A global sensitivity analysis of two-phase flow between fractured crystalline rock and bentonite with application to spent nuclear fuel disposal

Abstract

Geological disposal of spent nuclear fuel in deep crystalline rock is investigated as a possible long term solution in Sweden and Finland. The fuel rods would be cased in copper canisters and deposited in vertical holes in the floor of deep underground tunnels, embedded within an engineered bentonite buffer. Recent experiments at the Äspö Hard Rock Laboratory (Sweden) showed that the high suction of unsaturated bentonite causes a de-saturation of the adjacent rock at the time of installation, which was also independently predicted in model experiments. Remaining air can affect the flow patterns and alter bio-geochemical conditions, influencing for instance the transport of radionuclides in the case of canister failure. However, thus far, observations and model realizations are limited in number and do not capture the conceivable range and combination of parameter values and boundary conditions that are relevant for the thousands of deposition holes envisioned in an operational final repository.In order to decrease this knowledge gap, we introduce here a formalized, systematic and fully integrated approach to study the combined impact of multiple factors on air saturation and dissolution predictions, investigating the impact of variability in parameter values, geometry and boundary conditions on bentonite buffer saturation times and on occurrences of rock de-saturation. Results showed that four parameters consistently appear in the top six influential factors for all considered output (target) variables: the position of the fracture intersecting the deposition hole, the background rock permeability, the suction representing the relative humidity in the open tunnel and the far field pressure value. The combined influence of these compared to the other parameters increases as one targets a larger fraction of the buffer reaching near-saturation. Strong interaction effects were found, which means that some parameter combinations yielded results (e.g., time to saturation) far outside the range of results obtained by the rest of the scenarios. This study also addresses potential air trapping by dissolution of part of the initial air content of the bentonite, showing that neglecting gas flow effects and trapping could lead to significant underestimation of the remaining air content and the duration of the initial aerobic phase of the repository.