Advances in Bifurcation and Degradation in Geomaterials

Techniques from Complex Networks are used to study the evolving topology and functional connectivity in various granular systems in both two and three dimensions. A generic process of self-organization in all samples is realized and is characterized by the co-evolutionary synergy between force chains and 3-cycles. Three-cycles provide force chains a twofold benefit: (i) they prop-up force chains in the way that a counterfort or buttress supports a wall, and (ii) they frustrate rolling at contacts which is a critical mechanism for buckling. All samples reflect an inherent structural hierarchy where cyclic (minimal cycles) and linear (force chains) motifs of various length scales serve as basic building blocks for self-organization.

We investigate the rheology, force transmission and texture of granular materials composed of elongated particles by means of contact dynamics simulations. The particles have a rounded-cap rectangular (RCR) shape described by a single elongation parameter varying from 0 for a circular particle to 1 for an infinitely thin or long particle. We study the quasi-static behavior, structural and force anisotropies as a function of the elongation parameter for packings submitted to biaxial compression. The shear strength is found to increase linearly with this parameter whereas the solid fraction both at the initial isotropic state and in the critical state is nonmonotonous. We show that for these elongated particles a harmonic decomposition of the stress tensor provides a fairly good approximation of the internal state. Our data suggest that the increase of shear strength with reflects both enhanced friction mobilization and anisotropic particle orientation as the elongation of the particles increases.

Instability of sand can occur under drained or undrained loading conditions in loose sand or dense sand. A micromechanics approach is used for the analysis of local instabilities of inter-particle contacts and their relations to the global instability of assembly. The comparisons between experimental and predicted results on Toyoura Sand show the capability of the model in capturing the modes of instability at the assembly level. Analysis at inter-particle contact level for loose sand under an undrained triaxial loading condition show that the number of unstable inter-particle planes increases continuously, while the assembly remains stable. The assembly becomes unstable when the sum of all local second-order work becomes zero. After this point, the overall shear stress begins to decrease during a strain controlled test, and progressively, more inter-particle contact planes become instable.

It is known that unsaturated soil exhibits more brittle failure due to the collapse of the water meniscus, caused by shearing or the infiltration of water, than saturated soil. The aim of this paper is to observe the strain localization behavior and the microstructural changes in partially saturated soil during the deformation process using microfocus X-ray CT. The strain localization of fully saturated, partially saturated, and air-dried sand specimens during triaxial compression tests is observed and discussed. In addition, the microstructures in the shear bands of partially saturated specimen are discussed.

This paper presents a study on the anisotropic behavior of liqufiable sand subjected to undrained shear, by using a 3D Discrete Element Method with two different approaches describing particle rolling. By using a sliding and free-rolling model, the force network in relation to anisotropy in medium-loose or dense samples presents a clear bimodal character, while the liquefiable loose specimen behaves differently. Appreciable degree of anisotropy is found developed in the weak force network when the sample tends to liquefy. When the rolling resistance is considered, all samples show marked increases in anisotropy in both the weak and strong force networks as well as the overall shear strengths, as compared with the free-rolling case. The loose sample tends also to be more resistant to liquefaction in the latter case than in the free rolling case under otherwise similar conditions.

Effect of rotations of particles of non-spherical shape is considered. Under combined action of shear and compressive forces, the moment balance produces apparent negative stiffness. In the case when friction is controlled by rotating particles non-sphericity leads to oscillating friction force. In the case of particulate materials in compression this may lead, under certain magnitudes of compression, to the initiation of global negative stiffness and, subsequently, strain localization.

We offer experimental observations of meso-scale deformation and kinematic activity within sheared granular layers to investigate the nature and spatial periodicity of nonaffine displacement fields within shear bands in a granular material. Prismatic specimens of sands and glass beads are subjected to plane strain deformation in which zero-strain conditions are enforced by translucent glass walls. We use the Digital Image Correlation (DIC) to track movements of small, overlapping particle clusters. By subtracting a superimposed first-order shear displacement field from the observed non-affine displacement fields, co-rotational vortices appear and coordinate with previously-observed kinematic patterns. We undertake a preliminary assessment of the spatial periodicity of such patterns to glimpse the nature of an underlying length scale for granular material deformation.

The quest to understand the connections between the triumvirate of structure, dynamics and function continues to drive the forefront of research in Complex Systems. Crucial to these explorations is the development of graph-theoretic techniques that: (i) can detect communities and associated boundaries in the underlying network or graph, which represents the interactions of constituent units, and (ii) quantify shortest paths and related network measures within this graph. We report on a new study using data from high resolution digital image correlation (DIC) measurements of grain-scale kinematics in sand under shear. Preliminary results show that the nodes of the network in the shear band region exhibit high

closeness centrality

– a network measure of how efficient a given node is in spreading information to all the other nodes in the graph. It is thus reasonable to expect that the most efficient routes for spread of kinematical information within this network are those from nodes that correspond to the grid points that lie along the shear band. We believe these studies will ultimately lead to an improved understanding of self-organization, the nature of energy flow and dynamics in the critical state regime in the presence of persistent shear bands.

We use crystal plasticity to describe the slip and dislocation in crystalline solids under mechanical loading. The constitutive formulation involves linearly dependent slip systems from which we extract a group of linearly independent slip systems using the ultimate algorithm advocated by Borja and Wren (Int J Numer Methods Eng 36:3815–3840, 1993). We implement the ultimate algorithm into a 3D nonlinear finite element code with infinitesimal deformation. We use the code to compare the deformation fields generated by crystal plasticity formulation and the classic

J

2

plasticity model.

Gas hydrates, especially methane hydrates are viewed as a potential energy resource since a large amount of methane gas is trapped mainly within ocean sediments and regions of permafrost. In the present study, gas production process by heating-depressurizing method was simulated. The simulation was conducted for the model with inclined seabed ground with hydrate bearing layer in order to investigate the mechanical behavior during dissociation. The method has been developed based on the chemo-thermo-mechanically coupled analysis, taking into account of the phase changes from solids to fluids, that is, water and gas, the flow of fluids, heat transfer, and the ground deformation (Kimoto et al. 2010). As for the constitutive model for hydrate-bearing sediments, an extended elasto-viscoplastic model for unsaturated soils considering the effect of hydrate bonding is used.

Rock nailing of the borehole is considered as a means to reinforce the rock and increase its borehole failure strength. The technique is modeled using Cox’s original shear-lag method. In the continuum sense, nail reinforcement is viewed as a body force that acts as a confinement. Borehole stability is analyzed using an analytical solution that couples the effects of the nails and the rock. Results for a reinforced with a given nail density and an un-reinforced borehole are presented. They show that the nail length and the nail-rock contact parameter play an important role in the stability. Rock dilation is also important since the action of the nails is mobilized from the difference in displacement between the rock and the nail. The results show the potential for a significant increase in borehole strength.

The stability of creeping faults is studied under the effect of shear heating, pore fluid pressurization and mineral decomposition. Such reactions enhance the pore fluid pressurization because they release fluid, but they limit the temperature rise because they are endothermic. The stability of stationary slip is investigated by performing a linear perturbation analysis. It is shown that chemical reactions change a stable behaviour into an unstable one when the pore pressure effect is larger than the endothermic effect. It is shown that the opposite effect can also be observed when the dehydration reactions can trigger an arrest of the fault. Mineral decomposition can thus strongly modify the nucleation of seismic slip.

We present 3D grain-fluid simulations based on the discrete element method interfaced with the lattice Boltzmann method and applied to investigate the initiation of underwater granular flows. We prepare granular beds of 800 spherical grains with different values of the initial solid fraction in a biperiodic rectangular box. In order to trigger an avalanche, the bed is instantaneously tilted to a finite slope angle above the maximum angle of stability. We simulate the dynamics of the transient flow for different solid fractions. In agreement with the experimental work of Iverson (Water Resour Res 36(7):1897–1910, 2000) and Pailha et al. (Phys Fluids 20(11):111701, 2008), we find that the flow onset is controlled by the initial solid fraction.

In this communication, we perform a multi-scale periodic homogenization of Nernst-Planck and Poisson-Boltzmann equations describing the ionic transfer in saturated cementitious materials. Two models of ionic transfer are established successively. The first one by periodic homogenization from the Debye length scale to the capillary porosity scale, taking into account the electrical double layer phenomenon. The second model is obtained by upscaling the same ionic transfer equations from the capillary porosity scale to the material’s scale. Numerical simulations on porous media with more or less complex periodic microstructures are then carried out. Comparisons with existing experimental data are also presented and discussed.

In this paper, we report experimental results on the development in an immersed granular bed of a fluidized zone under the effect of a confined liquid flow upward. For this, two optical techniques (index-matching and laser induced fluorescence) have been combined to visualize the interior of a model granular medium made of glass beads. For small flow rates, the bed remains static while it gets fluidized in a vertical chimney at larger flow rates. For intermediate flow rates, a fluidized cavity can be observed in the vicinity of the injection hole. This cavity can either reach a steady-state or progressively expand up to the top at a small speed. The stability of such a cavity reveals a strong hysteretic behavior depending on whether the flow rate increases until fluidization or decreases after fluidization. We present here some characteristics of the phenomenon: phase diagram, regime of stable fluidized cavity, kinematics of fluidization front.

Contact erosion appears at the interface of two soil layers subject to a groundwater flow. Particles of the finer soil are eroded by the flow and transported through the pores of the coarser layer. Fluvial dykes are often exposed to this phenomenon. Small-scale experiments combining Refractive Index Matching medium, and Particle Image Velocimetry were carried out to measure the flow characteristics close to the interface between a porous medium and a sandy layer. Velocity and shear-stress distributions were obtained. They underline the spatial variability of stress exerted by the flow on the fine soil which is directly related to the variability of the pore geometry. These distributions can be helpful in better modelling contact erosion by going beyond the simple use of global mean values, such as Darcy velocity, as is usually proposed in the literature.

Hydraulic fracturing is a field technology widely used in the petroleum industry in order to increase the effective permeability of the reservoir and thus the production of gas by fracturing the rock by an injection of fluid. Hydrofracturing is often being monitored by detecting and analyzing microseismic events. We present a numerical technique that simulates the occurrence of microseismic events and their deformation modes during hydrofracturing and thus allows us to improve our understanding of hydrofracturing-related microseismicity. We can explain the time and spatial spread in the location of seismic events by the heterogeneity of the reservoir and the variability in the deformation modes as a natural process reflecting the reorganization of stresses in an elastic medium. We show that microseismic activity reflects the macroscopic description of hydrofracturing as a tensile crack even in highly heterogeneous reservoirs.

Segregation is a well known yet poorly understood phenomenon in granular flows. Whenever disparate particles flow together they separate by size, density and shape. If we wish to know how to separate particles more efficiently, or even how to keep them mixed together, we require a good understanding of both the phenomenology of the flow, and a quantitative analysis of the evolving particle size distribution towards a steady state. This chapter outlines the continuing effort towards this end, and provides a clue as to the future direction of our research.

A numerical study of the interaction between granular flow and an obstacle on an inclined plane is presented. A depth-averaged, two-dimensional Savage-Hutter-type model is used. The underlying differential equations are usually solved by finite difference. The model, which has been proved to perform well for steady granular flow, turns out to be inappropriate for the interaction between granular flow and obstacles. In this paper, the quality of the numerical solution is significantly improved by grid refinement. We make use of the Adaptive Mesh Refinement, where only local grid refinement around the obstacle is needed.

This study applies finite element analysis to examine the mechanism of crack propagations observed at the Glacier Point of the Yosemite National Park following the June 13, 1999 rockfall event. Extensive crack propagations were observed on a rectangular area of about 18 m by 23 m of rock exfoliation sheet on a rock face about 550 m above the Yosemite Valley 54 days following the rockfall by helicopter flights and by photographs taken by from the valley. This paper models the crack pattern on the hook-shaped rock sheet by using a plane stress analysis subjected to various ratios of vertical to horizontal strain increments. The present finite element analysis uses a code called RFPA that adopts a reduced modulus approach to model the damage process of rock. A mesh of 160 by 160 elements was used to model a rock face of about 35 m by 35 m, and a Weibull distribution is used to model the heterogeneity of the rock face. We found that when the vertical to horizontal strain ratio is smaller than one, the crack pattern closely resembles that observed at the Glacier Point. Thus, the present paper provides a plausible mechanism for the observed crack propagation.

Researchers at Caterpillar have been using Finite Element Analysis or Method (FEA or FEM), Mesh Free Models (MFM) and Discrete Element Models (DEM) extensively to model different earthmoving operations. Multi-body dynamics models using both flexible and rigid body have been used to model the machine dynamics. The proper soil and machine models along with the operator model can be coupled to numerically model an earthmoving operation. The soil – machine interaction phenomenon has been a challenging matter for many researchers. Different approaches, such as FEA, MFM and DEM are available nowadays to model the dynamic soil behavior; each of these approaches has its own limitations and applications. To apply FEA, MFM or DEM for analyzing earthmoving operations the model must reproduce the mechanical behavior of the granular material. In practice this macro level mechanical behavior is not achieved by modeling the exact physics of the microfabric structure but rather by approximating the macrophysics; that is using continuum mechanics or/and micromechanics, which uses length scales, that are larger than the physical grain size. Different numerical approaches developed by Caterpillar Inc. researchers will be presented and discussed.

Damage processes in geomaterials typically involve moving strong and weak discontinuities, multiscale phenomena, excessive deformation, and multi-body contact that cannot be effectively modeled by a single-scale Lagrangian finite element formulation. In this work, we introduce a semi-Lagrangian Reproducing Kernel Particle Method (RKPM) which allows flexible adjustment of locality, continuity, polynomial reproducibility, and h- and p-adaptivity as the computational framework for modeling complex damage processes in geomaterials. Under this work, we consider damage in the continua as the homogenization of micro-cracks in the microstructures. Bridging between the cracked microstructure and the damaged continuum is facilitated by the equivalence of Helmholtz free energy between the two scales. As such, damage in the continua, represented by the degradation of continua, can be characterized from the Helmholtz free energy. Under this framework, a unified approach for numerical characterization of a class of damage evolution functions has been proposed. An implicit gradient operator is embedded in the reproduction kernel approximation as a regularization of ill-posedness in strain localization. Demonstration problems include numerical simulation of fragment-impact of concrete materials.

A zone with significant irreversible deformations and significant changes in flow and transport properties is expected to be formed around underground excavation in the deep geological layers considered for the high level radioactive waste disposal. The present study concerns the modeling of this phenomena by a micromechanical damage model, based on a Mori-Tanaka homogenization on a cracked media. This anisotropic model is derived from Eshelby homogenized scheme, on which a coupling between damage and friction is taking into account on cracks. Compared to elastoplastic model on tunnel drilling modeling, micromechanical modeling seems very promising: both approaches provide similar EDZ sizes and shapes even if they do not have the same effects on perturbed mechanical behavior; micromechanical model also overcomes the elastoplastic one by a realistic characterization of crack processes.

The paper presents recent progress in the development of a computational multiscale modeling approach for simulating the interfacial mechanics between dense dry granular materials and deformable solid bodies. Applications include soil-tire/track/tool/penetrometer interactions, geosynthetics-soil pull out strength, among others. The approach involves a bridging scale decomposition coupling between a three-dimensional ellipsoidal discrete element (DE) model and a finite strain pressure-sensitive micromorphic constitutive model implemented in a new multi-field coupled finite element (FE) method. The concept borrows from the atomistic-continuum bridging scale decomposition methods, except for the relevant differences for our problem in granular materials: (1) frictional, large relative motion of DE particles/grains upon shearing by deformable solid; (2) open window representation of DE region in contact with deformable solid; (3) overlapping finite strain micromorphic constitutive model for granular material with additional kinematics and higher order stresses; and (4) adaptivity of DE-FE region. The paper focusses on a simpler subset problem of topics (1–3): a one-dimensional glued elastic string of spherical DEs, overlapped partially with a one-dimensional micropolar continuum FE mesh. A numerical example is presented.

This paper deals with numerical simulations of granular soils, idealized by a cyclic multi-mechanism anisotropic non-associated plasticity constitutive model (Aubry et al. 1982; Hujeux 1985) coupled to a dilation second gradient model (Fernandes et al. 2008), under a finite elements method framework. This constitutive model is integrated through an implicit scheme into the Finite Element software

Code_Aster

(Foucault 2009). The use of this constitutive model, in case of strain localization, exhibits mesh sensitivity as for any strain-softening model. First we verify the ability of the dilation second gradient model to (1) circumvent the problem of mesh sensitivity with this model and to (2) describe the post-peak behaviour of the studied loading process. We studied a biaxial laboratory test, on Hostun sand, under drained conditions and monotonic loading. Second we develop a methodology to determine the values of the second gradient parameter – i.e. characteristic length – for different sets of soil model parameters. We have adapted a 1-D theoretical approach proposed by Chambon et al. (2001), applied to our constitutive model, whose comparison with our numerical results provides a characteristic length. To check the ability of the dilation second gradient model to ensure objective results with this soil constitutive model, we performed an extensive validation on a bearing capacity case with excavation.

This paper deals with some recent results obtained with a simplified second gradient model to simulate localized patterns. The simplification is based on the use of the gradient of the volume variation only. It is first shown that this model is very efficient since it is less time consuming than a classical one. Consequently, such a model can be useful for 3-D computations. Finally some results show once more that even for 3-D computations, enhanced models such as second gradient ones do not restore the mathematical well posedness of the initial boundary value problem.

In the design of nuclear waste disposals, an important topic concerns the evolution of an Excavated Damage Zone (EDZ) with a thermal exchanges. In this paper, a new model of local second gradient coupling with a thermo-hydro-mechanical is presented. As for monophasic case, the use of enhanced media induce the objective of the computation but not the uniqueness of the solution. Some classical engineering problems are presented which exhibit several solutions.

The macroscopic behavior of granular materials, as a consequence of the interactions of individual grains at the micro scale, is studied in this paper. A two scale numerical homogenization approach is developed. At the small-scale level, a granular structure is considered. The Representative Elementary Volume (REV) consists of a set of N polydisperse rigid discs (2D), with random radii. This system is simulated using the

Discrete Element Method

(DEM) – molecular dynamics with a third-order predictor-corrector scheme. Grain interactions are modeled by normal and tangential contact laws with friction (Coulomb’s criterion). At the macroscopic level, a numerical solution obtained with the Finite Element Method (FEM) is considered. For a given history of the deformation gradient, the global stress response of the REV is obtained. The macroscopic stress results from the Love (Cauchy-Poisson) average formula including contact forces and branch vectors joining the mass centers of two grains in contact. The upscaling technique consists of using the DEM model at each Gauss point of the FEM mesh to derive numerically the constitutive response. In this process, a tangent operator is generated together with the stress increment corresponding to the given strain increment at the Gauss point. In order to get more insight into the consistency of the two-scale scheme, the determinant of the acoustic tensor associated with the tangent operator is computed. This quantity is known to be an indicator of a possible loss of uniqueness locally, at the macro scale, by strain localization in a shear band. The results of different numerical studies are presented in the paper. Influence of number of grains in the REV cell, numerical parameters are studied. Finally, the two-scale (FEM-DEM) computations for simple samples are presented.

Geomaterials are dissipative particulate systems due to internal forces that arise from intergranular friction or viscosity. As such, their mechanical behaviour is highlighted by various forms of failure with either localization of deformations or diffuse deformations. Intriguingly, the latter failure mode, which does not involve any localized deformations or discontinuities, is normally observed well before plastic limit conditions are met. This work examines failure as a material (constitutive) instability phenomenon giving way to a bifurcation problem whereby a multiplicity of material response is possible for the same initial loading history. We use a rate-independent elastoplastic constitutive model with plastic strain softening and non-associativity of plastic flow through a micromechanically derived stress-dilatancy equation. The dependencies of the latter on density, stress, and fabric provide essential mathematical sources of material instability to promote the capturing of discontinuous response. An example problem involving diffuse and localization deformations in a water saturated sand sample as a boundary value problem is presented.

This paper presents the results of a comprehensive study of the effects of rolling resistance in shear band formation in granular material using DEM (Discrete Element Modeling). The study used the two-dimensional Particle Flow Code (PFC) to simulate biaxial compression and simple shear tests in granular materials. A rolling resistance model was implemented as a user-defined model (UDM) in PFC. A series of parametric studies were performed to investigate the effects of different levels of rolling resistance parameters on the emergence and shear bands in granular materials. The results reinforce prior conclusions by Oda et al. (Mech. Mater. 1:269–283, 1982) on the importance of rolling resistance in promoting shear band formation in granular materials. It was shown that increased rolling resistance results in the development of columns of particles in granular soils during the strain hardening process. The buckling of these columns of particles then leads to the development of shear bands. It is concluded that the variation of rolling resistance parameters has a significant effect on the orientation, thickness and the onset of occurrence of shear bands. The PFC models were tested under a wide range of macro-mechanical parameters and boundary conditions, and how they influence shear band characteristics.

We rely on 3D simulations based on the Lattice Element Method (LEM) to analyze the failure of porous granular aggregates under tensile loading. We investigate crack growth by considering the number of broken bonds in the particle phase as a function of the matrix volume fraction and particle-matrix adhesion. Three regimes are evidenced, corresponding to no particle damage, particle abrasion and particle fragmentation, respectively. We also show that the probability density of strong stresses falls off exponentially at high particle volume fractions where a percolating network of jammed particles occurs. Decreasing the matrix volume fraction leads to increasingly broader stress distribution and hence a higher stress concentration. Our findings are in agreement with 2D results previously reported in the literature.

The evolution of patterns of shear zones in cohesionless sand for quasi-static earth pressure problems of a retaining wall under conditions of plane strain was analyzed with a discrete element method (DEM). The passive and active failure of a retaining wall was discussed. The numerical calculations were carried out with a rigid and very rough retaining wall undergoing horizontal translation, rotation about the top and rotation about the toe. The geometry of calculated shear zones was qualitatively compared with experimental results of laboratory model tests using X-rays and a Digital Image Correlation (DIC) technique, and quantitatively with corresponding finite element results obtained with a micro-polar hypoplastic constitutive model.

Snow avalanches generally result from the collapse of a weak layer underlaying a cohesive slab. We use the finite element code Cast3m to build a complete mechanical model of the {weak layer-slab} system including inertial effects. We model the weak layer as a strain-softening interface whose properties are spatially heterogeneous. The softening accounts for the breaking of ice bridges. The overlying slab is represented by a Drucker-Prager elasto-plastic model, with post-peak softening to model the crack opening. The two key ingredients for the mechanical description of avalanches releases are the heterogeneity of the weak layer and the redistribution of stresses by elasticity of the slab. The heterogeneity is modeled through a Gaussian stochastic distribution of the friction angle with spatial correlations. We first study the effect of the weak layer’s heterogeneity and the slab depth on the release on a simple uniform slope geometry. We observe two releases types, full slope releases corresponding to a crown rupture and partial slope releases for which the traction rupture occurs inside the slope and thus only a part of the slope is released. The influence of slab depth on the relative proportion of these two rupture types, as well as on the avalanche angle distributions is also studied.

The understanding of chemical effects on mechanical behavior of porous media is a key element in the solution of problems in geology, mining, soil, rock and environmental engineering. In order to develop this understanding, the current work employs a new set of equations to numerically investigate the coupled mechanical-hydrological-chemical (MHC) problem for soils and soft rocks. The objectives of this research are to observe the soil and soft rock behavior under mechanical loading and chemical erosion and also to validate the application of our solver algorithm written in a finite element programming environment “escript”.

Strain localization is a phenomenon occurring very often in anelastic media close to rupture and it is very important to take it into account in most of applications, particularly in geotechnics. Strain localization has been widely studied experimentally, see for instance Desrues (2004) or Desrues and Viggiani (2004). It has also been put in evidence in finite element simulations even if it presents mesh dependencies, see Chambon et al. (2001) for instance. From a theoretical point of view, the Rice condition (Rice 1976; Rudnicki and Rice 1975), with some assumptions, enables to determine the onset of strain localization but does not necessarily prove the existence of shear bands.

Shear banding in Santa Monica Beach sand deposited by dry pluviation in hollow cylinder specimens is studied in drained torsion shear tests with rotation of principal stress directions. Each test was conducted with the same, constant inside and outside confining pressure, σ

r

, thus tying the value of

$$b = ({\sigma }_{\mathit{2}} - {\sigma }_{\mathit{3}})/({\sigma }_{\mathit{1}} - {\sigma }_{\mathit{3}})$$

to the inclination, β, of the major principal stress. Shear bands can develop freely without significant restraint for soft rubber membranes. Strain localization and shear banding were observed in the hollow cylinder specimens, and this created failure conditions in plane strain and in tests with higher b-values. The results indicate the influence of the cross-anisotropic fabric on the stress-strain behavior, on the shear band inclination and on the shape of the failure surface.

It is known that the unstable behavior of unsaturated soil includes collapses behavior which is due to the decrease of suction as well as the shear failure. In addition, we often meet the numerical instability for the simulation of unsaturated soil during the wetting process. In the present study, we have performed a one-dimensional linear instability analysis for the air-water-soil coupled three-phase viscoplastic material. From the instability analysis of the viscoplastic unsaturated soil subjected to the water infiltration, we have found that the specific moisture capacity, the matric suction and the hardening parameter affect the instability. Then, we have performed numerical simulations of the behavior of unsaturated soil during the water infiltration using a multi-phase coupled elasto-viscoplastic finite element analysis method. The conditions under which the numerical instability occurs during the simulation agree well with the results by linear instability analysis.

The paper revisits the analysis of undrained shear banding in water saturated sand. By taking into account the coupling between potential volume change and excess pore water pressure as well as the continuity of stresses and pore pressure across the shear band, it is theoretically shown that incipient localisation may take place in loose sand but is precluded for dense sand under undrained conditions. The undrained shear band orientation primarily depends on the Poison’s ratio and the dilatancy characteristics of the material. Numerical examples are given to demonstrate the variation of shear band inclination with void ratio and the initial consolidation pressure.

This work uses a bifurcation approach to develop theoretical predictions for deformation band formation for a suite of true triaxial tests on Castlegate sandstone. In particular, the influence of the intermediate principal stress on strain localization is examined. Using common simplifying assumptions (localization occurs at peak stress, and the failure surface is similar to the yield surface), theoretical predictions captured the overall trends observed experimentally. However, agreement between predicted and observed band orientations for individual specimens was varied. This highlights the importance of detailed data analyses to accurately determine key material parameter values at the inception of localization.

Undrained plane strain compression experiments were conducted on slurry consolidated kaolin clay samples to elucidate the mechanism of shear localization in clays. The biaxial device was heavily internally instrumented, thus enabling the multiple boundary displacement measurements. These measurements were combined with boundary digital imaging to compute and verify displacements of the shear band. The orientation of shear band was determined from photographs thus enabling the computation of its dilatancy angle. Finally, the volumetric and shear strains of the shear band were computed for two limiting initial thicknesses of the shear band corresponding to 1 mm and 1.5 mm. The computed meso scale strains are several times larger than those reported in sand

A test series designed to investigate and quantify the effect of the intermediate principal stress on the failure of Castlegate sandstone was completed. Using parallelepiped specimens and a true triaxial testing system, constant mean stress tests were conducted. Stress states ranged from axisymmetric compression to axisymmetric extension. Results suggest a possible failure dependence on the third invariant of deviatoric stress at lower mean stresses.

Spectral effect of temporal localization of acoustic events associated with their spatial localization is investigated. It is shown the temporal localization produces blue shift in the frequency spectrum as compared to non-localized (random) generation of acoustic events. This opens a possibility to detect localization on the basis of a single channel recording without the need of determining the pulse locations. This can be done by analyzing the difference between the spectra of registered acoustic pulses and their randomized series.

The focus of this paper is on modeling the essential mechanical properties of weathered and moisture sensitive granular materials under dry and water saturated conditions using a state dependent solid hardness. Herein the solid hardness is related to the grain assembly in the sense of a continuum description and does not mean the hardness of an individual grain. Particular attention is paid to creep deformation initiated by water saturation of an initially dry material. First the constitutive model is outlined for isotropic loading paths. For general stress paths the influence of the stress deviator on creep deformation and stress relaxation is investigated with an extended hypoplastic model. The performance of the refined model is demonstrated for triaxial compression tests where the saturation of the material is investigated for different deviator stresses. The predictions of the model are compared with experiments.

This paper discusses results of experimental study. It highlights the influence of modes of reconstitution of samples on structural architecture of granular materials. Undrained triaxial tests were carried out on Hostun sand HN31. Different procedures for the preparation of the samples were used: moist tamping, dry pluviation and water pluviation. Bender elements which transmit and receive shearing waves have been use to study the structural anisotropy induced by each mode of deposition and its evolution during the loading path.Results obtained showed that the three different methods of preparation induce anisotropy of soil and lead to different mechanical behavior during the undrained shearing tests.

Failure by strain localization is commonly observed in geomaterials. In a previous workshop (IWBI Minneapolis St Paul, 2002), we presented an experimental characterization of strain localization in a porous sandstone (Bésuelle et al. 2000). This study was performed with classical axisymmetric triaxial compression tests. The effect of the confining pressure was observed on several aspects: onset of localization; pattern of localization; porosity evolution inside the localized bands. Complex patterns of localization were observed at high confining pressure in the transition between the brittle and ductile regimes, showing several deformation bands in the specimens. However the history (time evolution) of the localization was not accessible because the observations were post-mortem.

Combining integrated photoelasticity and high resolution particle image velocimetry (digital image correlation) provides a tool which can be used to study granular materials at different scales. We have used an experimental arrangement which subjects an assembly of small 3D glass grains to plane strain conditions at the macro-scale. While the quantitative interpretation of individual and incremental photoelastic images in terms of continuum stress quantities presents challenges, some comparisons of strain increment and qualitative stress increment data are presented with concentration on the regions of apparent strain localisation.