Thermo-Hydro-mechanical behaviour of loess is important for the serviceability and ultimate limit states of high speed railway embankment. In semi-arid areas, loess subgrade is subjected to daily heating-cooling and seasonally freezing-thawing. Many theoretical and experimental studies have demonstrated that soil behaviour is significantly affected by temperature. Most of these studies, however, focused on compacted soils. Loess, which is well known as a "problematic soil", has a metastable and high porosity structure under unsaturated conditions. Effects of soil structure (soil particle arrangement) on hydro-mechanical behaviour of loess at elevated and sub-zero temperatures has not been well studied.

The principal objectives of this research are to investigate the hydro-mecha...[ Read more ]

Thermo-Hydro-mechanical behaviour of loess is important for the serviceability and ultimate limit states of high speed railway embankment. In semi-arid areas, loess subgrade is subjected to daily heating-cooling and seasonally freezing-thawing. Many theoretical and experimental studies have demonstrated that soil behaviour is significantly affected by temperature. Most of these studies, however, focused on compacted soils. Loess, which is well known as a "problematic soil", has a metastable and high porosity structure under unsaturated conditions. Effects of soil structure (soil particle arrangement) on hydro-mechanical behaviour of loess at elevated and sub-zero temperatures has not been well studied.

The principal objectives of this research are to investigate the hydro-mechanical behaviour of loess at elevated and sub-zero temperatures. The oedometer, shear box and triaxial apparatus were modified by adding temperature control systems. With reference to the design of serviceability state, volume change behaviour of loess under cyclic heating-cooling and freezing-thawing was investigated through the newly developed oedometer and triaxial apparatus, respectively. For oedomter tests, loess with different specimen preparation methods (intact, compacted and reconstituted) was tested. With reference to the design of ultimate limit state, shear behaviour of intact and compacted loess at elevated temperatures were studied through the shear box apparatus. Furthermore, a new model for unfrozen water retention curve (UWRC) was proposed by considering stress effects. The performance of the new model and three popular existing models were compared through simulating measured UWRCs. Experimental results showed that the plastic strains of intact, compacted and reconstituted loess accumulated but at a reducing rate under cyclic heating-cooling. The accumulated plastic strain of the reconstituted loess was about 68% and 38% larger than those of the intact and compacted loess, respectively. Shear stiffness and dilatancy of compacted loess increase by up to 47% and 63% respectively with an increase in soil temperature from 20 to 60 ^oC. On the contrary, shear stiffness and dilatancy of intact loess decrease by 35% and 68%, respectively. The difference is because heating-induced plastic strain (0.4%), in addition to plastic strain-hardening effects, also destroys the resistant structure of intact loess. These differences are likely attributed to the different soil structures among the three types of specimens. Triaxial test results showed that phase change between liquid water and ice occurs during freezing-thawing. During phase change, the unfrozen water content of loess increased with an increase of confining stress. This is because higher confining stress results in smaller void ratios. More liquid water could be maintained due to the larger capillary forces (ice-water interface).

Comparisons between measured and calculated results illustrate that the new model improves the calculation of UWRC over existing UWRC models. Furthermore, the new model can well capture stress effects on UWRC, which could not be captured by existing models. The better performance obtained for the new model is mainly due to the separation of the freezing of capillary and adsorbed water.