
100
0.0
20
8
12
40
4 MPa
60
4
80
20
30
4 MPa
40
60
40
50
80
5
V = 51.1 mm/s
0.0
20
18
50
0
20
40
60
80
100
100
Fig. 47
Variation of temperature on the surface of the

strip (material A). To 190 C
25
20
15
10
100
Fig. 48
100
Variation of temperature along the central line

(material A). To 190 C
100
50
Temperature (T ), C
60
V = 51.1 mm/s
20
20
0.83
20
40
V = 51.1 mm/s
100
92
50
V = 77.2 mm/s
150
30
V = 77.2 mm/s
20
190 C
0
100
Fig. 45
200
V = 51.1 mm/s
20
Temperature (T ), C
V = 77.2 mm/s
20
Isostress line in the strip (material A). V 77.2 mm/s
25
20
15
10

12. S.R. Bodner and A. Merzer, Viscoplastic
Constitutive Equations for Copper with
Strain Rate History and Temperature Effects, J. Appl. Mech., Trans. ASME, Vol
100, 1978, p 388394
13. J.L. Chaboche and G. Rousseler, On the
Plastic and Viscoplastic Constitutive Equation (Part 1), J. Pressure Vessel Technol.,
Trans. ASME, Vol 105, 1983, p 153158
14. J.L. Chaboche and G. Rousseler, On the
15.
16.
17.
20
10
Stress, MPa
20
40
20
0
18.
20
40
19.
20
40
Plastic and Viscoplastic Constitutive Equations (Part 2), J. Pressure Vessel Technol.,
Trans. ASME, Vol 105, 1983, p 158164
P. Ackermann, W. Kurtz, and W. Heinemann, In Situ Tensile Testing of Solidifying
Aluminium and Al-Mg Shells, Mater. Sci.
Eng., Vol 75, 1985, p 7986
T. Suzuki, K.H. Tacke, and K. Schwerdtfeger, Influence of Solidification Structure
on Creep at High Temperatures, Metall.
Trans. A, Vol 19A, 1998, p 28572859
T. Inoue et al., Benchmark Project on the
Application of Inelastic Constitutive Relations in Plasticity-Creep Interaction Condition to Structural Analysis and the Prediction of Fatigue-Creep Life, Part I, Proc. of
Subcom. Inel. Anal. High Temp. Mater.,
JSMS, Vol 1, 1991, p 15
P.J. Wray and M.F. Holmes, Plastic Deformation of Austenitic Iron at Intermediate
Strain Rates, Metall. Trans. A, Vol 6A,
1975, p 11891196
P.J. Wray, Onset of Recrystallization Dur-
20.
21.
22.
23.
24.
0
20
200
40
Roll boundary
60
Strip thickness (), mm
20
0
x
y
z
20
40
20
0
80
20
10
25.
150
100
26.
Experiment
Calculation
100
50
20
Fig. 49
60
100
27.
Casting speed (v), mm/s
Stress distribution at several locations (material

A). V 190 mm/s
Fig. 50
Computing and measuring results of strip thickness (material A)
28.
100
100
29.
400
300
1200 C
200
1240 C
C
1280
100
1470 C
1400
1000
360 C
420
400
500
600
1.5
600
400
(a)
T.P. Battle and R.D. Pehlks, Mathematical of
100
200
300
1470 C
1400
1000
300 C
360
420
400
A
600
0
300
SELECTED REFERENCES
400
500
200
100
600
400
Fig. 51
1240 C
1280 C
1200 C
200
200
ing the Tensile Deformation of Austenitic

Iron at Intermediate Strain Rates, Metall.
Trans. A, Vol 6A, 1975, p 11971203
C.E. Pugh and D.N. Robinson, Some Trends
in Constitutive Equation Model Development for High Temperature Behavior of
First-Reactor Structural Alloys, Nucl. Eng.
Des., Vol 42, 1978, p 269278
T. Inoue and S. Nagaki, A Constitutive
Modeling of Thermo-Viscoelastic-Plastic
Materials, J. Thermal Stress., Vol 1, 1978,
p 5361
D. Kujawski and Z. Mroz, A Viscoplastic
Material Model and Its Application to Cyclic Loading, Acta Mech., Vol 36, 1980, p
213230
Grill, J.K. Brimacombe, and F. Weinberg,
Mathematical Analysis of Stresses in Continuous Casting of Steel, Iron Steel., Vol 1,
1976, p 3847
D.Y. Ju, T. Inoue, and N. Yoshihara, Simulation of Vertical Semicontinuous Direct
Chill Casting Process of Cylindrical Ingot
of Aluminum Alloy, Trans. Jpn. Soc. Mech.
Eng. (JSME), Vol 55 (No. 513), 1989, p
12361243 (in Japanese)
D.Y. Ju and T. Inoue, Simulation of Solidification and Evaluation of Residual
Stresses During Centrifugal Casting, Proc.
Third Int. Conf. Residual Stresses, Vol 1,
Elsevier, 1991, p 220225
D.Y. Ju and T. Inoue, Simulation of Solidification and Viscoplastic Deformation in
the Twin Roll Continuous Casting Process,
Trans. Jpn. Soc. Mech. Eng. (JSME), Vol
57 (No. 537), 1991, p 11471154 (in Japanese)
S. Redner and C.C. Perry, Factors Affecting
the Accuracy of Residual Stress Measurements Using the Blind Hole Drilling
Method, Proc. Seventh Int. Conf. Experimental Stress Analysis (New York), Aug
1982, p 604616
A. Boley and J.H. Weiner, Theory of Thermal Stresses, Wiley, 1960
L.I. Rubenstein, The Stefan Problem, American Mathematical Society, 1971
300
200
100

(b)
Distribution of temperature (material B). (a) Vc 400 mm/s. (b) Vc 600 mm/s
Microsegregation in Binary Metallic Alloys,

Metall. Trans. A, Vol 22A, 1991, p 957980
R.M. Bowen, Theory of Mixture, Continuum
Physics, A.C. Eringen, Ed., Vol 3, Academic
Press, 1976
M.C. Flemings, Behavior of Metal Alloys in
the Semisolid State, Metall. Trans. A, Vol
22A, 1991, p 957980
H. Gouin, Variational Theory of Mixture in
Continuum Mechanics, Eur. J. Mech., B/Fluids, Vol 9, 1990, p 469491
T. Inoue, in Thermal Stresses, R.B. Hetnarski, Ed., Vol 3, Elsevier Science, NorthHolland, 1989
D.Y. Ju and T. Inoue, A Thermomechanical
Model Incorporating Moving Liquid/Solid

200
100
100
100
200
0
100
s = 0.1
s = 1.0
200
300
400
0
100
s = 0.1
s = 1.0
200
Perzyna (n = 1)
Perzynna
Perzyna + creep
50
0
1.5
300
0.5
1.0
0.0

(a)
400
500
600
400
300
200
100
600
400
300
200
100

(b)
(a)
Fig. 52
Distribution of solid fraction (material B). (a) Vc 400 mm/s. (b) Vc 600 mm/s
100
500
Perzyna (n = 1)
Perzynna
Perzyna + creep
50
0
1.5
0.5
1.0
0.0

(b)
1500
Dependence of constitutive relationships on

stress distribution. (a) Vc 400 mm/s. (b) Vc
600 mm/s
Fig. 55
Center of slab
Surface of slab
1100
100
100
900
700
300
400
500
(a)
1500
Center of slab
Temperature (T ), C
1300
x = 1 MPa
600
Surface of slab
1100
x = 1 MPa
200
5
10
300
400
200
Stress (x ), MPa
100
0
1.5
900
1.0
0.5
300
400
0
1.5
500
5
10
200
Stress (x ), MPa
Temperature (T ), C
1300
500
1.0
0.5
700
0
100
200
300
400
500
600
600
160
(b)
Fig. 53
600 mm/s
Temperature variation at center and surface of

slab (material B). (a) Vc 400 mm/s. (b) Vc
120
80
40
600
160
Distance from meniscus (x), mm

(a)
Fig. 54
120
80
40
Distance from meniscus (x), mm

(b)
Distribution of stress rx (material B). (a) Vc 400 mm/s. (b) Vc 600 mm/s

Interface and Its Application to Solidification
Process, Proc. Sixth Int. Conf. Mechanical
Behaviour of Materials (Kyoto), JSMS, 28
July 1991, Vol 5, p 119120
D.Y. Ju, Y. Oshika, and T. Inoue, Simulation
and Temperature in the Centrifugal Casting
Process, J. Soc. Mater. Sci. (JSMS), Vol 40
(No. 449), 1991, p 1218 (in Japanese)
D.Y. Ju, S. Takemura, and T. Inoue, Analysis
of Coupled Mode of Solidification and
Stresses the Centrifugal Casting Process, J.
Soc. Mater. Sci. (JSMS), Vol 41 (No. 464),

1992, p 751757 (in Japanese)
J.B. Leblond, G. Mottet, J. Devaux, and J.C.
Devaux, Mathematical Models of Anisotropic Phase Transformation of Steels and
Predicted Plastic Behavior, Mater. Sci. Technol., Vol 1, 1985, p 815822
K. Miyazawa and J. Szekely, A Mathematical
Model of the Splat Cooling Process Using the
Twin-Roll Technology, Metall. Trans. B, Vol
20B, 1989, p 391390
D. Peirce, D.F. Shih, and A. Needlemen, A
Tangent Modulus for Rate Dependent Solid,

Int. J. Comput. Struct., Vol 18, 1984, p 875
887
T.L. Sham and H.W. Chow, A Finite Element
Method for an Incremental Viscoplasticity
Theory Based on Overstress, Int. J. Comp.
Mech., Vol 34, 1989, p 143156
O.C. Zienkiewicz, The Finite Element
Method, McGraw-Hill, 1977

DOI: 10.1361/hrsd2002p391

Residual Stress Formation Processes

during Welding and Joining
W. Zinn and B. Scholtes, Universitat Kassel, Germany
KNOWLEDGE ABOUT FORMATION and

consequences of welding or brazing residual
stresses has increased considerably during the
past decades. A number of papers have made
fundamental contributions to this topic (Ref 1
3), and general aspects about origin and distributions of welding and brazing residual stresses
are now classified. Considerable interest in this
subject is based upon the fact that welding or
brazing residual stresses have various impacts on
the behavior of joined components. In this context, the most important viewpoints are:
Temperature
Time t 0 = 0 s
High amounts of total stresses as a conse
quence of superimposing loading and residual

stresses
Influence on local yield strength of components, depending on sign and multiaxiality of
residual stress state
Formation of cracks without external loads
Increasing the risk of brittle fracture by high
local stress values and/or highly multiaxial
residual stress states
Influence of residual stress distributions on
loading modes of existing cracks
Influence on fatigue strength and fatigue life
Influence on elastic stability, for example,
buckling
Influence on stress corrosion sensitivity in the
case of corrosive environment
Consequences of existing residual stress distributions on distortion of joined components
Subsequently restricted to metallic materials, an

overview is given about origin and assessment
of welding and brazing residual stresses. A focal
point is put on the basic principles of residual
stress development and the influence of important process parameters on amount and distribution of residual stresses.
Residual stresses are always consequences of
inhomogeneous elastic and/or plastic deformations on a macroscopic or microscopic scale (Ref
4). Consequently, all processes or process parameters of the welding or brazing operation that
introduce or influence inhomogeneous deformations have to be considered, if the origin of
joining residual stresses has to be outlined. In
the following chapters, only macro residual
Time t 1 > t 0
HAZ
Cold
Seam
Hot
HAZ
Cold
Impeded longitudinal shrinkage

High value of longitudinal residual stresses
Low amount of shrinkage
Transverse shrinkage not impeded

Low value of transverse residual stresses
Noticeable amount of shrinkage
Fig. 1
Shrinkage and impeded shrinkage as a consequence of inhomogeneous temperature distribution. HAZ, heataffected zone. Source: Ref 6

stresses are considered, which are assumed to
have similar consequences on components as
loading stresses (Ref 4, 5).
Welding Residual Stresses

Welding residual stresses are the consequences of the fluctuating inhomogeneous temperature distributions arising in the course of the
different welding processes in and around the
weld seam. Heating, as well as the cooling period, is of importance. This section outlines different aspects and influencing parameters that
contribute to the amount and distribution of
welding residual stresses.
Residual Stresses Due to Shrinking Processes. A very important origin of welding residual stresses is impeded shrinkage processes,
which occur when heated and cooled regions are
neighbored. The basic principle is that heated
volumes shrink during the cooling-down process
according to their coefficient of thermal expansion and the existing temperature difference. It
is assumed that melted volumes do not exert
forces on surrounding volumes. Consequently,
weld seams and surrounding volumes that have
not been melted during the welding process exhibit different thermal and, thus, residual stresses
(Fig. 1) (Ref 6). If shrinking is not impeded, thermal stresses relax completely and no residual
stresses remain after cooling down to room tem-
Distributions of longitudinal and transverse residual stresses parallel and perpendicular to a two-pass welded
aluminum plate. TIG, tungsten inert gas; ry and rl, stresses parallel to the weld toe; rx and rt, stresses transverse
to the weld toe. Source: Ref 7
Fig. 2
perature. In the case of the weld seam, starting

from the stress-free state of the melting pool, tensile thermal stresses build up during the cooling
process, which reach the temperature-dependent
yield strength of the material. Consequently, at
the end of the cooling process, tensile residual
stresses exist in the weld seam. If the amount of
restricted shrinkage is high enough, they may
reach the yield strength of the weld seam material.
Materials volumes that are at a certain distance from the weld seam and are not melted
during the welding process undergo impeded
thermal strains during the heating period as well.
It follows that compressive thermal stresses
build up, yielding compressive plastic strains
when the compressive temperature-dependent
yield stress is reached. These volume shrinkages
build up tensile stresses when thermal contraction during cooling is impeded. The amount of
tensile residual stresses, in this case, is directly
related to compressive strains during the heating
process. For equilibrium reasons, longitudinal
residual stresses change sign over the width of
the plate (Fig. 2) and result in an inhomogeneous
distribution of longitudinal residual stresses (Ref
7).
Up to now, only longitudinal residual stress
components have been considered. But, as a consequence of inhomogeneous longitudinal
stresses, inhomogeneous shrinkage in transverse
direction results as well. Because inhomogeneous shrinkage also means impeding of shrinkage, residual stresses build up, also in the transverse direction, even if no external forces impede
shrinkage in the transverse direction. However,
residual stress amounts in transverse direction
reach only about one-third of the values in longitudinal direction. On the other hand, transverse
residual stresses may also reach the yield
strength of the material, for example, after multipass welding.
Because the amount of restraint is usually
higher at repair welds compared to new constructions, residual stresses are typically more
severe and especially have to be taken into account.
Residual Stresses Due to Quenching Processes. Particularly in the case of thicker plates,
considerable temperature differences build up
between the near surface and the core layers of
the plate during the cooling process. The resulting thermal residual stresses may exceed the
yield strength of the material, resulting in plastic
deformations. Obviously, these inhomogeneous
plastic deformations lead to residual stresses after cooling. If no other processes were active,
compressive residual stresses in near surface layers and tensile residual stresses in the core would
be expected. However, in most cases, by far,
quenching residual stresses develop together
with residual stresses due to shrinkage processes
and, if present, residual stresses due to phase
transformations.
Residual Stresses Due to Phase Transformations. Residual stresses as a consequence of
phase transformations during welding processes
Residual Stress Formation Processes during Welding and Joining / 393
Time = t 1
Temperature
Transformation
t1 < t2
Temperature
Time = t 2
Tra
S
n
Im sfor hrin
mi ma ka
g
Tra nent tion e
a
ns
t
for rans ffec
te
m
f
Sh atio orm d
a ti
rin n
o
ka aff
ge ec n
ted
Fig. 3
Shrinkage and expansion in example of phase transformation of sufficient highly treated areas. Source: Ref 6
Table 1 Materials data for the calculation

of brazing residual stresses
Material
E, GPa
,106K1
ZrO2
Si3N4
Ag
CK 45
210
280
73
188
0.3
0.27
0.3
0.3
10.6
2.5
20.0
14.0
E, Youngs modulus; m, Poissons ratio; , coefficient of thermal expansion. Source: Ref 12
Interaction of impeded shrinkage and transformation for different cooling velocities. Example
1, low velocity; Examples 2 and 3, high velocity. Re, yield
strength; r, stresses; , coefficient of thermal expansion; E,
Youngs modulus; DT, temperature differences; S, stiffness.
Source: Ref 6
occur if, during the cooling-down period of the

heated weld seam, phase transformation processes connected with volume changes occur locally (Fig. 3)(Ref 6). Inhomogeneity of transformation processes, which is a necessary
prerequisite for the formation of residual
stresses, may be attributed to different reasons,
for example, different peak temperatures during
the welding process, different cooling velocities,
or local changes of chemical composition. In the
case of steels, transformation from austenite to
ferrite, bainite, or martensite is connected with a
characteristic volume increase. Consequently,
one would always expect compressive residual
stresses in transformed volumes, if transformations occur simultaneously and balance tensile
residual stresses in adjacent volumes.
In practice, this is not the case. Moreover,

shrinkage and transformation processes overlap,
and the amount as well as the sign of the resulting residual stresses strongly depend on the temperature range in which transformation takes
place. The consequences are schematically outlined in Fig. 4. It shows, as a function of temperature, the tensile or compressive yield
strength of a ferritic-pearlitic or a bainitic microstructure. Starting at high temperatures, tensile
stresses build up due to shrinkage processes and
according to the temperature-dependent yield
strength of austenite. If phase transformations
occur as a consequence of volume increase, tensile stresses rapidly decrease and change to compression. If, in the transformed volume, a ferritic-pearlitic microstructure develops, maximum
compressive stresses are limited by the respective compressive yield strength of the phases created (Example 1 in Fig. 4). If a martensitic transformation takes place, because of the high yield
strength, the highest compressive stresses can
develop. At the end of the transformation process, compressive residual stresses decrease because of continuous temperature decrease along
with accompanying shrinkage (Examples 2 and
3 in Fig. 4). Depending on Youngs modulus, E,
the stiffness, S, of the component, the coefficient
of thermal expansion, , and the remaining temperature difference, DT, up to room temperature,
compressive or even tensile residual stresses result. Highest tensile residual stresses are expected if the transformation process is finished
at relatively high temperatures. In that case, the
tensile yield strength of the ferritic-pearlitic microstructure may be reached by the residual
stresses. From the schematic diagram in Fig. 5,
it can be seen that the lower the transformation
temperature, the more effective the consequences of transformation processes on the resulting residual stresses will be. Consequently,
pronounced effects of transformation processes
Fig. 4
Fig. 5
Influence of heat input on transformation temperature, Tt, and resulting consequences on shrinkage tensile or
transformation compressive residual stresses. RS, residual stress. Source: Ref 8

perpendicular to the weld seam. In the case of
low heat input (Curve 3, see also Fig. 4 and 5),
one can see that shrinkage and, especially, transformation processes lead to a reduction of tensile
residual stresses or even compressive residual
stresses, resulting in a characteristic W-shape of
the residual stress distribution. In the case of
high heat input (Curve 1), one can expect that
immediately after phase transformation in the
heated weld seam area compressive stresses exist
and that cooling down to room temperature results in a buildup of high tensile stresses across
the weld seam.
The influence of the chronological sequence
of the transformation processes is especially pronounced in voluminous components. In this
case, a considerable time may elapse before
transformation processes in the fast-cooling
outer layers and the slow-cooling core volumes
are finished. If transformation-induced stresses
resulting from the inhomogeneous volume
changes lead to plastic deformations, after temperature compensation residual stresses occur
with positive sign in volumes having undergone
transformations first and negative sign in volumes transformed last. In practical cases, however, this effect always occurs in connection with
residual stresses due to pronounced temperature
gradients in directions perpendicular to the weld
on the resulting residual stress distributions can

be expected, for example, for steels with high
yield strength at high temperature or for extremely high cooling rates.
The diagram sketched in Fig. 6 shows the influence of increasing heat input on the distribution of transverse residual stresses along a line
RS
1
RS
y
x
RS
Weld toe
HAZ
ea
ti
np
ut
RS
Fig. 6
Influence of heat input on transverse residual

stress distribution. RS, residual stress. Source:
Ref 9
Brazing temperature
Room temperature
seam. As a consequence, transformation-induced

compressive residual stresses are often observed
with smaller amounts in surface layers compared
with volumes near the core.
Summary. Welding residual stresses are a
consequence of the interaction of the following
separate processes:
Impeded expansion and shrinkage as a result
of inhomogeneous temperature distributions
Quenching effects
Consequently, the algebraic sign and amount of

residual stresses in the weld seam and the heataffected zone depend on the transformation temperature, Tt, which is determined by the timetemperature-transformation diagram of the
material and also the cooling velocity. Obviously, the local chemical composition plays an
important role. In addition, the local temperature-dependent yield strength of the material and
the degree of stiffness of the construction is decisive.
One can see that the basic processes of the
development of welding residual stresses are
widely understood. A prediction of welding residual stresses in individual cases, however, is
quite difficult because of the relatively great
number of influencing parameters and their interdependencies. To that end, further consequent
research and the correct inclusion of the experimental experiences in appropriate finite-element
(FE) models are necessary.
Brazing Residual Stresses
thA > thB
Fig. 7
Distortion and residual stresses shown schematically in a brazed two-layer compound. Source: Ref 12
Normalized thermal residual stresses

at the surface, MPa/K
2.0
Variations with hb (0 hb 0.3) mm
1.5
1.0
0.5
0.0
0.5
1.0
ZrO2
hc Si3N4
1.5
steel
hs steel
2.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
hb
The solidification behavior of the solder

The differences in the thermal expansion be3.5
hc/hs
Fig. 8
In modern structures, components are very often required to possess complex properties such
as sufficient ductility together with resistance
against abrasive wear and high temperatures.
These requirements cannot be satisfied by monolithic materials alone. In these cases, brazed
compounds have been demonstrated to be very
useful because materials of different properties
can be joined by this method. Very important
examples include cemented carbide tools and
brazed ceramic-metal compounds combining
hardness and chemical resistance with strength
and ductility of appropriate metallic partners. In
the case of brazing, joining of two partners is
achieved by a melted hard solder, while the materials joined remain well below their solidus
temperature. Special active solder materials have
been developed to meet the needs of economical
industrial applications. During the cool down
from joining temperature, brazed compounds develop a complex residual stress state. The most
important influencing parameters are:
Influence of the thickness ratio hc /hs of the ceramic and the steel on brazing residual stresses in the outer surfaces
of the compound. hc, thickness of ceramic; hs, thickness of steel; hb, thickness of solder. Source: Ref 12
havior of the joint materials
Elastic and plastic properties of the joint part-
ners
Geometry of the joint compound

Cooling down conditions (Ref 10, 11)
Residual Stress Formation Processes during Welding and Joining / 395

Thermal residual stresses during the cool down
from brazing temperature and residual stresses
after cooling may be detrimental for the compound and cause distortions or failure during the
manufacturing process or in service. Therefore,
knowledge about residual stress states and their
variation with brazing and geometrical conditions is of high importance in order to realize
compounds optimized for practical applications.
Characteristic Residual Stress Distributions in Brazed Components. Basic principles
of the development of brazing residual stresses
can be outlined with the aid of Fig. 7. It schematically shows a two-layer compound with layers A and B and thermal expansion coefficients
thA thB (Ref 12). Starting from a stress-free
state, thermal stresses build up as soon as the
solidified braze material leads to adhesion between the two layers, because of the differences
between the thermal expansion coefficients. This
is outlined on the right hand side of Fig. 7. In
addition to a global bending of the compound,
local bending effects on the lateral surfaces also
can be detected. Consequently, a relatively complex residual stress distribution results. In this
case and if similar thicknesses and elastic properties of the layers are assumed, compressive
(tensile) residual stresses occur at the outer free
surfaces of layer A (B). Near the joining face,
residual stresses with opposite sign develop.
Consequences of important parameters on resulting residual stress distributions and amounts
have been studied theoretically using analytical
as well as FE models (Ref 12). In most cases,
purely elastic behavior is assumed, which, however, is not always justified. As an example, Fig.
8 shows the influence of thickness ratio hc /hs (hc,
thickness of ceramic; hs, thickness of steel) for
compounds of steel Society of Automotive Engineers (SAE) 1045 and Zr02 or Si3N4 ceramic.
The calculations are based on a three-layer plate
with silver as braze material and the materials
properties listed in Table 1. The longitudinal residual stresses, normalized to the temperature
difference, DT, are plotted as a function of thickness of the compound layers. The coefficient DT
is the temperature difference during cooling
Examples of residual stresses determined in different brazed metal-ceramic compounds.

Source: Ref 12
Fig. 9
from brazing temperature, for which the brazing

solder shows elastic behavior. In each component, linear stress-depth distributions occur, resulting from a superposition of bending and tension or compression loading. In Fig. 8 one can
clearly see that tensile residual stresses in the
ceramic surfaces decrease considerably with decreasing thickness ratio. If hc /hs 0.5, compressive surface residual stresses in the ceramic
surfaces always exist. Because differences in
thermal expansion coefficients are smaller for
SAE 1045/Zr02 compounds compared with SAE
1045/Si3N4, amounts of brazing residual stresses
are also larger in the latter case. In most cases,
experimental residual stress determinations in
brazed compounds are carried out by x-ray or
mechanical methods. The results presented in
Fig. 9 confirm the calculations of Fig. 8. In this
case, x-ray measurements were carried out on the
outer surfaces of plates with the dimensions 20
20 mm2and the thicknesses indicated.
A considerable uncertainty of calculations exists in the ignorance about plastic deformations
in the brazed components during the cooling
process, if thermal stresses exceed the temperature-dependent yield strength. In a number of
cases, neutron stress analyses showed clear evidence of plastic deformations in brazed metallic
components and also in the metallic part of ceramic-metal joints (Ref 13). A typical example
is presented in Fig 10. The diagram shows the
distribution of residual stresses across the center
of a 19 19 mm2 brazed SAE 1045/WC-Co
compound. In this case, a Cu-Mn-Co hard solder
was used. In the WC-Co cemented carbide, a
linear distribution of the residual stresses as a
function of the distance from the surface is observed with a maximum tensile residual stress of
approximately 550 MPa immediately at the free
surface. Near the solder layer, compressive residual stress amounts of more than 600 MPa
are found. In the steel plate, compressive residual stresses of about 250 MPa exist at the free
surface, which decrease linearly and change sign
at a distance of about 1.5 mm from the surface.
Near the solder layer, a layer of about 2 mm
Fig. 10
thickness with constant tensile residual stresses

of about 250 MPa is observed. Measurements
using x-ray diffraction or hole-drilling techniques confirm these distributions in near surface
layers.
In the meantime, more complex residual stress
states in brazed components have also been analyzed. A characteristic example is the examination of joints of steel and cemented carbides
similar to cutting tools (Ref 13, 14). The typical
geometry investigated is shown in Fig. 11. Cemented carbide tungsten-carbon containing 10
wt% cobalt and 17.3 wt% titanium, tantalum,
and niobium was brazed into a slit of a steel
block made of SAE 4140. Brazing was carried
out using copper as braze foil together with a
nickel mesh in order to keep the thickness of the
brazing gap constant during the brazing process.
FE-calculations as well as experimental stress
analyses were carried out. Residual stresses near
the surfaces of the components are essentially
determined by local or global bending moments
due to different amounts of shrinkage during
cooling and also influenced by the geometry of
the parts under investigation. One can see a
rather good agreement between measurement
and calculations. Such investigations are very
helpful to assess strength or failure behavior of
real components. Already during the brazing
operation, thermal tensile stresses may produce
cracks in brittle materials. In addition, location
of crack initiation as well as direction of crack
propagation under fatigue conditions is clearly
determined by existing residual stress fields (Ref
16).
Summary. Brazing residual stresses play an
important role for the correct function of components during service. Because of the different
thermal expansion coefficients and the temperature cycles involved during the brazing process,
they are, in most cases, inevitable. Because of
the number of influencing parameters, however,
strategies can be developed to avoid detrimental
residual stress distributions, if the basic processes are well understood. In this context,
adapted geometries play an important role. Me-
Depth distribution of residual stresses rxRS in the center of an SAE 1045/WC- Co compound determined by
neutron diffraction. Source: Ref 13
Neutron diffraction
FE calculation
a
RS
X
RS
(MPa)
Y
RS
(MPa)
X
Neutron diffraction
FE calculation
a
y
RS
Y
Cemented carbide
x
z
a
Distance a, mm
Neutron diffraction
FE calculation
RS
(MPa)
Y
RS
(MPa)
X
RS
X
Distance a, mm
Fig. 11
Steel
a
RS
Y
y
x
x
z
2a
a = 12.7 mm
b = 5 mm
c = 2/3 a
Neutron diffraction
FE calculation
Distance a, mm
Distance a, mm
RS
Calculated and measured residual stress distributions in a model tool geometry made of steel SAE 4140 and cemented carbide. Left, rRS
x component; right, ry component;
above, residual stresses in steel; below, residual stresses in cemented carbide. Source: Ref 15
chanical as well as diffraction techniques for residual stress analysis are available. Analytical
and FE-methods to predict brazing residual
stress distributions and to assess the importance
of individual process parameters are also well
developed.
REFERENCES
1. E. Macherauch and H. Wohlfahrt, Origin of
Welding Residual Stresses, Materialprufung, Vol 19, 1977, p 272280 (in German)
2. H. Christian and F.-X. Elfinger, Residual
Stresses in Weld Seams, Der Maschinenschaden, Vol 51 (No. 3), 1978, p 124130
(in German)
3. H. Wohlfahrt, Welding Residual Stresses,
Origin, Calculation, Assessment, Eigenspannungen, E. Macherauch and V. Hauk,
Ed., DGM-Verlag, Oberursel, 1983, p 85
116 (in German)
4. E. Macherauch and K.-H. Kloos, Origin,
Measurement and Evaluation of Residual
Stresses, Proc. Int. Conf. Residual Stresses
in Science and Technology, Vol 1, Garmisch-Partenkirchen, DGM-Verlag, Oberursel, 1986, p 326
5. V. Hauk, Structural and Residual Stress
Analysis by Nondestructive Methods, Elsevier, Amsterdam, 1997
6. H. Wohlfahrt, Consequences of Austenite

Transformation for the Formation of Welding Residual Stresses, Harterei-TechnischeMitteilungen, Vol 41 (No. 5), 1986, p 248
257 (in German)
7. W. Zinn, Investigation of the Fatigue Behaviour of the Butt Welded Al-Alloys
AlMg3
and
AlMg4.5Mn,
Schweitechnische
Forschungsberichte,
Bd. 34, DVS-Verlag, 1990 (in German)
8. T. Nietschke-Pagel and H. Wohlfahrt, The
Generation of Residual Stresses Due to
Joining Processes, Residual StressesMeasurement, Calculation and Evaluation, V.
Hauk, H. Hougardy, and E. Macherauch,
Ed., DGM Informationsgesellschaft Verlag,
1991, p 121134
9. T. Nietschke-Pagel, Residual Stresses and
Fatigue Behaviour of Welded Fine-Grained
Steels, Ph.D. dissertation, University Kassel, 1994 (in German)
10. O.T. Janeu, D. Munz, B. Eigenmann, B.
Scholtes, and E. Macherauch, Residual
Stress State of Brazed Ceramic/Metal Compounds, Determined by Analytical Methods
and X-Ray Residual Stress Measurements,
J. Am. Ceram. Soc., Vol 73, 1990, p 1144
1149
11. L. Pintschovius, N. Pyka, R. Kumaul, D.
Munz, B. Eigenmann, and B. Scholtes, Experimental and Theoretical Investigation of
12.
13.
14.
15.
16.
the Residual Stress Distribution in Brazed

Ceramic-Steel Compounds, Mater. Sci.
Eng., A 177, 1994, p 5561
B. Scholtes, Recent Investigations on the
Origin and Distributions of Residual
Stresses in Ceramics and Metal-Ceramic
Compounds, Contraintes residuelles et nouvelles technologies, CETIM, 1990, p 6171
K. Bing, Residual Stresses and Distortion
of Different Metallic Materials Brazed with
Steel SAE 1045, Ph.D. dissertation, University Karlsruhe (TH), 1995 (in German)
L. Pintschovius, B. Schreieck, and B. Eigenmann, Neutron, X-Ray, and Finite-Element Stress Analysis on Brazed Components of Steel and Cemented Carbide,
MAT-TEC 97, A. Lodini, Ed., ITT-International, Gournay sur Marne, 1997, p 307
312
K. Bing, B. Eigenmann, B. Scholtes, and E.
Macherauch, Brazing Residual Stresses in
Components of Different Metallic Materials, Mater. Sci. Eng., A 174, 1994, p 95
101
B. Schreieck, B. Eigenmann, and D. Lohe,
Residual Stresses and Failure of Brazed
Joints of Cemented Carbide and Steel Similar to Cutting Tools, in Joining Ceramics,
Glass and Metal, DVS-Berichte Band 184,
DVS-Verlag GmbH, Dusseldorf, 1997, p
140144

DOI: 10.1361/hrsd2002p397

Residual Stresses in Powder-Metal

Processing
P. Ramakrishnan, Indian Institute of Technology, Bombay, India
POWDER METALLURGY (P/M) is concerned with producing metal and alloy powders
and converting them to semifinished or finished
parts for several consumer and industrial products, including simple household appliances,
computer disk drives, surgical implements, and
sophisticated components of satellites. Products
of P/M include porous materials, self-lubricating
bearings, filters, a variety of ferrous and nonferrous structural parts, electrical contacts, tool
steels, cemented carbides, diamond tools, friction materials, refractory-reactive and high-tem-
perature materials, and metal-matrix composites.

The P/M process conserves energy and raw materials, and minimizes machining and scrap
losses, resulting in cost reductions. Thus, the
process competes with traditional metalforming
processes such as casting, working, machining,
rolling, forging, and so forth. Powder processing
involves a sequence of operations depending on
the end-product requirements as shown in Fig.
1. These include the traditional low-cost components produced by the press-sinter technique,
another category of cost-effective and high-per-
Metal powder production

(Atomization, chemical, electrochemical, and mechanical processes)
Mixing
(additives and lubricants)
Compaction
Warm compaction
Hot
Die pressing
Isostatic pressing
Rolling
Spraying
Pressureless sintering
Cold
Die pressing
Isostatic pressing
Rolling
Slip casting
Injection molding
Sintering
(Atmosphere, vacuum, high temperature)
Fig. 1
Solid phases
Activated
Liquid phase
Infiltration
Rolling
Finishing:
Repressing
Extrusion
Machining
Resintering
Wire drawing
Heat treatment
Sizing
Forging
Rapid prototyping
General P/M process
formance parts that require advanced P/M processing techniques, and those components in
which the P/M approach will lead to enhanced
mechanical properties and improved performance characteristics. Thus, in addition to the
widely used press-sinter method a variety of processing techniques such as hot pressing, isostatic
compaction, powder rolling, powder extrusion,
powder preform forging, high-temperature sintering, hot isostatic pressing (HIP), injection
molding, warm compaction, rapid prototyping,
and so forth, have been developed, and their use
is increasing in P/M processing. Powder metallurgy has emerged as an advanced manufacturing technology for the mass production of precision net-shape high-performance components
for the automotive, general engineering, electrical, electronic, and aerospace industries. According to the information released by the U.S. Department of Commerce, Bureau of the Census,
the U.S. metal powder market in 1996 was of the
order of $1.9 billion and included iron, aluminum, copper, tungsten, molybdenum, titanium,
tantalum, superalloys, precious metal, and other
nonferrous metal powders, paste, and flakes. The
North American P/M parts business is estimated
to be of the order of $2 billion, and is the largest
P/M market in the world, twice the size of Europe and Japan, each of which account for about
$1 billion in sales of conventional P/M parts.
During the 1998 Powder Metallurgy World Congress and Exhibition in Granada, Spain, it was
reported that the metal powder shipments increased almost 12% in 1997 to 440,843 metric
tons. North American iron powder markets have
invested heavily in new capacity and new products with iron powder production capacity exceeding 590,000 metric tons (Ref 1). Powder
metallurgy is reported to be a high-growth industry and a 21st century technology. It has a
significant impact on design viability and economics in key industries. In the aircraft industry,
the cost of a design using P/M components is
37% lower than a machined component. This has
resulted in a market growth from a 0% share in
1986 to nearly 50% in 1997. It has been reported

that new commercial jet aircraft engines made
by Pratt and Whitney and General Electric Company contain a total weight of P/M superalloy
extruded forging ranging from 680 to 1995 kg/
engine (Ref 2). Powder metallurgy is the only
way to produce certain high-temperature alloys.
Powder-Metallurgy (P/M)
Processing
Ferrous P/M Parts
The use of high-performance sintered ferrous
materials has been increasing steadily in recent
years in the automotive sector because of the
cost-effective manufacturing capability of P/M
to produce products with improved structural
and functional performance along with reliability. The commercial use of sintered iron bearings
as a less expensive substitute to the expensive
bronze bearings initiated by General Motors,
USA, during early 1930s. Germany developed
sintered iron driving bands as a substitute to copper during the mid-1930s. During World War II,
both Germany and the United States produced
limited quantities of large sintered iron bearings
for machine components and war vehicles,
thereby demonstrating the economic viability of
sintered ferrous products as a substitute for cast
and machined steel components. A major consumer of P/M parts is the automotive market,
accounting for about 70 to 80% of total produc-
tion. One can expect the continuing influence of

the automotive industry on the fortunes of the
P/M industry. According to the forecast of industrial experts, the global car production will grow
from its 1994 level of 35 million units/yr to 40
million units/yr by the year 2005, thereby providing ample opportunities for P/M parts production (Ref 3). Furthermore, a typical 1998
American family vehicle contains 14.7 kg of P/M
parts, about 5% increase over 1997. The forecast
indicated a family vehicle will contain 22.6 kg
of P/M parts by the year 2000. Automotive major growth areas for P/M include engine transmission, exhaust systems, and antilock brake
systems (ABS). Powder-forged connecting rod
and steel bearing caps are recognized examples
of reliability, innovation, and cost savings. The
American automotive industry is seriously considering the use of P/M in pinions and pinion
carrier frames in automatic transmissions to improve dimensional accuracy and reduce manufacturing costs. The major material in the powder
blend would be steel, consuming about 81,700
metric tons, which represents about 24% of the
total current shipments of iron and steel powder
annually for all applications in North America
(Ref 4). It is expected that the pinion carrier
frame market will be an important new application for steel powder during the next decade.
The market for stainless steel products is increasing because of the newer demand of exhaust system flanges, sensor rings in ABS systems, and so forth. The stainless steel powder
shipments are expected to grow more than 10%
Metal powders, additives, and lubricants
Mixing
Compacting
Sintering
Infiltration
Repressing
Sizing
Resintering
Finished part
Fig. 2
Basic P/M process for steel parts
Finished part
Finished part
in 1998 in the United States. Powder metallurgy

high-speed steels are finding increasing usage for
cutting automotive gears, aerospace parts, tooling for injection molding, and P/M dies. The
world P/M tool steel market is of the order of
12,000 metric tons.
Powder Metallurgy Parts Manufacturing.
The widely used press-sinter P/M steel parts
manufacturing process is shown in Fig. 2. Carefully selected metal powders are mixed with suitable additives and lubricants to get the desired
alloy composition, compacting the powder mixture in a shaped die under pressure at room temperature so as to produce a green compact of
required density and strength. The compaction
pressure will depend on the characteristics of the
powder such as whether it is elemental or prealloyed, its particle size, shape, size and shape distribution, microstructure, phases present, chemistry, and residual stresses in the powder and the
size, shape, and green density of the compact.
After the application of pressure through the
punches, the compact has to be ejected from the
die to produce the green compact. Carbon is usually added as finely powdered graphite and 0.5
to 1% dry powdered lubricant such as wax, stearic acid, or metallic stearate to reduce the friction
during compaction. Green compacts are sintered
at elevated temperature below the melting point
of base metal in a sintering furnace with suitable
atmospheres. During sintering, the mechanical
bonds of the powder particles are converted to
metallurgical bonds, thereby leading to densification and improvement in mechanical properties. The dimensional changes during sintering
will depend on the composition of ferrous alloys
and the density of the green compact. Powder
metallurgy provides precise control of materials
and properties and permits wide variation in
physical and mechanical properties, at the same
time ensuring performance characteristics of
consistent uniformity. Practically any desired
metal, alloy, or mixture of metals, including
those combinations not available in wrought
form, can be processed through P/M.
A variety of controlled porosity products are
available, ranging from metal foams, filter, and
self-lubricating bearings to porous products to
diffuse the flow of gases or liquids. Powder metallurgy products permit quieter and smooth
operation by damping vibration and noise. Another advantage is the production of complexshaped parts and the capability to reduce the subassemblies by a proper die design and tooling or
joining during sintering or infiltration. The additional operations such as repressing and resintering and/or metal infiltration can provide an
increase in the density and mechanical properties
of the P/M product. The physical properties of
the P/M steel part can be optimized by a postsintering heat treatment, which requires proper
selection of materials, heat treating parameters,
and processing equipment. If necessary the parts
may be machined, heat treated, or densified by
infiltration, repressing, and resintering. Although
the P/M technology has been used to produce
net-shape products, occasionally some machining operations such as tapping, threading, and
transverse holes may be necessary. In such in-
Residual Stresses in Powder-Metal Processing / 399

stances, machinability of P/M parts can be improved with appropriate treatments. Powder metallurgy products may be subjected to mechanical
working as in the case of cast product or may be
directly processed from powders by rolling, extrusion, or forging with additional advantage.
Parts may be consolidated to full density by hot
pressing or by HIP as in the case of alloy steel
powders. Other relatively more recent consolidation techniques include high-temperature sintering, warm compaction, and metal-injection
molding (MIM) and rapid prototyping.
Powder Characteristics
The basic raw material for P/M processing is
the metal powder, and successful processing will
depend on the characteristics of the powders. A
knowledge of the characterization of powders is
fundamentally important in many critical stages
of the processes such as mixing, compaction,
sintering, and postsintering operations that will
govern the final structure and properties of the
products. However, the characterization of powders presents many challenges because of the
complexity of properties and the large number
of variables involved (Ref 5, 6). Even for a single
particle, one can distinguish between the characteristics of the material and the characteristics
introduced by the manufacturing method. The
material characteristics include chemical composition, theoretical density, melting point, and
structure. The characteristics introduced by the
manufacturing method include particle size,
shape, microstructure, density, reactivity, and
surface conditions. Since the powders generally
consist of a large number of particles with varying characteristics, it is essential to distinguish
the characteristics of a mass of powder, including size and shape distribution, surface area, apparent density, tap density; flow and friction conditions of the powders; size, shape, and
distribution of porosity and microstructure
within the powders; and internal stresses and
compressibility. Thus, there are large number of
variables and properties, and the general characteristics of the powders depend on the manufacturing method and treatment of the powders.
Powder Production. Important methods for
the production of powders can be classified into
(1) atomization of molten metal, (2) chemical
reactions, (3) electrochemical methods, and (4)
mechanical processing of solid materials. Several techniques are available in each category;
for example, the atomization of molten metal can
be achieved by liquids such as water, oil, different gases, vacuum, ultrasonic, and centrifugal
methods. Chemical methods are reduction of
metal oxides and other compounds with carbon
or other gases, decomposition of metal hydrides
and carbonyls, condensation of metal vapors,
and precipitation of metals from salt solutions.
Electrolytic deposition can be from aqueous solutions or fused salts. Various types of milling
and machining methods are used in the mechanical processes. In addition, a variety of techniques are used to produce ultrafine and nanopowders with special characteristics. The
selection of a particular method will depend on

the desired characteristics and intended application of the powders and the economics of the
process.
Iron and steel powders represent a major part
of the metal powders currently produced.
Among the steel powders, the stainless steel and
tool steel powders share is relatively low, of the
order of 5%; however, one can expect a high
growth rate on account of their increasing usage.
Although a large number of methods are available for the production of iron powders, the reduction of iron oxide and atomization of molten
metal dominated the market by virtue of their
inherent technical and economic development
potential. The major difference in the two powder production processes lies in the development
of the crude powders (Ref 7). In the sponge iron
process, crude powder is porous and irregular,
while in the atomizing process the particles are
denser and more regular. The raw material iron
ore or mill scale is reduced in one or two steps.
The single-step reduction process consists of reduction of oxides by hydrogen in a belt furnace
or in a fluidized bed, while in the two-step processsuch as Hoganasthe iron ore or mill
scale is first reduced to an intermediate product
known as sponge iron in tunnel kilns. The
sponge is crushed and ground, and the powders
further reduced and annealed to the final iron
powder.
In the atomization process, a high-carbon (4%
C) iron is air atomized and annealed for decarburizing the particle core and reduction of the
particle surface oxide. Another method consists
of shotting or atomizing high-purity pig iron into
rather coarse particles, which after ball milling
are treated in a reduction annealing process to
remove carbon and oxygen, as in the case of
Quebec metal powder process. In the Domfer
process, the shots are ball milled and mixed with
fine mill scale and then annealed to make iron
powder. Direct-water atomizing a molten steel
to the desired particle size distribution and then
annealing was subsequently adopted by the leading manufacturers throughout the world. For P/M
structural parts, iron powder or alloyed iron
powder is mixed with copper, graphite, nickel,
other elemental additives and a lubricant before
compaction. Recent advances are in the development of highly compressible powders, alloying methods, and mixing technology of the constituents.
To improve the homogeneity of steel powders,
prealloyed powders have been developed by
melting alloying elements with iron and then atomizing the melt. These powders have similar
chemical characteristics but a reduced compressibility, and this reduction in compressibility is
proportional to the type and amount of alloying
elements. However, the loss in compressibility
can be made up by techniques such as repeating
the basic compaction and sintering stages or by
hot pressing, HIP, or by powder forging. The
compressibility of prealloyed steel powders
could be improved with a small addition of molybdenum. Another development is the partially
prealloyed powders, in which alloying elements

are bonded by diffusion to iron powder particles
in order to minimize powder segregation and
compositional variations in the individual sintered components. Many ferrous alloy compositions with copper, nickel, and molybdenum
have been developed (Ref 8, 9). The mixing of
powders is influenced by the powder characteristics such as particle size, shape, size and shape
distribution, and composition. Often in the diffusion-alloyed powder, diffusion of alloying elements during sintering is incomplete, leading to
heterogeneous microstructures in the product.
The mixed powders have some drawbacks such
as demixing, poor flow and die fill, dusting of
fine particles, and relatively lower green strength
after compaction. These problems have been
solved by utilizing a new concept in which a
binder acts as a glue to bond the smaller particle
size alloying elements to the larger iron-base
particles (Ref 1012). Different alloying methods for ferrous powders are shown schematically
in Fig. 3. Thus, the alloying method of ferrous
materials significantly influences the distribution
of alloying elements, compressibility, and the
nature of porosity and microstructure of the sintered product.
Molybdenum and nickel are commonly used
in low-alloy P/M steel powders for increasing
the strength, hardenability, toughness, and fatigue resistance. The oxides of these elements
could be reduced during the annealing treatment
of the water-atomized powders. However, these
alloying elements are more expensive, and efforts are being made to replace them with less
expensive manganese and chromium. In P/M
steels, these elements are generally maintained
below 0.3% because of the difficulty in reducing
the oxides during annealing (Ref 13). However,
this problem can be solved by mixing the base
iron or prealloyed powders with higher-carbon
ferrochrome and ferromanganese and sintering at
1250 C for the reduction of chromium or manganese oxides. Sintered parts containing chromium and manganese have a tendency to produce larger dimensional changes, and the same
have been overcome with the addition of nickel
(Ref 14). Oxygen pickup of manganese- and
chromium-containing steels can be reduced by
using oil as the atomizing medium instead of water. Further oil atomization can be used for carbon-containing (6% C) rapidly quenched powder with low oxygen content (Ref 15). Iron
powder as well as high-alloy steel powders are
produced with an oxygen content lower than 100
ppm (0.01%) by oil atomization (Ref 16).
The P/M stainless steel sector is an important
and growing part of industry because of their
oxidation and corrosion resistance and good
physical and mechanical properties. Powders are
produced by water atomization with irregular
characteristics, while gas atomization using nitrogen or argon will provide spherical shape.
Considerable progress has been made in the atomization techniques including real-time visualization (Ref 17, 18). Commercial grades include AISI 410, 440C, 446, 304L, 304, 347,

316L, 316, and 317. Tool steel powders are also
produced by either water atomization, or gas atomization with argon or nitrogen. These powders have uniform composition and fine microstructure, resulting in excellent hardness with
less distortion during heat treatment. The P/M
approach also enables production of newer tool
steels with higher alloy content but without the
segregation of alloying elements.
Rapid solidification is another possibility in
the case of atomization of molten steels. This
will reduce the micro- and macrosegregation
continuously with increasing solidification rate.
Other advantages include increased solid solu-
bility and the formation of metastable or glassy

alloys. In principle, the latter case represents
zero segregation since the liquid state is preserved to room temperature with minor changes
in the atomic rearrangement. The dendrite arm
spacing as a function of solidification rate for
metals is shown schematically in Fig. 4. Ingots
and large castings have a solidification rate of 10
to 103 K/s; subsonic gas atomization induces
solidification at about 10 to 102 K/s, water atomization at 102 to 104 K/s, and ultrasonic gas
atomization at about 103 to 105 K/s. Metallic
substrates of 20 lm have a solidification rate of
about 104 to 106 K/s, while finer metallic sub-
Fe
Fe
Fe
Fe
Fe
Fe
Fe
Fe
(a)
(b)
Fe
Fe
Fe
Fe
(c)
Fig. 3
(d)
Schematic distribution of alloying elements in different alloying methods. (a) Admixed. (b) Partially alloyed. (c)
Binder bonded. (d) Prealloyed
103
Casting
Dendrite arm spacing, m
100
Gas atomization
Water atomization
10
Ultrasonic/centrifugal
atomization
Gun splat
0.1
0.01
103
101
10
103
105
107
109
1011
Cooling rate, K/s
Fig. 4
Cooling rates and secondary dendrite arm spacing of metal powders/substrates for different atomization techniques
strate of 10 lm to submicron thickness have a

solidification rate of the order of 106 to 1010 K/
s (Ref 19). A variety of centrifugal atomizing
processes have been developed to produce powders of finer particle size. Gas dissolution in
metal melt and gas atomization have been developed to produce steel powders of less than 20
lm in size (Ref 20). The growing interest in
MIM has increased the demand for finer iron and
steel powders. Iron micropowders produced by
decomposition of carbonyl iron process have
spherical particle shape, fine particle size, and
average size in the range of 4 lm or lower with
good sinterability (Ref 21). Improvements in the
commercial atomization process has enabled
production of steel powders with mean particle
size in the range of 4 lm for MIM (Ref 22).
Extremely fine iron powders of the order of 50
have been prepared by the evaporation and
A
condensation method (Ref 23).
Although mechanical processing is generally
used for the production of powders of hard and
brittle materials, the milling process is commonly used as a secondary operation in the production of iron powders from spongy cakes of
oxide-reduced, sintered, or agglomerated, atomized, and electrolytic powders. Milling is also
used for producing powders from cast iron scrap,
cryogenic milling, or hydrogen embrittlement of
steel swarf (Ref 24, 25). Different types of hammer, disk, rod, ball, vibratory, and fluid-energy
mills are used for this purpose. For making finer
powders or solid-state alloying, high-energy
mills such as attrition, vibratory, or large-diameter tumbler mills are used. There is an increasing use of high-energy mills for producing alloy
and composite powders having unique microstructures through mechanical alloying. The
technique is used as another nonequilibrium processing such as rapid solidification for reproducing a wide variety of materials such as solid solutions,
intermetallics,
amorphous,
and
nanocrystalline materials (Ref 26). Hard steel
materials have been fabricated with matrix of
chromium carbide by mechanically alloyed powders containing 30 to 50% Cr and 2.3 to 4.8% C
with better wear resistance than Stellite (Ref 27).
Mixing and Blending of Powders. Iron and
steel powders of the required characteristics are
to be produced by blending different size fractions of the metal powders to obtain the desired
apparent density and flow of the powders for die
filling. Powders of different constituents are to
be added to produce the desired alloys during
sintering. Lubricants are to be added to assist
compaction, reduce wear, and increase the die
life. Depending on the mixing devicesuch as
rotating drum, double cone or twin shell, screw
mixer, or a blade mixerthe mechanisms of
mixing may be predominantly diffusion, convection, or shear. Double-cone and twin-shell
mixers, with or without internal intensifier
blades, are used for dry mixing of iron and steel
powders. In forming processes such as injection
molding (Ref 28), powder-binder feedstock have
to be prepared with uniform distribution of
binder and ferrous powder particles throughout

the feedstock. The binders may be thermoplastic
mixtures of polymers, waxes, oils, lubricants,
and surfactants. Thermoplastic binders are
mixed at intermediate temperatures where shear
is dominant by using screw or plunger extruders
or sigma blade mixers. The powder binder mixes
and subsequently granulates before injection
molding. For injection molding, fine steel powders in the micron-size range are used to produce
intricate shape and improved sinterability with
the binder in the range of about 40 vol%, depending on the packing characteristic of the
powders.
Compaction in Rigid Dies

Unidirectional compaction is the most widely
used manufacturing method for structural steel
components. Powder pressing is carried out in
rigid steel or carbide dies under pressures of the
order of 150 to 900 MPa. Lower pressures are
used for porous ferrous bearings, and the pressure gradually increases with increasing density
such as low-density, medium-density, and highdensity products with less than 10% porosity.
The compacts maintain shape by cold welding
and interlocking of powder grains within the
mass. The green compacts must be sufficiently
strong to withstand ejection from the die and
subsequent handling and transport to the sintering furnace. The final shape and mechanical
properties are related to the level and uniformity

of the as-pressed density known as green density. Compaction cycle for a single-level part is
shown in Fig. 5. A desired amount of mixed iron
powders is automatically gravity fed into the die
along with core rod to produce a hollow shape
from the hopper through a feeder shoe and compacted with the help of upper and lower punches
at a pressure, depending on the density requirements of the part. The upper punch is withdrawn,
the lower punch ejects the pressed compact, and
the feeder shoe slides the part away from the die
cavity. The cycle is automatically repeated by
charging, powder filling of the die. Compaction
cycle can be classified as single, double, or multiple action depending on the complexity of the
part to be produced. Movement of only the upper
punch is usually termed single action, while relative motion of two memberseither both
punches with stationary die or upper punch and
die with lower punch stationaryis termed double action. Multiple action utilizes several tool
members arranged in order to support each level
of the component separately. Depending on the
complexity, P/M components can be classified
into one of the four classes (Ref 29):
Class I: components are single level with single shape and low length-to-diameter ratio
that will allow single-action compaction.
Class II: compounds are single level with
higher length-to-diameter ratio and require
double action.
Upper punch
Powder
from
hopper
Die
Lower
punch
Core
rod
Cycle start
Charging-powder
die filling
Compaction begins
Green
compact
Compaction
completed
Fig. 5
Compaction cycle for a single-level part
Ejection
of part
Recharging die
Class III: components have two levels and are
termed external- and internal-flanged components and require double-action compaction.

Class IV: components have three or more levels.
These are multilevel parts of different thickness
and contour, and they have to be pressed with
multiple forces from two directions. A variety of
compacting presses are available with wideranging pressing capacities and production rate
capabilities. Compacting presses are mechanically or hydraulically driven, but can incorporate
a combination of mechanically, hydraulically, or
pneumatically driven systems.
The tool design should take into consideration
the nonuniform transmission of pressure to the
mass of powder in the die, limited lateral flow,
as well as the ejection of the compacted part
from the die in the direction of pressing. The
density within the green compact decreases with
increasing length of the compact since the powders do not behave like liquids under pressure
and the applied pressure is not uniformly transmitted. Double-action compaction can improve
the pressure distribution, but generates a lowerdensity region in the middle of the compact.
Therefore, length-to-diameter ratios exceeding 3
to 1 are not generally recommended. Normally
reentrant grooves, reverse tapers, and lateral
holes are difficult to incorporate in the compact
because of the difficulty of ejection and, thus,
machining. Efforts have been made to overcome
this limitation by designing flexible die assemblies. Bevels requiring feather-edged tools that
are fragile should be modified to a small flat end.
Parts with abrupt changes in section should be
avoided as they introduce stress raisers that may
lead to crack formation as a result of the stresses
induced by the elastic expansion or springback
during ejection of the part from the die. The size
of production rate of the part will depend on the
press capacity, complexity of the part, and the
number of punch motions required during compaction.
The double-action tooling system is used to
produce class I and II parts. Force is applied to
the top and bottom of a part simultaneously. In
this case, the die is stationary and the punches
have the same travel rate. The floating-die tooling system can also give a similar effect in which
the die is mounted on a spring or a yielding
mechanism using hydraulic or pneumatic cylinders for an adjustable resisting force. When the
upper punch enters the die and starts to compact
the powder, friction between the powder and die
wall causes the die to move down, which will
have the same effect as an upward movement of
the lower punch. When the compaction is complete, the die moves upward to its powder fill
position, and the lower punch ejects the part
from the die. Multiple-motion die-set process
can produce complex P/M parts using floating
die and withdrawal tooling methods. For very
high production rates of simple parts of the order
of 60,000 to more than 100,000 pieces/h, me-

chanically driven rotary presses or anvil presses
are used.
Isostatic Compaction
Some of the constraints of rigid-die compaction such as long thin-wall cylinders and parts
with undercuts, and so forth, can be overcome
by isostatic compaction. In cold isostatic pressing, water or oil is used as the pressure medium.
The powder is contained in a flexible mold such
as polyurethane and is compacted with the same
pressure in all directions through the fluid pressurizing medium (Ref 30). Pressure ranging
from 200 to 400 MPa or more of the order of
800 MPa can be used. When the elastomeric
mold along with the powder can be removed
from the pressure vessel, the technique is known
as wet bag tooling, while the technique in which
the elastomeric mold is fixed to the pressure vessel is known as dry bag tooling. Since the pressure is applied isostatically over the entire surface of the part, more uniform density is
achieved through the entire part than in the case
of rigid die pressing. Isostatic pressing is used
in the manufacture of 30 cm cylinder liners (Ref
31), P/M tool steels, stainless steels, and alloy
steels. Other compaction processes include triaxial compression by simultaneous isostatic and
uniaxial compression, high-energy-rate compaction, explosive forming, slip casting, and vibration compaction (Ref 32, 33). Compaction in
rigid dies at very high pressures of the order of
3 GPa have also been carried out using T15 highspeed steel dies and punches. These steels are
manufactured by HIP of inert-gas-atomized
powders with a composition weight percentage
of 12 W, 5 V, 5 Cr, 5 Co, and 1.6 C. The process
known as cold sintering (Ref 34) or high-pressure consolidation at ambient temperature has
been used for processing T15 high-speed steel
and 410L stainless steel with improved properties. Full-density fine iron nickel powder composites with 80 vol% VC and Cr3C2 have been
successfully consolidated by cold sintering.
Sintering
Green ferrous parts produced by compaction
are sintered at elevated temperatures in a con-
trolled atmosphere so as to produce sintered

parts with the desired microstructure and mechanical properties. Sometimes loose powders
are also sintered to consolidate them for certain
applications, such as porous filters and electrodes or for model experiments to study sintering. Sintering is a key operation for the development of sintered ferrous alloy components
intended for structural applications. It is a thermal treatment of the powder or a compact at a
temperature below the melting point of the main
constituent to increase its strength by bonding
the particles together. A variety of physical,
chemical, and metallurgical transformations occur during sintering, and these are influenced by
sintering conditions such as temperature, time,
atmosphere, and cooling rate as well by the
chemical composition of the powder mass. During sintering, atomic diffusion takes place, and
the neck formed among the powder particles as
a result of bonding will grow, leading to improvement in mechanical properties. In the case
of atomized prealloyed powder particle, the microsegregation within the powder particle, if any,
will be eliminated. The reaction between the sintering atmosphere and the surface oxides or adsorbed gases on the powder particles will clean
the surface, thereby promoting the diffusion process. When the alloying additions are in the form
of elemental powders, the diffusion process will
lead to the beginning of alloying. As the sintering proceeds, recrystallization and grain growth
may follow, the pores tend to become rounded,
and the total porosity decreases, resulting in densification. Depending on the constituents and nature of initial powder mass, the lower-meltingpoint constituents melt with the formation of
liquid phase, or the liquid phase may be eutectic
or transient, which will facilitate rearrangement
and diffusion resulting in faster pore elimination
and densification. Alloying additions such as
copper and phosphorus in iron will result in the
formation of liquid phase during sintering at a
lower temperature than those of silicon, iron, or
tool steels. As the sintering continues with increasing temperature, there is likely to be more
densification and more homogeneous alloying.
After sintering, the sintered mass is cooled to
room temperature. Depending on the composition of ferrous alloys and cooling rate, the resultant structure may be annealed, normalized,
Slow cooling
zone
Loading
green parts
Fig. 6
Preheat, lubricant
burn-off zone
Sintering zone
Schematic presentation of a continuous mesh-belt sintering furnace
or quenched; the process chosen will govern the

mechanical properties of the final product.
Sintering Furnaces and Atmospheres. Different types of continuous production furnaces
are used for sintering. These include the mesh
belt conveyer furnace, the pusher furnace, roller
hearth furnace, and walking-beam furnace. In
addition, different batch-type furnaces and
vacuum furnaces are also used. The longitudinal
section of a muffle-type continuous production
furnace is shown in Fig. 6. The first section of
the furnace is the preheat or delubrication/burnoff zone, into which, after loading, the green
parts enter and the compacting lubricant is removed from the green P/M compact. Next is the
sintering zone, in which the parts are sintered at
a specific temperature and time in a suitable atmosphere. This is followed by a short, slowcooling zone to reduce the thermal shock and
develop the microstructure and a final cooling
zone with water cooling and a protective atmosphere to a sufficiently low temperature so that
oxidation does not occur when the sintered parts
are unloaded.
In a continuous mesh belt conveyor furnace
sintering is carried out at 1120 C. The function
of the sintering atmosphere is to control the
chemical reactions between the ferrous materials
being processed and the furnace surroundings.
An atmosphere may be used as source for one or
more chemical elements as in the case of addition of carbon to iron in a carburizing furnace
atmosphere. The atmosphere may also prevent
loss of alloying elements present in the materials, or the atmosphere may be used to remove
the decomposition products of the lubricants
from the furnace, thereby preventing their deposition in the furnace interiors. In the case of iron
and steel components, the atmosphere will aid
the reduction of the oxides on the surface of the
metal particles in the compact and control the
carburization and decarburization. The oxidation
and decarburization are caused when oxygen,
water vapor, and carbon dioxide are present in
improper proportions with respect to the hydrogen and carbon monoxide content in the sintering atmosphere. Iron oxides are reduced by hydrogen, carbon, and carbon monoxide, while
carburization is caused by hydrocarbons such as
methane or by carbon monoxide.
Typical atmosphere compositions, atmosphere functions, and corresponding tempera-
Hanging
curtains
Cooling
zone
Hanging
curtains
Final cooling
zone
Unloading
sintered parts

tures for sintering steel are shown in Fig. 7. The
most frequently used sintering atmospheres are
endothermic gas, exothermic gas, dissociated
ammonia hydrogen, and vacuum. The composition of principal furnace atmosphere constituents, such as conventional and synthetic nitrogen-base atmospheres, along with the chemical
requirements, such as the ratios of various atmospheric constituents for different furnace
operationsfor example, reduction of surface
oxides, carburization, decarburization, and burning lubricant vaporsare given in Tables 1 and 2.
Now consider sintering a green compact made
out of elemental powders of iron, copper, graphite, and lubricant. The product should be free
from soot and the admixed graphite should be
uniformly diffused into iron particles as combined carbon, resulting in an optimal pearlitic
microstructure during cooling with minimum
carbon loss. After melting, copper should be uniformly distributed over all particle surfaces in
the part. Good bonding between the particles and
relatively rounded pores are required for structural integrity. At least five stages are required
to complete the sintering process, which will occur in five different zones of the furnace, outlined in Fig. 7 and summarized in Table 2.
While sintering iron graphite powder, the temperature has a profound effect on the amount of
combined carbon formed for a given sintering
time. Furthermore, the cooling rate after sintering also influences the strength and hardness.
Sintering of iron-copper involves bonding of
iron to iron, iron to copper, melting of copper,
solution and diffusion of copper in solid iron,
and solution and precipitation of iron in liquid
copper. Iron-copper-carbon sintered products
425625 C
625
1040 C
give high strength due to the combined effect of

copper causing precipitation strengthening and
increasing hardenability of steel, but they lack
toughness. Sintering of alloy steel will depend
on the type of alloying such as elemental, prealloyed, diffusion alloyed, or a hybrid mix. Mixing
elemental powders provides high compressibility, but for homogenization and uniform composition good diffusion rates must be maintained
during alloying. During sintering, partially alloyed steels containing nickel, molybdenum, and
copper will form numerous nickel-rich areas of
austenite, sometimes surrounded by high-carbon
martensite because of the slow diffusion of
nickel and faster diffusion of copper and molybdenum.
High-Temperature Sintering. High-temperature sintering is currently used to sinter at
temperatures above 1120 C, which is difficult
to handle in the conventional belt furnaces. Depending on the alloy system and processing conditions, typical temperatures for high-temperature sintering of ferrous alloys are in the range
of 1180 to 1350 C. The advantages of hightemperature sintering include:
The use of reactive elements such as man-
ganese, chromium, silicon, and so forth,

which would otherwise oxidize at conventional sintering temperatures
Possibility of extending the sintering techniques to new alloy systems and materials
structures that are not possible at lower sintering temperatures, such as full-density tool
steels, stainless steel with improved corrosion
resistance, and metal-matrix composites
Improve the distribution of alloying elements
1040
1120 C
Preheating
zone
Loading green
compacts
Atmosphere
functions
Atmosphere
composition
Fig. 7
Delubrication
zone
1120815 C
815 C
to ambient
Slow cooling
zone
Water cooling
zone
Hot zone
Convey heat
quickly and
uniformly
Reduce
surface
oxides
Copper
melting
coating, or
infiltrating
Carbon
control
Cooling
Burn and
sweep out
lubricants to
front exit
Carbon
diffusion
Bonding
Cooling
rate
control
Prevent oxidation
or controlled
light oxidation
Lightly
oxidizing
Highly
reducing
Reducing
Reducing
Slightly reducing
or neutral or
slightly oxidizing
Neutral
to carbon
Neutral to
carbon
preferred
Neutral to
carbon
preferred
Carbon
control
Schematic of typical furnace condition for sintering steel
Unloading
sintered
products
such as nickel molybdenum, copper, and so

forth through enhanced diffusion at higher
temperatures; the increase in diffusion rate
also results in better densification and spheroidization of porosity
The formation of liquid phase at higher temperatures further helps in the densification of
the product.
However to take full benefits of high-temperature sintering, the critical processing steps must
be identified and controlled (Ref 36, 37). These
include the variations in sintering parameters
such as time, temperature, and atmosphere, as
well as their influence on the sintered product.
High-temperature sintering has been used as a
means of reducing the oxygen content in powder
preform forging, although the process was not
adopted for the production of sintered steel until
1970. This is mainly due to the lack of development in high-temperature sintering furnaces,
higher cost of production for making components in batch-type high-temperature sintering
furnaces, and inadequate knowledge on hightemperature sintering of ferrous alloys. Subsequently, interest in high-temperature sintering
revived with improvements in furnace technology for processing full-density tool steels and
specific ferrous alloys with improved properties.
Stainless Steels. Corrosion resistance of powder metallurgy stainless steels is inferior to that
of the wrought alloys of the same composition,
and so there is a need to investigate and improve
the properties of these sintered products. Hightemperature sintering in appropriate atmospheres
with controlled cooling rate can produce sintered
stainless steels with higher density and mechanical properties and improved corrosion resistance. Studies of stainless steel indicate that
higher sintering temperature of 1350 C reduces
the deleterious interstitial elements and increases
sintered density, strength, and corrosion resistance. Nitrogen in the sintering atmosphere provides alloy strengthening; however, nitrogen
control is important for corrosion resistance.
Rapid cooling from the sintering temperature of
the order of 200 C/min or faster is required to
prevent the absorption of excess nitrogen with
the formation of Cr2N. The formation of chromium oxide or nitride reduces the corrosion resistance. It has also been reported that the poor
corrosion resistance of 316 stainless part with
5100 ppm nitrogen, 3000 ppm oxygen, and
0.12% C can be improved by sintering in 75H/
25N atmosphere for 30 min at 1205 C followed
by fast cooling. The oxygen content is found to
be 3000 ppm, while nitrogen content reduced to
2100 ppm and carbon to 0.07%. Excessive carbon is generally caused by residual lubricant,
while high oxygen content is due to the high
dewpoint of the atmosphere. Low dewpoint gas
atmosphere of 40 C or vacuum sintering is
required for optimal corrosion properties. Precipitates of chromium nitride or carbide will not
only reduce the amount of chromium available
to form a protective oxide film, but will also

change the chemical potential of the matrix depleted of chromium from the remaining material
to form a protective oxide film and results in a
galvanic couple, thereby accelerating intergranular corrosion. Sintering studies based on relative contribution diagrams of 316L stainless
steel filters have shown that corrosion resistance
will depend on the average pore size (Ref 38).
Filters produced by sintering at higher temperature are found to have better corrosion behavior
in NaCl, although they had the same extent of
sintering as measured by the final porosity. The
improved corrosion resistance is attributed to the
healing of fine pores, which are detrimental. Improved corrosion resistance with increased sintering temperature of some stainless steels is reported in Fig. 8 (Ref 39).
Liquid-phase formation as a result of hightemperature sintering will lead to more complete
alloying, improved densification, and even attainment of full density. Theoretical density can
be achieved in the Fe-0.5B-0.6C, wt% at 1300
C, because of the formation of increasing
Table 1
amounts of the eutectic by the diffusion of iron

(Ref 40). The shrinkage and the sintered density
is a function of boron content and is strongly
affected by the carbon content. In the iron-silicon
system, sintering is promoted by a transitory liquid phase that disappears at higher temperatures.
However, the sintering is activated by the presence of silicon, which stabilizes the ferrite phase
where diffusion is higher. Other ferrite promoters such as tungsten and molybdenum improve
the density. However, the ferrite formation does
not allow heat treatment, but hardening can be
achieved by the formation of molybdenum and
tungsten carbides.
Tool Steels. Many of the problems related to
the conventional cast and wrought high-speed
steels can be overcome by adopting P/M methods. In this approach, powders of the required
composition are produced by gas or water atomization of the molten metal. The gas-atomized
powders are spherical and are densified by canning followed by HIP or hot extrusion in the
solid state at about 1100 to 1150 C. The primary
MC and M6C carbides in this process are finer,

of the order of 1 to 2 lm. However, these billets
require extensive finish machining to produce
the end products. The water-atomized powders
are irregular and can be die pressed or cold isostatically pressed to green compacts of 70 to
75% density followed by vacuum sintering at
about 1240 to 1330 C to full density. This
method can produce near-net-shape components,
but the use of higher sintering temperatures results in coarser carbides of the order of 2 to 15
lm. Hot isostatically processed components are
more expensive, but will have superior properties and performance as dies and cutting tools
particularly for complex broaches and gear hobs.
The general application of P/M tool steels can
be classified as: high-speed cutting and machining operations, cold working, processing and
molding of engineering plastic, hot (warm)
working, die casting, and precision wear parts.
The selection of a particular P/M tool steel for
a given application depends on a number of factors such as the service life required, the oper-
Composition of principal furnace atmosphere constituents

Key ratios(a)
Atmosphere
AGA
class
Air-tonatural
gas ratio
Dewpoint
Nominal composition, vol %
Nitrogen
Hydrogen
Water
Carbon monoxide
Carbon dioxide
Lean exothermic
Rich exothermic
Endothermic
Associated methanol
Associated ammonia
Hydrogen
Nitrogen
101
102
302
601
9.0
6.0
2.5
20(a)
20(a)
5
15.5
50
60
60
68(b)
68(b)
40
60
60
80
80
84.7
69.8
38.2
25.0
100
1.2
12.2
40.4
65.6
75.0
100
2.5
2.5
0.8
1.7
0.004
0.001
0.001
1.4
10.2
19.8
32.4
10.2
4.9
0.3
0.4
Methane
Hydrogen
to water
Carbon
monoxide to
carbon dioxide
Hydrogen
to carbon
monoxide
0.4
0.5
0.5
4.9
51
39
19,000
100,000
0.1
2.1
66
81
0.9
1.2
2.0
2.0
Note: For nitrogen-base atmospheres, composition, dewpoint, and key ratios can be synthesized over a broad range, depending on which of the above generated atmospheres is blended with the high-purity nitrogen gas. (a)
Control carburization or decarburization, oxidation or reduction, or combination of these. (b) Dewpoint is 6 C (10 F) above temperature of cooling water; dewpoint may be reduced to 5 C (40 F) by refrigeration or to 45
C (50 F) by absorbent tower dehydration. Hydrogen-to-water changes accordingly. Source: Ref 35
Table 2
Furnace zones, sintering phases, and atmosphere requirements to sinter an iron, 2% copper, 1% graphite, and 1% lubricant part
Atmosphere composition (a), %
Temperature
Furnace
zone
Time,
min
Sintering
phase
1 delube
425650
8001200
1030
2 preheat
6501070
12001960
515
3 high
temperature
10701150
19602100
1030
Particle
bonding
4 slow cool
1150815
21001500
515
Carbon
restoring
5 cool down
81550
1500120
3090
Cooling
down,
optionally
oxidizing
Delubing
Elimination
of particle
surface
oxides
Sintering
subphases
Atmosphere
requirement
Hydrogen
(b)
Methane
00.1
Water
0.51.5
Flow
Circulation
Lubricant vaporizing,
vapor burning,
flushing vapors to
entrance
Oxide reducing,
graphite diffusing,
copper melting, and
coating particle
surfaces
Copper diffusing, neck
(bond) growing,
pore rounding
Carbon transferring,
homogenizing
Fast, uniform heat

transfer, slightly
oxidizing
27
High toward Highly

entrance
desirable
Highly reducing to
surface oxides, neutral
to carbon fast, uniform
heat transfer
515
0.10.3
0.010.03 High toward Desirable

entrance
Neutral to carbon
reducing
38
0.20.5
Uniform slow cooling,

slightly carburizing
27
0.31.0
Pearlite forming, part

cooling down,
preventing oxidation
or controlling light
oxidation
Slightly reducing or
neutral (optionally
oxidizing) to iron,
uniform fast cooling
02
00.1
0.010.02 Medium
toward
entrance
0.010.02 Medium
toward
entrance
0.010.02 Low to
medium
partially
toward
exit
Desirable
Highly
desirable
Highly
desirable
(a) Balance is essentially nitrogen with or without small amounts of carbon monoxide and carbon dioxide. (b) Hydrogen can be derived from hydrogen storage vessel, dissociated ammonia, endothermic gas, or dissociated
methanol. (c) If slightly oxidizing atmosphere is required, water is relatively higher or a small amount of air is introduced along with nitrogen in water-cooled section at a selected location. Source: Ref 35

ating conditions anticipated, compatibility with
coating operations, materials availability, and
cost. Some of the P/M high-speed steel compositions are reported in Table 3 (Ref 41).
Processing of P/M tool steels requires controlled high-temperature sintering for obtaining
full-density products with controlled microstructure. The temperature range available for sintering to full density is extremely narrowof the
order of 3 to 5 Cnecessitating uniform temperature distribution in the furnace (Ref 42). Actual sintering temperature will depend on the
chemical composition, but may be between 1180
and 1300 C. The linear shrinkage for full densification can be of the order of 10%. The formation and sustenance of the liquid phase play
an important role in the density and desired microstructure of the sintered tool steel (Ref 43
45). A rise in the critical sintering temperature
of the order of 5 C can lead to rapid austenite
grain growth of continuous carbide films MC or
M6C, depending on the composition of steel, between adjacent grains. While sintering at lower
temperature below the solidus, the carbide dispersed in the structure will interfere with the
grain growth and densification. When the sinter-
ing temperature is raised to solidus, a certain

amount of the carbides will dissolve, thereby reducing the interference from the carbide dispersion, leading to grain growth and densification.
The required structure with small austenite
grains and finely dispersed carbides can be produced only by getting the correct amount of liquid phase and by controlling the temperature and
time of sintering. High-speed tool steel T15 has
a wider sintering temperature range than M2
steel because of the differences in the chemical
composition and carbides predominantly M6C in
the case of M2 and MC for T15 steel. Optimization studies on the vacuum sintering of a T15
steel (1.55C, 4.41Cr, 0.14Mo, 2.24Co, 4.27V,
12.80W, 0.094O, balance Fe) with a Fisher subsieve particle size of 13.6 lm have shown their
optimal microstructure and bend strength could
be achieved by sintering at 1265 C for 30 min
(Ref 45).
Another important development is the attainment of enhanced properties in high-speed steels
through improved cleanliness (Ref 46). Knowledge about the role of oxide inclusions on the
functional properties of high-speed steels has
initiated considerable measures to control the in-
clusion in the steel by improved refining. In a

high-alloy high-speed steel such as ASP 30
(8.5Co, 6.4W, 5Mo, 4.2Cr, 3.1V, 1.28C), inclusion control is achieved by employing a totally
closed system, from atomizing molten-metal
tundish to encapsulation of the powders, in order
to prevent the dust contamination and by the oxygen inclusion refining to an active slag treatment by the electroslag heating (ESH) tundish
process. These improvements in the P/M processing of high-speed steels have resulted in increasing the mechanical properties. Substantial
improvements in fatigue properties could be
achieved by electroslag heating of tundish. In
conventional high-speed steels (M2), when surface defects from tool preparation are eliminated,
individual carbides and stringers act as stress
raisers. In the P/M high-speed steels (ASP 30),
inclusions act as stress raisers and directly affect
the microchipping resistance. Minimization of
the unfavorable inclusions through the modified
process improves the microtoughness of P/M
steels (Ref 47). A reduction in the number of
inclusions with critical size is important in many
applications. Molds for optical lenses, medical
parts, compact disks, and encapsulation of integrated circuits may not permit nonmetallic inclusions as small as 15 to 20 lm, as the presence
of these will result in the rejection of the molds.
The newer P/M tool steels have a significant
positive effect on both manufacturing properties,
such as electrodischarge machining ability, polishability, and mechanical properties. They have
also exhibited improved tool performance in applications such as cutting tools, molds requiring
high surface finish fine blanking tools, powdercompacting tools, and highly stressed cold-forming tools.
Treatment of Sintered P/M Parts
Fig. 8
Corrosion resistance of various P/M stainless steels. Hours of immersion by open beaker method in 5% NaCl
solution until 1% of surface is covered by rust or stain. Source: Ref 39
Although sintered parts can be used as such,

some of the parts are sized or repressed in order
to maintain closer tolerances or higher densities.
Certain sintered products require additional
treatments such as infiltration, oil impregnation,
machining, joining, steam treatment, plating, and
other corrosion-protection treatments and heat
treatment.
Infiltration is a process of filling the pores of
P/M parts with a metal or alloy of lower melting
point to increase the density, strength, hardness,
ductility, or impact resistance of the component.
The properties of sintered iron P/M parts have
been improved with the infiltration of copper. It
is important to control the infiltration temperature for developing proper microstructure and
hence the mechanical properties (Ref 48).
Joining P/M components can be used to produce complex shapes or to produce products
with different properties by joining P/M parts
with different materials, including conventional
components. Joining can be achieved by sintering, infiltration, or by projection welding, arc,
resistance, or electron beam welding. Sintered
P/M parts can also be brazed with special braz-

ing materials. Other joining methods include
gluing, press fitting, upsetting, riveting, and bolting.
Oil impregnation will provide the part selflubrication, as in the case of bearings, or greater
resistance to corrosion. Impregnation can be accomplished by soaking the parts in an oil bath
or by a vacuum process.
Machining is minimal in P/M components
and is limited to operations such as undercuts,
cross holes, or threads. Frequently, machining
operations such as tapping, threading, drilling,
or surface grinding are necessary for sintered P/
M parts. Since the sintered materials are porous,
it is necessary to keep the cutting tool sharp. The
factors affecting the machinability of P/M steels
are composition, porosity, microstructure, processing conditions, inclusions, and machining
parameters (Ref 49). The effect of various machinability-enhancing additives or composition
Fig. 9
Effect of density on the hardenability of P/M

steels
Table 3
changes to sintered products has dominated most

of the investigations, and these additives have
included lead, bismuth, tellurium, selenium,
CaO, MoS2, and MnS. Although porosity is necessary for certain applications such as self-lubricants, it has a detrimental effect on machining,
leading to interrupted cut and acceleration of tool
wear and poor surface finish of the product. In
this context, studies on the machinability and enhancing additives are very significant for increasing the application range of sintered products.
Sintered materials are more susceptible to atmospheric corrosion than solid materials because of the porosity that will not only increase
the reactive surface, but will also provide favorable sites for condensation of liquids that may
cause corrosion. The extent of corrosion depends
on porosity, which is governed by the sintered
density of the material. The porosity can be
closed by resin impregnation or steam treatment.
For impregnation resins such as polyester, resin
diluted with styrene is found to be suitable, and
the impregnation is carried out in an autoclave
followed by hardening the resin using heat treatment. Resin-impregnated sintered components
are also useful for pressure-tight applications.
These components can also be plated with copper, nickel, chromium, cadmium, zinc, and so
forth for improving surface corrosion or the appearance of the sintered parts.
Steam treatment is another method to close the
porosity of sintered parts and/or to increase the
corrosion resistance and mechanical properties
(Ref 50). Steam treatment is carried out in special furnaces with overheated steam at around
550 C for 1 to 2 h. This treatment will provide
a black iron oxide (Fe3O4) layer of about 5 lm
thick on the surface that is hard and wear resistant.
Heat Treatment. Sintered steel components
are often subjected to heat treatment in order to
optimize the physical and mechanical properties

of the product. These heat treatments include annealing, through hardening, and case hardening.
While heat treating P/M parts, special consideration needs to be given to the density, composition, quenching or cooling method, temperature, atmosphere, and the equipment-related
variables such as temperature uniformity and atmosphere control.
The response of a P/M part to a particular heat
treatment will depend on its thermal conductivity, which is related to the density of the sintered
part. High-density P/M parts and wrought material parts have high thermal conductivity, ensuring rapid heating and cooling, while parts
with low density take longer to heat and cool,
thereby necessitating special quenching or cooling methods. Another factor to take into consideration is hardenability, which decreases with increasing porosity. During case hardening, case
depth may increase with increasing porosity, as
shown in Fig. 9 (Ref 51). Another factor to be
considered is the surface morphologythe
amount of open porosity at the surface. The
amount of surface open porosity increases with
decreasing density. Accordingly, decisions have
to be made about the type of quenchant, the type
of equipment (such as salt bath) to be used, and
so forth for hardening. The micrograph in Fig.
10 shows the effect of surface morphology on
the ability to carburize FC-0202 (0.2C, 2Cu) materials (Ref 51).
Alloying elements have a direct influence on
the hardenability of ferrous P/M materials. The
compositions of some P/M structural steels are
given in Table 4. The alloying elements are carbon, copper, nickel, and molybdenum. As the
alloying content increases the eutectoid of the
steel generally shifts to a lower carbon content
in the phase diagram, while the critical cooling
rate is reduced and the transformation nose of
Nominal compositions of P/M high-speed steels

Designation
Composition, wt%
Tradename(s)
AISI
UNS
JIS
Werk. No.
Mo
Wear-resistant high-speed steels containing 34% V

ASP23, APM 23, CPM M3, Micromelt M3, FAX 31, DEX 20, KHA 32
CPM M4, Micromelt M4, Isomatrix S690, HAP M4
M3
M4
T11323
T11304
SKH53
SKH54
1.3344
1.3
1.4
6.25
5.75
5
5
SKH 55
1.3243
1
1.5
1.3
6
5.75
6.25
SKH10
1.3202
1.3241
1.6
1.5
1.5
1.6
2.3
2.1
2.2
12
10
8
11
6.5
14
12
1.3
1.8
6.25
12.5
Heat-resistant and superhigh-speed steels containing 512% Co and 26.5% V

CPM M35
M35
CPM Rex 54
ASP30, APM30, CPM Rex 45, Micromelt HS 30, Isomatrix S790,
FAX 38, DEX 40, HAP 40, KHA 30

CPMT15, Micromelt T15, FAX 55, DEX 61, HAPT15, KHA 50
T15
T12015
CPM Rex 76, Micromelt HS 76
M48
T11348
HAP 50, DEX 62
Isomatrix S390
ASP60, APM60, KHA 60
DEX 80
HAP 70
Cobalt-free superhigh-speed steels

CPM Rex 20
CPM Rex 25
M62
M61
T11362
T11361
Co
Weq
Hardness, HRC
3
4
16.25
15.75
6567
6567
5
5
5
2
4
3
5
5
8
16
15.75
16.25
6567
6668
6668
5.25
6
2
7
6
9
5
3
4
5
6.5
5.5
5
5
8.5
8
8
10.5
12
12
12
20.5
20
15
20.5
26
30
6668
6769
6769
6668
6769
6870
6971
27.25
25.5
6668
6769
10.5
6.5
2
5
Note: All of the P/M high-speed steels contain about 4% Cr for hardenability in large sections. Silicon, manganese, and sulfur contents are typically 0.50%, 0.30%, and 0.03% maximum, respectively. For select applications
requiring improved machinability, sulfur contents are increased to 0.10 or 0.22% with corresponding increases in the manganese contents. Source: Ref 41

the continuous-cooling curve shifts to the right.
Increasing the carbon content tends to increase
the tensile strength. Maximum strength occurs at
a carbon level between 0.8 and 0.9%. However,
above the carbon level of 0.77%, the austenite
that forms at the grain boundaries results in a
brittle structure. The addition of copper also
strengthens and helps to sinter harden some lowalloy steels. Copper has a strong influence on
dimensional change, and precise control of copper and carbon is required to maintain consistent
dimensional change. Addition of nickel to 2%
and in certain cases up to 6% improves the heat
treated properties of the steel. Nickel lowers the
austenitizing temperature and increases hardenability. Molybdenum additions in the range of
0.5 to 1.5 are useful since it has a higher diffusion rate than nickel. However, more care should
be taken in the heat treatment of molybdenumcontaining steel, and the molybdenum carbide
formed during tempering can provide secondary
hardening. Powder metallurgy parts may be oil
quenched, which is less severe than water and
brine due to improved distortion control and

minimized cracking. Fast oil with improved heat
transfer characteristics is preferable. Since the P/
M part can absorb oil, proper care should be
taken during cleaning. In addition to oil, water,
brine, or aqueous polymer solutions can be used
with improved rate of heat transfer, but accelerate the part corrosion on account of the residual
quenchant that can be trapped in the surface of
pores. For hardening and case hardening, the
temperature must be high enough to fully austenitize the material so that it will quench to a
martensitic structure. A higher austenitizing temperature is required for oil quenching than for
water or brine quenching. Tempering and steam
treating are performed below the austenitizing
temperature, a uniform distribution and dissipation of heat are needed to ensure consistency of
the end product. With increasing carbon and
nickel content, the austenitizing temperature is
decreased, while other alloying elements slightly
increase the austenitizing temperature and time.
More than double the soaking time is required
for less-dense P/M products than for wrought
material because of the lower thermal conductivity. Considering the atmosphere of endothermic
gas, a nitrogen/methanol atmosphere or a nitrogen-base atmosphere can be used as long as the
carbon potential is controlled to prevent oxidation or reduction. During case hardening, the atmosphere should be neither carburizing nor decarburizing to the materials. The amount of
additive gases such as methane, propane, and/or
ammonia should be a function of density and
degree of open porosity in the part. Sintered steel
parts with low density and with highly open surface will have less uniform case depth and less
well-defined interface between case and core.
These defects can be minimized by controlling
the relative amounts of additive gases. Important
heat treatments of P/M steel parts include hardening by austenitizing and quenching; processes
Table 4
that improve surface properties such as the case

hardening methods of carburizing, carbonitriding, nitrocarburizing, and nitriding; tempering;
and annealing.
Carburizing is carried out on P/M parts with
relatively large cross-sectional thickness to obtain maximum fatigue and impact properties.
Parts with more than 15% porosity are not recommended for carburizing. Carburizing materials generally contain nickel, molybdenum, and
copper with a combined carbon level of 0.3 to
0.35% for a porosity of 10 to 15% and the combined carbon is reduced to 0.15 to 0.25% for
porosity of less than 10%. These steels are generally gas carburized at about 900 to 930 C for
shorter cycles with a higher carbon potential than
for wrought material.
Carbonitriding is a gas-carburizing process in
which about 10% ammonia gas is introduced
into the gas-carburizing atmosphere at about 800
to 850 C. The ammonia dissociates, and the nitrogen formed diffuses into the steel surface simultaneously with carbon, thereby retarding the
critical cooling rate on quenching, leading to a
more consistent martensitic transformation. Care
should be taken to avoid excessive nitrogen diffusion, which will lead to embrittlement. Carbonitriding, because of its lower operating temperature, provides better control of distortion
than carburizing.
In plasma nitriding, an ion discharge or
plasma of positive ions and electrons is set up
between two electrodes in a vacuum chamber
containing low-pressure nitrogen gas. The ions
bombard the P/M part that forms the cathode,
heating to temperatures of the order of 470 to
570 C, cleaning them by dislodging surface atoms and depositing active nitrogen, which is created by ion collision with nitrogen molecules in
the gas.
Induction hardening is recommended for
those parts requiring a hard and wear-resistant
Composition of P/M structural steels

Composition limits (b), %
Description
P/M iron and carbon steels
Iron-copper and copper steels
Iron-nickel and nickel steels
Copper-infiltrated steels
Low-alloy steels
Influence of surface effect on the ability to carburize on FC-0202 (0.2C, 2Cu) material. Top
to bottom open surface, closed surface, core. Closed surfaces allow for more uniform absorption and diffusion of
carbon. Etchant: 2% nital
Fig. 10
MPIF (a) designation
Ni
Cu
Fe
Mo
F-0000
F-0005
F-0008
F-0200
F-0205
F-0208
FN-0200
FN-0205
FN-0208
FN-0405
FN-0408
FX-1005
FX-1008
FX-2005
FX-2008
FL-4205
FL-4405
FL-4005
FLN-4205
00.3
0.30.6
0.60.9
00.3
0.30.6
0.60.9
00.3
0.30.6
0.60.9
0.30.6
0.60.9
0.30.6
0.60.9
0.30.6
0.60.9
0.40.7
0.40.7
0.40.7
0.40.7
1.03.0
1.03.0
1.03.0
3.05.5
3.05.5
0.350.55
1.702.00
1.352.50(c)
1.53.9
1.53.9
1.53.9
02.5
02.5
02.5
02.0
02.0
8014.9
8014.9
15.025.0
15.025.0
93.898.5
93.598.2
93.297.9
92.298.0
91.998.7
91.698.4
89.996.7
89.694.4
82.591.7
82.291.4
72.484.7
72.184.4
95.90.85
96.398.9
94.597.5
93.9597.75
0.500.85
0.701.00
0.400.80
0.500.85
(a) Metal Powder Industries Federation. (b) Other elements total by difference equals 2.0% max, which may include other minor elements added for
specific purposes. (c) At least 1% of the nickel is admixed as elemental powder.

surface such as cams, bevel gears, spur gears,
and so forth. Since the inductance of P/M materials, because of porosity, is lower than in
wrought materials of similar composition, a
higher power setting is required to reach a given
depth of hardening. Further rapid transfer to the
water-quench solution containing some rust-preventive additive to avoid internal corrosion is
also required.
Nitrocarburizing is carried out within the ferritic region at about 570 to 600 C, involving the
addition of nitrogen and carbon to the surface of
P/M parts. Nitrogen forms a thin layer of hard
and wear-resistant e-iron nitride on the surface.
The atmosphere usually consists of a 50:50 mixture of endothermic gas and anhydrous ammonia, and the metal powder should have a density
of more than 90%. Nitrocarburizing improves resistance to sliding wear and fretting and also increases the strength and reduces the notch sensitivity of P/M parts.
Annealing is used in cases such as P/M forging to improve machinability. Forgings are austenitized at a temperature appropriate for carbon
content, followed by rapid cooling to an isothermal transformation temperature determined by
the TTT diagram for the steel. A combination
atmosphere and vacuum continuous heat-treating system is shown in Fig. 11 (Ref 51). Typical
ferrous P/M applications include hardening, carburizing, carbonitriding, and tempering of automotive transmission and drive-train components,
hydraulic valves, cams, and actuators.
Sinter hardening allows P/M parts to be hardened in the sintering furnace while cooling from
the hot zone. The material should be selected
with sufficient alloy content so that it is air hardenable. Parts made out of nickel, molybdenum,
and low-alloy steel powders with blended addition of 2% Cu and 0.8% C have been found to
develop satisfactory properties by sintering at
1120 C followed by tempering at 177 C. During accelerated cooling from the sintering furnace, these parts transform to martensite. Although the sinter-hardening process may not
produce complete transformation to martensite,
the resultant microstructure/property relationship provides considerable flexibility to the P/M
fabricator. The incomplete transformation provides a tensile strength of the order of 70 to 80%
of tempered martensite, and the remaining microstructure consists of porosity, bainite, and
pearlite. The economic advantage of the process
includes the elimination of a separate heat treatment step such as quenching and tempering,
thereby providing a cleaner process with better
dimensional uniformity and yield. Depending on
the variations of belt speed and cooling system,
the percentage of martensite can be altered from
28 to 84% with corresponding hardness values
of 16 to 36 HRC (Ref 52).
It is essential to heat treat P/M tool steels to
develop their properties. Powder metallurgy tool
steels use the same heat treatments as those used
for conventional tool steels, but the former respond more rapidly because of the uniform microstructure and finer carbide size. The heat
treatments include preheating, austenitizing,
quenching, and tempering; the optimal heat
treating temperatures depend on the composition, and the recommendations by the manufacturers should be followed (Ref 53). Powder metallurgy grade tool steels are segregation-free,
and hence the variations in dimensional changes
after heat treatment are smaller than those in conventional tool steels. Cooling rates have a profound influence on the properties of P/M austenitic stainless steels. With very slow cooling,
precipitation occurs preferentially at grain
boundaries, resulting in an increase in strength
and loss in ductility. The carbide precipitation is
detrimental for machinability. When the cooling
Combination atmosphere and vacuum continuous heat treating system. Typical ferrous P/M applications
include hardening, carburizing, carbonitriding, and tempering of automotive transmission and drive-train
components, hydraulic valves, cams, and actuators
Fig. 11
rates are normal precipitation at grain boundaries

is reduced, while rapid cooling such as water
quenching suppresses the precipitation of carbides and nitrides, thereby providing maximum
ductility.
Hot Pressing
The majority of sintered parts are made by
cold compaction in rigid dies followed by sintering. Since metals are softer at elevated temperatures, it is possible to achieve higher densities using hot pressing or pressure sintering. The
powder or green compact is subjected to heat and
pressure. Resistance sintering under pressure,
electrically activated pressure sintering, and
spark sintering are the variations of hot pressing.
The die material in hot pressing may be alumina,
silicon carbide, superalloys, refractory metals
and their alloys, or graphite, and the latter can
be heated by direct resistance or by external resistance or induction heating. Depending on the
die materials, temperatures as high as 2200 C
and pressures of the order of 160 MPa can be
used. The atmosphere may be reducing, inert, or
vacuum.
Investigation on the hot pressing of iron powders between 100 and 1000 C indicates that the
compaction rate is larger during -c transformation than at temperatures just below or above
the transformation. Furthermore, the compaction
rate is much larger below the Curie point than
above (Ref 54). High-carbon steel powder containing 1.89% C produced by nitrogen gas atomization has been vacuum hot pressed at 700
C to more than 99% of theoretical density at 0.9
ks at 160 MPa and for 7.2 ks at 100 MPa (Ref
55). The analysis of densification showed the superplastic flow of each powder plays an important role in the densification process.
Hot Isostatic Compaction. Some of the limitations of hot pressing such as problems in producing higher length-to-diameter ratios, complex shapes, and larger components can be
overcome by HIP. The temperature of HIP varies
from 400 to 3000 C and pressures of the order
of 20 to 300 MPa, depending on the material.
Heating elements such as graphite, tungsten, tantalum, platinum, molybdenum, kanthal, and nichrome are used, and the pressurizing medium
can be inert gas such as nitrogen, argon, helium,
or vacuum. A typical HIP unit consists of a pressure vessel, gas storage, handling system, power
supply, controls, and instrumentation (Ref 56).
Schematic representation of the section of a HIP
pressure vessel with furnace is shown in Fig. 12.
Since the equipment must be able to withstand
high isostatic pressures at elevated temperatures,
a cold pressure vessel with an internal furnace is
preferred. The powder to be consolidated is
packed in a gastight sealed preshaped container
or cold isostatically pressed. The container materials may be glass, ceramic, or a metal or alloy
and should not react with the powder materials,
but should be able to transmit the load to the
powder at elevated temperature. The heat is supplied by the furnace, which is kept inside the

pressure vessel. The powder-filled container is
hot degassed and sealed under vacuum. The required gas pressure is generated either by pumping the gas to an intermediate pressure to the
workpiece, followed by heating to obtain the final temperature and pressure or by heating the
workpiece to the required temperature while
maintaining a low pressure of the order of 7 MPa
followed by pressurization that is hot loading.
Pressure-assisted sintering is another method
in which the P/M parts density could be increased to near-theoretical density at pressures
of the order of 2 to 12 MPa. This will reduce the
equipment cost of pressure vessels and associated facilities, gas consumption, and sintering
time, thereby reducing the operating cost substantially (Ref 57). Initially, the sintering is carried out to attain closed porosity, followed by
the application of pressure at or near the sintering temperature to produce a full-density product.
Since the HIP equipment and process is expensive, alternate approaches have been developed, such as Ceracon process, rapid omnidirectional compaction, and STAMP process. In the
Ceracon (CERAmic CONsolidation) process, a
granular ceramic medium is used to transfer the
pressure, instead of a gas, on HIP (Ref 58). The
preform is prepared by cold compaction and then
heated in a controlled atmosphere to the consolidation temperature. At the same time, the ceramic granules are also heated and charged into
the die. The heated preform is inserted into the
ceramic medium so that the preform is completely surrounded by the ceramic grains. The
heated assembly is then transferred to a consol-
idation press, and an axial load is applied to the

ceramic grain by moving the ram mechanically
or hydraulically. The preform experiences axial
and lateral forces, resulting in deformation and
full densification with the correct processing
variables. The consolidated part is ejected along
with the granules from the die and separated
from the ceramic granules. Preforms of alloy
steels have been consolidated to full density by
this process, and the surface finish of the product
is influenced by the size and shape of the ceramic
grains. The process is simple, and economic advantages include inexpensive tooling when low
dimensional tolerances are adequate. Products
such as low-alloy steel ratchet wrenches have
been produced economically by this technique.
Rapid omnidirectional compaction is another
isostatic compaction method in which a powderfilled cavity is surrounded by a mass of fluid die
material, such as AISI 1020 steel or a Cu-10Ni
alloy, which is capable of plastic flow at temperatures of the order of 1000 C and pressures
of the order of 800 MPa. This consolidation
pressure is applied to the fluid die from a uniaxial ram of a compaction or forging press and
tooling. The pressure is transmitted to the powder through the fluid die in an omnidirectional
way as in the case of isostatic compaction. The
thick fluid elements permit consolidation of
complex shapes, including holes, reentrant angles, and irregular shapes. The powder is filled
in the welded die assembly, evacuated and
sealed, and then heated to the consolidation temperature, placed in a pot die and subjected to
uniaxial ram force. The ram force is transmitted
through the plastic fluid die elements and reacted
End closure
Wire-wound vessel
W
o
r
k
l
o
a
d
Temperaturecontrolled
hot zone
W
o
r
k
Furnace insulation
mantle
Furnace heater
l
o
a
d
Support and bottom
insulation
Thermocouple
feedthrough
Power feedthrough
Fig. 12
Hot isostatic pressure vessel and furnace
by the pot die inner surface, thereby applying

pressure to the powder from all directions. Highspeed tool steel milling cutters have been processed by this technique to fully dense, near-net
shape with very fine microstructure resulting in
superior performance in comparison with a
coarse-grained sintered product with some residual porosity (Ref 59).
The STAMP process is an integrated one in
which the molten metal poured from a tundish
is horizontally atomized with nitrogen or inert
gas (Ref 60). The powders, after sieving, are hermetically sealed in a sheet metal container by
welding. These are then heated to about 1100 C
and transferred to a hydraulic press and consolidated to about 95% theoretical density. The billet
weight may range from 50 to 2500 kg. These
billets are then hot worked to final shape. Lowalloy steels, stainless steels, and tool steels have
been consolidated. The process is energy efficient and economical with improved hot workability of the billet on account of its microstructural homogeneity.
Roll Compaction
Direct rolling of powders into a compacted
green strip with sufficient strength for handling
is achieved in roll compaction. In this process,
the powder from the hopper is carried toward the
roll gap under the roll nip with its apparent density, depending on the powder characteristics in
the static condition forming the incoherent powder zone. Under dynamic conditions when the
rolls are rotating, the friction between rolls and
powder and interparticle friction along with
gravity will carry the powder into the roll gap,
and it will be compacted into a green strip when
using a horizontal mill (Ref 61). The incoherent
powder zone extends over an angular distance
and on the entry side of the rolls with diameter
R and is continuous with the compacted strip on
the exit side of the rolls (Fig. 13). Powder rolling
is used initially for single-layer strip with different porosity. Tooling has been added to guide
powders into the roll compaction zone, and control blades have been designed to regulate the
flow rate of the powder into the mill. Other modifications include a system of side entrants at the
ends of the rolls to prevent the powder from
spilling out of the compaction zone and a system
for continuously dressing the surface of the rolls
to maintain a steady-state friction condition between the rolls and powder. Multilayer strip is
also being produced by simultaneously feeding
separate streams of powders with different layers
into the roll gap. By modifying the powder feeding system, a vertical mill can also be used for
roll compaction. Considering the principle of
metal powder compaction in general, it may be
noted that in the case of compaction, in rigid dies
friction inhibits the compaction process, while
in roll compaction friction assists the compaction.
In addition to direct roll compaction of steel
powders, efforts have also been made to produce
a porous billet by sintering followed by different

mechanical working operations such as rolling,
cold or hot rolling to produce sheets, or forging
and extrusion to produce seamless tubes or cold
extruded products. The conventional way of producing thin steel from rough ingot to the final
coiled strip is capital and energy intensive with
considerable losses in the materials through edge
trimming, reheating, and pickling. Direct roll
compaction of powders followed by sintering
and hot and cold rolling as shown in Fig. 13, will
reduce the cost of capital and energy with substantial improvements in the materials utilization. Another method is a slurry process in which
a slurry is made out of the steel powder, which
is deposited on a suitable substrate followed by
roll compaction and sintering followed by hot
and cold rolling (Ref 62). Evaluation of the
large-scale production of mild steel sheets and
tubes including seamless tubes and extruded
steel parts have shown that the process is very
attractive and economical. Hot extrusion of powders to full density utilizes a combination of hot
pressing and hot mechanical working (Ref 63
65). The hydrostatic forces along with the unidirectional forces cause the compact to flow
through the die. Microstructural control of the
product is achieved by the manipulation of extrusion variables such as temperature, extrusion
ratio, and so forth. The frictional forces produce
a shear component, and the energy expended in
shear presents almost half of the energy needed
for extrusion. Different methods of powder extrusion are extrusion of loose powder through a
die, extrusion of cold or hot pressed compact,
and extrusion of powder in a sealed container,
canned powder extrusion. Seamless P/M stainless steel tubes are commercially manufactured
by hot extrusion. In a typical case, gas-atomized
stainless steel powders of composition 22Cr5Ni-3Mo-0.015N-0.003C, balance Fe, are
loaded into a mild steel can to a density of about
70% and sealed. The can is cold isostatically
compacted to a density of about 85% and then
hot extruded to full density at about 1200 C using glass lubricant. Technical and economic advantages of this process include improved mi-
crostructure and mechanical properties, along

with improved yield of the tubes (Ref 66). A
typical example of the use of these stainless steel
tubes is in the lift tower of the Grande Arche at
La Defence, Paris (Ref 67). More than 150 tons
of P/M grade stainless steel tubes ranging in diameter from 60 to 244 mm were used because
of their special surface finish, allowing freedom
from maintenance.
Powder Forging
Powder forging is a hot densification process
in which an unsintered, presintered, or sintered
preform made out of powders is forged to higher
densities. It has the advantage of powder metallurgy such as dimensional accuracy and minimum materials waste, along with the high
strength of forging. Plastic deformation and volume change of the sintered powder materials is
different from those of the conventional cast materials because of the porosity in the former case.
In powder preform forging, the preform design
has a significant effect on the metal flow, structure, distribution of the stresses in the material,
and properties of the final product. The mode of
densification during powder forging will depend
on the repressing or upsetting conditions. During
repressing, the extensive flow of the materials in
the lateral direction is prevented by die-wall constraints as the outside diameter of the preform is
close to the outside diameter of the forged part,
and with the approach of full density a hydrostatic condition prevails. A spherical pore becomes ellipsoidal on collapse, with the long axis
close to that of the original pore diameter. The
limited lateral flow of the repressed part restricts
anisotropic properties, while in upsetting forging
considerable lateral flow of the materials takes
place. The stress state around a pore is a combination of normal stress and shear stress, and
the spherical pore becomes flattened and elongated in the direction of lateral flow. This can
result in some amount of anisotropy in upset
forging that is much less than that in wrought
material.
Powder
Hopper
Compaction
rolls
Compacted strip
Hot rolling
Cold rolling
Sintering
Coiler
Fig. 13
Rolling of metal powders to strip
Conventional forging requires few dies and

forging steps, while powder forging can usually
produce a fully formed component with a single
forging step. Powder forging is energy efficient
and utilizes maximum materials with minimum
scrap. The actual savings in materials will depend on the complexity of the product. Powder
forging consumes less energy and provides
longer tool life. The properties of the powderforged component depend on a number of processing parameters such as powder characteristics, compaction, sintering, or forging
parameters such as forging temperature, pressure, speed of press, chilling of the preform by
the tooling, and the amount of material flow. A
variety of iron and steel parts have been powder
forged, and their number is increasing because
of the economic advantages (Ref 68). A large
number of powder-forged components are produced for automotive applications for transmissions, engines, and differentials. These include
connecting rods for engines, valve seats, stator
cams, chain saw sprockets, gears, and so forth.
The increasing demand for light weight, low
cost, close control of balance, weight, improved
material properties, and endurance limits by the
automotive industry has resulted in the introduction of powder-forged connecting rods. Powder
metallurgy offers homogeneous microstructure
without any material textures and anisotropic
properties. Powder forging in closed dies without lateral flow results in high density and
strength levels. Premixed powders such as FeCu-C or prealloyed powders such as Fe-Cr-NiMo-Mn with additional amounts of free carbon
and heat treatment enable generation of customer-tailored properties with preferable microstructures. Strength levels more than 1500 MPa
can be achieved by powder forging, and hence
individual alloy design offers the possibility of
lightweight connecting rods. Chemical compositions of some of the ferrous alloys used for
powder forging are given in Table 5.
The production sequence of the powder-forging process is shown in Fig. 14 (Ref 70). The
prealloyed powders are compacted on an eightaxis computer-numerical-controlled (CNC) servohydraulic multiplaten powder press to a density level of 6.4 to 6.6 g/cm3. A sophisticated
tool concept allows close control of the movements of the individual compacting punches and
guarantees a homogeneous distribution of sectional densities and therefore also determines the
total and balance weight of the components at
minimum weight scatter. A subsequent balance
system then provides a 100% weight check and
guarantees close tolerances and balance weights.
The highly precise weight control system guarantees consistent properties of the component as
well as optimal conditions and tool endurance
for the following precision net-shape forging in
closed dies. The preforms are then positioned by
a robot pick-and-place system directly onto the
conveyor belt of the sintering furnace or into an
intermediate buffer-system in order to provide
flexibility for tool changes or maintenance work.
Sintering is carried out at 1120 C for 15 min,

Table 5
Chemical composition requirements for powder forged parts

Composition, wt%
Element
Ni (max)
Mo (max)
Ma
Cu
Cr (max)
S (max)
Si (max)
P (max)
C
O
Fe
P/F-10xx
P/F-10Cxx
P/F-11xx
P/F-11Cxx
P/F-42xx
P/F-46xx
P/F-44xx
P/F-49xx
0.10
0.05
0.100.25
0.30 max
0.10
0.025
0.03
0.03
(b)
(c)
bal
0.10
0.05
0.10.025
1.82.2
0.10
0.025
0.03
0.03
(b)
(c)
bal
0.10
0.05
0.300.60(a)
0.30 max
0.10
0.23(a)
0.03
0.03
(b)
(c)
bal
0.10
0.05
0.300.60(a)
1.82.2
0.10
0.23(a)
0.03
0.03
(b)
(c)
bal
0.400.50
0.550.65
0.200.35
0.15
0.10
0.03
0.03
0.03
(b)
(c)
bal
1.752.00
0.500.60
0.100.25
0.15
0.10
0.03
0.03
0.03
(b)
(c)
bal
0.10 max
0.800.95
0.080.18
0.15
0.10
0.03
0.03
0.03
(b)
(c)
bal
0.10 max
1.41.6
0.080.18
0.15
0.10
0.03
0.03
0.03
(b)
(c)
bal
(a) Covers manganese sulfide additions of from 0.3 to 0.5 wt%. The manganese content in solution is similar to P/F-10xx or P/F-10Cxx, i.e., 0.100.25 wt%. (b) Carbon content specified by the purchaser. (c) When required,
maximum oxygen content shall conform to the amount specified by the purchaser. Source: Ref 69
Automatic
handling
100%
weighing
Sintering
Compacting
Buffer
Shot-peening
Fig. 14
Automatic
handling
Protective
atmosphere
Signing
Forging
Production route of powder-forged connecting rods
Powder
filling
Ejecting
Compacting
Preform
inserting
Forged
connnecting rod
Ejecting
Forging
Fig. 15
Tooling configuration for powder forging of connecting rods
and the process parameters such as temperature,

belt speed, dewpoint, and so forth are monitored
and registered. A second robot system transfers
the sintered hot parts under a protective atmosphere of nitrogen into the forging cavity of the
CNC hydraulic forging press.
Forging is done in a manner similar to the tool
concept of compaction in fully capsulated closed
dies when any lateral material flow and the possibility of generating flashes or burrs are
avoided. This compaction step therefore only
provides a reduction of the component height so
that all geometrical features and weight accuracy
of the green part are transferred into the forged
preform. The specified densities of more than 7.6
or even 7.8 g/cm3 are achieved in the most
stressed transition zone from gudgeon pin end to
shank section. Consistent forging results are obtained through the control of uniform tool temperature and proper lubrication. The tooling configuration for powder forging of the connecting
rod is shown in Fig. 15. After ejection, the forged
connecting rods will be cooled down to room
temperature under controlled conditions in a protective atmosphere. Shot peening is performed
to remove the connecting rod forging pale. Density and sectional distribution as well as carbon
and oxygen content are analyzed at regular intervals. All dimensions are checked on a statistical process control (SPC) base and recorded.
An assessment of the cost of manufacturing connecting rods by different manufacturing techniques indicates a minimum cost for powder
forging. Besides reduced energy and fluid costs
for machining, the environmental aspect due to
drastically reduced removal of chips and machining coolant volume gains more importance
under modern ecological management audit
schemes. Powder-forged connecting rods have
reduced engine weight and improved engine performance on account of their improved weight
control and dimensional control along with reduced component weight. Furthermore, powder
forging has proved to be competitive in relation
to conventional forging, resulting in an overall
cost reduction. Millions of powder-forged connecting rods are currently in use by leading auto
manufacturers throughout the world, and their
number is steadily increasing.

Metal-Injection Molding
Metal-injection molding (MIM) has been established over the past decade as an emerging
technology for the manufacture of small, precision components of complex shapes in both
small and large volumes. Metal-injection molding components are finding new applications in
industries such as automotive, business equipment, computer hardware, chemical, biomedical,
aerospace, and armaments. Specific components
include airbag components and other automotive
parts, micro-electronic packages, computer disk
drives, orthodontic brackets, firearm components, camera and business machine parts, cutting tools, wire bonding and other assembly
tools, wristwatch cases, jewelry, thread guides,
spray nozzles, wear components, oxygen sensors, surgical tools, biomedical implants, radiation shields, cutters, clippers, connectors, turbochargers, and valves. The important steps
involved in the MIM process are shown in Fig.
16. Ferrous powders of suitable characteristics
are hot mixed with a binder and granulated to
form the feedstock, which is molded somewhat
similarly to those used in plastic-injection molding. The molded part is then subjected to debinding and sintering to form the product. The
powder particle size and shape distributions are
important for injection molding, as spherical or
near-spherical shapes will provide a high packing density while fine powders sinter more readily than coarser powders. Generally, a mean particle size of less then 20 lm is preferred. Such
powders are produced by carbonyl process or by
water or gas atomization. The applications of
Steel
powder
these ferrous powders include soft magnetic materials, low-alloy steels, stainless steels, and tool
steels. The prominent binder categories are thermoplastic, thermosetting, water-based, inorganic, and gelatin systems. The binder systems
consist of a binder, a plasticizer to reduce the
viscosity of the binder, and surfactants to lower
the surface energy for mixing. Wax polypropylene system is a typical binder system for MIM.
The ratio of metal powder to binder or the solids
loading determines the viscosity, debinding rate,
dimensional control, and shrinkage. The mixing
should result in a homogeneous powder/binder
mixture free from segregation, and the binder
constituents should fill all the interstitial spaces
between particles, at the same time forming a
thin layer around each particle. The mix is granulated into solid pellet feedstock that can be
stored or fed into the injection molding machine.
The rheological properties of the feedstock are
important; the viscosity at the molding temperature should allow the feedstock to flow smoothly
into the die cavity without segregation, and the
mix should become rigid on cooling. The feedstock is extruded into the die cavity and is heated
with careful control of the nozzle and die temperature of the injection molding machine.
Molds with multiple cavities can produce several
parts during each injection, thereby reducing the
unit cost of the parts.
The parts are subjected to debinding by heating the green compact, and this will lead to the
melting, decomposing, and/or evaporation of the
binder. This process will take several hours, depending on the section thickness of the product.
Catalytic debinding could reduce the debinding
Binder
Premixing
Granulation
Hot mixing
Granules
Debinding
Mold
Hot plunger
Molding
Sintering
Fig. 16
Metal-injection molding process
Sintered parts
time. Other debinding techniques depending on

the binder system include wicking, solvent extraction, vacuum extraction, and air drying. After
removing the binder, the brown parts are sintered
at suitable temperature, time, and atmosphere to
obtain metallurgical bonding and densification.
Since the debound brown parts are porous, large
shrinkage occurs, and the sintering atmosphere
and temperature must be closely controlled to
retain the composition and shape of the product.
The density of the sintered part is usually greater
than 97% and approaches theoretical density.
The properties of the sintered parts can be improved by postsintering. Metal-injection molding has now become a mass-production technology for small net-shape parts of about 25 g
weight, exhibiting design and economic benefits
over cast and wrought components in such fields
as electronic equipment, machine tools, watches,
cameras, and medical equipment. Recent developments in binder systems and instrumentation
for injection molding and sintering equipment
have resulted in the expansion of MIM products
to larger parts such as a 2 kg boat propeller and
newer areas such as automotive engine parts
(Ref 71, 72.)
Spray Forming
This is a potential manufacturing process for
producing semifinished and finished engineering
components economically. In this process, molten metal or alloys are atomized in an inert gas
atmosphere at high pressure to give a spray of
liquid particles that is directed onto a substrate,
where they impinge, flatten into thin disks or
splats, coalesce, and solidify to produce a coherent deposit. The process gives a short route
from a melt to a product with fine grain size,
freedom from segregation, and enhanced mechanical properties. The product can be a strip
that can be subsequently hot and cold rolled, a
shape that can be forged, or a billet that can be
extruded or deposited on a cool substrate that can
be withdrawn as a continuous tube. The process
offers a reduced number of operations and increased material yields, resulting in significant
economic benefits (Ref 73). Many metallurgical
benefits are possible through rapid solidification
such as fine grain size, no segregation, and low
oxygen content for high-alloy materials. Spray
deposition is also attractive for new alloy composition, special products such as laminates, and
metal-matrix composites.
Spray forming in the Osprey mode has become a viable technology for the production of
near-net-shape preforms. Scientific knowledge
has been utilized in process control and optimization of spray forming. The process is being
developed commercially for the cost-effective
fabrication of tubular and billet-shaped products
from stainless steels, cast iron, and tool steels.
The alloys fabricated by spray forming frequently exhibit improved mechanical properties
over ingot metallurgy products (Ref 74).
High-quality seamless stainless steel tubes,
high-speed steels, and other high-alloy materials

with superior properties have been processed by
spray forming. Efforts are also being made to
produce billets up to 400 mm in diameter and
1200 kg in weight by spray forming of D2 (1.5C,
12Cr, l.0Mo, 1.0V, wt%) cold-work tool steels.
Other tools steels such as T15 high-speed steel
(1.5C, 4.5Cr, 12.5W, 5V, 5Co, wt%) and H13
hot-work tool steel (0.4C, 1.5Si, 5Cr, 1.5Mo,
1V, wt%) are also being evaluated (Ref 75).
Spray forming appears to offer a route for the
manufacture of billet products with structures
equivalent or superior to remelted materials and
comparable to P/M products. It is also expected
that these materials will have enhanced forgeability to convert the billet to final form.
Warm Compaction
Powder metallurgy end users are demanding
densities higher than the traditional values of less
than 7.1 g/cm3 and improved mechanical properties for ferrous P/M parts. Traditionally,
higher-density cores are obtained by copper infiltration, double pressing and double sintering,
high-temperature sintering, or by powder forging. However, these techniques involve secondary processing and a significant increase in the
cost of production. Relative costs of P/M processes versus part density is shown in Fig. 17
(Ref 76). Warm compaction is a technique by
which higher densities and improved mechanical
properties are achieved by single compaction,
using heated powders and tooling in the range of
75 to 150 C, in a standard compacting press.
When the powder is heated, the compressive
yield strength of the powder is reduced substantially, depending on the temperature, resulting in
higher densities at lower compaction pressures.
Since the powder and tooling are heated to 130
to 150 C, special lubricants have to be developed to perform at these temperatures. However,
the enhanced lubricity of the lubricant at elevated temperatures results in a lower amount of
lubricants of the order 0.6% instead of the traditional 0.75% in conventional compaction, resulting in the attainment of higher green and sintered densities (Ref 77). Among the advantages
of warm compaction, mention may be made of
Fig. 17
Relative cost versus ferrous part density of several P/M processes
the attainment of higher densities with reduced

pressing force, increased green strength, more
homogeneous density distribution, better dimensional control, improved static and dynamic
properties, and good hardening and machining
characteristics. The results of warm compacted
Densmix powders heated to 130 C and compacted in heated tooling could achieve in production scales of complex parts densities in the
range of 7.2 to 7.4 g/cm3 with 10 to 15% improvement in mechanical properties (Ref 78).
The usefulness of warm compaction has been
demonstrated in the production of automotive
hubs for high-performance engines, helical
gears, gears with complex forms, spiral gears,
lock components, and so forth. The warm compaction technology offers both technical and
economic possibilities to fill the gap between
conventional single pressing and double pressing
and extends the use of P/M into new applications.
Rapid Prototyping
Rapid prototyping (RP) processes allow for
the production of prototypes directly out of
three-dimensional
computer-aided
design
(CAD) data. Complex shaped parts, including
those with undercuts or hollow inserts, can be
manufactured without the use of molds or tools
within hours. The three-dimensional description
of the object can be CAD design, computer tomograph, finite-element analysis, or from the
digitization of a master pattern. The development of stereolithography (SLA) has enabled use
of the digital information to create a part directly
from the CAD solid model. The file of the part
is sectioned by a software program into thin horizontal slices. With the introduction of appropriate process parameters, the part is manufactured
by the RP machine. Subsequent to the introduction of SLA process, other RP processes were
developed that convert powders into objects by
using machine movements determined by algorithms that systematically section a computer
rendering of a part. The commercial RP processes produce parts in layers using free-forming
material. Among the established RP processes,
mention may be made of laser sintering, selective laser sintering (SLS), direct metal laser sintering (DMLS), laser cladding three-dimensional
printing, and multiphase jet solidification (MJS)
(Ref 79). Most MIM powder materials can be
processed by MJS. Excellent properties for 316
stainless steel, where yield strengths double that
of wrought annealed bar with no sacrifice in ductility are routinely achieved by laser-based direct-fabrication processing. A chromium/molybdenum hot-work die steel, H13, was directly
deposited for three-dimensional components, as
the material is commonly used for the die-casting tooling industry (Ref 80). This process can
reduce the time for die production by about 40%.
The advantages of RP include reducing the product cost and development time for a new product.
Residual Stresses in P/M Processing

Residual stresses are of considerable interest
and importance to both basic and applied studies
because of their influence on the geometrical
conditions, processing parameters, and service
behavior of products. They contribute over a
wide range from inhomogeneous microlevel
such as the lattice imperfections to homogeneous
micro- and macrolevels. Many processes such as
mechanical, thermal, chemical, or any combination of these treatments can lead to the origin
of residual stresses. A variety of techniques such
as x-ray diffraction, ultrasonic, acoustic emission, photoelasticity, nuclear magnetic resonance (NMR) imaging, neutron diffraction, and
so forth have been used to evaluate residual
stresses. The Versailles project on Advanced
Materials and Standards (VAMAS), technical
working area, has taken up residual stress measurement using neutron diffraction. It is reported
that this is the only technique currently available
for nondestructive measurement of residual
stresses deep inside components due to the high
penetration depths of neutron beams (Ref 80).
More details about residual stresses, their origin,
measurements, and their contribution to various
properties are reported in the literature (Ref 81,
82) as well as in different sections of this handbook.
In powder processing, residual stresses can be
generated during metal or alloy powder production, blending, compaction, sintering, mechanical working operations such as powder rolling,
forging, and extrusion, machining, heat treatment, and special treatments such as coatings to
improve the performance of the product. An understanding of the contribution of residual
stresses is of considerable importance since these
can influence the dimensional stability, distortion, and properties during processing and end
uses resulting in adverse functioning of the component and its reliability during service life.
The capability of powder metal processing for
making production rates of finished and semifinished parts with good dimensional tolerances is
considered to be one of the prominent economic
advantages. The effect of traditional P/M processing at various stages on the dimensional stability is shown in Fig. 18. It is evident that stages
such as heat treatment decrease the dimensional
stability, while stages such as sizing provide improved tolerances to the product. Furthermore,
as the number of processing steps increases from
the top to the bottom of figure the mechanical
properties also increase. Although residual
stresses may originate and contribute to the dimensional stability and properties of P/M product, relatively limited published literature is
available, and there is considerable scope for research and development in this underexplored
area.
Powder Production
Residual stress can originate during the production of metal and alloy powders by different

techniques. These include thermal gradients during atomization of liquid metals, evaporation and
condensation, nucleation, and growth processes
leading to micro- and macrolevel structural inhomogeneties. The magnitude of microlevel
residual stresses in powders will increase in nonequilibrium processes such as rapid solidification or due to environmental interactions with
oxygen, moisture, and so forth. The effects of
cooling rates and secondary dendrite arm spacing for production of metal powders for different
atomization techniques are given in Fig. 4. Different techniques such as gas, water, ultrasonic,
or centrifugal atomization can increase the cooling rate from 10 to 106 K/s, thereby reducing the
dendrite spacing from several microns to less
than 0.1 lm. This will have an effect on the
macro- and microsegregation as well as on the
homogeneous micro- and inhomogeneous residual micro stresses in the powders. The level of
residual stresses in powders is also influenced by
the powder characteristics such as size, shape,
internal porosity, surface conditions and reactivity of powder. More details about powder characteristics and powder production methods have
already been discussed in the earlier sections of
this article. Now the defect porosity, which is
very much detrimental to the P/M processing, is
considered. Many factors during powder manufacturing such as alloy composition, reactivity,
melt temperature, viscosity and surface tension
of the molten metal atomizing media, pressure,
design geometry of the atomization nozzle, flow,
surrounding atmosphere, and quenching rate can
contribute to the formation of porosity of powders. Furthermore, the porosity in the powder increases with particle size. The porosity in the
powder particles will increase the residual
Dimensional stability
Decreasing
Medium
Increasing
Pressing
stresses and reduce the apparent and tap densities, thereby creating problems during compaction, uncontrolled dimensional change during
sintering, and hot consolidation, and may give
rise to residual micropores in the final product,
resulting in inferior fatigue and fracture properties of the product.
Many commercial methods of production of
ferrous powders by electrolytic and chemical
methods require some amount of size reduction
by milling. In addition, high-energy milling
operations are used to produce alloy and composite powders by nonequilibrium process such
as mechanical alloying. Mechanical alloying is
carried out in the solid state to synthesize alloy
powders of very different melting temperatures,
nanocrystalline and amorphous materials, and a
variety of metal and nonmetal composite powders. The process is carried out using elemental
or alloy powders with other constituents in highenergy mills such as large-diameter horizontal
ball mill, planetary mill, SPEX shaker mill, or
attritor. During the process, the collision of milling balls with powder particles and the container
and impeller results in the fracturing and welding
of the particles. The repetitive deformation, fracturing, and welding of the particles during mechanical alloying produce an equilibrium composite powder structure in which the oxide or
other added dispersoid and/or naturally occurring oxide films is about the same as the welded
interlamellas. The laminated structure is further
refined as fracture takes place and the thickness
of the lamella is decreased. Schematic of variation of properties with processing time during
mechanical alloying is shown in Fig. 19. As the
processing time increases, initially welding predominates in ductile materials with an increase
in the particle size, followed by size reduction
and a decrease in particle size. The lamellar
spacing decreases, while the hardness of the
powder particle increases due to strain hardening
and improved distribution of dispersoids. The
mean strain at fracture of the particle initially
decreases rapidly as the larger particles initially
will have severe flow, and eventually the smaller

particles will contain less severe flow and are
more likely to withstand the severe strain required for welding. These powders will have
considerable residual stresses due to strain hardening during the process and the heterogeneity
in the microstructure of the composite powders.
Oxide-dispersion-strengthened iron-base alloys
for high-temperature applications are processed
by mechanical alloying. A typical example is the
Inconel alloy MA 956 with composition Fe20Cr-4.5Al-0.5Ti-0.5Y2O3, processed by mechanical alloying in bar, plate, sheet, tube, and
wire form (Ref 83).
In addition to powder production, stages such
as blending and mixing to prepare powders of
desired characteristics in terms of apparent density, or addition of lubricants and/or alloying
constituents such as carbon and other elemental
powders prior to compaction can also introduce
internal stresses to the powders. A longer mixing
time in an appropriate mixer device can lead to
more power consumption, cause excessive wear
and tear of the equipment, and introduce internal
stresses in the powder through strain hardening,
and the latter will adversely affect the compaction of the powders.
Compaction of Metal Powders

Residual stresses play an important role during compaction in widely used P/M structural
steel parts production. The compaction cycle for
a single-level part is shown in Fig. 5. Application
of pressure is followed by die filling with the
upper and lower punches, and the powder particles will initially undergo rearrangement followed by elastic and plastic deformation. With
the increase of pressure, the powder packing is
improved, resulting in a decrease in porosity and
an increase in strain hardening of the powders.
The work-hardened materials undergo fragmentation, resulting in further densification and bulk
deformation of the compact. After compaction,
the upper punch is removed and the lower punch
Sintering
Welding
predominates
Heat
treatment
Resintering
Sizing
Effect of P/M processing on the dimensional

stability
Mean strain at fracture
Particle size
Lamellar spacing
Heat
treatment
Fig. 18
Hardness
Variations in properties
Heat
treatment
Grinding predominates
Sizing
Processing time
Fig. 19
Schematic representation of variation of properties with processing time during mechanical alloying

is used to eject the compact, and the cycle is
repeated. During compaction, the powder particles undergo plastic and elastic deformation, and
the component is constrained in the radial direction by the die cavity. During ejection, more and
more of the component is freed from the die cavity, thereby the radial constraint is removed,
leading to the springback to relieve the elastic
deformation. When the compact is emerging
from the die cavity, it will expand in the horizontal direction due to internal elastic strain release accompanied by the weakening of the interparticle bond. If these stresses exceed the
green strength of the compact, then the compact
will develop cracks perpendicular to the pressing
direction. The effect of strain hardening during
compaction of iron powder has shown three
characteristic regions: an initial stage of rapid
densification, followed by an almost linear variation in slope, and the third stage in which the
slope decreases rapidly. By incorporating the
strain hardening of the materials, the second
stage of compaction behavior can be accurately
predicted, while the third stage is concerned with
a mechanism incorporating the closure of microscopic defects and pores (Ref 84). In iron powder, consolidation is promoted by the presence
of porous fines that deform readily and bind the
larger particles. The irregular particle shape
causes a symmetrical loading between particles,
creating shear stresses, interparticle sliding, and
consequent frictional welding. The green
strength is developed by a combination of mechanical interlocking and frictional welding.
During compaction, the transition from first
stage to second stage occurs when the pressure
exceeds the bulk yield stress of the material,
which is when the plastic flow becomes homogeneous instead of local. Hence the model of
compaction assuming homogeneous plastic flow
can become valid at the beginning of second
stage. In the case of gas-atomized spherical powders, plastic deformation occurs at the point of
interparticle contact, and the particle shape
changes from spherical to polyhedral with the
compaction pressure below the yield stress of the
material, indicating the first stage of compaction.
The plastic deformation and strain hardening of
the powders during compaction have been confirmed by x-ray diffraction and microhardness
analysis (Ref 85). In spite of the plastic deformation, the powders exhibited inadequate compaction and green strength on account of limited
interlocking, sliding, and frictional welding because of the spherical powder characteristics
leading to little or no shearing.
In rigid-die compaction, with the movement
of punches in the vertical direction the green
density within a compact decreases with increasing length of the compact, since the transmission
of stress through the mass of powder will not be
uniform. With increasing complexity of the
product, the die design and punches with multiple levels and motions will be more involved,
and the problems related to residual stresses and
their contribution during compaction also increase and should be tackled carefully to obtain
a sound green compact (Ref 86). Another factor

to be considered is the deflection in compaction
tooling. A better understanding of the densification of the powder and the resulting stress and
deflections in the tooling will provide a means
of balancing deflection and springback, thereby
avoiding some of the compaction defects. This
is particularly important because of the tendency
to produce high-performance parts with higher
green densities where the density gradients and
springback problems become increasingly common. Although the defects in the compacts are
due to many causes and are interrelated in a complicated way, the solution to quality problems
must be addressed in a simple relationship with
leading causes. In this respect, tool stress monitoringparticularly during pressure release and
the ejectioncan provide useful information for
identifying the main causes of cracks in the
green compact (Ref 87, 88). A majority of the
cracks originate primarily during consolidation
prior to sintering, although some cracks may not
become evident until sintering has occurred.
These defects can be eliminated if most of the
causes of cracks in green P/M components are
properly understood and addressed.
Nonuniform densification of metal powders
during die compaction is a problem to be addressed for a better understanding of the process.
Lower green strength may result from incomplete compaction or the development of residual
stresses that may cause either compact failure on
unloading or difficulty in subsequent handling.
Variations in densities may also interfere with
subsequent processing through inhomogeneous
sintering, infiltration, or deformation during
forging. The rate-independent models for powder compaction require a yield function to represent the transition from elastic to inelastic behavior. The experimental results suggested that
the current yield functions characterized by an
elliptical shape in pressure-derivatic stress space
may be incorrect and that a yield function reminiscent of granular behavior may be more appropriate. Furthermore, the two state variables
density and interparticle cohesionare necessary to characterize yield (Ref 89). Simulation
of compaction processes to predict the density
distribution in compact and/or shape of the compact through constitutive equations has been attempted. The two-dimensional simulation of
compaction behavior of powder by particulate
modeling has been improved by three-dimensional simulation that incorporates plastic deformation and actually measures interparticle friction. The constitutive equations applied to
simulation of powder behavior in forming processes using continuum approach could provide
useful information on microscopic behavior of
powder assemblies such as density distribution,
stress state, and shape of the compact during or
after compaction process. However, the effect of
particle characteristics such as friction and cohesion cannot be investigated, as these constitutive equations are not related to the particle
characteristics. The development of computers
has attracted further studies on the simulation of
compaction of powders based on the particle

modeling in three-dimensions, incorporating
plastic deformation of particles and experimentally measured interparticle friction. Simulations
have been carried out for various types of compaction such as closed-die isostatic and threedimensional compaction (Ref 90). Although the
pressure/density relationship obtained agrees
qualitatively well with the experimental observation and also with the prediction of yield surfaces, there is still much to be done to develop
a realistic interparticle force low in the simulation to quantify with the experimental observations. Generally, the optimization of the tool geometry and the die sequence in net-shape powder
compaction is currently carried out incrementally and empirically. Efforts have been made to
develop a physically based micromechanical
model of powder compaction to allow for the
computer-based design procedures. The volume
density of the compact D is defined as a ratio of
solid volume to the total volume, and the deformation of powder is viewed as the material indentation of powder particles. During the first
stage of compaction when the density is less than
0.8 D, the contacts between particles are assumed to be independent, and the plastic dissipation by frictional sliding or by plastic flow is
small relative to the work done by indentation
(Ref 91). A contact law has been proposed for
the normal frictionless indentation of dissimilar
rigid viscoplastic spheres based on a self-similar
deformation and within the infinitesimal strain
regime (Ref 92). Further investigations have included the effects of elasticity and finite deformation. During the second stage of compaction
when density is greater than 0.8 D, the compact
is viewed as a porous body. In a recent study on
the micromechanics model of powder compaction, finite-element calculations have been used,
taking into consideration the effects of die-wall
friction and the effect of material and friction
parameters on the final density distribution of a
syncor-hub and a valve guide (Ref 93). For a
highly constrained geometry of the valve guide,
the relative density quickly reaches the value D
1 in the vicinity of sharp corners as a result
of shearing. A die wall friction of 0.3 leads to
the competition between sliding and shear yielding near the wall. The top portion of the material
is fully densified, and when the frictional resistance of the material becomes larger than the
shear yield stress of the compacted material, the
latter shear within in a thin layer and stick to the
wall surface, leaving debris on the die wall. Because of the less-confined geometry of the syncor-hub, a high friction of 0.3 does not cause this
phenomenon. In addition, the variations in the
relative density of the syncor-hub are small in
the top wide portion and is of the order of 0.84
to 0.86 of D, while the variation is larger in the
thin outer cylinder. The final density distribution
is extremely sensitive to the choice of friction
coefficient, is moderately sensitive to the initial
density D start and is insensitive to the reference
density Do and the elastic modulus.

Although computer simulation of die compaction has been investigated in the past, the commercial possibility has been explored seriously
only recently. Finite-element analysis can be applied successfully to simulate powder die compaction. However, the simulation of the compaction stage must account for the complex
mechanical behavior of the powder, as well as
frictional interaction between the powder and
tool set, including complex tooling motion in the
case of multilevel part geometry. In the relaxation stage, the springback of the part is small,
but the simulation must also account for the
complex material behavior and friction, since
this can lead to further densification of the powders or green part failure as a result of cracking
or end capping. During this stage it is also necessary to take into account whether the part is
within the die, since as it emerges from the tool
set radial straining takes place. All of the above
stages are interdependent, and hence it is essential that a fully integrated model is developed to
represent the complete process, which will enable the industry to make products with consistent and improved properties.
Sintering of Metal Powders

Some aspects of sintering including furnaces,
atmospheres, and systems such as stainless steels
and tool steels, and other related high-temperature consolidation techniques have already been
reported. This section emphasizes the material
transport mechanisms, dimensional changes, the
role of stresses, and process modeling.
The materials transport during sintering of
powders is brought about by a combination of
processes of surface, grain boundary, and volume diffusions. Since the diffusion rates are exponentially rated to temperatures, high-temperature sintering is being used for high-alloy steels,
stainless steels, and tool steels to produce highperformance parts. Although changes in pore
shape can occur very much faster by surface diffusion than by volume diffusion, the former cannot contribute much to the changes in pore volume and to the overall densification of the
compact. Densification proceeds by a combination of volume and grain-boundary diffusion,
and significant reduction in pore size occurs
when the pores are connected to grain boundaries, suggesting that grain boundaries act as
sinks for vacancies. Thus, for high rate of sintering the initial grain size can be reduced by
reducing the particle size, but the chances of contamination and bulk flow characteristics of loose
fine powders may restrict their utilization. During sintering, when the pores reach a limiting
size they are unable to contribute to the grainboundary migration resulting in grain growth,
and a reduction in densification will take place.
In multicomponent alloy systems, the reactions
to sintering depend on the relative surface energies, diffusion rates, and solubility of the constituents. When the constituents are extensively
soluble in one another, Kirkendall effects become significant and give rise to vacancy fluxes
in response to concentration gradient. Eliminating the problem of shrinkage during sintering of

metallic powders in the solid state will be possible with a better understanding of creep phenomenon, viscous plastic flow, internal friction,
and diffusion.
In the case of liquid-phase sintering, three
stagessuch as the rearrangement of the particles of the residual solid phase by viscous flow
in the liquid phase, dissolution and reprecipitation processes, and coalescencecan be distinguished. A typical ferrous system of commercial
importance in which liquid-phase sintering can
lead to interesting dimensional changes is the
iron-copper system. During the sintering of iron
powder, shrinkage takes place at about 1150 C,
but the addition of about 2% Cu powder to iron
powder will eliminate the shrinkage. At the sintering temperature, the copper melts, enveloping
the iron particles by the rearrangement process,
which is accompanied by solution of iron from
convex parts of the surface and reprecipitation
on the concave neck surfaces. The other process
is the solution of copper into the c-iron, causing
the expansion of c to leave small pores at the
sites of copper and expansion of the compact.
Increasing the copper content from 2% to 8%
the limit of solid solubility of copper in c-iron
at about 1150 C will allow maximum expansion
of the compact to take place. This expansion of
the compact can be controlled by the addition of
1% graphite, which will result in the formation
of ternary Fe-Cu-C eutectic liquid phase at a
temperature lower than the sintering temperature, leading to shrinkage and thereby counteracting the expansion of the compact. In Fe-10Cu,
it has been reported that swelling of the compact
can take place by melt penetration. The additional melt in prior contact areas and at prior
grain boundaries in the interior of the iron particles caused dilation leading to microscopic
swelling (Ref 94). However, the melt penetration
along grain boundaries can be suppressed if carbon is present in the alloy.
During sintering of a green component, annealing, stress relieving, recovery, chemical reactions to form new phases or compounds, composition homogenization, shrinkage, and grain
growth may take place. On coolingdepending
on the systemchanges such as precipitation
hardening, or other phase transformation, chemical reaction, or the introduction of residual
stresses in the product may be expected. In the
initial stages of sintering atomic diffusion takes
place, the particle contacts formed during compaction grow, residual stresses are relieved, recrystallization and grain growth may follow, the
pores tend to become rounded, and the total porosity decreases, resulting in densification. Considerable emphasis is laid on sintering to obtain
maximum density, resulting in shrinkage, but
one of the advantages of P/M lies in the production of finished components requiring little
or no machining. There are many factors influencing the dimensional change of a component
during sintering. The decrease in dimensions as
a result of densification is due to the elimination
of pores. However, this shrinkage can be influenced by other factors such as entrapped gases,
which may expand resulting in an increase of
volume. Other factors influencing the expansion
include hydrogen reduction of surface oxides,
and oxidation as a result of poor atmosphere control. Shape changes can also occur due to variations in green densities of a component.
Finer powder generally provides higher
shrinkage than coarser material during sintering,
but the apparent powder density and green density of the former are lower than the latter powders. However, the atomic process in the fine
powders takes place more rapidly during sintering because of larger surfaces, grain boundaries,
and lattice distortions during compaction. The
initial stages of sintering at constant temperature
for a compacted nanometric iron powder of
mean particle size of 30 nm, giving a self-diffusion coefficient of D 6.1 1024 m2 /s at
325 C, provided an activation energy of 59 kJ/
mol, which is four times smaller than the one
traditionally used for sintering microscopic powders, that is, 239 kJ/mole (Ref 95). Thus, even
if surface diffusion is important and shrinkage
experiments cannot take this process into account, mass flux between grain boundary and
neck surface is an important process in the initial-stage sintering of ultrafine powders.
For a better understanding of the process, sintering may be subdivided into three stages. The
first stage consists of solid-state sintering when
interparticle necks are formed by surface and
grain-boundary diffusion. In the second stage,
the particles are still connected by isolated contacts with open pore space, and the diffusion
field of the necks now overlap. With further densification, the three dimensionally interlinked
pore channels close at the narrow sections with
a transition to the third stage, which is characterized by closed and isolated pores. The densification in the third stage is accompanied by
grain coarsening.
During compaction, the powder particles are
subjected to strain hardening and the extent of
strain hardening increases with increasing compaction pressure. In the initial stages of sintering
at lower temperatures, the particles will relieve
the strain hardening through recrystallization as
a result of annealing. However, the high shrinkage rates of real compacts in the initial stage of
solid-state sintering cannot be explained with
diffusion as the elementary materials transport
mechanism. This stage is characterized by the
existence and formation of high densities of various types of defects as well as high rates of densification. The quantitative defect analysis made
on the basis of positron lifetime measurements
demonstrates the existence of vacancy clusters
and dislocations as well as their time-related
density change and interaction (Ref 96). There
is a drastic decrease of the mean defect density
and considerable shrinkage as well as a high rate
of shrinkage. The excess vacancies originating
in the course of restructuring the mechanical
contact into a high-angle boundary agglomerate
into clusters in the contact region. They interact

with dislocations introduced by pressing and
newly formed dislocations in such a way that the
viscosity of contact substance is lowered, and a
cooperative materials transport mechanism will
take place. On the basis of dislocation flow, this
process should be assumed to allow the entire
particle to move along with the softened contact
boundary region, or if this mechanism is exhausted then the contact substance flows into the
pores in a dislocation viscous state. Studies on
the solid-state sintering assume the transport
mechanisms of grain boundary and surface diffusion have developed constitutive equations using finite-element simulations of the sintering
process for the three stages. Although some of
the predictions of the models are confirmed by
experiments, particularly the dependence of the
bulk viscosity on the density, the sintering stress
and the ratio of the shear to bulk viscosity could
not be predicted (Ref 97).
Modeling of sintering phenomena has been
developed from finite difference calculations of
geometric details of necks between two particle
models to finite-element simulation of shrinkage
of compacted powders. Earlier work on simulations could explain some of the deviations between idealized and more realistic neck geometry and other deficiencies of analytical solutions
for describing neck growth and shrinkage. Later
came the development of sintering diagrams in
which dominating mechanisms for neck growth
and shrinkage were outlined as a function of
temperature and neck radius or time (Ref 98).
However, the predictions from these simpler geometries were found to be deviating when quantitatively applied to real experimental results
where particles of different sizes, irregular
shapes, and of heterogeneous packing characteristics. Furthermore, the data collected from simple and independent experiments will differ from
those of the powder particles with considerable
lattice defects and surface contamination. Several groups are working on the development of
advanced numerical models for simulating solidand liquid-phase sintering (Ref 99). However, it
is necessary to check the physical soundness of
analytical and numerical approaches before their
application to industrial practice. These studies
have already led to the prediction of distortion
of sintering parts with complex shapes on the
basis of appropriately fitted parameters in the
constitutive equations.
The demand for better-quality P/M components with minimum dimensional scatter is increasing, and secondary operations such as sizing will increase the production cost.
Furthermore, it is difficult to carry out the sizing
operation on high-strength P/M components because of their high hardness. The requirement of
lowest cost coupled with high strength makes it
necessary to improve the dimensional tolerance
for P/M parts after sintering. The dimensional
scatter on P/M component can be estimated by
using a mathematical model that includes the
variations in mix composition and important
compacting and sintering variables (Ref 100).
By using the model, it is now possible to com-
pare different powders, alloys, and manufacturing processes to find the most efficient alternative to achieve the best dimensional precision of
P/M parts.
Pressure Sintering and Hot Isostatic

Pressing
Hot pressing, hot isostatic pressing (HIP), and
other hot consolidation processes are potential
methods for the manufacture of high-quality,
high-strength steel, stainless steel, and tool steel
components. During hot pressing or pressure sintering of powder compacts, the stress gradients
and the stress level present are much higher than
during pressureless sintering. The external pressure applied may be in the range of a few to
several hundred MPa and will alter considerably
the stress state caused by curved free surfaces of
coarse powders in the 100 lm range. These altered high stresses yield accelerated material
transport by major mechanisms, which are active
during pressureless sintering and can provide additional transport mechanisms as plastic deformation and power-law creep. A volume reduction of 30 to 40% may make the part susceptible
to nonuniform shrinkage and nonhomogenous
microstructural development and also can cause
considerable differences in local deformation. In
HIP the powder is packed in a thin shell container in the form of a preform, the gas pressure
is applied, and the temperature is raised. During
this operation, the densification is not uniform
because heat diffuses into the powder, the hotter
surface layer sinters faster than the interior, giving a denser skin. Heat is conducted through this
dense layer faster than through the less dense
interior, thereby further adding to the temperature difference between the surface and interior.
This will result in a densification front propagating inside, leading to large changes in the
shape. A simple one-dimensional model for coupled heat flow to a hydrostatic pressure gas has
been successfully developed (Ref 101). However, modeling of the densification of three-dimensional bodies during HIP when rapid densification of one part of the body may constrain
the remaining part in a way that inhibits densification, leading to unexpected changes of shape,
is yet to be worked out. The mechanisms that
contribute to the different stages of densification
have been identified and studied. However, it is
necessary to check and refine the calculations.
Furthermore, the materials properties are so variable that a refining process in which equations
are to be calibrated to the results of real industrial
HIP process.
The HIP process is expensive, and it is very
difficult to predict the shape and density of the
final product. Finite-element calculation of the
thermodynamic behavior of the powder has
proved very useful to optimize the process and
may result in important savings on energy and
materials. Theoretical and experimental work
has been carried out using appropriate constitutive equations for numerical simulation. A uni-
fied formalism has been proposed, and an

adapted numerical tool developed, integrating
different models that can be compared among
each other and tested against industrial practice
(Ref 102). These studies indicate that the constitutive equations on numerical simulation
should be more precisely analyzed, and more experimental and theoretical investigations are
necessary to improve the mechanical behavior of
metal powders at high temperatures.
Heat Treatment of P/M Steel Parts

Special consideration has to be given during
the heat treatment of sintered steel components
because of the presence of porosity. The hardenability of wrought steel is primarily influenced
by the chemical composition and grain size,
while in P/M steel the hardenability is significantly influenced by the interconnected porosity,
degree of sintering, graphite segregation, and so
forth. When a P/M steel part is quenched from
the austenitic range, the external surface may experience higher cooling rates than the interior,
resulting in mixed microstructures and inconsistent hardness values. These variations are also
related to the quantity and type of porosity
size, shape, open, or interconnectedand the
complexity of the P/M part. When the porosity
is eliminated by forging of P/M parts, the heat
treating practice will be the same as for those of
conventionally processed steel of similar composition.
The heat treatment of full-density P/M products including tool steels and problems related
to dimensional change and distortion is considered next. The quenching may experience large
temperature differences between surface and interior and between light and heavy sections,
which can create shape distortion because of the
thermal and mechanical stresses produced by
martensitic transformation. This problem is aggravated if the hardenability of steel is so slow
that a fast cooling rate is required to get the full
hardness. For controlling the distortion of the
part, special quenching procedures such as martempering and austempering may be useful. If
permissible, localized methods of heating and
quenching such as flame hardening, induction
hardening, electron beam, or laser hardening will
be helpful to reduce the distortion.
The presence of retained austenite in heat
treated tool steel can slowly transform and produce distortion if the materials are subsequently
heated or subjected to stresses. Tool steel products such as gages and blocks must retain their
exact size and shape over long periods. If the
composition of tool steel provides the required
hardness after tempering at relatively high temperature, then it is possible to reduce the amount
of retained austenite and the internal stresses by
multiple tempering. The first tempering reduces
the internal stresses and conditions the retained
austenite so that it can transform to martensite
on cooling from the tempering temperature. A
second or third tempering is carried out to reduce
the internal stresses set up by the transformation

brazing, welding, and so forth prior to heat treatment. These are removed by a stress-relieving
treatment in which the material is heated, and the
steel will yield plastically when the hot yield
strength is equal to the residual-stress level.
The P/M processing of steels, stainless steels,
and tool steel is discussed in other sections of
this article. Powder metallurgy steels, are produced from water- or gas-atomized alloy steel
powders by compaction, sintering, high-temperature sintering, extrusion, forging, or HIP.
Powder metallurgy tool steels have significantly
improved toughness and grindability over conventionally processed tool steels. Powder metallurgy tool steels are free from macrosegregation and porosity and are characterized by
uniform distribution of extremely fine carbides.
Microstructures of P/M T15 and conventional
T15 high-speed tool steels are shown in Fig. 20
(Ref 103). The conventional tool steel (b) microstructure consists of coarse carbides and carbide
segregation while the P/M microstructure (a) is
characterized by finer carbides with uniform distribution. These characteristics are favorable for
providing deeper hardening, less distortion,
higher hardness, and faster response to heat treatment and are particularly useful for molybdenum-containing high-speed steels, which tend to
decarburize rapidly at the austenitizing temperature. The out-of-roundness of heat treated conventional AISI M2 and P/M ASP 30 tool steels
are shown in Fig. 21 (Ref 103). It is evident from
the figure that the out-of-roundness distortion of
P/M tool steel is substantially lower than that of
the conventional tool steel. Powder metallurgy
also enables production of grades that are difficult to produce by the conventional method and
also newer grades with more alloying additions
to provide improved properties.
Surface hardening by carburizing is used to
enhance the fatigue strength and tribological
properties of P/M steel gears. The surface-hardened layers are characterized by microstructural
and microhardness evaluation along with residual-stress measurement by x-ray diffraction. The
results of the analysis indicate moderate com-
pressive surface residual stresses on the green

compacted gear. A slightly higher residual-stress
value obtained after sintering is attributed to the
limited oxidation registered at the end of sintering. A more pronounced increase in residual
stress is evident after carburizing the gear. It is
concluded that x-ray diffraction is a useful tool
for strict control at different production stages of
P/M technology (Ref 104). Special attention is
to be paid during heat treatment to minimize the
percentage of residual austenite and the component geometry, as the latter can influence thermal
shrinkage during quenching of the gear. The residual austenite and thermal shrinkage can influence the surface residual stresses to a great extent.
Although P/M technology has been used to
produce net-shape products of various sizes and
shapes, frequently it is necessary to use secondary machining operations to obtain the final
product as discussed earlier. Some of the P/M
products are also resin-impregnated in order to
seal the porosity for specific applications. These
resin impregnated products have improved
machinability, but not much information is
available regarding the level of residual stresses.
Investigations have been carried out by conducting drilling experiments to compare the residual stresses in P/M products with and without
resin impregnation (Ref 105). Residual stresses
were measured by using an advanced x-ray stress
analyzer based on solid-state linear image sensor
technique that enables conversion of x-rays directly into electrical signals. The basic system
consists of a goniometer and is capable of measuring residual stresses in metallic materials
without conducting 2h scanning of the detector.
The x-ray tube has a maximum output of 2 kW
and can measure residual stresses in microareas
by means of a small focal spot size capability.
An integrated PC-based control system is used
for measurement, data collection, analysis, and
display of results. The analysis of the results indicates that the magnitude of residual compressive stresses increases with resin impregnation,
and the machining of resin-impregnated parts
produces even greater residual compressive
stresses compared to nonimpregnated machined
parts. These findings are of significant value in
understanding machinability and mechanical
properties of sintered steels.
Microstructures of high-speed tool steel. (a) CPM T15. (b) Conventional T15. Carbide segregation and its
detrimental effects are eliminated with the CPM process, regardless of the size of the products. Source: Crucible
Material Corporation
Out-of-roundness measurements on test disks

after hardening. (a) Conventional, AISI M2. (b)
P/M high-speed tool steel ASP 30
of retained austenite. In plain carbon and lowalloy steel that must be tempered at low temperature to achieve the required hardness, single
or repeated low temperature treatment below
martensite finish (Mf) temperature can cause
most of the retained austenite to transform to
martensite. The low-temperature treatment is
carried out in a commercial refrigeration unit at
about 70 to 90 C, either before or after the
first temper. If the tool materials tend to crack
because of the additional stress induced by dimensional expansion during low-temperature
treatment, it is preferable to carry out the treatment after the first tempering of the tool. When
the low-temperature treatment is applied after
the first temper, the amount of retained austenite
that transforms the treatment may be considerably less than that desired because some of the
austenite might have been stabilized by tempering prior to low-temperature treatment, and in
such cases the tools must be retempered after
return to room temperature following low-temperature treatment so as to reduce the internal
stresses and to increase the toughness of the
newly formed martensite. It is desirable in some
tools to retain a small percentage of retained austenite for improving toughness and providing a
favorable internal stress pattern that will help the
tool withstand the service stresses, and in such
cases it is necessary to avoid the full stabilizing
treatment.
Distortion of the tool steel can also occur due
to a higher austenitizing temperature or due to
surface chemical reactions. Higher-temperature
austenitizing of tool steel will lead to grain
growth and more stabilized austenite. These conditions, along with larger thermal gradients during heating and quenching, will result in irregular dimensional changes of the products.
Surface reactions such as carburization and decarburization can change the surface temperature
and subsurface transformation, causing compressive or tensile stresses leading to cracking or
distortion. Residual stresses can also originate
from mechanical or thermal treatments such as
forming, grinding, and machining, or during
Fig. 20
Fig. 21

Other P/M Processes
Dimensional changes, defects, and residual
stresses are also important in P/M processes
such as metal infiltration, spray forming, warm
compaction, MIM, and rapid prototyping. However, very limited investigations have been carried out in the area of residual-stress evaluation.
Metal infiltration of ferrous materials is carried
out using copper-base alloys. The infiltrant is
placed in contact with the green ferrous compact
and sintered at a temperature above the melting
point of the infiltrant. The sintering of iron particles is promoted by the infiltrant drawn into the
ferrous compact by capillary action. Infiltration
can also be carried out as a second step after the
sintering of the green ferrous compact. The porosity can be partly or completely eliminated by
infiltration, resulting in improved density and
mechanical properties of the product. However,
as discussed in the sintering of the iron-copper
system, care should be taken to avoid the growth
as well as the possibility of surface erosion of
the compact. Adequate measures should be taken
to avoid the contribution of residual stresses as
a result of liquid-solid transformation of the infiltrant on cooling.
Metal injection molding consists of stages
such as mixing, molding, debinding, and sintering. The densities of the MIM green compact can
be as low as 50% in comparison with about 80%
green density of the traditional P/M part. One
way of reducing the distortion is to provide a
higher green density by using a higher powder
volume in the feedstock with a wider particle
size distribution in order to obtain a high packing
density. To obtain the desired dimension of the
final product, the mold dimension is kept oversized, and the solids loading and final densities
will govern the extent of oversize. The isotropic
linear shrinkage (LS) for the molded dimension
to the sintered part dimension (Ref 106), can be
given by:
LS 1 [Di(1 WL)/Df]1/3
where Di is the density of green molded part, Df

is the density of sintered part, and WL is the total
weight loss from molded to the sintered part expressed as a fraction.
The mold shrinkage factor is (green dimension/sintered dimension) 1/[1 LS]. However, for complex components the mold shrinkages are not isotropic. The anisotropic shrinkage
arises due to variations in powder orientation,
binder orientation, nonhomogeneous powder/
binder separation, gravitational force, thermal
gradients during sintering, and so forth. In such
cases, the mold shrinkage factors for each particular dimension are to be evaluated by initial
trials. The debinding process should be selected
carefully, and the debinding cycle has to be optimized so that the product is free from macroscopic defects such as blisters, cracks, and so
forth, as well as from microscopic defects arising
out of powder segregation. Computer simulation
has also been carried out for the sintering den-
sification of injection-molded M2 tool steel, taking into consideration sintering densification,

distortion, and microstructure coarsening (Ref
107). During heating, solid-state sintering occurs
initially, followed by liquid-film formation and
spread on the grain boundaries, resulting in rapid
grain growth. A time-temperature sintering window is computed for full density without extensive carbide coarsening or shape distortion. The
simulation compares favorably with distortion
and densification findings of this steel, and is
characterized by a narrow sintering window.
The problems related to the dimensional
changes and stress levels in warm compaction
are of a different magnitude in relation to cold
compaction. The expansion of the green compact
for equivalent density is lower for warm compaction, because lower compaction pressures are
required to achieve the same density. However,
as the green density increases, since the final
green densities of the warm compacts are higher
than cold compacts, with the increased density
the tooling load will increase, resulting in greater
expansion of the product. Again, with the increase in tooling loads it is necessary to use
thicker stress rings for greater tool deflection.
Adequate care should be taken for the increased
expansion as the same can cause microlaminations particularly in multilevel parts. Microlamination can lead to reduced structural integrity of
the sintered product. A major advantage of warm
compaction is not only the attainment of increased density, but also the increased uniformity of density, which will result in increased
load-carrying capacity of the component.
The management of residual stresses and resultant distortion is an important factor in the
rapid prototyping of three-dimensional components. The chromium-molybdenum hot-work
H13 die steel is widely used by the tool and die
industry, but at the same time it is a difficult
alloy for direct-metal deposition because of the
residual-stress accumulation from martensitic
transformation. Direct-metal deposition of H13
tool steel was carried out with a laser power of
1000 W and powder feed rate of 5 g/min for fine
processing. The thickness of each deposited
layer was 250 lm, and the pattern was repeated
to create a three-dimensional component (Ref
108). Since residual-stress accumulation is the
biggest cause of cracking, an estimation of residual stress per layer was first established. After
the deposition of a predetermined number of layers, stress relieving was carried out before further layers were deposited. This strategy is useful
in the fabrication of a full-size component. Thus,
understanding and estimation of residual stress
play an important role in the successful fabrication of three-dimensional components by direct metal-deposition of alloys such as H13 tool
steel.
Microstructural Development and

Properties
Microstructural development of elastically
strained crystalline system has been of great in-
terest, as in the case of the coarsening of coherent

precipitates. The equilibrium morphology of a
coherent precipitate is dictated by the interfacial
free energy and the elastic strain energy. Computational technique has been used to analyze the
elastic state associated with arbitrarily shaped
precipitates whose elastic constants are different
from those of the matrix phase. Earlier approaches failed to describe the coherency strain
accurately. Recently, a new technique termed the
discrete atom method (DAM) has been developed to examine the general coherency problem,
based on chemical statistical mechanics and linear elasticity (Ref 109). The DAM appears to be
an answer for the study of predicting microstructural evolution in many stressed alloy systems.
The phase decomposition behavior from d to
(d c) phases at elevated temperature has been
investigated in the two-phase stainless steel
powder of Fe-24Cr-8Ni alloy, associated with
the amount of strain stored through mechanical
alloying. Mechanical alloying was carried out in
a vibrating ball mill, and with increasing milling
time the strain stored within the d powder increased and the hardness increased from 250 to
600 HV for 360 ks milling, without any change
in the crystal structure of the powder. An allotropic transformation from d to c can be induced
by heating the powder to high temperature, owing to the increased driving force in c nucleation,
prior to the equilibrium-phase decomposition to
(d c) phases. The strain energy stored through
mechanical milling can influence the allotropic
d-c transformation, resulting in grain refinement
(Ref 110). At temperatures higher than 1300 K,
the microduplex structure of (d c) markedly
coarsened, but addition of the 2 vol% of alumina
powder was found to be very effective in suppressing grain growth.
The influence of residual stresses on recrystallization has been a controversial issue (Ref
111) and recently this has been investigated in
oxide-dispersion-strengthened mechanically alloyed sheet steel MA 956 containing 20.8Cr,
5.0Al, 0.5Y2O3, and 0.5Ti. Mechanically alloyed
powders were hot isostatically pressed, and the
billets were cold rolled to produce 0.5 mm thick
sheet. These were annealed at temperatures between 1300 and 1380 C, and the recrystallization was established by light and transmission
electron microscopic investigations (Ref 112).
Recrystallization initiated rapidly at the sheet
center line, designated as stage I, resulted in
plate-shaped grains oriented parallel to the rolling direction at the sheet center line. Stage II is
characterized by slower planar growth through
the sheet thickness and finally a very coarse
grain structure, sometimes with a single grain
through the sheet thickness. The recrystallization
kinetics are characterized by incubation time
with an activation energy of 506 kJ/mole followed by a decreasing rate of boundary migration with further increasing time and temperature. The microstructural evolution is further
discussed in terms of residual stresses, deformation texture, dislocation structure, and oxide

dispersion on the recrystallization process. The
center-line-initiated recrystallization might be
attributed to the residual compressive stresses at
the center line of the sheet, although the compressive stresses are maximum at the surface of
the sheet as determined by hardness measurements. The residual-stress profile from the sheet
surface to the center line indicates that compressive stress at 725 MPa was measured at the
surface, while the same becomes negligible at a
depth of 0.02 mm. Two opposing regions of tensile stress, which balance the compressed stress,
were broader, ranging from 0.02 to 0.19 mm
with a peak tensile stress of 80 MPa. The remaining region around the center line of the
sheet is slightly compressive at 20 to 30
MPa. The high surface compressive stresses are
consistent with the highly deformed near-surface
areas of the sheet as indicated by the high-density shear bands. This unique recrystallization
process in MA 956 sheet appears to be influenced by the strong deformation texture gradient
and powerful high- and low-angle boundary pinning by the fine oxide dispersions.
In the development of new types of high-temperature materials for the propulsion system and
so forth, there is a need to withstand severe heating and oxidation along with a steep temperature
gradient. Appropriate combination of functionally gradient materials is a possible solution to
withstand such severe conditions. Functionally
gradient materials from 304 stainless steel powders and stabilized zirconia (ZrO2-3mol%Y2O3)
have been prepared by hot pressing. These materials were evaluated by burner heating using
H2 /O2 combustion flames. The heat insulating
properties of these materials were satisfactorily
evaluated by the thermal resistance value, but the
sample surface developed cracks during the
cooling cycle (Ref 113). Thermal-stress analysis
showed that compressive stress in the biaxial
state was generated in the areas channeled by
combustion flame at the sample surface, yielding
appreciable plastic deformation, which is responsible for the generation of large tensile
stresses exceeding the fracture strength of zirconia during the cooling cycle.
Improvements in the contact wear resistance
of high-speed tool steel can be achieved by reinforcement with aluminum oxide powder, but
will result in a decrease in the tensile strength.
The loss in strength can be recovered by selective lamination of layers of different reinforcing
content, which will make a functionally gradient
material (Ref 114). The sintering time and temperature could influence residual-stress distribution, taking full account of relaxation process
and volume change due to phase changes. Experimental stress analysis measurements served
to validate the analytical predictions.
Considerable attention has been paid to investigate the improvements in the fatigue life of sintered steel products by appropriate surface modification treatments such as surface rolling or
shot peening. The enhanced fatigue behavior of
the steel could be attributed to the microstructural features and residual-stress distributions.
Surface rolling could improve the fatigue

strength by a factor of 1.6 in the case of sintered
Fe-1.5Cu alloy and a factor of 1.3 in the case of
Fe-8Ni-1Mo-0.5C alloy (Ref 115). The improvement is caused by interactions such as surface
densification, increased hardness, improved surface roughness, and buildup of compressive residual stresses. The residual stresses were determined by x-ray, and through-thickness
residual-stress measurements were carried out by
incremental removal of very thin surface layers.
The hardness of high-strength alloy steel is much
higher than that of iron-copper alloys because of
the higher content of martensite. The maximum
values of the compressive residual stresses at the
surface of as-sintered states were caused by surface machining. However, when the surfaces are
rolled, the compressive residual stresses are no
longer determined by prior machining, but by the
level of yield stress and rate of densification, and
therefore these values are higher than those of
the sintered and machined surfaces. The amount
of improvement in fatigue life is related to the
stress-concentration factor, and the compressive
residual stress fields at a depth of 0.2 to 0.6 mm
will retard the propagation of surface cracks.
In the development of high-strength powderforged connecting rods, the fatigue strength is
governed by the surface defects such as decarburized layer, surface roughness, and nonsintered areas. The nonsintered area is attributed to
defective bonding caused by oxidation of the interface in the manufacture of connecting rods.
The oxidation may begin at the surfaces and proceed deeper during the transfer of porous preforms, heated to high temperature, from the sintering furnaces to the forging die in the normal
atmosphere. Controlling nonsintered areas and
imparting residual compressive stresses to the
surface are found to be the most effective means
of improving the fatigue strength (Ref 116).The
compressive stresses are produced by shot peening, which could improve the fatigue strength by
about 30% and could result in the production of
a 10% lighter connecting rod. The continuous
efforts to improve the strength/weight ratio
without affecting reliability have motivated further work, including the optimization of shot
peening and the evaluation of the magnitude of
residual stresses of powder-forged connecting
rods. Residual-stress measurements were made
using x-ray diffraction techniques. Maximum fatigue strength is obtained at a shot-peening intensity in the range of 15 to 20 A (Almen). Further increase in peening intensity resulted in a
decrease in fatigue life (Ref 117). This decrease
in properties is attributed to overpeening, resulting in surface extrusion faults. This phenomenon
can form cracks at the surface of the component
and lower its fatigue life. Optimization of the
peening process on fatigue test specimens resulted in an increase of 55% improvement; when
the process was carried out on forged connecting
rods (composition Fe-0.5C-0.1S-2Cu) using an
optimal peening intensity, an improvement of
27% of fatigue strength over the unpeened rods
at 90% reliability level resulted.
Conclusions
The P/M process has emerged as an advanced
manufacturing technology for the mass production of precision parts for general engineering,
automotive, electrical, electronic, and aerospace
industries. The process has become competitive
with other metalforming technologies, and there
is a continuous growth in the areas of steels,
stainless steels, tool steels, and metal-matrix
composites, which are difficult or impossible to
produce by conventional methods. The growing
powder-processing technologies include warm
compaction, liquid-phase sintering, high-temperature sintering, hot isostatic pressing, powder
forging, metal injection molding, and rapid prototyping.
Studies of residual stresses during the various
stages of powder metal processing are very fascinating and important from the view point of
producing quality products. These include the
production of metal powders by atomizing, nonequilibrium processing such as rapid solidification and mechanical alloying, compaction of
metal powders, sintering, hot pressing and hot
isostatic pressing, heat treatment and working
such as powder forging, and other P/M processes. However, only very limited investigations have been carried out in P/M processing,
and the major reported areas include heat treatment, microstructural evolution, and increasing
the fatigue life of sintered and powder-forged
components. In these cases, the residual-stress
measurements have been carried out using hardness and x-ray techniques, and more powerful
techniques such as neutron diffraction are yet to
be investigated. There is a need to develop prediction methods and measurement techniques for
residual stress measurements in P/M products,
and a better understanding of optimized residual
stresses can offer benefits such as enhanced service behavior and reliability to the components.
REFERENCES
1. D.G. White, Global Growth Trends at PM98, Granada Spain, Int. J. Powder Metall.,
Vol 34 (No. 8), 1998, p 9
2. A.J. Clayton, P/M Growth Reported in Las
Vegas, Int. J. Powder Metall., Vol 34 (No.
5), 1998, p 9
3. S. Pickering, The Road Ahead: PMs Future in the Automotive Industry, Powder
Metallurgy in Automotive Applications, P.
Ramakrishnan, Ed., Science Publishers,
1998, p 15
4. D.G. White, The Challenges of Growth:
State of the P/M Industry North America,
Int. J. Powder Metall., Vol 34 (No. 5),
1998, p 27
5. P.N. Singh, P. Ramakrishnan, and J.K.
Beddow, Characterization of Iron Powders
by Fourier Analysis Techniques, Advances
in Particulate Technology, M. Ramanujam, Ed., I.I.T. Madras, 1986, p 17
6. P.N. Singh and P. Ramakrishnan, Powder
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
Characteristics by Particle Shape Assessment, Kona, Powder and Particle, Hosokawa Powder Technology Foundation, Japan, No. 14, 1996, p 16
V.K. Sud, Expanding Horizon of Iron and
Alloy Iron Powders, Metal , Ceramic and
Composite Powders, P. Ramakrishnan,
Ed., Oxford and IBH Publishing Co. Pvt.
Ltd., New Delhi, 1990, p 13
S.I. Hulten, Evaluation of Iron Powders for
Structural Components, Powder Metall.,
Vol 13 (No. 6), 1970, p 234
F. Miami, A. Zambon, and E. Ramous,
Diffusion Bonding of Nickel Metal Powder on Iron Powders: A Microstructural
Study, Proc. Powder Metallurgy World
Congress (Granada, Spain), Vol 1, European Powder Metallurgy Association,
1998, p 245
G. Zapf and K. Dalal, Introduction of Oxygen Affinity Elements Manganese, Chromium and Vanadium in the Powder Metallurgy
of
P/M
Parts,
Modern
Developments in Powder Metallurgy, Vol
10, Metal Powder Industries Federation,
1976, p 129
K.S.V.L. Narasimhan and J. Tengzlius, P/
M Low Alloy Steel Alloying Methods and
Continuous Improvement, Advances in
Powder Metallurgy and Particulate Materials, Vol 5, Metal Powder Industries
Federation, 1992, p 153
S.H. Luk and J.A Hanil Jr., Dust and Segregation Free Powders for Flexible P/M
Processing, Advances in Powder Metallurgy and Particulate Materials, Vol 1,
Metal Powder Industries Federation, 1993,
p 153
Y. Trudel, M. Gangne, and F. Racicot, Design Criteria for the Manufacturing of Low
Alloy Steel Powders, Advances in Powder
Metallurgy, Vol 5, Metal Powder Industries Federation, 1991, p 45
R.J. Causton and T. Cimino, High Density
Processing of Cr-Mn P/M Steels, Advances in Powder Metallurgy and Particulate Materials, Vol 5, Metal Powder Industries Federation, 1992. p 89
K.U. Kainer and B.L. Mordike, Oil AtomizationA Method for the Production
of Rapidly Solidified Iron-Carbon Alloys,
Modern Developments in Powder Metallurgy, Vol 20, Metal Powder Industries
E.V. Axmin, Iron Powder with Low Oxygen Content, Proc. Powder Metallurgy
World Congress (Granada, Spain), Vol 1,
European Powder Metallurgy Association,
1998, p 72
A.J. Yule and J.J. Dunkley, Atomization of
Melts for Powder Production and Spray
Deposition, Clarendon Press, Oxford,
1994
S.A. Miller and R.S. Miller, Real Time Visualization of Close-Coupled Gas Atomization, Advances in Powder Metallurgy
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
and Particulate Materials, Vol 1, Metal

Powder Industries Federation, 1992, p 113
N.J. Grant, Powder and Particulate Production of Metallic Alloys, Advances in
Powder Technology, G.Y. Chin, Ed.,
American Society for Metals, 1982, p 1
C.P. Ashdown, J.G. Bewley, and G.B.
Kenney, The Tandom Atomization Process for Ultrafine Metal Powder Production, Modern Developments in Powder
Metallurgy, Vol 20, Metal Powder Industries Federation, 1988 p 169
J.E. Japka, Iron Powder for Metal Injection
Molding, Int. J. Powder Metall., Vol 27
(No. 2), 1991, p 107
Y. Kato, Ultrafine Powder for MIM and
Large Components Made Using MIM,
Powder Metallurgy in Automotive Applications, P. Ramakrishnan, Ed., Science
Publishers, 1998, p 75
J.D. Holmgren, J.O. Gibson, and C. Sheer,
Some Aspects of Arc Vaporised Submicron Particulates, Ultrafine Particles, W.
Kuhn, Ed., John Wiley & Sons, 1963, p
129
A.K. Joseph, R.R. Kapoor, V. Gopinathan,
and P. Ramakrishnan, Some Studies on the
Size Reduction of Steel Swarf to Powder,
Trans. Powder Met. Assoc. India, Vol 14,
1987, p 49
R. Anand, A.N. Tiwari, and P. Ramakrishnan, Investigation on Dual Phase Steel
through P/M Route, Key Engineering Materials, P. Ramakrishnan, Ed., Vol 2931,
Trans Tech Publication, Aedermannsdorf,
Switzerland, 1989, p 443
J.S. Benjamin, Mechanical AlloyingA
Perspective, New Materials by Mechanical
Alloying Techniques, E. Arzt and L.
Schultz, Ed., Informations Gesellschaft
Verlag, Adenaueralle, Germany, 1989, p 3
S. Takaki, Yu. Kimura, and T. Kama, Production of Hard Materials with High Chromium Carbide Matrix by MA Process,
Proc. Powder Metallurgy World Congress
(Granada, Spain), Vol 1, European Powder
Metallurgy Association, 1998, p 350
R.M. German, Powder Injection Molding,
Metal Powder Industries Federation, 1990
R. Burns, Production Presses and Tooling,
Powder Metallurgy, Vol 7, Metals Handbook, American Society for Metals, 1984,
p 332
P.J. James, Principles of Isostatic Pressing,
Isostatic Pressing Technology, P.J. James,
Ed., Applied Science Publishers, London,
1983, p 1
W.B. James, Overview of High Density P/
M Forming Processes, Int. J Powder Metall. Powder Technol., Vol 21 (No. 3),
1985, p 163
H.H. Hausner, K.H. Roll, and P.K. Johnson, Perspectives in Powder Metallurgy,
Vol 1, Consolidation of Metals Powders,
Plenum Press, 1967
K.S. Baird and J.D. Williams, Relation-
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
ships between Process Variables and Density of Explosively Compacted Iron Powder, Int. J. Powder Metall. Powder
Technol., Vol 20 (No. 1), 1984, p 23
E.Y. Gutmanas, Cold SinteringHigh
Pressure Consolidation, Powder Metal
Technologies and Applications, Vol 7,
ASM Handbook, ASM International, 1998,
p 574
H.S. Nayar, Sintering Atmospheres, Powder Metal Technologies and Applications,
Vol 7, ASM Handbook, ASM International, 1998, p 457
H.I. Sanderow, High Temperature Sintering, New Perspectives in Powder Metallurgy, Vol 9, Metal Powder Industries Federation, 1990
P. Ramakrishnan, High Temperature Sintering of Ferrous Alloys, Progress in Thermal Treatment of Materials, P. Ramakrishnan, Ed., New Age International (P) Ltd.,
New Delhi, 1996, p 115
S.N. Malhotra and P. Ramakrishnan, Corrosion Resistance of Stainless Steel Filters
Made by Powder Metallurgy, Proc. Int.
Conf Stainless Steels, Vol 2, The Iron and
Steel Institute, Japan, 1991, p 1117
E. Klar and P.K. Samal, Powder Metallurgy Stainless Steels, Powder Metal Technologies and Applications, Vol 7, ASM
Handbook, ASM International, 1998, p
774
M. Cowbiere and J.F. Faure, Activated
Sintering of Steel through the Presence of
Liquid Phase, Advances in Powder Metallurgy and Particulate Materials, Vol 3,
Compiled by J.M. Capus and R.M. German, Metal Powder Industries Federation,
1992. p 57
R.B. Dixon, W. Stasko, and K.E. Pinnow,
Particle Metallurgy Tool Steels, Powder
Metal Technologies and Applications, Vol
7, ASM Handbook, ASM International,
1998, p 786
L.A. Muller, R. Walling, N. Katrak, and R.
Chattopadhyaya, Manufacturing and
Properties of Sintered High Speed Steel
Products, Powder Metallurgy and Related
High Temperature Materials, P. Ramakrishnan, Ed., Trans Tech. Publications, Aedermannsdorf, Switzerland, 1989, p 565
M.T. Podob, L.K. Woods, P. Beiss, and W.
Happmann, The Mechanism of Sintering
of High Speed Steel Powders, Modern Developments in Powder Metallurgy, H.H.
Hausner, H.W. Antes, and G.P. Smith, Ed.,
Vol 13, Metal Powder Industries Federation, 1981, p 71
C.S. Wright, The Production and Application of PM High Speed Steels, Powder
Metallurgy World Congress (Paris), Vol
11, European Powder Metallurgy Association, 1994, p 937
M. Santos, M.M. Oliveria , M.M. Rebeck,
and A.S. Ernoski, Optimisation of Processing Parameters of T15 High Speed
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
Steel, Powder Metall., Vol 34 (No. 2),

1991, p 93
P. Ramakrishnan, Developments in P/M
High Speed Steels, Tool and Alloy Steels,
Vol 28 (No. 12), 1994, p 381
P. Billgren and O. Sandberg, A New Generation of PM Steels for New PM Compacting Tools and Other Tool Applications, Powder Metallurgy World Congress
(Paris), Vol 11, European Powder Metallurgy Association, 1994, p 1023
P.W. Tanbenblat and W.E. Smith, Infiltration of Iron P/M Parts, Copper Base Powder Metallurgy, Metal Powder Industries
H.I. Sanderow, J.R. Spirko, and R. Corrente, Machinability of P/M Materials as
Determined by Drilling Tests, Int. J. Powder Metall., Vol 34 (No. 3), 1998, p 37
P.J. Orgeske, Steam Treatment Improves
P/M Part Quality, Heat Treatment, May
1986, p 34
D.H. Herring and P.T. Hansen, Heat-Treating Ferrous P/M Parts, Adv. Mater. Process., Vol 4, 1998, p 44cc
A.B. Davala, A.H. Graham, and R.J. Causton, Effect of Process Conditions on Sinter
Hardening Response of P/M Alloy FLC
4608, Ind. Heat., Sept 1998, p 111
H. Chandler, Heat Treaters Guide, Practices and Procedures for Iron and Steel,
R. Trembley and R. Angers, Hot Pressing
of Iron Powder, the Role of Transformations, Int. J. Powder Metall., Vol 8 (No.
4), 1972, p 235
K. Isonishi and M. Tokizane, Superplastic
Deformation of a P/M Ultrahigh Carbon
Steel, Int. J. Powder Metall., Vol 25 (No.
3), 1989, p 187
F.X. Zimmerman, Design of Some Hot
Isostatic Pressing Equipment, Ind. Heat.,
May 1982, p 38
I.F. Anderson and A.C. Nyce, Pressure Assisted High Temperature Sintering of
Metal Alloys and Ceramics, Met. Powder
Rep., Vol 43 (No. 9), 1988, p 581
B. Ferguson, A. Kuhn, O.D. Smith, and F.
Hofstatter, Hot Consolidation of Porous
Preforms Using Soft Tooling, Int. J. Powder Metall. Powder Technol., Vol 20 (No.
2), 1984, p 131
K.M. Kulkarni, Powder Metallurgy for
Full Density Products, Vol 8, Perspectives
in Powder Metallurgy, Metal Powder Industries Federation, 1987, p 237
T.G. Johannission, H.G. Larson, and E.
Westman, Steel Billets Produced by New
P/M Process Using Horizontal Gas Atomization, Ind. Heat., May 1982, p 33
P.E. Evans, The Continuous Roll Compacting of Powders, Met. Mater., August
1968, p 235
W.G. Jaffery, I. Davis, and R.L.S. Taylor,
Thin Steel Strip from Powder, Sci. J., Vol
3A ( No. 3), 1969, p 61
63. C.A. Weaver, R.G. Butters, and J.A. Lund,

Hot Rolling Behaviour of Iron Powder Preforms, Int. J. Powder Metall., Vol 8 (No.
1), 1972, p 3
64. W.R. Morgan, An Evaluation of the Large
Scale Production of Mild Steel Tubes from
Powders, Powder Metall., Vol 12 (No. 23),
1969, p 107
65. H.G. Schuhbauer, Technological and Economic Comparison of Sintered and Cold
Extruded Parts, Powder Met., Third European Powder Metallurgy Symposium,
Part II, Institute of Metals, London, 1971,
p 469
66. C. Aslund, A New Method for Producing
Stainless Seamless Tubes from Powder,
Fifth European Symposium on Powder
Metallurgy (Stockholm), Vol 1, Jernkontoret, Stockholm, Sweden, 1978, p 278
67. Paris Arch Uses PM Tubing, Powder Metall., Vol 37 (No. 1), 1994, p 8
68. W.J. Huppmann, The Technical and Economic Development of Powder Forging,
Powder Metall. Int., Vol 13, 1992, p 186
69. W.B. James, M.J. McDermott, and R.A.
Powell, Powder Forged Steel, Powder
7, ASM International, 1998, p 803
70. L.A. Muller, E. Schneider, and O. El-Eibrich, Powder Metallurgy Connecting
Rod, Powder Metallurgy in Automotive
Applications, P. Ramakrishnan, Ed., Science Publishers, 1998, p 263
71. J. Lasalle, B. Sherman, and C. Jones, Net
Shape Processing Using an Aqueous
Based Injection Molding Binder in a Batch
Furnace, Ind. Heat., Feb 1999, p 57
72. Y. Kato, Ultrafine Powder for Metal Injection Molding of Automobile Parts, Powder
Ramakrishnan, Ed., Science Publishers,
1998, p 295
73. A.R.E. Singer, Recent Developments in
the Spray Forming of Metals, Int. J. Powder Metall. Powder Technol., Vol 21 (No.
3), 1985, p 219
74. A.G. Leatham and A. Lawley, The Osprey
Process: Principle and Applications, Int. J.
Powder Metall., Vol 29 (No. 4), 1993, p
32
75. L. Shaw and C. Spiegelhauer, Spray Forming Large Diameter Special Steel Billets,
Int. J. Powder Metall., Vol 33 (No. 3),
1997, p 31
76. J.C. Kosco, Mechanical Properties of
High-Performance Powder Metallurgy
Parts, Powder Metal Technologies and Applications, Vol 7, ASM International,
1998, p 947
77. F.C. Hanejko, Warm Compaction, Powder
7, ASM International, 1998, p 376
78. U. Engstrom and B.J. Johonson, Warm
Compaction of Densmix Powders, Powder
Ramakrishnan, Science Publishers, 1998,
p 133
79. M. Greul, F. Petzolett, M. Grenbich, and

W. Wander, Rapid Prototyping Moves on
Metal Powders, Met. Powder Rep., Vol 52
(No. 10), 1997, p 24
80. Powder Metallurgy Methods for Rapid
Prototyping, Powder Metal Technologies
and Applications, Vol 7, ASM International, 1998, p 426
81. E. Kula and V. Weiss, Residual Stress and
Stress Relaxation, Plenum Press, 1982
82. E. Macherauch and V. Huk, Residual
Stresses in Science and Technology, Vol 1,
Informations Gesellschaft Verlag, Adenaueralle, Germany, 1987
83. G.M. McColvin and M.J. Shaw, Volume
Manufacturing and Applications of Mechanically Alloyed Materials, Mechanical
Alloying, P.H. Shingu, Ed., Materials Science Forum, Vol 8890, Trans Tech Publications, Zurich, Switzerland, 1992, p 235
84. R.L. Hewitt, W. Wallace, and M.C. de
Malherbe, The Effect of Strain Hardening
in Powder Compaction, Powder Metall.,
Vol 16 (No. 31), 1973, p 88
85. R.L. Rewitt, W. Wallace, and M.C. de
Malherbe, Plastic Deformation in Metal
Powder Compaction, Powder Metall., Vol
17 (No. 33), 1974 p 1
86. B.L. Ferguson, G.J. Petrus, T.M. Krauss,
W.J. Porter Jr., and R. Brown, Deflections
in Compaction Tooling, Advances in Powder Metallurgy and Particulate Materials,
Vol 2, Metal Powder Industries Federation, 1992, p 251
87. H. Tsuru, and T. Nakagawa, Studies in
Crack Formation in P/M Compacting, Advances in Powder Metallurgy and Particulate Materials, Vol 2, Metal Powder Industries Federation, 1992, p 291
88. D.C. Zernger and H. Cai, Common Causes
of Cracks in P/M Compacts, Int. J. Powder
Metall., Vol 34 (No. 4), 1998, p 33
89. S.B. Brown and G. Abou-Chedid, Constitutive Behavior of Powders During Compaction, Proc. Powder Metallurgy World
Congress, Y.-Ch. Bando and K. Kosage,
Part 1, Japan Society of Powder and Powder Metallurgy, 1993, p 306
90. S. Shima and H. Kotera, 3-D Simulation
of Powder Characteristics and Compaction
Behavior, Powder Metallurgy World Congress (Paris), Vol 1, European Powder
91. N.A. Fleck, J. Mech Phys. Solids, Vol 43
(No. 5), 1995, p 1409
92. B.S. Torakers, S. Biwa, and P.L. Larsson,
Int. J. Solids Structures, Vol 34 (No. 24),
1997, p 3061
93. S.D. Mesarovic, N.A. Fleck, and J. Tweed,
Micromechanics of Powder Compaction,
Special Seminar on Modelling and Simulation of Compaction, Powder Metallurgy
World Congress (Spain), European Powder Metallurgy Association, 1998, CDROM
94. W.A. Kaysser, W.J. Huppmann, and G.
Petzow, Analysis of Dimensional Changes
95.
96.
97.
98.
99.
100.
101.
102.
During Sintering of Fe-Cu, Powder Metall., Vol 23, 1980, p 86

O. Dominguez and J. Bigot, The Effect of
Compaction Pressure and Heating Atmosphere on the Sintering Behavior of Iron
Nano Metric Powders, Proc. Powder Metallurgy World Congress (Paris), Vol 111,
European Powder Metallurgy Association,
1994, p 1785
J. Boiko, W. Schatt, F. Aldinger, and G.
Petzow, Densification Mechanisms During
the Initial Stage of Solid Phase Sintering,
H. Riedel, J. Svoboda, and H. Zipse, Numerical Simulation of Die Pressing and
SinteringDevelopment of Constitutive
Equations, Proc. Powder Metallurgy
World Congress (Paris), Vol 1, European
Powder Metallurgy Association, 1994, p
663
F.B. Swinkels and M.F. Ashby, A Second
Report on Sintering Diagram, Acta Metall.,
Vol 29, 1981, p 259
H.E. Exner and T. Kraft, Review on Computer Simulation on Sintering Process,
J. Tengzelius and M. Larson, Net Shape
after Sintering by Consistent Powder
Mixes, J. Jpn. Soc. Powder Powder Metall., Vol 44, 1997, p 313
W.B. Li, M.F. Ashby, and K.E. Easterling,
On Densification and Shape Change During Hot Isostatic Pressing, Acta Metall.,
Vol 34 (No. 12), 1987, p 2831
C. Argento, D. Bouvard, P. Stutz, R. Charlier, and A.M. Habraken, Numerical Simu-
103.
104.
105.
106.
107.
108.
109.
lation of Hot Isostatic Pressing of Metal

Powder: Influence of Constitutive Equations, Proc. Powder Metallurgy World
Congress (Paris), Vol 1, European Powder
Harry Chandler, Ed., Heat Treaters
GuidePractices and Procedures for Iron
and Steels, 2nd ed., ASM International,
1995, p 844
G.M. La Vecchia, R. Roberts, A. Bertasio,
and G.P. Marconi, Surface Residual Stress
Measurements on As-Sintered and Carburized Steel Parts, Proc. Powder Metallurgy
World Congress (Paris), Vol 11, European
Powder Metallurgy Association, 1994, p
855
I. Shraf and K.H. Boswell, Residual Stress
Measurements in Machined P/M Steel,
Advances in Powder Metallurgy and Particulate Materials, Vol 8, Metal Powder
Industries Federation, 1995, p 183
K.M. Kulkarni, Factors Affecting Dimensional Precision of MIM Parts under Production Conditions, Advances in Powder
Metallurgy and Particulate Materials, Vol
5, Part 19, Metal Powder Industries Federation, 1996, p 157
R.M. German, Computer Model for the
Sintering Densification of InjectionMolded M2 Tool Steel, Int. J. Powder Metall., Vol 35 (No. 4), 1999, p 57
J. Mazumder, J. Choi, K. Nagarathanam,
J. Koch, and D. Hetzner, The Direct Metal
Deposition of H13 Tool Steel for 3D Components, JOM, Vol 49 (No. 5), May 1997,
p 55
J.K. Lee, Stress Induced Microstructural
Evolution via Desecrate Atom Method,
Int. Mater. Rev., Vol 42 (No. 60), 1997, p
22l
110. S. Takaki and Y. Kimura, Grain Refining

of (d c) Two Phase Stainless Steel Powder Compact through Mechanical Milling
Treatment, Proc. Powder Metallurgy
World Congress, Y.O. Bando and K. Kosuge, Ed., Jpn. Soc. Powder Powder Metall., Part 1, 1993, p 151
111. J. Nutting, S. Ubhi, and T.A. Hughes,
Frontiers of High Temperature Materials,
J.S. Benjamin, Ed., Inco MAP, 1981, p 33
112. R.C. Klug, G. Krauss, and D.K. Matlock,
Recrystallization in Oxide Dispersion
Strengthened Alloyed Sheet Steel, Met.
Mater. Trans., Vol 27A, July 1996, p 1945
113. R. Watenabe, A. Hibino, and A. Kawasaki,
Thermomechanical Properties of Disc
Shaped Functionally Gradient Material
Fabricated by Hot Pressing, Advances in
Powder Metallurgy and Particulate Materials, Vol 9, Metal Powder Industries
114. R.A. Queeney, Residual Stresses and Mechanical Responses in Al2O3 Reinforced
M2 HSS Gradient Components, Int. Conf.
Powder Metallurgy and Particulate Materials (Vancouver), 2024 June 1999
115. C.M. Sonsino, F. Muller, and R. Muller,
The Improvement of Fatigue Behavior of
Sintered Steels by Surface Rolling, Int. J.
Fatigue, Vol 14 (No. 1), 1992, p 3
116. K. Takahashi, E. Hoshina, and M. Ogino,
Development of High Strength Powder
Forged Connecting Rods, Proc. Powder
Metallurgy World Congress (Paris), Vol
111, European Powder Metallurgy Association, 1994, p 2227
117. R.A. Chernekoff, S. Mocarski, and D.A.
Yeager, Increased Fatigue Strength of
Powder Forged Connecting Rods by Optimized Shot Peening, Powder Metall., Vol
38 (No. 3), 1995, p 196

DOI: 10.1361/hrsd2002p424

Residual Stress Formation and

Distortion of Rail Steel
F.D. Fischer and G. Schleinzer, Montanuniversitat Leoben, Austria
RESIDUAL STRESSES AND DISTORTION

OF RAIL STEEL are important to security and
comfort in railway traffic. In fact, some derailments were found to be initiated by residual
stresses, and the buckling of rails has already
caused some serious accidents. Fortunately,
great disasters are a rare occurrence in railway
traffic. Still wear, and thus maintenance costs,
are influenced by residual stresses. Due to increasing speed and rising wheel loads of freight
cars, more and more research interests have been
directed to the railway system, and railway administrations demand straightness and low tensile oreven bettercompressive residual
stresses in the head and in the foot near the surface to improve crack growth resistance.
In Fig. 1 different steps of the life of a rail are
illustrated. Two production steps in particular offer the possibility of enhancing the properties of
a rail of given metal composition (i.e., the cooling process and the straightening of the rail). Of
course, nobody expects significant deformation
during the service of a rail (that is, the deformation of a rail under wheel loads); thus, the
residual stress state is highly influenced by the
very localized nature of the stresses caused by
wheel/rail contact. In the case of a wavy or
rough rail surface (e.g., due to asperities), the
localization of the contact is even stronger.
The Cooling Process

The main geometrical feature of a rail is obviously the fact that it has only one symmetry
with respect to a plane orthogonal to the rail foot.
In addition to the single symmetry, the mass distribution along this plane varies significantly.
The rail head represents a mass concentration of
roughly square shape. However, the web and the
two halves of the foot can be approximated by
plates. Inevitably the cooling process of a rail
Metal
composition
Fig. 1
Hot rolling
A rail from production to service
profile with an initially spatially constant temperature distribution results in a strong inhomogeneous temperature field and a thermal
stress state. If the rail is free from external constraint, of course, neither a resultant moment nor
a resultant force can appear due to pure cooling.
Although the understanding of such a temperature field is of a certain technical interest and
has been a topic of research for many years, little
literature exists on the residual stress state due
to cooling. We use the word certain, because
some experts argue that only the stress state after
the roller straightening process, which follows
the cooling process, is of any relevance. However, if one wants to understand why the rail is
curved after cooling and if there are ways to reduce the curvature as much as possible, one has
to study the cooling process very carefully. Evidently, the curved size after cooling is the input
information for the straightening process. The
residual stress state due to cooling, however, is
fully eliminated during the straightening process.
One of the first papers dealing with residual
stresses in rails after cooling was published by
Mayer in the 1930s (Ref 1). Mainly experimental
work is reported. The method of cutting the rail
profile into small square columns (the sectioning method) was used, which obviously is too
inaccurate to give a really clear picture. A paper
later published by Asbeck et al. (Ref 2) also discusses experimental results.
One of the reasons for the near-absence of numerical and analytical work on residual stresses
in rails is the difficult geometry. Therefore, the
first real progress toward a better understanding
of the history of the residual stress state and the
deformation of the rail could be achieved by applying the finite element method (FEM), as in
one of the first works in this field by Marcelin et
al. (Ref 3). This paper could not provide a clear
picture about the development of the residual
Cooling
Straightening
stress state and the distortion of the rail because

the authors investigated a single rail in vertical
position; however, its merit has been to show a
way of applying FEM to this difficult problem.
To the knowledge of the writers the most comprehensive numerical study was presented in a
thesis by Hinteregger (Ref 4). Two accompanying papers were published, one dealing mainly
with the transient temperature field and the related residual stress state in a rail profile (Ref 5),
and the second paper considering long rails and
the variation of the residual stress state along the
rail (Ref 6). After the discussion of the temperature field and the method for calculating the residual stress state, the problem of long rails, the
rail end problem, is treated again in detail. Furthermore, it should be mentioned that mainly the
results of Hinteregger et al. (Ref 6) are discussed
in this chapter. Of course, finer meshes would
be applied today; however, the concept followed
can be considered as absolutely up-to-date. Also,
a distinct rail profile must be assumed for the
numerical calculation. The widely used UIC54E
profile is applied. The thermal material behavior
of this fine pearlitic material can be found in Fischer et al. (Ref 5). The austenite to pearlite
transformation occurs during cooling between
ca. 690 C and 610 C. The rail is represented
by a layer of generalized plane-strain elements
(see the mesh in Fig. 2). The base plane of this
layer is defined to remain in the plane of the
coordinate system. The cover plane, however,
Mounting
and welding
Fig. 2
Finite element mesh. Source: Ref 4
Residual Stress Formation and Distortion of Rail Steel / 425

dependence of the heat conductivity k, the specific heat capacity cp, and the mean coefficient
of thermal expansion m as a function of the current temperature T, m (T), expressed by:
(1)
(3)
(2)
Fig. 3
( kgkJ K (
1.0
Tref
(H)dH/(T Tref)
(Eq 1)
ri,j ui,j ei ej r Ai(T i4 T j4)
Q
Rails on the cooling bed. Source: Ref 4
may displace in the longitudinal direction and

may also rotate with respect to a rotation vector,
which is fully free. The global equilibrium conditions enforce no resultant longitudinal force,
no shear force on the cross section and no resulting moment. Consequently the rotation of the
cross section, and thus the distortion of the rail,
can be calculated. The thermal boundary conditions are explained in the next subchapter. All
calculations (thermal and mechanical) were performed with the program ABAQUS (Ref 7).
The Cooling Boundary Conditions and
Heat Transfer. It is assumed that the temperature of the rail is spatially constant at 1030 C
(1303 K) over the entire cross section at the beginning of the cooling procedure. This is the
measured surface temperature after the rail has
left the rolling mill. Cutting and transportation
take usually about 4 min until the rail reaches
the cooling area. Most of the rails cool down
with contact and thermal interactions with the
neighboring rails as shown in Fig. 3. A single
cooled rail is a special case that rarely occurs in
the course of rail production.
At temperatures higher than 500 C cooling is
mainly determined by radiation. Heat transfer by
convection becomes more important at lower
temperatures. Figure 4 shows the temperature
Cp
surface radiates to another area of the same body.

Because of the small temperature differences the
heat flux by self-radiation is rather small. Therefore, mainly the radiation into the surroundings
is of interest. The heat flux is governed by
(Ref 8):
where (H) is the instantaneous linear coefficient of thermal expansion at temperature H including both the temperature-induced and the
transformation-induced volume changes. Tref is
chosen to be 900 C.
As mentioned, the austenite-pearlite transformation takes place in the temperature range between 690 and 610 C. The latent heat q in W/
m3 set free during this phase transformation can
be represented either as a heat source on the right
side of the heat conduction equation or, approximately, by an artificial increase qDcp of the specific heat capacity qcp according to the relation
with the time t:
q(cp Dcp)
qDcp
T
T
div(k grad T) q qcp
t
t
div(k grad T)
T
qDcp
q
(Eq 2)
t
DT
q
Dt
qDcp
q Dt
DT
(Eq 3)
This practice was applied in the study explained

here.
Each point of the rail surface absorbs or transmits energy by radiation to the surroundings and
by self-radiation, which means that an area at the
( mW K (
k(K 1)
25.016
50
and
ui, j 12(sin 1 sin 2)
Cp
20.016
40
2 3
6.1
0.6
15.106
30
6
= 15
2 = 83
0.4
10.106
20
0
200
400
600
800
1000
1100
Temperature, C
15 14 13
Fig. 4
Dependence of the heat capacity, cp; conductivity, k, and the mean coefficient of the thermal expansion, m,
on temperature. Source: Ref 4
(Eq 5)
ri, j is the heat flux from the plane i to the

where Q
plane j; ei,ej are the emissivity coefficients; r
5,67 108(W/m2K2) is the Stefan-Boltzmann
constant; Ai is the area of plane i; Ti,Tj are the
temperatures of the planes i and j, respectively;
ui,j is the radiation shape factor; and 1,2 are
the angles between the normal of the examined
element and the tangent to the rail surface as
shown in Fig. 5.
The comparison between the total heat flux
and the heat flux due to radiation into the surroundings after 6 min cooling time is shown in
Fig. 6. The numbers indicate the fraction of heat
flux into the surroundings in percent. On the
lower side of the rail head 87% of the total heat
flux is radiated into the surroundings. The other
part is the heat flux from the head to the web and
the foot, respectively. The surface temperatures
at the web and the foot are lower than the head
temperature, and therefore, the heat flows from
the head to the other points.
A positive heat flux leaves the body. The total
heat flux in a point at the web is the sum of the
positive heat flux by radiation into the surroundings and the negative heat flux by radiation from
the neighboring areas of the profile. Because the
total heat flux is smaller than the heat flux into
the surroundings, values greater than 100% may
0.8
(Eq 4)
Fig. 5
10
12 11
Angles 1 and 2. Source: Ref 4

occur. The heat flux leaving the body is approximately the same as the heat flux entering the
body in a point at the rail foot. The total heat
flux is very small or zero, due to a thermal insulation, symbolized in Fig. 6 by two thick lines
at the foot surface. The percentage tends to infinity in this area.
The heat transfer is predominantly conductive
in the contact area to the neighboring rails. This
fact is taken into account in the computer model
by combining the head and the foot surface of
the same rail with six node interface elements
for heat transfer, as symbolized in Fig. 6 by the
shaded area.
Below 500 C convection becomes more and
ci from a plane to
more relevant. The heat flux Q
the surroundings is defined as:
ci Aih(T ) (Te T )
Q
(Eq 6)
where h(T ) is the heat-transfer coefficient, and

Te is the temperature of the surroundings. It is
convenient to approximate the contour of the
cross section by straight lines for calculating the
heat-transfer coefficients (Fig. 7). Simplified empirical relations for the heat transfer coefficient
h for free convection from various surfaces to
the air at different temperatures are given in Holman (Ref 8) as:

1/4
DT
h 1.42
L
(Eq 7)
for a vertical plane,

1/4
L
DT
h 1.32
(Eq 8)
2o
3o
6o
4u
2u
Fig. 7
7
5u
3u
6u
Approximated cross section for convection.
Source: Ref 4
96
57
25
89
5o
4o
100
100
93
for a horizontal plane facing upward, and

1/5
h 1.61
DT
(Eq 9)
for a horizontal plane facing downward. DT (C)

is the temperature difference between the surface
and the ambient temperature; L is the vertical or
horizontal dimension (m). The unit of the heat
transfer coefficient h is W/(m2K).
No further details are given with respect to the
calculation procedure because a transient temperature calculation is now state of the art. For
more information see Ref 7. It should be mentioned that the thermal and mechanical processes
are uncoupled in so far as the thermoelasticity
effect (contribution to entropy via strain energy)
is neglected since it is of no practical relevance.
The cooling process starts with free cooling
during the first four minutes. Then an interaction
takes place with the neighboring rails at the same
temperature level (Fig. 4, 8). After 100 min the
surface temperature reaches approximately 200
C, and the rail is lifted off from the cooling area.
The cross section has reached ambient temperature after 10 h cooling time. Generally the rail
foot cools down more quickly than the head due
to the mass difference of the head in relation to
the foot. Measured and computed temperaturetime histories are in good agreement. Figure 8
shows the temperature distribution after 4 min.
However, the situation is different in the case
of head-hardened rails. After hot-rolling, the
head of the rail is plunged into a quenching fluid
for 212 minutes. Then the rail is air-cooled as
usual. The temperature distribution in such a
head-hardening rail after 412 minutes cooling
time can be seen in Fig. 9. The composition of
the quenching fluid and, therefore, the heat transfer coefficient are treated as a certain industrial
secret. More details on the head-hardened rail
can be found in Moser et al. (Ref 9).
87
87
800
780
760
103
103
740
740
720
97
97
680
680
112
107
720
660
640
Fig. 8
Fig. 6
Heat flux into the surroundings in percent of the total heat flux. Source: Ref 4
Ref 4
680
740
720
680
660
Temperatures in C of a rail with interaction with

other rails after 4 min cooling time. Source:

The analysis of a residual stress state during
quenching or cooling can now be assumed as a
well-established concept. Specifically, the kinetics of the phase change phenomena, as well as
their mechanical appearance due to a transformation eigenstrain, must be understood. Here,
the writers refer, for example, to a widely accepted paper by Inoue et al. (Ref 10) and for
details to a paper by Tanaka et al. (Ref 11). In
the case of a diffusional transformation such as
the austenite-pearlite transformation, a volume
change of a few percent occurs. Note that the
transformation volume change itself is included
in the mean coefficient of thermal expansion.
This volume change cannot be accommodated
by elastic straining. Therefore, a local plastifi-
h
pl
TP
eij eel
ij eij eij eij
(Eq 10)
where eel
ij is the elastic part of the strain tensor;
ehij the thermal part being m(T Tref)dij; dij denotes the unity tensor, epl
ij the plastic part and, as
mentioned, eTP
ij the TRIP part. The TRIP part
(see Fischer et al., 1996) is derived as:
e TP
ij
490
where ry* is a weighted yield stress of both

phases; f represents the volume fraction of pearlite; w(
f ) describes the pearlite transformation kinetics; sij is the global (locally averaged) stress
deviator; and d is the transformation volume
change. Such a modification of the constitutive
law was not available when Hinteregger (Ref 4)
worked out his thesis. Therefore, the softening
due to the TRIP-effect was taken into account by
a reduction of the yield stress, as seen in Fig. 10.
Also, the Youngs modulus, Poissons ratio, the
mean coefficient of thermal expansion, and the
stress-strain curves are taken into account with
respect to their temperature dependence. For the
stress calculation the generalized plane-strain
model explained previously is used. It allows a
longitudinal displacement and rotations of the
cross-section. Figures 11 and 12 show the thermal longitudinal stresses along the symmetry
cation is activated which interacts with the

global load stress state. This phenomenon is denominated as transformation induced plasticity
(TRIP). Refer to a detailed overview by Fischer
et al. (Ref 12). As a practical device, an additional plasticity term eTP
ij is introduced with respect to the additive strain tensor eij decomposition:
3
dw

K
f sij
2
df
5
1
d
6 ry*
(Eq 11)
520
580
610
670
640
670
640
640
Temperatures in C of a head-hardened rail after

4.5 min cooling time. Source: Ref 4
Fig. 9
100
Yield stress, y 0.001(N/mm2)
100
50
150
100
50
Fig. 11
Longitudinal stresses in MPa along the axis of symmetry after 20 min cooling time. Source: Ref 4
Fig. 12
Longitudinal stresses in MPa along the axis of symmetry at the end of cooling. Source: Ref 4
75
"Mixture"
rule
50
Trip
25
0
600
650
700
Temperature, C
750
Temperature dependence of the yield stress

during the austenite-pearlite transformation.
Source: Ref 4
Fig. 10
100
50
50
100

axis for different time steps of a cooling procedure, taking into account full interaction with the
surrounding rails as shown in Fig. 3. It can be
shown that the influence of TRIP on the final
residual stress state is significant; this may be
one of the reasons for the wide scatter of measured residual stresses in railroad rails. These
results were calculated on the assumption of a
rail-by-rail arrangement on the cooling bed,
however, without frictional contact, which is discussed subsequently. It is important to note that
the residual stress distribution in a single rail differs significantly from Fig. 11 and 12 due to the
different temperature field of a simple rail. If the
rail is fixed with respect to its transversal deformation (a fork-type positioning of the rail), an
additional constraint is imposed on the rail leading also to a modified picture of the residual
stress state.
Head-hardening changes the picture of the residual stress distribution, as shown in Fig. 13. It
is interesting to note that the residual stress distribution is nearly independent of the lateral constraint of the rail in the case of head-hardened
rails.
Due to the changing temperature differences
between the head and the foot, the contraction of
the foot at the beginning of the cooling process
is greater than the contraction of the head. This
tendency changes several times during the cooling period. The sag f for a rail section of 30 m
in length versus time t is plotted in Fig. 14, considering a positive sag if the center of the curvature is at the head-side. One can see that the
sag changes significantly with time.
Weight and Friction: Rail End Problem. In
the treatment just described a cross section of a
rail is investigated under a given time-dependent
temperature field, and the assumption of a generalized plane-strain state only allows the analysis of an infinitely long rail. Furthermore, no
weight or friction is taken into account. However, actual rails have a length of 30 m (98 ft) to
more than 100 m (328 ft) and are subjected to
their own weight. The friction on the cooling bed
prevents the rail from deforming in transversal

direction. This fact leads to the surprising effect
that very long rails (approximately 100 m, or 328
ft, length) show an almost perfectly straight middle section M of at least 60 m (195 ft) length
on the cooling bed and two deformed rail ends.
Of course, after lifting the rail with a crane and
releasing the frictional forces, the rail becomes
suddenly curved with a sag of some meters.
Also, cutting a very long rail on a cooling bed
leads to a sudden change of the deformation
shape because the friction force distribution depends on the length of the rail. This phenomenon
has caused some severe accidents in the shops.
The first authors to describe the rail end problem
of a heavy rail on a rough surface during a cooling process were Fischer and Rammerstorfer in
1989 (Ref 13). The rail is modeled by a bar with
cross section A. For easier understanding assume
an elastic bar; its stiffness is EJ; E, Youngs
modulus; and J, cross-section inertia moment.
The bar is primarily loaded by a temperature
field T(z,t), which does not depend on the longitudinal coordinate x, leading to a temperature
moment:
mh(t)
E T(z,t)z dA
A
(Eq 12)
Since the bar may move on the rough surface

(representing the cooling bed) a transversal friction force, p, can be activated during the sliding,
thus:
| p(x)| lq
(Eq 13)
where l is the coefficient of Coulomb friction

(e.g., 0.45), and q is the weight per length unit
of the rail (e.g., 54 kg mass for a UIC54E rail
per meter length).
Since this is a sliding process, p(x) is always
oriented in the negative direction of the transversal velocity w/t with w being the transversal displacement. p(x) is, therefore, sectionally
constant and changes its sign at locations xi(t), i

1, 2, N where
w
t
xi
w
| xi 0
(Eq 14)
Fischer et al. (Ref 13) were able to show that

there are no sticking intervals. The differential
equation for the rail is:
2w
1
(M(x,t) mh(t))
x2
EJ
(Eq 15)
where M(x,t) is the moment due to the friction

force distribution (Fig. 15). Now in the middle
section M of the rail w(x,t) is observed to be
w(t)|M 0. Since also its second derivative is
zero, Eq 15 leads to:
M
mh

t
t
(Eq 16)
Since equilibrium enforces

M
Q
x
2M
Q
tx
t
(Eq 17)
Q being the resultant shear force in transversal

direction, it follows immediately from Eq 16
that:
Q
t

mh
x
t
(Eq 18)
Due to the initial conditions mh(0) 0, w(x,0)

0, M(x,0) 0, Q(x,0) 0, it follows also
that:
Q(x,t)|M 0
(Eq 19)
and, therefore:
Q
x
p(x)|M 0
(Eq 20)
Equation 20 allows the conclusion that the middle section M is not affected by any friction
force during cooling:
M(x,t)|M mh
Q(x,t)|M 0
p(x,t)|M 0
(Eq 21)
L 30 m
f, cm
f
50
0
3
9.103
t, s
50
100
150
Fig. 13
100
50
50
100
150
200
Longitudinal stresses in MPa of a head-hardened rail along the axis of symmetry at the end of cooling. Source:
Ref 4
Fig. 14
Sag, f, of a rail with a length of 30 m on a frictionless support. Source: Ref 4

Thus, the middle section M of the rail can be
considered as a long beam with an external moment mh on both ends.
It is now a very difficult problem to find the
sequence xi(t), i N. The most surprising effect
is that the two equilibrium conditions:
xN
xN
p(x)dx 0
p(x)(xN x)dx mh
together with the conditions (Eq 14):

w
|xi 0, i 1, N
yield a system of N 1 differential/algebraic

equations for N unknowns xi, i 1, N (Ref
14). This obvious paradox can be solved only if
N is assumed as N . A detailed description
of the solution routine is presented in Ref 14
going back to an earlier work (Ref 15).
Dimensionless interval boundaries n i:
ni
xi
l*
l* (mh(t)/lq)1/2
(Eq 22)
are introduced. A numerical procedure gives:

n 1 1.041,
n 2 2.387,
n 3 2.802,
n 4 2.948,
n 5 3.003,
n 6 3.024,
It can be seen that the interval boundaries come

closer and closer with increasing n. The se-
quence n n, n , reaches a limit value n

3,029. It is interesting to note that Stupkiewicz
and Mroz (Ref 16) found the same solution independently. Recently Mogilevsky et al. (Ref
17) published a concept for a finite-length beam
with a sequence of friction intervals with changing friction force sign.
The finite value of n points to a perfectly
straight middle section M of the rail. Of
course, since the material is assumed to be elastic
and mh vanishes for a rail at room temperature,
no final deflection would occur. Nevertheless,
this elasticity friction study leads to a better understanding of the complicated phenomenon of
the moving/sliding of the rail on the cooling bed.
The investigation of the real situation of an
elastic-plastic rail on a real cooling bed consisting of a grid with discrete supporting bars with
spacing being orthogonal to the rail was elaborated by Hinteregger (Ref 4), Fischer et al. (Ref
18), and Hinteregger et al. (Ref 19). In the following text the results for a 100 m (328 ft) rail,
which is placed on a grid, are reported. The rail
was modeled by an overlay of six bars (B32
beam elements of Type ARBITRARY) adequately representing the area, as well as the moment of inertia of a UIC54E rail profile. The
same constitutive law was assumed for the beam
cross section as in the two-dimensional study
just described. The stick-slip effect of a supporting bar is represented by an elastic-ideally
plastic material with the yield stress rpy (lqa/
As), As being the cross section of the supporting
y
z
W(x, t )
z
T(z, t )
z
x
x0
m (t )
bar. The Youngs modulus E p of the supporting

bar is selected as E p 100E. Since the supporting bar is modeled as a truss element fixed
at one end, using the constitutive law explained
previously, the correct sign of the friction force
with the magnitude lqa is automatically taken
into account. According to the calculated temperature field, the temperature is defined in 32
points of the cross section with spatial linear interpolation. In each equivalent point with respect
to the longitudinal x-direction, the same time history of the temperature T( y,z,t) is defined. Both
the transversal displacement w(x,t) and the longitudinal stress rx(x,y,z,t) are calculated. In the
middle section M the rail remains almost totally straight, which corresponds to a fork-type
positioning of the rail. Therefore, the residual
stress state calculated due to this constraint in
the previous section can be used to check
rx( y,z,t)|M.
The partial plastification of the rail during
cooling leads to a deformed shape. Figure 16
shows the deformation shape of a half 100 m rail,
with the length counted from the middle of the
rail. Furthermore, Fig. 16 shows that only the
rail end is deformed, covering roughly a range
of 25 m. Figures 17 to 19 demonstrate the longitudinal residual stress rRx (x,y,z) at room temperature after lifting of the rail. rRx calculated by
the overlay-beam-model at hand agrees well
with the calculated values from the two-dimensional model with a fork-type positioning of
the rail in the middle section M.
Experimental Results. The findings of the
previous section on a long rail led to a new view
on the results of experimental investigations of
the residual stress state. Usually a certain part of
a rail (with a length of about 1 m, or 3.28 ft) is
cut out and experimentally investigated. Figures
17 to 19, however, demonstrate that the residual
stress state varies significantly within the last 25
m (82 ft) of a long rail. That means the place
where a part of the rail has been cut out for experimental investigation must be defined precisely. Of course, for sake of comparison, it
should be cut out from the middle section M.
22
q
+ q
11
x2(t )
x3(t )
x i (t )
xN(t )
End portion
L( t )
Straight
middle
part
Width, cm
x1(t )
6
11
22
w
t
33
0.0
4
10.0
20.0
30.0
40.0
50.0
Length, m
Displacement of a 100 m rail depending on the
cooling time. 1, 400 s; 2, 660 s; 3, 920 s; 4,
1200 s; 5, 3600 s; 6, 6000 s; 7, 36,000 s. Source: Ref 4
Fig. 16
Fig. 15
Frictional forces on a continuous cooling bed. Source: Ref 4

The strong variation of the residual stress state
in the rail-end region is obviously one of the reasons for a significant scatter of the experimental
results reported in the literature (Ref 1, 2, 20). If
the reported results on residual stresses are depicted in one graph, a scatterband emerges depicted as a shadowed region in Fig. 20. It is interesting to note that specifically the results
published by Kolmogorov et al. (Ref 20) fit well
the calculated results of this work relating to the
middle of a fairly long rail. Also, the results
found by Marcelin et al. (Ref 3) agree well, both
100
3
, N/mm2
50
0
2
1
2
50
3
100
1
150
0
10
20
30
50
40
qualitatively, and in part, quantitatively, with the

results reported here.
A further comment is necessary with respect
to experimental investigations. Often, the holedrilling method is applied, which is assumed to
be an established technique (Ref 21). However,
a numerical simulation of this hole-drilling process by Fischer et al. (Ref 22) has revealed that
the evaluation method of surface deformations
by strain gages leads to wrong results if there is
a steep stress gradient in the virgin material. This
is a real shortcoming and has led to many misinterpretations of the residual stress state in rails
since just near the head of the rail one can often
observe a very steep gradient after the cooling
process and the subsequent straightening process.
As a conclusion, one must accept that rails,
after cooling on a cooling-bed, are curved at
room temperature. However, several attempts
are under way to achieve a straight rail after the
cooling process. This can only be performed
with a specific cooling procedure, such as by
contact with copper bars as discrete heat extractors on the surface of the rail. Several rail
producers are working on such a procedure but
maintain strict silence on the outcome.
Length, m
Fig. 17
Longitudinal stresses of a rail lifted off the cooling bed. Source: Ref 6
After hot rolling, the rails cool down on a

cooling bed. Due to the different cooling velocities in the cross section, the rails are distorted
after having reached room temperature. Since
there are certain requirements for the rail
straightness, not the least of which is to ensure
passenger comfort, a further production stepa
straightening stepmust be added. Although
there are different possibilities (e.g., stretch
100
2
, N/mm2
50
0
2
3
1
50
3
100
Roller Straightening
straightening), roller straightening, as used for

other long products, is the most common
method. Roller straightening is much faster than
other methods because it allows continuous
operation, and the underlying principle is fairly
simple. The cool, distorted rails are drawn
through the straightener by a series of driven
rollers, which are located alternately below and
above the rail. While the rail is passing through
the rollers in the longitudinal direction, it is bent
up and down. Thus, the initial distortion is homogenized and decreased by the plastic deformation caused by a combination of bending,
shear, and roll contact stresses. This operation is
quite important because it is the final deformation process of the production route, which determines the residual stress pattern (Fig. 21).
There is always a springback effect when the
outer forces are released after a plastic deformation. Thus, a new equilibrium of forces and
moments caused by residual stresses are
achieved. A typical residual stress pattern develops during the straightening, which is nearly
independent of the initial rail distortion before
the straightening operation. As tensile residual
stresses promote the initiation and propagation
of cracks, railway producers try to achieve low
residual stresses. The most critical point here is
the foot, where both the longitudinal tensile
stresses due to the wheel loads and the residual
stresses superpose. However, obtaining a
straight rail and reducing residual stresses are
Roller deflection
Fig. 21
Roller-straightening process
150
0
10
20
30
50
40
Length, m
Fig. 18
100
3
3
50
, N/mm2
1
0
2
2
50
100
1
150
0
10
20
30
40
50
Length, m
Fig. 19
150
Fig. 20
100
50
0
N/mm2
50
100
150
Scatterband of longitudinal residual stresses of unstraightened rails from literature. Source: Ref 4

difficult tasks because the straightness also depends on the distribution of the residual stresses.
The roller deflections can be changed on an
existing roller straightener in order to improve
the residual stress pattern and the straightness of
the rail. Up to now the setup of the roller positions has been a matter of trial and error, largely
influenced by the skills and expertise of the operator.
Railway administrations demand low residual
stresses in rails (e.g., that the longitudinal resid-
Strain
gage
1m
20 mm
Saw cut
Fig. 22
Measurement of the longitudinal stresses at the

rail foot according to a CEN draft standard
300 200 100
100
200
300 MPa
Head
Web
Foot
Fig. 23
300
370 BHN, neutron

370 BHN, sectioning
200
Stress, MPa
Scatterband of the longitudinal residual stresses

taken from Hodgson. Source: Ref 25
100
0
100
200
300
Fig. 24
100
50
Rail height, mm
150
Longitudinal residual stresses of a 370 BHN

new roller straightened rail. Source: Ref 30, 31
ual stress, which is the most critical component,

must not exceed 250 MPa, or 36 ksi, in the middle of the foot). According to a European Committee for Standardisation (CEN) draft standard
(Ref 23), the measurement of this residual stress
component has to be performed in the following
way: A strain gage is attached in longitudinal
direction in the middle of the rail foot. Then, a
saw cut in front of and behind the gage position
and normal to the longitudinal direction relieves
the residual stresses. The stress relief causes an
elastic strain, which can then be measured by the
strain gage. Finally, the original longitudinal
stresses can be calculated using the Youngs
modulus (Fig. 22).
Residual Stresses in Unused RollerStraightened Rails. Generally speaking, the
longitudinal residual stress pattern in the cross
section is a C shape with tensile stresses in the
head and the foot and compressive stresses in the
web. A comprehensive study about stresses in
roller-straightened rails was done by the ORE
Committee (Ref 24), which used destructive sectioning methods. Due to the sectioning, the rail
parts expand or shrink elastically when separated
from one another. The resulting elastic strains
can be calculated by measuring the length
changes, or they can be directly measured by
strain gages. Of course, these measurements
must be done very carefully so that the cutting
or frictional heating does not influence the results. Also, the spatial resolution is not very
high. Hodgson (Ref 25) summarized the results
of the ORE Report in a scatterband of the longitudinal stresses (Fig. 23).
Similar results are reported in Ref 26 to 28.
An equivalent scatterband can be found in Wineman et al. (Ref 29). The accuracy of the measurements can be verified by checking the force
and moment equilibrium in the cross section.
Later, in 1992, the European Rail Research
Institute (ERRI) (Ref 30) reported new measurements by neutron diffraction performed by Webster et al. (Ref 31). Neutron diffraction is a nondestructive testing method. In contrast to x-rays,
neutrons can penetrate material more easily, allowing the measurement of residual stresses in
the interior of the material. However, it is a
rather expensive and very time-consuming
method; moreover, a nuclear reactor is necessary
to provide the neutrons. Monochromatic neutrons can be scattered at the lattice planes of a
crystalline material. If the beam path length between the planes is a multiple of the wavelength
of the neutron, then constructive interference occurs. Therefore, only under certain beam angles
will diffraction take place. Also, if stresses in the
material alter the lattice distances, the scatter angle changes. For more details about the neutron
scattering method see the article Methods for
Determination of Inhomogeneous Residual
Stress Fields in this Handbook. Unfortunately,
the investigation of the interior of the specimen
is limited. Due to the scattering, fewer and fewer
neutrons reach points deep inside the material.
To ensure that a reasonable amount of neutrons
reach the detector, a certain total beam path
length should not be exceeded. In steel, for example, the maximum path length is only a few
centimeters. Webster (Ref 31), therefore, chose
a longitudinal and a normal plate of the rail for
the analysis. The gage volume was 2 2 2
mm3, which is typical for neutron diffraction.
Hence, the results are averaged values inside this
volume. In qualitative terms, his results are well
in accordance with the ORE Report (Fig. 23 and
24). Just near the head surface a certain stress
reversal was observed. The longitudinal tensile
stresses dropped from a maximum beneath the
surface to a lower value at the surface. Already
the ERRI report mentioned that results near the
specimen surface must be interpreted with caution due to the extrapolation of the data. An additional problem near the surface is parasitic
peak shifts if the neutron gage is not entirely
immersed in the specimen. Hauk and Kockelmann (Ref 32) also reported a steep stress gradient near the surface with compressive stresses
at the surface. The stresses were measured by xray diffraction. But in their paper the longitudinal residual stresses changed from compressive
to tensile stresses within 0.1 mm beneath the surface. Thus, the general stress pattern seems to be
clear, although near the surface there are still uncertainties (Fig. 24).
Behavior of Rail Steel under Plastic Deformation. There are different types of rail steel:
the plain carbon, the alloy, and the heat treated
qualities. The different qualities should maximize the lifetime of the rails under different operating conditions. In this case the focus is on
describing the behavior of these steels under cyclic plastic deformation. Plastic deformation
causes the material to undergo a certain change
in shape and remains after the removal of the
outer forces. In the case of straightening of rails
and rails in service, the loading occurs repeatedly
and with changing signs. Therefore, the emphasis is on the cyclic plastic deformation.
Plastic deformation is based mainly on the
gliding of lattice planes in the crystalline structure of metals; this gliding requires dislocations
in the lattice. Although a certain amount of dislocations are always present in a metal, they are
also generated by the plastic deformation itself.
On the other hand, the piling up of dislocations
decreases the mobility of the dislocation again,
and the material shows hardening. If loading is
reversed, it has often been observed that the elastic limit increases in one loading direction while
it decreases in the opposite loading direction.
This is called the Bauschinger effect and has to
be considered in the case of cyclic loading (Fig.
25).
Polycrystalline materials in the virgin state are
isotropic. von Mises assumed that only the second invariant of the stress tensor accounts for the
elastic limit. The material shows elastic behavior
as long as the stress state stays within the von
Mises yield surface. Thus, his theory is also
called J2-flow theory, which is a good approximation of real metal behavior.
After plastic deformation the material is no
longer isotropic, and additional internal state

variables must be used to define the hardening
of the material. A common approach for cyclic
plasticity uses the decomposition of the hardening into kinematic and isotropic parts. These
models still contain the von Mises yield function
and evolution laws for the internal state variables
such as the yield stress rY, which defines the
radius of the elastic range (Mises cylinder) and
the backstress tensor ij, which describes the
shift of the Mises cylinder in the principal stress
space (Fig. 26).
The main issue of a material model is to describe the evolution of elastic yielding and the
state variables during plastic deformation. In
1979 Chaboche et al. proposed a multiple component non-linear hardening model. Today this
model is widely accepted and a basis for further
developments in plasticity models.
The decomposition of the stress tensor into a
hydrostatic and deviatoric part gives:
2
J2(Sij ij) 12(Sij ij)(Sij ij) 13rY
(Eq 24)
rY rY,0 Q(1 e beeqv)
The flow rule is modified to:

e pij k(Sij ij)
(Eq 25)
where k is a constant dependent on deformation
400
nij
0
0.005 0.000
200
0.005
0.010
400
600
800
/Pa
800
Stress, MPa
600
600
400
400
200
200
0
0.005
0.005
0.010
0.005 0.000
200
0.005
0.010
400
200
600
400
800
600
Fig. 25
Cyclic loading of rail steel UIC 900A
/Pa
800
600
p
ij
400
200
Sij
Sij ij
0.010
0.005 0.000
200
0.005
0.010
400
ij
600
Fig. 26
800
(a) Measurement: cyclic tension-compression

test (b) compared with a linear kinematic material model and (c) Multiple component model of Chaboche.
Fig. 27
von Mises yield surface and its shift due to kinematic hardening in the principal stress space
(Eq 27)
2 n p
p
c e ij c nnije eqv
3
(Eq 28)
200
Then a modified von Mises yield surface for kinematic hardening is necessary, where rY is the
common elastic yield value in a simple tensile
test:
0.010
nij
n1
where each backstress component evolves due

to a nonlinear law:
600
(Eq 23)
Strain
0.010
(Eq 26)
where rY,0 is the initial yield stress, Q and b are

hardening parameters, and epeqv is the accumulated plastic strain.
Next, the backstress tensor is decomposed into
an arbitrary number of components:
ij
/Pa
800
0.010
rij rdev
rhyd
Sij dij 13tr(rij)
ij
ij
history. The evolution of the yield radius is assumed to be:
c n and c n are the kinematic hardening parameters, which are different for each backstress component.
The kinematic hardening equations take into
account the transient hardening effects in each
stress-strain loop. The merit of Chaboches
model is its easy mathematical structure, which
allows a simple way to determine model parameters. The parameters can be determined from a
simple cyclic uniaxial tension-compression test.
Chaboches model allows an almost exact representation of the hysteresis curve under cyclic
loading if the material shows no out-of-phase
hardening. This out-of-phase hardening or additional hardening may occur if the strain paths
change from longitudinal to shear paths and vice
versa during the loading. Such nonproportional
paths occur during the roller contacts. A comprehensive study about the latest developments
in material models can be found in the thesis of
Jiang (Ref 33) and in Chaboche (Ref 34) (Fig.
27).
Because rail steel shows no out-of-phase
hardening, a good representation of the plastic
response of the material can be obtained with
Chaboches multiple component plasticity
model. The use of a simpler material model requires caution because it can strongly influence
the calculated results on residual stresses.
Simulation of the Roller Straightening. Different attempts were made to simulate the roller
straightening. Brunig (Ref 35) used beam elements of different widths accounting for the special cross section of the rail. He already used an
enhanced two-surface plasticity model. Because
elements covering the whole width do not allow
a variation of stresses along the width, his results
are, at best, average values. He also stated that
the contact pressure at the rollers had only a minor influence on the residual stresses. Naumann
(Ref 36) investigated a full three-dimensional
FEM model with an explicit integration method,
a coarse mesh, and linear kinematic hardening.
However, he did not publish residual stress results from his calculations. In contrast to Brunig,
Naumann stated that the pressing of the rollers
into the rail head and foot is in fact responsible
for the residual stress pattern. Weiser (Ref 37)
applied straightening mechanics on a slice of

that the submodel was loaded with the correct
forces and bending moments (Fig. 28).
The straightening simulation was repeated
with the submodel, still including the contact of
the rollers with the rail. The global model was
calculated with ABAQUS/Explicit. The explicit
integration had the advantage of being able to
handle contact problems more easily. For the
submodel ABAQUS/Standard was used because
the standard integration allowed a fine mesh.
Eight node brick elements with reduced integration were used for both models. The rollers were
modeled with analytical rigid surfaces. Thus, the
total calculation time on a single node workstation could be reduced to about two weeks (Fig.
29).
The results of this simulation showed a Cshaped residual stress pattern. The submodeling
technique was also tested with two-dimensional
models and was found to be correct. Thus, new
evidence was found with this model that the general idea of the residual stress pattern is true.
Already Wineman (Ref 29) and others stated that
a shortening of the material in the rail head and
foot due to the rolling contact leads to tensile
residual stresses. In Fig. 30 the total strains of a
subsurface element in the middle of the rail head
during the straightening process are shown. The
dominant mechanism is that a small element in
the head or base near the roll becomes shorter in
length (e33), shorter in height (e22), and wider
(e11). This leads to a mismatch in length between
200
200
150
150
50
/MPa
100
50
100
0
50
50
50
150
100
0
0.2
0.4
0.6
Foot
0.8
1.0
Mesh of the submodel of a UIC60 rail and longitudinal residual stresses in MPa along the symmetry line after
the straightening and as a contour plot
Fig. 29
20
16
11
12
8
3
4
0
33
4
8
22
16
Submodel
+
Fig. 28
+
+
0
200
Head
12
50
Strain, 10
rail. He implemented the roller contact with an

overlay of Hertz-type contact pressure. The results correspond qualitatively with the measurements. Finstermann et al. (Ref 38) applied moments and forces of a beam model on a
three-dimensional FEM model of a piece of rail
in order to bend it around a top and bottom roller.
The repeated contact with the top and bottom
roller was supposed to give a good approximation of the real roller contact. The results show
compressive stresses at the head and foot surface
that are much higher than the ones hinted at in
the measurements of Webster or Hauk. Varney
et al. (Ref 39) again used a beam model with a
linear kinematic hardening law. The result is a
zig-zag pattern of longitudinal stresses. Henderson et al. (Ref 40) modeled the process by a series of three-point bendings and linear kinematic
hardening. A first result had only little resemblance to measured U- (or C) shaped patterns.
After the material parameters had been adjusted
to unstraightened and straightened conditions,
the simulation gave better results.
Although these studies explain many mechanisms of straightening and offer help in understanding the process, none of them has yet
solved the formation of residual stresses quantitatively. The key problem is that limited computing resources made various simplifications
necessary, either by using beam or two-dimensional elements, neglecting the rolling of the
rollers or using linear kinematic hardening for
the material. Especially two-dimensional simulations with linear kinematic material hardening
lead to zig-zag patterns of the longitudinal residual stresses, which are not in agreement with the
measurements. In addition, the residual stresses
after straightening, which are calculated in a
two-dimensional model, strongly depend on the
plasticity model used. However, an FEM simulation is quite time consuming, and contact problems make the time problem even worse. Nevertheless, the authors tested a new concept for a
full three-dimensional simulation of the roller
straightening process with a minimum of simplification.
In order to save computation time, the simulation was split into two FEM processes. In the
first step a global model, 16 m long with a coarse
mesh and linear kinematic hardening law, was
applied. In the second step a submodel, only
three times as long as the height of the rail with
a fine mesh and Chaboches multiple component
hardening law, was used. The front and back surface of the submodel was driven by the nodal
displacements of the global model. This ensured
Roller 2
Driving a small submodel through the roller

straightener
Roller 3
Roller 4
Roller 5
Roller 6
Roller 7
Roller 8
Time
Fig. 30
Total strains in the subsurface layer in the middle of the head during roller straightening. e11 is the vertical
strain (width), e22 is the horizontal strain (height), and e33 is the longitudinal strain.

flanges and web and gives the C-shaped longitudinal stress distribution.
Rails in Service
Temperature Influence. The influence of
ambient temperature on the residual stresses
should be mentioned. Like all other materials,
steel has a certain thermal expansion coefficient.
Therefore, rails expand or shrink due to temperature changes. With an expansion coefficient
of 12 106 the longitudinal stresses shift
by 25 MPa (4 ksi) with a temperature change of
10 C. Under hot conditions rails can buckle,
while under cold conditions again tensile
stresses increase the possibility of rail fracture.
Therefore, railway administrations try to mount
the rails at average temperatures.
Residual Stresses due to Welding. Railway
rails are produced in lengths up to about 100 m.
In the past bolting was the common method of
joining rails on the track, but at these joints repeated impacts occur, increasing the possibility
of rail failure. Also, a different residual stress
pattern at the rail endsthe longitudinal stresses
in the head and in the base change to vertical
tensile stresses in the websupport web cracks.
And finally, passenger comfort is decreased by
the bumps and additional side forces in curves.
Currently rails are joined by welding. As a
result, the rail is continuous, and no bumps or
vertical residual stresses at the rail ends occur.

During welding the rail ends are heated up and
the former residual stresses vanish. Then, the
new joint cools down again, and new residual
stresses develop. The most common method for
joining rails on the track is aluminothermic (termite) welding: The two rails become bonded by
superheated metal, which is the result of a highly
exothermic reaction of iron alloy, iron oxide, and
aluminum powder. The cooling of the welded
gap then forms new residual stresses. Although
this welded joint is of superior quality, some
welds fail. Again, Webster et al. (Ref 41) investigated residual stresses in and near the welded
joint of a rail. They reported tensile longitudinal
stresses in the web region with steep gradients
to high compression in the head and the foot.
The stress pattern almost looks completely opposite to the distribution in new rails (compare
Fig. 23 and 31). The length of this zone depends
on the length of the mold, which has an insulating effect. However, it was found that approximately 90% of aluminothermic welds fail due to
porosity caused by dampness that can arise if the
molds are not thoroughly dried before welding.
Residual Stress Formation in Rolling Contact. Many investigations of residual stresses of
rails in service have been made. The motion of
a wheel on a rail has a complex character consisting of elasto-plastic rolling and sliding under
many circumstances (braking, acceleration of the
locomotive, running in curves). Wheel passages
on a rail cause plastic deformation in the running
surface layer and lead to particular changes of
the stress distribution in a rail head. Under heavy
traffic, wheel loads can be so severe that rails
fail due to the plastic deformation. A zone 1 to
2 mm (0.04 to 0.08 in.) deep is sheared relative

to the bulk material, and the strains can exceed
100% deformation. This deformation is caused
by the high normal and tangential forces of the
wheels. Each pass of a wheel leads to additional
plastic flow until the shakedown limit is reached.
The following is a brief overview of the results
of experimental and theoretical investigations
that are an important step toward understanding
the mechanisms of railway track failures.
Webster et al. (Ref 31) reported neutron measurements of used 370HB and 340HB rails. The
tensile longitudinal stresses in the subsurface
layer of the rail head of new rails are reduced to
zero or even negative values during operation.
This of course enhances the crack resistance of
the running surface (Fig. 32). Also, Bower et al.
(Ref 42) showed measurements of residual
stresses of used rails, which are definitely compressive near the running surface.
Looking at theoretical investigations different
approaches have been proposed to evaluate residual stresses after two-dimensional rolling
contact. Sehitoglu and Jiang (Ref 43) proposed
an analytical method for the determination of residual stresses in elasto-plastic rolling contact.
They found compressive longitudinal stresses
that reach a maximum at a distance of approximately 0.5 to 1 times the contact half length a
beneath the surface. In 1997 McDowell proposed a new hybrid scheme (Ref 44) that asymptotically approaches the solution of Sehitoglu and Jiang. Zochowski et al. (Ref 45), who
applied FEM, showed the additional effects of
acceleration or braking of a wheel on residual
stresses (Fig. 33).
In the free-rolling case Zochowski found ten-
0
3
2
3
1
, MPa
200
200
1
400
Longitudinal residual stresses in an aluminothermic weld. Source: Ref 42
y/a
Fig. 31
y/a
400
2
2
300
New
Used
200
3
100
0
4
100
Foot
200
Fig. 32
Web
Head
50
100
Rail height, mm
150
Change of the longitudinal residual stresses

during service. Source: Ref 31
1
2
3
4
3
Free rolling
q = 0.3
q = 0.75
q = 1.0
0.255 0.17 0.085 0
4
0.085 0.17
res
x/po
(a)
1
2
3
Free rolling
q = 0.3
q = 0.7
0.34 0.255 0.170.085 0
0.085
res
x/po
(b)
Longitudinal residual stresses rx due to free rolling, q 0, (a) acceleration q 0 and (b) braking q 0. y is
the depth, a is the contact half length, p0 is the maximum of the Hertz contact pressure distribution, and q is
the ratio of normal and tangential force. Source: Ref 46
Fig. 33

sile stresses at the surface, where McDowells
results tend to zero. Nevertheless, the general residual stress patterns of both investigations are
similar. Zochowskis results also show that braking and acceleration lead to compressive stresses
at the surface. These theoretical results agree
with the measurements of Webster et al. (Fig.
32) in the sense that where the theoretical results
show the generation of compressive stresses in
the subsurface layer of a formerly stress-free material, the measurements exhibit a decrease of the
tensile residual stresses.
ACKNOWLEDGMENT
We gratefully appreciate the cooperation with
Voest-Alpine Schienen GmbH & Co KG, which
has been supporting our research for more than
15 years. We also want to express our gratitude
stertoward the Austrian Science Foundation O
reichischer Forschungsforderungsfonds (FFF
project no. ZL 3/12384) for its financial support.
10.
11.
12.
13.
14.
REFERENCES
1. H. Mayer, Residual Stresses in Railroad
Rails, Organ f.d. Fortschritt des Eisenbahnwesens, Vol 91, 1936, p 320329 (in German)
2. H.O. Asbeck and M. Heyder, Residual
Stresses in Unused Rails after Cooling and
Straightening, Eisenbahntechnische Rundschau, Vol 26, 1977, p 217222 (in German)
3. J.L. Marcelin, M. Abouat, and J.L. Chenot,
Analysis of Residual Stresses in Hot-Rolled
Complex Beams, Comp. Meth. Appl. Mech.
& Eng., Vol 56, 1986, p 116
4. E. Hinteregger, Residual Stresses and Distortion of Rails after Rolling before
Straightening, Ph.D. thesis, Montanuniversitat Leoben, 1990 (in German)
5. F.D. Fischer, E. Hinteregger, and F.G. Rammerstorfer, The Influence of Different Geometrical and Thermal Boundary Conditions
and the Phase Transformation on the Residual Stress State in Railroad Rails after HeatTreatment, International Conference on Residual Stresses, ICRS 2, G. Beck, S. Denis,
and A. Simon, Ed., Elsevier Applied Science, London and New York, 1989, p 467
471
6. E. Hinteregger, F.D. Fischer, F.G. Rammerstorfer, and A. Joller, Calculation of the
Longitudinal Residual Stresses of Long
Rails during Cooling, Berg- und Huttenmannische Monatshefte (BHM), Vol 135,
1990, p 437441 (in German)
7. ABAQUS Finite Element Analysis Products, Hibbit, Karlsson & Sorensen Inc.,
www.hks.com
8. J.P. Holman, Heat Transfer, 4th ed., McGraw-Hill-Kogakusha Ltd., Tokyo et al.,
1976
9. A. Moser, P. Pointer, and G. Prskawetz,
15.
16.
17.
18.
19.
20.
21.
22.
Production of Head Hardened Rails from

the Rolling Heat, Berg- und Huttenmannische Monatshefte (BHM), Vol 133, 1988,
p 321326 (in German)
T. Inoue, S. Nagaki, T. Kishino, and M.
Monkawa, Description of Transformation
Kinetics, Heat Conduction and Elastic-Plastic Stress in the Course of Quenching and
Tempering of Some Steels, Ingenieurarchiv, Vol 50, 1981, p 315327
K. Tanaka and Y. Sato, A Mechanical View
of Transformation-Induced Plasticity, Ingenieurarchiv, Vol 55, 1985, p 147155
F.D. Fischer, Q.-P. Sun, and K. Tanaka,
Transformation-Induced Plasticity (TRIP),
Appl. Mech. Rev., Vol 49, 1996, p 317364
F.D. Fischer and F.G. Rammerstorfer, The
Thermally Loaded Heavy Beam on a Rough
Surface, Trends in Applications of Mathematics to Mechanics, W. Schneider, H. Troger, and F. Ziegler, Ed., Longman Scientific
& Technology, Harlow, 1991, p 1021
L.V. Nikitin, F.D. Fischer, E.R. Oberaigner,
F.G. Rammerstorfer, M. Seitzberger, and
R.I. Mogilevsky, On the Frictional Behaviour of Thermally Loaded Beams Resting on
a Plane, Int. J. Mech. Sci., Vol 38, 1996, p
12191229
L.V. Nikitin, Bending of a Beam on a
Rough Surface, Dokl. Ak. Nauk, Vol 322,
1992, p 10071016 (in Russian)
S. Stupkiewicz and Z. Mroz, Elastic Beam
on a Rigid Frictional Foundation under
Monotonic and Cycling Loading, Int. J. Solids Struct., Vol 31, 1994, p 43194342
R.I. Mogilevsky and L.V. Nikitin, In-Plane
Bending of a Beam Resting on a Rigid
Rough Foundation, Arch. Appl. Mech., Vol
67, 1997, p 535542
F.D. Fischer, E. Hinteregger, and F.G. Rammerstorfer, A Computational Study of the
Residual Stress Distribution in Thermally
Loaded Beams of Arbitrary Cross-Section
on Frictional Support, Nonlinear Computational Mechanics, P. Wriggers and W. Wagner, Ed., Springer Verlag, 1991, p 737750
E. Hinteregger, F.D. Fischer, and F.G. Rammerstorfer, The Influence of Frictional
Forces on the Longitudinal Distribution of
Thermally Induced Residual Stresses in
Beams, Residual Stresses III, Science and
Technology, Vol 2, H. Fujiwara, T. Abe, and
K. Tanaka, Ed., Elsevier Applied Science,
1992, p 12781283
S.V. Kolmogorov, D. Makarov, and O.N.
Mikhailov, Formation of Residual Stresses
in Heat Treatment Hardened Rail, Steel in
the USSR, Vol 17, 1987, p 271273
G.S. Schajer, Measurement of Non-Uniform
Residual Stresses Using the Hole-Drilling
Method, Part I and Part II, J. Eng. Mater.
Technol. (Trans. ASME), Vol 110, 1988, p
338343, 344349
F.D. Fischer, E. Hinteregger, and F.G. Rammerstorfer, Numerical Simulation of an
Experimental Stress Analysis, Material-
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
prufung, Vol 32, 1990, p 181185 (in German)

Railway ApplicationsTrack Heavy Rail,
Part 1: Flat Bottom Symmetrical Railway
Rails 46 kg/m and Above, CEN/TC256/
WG4 Draft Standard, 1997
Studies Concerning the Measurement and
Improvement of the Level of Residual
Stresses, ORE D 156/Report 4, Utrecht, the
Netherlands, 1987
W.H. Hodgson, Residual Stresses in Rail.
Rail Quality and Maintenance for Modern
Railway Operation, J.J. Kalker et al., Ed.,
Kluwer Academic Publishers, the Netherlands, 1993, p 6173
J. Szelazek, Ultrasonic Measurement of
Thermal Stresses in Continuously Welded
Rails, Nondestruct. Test. Eval., Vol 25,
1992, p 7785
C. Urashima and K. Sugino, Generation
Mechanism of Residual Stresses in Rails,
International Conference on Residual
Stresses, ICRS 3, H. Fujiwara, T. Abe, and
K. Tanaka, Ed., Vol 2, 1992, p 14891493
S.V. Kolmogorov, Y.D. Makarov, and O.N.
Mikhailov, Formation of Residual Stresses
in Heat Treatment Hardened Rail, Steel
USSR, Vol 17, 1987, p 271273
S.J. Wineman and F.A. McClintock, Residual Stress and Web Fracture in RollerStraightened Rail, Residual Stress in Rails,
O. Orringer, J. Orkisz, and Z. Swiderski,
Ed., Vol 2, 1992, p 122
Measurement of Residual Stresses of
Grade A and Two Head Hardened Rails
with Neutron Diffraction, ERRI D 173/Report 4, Utrecht, the Netherlands, 1992 (in
German)
P.J. Webster, G. Mills, X. Wang, and W.P.
Xang, Residual Stress Measurements in
Rails by Neutron Diffraction, Rail Quality
and Maintenance for Modern Railway
Operation, J.J. Kalker et al., Ed., Kluwer
Academic Publishers, the Netherlands,
1993, p 307314
V. Hauk and H. Kockelmann, Residual
Stress State in the Running Surface of a
Rail, HTM-Harterei-technische Mitteilungen, Vol 49, 1994, p 340352 (in German)
Y. Jiang, Cyclic Plasticity with an Emphasis on Ratchetting, Ph.D. thesis, University
of IllinoisUrbana, 1993
J.L. Chaboche, Constitutive Equations for
Cyclic Plasticity and Cyclic Viscoplasticity,
Int. J. Plast., Vol 5, 1989, p 247302
M. Brunig, An FE-Model for the Simulation of the Roller Straightening of Heavy
Profiles, Mitteilung, 895, Ruhr-Universitat Bochum, 1989 (in German)
N. Naumann, Straightening of Long Products about the Main Axis of Maximum Inertia, Ph.D. thesis, Montanuniversitat Leoben, Austria, 1998 (in German)
J. Weiser, Analysis of the Formation of Residual Stresses During the Roller Straightening of Rails, Ph.D. thesis, Otto-von-

Guericke-Universitat Magdeburg, Germany, 1997 (in German)
38. G. Finstermann, F.D. Fischer, G. Shan, and
G. Schleinzer, Residual Stresses in Rails
Due to Roll Straightening, Steel. Res., Vol
69, 1998, p 272278
39. B.E. Varney and T.N. Farris, Mechanics of
Roller Straightening, 39th Mechanical
Working and Steel Processing Conf., Iron
and Steel Society of AIME, Warrendale,
PA, Vol 35, 1998, p 11111121
40. S. Henderson, W.T. Cook, L. Woollard,
A.D. Richardson, and M.J. Wood, Control
of Straightening Operations for Optimisa-
tion of Product Properties, EUR 18540 EN,

Office for Official Publications of the European Commission (eur-op.eu.int), Luxembourg, 1998
41. P.J. Webster, G. Mills, X.D. Wang, W.P.
Kang, and T.M. Holden, Residual Stresses
in Alumino-Thermic Welded Rails, J.
Strain Anal., Vol 32, 1997, p 389400
42. A.F. Bower and K.L. Johnson, Shakedown,
Residual Stress and Plastic Flow in Repeated Wheel-Rail Contact, Rail Quality
1993, p 339349
43. H. Sehitoglu and Y. Jiang, Residual Stress

Analysis in Rolling Contact, Rail Quality
1993, p 349358
44. D.L. McDowell, An Approximate Algorithm for Elastic-Plastic Two-Dimensional
Rolling/Sliding Contact, Wear, Vol 211,
1997, p 237246
45. M. Bijak-Zochowski and P. Marek, Residual Stress in Some Elasto-Plastic Problems
of Rolling Contact with Friction, Int. J.
Mech. Sci., Vol 39, 1997, p 1532

DOI: 10.1361/hrsd2002p437

Residual Stresses during Gear

Manufacture
K. Funatani, Nihon Parkerizing Co., Ltd., Japan
peak of residual stress vary, depending on distribution of carbon, quench condition, and properties of the steel used. Various investigations on
the residual stress state of case-hardened steels
have been done, and many simulation models are
being developed to estimate the distribution of
residual stress in relation to various controlling
factors as follows (Ref 17):
Distribution of longitudinal, tangential, and
axial residual stresses
Difference of residual stress of carburized and
pseudocarburized steel
Influence of carburizing condition and carbon
gradient
Influence of quenching media and cooling
condition
Influence of steel grades and their hardena-
bility
ments of the core. When the carbon concentration becomes higher than 0.75 or 0.8% C, retained austenite is observed in relation to alloy
elements and results in a decrease of surface
hardness as retained austenite volume increases.
Generally, 20 to 30% of residual austenite is observed when surface carbon percentage is around
0.85 to 0.9%. The amount of retained austenite
increases by increasing carbon concentration at
the surface layer.
Distribution of Residual Stress in
Carburized-and-Hardened Steels. The surface
layer of carburized-and-hardened steels has compressive residual stresses as shown in Fig. 1, and
the compressive residual stresses counteract applied tensile stresses and improve bending fatigue strength. The typical profile of residual
stress produced in carburized-and-hardened steel
Influence of specimen or work size and case
depth
800
Residual Stress in Carburized-andHardened Steels

Carbon, %
Carbon Gradient, Hardness, and Residual

Stress in Carburized Steels. The typical distribution of hardness, carbon gradient, and longitudinal residual stress in case-hardened steels are
shown in Fig. 1. The surface hardness is about
700 to 850 HV and gradually decreases to the
same hardness as the core or center of hardened
steel. Effective case depth (ECD) is the depth
from the surface at which 513 or 550 HV, equivalent 50 or 55 HRC is maintained. Total case
depth (TCD) is the depth at which the carbon
content is 0.04% higher than that of the core. The
TCD varies depending on steel chemistry, time,
temperature, and carbon potential of the carburizing process. The ECD is affected by the carburizing condition similar to that of TCD,
quenching condition, hardenability of the core,
and mass and size of the workpiece. Surface
hardness varies depending on carbon concentration, which is controlled by carburizing atmosphere and the amount and types of alloying ele-
700
A
0.8
600
500
0.6
Hardness, HV
Influence of microstructure and other factors
0.4
B
0.2
Residual stress, kg/mm2
CARBURIZING is a commonly used process

for improving fatigue strength and wear resistance of mechanical components such as gears
and many other automotive components. The
surface and core hardness and case depth are
specified in design drawings, which influence the
strength of carburized-and-hardened steels. The
hardness and case depth are influenced by steel
hardenability and various processing variables
such as carburizing and quench condition.
Along with heat treatment results affecting
with hardness, carburizing and quenching operations introduce compressive residual stress into
the surface layer of steels. Compressive stress
contributes very effectively to the fatigue
strength of case-hardened products by reducing
applied tensile stresses. Surveys of various investigative results showed that measured peak
compressive residual stresses ranged from 200
to 550 MPa.
It is very important to introduce compressive
residual stress into the surface of carburized
parts; this is performed by appropriate component design and proper processing technologies.
Similar compressive residual stress is attainable
by surface induction hardening, which also has
a large influence on bending fatigue strength.
Additional posttreatment such as shot blasting
and/or double-hard peening can further increase
the surface compressive residual stress and contribute more to the improvement of fatigue
strength than can the traditional shot peening
methods. This article explains the fundamental
state of residual stress of carburized-andhardened steels in relation to each factor.
Case hardening by carburizing or carbonitriding and quench processes provides the major
benefit of introducing large compressive residual
stress into the surface of the parts. It results from
uneven volume expansion caused by transformation from austenite to martensite. The residual
stress at the surface of case-hardened steel is
generally compressive because of the larger expansion of high-carbon and/or nitrogen layer
into martensite during transformation. The compressive residual stress at the surface layer is effective for increasing fatigue durability of the
case-hardened steels. However, distribution and
0
10
C
20
30
0
0.4
0.8
1.2
Depth from surface, mm
1.6
Carbon gradient, hardness, and longitudinal residual stress produced in carburized-andhardened steel. Carburized 5 h at 925 C, quenched at 860
C. A: SCM 22, 9 mm diam; B: S15CK, 25 mm diam; C:
SCM 420, 9 mm diam. Source: Ref 6
Fig. 1
Tensile
stress
Case
Core
Radial
0
Compressive
stress
Longitudinal
Tangential
some distance from the surface.
Distribution of residual stresses in longitudi-
nal and tangential directions is nearly the

same, but that of radial direction differs.
The excessive case carbon concentration increases retained austenite, which decreases
surface hardness and compressive residual
stresses.
Maximum value and distribution of residual
stress are influenced by heat treatment methods such as carbon potential of carburizing
atmosphere and quenching.
Although some investigators concluded that the
influence of steel type and the conditions of carburizing and quenching are minor (for example,
Ref 3, 7, and 9), these factors have a profound
influence on the performance of carburized-andhardened products as explained later in this article.
Influence of Steel Grade on Residual
Stress. Steel grade has a profound influence on
distribution of residual stresses. Figure 3 shows
the distribution of residual stresses of two casehardening grades steels, nickel-chromium alloy
K13NiCr12 steel (Fig. 3a) and straight carbon
C15 steel (Fig. 3b). Residual stress of alloy steel
shows lower stress near the surface due to retained austenite, and plain carbon steel shows the
highest stress at the surface.
Influence of Size and Workpiece Shapes.
Size and shape of testpiece or work have profound influences on magnitude and profiles of
residual stresses. Figure 4 shows the influence of
specimen size on distribution of residual
stresses. Generally, as the size of the specimen
increases, the maximum compressive stress increases, and also the highest stress peak gradually moves inward. However, residual stress profile near the surface not only varies depending
on carbon concentration and cooling process, but
also is affected by complicated vaporizing and
rewetting phenomena during cooling.
Methods of Measuring Residual

Stress
Parallel beam counter methods

Cr K, 30 kV, 8 or 10 mA, vanadium filter
Count range: 200 cps, time constant: 16 s
Dispersion angle: 0.35, focus: l 2 w
5 or 6 mm
Sin2 W: W 0, 15, 30, 40
The distribution of residual stresses are measured under these conditions, and about 5 to 10
20
ra*
0
Compressive residual stresses have a peak at
for the effective penetration depth of x-ray and

changes in residual stress resulting from layer
removal. Layer removal with the x-ray method
has shown excellent agreement when used to
evaluate residual stress distribution.
Measurement of the residual stress discussed
in this section used scanning x-ray method under
the following conditions (Ref 5,6):
tg
ax*
20
tg*
40
ax
60
80
0.4
0.2
0.6
0.8
1.0
0.8
1.0

(a)
20
ra *
0
20
tg*
40
tg
60
ax*
80
ax
100
120
0.2
0.4
0.6

(b)
Depth
Fig. 2
nonmartensitic phases such as bainite, troostite,

or pearlite to form. Also, when surface carbon
concentration increases to higher than 0.9% the
surface microstructure retains more austenite and
reduces hardness and compressive residual stress
by reducing the expansion that accommodates
the transformation to martensite.
Generally, distribution of residual stress as
shown in Fig. 1 and 2 has the following characteristics:
is shown in Fig. 2. Residual stress distribution,

especially at the surface, varies depending on
steel properties and carburizing-and-hardening
conditions such as carbon potential, time,
quench media, and cooling method as explained
later in this article. Generally, residual stress profiles in the longitudinal and tangential direction
are similar, while that of axial direction differs
from the other two directions. High compressive
residual stresses are produced by expansion of
carburized case at the end of quench-cooling
process, which balance with the tensile stress in
the core.
The state of residual stresses introduced in
hardened case of carburized-and-hardened steel
is generally in compression. Low-carbon casehardening steels are carburized and quenched in
selected quench media. During quenching, the
subsurface portion of the core transforms first at
a high temperature to ferrite and pearlite, bainite,
or low-carbon martensite (700550 C, depending on steel grade) in relation to transformation
plasticity. Later, at a lower temperature (lower
than 300 C), the carburized case transforms to
martensite with volume expansion under influence of the transformation plasticity caused by
the preceding transformation phase of the core.
High-carbon case of 0.8% C expands nearly
1.2%, while that of low-carbon core of 0.2% C
is about 0.3%, and this large difference in carbon
level and expansion determines the sequence of
phase transformation and resultant development
of compressive stress in the case. Compressive
residual stress in the case is influenced by core
hardness or strength; however, the relation between core strength and residual stress is not
simple and is explained in later sections.
However, the control of carburizing-andhardening conditions is very important to prevent introduction of tensile residual stress at the
surface. Especially, when the surface layer has
considerably thick grain-boundary oxidation, the
quenching time after pullout from the furnace
and the subsequent cooling speed should be optimized to reduce formation of nonmartensite
phases. The lack of surface hardenability results
from grain-boundary oxidation and low-carbon
contents easily cause transformation products of
Schematic of residual stresses in carburized-andhardened steels. Source: Ref 8
X-Ray Stress Analysis Methods. There are

several methods of measuring residual stresses,
such as distortion measurement, strain gage, xray diffraction, and neutron radiation methods.
Distribution of residual stress can be measured
by repeating mechanical or chemical methods to
remove core portion or outer layer and correcting
Distribution of residual stresses measured by xray produced in gas-carburized hardened steels.

(a) K13NiCr12; ECD: 0.40.6 mm, carburized and double
quenched. (b) CK15 (AISI 1015); ECD: 0.50.6 mm carburized and hardened. rax, rtg, and rra are measured residual stresses in axial, tangential, and radial directions, respectively. Curves with asterisk show corrected curve of
measured residual stress distribution. Source: Ref 7
Fig. 3
Residual Stresses during Gear Manufacture / 439

lm surface layer of specimen is chemically
etched repeatedly. Calibration on the influence
due to removal of surface layer is calculated
from the thickness of removed layer determined
by the diameter and residual stress measurements before chemical etching. This procedure
is repeated to obtain distribution of residual
stresses. This conventional method measures
only one of the W incidental angle directions.
It is thus unclear whether the direction of the
principal stress is perpendicular or inclined to the
sample surface. Therefore, the measured residual
stresses have certain variance and are not as accurate as measurements by the recently developed position-sensitive proportional counter
(PSPC) method.
The technology in residual stress measurement by x-ray diffraction has made great advancement, and PSPCs or position-sensitive proportional detectors (PSPDs) are widely accepted.
Compared to old scanning methods, which took
many hours, equipment with a PSPD, such as
PSPC, can measure residual stress in a short
time. The PSPC is a linear counter that obtains
diffracted x-rays simultaneously, counting over
a section of 50 to 100 mm or 20 range (2h
range) in the longitudinal direction (Fig. 5). New
systems measure both W tilt directions to ascertain the direction of the principal stress. Also,
recently developed x-ray analysis equipment enables residual stress measurement of microscopic areas of 0.15 mm diameter in combination with incident collimator (Ref 10).
In addition, recently developed equipment
uses computers for immediate data processing
with a data base on diffraction and materials constant, and it is a very effective tool for residual
stress research. However, it is very important to
set the focus precisely when investigating ma-
terials with fiber flow or sharp distribution of

residual stresses, and also care should be taken
to avoid errors in measurement of small diameter
bar or rod and corner or fillet of workpieces such
as gear tooth roots. It is necessary to apply lowangle diffraction methods to measure residual
stress for investigation of highly oriented microstructures and sharp stress profiles. The positivepole figure analysis for highly oriented microstructures is recommended.
Residual Stress Variation in

Testpieces and Gears
Residual Stresses in Testpieces
The distributions of carbon content, hardness,
and residual stress in three 9 mm fatigue test
specimens carburized and quenched in cold oil
shown in Fig. 6 have small variations (3 kg/
mm2). On the other hand, the residual stress distribution of rectangular plate and cylinder rod
carburized and quenched in hot oil shown in Fig.
7 has large fluctuation. This variation seems to
be caused by the difference in the cooling process; the 9 mm testpieces were cooled more uniformly than the rectangular plate. During cooling
of the plate and rectangular bars in hot oil, a
vapor blanket formed in the early stage of cooling at the surface broke nonuniformly, and the
resulting fluctuation of cooling speed caused the
variation of residual stresses observed in Fig. 7.

The wavy distribution of residual stresses
shown in Fig. 7 was introduced in the testpiece
quenched in still hot oil, where the vapor blanket
surrounding the workpiece varied, depending on
the testpiece shape, causing an unstable oil flow
and temporary change of surrounding quenching
condition. However, the surface residual stresses
produced in carburized plates and rods quenched
in hot oil are within a range of agreeable variance. Table 1 compares total area in compression
of agitated conditions with that of quenching in
still hot oil, which increases slightly with increasing quench oil agitation.
Residual Stresses in Production

Gears
The durability of gears used for automotive
transmission should be qualified by standardized
processing condition of carburizing-andhardening operation carried out in the heat treatment plant. Fatigue strength of those gears varies
as influenced by processing variables such as
carburizing and quenching conditions. The quality of produced gears is usually evaluated based
on surface and core hardness, case depth, and
microstructure of testpieces carburized simultaneously with gears. In some cases, metallurgical
quality is checked and residual stresses of gear
Preamp A
Preamp B
Delay line

1
Residual stress, kpsi
Diffracted
x-rays
Center wire (anode)

Detector gas
C
0
Cathode
10
E
10
20
Residual stress, k/mm2
20
X-ray tube
Incidence
x-rays
20
40
Crystal grain
0
0.08
0.04
Depth from surface, in.
Influence of specimen size on the distribution of

residual stress. A: 12.7 mm diam, AISI 8610 steel,
TCD 1.524 mm, tempered at 148.9 C. B: 19.1 mm
diam, AISI 8610 steel, TCD 1.524 mm, tempered at
148.9 C. C: 25.4 mm diam, AISI 8610 steel, TCD 1.524
mm, tempered at 148.9 C. D: 6.35 mm diam, AISI 8610
steel, TCD 1.27 mm, tempered at 190.6 C. E: 38.1 mm
diam, AISI 8610 steel, TCD 1.02 mm, tempered at 148.9
C. AD, source: Ref 3; E, source: Ref 2
TA = D, TB = D(L )
0.12
TA TB = D(L 2)
d
Fig. 4
Principle of detector used in x-ray stress measurement by PSPC system. Delay time per unit: Dns /mm; time to
produce output at A: TA; at B: TB; TB TA position v. h, x-ray angle; v, distance from edge of counter end;
g, reflection angle of x-ray; L, length of counter; D, delay time (n s/mm); d, lattice space of the crystal grain. Source: Ref
10
Fig. 5

ference results from reproducibility of heat treatment condition. The variation of treatment condition of the induction-hardened gear is more
stable than in the carburizing process, where
many gears are hung or mounted on a tray and
quenched in oil at once. The cooling condition
of each gear differs depending on position in the
tray, and also the cooling rate at the foot of gear
tooth is slower than at the tooth tip, reducing
transformation speed and resulting in the increases of the formation of nonmartensite phases
at the surface.
The variation in average value and range of
group A is smaller than that of group B. Average
value and variation range of group B varies
much more widely and seems to indicate that the
variation of quenching condition of B group was
not the same for each gear.
Residual Stress of Gas-Carburized and
Direct-Quenched Gears. An average residual
stress was measured on production gears made
of JIS SCr 420 of module 3. Gears were heat
treated in a continuous gas-carburizing furnace
with gas atmosphere using the boost-and-diffuse
method, in which surface carbon content before
itself are measured on limited samples. Measurement of residual stresses of gears takes a
long time, and a lot of laboratory work is necessary; therefore, quality control of gear strength
needs precise process control of key factors.
Bending fatigue strength of gear teeth depends
on hardness, case depth, and magnitude of compressive residual stresses at the tooth root portion. Measurement of residual stress at this point
is not easy because of limited space and round
root shape, and the measured value is not as accurate as measured stresses on flat testpieces or
other open faces. Surface residual stress of production gears varies more than that of simple
shape test samples heat treated in precisely controlled conditions.
Residual Stress Variation at Tooth Root
Surface of Induction-Hardened and Carburized Gears. Table 2 shows the variation in residual stress measured at the foot of gear teeth
of induction-hardened gears and carburized-andhardened gears. Compressive residual stresses
produced in induction-hardened gears are higher
than that of carburized gears, and their variation
is smaller than that of carburized gears. This dif-
quenching is controlled to about 0.9% C and directly quenched in hot oil of 130 C. The residual
stress level was 24.8 4.25 kg/mm2 in compression, which is quite an agreeable result compared with that of other testpieces carburized and
quenched in same furnace.
Axial Stress Distribution Estimated in Carburized Gear during Quenching (Ref 11).
Figure 8 shows the details of generation of axial
stress distribution estimated by calculation of a
carburized gear during quenching. In the early
stages, the contour areas of equal stress were almost unaffected by the surface profile. Later, a
zone of high compressive stress distribution occurred in the central subsurface portion of the
teeth. Such high stress has a large influence on
transformation plasticity and determines distortion, sometimes resulting in cracking.
During cooling, stress distribution changes are
caused by shrinkage and expansion by transformation to martensite. The state of stress distribution is estimated by a simulation model, as
shown in Fig. 8. However, in actual quenching,
the cooling speeds at the tooth tip and tooth foot
differ extensively, and vapor generation accel-
20
SCM 420
600
50 HRC (513 HV)
500
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
10
20
S
O
H
30
40
300
0.5
1.0
1.5
2.0
S15CK
(a)
TCD
20
10
10
400
Carbon content, %
ECD
Hardness, HV (5)
700
10
800
20
SCM 420
30
0
10
20
40
40
S
O
H
30
0.5
1.0
1.5
2.0

50
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
20

Carbon gradient, hardness, and residual stress variation produced in carburized specimens. Steel: SCM 420
(Jominy hardenability 12 in. 34.8 HRC), rotating beam bending fatigue test specimen; 9 mm diam. Carburizing
process: 920 C, 0.9% CP for 5 h. Quenched from 880 C in JIS-1-1 cold oil (3550 C) and tempered at 160 C for 2 h
(7.2 ks). Carbon concentration is measured in S15CK straight carbon steel 25 mm diam specimen. ECD 1.22 mm,
surface 824 HV (10), core 468 HV (10). CP, carbon potential of gas carburizing atmosphere. Source: Ref 9
Fig. 6
(b)
Distribution of residual stress produced in carburized plate and cylinder rod. Steel: JIS SCM
420 (Jominy hardenability 12 in. 26.5 HRC), Test piece
size: (a) 20 by 10 by 200 mm, (b) 14 mm diam by 200 mm.
Carburizing process: 920 C for 5 h, quenched from 880
C in JIS 2-1 hot oil (135 C). S, still hot oil; O, mild agitation; H, fast agitation.
Fig. 7

erates this difference. It is important to prevent
slow cooling at tooth foot portion to reduce the
introduction of surface soft layer and tensile residual stresses.
reported comparing different types of steels and

heat treatment methods. However, the chemical
contents and related properties of steels, even in
one type or grade, has considerable influence on
carburized-and-hardened steels as explained in
this section.
Influence of Steel Grade and

Hardenability on Residual Stress
Influence of Steel Grade
Case depth, core hardness, residual stress, and

distortion vary depending on chemical composition of steels, hardenability, and processing
variables even if parts are heat treated in the
same manner. Those characteristics have profound effects on strength and durability of casehardened components. Much research has been
The steel grade, in relation to its hardenability,

directly and indirectly influences surface and
core hardness and hardness profile through the
transformation characteristics of the steel. Along
with the steel grade as investigated in the past,
the hardenability of steels such as Jominy hardenability also has profound influence on the
Table 1 Surface residual stress and total area in compressive stress introduced in carburized
plate and rod
JIS SCM 420, Jominy hardenability (12 in.) 26.5 HRC, carburized at 920 C, in continuous carburizing furnace and
quenched in hot oil (135 C)
Agitation
Stress/area
Still oil
Slow agitation
Fast agitation
Plate
Surface residual stress, kg/mm2
Compressive area, %
32, 33
100
21, 32
113.6
36, 45
133.6
Cylinder rod
Compressive area, %
32, 28
100
24, 17
102.4
28, 22
106.8
Table 2
Residual stress measured at the tooth foot surface of carburized gears
JIS SCr 420, carburized in continuous gas-carburizing furnace and tempered at 160 C for 2 h
Residual stress of sample, kg/mm2
Heat treatment
A-1
A-2
A-3
B-1
B-2
B-3
33.8
2.8
42.7
2.4
31.3
1.6
40.0
5.6
32.7
8.8
32.8
4.4
Carburized and oil quenched

Average
30.0
Deviation
3.1
15.4
2.3
12.4
1.4
0.2
5.0
19.5
17.2
27.1
5.4
Induction hardened
Average
Deviation
Measured at three positions of the foot of one gear. Average stress value r (in kg/mm2). A and B, heat treated in different lots. 13 show different
portions.
900
z = 200 MPa
100
0
100
0
0
0
600
300
300
300
0
300
600
300
300
900
600
600
0
0
900
300
0
300
300
600
900
t=3s
Fig. 8
30 s
60 s
600
Estimated axial stress distribution in carburized gear (SNC815) during quenching. Source: Ref 12
hardness pattern resulting in fluctuation of residual stress profile and distortion. Generally, the
higher the hardenability of steel, the larger the
distortion. On the other hand, the surface residual stress and residual stress profiles produced
by carburizing-and-hardening processes are
complicated, and there seems to be no simple
proportional relation between hardenability and
residual stresses.
Residual stress distribution of two different
grade steels with same alloying elements are
compared in Fig. 9 (Ref 2), which shows a
higher-carbon smaller-diameter AISI 8620 sample with higher surface residual stresses compared with the other two larger AISI 8617 samples. This result indicates the influence of core
hardness or strength. Surface residual stresses of
surface carbon content of 0.95 % C are lowest
compared with the other two samples. In particular, the surface residual stress of deepest case
depth is the lowest compared with thinner case
depth, which seems to be caused by a larger volume percentage of retained austenite.
Areas in compressive stress increase as case
depth increases; however, the residual stress profile of 12.7 mm diam 8620 specimen is a little
different from that of the other two 19.05 mm
diam 8617 samples, where the area in compressive stress increases with increasing case depth.
The differences of three residual distributions indicates the influence of core hardness, case
depth, and specimen size.
Figure 10 shows the influence of steel grade
on residual stress (Ref 3). When surface compressive stress of higher hardenability AISI 8640
is lower than that of AISI 8620, the depth in
compressive residual stresses is deeper than that
of 8620. Similar to Fig. 9 the difference of residual stress distribution in Fig. 10 suggests the
influence of steel hardenability on residual
stress.
The impact strength of carburized-andhardened steels simply decreases as the case
depth and core hardness increase. However, fatigue strength of carburized-and-hardened steels
is influenced by hardenability of steel in relation
with the hardness profile and microstructure produced during the carburizing-and-hardening
operation. Influence of steel hardenability on residual stress distribution is partly shown in Fig.
9 and 10.
Influence of Hardenability Variation of

a Steel Grade
Influence of Steel Hardenability on Core
Hardness and Case Depth. Steel hardenability
is controlled by keeping the content of alloying
elements within a specified limit. However, standard specifications of steels have a range, and
this affects the variation of resulting hardenability of steel. Thus it is quite difficult to produce
exactly the same impact and fatigue strength,
distortion, and residual stresses even under the
same heat treat conditions.
The variation of steel hardenability due to
variation of chemical content investigated on a

the core has a direct influence on the distribution
of residual stresses and distortion in relation with
transformation plasticity at higher temperature,
where the surrounding matrix is soft and under
large temperature difference between surface
and core.
Influence of core hardness (core strength) on
fatigue strength investigated using the rotary
bending fatigue test is shown in Fig. 12 (Ref 9).
Variation in endurance limit reaches maximum
when the core hardness of a beam specimen is
around 430 to 450 HV. In the lower core hardness zone, a large percentage of fracture starts
from the inner side of hardened case, called fisheye fracture. By increasing the core hardness or
case depth, fisheye fracture disappears and endurance limit increases. In the higher core hardness zone, endurance limit seems to decrease
when carburizing time exceeds 5 h in this fatigue
test condition. Besides the important contribution of core hardness, the influence of case depth
also has a profound influence on fatigue strength,
which is explained later in this article. The fatigue test results of rotating beam evaluated in
relation to core strength has a peak near 140 to
150 kg/mm2, which coincides with the fatigue
test results of module 3 gears as shown in Fig.
13 (Ref 1, 13).
Influence of Core Hardness on Residual
Stress. Steel hardenability has direct influence on
core hardness and indirect influence on surface
residual stress and distribution. However, little

difference is observed in surface hardness and
residual stress where the carbon concentration at
the surface layer of longer carburizing time tends
to increase retained austenite and to reduce hardness and compressive residual stresses. High
hardness case produced by the carburizing-andquenching operation introduces compressive residual stresses in the surface layer, which increase as case depth increases; however, the
highest residual stresses are in the same level.
Influence of Carburizing Processes

Influence of Carburizing and
Quenching
The profiles of residual stresses and maximum
peak values vary depending on miscellaneous
20
10
A
Table 3 Chemical composition and hardenability of five lots of chromium-molybdenum

steel JIS SCM 415, 420H
Lot
A
B
C
D
E
Si
Mn
Cr
Mo
(12 in.), HRC
0.13
0.17
0.21
0.23
0.23
0.26
0.28
0.20
0.33
0.35
0.80
0.75
0.80
0.79
0.82
0.013
0.014
0.016
0.014
0.013
0.011
0.011
0.013
0.009
0.014
1.03
1.04
1.11
1.06
1.11
0.21
0.20
0.19
0.19
0.28
20.0
25.7
29.1
34.8
40.1
30
20
30
30
40
60
0
0.04 0.08
10
Fig. 9
30
10
20
30
30
40
60
0
0.12

(a)
0.04 0.08
3
10
0.12
10
20
30
30
40
60
0

(b)
3.0
Residual stresses produced in gas-carburized

and oil-quenched 6.5 mm diam rods made
from AISI 8620 (curve A) and AISI 8640 (curve B) steel.
Source: Ref 3
10
2.0
1.0
Fig. 10
10
30
20
40
Source: Ref 6
10
30
Jominy
hardenability
Composition, %
0.04 0.08
JIS SCM 420 chromium-molybdenum casehardening steel grade of different melting lot is
shown in Table 3. Case depth and core hardness
of higher-hardenability steel increase even when
they are carburized and hardened in the same
conditions, as shown in Fig. 11. However, the
surface hardness is not influenced by hardenability of steel produced under a controlled carburizing atmosphere (Ref 6).
Influence of Steel Hardenability on Fatigue Limit and Residual Stresses. Generally,
the residual stress at the surface layer of
carburized-and-hardened steel is compressive
due to the larger expansion caused by transformation to martensite at the near-surface highercarbon layer that balances out the already transformed subsurface lower-carbon concentration
region. Even though the amount of expansion by
transformation is reduced to some extent due to
transformation plasticity, still the carburized
case introduces compressive stresses at the surface layer and tensile stresses at the inner side of
the high-carbon hard case.
The compressive residual stresses introduced
in the outer carburized layer are caused by the
timing of transformation at the last part of
quenching balancing to strength of core. The
core strength is related to the steel hardenability,
and the timing of transformation to low-carbon
martensite proceeds earlier than that of highcarbon case. This preliminary transformation of
0.12

(c)
Influence of steel grade and case depth on residual stress profile. (a) 12.7 mm ( 2 in.) diam; AISI 8620; surface 0.80% C; ECD 0.76, 1.27, 2.54 mm. (b) 19.05 mm (34 in.)
diam; AISI 8617; surface 0.95% C; ECD 0.76, 1.40, 2.29 mm. (c) 19.05 mm (34 in.) diam; AISI 8617; surface 0.70% C; ECD 0.76, 1.52, 2.54 mm. Source: Ref 2
1
70
Carburizing time, h
2.5
5.0
8.0
20.0
150
200
60
100
Core strength, kg/mm2
800
Fig. 12 Relation between core hardness and endurance limit of 9 mm diam rotating beam fatigue
test. JIS SCM 420, carburized in muffled gas-carburizing
furnace at 920 C for selected duration and quenched directly in JIS 1-1-1 cold oil (4560 C), and tempered at 160
C for 2 h. Source: Ref 9
700
500

0
20
Endurance limit, kg/mm2
80
The control of residual stress profile is not

easy even if the carburizing-and-hardening
operation is the same and is influenced by steel
lots. Difficulty of control of residual stresses is
shown by two gears processed to same specifications, but from different but acceptable heats
of 43BV14 steels in Fig. 16. Residual stress distribution of lot C is much lower, and surface residual stress is in tension if shot peening is not
provided. Residual stress distribution of lot D is
much better even before shot peening. Residual
stresses at the surface are increased by postquench shot peening to the same level and depth
of about 0.13 to 0.25 mm (0.0050.01 in.) and
resulted from work hardening and stressing by
the impact of blasted shots.
Figure 17 shows the two extremes in residual
stress distribution found in 40 production gears
of various specifications and steels. Residual
stress distribution of A (large gear of approximately 3 pitch AISI 4320 reheat quenched and
shot peened) is very poor, and residual stress at
the surface layer is tensile before shot peening.
Residual stress of B (medium gear of approximately 10 pitch marquenched and not peened) is
highly compressive. The carburizing process
specification including carburizing and quenching should be well standardized so as not to produce type A distribution.
10
0
10
20
20
40
30
60
40
50
80
60
1.2
70
400
2.0
1.0
20
25
30
35
40
Hardenability (Jominy 1 / 2 in.), HRC
Fig. 11 Influence of steel hardenability on effective

case depth, core hardness, and surface hardness of carburized-and-hardened gears. Carburizing temperature at 920 C with carburizing atmosphere of 0.9% C
potential for 2.5, 5, 8, and 20 h, quenched in JIS-1-1 cold
oil, held at 4060 C with agitation and tempered at 160
C for 2 h. ECD: A cycle, 1.4 mm; B cycle, 1.2 mm; C cycle,
1.0 mm. Source: Ref 6
1.0
60
Carbon ccontent, %
A cycle
B cycle
C cycle
300
sidual stresses of the pack-carburized specimens.

The influence of steel grade and quench methods on three-dimensional residual stresses are
compared in Fig. 15. Three quench media10%
saltwater, water, and oilare compared on a
pack-carburized specimen. Surface carbon concentration is about 1.0% C and produced about
10 to 16% retained austenite in an oil-quenched
specimen. However, the longer carburizing time
seems to introduce a greater percentage of retained austenite and reduces hardness.
The longitudinal and tangential residual stress
near the surface differs where a saltwater quench
produces highest residual stresses, and an oil
quench seems to produce the lowest residual
stress. Saltwater quench for AISI 1020 (JIS
S20C) and water quench for AISI 5120 (JIS
SCr420) can produce higher surface residual
stresses compared with that of oil quench, as
shown in Fig. 15 (Ref 14).
Fatigue bending strength, kg/mm2
Case depth, mm
Core hardness, HV
Surface hardness, HV
processing factors and steel properties. Even

when a certain number of parts are heat treated
in the same condition or batch, it is quite difficult
to get the same residual stress magnitude and
profile, for it is almost impossible to completely
control every factor of carburizing and quenching in the same manner.
Figure 14 compares the distribution of residual stresses introduced in AISI 8620 steel by different carburizing-and-quenching methods. A
pack-carburized testpiece has very high carbon
content (1.2% C) compared with that of gas carburizing (0.8% C) and salt-bath carburizing
(0.5% C). Residual stresses produced in packcarburized specimens differ depending on
quench methods. Where water quenching produces deeper and larger compressive residual
stresses compared with that of oil-quenched
specimen, even the carbon gradient is the same.
Residual stresses near the surface of both packcarburized specimens drop sharply because of a
high-carbon layer with a large volume of retained austenite. The difference in carburizing
methods such as liquid and gas methods produce
different carbon distribution, and the resulting
distributions of residual stresses differ as observed in Fig. 14 (Ref 2). Salt-bath carburizing
has a very thin case and low carbon gradient;
however, the case hardenability of salt-bath carburizing is improved by nitrogen diffused during
the carburizing process and then the near-surface
residual stresses drop. Also, residual stresses of
gas-carburized specimens have a mild drop of
near-surface residual stresses compared to the re-
40
0.8
0.6
0.4
0.2
0
20
50
100
150
200
0.2
50
Core strength, kg/mm2

Influence of core strength on fatigue strength of
carburized-and-hardened gears. Fatigue limit
of gears of different core strength. Nonalloyed and alloyed
case-hardened steels. Modulus 3. Source: Ref 1, 13
Fig. 13
Pack carburized, water quenched

Pack carburized, oil quenched
Gas carburized, oil quenched
Liquid carburized, oil quenched
0.02 0.04 0.06 0.08 0.10 0.12

Comparison of residual stresses produced in

AISI 8620 steel 38.1 mm diam rods by different
carburizing-and-quenching methods. Pack-carburized and
reheat quenched in water or oil, and gas or liquid carburized and oil quenched. Source: Ref 2
Fig. 14

Water quenched SCr 420
40
D = 0.3 mm
80
60
40
20
Sectional area, mm2
10% Salt water quenched 1020
40
40
D = 0.3 mm
80
60
40
20
40
D = 0.3 mm
40
80
80
40
80
Influence of Carbon Potential During

Carburizing
Oil quenched SCr 420
40
80
60
40
Sectional area, mm2
20
Water quenched AISI 1020
40
D = 0.3 mm
80
80
Sectional area, mm2
60
40
20
Sectional area, mm2
Residual stress distribution produced in AISI 1020 (JIS S10C) and AISI 5120 (JIS SCr 420) test specimen.
Specimen: 11.3 mm diam, 80 mm long. Pack-carburized (surface carbon: 1.0% C) and furnace cooled.
Reheated up to 880 C and quenched in 10% saltwater (20 C), water (20 C), and 40 C oil. Case depth D is distance
from surface to the point of 0.6% C. Source: Ref 14
Fig. 15

0
0.5
1.0
1.5
2.0
2.5
3.0
30
20
10
20
20
30
D
40
10
C
10
Compression
Tension
10
10
30
50
40
60
70
0
0.010
0.030
0.050
0.070
0.090
0.110
0.130

Two gears processed to the same specifications made from different but acceptable heats of steels. C and D:
Different heats of 43BV14 steel gears processed in the same schedule. Open circles, total case depth; closed
circles, effective case depth (50 HRC). Source: Ref 15
Fig. 16
Precise control of carburizing cycle is very

important for producing quality components
with optimal distribution of carbon and residual
stress and for minimizing carburizing time by the
boost-and-diffuse control method (Ref 16).
Carbon Potential Control of Carburizing
Processes. Usually, the carburizing process is
controlled by carbon potential by measurement
of dew point, methane (CH4), carbon dioxide
(CO2), or oxygen sensor, and various other gas
elements contained in carburizing atmosphere in
recent gas-carburizing processes. Compared
with pack-carburizing and salt-bath-carburizing
processes, gas-carburizing processes are more
stable by the application of such gas analysis and
control methods. However, throughout the carburizing operation the carbon potential of the
used gas atmosphere varies depending on carrier
gas carbon potential, enriching gas types and
flow volume, total surface area of works, and
furnace design and construction materials.
The specificationes usually designated in part
drawings are case depth, surface hardness, and
core hardness. There are various control methods
to get a designated case depth, for example, the
selection of steel grades, carbon potential in a
time frame, and quench media. Vacuum or
reduced-pressure carburizing methods are becoming popular, but have difficulty in carbon potential control due to the lack of an appropriate
carbon sensor. Therefore, the establishment of
carburizing control methods has been dependent
on trial and error. The boost-and-diffuse carburizing method is popular, but the development of
appropriate control measures for vacuum and reduced pressure carburizing is requested.
Alloying elements such as nickel and carbon
concentration above the eutectic percentage increases retained austenite in carburized case.
Therefore, carbon potential during carburizing
should be carefully controlled to prevent excessive retained austenite in carburized gears.
Figure 18 shows the influence of carbon content on hardness distribution by two different
carbon potential atmospheres. The hardness and
carbon distribution treated in higher 0.9% C potential has lower hardness due to a larger amount
of retained austenite and deeper carbon diffusion, while that of 0.7% C potential hardness is
higher in surface layer to the depth of 1 mm but
carbon distribution is lower than in the former
condition. Careful carbon potential control is
very important for deep-case treatment to prevent excessive retained austenite, and optimal
control of diffusion process is the key to fabricating quality products.
Relation between Carburizing Time, Carbon Potential, and Residual Stress. Carbon potential control is important not only for hardness
and residual stresses, but also for reducing treating time. Boost-and-diffuse carburizing methods
have been widely applied since the late 1950s
(Ref 16). The influence of carbon potential con-
Carbon content, %
800
800
700
A SCM 420, 9 mm
600
600
500
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
10
700
Hardness, HV(5)
1.7 kg/mm2. Similar results were reported when

quench oil temperature was raised from 50 to
270 C, and the most significant increase is attained above 200 C (Ref 18). However, influence of quenching media such as hot or cold oil,
salt-bath, and water-base polymer solution is not
simply defined. Not only the types of quench
media but their viscosity, heat transfer coefficient, Leidenfrost point, and flow speed change
corresponding to the solution temperature.
Influence of Tempering on Residual
Stresses. Carburized-and-hardened components
are usually tempered at low temperatures (130
190 C) for about 1.5 to 2 h to prevent the chipping or cracking often observed in grinding
400
B
A
S15CK,
25 mm
300
0
C
0
trol on carbon distribution and their relation with

residual stresses is shown in Fig. 19. In spite of
large differences in carbon potential during carburizing and carburizing time where that of
shortest method is less than half of the longest,
the case depth and distribution of compressive
residual stresses differ only slightly. The variation of residual stress is within 10 kg/mm2, and
the total depth of compressive layer is similar.
Additionally, the endurance limit of shortest cycle A is 6 to 7% higher than that of B and C. It
means optimizing the carburizing method by
controlling carbon potential is very important
not only for the improvement of durability, but
is also effective for improving processing efficiency and manufacturing cost.
The profiles of longitudinal residual stresses
produced in carburized-and-hardened steel differ
from each other depending on case depth, as
shown in Fig. 20. The surface compressive residual stress of ECD 0.8 mm is highest and decreases as case depth increases, while the compressive residual stress zone expands inward as
case depth increases.
Influence of Quenching Oil on Residual
Stress. Quench methods influence the residual
stress state. Surface residual stress produced in
9 mm diam rotating beam bending fatigue specimens quenched in hot and cold oils are compared. The surface compressive residual stresses
in the carburized fatigue specimen made of JIS
SCM 420 increase as the quench oil temperature
is raised from 30 to 135 C. Surface residual
stresses of specimens carburized and quenched
in 30 C oil was 39.7 0.7 kg/mm2 and that of
specimens quenched in 130 C oil was 46.0
Hardness, HV(5)
10
20
Carbon potential
0.9% C
0.7% C
30
40
1.0
2.0
3.0
10
20
SCM 420, 9 mm
30
B
40
0.4
4.0
0.8
1.0
1.4
1.8

Influence of carbon potential on longitudinal
residual stress distribution produced in AISI
8617 19 mm diam specimen carburized and quenched directly in oil. Source: Ref 3
Fig. 18
Surface residual stress of test specimen in relation to effective case depth. Material: JIS SCM
420. 0.9 mm diam rotating beam fatigue testpieces. Carburizing condition in three different carbon potential histories: A, 1.20.9% C/6 h; B, 0.9% C/constant 8 h; C, 0.6
0.9% C/13 h. Source: Ref 17
Fig. 19

0
0.5
1.0
1.5
2.0
2.5
30
3.0
10
10
Tension
20
20
30
40
30
B
50
40
60
70
0
0.010
0.030
0.050
0.070
0.090
0.110
0.130
10
10
10
Compression
20
A
0
10
20
30
0.8 mm
1.3 mm
2.0 mm
40
50

Extremes in residual stress distributions found among 40 production gears of various modules and heats made
of the same steel. Open circles, total case depth; closed circles, effective case depth (50 HRC). A, large gear
(approximately 3 pitch) AISI 4320 reheat, quenched and shot peened; B, medium gear (approximately 10 pitch) marquenched, no peening. Source: Ref 15
Fig. 17
1.0
2.0
3.0

Influence of case depth on residual stress distribution produced in 6.5 mm diam cylinder
rod made from AISI 8620 carburized and directly
quenched in oil. Source: Ref 3
Fig. 20

operations. Tempering after quenching at 160 C
reduces residual stresses about 1 to 2 kg/mm2 and
reduces the volume of retained austenite. Reduction of surface residual stresses measured at each
of the surfaces of 12 by 35 by 80 mm plates and
20 mm diam rods are shown in Table 4.
Tempering temperature has considerable influence on the decrease of residual stress and the
increase of temperature resulting in a larger decrease of residual stress. Tempering after
quenching at 165 C reduced a compressive residual stress produced in AISI 4130 and 1526
about 50 to 60 MPa in both steels (Ref 18). However, the influence of temperature depends on
tempering methods such as type and construction of furnace corresponding to the heat capacity, temperature control system, and measuring
methods of holding time, which is explained in
a later section.
Influence of Carburizing Time and

Case Depth on Fatigue Strength
The influence of case depth on fatigue properties has been investigated for about a half cenTable 4
tury, and various conclusions were drawn. However, Fig. 21 shows a clear relationship between
ECD and endurance limit investigated by 9.00
mm diam rotating beam bending fatigue tests.
The approximate value of the optimal ECD for
the 9 mm diam specimen is 1 to 1.2 mm. Endurance limits of shallow ECD and low core
hardness specimen are low due to subsurface
fracture and by increasing case depth or core
hardness fatigue strength. Even though the carburized case is strengthened, the core to case
boundary often originates a fracture, which propagates outward leading to a breakage called fisheye fracture. By increasing case depth, the distribution of materials strength inside the
specimen increases, and all fracture starts from
the surface. The strength of the surface becomes
the key factor, and damage such as grainboundary oxidation, nonmartensite phases at the
surface, and retained austenite reduce the endurance limit. By increasing case depth or core
hardness, fatigue strength can be increased.
However, the peak ECD becomes deeper as the
core hardness increases, and there seems to be a
possibility of obtaining higher endurance limit
Surface compressive residual stresses in relation to effective case depth
JIS SCM 420, Jominy hardenability (12 in.) 25.0. Plate 13 by 35 by 80 mm long and rod 20 mm diam by 80 mm long.
Gas-carburizing temperature: 920 C with carburizing atmosphere of 0.9% C potential for 2.5, 5, 8, and 20 h, quenched
from 860 C in JIS-1-1 cold oil held at 3550 C with agitation and tempered at 160 C for 2h
Residual compressive stress, kg/mm2
Average
Effective case depth, mm
0.2
0.5
1.0
2.0
Average
Plate
Rod
As carburized
Tempered
As carburized
Tempered
As carburized
Tempered
30.3
28.9
34.7
29.8
30.9
27.0
30.9
30.3
30.0
29.6
30.832.0
27.829.5
34.537.5
30.033.3
32.0
23.929.5
27.033.5
31.034.3
29.533.5
30.2
26.532.8
27.833.0
32.335.5
20.833.0
29.8
26.329.0
29.033.0
28.232.3
27.331.8
29.1
Jominy hardenability (1/ 2 in.)

20.0 HRC
29.1 HRC
25.7 HRC
40.1 HRC
Endurance limit, kg/mm2
80
10
20
B
C
30
D
40
F
50
0.5
1.0
1.5
2.0
2.5

60
0.5
1.0
1.5
2.0
2.5
3.0

Influence of effective case depth on endurance
limit. Material: JIS SCM 420 9 mm diam fatigue
test specimen carburized at 920 C, quenched in cold oil
and tempered at 160 C. Source: Ref 6
Fig. 21
Relation between effective case depth and surface residual stresses produced in carburizedand-hardened steel. A: 10 by 10 mm, Jominy hardenability
1
1
( 2 in.) (J- 2) 14 HRC; B: 10 by 10 mm, J-12 22 HRC;
C: 20 mm diam, J-12 25 HRC; D: 10 by 35 mm, J-12
25 HRC; E: 10 by 10 mm, J-12 20 HRC; F: 10 by 10 mm,
J-12 35 HRC. Source: Ref 6
Fig. 22
Influence of Case Depth on Residual

Stress
The goal of the carburizing process is to produce high hardness case and also give high compressive residual stresses in the hardened case.
The influence of case depth on fatigue strength
is also related to the influence on residual stress.
Figure 22 shows the influence of ECD on surface
residual stresses, indicating the peak of the residual stresses is around an ECD of 1.0 to 1.5
mm, where the highest endurance limit exists.
The figure also shows another important point
that the optimal ECD and residual stresses are
influenced by core hardness, and by increasing
core hardness the surface compressive residual
stresses increase higher than that of lower core
hardness peak. The region over the peak of case
depth influence turns negative because of grainboundary oxidation and retained austenite.
Fatigue durability of testpieces or parts with
stress concentration becomes a little different
from that of unnotched specimens. It is related
to the balance of applied stress and distribution
of materials strength leading to the shift of optimal case depth. Carburized components such
as gears have stress concentration at the tooth
foot portion, and the relation between ECD and
endurance limit differs from the results observed
in Fig. 22. However, the basic factors related to
the influence of case depth, residual stress, and
grain-boundary oxidation play the key role as
explained in a later section.
Influence of Microstructure on
Residual Stress and Fatigue Strength
Influence of Retained Austenite on
Residual Stress and Strength of
Carburized Steels
70
by a combination with higher-hardenability

steels.
The distribution of carbon has an important

influence on fatigue strength and the profile of
residual stresses and is controlled by the carburizing condition. In particular, the variation of
carbon potential in relation to operating temperature is the key to designed strength and durability. Microstructural changes introduced by
carburizing and quenching vary depending on
carbon gradient and quench severity, in relation
to alloy elements. The carbon content of the surface layer decides the percentage of high-carbon
martensite and retained austenite changes and affects wear properties and strength of carburized
products.
Retained Austenite in Carburized-andHardened Case. Generally, carburized-andhardened case contains 20 to 35% retained austenite phase that has considerable influence on
the fatigue properties of carburized steels.
Retained austenite at the surface layer due to
high carbon content reduces compressive residual stresses as shown in Fig. 23. Therefore, car-
Influence of Tempering and Cryogenic

Treatment on Retained Austenite
Carbon content
(estimated)
10
25
Residual
stress
10
20
Retained
austenite
15
20
10
30
40
50
60
10
20
30
40
50
0
10
20
10
Retained austenite, vol.%
30
30

(a)
0.5
1.0
1.5
0
60
20
SAE 1118
10
Residual
stress
40
10
30
Carbon content
20
10
20
30
10
Retained
austenite
20
10
Retained austenite, vol.%
50
10
20
30
40
50
40
60
30

(b)
Relation among carbon gradient, retained austenite, and residual stress produced in
carburized-and-hardened steels. (a) AISI 8620. (b) AISI
1118. Source: Ref 4
Fig. 23
Low-Temperature Tempering and Retained Austenite. Usually, carburized parts are

tempered at about 150 to 190 C to prevent problems such as chipping or cracking in the finish
machining process. Tempering at this temperature range after carburizing and hardening reduces hardness, residual stresses, and retained
austenite depending on temperature and holding
time. Figure 25 shows the decrease of residual
stress by tempering at each temperature. Retained austenite also decreases by tempering at
higher temperatures.
Cryogenic (Subzero) Treatment. Retained
austenite transforms to martensite by subzero or
cryogenic treatment, cooling down to deep freezing temperature and increasing martensite volume and residual stresses. Retained austenite
volume at the surface is lower than that of subsurface results from nonmartensite anomaly layers. Austenite volume increases by polishing
away a few microns of the surface anomaly
layer. From the surface to a depth of 10 to 30
lm, the retained austenite volume decreases to
less than half. Although the retained austenite
volume is governed by the carburizing and
quenching conditions the values listed in Table
5 seem to be in the generally acceptable range.
The volume of retained austenite changes by
tempering and subzero treatment as shown in Table 5. The carburized-and-hardened and tempered conditions retained almost the same level
of austenite volume (10%). Cryogenic treatment reduces retained austenite, and the volume
decreases as treatment temperature is lowered.
Tempering after cryogenic treatment is important, for there is a possibility of introducing
microdefects caused by forced transformation by
very-low-temperature subzero treatment. Microextrusion is often observed at the grain
boundary, caused by transformed martensite and
Distance below surface, in.

40
0.010
0.020
0.030
0.040
0.050
Direct quench
Single reheat
Double reheat
35
30
SAE 8620
reducing fatigue life. Tempering after subzero

treatment has no observable effects on retained
austenite volume and hardness.
Influence of Rolling, Shotblasting, and
Stress Loading. Excessive retained austenite is
hazardous for fatigue strength and wear resistance. Retained austenite transforms to martensite by repeated loading such as rolling and improves load-carrying capacity under rolling
fatigue applications (Ref 20). For this purpose,
25
20
15
10
5
0
0
0.25
0.50
0.75
1.0
1.25
Distance below surface, mm

Profile of retained austenite in carburized-andhardened steel produced by different quench
methods. Retained austenite, measured by x-ray diffraction, as a function of distance from the surface of an AISI
8620 steel carburized at 925 C. The single and double
reheats were accomplished by heating to 845 and 790 C,
respectively. Source: Ref 19
Fig. 24

0.5
1.0
1.5
sensing methods is improving the reliability and

reproducibility of carburizing processes.
Figure 23 shows distribution of residual
stresses, carbon gradient, and retained austenite
of two carburized-and-hardened steels. The distance from surface to the peak of residual
stresses is influenced by the volume of retained
austenite. Increase of retained austenite decreases hardness and residual stresses by reduction of transformation to martensite that accompanies expansion and increases the compressive
residual stress (Ref 4). Retained austenite transforms to martensite or other phases by succeeding tempering, cold working, or stressing, which
is explained in a later section.
Figure 24 compares the distribution of retained austenite, measured by x-ray diffraction,
resulting from different quenching methods. Direct or single reheat quench after carburizing
produces more retained austenite, while double
reheat and quench operations can reduce the
amount of retained austenite to some extent.
40
Untempered
30
Tempered
20
0
0
50
100
150
200
250
Quenchant temperature, C
300
(a)
bon potential control during the carburizing process is very important to get quality products.
Alloy elements added to case-hardening steels
such as nickel, chromium, and manganese increase retained austenite; therefore, surface carbon content should be carefully controlled in carburizing alloyed steels containing nickel,
manganese, and chromium.
Compared with pack carburizing and salt-bath
carburizing, the recent technology in gascarburizing processes are much more stable, and
carbon potential control is easy via the use of
various sensors. However, periodic maintenance
of the oxygen sensor is very important to prevent
excess or shortage of carbon potential due to the
deterioration of any controlling sensors. Therefore, the recent trend toward the use of multi-
40
Untempered
30
Tempered
20
0
0
50
100
150
200
250
Quenchant temperature, C
300
(b)
Influence of quench oil temperature and tempering on residual stresses. Variation in surface
retained austenite content of untempered and tempered
carburized (a) AISI 4130 steel, and (b) AISI 1526 steel for
various quenchant temperatures. Source: Ref 18
Fig. 25

Table 5
Influence of tempering and subzero treatment on volume of retained austenite
JIS SCM 420, gas carburized at 920 C, quenched from 850 C in cold oil of 3550 C, tempered at 160 C for 2h
Retained austenite (a), vol%
Treatment
At carburized surface
At 10 lm, etched
At 30 lm, etched
14.7
16.5
10.6
6.7
21.8
18.5
18.6
11.0
9.15
10.7
11.3
6.7
Carburized-and-hardened surface
Tempered
Subzero, 25 C, treated
Subzero, 75 C, treated
(a) Reliability range: vol% 10%
40
30
Shot blasted
20
Shot peened
(45 HRC, 55 m/s)
Hard shot peened
(53 HRC, 90 m/s)
10
0
0
100
50
150

Change of retained austenite content by hard
shot peening investigated on carbonitrided
steel. 20 to 50 vol% of austenite retained at surface layer
of carbonitrided steel. Retained austenite volume decreases
by shot peening, depending on shot peening condition.
Hard shot peening is more effective for reducing retained
austenite than traditional shot peening. Source: Ref 21
Maximum residual stresses after shot peening, kg/mm2
Fig. 26
Influence of Grain-Boundary Oxidation

on Residual Stress and Fatigue Strength
120
Hard shot peened
(53 HRC, 90 m/s)
(57
Hard shot peened
(53 HRC, 90 m/s)
100
Hard shot peened

(53 HRC, 90 m/s)
(45
80
Traditional
shot peening
(45 HRC, 55 m/s)
60
Shot
Shotblasted
blast
40
Carburizing time, (t ), h
10
20
30
Volume change of retained austenite

by shot peening, %
Relation between volume of change of retained austenite and maximum residual stress
produced by shot peening of carbonitrided steel. Retained
austenite before shot peening was 32 vol% at the point of
measurement 25 lm from surface. Shot hardness and shot
speed are as indicated. Hardness of shot is indicated by
HRC values. Source: Ref 21
Fig. 27
Variations of surface hardness and volume of

retained austenite of carburized-and-hardened
steels are related to the steel type and grainboundary oxidation. The formation of nonmartensite phasessuch as bainite, troostite, or
pearlite phases at the grain boundary in the surface layerdirectly results in decrease of hardness and compressive residual stress.
Grain-boundary oxidation accompanying traditional carburizing processes is caused by reaction of the steel surface with the carburizing
atmosphere containing carbon monoxide. The
grain-boundary oxidation itself and the associated introduction of surface anomalies such as
nonmartensite phases are closely related, but the
basic phenomena are different from each other
(Ref 2224).
Grain-Boundary Oxidation. During carburizing in an atmosphere containing carbon monoxide, gas grain-boundary oxidation progresses
simultaneously with carburizing by metal gas reaction caused by dissociation of carbon mon-
30
Depth of anomaly, m
Retained austenite, vol %
As quenched
30 to 40% of retained austenite is selected in

special components used for rolling fatigue applications. Retained austenite transforms by mechanical stressing such as rolling and shot peening, causing stress-induced transformation into
martensite. The stress loading that accompanies
rolling fatigue or bend loading during mechanical operation also assists transformation of austenite.
The shot peening process is effective to increase hardness, reduce retained austenite, and
improve fatigue durability (Ref 11). Figure 26
shows the volume change of retained austenite
by shot peening using different hardness shots.
Retained austenite of the as-quenched condition
can be reduced by various shot peening methods,
and a great decrease is observed by hard peening.
Figure 27 shows the relation among hardness increment results from shot peening blasted on
various levels of retained austenite. It shows that
the hardness increase results from transformation of retained austenite to martensite accelerated by shot impact. The depth of hardness increase caused by shot peening reaches down to
nearly 200 lm by the hardest peening condition
(Ref 21).
oxide. Also, the amount of carbon monoxide has

a direct relation to the degree of grain-boundary
oxidation (Ref 25), and as carburizing time or
concentration of carbon monoxide increases, the
depth of grain-boundary oxidation increases.
Grain-boundary oxidation decreases the hardenability of near-surface grain boundary and reduces the volume of transformation into martensite and reduces surface hardness and
compressive residual stresses. Therefore, grainboundary oxidation affects the fatigue property
of carburized steel directly by reducing hardness
and compressive residual stress and indirectly
weakens grain-boundary strength to some extent.
The depth of oxidation increases as carburizing time increases, as shown in Fig. 28. However, the depth of grain-boundary oxidation varies by condition. Etched microstructure shows a
deeper anomaly than that of an unetched microstructure. Diffusion depth of oxygen investigated by electron probe microanalysis is much
deeper than that observed by optical microscopy.
Formation of Nonmartensite Phases. Depth
of grain-boundary oxidation measured on nonetched and etched microstructures differs as shown
in Fig. 29. Alloying elements such as chromium,
silicon, and manganese are oxidized rather easily
by the reaction with carbon monoxide in carburizing atmosphere (for example, Ref 2225).
Those alloy elements move to the grain boundary where oxygen is supplied from atmosphere
through the grain boundary and forms metallic
oxides. The depth of oxygen diffusion is far
deeper than that measured by optical microscopy
observing nonmartensite phases such as upper
bainite, troostite, or pearlite. The transformation
from austenite to nonmartensite phases is caused
16
25
20
10
SCM 420
cold oil
No etching
10% nital
SCr 420
cold oil
No etching
10% nital
SCr 420
hot oil
No etching
10% nital
0
0
t
Relation between carburizing time and depth
of grain-boundary oxidation. Material: JIS SCM
420, SCr 420 steels, gas carburized at 920 C in 0.9% C
atmosphere for various times. Quenched from 860 C in
hot and cold oil. Grain-boundary oxidation measured aspolished and etched with nital 10%. Source: Ref 26
Fig. 28

by reduced hardenability resulting from a decrease in alloy elements from austenite grains.
This phenomenon is accelerated by a poor cooling rate caused by delayed quenching operation,
prolonged vapor blanket stage during quench,
and poor quench power. The hardenability of
grain boundary of surface layer to obtain martensite can be recovered by the addition of other
alloying elements such as nickel, molybdenum,
boron, and nitrogen or by maintaining sufficient
carbon higher than eutectic content in austenite.
In addition to the influence of grain-boundary
oxidation, the case hardenability of the surface
layer is related to the surface carbon concentration and has considerable influence on the case
hardenability. However, the influence of grainboundary oxidation is overstated because of lack
of knowledge on the difference of oxidation itself and nonmartensite formation related to poor
cooling speed or quench operation.
In a comparison of the influence of quench
media, hot oil quench produces a more nonmartensitic structure than do cold oil quench and salt
quench. In the early vapor blanket stage in
quenching, the cooling speed is slow and the low
hardenability surface with reduced alloy elements tends to produce the nonmartensite phases
rather easily.
Influence of Nonmartensite Phases on Residual Stress. The compressive residual stress
produced in carburized-and-hardened case decreases if the surface layer has nonmartensite
phases or excessive retained austenite; the surface residual stresses are lost and turn to tension
in an extreme case. Therefore, grain-boundary
oxidation, the associated formation of nonmartensite, and excessive residual austenite result
from excessive high-carbon content should be
prevented by appropriate processing specification (Ref 28).
Removal of the nonmartensite layer can improve residual stress distribution and results in improvement of fatigue durability. Figure 28 shows
how to eliminate negative effects of the nonmartensite layer. It is clear that shot peening can easily
eliminate the influence of grain-boundary oxida-
tion and works better than removal of the layer

by chemical etching or grinding.
The influence of nonmartensite phases is easily eliminated by various methods as shown in
Fig. 30 and 31. However, shot peening and hard
peening processes directly shot on as a
carburized-and-hardened surface give the best
result, as observed in the figures. Recently, various gears have applied hard or double peening
to improve fatigue strength of carburized gears.
Optimization of shot peening conditions such as
size, shape, and hardness of shots, shot peening
machine, and its operating speed and combination with those of the double peening condition
is being investigated, and further progress is anticipated.
Benefit and Change of Compressive

Residual Stress after Heat
Treatment
The particular benefit of carburizing is the introduction of high hardness case and introduction of compressive residual stresses. The compressive residual stresses produced in the
hardened case counteract the applied stresses and
improve bending fatigue strength. As the necessity of improving the performance of carburized
parts emerges, much effort has been devoted to
developing quick measuring methods, modeling,
and simulation technologies to understand and
increase compressive residual stresses. The benefit of compressive residual stresses in addition
to the fatigue strength of the hardened surface
layer is that longitudinal compressive stresses increase the endurance limit about 10% of the residual stress (Fig. 32) (Ref 1, 30).
Changing Residual Stress by Aging and

Tempering
Aging. Residual stress decreases by aging and
by tempering, as internal stresses are released by
aging effects that precipitate over saturated carbon and other elements from martensite phases.
Aging effects proceed gradually and reduce residual stresses as shown in Fig. 33; the change
of residual stress distribution due to aging is different depending on steel grade. As observed in
Fig. 33(a), straight carbon steel loses residual
stress more easily than does nickel and chromium alloy steel (Fig. 33b).
Tempering. Similar to aging phenomena,
tempering also reduces residual stress as observed in Fig. 34 and 35. The reduction of residual stress resulting from tempering differs depending on factors such as steel grade and state
of carbon gradient in relation to microstructures
and tempering condition, as observed in Fig. 34
(Ref 3). The decrease in compressive stress influences distribution, and the stress peak disappears at tempering at temperatures higher than
200 C.
Changing Residual Stress Profile by

Repeated Stressing
Residual stress decreases during fatigue (repeated) loading. However, compressive residual
Limited time life strength, kg/mm2 106
J- 1/ 2 = 20 HRC
J- 1/ 2 = 40 HRC
A
B
C
D
90
100
110
120
J 1/ 2 , Jominy hardenability (1/ 2 in.)

Influence of posthardening finishing method
on rotating beam bending fatigue life. Surface:
0.9% C. 920 C for 8 h. Cold oil quench: 860 C. J-12,
Jominy hardenability (12 in.). A: as-carburized and
quenched; B: carburized, quenched, and shotblasted; C:
carburized, quenched, and chemically etched; D: carburized, quenched, etched, and shotblasted. Source: Ref 26
Fig. 30
(c)
(a)
(b)
Microstructures of grain-boundary anomalies. (a) AISI 8620 steel, gas carburized at 955 C. Specimen shows grain-boundary anomaly to a depth of approximately 0.02
mm. 1% nital. 750. (b) AISI 4118 steel, gas carburized at 955 C. Grain-boundary anomaly to a depth of approximately 0.02 mm. As polished. 750. (c) AISI 8822 steel,
gas carburized 15 h at 925 C, reheated to 840 C, and held 40 min, oil quenched, and tempered 2 h at 150 C. Bainite-pearlite mixture near the surface is visible because of the lighter
etching. 2% picral. 500. Source: Ref 27
Fig. 29
(a)
(b)
(+1.4%)
(c)
(+10%)
(d)
(+12%)
(+25%)
(+47%)
(e)
50
100
Rotate bending fatigue strength, kg/mm2

Comparison of various finishing methods to
eliminate problems caused by surface anomalies. (a) Basic process SCM 420 carburizing. Traditional
carburizing as the base data. (b) Carburizing and NH3 addition. Ammonium gas addition at the final stage of carburizing. (c) Vacuum carburizing. (d) Alloying. Experimental steel with Ni-0.2%, reduce Si to 12 of standard steel. Mo
increased. (e) Electrochemical etching and shot peening
compared by fatigue test. Source: Ref 29
Improvement of endurance limit, kg/mm2
Fig. 31
Fig. 32
2
1
0
30
40
20
Axial residual stresses, kg/mm2
stresses do not disappear completely, even after

fatigue failure. The process of stress decrease is
not simple, and reduction and partial increase is
observed until fatigue failure.
Changing Surface Residual Stress by Repeated Stressing. The change of surface residual stress investigated by rotating beam fatigue
test samples made of JIS SCM 420 is shown in
Fig. 36 (Ref 5). The surface compressive residual stresses decrease at first and then turn to increase at around 0.5 million cycles, then turn to
decrease again after a million cycles. The point
of fatigue failure does not correspond to the upand-down trend of residual stresses (Ref 5).
The process of reducing surface residual stress
differs by testpieces as decrease, increase and decrease again until final failure. The data indicated
in Fig. 37 show amplitude of repeated stress. Occasionally, some fatigue failure occurs during
the increase stage of surface residual stress that
results from partial irregular distribution of the
residual stress measured at the portion near the
fracture face.
Changing Residual Stress Distribution by
Repeated Stressing. The decrease of residual
stresses is not only a phenomenon that occurs at
the surface, but the decrease of compressive residual stress is also observed as the change of
residual stress distribution as shown in Fig. 38
and 39. Figure 38 shows the process of reducing
residual stresses at several layers in a fatigue
testpiece made of nickel-chromium casehardening steel (SNC21: 0.13C, 2.16Ni, 0.38Cr).
Near the surface, the change or decrease of residual stresses is clear, but change at the inner
core is not so large and is rather simple compared
with the surface changes (Ref 31). The progress
of relaxation of compressive residual stress by
repeated stressing is clear in residual stress distribution as shown in Fig. 39. The decrease of
area in compressive stress by fatigue is compared with the residual stress distribution pro-
duced in the same lot of carburizing and hardening as shown in Fig. 6.

The change in the surface residual stresses
does not directly correspond to the change in distribution of residual stresses shown in Fig. 37.
The change of residual stress distribution is not
simple, and their wavy distribution patterns have
no characteristic tendency. The general trend in
decrease of residual stresses is in Table 6, where
the total area in compressive stresses decreases
as stress cycle increases, even when the cyclic
stress level is lower than that of high-stress
short-life test condition (Ref 5). The reduction
of area in compressive stress is compared with
that of average distribution of three carburizedand-hardened samples shown in Fig. 6.
Influence of Residual Stress and

Grain-Boundary Oxidation on Gear
Strength
Case depth influences the endurance limit and
fatigue lifetime of gear teeth as shown in Fig.
40. When the endurance limit exhibits double
peak, the first peak exists at very thin case depth
and second peak at about 1.0 mm ECD. This
tendency is different from unnotched rotating
beam fatigue test results; however, the case
depth of the second peak is similar to that of
rotating beam bending fatigue test as shown in
Fig. 21. The same tendency is observed in impact bending fatigue test results of carburized
notched square bar (Ref 25).
Carburized-and-hardened case gives high
hardness, strength, and compressive residual
stress and contributes to the improvement of fatigue durability, while the static and impact
strength deteriorate. The compressive residual
stresses possess an independent feature that has
the highest value when certain case depth is obtained at the surface layer, as shown in Fig. 22.
Improvement in bending fatigue life by surface

residual stresses. Source: Ref 30
10
10
20
1 day
3 days
1 week
15 days
23 days
1 month
after quenching
30
40
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Fig. 33
20
1 day
1 week
15 days
23 days
1 month
after quenching
30
40
0.8

(a)
10
0.1
0.2
0.3
0.4
0.5
0.6
10
0
0
10
20
40
0
Decrease of residual stress distribution due to aging. (a) AISI 1015 (S15CK) carbon steel. (b) AISI 8620 (JISSNC21) nickel-chromium steel. Source: Ref 31
1.0
2.0
3.0

(b)
As quench
150 C for 1 h
190 C for 1 h
30
Influence of tempering on longitudinal residual

stress of gas-carburized AISI 8617 steel. Gas
carburized and directly oil quenched AISI 8617 19 mm
cylinder rod. Source: Ref 3
Fig. 34
Influence of Residual Stresses

Produced by Hard Shot Peening
Table 6 Decrease of area in compressive

residual stress by fatigue test in relation to
nominal stress and repeated cycles
General strengthening effects of compressive

residual stress are well known, as shown in Fig.
32. The important influence of residual stresses
have been reevaluated and investigated intensively (Ref 21, 29, 3338). Figure 41 shows the
relation between compressive residual stress introduced by intensive peening methods and fatigue strength investigated via control of compressive residual stress profile and simulation of
their effects on fatigue strength (Ref 35).
Testpiece
No.
Nominal stress,
kg/mm2
Average of 3
1
2
3
Stress
cycles, N
Area in
compression, %
Before fatigue test

86.6
2.6 106
78.2
4.6 106
74.1
1.0 107
100
4.8
15.1
21.2
Source: Ref 5
Posthardening Shotblast Finish

In mass production, transmission gears are
blast cleaned to eliminate adhered soot and/or
color resulting from the dissociation of quench
oils. Shotblasting and peening increase compressive residual stresses. However, general
shotblasting equipment cannot introduce the uniform arc height and coverage necessary to improve surface fatigue durability.
By appropriate control of blasting condition
to attain uniform coverage and arc height, shotblasting can increase fatigue durability of production gears by increasing surface compressive
residual stresses, although the durability is far
less than that attained by shot peening and hard
peening. The details of shot peening and hard
peening operations are explained later in this article. For example, compressive residual stresses
of production gears that were 0.1 6 kg/mm2
can be improved up to 30 kg/mm2. If the condition is controlled in an optimal level, it can
increase surface residual stresses up to 48 5
kg/mm2 in compression even by traditional shotblasting machines. It is very important to notice
40
The depth of grain-boundary oxidation increases

as carburizing time is elongated, as shown in Fig.
28, and their influence on fatigue strength increases too. These two main factors combine and
result in a complicated influence on fatigue properties. Apparently, although grain-boundary oxidation has deleterious effects, the importance of
residual stress is more profound and should be
taken into account when designing gear durability. Figure 40 shows the influence of residual
stress and grain-boundary oxidation on gear
tooth bending fatigue strength.
The bending fatigue strength of a carburizedand-hardened gear tooth depends on the fatigue
strength of gear tooth root where applied stresses
concentrate. Fatigue strength of notched specimen or mechanical components has the highest
peak at thin case depth zone as observed in Fig.
40. On the other hand, tooth contact areas have
rolling fatigue strength in relation to contact
pressure and Hertz shear stress, which means the
tooth pitch area needs a thick case depth to withstand high rolling loads. Therefore, the design
specification of ECD for gear teeth is not as simple as the optimal case depth for flat specimens.
Specification of ECD should take the operation
condition into account to satisfy application
needs to guarantee total gear life against rolling
fatigue and bending fatigue life. In addition to
the fatigue strength given by carburizing and
hardening, the influence of the postheat treatment process is increasing in importance. A part
of postheat carburizing-and-hardening treatment, such as shotblasting, is explained earlier
in this article; however, the recent hard peening
technology has made great advancement and
contributes to the improvement of load-carrying
capacity of carburized-and-hardened gears. The
following section explains the influence of recent shot peening technology.
77.7
78.8
30
68.7
20
73.3
86.3
x
10
0
105
106
Number of cycles, N
107
Changes in surface residual stresses during fatigue. Gas carburized at 920 C for 5 h (18 ks)
in 0.8% C atmosphere, quenched from 860 C in cold oil
of 35 C , and tempered at 160 C for 2 h (7200 s). x,
fracture point. Source: Ref 5
Fig. 36
10
0
40
20
0
D = 0.3 mm
As quenched
20
40
Axial stress
Tangential stress
Radical stress
80
40
20
0
20
60
40
D = 0.3 mm
250 C
C tempered
tempered
60
80
60
40
Sectional area, mm
20
As quenched
After 2.15 104 cycles
30
0
20
D = 0.3 mm
300 C
C tempered
tempered
60
20
80
100
80
60
40
40
20
0.1
0.2
0.3
0.4
0.5
0.6
Sectional area, mm
Change of residual stress distribution by tempering. Pack carburized, oil quenched from 880 C, and tempered
at each temperature for 2 h. Surface carbon 1.0% C, retained austenite 916.5%, case depth (0.6% C) 0.3
mm, TCD 0.9 mm. JIS SCr 420 steel, 11.3 mm diam by 80 mm. Source: Ref 32
Fig. 35
20
40
40
80
100
D = 0.3 mm
180 C
C tempered
tempered
40
10
Change in residual stresses due to repeated

stressing. Materials: SNC21, pack carburized
for 2 h at 900 C, annealed, reheated up to 750 C, and
quenched in cold oil of 15 C and tempered at 150 C for
30 min. Stress amplitude: 40 kg/mm2. Source: Ref 31
Fig. 37
Contribution of Residual Stresses

Introduced by Hard Shot Peening
Bending fatigue durability of carburized gears
is improved by surface compressive stresses introduced by hard shot peening. The influence of
compressive residual stress introduced by shot
From surface:
0.01 mm
0.1 mm
0.2 mm
0.3 mm
30
20
10
106
rW 0.931rR 0.275
700
700
600
600
10
500
400
300
0.4
Micro hardness, HV (500 g)
800
Carbon content, %
800
Micro hardness,
HV (500 g)
Fig. 38 Change in residual stresses due to repeated

stressing. Materials: SNC21, pack carburized
for 2 h at 900 C, annealed, reheated up to 750 C, and
quenched into cold oil of 15 C and tempered at 150 C
for 30 min. Stress amplitude: 40 kg/mm2. Source: Ref 31
0.6
0.2
0
After 107
After 4.6 106
20
After 2.6 106
30
Initial
40
0
0.4
0.8
1.2
1.6
2.0

Changes in residual stress distribution due to
fatigue. Materials and test conditions are the
same as that of Fig. 6 and 37. Source: Ref 5
Fig. 39
800
As carburized, hardened,
and tempered
700
600
500
400
(Eq 1)
107
Number of cycles
0.8
900
where rW is the endurance limit (MPa) and rR

is the peak compressive stress (MPa).
Carburizing time, h
0
50
30
10
Carburized, hardened, tempered,

and shotblasted
20
10
200
0
Carburized,
As carburized,
hardened,
hardened,
tempered,
and
andshotblasted
tempered
10
20
30
40
sCarburized,
carburized,hardened,
hardened,
tempered,
and
and tempered
shotblasted
50
60
0.5
1.0
1.5
2.0
Fig. 41
Change of hardness and residual stress distribution by shotblasting. Source: Ref 11
70
2.0
105
60
Fatigue strength, GPa
105
Distortion is an irreversible dimensional

change in the component during heat treatment.
From a metallurgical standpoint, there are thermally induced distortion, transformationinduced distortion, and distortion induced by
both principal causes. In many cases, uneven
cooling often causes distortion corresponding to
any of the three principal causes. Dimensional
change means change in both shape and size. It
is not easy to solve the distortion problems confronting the heat treater and heat treat industries
on a daily basis. Changes in straightness or
warpage of shafts are reversible by applying
stress in the elastic range or during tempering,
for size and shape are not reversible by applied
stresses.
300
2.82 1046
9.27 104
0
104
1.15 1047
The conclusive result proved by a linear relationship between maximum compressive residual stress and bending fatigue strength of gear
tooth leads to Eq 1 (Ref 35). This result fits well
with that of Fig. 42 concluded from various materials and heat treatment.
Distortion of Carburized-andHardened Steels
Residual stresses, kg/mm2
6.05 1045
maximum values of compressive residual

stress introduced by shot peening and bending fatigue strength of gears.
Shot peening was investigated on shot size
and shape, materials, peening machine, shot
speed, arc height, coverage, and the interrelationship among those factors, and these
were simulated to give optimal results.
The maximum compressive residual stress
does not depend simply on the shot peening
condition, but also on the relative effects of
retained austenite before peening. Retained
austenite volume is increased by carbonitriding or by increase of carbon content in relation to other alloying elements such as nickel
or molybdenum.
Grain-boundary oxidation, m
2.15 104
Linear relationship is concluded between
Nominal bending stress, 105 Pa
40
peening on bending fatigue strength of module

2.55 gears was studiedusing JIS SCr 420 and
experimental steels to analyze retained austenite
effects, taking various shot peening conditions
into account and using statistical analysisand
led to the following conclusive results (Ref 35):
Hardness, HV
those accomplishments that can almost eliminate

the negative effect of surface anomalies due to
intergranular oxidation (Ref 28).
5 105
50
40
Endurance limit
30
0.5
1.0
1.5
2.0
1.8
1.6
1.4
Calculated
Fatigue test
2.5
Fig. 40 Influence of residual stress and grain-boundary

oxidation on gear tooth fatigue strength, tested
hydraulic pulsating machine. Module 3 JIS SCr 420 gears
are gas carburized at 920 C and quenched to JIS 1-1-1
cold oil, and tempered at 160 C for 2 h. Source: Ref 26
1.2
1.0
Fig. 42
Ref 35
1.2
1.4
1.6
1.8
Peak compressive stresses, GPa
2.0
Relation between compressive residual stress

(rR) and fatigue strength (rW). See Eq 1. Source:
Maximum pitch error, m
Recent advancement in modeling and simulation to estimate hardness, microstructure, and

stresses is offering new ways to understand the
mechanism of distortion and appropriate measures to reduce distortion problems. Transformation plasticity influences distortion at the
higher-temperature region, where the temperature difference influences the expansion due to
transformation, and at the lower-temperature region where the region transformation at the final
stage is affected by surrounding stress field. Depending on the type of heat treatment, causes of
distortion are different, and the solution to dis-
tortion problems vary. However, by the application of appropriate countermeasures in jigging

and quenching methods, many distortion problems can be minimized.
Types of Distortion
Distortion can be classified into several categories, such as geometrical change as warpage,
twisting, bending, waving, expansion or shrinkage, and nonsymmetrical dimensional change.
Shape and small size changes such as gear tooth
profiles and changes in lead angles also quite
common phenomena that always raise problems.
Distortion related to gears are classified into
three basic categories:
Body distortion in the form of warpage
Twisting introduced in body shape itself
Body shape distorted out of concentricity
Spec. limit
The most serious complicated distortions include

tooth shape change on lead angle, tooth contour
shape as pressure angle, teeth width and height
balance, and pitch errors, which affect gear
noise, vibration, and tooth fatigue life. These
types of distortion have a profound influence on
tooth contact, fatigue durability, local wear and
scuffing problems, and contact noise.
Change of pitch error, m
Examples of Methods to Reduce

Distortion
0
20 25 30 35 40
Jominy (1/ 2 in, HRC)
(a)
S M L
Quench
oil
C B A
Carburizing
cycle
Gear distortion is affected by various factors

such as steel grade, hardenability, grain growth
during carburizing, and quench methods. Most
automotive gears are carburized in continuous
gas-carburizing furnaces and quenched into cold
or hot oil. Recent new technologies applying
vacuum or reduced-pressure carburizing furnaces use a pressurized gas quench system and
the differences of the cooling media from the
traditional process necessitate new cooling technologies.
Influence of Steel Hardenability on Distortion. The following paragraphs discuss the
Fastener hole
Pitch error
(b)
Influence of steel hardenability, quench oil
flow, and case depth on distortion of hypoid
ring gears. Gas carburized at 920 C and press quenched
with cold oil. (a) Maximum pitch error and change of pitch
error in relation to steel hardenability (Jominy, 12 in.)
quench oil flow volume, and effective case depth. (b) Radial distribution of pitch error in relation to fastener hole
placement. Press quench oil flow is adjusted from minimum (S), regular (M), and about 1.5 volume (L) compared with regular flow volume. Approximate effective
case depth is A, 1.1 mm; B, 0.9 mm; C, 0.7 mm. Source:
Ref 39
Fig. 43
Change of overball diameter, m
100
80
110 mm
60
40
20
HT = carburizing (920 C) and

semihot oil quenching (90 C)
0
0
24
26
28
30
32
Jominy (1/ 2 in.) HRC

Influence of steel hardenability of SCr 420, on
over ball diameter of module 2.8 carburizedand-hardened transmission spur gears. Source: Ref 40
Fig. 44
influence of steel hardenability on the distortion

of differential ring gears and its affect on the
over ball diameter (OBD) of transmission gears.
Distortion of Differential Ring Gears (Ref 39).
Differential ring gears have been press quenched
to prevent warpage caused by uneven geometry,
for example, when one side has a gear tooth and
the other side is flat but has many fastener holes.
However, even with press quenching, distortion
of ring gear causes many problems. Most of the
distortion types resulting from quenching are
flatness, taper, inner diameter, ellipticity, pitch
error, and tooth angle and pressure angle.
The pitch error of a ring gear increases with
increasing steel hardenability, as shown in Fig.
43. It also affected by gear design with regard to
placement of fastener holes. The decrease of
thickness behind the gear tooth increases hardness, causes deflection of gear teeth, and increases pitch error. Oil flow rate also has considerable influence, and pitch error increases with
increasing the volume of oil flow. Even with
press quenching, the minimum volume of oil
flow exceeds the level of oil flow necessary to
quench small ring gears used for popular passenger vehicles.
Distortion increases as hardenability increases. Observations show that a quench
method appropriate for small ring gears is hot oil
quench with a rather high cooling rate. As ring
gears have gear teeth on one side and cooling
severity of the tooth side and the fastener hole
side differ very much, it is important to prevent
overcooling of gear tooth side. Therefore, application of a hot oil jig quench is recommended
for small-pitch hypoid ring gears.
Distortion Affecting OBD of Transmission
Gear (Ref 40). Steel hardenability has a profound influence on heat treatment distortion.
Generally, the higher the hardenability of steel,
the larger the core hardness and distortion increase. The average variation of steel hardenability indicated by Jominy (12 in.) value has a
hardness of about 15 to 20 HRC if appropriate
chemical composition control is not provided.
Therefore, the control or selection of steel hardenability is important to reduce lot-by-lot scatter of distortion. The same influence is observed
in many components, especially on gears having
high core hardness and ring gears that have a
fastener hole behind the gear teeth. The distortion indicated in Fig. 44 is JIS SCr 420 (chromium case-hardening steel), and the hardenability is Jominy 12 in. HRC. Transmission gear of
module 2.8 are carburized at 920 C and
quenched into semihot oil of 90 C (Ref 40).
Influence of Machining Condition on Distortion (Ref 40). Residual stress resulting from
previous machining has profound effects on distortion, and as shaved thickness increases, distortion such as OBD and its variations also increases (Fig. 45). Such effects may result from
difference of machinability, tool sharpness, and
also by other machining conditions. Therefore,
the process control for stable quality production
even before heat treatment should be carefully
maintained.

Influence of Setting Methods
Jig Design for Gears (Ref 40). Jig design to
set components is very important to reduce part
shape distortion. Usually, gears are stacked or
hung on several layers of holding shafts depending on the size and shape of components. The
influence of gear-setting methods on lead error
and pitch error is compared to minimize distortion due to the difference of lower side and upper
side of gear flange. A vertical holding method
can reduce the difference of cooling speed at
both sides of the gear flange and seems to reduce
gear tooth distortion (Fig. 46) (Ref 40).
Improvement of Ratchet Gear Roundness
by Jig Design Change (Ref 41). Distortion of
ratchet gear made from straight carbon plate was
corrected by a special jig design developed for
the parts. The hole diameter varies depending on
the fine blanking direction and the cooling condition after carbonitriding. The distortion of
shaft hole diameter (12.06 0.01 mm) of carbonitrided (900 C) hot-rolled plate (5 mm thick)
results in additional reamer finishing to maintain
80
54
OBD error, m
Input shaft
60
tight fit. The first design to hold 15 ratchet plates

could not maintain roundness and necessitated
additional reamer finishes. This type of hole distortion, called out-of-roundness (OOR) can be
corrected by changing the design of holding jigs.
Reducing the cooling speed by using the type C
jig shown in Fig. 47 successfully reduces the
OOR and contributed to the elimination of
reamer finish after treatment. Change in distribution of jig mass near the assembly hole results
in extensive improvement of hole roundness and
completely eliminated rejection.
Direct Hot Quench of Hypoid Gear

(Ref 40)
Hypoid ring gears have been using press
quench methods for more than 60 years. The
elimination of press quench enables reduction of
production cost via the improvement in production efficiency. A stacking jig using a vertical
stacking shaft with massive base plate enables
direct hot oil quench of ring gears. This stacking
jig enabled the reduction of setting work as observed from direct automatic gear-mounting
methods after machining compared to gear setting and removing with each spacer jig gear. The
repeated quenching of the jig distorted the massive base (Fig. 48) and necessitated periodic re-
grinding to keep flatness of quenched gears in

specification (Ref 40).
Influence of Quenching Media and

Cooling Condition on Distortion
Ultrahigh Temperature Oil Quenching of
Ball Bearing Race (Ref 42). Generally, saltbath quenching can produce minimum distortion
products compared with oil quench. The oilquench temperature is usually lower than that of
salt bath. However, new oil usable at 210 C enables the reduction of distortion to a level almost
the same as salt-bath quenching, as shown in Fig.
49. Compared with distortion results from
quenching in the traditional temperature of 80
C, each of the ultrahigh temperature oil and salt
kept at 210 C reduced distortion to almost the
same level (Ref 42).
Influence of Quenching Oils on Distortion
of Steering Sector Shaft. Quench severity of
quenching oil has profound effects on case
depth, core hardness, and distortion of carburized components. Figure 50 shows the relation
between the H value (quench severity index) and
core hardness, case depth, and distortion, respectively, of a carburized shaft. The shaft has
gear teeth on one limited portion, which accelerates local cooling faster than on the other side
356
40
20
A jig
x-y diameter after carburizing and quenching

B jig
C jig
A jig
B jig
C jig
As fine boring
0
60
80
100
120
140
160
180
200
Shaved thickness, m
Fig. 45
0.5
Influence of shaving thickness on over ball diameter. Source: Ref 40
= 12.06 mm
0.3
0.2
120
0.1
0
Lead error, m
A
60
50
Drive face
40
Coaster face
Out-of-roundness
spec = 0.01 mm
30
Upper limit
20
0.01
10
0
A
Pitch error, m
0.01
80
60
0.02
Lower limit
0.03
40
0.04
20
0
0
Fig. 46
0.05
Influence of setting jig design in gear tooth distortion. Source: Ref 40
Fig. 47
Influence of holding jig design on ratchet gear distortion. Hot-rolled plate gas carbonitrided at 900 C and
quenched in cold oil. Source: Ref 41

Distortion Control by Control of Vapor
Blanket Stage. By controlling quench chamber
pressure, the vaporizing character of oil changes
and the cooling power of quench oil can be adjusted to get better results (Fig. 51) (Ref 43, 44).
With quenching after vacuum heat treatment, the

cooling power of the quenching oil does not fit
the necessary cooling speed to increase hardness
and reduce distortion. By changing the quench
chamber pressure from 500 to 100 torr, distor-
450
Core hardness, HV
and causes bending of straight shaft. Figure 50

shows the relation between case depth and core
hardness as almost linear as the H value increases, resulting in deeper case depth and higher
core hardness, while the relation between H
value and distortion is nonlinear, as observed in
Fig. 50. Some oils increase distortion, and other
oils can reduce distortion even if their H value
is high. The characteristic feature of the best oil
that enables distortion reduction is a long vapor
blanket stage; this seems to reduce overcooling
of the gear tooth portion. A similar phenomenon
(Fig. 51) enables the control of vapor blanket
stage of oil quench by the adjusting quench
chamber pressure (Ref 41).
400
TP
350
300
Flatness, m
100
Ring
gear
80
60
40
120
1.0
Shaft warpage, 1/100 mm
250
15
0.9
0.8
0.7
0.6
0.5
20
0
0
2
3
4
5
6
Jig usage, months
8 Stacking
jig base
Fig. 48 Influence of base jig distortion on flatness of

ring gears. Jig: second generation; eight gears
stacked, no spacer. DHR gear: SCM 420H (Cr-Mo casehardening steel, Jominy 12 in. controlled. Heat treatment:
carburized and hot oil direct quenched (130 C). Source:
Ref 40
10
0.06
0.08
15
0.12
0.14
0.1
10
0.4
Agitator pump
pressure, kg/mm2
Relation between H value of quenching oil and ECD, core hardness, and distortion, respectively. Gas carburized at 920 C and quenched from 850 C in cold oil JIS SCr 420 shaft. Warpage is measured at center of
shaft. TP, test piece; R, range of variation; x, average value. Source: Ref 41
Fig. 50
5
Top
Middle
Lower
(a)
200
Nonagitation
Agitation
150
1 atom 130 C
900
Top
800
Middle
700
Lower
Top part
cooled fast
600
500
100
50
0
Sample 2
(80 C)
Sample 4
(210 C)
Salt
(210 C)
(b)
Fig. 49 Influence of quenching media on distortion of

helical transmission gears. (a) Cylindrical distortion. (b) Elliptical distortion. Source: Ref 42
Fig. 51
Temperature, C
Salt
(210 C)
20 40 60 80
Cooling rate, C/s
100
Top
Middle
Lower
0.2 atom 160 C oil
900
800
Lower
700
Top
Middle
600
Uniform vapor
blanket
500
0
20 40 60 80 100
Cooling rate, C/s
10
Lead error, m
Sample 4
(210 C)
Temperature, C
Pressure angle, m
Cylindrical distortion, m
Nonagitation
Agitation
20
Sample 2
(80 C)
Elliptical distortion, m
0.1
H value,cm2
25
500
10
760
600
400
Pressure, torr
200
0
15
30
760
600
400
Pressure, torr
Reduction of distortion by adjustment of quench chamber pressure. Source: Ref 44, 45
200

tion of bearing steel parts was reduced and hardness was also improved (Fig. 52) (Ref 43). Similar pressure-control quenching is applied to
carburizing of automotive gears. The distortion
in pressure angle and helical angle becomes
smallest at 100 torr as shown in Fig. 51.
Distortion Control by Plug Quenching

Roundness and taper distortions of ringshaped products can be corrected by plug
quenching. This method enables precise control
of diameter and roundness by putting a prepared
plug inside the ring work before quenching and
dip or supply cooling media in a way similar to
the press quench machine. Mesh-belt-type continuous reheating furnace or induction reheating
can be fit with plug quench station, and the plug
is pulled out of work after quenching (Ref 41).
Ways to Reduce Distortion by Various

Measures
ables understanding of the effects of cooling on

distribution of temperature, pressure, microstructure, and progress of transformation.
In quenching carburized components, the
early stage of about 30 s after dipping the outer
portion of core materials just inside of carburized
case starts to transform. The influence that high
compressive pressure caused by cooling from
the outer austenite shell has on transformation
plays a very important role in distortion. Under
high compressive stress, transformation starts
from the outer noncarburized core portion and is
affected by the pressure and direction of new
martensite phase expansion to the direction to
reduce pressure. After 50 to 60 s from immersion
and following cooling, the outer-carburized case
reaches the transformation start temperature under the restriction of transformed core, and the
influence of transformation plasticity plays another role on transformation of high-carbon case
and affects residual stresses and distortion (Ref
46).
Depending on component shape and size, design and materials such as shafts, flanged gears,
sector shaft, weld products, and pressed sheet,
quenching or cooling methods are selected to reduce distortion. Also, the cooling process should
not only be simple, but various cooling stages
should be designed for distortion reduction.
Such cooling methods as quick-slow-quick
quenching, immersion time quench, delayed
quench, and many other methods can be used for
reduction of distortion. In the future, technology
to reduce distortion should be established with
the computer-simulation technology to understand the cooling and transformation processes
that may open better routes to minimizing distortion.
Figure 53 shows the difference of ring distortion (outer diameter, 75 mm; inner diameter, 25
mm; and thickness, 10 mm) comparing calculated results with and without transformation
plasticity. Simulation results of distortion estimated for carburized-and-hardened ring showed
large differences in diameter change and distortion of ring body. The simulated results show
that outer and inner diameter without transformation plasticity expand, and the final shape of
the cut section becomes like a hand drum. On
the other hand, simulated results with transformation plasticity shrink both the inner and outer
diameter, and shape of cut section expands and
becomes drum shape. The simulation results of
the same specimen for estimating quenching and
tempering are also shown in Fig. 53. The distortion in diameter change simulated with and without transformation plasticity is quite similar to
that of a carburized-and-hardened ring, while
that of changes in cut section does not differ as
much as that observed in carburized case. Simu-

Initial shape
Experimental
Carburized quenching
Normal quenching
t = 10 s
t = 60 s
t = 2400 s
Influence of Transformation Plasticity

Recent advancement in computer simulation
has enabled the analysis of progress of cooling,
internal stress, and transformation timing (Ref
12, 46). Distortion of heat treated steel can be
analyzed by watching intermittent steps of
quenching by computer simulation, and this en-
No. of works
500 torr
250 torr
Carburized quenching
Normal quenching
100 torr
20
10
0
0.01 0.03 0.05 0.07 0.01 0.03 0.05
0.01 0.03 0.05
Distortion, mm
Fig. 52
Ref 43
Distortion of SUJ 2 part quenched under different pressures. Hardness, 66 HRC. Source:
Influence of transformation plasticity on distortion. (a) Simulated distortion during quenching and measured
data after quenching. (b) Comparison of the simulated results depending on the effect of transformation plasticity. Dimension is enlarged 100, and central axis is on left. Source: Ref 46
Fig. 53

lation technologies should be improved further
to get satisfactory results through the improvement of model and validation efforts and construction of the necessary database, but even at
this stage it can contribute greatly to understanding and improving heat treatment conditions for
reduction of distortion.
REFERENCES
1. H. Sigwart, Influence of Residual Stresses
on the Fatigue Limit, Proc. Int. Conf. Fatigue of Metals, American Society of Mechanical Engineers, 1954, p 272281
2. J.E. Campbell and H.O. McIntire, The Iron
Age, Nov 1953, p 102103
3. W.S. Coleman and M. Simpson, Residual
Stresses in Carburized Steels, Fatigue Durability of Carburized Steel, American Society for Metals, 1957, p 33
4. D.P. Koistinen, The Distribution of Residual Stresses in Carburized Cases and Their
Origin, Trans. ASM, Vol 50, 1958, p
9297
5. K. Funatani, Change of Residual Stress of
Case Hardened Steel by Fatigue, X-Ray Materials Strength, Symposium of Society of
Materials Science, Japan, Nov 1967
6. K. Funatani and F. Noda, The Influence of
Residual Stress on the Fatigue Strength of
Carburized Hardened Steels, J. Soc. Mater.
Sci. Jpn., Vol 17 (No. 183), 1968, p 1124
1128
7. K. Hammer, Untersuchungen zum Zusammenhang zwischen Biegewechselfestigkeit
und Eigen-spannungsverteilung einsatzgehaerteter stahle, Hart-Tech. Mitt., Vol 22
(No. 2), 1967, p 161162
8. G. Parrish and G.S. Harper, Production Gas
Carburizing, Pergamon Press, 1985
9. K. Funatani and H. Nakamura, Influence of
Case Depth and Core Strength to the Fatigue
Strength of Carburized Steels, Toyota Eng.,
Vol 17 (No. 4), 1966, p 293301
10. PSPC/MSF System, Rigaku Catalog
11. K. Funatani and F. Noda, Residual Stresses
and Fatigue Durability of Carburized Steel,
Symposium on X-Ray Study of Material
Strength, Society of Materials Science, Japan, 12 Nov 1968, p 115120
12. T. Yamaguchi, Z.G. Wang, and T. Inoue,
Proc. 27th Japan Congress on Materials
Research, 1984, p 147; Mater. Sci. Technol., Vol 1, 1985, p 872876
13. M. Ulrich and H. Glaubitz, Stand der Induktionshaertung
von
ZahnraedernFestgkeit- und Verschleissverhalten, Z. Verein Deutscher Ingenieure, Vol 91 (No. 22),
1949, p 577583
14. M. Motoyama and H. Horisawa, Residual
Stress Measurements in Case-Hardened
Steels, SAE 710281, Society of Automotive Engineers, 1971
15. R.L. Mattson, Fatigue, Residual Stresses
and Surface Cold Working, Proc. Int. Conf.
Fatigue of Metals, American Society of Mechanical Engineers, 1956, p 593603

16. I. Niimi, Speed Up Carburizing by Controlling Gas Atmosphere in Gas Carburizing,
Toyota Eng., Vol 13 (No. 4), 1962, p 339
350
17. K. Funatani, Key Features in Carburizing
and Carbonitriding, Proc. Carburizing and
Nitriding with Atmosphere, American Society for Metals, 1995, p 255260
18. M.M. Shea, Influence of Quenchant Temperature on the Surface Residual Stress and
Impact Fracture of Carburized Steel, J. Heat
Treat., Vol 3 (No. 1), June 1983, p 3847
19. C.A. Apple and G. Krauss, Microcracking
and Fatigue in a Carburized Steel, Metall
Trans., Vol 4, 1973, p 11951200
20. C. Razim, Ueber den Einfluss von Restaustenit auf das Festigkeitsverhalten einsatzgehaerteter Probe-koerper bei schwingender
Beanspruchung, Hart-Tech. Mitt., Vol 23
(No. 1), 1968, p 18
21. Y. Kojima, N. Miwa, M. Suzawa, K. Nishimura, and Y. Arimi, Carbonitriding and
Hard Shot Peening for High-Strength
Gears, Technical Report No. 5, Mazda,
1987, p 165173
22. A. Hultgren and E. Hagglund, Carbide and
Oxide in Surface Zone of Carburized Alloy
Steels, Trans. ASM, Vol 39, 1947, p 820
842
23 J.J. Kary, Nonmetallic Grain Boundary Material in Carburized Cases, Met. Prog.,
1948, p 218222
24. S. Gunnarson, Structure Anomalies in the
Surface Zone of Gas-Carburized, Case
Hardened Steel, Metal Treat. Drop Forging,
June 1963, p 219229
25. M. Ichihara and S. Kakuwa, Study on the
Abnormal Surface Structure in GasCarburized Steel, J. Jpn. Inst. Met., Vol 30
(No. 3), 1966, p 307312
26. K. Funatani, Einfluss von einsatzgaehaertungstiefe und Kernhaerte auf die Biegedauerfestigkeit von aufgekohlten Zahnraedern, Hart-Tech. Mitt. Vol 25 (No. 2), 1970
27. Case Hardening of Steel, Metallography
and Microstructures, Vol 9, ASM Handbook, American Society for Metals, 1987, p
222
28. K. Funatani, The Influence of Post Carburizing Processes on Fatigue Strength, No.
61, Fall Conf., Japan Institute of Metals, Oct
1967
29. S. Hisamatsu and T. Kanazawa, Improvement of Carburized Gear Strength by Shot
Peening, J. SAEJ, Vol 41 (No. 7), 1987, p
722728
30. H. Buehler and H. Buchgholtz, Longitudinal
Residual Compressive Stresses and Fatigue
Strength, Mitt. Forsch., Vol 3, 1933, p 235
31. S. Taira and H. Murakami, Change of Residual Stress by Aging and Repeated
Stresses, J. Mater. Test., Vol 7 (No. 62),
1958, p 591
32. S. Motoyama and S. Yonetani, Residual

Study Committee Report, 1 Feb 1970, Japan
Society for Mechanical Engineering
33. D.P. Townsend and E.V. Zaretsky, Effect
of Shot Peening on Surface Fatigue Life of
Carburized and Hardened AISI 9310 Spur
Gears, NASA Technical Paper 2047, National Aeronautics and Space Administration, 1982
34. M.D. Lawrenz, Shot Peening and Its Effect
on Gearing, SAE 841090, SAE Trans.,
1984, p 5767
35. H. Aihara, M. Ogino, M. Mitsubayashi, H.
Inagaki, and K. Ogawa, The Strengthening
of Gears by Shot Peening, No. 921125,
Proc. Spring Conf., Society of Automotive
Engineers, Japan, 1992, p 6971
36. M. Mitsubayashi, T. Miyata, and H. Aihara,
Phenomenal Analysis of Shot Peening:
Analysis of Fatigue Strength by Fracture
Mechanics for Shot-Peened Steel, JSAE No.
9430040, JSAE Rev., Vol 15, 1994, p 67
71
37. K. Ogawa and T. Asano, Theoretical Prediction of Residual Stress Produced by Shot
Peening and Experimental Verification for
Carburized Steel, Mater. Sci. Res. Int., Vol
6 (No. 1), 2000, p 5562
38. M. Mitsubayashi, M. Ohnishi, and H. Aihara, Improvement of Pitting Strength by
Combined Shot Peening for Transmission
Gears, Toyota Tech. Rev., Vol 50 (No. 1),
2000, p 106
39. K. Funatani, The Effect of Hardenability of
Steels on Distortion of Differential Gears,
Toyota Tech. Rev., Vol 18 (No. 1), 1966, p
1118
40. A. Takahashi, T. Morishima, and H. Yamada, Quality Control in Heat Treatment, J.
Jpn. Soc. Heat Treat., Vol 30 (No. 6), 1990,
p 301308
41. K. Funatani, The Trends and Tasks of Modeling and Simulation for Heat Treatment Processes, First Int. Conf. Thermal
Process Modeling (Shanghai), March 2000,
Intl Fed. for Heat Treatment and Surface
Eng.
42. E. Nakamura, H. Uchida, and S. Koyama,
The Improvement of Distortion by HighTemperature Oil Quenching, 2nd Intl Conf.,
Quenching and the Control of Distortion,
ASM International, Nov 1996
43. M. Sugiyama, H. Iwata, and M. Uchigaito,
Controlling Oil Surface Pressure in Vacuum
Oil Quenching, Heat Treat., July 1987
44. M. Ogino, State and Future of Materials and
Heat Treatment of Toyota Motor Corp. Automobile and Heat Treatment, Proc. for
Symp. of Japan Society of Industrial Furnaces, Feb 1995, p 20
45. S. Asada and M. Ogino, Reduced Pressure
Quenching Oil and Distortion, 2nd Intl
Conf., Quenching and the Control of Distortion, ASM International, 1996
46. S. Yamanaka, T. Sakanoue, and S. Yoshii,
Influence of Transformation Plasticity on
the Distortion of Carburized Quenching
Process of Cr-Mo Steel Ring, J. Jpn. Soc.
Mater. Sci., Vol 48 (No. 7), 1999, p 733
739

DOI: 10.1361/hrsd2002p459

Metric Conversion Guide

This Section is intended as a guide for expressing weights and measures in the Syste`me International dUnites (SI). The purpose of SI units,
developed and maintained by the General Conference of Weights and
Measures, is to provide a basis for worldwide standardization of units
and measure. For more information on metric conversions, the reader
should consult the following references:
Standard for Metric Practice, E 380, Annual Book of ASTM Stan-
dards, American Society for Testing and Materials, 1916 Race Street,
Philadelphia, PA 19103
Metric Practice, ANSI/IEEE 2681982, American National Standards Institute, 1430 Broadway, New York, NY 10018
The International System of Units, SP 330, 1986, National Institute
of Standards and Technology. Order from Superintendent of Documents, U.S. Government Printing Office, Washington, DC 204029325
Metric Editorial Guide, 4th ed. (revised), 1985, American National
Metric Council, 1010 Vermont Avenue NW, Suite 1000, Washington, DC 20005-4960
ASME Orientation and Guide for Use of SI (Metric) Units, ASME
Guide SI 1, 9th ed., 1982, The American Society of Mechanical
Engineers, 345 East 47th Street, New York, NY 10017
460 / Metric Conversion Guide
Metric Conversion Guide / 461
Metric Conversion Guide / 463

Index
A
ABAQUS computer program .. 21, 155, 202,
303, 425
ABAQUS/Explicit software program, to
model rail steel residual stresses ....... 433
ABAQUS/Standard software program, to
model rail steel residual stresses ....... 433
ABB ................................................11
Abels kernel ................................... 128
Abrasive-jet drilling .......................... 123
Abrasive jet machining (AJM) ............. 111
Ac3 ............................................... 189
Acoustic emission ...................... 13, 14(F)
crack growth in low-alloy steel .............75
evaluating residual stresses in P/M
products ................................. 413
measuring kinetics of stress-corrosion
cracking ...................................80
for studying delayed fracture in steels .....77
Activation enthalpy ................... 56, 57, 58
Activation enthalpy for residual stress
relaxation .......................... 56, 58(F)
Activation enthalpy of self-diffusion ........58
Active-part corrosion (APC)
main gas-pipe line steels .....................82
mechanism .....................................79
Additional light deformation
drawing ............................... 146(F)
Adhesion, effect of thermal treatment on
coatings .................................. 16(F)
Admixed powder .......................... 400(F)
Ae1, carbon and alloying element
effects ...................................... 200
Ae3, carbon and alloying element
effects ...................................... 200
Aging
and distortion ................................ 172
effect on residual stress of carburized
steels .......................... 449, 450(F)
Agitation .............. 258, 259(F), 274, 279(F)
effect of heat transfer ...... 271273, 274(F),
275(F), 276(F), 277(F)
effect on quench oils ............. 264, 268(F)
effect on residual stress and
distortion .......... 275277(F), 283(F),
284(F), 285(F, T)
Agitation rate .............. 262, 266(F), 267(F)
of aqueous polymer quenchants ......... 264,
269(F)
Aggregates of impurities .................... 260
Ain calibration coefficient .....................12

Aircraft coupling rods, shot peening effect on
fatigue life ...............................20(T)
Aircraft engines, P/M parts ................. 398
Aircraft Material Specifications (AMS),
specific types
AMS-S-131 65, Almen strip system .... 349,
350(F)
2770, quench media selection ............. 257
Air drying ...................................... 412
Aireys stress function u(x, y) .............. 106
Air quenching ....... 257, 259(T), 260, 264(F)
Alkaline-earth elements, addition effect on
stress-corrosion cracking ..................81
Alkaline solutions, and stress-corrosion
cracking ......................................79
Alligatoring .................................... 147
Alloying Mo 0.5% Si, rotate bending
fatigue strength ....................... 450(F)
Alloying Ni 2% Si, rotate bending fatigue
strength ................................ 450(F)
Alloy steel reactor pressure vessel,
inhomogeneous residual stress
investigated ..................... 136137(F)
Alloy steels
cryogenic cooling ........................... 331
dimensional instability ................. 337(F)
martensitic transformation ....... 332, 333(F)
nitriding ....................................... 212
plasma nitriding ............................. 210
quenching materials ......................... 312
residual stresses due to turning ....... 109(F)
time of intension cooling
calculation ....................322323(T)
Almen block(s) with the strip .......... 349(F)
Almen intensity ..................................19
Almen strip ......... 91, 96, 151, 152, 351, 352
controlling shot peening intensity .... 349(F)
development of .............................. 345
monitoring degree of shot peening ........91,
93(F)
system ............................... 349, 350(F)
types ....................................... 349(F)
Alpha ferrite (ferrite-martensite), first-order
residual stress parameters ........... 335(T)
Alternating bending fatigue ..................48
Alternating bending tests .........64(F), 65(F)
Alternating magnetic fields, for inducing
residual stress relaxation ..................54
Alumina grinding wheel ..................... 150
Aluminothermic (termite) welding, of rail
steel .................................... 434(F)
Aluminum
alloying effect on hydrogen embrittlement
of steel ....................................83
alloying element effect on crack growth ..80
as alloying element to help obtain fine grain
at heating ............................... 325
density .................................... 367(T)
effect on transition temperature .........76(T)
first-order residual stress parameters 335(T)
heat capacity in liquid .................. 367(T)
heat capacity in solid ................... 367(T)
initial temperature ....................... 367(T)
latent heat ................................ 367(T)
mechanical properties .................. 367(T)
melting temperature ..................... 367(T)
plasticity loss after hydrogenation ......76(T)
Poissons ratio ........................... 367(T)
resistance to hydrogen embrittlement ..76(T)
thermal conductivity in liquid ......... 367(T)
thermal conductivity in solid .......... 367(T)
thermal expansion coefficients ........ 367(T)
thermal properties ....................... 367(T)
time up to fracture ........................76(T)
work of propagation of a ductile
crack ..................................76(T)
Youngs modulus ....................... 367(T)
Aluminum alloys
modulus of elasticity .....................96(T)
Poissons ratio .............................96(T)
residual stresses in castings with
solidification time ................. 362(F)
shear modulus .............................96(T)
Aluminum alloys, specific types
2014-T6
residual stress of soft metallic
material .......................... 346(F)
7010-T76
material library screen printout .... 352(F)
7075
casting speed .......................... 378(T)
density ................................. 378(T)
dilation due to solidification ........ 378(T)
gradient of liquidus time ............ 378(T)
gradient of solidus line .............. 378(T)
hardening coefficient ................ 378(T)
heat conductivity ..................... 378(T)
initial yield stress .................... 378(T)
latent heat due to solidification .... 378(T)
liquidus temperature ................. 378(T)
pipe and mold of centrifugal
casting ..................380, 382383,
384(F), 385(F), 386(F)

466 / Index
Aluminum alloys (continued)

Poissons ratio ........................ 378(T)
solidus temperature .................. 378(T)
specific heat ........................... 378(T)
thermal expansion coefficient ...... 378(T)
viscosity ............................... 378(T)
Youngs modulus .................... 378(T)
Aluminum-magnesium alloys, macro and
micro residual stress relaxation ..........55
Aluminum nitride (AlN) ............... 209, 325
influence on residual stress ............ 214(T)
crack ..................................76(T)
Aluminum oxide coating .................... 118
Aluminum-silicon alloys, permanent mold
casting ..................................... 369
American Standard of Testing Materials
(ASTM) test methods, specific types
D6200, cooling-curve analysis ............ 266
E 83792 hole-drilling method ....... 123(T)
Amorphous materials ........................ 400
Anharmonic property of the solid ........ 113
Anisotropic transformation strain
modeling of 79(F)
and transformation plasticity ............. 7(F)
Annealing
eliminating residual stresses ............... 144
factors affecting ...............................54
hardened or machined materials ........ 55(F)
inducing residual stress
relaxation ..................... 5460(F,T)
before nitriding .............................. 209
P/M parts .........................406, 407, 408
soft, and distortion ...... 153, 154(F), 155(F)
temperature .....................................56
time ......................................... 54, 56
Annealing twinning, shape change ........ 5(T)
Anodizing ....................................... 118
ANSYS software program ............ 127, 202
Antimony
alloying effect on hydrogen embrittlement of
steel .................................. 83, 84
crack ..................................76(T)
Antioxidants, in quench oils .......... 263, 264
Approximation method ..................... 334
Aqueous acid solutions, and stress-corrosion
cracking ......................................79
Aqueous polymer quenchants ....... 257, 264,
268(F), 269(F), 314316, 319320,
323324, 326327
Aqueous polymer quenching ........ 269, 270,
272277(F), 279(F), 280(F), 283(F),
285(F,T)
computer simulation of steel cylinder ... 289,
291(F)
Ar1 ............................................. 91(F)
Ar3 ............................................. 91(F)
ARBITRARY computer modeling

program ................................... 429
Arbitrary cut-out indicator
method ..............................132133
Arc height ...................................... 349
Arc welding .................................... 405
Area reduction ...... 144(F), 145, 146, 147(F)
Argon, as quenchant ................ 261, 265(F)
Armco iron, nitriding ......................... 212
Armor-iron, delayed fracture ..................74
Arsenic, alloying element effect on crack
growth .......................................80
Associated ammonia furnace atmosphere,
composition ........................... 404(T)
Associated methanol furnace atmosphere,
composition ........................... 404(T)
Asymmetry
of axle components ..... 159161, 163(F,T),
164(F)
and distortion of components ............ 165,
167(F,T), 168(F)
Atmosphere
composition of constituents ........... 404(T)
gas velocity differences and distortion ... 184
hot pressing .................................. 408
oxygen-containing, and carbon fibre
composite materials ................... 184
requirements for sintering .............. 404(T)
sintering ......................402403, 404(T)
Atmosphere control, and
distortion ..................256257, 259(T)
Atomization of molten metal ............... 399
Auger spectroscopy, to study brittle fracture
in hydrogen embrittlement in steel ......79
Austempering .. 252253, 257, 261263, 280,
281, 417
Austenite
anisothermal decomposition ..... 200202(F)
atomic volume in ferrous alloys ...... 250(T)
electronic states of .......................... 3, 4
expansion coefficient ...................... 4(F)
heat capacity and magnetic
properties ........................... 34(F)
homogenization with induction
hardening ..................... 233234(F)
nitrocarburizing of, deformation
tendency ............................ 170(T)
in schematic representation of relative
transformation of mild steel
plate ................................... 91(F)
thermal expansion coefficient ............ 7(F)
transformation products .............4(F), 5(F)
transformation to acicular lower bainite,
volumetric change ................. 250(T)
transformation to feathered upper bainite
volumetric change ................. 250(T)
transformation to martensite at cryogenic
temperatures ................. 331, 332(F)
transformation to martensite volumetric
change .............................. 250(T)
Austenite grain growth ...................... 325
Austenite grain size ........................... 325
Austenite transformation
after carburizing and carbonitriding ...... 189
during hardening processes, computation
of .............................. 200202(F)
Austenitic stainless steels .................... 110
abrasive jet machining ...................... 111

residual stresses due to grinding ...... 109(F)
strain hardening ............................. 109
Austenitic steels
floccules not observed ........................73
residual stress vs. temperature ........... 7(F)
shot peening of chemical process
vessels ........................ 354, 356(T)
Austenitic steels, specific types
X 2 CrNiMoN 225-3 (1.4462)
stress-corrosion cracking test
results ............................ 356(T)
X 6 CrNiTi 17122 (1.4571)
results ............................ 356(T)
X 6 CrNiTi 1810 (1.4541)
results ............................ 356(T)
Austenitizing ................................... 407
of high-speed steels ......................... 337
high-temperature ..............................81
of low-alloy steels ...................... 337(T)
supercarburized high-speed steels .... 337(T)
two-stage .......................................81
Austenitizing temperature ....77, 248(F), 251
and volume changes .............. 252, 254(F)
Autofrettaged tube, crack propagation .. 35(F)
Autofrettaging process, and crack propagation
35(T)
Autofretting ......................................12
Automated stress mapping ................. 100
Automobiles, ferrous P/M parts ............ 398
Automotive gears, carburization of ........ 453
Avrami approach ......... 56, 57, 59(F), 60(T)
parameters ............................. 56, 58(F)
Avrami equation .............................. 217
Avrami kinetic law ........................... 201
Axial internal residual stresses from
induction
hardening ............. 234235(F), 236(T)
Axial residual stresses, after induction
hardening .............................. 240(F)
Axial stresses .. 144(F), 145, 146, 147, 148(F),
149
for bearing ring .......................... 317(F)
residual stress distribution in radial direction
at t8.32 s using a plane-strain
assumption ......................... 368(F)
residual stress variation in the axial direction
at t8.3202 s with a traction-free top
surface .............................. 368(F)
semicontinuous casting distribution ..... 378,
380(F)
specimen heating ........................ 313(F)
thermal stress history with a plane-strain
assumption ......................... 368(F)
thermal stress history with a traction-free
top surface .......................... 369(F)
Axle components
asymmetry of ... 159161, 163(F,T), 164(F)
distortion ........ 159161, 163(F,T), 164(F)
quenching of ............................. 174(F)
B
Back stress ........................... 299, 300(F)
Backstress tensor, of rail steel .............. 432

Bain distortion ................................. 332

Bainite ................................ 4(F), 250(T)
deformation systems associated with
transformations ....................... 5(T)
transformation of mild steel plate
.......................................... 91(F)
shape change ................................ 5(T)
transformation-start temperature, stress
effects on ...........................6(F)(T)
Bainite habit plane, volume fraction of,
tensile load effects ................78, 9(F)
Bainitic steel, residual stress vs.
temperature ............................... 7(F)
Ball
Kondratiev form factor, K, ............ 323(T)
quenching and heating computational and
experimental results ...... 321323(F,T)
Ball-bearing steels
dimensional stability of French
specimens .................339(F), 340(F)
Ball screw, distortion of ............ 166, 168(F)
Banded segregation of carbide, and
quenching distortion ........... 168, 169(F)
Barkhausen noise analysis (BNA) ........ 112,
114115, 156(F)
Bars
drawing ................................... 141(F)
hollow cylinder, rotationally symmetric
stresses, measurement ................ 106
solid cylinder, with/without rotationally
symmetric stresses, measurement ... 105
Basquin relation ............................ 30, 34
Batch quench system ............... 258, 259(F)
Bauer-Heyn method .......................... 102
for destructive residual-stress
measurement ....................... 102(T)
Bauschinger effect ...... 62, 66, 142143, 299
in rail steel ......................... 431, 432(F)
Bauschinger-induced work-softening .......66
Beam stress ..................................9293
Bearing rings
deformation prediction during
hardening .................. 317320(F,T)
distortion ........................... 152155(F)
ovality after hardening ........... 153, 155(F)
quenching in a hardening press ....... 185(F)
quenching process and
hardening .................. 315317(F,T)
residual stresses and conicity .......... 319(T)
service life .................................... 319
surface layer removal and service life ... 319
Beilby layer .................................... 214
Bending ............................... 151, 154(F)
mechanical, thermal, or structural origins of
residual stress ........................13(T)
modeling and measurement of residual
stresses ........................ 154156(F)
with superimposed tension ................ 143
Bending deflection, asymmetrical components
165, 167(F,T), 168(F)
Bending fatigue ................................ 189
crack initiation and residual stresses ... 34(F)
increased by shot peening .......... 198199,
201(F)
Bending fatigue crack initiation,

types ..................................197198
Bending fatigue limit, in medium-strength
steel ..............................49(F), 50(F)
Bending fatigue resistance
of carburized steels ......................... 195
and residual stresses of carburized
steels ..............................197198
Bending fatigue strength
deep rolling effect ......................... 37(F)
of gear teeth, grain-boundary oxidation
effect ....................450451, 452(F)
gear wheels ............................... 244(F)
grinding effect ............................. 37(F)
of induction-hardened gear teeth and
machine parts ................ 243244(F)
and milling residual stress
influence ......................... 3637(F)
milling vs. grinding of steels ................38
roughness effect on surface of low-strength
steels .................................. 39(F)
shot peening with grinding, S-N curves and
residual stress effect ........... 3738(F)
vs. surface hardness in low-strength
steel ................................... 40(F)
Bending fatigue testing, quenched-andtempered steels in seawater ........... 34(F)
Bending moment .....9192, 93, 94(F), 95(F),
143(F), 145
symbol and units of .......................93(T)
Bending of sheet .... 142(F), 143(F), 144145
Bending out (d) ...................... 152, 154(F)
Bend testing ..........................64(F), 65(F)
Bent-beam stress corrosion test, specimen
preparation practices ..................... 110
Biaxial residual-stress state, determination
of .................................... 9697(F)
Biaxial stress analysis ........................ 122
Bidimensional dislocation networks ...... 342
Bielastic bands ........................... 129, 131
piecewise-homogeneous, residual stress
determination .......................... 131
Biharmonic differential equation .......... 106
Bimetallic bands, edge cracking ........ 129(F)
Bimetallic materials (bimetals)
inhomogeneous residual stress
fields ................................ 137(F)
used in power installation shells with
corrosion-proof surfacing ............ 129
Bin calibration coefficient .....................12
Binder bonded powder ............ 399, 400(F)
Binders .................................... 399, 401
Binder systems, for metal-injection
molding .................................... 412
Biot criterion ................................... 322
Biot number ........265, 282, 314(F), 316, 323
circular residual stresses compared to
generalized value ............ 314315(F)
generalized ................... 322, 323, 324(T)
usual ..................................... 314, 315
Birefringence
shear wave ................................... 114
texture-induced .............................. 114
Birefringent gages ............................ 109
Birefringent methods ........................ 110
Blank manufacturing methods ... 167169(F)
Blasting ......................................... 150
Index / 467
Blind hole drilling ................. 99, 110111
Blisters ....................................... 72, 83
Blowpipe positioning, in induction
hardening .................................. 220
Blue tempering ..................................91
Bodner inelastic constitutive equations .. 372
Bodners model ...................... 373, 374(F)
Body-centered cubic (bcc) materials, residual
stresses ..................................... 333
Boiling film ..................................... 314
Bolt, distortion of .......................... 181(F)
Boost-and-diffuse carburizing method .. 189,
190(F), 440, 444
Boreholes, for quenching after induction
hardening .................................. 232
Boriding, deformation tendency ......... 170(T)
Boring, mechanical, thermal, or structural
origins of residual stress ..............13(T)
Boron
alloying effect on grain-boundary
oxidation ................................ 449
content effect on shrinkage and sintered
density .................................. 404
Boronizing, deformation tendency ...... 170(T)
Bosch, Robert ....................................11
Boundary-value problems for twodimensional elasticity theory ........ 127,
128, 129
Bowing ............................................91
Bragg angles ............................... 112(F)
Bragg reflection ............................... 120
indices of ..................................... 120
Braggs Law ............................. 113, 333
Brass
third-order residual stresses ...... 334335(F)
Brass 46,
first-order
residual stress ......................... 335(T)
Brass 73,
first-order
Brazing ...................... 393(T), 394396(F)
finite element analysis for modeling residual
stresses .................. 394, 395, 396(F)
metal-ceramic compounds ....... 394395(F)
parameters influencing complex residual
stress state ........................394395
P/M parts ...............................405406
shrinkage of joints ................ 394(F), 395
tensile residual stresses and
thickness ratio ............394(F), 395(F)
thermal expansion coefficients and their
differences .................394(F), 395(F)
thickness ratio
influence for
ceramic and steel .........394(F), 395(F)
tungsten carbide-carbon cemented carbide
with steel ..................... 395396(F)
tungsten carbide-cobalt with steel ....... 394,
395(F)
Brine, as quenchant ...........264265, 270(F)
Brine quenching, ..............264265, 270(F)
British Aerospace ...............................11

468 / Index
Brittle coatings ................................ 110

Brittle fracture, carburized steels ........... 196
Bubbles ............................................72
Buckling ..........................................27
Built-up crack method .................128129
Bushings
eccentric cylinder, distortion ..159, 161(F,T)
plug quenched ......................... 257
Butt curl ........................................ 363
C
CAD. See Computer-aided design.
Calcium
steel .......................................83
Calcium chloride, as quenchant ............ 326
Calibration coefficients ........................12
Cam springs, shot peening effect on fatigue
life ........................................20(T)
Canning ......................................... 404
Cantilever beam bending test ........ 102, 119
Carbide coagulation, and nitriding ......... 209
Carbide precipitation
and nitriding ................................. 209
of P/M tool steels ........................... 408
Carbide segregation .................... 168, 260
Carbon
composition requirements for powder forged
parts ................................. 411(T)
content effect in nitrided compound
layer ..................................... 213
content effect on cracking propensity ... 254,
256(T)
content effect on cylinder distortion .... 159,
163(F)
content effect on quench cracking ....... 279,
286(F)
content effect on residual stresses and
distortion ..................... 249, 250(F)
content effect on shrinkage and sintered
density .................................. 404
crack ..................................76(T)
Carbonate-bicarbonate earth solutions, pH
media for gas pipe line steels ............82
Carbonate solutions, and stress-corrosion
cracking ......................................79
Carbon atoms, effect on micro residual
stresses in hardened steels ................63
Carbon diffusion model ........... 199, 202(F)
Carbon distribution in steel, and stresscorrosion cracking .........................80
Carbon fibre composite (CFC), not used with
oxygen-containing atmospheres ........ 184
Carbon fiber/epoxy composites, incremental
hole-drilling method for residual stress
evaluation .......................... 12, 14(F)
Carbon gradient, of carburized-and-hardened

steels ................................... 437(F)
Carbonitrided steels, cryogenic cooling .. 340
Carbonitriding ......... 189, 190191(F), 407,
408(F)
of gears ................................... 454(F)
increasing retained austenite volume and
maximum compressive residual
stresses .................................. 452
introducing compressive residual stress 437
process temperatures ..................190191
temperature differences and distortion ... 184
Carbon-manganese steel, stress relief and
yield strength ................... 251, 254(F)
Carbon monoxide, and carburizing steel
process ..................................... 448
Carbon potential .............................. 407
and carburizing, influence
during ...................... 444446(F,T)
control in carburized steels ...... 446447(F)
effect on carburized-and-hardened
steels .................................... 438
Carbon profile ................................. 193
after boost/diffuse carburizing ........ 203(F)
after carburizing of 15CD4
steel cylinder ................. 203, 204(F)
Carbon steels
aging effect on carburized steel .......... 449,
450(F)
austempering ................................. 281
carbonitriding ............................ 191(F)
case hardening ........................... 438(F)
cold plastic deformation ......................79
composition .............................. 407(T)
for gear wheel, induction hardened,
computer simulation ....... 296, 306(F),
307309(F)
hydrogen embrittlement .................. 76(F)
quenching ...... 275, 281(F), 282(F), 284(F),
285(F)
quenching, agitation effect ...... 276, 283(F),
284(F), 285(F)
quenching of disk and cylinder computer
simulation ......... 288289(F), 290(F),
291(F)
for ring, induction hardened and carburized
quenched 303(F), 304(F), 305307(F)
stress-corrosion cracking ................... 110
tight scale formation .............. 255, 258(F)
unalloyed, carburized
CCT diagram ..........192193, 194(F)
Carbonyl iron process,
decomposition of ......................... 400
Carbonyl process ............................. 412
Carburization .................. 407, 408(F), 418
atmosphere influence ....................... 402
of P/M parts ....................... 406, 407(F)
Carburized-and-hardened steels
carbon gradient, hardness, and residual
stress ................ 437438(F), 439(F)
distortion ........................... 452457(F)
residual stress distribution ....... 437438(F)
Carburized bushing, quenching and thermal
stresses ........................ 320321(F,T)
Carburized case ............................. 92(F)
Carburized parts .................... 196, 199(F)
Carburized products, hardening

methods .......................... 189, 190(F)
Carburized steels. See also Carburized-andhardened steels, Carburizing.
aging effect on residual stresses ......... 449,
450(F)
bending fatigue resistance .... 195, 197198
contact fatigue resistance .................. 198
induction hardening of cylinder,
model ......... 203204, 205(F), 206(F)
quenching temperature effect on mold hole
distortion ........................... 171(T)
repeated loading effect on residual
stresses ........ 449450, 451(F), 452(F)
repeated stressing effect on residual
stresses ........ 449450, 451(F), 452(F)
retained austenite influence ...... 446447(F)
tangential stresses of cylinder during
quenching, model ........... 202, 203(F)
tempering effect on residual stresses .... 449,
450(F), 451(F)
tempering evaluation by x-ray
diffraction .............................. 112
Carburizing. See also Carburized-andhardened steels, Carburized steels. .... 189
carbon potential influence ..... 444446(F,T)
compressive residual stresses from ....... 437
compressive stress development ........ 92(F)
diffusion process computation .. 199, 202(F)
distortion tendency ...... 176(T), 178179(F)
duration effect on fatigue strength
of steels ................................. 446
heating device distances .................... 184
heat-transfer model ...................199200
and nitriding, compared .................... 209
phenomenological coupling (thermal,
metallurgical, and mechanical
interactions) .................. 199, 201(F)
pre-cooling quenching ...................... 177
rotate bending fatigue strength ........ 450(F)
techniques .................................... 189
temperature differences and distortion ... 184
temperature field computation .......199200
transformations and stress
evolution ........... 191194(F), 195(F),
196(F), 197 (F)
Carburizing and NH 3 addition, rotate
bending fatigue strength ......... 450(F)
Carburizing and quenching, variation
influence on residual
stresses .............. 442444(F), 445(F)
Carburizing at low temperature, deformation
tendency ............................... 170(T)
Carburizing-quenching process
of chromium-molybdenum steel ring,
distortion computer simulation ..... 290,
292(F)
of cylinder, computer simulation ........ 296,
302(F), 305
deformation tendency ................... 170(T)
of ring, computer simulation ... 296, 303(F),
304(F), 305307(F)
simulation of ...........................285286
Case depth ...................................... 407
of carburization ........................178179

in carburized gears, residual stresses .... 441,

442(F)
and carburized steel residual stress ... 445(F)
and fatigue strength of carburized
steels ................................ 446(F)
heat treatment effect on residual stress
distribution ............................. 194
after induction hardening rollers ..... 239(T),
240(F)
influence on residual stresses of carburized
steels ................................ 446(F)
unevenness and distortion potential ...... 184
Case hardening ........................... 12, 211
compressive stresses from ...................90
and crack initiation ...........................29
P/M parts ............................... 406, 407
Case-hardening depths, influence on residual
stresses ........................... 194, 197(F)
C.A.S.E. process .... 350351, 352(T), 353(F),
354(F)
Case thickness, defined ....................... 194
Casting ............... 99, 361370(F,T), 400(F)
air-gap width ............. 365(F), 366(F), 369
axial, hoop, and radial stresses of
solidification ................. 368369(F)
continuous .................................... 370
defects due to porosity formed from
shrinkage ............................... 370
distortion ..................................... 361
elastic-plastic deformation regimes in
solidifying material ......... 361362(F)
elastoplastic behavior ....................... 363
elastoviscoplastic behavior ................ 363
ferrostatic pressure effects ................. 370
finite-element analysis for
modeling ................... 364369(F,T)
finite-element analysis of defects ......... 363
hysteresis loop of stress ................ 375(F)
ingot ........................................... 370
interdendritic voids ......................... 370
locking problems ............................ 369
mixed-section, uneven, cooling causing
constraints .......................... 361(F)
permanent mold, of aluminum-silicon
alloys .................................... 369
production of residual stress ..... 362364(F)
residual porosity ............................. 363
shrinkage defect location and size
determination .......................... 370
single-component heavy-section .......... 361
static ........................................... 370
stress concentration regions predicted
367(F)
thermal stress-generated defects, dollar
loss ...................................... 361
Castings
boundary condition of heat transfer and
thermal radiation ...................... 375
creep ...................................... 373(F)
distortion factors contributing to ....167168
L-shaped, shrinkage cavity prediction ... 370
strain rate effect on hardening
coefficient .......................... 373(F)
stress related to creep strain in a logarithmic

scale ................................. 373(F)
tension testing ............................... 373
T-shaped, shrinkage cavity prediction ... 370
yield stress ............................... 373(F)
Casting speed, of aluminum alloy ...... 378(T)
Cast iron
roller burnishing effect on
fatigue strength ................. 19, 20(F)
spray forming ..........................412413
Cast steels, stress-corrosion cracking .........81
Catalytic debinding ........................... 412
Catastrophic fracture ..........................90
Cathode polarization method ........... 73, 74
Cauchy (C) distribution ..................... 334
Cauchy method ................................ 122
Caustic quenching ............264265, 270(F)
Caustic soda, as quenchant ..264265, 270(F)
Cavity of concave dies, distortion of ...... 165,
168(F)
Cementation steel, induction
hardening .............................. 244(F)
Cemented carbides, in brazed joint
with steel .....................395(F), 396(F)
Cementite ....................................... 249
Cementite plates, shape change ............ 5(T)
CEN. See European Committee for
Standardization.
Center of gravity method ................... 122
Centrifugal atomization ........... 400(F), 414
Centrifugal casting ..................... 167, 168
products formed ............................. 380
and residual stress formation ... 379383(F),
384(F), 385(F), 386(F)
Ceracon (CERAmic CONsolidation)
process .................................... 409
Ceramic coatings ................................12
Ceramic shot, hardness range ............... 349
Cerium
steel .......................................83
work of propagation of a
ductile crack .........................76(T)
Chaboches multiple component plasticity
model ............................ 432(F), 433
Characteristics data file format ............ 304
Characteristic x-ray radiation, of first-order
residual stresses of
metallic materials .................... 335(T)
Charging basket, and distortion .. 172, 173(F)
Chasing, mechanical, thermal, or structural
Chemical driving forces ........................ 8
limits affected by temperature ............ 6(F)
Chemical etching .............................. 439
Chemically Assisted Surface Engineering
(C.A.S.E.) process ...... 350351, 352(T),
353(F), 354(F)
Chemical polishing ........................... 109
Chemical species flow .............. 362, 363(F)
Chemical strain-measurement methods .. 110
Index / 469
Chemical vapor deposition (CVD) ...25, 118,
120
Chemical vapor deposition (CVD)
coatings ................................ 11, 12
Chen method, for destructive residual-stress
measurement .......................... 102(T)
Chills, in L-shaped and T-shaped castings with
shrinkage defects ......................... 370
Chip formation, in milling .................. 150
Chip-removal methods .................. 109(F)
Chrome plating ..................................12
Chromic steels, floccules present .............73
Chromium
first-order residual stress parameters ....335(T)
increasing retained austenite in casehardening steels ....................... 447
for low-alloy P/M steel powders ......... 399
oxidation in carburizing atmosphere ..... 448
crack ..................................76(T)
Chromium-alloyed steels, nitriding ........ 216
Chromium carbides ....................403404
in high-speed steels after cryogenic
cooling ..................340341(F), 343
Chromium-manganese steels
floccules present ...............................73
nitriding ....................................... 212
Chromium, (max), composition requirements
for powder forged parts ............. 411(T)
Chromium-molybdenum steels
hardenability effect on core hardness and
case depth .......... 441442(T), 443(F)
nitriding ....................................... 212
ring distortion, carburized-quenched,
computer simulation ........ 290, 292(F)
quenched .... 303(F), 304(F), 305307(F)
Chromium-molybdenum-nickel steels,
induction hardening .................. 244(F)
Chromium-molybdenum-vanadium steels,
nitriding ................................ 212(T)
Chromium nickel (cementation) steels,
Chromium-nickel steels, floccules
present .......................................73
Chromium-nickel-tungsten steels,
floccules present ............................73
Chromium nitrides .............. 209, 403404
Chromium nitride (CrN), influence on
residual stress ......................... 214(T)
Chromium nitride (Cr 2N), influence
on residual stress .................. 214(T)
Chromium-silicon spring wire, residual stress
distribution after shot peening .. 91, 94(F)
Chromium steels, cryogenic cooling ....... 340
Circular hole drilling method ........ 130, 133
Circumferential hoop stress ......... 94, 95(F)
Circumferential (hoop) stress in tube,
formula for calculating residual stress
.................................... 95(F), 96(T)

470 / Index
Circumferential stress
for bearing ring .......................... 317(F)
from specimen heating ................. 313(F)
Circumferential stress in tube from net
opening displacement X, formula for
calculating residual stress .............96(T)
Cladding, crack-indicator method
for determining inhomogeneous
residual stress fields ........... 136137(F)
Clamping, and distortion ..................... 166
Clausius-Duhem inequality ................. 298
Clay-coating method .........278(F), 281282
quenching, and quench cracking ......... 278,
286(T)
Climb ..............................................58
Coagulation phenomenon ......... 342(F), 343
Coalescence ......................................15
Coatings
brittle .......................................... 110
compressive stress at elevated
temperature ............................. 118
compressive stresses ........................ 118
and quenching, effects on ........ 273, 278(F)
stress determination in ........ 118124(F,T)
tensile stress at elevated temperature ..... 118
Coating separation ........................... 118
Cobalt
steel .......................................83
crack ..................................76(T)
Cobalt-chromium-aluminum-yttrium
coating ..................................... 118
Cobalt-free superhigh speed steels
composition .............................. 406(T)
designation ............................... 406(T)
hardness .................................. 406(T)
Weq ........................................ 406(T)
Coefficient K, of first-order residual stresses of
metallic materials .................... 335(T)
Coefficient of linear expansion, of bearing
ring being hardened .................. 318(F)
Coefficient of thermal expansion .......... 427
Coffin-Mason relation .........................30
Coherent-optical interference methods
..................................... 134136(F)
Cold cracks .......................................73
Cold die pressing, in P/M processing .....397(F)
Cold-die quenching ........................... 281
Cold forging .................................... 141
Cold forming (working)
definition ..................................... 141
steel ................................. 147148(F)
Cold injection molding, in powder metallurgy
processing ............................. 397(F)
Cold isostatic pressing ................. 402, 404
in powder metallurgy process ......... 397(F)
Cold oil quenching ................. 449, 453(F)
Cold-rolled steel ............................... 111
Cold-rolled steel sheet stampings, distortion
of ................................. 168169(F)
Cold rolling .............................. 141, 419

and distortion ...... 152, 153, 154(F), 155(F)
in powder metallurgy processing ..... 397(F)
Cold-shortness threshold, antimony and tin
alloying effects .............................83
Cold sintering .................................. 402
Cold slip casting, in powder metallurgy
processing ............................. 397(F)
Collectors. See Pores.
Compaction, in powder metallurgy
processing ............................. 397(F)
Compaction cycle ......................... 401(F)
Compatibility equations of ............ 105, 106
Compliance method ............................91
Component design, and distortion ... 252253,
255(F), 256(F)
Compressive hoop stress ......................95
Compressive residual stresses ... 193, 196(F),
211, 216
area decrease by fatigue test, repeated
stressing ...................... 450, 451(T)
in carburized steels after
heat treatment ...... 449450(F), 451(F)
after induction hardening .................. 239
induction hardening of gear tooth .... 242(F)
induction hardening of machine parts .... 243
in spring steel ..................................27
Compressive residual stress zone ....194195
Compressive shrinking stresses ............ 193
Compressive stresses
increase by shot peening ...198199, 200(F)
induction hardening ............... 232242(F)
Compressive thermal stresses, from
welding ................ 392, 393(F), 394(F)
Computational fluid dynamics (CED) .... 258
Computer-aided design (CAD) ............ 413
Computer-aided engineering (CAE)
systems ..............................302303
HEARTS ........... 296, 301(F), 302304(F)
Computer-controlled robotics system, fiveaxis for lasershot peening ............. 355
Computer modeling ......................... 100
Computer-numerical-controlled (CNC)
servohydraulic multiplaten powder
press ............................... 410, 411
Computer programs. See also Computer
simulation.
modeling ...................................... 202
Computer simulation ........................ 202
of carburizing-quenching process ...285286
chromium-molybdenum steel ring distortion
carburized-quenched ........ 290, 292(F)
of cylinder, carburized ..... 296, 302(F), 305
discrepancies of calculated models with
measured amounts .......... 204206(F)
of gear wheel, dual frequency induction
hardened ...... 296, 306(F), 307309(F)
heat-transfer coefficients modeling ....... 268
for jet-hardening and fog-quenching
process .................................. 174
of quenching a carbon steel disk and
cylinder ... 288289(F), 290(F), 291(F)
rail steel roller straightening ..... 432434(F)
of residual stresses/distortion of scanning
induction hardened cylinder ........ 290,
293(F), 294(F)
of ring, induction hardened and carburized

quenched ...........296, 303(F), 304(F),
305307(F)
software for residual-stress
measurement ........................... 113
stress/strain prediction in case-hardened
parts ........................... 199, 201(F)
warpage of steel shafts .....289290, 291(F)
Concurrent engineering .......................11
residual stress-integrated design and other
sectors connected ............... 24(F), 25
Conductance-time curves ................... 269
Conformal mapping functions ............. 132
Consistency relation held during creep
loading .................................... 302
Constantinescu and Ballard method ..... 105
Construction, and distortion potential ..... 183
Contact fatigue ............. 150, 196, 215216
carburized steels ............................. 196
Contact fatigue resistance, and residual
stresses of carburized steels ............ 198
Contact length ................................. 143
Contamination, of quenchants ............. 259
Continuous casting
and crack initiation in gas main pipes .....82
finite-element method
formulation .......................376377
Continuous casting by twin-roll method,
residual stress formation ......... 383385,
386(F), 387(F), 388(F), 389(F)
Continuous cooling ................. 191193(F)
Continuous cooling curves,
superposition of .......................... 249
Continuous-cooling-transformation (CCT)
diagrams ........................ 248249(F)
of carburized steels ........ 191192, 193(F),
196(F)
cooling austenite transformation
for steel .........................45, 314(F)
of supercooled austenite .................... 320
Continuous gas-carburizing furnaces, to
carburize automotive gears ............. 453
Continuous weight functions ............... 127
Continuum approach ........................ 415
Contour hardening ........................... 241
Control/analysis condition data file
format ..................................... 304
Convection-cooling stage ....... 268, 269, 270,
271, 274
Convection heat transfer, area of .......... 312
Convective heat transfer (cooling)
(CONV) ......................... 258, 261(F)
Converters ...................................... 221
Cooling ................................... 8990(F)
circumferential stresses at characteristic
points versus time ................. 314(F)
effect on distortion of
components ............... 172177(F,T)
residual stresses experimental vs. calculated
values ............................... 314(F)
temperature field changes ........ 313314(F)
and thermal stress buildup relieved ......... 7
without transformation ........... 250, 252(F)
with transformation ........ 250251, 252(F),
253(F)
Cooling acceleration effect ........ 273, 278(F)

Cooling atomic reactors, stress-corrosion

cracking of steel ............................81
Cooling characteristics, effect on residual
stress and distortion from quenching
275278, 281(F), 282(F), 283(F),
284(F), 285(F),T) 286(T)
Cooling condition, effect on distortion
of gears .......................... 454456(F)
Cooling curve ........... 192, 193, 196(F), 259,
262(F), 265271(F), 272(F)
agitation effect ..................... 271, 274(F)
water temperature effect ............... 270(F)
Cooling curve analysis ..........265267, 268,
270(F), 271(F)
Cooling-curve tests ..................... 265, 274
Cooling factor ................................. 322
Copper
alloying addition effect on sintering ..... 402
alloying effect on P/M structural
steels ..........................406407(T)
alloying element effect on crack
growth ....................................80
composition requirements for powder
forged parts ........................ 411(T)
Poissons ratio .............................96(T)
quenching .......................... 273, 279(F)
shear modulus .............................96(T)
work of propagation of a ductile crack
76(T)
Copper-infiltrated steels,
composition ........................... 407(T)
Copper infiltration, cost, relative ....... 413(F)
Copper-nickel, first-order residual stress
parameters ............................. 335(T)
Copper steels, composition .............. 407(T)
Copper-zinc alloys, macro and micro residual
stress relaxation ............................55
Core hardness, effect on residual stress of
carburized-and-hardened steels ......... 442
Correction in stress c(Z1) ................... 108
Corrosion, nitrided surfaces ................. 215
Corrosion fatigue ............................. 100
resistance improved by shot peening ..... 354
Corrosion pittings, and crack initiation in
seawater .....................................34
Cost in Euro for production
1. 4462 (X 2 CrNiMoN 225-3)
cost comparison of 5000L vessel .. 356(F)
1. 4539
1. 4541 (X 6 CrNiTi 1810)
cost comparison of 5000 L vessel 356(F)
1. 4571 (X 6 CrNiTi 17122)
Crack arrest ...... 35, 47, 48, 49(F), 50(F), 52
and deep rolling ...................... 4950(F)
and shot peening .................. 353, 354(F)
in steels .........................................75
Crack growth equation ........................29
Crack growth rate ..............................29
Crack-indicator method ................. 137(F)
Crack initiation ...............27, 28, 3334(F)

carburized steels .............196197, 199(F)
at notch root, and crack arrest ..........4748
quenched-and-tempered steel
in seawater ........................... 34(F)
in steels .........................................75
Crack propagation ................... 16, 27, 28
carburized steels .......................196197
in high-strength steels ............... 4849(F)
medium-strength steels ................... 49(F)
residual stresses influence ........... 3436(F)
Crack propagation rate (velocity), work
hardening effect and
residual stresses ........................ 35(F)
Crack tip plastic zone ..........................28
Crack-tip state ................................. 131
Crack-Nicholson method .................... 376
Crankshafts
induction hardening and
grinding ....................... 244246(F)
roller burnishing effect on fatigue
strength .......................... 19, 20(F)
shot peening effect on fatigue life ......20(T)
C-ray stress corrosion test specimens,
practices for making and using ........ 110
Creep ............ 4, 5859, 60(F), 68, 213, 251
in casting models ............................ 364
high-temperature ..............................60
modeling distortion of bearing rings ..... 155
thermally controlled ........................ 216
Creep constitutive equation ................ 302
Creep function ................................. 302
Creep-hardening parameter ................ 302
Creep-resistant austenitic steel, stress relief
and yield strength .............. 251, 254(F)
Creep strain ..............................301302
in strip continuous casting by twin-roll
method .................................. 384
Creep strain limit ................. 5859, 60(F)
Creep strain rate ........................ 302, 374
C-ring test ............................... 91, 94(F)
Critical compressive loading
stress ................................ 6162(F)
Critical loading amplitude, cyclic
deformation .................................66
Critical stress intensity factor ................30
Critical temperature of brittleness ..........76
Cross correlation method ................... 122
Crossland criterion ........................ 17, 19
Cross slip .........................................58
Cryogenic cooling ................ 331344(F,T)
carbide formation ................. 340341(F)
influence on dimensional stability
of steels ...........336339(F,T), 340(F)
influence on residual stresses of steels ... 336
influence on structure and substructure
of steels .................... 339343(F,T)
purpose ....................................... 331
residual stresses evaluated
after ........................ 333336(F,T)
residual stresses induced ...... 331336(F,T)
and retained austenite in carburized
steels .......................... 447, 448(T)
steels ................................ 331, 332(F)
thermal range ................................ 331
Index / 471
Crystallites ..................................... 333
Crystal structure .................249(F), 250(F)
Cube, Kondratiev form factor, K ........ 323(T)
Cubic boron nitride (CBN) grinding
wheel ....................................... 150
Cups, deep drawing ....... 147(F), 148149(T)
Curie temperature ............................... 3
Curie transition temperature .... 203, 205(F)
Current surface stresses, vs. time ..... 315(F)
Cut-out indicator method .............132133
Cutting .......................................... 150
and distortion .......................... 155, 166
Cyclic bending tests, and residual stress
relaxation ........................... 66, 67(F)
Cyclic creep ............................ 6364, 66
Cyclic deformation .................... 2829(F)
for inducing residual stress relaxation ....54,
6368(F,T)
residual stresses influence ........... 3133(F)
residual stress relaxation ......... 6368(F,T)
Cyclic hardening ................................27
Cyclic hardening material ....................20
Cyclic loading ........................22(F), 23(F)
and fatigue life ...........................2021
Cyclic plasticity
in low-strength steels ................ 3940(F)
in medium-strength steels ....................41
Cyclic softening ................ 27, 31(F), 32(F)
Cyclic softening material ......................20
Cyclic stress-strain curve, of plain carbon
steels ................................ 2829(F)
Cyclic surface yield strength ........ 66, 68(T)
Cyclic work softening ............ 6364(F), 66
Cyclic yield strength, and fatigue
behavior ................................. 32(F)
Cylinders (cylindrical components)
of carbon steel, quenched, computer
simulation .......... 288289(F), 291(F)
computer simulation of metallothermo-mechanics during
quenching .............. 296, 302(F), 305
destructive residual-stress measurement
procedures .......................... 102(T)
eccentric hollow, distortion of .. 165, 166(F)
heat treatment distortion ...... 159, 162(F,T),
163(F), 164(T)
hollow distortion ..... 161165(F,T), 166(F),
167(F)
hollow with different cross sections,
distortion of .................. 165, 167(F)
residual stresses/distortion in scanning
induction hardening .................. 290,
293(F), 294(F)
solid, residual stress measurement ........ 108
D
DAM. See Discrete atom method.
Damage parameters ............................45
Dang Van criterion .................... 17(F), 19
Dang Van diagram ......................... 18(F)
DANTE (Deformation Control Technology,
Inc.), software program ..............9596

472 / Index
Data processing procedure, accuracy

evaluation ............................. 129(F)
Debinding ................................. 412, 419
Debye specific heat function ............. 3, 335
Debye temperature ..................3, 335, 336
Decarbonization ............................... 278
Decarburization .........................254256
atmosphere influence ....................... 402
and distortion ..........................255256
Decohesion .......................................78
Decreasing plasticity under tension .........72
Deep cooling ................................... 195
Deep drawing ........................ 141142(F)
cups ....................... 147(F), 148149(T)
Deep freezing process, and distortion ...... 178
Deep rolling ......................................55
and crack arrest ....................... 4950(F)
and cyclic deformation behavior ............33
and fatigue strength ...........................52
generating compressive macro residual
stresses ....................................54
microcrack initiation in medium-strength
steels ......................................42
residual stress influence on cyclic
deformation behavior .......31(F), 32(F)
S-N curves and bending
fatigue strength ...................... 38(F)
S-N curves and residual
stresses effect ........................ 37(F)
Deflection formula derivation ....... 93, 95(F)
Deflection methods ........... 8998(F,T), 120
to determine residual stress in
coating ............................118119
Deflection of Almen strip ......................91
Deformation. See also Distortion.
adiabatic ......................................... 3
characteristics of different modes ........ 5(T)
interaction with time, temperature and
microstructure ........................ 3(F)
plastic ............................................ 5
Deformation-induced
transformation ......... 198199, 200(F),
201(F)
Deformation zone
and drawing .................................. 145
shape of ....................................... 143
Deformed material, interplanar spacings and
residual stresses .......................... 333
DEFORM-HT (SFTC Company), computer
program ....... 289290, 291(F), 302303
Delayed fracture ................................72
in steels ............................. 7379(F,T)
Delayed quench ............................... 456
Dendrite arm spacing, of metal powders/
substrates for atomization
techniques ............................. 400(F)
Densification process ......................... 408
Density
of aluminum alloy ...................... 378(T)
and expansion coefficient .......... 3(T), 4(F)
Densmix powders, warm compacted ....... 413
Deviatoric stress ...................288, 297, 299
Deviatoric stress tensor ................ 202, 288
DHR. See Direct hot quench.
Diamond-like carbon (DLC) coating .......25
Diamond pyramid hardness test .......... 223
Diamonds, artificial, high-speed steel dies for
growing, service life ..................... 326
Die pressing .................................... 404

Differential equation of
heat conductivity ........................ 312
Diffraction angle .............................. 333
of first-order residual stresses of metallic
materials ............................ 335(T)
Diffraction line ................................ 342
physical width of .................. 334, 335(F)
Diffractogram ........................... 333, 334
Diffractometers
to measure residual stresses in coatings
portable ................................. 122
Diffusion, during carburizing ...... 199, 202(F)
Diffusion-alloyed powder ................... 399
Diffusion coefficient of carbon ... 199, 202(F)
Diffusion-controlled dislocation creep ......60
Diffusion creep ............................. 54, 58
Diffusion equation
for carbon content ........................... 298
for carbon content during carburizing ... 285
Diffusion zone .....................211, 214, 215
Dilatational component .................. 6(F), 7
Dilation due to solidification, of aluminum
alloy .................................... 378(T)
Dimensional stability ...........................16
of high-alloy tool steels ................ 343(F)
of steels, cryogenic cooling
influence ..........336339(F,T), 340(F)
Dimensional tolerance limits ............... 186
Direct chill (DC) casting ...............363364
finite-element analysis predictions of
defects ............................363364
Direct hot quench (DHR), of
hypoid gear ...................... 454, 455(F)
Direct-metal deposition ...................... 419
Direct metal laser sintering (DMLS) ..... 413
Direct method ................................. 334
Direct quenching .............................. 189
Direct stretching (plastic deformation), after
shot peening .................... 151, 153(F)
Dirty steels ..................................... 254
Discrete atom method (DAM) .............. 419
Disk, of carbon steel, quenched, computer
simulation ...................... 288289(F),
290(F), 291(F)
Disk deflection method ...................... 120
Disk deflection test ........................ 119(F)
Dislocation
and crack initiation ......................3334
in quenched surface layer .................. 326
Dislocation arrangement, and thermally
induced stresses from annealing .........58
Dislocation-core diffusion-controlled climb
by edge dislocations ......................58
Dislocation creep ........................... 54, 58
Dislocation density ...................... 342, 343
of high-alloy tool steels ................ 343(F)
Dislocation movement .........................57
Dislocation pinning ........................6667
Dislocation slip .............................5468
stable obstacles to .............................59
Dispersion hardening, niobium alloying
effect ....................................... 325
Dispersoids ..................................... 414
Displacement rate vector ..............376377
Displacement vector, of invariant-plane
strain .......................................... 5
Disruption cracks ...............................72

Dissection .........................................89
Dissociated ammonia hydrogen
atmosphere ............................... 403
Distance from neutral axis, symbol and
units of ...................................93(T)
Distortion See also Deformation. .............27
asymmetrical components .... 165, 167(F,T),
168(F)
asymmetry increasing tendency for
component .............................. 159
and atmosphere control .....256257, 259(T)
ball screw ........................... 166, 168(F)
of bearing ring after
quenching ................. 315317(F,T)
of bearing rings, prediction
during hardening ......... 317320(F,T)
bolt ........................................ 181(F)
carbon content effect ............. 249, 250(F)
of carbon steel from quenching .......... 275,
281(F), 282(F), 283(F), 284(F), 285(F)
of carburized-and-hardened
steels .......................... 452457(F)
of carburized bushing due to high-rate
quenching ................. 320321(F,T)
of carburized workpiece ............... 176(T),
178179(F)
castings .......................... 167168, 361
causes ......................................... 453
cavity of concave dies ............ 165, 168(F)
of chromium-molybdenum steel ring,
carburized-quenched, computer
simulation .................... 290, 291(F)
clamping apparatus choice ................. 166
cold-rolled steel sheet
stampings ..................... 168169(F)
and component design ........... 252253(F),
255(F), 256(F)
component design as
control method ...................278279
cooling process and
equipment ................. 172177(F,T)
and decarburization ...................255256
and deep freezing ........................... 178
definition ..................................... 452
of eccentric cylinder bushing ..159, 161(F,T)
as effect of process equipment
design ......................... 183186(F)
elastic chuck ....................... 166, 168(F)
after final shaping .............. 151157(F,T)
forgings ....................................... 168
of gears ...... 176(T), 179, 253, 255, 256(F),
307(F), 309, 441442
generating process equipment .......183185
gravity causing .......... 178(F), 179180(F)
heating rate and
uniformity factors ...............170171
heat treating procedure influence .. 165166,
168(F), 169172(F,T), 173(F)
by heat treatment of component
shapes ......... 159161(F,T), 162(F,T),
163(F), 164(F)
high-speed gear wheel ............ 166, 169(F)
hob ......................................170171
hot-rolled steels .............................. 168
hypoeutectoid steels ........................ 168

after induction hardening ....... 238239(F),

242(F), 243
influences of heating rules ....... 171(F), 172
of Japanese sword steel ....308(F), 309310
loading methods contributing to ......178(F),
179182(F,T)
of machine parts ............................. 221
machining process influence ........ 165166,
168(F), 169(F), 254, 256(F)
measurement ................................. 438
metallurgical sources during reheating and
quenching ..............251252, 254(F)
methods to reduce ..............453(F), 454(F)
minimized by press and plug quenching
dies ...................................... 281
of nitrided workpiece ....................... 179
at notch, with quenching ......... 253, 256(F)
one-dimensional ............................. 260
out-of-roundness ........................ 454(F)
and overheating, localized ...........256257
pinion shafts ....................... 255, 257(F)
plug quenching for control of ............. 456
prediction, after quenching ...... 284293(F)
quench ..................................259260
quenchant effect on ............ 174175(F,T)
quench chamber pressure control of vapor
blanket stage ................. 455456(F)
and quenching ..................... 252, 255(F)
quenching, agitation effect ...... 276, 283(F),
284(F), 285(F)
from quenching, cooling characteristics
effects ........ 275278, 281(F), 282(F),
283(F), 284(F), 285(F,T), 286(T)
quenching media effect on
gears ........................... 454456(F)
and quench press machines ... 181182(F,T)
and quench severity ...................279280
of rail steel ......................... 424435(F)
ratchet gear roundness improved by jig
design ............................... 454(F)
reduction and prevention in quenched
parts ..................................... 323
reduction in quenched parts ............... 328
reduction measures for gears .............. 456
and retained austenite content ... 257, 259(T)
of ring ............... 303(F), 305(F), 306, 307
of scanning induction hardened steel
cylinder, computer simulation ...... 290,
293(F), 294(F)
setting method influence ............... 454(F)
shape ....................................259260
of shaped components .........159161(F,T),
162(F,T), 163(F), 164(F)
size ............................................ 260
and stabilization ............................. 178
steel grade selection as
control method ......... 278, 279, 286(F)
steel hardenability influence ........... 453(F)
of steering sector shaft ........... 454455(F)
stop ring ............................ 180(F), 181
and tempering .........177178(F), 283284,
287(F)
transformation plasticity
influence ...................... 456457(F)
two-dimensional ............................. 260
types ........................................... 452
and welding ......................391 to 394(F)
in welds .................................... 7(F,T)

zero value fixed by heat transfer
intensification .......................... 328
DistSIMR (heat treatment simulation
code) ....................................... 155
DMLS. See Direct metal laser sintering.
Domfer process ................................ 399
Double-action tooling system .........401402
Double-exposure technique (DET) .. 112, 135
Double-hard peening, to improve fatigue
strength .................................... 437
Double peening ................................ 449
Double-press/double-sinter, cost,
relative ................................. 413(F)
Downcut milling, residual stresses
generated ................................ 36(F)
Drawing .................................... 99, 141
advantages .................................... 142
bars ........................................ 141(F)
contact length ................................ 143
deformation zone geometry ................ 143
for inducing residual stress relaxation .....54
mean thickness or diameter ................ 143
residual stresses ............................. 143
rod ......................... 143(F), 145146(F)
tube .................................. 145146(F)
wire ............................141, 145146(F)
Drilling .................................... 109, 150
and distortion potential ..................... 183
Driving rods, shot peening effect on fatigue
life ........................................20(T)
Dry bag tooling ................................ 402
Dual shot peening ............................. 347
Ductile-brittle transition temperature,
alloying elements effects ..................82
Dummy gage ................................. 110
Duplex surface treatments, carburizing
and induction heating and
quenching ....... 203204, 205(F), 206(F)
Durability, of high-speed steels ... 343344(F)
Durability till complete failure ...............80
Dyescan fluorescent tracer liquids, for
inspecting shot peening coverage ...... 349
Dynamic deformation ........................ 335
DYNA2D (computer program) .. 156157(F)
E
Eddy currents, and induction
hardening ..................... 242243, 246
Edge dislocations ........................... 58, 71
Edge effect .........271, 273274, 276, 279(F),
280(F), 283(F)
EDM. See Electric discharge machining.
Effective case depth (ECD) ................. 437
and carburized steel residual
stresses .............................. 445(F)
and fatigue strength of carburized
steels ................................ 446(F)
effect .......................... 451, 452(F)
Index / 473
of scanning induction hardened cylinder,
294(F)
Effective diffusion coefficient .................71
Effective stress intensity range
(DKeff) .............................. 3536(F)
and crack propagation rates
with welding .................... 3536(F)
quantitative determination ...................36
Eight-noded quadratic isoparametric
elements, in casting thermal model ... 364
Elastic chuck, distortion of ........ 166, 168(F)
Elastic compression, regime in simple slab
casting ................................. 362(F)
Elastic constants ....................... 95, 96(T)
Elastic crack tip, as mechanical driving force
coefficient ................................ 6(T)
Elastic distortion .............................. 333
Elastic modulus .................................. 3
of bearing ring being hardened ....... 318(F)
effect on residual stress development
in steels ................................ 3(T)
Elastic-plastic time-independent model,
semicontinuous casting calculated
stress distribution ............... 378, 381(F)
Elastic recovery, after straightening ....... 151,
154(F)
Elastic residual strains (Ee) ...................54
Elastic strain ... 101, 111112, 113, 125, 285,
298299, 373
Elastic strain component .................... 217
Elastic strain rate ....................... 202, 312
Elastic stress-strain relation of the
mixture .................................... 298
Elastic tension, regime in simple slab
casting ................................. 362(F)
Elastic theory ....................................12
Elastic theory problem solution for regions
with the wedge cutouts ................ 130
Elastic-viscoplastic constitutive model ... 378
semicontinuous casting calculated stress
distribution ................... 378, 381(F)
Elastic-viscoplastic solids .................... 372
Elastic-viscoplastic stress-strain matrix .. 376
Elastoplastic behavior ....................... 363
Elasto-plastic rolling contact,
of rail steel ...................... 434435(F)
Elastoviscoplastic behavior ................. 363
Elastoviscoplastic constitutive equation .. 363
Elastoviscoplastic stress model ............. 364
finite-element analysis ...................... 365
Electrical conductance ............. 259, 262(F)
measurement ....................... 269, 273(F)
Electrical discharge machining ..............91
Electrically activated pressure
sintering ................................... 408
Electrical-resistance strain gages ... 101, 102,
103, 104
for strain measurement ............... 109, 110
Electric discharge machining (EDM) ..... 147
Electrochemical etching, rotate bending
fatigue strength ....................... 450(F)
Electrochemical metal corrosion ............82
Electrodischarge machining ................ 405
Electrodischarge milling (EDM) ........... 109
Electroless deposition ........................ 118

474 / Index
Electrolytic etching method, to measure

Electrolytic polishing ......................... 109
Electromagnetic-acoustic transducers ......91
Electromagnetic (EM) casting, finite-element
analysis predictions of defects ......... 363
Electron beam, mechanical, thermal, or
structural origins
of residual stress .......................13(T)
Electron beam hardening ................... 220
Electron beam heating-quenching,
deformation tendency ................ 170(T)
Electron beam welding ...................... 405
of maraging steel ..................... 6061(F)
Electronic digital speckle
interferometry .........130, 132, 135, 136
Electronic packaging ...........................25
Electronic specific heat coefficient ........... 3
Electron probe micro-analysis (EPMA), to
calculate time-dependent distribution of
diffused carbon content in rings ...304(F),
306
Electroplating ............................ 118, 120
Electropolishing .................. 104105, 109
combined with x-ray diffraction .......... 109
Electroslag heating (ESH)
tundish process .......................... 405
Enameling ........................................12
End effect ............. 252, 255(F), 276, 283(F)
Endothermal heat of solution ................70
Endothermic furnace atmosphere ......... 403
composition .............................. 404(T)
Endothermic nitrogen-base
atmosphere ............................... 407
Endothermic nitrogen/methanol
atmosphere ............................... 407
Endurance limit ........................ 28, 30(F)
of carburized-and-hardened steels, 442,
443(F)
case depth influence in carburized
steels ................................ 446(F)
hard shot peening effect on carburized
gears ................................. 452(F)
Endurance ratio .................................17
ENSPED program in Europe ................25
Enthalpy ........................................... 4
Entropy ............................................ 4
Epsilon carbides ............................... 342
e-carbonitrides ................................ 213
e-nitrides ........................................ 213
Equilibrium
equations of ............................ 105, 106
of loading mean stresses vs. residual
stresses ....................................28
Equivalent creep strain ................ 288, 374
Equivalent loading stress .....................62
Equivalent octahedral stress .................21
Equivalent stress .............. 144(F), 145, 146
ERRI. See European Rail Research Institute.
Eta (g) carbides ............................... 342
Etching .......................................... 439
Euler backward method ..................... 304
Eulers equation ............................... 127
European Aeronautic, Defense and Space
(EADS) ......................................11
European Committee for Standardization

(CEN), draft standard, residual stress
measurement of rail steel ........... 431(F)
European Network of Surface and Prestress
Engineering and Design (ENSPED) ..11
European Rail Research Institute
(ERRI) .................................... 431
report on rail steel residual stresses ...... 431
Eutectoid, transformation-start temperature
stress effects on .......................... 6(T)
Eutectoid tool steels, distortion of ....171172
Exfoliation ...................................... 215
Exothermic gas atmosphere ................ 403
Expansion coefficient, and density .... 3(T), 4(F)
Explosion welding, to apply cladding to
nuclear reactor casing ............... 137(F)
Explosive forming ............................ 402
Explosive Newtonian wetting ..... 259, 263(F)
Explosive-type wetting ............. 259, 262(F)
Extensometer .................................. 102
Extrusion ....... 141, 142(F), 146147(F), 399
advantages .................................... 142
contact length ................................ 143
mean thickness or diameter ................ 143
open-die ............................ 146147(F)
of rods .................................... 147(F)
F
Fabrication, residual stress introduction ...... 7
Face-centered cubic (fcc) materials, residual
stresses ..................................... 333
Facing, residual stresses on
stainless steels ........................ 109(F)
Failure modes, castings .................. 361(F)
Fatigue ..............................100, 150, 152
carburized gears and
shot peening effect .......... 353, 356(F)
crack initiation ........................ 3334(F)
crack propagation .................... 3436(F)
critical layer thickness approach
for steel ............................... 18(F)
cyclic deformation in steels ........ 2829(F)
cyclic plastic deformation ....................27
cyclic softening of plain
carbon steels ......................... 29(F)
determined by residual stress interaction
with cyclic loading stresses ............28
local fatigue strength ................ 4547(F)
macrocrack propagation
in steels ..................2829(F), 30(F)
mechanisms of residual stress formation ..27
microcrack initiation in steels ...... 28(F), 29
modeling residual stress
influence ......................... 4350(F)
nitrided steel .................... 214216(F,T)
phase-hardening mechanism .................29
quenched-and-tempered steels .......... 29(F)
residual stress alterations of behavior leading
to ...........................................27
residual stress on steel surface taken into
account ................................ 18(F)
residual stress stability .......................27

of steels ................................ 2831(F)
unstable crack propagation and failure in
steels .................................. 28(F)
Fatigue cracking
and shot peening application .... 353, 354(F)
and subzero cooling ........................ 195
Fatigue damage parameters .............3031
Fatigue ductility coefficient ...................30
Fatigue ductility exponent ....................30
Fatigue failure, from repeated stressing in
carburized steels ................ 450, 451(F)
Fatigue life .................................. 19, 28
calculation while taking residual stresses into
account ................................ 21(F)
of cold-formed products .................... 144
deep rolling effect ......................... 37(F)
distribution of ..................................22
grinding effect ........................ 3738(F)
heat treatment effect on carburized
steels .......................... 449, 450(F)
of high-strength steel after shot peening and
grinding .......................... 3839(F)
increase for mechanical components from
shot peening .............................20
prediction calculations affected by residual
stress ............................. 1619(F)
prediction of .................. 22, 23(F), 24(F)
of shot-peened steel, Dang Van
criterion ............................... 18(F)
shot peening effect on stainless steel .......16
Fatigue life distribution .......................21
Fatigue limit
hardenability influence on carburized-andhardened steels .............. 442, 443(F)
in high-strength steels ............... 48(F), 49
of medium-strength steels ....................49
nitrided layer ................................. 216
Fatigue notch factor
of high-strength steels ........................51
of low-strength steel ..........................51
of medium-strength steel .....................51
Fatigue slip bands ..................... 28(F), 29
Fatigue strength ..................... 1415(F,T)
of carburized-and-hardened steels ........ 441
case depth effect on
carburized steels ................... 446(F)
gears with induction hardened
flanks ..............................241242
effect ....................450451, 452(F)
of gear wheels, and tangential residual
stresses .................................. 238
of induction-hardened machine
parts ........................... 243244(F)
of induction-hardened steel ................ 224
after induction hardening ............ 239, 246
prediction taking residual stress into
account ....................................24
residual stress effect ................. 1619(F)
surface finish and strain hardening
effects ................................. 19(F)
Fatigue strength coefficient ...................30
Fatigue strength diagrams ................ 30(F)
Fatigue strength exponent ....................30
Fatigue testing

carburized steels ................... 195, 199(F)

induction-hardened steels .............. 244(F)
FATIGUE3D (computer program) .... 21, 22
Ferric carbide (Fe 3C), dissolving, influence
on residual stress .................. 214(T)
Ferric nitride (Fe 4N), influence on residual
stress ................................ 214(T)
Ferrite
acicular ...................................... 4(F)
acicular, shape change ..................... 5(T)
acicular, transformation-start temperature,
stress effects on ....................... 6(F)
allotriomorphic ............................. 4(F)
allotriomorphic, shape change ........... 5(T)
expansion coefficient ...................... 4(F)
heat capacity and magnetic
properties ........................... 34(F)
idiomorphic ................................. 4(F)
idiomorphic, shape change ............... 5(T)
massive ...................................... 4(F)
nitrocarburizing of,
deformation tendency ............. 170(T)
transformation of mild
steel plate ............................. 91(F)
thermal expansion coefficient ............ 7(F)
Widmanstatten .............................. 4(F)
Widmanstatten, deformation systems
associated with transformations .... 5(T)
Widmanstatten, shape change ............ 5(T)
Widmanstatten, transformation-start
temperature, stress effects on ....... 6(F)
Ferrite, allotriomorphic, shape change 5(T)
Ferrite, idiomorphic, shape change ...... 5(T)
Ferrite, nitrocarburizing of, deformation
tendency ............................... 170(T)
Ferrite, Widmanstatten, deformation
systems associated
with transformations .................. 5(T)
Ferrite, Widmanstatten, shape change .. 5(T)
Ferrite, Widmanstatten, transformation-start
temperature stress effects on ...........6(F)
Ferrite, 0. 1N, influence
on residual stress ..................... 214(T)
Ferrite-pearlite steels
Barkhausen noise analysis for stress
measurement ........................... 114
hydrogen embrittlement ......................72
hydrogen states ................................72
Ferrite carbides, atomic volume in
ferrous alloys ......................... 250(T)
Ferritic steels, floccules not observed ........73
Ferrochrome ................................... 399
Ferromagnetics, for inducing residual stress
relaxation ....................................54
Ferromanganese .............................. 399
Fiat .................................................11
Ficks second law ............................. 199
Film-boiling (FB) process ........ 258260(F),
261(F), 262(F), 264265, 268, 269,
270271, 312
Film cooling .......................... 258, 260(F)
Final instability effect ..........................30
Final part shaping
asymmetry influence .. 165, 167(F,T), 168(F)
auxiliary equipment effects and loading

methods ...........178(F), 179182(F,T)
axle components ......... 159161, 163(F,T),
164(F)
blank manufacturing methods ... 167169(F)
cavity of concave dies ............ 165, 168(F)
chemical heat treatment ............... 176(T),
178179(F)
cooling process and equipment
influence ................... 172177(F,T)
cutting ......................................... 166
cylinders ...... 159, 162(F,T), 163(F), 164(T)
cylinders, hollow ..... 161165(F,T), 166(F),
167(F)
factors and relationships between heat
treatment and distortion .... 159, 160(F)
heat treating procedure influence .. 165166,
168(F), 169172(F,T), 173(F)
heat treatment distortion ......159161(F,T),
162(F,T), 163(F), 164(F)
machining process influence ........ 165166,
168(F), 169(F)
screw ................................ 161, 164(F)
stress relieving ............................... 166
Final shaping
distortion after .................. 151157(F,T)
prior to heat treatment ......... 150158(F,T)
Final steady-state temperature (Ta) ....... 367
Finding coefficients ........................... 130
Fine-grain steels ............................... 325
Finishing, in powder metallurgy
processing ............................. 397(F)
Finite cylinder, height: z, Kondratiev form
factor, K ............................... 323(T)
Finite-difference two-stage Crank-Nicolson
predictor-corrector method ........... 364
Finite-element method (FEM), (finiteelement analysis) ........................ 144
for brazing, residual stress prediction ... 394,
395, 396(F)
to calculate residual stresses after shot
peening ................................. 151
calculation of time variation of internal
stresses ......................234, 235, 236
of casting defects ............................ 363
casting elastoviscoplastic stress model .. 365
for casting modeling ........... 364369(F,T)
centrifugal casting modeling ....... 379, 380,
383(F)
determining stabilized residual stress after
fatigue loading ...........................19
formulation for unsteady-state casting ... 376
heat-flow problem in casting .............. 364
measuring distortion of ring ..... 290, 292(F)
modeling bending .............. 155, 156157
modeling elasto-plastic rolling contact of rail
steel ........................... 434435(F)
modeling phase transformations ... 296, 300,
301(F), 302, 303, 306(F), 307(F), 309
modeling relaxation behavior of multiaxial
macro residual stresses ........ 6263(F)
for modeling roller straightening ...432433
modeling straightening ........... 156157(F)
predicting distortion and residual stress after
quenching .................... 284, 288(F)
predicting relaxation of residual stresses ..11
quench-crack formation .................... 313
Index / 475
quenching of carbon steel disk
and cylinder ........ 288289(F), 290(F)
quenching simulation for
carburized bushing ......... 320321(F,T)
rail steel deformation by residual
stresses .............................. 424(F)
residual stress distribution modeling ... 14(F)
semicontinuous casting numerical calculation
for temperature ........................ 377
for simulating particulate, compaction .. 413,
416, 417, 419
for simulating thermomechanical problems
in casting .........................374375
software programs .......................... 202
solidification modeling ..................... 372
for straightening, Swedish steel industry
research project ........................ 156
stress/strain prediction in case-hardened
parts ........................... 199, 201(F)
for strip continuous casting by twin-roll
method model ........ 383385, 386(F),
387(F), 388(F), 389(F)
thermoplasticity of carburized bushing .. 320
through-thickness temperature variation of
plate ........................... 223224(F)
unsteady-state heat-conduction problem of
carburized bushing .................... 320
warpage prediction of steel shafts with
keyway ..................289290, 291(F)
Finite plate, dimensions of sizes: L1, L2, L3,
Kondratiev form factor, K ............. 323(T)
Fisheye fracture ......................... 442, 446
Fishtailing ............................... 89(F), 93
Fixture hardening ............................ 185
Fixtures
and distortion ...................... 172, 173(F)
for hardening of ring-shaped bodies .....186(F)
Flame cutting .................................. 109
Flame hardening .............................. 220
Flame heating-surface quenching,
deformation tendency ................ 170(F)
Flat rolling ..................................... 143
Flocculation .................................7273
prevention in steels ...........................73
Floccules ................................. 7273(F)
prevention of formation in steels ...........73
Flow rule for plastic strain rate .....299300
Flow speed, of quenching media and
carburizing ................................ 445
Flow stress ...................................... 299
Fluid-flow analysis, for solidification
modeling, benefits listed ............ 363(F)
Fluidity coefficient ............................ 366
Fluidized-bed quenching .......... 263, 267(F)
Foaming, of quenchants ...................... 259
Fog quenching ................................. 282
equipment .................................... 174
Foreman equation, crack propagation taking
mean stresses into account ................35
FORGE (software program) ............... 202
Forging ..... 99, 109, 141142, 148149(F,T),
399
advantages .................................... 142
process conditions ....................... 148(T)

476 / Index
Forging (continued)
workpiece geometry .................... 148(T)
Forging reduction ratio ...................... 254
Forgings
cylindrical steel neutron diffraction
methods ................................. 113
distortion ..................................... 168
steel rocket case, neutron diffraction
methods ................................. 113
Forming ............................ 141149(F,T)
bending ............. 142(F), 143(F), 144145
deep drawing ............ 141142(F), 147(F),
148149(T)
drawing ... 141(F), 142, 143(F), 145146(F)
electric discharge machining .............. 147
extrusion ...... 141, 142(F), 143, 146147(F)
forging ...............141142, 148149(F,T)
press braking ....................... 147148(F)
pressing .................................141142
rolling ............ 141, 142, 143, 146148(F)
Forward solution ................................91
Fourier coefficients ........................... 334
Fourier law ............................... 199, 375
Fourier number ............................... 321
generalized ................................... 322
Fouriers heat conduction law ............. 298
Fredholm integral equation, method of the
first type ......................... 126128(F)
Fredholm integral equation (computation
experiment) residual stress
determination .................. 126, 127(F)
Free quenching ................................ 257
French specimens ...... 338(F), 339(F), 340(F)
French Technical Center for Mechanical
Industry (CETIM) .............. 15, 16, 17
Fretting corrosion ...................... 215, 216
Fretting fatigue, shot peening to promote
resistance .................................. 354
Friction ............................................16
Fringe effect .................................... 134
Function smoothness measure ............. 127
Furnace hardening, carbon content
limits ................................... 256(T)
Furnaces
atmosphere-hardening ...................... 257
austenitizing ............................155156
batch-type .................................... 402
belt ............................................ 257
combustion-heated, and distortion ........ 172
continuous, and distortion ................. 172
continuous gas-carburizing ................ 440
continuous mesh-belt sintering ........ 402(F)
design and operation to control heat treating
problems ................................ 257
direct-fired gas ............................... 255
distortion of bearing rings ...........155156
hardening ..................................... 257
heat irradiating areas in ratio to the size of
the batch ................................ 184
for hot isostatic compaction ........... 409(F)
insulation ..................................... 184
muffle-type continuous production ... 402(F)
passing-type, and distortion ............... 172
pit-type and distortion potential ........... 184
pusher ......................................... 402
roller hearth .................................. 402
shaker hearth ................................. 257
sintering .................. 402403(F), 404(T)

vacuum ....................................... 402
vacuum, and distortion potential .......... 184
walking-beam ................................ 402
zones ...................................... 404(T)
G
Galerkin method .............................. 364
Galerkin type formulation .................. 376
Galvanic corrosion, stainless steel P/M
parts ..................................403404
Galvanopairs, in stress-corrosion
cracking ......................................79
Gamma ferrite (austenitic), first-order
Gas atomization ............... 400(F), 412, 414
Gas blending ................................... 261
Gas carburizing ........ 189, 190, 194, 197(F),
443(F), 450(F), 451(F), 453(F)
carbon potential control .................... 444
diffusion ............................ 199, 202(F)
and retained austenite content ............. 447
Gas dissolution in metal melt .............. 400
Gaseous nitrocarburizing ............... 215(F)
Gas jet and fog jet cooling
equipment ...................... 172, 174(F)
Gas lances ...................................... 184
Gas-pipe lines steels, stress-corrosion
cracking ............................. 8182(F)
Gas quenching ...... 256(T), 257, 261, 265(F),
266(F), 271
distortion minimizing ....................... 185
equipment .......................... 172174(F)
high-pressure ....................... 172174(F)
Gas turbine components, shot peening and
distortion .................................. 156
Gauss (G) distribution ....................... 334
Gaussian method .............................. 122
Gausss theorem ........................ 375, 376
Gear box, shot peening effect on fatigue
life ........................................20(T)
Gears
automotive ............................... 453(F)
automotive, pressure-control
quenching .................... 455456(F)
carburized, and shot peening effect on
fatigue ......................... 353, 356(F)
carburized, axial stress distribution during
quenching .................... 440441(F)
carburized, shotblasting and
peening ....................... 451452(F)
carburized variation of residual
stresses ........................ 439, 440(F)
case hardening, transformations and stress
evolution model ... 202203(F), 204(F)
differential ring .......................... 453(F)
distortion ..... 253, 255, 256(F), 307(F), 309
distortion after carburizing and
quenching .................... 176(T), 179
dual-frequency induction hardening,
307309(F)
gas-carburized and direct-quenched, residual
stress .................................... 440
grain-boundary oxidation effect on fatigue

strength .................450451, 452(F)
hypoid ring ............................... 453(F)
hypoid ring, direct hot quench
(DHR) ......................... 454, 455(F)
induction-hardened and carburized residual
stresses ........................ 440, 441(T)
loading methods ................179(F), 180(F)
manufacture of .................. 437457(F,T)
quenched in quench press
machine .................... 181182(F,T)
quenching .................................... 257
ratchet ..................................... 454(F)
residual stresses in ............. 439441(F,T)
ring .................................. 255, 256(F)
setting methods effect on distortion .. 454(F)
shot peening and C.A.S.E. process ...... 350,
351, 352(T), 354(F)
shot peening application ......... 353, 355(F)
steel grade and hardenability
influence ..........441442(F,T), 443(F)
stress profile .............................. 354(F)
transmission spur ........................ 453(F)
Gears, modulus, m5 to 8 mm,
advantages ................................ 327(T)
former steel and process ............... 327(T)
new steel and process .................. 327(T)
Gear wheels
bending fatigue strength when induction
hardened ............................ 244(F)
dual-frequency induction hardening,
307309(F)
fatigue strength after induction
hardening ..................... 243244(F)
induction hardening ........ 226, 227231(F),
236239(F), 241242(F)
shot peening ...........................347348
temperature variations during cooling after
induction hardening ......... 230231(F)
Generatrix ...................................... 318
Gerber parabola ............................ 30(F)
Germanium, alloying element effect on crack
growth .......................................80
German steels. See Steels, series and classes:
1023, 1045, C80, and CK01.
Gibbs free energy .......................297298
Glassy alloys ................................... 400
Glide ...............................................58
Gliding, irreversible .............................29
Global momentum balance in combined
domain .................................... 375
Glow discharge spectroscopy ............... 216
Goodman approximation .. 30(F), 43, 44, 45,
48, 49, 50, 51
Goodman endurance diagrams .... 1415, 17
Goodman relationship .................... 16, 17
Grade, influence on steel gear
stresses ...............441442(F,T), 443(F)
Gradient of liquidus line, of aluminum
alloy .................................... 378(T)
Gradient of solidus line, of aluminum
alloy .................................... 378(T)
Grain boundaries, and crack
initiation ................................3334
Grain-boundary glide ..........................58

Grain-boundary oxidation
in carburized-and-hardened steels ....... 438,
448449(F), 450(F)
of carburized steels ......................... 446
effect on fatigue strength of gear teeth
450451, 452(F)
Grain-boundary sliding .......................54
Grain growth, and sintering ................. 402
Grain refinement .............................. 189
Grain size, and stress-corrosion cracking
resistance ....................................81
Grain slips ........................................15
GRANTAS (Komatsu Company) computer
program ................................... 303
Gravity line method .......................... 122
Gray iron
residual stresses in castings with
solidification time ................. 362(F)
three-bar frame casting model ............. 362
Green density .................................. 401
Greenwood-Johnson transformation ..... 320
Grids, as distortion sources ............184185
Grinding 55, 57(F), 101, 109, 150(F), 151(F)
carburized steels ................... 198, 199(F)
costs dependent on overmeasures
knowledge .............................. 183
and distortion .......................... 152, 156
drifting ................... ................... 198
effect on local fatigue strength and crack
arrest, with shot peening ........... 48(F)
factors causing residual stresses .......... 150
and induction hardening ...220, 244246(F)
with milling and shot peening, S-N curves
and residual stress effect ...... 3738(F)
power ..................................... 150(F)
and quenching in a hardening press ...... 185
residual stresses in austenitic
stainless steels ..................... 109(F)
S-N curves and residual stress effect in highstrength steel .................... 3839(F)
S-N curves and residual
stresses effect ................37(F), 38(F)
wheels ......................................... 150
Grossman factor .......................... 176(T)
Grossman n number (H value) .. 257, 259(T),
262, 265
definition ..................................... 265
Grossman Quench Severity factor ........ 174
Gunnert method ........................ 102, 103
measurement ....................... 102(T)
Gun splat atomization ................... 400(F)
Gyaku-sori (reverse bending), of Japanese
sword ............................. 309310(F)
H
Habit plane ........................................ 5
of invariant-plane strain ....................... 5
Habit plane indices ....................... 5(T), 6
Haigh diagram, (fatigue strength) .........14,
17(F), 30(F), 4345(F), 4748, 52
Half-width values ........................... 61(F)
Hammer peening ................................12

Hamon (border of martensite structure), of
Japanese sword ................. 308(F), 309
Hardenability
effect on fatigue limit and residual
stresses ........................ 442, 443(F)
grade variation effect on carburized
gears ........................ 441442(F,T)
influence on gear stresses .....441442(F,T),
443(F)
Hardenable steel, tempering ...... 284, 287(F)
Hardened steels, micro residual stresses ....63
Hardened surface layer depth ............. 327
Hardening
of carburized components ........ 189, 190(F)
and distortion ............ 153, 154(F), 155(F)
mathematical model ........................ 315
quench-crack formation
prediction .................. 315317(F,T)
of ring-form workpieces ............... 186(F)
thermal stress state calculation ............ 315
Hardening coefficient (H) ........ 300(F), 301
Hardening function ........................... 300
Hardening modulus, of bearing ring being
hardened ............................... 318(F)
Hardening parameter ........................ 299
Hardening regimes ........................... 327
Hardening residual stress state ............ 191
Hardening rule ................................ 299
Hard layer, optimal depth, and maximum
compressive surface stresses ....... 315(F)
Hardness
and bending fatigue strength in low-strength
steel ................................... 40(F)
of carburized-and-hardened steels .... 437(F)
carburized and induction-hardened steel
cylinder ....................... 203, 205(F)
Hard shot peening
austenite transformation in carburized
steels .......................... 448(F), 449
of carburized steels ............... 451152(F)
Harmonic analysis method ....... 334335(F)
Hatched temperature range, after induction
hardening ........................ 240241(F)
Head-hardening, of rail steel ............ 428(F)
HEARTS (HEAt tReaTment Simulation
computer program) ... 202, 288289(F),
290294(F), 296, 301304(F), 306(F),
308
Heat-affected zone (HAZ)
and crack propagation rates .............. 35(F)
of induction-hardened and carburized steel
cylinder ............. 204, 205(F), 206(F)
induction hardening and residual stresses
distribution ............................. 245
and shrinkage due to temperature
distribution ............. 391(F), 392, 394
Heat capacity ........................... 34, 288
Heat conduction ............................... 288
Heat-conduction analysis ..............303304
Heat-conduction equation ............ 284285,
297298
finite element formula of ................... 376
Heat conductivity
Index / 477
differential equation of ..................... 312
Heat-flow analysis, for solidification
Heat flux ............267, 288, 376, 377, 378(F)
in rail steel ......................... 425426(F)
Heating factor ................................. 322
Heat-resistant and superhigh-speed steels
containing 512% Co and 26.5% V
composition .............................. 406(T)
designation ............................... 406(T)
hardness .................................. 406(T)
Weq ....................................... 406(T)
Heat transfer, boundary conditions on outer
surface ..................................... 285
Heat-transfer coefficient .. 264, 323, 375, 377,
378(F)
aluminum alloy pipe centrifugal casting
380382
in casting ........................... 365, 366(F)
definition ..................................... 265
determination of ...............................23
estimation techniques ..........267268, 269,
271(F), 274(F)
nickel-chromium alloy pipe centrifugal
casting ........ 380, 382, 383(F), 384(F)
of quenching media, and carburizing .... 445
quenching of carbon steel disk and cylinder
computer simulation .............. 289(F)
and quenching process modeling ......... 319
Heat-transfer-coefficient curve ...............23
Heat-transfer intensification ................ 312
Heat transfer rates ................. 257, 259(T)
Heat treatment ................................ 109
as cause of residual stress ....................12
distortion caused by procedure ..... 165166,
168(F), 169172(F,T), 173(F)
distortion of component shapes ................
159161(F,T), 162(F,T), 163(F), 164(F)
effect on residual stress
distribution .. 194196, 197(F), 198(F),
199(F)
effect on residual stress of plasma-sprayed
coatings ............................... 16(F)
to eliminate residual stresses .............. 144
media influences on distortion ........ 171(T)
modeling of transformation and stress
evolution processes ................... 202
of powder metallurgy steel
parts ........................... 417418(F)
of sintered P/M steel
components ............... 406408(F,T)
variant A ...... 331, 332(F), 337(F,T), 342(T)
variant B ...331, 332(F), 337(F,T), 338(F,T),
339, 342(F,T), 343(F), 351(T)
variant C ........331, 332(F), 337(T), 342(F),
343(F)
variant D ..... 331, 332(F), 337(F), 338(F,T),
339340, 341(F,T), 342(F,T), 343(F)
variant E ..... 331, 332(F), 341(T), 342(F,T),
343
variant F .........331, 332(F), 342(F), 343(F)
variant G ...331, 332(F), 338(F,T), 339340,
341,(F,T) 342(F,T), 343(F)
variant H .. 331, 332(F), 339, 340(F) 341(T),
342(T)

478 / Index
Heat treatment (continued)

variant ........ 331, 332(F), 341(T), 342(T)
variant K ............ 331, 332(F), 339, 341(T)
variant L ............................ 331, 332(F)
variant M ........... 331, 332(F), 339, 341(T)
variant I ................................... 339(F)
variant II .........................339(F), 340(F)
variant III ........................339(F), 340(F)
variant IV ........................339(F), 340(F)
variant V .........................339(F), 340(F)
HEAt tReaTment Simulation program
(HEARTS) ............... 202, 288289(F),
290294(F), 296, 301304(F),
306(F), 308
Helical springs, shot peening effect on fatigue
life ........................................20(T)
Helium, as quenchant ............... 261, 265(F)
Hertzian pressure, and microcrack initiation
in steels ......................................29
Hertzian pressure formation ..... 151, 153(F)
Hertzs compression ...................... 346(F)
Heteromodulus materials ................... 129
Hexagonal close-packed (hcp) materials,
High-alloy austenitic steels, stress-corrosion
cracking ......................................80
High-alloy steels
delayed failure ............................7879
distortion ........................... 171(T)
High-alloy tool steels. See also High-speed
steels.
cryogenic cooling .... 339, 340(F), 341(F,T),
342(F,T), 343(F)
dimensional stability .................... 343(F)
dislocation density ...................... 343(F)
retained austenite .................. 343(F), 344
High-carbon alloy steels, hardening steelquenching method, arbitrarily
hard matrix ................................ 327
High-carbon low-alloy steels, quenching
temperature effect on mold
hole distortion ........................ 171(T)
High-carbon steel
dimensional variation with time ......... 257,
259(T)
distortion ........................... 171(T)
High cycle fatigue ...............................16
High-energy-rate compaction .............. 402
High-frequency quenching .................. 169
High-pressure consolidation, at ambient
temperature ................................ 402
High-pressure gas quenching .... 172174(F)
High-speed gear wheel, distortion of ..... 166,
169(F)
High-speed steels. See also High-alloy tool
steels.
cryogenic cooling ................. 331, 332(F)
cryogenic cooling effect on
microstresses ....................... 336(F)
cryogenic cooling twice repeated ........ 339,
340(F)
cryogenic heat treatment and dimensional
stability ........................... 338(F,T)
distortion after straightening ........... 156(F)

durability ........................... 343344(F)
flocculation not observed ....................73
heat treatment variants ........... 331, 332(F)
inclusion control in P/M processing ..... 405
microstress evaluation and calculation .. 334,
335(F)
retained austenite decomposition at 823 to
973 K ...................... 337339(F,T)
High-strength materials, thermal
fundamentals of making ..........327328
High-strength spring steel,
crack initiation ..............................34
High-strength steels
delayed fracture ................. 74(F), 7677
effective stress intensity factor range .. 48(F)
Haigh diagram model of residual stress
sensitivity influence ................. 44(F)
local fatigue strength ................ 4647(F)
macro residual stress effect ..................27
macro residual stress effect on fatigue
strength .......................... 5051(F)
S-N curves and residual stress
effect ................3839(F), 4243(F)
stress-corrosion cracking ............ 7982(F)
High-strength structural steel, crack
propagation .............................. 35(F)
High-temperature sintering ................ 399
High-temperature thermo-mechanical
processing ..................................81
High-velocity oxygen fuel (HVOF),
mechanical, thermal, or structural origins
Hob, distortion of ........................170171
Hoganas reduction process ................. 399
Hole-drilling strain gage, to measure residual
stresses in coatings ....................... 118
Hole-drilling strain-gage technique, to
calculate elastic-viscoplastic
stresses ..................... 377, 378, 381(F)
Hole-drilling technique .... 99, 100, 110111,
144
applied to rail steel .......................... 430
for evaluating brazed joint of steel and
cemented carbide ...................... 395
for measuring residual stresses ..122124(T)
Holographic interferometry ... 130, 131, 133,
136
Holographic speckle-interferometry ...... 132
Homologous annealing temperature .... 55(F)
Hookes law ........... 202, 298, 333, 334, 373
Hookes law of elasticity .................... 113
Hoop stress ..............91, 92(F), 96(F), 97(F)
at t8.32 s using a plane-strain
assumption ......................... 368(F)
at t8.3202 s with a traction-free top
surface .............................. 368(F)
assumption ......................... 368(F)
thermal stress history with a traction-free top
surface .............................. 369(F)
Hot consolidation ............................. 414
Hot die pressing, in powder metallurgy
processing ............................. 397(F)
Hot extrusion ............................ 404, 410
Hot isostatic compaction .......... 408409(F)
Hot isostatic pressing (HIP) ... 399, 402, 404,

408409(F), 417
Hot oil jig quench ............................. 453
Hot oil quenching .................257, 449, 453
Hot pressing ........................399, 417, 420
P/M parts ........................... 408409(F)
Hot pressureless sintering, in powder
metallurgy processing ............... 397(F)
Hot-rolled steels, and distortion ............. 168
Hot rolling ........................................90
and distortion ............ 153, 155(F), 156(T)
Hot spraying, in powder metallurgy
processing ............................. 397(F)
Hull plate steel, residual stress
investigations ............................. 136
Hydro Aluminum ...............................11
Hydrogen
atomic ...................................... 78, 81
desorbent amount over the period
of holding ................................77
diffusive-mobile .......................... 78, 79
diffusive-mobile source in quenched steel
structure ..................................77
influence on crack propagation .............71
ionized ..........................................78
lattice solubility ...............................77
as quenchant ....................... 261, 265(F)
solubility in steel ..............................71
states in steel ..............................7172
Hydrogenation ............................... 74(F)
Hydrogen corrosion ............................72
Hydrogen diffusion, in iron, pure
annealing ................................ 71(F)
Hydrogen embrittlement ......... 70, 100, 256
alloying effect on steels .................8284
alloying elements influence ..............76(T)
aspects of theory ..............................72
carbide-forming elements effects ......8283
carbide-non-forming elements ..............83
external .........................................75
ferrite-pearlite steels ..........................72
and hydrogen-induced corrosion
cracking ...................................79
impurities effects ..............................83
impurity elements effects on steels ....8284
internal ..........................................75
martensitic steels ..............................72
mild steels ......................................83
types of .........................................72
Hydrogen furnace atmosphere,
composition ........................... 404(T)
Hydrogen-induced corrosion cracking
(HISCC) ....................................79
main gas-pipe line steels .....................82
Hydrogen-induced stress cracking ........ 110
Hydrogen solubility
in iron ....................................... 70(F)
in steel ..........................................70
Hydrogen stress cracking .....................83
Hydrogen sulfide, effect on stress-corrosion
cracking of steels ...........................81
Hydrogen-sulfide cracking ....................83
Hydrogen traps ..................................71
Hydrostatic pressure

effect on transformation kinetics ... 249, 250,

251(F)
maximum .......................................17
mean ............................................17
Hydrostatic stresses ... 6, 379, 380(F), 381(F)
Hydro-sulfide ....................................70
Hypereutectoid tool steels,
distortion of .........................171172
Hypoeutectoid 0.43% carbon steel, induction
hardening-temperature cycle
variations .............................. 223(F)
Hypoeutectoid steel
distortion of ............................ 168, 171
hot rolling and air cooling .......... 89, 91(F)
Hysteresis loops .................................32
I
I-DEAS computer program ................ 304
Immersion cooling ........ 258, 260(F), 261(F)
Immersion quenching, after induction
hardening .................................. 232
Immersion time quench ..................... 456
Impact bending fatigue test, fatigue limit and
life of carburized notched bar .......... 450
Impact strength, of carburized-and-hardened
gears .............................. 441, 442(F)
Impregnation resins .......................... 406
Inclusions, crack initiation in steels ..........29
Incremental hole-drilling method ..... 11, 12,
13(F), 18
Incremental isotropy ......................... 374
Incubation period
of crack growth in steels ........... 74, 75, 77
of hydrogen embrittlement .............. 74, 75
Indicator crack method .....128132(F), 134
Induction, mechanical, thermal, or structural
Induction-hardened steels,
contact fatigue ............................ 198
Induction hardening .............. 220247(F),
407408
advantages over other procedures ....... 220,
221222
applications .................................. 220
to attain compressive residual stress ..... 437
austenization and residual stresses .... 245(F)
carbon content limits ................... 256(T)
of carburized steel cylinder
model ......... 203204, 205(F), 206(F)
case depths ................................... 220
case hardnesses .............................. 220
characteristics ...................... 221222(F)
compressive stresses .............. 232233(F)
contour hardening ........................... 241
current frequencies .......................... 223
dual-frequency ........... 226, 227(F), 228(F)
dual-frequency distortion ........ 238239(F)
dual-frequency gear tooth hardness
profiles .............................. 229(F)
dual-frequency gear wheel computer
simulation .... 296, 306(F), 307309(F)
dual-frequency martensite volume fractions
simulated ..................228(F), 229(F)
dual-frequency of gear wheels ........... 226,

227(F), 228(F)
dual-frequency, tangential residual
stresses .............................. 238(F)
dual-frequency temperature
distributions ........ 227, 228(F), 229(F)
electromagnetic analysis ......... 226, 227(F)
electrothermomechanical model .......... 226
and finish grinding ................ 224246(F)
flame hardening ............................. 220
gear wheel, computer simulation ........ 296,
306(F), 307309(F)
generator frequency ............... 222(F), 223
grinding ....................................... 220
hardness of surface, roller specimen ....239(F)
homogeneous austenitization
temperature ............................. 227
induction heat treatment .................... 221
Langeots method of mathematical modeling
with quenching .................... 225(F)
machine parts ................................ 220
magnetic field relation to thermomechanical
state of workpiece material ...... 226(F)
martensite distribution and hardness 240(F)
martensite volume-fraction
distribution ......................... 231(F)
mass of specimen effect on heating
time ............................ 224225(F)
materials used ................................ 220
metallothermomechanical analysis ...... 226,
227(F)
metallothermomechanical model ......... 226
numerical calculation flow chart ......... 226,
227(F)
parameters of system ....................... 222
power/frequency regions of heat treatment
applications ........................ 221(F)
and quenching ......................... 227, 228
quenching systems ................ 231232(F)
quenching temperature
distributions .................. 230231(F)
of ring computer simulation .... 296, 303(F),
304(F), 305307(F)
scan hardening ........................... 241(F)
scanning ...................................... 220
shape distortion .................... 238239(F)
shot hardening ............................... 241
single-frequency, distortion ...... 238239(F)
single-frequency, gear tooth hardness
profiles .............................. 229(F)
single-frequency, gear wheels .. 226, 227(F),
228(F), 229(F)
single-frequency, martensite volume
fractions simulated .......228(F), 229(F)
single-frequency, tangential residual
stresses .............................. 238(F)
single-frequency, temperature
distributions ........ 227, 228(F), 229(F)
single-shot .... 220, 222, 223, 225(F), 233(F)
single-shot, time-temperature variations at
various high-frequency
generation .................... 240241(F)
techniques for machine parts .............. 220
temperature cycle variations of surface and
core considering enthalpy of phase
transformations .............. 223, 224(F)
Index / 479
temperature distributions during cooling
process ........................ 230231(F)
time/temperature dependence ... 222231(F)
time/temperature variations in scan
hardening at various speeds and HF
generation power .................. 241(F)
time variation of residual stresses in a
material .................... 232242(F,T)
time variation of stresses in
a material .................. 232242(F,T)
tooth-by-tooth ...................... 253, 256(F)
tooth gap hardening ......................... 241
volume changes after ....................... 222
volume changes and distortion,
gear teeth ........................... 242(F)
water jet quenching ............... 241, 242(F)
Induction heatingsurface quenching,
deformation tendency ................ 170(T)
Industrial heat treatment temperature,
cryogenic cooling influence
on steels ................................... 336
Inelastic strain rate ...............202, 302, 373
Infiltration ...................................... 405
Infinite cylinder ........................... 323(T)
Infinite plate, thickness of 2R, Kondratiev
form factor, K ......................... 323(T)
Inhomogeneous diffusion equation ........ 217
Inhomogeneous microstresses,
in coatings ................................. 118
Inhomogeneous residual stress fields,
determination methods ........ 125137(F)
Initial incompatible strain .................. 125
Initial method .................................. 377
Initial strain .................................... 125
Initial yield stress, of aluminum
alloy .................................... 378(T)
Injection molding .......................400401
Inspection plan, for quality assurance ......13,
14(F)
Institute for Strength Problems/Engineering
Thermophysics Institute, National
Academy of Sciences of Ukraine,
software flow chart ................... 313(F)
Insulation, nonuniformity as cause of
distortion .................................. 184
Integral equation kernel ..................... 126
Intensive cooling .... 312, 316, 317, 319(F), 320
advantages .................................... 317
effect on distortion of quenched
parts ...............................323324
Intensive jet cooling .................... 327, 328
Intensive quenching .......................... 282
Interaction energies ........................ 8, 71
Interatomic bonds, stress-corrosion cracking
role ...........................................80
Interfacial wetting, kinematics .............. 258
Interference fringe .. 120, 135136(F), 137(F)
patterns ............................. 119120(F)
Interferencial-optical methods ............. 132
Interferometer ................................. 135
high sensitivity moire ........................12
Interferometry .....................110, 119, 120
Intergranular effects ......................... 260
Intergranular fatigue crack initiation .... 197
Intergranular fracture ....................... 196

480 / Index
Intermediate file format ..................... 304

Intermetallic materials ................. 25, 400
Internal cracking ................... 193, 196(F)
Internal deflection method of shot
peening ................................ 348(F)
Internet Web site ............................. 100
Interplanar spacing ................ 333, 334(F)
nonhomogeneity degree .................... 334
Interrupted quenching .................280281
definition ..................................... 280
Intracoating spallation ....................... 118
Invariant-plane strain (IPS) ......... 4, 5(T), 6
and bainite growth ..........................78
Inverse heat conduction problem
solution .................................... 268
Inverse residual stress problem ......125126
Inverse solution .................................91
Inverters ........................................ 221
Inv-Probe-2D computer program ........ 268,
289(F)
Ion implantation, deformation
tendency ............................... 170(T)
Ion implantation enhanced vapor phase
deposition, deformation
tendency ............................... 170(T)
Iron. See also Pure Iron.
parts ................................. 411(T)
nitrogen concentration ............ 216(F), 217
Iron-aluminum-carbon alloy, nitriding
Iron-carbon-chromium-manganese steels,
delayed fracture ........................ 77(F)
Iron-carbon-chromium-manganese-silicon
steels
delayed fracture ........................... 77(F)
stress-corrosion cracking ................. 81(F)
Iron-carbon-chromium-nickel-manganese
steel, stress-corrosion cracking ....... 81(F)
Iron-carbon-chromium-nickel-molybdenum
steel
manganese sulfides causing corrosion ........81
stress-corrosion cracking .....................81
Iron-carbon-chromium-silicon steel, delayed
fracture ................................... 76(F)
Iron-carbon-manganese-nickel-molybdenum
steel, stress-corrosion cracking ...........81
Iron-carbon-manganese-silicon-nickel steel,
stress-corrosion cracking ..................81
Iron-carbon-manganese-silicon-nickelvanadium steel, stress-corrosion
cracking .................................. 82(F)
Iron-carbon-manganese-silicon steel, stresscorrosion cracking ..................... 82(F)
Iron-carbon-manganese-silicon-vanadiumniobium steel, stress-corrosion
cracking ............................. 8182(F)
Iron-carbon-manganese-silicon-vanadium
steel, stress-corrosion cracking .. 8182(F)
Iron-carbon-nickel-chromium-manganesesilicon steel
delayed fracture ..........................7677
stress-corrosion cracking in sodium chloride
solution ...................................80
Iron-carbon-nickel-chromium-molybdenumsilicon steel, stress-corrosion

cracking ......................................81
Iron-carbon-nickel-chromium-molybdenum
steel, stress-corrosion cracking ...........81
Iron-carbon-nickel-chromium-molybdenumvanadium steel, stress-corrosion
cracking ......................................81
Iron-chromium alloys, nitrogen
concentration .................... 216(F), 217
Iron-chromium-aluminum yttrium
coating ..................................... 118
Iron-chromium-nickel-molybdenumaluminum-titanium steel, delayed
failure ...................................7879
Iron-copper steels, composition ........ 407(T)
Iron-copper system, liquid-phase
sintering ................................... 416
Iron micropowders ........................... 400
Iron-nickel-cobalt-molybdenum-titanium
steel, delayed fracture ................. 77(F)
Iron-nickel-molybdenum-tungsten-titanium
steel, delayed fracture ................7778
Iron-nickel powder, P/M processing ...... 402
Iron-nickel steels, composition ......... 407(T)
Iron-nickel-tungsten-molybdenum-titanium
steel, delayed fracture ............ 7778(F)
Iron nitrides .............................. 209, 216
Iron powder, production of .................. 399
Isochronal lines ................................ 192
Isochronous curves, mechanical properties
dependencies on temperature .......... 313,
314(F)
Isolines for temperature, for bearing ring
(317(F)
Isoparametric elements ...................... 364
ISO test methods
ISO 9950, cooling-curve
analysis ....................... 266270(F)
Isothermal cooling in the air ............... 328
Isothermal nitriding .......................... 170
Isothermal quenching
and distortion ............................ 177(F)
tanks ........................................... 172
Isothermal transformation ........ 331, 332(F)
Isothermal transformation (IT) diagrams.
See also Time-temperature transformation
diagrams. ........ 248(F), 249, 250, 251(F)
Isotherm fields ............................. 317(F)
of bearing ring ........................... 317(F)
Isotherms, temperature distributions during
induction heating of gear teeth .....230(F),
231(F)
Isotropic function theory .................... 298
Isotropic-hardening hypothesis .. 299(F), 301
Isotropic strain ................................ 298
Iteration method .............................. 268
J
J2-flow theory of von Mises ................. 431
Jakibatsuchi (Japanese clay
mixture) ............... 307(F), 308(F), 309
Japanese Industrial Standard (JIS) steels.

See Steels, series and classes, JIS
numbers.
Japanese Industrial Standard (JIS) test
methods, JIS K 2242
cooling-curve analysis ......266267, 271(F)
Japanese swords ........................281282
distortion .....................308(F), 309310
quenched, computer simulation of metallothermo-mechanics ...... 96, 304, 307(F),
308(F), 309310(F)
Japan Heat Treatment Society, Quench
Cracking Working Group .. 227, 285(F)
Jet cooling ...................................... 314
Jet-hardening and fog-quenching
equipment ...................... 172, 174(F)
Jet shot peening, optimal peening
angle ................................... 348(F)
J-intergal, resistance to crack growth ........80
Johanssen method ........................ 106(F)
measurement ....................... 102(T)
modification of ........................108109
Johnson-Mehl relation, modified .... 288, 297
Joining ............................. 391396(F,T)
P/M components ......................405406
Jominy hardenability, influence on gear
stresses ............... 441442(FT), 443(F)
Joule heating effect ........................... 225
K
Kakuno criterion ................................17
Kernel ........................................... 147
Kernels of integral operators ......... 126, 127
Keyslot, effect on hollow cylinder distortion
after quenching ................. 165, 166(F)
Kinematic hardening equations ........... 432
hypothesis .......................... 299300(F)
Kinetic diagrams of cracking ............ 74(F)
Kinetics of transformations .................... 3
Kirkendall effects ............................. 416
Koistinen-Marburger law, ............201202
Kolosov-Muskhelishvilis relations
(potentials) ............................... 132
Kondratiev form coefficient ................ 315
Kondratiev form factor ............ 322, 323(T)
analytical calculation of, functions
for ................................... 323(T)
Kondratiev number .....322, 323, 324(T) 328
L
Laboratory of Mechanical Systems
Engineering (LASMIS) ..................22
design tool on fatigue ................... 20, 21
k stress singularity ........................... 131
Lance method of shot peening ......... 348(F)
Lanthanum, alloying element effect on crack
growth .......................................80
Lapping ......................................... 278
Large-modulus gears, m 10 to 14 mm
advantages ................................ 327(T)


Laser cladding three-dimensional
printing ................................... 413
Laser cutting ................................... 109
Laser hardening ............................... 220
applications .................................. 220
Laser heat treatment, deformation
tendency ............................... 170(T)
Laser shock treatment .........................25
Laser shot peening ...........354355, 357(F)
Laser sintering ................................ 413
Laser spraying, mechanical, thermal, or
structural origins of residual stress ..13(T)
Latent heat due to solidification, of
aluminum alloy ....................... 378(T)
Latent heat of transformation ................ 3
Lathe turning ............................ 101, 109
Lathe martensite .............................. 254
Lattice parameter, of first-order residual
stresses of metallic materials ....... 335(T)
Lead
as quenchant ................................. 261
error ....................................... 454(F)
Lean exothermic furnace atmosphere,
composition ........................... 404(T)
Least-square method ................... 107, 108
Ledeburite ...................................... 168
in induction-hardened and ground
microstructure ......................... 245
Lehrer phase diagram ....................... 209
Leidenfrost point, of quenching media
and carburizing ........................... 445
Leidenfrost Temperature,
definition ..................................... 258
Length of curved beam
Lever rule ................................ 376, 377
Life cycle simulation with residual stress
consideration ..........................1112
Lift height ........................................91
Linear absorption coefficient ............... 121
Linear elastic fracture mechanics ...... 29, 30
Linear mixture rule .......................... 200
Linear shrinkage (LS) (isotropic) for molded
dimension to sintered part
dimension ................................. 419
Liquid carburizing ................. 190, 443(F)
Liquid flow velocity ......... 271273, 274(F),
275(F), 276(F), 277(F)
Liquid-phase sintering ................. 416, 417
Liquid tracer system, to inspect shot peening
coverage control .......................... 348
Liquidus temperature, of aluminum
alloy .................................... 378(T)
List image file format ........................ 304
Loading, and distortion ............ 255, 256(F),
Loading mean stresses, and residual stresses
compared ...............................2728
Loading methods, and distortion .......178(F),
179182(F,T)
Loading pattern, and distortion .. 171(F), 172,
173(F)
Loading stresses, and residual stress

relaxation ....................................59
Local fatigue strength
and crack initiation ...........................34
high-strength steels .................. 4647(F)
medium-strength steels ................... 47(F)
Locally effective fatigue strength, calculation
of .............................................45
Local quenching, deformation
tendency ............................... 170(T)
Logarithmic creep law ....................6364
Longitudinal compressive stresses, effect on
bending fatigue life in carburized
steels ............................. 449, 450(F)
Longitudinal electricity modulus
material ................................... 334
Longitudinal residual stresses .......... 9394,
95(F), 146
of carburized-and-hardened steels .... 437(F)
of Japanese sword ................. 309310(F)
of rail steel .....427(F), 428(F), 429430(F),
431(F), 434(F)
Longitudinal stress in plate, formula for
calculating residual stress .... 95(F), 96(T)
Longitudinal stress in solid bar where r is
radius of bar, formula for calculating
residual stress .................. 95(F), 96(T)
Longitudinal stress in thin-walled
tubing ............................... 94, 95(F)
formula for calculating residual
stress ......................... 95(F), 96(T)
Lorentz (L) distribution ..................... 334
modified method ............................ 122
Low-alloyed pipe steels, hydrogen
embrittlement resistance ..................76
Low-alloy high-strength steels, stresscorrosion cracking ....................... 110
Low-alloy plasma-nitrided steels, residual
stress modeling ........................... 216
Low-alloy steels
austempering ................................. 281
austenitizing .............................. 337(T)
carbonitriding ............................ 191(F)
composition .............................. 407(T)
crack growth shown by acoustic
emission ..................................75
cryogenic cooling ................. 331, 337(T)
cyclic deformation behavior .................28
metal-injection molding .................... 412
nitriding ....................................... 209
Poissons ratio .............................96(T)
quenching ................................ 337(T)
shear modulus .............................96(T)
stress relief and yield strength .. 251, 254(F)
tempering ................................. 337(T)
tempering effect on ultrasonic velocity .. 114
Vickers hardness ........................ 337(T)
Low-alloy structural steel, nitriding ... 210(F)
Low-angle diffraction methods ............ 439
Low-carbon ferritic steels, Barkhausen noise
analysis for stress measurement ........ 114
Low-carbon steel
delayed fracture ...............................74
nitriding ................................... 209(F)
Index / 481
distortion ........................... 171(T)
Low-cycle fatigue(LCF)
cyclic deformation behavior of shot peening
and deep rolling on plain carbon
steel .............................. 31(F), 32
effect on bending fatigue strength of
medium strength steel ..................41
Lower bainite ........................ 250(T), 257
Low-strength steel
macro residual stress effect ..................27
strength .......................... 5051(F)
effect ................3637(F), 3940(F)
LS-DYNA 3D (software program) ........ 363
Lubrication, effect modeled for casting ... 364
Luders bands ....................................28
Lumped-heat-capacity method (computer
program) ........... 267, 268, 269, 271(F),
274(F), 279(F)
LUMP-PROB (computer program) ..... 268,
289(F)
M
Machine driving system, and distortion ... 172
Machine parts, induction-hardened stress
profiles in the loaded state .... 242243(F)
Machining ................................... 19, 99
and distortion ...................... 254, 256(F)
distortion caused by process ........ 165166,
168(F), 169(F)
effect on distortion of gears ..... 453, 454(F)
P/M parts ............................... 405, 406
Mackenzie and Bowles notation, for
deformation matrix describing shape
deformation .................................. 8
Macroresidual stress ...........................45
and annealing ......................... 5455(F)
relaxation in notch root .................. 33(F)
state .............................................27
vs. fatigue strength ........................ 50(F)
Macroscopic residual stress ..12, 331, 332(F)
Macrosegregation, reduced by rapid
solidification .............................. 400
Macrostresses (first-order residual stresses)
89, 143, 145, 215, 331, 332(F), 333334,
335(T)
calculation of ................................ 217
carburizing and carbonitriding ............ 189
in coatings .................................... 118
definition ..................................... 125
and hydrogen embrittlement .................72
residual or tensile, in carburized steels .. 197
Magees rule, modified ....................... 288
MAGNA (computer program) ... 306(F), 308
Magnetic Barkhausen noise, .. 112, 114115
Magnetic domain temperature ......... 225(F)
Magnetic-flux concentrator ................. 222
Magnetic measurement methods ............25
Magnetic methods ............... 13, 14(F), 109

482 / Index
Magnetic permeability, and induction

heating ..................................... 226
Magnetic transformation temperature .. 223,
225, 226227
Magnetomechanical acoustic emission
(MAE) ..................................... 115
Magniflux inspection ......................... 256
Mandrels, and quenching in a hardening
press .............................. 185186(F)
Manganese
steel .......................................83
growth ............................... 80, 81
parts ................................. 411(T)
and flocculation susceptibility of steels ....73
crack ..................................76(T)
Manganese sulfides ........................ 72, 76
effect on hydrogen embrittlement
of steels ...................................83
effect on iron-carbon-chromium-nickelmolybdenum steel .......................81
Manson-Coffin relationship ..................45
Manual on Shot Peening (1952) SAE
publication ............................... 345
Maraging steels
crack growth mechanism in delayed
failure ................................. 78(F)
delayed fracture .......................... 75, 78
high-strength, delayed fracture of ..........77
hydrogen states ................................72
stress-corrosion cracking .....................81
welding of ............................. 6061(F)
MARC (software program) .... 202, 363364
Marquenching. See also
Martempering. ........... 261263, 266(T),
280281, 287(F)
and distortion .................. 176(T), 177(F)
modified ............................ 281, 287(F)
Martempering. See also
Marquenching. ........... 261263, 266(T),
280281, 287(F), 417
Martensite ..................................... 4(F)
contact fatigue effect ....................... 198
deformation systems associated with
transformations ....................... 5(T)
formation, causing compressive residual
stresses .................................. 150
formation during hardening processes .. 200,
201202
hardening from grinding power ........... 150
in induction-hardened and carburized steel
cylinder ....... 203204, 205(F), 206(F)
after induction hardening .............. 240(F)
plate ................................... 91(F)
shape change ................................ 5(T)

transformation affected by stress ..... 297(F)
transformation-start temperature, stress
effects on ............................ 6(F,T)
transformation through sinter-hardening
process .................................. 408
volume fraction after induction
hardening ........................... 228(F)
volume fraction after water quenching of
disk .................................. 289(F)
volume-fraction distribution after induction
hardening ........................... 231(F)
volume fraction in induction hardened gear
wheel ................ 306(F), 307(F), 309
volume fraction in quenched Japanese
sword ................ 308(F), 309310(F)
Martensite finish (Mf) temperature .. 251, 260,
263(F), 332, 333
Martensite formation, in bearing ring
hardening study .............. 314317(F,T)
Martensite start (Ms) temperature ...... 248,
249, 250(F), 260, 263(F), 332
carbon and alloying element effects ...... 200
of carburized steels ...................191192
and quench cracking ....... 279, 281, 286(F),
287(F)
and quenching in a hardening press ...... 186
after induction hardening and grinding .. 245
Martensitic hardening .........................99
Martensitic steels
hydrogen states ................................72
residual stress vs. temperature ........... 7(F)
Martensitic transformation ............. 314(F)
alloy steel ........................... 332, 333(F)
mixed theory ................................. 332
stress theory .................................. 332
thermal theory ............................... 332
tool steel ............................ 332, 333(F)
total energy exchange ...................... 332
Martins process ................................70
Mass effect ..................................... 274
Material removal ............................. 109
Maximum shearing .............................17
Maximum static pressure .....................21
MC carbides ............................. 404, 405
M6C carbides ............................. 404, 405
Mean stress .............................30(F), 297
Mean stress sensitivity ....................4546
Measured deflection, symbol and units
of .........................................93(T)
Measurement
destructive procedures ......... 101110(F,T)
destructive residual-stress
procedures .......................... 102(T)
elastic strain .................................. 101
generic destructive stress-relief
procedure ............................... 101
indentation methods ........................ 111
isolation of gaged element ................. 101
material-removal methods ................. 109
nature of residual stresses ..........99(F), 100
need for ....................................... 100
nondestructive procedures ...........111115
post-stress relaxation ....................... 101
of residual stress .................. 438439(F)
sectioning ..................................... 109

semidestructive procedures .......... 110, 111
spot annealing ............................... 111
strain-measurement methods ........109110
strain-measurement technique ............. 101
stress-field condition
assumptions ............... 101109(F,T)
Mechanical alloying .......... 400, 414(F), 419
Mechanical alloying of Japanese sword
steel ........................................ 309
Mechanical chip-removal processes ...... 109
and isolation of gaged element ............ 101
Mechanical deflection ........................ 124
to measure residual stresses
in coatings .............................. 118
Mechanical driving force ................. 6(F,T)
limits affected by temperature ............ 6(F)
Mechanical gages, for strain
measurements ....................... 109, 110
Mechanical-gaging technique .............. 102
Mechanical strength of material, residual
stress effect ...................... 1416(F,T)
Mechanical stresses ........................... 332
Mechanical twinning ............................ 4
characteristics ............................... 5(T)
shape change ................................ 5(T)
Mechanisms of transformations .............. 3
Medium-carbon low-alloy steel, quenching
temperature effect on mold
hole distortion ........................ 171(T)
Medium-carbon steels
distortion .......................... (171(T)
residual stresses due to milling ...... (109(F)
tempering ........................... 284, 287(F)
Medium-strength steels
bending fatigue limit .........................50
bending fatigue strength vs. residual
stress .................................. 43(F)
crack propagation .................... 4849(F)
Haigh diagram model of residual stress
influence ......................... 4445(F)
local fatigue strength ..................... 47(F)
strength .......................... 5051(F)
relaxed residual stress sensitivity and mean
stress sensitivity of bending fatigue
strength ............................... 45(F)
effect ................3738(F), 4042(F)
Melting, and distortion potential ............ 183
Melting temperature (Tm) .............. 54, 367
Melt penetration .............................. 416
Melt pressure, in casting ........... 368369(F)
Mesnager-Sachs boring-out technique ... 102
measurement ....................... 102(T)
Metal infiltration, residual stresses ......... 419
Metal-injection molding (MIM) .... 399, 400,
412(F)
Metallic cementation, deformation
tendency ............................... 170(T)
Metallic mold casting ........................ 167
Metallothermomechanical
theory ............................ 284288(F)
Metallo-thermo-mechanics

application to quenching ......... 296310(F)

parameters in schematic diagram ..... 296(F)
Metal-matrix composites ............ 11, 12, 25
Metal-oxide semiconductors ................ 118
Metal powder production, in powder
metallurgy processing ............... 397(F)
Metal welding ....................................70
Metastable alloys .............................. 400
Michelson interferometer ................... 135
Microalloying, to preserve fine-grain
austenite ................................... 325
Microcarbides ................................. 339
Microcracking ......................... 32, 33, 70
of carburized steels ...................196197
of gas main pipes .............................82
in low-strength
steels ......................................41
in maraging steels .............................75
propagation in medium-strength
steels .................................4142
in steel ..........................................71
Microcreep .................................. 59, 60
Microdefects, and cryogenic treatment .. 447
Micro distortion quenching ................. 171
Microextrusion, and cryogenic
treatment ................................. 447
Microhammering .......................325326
Microlamination .............................. 419
Micronotches, crack initiation enhanced by
shot peening ................................34
Micro plastic strains (ep) ......................54
Micropores, hydrogen accumulation
in steel .......................................72
Micro-quasi-chips ...............................73
Micro residual stress
cyclic deformation .......................6768
uniaxial deformation for relaxation .........63
Micro residual stress state ............... 27, 55
Microsegregation
of prealloyed powder ....................... 402
reduced by rapid solidification ............ 400
Microstrains ...... 112, 331, 332(F), 335336
Microstresses (second-order residual
stresses) .......77, 89, 112, 114, 125, 143,
331, 332(F), 334
in coatings .................................... 118
definition ..................................... 125
and hydrogen embrittlement .................72
residual .........................................19
of steels, cryogenic cooling
influences ........................... 336(F)
Microstructural evolution
analysis ..............................303304
Microstructure
influence on residual stress and carburized
steel strength ..... 446449(F,T) 450(F)
interaction with time, temperature and
deformation ........................... 3(F)
of powder metallurgy processed
steels ..............................419420
transformations ....................... 36(F,T)
Microstylus profilometry .................... 119
Microstylus traces ........................ 120(F)
Microvoids ..................................... 339
Mikus diagram ............................ 337(F)
Milam system ............................351352
Mild steels
cold drawing ............. 147(F), 148149(T)

Military specifications, MIL-S-13165
shot peening ................................. 345
Miller indices .............................. 334(F)
Miller indices of the diffraction planes hkl,
materials ............................... 335(T)
Milling ...........55, 57(F), 101, 109, 150151,
152(F), 414
and distortion ................................ 152
down-cut ......................150151, 152(F)
effect on fatigue behavior of high-strength
steels ......................................43
face-milling ..................150151, 152(F)
with grinding, S-N curves and residual
stresses effect ...........3738(F), 39(F)
and local fatigue strength ....................47
and P/M processing .................. 400, 414
residual stresses effect on S-N
curves ............................ 3637(F)
residual stresses in medium-carbon
steels ................................ 109(F)
up-cut .........................150151, 152(F)
Mineral oils, for quenching ............319320
Mises criteria .......................... 6, 17, 155
von Mises equivalent stress ................. 374
Mises-type yield function .................... 288
von Mises flow criterion ............. 6, 17, 155
von Mises J2-flow theory .................... 431
von Mises yield function ........299, 301, 302
Mixing, in powder metallurgy
processing ............................. 397(F)
Mixture rule ................................... 284
Mixture theory ................................ 379
MJS. See Multiphase jet solidification.
Mode I crack, crack propagation rate ........36
Modeling.See also Computer simulation,
Finite-element method. ....................25
residual stress field ................ 199206(F)
Modified Bessel functions of the
first type .................................. 132
Modified Lorentzian method ............... 122
Modified Struve functions .................. 132
Moire interferometry ..........................12
Mold shrinkage factor ....................... 419
Molten metal quenching .........261, 271, 275
Molten-salt baths. See also Salt-bath
quenching.
as quenchant ...........................261262
Molybdenum
oxidation ................................ 449
alloying effect on P/M structure
steels ..........................406407(T)
effect on sintering of stainless steels ..... 404
crack ..................................76(T)
Index / 483
Molybdenum carbides, in high-speed steels
after cryogenic cooling ..340341(F), 343
Molybdenum (max), composition
requirements for powder forged
parts .................................... 411(T)
Molybdenum nitrides ........................ 209
Moment of inertia .........................9192
Moore and Evans method .................. 105
measurement ....................... 102(T)
Mosaic blocks ............. 71, 342(F), 343, 344
Multiangle technique ......................... 112
Multimaterials ...................................11
inducing residual stress ......................24
Multiphase jet solidification (MJS) ....... 413
Multiple-motion die-set process ........... 401
Multiroll straightening .................. 146(F)
Mushy zone .................... 362, 375, 376(F)
centrifugal casting ........... 379, 382, 383(F)
cooling water discharge effect .. 379, 381(F)
Muskhelishvili equation of the elastic
theory ...................................... 132
N
NACTRAN software program ............. 127
Nanocrystalline materials ................... 400
National Center for Manufacturing Sciences
(NCMS) ................................... 202
Navy C-ring test ........................ 91, 94(F)
Near-net-shape components ................ 404
Neel temperature ................................ 3
Net opening (x) ..................................95
Net opening deflection method ..... 9596(F),
97(F)
Net opening displacement, simulation of
residual stress effects .........96(F), 97(F)
Net-shape powder compaction ............. 415
Net-shape products .............. 398399, 418
Neutron bombardment, for inducing residual
Neutron diffraction (ND) .....11, 12, 109, 144
to evaluate brazed joint of steel and tungsten
carbide-carbon cemented
carbide ........................ 395396(F)
to evaluate residual stresses after
grinding ............................. 151(F)
to evaluate residual stresses in P/M
processing .............................. 413
to evaluate steel/tungsten carbide-cobalt
brazed compounds ................ 395(F)
to measure actual crystal
dimension ................. 111112, 113
to study rail steel residual stresses ........ 431
Neutron radiation method .................. 438
Neutron scattering method ................. 431
Neutron stress analysis, brazed metallic
components and ceramic-metal
joints ................................. 395, 396
Newtonian wetting .................. 259, 263(F)
Newton-Raphson method .......131, 288, 377
Newtons iteration method .................. 304
Newtons law of heat transfer .............. 200

484 / Index
Nickel
oxidation ................................ 449
steel .......................................83
alloying effect on P/M structural
steels ..........................406407(T)
quenching, surface oxidation effect ..... 273,
277(F)
crack ..................................76(T)
Nickel-base superalloys, specific types
Inconel 600, as ISO probe for computer
simulation .......................... 289(F)
Inconel 600, as quench temperature
probe .................... 266, 267, 270(F)
Inconel 718, lasershot peening .. 355, 357(F)
Inconel 718, shot peening ....... 355, 357(F)
Inconel 718, shot-peened gas turbine
components and distortion ........... 156
Inconel 718, turning .............. 151, 153(F)
Nickel-chromium alloy
case-hardening ........................... 438(F)
centrifugal casting of pipe .......... 380, 382,
383(F), 384(F)
Nickel-chromium-aluminum yttrium
coating ..................................... 118
Nickel-chromium steel
aging effect on carburized steel .......... 449,
450(F)
repeated stressing effect in carburized
steel ........................... 450, 452(F)
quenched .... 303(F), 304(F), 305307(F)
Nickel-containing steels
Nickel (max), composition requirements for
powder forged parts .................. 411(T)
Nickel plating ....................................12
Nickel steels
composition .............................. 407(T)
Niobium
alloying effect in steels dispersion
hardening ............................... 325
critical amount in steel for affecting
recrystallization and energy of
activation ............................... 325
76(T)
Niobium nitride (NbN) ...................... 209

crack ..................................76(T)
Nitrate solutions
and stress-corrosion cracking ................79
Nitriding .......................12, 209219, 407
alloy steels ................................... 212
Armco iron ................................... 212
and carburizing, compared ................. 209
C45 steel ..................................... 212
chromium-alloyed steels ................... 216
chromium-manganese steels ............... 212
chromium-molybdenum steels ............ 212
chromium-molybdenum-vanadium
steels ................................ 212(T)
compound layer residual
stresses ..................... 211213(F,T)
computer modeling ............ 211214(F,T)
diffusion layer of residual
stresses ............................213214
and distortion of surface layer ............ 172
distortion tendency ................ 179, 181(F)
EN 19 steel ................................... 216
gaseous nitriding in NH3-N2-H2
mixtures ............................ 209(F)
gaseous nitrocarburizing ............... 215(F)
heating device distances .................... 184
low-alloyed steels ........................... 209
low-alloy structural steel ............... 210(F)
low-carbon steel ......................... 209(F)
modeling of residual stresses .... 216218(F)
nitrocarburizing (ferritic) ........ 209210(F),
211(F)
plain carbon steel ....................... 210(F)
plasma nitriding ... 210211(F), 215(F), 216
plasma nitrocarburizing ... 210211, 212(F),
215(F)
process temperature ......................... 209
residual stress influence on steel
fatigue ...................... 214216(F,T)
salt bath nitrocarburizing ........ 210, 211(F),
215(F)
short time, deformation tendency ..... 170(T)
stainless steels ............................... 211
steel .............................. 214216(F,T)
structure of nitrided layers ....... 209211(F)
unalloyed steels .............................. 209
Nitrocarburizing ........................ 407, 408
and cyclic deformation behavior after deep
rolling ................................. 31(F)
and distortion of surface layer ............ 172
modeling of residual stresses .... 216218(F)
Nitrocarburizing of austenite, deformation
tendency ............................... 170(T)
Nitrocarburizing of ferrite ....... 209210(F),
211(F)
Nitrogen
oxidation ................................ 449
as quenchant ....................... 261, 265(F)
Nitrogen furnace atmosphere,
composition ........................... 404(T)
Nitrogen/helium, as quenchant ............. 261

Nodal point/element data file format ..... 304
Nonhomogeneity degree, of interplanar
spacing ..................................... 334
Non-isothermal phase transformations,
computation of .................. 200202(F)
Nonmartensite phases .............. 448449(F)
and residual stresses ...........449(F), 450(F)
at surface ..................................... 446
Nonmeteallic impurities, ..... effect on stresscorrosion cracking .........................81
Nonmetallic stringers, and distortion ..... 255,
256, 258(T)
Nonnegative continuous weight
functions .................................. 127
Non-Newtonian wetting 259, 262(F), 263(F)
Nonuniform quenching ................ 268, 280
Normalization ....................... 147(F), 254
Normalized steel, crack initiation .............34
Normalized stress gradient ...................36
Normalizing ......................................81
and distortion ...................... 154, 155(F)
Nortons creep equation ........... 217, 374(F)
Nortons law .............................. 60, 374
Notch, quenching and distortion ... 253, 256(F)
Notching ........................... 125126, 128
Notch line ....................................... 126
Notch root residual stresses .............. 32(F)
Nucleate-boiling (NB) stage ...... 258, 260(F),
261(F), 269271, 312, 314
Nuclear magnetic resonance imaging
(NMR), to evaluate residual stresses in P/
M products ................................ 413
Nuclear power plant reactor casing, residual
stresses in weld zone of bimetallic
material ................................ 137(F)
Number of cycles at a crack length,
formula ......................................34
Number of cycles to crack initiation, residual
stresses effect from mechanical surface
treatments ...................................34
Numerical analysis, formulation of ........ 376
Numerical analysis algorithm .............. 377
Numerical analysis method incorporating
solidification .............................. 372
Numerical analysis method of
thermomechanical problems in casting
374377(F)
Numerical inverse methods ........... 268, 271
O
OBD. See Over ball diameter.
Octahedral shearing ............................17
Octahedral shearing plane ....................17
Oil cooling ...................................... 315
Oil impregnation ........................ 405, 406
Oil quenching ....... 263264, 267(F), 268(F),
269, 272273(F), 275, 277(F), 282(F),
316
of carburized testpieces and gears ....... 439,
440(F)
edge effect .......... 273274, 279(F), 280(F)
steering sector shaft ............... 454455(F)

tanks ........................................... 172

ultrahigh temperature ............. 454, 455(F)
Oils. See Oil quenching.
(X)-diffractometer ............................ 121
One-angle technique (SET) ............. 112(F)
One-phase convection heat transfer ..... 312,
314
OOR. See Out-of-roundness.
Open-hearth process ...........................70
Operator kernel
of Fredholm integral equation of the first
type ...................................... 126
of Volterra integral equation .............. 126
Optical gages ............................. 109, 110
Optical-geometrical methods, to measure
strains and displacements fields in area of
crack indicator ............................ 130
Optical interferometry
methods ......................... 134136(F)
Optically sensitive coating method ........ 132
Optically sensitive coatings ................. 133
ORE Committee .............................. 431
ORE Report .............................. 431(F)
Organic matrix composites ...................12
Orikaeshi ........................................ 309
Oscillating stress .............................. 345
Oscillators ...................................... 221
Ostergren damage parameter ...... 3031, 45
Out-of-phase hardening ..................... 432
Out-of-roundness (OOR) .......418(F), 454(F)
Over ball diameter (OBD) ........ 453, 454(F)
Overheating, localized, and
distortion ............................256257
Overload-induced compressive residual
stresses ......................................35
Overload ratio ............................... 35(F)
Overpeening effects .......................... 347
Oxide-dispersion-strengthened iron-base
alloys ....................................... 414
Oxides, role in stress-corrosion cracking .....81
Oxide scale ..255, 258(F), 273, 277(F), 278(F)
and quench cracking .......... 278(F), 286(T)
Oxygen
parts ................................. 411(T)
effect on gaseous nitrocarburizing .... 214(F)
P
Pack carburizing ..........189, 443(F), 444(F),
451(F)
Parabola method .............................. 122
Parallelepiped, quenching and heating
computational and experimental
results ......................... 321323(F,T)
Paris equation ...............................29,30
describing crack propagation ...........3435
Partial damage, for inducing residual stress
relaxation ....................................54
Partially alloyed powder ................ 400(F)
Patch test ....................................... 369
Patel and Cohens method ..................... 8
Patenting ........................................ 261

PA-TRAN computer program ............. 304
PCD. See Pitch circle diameter.
Peak shift method ..........................1213
Pearlite ................................... 4(F), 448
in carburized cases .......................... 189
plate ................................... 91(F)
shape change ................................ 5(T)
transformation to austenite, volumetric
change .............................. 250(T)
volume fraction in quenched Japanese
sword ................ 308(F), 309310(F)
volume fraction of .......................... 297
Pearson VII method .......................... 122
Peen forming ................... 151, 347, 348(F)
Peening, fatigue ..................................27
Peening-induced microstructural workhardening effect ............... 6263(F,T)
Peenscan process .......... 349350(F), 351(F)
Peenstress (computer program) ......91, 350,
352(F), 353(F)
Penetration depth ............................. 121
Persistent slip bands (PSBs) ........ 28(F), 29,
3334
Perzynas constitutive model, modification
for semicontinuous casting ............. 377
Perzynas inelastic constitutive
equations .................................. 372
Perzynas viscoplastic constitutive
relationship, modified .................. 385
Perzynas viscoplastic model on excess stress
theory ............................ 373374(F)
temperature-dependent viscosity
variation ............................ 374(F)
pH, effect on main gas-pipe line steels
corrosion .....................................82
due to welding ........... 392393(F), 394(F)
and fatigue .....................................27
finite-element method for modeling ..... 296,
300, 301(F), 302, 303, 306(F), 307(F),
309
during heat treating .. 248251(F,T), 252(F),
253(F), 254(F)
material movements ..................251252
volume changes ................... 252, 254(F)
Phonons ............................................ 3
Phosphate solutions, and stress-corrosion
cracking ......................................79
Phosphorus
steel .......................................83
crack ..................................76(T)
Phosphorus (max), composition requirements
for powder forged parts ............. 411(T)
Photoelastic coating method ... 127, 128, 130,
131, 133134(F), 135, 136(F), 137
Index / 485
Photoelastic coatings ...........................91
thickness selection .......................... 134
Photoelasticity, to evaluate residual stresses in
P/M products ............................. 413
Photoelastic technique ....................... 110
Photographs, to measure surface wetting .... 269
Photometer parameter ....................... 334
Physical vapor deposition (PVD) .....25, 118,
119, 120, 121
coatings .................................... 11, 12
Pickel method .................................. 106
Piecewise-homogeneous materials,
peculiarities in study of ....... 130(F), 131
Pinion shafts, distortion ............ 255, 257(F)
Pinned dislocations ........................6667
Pinning ............................................55
Pipe
weldment residual stresses ............. 100(T)
weldments .................................... 102
Pipe steels, stress-corrosion cracking .........81
Pisarenko-Lebedev criterion ...312, 317, 328
as failure criterion during hardening ..... 315
Piston ring, tempering and distortion ..177(F),
178
Pitch circle diameter (PCD) ..........308309
Pitch error .........................453(F), 454(F)
Pitting ........................................... 215
Plain carbon steels
crack initiation .................................34
cyclic deformation behavior ....... 2829(F),
31(F), 32
nitriding ................................... 210(F)
Poissons ratio .............................96(T)
residual stress relaxation produced by
annealing ........................ 5455(F)
shear modulus .............................96(T)
Plain carbon tool steels, cryogenic
cooling ..................................... 340
Plane-strain fracture toughness .... 74(F), 80
Plane-strain stress analysis, with von Mises
yield criterion for casting ............... 366
Planing .......................................... 150
PLASMA (plasma-nitriding computer
program, simulation) .................. 216
Plasma carburizing ........................... 189
Plasma cutting ................................. 109
Plasma nitriding ... 210211(F), 215(F), 216,
407
stainless steels ............................... 211
Plasma-nitriding simulation program
(PLASMA) ............................... 216
Plasma nitrocarburizing ... 210211, 212(F),
215(F)
Plasma-sprayed coatings ................. 11, 25
Plasma spraying ............................. 16(F)
Plastic Ball Grid Array (PBGA)
package .....................................12
Plastic compression, regime in simple slab
casting ................................. 362(F)

486 / Index
Plastic deformation ........... 45(F,T), 19, 27

and cyclic softening ...................... 32(F)
nonhomogeneous ..............................12
of rail steel ......................... 431432(F)
Plastic flow, nonhomogeneous ................12
Plastic flow theory ............................ 315
with kinematic strengthening .............. 315
Plastic (inelastic) strain ...................... 299
Plasticity ........................................... 3
Plasticity-creep interaction condition .... 302
Plasticity theory equations .................. 312
Plasticizing ..................................... 220
Plastic potential, of casting .................. 365
Plastic potential theory ...................... 300
Plastic strain rate ...... 202, 286, 299300(F),
302, 312
Plastic stretching of immediate workpiece
surface ................................. 346(F)
Plastic tension, regime in simple slab
casting ................................. 362(F)
Plastification ................................... 230
after induction hardening and grinding .. 245
of subsurface after induction hardening .... 239
Plate
carburized, residual stresses .... 439, 440(F),
441(T)
procedures .......................... 102(T)
flat, biaxial stresses measurement ..106108
flat, residual stress measurement .......... 108
flat, shear stress measurement ............. 106
piecewise-homogeneous, residual stress
close to edges .......................... 129
residual-stress distributions ............ 107(F)
residual stress in inner regions,
determination of ............. 128129(F)
steel, plastically deformed, neutron
diffraction methods ................... 113
stress profile from Peenstress
software ............................. 353(F)
thick, stress measurement .................. 106
through-thickness temperature
variation ...................... 223, 224(F)
welded, shot peening effect on residual
stress .......................... 354, 356(F)
Plate weldments, stress measurement ...... 106
Plating, of P/M parts ......................... 406
Plug quenching ..............257, 280, 281, 456
Poisson contraction .............................92
Poissons ratio .................................... 3
effect on residual stress development in
steels ................................... 3(T)
materials ............................ 335(T)
Polarization curves, active-passive region of
potentials ....................................79
Polishing ........................................ 405
Polyakov rules for quenching .... 260, 264(F)
Polyalkylene glycol (PAG), as
quenchant .... 264, 268(F), 276(F), 279(F),
283(F), 285(F,T)
Polyethyloxazoline ............................ 264
Polymer additive ratio, effect on cooling rate
after induction hardening ........... 232(F)
Polymer quenchants, aqueous. See Aqueous

polymer quenchants; Aqueous polymer
quenching.
Polymers, aqueous solutions as quenchants.
See Aqueous polymer quenchants;
Aqueous polymer quenching.
Polysodium acrylate .......................... 264
Polyvinylpyrrolidone ......................... 264
Pores (collectors) ................................72
Porosity
in castings .................................... 370
of P/M parts ................................. 406
Portable optical method for strain
measurement in destructive
techniques ..................................25
Position-sensitive proportional counter
(PSPC) method ...................... 439(F)
Position-sensitive proportional detector
(PSPD) ................................ 439(F)
Positive-pole figure analysis ................ 439
Positron lifetime measurements ........... 416
Post-processing file format .................. 304
Postsintering, and metal-injection
molding .................................... 412
Powder forging ................... 410411(F,T)
cost, relative ............................. 413(F)
Powder metallurgy iron,
composition ........................... 407(T)
Powder-metallurgy (P/M) processing
advantages .............................. 397, 398
applications ............................397398
automotive applications .................... 398
classes of components ...................... 401
compaction in rigid dies ......... 401402(F)
dendrite arm spacing .................... 400(F)
density distribution ......................... 415
double-action tooling system ........401402
ferrous P/M parts ................. 398399(F)
heat treatment of steel parts ..... 417418(F)
hot pressing ........................ 408409(F)
interparticle cohesion ....................... 415
isostatic compaction ........................ 402
lubricants ..................................... 400
market value ................................. 397
metal-injection molding ................ 412(F)
metals processed ............................ 397
microstructural development and
properties .........................419420
mixing and blending ..................400401
nonsintered areas ............................ 420
operations sequence ..................... 397(F)
powder characteristics ............ 399401(F)
powder forging ................. 410411(F,T)
press-sinter manufacturing process ... 398(F)
production capacity ......................... 397
production of powder ............. 399400(F)
products formed ............................. 397
rapid prototyping (RP) ..................... 413
residual stresses ................... 413420(F)
roll compaction .................... 409410(F)
simulations, particulate modeling ........ 413,
415416, 417
sintering ......................... 402408(F,T)
spray forming ..........................412413
structural steels, compositions ........ 407(T)
superalloy weight in commercial jet aircraft
engines .................................. 398
tool steel market ............................. 398

warm compaction ....................... 413(F)
Power density .............................. 222(F)
definition ..................................... 222
Prager model, of back stress rate .......... 299,
300(F)
Pragers consistency relation ............... 300
Prandtl-Reuss flow rule ..................... 301
Prealloyed powders ................. 399, 400(F)
powder forging .............................. 410
sintering effect on microsegregation ..... 402
Pre-cooling quenching .................176177
Prediction, residual stress field .... 199206(F)
Pre-heating
to control distortion ...................155156
and distortion ...................... 171172(F)
Pre-heat-treating, and
distortion ............169172(F,T), 173(F)
Premixed powders, powder forging ........ 410
Press braking ........................ 147148(F)
Pressing ...................................141142
applications .................................. 142
Press quench ................................... 454
Press quenching .............183, 280, 281, 453
system ............................... 253, 256(F)
Press-sinter technique ....................... 397
Pressure-assisted sintering ............ 409, 417
Pressure casting ......................... 167, 168
Pressure rolling ..................... 151, 153(F)
and distortion ................................ 152
Pre-straining .....................................19
Prestress engineering of structural materials
(PESM or PRESMA) ....................25
Prestress processing ............................11
Primary gas porosity ...........................72
Primary microcreep ....................... 59, 60
Process equipment design ......... 183186(F)
Processing analysis ........................1112
Processing modeling .......................1112
Processing parameter optimization ....1112
Production gears, residual stresses
in ............................... 439441(F,T)
Profilometry .................................... 120
Programmable control units, storage of
workpiece- and material-specific treatment
parameters in a database ................ 186
Projection welding ............................ 405
Proportionality, coefficient ....................75
Protective films, rate of forming and stresscorrosion cracking .........................80
(w)-diffractometer ............................ 121
(w)-splitting .............................. 121, 122
PSPC. See Position-sensitive proportional
counter method.
PSPD. See Position-sensitive proportional
detector.
Pull cracking ............... 260, 263(F), 264(F)
Pull stress ............................. 260, 263(F)
Pure iron
277(F)
specific heat capacity, as function of
temperature ......................... 34(F)
stress-corrosion cracking insensitivity .....79
Push cracking .............. 260, 263(F), 264(F)
Push-pull fatigue testing ..........28, 63(F), 66

and cyclic deformation behavior ........ 33(F)

Push stress ............................ 260, 263(F)
Q
Quality assurance, inspection plan .. 13, 14(F)
Quality control .............................1112
of gear strength .............................. 440
Quantimet 720 quantitative microscope, with
computer interface ....................... 341
Quantitative electronic microscopy, to
identify and count microcarbides ...... 339
Quasi-spall .......................................76
Quasi-static surface yield strength in
compression ....................... 66, 68(T)
Quebec metal powder process ............. 399
Quenchants. See also Quenching; Quench
severity; Quench
uniformity. ......... 260265(F,T), 266(F),
267(F), 268(F), 269(F), 270(F)
agitation rate influence ................. 176(T)
aqueous polymer ..257, 264, 268(F), 269(F),
272(F)
brine ...........................264265, 270(F)
caustic soda ..................264265, 270(F)
circulation uniformity effect ............... 176
cooling rate and distortion
compared ........................... 175(T)
and distortion ................... 174175(F,T)
effect on quenching performance ........ 259,
262(T)
factors influencing cooling intensity .... 174,
175(T)
influence on distortions of C-style
sample .............................. 175(T)
influence on distortion trend of cavity of
cold forging dies ............ 175, 176(T)
intensively cooling .......................... 314
oils ... 263264, 267(F), 268(F), 269, 272(F)
selection, and cooling characteristics ... 275,
281(F), 282(F)
selection, and distortion ..............279280
selection effect ..... 269270, 272(F), 274(F)
temperature effect .......176, 268(F), 269(F),
270271(F)
uniformity ...... 258259(F), 260(F), 261(F)
vaporizable liquid ........................... 258
viscosity ...................................... 270
water ................ 264, 269, 270(F), 272(F)
Quenchant solution circulation rate ...... 328
Quench cracking ... 253256(F,T), 257, 259
260, 263(F), 264(F), 285(F), 312
elimination by intensification of heat transfer
at phase transformations ............. 328
prevention .................................... 323
probability of ............................ 313(F)
and quench nonuniformity ................. 258
and surface roughness ......277278, 286(T)
techniques of steel quenching for prevention
of .............................. 325, 326(F)
Quench Cracking Working Group, Japan
Heat Treatment Society ..... 277, 285(F)
Quench distortion .......................259260
Quenched-and-tempered steel
bending fatigue tests in sea water ...... 34(F)
cyclic deformation ........................ 29(F)

fatigue strength .......................... 244(F)
local fatigue strength ..................... 47(F)
Quenched steels, delayed fracture ........ 77(F)
Quench embrittlement ....................... 196
Quenching. See also Air quenching; Aqueous
polymer quenching; Brine quenching;
Carburized quenching; Cold-die
quenching; Direct quenching; Distortion;
Fluidized-bed quenching; Fog quenching;
Free quenching; Gas quenching; Hot oil
quenching; Immersion quenching;
Intensive quenching; Interrupted
quenching; Marquenching; Martempering;
Molten metal quenching; Nonuniform
quenching; Oil quenching; Plug
quenching; Pre-cooling quenching; Press
quenching; Quenchants; Quench cracking;
Quench distortion; Quench oils; Quench
severity; Quench uniformity; Restraint
quenching; Salt-bath quenching; Selective
quenching; Spray quenching; Time
quenching; Two-step quenching; Water
quenching. ..... 99, 257272(F,T), 273(F),
274(F), 275(F), 336
agitation effect .... 258, 259(F), 275277(F),
283(F), 284(F), 285(F,T)
boundary conditions selection for
simulation .............................. 325
of carburized bushing .......... 320321(F,T)
of carburized cylinder, computer
simulation .............. 296, 302(F), 305
carburized gear axial stresses .... 440441(F)
of case-hardened gear, model .. 202203(F),
204(F)
chloride solutions for tool steel
punches ................................. 326
coating effects ..................... 273, 278(F)
cooling characteristics effect on residual
stress and distortion .. 275278, 281(F),
282(F), 283(F), 284(F), 285(F,T),
286(T)
dimensional changes ....... 252, 255(F), 274,
275, 282(F)
direct or single reheat after carburizing, and
retained austenite .................. 447(F)
and distortion ...................... 252, 255(F)
distortion, components 159, 161(T), 162(F)
distortion, component with simple
shape .......... 159161(F,T), 162(F,T),
163(F), 164(F)
distortion, metallurgical sources ... 251252,
254(F)
and distortion potential ...............184185
distortion prediction .............. 284293(F)
double reheat and quench after carburizing,
and retained austenite ............. 447(F)
geometry effects ... 273274, 279(F), 280(F)
in hardening press, and distortion
minimizing ................... 185186(F)
after induction hardening ....... 236238(F),
241, 242(F), 246
for induction hardening .......... 231232(F)
from intercritical temperature interval .....81
of Japanese sword, computer
simulation ....296, 304, 307(F), 308(F),
309310(F)
Index / 487
kinetics of process .......................... 288
material effect on distortion ........ 252256,
257(F), 258(F,T)
material effects .............. 274, 280(F), 312
mathematical modeling of process ....... 319
media, 449
media, influence of distortion of
gears ........................... 454456(F)
media measurement and evaluation of
power ........ 265269, 270(F), 271(F),
272(F), 273(F)
methods .................................280282
methods, effect on carburized-and-hardened
steel residual stresses ....... 443444(F)
methods, Ukraine Patent (No. 4448, Bulletin
6-I), 1994 ............................... 328
model of stress/strain prediction in casehardened parts ............... 199, 201(F)
no phase transformation, mechanical,
thermal, or structural origins of residual
stress ..................................13(T)
one-step ....................................... 319
Polyakov rules ..................... 260, 264(F)
powder metallurgy processed parts ....... 407
quenchant and material properties
effects ......................... 259, 262(T)
quenchant choice and
distortion .................. 174175(F,T)
residual stress prediction ......... 284293(F)
simple shape bodies, computational and
simulation, inverse technique ...............23
steel chemical composition effect ........ 274
surface conditions effect ........ 273, 277(F),
278(F), 279(F)
surface distortion comparison ......... 170(F)
tank for immersion ................ 172, 173(F)
temperature influence on distortion of mold
hole .................................. 171(T)
two-step ................. 312, 319(F), 320, 328
of carburized bushing .......... 320321(F,T)
workpiece size effects ..... 252, 255(F), 274,
275, 282(F)
Quench oils .......... 263264, 267(F), 268(F),
323324
accelerated ......................... 264, 267(F)
and carburized steel residual
stresses .............................. 445(F)
conventional ................................. 263
martempering ...................... 264, 268(F)
MS-20 ...................................320321
MZM-16 .................................. 318(F)
quench chamber pressure
control ........................ 455456(F)
quench severity index
(H value) ..................... 454455(F)
vapor blanket stage ............... 455456(F)
Quench press machines, and
distortion ..................... 181182(F,T)
Quench severity ........ 257, 259(T), 262, 265,
266(F), 269
of aqueous polymer quenchants .......... 264
definition ............................... 257, 265
and distortion ..........................279280

488 / Index
Quench severity (continued)

of gas quenching .................. 261, 266(F)
water content effect ........ 262263, 266(F),
267(F)
Quench severity index (H value) of
quenching oil .................. 454455(F)
Quench uniformity ...... 258259(F), 260(F),
261(F), 266(F), 269
Quick-slow-quick quenching ............... 456
R
Racking, and distortion ... 171(F), 172, 173(F)
Radial forging .................... 148149(F,T)
Radial stress ......................... 144(F), 145
from induction hardening ........ 234, 235(F)
at t 8.32s rising a plane-strain
assumption ......................... 368(F)
at time t 8.3202 s with a tractionfree top surface .................... 368(F)
380(F)
from specimen heating ................. 313(F)
assumption ......................... 368(F)
thermal stress history with a traction-free top
surface .............................. 369(F)
Radiation, monochromatic, wavelength
of ........................................... 113
Radiation thermometer, to measure
temperature of centrifugal castings .... 382
Radius of curvature of beam on displaced
section, symbol and units of .........93(T)
Radius of round bar cross section, symbol
and units of .............................93(T)
Rail end problem ............................. 424
Railroad wheels, ultrasonic velocity
measurement technique ................. 114
Rail steel
cooling process .................... 424430(F)
cyclic plastic deformation ........ 431432(F)
distortion and residual stress
formation ..................... 424435(F)
head-hardened ............................... 426
heat flux and cooling process ......... 425(F)
moving/sliding on cooling bed ........... 429
production steps ......................... 424(F)
rail end problem ............................. 428
roller straightening process ........ 424, 430
434(F)
in service ........................... 434435(F)
temperature influence ....................... 434
welding as cause of residual
stresses .............................. 434(F)
Rail steel stress tensor, decomposition into
hydrostatic and deviatoric parts ........ 432
Railway tank cars, for transporting ammonia,
shot peening effect on residual
stress ............................. 354, 356(F)
Ram force ...................................... 409
Rapid induction heating/intensive
cooling ............................... 327, 328
Rapid omnidirectional compaction ....... 409
Rapid prototyping ...................... 399, 413

Rapid rewetting process ........... 259, 262(F)
Rapid solidification ...............400, 412, 414
Rare-earth metals
addition effect on stress-corrosion
cracking ...................................81
steel .......................................83
76(T)
Rate form of stress-strain
relation .......................... 300301(F)
Rayleigh waves ................................ 114
Real hydrogen embrittlement ................73
Real-time visualization ................ 135, 399
Recovery mechanisms .........................58
Recrystallization .............................. 416
residual stress influence ..............419420
and sintering ................................. 402
Recrystallization temperature, of ferritic
steels .........................................54
Rectangular cross-section bar, destructive
residual-stress measurement
procedures ............................. 102(T)
Redrawing ........................................60
Reduced-pressure carburizing
furnace ........................................ 453
Reduction in area ............................. 146
Reduction of iron oxide ..................... 399
Reflection polariscope ........................ 110
Reflection polariscope
Photoelastic Inc. ............ 136137(F)
Refrigeration treatments .................... 195
Regular conditions, theory of ............... 322
Regularization algorithm, when interpreting
distribution containing random
disturbance ...................... 127128(F)
Regularization methods .........126, 127, 318
Regularization parameter ................ 127
Reheating treatments ..................189190
distortion, metallurgical sources ... 251252,
254(F)
Reheat temperatures ......................... 189
Repeated bending test ..................... 17(F)
Repeated loading, effect on residual stresses
of carburized steels ..... 449450, 451(F),
452(F)
Repeated stressing, effect on residual stresses
of carburized steels ..... 449450, 451(F),
452(F)
Repeated tension test ....................... 17(F)
Repressing ................................. 60, 410
Rerolling ..........................................60
Reserve strength factor, of bearing ring
317(F)
Residual austenite ..............................81
Residual microstresses ....................... 335
Residual stresses
admissible prediction of ............. 2224(F)
compressive macro ............................54
from cooling process, calculated vs.

experimental ....................... 314(F)
cumulated value at any point .............. 331
definition ............................... 143, 221
distribution, simulation models ........... 437
effect on fatigue strength ................ 19(F)
first-order (macroscopic) ......... 331, 332(F)
induction hardening ............ 232242(F,T)
and loading mean stresses
compared ............................2728
manufacturing implications ....9091, 92(F),
93(F)
measurement methods from deflection
data ........................................91
origins of ..............................1213(T)
prediction, after quenching ...... 284293(F)
from quenching, cooling characteristics
effects ........ 275278, 281(F), 282(F),
283(F), 284(F), 285(F,T), 286(T)
of scanning induction hardened cylinder,
294(F)
second-order (microscopic) ...... 331, 332(F)
state of .............................. 331, 333(F)
third-order (ultramicroscopic) .. 331, 332(F),
335336
types ................................. 331, 332(F)
Residual stress field, modeling and
prediction ........................ 199206(F)
Residual-stress-free condition, fatigue
strength ......................................45
Residual stress relaxation phenomenon. See
also Stress relaxation. .....................17
Residual stress restitution
(inverse problem) ....................... 127
Residual stress sensitivity ................4546
Residual stress sensitivity factor (m) ........66
Resin impregnation ..................... 406, 418
Resintering, in powder metallurgy
processing ............................. 397(F)
Resistance sintering under pressure ...... 408
Resistance welding ............................ 405
Restart file format ............................ 304
Restraint fixturing ............................ 281
Restraint quenching .......................... 281
Restretching ......................................60
Retained austenite ......................336337
in carburized gears ......... 441, 442(F), 443,
444(F), 446
in carburized plain carbon and low-alloy
steels .......................... 194, 197(F)
in carburized steels .................... 437, 438
in case region of carburized and hardened
steels ....................195196, 198(F)
contact fatigue effect ....................... 198
decomposition in steels at 823973 K ... 337
decreased by shot peening .......... 198199,
200(F)
and deep freezing ........................... 178
and distortion ...................... 257, 259(T)
effect on carburized steel cylinder stresses,
model ......................... 202, 203(F)
effect on transgranular fatigue crack
initiation ..........................197198
factors affecting formation ................. 251
formation with austempering processes .. 257

as function of tempering temperature ... 251,

254(F)
in heat treated P/M tool steel .......417418
heat treatment effect on residual stress
distribution ............................. 194
heat treatment variant effect in high-alloy
tool steels ..................... 340, 341(T)
of high-alloy tool steels .......... 343(F), 344
in high-speed steels ................... 338, 339
in induction-hardened and carburized steel
cylinder ....... 203204, 205(F), 206(F)
influence on carburized steel residual stress
and strength .................. 446447(F)
reduced by cryogenic cooling ............. 331
in reheated carburized parts ............... 190
and stabilization ............................. 178
and surface residual stresses ............... 449
tempering influence on
distortion ..................... 177178(F)
transformation affecting volume during
tempering ...............283284, 287(F)
transformation by mechanical stressing in
carburized steels ............. 447448(F)
Reverse bending, of Japanese
sword ............................. 309310(F)
Reverse impending method ................. 215
Rewetting, effect on carburized steels ...... 438
Rich exothermic furnace atmosphere,
composition ........................... 404(T)
Ridder method ................................ 129
Rigid-die compaction ........................ 415
Rimmed carbon steels, Barkhausen noise
analysis for stress measurement ........ 114
Ring
of chromium-molybdenum steel, carburizedquenched, computer simulation of
distortion ..................... 290, 292(F)
distortion ........... 303(F), 305(F), 306, 307
induction hardening and carburized
quenching, computer simulation ... 296,
303(F), 304(F), 305307(F)
Ring coring ...............................110111
Rings and races of bearings thicker than 12
mm
advantages ................................ 327(T)
Robert Bosch ....................................11
Rocker arm, shot peening effect on fatigue
life ........................................20(T)
Rods
cylindrical, carburized, residual
stresses .............. 439, 440(F), 441(T)
procedures .......................... 102(T)
drawing ................... 143(F), 145146(F)
extrusion .................................. 147(F)
Roll compaction ...............409410(F), 414
friction role ................................... 409
and residual stresses ..................414416
Roller burnishing .......................... 12, 19
effect on cast iron fatigue strength ........19,
20(F)
to introduce residual compressive stress ...14
Roller straightening ..........424, 430434(F)

simulation .......................... 432434(F)
Rolling ..... 99, 141, 146147(F), 147148(F),
399
advantages .................................... 142
screwplate, jig and fixture method for
quench ........................ 159, 161(F)
Rolling bearing steel, residual stress relaxation
produced by annealing ................ 55(F)
Rolling contact, elasto-plastic, of rail
steel .............................. 434435(F)
Rolling-contact fagitue
carburized parts .............................. 196
Rolling faces pulling ......................... 363
Romberg algorithm .......................... 129
Rosettes ....................................... 34(F)
Rosenthal and Nortons method ....... 103(F)
measurement ....................... 102(T)
enhancement of .............................. 104
variation for determining triaxial residualstress condition of plate .............. 105
Rotary bending fatigue test ....... 442, 443(F)
Rotating beam fatigue test ....... 446(F), 450,
451(F)
unnotched gear teeth ........................ 450
Rotation velocity, of workpieces in induction
heating ..................................... 232
Rough-condition temperature .............. 325
Roughness
and bending fatigue strength of mediumstrength steels .......................4142
effect on fatigue strength of high-strength
steel .................................. 51, 52
effect on fatigue strength of mediumstrength steel .............................51
and quench cracking ........277278, 286(T)
of surface, and quenching ........ 273, 279(F)
R ratio .............................................19
R ratio effect .....................................30
R6M5 punches ............................. 326(F)
tool life in automatic forming
machine ....................... 326, 327(T)
Rupture strength ....................... 1619(F)
Ruud-Barrett Position-Sensitive Scintillation
Detector ............................... 112(F)
S
Sachs method ................................. 148
Safety factor
distribution of ..................................22
on fatigue ................................... 21(F)
field of .............................. 313(F), 316
Sag, of rail steel section .................. 428(F)
Salt bath carburizing .............. 189, 443(F)
Salt bath nitrocarburizing ....... 210, 211(F),
215(F)
Index / 489
Salt bath quenching ......... 261263, 264(F),
266(T), 271, 275, 449, 454
with austempering ........................... 281
Sandblasting, as predecessor of shot
peening .................................... 345
Sand casting ................................... 167
Sandwich-hologram method ................ 135
Saw curf width .............................9495
Saw cutting .......................................91
Sawing ..................................... 101, 109
Scale formation. See Oxide scale.
Scanning induction hardening, of cylinder,
computer simulation of residual stresses/
distortion .............. 290, 293(F), 294(F)
Scatter band ................................. 36(F)
Scheil equation, to determine volume fraction
of solid phase ............................. 375
Screw, distortion after quenching ....161, 164(T)
Screw dislocations ..............................58
Sea water, quenched-and-tempered steel and
crack initiation .......................... 34(F)
Secondary ion mass-spectrometry, hydrogen
ion behavior .................................78
Secondary porosity .............................72
Sectioning method ...................... 109, 424
to study rail steel residual stresses .... 431(F)
Segregation effect, distribution coefficient
representing .........................375376
Selective laser sintering (SLS) ............. 413
Selective quenching .....................281282
Self-agitation effect ........................... 271
Self-regulated thermal process ............. 328
Self-tempering ................................. 328
Semiconductor gages ......................... 110
Semicontinuous casting
cooling water discharge and stress
distributions .................. 379, 382(F)
displacement mode ................ 378, 379(F)
ingot radii, effect on stress
distributions .................. 379, 382(F)
ingot radii, isothermal lines for .... 379, 381(F)
residual stress formation ........ 377379(F),
380(F), 381(F), 382(F)
speed of casting, and stress
distributions .................. 379, 382(F)
speed of casting, effect on stress
distributions .................. 379, 382(F)
speed of casting, isothermal lines for ... 379,
381(F)
Semicontinuous direct-chill casting ....... 377
residual stress formation ........ 377379(F),
380(F), 381(F), 382(F)
Semiferritic steels, floccules not observed ..73
Sensors, for carbon potential control in gas
carburizing ................................ 447
Service life, effect of quenching method ... 316
Servohydraulic test systems, computerized
28
Shafts
warpage, computer simulation ..... 289290,
291(F)
Shape, influence on carburized steel residual
stresses ........................... 438, 439(F)
Shape-memory alloys ........................... 6
Shear component ......................... 6(F), 7
Shearography .................................. 110

490 / Index
Shear strains ...................................... 5

Shear wave birefringence ................... 114
Sheet
bending of ....................142(F), 144145
procedures .......................... 102(T)
Shell hardening ................................ 327
Shock laser treatment ..........................12
Shotblasting
and austenite transformation in carburized
steels .......................... 447448(F)
of carburized steels ............... 451452(F)
to improve fatigue strength ................ 437
Shot impact angle ............................. 348
Shot peening 12, 19(F), 55, 57(F), 6163(F),
90
applications .... 352354(F), 355(F), 356(F)
area ............................................ 347
and austenite transformation in carburized
steels .......................... 448(F), 449
automatic process .....................348349
callouts ..........................................91
carburized steels .. 198199, 200(F), 201(F),
451452(F)
C.A.S.E. process .. 350351, 353(F), 354(F)
ceramic shot .................................. 349
characterization and key data .... 346, 347(F)
compressive stresses from .............. 90, 91
computer-controlled machine ... 350, 351(F)
computer software program .... 350, 352(F),
353(F)
coverage control inspection methods .... 349
crack initiation earlier in quenched-andtempered steel ....................... 34(F)
degree monitored by Almen strips ........91,
93(F)
and distortion .......................... 152, 157
documentation by computer printout .... 350,
352(F)
effectiveness in increasing fatigue life of
mechanical parts ................ 19, 20(T)
effect on bending fatigue strength in highstrength steels ................... 4243(F)
effect on fatigue life of stainless steel .....16
elementary processes ............. 345346(F)
and fatigue strength ......46(F), 47(F), 48(F),
49, 52
gas turbine components .................... 156
gear wheels .............................347348
generating compressive macro residual
stresses ....................................54
with grinding, S-N curves and residual
stresses effect .......................3739
hardness of material ..................346347
hardness ranges of shot ..............348349
historical development ..................... 345
impingement angle vs. intensity ...... 348(F)
inspection methods for coverage
control .................................. 349
intensity in proximity of edge layer ...... 348
to introduce residual compressive stress ...14
lasershot peening ............354355, 357(F)
media .............................348(F), 349(F)
modeling capability ..... 350, 352(F), 353(F)
monitoring of process ........... 348350(F),

351(F), 352(F)
nonferritic materials, shot media used ... 349
numbers of cycles to crack initiation affected
by ..........................................34
optimization of effects ..... 345, 350, 352(F),
353(F), 354(F)
order of addressing parameters ............ 348
patents for machines ........................ 345
postquench for carburized gears ......... 443,
444(F)
after powder forging ........................ 411
prior to heat treatment ...... 150, 151, 153(F)
process parameters .......................... 350
prolonging life span under fatigue ........ 346
purposes ...................................... 345
residual stress advantages ........ 346, 347(F)
residual stress distribution modeling ... 14(F)
residual stresses having relatively large
absolute value ...............54, 55, 56(F)
residual stress influence on cyclic
deformation behavior .......31(F), 32(F)
S-N curves and residual stress effect in
medium-strength steel .............. 38(F)
software path flow diagram ...... 350, 351(F)
stress distributions ................ 346, 347(F)
work hardening after process .............. 151
work-hardening effect in medium-strength
steels .................................. 41(F)
workpiece material and process parameters
effect on residual compressive
stresses ........................ 346, 347(F)
Shot size ..................................... 347(F)
Shot velocity ............................... 347(F)
Shrinkage
of brazed joints .................... 394(F), 395
welding residual stresses and temperature
distribution ... 391(F), 392, 393(F), 394
Shrinkage factors ............................. 419
Shrinking-type residual stress
distribution ..................... 193, 195(F)
Siemens ...........................................11
Silicates, role in stress-corrosion cracking ...81
Silicon
steel .......................................83
crack ..................................76(T)
Silicon (max), composition requirements for
Silicon nitride (Si3N4)
brazing residual stresses ................ 393(T)
coating ........................................ 118
Silicon nitride ceramic, brazing ........394(F),
395(F)
Silver (Ag)

277(F)
Simple beam theory .............. 9192, 94(F)
Simple saw cutting .............................91
Simulations. See Computer simulations.
Sines criterion ...................................17
Sine-square-psi technique ............. 112, 125
with CrK radiation, to investigate residual
stresses ....................................54
Single-exposure technique (SET) ...... 112(F)
Single press and sinter, cost, relative .. 413(F)
Single-stage carburizing ........... 189, 190(F)
Singularity degree of k* stress ............. 131
Sintering ............... 400, 402408(F,T), 414
atmospheres ..................402403, 404(T)
furnace condition for steel ............. 403(F)
furnaces ................... 402403(F), 404(T)
high-temperature ............................ 403
iron-copper ................................... 403
iron-copper-carbon .......................... 403
iron graphite powder ........................ 403
phases and subphases ................... 404(T)
residual stresses .......................416417
sinter hardening ............................. 408
stainless steels ...............403404, 405(F)
tool steels .....................404405, 406(T)
treatment of P/M parts ........ 405408(F,T)
Size of specimen, effect on carburized steel
residual stress ................... 438, 439(F)
Sizing ..............................................90
Skin effect ............................ 220, 222(F)
Slice depth, appropriate ...................... 108
Slip ............................................... 4, 5
crack initiation present ..................3334
Slip band formation ........................ 28(F)
Slip casting ..................................... 402
Slip deformation, characteristics ........... 5(T)
Slip planes ...................................... 336
Slip systems ....................................... 6
Slitting ............... 91, 92(F), 94, 95(F), 96(F)
axial ......................................... 96(F)
point of ....................................... 131
Slow strain rate testing procedure (SSRT),
for tests at constant load rate .............80
SLS. See Selective laser sintering.
Slurry process, with roll compaction ...... 410
Small-base tensiometry ...................... 133
Smearing, in milling ....................150151
Smith-Watson-Topper (SWT)
parameter .................... 3031, 45(F)
Smoothing functional method ........ 127, 128
S-N curves .................................... 30(F)
gas-carburized 4320 steel .............. 201(F)
residual stress influence ............. 3643(F)
shot-peened vs. ground steel condition,
cyclic deformation ............. 66, 67(F)
S-N diagram .....................................30
slip band formation, microcrack formation
and failure of carbon steel ......... 28(F)
SNECMA .........................................11
Soft annealing, and distortion .... 153, 154(F),
155(F)

Soft magnetic materials, metal-injection

molding .................................... 412
Solidification
cylindrical body residual stress
distribution ......................... 368(F)
cylindrical body uniformly cooled ... 368(F)
kinetics of phase transformation ....375376
Solidification kinetics ..................362363
for solidification modeling, benefits listed
........................................ 363(F)
Solidification modeling
analysis methods available with their
benefits ........................ 362, 363(F)
architecture of a comprehensive
system ......................... 362, 363(F)
Solidification process, in steel ingots,
mathematical models .................... 364
Solid quenching, deformation
tendency ............................... 170(T)
Solid solutions ................................. 400
Solid-state linear image sensor
technique ................................. 418
Solid-state sintering ..............416, 417, 419
Solidus temperature (Tm) ......................54
Solubility curve equation .....................70
Solvent extraction ............................. 412
Sori (complicated deformation) ............ 309
Spark sintering ................................ 408
Spatial distributions of residual stress,
problems of determining ................ 128
Species flow analysis, for solidification
Specific heat capacity ........................ 312
due to magnetism .............................. 3
steels ................................... 3(T)
of pure iron, as function of
temperature ......................... 34(F)
Speckle-correlation interferometry ...... 110,
133, 136(F)
Speckle interferograms ............ 135, 136(F)
Speckle interferometer ............. 135(F), 136
Speckle interferometry ...... 110, 133, 136(F)
Speckle structures .................. 135, 136(F)
Spheroidization ................................ 170
and nitriding ................................. 209
Spheroidize annealing .........................90
Spindles, shot peening effect on fatigue
life ........................................20(T)
Spline approximation ........................ 129
Sponge iron process .......................... 399
Spot annealing ................................. 111
Spot mode of x-ray beam method ......... 122
Spray deposition ........................412413
Spray forming ...........................412413
Spray hardening, and distortion .. 176(T), 177
Spray quenching .............................. 282
Springback effect ...........145, 401, 415, 416
Springback effect of rail steel .............. 430
Spring steel, compressive residual
stresses .......................................27
Stabilization, and distortion ................. 178
Stabilized zirconia, P/M processed ........ 420

Stablein method, for destructive residualstress measurement .................. 102(T)
Stacking faults ................................. 333
Stage I crack growth ...........................29
Stage II cracks ..................................29
Stainless steels
cyclic deformation behavior ............. 32(F)
nitriding ....................................... 211
P/M processing .............................. 399
Poissons ratio .............................96(T)
quenching, agitation effect .......... 272273,
277(F)
277(F)
residual stresses due to facing ......... 109(F)
quenched .... 303(F), 304(F), 305307(F)
shear modulus .............................96(T)
shot peening effect on fatigue life ..........16
sintering ......................403404, 405(F)
spray forming ..........................412413
tension testing ............................... 373
weldment residual stresses ............. 100(T)
Stainless steels, series and classes
AM-355. See S35500.
S30303 (303L), corrosion resistance of P/M
parts ................................. 405(F)
S303LSC, corrosion resistance of P/M
parts ................................. 405(F)
S30400 (SUS 304)
chemical composition ............... 372(T)
crack initiation after mechanical surface
treatments ......................... 33(F)
cyclic deformation behavior after shot
peening, and deep rolling ....... 32(F)
fatigue crack propagation rates ....... 35(F)
mechanical properties ............... 372(T)
melting point .............................. 372
P/M processing .............. 399400, 420
for ring, induction hardened and
carburized quenched ...........303(F),
304(F), 305307(F)
strip continuous casting by twin-roll
method ............................... 383
SWT parameter vs. number of cycles to
failure .............................. 45(F)
titanium nitride-coated, microstylus trace
of residual stresses ............. 120(F)
S30403 (304L)
corrosion resistance of P/M parts ....405(F)
P/M processing .....................399400
S31600
P/M processing .....................399400
sintering of powder metallurgy parts and
corrosion resistance ................ 403
S31603 (316L)
corrosion resistance of P/M parts ....405(F)
P/M processing .....................399400
sintering effect on corrosion
resistance ............................ 404
S31700, P/M processing .............399400
S31703 (317L), corrosion resistance of P/M
parts ................................. 405(F)
Index / 491
S34700, P/M processing .............399400
S35500 (AM-355), PVD titanium nitride
coating .................................. 121
S41000, P/M processing .............399400
S41003 (410L), P/M processing .......... 402
S44004 (440C), P/M processing ...399400
S44600, P/M processing .............399400
188, tight scale formation ...... 255, 258(F)
SS-100, corrosion resistance of
P/M parts ........................... 405(F)
22CR-5Ni-3Mo-0.015N-0.003C, balance Fe,
hot extrusion ........................... 410
Fe-24Cr-8Ni, phase decomposition behavior
of P/M powder ........................ 419
STAMP Process ............................... 409
Static loading, plasticity degree decrease ....75
Statistical process control (SPC), of powder
forging ..................................... 411
Steady-state continuous casting, finiteelement method formulation ......376377
Steam treatment, of P/M parts ...... 405, 406,
407
Steel ring, distortion prediction, carburizedquenching ....................... 290, 292(F)
Steels
cold forming ....................... 147148(F)
crack growth ...................................75
cryogenic cooling influence on dimensional
stability ....... 331, 332, 336339(F,T),
340(F)
cryogenic cooling influence on residual
stresses .................................. 336
cyclic behavior ............................ 22(F)
cylindrical forgings, neutron diffraction
methods ................................. 113
fatigue behavior ...................... 2831(F)
fatigue, Dang Van diagram .............. 18(F)
grade, effect on residual stress ........ 438(F)
grade, influence on gear
stresses ............ 441442(FT), 443(F)
hydrogen states in ........................7172
martensitic transformation of carburized
surface ..................................... 5
nitriding .......................... 214216(F,T)
physical properties affecting residual stress
development .......................... 3(T)
plate, plastically deformed, neutron
diffraction method .................... 113
P/M processed, heat
treatment of .................. 417418(F)
production and distortion potential ....... 183
residual stress magnitudes and
distributions ....................99(F), 100
rocket case forgings, neutron diffraction
methods ................................. 113
shot-peened, fatigue life .................. 18(F)
weldments, neutron diffraction methods
for ....................................... 113
Steel shafts, warpage prediction, computer
simulation .................289290, 291(F)
Steels, series and classes
0.18C-0.25Si-0.8Mn-0.5Cr-0.22Mo-0.62Ni,
carburized cylinder tangential stresses
during quenching, model ... 202, 203(F)
0.75Cr-3.05Ni (SNC815)

492 / Index
Steels, series and classes (continued)

carburized gear axial stresses during
quenching ....................... 441(F)
gear application, model of case hardening
and stresses from ........ 202203(F),
204(F)
304(F), 305307(F)
Cr12Mo, hollow cylinder distortion ...... 163
CrWMn, thermal straightening ............ 170
9CrSi, asymmetrical component
distortion ......... 165, 167(F,T), 168(F)
15CrNi6, residual stress of hard metallic
material ............................. 346(F)
15CrV6, volume changes during phase
transformation ............... 252, 254(F)
17CrNiMo6, shot peening .......... 350, 351,
352(T), 354(F)
18CrNiMo5, residual stress of hard metallic
material ............................. 346(F)
18Cr2Ni4W, gear wheel, distortion of ... 166
38CrMoAl, axle of boring lathe, distortion
of ........................................ 166
50CrV4
residual stress of soft metallic
material .......................... 346(F)
stress peening ......................... 348(F)
9Mn2V, thermal straightening ............ 170
16MnCr5, carburized steel CCT
diagram ....................... 192, 193(F)
26MnCr4, hardened and case-hardened
plate ........................... 191, 192(F)
diagram ....................... 192, 193(F)
90MnV8, volume changes during phase
transformation ............... 252, 254(F)
90MoCrV8, fluidized-bed quenching ... 263,
267(F)
diagram ....................... 192, 193(F)
20MoCr4, carburized cylinders and casehardening residual stress ... 195, 198(F)
1015 (CK15, S15CK)
aging effect on carburized steel ....... 449,
450(F)
case hardening ........................ 438(F)
cold-rolled steel sheet stampings,
distortion of ............... 168169(F)
distortion of carburized steel ........... 159
hollow cylinder distortion .............. 162
S-N curves and residual stresses effect,
after deep rolling ................. 37(F)
work-hardening effect on fatigue
behavior ...............................40
1018
carbonitriding ......................... 191(F)
carbonitriding and tempering .......... 191,
192(F)
tempering carburized cases ......... 191(F)
1019, crack propagation rate and stress
intensity in welded
specimens ....................... 3536(F)
1020 (JIS S20C)
carburized and quenched ...... 443, 444(F)
S-N diagram (Wohler curves) for slip band
formation, microcrack initiation, and
failure .............................. 28(F)
1023 (CK22 German grade)

cyclic stress-strain curve ......... 2829(F)
residual stress relaxation ..................59
annealing .......................... 55(F)
1040, water quenching ...............253254
1045 (CK45 German grade)
brazed compound with tungsten carbidecobalt ............................. 395(F)
brazing residual stresses ............. 393(T)
brazing with ceramics
compounds .............394(F), 395(F)
bushing distortion, eccentric
cylinder ..................159, 161(F,T)
continuous-cooling-transformation
diagram .................... 248249(F)
crack initiation .......................... 34(F)
cyclic deformation behavior .......... 32(F)
cyclic deformation behavior after deep
rolling .............................. 31(F)
cyclic deformation behavior in low-cycle
fatigue regime ............... 31(F), 32
cyclic stress-strain curve ......... 2829(F)
distortion of cylinder ...... 162(T), 163(F)
distortion of stop ring ......... 159, 161(F)
eccentric hollow cylinder,
distortion of ............... 165, 166(F)
effective stress-intensity factor
range ............................... 48(F)
fatigue notch factor vs. surface residual
stress .............. 40(F), 41(F), 42(F)
grinding and macro residual stresses and
half-width values ............ 55, 57(F)
grinding and S-N curves influenced by
residual stresses .................. 37(F)
grinding-induced work softening ........42
Haigh diagram of bending fatigue strength
of high-strength steel ............ 44(F)
half-width values vs. annealing time, after
shot peening ................. 55, 56(F)
hollow cylinder distortion ....... 161, 162,
164(F), 165(F,T)
induction hardening ............ 223224(F)
local fatigue strength ..........46(F), 47(F)
macro residual stress and half-width
values, shot peened and
hardened ................. 5556, 58(F)
values, shot peened and
normalized .............. 5556, 58(F)
values, shot peened and quenched
and tempered ............ 5556, 58(F)
macro residual stress ratio vs. annealing
time, after shot peening .... 55, 56(F)
micro residual stresses, annealing
temperature ............................55
milling and macro residual stresses and
half-width values ............ 55, 57(F)
quenched-and-tempered, effective stressintensity factor range ............ 49(F)
quenched-and-tempered, local fatigue
strength ............................ 47(F)
quenched-and-tempered, shot peened,
effective stress-intensity factor range
....................................... 49(F)
quenching ....................... 275, 282(F)
relaxation of residual stresses after shot

peening ............................ 33(F)
residual stress relaxation, delayed .. 55(F),
59
annealing .......................... 55(F)
shot peened bending specimen and crack
initiation ........................... 34(F)
shot peening .......................... 347(F)
shot peening and macro residual stresses
and half-width values ....... 55, 57(F)
shot peening effect on bending fatigue
strength ....................... 4243(F)
S-N curves after grinding, residual stresses
effect .......................38(F), 39(F)
S-N curves and residual stresses effect
from milling ................. 3637(F)
stop ring, distortion of ......... 180(F), 181
stress-controlled loaded in push-pull tests,
cyclic deformation ............... 33(F)
stress relief heat treatment effect on
fatigue life and fatigue limit ........40
surface residual stress vs. number of
cycles .............................. 40(F)
SWT parameter vs. number of cycles to
failure .............................. 45(F)
time-temperature-transformation
diagram .......................... 248(F)
valve stem distortion ....160161, 164(F)
water quenching ....................253254
1055
for cylinder scanning induction hardened,
computer simulation .... 290, 293(F),
294(F)
grinding .......................... 150, 151(F)
1070, Almen strip material ............ 349(F)
1080
austempering .............................. 281
crack initiation ..............................34
crack initiation and macro residual
stresses .................................34
elastic chuck, distortion of .... 166, 168(F)
1095, tight scale formation ...... 255, 258(F)
1118, residual stress distribution in
carburized-and-hardened steel ... 447(F)
1137, water quenching ...............253254
1141
quench cracking .......................... 254
water quenching ....................253254
1144
quench cracking .......................... 254
water quenching ....................253254
1340, quench cracking ..................... 254
1345, quench cracking ..................... 254
1526
quenching and tempering effects on
residual stresses .......... 195, 198(F)
tempering after quenching of carburized
steels ................................. 446
tempering effect on residual
stress ............................. 447(F)
1536
quench cracking .......................... 254
water quenching ....................253254
1541
quench cracking .......................... 254
water quenching ....................253254

4024, tempering carburized cases .... 191(F)

4118, grain-boundary anomaly, carburized
steel ........................... 448, 449(F)
4130
quenching and tempering effects on
residual stresses .......... 195, 198(F)
tempering after quenching of carburized
steels ................................. 446
tempering effect on residual
stress ............................. 447(F)
4140
agitation rate effect on hardening ..... 262,
267(F)
alternating bending tests and cyclic
deformation .................. 66, 67(F)
Bauschinger effect .........................62
brazed compound with cemented carbide
tungsten-carbon ........... 395396(F)
cyclic deformation ..................... 29(F)
cyclic deformation behavior ..28(F), 29(F)
cyclic deformation behavior after shot
peening ............................ 31(F)
distortion of cylinder ................ 162(T)
half-width vs. distance from surface under
cyclic deformation ............... 67(F)
induction hardening and
grinding .................... 244246(F)
loading stress effect on shot peening
macro residual stresses ..... 61, 62(F)
loading stresses and macro residual
stresses ............................. 61(F)
macro and micro residual stress
relaxation ..............................55
macro residual stresses at surface vs.
number of cycles during alternating
bending tests ..............64(F), 65(F)
macro residual stresses in longitudinal
direction vs. distance from
surface ........................ 65, 66(F)
macro residual stresses, measured and
calculated ..................... 57, 60(F)
macro residual stresses measured vs.
modeled ....................... 6263(F)
peened and normalized, critical loading
stresses for initiation of macro
residual stress
relaxation ..................... 62, 63(T)
peened and quenched and tempered,
critical loading stresses for initiation
of macro residual stress relaxation
.................................. 62, 63(T)
peened condition, cyclic surface yield
strength ....................... 66, 68(T)
plastic strain rate vs. mean residual stress,
shot peened ........................ 60(F)
push-pull fatigue testing .............. 63(F)
quench cracking .......................... 254
relaxation of half-widths and mean
strains .................... 5657, 59(F)
shot peened and hardened, material
properties of Avrami
approach ...........................60(T)
shot peened and normalized, macro
residual stress ................ 61, 62(F)
shot peened and normalized, material
properties of Avrami
approach ...........................60(T)
shot peened and normalized, ratio of halfwidth values .................. 62, 63(T)
shot peened and quenched and tempered
at 450 C, macro residual
stresses ........................ 61, 62(F)
shot peened and quenched and tempered
at 650 C, macro residual
stress .......................... 61, 62(F)
shot peened and quenched and tempered,
material properties of Avrami
approach ...........................60(T)
shot peened and quenched and tempered,
ratio of half-width values .. 62, 63(T)
shot-peened condition, quasi-static surface
yield strength in compression .....66,
68(T)
S-N curves, shot peened vs. ground
condition ...................... 66, 67(F)
strain aging effects on cyclic deformation
and macro residual stresses ... 6566,
67(F)
thermal relaxation of macro residual
stresses, quenched, tempered, and
shot peened .............. 5657, 59(F)
total deformation and macro residual
stress ............................... 61(F)
unpeened and normalized, yield strength
ratio ........................... 62, 63(T)
unpeened and normalized, yield
strengths ...................... 62, 63(T)
unpeened and quenched and tempered,
yield strength ratio .......... 62, 63(T)
unpeened and quenched and tempered,
yield strengths ............... 62, 63(T)
volume diffusion-controlled dislocation
creep in residual stress field ........57
warm-peened condition, cyclic surface
yield strength ................ 66, 68(T)
warm-peened condition, quasi-static
surface yield strength in
compression .................. 66, 68(T)
4142
straightening of bars, distortion, and
residual stresses ................ 156(T)
4150, quench cracking ..................... 254
4320
reheat quenched and shot-peened
carburized gears .......... 443, 445(F)
shot peening effect on carburized steel
198199, 200(F), 201(F)
4337, crack propagation of autofrettaged
tube .................................... 35(F)
5100, grinding ............................... 150
5115
crack arrest and bending fatigue
limit ................................ 50(F)
shot peening effect on S-N curves .. 38(F),
39(F)
5120 (JIS SCr 420)
carburized and quenched ...... 443, 444(F)
carburized gears, residual stresses at tooth
foot surface ................ 440, 441(T)
grain-boundary oxidation ........... 448(F)
grain-boundary oxidation effect on gear
tooth fatigue strength .......... 452(F)
Index / 493
hard shot peening effect on gear fatigue
strength .............................. 452
steel hardenability effect on distortion of
carburized steels ................ 453(F)
tempering effect on carburized steel 449,
451(F)
5132, pressure rolling .................. 153(F)
5135
crack arrest .............................. 49(F)
deep rolling effect on S-N curves and
bending fatigue strength ........ 38(F)
quench-crack formation ............. 313(F)
5140
C-ring test specimen distortions ... 175(F)
hollow cylinder distortion ....... 161162,
165(F)
52100 (100 Cr6, ShKh15)
austenitizing temperature .........171172
ball screw distortion of ........ 166, 168(F)
bearing ring, prediction of deformation
during hardening ...... 317320(F,T)
for bearing rings, distortion .. 152, 154(F),
155(F)
bearing rings, quenching and residual
stress distribution ................... 321
die service life after intense quenching
327(F)
distortion of axle components ......... 160,
164(F)
hollow cylinder distortion ....... 161, 162,
164(F)
inner bearing ring 7515/02, hardened and
water cooled ........... 315317(F,T)
quenching in a hardening press of bearing
rings .............................. 185(F)
annealing .......................... 55(F)
tempering effect ................ 251, 254(F)
thermal straightening .................... 170
6150, crack initiation .........................34
8610, carburized-and-hardened, residual
stresses .............................. 439(F)
8617
carbon potential effect in carburized and
quenched gears ........... 444, 445(F)
steel grade and case depth in carburized
gears ........................ 441, 442(F)
tempering effect on gas-carburized steel
449, 450(F)
8620 (JIS-SNC 21)
aging effect on carburized steel ....... 449,
450(F)
carbonitriding ......................... 191(F)
carburizing and quenching variation
effects ............................ 443(F)
case-depth influence on carburized steels
445(F)
grain-boundary anomaly, carburized steel
448, 449(F)
repeated stressing of carburized steel
..................... 450, 451(F), 452(F)
residual stress distribution in carburizedand-hardened steels ............ 447(F)
retained austenite in carburized-andhardened steels ................. 447(F)

494 / Index
Steels, series and classes (continued)

steel grade and case depth in carburized
gears ........................ 441, 442(F)
steel grade effect on carburized
gears ........................ 441, 442(F)
8640, steel grade effect on carburized
gears ........................... 441, 442(F)
8719, carburized and hardened ........ 190(F)
8822, grain-boundary anomaly, carburized
steel ........................... 448, 449(F)
10100, hollow cylinder distortion ........ 162
10120, hollow cylinder distortion ........ 162
AZ 31, residual stress of soft metallic
material ............................. 346(F)
43BV14, carburized and quenched with
various heats ................. 443, 444(F)
C15, case hardening .................... 438(F)
C22, residual stress vs. distance from surface
with different case-hardening
depths ......................... 194, 197(F)
C80 (German grade), cyclic stress-strain
curve ............................. 2829(F)
15CD4, induction hardening of carburized
cylinder model ........ 203204, 205(F),
206(F)
CK01 (German grade), cyclic stress-strain
curve ............................. 2829(F)
CK15. See 1015.
CK22. See 1023.
CK45. See 1045.
100Cr6. See 52100.
EN9, nitriding ........................... 212(F)
EN19, nitriding .............................. 216
EN-S355, crack propagation and welding
residual stresses ..................... 35(F)
FC0202, carburization of P/M parts .... 406,
407(F)
JIS S20C. See 1019.
JIS S38C, quenching and distortion ..... 284,
287(F)
JIS S45C
coated cylinders, quenching
of ........................... 273, 278(F)
cylinder quenched, computer simulation
............................... 289, 291(F)
disk quenched, computer
simulation ................ 288289(F),
290(F), 291(F)
for gear wheel, induction hardened,
computer simulation .............. 296,
306(F), 307309(F)
quenching ............. 275, 281(F), 282(F)
quenching, agitation effect ... 276, 283(F),
284(F), 285(F,T)
quenching, clay coating, effect on quench
cracking .................... 278, 286(T)
quenching, surface oxidation and quench
cracking ................ 278(F), 286(T)
quenching, surface texture and roughness
and quench cracking tendency .. 278,
286(T)
304(F), 305307(F)
selective quenching ......278(F), 281282
steel shafts, warpage prediction by
computer simulation ........ 289290,
291(F)
surface finish effect on cooling curves

............................... 273, 279(F)
JIS SCM420
case depth influence on endurance limit in
carburized steel ................. 446(F)
core hardness effect on fatigue strength
(endurance limit) ......... 442, 443(F)
grain-boundary oxidation ........... 448(F)
hardenability effect on core hardness and
case depth .................441442(T)
quenching oil influenced on carburized
steel residual stresses .............. 445
repeated stressing effect in carburized
steel ........................ 450, 451(F)
JIS SCM 420H, distortion of gears ...... 454,
455(F)
JIS SCM435
quenching ....................... 277, 285(T)
quenching, clay coating effect on quench
cracking .................... 278, 286(T)
JIS SCM440
distortion .............................. 256(F)
304(F), 305307(F)
JIS SCr 420. See 5120.
JIS SK4
quenching ....................... 277, 285(T)
quenching, clay coating, effect on quench
cracking .................... 278, 286(T)
quenching, surface oxidation and quench
cracking .................... 278, 286(T)
quenching, surface texture and roughness
and quench cracking ..... 278, 286(T)
JIS SNC 21. See 8620.
JIS SNCM439, quenching and distortion
284, 287(F)
K13NiCr12, case hardening ........... 438(F)
MA956
mechanical alloying ...................... 414
microstructure and residual
stresses .........................419420
90MoCrV8, fluidized-bed quenching ... 263,
267(F)
60NCD11, dilatometer curves ...... 249250,
251(F)
S35, induction hardening .................. 229
S15CK. See 1015.
S690QL1, crack propagation ............ 35(F)
SF-Cu F20, residual stress of soft metallic
material ............................. 346(F)
ShKh15. See 52100.
SS44, distortion after straightening ... 156(F)
STE460 (P460N), welded plate, shot peening
effect on residual stress .... 354, 356(F)
Steel 14KhGSN2MA, carburized bushing
having high-rate
quenching ................. 320321(F,T)
Steel 35Kh2NgSM, austenite grain size
effect .......................... 325, 326(F)
Steel 40, austenite grain size effect ...... 325,
326(F)
Steel 40Kh2NgSM, austenite grain size
effect .......................... 325, 326(F)
Steel ShKh15. See 52100.
Steering sector shaft, distortion .. 454455(F)
Stefan-Boltzmann coefficient ............... 375
Step-by-step time-integration method .... 288

Step-quenching tanks ........................ 172
Stereolithography (SLA) .................... 413
Stick-slip effect ................................ 429
of supporting bar ............................ 429
Stiffness ......................................... 288
Stiffness equation ............................. 377
Stoneys formula ..................119, 152, 157
Stop ring, distortion of ............. 180(F), 181
Straightening ...60, 90, 150, 151, 154(F), 220
of bars ......................................... 150
and distortion ...................... 156(T), 157
parallel-roll ......................... 156157(F)
Strain
chemical measurement methods .......... 110
due to differential contraction ............. 362
due to volume change ...................... 217
measurement methods ................109110
Strain aging effects, cyclic deformation ....65,
66, 67(F)
Strain gages ......................................91
to measure residual stress magnitude .... 238
Strain gage technique .................. 100, 438
Strain gaging/sectioning technique ... 103(F)
Strain hardening ......................19(F), 109
effect on fatigue strength ................ 19(F)
and P/M processing ........................ 416
in roll compaction ........................... 415
Strain-hardening parameter ......... 299, 300
Strain peening ................................. 151
Strain rates ...............................285287
computation of ............................... 202
Strain rate tensor ............................. 202
Strain relaxation, measurement of ...........91
Strength-differential effect ....................61
Stress amplitude, and crack initiation .......34
Stress-concentration factor ................. 420
and bending fatigue strength ....... 3637(F)
Stress corrosion .......................... 16, 100
Stress-corrosion cracking (SCC) .....72, 110,
150, 152
active-part corrosion (APC) .................79
alkaline and acid solutions for failure ......79
alloying elements contribution ..............80
alloying elements effect on resistance to ..80
crack growth mechanism .....................79
crystal structure role .....................8081
estimation of sensitivity to, criteria and
methods for .........................7980
high-strength steels .................. 7982(F)
hydrogen-induced corrosion cracking
(HISCC) ..................................79
insensitivity of high-pure metals ............79
of low-alloy steels ................... 7982(F)
resistance by shot peening in chemical
industry applications ........... 353354,
356(F,T)
selectivity of material relative to a corrosion
medium ...................................79
steels for main gas-pipe lines ...... 8182(F)
test methods .................................. 110
thermal processing role .............. 8081(F)
Stress-corrosion hydrogen-induced
cracking ................................... 110

Stress-corrosion testing, preparation of test

specimens for weldments ............... 110
Stress-free interplanar spacing, for a
coating ..................................... 122
Stress gradient ..18, 36(F), 37(F), 40, 51, 101,
111
at notch root ...................................36
Stress in a simple beam .......................92
Stress in material, symbol and
units of ...................................93(T)
Stress intensity coefficient, threshold
value ............................... 73, 74, 75
Stress-intensity factor .....................2930
for normal breakoff ......................... 128
for transverse shear ......................... 128
Stress-intensity factor range
(Mode I) ................................2930
Stress loading, and austenite transformation in
carburized steels ....................447448
Stress mapping ................................ 100
Stress measurement temperature ......... 118
Stress peening ....................... 345, 348(F)
Stress profile, by Peenstress software .. 353(F)
Stress raisers ............................. 401, 405
Stress-reconstruction procedure (series
method) ................................... 108
Stress relaxation ......................... 17, 150
for changing or disappearing residual
stresses ....................................28
induced by phase transformations in
induction-hardened and carburized steel
cylinder ....................... 204, 206(F)
residual stress, prediction of ........ 2223(F)
resistance to ...................... 5859, 60(F)
and tempering ...................... 284, 287(F)
uneven, and distortion resulting from .... 156
welding of maraging steel .......... 6061(F)
Stress relieving ..95, 110111, 120, 224, 251,
254(F)
for avoiding changes in dimensional
stability ...................................16
castings ....................................... 168
dimensional stability influenced by ...... 339
and distortion ................................ 166
to eliminate residual stresses .............. 144
by tempering ....................... 284, 287(F)
Stress risers .....254, 255, 256, 257, 260, 264,
278
Stress Scan 500 equipment ................. 156
Stress/strain analysis ..................303304
Stress-strain constitutive
relation .......................... 298302(F)
Stress-strain curves
of castings ................................ 373(F)
under strain rates for castings ......... 374(F)
Stress-strain-induced transformation .... 296
Stress-strain state, computation
of ................................. 313315(F)
Stress tensors .................................. 121
Stress-transient method .............. 57, 60(F)
Stretch forming, and distortion ......... 169(F)
Stretching, for inducing residual stress
relaxation ....................................54
Strip, rolling of ............................ 147(F)
Strip continuous casting by twin-roll method,
residual stress formation ........... 383385,
386(F), 387(F), 388(F), 389(F)
Structural dilatation ...................286288

due to phase transformations, strain rates
for ....................................... 312
due to transformation plasticity, strain rate
for ....................................... 312
Structure stress, as cause of
distortion ............159161(F,T), 162(F)
Subcritical annealing ......................... 254
Subgrains ....................................... 333
Subsonic gas atomization ............... 400(F)
Subzero cooling, and residual stresses ..... 195
Sulfides
as fracture origin sites ........................75
role in stress-corrosion cracking ............81
Sulfide stress cracking .........................83
Sulfur
steel .......................................83
crack ..................................76(T)
Sulfur (max), composition requirements for
Sunk tubes ................. 147(F), 148149(T)
Superalloys .......................................73
Supercarburized high-speed steels
austenitizing .............................. 337(T)
cryogenic cooling ....................... 337(T)
quenching ................................ 337(T)
tempering ................................. 337(T)
Vickers hardness ........................ 337(T)
Superhardening ............................... 328
Superplasticity, of bearing ring ............. 318
Superposed model of plasticity and
creep ....................................... 302
Superposition model ............... 373, 374(F)
Superstrengthening, by intensive quenching
326
Surface crack formation .................... 215
Surface integrity .............................. 221
Surface nanocrystallization ...................25
Surface oxidation ......... 273, 277(F), 278(F)
Surface quenching with a phase change,
mechanical, thermal, or structural origins
Surface texture
and quench cracking ........277278, 286(T)
and quenching ..................... 273, 279(F)
Surface values ................................. 108
Surface wetting ................................ 258
Surface work hardening ..................... 198
Swell ............................................. 363
SYS-WELD (FRAMATOME Company)
(computer program) ............. 202, 303
T
Tamahagane (Japanese steel) ............... 309
Tangential internal residual stresses, from
induction hardening ..235(F), 236, 238(F),
239(F)
Index / 495
Tangential stress .............. 144(F), 145, 146
380(F)
Tank chain, shot peening effect on fatigue
life ........................................20(T)
Tatara system .................................. 309
Taylors expansion, for viscoplastic strain
increment of element at Kth time
step ......................................... 376
Taylor/von Mises criterion .................... 6
Temperature
carburizing heat-transfer model .....199200
cooling process changes .......... 313314(F)
effect on chemical and mechanical driving
forces ................................... 6(F)
effect on distortion in welds .............. 7(T)
effect on rail steel residual stresses ....... 434
effect on specific heat capacity of pure
iron .................................. 34(F)
induction hardening of carburized steel
cylinder ....................... 203, 205(F)
interaction with deformation and
quenchant ....... 268(F), 269(F), 270271(F)
and residual stress level in nitriding ...... 214
rough-condition .............................. 325
semicontinuous casting profile .. 378, 379(F)
Temperature-dependent yield function .. 286
Temperature factor, for third-order residual
stresses ..................................... 235
Temperature field, nonsmoothness or
unevenness ................ 322, 323, 324(T)
Tempered martensite .................. 195, 336
after sinter hardening of P/M parts ...... 408
Temper embrittlement .........................76
Tempering ................282284, 287(F), 336
carburized-and-hardened steels ..445446(T)
and carburizing procedures ...... 190, 191(F)
dimensional variation and retained austenite
content ........................ 251, 254(F)
and distortion .. 154, 177178(F), 283284,
287(F)
effect on crack toughness ........... 8081(F)
effect on residual stress distribution ..... 195,
198(F)
effect on residual stress of carburized steels
449, 450(F), 451(F)
of marquenched steels ............ 280, 287(F)
microstructural variation effect on volume
changes ....................... 251, 253(F)
niobium alloying effect .................... 325
P/M parts ........................... 407, 408(F)
P/M steel parts ........................417418
and retained austenite content in carburized
steels ................................ 447(F)
stages .......................... 251, 253(F), 283
stress relief ......................... 284, 287(F)
transformation process prediction ........ 202
volume changes .............283284, 287(F)
Tensile applied stress ....................... 6(F)
Tensile residual stresses ..... 193, 196(F), 211
and brazing ......................394(F), 395(F)
from welding .......... 392(F), 393(F), 394(F)
Tensile strength .................................16

496 / Index
Tension testing ...................... 251, 254(F)

Termite welding, of rail steel ........... 434(F)
Test methods, stress-corrosion cracking ... 110
Testpieces, carburized, variation of residual
stresses ................. 439, 440(F), 441(T)
Tetragonality degree ......................... 338
Tetragonal martensite ....................... 338
tempering of ................................. 224
Theory of regular conditions ............... 322
Thermal conductivity ..................... 3, 312
steels ................................... 3(T)
nonlinear equations ......................... 315
nonstationary, equation for ................ 312
Thermal convection and radiation, in
carburizing ................................ 200
Thermal cycles, as description of surface heat
treatment process ......................... 223
Thermal cycling, for inducing residual stress
relaxation ....................................54
Thermal diffusivity coefficient ............. 322
Thermal expansion coefficient
differences affecting brazing residual
stresses .....................394(F), 395(F)
Thermal expansion mismatch ................12
Thermal expansivity ............................ 3
steels ................................... 3(T)
Thermal gradients, nonuniform ............ 332
Thermal growth ............................... 118
Thermal inversion in workpiece
quenching ............................. 223(F)
Thermal load .................................. 288
Thermally activated processes, for inducing
residual stress relaxation ...... 5460(F,T)
Thermal oscillations, influence eliminated in
microstrains ............................... 335
Thermal plasticity ............................ 313
Thermal radiation coefficient, for centrifugal
casting pipe models ...................... 382
Thermal shocking, ............................ 156
Thermal shrinking stresses ................. 191
Thermal spraying ....................... 118, 120
Thermal straightening ....................... 170
Thermal strain .......................... 286, 298
Thermal strain rate .................... 202, 312
Thermal stress
of carburized bushing due to high-rate
quenching ................. 320321(F,T)
castings, segregation in ..................... 167
as cause of distortion ..........159161(F,T),
162(F,T), 163(F), 164(F), 165
cooling with transformation ........ 250251,
252(F), 253(F,T)
nonuniform ................................... 260
and quench severity ......................... 257
Thermal-stress analysis, to evaluate residual
stresses in P/M products ................ 420
Thermal surface treatment .......... 19, 20(F)
Thermochemical surface treatment ........19,
20(F)
Thermocouples ................................ 223
for measuring temperature on gear-wheel
teeth, induction hardening ....... 230(F)
and temperature monitoring to control

distortion ............................... 184
and time variation of temperature of gearwheel teeth in induction
hardening ........................... 230(F)
Thermodynamics ................................ 3
Thermoelasticity effect, on rail steel ....... 426
Thermokinetic diagram of the supercooled
austenite transformations ............. 313
Thermomechanical analysis, for solidification
Thermomechanical state, during and after
surface induction heating ............... 226
Thermomechanical theory .................. 372
Thermoplasticity, of carburized bushing with
two-step quenching ......... 320321(F,T)
Theta-method .................................. 304
Thickness of beam/plate/tube wall, symbol
and units of .............................93(T)
Thin-slab casting process, residual stress
formation ....... 383385, 386(F), 387(F),
388(F), 389(F)
Three-point bending ......................... 155
Three-time rule ................................ 109
Threshold (critical) stress-intensity factor
(KISCC), ......................................80
Threshold effect .................................30
Threshold stress valve .........................80
Through-hardened steels, contact
fatigue ...................................... 198
Through-hardening .......................... 316
P/M parts ..................................... 406
Through-thickness temperature, from
induction hardening ............ 223224(F)
Tight scale formation ........... 255, 258(F)
Tikhonovs method ........................... 127
Time
of carburizing and fatigue strength ....... 446
interaction with deformation and
and residual stress level in nitriding .. 214(F)
Time discretization, in unsteady-state casting
analysis .................................... 376
Time function of temperature
change ........................... 321322(F)
Time of diffusion, and hydrogen critical
concentration ................................75
Time quenching ......................... 280, 281
Time-temperature-transformation (TTT)
diagrams ........... 191, 248(F), 249, 250,
251(F), 288, 297
computation of, during hardening
processes ............................... 200
Time till a crack initiation ....................80
Tin
steel .................................. 83, 84
76(T)
Tin-bismuth alloy, strip continuous casting by
twin-roll method .......................... 383
Titanium
alloying effect on susceptibility to austenite

grain growth ..................... 325, 328
alloying influence on delayed fracture of
steel ................................... 78(F)
Poissons ratio .............................96(T)
shear modulus .............................96(T)
crack ..................................76(T)
Titanium alloys, macro and micro residual
Titanium carbide coating ................... 118
Titanium hydrides, formation at grain
boundaries ...................................78
Titanium nitride ....118, 119, 120(F), 122(F),
209
Titanium nitride coatings ............... 121(F)
Tool gap hardening .......................... 241
Tool steels
contact wear resistance improved ......... 420
cryogenic cooling ............. 340(F), 341(T)
heat treatments applied for dimensional
stabilization .............. 338(F), 338(T)
martensitic transformation ....... 332, 333(F)
P/M processed, applications ........404405
P/M processed, heat treatment of ....... 408,
417418(F)
P/M processing .............................. 400
sintering ......................404405, 406(T)
spray forming ..........................412413
Tool steels, series and classes
ASP30 (P/M tool steel)
out-of-roundness measurements ... 418(F)
D2, spray forming ........................... 413
H13
direct-metal deposition .................. 419
rapid prototyping ......................... 413
spray forming ............................. 413
M2, out-of-roundness
measurements ...................... 418(F)
M20, high-speed steel for R6M5
punches ....................... 326(F), 327
T12, distortion ............................... 159
T15
microstructures of convertional vs. P/M
processed ........................ 418(F)
P/M processing .............................. 402
spray forming ................................ 413
Tool steel UHB Chipper ................. 156(F)
Torque rods, shot peening effect on fatigue
life ........................................20(T)
Total case depth (TCD) ..................... 437
Total strain increment ....................... 217
Total strain rate .............285, 300, 312, 373
Total strain Wohler curve ................ 30(F)
Tracked and road vehicles, shot peening for
improving vibratory fatigue
properties .................................. 345
Traction-compression loading ..... 22, 23(F),
24(F)

Transformation eigenstrain, of rail

steel ........................................ 427
Transformation induced plasticity
(TRIP) ...................... 90(F), 427428
Transformation plasticity ............36, 201,
249250, 251(F), 286288, 302(F),
304307(F)
and anisotropic strain ...................... 7(F)
definition ..................................... 288
distortion of steel ring, computer
simulation .................... 290, 292(F)
of gears ............................. 456457(F)
Transformation plasticity coefficient
for martensitic or pearlitic
transformation .......................... 288
Transformation plasticity effect ........... 296
Transformation plasticity strain rate .... 202,
288
Transformation plastic strain .... 301(F), 302
Transformations
deformation systems associated with ... 5(T)
diffusionless, in
metallo-thermo-mechanics ....... 297(F)
diffusion-type, in
metallo-thermo-mechanics ....... 297(F)
displacive ..............................45(F), 9
mechanical driving force ...................... 6
stress components interacting with volume
change ..................................... 6
intervariant ...................................... 6
reconstructive ............................ 45(F)
stress components interacting with volume
change ..................................... 6
shape change due to ....................... 5(T)
stress-assisted, limits to ................... 6(F)
stress-(or strain) induced ................. 296
unconstrained, and shape changes ....... 5(F)
under influence of a mild stress .............. 8
Transformation-start temperature ....... 6(F)
Transformation stress ..........159160, 161,
163(F,T), 164(F), 191
of carburized parts .......................... 178
castings, segregation in ..................... 167
as cause of distortion ....................... 165
cooling with transformation ........ 250251,
252(F), 253(F,T)
Transformation superplastic strain ....... 302
Transformation temperature range ...... 192
Transformation volume strain ................ 5
Transgranular fatigue crack initiation ... 197
Transgranular spell mechanism .............79
Trapezium method ........................... 129
Trapezoid method ............................ 304
Traps ......................................... 71, 72
for hydrogen ......................... 75, 76, 82
TRAST (software program) ................ 202
Trepanning ...............................110111
Tresca criteria ...................................17
Tresca yield function ................... 299, 302
Treuting and Read method .............. 89, 96
measurement ....................... 102(T)
Triaxial compression, by simultaneous
isostatic and uniaxial compression .... 402
Triaxial residual stress state, and crack
propagation .............................. 35(F)
Triaxial stress analysis ...................... 122

Triaxial stress state without shear stress, of
coatings ..............................120121
Triaxial stress state with shear stress, of
coatings .................................... 121
TRIP. See Transformation induced plasticity.
TRIP (transformation) induced plasticity
effect ..................................... 90(F)
Troostite .................................. 438, 448
Truck leaf springs
advantages ................................ 327(T)
Tubes
autofrettaged, and crack propagation ... 35(F)
procedures .......................... 102(T)
drawing ............. 141, 142(F), 145146(F)
sunk, deep drawing ..... 147(F), 148149(T)
Tungsten
Poissons ratio .............................96(T)
shear modulus .............................96(T)
Tungsten carbide-carbon cemented carbide
in brazed joint with steel ......... 395396(F)
brazed compound with steel residual
stresses .............................. 395(F)
Tungsten carbide-cobalt
coating ...........................123124(T)
thermal spraying ....................... 118, 123
Tungsten carbides
cooling ..................340341(F), 343
Tungsten inert gas welding (TIG) ..... 392(F)
of maraging steel ..................... 6061(F)
Turning ............... 150151, 152(F), 153(F)
and distortion ................ 152, 155(F), 156
residual stresses in alloy steels ........ 109(F)
12% chromium steel, thermally activated
relaxation after shot peening .... 56, 59(F)
Twin boundaries, and crack initiation ..3334
Twinning ........................................... 5
Twin-roll method of strip continuous
casting, residual stress formation .... 383
385, 386(F), 387(F), 388(F), 389(F)
Two-angle technique (DET) ................ 112
Two-dimensional finite difference
method .................................... 268
Two-dimensional inverse method ......... 268
Two-stage boost/diffuse method of vacuum
carburizing model .........203(F), 204(F)
Two-stage nitriding ........................... 170
Two-step quenching ......... 176, 312, 319(F),
320321(F,T), 328
Twyman-Green interferometry ..............12
U
U-bend stress corrosion test, specimen
making and using ........................ 110
Index / 497
UIC54E rail steel profile .........424, 428,429
Ultimate strength, of bearing ring being
hardened ............................... 318(F)
Ultrahigh temperature oil quenching, of ball
bearing race ..................... 454, 455(F)
Ultramicroscopic stresses (third-order
residual stresses) ..331, 332(F), 335336
Ultrasonic gas atomization, ....... 400(F), 414
Ultrasonic inspection, to evaluate residual
stresses in P/M products ................ 413
Ultrasonic instrumentation ................. 100
Ultrasonic methods ........................... 109
Ultrasonic shot peening ........................25
Ultrasonic velocity ............... 112, 113114
Ultrasound, ..........................13, 14(F), 25
Unalloyed steels, nitriding ................... 209
Undeformed material, interplanar spacings
and residual stresses ..................... 333
Undercooling, of austenite .............331332
Uniaxial compression, as mechanical driving
force coefficient ......................... 6(T)
Uniaxial deformation .............. 6063(F,T)
for inducing residual stress relaxation ....54,
6063(F,T)
Uniaxial loading, resistance to residual stress
relaxation ........................... 6162(F)
Uniaxial stress .................................... 6
Uniaxial tension, as mechanical driving force
coefficient ................................ 6(T)
Unidirectional compaction .............. 401(F)
Unified constitutive theory .................. 372
Uniform casting, .............................. 377
Universal joint shaft, shot peening effect on
fatigue life ...............................20(T)
Unsteady-state centrifugal casting
finite-element method formulation ....... 376
and residual stress formation ... 379383(F),
384(F), 385(F), 386(F)
Upcut milling ................................ 39(F)
producing residual stresses .............. 32(F)
residual stresses generated ............... 36(F)
Upper bainite .................. 250(T), 257, 448
Upsetting forging ............................. 410
V
Vacuum atmosphere ......................... 403
Vacuum carburizing ......................... 189
furnace ........................................ 453
Vacuum extraction ........................... 412
Vacuum quenching ........................... 172
Valve, shot peening effect on fatigue
life ........................................20(T)
Vanadium
growth ............................... 80, 81
crack ..................................76(T)

498 / Index
Vanadium carbides (VC)

cooling ..................340341(F), 343
resistant to dissolution ................ 325, 328
temperature for full dissolution in 30 2
steel ..................................... 325
Vanadium nitrides (VN)
resistant to dissolution ................ 325, 328
temperature for full dissolution in 30 2
steel ..................................... 325
crack ..................................76(T)
Vapor blanket cooling stage ...258259, 268,
269, 272(F), 273(F), 274, 439
Vapor blanket cooling stage of quenching
oil ................................. 455456(F)
Vaporizing, effect on carburized steels .... 438
Variational principle ................... 127, 376
Variation of residual stresses in test pieces
and gears ....................... 439, 440(F)
Versailles project on Advanced Materials
and Standards (VAMAS) ............. 413
Vertical semicontinuous direct-chill casting
process ......377379(F), 380(F), 381(F),
382(F)
Vestergards asymptotic relationships .... 130
Vibration, for inducing residual stress
relaxation ....................................54
Vibration compaction, ....................... 402
Vickers hardness
decreasing as relative shrinkage increases
during tempering .................. 338(F)
Video .....to measure surface wetting ..... 269
Virtual work principle ....................... 377
Viscoplastic flow laws ........................ 365
Viscoplasticity ................................. 302
unified theories of .....................373374
Viscoplastic strain rate ..........202, 366, 374
Viscoplastic stress analysis .............. 366(F)
Viscosity
of quenching media, and carburizing .... 445
Viscous cup fracture ...........................76
Viscous fluids .................................. 372
Voids, in castings .............................. 370
Volterra equation of the first type 126, 127,
128
Volterra integral equation .....126(F), 136(F)
Volterra integral equation method ........ 133
design diagrams ......................... 126(F)
Volvo ..............................................11
W
Warm compaction .................. 399, 413(F)
Warm compaction and sinter, cost,
relative ................................. 413(F)
Warm-peening .................. 65(F), 66, 67(F)

and cyclic deformation ...........64(F), 65(F)
fatigue ...........................................27
Warpage, of steel shafts, computer simulation
289290, 291(F)
Wartisla ...........................................11
Water
content effect in salt-bath
quenching .... 262263, 266(F), 267(F)
as quenchant ..........262264, 269280(F),
282(F), 283(F), 284(F), 285(T), 286(T)
Water atomization ............ 400(F), 412, 414
Water-based polymer quenching tanks .. 172
Water jet quenching of steels .............. 327
Water quenching ... 264, 269280(F), 282(F),
283(F), 284(F), 285(F), 286(T)
agitation effects .......................... 283(F)
computer simulation ... 288289(F), 290(F),
291(F)
under pressure ......................... 315, 316
Wear .................................. 16, 150, 189
Wear-resistant high-speed steels containing
34% V
composition .............................. 406(T)
designation ............................... 406(T)
hardness .................................. 406(T)
Weq ........................................ 406(T)
Welded assembly, crack propagation speed
affected by residual stresses .......... 15(F)
Welded joints
fatigue strength and residual stress ..... 15(F)
stress relief of .............................. 15(F)
Welding ...........................................99
crack propagation and residual
stresses ........................35(F), 36(F)
and distortion ...................... 391394(F)
effect on rail steel residual stresses ... 434(F)
longitudinal and transverse residual stresses
on aluminum plate .......392(F), 394(F)
macro residual stresses set up ...... 6061(F)
multipass ..................................... 392
phase transformations and residual
stresses .............. 392393(F), 394(F)
P/M parts ..................................... 405
quenching processes and residual
stresses ............................ 392, 394
repair .......................................... 392
shrinkage and impeded shrinkage due to
temperature distribution ... 391(F), 392,
393(F), 394(F)
tensile or compressive yield strength as
function of temperature .... 393394(F)
tungsten inert gas ....................... 392(F)
Weldments
procedures .......................... 102(T)
distortion dependent on temperature of
transformation ........................ 7(T)
hydrogen-induced stress cracking ........ 110
pipe ............................................ 102
residual stresses measurements ...99, 100(T)
Rosenthal and Nortons method,
variation of ............................. 105
shot peening effect on residual stress ... 354,
356(F)
steel, neutron diffraction weldments of .. 113

stress corrosion test specimen
preparation ............................. 110
Wet bag tooling ............................... 402
Wetting, period of (tw), ............. 258, 261(F)
Wetting front .................. 258, 260(F), 269
Wetting kinematics ........................... 259
measurement methods .. 269, 272(F), 273(F)
Wetting process ........ 268, 269, 270, 272(F),
273(F)
and quenchant selection .................... 275
Wetting time ............................. 259, 272
Wetting velocity .. .. 269, 270, 271272, 274
Wicking ......................................... 412
Widmanstatten ferrite. See Ferrite,
Widmanstatten.
Williams functions ........................... 130
Wire drawing ..................141, 145146(F)
Wohler diagram (S-N diagram) ..............30
Wohler stress curves .................. 17, 30(F)
for induction hardened, carburized, or
quenched-and-tempered steels .. 244(F)
for slip band formation, microcrack
initiation, and failure ............... 28(F)
Work hardening ................................55
and deformation systems ...................... 6
effect on fatigue life in low-strength
steel .......................................40
and fatigue strength ................. 27, 28, 52
parameter ............................... 299, 300
P/M parts ........................... 407, 408(F)
Workpiece size, influence on residual
stresses ........................... 195, 198(F)
Work-softening processes, and fatigue .....27,
28(F)
www.residualstress.com ..................... 100
X
X-ray diffraction (XRD) ......99101(F), 104,
124, 144, 146, 339, 438439(F)
brazed metal-ceramic compounds .... 395(F)
for calculating longitudinal residual stresses
in Japanese sword ........309(F), 310(F)
combined with electropolishing ........... 109
for determining residual stress variations
after induction hardening ............ 233
for evaluating brazed joint of steel and
cemented carbide ...................... 395
for evaluating residual stresses after
grinding ............................. 151(F)
for evaluating residual stresses after
straightening ........................... 156
for evaluating residual stresses in P/M
products .....................413, 415, 418
of high-alloy tool steels .................... 342
for measuring actual crystal
dimension ........................111113
for measuring axial residual
stresses ........................ 236(T), 238
for measuring compressive stresses in rail
steel ..................................... 431

for measuring computerized quenching of

cylinder ....................... 302(F), 305
for measuring initial residual stress
distribution ........................... 22(F)
for measuring plate weldment strain ..... 106
for measuring residual stress distribution
after quenching and tempering ..... 284,
287(F)
for measuring residual stress distribution in
scanning induction hardened
cylinder ....................... 290, 294(F)
for measuring residual stresses ... 1113, 18,
54, 211, 212(F), 217, 223
for measuring residual stresses
in coatings ..............118, 120122(F)
for measuring residual stresses in quenched
steel disk ..................... 289, 290(F)
for measuring retained austenite distribution
in carburized-and-hardened
steels ................................ 447(F)
X-ray diffraction profiles ..............351352
X-ray diffractometer, miniature ........112(F),
113(F)
X-ray diffractometry ......................... 333
X-ray electron spectroscopy, grain boundary
embrittlement ...............................78
X-ray interference lines

to determine micro residual stresses .......55
and micro residual stresses and cyclic
deformation ..............................67
X-ray profile analysis, micro residual stresses
relaxation behavior .........................63
Index / 499
materials ............................ 335(T)
Yttrium oxide/partially stabilized zirconia
coating ..................................... 118
Yuriev formulas ............................... 313
Z
Y
Yield function ..................299300(F), 301
of casting process ........................... 365
Yield point ...................................... 317
Yield strength
of austenite after welding ........ 393394(F)
of bearing ring being hardened ....... 318(F)
steels ................................... 3(T)
temperature effect ................. 251, 254(F)
Yield stress
of casting process ........................... 365
in cold forming .............................. 148
Youngs modulus ................... 142(F), 145
ZDT. See Zero ductility temperature.

Zener-Wert-Avrami function, time effect on
residual stress relaxation ......... 56, 58(F)
Zero ductility temperature (ZDT) ........ 363
Zero strength temperature (ZST) ......... 363
Ziegler model of back stress
rate ............................... 299300(F)
Zinc, alloying element effect on crack
growth .......................................80
Zirconium, alloying effect on susceptibility to
austenite grain growth ............. 325, 328
Zirconium oxide (ZrO2)
brazing ...........................394(F), 395(F)
Zone segregation .............................. 168
ZST. See Zero strength temperatures.
ASM International is the society for materials
engineers and scientists, a worldwide network

dedicated to advancing industry, technology, and
applications of metals and materials.
ASM International, Materials Park, Ohio, USA
This publication is copyright ASM International. All rights reserved.
Publication title
Product code
Handbook on Residual Stress and Deformation of Steel
06700G
To order products from ASM International:

Online Visit www.asminternational.org/bookstore
Telephone 1-800-336-5152 (US) or 1-440-338-5151 (Outside US)
Fax 1-440-338-4634
Mail
Customer Service, ASM International

9639 Kinsman Rd, Materials Park, Ohio 44073-0002, USA
Email CustomerService@asminternational.org
American Technical Publishers Ltd.

27-29 Knowl Piece, Wilbury Way, Hitchin Hertfordshire SG4 0SX,
In Europe United Kingdom
Telephone: 01462 437933 (account holders), 01462 431525 (credit card)
www.ameritech.co.uk
Neutrino Inc.
In Japan Takahashi Bldg., 44-3 Fuda 1-chome, Chofu-Shi, Tokyo 182 Japan
Telephone: 81 (0) 424 84 5550
Terms of Use. This publication is being made available in PDF format as a benefit to members and
customers of ASM International. You may download and print a copy of this publication for your
personal use only. Other use and distribution is prohibited without the express written permission of
ASM International.
No warranties, express or implied, including, without limitation, warranties of merchantability or
fitness for a particular purpose, are given in connection with this publication. Although this
information is believed to be accurate by ASM, ASM cannot guarantee that favorable results will be
obtained from the use of this publication alone. This publication is intended for use by persons having
technical skill, at their sole discretion and risk. Since the conditions of product or material use are
outside of ASM's control, ASM assumes no liability or obligation in connection with any use of this
information. As with any material, evaluation of the material under end-use conditions prior to
specification is essential. Therefore, specific testing under actual conditions is recommended.
Nothing contained in this publication shall be construed as a grant of any right of manufacture, sale,
use, or reproduction, in connection with any method, process, apparatus, product, composition, or
system, whether or not covered by letters patent, copyright, or trademark, and nothing contained in this
publication shall be construed as a defense against any alleged infringement of letters patent,
copyright, or trademark, or as a defense against liability for such infringement.

Handbook of Residual Stress and Steel Deformation

Transféré par

Informations du document

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Handbook of Residual Stress and Steel Deformation

Transféré par

Droits d'auteur :

Formats disponibles

2002 ASM International. All Rights Reserved.

Handbook of Residual Stress and Deformation of Steel (#06700G)