Microsoft Word - numero_33_art_38 Y. Wang et alii, Frattura ed Integrità Strutturale, 33 (2015) 345-356; DOI: 10.3221/IGF-ESIS.33.38 345 Focussed on multiaxial fatigue Critical plane approach to multiaxial variable amplitude fatigue loading Yingyu Wang Key Laboratory of Fundamental Science for National Defense-Advanced Design Technology of Flight Vehicle, Nanjing University of Aeronautics and Astronautics, Nanjing, 210016, China yywang@nuaa.edu.cn Luca Susmel Department of Civil and Structural Engineering, the University of Sheffield, Sheffield S1 3JD, UK l.susmel@sheffield.ac.uk ABSTRACT. A new critical plane approach based on the modified Manson-Coffin curve method (MMCCM) is presented in this paper for predicting fatigue lifetime under variable amplitude (VA) multiaxial fatigue loading. The critical plane is assumed to coincide with that material plane experiencing the maximum variance of the resolved shear strain. Fatigue damage is hypothesized to be a function of both the amplitude of the resolved shear strain and the so-called critical plane stress ratio. The latter quantity depends on the mean value and the variance of the stress perpendicular to the critical plane as well as on the variance of the shear stress resolved along the direction experiencing the maximum variance of the resolved shear strain. Load cycles are counted from the resolved shear strain time history by using the classic rain flow counting method. Palmgren-Miner’s linear damage rule is applied to estimate cumulative fatigue damage. The accuracy and reliability of the proposed approach is checked by using several experimental data taken from the literature. The estimated fatigue lives based on the new approach are seen to be in sound agreement with the experimental results. KEYWORDS. Multiaxial fatigue; Variable amplitude loading; Critical plane; Fatigue life prediction. INTRODUCTION atigue life prediction approaches based on the concept of the critical plane are generally considered to be accurate enough to estimate fatigue lifetime under multiaxial load histories. The critical plane concept is based on the physical observation that cracks initiate and grow on some specific material planes. We believe there are three key aspects needed to be considered when applying the critical plane approach to estimate fatigue lifetime under multiaxial variable amplitude loading, i.e.: (i) determining the orientation of the critical plane, (ii) counting the fatigue cycles and (iii) calculating the amplitude and mean value of the stress/strain components relative to the critical plane. As to the available critical plane approaches, Findley [1] determined the critical plane by maximizing a linear combination of the shear stress amplitude and the maximum normal stress. Brown and Miller [2] defined the critical plane as the plane experiencing the maximum shear strain amplitude. Wang and Brown [3] have also proposed a modified version of their criterion by defining the critical plane as the one experiencing not only the maximum shear strain range, but also the largest value of the normal strain excursion. Fatemi and Socie [4] assumed that the critical plane is the plane of maximum shear strain amplitude when the crack initiation process is Mode II governed. From the brief review presented above, it F Y. Wang et alii, Frattura ed Integrità Strutturale, 33 (2015) 345-356; DOI: 10.3221/IGF-ESIS.33.38 346 becomes clear that there are different ways to define the critical plane itself. However, it is not an easy task to determine the orientation of the critical plane when the applied multiaxial load history varies randomly. Several investigations [5-9] have been published by focusing attention on the determination of the critical plane under multiaxial random loading. Susmel [9] have recently formalized a novel technique, the Shear stress-Maximum Variance Method (-MVM). The Maximum Variance Method (MVM) postulates that the critical plane can be defined as that plane containing the direction (passing through the assumed critical point) that experiences the maximum variance of the resolved shear stress. The peculiarity of the MVM is that it is not necessary to calculate the stress/strain components relative to any plane passing through the assumed critical point to determine the critical plane itself. Therefore, from a computational point of view, it is more efficient than the other existing methods. In order to take into account cyclic hardening or softening in the fatigue failure criteria, both stress and strain components are recommended to be used to estimate the fatigue damage extent. As to the in-field usage of the available VA multiaxial fatigue life estimation techniques, correctly performing the cycle counting under such complex loading conditions is one of the trickiest aspects [10]. Even though there are a few methods suitable for counting the cycles under uniaxial loading, among these the rainflow cycle counting method [11] has been most widely and successfully used. The number of published papers dealing with VA multiaxial loading histories is small [12]. Bannantine and Socie [13] proposed a method based on the critical plane concept and the rainflow cycle counting. Wang and Brown [14] proposed a cycle counting method based on the rainflow and a modified von Mises equivalent strain. The accuracy of the results predicted by using the critical plane approach to a great extent relies on the determination of the amplitude and mean value of the normal and shear stresses and strains acting on a particular plane [15]. For a particular plane under complex loading, the direction of the normal stress/strain does not change over time, since just the magnitude of the normal stress/strain vector varies. However, in the most general case, the shear stress/strain vector on this particular plane changes in both magnitude and direction. Therefore, the tip of the shear stress/strain vector draws an imaginary curve on that plane. In this situation, it is not an easy task to calculate the amplitude and mean value of the shear stress/strain on a given plane. There are several techniques proposed to address this problem, namely the Longest Chord Method, the Longest Projection Method, the Minimum Circumscribed Circle concept, and the Minimum Circumscribed Ellipse Method. These methods have been discussed in the Refs [9, 15]. The present paper summarizes an attempt of extending the use of the Modified Manson-Coffin Curve Method (MMCCM) to those situations where complex variable amplitude loadings are involved. By making the most of our previous experience [10, 16-18], now we intend to use the stain-based MVM in conjunction with MMCCM to predict fatigue lifetime under VA multiaxial fatigue loading. In more detail, the MMCCM is suggested here as being applied in conjunction with the maximum variance method (MVM) to estimate fatigue lifetime by directly post-processing the strain state relative to that material plane containing the direction along which the variance of the resolved shear strain is maximized. Since, by definition, the resolved shear strain is a monodimensional quantity, the rainflow method can directly be used to count the loading cycles. After counting the cycles and estimating the fatigue damage for each counted cycle, fatigue lifetime can be predicted directly by using the Palmgren-Miner’s linear damage rule. Finally, the validation of the new approach is presented. THE SHEAR STRAIN_MAXIMUM VARIANCE METHOD (-MVM) he shear strain-Maximum Variance Method (MVM) assumes that the critical plane can be defined as that plane containing the direction experiencing the maximum variance of the resolved shear strain. In order to calculate the shear strain relative to a generic material plane,  , and resolved along a generic direction, q, consider a body subjected to an external system of forces resulting in a triaxial strain state at point O (Fig. 1a). Point O is taken as the center of the absolute system of coordinates, Oxyz. The time-variable strain state at point O is defined through the following strain tensor:   x xy xz x xy xz yx y yz xy y yz zx zy z xz yz z 1 1 ε (t ) (t ) (t ) 2 2ε (t ) ε (t ) ε (t ) 1 1 ε t ε (t ) ε (t ) ε (t ) γ (t ) ε (t ) γ (t ) 2 2 ε (t ) ε (t ) ε (t ) 1 1 γ (t ) γ (t ) ε (t ) 2 2                              (1) T Y. Wang et alii, Frattura ed Integrità Strutturale, 33 (2015) 345-356; DOI: 10.3221/IGF-ESIS.33.38 347 where x , y and z are the three normal strains, and xy , xz and yz are the total shear strains. The orientation of a generic material plane,  , having normal unit vector n can be defined through angles  and  (Fig. 1b).  is the angle between axis x and the projection of unit vector n on plane x-y.  is the angle between n and axis z. A new system of coordinates, Onab, can now be defined. The unit vectors defining the orientation of axes n, a and b can be expressed as follows:           sin cos sin sin cos x y z n n n                        n ;     sin cos 0 x y z a a a                      a ;           cos cos cos sin sin x y z b b b                        b (2) Figure 1: Generic plane and shear strain resolved along one generic direction in a body subjected to an external system of forces. Consider now a generic direction q lying on plane  and passing through point O. α is the angle between direction q and axis a. The unit vector defining the orientation of q can be calculated as follows:                         cos sin sin cos cos cos cos sin cos sin sin sin x y z q q q                                 q (3) According to the definition reported above, the instantaneous value of the shear strain resolved along direction q, q ( )t , can then be calculated as: x xy xz q y xy y yz xz yz z 1 1 ε (t ) (t ) (t ) 2 2 γ ( ) 1 1 γ (t ) ε (t ) γ (t ) 2 2 2 1 1 γ (t ) γ (t ) ε (t ) 2 2 x x z y z n t q q q n n                         (4) In order to make the calculation easier, it is useful to express q ( )t through the following scalar product:  q γ ( ) 2 t t d s (5) where d is the vector of direction cosines, that is: Y. Wang et alii, Frattura ed Integrità Strutturale, 33 (2015) 345-356; DOI: 10.3221/IGF-ESIS.33.38 348                                 2 1 2 2 3 4 5 6 1 sin θ sin 2φ cos α sin 2θ cos φ sin α 2 d 1 sin θ sin 2φ cos α sin 2θ sin φ sin α d 2 1d sin 2θ sin α 2d 1d sin 2θ sin 2φ s 2d x x y y z z x y y x x z z x y z z y n q n q n q n q n q n q n q n q n q                                                    d                                 in α sin θ cos 2φ cos α cos 2θ cos φ sin α cos θ sin φ cos α cos 2θ sin φ sin α cos θ cos φ cos α                                       (6) and s(t) is a six-dimensional vector process depending on  [ε ]t and defined as: x y z xy xz yz 1 1 1 ( ) ε (t ) ε (t ) ε (t ) (t ) (t ) γ (t ) 2 2 2 t       s (7) According to the quantities defined above, the variance of the shear strain, q ( )t , resolved along direction q can then be calculated directly as:      Var Var ( ) Cov , ( ) 2 q T k k i j i j k i j t d s t d d s t s t C                   d d (8) where [C] is a symmetric square matrix of order six, and the terms of the covariace matrix are defined as    Cov ,  ij i jC s t s t    (9) where, when i j then      Cov ,   Vari j is t s t s t       , whereas when i j then        Cov ,   Cov ,  i j j is t s t s t s t       . Now [C] can be rewritten in explicit form by using both the variance and covariance terms:   x x , y x ,z x ,xy x ,xz x , yz x , y y y ,z y ,xy y ,xz y , yz x ,z y ,z z z,xy z,xz z, yz x ,xy y ,xy z ,xy xy xy ,xz xy , yz x ,xz y ,xz z,xz xy ,xz xz xz, yz x , yz y , yz z, yz xy , yz xz, yz yz V C C C C C C V C C C C C C V C C C C C C C V C C C C C C V C C C C C C V                     (10) where  Var     for  ,  ,  ,  ,  , i iV t i x y z xy xz yz    (11)    ,  CoVar ,       for  ,  ,  ,  ,  , i j i jC t t i x y z xy xz yz     (12) Then Eq. (8) can be rewritten in the following simple form:    Var Tq t C    d d (13) Eq. (13) makes it evident that the determination of the direction experiencing the maximum variance of the resolved shear strain is a conventional multi-variable optimization problem. It can be solved satisfactorily by simply using the so-called Gradient Ascent Method [19]. Fig. 2 reports the flowchart summarising the algorithm which was proposed in Ref. [9] to be used to determine the orientation of the critical plane. Y. Wang et alii, Frattura ed Integrità Strutturale, 33 (2015) 345-356; DOI: 10.3221/IGF-ESIS.33.38 349 After determining the direction of the maximum variance, the shear strain resolved along this direction, MV ( )t , can directly be evaluated, at any instant, t, of the load history, through the following relationship:   x xy xz MV xy y yz xz yz z 1 1 ε (t ) (t ) ( t ) 2 2 1 1 γ ( t ) ε ( t ) γ (t ) 2 2 2 1 1 γ (t ) γ (t ) ε (t ) 2 2 x x y z y z n t q q q n n                             (14)                                                              nnn nnn nnn n n n n n n t t t k                ,, )(Var ,, )(Var ,, )(Var q q q 1 1 1 Figure 2: Flowchart summarizing the algorithm for the determination of the critical plane orientation. The shear stress resolved along the maximum variance direction, MV ( )t , and the stress normal to the plane experiencing the maximum variance of resolved shear strain, n ( )t , can be evaluated, at any instant, t, of the load history, through the following relationships:   x xy xz MV xy y yz xz yz z (t ) (t ) (t ) (t ) (t ) (t ) (t ) (t ) (t ) x x y z y z n t q q q n n                            (15) Y. Wang et alii, Frattura ed Integrità Strutturale, 33 (2015) 345-356; DOI: 10.3221/IGF-ESIS.33.38 350   x xy xz n xy y yz xz yz z (t ) (t ) (t ) (t ) (t ) (t ) (t ) (t ) (t ) x x y z y z n t n n n n n                            (16) EVALUATION OF STRAIN AND STRESS COMPONENTS RELATIVE TO THE CRITICAL PLANE onsider the body subjected to a complex system of time variable forces shown in Fig. 3. The applied load history is defined in the time interval [0, T ] and it is assumed to result in a VA multiaxial strain state at point O. By using the -MVM, the orientation of the potential critical plane can be determined through that direction which experiences the maximum variance of the resolved shear strain. After determining the direction of maximum variance (MV), the shear strain resolved along this direction, MV ( )t , the shear stress resolved along the maximum variance direction, MV ( )t , and the stress normal to the plane experiencing the maximum variance of the resolved shear strain, n ( )t , can be evaluated, at any instant, t, of the load history, through Eq. (14) to (16). Attention can initially be focused on stress component, n ( )t , which is defined in the time interval [0, T]. The mean value of this stress component is equal to T n , m n 0 1 σ σ ( ) T t dt  (17) The equivalent amplitude of n ( )t is suggested here as being calculated as follows:  n ,a nσ 2 Var[σ ]t  (18) where Var[ n ( )t ] is the variance of stress component n ( )t , i.e. T 2 n n n ,m 0 1 Var[σ ( )] [σ ( ) σ ] T t t dt  (19) Consider now the shear stress, MV ( )t , resolved along direction MV. Owing to the fact that also MV ( )t is a monodimensional stress quantity defined in the time interval [0, T], its mean value and its equivalent amplitude can directly be determined by following a strategy similar to the one adopted above to calculate n ,m and n ,a . In particular, m is equal to T m MV 0 1 τ τ ( ) T t dt  (20) whereas the equivalent amplitude of the shear stress resolved along direction MV takes on the following value:  a MVτ 2 Var[τ ]t  (21) where T 2 MV MV m 0 1 Var[τ ( )] [τ ( ) τ ] T t t dt  (22) Then a stress ratio  ρ , which will be used to adapt the Manson-Coffin curve to the degree of multi-axiality and non- proportionality of the stress/strain state at the assumed crack initiation site can be defined as follows: n , m n , a a σ σ ρ τ   (23) C Y. Wang et alii, Frattura ed Integrità Strutturale, 33 (2015) 345-356; DOI: 10.3221/IGF-ESIS.33.38 351 To conclude the present section it is important to highlight that, in general, there exist two or more different directions which experience the maximum variance of the resolved shear strain, so that, it is always possible to locate, at least, two potential critical planes: amongst all the potential critical planes, the one which has to be used to estimate fatigue lifetime is that experiencing the largest value of ρ . THE MMCCM TO ADDRESS THE VARIABLE AMPLITUDE PROBLEM The use of the MMCCM in those situations involving VA multiaxial load histories is based on the following assumptions: (i) initiation and initial propagation of Stage  cracks occur on that material plane containing the direction experiencing the maximum variance of the resolved strain, MV ( )t ; (ii) the fatigue damage depends also on the stress ratio ρ which is defined in the previous Section. Fig. 3 summarizes the approach proposed in the present paper to be followed to estimate fatigue lifetime of engineering materials subjected to in-field VA multiaxial fatigue loading. In more detail, consider a body subjected to a complex system of time variable forces resulting in a VA multiaxial strain state at the assumed crack initiation site (i.e. point O in Fig. 3a). Initially, by making use of the shear -MVM, the orientation of the candidate critical planes can directly be determined through that direction, MV, experiencing the maximum variance of the resolved shear strain (Fig. 3b). After determining the direction of maximum variance of the shear strain, the shear strain resolved along this direction, MV ( )t , the shear stress resolved along the maximum variance direction, MV ( )t , and the stress normal to the critical planes, n ( )t , can be evaluated, at any instant, t, of the load history, through Eq. (14) to (16) (Figs 3c and d). Subsequently, the calculated values for n ,m (Eq. 17), n ,a (Eq. 18) and a (Eq. 21) have to be used to estimate, according to Eq. (23), critical plane stress ratio ρ . According to the determined value for ρ , the profile of the corresponding modified Manson-Coffin curve can be described by using the following general relationship [18]:     ' ( )f ( ) ' a f f f( 2 ) ( ) 2 cbN N G       (24) where 'f ( )  , ' f ( )  , ( )b  , ( )c  are material functions, which can be determined through the following equations [18]:     0 0 b b b b b b       (25)     A , ref' f ( ) A ( ) 2 b N       (26)       0 ' 'f A , ref A f A2 (1 ) 2 2 b b N N         (27)    ' ' 'f p f f1 ν (1 )          (28)     0 0 c ρ c c c c c     (29) where p is Poisson’s ratio for plastic strain, NA is the reference number of cycles to failure. After determining the appropriate reference fatigue curve as above, by post-processing the shear strain resolved along direction MV, the classical rain-flow method allows the corresponding shear strain spectrum to be built directly (Figs 3g and h). Finally, from such spectrum fatigue strength under VA multiaxial fatigue loading can be predicted according to Palmgren-Miner’s rule (Fig. 3i). Here it is assumed that breakage takes place when the damage sum equals unity. Y. Wang et alii, Frattura ed Integrità Strutturale, 33 (2015) 345-356; DOI: 10.3221/IGF-ESIS.33.38 352 VALIDATION BY EXPERIMENTAL DATA n order to check the accuracy of the proposed approach in estimating fatigue lifetime under VA multiaxial fatigue loading, a number of experimental data were selected from the technical literature [12, 20-22]. The summary of the static and fatigue properties of the considered materials is reported in Tab. 1 and 2. When the torsional material constants listed in the above tables were not directly available in the original sources, they were calculated as follows [23]: 2(1 ) E G    ' '3f f  ' ' 3 f f    0b b 0c c Figure 3: In-field use of the MMCCM to estimate fatigue lifetime under VA multiaxial fatigue loading. The investigated materials’ non-proportional hardening was taken into account by making the following assumption [23]: I Y. Wang et alii, Frattura ed Integrità Strutturale, 33 (2015) 345-356; DOI: 10.3221/IGF-ESIS.33.38 353 ' 'NP 1.25K K  ' ' NPn n Both the strain loading history and the stress loading history at the assumed crack initiation site are needed. If the stress loading histories were not given in the original sources, they were calculated by using the model proposed by Jiang and Sehitoglu [24]. In Fig. 4, the predicted fatigue lives are compared with the experimental ones for S45C steel and SNCM630 steel under VA multiaxial fatigue loading. Fig. 4 also shows the loading paths applied to specimens made of S45C steel and SNCM630 steel. As it can be seen from this figure, 81% of the data are within a scatter bands of 3. Data points under loading blocks with a considerable portion of axial loading such as AT, TA and AV are outside of the target scatter band, being on the conservative side. Material Ref. E (MPa) G (MPa) ν S45C [20] 186,000 70,600 0.28 SNCM630 [21] 196,000 77,000 0.273 1050 QT Steel [12] 203,000 81,000 0.27 304L steel [12] 195,000 77,000 0.27 Pure titanium [22] 112,000 40,000 0.4 Titanium alloy BT9 [22] 118,000 43,000 0.37 Table 1: Elastic static constants for the considered materials. Material Ref. ’f ’f (MPa) b c ’f ’f (MPa) b0 c0 K’ (MPa) n’ K’NP (MPa) n’NP S45C [20] 0.359 923 - 0.099 - 0.519 0.198 685 -0.12 -0.36 1215 0.217 1519 0.217 SNCM630 [21] 1.54 1272 - 0.073 - 0.823 1.51 858 -0.061 - 0.706 1056 0.054 1320 0.054 1050 QT St. [12] 2.01 1346 - 0.062 - 0.725 3.48 777 -0.062 - 0.725 1461 0.06 1420 0.113 304L St. [12] 0.122 1287 - 0.145 - 0.394 0.211 743 -0.145 - 0.394 680 0.214 5056 0.373 Pure Ti [22] 0.548 647 - 0.033 - 0.646 0.417 485 -0.069 - 0.523 - - - - BT9 [22] 0.278 1180 - 0.025 - 0.665 0.18 881 - 0.0082 -0.47 - - - - Table 2: Fatigue constants for the considered materials. Figure 4: Comparison of observed and predicted fatigue lives by the MMCCM for S45C and SNCM630. Y. Wang et alii, Frattura ed Integrità Strutturale, 33 (2015) 345-356; DOI: 10.3221/IGF-ESIS.33.38 354 The predicted versus experimental fatigue lifetime diagram for 1050 QT steel and 304L steel under discriminating axial- torsion stain paths with random and incremental changes in straining direction is reported in Fig. 5. As it can be seen from Fig. 5, all the data are within scatter bands of 3 from predictions. Fig. 6 shows the predicted versus experimental fatigue lifetime diagram for pure titanium and titanium alloy BT9 under step loading and block loading composed of different combinations of axial, torsion and 90°out-of-phase axial-torsion strain paths. As it can be seen from Fig. 6, 90% of the data are within a scatter band of 3. Figure 5: Comparison of observed and predicted fatigue lives by the MMCCM for 1045 QT steel and 304L steel. Figure 6: Comparison of observed and predicted fatigue lives by the MMCCM fo pure titanium and titanium alloy BT9. CONCLUSIONS (1) The MMCCM originally proposed by Susmel et al. [18] for multiaxial constant amplitude loading was reformulated to estimate fatigue lifetime of components subjected to VA multiaxial fatigue loading. Satisfactory fatigue life predictions were obtained by using the MMCCM, when coupled with the Rain-Flow cycle counting method and Palmgren-Miner Y. Wang et alii, Frattura ed Integrità Strutturale, 33 (2015) 345-356; DOI: 10.3221/IGF-ESIS.33.38 355 linear cumulative damage rule for different materials tested under VA multiaxial fatigue loading. The critical plane was determined by using the -MVM. (2) The MVM is an efficient and valid method to determine the orientation of the critical plane when it is used to evaluate fatigue damage under complex variable amplitude multiaxial fatigue loading. (3) Since the controlling parameter of the MMCCM is just the shear strain resolved along the direction of maximum variance of the resolved shear strain, when the MMCCM is used to estimate fatigue damage under variable amplitude multiaxial fatigue loading, cycles can directly be counted by using rainflow method. ACKNOWLEDGEMENTS his work is partly supported by Jiangsu Oversea Research & Training Program for University Prominent Young & Middle-aged teachers and Aviation Science Funds of China (No. 2013ZA52008). REFERENCES [1] Findley, W.N., Modified theory of fatigue failure under combined stress, In: Proc of the society of experimental stress analysis, 14 (1956) 35-46. [2] Brown, M.W., Miller, K.J., A theory for fatigue under multiaxial stress-strain conditions, In: Proc institution of mechanical engineering, 187 (1973) 745-56. [3] Wang, C.H., Brown, M.W., A path-independant parameter for fatigue under proportional and non-proportional loading, Fatigue of Engineering Materials and Structures, 16 (1993)1285–1293. [4] Fatemi, A., Socie, D.F., A critical plane approach to multiaxial fatigue damage including out-of-phase loading, Fatigue Fract Eng Mater Struct, 11 (1988) 149-65. [5] Marciniak, Z., Rozumek, D., Macha, E., Verification of fatigue critical plane position according to variance and damage accumulation methods under multiaxial loading, Int J Fatigue, 58 (2014) 84-93. [6] Bedkowski, W., Macha, E., Ohnami, M., Sakane, M., Fracture plane of cruciform specimen in biaxial low cycle fatigue-estimate by variance method and experimental verification, Trans ASME J Eng Mater Technol, 117 (1995) 183–90. [7] Carpinteri, A., Macha, E., Brighenti, R., Spagnoli, A., Expected principal stress directions under multiaxial random loading Part I: Theoretical aspects of the weight function method, Int J Fatigue, 21(1999) 83–88. [8] Carpinteri, A., Macha, E., Brighenti, R., Spagnoli, A., Expected principal stress directions under multiaxial random loading Part II: numericalsimulation and experimentally assessment through the weight functionmethod, Int. J. Fatigue, 21(1999): 89–96. [9] Susmel, L., A simple and efficient numerical algorithm to determine the orientation of the critical plane in multiaxial fatigue problems, Int. J. Fatigue, 32 (2010) 1875-1883. [10] Susmel, L., Tovo, R., Estimating fatigue damage under variable amplitude multiaxial fatigue loading, Fatigue & Fracture of Engineering Materials & Structures, 34 (2011) 1053-1077. [11] Matsuishi, M., Endo, T., Fatigue of Metals Subjected to Varying Mar. Stress, Presented at Japan Society of Mechanical Engineers, Fukuoka, Japan, (1968). [12] Shamsaei, N., Fatemi, A., Socie, D.F., Multiaxial fatigue evaluation using discriminating strain paths. Int. J. Fatigue, 33 (2011) 597-609. [13] Bannantine, J.A., Socie, D.F., Multiaxial fatigue life estimation technique, In: Mitchel M, Landgraf R, editors. ASTM symposium on advances in fatigue lifetime predictive techniques, ASTM STP 1122 (1991) 249–75. [14] Wang, C.H., Brown, M.W., Life prediction techniques for variable amplitude multiaxial fatigue – part I: theories, J. Eng. Mater. Technol. 118 (1996) 367–70. [15] Papadopoulos, I.V., Critical plane approaches in high-cycle fatigue: on the definition of the amplitude and mean value of the shear stress acting on the critical plane, Fatigue Fract. Eng. Mater. Struct., 21 (1998) 269-285. [16] Susmel, L., Tovo, R., Benasciutti, D., A novel engineering method based on the critical plane concept to estimate the lifetime of weldments subjected to variable amplitude multiaxial fatigue loading, Fatigue & Fracture of Engineering Materials & Structures, 32 (2009) 441-459. [17] Susmel, L., Taylor, D., A critical distance/plane method to estimate finite life of notched components under variable T Y. Wang et alii, Frattura ed Integrità Strutturale, 33 (2015) 345-356; DOI: 10.3221/IGF-ESIS.33.38 356 amplitude uniaxial/multiaxial fatigue loading, International Journal of Fatigue, 38 (2012) 7-24. [18] Susmel, L., Meneghetti, G., Atzori, B., A simple and efficient reformulation of the classical Manson-Coffin curve to predict lifetime under multiaxial fatigue loading-Part I: plain materials, Journal of Engineering Materials and Technology, 131 (2009) 021009-1-021009-9. [19] Snyman, J.A., Practical mathematical optimization: an introduction to basic optimizaion theory and classical and new gradient-based algorithms, Springer Publishing, (2005). [20] Kim, K.S., Park, J.C., Lee, J.W., Multiaxial fatigue under variable amplitude loads, Trans ASME J. Eng. Mater. Technol., 121 (1999) 286-93. [21] Han, C., Chen, X., Kim, K.S., Evaluation of multiaxial fatigue criteria under irregular loading, Int. J. Fatigue, 24 (2002) 913-922. [22] Shamsaei, N., Gladskyi, M., Panasovskyi, K., Fatemi, A., Multiaxial fatigue of titanium including step loading and load path alteration and sequence effects, Int. J. Fatigue, 32 (2010) 1862-1874. [23] Socie, D.F., Marquis, G.B., Multiaxial fatigue, Warrendale, USA, (1999). [24] Jiang, Y., Sehitoglu, H., Modeling of cyclic racheting plasticity, Part : development of constitutive equations, Journal of Applied Mechanics, 63 (1996) 720-725.