A revisited thermal, hydrodynamic, chemical and mechanical model of compacted bentonite for the entire duration of the FEBEX in situ test

[Abstract:] The safety assessment of geological repositories for radioactive waste requires quantifying the geochemical evolution of the engineered clay barriers where thermal (T), hydrodynamic (H), mechanical (M) and chemical (C) phenomena play a major role during the early stages of the repository. Compacted bentonite is one of the components of the engineered barrier system. The FEBEX (Full-scale Engineered Barrier EXperiment) in situ heating and hydration test was performed at the Grimsel underground laboratory (Switzerland) in two operation periods. The 1st period lasted from 1997 to 2002 when heater 1 was switched off and the area around it was dismantled. Heater 2 was turned off and the full test was dismantled in 2015. Here we present a revisited THCM model of the two operation periods of the FEBEX in situ test, which improves the THCM model reported by Zheng et al. (2011) for the 1st operation period by revising the boundary condition at the heater/bentonite interface, refining the spatial discretization of the finite element mesh near the heater, updating the dispersivities of the bentonite and the granite and revisiting the back-diffusion of solutes from the bentonite barrier into the granite. The concentrations of dissolved Cl− computed with the revisited model at the end of the 1st operation period almost coincide with those of the previous model, except near the heater where the concentrations computed with the revisited model are much larger than those of the previous model. Both models provide a similar fit to the pore water concentration data inferred from aqueous extracts because the model differences occur in a band 0.045 m thick near the heater where there are no measured pore water chemical data. Solute back-diffusion from the bentonite pore water to the granite groundwater is not uniform along the FEBEX gallery due to the heterogeneity of the surrounding granitic formation. The revisited model reproduces the trends of the experimental observations of solute back-diffusion by adopting diffusion coefficients of the bentonite and granite larger than those of the simulation run without the solute back-diffusion. The results of the revisited THCM model match the gravimetric water content and dry density data measured at the end of both operation periods in 2002 and 2015 and fit the general trends of the on line data of temperature and volumetric water content in the bentonite and pore water pressure in the granitic rock collected from 2002 to 2015. The predicted concentrations of the dissolved species in 2002 are sensitive to changes in the vapour tortuosity factor and the retention curve of the bentonite, but are less sensitive to smectite dissolution. The predicted concentrations of the dissolved species at the end of the 2nd operation period in 2015 computed with the revisited THCM model are large near the heater and decrease towards the bentonite/granite interface. The concentrations are generally smaller than those computed at the end of the 1st operation period, except for the concentration of dissolved HCO3– which increases due to a calcite dissolution front. The predicted pH at the end of the 2nd operation period is similar to that of the 1st operation period. The concentrations of dissolved Cl− predicted at the end of the 2nd period with the revisited model reproduce the sharp increase of the measured Cl− data near the heater and match the data measured in 5 out of 6 radial distances.