Search

Discrete element method simulations are used to investigate the effects of particle shape on the plastic behaviour of sand. This is achieved by using four different shapes of crushable particles. The behaviours of the different particles when subjected to isotropic compression, triaxial shearing, and a range of stress path tests are compared, revealing the important role of particle shape on key aspects of behaviour. Yield surfaces are established and analysed with respect to different modes of deformation, and are also shown to depend on the particle shape. A simple state boundary surface is then obtained, which is confirmed by a performing an additional stress path test on a sample that had been sheared to a critical state.

In this paper, a shakedown analysis is conducted for both cylindrical and spherical cavities in cohesive-frictional materials subjected to cyclic internal pressures. Based on Melan’s lower-bound shakedown theorem, theoretical shakedown conditions for a two-layered cavity system are developed using the stress and strain solutions obtained from cavity expansion and contraction analyses. The shakedown limits are investigated in relation to the dimensions of the two-layered cavity system and the elastic/plastic material parameters. Numerical validation is conducted through a comparison of the stress path of the critical element and the failure lines of the materials. In addition, closed-form solutions for special cases, namely, the analytical shakedown conditions of cavities in an unbound cohesive-frictional material and in a bound material with different boundary conditions, are explicitly expressed. Finally, the shakedown solutions are applied to solve the thickness design problem of underground compressed air energy storage in lined and unlined caverns.

It is well known that the deformation and stress-resistant characteristics of fine-grained soils, especially soft clays, are significantly influenced by the soil softness. It is therefore very important to employ a model which can accurately simulate the effects of this phenomenon. A constitutive model must be able to create a balance among stress paths, the number of parameters, the process of parameter determination, and finally, the simplicity of the computational calculations. The current study investigates the performance of a two-yield surface (cone and cap yield surface) model for soft soils. The efficiency of the cap yield surface has been studied as well. The model has been calibrated by employing the data derived from previous researches for Bangkok clay. The incorporated data have been obtained from the results of CD triaxial, CU triaxial, and oedometer tests. The proposed method for the model calibration can accurately predict the triaxial test results and oedometer test stress path simultaneously. This method for predicting the soil behavior is based on the main soil characteristics taken from common soil mechanics tests. It can be widely employed for engineering practices, especially when complicated soil behavior is encountered.

As a modification of the deviatoric hardening plasticity model, a material state-dependent model was proposed to simulate the response of unsaturated sand containing gas bubbles under undrained triaxial conditions. Affected by the compressibility of gas, the stress paths under undrained conditions approach the drained response of sand when the initial degree of saturation is low. Upon an increase in the degree of saturation, the stress path gradually approaches the saturated undrained response. According to the prediction based on the second-order work criterion, static liquefaction occurs in loose sand, but not in dense sand. Increases in the degree of saturation and the initial gas pressure reduce the stress ratio at the instability points. The instability line obtained by connecting those instability points in the p-q space is nonlinear, and its slope depends on the initial void ratio, the initial degree of saturation, the initial gas pressure, and the confining stress. After comparing the experimental results in the literature with the theoretical prediction, the proposed model was shown to precisely predict the onset of the static liquefaction of unsaturated sand containing gas bubbles.

This paper reports on the small-strain stiffness (Gmax) of a well graded crushed rock reinforced with polypropylene fibres. The experimental study was conducted on twenty-five fully saturated samples subjected to bender element tests. The contents of fibre used were equal to 0%, 0.5%, 1.0%, 1.5% and 2.0%. The bender elements were inserted in a resonant column and a stress path triaxial apparatus accommodating samples of 70 and 50 mm in diameter, respectively, with a ratio of length to diameter equal to 2:1. The samples were initially consolidated isotropically to pressures of 50, 100, 400 and 700 kPa. At a given isotropic pressure, Gmax measurements were conducted increasing the deviatoric compressive stress (q) and keeping a constant mean effective confining pressure (p'). This allowed the study of the effect of stress anisotropy, expressed with the ratio (q/p'), and the role of fibre inclusion and content on Gmax of the reinforced samples under variable mean effective confining pressures. The data were analysed by means of the normalized shear modulus, Gmax,normalized, against the stress ratio, where Gmax,normalized expressed the increase of the small-strain shear modulus under an anisotropic stress state with respect to the corresponding modulus under an isotropic stress state. For a given confining pressure, the results showed a clear increase of Gmax,normalized with increased content of fibres which implied that the sensitivity of modulus to the stress anisotropy became more pronounced for greater contents of fibres. However, this influence of fibre inclusion and content on the small-strain stiffness of the samples was found weaker at greater magnitudes of p'.

Mixing a loose clean sand with random discrete flexible fibres has been found beneficial in decreasing the susceptibility to the phenomenon of liquefaction under monotonic loading. The addition of fibres can convert the strain softening response, typical of a loose unreinforced sand, into a strain hardening response by affecting the pore pressure generation and the effective stress path response. A new constitutive model based on the rule of mixtures has been used to simulate the undrained response of fibre reinforced sands. The model superimposes the individual contributions of the sand and the fibres according to their volumetric fraction. An apparent densification of the sand matrix induced by the presence of the fibres is accounted for in the model by assigning some of the void space to the fibres. This apparent densification is considered responsible for the observed strain hardening behaviour of reinforced sands. The proposed model is able to accommodate any distribution of fibre orientation: the orientation of fibres plays a key role in explaining the experimentally observed effective stress paths.

A stress path method for computing the settlement of normally consolidated clays is presented. The proposed method introduces the characterization of the stress-strain behavior that eliminates unnecessary experimental cost due to uncertainty in the design stress path. For arbitrary control of the effective stress during consolidation, a technique of back-pressure equalization, which enforces the immediate drained condition on the soil specimen without change of the effective stress state, was used. To account for nonlinear anisotropic response in computing the consolidation strains, the iteration procedure was designed. For the proposed method, the stress path experiments on the normally consolidated kaolin were conducted. An example illustrating the feasibility of proposed method in precisely calculating the settlement is shown for a circular footing on the clay stratum.

Ageing effects on the elasto-viscoplastic property of compacted moist cement-mixed granular material (GM) were evaluated by performing a series of non-standard drained triaxial compression (TC) tests. Two types of GM, crushed gravelly soil from a quarry and crushed concrete aggregate (i.e., a recycled material), were used. The specimens were produced by moist-compaction and then cured at constant water content under unstressed conditions for seven days. They were re-cured basically for two days under different stress states during otherwise drained TC loading at a constant strain rate. Yielding characteristics upon the restart of drained monotonic loading (ML) at a constant strain rate toward ultimate failure at the same or increased or decreased confining pressure were evaluated. The stress-strain behaviour before the stress state reaches the current yield locus is very stiff and highly reversible. Unlike elasto-plastic materials exhibiting no ageing effects, the yield locus expands during sustained loading at a fixed effective stress state due to not only yielding associated with creep deformation, controlled basically by the viscous property, but also ageing, controlled basically by time-elapsing. The shape and location of current yield locus depends on the location of the current stress state relative to the current ultimate failure envelope. The observed yield characteristics were analyzed based on a newly introduced interactive double-yield concept while in the framework of the non-linear three-component elasto-viscoplastic model that takes into account ageing effects as well as an interaction between ageing and inviscid yielding (and its potential decay by irreversible straining). The trends of stress-strain-time behaviour observed with the two types of cement-mixed GMs are essentially the same.

It is shown that the use of visco-plastic shear or volumetric strain as the stress history-independent hardening parameter in an elasto-viscoplastic model for sand may result in inaccurate numerical simulations of geotechnical boundary value problems. A new elasto-viscoplastic constitutive model for sand is proposed, formulated based on a stress path-independent irreversible (or visco-plastic) strain energy-based hardening function. The function was derived based on results from drained plane strain compression (PSC) tests on saturated dense Toyoura sand along a wide variety of stress path. The model is coupled with an existing isotropically work-hardening and -softening, non-associated, elasto-plastic model for sand. The constitutive model takes into account the effects of loading rate due to viscous properties on the stress-strain behaviour as well as those of pressure level, inherent anisotropy and void ratio and work softening associated with strain localization into a shear band. It is shown that the proposed model can much better simulate the effects of stress history on the deformation characteristics of sand than many previous models. The FEM code incorporating the model is validated by simulating physical PSC tests and bearing capacity model tests of a strip footing on sand performed by previous studies.

To investigate liquefaction properties of sandy soils, undrained cyclic loading tests are usually performed. However, it would be difficult to simulate fully the actual soil behaviour through the laboratory tests because the tested sample can be disturbed even though it is taken by in-situ freezing technique. In this study, by using three kinds of in-situ frozen sandy soils which were taken from Holocene and Pleistocene deposits and their reconstituted samples, their volume change properties were measured during freeze and thaw processes at different confining pressures of 30 kPa and 98 kPa. In order to investigate the effects of the possible sample disturbance on the liquefaction resistance, small strain characteristics were measured as well, which would reflect the soil structure. Decreases in the small strain characteristics and the liquefaction resistance were observed in case of the Holocene specimens that were thawed at the confining pressure that was lower than the in-situ overburden stress. On the other hand, in case of the Pleistocene specimens, the effects of the confining pressure during the thaw process on the small strain characteristics and the liquefaction resistance were small. Such contrastive feature between the Holocene and the Pleistocene samples could be linked with the difference in the types of their natural aging effects.

In order to investigate the effects of different geological ages on liquefaction properties of sandy deposits, a series of undrained cyclic triaxial tests was performed on three kinds of in-situ frozen and their reconstituted samples which were retrieved from Holocene (Tone-river sand) and Pleistocene (Edo-river B and C sands) deposits. The specimens were subjected to isotropic consolidation at a specified confining stress which is equivalent to the in-situ overburden stress at the depth of sampling, and small strain shear moduli were measured before and during the undrained cyclic loading tests. The liquefaction properties and the small strain shear moduli were affected by not only the natural aging effect of the specimen but also the inter-locking effect that was enhanced by applying drained cyclic loading before the undrained cyclic loading tests. During liquefaction, different tendencies of degradation in the small strain shear moduli which would reflect the aging effects of the specimen were observed between Tone-river Holocene sand and Edo-river B and C Pleistocene sands. The applicability of reconstituted samples as substitutes for in-situ frozen samples was confirmed with Tone-river Holocene sand that has no cementation effect between soil particles, whereas it seems difficult to simulate fully the liquefaction behaviour of Edo-river B and C Pleistocene sands which have higher cementation effect.

The influence of anisotropy on the elastic shear modulus of clays was addressed using test results obtained from a series of laboratory tests; i.e., triaxial test and Square oedometer. Under various stress stages imposing under triaxial condition, the deviator stress had an indiscernible influence on the elastic shear modulus. The path of vertically logged elastic shear modulus of clay could reasonably be expressed in terms of mean effective stress. Two pairs of bender elements were installed in the square oedometer so that shear wave velocities of samples which were trimmed parallel or perpendicular to their bedding direction can be measured in the vertical and horizontal directions. For sample trimmed parallel to the bedding direction (conventional one), the horizontal shear wave velocity was larger than the vertical one. The opposite observation, however, was observed for sample trimmed 90 degrees apart. The extents of different in vertically and horizontally logged shear wave velocities from both types of samples were almost similar. This clearly indicated the influence of inherent anisotropy rather than the stress induced. The mean effective stress can well represent the stage variable of elastic shear modulus.

The roles of dilatancy in granular materials are experimentally and numerically explored with regard to material instability and failure. Experimental results of sand corresponding to proportional strain paths show that either pre- or post-failure instability may take place, depending on the density and the imposed rate of volume change. Pre-instability strain softening is observed for both dense and very loose sand, however, strain softening is not a necessary condition for the occurrence of material instability. Both experimental results and constitutive modeling show that the difference between the forced rate of volume change and the inherent potential of dilation dominates the occurrence of strain softening and material instability. The dependency of inherent dilation on void ratio and stress level is a key factor in predicting the complicated stress responses of sand deforming along imposed strain paths.

Two-dimensional (2D) and three-dimensional (3D) model tests and the corresponding elastoplastic finite element analyses were carried out to investigate the influences of 3D effects, wall deflection process and wall deflection mode on the earth pressures and the ground movements in retaining wall problems. Aluminum rod mass were used in 2D model tests, and alumina balls were used in 3D model tests. The stress-strain behavior of these materials is similar to that of dense sands with positive dilatancy. In the finite element analyses, a recently developed elastoplastic constitutive model, named subloading tij model, were used. This model can describe typical stress deformation and strength characteristics of soils such as the influence of intermediate principal stress, the stress path dependency of plastic flow and the influence of density and/or confining pressure properly. The test results show that the earth pressures on the retaining wall in 3D condition are much smaller than those in 2D condition, and these distributions of earth pressure are more influenced by the wall deflection process than by the wall deflection mode. Observed surface settlements at the backfill just behind the wall in 3D condition are larger than those in 2D condition, but 3D surface settlements occur more locally. These 2D and 3D surface settlements are influenced by the wall deflection process more than the wall deflection mode. However, at the same wall deflection process and the same wall deflection mode, there is not much difference in the shapes of the earth pressure distributions and the surface settlement troughs between 2D and 3D conditions. These differences and similarities of the earth pressures and the surface settlements are simulated not only qualitatively but also quantitatively in the analyses using the above constitutive model.

The conventional elastoplastic model premising that the interior of yield surface is a purely elastic domain is incapable of describing the plastic deformation by the rate of stress inside the yield surface. Thus, it is inapplicable to the description of cyclic loading behavior. Besides, the traditional elastoplastic model is independent of the stress rate component tangential to the yield surface. Therefore, it predicts an unrealistically high stiffness modulus for nonproportionalloading process deviating significantly from proportional one. The extended tangential-subloading surface model proposed by Hashiguchi and Tsutsumi (2001) would be capable of describing the cyclic loading behavior and the inelastic strain rate induced by the stress rate component tangential to the subloading surface. In this article, the extended tangential-subloading surface model is applied to the prediction of deformation behavior of sands subjected to various loading ranging from proportional to cyclic nonproportionalloading. The validity is verified by comparing with the various test data. Then, it is revealed that the incorporation of the strain rate due to the stress rate component tangential to the subloading surface is of importance for the description of nonproportional loading behavior., 論文

In this paper we will firstly consider the range of validity of Taylor's “stress-dilatancy” hypothesis. From this study we conclude that Taylor's rule is strictly valid only for coaxial deformations. For smooth non-coaxial deformations Taylor's rule must be modified according to the rule proposed by Gutierrez and Ishihara (2000), which includes a “non-coaxiality” correction factor. Strictly speaking the stress-dilatancy rule breaks down if in a process an abrupt rotation of principal axes is imposed, as this is the case at the onset of localization. During this phase the corresponding stress function decreases monotonously, whereas the dilatancy oscillates between large negative and large positive values. Finally it is shown that in globally undrained, displacement controlled tests this incipient post-failure contractancy to dilatancy oscillation should result in a dynamic instability.

A series of strain controlled undrained true triaxial tests on cubical specimens have been performed using a fully automated flexible boundary experimental setup, with real-time feedback control system to evaluate the three-dimensional mechanical behavior of kaolin clay. This paper concentrates on the comparative response of the soil elements under monotonic shear loading observed at various overconsolidation levels. A possible shape of the initial yield surface is explored based on the limit of purely elastic deformations. The discussion includes the influence of the relative magnitudes of principal stresses on the clay behavior observed at different overconsolidation ratio values. Failure conditions are shown to be governed by the onset of localizations developed within the specimens, and because of which the specimens show sudden failure response before reaching a perfectly plastic deformation conditions. A reference surface constraining the ultimate plastic yielding of clay is found to be different from the observed failure surface. Based on the experimental observations, a 3-D failure criterion is developed that grows in size as a function of pre-consolidation stress.

An isotropic hardening elastoplastic model, named subloading tij, has been recently proposed (Nakai and Hinokio, 2004). Three features differentiate subloading tij model from the conventional ones: (a) the use of a modified stress space given by tensor tij; (b) the split of the plastic strain increments in two components and (c) the use of two yield surfaces based on the concept of subloading. These three characteristics greatly improve the prediction capabilities of the model, with respect to those of the well-known Cam-clay model. However, the model formulation and implementation becomes a little more complex. In this paper the basic equations for the evolution of stresses and strains and for the evolution of the internal variables that control the size of the yield surfaces of the model are reformulated in such a way as to resemble those of any conventional elastoplastic model. This facilitates significantly the numerical integration (explicit or implicit) of the corresponding system of differential equations of the model. The present formulation also identifies the physical meaning of each concept and clearly shows where they intervene in the deduction of the elastoplastic constitutive tensors. The model is initially formulated without taking into account the plastic strain increment split, in order to emphasize the effect of adopting the subloading concept. Then the plastic split is considered using a different approach that allows writing the constitutive tensors in a simpler manner. The procedures proposed were successfully tested and all derivatives and algorithms necessary to implement the model are given in the appendices. The authors hope that these procedures make it easier for other researchers to implement and use the model as it is or to use the basic concepts in any other model.

The roles of dilatancy and fabric on the behaviour of granular materials are both numerically and experimentally explored for the study of material instability and failure. This investigation has two basic ingredients: namely a stress dilatancy model with microstructural information embedded through a fabric tensor, and an experimental rendition of force transmission and structure in an assembly of 2-D photoelastic disks. In order to highlight material instability, model simulations of sand behaviour are carried out in axi-symmetric stress conditions along proportional strain paths with varying degrees of controlled dilation (or compaction) including isochoric deformations as a special case. It is shown that sand, otherwise stable under isochoric (undrained) conditions, can actually succumb to an instability or a liquefaction behaviour under other loading paths. This suggests that flow type of failures in soils may not be necessarily restricted to the classic saturated loose sand case in undrained conditions, but could manifest itself under other conditions as well.

A series of triaxial compression (TC) tests were performed on Toyoura sand to evaluate the viscous effects on the stress-strain behaviour following the shear yielding mechanism. It is shown that very different shear stress and shear strain relations obtained from TC tests performed along different stress paths become rather unique when converted into relationships between the stress parameter Xst = (q/pa)/(p ' /pa)βsf, which is deemed to control the shear yield mechanism, and the normalized irreversible shear strain energy κs , which is the stress path-independent strain hardening parameter for shear yielding. Along each inviscid yield locus quantified based on the test results, with an increase in the effective mean principal stress p', the deviator stress q increases at a relatively large rate while the stress ratio q/p' decreases at a relatively small rate. According to the non-linear three-component model described in the paper, the stress parameter Xst is decomposed into the inviscid and viscous components, Xsf and Xsv. The inviscid shear yield loci are described in terms of Xsf and develop with an increase in Ks. The viscosity function g v κ s is introduced to relate Xsv to Xsf in the form of X s v = X s f ⋅ g v κ s based on the test results. It is shown that the viscous effects on the Xst − κs-Ks relations obtained from the TC tests performed along a wide variety of stress path could be simulated rather well by the model taking into account the decay of the viscous component Xsv with an increase in κs.

International audience; The object of this study was to understand and characterize dilatancy in clay in relation to the overconsolidation ratio. This phenomenon was investigated along a large range of loading paths, with strong emphasis placed on constant mean stress paths given that it was possible to measure the volumetric strain created by the sole deviatoric stress. Triaxial tests performed on remolded clay (Kaolinite P300) allowed us to determine three different types of behaviour for specimen submitted to a deviatoric stress: no volume change, dilatancy and contraction. In the (p', q) plane these domains are limited by the maximum strength envelope zeta, which coincides with the perfect plasticity line M for OCRs < 2. The experimental results were then compared to plastic flow theories used in the Cam Clay model and in Rowe's dilatancy theory. Finally, by measuring the plastic strain increment vectors along different stress paths in the (p', q) plane, we demonstrated that the uniqueness hypothesis of the plastic potential is not valid. The results could be explained by using two plastic strain mechanisms: a deviatoric one and an isotropic one.

An isotropic hardening elastoplastic model for soil, which can describe typical deformation and strength behavior of normally and over-consolidated soils in general stress conditions, is presented. This model can take into consideration the influence of density and/or confining pressure on the deformation and strength characteristics of soils. Further, it also takes into account the influence of intermediate principal stress on the deformation and strength of soil and the stress path dependency on the plastic flow, as did other previous versions of this model. For considering the intermediate principal stress on the deformation and strength, the concept of modified stress tij is employed in the same way as the previous models. With respect to the subloading surface concept, the previous models are revised and extended into one in which the influence of density and/or confining pressure on the deformation and strength of soil can be considered. Furthermore, the influence of stress path dependency of the direction of plastic flow is considered by dividing the plastic strain increment into two components – i.e., a component satisfying associated flow rule and a component of isotropic compression. In the present model, only one material parameter a for representing the influence of density and/or confining pressure is added to the parameters of the previous model, which are fundamentally the same as those of the Cam clay model. The validity of the present model is checked by monotonic and cyclic loading tests not only on normally and over-consolidated clay but also on loose and dense sands in three-dimensional stresses.

Effects of viscous properties on the yielding characteristics of sand under general stress conditions were evaluated by performing unconventional drained triaxial compression tests on saturated dense Toyoura sand. Shear yield stresses evaluated by applying unload/reload cycles of deviator stress at a fixed confining pressure were noticeably affected by a strain rate change between primary loading and reloading as well as creep deformation at the maximum deviator stress before unloading due to the viscous properties of sand. It is shown that the double yielding concept, describing the yielding of soil in terms of shear and volumetric yielding mechanisms, is relevant to sand. The shear yield loci on the q—p' stress plane were obtained by decreasing and then increasing the deviator stress respectively at different confining pressures. The shape of shear yield loci was also affected by the viscous properties. With an increase in the effective mean principal stress, p', along a given inviscid shear yield locus, the deviator stress, q, increases while the stress ratio, q/p', slightly decreases. It is demonstrated that it is necessary to incorporate the effects of viscous properties when developing a realistic constitutive model for sand. An interaction between shear and compression yielding mechanisms is discussed based on experimental results.

Effects of wet conditions (air-dried or saturated), pressure level and dry density on the viscous properties of sand were evaluated by performing unconventional triaxial compression tests on Toyoura sand. Effects of stress path were studied to a limited extent. During otherwise monotonic loading at a constant strain rate, the strain rate was changed stepwise and/or creep loading was performed after different strain rate histories. The test results were analysed in the framework of a three-component model, called the TESRA model, that had been developed based on the results from plane strain compression tests on dense Hostun and Toyoura sands at a confining pressure of 400 kPa. The manner of viscous stress change by a change in the irreversible shear strain rate observed under the various test conditions could be represented by the same viscosity function representing the viscous properties. The viscous stress decayed with an increase in the irreversible shear strain in all the tests, while the rate of decay was noticeably larger with air-dried sand than with saturated sand. The TESRA model using the same viscosity and decay functions simulates very well the effects of viscous properties observed in the whole tests.

At present, finite-element methods are frequently used in the analysis of geotechnical engineering problems. The selection of an appropriate constitutive law primarily involves balancing simplicity with accuracy. Given that many practitioners still use the hyperbolic model, the adequacy of a modified version of this simplified stress-strain relation-ship to predict the nonlinear behavior of granular soils, is herein examined. Drained triaxial tests were performed on specimens composed of Maipo River sand at different relative densities. These specimens were subjected to different stress-paths in order to an extensive comparison of the measured strains and the predicted values. In addition, laboratory plate-load tests were conducted on rough, circular and strip, surface footings, which were 100-mm in diameter and 60-mm wide, respectively. The testing box, with dimensions 1.05 m × 1.05 m × 0.65 m, was filled with air-dried sand by using a raining apparatus. A field plate-load test performed on dense sandy gravel was also analyzed. From the good agreement achieved between the empirical observations and both the calculated strains and load-settlement relationships, it is concluded that the proposed hyperbolic model predicts with sufficient accuracy the nonlinear response of granular soils in most practical cases, as long as the soil mass is not close to failure.

The applicability of the SHANSEP method was examined for six different natural clays, namely Louiseville (Quebec, Canada), Bothkennar (Scotland, UK), Yamashita (Yokohama, Japan), Amagasaki (Japan), Rakusai (Kyoto, Japan) and Singapore. All of these clays are somewhat overconsolidated and their OCR range between 1.9 and 5.0. It was confirmed that post-depositional stress change had taken place for the latter three clays, although their OCRs could not be fully explained only by the stress history. Some or all parts of the overconsolidation for the objective clays may have been created by reasons other than the stress change, such as ageing or cementation. The undrained shear strength (su) of these clays was examined using the SHANSEP method and compared with that obtained using the recompression method. All the soil samples were recovered using the Japanese standard fix piston or equivalent samplers, and strength evaluation was done using the triaxial and the direct shear apparatuses. It was found that the stress-strain relations as well as stress paths are somewhat different between the SHANSEP and the recompression methods : that is, larger strain and smaller internal friction angle at failure for the SHANSEP method. However, difference in su measured by the two methods was surprisingly small, the ratio ranging between 0.9 and 1.1. This research has led us to the conclusion that considering the variation in sample quality in practical soil investigation, the SHANSEP method is a quite useful method in evaluating the design undrained shear strength of a soil, which is free from sample disturbance. It should be noted, however, that the yield consolidation stress, which is an important soil parameter in SHANSEP method, is affected by sample quality.

A new devise is developed for conducting a strain-path controlled test with a conventional triaxial test apparatus. By controlling lateral stress during axial strain controlled test, this devise can keep the principal strain increment ratio, Re , defined as a ratio between axial and lateral strain increments, constant. Toyoura sand specimens at relative densities of 55% and 80%, consolidated under four different initial confining stresses, are axially loaded with a Re -value between — 0.5 and 1.0. The specimens are then unloaded to the initial stress states and reloaded with different Re -values. The test results during compression show that the principal stress ratio, K, converges at a constant value. The converged K- value is strongly dependent on the Re -value but insensitive to the strain history and the initial stress state. There is a well-defined relationship between the principal stress ratio K and the total strain ratio Re,t where the K-value decreases with decreasing Re,t-value. The K-Re,t relation is independent of its strain path and uniquely expressed in terms of the internal friction angle and the dilatancy angle of the sand. Based on the K-Re,t relationship, a simple equivalent linear constitutive model for axisymmetric condition is developed. Two series of stress-controlled drained triaxial compression tests are performed, and the results are compared with those simulated by the model. The computed results show excellent agreements with the test results, indicating that the proposed constitutive model, although based on the equivalent linear concept, is promising for simulating a stress-strain relation including dilatancy behavior of sand in the stress-controlled triaxial test.

Undrained anisotropy, together with the undrained response to monotonic as well as cyclic rotation of principal stress directions, of an isotropically consolidated loose Ham River sand has been investigated by performing three series of hollow cylinder tests. The initial anisotropy of the sand when subjected to a single consolidation pressure was successfully established in the first series of tests, each using different, but fixed in each test, values of α (i.e., the direction of major principal stress relative to the vertical deposition direction) at different b values (i.e., the relative magnitude of intermediate principal stress). The response against monotonic principal stress rotation was also examined in the second series of tests, in which the total stress path defined using the shear stress, t, and α was varied in various manner with b fixed at 0.5. In the third series of tests, the response to cyclic changes of both α and b was examined, and the results were interpreted with the comparative behaviour of the monotonic loading tests with different combinations of α and b. It was found that the sand’s undrained anisotropy and cyclic response could be interpreted by invoking the concept of undrained local boundary surface (LBS). The four-dimensional LBS could be visualized in (t, p′, α) space as a family of superimposed three-dimensional shapes with different b values.

A series of K0-consolidation tests in a triaxial cell was carried out for marine clays collected from different areas in the world in order to investigate the behavior during K0-consolidation tests and evaluate K0-values that are required for laboratory tests with recompression techniques, FEM analysis using elasto-plastic constitutive equations, etc. It was confirmed that the famous equation K0NC = 1–sin ϕ′ is useful; however, the definition of ϕ′ should be concerned. In the case using ϕ corresponding to a critical state line, i.e. (q/p)max, the equation underestimates K0NC by 0.05; in contrast, in the case using ϕ′ corresponding to a peak strength, i.e. (q)max, the equation overestimates K0NC by 0.05. This study proposes a procedure for estimating the in-situ K0-value. For a representative undisturbed clay sample, K0-consolidation and K0-overconsolidation tests (i.e. SHANSEP test) are consecutively carried out to obtain the K – OCR relationship. Using the profile of OCR obtained from a series of oedometer tests, the K-value corresponding to the OCR can be calculated as the in-situ K0. The K0-values obtained from both the K0-overconsolidation test corresponding to the OCR in the laboratory and the fiat dilatometer test (DMT test) in the field compared well.

This study examines the phenomenon of settlement induced by traffic loading on low embankment roads built on soft cohesive soils deposited in a reclaimed lowland. A series of cyclic triaxial tests were conducted on undisturbed soft Ariake silty-clay under undrained and partially-drained conditions. A model based on the test results is proposed to simulate the partially drained behavior of silty-clay under a variety of lateral stresses. It is shown that the maximum excess pore pressure generation ratios and volumetric changes increase until the cyclic stress ratio reaches a value of 0.73, which is independent of the effects of lateral stresses under the same anisotropic consolidation ratio. Numerical analyses applying the proposed model were performed to compare the effects of soil thickness and cyclic loading time during one day and one cycle on partially drained cyclic behavior. It was found that a large ratio of cyclic to non-cyclic load­ing time during one cycle exerts a smaller influence on volumetric strain than the total time of cyclic loading during one day. Observed settlement results of a low embankment road are analyzed to verify the applicability of the proposed model.

In order to study stress-strain-strength and yielding characteristics of lime treated clay, an extensive testing program was conducted on lime treated clay. It was found that the main effect of lime treatment was to change the soft clay from normally consolidated to overconsolidated behavior. Heavily overconsolidated characteristics were observed for stress states inside the volumetric yield locus obtained from anisotropic consolidation tests. The volumetric yield loci of lime treated clays were found to be more pronounced than distortional or strain path yield locus. Outside the volumetric yield locus, the behavior was found to consist of an initial pseudo-elastic phase followed by a stage where the behavior appears to be similar to that of a work-hardening plastic material as the stress path proceeds towards the curved failure envelopes. The treated clays strain-softened after failure with the residual stress states lying close to the critical state line of the untreated clay. A conceptual model to describe the behavior of lime treated clay is introduced, in which the presence of distortional yield locus shifts with lime content and curing time.

A micromechanics model is used to analyse the stiffness anisotropy of soils at small strains. Five material constants for a cross-anisotropic elastic material are related to micromechanics variables such as fabric anisotropy, contact stiffness, particle radius, and the number of contacts in a given volume of particulate assembly. The analytical results from the model are compared with the published experimental data on small-strain stiffness anisotropy in order to estimate typical soil fabric conditions of sands and clays. The relationship between the small-strain shear modulus obtained from triaxial tests and shear tests is examined using the micromechanics model. The analysis shows that, when a soil is stiffer in the horizontal direction, the shear modulus evaluated from the conventional triaxial drained tests underestimates Gvh and Ghh. The opposite is true when a soil is stiffer in the vertical direction. When a soil is sheared in undrained condition, the measured shear modulus is closer to Gvh than Ghh, especially when the soil is stiffer in the horizontal direction. The effect of soil anisotropy on the stiffness measured from different stress paths in triaxial condition is investigated.

The use of a compressible layer such as expanded polystyrene blocks behind a rigid retaining wall and geogrid layers embedded in a dense granular backfill is examined as a reinforcement technique for retaining wall structures. The mobile model retaining walls adjacent to reinforced model specimens are subjected to different surcharge pressures, and are caused to move laterally to measure the lateral earth pressure during the wall movement. The coefficients of earth pressure at rest and active earth pressure are carefully inferred from test results. Three series of tests are conducted; one test series with expanded polystyrene blocks installed behind the wall, another with geogrid layers embedded within model specimens, and the last series with expanded polystyrene blocks installed behind the wall and geogrid layers fixed between two adjacent expanded polystyrene blocks and embedded within model specimens. The reductions in the earth pressure at rest and the active earth pressure due to various patterns of reinforcement are interpreted in relation to the concept of controlled yielding of compressible expanded polystyrene blocks, tensile strains induced along geogrid layers, fixity between expanded polystyrene blocks and geogrid layers, and a facing unit consisting of expanded polystyrene blocks.

Monotonic and cyclic loading tests have been carried out on a silica sand over a wide range of stresses in order to compare non-crushing and crushing behaviour. Samples were sieved at several stages of testing to theoretically determine the increase in particle surface area and hence degree of crushing. Undrained shear behaviour was compared for crushing and non-crushing regions above 3 MPa. Samples consolidated to 0.1 MPa demonstrated strong dilative behaviour, while above the yield stress of 3 MPa dilation was suppressed and considerable particle crushing occurred. Shearing caused a marked increase in particle crushing particularly after the phase transformation point. Crushing at the steady state was similar for isotropic and anisotropically consolidated sands. At high confining pressures the cyclic strength curves were similar to those for loose sands except for confining pressure dependency due to particle crushing. For low confining stresses cyclic strength increased with initial stress ratio, while for high confining stresses it decreased with initial shear stress ratio. In the cyclic tests there was no significant crushing for 0.1 MPa. Crushing was seen to increase rapidly after the phase transformation point, where high strains developed and where particle rotation and translation contributed to the crushing process.

In very loose sand, the ratio Mp of shear stress to mean normal stress at the peak point of the undrained effective stress path (UESP) is very close to the stress ratio Mat the peak point of the capped yield surface. Stress ratios Mp can therefore be used in constructing yield surfaces of sands. These stress ratios have also been used in the past in evaluating flow potential of loose sand. Application of Mp for these purposes requires that factors affecting this stress ratio, and quantitative relationships for the variation of Mp with these factors be determined. In this paper, effects of the intermediate principal stress and direction of loading on Mp are investigated, and models are developed by which these effects can be quantified. It is shown that variations of Mp with these factors are similar to the variations of yielding stresses obtained from stress-strain data. Yield surfaces obtained from the variation of Mp indicated a strong dependency of yielding stresses on inherent anisotropy. Data examined in this paper also suggest that the effects of inherent anisotropy on yielding stresses are controlled primarily by the relative magnitudes of the normal stresses applied in the principal directions of material anisotropy.

This paper describes a geotechnical survey evaluating the shear strength of Pleistocene clay in Osaka Bay. In the stability evaluation for a seawall structure of Kansai International Airport, as the slip circle giving the minimum safety factor possibly passes through the Pleistocene clay layer, the stability could be greatly affected by the shear strength profile of the Pleistocene clay. The result of the stability analysis is very sensitively affected by the undrained shear strength profile, which is roughly estimated based on a ratio of undrained shear strength to consolidation yield stress (s u /py) of natural clay deposit (Mesri, 1975; Tanaka and Tanaka, 1994) and the average apparent overconsolidation ratio (OCR = py / σ′v0) at this site (Horie et al., 1984). In this study, in order to evaluate the undrained shear strength obtained by typical testing methods, UC (unconfined compression) test, UU, CU, K0CU compression and extension triaxial tests and DS (direct shear) test were carried out for undisturbed samples of the Pleistocene clays collected for the second phase construction of Kansai International Airport. Mechanical behaviors and shear strength profiles obtained by each test were compared and discussed. The undrained shear strength and its scattering depend on the testing methods. The recompression method, in which a specimen is consolidated under the in-situ stress, gives many reliable test results. The UC and UU tests, in which a specimen is not consolidated, give very scattered and unreliable test results. From the comparison between undrained shear strengths for design determined by each test with correction factors, it is found that the strength of the DS test is smaller than those of K0CU and CU tests. This fact is derived from the relatively larger strength anisotropy of the Pleistocene clay. The undrained shear strength increase ratio with depth Δsu/Δz for design is obtained as 2.09 kPa/m, considering strength anisotropy.

A site investigation program was carried out at the Yamashita site, Yokohama, Japan, by means of laboratory as well as field tests. It was found from the investigation that the clayey layer is overconsolidated with an OCR of about 2.0. It is evident from geological consideration that this overconsolidation may not have been caused by stress change, but was probably due to structure effects such as cementation or delayed consolidation. The physical and mechanical properties of the Yamashita clay is compared with those of Louiseville clay, a Champlain Sea Clay widely distributed in Quebec, eastern Canada, and a well known cemented clay. Louiseville clay was retrieved by Japanese sampler and transported to the Port and Harbour Research Institute (PHRI) for laboratory investigation. Comparison of the two clays shows that strength properties such as the stress-strain curve or stress path for both clays are very similar to each other. However, the pattern of the e-log p curve measured by Constant Rate of Strain oedometer, particularly beyond the yield consolidation pressure, is remarkably different between the two clays.

Several studies have concluded that the effect of matric suction on the shear strength of unsaturated soil is very important. However, little attention is given to the stress history due to unsaturation and drainage condition in the arrangement of these experimental results. This paper attempts to show the relationship between shear properties and matric suction under a specified stress history and drainage condition. In this study, an unsaturated specimen is set up by dehydrating the normally consolidated soil with controlled matric suction to a specified stress history. Triaxial apparatus and hollow cylindrical torsional shear apparatus are employed to examine the effects of stress history and drainage condition, respectively. Through systematic experiments, the effects of stress history and drainage condition on the stress-strain relationship are clarified. The increase in shear strength by matric suction which is measured at the point of maximum shear stress is observed to be independent of stress history and drainage condition. However, in strain softening behavior it was found that the ultimate strength is independent of matric suction and traces a unique line which is parallel to the critical state line on the pnet-q plane.

A series of undrained triaxial compression tests was performed on Toyoura sand in order to investigate the behavior of sand under large deformation. The present study focuses on the effects of anisotropic consolidation on the undrained behavior of sand. A wide range of initial states of sand is covered and taken into account with the behavior of sand varying from contractive to dilative. Different states of consolidation stress were shown to affect the stress-strain behavior of sand and the development of excess pore water pressure up to an axial strain of 5%. Beyond a strain in excess of 10%, the behavior of sand was shown to become independent of the stress state at consolidation. Consequently, the relation between void ratio and confining stress at steady state and quasi-steady state are independent of the extent of anisotropic consolidation. Moreover, the initial dividing curve between dilative and contractive behaviors in an e ~ p' diagram was shown to move down as the sand is more anisotropically consolidated.

A series of consolidated-undrained torsion shear tests was performed on large, hollow cylindrical specimens of undisturbed San Francisco Bay Mud to investigate the influence of stress rotation and changes in b-values on the stress-strain, pore pressure, and strength characteristics of a cross-anisotropic, natural clay. The specimens were first K0-consolidated and then sheared along various undrained stress paths to study the clay behavior in the full range of stress rotation from 0° to 90° relative to vertical. For this cross-anisotropic clay, the test results indicate that the major principal strain increment directions in physical space do not coincide with the directions of major principal stress at failure. The deviation between these two directions varied continuously and the maximum deviation was measured at approximately 20°. The normalized, undrained shear strength decreases systematically with inclination of the major principal stress at failure, ψ, from triaxial compression with ψ = 0° and b = 0 to triaxial extension with ψ = 90° and b = l. The effective friction angle remains constant at approx. 35° in the range of ψ from 0° to 30° after which increases gradually to almost 50° at ψ = 90°. These results are unlike those obtained for a laboratory prepared deposit of kaolin, which tended to behave more like an isotropic material despite the initial A0-consolidation. Thus, effects of aging are very important in the behavior of clays, and it may not be possible to simulate such effects with sufficient accuracy and reliability with laboratory prepared clay deposits.

Pile foundations in a soft ground will usually be subjected to lateral cyclic loading during earthquakes. In a major earthquake, it is reasonable to think that the mechanical behavior of the pile foundation and the surrounding ground is nonlinear. In recent years, the limit state design method has become predominant in the design of foundations for railway bridges and other structures. Investigations of the mechanical behavior of pile foundations subjected to lateral cyclic loading up to the ultimate state, therefore, are very important in providing evidence for the design method. In this paper, field tests on a pile foundation, composed of cast-in-place reinforced concrete piles and subjected to one-side cyclic lateral loading up to the ultimate state, are simulated with the three-dimensional elasto-plastic finite element analysis (DGPILE-3D). In the numerical analysis, particular attention is paid to the stress-strain relation of the soil which, to the author’s point of view, plays a dominant role in the mechanical behavior of the pile foundation. Such constitutive models as the Drucker-Prager model, Cam-clay model, and tij model are adopted for the analysis in order to find the differences in the results due to the application of different constitutive models. Based on the analysis, the authors try to provide an applicable numerical way of evaluating the mechanical behavior of a pile foundation subjected to cyclic lateral loading at the ultimate state.

In order to evaluate the effects of increasing cementation on the mechanical properties of lightly cemented soils, artificially cemented clay was prepared and a series of standard consolidation tests, undrained triaxial compression tests and constant mean effective stress tests was performed. Based on the experimental considerations, the effects of light cementation on the stress-strain behaviour and strength property are discussed. In addition, based on the theoretical considerations within the framework of the critical state concept, several basic concepts for the development of a constitutive model for lightly cemented clays are presented. The special characteristic of the presented model is the introduction of an internal variable which controls the effects of cementation. The following conclusions are obtained: 1) The failure state of lightly cemented clay may be determined in the balance of cementation, void ratio and stress condition. 2) The failure state line in the p — q space is parallel to that of an uncemented clay and has a certain intercept pr, which characterizes the cementation effect. The slope of the failure state line in the e—In (p + pr) space becomes steeper with any increase in cementation effect. 3) A constitutive model incorporating cementation is proposed. Introduction of a soil constant pr is effective in estimating the mechanical properties of lightly cemented clay.

The effects of tube sampling disturbances on undrained shear properties of reconstituted samples of coastal soils were investigated. The values of undrained shear strength (su), initial tangent modulus (Ei) and secant modulus (E50) were reduced while axial strain at peak deviator stress (ɛp) increased because of disturbance caused by the penetration of samplers. Disturbances also considerably reduced initial effective stress (σi') and pore pressure changes. The changes in soil parameters between the “in-situ” and “tube” samples depended markedly on the characteristics of the tube samplers. Values of su, Ei E50 and σi' decreased while the values of ɛp increased with increasing area ratio (or decreasing external diameter to thickness ratio) of the samplers. It was also found that the reductions in the values of su, ɛp, Ei, E50 and σi' increased with decreasing plasticity of the soils. A number of reconsolidation procedures, both isotropic and anisotropic including SHANSEP methods, were adopted in order to assess the suitability of reconsolidation of “tube” samples to recover the “in-situ” properties of the soils. It appeared that reconsolidation using SHANSEP procedures may not be applicable to these coastal soils. However, K0-reconsolidation to “in-situ” stresses provided a better estimate of “in-situ” properties than other reconsolidation techniques.

Sample quality was studied at three sites; Ariake (Japan), Bothkennar (Britain) and Drammen (Norway), using various types of samplers. It was found that the unconfined compression strength (qu) is significantly affected by the sample quality. Sample quality obtained by the Japanese standard (JPN) sampler is the same as that by the Laval or the Sherbrooke samplers, which have a worldwide reputation for their ability to get high quality samples. However, there is a remarkable difference in qu for the lean clay layer of the Drammen clay between the JPN and the Sherbrooke samplers. It is concluded from a series of soil investigations that the reduction of the qu value may be caused by two different reasons: loss of the residual effective stress and destruction of the soil structure. In most of the previous studies, these two phenomena were thought to take place concurrently. When the structure is kept unchanged and only the residual effective stress is lost, the recompression technique, where a specimen is consolidated under the same stress level as in situ, is effective in producing the in situ soil behavior. However, if the soil structure is destroyed and its e—log p curve is different from the original one, the recompression technique is not capable of providing real soil behavior.

Undrained cyclic torsional shear and triaxial tests were performed on hollow cylindrical specimens of dense Toyoura sand, which were consolidated isotropically after preparation by air-pluviation and saturation. At several stress states, small strain quasi-elastic deformation properties were measured by applying very small amplitude cyclic torsional and vertical loads. During the liquefaction process, the observed degree of degradation of quasi-elastic shear modulus and vertical Young's modulus, was similar to that measured during isotropic consolidation. This suggests that the soil structure was damaged during the process of liquefaction. A larger extent of degradation was observed under triaxial extension or torsional shear than under triaxial compression condition. Such a change in the quasi-elastic deformation properties should be properly considered when analyzing the liquefaction process of sands based on elasto-plastic modeling. The behaviors observed during small amplitude cyclic vertical loading under undrained condition could be explained by considering inherent and stress state-induced anisotropy in the modeling of quasi-elastic deformation characteristics, and by correcting for the effects of membrane penetration. The latter correction was necessary, although the involved strain level was as small as 0.001%.

Plane strain compression tests were performed on large specimens that were either unreinforced or reinforced with 6 or 11 layers of geogrid, both 57.0 cm in height and 24.4 cm×21.4 cm in cross-section. It is shown that the effects of covering ratio for each grid layer is much more important than the total tensile stiffness of grid within the limits of the test conditions in this study. Numerical analysis of the test results by a plane strain non-linear elasto-plastic FEM was performed considering strain localisation as well as anisotropic stress-strain behaviour of sand and interface properties. The geogrid was modelled as a planar reinforcement. Not only the pre-peak stress-strain behaviour of the unreinforced and reinforced specimens, but also the peak strength, post-peak behaviour and dilatancy characteristics from the FEM analysis all compared well with those from the physical tests. The effects of reinforcement rigidity and covering ratio were also well simulated. The relationship between the reinforcement covering ratio in the physical tests and the equivalent interface friction angle for the FEM analysis that provides the same reinforcing effects is presented. The mechanism of tensile-reinforcing is analysed based on local stress paths within the reinforced sand obtained from the FEM analysis.

A series of drained stress path plane strain tests was performed on saturated dense specimens of Toyoura sand with precise stress and strain measurements. It is shown that all the strain increments (i.e., axial, lateral, shear and volumetric) that occurred by loading between two stress states were dependent on the intermediate stress paths. It is suggested that the use of a strain quantity as the hardening parameter that is independent of stress history in an elasto-plastic model for sand may not be relevant. Based on the test results, one form of energy function is shown to be stress history-independent. This quantity is a state parameter, being a unique function of one form of stress parameter for different stress paths. How this function can be used as a stress history and stress path-independent hardening function in an elasto-plastic model briefly discussed. The effects of stress history and instantaneous stress path on the friction angles at the failure and residual states are negligible. The effects of stress path on stress-dilatancy relationships based on plastic strain increments are found to be small but noticeable.

The paper presents a theoretical and experimental study of the effects of non-coaxiality or non-coincidence of the principal stress and the principal plastic strain increment directions on the behaviour of granular materials. Experimental results from hollow cylindrical tests on sand involving principal stress rotation which support previously published results on non-coaxiality are presented. These results imply that constitutive relations cannot be sufficiently formulated in the principal stress space unless the deviations between the principal stress and plastic strain increment directions are taken into consideration. It is shown that plasticity formulations with plastic potentials that are scalar functions of the stress invariants alone implicitly assume coaxiality and cannot be used for loading involving principal stress rotation. The paper presents a comprehensive analysis of the effects of non-coaxiality on the energy dissipation of sand. The paper shows that energy dissipation calculated from the principal stresses and the principal plastic strain increments or from the stress and plastic strain increment invariants, would be erroneous and would over-estimate the amount of dissipated energy during loading in the case of non-coaxial flow. A non-coaxiality factor is introduced in order to account for the effects of non-coaxiality on the energy dissipation equation and in a stress-dilatancy relation for granular materials. Explicit expressions of the non-coaxiality factor for two-and three-dimensional loading conditions are given at the end of the paper. Experimental results are presented to show the validity of the proposed energy dissipation and stress-dilatancy equations.

Sampling induces structural modifications in clay which depend on the stress and strain paths followed during sampling history. A complete description of disturbance is very difficult, because it originates in situ during drilling and is completed in the laboratory with reloading at the pre-existing stress conditions. In this paper, only the effects of undrained relief of confining effective stresses are investigated for reconstituted natural clays of varying plasticity. During undrained stress relief, negative pore pressure (suction) occurs; this residual stress was selected as a reference parameter for disturbance appraisal and compared with initial conditions before unloading. For the same type of soil, consolidation was carried out in various stress conditions (isotropic and anisotropic), and both normally consolidated and overconsolidated specimens were made. After undrained unloading, residual stress was evaluated by means of a controlled swelling technique using a modified oedometer and a triaxial device. This method was validated by comparisons with suction plate measurements performed on reconstituted kaolin clay specimens. Experiments showed various causes of loss of suction in samples, related to plasticity index, stress history, stress conditions and released stress level. For normally consolidated non-structured clays, undrained stress relief gives rise to some important disturbance processes, more evident for anisotropically consolidated samples. Residual stress may be considered meaningful for stress relief disturbance appraisal.