Evaluation of Thermal Conductivity Coefficient Prediction Models of Soil and Their Parameter Analysis

Soil thermal conductivity is an important thermophysical parameter for soil heat transfer analysis， widely used in fields involving thermal effects. Accurate prediction of soil thermal conductivity is crucial due to numerous challenges in measuring soil thermal conductivity. In terms of that， this study utilized experimental data on soil thermal conductivity to evaluate the prediction accuracy of four classical models， then examined the impact of model parameters on the predicted values of thermal conductivity， and porosity on the value selecting of model parameters. To enhance the prediction accuracy， this research used the best-performing model to predict the soil thermal conductivity with different soil textures based on the evaluation results. A parameter calculation model was also developed to improve the model's applicability. The results indicated that： 1） the Bi-Zhang-Chen model exhibited the highest overall calculation accuracy among different soils， followed by the Côté-Konrad model， the Lu-Ren-Gong model， and the Chen model. More specifically， the Bi-Zhang-Chen model showed small calculation errors for sandy soil， loess， silty soil， and clay， while the Côté-Konrad model yielded more accurate results for quartz sand soil. 2） Soil porosity had a significant impact on model parameter values in the same region， particularly showing strong regularity in loess. 3） Selecting different prediction models based on soil types can better predict the thermal conductivity of soils with different textures. The parameter calculation model established through model evaluation and parameter analysis can effectively predict the thermal conductivity of sandy soil， loess， silty soil， and clay. This study provides theoretical references for the improvement and establishment of soil thermal conductivity models.