Lecture Notes

PART A: ENGINEERING FRACTURE MECHANICS

AND

PART B: FATIGUE ANALYSIS

Prepared by
Dr. S. M. Murigendrappa
Department of Mechanical Engineering
National Institute of Technology Karnataka, Surathkal

14-08-2019
 Greatest Memories with Eminent Personalities

With Late Prof. Paul C. Paris With Prof. James R. Rice

Developed fatigue crack growth model. Developed theoretical concept of J-integral for
elastic-plastic materials.

With Prof. S.K. Maiti With Prof. J.N. Reddy

One among a few Indian legends in FEM and Contributor in the Theory of Solid Mechanics and 2
Fracture Mechanics. FEM
 CONTENTS

❖ Part 1- Role of Failure Prevention Analysis in Mechanical Design

❖ Part 2- Review of Stress and Strain
❖ Part A: Engineering Fracture Mechanics
 Introduction
 Fracture Criteria and Stress Analysis of Cracks
 Fracture Testing
 Plasticity Effects
 Elastic-Plastic Fracture Mechanics
 Computational Fracture Mechanics

❖ Part B: Fatigue Analysis

 Introduction
 Uniaxial High-cycle Fatigue
 Low-cycle Fatigue
 Fatigue Growth Laws
 Notches and its Effects
 Multiaxial High-cycle Fatigue
 Fatigue Design Methodologies
 Fatigue of Non-metallic Materials
3
 Nomenclatures
 - Normal stress u- translational displacement in x -direction
ys – yield stress S – Surface energy per unit area
- Shear stress
- Shear modulus KI - Stress intensity factor of Mode I loading
 - Normal strain KcI – critical Stress intensity factor of Mode I loading
 - Shear strain a- crack length
- Potential energy of body with crack ac- critical crack length
0 - Potential energy of body with out crack ai - initial crack length
A - Projected crack area
E- Young’s modulus af- final crack length
G- Energy release rate aeff- effective crack length
Gc – Critical energy release rate N- Fatigue cycles
U – Strain energy Sf – Fatigue limit
W ext – Work potential Nf – fatigue life
W – Width
B - thickness SNF – Fatigue strength
P- Load Sm- mean stress
Pmax- Maximum load Sa- alternating or amplitude stress
PQ- critical load S – stress range
/2 = a - Total strain amplitude Smax – maximum stress
e/2 – elastic strain amplitude = S/2E = Sa/E
p/2 – plastic strain amplitude = /2 - e/2 Smin – minimum stress
'f – Fatigue ductility coefficient R – Stress ratio
S'f – Fatigue strength coefficient

4
 Part 1-Role of Failure Prevention Analysis in Mechanical Design

Mechanical Failure:
• Mechanical failure may be defined as any change in size, shape or material
properties of a structure, machine or machine part that renders it incapable
of satisfying performing its intended function.

• It is the primary responsibility of any mechanical designer to ensure that his

or her design functions as intended for the prescribed design life-time and
at the same time, that is competitive in the market place.

• Success in designing competitive products while averting premature

mechanical failures can be consistently achieved only by recognizing and
evaluating all potential modes of failure that might govern the design.

• To be effective in averting failure, it is essential for a designer to have a

good working knowledge of analytical and/or empirical techniques of
predicting failure so that failure during the prescribed design life may be
prevented.

• Thus, it is clear that failure analysis, prediction, and prevention are of

critical importance to any designer.
Stress Analysis
• Stress analysis is an engineering discipline that determines the stress in
materials and structures subjected to static or dynamic forces or loads.

• A stress analysis is required for the study and design of structures, under
prescribed or expected loads.

• Stress analysis may be applied as a design step to structures that don’t yet
exist.

• Aim of the analysis is usually to determine whether the member or collection

of members, usually referred to as a system, behaves as desired under the
prescribed loading.

• For example, this might be achieved when the determined stress from the
applied forces(s) is less than the tensile yield strength or below the fatigue of
the material.

• Analysis may be performed through experimental testing techniques, through

analytical mathematical modelling or computational simulation or a
combination of methods.
• Stress analysis may be carried out based on:
1) Analytical Methods
i) Strength of Materials
ii) Solid Mechanics:
Theory of Elasticity
Theory of Plasticity
Theory of thermo-elasticity
Engineering Fracture Mechanics

2) Numerical Methods
i) Finite Element Methods
ii) Boundary Element Methods
iii) Mesh-free Element Method
iv) Finite Difference Method

3) Experimental methods
 Images of Stress Distribution in Simply Supported
Beams subjected to Point Load applied at the Centre

Solution is based on strength of material approach for the cases of

uniform cross-section geometries.
 Solution is based on Stress
Concentration approach for the
case only when hole geometry is
comparable with depth of beam

Enlarged view showing the

distribution of stress around
circular hole
 Solution is based on theory of
elasticity approach for the case by
treating the hole geometry is very
small compared with depth of
beam.

Enlarged view showing the

distribution of stress around
circular hole
 Solution is based on fracture
mechanics approach.

Enlarged view showing the

distribution of stress around crack
zone.
 Part 2- Review of Stress and Strain
STRESS
• Consider a body shown in Fig. (2.1), is in equilibrium.
• Under the action of external forces, P1, P2, … Pn, as surface forces and/or body
forces, internal forces will be produced between the parts of the body.
• Knowledge of the internal forces at all points in the body is essential because
these forces need to be less than the strength of the material used in the
structure.
• To study the magnitude of these forces at any point, o , let us imagine the body
is divided into two parts 1 and 2, by a cross-section through this point.
• It will be assumed that the internal forces are continuously distributed over this
cross-section.
• Magnitude of such forces is usually defined by their intensity, i.e., by the amount
of force per unit area of the surface on which they act. This intensity is
called stress. y P4 z
P3
y
Py P
1
o Px
Pz 2

P2
Pn

P1
x
z Fig. 2.1 Body with external forces
• In general, if the stress is not uniformly distributed over cross-section, taking
small area on the section and is A (=yz), the resultant force acting on it is
P.
• Limiting value of the ratio P/A. gives the stress acting on the cross-section at
the point o.
• Limiting direction of the resultant P is the direction of the stress.
• In general case, the direction of stress is inclined to the area A. on which it
acts.
• Resolving it into a normal or direct stress perpendicular to the area and other
two, shear stresses acting in the plane of the area A..

y P4 z
P3
y
Py P
1
o Px
Pz 2

P2
Pn

P1
x
z Fig. 2.1 Body with external forces
Notation of Stresses
• Letter  used for normal stress and the letter  for shearing stress.
• To indicate the direction of the plane on which the stress is acting, subscripts to
these letters are used.
• First subscript denotes the direction of the normal to the face and second
denotes the direction in which the stress component acts.
Px
 xx =  x = lt (2.1)
A→0 A

• Normal stress component xx acting on the x-face will act in the x-direction .
• Similarly, Py P
 xy = lt and  xz = lt z
(2.2)
A→0 A A→0 A
where xy and xz shear stresses acting on the faces in y- and z-direction,
respectively.
y P4 z
P3
y
Py P
1
o Px
Pz 2

P2
Pn

P1
x
z Fig. 2.1 Body with external forces
• Similarly, stresses can be defined for cross-sections parallel to the xy and xz
planes.
• For defining all these stresses, the stress at a point is defined generally by
taking an infinitesimal cuboid as shown in Fig.2.2. y
dz
• Nine different stresses act at a point in the element. yz
yx

• The six shear stresses are related as xy

zy
  x
zx xz
 xy =  xy ,  yz =  zy and  xz =  zx
dy
z

dx

Fig. 2.2 Infinitesimal element under stresses.

• Preceding three relations are found by equilibrium of moments of the

infinitesimal cube.
• There are thus six independent stresses.
• Stresses σx, σy, and σz are normal to the surfaces of the cuboid and the stresses
yz, zx, and xy are along the surfaces of the cuboid.
• A tensile normal stress is +ve, and a compressive normal stress is -ve. A shear
stress is positive, if its direction and the direction of the normal to the face on
which it is acting are both in positive or negative direction; otherwise, the shear
stress is negative.
State of Stress in Two-dimensions or Plane Stress:
• A two-dimensional stated-stress exists when the stresses and body forces are
independent of one of the coordinates. Such a state is described by stresses x,
y and xy, and the body forces Fx and Fy ( z is taken as the independent
coordinate axis). This combination of stress components which are functions of
only x- and y- coordinates called plane stress in the xy-plane(Fig.2.3).
• Example In a thin plate located in the plane of the plate there will be no stress
acting perpendicular to the surface of the plate.

y dx and dy  dz
dz
yx
yz=0

zy=0
Fy
Slim body xy
Plates
Fz=0
Almirah body  Fx  x
dy zx=0 xz=0

z=0

Paper sheets dx
Fig. 2.3 Two-dimensional plane stress model..
Boiler shell
STRAIN
• Similar to the need for knowledge of forces inside a body, knowing the
deformations because of the external forces is also important.
• Knowledge of deformations is specified in terms of strains, that is, strain is
measure of relative change in the size and shape of the body.
• Strain at a point is also defined generally on an infinitesimal cuboid in a right-
hand coordinate system.
• Under loads, the lengths of the sides of the infinitesimal cuboid change and
faces of the cube also get distorted.
• Change in length corresponds to a normal strain and the distortion corresponds
to the shearing strain.
• Figure 2.4 shows the two-dimensional state-of-strains on one of the faces,
ABCD, of the cuboid. dy u
y
C
v
y
dy
Deformed
y D

D u
y Undeformed
C
B
dy

v
u v x
dx
x
A v
B
A(x,y) u
dx
dx x

o x
Fig. 2.4 Cuboid with strains
• Strains and displacements are related to each other.
• Take the two perpendicular lines AB and AD.
• When the body is loaded, the two lines become A′B′ and A′D′.
• Normal strain in the x –direction, εx, is defined as the change of length of line AB
per unit length of AB as
AB − AB
 x = lt (2.3a)
dx →0 AB
• Similarly, εy in the y-direction is
ADB − AD
 y = lt (2.3b)
dy →0 AD

u
y
dy C
v
y
dy
Deformed
y D

D u
y Undeformed
C
B
dy

v
u v x
dx
x
A v
B
A(x,y) u
dx
dx x

o x
Fig. 2.4 Cuboid with strains
• Shear strain is defined as the tangent of the change in angle between the two
originally perpendicular axes.
• This component is specified with respect to two axes which are perpendicular on
the undeformed body and is designated by the symbol , with two subscripts to
indicate these two axes.
 
 xy = lt BAD − BAD = lt  − BAD (2.3c)
dx→0 dx →0  2 
dy→0 dy →0

• In the cuboid ABCD, before strain, the length of AB is dx, after strain, A is
displaced to A.
• Let us denote the xy components of the displacement of the particle at A by u
and v.
• As u and v vary from point to point in the body.
u
y
dy C
v
y
dy
Deformed
y D

D u
y Undeformed
C
B
dy

v
u v x
dx
x
A v
B
A(x,y) u
dx
dx x

o x
Fig. 2.4 Cuboid with strains
• Similarly, the displacements from B to B can be written as
u v
u= dx and v = dy
x y
u
• x-projection of A B is therefore, dx + dx
x
v
• and the y-projection is dx.
x
• Therefore, length of line A B can be expressed as
2 2
 u   v 
(AB)2 =  dx + dx  +  dx  (2.4a)
 x   x 

u
y
dy C
v
y
dy
Deformed
y D

D u
y Undeformed
C
B
dy

v
u v x
dx
x
A v
B
A(x,y) u
dx
dx x

o x
Fig. 2.4 Cuboid with strains
• By definition, normal strain in x-direction, is given by
AB  − AB
x =
AB
• Rearranging AB = AB( x + 1) = dx( x + 1)

• Squaring on both sides,

(AB)2 = ( x + 1)2 dx 2 (2.4b)

• Equating (2.4a) and (2.4b) we have

2 2
 u   v 
( x + 1)2 dx 2 =  dx + dx  +  dx 
 x   x 
u
y
dy C
v
y
dy
Deformed
y D

D u
y Undeformed
C
B
dy

v
u v x
dx
x
A v
B
A(x,y) u
dx
dx x

o x
Fig. 2.4 Cuboid with strains
• Simplification of above equation, we get
2 2
u  u   v 
( x2 + 2 x + 1) = 1 + 2 +  + 
x  x   x 
• Ignoring higher order terms due to small quantities, we get
u
x = (2.5a)
x

• Similarly, the longitudinal strain component in y-direction is obtained as

v
y = (2.5b)
y

u
y
dy C
v
y
dy
Deformed
y D

D u
y Undeformed
C
B
dy

v
u v x
dx
x
A v
B
A(x,y) u
dx
dx x

o x
Fig. 2.4 Cuboid with strains
• Shearing strain in the xy-plane, is as defined earlier the change in the angle
between sides AB and AD from 90°.

    u v 
 xy = − BAD = −  − − 
2 2  2 y x 

u v
  xy = + (2.5c)
y x
• Shearing strain is positive when the angle between the sides AD and AB
decreases; otherwise, the shearing strain is negative.

u
y
dy C
v
y
dy
Deformed
y D

D u
y Undeformed
C
B
dy

v
u v x
dx
x
A v
B
A(x,y) u
dx
dx x

o x
Fig. 2.4 Cuboid with strains
• Definitions of the remaining normal and shearing strains for three-dimensional
state-of-strains can be found by noting the change in size and shape of the other
sides of the infinitesimal cuboid in Fig. 2.5 as
w
z = (2.5d)
z
v w
 yz = + (2.5e)
z y
u w
 zx = + (2.5f)
z x

y
dz
yx
yz
xy
zy
  x
zx xz
dy
z

dx

Fig. 2.5 Infinitesimal element strains.

State of Strain in Two-dimensions or Plane Strain:
• A two-dimensional stated-strain exists when the strains are independent of one
of the coordinates. Such a state is described by strains x, y and xy, which are
assumed to be functions of only the x- and y- coordinates, and remaining strains
described by independent coordinate, z are zero(Fig.2.6).
• Example: Assumptions of plane strain is applicable for bodies that are long and
whose geometry and loading do not vary significantly in the longitudinal
direction.

dz  dx and dy
Beam

y
Shaft dz yx
Fat body yz=0 xy

Pipe   x
xz=0

zy=0
zx=0
dy
z=0

dx

Bridge Dam Fig. 2.6Two-dimensional plane strain model.

 Design Preference:
• Good design engineer prefers plane stress case of members used in machine,
such members always fail in ductile fracture type thereby ensures integrity in the
machine before becoming pieces.
• Analogy: Doctors recommends to keep good health and to look young by
maintaining slim body and such people may not suffer from serious health issues.
• Whereas, design engineer won’t prefer plane strain case of members used in
machine, such members always fail in brittle fracture type and thus, fails the
integrity in the machine with catastrophic manner.
• Analogy: If the person become fat by ignoring health consciousness then such
person may suffer from serious health issues like heart malfunctioning, BP, etc.
and may die in catastrophic manner.

Plane stress -Slim Plane strain-Fat

Mirror is keeping happy for plane-strain through images to look as plane-stress.

 PART A
ENGINEERING FRACTURE MECHANICS

27
 MOTIVATION

Failure of Load Bearing Structures

a) Excessive Deformation b) Breakage of Structure

i) Yielding – Irreversible deformation i) Fracture of the structure-

ii) Buckling- Sudden loss of shape causing material separation
due to excess deformation (with or without yielding)

We want to know the safety or load carrying capability of

“damaged/flawed/cracked” structure

28
• Historically Speaking

– Every civilization in the world had to confront with the problem of

breakage of solids/structure

– Scientific approach to understanding fracture and limited

strength of materials- Galieo and others in the 17th century and
later

– 19th century railroad fracture and ship breakages

– In the 20th century problem of fatigue and brittle fracture

– Aircraft failure post 1950 problem of ductile fracture

29
 • Examples of failure

Liberty ship failure during WW-II Aloha airline flight 243 failure
(http://www.sozogaku.com/fkd/en/cfen/CB1011020.html) (http://www.disastercity.info/ghost/sequence/)

Crankshaft failure Crack in wall

30
 • Example of Failure of Titanic ship

• On April 14, 1912, the R.M.S. Titanic

collided with a massive iceberg and
sank in less than three hours.
• At the time, more than 2200
passengers and crew were aboard
the Titanic for her maiden voyage to
the United States.
• Only 705 survived.
• According to the builders of the
Titanic, even in the worst possible
accident at sea, the ship should
have stayed afloat for two to three
days.
• The material failures and design
flaws that contributed to the rapid
sinking of the Titanic.
• In addition, the changes that have
been made in both the design of
ships and the safety regulations
governing ships at sea as a result of
the Titanic disaster (Source from
following URL).

31
For more details: http://writing.engr.psu.edu/uer/bassett.html
http://www.charlesapple.com/2012/04/a-terrific-titanic-anniversary-graphic-from-south-africa/
• Example of Failure of Titanic ship

32
For more details: http://writing.engr.psu.edu/uer/bassett.html
http://www.charlesapple.com/2012/04/a-terrific-titanic-anniversary-graphic-from-south-africa/
• Examples of failure

Aloha 737 fatigue failure

Landing wheel failure

HCF in turbine
Tooth interior fatigue failure Spring failure

For more cases visit 1), http://plane-truth.com/fatigue_details.htm

2) http://www.disastercity.info/ 33
• Fracture Analysis- Historical Development (Pre-1945)

Pre-1900s
➢ Galileo “ Why do solids break ?”
- some understanding of brittle nature of many structural materials
➢ Development of continuum mechanics and elasticity in the 18th and 19th
centuries
➢ Failure theories of solids for yielding
➢ Fatigue of materials (Wohler and others)

1900-1945
➢ C. E. Inglis, 1913,
I) stresses in a plate due to presence of crack and sharp corners
II) Classical elasticity cannot correctly give correct stresses near the edges
of a crack-like flaw in a structure (work on elliptical holes)

34
➢ A. A. Griffith, 1921
The phenomena of rupture and Flow in solids based on energy approach to
explain brittle fracture of glass
➢ H. M. Westergaard, 1939, Bearing pressure and cracks – application of a
complex stress function approach for two-dimensional problems

1945-1967 (Linear Elastic Fracture Mechanics)

➢ G. R. Irwin (1950s – 1960s)
- Laid the foundations for LEFM, concept of stress intensity factor, strain
energy release rate, etc.
➢ Orowan
- Energy criteria for brittle fracture
➢ M. L. Williams
- Series form for stresses in the region around a crack tip
➢ P. C. Paris
- Relating fatigue and crack growth

35
1967 (Non-linear Fracture Mechanics, Dynamic Fracture, Fracture
of Composites, etc.)

➢ J. R. Rice, et al
- J-Integral as a fracture criterion
➢ Analysis of ductile fracture
➢ Fracture in heterogeneous, anisotropic materials
➢ Interfacial fracture
➢ Numerical and computational methods
➢ Damage Tolerant Design of Load bearing structures

36
 INTRODUCTION

• For component subjected to uniaxial loading system, conventional

design failures observed during the tensile testing are:
➢ Yielding (yield) and
➢ Necking (ultimate) Lo

d do
 

Fig. 1 Component subjected to uniaxial loading

• Onset of yielding is characterized by the applied stress  reaching a

critical value y the yield strength of the materials.
• Similarly the necking phenomenon begins when the applied stress 
reaches the value u, ultimate strength of the material.
• For designing a component against permanent deformation, yield
strength y becomes the basis for selecting the allowable stress.
• Similarly, for designing against the aforesaid mechanical instability
ultimate strength, u, is the basis for selecting the allowable stress.
37
 Pure brittle
 f =  ys
d d
e Tough Shape memory alloy
f

Stress, 
c
 ys
c

 ys
b b Ductile
g
 ys
a
a

0
 fd  f  ts  fg
e
Strain, 

Fig. 2 Stress-strain curves for different materials

• For components subjected to complex loading system, the classical design
based on the prevention of failure initiation at the most critical point.
• This is done by using the one of the conventional failure criteria:
➢ Max normal stress theory
➢ Max shear stress theory, etc.
• Classical approach to design, which is based on the assumptions that the
material is homogeneous and defect-free
• According to such design requirements, materials with any defect are
useless
• These design practices don’t provide any basis for the prevention of failure,
initiation of the flaws. 38
• A component containing flaws/defects may not lead to a catastrophic
failure immediately on loading-
− There may be a stable crack growth before the instability sets in
− there is possibility that crack may not grow at all if the load is below a
certain critical value

• The classical approaches don’t give any basis for the calculation of this
critical load or the stable crack growth rate

• Thus, there is a tremendous scope for the material utilization if one can
predict the failure behaviour or provide the basis for calculation of
strength of components containing crack-like defects.

• Such a filling-up of the gap will enable one to design reliably even if the
flaws were to come up during manufacturing or fabrication.

39
• Theoretically, strength of material is of the order E/10 to E/100 (where E
is the modulus of elasticity).
• But, actual fracture strength of material which occurs without any plastic
deformation of the order E/100 to E/1000 and much below yield point of
the material
• Such a situation could not be explained using classical failure theories

Specimen with lower

number of defects
e
a
Stress, 

Specimen with larger

E
c number of defects

0 Strain,  40
Fig. 2a Stress-strain curves obtained for five specimens made from same material.
• This leads to the two important enquiry
✓ Why is there so much weakening of the material?
✓ Is this type of failure governed by a certain parameter reaching a critical value?
• These are the questions which have given rise to the discipline of Fracture
Mechanics
• This subject has given rise to the new material parameters, in terms of which
brittle fracture of bodies containing crack-like defects is defined.
• It has widened the scope of design and it has resulted in designing even with
defective materials.
Specimen with lower
number of defects
e
a
Stress, 

Specimen with larger

E
c number of defects

0 Strain,  41
Fig. 2a Stress-strain curves obtained for five specimens made from same material.
 FRACTURE
• Fracture is the separation or fragmentation of a solid body in to two or more
parts under the action of load
• Process of fracture
➢ Crack initiation
➢ Crack propagation
➢ Fracture/separation
• Classification
i) Ductile fracture ii) Brittle fracture iii) Mixed fracture
➢ Ductile fracture:
- It is characterized by appreciable plastic deformation prior to and during the
propagation of the crack
- An appreciable amount of gross deformation is usually present at the fracture
surfaces.
- It is high-energy process in which a large amount of energy dissipation is
associated with a plastic before crack instability occurs
➢ Brittle fracture
- It is characterized by a rapid rate of crack propagation (catastrophic
failure) and with very little micro deformation
- it is similar to cleavage in ionic crystals
- It is low-energy process in which a low energy dissipation occurs
42
 (i) (ii) (iii) (iv)
Cast iron Bronze Copper Aluminium

Fig. 3 Types of fractures observed in metals

i) Brittle fracture ii) Shearing fracture iii) Completely ductile fracture in polycrystals
and iv) Ductile fracture in polycrystals

43
 FRACTURE MECHANICS

• Strength of Material, Theory of Elasticity and Theory of Plasticity are used to

serve the purpose of illustrating the close form of analytical procedures in
order to develop constitutive equations for predicting failure of crack-free
solids.
• However, when solids contain flaws or cracks, the field equations are not
completely defined by these theories since they do not consider the stress
singularity phenomenon near a crack tip.
• They only provides the means to predict general yielding as a failure
criterion.
• Despite the usefulness of predicting yielding, it is necessary to use the
principles of fracture mechanics to predict fracture of solid components
containing cracks.

• Fracture Mechanics is an interdisciplinary subject which is concerned with

the effect of loading, configuration and size of the fracture of a load-bearing
body containing a flaw or crack

44
• Most static failure theories assume that the solid material to be analyzed is
perfectly homogeneous, isotropic and free of stress risers or defects, such
as voids, cracks, inclusions and mechanical discontinuities (indentations,
scratches or gouges).
• Actually, fracture mechanics considers structural components having small
flaws or cracks which are introduced during processing of materials and
manufacturing (e.g. solidification, quenching, welding, machining or
handling) . However, cracks that develop in service are difficult to predict and
account for preventing crack growth.

• Important aspect of this field is failure analysis where it able to answer the
following questions

– What is the residual strength as a function of crack size?

– What size of crack can be tolerated at the expected service load?
– How long does it take for a crack to grow from a certain initial size to the critical
size?
– What size of pre-existing flaw can be permitted at the moment the structure starts
its service life?
– How often should the structure be inspected for cracks?
45
 Comparison

Applied Stress Yield or Tensile Strength

Fig. 4a Traditional approach to structural design

Applied Stress

Flaw Size Fracture Toughness

Fig. 4b Fracture mechanics approach to structural design

46
 Family Tree of Fracture Mechanics

Linear time-independent
Linear Elastic Fracture Mechanics materials

Non-Linear time-
Elastic-Plastic Fracture Mechanics independent materials

time-
Dynamic Fracture Visco-elastic Visco-plastic dependent
Mechanics Fracture Mechanics Fracture Mechanics materials

Fig. 5 Simplified family tree of Fracture mechanics

47
 FRACTURE MECHANICS AT DIFFERENT SCALE LEVEL

• Phenomenon of the fracture of solids is complicated and depends on a

wide variety of factors, including the macroscopic effects, the microscopic
phenomena which occur at the locations where the fracture nucleates or
grows, and the composition of the material.
• Study of the fracture process depends on the scale level at which it is
considered.
• At one extreme there is a rupture of cohesive bonds in the solid, and the
associated phenomena take place within distances of the order of cm.
• For such studies the principles of quantum mechanics should be used.
• At the other extreme the material is considered as a homogeneous
continuum, and the phenomenon of fracture is studied within the
framework of continuum mechanics and classical thermodynamics.
• Fracture studies which take place at scale levels between these two
extremes concern movement of dislocations, formation of sub-grain
boundary precipitates and slip bands, grain inclusions and voids.

48
• Size range of significant events involved in the process of crack extension
is shown in following Figure.
• An understanding of the phenomenon of fracture depends to a large
extent on the successful integration of continuum mechanics with
materials science, metallurgy, physics and chemistry.
• Due to the great difficulties encountered in an interdisciplinary approach,
the phenomenon of fracture is usually studied within one of the three
scale levels: the atomic, the microscopic or the continuum.

49
Fig. 6 Scale Levels
 FRACTURE TERMINOLOGIES

• Some basic concepts referring to fracture will be defined as follows.

a) Fracture Process Zone: It is a small region surrounding the crack
where fracture develops through the successive stages of
inhomogeneous slip, void growth and coalescence, and bond braking
on the atomic scale
b) Crack Front: It is the line connecting all adjacent sites where separation
may occur subsequently
c) Fracture Surface: During continued separation, crack front will move
along a geometric surface termed the fracture surface. Area of this
surface i.e., the developed crack area, will increase as the crack grows.


Crack
Crack Front
Surface

Crack
Stage
Stage II
I
 50
Plate under load, 
d) Fracture Mode:
- Fracture mode designates the separation of geometrically
- In Irwin’s notation,
1) Mode I: It denotes a symmetric opening, the relative
displacements between corresponding pairs being normal to the
fracture surface, i.e., when two surfaces of a crack are being
separated by tensile forces which are applied perpendicularly to the
plane of the crack
2) Mode II: Occurs when in-plane shear forces are applied
3) Mode III: Occurs when out-of-plane shear forces are acting

y
x y y
x x
FI
z
Fracture z z
FII
surface
Crack tip
FIII FIII
FI FII
Mode I Mode II Mode III
(Opening or (In-plane shear or (Out-of-plane shear 51
Bending mode) Sliding mode) or Tearing mode)
 1.1 FRACTURE CRITERION

I. Theoretical Cohesive Strength (An Atomic View of Fracture)

– Metals are of great technology value, primarily of their high strength
combined with a certain measure of plasticity
– In most basic terms, strength is due to cohesive stress between atoms
– In general, high cohesive stresses are related to large elastic
constants, high melting points and small coefficients of thermal
expansion
– Following Fig.9 shows the variation of the cohesive stress between two
atoms as function of the separation between these atoms
+

Cohesive strength, 
max
x0
x0
0 x

Atoms
–
Fig.9 Cohesive strength as a function of the separation on between atoms
– A good approximation to the theoretical cohesive strength can be
obtained, if it is assumed that the cohesive strength curve can be
represented by a sine curve,
 x 
 =  max sin 
 /2
– For small atom displacements,
 x 
   max  
  / 2 
– and slope of the curve in this region becomes
+

Cohesive strength, 
 2 max max
= (1.1)
x  x0
0 x
/2

–
Fig.10 Cohesive strength as a function of the separation on between atoms

53
– From Hooke’s law for a brittle elastic solid,
x
E = Stress =  = 0
Strain x / x0 x
– Rearranging above equation and differentiating we get
 E
= (1.2)
x x0
– By combining equations (1.1) and (1.2),
E
Max = (1.3)
2x0

– If we make that x0 /2,

Max = E  E
 3
where x0 - distance between the atoms in the unstrained condition, and E -
Young’s modulus.
– Therefore, the potential for withstanding extremely high stresses before
fracture.

54
– When fracture occurs in a brittle solid all of the work expanded in
producing the fracture goes into the creation of two new surfaces.
– Each of these surfaces has a surface energy of S
– Work done per unit area of surface in creating the fracture is the area
+
under the stress-displacement curve.

Cohesive strength, 
max
– i.e.,
2
 2x 
WS =   Max sin  dx
x0

  
0 x
0 /2

 Max
=
 –
Fig.10 Cohesive strength as a
function of the separation on
between atoms
– But this energy is equal to the energy required to create the two new
fracture surfaces, thus
 Max
= 2 S

2  S
– or = (1.4)
Max 55
– Substituting equation (1.4) in to equation (1.3), we get the maximum
cohesive strength in the material as
– i.e.,
E 2  S
 Max = 
2 x0  Max

E S
  Max = (1.5)
x0

– Using expressions for the stress-displacement curve which are more

complicated than the sine-wave approximation results in estimation of
maximum stress from E/4 to E/15.

56
II. Stress Concentration Effect of Flaw
– Materials possess low fracture strengths relative to their theoretical
capacity because most materials deform plastically at much lower
stress levels and eventually fail by an accumulation of this irreversible
damage
– In addition, components and structures are not perfect
– They contain infinite material defects (e.g., pores, slag particles,
inclusions, and brittle particles), manufacturing flaws (e.g., scratches,
gouges, weld torch arc strikes, weld under cutting and machining
marks), and design defects (e.g., excessive stress concentrations
resulting from inadequate fillet radii and discontinuous changes in
section size).
– Stress concentration factor describes the effect of crack or flaw
geometry on the local crack tip stress level in a plate containing a hole
which represents a crack or flaw.

57
– In an infinite plate containing an elliptical hole with major axis 2a and minor axis
2b, expressions for the stress distribution are given by (Inglis)

 1 x( x 2 − a2 + b2 )−1 2 a2 x( x 2 − a2 + b2 )−3 2 
x = − 2 + 2 − 
 A A A 
 2 −1 2 a2 x( x 2 − a2 + b2 )−3 2 
y = 1 − B + Bx( x − a + b ) −
2 2

 A 
1 − 2b a
where, A = 1− b , B = 2 and  is the minimum radius of curvature
a (1 − b a )
at the end of major axis.

2a

2b x

W
B
 58
 - Radius of curvature,
– The resultant maximum axial stress at the edge of the ellipse is

(
max = 1+ 2a 
b
)
– Since the radius of curvature  at the end of the ellipse is given by
b2
=
a
– Substitution of above quantity in to max. stress expression we get
(
max = 1+ 2 a /   )
– In most cases a  , therefore
a
 Max = 2

 Kt  (1.6)

where Kt is the stress concentration factor

– By substituting  = x0 into equation (1.6), we obtain an estimate of the local stress
concentration at the tip of an atomically sharp crack

a
 Max = 2 (1.7)
x0
59
– If it is assumed that facture occurs when the local stress concentration at the tip of
an atomically sharp crack equal to the theoretical cohesive strength.

– Thus equating equations (1.5) and (1.7), resulting in the following expression for the
fracture stress at the failure.
– i.e., rough estimate of failure stress, f is obtained by

E S a
= 2 f
x0 x0

– Simplification yields, failure stress as

E s
f =
4a
– Thus, rough estimate of failure stress
Es
f = 
a
– Because of the continuum assumptions upon which the Inglis is based is not valid at
the atomic level.

60
III. Griffith Energy Balance (1921)
➢ Griffith's main achievement in providing a basis for the fracture
strengths of bodies containing cracks.

➢ He proposes approach based on First law of Thermodynamics criterion

for fracture by considering the total change in energy of a cracked body
as the crack length increases.

➢ First law of thermodynamics gives: The change in energy is proportional

to the amount of work performed.

➢ Since, only the change of energy is involved, any datum can be used as
a basis for measure of energy. Hence energy is neither created nor
consumed.

➢ Griffith showed that material fail not because of a maximum stress, but
rather because a certain energy criteria was met.

61
 • If we consider an initial crack length 2a, is increased to a length of 2a+2da in
a plate of elastic material under arbitrary loading P, then according to the
first law thermodynamics, a balance must be struck between the decrease in
the potential energy (related to the release of stored elastic energy and work
done by movement of the external loads) and the increase in surface energy
resulting from the presence of the crack.
• Thus, the Griffith model for elastic solids show that crack propagation is
caused by a transfer of energy from external work and/or strain energy to
surface energy.
• An existing crack would grow by some increment if the necessary additional
surface energy is supplied by the system.
• Surface energy results from the presence of a crack as shown in figure
below.
P
 - Radius of curvature,

2a
2b
2a 2a+2da

W
Representation of crack by 62
Inglis plate with elliptical hole. B P
collapsing minor axis to zero.
➢ Problem statement: A large plate containing a crack is subjected to a
remote and uniform tensile load in the direction of the y-axis and
perpendicular to the crack line along the x-axis. What is the external
stress that will cause crack instability value?

Solution:
❖ Surface energy is required to create two new surfaces,
ES = 2 2a B   S = 4a B  S (1.8)

❖ Using Inglis’s Stress analysis approach,
Stress
Total release of strain energy, free zone

1  
2
2a
ER = 2  (2a ) (2a ) B  
2  2E
2a

If  = /2, then

a 2B 2
ER = (1.9)
E 
Large plate containing a crack
is subjected to a remote and 63
uniform tensile load.
❖ For an incremental increase in the crack area, dA (=2 da B), under the
equilibrium conditions:
dE = d + dES = 0
dA dA dA
d dES
or − = (1.10)
dA dA
❖ Potential energy of a body with crack,
a 22B
 =  0 − ER =  0 −
E
❖ Differentiation of above equation w.r.t crack area A yields
d d 0 d  a 22B  1 d 0 1 d  a 22B 
= −   = −  
dA dA dA  E  2B da 2B da  E 
d  2a
 =− (1.11a)
dA E
❖ Differentiating equation (1.8) w.r.t. crack area A, we get
dES
=
d
(4a B  S ) = 1 d (4a B  S ) = 2 S (1.11b)
dA dA 2B da 64
❖ Substitution of equations (1.11a) and (1.11b) into equation (1.10), yields
 2a
= 2 S
E
❖ above equation can be written for fracture stress for the plane stress case as

2E  S
f = (1.12a)
a
❖ And for the plane strain case we have

2E  S
f = (1.12b)
 a (1 − 2 )
where f is the failure strength to cause crack instability.

➢ Griffith equation (1.12) shows a strong dependence of fracture strength on

crack length
➢ Griffith equation satisfactory predicts the fracture strength of a completely
brittle material such as glass.
➢ However, the Griffith equation severely underestimates the fracture strength of
65
metals.
III. Modified Griffith Theory
– It is well-known that plastic deformation occurs in engineering metal, alloys and
some polymers prior to fracture.
– This is substantiated by x-ray diffraction studies of fracture surfaces and metals
graphic studies of fracture.
– Therefore Griffith equation for the fracture strength doesn’t apply for metals.
– Orowan modified the Griffith equation to account for materials that are capable
of plastic flow by the inclusion of a term plastic deformation energy or plastic
strain work denoted by P, expressing the plastic work required to extend per
unit crack area of surface created.
– Revised expression for fracture strength for plane stress case, is given by

2 E ( S +  P )
f = (1.12c)
a 
Plastic zone
Thickness, B

66
 ➢ In an ideally brittle solid, a crack can formed merely by breaking atomic
bonds (Ref. Fig. i), thus S reflects the total energy of broken bonds in a unit
area
➢ When a crack propagates through a metal, dislocation motion occurs in the
vicinity of the crack tip, resulting in additional energy dissipation (Fig. b)
➢ Although, Orowan originally derived equation (1.12c) for metals, it is
possible to generalize the Griffith model to account for any type of energy
dissipation
2 E Wf
f = (1.12d)
a
where Wf is the fracture energy which could include plastic, visco-elastic, or
visco-plastic effects depending on the material.
➢ Fracture energy can also be influenced by crack meandering and
branching, which increases the surface area.
Crack propagation
Broken bonds

Crack v
v
Wf = S
Plastic deformation
Wf = S  (True area/Projected area)
Wf = S + P

Fig. (i) Ideally brittle material Fig. (ii) Quasi-brittle elastic-plastic Fig. (iii) Brittle material with 67
crack
material meandering and branching
V. Energy Release Rate
➢ In 1956, Irwin proposed an energy approach for fracture
➢ He defined an energy release rate: energy released per unit increase
in area during crack growth

❖ Mathematical formulation;
with an advancing crack, the following happen in a general case:
− Strain energy in the component decreases or increases
− Stiffness of the component decreases
− Points of the component at which external loads are applied may or
may not move, work is being done on the component by these
forces if the points move
− Energy is being consumed to create new crack surfaces.

68
➢ To cause an incremental increase in the crack area A, an external
incremental work is done, Wext and the strain energy within the body
increases by U.
Thus, the available energy provides the energy balance as

G A = − (U − Wext )

where G is energy release rate per unit area of growth given by

d 
G=− (U − Wext )
dA Thickness, B

➢ Substitution of equation (1.11) in to above equation, yields

a da

d
G=− (1.13)
dA

➢ In many cases, fracture mechanics is applied to plates of uniform

thickness and then crack area A can be expressed as Ba , where B is
the thickness and a is the incremental in crack length, we have

1 d  2a
G=− = (1.14) 69
B da E
➢ Crack extension occurs when G reaches a critical value, that is,
 f2ac
G  GC = (1.15)
E
where GC - critical energy release rate and is measure of the fracture
toughness of the material and is constant, aC- critical crack length, and
C- failure stress

▪ Two approaches to find the value of G

i. Change in compliance approach

a) Constant load method
b) Constant displacement method
ii. Change in strain energy approach
a) Constant load method
b) Constant displacement method

70
❖ Displacement u of the point at which the load is applied can be expressed as
P
u= =CP (1.16)
K
where P is the applied load, K is the stiffness of the body and C is compliance

i. Change in Compliance Approach

u+du
• Constant load method: u

Strain energy, U = 1 Pu and a a a +da

2
P
P
External Work, Wext = P u
1
dU = P du
2
Potential energy of the component,
P
1
 = U - Wext = − P u (1.17) dWext= P du
2

Load
Substituting eqns. (1.16) and (1.17) into eq. (1.13),
d 1 d 1 d  Pu 
G=− = − = − − 
dA B da B da  2  u du
Displacement
P du P 2 dC
=  = (1.18)
2B da 2 B da
b) Constant displacement method:
Strain energy, U = 1 P u and
2
External Work, Wext = 0 u u

Potential energy of the component, a a+da

1
 = U - Wext = P u (1.19) 1
2 dU = u dP
2
Substituting eq. (1.19) into eq. (1.13), we have P
dP
d 1 d 1 d  Pu 
G=− =− =−  
dA B da B da  2 

Load
u dP
=− 
2B da u
Displacement

From eq. (1.16), substituting value of P into above equation, we get

u d u u2 d  1 
G=−    =−   
2B da  C  2B da  C 
simplifying yields energy release rate in terms of change in compliance,
u 2 dC
G= 2
(1.20) 72
2 BC da
➢ Substituting value of displacement u from equation (1.16) in to equation
(1.20) we obtain energy release rate in terms of rate of change of
compliance as ,
(CP )2 dC P 2 dC
G= =
2 BC 2 da 2 B da

ii. Change in Strain Energy Approach

a) Constant load method:
In this case we have

Strain energy, U = 1 P u = 1 Wext

2 2
or External Work, Wext = 2U
Thus the potential energy of the component with crack is given by
Π = U - Wext
Energy release rate,
dΠ d dU
G=− =− (U - Wext ) = (1.21)
dA dA dA
− Thus the strain energy increases with as the crack advances
73
b) Constant displacement method:

In this case we have

Strain energy, U = 1 Pu
2

or External Work, Wext = 0

Potential energy of the component with crack

 = U - Wext

Energy release rate,

d d dU
G=− =− (U - Wext ) = − (1.22)
dA dA dA

− Strain energy decreases with as the crack advances

74
VI. Stress Intensity Approach
• Certain cracked configurations, expressions for stress components in
the vicinity of crack tip in the body under external load were reported by
Irwin, Sneddon, Williams and Westergaard
• Stress components in the vicinity of crack tip is given by
ij = f (, a, r , , geometry ) 

• For the case of flat plate with a crack of length, y

22
2a and for field stress, , the stress field at a
12
general point H near the crack tip for Mode I H 11
case is given by r

x
 (a)  (a) 2a
11 = f11() , 22 = f22 ()
2r 2r

 (a)
and 12 = f12 () 
2r
75
• Displacement field for plane strain near the crack tip for the Mode I problem
is given by

u11 =

( )
 (a) r 1 / 2 ˆ
2
f11() , u22 =

( )
 (a) r 1 / 2 ˆ
2
f22 () and u12 = 0

• where  is the shear modulus

❖ Irwin pointed out that the local stresses near a crack in an elastic body
with crack depend on the product of the normal stress () and the
square root of the half-flaw length (a)
❖ He called this relationship as Stress Intensity Factor (SIF) denoted by K
❖ For the crack configuration shown in figure, the SIF is given by

K = a (1.23)

➢ It defines the amplitudes of the crack tip singularity

➢ It characterizes the crack tip conditions in a linear elastic material
➢ If K is known, it is possible to solve for all components of stress, strain
and displacement
➢ If two different flaw configurations have same value of K, then stress
field around each flaw are identical 76
▪ For general case the SIF is given by
K =  a (1.24)
where  - crack parameter that depends on the specimen and crack
geometry.

❖ Critical SIF (KC)

➢ If SIF of a crack approaches or exceeds an upper limit of SIF, the crack

may grow. The Upper limit is known as critical SIF or Fracture
toughness
➢ Critical values of SIF can be used to define the condition for failure, i.e
K  KC (1.25)

➢ Critical SIF is considered as material property which describes the

inherent resistance of the material to failure in presence of a crack

▪ In comparison with SIF, the energy release rate, G has a more direct
physical significances to the fracture process.
▪ However, SIF is preferred in working with fracture mechanics because
it is more agreeable to analytical determinations.
77
▪ Comparing equations (1.14) and (1.23) results in a relationship
between energy release rate, G and SIF, K.
▪ i.e., we have from equation (1.14) for energy release rate as
 2a
G=
E
▪ or GE =  2a (i)

▪ Squaring equation (1.23) on both sides, we get,

K2 =  2 a (ii)

▪ Equating equations (i) and (ii), we obtain relationship between SIF and
energy release rate for mode I loading as

K2
G= (1.26)
E
▪ This same relationship holds for critical values, as

KC2
GC = (1.27)
E
78
• In general for Mode I, II and III, the stress intensity factor is written as
KI, KII, and KIII, respectively.
• Stress fields ahead of a crack tip in linear elastic material can be
written as

fij(II )() and (ijIII ) = III fij(III )()

KII
fij(I )(), (ijII ) =
KI K
(ijI ) =
2r 2r 2r

• In a mixed-mode problem (more than one loading mode is present),

then the stress components are additive:
(ijtotal ) = (ijI ) + (ijII ) + (ijIII )

➢ For given type of loading and crack geometry, the relations

For Mode I:
KIC =  f  aC = or = f  aC f a
h
()
For Mode II:
KIIC =  f  aC or = f  aC f a
h
()
()
For Mode III:
KIIIC =  f  aC or = f  aC f a
h

where  - crack parameter, f - failure normal stress, f - failure shear 79

stress, f(a/h) - geometrical parameter and h - depth of component
 R-Curve

Crack Resistance:
• Energy required for a crack to grow per unit area extension is called crack
resistance and is denoted by the symbol, R.
• It characterizes the material behaviour.
• For most of the engineering materials crack resistance increases with crack
length as shown in Fig (i).
• A minimum value, Ri is needed to make the crack to grow.
• Crack resistance depends on the plastic zone size: for a large plastic zone
size, high energy is required to grow the crack because more material is
subjected to plastic deformation
• As the crack advances, plastic zone size becomes larger which in turn
requires higher energy for growth of the crack.

R R-curve

Ri

80
a0 Crack size, a
• Shape of the R-curve is not found to be depend on the initial length of a
crack for most of the engineering materials
• However, R-curve depends considerably on temperature and thickness
of a plate
• Dependence on temperature is because many other properties such as
yield stress, etc., depend on temperature.

Stable and Unstable Crack Growth:

• Crack extension occurs when energy release rate is twice the surface
energy (i.e., G = 2Wf), but crack growth may be stable or unstable,
depending on how G and Wf vary with crack size.
• To illustrate the stable and unstable crack growth, it is convenient to
replace 2Wf with R, material resistance to crack growth
• For a crack to grow the energy release rate G of a crack must be
greater than the crack resistance R. i.e.,
GR (1.28)

81
• Consider a large plate with a initial centre crack length 2ao loaded in
Mode I by stress  as shown in Fig (ii).
• Referring to equation (1.14), at fixed remote stress,  the energy release
rate G varies linearly with crack.
 2a
G= (1.14)
E
• Figs. (iii) and (iv) shows schematic of G/R-curves for two types of material
behaviour.
• A plot of G vs. a is the driving force curve or G-curve, corresponding a
plot of R vs. a is the resistance curve or R-curve.
4

 2

G, R
Instability 3
G, R

Unstable
GC=Ri R GC
1
2ao 2
Stable
A B
a0
Crack size, a
 Ri
1

82
a0 a3 a4
Crack size, a
• Referring to Fig. (iii), a flat R-curve indicates the material resistance is
constant with crack growth
• If applied stress is equal to 1 the crack growth is stable.
 12a
G=
E
• Facture occurs when the applied stress reaches 2 the crack propagation
is unstable because the driving force increases with crack growth, but the
material resistance remains constant.

2
G, R

Unstable
GC=Ri R
1
Stable
a0
Crack size, a

83
• Referring to Fig. (iv), illustrates a material with a rising R-curve.
• If stress 1 is small and the G-curve intersects the R-curve below Ri, thus
crack will not propagate because the energy release rate is not adequate.
• Curve of high stress, 2 just passes through Ri and still the crack does not
advance because with a increase in crack length, the energy release is
smaller than the requirement
• If the stress is further increased to 3 , G exceeds R for crack having length
between a0 and a3, and the crack is likely to grow
• However, with the increase in crack length the difference between G and R
diminishes to zero at point B
• There will not be any further advancement of the crack.

G, R
3

2
A B

Ri
G 1

84
a0 a3
Crack size, a
• Therefore, the crack may advance from length a0 to a3 only , if the applied
stress increases gradually, point B shifts to right, that is the crack grows
slowly(stable crack growth).
• For stress 4 G-curve just touches the R-curve which means energy
release rate is higher for all crack lengths except at the crack length a4
where they are equal
• As soon as the crack length increases the difference between G and R
grows which provides excess energy to the crack
• As a result, the crack gains velocity ending up in a catastrophic failure
• Condition for stable crack growth can be expressed as follows:
4
dG dR
G=R and 

G, R
da da Instability 3
• Unstable crack growth occurs when
GC

dG dR 2
 A B
da da
Ri
G 1

85
a0 a3 a4
Crack size, a
 1.2 CRACK TIP STRESS ANALSYS

• A variety of techniques are available for analyzing stresses in cracked

bodies.
• Among these techniques, Westergaard approach provides a means for
connecting the local fields to global boundary conditions in certain
configuration
• While in William approach considers the local crack tip fields under
generalized in-plane loading

• In this study, Westergaard approach is discussed to derive crack tip

stresses.

86
 Westergaard Approach

• This approach provides a means for connecting the local fields to

global boundary conditions in certain configurations

• For limited class of problems to solve the biharmonic equation is to

express Airy stress function , in terms of another complex stress
function, ZI(z) for Mode I problems, where z = x + iy.

• For specific problem, the form of ZI(z) is chosen to satisfy all the
boundary conditions.

• Similarly, for Mode II and Mode III, the form of ZII(z) and ZIII(z),
respectively, are chosen.

87
 Mode I (Opening Mode)

➢ Westergaard stress function is related to

 = ZI (z) = ReZI + y Im ZI (1.29)

where Re and Im denote real and imaginary parts of the function and bars
over the Z denotes integration with respect to z, i.e.,

dZI 
ZI = 
dz 
 (1.30)
ZI =
dZI 
dz 

➢ Making use of following equations

 Re Z  Im Z 
= Re Z  , = Im Z  ,
x x 

 (1.31)
 Re Z 
 Im Z 
= − Im Z  , = Re Z 
y y
88
• Equation (1.29) should satisfy following biharmonic equation
 4  4  4
+2 2 2 + 4 =0 (1.32)
x 4 x y y
• To check, differentiating equation (1.29) w.r.t. x four times we get
  2
  3

= ReZI + y Im ZI = Re Z + y Im Z  = ReZI + y Im ZI
x x 2 I I
x 3

 4
= ReZI+ y Im ZI
x 4

• Similarly, differentiating equation (1.29) w.r.t. y four times we get


= − Im ZI + y Re ZI + Im ZI = y Re ZI
y
 2
= − y Im Z  + Re ZI
y 2

 3
= − yReZI − Im ZI − Im ZI = − yReZI − 2Im ZI
y 3

 4
= y Im ZI− ReZI− 2 Re ZI = y Im ZI− 3ReZI
y 4 89
• Similarly we obtain other terms by differentiating equation (1.29) w.r.t x
and y with single and two times as

 2
= y Re ZI
xy

 4
= −y Im ZI + Re ZI
x y
2 2

• Substituting above values in to equation (1.32), we obtain the satisfaction

of biharmonic condition. Thus, equation (1.29) may be used to define the
stress field as
2
x = 2 = ReZI − y Im ZI , 
y 

2
y = 2 = ReZI + y Im ZI and  (1.33)
x 

2 
 xy = − 
= − y ReZI 
xy

90
➢ To determine displacement field (u, v) in the vicinity of the crack tip, let
us convert stress field to strain field with the help of stress-strain
relations.
➢ Plane stress case:
➢ Strain-stress relationships in terms of Westergaard functions, are given
by
u 1
= ( x −  y ) = (ReZI − y Im ZI ) −  (ReZI + y Im ZI )
1
x =
x E E
1− 1+
= ReZI − y Im ZI (1.34a)
E E
v 1
= ( y −  x ) = (ReZI + y Im ZI ) −  (ReZI − y Im ZI )
1
y =
y E E
1− 1+
= ReZI + y Im ZI (1.34b)
E E

u v 2(1 +  ) 2(1 +  )
 xy = + =  xy = − y ReZI (1.34c)
y x E E
91
• Making use of the relation
E
=
2(1 +  )
• We re-write equations (1.34a) and (1.34b) as
u 1 1 −  
=  ReZI − y Im ZI  (1.35a)
x 2 1 +  

v 1 1 −  
=  Re Z + y Im Z 
I  (1.35b)
y 2 1 + 
I

• Integrating equation (1.35a) w.r.t x, we have,
1 1 −  
2 1 +    I 
u=  Re ZI dx − y Im Z  dx

1 1 −  Im ZI 
= 
2 1 +   ReZI dx − y  x dx 
1 1 −  
=  Re Z − y Im ZI  + f (y ) (1.36a)
2 1 + 
I

92
• Similarly, integrating equation (1.35b) w.r.t y, we have,

1 1 −  
2 1 +   
v=  Re ZI dy + y Im ZI
 dy 

1 1 −  ReZI  
= 
2 1 +   
ReZI dy + y  −
 y
 dy 
 

1  2 
=  Im Z − y Re ZI  + g( x ) (1.36b)
2 1 + 
I


• Where f(y) is the constant of integration which can be function of y only,

similarly, and g(x) is the function of x only.
• Equation (1.36) contains the unknown functions f(y) and g(x) and these
values are obtained by substituting equation (1.36) in to equation
(1.34c),

93
• i.e., differentiating equation (1.36a) w.r.t. y, we get
u 1 1 −     f ( y )
=  (Re Z ) − ( y Im Z I +
)
y 2 1 +  y y y
I


=
1 1 −  

2 1 +  y
( Re Z dx ) − y (y Im Z ) + f(yy )
I I


1 1 −    f ( y )
=  (0 ) − y ( Im Z ) − Im Z I+
2 1 +  y y
I


=
1
− y Re ZI − Im ZI  + f ( y )
2 y

• Ignoring imaginary part, we obtain

u
=
1
− y Re ZI + f ( y ) (1.37a)
y 2 y

94
• i.e., differentiating equation (1.36b) w.r.t. x, we get

v 1  2    g ( x )
=  (Im Z ) − ( y Re ZI +
)
x 2 1 +  x x x
I


=
1  2 
2 1 +  x 
 (Im ZI dy − y )
x
(Re Z ) g ( x )
I +
 x

=
1  2
 (0) − y  (Re ZI ) + g ( x )
2 1 +  x  x

=
1
− y Re ZI + g( x ) (1.37b)
2 x
• From equation (1.34c), we have

u v 2(1 +  )
 xy = + =− y ReZI (1.34c)
y x E
1
=− y ReZI

• Substituting equations (1.37a) and (1.37b) in to above equation we get
95
  1 f ( y )   1 g ( x ) 
 2 − y Re Z   +  +  − y Re Z   +  = −
1
y ReZI
y   2 x  
I I

f ( y ) g ( x )
 + =0
y x
• Above equation is re-arranged as
f ( y ) g ( x )
=− =A
y x

• where A is a constant on integration, we have

f ( y ) = Ay + B
g( x ) = −Ax + C
• Again B and C are constants of integration. If we substitute these
equations in to displacement equations (1.36a) and (1.36b), we observe
that all the points of the component are displaced by same distance, given
by u=B and v=C.
• Therefore, B and C corresponding to rigid body translation and they may
be set to zero without losing any generality.
96
• Therefore, B = C = 0 and thus,
f ( y ) = Ay
g( x ) = − Ax
• Furthermore, rigid body rotation  is governed by the equation
1  u v 
=  − 
2  y x 

1  1   1 
= 
 − y Re Z  + A 
 − 
 − y Re Z  − A 
2  2   2
I I

 =A
• Therefore, A corresponds to rigid body rotation of the component and
may be set equal to zero without losing any generality.
• Thus for the plane stress cases, displacement are
1 1 −  
u=  Re Z − y Im ZI (1.38a)
2 1 + 
I


1  2 
v=  Im ZI − y Re ZI  (1.38b)
2 1 +   97
• Plane strain case:
• Strain-stress relationships in terms of Westergaard functions, are given by
u 1 + 
x = = (1 −  ) x −  y  1 
=  (1 −  )ReZI − (1 +  )y Im ZI
x E E
v 1 + 
y = = (1− ) y −  x  1 
=  (1 −  )ReZI + (1 +  )y Im ZI
y E E

u v 2(1 +  ) 2(1 +  )
 xy = + =  xy = − y ReZI
y x E E
 E  
• where E = − and  =
1− 2 1−
• Integrating above equations, we obtain the displacements for the plane
strain as
u=
1
2

(1 − 2 )ReZI − y Im ZI 
(1.39a)

v=
1
2

2(1 − )Im ZI − y Re ZI 
(1.39b)

• Thus all the stress and displacements are expressed in terms of the
Westergaard function which is still not a known function. 98
• Consider an infinite plate with through the thickness crack of length 2a
loaded under a biaxial field stress .
• Boundary conditions:
i. At the crack tip ( i.e., x= -a and x=+a): y= 
ii. On the cracked surface (i.e., y=0, -a  x  a): y= 0 and xy= 0
iii. For away from the crack ( large IzI ):
x= , y=  and xy= 0 

iy
y
xy
• Westergaard function which satisfies all the x
boundary conditions is z 

z  x
ZI ( z ) =
(z − a )1/ 2 (z + a )1/ 2
2a

or
z
ZI ( z ) = (1.40)
(z 2
−a )
2 1/ 2 

99
 
• Transformation of origin from center of crack to its tip
y
z = a + z0 (1.41) y
xy
• Then equation (1.40) is modified to
x
 (a + z0 ) z=a+z0
ZI ( z0 ) =
r
a  x 

(a + z0 − a) (a + z0 + a)
1/ 2 1/ 2
2a

 z0 
 a 1 + 
=  a  
1/ 2
 z 
(2z0 )1/ 2 1 + 0 
 2a 
• Since z0 << a in the vicinity of the crack tip, the above equation is
simplified to the approximate relation as
 a KI
ZI ( z0 )  = (1.42)
2z0 2z0

• Expressing z0 in polar co-ordinates

zo = r cos  + i sin  100
• Thus, 
KI
ZI =
2r (cos  + i sin ) y
y
or
KI    xy
ZI = cos − i sin
2 
x
2r  2 z=a+z0
r
a  x 

• Separating real and imaginary terms, we get 2a
KI 
Re ZI = cos (1.43a)
2r 2
KI  
Im ZI = − sin (1.43b)
2r 2
• Similarly, the first derivative of equation (1.42) yields
 a2  a2
ZI( z0 ) = − −
z03 / 2 ( z0 + 2a )3 / 2 z03 / 2 ( 2a )3 / 2
• Expressing in polar coordinates, we get

 a  3 3 
ZI(r ) = − cos − i sin
2(2 )1/ 2 r 3 / 2  2 2 
101
 KI  3 3 
• ZI = − cos − i sin
2 
or
2(2 )1/ 2 r 3 / 2  2

• Separating real and imaginary terms we get,

KI 3
Re ZI = − cos (1.44a)
2(2 )1/ 2 r 3 / 2 2
KI 3
Im ZI = sin (1.44b)
2(2 )1/ 2 r 3 / 2 2
• Substituting equations (1.43a) and (1.44b) in to equation (1.33) for x and
realizing that y = r sin , we get
KI  KI 3
x = cos − ( r sin  ) sin
(2r )1/ 2 2 2(2 )1/ 2 r 3 / 2 2

• Using trigonometric relation, we obtain

KI   3 
x = cos 1 − sin sin  (1.45a)
(2r )1/ 2 2 2 2
102
• Similarly, for stresses y and xy we obtain

KI   3 
y = cos 1 + sin sin  (1.45b)
(2r )1/ 2 2 2 2

KI    3 
 xy =  sin cos cos  (1.45c)
(2r )1/ 2  2 2 2
• For obtaining the displacement field in the vicinity of the crack tip, we
obtain integral of Westargaard function from equation (1.42) as
1/2
KI 2
ZI ( z0 ) =  ZI ( z0 ) dz =  2z0
dz =  
 
K I z0

• Expressing in terms of polar coordinate, we get

 
1/2
 2r  
ZI (r ) =   KI cos + i sin 
   2 2
• Separating real and imaginary terms, we get

 
1/2 1/2
 2r   2r 
Re ZI =   KI cos and Im ZI =   KI sin (1.46)
  2   2
103
• Substituting equations (1.43) and(1.46) in to equation (1.38) we obtain
displacement field for plane stress as
KI r  1 −  
u= cos  + sin 2  (1.47a)
 2 2 1 +  2

KI r  2 
v= sin  − cos2  (1.47b)
 2 2 1 +  2
• and for plane strain case, we obtain displacement field by substituting
equations (1.43) and(1.46) in to equation (1.39), i.e.,

KI r  
u= cos 1 − 2 + sin2  (1.48a)
 2 2 2

KI r  
v= sin 2 − 2 − cos 2  (1.48b)
 2 2 2

• Note that above solution is valid for only in the vicinity of the crack tip and
u does not depend on , because Mode I problem is symmetric, however,
v changes since it is dependent on .
104
 Mode II (Sliding Mode):

• Westergaard function for Mode II loading is given by



ZII ( z ) = −y Re ZII (1.49) y

y
xy
x
Stress field: z
 x = 2Im ZII + y ReZII 
 x
 y = − y ReZII  2a
 (1.50)

 xy = ReZII − y Im ZII 


and displacement field for plane strain case:

u=
1
2
(
2(1 − )Im ZII + y ReZII ) 

 (1.51)
v=
1
(− (1+ 2 )Re ZII - y Im ZII ) 
 105
2
• Suitable Westergaard function is given by 

z y
ZII ( z ) = (1.52)
(
z −a
2
)
2 1/ 2 y
xy
x
Transformation of origin from center of crack to its tip r

z = a + z0 (1.53) 2a
x

Then equation (1.52) is modified to

 a K II
ZII ( zo )  = 
2zo 2zo
Expressing in polar co-ordinates and substitution of ZII and its derivatives into
eq. (1.50) yields stress field as
K   3  
 x = − II sin 2 + cos cos  
2r 2 2 2 
KII   3 

y = cos sin cos  (1.54)
2r 2 2 2 

K II   3  
 xy = cos 1 − sin sin   106
2r 2 2 2 
• Substitution of value of ZII and it integral in to eq. (1.51) yields
displacement field for plane strain case as

K II r   
u= sin 2 − 2 + cos 2  
 2 2 2 
 (1.55)
K II r    
v= ( − sin )1 − 2 − sin2  
 2 2  2 


• With the similar procedure, displacement field for plane stress case can
be obtained as
 2 
K II r 2 
u= sin  + cos  
 2 2 1 −  2 
 (1.56)
K r    1 −   
v = II  − cos   − sin2  
 2  2  1 +  2 


107
 Mode III (Tearing Mode) 

• For Mode III loading: y

x
− Displacements: u=0, v=0 and w(x,y)  0 r 

− Strains: x = 0, y = 0, z = 0,  2a z

1 w 1 w  (1.57)
 xy = 0,  xz = ,  yz =  
2 x 2 y 

− Stresses:  x = 0,  y = 0,  z = 0, 

w w  (1.58)
 xy = 0,  xz =  ,  yz =  
x y 
• Note that problems of Mode III is not a case of plane stress or plane
strain case.
• Many stresses, strains and displacements are zero thereby the
problem is simple
• Thus, not required to use biharmonic equation
• Instead, the problem will be solved in displacement component, w and
no need to check for compatibility conditions.
108
• Out of three equilibrium equations for general three-dimensional body,
only one provides non-trival equation as
 xz  yz
+ =0
x y
• Converting in to displacement components using eq.(1.58), yields
 2w  2w
+ =0 (1.59)
x 2 y 2
• General Westergaard function is of the form
1
w= Im ZIII (1.60)

• Shear stresses are given by
 xz = Im ZIII 

 (1.61)
 yz =ReZIII 

• Suitable Westergaard function is given by
z
w = ZIII ( z ) = (1.62)
(z −a
2
)
2 1/ 2
109
which satisfies all boundary conditions
• By transformation of the origin and after neglecting the small terms, we
obtain
 a K III
ZIII ( zo ) = = (1.63)
2zo 2 z o

• Expressing in polar coordinates, we get

K III   
ZIII = cos − i sin (1.64)
2r  2 2 

• Substituting above value into eq. (1.61) yields the stress field as
K  
 xz = − III sin 
2r 2 
 (1.65)
KIII  
 yz = cos
2r 2 

• Integrating eq. (1.63) and expressing in polar coordinates then substitution
of imaginary term in to eq. (1.60), we get displacement field

K III 2r 
w= sin (1.66)
  2
110
with u = 0 and v = 0.
 SIGNIFICANCE OF KI
• The stress intensity factor defines the amplitude of the crack tip singularity that is,
stresses near the crack tip increase in proportion to KI
• Moreover, KI is known, it is possible to solve for all components of stresses, strains
and displacements as a function of r and 
• Referring to figure is a schematic plot of the stress normal to the crack plane, y
versus distance from the crack tip, r in the plane of crack (i.e.,  =0), equation
(1.45b) is only valid near the crack tip where the 1/r1/2 singularity dominates the
stress field.
KI   3 
y = cos 1 + sin sin  (1.45b)
(2r )1/ 2 2 2 2


KI
y =
2r

 =

2a x

Singularity dominated zone

111

 KI VALUES FOR SPECIFIC CRACK GEOMETRIES

• In general, the successive application of LEFMs to structural analysis, fatigue and

stress corrosion cracking requires a known stress intensity factor (SIF) equation for
a particular specimen configuration. The SIF can be related to the global stress by
KI=  (a)1/2 for various crack geometries, where  is the geometric correction
factor. The expressions for  for some selected and practical crack configurations
for mode I loading are shown in Table:

112
 Fracture Toughness KC in Plane Stress and Plane Strain cases

• It is found that the critical SIF or fracture toughness (KC) is strongly dependent on
the material thickness up to a limiting value.
• For a thin plate, plane stress condition (z=0) governs the fracture process
because the plate is too thin to sustain through-the-thickness stress thus the
fracture toughness is depends on the thickness up to a limiting value
• For a thick plate, plane strain condition ((z0) prevails in which fracture
toughness becomes a material’s property, i.e., fracture toughness is independent
of the thickness beyond limiting value
• It is this property, KC, that the designer must use to assure structural integrity.


y y

B
y x x

x z=0 z
z Plane stress case Plane strain case

Illustration of the state of stress at the crack tip for plane

stress and strain conditions


113
 • Figure (i) shows the variation in fracture toughness for three regions called plane strain,
mixed mode and plane stress as a function of the material’s thickness with the amount
of the flat and slanted surfaces corresponding to each region
• Figure (ii) shows the variation in the amount of flat fracture as a function of thickness.

Specimen under no load Specimen under load

Maximum flat surface

Thin B1B2  B3
B1

Amount of flat fracture (%)

B2 Medium
B2 B3
Thick B1
B3
Almost
Mixed Almost flat
KIC slanted
Plane stress (Mixed mode fracture) Plane strain
(Ductile fracture) (Brittle fracture)
Plane stress Plane strain
(Mixed mode fracture)
(Ductile fracture) (Brittle fracture)

Thickness, B (mm)
114
Fig. (i) Effect of specimen thickness over fracture toughness Fig. (ii) Effect of specimen thickness over flat fracture surface
 – For thin sections where the state of stress at the crack tip is not triaxial, the
constraint to plastic deformation lessens and failure is associated with plane stress
– For the plane strain failure, the portion of the flat surfaces is much larger than the
slanted section
– For sections with adequate thickness, in which plane strain and plane stress are
combined, the state of stress is termed mixed mode

Specimen under no load Specimen under load

Maximum flat surface

Thin B1B2  B3
B1

Amount of flat fracture (%)

B2 Medium
B2 B3
Thick B1
B3
Almost
Mixed Almost flat
KIC slanted
Plane stress (Mixed mode fracture) Plane strain
(Ductile fracture) (Brittle fracture)
Plane stress Plane strain
(Mixed mode fracture)
(Ductile fracture) (Brittle fracture)

Thickness, B (mm)
115
Fig. (i) Effect of specimen thickness over fracture toughness Fig. (ii) Effect of specimen thickness over flat fracture surface
 1.3 FRACTURE TOUGHNESS TESTING OF METALS

• Critical SIF or Fracture toughness test measures the resistance of a

material to crack extension
• There are difficulties faced to measure the fracture toughness:

• KC decreases with decreased temperature,

• KC decreases with increased strain rate and
• KC is found to be dependent for up to significant value of thickness
• Beyond this thickness i.e., for thick plate, it is independent of thickness

• A variety of organizations published standardized procedures for

measurements of Kc
• American Society for Testing and Materials (ASTM)
• British Standards Institution (BSI)
• International Institute of Standards (ISO)
• Japan Society of Mechanical Engineers (JSME)

116
 KIC Test Technique

• Code on Fracture toughness, KIC test will be discussed in this section from
the ASTM Code Designation E399-83 to determine KIC of a material
• Basis:
– Linear elastic fracture mechanics is applied to the analysis
– Entire body of the specimen is assumed to be deformed elastically and small
strain theory having linear stress-strain relation is invoked
– Loading in plane strain only
– Plastic zone size in front of the crack tip is quite small in comparison with
specimen thickness
▪ Test Technique:
a) Specimen Configurations
b) Constraints on Specimen Dimensions
c) Specimen Orientations
d) Fatigue Pre-cracking
e) Instrumentations
f) Test Procedure
g) Analysis of data record and Calculation of conditional KIC
h) Final Check for KIC

117
a) Specimen Configurations:
❖ There are five types of specimens that are permitted in ASTM
Standards that characterize fracture initiation and crack growth.

i. Compact Tension Specimen (CTS)

ii. Single Edge Notched Bend (SENB) Specimen

(ii)
(i)

118
 iii. Arc-shaped Tension (AT) Specimen
iv. Disk-shaped Compact Tension (DCT) Specimen
v. Middle Tension (MT) Panel

(iii) (iv) (v)

❖ Specimens configuration has three important characteristic dimension:

Crack length, a, Thickness, B and Width, W.
In most cases, W=2B and a/W  0.5

119
b) Constraints on Specimen Dimensions:
• Specimen plate thickness B, crack length a and width W must
satisfy the following requirements

2
 K IC 
B  2 .5  
 
 ys 

2
K 
a  2.5  IC 
 
 ys 

2
K 
W  5.0  IC 
 
 ys 

a
0.45   0.55
W

120
c) Specimen Orientation :
– Engineering materials are seldom homogeneous and isotropic
– Microstructure and mechanical properties are sensitive to
direction
– Sensitivity to orientation is pronounced in fracture toughness
measurements, because a microstructure with a preferred
orientation may contain planes of weakness where crack
propagation is relatively easy
– specimen orientation is such an important variable in fracture
toughness measurements
+
Before cold rolling a bar Elongated grains
after cold rolling

L
+

T
L- Longitudinal direction
T- Transverse direction
121
Illustration of the anisotropic nature of the material that evolved during the manufacturing.
 • Specimens are extracted from a rolled or forging operation

L-T T-L C
R-L
L-S
R-C C-R
T

L S
R
C-L
S-L
L-R L-C
T-S S-T L
L-Longitudinal
T-Transverse L-Longitudinal
S-Short Transverse C-Circumferential
R-Radial
ASTM crack plane orientation code
ASTM crack plane orientation code designation for
designation for cylindrical cross-section
rectangular cross-section
122
 d) Fatigue Pre-cracking:
– Linear elastic fracture mechanics theory applies to cracks that
are infinitely sharp prior to loading,
– While laboratory specimens invariably fall short of this ideal,
– It is possible to introduce cracks that are sufficiently sharp for
practical purposes
– Most efficient way to produce such a crack is through cyclic
loading
– Procedure involves a fatigue crack initiates at the tip of a
machined notch and grows to the designed size through careful
control of the cyclic loads
P P

Applied load, P
300 300
W/16 W/16
anotch anotch a
p
a0

0 cycle
Fatigue load
W W
Precracking length, ap 0.05W
123
(i) Before pre-cracking (ii) After minimum pre-cracking
– In order for a fracture toughness to reflect true material
properties, the fatigue crack must satisfy the following conditions:

• Crack tip radius at failure must be much larger than the initial radius
of the fatigue crack
• Plastic zone produced during fatigue cracking must be small
compared to the plastic zone at fracture

124
e) Setup and Instrumentations:
• At a minimum, the applied load and a characteristic displacement on
the specimen must be measured during a fracture toughness test
• Additional instrumentation is applied to some specimens in order to
monitor crack growth or to measure more than one displacement

➢ Load cells can be used to measure the load

125
The set-up of measuring load-displacement at the mouth of the loaded crack
 ➢ Displacement transducers (e.g., clip gauge, linear variable differential
transducer [LVDT], Potential drop technique) can be used to measure the
displacement
Specimen
Galvano-
meter
Strain gauge T1 C1
T1
T1
C1 T2 C2
C1
C2
T2 Orientation of strain
gauges on beam-1
Knife edge
V
a) Measurement of crack mouth opening displacement with clip gauge Wheatstone bridge

Coil-1 (secondary) Primary coil Coil-2 (secondary)

Current constant

Motion Secondary
Core V
V1 Difference voltage,
Primary V0=V1-V2
Core

Insulator V2

Voltage
Constant
AC voltage Secondary

Difference voltage,
V0=V1-V2 Crack length

b) Schematic of LVDT. Displacement of the central c) Potential drop method for

core causes a variation in the out put voltage monitoring crack growth
f) Test Procedure:
▪ Test set up -
✓ Centre the specimen in the loading fixtures to ensure concentricity of
loading
✓ Seat the displacement gauge such that a continuous load-displacement
record will be obtained throughout the test
▪ Loading rate -
✓ Load the specimen at a rate such that the rate of increase of stress
intensity, KIC is within the range from 0.55 to 2.7 MPam1/2/sec.
▪ Test record -
✓ A test record consisting of an autographic plot of the out put of the
loading-sensing transducer versus the out put of the displacement
gauge should be obtained
✓ Initial slope of the linear portion should be between 0.7 and 1.5.
✓ Select a combination of load-sensing transducer and autographic
recorder so that the maximum load can be determined from the test
record with an accuracy of 1.0.
✓ Continue the test until the specimen can sustain no further increase in
load
▪ Measurement -
✓ Measurements of the specimen dimensions and fracture surfaces to be
recorded. 127
 g) Analysis of data record and Calculation of conditional KIC
➢ A critical load, PQ is determined from the experimental records
➢ A criterion should guarantee that either the crack has already moved by
a small amount or it is definite to move if load is increased by a small
amount
➢ For the case of schematic test record of ductile material is presented in
following figure of Type I material with a 5% secant offset line and its
intersection with load-displacement curve gives PQ = P5.
-Modulus of elasticity(E) represents the slope of
the straight line in a - curve.
-Secant modulus(Es) represents the slope of the
line that joins the origin to the point on the -
curve at a given stress value.
-Tangent modulus(Et) represents the slope of
Pmax D
D
the tangent line drawn to the - curve at a
PQ= P5 Pmax
C given stress value.
B B
PQ Pmax=PQ
C C
P5 P5
Tangent
modulus, Et
(ys)0% secant line D
Load

Load
Load

(ys)2% secant line C F

Offset 5% Offset 5% Offset 5% ys B Secant modulus, Es

secant line secant line secant line

Stress, 
Offset 0% secant line
A A A
0.00

0.00
0.15
0.00

0.15
0.15

0.10
0.05

0.10
0.05
0.10
0.05

Displacement Displacement Displacement Modulus of elasticity, E

Type I Type II Type III Offset 2% secant line

A
Strain, 

0.00

0.06
0.04
0.02

0.08

0.14
0.10
0.12
Load-displacement test record of various materials
 ➢ The provisional fracture toughness, KQ can be computed from the
following relationship:
PQ  a 
KQ = f  
BW 1/ 2  W 
h) Final Check for KIC:
➢ The value KQ computed from above equation is a valid KIC result only
if all validity requirements in the standard are met including:
0.45  a  0.55
W 2
 KQ 
B, a = 2.5  
  ys 
Pmax  1.10 PQ
where ys =0.2% offset yield strength in tension.
➢ If this quantity is less than both thickness and the crack length of the
specimen, then
KQ = K IC

129
➢ Otherwise, it is necessary to use a larger specimen to determine KIC
in order to satisfy this requirement
➢ The dimensions of the larger specimen can be estimated on the
basis of KQ:
✓ KQ is valid if the strength or stress ratio, R < 1 and specimen has failed
before yielding.
✓ For R-values slightly above 1, still the value KQ is valid for some extent of
yielding is small
✓ For larger values of R, i.e., above 1.5, the behaviour is increasingly
elastic-plastic, indicating ductile-type behaviour.
where R is defined by the following relation:

Bending moment at failure

R=
Bending moment at yielding

130
 2.1 CRACK TIP PLASTICITY

• Linear elastic stress analysis of sharp cracks predicts infinite stresses at

the crack tip
• In practice, materials (especially metals) tend to exhibit a yield stress,
above which they deform plastically
• This means that there is always a region around the tip of a crack in a
metal, where plastic deformation occurs and hence a stress singularity
cannot exist, i.e., stresses at the crack tip are finite because the crack tip
radius must be finite
• This plastic region around the crack tip is known as crack tip plastic zone

Surface
Plastic zone
131
 Plastic zone shape in stainless steel specimen

• Elastic stress analysis becomes increasingly inaccurate as the inelastic

region at the crack tip grows
• Simple corrections to linear elastic fracture mechanics are available
when moderate crack tip yielding occurs:
– Irwin approach
– Dugdale approach or Strip yield model
• For more extensive yielding, one must apply alternative crack tip
parameters that take non-linear material behaviour into account:
– Crack Tip Opening Displacement (CTOD)
132
– J-Integral
❖ Approximate Shape and Size of Plastic Zone:
• Plastic zone plays a vital role in fracture
• The factors which control the size of the plastic zone are the shape of
the plastic zone and yield planes

• Stress field in the vicinity of a crack tip is determined by assuming the

component remains elastic at the crack tip knowing the stress field

• For evaluating the size and shape of the plastic zone by invoking one of
the following two commonly accepted yield criteria:
– Von Mises yield criterion
– Tresca yield criterion

• Since the rigorous analysis is complex and closed form solutions are
not available for most of the problems,
• Alternatively, an approximate solutions are used to find the interface
between plastic and elastic zone
133
• Let us first consider a two-dimensional body subjected with Mode I
loading.
• To find the principal stresses in xy-plane, the centre of the Mohr’s circle
for which x, y and xy are known through equation (1.45) is given by
x +y
0 =
2
1 KI    3    3  
= cos  1 − sin sin +
 1 + sin sin 
2 (2r )1/ 2 2   2 2   2 2  

KI 
= cos (2.1) 
(2r )1/ 2 2
y
y
 xy
x
yx y
r
R  x
2a
x
2 y 0 1 

xy
x
134

 • Radius of circle, R is given by
2
x −y 
R = 
2
 +  xy
2

 2 
2
 1 KI    3    3   
=  cos 
 1 − sin sin −
 1 + sin sin    +
 2 (2r ) 2  2   2 
1/ 2
2 2
2
 KI    3  
 1/ 2 
sin cos cos  
 (2r )  2 2 2 
2
 KI     2 3 2 3 
 R = 
2
cos sin 
 sin + cos 
 ( 2r )1/ 2
2 2   2 2  
 2 3
sin + cos2
3 
 =1
 2 2
KI    
 R= cos  sin  (2.2)
(2r )1/ 2
2 2
• Thus, two principal stresses are given by
KI      
1 = 0 + R = cos  1 + sin  (2.3a)
(2r )1/ 2  
2  2 
KI      
 2 = 0 − R = cos  1 − sin  (2.3b) 135
(2r )1/ 2 2   2 
• and third principal stress 3 can be determined by invoking the values
of x, y and xy:

for plane stress case:

3 = 0 (2.3c)

for plane strain case:

KI 
 3 = 2 cos (2.3d)
2r 2

136
• Plastic Zone Shape and size for Plane Stress:
➢ Von Mises yield criterion:
(1 −  2 )2 + ( 2 −  3 )2 + ( 3 − 1 )2  2 ys2 (2.4)
– Substitution of eqs. (2.3) in to eq. (2.4) and replacing r by rpz yields
plastic zone size as
KI2 
        
2
   

2
cos2   4 sin2   + 1 − sin   + 1 + sin    = 2 ys2
2rpz 2  2   2    2  
 

– Simplification yields to
max
1 K  3 2
2

rpz = I
1 + sin  + cos  (2.5)
4   2
2
ys


3 +
➢ Tresca yield criterion: 0 2 1 
 ys
 max  (2.6)
2
– For plane stress, max is given by the biggest Mohr circle which is
between 1 and 3.
– Thus at r = rpz
1 −  3  ys
= (2.7) 137
2 2
 ➢ Substituting eqs. (2.3a) and (2.3c) in to eq. (2.7), we get

1 KI       1  ys
cos  1 + sin  − (0) =
2 (2rpz )1/ 2
2   2  2 2

➢ Simplification yields plastic zone size as

2
1 K  2
  
rpz = 
I
cos  1 + sin  (2.8)
2   2
ys 2 2 

• Plastic Zone Shape and size for Plane Strain:

➢ Von Mises yield criterion:
(1 −  2 )2 + ( 2 −  3 )2 + ( 3 − 1 )2  2 ys2 (2.4)

Substitution of eqs. (2.3) in to eq. (2.4) and replacing r by rpz yields

plastic zone size as

1 KI2  3 2 
rpz = 2 
sin  + (1 − 2 )2
(1 + cos  ) (2.9)
4  ys 2 
138
➢ Tresca yield criterion:
 ys
 max  (2.6)
2
– If the Tresca yield criterion is applied then it is require to find out which
is the largest Mohr circle

– Looking carefully for 1, 2 and 3 in equation (2.3), always 1 is larger

than 2 and 3.

– Subtracting equation (2.3d) from equation (2.3b) and replacing r by rpz

we obtain
       
( 2 −  3 ) = KI
cos  1 − sin  − 2
KI
cos 
(2rpz )1/ 2
2   2   2r 2
– or
     
( 2 −  3 ) = KI
cos  1 − 2 − sin  (2.10)
(2rpz )1/ 2  
2  2 

139
▪ Equation (2.10) is used for following cases:
➢ Case (i): For small value of  (  38.90 for  = 0.33), 3 is smallest principal
stress and yielding is governed by 1 and 3 , and thus we have
equation(2.6)
1 −  3  ys
 max = =
2 2
– Substituting equations (2.3a) and (2.3d) in to above equation and
replacing r by rpz we obtain

KI      
cos  1 − 2 + sin  =  ys
(2rpz )1/ 2
2   2 

– Thus the plastic zone yields to

2  
2
1 K I2
rpz = cos 1 − 2 + sin (2.11)
2  ys
2
2  2 

140
➢ Case (ii): For a large value of  (  38.90 for  = 0.33), 2 is smallest principal
stress and yielding is governed by 1 and 2, thus equation (2.6) is given by
1 −  2  ys
 max = =
2 2
– Substituting equations (2.3a) and (2.3b) in to above equation and replacing
r by rpz we obtain
KI             
cos   1 + sin  − 1 − sin   =  ys
(2rpz )1/ 2
 2    2    2  
– Thus the plastic zone yields to
Plane Von Mises
1 KI2 Plane
rpz = sin2  (2.12) strain
stress
2  ys
2
Tresca

• Corresponding plastic zone shapes are shown  

in following Figs.
• Plastic zone size of plane strain is much
smaller than that of plane stress. (i) Von Mises (ii) Plane strain

Crack tip plastic zone for Mode I

141
 2.2 EFFECTIVE CRACK LENGTH

• Appearance of the plastic zone at the crack tip does not allow material
to bear high stresses predicted by the elastic analysis
• Presence of the plastic zone, the stiffness of the component decreases
or the compliance increases
• Consequently, the crack is equivalent to a length that is longer than
actual length
• Size of the plastic zone in front of the crack tip determines how much
longer is the effective crack length ?
• Therefore, considerable efforts have been made by many investigators
to determine the plastic zone size accurately in front of the crack tip
along say x-axis.
• Size of the crack tip yielding zone can be estimated by approximate
approach, Irwin approach and strip yield model

Plastic zone

 2a 

Crack tip plastic zone

142
(a) Approximate Approach:
• One of the simplest expression for the plastic zone size is found by looking
at the y vs. r curve for mode I loading
• On the crack plane  = 0, the normal stress y in a linear elastic material is
given by
KI
y = (2.12)
2 r

• As the first approximation, we can assume that the boundary between the
elastic and plastic behaviour occurs when the stresses given by equation
(2.12) satisfy a yield criterion
• For plane stress condition, yielding occurs when
y
 y =  ys KI
y =
2r

where ys is the uniaxial yield strength of

y = 
material.
r

First order estimate of plastic zone size for  = 0

143
• Substituting above value in to equation (2.12) and replacing r by ry gives a
first order estimate of plastic zone size as
2
1  KI 
ry = (2.13)
2   ys 
• If strain hardening is neglected, the stress distribution for r  ry can be
represented by a horizontal line at y =ys as illustrated in Fig.
• Singularity is truncated by yielding at the crack tip.
• It is clear that the actual plastic zone size must be larger than ry
• The load represented by the shaded area must still be carried through
• This can be achieved if the material immediately y
ahead of the plastic zone carries some more KI
y =
stress, which will bring this material above the 2r

yield stress. ys

y = 

a ry
First order estimate of plastic zone size for  = 0

144
(b) Irwin Plastic Zone Correction:
• Irwin suggested improved expression for the plastic zone size along the x-
axis through a model which accountable for absence of high stresses
within the yield zone
• When the yielding occurs, stresses must redistribute in order to satisfy
equilibrium
• As per the Irwin, the occurrence of plasticity makes the crack behave as if
it were longer than the physical size
• As a result of the crack tip plasticity the displacements are larger and the
stiffness is lower than in the elastic case
• i.e., the plate behaves as if it contained a crack of some what larger crack
size y
K eff
y =
• Effective crack size aeff is equal to a + ., 2 (r − ry )
the physical crack size a plus a
correction  .
y = 
• An expression for  can be easily be
derived. r

a 
aeff
145
First order estimate of plastic zone size for  = 0
• Referring to the Fig,, elastic stress distribution (y) at the crack tip of the
effective crack and stress at the tip of the effective crack is limited to the
yield stress ys
• Similarly, the stress acting on the part  in front of the physical crack is
equal to the yield stress, since forces that would be present in an elastic
material but can’t be carried in the elastic-plastic material because the
stress can’t exceed yield.
• Thus, the plastic zone must increase in size in order to accommodate these
forces
• Consequently,  must be large enough to carry the load that is lost by
cutting the area A1 from the elastic stress distribution.
y
• Hence, area A1 is equal to area A2. K eff
y =
• Distance,  in computed from following 2 (r − ry )
equation A2
A1
K eff ys
 ys = (2.14) y = 
2  r
where Keff is effective SIF accounted for
effective crack length, a + . a  
aeff
146
First order estimate of plastic zone size for  = 0
• The load not sustained on length  becomes
 
PA1 = B    y dx −  ys   (2.15)
0 
• The load sustained on length  is
PA2 = B ys  (2.16)
• Irwin correction is therefore,
PA1 = PA2
• Substituting equations (2.15) and (2.16) in to above equation
 
B    y dx −  ys   = B ys  y
y =
K eff
0  2 (r − ry )

• or A2
 A1
ys
 ys  =   ydx −  ys  y = 

0 r

 K eff 
 ys  =    dx −  ys 
1/ 2 
0  (2 x )
a  
 aeff
147
First order estimate of plastic zone size for  = 0
• Substituting value for effective SIF from equation (2.14) in to above
equation we get,
  ys (2  )1/ 2 
 ys  =    dx −  
0  (2 x )
 1/ 2  ys

• Therefore

 1/ 2 
 =   1/ 2  dx − 
0
x 
• Evaluation of integration yields,
 = y
• Therefore y =
K eff
2 (r − ry )
 =  = ry (2.17) A2
A1
• Overall plastic zone size becomes ys
y = 
r p =  +  = ry + 
r
2
1  K eff 
rp = 2 = 2   rp
2   a  
 ys  aeff
148
First order estimate of plastic zone size for  = 0
• Thus the plastic zone size is given by
2
1  K eff 

rp = (2.18)
   ys 

• Note that the size of the plastic zone rp is found to be twice as large as
the first estimate given ry.

• Effective crack length aeff is given by

aeff = a + 

=a+ (2.19a) y
K eff
y =
2 (r − ry )
2
Keff A2
=a+ (2.19b) A1
2 ys2 ys
y = 

rp
a  
aeff
149
First order estimate of plastic zone size for  = 0
• Expression for effective SIF, Keff based on the effective crack length is still
unknown.
• However, plastic zone size rp is small in comparison to the crack length, a,
effective SIF Keff may be taken same as of KI of crack length a.
• Expression for Keff can be obtained in closed form for the case of an
infinite plate.
• Effective or corrected SIF Keff, is written as
Keff =   (a +  )
1/ 2

1/ 2 
2
K eff 
= a+ 
 2 ys 
2
 y
y =
K eff
2 (r − ry )
• Simplifying we obtain effective SIF as
KI
ys y =
2 ry
 a
K eff = 1/ 2
(2.20) r
 1  
2

 
1 −  
  rp

 2   ys   a
aeq
ry= 

• Fig. illustrates the Irwin plastic zone correction. Irwin plastic zone size for  = 0
150
(c) Dugdale Strip Yield Model:
• In this model, it is assumed that a long slender plastic zone at the crack tip
in a non-hardening material in plane stress as shown in Fig.

Plastic zone

 2a 

Strip yield model

• In this approach, singularity at the tip of the effective crack is nullified by a

uniform pressure equal to yield stress ys on position of  of the crack as
shown in Fig. ys

ys

 2a 

Plastic zone is modelled by yield compressive stress at each crack tip

• Plastic zone length must be chosen such that the KI from the remote
tension ( ) and closure stress (ys ) cancel one another.
151
• Let K - singularity due to the pressure ys and K singularity due to the
remote stress  on the effective crack, we have

K + K  = 0 (2.21)

• SIF from the remote stress is given by

K =   (a +  ) (2.22)
• Stress intensity due to the closure stress can be estimated by considering
a normal force P applied to the crack at a distance x from the centre line of
the crack.
• Stress intensities for the two crack tips are given by

P a+x
K +a = (2.23a)
(a )1/ 2 a−x
P
x
P a−x
K −a = (2.23b)
(a )1/ 2 a+x
2a

Crack opening force applied at a distance x

from the centre line 152
 • Assuming plate is of unit thickness, the closure force at a point within the
strip yield zone is equal to
P = − ys dx (2.24)
• Substituting equation (2.23) we get,
−  ys dx a+x −  ys dx a−x
K +a = and K −a =
(a )1/ 2 a−x (a )1/ 2 a+x
• Total stress intensity at each crack tip resulting from the closure stresses is
obtained by replacing a with (a + ) in above equations and adding the
contribution from both crack tips, we get
a+ 
−  ys  a+ +x a+ −x
ys ys

 (a +  )1 2 a
K =  a+ −x +  dx
 a+  + x  
2a

a+ 
a+
Plastic zone is modelled by yield
dx

compressive stress at each crack tip.
= −2 ys
 a ( a +  )2 − x 2 x
P
dx

a+  a 
= −2 ys cos−1  (2.25) 2a
 a+   Crack opening force applied at a distance x
from the centre line 153
• Substituting values of K and K from equations (2.21) and (2.25) in to
equation (2.21), i.e.,

K + K  = 0

 a+ −1 a 
  (a +  ) + − 2 ys cos    = 0
  a+   

1  a 
  (a +  ) = 2 ys a +  cos −1 
 a+  

  a 
= cos−1 
2 ys a+  

a   
• Thus, = cos  (2.26)
a+  2 
 ys 
• Note that  approaches to infinity as  → ys.
154
• To find the effective SIF Keff, performing a Taylor series solution on
equation(2.26) 2 4 6
a 1    1    1   
= 1− + - +
a+      
2!  2 ys  4!  2 ys  6!  2 ys 
• Neglecting all but the first two terms
2
a 1   
= 1−
a+ 2!  2 ys 
• and solving for the plastic zone size yields,
2
a       2 a    K I2 

=  
=
  = (2.27)
8   ys  8   ys 
2
8   ys
2 

for  < < ys.
• Thus the effective SIF Keff is set to aeff equal to (a + ):

Keff =   aeff =   (a +  )

  2 2 a    
=   a +  =   a sec  

(2.28)
 8 ys 
2
 2 ys 
155
 3 ELASTIC-PLASTIC FRACTURE MECHANICS

• Linear elastic fracture mechanics (LEFM) is valid only as long as

nonlinear material deformation is confirmed to a small region surrounding
the crack-tip

• In many materials, it is virtually impossible to characterize the fracture

behaviour with LEFM and an alternative fracture mechanics is required.

• Elastic-plastic fracture mechanics applies to materials that exhibit time-

independent, nonlinear behaviour (i.e., plastic deformation).

• Following two elastic-plastic parameters are used to describe crack tip

conditions in elastic-plastic materials:
a) Crack-tip opening Displacement
b) Nonlinear energy release rate or J-integral

• In these cases the fracture criterion can be established

156
 3.1 Crack-tip Opening Displacement(CTOD)

• In 1961, Wells proposed that the fracture behaviour in the vicinity of a

sharp crack could be characterized by the opening of the notch faces,
namely CTOD.
• While examining fractured test specimens, Well noticed that the crack face
had moved apart prior to fracture: plastic deformation blunted an initially
sharp crack as shown in Fig.3.1.
• Degree of crack blunting increased in proportion to the toughness of the
material.
• This observation led Wells to propose the opening at the crack tip as a
measure of fracture toughness.
• This parameter is known as the crack-tip opening displacement(CTOD).

aeff
a
Fig. 3.1 An initially sharp crack blunts
CTOD with plastic deformation, resulting in a
finite displacement at the crack tip.

157
• Expression for CTOD can be obtained by using the expression of crack
opening displacement (COD) of mode I.
• COD is the distance between the two crack faces as illustrated in Fig. 3.2.
• COD of a blunt cracked plate subjected to plane stress loading will be
obtained by invoking equation (1.38b) for y=0, thus the displacement in y-
direction is given by
1  2 
v=  Im ZI (3.1) 
2 1 +  
• For this problem, the typical Westergaard
function, ZI and ZI are given by
iy

z
ZI ( z ) =
( )
1/ 2
v
x 
z2 − a2 x
COD

− i x
=  z = x + iy = x
(a 2
−x )
2 1/ 2

• Separating real and imaginary terms, we have

Fig. 3.2 Centre crack in an infinite plate.
− x
Re ZI = 0 and Im ZI =
(a 2
− x2 )
1/ 2
158
• Integral value of Westergaard function is given by

(
ZI (z) =  z2 − a2 )
1/ 2
(
= i  a2 − x 2 ) 1/ 2

• Separating real and imaginary terms, we have

Re ZI = 0 and Im ZI =  a2 − x 2 ( )
1/ 2

• Substituting above values in to equation (3.1) , we obtain displacement in

y-direction for plane stress case as 

1  2 2
v= 
2 1 + 
(
a − x2 
1/ 2 
)
(3.2)
 iy
• Thus, the crack opening displacement at a
distance of x from the crack centre is v
x 
therefore, COD

1  2 2
( )
x
1/ 2 
COD = 2v = 2  a − x2 
2 1 +  
1  2 2
= 
 1 + 
a − x2 ( ) 1/ 2 

 Fig. 3.2 Centre crack in an infinite plate.

4 2
 COD =
E
a − x2 ( )
1/ 2
(3.3)  =
E
2(1 +  )
159
• Maximum crack opening displacement occurs at x=0, thus, we have
4 a
CODmax = (3.4)
E
• Referring to Fig. 3.3, for plane stress the COD at a distance x is given by
4 2
COD =
E
(
aeff − x 2 )
1/ 2
(3.5a)

• and for plane strain is given by
4 (1 − 2 ) 2
COD =
E
(
aeff − x 2 )
1/ 2
(3.5b) y Blunt crack
Actual crack COD
where aeff is effective crack length and
can be estimated using Irwin’s x

correction(Equation 2.19a) as 2a x
2aeff
aeff = a + 
rp
=a+
2
• Substituting value of in equation (3.5a), Fig. 3.3 Effective crack length, COD and CTOD.

we have
160
 1/ 2
 2 
COD =
4  a + rp  − x 2 
  (3.6)
E  2  
 
• For evaluating CTOD, replace x by a and we obtain
1/ 2
4   
2
rp  2
CTOD = a +  − a 
E  2
 
1/ 2 
4  rp 
=  4a + rp  (3.7)
E 4 
• For LEFM, rp may be neglected in
y Blunt crack
comparison to 4a and the expression Actual crack COD
CTOD
simplifies to
 4 rp a
1/ 2 x
4  rp
CTOD =  4a = (3.8) 2a x
E 4  E 2aeff

• where rp is given by equation (2.18)

KI2
rp = (3.9)
 2
ys
Fig. 3.3 Effective crack length, COD and CTOD.

• In this analysis small scale yielding is considered. 161

• Approximate relation for CTOD can be obtained by determining the KI on
the actual crack length, a, then

CTOD =
4
a
K I
2
=
4
a
( a )
2

=
4
a
 2a
=
4  2a 2
E  2
ys E  ys2 E  ys2 E  ys2

4 2a 4KI2
= =
E ys  E ys
4G
 CTOD = I
(3.10a)
 ys

• where GI is the energy release rate for mode I loading.

• If Dogdale approach is used, it can be shown that

KI2
CTOD =
E ys
2

G
= I
(3.10b)
 2
ys
162
 3.2 J-CONTOUR INTEGRAL
• James R. Rice has developed J-contour integral to characterize fracture
parameters for nonlinear materials.

• He applied deformation plasticity i.e., by idealizing the elastic-plastic

deformation as nonlinear elastic for the analysis of a crack in a nonlinear
elastic body.

• He showed that the nonlinear energy release rate, which he called J, could
be written as path-independent line integral.

• Further he showed that J-integral uniquely characterizes crack tip stresses

and strains in nonlinear materials.

• Thus the J-integral can be viewed as both an energy parameter and a

stress intensity parameter.

163
• Fig. 3.4 illustrates the uniaxial stress-strain behaviour of elastic-plastic and
nonlinear elastic materials.
• Loading behaviour for the two materials is identical, i.e., follow same path (A
to B).
• But the material responses differ when each is unloaded.
• Elastic-plastic material follows a linear unloading path (B to C) with the slope
equal to Young’s modulus, while the nonlinear elastic material unloads along
the same path as it was loaded (B to A).
• There is unique relationship between stress
and strain in an elastic material, but a given Nonlinear elastic

strain in an elastic-plastic material can B

correspond to more than one stress value if

Stress
the material is unloaded or cyclically loaded.
• Consequently, it is much easier to analyze Elastic-plastic

an elastic material than a material that

Linear elastic
exhibits irreversible plasticity.
• As long as the stresses in both materials A C Strain
increases monotonically, the mechanical Fig. 3.4 Comparison of the stress-
response of the materials is identical. strain behaviour of two materials
164
• Thus, an analysis that assumes nonlinear elastic behaviour may be valid
for an elastic-plastic material, provided no unloaded occurs.
• Deformation theory of plasticity, which relates total strains to stresses in a
material, is equivalent to nonlinear elasticity.

Mathematical Model:
• Energy release rate for linear material given by equation (1.13), same
holds for nonlinear elastic materials, except that symbol G is replaced by
J. i.e.,
d
J =− (3.11)
dA
where  is the potential energy of the body with crack and A is the crack
area.
• The potential energy is given by
 = U − Wext (3.12)

where U- strain energy stored in the body and Wext –work done by external
forces.
165
• Consider a plate with crack (Fig. 3.5a) which exhibits a nonlinear load-
displacement curve as shown in Fig. 3.5b.
• If the plate has unit thickness, then area, A=a, for load control, the
potential energy is given by
 = U − Pu  Wext = Pu

  = −U  (3.13)
• where U is the complimentary strain energy defined by
P
U  =  u dP (3.14)
0

P dP
U
u+du
u dU= -dU

Load
a a +da U
P
P
u Disp.

Fig. 3.5a Plate with crack under loading Fig. 3.5b Nonlinear energy release rate
166
• Thus, if the plate shown in Fig. 3.5a, is in load control, the J-integral is given
by  dU  
J =   (3.15)
 da P
• Similarly, if the crack advances at a fixed displacement, i.e., Wext =0, and J-
integral is given by
 dU 
J =  (3.16)
 da u
• According to Fig.3.5b, dU for load control
differs from –dU for displacement control by
the amount of (1/2)dPdu, which is
du
vanishingly small compared to dU P dP
U

• dU= -dU

Load
Therefore, J for load control is equal to J for
displacement control.
U

u Disp.
Fig. 3.5b Nonlinear energy release rate
167
• By invoking the definition for U and U , we can express J in terms of load
control and displacement control as
  P  P
 u 
J =   u dP  =    dP (3.17)
 a 0 P 0
a P

  u  u
 P 
or J = −  P du  = −   du (3.18)
 a 0 u 0
a u
• Integrating both equations (3.17) and (3.18) by parts leads to equal and J
the same for fixed load and fixed grip conditions.
• Thus, J is a more general version of the energy release rate.
• For special case of linear elastic material,
J =G
• Also for Mode I loading, we have
KI2
J= (3.19)
E

168
J as a Path-Independent Line Integral:
• Consider an arbitrary counter-clockwise path ( ) around the tip of a crack
as shown in Fig. 3.6
• The J – integral is given by
 u 
J =  w dy − Ti i ds (3.20)

x 
• where w- strain energy density, Ti are components of the traction vector,
ui are the displacement vector components and ds is a length increment
along the contour i. y
• The strain energy density is defined as
ij

w =   ij dij (3.21)
0 
• where ij and ij are the stress and strain x
tensors, respectively.
ds
• Traction is a stress vector acting normal to Ti
the contour i, and are given by ni

Ti =  ij n j (3.22)
Fig. 3.6 Path around the crack-tip with
• where ni are the components of the unit outward normal ni and traction Ti
vector normal to . 169
 • To prove that the J-integral is independent path, let us consider an area A
of a plate not containing any crack or damage or void and surrounded by a
closed path  as shown in Fig. 3.7a.
• Let the value of integral J having same integrand as that of J-integral but
evaluated along the closed path  vanishes, i.e.,
 u 
J  =  w dy − Ti i ds = 0 (3.23)
Γ
 x 
• Ref. Fig. 3.7b, substituting dy=n1ds and Ti =ij nj in the above equation, we
get
u 


J  = w n1 −  ij n j i ds (3.24)
 x 

Γ
y
n

*
A* x 1
 n2
dy
S=0

dx n1
Fig. 3.7a Area A* without any crack and enclosed by 170
Fig. 3.7b Relation between n1 and dy.
contour  *
• Above integral will be converted from the line integral to area integral
using the Divergence theorem which is stated in general form as
M
 M n j ds =  xi =  M,i dA

dA (3.25)
Γ A A
• where M is the any variable.
• Invoking the Divergence theorem to equation (3.24), we get

 w  ui    
J = 

−   ij  dA (3.26)
 x 
A 
x , j 

• By invoking the definition of strain energy density given by equation (3.21)
yields to

 w ij  ui  ui 
 
J = 

−  ij   −  ij , j dA (3.27)
 ij x
A 
 x , j x 
• For elastic materials (linear or nonlinear)
w
 ij = (3.28)
ij
• and invoking the equilibrium equation,
 ij
=  ij , j = 0 171
x j
• Simplifying equation (3.27) leads to

  u 
J  =   ij ij −  ij i , j dA (3.29)
A 
x x 
• Further simplification, we note that
w ij 1 (ui , j + u j ,i )
=  ij =  ij (3.30)
x x 2 x
• And owing to symmetry of
 ji =  ij
• also
u j ,i u j ,i
 ij =  ji (3.31)
x x
• Interchanging the dummy indices on right-hand side, we obtain
u j ,i ui , j
 ij =  ij (3.32)
x x
• And then equation (3.30) is simplified to
ij ui , j
 ij =  ij (3.33)
x x 172
• Substituting above values in equation (3.29), we have
 u u 
J  =   ij i , j −  ij i , j dA (3.34)
A 
x x 

 J = 0 (3.35)

• This proves that the left-hand side of equation (3.34) is zero for a closed
curve and is valid for all materials whose constitute equation can be defined
through the relation equation (3.28).
• This result will now be used to prove the path-independent of J-integral.

173
• Consider an area A shown Fig. 3.8, which is closed by four segments,
BCD, DE, EFG and GB.

• Segments DE and GB are coincide with the crack surfaces

• Segments BCD and EFG can be chosen arbitrarily within the material as
long as they don’t cross each other.

• Referring to the contour, applying the A*

EFG
equation (3.23) by splitting in to four
segments as
BCD
D DE F
J = JBCD + JDE + JEFG + JGB E

B GB G
• or

J= 
BCD
+ 
DE
+ 
EFG
+ 
GB
=0 (3.34) C

Fig. 3.8 Area A* enclosed by two curves around the

crack-tip and two segments along the crack surfaces.
174
• Referring to the Fig. 3.9, integration along DE and GB don’t contribute
because, dy=0 and Ti = 0 for traction free crack surfaces considered here.
• Thus two surviving integral are


BCD
+ 
EFG
=0 (3.35)

• Integral along BCD is counterclockwise direction with its normal pointing in

to the outside region and integral along EFG is clockwise with its normal
pointing inside.
A*
EFG

BCD
BCD D DE F
E
EFG
B GB G

F
D E
C
B G

Fig. 3.8 Area A* enclosed by two curves around the

C
crack-tip and two segments along the crack surfaces.
175
Fig. 3.9 Path independence of J-integral.
• The direction of the path EFG can be changed from clockwise to
counterclockwise by introducing a negative sign.
• Incorporation of this into equation (3.35) we get


BCD
−
EFG
 =0

• or

BCD
= 
EFG
(3.37)

• Alternatively,

Γ1
= 
Γ2
(3.38)
EFG A*

BCD
BCD = 1 D DE F
E
EFG =2
B GB G

F
D E
C
B G

Fig. 3.8 Area A* enclosed by two curves around the

C
crack-tip and two segments along the crack surfaces.
176
Fig. 3.9 Path independence of J-integral.
• Thus, that the path  of J – integral can be chosen anywhere within the
material.
• It is noted that path independent of J-integral is valid only for cases in
which equation (3.28) is applicable, i,e.,
w
 ij = (3.28)
ij
• Integration path  can be chosen far away from the crack-tip to avoid the
plastic zone or it may be taken very close to the crack-tip so that it poses
through the plastic zone.
• In any case J-integral is same provided there
is no unloading. 1=BCD
2=EFG
• This suggest that J-integral is fully capable of
characterizing a crack. F
D E

B G

Fig. 3.9 Path independence of J-integral.

177
Equivalence of G and J for Elastic Materials:
• For linear elastic material, J-integral represents the potentiality for release
of energy from the system per unit area extension of the crack growth and
is same as G.
Proof:
• Consider the general case of a planner body of area A with no body
forces (Fig. 3.10).
• Contour of this body is represented with curve 0 which is split into two
parts, T where traction is prescribed and u on which displacement
conditions are given.
• Note that the free surface belongs to T . 

• Potential energy  is then given by  = +

0 u T
 T

 =  w dA −  Ti ui ds (3.29)
A ΓT 0
Ti
d
G=−
da

 
u
dw dui
=− dA + Ti ds (3.30) Fig. 3.10 Path 0 chosen to coincide with
da da the body contour and cracked surface.
 ΓT 178
A
• Transforming the origin to the crack tip by the relation, 
0=u+T
X = x −a (3.31) T

• We obtain
0
d    X 
= + = + Ti
da a a a a X
• From equation (3.31), we have
u
X   Fig. 3.10 Path 0 chosen to coincide with
= −1 and =
a X x
the body contour and cracked surface.

• Then the differentiation relation simplifies to

d  
= -
da a x
• Operating it on equation (3.30), we obtain
w w u u
G = − dA +  dA +  Ti i ds −  Ti i ds (3.32)
A
a A
x ΓT
a ΓT
x
• Integral of the first term is modified as follows:
w w ij 1 (ui , j + u j ,i )
= =  ij
a ij a 2 a 179
• Making use of the symmetry of stress tensor, we obtain 
0=u+T
w ui , j  u  T
=  ij =  ij  i  (3.33)
a a  a , j
0
• Invoking the principle of virtual work for third Ti

term in equation (3.32), we have

u  u 
 
u
Ti i ds =  ij  i  dA (3.34) Fig. 3.10 Path 0 chosen to coincide with
a  a , j the body contour and cracked surface.
ΓT 
A
• Substituting equation (3.33) and (3.34) into equation (3.32), we obtain
w ui Note that ui =0 on u
 a Γ i x ds

G= dA − T (3.35)
A 0 and  0=T .
• First term now being converted into line integral by invoking the divergence
theorem for which closed contour 0 is chosen to coincide with the body
contour and the crack faces as shown with the dashed lines.
• Although the singularity of the crack-tip lies on the path 0 it is not serious
for the first term of the equation (3.35)
• For small values of y, strain energy density w varies as 1/r, because it is a
product of stress and strain, each varying as 1/r(1/2) , 180
• Therefore, w/x varies as 1/r but dA varies as r2 and thus singularity is
nullified when r→0.
• Then the divergence theorem modifies equation (3.35) to 
0=u+T
u
G =  w n1 ds −  Ti i ds
T
(3.36)
Γ0 Γ0
x
0
• The contour 0 can be split into two parts, Ti

Γ0 = Γ + Γc u
Fig. 3.10 Path 0 chosen to coincide with
the body contour and cracked surface.
• Where c is the portion on crack surfaces and  is the usual integration path
of J-integral.
• Also noting that on c path dy=0 and Ti=0.
• Thus the equation (3.36) is simplified to
ui
G =  w dy −  Ti ds (3.37)
Γ Γ
x
• The right-hand side has the expression of J which is path independent ,
thus for linear elastic materials, expression of J-integral is restatement of G.
• That is, 181
J =G (3.38)
 Equivalence Between CTOD and J-integral:
• Both CTOD and J-integral characterize elastic–plastic fracture and
therefore a relation should exist between them.
• Consider a crack of actual length a and an effective crack length aeff whose
crack-tip is located at point B as shown in Fig. 3.11.
• Distance AC gives CTOD.
• An approximate integration path ABC is chosen for determining J-integral
y
which is restated as
 ys
 u 
J =  w dy − Ti i ds  (3.39) C
Γ 
x  
  B x
A
• Along the integration path dy is negligibly a
 ys
small making the first term of J to be aeff
negligible. Fig. 3.11 Finding J-integral along
with the path ABC.
• If material is considered to be elastic-
plastic, Ti =ys on  and J simplifies to
y
v C

J = −  Ti ds =   ys d (v BC − v AB ) =  ys (v BC + v AB )
dy B

x dx x
Γ
Detail of upper side of crack zone
 J =  ysCTOD (3.40) along y=0.
182
• Leading to
J
CTOD = (3.41)
 ys

• However, the material within the plastic zone may deform with work
hardening characterized by factor  .
• The relation is then written in more general form as

J
CTOD = (3.42)
  ys
• where  is a dimensionless factor that depends on .

183
COMPUTATIONAL FRACTURE MECHANICS

184
 INTRODUCTION
– It is often necessary to determine the distribution of stresses and
strains in a body that is subjected to external loads or displacements
– In limited cases, it is possible to obtain a closed-form analytical solution
for the stresses and strains
– Westergaard and William used an approaches to derive solutions for
the stresses and strains near the tip of a sharp crack in elastic material
– However, closed-form solutions are not possible and the stresses and
strains in the body must be estimated numerically
– Rapid advances in computer technology are primarily responsible for
the exponential growth in application of computational fracture
mechanics
– Numerical modelling has become an potential tool in fracture analysis,
since relatively few practical problems have closed-form analytical
solutions
– A variety of numerical techniques have been applied to problems in
solid mechanics, including Finite Difference, Finite Element and
Boundary Integral Equation Methods,
– Vast majority of analyses of cracked bodies utilize Finite Element
Method and Boundary Integral Method may be useful in limited
185
circumstances
 APPLICATIONS

• Importance of fracture mechanics techniques for the analysis of

mechanical components containing defects like flaw/crack has increased
considerably over recent years.
• If the value of fracture parameters e.g., stress intensity factor exceeds a
critical value, crack instability is assumed to exist.
• the evaluation of fracture parameters at a given crack-tip configuration is all
that is required to asses whether failure will occur or not.
• There are several analytical and experimental solutions for evaluation of
stress intensity factor. However, the experimental solutions are
considerably expensive and the set of analytical solutions are restricted to
a few number of practical situations.
• Into this grain the computational mechanics appear such as powerful and
economic tool on stress intensity factor evaluation. Using finite element
method is possible to obtain the stresses and displacements fields at the
crack tip
• The stress intensity factor can be derived by substituting computed values
of stresses or displacements into the known crack-tip field equations or
from energy considerations..
186
Potential use of Fracture Mechanics found in
a) Engineering:
• Aerospace components
• Ship components
• Critical locomotive components
• Critical automobile components
b) Medical sciences
• Analysis of bones
c) Dental sciences
• Analysis of tooth
Component made from Material of the type:
• Elastic-linear materials
• Elastic-plastic materials
• Visco-elastic-materials
• Visco-plastic materials
To Analyze the problems under loading of type
• Static (time-independent)
• Dynamic (time-dependent) 187
 Derivation of Finite Element Differential Equations for Static
Problems using Principle of Minimum Potential Energy

Step 1. Select the element type and discretize the body or system into “E”
number of finite elements.

Step 2. Select the displacement or interpolation model of eth element as

 u ( x, y , z ) 
 
{U e ( x, y , z )} =  v ( x, y , z )  = [N ( x, y , z )]{q e } (5.1)
w ( x, y , z )
 
where [N]- shape function matrix, {qe}- nodal displacement vector

188
Fig. 1a Beam
Fig. 1d FE modelling with 3-D elements
Step 3. Define the strain-displacement relations and stress-strain relations.
 
 x 0 0
  
x 
e
0 0
   y 
 y     u e
 z  0 0
z   v 
  =  
 
 xy   0  w 
 yz   y x  
     
 zx  0 z y 
 
 0
 
 z x 
or { e } = [B]{q e } ...... (5.2)

1 −    0 0 0 
 x      x 
e e
1−  0 0 0
    
 y    1− 0 0 0   y 
 z   1 − 2  z 
  =
E
 0 0 0 0 0   
 xy  (1 +  )(1 − 2 )    xy 
2
1 − 2
 yz   0 0 0 0 0   yz 
   2  
 zx   0 1 − 2   xz 
0 0 0 0
 2 
or { e } = [D]{ e } = [D][B]{q e } (5.3)
Step 4: Derive element stiffness matrices and load vectors from total
potential energy of the element:
– The potential energy of an eth element can be expressed as

 eP = SE e + WP e
Strain energy stored in an element eth
1
2 V
SE e = { e T
} { e
} dV
e

1 e T 
{q }   B  D B dV  {q e }
or
=
T

2 V e

Work potential of an element subjected to external loads such as
traction force {Te}, body force {Fe} and concentrated force {Pe}, is
given by

WP e = −  {U e } T {T e }dS −  {U e } T {F e }dV −  {Uie } T {Pi e }

Se Ve i

   
or WP = −{q }   {N } {T }dS  − {q }   {N } {F }dV  − {q e } T {P e }
e
 e TT e
 e T
 T e

S e
 V  e
Total potential energy of an element is given by

1 e T  e  
 = {q }   [B ] [D][B ]dV {q } − {q }   {N } {T }dS 
e
P
 T
 e T
 T e

2 V e
 S  e

 
− {q }   {N } {F }dV  − {q e } T {P e }
e T T e

V e

where,
Element stiffness matrix, [K ] =  [D][B]dV
e T
[ B ]
Ve

Element force vector due to traction, {FT } = 

e T e
{N } {T }dS
Se

Element force vector due to body force, {FB } = 

e T e
{N } {F }dV
Ve

Element force vector due to point load, {FP } = {P }

e e
 Step 5: Derive global stiffness matrix and load vectors from the total potential
energy of the whole elastic body.
– If elastic body is divided into E number of elements, then the total
potential energy of the whole body is obtained by direct addition of the
individual element potential energy, i.e.,
E
1  E   E 
P =   = {q }    [B ] [D][B ]dV {q } − {q }    {N } {T }dS 
  
e T T T T e
P
e =1
2  e =1 V
e
  e =1 S e

 E  E
− {q }    {N } {F }dV  − {q }  {P e }
T T e T

 e =1 V e
 e =1
1
Or = {q } T [K ]{q } dV − {q } T {FT } − {q } T {FB } − {q } T {FP } (5.4)
2
E

  [D][B]dV
where
Global stiffness matrix, [K ] =
T
[ B ]
e =1 V e
E
Global force vector due to traction, {FT } =  {N }
e =1 S e
T
{T e
}dS
E
Global force vector due to body force, {FB } =  {N } T

e =1 V e
{F e
}dV
E
Global force vector due to point load, {FP } =  {P
e =1
e
}
Step 6: Derive finite element governing equation of the whole body
– Applying Principle of minimum potential energy

 =   {q } T [K ]{q } dV − {q } T {FT } − {q } T {FB } − {q } T {FP }  = 0

1
2 
we get the governing equilibrium equation as

[K ]{q } − {FT } − {FB } − {FP } = 0 (5.5)

– Let us define global load vector {F} = {FT} + {FB} + {FP}, and also use
subscripts with “o” to denote original matrix and vectors, therefore,

[K ]o {q } o − {F } o = 0o (5.6)
Step 7: Imposing boundary conditions and eliminating the rows and columns
corresponding to those known null displacements, results in reduced global
equilibrium equation as
[K ]r {q } r − {F } r = 0r (5.7)
– Above equation can used to solve for global nodal displacements vector
using either Inversion method or Gauss elimination method or
Cholesky’s method, etc.
Step 8: Computation of strain and stress vectors for each elements using
step 3
– Once the global nodal displacements are computed, the nodal
displacement vector for eth element is extracted separately.
– Substituting nodal displacement vector of eth element {qe} in to
equations (5.2) and (5.3) to get element strain vector and stress
vector, respectively.

Step 9: Computation of reaction vector using step 6

– Reactions at the supports are computed by using original global
equilibrium equation (5.6). That is the global reaction vector is given
by
{R } o = [K ]o {q } o − {F } o (5.8)
 Development of Eight-Noded Quadrilateral Singular Finite Element

• Importance of fracture mechanics techniques for the analysis of machine

component containing cracks has increased considerably over recent
years.
• For situations where yielding in the region of the crack tip is very
restricted, linear elastic stress analysis implies that the local field of
stresses and displacements depend on a parameter termed the stress
intensity factor, namely, K.
• Value of K depends on the loading and the component geometry.
• If the value of K exceeds a critical value, Kc, crack instability is assumed to
exist.

y
Lc= L
a
x
h

L
b

195
• There are several analytical and experimental solutions for evaluation of
stress intensity factor.
• However, the experimental solutions are considerably expensive and the
set of analytical solutions are restricted to a few number of practical
situations.
• Using finite element method, it is possible to obtain the stresses and
displacements fields at the crack tip.
• Thus stress intensity factor can be derived by computed values of stresses
or displacements into the known crack tip field equations or from energy
considerations.
• Quadratic iso-parametric element developed by Barsoum was used in the
solution of two dimensional crack problems giving very accurate results.
• Original Barsoum element is a eight-noded iso-parametric rectangle,
however, it is possible to derive a six-noded iso-parametric triangle by
collapsing one of its sides.

196
• Use of conventional quadratic quadrilateral isoparametric elements near
the crack tip, even with very fine mesh near would not simulate the stress
field conditions of the crack tip 1/r.
• Barsoum showed that by moving mid-side nodes of isoparametric
element to quarter points induced 1/r singularity and improved the
performance of the analysis enormously.




y
y KI
y =
2r
xy
x  =
r
 x
2a 2a x

Singularity dominated zone

 

197
Formulation of the isoparametric finite element
• Geometry of the element can be mapped into a normalized square element
using natural coordinates through the transformations as
8 8
x =  Ni ( , )xi and y =  Ni ( , )y i (5.9)
i =1 i =1

• Shape functions at ith node are defined by

 i2i2
Ni = (1 +  i )(1 + i ) − (1 − 2 )(1 + i ) − (1 −  )(1 +  i )2
2 2

+ (1 −  2 )(1 + i )(1 −  i2 ) i + (1 −  2 )(1 +  i )(1 − i2 ) i 4 (5.10)

2 2
7
• Field variables are interpolated by y 4 3

8 
u =  Ni ( , )ui (5.11a )
i =1
8 6
8 (0,0) 
v =  Ni ( , )v i (5.11b ) v(,)
i =1
u(,)

1 2
5
L x
198
• Element strain vector is given by

ui 
  = B  (5.12)
e

v i 
 Ni 
 0 
 x 
Ni 
• Where B  =  0
 y 
 Ni Ni 
 
 y x 

 Ni   Ni 
 x   
−1   
• and  N  = J   N  (5.13)
 i  i
 y    
 x y 
   
• where Jacobian matrix is J  =  x y 


    199
 • Element stress vector is given by
{ e } = [D]{ e } (5.14)
• where [D] is material property matrix
• Element stiffness matrix is then obtained as
+1 +1
[K ] =   [B]T [D][B] J dζ dη
e
 (5.15)
-1 -1
• In order to obtain stress or strain singularity at the crack-tip, equations
(5.12) and (5.14) should be singular.
• This singularity is achieved placing the mid-side node at the quarter-point
position of7 the sides adjacent to a crack tip. 7
y 4 3 4 3

 

8 6 6
(0,0)  (0,0) 
v(,)
8
u(,) L/4
5
1 2 1 2
5 Crack tip L/4
L
x 200
• Shape functions of Ni(,) in all isoparametric elements are polynomials and
hence, at Ni/ and Ni/ are all nonsingular.
• On the other hand, combining equations (5.12) and (5.13) the strain
equation (12) can be rewritten as
ui 
  = J B ( , ) 
e −1 *
(5.16)
v i 
 Ni 
  0 
 
Ni 
where B *  =  0  
 Ni Ni 
 
   

• Singularity for {e} in equation (5.16) could be achieved by requiring that the
Jacobian [J] be singular at the crack tip. i.e., the determinant of the
Jacobian |J| has to vanish at the crack tip.

201
 • For the 8-noded quadrilateral element with mid-side nodes on two sides at
the quarter-point the singularity may occur on both sides 1−5−2 or 1−8−4,
but, for simplicity, evaluate the singularity along the line 1−5−2 ( =−1).
• Shape functions evaluated along the line at nodes 1, 5 and 2 (x-direction)
are 1 1
N1 = −  (1 −  ); N2 =  (1 +  ); N5 = (1 −  2 ).
2 2
• Substituting above values into equation (5.9) we have
1 1
x = −  (1 −  )x1 +  (1 +  )x2 + (1 −  2 )x5 (5.17)
2 2
7 7
y 4 3 4 3

 

8 6 6
(0,0)  (0,0) 
v(,)
8
u(,) L/4
5
1 2 1 2
5 Crack tip L/4
L
x 202
 L
• Referring the geometry of the element by setting x1 = 0, x2 = L and x5 =
4
• Substituting above values in to equation (5.17)
1 2 L
x =  (1 +  )L + (1 −  ) (5.18)
2 4
• Solving for  we have
x
 = −1 + 2 (5.19)
L

7 7
y 4 3 4 3

 

8 6 6
(0,0)  (0,0) 
v(,)
8
u(,) L/4
5
1 2 1 2
5 Crack tip L/4
L
x 203
 • and therefore Jacobian is given by
x L
J= = (1 +  ) = xL (5.20)
 2

• As x → 0, det [J ]= 0, i.e., determinant of the Jacobian vanish at the crack

tip leading to singularity.

7 7
y 4 3 4 3

 

8 6 6
(0,0)  (0,0) 
v(,)
8
u(,) L/4
5
1 2 1 2
5 Crack tip L/4
L
x 204
 • Considering only the displacement along the line of nodes 1, 5 and 2, we
have
1 1
u = −  (1 −  )u1 +  (1 +  )u2 + (1 −  2 )u5 (5.21)
2 2
• Substituting equation (5.19) for value of  into above equation, we get

1 x  x  x  x   x x
u = −  − 1 + 2  2 − 2 u1 +  − 1 + 2  2 u2 + 4  − u5 (5.22)
2 L   L  L  L   L L

7 7
y 4 3 4 3

 

8 6 6
(0,0)  (0,0) 
v(,)
8
u(,) L/4
5
1 2 1 2
5 Crack tip L/4
L
x 205
• Thus, the strain in the line 1-5-2 is obtain as
u u  u
x = = J −1 =
x  x 
1 3 4 1 1 4  2 4
=−  − u1 +  − + u2 +  − u5 (5.23)
2  xL L  2 xL L   xL L 
• The singularity achieved for the strain field on equation (5.23) is of order
1/x, therefore, at the crack tip, the stress on equation (5.14) is infinite,
which is the same singularity achieved beyond the elastic plane analysis.
• As x→ 0, x →  and x → , this shows the stress and strain at the
crack-tip leads to singular.
• Six-noded triangular singular elements are placed around the crack-tip
and are generated by 8-noded quadrilateral element.

206
 • These are generated by collapsing the side 1−8−4 of the quadrilateral
element.
• The geometry along the x-axis, where = 0, and for this line, is given by
1 1 1 L 1 1 L
x = − (1 +  )(1 −  )L1 − (1 +  )(1 −  )L1 + (1 −  2 ) 1 + (1 +  )L1 + (1 −  2 ) 1
4 4 2 4 2 2 4
• Simplification which yield to as similar to equation (5.18), i.e.,
1 L
x =  (1 +  )L1 + (1 −  2 ) 1
2 4
• and solving for  we obtain as identical to equation (5.19), i.e.,
x
 = −1 + 2
L1
7
y 4 3
3 3
y y

7
7
 
8 6 1,8,4 1,8,4
(0,0)  6 6
v(,) (0,0)  (0,0) 
x
5
u(,) L1/4
5 Crack tip
1 2
5 2 2
L L1 207
x
 MESHING RULES FOR CRACK MODELLING
• Moving of the nodes from midpoint to the quarter point, the shape function
in global coordinate becomes more similar to the 1/√r function.
• However, when the nodes are collapsed to produce the 1/√r singularity, the
element reflects only the near field behavior and its size should be restricted
to that of the region of validity of the near‐field equation.
• Pu and Hussain noted that the 12‐node collapsed element gave good
results for problems they investigated if r0 were restricted to 1 – 2% of the
crack length, a, which is comparable with the size of the singularity
dominated‐zone
• In particular, a small number of Quarter-point singular finite element (QPE)
surrounding the crack tip results in inadequate modeling of the
circumferential displacements, while too small a span angle introduces
errors due to excessive element distortion.
• Therefore, it would seem that employing between 6 and 8 QPEs are
reasonable [Lim et al.]

208
• Under special configuration, transitional elements improve the accuracy of
stress intensity factor computations.
• These elements are located in the immediate vicinity of the singular
elements with the mid‐side nodes so adjusted as to reflect or extrapolate
the square root singularity on the stress and strains at the tip of the crack.
• Midpoint of the transitional element can be obtained by using following
equation. L + 2 L +1
= (5.24)
4L
• Finally, rest of the region of the component meshed using regular
isoparametric elements.
Transitional element

Quarter- point y
singular element

Crack-tip
(L,0) (L,0) x
1
L

Isoparametric element
209
Mesh Size:
• In general, there is no optimal numbers for the size of the each element.
• Mesh size should be determined in each problem so that good accuracy is
obtained.
• However, there exist some suggestions in order to define the size of the
mesh.
• For the collapsed quadrilateral QPE, the recommended ratio for the crack
length and the distance between crack-tip and quarter point is about
L − QPE
= 0.05 ~ 0.10 (5.25)
a
• where a is distance between crack
tip and quarter point.
• Also, since larger number of
elements in hoop direction will
make those elements excessively
distorted, number of elements in
circumferential direction should be
6~8.

210
• Transition elements are expected to be larger than QPE and it is known
that the dimension should be in the order of
L − Tra
~ 2.5 (5.26)
L − QPE
• Fig. depicts the example of the effect of the elements size.
• In this graph, L‐QPE/a appears on the x‐axis and the error SIF (Stress
Intensity Factor) appears on the y‐axis.
• Regardless of method to find the SIF, a small size (L-QPE/a is less than
0.05) QPE, the error in SIF increases a lot.

211
 TRADITIONAL METHODS IN COMPUTATIONAL LEFM

• There are two approaches:

➢ Point Matching Method

• In this approach, the stress intensity factor (K) is computed from the
stress or displacement fields in the body

➢ Energy Method
• The method compute the energy release rate (G) in the body and
relate G to stress intensity factor (K)

212
➢ Stress Based Point Matching Method
– Consider a cracked body subjected to pure Mode I loading, the stress
component in the y-direction is given by
KI   3 
y = cos 1 + sin sin  (5.27)
(2r )1/ 2 2 2 2

– On the y-plane ( = 0), we have

r1 r2 r3 r4 x
KI
y = (5.28)
(2r )1/ 2
– Rearranging above equation we get the SIF as

KI =  y 2r (5.29)

213
– At various points, r1, r2, r3, etc., measure stresses y1, y2, y3, etc., and
thus stress intensity factor KI(r1), KI(r2), KI(r3), etc., can be computed.
i.e.,
KI (ri ) =  yi 2ri for i = 1, 2, 3,  (5.30)

y
y1 y2 y3 y4

r1 r2 r3 r4 x y

r1 r2 r3 r4 x

– Plotting the values KI(r1), KI(r2), KI(r3), etc., against nondimension (r/a)
and extend the best fit line to meet the abscissa to get required value of
SIF, KI for given crack configuration and loading.

KI 
   

r
214
a
➢ Displacement Based Point Matching Method
– Alternatively, KI can be estimated from a similar extrapolation of crack
opening displacement
– Expression for displacement in the y-direction for mode I loading and
plane stress case is given by
KI r  2 
v (r ) = sin  − cos2  (5.31)
 2 2 1 +  2

– Rearranging for SIF, we get

2
KI =  v (r ) f ( ) (5.32)
r y
r
– At some given angle  = 1 1
x
2
K I (r ) =  v (r ) f (1 ) (5.33)
r

215
• At various points, r1, r2, r3, etc., measure displacements v1, v2, v3, etc., and
thus stress intensity factor KI(r1), KI(r2), KI(r3), etc., can be computed. i.e.,
2
K I (ri ) =  v i (ri ) f (1 ) for i = 1, 2, 3,  (5.34)
ri
v4
v3
y v2
v1 r
r4
r3 y
r2  1 r
r1 x
1
x

• Plotting the values KI(r1), KI(r2), KI(r3), etc., against nondimension (r/a) and
extend the best fit line to meet the abscissa to get required value of SIF, KI
for given crack configuration and loading.
KI  
   

r
a

• Eq. (5.33) tends to give more accurate estimates of KI than eq.(5.29) because nodal
displacements can be inferred with a higher degree of precision than stresses.216
➢ Energy Method (Elemental crack advance)
− Energy release rate can be inferred from the rate of change in global
potential energy with crack growth
− If two separate numerical analyses of a given geometry are performed,
one with crack length a and the other with crack length a +da, the
energy release rate is given by
d (a + da) − (a)
GI = − = (5.35)
da a
− SIF can evaluated from following relations
EGI
KI = EGI for plane stress; KI = for plane strain; (5.36)
1− 2

 

1 2
1 2

a a + a 217
− This technique requires minimal post-processing, since total strain
energy is output by many commercial analysis codes

− This technique is also more efficient than the point matching methods
since global energy estimates do not require refined meshes

− Numerical differentiation in eq. (5.35) can result in significant errors

unless the crack length intervals (da) are small.

218
• MESH DESIGN
– Figure shows a half-symmetrical model of a simple cracked body, in
which a spider-web mesh transitions to coarse rectangular elements

Detail of crack tip fine mesh, showing displaced

shape and positions of nodes, moved to element
quarter points.

Half-symmetrical model of a cracked body

219
(b)
– Figure shows a finite number of finite elements of finite size,
interconnected only at their nodal points

Coarse Fine

220
 PART B
FATIGUE ANALYSIS

221
 FATIGUE
• Term fatigue of materials and structural components means damage and fracture
due to cyclic, repeatedly applied stresses
y T
P(t) C
y
(iv) Hogging
x (i) P(t)

y Neutral axis x (iii)

P(t)
(ii) Sagging
C y P(t)
T
T- Tensile region
C-Compressive region (ii) Sagging
C
(a) First-quarter of cycle T

Beam subjected to fatigue load. P(t) (b) Second- and third-quarter of cycle

P(t)
(ii)
+ y T
C
(iv) Hogging
Stress, 

0 (i) (iii) (v) y

Cycle
or time x (v)

-
(iv)
Bending axial stress in material below neutral axis. (c) Fourth-quarter of cycle
• According to ASTM, the definitions of fatigue:
– The process of progressive localized permanent structural change occurring
in a material subjected to conditions that produce fluctuating stresses and
strains at some point or points and that may culminate in cracks or complete
fracture after a sufficient number of fluctuations

• Six key words:

– Progressive
• Implies that the fatigue process occurs over a period of time or usage
• A fatigue failure is often very sudden, with no obvious warning
• Mechanisms involved may have been operating since the beginning of
the component’s or structure’s usage

– Localized
• Implies that the fatigue process operates at local areas rather than
throughout the entire component or structure
• These local areas can have high stresses and strains due to external
load transfer, abrupt changes in geometry, temperature differentials,
residual stresses and material imperfections
– Permanent
• Implies that once there is a structural change due to fatigue, the
process is irreversible
– Fluctuating
• Implies that the process of fatigue involves stresses and strains
that are cyclic in nature and requires more than just a sustained
load
• Magnitude and amplitude of the fluctuating stresses and strains
must exceed certain material limits for the fatigue process to
become critical
– Cracks
• Ultimate cause of all fatigue failures is that a crack has grown to a
point at which the remaining material can no longer tolerate the
stresses or strains, and sudden fracture occurs
• Crack had to grow to this size because of the cyclic loading
– Fracture
• Indicates that the last stage of the fatigue process is separation of
a component or structure in to two or more pieces or parts
• Fatigue Damage Mechanisms:
– It has been observed that the fatigue process involves the
following stages
i. Crack nucleation
ii. Short crack growth (Stage I)
iii. Long crack growth (Stage II)
iv. Fracture

Crack

Stage
Stage II
I

Plate under load,  

Enlarged view 225
− Crack Nucleation Stage:
▪ In metal alloys, when there are no macroscopic or microscopic
discontinuities, crack starts with dislocation movement on the localized shear
plane at or near high stress concentrations such as persistent of slip bands,
inclusion, porosity or discontinuity
▪ The localized shear plane usually occurs at the surface or within grain
boundaries
▪ This step, crack nucleation, is the first step in the fatigue process
▪ Once nucleation occurs and cyclic loading continues, the crack tends to grow
along the plane of maximum shear stress and through the grain boundary

 

(a) (b) (c) (d)

226
 
Fatigue mechanism in ductile materials
– Crack Growth Stage
▪ This stage is divided between the growth of Stage I and Stage II cracks
▪ Stage I
▪ Crack nucleation and growth are usually considered to be the initial short
crack propagation across a finite length of the order of a couple of grains on
the local maximum shear stress plane
▪ Crack tip plasticity is greatly affected by the slip characteristics, grain size,
orientation, and stress level
▪ Because the crack size is comparable to the material microstructure

▪ Stage II

▪ Crack growth refers to long crack
propagation normal to the principal
tensile stress plane globally and in the
maximum shear stress direction locally
▪ Characteristics of the long crack are
less affected by the properties of the
microstructure than the Stage I crack Crack
▪ It is because the crack tip plastic zone Stage
Stage II
I
for Stage II crack is much larger than
the material microstructure
– Fracture
▪ Last stage of the fatigue process is
separation of a component or structure in to 227
two or more pieces or parts
• Fatigue Life
– In engineering applications, for the plane strain case, the amount of
component life spent on crack nucleation and short crack growth or for
the case of plane stress condition, the amount of component life spent
only on crack nucleation is usually called the crack initiation period, Ni
– Whereas, the component life spent during only long crack growth in
case of plane strain or life spent both on short and long crack growth
in case of plane stress, is called the crack propagation period, Np
– The total fatigue life, NT is the sum of the initiation and propagation
live:
NT = Ni + Np

– There is no simple or clear delineation of the boundary between

fatigue crack initiation and propagation
– Further, a pre-existing crack in a structural component can reduce or
eliminate the fatigue-crack initiation life and thus decrease the total life
of the component

228
• Factors Affecting Fatigue Performance
– Many parameters affect the fatigue performance of structural
components
▪ Stress (load) parameters
* State of stress
* Stress range
* Stress ratio
* Constant or variable loading
* Frequency
* Maximum stress
▪ Geometry and Properties of component
* Stress (strain) raisers
* Size
* Stress gradient
* Metallurgical and mechanical properties of base metal and
weldments
▪ External environment
* Temperature 229
* Aggressiveness of the environment
– Primary factor that affects the fatigue behaviour of structural
components is the fluctuation in the localized stress or strain

– Consequently, the most effective methods for increasing the fatigue

life significantly are usually accomplished by decreasing the severity
of the stress concentration and the magnitude of the applied normal
stress

– In many cases, decrease in the severity of the stress concentration

can be accomplished by using transition radii fillet regions as for
keyways and geometrical changes and by minimizing the size of weld
imperfections

– Corrosion-fatigue behaviour is affected by the same parameters that

affect the fatigue behaviour as well as factors that do not affect
fatigue. Example, Frequency and waveform and effects of the
environment

230
 Recognition of Fatigue Failure
• Two fatigue zones are evident when investigating a fracture surface due to fatigue,
the fatigue zone and the rupture zone
– Fatigue zone is the area of the crack propagation
– Area of final failure is called the rupture or instantaneous zone
• In investigation of a failed specimen, the rupture zone yields the ductility of the
material, the type of loading, and the direction of loading.
• Relative size of the rupture zone compared with the fatigue zone relates the degree
of overstress applied to the structure
• Fatigue zone can be described as follows
– A smooth rubbed, and velvety appearance, the presence of waves known as
"clam-shells" or "oyster-shells", "stop marks" and "beach marks," and the
herringbone pattern or granular trace which shows the origin of the crack
– In general, stop marks indicate the variations in the rate of crack propagation
due to variations in stress amplitude in a cyclic application varying with time.
Rupture zone
Fatigue zone

Origin of
fatigue crack

Herringbone pattern
or Granular face
Stop marks

Schematic representation of the fatigue zone

• Amount of overstressing can be determined from the fatigue zone as follows
– highly overstressed if the area of the fatigue zone is very small compared with the area
of the rupture zone
– medium overstress if the size or area of both zones are nearly equal
– low overstress if the area of rupture zone is very small

232
Figure describes relations between the fatigue and rupture zones
 Fatigue failure of automobile components
Lorry spring failure: Springs suffer a combination of bending
and torsion and fatigue cracks propagate at right angles to the
principal stresses.
Thermal fatigue: Thermal fatigue damage on cast iron clutch plate
caused by rapid heating and cooling of disc by friction material.
Steel bolt: This high tensile steel bolt failed under
low stress high cycle conditions with a fatigue
crack running from 9 o'clock as shown by the
beach marks.
Stub axle: This is the classic reverse bending
fatigue of a steel stub axle from a road vehicle.
Notice cracks have grown from 8 o'clock upwards
and to a lesser extent from 2 o'clock downwards.
The rough central region is the final 233 ductile
rupture.
 Fatigue Loading

• Structural components are subjected to a variety of load (stress)

histories

• Two types of loadings:

a) Constant Amplitude Loading:
b) Variable Amplitude Loading:

234
a) Constant Amplitude Loading:
– Simplest of these histories is the constant-amplitude cyclic-stress
fluctuation shown in Figure (i)
– This type of loading usually occurs in machinery parts such as shafts
and rods during the periods of steady-state rotation
– Constant amplitude load histories can be represented by constant load
(stress) range, P (or S); a mean stress Smean; an alternating stress or
stress amplitude, Sa; and a stress ration, R
– TERMINOLOGIES:
➢ Stress range, is the algebraic difference between successive valley
and peak stresses (positive range) or between successive peak and
valley stresses (negative range).
S = Smax − Smin
➢ Mean stress is the average of
Smax
maximum and minimum
stresses. Stress, S
S + Smin
Sm = max Sm
2 Cycle
or time
➢ Alternating stress is half of the S
stress range S Smin
Sa =
2
➢ Stress ratio is the ratio of minimum stress to the maximum stress.
Smin
R=
Smax
➢ Reversal is the derivative of the stress-time history changes sign Point A
or B).
➢ Peak is the derivative of the stress-time history changes from positive to
negative sign (Point B).
➢ Valley (or trough) is the derivative of the stress-time history changes
from negative to positive sign (Point A).
➢ Mean crossing or zero crossing, is the number of times that the stress-
time history crosses the mean load level during a given duration of the
history. Normally only crossings with positive slopes are counted.

• Stresses in above relations can be Reversal

replaced with load, moment, torque, Smax
strain, deflection or stress intensity B Mean crossing
factors.
Stress, S

Sm
Cycle
or time
S
A C
Smin
1 cycle
Reversal
 Smax

Applied stress
Sm=0
Time or
Smin cycle

(a) R= -1, Complete or fully reversal of load

Smax

Applied stress
Applied stress

Smax
Sm
Sm
Smin
Time or Time or
Smin=0 0
cycle cycle
(b) R= 0, Zero or full tensile load (c) 0  R  1, Tensile to tensile load

Time or
Applied stress

Smax=0 0 Time or

Applied stress
cycle Smax
cycle
Sm
Sm
Smin
Smin
(d) 1/R= 0, zero or full compression load
(e) 0  1/R  1, compression to compression load

Comparison of R-ratio for various loadings

 Smax

Applied stress
Sm=0
Time or
Smin cycle

(a) R= -1, Complete or fully reversal of load

Shaft

Oil engine

Flywheel

W
Schematic of oil engine with flywheel
Oil Engine with overhung flywheel: Crank shaft is
subjected to fatigue load due to self weight of flywheel

Bus axles and chassis under fatigue load

Field examples of fatigue loading
 Eccentric crank
Applied stress

Smax
Cantilever
beam
Sm

Time or Zero stress position

Maximum stress position
Smin=0
cycle
Bending cantilever beam: Material below neutral
(b) R= 0, Zero or full tensile load axis is subjected to full tensile stress

Fulcrum

Roller
Roller

Spring
cam cam
Zero stress position Maximum stress position

Tension Springs in dynamic ignition timer

Translational spring subjected to full tensile stress

Field examples of fatigue loading

 Smax
Applied stress

Sm

Smin
Time or
0 cycle
(c) 0  R  1, Tensile to tensile load

P- Pre-tension
P- Pre-tension W- Normal load
- Rim deflection (P=W)

P P+W

P P P+W/2 P+W/2
W

P P-W

Pre-stressed spokes and rim Wheel under normal load Deformed state of rim

Bicycle spokes and rim under fatigue stress

Field examples of fatigue loading

 Pressed and
Smax Pressed load normal load
Applied stress

Sm Pre-
Tensile
Stressed
wall Compressive Tensile
Smin
Time or
0 cycle Bicycle tyre Sectional view of tyre: Tensile stress
(c) 0  R  1, Tensile to tensile load on side faces of tyre

Contact stress in landing wheels

Contact stress in automobile wheels

Belt drive: Fatigue stress in belt

Field examples of fatigue loading
 Time or

Applied stress
Smax=0
cycle

Sm

Smin

(d) 1/R= 0, zero or full compression load

Contact stress in gears teeth Contact stress in locomotive wheel

Gear train: Gear teeth

Field examples of fatigue loading

 Applied stress 0 Time or
Smax
cycle

Sm

Smin

(e) 0  1/R  1, compression to compression load

Locomotive suspension system: compression springs

IC engine: Springs, cam and follower Motor bike: Shock absorbers

Field examples of fatigue loading

 Heel strike Heel on ground Heel off

Side face of heel under fatigue loading

Applied stress

Smax

Heel cracks
Sm

Time or
Smin=0
cycle
(a) R= 0, Zero or full tensile load on side face of heel
Heel cracks under fatigue loading

Flatfoot
Heel strike

Time or
Applied stress

Smax=0
cycle
Toe off

Ground reaction, N
800
Sm 600
400
Smin 200
0
(b) 1/R= 0, zero or full compression load on foot 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Time in second
http://www.barefootrunning.fas.harvard.edu/4BiomechanicsofFootStrike.html
Ground reaction on foot
Examples of fatigue loading on human body
 Smax
Leg is in swing period
(Self weight of leg)
Applied stress

0 Time or cycle
Sm Leg is in heel strike to toe off period
(Body weight)

Smin

(c) R < -1,Partial reversal of load on left leg.

Walking stages

Smax=0
Applied stress

Sm

Smin
0 Time or cycle

(d) 0  R  1, Tensile to tensile load on right-hand

Bowling stages

Smax
Applied stress

Sm

Smin=0 Time or cycle

During freestyle swimming hands are under bending
fatigue loading
(e) R= 0, Zero or full tensile load on hand
Examples of fatigue loading on human body
b) Variable Amplitude Loading:
− Most complex fluctuating-load history is a variable-amplitude random
sequence as shown in Figure (ii)
− This type of loading is experience by many structures, including
offshore drilling rigs, ships, aircraft, bridges and earth moving
equipments

Applied stress 1 Cycle

Time or cycle
Figure (ii)
 – In which the probability of the same sequence and magnitude of stress
ranges recurring during a particular time interval is very small
– Such histories lack a describable pattern and can not be represented
by an analytical function
– Between the extremes of constant amplitude cyclic stress histories and
variable amplitude random-sequence stress histories, there is a
multitude of stress pattern of varying degrees of complexity
– Many of these histories can be described by analytical functions and
represented by various parameters
– Some simple variable amplitude stress histories are these
corresponding to a single cycle or multi-high tensile load cycles
superimposed upon constant amplitude cyclic-load fluctuations

Stress
Stress
Stress

0 0 0
Time or cycle
(a) Single high cycle (b) Multiple high cycle (c) Variable multiple high cycle
 Fatigue load spectra
Water wave

Crest Crest
Trough

Compression Water level

Weight
Bouncy Bouncy
Tension

(b) Moment-line (a) Sagging

(a) Ground loads

(c) Stress in lowering wing skin during flight and landing

Fig. (I) Landing loads on aircraft.
Visit : https://www.youtube.com/watch?v=KgIhWqvNGRc Crest
Trough Trough

Water level
Tension

Compression

(b) Hogging

Fig. (III) Longitudinal bending stresses on ship.

Fig. (II) Load due to water wave on offshore platform Visit : https://www.youtube.com/watch?v=K6AfSckWSZA
248
Visit: https://www.youtube.com/watch?v=Ar79VQSa6xQ https://www.youtube.com/watch?v=aBM7NgMhg90#t=28.416
 Fatigue Testing Machines
• Systematic, constant amplitude fatigue testing was first initiated by Wahler
on railway axles in the 1850s.
• Commonly used constant amplitude fatigue test machines are as follows.
(a) Cantilever rotating bending machine:
➢ Schematic of machine is as shown in Fig. (i).
➢ Machine produces a non-uniform bending moment along the
specimen’s length.
(b) Rotating pure bending machine:
➢ Test machine as shown in Fig. (ii), produces a uniform, pure bending
moment over the entire test length of the specimen.
− Both of test machines are known as ‘constant load amplitude machines’,
because, despite changes in material properties or crack growth, the
load amplitude do no change.
Coupling Main bearing Coupling Main bearing Load bearings Main bearing

Test specimen

Motor Motor

Test specimen
Load bearing

W
(i) Cantilever rotating machine W
(ii) Rotating pure bending machine
(c) Bending cantilever eccentric crank machine:
➢ Test machine as shown in Fig. (iii) produces a non-uniform bending
moment along the specimen’s length.
➢ Since the rotating eccentric crank produces constant deflection
amplitude, the load amplitude changes with specimen cyclic hardening
or softening and decreases as cracks in the specimen nucleate and
grow.
➢ For a given initial stress amplitude, constant deflection test machines
may give longer fatigue life than load-control machines because of the
decrease in load amplitude.
➢ Eccentric crank test machines, however, do have an advantage over the
rotating bending test machines in that the mean deflection, and hence
the initial mean stress, can be varied.

Test specimen

(iii) Bending cantilever eccentric crank.

(d) Axial loaded test machine:
➢ Fig. (iv) shows a schematic of an axial loaded fatigue test machine
capable of applying both mean and alternating axial loads in tension
and/or compression.
(d) Combined in-phase torsion and bending test machine:
➢ A common test setup for combined in-phase torsion and bending with or
without mean stress loading is shown in Fig.(v).
➢ Test machine provides a uniform torque and a nonuniform bending
moment along the specimen’s length.

Actuator
Test specimen

Grip Grip

(iv) Axial loading machine (v) Combine torsional and bending loadings
• Many closed-loop servohydraulic test machines have been designed over
the years and which utilizes personal computer shown in Fig.(vi).
• Principle of operation includes generating an input signal of load, strain, or
displacement using a function generator and applying this input through a
hydraulic actuator; measuring the specimen response via a load cell, a clip
gage, or a linear variable differential transducer (LVDT); and comparing this
with the specific input.
• Control and test data outputs are usually through a personal computer and
commercial or in-house software.
• Test frequency can range from mHz to kHz for constant or variable
amplitude load, strain, displacement, or stress intensity factor.

(iv) Servohydraulic test machine.

 Fatigue Testing Specimens

• Common test specimens for obtaining fatigue data (e.g., total fatigue life, that
includes crack nucleation life and crack growth life) are shown in Fig.
• These specimens usually have finely polished surfaces to minimize surface
roughness effects.
• In all cases a thin slit, notch, or groove with a very small root radius is
machined into the specimen.
• A small pretest fatigue crack is then formed at this root radius by cycling at a
low stress intensity factor range.
• After this sharp pretest fatigue crack has been formed, the real fatigue test
can begin.

Fig. Fatigue test specimens. (a) Rotating bending. (b) Axial uniform. (c) Axial hourglass. (d) Axial
or bending with circumferential groove. (e) Cantilever flat sheet/plate. (f) Tubular combined
axial/torsion with or without internal/external pressure. (g) Axial cracked sheet/plate. (h) part-
through crack. (i) Compact tension. (j) Three-point bend.
 Fatigue Life Behaviour
• Fatigue damage is strongly associated with the number applied stress cycle
and the fatigue life
• Since the mid-1800s, there are four fatigue life models exist for design
engineer:
▪ The nominal stress-life (S-N) model ( Formulated in 1850s by Wahler)
− Uses the nominal stresses and relates these to local fatigue strengths for
notched and unnotched components
▪ The local strain-life (-N) model (Formulated in 1960s)
− Deals directly with local strain at a notch and this is related to smooth
specimen strain-controlled fatigue behaviour
▪ The fatigue crack growth (da/dN - K) model (Formulated in 1960s)
− Requires the use of fracture mechanics and integration of the fatigue crack
growth rate equation to obtain the number of cycles required to grow a crack
from a given length to another length and/or to fracture
− This model can be considered a total fatigue life model when it is used in
conjunction with information on the existing initial crack size following
manufacture
▪ The Two-stage model
− Which consists of combining models above second and third models to
incorporate both microscopic fatigue crack formation (nucleation) and fatigue
crack growth
• Cyclic stresses or strains are the governing parameter for fatigue failure
• In general, fatigue is classified as either low-cycle or high-cycle
▪ Low-cycle fatigue
➢ Is more closely associated with the modelling and analysis of
localized plastic deformation, which is essentially the cause of
fatigue crack initiations such, low-cycle fatigue behaviour is
typically determined from strain-based fatigue testing

➢ This method is useful because it provides a basis to relate

component fatigue behaviour to localized regions of plastic
deformation at notches or other design details

▪ High-cycle fatigue
➢ High-cycle fatigue of steels is often expressed by a fatigue limit,
which is the value of stress below which the steel can presumably
tolerate an infinite number of cycles

➢ If a fatigue limit is observed in steel for below about 1100MPa

tensile strengths, it typically occurs above 106 cycles between the
range of 107 and 108 cycles
❖ The fatigue life behaviour is further categorized in to uniaxial or
monotonic cycle fatigue and multiaxial cycle fatigue

▪ In the uniaxial high or low cycle fatigue, the component is under the
unidirectional cyclic loading (stress or strain)

▪ In the case of multiaxial high or low cycle fatigue, the component is

under the multidirectional cyclic loading
 HIGH-CYCLE FATIGUE
Hand Biceps Curl
• Term "biceps curl" may refer to any of a number of weight training
exercises that target the biceps brachii muscle.

Dumbbells weight
25
20

15
10
5

1 50 100 150 200

Cycles to strain
Hand biceps weight-cycles to
strain curve
 Body-Building and Yoga
Body-building: Then

• Advantage of muscle building is stiffening of the muscles by

accumulating stress/strain in the muscles through fatigue loads.
• In young age the body is maintained in good muscular shape due
to stiffening effect however, in old age body shape changes to a
flabby or flexible shape. Now
• Muscle builders in general suffer from health issues in old age.
Yoga:
• Yoga is ascetic discipline, a part of which, including breath
control, simple meditation, and the adoption of specific bodily Body-builder and Actor
postures, is widely practiced for good health and relaxation or Arnold Schwarzenegger
relief from stress.
• Practicing of yoga regularly, a total relief from the stress or strain
in the body and mental stability can be achieved.

Attention to yoga's holistic benefits,

the United Nations Declared June 21
as “International Day of Yoga”.
‘To Keep Good Health - Practice Yoga’
‘Over-stressed fatigue life reduces age At the age of 50yrs At the age of 93 yrs
Free-stressed fatigue life increases age’ Yoga guru BKS Iyengar (1918-2014)
 HIGH-CYCLE FATIGUE
• Stress-Life (S-N) Curve developed by Wahler, Railway Engineering
− Typical schematic S-N curve obtained under axial load or stress
controlled test conditions with smooth specimens.
Original specimen
Sa Failure
Elongation state
Lo
SaA=Sut
Nf =¼  1 do
Sa dt Sa
0 N
Lt

Specimen-A dc Sa
SaB
Nf =2 Failure
Lc
SaA=Sut Contrast state
0 N
SaB
SaC
Specimen-B SaD
SaC Failure SaE
Nf =5
Sa S
0 N aF

SaG
Specimen-C SaH

SaH Nf =24 Failure

1 5 10 15 20 25
0 N Nf
Specimen-H Typical S-N curve
▪ General S-N Behaviour:
− Two typical schematic S-N curves obtained under axial load or stress
controlled test conditions with smooth specimens are shown in figure
(i)
Log(S)

Log(S)
Sf

100 102 104 106 108 100 102 104 106 108
Nf Nf

(a) (b)
Figure (i) Typical S-N curves 260
– S is the applied normal stress, usually taken as alternating stress Sa
and Nf is the number of cycles or life to failure where failure is defined
as fracture
– Constant amplitude S-N curves of these types plotted on semi-log or
log-log coordinates and often contain fewer data points than shown
above figure
– S-N curves obtained under torsion or bending load-control test
conditions often don’t have data at the shorter fatigue lives (say, 103 or
104 cycles or less) due to significant plastic deformation: the torsion
and bending stress equations  = T r /J and  = M y / I can be used
only for nominal elastic behaviour
– Figure (ia) shows a continuously sloping curve, while figure (ib) shows
a discontinuity or ‘knee’ in the S-N curve
– Knee has been found in only a few materials between 106 and 107
cycles under noncorrosive conditions

261
– Definition of terms
• Nf – fatigue life is the number of cycles of stress or strain of a
specified character that a given specimen sustains before failure of a
specified nature occurs

• SNf – Fatigue strength is a hypothetical value of stress at failure for

exactly Nf cycles as determined from the S-N diagram

• Sf – Fatigue limit is the limiting value of the median fatigue strength

as Nf becomes very large

Log(S)
Sf

100 102 104 106 108

Nf

(b)
262
▪ Fatigue Limit under Fully Reversed Uniaxial Stressing
− Fatigue limit has historically been a prime consideration in long-life
fatigue design

− For a given material the fatigue limit has an enormous range,

depending on surface finish, specimen or component size, type of
loading, temperature, corrosive and other aggressive environments,
mean stresses, residual stresses and stress concentrations

− Considering the fatigue limit based on a nominal alternating stress, Sa,

this value has ranged from essentially 1 to 70% of the ultimate tensile
strength

− Most long-life S-N fatigue data available in the literature consists of fully
reversed (R= -1, or Sm =0) uniaxial fatigue strengths or fatigue limits of
small, highly polished, unnotched specimens based on 106 to 5108
cycles to failure in laboratory air environment

263
− Representative monotonic tensile properties and bending fatigue
limits of selected engineering alloys obtained under above
condition are given in the following table

264
▪ Mean Stress Effects on S-N Behaviour
− Alternating stress, Sa and the mean stress, Sm are defined in following figure
− Mean stress can have a substantial influence on fatigue behaviour
− This is shown in the following figure, where alternating stress is plotted against
the number of cycles to failure for different mean stresses
− It is seen that the tensile mean stresses are harmful and compressive mean
stresses are useful, this is shown by the three vertical lines indicating fatigue
life: Nft, Nf0 and Nfc representing the fatigue life for tensile, zero and compressive
mean stress for a given alternating stress

Sm = compression
Alternating stress, Sa

Sm = 0
Sm = tension

Nft Nf0 Nfc

100 102 104 106 108
Cycles to failure, Nf 265
Effect of mean stress on fatigue life
• Substantial investigation of tensile mean stress influence on long-life fatigue
strength has been made
• Typical dimensionless plots are shown in the following figures for steels and
aluminium alloys, where Sa/Sf versus Sm/Su is plotted, [Sf is the fully
reversed (Sm =0, R= -1), fatigue limit of smooth specimens, and Su is the
ultimate tensile strength]

1.0 1.0
Fully reversed fatigue limit, Sf

Fully reversed fatigue limit, Sf

Alternating stress, Sa

Alternating stress, Sa
0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
Mean tensile stress, Sm Mean tensile stress, Sm
Ultimate tensile strength, Su Ultimate tensile strength, Su
(a) Steels based on  107 cycles (b) Aluminium alloys on  5107 cycles

Effect of mean stress on alternating fatigue strength at long life 266

• Substantial scatter exists, but the general trend indicating that tensile mean
stresses are harmful is quite evident
• Small tensile mean stresses however often have only a small effect
• It appears that much of the data fall between the straight and curved lines
• The straight line is modified Goodman line and the curve is the Gerber
parabola

1.0 1.0
Fully reversed fatigue limit, Sf

Fully reversed fatigue limit, Sf

Alternating stress, Sa

Alternating stress, Sa
0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
Mean tensile stress, Sm Mean tensile stress, Sm
Ultimate tensile strength, Su Ultimate tensile strength, Su
(a) Steels based on  107 cycles (b) Aluminium alloys on  5107 cycles

Effect of mean stress on alternating fatigue strength at long life 267

• An additional popular relationship has been formulated by replacing Su with
f known as Morrow line, where f is the true fracture strength
• Following equations represent these tensile mean stress effects for uniaxial
state of stress: 2
Sa Sm Sa  Sm 
Modified Goodman : + = 1; Gerber : +   = 1;
Sf Su Sf  Su 
Sa Sm
Morrow : + = 1.
Sf  f

1.0 1.0
Fully reversed fatigue limit, Sf

Fully reversed fatigue limit, Sf

Alternating stress, Sa

Alternating stress, Sa
0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
Mean tensile stress, Sm Mean tensile stress, Sm
Ultimate tensile strength, Su Ultimate tensile strength, Su
(a) Steels based on  107 cycles (b) Aluminium alloys on  5107 cycles

Effect of mean stress on alternating fatigue strength at long life 268

• All three expressions have been used in fatigue design when modified for
notches, size, surface finish environmental effects and finite life.
• A yield criterion has also been used in conjunction with these expressions
• Above figures does not provide information on compressive mean stress
effects as like shown in the following figure for several steels and
aluminium alloys
Reversed fatigue strength, (Sm=0)

Aluminium alloys
Axial stress amplitude, Sa

Steels

1.5

1.0

0.5

0 -1.5 -1.0 -0.5 0 0.5 1.0

Compression Tension
Mean stress, Sm
Yield strength, Sy
Compressive and tensile mean stress effect 269
• It is seen that these compressive mean stresses cause increases of up
to 50% in the alternating fatigue strength
• Based on the fact that compressive mean stresses at long lives are
useful, the modified Goodman or Morrow equations can be
conveniently extrapolated to the compressive mean stress region
• The Gerber equation incorrectly predicts a harmful effect of
compressive mean stresses and does not properly represent notched
component tensile mean stress fatigue behaviour
Reversed fatigue strength, (Sm=0)

Aluminium alloys
Axial stress amplitude, Sa

Steels

1.5

1.0

0.5

0 -1.5 -1.0 -0.5 0 0.5 1.0

Compression Tension
Mean stress, Sm
Yield strength, Sy
Compressive and tensile mean stress effect 270
• Thus, we deemphasize the use of the Gerber equation for design. The
Modified Goodman and Morrow equations are shown in following figure
for a given long life (e.g., 106, 107, and 108) along with the criterion for
uniaxial yielding:
Sa Sm
+ =1
Sy Sy
where Sy is the monotonic tensile yield strength and S‘y is the cyclic yield
strength

Sa
S’y

Yielding

Sf
Modified Goodman line

Morrow line

-sy 0 Sy Su f
-Sm +Sm 271
Fatigue and yielding criteria for constant life of unnotched parts
• In fatigue design with constant amplitude loading and unnotched parts, if
the coordinate of the applied alternating and mean stresses fall within the
modified Goodman or Morrow lines shown in above figure, then fatigue
failure should not occur prior to the given life
• The difference between the modified Goodman and Morrow equations is
often rather small thus, either model often provides similar results
• If yielding is not to occur, then the applied alternating and mean stresses
must fall within the two yield lines connecting Sy to S'y
• If both fatigue failure and yielding are not to occur, then neither criterion,
as indicated by the bold lines should be exceeded
Sa
S’y

Yielding

Sf
Modified Goodman line

Morrow line

-sy 0 Sy Su f
-Sm +Sm 272
Fatigue and yielding criteria for constant life of unnotched parts
▪ S-N Curve Representation and Approximations:
• Actual fatigue data from either specimens or parts should be used in design, if
possible
• This information may be available through data handbooks, design codes,
industry standards, company data files, and previous tests
• Often this information is not available and must be generated or
approximations of S-N behaviour must be made
• There are many models depicting S-N curves and these usually imply a
median fatigue life
• Following figure shows common reasonable S-N median fatigue life curves
based upon straight-line log-log approximation

A, Sa or f
Log(Sa) or Log(SNf)

Sf

100 102 104 106 108

Cycles to failure, Nf 273
Basquin type S-N curve
• Basquin in 1910s suggested a log-log straight line S-N relationship
such that
SNf = A(Nf )
B
Sa or

where Sa is an applied alternating stress and SNf is the fully reversed;

R =-1, fatigue strength at Nf cycles, A is the coefficient and represents
the value of Sa or SNf at one cycle, and B is the exponent or slope of
the log-log S-N curve

A, Sa or f
Log(Sa) or Log(SNf)

Sf

100 102 104 106 108

Cycles to failure, Nf 274
Basquin type S-N curve
• One approximate representation of the median S-N curve is a tri-slope
model with one slope between 1 cycle and 103 cycles, and one slope
between 103 and 106 or 108 cycles, and another slope after 106 or 108 cycles
(This model is represented by the dashed lines in following figure.)
• Tri-slope model indicates that a fatigue limit does not exist, which may be
the case for in-service variable amplitude loading, however, in some design
codes such as those for gears and welds, it can exist.
• In the tri-slope model, the third, or long-life, slope could also be horizontal
after 106 or 108 cycles.
• The intercept, A, at Nf=1 could be chosen as the ultimate tensile strength,
Su, the true fracture strength, f, an intercept found from regression of the
fatigue data

A, Sa or f
Log(Sa) or Log(SNf)

Sf

100 102 104 106 108

Cycles to failure, Nf 275
Basquin type S-N curve
• Constant fatigue life diagrams relating Sa and Sm are often modelled
as shown schematically in the following figure, by using one of the S-N
models from the above figure and the modified Goodman or Morrow
equations for mean stress
• In the above figure, the intercepts SNf at Sm=0 for a given life are found
from the fully reversed, R= -1, modelled S-N curves.
• Modified Goodman straight lines are shown passing through these
intercepts and the ultimate tensile strength, Su in the figure.

Sa
S’y
103
SNf 104
105
106 - 108

Sf

-sy 0 Sy Su
-Sm +Sm 276
Constant life diagrams with superimposed yield criterions
• For these finite fatigue lives, the modified Goodman or Morrow
equations should have Sf replaced with SNf for a given life.
• Using these equations and Basquin’s equation for the fully reversed,
R = -1, finite life region, gives the following equations:
SNf = A(Nf )
B

and
Sa Sm Sa Sm
+ = 1 or + =1
SNf Su SNf  f

Sa
S’y
103
SNf 104
105
106 - 108

Sf

-sy 0 Sy Su
-Sm +Sm 277
Constant life diagrams with superimposed yield criterions
• Above equations provides information to determine estimates of
allowable Sa and Sm for a given fatigue life of unnotched parts
• When above equations are used, the S-N curve for a given mean
stress is parallel to the fully reversed, R= -1, i.e., the slope, B, remains
unchanged.
• Thus the mean stress effect is handled in the same manner at both
short and long lives
• This may not represent the actual behaviour in situations where mean
stress relaxation occurs at short lives due to plastic deformation, such
as when the mean stress represents a residual stress
• For such cases, it may be more realistic to apply the mean stress
correction to the fatigue strength at long life, i.e., the fatigue limit, and
connect it by a straight line to Su or f at one cycle
• This assumes a full effect of mean stress at long life, with gradually
decreasing effect, to no effect at 1 cycle.
• Following yield criterion is also superimposed in the above figure
Sa Sm
+ =1

Sy Sy
278
 LOW CYCLE FATIGUE

• Phenomenon of low-cycle fatigue can be in principle described in terms of

the theory of plasticity.
• Low-cycle fatigue is a cyclic elasto-plastic deformation occurring until the
expenditure of plasticity reserves.
• Material behavior at unloading and reversed loading (cyclic loading), in
particular the shape and size of hysteresis loops, is of essential significant
in low-cycle fatigue.
• Following figure depicts the Hysteresis loop generated as a result of a
complete cycle under strain controlled.

S Yield 1, 3, 5

+ 1 3 5 Plastic
Elastic


S

Time 

-
2 4 6 Elastic
Plastic
Strain controlled condition 2, 4,6
 = e +p 279
Hysteresis loop for one cycle
• Illustrates the cyclic stress-strain curve obtained for various strain level

Time
1 2 3

(1) (2) (3)

Strain controlled conditions

S (3)
(2)
(1)

1 2 3 

Illustration of the cyclic stress-strain curve and the method for obtaining it. 280
• Cycle-deformation relations depend on the type of loading process

• They change whether this process is load- or displacement- controlled

– Some materials reveal a tendency to cyclic hardening: under the
displacement controlled loading the maximal cycle stresses grow with
the cycle number
– Other materials reveal a tendency to cyclic softening

• An intermediate place is occupied by the plastically stabilized materials

• Depending on the micro-structural state and temperature, the same

material may behave in various ways

281
 • Typical diagrams of uniaxial tensile/compression deformation S() are
shown in the following figure
• Figure (i) shows the behaviour of a cycle-dependent hardening materials
and figure (ii) that of a softening material

Stress response
Stress response

+ +

S S
Time Time

1 3 5 - -
+

 5
Time
3
- 1 1
2 4 6 S S
3
5
Strain controlled condition

 

6
4
2 2
4
282
6 Figure (ii) Softening behavior
Figure (i) Hardening behavior
• The relationship between maximal stresses and strains within a cycle
generally differs from that in monotonic quasi-static loading

Hardening behavior

Softening behavior
Stress

Cyclic stress-strain of soft material

Cyclic stress-strain of hard material
Monotonic stress-strain

Strain
Comparison of monotonic and cyclic stress-strain curves
• For the low-cycle fatigue region (N<104 cycles) tests are conducted with
controlled cycles of elastic-plastic strain rather than with controlled load or
stress cycles

• When the strain level is high, the cycle number at fatigue failure is
comparatively small and significant one-sided residual deformations
accumulate in the specimen

• At a moderate strain level, test results are convenient to represent with

fatigue curves

• Compared with high-cycle fatigue, low-cycle fatigue curves are usually

plotted on the plane of characteristic strain versus cycle number at failure

• Standard test in tension or tension/compression are usually performed

maintaining a constant range  of the nominal strain 

284
• Under controlled strain testing, fatigue life behaviour is represented by a
log-log plot of the total strain range , versus the number of cycles to
failure Nf as shown in figure (i)
• At given number of cycle to failure, the total strain is the sum of the elastic
and plastic strains
• Both the elastic and plastic curves can be approximated as straight lines

Low-cycle fatigue
f (2Nf=1)
High-cycle fatigue

p
p
Strain amplitude, a (Log scale)


Total strain-life curve

S’f /E 
Elastic line
S’f (2Nf)b/E

Plastic line
f (2Nf)c

100 102 2NT 104 106 108 1010

Reversals to failure, 2Nf (Log scale) 285
Illustration of total strain-life curve
• At large strains or short lives the plastic strain component is
predominant and at small strains or longer lives the elastic strain
component is predominant
• The intercepts of the two straight lines at 2Nf = 1 are S'f/E for the elastic
component and 'f for the plastic component
• Let, the slopes of the elastic and plastic line are b and c, respectively.
This provides the following equation or strain-life data of small smooth
axial specimens:
Low-cycle fatigue
f (2Nf=1)
High-cycle fatigue

p
p
Strain amplitude, a (Log scale)


c
1 Total strain-life curve

S’f /E 
b
1 Elastic line
S’f (2Nf)b/E

Plastic line
f (2Nf)c

100 102 2NT 104 106 108 1010

Reversals to failure, 2Nf (Log scale) 286
Illustration of total strain-life curve
   e  p Sf
= a = + = (2Nf )b +  f (2Nf )c
2 2 2 E
where, /2 = a - Total strain amplitude,
e/2 – elastic strain amplitude = S/2E = Sa/E
p/2 – plastic strain amplitude = /2 - e/2
'f – Fatigue ductility coefficient, c – fatigue ductility coefficient,
S'f – Fatigue strength coefficient, d – fatigue strength coefficient,
E- Young’s modulus, and
S/2 – stress amplitude = Sa
Low-cycle fatigue
f (2Nf=1)
High-cycle fatigue

p
p
Strain amplitude, a (Log scale)


c
1 Total strain-life curve

S’f /E 
b
1 Elastic line
S’f (2Nf)b/E

Plastic line
f (2Nf)c

100 102 2NT 104 106 108 1010

Reversals to failure, 2Nf (Log scale) 287
Illustration of total strain-life curve
• The straight line elastic behaviour can be transformed to

S
= Sa = Sf (2N )
b

2
which is equation is Basquin’s equation proposed in 1910s.

Low-cycle fatigue
f (2Nf=1)
High-cycle fatigue

p
p
Strain amplitude, a (Log scale)


c
1 Total strain-life curve

S’f /E 
b
1 Elastic line
S’f (2Nf)b/E

Plastic line
f (2Nf)c

100 102 2NT 104 106 108 1010

Reversals to failure, 2Nf (Log scale) 288
Illustration of total strain-life curve
• The relation between plastic strain and life is
 p
=  f (2N )
c

2
which is the Manson-Coffin relationship first proposed in the early 1960s.

− The exponent c ranges from about -0.5 to -0.7 with -0.6 as a

representative value
− The exponent b ranges from about -0.06 to -0.14, with -0.1 as a
representative value
Low-cycle fatigue
f (2Nf=1)
High-cycle fatigue

p
p
Strain amplitude, a (Log scale)


c
1 Total strain-life curve

S’f /E 
b
1 Elastic line
S’f (2Nf)b/E

Plastic line
f (2Nf)c

100 102 2NT 104 106 108 1010

Reversals to failure, 2Nf (Log scale) 289
Illustration of total strain-life curve
− The term 'f is related to the true fracture strain, f in a monotonic tensile
test and in most cases ranges from about 0.35 to 1.0 times f
− S'f is somewhat related to the true fracture stress, Sf in a monotonic tensile
test

Low-cycle fatigue
f (2Nf=1)
High-cycle fatigue

p
p
Strain amplitude, a (Log scale)


c
1 Total strain-life curve

S’f /E 
b
1 Elastic line
S’f (2Nf)b/E

Plastic line
f (2Nf)c

100 102 2NT 104 106 108 1010

Reversals to failure, 2Nf (Log scale) 290
Illustration of total strain-life curve
 FATIGUE GROWTH LAWS

a) Extended Service Life:

− Different inspection methods that use Visual Inspection, X-ray, Ultrasonic
waves and electric potential drop methods can be employed to detect
cracks
− Once cracks are detected, FM methods can be used to determine the
extended life of the structures or components
− Figure shows the concept by plotting the crack length a as a function of
the number of cycles, N
− In the figure ai represents the initial crack length within a structure, ad a
detectable crack length, and ac the critical crack length within the stress
intensity factor of the crack reaches the critical value.

Fatigue life
Crack size, a

af

Useful life

ad ac
ab
ai

Inspection interval

Ni No. of cycles, N Nb Nd Nc Nf
 – The numbers of cycles when a crack length reaches the critical length ac
is denoted as Nc
– The extended life Ncd equals Nc – Nd

b) Crack Growth Behaviour:

• LEFM has been successfully used to model the fatigue crack growth
behaviour
❑ Consider a cracked structure with a crack length a under cyclic stress
conditions, as shown in figure (i)
❑ The applied stress  as a function of time t is shown in figure (ii)
❑ The maximum and minimum stress intensity factors Kmax and Kmin are
linearly related to the maximum and minimum applied stresses, max and
min, respectively

Stress intensity factor, K

Kmax
a

Km
Kmin

No. of cycles, N`
(i) (ii)
❑ According to LEFM,

K max = max a f a
W
( )
K min = min a f ( a )
W

❑ Stress intensity factor range, K is defined as

K = Kmax − Kmin

❑ The load stress ratio R is the stress-life approach is defined as

 min
R=
 max
Kmin
or =
Kmax
 – Based on experimental results figure (i) shows schematically the crack growth
rate da/dN as a function of the stress intensity factor range K in a log-log scale
– In region I:
• when K decreases, the crack growth rate drops significantly
• The asymptote Kth is the threshold of the stress intensity factor range
below which no fatigue crack growth should occur
• The values of Kth are 5-16 MPa(m)1/2 for steels and 3-6 MPa(m)1/2
aluminium alloys 10 -3

Fracture
– In region III:
• when K is large, the crack growth 10-4

rate accelerates significantly

Fatigue Crack Growth Rate, da/dN

• This can happen when Kmax C
10-5
approaches KC which represents
the critical K for crack initiation for a
given thickness 10-6

Threshold
10-7

10-8
100 101 102
Stress Intensity Factor Range, K
– In region II:
• The crack grow rate da/dN can be approximately linear related to the
stress intensity factor range, K in the log-log scale plot
• Paris law for crack growth is therefore,
da = C(K )m
dN
• where C and m represent the material constants and are obtained by
adopting Least-square method to fit the fatigue data in the log-log scale
plot of da/dN as a function of K
• Thus,
Y = A+B X

( )
where Y = log da , X = log (K), A = log C and B = m
dN

Typical values of material constants C and m in the Paris law for R=0

Material C m
1. Martensitic steel 1.3610-7 2.25
2. Ferritic-pearlitic steel 6.8910-9 3
3. Austentic stel 5.6110-9 3.25
c) Life Estimation:
– Life estimation under constant load amplitude loading conditions usually
needs numerical integration
– In general K increase as a increases with  being kept as constant
– Crack growth rate da/dN can be expressed as 10 -3

Fracture
da = f ( K,R )
dN 10-4
– rearranging, we get
da

Fatigue Crack Growth Rate, da/dN

dN =
f ( K,R ) 10-5
C

– Integration of above eq., gives

Nf af 10-6

 dN =  f ( da

Threshold
Ni ai
K,R )
– Left hand side of the above equation can be 10-7

integrated as N f

Nif = Nf − Ni =  dN 10-8
100 101 102
Stress Intensity Factor Range, K
Ni
where Nif represents the number of cycles for crack propagation from ai to af
– Paris law gives a specific form for the crack growth rate as
= C (K )
da m

dN
– Consider the solution for K in the form as
a 
K =  a f  
W 
where f(a/W) is function of geometry and loading

– Crack growth rate now can be written as

m
da   a 
= C   f   a m / 2
dN   W 
– Number of cycles Nif for crack propagation from the crack length ai to af can
be integrated as
Nf af
dN
Nif =  dN = a da da
Ni i
af
1
=    a 
m
a −m / 2 da
C   f
ai

 W 
– Integrating explicitly above equation yields as

 m −1

 ai  2 
1 −  a  
  f  
Nif =

C   f a
W
( ) ( )
m
m
2
m −1
− 1 ai 2

– Above equation indicates that Nif is very sensitive to the selection of ai.
Because f(a/W) is a function of geometry, loading and crack length.
– A numerical integration is usually carried out for solution
 The bar shown in following figure is subjected to a repeated moment 0 ≤ M ≤ 1200
Nm. The bar is AISI 4430 steel with Sut = 185 kN/m2, Sy = 170 kN/m2, and KIc = 73
kN/m2 √m. Material tests on various specimens of this material with identical heat
treatment indicate worst-case constants of C = 3.8(10−11)(m/cycle)(kN/m2 √m)m and m
= 3.0. As shown, a small cut of size 0.004m has been discovered on the bottom of the
bar. Estimate the number of cycles of life remaining.
• The stress range σ is always computed by using the nominal (no-cracked) area.
Thus, from bending moment equation, we have,
I bh2 0.25(0.5)2
= = = 0.01042m3 0.25m
y 6 6
• Therefore, before the crack initiates, the stress range is 0.5m
M 1200
 = = = 115.2 kN / m 2 M
I 0.01042 Small cut
y
which is below the yield strength.
• As the crack grows, it will eventually become long enough such that the bar will
completely yield or undergo a brittle fracture. For the ratio of Sy/Sut it is highly unlikely
that the bar will reach complete yield. For brittle fracture, designate the crack length as
af. If  = 1, then with KI = KIc, we approximate af as
2
 KI =   a
1  K Ic 
2

 = 
1 73 
af =   = 0.1278m
   max    115.2 
• From following figure, compute the ratio af/h as

af 0.1278
= = 0.256
h 0.5

• Thus af/h varies from near zero to approximately 0.256. From above figure for this
range  is nearly constant at approximately 1.07. Assume it to be so, and re-evaluate
af as 2 2
1  K Ic  1 73 
af =   =   = 0.112m
   max    1.07(115.2) 
• Thus, the estimated remaining life is
1a f
da 1 0.112 da
Nfi =  = 
C a (  a )m
i
3.8(10−11) 0.004 (1.07  115.2 a )3
0.112
5.047(103 )
= = 64.7(103 ) cycles
a 0.004
 NOTCHES AND THEIR EFFECTS
1. NOTCHES
• Notch refers to any geometric contour that disrupts the stress flow through
the loaded machine part (Fig. 1).
• Notches of concern here are those that are deliberately introduced to obtain
engineering features such as fillet, groove, hole, an abrupt change in cross-
section or any disruption to the smooth contours of a part (Fig. 2).
• Any one of these changes in cross-sectional geometry will cause localized
stress concentrations.

Fig. 1 Stress flow

in plate.

Fig. 2 Notches in machine parts

• It is assumed that the engineer will follow good design practice and keep the
radii of these notches as large as possible and reduce the stress
concentrations.
• Notches concentrate stresses and strains.
• For static loading, it is measured by the elastic or geometrical or theoretical
stress concentration factor, Kt, defined as the ratio of the actual maximum
stress, max, or strain, max, at the notch or discontinuity to the nominal
stress, 0, or strain, 0. i,e.,
 max or K = max
Kt =
0
t
0
• For dynamic loading, notched fatigue stress-concentration factor also known
as fatigue notch factor, denoted by Kf can be applied.
• It is the ratio of smooth to net notched fatigue strengths based on the ratio of
alternating stresses.
Smooth fatigue strength Max. stress in notched specimen
Kf = or Kf =
Notched fatigue strength Stress in notch - free specimen
• The fatigue notch factor is usually estimated for zero mean stress and long life
(106-108 cycles).

• Difference between Kt and Kf is related to the stress gradient and localized

plastic deformation at the notch root.

• Reasoning for stress gradient influence is that the notch stress controlling the
fatigue life may not be maximum stress on the surface of the notch root, but
rather an average stress acting over a finite volume of the material at the
notch root.

• This average stress is lower than the maximum surface stress, calculated
from Kt.

• When small cracks nucleate at the notch root, they grow into regions of lower
stress due to the stress gradient.

• Localized plastic deformation and notch blunting effect due to yielding at the
notch root reduce the notch root stress amplitude, particularly at short lives.
2. NOTCH SENSITIVITY
• Materials have different sensitivity to stress concentrations, which is referred
to as the notch sensitivity of the material.

• In general, more ductile material have less notch sensitivity on the other hand,
brittle materials are more notch sensitive.

• Since, ductility and brittleness in metals are roughly related to strength and
hardness, low-strength, soft materials tend to be less notch sensitive than
high-strength, hard ones.

• Notch sensitivity also depends on the notch radius which is measure of notch
sharpness; as notch radii approaches to zero, the notch sensitivity of materials
decreases and also approaches to zero.
• Neuber made the first study of notch effects and published the equation for
fatigue notch factor in 1937.

• Subsequently, Perterson refined the approach and developed the concept of

notch sensitivity denoted by q defined as
Kf − 1
q=
Kt − 1
• A value of q = 0 (or Kf = 1) indicates no notch sensitivity, whereas a value of q
= 1 (or Kf = Kt) indicates full notch sensitivity.

• Procedure is to first determine the Kt for the particular geometry and loading,
then establish the notch sensitivity for the chosen material and then find the Kf
from the above equation.

• That is, the fatigue notch factor can be described in terms of the material
notch sensitivity as

K f = 1 + q (K t − 1)
• Neuber developed the following approximate formula for the notch factor for
stress ratio, R = -1 loading (i.e., mean stress is zero),
1 Kt − 1
q= or Kf = 1 +
1+  r 1+  r
where r is the radius at the notch root and  is the characteristic length
depends on the material.

• Peterson has observed that good approximations for R = -1 loading can also
be obtained by using the somewhat similar formula
1 Kt − 1
q= or Kf = 1 +
1+ a r 1+ a r
where a is another material characteristic length.
• Values for a are given by Peterson. An empirical relationship between
ultimate tensile strength Su, and a for steels is given as
1 .8
 2070 
a = 0.0254 
 Su 
where Su in MPa and a in mm.
 EXAMPLE
A rectangular, stepped bar with D=0.2m, d=0.18m and r=0.025m and Su
=700MPa., is to be loaded in bending. Determine the fatigue stress-
concentration factor for the given dimension.
• Data: D= 0.2m, d=0.18m and r=0.025m and Su =700MPa
• Theoretical stress concentration factor is computed from,
b
r 
K t = A 
d 
• Using D/d= 0.2/0.18=1.11, we have A=1.02 and b= -0.218
−0.218
 0.025 
K t = 1.02  = 1.56
 0 .18 
• Neuber factor is given as  =0.0096 for Su =700MPa.
• Notch sensitivity q of the material can be found by
1 1 0.08

q= = = 0.94

Neuber constant,  in m½
 0 . 0096 0.064

1+ 1+ 0.048
r 0.025
0.032

• Fatigue stress-concentration can be determined as 0.016

K f = 1 + q (K t − 1) = 1 + 0.94(1.56 − 1) = 1.526
275 550 825 1100 1375 1650
Ultimate tensile strength, Sut MPa

• As expected, we find for that always Fig. 3 Neuber's constants for

Kf  Kt steel and aluminum.
3. EFFECTS OF STRESS ON NOTCH FACTOR
• For a fatigue life of 106 to 108 cycles with R = -1, the notched fatigue strength
SfIKf can be estimated.
• Nominal stress calculated for these conditions is less than the yield strength.

• In monotonic testing, notches may increase or decrease the nominal

strength.
• A sharp groove in a notched tensile bar with ductile behavior produces a
higher nominal ultimate tensile strength than a uniform bar of the same
minimum diameter.
• Smooth bar necks down and finally fails with a much reduced area.
• Notched bar cannot neck down because of the un-yielded metal above and
below the notch as well as its ultimate tensile strength is greater.
• For a material behaving in a brittle manner, necking does not occur in the
smooth bar and greater notch sensitivity exists.
• Therefore, the ultimate tensile strength of the notched bar made of such a
material is generally lower than that of the smooth bar.
• Data on the behavior of notched parts and smooth parts at different constant
amplitude stress levels generally show that the notched fatigue strength is a
greater fraction of the smooth fatigue strength as the nominal stresses get
higher.
• Where data are available, the designer will of course use them, but in the
absence of data the behavior of notched parts must be estimated.

4. CONSTRUCTION OF S-N CURVE

• One can estimate Kf the fatigue notch factor, for fully reversed stresses at
long life.
• In absence of other data, one can estimate the monotonic tensile strength of
the notched part for a metal behaving in a ductile manner to be equal to the
strength of the smooth part in monotonic testing.
• This corresponds to one-quarter of a fully reversed cycle but it may be
assumed to be 1 cycle.
• If a monotonic test can be made, one can use its result rather than the
estimate.
• Using two endpoints; one at 1 cycle and the other at 106 to 108 cycles
connecting them by a straight line, can obtain the notched S-N curve,
•
• Following examples demonstrates the construction of the S-N curve for an
80 mm wide sheet of 1020 hot-rolled steel with a 10 mm central hole:
• From material property table, for un-notched fatigue strength at
approximately 106 cycles and ultimate tensile strength are 241 and 448 MPa,
respectively, under application of bending loads.
• For axial loading, fatigue strength is 10 to 25% of fatigue strength under
bending loading.
• In this example let us use 15 percent reduction resulting in an unnotched
fatigue limit of 0.85 (241) = 205 MPa at 106 cycles which defines the lower
limit of the line and upper limit of line is 448 MPa at 1 cycle.
• From Fig. 4, theoretical stress concentration factor is 2.7 for d/w=0.125.
• From Fig. 5, material characteristic parameter, = 0.49 for Su=448MPa.
448

0.49

2.7

Fig. 5. Neuber's material characteristic length 

0.125

Fig. 4. Theoretical stress concentration vs. tensile ultimate strength for steel alloys
factor for a central circular hole in a plate
• Fatigue notch factor Kf is given by
K −1 2 .7 − 1
Kf = 1+ t = 1+ = 2.39
 1+
0.49
1+
r 5
• Fig. 6 depicts the S-N line for the plate with the hole goes from 448 MPa at 1
cycle to 205/2.39 = 85.7 MPa at 106 cycles.
• It should be noted that for metals behaving in a brittle manner, the notch
effect at short lives is usually more pronounced than that shown in Fig. 6.
• Also, an alternative estimate of the S-N curve for a notched part made of a
ductile material assumes equal fatigue strengths of the notched and smooth
members at 103 reversals.

without a notch.

with notch.

Fig. 6. S-N curve for 1020 hot-rolled sheet steel

with and without a notch.
5. MODIFIED GOODMAN MODEL FOR NOTCH
• A simpler approach for estimating the long life fatigue strength with a tensile
or compressive mean stress which involves the use of the modified
Goodman equation for unnotched components is given by

Sa Sm
+ =1
Sf Su

• For a notched member, the long life smooth fatigue strength, Sf is simply
divided by the fatigue notch factor, Kf. i.e.,

Sa S
+ m =1
(Sf Kf ) Su
where Sa, and Sm are unnotched alternating and mean stresses,
respectively.
• Estimates for both smooth and notched parts based on the modified
Goodman equation, along with the yield limits, are shown in Fig. 7.
• Estimated long life fatigue strength based on this equation, however, is
usually less conservative in the tensile mean stress region but more
conservative in the compressive mean stress region.
• With Fig. 7, one can estimate the long life fatigue strength of parts with
notches for any combination of mean stress and alternating stress.
• A point on the S-N curve also can be estimated for long fatigue life.
• For an estimate of a short life at a high stress, static fracture is one point that
can be used.
Sa
S’y

Yielding

Sf Unnotched

Notched
Sf /Kf
-sy 0 Sy Su f
-Sm +Sm
Fig. 7 Fatigue and yielding criteria for constant life of with and without notched parts.
 SHIGLEY EXAMPLES FROM 6-6 PAGES FROM-289

• A notched part is shown in Fig. , is made from quenched and tempered hot-
rolled RQC-100 steel. The elastic stress concentration factor, Kt is 3. The
nominal stress, S = P/A + Mc/I, is 11.25P (MPa), where P is the load in kN.
Relevant data for RQC-100 steel in the long transverse direction are given
by Su = 931MPa, Sy = 883MPa, f = 1330MPa, Sy = 600MPa, f =1240MPa,
and b= -0.07. Construct the S-N line for completely reversed, constant
amplitude loading.

 max max
Kt = =
 
  max max
Kt = =
 
 FATIGUE FROM VARIABLE AMPLITUDE LOADING

1. SPECTRUM LOADS
• In virtually every engineering applications where fatigue is an important
failure mode, the alternating stress amplitude may be expected in vary or
change in some way during the service life.
• Such variations and changes in load amplitude, often referred as Spectrum
loading.
• Service load histories are usually variable amplitude, as shown in Fig. 1.

(a) (b)

(c)

Fig. 1 Load-time responses in service.

• To measure the load history, transducers (most commonly electrical
resistance strain gages) are attached to the critical areas of the component.
• These critical areas are often identified either analytically, by FEA or
experimentally.
• Acquired data from the transducers are usually recorded and stored by a
computer or by other devices.
• Recorded data may be filtered to isolate the primary loads from noise, and
then are often summarized or compressed by cycle counting methods in order
to simplify the fatigue damage computations.
• If closed-loop electro-hydraulic test systems are used, real-life load histories
can be applied directly to small test specimens, components, subassemblies,
and even entire products.
• Historically, complex load histories were often, and still are sometimes,
replaced in test programs by more simplified loadings, such as the block
programs shown in Fig.1. or by constant amplitude tests.

Programmed six load level test. Random block program loading.

Fig. 2 Block program load spectra
2. CUMULATIVE DAMAGE THEORIES
• It is important to a designer to have available a theory or hypothesis, verified
by experimental observations, that will permit good design estimates to be
made for operation under conditions of spectrum loading using the standard
constant-amplitude S-N curve that are more readily available.

• Under spectrum loading, the basic postulate adopted is that operations at any
given cyclic stress amplitude will produce fatigue damage, the seriousness of
which will be related to the number of cycles of operation at that stress
amplitude and also related to the total number of cycles that would be required
to produce failure of an undamaged specimen at that stress amplitude.

• It is further postulated that the damage incurred is permanent and operation at

several different stress amplitudes in sequence will result in an accumulation
of total damage equal to the sum of the damage increments occurred at each
individual stress level.
• For variable load histories, the cumulative damage is defined as the fraction
of life (cycle ratio) by an event or a series of events or is the ratio of
instantaneous to final crack length. i.e.,
a
D= (1)
ac
• where ac, is the critical crack length for the highest expected load, a is
extended the crack or crack growth for a loading event and D is damage
fraction.
• In addition to the life fraction (or cycle ratio) and crack length or crack
population, many other measures have been used to quantify fatigue
damage.
• These include metallurgical parameters, mechanical parameters, and
physical measures. Metallurgical parameters include the size or number of
dislocations and the spacing or intensity of slip-bands.
• These measures represent the physical nature of fatigue damage explicitly,
but their quantification usually involves destructive evaluation techniques.
• Mechanical parameters directly reflect damage through changes in the
mechanical responses of the material such as hardness, stress, strain,
stiffness, and strain energy as the material degrades with fatigue damage
accumulation.
• Physical measures indirectly quantify the fatigue damage and consist of
mainly nondestructive techniques such as X-radiography, acoustic emission,
ultrasonic techniques, magnetic field methods, potential drop, and eddy
current techniques.
• Many different cumulative damage theories have been proposed for the
purpose of assessing fatigue damage caused by operation at any given
stress level and the addition of damage increments to properly predict failure
under conditions of spectrum loading.
• There are many cumulative damage theories based on linear and non-linear
damage rules, are as follows:
(i) Palmgren-Miner Linear Damage Theory
• First cumulative damage theory was proposed by Palmgren in 1924 and
later developed by Miner in 1945.
• The theory is described using the S-N Sa f
curve (Fig. 3).
• By the definition of the S-N curve, operation
at a constant stress amplitude S1 will Sf
produce complete damage, or failure, in N1
cycles. Cycles to failure, Nf
Fig. 3 S-N curve.
• Operation at stress amplitude Sa1 for a number of cycles n1 smaller than Nf1
will produce a smaller fraction of damage, say D1.
n1
D1 = (2)
Nf 1
where Nf1 is the number of repetitions of this same cycle that equals the
mean life to failure and D1 is usually termed the damage fraction.
• Operation over a spectrum of different stress levels results in a damage
fraction Di, for each of the different stress levels Sai in the spectrum.
• When these damage fractions sum to unity, failure is predicted, i.e., failure
is predicted to occur, if;

 Di = D1 + D2 + D3 +  = 1 (3)

• Palmgren-Miner hypothesis asserts that the

damage fraction at any stress level Sai is Sa f
linearly proportional to the ratio of number
Sa1
of cycles of operation to the total number of Sa2
cycles that would produce failure at that Sf
stress level, i.e.,
n1 Nf1 n2 Nf2
n
Di = i (4) Cycles to failure, Nf
Nf i Fig. 3 S-N curve.
 • Fig.4 shows two blocks of constant amplitude stress cycles, and the
corresponding S-N curve, with fatigue lives at stress amplitudes Sa1, and
Sa2 denoted by Nf1 and Nf2, respectively.
• Damaging effect of n1 cycles at Sa1 stress amplitude is assumed to be
D1 = n1/Nf1, while the damaging effect of n2 cycles at Sa2 stress amplitude
is assumed to be D2 = n2/Nf2.
• Similarly, the cycle ratio or damage caused by ni cycles at Sai stress
amplitude is Di = ni/Nfi.
• Thus, the failure is predicted when the sum of all ratios becomes 1or
100%, we have,
 n n n n
 i = 1 + 2 + 3 +  = 1 (5)
i =1 N N N N
fi f1 f2 f3

Sa
Sa1 Sa f
Sa2
Sa1
0 N Sa2
Sf

n1 Nf1 n2 Nf2
n1 n2
Cycles to failure, Nf
Fig. 4(a) Constant amplitude stress-cycle blocks. Fig. 4(b) S-N curve.
• Eq. (5) is also used with other fatigue curves such as load-life or -N curves.
• Assumption of linear damage is open to many objections.
• For example, the sequence and interaction of events may have major
influences on the fatigue life also, the rate of damage accumulation may be a
function of the load amplitude such that at low load levels most of the life is
involved in crack nucleation, while at high load levels most of the life is spent
in crack growth.
• Experimental evidence under completely reversed loading conditions for
both smooth and notched specimens often indicates that nilNfi  1 for a low-
to-high or a high-to-low loading sequence.
• Even though the linear damage rule ignores these effects, it is commonly
used because none of the other proposed methods achieves better
agreement with data from many different tests.
• Palmgren and Miner use an average damage, and their method is the
simplest and still the most widely used approach to predicting fatigue life to
the appearance of cracks and in many cases to predicting the total fatigue
life to fracture.
(ii) Nonlinear Damage Theories

• To remedy the deficiencies associated with the linear damage assumption,

many nonlinear cumulative fatigue damage theories have been proposed.

• These theories account for the nonlinear nature of fatigue damage

accumulation by using nonlinear relations such as

i
n 
D =   i  (6)
 Nfi 
where the power I depends on the load level, proposed by Marco and
Starkey, rather than the linear relation in Eq.4.
• Fig. 5 shows the linear damage rule and the aforementioned nonlinear rule
at three stress levels in a plot of fatigue damage vs. cycle ratio.
• This Fig. 5 indicates that according to this nonlinear rule, a cycle ratio of n/Nf
= 0.6, for example, produces damage fractions of D = 0.77, 0.36, and 0.13 at
S1, S2, and S3 stress levels, respectively.
• Damage fraction, 0.6 is independent of the stress level, according to the
linear damage rule.

0.77

0.36

0.13

Fig. 5 Fatigue damage fraction vs. cycles.

 3. CYCLE COUNTING METHOD
• Purpose of all cycle counting methods is to compare the effect of variable
amplitude load histories to fatigue data and curves obtained with simple
constant amplitude load cycles.
• Different counting methods can change the resulting predictions by an order of
magnitude.
• Simple load history is shown in Fig.6a and can be analyzed as cycles, which
uses only the segment from point 1 to point 11, which is the beginning of a
repetition of the same events.
• In the first method of counting shown in Fig. 6b, one cycle with each of the
following five pairs of extreme values can obtained:
100 / 200, 100 / 300, − 200 / + 200, − 200 / − 100, − 300 / − 100.
• In second counting method produces five pairs shown in Fig. 6c:
− 300 / 300, 100 / 200 twice, − 200 / − 100 twice,
• Both the methods result in very different calculations, and the results of the
latter method (Fig. 6c) correlate much better with experimental results.
3 3 3 3 3 3
300 300 300
1 5 11 5 11 1 5 5 11 11 5 1 a
a 200 200 a
100 b 100 100 b
2 4 2 4 2 24 2 4 4 4 b 2
0 0
7 9 7 9 7,9 7 9 9 7 9 d
-100 d -100 -100
d
c -200 -200 6 c
6 10 6 10 6,10 6 10 6 c 10
-300 8 8
-300 8
-300 8
Fig. 6a Load history Fig. 6b First type of counting Fig. 6c Second type of counting
• There are five types of cycle counting methods:
(i) Rainflow Method
(ii) Range-Pair Method
(iii) Racetrack Counting Method
(iv) Level-Crossing Method
(v) Peak Counting Method

• All counting methods must count a cycle with the range from the highest
peak to the lowest valley and must try to count other cycles in a manner that
maximizes the ranges that are counted.

• In addition, all counting methods count every part of every overall range
once and only once.

• They also count smaller ranges down to some predetermined threshold once
and only once.
(i) Rainflow Method
• First proposed by Matsuishi and Endo, with the load-time, stress-time, or
strain-time history plotted such that the time axis is vertically downward.
• Rainflow method is a method for counting fatigue cycles from a time history.
• Consider three pagoda or pyramid roofs as shown in following Figures.
• As rain flowing down from first roof and vertical drops on to second roof and
continues to flow further, it vertical drops on to third roof, continues to flow
over it and finally, flow ends with vertical drop from third roof to ground.
• Each of the “pagoda roofs” so defined, until either a roof extends opposite
beyond the vertical of the starting point, or the flow reaches a point that is
already wet.
• Each time, a half-cycle is so defined.

First roof

Second roof

First roof Third roof

Second roof

Third roof

View from rear corner View from rear end

Moodabidri Thousand pillar Temple
 • Lines going horizontally from a reversal to a succeeding range as rain flowing
down a pyramid roof represented by the history of peaks and valleys
therefore, the method was called “rainflow counting”.
• Operation of the rainflow method is shown in Fig. 7a for a history consisting of
four peaks and four valleys and the rules are explained in following algorithm.
Algorithm:
1) Reduce the stress-time history to a sequence of (tensile) peaks and
(compressive) valleys.
2) Imagine that the stress-time history is a pagoda.
3) Turn the sheet clockwise 90°, so the starting time is at the top.
Stress, S

A(25) I(25)
25

20
E(16)
15 C(14)

10
G(7)
5
B(5)
Stress, S

0 F(2) Time

-5

-10
H(-12)
-15
D(-14)
-20
Time

Fig. 7a Stress history

Fig. 7b Stress history

 4) Each tensile peak is imagined as a source of water that "drips" down the
pagoda.
5) Count the number of half-cycles by looking for terminations in the flow
occurring when either:
It reaches the end of the stress-time history
It merges with a flow that started at an earlier tensile peak; or
It encounters a valley of greater magnitude.
6) Repeat step 5 for compressive valley.

Stress, S

A(25) I(25) A
25

20 B
E(16)
15 C(14)
B C
10
G(7) D
5
B(5)
Stress, S

Time E
0 F(2)

-5 F

-10
F G
H(-12)
-15 H
D(-14)
-20 E
Time

Fig. 7a Stress history I

Fig. 7b Stress history

 7) Assign a magnitude to each half-cycle equal to the stress difference
between its start and termination.
8) Pair up half-cycles of identical magnitude (but opposite sense) to count
the number of complete cycles. Typically, there are some residual half-
cycles.

Stress, S

A(25) I(25) A
25

20 B
E(16)
15 C(14)
B C
10
G(7) D
5
B(5)
Stress, S

Time E
0 F(2)

-5 F

-10
F G
H(-12)
-15 H
D(-14)
-20 E
Time

Fig. 7a Stress history I

Fig. 7b Stress history

 9) Fig. 7c presents the load cycles for different load ranges
10) Resulting range and mean values are as follows:

Path SMax SMin Sm S No. of

Cycles, ni

A −D −I 25 − 14 5 .5 39 1
B − C − B 14 5 9 .5 9 1
E − H − E 16 − 12 2 28 1
F − G − F 7 2 4 .5 5 1
Stress, S

A(25) I(25) A A I
25 25
20 B
20
E(16) E E
15 C(14) 15 C
B C
10 10
G(7) D G
5 5
B(5) B B
Stress, S

Time E
Time

Stress, S
0 F(2) 0 F F

-5 F -5
-10 -10
F G
H(-12) H
-15 H -15
D(-14) D
-20 E
Time

Fig. 7a Stress history I

Fig. 7c Stress history
Fig. 7b Stress history
 11) Cumulative damage fraction is computed as follows:
Path SMax SMin Sm S No. of
Cycles, ni

A −D −I 25 − 14 5 .5 39 1
B − C − B 14 5 9 .5 9 1
E − H − E 16 − 12 2 28 1
F − G − F 7 2 4 .5 5 1

▪ Cumulative damage fraction is computed by following equation

5

D= 
i =1
ni n n n
= 1 + 2 + 3 + 
Nfi Nf 1 Nf 2 Nf 3

▪ For computation of No. of cycles to failure, we have

Sa f
S
= A(Nfi )
B
Sai or Sa1
2
Sa2
where A or f and B are obtained from S-N curve Sf

of the given material using experimentation. n1 Nf1 n2 Nf2

Cycles to failure, Nf
Fig. 8 S-N curve.
 • There are advantage of rainflow counting when it is combined with a notch
strain analysis.
• Load history in Fig.8a is applied to the notched component in Fig. 8b, resulting
in the notch strain history shown in Fig. 8c.

1 1
1 1 3
3 5
5
+
+

Strain, 
7
7
Load, P

P P
Time
Time 2
2 - 8
- 8
6
6 4
4 Fig. 8(b) Notched component.
Fig. 8(c) Notch strain history.
Fig. 8(a) Load-time history.
 • This notch strain history is then applied to the smooth axial specimen shown
in Fig. 8d, resulting in the stress response shown in Fig. 8e and the stress-
strain hysteresis loops shown in Fig. 8f:

1 1
1 1 3
3 5
5
+ 5 -
+

Strain, 
7
7 - 7
Load, P

P P

- Time
Time 2 2
2 - 8
- 8
6
6 4
4 Fig. 8(b) Notched component.
Fig. 8(c) Notch strain history.
Fig. 8(a) Load-time history. 1 1
3 S 1
+ 5 5 5
5 - 1
3
Stress, S

7 7
- 7 7

- Time
2 2 
 
- 8 8 2
6 2
Fig. 8(d) Notch strain history 6 4
applied to a smooth specimen. 4
Fig. 8(e) Notch stress response Fig. 8(f) Notch stress-strain
obtained from a smooth specimen. hysteresis loops.
 • Stress response can also be obtained from the cyclic stress-strain curve or
equation, if available.
• Note that rainflow counting results in closed hysteresis loops, with each closed
loop representing a counted cycle.
• Therefore, the closed hysteresis loops can also be used to perform the cycle
counting.
1 1
1 1 3
3 5
5
+ 5 -
+

Strain, 
7
7 - 7
Load, P

P P

- Time
Time 2 2
2 - 8
- 8
6
6 4
4 Fig. 8(b) Notched component.
Fig. 8(c) Notch strain history.
Fig. 8(a) Load-time history. 1 1
3 S 1
+ 5 5 5
5 - 1
3
Stress, S

7 7
- 7 7

- Time
2 2 
 
- 8 8 2
6 2
Fig. 8(d) Notch strain history 6 4
applied to a smooth specimen. 4
Fig. 8(e) Notch stress response Fig. 8(f) Notch stress-strain
obtained from a smooth specimen. hysteresis loops.
 • The tips of the largest hysteresis loop are at the largest tensile and
compressive loads in the load history (points 1 and 4).
• Also, note that the notch strain-time history (Fig.8c) is quite different from the
corresponding notch stress-time history (Fig.8e).
• During each segment of the loading, the material “remembers” its prior
deformation (this is called “material memory”).
1 1
1 1 3
3 5
5
+ 5 -
+

Strain, 
7
7 - 7
Load, P

P P

- Time
Time 2 2
2 - 8
- 8
6
6 4
4 Fig. 8(b) Notched component.
Fig. 8(c) Notch strain history.
Fig. 8(a) Load-time history. 1 1
3 S 1
+ 5 5 5
5 - 1
3
Stress, S

7 7
- 7 7

- Time
2 2 
 
- 8 8 2
6 2
Fig. 8(d) Notch strain history 6 4
applied to a smooth specimen. 4
Fig. 8(e) Notch stress response Fig. 8(f) Notch stress-strain
obtained from a smooth specimen. hysteresis loops.
 • For example, after unloading from point 1 to point 2, loading from point 2 to
point 3 follows the hysteresis loop shown in Fig.8f:
• But upon unloading from point 3 to point 4, the unloading deformation path
continues to point 4 along the same hysteresis path from point 1 to point 2.

1 1
1 1 3
3 5
5
+ 5 -
+

Strain, 
7
7 - 7
Load, P

P P

- Time
Time 2 2
2 - 8
- 8
6
6 4
4 Fig. 8(b) Notched component.
Fig. 8(c) Notch strain history.
Fig. 8(a) Load-time history. 1 1
3 S 1
+ 5 5 5
5 - 1
3
Stress, S

7 7
- 7 7

- Time
2 2 
 
- 8 8 2
6 2
Fig. 8(d) Notch strain history 6 4
applied to a smooth specimen. 4
Fig. 8(e) Notch stress response Fig. 8(f) Notch stress-strain
obtained from a smooth specimen. hysteresis loops.
 • Damage from each counted cycle (or each closed hysteresis loop) can be
computed from the strain amplitude and mean stress for that cycle as soon as
it has been identified in the counting procedure.
• The corresponding reversal points can then be discarded.

1 1
1 1 3
3 5
5
+ 5 -
+

Strain, 
7
7 - 7
Load, P

P P

- Time
Time 2 2
2 - 8
- 8
6
6 4
4 Fig. 8(b) Notched component.
Fig. 8(c) Notch strain history.
Fig. 8(a) Load-time history. 1 1
3 S 1
+ 5 5 5
5 - 1
3
Stress, S

7 7
- 7 7

- Time
2 2 
 
- 8 8 2
6 2
Fig. 8(d) Notch strain history 6 4
applied to a smooth specimen. 4
Fig. 8(e) Notch stress response Fig. 8(f) Notch stress-strain
obtained from a smooth specimen. hysteresis loops.
 (ii) Range-Pair Method
• In this method, cycles containing small ranges are counted first, and their
reversal points (peaks and valleys) are eliminated from further consideration.
• Consider a stress-time history as presented in Table 1 and Fig.9a depicts
the plot.
Table 1
Time Stress Name
0 (0) A F
H
1 10 B
D
2 4 C J L R
B E G P
3 16 D
Stress, S

Time
4 14 E K
C I T V
5 25 F A Q
N
6 14 G
S
7 22 H
O
8 4 I
9 13 J M U
10 8 K
Fig.9(a) Counting smaller cycles and
11 13 L
discarding the corresponding reversal points.
12 -18 M
13 -6 N
14 -12 O
15 7 P
16 0 Q
17 13 R
18 -6 S
19 0 T
20 -18 U
21 (0) V
 • In Fig. 9a, there are 20 reversals, fourteen of them are counted and
eliminated by counting the seven pairs indicated by crosshatching.

Table 1
Time Stress Name
0 (0) A F
H
1 10 B
D
2 4 C J L R
B E G P
3 16 D
Stress, S

Time
4 14 E K
C I T V
5 25 F A Q
N
6 14 G
S
7 22 H
O
8 4 I
9 13 J M U
10 8 K
Fig.9(a) Counting smaller cycles and
11 13 L
discarding the corresponding reversal points.
12 -18 M
13 -6 N
14 -12 O
15 7 P
16 0 Q
17 13 R
18 -6 S
19 0 T
20 -18 U
21 (0) V
 • This leaves the six reversals as shown in Table 2 of Fig. 9b and looking only
at Fig. 9b, we see three peaks and three valleys.

Table 1
Time Stress Name
0 (0) A F
H F
1 10 B
D
2 4 C J L R L R
B E G P
3 16 D
Stress, S

Time Time

Stress, S
4 14 E K
C I T V V
5 25 F I
A Q A
N
6 14 G
S
7 22 H
O
8 4 I
M U M U
9 13 J
10 8 K Fig.9(b) Further counting of smaller cycles and
Fig.9(a) Counting smaller cycles and
11 13 L discarding the corresponding reversal points.
discarding the corresponding reversal points.
12 -18 M
13 -6 N
14 -12 O
15 7 P Table 2
16 0 Q Time Stress Name
17 13 R 0 (0) A
18 -6 S 5 25 F
19 0 T 8 4 I
20 -18 U 11 13 L
21 (0) V 12 -18 M
17 13 R
20 -18 U
21 (0) V
 • Further, six reversals, two of them are counted and eliminated by counting
the four pairs indicated by crosshatching. This leaves the two reversals as
shown in Table 3 and looking only at Fig. 9c, we see one peak and one
valley.
Table 1
Time Stress Name
0 (0) A F
H F
1 10 B
D
2 4 C J L R L R
B E G P
3 16 D
Stress, S

Time Time

Stress, S
4 14 E K
C I T V V
5 25 F I
A Q A
N
6 14 G
S
7 22 H
O
8 4 I
M U M U
9 13 J
10 8 K Fig.9(b) Further counting of smaller cycles and
Fig.9(a) Counting smaller cycles and
11 13 L discarding the corresponding reversal points.
discarding the corresponding reversal points.
12 -18 M
13 -6 N
14 -12 O F
15 7 P Table 2 Table 3
16 0 Q Time Stress Name Time Stress Name
17 13 R 0 (0) A 0 (0) A
5 25 F Time

Stress, S
18 -6 S 5 25 F V
19 0 T 8 4 I 12 -18 M A
20 -18 U 11 13 L 21 (0) V
21 (0) V 12 -18 M
17 13 R
20 -18 U M
21 (0) V
Fig.9(c) Final cycle left from range-pair counting.
 • Some times, they would be counted as cycle (shown by maximum and
minimum loads separated by a slant) as 25/-18, which is shown in Fig.9c.
• Result of a range-pair count is a table of the occurrence of ranges and, if
desired, of their mean values.
Table 1
Time Stress Name
0 (0) A F
H F
1 10 B
D
2 4 C J L R L R
B E G P
3 16 D
Stress, S

Time Time

Stress, S
4 14 E K
C I T V V
5 25 F I
A Q A
N
6 14 G
S
7 22 H
O
8 4 I
M U M U
9 13 J
10 8 K Fig.9(b) Further counting of smaller cycles and
Fig.9(a) Counting smaller cycles and
11 13 L discarding the corresponding reversal points.
discarding the corresponding reversal points.
12 -18 M
13 -6 N
14 -12 O F
15 7 P Table 2 Table 3
16 0 Q Time Stress Name Time Stress Name
17 13 R 0 (0) A 0 (0) A
5 25 F Time

Stress, S
18 -6 S 5 25 F V
19 0 T 8 4 I 12 -18 M A
20 -18 U 11 13 L 21 (0) V
21 (0) V 12 -18 M
17 13 R
20 -18 U M
21 (0) V
Fig.9(c) Final cycle left from range-pair counting.
 Examples on Fatigue Crack Growth

• Example 1

345
Taking the logarithm of both sides of Paris equation, we obtain

346
• Example 2

347
348
1.

349
Figure 1 Crack size vs. number of cycles

350
• Example 3

2.

351
 2

as

352
• Example 4

353
Figure 3 Triangular stress cycle profile
354
355
 MULTIAXIAL HIGH CYCLE FATIGUE

• Multiaxial states of stress are very common

• The strains are triaxial in a tensile bar
– With longitudinal strain  we have two transverse strains, -  , where 
Poisson’s ratio
• In a shaft that transmits torque
– We have two principal stresses S, equal in magnitude but opposite in sign
– There is neither stress nor strain in the third principal direction
• In a crankshaft we have torsion and bending
– On points of its surface we have two principal stresses that vary in
magnitude and direction,
– The frequencies of the bending cycles and torsion cycles are not the same,
Most points on the surface have triaxial strains

356
• The subject of multiaxial fatigue has developed over many decades

• Initial approaches to modelling multiaxial fatigue behaviour were based

on the static yield criteria developed a century ago, e.g., Von Mises,
Tresca criteria, etc.
• Another class of approaches called critical plane theories devote
specific attention to the orientation of small cracks in multiaxial fatigue

• These theories assert that the most critically damaged plane is one of
maximum shear stress or strain amplitude that experiences the
maximum normal strain and/or normal stress

• These critical plane theories were preceded by some 30 years by the

high-cycle fatigue theories

357
 Multiaxial High-Cycle Fatigue Models

• Most experimental data consists even today, in the stressing system

subjected to of at least four variables, singly or combinations:
i. Reversed bending stresses
ii. Reversed torsion stresses
iii. Static bending stresses
iv. Static torsion stresses
• Fatigue theories are then used to extrapolate into more complex
states of stress
• Starting in 1970s, much development effort was devoted to strain-
based multiaxial fatigue models which are more useful for low-cycle
fatigue analysis
• However, the stress-based models continue to be more widely used
and are suitable for the large class of components that must operate
near or below the fatigue threshold
• Many of the stress-based models can be used successfully in the
finite life regime if the plastic strains are small
• Emphasis is given in this section to presenting those fatigue models
that have gained widespread acceptance 358
I. Gough Models (1935, 1950):
➢ Gough performed experiments on several materials under different
ratios of bending stress/torsional stress to establish the fatigue lime
in combined loading
➢ Based on experimental study, he proposed two formulations:
i. An elliptical quadrant (shown as the solid line in the following figure)
2
Sbs Sss2
+ =1
Sbfl2 Stfl2
ii. An elliptical arc (given as the dashed line)

Sss2 Sbs  Sbfl  Sbs  Sbfl 

2
+  − 1 +  2 −  =1
Stfl2 Sbfl2  Stfl  Sbfl 
Stfl 

Semi-range of shear stress

where Sbfl – bending fatigue limit
Stfl – fatigue limit in tension
Sbs – bending stress
Sss – shear stress in torsion, Sss

359 Sbs
Semi-range of direct stress due to bending,
Gough Ellipse
 ➢ Gough motivated to provide engineering methods that could be
applied to the design of shafts under mixed torsion and
bending loading
➢ The elliptical quadrant and ellipse arc are empirical, having
been chosen simply to fit the experimental data for two
apparently distinct groups of materials: the ellipse quadrant for
ductile materials and the ellipse arc for brittle cast irons or
notched shafts

II. Static Yield Criteria Models:

➢ Numerous investigators, including Gough, have attempted to
correlate multiaxial fatigue test data using parameters originally
developed for static yield under complex stress states
➢ Although, many yield criteria have been suggested, the three
most common criteria are the maximum normal stress theory,
the maximum shear stress theory and the octahedral shear
stress theory

360
➢ The maximum normal stress theory applied to cyclic loading may be
expressed as
eq = 1
➢ The maximum shear stress theory may be expressed as
eq 1 − 3
 = =
2 2
➢ The most popular theory is the octahedral shear stress theory (von
Mises) can be computed

2
 2 2

eq = 1 (1 − 2 ) + (2 − 3 ) + (3 − 1)
2 1/ 2

➢ Above equations are valid only for proportional or in-plane loading

➢ During nonproportional or out-of-loading
1 − 3  (1 − 3 )
so that no unique definition of the shear stress exists
➢ When the principal stress directions remain fixed as they do in a
pressurized pipe, the ASME pressure vessel code recommends find
and using the maximum value of
(1 − 3 )
during any point in the loading cycle 361
➢ Following figure shows a typical set of Gough’s test data, illustrating
the effect of various ratios of bending and shear stress.
➢ This data shows that octahedral shear stress results in a better
correlation than either the maximum shear stress or the maximum
principal stress when comparing tension and torsion data
Shear stress in bending
½ bending fatigue limit

Shear stress in torsion

1/2 bending fatigue limit
Correlation of bending and torsion data

➢ One feature common to the static yield theories when extended to

fatigue is that they predict a constant ratio of fatigue strength in
bending/fatigue strength in torsion, Sbfl/Stfl
362
➢ Table gives the relationship for normal and shear stresses for the
three most common yield criteria

Sbfl
Stfl

➢ Tests showed that Sbfl/Stfl was strongly material dependent: for steel
and aluminium, the ratio varies from 1.3 to 2.5, for cast iron 0.9-1.3
and for copper alloys 1.7-2.7 363
III. Sines Model (1955, 1959):
➢ He proposed that the octahedral shear stress be used as
fatigue damage criteria
➢ Physical significance of the octahedral shear stress is that it
expresses the average effects of slippage on different planes
and in different directions of all crystal in the aggregate, with
slip in any given grain caused by the critical resolved shear
stress in that grain
➢ In three dimensional stress state, it is more difficult to compute
the maximum shear stresses because a cubic equation must
be solved for the three principal stresses, whereas octahedral
stresses can be obtained from direct computations using the
six stress components for any coordinate system
➢ To establish the effects of mean stresses, Sines grouped the
experimental results as follows:
• Cyclic axial with static tension and compression
• Cyclic torsion with static torsion
• Cyclic bending with static torsion
• Cyclic torsion with static bending

364
 ➢ Figure (i) shows the effect of static tensile or compressive stress on
axial fatigue strength ( different symbols in the figure represents
various materials)
➢ Figure (ii) shows few test data showing the effect of static tension
superimposed on alternating torsion suggest a linear dependence
between the amplitude of alternating torsion and the static tensile
stress
➢ The relationship is linear as long as maximum stress during a cycle
does not exceed the yield strength of the material

Figure (i) Cyclic axial stress with static tension and Figure (ii) Cyclic torsion with static tension
365 and
compression compression
➢ Sines concluded that the mean hydrostatic stress during a cyclic
had an effect on fatigue life.
➢ His resulting failure criterion can be expressed as
oct
+ (3h ) = 
2
or

6 x y x z y z ( xy yz xz ) (x y z )
1 ( −  )2 + ( −  )2 + ( −  )2 + 6 2 + 2 + 2 +  mean + mean + mean = 

➢ In the case of combined alternating axial and shear stress, the

damage parameter leads to a series of concentric ellipse as
shown in following figure

366
Sines criterion for the case of alternating tension and torsion
➢ Referring to the above figure, the ordinate and abscissa represent
the amplitude of the alternating components of stress, and the size
of the ellipse is linearly proportional to the sum of the tensile static
stresses

➢ The use of hydrostatic stress predicts that the damaging effect of a

tensile mean stress, mx ean , in one direction can be offset by a
compressive mean stress, myean , in a different direction

➢ Sines formulation has the advantages that it is easily solved for

complex stress states and only relatively simple tests must be
done to determine the two constants  and . The above equation
is defined only in the case of proportional loading.

367
IV. Findley Model (19956, 1959):
➢ Findley reviewed much of the same experimental data as Sines
but came to a slightly different model
➢ He suggested that the normal stress, n on a shear plane might
have a linear influence on the allowable alternating shear
stress, /2.
(2 +  )
n
max
= Sbs

➢ Parabolic forms were investigated, but Findley concluded that

the linear form was sufficient to describe the experimental data.
This model differs from Gough’s, Sines’, or the yield criteria
models, in that it identifies the stress acting on a specific plane
within the material.
➢ These are termed critical planes and can be defined as one or
more planes within a material subject to a maximum value of a
damage criterion. Fatigue life is controlled by the combination
of stresses and strains acting on a critical plane.

368
 ➢ Findley identifies a critical plane for fatigue crack initiation and
growth that is dependent on both alternating shear stress and
maximum normal stress
➢ The combined action of shear and normal stresses is responsible
for fatigue damage and the maximum value of the quantity in
parenthesis in above equation is used rather that the maximum
value of shear stress

V. McDiarmid Model (1991, 1994):

➢ McDiarmid analyzed much of the available high-cycle fatigue data
and proposed a criterion based on the shear stress amplitude
max/2, on the plane of maximum range of shear stress and the
normal stress n,max on the same plane.

max 
+ n, max = 1
2Ssfs, A, B 2uts

where Ssfs, A,B - Shear fatigue strength for Case A or Case B cracking
uts – Ultimate tensile strength

369
› Regarding Case A and Case B crack growth
 Shear crack can grow in one of two ways, illustrated by Case A and
Case B in the following figure.
 In a Case A, the shear stress acts on the free surface in a direction
parallel to the length of the crack. This is an in-plane shear stress. No
shear stress is acting perpendicular to the free surface along the crack
depth. As a result, these types of cracks tend to be shallow and have a
small aspect ratio
 Case B cracks are the result of out-of-plane shear and care the ones
described by the intrusion-extrusion model.
 Tension loading has the same shear stress for both Case A and Case
B and can display either mode of crack growth
 Torsion or combined tension-torsion loading always has Case A cracks

Case A Case B

Case A and Case B crack growth 370

➢ The McDiarmid model is similar to Findley’s model, with the following
modifications:
 It considers both A and B type cracking modes, Case A cracking
propagates cracks that grow along the free surface, and Case B cracking
results in cracks that penetrate in to the material Ssfs, A, B
 Findley’s material parameter  is replaced by the quantity 2 . Thus,
uts
only one set of fatigue tests and one static test are necessary, as
compared to the two sets of fatigue tests needed to define 
 The critical plane is defined as the plane with that maximum shear stress
amplitude and not the plane on which the damage quantity is maximized.
In some cases, this may be significantly different from the plane defined
by Findley.
➢ In the finite life region, this equation can be combined with Basquin;s
equation
max  Ssfs, A, B 
+  n, max = f (2Nf )
b
2  2uts 
where f - Shear fatigue strength coefficient, Nf - Cycles to failure

371
VI. Dang Van Model (1989, 1993, 1999):
➢ Dang Van has proposed an endurance limit criterion based on
the concept of microstress within a critical volume of the
material
➢ This model arises from the observation that fatigue crack
nucleation is a local process and begins in grains that have
undergone plastic deformation and form characteristic slip
bands. It is hypothesized that because cracks usually nucleate
in intraganular slip band, the microscopic shear stress on a
grain must be an important parameter
➢ In the same way, it is reasoned that the microscopic hydrostatic
stress will influence the opening of these cracks or slip bands.
➢ The simplest failure criterion involving these two variables is a
linear combination
(t) + a h(t) = b
where (t) and h(t) are instantaneous microscopic shear stress and
hydrostatic stress respectively, a and b are constants, t is time.

372
➢ The microscopic stresses and strains within critical grains
are different from the macroscopic stresses and strains
commonly computed for fatigue analysis
➢ Two size scales (following figure) have been distinguished in
the model:

ij(m,t)
ij(m,t)
m
V(M)

ij(M,t)
Eij(M,t)

Macroscopic and microscopic scales

 A macroscopic scale is characterized by an elementary volume V(M)

surrounding a point where the fatigue analysis is to be performed. This
size scale is on the order of a strain gage or FEM element. This is the
scale considered by engineers and is the same as local stresses and
strains as defined in the "local strain" approach to fatigue analysis.
Macroscopic stresses are denoted (M,t) and strains E(M,t). Both are
functions of position within the structure, M, and time, t.
 A microstructural scale on the order of a grain or other suitable microstructural
unit corresponding to a subdivision of V(M). Microscopic stresses (rn,t) and373
strains (m,t) are related but not equal to (M,t) and E(M,t).
 ▪ At the microstructural scale, a material is neither isotropic nor
homogeneous, and the values  (m,t) and (m,t) will be different from
the local macroscopic variables (M,t) and E(M,t). Plastic shear
deformation of the grains with the most severe orientations is
constrained due to the elastic behavior of neighboring grains having less
severe orientations. If fatigue failure is to be avoided, the stresses and
strains in the critically oriented grains must stabilize by the process of
elastic shakedown and thus prevent crack growth to the neighboring
grain.

➢ The microscopic shear stress used in the failure criterion is

computed from the microscopic principal stresses using the Tresca
maximum shear stress theory

(t ) = 1 1(t ) − 3(t )

2

374
➢ The criterion can be expressed as combinations of‘ (t) and
hydrostatic stress, as shown in following figure. The constants a
and b are obtained from materials tests at two different stress
states. A loading path that remains within the two bounding failure
lines is expected to have infinite life, whereas any path that extends
outside the damage line will have fatigue failures because
microscopic plastic strains occur.

 (t) + a h(t) = b  Failure predicted

h h
Loading path

Dang Van criteria

➢ The Dang Van criterion is intended as a method of predicting the
endurance limit under complex loading situations. However, it is
sometimes applied in the long but infinite life regime as well, with
different values of constants a and b being determined for different
fatigue life ranges. 375
 Small Crack Growth in Multiaxial Fatigue

• Crack nucleation processes are associated with the generation and

coalescence of excess vacancies along persistent slip bands in ductile
single crystals or coarse grain polycrystals
• In polycrystals, cracks may nucleate via fracture during processing at
intersecting slip bands or by blockage of a slip band by second-phase
particles
• A second type of microcracking in polycrystals occurs along grain
boundaries due to impurity embrittlement or the presence of voids

• We focus here on the growth of small cracks in fatigue rather than the
nucleation problem, that is, the propagation of small cracks of a fatigue
crack on the order of 1mm

376
• Characteristics of small fatigue cracks
– Mixed-mode fatigue crack propagation studies have largely focused on the
behaviour of mechanically long cracks
– The problem of the growth of small cracks in fatigue (from lengths on the
order of 1 to 500-1000m) has received increased attention
– Cracks are considered to be small when all pertinent dimensions are small
compared to some characteristic length scale
– Typical propagation behaviour of small cracks is shown in following figure.
– Considerable scatter of the fatigue crack growth rate of small cracks at a
given K is noticeable
– At low stress amplitudes (high-cycle fatigue), deceleration of crack growth
is often observed, associated with a dip in the da/dN versus K behaviour

Typical propagation behavior of small

cracks. Note that da/dN is higher for a
given K than for long cracks, and the
apparent scatter in da/dN is significant.
The bottom dashed line is a linear
extension of Paris regime
377
 – Subsequently, crack growth may accelerate prior to merging with the long
crack data

– At sufficiently low amplitudes, small cracks may become arrested

– As small cracks propagate, their da/dN versus K responses are typically

observed to merge with the long crack response

• Experimental observations indicate that the propagation behaviour of

micro-structurally small and physically small cracks depends
significantly on both the R-ratio and stress amplitude, in addition to
stress state

378
• The data in following figure clearly illustrate some important aspects of
multiaxial fatigue crack growth for constant-amplitude loading of a
ductile alloy in tension-compression and in torsion
• The curved contours represent the locus of normalized cycles, N/Nf, to
growth to a 0.1 mm surface crack, with Nf corresponding to the number
of cycles of growth to a 1 mm surface cracks
• Regimes of shear-dominated growth (stage A) along maximum shear
strain range planes and normal stress-dominated growth (stage B)
normal to the range of maximum principal stress are shown
• The curve representing the fraction of life to a 0.1 mm crack is termed
“crack nucleation”

Data of Socie on 1045 steel for life to 0.1 mm and 1 mm cracks (N/Nf = 1) for torsional
and uniaxial loading 379
• There are different stages of fatigue damage in an engineering
component where defects may nucleate in an initially under-damaged
section and propagate in a stable manner until catastrophic fracture
occurs
• For this most general situation, the progression of fatigue damage can be
broadly classified in to the following stages:
– Sub-structural and micro-structural changes which cause nucleation
of permanent damage
– The creation of microscopic cracks
– The growth and coalescence of microscopic flaws to form dominant
cracks which may eventually lead to catastrophic failure
– Stable propagation of the dominant macro-crack
– Structural instability or complete fracture

380
• The conditions for the nucleation of micro-defects and the rate of advance
of the dominant fatigue crack are strongly influenced by a wide range of
mechanical, micro-structural and environmental factors.
• Principal differences among differences among different design philosophy
rest on how the crack initiation and the crack propagation stages of fatigue
are quantitatively treated

• The major obstacle to the development of life prediction models for fatigue
lies in the choice of a definition for crack initiation:
– Material scientists concerned with the microscopic mechanisms of
fatigue are likely to regard the nucleation of micrometer-size flaws
along slip bands and grain boundaries and the roughening of fatigue
surfaces as the crack inception stage of fatigue failure
– A Practicing engineer tends to relate the limit of resolution of the crack
detection with the nucleation of a fatigue crack and with initial crack
size (typically of fraction of a millimeter) used for design
– Scattered within the limits of this broad range of choices, there lies a
variety of definitions for crack nucleation which are specific to certain
classes of fatigue-critical engineering applications
381
 – The total fatigue life is defined as the sum of the number of cycles to
initiate a fatigue crack and the number of cycles to propagate it to some
final crack size.
– Following methods constitutes crack initiation, making a clear
relationship between crack initiation and crack propagation has
become a critical task
➢ Total-life or Infinite-life Design
➢ Safe-life Design
➢ Fail-safe Design
➢ Defect-tolerant or damage-tolerant Design

➢ Total-life Design:
▪ Classical approaches to fatigue design involve the characterization of
total fatigue life to failure in terms of the cyclic stress range (i.e., S-N
curve approach) or the strain range (i.e., -N curve approach).
▪ It requires local stresses or strains to be essentially elastic and safely
below the pertinent fatigue limit
▪ For parts subjected to millions of cycles, this is still a good design
criterion
▪ This design criterion may not be economical or practical in many
design situations
382
▪ Examples: Automotive and locomotive components
➢ Safe-life Design:
▪ The practice of designing for a finite life is known as safe-life design
▪ Safe-life design may be based solely or partially on field and/or
simulated testing
➢ In the safe-life design approach, the typical cyclic load spectra,
which are imposed on a structural component in service are first
determined
➢ On the basis of information, the components are analyzed or
tested in the laboratory under load conditions which are typical of
service spectra,
➢ A useful fatigue life is estimated for the component
➢ Estimated fatigue life, suitably modified with a factor of safety
➢ Then provides a prediction of safe-life for the component
▪ Examples, railroad axles, automobile parts, etc.

383
➢ Fail-safe Design:
▪ In the fail-safe concept, even if an individual component of a large
structure fails there should be sufficient structural integrity in the
remaining parts to enable the structure to operate safely until the
crack is detected
▪ Components which have multiple load paths are generally fail-safe
because of structural redundancy
▪ The fail-safe approach mandates periodic inspection along with the
requirement that the crack detection techniques be capable of
identifying flaws to enable prompt repairs or replacements
▪ This step eliminates dangerous consequences arising from false
estimates and errors in the design stage, especially with the safe-life
approach
▪ Finally, when a component, structure or vehicle reaches its allowable
safe-life, it must be retired from service
▪ Examples, aircraft and nuclear industries

384
➢ Defect-tolerant Design:
▪ This philosophy is refinement of the fail-safe philosophy
▪ It assumes that cracks will exist, caused either by processing or by
fatigue, and uses Fracture Mechanics analyses and tests to
determine such cracks will grow large enough to produce failures
before they are detected by periodic inspection
▪ Three key items are needed for successful defect-tolerant design
− Residual strength, fatigue crack growth behaviour and detection
involving non-destructive inspection.
− In addition, environmental conditions, load history, statistical
aspects, and safety factors must be incorporated in this design
▪ Examples include the aero-space and nuclear industries

385
• Traditional structural metals such as steel and aluminium are being
replaced with plastics, ceramics and composites in a number of
applications

• Engineering plastics have a number of advantages, including low cost,

ease of fabrication, and corrosion resistance

• Composites offers high strength/weight ratios and enable engineers to

design materials with specific elastic and thermal properties

• Traditional nonmetallic materials such concrete continue to see

widespread use

386
• Nonmetals, like metals, are not immune to fracture
– Pinch clamping of polyethylene pipe lead to time-dependent fracture
– So-called high toughness ceramics that have been developed in recent
years have lower toughness than even the most brittle steels

• Compared with fracture of metals, research into the fracture behaviour

of nonmetals is in its infancy

• Much of the necessary theoretical frame work is not yet fully developed
for the nonmetals and there are many instances where fracture
mechanics concepts that apply to metals have been misapplied to other
materials

387
 Engineering Plastics

– Fracture behaviour of polymeric materials has only recently become a

major concern as engineering plastics have begun to appear in critical
structural applications

– In most consumer products made from polymers (e.g., toys, garbage bags,
ice chests, lawn furniture, etc.) fracture may be an annoyance, but it is not a
significant safety issue

– Fracture in plastic natural gas piping systems or aircraft wings, however,

can have dire consequences

388
 Yielding and Fracture in Polymers

• In metals, fracture and yielding are competing failure mechanisms

• Brittle fracture occurs in materials in which yielding is difficult
• Ductile metals, by definition, experience extensive plastic deformation
before they eventually fracture
• Low temperatures, high strain rates and triaxial tensile stresses tend to
suppress yielding and favor brittle fracture
• From a global point of view, the forgoing also applies to polymers, by
the microscopic details of yielding and fracture in plastics are different
from metals
• Polymers do not contain crystallographic planes, dislocations, and grain
boundaries; rather, they consist of long molecular chains
• In metals, fracture on the atomic level involves breaking of bonds and
polymers are no exception

389
• A complicating feature for polymers is that two types of bond govern
the mechanical response; the covalent bonds between carbon atoms
and the secondary van der Waals forces between molecule segments

• Ultimate fracture normally requires breaking the molecule segments,

but the secondary bonds often a major role in the deformation
mechanisms that lead to fracture

• The fracture that govern the toughness and ductility of polymers include
strain rate, temperature and molecular structure

• At high rates or low temperatures, polymers tend to be brittle, because

there is insufficient time for the material to respond to stress with larg-
scale viscoelastic deformation or yielding

390
 FATIGUE OF ENGINEERING PLASTICS

• Time dependent crack growth in the presence of cyclic stresses is a

problem in virtually all material systems
• Two mechanisms control fatigue in polymers; chain scission and
hysteresis heating
– Crack growth by chain scission occurs in brittle systems, where crack tip
yielding is limited. A finite number of bonds are broken during each stress
cycle, and measurable crack advance takes place after sufficient cycles
– Unlike elastic materials, where the unloading and loading path coincide and
the strain energy is recovered, a viscoelastic material displays a hysteresis
loop in the stress-strain curve (figure 1); the area inside this loop represents
energy that remains in the material after it is unloaded

Absorbed energy
Stress

Strain 391
• When a viscoelastic material is subject to multiple stress cycles, a
significant amount of work is performed on the materials

• Much of this work is converted to heat and the temperature in the

material rises

• The crack tip region in a polymer subject to cyclic loading may rise to
well above Tg (glass transition temperature at which the relative motion
of the molecules becomes restricted), resulting in local melting and
viscous flow of material

• The rate of crack growth depends on the temperature at the crack tip,
which is governed by the loading frequency and the rate of heat
conduction away from the crack tip

392
 Thermal Failure

– The major cause of thermal failure is believed to involve the

accumulation of hysteresis energy generated during each loading
cycle

– Since this energy is dissipated largely in the form of heat, an

associated temperature rise will occur for every loading cycle when
isothermal conditions are not met

– As shown in figure (i), the temperature rise can be great enough to

cause the sample to melt, thereby, preventing it from carrying any
load

– Failure is presumed, therefore, to occur by viscous flow, although

the occurrence of some bond breakage cannot be excluded

393
 (a)

(b)
Figure (i) Effect of temperature rise during fatigue testing. (a) Temperature rise to cause thermal
failure at different stress levels ( No failure seen in F where sample temperature stabilized (b)394
Loss
compliance and temperature rise
 Mechanical Failure – Fatigue Crack Propagation (FCP)

• During the past few years, a number of FCP studies of engineering

plastics have been conducted and data are now available
• With such a body of data, certain conclusions and generalities may be
drawn:

Relationship between crack growth

rate in several polymers as a function
of range stress intensity factor.
PMMA- Plymethyl methacrylate
LDPE- Low-density polyethylene
ABS- Acrylonitrile-butadiene-styrene
PC- Polycarbonate

395
– A in metals, the FCP rates of polymers are strongly dependent on
the magnitude of the stress intensity factor range, regardless of
polymer chemistry or long-range architecture

– The superior FCP behaviour exhibited by polycarborate and Nylon

66 was related to the availability of significant energy dissipation
mechanism that are linked to main chain segmental motions

– For the case of Nylon 66, superior FCP behaviour also may have
been due to crystalline regions acting in a manner that would retard
crack advance.
– PS and PMMA the low fatigue resistant materials, since these
posses this dubious distinction as a result of their low fracture
energy, amorphous structure, and/or tendency to craze.

396
ASTM Standards for determination of Fracture
and Fatigue Properties
Category 1: using smooth and notched specimens
E466 Conducting constant amplitude axial fatigue tests of metallic
materials
E606 Constant amplitude low-cycle fatigue testing
E338 Sharp-notch tension testing of high strength sheet materials
E602 Sharp-notch tension testing with cylindrical specimens
E208 Conducting drop weight test to determine nil-ductility transition
temperatures of ferritic steels
E23 Notched bar impact testing of metallic materials
E436 Drop weight test of ferritic steels
E604 Dynamic tear testing of metallic materials
E1304 Plane strain chevron-notch fracture toughness of metallic
materials
397
 Category 2: using cracked specimens

E740 Fracture testing with surface crack tension specimens

E399 Plane strain fracture toughness of metallic materials
E1290 Fracture toughness measurement crack tip opening
displacement (CTOD)
E813 JIc, a measure of fracture toughness
E561 R-curve determination
E1221 Determining the plane strain crack arrest fracture toughness,
KIa, of ferritic steels
E1457 Measurement of creep crack growth rates in metals
E647 Measurement of fatigue crack growth rates
E1820 Standard test method for the measurement of fracture
toughness

398
 REFERENCES
Scientific journals
–International Journal of Fracture
–Engineering Fracture Mechanics
–ASME Journal of Applied Mechanics
–International Journal of Solids and Structures
–Journal of the Mechanics and Physics of Solids
Series on fracture mechanics
–“Fracture : an advanced treatise”, 7 volumes edited by H. Liebowitz
–“Mechanics of fracture”, 7 volumes edited by G.C. Sih
Tables of stress intensity factors
–X.-R. Wu and A.J. Carlsson, (1991), “Weight functions and stress intensity factor
solutions”,
Pergamon
–“Stress intensity factor handbook”(2 volumes), (1987), Pergamon
–D.P. Rookeand D.J. Cartwright, (1976), “Compendium of stress intensity factors”,
London
–D.P. Rooke, “Compounding stress intensity factors : applications to engineering
structures”,
(1986), Parthenon Press
–G.C. Sih, ed., “Plates and shells with cracks : a collection of stress intensityfactor
solutions for
cracks in plates and shells”, (1977), Noordhof
–G.C. Sih, “Handbook of stress intensity factors”, (1973), Bethleem
399
Books: General
–T.L. Anderson, (1995), “Fracture mechanics : Fundamentals and applications”, CRC
Press
–David Broek, (1986), “Elementary fracture mechanics”, Nijhof/Kluwers
–E.E. Gdoutos, (1993), “Fracture mechanics : An introduction”, Kluwers
–Melvin F. Kanninen, (1985), “Advanced fracture mechanics”, Oxford University Press
–Bertram K. Broberg, (1999), “Cracks and Fracture”, Academic Press
Books: Introduction-History
–H. Petrosi, (1994), “Design paradigms : Case histories of error in judgment in
engineering”, Cambridge
–J.M. Barsom, (1987), “Fracture mechanics retrospective : Early classic papers (1913-
1965)”, ASTM RPS1
Books: Dynamic Fracture
–L.B. Freund, (1990), “Dynamic fracture mechanics”, Cambridge
–V.Z. Partonand V.G. Borikovski, (1989), “Dynamic fracture mechanics. Volume 1:
Stationary cracks. Volume 2: Propagating cracks”, Hemisphere Publ.
Books: Numerical Methods in Fracture Mechanics
–M.H. Aliabadiand D.P. Rooke, (1991), “Numerical Fracture Mechanics”, Kluwers
–D.R. Owen and A.J. Fawkes, (1983), “Engineering fracture mechanics: Numerical
methods and applications”, PineridgePress
Books: Micromechanics of Fracture
–B. Lawn, (1993), “Fracture of brittle solids -Second edition”, Cambridge
–S. Suresh, (1991), “Fatigue of materials”, Cambridge
400
Papers on Fundamental Equations:
• R. S. Barsoum. On the use of isoparametric finite elements in linear
fracture mechanics. International Journal of Numerical Methods in
Engineering, 10(1), 25-37, 1976
• J. R. Rice, A path independent integral and the approximate analysis of
strain concentration by notches and cracks. Journal of Applied Mechanics,
10, 379-386, 1968.
• I.L.Lim, I.W.Jhonston and S.K.Choi, Application of singular quadratic
distorted isoparametric elements in linear fracture mechanics. International
journal for numerical methods in engineering, 36, 2473‐2499, 1993.
• S. L.Pu, and MA Hussain, The Collapsed cubic isoparametric element as
a singular element for crack problem, International Journal of Numerical
Methods in Engineering, 12,1727-1742, 1978

401
• R. S. Barsoum. On the use of isoparametric finite elements in linear
fracture mechanics. International Journal of Numerical Methods in
Engineering, 10(1), 25-37, 1976
• J. R. Rice, A path independent integral and the approximate analysis of
strain concentration by notches and cracks. Journal of Applied Mechanics,
10, 379-386, 1968.
• I.L.Lim, I.W.Jhonston and S.K.Choi, Application of singular quadratic
distorted isoparametric elements in linear fracture mechanics. International
journal for numerical methods in engineering, 36, 2473‐2499, 1993.
• S. L.Pu, and MA Hussain, The Collapsed cubic isoparametric element as
a singular element for crack problem, International Journal of Numerical
Methods in Engineering, 12,1727-1742, 1978
• Ralph I. Stephens, Ali Fatemi, Robert .R. Stephens and Henry O Fuchs,
Metal Fatigue in engineering, John Wiley, New York.
• Jack. A. Collins, Failure of Materials in Mechanical Design, John Wiley,
New York.
• Robert L. Norton, Machine Design, Pearson.