Investigating thermal effects on bentonite buffer's swelling and hydration response: A coupled thermo-hydro-mechanical modelling study

Comprehending coupled thermo-hydro-mechanical (THM) processes in geomaterials is essential for addressing various geoenvironmental problems. Predicting the THM response of bentonite buffers in nuclear waste repositories (NWRs) requires accounting for temperature-sensitive hydraulics that affect swelling behaviour. In this study, a temperature-dependent soil water retention curve (SWRC) model was formulated to attain an implicit thermo-hydraulic coupling in the THM governing equations. Furthermore, a parameter-sensitivity study was conducted to investigate the influence of the SWRC fitting parameters on the bentonite’s water retention capacity. Building on this constitutive law, the study developed a simplified, rigorous THM framework that augments Richards’ flow equation with explicit vapour diffusion, employs the developed non-isothermal SWRC relation and temperature-dependent fluid properties to characterise the bentonite’s hydration response at elevated temperatures, and couples hydraulics to the Extended Barcelona Basic Model (BBMx) to predict suction-driven swelling pressures. The performance of the proposed framework was demonstrated through two numerical scenarios: (i) a laboratory-scale domain with a simplified representation of NWR conditions under coupled thermo-hydraulic gradients, and (ii) an isothermal scenario focused on evaluating swelling pressures. Results revealed that incorporating non-isothermal hydraulic constitutive relationships improved suction predictions and precisely corroborated experimental relative humidity data. The thermal dependence of the SWRC was identified as the key sensitive parameter governing the hydraulic flow and mechanical behaviour. Neglecting thermal effects led to an overestimation of swelling pressure, underscoring the importance of thermal-hydraulic interactions in governing the mechanics. The proposed formulation captured the THM behaviour with accuracy comparable to complex two-phase models, with significantly lower computational demands, making it adaptable to other geoenvironmental systems.