Uncertainty and sensitivity analysis of a postulated severe accident in a generic German PWR with the system code AC2

Abstract

Severe accident analysis is an important component in ensuring high standards of safety in nuclear power plants. Since the accident in the Fukushima-Daichi NPP even more attention has been paid to this highly complicated and complex field. Numerical simulation tools are widely used to analyse postulated accident sequences, including severe accidents, as well as the evaluation of their possible radioactive impact on the environment. The system code package AC² developed by GRS can simulate the whole reactor in detail, both the core region with the RCS, as well as the containment, starting from normal operational conditions up to severe accidents including core melting, making it a highly valuable tool. At the same time, such tools and their models are developed based on a limited number of experiments and available data, particularly models related to severe accident phenomena. Therefore, it is of great interest to be able to evaluate their accuracy and uncertainty on their respective application fields. In this paper an approach developed and tested on the Phébus FPT1 experiment is applied to a reactor scenario in a generic German PWR to assess source term related uncertainties. A full scale AC² simulation (ATHLET-CD/COCOSYS) is carried out for a medium break LOCA with station blackout in a generic German PWR and used as a best estimate case for the BEPU analysis. The uncertainty and sensitivity analysis focuses on source term related phenomena and figures of merit. Altogether over 80 uncertain input parameters directly related to the modelling of fission product behaviour are considered. In addition to the 95/95 tolerance limits, sensitivity measures (Spearman Rank Correlation Coefficient) are derived to further analyse the dependency of the simulation results on different input parameters.