International Journal of Geomechanics

Pavement foundation geomaterials, i.e., fine-grained subgrade soils and unbound aggregates used in untreated base/subbase layers, exhibit nonlinear behavior under repeated wheel loads. This nonlinear behavior is commonly characterized by stress-dependent resilient modulus material models that need to be incorporated into finite element (FE) based mechanistic pavement analysis methods to predict more accurately the pavement resilient responses, such as stress, strain, and deformation. Many general-purpose FE programs have been used to predict such pavement responses under various traffic loading conditions while not considering properly material characterizations of the unbound aggregate base/subbase and subgrade soil layers. This paper describes the recent pavement FE modeling research efforts at the University of Illinois focused on using both the specific-purpose axisymmetric and general-purpose three-dimensional (3D) FE programs for flexible pavement analyses. To properly characterize the resilient behavior of pavement foundations, nonlinear stress-dependent modulus models have been programmed in a user material subroutine (UMAT) in the commercial general-purpose finite-element program ABAQUS. The results indicated that proper characterizations of the nonlinear stress-dependent geomaterials significantly impacted accurate predictions of critical pavement responses. The prediction ability of the developed nonlinear UMAT characterization was next validated by predicting similar pavement critical responses to those measured from field instrumented pavement test sections. Different resilient modulus models, considering both axisymmetric and 3D stress states, developed from true triaxial test data on unbound granular materials were also studied. When the intermediate principal stresses were taken into account in the 3D modulus model development unlike in the axisymmetric models, somewhat lower asphalt concrete tensile strains were obtained from 3D nonlinear FE analyses of flexible pavements with unbound aggregate bases.

The critical state of a granular material was investigated by performing cubical triaxial simulations using the discrete element method. These samples, assemblages of ellipsoids of two kinds, were prepared either by settling the randomly generated particles under gravity or by compressing these particles without gravity. Gravity was set to zero and the samples were further consolidated isotropically. Then, numerical drained tests were carried out until the critical state was reached. Macroscopic and microscopic data in the critical state were examined. The result shows a unique critical state line in the void ratio-mean stress space regardless of the initial conditions. The obliquity (ratio of major and minor principal stresses) as a function of strain is very similar for these samples at different confining pressures. The most interesting result is the linear critical state line in the void ratio-mean stress space. Three microscopic parameters including particle orientation, branch vector, and contact normal force were examined. At critical state, the long axes of most ellipsoids are perpendicular to the major principal stress direction, the distribution of branch vectors is random, and the distribution of contact normal forces shows a concentration along the major principal stress direction. The micromechanical descriptor based on contact normal forces is closely related to the macroscopic variable, obliquity.

A numerical model for the loose and the cemented component of the Torino subsoil is presented in this paper. The model is represented by a dense packing of nonuniform sized cylindrical particles with clumps by using the particle flow code in two dimensions. The paper shows how the synthetic specimens were calibrated to reproduce true material behavior giving insights over the influence of various parameters (such as platens velocity, particle shape, bond strength, etc.) on the mechanical behavior of bonded and unbonded assemblies. Loose and cemented ground are basically characterized by the same grain-size distribution, therefore the same numerical model is adopted. The effect of cementation is taken into account by introducing parallel bonds at contacts between the different particles. It is shown that the mechanical behavior of the Torino subsoil is well reproduced at the volume element scale. The results can be used to build representative numerical models to study engineering problems at the site scale.

The vertical uplift capacity of an infinite numbers of rigid strip plate anchors placed in a group and embedded in sand, has been determined. An upper bound theorem of limit analysis in combination with finite elements and linear programming has been used to perform the analysis. All the anchors are loaded to failure simultaneously with the same magnitudes of the failure loads. For different clear spacing $\left(S\right)$ between the anchors, the magnitude of the efficiency factor $\left({\xi }_{\gamma }\right)$ is determined; where ${\xi }_{\gamma }$ is the ratio of the failure load for an interfering strip anchor, for given width $\left(B\right)$ and depth $\left(d\right)$ , to that of an isolated strip anchor having the same width and depth. The magnitude of ${\xi }_{\gamma }$ has been found to reduce substantially with a decrease in the spacing between the anchors. It was noted that a rigid soil wedge just above the anchor plate and bounded by planar rupture surfaces separates out and it moves with the velocity same as that of the anchor plate itself. The results available from the analysis were found to compare reasonably well with the available theoretical and experimental data from the literature.