Parametric calibration of a thermo-hydro-mechanical coupled swelling model for viscous soil using Bayesian methods

Following the large-scale development and utilisation of underground space, simulation of the freezing process based on multi-field coupling theory has been widely applied in artificially frozen tunnelling. However, the inherent variability of geotechnical parameters often leads to significant discrepancies between numerical results and monitoring data. To address this issue, a Bayesian method for calibrating the key geotechnical parameters of cohesive soil in Thermo-Hydro-Mechanical (THM) coupling calculations is proposed. The shield tunnel project of Metro Line 18 is taken as the objective in Shanghai, China, which passes underneath the existing Metro Line 10 at Guoquan Road Station. The bearing soil layer ⑤1 of grey clay is investigated for the impact of the ground freezing technique by the underpassing tunnels. The THM model is improved by treating void ratio, freezing temperature, and average water pressure as coupling variables, which can take water migration and freeze-thaw deformation into account. Based on stochastic mechanics-Bayesian method, key geotechnical parameters are inversely calibrated to reduce intrinsic uncertainty. Finally, numerical simulation of the coupled THM fields is conducted by the weak form module of COMSOL Multiphysics software. Results show that the simulated values are in coincidence with the monitored data and below the preset thresholds, which verifies the rationality of the coupling theory. This research provides methodological support for analysing the frozen tunnelling underneath an existing Metro station by multi-physical field coupling theory.