Prediction of thermal conductivities of insulation materials under water immersion and freeze–thaw conditions for tunnels in cold regions

Abstract

The processes of groundwater flow and freeze–thaw cycles negatively affect the thermal performances of insulation materials in cold-region tunnels and notably reduce their long-term service performances. Therefore, this study investigated the water absorption and microscopic characteristics of three insulation materials (namely, polyphenolic materials, polyurethane, and polystyrene) under water immersion and freeze–thaw cycles based on indoor tests, and a model was developed to calculate the thermal conductivity considering environmental factors. Four machine learning methods were also used to verify its accuracy in thermal conductivity prediction, and model evaluation and multi-factor analysis were performed. The results showed that the mass water contents of the three materials were positively correlated with the head pressure and number of freeze–thaw cycles, and the maximum values for the polyphenolic insulation material, polyurethane, and polystyrene reached 1585.37 %, 373.42 %, and 286.86 %, respectively. Compared with the values in the initial dry state, the thermal conductivities of the three materials with the maximum water content increased 11.52, 3.00, and 0.67 times in the thawed state, respectively, while in the frozen state, they increased 26.59, 6.25, and 2.45 times, respectively. The thermal conductivities of the insulation materials had good correlations with the water contents, and the introduction of the environmental factor could improve the accuracy of the calculation model. The backpropagation neural network had the highest accuracy for predicting the thermal conductivity, and the order of importance of the factors was as follows: frozen and thawed states > mass water content > head pressure > material type > freeze–thaw cycles. The results can provide guidance for the selection and design of the insulation materials for tunnels in cold regions.