Verification of generalized soil-water characteristic curve model considering two stress state variables for unsaturated soils

Abstract: This study investigated the influence of geostatic stress on soil structural evolution, water retention characteristics, permeability properties as well as the air-entry value and validated the reliability of the proposed generalized soil-water characteristic curve model (generalized SWCC). Nanyang expansive clay (depth of 0-10 m) was collected for determination of SWCC by a Geo-Expert high-type pressure plate apparatus for stress-dependent soil-water characteristic curve (SDSWCC-H) based on axis translation technique. A series of SWCCs for Nanyang expansive clay were obtained under the vertical depth of 0, 2, 5, 10 m (the geostatic stress was 0, 40, 100, 200 kPa respectively). The coefficient of nonuniformity and the coefficient of curvature of the expansive clay were 60.44 and 9.08, respectively. With the obtained data and literature data, the generalized SWCC considering two stress state variables and multimodal pore-size density probability function (M-PDF) was validated against a wide range of experiments under the following different conditions: 1) the confined compression tests and drying-SWCCs for Nanyang expansive clay under different geostatic stress; 2) Double-series porosity diatomaceous earth mixtures SWCCs with zero net normal stress state; 3) One porosity Korean residual soil SWCCs at different net confining stress; and 4) Drying-wetting path of France Bapaume loess under different stress path, and Japan Edosaki sandy soils under different dry density. Performances of the proposed equation, van Genuchten equation and Fredlund et al’s equation were compared. The results showed that the void ratio of the expansive clay at the matric suction of 0.1 kPa was 0.688, 0.678, 0.627 and 0.576 under the geostatic stress of 0, 40, 100, 200 kPa, respectively. The void ratio decreased with increasing matric suction and increasing geostatic stress. During the drying process, the water content of the expansive clay at the geostatic stress of 40, 100 and 200 kPa was lower than that at the geostatic stress of 0 kPa. It suggested that the net normal stress had great influences on the water content-SWCC curves. The influence of the net normal stress on water content in boundary effect zone and transition zone was different. In the first boundary effect zone, the water content almost equaled to the initial water content; In the zone at matric suction of 1-10 kPa, water content decreased rapidly with increasing matric suction; In the zone with matric stuction of 10-100 kPa, the water content change was smaller; In the zone with matric suction of 100-450 kPa, the change of water content tended to the same decreasing trend under the different geostatic stress. The air entry of the SWCC was about 1 kPa at the first peak and 25.25-92.48 kPa at the second peak under the different geostatic stress. For the Korean residual soil, the generalized SWCC could fit well with the R2>0.97 and root mean square error (RMSE) smaller than 1.5%. Based on 3 data points, the SWCC could be well fitted with the R2 of 0.997 and RMSE of 0.55%, indicating the feasibility of using 3 data points to fit SWCC curves. For the diatomaceous earth mixtures, the generalized SWCC model showed high fitting accuracy of R2 higher than 0.97 and RMSE smaller than 2.9%. Under different net confining stress, the generalized SWCC model well fitted the SWCC in the expansive soil with R2 higher than 0.99 and RMSE smaller than 0.28%. Similarly, the generalized SWCC model had high accuracy in fitting drying path and wetting path of Japan Edosaki sandy soil under different dry density and France Bapaume loess. Compared with the van Genuchten and Fredlund et al’s equations, the generalized SWCC model had higher accuracy (R2=0.98 and RMSE<1.3%) in France Bapaume loess and Korean residual soil at the net confining stress of 0 kPa. This study validated the feasibility and reliability of the proposed generalized SWCC model in simulating SWCC under different conditions and in different soils. The model would be helpful in quantitatively describing soil water retention property, permeability, strength and soil erosion mechanism.