Differential assessment of effects of increasing model complexity in THM coupled models of the FE experiment at Mt. Terri

Abstract

The Full-scale Emplacement (FE) experiment is studied numerically using large 3D meshes with over one million elements and degrees of freedom. Non-isothermal Richards flow coupled with mechanics is the applied physical process model. Systematic model variations are performed to generate results for a systematic comparison to a large set of measurement data utilising statistical methods. These variations are based on several hypotheses concerning potential sources of uncertainty or mismatch between modelled and measured data. Among them are geometrical aspects as well as aspects of model complexity, such as including additional materials or physical processes. The geometrical aspects include the effects of varying the inner tunnel radius, changing the heater representation, examining the uncertainty of the sensor locations and investigating geometrical aspects of adding a shotcrete layer. Complexity is increased by including shotcrete and a simplified excavation damaged zone (EDZ) as additional materials, by accounting for a ventilation phase and by providing an excavation-induced non-homogeneous initial liquid pressure field. We discuss the nature and magnitude of the effects of each of these aspects as well as their combinations. Notably, we find a large influence of the geometrical aspects on the temperature results in the tunnel on the one hand and on the other the importance of the initial liquid pressure field in the Opalinus clay. Overall, the comparison to the measurement data can be improved by incorporating the investigated aspects into models of the FE experiment.