Efficient determination of frozen soil hydraulic conductivity via transient-state computational methods

The frozen soil hydraulic conductivity (FSHC, k w ( T )) is a fundamental constitutive parameter for the high-fidelity thermo-hydro-mechanical (THM) modeling of cold-region geotechnical structures. However, conventional steady-state methods are time-consuming, often requiring hours to reach equilibrium, and provide only a limited number of data points at discrete temperatures. These limitations significantly restrict their integration into advanced computational solvers. This study proposes transient- state computational methods based on one-dimensional freezing tests: (1) Freezing Energy Demand Analysis Method (FEDM), utilizing energy conservation and temperature fields for flow derivation; (2) Freezing Front Advancing Method (FFAM), determining velocities via ice content evolution at the advancing front. In addition, the proposed methods are examined through complementary validation, including comparison with an independent dataset, cross-verification using an alternative method based on suction–temperature equivalence, and comparison with published steady-state data. Theoretical analysis indicates a practical complementarity between FEDM and FFAM. FEDM is based on spatial differentiation and requires relatively slow freezing rates and dense temperature measurements, whereas FFAM relies on temporal differentiation and benefits from rapid freezing and high-frequency data acquisition. Experimental results show that two methods yield consistent k w ( T ) curves, which are validated through multiple independent approaches. For the tested silty clay, a distinct “residual point” is observed at approximately − 6 °C, which may indicate a transition in seepage behavior from capillary-dominated flow to adsorbed-film-dominated flow under the present conditions. In addition, the results show a dependence of FSHC on initial saturation, where the initial water saturation influences the evolution of k w ( T ). These transient methods reduce the testing duration from weeks to one day, providing high-efficiency computational method for characterizing water migration and frost heave in frozen soils.