Probabilistic assessment of multiphysics-driven stochastic corrosion for the safety evaluation of nuclear waste canisters
Über dieses Buch
Ensuring the long-term safety of deep geological disposal systems requires reliable prediction of corrosion in nuclear waste canisters. Their interfacial corrosion is governed by coupled thermal-hydro-chemical (THC) processes and is strongly affected by uncertainties arising from heterogeneous materials, spatially variable conditions and environmental parameters. However, existing numerical approaches face substantial challenges when addressing large computational domains, high-dimensional random inputs, and long-term simulations required to represent geological time scales. This study presents an efficient stochastic finite element framework for quantifying the probabilistic evolution of corrosion depth at canister interfaces. The framework solves coupled stochastic THC PDEs and models the interface recession as an evolving random field. Uncertainty in initial conditions, material-environment interaction coefficients and corrosion mechanisms is represented through structured high-dimensional random inputs. The proposed method separates the spatial, temporal and random components of the solution space, which enables scalable and computationally efficient simulations while maintaining accuracy in estimating probability distributions and ensemble statistics of corrosion progression. The resulting stochastic solutions further facilitate reliability evaluation, sensitivity analysis and inverse identification of key corrosion parameters for safety assessment.
