Full scale 3D modelling of a repository in the Callovo-Oxfordian clay with focus on two-phase transport of 14C and 129I

Document available in extended abstract form only. Andra studies the long-term safety performance of a planned repository for high-level and intermediate-level long-lived radioactive waste situated at around 500 metres below the ground level in the Callovo-Oxfordian clay in north-eastern France. The envisaged repository encompasses a horizontal area of 10 to 20 square kilometres and consists of several thousand waste emplacement cells connected by numerous drifts and shafts for access and operation. The construction and operation of the repository will alter the natural conditions and the properties of the affected host rock. Moreover, the generation and release of heat and gas by the stored waste, which continues for several thousand years into the post-operational phase, will persistently influence the evolution of the physical state inside the repository and surrounding host rock and potentially decrease the retention capacities of the engineered barriers and host rock for the radionuclides released by the waste. Important processes taking place are two-phase flow (gas, liquid) with advective and diffusive transport of water, vapour, hydrogen, and radionuclides, (de-)sorption, evaporation/condensation, volatilisation/dissolution, radionuclide decay, and heat dissipation by convection, conduction, and radiation. Driving forces are pressure differences, viscous, buoyant, and gravity forces. Liquid and gas phases interact by saturation dependent capillarity and relative permeability. Porosity may change due to thermal expansion and compressibility. The involved processes take place on different spatial scales - from the centimetre scale of a waste package to the kilometre scale of the entire repository - and on a large time scale of several hundred thousand years. We developed a model concept to assess the release and transport of heat, gas, and radionuclides from the planned repository by numerical simulation that considers the relevant physical processes and the impact of construction. We consider the general repository architecture of the drifts and shafts (Figure 1); the amount, type, and arrangement of the waste packages; the parametric uncertainty of selected flow and transport properties, and of the heat, gas, and radionuclide sources. We performed a series of simulation cases using a modified version of TOUGH2-MP EOS7R that accounts for hydrogen instead of air and that can account for two individual radionuclides, here chosen to be 14C and 129I. The simulation results illustrate the transfer paths of the radionuclides from the waste into the host rock and along the backfilled drifts and shafts to the model boundaries (Figure 1) and allow determining and quantifying the dominant radionuclide transport paths for different conditions and scenarios such as early canister failure and varied gas generation regimes. The study and simulation results demonstrate the feasibility of simulating the coupled thermohydraulic two-phase flow and radionuclide transfer processes on the scale of the entire repository and the embedding host rock between the Oxfordian and the Dogger. Numerical sensitivity analyses and parameter studies help to identify the most relevant parameters and mechanisms influencing radionuclide transport from the waste to the environment and to quantify their impact on the full scale of the repository. The task, the novel model concept, and an overview of the simulation results will be presented. Furthermore, we assess the robustness of the repository layout by comparing the impact of the variation of selected parameters on the modelled radionuclide transport and release. (authors)