Transactions Journal of of the ASME Engineering Materials

and Technology
Technical Editor DAVID L. MCDOWELL GEORGE J. WENG (Past Technical Editor) Materials Division Associate Technical Editors R. BATRA 2001 C. BRINSON 2002 E. BUSSO 2001 K. CHAN 1999 N. CHANDRA 2000 S. DATTA 2000 G. JOHNSON 2001 J. W. JU 2000 S. MALL 2001 D. MARQUIS 2001 S. MEGUID 2001 A. M. RAJENDRAN 2001 G. RAVICHANDRAN 1999 H. SEHITOGLU 2000 E. WERNER 2002 H. ZBIB 2000 BOARD ON COMMUNICATIONS Chairman and Vice-President R. K. SHAH OFFICERS OF THE ASME President, R. E. NICKELL Executive Director D. L. BELDEN Treasurer J. A. MASON PUBLISHING STAFF Managing Director, Engineering CHARLES W. BEARDSLEY Director, Technical Publishing PHILIP DI VIETRO Managing Editor, Technical Publishing CYNTHIA B. CLARK Managing Editor, Transactions CORNELIA MONAHAN Production Assistant MARISOL ANDINO

Published Quarterly by The American Society of Mechanical Engineers

VOLUME 122 NUMBER 2 APRIL 2000

TECHNICAL PAPERS
149 Predictive Modeling of the Nonuniform Deformation of the Aluminum Alloy 5182 M. E. Bange, A. J. Beaudoin, M. Stout, S. R. Chen, and S. R. MacEwen Multiaxial Cyclic Ratcheting in Coiled TubingPart I: Theoretical Modeling Radovan Rolovic and Steven M. Tipton Multiaxial Cyclic Ratcheting in Coiled TubingPart II: Experimental Program and Model Evaluation Radovan Rolovic and Steven M. Tipton Thermomechanical Behavior and Modeling Between 350C and 400C of Zircaloy-4 Cladding Tubes From an Unirradiated State to High Fluence 0 to 85" 1024 nm2, E 1 MeV fer, P. Geyer, P. Boufoux, and P. Delobelle I. Scha The Inuence of Intrinsic Strain Softening on Strain Localization in Polycarbonate: Modeling and Experimental Validation L. E. Govaert, P. H. M. Timmermans, and W. A. M. Brekelmans Low-Cycle Fatigue of TiNi Shape Memory Alloy and Formulation of Fatigue Life Hisaaki Tobushi, Takafumi Nakahara, Yoshirou Shimeno, and Takahiro Hashimoto On the Nonuniform Deformation of the Cylinder Compression Test Fuh-Kuo Chen and Cheng-Jun Chen Fracture Static Mechanisms on Fatigue Crack Propagation in Microalloyed Forging Steels M. A. Linaza and J. M. Rodriguez-Ibabe Damage of Short-Fiber-Reinforced Metal Matrix Composites Considering Cooling and Thermal Cycling Chuwei Zhou, Wei Yang, and Daining Fang Probabilistic Mesomechanics for High Cycle Fatigue Life Prediction Robert G. Tryon and Thomas A. Cruse Study of Surface Residual Stress by Three-Dimensional Displacement Data at a Single Point in Hole Drilling Method Z. Wu and J. Lu A Study of Burr Formation Processes Using the Finite Element Method: Part I I. W. Park and D. A. Dornfeld A Study of Burr Formation Processes Using the Finite Element Method: Part IIThe Inuences of Exit Angle, Rake Angle, and Backup Material on Burr Formation Processes I. W. Park and D. A. Dornfeld Densication Behavior of Ceramic Powder Under Cold Compaction K. T. Kim, S. W. Choi, and H. Park

157

162

168

177

186

192 198

203

209 215

Transactions of the ASME, Journal of Engineering Material and Technology (ISSN 0094-4289) is published quarterly (Jan., Apr., July, Oct.) for $205.00 per year by The American Society of Mechanical Engineers, Three Park Avenue, New York, NY 10016. Periodicals postage paid at New York, NY and additional mailing ofces. POSTMASTER: Send address changes to Transactions of the ASME, Journal of Engineering Materials and Technology, c/o THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS, 22 Law Drive, Box 2300, Faireld, NJ 07007-2300. CHANGES OF ADDRESS must be received at Society headquarters seven weeks before they are to be effective. Please send old label and new address. PRICES: To members, $40.00, annually: to nonmembers, $205.00. Add $40.00 for postage to countries outside the United States and Canada. STATEMENT from By-Laws. The Society shall not be responsible for statements or opinions advanced in papers or ... printed in its publications (B7.1, Par. 3). COPYRIGHT 2000 by the American Society of Mechanical Engineers. For authorization to photocopy material for internal or personal use under those circumstances not falling within the fair use provisions of the Copyright Act, contact the Copyright Clearance Center (CCC), 222 Rosewood Drive, Danvers, MA 01923, tel: 978-750-8400, www.copyright.com. Request for special permission or bulk copying should be addressed to Reprints/Permission Department. INDEXED by Applied Mechanics Reviews and Engineering Information, Inc. Canadian Goods & Services Tax Registration #126148048

221

229

238

M. E. Bange A. J. Beaudoin
Department of Industrial and Mechanical Engineering, University of Illinois at Urbana-Champaign, Urbana, IL 61801

Predictive Modeling of the Nonuniform Deformation of the Aluminum Alloy 5182

We pose an experimental model for hot deformation that, in complexity, falls between a homogeneous laboratory test and an industrial process. Our objective is to document a transition between two distinct modes of thermally-activated deformation: diffusion controlled solute drag and hardening with concurrent recovery of dislocations. We demonstrate that constitutive equations for plasticity, describing different regimes of dislocation kinetics and calibrated with a minimum of adjustable constants, can be incorporated into nite element analysis and used reliably to predict the mechanical response of a nonuniform body. S0094-42890000302-9

M. Stout S. R. Chen
Los Alamos National Laboratory, Los Alamos, NM 87545

S. R. MacEwen
Alcan International Ltd., Kingston, Ontario, Canada

Introduction

There exists considerable difference between the laboratory setting used for characterization of metals and the industrial plant where the resulting material models are applied. In the laboratory, it is possible to pose a thermomechanical history that elicits a particular microstructural response. In contrast, the typical industrial process will span a considerable range of conditions in deformation, time, and temperature space to achieve desirable properties. A thermomechanical history encompassing recovery, recrystallization, and work hardening will likely be encountered during hot rolling, for example. Coordination throughout the process is critical to achieve optimum properties. Taking aluminum alloys used in automotive and packaging applications as an example, critical demands on formability, strength, and surface nish are achieved through careful manipulation of casting, hot rolling, and heat treatment. It is only through understanding of the interaction between metallurgical actions at the scale of the microstructure and thermomechanical conditions dictated by the macroscale process pathway that target properties can be achieved. In the following, we pose a deformation model that, in complexity, falls somewhere between a homogeneous laboratory test and an industrial process. Our objective is to document a transition between two distinct modes of thermally-activated deformation: diffusion controlled solute drag and hardening with concurrent recovery of dislocations. A further objective is the validation of the constitutive model and nite element implementation. The Al-Mg alloy AA 5182 is studied, thereby providing a material with both microstructural complexity and industrial relevance. The end result is a test procedure that serves as a rigorous validation of the constitutive model and nite element implementation at a level of complexity that lies between homogeneous testing and industrial processing. The approach to the design and experimental evaluation of the inhomogeneous compression test as well as a general outline of this manuscript follows. The constitutive response of AA 5182 is detailed and two distinct regimes of high temperature deformation are identied. Description of the constitutive response is achieved through two physically-motivated relationships for plastic ow. A nite element formulation for the simulation of a viscoContributed by the Materials Division for publication in the JOURNAL OF ENGINEERING MATERIALS AND TECHNOLOGY. Manuscript received by the Materials Division April 13, 1999; revised manuscript received November 13, 1999. Associate Technical Editor: H. M. Zbib.

plastic deformation is cast and simulations are conducted to predict shape change for an inhomogeneous deformation. The resulting design offers a specimen shape in which the active deformation mechanism, by virtue of the rate sensitivity, creates variations in the deformed geometry. Experimental testing is undertaken using the design developed from the nite element simulations. Detail is also given to procedures necessary to minimize both friction effects and thermal gradients. The deformation is characterized by the shape of the sample after compression and by optical metallography. Comparison is drawn between the experimental and simulation results. Finally, the implications of this work are discussed with regard to the signicance of inhomogeneity inherent in the experimental design and the predictive capability of the numerical simulations.

Constitutive Response of AA 5182

The aluminum alloy AA 5182 4.5 Mg, 0.35 Mn Hatch 1 has been studied using compression testing at constant strain rate Chen et al. 2, Stout et al. 3. It was found that the strain rate sensitivity ranges from a value of 1/3 a relatively high rate sensitivity at high temperatures and low strain rates to a small negative value at room temperature. As the experimental observations and resulting constitutive formulation are key to the present research, this prior work will be briey reviewed. At low stresses, plastic ow is diffusion controlled through solute drag SD. In this regime, the stress is characterized by the relationship A

b3 QD exp kT RT

(1)

where T is the absolute temperature, b is the Burgers vector, k is Boltzmans constant, and R is the universal gas constant. Q D is the activation energy of 131 kJ/mol, a handbook value for selfdiffusion of magnesium in aluminum, and A is an adjustable constant. The temperature dependent shear modulus, , is given by the empirical relation Varshni 4

D T0 exp 1 T

(2)

where 0 , D, and T 0 are tted parameters. Shown in Fig. 1 are yield stress data taken from uniaxial compression tests. The stress-strain rate relationship embodied in Eq. 1 is shown as a dashed line. Below a stress level of APRIL 2000, Vol. 122 149

Journal of Engineering Materials and Technology Copyright 2000 by ASME

 ,T 1 S

kT 0 ln b 3g 0

1/q 1/p

(5)

In Eq. 3, a represents an athermal contribution and the following two terms, S i and S , provide thermal components of the ow stress arising from dislocation interactions with obstacles. S i and S contribute to yield and hardening response of the stress, respectively. The evolution of the state d / d takes the form Bronkhorst et al. 7

0 1 0 s 0

s
(6)

where s is a function of temperature and strain rate,

Fig. 1 Yield stress data for AA 5182 plotted in the appropriate manner for solute drag. The line, n 3, represents a strain rate exponent of 13. Deviation from the dashed line provides an indication of an active mechanism other than solute drag.

s s0 0s

kT / b 3 g 0 s

(7)

( / ) 4 10 3 , the trend in the data demonstrates a similar power-law behavior with a stress exponent close to 3 solid line. Beyond this threshold stress level, however, there is an apparent transition away from solute drag. Increasing the stress, through reducing the temperature and/or increasing the strain rate, leads to plastic ow with thermallyactivated hardening and recovery HR. There is a corresponding reduction in the strain rate sensitivity. Such a regime is accurately represented by the Mechanical Threshold Stress MTS model Follansbee and Kocks 5. The evolution of material state in the HR regime is described through a single scalar variable. Only a very brief outline of the MTS model will be provided here, more detail may be found in references Follansbee and Kocks 5, Chen and Gray 6, Chen et al. 2. The ow stress, , at constant structure is a function of temperature T, strain rate , and an internal variable which characterizes the structure, . The ow stress is given by

The parameters for the model were tted for AA 5182 over an appropriate range of strain rates and temperatures Chen et al. 2. These parameters are listed in Table 1. While the number of parameters is at rst sight a bit disturbing, it must be noted that many of the parameters are relatively xedas dictated by the physical motivation behind the MTS model. As an example, the parameters p i , p , q i , and q follow from a phenomenological description of the dislocation glide resistance prole and are limited to the range Kocks et al. 8. 0 p i , p 1; 1 q i , q 2 (8)

a i S i ,T S ,T 0 0
where

(3)

,T 1 S i and

kT 0i ln b 3g 0i

0 , and s 0 are typically in the range of The terms 0i , 106 107 s1. Parameters a , i / 0 , and g 0 i are readily determined from simple linear regression applied to a so-called Fisher plot of yield stress data MacGregor and Fisher 9. With regard to the forthcoming simulation procedures, an approach must be adopted to transition between constitutive relations for solute drag and HR. For a given temperature and strain rate the active mode of deformation is determined using Eqs. 1 and 3. Because of the high rate sensitivity, application of the power law outside of the regime of solute drag quickly leads to very high stress levels dashed line of Fig. 1. Hence, the mechanism presenting the lower stress is adoptedan algorithmically easy choice. The state variable, is only evolved when HR is the active deformation mechanism. Finally, it should be noted that dynamic strain aging is observed for conditions

1/q i 1/p i

(4)

kT 10 7 s 1 3 ln b

2/3

0.32

(9)

Table 1 Parameters for mechanical threshold stress model

a
0i g 0i pi qi

3440 MPa 215K exp 1 T

0
i 0 0
g 0 p q

2.8815 104 MPa

k b3

0.5899 MPa/K

10 MPa 1 107 s 1 1.196 1/2 3/2

0.010315 1 107 s 1 1.6 2/3 1

s0
0s g 0s

1996 MPa 1 107 s 1 0.1058 6800 MPa 2

150 Vol. 122, APRIL 2000

Transactions of the ASME

Constitutive response is characterized by negative rate sensitivity and jerky ow. This deformation mechanism is not addressed in the present work.

Design of the Test Specimen

Initially, an axisymmetric test specimen was postulated wherein cross sections along the compression axis would have different diameters Fig. 2. This difference would lead to a gradient in strain rate through the specimen necessary to maintain equilibrium. The notion is that a balance of axial force between the end larger and center cross sections would reect the rate sensitivity associated with the underlying mode of deformation. For example, the high strain rate sensitivity associated with solute drag would lead to a relatively smaller difference in the compressive deformation rate between cross sections as compared to the HR regime. A similar approach has been used by Lalli and DeArdo 10 to investigate the effects of hardening evolution in the deformation of pure aluminum. Prior homogeneous compression tests demonstrated a transition in the dominant kinetics at a temperature of roughly 400C Chen et al. 2. At this temperature, a strain rate of 10 3 s1 resulted in a stress and strain-rate sensitivity indicative of solute drag Fig. 1. Similarly, a strain rate of unity placed material response rmly within the HR regime characterized by the MTS model. The constitutive models described above were introduced into a hybrid nite element formulation with a viscoplastic description of material constitutive response an outline of the formulation is given

in Appendix A. Numerous simulations were carried out in the course of designing the test specimen. Geometries with both linear and curved transitions from the end larger diameter to the center diameter were considered. All specimen designs left some material at the ends with a constant cross section. The temperature in the simulations was specied as 400C and velocity boundary conditions were manipulated to place constitutive response in either the realm of solute drag or HR. The nal choice for specimen geometry consisted of a constant radius transition between the end and center diameters Fig. 3A. Four dimensions were parametrically varied to arrive at the specimen design: the end diameter, the diameter of the reduced middle section, the radius of curvature between these two sections, and the total height of the specimen. For the conguration adopted, the overall height of the specimen was 12.5 mm and the end diameter was 10 mm Fig. 4. A reduced section, centered axially along the specimen, had a diameter of 4 mm. The radius of curvature between the end diameter and reduced section was 4 mm. Associated with this geometry are cross-head velocities that render deformation within range of the desired model kinetics solute drag Fig. 3B or the MTS model Fig. 3C. In both gures the stress invariant

II

3 : 2

(10)

is shown, where is the deviatoric stress tensor and : indicates the inner product. Constant crosshead velocities of 0.025 mm/s and 25.0 mm/s lead to specimen stresses in the domain of Eqs. 1 and 3, respectively.

Inhomogeneous Compression Tests

Fig. 2 Simple notion that the deformed geometry at two distinct cross sections 1 and 2 will reect the strain rate sensitivity m , following from force balance

4.1 Test Procedure. An experimental procedure was developed for the elevated temperature inhomogeneous compression tests. This rst series of tests was performed on an Instron model 1331 servo-hydraulic machine equipped with a resistive furnace AMTEL Laboratory at the University of Illinois at UrbanaChampaign. A second series of tests was performed on a MTS model 880 servo-hydraulic machine Los Alamos National Laboratory, Materials Science and Technology Division. The second set of experiments used cartridge heaters located in the ends of metal platens in order to achieve testing temperatures. The AA 5182 specimens were turned from the center of a 25 mm thick plate and were oriented such that the compression axis was parallel to the rolling direction. Load rings were also machined in both planar ends of this specimen Detail A of Fig. 4. The load

Fig. 3 Effective stress from nite element simulations plotted in units of lnII : A undeformed specimen; B solute drag regime SD; C thermally activated hardening and recovery HR

Journal of Engineering Materials and Technology

APRIL 2000, Vol. 122 151

Fig. 4 Final specimen geometry based on simulations

ring creates a recess on the ends of each specimen which is lled with lubricant Rastegaev 11. The lubricant used in these tests was a mixture of vacuum grease and colloidal graphite 2:1, respectively. For consistency between experiment and simulation procedure, the presence of a thermal gradient in the specimen must be avoided. An assessment of thermal gradients in the specimen was undertaken during the rst series of tests. Chromel-alumel thermocouples were placed along the compression axis in the center of the specimen at three locations Fig. 5. In order to reduce heat conduction from the specimen to load frame, alumina platens were placed on either side of the test specimen. The furnace control thermocouple was spot welded to a metal washer which was located on the upper surface of the bottom alumina platen. A temperature controller was programmed to ramp the resistive furnace to a test temperature of 400C. The program consisted of a ramp to 200C with a short soak followed by a ramp to 400C. The plot of the specimen temperature is shown in Fig. 6. It can be see from the plot that no overshoot of the testing temperature was observed. Also, the specimen remained within 2C of the test temperature for over six minutes. After each test, the specimens were quenched in water. The specimen entered the quench bath in under three seconds. Corresponding to the simulations, all inhomogeneous compression tests were performed using constant crosshead velocity. Crosshead velocities of v z 0.025 mm/s and v z 25.0 mm/s placed specimen stress in the SD and HR regimes, respectively. Load and crosshead displacement data were collected for all tests.

Fig. 6 Specimen temperature recorded at the three thermocouple positions shown in Fig. 5

4.2 Experimental and Numerical Results. The numerical simulations resulted in two distinct deformed shapes for the SD and HR regimes Figs. 3B and 3C, respectively. The diameters of the center cross section from the two simulations are recorded in Table 2. For the experimental tests, the stock material is textured Stout et al. 2 so some ovaling of the compression sample is expected Beaudoin et al. 12, Kalidindi and Anand 13. Major and minor diameters of the center cross section are listed for the experimental samples. Finally, the ratio of diameters associated with HR and solute drag is given for each pair of measurements. The simulationfor the most partunderpredicts the measured dimensions. Yet, the ratio of the measured diameter for HR to that for solute drag shows a common trend between the simulations and experimental tests Table 2. Shadow graphs of the outline of each specimen were constructed in order to provide rigorous comparison between deformation in the two constitutive regimes. This process involves back-lighting each specimen, focusing the image on a screen, and digitally scanning the image. The difference in deformation in the center region is clearly rendered in Figs. 7A and 7B for both the simulation and experimental results, respectively. This trend was repeated in all shadow graphs taken with the major and minor diameters, as well as the 45 deg bisector. The shadow graphs of the experimental specimens were overlayed with the side proles from the simulations Fig. 8. Some ne details in the experimental tests are not fully rendered by the simulations. The model underpredicts the deformation of the curved transition between the major and minor diameters. This is possibly due to neglecting the load ring geometry in the nite element simulations. It is likely that deformation of the load ring at the beginning of the test enhanced deformation near the platen/sample interface upper and lower portions of the sample in Fig. 8. This may be a

Table 2 Measured diameter mm of the center bulged cross section from simulation and experiment. Measurements on the experimental specimens were performed using a machinist microscope. Experiment Simulation Hardening and recovery ( D HR ) Solute drag ( D SD ) Ratio ( D HR / D SD ) 7.619 7.271 1.048 Major 8.133 7.744 1.050 Minor 7.608 7.287 1.044 Average 7.871 7.516 1.047

Fig. 5 Thermocouple positions and general experimental setup used for temperature verication drawing not to scale

152 Vol. 122, APRIL 2000

Transactions of the ASME

result of over-compensation for the effects of friction. That is, the hydrostatic pressure of the lubricant increases the radial force on the load ring. The load-displacement response is shown in Fig. 9. Displacement data collected from experimental tests was shifted 0.508 mm in order to correct for the deformation of the load ring geometry. This value is equal to the total height of the load ring. The simulation result slightly underpredicts the load developed in SD Fig. 9. A slightly greater rate of hardening appears for the simulation, as compared to experiment, when HR is the active mechanism. However, the overall comparison is favorable. Control of friction

Fig. 7 Comparison of the deformed geometry: A for the simulations and B for the experiments. Shadow graphs shown in B contain the minor axis of the center section.

Fig. 8 Comparison of the deformed geometry from the experiment and simulation for solute drag and HR

Fig. 9 Comparison of load-displacement curves between the simulation and the experiment for both HR and SD regimes

Fig. 10 Micrograph of specimen deformed in the solute drag regime with ram velocity of v z 0.025 mms: A unrecrystallized structure with compression axis oriented vertically; B magnication of bulged section with compression axis oriented horizontally; C further magnication of the bulged region, with compression axis oriented horizontally, showing serrated grain boundaries

Journal of Engineering Materials and Technology

APRIL 2000, Vol. 122 153

appearance than grains in the upper section, reecting a greater degree of plastic strain in the specimen center. A rather atypical feature lies in the appearance of the grain boundaries Fig. 10C. A coarse subgrain structure with planar arrays impinging on the grain boundaries creates serrations. In marked contrast, deformation at v z 25.0 mm/s led to a recrystallized microstructure Figs. 11A and 11B. The stress state predicted by the simulation is reected in the recrystallized grain structure. Grains in the center region are ner and reect the history of locally elevated stress. The microstructure in this region is equiaxed Fig. 11C. The experimental conditions chosen for testing in the HR regime encourage recrystallization. For deformation taking place at 411C and a strain rate of 10.2 s1, the time needed for 50 percent recrystallization is 2.5 seconds Wells et al. 14. While every attempt was made to quench quickly after the completion of the inhomogeneous compression tests, the time to quench was greater than the rapid recrystallization times noted in the referenced work. We surmise that the microstructure shown in Fig. 11 follows from static recrystallization after completion of the test.

Discussion

Fig. 11 Micrograph of specimen deformed in the hardening and recovery regime with a ram velocity of v z 25 mms: A recrystallized structure with variation in grain size along vertical compression axis; B magnication of bulged section with compression axis oriented horizontally and revealing ne grains along regions of shear; C further magnication of the bulged region, with compression axis oriented horizontally

effects in the experiment and accurate representation of the constitutive response in the nite element model are indicated by the load-displacement results. 4.3 Metallographic Analysis. Metallographic samples were prepared by making a longitudinal cut through the specimen, at roughly 45 deg to the major and minor axes. A Barkers reagent was used as an etchant. Optical metallographs are shown in Figs. 10 and 11. In the solute drag regime, grains take on a general aspect ratio that reects the deformation history corresponding to location in the sample. Grains in the central section have a more elongated 154 Vol. 122, APRIL 2000

The key result of this work is the development of an experimental test that highlights the constitutive response of a solutionstrengthed aluminum alloy. As a furtherand equally importantobjective, the test regime exercised both a constitutive model for viscoplastic response and implementation of that model in a nite element code. Predictive capability of the computer code was demonstrated both in the use of the implementation as a design tool and subsequent validation through experiment. Clearly, many repeated experimental iterations would be required to settle on a specimen design that gave a clear indication of the dominant mechanism of deformation. Use of the nite element model as a design tool was indispensable in development of the test procedure. Uniaxial compression tests were used to develop the constitutive response Chen et al. 2. The original concept of the present experiment followed a line of thought based on these homogeneous tests. That is, to maintain equilibrium, the compressive strain rate would be different in specimen cross sections of different diameter. This logic assumes that the shearing in the specimen is relatively small, and is reected to some degree in the deformed mesh of Fig. 3B. However, a similar situation does not carry over to the case of deformation in the HR regime. From close inspection of Fig. 3C, distortion in the mesh indicates that shearing is prevalent in proximity to the specimen center. The variation in recrystallized grain size, within the bulged region of Fig. 11B, emphasizes the non-uniformity of deformation. Simply put, the deformation detailed by the simulation was rather more complex than the simple stacking of homogeneous deformations that posed the original conception of the test. Finally, the approach to capturing the constitutive response through distinct relations in accordance with distinct mechanisms warrants some comment. As noted above, for solute drag Eq. 1 involves a single free parameter, A. For the single temperature consideredand when HR is the active mechanismevolution of state, , is relatively modest. Stress response is dominated by the rst two terms of Eq. 3 and not the last term involving S . In the HR regime, the free parameters a , g 0 i , and i follow from use of a Fisher plot MacGregor and Fisher 9, Kocks et al. 8. Other parameters are xed for the most part. Accurate modeling of the constitutive response follows from a physicallymotivated description, not a general curve t. Looking toward a more general application, one needs to consider the handling of the state variable with excursions of stress between HR and solute drag. Jump tests for a sequence of strain rates from 1 s1 to 10 3 s1 and a return to 1 s1 show that a steady-state value of the ow stress is achieved after a slight transient Fig. 12. This behavior, rapid establishment of a ow stress Transactions of the ASME

where are weighting functions. The constitutive response is written as

2 eff , T D

(14)

where T is the temperature, D is the deviatoric deformation rate tensor, and is taken as the second invariant of the deformation rate,

2 D : D 3

(15)

The effective viscosity, eff , is evaluated according to the appropriate regime of behavior dictated by the stress, as described above. A weighted residual is formed on Eq. 14 giving

1 : dV 0 2 eff

(16)

To complete the formulation, the constraint

Fig. 12 Strain rate jump test at 400C with strain rates of 1, 10 3 , and 1 s1 on homogeneous compression specimens

ui 1 p i 1 p i

(17)

with little hardening, is central to the selection of 400C as a test temperature. Yet, recovery of prior hardening is indicated by the transient present in the second jump from 10 3 s1 to 1 s1. The HR model is dependent on the concept of a state variable that represents the resistance to further deformation; the SD model is not, depending primarily on the amount of mobile solute in solid solution. A general application of the constitutive formulation adopted herein must consider evolution of the state variable, , for excursions into the regime of solute drag. The current work also does not model plastic ow which occurs in the regime of dynamic strain aging high stress developed at lower test temperatures. Future efforts will be aimed at the incorporation of a static recovery term into the evolution Eq. 6. Such a step will further the applicability of the constitutive model to a broad base of industrially relevant problems.

Acknowledgments
This work was sponsored by NSF Career Award DMII-75154, by a grant from Alcan International, Ltd., and with support from U.S. Department of Energy, Basic Energy Sciences, Division of Materials Sciences, Center for Synthesis and ProcessingMetal Forming Project. Conversations with Prof. Paul R. Dawson prompted consideration of thermal aspects of the test. Finally, the careful attention paid by Dr. Peter Kurath, Mr. Manuel Lovato, and Dr. George Kaschner to the experimental tests is sincerely appreciated.

is employed where convergence of the pressure in the i 1 iteration provides near incompressible response Zienkewicz et al. 15. Trial functions are introduced into the residuals for the interpolated eld variables. An eight node brick was employed with interpolation of the continuous linear velocity eld, piecewise linear deviatoric stresses, and piecewise constant pressure. Subsequently, the discretized form of Eq. 16 may be inverted and the result used to solve for the deviatoric stress in Eq. 13. A parallel implementation was used to conduct the simulations. All communication is effected through the Message Passing Interface MPI Gropp et al. 16. Global data structures from assembly of the nite element system of equations are contained in parallel objects. The maintenance of these objects through time stepping of the solution is enabled through the Portable Extensible Toolkit for Scientic Computing PETSc Balay et al. 17. The object-oriented approach enables access to a variety of Krylov subspace methods and preconditioners. The conjugate gradient method combined with Jacobi preconditioner was applied to these simulations.

References
1 Hatch, J. E., 1984, Aluminum: Properties and Physical Metallurgy, American Society For Metals, Metals Park, OH, p. 353. 2 Chen, S. R., Kocks, U., MacEwen, S., Beaudoin, A. J., and Stout, M. G., 1998, Constitutive Modeling of a 5182 Aluminum as a Function of Strain Rate and Temperature, Hot Deformation of Aluminum Alloys II. T. R. Bieler et al., eds., The Minerals, Metals & Materials Society, Warrendale, PA, pp. 205 216. 3 Stout, M. G., Chen, S. R., Kocks, U. F., Schwartz, A. J., MacEwen, S. R., and Beaudoin, A. J., 1998, Mechanisms Responsible for Texture Development in a 5182 Aluminum Alloy Deformed at Elevated Temperature, Hot Deformation of Aluminum Alloys II. T. R. Bieler et al., eds., The Minerals, Metals & Materials Society, Warrendale, PA, pp. 243254. 4 Varshni, Y. P., 1970, Temperature Dependence of the Elastic Constants, Phys. Rev. B, 2, pp. 39523958. 5 Follansbee, P. S., and Kocks, U. F., 1988, A Constitutive Description of the Deformation of Copper Based on the Use of the Mechanical Threshold Stress as an Internal State Variable, Acta Metall., 36, pp. 8193. 6 Chen, S. R., and Gray, G. T., 1996, Constitutive Behavior of Tantalum and Tantalum-Tungsten Alloys, Metall. Trans. A, 27A, pp. 29943006. 7 Bronkhorst, C. A., Kalidindi, S. R., and Anand, L., 1992, Polycrystal Plasticity and the Evolution of Crystallographic Texture in Face Centered Cubic Metals, Philos. Trans. R. Soc. London, Ser. A, 341, pp. 443477. 8 Kocks, U. F., Argon, A. S., and Ashby, M. F., 1975, Thermodynamics and Kinetics of Slip, B. Chalmers, J. W. Christian, and T. B. Massalski, eds., Progress in Materials Science, Vol. 19, Pergamon Press Ltd., Oxford, England, p. 142. 9 MacGregor, C. W., and Fisher, J. C., 1946, A Velocity-Modied Temperature for the Plastic Flow of Metals, ASME J. Appl. Mech., 13, pp. 1116. 10 Lalli, L., and DeArdo, A., 1990, Experimental Assessment of Structure and Property Predictions During Hot Working, Metall. Trans. A, 21A, pp. 3101 3114.

Appendix A: Finite Element Formulation

Neglecting inertial effects and body forces, the motion of the body is described through 0 n t u u in B on B on Bu (11)

Partitioning the deviatoric, , and hydrostatic, p, components of the Cauchy stress as

p I

(12)

then forming a weighted residual in Eq. 11 and applying the divergence theorem gives the equilibrium statement

: dV

p dV

dS t

(13)

Journal of Engineering Materials and Technology

APRIL 2000, Vol. 122 155

11 Rastegaev, M. V., 1940, A New Method of Homogeneous Compression of Specimens for Determining Flow Stress and the Coefcient of Internal Friction, ZADOVSK Lab., Vol. 6, p. 345. 12 Beaudoin, A. J., Mathur, K. K., Dawson, P. R., and Johnson, G. C., 1993, Three-Dimensional Deformation Process Simulation with Explicit Use of Polycrystal Plasticity Models, Int. J. Plast., 11, pp. 501521. 13 Kalidindi, S. R., and Anand, L., 1994, Macroscopic Shape Change and Evolution of Crystallographic Texture in Pre-Textured fcc Metals, J. Mech. Phys. Solids, 42, pp. 459490. 14 Wells, M. A., Lloyd, D. J., Brimacombe, I. V. S. J. K., and Hawbolt, E. B., 1998, Modeling the Microstructural Changes During Hot Tandem Rolling of

aa5xxx Aluminum Alloys: Part 1. Microstructural Evolution, Metall. Mater. Trans. B, 29B, No. 3, pp. 611620. 15 Zienkewicz, O. C., Vilotte, J. P., and Toyoshima, S., 1985, Iterative Method for Constrained and Mixed Approximation and Inexpensive Improvement of fem Performance, Comput. Appl. Mech. Engrg., 53, pp. 329. 16 Gropp, W., Lusk, E., and Skjellum, A., 1994, USING MPI: Portable Parallel Programming with the Message-Passing Interface, The MIT Press, Cambridge, MA. 17 Balay, S., Curfman-McInnes, L., Gropp, W. D., and Smith, B. F., 1995, PETSc 2.0 Users Manual ANL Report ANL-95/11 ed., Argonne National Laboratory, Argonne, IL.

156 Vol. 122, APRIL 2000

Transactions of the ASME

Radovan Rolovic1
Research Associate

Multiaxial Cyclic Ratcheting in Coiled TubingPart I: Theoretical Modeling

Coiled tubing is a long, continuous string of steel tubing that is used in the oil well drilling and servicing industry. Bending strains imposed on coiled tubing as it is deployed and retrieved from a well are considerably into the plastic regime and can be as high as 3 percent. Progressive growth of tubing diameter occurs when tubing is cyclically bentstraightened under constant internal pressure, regardless of the fact that the hoop stress imposed by typical pressure levels is well below the materials yield strength. A new incremental plasticity model is proposed in this study that can predict multiaxial cyclic ratcheting in coiled tubing more accurately than the conventional plasticity models. A new hardening rule is presented based on published experimental observations. The model also implements a new plastic modulus function. The predictions based on the new theory correlate well with experimental results presented in Part II of this paper. Some previously unexpected trends in coiled tubing deformation behavior were observed and correctly predicted using the proposed model. S0094-42890000402-3

Steven M. Tipton
Professor Department of Mechanical Engineering, University of Tulsa, Tulsa, OK 74104

Introduction
When structural components are cyclically loaded in the plastic regime, progressive plastic deformation can occur under a combination of a primary steady loading and a secondary cyclic loading. In cases where maximum load levels induce plastic deformation, small amounts of plastic strain are sometimes not recovered during each cycle, which can lead to large accumulated plastic strains. This can occur even when the primary load levels are well within the elastic regime. This phenomenon is called cyclic ratcheting. Coiled tubing undergoes such loading conditions in routine service. A schematic layout drawing of the coiled tubing deployment equipment is shown in Fig. 1. The operational concept of the coiled tubing system involves running a continuous string of steel tubing into a well to perform a specic operation. When the operation is complete, the tubing is retrieved from the well and spooled onto its reel for storage and transport. Throughout its life, coiled tubing endures severe cyclic deformation imposed by the above-surface deployment hardware, high internal pressure, and axial forces. Due to tube dimensions and curvatures of the spool and guide arch, the tubing is plastically deformed every time it is deployed and retrieved from a well, with bending strains as high as 3 percent. The imposed internal pressure can generate hoop stresses on the order of 50 percent of the materials monotonic yield stress. When the tubing is cyclically bent under imposed internal pressure, a progressive growth of tubing diameter occurs. In some cases the tubing diameter can increase as much as 30 percent before a tubing failure occurs. Severe diametral growth can indicate the onset of fatigue failure and can cause other problems, such as tubing incompatibility with implementation hardware. A sophisticated multiaxial cyclic plasticity algorithm is needed to model the elastic-plastic behavior of coiled tubing. Investigation of plastic deformation of engineering materials under multiaxial states of stress has mainly been conned to situations where the load history is relatively short and simple. In these cases, a small inaccuracy in the theoretical representation of observed
1 Currently at Schlumberger Technology Corporation, Sugar Land, Texas, 77478 e-mail: rolovich2@slb.com. Contributed by the Materials Division for publication in the JOURNAL OF ENGINEERING MATERIALS AND TECHNOLOGY. Manuscript received by the Materials Division November 19, 1998; revised manuscript received September 21, 1999. Associate Technical Editor: H. Sehitoglu.

physical phenomena usually results in satisfactory predictions. However, when the load history is more complex or repeated many times, inaccuracies can accumulate and nal results can differ signicantly from physical observations. This inconsistency between theoretical predictions based on the existing plasticity models and physical observations is strongly manifested in the case of multiaxial cyclic ratcheting in coiled tubing. Conventional plasticity models based on kinematic hardening rules proposed by Prager 1, Ziegler 2, Armstrong and Frederic 3, Mroz 4, Phillips and Lee 5, Garud 6, and Tseng and Lee 7, cannot accurately describe multiaxial cyclic ratcheting. Tipton 8 showed that the Mroz hardening rule signicantly overestimates the diametral growth of coiled tubing, especially for low and moderate hoop to yield stress ratios. Bower 9, Chaboche 10, Hassan et al. 11, Ohno and Wang 1214, Moreton 15, and Jiang and Sehitoglu 1618 also reported that conventional plasticity models tend to overestimate cyclic ratcheting rates. Recently, Bower 9, Chaboche 10, Ohno and Wang 1214, and Jiang and Sehitoglu 17,18 modied the nonlinear kinematic hardening model by introducing new terms to the basic Armstrong-Frederic model. The ability of these new versions of the nonlinear kinematic model to describe multiaxial cyclic ratcheting improved, but the complexity and the number of experimental parameters increased signicantly. Consideration of these factors led to a new incremental plasticity model for multiaxial cyclic deformation of coiled tubing.

New Plasticity Model

A yielding criterion is needed to dene the onset of plastic ow during an increment of loading. When the von Mises yield criterion is assumed, the kinematic yield function can be expressed as 2 F 3/2( S ):( S ) Y 0, where S is the deviatoric stress tensor, represents the deviatoric quantity of the back stress tensor center of the yield surface, and Y is the yield strength in uniaxial tension. A bold letter denotes a Cartesian tensor. A colon between two second-order tensors denotes their scalar product. A ow rule relates the plastic strain increment to the current state of stress and to the stress increment. The generally accepted ow rule, the normality ow rule, proposed by Prandtl 19 and later generalized by Reuss 20, can be stated as d p 1/H n:d n, where d p and d are the tensors of incremental plastic strain and stress, H is a proportionality factor called the APRIL 2000, Vol. 122 157

Journal of Engineering Materials and Technology Copyright 2000 by ASME

Fig. 1 Schematic layout drawing of coiled tubing deployment equipment

plastic modulus, and n is the unit normal to the yield surface. The MacCauley bracket, , is dened by x ( x x )/2. A general stress-strain relationship for plastic deformation must be in the incremental form because plastic deformation is history path dependent. Assuming the additive decomposition of the strain tensor, d d e d p , where d e is the elastic strain tensor and d p is the plastic strain tensor, and the normality ow rule, the general stress-strain relation can be written as d De 1/H nn :d , where De is the elastic compliance tensor, and nn is the dyadic product of two vectors, which is a special form of second-order tensor whose components are given as nn n n T . New Hardening Rule. A hardening rule describes the evolution of subsequent yield surfaces. Based on experimental observations, a new hardening rule is proposed in this study. Phillips and Tang 21, Phillips and Lee 5, and Phillips and Lu 22 conducted experiments on commercially pure aluminum, under both proportional and nonproportional, axial-torsional loading conditions. They found that the center of the yield surface tends to translate in a direction that lies between the stress increment d and the plastic strain increment d p . They also found that in most cases the motion of the yield surface is predominated by the stress increment d ; i.e., the center of the yield surface moves in the direction that is closer to d than to d p . Similar observations were reported by McDowell 23,24 for type 304 stainless steel under proportional and nonproportional, axial-torsional loading conditions. Based on these observations, a new hardening rule is proposed herein. The new hardening rule states that the motion of the center of the yield surface, , is given by d d

Fig. 2 Graphical representation of the new hardening rule

d d p q , d d p

(1)

where d is a scalar which can be determined from the consistency condition F ( d , d , Y d Y ) 0, q is a hardening parameter which, in general, is a function of material characteristics and loading, the norm of the second order tensor d is denoted by d d :d , and d p d p :d p . The graphical representation of the new hardening rule is shown in Fig. 2. When the normality ow rule is assumed, it follows that d p / d p n, where n is the normal unit vector to the yield surface, and the new hardening rule can be written as d d ( d / d q n). From the experimental observations mentioned above, it can be concluded that in most cases 0 q 1. Anisotropic distortion of yield surface, which manifests in an increased curvature of the yield surface in the direction of plastic loading, has been observed by many researchers, for instance by Phillips and Tang 21, Phillips and Lee 5, Phillips and Lu 22, and Eisenberg and Yen 25. This anisotropic distortion of yield surface appears to cause the normal to the yield surface, n, at the current stress point and the plastic strain increment, d p , to move closer to the direction of loading see Fig. 3. The direction of yield surface motion during plastic loading also becomes closer to the loading direction as the anisotropic distortion progresses. 158 Vol. 122, APRIL 2000

Thus, the hardening parameter q in Eq. 1 should change accordingly if an ideal, undistorted yield surface is assumed. Otherwise, small inaccuracies in the calculation of yield surface motion can accumulate and lead to large errors in the calculation of plastic strain. It is very difcult to examine and model the anisotropic distortion of yield surface, especially for long load histories. The inuence of anisotropic distortion of yield surface on the shift of yield surface motion toward the loading direction can be approximated by using an undistorted yield surface and decreasing the hardening parameter q in Eq. 1 as plastic deformation progresses. The change of the hardening parameter q can be modeled as a decreasing function of, for instance, the accumulation of equivalent plastic strain. If during the subsequent loading the stress increment impinges on a different, less distorted or undistorted part of the yield surface, some or all memory of previous anisotropic distortion can be lost, depending on how far the new loading point is from the center of the previously distorted region of the yield surface. This memory of anisotropic distortion of yield surface can be modeled by updating the accumulated equivalent plastic p p e e strain, , acc , if , acc is used to describe q when the stress increment impinges on a different part of the yield surface. This study primarily deals with tubular structures submitted to a combination of steady or quasi-steady internal/external pressure primary loading and a cyclic bending/axial loading secondary loading. In this case the hardening parameter q can be described as a decreasing function of the accumulation of equivalent plastic p strain, e , acc , and the memory of anisotropic distortion of yield p surface can be modeled by updating e , acc as the steady primary stress level, P , is changed. For coiled tubing applications the primary stress is the hoop stress, i.e., P h . Based on experimental observations of multiaxial cyclic ratcheting in coiled tubing, the general form of the hardening parameter q should be able to describe the behavior exhibited in Fig. 4.

Fig. 3 Schematic illustration of anisotropic distortion of yield surface and the corresponding shift of the associated normal to the yield surface towards the loading direction

Transactions of the ASME

Fig. 4 Hardening parameter q as a function of the primary p e stress P and , acc

In a mathematical form, a hardening parameter q that satises the above considerations can be expressed as q C q0 C q1
p 2 1 C q 2 1 C q 3 P / Y C q 4 e , acc

Fig. 6 Parameters used to determine the plastic modulus in this study

(2) ed according to experimental observations, which may not be possible if auxiliary yield surfaces are placed around the active yield surface. Plastic Modulus. In this study, the plastic modulus, H, is dened as a function of back stress . The plastic modulus, H, is related to the uniaxial plastic modulus, E p , as H 2/3E p . The Ramberg-Osgood equation for the uniaxial stress-plastic strain curve, K ( p ) n , can be used to calculate the uniaxial plastic modulus as E p d / d p , where K is the cyclic strength coefcient and n is the cyclic strain-hardening exponent. Thus, the plastic modulus, H, can be obtained as H 2 K n 3 K

where C q 0 , C q 1 , C q 2 , C q 3 , and C q 4 are material constants, and p e , acc is equivalent plastic strain accumulated at a primary steady or quasi-steady stress level P . The sum of C q 0 and C q 1 is used to dene the initial value of the hardening parameter q see Fig. 4. The constant C q 0 is also used to dene the lowest value of qwhen anisotropic distortion of yield surface has been saturated at the current primary stress level. The other three constants, C q 2 , C q 3 , and C q 4 , are used to dene the nonlinear reduction of q as the distortion of yield surface progresses. An explanation of the empirical determination of the material constants in Eq. 2 is given in Part II of this paper. When the primary stress level is changed by P relative to the previous load increment in the p plastic regime, e , acc is recalculated to account for previous anisotropic distortion of yield surface as
p p 2 e e , acc , acc i 1 exp 0.5 E / Y P / Y ,

n1 /n

(4)

(3)

p where ( e , acc ) i 1 is accumulated equivalent plastic strain up to the current load increment. The function given in Eq. 3, which is used to update accumulated equivalent plastic strain between two plastic loading steps, is shown schematically in Fig. 5. If there is no steady or quasi-steady stress, the primary stress P in Eq. 3 p changes considerably with each load step, resulting in e , acc 0 and q C q 0 C q 1 . Thus, when plastic loading is distributed around the yield surface, it may be sufcient to assume that the hardening parameter q is constant. It should be pointed out that the new hardening rule, given by Eq. 1, would not ensure nesting of the yield surface with the next inactive yield surface or limit surface in a multiple-surface or two-surface model. The new model uses a single yield surface to determine if plastic ow will occur for an increment of loading and the direction of plastic strain increment. The plastic modulus and material memory effects are modeled based on the back stress . This way, the motion of the active yield surface can be speci-

where is used instead of to account for material memory effects. In this study, is calculated based on the cyclic stressstrain curve whenever max and e e ,max as

3 2

(5)

and in all other cases based on the hysteresis curve of the material as

Y 0.5

3 , 2 in

(6)

where Y denotes the size of the yield surface, is the deviatoric back stress tensor the deviatoric location of center of the current is the yield surface, e is the effective von Mises stress, in deviatoric back stress at the beginning of load reversal see Fig. 6. The max subscript is used to indicate the maximum value is upof the corresponding quantity during previous loading. in ) every time the loading is changed from elastic to dated ( in plastic, i.e., when F ( ) 0 is followed with F ( ) 0 and F / :d 0, where F ( ) is the yield function. A new reversal of loading begins when elastic unloading occurs, i.e., when F ( ) 0 and F / :d 0. Thus, once plastic deformation has been initiated, a load reversal extends as long as the deformation stays plastic.

Analysis of Service Loading Applied to Coiled Tubing

Deformation of coiled tubing caused by bending loading is strain-controlled because the tubing is bent over a cylinder or an arch with a constant radius of curvature. Therefore, the bending strain history in the axial direction of tubing, b x , is known, based on the assumption that plane transverse sections of coiled tubing remain plane. This assumption has been supported experimentally by Newburn 26. Applied internal pressure will produce stresses APRIL 2000, Vol. 122 159

Fig. 5 Accumulated equivalent plastic strain as a function of the primary stress level change

Journal of Engineering Materials and Technology

the equilibrium condition in the axial direction. In practice, an identical axial strain a x is applied to every section around the circumference in order to establish axial equilibrium.

Discussion
The plastic modulus function, as described in Eq. 4, accurately describes cyclic material responses for the load histories examined in this study. However, the characterization of the material memory by only using Eqs. 5 and 6 can result in overestimating the plastic modulus for certain load histories. For these load histories, a more robust description of the plastic modulus may be employed. For instance, a more rigorous set of rules with additional memory factors can replace Eqs. 5 and 6. The introduction of the function q in the new hardening rule is based on observations of data from multiaxial yield surface probing experiments conducted on aluminum and steel tubular samples in axial-torsional and axial-bending-internal pressure loading. Specic multiaxial experiments are required to identify the parameters dening the hardening function q, but these can be standardized for particular applications. With coiled tubing, for example, diametral growth data from simple constant pressure, cyclic bending tests were used to identify parameters in Eq. 2 empirically. The new hardening rule and plastic modulus function did an excellent job correlating with diametral growth measurements, as will be discussed in Part II of this paper. The approach just described, with its parameters as identied from simple testing, was used by Rolovic 27 to make predictions of tubing behavior under complex combinations of bending, axial force, and internal pressure. Again, good correlation was demonstrated with experimental results. The routine made predictions of coiled tubing behavior that were previously unobserved and unexpected. These include diametral contraction during bending cycles with positive internal pressure and axial contraction due to bending cycles following axial force applications. This mechanical behavior was subsequently substantiated experimentally. The routine also predicted accurately that slightly greater wall thinning occurs at the compressive wall relative to the tensile wall, a trend previously noted in literature. Diametral shrinkage and wall thinning data and predictions are presented and discussed in Part II.

Fig. 7 Tubing geometry and the orthogonal coordinate system for an element of coiled tubing x axial, h hoop, and r radial direction

in the hoop and radial direction of tubing. Thus, hoop stress, h , and radial stress, r , histories are assumed to be known. In this study, in order to decrease the computational time, coiled tubing deformation behavior is modeled based on the components of stress and strain calculated only as their average quantities at the midthickness radius of tubing. When tubing is constrained onto a circular arch, the magnitude of bending strain in the axial direction at the midthickness radius of tubing is given by

b x

Dt cos , 2Rb

(7)

where D is the outer diameter of tubing, t is the wall thickness, R b is the radius of curvature of the bending neutral plane of tubing, and denotes the angular position around the circumference of tubing relative to the top midthickness point on the convex side of tubing see Fig. 7. When tubing is subjected to internal pressure, p, hoop and radial stresses are developed. Hoop stress at the midthickness of tubing can be estimated fairly accurately as average stress obtained from static equilibrium as

h p

D2t . 2t

(8)

Since Eq. 8 is derived from equilibrium consideration, it can be used for both elastic and plastic deformation. The stress in the radial direction of tubing is caused by internal pressure and varies from p on the inside wall of tubing to zero on the outside wall. The radial stress distribution through the wall thickness is nonlinear and different for elastic and plastic deformation. The inuence of radial stress on plastic deformation of coiled tubing is much smaller than the inuence of hoop and axial stress. Hence, radial stress at the midthickness radius of tubing for both elastic and plastic deformation can be approximated as

Summary
A single surface incremental plasticity model has been developed. The model is specialized for the prediction of multiaxial cyclic ratcheting of tubular structures submitted to pressure and bending/axial loading. A new hardening rule and plastic modulus function are implemented in the model. The hardening rule is based on observations of experimental data available in the literature. The model was used to successfully describe the mechanical behavior of coiled tubing under complex loading. The model made more accurate predictions of coiled tubing diametral growth than conventional plasticity theory and predicted a diametral shrinkage phenomenon that had never before been observed. This observation was subsequently validated experimentally and is presented in Part II of this paper.

r 0.5p .

(9)

When the ratio of the wall thickness to the tubing diameter is small thin-walled tubing, Eq. 9 is fairly accurate. Based on the tubing geometry, bending radius, and applied internal pressure, the input loading b x , h , and r can be calculated from Eqs. 7 9. The three remaining stress and strain components ( x , h , and r ) can be obtained from the general stress-strain relationship given in the introductory part of the previous section. The axial force resulting from the pressure on the closed ends of coiled tubing is computed by F x,p p
D 2 t 2 . 4

Acknowledgments
The authors express appreciation for support of this work to The University of Tulsa, The United States Department of Energy, and members of the University of Tulsa Coiled Tubing Mechanics Research Project: Elongation and Diametral Growth.

(10)

References
1 Prager, W., 1955, The Theory of Plasticity: A Survey of Recent Achievements, Proc. Inst. Mech. Engr., 169, pp. 4157. 2 Ziegler, H., 1959, A Modication of Pragers Hardening Rule, Q. Appl. Math., 17, pp. 5565. 3 Armstrong, P. J., and Frederic, C. O., 1966, A Mathematical Representation of the Multiaxial Bauschinger Effect, C.E.G.B., Report RD/B/N 731.

This axial force, together with any externally applied axial force, will cause additional uniform strain, a x , in the axial direction of tubing. In general, axial strain a x cannot be determined directly from the applied axial load. It has to be calculated iteratively from 160 Vol. 122, APRIL 2000

Transactions of the ASME

4 Mroz, Z., 1967, On the Description of Anisotropic Workhardening, J. Mech. Phys. Solids, 15, pp. 163175. 5 Phillips, A., and Lee, C., 1979, Yield Surfaces and Loading Surfaces. Experiments and Recommendations, Int. J. Solids Struct., 15, pp. 715729. 6 Garud, Y., 1981, A New Approach to the Evaluation of Fatigue Under Multiaxial Loadings, ASME J. Eng. Mater. Technol., 103, pp. 118125. 7 Tseng, N. T., and Lee, G. C., 1983, Simple Plasticity Model of Two-Surface Type, Journal of Engineering Mechanics, 109, No. 3, pp. 795810. 8 Tipton, S. M., 1995, Multiaxial Plasticity and Fatigue Life Prediction in Coiled Tubing, Fatigue Lifetime Predictive Techniques: 3rd Volume, ASTM STP 1292, M. R. Mitchell and R. W. Landgraf, eds., American Society for Testing and Materials, Philadelphia, pp. 283304. 9 Bower, A. F., 1989, Cyclic Hardening Properties of Hard-Drawn Copper and Rail Steel, J. Mech. Phys. Solids, 37, pp. 455470. 10 Chaboche, J. L., 1991, On Some Modications of Kinematic Hardening to Improve the Description of Ratcheting Effects, Int. J. Plast., 7, pp, 661678. 11 Hassan, T., Corona, A., and Kyriakides, S., 1992, Ratcheting in Multiaxial Cyclic Plasticity, Part II: Multiaxial Behavior, Int. J. Plast., 8, pp. 117146. 12 Ohno, N., and Wang, J. D., 1991, Nonlinear Kinematic Hardening Rule: Proposition and Application to Ratcheting Problems, Structural Mechanics in Reactor Technology, Transaction of the 11th International Conference on Structural Mechanics in Reactor Technology, Shibata, L, Tokyo, Japan, L22/1, pp. 481486. 13 Ohno, N., and Wang, J. D., 1993, Kinematic Hardening Rules with Critical State of Dynamic Recovery, Part IFormulation and Basic Features for Ratcheting Behavior, Int. J. Plast., 9, pp. 375390. 14 Ohno, N., and Wang, J. D., 1993, Kinematic Hardening Rules with Critical State of Dynamic Recovery, Part IIApplication to Experiments of Ratcheting Behavior, Int. J. Plast., 9, pp. 391403. 15 Moreton, D. N., 1993, The Ratcheting of a Cylinder Subjected to Internal Pressure and Alternating Axial Deformation, J. Strain Anal., 28, No. 4, pp. 277282.

16 Jiang, Y., and Sehitoglu, H., 1994, Multiaxial Cyclic Ratcheting Under Multiple Step Loading, Int. J. Plast., 10, No. 8, pp. 849870. 17 Jiang, Y., and Sehitoglu, H., 1996, Modeling of Cyclic Ratcheting Plasticity, Part I: Development of Constitutive Relations, ASME J. Appl. Mech., 63, pp. 720725. 18 Jiang, Y., and Sehitoglu, H., 1996, Modeling of Cyclic Ratcheting Plasticity, Part II: Comparisons of Model Simulations With Experiments, ASME J. Appl. Mech., 63, pp. 726733. 19 Prandtl, L., 1925, Spannungsverteilung in Plastischen Koerpern, Proc. 1st Int. Congr. On Appl. Mech., Delft, pp. 4354. 20 Reuss, E., 1930, Beruecksichtigung der Elastischen Formaenderungen in der Plastizitaetstheorie, Z. Angew. Math. Mech., 10, pp. 266274. 21 Phillips, A., and Tang, J. L., 1972, The Effect of Loading Path on the Yield Surface at Elevated Temperature, Int. J. Solids Struct., 8, pp. 463474. 22 Phillips, A., and Lu, W., 1984, An Experimental Investigation of Yield Surfaces and Loading Surfaces of Pure Aluminum with Stress-Controlled and Strain-Controlled Paths of Loading, ASME J. Eng. Mater. Technol., 106, pp. 349354. 23 McDowell, D. L., 1985, An Experimental Study of the Structure of Constitutive Equations for Nonproportional Cyclic Plasticity, ASME J. Eng. Mater. Technol., 107, pp. 307315. 24 McDowell, D. L., 1987, An Evaluation of Recent Developments in Hardening and Flow Rules for Rate-Independent, Nonproportional Cyclic Plasticity, ASME J. Appl. Mech., 54, pp. 323334. 25 Eisenberg, M. A., and Yen, C.-F., 1984, The Anisotropic Deformation of Yield Surfaces, ASME J. Eng. Mater. Technol., 106, pp. 355360. 26 Newburn, D. A., 1990, Post Yield Cyclic Strain Response of Pressurized Tubes, Master of Science Thesis, The University of Tulsa, Tulsa, Oklahoma. 27 Rolovic, R., 1997, Plasticity Modeling of Multiaxial Cyclic Ratcheting in Coiled Tubing, Ph.D. dissertation, The University of Tulsa, Tulsa, Oklahoma.

Journal of Engineering Materials and Technology

APRIL 2000, Vol. 122 161

Radovan Rolovic1
Research Associate

Multiaxial Cyclic Ratcheting in Coiled TubingPart II: Experimental Program and Model Evaluation
An experimental program was conducted to evaluate the plasticity model proposed in a separate paper (Part I). Constant pressure, cyclic bend-straighten tests were performed to identify material parameters required by the analytical model. Block pressure, bendstraighten tests were conducted to evaluate the proposed model. Experiments were performed on full-size coiled tubing samples using a specialized test machine. Two commonly used coiled tubing materials and four specimen sizes were subjected to load histories consisting of bending-straightening cycles with varying levels of internal pressure. It was observed that cyclic ratcheting rates can be reversed without reversing the mean stress, i.e., diametral growth of coiled tubing can be followed by diametral shrinkage even when the internal pressure is kept positive, depending on the loading history. This material behavior is explained in the context of the new theory. The correlation between the predictions and the test data is very good. S0094-42890000502-8

Steven M. Tipton
Professor Department of Mechanical Engineering, University of Tulsa, Tulsa, OK 74104

Introduction
In Part I of this paper, a single surface plasticity model was presented which predicts multiaxial cyclic ratcheting in tubular structures under a combination of pressure and bending/axial loading conditions. In Part II, experimental results from constant pressure cyclic bending tests are presented that were used to identify material parameters for two coiled tubing materials. Results from more complex experiments are also presented which demonstrate the excellent predictive capabilities of the model. Coiled tubing tends to increase in diameter as it is cycled in bending with constant internal pressure. This occurs in spite of the fact that typical pressure levels cause hoop stress below 50 percent of the yield strength. However, the shape of the curve has not been closely studied and is usually considered linear during a constant pressure test. The data presented in this paper show nonlinear diametral growth, which varies with pressure level and bending strain range. Predictions from the model correlate well with the experimental results. The model predicted another previously unobserved phenomenon: incremental diametral shrinkage during cycling with positive internal pressure. This is predicted to occur when cycling follows previous cycling at signicantly higher internal pressure. Experimental results are presented from block pressure experiments to validate this prediction. It should be noted that additional complex experiments were conducted involving bend-straighten cycling at various pressures with intermittent axial loading, which are not included in this paper. Strain gages monitored axial elongation behavior during these tests and predictions from the analytical model correlated extremely well with those data Rolovic 1.

Experimental Program
Two coiled tubing materials, designated as QT-800 and QT1000, were used in this study. QT-800 is modied ASTM A-606 Type 4 steel, normalized to a minimum nominal monotonic yield
1 Currently at Schlumberger Technology Corporation, Sugar Land, Texas, 77478 e-mail: rolovich2@slb.com. Contributed by the Materials Division for publication in the JOURNAL OF ENGINEERING MATERIALS AND TECHNOLOGY. Manuscript received by the Materials Division November 19, 1998; revised manuscript received September 21, 1999. Associate Technical Editor: H. Sehitoglu.

strength of 552 MPa and minimum nominal tensile strength of 620 MPa. QT-800 has a minimum elongation of 30 percent and the maximum hardness of Rockwell C22. QT-1000 is modied ASTM A-607 micro-alloyed steel, quenched and tempered to minimum yield strength of 690 MPa and minimum tensile strength of 758 MPa. Low-cycle fatigue data were available for these materials from a previous study Tipton 2. Diametral growth ratcheting tests were performed using fullsize coiled tubing specimens. Straight specimens were cut at the end of the production line from the continuous tubing that would have otherwise been spooled onto a take-up reel. Specimen geometry is shown schematically in Fig. 1. Tubing samples had nominal outside diameters, D, of 31.75 mm nominal wall thickness, t, of 2.41 and 3.96 mm and 60.33 mm nominal wall thickness, t, of 3.40 and 3.96 mm. The specimen length, L, was 1300 mm for the 31.75 mm diameter tubing and 1400 mm for the 60.33 mm diameter tubing. The gage length, L gage , was approximately 500 mm for all tubing samples. This section received cyclic bending over the full strain range associated with the curvature of the bending mandrel. Tubing samples were tested as received from the manufacturer without any modication. In some cases, the ends of specimens were machined to t into the pressure caps. The actual diameter and the wall thickness of the specimens were slightly different than their nominal values. Experiments were conducted using a specialized test machine. Schematic layout drawings of the test machine with a specimen in the straight and bent positions are shown in Fig. 2. The specimen is xed at its upper end. The bending mandrel has varying curvature along its length to provide a transition into the gage section curvature. The form is straight at the end, blending tangentially into a curvature of twice the gage radius for 50 mm and tangentially to its gage radius of curvature. Two forms were used for testing with gage radii of 1200 and 1800 mm. Samples were bent and straightened using a hydraulic actuator with two V -prole rollers. Two mechanical limit switches were positioned to reverse the travel of the actuator, thus dening the maximum bent and straight positions of the specimen. A pressure transducer was used to monitor pressure inside the specimen. Internal water pressure was controlled manually using proportional valves. A digital meter was used to display the instantaneous value of internal pressure. Transactions of the ASME

162 Vol. 122, APRIL 2000

Copyright 2000 by ASME

Fig. 1 Specimen geometry

Fig. 4 Variation of tubing diametral growth along the gage length

Constant pressure diametral growth test data were collected at regular intervals throughout the life of each test. Measurements were taken manually at the middle of the gage section using digital calipers. The accuracy of the calipers was 0.001 mm. Since tubing becomes approximately 1 to 2 percent oval during bending cycles, orthogonal measurements were taken across the tube and their average values are reported herein. One diameter, D na , was measured across the bending neutral axis and the other, D tc , was measured 90 degrees from the bending neutral axis from the convex tensile side of the tubing to the concave compressive side see Fig. 3. The average outside diameter was calculated as D 0.5( D na D tc ). After the specimens were taken out of the test machine, it was observed that the diametral growth varied along the gage length as shown in Fig. 4. This effect referred to as ballooning caused the location of the maximum diameter to vary randomly from test to test along approximately 200 mm of the gage length. Measurements could not be taken at the location of the tubing where the maximum growth occurred due to restricted physical access to the tubing and the random nature of the phenomenon. However, the growth trend demonstrated by the measurements was useful for validating the trend predicted by the analytical model. Diameter measurements were taken at the maximum growth location after

Fig. 5 Typical pressure variation during block-pressure tests

the specimen was removed from the machine. These points were used as targets for dening parameters in the incremental plasticity model. Block-pressure diametral growth tests were conducted on tubing samples by applying a nite number of bend-straighten cycles at constant internal pressure. The pressure level was changed several times during a test as shown in Fig. 5. Pressure levels were changed to study the load sequence inuence. High-low, lowhigh, and random pressure level sequences were investigated in this study. Again, for these tests, diameter measurements were taken at the middle of the gage section and the maximum diameter was measured at the end of each test after the specimen was removed from the test machine.

Test Results and Model Evaluation

Experimental results were used to examine coiled tubing deformation behavior and evaluate the proposed analytical model. Based on the proposed model, a computer program was developed to simulate coiled tubing deformation behavior under any applied loading history. Diametral growth data from relatively simple, constant pressure fatigue tests were used to determine parameters needed for the model. Diametral growth data from more complex block load histories were used to evaluate the model. Computational Procedure. Based on the analytical model described in Part I, a computer program was developed to describe deformation of coiled tubing. The plasticity model was applied to discrete regions around the cross-section. Several load histories were simulated with the cross-section divided into 20, 30, 40, 60, 100, and 200 segments. The results based on 40 segments were negligibly different from the solutions obtained with 200 segments. Based on this observation and the advantage of shorter run times, the cross-section with 40 segments was used in this study. Because of symmetry, only one half of the crosssection was modeled as shown in Fig. 6. Stress and strain quantities were calculated as their average values at the center of each segment. All calculations assume that stress and strain are uniform along the length of tubing. Since the tubing has closed ends, there is an axial force acting on the ends of tubing due to internal pressure. The inuence of internal pressure acting on the ends of tubing is modeled using equilibrium of forces in the axial direction of tubing. This can only be achieved iteratively, which increases the execution time of the computer program considerably compared to a closed form solution. In order to reduce the number of iterations and the execution time, the axial equilibrium was satised by maintaining the calculated axial force to within 0.5 percent of the axial force whenever the axial force was not equal to zero. When the axial force was equal to zero, the calculated axial force was maintained within 222 N. Based on the input data material characteristics, tubing dimensions, and load history, the program computed APRIL 2000, Vol. 122 163

Fig. 2 Schematic layout drawing of bendingtensile test machine

Fig. 3 Tubing diameters measured during testing program

Journal of Engineering Materials and Technology

Fig. 6 Cross-section of tubing divided into equal segments Fig. 8 Diametral growth test data and predictions for QT-800 under high constant pressure, cyclic bending loading

changes to the state of stress and strain for each segment. The geometry of each segment, as determined from its updated strain state, was used to compute updated cross-sectional dimensions. These, in turn, were used to compute hoop and axial stresses caused by pressure for the next load application. Determination of Material Constants and Model Parameters. Uniaxial cyclic constants used with the proposed model are given in Table 1. The model requires ve material constants for multiaxial loading conditions to dene the hardening parameter q, Eq. 2 of Part I. To determine these ve material constants, three constant pressure, diametral growth tests were conducted for each material used in this study: i one test at a low hoop stress level ( h / Y 0.1), ii one test at an intermediate hoop stress level ( h / Y 0.3), and iii one test at a high hoop stress level ( h / Y 0.5). Through trial and error comparisons of model predictions with these diametral growth data, the material constants required by the model are obtained as: C q 0 0.002, C q 1 0.218, C q 3 75, and C q 4 2.4 for both QT-800 and QT1000; C q 2 9 for QT-800 and C q 2 4 for QT-1000. A total of six tests three tests for each material were used to establish all material parameters required by the model. Additional constant pressure tests were conducted on tubing specimens with different diameter to wall thickness ratios and with different maximum bending strains to verify that the above material parameters are not affected by the tubing size and loading conditions. Two sets of diametral growth measurements two tests are shown here in Figs. 7 and 8. Additional test results are given by Rolovic
Table 1 Uniaxial cyclic constants Material QT-800 QT-1000 E MPa 200000 200000

1. Tests were not replicated in most cases because diametral growth data generated by tubing manufacturers generally show excellent repeatability. As mentioned earlier, diametral growth measurements were taken at the center of the gage section ( L gage/2) because of the limited physical access to the specimen. Since the maximum diameter in all tests occurred at a different and varying location, it could only be measured at the end of each test, after the specimen was removed from the test machine. The maximum diameter measurements are shown as diamond symbols in Figs. 7 and 8. All predictions in this study are based on the maximum diameter, D max . Model Evaluation. Internal pressure during service loading of coiled tubing is usually not constant and, in fact, can vary from

0.29 0.29

Y MPa
379 517

K MPa 785 806

n 0.10 0.055 Fig. 9 Diametral growth test data and predictions for QT-1000 under block pressure, cyclic bending loading

Fig. 7 Diametral growth test data and predictions for QT-800 under low constant pressure, cyclic bending loading

Fig. 10 Diametral growth test data and predictions for QT1000 under block pressure, cyclic bending loading

164 Vol. 122, APRIL 2000

Transactions of the ASME

Fig. 11 Diametral growth test data and predictions for QT-800 under block pressure, cyclic bending loading

Fig. 12 Diametral growth test data and predictions for QT-800 under block pressure, cyclic bending loading

tubing to substantially grow in diameter when subjected to bendstraighten cycles at internal pressure can be explained schematically using Fig. 14 which shows a von Mises yield surface. At zero internal pressure, the stress state in a highly strained region of tubing impinges on the yield surface at points A and B. Since the plastic strain increment is normal to the yield surface at the loading point normality ow rule, the components of the hoop strain increments at A and B are equal and opposite, which means that negative hoop strain increments at B are offset by positive hoop strain increments at A. Therefore, no diametral growth occurs at zero internal pressure. However, if internal pressure is applied, resulting hoop stress brings cyclic axial loading at the CD level. At this level, the negative hoop component of the plastic strain increment at D is smaller than the positive hoop component at C. This leads to accumulation of plastic strain in the positive hoop direction during each cycle and, therefore, to diametral growth. Block pressure, cyclic bending tests Figs. 913 clearly show the importance of loading sequence on diametral growth rates. In some cases when bending cycles under internal pressure are followed by bending cycles under lower or zero internal pressure, diametral growth ratcheting rates can be reversed, which means that diametral growth can be followed by incremental diametral shrinkage. The amount of diametral shrinkage depends on the hoop stress difference between two load steps and the number of bending cycles imposed at the higher hoop stress level. It increases when the hoop stress difference is increased and when more bending cycles are applied at the higher hoop stress level. Diametral shrinkage tapers off with cycling at a given hoop stress level and cannot cause the tubing to shrink below the original diameter so long as hoop stress is greater than or equal to zero. If the hoop stress level for subsequent bending cycles is not lowered enough, diametral growth can continue, but this time at a lower rate see Fig. 11. Low-high, high-low, and mixed hoop stress sequences were used with different materials and tubing sizes to examine coiled tubing behavior and evaluate the proposed model. The predictions based on the proposed analytical model are very good. In one case the proposed model predicted small diametral shrinkage, while the measurements showed slight diametral growth for one part of the test second hoop stress level in Fig. 13. However, notice that the growth rate for this segment, conducted at a hoop stress of 21 percent of the yield strength, slowed considerably following the 50 cycles imposed with a hoop stress of 34 percent of the yield strength. Note in Fig. 12 that initial cycles at about the same hoop stress 20 percent of the yield strength caused a considerably higher diametral growth rate than the 21 percent block in Fig. 13. The predicted nal diameter was very close to the measured value for all tests. The physics underlying diametral shrinkage under positive hoop stress is explained by considering the behavior of the kinematic yield surface, as dictated by the new hardening rule. Cyclic

Fig. 13 Diametral growth test data and predictions for QT-800 under block pressure, cyclic bending loading

cycle to cycle. In order to evaluate the proposed model for more realistic loading, a number of block pressure tests were conducted. All material constants used by the model were determined from constant pressure, diametral growth tests. The model was then used to predict coiled tubing behavior under more complex pressure histories. Some diametral growth measurements and predictions from the proposed model are shown here in Figs. 913. Additional test results and model predictions are given by Rolovic 1.

Discussion
Coiled tubing routinely exhibits a signicant amount of ratcheting when utilized with internal pressure. The tendency of the Journal of Engineering Materials and Technology

Fig. 14 Graphical representation of diametral growth mechanism when axialbending cycling is applied at CD hoop stress level

APRIL 2000, Vol. 122 165

Table 2 Wall thickness measurements and predictions Original wall thickness mm 2.46 4.11 2.49 4.16 4.01 3.51 4.09 3.53 2.49 Measured wall thickness t 1 mm t 2 mm 1.73 2.79 1.98 3.18 3.23 2.97 3.23 2.92 1.96 1.98 3.12 2.13 3.40 3.28 3.00 3.30 2.97 2.01 Predicted wall thickness t 1 mm t 2 mm 1.75 2.82 1.93 3.30 3.20 2.95 3.15 2.72 1.93 1.80 2.92 1.96 3.33 3.28 2.97 3.20 2.82 1.96

Fig. 15 Graphical representation of diametral shrinkage mechanism under positive hoop stress. Axialbending cycling at hi hoop stress a is followed by axialbending cycling at hj hoop stress b.

axial stress due to bending in the presence of a constant hoop stress due to internal pressure tends to cam the center of the yield surface, , upward progressively hi in Fig. 15a. This effect correctly causes a lower predicted strain ratcheting rate in the hoop direction than conventional hardening rules which tend to overpredict transverse ratcheting. The other consequence of the progressive upward shift in the center of the yield surface is the fact that when the hoop stress level, h , is dropped, the stress state can impinge on the yield surface below its center h j h j in Fig. 15b. The outward normal during axial stress excursions in the tension direction now has the negative hoop component that is greater than the positive hoop component during axial stress excursion in the compression direction. This results in negative hoop ratcheting and diametral shrinkage. As the loading at this hoop stress level proceeds, the yield surface is progressively moved downward, showing the predicted decreasing rate of diametral shrinkage. The characterization of coiled tubing materials using the hardening parameter q described in Part I of this paper is promising in that all but one of the parameters dening q were identical for the two materials examined. Furthermore, this parameter, C q 2 , tends to decrease with the higher yield strength of the material. Having examined only two materials, it would be premature to suggest that this trend would hold up for an entire class of coiled tubing materials. However, further research to identify a robust formulation for q is denitely warranted. Variation of tubing diametral growth along the gage length of the specimens and localized occurrence of the maximum diameter is not a result of the specic specimen gage length or the test machine design. The localization of tubing diametral growth has been observed in eld applications when long tubing sections were exposed to the same loading conditions. Similar tubing behavior has been observed in different test machines that use tubing specimens with different gage lengths. In all of these cases the proposed model accurately predicted the maximum tubing diam-

eter without any adjustments to the model. The random localization of the maximum tubing diameter in the case of tubing ballooning is analogous to the random localization of specimen necking during a simple tensile test. Subtle differences in the material microstructure result in the existence of a tubing section that is less resistant to plastic deformation than the surrounding sections, which leads to locally higher plastic deformation and ballooning. Overall diametral growth ratcheting predictions in this study are very good considering the extremely complex nature of plastic deformation in coiled tubing. In order to improve predictions beyond the current level, more fundamental research is needed. The normality ow rule, which is widely accepted in the theory of plasticity, provides a good estimate for the direction of plastic ow, but yield surfaces are not simple to characterize. Surfaces can distort during plastic deformation, making it more difcult to predict plastic ow with certainty. Material anisotropy and transient behavior hardening, softening, or both further complicate analytical modeling. Even though these phenomena may not be pronounced, they can cause computational errors to accumulate when long load histories are considered. One or two degrees deviation from the normality ow rule, or a small distortion of the yield surface can give negligible errors for a few load cycles, but it can cause large errors to accumulate over a long load history. Another subtle characteristic of coiled tubing deformation behavior was accurately described by the proposed model. Several fatigue test specimens for constant pressure tests were cut transversely after the failure, and the wall thickness was measured at two locations as shown in Fig. 16. The measurements and the predictions from the proposed model are presented in Table 2. It can be seen that the predictions are very good, within 10 percent from the corresponding measured values. The model also predicts accurately that for constant pressure cyclic bending tests, the wall on the concave side of the tubing becomes slightly thinner than the wall on the convex side of tubing t 1 t 2 in Fig. 16. This can be explained by considering that an element at t 1 feels a combined effect of radial compressive stress from pressure and a mean compressive axial strain. This slightly accelerates the driving forces for plastic ow in the circumferential direction, relative to the combination of radial compression and mean tension at t 2 .

Conclusions
Multiaxial cyclic ratcheting in coiled tubing was investigated in this study. Diametral growth ratcheting of coiled tubing was measured during cyclic bending tests under internal pressure. The new incremental plasticity model developed in Part I was used to predict coiled tubing deformation behavior. Based on the experimentally observed coiled tubing behavior and predictions from the proposed analytical model, the following conclusions can be made:
Fig. 16 Wall thinning during cyclic bendinginternal pressure loading

1 When coiled tubing is cyclically bent-straightened in the plastic regime under imposed internal pressure, its diameter increases and wall thickness decreases. The maximum increase in Transactions of the ASME

166 Vol. 122, APRIL 2000

diameter observed in this study was approximately 15 percent and the maximum reduction in wall thickness was about 30 percent. 2 Cyclic ratcheting rates can be reversed without reversing the primary mean stress, depending on the load history. Tubing diametral growth caused by cycling at high internal pressure can be followed by incremental diametral shrinkage caused by bend cycling at a lower positive or zero internal pressure. 3 Plastic deformation of coiled tubing is highly dependent on load history; i.e., diametral growth of coiled tubing differs considerably when load sequence is changed. 4 The proposed analytical model demonstrates excellent capabilities in terms of predicting the complex nature of coiled tubing mechanical behavior. 5 More research would be useful in the following areas: a thin-walled tubes subjected to cyclic axial loading, internal pressure, and torsion would provide additional important validation and clues for renement of the proposed model; b transient cyclic behavior of materials softening, hardening, or both needs to be examined under both uniaxial and multiaxial loading conditions;

c a standard procedure needs to be identied for obtaining empirical parameters required by the plasticity model. Additional data sets for new materials would be useful towards this end.

Acknowledgments
The authors wish to express their appreciation for the support from The University of Tulsa, The United States Department of Energy, and members of the University of Tulsa Coiled Tubing Mechanics Research Project: Elongation and Diametral Growth.

References
1 Rolovic, R., 1997, Plasticity Modeling of Multiaxial Cyclic Ratcheting in Coiled Tubing, Ph.D. dissertation, The University of Tulsa, Tulsa, Oklahoma. 2 Tipton, S. M., 1998, Low-Cycle Fatigue Testing of Tubular Materials using Non-Standard Specimens, Effects of Product Quality and Design Criteria on Structural Integrity, ASTM STP 1337, T. L. Panontin and S. D. Sheppard, eds., American Society for Testing and Materials.

Journal of Engineering Materials and Technology

APRIL 2000, Vol. 122 167

 ffler I. Scha P. Geyer P. Bouffioux

de France, Electricite Direction des Etudes et Recherches, partement MTC Route de Sens, De 77250 Moret/Loing, France

Thermomechanical Behavior and Modeling Between 350C and 400C of Zircaloy-4 Cladding Tubes From an Unirradiated State to High Fluence (0 to 85" 1024 nm2, E 1 MeV
This paper rst describes the effect of neutron irradiation on the thermomechanical behavior of stress-relieved Zircaloy-4 fuel tubes that have been analyzed after exposure to ve different uences ranging from nonirradiated material to high burnup. In the second part, a viscoplastic model is proposed to simulate, for different isotherms, 350CT400C, out-of-ux anisotropic mechanical behavior of the cladding tubes over the uence range 01001024 nm2 (E1 MeV). The model, identied for tests conducted at 350C, has been validated from tests made at 380C and 400C. The model is capable of simulating strain hardening under internal pressure followed by a stress relaxation period, the loading producing an interaction between the pellet and cladding. Introduction of a state variable characterizing the damage caused by a bombardment with neutrons into the model has allowed us to simulate the irradiation-induced hardening and creep rate decrease, as well as the saturation noticed after two cycles of irradiation (451024 nm2 (E1 MeV)) in a pressurized water reactor (PWR). Finally, the numerical simulations show the model is able to reproduce the totality of the thermomechanical experiments. S0094-42890000202-4

P. Delobelle
canique Applique e R. Chale at, Laboratoire de Me UMR 6604, 24 chemin de lEpitaphe, 25030 Besanc on Cedex, France

Introduction

Today, over 70 percent of French electric power has nuclear origin and is generated by 58 pressurized water reactor PWR units. This capacity makes it necessary for reactors to adapt their production to network demand by operating under load following conditions. Moreover, due to economic reasons, Electricity of France EDF seeks to extend the fuel assembly life to ultimately reach a 60 GWd/tU Giga-Watt-day/ton Uranium burnup. These new conditions of use bring about more severe mechanical loadings of the cladding, owing to the greater number of power ramps. Behavior laws currently used for predicting the mechanical behavior of fuel rod cladding do not consider the loading modes inherent under operating conditions Baron and Boufoux 1. This requires developing a behavior model that incorporates the anisotropic behavior of the material more specically, the stress relaxation in cladding following loading from pellet-cladding interaction PCI resulting from a power ramp. Such a model should also be able to take into account the irradiation effects upon the mechanical properties at uences, which may reach 100 1024 nm2 ( E 1 MeV), and at the cladding operating temperatures ranging from 350C to 400C. Conventionally, in reactor deformations have been classied as swelling, thermal or irradiation enhanced. Thermal deformation models have been used to simulate PCI Baron and Boufoux 1. The literature gives many results of the mechanical testing of irradiated claddings Higgy and Hammad 2, Northwood 3, Franklin 4, Petterson 5, Baty et al. 6, Yasuda et al. 7.
Contributed by the Materials Division for publication in the JOURNAL OF ENGINEERING MATERIALS AND TECHNOLOGY. Manuscript received by the Materials Division November 12, 1998; revised manuscript received July 6, 1999. Associate Technical Editor: Kwai S. Chan.

However, the diversity of alloys tested, as well as the irradiation and testing conditions examined were insufcient to provide the data needed to construct a model. EDF has consequently launched, with three industrial partners, new testing programs on stress-relieved Zircaloy-4 claddings used in EDF-PWR. This database is now available and has allowed us to develop a behavior model which takes into account the anisotropy, the temperature, and the irradiation effects upon the cladding mechanical properties. A mechanical behavior model was developed during the rst phase of the project from uni- and biaxial tests conducted on recrystallized and stress-relieved Zircaloy-4 tubes Delobelle et al. 8,9. This paper presents the second phase of the project during which the model has been applied to unirradiated tubes before taking into account the temperature and the irradiation effects upon the mechanical behavior. The model will rst be applied to tests performed at 350C on nonirradiated tubes, which are representative of those in service in PWRs. The analysis will then be expanded to describe the effect of variable temperature between 350C and 400C on the model parameters. The model will be shown to describe the anisotropic thermomechanical behavior of unirradiated claddings. The irradiation effects will be introduced into the model through a damage variable which will affect the viscoplastic state equation and the static recovery of the kinematic strain hardening variables.

Tested Materials and Experimental Results

2.1 Tested Materials. The cladding tubes are stressrelieved, low-tin content, Zircaloy-4. Their chemical composition is given in Table 1. The thermal treatment consists of a 5 hr anneal at 460C. The content of alloying elements is consistent with ASTM B 353 specication. The crystallographic texture of Transactions of the ASME

168 Vol. 122, APRIL 2000

Copyright 2000 by ASME

Table 1 Weight composition of low tin content Zircaloy-4 cladding tubes Alloying elements Cr 0.10 Fe 0.21 Sn 1.25 O 0.109 Zr bal. Impurities ( pp ) C 120 N 31 H 2

Table 2 Kearns factors fr 0.57

such that: f a f r f t 1

fa 0.35

ft 0.08

the tubes has been analyzed through x-ray diffraction. The analysis shows two preferred orientations of the C axis of the hexagonal crystallites at 40 deg in plane ( r , ) and 15 deg in plane ( r , z ), where r is the radial direction, the tangential direction and z the longitudinal direction. The Kearns factor values which represent the average of the squares of the cosines of the angles between the poles 0002 and the direction i Van Swam et al. 10, are given in Table 2 for the three directions r , z , : f r , f a , f l . The grain structure is ne and elongated in the rolling direction. The microstructure is strain hardened similarly to that of a stressrelieved material Boufoux 11. In summary, the material shows a strong texture due to the preferred crystallite orientations and the shape of polycrystal grains. This structure will produce anisotropic mechanical behavior. In order to characterize this anisotropy, tests on tubing have been conducted in the longitudinal direction and under internal pressure. The nominal dimensions of test specimens are an outside diameter ext of 9.5 mm for a wall thickness, e, of 0.57 mm. 2.2 Mechanical Tests Performed on Nonirradiated Cladding Tubes. Many tests have been performed on nonirradiated cladding tubes so as to describe the reference state of the material before irradiation. 2.2.1 The Longitudinal Tests. Longitudinal deformation was measured using LVDT type sensor. The effective specimen length is 50 mm and the required deformation rate is controlled by the sensors. The stress state, which is homogeneous in the test area, is obtained directly by the ratio of the axial force divided by the section of the sample. Monotonic tests were performed at 350C, 380C and 400C with strain rates between 2 10 6 and 2 10 3 s1 to obtain the materials inelastic properties. Figure 1 shows the stress-strain

Fig. 2 Longitudinal creep tests performed at 350C for different stress levels: zz 275, 350, 380, and 400 MPa. Experimental results and simulations.

curves for different strain rates. The data show the strain-rate dependence of the material at these temperatures Delobelle et al. 8. Creep tests were also conducted at stress levels ranging from 140 to 400 MPa for these three temperatures. An example is given in Fig. 2. 2.2.2 Biaxial Tests Under Internal Pressure. Since the cladding is anisotropic, strain hardening tests were performed at various biaxial stress ratios zz / the longitudinal stress divided by the tangential stress at 350C. The diametral deformation rates which were investigated, correspond to those imposed by the pellet on the cladding. Deformation rates ranged from 2.10 5 to 2.10 4 s1. Additional tests were performed at 380C and 400C to determine temperature dependence. Many creep and relaxation tests were also performed. The test specimen is tted with four diametral point sensors of the LVDT type that randomly contact four points along the tube. During the strain-hardening tests, the testing machine exerts the diametral deformation rate while keeping the biaxial stress ratio, , constant. During testing under internal pressure, the stress state is not entirely uniform throughout the tube thickness. One uses the mean variables of rational stresses as in Eq. 1 Boufoux 12:

rr p /2 0,

r ext e exp 1 C , e (1)

zz

Fig. 1 Longitudinal tensile tests performed at 350C for differ3 T , 2 " 10 4 , 2 " 10 5 and 2 " 10 6 s1. ent strain rates zz 2 " 10 Experimental results and simulations.

rr , , and zz are, respectively, radial, hoop, and axial stresses, p is the internal pressure, r ext ext/2 and, e, the wall thickness of the tubes. In order to calculate , knowing the deformation , a constant C which varies with the biaxiality ratio , is introduced. Figure 3 shows an example of biaxial results obtained with 0.5, 0.25, 0, 0.25, 0.5, 0.75 at 350C. At 350C, creep tests were also performed with 0.5 and tangential stress levels ranging from 100 to 275 MPa Fig. 4. Repeated tests at 380C and 400C were conducted to better estimate the temperature effects. Lastly, tensile tests under internal pressure ( 0.5) followed by a relaxation phase were performed for a future validation of the model on loadings representative of a pellet-cladding interaction at two required diametral rates: 2 10 4 and 2 10 5 s1 at 350C, 380C, and 400C. Figure 5 gives an example of data obtained with 2 10 5 s1.
APRIL 2000, Vol. 122 169

Journal of Engineering Materials and Technology

Fig. 3 Biaxial tensile tests performed at 350C for different stress-biaxiality ratio zz , 0.25, 0.5, 0, 0.25, 0.5, and 0.75. Experimental results and simulations.

2.3.1 Program A. Tests specimens taken from tubes similar to those described earlier are introduced into capsules and irradiated in an experimental reactor. Some specimens have been removed after 500, 1000, and 3000 hours so as to obtain claddings irradiated at different uence levels, say about 41024, 8 1024, and 21 1024 nm2 ( E 1 MeV). The upper uence level corresponds to that of one cycle of irradiation in PWR reactor. As the test samples had been irradiated in a NaK medium the samples are corrosion resistant to this medium, they have undergone no hydriding or corrosion and have not been exposed to any mechanical loading. Strain-hardening tests under internal pressure ( zz / 0.5) were conducted at two loading rates, namely: 2.10 4 and 5.10 6 s1, at 350C and 380C. Figure 6 gives an example of the inuence of the irradiation on the stress-strain curves at 350C and 2 10 4 s1. These tests have been complemented by twelve creep tests performed at 350C and three tests at 380C. Some results of creep experiments are given in Fig. 7 ( 8.1024 nm2 and T 350C). Thus, this program allows us to estimate the evolution of the mechanical behavior of the material during the rst cycle (21 1024 nm2 ( E 1 MeV)) in PWR reactor. 2.3.2 Program B. Tests are conducted on specimens from fuel rods taken from a single assembly following two cycles ( 45 1024 nm2 ( E 1 MeV)) in a PWR reactor. Samples are

Fig. 4 Biaxial creep tests performed at 350C with 0.5 and for different stress levels: 100, 140, 170, 200, and 275 MPa. Experimental results and simulations.

Fig. 6 Biaxial tensile tests performed at 350C with 0.5 for an imposed hoop strain rate of 2 " 10 4 s1 and for different uences: 0, 5.1" 1024, 7.21" 1024, and 19" 1024 nm2 E 1 MeV . Experimental results and simulations program A .

Fig. 5 Biaxial tensile tests performed at 350, 380, and 400C with 0.5 and 2"10 5 s1 followed by a relaxation phase. These loadings are representative of a PCI transient. Experimental results and simulations.

2.3 Testing Program Performed on Irradiated Cladding Tubes. This database consists of three programs respectively designated by A, B, and C CEA, Framatome and EDF 13. These programs will allow us to quantify the effects of irradiation upon a fuel rod cladding throughout the life span of assemblies up to high burnup. 170 Vol. 122, APRIL 2000

Fig. 7 Biaxial creep tests performed at 350C with 0.5 on irradiated cladding tubes 1000 h of irradiation for different stress levels: 359, 397, 445, and 517 MPa. Experimental results and simulations program A .

Transactions of the ASME

Fig. 8 Biaxial tensile tests performed at 350C with 0.5 on irradiated cladding tubes until two cycles in PWR 45" 1024 nm2, E 1 MeV and for different hoop strain rates: 2 " 10 4 , 3 " 10 4 , and 5 " 10 6 s1. Experimental results and simulations Program B .

Fig. 10 Biaxial creep tests performed at 350C with 0.5 and 415 MPa on irradiated cladding tubes at different uences: 4.4" 1024, 20.8" 1024, 45" 1024, and 85" 1024 nm2 E 1 MeV , programs A , B , and C . Experimental results and simulations.

then cut and a chemical decladding of the fuel is made. These operations therefore allow us to obtain specimens that have been irradiated upto a uence of 45 1024 nm2 ( E 1 MeV). It is worth noting that claddings are corroded and hydrided as they have been in contact with the primary coolant. However, the hydrogen content of the irradiated fuel cladding has not been measured. Irradiated test specimens are subjected to biaxial tensile tests ( zz / 0.5) at three deformation rates: 2 10 4 , 5 6 1 2 10 , and 5 10 s at 350C, 380C Fig. 8 and 400C, and to creep tests at 350C for ve stress levels ranging from 300 to 520 MPa. Other creep tests were conducted at 380C. Therefore, this program describes the mechanical properties of tubes after two irradiation cycles ( 45 1024 nm2 ( E 1 MeV)) in PWR reactor. 2.3.3 Program C. Like program B, tests are performed on specimens cut from a fuel rod irradiated during four cycles ( 85 1024 nm2 ( E 1 MeV)) in a PWR reactor. These claddings are consequently corroded and hydrided also. Biaxial tensile tests ( zz / 0.5) were performed at 350C, 380C, and 400C at deformation rates ranging from 1.4 10 7 s1 to

1.4 10 4 s1. An example of the stress strain curve at different rates is given in Fig. 9. Creep tests were also performed at these temperatures for zz / 0.5 and various tangential stress levels ranging from 300 to 550 MPa. Figure 10 gives some creep curves obtained at T 350C for 415 MPa and for four uences corresponding to programs A, B, and C. This program allows us to evaluate the mechanical properties of a cladding tube irradiated during four cycles ( 85 1024 nm2 ( E 1 MeV)) in PWR reactor. One will note that specimens obtained within the framework of programs B and C have been oxidized, hydrided, and exposed to in-reactor loadings which remain unknown. This is why we must be careful with regard to a precise analysis of these programs. This question will be considered in detail in the next paragraphs and we will try to check the data base consistency as regards the evolution of irradiation-induced mechanical properties.

Analysis of the Results

3.1 Case of Unirradiated Tubes. As shown in Fig. 1 the material presents a strong strain-rate dependence at the temperatures of the tests. Uni- and biaxial tests show the isotropy of the elastic properties of the material and a linear decrease of Youngs modulus between 350C and 400C while Poissons ratio remains constant. The Rp ratio of the plastic hoop strain to the axial one, p p R p / zz , is found to be, as a rst approximation, independent of temperature, plastic strain, and strain rate and equal to R p 0.6 0.02. The knowledge of the Kearns factors allows us to evaluate ratio R p and then, the material incompressibility permits to write the contractile strain ratio, CSR, Van Swam et al. 10 as: CSR fr Rp 1 f r f a 1Rp (2)

Fig. 9 Biaxial tensile tests performed at 380C with 0.5 on irradiated cladding tubes until four cycles in PWR 85" 1024 nm2, E 1 MeV and for different hoop strain 1.4" 10 6 , 1.4" 10 7 and 1.4" 10 8 s1. Experimental rates: results and simulations program C .

Kearns factors in radial f r and axial f a directions are recalled in Table 2. The calculation gives a value of R p 0.62 which is in accordance with the value reported above. To simplify the description of anisotropy in the model, the set of anisotropy coefcients will be considered as independent of temperature and strain rate, which is corroborated by the results of Beauregard et al. 14 and Murty 15. In fact, this alloy is quite anisotropic for the observed texture, but the failure to detect the elastic anisotropy is an issue of experimental sensitivity. For modeling, the elastic properties will be considered as isotropic, but for the opposite case, the generalized Hookes law can be used: APRIL 2000, Vol. 122 171

Journal of Engineering Materials and Technology

 C e ,

where C is a fourth rank tensor. The use of uniaxial creep tests conducted at 350C, 380C, and 400C allows us to determine the apparent activation energy H and the stress exponent n, according to the relations 3: H ln zz / 1/kT zz , n ln zz / ln zz T (3) zz is the steady creep rate and k the Boltzmann constant. We nd an energy value of 2.0 0.1 eV between 350C and 400C and n 3.4 for the lowest stresses (85 zz 174 MPa). Then, n increases with increasing stress Fig. 12 and can be represented as a hyperbolic sine dependence of zz versus zz . The H value is close to the one obtained in internal friction for the Snoek peak in impure Zr and caused by the reorientation of the oxygen-impurity pairs in the hexagonal matrix; 0-Sn is a possible pair Yi et al. 16, Prioul 17. This value suggests that the dislocation substructure recovery is controlled either by the self-diffusion or by the diffusion of a species within the matrix. This value is used in the model formulation. Note that Murty et al. 18 and Matsuo 19 reported a value of 2.68 eV for the same temperature range, which is in close agreement with the value obtained for selfdiffusion Lyashenko et al. 20. 3.2 Case of Irradiated Claddings. The elastic behavior of the material does not seem to be signicantly modied by a neutron irradiation. However, the inelastic behavior of the claddings is strongly affected by the irradiation. We have plotted on Fig. 11 the variations of the ow stress required to obtain a given deformation versus the uence. To do so, we have used biaxial tensile tests performed at 350C and obtained through programs A and B

at diametral loading rates of 2 10 4 and 5 10 6 s1. The graphs prove that the irradiation makes the material harder at a high rate at uences ranging from 10 to 201024 nm2 ( E 1 MeV) and that it tends to saturate at higher uences. This saturation in the evolution of mechanical properties during tensile tests relative to the uence conrms the conclusions obtained on other types of Zircaloy EPRI 21, Yasuda et al. 7. In addition, as in the case of nonirradiated stress-relieved claddings, the material shows a stabilized ow stress during strain hardening tests even at high loading rates. The comparison of tests performed at 2 10 4 and at 5 10 6 s1 Fig. 11 proves that we mainly obtain a strain hardening effect when the total hoop strains are not important ( 0.6 percent), whereas the recovery prevails when the strains are greater. The results of tests within the framework of program C also display a hardening and a considerable loss in material ductility. We have also seen that the irradiation temperature affects the hardening amplitude through the original position of the tested specimen along assembly of cladding tubes Higgy and Hammad 2, Petterson 5, Franklin 14. The saturation is due to a balance of the emergence of defects and their recombination. As the defect mobility is high, the higher the irradiation temperature, the smaller is the number of irradiation-induced defects present in claddings. These defects will thus promote a quick growth of a dislocation loops. Per contra, at low temperatures, a greater density of the smallest loops, which will slowly grow, can be expected Northwood 3. This might explain why the hardening due to irradiation is slower for claddings irradiated at low temperatures and this conrms the importance of the original state of the test specimen, depending on its position along the cladding in the assembly, with regard to the mechanical behavior. In these programs no reported data obtained on recrystallized samples, one also checks that the variations in the behavior of stress-relieved and recrystallized claddings become smaller as irradiation progresses and that the overall hardening due to irradiation is much more important in the case of recrystallized material than for stress-relieved material when irradiation starts. These observations conrm the conclusions obtained by Higgy and Hammad 2, Franklin et al. 22. During the secondary creep phase, strain hardening could be regarded as being constant and therefore it is possible to analyze the recovery phenomenon by plotting the diametral steady-state creep rate versus the diametral constant stress for different uences Fig. 12. We also take into consideration the steady ow stress obtained during tensile tests as a function of the required strain rate; thus the points lie in continuation of those derived

Fig. 11 Evolution at 350C of the ow stress for different total strain levels versus the uence a 2 " 10 4 s1, b 5 " 10 6 s1

Fig. 12 Representation of the equivalent Mises steady creep rates versus the equivalent Mises stress at 350C and for different uence levels, 0 85" 1024 nm2

172 Vol. 122, APRIL 2000

Transactions of the ASME

Table 3 Equations of the model: Voigt matrix notations Equations of the strains T e

1 t ET ET

3 M , 0 T , D sinh 2 NT,D In these equations: 1 t , 3 1 t , 3

Fig. 13 Evolution of the equivalent steady creep rates versus the uence programs A , B , and C at 350C and for different stress levels; 313, 334, 368, and 415 MPa

3 t M 2

1/2

from creep tests. Except for the case of nonirradiated material ( 0), all the tests for irradiated material were performed with zz / 0.5, thus the translation of the curves Ln Mises f (Ln Mises) Fig. 12, is due to the effect of irradiation and not to the anisotropy. As in the case of nonirradiated tubes, n increases with increasing stress (3.8 n 14). This gure shows that the state recovery of the material during creep tests is a decreasing function of the uence. Some points derived from tensile or creep test results do not lie in continuation of the other tests performed at the same uence. This may be due to experimental uncertainties regarding the temperature and stress which exist during creep tests, the improper knowledge of corrosion and hydriding effects Chen et al. 23 upon the mechanical behavior, and generally speaking, the difculties met when conducting hot-lab tests on specimens which have been weakened by irradiation. In addition, as for strain hardening tests, the cladding irradiation temperature in reactors substantially affects the creep rate Higgy and Hammad 2, Petterson 3, Franklin et al. 21. Figure 13 shows the decrease of the creep rate, at 350C, as a function of the uence. The points derived from tests performed within the framework of programs A, B, and C have been plotted. We note that the creep becomes much lower before stabilizing for high uences according to the results given by Franklin et al. 21. In fact, the creep rate decreases during two irradiation cycles to about 45 1024 nm 2 ( E 1 MeV), and then decreases more slowly. Baty et al. 6 also report a saturation of the hoop creep rate at 350C between 30 1024 and 50 1024 nm2. In conclusion, an analysis of the test results has allowed us to determine, at uence ranging from 0 to 85 1024 nm 2 ( E 1 MeV), consistent evolutions with regard to the mechanical properties of claddings made of Zircaloy-4. The model will be specied, after a formulation has been proposed, on the bi-axial monotonic tensile and creep tests conducted between 350C and 400C for programs A, B, and C.

Equations of the kinematical hardening variables , (1) , and (2) m0 2 p Y *TN Q 1 r m T , D sinh 3 0T,D

N R

1 p 1 2 p 2 and

2 Q 1 2 Y * T N 3 2 Y * T N Q 2 3

k t0 0,0,0,0,0,0

with

3 t R 2

1/2

Equation of the scalar variable Y * Y *Y 0TY bY sat Y Y , Y 0 0

The Voigt notation is adopted, so t 11 1 , 22 2 , 33 3 , 12 4 , 13 5 , 23 6

t 11 1 , 22 2 , 33 3 ,2 12 4 ,2 13 5 ,2 23 6 t 1,1,1,0,0,0

The matrices M, N, Q, and R have an orthotropic symmetry. For example, for M

Modeling of the Temperature and Iradiation Effects

We worked with a behavior model developed on other types of materials and adapted to the Zircaloy case Delobelle et al. 9. This model has proven its tness to simulate the mechanical behavior of tubes made of Zircaloy-4 in two different metallurgical states. 4.1 Initial Model. The equations of the initial model are given in Table 3. Initially in this model, the different parameters of the equations are constant and therefore, independent of the temperature and of the irradiation, i.e., T and D. In the case of small deformations, the total strain T can be divided into two terms, an elastic strain e considered isotropic Journal of Engineering Materials and Technology

M 11 M 12 M 13

M 12 M 22 M 23 0

M 13 M 23 M 33 M 44 M 55 M 66 0

with the incompressibility relations M 11 M 12 M 13 0 M 12 M 22 M 23 0 M 13 M 23 M 33 0

APRIL 2000, Vol. 122 173

and a plastic strain which is anisotropic. Experimental studies including cyclic tests clearly indicated the kinematical nature of the hardening. So, the stress is the sum of two components, a strain rate dependent stress, , and an internal stress . Three kinematical variables are necessary to simulate the strong nonlinearity of the stress-strain curves. The evolution laws of the kinematical hardening variables , ( 1 ) , and ( 2 ) dene the mechanical behavior completely. The rst term of these equations is the linear part of the hardening, the second term represents the dynamic recovery and the last one represents the time dependent static recovery. The introduction of the anisotropy in this model Delobelle et al. 9 is made via the four fourth rank tensors affecting the ow directions M, the linear part of the kinematical hardening N, and the dynamic and static recoveries, Q and R, of these same hardening variables. The material has a mostly radial-tangential texture, but the symmetry of the crystallographic texture in the tubing reference system r, , z imparts the orthotropic properties, which explains the form of the anisotropy matrix. Y is a scalar variable which describes the cyclic softening of the material. In this study, only monotonic loadings are considered and therefore Y is equal to zero and Y * Y 0 . 4.2 Modeling of the Temperature Effect. A detailed description of the general method to identify the parameters of the model is beyond the scope of this paper but is given by Delobelle et al. 9. Only the parameters E, 0 , N, r m , 0 , and Y 0 have been considered as temperature dependent Table 4. Tensile tests show a linear decrease of Youngs modulus between 350C and 400C while Poissons ratio remains constant. As mentioned earlier, in a rst approximation, it has been asTable 4 Modications introduced into the model to take into account the temperature and the irradiation effects Temperature dependence of parameters ETE0E1T 0T 1 exp NTN0N1T Y 0TY 1Y 2T

H* NTn kT

sumed that the temperature does not affect R P between 350C and 400C. For an easier acknowledgment of the temperature in the model, we consider that all the anisotropy coefcients are not linked to the temperature. During the steady creep phase, the use of uniaxial tests performed at 350C, 380C, and 400C allow us to determine the value of H Eq. 3 and consequently the evolution of r m versus the temperature. This is based on the hypothesis that during steady state rst, the strain hardening may be considered as constant and does not vary much with the temperature and secondly the strain-rate dependent stress level is low ( ). Dip-tests conducted at different temperatures Delobelle et al. 9 during creep tests allow us to determine the evolution of as a function of v for each isotherm, and consequently, the determination of H * . To take into account the decrease of the yield stress with the temperature, we have considered that Y 0 is a linearly decreasing function of the temperature. The strainrate dependent stress increases with the temperature and thus N is considered as an increasing function of the temperature. The ve temperature dependent parameters are identied numerically with the SIDOLO software Pilvin 24 over all the uni- and biaxial tensile and creep tests performed at 350C and 400C. Then, knowing the values of these parameters, we determine the coefcients of their evolution laws Table 4. The identication is then validated by comparing the calculation results and the test performed at 380C. The other parameters which are not temperature dependent are identied at 350C. Numerical simulations of all tests have been performed, but only a few gures are presented Figs. 15. The model is in good agreement with the results of the tensile tests at different strain rates Fig. 1, of the creep tests at different stress levels Fig. 2, of the biaxial tests at different biaxiality ratios Fig. 3, and of the biaxial creep tests with 0.5 and different stress levels Fig. 4. At this stage of the model development and of its identication on non irradiated claddings, the formulation is validated by comparing the model predictions and the tensile tests performed at different temperatures under internal pressure ( 0.5) and followed by a relaxation phase Fig. 5. These tests simulate a pellet-cladding interaction type loading. The results can be considered as satisfactory. It should be noted that the whole terms of the model are important and activated for the simulation of the data base, as it was shown in a previous paper Delobelle et al. 9. 4.3 Modeling of the Irradiation Effects. As seen earlier, irradiation causes a marked hardening while reducing the creeprate or strain of the material. These effects are particularly apparent during two cycles in PWR ( 45 1024 nm 2 ( E 1 MeV)) and become less perceptible when higher uences are reached. To take account of the saturation observed in the evolution of the mechanical properties, we do not use the uence as a damage variable creation of random or linear defects linked to bombard ck and Anderment with neutrons, Franklin et al. 22 and Limba son 25 but a state variable D Table 4, such that D D 0 1 exp D 1 , with D 0 0 (4)

0T12T

rmTrm0 exp

H 0Tm0 kT

Equation of the irradiation damage variable D D1T*D0T*D with D 0 0 D The integrated form is given by D D o ( T * )(1 expD1(T*)) In these equations, is the neutron ux, the uence and T * the irradiation temperature which may be different from the temperature T of the test. The evolution of D 0 ( T * ) and D 1 ( T * ) is not specied, these two parameters are considered as constant in the present study. Irradiation dependence of the parameters 0T,D 0TNT,Dn exp1D2 NT,DNTN21expN3D rmT,DrmTexp1D2

0T,D0Texp3D4

A number of model parameters will depend upon this variable Table 4. We have seen that the irradiation temperature T * also affects the cladding hardening amplitude, so the two parameters D 0 and D 1 must depend on this temperature. In this study T * is nearly constant, ( T * 330C), therefore these two parameters are considered as constant. Such a formulation of D variable corresponds to an isotropic representation. It expresses the fact that the number of defects existing in the material is equal to the number of defects created less the number of defects which recombine Matzke 26. A possible anisotropy of the damage might be easily integrated into the model by introducing a second rank tensor A such that D A D . This construction is not necesTransactions of the ASME

174 Vol. 122, APRIL 2000

Lastly, the set of these results conrms the choice made to describe the temperature and irradiation effects, as well as the identication of evolution law parameters with T and . Note that there are no systematic deviations in the simulations, the scattering is due to the experimental uncertainties regarding the temperature 5C and the stress 5 MPa which exist during creep tests on irradiated tubes, the improper knowledge of corrosion and hydridation effects upon the mechanical behavior and, generally speaking, the difculties met when conducting hot-lab tests on specimens which have been weakened by the irradiation.

Conclusions

Fig. 14 Biaxial tensile tests performed at 350 and 380C with 0.5, 2 " 10 5 s1, followed by a relaxation period on irradiated cladding tubes until four cycles in PWR 85" 1024 nm2, E 1 MeV . These loadings are representative of a PCI transient on irradiated tubes. Experimental results and simulations.

sary to simulate the data base described above as only tensile and creep tests under internal pressure ( 0.5) are available. Figure 12 has allowed us to demonstrate that the recovery drastically decreases as irradiation progresses. To take this observation into account, as well as the hardening noticed Fig. 11, we propose to rewrite the evolution law of the kinematical strain hardening variable by making the parameters, 0 and r m of the static recovery term depend on D Table 4. To take account of the strain-rate dependent stress evolution with irradiation, the two parameters 0 and N also depend on the variable D Table 4. One should note that the four parameters, 0 , r m , 0 , and N, which depend on variable D, also depend on the temperature. The formulation of the evolution of these four parameters, given in Table 4, is fairly nonlinear like the phenomena to be modeled. The model parameters linked to irradiation through variable D, as well as those which control the evolution law of D, are identied numerically on biaxial tensile and creep tests made at 350C. Figure 6 compares biaxial strain hardening tests performed at 350C and their simulation at a specied hoop strain rate, 2 10 4 s1, on a nonirradiated tube and on irradiated tubes within the framework of programs A and B. Figure 7 presents simulations of biaxial creep tests derived from program A. From these two gures, the correlation of the simulation with the tests used to identify the model is satisfactory. Figure 8 shows the simulation of the inuence of the hoop strain rate on an irradiated cladding up to two cycles ( 45 1024 nm 2 ( E 1 MeV) in PWR at 350C. The test has not been used for identication purposes and it allows us to check that the model and the parameter are satisfactory. Temperature-controlled evolution laws and irradiation being determined, one checks that the parameters calculated at 380C and 400C allow the simulation of the irradiation-controlled evolution laws with the biaxial tensile and creep tests conducted at these temperatures. Figure 9 compares the simulation of biaxial tensile tests performed at 380C on tubes that have been irradiated during four cycles (85 1024 nm 2 ( E 1 MeV), Program C. One similarly plots on Fig. 10 the diametral strains versus time for creep tests performed at 350C, for 415 MPa on claddings irradiated within the framework of programs A, B, and C. The predictions made for these gures are realistic. Figure 14 shows the simulations of biaxial tensile tests performed at 350C and 380C with 2 10 5 s1 followed by a relaxation period on irradiated cladding tubes until four cycles (85 1024 nm 2 ( E 1 MeV)) in PWR, which are representative of a PCI transient on irradiated tubes. There is a fairly good agreement between the test results and the simulations. Journal of Engineering Materials and Technology

The mechanical behavior of Zircaloy-4 cladding tubes was determined experimentally. The effect of the temperature on the mechanical properties of nonirradiated and irradiated claddings were reported. The analysis of the irradiated cladding tubes data, composed of biaxial tensile, creep, and relaxation tests, exhibited radiation hardening and a decrease in creep strain. A unied viscoplastic model was constructed to describe the experimental observations. The model is able to simulate uni- and biaxial tensile, creep, and relaxation tests for temperatures between 350C and 400C. The effects of irradiation are integrated into the model by means of an internal state variable representing neutron damage. This variable simulates a greater strain hardening and a smaller static recovery. Finally, we demonstrate the ability of the model to evaluate the anisotropic behavior of cladding tubes during tensile, high hoop stress creep and relaxation tests in the temperature range of 350 400C and uences from 0 to 85 1024 nm 2 ( E 1 MeV). The model allows the simulation of a pellet-cladding interaction type loading. The model has already been incorporated at EDF in the nite element code ASTER and implemented in the cladding tube calculation code CYRANO 3. This code is used to predict PCI transients. Future developments of this model will incorporate irradiation creep due to fast neutron ux.

Acknowledgments
The authors want to thank their industrial partners for nancial and technical supports in the research and development programs used to develop this study, i.e., Framatome Nuclear Fuel, Com` lEnergie Atomique, EDF/SEPTEN. missariat a A condensed version of this article has been presented at SMIRT XIV at Lyon France in 1997.

Nomenclature
Mathematical Functions A scalar variable A derivative of the scalar variable A A A vectorial representation of the second rank tensor A A t transpose of the vector A derivative of the vector A with respect to time A vectorial representation of the Kronecker symbol A matrix representation of the fourth rank tensor A A 1 inverse of the matrix A sinh hyperbolic sine function exp exponential function A (0), A t0 initial values of the variables A and A t Multiaxial Strains T, e, total, elastic, and inelastic strain rate vectors of the strain rate; norm 2/3 t M 1 1/2 zz , rr , axial, radial, and tangential strains APRIL 2000, Vol. 122 175

Multiaxial Stresses stress vector deviatoric stress vector , ( 1 ) , ( 2 ) vectorial notation of the three kinematical hardening variables deviatoric kinematical hardening variable norm of the kinematical hardening variable ; 3/2 t R 1/2 norm of the overstress viscous stress ; 3/2 t M 1/2 M, N, Q, R matrix representation of the four fourth rank tensors describing the material anisotropy, i.e., orthotropic symmetry zz , rr , axial, radial, and tangential stresses zz / biaxiality ratio Physical Parameters T T* E H, H* temperature of the test temperature of the irradiation neutron ux uence, dt Young modulus Poisson ratio apparent creep activation energies respectively at and v constant Y * , Y scalar parameters describing the evolution of the amplitude of the kinematical hardening variables during cycling D irradiation damage variable scalar k Boltzmann constant

References
acteurs a ` 1 Baron, D., and Boufoux, P., 1989, Le Crayon Combustible des Re e de Grande Puissance, Rapport EDF, HT M2/88-27. Eau Pressurise 2 Higgy, R., and Hammad, F. H., 1972, Effect of Neutron Irradiation on the Tensile Properties of Zircaloy-2 and Zircaloy-4, J. Nucl. Mater., 44, pp. 215277. 3 Northwood, D. O., 1977, Irradiation Damage in Zirconium and its Alloys, AT. Energy Review, p. 154. 4 Franklin, D. G., 1982, Zircaloy-4 Cladding Deformation During Power Reactor Irradiation, ASTM-STP 754, pp. 235267. 5 Petterson, K., 1982, An Evaluation of Irradiation Temperature on the Irradiation Hardening of Zircaloy, Studvik Super-Ramp Project, SR 82/3. 6 Baty, D. L., Pavinick, W. A., Dietrich, M. R., Clevinger, G. S., and Papazoglou, T. P., 1984, Deformation Characteristics of Cold-Worked and Recrystallized Zircaloy-4 Cladding in Zirconium in the Nuclear Industry, 6th International Symposium, ASTM-STP, 824, 306339.

7 Yasuda, T., Nakatsuka, M., and Yamashita, K., 1987, Deformation and Fracture Properties of Neutron-Irradiated Recrystallized Zircaloy-2 Cladding Under Uniaxial Tension, Zirconium in the Nuclear Industry, VIIth Int. ASTM 939, pp. 734747. 8 Delobelle, P., Robinet, P., Boufoux, P., Geyer, P., and Le Pichon, I., 1996, A Unied Model to Describe the Anisotropic Viscoplastic Behavior of Zircaloy-4 Cladding Tubes, Zirconium in the Nuclear Industry, 11th International Symposium, ASTM-STP 1295, pp. 373393. 9 Delobelle, P., Robinet, P., Geyer, P., and Boufoux, P., 1996, A Model to Describe the Anisotropic Viscoplastic Behavior of Zircaloy-4 Tubes, J. Nucl. Mater., 238, pp. 135162. 10 Van Swam, L. F., Knorr, D. B., Pelloux, R. M., and Shewbridge, J. F., 1979, Relationship between Contractile Strain Ratio R and Texture in Zirconium Alloy Tubing, Metall. Trans. A, 10, p. 483. 11 Boufoux. P., 1995, An Experimental Method to Investigate the Anisotropic Viscoplastic Behavior of Zircaloy Cladding Tubes, 11th International Symposium, ASTM-STP. des Tubes de 12 Boufoux, P., 1994, Etude du Comportement en Plasticite es, Report DER-EDF A4/94/ Gainage en Zircaloy-4 Sous Sollicitations Biaxe 012A. 13 CEA, Framatome, and EDF Cooperative program, 1995, Proprietary data. 14 Beauregard, R., Clevinger, G. S., and Murty, K. L., 1977, Effect of Annealing Temperature on the Mechanical Properties of Zircaloy-4 Cladding, Proceedings of the SMIRT IV, paper C3/5. 15 Murty, K. L., 1989, Applications of Crystallographic Textures of Zirconium Alloys in Nuclear Industry, Zirconium in the Nuclear Industry, VIIIth International Symposium, ASTM-STP 1023, p. 570. 16 Yi, J. K., Park, H. B., Park, G. S., and Lee, B. W., 1992, Yielding and Dynamic Strain Aging Behavior of Zircaloy-4 Tube, J. Nucl. Mater., 189, pp. 353. 17 Prioul, C., 1995, Le Vieillissement Dynamique Dans les Alliages de Zirco s Me caniques, SF2M, Journe e nium: Consequences Sur Les Propriete te s-Microstructures-Les Edit, Physique, pp. 2534. dEtudes Proprie 18 Murty, K. L., Clevinger, G. S., and Papazoglou, T. P., 1977, Thermal Creep of Zircaloy-4 Cladding, SMIRT IV, San Francisco, Aug. 1519, paper C3/4. 19 Matsuo, Y., 1987, Thermal Creep of Zircaloy-4 Cladding Under Internal Pressure, J. Nucl. Sci. Technol., 24, No. 2, pp. 111119. 20 Lyashenko, V. S., Bykov, V. N., and Paulinov, L. B., 1959, Fiz. Met. Metalloved, 8, p. 362. 21 EPRI, B and W, 1983, Cooperative Program on PWR Fuel Rod Performance, NP 2848, Project 711-1. 22 Franklin, D. G., Lucas, G. E., and Bement, A. L., 1983, Creep of Zirconium Alloys in Nuclear Reactors, ASTM STP 815, p. 35. 23 Huang, P., Mahmood, T., and Adamson, R., 1996, Effects of Thermomechanical Processing on In-Reactor Corrosion and Post-Irradiation Mechanical Properties of Zircaloy-2, 11th International Symposium, ASTM-STP 1295, pp. 726755. ` les de Comporte24 Pilvin, P., 1988, Identication Des Parametres de Mode ` camat, International Seminar on Inelastic Behavior ment, Proceedings of Me of Solids, Models and Utilization, pp. 155164. ck, M., and Anderson, T., 1996, A Model for Analysis of the Effect of 25 Limba Final Annealing on the in- and out-of- Reactor Creep Behavior of Zircaloy Cladding, ASTM-STP 1295, pp. 448468. 26 Matzke, H., 1993, Radiation Damage in Nuclear Fuel Materials, Solid State Phenomena, 30 31, pp. 355366.

176 Vol. 122, APRIL 2000

Transactions of the ASME

L. E. Govaert
Eindhoven University of Technology, Faculty of Mechanical Engineering, P.O. Box 512, 5600 MB Eindhoven, The Netherlands

The Influence of Intrinsic Strain Softening on Strain Localization in Polycarbonate: Modeling and Experimental Validation
Intrinsic strain softening appears to be the main cause for the occurrence of plastic localization phenomena in deformation of glassy polymers. This is supported by the homogeneous plastic deformation behavior that is observed in polycarbonate samples that have been mechanically pretreated to remove (saturate) the strain softening effect. In this study, some experimental results are presented and a numerical analysis is performed simulating the effect of mechanical conditioning by cyclic torsion on the subsequent deformation of polycarbonate. To facilitate the numerical analysis of the mechanical rejuvenation effect, a previously developed model, the compressible Leonov model, is extended to describe the phenomenological aspects of the large strain mechanical behavior of glassy polymers. The model covers common observable features, like strain rate, temperature and pressure dependent yield, and the subsequent strain softening and strainhardening phenomena. The model, as presented in this study, is purely single mode (i.e., only one relaxation time is involved), and therefore it is not possible to capture the nonlinear viscoelastic pre-yield behavior accurately. The attention is particularly focused on the large strain phenomena. From the simulations it becomes clear that the preconditioning treatment removes the intrinsic softening effect, which leads to a more stable mode of deformation. S0094-42890001002-1

P. H. M. Timmermans
Philips Center for Manufacturing Technology, Eindhoven, The Netherlands

W. A. M. Brekelmans
Eindhoven University of Technology, Faculty of Mechanical Engineering, P.O. Box 512, 5600 MB Eindhoven, The Netherlands

Introduction

The deformation behavior of glassy polymers is generally strongly dominated by localization phenomena like necking, shear band formation, or crazing. This susceptibility to localization is directly related to the intrinsic large strain behavior of glassy polymers visualized in Fig. 1. These true stress-strain curves can be obtained in uniaxial extension using a video-controlled tensile test 1 or in uniaxial compression 2,3. Typically, the yield stress depends on strain rate, temperature, and pressure 4. The postyield behavior of glassy polymers is governed by two characteristic phenomena 1,5. Immediately after the yield point the true stress tends to decrease with increasing deformation, an effect that is usually referred to as intrinsic strain softening. At large deformations the softening effect is saturated and the true stress starts to rise again with increasing deformation. This strain hardening effect has been subject of a number of studies in the past e.g., 3,6,7, and is generally interpreted as a rubber elastic contribution by the molecular entanglement network. Although the origin of the intrinsic softening effect is not yet completely clear, it seems to be closely related to the physical aging process volume relaxation that occurs in the glassy state 8. With physical aging the specic volume decreases leading to an increase of the elastic modulus, a decrease of the time dependence age-shift, and an increase of the yield stress 8. The increase of the yield stress seems to develop simultaneously with the enthalpy overshoot that is observed around the glass transition temperature in DSC experiments on aged amorphous polymers 9,10. The effect of aging on the deformation behavior of a glassy polymer is schematically represented in Fig. 2. During aging, the yield stress increases and the intrinsic softening effect appears. As a result of intrinsic softening the large strain behavior
Contributed by the Materials Division for publication in the JOURNAL OF ENGINEERING MATERIALS AND TECHNOLOGY. Manuscript received by the Materials Division May 5, 1999; revised manuscript received November 23, 1999. Associate Technical Editor: H. M. Zbib.

of the different samples is exactly the same: the effect of physical aging has been removed and the material is rejuvenated 1012. The same effect can be achieved by heating the sample above the glass transition temperature and cooling it rapidly to the glassy state quenching. In some glassy polymers, as, for instance, PVC, intrinsic softening completely disappears after this quenching from the rubbery into the glassy state 13. Although these experimental observations clearly connect the intrinsic softening effect to the physical aging process, the effect cannot be rationalized completely in terms of an increase of free volume as a result of the imposed strain. The inability to explain intrinsic softening in experiments with a negative dilatational strain compression is probably the strongest argument. Xie et al. 14 measured a decrease of the actual free volume in polycarbonate under compression by means of positron annihilation lifetime spectroscopy PALS, whereas polycarbonate is known to display intrinsic softening in compression 3. In PALS measurements during compression tests on polymethylmethacrylate, however, Hasan et al. 15 observed an increase of the number of areas of local free volume evolving to a steady value. Based upon these observations they postulated a phenomenological law for the evolution of the density D of these areas in a glassy polymer during deformation. During elastic deformation, D is constant the material state does not change. During plastic deformation, D evolves to a saturation value D , indicating a maximum amount of regions with elevated levels of free volume, which is independent of strain rate or thermal history. Inclusion of D in an originally nonintrinsic softening model resulted in a constitutive model that exhibited intrinsic strain softening 15. Intrinsic strain softening is an important factor in the initiation of strain localization. As during softening the deformation is allowed to proceed at a decreasing level of the true stress, small stress variations will inherently lead to large differences in the local strain rate, nuclei for localized plastic deformation zones. In the absence of intrinsic softening, the deformation will be homogeneous if the strain-hardening behavior is large enough to comAPRIL 2000, Vol. 122 177

Journal of Engineering Materials and Technology Copyright 2000 by ASME

porate the phenomena of intrinsic strain softening and strain hardening. The material characterization, including the determination of the necessary parameters for the extended model, will be discussed. With the parameters for polycarbonate known, the model is employed to simulate neck formation and to predict the deformation behavior of a mechanically preconditioned sample torsion cycling in a subsequent tensile or torsion test.

Constitutive Modeling

Fig. 1 Schematic representation of the effect of strain rate on the true stress-strain curve of a glassy polymer

2.1 The Compressible Leonov Model. For an arbitrary material element of a loaded conguration the local actual deformation with respect to a predened reference state is determined by the deformation gradient tensor F e.g., Hunter 20. This tensor F is multiplicatively decomposed into an elastic part Fe and a plastic part Fp , according to: F Fe Fp (1) The plastic contribution Fp indicates the deformation with respect to the reference state of the relaxed stress-free conguration, which is dened as the state that would instantaneously be recovered when the stress is suddenly removed from the element considered. The decomposition in Eq. 1 is not unique because rotational effects can be assigned to Fp as well as to Fe . Uniqueness is achieved by the extra requirement that the plastic deformation occurs spin-free 21. The Cauchy stress tensor is elastically expressed in the left Cauchy Green tensor Be associated with the tensor Fe which is dened by Be Fe Fc e Fc e (2) where denotes the conjugate of Fe which is equivalent to the transpose of the matrix representation of the tensor. In this equation it is presupposed that the elastic behavior is isotropic. In that case the application of expressions of the type ( Be ) guarantees the conservation of objectivity if indeed the total spin and consequently superimposed rigid body rotations are completely attributed to the elastic part of the deformation. To specify the dependence of the stress on the deformation, a neo-Hookean relationship is chosen 19: d K J e 1 I G B e (3)

Fig. 2 Schematic representation of the effect of physical aging on the true stress-strain curve of a glassy polymer

pensate for the geometrical softening during a tensile test 4,7. An extensive numerical study on the inuence of strain softening and strain hardening on neck formation in plane strain extension was performed by Wu and van der Giessen 16. They showed that intrinsic strain softening always leads to strain localization, whereas in the absence of softening strain localization can be suppressed if the amount of strain hardening is sufcient. There is also some experimental evidence concerning the inuence of strain softening on neck formation. Cross and Haward 13 used samples of quenched PVC that display no intrinsic softening and observed uniform deformation in a tensile test whereas slowly cooled samples necked. An alternative method to prevent inhomogeneous behavior in glassy polymers is based on the initial elimination of intrinsic softening by raising the value of the softening parameter D to its saturation value D by application of plastic deformation mechanical preconditioning. A good example of the effect of mechanical preconditioning is the alternated bending of PVC samples by Bauwens 17, which suppressed necking in a subsequent tensile test. GSell 11 achieved the same effect after plastic cycling in simple shear on polycarbonate. Recent experimental research 18 also shows the effect of the elimination of intrinsic softening by mechanical preconditioning: axisymmetrical samples were plastically cycled in torsion; tensile tests on these rejuvenated samples resulted in homogeneous deformations and allowed for the characterization of the strain hardening behavior of polycarbonate. The present study addresses the inuence of intrinsic strain softening on the macroscopic deformation behavior of axisymmetric polycarbonate bars. To facilitate a numerical analysis, a constitutive model which was derived in a previous study, the so-called compressible Leonov model 19, is extended to incor178 Vol. 122, APRIL 2000

where the superscript d indicates the deviatoric part. In this equation, K and G are the bulk modulus and the shear modulus, respectively. The elastic volume change factor J e is dened by J e det Fe det Be (4)

e denotes the isochoric fraction of the elastic left The tensor B Cauchy Green tensor Be according to
2/3 e J B Be e

(5)

Based on purely kinematical considerations 19 the following differential equation can be derived to calculate the evolution of e : B Dd Dd B e B e Dd Dd B e p p (6)

The left-hand side of this equation represents the objective Jaumann derivative of the isochoric elastic left Cauchy Green tensor. The tensor Dp denotes the plastic deformation rate tensor. The initial condition necessary for the solution of the differential equa e I. tion 6 reads: B To complete the constitutive description the plastic deformation rate is expressed in the Cauchy stress by a generalized nonNewtonian ow rule 22 Dp

d 2 eq

with

d d eq 1 2 tr

(7)

Transactions of the ASME

Gaussian chain statistics leading to a neo-Hookean strain hardening response to experimental uniaxial stress-strain curves, and concluded that some amorphous and most semicrystalline polymers obeyed this formulation. The large amount of softening, observed in some glassy polymers, prevented the successful application of the Gaussian model to these polymers. However, as mentioned in the Introduction, it can be shown by means of mechanical preconditioning 18 that this approach is also valid for polycarbonate. Gaussian statistics leads to a neo-Hookean relation between stresses and strains. Generalization to three dimensions, in the assumption that the network is incompressible, this neoHookean relationship for the hardening stress tensor r can be written as d r H B
Fig. 3 Schematic representation of the response in uniaxial extension from the Leonov model

(12)

The viscosity depends on the equivalent stress eq according to an Eyring relationship 19:

eq A 0

eq / 0 sinh eq / 0

(8)

with H the strain hardening modulus assumed to be temperature independent. Contrary to Boyce et al. 21 the hardening stress is not related to the plastic deformation but to the total deformation. This adaptation is introduced because in the present approach both elastic and plastic deformations are assumed to decrease the congurational entropy of the polymer. To complete the constitutive description the plastic deformation rate is still expressed in the Cauchy stress tensor by a generalized non-Newtonian ow rule Dp sd 2 eq , D , p (13)

In this equation A is a time constant and 0 a characteristic stress, respectively related to the activation energy H and the shear activation volume V according to 4,20 H A A 0 exp ; RT

RT 0 V

(9)

with R the gas constant, A 0 a constant preexponential factor involving the fundamental vibration energy, and T the absolute temperature. It is emphasized that Eq. 7 implies that plastic deformation occurs at constant volume: tr( Dp ) 0 as tr( d ) 0. Consequently Dd p in Eq. 6 may be replaced by Dp . For the same reason, J e det(Fe ) in Eq. 3 may be replaced by J det(F). The model derived above was referred to as the compressible Leonov model in the original paper by Tervoort et al. 19. To demonstrate the typical behavior of this compressible Leonov model, an application to uniaxial extension is performed. This leads, for constant strain rate, to the response schematically visualized in Fig. 3. The response of this Leonov model shows a sudden transition from elastic-to-plastic behavior, which is very similar to that of an elastic-perfectly plastic material with a ratedependent yield stress. 2.2 Extension to Intrinsic Strain Softening and Strain Hardening. This section describes the extension of the compressible Leonov model to include both the intrinsic strain softening and the strain hardening effect. Complementary to the outline in Section 2.1 the Cauchy stress tensor is now redened to be composed of two distinguishable parts in a parallel assemblage, the driving stress tensor s and the hardening stress tensor r, according to

where eq , D, and p are state variables to be dened in the following. Particularly the driving stress tensor s is relevant for the incorporation of softening in the model. As suggested by Hasan et al. 15 a history variable D is specied, the softening parameter, which inuences the viscosity . During plastic deformation D evolves to a saturation level D , which is independent of the strain history. The result for reads

eq , D , p A m D , p 0

eq / 0 sinh eq / 0

(14)

where the equivalent stress eq is redened by

d d eq 1 2 tr s s

(15)

and with A m D , p A exp

p D 0

(16) (17)

1 1 p 3 tr 3 tr s

where p is the pressure positive in compression. The parameter is a pressure coefcient, related to the shear activation volume V and the pressure activation volume according to

s r

(10)

(18)

The expression for the driving stress s is adopted from the compressible Leonov model described above, see Eq. 3: d s K J 1 I G B e (11)

The evolution of the softening parameter D is specied by 15 h 1 D

D D p

(19)

The expression for the hardening stress r is obtained in the following. In studies on the deformation behavior of glassy polymers, it is common practice to model the hardening behavior as a generalized rubber elastic spring with nite extensibility, like the so-called three-chain and eight-chain models of Arruda and Boyce 3, or the full chain model of Wu and van der Giessen 6. On the other hand, Haward 7 applied network models employing Journal of Engineering Materials and Technology

with initially D 0; h is a material constant describing the relative p is the equivalent plastic strain rate, accordsoftening rate and ing to

p tr Dp Dp

eq

&

(20)

APRIL 2000, Vol. 122 179

Experimental

3.1 Materials and Sample Preparation. The material used in this study was polycarbonate, purchased as extruded rods 10.4 mm in diameter from Eriks BV Alkmaar, The Netherlands. Additional to the mechanical parameters K and G, the values of the density , the thermal conductivity k, the thermal expansion coefcient , the specic heat c, and the glass-transition temperature T g are given in Table 1. The material properties, with the exception of T g , G, and K, were provided by the supplier and are in good agreement with values reported in literature 2326. T g was determined by dynamic mechanical thermal analysis DMTA and G and K were determined from the Youngs modulus E and the Poissons ratio v measured in the initial stages of a tensile test 19. The thermal material parameters will be used to perform a thermomechanical analysis in the subsequent sections. For the uniaxial extension and torsion experiments, the specimens were designed as dog-bone shaped axisymmetric bars, depicted in Fig. 4a. For the uniaxial compression experiments, cylindrical test specimens were used, the geometry shown in Fig. 4b. 3.2 Uniaxial Extension. Uniaxial tensile tests were performed on a FRANK 81656 tensile tester at strain rates varying from 10 4 to 10 2 s 1 and at temperatures of 22, 32, and 40 C 295, 305, and 313 K, respectively. The true stress at the yield point, required for the determination of the yield parameters, was determined by assuming incompressibility in the viscoelastic area, which introduces a small error approx. 2 percent compared to a compressible approach. Neck formation and propagation was recorded by means of a video camera. From the images, the elongation factor in the neck is calculated from the diameter reduction in combination with the assumption of incompressibility. At the end of the test, the neck diameter was measured with the specimen still in the load frame as an assessment of the video images. 3.3 Uniaxial Compression. Uniaxial compression tests were performed at room temperature at strain rates in the range from 10 4 to 10 2 s 1 , also on a FRANK 81656 tensile tester. A high performance lubricant Hasco Z260 between the sample and the polished stainless-steel shaft of the compression device could not prevent the samples from barreling at compressive strains of approximately 0.20 -. Since this phenomenon occurred at compressive strains beyond the yield point, it does not affect the measured value of the true stress at the yield point. To avoid inuence on the determination of the softening and hardening parameters, data measured at compressive strains beyond 0.20 - are omitted.

3.4 Torsion. Torsion experiments were performed on a testing machine consisting of an adjustable rigid support and a rotating clamp. The sample is installed in the machine in a way that initial axial forces in the sample are avoided. In the testing device the length of the sample is xed during deformation and torque and axial load on the sample are measured during deformation by two independent load cells in the support. To determine the rotation, the angular displacement of the clamp is monitored. As a reference, an axial line was drawn on the specimen, and it appeared that the torsion was restricted to the gauge section of the sample. During the torsion of polycarbonate initially narrow circumferential shear bands were observed that broadened with ongoing rotation, a phenomenon that has also been reported by Wu and Turner 27. In order to obtain isothermal conditions, the rotation speed was limited to 360 degrees per minute resulting in a nominal shear rate of 0.56 s1 at the outer surface of the bar. In the case of the mechanical pretreatment rejuvenation, the torsion experiments were performed by twisting polycarbonate specimens to and fro over 720 degrees. After reversing the direction of the twist, the rotation rate was the same as during loading. Heating of the rejuvenated samples above the glass transition temperature did not induce any residual motion, from which it was concluded that the specimens rejuvenated in this way regain isotropy. After mechanical conditioning, the rejuvenated samples were allowed to relax unconstrained for 3 hrs. Subsequently, they were subjected to either uniaxial extension or torsion.

Material Characterization

4.1 Yield Parameters. The yield or Eyring parameters can be determined by measuring the true stress at the yield point during tension and compression experiments as a function of strain rate at different temperatures 28,29. The strategy is based on the application of the incompressible non-Newtonian viscous ow rule, Eq. 13, which can be reformulated in axial direction Fig. 4a by p , zz 1 s s p , zz 3 zz rr (21)

with p , zz the axial plastic elongation factor and with s zz and s rr the axial and radial components of the driving stress tensor s, respectively. The expression for the viscosity, Eq. 14, can then be replaced by:

s zz , s rr , p , D , T

Table 1 Material temperature

properties

of

polycarbonate

at

room

0) A0 p s zz s rr exp D sinh 0 0)

s zz s rr

(22)

k kg m3 W m1 K1
1200 0.21

K1
65 10 6

c G K Tg J kg1 K1 MPa MPa C 1200 860 4000 150

To facilitate a straightforward analysis of the yield data, the following considerations are made: At the yield point, the contribution of hardening is negligible, and therefore the components of the driving stress are equal to the components of the Cauchy stress. At the yield point, the plastic strain rate is equal to the nomi0 nal strain rate zz applied. At the yield point the value of the softening parameter D equals 0. The argument of the hyperbolic sine in the viscosity function is large, and therefore the hyperbolic sine may be approximated by an exponential function. During uniaxial tension and compression, the pressure is given by p 1/3 zz . The incorporation of these considerations into the non-Newtonian ow rule leads to Transactions of the ASME

Fig. 4 Geometry of axisymmetrical specimens for a uniaxial extension and torsion experiments and b uniaxial compression experiments. Dimensions in mm.

180 Vol. 122, APRIL 2000

Fig. 5 Yield stress over absolute temperature zz T as a function of strain rate zz . The solid lines are a best t using a single set of yielding parameters for each polymer.

1 Fig. 6 True stress zz versus 2 zz zz during a tensile test at 2.2" 10 3 s 1 of a polycarbonate tensile bar, preconditioned in torsion

Table 2 Yield parameters for polycarbonate A0 s 3.6 10 25 V m3 mol1 3.4 10 3 m3 mol1 2.4 10 4 H kJ mol1 290

quently supports the neo-Hookean approach and directly reects the value of the hardening modulus: H 29 MPa . 4.3 Strain Softening Parameters. The softening parameters of polycarbonate are provisionally determined from a uniaxial compression experiment. As was mentioned before, the uniaxial compression experiments showed barreling of the specimen at compressive strains over 0.2 - and therefore the data at larger compressive strains were omitted. The values of the softening parameters were determined by a tting procedure on the post yield behavior of a compression test at a rate of 10 3 s 1 : h 200 - and D 28 - see Fig. 7a. To facilitate the comparison with the experimental data, the predictions by the Leonov model were shifted along the strain axis in order to overlap the predicted and the measured yield points. The actual comparison between the experimental data and the prediction using the compressible Leonov model is shown in Fig. 7b. The simulation was performed using the yield and hardening parameters obtained in the previous sections and the softening parameters mentioned above. Note the difference between the strains at yield in Fig. 7b. Since the single mode compressible Leonov model displays elastic rate independent behavior up to the yield point, it is not able to describe the multirelaxation time viscoelastic behavior displayed by the material. There are several possibilities to correct for this deciency. One is the extension of the Leonov model to a spectrum of relaxation times multimode as was suggested by Tervoort et al. 30. Another possibility was demonstrated by Hasan and Boyce 31 who considered a distribution of activation energies. As the use of either of these extensions would dramatically increase the computation time for the nite element analysis in the next sections the deciency at low strain levels will not be addressed.

H zz 1 3R 0 ln A 0 zz ln ) T RT 6 )Va

(23)

with a sign( zz ). This expression suggests that plots of zz / T against the logarithm of the strain rate for a series of temperatures should give a set of parallel lines. The result for polycarbonate is shown in Fig. 5, where the measured values of zz / T of both the uniaxial tension as the uniaxial compression tests are plotted 0 against the logarithm of zz . The solid lines represent the best t of the experimental results using a single set of the yield parameters A 0 , V , , and H , and seem to represent the actual yield behavior well over the entire range of strain rates experimentally covered. It should be noted however, that the yield parameters should be used with care outside the experimentally covered region, as it has been shown in experiments by Bauwens-Crowet et al. 29 that the yield stress of polycarbonate tends to have a more substantial strain rate and temperature dependence at low temperatures and high strain rates than observed at high temperatures and low strain rates. This is related to secondary glasstransitions and implies that actually more than one Eyring ow process should be taken into account. For the range of strain rates considered in the present work a single ow process seems to sufce. The values of the yield parameters obtained from the t are given in Table 2. The values are in good agreement with values reported by Bauwens-Crowet et al. 29 and Duckett et al. 28 4.2 Hardening Parameter. As mentioned before in the Introduction, the hardening parameter H is, in the present study, determined from a uniaxial tensile test on a rejuvenated polycarbonate sample. To rejuvenate the material, axisymmetric samples were subjected to a 720 deg to and fro xed-end torsion treatment. After this mechanical pretreatment the intrinsic strain softening behavior has disappeared saturated with the astonishing result that the polycarbonate bars deform homogeneously without necking in a subsequent tensile test. Figure 6 shows the result of such an experiment, where zz is plotted as a function of the strain 2 1 measure zz zz . This strain measure is, in a uniaxial tensile or compression test, the component of the deviatoric isochoric left d in the load direction e z . The Cauchy Green deformation tensor B constant slope of the tensile curve at large strain levels conseJournal of Engineering Materials and Technology

Fig. 7 Determination of the softening parameters in polycarbonate at a compressive strain rate of 10 3 s 1 at room temperature. a tting the post-yield softening behavior; b simulation using the compressible Leonov model.

APRIL 2000, Vol. 122 181

Numerical Simulations

5.1 Uniaxial Extension of Untreated Polycarbonate. For the analysis an axisymmetric specimen, as used in the experiments Fig. 4, is considered. In the center of the bar, a Cartesian coordinate system is dened, in which z refers to the axial direction and r to the radial direction. Because of symmetry of the material geometry and loading conditions, in Fig. 8 only one quarter of the longitudinal cross-section is considered up to z 0.5 L 0 initial geometry. At the end face on this position the displacements in z direction are prescribed constant velocity. In the simulations, near z 0, a geometric imperfection is introduced to initiate necking. This imperfection is cosine shaped, dened by R i R 0 1

1 z 1 cos 2 R0

0zR0

(24)

with R i the outer radius of the imperfection, R 0 the outer radius of the gauge section of the perfect bar, the measure of the imperfection, (2 R i ( z 0) R 0 )/ R 0 , and controls the length of the imperfection. In the present calculations 0.9925 and 0.85 which is equivalent to an area reduction of 1.5 percent at z 0. The nominal or engineering stress is dened by F z /( R 2 0 ), where F z is the applied tensile force. The result of a numerical simulation at a nominal strain rate in the gauge section of 7.510 3 s1 is given in Fig. 9. The nominal stress and the elongation factor in the neck are depicted as a function of the nominal strain, together with the deformed meshes at different stages of the deformation. The nominal stress is dened as the tensile force divided by the original cross-sectional area and the draw ratio in the neck N is calculated by division of the original cross-sectional area of the bar by the actual crosssectional area in the middle of the specimen. During the simulation, the specimen initially deforms homogeneously, both in the elastic region and in the rst part of the viscoplastic region ( a b ). At some stage, the deformation localizes and a neck is formed ( b d ), which propagates along the specimen as deformation continues ( d e ). The neck propagation takes place under approximately steady-state conditions. The steady-state value of

Fig. 10 Comparison of simulated and experimental values of the nominal stress during neck propagation at room temperature as a function of the nominal strain rate using different values of D . The lines are tted through the results of the simulations.

Fig. 11 Simulated and experimental values of the nominal stress during neck propagation versus nominal strain rate at different temperatures using D 36. The lines are tted through the results of the simulations.

Fig. 8 Denition of the geometry of the longitudinal crosssection of the axisymmetrical tensile bar

the nominal stress during propagation of the neck proved to be independent of the geometry of the initial imperfection. The level of the draw ratio, on the other hand, was slightly inuenced. The simulated levels of the nominal stress during neck propagation are compared to experimental data taken at room temperature in Fig. 10. Using the parameter set determined in the previous section, it proved impossible to predict the right stress level see Fig. 10, D 28 - . If it is assumed that the constitutive model is adequate, the quantication of the parameter set is indicated as the source of this discrepancy. Apparently, the friction between the compression platens and the sample also inuenced the results of the compression test at low strain levels. To improve the description of the material behavior, the value of the softening parameter D was varied see Fig. 10. A good description of the experimental data was obtained with a value of D 36 - . To check this value, the simulations were repeated at temperatures of 305 and 312 K and compared to the experimental values of the nominal stress level during neck propagation at these temperatures. As can be observed in Fig. 11, the value D 36 - yields a reasonable description for all temperatures. This value was therefore adopted for the numerical simulations in the next sections. 5.2 The Mechanical Pretreatment: Mechanical Rejuvenation. The geometry of the axisymmetric polycarbonate samples that were preconditioned by one cycle of xed-end torsion was shown in Fig. 4. The cylindrical surface of the bar is traction free, Transactions of the ASME

Fig. 9 Simulated tensile response of polycarbonate in terms of the nominal stress 0 zz and the draw ratio in the neck N versus the nominal strain 0 0 zz at a nominal strain rate zz 7.5 3 1 " 10 s . The deformed meshes at different stages of the simulation a e are also shown.

182 Vol. 122, APRIL 2000

and it is assumed that axial displacements are negligible. The twist ( t ), dependent on the time t and dened per unit length of the bar, is applied at a low, constant, angular velocity . Consequently, temperature effects have not been taken into account and the simulations were performed under isothermal conditions. The applied twist ( t ) leads to a torque M ( t ), and an axial force F ( t ) resulting from the axially constrained ends. To simulate the mechanical rejuvenation treatment a dedicated nite difference scheme was developed, assuming the relevant stress and strain quantities to be only a function of the radius and the loading time homogeneous deformation over the length of the bar. The axisymmetric specimens were subjected to one full cycle of large strain torsion. First, the specimen was deformed up to a maximum twist dened per unit length of 0.25 rad mm1 , which was applied at a constant angular velocity equal to 9.1 10 4 rad s1 mm1 . After this, the direction of the twist was reversed, deforming the sample to a small negative twist value. Upon unloading of the sample, the cylindrical outer surface of the specimen approximately regains its initial state, which was veried during the experiments by monitoring a reference line on the sample. Figure 12a shows the variation of the torque M as a function of the applied twist during the applied process history. Path ABC corresponds to the loading stage of the rejuvenation experiment and shows clearly the effect of intrinsic softening as the torque decreases after the yield point B has been reached. After a subsequent twist of about 0.1 rad mm1 the torque level increases again as strain hardening sets in. The reversed twisting stage is represented by path CDE. The elastic unloading is shown by path EF. During this stage the cylindrical surface of the specimen regains its original geometry. Figure 12b shows the distribution of the softening parameter D over the dimensionless radius r / R a R a is the actual outer radius at stages C and F in Fig. 12a. It is clear that after the loading path ABC the softening parameter D in the outer layer of the specimen has almost reached its saturation level of D 36. After the return twist, path CDEF, the softening parameter reaches its saturation level over 0.4 r / R a 1 indicating that the intrinsic softening effect has been removed over approximately 84 percent of the specimen volume. The effect of this rejuvenation on a subsequent twist is shown in Fig. 12a, path FGH, where it is clear that a the onset of yield begins at a considerably smaller torque than for the original material and b no intrinsic softening is observed. The distribution of the stress components over the dimensionless radius at stage C and of the residual stress components at the end of the rejuvenation by xed-end torsion at stage F are shown in Fig. 13. It is clear that at the end of the mechanical pretreatment the residual stress level of most components is negligible, whereas the component z has a relatively high value in the core of the

Fig. 13 Distribution of the relevant components of the Cauchy stress tensor over the dimensionless radius at stage C and the residual stress distribution after mechanical preconditioning at stage F in Fig. 12a

specimen. Although these residual stresses may effect the subsequent deformation behavior of the rejuvenated sample through the stress dependence of the viscosity Eq. 14, the elevated levels of z only inuence a small part of the specimen approx. 10 percent of the volume. In order to minimize the inuence of residual stresses in the experiments performed, the samples were allowed to relax unconstrained for a period of 103 104 s before subsequent mechanical experiments were performed, thus allowing the residual stress levels to decrease. Therefore, residual stresses will not be taken into account in simulations of tension and torsion tests on rejuvenated samples. Only the distribution of the softening parameter D over the radius after the rejuvenation, depicted by stage F in Fig. 12b, will be used to characterize the state of the rejuvenated material. 5.3 Fixed-End Torsion of Rejuvenated Polycarbonate. The xed-end torsion simulations on rejuvenated samples were realized with a constant twist rate of 910 4 rad s 1 mm 1 at room temperature to a maximum twist of 0.4 rad mm1. The simulations were performed employing the same method as in the previous section, whereas the prole of the softening factor D, depicted in Fig. 12b at stage F, was taken as the initial situation. It should be noted that the residual stresses were not taken into account. In Fig. 14a the variation of torque as a function of the twist is given for both the simulation and the experiment. Both compare well, which is a support for the values of D 36 - and H 29 MPa determined previously. The small deviations at lower twist levels, 0.05 rad mm 1 , are attributed to viscoelastic effects, which, as already stated before 30 are not adequately addressed in the single mode compressible Leonov model.

Fig. 12 Results of simulations of the mechanical preconditioning by torsion at an applied twist rate 9.1 " 10 4 rad s1 mm 1 at room temperature. a Torque versus twist per unit length and b distribution of the softening parameter D over the dimensionless radius at stages C and F in a.

Fig. 14 Simulated and experimental curves of a torque versus twist per unit length and b compressive normal force versus twist per unit length for rejuvenated polycarbonate

Journal of Engineering Materials and Technology

APRIL 2000, Vol. 122 183

Figure 14b shows a comparison of the experimental and numerical results of the normal force versus the twist , which is a more critical assessment of the constitutive model. It is clear that there are strong deviations between simulation and experiment. Again it should be emphasized that the viscoelastic behavior at low twist levels is not accurately described by the single mode compressible Leonov model. Also, at higher twist levels the model is not able to capture these secondary effects accurately, although the description improves and the deviations are, strangely enough, observed to be constant with a value of approximately 100 N for 0.2 0.4. 5.4 Uniaxial Extension of Rejuvenated Polycarbonate. For the simulation of uniaxial extension of a rejuvenated sample of polycarbonate, again the nite element method is employed. In the nite element model, the distribution of the softening parameter D over the radius, depicted in Fig. 12b at stage F is specied as the initial condition. The distribution has been restricted to the gauge section of the specimen, since torsional deformations were only observed in this part. In the experiment and in the numerical simulation, the deformation rate imposed at the free end of the specimen corresponded with a nominal strain rate in the gauge section of 2.25 10 3 s 1 . To possibly trigger necking, exactly the same imperfection was used as in the tensile simulations of the untreated samples. For the present problem it was found that only meshes composed of ten elements or more in radial direction facilitated an accurate description of the initial distribution of D over the radius. Since during these experiments deformations were observed over the entire sample length, the mesh is equally distributed over the specimen. Figure 15 shows the simulated and experimental true stress zz 2 1 versus zz zz for the mechanically rejuvenated polycarbonate tensile bar at a nominal strain rate of 2.25 10 3 s 1 . Small deviations between simulations and experiment can be observed,

2 1 zz may be attributed to inaccurawhich at small values of zz cies in the calculated value of D after rejuvenation by torsion. Other causes for the differences between simulated and experimental results may be viscoelastic effects, where the mechanical pretreatment might still inuence the deformation behavior through a memory effect. As mentioned before, the single mode model used here will not capture these effects adequately. Figure 15 also includes the deformed meshes at different stages of the deformation, which conrms the absence of necking, despite the presence of the imperfection. In contrast to the ndings of Lu and Ravi-Chandar 32, this result strongly suggests that, in polycarbonate, strain softening is the main reason for localization phenomena.

Conclusion

In this study an extension of the compressible Leonov model has been presented, that captures the typical characteristics of the post-yield behavior of glassy polymers: intrinsic strain softening and strain hardening. Regarding the experimental assessment of the parameters needed for the model, it was found that the postyield behavior during a compression test was too strongly inuenced by barreling. Mechanical rejuvenation by cyclic xed-end torsion has been simulated, and the results have been used as initial conditions for numerical simulations of xed-end torsion and uniaxial extension of rejuvenated polycarbonate. Comparison of these simulations with the experiments shows that although the post-yield behavior is described correctly by the compressible Leonov model, the deformation behavior at small strains is not captured. This is especially observed when comparing the axial forces during xed-end torsion on rejuvenated samples. Although at higher strain levels there appeared to be a qualitative agreement, the behavior at low strain levels deviated strongly. It can therefore be concluded that the single mode compressible Leonov model is not valid with respect to second order effects during strain hardening. On the other hand, the compressible Leonov model, extended with intrinsic strain softening and strain hardening, seems to be able to predict the transition from inhomogeneous necking to homogeneous deformation as a result of a mechanical pretreatment. It is therefore concluded that the intrinsic strain softening effect is the main cause for localization phenomena in polycarbonate.

References
1 GSell, C., Hiver, J. M., Dahouin, A., and Souahi, A., 1992, Video Controlled Tensile Testing of Polymers and Metals Beyond the Necking Point, J. Mater. Sci., 27, p. 5031. 2 Boyce, M. C., Arruda, E. M., and Jayachandran, R., 1994, The Large Strain Compression, Tension, and Simple Shear of Polycarbonate, Polym. Eng. Sci., 34, pp. 716725. 3 Arruda, E. M., and Boyce, M. C., 1993, Evolution of Plastic Anisotropy in Amorphous Polymers During Finite Straining, Int. J. Plast., 9, pp. 697720. 4 Ward, I. M., 1983, Mechanical Properties of Solid Polymers, 2nd ed., Wiley, Chichester. 5 Haward, R. N., and Young, R. J., 1997, The Physics of Glassy Polymers, 2nd ed., Chapman & Hall, London. 6 Wu, P. D., and van der Giessen, E., 1993, On Improved Network Models for Rubber Elasticity and Their Applications to Orientation Hardening in Glassy Polymers, J. Mech. Phys. Solids, 41, pp. 427456. 7 Haward, R. N., 1993, Strain Hardening of Thermoplastics, Macromolecules, 26, pp. 58605869. 8 Struik, L. C. E., 1978, Physical Aging in Amorphous Polymers and Other Materials, Elsevier, Amsterdam. 9 Struik, L. C. E., 1986, Physical Aging: Inuence on the Deformation Behavior of Glassy Polymers, Failure of Plastics, Brostow, W., and Corneliussen, R. D., eds., Hanser Publishers, Munich. 10 Hasan, O. A., and Boyce, M. C., 1993, Energy Storage During Deformation in Glasses, Polymer, 34, pp. 50855092. 11 GSell, C., 1986, Plastic Deformation of Glassy Polymers: Constitutive Equations and Macromolecular Mechanisms, H. J. Queen et al., eds., Strength of Metals and Alloys, Pergamon Press, Oxford, pp. 19431982. 12 Aboulfaraj, M., GSell, C., Mangelinck, D., and McKenna, G. B., 1994,

Fig. 15 Simulated and experimental true stress zz versus 1 2 during a tensile test at a strain rate of 2.25 zz zz 3 " 10 s 1 of a polycarbonate tensile bar, preconditioned in torsion. The indications a d are related to the deformed meshes at different stages of the deformation.

184 Vol. 122, APRIL 2000

Transactions of the ASME

13 14

15 16 17 18

19 20 21

Physical Aging of Epoxy Networks After Quenching and/or Plastic Cycling, J. Non-Cryst. Solids, 172 174, pp. 615621. Cross, A., and Haward, R. N., 1978, Orientation Hardening of PVC, Polymer, 19, pp. 677682. Xie, L., Gidley, D. W., Hristov, A., and Yee, A. F., 1995, Evolution of Nanometer Voids in Polycarbonate Under Mechanical Stress and Thermal Expansion Using Positron Spectroscopy, J. Polym. Sci.: Part B: Polym. Phys., 33, pp. 7784. Hasan, O. A., Boyce, M. C., Li, X. S., Berko, S., 1993, An Investigation of the Yield and Postyield Behavior and Corresponding Structure of PolyMethyl Methacrylate, J. Polym. Sci., Part B: Polym. Phys., 31, pp. 185197. Wu, P. D., and van der Giessen, E., 1995, On Neck Propagation in Amorphous Glassy Polymers Under Plane Strain Tension, Int. J. Plast., 11, pp. 211235. Bauwens, J. C., 1978, A New Approach to Describe the Tensile Stress-Strain Curve of a Glassy Polymer, J. Mater. Sci., 13, pp. 14431448. Govaert, L. E., van Aert, C. A. C., Boekholt, J., 1997, Temperature and Molecular Weight Dependence of the Strain Hardening Behavior of Polycarbonate, Proc. 10th Int. Conf. On Deformation, Yield and Fracture of Polymers, The Institute of Materials, pp. 424426. Tervoort, T. A., Smit, R. J. M., Brekelmans, W. A. M., and Govaert, L. E., 1998, A Constitutive Equation for the Elasto-Viscoplastic Deformation of Glassy Polymers, Mech. Time-Dep. Mat, 1, pp. 269291. Hunter, S. C., 1983, Mechanics of Continuous Media, 2nd ed., Ellis Horwood Ltd., Chichester, U.K. Boyce, M. C., Parks, D. M., and Argon, A. S., 1988, Large Inelastic Deformation of Glassy Polymers. Part I: Rate Dependent Constitutive Model, Mech. Mater., 7, pp. 1533.

22 Bird, R., Armstrong, R., and Hassager, O., 1987, Dynamics of Polymer Liquids, Vol. 1: Fluid Mechanics, Wiley, New York. ginger, B., and Fritz, U., 1991, Thermal Properties of Strained Thermo23 Mo plastic Polymers, Polym. Int., 26, pp. 121128. 24 Koenen, J. A., 1992, Observation of the Heat Exchange During Deformation Using an Infrared Camera, Polymer, 33, pp. 47324736. 25 Boyce, M. C., Montagut, E. L., and Argon, A. S., 1992, The Effects of Thermomechanical Coupling on the Cold Drawing Process of Glassy Polymers, Polym. Eng. Sci., 32, pp. 10731085. 26 Van Krevelen, D. W., 1990, Properties of Polymers: Their Correlation with Chemical Structure, Their Numerical Estimation and Prediction from Additive Group Contributions, 3rd ed., Elsevier, Amsterdam. 27 Wu, W., and Turner, A. P. L., 1973, Shear Bands in Polycarbonate, J. Polym. Sci., Polym. Phys. Ed., 11, pp. 21992208. 28 Duckett, R. A., Goswami, B. C., Smith, L. S. A., Ward, I. M., and Zihlif, A. M., 1978, The Yielding and Crazing Behavior of Polycarbonate in Torsion Under Superimposed Hydrostatic Pressure, Br. Polym J., 10, pp. 1116. ` s, G., 1969, Tensile Yield 29 Bauwens-Crowet, C., Bauwens, J. C., and Home Stress Behavior of Glassy Polymers, J. Polym. Sci.: Part A-2, 7, pp. 735 742. 30 Tervoort, T. A., Klompen, E. T. J., and Govaert, L. E., 1996, A Multi-Mode Approach to Finite, Three-Dimensional, Nonlinear Viscoelastic Behavior of Polymer Glasses, J. Rheol., 40, pp. 779797. 31 Hasan, O. A., and Boyce, M. C., 1995, A Constitutive Model for the Nonlinear Viscoelastic Viscoplastic Behavior of Glassy Polymers, Polym. Eng. Sci., 35, pp. 331344. 32 Lu, J., and Ravi-Chandar, K., 1999, Inelastic Deformation and Localization in polycarbonate Under Tension, Int. J. Solids Struct., 36, pp. 391425.

Journal of Engineering Materials and Technology

APRIL 2000, Vol. 122 185

Hisaaki Tobushi
e-mail: tobushi@me.aitech.ac.jp

Takafumi Nakahara Yoshirou Shimeno

Department of Mechanical Engineering, Aichi Institute of Technology, 1247 Yachigusa, Yagusa-cho, Toyota 470-0392 Japan

Low-Cycle Fatigue of TiNi Shape Memory Alloy and Formulation of Fatigue Life
The low-cycle fatigue of a TiNi shape memory alloy was investigated by the rotatingbending fatigue tests in air, in water and in silicone oil. (1) The inuence of corrosion fatigue in water does not appear in the region of low-cycle fatigue. (2) The temperature rise measured through an infrared thermograph during the fatigue test in air is four times as large as that measured through a thermocouple. (3) The fatigue life at an elevated temperature in air coincides with the fatigue life at the same elevated temperature in water. (4) The shape memory processing temperature does not affect the fatigue life. (5) The fatigue equation is proposed to describe the fatigue life depending on strain amplitude, temperature and frequency. The fatigue life is estimated well by the proposed equation. S0094-42890001102-6 Keywords: Shape Memory Alloy, Fatigue, Titanium-Nickel Alloy, Martensitic Transformation, Rotating-Bending, Low-Cycle Fatigue, Frequency, Fatigue Equation, Atmosphere

Takahiro Hashimoto
Takiron Co., Ltd., 2-3-13 Azuchi-cho, Chuo-ku, Osaka 541-0052 Japan

Introduction

The shape memory effect and superelasticity appear in a shape memory alloy SMA Perkins 1, Funakubo 2, Doyama et al. 3, Chu and Tu 4. The development of intelligent materials and machine systems using these properties has attracted interest Pelton et al. 5. In applications of SMAs, we use not only shape recovery but also recovery force, energy storage, and dissipation of work. In applications to a robot, an actuator, and a solid-state heat engine, a SMA element performs cyclic motions. In order to evaluate the reliability of the SMA element, fatigue properties of the material are important. SMAs are subjected to both mechanical cycle with cyclic loading and thermal cycle due to heating and cooling. The thermomechanical cycle affects fatigue properties signicantly McNichols and Brookes 6, Melton and Mercier 7, Miyazaki 8. From the viewpoint of thermal response of SMA elements, thin wires are widely used in practical applications. The SMA elements working under a certain constant stroke are more widely used than those subjected to stress control. Considering these points, the authors carried out the rotating-bending fatigue tests of TiNi SMA wires subjected to strain control and ascertained that the fatigue life is longer than 107 cycles in the rhombohedralphase transformation RPT region Otsuka 9, Miyazaki and Otsuka 10, Tobushi et al. 11, Tobushi et al. 12, Tobushi et al. 13. They conrmed also that the low-cycle fatigue properties associated with the martensitic transformation MT are different between in air and in water. The fatigue life in air shortens due to temperature rise. In the previous paper Tobushi et al. 12, temperature was measured through a thermocouple. Therefore temperature was not measured with high accuracy. In this work, by carrying out rotating-bending fatigue tests of TiNi SMA wires, the relationship between strain amplitude and fatigue life is investigated. The fatigue life between in water and in silicone oil is compared and the inuence of corrosion fatigue in water is discussed. By measuring temperature rise of the wires
Contributed by the Materials Division for publication in the JOURNAL OF ENGINEERING MATERIALS AND TECHNOLOGY. Manuscript received by the Materials Division July 27, 1998; revised manuscript received November 8, 1999. Associate Technical Editor: C. Brinson.

in air through an infrared thermograph, the main cause of the temperature rise that markedly affects the fatigue life is discussed. Inuence of frequency and shape-memory processing temperature on fatigue life is also investigated. Based on the results, the lowcycle fatigue life is formulated.

Experimental Method

2.1 Materials and Specimen. The material was a Ti-55.4 wt percent Ni SMA wire, 0.75 mm in diameter. The specimens were given shape memory of a straight line through shape memory processing SMP. This was done by holding the wires in the straight line at a certain temperature for 60 min and cooling in a furnace. The specimens were straight lines with uniform cross section. The reverse-transformation completion temperature A f was about 323 K at a SMP temperature of 673 K. 2.2 Experimental Apparatus. The SMA properties testing machine composed of a tensile machine and a heating-cooling device Tobushi et al. 14 was used for the tensile test. The specimen was heated by hot air or cooled by liqueed carbon dioxide. Temperature was measured through a thermocouple, 0.1 mm in diameter, put on the surface of the specimen. The gauge length of the extensometer was 20 mm. The rotating-bending fatigue machine Tobushi et al. 12, Tobushi et al. 13 was used for testing fatigue. One end of the specimen bent to a prescribed curvature was mounted to a motor and the other end, at which the number of cycles to failure was measured, rotated freely. By keeping the bent form during the experiment, the specimen was rotated. The surface element of the specimen was subjected to tension and compression during one revolution. Maximum bending strain of the surface of the specimen was prescribed by using the radius of curvature in the center between the supports. The rupture occurred in the central part of the specimens between the supports. Maximum bending strain of the ruptured specimens was obtained by using the radius of curvature of a ruptured part. The infrared thermograph was used to measure temperature of the specimen during the fatigue test in air. By receiving the infrared energy radiated from the specimen through the thermograph, temperature at each part was measured without contact. Transactions of the ASME

186 Vol. 122, APRIL 2000

Copyright 2000 by ASME

2.3 Experimental Procedure. To investigate the rotatingbending fatigue properties of a TiNi SMA wire and temperature rise during the fatigue test, the following four experiments were carried out. (1) Fatigue Test in Liquid. The fatigue properties were investigated in water and in silicone oil. Test temperature was 303 K, frequency 1000 cpm and SMP temperature 673 K. (2) Test for Temperature Rise During Rotating-Bending. The temperature rise of the specimen in air was measured during the rotating-bending test. The test was carried out at various values of frequency and strain amplitude at room temperature. The temperature was measured at every 10 s for 3 min from the start of the test. (3) Fatigue Test at Various Frequencies. The fatigue properties were investigated at various frequencies in air and in water. The value of frequency f was 1001000 cpm. The test temperatures were room temperature (RT 303 K), 333 K and 353 K. The test in air was carried out in a state of natural radiation of heat. The value of strain amplitude a was 0.5 2.5 percent. The SMP temperature of the wire was 673 K. (4) Fatigue Test at Various SMP Temperatures. The fatigue properties were investigated in water for the specimens at various SMP temperatures. The SMP temperatures T p were 623 K, 673 K, and 723 K. Test temperature T was 333 K and frequency f 500 cpm.

3.2 Temperature Rise of Specimen in Air. The temperature distribution in the specimen obtained through the infrared thermograph is shown in Fig. 2. As seen in Fig. 2, the temperature rise is distributed almost uniformly in the whole specimen. The relationship between temperature rise of the specimen and lapse of time during the fatigue test at various frequencies f measured through the infrared thermograph is shown in Fig. 3. As seen in Fig. 3, temperature increases rapidly until 2030 s and is saturated in a certain value thereafter. Therefore the specimen is kept at an elevated temperature during almost the whole time. Temperature rise is large as f is large. In the case of f 1000 cpm, temperature rise is about 25 K. The relationship between the MT stress M and temperature T is expressed by the following equation called the transformation line Tanaka et al. 15

MC M T M s

(1)

where C M and M s denote a slope of the transformation line and the MT starting temperature under no stress, respectively. As found from Eq. 1, M increases in proportion to T. For TiNi SMA, C M is 6 MPa/K Tobushi et al. 16. If temperature increases by 25 K, the MT stress increases by about 150 MPa. Therefore temperature rise markedly affects the fatigue properties. The relationship between the saturated temperature rise T RT and frequency is shown in Fig. 4. In the case of a 2.04 percent and f 1000 cpm, because the specimen was ruptured at several ten seconds, the saturated temperature rise is not plotted. As seen in Fig. 4, temperature rise increases with increasing both frequency and strain amplitude. The relationship is not

Experimental Results and Discussion

3.1 Inuence of Liquid Atmosphere on Fatigue Life. The relationship between strain amplitude a and the number of cycles to failure N f obtained by the fatigue test in water and in silicone oil is shown in Fig. 1. As seen in Fig. 1, the distinct difference in fatigue life between in water and in silicone oil does not appear. It may be assumed that inuence of corrosion on fatigue appears in water and therefore fatigue life shortens in water. However, in the range of low-cycle fatigue, the time to failure is short and failure may occur before the inuence of corrosion fatigue appears. Because coefcient of heat transfer in liquid is larger than that in gas, temperature rise in water and in silicone oil is small compared with that in air. Therefore the difference in fatigue life does not clearly appear between in water and in silicone oil. As observed in the previous paper, because the low-cycle fatigue properties of SMA differs markedly between in air and in water, the inuence of test atmosphere on fatigue life is important in applications. In the case of using shape memory effect in applications, because SMAs are subjected to loading-unloading and heating-cooling, it is also necessary to take account of the inuence of these thermomechanical paths. In the case of high-cycle fatigue, it is necessary to take account of the inuence of fatigue limit and corrosion fatigue which depends on atmosphere. These points are future problems.

Fig. 2 Temperature distribution in the specimen obtained through the thermograph a 1.54 percent, f 1000 cpm, t 60 s

Fig. 1 Strain amplitude versus fatigue life in water and in silicone oil

Fig. 3 Temperature rise with lapse of time during fatigue test a 1.54 percent

Journal of Engineering Materials and Technology

APRIL 2000, Vol. 122 187

Fig. 4 Relationship between saturated temperature rise T RT and frequency

proportional. As found from the relationship, the rate of temperature rise decreases with increasing frequency. The temperature rise shown in Fig. 4 is almost four times as large as that measured through a thermocouple in the previous paper. The reason why temperature increases in air can be explained as follows. The surface elements of the wires under rotatingbending are subjected to tension and compression repeatedly. The stress-strain curve of SMA shows a hysteresis loop with wide width. The area surrounded by the hysteresis loop of the stressstrain curve represents dissipated work per unit volume. The dissipated work is lost as heat, resulting in temperature rise of the specimen. In air, because coefcient of heat transfer is small and heat is hardly diffused, temperature increases due to the generated heat. The temperature rise can be well estimated by considering the heat transfer characteristics around the specimen Mikuriya et al. 17.

3.3 Fatigue Properties at Various Frequencies in Air. The relationship between strain amplitude a and the number of cycles to failure N f obtained by the fatigue test at various frequencies f in air is shown in Fig. 5. As seen in Fig. 5, fatigue life depends markedly on frequency and shortens with increasing f. This is different from both the fatigue property of SMA in water that fatigue life does not depend on frequency Tobushi et al. 12 and that of stainless steel that fatigue life degrades with decreasing frequency Cofn 18. The slope of the curves becomes gentle with increasing f. The difference in N f due to the difference in f is large in the region of large a . For a 2 percent, N f at f 100 cpm is more than ten times larger than that at f 1000 cpm. These properties appear due to temperature rise of the specimen. On the other hand, the inuence of f on N f decreases in the region of small a . Therefore the strain amplitude at fatigue limit may depend little on frequency. This is due to the fact that strain amplitude at fatigue limit is in the region of the RPT Tobushi et al. 12 and that temperature increases little at small a which was shown in Section 3.2, resulting in slight inuence of f on N f . The fatigue life at an elevated temperature in air coincides with the fatigue life at the same elevated temperature in water. 3.4 Inuence of SMP Temperature on Fatigue Properties. The relationship between strain amplitude a and the number of cycles to failure N f obtained by the fatigue test at various SMP temperatures T p is shown in Fig. 6. As seen in Fig. 6, the strainlife relationships at each T p are almost the same. The slope of the approximate line in the low-cycle region is 0.49 at T p 723 K, 0.47 at T p 673 K, and 0.48 at T p 623 K. The strain-life relationship at each T p has a knee in the region of a 0.50.7 percent and N f 104 105 . The a N f relationship approaches the horizontal line in the region of a smaller than the knee. The stress-strain curves obtained by the tensile test for the specimens at various T p are shown in Fig. 7. The tensile test was carried out at the same temperature T 333 K as the fatigue test. As seen in Fig. 7, the MT stress increases with decreasing T p . Because dislocations with high density induced through cold drawing to produce a SMA wire construct the stable internal

Fig. 6 Strain amplitude versus fatigue life for various shapememory processing temperatures T p

Fig. 5 Strain amplitude versus fatigue life at various frequencies f in air

Fig. 7 Stress-strain curves for various shape-memory processing temperatures T p

188 Vol. 122, APRIL 2000

Transactions of the ASME

structure at low T p , the MT stress increases. Therefore it may be assumed that the fatigue life at low T p shortens because the MT stress is high. However, as observed in Fig. 6, the difference in fatigue life does not clearly appear. The reason why T p does not affect the fatigue life may be explained as follows. The dissipated work E d per unit volume under tension is obtained from the stress-strain curves shown in Fig. 7. The value of E d is 6.9 MJ/m3 at T p 723 K, 5.4 MJ/m3 at T p 673 K, and 4.2 MJ/m3 at T p 623 K. E d decreases with decreasing T p in inverse proportion to increase in the MT stress. If E d is large, fatigue damage is large, resulting in short fatigue life Sakuma et al. 19. Therefore, because the MT stress is low but E d is large at high T p , the fatigue properties do not clearly depend on T p due to both effects.

Formulation of Low-Cycle Fatigue Life

Let us discuss the formulation of the rotating-bending low cycle fatigue life of TiNi SMA wires for constant SMP temperature of 673 K. 4.1 Dependence on Strain Amplitude. As found from the strain-life curves in Section 3, the relationship between strain amplitude a and the number of cycles to failure N f in the region of low-cycle fatigue on a logarithmic graph is almost expressed by a straight line. Therefore, as similar as the Manson-Cofn relationship for normal metals in low-cycle fatigue, the relationship between a and N f is express as follows a N f (2) where and represent a in N f 1 and the slope of the log a log N f curve, respectively. The value of is about 0.5 at each temperature which is valid for normal metals. 4.2 Dependence on Temperature. The dependence of the fatigue life on temperature will be discussed in water where temperature of the specimen increases little. The exponent in Eq. 2 is about 0.5 at each temperature. The value of decreases with increasing temperature T. As expressed by Eq. 1, the MT stress increases in proportion to T. If the MT stress is high, fatigue damage is large and the fatigue life is short, resulting in small . Based on these considerations, if the relationship between and T is plotted on a semilogarithmic graph, it is found that the relationship is expressed by a straight line. Therefore it becomes as follows

Fig. 8 Relationship between saturated temperature rise T RT and frequency in a logarithmic scale

surface element under rotating-bending, strain which varies in one cycle for strain amplitude a is 4 a . Therefore average strain rate at frequency f is 4 a f 4 f a (6) From these relationships, E d becomes a function of a and f as follows E d E d a , f (7) Because temperature increases based on E d , temperature rise T is also a function of a and f TT a , f (8) Let us obtain the concrete form of Eq. 8. The relationship between temperature rise at room temperature T RT and frequency f is shown in Fig. 8 which is rearranged in a logarithmic scale from Fig. 4. As seen in Fig. 8, the relationship between log T RT and log f is expressed by a straight line. Therefore T RT becomes T RT

f f0

(9)

s 10 a T M s

(3)

where M s is 253 K which was obtained by the DSC test. Based on the results of the fatigue test in water, the coefcients in Eq. 3 are determined as s 8.56 and a 0.012 K1. Therefore, from Eq. 2, the relationship between a and N f is expressed by the following equation
0.012 T M s a N 0.5 f 8.56 10

where f 0 represents a value on the frequency axis. As seen in Fig. 8, f 0 20 cpm. In Fig. 8, b represents the slope of the straight line and changes depending on strain amplitude. The slope b increases in proportion to strain amplitude in a logarithmic scale. Therefore b is expressed as follows b c log a log l (10) where c 2.13 and l 0.006. Because the RPT strain is 0.6 percent, Eq. 10 reveals the fact that temperature increases markedly in the MT region of strain amplitude above l 0.6 percent. From Eqs. 9 and 10, temperature rise at room temperature is expressed as follows T RT

(4)

Thus the dependence of the fatigue life on a and T is described by Eq. 4. 4.3 Dependence of Temperature Rise in Air on Frequency. Because temperature increases little in water due to high heat transfer, temperature rise is ignored. On the contrary, temperature increases depending on strain amplitude and frequency in air, resulting in short fatigue life. The temperature rise appears due to dissipated work E d . In order to consider the fatigue properties in air, it is important to take account of the temperature rise. Based on tensile deformation properties of SMA, dissipated and maximum strain m Towork E d depends on strain rate bushi et al. 20. Therefore E d is expressed as follows E d E d ,m (5) is proportional where m corresponds to strain amplitude a and to frequency f in rotating-bending fatigue test. With respect to the Journal of Engineering Materials and Technology

f f0

c log a / l

(11)

4.4 Dependence on Temperature in Air. In order to obtain temperature rise at an arbitrary temperature, dissipated work E d at each temperature T will be discussed. E d is obtained from tensile test results for maximum strain of 2 percent. Assuming that stressstrain curves are symmetrical with respect to tension and compression, E d in one cycle is determined. The relationship between E d and T is shown in Fig. 9. As seen in Fig. 9, log Ed decreases in proportion to T at temperatures above T l 320 K. This means that E d decreases in the superelastic region because T l 320 K is close to the reverse-transformation completion temperature A f 323 K. Taking E d as unit in the region of temperature below APRIL 2000, Vol. 122 189

Fig. 11 Strain-life curves at various temperatures in air

Fig. 9 Relationship between dissipated work and temperature

320 K at which the shape memory effect is observed and room temperature exists, E d decreases at the rate of r at temperatures above 320 K. From Fig. 9, r is r 10 h T T l
1

(12)

where h 0.0054 K and T l 320 K. Temperature rise T appears based on E d . Therefore, compared with temperature rise at room temperature T RT , temperature rise T at temperatures in the superelastic region decreases at the rate of r as follows T T RT 10 h T T l From Eqs. 11 and 13, T becomes T (13)

region of low frequency and small strain amplitude. This occurs due to the assumption of Eq. 15 in which Eq. 4 in water is used as the condition of no temperature rise. As observed in the previous paper Tobushi et al. 12, in the high-cycle region, the fatigue life in water is shorter than that in air. If the state of air ow changes, coefcient of heat transfer changes and accordingly temperature rise changes. Therefore it is important to decide the fatigue life by considering carefully the atmosphere. Furthermore, it is necessary to formulate the fatigue limit for high-cycle applications. It is necessary to consider corrosion for the fatigue limit in water. These are future problems.

Conclusions

The low-cycle fatigue properties of TiNi SMA wires subjected to rotating-bending have been investigated in water and in air. The main results obtained are summarized as follows. 1 The inuence of corrosion fatigue in water does not appear in the region of low-cycle fatigue. 2 The temperature rise measured through an infrared thermograph during the fatigue test in air is four times as large as that measured through a thermocouple. 3 The fatigue life at an elevated temperature in air coincides with the fatigue life at the same elevated temperature in water. 4 Although stress-strain curves are different at various SMP temperatures, fatigue lives are almost the same. 5 The fatigue equation is proposed to describe the fatigue life depending on strain amplitude, temperature and frequency in water and in air. The fatigue life is estimated well by the proposed equation. The proposed fatigue equation is useful for the design of SMA elements.

f f0

c log a / l

10 h T T l

(14)

Considering temperature rise, the fatigue equation at an arbitrary temperature T in air becomes from Eq. 4 as follows
0.012 T T M s a N 0.5 f 8.56 10

(15)

The fatigue life of SMA wires in the low-cycle region in air is determined from Eq. 15. 4.5 Calculated Results and Discussion. The calculated results between strain amplitude and fatigue life in water obtained from Eq. 4 are shown in Fig. 10. In Fig. 10, solid curves represent the calculated results and plotted points the experimental results, respectively. As seen in Fig. 10, the low-cycle fatigue life in water is well expressed by Eq. 4. The calculated results between strain amplitude and fatigue life at various temperatures in air obtained from Eq. 15 are shown in Fig. 11. The calculated results at various frequencies in air are shown in Fig. 5. In both gures, the calculated results are shown by solid curves. As seen in Figs. 11 and 5, the overall fatigue life in the low-cycle region in air is expressed by the equation in which temperature rise is considered. As seen in Fig. 5, the calculated fatigue life is shorter than the experimental one in the

Acknowledgments
The experimental work of this study was carried out with the assistance of the students of Aichi Institute of Technology, to whom the authors wish to express their gratitude. The authors also wish to extend thanks to the Scientic Foundation of the Japanese Ministry of Education, Science and Culture for nancial support.

References
1 Perkins, J., 1975, Shape Memory Effects in Alloys, Plenum Press, New York. 2 Funakubo, H., 1987, Shape Memory Alloys, Gordon and Breach Science, New York. 3 Doyama, M., Somiya, S., and Chang, R. P. H., 1989, Shape Memory Materials, MRS Intl. Mtg. on Adv. Mats., Vol. 9, Materials Research Society, Pittsburgh. 4 Chu, Y. Y., and Tu, H. L., 1994, Shape Memory Materials 94, Proc. of Inter. Sym. on Shape Memory Materials, International Academic, Beijing. 5 Pelton, A. R., Hodgson, D., and Duerig, T., 1995, SMST-94, Proc. of First Inter. Conf. on Shape Memory & Superelastic Technologies, MIAS, Monterey. 6 McNichols, J. L., Jr., and Brookes, P. C., 1981, NiTi Fatigue Behavior, J. Appl. Phys., 52, No. 12, pp. 74427444. 7 Melton, K. N., and Mercier, O., 1979, Fatigue of NiTi Thermoelastic Martensites, Acta Metallurgica, 27, pp. 137144. 8 Miyazaki, S., 1996, Development and Characterization of Shape Memory Alloys, Shape Memory Alloys, M. Fremond and S. Miyazaki, eds., Springer Wien, New York, pp. 69147.

Fig. 10 Strain-life curves at various temperatures in water

190 Vol. 122, APRIL 2000

Transactions of the ASME

9 Otsuka, K., 1990, Introduction to the R-Phase Transition, Engineering Aspects of Shape Memory Alloys, T. W. Duerig, K. N. Melton, D. Stockel, and C. M. Wagman, eds., Butterworth-Heinemann, London, pp. 3645. 10 Miyazaki, S., and Otsuka, K., 1986, Deformation and Transition Behavior Associated with the R-Phase in Ti-Ni Alloys, Metall. Trans. A, 17, pp. 53 63. 11 Tobushi, H., Yamada, S., Hachisuka, T., Ikai, A., and Tanaka, K., 1996, Thermomechanical Properties Due to Martensitic and R-Phase Transformations of TiNi Shape Memory Alloy Subjected to Cyclic Loadings, Smart Mater. Struct., 5, pp. 788795. 12 Tobushi, H., Hachisuka, T., Yamada, S., and Lin, P. H., 1997, RotatingBending Fatigue of a TiNi Shape-Memory Alloy Wire, Mech. Mater., 26, pp. 3542. 13 Tobushi, H., Hachisuka, T., Hashimoto, T., and Yamada, S., 1998, Cyclic Deformation and Fatigue of a TiNi Shape-Memory Alloy Wire Subjected to Rotating Bending, ASME J. Eng. Mater. Technol., 120, pp. 6470. 14 Tobushi, H., Tanaka, K., Kimura, K., Hori, T., and Sawada, T., 1992, StressStrain-Temperature Relationship Associated with the R-Phase Transformation in TiNi Shape Memory Alloy, JSME Int. J. Ser. I, 35, No. 3, pp. 278284.

15 Tanaka, K., Kobayashi, S., and Sato, Y., 1986, Thermomechanics of Transformation Pseudoelasticity and Shape Memory Effect in Alloys, Int. J. Plast., 2, pp. 5972. 16 Tobushi, H., Lin, P. H., Tanaka, K., Lexcellent, C., and Ikai, A., 1995, Deformation Properties of TiNi Shape Memory Alloy, J. Phys. IV, C2, No. 5, pp. 409413. 17 Mikuriya, S., Nakahara, T., Tobushi, H., and Watanabe, H., 1999, The Estimation of Temperature Rise on Low Cycle Fatigue of TiNi Shape Memory Alloy, Trans. Jpn. Soc. Mech. Eng., Ser. A, 65, No. 633, pp. 10991104. 18 Cofn, L. F., 1978, Fatigue in Machines and Structures-Power Generation, Materials Science Seminar, Fatigue and Microstructures, St. Louis, American Society for Metals, pp. 128. 19 Sakuma, T., Iwada, U., Kariya, N., and Ochi, Y., 1998, Fatigue Life of TiNiCu Shape Memory Alloy under Thermo-mechanical Conditions, Proc. of 11th Inter. Conf. on Exp. Mech., Oxford, Vol. 2, pp. 11211126. 20 Tobushi, H., Shimeno, Y., Hachisuka, T., and Tanaka, K., 1998, Inuence of Strain Rate on Superelastic Properties of TiNi Shape Memory Alloy, Mech. Mater., 30, pp. 141150.

Journal of Engineering Materials and Technology

APRIL 2000, Vol. 122 191

Fuh-Kuo Chen
Professor

On the Nonuniform Deformation of the Cylinder Compression Test

A theoretical model based on Hills general method is developed in the present study to calculate the ow stress in a cylindrical specimen under axial compression in the presence of friction at the die-specimen interface. Unlike most of the published methods which studied the incipient barreling only, the proposed theoretical model takes the barreled shape of the deforming specimen into account. In order to construct the stress-strain curve, the mean effective strain of the barreled specimen was also calculated on the basis of an assumed velocity eld. As the present study shows, the proposed theoretical model provides good results, both in magnitude and in trend, for the prediction of ow stresses in the barreled specimen during the compression test. S0094-42890000602-2

Cheng-Jun Chen
Graduate Student Department of Mechanical Engineering, National Taiwan University, Taipei, Taiwan

Introduction

The cylinder compression test has been widely adopted for determining a materials ow stress as a function of the strain during the compression of cylindrical specimens between two at dies. If the friction is absent at the die-specimen interface, the deformation in the cylinder is uniform and the free surface of the cylinder remains straight during the compression. Thus, the ow stress of the specimen is simply the value obtained by dividing the applied force P by the contact area of die-specimen interface A, and the effective strain is the same as the axial compressive strain given by ln(2Ho/2H ), where 2 H 0 and 2 H are the initial height and the instantaneous height of the deforming specimen, respectively. In reality, due to the presence of friction at the diespecimen interface, the radial displacement of the material near the interface is restrained while the remaining portion of the specimen bulges out changing the free surface into the form of a barrel. The barreled shape leads to an inhomogeneity of deformation, and therefore, the method used to calculate the ow stress and the effective strain for the uniformly deformed specimen is not applicable. Many theoretical methods have been proposed to derive the ow stress for a barreled specimen. Siebel 1 assumed no barreling and a low coefcient of friction, and derived a simple wellknown formula for the relationship between the mean axial pressure acting at the die-specimen interface and the ow stress of the material under frictionless uniaxial compression. With the use of the upper-bound method, Avitzur 2 has determined a relationship between the applied load and ow stress of the material for a cylinder under uniaxial compression with friction at the diespecimen interface. Later, Lee and Altan 3 also proposed an upper-bound velocity eld that considers bulging of the cylinder during the upsetting. They developed computer programs to determine the strain, strain-rate, velocity, and ow-stress distributions. Although the previous works mentioned above have taken the effect of friction into consideration, the efforts were mainly focused on the incipient barreling of the free surface. In other words, the specimen with a straight free surface was used for the analysis, and no attempt was made to consider a barrel-shaped specimen for the analysis. Since the compression load acting on any cross-sectional area of the barreled specimen has the same value, another way to calculate the ow stress of the specimen may rest on the analysis for the mean cross sectional area. Following this concept, a simple expression for the ow stress Y is therefore suggested as
Contributed by the Materials Division for publication in the JOURNAL OF ENGINEERING MATERIALS AND TECHNOLOGY. Manuscript received by the Materials Division October 6, 1998; revised manuscript received February 13, 1999. Associate Technical Editor: G. Ravichandran.

P A

(1)

where A is the mean cross-sectional area obtained by dividing the volume of the specimen ( V ) by the instantaneous height 2 H of the specimen. This method is termed the mean ow stress method thereafter in this paper. In 1983, Ettouney and Hardt 4 extended the concept, which Bridgman 5 had developed for calculating the stress distribution at the neck of tensile specimen, to determine the ow stress of the barreled specimen. Considering the barreled shape in the analysis, they derived a formula of the following nature: Y zm where
z m

2R R2 ln 1 R2 2R

(2)

P , R2 2

Y is the ow stress of the material under frictionless uniaxial compression, P is the compression load, R is the radius of the bulge curvature, and R 2 is the maximum radius of the barreled specimen, as shown in Fig. 1. In the present study, a theoretical model based on Hills general method was developed to calculate the ow stress of the barrelshaped specimen, which is compressed under frictional condition. In the theoretical model, the longitudinal nonuniformity was under investigation, and an expression for the ow stress in terms of the compression load and the geometric parameters dening the barreled specimen was obtained. Finite element simulations were also performed in the present study in order to validate the theoretical calculation. The ow stresses calculated according to the proposed theoretical model are compared with the true ow stress and with those predicted by the Ettouney and Hardts formula termed the E-H formula for short, and the mean ow stress method, based on the compression loads and the geometric parameters obtained from the nite element simulations. It should be noted that the strain distribution of the barreled specimen is also not uniform and the effective strain cannot therefore be determined explicitly. Nevertheless, the overall axial compressive strain has been commonly used by researchers, without proof, to approximately represent the effective strain. In the present study, an admissible velocity eld was proposed to calculate the mean effective strain, and the detailed derivation is presented following the calculation of the ow stress. Transactions of the ASME

192 Vol. 122, APRIL 2000

Copyright 2000 by ASME

1 The deforming material is isotropic and incompressible. 2 The elastic deformation is negligible, i.e., the material is considered as rigid-plastic. Now, consider a barrel-shaped specimen under compression between two at dies which move toward each other with a unit speed at any stage of compression. The specimen has an instantaneous height of 2 H and a radius of R 1 at the die-specimen interface, as shown in Fig. 1. Since the nonuniform deformation in the z-direction is our main concern, the selected velocity eld must be a function of z. A simple velocity eld representing the incipient barreling of the specimen has been shown by Kobayashi et al. 8 to be in the form
v r 2 r z ,
1

v 0,

and v z z ,

(4)

Fig. 1 Congurations specimens

of

barrel-shaped

and

equivalent

Calculation of the Flow Stress

In 1963, Hill 6 proposed a new method of analysis for metalforming processes based on a variational principle. Although Hill believed that the proposed method comes close to the ideal, very few investigations on the method have appeared in the published literature. Among them, Lahoti and Kobayashi 7 carried out the analysis of ring compression with barreling, spread in Steckel rolling, and thickness change in tube sinking. Hills method begins by selecting a class of velocity eld v i from which the best approximation will eventually be taken. The selected velocity eld must satisfy all the kinematic conditions. However, the associated stress eld in the deformation zone, which is determined according to the material constitutive law, will not generally satisfy all the static requirements. Hence, the selection criterion for the best approximating stress eld i j may be considered as

where v r , v , v z are the velocities in the r, , and z directions, respectively, the prime denotes differentiation, is a function that is sufciently differentiable. The velocity eld given in Eq. 4 implies the incompressibility of material, while ( z ) ( z ), ( H ) ( H ) 1 due to the symmetry with respect to the mid-plane and the velocity boundary conditions, respectively. It is reasonable to suppose that the velocity eld given in Eq. 4 continues to hold during a nite compression of the barreled cylinder, and it is therefore adopted in the present study to calculate the ow stress for the barreled cylinder. One way to choose the virtual orthogonalizing velocities is just to take the similar form as that given in Eq. 4, such as
1 w r 2 r z ,

w 0,

and w z z ,

(5)

with ( z ) ( z ). Making use of the following conditions: S I 0, n r , n , n z 0,0,1 , and l r , l , l z 1,0,0 , Eq. 3 can be written as 2

H 0 A

wr wr wz wr wz z rz r r z z r

dA dz

z w z mkw r z H dA ,

(6)

ij

w j dV xi

j w j dS

SI

n i i n j mkl j w j dS (3)

noting that n j and l i are the unit normal vectors in the z and r directions, respectively. Substituting Eq. 5 into Eq. 6, yields 2

SC

for a sufciently wide subclass of virtual orthogonalizing motion w j , where i denotes the surface traction computed from the considered approximating stress eld i j , n i , and n j are the local unit outward normals, l j is a unit tangent vector opposite in sense to the relative velocity of sliding in the approximating eld, and mk represents the constant frictional stress with 0 m 1, k being the shear yield stress. For metal-forming process, the surface S of the deforming zone usually consists of three distinct parts: S C is the interface between the die and the workpiece; S F is the unconstrained surface; S I is the interface between the deforming zone and the rigid zone. It is to be noted that use of the reverse of the virtual work-rate principle has been made to derive Eq. 3, and although the method is applicable to all types of friction, constant friction is adopted in Eq. 3 for simplicity. In applying Eq. 3, the orthogonalizing family w j must be sufciently wide and extensive to identify a single approximating velocity eld in the particular class constructed for satisfying the kinematic conditions. Once the orthogonalizing family is chosen, the calculus of variations technique is applied to Eq. 3, treating w j as a variation parameter. This furnishes a system of equilibrium equations and boundary conditions, suited to the particular approximating class, allowing us to determine its best member Lahoti and Kobayashi 7. The detailed description of the method can be found in Hill 6. To facilitate the analysis, the following assumption are made: Journal of Engineering Materials and Technology

H 0 A

r rz r z dA dz 2 2 2

mkr 2

dA .
zH

(7)

Applying integration by parts and rearranging terms, Eq. 7 becomes

H 0 A

r 2 z r

rz dA dz z

H 0

rz r dA dz z

2 P H mk H

r dA ,

(8)

AH

where P is the compression load given by P A H z dA , and A H is the contact area at the die-specimen interface. Since the area of cross section of the barreled cylinder is itself a function of z, it is convenient to consider Eq. 8 in relation to an equivalent specimen having a uniform cross section with a mean area A V /(2 H ), where V is the volume of the specimen which remains constant during compression. The conguration of the equivalent specimen is also shown in Fig. 1 by the dashed-line, . With the above asthe equivalent radius being denoted by R APRIL 2000, Vol. 122 193

sumption in mind, the integrations with respect to dA and dz in Eq. 8 can then be interchanged. Consequently, the second integral on the left-hand side of Eq. 8 can be written as

Calculation of Effective Strain

H 0

rz r dA dz z A

A A

rz r dz dA z
(9)

rz H r H dA ,

Since the barreling of the cylinder leads to a nonuniform deformation, the stress-strain relation may be best represented by adopting the mean ow stress and the associated mean effective strain of the deforming cylinder. In order to determine the mean effective strain, the upper half of the barreled specimen, as shown in Fig. 1, was considered, with a velocity eld suggested by Chakrabarty 9 being adopted, which has the form v r H and v z H

with rz 0 at z 0. In the cylinder compression with friction at the die-specimen interface, Hill 6 has shown that the von Mises yield criterion can be taken in the form
1 z 2 r Y ,

r z 2H H

v 0,

(10)

to a close approximation. Substituting Eq. 9 and Eq. 10 into Eq. 8 and noting that P(H) PH 0 ( z ) dz , results in

z H z H r H

(19)

H 0 A

2Y r

H mk

rz dA 2 P dz z
r dA

AH

rz H r dA 0,
(11)

is the rate of change of the specimen height, is an where H arbitrary small coefcient representing the frictional condition at the die-specimen interface, and is an approximating function of z / H . Also, (0) (1) ( 1) 0. The arbitrariness of the function , which represents the barreling effect, makes the velocity eld exible enough to include the whole range of possible shapes of the barreled specimen. The velocity eld given in Eq. 19 characterizes the feature of longitudinal nonuniformity in the barreled specimen, and the associated strain rates are

where the rst and the third integrals are evaluated over the mean area of the barreled cylinder. Since is arbitrary, it follows from Eq. 11 that

vr H , r 2H

rz 2Y r dA 2 P 0, z

(12)

1 H z H vr r H 2 r H vz H H z 1 z H r H

and mk

R1

2 r 2 dr

z
R

rz z H 2 r 2 dr 0,

(13)

rz

, and R 1 is where R is the radius corresponding to the mean area A the radius of the surface of contact, as shown in Fig. 1. Integrating out Eq. 13, yields
rz z H

H z z 1 vr vz H2 2 2 z r 2H H r H

, ,

. (20)

R1 R

The corresponding expression for the effective strain rate is given by z 2 2 2 2 2H 2 2 z 2 rz 1 3 r r H

2

mk .

(14)

H H

, (21)

Integrating Eq. 12 with respect to z between the limits 0 to H, and noting the fact that rz 0 at z 0, gives

the terms of order being omitted since they are small compared to unity. This gives the effective strain rate as

2 Y H r rz z H dA 2 PH 0.

(15)

Evaluating Eq. 15 at z H and substituting from Eq. 14, we obtain the expression for the compression load as PY A

H H z 1 H r H

(22)

R3 1 mk
3H

(16)

R , and mk is replaced When the barreling is disregarded, i.e., R 1 with Y , Eq. 16 reduces to the well known Siebel formula. For the von Mises yield criterion, k Y / ) , Eq. 16 becomes PY A

with the terms of order 2 being omitted. The negative sign on the left-hand side of Eq. 22 is used to ensure a positive value for the effective strain rate. A mean effective strain rate may be dened with sufcient accuracy as

R3 1m
3)H

R 2H
H H H H

H 0

2 r dr dz

(17)

The expression of the ow stress therefore becomes YP A 194 Vol. 122, APRIL 2000

R3 1m
3)H

2 2 R

R H

dr

dz

(23)

(18)

Transactions of the ASME

since vanishes at both limits of integration, i.e., (0) (1) 0, R being the radius of the mean area. The total mean effective strain at any stage is therefore given by

dt

1 dH dt H dt

dH H0 ln , H H0 H

(24)

where 2 H 0 is the initial height of the underformed specimen. Equation 24 indicates that the mean effective strain for a barreled specimen is the same as in the case of compression without barreling to a rst order of approximation. This result provides the theoretical justication for using the axial compressive strain, ln(Ho /H), or ln(2Ho/2H ), as the effective strain to construct the stress-strain curve for a barreled specimen in the compression test.

obtained from the nite element simulations are used to calculate the ow stresses Y according to the proposed theoretical model, the E-H formula, and the mean ow stress method. The predicted ow stresses are then compared with each other and with the given material stress-strain relation for the required validation. The nite element program DEFORM was adopted to perform the simulations, and the program was run on a SGI R4000 workstation.

Results and Discussions

Finite Element Model

In an actual compression test, it is very difcult to determine the exact frictional condition at the interface between die face and specimen; while the coefcient of friction can be easily incorporated in the nite element method which has been considered as a well-developed technology for the analysis of metal forming processes. It is therefore convenient to perform the nite element simulations, instead of conducting experiments, to validate the proposed theoretical model. The procedure begins with the construction of the nite element model, as shown in Fig. 2. Because of symmetry, only a quarter of the specimen is analyzed. Both the top and bottom dies are considered as rigid and the specimen is modeled by 4-node axisymmetric elements, as shown in Fig. 2. It is seen in this gure that the mesh is very dense so that the prole of the deformed specimen can be determined with sufcient accuracy. The material constitutive relation of the specimen used in the simulations is Y K n with K 516 MPa and n 0.23, where is the effective plastic strain. Since the barreling of the specimen depends also on the ratio of the initial height (2 H o ) to the initial diameter ( d o ), known as the aspect ratio, in addition to the friction coefcient, specimens of 10 mm in diameter and of two different heights of 15 mm and 10 mm were used in the simulations, the corresponding aspect ratios being 1.5 and 1.0, respectively. The simulations were performed for both the specimen sizes under the frictional conditions specied by m 0, m 0.3, and m 0.5, respectively. The compression load and the geometric parameters of the deforming specimen, such as the radius of the contact area between die and specimen ( R 1 ), the maximum radius corresponding to the central section of the specimen ( R 2 ), and the instantaneous height of the specimen (2 H ), as shown in Fig. 1,

In order to validate the nite element simulation itself as an effective method for providing the necessary information for the compression loads and the geometric parameters of the barreled specimen, the simulation was rst performed for the specimen that is 10 mm in height (2 H o / d o 1.0) under frictionless condition, so that the simulation results can be compared with theoretical predictions. The free surface of the deforming cylinder, obtained from the nite element simulation, remains straight even under a large reduction in height to 4 mm. The ow stresses calculated according to the relation P / A at various effective strains or axial compressive strains during the compression are almost the same as those given by the stress-strain curve, as shown in Fig. 3, where P is the compression load, and A is the cross-sectional area of the specimen obtained from the nite element simulation. Although this good agreement between the simulation results and the theoretical predictions only validates the nite element simulations under the frictionless condition, it also implies the feasibility of using this method to simulate the cylinder compression in the presence of friction. It is therefore meaningful to compare the true ow stress given in the nite element method as input data with the ow stresses calculated according to the present theoretical model, the E-H formula, and the mean ow stress method, using the compression loads and the geometric parameters of the deforming specimen furnished by the nite element simulation results. The compression load at various height reduction ratios ( H o H )/ H o for both the specimens under different frictional condi-

Fig. 3 Comparison of the simulated ow stresses P A with the true ow stresses

Fig. 2 Initial mesh for the nite element simulation unit: mm

Fig. 4 Compression load at various height reduction ratios 2 H o d o 1.0

Journal of Engineering Materials and Technology

APRIL 2000, Vol. 122 195

Fig. 5 Radius of bulge prole at various effective strains 2 H o d o 1.0

Fig. 7 Comparison of ow stresses at various effective strains 2 H o d o 1.5; m 0.5

tions were obtained from the nite element simulations. The results for the specimen of aspect ratio of 1.0 are plotted in Fig. 4. As seen in the gure, the compression load increases as the coefcient of friction increases. In the E-H formula, the bulge prole is treated as a circular arc and the radius of the bulge curvature R is determined from an empirical formula Horton et al. 10 which has the form R
2 H 2 d 2 d 1 2 , 4 d 2 d 1

(25)

where d 1 , d 2 are the minimum and maximum diameters of the barreled specimen, respectively, and 2 H is the instantaneous height of the specimen. The nite element simulation results were utilized to calculate the corresponding radius of the bulge curvature for the specimens under different frictional condition according to Eq. 25. Figure 5 shows the calculated radii of the bulge prole versus the mean effective strain of the cylinder with an aspect ratio of 1.0 under constant shear friction given by m 0.1, 0.3, and 0.5, respectively. As seen in Fig. 5, the radius of the bulge prole decreases almost exponentially with the current height, and the specimen under higher friction coefcients has smaller radii of curvature as expected. With the compression loads and the geometric parameters generated from the nite element simulation results, the ow stresses calculated according to the proposed theoretical model, the E-H formula, and the mean ow stress method under various frictional conditions were evaluated. To limit the length of this paper, only the results of higher friction coefcients are discussed. The ow stress calculated from the three methods and the true ow stress are plotted against the mean effective strain in Figs. 6 and 7, respectively, for the specimen of aspect ratio 1.5 under the frictional conditions corresponding to m 0.3 and m 0.5. It is seen in Fig. 6 and Fig. 7 that the mean ow stress method shows good results over a range of low compressive strains in which the barreling is not signicant. However, when the cylinder is deformed

Fig. 6 Comparison of ow stresses at various effective strains 2 H o d o 1.5; m 0.3

0.8, the calculated mean ow to large effective strains, say stress deviates from the true ow stress and the difference becomes larger as the reduction increases, and the overestimated value of the ow stress at higher compressive strains becomes more signicant as the coefcient of friction increases. It follows ) is that the calculation based on the mean cross sectional area ( A not sufciently accurate, and the ow stress calculated from this method requires correction at higher reductions in height. It is also seen in Fig. 6 and Fig. 7 that the ow stresses predicted by the E-H formula agree reasonably well with the true ow stresses at lower compressive strains for both the frictional conditions of m 0.3 and m 0.5, but signicant differences are noted at compressive strains higher than 0.4. As noted in both the gures, the calculated ow stress according to the E-H formula increases very rapidly when the compressive strain is greater than 1.0 for m 0.3, and greater than 0.8 for m 0.5. The difference between the predicted ow stresses and the true ow stresses may be attributed to the magnication of error in the approximate formula Eq. 25 for the radius of curvature R of the bulge curvature. The rapid increase in the calculated ow stress is due to the fact that R 2 approaches the value of 2 R as the strain increases, which makes the term 1 R 2 /(2 R ) appearing in Eq. 2 tend to zero. When R 2 2 R the ow stress is not obtainable from Eq. 2. This suggests that the E-H formula is applicable only to cylinders of relatively large aspect ratios. The ow stresses calculated according to the proposed theoretical model for m 0.3 are also plotted in Fig. 6. It is seen in this gure that the theoretical model underestimates the ow stresses over a range of axial compressive strains lower than 1.2, and overestimates the ow stresses at higher axial compressive strains. However, no marked difference is observed almost throughout the compression process. Although the difference would tend to increase with the increasing compressive strains, the variation is small and the ow stress still agrees reasonably well with the true ow stress. When the friction coefcient is increased to m 0.5, the ow stress calculated by the proposed theoretical model agrees very well with the true ow stress, even for large compressive strains, as indicated in Fig. 7, while the other two theories result in signicant deviations from the true ow stress. Figure 7 also suggests that the proposed theoretical model would provide better results for specimens with signicant barreling. On the whole, the proposed theoretical model is an improvement over the E-H formula and the mean ow stress method, both of which predict signicantly higher ow stresses at higher compressive strains, and therefore increasingly deviate from the true ow stress of the material. The comparison of the ow stresses calculated by the three methods with the true ow stress is also made in Fig. 8 for the specimen with an aspect ratio of 1.0 and a friction coefcient of 0.5. The lower the aspect ratio, the less is the radius of the barrel curvature, implying that an innitely long cylinder would not barrel at all, as suggested by Johnson and Mellor 11. Hence, decreasing the aspect ratio from 1.5 to 1.0 signicantly decreases the
Transactions of the ASME

196 Vol. 122, APRIL 2000

Fig. 8 Comparison of ow stresses at various effective strains 2 H o d o 1.0; m 0.5

barrel radius when subjected to the same height reduction ratio, thereby signicantly increasing the inhomogeneity of deformation. It is seen in Fig. 8 that the proposed theoretical model still predicts the ow stress consistently in close agreement with the true ow stress, and is superior to those given by the other two theories. This conrms the validity of the proposed theoretical model for calculating the ow stress for a barreled specimen under axial compression over the whole range of reductions in height.

troducing various simplifying assumptions. The satisfactory estimate of the ow stress for the barreled specimen under inhomogeneous compression indicates that the assumptions made in the proposed theoretical model are sufciently realistic. The velocity eld adopted in the present study to calculate the mean effective strain for the barreled specimen also proves to be effective. The fact that the mean effective strain of the barreled specimen is approximately equal to the overall axial compressive strain to the rst order simplies the construction of the stressstrain relation in the compression test. Furthermore, although the analysis correctly predicts the main trend for the ow stresses in a barreled specimen during the compression test, the proposed theoretical model would tend to underestimate the ow stress of specimen at lower compressive strains. However, the difference between the calculated ow stresses and the true ow stresses is not signicant. It is suggested that the inclusion of a shape factor in Eq. 18, which can allow for the change in the barrel geometry during the compression, might improve the accuracy of the proposed theoretical model.

Acknowledgment
The authors wish to thank the National Science Council of the Republic of China for their grant under the project NSC85-2216E-002-027, which makes this research possible. They also would like to thank Professor J. Chakrabarty for his helpful discussions.

Summary and Concluding Remarks

The purpose of this paper is to calculate the ow stresses for a barreled specimen, resulting from the presence of frictional force at the die-specimen interface during the compression test. The nite element simulations, instead of experimental results, were employed to generate the compression loads and the geometric parameters of the barreled specimen to validate the theoretical model developed in the present study. Unlike most of the published methods, which studied the incipient barreling only, the proposed theoretical model takes the continued barreling of the deforming specimen into account. As the present study shows, the analysis which is based on Hills general method for metal forming problems has yielded better results, both in magnitude and in trend, for the prediction of ow stresses in the barreled specimen in a compression test. It is also noted from the comparison made between the three methods that the proposed theoretical model can be used more effectively to the cylinder compression test with signicant degrees of barreling. In spite of extensive derivations and complex integrations required in its application, Hills general method is certainly based on sound mathematical principles, and is exible enough for in-

References
1 Siebel, E., 1923, Stahl und Eisen, Duesseldorf, 43, p. 1295. 2 Avitzur, B., 1968, Metal Forming: Process and Analyses, McGraw-Hill, New York, pp. 102111. 3 Lee, C. H., and Altan, T., 1972, Inuence of Flow Stress and Friction Upon Metal Flow in Upset Forging of Rings and Cylinders, ASME J. Eng. Ind., 94, Aug., pp. 775782. 4 Ettouney, O., and Hardt, D. E., 1983, A Method for In-Process Failure Prediction in Cold Upset Forging, ASME J. Eng. Ind., 105, pp. 161167. 5 Bridgman, P. W., 1952, Studies in Large Plastic Flow and Fracture, McGrawHill, New York, pp. 986. 6 Hill, R., 1963, A General Method of Analysis for Metal-Working Processes, J. Mech. Phys. Solids, 11, pp. 305326. 7 Lahoti, G. D., and Kobayashi, S., 1974, On Hills General Method of Analysis for Metal-Working Processes, Int. J. Mech. Sci., 16, pp. 521540. 8 Kobayashi, S., Oh, S. I., and Altan, T., 1989, Metal Forming and the FiniteElement Method, Oxford University Press, Oxford, pp. 7883. 9 Chakrabarty, J., 1996, private communications. 10 Horton, H. L., Ryffel, H. H., and Schubert, P. B., 1959, Machinists Handbook, 16th Edition, The Industrial Press, New York, p. 152. 11 Johnson, W., and Mellor, P. B., 1975, Engineering Plasticity, Van Nostrand Reinhold, London, pp. 110114.

Journal of Engineering Materials and Technology

APRIL 2000, Vol. 122 197

M. A. Linaza J. M. Rodriguez-Ibabe
n, CEIT and ESII de San Sebastia Pde Manuel de Lardizabal, 15, 20018, n, Basque Country, Spain San Sebastia

Fracture Static Mechanisms on Fatigue Crack Propagation in Microalloyed Forging Steels

The inuence of static mechanisms on fatigue crack propagation in Ti and Ti-V microalloyed steels is considered. Small inclusions originate void nucleation. In contrast, TiN coarse particles contribute to the formation of bursts of cleavage in the fatigue zone. Taking into account the microstructural characteristics of the matrix that surrounds the particle, the microcrack can be conned within the particle or propagate along the matrix forming a cleavage burst. The inuence on macroscopic crack propagation of both types of static micromechanisms is considered. S0094-42890000902-6

Introduction

In the fatigue crack propagation corresponding to the Paris regime, together with the conventional fatigue mechanisms, other processes of static nature can take place in some materials. The presence of these static mechanisms can contribute in a relevant way to the crack behavior, giving place to higher macroscopic propagation rates. For example, it is well known that in low toughness materials the local cleavage mechanisms contribute in increasing the m exponent value Ritchie and Knott 1, Beevers et al. 2. On the other hand, ductile voids are related to nonmetallic particle decohesion and their inuence on fatigue is not so well dened as in the case of brittle fracture mechanisms. This can be as a consequence of the different behavior of the nonmetallic inclusions depending on their nature Nicholson and Gladman 3 and on testing orientation Wilson 4. In the case of forging steels, microalloying addition has become one of the main procedures in achieving ferrite-pearlite microstructures, with strengths similar to the more expensive quenched and tempered steels. Among these types of steel, Ti, V, and Ti-V microalloyed steels must be considered. Although equivalent mechanical strength has been achieved with these steels, usually fracture toughness behavior remains lower Naylor 5,6. In these types of steels, the cleavage process has been attributed to the nucleation of microcracks in some nonmetallic particles and their dynamic brittle propagation across a coarse ferrite-pearlite microstructure Linaza et al. 7. This type of brittle mechanism, acting under static conditions, can play a relevant role during fatigue crack propagation in the as-hot worked microstructures. On the other hand, in order to improve machinability or toughness properties by promoting intragranular ferrite nucleation LaGreca et al. 8, a signicant volume fraction of nonmetallic inclusions mainly MnS inclusions can be present in forging steels. These inclusions can give rise to the nucleation of ductile voids, which can increase crack growth rates. In this work, the inuence of both ductile and brittle static mechanisms, on the fatigue crack propagation, is considered for the case of two microalloyed medium carbon steels Ti and Ti-V with different microstructures.

strength. In this latter case, a higher sulfur content improves the machinability of the steel. The as-rolled ferrite-pearlite condition was analyzed in both steels and, for the Ti-V microalloyed steel, three additional microstructures were considered. These microstructures were obtained by laboratory thermomechanical simulations performing plane strain compression tests (50 50 10 mm3) at 850, 950, and 1000C, developing different ne ferrite-pearlite and acicular ferrite microstructures. The main microstructural parameters d ferrite grain size and f ferrite volume fraction and mechanical properties y yield stress, UTS ultimate tensile strength and J integral fracture toughness parameter ASTM E813 standard, three point bending specimen with B6 mm are listed in Table 2. A more detailed description of the microstructures and their corresponding toughness behavior have been reported elsewhere Linaza et al. 9,10,7 Specimens for fatigue crack growth analysis were machined into CT geometries B6 mm and W36 mm from the as-rolled bars and plane strain compression samples. With this geometry, both the threshold and the Paris equation were determined at room temperature, in a 20 kN resonant testing machine at a load ratio of R0.03. In the case of as-rolled microstructures, in order to minimize the crack closure effect on crack propagation, some tests were performed at R0.5. Tests for the threshold measurement were carried out applying a manual load shedding technique. For the determination of the Paris equation, the tests were carried out under load control mode, using a sinusoidal load signal, in agreement with the ASTM E647 standard. During the test, the fatigue crack length was measured periodically and the corresponding number of cycles recorded. The fracture surface of the tested specimens was extensively studied by scanning electron microscopy.

Results

Material and Experimental Procedure

Figures 1 and 2 show the K da / dN curves corresponding to the Ti steel and the Ti-V steel, respectively. In Fig. 1, the results have been obtained with the same ferrite-pearlite microstructure, considering the inuence of the load ratio R. As can be observed, in the analyzed K range, the crack propagation rate is smaller for R0.03 than for R0.5. In Fig. 2 the inuence of four different microstructures is considered. It is worth emphasizing that
Table 1 Chemical compositions of the studied microalloyed medium C steels wt. percent Steel C 0.35 0.37 Mn 1.56 1.45 Si 0.33 0.56 P 0.004 0.010 S 0.007 0.043 V 0.11 Al 0.027 0.024 Ti 0.028 0.015 N ppm 89 162

Two hot-rolled bar steels 50 mm square with compositions listed in Table 1 were used in this investigation. Both materials are medium carbon steels with Ti additions and, in the case of the Ti-V steel, an addition of V is done in order to obtain a higher
Contributed by the Materials Division for publication in the JOURNAL OF ENGINEERING MATERIALS AND TECHNOLOGY. Manuscript received by the Materials Division April 13, 1999; revised manuscript received October 14, 1999. Associate Technical Editor: S. Mall.

Ti Ti-V

198 Vol. 122, APRIL 2000

Copyright 2000 by ASME

Transactions of the ASME

Table 2 Microstructural parameters and mechanical behavior of analyzed microstructures Steel Ti Ti-V Process As-rolled As-rolled HW 950CSC HW 850CSC HW 1000CFC d m 5.3 5.5 5.1 5.6 f % 26.1 38.0 26.0 39.0 -

y MPa
440 590 655 607 560

UTS MPa 740 875 927 907 920

J kJ/m2 58.1* 23.7* 78.5 103

KJ MPam 116.3 74.3 135.2 154.9

HW: hot worked; SC: slow cooled; FC: fast cooled * J J c J J Ic

the ne ferrite-pearlite microstructure obtained by hot working at 850C shows a lower crack growth rate for a given cyclic stress intensity value. In a similar way, the higher crack growth rate corresponds to the coarse as-rolled ferrite-pearlite microstructure. The acicular ferrite and the ferrite-pearlite microstructures obtained by hot working at 950C lie between these two extremes. A summary of the results C and m Paris coefcient values and K th threshold data is shown in Table 3. As can be seen in the table, m exponent values are between 3.4 and 4.7 for R0.03; in contrast, when R0.5, the m exponent value ranges between 2.7 and 3. In relation to the fatigue threshold, the inuence of load ratio is more relevant than the microstructure effect. The fractographic analysis shows the classical at and quasiamorphous aspect of the fatigue fracture surfaces. Apart from the presence of striations, more visible for high K values, ductile voids nucleated at inclusions and brittle islands, formed by several

cleavage facets, have been identied. In relation to the ductile voids, they are more frequently observed in the Ti-V steel, which exhibits a higher density of inclusions due to the higher sulfur concentration see Fig. 3 corresponding to voids nucleated at MnS inclusions. On the other hand, the void volume fraction remains constant during the entire K range corresponding to the Paris regime, although there are some local heterogeneities. In the case of the Ti-V microalloyed steel, an area fraction of around 1 percent of cavities in the fracture surfaces has been estimated by point counting technique in the K max range from 20 to 50 MPa m . Brittle cleavage islands can be observed in both steels with ferrite-pearlite microstructures they have not been identied in the case of an acicular ferrite microstructure, although more in the Ti-V steel, and usually they appear for K max values higher than

Fig. 1 da dN K fatigue crack propagation curves of asrolled ferrite-pearlite Ti steel Table 3 Steel Ti-V Process As-rolled As-rolled HW 1000CFC HW 850SC HW 950SC As-rolled As-rolled As-rolled As-rolled As-rolled

Fig. 2 da dN K fatigue crack propagation curves of Ti-V microalloyed steel for different ferrite-pearlite and acicular ferrite microstructures developed by thermomechanical treatments

C and m coefcient values of Paris equation and fatigue threshold values

Microstruct. ferrite-perl. ferrite-perl. acicular ferrite ferrite-perl. ferrite-perl. ferrite-perl. ferrite-perl. ferrite-perl. ferrite-perl. ferrite-perl. R 0.03 0.03 0.03 0.03 0.03 0.5 0.5 0.03 0.5 0.5 C 4.74 10 13 1.17 10 12 5.15 10 13 1.02 10 14 8.05 10 14 1.31 10 11 8.26 10 12 4.14 10 14 6.84 10 12 1.59 10 11 m 3.7 3.4 3.6 4.7 4.2 2.8 3.0 4.3 3.0 2.7 K th MPam 8.8 8.5 11.6 4.7 10.5 5.6 -

Ti

Journal of Engineering Materials and Technology

APRIL 2000, Vol. 122 199

50 MPa m . In all the cases, they are originated by the rupture of TiN particles coarser than 3.5 m see Fig. 4 and consist of several facets. In order to analyze the characteristics of these brittle islands, some geometrical measurements have been performed for the case of the Ti-V steel with a ferrite-pearlite microstructure. The measured parameters, listed in Table 4, are the K max value at which the island has been formed, the dimensions of the TiN particle responsible for the cleavage initiation, the number of cleavage facets constituting the island and the maximum and minimum dimensions of the island. Another important feature,

common to all the identied islands, is the small misorientation between the facets and the macroscopic fatigue crack plane. Finally, some coarse TiN broken particles have also been observed in the fatigue fracture surface, which have not promoted the formation of cleavage facets. This behavior is more common in the case of acicular ferrite and ne ferrite-pearlite microstructures hot-worked samples at 850 and 950C.

Discussion

Referring to the results obtained in the Paris region, the most relevant parameter inuencing crack propagation is the load ratio, as can be seen in Figs. 1 and 2. This behavior has also been observed in other ferrite-pearlite steels and it can be correlated with the presence of a crack closure effect. Although crack closure measurements have not been carried out, taking into account the results obtained by Tanaka and Soya 11 with different V microalloyed structural steels, the slight differences observed in the Paris zone between the different microstructures can be considered through the crack closure. Similarly, Costa and Ferreira 12 have shown that in the case of thin specimens (B6 mm) there is a stronger inuence of the crack closure effect in the Paris regime. These authors have proposed the following empirical relationships to take into account the crack closure effect: U 0.433 0.716R 0.012 K 0.144 B/W U1 if 0.567 if 0.567 (1)

Fig. 3 Fracture surface in the Paris zone of Ti-V steel with as-rolled ferrite-pearlite microstructure showing ductile voids nucleated at inclusions R 0.5, da dN 1.2 10 7 mc

Fig. 4 Brittle island in the Paris region originated by the rupture of a TiN coarse particle Ti steel, R 0.03

where 0.716R 0.0121 K 0.144(B/W) and K effUK Taking into account these expressions, it can be concluded that for R0.5, in the Paris regime, there is not a closure effect. On the other hand, if K eff is considered instead of K , the data corresponding to R 0.03 overlap the data obtained for R0.5 for both steels, showing that the R effect in the Paris zone corresponds to the crack closure. All the data are within the scatter band of fatigue crack growth rates published in the bibliography corresponding to ferritepearlite microstructures obtained with different hypoeutectoid steels Rodriguez-Ibabe and Gil-Sevillano 13. Similarly, threshold data exhibit fair agreement with reported data for ferritepearlite steels Bulloch 14. Figure 5 shows the Paris equations corresponding to the mean values of the different tests of both steels. In the gure the results obtained with a ferrite-pearlite C-Mn steel without static mechanisms are also included for comparison Linaza et al. 15. To avoid the possible effect of crack closure, only the results measured for R0.5 have been considered. The Paris straight lines corresponding to the three materials are parallel and the propagation rate is fastest for the Ti-V steel, followed by the Ti steel and by the C-Mn steel, respectively. In consequence, while the m exponent is very similar in all cases, there are some differences in C coefcient values that can be correlated to the contribution of static mechanisms in crack propagation.

Table 4 Characteristics of the brittle cleavage islands on the fatigue surface of Ti-V steel with ferrite-pearlite microstructure TiN R 0.03 K max MPa m 58 60 80 86 53 59 60 51 53 a min m 6.1 4.4 5.4 8.0 3.3 4.2 2.2 2.2 1.6 size a max m 6.9 6.5 6.8 8.0 8.3 8.3 5.8 10.4 8.3 facet number 8 9 1 7 7 10 8 10 2 d min m 45 50 22 44 90 68 50 40 40 Island size d max m 89 90 30 90 142 97 80 128 52

0.5

200 Vol. 122, APRIL 2000

Transactions of the ASME

Fig. 6 Histograms of the minimum and maximum dimensions of coarse TiN particles in the Ti-V steel

Fig. 5 Paris equations of Ti-V and Ti steels for R 0.5. Results obtained with a ferrite-pearlite C-Mn steel are included Linaza et al. 15.

4.1 Brittle Static Mechanisms. In relation to the static fracture mechanisms, the number of brittle features identied in each specimen is very small and, as a result, they did not inuence the macroscopic crack propagation, as can be observed in the results of Figs. 1 and 2 and in the m exponent values of Table 3. Nevertheless, the characteristics of these islands can contribute to the study of the micromechanisms taking place in the static brittle fracture of microalloyed forging steels. The brittle islands are nucleated at coarse TiN particles and, in consequence, the presence of these cleavage facets, completely surrounded by fatigue mechanisms, must be related to the nature of these TiN particles. Coarse TiN particles are formed during the solidication process of the steel Herman et al. 16 and, due to their size, they are ineffective in controlling austenite grain coarsening during forging. On the other hand, as a consequence of the thermal expansion coefcient mismatch Brooksbank and Andrews 17, TiN particles have a very strong bonding with the steel matrix. Brittle fracture requires the nucleation of a microcrack in some microstructural feature, usually as a consequence of a plasticity induced process Bowen et al. 18, and its propagation across the matrix. In the case of TiN particles, once the particle is broken, its good inclusion/matrix bonding facilitates the brittle propagation of the microcrack across the interface. In a previous study, Linaza et al. 9,20 identied that the static brittle fracture toughness of both steels was related to the nucleation of microcracks at coarse TiN particles and their propagation across different microstructural barriers particle-matrix and high angle matrix-matrix barriers. Once one of these microcracks has traversed several grains sometimes one is enough, its length together with the local stress state is sufcient for catastrophic propagation to failure. This behavior takes place in the completely brittle regime and in the ductile-brittle transition, and has been identied as one of the main microstructural features promoting poor toughness values in the as-forged ferrite-pearlite microstructures Linaza et al. 20. Figure 6 shows the histograms of the measured minimum and maximum dimensions of coarse TiN particles present in the Ti-V steel. In the static fracture toughness tests performed with the same steel in a previous work Linaza et al. 10, the interval of TiN particles identied as responsible for cleavage nucleation ranged from 1 to 4 m. Comparing these previous results with the data of Table 4, it is worth emphasizing that the particles observed in the fatigue surfaces promoting brittle islands are coarser than those identied in the static tests, and that they belong to the Journal of Engineering Materials and Technology

group of largest particles present in the steel. In order to analyze this size difference of the TiN particles acting as brittle fracture nucleators, a probability effect must be considered originating as a consequence of testing samples of different volume in the fatigue and fracture toughness experiments. Taking into account the geometries tested, in the static fracture toughness tests three point bending specimen with B6 mm and W11 mm the material volume sampled by the peak stress is at least three times smaller than in the case of fatigue specimens considering the material volume corresponding to the propagation of the crack for K max 50 MPa m , minimum value observed for the appearance of brittle islands in the fatigue tests. In consequence, the probability of having coarse TiN particles able to nucleate brittle fracture bursts is higher in the fatigue samples, leading to the differences observed between the static fracture toughness results and the values of Table 4. Another relevant characteristic is that there are broken TiN particles which have not promoted the formation of cleavage facets. This means that the nucleation of a sharp microcrack by the rupture of a TiN particle is not a sufcient criterion to cause the formation of brittle facets. In order to explain this behavior, it must be considered that, as a consequence of a Weibull volume effect Knott and King 21, large TiN particles can break at lower stresses than small particles. If cracks are nucleated in large particles at stress values too low to be able to promote their propagation across particle-matrix interfaces, microcracks will blunt without forming cleavage facets. This behavior can explain why a minimum K max value is required for the appearance of brittle islands. In relation to the characteristics of the islands, independent of the number of facets, in all the cases they are well oriented in relation to the macroscopic plane. This means that the rst facet formation only takes place in those cases in which there is a good crystallographic orientation of the 100 planes surrounding the broken TiN particle. In consequence, in those cases without an appropriate crystallographic orientation for example in the case of an acicular ferrite microstructure, the local stress value acting normally to the 100 plane would be smaller, increasing the difculty for propagation of the microcrack from the broken TiN particle to the matrix and across the different microstructural barriers. This is the reason why after some brittle propagation, cracks are arrested at matrix-matrix boundaries and the cleavage process ahead of the fatigue crack is completely stopped. Similarly, brittle islands nucleated at TiN broken particles and arrested after crossing some matrix-matrix boundaries have been identied in static tests in the ductile-brittle regime San Martin and RodriguezIbabe 22. Finally, although these brittle fracture mechanisms have not modied the m slope value of the Paris curve, it is worth emphaAPRIL 2000, Vol. 122 201

sizing the deleterious effect that they can have in industrial applications, promoting premature brittle fracture of components. It is well documented that coarse TiN particles are deleterious from the point of view of fracture toughness when they are combined with coarse microstructures Linaza et al. 9,20. This refers to the situation corresponding to the lower part of the ductile-brittle regime of microstructures at room temperature. In this zone, there is a large dispersion of toughness values, dependent upon the probability of activating the brittle process after some ductile propagation Landes et al. 19. As a consequence of the fatigue crack propagation process, during the crack extension a higher material volume could be sampled by the peak stress than is the case for a stationary crack, thus increasing the probability of having adequate conditions to promote cleavage fracture a combination of the presence of a TiN particle of adequate size surrounded by a coarse ferrite-pearlite matrix. 4.2 Ductile Static Mechanisms. Ductile voids nucleated at inclusions appear in the entire range of measured K values. The order of Paris straight lines of Fig. 5 can be correlated to the cleanness of the steels: while the lower growth rate for a given K value corresponds to C-Mn steel, the higher rate propagation is achieved with the Ti-V steel. The Ti-V steel has a higher volume fraction of MnS and oxides, while the C-Mn steel exhibits a very low volume fraction of inclusions 0.020 percent. The different volume fraction of inclusions results in the appearance of different percentages of ductile cavities on the fatigue surface. These cavities have been nucleated ahead of the crack tip in the plastic zone. In consequence, fatigue crack growth will be easier, due to the fact that the crack must propagate across a lower effective surface. If the effect produced by a cavity nucleated from an inclusion is assimilated to a pore in a sintered material, the propagation rate can be obtained by the equation proposed by Bompard and Franc ois 23 for those types of materials: da C K m dN D m1 (2)

probability effect, the toughness of these steels, in the ductilebrittle regime, can correspond to the lower bound of the values measured by static fracture toughness tests. 3 The formation of ductile cavities at non-metallic inclusions increases the crack propagation rates slightly by increasing the C coefcient of the Paris equation. In the Paris regime, the formation of the ductile cavities is independent of the applied K value.

Acknowledgments
Part of this work has been carried out under an ECSC Research Project. Financial support by CICYT is also acknowledged.

References
1 Ritchie, R. O., and Knott, J. F., 1973, Mechanisms of Fatigue Crack Growth in Low Alloy Steels, Acta Met., 21, p. 639. 2 Beevers, C. J., Cooke, R. J., Knott, J. F., and Ritchie, R. O., 1975, Some Considerations of the Inuence of Subcritical Cleavage Growth During Fatigue-Crack Propagation in Steels, Met. Sci., 9, p. 119. 3 Nicholson, A., and Gladman, T., 1986, Non-Metallic Inclusions and Development in Secondary Steelmaking Ironmaking and Steelmaking, 13, p. 53. 4 Wilson, A. D., 1984, Fatigue Crack Propagation in Steels: The Role of Inclusions, Fracture: Interactions of Microstructure, Mechanisms and Mechanics, J. M. Wells and J. D. Landes, eds., AIME, p. 235. 5 Naylor, D. J., 1989, Review of International Activity on Microalloyed Engineering Steels Ironmaking and Steelmaking, 16, p. 246. 6 Naylor, D. J., 1998, Microalloyed Forging Steels Mater. Sci. Forum, 284 286, p. 83. 7 Linaza, M. A., Rodriguez-Ibabe, J. M., and Urcola, J. J., 1997, Determination of the Energetic Parameters Controlling Cleavage Fracture Initiation in Steels, Fatigue Fract. Eng. Mater. Struct., 20, p. 619. 8 LaGreca, P. D., Matlock, D. K., and Krauss, G., 1996, Short-rod Fracture Toughness Testing of Microalloyed Steels as a Function of Sulfur and Intragranular Ferrite Content, Fundamentals and Applications of Microalloyed Forging Steels, C. J. Van Tyne et al., eds., TMS, Warrendale, p. 357. 9 Linaza, M. A., Romero, J. L., Rodriguez-Ibabe, J. M., and Urcola, J. J., 1993, Inuence of the Microstructure on the Fracture Toughness and Fracture Mechanisms of Forging Steels Microalloyed with Ti with Ferrite-Pearlite Structures, Scr. Metall. Mater., 29, p. 451. 10 Linaza, M. A., Romero, J. L., Rodriguez-Ibabe, J. M., and Urcola, J. J., 1995, Cleavage Fracture of Microalloyed Forging Steels, Scr. Metall. Mater., 32, p. 395. 11 Tanaka, Y., and Soya, Y., 1990, Metallurgical and Mechanical Factors Affecting Fatigue Crack Propagation and Crack Closure in Various Structural Steels, Fatigue 90, H. Kitagawa and T. Tanaka, eds., MCE Publications, Vol. 2, p. 1143. 12 Costa, J. D. M., and Ferreira, J. A. M., 1998, Effect of Stress Ratio and Specimen Thickness on Fatigue Crack Growth of CK45 Steel, Theor. Appl. Fract. Mech., 30, p. 65. 13 Rodriguez-Ibabe, J. M., and Gil-Sevillano, J., 1984, Fatigue Crack Path in Medium-high Carbon Ferrite-Pearlite Structures, Advances in Fracture Research, S. R. Valluri et al., eds., Pergamon Press, 3, p. 2073. 14 Bulloch, J. H., 1992, Effects of Mean Stress on the Threshold Fatigue Crack Extension Rates of Two Spherical Graphite Cast Irons, Theor. Appl. Fract. Mech., 18, p. 15. 15 Linaza, M. A., Rodriguez-Ibabe, J. M., and Fuentes, M., 1992, Fatigue Crack Growth and Closure Behavior of Pressure Vessel C-Mn Welded Steels, Reliability and Structural Integrity of Advanced Materials, S. Sedmak et al., eds., EMAS, Vol. 1, p. 397. 16 Herman, J. C., Messien, P., and Greday, T., 1982, HSLA Ti Containing Steels, Thermomechanical Processing of Microalloyed Austenite, A. J. DeArdo et al., eds., AIME, Warrendale, p. 655. 17 Brooksbank, D., and Andrews, K. W., 1968, Thermal Expansion of Some Inclusions Found in Steels and Relation to Tessellated Stresses, JISI, 206, p. 595. 18 Bowen, P., Druce, S. G., and Knott, J. F., 1987, Micromechanical Modelling of Fracture Toughness, Acta Metall., 35, p. 1735. 19 Landes, J. D., Heerens, J., Schwalbe, K. H., and Petrovski, B., 1993, Size, Thickness and Geometry Effects on Transition Fracture, Fatigue Fract. Eng. Mater. Struct., 16, p. 1135. 20 Linaza, M. A., Romero, J. L., Rodriguez-Ibabe, J. M., and Urcola, J. J., 1997, Inuence of Thermomechanical Treatments on the Microstructure and Toughness of Microalloyed Engineering Steels, Thermomechanical Processing in Theory, Modelling and Practice, B. Hutchinson et al., eds., SFMC, p. 351. 21 Knott, J. F., and King, J. E., 1990, Fatigue in Metallic Alloys Containing Non-Metallic Particles, Fatigue 90, H. Kitagawa and T. Tanaka, eds., MCE Publications, Vol. 4, p. 2557. 22 San Martin, I., and Rodriguez-Ibabe, J. M., 1999, Determination of the Energetic Parameters Controlling Cleavage Fracture in a Ti-V Microalloyed Ferrite-Pearlite Steel, Script. Mat., 40, p. 459. 23 Bompard, P. H., and Franc ois, D., 1984, Effect of Porosity on Fatigue Crack Propagation in Sintered Nickel, Advances in Fracture Research, S. R. Valluri et al., eds., Vol. 3, p. 2049.

where D is the ratio between the effective area and the total area. In this equation, it is supposed that the ductile static fracture events are independent of stress intensity and, as a result, they contribute only to increasing the coefcient of the power-law, but not its exponent. If the area fraction of cavities of around 1 percent, estimated for Ti-V steel, is substituted in Eq. 2 the differences between Ti-V steel and C-Mn can be explained the C coefcient value for the Ti-V steel changes from 1.0 10 11 mean value of tested specimens to 5.9 10 12, very close to the 7.5 10 12 value obtained for the C-Mn steel. A similar behavior is observed for the Ti steel. Summarizing, the existence of ductile mechanisms in the fatigue crack propagation in the microalloyed forging steels considered in this study has a very limited effect, introducing only a slight increase in the C coefcient value of Paris equation.

Conclusions

The inuence of static mechanisms on fatigue crack propagation in the considered microalloyed steels leads to the following conclusions: 1 These steels exhibit fatigue threshold values and Paris equations close to those found in plain C-Mn steels with ferrite-pearlite microstructures. The brittle islands nucleated at TiN coarse particles which appear in both microalloyed steels have no inuence on the fatigue crack propagation rate. 2 The formation of these brittle islands agrees with the brittle fracture process, which takes place in Ti microalloyed steels with coarse TiN particles. During fatigue crack propagation, due to a 202 Vol. 122, APRIL 2000

Transactions of the ASME

Chuwei Zhou Wei Yang Daining Fang

FML, Department of Engineering Mechanics, Tsinghua University, Beijing 100084, China

Damage of Short-FiberReinforced Metal Matrix Composites Considering Cooling and Thermal Cycling
Mechanical properties and damage evolution of short-ber-reinforced metal matrix composites (MMC) are studied under a micromechanics model accounting for the history of cooling and thermal cycling. A cohesive interface is formulated in conjunction with the Gurson-Tvergaard matrix damage model. Attention is focused on the residual stresses and damages by the thermal mismatch. Substantial stress drop in the uniaxial tensile response is found for a computational cell that experienced a cooling process. The stress drop is caused by debonding along the ber ends. Subsequent thermal cycling lowers the debonding stress and the debonding strain. Micromechanics analysis reveals three failure modes. When the thermal histories are ignored, the cell fails by matrix damage outside the ber ends. With the incorporation of cooling, the cell fails by ber end debonding and the subsequent transverse matrix damage. When thermal cycling is also included, the cell fails by jagged debonding around the ber tops followed by necking instability of matrix ligaments. S0094-42890001202-0

Introduction

Formulation on Interface and Matrix Damage

Short-ber-reinforced Metal-Matrix-Composite MMC has advantages in both formability and enhanced mechanical properties, such as high specic stiffness and high specic strength. One critical requirement for short-ber-reinforced MMC is the survivability under a severe history of thermal cycling. For example, a space vehicle may encounter thermal cycling ranging from an elevated temperature to a cryogenic temperature. The mismatch in thermal expansions between the ber and the matrix produces residual stresses in MMC. Damage accumulation during a severe thermal cycling history may degrade the mechanical properties, such as the strength and the toughness, of MMC. Multiple damage sources exist in the MMC: the interfaces may debond, voids may occur in the metal matrix, and bers may break under unbearable tensile stress. The present work is focused on the interface and matrix damages, since ber breakage does not dominate the damage pattern for the MMC reinforced by short-bers. The evolution of damage is dictated by the local stress and strain elds in MMC, thus raising the issue of sensitivity to the FEM scheme in the simulation. A viable FEM scheme should have the resolution on stress, strain, and damage elds ner than the diameter of a ber, and have the ability to account for the history of thermal-mechanical processing. FEM studies of MMC were initiated by Christman et al. 1, by Levy and Papazian 2, and by Tvergaard 3. These pioneer works did not discuss the issue of residual stresses and strains in MMC when cooled from the solution-treatment temperature to the room temperature. The importance of the residual stresses and the thermal cycling was soon recognized. Levy and Papazian 4, and Davis and Allison 5 discussed the residual stresses in MMC after cooling, and Levy and Papazian 6 considered the effect of thermal cycling. Unfortunately, those works neither referred to the damage evolution within the material, nor accounted explicitly the damages in interface and matrix. That provides a strong impetus for the present work to investigate the effects of cooling and thermal cycling on the strength and damage evolution of MMC.
Contributed by the Materials Division for publication in the JOURNAL OF ENGINEERING MATERIALS AND TECHNOLOGY. Manuscript received by the Materials Division March 24, 1999; revised manuscript received November 4, 1999. Associate Technical Editor: E. Busso.

The ber/matrix interface is usually a thin layer of different material properties, or a bimaterial interlayer Yang and Shih 7. The present work adopts elements based on the cohesive interface model Needleman 8, Tvergaard 9, Chaboche et al. 10 with minor modications, that enables numerical simulation of interface damage and debonding. The cohesive interface is formulated under a phenomenological model. It relates to the interface properties and the damage of MMC by choosing the interface parameters properly. Damages in the normal and the tangential directions are coupled. The interface traction components, T n and T t , are correlated to the displacement jumps across the interface, n and t , by T n 1 max 2 E n n H n K n n H n T t 1 max 2 E t t where max (1) (2)

max n

H max n

tmax

(3)

and tmax denote the maximum displacement The quantities max n jumps of n and t in the history. In Eqs. 1 and 2, subscripts n and t label the normal and the tangential components. Symbol H denotes the Heaviside function that is utilized to distinguish interface responses under tension and compression. Symbols E n and E t , n and t , are two sets of interface parameters that represent the interface moduli and the critical interface separations, respectively. The quantity K n denotes the normal compressive stiffness and is assigned a large value to minimize the ber/matrix penetration. The scalar max is a monotonically increasing parameter indicative to the interface damage: max0 refers to a perfect interface and max1 a complete separation. Figure 1 depicts the traction-displacement jump curve for the case of normal separation. As the interface displacement jump increases from zero, the interface traction rises at rst, then declines after the traction peak, and eventually diminishes when n n . The descending regime in the cohesive curve describes the interface weakening. The situation that max ceases to evolve refers to a pseudo-elastic state, where the interface traction and interface displacement jump are APRIL 2000, Vol. 122 203

Journal of Engineering Materials and Technology Copyright 2000 by ASME

ated void growth, and f F the void volume fraction to lose the stress carrying capacity. Numbers q 1 , q 2 , and q 3 are the adjustable parameters introduced by Tvergaard 12. The rate f consists of the contributions from the growth of the existing voids and the nucleation of the new ones. That is
p p f 1 f kk A M p kk

(9)

where is the plastic part of the macroscopic volume deformap tion rate and M the equivalent plastic strain rate of the matrix material. The latter is obtained by equating the macroscopic and the microscopic plastic powers:
p M

i j ipj 1 f M

(10)

Fig. 1 Cohesive force curve during normal separation

The rst term in Eq. 9 is derived from the condition of plastic incompressibility of the matrix material. For the strain-controlled nucleation, the second term in Eq. 9 takes the Gaussian distribution form proposed by Chu and Needleman 15 A fN S N h 2 exp

linked by linear unloading or linear reloading, as marked in Fig. 1. The shaded area represents the energy dissipated during normal interface separation. Once max reaches unity, the interface is separated freely except blocked by contact, then Eqs. 1 and 2 are replaced by: T n K n nH n tH n T t K n n sgn (4) (5)

p N 1 M 2 SN

(11)

where f N is the volume fraction of void nucleating particles, N the profusion strain for nucleation, S N the standard deviation of the distribution, and h the matrix hardening modulus. The plastic ow within the matrix is modeled by an isotropic hardening law:
p Mh M 0 1 p EMM

(12)

The scalar denotes the friction coefcient. If a fully detached interface is recompressed, the interface will cease to slide when T t T n . The maximum strengths that a cohesive interface can sustain in pure normal and tangent directions can be derived as 4 n E n n , 27 4 t E t t 27 (6)

with the strain-hardening exponent N, the Youngs modulus E M and the initial yield stress 0 . The ductile matrix may withstand large plastic ow before failure, so a nite deformation framework is adopted Simo and Ortiz 16, Moran et al. 17.

Computational Cell

A periodic distribution of bers is assumed for simplicity. A longitudinal section is shown in Fig. 2a, where short bers of

One can use the interface strengths n and t , instead of the interface moduli E n and E t , as the independent interface parameters. Nucleation, growth and coalescence of voids characterize the matrix damage. The matrix damage is treated in a distributed manner in the present work. Within metallic matrix, the void evolution is described by Gurson-Tvergaard model Gurson 11, Tvergaard 12,13, Needleman and Tvergaard 14. The only damage indicator is the volume fraction of voids. This averaging treatment of damage makes sense if the initial void size is much smaller than the ber diameter in the order of one micron. One should bear in mind that when coupling with the nite element method, the Gurson-Tvergaard model can also handle large, discrete cavities formed due to void coalescence. Void nucleation and growth in the matrix of MMC result in macroscopic dilatancy, which should be incorporated into the constitutive relation of matrix materials. The yield condition of the Gurson-Tvergaard model is expressed as

2 eq 2 M
f

f *

2 f * q 1 cosh ffC

q 2 kk 1 q3 f *20 2M

(7)

if

f C

2 2 q 1 q 2 1 q 3 / q 3 f C

f F f C

f f C

if

ffC

(8)

where i j is the macroscopic stress tensor, eq the macroscopic equivalent stress, and M the actual yield stress of the matrix material. Symbols f and f * represent the actual and the effective void volume fractions, f C the critical volume fraction for acceler204 Vol. 122, APRIL 2000

Fig. 2 Aligned short bers arranged in a transverse hexagon array. a A longitudinal section shows the ber alignments; b a transverse section shows the hexagon array; c nite element mesh with indication of boundary conditions.

Transactions of the ASME

radius r f and half-length l f are aligned in a rectangular array of a horizontal spacing 2 l c . In the transverse section, the ber centers are arranged in a hexagonal array as shown in Fig. 2b, with a center-to-center spacing a s . This conguration is further simplied by the axisymmetric cells, namely, to replace the hexagons by circles of the same area as shaded in Fig. 2. The radius of the circle in nondeformed state is r c (3 1/4/ 2 ) a s . The volume fraction of bers is
2 V fr2 f l f /rclc

Results and Discussion

(13)

The aspect ratios of the bers and the initial cells are

fl f /r f ,

c l c / r c

(14)

Subjected to macroscopically uniform deformation, the periodicity dictates that each cell remains cylindrical though its aspect ratio varies with the deformation. Consequently, the local deformation and the damage are also periodic. In a nite deformation state, the present cell model cannot be used to the place where the macroscopic strain localizes.

Material Characterization

We assume the MMC is stress free at 525C, the annealing temperature of the matrix material. It rst undergoes a cooling phase, from 525C to the room temperature 25C. Then it is subjected to a specied thermal cycling phase that terminates at the room temperature. After cooling and thermal cycling, the mechanical behavior of the MMC is evaluated by a tensile test. The simulation is conducted for SiC whisker-2124 aluminum alloy, whose material constants at the room temperature are summarized by Tvergaard 18. The SiC whisker has a Poissons ratio f of 0.2 and a coefcient of thermal expansion CTE f of 4.9 10 6 /C. The whisker has a diameter of 0.5 m and an average length of about 2.5 m, resulted from the breakage in the extruding process. Accordingly, the ber has an aspect ratio of f 5 Tvergaard 18. The matrix of 2124 aluminum alloy has a E m / 0 Youngs modulus to the initial yield stress ratio of 200, a Poissons ratio m of 0.30, a work-hardening exponent N of 0.13 and a CTE m of 23.2 10 6 /C. The Youngs modulus of the ber is 5.7 times of that of the matrix. A value of 20 percent is selected for the volume fraction of bers. A cell aspect ratio of c 4.3 is chosen to give uniform longitudinal and transverse spacings among bers. The nite element mesh and the boundary conditions are depicted in Fig. 2c. The periodicity and the predominately tensile loading leave the geometry of the computation cell undistorted. The horizontal cell boundary remains horizontal and the vertical boundary remains vertical during the deformation. The elements within the ber are highlighted by gray color. Between the ber and the matrix, a layer of interface elements with zero initial thickness is inserted. Determination of the parameters in Eqs. 7, 8, and 11 requires detailed experiments on the matrix material. Several measurable material parameters such as the yield strength, the ultimate strength, the ductility and the void volume fraction at failure of the Chinese equivalent of 2124 aluminum alloy was summarized by Du et al. 19. Specically, the f F value is reported to be 0.20. From those experimental data, and the numerical simulation for the uniaxial tensile response of a bar made of 2124 aluminum alloy, Zhou et al. 20 correlated the following matrix damage parameters: the Tvergaards parameters for modied Gurson yield surface are found to be q 1 q 3 1.25 and q 2 1.0; the void nucleation parameters are found to be f N 0.04, S N 0.1 and N 0.3, identical to those used by Biner 21 for MMC; and the critical void volume fraction for accelerated void growth is found to be f C 0.15. Following Tvergaard 3,18, the interface parameters are chosen as n t 0.02r f , n t 2.5 0 and 0.3. Journal of Engineering Materials and Technology

We now present the simulation results on the computational cell made of SiC whisker-2124 aluminum alloy. When cooled from the annealing temperature 525C to the room temperature 25C, residual stresses develop by the CTE mismatch between the ber and the matrix. The residual shear stress concentrates near the corner of the ber end and induces the initial damage along the interface and within the matrix. The ber/matrix interface is compressed at this stage. We next consider a thermal cycling about the room temperature. The MMC is rst cooled from the room temperature to a cryogenic temperature. The residual stress and the damage grow as the temperature declines. Then the procedure is reversed, the MMC is heated from the cryogenic temperature, through the room temperature, to an elevated temperature. Tensile stress develops along the ber/matrix interface. The residual stresses in matrix relaxes rst and then changes to opposite signs. Figure 3 shows the most severe thermal stress distributions during a thermal cycling of 200C. Graphs a, b, and c plot the contours of radial, longitudinal, and shear stress components. The results indicate a considerable axial tensile stress ahead of the ber end, and a large shear stress around the corner of the ber. The effective stress near the ber corner surpasses the yield stress, and promotes void nucleation and growth in the matrix surrounding the ber corner. Graph d delineates the distribution of max along the interface. The suppression of matrix damage would raise the interface damage slightly, as shown in Graph e. Graphs f and g plot the contours of the matrix damage with and without the consideration of the interface damage. The occurrence of interface damage slightly reduces the level of matrix damage. Both the interface and the matrix damages concentrate near the ber end, especially near the ber corner. The highest value of interface damage is 0.178 against the value of max1 for the interface debonding; while the highest value of matrix damage is 0.00105 against the value of f F 0.20 for the matrix failure. Accordingly, the matrix damage is small compared with the interface damage and consequently secondary for the subsequent failure of MMC. x versus Figure 4 plots the curves of the average tensile stress the average logarithmic tensile strain x under different histories of thermal cycling. Curve A refers to the case without cooling and thermal cycling, and consequently free of the initial damage. The absence of initial damage suppresses the interface damage, and debonding never occurs when the tensile loading is applied to the computation cell. The stress strain curve attains the highest ultimate strength and the gradual declination of tensile stress is caused by the matrix damage. The MMC exhibits good strength and good toughness measured by the energy dissipation in the failure process in this case. Curves B, C, and D in Fig. 4 are all preceded by a cooling process from 525C to 25C. Interface damage takes place during cooling, and leads to debonding in the subsequent tensile loading. As the result, the average tensile curve will suffer a sudden drop in the average tensile stress. Since the rst stress drop is triggered by debonding along the ber end, we term the average tensile stress and strain corresponding to this event the debonding stress d and the debonding strain d . After the rst drop and a plateau of sustained strength, the second drop in the stress strain curve appears by damage coalescence in the matrix, which destroys the loading capacity of the composite. The effect of thermal cycling further reduces the strength and the toughness of MMC. Curve B refers to the case without thermal cycling, while curves C and D correspond to thermal cyclings with ranges of 100C and 200C, respectively. Figure 4 indicates that thermal cycling reduces the ultimate strength of MMC, as well as the average tensile strain at the rst stress drop. The wider the thermal cycling range, the lower the ultimate strength and the debonding strain. Figure 5 plots the x versus x curves after zero, one, two and twenty thermal cycles of 200C. The simulations indicate that the effect of thermal cycling stabilizes after two complete temperature cycles. Fatigue effects are not APRIL 2000, Vol. 122 205

Fig. 3 Stress and damage elds during two thermal cycles of 200C. a Maximum radial stress distribution in matrix; b maximum longitudinal stress distribution; c maximum shear stress distribution; d distribution of the interface damage max with matrix damage; e distribution of max without matrix damage; f distribution of the matrix damage f with interface damage; g distribution of f without interface damage.

included in this research since neither the cohesive interface model nor the Gurson-Tvergaard matrix damage model accommodates fatigue damage. In the present formulation, phenomena such as interface separation, matrix plastic deformation, residual stress and residual strain are dominated by the thermal cycling range rather than the number of thermal cycles. Under a given thermal cycling range, interface separation and matrix plastic deformation shake down after about two thermal cycles. The qualitative changes in the uniaxial tension curve of MMC by cooling and thermal cycling become transparent when their respective failure modes are examined. Figure 6 delineates the matrix damage distribution at the failure stage when the computation cell undergoes different thermal processes. If the effects of cooling and thermal cycling were ignored, the failure mode would 206 Vol. 122, APRIL 2000

assemble the damage pattern in Fig. 6a. It is featured by matrix damage that circumvents a cap ahead of the ber end, without any trace of interface debonding. The incorporation of a cooling phase alters the failure mode, as shown in Fig. 6b. The ber end debonds completely, and the matrix damage progresses across the ligament to form a at failure surface. Thermal cycling of a sufcient temperature range further twists the failure mode, as shown in Fig. 6c. Debonding occurs not only along the ber end, but also along the side surface of the ber. The cell eventually fails by necking instability of the matrix ligaments, leading to a jagged fractography. The effect of thermal cycling can be measured by a dimensionless ratio / L f ( m f ) T , where the numerator represents the largest separation the interface can sustain without debonding, Transactions of the ASME

Fig. 4 Average tensile stress x versus average logarithmic strain x curves f f 20 percent, n t 2.5 0 , f 5 and R 4.3 after various thermal histories. A, without cooling and thermal cycling; B, with cooling from 525C to 25C but not thermal cycling; C, cooling followed by thermal cycling of 100C; D, cooling followed by thermal cycling of 200C.

and the denominator measures the mismatch at the ber/matrix interface due to thermal cycling of a temperature range T . The characteristic length L f should be assigned as l f for debonding along the ber end, and r f for debonding along the ber side. When is large, the effect of thermal cycling is insignicant; when is small, the effect of thermal cycling may change the failure mode. In the present simulations, / r f 0.02, / l f 0.004 and m f 18.3 10 6 /C. A thermal cycling of 100C can affect the debonding at the ber end ( 1.09), but is insufcient to alter the debonding status along the ber side ( 5.46). A thermal cycling of 200C, on the other hand, affects not only the debonding status at the ber end ( 0.55), but also the debonding status along the ber side ( 2.73). The dimensionless group depends critically on the relative separation / r f that governs the debonding along the side surface of bers. For a given MMC after a prescribed manufacturing process namely, a xed history of cooling and thermal cycling, the ratio / r f becomes the critical parameter. Since the value of can be altered by interface coating, the ratio / r f is an adjustable parameter. The inuences of / r f on d and d are plotted in Figs. 7a and 7b, respectively. From Fig. 7a, one observes d declines as decreases. The declination becomes faster that

Fig. 6 The interface and matrix damage distributions at the failure stage after a thermal history A, b thermal history B , c thermal history D , preceding the uniaxial tension at the room temperature.

Fig. 5 Average tensile stress x versus average logarithmic strain x curves after zero, one, two, and twenty thermal cycles

Fig. 7 Average debonding tensile stress and strain versus the relative interface thickness r f curves under thermal histories B , C , and D . a d 0 versus r f curves; b d versus r f curves.

Journal of Engineering Materials and Technology

APRIL 2000, Vol. 122 207

under larger thermal cycling, as predicted by the form of the dimensionless ratio . The same trend is observed in Fig. 7b for the debonding strain. Under a large thermal cycling range, Figs. 7a and 7b indicate that the ratio / r f has large inuence on the debonding stress and strain.

References
1 Christman, T., Needleman, A., and Suresh, S., 1989, An Experimental and Numerical Study of Deformation in Metal-Ceramic Composites, Acta Metall., 37, pp. 30293050. 2 Levy, A., and Papazian, J. M., 1990, Tensile Properties of Short FiberReinforced SiC/Al Composites: Finite Element Analysis, Metall. Trans. A, 2, pp. 411420. 3 Tvergaard, V., 1990, Effect of Fiber Debonding in a Whisker-Reinforced Metal, Mater. Sci. Eng., A, 125, pp. 202213. 4 Levy, A., and Papazian, J. M., 1991, Elastoplastic Finite Element Analysis of Short-Fiber-Reinforced SiC/Al Composites: Effect of Thermal Treatment, Acta Metall. Mater., 39, pp. 22552266. 5 Davis, L. C., and Allison, J. E., 1993, Residual Stresses and Their Effects on Deformation in Particle-Reinforced Metal-Matrix Composites, Metall. Trans. A, 24, pp. 24872496. 6 Levy, A., and Papazian, J. M., 1993, Finite Element Analysis of WhiskerReinforced SiC/Al Composites Subjected to Cryogenic Temperature Thermal Cycling, ASME J. Eng. Mater. Technol., 115, pp. 129133. 7 Yang, W., and Shih, C. F., 1994, Fracture along an Interlayer, Int. J. Solids Struct., 31, pp. 9851002. 8 Needleman, A., 1987, A Continuum Model for Void Nucleation by Inclusion Debonding, ASME J. Appl. Mech., 54, pp. 525531. 9 Tvergaard, V., 1990, Analysis of Tensile Properties for a WhiskerReinforced Metal Matrix Composite, Acta Metall. Mater., 38, pp. 185194. 10 Chaboche, J., Girard, R., and Levasseur, P., 1997, On the Interface Debonding Models, Int. J. Damage Mech., 6, pp. 220257. 11 Gurson, A., 1977, Continuum Theory of Ductile Rupture by Void Nucleation and Growth: Part I-Yield Criteria and Flow Rules for Porous Ductile Media, ASME J. Eng. Mater. Technol., 99, pp. 215. 12 Tvergaard, V., 1982, Ductile Fracture Cavity Nucleation Between Large Voids, J. Mech. Phys. Solids, 30, pp. 265286. 13 Tvergaard, V., and Needleman, A., 1984, Analysis of the Cup-cone Fracture in a Round Tensile Bar, Acta Metall., 32, pp. 157169. 14 Needleman, A., and Tvergaard, V., 1984, An Analysis of Ductile Rupture in Notched Bars, J. Mech. Phys. Solids, 32, No. 6, pp. 461490. 15 Chu, C., and Needleman, A., 1980, Void Nucleation Effects in Biaxially Stretched Sheets, ASME J. Eng. Mater. Technol., 102, pp. 249256. 16 Simo, J., and Ortiz, M., 1985, A Unied Approach to Finite Deformation Plasticity Based on the Use of Hyperelastic Constitutive Equations, Comput. Methods Appl. Mech. Eng., 49, pp. 221245. 17 Moran, B., Ortiz, M., and Shih, C. F., 1990, Formulation of Implicit Finite Element Methods for Multiplicative Finite Deformation Plasticity, Comput. Methods Appl. Mech. Eng., 29, pp. 483514. 18 Tvergaard, V., 1993, Model Studies of Fiber Breakage and Debonding in a Metal Reinforced by Short Fibers, J. Mech. Phys. Solids, 41, pp. 13091326. 19 Du, Z. Y., Liu, J. Y., Zhang, S. L., and Zhu, X. W., eds., 1995, A Concise Handbook of Engineering Materials, Electronical Industry Press in Chinese. 20 Zhou, C., Yang, W., and Fang, D., 1999, Strength and Damage of Metal Matrix Composites, Acta Mech. Solida Sinica, 31, pp. 372377 in Chinese. 21 Biner, S. B., 1994, The Role of Interfaces and Matrix Void Nucleation Mechanism on the Ductile Fracture Process of Discontinuous Fiber-Reinforced Composite, J. Mater. Sci., 29, pp. 28932902.

Conclusions

Incorporating interface and matrix damage, the present analysis reveals the damage evolution in cooling, thermal cycling, and the subsequent tension of MMC. During cooling, the CTE mismatch plays a central role in provoking a high residual stress distribution tensile in front of the ber end, and shear around the ber corner. Interface damage occurs along the ber end and matrix damage initiates near the ber corner. The initial damage results in substantial stress drop nearly a half of the ultimate strength in the overall ow curve of the computation cell when subjected to mechanical loading at the room temperature. The drop is caused by the interface debonding along the ber ends. The stress drop reduces the toughness of the MMC. The ultimate strength corresponds to the debonding stress, the critical strain for the stress drop corresponds to the debonding strain. Both debonding stress and debonding strain decrease as the range of thermal cycling increases, and increase as the relative interface thickness / r f , adjustable by interface coating, increases. Cooling and thermal cycling change the failure modes of MMC. Under uniaxial tension, the SiC whisker-2124 aluminum alloy fails by matrix damage in front of the ber end when cooling and thermal cycling are absent. With cooling but not thermal cycling, it fails by debonding along the ber ends and transverse damage across the matrix. With the further participation of thermal cycling, the MMC fails by debonding along the ber ends and ber sides, followed by necking instability of the matrix ligaments. Though this conclusion is specied to the MMC of SiC whisker-2124 aluminum alloy, and the results might change if the damage parameters of matrix material are altered, we still expect the occurrence of three failure modes mentioned above in other MMC when cooling and thermal cycling are involved. The effect of thermal cycling can be measured by a dimensionless ratio of / L f ( m f ) T . The numerator indicates the largest separation the interface can sustain without debonding, and the denominator measures the mismatch due to a thermal cycling of temperature range T .

208 Vol. 122, APRIL 2000

Transactions of the ASME

Probabilistic Mesomechanics for High Cycle Fatigue Life Prediction

Robert G. Tryon Thomas A. Cruse
Fellow ASME Brentwood Technologies, Inc., Brentwood, TN 37027

This paper presents an analytical modeling approach to characterize and understand high cycle fatigue life in gas turbine alloys. It is recognized that the design of structures subjected to fatigue cannot be based on average material behavior but that designs must consider 3 or some other appropriate extreme value (tail of the distribution) loading and/or material properties. Thus, a life prediction capability useful in a design application must address the scatter inherent in material response to fatigue loading. Further, the life prediction capability should identify the key micromechanical variables that are critical in the tail of the materials durability distribution. The proposed method addresses the scatter in fatigue by investigating the microstructural variables responsible for the scatter and developing analytical and semi-analytical models to quantitatively relate the variables to the response. The model is general and considers the entire range of damage accumulation sequences; from crack nucleation of the initially unawed structure to nal fast fracture. S0094-42890001302-5

Introduction
Many material and structural design factors inuence component reliability in terms of the dened durability problems. From a material performance standpoint, many of these factors are at work in the durability size effect. Two important aspects of the size effect inuence high cycle fatigue HCF in mechanical components: the relative size of the stressed area compared to the size of the component and the relative size of the damage crack compared to the size of the microstructure. The size effect was rst reported by Peterson 1 when he noticed that the mean fatigue life and variation in fatigue life were a function of the stressed volume. The size effect must be carefully considered in regards to HCF of such components as aeroengine airfoils. The stresses that cause HCF are often mode shapes bending stresses induced by vibratory excitation. Only a very small portion of the total airfoil area is subjected to the high stresses. The size effect has another fundamental role in controlling HCF because damage accumulation often starts on a small scale. HCF failures are not usually initiated by the large microstructural defects associated with low cycle fatigue failures but often nucleate naturally at local regions of high stress. The local regions of high stress may be caused from vibratory resonance or foreign object damage. The damage grows through various mechanisms, including crack nucleation, microstructurally small crack growth, and linear elastic long crack growth. Each mechanism is associated with a characteristic size and each characteristic size has its own geometric complexities, constitutive laws and heterogeneities. Fatigue behavior cannot be fully understood and predicted without obtaining information about each of the characteristic sizes, or what can be called mesodomains. Nested models can link each of the mesodomains to determine the response of the macrodomain. The overall fatigue response of a component is predicted by nesting the individual mesoscale models. The lowest level model uses the appropriate mesoscale parameters to determine the initial state of the next level. This level uses the results from the previous level along with the appropriate parameters to determine the initial state of the next level and so on. Using nested models, eet reliability can be linked to the heterogeneities at each mesoContributed by the Materials Division for publication in the JOURNAL OF ENGINEERING MATERIALS AND TECHNOLOGY. Manuscript received by the Materials Division June 23, 1998; revised manuscript received November 1, 1999. Associate Technical Editor: H. Sehitoglu.

domain. Additionally, by modeling each level of the fatigue process individually, and rigorously linking the levels, various size effects are included.

Three Level Fatigue Model

Figure 1 shows the three levels of damage accumulation that are assumed in the present study. First, the crack nucleates on a small scale on the order of the grain size. Then the crack grows as a microstructurally small crack in which the crack front lies in relatively few grains. The material properties, averaged along the crack front, approach bulk material properties as the crack grow and the number of grains interrogated by the crack front increase. At this point, linear elasticity can be assumed and the crack grows as a typical long crack until nal failure. The models used to predict the behavior for each of the three levels of damage accumulation have been discussed elsewhere 2 and only those aspect of the models related to HCF of gas turbine alloys will be discussed. Crack Nucleation Model. Models used in the research must have two attributes. They must be quantitative with regards to the number of cycles needed to produce a crack to a specic size if they are to be used for lifetime predictions. The models must also be able to address the microstructural parameters in order to provide a physical link between the microstructure and the fatigue behavior. The crack nucleation model used in the current research addresses slip band cracking within a grain that is a preferred mode of damage accumulation for HCF in gas turbine alloys 3. The model used in this effort is based on a model proposed by Tanaka and Mura 4 and extended to account for grain orientation by Tryon and Cruse 3 as

Fig. 1 Three-stage mesomechanical fatigue model

Journal of Engineering Materials and Technology Copyright 2000 by ASME

APRIL 2000, Vol. 122 209

N n

4 GW s 2 1 2k 1 d MS

(1)

a j k j arcsin 2 c

where N n is the number of cycles needed to grow a crack to the size of the grain, G is the shear modulus, W s is the specic fracture energy per unit area, is the local applied normal stress, M S is grain orientation factor reciprocal Schmid factor, k is the frictional stress which must be overcome to move dislocations, is Poissons ratio, and d is the grain diameter. Small Crack Growth Model. The behavior of small cracks differs from the behavior of long cracks. Long crack behavior can be predicted using conventional continuum based LEFM techniques. Small crack growth rates vary widely, from several orders of magnitude greater than that predicted by continuum based K to complete arrest. A small crack can be thought of as a crack with a size on the order of the microstructure. The anomalous growth of small cracks has been attributed to two competing factors: high growth rates due to lack of closure and growth retardation due to microstructural obstacles. Crack Tip Opening Displacement. The experimentally observable parameter that has been correlated to the varying small crack growth rate is the crack tip opening displacement CTOD 5 da C t dN

i j1

i 1 k i 1 i k i arcsin

L i1 c

(3)

where, i is the applied resolved shear stress in the i th grain, k i is the frictional stress of the i th grain, a is the crack length, c is the crack length plus slip band length, L i is the distance from the free surface to grain boundary of the i th grain preceding the slip band tip as shown in Fig. 2. The CTOD is given by

2k ja c i1 k i1 i k i ln g a ; c , L i1 2A a i j1 2A

g a ; c , L L ln

c 2 L 2 c 2 a 2 c 2 L 2 c 2 a 2

a ln

a c 2 L 2 L c 2 a 2 a c 2 L 2 L c 2 a 2

(4)

A G /2 1 for edge dislocations A G /2 for screw dislocations For the slip band blocked by the grain boundary, the size of the slip band zone is

(2) The CTOD is given by

L n a

(5)

where a is the crack length, N is cycles, t is the CTOD, and C is a material constants derived from test data. The CTOD is a measure of the amount of damage associated with the crack tip. The larger the CTOD, the higher the crack growth rate. This phenomenon was rst observed by Laird and Smith 6 and has been well established in long crack growth behavior 7. The direct proportionality of Eq. 2 has been observed in small crack growth of aluminum, nickel and titanium alloys 8. Nisitani and Takao 9 showed that small crack arrest could be associated with zero CTOD. In the current research, the CTOD is modeled as a function of the random microstructural variables based on the approach used by Tanaka et al. 10 and extended by Tryon 2. Consider a crack of length a with the crack tip in the j th grain as shown in Fig. 2. The slip band has a length of w with the slip band tip in the n th grain. The total length of the damage, c, is the crack length plus the slip band length. If the slip band is propagating not blocked by the grain boundary, the size of the slip band zone can be found from

2k a c c 2 a 2 2j ln A A a

i j1

i1 k i1 i k i g a ; c , L i1 2A

2k j a arccos j c
n

i j1

2 i 1 k i 1 i k i L i1 arccos j c K m c

(6)

The microscopic stress intensity factor at the slip band tip is (7)

To account for the crystallographic orientation of the individual grains, the applied resolved shear stress in Eqs. 3 through 7 can be replaced with

for surface grains MS for interior grains MT

where is the local normal applied stress, M S is the reciprocal Schmid factor and M T is the Taylor factor 12.

Modeling the Physical Microstructure

Consider a random array of grains as shown in Fig. 3. A crack nucleates in the surface grain X 0 and then grows along the x axis as a semi-circle through zones in which the effective material properties are uniform. The boundaries of the zones are represented by the concentric half circles. The zones are composed of grains represented by the semi-circular arc segments. The arc length of the semi-circular segments is a random variable equal to the grain diameter. The surface grains are represented by the inTransactions of the ASME

Fig. 2 Crack tip slip band in multiple grains

210 Vol. 122, APRIL 2000

As the crack becomes long, P n eff approaches the bulk properties. Using the concepts of effective material properties, crack growth is modeled as one-dimensional. Consider a cut along the x axis Section A-A in Fig. 3. The fatigue damage is modeled as a one-dimensional crack growing through zones of varying size l n and varying effective material properties P n eff . The above microstructural modeling technique is approximate and does not capture some of the nuances of crack/microstructure interactions. In particular, the model does not allow for spatial variation of properties along the crack front that can cause a nonsmooth or ragged crack front shape. If the crack front encounters strong grains due to unfavorable orientation or high frictional stress in a matrix of weaker grains, the crack front will retard in the region near the strong grains and tunnel in the region of the weak grains. However, crack growth mechanisms tends to have a smoothing effect on the crack front shape. The crack front will tunnel around the blockage until the shape of the crack front at the blockage is such that the stress intensity overcomes the blockage and the crack front resumes its smooth shape 11. If the blockage is not overcome, the crack front will not continue to tunnel. The crack growth will arrest. Long Crack Growth Model. The long crack growth is modeled using the Paris law representation of a surface crack in a semi-innite body subjected to a constant stress cycle. If the nal crack size is much greater than the initial crack size, Tryon and Cruse 12 showed that N g tersection of the zones and the surface. Note that because the crack grows as a semi-circle, the surface crack growth is the same as the in-depth crack growth through the zones represented by Section A-A. After successful crack nucleation, the crack grows from grain X 0 into zone 1. In the example shown in Fig. 3, zone 1 contains three grains. The effective material property P 1 eff , of zone 1 is the average of the properties of the individual grains. P 1 i weighted with the area of the i th grain. In the current study P 1 eff represents the local frictional strength k or the local applied stress .
2 2 P 11d 2 1 P 12d 12 P 13d 13 P 1 eff 2 2 d2 1 d 12 d 13

Fig. 3 Array of random grains

a i1 n /2 C n n

n 1 2

(10)

where, N g is the number of cycles needed for the crack to grow to failure, a i is the initial crack size at the start of the long crack growth phase, is the global stress range, is the geometry constant 1.12, and C and n are based on material properties.

Monte Carlo Simulation Model

The statistical characteristics of variables used in the Monte Carlo simulation have been discussed in detail in Tryon and Cruse 13. The random variables and the associated statistical distributions are shown in Table 1. Normalized distributions are used for d, a, k, and C. This allows the average values to be easily changed without having to re-evaluate the distribution parameters. The distribution parameters only need to be re-evaluated if a change in coefcient of variation COV is desired. The orientation factors are not normalized. A change in the average value of the orientation factor would require texturing the microstructure. The values

(8)

In the n th zone composed of j grains, the effective material property is P n eff

2 ij 1 P ni d ni 2 ij 1 d ni

(9)

Table 1 Values used in the Monte Carlo simulation Variable C C d da G k M K crit MS MT n WS Description Paris Law Coefcient CTOD Law Coefcient Grain diameter Small Crack Growth Interval Bulk Shear Modulus Frictional Strength Critical Microstructural Stress Intensity Factor Schmid Orientation Factor Taylor Orientation Factor Paris Law Exponent Specic Fracture Energy Poissons Ratio Applied Micro-stress Distribution type Lognormal Deterministic Lognormal Deterministic Deterministic Weibull Deterministic Distribution Parameters 0.034 0.076 N/A N/A N/A N/A 12 12 N/A N/A N/A Average COV 0.30 N/A 0.40 N/A N/A 0.30 N/A 0.08 0.13 N/A N/A N/A 0.30

0.30 0.39 3.7

1.12

4.4 10 9 MPam 0.10 55.8 m 0.5 76 10 3 MPa 69 MPa 769 MPam 2.21 3.07 3 440 kN/m 0.3 Variable*

Curve Fit See Ref. Curve Fit See Ref. Deterministic Deterministic Deterministic Normal 1

0.3

*Note: Several different stress levels are modeled as discussed in the Section on Model Results.

Journal of Engineering Materials and Technology

APRIL 2000, Vol. 122 211

in Table 1 would no longer be valid and new representations would be required. The deterministic input variables are also shown in Table 1. The basic ow of the Monte Carlo simulation is outlined as follows. A crack is nucleated in each surface grain of a component. Each crack goes through the small crack growth phase and long crack growth phase. The total life associated with each crack is the summation of the cycles in the crack nucleation, small crack growth and long crack growth phases. The life of the component is equal to the minimum total life of all of the cracks. The details of the simulation are described in Tryon 2.

Model Results
The predictions of the individual crack nucleation, small crack growth, and long crack growth simulations have been shown to correlate very favorably with experimental observations and are discussed elsewhere 3,12. In this section, we will discuss the prediction of the total fatigue life for a simple test specimen. Predicted Total Fatigue Life of a Test Specimen. It is difcult to compare the probabilistic model predictions for total fatigue life directly with experimental data because the parameters used in the model are usually not reported. However, the predicted scatter in fatigue data is compared with trends in the experimental data and the predicted mean life for different size specimens is compared with size effect observations. The distribution of fatigue life for test specimens was predicted by assuming the parameters in Table 1 in the Monte Carlo analysis. These values are characteristic of a stainless steel. The specimen has a circular cross section with radius 7.62 mm, and a shallow notch with a gauge surface area of 1.61 mm2. This gauge surface area results in about 4000 grains per specimen. Different specimens will have a different number of surface grains and therefore the number of surface grains is a random variable. The predicted CDF of fatigue life for the specimens is shown in Fig. 4. The mean life of the specimens is 60,000 cycles with a COV of 0.17. Figure 4 shows that tting the model results to a lognormal distribution give a correlation coefcient of 0.993. Fitting the model results to a normal distribution not shown gives a correlation coefcient of 0.999. Both the normal and lognormal distributions provide an adequate representation of the model results and both have been used to represent experimental data 14,15, p. 380. A thorough investigation of the scatter in fatigue life is not available in the literature for most alloys. Many manufacturers, particularly in the aerospace industry, have the large compilation
Fig. 5 Predicted fatigue life distribution plotted on lognormal paper

Fig. 4 Fatigue life distribution of the specimens plotted on lognormal probability paper

of data used for statistical characterization. However, the cost associated with such test is considerable and the data is tightly held. However, Bastenaire 14 performed a thorough investigation of the scatter in fatigue life for ve different grades of low alloy steel. Steels may nucleate cracks by mechanisms other than slip band cracking depending on the alloy composition and the impurities. However, the trend in the scatter in steel data has been observed in other metallic alloys 16. Bastenaire performed rotating bending fatigue experiments for many stress levels for each grade of steel with several hundred specimens for each stress level. Figure 6 shows the trends in the scatter exhibited in Bastenaires data plotted on lognormal paper. The curves are replotted from the data in Fig. 7 of 14. If data plots as a straight line in Fig. 6, the lognormal distribution is valid. If data plots as a nonstraight line in Fig. 6, the lognormal distribution is no longer valid. The general trend is that the COV indicated by the slope of the curves is fairy constant for applied stresses well above the fatigue limit 363324 MPa. As the applied stress decreases, the COV starts to increase 304285 MPa. As the applied stress approaches the fatigue limit, the fatigue life increases and run-outs start to occur. The right tail of the distribution becomes more heavily populated than a lognormal distribution, which causes a line through the data to bend to the right 265245 MPa. The 363 MPa data curves slightly to the left indicating the right tail of distribution is less populated than a lognormal distribution and the data can also be tted to the normal distribution. As the applied stress decreases, the curvature shifts to the right. Comparison of Fig. 5 with the results in Fig. 6 shows that the model predicts all of the above trends observed in the experimental data. Figure 7 presents the same data in the familiar form of a SN diagram. The runouts suspensions are the percentage of specimens that did not fail at 106 cycles. The Monte Carlo simulation showed that most of the failures were caused by the largest grain in the specimen and almost all the failures were initiated in one of the ve largest grains. The lower the stress the more failures initiated in the largest grain. This indicates that the weak links in crack nucleation are the largest grains. Experimental evidence shows that failures can be associated with the largest grains 11. The distribution of the largest defects or the largest grains in the present model lead to the size effect model developed by Weibull 17. Size effect is the phenomenon that small components have a higher fatigue life than larger geometrically similar components. Weibull assumed that the larger component is more Transactions of the ASME

212 Vol. 122, APRIL 2000

Fig. 6 Fatigue life test data plotted on lognormal paper data from Bastenaire

Fig. 8 Predicted stress versus life curve

likely to have a larger life-controlling defect. This approach is widely used in the design of ceramics and it has also been applied to ductile materials 18. The reliability of different size dened by the mean number of surface grains specimens was determined and the mean fatigue strength at an arbitrary life is plotted against size in Fig. 8. The model indicates that very large structures have zero life. This is because a lognormal distribution of grains allows an innitely large grain in an innite population. In reality, the grain size cannot be innite and the true distribution of grain size is truncated at a size no larger than the component. Maximum truncated grain size would control the fatigue life of a very large structure. The predicted size effect on fatigue strength is linear in log space as shown in Fig. 8. Trantina 18 predicted the same relationship using a weakest link theory. Trantinas experimental observations on smooth, bolt hole and sharp notched specimens have been scaled with respect to fatigue strength for comparison with the model predictions in Fig. 8. A direct comparison cannot be made because the data exhibited is for a different material than that modeled. The important point demonstrated by Fig. 8 is that the model predicts that the fatigue life decreases linearly with an increase in the log of volume or surface area. The intercept of the line depends on the specied fatigue life. The slope of the line, which represents the sensitivity of the material to size effect, de-

pends on the scatter of the fatigue strength controlling variables and can vary with material processing and material alloy 18. Sensitivities. The sensitivities shown in Figure 9 represent the sensitivity of the total fatigue life COV to the random variable COV. To change in total life COV a Monte Carlo simulation was performed using the nominal variations in Table 1. Then a separate Monte Carlo simulation was performed for each of the random variables in which the COV the random variable was decrease by 5 percent. The sensitivities have been normalized such that the summation of sensitivities is one. Figure 9 shows that at low stress high cycle fatigue, the variation in fatigue life is most sensitive to the variation in the grain orientation. It is well known that texturing can greatly effect high cycle fatigue life. The variation in high cycle fatigue life is shown to be least sensitive to the variation in grain size. The Monte Carlo simulation showed that at low stress, the largest grains were responsible for the failure-causing crack. It would seem that the fatigue life would be sensitive to the grain size distribution. However, the distribution of the largest grains in each specimen is an extreme-value distribution and will only change slightly with a 5 percent decrease in the COV of grain size for all of the grains in the specimen. Figure 9 shows that at high stress low cycle fatigue, the variation in fatigue life is most sensitive to the variation in the applied

Fig. 7 Predicted mean fatigue life for various size specimens

Fig. 9 Importance of the random variable variation on the fatigue life variation

Journal of Engineering Materials and Technology

APRIL 2000, Vol. 122 213

microstress. In low cycle fatigue, the crack tip plastic zone is large and not as sensitive to the local material property variations. The scatter in fatigue life is more sensitive to grain size variations in low cycle fatigue than in high cycle. This is because the failurecausing crack is less likely to be associated with the largest grain in the specimen. The distribution for the entire population of grain size in the specimen will effect the failure causing cracks in low cycle fatigue. High cycle fatigue is a function of the extreme value distribution of grain size.

Acknowledgments
This work was supported by NASA SBIR Contract NAS 399040, Dr. P. L. N. Murthy, Technical Representative, and GSRP Project No. NGT-51053, Glenn Research Center, Dr. C. C. Chamis, Technical Advisor.

References
1 Peterson, R. E., 1939, Methods of Correlating Data from Fatigue Test of Stress Concentration Specimens, Contributions to the Mechanics of Solids, Macmillan, pp. 179183. 2 Tryon, R. G., 1996, Probabilistic Mesomechanical Fatigue Model, Ph.D. thesis, Vanderbilt University. 3 Tryon, R. G., Cruse, T. A., 1998, A Reliability-Based Model to Predict Scatter in Fatigue Crack Nucleation Life, Fatigue Fract. Eng. Mater. Struct., 21, pp. 257267. 4 Tanaka, K., Mura, T., 1981, A Dislocation Model for Fatigue Crack Initiation, ASME J. Appl. Mech., 48, pp. 97103. 5 Chan, K. S., Lankford, J., 1983, A Crack Tip Strain Model for the Growth of Small Fatigue Cracks, Scr. Metall., 17, pp. 529532. 6 Laird, D., Smith, G. C., 1962, Crack Propagation in High Stress Fatigue, Philos. Mag., 7, pp. 847857. 7 Weertman, J., 1979, Fatigue Crack Propagation Theories, Fatigue and Microstructure, ASM, Metals Park, Ohio, pp. 279206. 8 Hicks, M. A., Brown, C. W., 1984, A Comparison of Short Crack Growth Behavior in Engineering Alloys, Fatigue 84, Engineering Materials Advisory Services Ltd., England, pp. 13371347. 9 Nisitani, H., and Takao, K-I., 1981, Signicance of Initiation, Propagation and Closure of Microcracks in High Cycle Fatigue of Ductile Metals, Eng. Fract. Mech., 15, No. 3, pp. 445456. 10 Tanaka, K., Kinefuchi, M., and Yokomaku, T., 1992, Modelling of Statistical Characteristics of the Propagation of Small Fatigue Cracks, Short Fatigue Cracks, Miller, K. J., and de los Rios, E. R., eds., ESIS 13, Mechanical Engineering Publications, London, pp. 351368. 11 Forsyth, P., 1969, The Physical Basis of Metal Fatigue, American Elsevier Publ., New York. 12 Tryon, R. G., Cruse, T. A., 1997, Probabilistic Mesomechanical Fatigue Crack Nucleation Model, ASME J. Eng. Mater. Technol., 19, No. 1, pp. 6570. 13 Tryon, R. G., Cruse, T. A., 1995, Probabilistic Mesomechanical Fatigue Crack Initiation Model, Phase 1: Crack Nucleation, ASME/JSME Pressure Vessel and Piping Conference, Honolulu, HI, Published in PVP-95-MF2. 14 Bastenaire, F. A., 1972, New Method for the Statistical Evaluation of Constant Stress Amplitude Fatigue-Test Results, Probabilistic Aspects of Fatigue, Ed., Heller, R. A., ASTM STP 511, pp. 328. 15 Dieter, G. E., 1986, Mechanical Metallurgy, McGraw-Hill, Third Edition. 16 Sasaki, S., Ochi, Y., Ishii, A., Hirofumi, A., 1989, Effects of Material Structures on Statistical Scatter in Initiation and Growth Lives of Surface Cracks and Failure Life in Fatigue, JSME Inter. J., Series I, 32, No. 1, pp. 155161. 17 Weibull, W., 1961, Fatigue Testing and Analysis of Results, Pergamon Press. 18 Trantina, G., 1981, Statistical Fatigue Failure Analysis, J. of Test. Eval., 9, No. 1, DD. 4449.

Summary
This study develops a probabilistic mesomechanical fatigue model to relate the variation in the material microstructure to the variation in the fatigue life of macrostructural components. Only the microstructural effects were investigated. Variations in the applied loading, stress concentrations, residual stresses, and global geometry are not considered. Single-phase polycrystalline components are modeled. Grain shape is assumed equiaxial and the grain orientation is untextured. Loading and material properties within a grain are homogeneous although not isotropic and vary from grain to grain. Component geometries are simple smooth test specimens. The fatigue process is divided into three phases. The rst phase is the crack nucleation phase. The second phase is the small crack growth phase. Local microstructural variables considered random are grain size, grain orientation, micro-stress and frictional stress. The variables are common to both the crack nucleation and small crack growth models. The third phase is the long crack growth phase. Long crack growth rate is modeled using Paris law and microstructural variations are not explicitly considered. All variation in long crack growth is model by allowing the Paris law coefcient to be a random variable. The model predicted many aspects of fatigue observed in the experimental data. These include: The shape of the total fatigue life distribution. The applied global stress effects on the shape of the total fatigue life distribution. The knee in the SN curve and run-outs. The size effect. This study demonstrates the feasibility of developing probabilistic mesomechanical material models that can link the variation in the material microstructure to the scatter in fatigue life.

214 Vol. 122, APRIL 2000

Transactions of the ASME

Z. Wu J. Lu
Ph.D. Student. Professor. Mem. ASME LASMIS, University Technologie de Troyes, 12 Rue Marie Curie-BP2060-10010, Troyes, Cedex, France e-mail: lu@univ-troyes.fr

Study of Surface Residual Stress by Three-Dimensional Displacement Data at a Single Point in Hole Drilling Method
interferometry, Twyman Green interferometry, and blind hole A method combining moire drilling method is proposed for simple and accurate determination of residual stress. The relationship between the three-dimensional surface displacements produced by introducing a blind hole and the corresponding residual stress is established by employing the Fourier expansion solution containing a set of undetermined coefcients. The coefcients are calibrated by 3D nite element method. The surface in-plane displacements U x , U y , and the out-of-plane displacement U z produced by the relaxation of residual stress are interferometry and Twyman Green interferometry, respectively, after measured by moire the hole-drilling procedure. The complete three-dimensional displacement data at any single point around the hole can be used for residual stress determination. The accuracy of the method is analyzed and the experimental procedure is described to determine the sign of residual stresses. As an implementation of the method, a shot peening residual stress problem is studied. S0094-42890000802-1

Introduction
Strain gage hole-drilling method was rst proposed by Soete 1 and continuously developed by many contributors 28. Strain gage hole drilling method is the most widely used residual stress measurement method. For planar uniform residual stresses, it is believed reliable, simple, and fast to implement. It requires relatively simple equipment and modest operator skills. However, in the method, strain data are only obtained at three specied directions and random measurement errors are inevitable. Moreover, for blind-hole problems, the radial distribution of surface deformations due to the relaxation of residual stress is very nonuniform and highly localized to the boundary of the hole. The nonlinear distributions of the deformations depend on both the hole depth-to-diameter ratio and the residual stress which is unknown previously. As the size of a standard strain gage rosette is typically two to four times the hole diameter, r 0 and 30 deg of circumferential angle, ASTM E837-92 9, the region covered by a strain gage is too large compared to the released strain eld. Thus, the average strain measured within the covered area, which will be used to determine residual stress, might not be equivalent to the possible strain at the calibration point. Another commonly recognized drawback of the method is the eccentric hole-drilling error. inRecently, some laser interferometry methods such as moire terferometry and holographic interferometry have been used in conjunction with hole-drilling technique as an alternative means for residual stress studies 1016. These optical methods can provide whole-eld displacements around the hole to cope with the drawbacks of strain gage method. However, they often require three measurement points with large angle intervals, which is time consuming and easy to introduce measurement errors. interferometry and TwymanGreen interIn this study, moire ferometry are to be used together to provide complete threedimensional displacement elds when a blind-hole is drilled. The three-dimensional displacement data at any single point is to be
Contributed by the Materials Division for publication in the JOURNAL OF ENGINEERING MATERIALS AND TECHNOLOGY. Manuscript received by the Materials Division September 29, 1997; revised manuscript received October 14, 1999. Associate Technical Editor: Kwai S. Chan.

used for residual stress determination. This approach can save measurement time and improve the measurement accuracy. By using multipoint analysis, random measurement errors can be reduced. The proposed method is to be implemented for a shot peening residual stress problem.

Interferometry and Twyman Background: Moire Green Interferometry

interferometry is an optical method, providing real time Moire and whole eld contour maps of in-plane displacements. The high displacement measurement sensitivity and high spatial resolution made it suitable for a broad range of problems in solid mechanics. interferometry is A typical setup of an optical system for moire depicted schematically in Fig. 1. The specimen grating of frequency f s is replicated on the surface of the specimen. According to grating diffraction equation, Eq. 1, two beams of symmetrically incident light are mutually coherent and are diffracted by the specimen grating to different diffraction orders, sin m sin m f s (1) where is the incident angle, m is diffraction orders, m is the direction of the m th order diffraction beam, is the wavelength of the laser, and f s is the frequency of the specimen grating as shown in Fig. 1. When the relation of Eq. 1a is satised, the 1 and 1 diffraction orders emerge as normal to the specimen, sin f s (1a)

If the specimen grating is perfect, the two diffracted beams coexist in space but their angle of intersection is zero. Their mutual interference produces a uniform intensity throughout the eld, which is called the null eld. When the specimen deforms, the specimen grating deforms correspondingly and the resulting interference patterns represent the contours of constant U x and U y displacements. The displacements can then be determined from the fringe orders by the following relationships 17 et al., U x x , y 1 N x, y , 2 fs x U y x , y 1 N x,y 2 fs y (2)

Journal of Engineering Materials and Technology Copyright 2000 by ASME

APRIL 2000, Vol. 122 215

and u z , respectively, in the cylindrical coordinate system. The corresponding load distribution for each value of n has the same harmonic form, and thus n 0, 2, 4, . . . For biaxial planar residual stresses, as the rst order approximation, n 0, 2, the displacement eld near a blind hole can be expressed as 12 u r r , A xx y y B xx y y cos 2 2 xy sin 2 u r , C xx y y sin 2 2 xy cos 2 (5)

u z r , F xx y y G xx y y cos 2 2 xy sin 2 where, A, B, C, F, and G are undetermined coefcients which depend on the material constants and the geometrical parameters of the blind hole; xx , y y , and xy are residual stress components in a Cartesian coordinate system; u r , u , and u z are surface displacements in a cylindrical coordinate system.
Fig. 1 Schematic setup of an optical system combining moire interferometry and TwymanGreen interferometry

Determination of Calibration Coefcients

where N x and N y are fringe orders in U x and U y elds, respectively. In routine practice, a frequency of 1200 lines/mm is used, providing a contour interval of 0.417 m/fringe order. TwymanGreen interferometry was originally introduced to test optical elements 18. A combined setup of TwymanGreen interferometry is also depicted in Fig. 1. interferometry and moire The collimating beam of laser light is separated by a half mirror. One beam of light, traversing the half mirror, is intersected by a high quality reference mirror. The returning light from the mirror is reected by the half mirror to the CCD camera. Another beam of light reected from the half mirror is intersected by the specimen grating at the normal of the grating plane. A high quality grating with a metal coating is similar to a at mirror since the grating frequency is very high. The returning light from the grating traverses from the half mirror to the CCD camera. The half mirror is at 45 deg to the incident light. Both the normal of the specimen grating and the reference mirror coincide to the incident and emergent light, thus, their mutual interference produces a uniform intensity throughout the eld, too. When the specimen has an out-of-plane deformation, the change of the light path will produce a interfering fringe pattern representing contours of U z displacement by the following relationship 18, U z x , y N x,y 2 z (3) The coefcients A, B, C, F, and G in Eq. 5 can be determined by three-dimensional nite element analyses by using two specic loading cases: 1 xx y y , xy 0, an equibiaxial residual stress eld, which is equivalent to a uniform pressure p acting on the hole boundary. By using Eq. 5, the coefcients A and F are determined as, A E,,r0 ,r,

h u r r , r0 2

F E,,r0 ,r,

h u z r , (6) r0 2

2 xx y y , xy 0, a pure shear residual stress eld, which is equivalent to the harmonic distributions of the normal stress rr cos 2 and the shear stress r sin 2 acting on the hole boundary. By using Eq. 5 again, the coefcients B, C, and G are determined as, B E,,r0 ,r, C E,,r0 ,r, G E,,r0 ,r,

where is the wavelength of the laser and N z is the fringe orders representing the out-of-plane displacement contours. The combination of Eqs. 2 and 3 provides complete threedimensional displacements in the whole eld with the same order of sensitivity.

h u r r , r0 2 cos 2 h u r , r0 2 sin 2 h u z r , r0 2 cos 2 (7)

3D Displacement Field for Blind Hole Drilling Method

For blind hole problems, a Fourier expansion solution was proposed by Schajer 3 to account for the effect of out-of-plane displacement and was expressed as u r r , , z

In Eqs. 6 and 7, u r , u , and u z are the surface displacements calculated from the nite element analysis. It is to be noted that the calibration coefcients are independent of angle . The coefcients depend on the material constants of Youngs modulus E and Poisson ratio , the geometrical parameters of the hole radius r 0 , hole depth h, and the radial coordinate r at the calibration point.

n0

u
n0

Determination of Residual Stress

interferometry provides displacement data with the CarMoire tesian coordinates. The radial displacement U r , the circumferential displacement U can be expressed in terms of U x and U y as,

rn r , z cos n

u r , , z u z r , , z

n r , z sin n

(4)

U r r c , U x x , y cos U y x , y sin U r c , U x x , y sin U y x , y cos

(8)

n0

zn r , z cos n

where u rn , u n , and u zn are the contributions of the n th displacement components to the total displacement components u r , u , 216 Vol. 122, APRIL 2000

1 where x 2 y 2 r 2 c and tan (y/x). By combining Eqs. 2, 3, 5, and 8, the relationship between the residual stress and the fringe orders at a surface point ( r c , ) around the hole can be rewritten in a matrix form as,

Transactions of the ASME

 N x x , y N y x , y

cos sin A B cos 2

2 f s A B cos 2

N x x , y N y x , y

sin cos

xx 2 B sin 2 y y xy

2 f s C sin 2 N x , y F G cos 2 2 z

C sin 2

xx 2 C cos 2 y y xy

F G cos 2

xx 2 G sin 2 y y xy

(9)

The three unknown residual stress components xx , y y , xy can be determined by solving the above three homogeneous linear equations.

Fig. 3 U z eld fringe pattern modied by carrier fringes of rotation with orthogonal fringe vectors

Determination the Sign of Residual Stress

interferometry and TwymanGreen Fringe patterns of moire interferometry, produced by residual stresses with the same distribution and the opposite signs, are identical to each other. The sign of the fringe orders should be determined during the experiments since the sign of residual stresses is previously unknown. There interare many methods available for sign determination in moire ferometry and TwymanGreen interferometry, but the simplest and the most convenient way is the carrier fringe method. In interferometry, if the specimen grating and the incident moire beams have a relative rigid body rotation around each other without the change of incident angle, a series of uniformly spaced fringes perpendicular to the grating lines will be created, which are called the carrier fringes of rotation. Similarly, if the angle of the incident beams is changed, a series of uniformly spaced fringes parallel to the grating lines will be created, which are called the carrier fringes of extension or compression. When carrier fringes are added to a load induced fringe pattern, the resultant fringe gradient vector at every point of the fringe pattern will be the superposition of the original and the carrier fringe gradient vectors. This is clearly illustrated in Fig. 2, where the moire fringes which represent the released U x displacement eld under the equibiaxial residual stress state are modied in different ways by two known carrier fringes of rotation with opposite signs. At two typical regions of Fig. 2, the original displacement gradient F 0 is changed in both direction and magnitude due to the effect of carrier fringe of rotation F c as illustrated in the inserted parts. In interferTwymanGreen interferometry, similar to those in moire ometry, the carrier fringes can be created by rotating the reference mirror and the fringe lines are perpendicular to the direction of rotation. The carrier fringes of ration can be superposed to the load induced out-of-plane displacement as illustrated in Fig. 3 for the same residual stress problem, where the originally circular U z fringes are altered in different ways when the orthogonal carrier fringes are introduced. For residual stresses have identical magnitude but opposite signs, the fringe gradient at any point of the eld should have opposite directions. Thus, the carry fringes of rotation or extension can alter the residual stress induced fringes in opposite ways. Therefore, the sign of residual stress is easy to determine by this method.

Accuracy Analysis
From Eq. 9 we know that the accuracy of residual stress result depends on the accuracy of calibration coefcients and fringe orders. Generally, calibration coefcients can be determined accu interrately for given material geometrical parameters. In moire ferometry and TwymanGreen interferometry, the integral and fractional fringe orders can be determined accurately through the intensity distribution analysis of the fringe patterns or phase shifting technique. The reasonable accuracy in fringe order determination will be 0.05 fractional fringe orders which represents interferometry ( f s 0.02 m of displacement in moire 1200 lines/mm), and 0.016 m in TwymanGreen interferometry ( 632.8 nm). The accuracy would be sufcient for many residual stress problems. As the fringe counting uncertainties are random measurement errors, the average value of the random displacement errors should tend to zero if a large amount of measurement points are considered. Therefore, the residual stress measurement accuracy can be further improved through multipoint analysis. It is important to notice that, in the previous hole-drilling methods, the three measurement points should be chosen with relatively large angle intervals so that the homogeneous linear equations are not ill-conditioned. However, in the proposed method, the homogeneous linear equations, Eq. 9, composed by the complete three-dimensional displacements at a single point are always well-conditioned. APRIL 2000, Vol. 122 217

pattern modied by carrier fringes of roFig. 2 U x eld moire tation with opposite fringe vectors

Journal of Engineering Materials and Technology

Implementation of the Method for Shot-Peened Specimen

In order to implement this method, a shot peening residual stress problem was studied. Shot peening processing is widely used in engineering to increase the working life under fatigue load or to enhance the resistance to surface corrosion of metallic or metal based composite components by introducing moderate compressive residual stresses in the near surface layer of the material 19. The specimen was cut from a large plate of AS10U3NG aluminum alloy, one side of which was shot peened. The material properties were: E 79.5 GPa, 0.3. The holographic specimen grating with a frequency of 1200 lines/mm was replicated on the shot peened surface of the specimen at room temperature. Based on a previous study, compressive residual stress existed within a 0.3 mm surface layer, thus a blind hole of 2.0 mm in diameter and 0.3 mm in depth was carefully drilled at the grating side of the specimen. It was drawn schematically in Fig. 4. The drill bit was controlled precisely at a speed of 1.0 m/s during the hole drilling procedure and the machining stress proved experimentally to be very small. The inuence of grating delamination and the plastic zone was diminished within 0.1r 0 . The bottom of the hole was at.

Fig. 6 U y displacement eld obtained from shot peened AS10U3NG aluminum alloy specimen with the hole radius of 1.0 mm and the hole depth of 0.3 mm

Fig. 4 Schematic drawing of the shot peened specimen containing compressive residual stress

Fig. 7 U z displacement eld obtained from shot peened AS10U3NG aluminum alloy specimen with the hole radius of 1.0 mm and the hole depth of 0.3 mm

Fig. 5 U x displacement eld obtained from shot peened AS10U3NG aluminum alloy specimen with the hole radius of 1.0 mm and the hole depth of 0.3 mm

After the hole was drilled, the specimen was put into the com interferometry and TwymanGreen bined optical system of moire interferometry as shown in Fig. 1. The fringe patterns representing the contour maps of U x , U y , and U z displacements were captured by a high resolution CCD camera (1317 1035) as shown in Figs. 57. The sign of the fringes was determined by carrier fringe method. Generally, in order to improve the accuracy in fringe counting, the displacements should be measured as close as possible to the boundary of the hole. By considering the inuence of grating delamination, the displacement data in this experiment were obtained at r c 1.2r 0 . Shot peening was a random process and thus, the in-plane residual stress produced by the process is regarded to be a uniform equibiaxial residual stress state xx y y , xy 0 . Based on the above analysis, in order to improve the measurement accuracy, residual stress was determined separately from 16 points around the hole with an identical angle interval of /8. The fringe orders of U x , U y , and U z displacements at the points (1.2r 0 , n /8) are listed in Table 1. The calibration coefcients Transactions of the ASME

218 Vol. 122, APRIL 2000

Table 1 Fringe orders obtained from the U x , U y , and U z fringe patterns at the corresponding points

N Nx Ny Nz 0 2.1 0 0.7

/8
2.05 0.75 0.7 9/8 2.3 0.9 0.7

/4
1.8 1.7 0.7 5/4 1.7 1.6 0.7

3/8 1.05 2.2 0.7 11/8 1.1 2.0 0.65

/2
0.1 2.4 0.7 3/2 0 2.2 0.65

5/8 0.9 2.2 0.7 13/8 0.6 2.0 0.65

3/4 1.95 1.95 0.7 7/4 1.5 1.55 0.7

7/8 2.4 1.0 0.7 15/8 1.95 0.8 0.7

N Nx Ny Nz

2.45 0 0.7

Table 2 Calibration coefcients determined by 3D nite element analyses when r 0 1.0 mm and h 0.3 mm at r c 1.2 r 0 A 0.4566 10
5

B 0.644 10
5

C 0.3376 10
5

F 0.1075 10
5

G 0.4852 10 5

1 Nonuniform Planar Residual Stress Evaluation. For planar nonuniform residual stresses, which exist in some local regions, when a hole is drilled through these regions, the relaxation of residual stress at different parts has no effect on each other as shown in Fig. 9. The relationship between the local residual stress and the surface displacements can be proposed as, u r r , A r rr B r u r , C r r u z r , F r rr G r where A , B , C , F , and G are calibration coefcients which can be determined by using the similar nite element analysis FEA. The three-dimensional displacement data at the coefcient calibration point can be used for the local residual stress determination.
Fig. 8 Residual stress determined at the radial coordinate r c 1.2r 0 and the angle interval 8 for AS10U3NG aluminum alloy specimen

(10)

determined at r c 1.2r 0 by Eqs. 6 and 7 are listed in Table 2. By using Eq. 9, residual stress is determined, respectively. The determined residual stress results at these points are plotted together in Fig. 8. It is obvious that the maximum compressive residual stresses are 140.0 MPa for xx and 148 MPa for y y , respectively. The minimum compressive residual stresses are 113.7 MPa for xx and 109.7 MPa for y y , respectively. The average values of xx and y y at the points around the hole are 129.5 MPa and 128.6 MPa, respectively. The determined shearing stresses from these points are within 30.0 MPa and the average value of the shear stresses at the points around the hole is 2.2 MPa. Therefore, the average results can reect the equibiaxial compressive residual stress induced by a shot peening procedure.

Discussion
The proposed method in this study could be used in the following cases: Journal of Engineering Materials and Technology
Fig. 9 Potential application of the proposed method for planar nonuniform residual stresses

APRIL 2000, Vol. 122 219

2 Determination of Residual Stress Distribution in Depth. As the accuracy of the proposed method is sufciently high, by combining the method with incremental hole-drilling technique, the distribution of residual stress in depth can be determined by using the similar procedure as above. The accuracy can also be improved by multipoint measurements.

6 7 8 9

Conclusions
A method is proposed to determine residual stress by threedimensional displacement data at a single point. The relationship between the three-dimensional surface displacements produced by introducing a blind hole and the corresponding residual stress is established by using a Fourier expansion solution with a set of calibration coefcients. The three-dimensional surface displace inments are provided by whole-eld and high sensitivity moire terferometry and TwymanGreen interferometry. The method is sufciently accurate. Through multipoint analysis and average processing, accuracy of residual stress result is further improved. As an implementation, a shot peening residual stress problem is studied. The equibiaxial compressive residual stress is found. More applications of the method are anticipated.

10 11 12 13 14 15 16 17 18 19

References
1 Soete, W., 1949, Measurement and Relaxation of Residual Stress, Sheet Met. Ind., 26, No. 266, pp. 12691281. 2 Kelsey, R. A., 1956, Measuring Non-Uniform Residual Stresses by the Hole Drilling Method, Proc. SESA, No. 1, pp. 181194. 3 Schajer, G. S., 1981, Application of Finite Element Calculations to Residual Stress Measurements, ASME J. Eng. Mater. Technol., 103, No. 4, pp. 157 163. 4 Lu, J., Niku-Lari, A., and Flavenot, J. F., 1985, Mesure de la Distribution des thode du Trou Incre mentale, Contraintes Residuelles en Profondeur par la Me moire et Etudes Scientiques, Revue de la Me tallurgie, Feb., pp. 6981. Me 5 Lu, J., and Flavenot, J. F., 1989, Applications of the Incremental Hole-

Drilling Method for Measurements of Residual Stress Distribution, Exp. Tech., 13, No. 11, pp. 1824. Flaman, M. T., and Manning, B. H., 1985, Determination of Residual Stress Variation with Depth by Hole-drilling Method, Exp. Mech., 25, No. 3, pp. 205207. Schajer, G. S., 1988, Measurement of Non-Uniform Residual Stresses Using the Hole Drilling Method, Part IStresses Calculation Procedures, ASME J. Eng. Mater. Technol., 110, No. 4, pp. 338343. Schajer, G. S., 1988, Measurement of Non-Uniform Residual Stresses Using the Hole Drilling Method, Part IIPractical Applications of the Integral Method, ASME J. Eng. Mater. Technol., 110, No. 4, pp. 344349. ASTM E837-92, 1992, Standard Test Method for Determining Residual Stresses by the Hole Drilling Strain-Gage Method, Annual Book of ASTM Standards, Section 3, 03.01, pp. 747753, American Society for Testing and Materials, Philadelphia, PA. Antonov, A., 1983, Inspecting the Level of Residual Stresses in Welded Joints by Laser Interferometry, Weld Prod., 30, No. 9, pp. 2931. Interferometry Determination of Residual Stresses Nicoletto, G., 1991, Moire in the Presence of Gradients, Exp. Mech., 31, No. 3, pp. 252256. Makino, A., and Nelson, D., 1994, Residual Stress Determination by Single Axis Holographic Interferometry and Hole Drilling, Part I: Theory, Exp. Mech., 3, pp. 6678. Makino, A., and Nelson, D., 1997, Determination of Sub-surface Distributions of Residual Stresses by Holographic Hole-Drilling Technique, ASME J. Eng. Mater. Technol., 119, No. 1, pp. 95103. Nelson, D., Fuchs, E., Makino, A., and Williams, D., 1994, Residual Stress Determination by Single Axis Holographic Interferometry and Hole Drilling, Part II: Experiments, Exp. Mech., 3, pp. 7988. Wu, Z., Lu, J., and Han, B., 1998, Study of Residual Stress Distribution by a Interferometry and Incremental Hole-Drilling, Combined Method of Moire Part I: Theory, ASME J. Appl. Mech., 65, No. 4, pp. 837843. Wu, Z., Lu, J., and Han, B., 1998, Study of Residual Stress Distribution by a Interferometry and Incremental Hole-Drilling, Combined Method of Moire Part II: Implementation, ASME J. Appl. Mech., 65, No. 4, pp. 844850. Experimental Post, D., Han, B. and Ifju, P., 1994, High Sensitivity Moire Analysis for Mechanics and Materials, Springer-Verlag. Brown, M., and Wolf, E., eds., 1985, Principles of Optics, Sixth Edition, Pergamon Press, pp. 302. Metal Improvement Company, Inc., 1981, Applications du Shot Peening.

220 Vol. 122, APRIL 2000

Transactions of the ASME

I. W. Park
Integrated Surgical Systems, Inc., 1850 Research Park Dr., Davis, CA 95616

A Study of Burr Formation Processes Using the Finite Element Method: Part I
A nite element model of orthogonal metal cutting including burr formation is presented. A metal-cutting simulation procedure based on a ductile failure criterion is proposed for the purpose of better understanding the burr formation mechanism and obtaining a quantitative analysis of burrs using the nite element method. In this study, the four stages of burr formation, i.e., initiation, initial development, pivoting point, and nal development stages, are investigated based on the stress and strain contours with the progressive change of geometry at the edge of the workpiece. Also, the characteristics of thick and thin burrs are claried along with the negative deformation zone formed in front of the tool edge in the nal development stage. S0094-42890000702-7

D. A. Dornfeld
Fellow ASME Department of Mechanical Engineering, University of California, Berkeley, Berkeley, CA 94270-1740

Introduction

Burrs are dened as undesirable projections of the material formed as the result of the plastic ow from cutting and shearing operations. Manufacturing of precision components often requires a deburring or nishing operation. Studies have shown that for manual deburring of parts, the time required to deburr a part increases exponentially with an increase of burr thickness. Hence automation of the deburring process has become a prime objective as part of efforts to automate the entire production system. However, deburring automation is quite difcult, especially forcebased control technique for precision components due to burr prole variation. Thus, under any circumstance, any attempts to automate deburring requires a reliable model of burr formation. More important, understanding of the burr formation mechanism may allow prevention or minimization of burrs. In most machining operations, rollover burrs are formed at the end of cut as a result of a chip pushed out of the cutters path rather than shearing. There have been many studies to predict the general burr formation characteristics in both ductile and brittle materials. The rst quantitative analysis of burr formation was performed by Gillespie 1 assuming that the burr is formed by bending deformation of the end portion of the workpiece. He characterized the burr by the distance of the tool from the end of the workpiece at which bending deformation starts. Despite the unsuccessful quantitative prediction of the burr size, the qualitative effect of each parameter is quite often observable. Iwata et al. 2 observed the burr formation during machining inside a scanning electron microscope SEM, and experimentally determined the effect of exit angles of the workpiece on the burrs. They explained the fracture of the workpiece at the tool edge using the strain obtained by the nite element method FEM analysis. Ko and Domfeld 3 developed a model for burr formation in materials exhibiting ductile and brittle behavior, and evaluated the model at slow machining speeds. Later, Chern and Dornfeld 4,5 continued the development of burr formation models, extended Ko and Dornfelds model with more realistic machining operations and conditions. They also included the study of edge breakout phenomenon as well as burr formation. Pekelharing 6 tried to explain poor tool life and described the formation of foot in interrupted cutting with sharp tools at the exit of cut. In general, closed-form analytical solution for general problems of elastic-plastic with large deformation such as chip and burr
Contributed by the Materials Division for publication in the JOURNAL OF ENGINEERING MATERIALS AND TECHNOLOGY. Manuscript received by the Materials Division June 26, 1998; revised manuscript received December 31, 1999. Associate Technical Editor: Kwai S. Chan.

formation processes are very difcult to derive. Thus, simulation analysis using the nite element method technique is a reasonable approach to model the metal-cutting and burr formation processes. In the last two decades, a great deal of research has gone into the establishment of FEM techniques for orthogonal cutting. In 1973, Klamecki 7 introduced the study of metal cutting using FEM by presenting nite element models in his Ph.D. dissertation. In 1984, Usui and Shirakashi 8 made a signicant contribution to the development of FEM as the best predictive theory because it can predict chip formation, cutting force, and distributions of stress and strain without any input from cutting experiments. Later, Iwata et al. 2 presented a rigid-plastic nite element model for steady-state plane strain orthogonal machining. Strenkowski and Caroll 9 also presented nite element models based on general purposed two-dimensional nite element code NIKED2D. Recently, Park et al. 10 conducted a nite element analysis of burr formation in ductile material. In this study, a general purpose nite element software package, ABAQUS/ Standard, was used to simulate the chip and burr formation processes in orthogonal cutting. Based on the geometrical change at the edge of the workpiece and a series of stress and strain contours, the fundamental burr formation mechanism was found and divided into four stages as the tool approaches the edge of the workpiece. The stages are initiation, initial development, pivoting point, and nal development stages. The results from the nite element analysis were also qualitatively veried with the experimental data. Although the nite element analysis provided physical insight into the fundamental burr formation mechanism, a nal burr conguration could not be simulated due to assumptions such as an isothermal condition and ignorance of strain-rate inuence. The isothermal condition is assumed so that the temperature of the workpiece remains constant at ambient throughout the simulation. However, during the cutting of metal, high temperatures are generated in the region of the tool cutting edge, and these temperatures have a controlling inuence on the friction between the chip and tool. Also, considering the geometry of the process, plastic ow at these high temperatures implies that the strain rate is quite high for the chip to form because of the geometry of the process. The most signicant problem in the previous model was the concept of the parting line and a separation criterion. During the burr formation process, the distorted parting line was constantly rezoned to a predened location associated with the metal-cutting process involving extensive computing. Further, the distance separation criterion 11 was adopted to simulate the metal-cutting and burr formation processes under the assumption that the tool tip radius is quite small compared to the depth of the cut. This APRIL 2000, Vol. 122 221

Journal of Engineering Materials and Technology Copyright 2000 by ASME

approach was sufcient to nd the fundamental burr formation mechanism but is inappropriate to give a quantitative estimate of burr dimensions. The objective of this study is to develop a nite element model to further investigate the burr formation process including the analysis of a burr or edge breakout. In this study, the burr formation process in each stage is reexamined. An explicit dynamics FEM software package, ABAQUS/Explicit, is used for the simulation of the burr formation. The ABAQUS/Explicit code has several important features, which make it more suitable for studying burr formation problems. First, instead of the parting line criterion, a ductile failure model is adopted to separate the chip from the workpiece. Second, an adiabatic heating model is adopted to simulate the heat generation effects due to plastic work of the workpiece and chip. Thus, the temperature-dependent material properties in this case, 304 L stainless steel SS304L and strainrate property are incorporated into the nite element analysis.

Table 1 Material properties of SS304L and cutting conditions

Finite Element Modeling

The two-dimensional nite element model is generated under the plane strain assumption that the width of the cut is at least ve times greater than the depth of the cut 0.5 mm, i.e., w 5 t . The mesh conguration and tool geometry is shown in Fig. 1. The element type used in this model is a four-node bilinear and reduced integration with hourglass control to deal with the large deformation caused by the chip and burr formation processes. Also, the tool is assumed to be perfectly rigid. Total number of elements and nodes are 1054 and 1180, respectively. In this model, the boundary ABC is constrained against displacements ( U 0), and the boundary CDA is unconstrained ( F 0) to allow a burr or edge breakout to form. The nite element model with an incorporated initial chip geometry is large enough to quickly reach steady-state cutting conditions before the burr initiation occurs. Also, total ve element groups for the workpiece part of the model are assigned and tied together with *CONTACT PAIR denition and *TIED option using ABAQUS/Explicit 12. The perfectly rigid tool with a rake angle of 5 degrees contains a slide-line capability along the rake face to describe the chip ow as the tool advances through the workpiece. Also, a slide-line on the rake face allows friction to form a secondary shear zone due to interaction on the tool-chip interface. Coulomb friction of 0.3 is used in this model. However, it should be pointed out that higher Coulomb friction would be more appropriate to describe the interaction on the tool-chip interface in actual metal cutting. The rigid tool advances through the stationary workpiece with a uniform velocity of v 5.0 m/s. The heat generated during metal cutting, especially in high speed cutting, is of importance in the chip formation process. The

thermal model assumes adiabatic conditions within each element so that no heat transfer occurs within the workpiece and on the workpiece free surface. Adiabatic assumption is typically used in the simulation of high speed manufacturing processes involving large amounts of inelastic strain, where the heating of the material due to its deformation is an important effect because of temperature-dependence of the material properties, i.e., thermalelastic-plastic material behavior. The fraction of inelastic dissipation rate that appears as heat ux per volume, i.e., the work-toheat conversion factor, is 0.8 in the nite element model. For adiabatic heating, the temperature change can be written as T Q , Cp (1)

where T is the temperature change, Q is the heat generated per unit volume, and and C p are the material density and specic heat, respectively. The temperature increase is calculated directly at the material integration points, and temperature is not a degree of freedom in this model. As strain rates increase during metal cutting, many metals show an increase in their yield strength. In primary shear zone, material is usually deformed at a bulk shear strain rate on the order of p 103 to 105 s1 13. This effect is very important as it commonly occurs in high energy dynamic events or in manufacturing processes. The model for this purpose is pl D

1 o

for o ,

(2)

Fig. 1 A nite element model of burr formation

where is the effective yield stress at a nonzero strain rate, o is the static yield stress, pl is the uniaxial equivalent plastic strain rate, and D and p are material parameters that may be functions of temperature and, possibly, of other predened state variables. In Transactions of the ASME

222 Vol. 122, APRIL 2000

 and p are used under the assumpthis study, constant values of D tion that they are temperature-independent and not inuenced by any other predened state variables. Due to the adiabatic assumption, the workpiece boundary of the nite element model is insulated as shown in Fig. 1. The thermalelastic-plastic material properties of SS304L along with the cutting conditions are given in Table 1. Finally, no build-up-edge BUE formation or tool wear is assumed.

Solution Procedure

For the simulation of the metal cutting and burr formation processes, ABAQUS/Explicit, an explicit dynamics nite element program is used 12. The explicit dynamics analysis procedure is based upon the implementation of an explicit integration rule together with the use of diagonal or lumped element mass matrices. The equations governing motion of the body are integrated using the explicit central difference integration rule u i 1/2 u i 1/2 and u i 1 u i t i 1 u i 1/2 , (4) t i 1 t i i u 2 (3)

Fig. 2 Stress-strain curves of the ductile failure model

where the subscript i refers to the increment number, u represents the displacement vector, and t represents the time increment. The acceleration at the beginning of the increment is computed by u M 1 F I , (5)

where M is the diagonal lumped mass matrix, F is the applied load vector, and I is the internal force vector. The central difference integration operator is explicit in that the kinematic state may be advanced using known values of u ( i 1/2) and u ( i ) from the previous increment. From Eqs. 3 and 4, it can be noticed that the explicit integration rule is quite simple; furthermore, the computational efciency of the explicit procedure is obtained by the use of a diagonal lumped matrix. The explicit procedure requires no iterations and no tangent stiffness matrix. The explicit procedure integrates through time by using many small increments. The central difference operator is conditionally stable, and ABAQUS/Explicit uses an adaptive algorithm to determine conservative bounds for the stable time increment. The time incrementation scheme in ABAQUS/Explicit is fully automatic and requires no user intervention. The use of small increments allows the solution to proceed without iteration. Although the explicit dynamics analysis procedure can apply to the analysis of slower processes, it is obvious that it is ideally suited for analyzing high speed dynamic events. The analysis proceeds by integrating the equations of motion 3 and 4 with a time increment size determined by the time incrementation scheme. While the analysis takes an extremely large number of increments, each increment is relatively inexpensive and often economical. For the analysis of the metal cutting and burr formation presented here, approximately 400,000 iterations are performed.

tion under the assumption that the tool edge is relatively sharp compared to the depth of cut, that is, the ratio of the tool tip radius to the depth of cut is on order of 10 2 . Considering the work in previous FEM studies, a new approach to the simulation of a metal cutting process including both steadystate cutting and burr formation is proposed here. The metalcutting process developed in this study is based on the ductile failure model using the *FAILURE option in ABAQUS/Explicit 12. This simple failure criterion is designed to allow the stable removal of elements from the mesh as a result of tearing or ripping of the structure. ABAQUS/Explicit automatically deletes elements from the mesh as they exceed the failure criterion. The failure model is based on the value of the effective plastic strain. As shown in Fig. 2, when any element of the mesh reaches pl o , damage on the corresponding element initiates. And when this damage reaches pl f due to further deformation, the material point fails, and the element loses its ability to resist any further load. The damage is calculated from the effective plastic strain as D pl pl o pl pl f o , (6)

where pl is the current effective plastic strain of the material. Also, the materials elastic response is based on the damaged elasticity. The damaged elastic moduli are given by G D 1 D G and K D 1 D K , (8) (7)

where G D and K D are the damaged shear modulus and bulk modulus, respectively. Figure 2 shows the unloading path along the damaged modulus. The damaged plastic yield surface is dened as

yd 1 D y pl .

(9)

Metal-Cutting Simulation Procedure

In order to separate the chip from the workpiece in nite element analysis of metal cutting, various material separation criteria, such as an energy approach method 8, a total effective plastic strain approach 9, a distance separation method 11, etc. 14,15, have been proposed. In steady-state cutting, the proposed material separation criterion remains constant and ensures the smooth chip ow on the rake surface of the tool in order to predict the shear angle formed from the tool edge to the free undeformed surface. However, to include the burr formation process, implementing a suitable separation criterion becomes the most difcult obstacle in the nite element simulation. Recently, Park et al. 10 used the distance separation criterion to simulate the burr formaJournal of Engineering Materials and Technology

This causes the yield surface to shrink to a single point in stress space when the damage reaches a value of one. In this study, the metal-cutting procedure is performed based on the ductile failure criterion by removing the elements in contact with the tool edge as the tool advances through the workpiece. As a matter of convenience, the ductile failure criterion is dened as FC i pl pl o , f , (10)

where i represents the element number. It should be noted that the ductile failure model approach does not necessarily reect the mechanism involving the actual material separation during the metal cutting operation. Instead, the ductile failure criterion values are assigned to delete the elements contacted with the tool tip and should assure that the surrounding elements exhibit steady-state cutting and burr formation characteristics as shown in Fig. 3. APRIL 2000, Vol. 122 223

Fig. 4 Mesh deformation in steady-state cutting

ing on the machined surface and the normal stress built up on the rake surface of the tool. In steady-state cutting, two constraints should be enforced to insure the validity of the ductile failure criterion as follows:
Fig. 3 Metal-cutting simulation procedure by ductile failure model

pl pl o f f T , and pl o d ,

(15)

(16)

In order to determine the values of and in 10, the metal-cutting process in the nite element model is divided into three parts: steady-state cutting, initiation to pivoting point stage, and pivoting point to nal development stage. 4.1 Ductile Failure Criterion in Steady-State Cutting. Instead of attempting to predict the shear angle or chip formation in orthogonal cutting, the ductile failure criterion values are determined with shear angle information obtained from the metalcutting handbook 16 according to the cutting conditions and tool geometry. With the information of the shear angle, pl o is found under the assumption that the only shear occurs on the shear plane and can be represented by and d pl d ) , (12) cos sin cos (11)

pl o

pl f

where f represents the effective plastic strain at the maximum fracture strain value a function of the temperature, T, and d is the effective plastic strain value of the element d, shown in Fig. 3. Constraints 15 indicate that pl o should not exceed f ( T ) which is considered as the upper bound value of the ductile failure criterion. Also, constraint 16 indicates that during steady-state cutting, any contact mechanism characteristics should be avoided; otherwise, the element d is removed before the element h is removed in Fig. 3. This phenomenon takes place if pl o is overestimated so that the workpiece is largely dragged by the tool. Therefore, although Eq. 11 and 13 provide the initiation point for the steady-state cutting simulation, the ductile failure criterion values should be accordingly adjusted by constraints 15 and 16. Figure 4 shows the mesh deformation in steady-state cutting and exhibits the characteristics of orthogonal cutting such as primary shear zone, secondary shear zone, etc. The analysis shows that the ductile failure criterion approach to chip separation from the workpiece is as effective as the other separation criteria discussed ealier. 4.2 Ductile Failure Criterion From the Initiation to Pivoting Point Stage. In this part, we utilize the information obtained from the previous FEM study. The initiation stage represents the point where the plastically deformed region appears on the part edge. The main contributor to the initiation is the high compression developing at the edge of the workpiece; whereas, high tensile stress beneath the tool edge develops. This equilibrium state indicates that the burr initiation is similar to bending deformation. Hence, after the initiation stage, bending takes place at the edge of the workpiece. This stage is called the initial development stage. As the tool approaches the edge of the workpiece, it is found that the effective plastic strain of the element closest to the tool edge monotonically increases with an increase of the amount of bending: dc 0, de (17)

represent a shear strain and shear strain on the where and shear plane, respectively, and and are rake and shear angles, respectively. Assuming that the shear deformation on the shear plane begins with o 0, Eq. 12 is integrated from 0 to and gives
pl o ) . (13)

Thus, the ductile failure criterion on the i th element is determined by FC i pl pl o , o , pl pl f o . (14) where 0. Hence In this study, is chosen not only to assure the cutting characteristics of the surrounding elements, but also to control the plastic work rate for the chip separation from the workpiece. It is obvious that with an increase of , the total strain energy increases for the ductile failure, i.e., chip separation from the workpiece. Hence, the energy rate required for the chip separation in the nite element model also increases. As a result, the numerical results indicate that the size of gives some effects on the amount of residual stress remain224 Vol. 122, APRIL 2000

where c represents the effective plastic strain closest to the tool tip and forms a part of the chip, and e is the largest effective plastic strain at the edge of the workpiece, representing the amount of bending at the edge. It now follows that we introduce the interpolating function: Transactions of the ASME

Fig. 5 Constraint for allowable machined surface nish

pl o F 1 ,2 , . . . ,n

,
i1 i i

(18)

where i is the weighting factor, and i represents the element number surrounding the element contacted with the tool edge in Fig. 3. The weighting factor should be subjectively chosen and should be validated by constraints 15, 16, and 17. However, one more constraint should be enforced to assure the consistency of cutting process in this part by
R R e ,

(19)

where R represents the distance from the reference point to predened machining surface, R e is the distance from the reference point to the machining surface obtained from the nite element simulation, and is the element size near the machining surface, shown in Fig. 5. Constraint 19 indicates that the allowable surface nish in the nite element simulation should be less than the element size on the machined surface. Therefore, pl o should be accordingly determined to assure the burr formation environment specied by the constraints. Also, for the element damage procedure should be determined to assure a continuous burr formation process. In this part of the simulation, the four constraints 15, 16, 17, and 19 are enforced to validate the ductile failure criterion values obtained from the interpolating function 18. It should be pointed out that constraint 19 would be the most important one because the smaller the element size, the more difcult the choice of the ductile failure criterion values become despite obtaining more reliable results. In this study, the element size is determined based on constraint 19 with consideration of the computation time. 4.3 Ductile Fracture Criterion From the Pivoting Point to Final Development Stages. The pivoting point stage represents the point where material instability at the edge of the workpiece takes place due to high stress built-up. From this point, catastrophically large deformation, known as the rollover process, occurs instead of bending deformation. In the pivoting stage, the length from the pivoting point and cutting surface is known as burr thickness. After the pivoting point stage, the chip no longer forms, and material in front of the tool is pushed ahead by both shearing and plowing processes. This stage is called the nal development stage where the rollover process continues to give nal burr shape or edge breakout. For the simulation from the pivoting point to nal development stages, constraints 15 and 19 are enforced to determine the ductile failure criterion values from Eq. 18. However, constraint 17 is not enforced since no information is available with respect to the relationship between the effective plastic strain near the tool edge and the amount of rollover at the edge. Constraint 16 is not appropriate in this case because the edge breakout can take place. Therefore, we introduce another constraint instead of constraint 16 that Journal of Engineering Materials and Technology

Fig. 6 Element groups of a FEM model and cutting path in thick burr simulation

pl f fT pl f fT

burr breakout. (20)

Constraint 20 can be considered to be the criterion to predict the formation of a burr or edge breakout. It should be noted that the edge breakout phenomenon involves fracture, and a crack forms near the tool edge and propagates into the workpiece. Although the ductile failure criterion would be inappropriate as a burr/ breakout criterion from the viewpoint of fracture mechanics, it is adopted in this study since the precise criterion is not known yet. As shown in Fig. 6, the nite element model is divided into three regions. Region I simulates the chip formation, and the ductile failure model is not applied by assuming that a continuous chip forms. Region II supports the metal-cutting process. Since this region primarily involves elastic deformation, the ductile failure model is omitted. The ductile failure model is applied in region III to simulate metal cutting and burr formation. Also, Fig. 6 shows that a total of seven ductile failure criteria are determined along the cutting path for the simulation of a thick burr with their locations.

Simulation Procedure

The owchart of the nite element simulation of chip and burr formation is shown in Fig. 7. First, the simulation starts with assigning the ductile failure criterion values for steady-state cutting. If the ductile failure criterion values are not properly assigned, then they are adjusted according to the constraints. Otherwise, the ductile failure criterion values are kept and used to describe the steady-state cutting process. When it reaches the initiation stage, the ductile failure criterion values for steady-state cutting are no longer valid. From this point, the ductile failure criterion values must be constantly adjusted based on the constraints to describe the burr formation process. This procedure is continuously repeated until all ductile failure criterion values from APRIL 2000, Vol. 122 225

Fig. 7 Flowchart of metal cutting and burr formation simulation

the steady-state cutting to nal development stages are found. Table 2 shows the ductile failure criterion values applied to the simulation of the thick and thin burrs. However, it should be noted that the complete temperature-dependent maximum fracture strain values of SS304L are not available for this study. Therefore, we assume that the ductile failure criterion values are validated for a burr formation process in Table 2. The computation is performed on an IBM RS6000 Model 590 workstation.

point and the negative deformation zone. The simulated results are compared and veried with SEM micrographs obtained from the actual metal-cutting data using the micro machining stage designed by Stiles 17 and modied by Ko 3. In order to explain the development of the pivoting point, a simple model, shown in Fig. 10, is introduced. In the initiation stage, yielding occurs at the edge of the workpiece, and the main contributor to the yielding is the compressive stress developing at the edge of the workpiece. This phenomenon is similar to the bending mechanism. Also, in this stage, the initial pivoting point is established. In Fig. 10, the cutting force acting on the machined surface causes the bending or deection of the workpiece. On the other hand, since the workpiece is elastically deformed below the yielding zone, it is assumed that the workpiece from the yielding zone to the unmachined workpiece surface is attached and xed at the workpiece below the yielding zone. Therefore, in order to statically balance forces and moments, i.e., F M 0, the negative moment must be introduced around the yielding zone. In the nite element model, this moment is displayed in terms of the positive shear. The development of the positive shear is quite important to determine the nal pivoting point in burr formation. Figure 11 shows the development of the pivoting point which initially forms quite a bit below the predened machined surface and gradually moves up to form at the edge of the workpiece in the pivoting point stage. The initial formation of the pivoting point is quite important not only for determining the burr thickness but also for the development of burrs in the nal development stage. The prediction of the nal location of the pivoting point is quite important for the deburring process because the length of the burr thickness is measured from the machined surface to the nal location of the pivoting point. As mentioned previously, the time required to deburr a part increase exponentially with an increase of burr thickness. It is found that the location of the pivoting point would be largely determined by the plastic work rate required in the steady-state cutting since the burr initiation is similar to bending deformation. Hence, in order examine the burr thickness characteristics without any signicant change of shear angle, two dif-

Results

The results from the nite element simulation are presented in this section to enhance the understanding of the burr formation mechanism. With the proposed metal-cutting simulation procedure, the study is extended to investigate the physical phenomenon occurring in the nal burr development stage. The characteristics of thick and thin burrs in orthogonal cutting are claried along with the investigation of the development of the pivoting
Fig. 8 Progressive change of geometry in thick burr simulation Table 2 Ductile failure criterion for thick and thin burrs

Fig. 9 Progressive change of geometry in thin burr simulation

226 Vol. 122, APRIL 2000

Transactions of the ASME

Fig. 12 Negative deformation zone and SEM micrograph of a thin burr

Fig. 10 Formation of positive shear on the workpiece edge

ferent values of in Eq. 10 are assigned. However, it should be noted that is chosen to examine the inuence of the total strain energy for ductile failure modes on the burr formation processes within the bounds of the constraints. Figures 8 and 9 show the progressive changes of geometry at the edge of the workpiece of the thick and thin burr formation, respectively. With a small value of , the thin burr is formed; whereas, with a large value of , the thick burr is formed. They clearly exhibit that a plastic

Fig. 13 Negative deformation zone and SEM micrograph of a thick burr

Fig. 11 Development of positive shear on the workpiece edge in burr formation

hinge, known as the pivoting point is formed at the edge of the workpiece, as described in Fig. 10. Also, according to numerical results, for a thick burr, the location of the pivoting point from the predened machined surface appears further below the depth of cut ( l c t ), and the edge of the workpiece exhibits a smooth contour. On the other hand, in the thin burr case, l n t , and the edge of the workpiece shows a sharp corner. Chern and Dornfeld 4 have developed a formula to predict the burr size in orthogonal cutting. One of the parameters in the equation is the angle of the negative deformation zone. After the initiation stage, the shear angle tends to decrease as the edge of the workpiece deects. As a result, high negative shear stress in front of the tool edge develops; whereas, positive shear is formed around the pivoting point in Fig. 11. As the tool further approaches the edge of the workpiece, the negative shear formed near the tool edge increases and greatly inuences the rollover process. In a thin burr case, Fig. 12 shows that the negative shear formed near the tool edge extends to the edge of the workpiece as seen in the deformed mesh, and the result is compared to a SEM micrograph of micro-burr formation. As a result, the rollover process is initiated by the shearing effect, which results in a sharp corner at the edge of the workpiece. On the other hand, in a thick burr case, the pivoting point is formed quite below the predened machined surface; consequently, the negative deformation zone does not extend to the edge of the workpiece, as shown in Fig. 13. Since the shearing effect is not accounted for during the rollover process, the edge shows a smooth contour. This is also compared to a micrograph from micromachining stage inside the SEM. Since a burr generally forms for machining conditions used here for SS304L, edge breakout phenomenon is not considered in this study. However, it now becomes clear that if the edge breakout occurs, a crack would form in the negative deformation zone due to excessive distortion of the material near the tool edge. As the tool further approaches the edge of the workpiece, the crack would initially propagate through the negative deformation zone by mode II shear rupture. And later the edge of the workpiece would be separated from the main body by mode I opening type of fracture 18 normal rupture. APRIL 2000, Vol. 122 227

Journal of Engineering Materials and Technology

Conclusion

References
1 Gillespie, L. K., and Blotter, P. T., 1976, The Formation and Properties of Machining Burrs, ASME J. Eng. Ind., 98, Feb., pp. 6474. 2 Iwata, K., Ueda, K., and Okuda, K., 1982, Study of Mechanism of Burrs Formation in Cutting Based on Direct SEM Observation, J. Japan Society of Precision Engineering, 48-4, pp. 510515. 3 Ko, S. L., and Dornfeld, D. A., 1991, A study on Burr Formation Mechanism, ASME J. Eng. Ind., 98, No. 1, pp. 6674. 4 Chern, G. L., 1993, Analysis of Burr Formation and Breakout in Metal Cutting, Ph.D. Thesis, University of California, Mechanical Engineering Department, pp. 2542. 5 Chern, G. L., and Dornfeld, D. A., 1996, Burr/Breakout Development and Experimental Verication, ASME J. Eng. Mater. Technol., 118-2, pp. 201 206. 6 Pekelharing, A. J., 1978, The Exit Failure in Interrupted Cutting, CIRP Ann., 27, pp. 510. 7 Klamecki, B. E., 1973, Incipient Chip Formation in Metal CuttingThree Dimension Finite Element Analysis, Ph.D. dissertation, Univ. of Illinois, Urbana-Champaign. 8 Usui, E., and Shirakashi, T., 1982, Mechanics of Machining-From Descriptive to Predictive Theory, On the Art of Cutting Metals75 Years Later, ASME Publication PED, 7, pp. 1335. 9 Strenkowski, J. S., and Carroll, III, J. T., 1985, A Finite Element Model of Orthogonal Metal Cutting, ASME J. Eng. Ind., 107, pp. 349354. 10 Park, I. W., Lee, S. H., and Dornfeld, D. A., 1994, Modeling of Burr Formation Processes in Orthogonal Cutting by the Finite Element Method, ESRC Report No. 9334, Univ. of California, Berkeley, Dec. 11 Komvopoulos, K., and Erpenbeck, S. A., 1991, Finite Element Modeling of Orthogonal Metal Cutting, ASME J. Eng. Ind., 113, No. 3, pp. 253273. 12 Hibbitt, Karlsson, and Sorenson, Inc., 1988, ABAQUS/Explicit Users Manuals, Version 5.3, Providence, RI. 13 Wright, P. K., 1982, Predicting the Shear Plane Angle in Machining from Workmaterial Strain-Hardening Characteristics, ASME J. Eng. Ind., 104, No. 3, pp. 285292. 14 Lin, Z. C., and Lin, S. Y., 1992, A Coupled Finite Element Model of Thermo-Elastic-Plastic Large Deformation for Orthogonal Cutting, ASME J. Eng. Mater. Technol., 114, pp. 218226. 15 Cockcroft, M. G., and Latham, D. J., 1968, Ductility and Workability of Metals, J. Inst. Met., 96, pp. 3339. 16 ASME Research Committee on Metal Cutting Data and Bibliography, 1952, Manual on Cutting of Metals with Single-Point Tools, The American Society of Mechanical Engineering. 17 Stiles, T. A., 1985, A Scanning Electron Microscope SEM Machining Substage for Metal Cutting Observation and Acoustic Emission Analysis, MS thesis, Univ. of California, Berkeley, Department of Mechanical Engineering. 18 Barsom, J. M., and Rolfe, S. T., 1987, Fracture & Fatigue Control in Structures, Prentice-Hall Inc., New Jersey.

A nite element method of a metal-cutting process including burr formation has been presented. A new procedure for a chip separation criterion has been proposed in this study. This procedure is based on the ductile failure criterion included in ABAQUS/Explicit FEM package. The nite element simulation is divided into three parts: steady-state cutting, initiation to pivoting point stage, and pivoting point to nal development stage for burr formation. In each part, several constraints are enforced to determine the ductile failure criterion values in order to assure that the metal cutting simulation is as close to reality as possible. And an adiabatic heating model is adopted to include the heat generated by the element distortion consistent with the temperature dependent material properties of SS304L. The physical phenomenon occurring in the nal development stage is more focused than the other stages, i.e., initiation, initial development, and pivoting point stages. The characteristics of the thick and thin burrs are claried along with the investigation of the negative deformation zone based on the numerical results. It is found that the formation of thick and thin burrs are signicantly inuenced by the size of the negative deformation zone. The thin burr forms when the development of high negative shear near the tool edge extends to the edge of the workpiece; consequently, a sharp corner is formed at the edge due to the shearing effect during the rollover process. On the other hand, a thick burr exhibits a smooth contour on the edge of the workpiece because the negative deformation zone does not extend to the edge of the workpiece. This work attempts to clarify the burr formation mechanism in orthogonal cutting, especially, for the nal development stage, and also to introduce metal cutting simulation using the ductile failure criterion in order to obtain a quantitative analysis of burr/ breakout formation.

Acknowledgments
The authors would like to thank the members of the Consortium on Deburring and Edge Finishing CODEF at the University of California, Berkeley for their nancial support. The authors would also like to thank Dr. N. Rebelo Hibbitt, Karlsson, and Sorenson, Inc. for his helpful comments and suggestions.

228 Vol. 122, APRIL 2000

Transactions of the ASME

I. W. Park
Integrated Surgical Systems, Inc., 1850 Research Park Dr., Davis, CA 95616

A Study of Burr Formation Processes Using the Finite Element Method: Part IIThe Influences of Exit Angle, Rake Angle, and Backup Material on Burr Formation Processes
Finite element models in orthogonal cutting are presented in order to examine the inuences of exit angles of the workpiece, tool rake angles, and backup materials on burr formation processes in 304 L stainless steel in particular. Based on the metal-cutting simulation procedure proposed by the authors, a series of stress and strain contours and nal burr/breakout congurations are obtained. The burr formation mechanisms with respect to ve different exit angles are found, and duration of the burr formation process increases with an increase of exit angle, resulting in different burr/breakout congurations. Based on the development of negative shear stress in front of the tool tip, the tool tip damage, what is called chipping, is investigated. Also, with xed cutting conditions and workpiece exit geometry, the inuence of the rake angle is found to be closely related to the rate of plastic work in steady-state cutting because the larger the rate of plastic work in steady-state cutting, the earlier the burr initiation commences. Furthermore, in order to effectively minimize the burr size, three cases of backup material inuences on burr formation processes are examined. It is found that the burr size can be effectively minimized when the backup material supports the workpiece only up to the predened machined surface. S0094-42890001402-X

D. A. Dornfeld
Fellow ASME Department of Mechanical Engineering, University of California, Berkeley, Berkeley, CA 94270-1740

Introduction

Rollover burrs, the result of material behavior at the end of an orthogonal-like cutting process, are formed in most machining operations. Some attempts have been made to analyze rollover burrs by controlling the cutting conditions and material properties. The rst quantitative analysis of burr formation was carried out by Gillespie 1 assuming that the burr is formed by bending deformation of the end portion of the workpiece. Ko and Dornfeld 2 developed a model for burr formation in material exhibiting ductile or brittle behavior and evaluated it at slow machining speeds. Further, Chern and Dornfeld 3,4 continued the development of burr formation models, extended Ko and Dornfelds model with more realistic machining operations and conditions. Within the boundary of the model and cutting conditions, the predictions from Chern and Dornfelds orthogonal cutting model correspond well with the actual measured burr/breakout values from orthogonal impact machining tests. Pekelharing 5 tried to explain the poor tool life often seen in interrupted cutting with sharp tools at the exit of a cut. In his study, the interrupted cutting was carried out on the workpieces with various exit angles, dened as the angle between the cutting velocity and the edge of the workpiece. Iwata et al. 6 observed the burr formation during machining inside a scanning electron microscope SEM and experimentally determined the effect of the exit angles of the workpiece on the
Contributed by the Materials Division for Publication in the JOURNAL OF ENGINEERING MATERIALS AND TECHNOLOGY. Manuscript received by the Materials Division June 26, 1998; revised manuscript received December 31, 1999. Associate Technical Editor: Kwai S. Chan.

burr size. They explained the fracture of the workpiece at the tool edge using the strain obtained by a nite element method FEM analysis. Park et al. 7 conducted a nite element simulation of the burr formation using the general purpose nite element software package, ABAQUS/Standard, and attempted to model the process as close to reality as possible. In this study, based on the progressive changes of the edge of the workpiece along with a series of stress and strain contours, the fundamental burr formation mechanism was observed and divided into four stages: initiation, initial development, pivoting point and nal development. However, due to several assumptions used in the nite element model, the simulation study on the burr formation process could not be completed to provide the nal burr/breakout conguration. Recently, Park and Dornfeld 8 continued the development of the nite element model and extended the previous model with explicit dynamics nite element software, ABAQUS/Explicit. In this study, the metal-cutting simulation procedure form the steady-state to nal development stage was developed not only to conrm the fundamental burr formation mechanism obtained from the previous study 7, but also to provide a quantitative analysis of the nal burr/breakout conguration. Furthermore, the physical phenomenon occurring in the nal development stage was intensively studied. As a result, the characteristics of thick and thin burr formation were claried along with the investigation of the negative deformation zone. The objective of this study is to examine the inuences of exit angles of the workpiece edge, tool rake angles, and backup materials on the burr formation processes using the nite element method. Finite element models with various exit angles at the workpiece edge and tool rake angles are developed in this study. APRIL 2000, Vol. 122 229

Journal of Engineering Materials and Technology Copyright 2000 by ASME

In a drilling operation, backup material is often placed behind the drill exit surface of the workpiece not only to minimize the burr size but also to reduce the residual shear stress formed around the hole. Thus, in this study, the use of backup material, which comes into perfect contact with the workpiece edge, is modeled to observe the inuence of the backup material in orthogonal cutting. In each simulation, the metal cutting simulation procedure developed in the burr formation study 8 is applied, and the temperature-dependent material properties of 304 L stainless steel SS304L are also incorporated into the nite element analysis. Furthermore, the nite element study includes the effect of damage on the tool edge occurring at the exit of cut, and the results are also compared with the experimental results obtained by Pekelharing 5.

Finite Element Modeling

Fig. 2 Finite element models of 5 and 20 degree rake angles

Two-dimensional nite element models with plane strain assumption to examine the effects of exit angles, rake angles, and backup materials are developed and displayed in Figs. 13. Total number of elements and nodes in each two-dimensional model are 1080 and 1160 with plane strain assumption, respectively. In the nite element model for the workpiece, total ve element groups are assigned and tied together with *CONTACT PAIR denition and *TIED option 9. The boundary BCD is constrained against displacements ( U 0), and boundary DEFGA is unconstrained ( F 0) to allow a burr to form. For the backup material simulation, the boundary QR is constrained against displacements, and the boundary RSPQ is unconstrained. The depth of cut is 0.5 mm, and the tool is assumed to be perfectly rigid with a slide-line capability. Also, Coulomb friction of 0.3 is used on the rake surface of the tool to describe the interaction on the tool-chip interface. Furthermore, the heat model assumes adiabatic conditions within each element so that no heat transfer occurs within the workpiece and on the workpiece free surface for moderate cutting speed processes ( 5.0 m/s). The cutting process including burr formation involves large amounts of inelastic strain, where the heating of the material caused by its deformation is an important effect because of temperaturedependence of the material properties, i.e., thermal-elastic-plastic material behavior. The strain rate effect is also considered in the model because many metals show an increase in their strength as the strain rate increases during the metal-cutting process. In order to investigate the exit angle inuence on burr formation, a total of six nite element models with exit angles of 60, 80, 90, 100, and 120 degrees are introduced and displayed in Fig. 1. The gure shows that the exit angle of each model is referenced to point E at the edge of the workpiece. The distance from the predened machined surface BF to point E is determined so as to assure that the signicant part of burr formation processes, i.e., burr initiation, pivoting point, etc., takes place above point E.

Hence the workpiece below point E is furnished simply to support the metal-cutting process. For the investigation of rake angle inuence, two nite element models are developed with rake angles of 5 and 20 degrees, displayed in Fig. 2. In this part of the simulation, the exit angle of the workpiece and cutting speed are xed to 90 degrees and 5.0 m/s in each case, respectively. In the investigation of the backup material inuence, a total of three nite element models for backup material inuences are developed, displayed in Fig. 3. The rst model, Fig. 3a, examines the burr formation mechanism with a 10 mm thick backup material which comes into perfect contact with the workpiece material. The second model, Fig. 3b, use a backup material is half as thick as that in Fig. 3a in order to examine the effect of the backup material. Finally, the last model, Fig. 3c, examines the effect of the backup material supporting the workpiece up to the predened machined surface under the condition of perfect contact. The most important part of the nite element simulation is the description of the metal cutting process including burr formation. In this study, the metal-cutting simulation procedure developed in

Fig. 1 Finite element models of 60, 80, 90, 100, and 120 degree exit angles

Fig. 3 Finite element models of backup materials

230 Vol. 122, APRIL 2000

Transactions of the ASME

Table 1 Material properties of SS304L and cutting conditions

between the cutting velocity vector in orthogonal cutting and the workpiece edge, Fig. 1. It is important to observe the location of the pivoting point since it determines burr thickness. Also, it is important to observe the duration of the burr formation process and how the edge deection and rollover develop with regard to the exit angle variation. These play an important part in the development of negative shear near the tool edge. As Pekelharing 5 pointed out, the negative shear near the tool edge is the main cause of the tool tip damage, known as chipping, at the exit of a cut. In the 60 degree exit angle simulation, the pivoting point at the edge of the workpiece appears quite close to the machined surface, Fig. 4. Also, Fig. 5 shows that the shear stress contour in front of the tool edge reveals a similar pattern seen in the contact mechanisms 10, which indicates that the workpiece material in front of the tool would no longer form a chip. Moreover, since the pivoting point is too close to the machined surface, this particular stress environment could not allow the workpiece edge to roll over. Hence since the tool moves with the uniform speed, a mode II shearing type of fracture 11 would most likely take place. Consequently, it leads to edge breakout phenomenon displayed in Fig. 6. As a result, due to the edge breakout, the burr size, thickness or height, for example, can be effectively minimized with a low exit angle at the workpiece edge. When the exit angle is 80 degrees, the pivoting point appears to form a little below the machined surface in Fig. 7 compared with the 60 degree exit angle case. And this environment could allow

the study done by Park and Dornfeld 8 is applied to each nite element simulation. The temperature dependent material properties of SS304L and cutting conditions are displayed in Table 1. Table 2 shows the sequence of the ductile failure criterion determined for the simulation of 80 degree exit angle and 100 degree exit angle as the tool advances toward the edge of the workpiece.

Finite Element Simulation Results

3.1 Exit Angle Inuences. The focus of the nite element study in this section is to investigate the location of a plastic hinge, known as the pivoting point, with respect to the exit angle of the workpiece edge. The exit angle is dened as the angle

Fig. 4 Equivalent stress contour of 60 degree exit angle

Table 2 Ductile failure criterion for 80 and 100 degree exit angles

Fig. 5 Shear stress contour of 60 degree exit angle

Journal of Engineering Materials and Technology

APRIL 2000, Vol. 122 231

Fig. 9 Edge breakout of 80 degree exit angle

Fig. 6 Edge breakout of 60 degree exit angle

Fig. 10 Equivalent stress contour of 90 degree exit angle

Fig. 7 Equivalent stress contour of 80 degree exit angle

the workpiece edge to roll over to certain degree. Also, with the development of positive shear as shown in Fig. 8, the negative deformation could appear in this case. However, as the tool approaches the edge of the workpiece, according to the shear stress contour in front of the tool edge in Fig. 8, the contact mechanism characteristics still appears and dominate for 304 L stainless steel, which would lead to the edge breakout phenomenon. It is also important to note that a mode II shearing type of fracture initially occurs; however, due to the edge rollover, the crack later propagates with a mode II opening type of fracture 11, shown in Fig. 9. As a result, the burr size can be also minimized due to the

Fig. 8 Shear stress contour of 80 degree exit angle

Fig. 11 Shear stress contour of 90 degree exit angle

232 Vol. 122, APRIL 2000

Transactions of the ASME

Fig. 12 Formation of negative deformation zone in burr formation

breakout. However, there exists a small burr, a remnant from the edge rollover, as shown in Fig. 9. When the workpiece with a 90 degree exit angle is machined, Fig. 10 shows that the pivoting point forms quite below the machined surface compared to the 60 and 80 degree exit angle case.

Furthermore, the magnitude and size of the positive shear stress considerably increase, and the effect of the contact mechanism decrease, shown in Fig. 11. Consequently, the mode II shearing type of fracture would not take place, and the workpiece edge would most likely roll over. Thus the burr formation would be more likely to occur than edge breakout phenomenon around the 90 degree exit angle under the given machining conditions and material properties. During the rollover process, the magnitude of the negative shear stress near the tool edge further develops, leading to the formation of the negative deformation zone, shown in Fig. 12. This has been experimentally observed by Chern and Dornfeld 3 and veried by Park and Dornfeld 8 using the nite element method. As the exit angle increases up to 100 and 120 degrees, the magnitude and size of the positive shear stress further increase, Fig. 13, compared with the lower exit angle cases. Also, it can be observed from Fig. 14 that the higher the exit angle becomes, the lower the location of the pivoting point becomes from the machined surface. As a result, Fig. 15 shows that the burr size becomes relatively thick and long compared with the 90 and lower exit angle case although the amount of edge rotation during the rollover process decreases due to the exit geometry of the workpiece. Also, it can be observed that with an increase of exit angle, the inuence of negative deformation zone on the edge of the workpiece decreases leading to a smooth edge contour. Therefore, from the viewpoint of burr minimization and economical deburring, the workpiece with a large exit angle should be avoided. Although the tool is assumed to be perfectly rigid in the nite element model, tool tip damage during the burr/breakout forma-

Fig. 13 Shear stress contours of 100 and 120 degree exit angle

Fig. 14 Equivalent stress contours of 100 and 120 degree exit angle

Journal of Engineering Materials and Technology

APRIL 2000, Vol. 122 233

Fig. 16 Pekelharings tool life experiment results in orthogonal cutting, from 5

Fig. 15 Final burr conguration of 100 and 120 degree exit angles

tion can be indirectly examined through a series of shear stress contours of the workpiece near the tool edge. Pekelharing 5 conducted an experiment with regard to the number of cuts before tool tip damage, i.e., chipping, occurs with respect to the exit angle, Fig. 16. The gure shows that the most severe damage on the tool edge occurs around an exit angle of 90 degrees. On the other hand, the tool tip damage is considerably reduced with both increasing and decreasing the exit angles around 90 degrees despite the different burr/breakout congurations. It is known that the tool tip damage occurring at the exit of cut is mainly associated with the development of the shear stress near the tool edge during the rollover process. The development of the negative shear stress of the element closest to the tool edge was investigated as the tool approaches the edge of the workpiece. The numerical results are displayed in Fig. 17. First of all, it is important to observe that the duration of the burr formation process from the edge where burr formation commences, i.e., the burr initiation, increases with an increase of exit angle, which is also closely related to the development of the shear stress closest to the tool edge. With a 60 degree exit angle, the negative shear remains almost constant prior to edge breakout. In other words, when the tool edge is approximately 1.5 mm behind the workpiece edge, the breakout occurs immediately after the pivoting point stage followed by burr initiation, and its development. Second, any signicant change in shear stress of the element closest to the tool edge is not observed. Instead, due to the contact mechanism characteristics, the higher shear stress in front of the tool edge develops prior to the edge breakout. Thus there is no time for the negative shear to develop and effectively damage the tool tip. With the 80 234 Vol. 122, APRIL 2000

Fig. 17 Development of negative shear stress closest to the tool edge

Transactions of the ASME

degree exit angle, it is observed that the edge rollover initiates prior to the edge breakout. As shown in Fig. 17, after the rapid increase of the magnitude of the negative shear stress caused by the rollover process, the edge breakout immediately occurs. This could lead to another sudden change in the stress eld at the tool edge. Under the circumstances, the tool tip would undergo a mechanical impact-shock at the exit of cut, which could be substantially harmful to the tool tip. From the investigation of the 60 and 80 degree exit angles, it can be deduced that the magnitude of a mechanical impact-shock would depend on the amount of the edge rollover prior to the breakout. Thus the lower the exit angle becomes, the less edge rollover there is, which obviously leads to longer tool life. When the exit angle is 90 degrees, Fig. 17 shows that the burr initiation occurs early compared with the 80 degree exit angle case. The sudden decrease of the negative shear stress takes place immediately after the rapid increase of the negative shear stress due to the rollover process. Hence the development of the negative shear stress in the 90 degree exit angle would be similar to that in the 80 degree exit angle although a burr forms instead of breakout. Thus the tool tip could be still vulnerable to fracture due to a mechanical impact-shock. On the other hand, as the exit angle increases to 100 and 120 degrees, the burr initiation occurs quite early compared with the lower exit angle cases. Instead of the rapid change in the shear stress closest to the tool edge, the magnitude of the negative shear stress gradually increases during the initial development stage and gradually decreases during the edge rollover. This indicates that a mechanical impact-shock could be considerably relieved by the gradual increase and decrease of the negative shear stress compared with the 80 and 90 degree exit angle cases. As a result, the damage on the tool tip could be considerably reduced with an increase of the exit angle. 3.2 Rake Angle Inuences. In this section, the nite element simulation is conducted to examine the rake angle inuences on the burr formation. The rake is dened as the angle from the line perpendicular to the cutting velocity vector in orthogonal cutting to tool rake surface, Fig. 2. The rake angles of 5 and 20 degrees are chosen with a xed 90 degree exit angle in both cases. In steady-state cutting, the rate of plastic work is dependent on the shear angle forming from the cutting edge to workpiece free surface. According to Rowe and Spicks simple model 12 based on the minimum energy approach, i.e., the Hamiltonian principle being used, Wright 13 showed that the minimum power of primary shear for machining commercial steel with low friction on the tool-chip interface arises when the shear angle is around 48 degrees for a rake angle of 6 degrees. It is important to consider the rate of plastic work in steady-state cutting in the study of the burr formation because the mechanism of burr initiation is quite similar to the bending mechanism. Thus, it can be deduced that the larger the rate of plastic work in steady-state cutting, the earlier burr initiation begins. The simulation of steady-state cutting of SS304L in this study yielded the shear angles of approximately 28 and 40 degrees for rake angles of 5 and 20 degrees, respectively. Therefore, under the cutting conditions specied in this study, especially, with low friction on the tool-chip interface, the rate of plastic work for a 5 degree rake angle tool would be greater than that for a 20 degree rake angle tool in steady-state cutting. This certainly makes qualitative sense. In case of the 20 degree rake angle tool, the numerical result shows that the burr initiation takes place when l 20 3.3t in Fig. 18, where l 20 represents the burr initiation distance from the tool tip to the edge of workpiece with a 20 degree rake angle tool, and t is the depth of cut. For a 5 degree rake angle tool, the burr initiation takes place when l 5 4.5t , Fig. 19. Also, it can be observed from Figs. 18 and 19 that the location of yielding at the edge of the workpiece with the 20 degree rake angle is closer to the machined surface than that with the 5 degree rake angle. As a result, a thinner burr would be obtained with the 20 degree rake Journal of Engineering Materials and Technology

Fig. 18 Location of burr initiation of 20 degree rake angle

Fig. 19 Location of burr initiation of 5 degree rake angle

angle tool than that with the 5 degree rake angle tool. Gillespie 14 pointed out that for hand deburring of precision parts, the time required to deburr a part increases exponentially with increased burr thickness and burr accessibility. Therefore, it would be helpful to machine with a higher rake angle tool. However, from a viewpoint of tool wear, as the rake angle increases, the tool tip is mechanically more vulnerable to fracture. Furthermore, a mechanical impact-shock at the exit of cut could further damage the tool tip if the duration of the burr formation process is short. Hence it would be desirable to nd the optimal selection of cutting tool geometry with the consideration of both deburring efforts and tool wear. 3.3 Back-Up Material Inuences. The most effective way to minimize the burr size would be to put a back-up material behind the edge of the workpiece. Gillespie 15 conducted experiments to examine the backup material inuence for the purpose of burr minimization in drilling. Based on the Gillespies experiments, nite element models are analyzed to investigate the backup material inuence on burr formation. In this study, a backup material, whose material properties are twice as stiff Youngs modulus and twice as strong yielding and ultimate strength as those of the workpiece material, is used. Backup materials with a thickness of 10 mm and 5 mm are APRIL 2000, Vol. 122 235

Fig. 20 Equivalent stress contour with 10 mm backup material

Fig. 21 Normal stress contour in y -direction with 5 mm backup material

placed at the end surface of the workpiece with condition of perfect contact between the two materials. As seen earlier, after the burr initiation, deection at the edge of the workpiece occurs. With the 10 mm thick backup material, the deection of the workpiece causes local deformation of the backup material at point P, Fig. 20, and the backup material signicantly reduces the amount of deection at the workpiece edge. As a result, continuous chip formation continues until the tool advances toward the very end of the workpiece, and fracture takes place at the last moment. Consequently, the burr can be effectively minimized. However, the slight deection at the workpiece edge creates a small gap between the two materials. Therefore, depending on the size of the gap, a small burr, a remnant from the fracture, would be expected since the size of the gap is an indirect measure of the amount of deection or rollover at the workpiece edge. With the 5 mm thick backup material, instead of local deformation, the whole backup material exhibits the bending characteristics. Around the xed boundary QR, the normal stress contour in y-direction in Fig. 21, shows that the tensile and compressive stress develop at point Q and R, respectively. The bending of the back-up material results in a large gap between the two materials compared with the 10 mm thick backup material case, which indicates that the workpiece edge with the 5 mm thick backup material deects more than with the 10 mm thick backup material prior to the end of cut. In this case, the burr size also can be effectively minimized although a relatively large remnant, compared to the 10 mm backup material case, would be expected to be left at the edge. As a result, it would be desirable to have backup materials thick enough to cause only local deformation near the edge of the workpiece by avoiding the bending of the backup material. The nal simulation examines the case when the 5 mm backup material contacted with the workpiece only up to the predened machined surface. A similar case has been experimentally carried out by Gillespie 15 to minimize the size of a drilling burr. Although orthogonal cutting is quite different from drilling, the characteristics of the rollover process in orthogonal cutting would be similar to those seen in drilling. Initially, the deection of the edge above the predened machined surface takes place, which brings point P into contact with point F instead of forming a gap between the two materials. As the tool further approaches the edge, the burr size is effectively minimized by a mode II shearing type of fracture before the tool edge reaches the edge of the workpiece, in Fig. 22, which quite resembles the mechanism in the case of a 60 degree exit angle. As a result, this effectively reduces the size in any resulting burr and would be also the mechanism behind minimizing the size of a drilling burr in Gillespies experiment.

Conclusion

Fig. 22 Inuence of backup material partially supporting workpiece

A nite element analysis has been performed to examine the inuences of exit angles at the edge of the workpiece, rake angles, and backup materials on the burr formation processes based on the metal cutting simulation procedure proposed in the study done by Park and Dornfeld 8. In the study of exit angle inuences, when the exit angle at the edge is 60 degrees, the edge is the more susceptible to breakout than burr formation. It has been found that the pivoting point forms close to the machined surface as the exit angle decreases. As a result, the burr size can be effectively minimized because this particular environment triggers breakout by a mode II shearing type of fracture at the edge. On the other hand, when the exit angle is 90 degrees, burr formation is more favorable than breakout. From this point, considerable deburring efforts would be required. As the exit angle further increases, the amount of edge rotation during the rollover process is reduced due to the exit geometry; however, a burr becomes relatively thick and long compared with the 90 degree exit angle case since the pivoting point forms quite below the machined surface. From the viewpoint of Transactions of the ASME

236 Vol. 122, APRIL 2000

tool wear, it has been found that 80 to 90 degree exit angle would be the most damage to the tool tip because the sudden change of shear stress in front of tool tip creates a mechanical impact-shock characteristic. On the other hand, as the exit angle increases, the tool tip damage can be substantially reduced because a mechanical impact-shock on the tool tip is relieved due to the long duration time of the burr formation process. In other words, the shear stress eld closest to the tool edge does not involve impact-shock characteristics. As the exit angle decreases, the tool tip damage can be also reduced because a mechanical impact-shock hardly develops due to a decrease in the amount of edge rotation due to the development of the contact mechanism characteristics in front of the tool edge. With a xed exit angle of 90 degrees, the location of burr initiation is dependent on the rake angle in terms of the rate of plastic work, i.e., the larger the rate of plastic work in the steadystate, the earlier the burr initiation commences. Thus in order to obtain a thin burr for an economical deburring process, it would be desirable to nd cutting conditions which assure an optimal rate of plastic work while considering the tool wear. The use of backup material is the most effective way to minimize the burr size. It has been found that the backup material should be thick enough to allow only local deformation without bending of the backup. Also, an effective way to minimize burr size is to place the backup material in contact with the workpiece at the predened machined surface. This study has shown that the nite element method is a valuable approach to understand the burr formation mechanism in machining and the inuences of workpiece exit angles, tool rake angles, and backup materials. Within the limits of the nite element models, the numerical results provide the physical insight into the burr/breakout formation and qualitatively agree with the experimental results.

support at the University of California, Berkeley. The authors would also like to thank Dr. N. Rebelo of Hibbitt, Karlsson, and Sorenson, Inc. for his helpful comments and suggestion.

References
1 Gillespie, L. K., and Blotter, P. T., 1976, The Formation and Properties of Machining Burrs, ASME J. Eng. Ind., 98, pp. 6474. 2 Ko, S. L., and Dornfeld, D. A., 1991, A study on Burr Formation Mechanism, ASME J. Eng. Ind., 112, No. 1, pp. 6674. 3 Chern, G. L., 1993, Analysis of Burr Formation and Breakout in Metal Cutting, Ph.D. Thesis, University of California, Mechanical Engineering Department, pp. 2542. 4 Chern, G. L., and Dornfeld, D. A., 1996, Burr/Breakout Development and Experimental Verication, ASME J. Eng. Mater. Technol., 118-2, pp. 201 206. 5 Pekelharing, A. J., 1978, The Exit Failure in Interrupted Cutting, Ann. CIRP, 27, pp. 510. 6 Iwata, K., Ueda, K., and Okuda, K., 1982, Study of Mechanism of Burrs Formation in Cutting Based on Direct SEM Observation, JSPE, 48-4, pp. 510515. 7 Park, I. W., Lee, S. H., and Dornfeld, D. A., 1994, Modeling of Burr Formation Processes in Orthogonal Cutting by the Finite Element Method, ESRC Report No. 93-34, Univ. of California, Berkeley, Dec. 8 Park, I. W. and Dornfeld, D. A., 1995, A Study of Burr Formation Processes Using the Finite Element Method Part I, ESRC Report No. 95-32, Univ. of California, Berkeley, Sept., 1995. 9 Hibbitt, Karlsson, and Sorenson, Inc., 1988 ABAQUS/Explicit Users Manuals, Version 5.3, Providence, RI. 10 Hills D. A., and Novell D., 1994, Mechanics of Fretting Fatigue, Kluwer Academic Publisher, London. 11 Barsom, J. M., and Rolfe, S. T., 1987, Fracture & Fatigue Control in Structures, Prentice-Hall, Englewood Cliffs, NJ. 12 Rowe, G. W., and Spick, P. T., 1967, A New Approach to Determination of the Shear Plane Angle in Machining, ASME J. Eng. Ind., 89, pp. 530538. 13 Wright, P. K., 1982, Predicting the Shear Plane Angle in Machining from Workmaterial Strain-Hardening Characteristics, ASME J. Eng. Ind., 104, pp. 285292. 14 Gillespie, L. K., 1975, Hand Deburring Precision Miniature Parts, Precis. Eng., 1, No. 4, pp. 189198. 15 Gillespie, L. K., 1975, Burrs produced by Drilling, Bendix Corporation, Unclassied Topical Report BDX-613-1248, Dec., 1975.

Acknowledgments
The authors would like to thank the members of the Consortium on Deburring and Edge Finishing CODEF for their nancial

Journal of Engineering Materials and Technology

APRIL 2000, Vol. 122 237

K. T. Kim
Professor, Assoc. Mem. ASME e-mail: korean@postech.ac.kr

Densification Behavior of Ceramic Powder Under Cold Compaction

Densication behavior of ceramic powder under cold compaction was investigated. Experimental data were obtained for zirconia powder under triaxial compression with various loading conditions. For densication of ceramic powder during cold compaction, a novel hyperbolic cap model was proposed from the iso-density curves based on experimental data of zirconia powder under triaxial compression. The proposed model was implemented into a nite element program (ABAQUS) to study densication behavior of zirconia powder under die compaction. The modied Drucker Prager/cap model was also employed to compare with experimental data and the nite element results from the proposed model in the present work. By including the effect of friction between the powder and die wall, density distributions of a zirconia compact were measured and compared with nite element results under die compaction. S0094-42890000102-X Keywords: Cap Model, Ceramic, Densication, Die Compaction, Drucker-Prager/Cap Model, Finite Element Analysis, Triaxial Compression

S. W. Choi H. Park
Department of Mechanical Engineering, Pohang University of Science and Technology, Pohang 790-784, Korea

Introduction
Ceramics have become increasingly important in modern industry because of their good mechanical and physical properties 1. Ceramic parts are generally produced by cold die compaction with subsequent sintering and nishing or by hot pressing or hot isostatic pressing and nishing. Among these procedures, sintering has been investigated theoretically and experimentally by many researchers. The study on densication behavior of ceramic powder under cold die compaction prior to sintering, however, has been dependent mainly on experimental trial-and-error method. The friction between the powder and die wall during die compaction typically causes residual stress and inhomogeneous densication. Inhomogeneous density distributions in a powder compact lead to nonuniform shrinkage or distortion during sintering process making it difcult to control the shape of nal ceramic parts. Residual stress causes cracks in a powder compact during the ejection of the compact from the die or during the sintering process and thus affects the mechanical properties of the compact. The density that can be achieved under cold die compaction of ceramic powder is fairly low compared with that of metal powder compacts, and the ceramic compact then undergoes a signicant volume change during sintering. Therefore, the effects of inhomogeneous density distributions and residual stress on ceramic powder compacts are more serious than those on metal powder compacts. Process simulations by using a nite element analysis may be useful for optimizing the mold design, minimizing the required compaction pressure and density gradients, and controlling shapes of nal ceramic parts. The numerical modeling of the powder compaction processes requires appropriate constitutive models that can describe densication behaviors of ceramic powder. The densication behaviors of ceramic powder may be studied by investigating interactions between powders, frictional behavior, and so on after characterizing the powder by size, hardness, and shape 2. An alternative approach is to treat powders as a homogeneous continuum. The numerical analysis from constitutive models based on continuum mechanics has been widely used in soil mechanics to obtain stress distributions under the complex loading conditions 3. So far, the constitutive models for densication of ceramic
Contributed by the Materials Division for publication in the JOURNAL OF ENGINEERING MATERIALS AND TECHNOLOGY. Manuscript received by the Materials Division March 12, 1998; revised manuscript received September 13, 1998. Associate Technical Editor: G. Ravichandran.

powder have been generally adopted from those of soil mechanics and powder metallurgy. To study densication behaviors of ceramic powder, Strijbos et al. 4 used the double hardening model for sand 5 and Schwartz and Weinstein 6 and Broese 7 used the MohrCoulomb model. Shima and Mimura 8 adapted a model from experimental data of iron powder and copper powder to ceramic powder. Ceramic powder is densied by rearrangements including sliding and rolling of powders 9. Ceramic powder also shows a tendency to be in a state of agglomerates caused by van der Waals attraction between them 10. Thus, it is essential to obtain the constitutive model from experiments to investigate densication behaviors of ceramic powders. From experimental data of zirconia powder under triaxial compression, Bortzmeyer 11 proposed a roughly linear cap model and compared experimental data with nite element calculations from the proposed model. Recently, Aydin et al. 12 investigated densication behavior of alumina powder under die compaction by comparing nite element calculations from the modied DruckerPrager/cap model with experimental data. The present paper reports on densication behaviors of zirconia powder under cold compaction. To analyze densication behavior of ceramic powder under cold compaction, a novel hyperbolic cap model was proposed based on experimental data of zirconia powder under triaxial compression. The proposed model was implemented into a nite element program ABAQUS to compare with experimental data of zirconia powder under die compaction. The modied DruckerPrager/cap model was also employed to compare with the proposed model and experimental data in the present work. By including the effect of friction between the powder and die wall, nite element calculations were compared with experimental data for density distribution of a zirconia powder compact under die compaction.

Analysis
The strain rate in a ceramic powder compact during cold compaction may be decomposed into an elastic part and an inelastic part 1315. Thus,
i j iel iin j j iel j iin j

(1)

where and are elastic and inelastic strain rate tensors, respectively. By using the generalized Hookes law in the elastic part, the stress tensor can be written Transactions of the ASME

238 Vol. 122, APRIL 2000

Copyright 2000 by ASME

 el i j Diel jkl kl

(2)

where Diel jkl is a fourth-order elastic modulus tensor. The inelastic strain rate can also be written iin j

tion of zirconia powder. Table 2 shows mechanical properties of zirconia powder. Triaxial Compression. Ideal triaxial compression test should permit independent control of the three principal stresses ( 1 , 2 , 3 ) or strains ( 1 , 2 , 3 ). However, it was not feasible to fabricate such an apparatus. Thus, a standard triaxial compression apparatus that applies an axial load to the compact under constant conning pressure was used in this work. Figure 1 shows a schematic drawing of the triaxial compression apparatus used in this work. Samples for triaxial compression test were produced as follows: 6.4 g of zirconia powder was poured into a die, then the ceramic preforms were produced by cold die compaction. Samples have the average initial relative density of D o 0.31. The sample is 13.1 mm in diameter and 25.3 mm in height with the aspect ratio about 2 to reduce the friction between the specimen and platens 20. A rubber mold of 0.7 mm in thickness was used to prevent the penetration of water into a sample during triaxial test. Teon sheets were also placed between the sample and platens to reduce the friction. Then, the sample assembly was placed in the triaxial compression apparatus after vacuum sealed. The conning pressure of 20200 MPa and the axial force of 0250 kN were ap-

f i j

(3)

where f is the yield function for ceramic powder under cold com is a positive scalar. From the mass conservation, the paction and densication rate can be written
in D D kk

(4)

where D is the relative density of a powder compact. Considering the relative density as a hardening parameter 8,9, the yield function for densication of ceramic powder can be written f i j , D 0. (5)

Assuming that ceramic powder is isotropic and neglecting the third invariant of stress tensor 8, Eq. 5 can be written f p,q,D 0 (6) where p and q, respectively, are the rst and second invariant of stress tensor. When the direction 1 is regarded as the axial direction and the directions 2, 3 are regarded as the lateral directions, respectively, then 2 3 and 2 3 in the standard triaxial compression test. Hence, the hydrostatic pressure p in Eq. 6 can be written
1 p 3 1 1 2 3 3

1 2 2 .

(7)

Table 1 Physical properties and chemical composition of HSY-3.0 zirconia powder Specic surface area m3g1 Average particle size m Chemical composition wt% ZrO2 Y2O3 CaO Na2O Loss of ignition 6.4 0.53 94.06 5.41 0.02 0.01 0.13

The effective stress q in Eq. 6 can also be written: q 2 ij ij 1/2 1 2

3

(8)

where ij is the deviatoric components of stress, i.e.,

1 ij i j 3 kk i j .

(9)

Finite Element Analysis. The constitutive equations 1 9 were implemented into the user subroutine UMAT of ABAQUS 16 to analyze densication behavior of ceramic powders under cold compaction. The numerical scheme used in this paper can be found elsewhere 1719. Finite element calculations were also obtained by using the modied DruckerPrager/cap model in the constitutive library provided in ABAQUS 16. For the modied DruckerPrager model, we determined the parameters in the model from experimental data of zirconia powder under triaxial test, for instance, 0.03, 54.3, d 1.5 MPa, and R 0.835. The denitions of these parameters can be found in Fig. 4.4.4-1 in ABAQUS 16. To study densication behavior of ceramic powder under die compaction by single action pressing, we used 200 four-node axisymmetric elements in the nite element analysis. The axisymmetry for the y axis allowed a half model of the powder compact. The die wall was assumed to be a rigid surface. The relative density of the powder compact in the nite element analysis was obtained from the volume average of relative density at each element. Thus, D avg
m j1D jV j m j1V j

Table 2 Mechanical powder 32

properties

of

HSY-3.0

zirconia

Theoretical density gcm3 Youngs modulus GPa Poissons ratio

6.08 206 0.31

(10)

where m is the number of total element and D j and V j , respectively, are relative density and the volume of the j th element.

Experiments
Zirconia powder HSY-3.0, Daiichi-Kigenso Kagaku Kogyo Co. Ltd., Japan stabilized by 3 mol percent Y2O3 was used in this work. Table 1 shows physical properties and chemical composiJournal of Engineering Materials and Technology
Fig. 1 A schematic drawing of the triaxial compression apparatus.

APRIL 2000, Vol. 122 239

plied to the sample simultaneously. The stress path during triaxial compression test was a straight line with the slope q 3 p in the p, q plane. The frictional force caused by the O-ring between the pressure chamber and the moving ram to prevent the leakage of water varied with the conning pressure. Thus, the frictional force was considered for calculating the axial force on the sample. The ram speed was set at 1.2 mm mn1 during the test. Die Compaction. The cold die compaction of zirconia powder was carried out by single action pressing. 4 g of zirconia powder was poured into the die, then the compaction pressure of 10350 MPa was applied to the powder. The relative density of the powder compact was measured by Archimedes method. The sample was coated by alkyd lacquer to prevent from water penetration. The mass and volume of the coating material were compensated in calculating the density of the sample. Density Distributions. The density distribution in a ceramic compact may be obtained indirectly from the relationship between hardness and relative density. Abe et al. 21 obtained density distribution in a ceramic compact by using Vickers hardness. Rajab and Coleman 22 also used Vickers hardness to show the density distribution in an iron powder compact with a complicated shape. The samples for hardness test were prepared as follows: A ceramic compact was sintered for 40 min at 1000C in a vacuum furnace without changing its relative density. The sample was cut vertically by a diamond wheel. Then, the sample was coated by alkyd lacquer 0.005 mm thick, because the marks by a diamond indenter on the cut plane was very obscure. The sample has grid lines with 1 mm distance along the radial direction. A total of 11 points was measured for a sample by applying 25 g force on the sample for 15 s. The average value of measured hardness was used to obtain the relationship between hardness and relative density. To obtain the density distribution in a ceramic compact under cold die compaction, a total of 84 points was measured in each sample. The cut sample has grid lines with 0.5 mm distance near the edge and 1 mm distance in the central portion of the sample.

Fig. 2 Variation of Youngs modulus with relative density for a zirconia powder compact obtained from triaxial compression test

Fig. 3 Variation of relative density with hydrostatic stress for zirconia powder at various conning pressures P c

Results and Discussion

Triaxial Compression. The tensile direction was regarded as the positive direction in this paper. Figure 2 shows the variation of Youngs modulus with relative density for zirconia powder compacts. Youngs modulus of the powder compact was obtained from measuring the variation of the axial strain with the axial stress in innitesimal unloading during triaxial compression test, because unsintered ceramic compacts show very low uniaxial compressive strengths. The solid curve in Fig. 2 was obtained from the curve tting 23. Thus, E E 0 exp b c 2 where E 0 206 GPa, b 12.6, c 6.99. In Eq. 11, E 0 is Youngs modulus of the matrix material, b and c are constants, and ( 1 D ) is the porosity of the powder compact. The Poissons ratio for a porous ceramic compact was obtained from the literature 24. Thus, (11) pressure than high conning pressure. The variations of relative density under various conning pressures were found to be almost linear as shown in Fig. 3. During triaxial compression of ceramic powder, the critical state was observed as shown in Fig. 4a. In the critical state, the compact deforms continuously without changing its volume 25. The critical state was observed at several conning pressures in Fig. 3. The critical state in this experiment can be represented by the following equation in the p, q plane see: Fig. 4b: qM p (13) where M 1.41. In the ideal critical state, the volumetric strain must be zero as in Fig. 4a. However, experimental data show that the relative density increased slightly while the stress remains almost constant. This may be due to the friction between the powder compact and platens although teon sheets were placed between them. Figure 5 shows zirconia compacts a before deformation, b after isostatic compression under 200 MPa, and c after triaxial compression under conning pressure of 60 MPa with p 117 MPa and q 171 MPa. The friction between the specimen and teon sheets prevented the ideal shrinkage of the sample under isostatic pressing see: Fig. 5b and also caused barreling see: Fig. 5c. Figure 6 shows the variation of the effective stress q with axial strain 1 and lateral strain 2 for zirconia compacts at various conning pressures P c during triaxial compression. Here the average lateral strain was used because of barreling as shown in Fig. Transactions of the ASME

1 4 0 3 7 0 4 1 2 3 0

(12)

where 0 is the Poissons ratio of the matrix material. Figure 3 shows the variation of relative density with hydrostatic pressure for zirconia compacts at various conning pressures under triaxial compression. It is observed in Fig. 3 that the densication is enhanced by adding shear stress. The effect of shear stress on densication is more pronounced under low conning 240 Vol. 122, APRIL 2000

Fig. 6 Variation of the effective stress q with axial and lateral strains for zirconia powder compacts at various conning pressures P c

with A D 1.92 10 4 D 7.008, B D 3.196 105 D 7.915 (15)

Fig. 4 a Description of the critical state in the stress and strain plane and b stress path in the p , q plane during triaxial compression

5c. As conning pressure increases, the effective stress increases rapidly as strains increase. As strains further increase, the increase in the effective stress slows down and then plateaus eventually. Figure 7 shows the iso-density curves of zirconia compacts in the range of relative density D 0.42 0.51. Data points were obtained from the relationship between relative density and hydrostatic stress at various conning pressures under triaxial compression. The iso-density curves solid curves at various relative densities in Fig. 7 were obtained from the least-square curve t of a hyperbolic function. Thus, q cosh A D p B D (14)

The dashed curves were obtained from the modied Drucker Prager/cap model with parameters 0.03, 54.3, d 1.5 MPa, and R 0.835 in ABAQUS 16. Bortzmeyer 9 showed that relative density of ceramic powder compacts behaves as the hardening parameter. The iso-density curve in Eq. 14 can be regarded as the yield function for densication of zirconia powder under cold compaction. The yield function 14 has a special form of the cap model, which is widely used in soil mechanics and powder metallurgy 26,27. In particular, Eq. 14 represents hyperbolic caps instead of roughly linear caps as observed by Bortzmeyer 11 for zirconia powder under triaxial compression. The iso-density curves in Fig. 7 were obtained for zirconia powder with relative density greater than D 0.42. Thus, the extrapolation of Eq. 14 is necessary for ceramic compacts with initial relative density lower than those in Fig. 7. To show the validity of the extrapolation for ceramic compacts with low relative density, we compared the extrapolation of Eq. 14 with experimental data under hydrostatic compression. Figure 8 shows comparisons between the iso-density curves extrapolated from Eq. 14 with Eq. 15 and from the modied DruckerPrager/cap model for experimental data of zirconia compacts under hydrostatic pressing. The extrapolations of both Eq.

Fig. 5 Zirconia powder compacts a before deformation, b after isostatic compression under 200 MPa, and c after triaxial compression under conning pressure of 60 MPa.

Fig. 7 Comparison between the proposed hyperbolic cap model solid and the modied DruckerPragercap model dashed for iso-density curves of zirconia powder compacts under triaxial compression

Journal of Engineering Materials and Technology

APRIL 2000, Vol. 122 241

Fig. 8 Comparison between the extrapolated curves from the proposed hyperbolic cap model solid and the modied DruckerPragercap model dashed for experimental data of zirconia powder under cold isostatic pressing

Fig. 10 Comparison between nite element calculations from the proposed hyperbolic cap model solid and the modied DruckerPragercap model dashed for the variation of relative density with hydrostatic pressure for zirconia powder during cold isostatic compression

14 and the modied DruckerPrager/cap model agree reasonably well with experimental data for zirconia compacts with low relative density under hydrostatic compression. Cold Isostatic Pressing. The yield function in Fig. 7 show vertices in the p axis. Plastic strain rate under hydrostatic loading may have a deviatoric component as well as a volumetric component, since the direction of plastic strain rate is not unique at the vertex. There also may be discontinuity when the stress moves from hydrostatic pressure to triaxial stress. To solve this problem, the vertex on the yield surface was replaced by a small spherical cap see Fig. 9 as suggested by Govindarajan and Aravas 28. Figure 10 shows comparisons between experimental data and nite element calculations for the variation of relative density with hydrostatic pressure for zirconia compacts with initial relative density D 0 0.31 during cold isostatic pressing. The solid curve was obtained from nite element calculations by using the proposed cap model, i.e., Eq. 14 with Eqs. 11 and 12. The dashed curve was obtained from nite element calculations by using the modied DruckerPrager/cap model. The proposed model agrees well with experimental data for zirconia compacts under hydrostatic compression, while the modied Drucker Prager/cap model slightly underestimates experimental data. Cold Die Compaction. Many researchers investigated densication behaviors of ceramic powders under cold die compaction. For instance, Thompson 29 proposed a theoretical model based on the model by Janssen 30 and experimental data by Unckel 31. Thus,

z r , z B

r2 4 r2 z C 1 2 2 exp R R R

(16)

where z is the axial stress and r and z are the radial and axial coordinates, respectively. Also, R is the radius of the sample, and are material constants, and B and C are constants. In Eq. 16, the stress distributions in the radial and axial direction have the parabolic and exponential forms, respectively. At the center line ( r 0), the axial stress z is constant along the z axis. To compare nite element calculations by using the proposed cap model, Eq. 14 with experimental data of a zirconia compact under die compaction, we determined the friction coefcient between the powder and die walls during die compaction by following the approach of Kwon et al. 19. Thus, the friction coefcient was obtained indirectly from the relationship between the compaction pressure and ejection pressure by comparing experimental data with nite element results. Figure 11 shows comparisons between experimental data and calculated results for the variation of ejection stress with axial stress of zirconia powder during die compaction. The ejection stress was measured by pushing out the compact out of the die. The data points in Fig. 11 were obtained for zirconia powder under single action die pressing. The dotted curve, solid curve, and dashdotted curve were obtained from nite element calculations by using Eq. 14 with various friction coefcients. Here, the solid curve with the friction coefcient 0.2 shows a good

Fig. 9 The modied yield surface in the vicinity of hydrostatic stress axis

Fig. 11 Comparison between experimental data and calculated results with various friction coefcients for the variation of ejection stress with axial stress of zirconia powder under die compaction

242 Vol. 122, APRIL 2000

Transactions of the ASME

Fig. 12 Comparison between nite element calculations from the proposed hyperbolic cap model solid and the modied DruckerPragercap model dashed for the variation of relative density with axial stress of zirconia powder under die compaction

agreement with experimental data for zirconia powder in Fig. 11. The friction coefcients were found by a trial-and-error method. Figure 12 shows comparisons between experimental data and nite element calculations for the variation of relative density with axial stress of zirconia powder during die compaction. The friction coefcient 0.2 was used for zirconia powder. The solid curve was obtained from nite element calculations by using Eq. 14 and the dashed curve from the modied DruckerPrager/ cap model. The proposed model slightly overestimates experimental data, while the modied DruckerPrager/cap model underestimates. Figure 13 shows the variation of relative density with Vickers hardness ( H v ) for zirconia powder compacts. The relationship of relative density and Vickers hardness can be represented by the following equation: D 0.31349 0.01781 H v (17)

Fig. 14 Comparison between a experimental data and nite element calculations from b the proposed hyperbolic cap model, and c the modied DruckerPragercap model for relative density contour plots of a zirconia powder compact ejected from a die after compacted by single action pressing under axial stress of 100 MPa

Figure 14 shows a comparison between experimental data and nite element calculations for relative density contour plots of a zirconia powder compact ejected from a die after compacted by single action pressing under axial stress of 100 MPa. It is understood only right halves of the compact are shown in Figs. 14a, 14b, and 14c. Figure 14a was obtained from Eq. 17 with Vickers hardness. Figures 14b and c, respectively, were obtained from nite element calculations by using Eq. 14 and the modied DruckerPrager/cap model with 0.2. In Fig. 14a, relative density is the highest at the corner of contact surface between the upper punch and the die wall and the lowest at the

corner of contact surface between the lower punch and the die wall. A similar trend of the density distribution was also observed for a stainless steel powder compact by single action pressing 19. Thompson 29 assumed that the relative density is constant along the center line of the sample during die compaction, however, it varies as shown in Fig. 14a. The nite element results in Figs. 14b and 14c show some of the same trends with experimental data in Fig. 14a. The overall density distribution in Fig. 14b is slightly overestimated experimental data in Fig. 14a, while that in Fig. 14c is underestimated experimental data. Figure 15 shows nite element calculations for Figs. 15a normal stress and 15b shear stress distributions on a zirconia compact under axial stress of 100 MPa by single action pressing. Figure 15 was obtained from nite element calculations by using Eq. 14 with 0.2. In Fig. 15a, the normal stress was represented by the ratio to the axial stress of 100 MPa. The normal stress on the die wall was the highest in the top corner and the lowest in the bottom corner due to the effect of friction. The shear

Fig. 13 Variation of relative density with Vickers hardness H v for zirconia powder compacts

Fig. 15 Finite element calculations from the proposed model for a normal stress and b shear stress distributions on a zirconia powder compact under axial stress of 100 MPa by single action pressing

Journal of Engineering Materials and Technology

APRIL 2000, Vol. 122 243

References
1 Klocke, F., 1997, Modern Approaches for The Production of Ceramic Components, J. Eur. Ceram. Soc., 17, pp. 457465. 2 Thornton C., 1987, Computer Simulated Experiments on Particulate Materials, Tribology in Particulate Technology, B. J. Briscoe and M. J. Adams, eds., Adam Hilger Publishers, Bristol, pp. 292302. 3 Lewis, R. W., and Schreer, B. A., 1987, The Finite Element Method in The Deformation and Consolidation of Porous Media, Wiley, New York. 4 Strijbos, S., Broese, A. V. G., and Vermeer, P. A., 1979, Recent Progress in Understanding Die Compaction of Powders, J. Am. Ceram. Soc., 62, pp. 5759. 5 Vermeer P. A., 1977, A Double Hardening Model for Sand, Delft Prog. Rep. Civil Eng., 2, pp. 303320. 6 Schwartz, E. G., and Weinstein, A. S., 1964, Model for Compaction of Ceramic Powders, J. Am. Ceram. Soc., 48, No. 7, pp. 346350. 7 Broese, A. V. G., 1978, Pressing of Ceramic Powders: A Review of Recent Work, Powder Metall. Int., 10, No. 4, pp. 206211. 8 Shima, S., and Mimura, K., 1986, Densication Behavior of Ceramic Powder, Int. J. Mech. Sci., 28, No. 1, pp. 5359. 9 Bortzmeyer, D., 1990, Compaction des Poudres Ceramiques, Doctoral thesis, Ecole Nationale Superieure des Mines de Paris. 10 Yen, K. Z. Y., and Chaki, T. K., 1992, A Dynamic Simulation of Particle Rearrangement in Powder Packings with Realistic Interactions, J. Appl. Phys., 71, No. 7, pp. 31643173. 11 Bortzmeyer, D., 1992, Modeling Ceramic Powder Compaction, Powder Technol., 70, No. 2, pp. 131139. rk, K. Y., 1996, The Internal Form of 12 Aydin, I., Briscoe, B. J., and Sanlitu Compacted Ceramic Components: A Comparison of a Finite Element Modelling with Experiment, Powder Technol., 89, No. 3, pp. 39254. 13 Broese, A. V. G., and Lissenburg, R. C. D., 1983, Inhomogeneous Density in Die Compaction: Experiments and Finite-Element Calculations, J. Am. Ceram. Soc., 66, No. 9, pp. c156c158. 14 Hehenberger, M., Samuelson, P., Alm, O., Nilsson, L., and Olofsson, T., 1982, Experimental and Theoretical Studies of Powder Compaction, IUTAM Conference on Deformation and Failure of Granular Materials, Delft, pp. 381390. 15 Strijbos, S., and Vermeer, P. A., 1978, Stress and Density Distribution in The Compaction of Powders, Processing of Crystalline Ceramics, H. Palmour, R. F. Davis, and T. M. Hare, eds., Materials Science Research, New York, pp. 113123. 16 Hibbitt, Karlsson, and Sorensen, 1996, ABAQUS Users I and I Manual. 17 Aravas, N., 1987, On The Numerical Integration of A Class of PressureDependent Plasticity Models, Int. J. Numer. Methods Eng., 24, pp. 1395 1416. 18 Govindarajan, R. M., 1992, Deformation Processing of Porous Metals, Doctoral thesis, University of PA. 19 Kwon, Y. S., Lee, H. T., and Kim, K. T., 1997, Analysis for Cold Die Compaction of Stainless-Steel Powder, ASME J. Eng. Mater. Technol., 119, No. 4, pp. 366373. 20 Meerman, W. C. P. M., and Knaapen, A. C., 1979, A High Pressure Triaxial Testing Cell, Powder Technol., 22, pp. 271278. 21 Abe, O., Iwai, S., Kanzaki, S., Ohashi, M., and Tabata, H., 1986, Inuence of Size and Shape on Homogeneity of Powder Compacts Formed by Cold Isostatic Pressing, Yogyo Kyokaishi, 94, No. 10, pp. 10921098. 22 Rajab, M., and Coleman, D. S., 1985, Density Distribution in Complex Shaped Parts Made from Iron Powders, Powder Metall., 28, No. 4, pp. 207 216. 23 Wang, J. C., 1984, Youngs Modulus of Porous Materials, J. Mater. Sci., 19, pp. 801814. 24 Ramakrishnan, N., and Arunachalam, V. S., 1993, Effective Elastic Moduli of Porous Ceramic Materials, J. Am. Ceram. Soc., 76, No. 11, pp. 2745 2752. 25 Schoeld, A., and Wroth, P., 1968, Critical State Soil Mechanics, McGrawHill, London. 26 DiMaggio, F. L., and Sandler, I. S., 1971, Material Model for Granular Soils, J. Eng. Mech. Div., Proc. ASCE, 97, EM3, pp. 935950. 27 Brekelmans, W. A. M., 1990, Comparison of Two Drucker-Prager-type Constitutive Models for Granular Material Behavior in Uniaxial Compression, Powder Technol., 62, pp. 2126. 28 Govindarajan, R. M., and Aravas, N., 1994, Deformation Processing of Metal Powders: Part ICold Isostatic Pressing, Int. J. Mech. Sci., 36, No. 4, pp. 343357. 29 Thompson, R. A., 1981, Mechanics of Powder Pressing: I, Model for Powder Densication, Ceram. Bull., 60, No. 2, pp. 237243. 30 Ozkan, N., and Briscoe, B. J., 1997, Characterization of Die-Pressed Green Compacts, J. Eur. Ceram. Soc., 17, pp. 697711. 31 Unckel, H., 1945, Experimental Results of Pressing Metal Powders Arch. Eisenhuettenwes., 18, No. 7, pp. 161167. 32 1989, Ceramic Sources, Vol. 4, American Ceramics Society.

Fig. 16 Finite element calculations from the proposed model for distributions of a residual Mises stress and b residual hydrostatic stress of a zirconia powder compact ejected from a die after compacted under 100 MPa

stress which affects the life of the die by wear is zero at the bottom corner and the highest at the top corner. The shear stress is zero at the center line due to the axisymmetric condition. Figure 16 shows nite element calculations for distributions of Figs. 16a residual Mises stress and 16b residual hydrostatic stress of a zirconia powder compact ejected from a die after pressed under 100 MPa. Figure 16 was obtained from nite element calculations by using Eq. 14 with 0.2. In Fig. 16a, the highest residual Mises stress is observed at the upper corner of the compact. In Fig. 16b, tensile residual hydrostatic stress is observed at the top and bottom of the compact, although it is small. Almost no residual hydrostatic stress was observed at the central part of the compact. It is also observed that the residual stress in the compact are much smaller than the compaction pressure.

Conclusions
Densication behaviors of zirconia powder during cold compaction were investigated. Based on experimental data of zirconia powder under triaxial compression, a hyperbolic cap model was proposed. The proposed model was implemented into a nite element program to compare with experimental data of zirconia powder under die compaction. The modied DruckerPrager/cap model was also employed to compare with the proposed model. From the relationship between the compaction and ejection pressures, the friction coefcient 0.2 was obtained for zirconia powder between the powder and die walls. Finite element calculations from the proposed model agreed reasonably well with experimental data of zirconia powder under cold compaction for both global densication and local density distribution. Finite element calculations from the modied DruckerPrager model showed some of the same trends with experimental data, however, underestimated.

Acknowledgments
This work was nancially supported by a grant from the Korean Science and Engineering Foundation KOSEF under grant no. 971-1007-042-2. We are grateful for this support.

244 Vol. 122, APRIL 2000

Transactions of the ASME