Experimental Study on Moisture Migration in Unsaturated Loess under Effect of Temperature

In this paper, moisture migration in loess considering temperature effect is studied by tests on unsaturated loess samples with different densities and initial moisture contents. Test results reveal that obvious changes in moisture content distribution in a loess sample can be observed after temperature difference is exerted on the two ends of the sample. Moisture content at the cold end increases and that at the hot end decreases. Under the effect of temperature difference, moisture content difference at the two ends of a soil sample is related to the initial moisture content, soil density, and magnitude of the temperature difference. Generally speaking, larger temperature differences and smaller soil densities result in more obvious moisture migration and larger moisture content differences at the two ends of the soil sample. When the initial moisture content is large, the moisture content difference caused by a temperature difference is small; when the initial moisture content is small, the moisture content difference caused by a temperature difference is also small; when the initial moisture content is moderate, the moisture content difference caused by a temperature difference is large. After the analysis of test results, taking the soil density and moisture content into account, a formula is obtained to determine the moisture content gradient resulting from the temperature gradient. Reliability of the formula is verified by comparing the measured and calculated data. Because of the reverse migration of liquid water and water vapor at the end of the experiment, it is difficult to determine the thermal potential and matrix potential. Based on the experimental data, this paper probes into the water potential equation that can be used for stability analysis. The equation considers the comprehensive impact of soil density, temperature gradient, moisture content, and moisture content gradient on water potential. It only applies to analyze stable distributions of temperature and does not apply to unstable temperature distributions.