In this paper, the genetic algorithm (GA) is applied to determine the parameters of unified model for clay and sand, referred to as CSUH constitutive model. When determining the parameters of CSUH model by conventional methods, many conventional tests are required to be done, including low and high confining pressure triaxial tests and isotropic compression tests with specific initial void ratio. That will take more efforts and long time, and a basic understanding of CSUH model should be requested at the same time. However, using GA to automatically determine those parameters only needs a small amount of conventional triaxial tests. Finally, taking the parameters determination of Toyoura sand in CSUH model as an example, the GA is used to determine the parameters according to the partial experimental data and then to predict all the experimental data, the prediction result is acceptable.

This paper is centered on the assessment of 3 soil-water retention curves (SWRC) equations for unsaturated soils, for its predictive capability with respect to different soil types. The fitting effects of van Genuchten equation, Fredlund and Xing equation and Brooks and Corey equation on clay, silt and sand are studied. The influence of different soils with different initial dry density on parameters in different equations is analyzed. A new SWRC prediction method considering the influence of initial void ratio is proposed, which assumes a linear relationship between the initial void ratio and the saturation under the same suction. By analyzing the test data of sand, the predicted results are compared well with test results.

Frozen soil’s mechanical behavior is characterized by interactions between solid grains, ice and unfrozen water. It is strongly affected by temperature and ice content, indicating pronounced differences between frozen and unfrozen soils. Rate-sensitive behavior of frozen soil is expected, given the highly rate-dependent behavior of ice. The consequence is reflected in the peculiar features seen in frozen soil’s strength and deformation characteristics under transient temperature and load. To capture these features, an elasto-viscoplastic constitutive model for estimating temperature- and strain rate-dependent behavior of water-saturated clays is presented that is applicable continuously to both frozen and unfrozen states. This model adopts the p’:q plane with a Critical State Line (CSL) that moves with temperature and strain rate while converging to a unique unfrozen CSL, thus it is seamlessly continuous to a conventional elasto-viscoplastic critical state model in unfrozen states. This model is based on an isotach over-stress approach with the cryogenic suction as additional state variable, and is potentially capable of describing varying-load and varying-temperature behaviour seen under combined influence of these two factors.

In previous studies, almost all they had carried out the crushing tests under usual tri-axial compression, Ko and simple shear conditions. However, it is necessary to perform the crushing test under general stress conditions in ground with controlled each principal stresses individually to find the mechanical characteristics of the particle crushing. Therefore, the purpose of this study is to confirm the effect for particle crushing under various stress paths in the combination of principal stresses. Authors plan the particle crushing tests with combinations of principal stresses operating the mean stress p, deviatoric stress q and Lode angle θ up to the Shear Failure Line (SFL) on the pie-plane. These particle crushing tests were carried out using the high pressure true tri-axial compression apparatus under the planned stress paths. As a result, the degree of Lode angles has little traction to the progress of the particle crushing and authors found that deviatoric stress q has essential effect on the particle crushing. In addition, the degree of particle crushing are little effected by the various monotonic stress paths on p-q plane when the maximum values of p and q are same.

In the present study, water retention tests on Toyoura sand with high bulk density and low bulk density are conducted. Three-phase microstructures in sand specimens are visualized using an x-ray micro tomography at the different water retention states. Distributions of the degree of saturation in each pore are quantified using a segmentation technique and a Voronoi tessellation technique. The pore-scale distribution of degree of saturation in partially saturated sands with different bulk density and its variation during drying and wetting processes are discussed.

To investigate the macroscopic and microscopic mechanical behaviors of unsaturated structural loess, three- dimensional (3D) distinct element analyses of constant stress ratio (CSR) compression and collapsibility tests were performed. The evolutions of void ratio, deviator strain and bond failure number against vertical stress and water content were analyzed. The results show that the simulations reproduce the main mechanical behaviors of the unsaturated structural loess. The compression tests show that, with stress ratio decreases, the deviator strain and bond failure number increase. The collapsibility tests show that the void ratio decreased with an increases of water content.

The discrete element analysis of the true triaxial and wetting tests is conducted to study the macroscopic and microscopic mechanical properties of unsaturated structural loess under complex stress states. The change of axial strain after wetting and the variation of void ratio and bond breakage number with water content were studied in this work. The results show that the collapsing failure occurs when the deviatoric stress approaches or exceeds the peak shear strength of the corresponding saturated structural loess sample. There is little difference in the axial strain between the quick wetting (QW) method and gradual wetting(GW) method, and the volumetric strain of the samples wetted by QW are larger than that by GW specimen. The bond breakage number of the samples wetted by QW are slightly higher than that by GW. However, the number of bond breakage wetted by the two wetting methods is slightly higher than that of the corresponding saturated structural loess sample.