Experimental study and simulation of thermal conductivity of saturated frozen soil

Abstract

The aim of this study was to enhance the accuracy of predicting the temperature field of frozen soil and to reduce the workload of thermal parameter testing. To achieve this, we employed a three-phase model comprising soil, water, and ice. The unfrozen water content in frozen soil at varying temperatures was determined using nuclear magnetic resonance spectroscopy (NMR), while the thermal conductivity was measured by a thermal characteristic analyzer. A MATLAB software-based random model of the frozen soil was then established and imported into COMSOL simulation software. The repeatability and reproducibility of the established model were verified by varying the proportions of pore water and frozen ice to determine the degree of simulation accuracy.The results demonstrated that the unfrozen water content maintained a dynamic equilibrium relationship with temperature, which influenced the thermal conductivity of frozen soil. The simulation results were consistent with those obtained from instrument measurements of trends with respect to temperature. The average PBIAS value between the calculated and measured values was 0.0139, indicating theoretical feasibility. Comparison with experimental data confirmed the effectiveness of our approach, providing a novel concept and a simple method for predicting the temperature field of frozen soil engineering in areas that experience seasonal freezing.