MODELLING STRAIN-RAT E-DEPENDENCY OF

Acta Mechanica Solida Sinica,Vol.24,No.3,June,2011ISSN0894-9166 Published by AMSS Press,Wuhan,ChinaMODELLING STRAIN-RATE-DEPENDENCY OF NATURAL SOFT CLAYS COMBINED WITHANISOTROPY AND DESTRUCTURATIONZhen-Yu Yin1 Minna Karstunen2(1Department of Civil Engineering,Shanghai Jiao Tong University,Shanghai200240,China) (2Department of Civil Engineering,University of Strathclyde,John Anderson Building,107Rottenrow,Glasgow G40NG,UK)Received16November2009,revision received30October2010ABSTRACT The paper aims to investigate modelling the strain-rate-dependency of natural softclays combined with anisotropy and destructuration using an elasto-viscoplastic model.The modelis based on Perzyna’s overstress theory and the elastoplastic model S-CLAY1S.Tests at constantstrain-rate and creep tests under both one-dimensional and triaxial conditions on several claysare simulated.Simulations highlight the loading scenarios in which it is necessary to account foranisotropy and/or destructuration in order to get accurate parisons between thepredicted and measured results demonstrate that the proposed model can successfully reproducethe time-dependent behaviour of natural soft clays under different loading conditions.KEY WORDS anisotropic materials,creep,debonding,time-dependentI.INTRODUCTIONNatural soft clays exhibit a significant degree of anisotropy developed during their deposition,sed-imentation,consolidation history and any subsequent straining[1–3].Based on experimental studies on natural Otaniemi clay,Wheeler et al.[4]showed that the changes in anisotropy can be linked to plastic straining,and proposed an elastoplastic model S-CLAY1extended from Modified Cam Clay[5],with a rotational hardening law describing the changes in the inclination of the yield surface.The calibration of the parameters for S-CLAY1model is straightforward and the model has been thoroughly validated experimentally by Karstunen and Koskinen[6].In addition to anisotropy,most natural soils also exhibit some apparent bonding.Experimental observations[1,7]show that as natural clays are deformed,the initial apparent bonding is progressively lost,and at large strains,soil starts to behave as a fully reconstituted material.According to Leroueil et al.[8],the process,during which the bonds are progressively destroyed by plastic straining,is referred to as destructuration.Incorporating the concept of an intrinsic yield surface proposed by Gens and Nova[9],Karstunen et al.[10]developed an S-CLAY1S model for structured soils extended from S-CLAY1 model.The initial amount of bonding can be related to the sensitivity of natural clay.Corresponding author.E-mail:zhenyu.yin@The work presented was sponsored by the Academy of Finland(Grants210744and1284594)and the European Community through the program‘People’as part of the Industry-Academia Pathways and Partnerships project GEO-INSTALL(PIAP-GA-2009-230638).Vol.24,No.3Zhen-Yu Yin et al.:Modelling Strain-Rate-Dependency of Natural Soft Clays·217·The fundamental stress-strain-strength behaviour of soft soils is however also strain-rate-dependent. The shear strength of clays from undrained triaxial tests is highly dependent on the applied axial strain-rate,as shown by numerous researchers[11–14].Similar results by Leroueil et al.[15]demonstrate the strain-rate-dependency of the apparent preconsolidation pressure measured from constant strain rate oedometer tests.The strain-rate dependency of soft clays has significant impact on geotechnical engineering practice:the response of geotechnical structures is dependent on the rate of loading,as the fundamental stress-strain-strength behaviour of soils is strain-rate-dependent.As a result,the selection of design parameters should take into account the strain-rate-dependency of deformations and strength. Therefore,a constitutive model taking into account the strain-rate-dependent behaviour,in addition to anisotropy and destructuration,should be recommended.In the literature,numerous constitutive models have been proposed to describe some of these features: for instance,rate-independent models[4,5,10];rate-dependent isotropic models[16–22];rate-dependent anisotropic models[23–26];etc.Kimoto and Oka[24]developed a rate-dependent model accounting for destructuration and inherent anisotropy.However,the induced anisotropy during loading has not been considered.Neglecting one feature may lead to inaccurate predictions of soil response under loading, especially for foundations and embankments on soft sensitive soils.In order to model all these features,an elasto-viscoplastic model based on Perzyna’s overstress theory[27,28]and the S-CLAY1S model[10],has been developed by Yin and Karstunen[20].In works by Yin and Karstunen[29],the development of the model was presented with an application to a hypothet-ical embankment.No validation was carried out in terms of the comparison between simulations and observations.In this paper,the elasto-viscoplastic model is used to investigate modelling the strain-rate-dependency of natural soft clays combined with anisotropy and destructuration at element test-level.First,the development of the model is briefly presented.The calibration of model parameters is discussed.The experimental validation of the model considers strain-rate-controlled oedometer and triaxial tests on several clays.Long-term oedometer tests and undrained triaxial creep tests are also simulated.The simulations highlight the loading scenarios in which it is necessary to account for anisotropy and/or destructuration in order to get accurate predictions.II.CONSTITUTIVE MODEL EVP-SCLAY1S2.1.Model DescriptionThe constitutive model,called EVP-SCLAY1S,accounting for strain-rate-dependency,anisotropy and destructuration of natural soft soils,is based on the overstress theory of Perzyna[27,28]and elasto-plastic model S-CLAY1S[10].According to Perzyna’s overstress theory,the total strain rate is the sum of the elastic strain rates and viscoplastic strain rates.The elastic behaviour in the proposed model is assumed to be isotropic,similarly to the Modified Cam Clay model.The viscoplastic strain rate˙εvpij is assumed to obey an associatedflow rule with respect to the dynamic loading surface following the original proposal by Perzyna:˙εvp ij =μ Φ(F)∂f d∂σp s−·218·ACTA MECHANICA SOLIDA SINICA2011Fig.1.Definition of surfaces in the p -q space.Following the ideas by Wheeler et al.[4],an elliptical surface is adopted as the dynamic loading surface corresponding to the current stress state:f d=32[σM2−32[α(M2−α2)p+3/2(α:α)de 1+c4+(1−c4)sin(3θ)M c,−π6≤3sin3¯J¯3/26w2¯s i3¯s iλ−κVol.24,No.3Zhen-Yu Yin et al.:Modelling Strain-Rate-Dependency of Natural Soft Clays·219·whereλi is the slope of the intrinsic normal compression curve in the e-lnσ v plane for a stress path at a constant stress ratio involving no change of anisotropy(such as isotropic loading of an isotropic sample,or oedometer test on a reconstituted sample,see Fig.2),κis the slope of the swelling-line and e0is the initial void ratio.The rotational hardening law,based on the formulation proposed by Wheeler et al.[4],describes the development or erasure of anisotropy caused by viscoplastic strains.Both volumetric and deviatoric viscoplastic strains influence the rotation of the yield curve.dαd=ω3σd4p−3p−Fig.2.One-dimensional behaviour of intact and reconstituted samples of soft sensitive clays.where the soil constantξcontrols the absolute rate of destructuration andξd controls the relative effect of viscoplastic deviatoric strains in destroying the bonds;χ0is the initial amount of bonding,which relates the sizes of the intrinsic yield surface and static yield surfaces asp m0=(1+χ0)p mi0(8) The numerical implementation of the model EVP-SCLAY1S as a user-defined soil model into the 2D v8PLAXISfinite element code is based on a numerical solution proposed by Katona[32].The basic·220·ACTA MECHANICA SOLIDA SINICA2011finite element scheme for the proposed model is similar to the ones presented by Oka et al.[33]and Rowe and Hinchberger[34].For a coupled consolidation analysis based on Biot’s theory,the relationship of the load increment is given by applying the principle of virtual work to the equilibrium equation as shown by Oka et al.[33].The coupledfinite element equations are well documented by several researchers[33–35], and not repeated here.2.2.Principle of the Elasto-viscoplastic ModelThe principle of EVP-SCLAY1S is illustrated in Fig.3(a).Again for simplicity,the soil is assumed to be initially cross-anisotropic using the definitions in Fig.1.In Fig.3(a),point‘A’represents the initial stress state that is assumed to correspond to the normally consolidated K0state(K0is the lateral earth pressure at rest).Along the stress path of an undrained constant strain-rate shear test,‘A-B’in Fig.3(a),viscoplastic volumetric strains occur during loading and cause the initial intrinsic yield surface to expand in the stress space.The viscoplastic volumetric and deviatoric strains cause changes in the inclination of the yield surfaces and degradation in bonding.The latter may reduce the size of the static yield surface.Supposing that the stress state‘B’approaches the critical state line at‘C’,the size of the intrinsic yield surface becomes constant,given that the increments of the viscoplastic volumetric strains tend to zero.The deviatoric viscoplastic strains cause debonding of soil particles until the intrinsic and static yield surfaces coincide.The critical state for this applied strain-rate is reached.The soil is then subjected to a constant amount of overstress,which provokes an increase in the viscoplastic deviatoric strains at constant strain-rate.All through this process,the yield surfaces rotate tending towards a final orientation of M c/3which was explained and experimentally supported by Wheeler et al.[4].Fig.3.Schematic representation of the behaviour of the EVP-SCLAY1S model in the p -q space.In the case of an undrained creep test,represented by stress path‘A-B-C’in Fig.3(b),the size of the intrinsic yield surface changes with viscoplastic volumetric strains,related to the amount of overstress. The size of the static yield surface is,however,also influenced by destructuration.Therefore,viscosity and destructuration have strong coupling.In the case of a highly structured clay,destructuration will dominate,while for unstructured soils viscosity dominates.Ultimately the static yield surface and the dynamic loading surface coincide,equilibrium is obtained and the strains stabilize with time.If the effective stresses reach the critical state line,there will be naturally no changes in the size of the intrinsic yield surface.The viscoplastic deviatoric strains will continue to increase and rotate the yield surfaces towards their ultimate target orientation.An attractive feature of the proposed model is that it is hierarchical.By giving a sufficiently small viscosity(via assigning suitable values for N andμ)the proposed model converges to the elastoplastic model S-CLAY1S.Similarly,by setting the initial value for bondingχ0to zero,and using an apparent value ofλdetermined from a consolidation test on a natural clay sample,instead of the intrinsic value of λi from a reconstituted clay sample(Fig.3),the proposed model will account for anisotropy and viscosityVol.24,No.3Zhen-Yu Yin et al.:Modelling Strain-Rate-Dependency of Natural Soft Clays·221·only(EVP-SCLAY1).Likewise,if the initial anisotropy(described via scalarα0for cross-anisotropic soils)is set to zero and the anisotropy parameterωis set to zero,the model can reduce to a viscoplastic version of the Modified Cam Clay(MCC)model,denoted as EVP-MCC.If this is combined with a sufficiently small viscosity,the model ultimately reduces to the elastoplastic MCC model.2.3.Evaluation of Model ParametersThe proposed model involves a number of soil constants and state variables which can be divided into four main groups.Thefirst set of parameters are the same as in the MCC model,withλ(slope of the normal compression line)replaced by the intrinsic compressibilityλi,which can be determined for example from a1D consolidation test on a reconstituted sample or from a1D consolidation test on an intact soil sample executed to very high stress levels so that virtually all the bonds have been destroyed.The initial size of the static yield surface can be calculated by Eq.(4)assuming the stress state corresponding to in situ effective stresses,since soils are under these stresses for years.The second set of parameters is related to anisotropy.According to Wheeler et al.[4],the initial anisotropyα0can be calculated from M c by assuming that the initial anisotropy has been created by one-dimensional compression,which implies in the normally consolidated region thatα0=αK0=ηK0−M2c−η2K038(η2+2η−M2)·222·ACTA MECHANICA SOLIDA SINICA 2011Fig.4.The effect of viscosity parameters on vertical preconsolidation stress/undrained shear strength at high strain-rates.Table 1.State parameters and soil constants of EVP-SCLAY1S modelGroup Parameter DefinitionDeterminationInitial size of yield surface Anisotropyparameter ωAbsolute rate of yield surface rotation Typically [10/λi -20/λi ]Calculated from soil sensitivity χ0=S t −l χ0Initial amount of bonding with S t determined e.g.with a fall cone test Destructuration (state parameter)(or from 1D consolidation tests,see Fig.2)parameters From curve fitting of consolidation tests:ξAbsolute rate of bond degradation oedometer test and isotropic consolidation test,ξdRelative rate of bond degradation typically ξ=[8-12]and ξd =[0.2-0.3]ViscosityN Strain-rate coefficient From curve fitting of tests at different strain and destructuration.For each set of tests,the experiments are firstly described followed by discussion on the estimation of model parameters.Simulations are made with model features switched on and off,to explore the relative importance of anisotropy and destructuration on the viscous behaviour as follows:(a)for the case referred to as EVP-MCC,the soil is assumed to be isotropic and only viscosity is considered (with α0=0,ω=0and χ0=0);(b)for the case referred to as EVP-SCLAY1,only viscosity and anisotropy are considered (with χ0=0);(c)for the case referred to as EVP-SCLAY1S,all features are incorporated.In addition,long-term oedometer tests and undrained triaxial creep tests were also simulated to examine the model performance on describing the creep behaviour of soft soils.3.1.Oedometer Tests on Saint-Herblain ClayRangeard [37]conducted a constant strain-rate (CSR)oedometer test on Saint-Herblain clay at two constant strain-rates (d εv /d t )by using an oedometric cell providing horizontal stress in addition to vertical stress [37].The clay properties are as follows:water content 87%,liquid limit w L =90%,plastic limit w P =48%,unit weight γ=14.9kN/m 3.The test was performed at d εv /d t =3.3×10−6s −1until εv reaches 12%,then changed to d εv /d t =6.6×10−7s −1until a vertical strain of 15.5%,and finally changed back to the initial strain-rate.The parameters λi ,κ,e 0in Table 2were determined from the results of oedometer tests on natural samples with χ0=0as the clay is slightly structured [37].The slope of the critical state line in compression (M c )was taken as 1.25from triaxial tests [20].For the simulations with anisotropic versions of the model p m 0=27.4kPa correspond to σ v 0=39kPa,whilst for the simulations assuming isotropy of the soil,Vol.24,No.3Zhen-Yu Yin et al.:Modelling Strain-Rate-Dependency of Natural Soft Clays·223·the value needs to be changed to p m0=35.7kPa(by Eq.(3)withα0=0).Poisson’s ratio was assumed to be0.3.The actual value of parameterωhas no influence on theεv-σ v curve of an oedometer test, because during one-dimensional consolidation no further rotation of the yield surface is expected[4]. The viscosity parameters N andμwere determined from the tests at two strain-rates byfixing Nfirst corresponding to the slope of theσ p-dε/d t curve and adjustingμfor the value ofσ p.Table2.Values of state parameters and soil constants for the clays simulatedParameters p (kPa)eν κλMωχξξNμ(s−1)St-Herblain27.4 2.260.30.0220.41 1.25----72×10−8Fig.5.CSR oedometer tests on Saint-Herblain clays.Experimental data versus simulations.3.2.Oedometer Tests on Batiscan ClayLeroueil et al.[15]presented a set of CSR oedometer tests on Batiscan clay with strain-rates varying between1.7×10−8and4×10−5s−1.The clay properties are as follows:water content80%,liquid limit w L=43%,plastic limit w P=21%,unit weightγ=17.5kN/m3,clay-sized particle content80%.The initial state was taken atσ v0=65kPa,a stress equal to the in situ vertical effective stress at depth of the specimens.The parametersλi,κ,e0andχ0in Table2were determined from the results of oedometer tests on natural samples,withλi=0.41determined at a strain level of20-25%.For the cases in which bonding and destructuration are ignored(χ0=0),the value ofλneeds to correspond to theλ-value of the natural clay(λ=1.64was measured from theεv-logσ v curves corresponding to a vertical strain range 3-9%).The slope M c was taken as0.98,as proposed by Rocchi et al.[19].The value of p m0was taken equal to50kPa corresponding toσ v0=65kPa.Poisson’s ratio was assumed to be0.3.Because results are available for only one stress path,the values ofξandξd were chosen as coupled values via curve fitting.The viscosity parameters N andμwere determined from the tests at two selected strain-rates by curvefitting byfixing Nfirst corresponding to the slope of theσ p-dε/d t curve and then adjusting μfor the value ofσ p.Figure6(a)shows that predictions by models(EVP-SCLAY1and EVP-SCLAY1S)are in good agreement with the experimental data:i.e.,the strain-rate effect on the apparent preconsolidation pressure is correctly represented.For thefinal part of this curve,the predicted preconsolidation pressure converges to a stable value for decreasing strain-rates(dεv/d t<10−9s−1in this case).The dependency·224·ACTA MECHANICA SOLIDA SINICA2011ofσ p on the strain-rate is similar to the strain-rate effect of the undrained shear strength from triaxial shear tests[11,13].Fig.6.CSR oedometer tests on Batiscan clays.Experimental data versus simulations.However,the results in Fig.6(b)show that accounting for destructuration is necessary for getting accurate predictions of the stress-strain behaviour over the whole stress range for such a soft sensitive clay.The parameters determined from strain-rate tests were examined by simulating long-term oedometer tests conducted on the same material by Leroueil et al.[15].The applied vertical stress varies from78to 121kPa.The permeability of soil is taken equal to1×10−9m/s corresponding to the full dissipation of excess pore pressure at t=1×106s for the applied stress levels.Figure7shows a good agreement between the experimental results and the simulations.It is to be noted that,although the model is based on the strain-rate-dependency phenomenon of soils,the creep phenomenon is also revealed.Fig.7.Long-term oedometer tests on Batiscan clays.Experimental data versus simulations.3.3.Undrained Triaxial Tests on Reconstituted Kawasaki ClayCSR triaxial tests with the axial strain-rate varying from0.7to0.007%/min were performed by Nakase and Kamei[12]on reconstituted Kawasaki clays.The clay properties are as follows:liquid limit w L=55.3%,plastic limit w P=25.9%,specific gravity G s=2.69,clay-sized particle content22.3%. The specimens were consolidated under the K0-consolidation until the vertical effective consolidationpressure of392kPa,and then subjected to undrained shearing in compression and extension.The tests were simulated fully including the consolidation stage(assuming one week for the whole consolidation stage).The values of parametersλi,κand e0were given by Kamei and Sakajo[38].The slope M c=1.64was obtained from triaxial tests.The value of p m0was assumed equal to one third of the preconsolidation pressure since the tests were performed on reconstituted samples and the fullVol.24,No.3Zhen-Yu Yin et al.:Modelling Strain-Rate-Dependency of Natural Soft Clays·225·consolidation stage was simulated.χ0=0was given as any initial bonding is considered to be destroyed during remoulding.The parameterωwas taken equal to77,obtained fromfitting the triaxial tests in extension.The viscosity parameters N andμwere determined considering two selected strain-rates. The values of the parameters are summarized in Table2.Figure8shows the comparison between the predicted and measured results.Both isotropic and anisotropic models well capture the compression behaviour.However,for extension tests,the model accounting for anisotropy(EVP-SCLAY1)improves the match in terms of the stress-strain and effective stress paths curves compared with the isotropic version of the model(EVP-MCC).The comparison demonstrates that even for reconstituted clays,it is important to consider the induced anisotropy due to K0-consolidation and subsequent loading in combination with strain-rate-dependency.Fig.8.Undrained triaxial tests at constant strain-rate on reconstituted Kawasaki clays.Experimental data versus simu-lations.3.4.Undrained Triaxial Tests on Haney ClayVaid and Campanella[11]conducted undrained triaxial tests at constant strain-rate varying from 0.0001to10%/min on undisturbed Haney clay.The clay properties are as follows:plastic limit w P=26%, liquid limit w L=44%,the preconsolidation pressure of340kPa and a sensitivity S t from6-10.All triaxial tests considered were conducted by initially consolidating the samples under an effective isotropic confining pressure of515kPa for36hours,and then allowing them to stand for12hours under undrained conditions before starting the shearing part of the test.Two triaxial tests at constant strain-rates were selected to calibrate parameters.The values of parameters forκ,λi,e0andν provided by Vermeer and Neher[17]were used.M c=1.28was estimated from triaxial tests.ωwas calculated fromλi(ω=15/λi)as suggested by Zentar et al.[39].The initial bonding amountχ0was determined from sensitivity byχ0=S t−1.The parametersξandξd were chosen as coupled values via curvefitting.The viscosity parameters N andμwere chosen from tests at two levels of constant strain-rate.Noting that the soil anisotropy can be modified during the isotropic consolidation stage,and the preconsolidation pressure depends on not only the applied maximum consolidation stress but also the duration of consolidation stage,p s m=410kPa andα=0were estimated at the end of the consolidation stage.Model parameters are summarized in Table2.Figure9(a)shows good performance of both models(EVP-SCLAY1and EVP-SCLAY1S)on simu-lating the effect of strain-rate on the undrained shear strength(maximal deviatoric stress q max for each strain-rate)in two phases:for the tests with small strain rates,the shear strength converges to a con-stant stress value representing the long-term strength of the material,while for tests with middle-level strain rates,the strain-rate effect can be represented by a linear relationship between the shear strength and the strain-rate in log scales.This predicted two-phase phenomenon agrees with the experimental observations.However,in Fig.9(b)simulations marked by EVP-SCLAY1,which have not considered destructura-tion,are unable to capture the strain softening phenomena for isotropically consolidated samples of soft sensitive clays.Fig.9.Undrained triaxial tests at constant strain-rate on Haney clays.Experimental data versus simulations.Fig.10.Undrained triaxial creep tests on Haney clasy.Experimental data versus simulations.The parameters determined from strain-rate tests were examined by simulating undrained triaxial creep tests conducted on the same material with the same procedure of the consolidation stage as tests at a constant strain-rate[11].The applied deviatoric stress q varies from193to329kPa(See Fig.10). The applied deviatoric stress q varies from193to329kPa.Although some discrepancies were found between predicted and measured results which is possibly due to the variation of natural samples for structured clays,the model accounting for destructuration well captured the undrained creep behaviour of soft sensitive clays in terms of strain and strain rate evolutions versus time:for low applied stress levels,only the primary creep stage(defined by a continuous decrease of the strain rate)occurs,while for higher applied stress levels,a tertiary creep stage(defined by an increase of the strain rate)can be seen after a short duration of a secondary creep stage(defined by a constant strain rate).The model performance in predicting the tertiary creep stage is due to the incorporation of soil destructuration (predictions by EVP-SCLAY1vs.predictions by EVP-SCLAY1S):for higher applied stress levels,if the effective stress reaches the critical state line,there will be no change in the size of the intrinsic yield surface;the viscoplastic deviatoric strain will continue to increase and cause a reduction in the size of the static yield surface towards the intrinsic yield surface,which in turn increases the amount of overstress p d m/p s m.As a result,it is possible to reproduce the so-called tertiary creep stage.Therefore, the models explained one possible reason of the tertiary creep:the destructuration of soft sensitive clays.3.5.Multistage Constant Strain-rate Triaxial Tests on Hong Kong Marine DepositThe triaxial compression and extension tests performed on Hong Kong marine deposit were presented by Zhou et al.[25]and Yin and Cheng[14].The clay properties are as follows:water content54.3%,liquid limit w L=57%,plastic limit w P=25%,unit weightγ=16.5kN/m3,clay-sized particle content30%. All the specimens taken from a depth of10m at a location near Tsing Yi Island werefirst saturated by standard increments of cell pressure up to205kPa and back pressure up to200kPa.After that,eachspecimen wasfirst consolidated under K0-condition with a vertical stress-rate of2.5kPa per hour,i.e., axially compressed with approximately zero radial strain with afinal effective confining pressure from50 to400kPa.Thefinal consolidation stresses were then held constant for24hours.After K0-consolidation, the cell pressure was held constant,and the specimens were sheared up to failure with step-changes in strain-rates,including unloading and reloading loops,either in compression or extension with strain rates varying from0.2to20%/h.The values for parametersλi,κand e0were determined from the results of oedometer tests on reconstituted samples[18,25].The initial amount of bondingχ0was measured from the consolidation stage of triaxial tests,as presented in Fig.11(a)using test E400.The slope M c was measured as1.24, and M e=0.94.The value of p m0was taken equal to46kPa corresponding to the in situ condition of samples.The coupled values forξandξd were determined by curvefitting the consolidation stage of test E400(at stress-rate of2.5kPa/h,see Fig.11(b)).The value of parameterωwas estimated via curve fitting of triaxial tests E400in extension.The viscosity parameters N andμwere determined from C400 (or E400alternatively)considering three different constant strain-rates(Fig.12).The selected values of model parameters,summarized in Table2,were used to predict triaxial tests C150and E150.Fig.11.Consolidation stage for triaxial test E400on Hong Kong marine deposit.Fig.12.Undrained triaxial tests C400/E400on Hong Kong marine deposit.Simulations by EVP-SCLAY1S versus exper-imental data for the calibration of parameters.Comparison between the measured results and the predicted ones was carried out for an undrained triaxial test in compression(C150)and in extension(E150).In Figs.13and14,results of3simulations are shown:EVP-MCC,EVP-SCLAY1and EVP-SCLAY1S considering axial strainεa vs.deviator stress, effective stress path in the triaxial parison between the prediction by anisotropic model EVP-。