A COMSOL-based numerical approach to improve heat-pulse measured frozen soil thermal properties

Abstract

Heat pulse (HP) is the most widely used transient technique determining soil thermal properties (STPs) in unfrozen conditions, yet its application to frozen soils introduces significant challenges. At high subfreezing temperatures (−5 to 0 °C), the HP measurements induce thawing and refreezing of ice, dynamically altering the frozen soil thermal properties (FSTPs) being measured. Conventional analytical solutions fail to account for these phase change effects, leading to substantial errors in estimation. Although various approaches have been developed to improve FSTPs determination, achieving accurate measurements remain challenging. This study employed a COMSOL-based numerical model to solve heat conduction equations incorporating latent heat and compared the results with that obtained with traditional analytical solutions. The results revealed that analytical solutions consistently underestimate frozen soil thermal conductivity (FSTC) at temperatures above −3 °C, even with optimized heating strategies. Numerical simulations demonstrated that phase transition parameters critically influence temperature evolution, particularly above −5 °C, and the COMSOL improved FSTC estimates between −4 and 0 °C, though the performance depended on heating strategies. To facilitate parameter selection, linear regression models were derived for phase transition interval (ΔT_p, R² = 0.37) and phase transition point (T_pc, R² = 0.30). These advancements enhance the accuracy of HP-measured FSTPs, providing a more reliable approach for cold-region researches and applications.