LAUSR

Abstract In France, nuclear energy is extensively used for electricity production, and consequently high-level radioactive waste are generated by the fleet of nuclear powerplants. This waste is planned to be disposed of in a deep geological repository. Engineers have only a limited feedback from similar disposal facilities, as this project is innovative. This lack of knowledge and the timescales involved induce uncertainties about the disposal environment and about the conditions of operation. Reliability analysis is used here as it provides a suitable framework to deal with such uncertainties. This paper focuses on the overpack, which isolates the waste from the environment during the early stage of disposal (at least 500 years). The failure probability is investigated through a crack propagation criterion using a Monte Carlo analysis performed on two levels. The approach is based on a dedicated stochastic structural model including the corrosion rate, the mechanical loading and initial crack conditions. The robustness of this approach is evaluated through sensitivity analysis, and by considering multiple distributions for the variables elicited from expert’s opinion.