ABSTRACT The phenomenon of water–ice phase transition in frozen soils is the key to explaining the mechanism of frost heaving and thawing settlement disaster. However, numerical analysis of water–ice phase transitions in frozen soils at mesoscale is rarely reported. This study combines the modified Gibbs-Thomson equation and the enthalpy-based lattice Boltzmann model, and develops a mesoscale numerical method to simulate the water–ice phase transition in the local pores of saturated frozen soil during freezing and thawing process. In order to accurately simulate this process, a new η coefficient is proposed to modify the Gibbs-Thomson equation, which is unrelated to the type of soil sample and gradation characteristics. According to the particle size distribution curve, the two-dimensional random circle generation method is used to reconstruct the soil pore structure. The freezing and thawing processes are verified with the experimental data. A larger non-uniformity coefficient C u and curvature coefficient C c of the grain size distribution result in a lower degree of subcooling and lower residual water content of the soil during the freezing process. The new model provides an effective means to understand the water–ice phase transition in frozen soil at a mesoscopic scale.
Read full abstract