Coupled THM modelling of the FE Experiment at Mt. Terri in Decovalex Task C

The theory of non-isothermal Richards flow with mechanics (TRM) assumes that the mobility of gases in unsaturated porous media is always large enough to allow sufficiently fast drainage so that the gas pressure does not increase significantly but remains constant. The benefit of an implementation using this simplified approach is a faster numerical model compared to an implementation using the more general theory of non-isothermal two-phase two-component flow with mechanics (TH²M), which represents the gas phase explicitly. The Full-scale Emplacement (FE) experiment conducted at the Mont Terri Underground Rock Laboratory was designed to simulate an emplacement tunnel for high-level radioactive waste at full scale using the reference repository design of the National Cooperative for the Disposal of Radioactive Waste (Nagra) of Switzerland. In this experiment, which is numerically studied in Task C of the Decovalex-2023 project, water vaporisation in the bentonite backfill near the heaters might potentially increase the gas pressure. Thus, justification for the applicability of Richards' assumption for this environment should be demonstrated before attempting to use a TRM process to model the coupled thermo-hydro-mechanical problem of this experiment. We demonstrate the applicability for the given conditions by comparing model results from OpenGeoSys (OGS-6) obtained by a TRM process with those obtained by a TH²M process implementation. The comparison of the TRM with the TH²M process provides the basis for subsequent extensive analyses using the TRM process of OGS-6 for modelling the FE experiment with large 3D meshes with more than one million degrees of freedom in parameter variation studies. This model has 75 parameters and many initial / boundary conditions, of which single ones or several at a time are varied. By comparing model results of these variations at 200 observation points with each other and to five years of hourly measurement data using statistical methods, sensitivities to single parameters are studied. But also relevant features not included in the current setup are revealed by using parameter settings that act as proxies for these anticipated missing features. These proxies give results that fit well to the observations but are without physical justification themselves. Due to the great possibilities for validation of existing coupled THM models arising thereof, this also leads to significant further development of model capabilities for the use in integrity evaluations of geotechnical and geological barriers.