Multiaxial Fatigue Damage Models D. Socie Department of Mechanical and Industrial Engineering, University of Illinois at Urbana-Champaign, Urbana,IL 61801 Two multiaxial fatigue damage models are proposed: a shear strain model for failures that are primarily mode II crack growth and a tensile strain model for failures that are primarily mode I crack growth. The failure mode is shown to be dependent on material, strain range and hydrostatic stress state. Tests to support these models were conducted with Inconel 718, SAE 1045, and AISI Type 304 stainless steel tubular specimens in strain control. Both proportional and nonproportional loading histories were considered. It is shown that the additional cyclic hardening that accompanies out of phase loading cannot be neglected in the fatigue damage model. Introduction A large number of multiaxial fatigue damage models have been proposed. Stress, strain, and energy have all been used to correlate test data. Despite much research in multiaxial fatigue, no single theory has been able to correlate the data for a wide variety of materials and loading conditions. Early investigators (Gaugh, 1933; Stulen, 1954; Sines and Waisman, 1959) were interested in long life problems and proposed shear stress based theories for ductile materials and principle stress based theories for brittle materials. In the low cycle fatigue region, equivalent strain (Taira et al. 1967; Pascoe and DeVilliers, 1967; Yokobori et al. 1965), plastic work (Garud, 1979), plastic strain energy, (Ellyin and Valaire, 1982) and critical plane approaches (Brown and Miller, 1973; Lohr and Ellison, 1980) have been used to correlate the data. Since plastic shear strains play an important role in crack nucleation and early growth, it is not surprising that all of these approaches are essentially shear stress and strain theories with the exception of plastic work. This approach includes a contribution from the tensile stresses and strains. Material dependency is introduced only in the constants required to fit each set of data. By adjusting the constants some shear strain theories can be made to be numerically equivalent to a maximum principal strain theory. Observations of crack formation and early growth for a number of materials by Bannantine and Socie (1985) show two types of behavior. Materials such as AISI Type 304 stainless steel fail by the growth of a mode I tensile crack or a mode II shear crack depending on the stress state and cyclic strain amplitude. Mode I failures were observed for tensile loading at all strain amplitudes. In torsion, mode II, shear failures were observed at high strain amplitude and mode I failures at low strain amplitudes. Other ductile materials such as Inconel 718 fail by the growth of a mode II shear crack in both tension and torsion loading. A multiaxial fatigue damage parameter has been developed for Inconel 718 at room temperature. Extensive observations have shown that the majority of the fatigue life in tubular specimens of this material is spent in growing cracks from about 0.01 mm to 1.0 mm on planes of maximum shear strain amplitude. Stresses and strains perpendicular to the shear crack influence the rate of crack growth. A schematic illustration of the damage model is shown in Fig. 1. The crack surfaces are irregularly shaped as the crack advances through the individual grains. Mechanical interlocking occurs during shear loading and high frictional forces are developed. This reduces the stresses and strains at the crack tip and is responsible for a lower growth rate. Stresses and strains perpendicular to the shear crack can be developed for loading situations other than torsion. This opens the crack surfaces and reduces or eliminates the friction forces. Higher stresses and strains will be developed at the crack tip and higher growth rates are observed. Fractographic evidence has been presented by Socie et al. (1985a), where the individual slip bands are clearly visible on the fracture surfaces of the tension test while the torsion test has a fracture surface that has been burnished by the rubbing of the two crack surfaces. A three parameter description of this damage process has been proposed by Socie et al. (1985b). y + e„+^ = y}(2Ny + T}/G{2N)b Torsion (1) y- -y CT,€ Tension y-»- -*~Y Contributed by the Materials Division for publication in the JOURNAL OF ENGINEERING MATERIALS AND TECHNOLOGY. Manuscript received by the Or, 6 Fig. 1 Shear damage model Materials Division July 22, 1987. Journal of Engineering Materials and Technology OCTOBER 1987, Vol. 109 / 293 Copyright © 1987 by ASME Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use TORSION TENS la, 101, - - - - - - , - - - - - - , - - - - - - - , - - - - - - , Inconel 718 ;:<>I'H."l\lI'" ,,-~L~ "l".'tf.""Ufl AI~t.1lW I>W1, __• ,.1 J. • ff';:" • . .i'J.«tJ ~ 163L.;,-----~----...l...;------l.::-----...J ~ d ~ Nl.O ~ ~.US.: ~ Correlalion of Inconel 718 Test Results with Shear Strain Parameter Fig. 2 Correlation of Inconel 718 test data The right-hand side is the description of the strain life curve generated from torsion testing with the following nomenclature: 'Yj shear fatigue ductility coefficient c fatigue ductility exponent shear fatigue strength exponent b fatigue strength exponent G shear modulus 2N reversals to the formation of a 1.0 mm surface crack. 7j The terms on the left-hand side represent the loading parameters defined on the plane experiencing the largest range of cyclic shear strain and have the following definitions: r En anD E maximum shear strain amplitude tensile strain perpendicular to the maximum shear strain amplitude mean stress perpendicular to the maximum shear strain amplitude elastic modulus. Fig. 3 Table 1 Material properties, 304 stainless steel MONOTONIC TENSILE PROPERTIES E, Elastic Modulus 183 GP. ai%' 0.2% Offset Yield Strength 325 MPa au' of' E , f % RA, Ult imate Strength True Fracture Strength True Fracture Strain % Reduction in Area Strength Coefficient Strain Hardening Exponent K, Fifteen proportional and nonproportional loading histories have been used to evaluate the model. Fifty-five tests were performed on tubular specimens using these loading histories. An additional twenty tests were performed on solid axial specimens to evaluate mean stress effects in uniaxial tension. Correlation of all of the test data with the shear parameter given in equation (1) is shown in Fig. 2. The solid line represents the best fit through the torsion data that was used to establish the material constants. Tension and torsion tests were also conducted on tubular AISI Type 304 stainless steel specimens and reported by Bannantine and Socie (1985). Although both materials have considerable ductility, the stainless steel tends to crack on planes of maximum principal strain except for the high strain torsion tests. The cracking behavior of both materials is shown in Fig. 3 for both tension and torsion tests. Note the extensive shear cracking in the Inconel 718 tension and torsion tests. All of these photos were taken from tests where the elastic strains were approximately equal to the plastic strains. The fracture surfaces of the AISI Type 304 stainless steel tests were examined and showed no evidence of shear cracking in the tension tests. Shear nucleation was observed in the AISI Type 304 stainless steel torsion tests but the majority of the life is consumed in mode I tensile growth at this strain amplitude. These observations suggest that the shear parameter used to correlate the Inconel 718 data would be inappropriate for AISI Type 304 stainless steel. This paper reports results, observations and correlations for both proportional and nonproportional loading tests on AISI Type 304 stainless steel specimens and makes comparisons between these two materials and SAE 1045 steel. 294/VoI.109, OCTOBER 1987 Cracking behavior of Inconel 718 and 304 stainless steel n, 650 MPa 1400 MPa 1. 731 80% 1210 MPa 0.193 AXIAL CYCLIC PROPERTIES (R = -I) E E, Elastic Modulus 185 GPa Of' b Fatigue Strength Coefficient Fatigue Strength Exponent 1000 MP. -0.114 <f' c K' n' Fatigue Ductility Coefficient Fatigue Ductil ity Exponent Cycl ic Strength Coefficient Cyclic Strain Hardening Exponent 0.171 -0.402 TORSIONAL CYCLIC PROPERTIES (R y = -I) G, Torsional Modulus f, 1660 MP. 0.287 82.8 GPa Fatigue Strength Coefficient Fatigue Strength Exponent 709 MP. -0.121 c, Fatigue Ductility Coefficient Fatigue Ductil ity Exponent 0.413 -0.353 K', n', Cyclic Strength Coefficient Cyclic Strain Hardening Exponent T b Yf' 785 MP. 0.296 Experimental Procedure Strain controlled tension and torsion tests were performed on AISI Type 304 stainless steel tubular specimens with an internal diameter of 25 mm and a wall thickness of 3.8 mm. A complete set of material properties obtained from monotonic and strain controlled tension, and torsion fatigue tests are given in Table 1. Additional materials description is given by Transactions of the ASME Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use Biaxial Loading Paths for 3 0 4 Stainless Steel Table 2 Fatigue test data, 304 stainless steel CONTROL PARAMETERS LOADING SPEC fiE/2 EQ AH/2 STEADY STATE STRESSES (HPa) XQ to/Z aQ AT/2 H TQ HISTORY Axial 1 0 (cycles) SS-06 0.0035 SS-15 0.00-16 0 A-03 0.0100 2.0 A-09 0.0100 -1.4 0 0 A-06 0.0060 -21.9 0 0 A-01 0.0035 5.8 0 0 A-10 0.0035 -9.1 0 0 A-11 0.0035 8.4 0 0 0 0 240 -2.2 0 37,600 38,500 0 0 10,000 10,300 0 0 '- . " 1,070 A-12 0.0020 16.5 0 0 A-20 0.0020 29 0 0 0 248 SS-08 0.017 "y/Ji 0 1.2 1,370 1,167 30,700 33,530 29,000 286,400 333,100 0.008 0 191 1.9 42,000 SS-21 0.0079 0 156 0.4 7,000 32,100 SS-22 0.0080 0 157 0 5,000 33,900 SS-07 0.006 0 157 0 SS-16 0.0061 0 140 0 1,000 SS-09 0.0034 0 138 0 883,000 >1.0 X 10 SS-20 0.0034 0 125 0.6 650,000 824,200 -0.6 940,000 >1.1 x 10 0.0034 0 127 SS-02 0.0025 0.0043 2.7 109 SS-12 0.3 44,300 1 48,500 133,000 83,400 53,000 0.0025 0.0043 -3.7 101 SS-25 0.00145 0.0023 0.6 89 SS-26 0.00145 0.0023 0.8 90 SS-10 0.0035 0.0061 2.6 267 SS-13 0.0035 0.0061 -4.5 256 1.3 3,600 SS-13 0.0020 0.0035 18.6 168 - 2.1 38,600 50,000 SS-29 0.0020 0.0033 29.3 176 - 3.5 35,300 45,000 SS-03 0.0025 0.0043 0 226 0 5,110 5,110 SS-11 0.0025 0.0043 0 218 0 6,000 6,200 SS-30 0.0014 0.0024 20.6 139 1.8 82,100 89,312 SS-31 0.0014 0.0024 17.1 137 2.2 90,100 100,000 SS-01 0.0025 0.0043 32 199 -13.2 SS-04 0.0025 0.0043 35 209 -21.8 7,340 SS-32 0.0014 0.0024 47.7 114.7 -17.0 182,000 200,000 SS-33 0.0014 0.0024 44.5 115.6 -15.7 192,200 205,000 0.7 44,000 52,900 1.2 420,000 440,000 2.9 340,000 356,000 - 4.4 3,340 3,560 - P I 4,090 i 101,000 Proportional 6,080 SS-05 SS-28 Torsion A • < t 1 , 90° out of Phase Box Two Box Fig. 4 Loading histories for 304 stainless steel tests lOOOr 3,730 500 11,080 9,800 Bannantine (1986). Tests were performed with a computer controlled axial-torsion servo-hydraulic testing system that was automated for test control and data acquisition. The strains were measured with an internal extensometer that is described by Socie et al. (1985b). This technique allows the outside surface of the specimen to be monitored for crack formation and growth. Most of the axial tests were conducted on standard 6 mm diameter solid tensile specimens and are denoted in Table 2 with specimen numbers beginning with A. All tests were conducted at room temperature. Six loading histories were used for these tests and are shown in Fig. 4. Two strain levels were used for each multiaxial loading condition resulting in fatigue lives ranging from 3 x 103 to 2 x 105 cycles. Ranges of stress and strain are indicated with the A symbol and mean values of stress and strain by the subscript 0 in Table 2. The first observation of a surface crack 1.0 mm long is denoted by 7V,]0 and final failure by Nf. All of the stresses and strains given in Table 2 were computed at the midsection of the thin walled tube from the stable hysteresis loop of the measured axial load and torque, gage section deflection and angle of twist. Results and Discussion Results from the fatigue tests are given in Table 2. The fatigue test data initially was correlated in terms of the maximum principal strain amplitude. This correlation was satisfactory for all of the in-phase tests for histories A, B, and C. The nonproportional tests, however, were more damaging by about a factor of ten in fatigue life when compared on the basis of principle-strain range alone. It is now well established that in many materials, including AISI Type 304 stainless steel, out-of-phase loading produces additional cyclic hardening that is not found in simple proportional tests. The stable effective cyclic stress strain curve for both in-phase and 90 Journal of Engineering Materials and Technology 0,01 Strain Fig. 5 Effective stress strain curve for in and out of phase tests deg. out-of-phase loading is shown in Fig. 5. The stable curve is doubled during the nonproportional loading. This increase in stress must be included in the damage parameter. The simplest parameter which incorporates both cyclic strain range and maximum stress was proposed by Smith et al. (1970), and is commonly referred to as the Smith-Watson-Topper (SWT) parameter. <>? afax Ae, / 2 = a'rt'r (27V) * + c + - £ - (27V) (2) The right-hand side is a description of the uniaxial strain life curve generated from tensile testing with the following nomenclature: tensile fatigue ductility coefficient fatigue ductility exponent tensile fatigue strength exponent b fatigue strength exponent E elastic modulus 2N reversals to the formation of a 1.0 mm surface crack. c The terms on the left-hand side represent the loading parameters and have the following definitions: Ae,/2 maximum principal strain amplitude CTmax maximum stress on the maximum principal strain plane A considerable data base exists for this parameter for effects of mean stress during uniaxial loading. It can be reasoned that the larger stresses during nonproportional loading will have an effect similar to a tensile mean stress. OCTOBER 1987, Vol. 109/295 Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use The principal stresses and strains continuously rotate during a nonproportional test. The damage parameter can be interpreted in two ways. Analogous to the critical plane shear strain theory, the SWT parameter can be determined on the plane that has the greatest range of principal strain. An alternate definition could be the maximum value of the SWT parameter. In these tests, both definitions result in nearly the same damage parameter. This parameter is proportional to plastic work per cycle for in-phase fully reversed loading. It differs in concept because the plastic work approach does not consider a critical direction or plane for crack nucleation and growth. This parameter also has the capability to directly model mean stress effects. Failure cracks and critical planes for histories C, N, P, and Q are shown in Fig. 6. The cracks have a surface length of approximately 500 /-tm in these photos. The plane that intersects the surface experiencing the largest range of principal strain is /' /' /' /' a b c d Fig. 6 Failure cracks (a) proportional, (b) 90· out of phase, (e) box, (d) two box Axial Proporlionol Loading ~roo A~iol 80x Polh TO/sian shown as a dashed line. Cracks initially form and grow on this plane for all of the tests. Observations made during the tests indicated that the majority of the life is consumed in forming a 1.0 mm surface crack. Cracks continue to grow on the maximum principal strain range plane for in-phase loading. During 90 deg. out-of-phase loading the principal strain plane continuously rotates and cracks could be expected in any direction depending on the magnitude of the applied axial and torsional strain. After nucleation, cracks grow during the axial strain part of the load cycle for paths P and Q after they become large enough to generate their own stress fields. During this part of the load cycle the cracks are subjected to an axial strain cycle with a mean torsion stress. The other half of the cycle can be modeled as a torsion crack subjected to a cyclic shear stress with a mean axial stress. Thus, the cracks turn and tend to grow normal to the specimen axis. The observed cracking directions provide physical justification for using a principal strain based theory. Axial and torsional response from the four loading histories are given in Fig. 7 for the stable hysteresis loops. Note that the maximum stress and strain do not occur at the same point in time during out-of-phase loading. Axial stress was computed from the measured loads. Average shear stress was computed from the measured torque employing thin walled tube assumptions. The hoop strain required to determine the strains on the critical plane was computed from the measured axial stresses and strains. Stress strain response on the critical plane for these- tests was computed from the test data with the results given in Fig. 8. Here, the critical plane is defined as the plane experiencing the largest range of normal strain. The additional cyclic hardening is clearly evident when compared to in-phase and out-of-phase tests. The amount of hardening is nearly the same for all of the out-of-phase tests. The proportional loading test has the largest plastic strain range and has the longest life for any of the tests. This conflicts with the widely held notion that plastic strain alone is responsible for fatigue damage. Cyclic stress plays an important role in the damage process and cannot be neglected. The SWT parameter was computed from the hysteresis loops shown in Fig. 8. The maximum stress in the loop was used in the analysis even though it did not occur at the maximum strain. Correlation of the test data is given in Fig. 9. Correlation is good for all tests. This parameter does not require the determination of plastic strains so that visco elastic and plastic constitutive equations could be used to evaluate the damage parameter. In plastic strain approaches, small errors in the stress calculation can result in large errors in estimated life. Axial 90" aul of Phose Loading ~"'o Axial Two Box Poth Torsion ~"'" Torsion Fig. 7 Axial and torsion stress response (a) proportional, (b) 90 out of phase, (e) box, (d) two box 29B/Vol. 109, OCTOBER 1987 Transactions of the ASME Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use : ' 1 1 1 1 1 U 1 i 11 m i Inconel 718 . - 1 1 THI| "1 o % o ; : o <& Critical Plane, Proportional Loading CO °%0 o o - Cfilical Plane, 90° oul of Phase Loading o oc? o oo £* 0 b 10 qp o o <& o _ : •1 1 1 Mill 10= 10° Ni Fig. 10 J-600 Cfilical Plane, Box Palh -~ j_600 „ . , _, _ Critical Plane, Two Box Path Fig. 8 Stress strain response on the critical plane, (a) proportional, (b) 90 out of phase, (c) box, (d) two box Fig. 9 Correlation of 304 stainless steel test data with SWT parameter The details of constitutive modeling are beyond the scope of this paper but it is worth noting that the correlations given in Fig. 9 for the experimentally measured stress and strain and analytical estimates of stress and strain are nearly identical. The constitutive model described by McDowell and Socie (1985) is based on Mroz kinematic hardening and an out-ofphase isotropic hardening parameter proposed by McDowell (1983). The SWT parameter would not be expected to correlate the test data from the Inconel 718 tests because of the difference in cracking (Fig. 3). This is indeed the case as shown in Fig. 10 where the data given in Fig. 2 has been replotted in terms of the SWT parameter. These results show the importance of choosing a damage parameter that is based on the physical mechanism for the material and loading of interest. It has been reported that in-phase loading is more damaging at low strain amplitudes and out-of-phase loading is more damaging at higher strain amplitudes (Garud, 1979). In the case of low strains this is misleading because the comparison is based on the same axial and torsion strains rather than maximum shear strain ranges. If the comparison is made on the basis of maximum strain ranges rather than applied tensile and torsion strain ranges out-of-phase loading is equally or more damaging at low strain levels. For higher strains, out-of-phase loading is expected to be more damaging because of the inJournal of Engineering Materials and Technology Correlation of Inconel 718 test data with SWT parameter creased cyclic hardening. At small plastic strain levels the additional hardening will also be small and no difference in fatigue lives should be observed. For larger plastic strains, the hardening can be as much as a factor of two and a considerable effect on the fatigue life should be observed. Materials that show small amounts of additional cyclic hardening for out-of-phase loading would be expected to show small differences in fatigue life between in-phase and out-ofphase loading. Fatima (1986) tested 1045 Steel in 90 deg. outof-phase loading and found that the additional cyclic hardening was only 10 to 15 percent and the fatigue lives were reduced by about 1/2 when compared to the in-phase tests. In this investigation the stresses doubled during out-of-phase loading and the fatigue life is reduced by a factor of 10. The shear strain based model presented in equation (1) predicts no effect from the additional cyclic hardening. The effects could be included by combining the cyclic normal strain term and the mean stress term into a maximum stress term (Fatima and Socie, 1986). Materials tested to date in our laboratory that have shear dominated failures show small amounts of out-of-phase hardening so the importance of an out-of-phase hardening term in the shear model cannot be confirmed. Out-of-phase loading will always be more damaging in the shear model because the normal strain term is increased during this type of loading. Test data reported here show at least two multiaxial damage criteria are required: one for shear cracking failure modes and another for tensile cracking failure modes. Knowledge of the cracking behavior allows the selection of the most appropriate damage parameter. It is suggested that fatigue damage in most finite life situations could be described by one of these two simple models. If both tension and torsion test data is available, good life estimates can be made by considering both tensile and shear failure modes. The shear model overestimates the life in situations dominated by tensile crack growth and the tensile model overestimates the life for situations dominated by shear crack growth. This suggests that a possible engineering approach is to calculate the two damage parameters and estimate the life from each one. The expected life will then be the lower of the two estimates. Results for three materials are shown in Fig. 11. Correlation of the data for 167 datapoints is good. The dashed lines represent a factor of 2 in life. Test data for SAE 1045 steel is given by Fatima (1986), Fash et al. (1985), and Leese and Morrow (1985). The primary drawback to this approach is that it will require test data from both tension and torsion testing to evaluate both sets of constants in equations (1) and (2). Knowledge of the cracking behavior would allow a single set of material parameters to be used to correlate all of the data. OCTOBER 1987, Vol. 109/297 Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use i-ft-ni Inconel 718 SAE 1045 Steel 304 Stainless Steel References §• 167 Datapoints Bannantine, J. A., and Socie, D. F., 1985, "Observations of Cracking Behavior in Tension and Torsion Low Cycle Fatigue," ASTM Symposium on Low Cycle Fatigue-Directions for the Future. Bannantine, J. A., 1986, "Observations of Tension and Torsion Fatigue Cracking Behavior and the Effect on Multiaxial Damage Correlation," M.S. thesis, Mechanical Engineering Department, University of Illinois at UrbanaChampaign. Brown, M. W., and Miller, K. J., 1973, " A Theory for Fatigue Failure Under Multiaxial Stress and Strain Conditions," Proc. Inst. Mechanical Engineers, Vol. 187, pp. 746-755. Ellyin, F., and Valaire, B., 1982, "High Strain Multiaxial Fatigue," ASME o Tensile aShear Iff 10 Fig. 11 10 2 stresses can be increased by a factor of two during nonproportional loading. JOURNAL OF ENGINEERING MATERIALS AND TECHNOLOGY, Vol. 104, pp. 165-173. 103 ID 4 Estimated 105 Ifj 3 Correlation of test data using tensile and shear parameters This work has been restricted to the finite life region of less than 106 cycles. It is likely that a third damage model is required for longer lives where crack nucleation may be the primary failure mechanism. The possibility of nonpropagating cracks should also be considered in some materials. Fracture mechanics approaches could also be used rather than the bulk stress-strain approaches suggested here. Some success has been attained for in-phase loading by Socie et al. (1987), using this approach. The analytic procedures are straight forward but have some difficulties in applications to shear cracks. Referring to Fig. 3, it is difficult to estimate the stress intensity factor for the multiple cracking observed in both the tension and torsion tests of Inconel 718. Summary This paper has shown the importance of considering the manner in which cracks form and grow in the selection of the most appropriate multiaxial damage model. Two simple models based on the critical plane for crack nucleation and early growth have been presented. A shear strain model is used for materials and loadings that result in failure by mode II cracking and a tensile strain model for mode I failure. The importance of considering the additional cyclic hardening observed during out-of-phase loading was shown. For equal maximum shear strain ranges, proportional loading tests may have both larger plastic stain ranges and longer lives than non-proportional loading tests. Cyclic 298/Vol. 109, OCTOBER 1987 Fash, J. W., Socie, D. F., and McDowell, D. L., 1985, "Fatigue Life Estimates for a Simple Notched Component under Biaxial Loading," ASTM STP 853, pp. 497-513. Fatima, A., 1985, "Fatigue and Deformation Under Proportional and Nonproportional Biaxial Loading," Ph.D. thesis, University of Iowa. Fatima, A., and Socie, D. F., 1986, "Biaxial Fatigue of 1045 Steel," (in press) Fatigue of Engineering Materials and Structures. Garud, Y. S., 1979, " A New Approach to the Evaluation of Fatigue under Multiaxial Loadings," Methods for Predicting Material Life, ASME, pp. 247-263. Gaugh, H. J., 1933, "Crystalline Structure in Relation to Failure of Metals Especially Fatigue," Proc. ASTM 33, Part II, pp. 3-14. Leese, G. E., and Morow, J., 1985, "Low Cycle Torsional Fatigue of 1045 Steel in Shear Strain Control," ASTM STP 853, pp. 482-496. Lohr, R. D., and Ellison, E. G., 1980, " A Simple Theory for Low Cycle Multiaxial Fatigue," Fatigue of Engineering Materials and Structures, Vol. 3, pp. 1-17. McDowell, D. L., and Socie, D. F., 1985, "Transient and Stable Deformation Behavior under Cyclic Nonproportional Loading," ASTM STP 853, pp. 64-87. McDowell, D. L., 1983, "Transient Nonproportional Cyclic Plasticity," Ph.D. thesis, University of Illinois at Urbana-Champaign. Pascoe, K. J„ and DeVilliers, J. W. R., 1967, "Low Cycle Fatigue of Steels under Biaxial Strainings," J. Strain Analysis, Vol. 2, pp. 117-126. Stulen, F. B., 1954, " A Failure Criterion for Multiaxial Stresses," Proc. ASTM, Vol. 54, pp. 822-830. Sines, G., and Waisman, J. L., 1959, Metal Fatigue, McGraw-Hill, pp. 145-167. Smith, R. N., Watson, P., and Topper, T. H., 1970, " A Stress Strain Function for the Fatigue of Metals," /. of Materials JMLSA, Vol. 5, No. 4, pp. 767-778. Socie, D. F., Kurath, P., and Kock, J. L., 1985a, " A Multiaxial Fatigue Damage Parameter," Second International Conference on Multiaxial Fatigue. Socie, D. F. Wail, L. E., and Dittmer, D. F., 1985b, "Biaxial Fatigue of Inconel 718 Including Mean Stress Effects," Multiaxial Fatigue ASTM STP 853, pp. 463-481. Socie, D. F., Worthem, D. F., and Hua C. T., 1987, "Mixed Mode Small Crack Growth," Fatigue of Engineering Materials and Structures, Vol. 10, No. 1, pp. 1-16. Taira, S., Inoue, J., and Takashashi, M., 1967, "Low Cycle Fatigue under Multiaxial Stresses (In the Case of Combined Cyclic Tension-Compression and Cyclic Torsion in the Same Phase at Elevated Temperature)," The 10th Japan Congress on Testing Materials, pp. 18-23. Yokobori, T., Yamanouchi, H., and Yammamoto, S., 1965, "Low Cycle Fatigue of Thin-walled Hollow-cylinder Specimens of Mild Steel in Uniaxial and Torsional Tests at Constant Strain Amplitude," Int. J. Fracture Mechanics, Vol. 1, pp. 3-13. Transactions of the ASME Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use