Influence of anisotropic stress states on the thermal volume change of unsaturated silt

Abstract

This study is focused on understanding the influence of anisotropic stress states on the thermal volume change of unsaturated, compacted silt specimens. A thermo-hydro-mechanical true-triaxial cell was used that permits control of the temperature on all six boundaries of a cubical soil specimen as well as control of the suction within the specimen to provide drained conditions during mechanical loading and temperature changes. Six non-isothermal tests were performed as part of this study, each involving suction application, consolidation to a given isotropic or anisotropic stress state, heating and cooling in stages under drained conditions, and unloading. Specifically, tests having minor to major principal stress ratios of 1.0, 0.7, and 0.5 were performed on specimens having initial degrees of saturation of 0.7 and 0.8, complementing tests on the same soil under similar stress states but saturated conditions published in a previous study. Although compressive thermal axial strains were measured in both the major and minor stress directions, a greater thermal axial strain was observed in the direction of the major principal stress for stress ratios less than 1.0. However, similar thermal volumetric strains were observed in all of the tests regardless of the stress state. A small effect of inherent anisotropy was observed due to the formation of the specimens using compaction. Specimens with a lower initial degree of saturation experienced greater thermal volume changes than specimens closer to saturation, possibly due to thermal collapse of the air-filled voids during heating or thermally accelerated creep after application of a given plastic strain during mechanical loading. An empirical relationship to consider the effects of anisotropic stress states and variable saturation was incorporated into an established elasto-plastic model developed for saturated soils under isotropic conditions, and a good fit was obtained between the measurements and predictions.