A coupled THMC model of a heating and hydration laboratory experiment in unsaturated compacted FEBEX bentonite

Unsaturated compacted bentonite is foreseen by several countries as a backfill and sealing material in high-level radioactive waste repositories. The strong interplays between thermal (T), hydrodynamic (H), mechanical (M) and chemical (C) processes during the hydration stage of a repository call for fully coupled THMC models. Validation of such THMC models is prevented by the lack of comprehensive THMC experiments and the difficulties of experimental methods to measure accurately the chemical composition of bentonite porewater. We present here a non-isothermal multiphase flow and multicomponent reactive solute transport model for a deformable medium of a heating and hydration experiment performed on a sample of compacted FEBEX bentonite. Besides standard solute transport and geochemical processes, the model accounts for solute cross diffusion and thermal and chemical osmosis. Bentonite swelling is solved with a state-surface approach. The THM model is calibrated with transient temperature, water content and porosity data measured at the end of the experiment. The reactive transport model is calibrated with porewater chemical data derived from aqueous extract data. Model results confirm that thermal osmosis is relevant for the hydration of FEBEX bentonite while chemical osmosis can be safely neglected. Dilution and evaporation are the main processes controlling the concentration of conservative species. Dissolved cations are mostly affected by calcite dissolution-precipitation and cation exchange reactions. Dissolved sulphate is controlled by gypsum/anhydrite dissolution-precipitation. pH is mostly buffered by protonation/deprotonation via surface complexation. Computed concentrations agree well with inferred aqueous extract data at all sections except near the hydration boundary where cation data are affected by a sampling artifact. The fit of Cl⁻ data is excellent except for the data near the heater. The largest deviations of the model from inferred aqueous extract data occur for dissolved [formula removed] which is underpredicted by the model. There are uncertainties on the amount of gypsum available for dissolution and its dissolution mechanism (kinetics or local equilibrium).