Development of dynamic PRA methodology for external hazards in sodium-cooled fast reactor via applying Markov chain Monte Carlo method to severe accident analysis code; Assessment of accident management of assigning independent emergency diesel generators to each air cooler

Li, C.-Y.; 渡部 晃*; 内堀 昭寛; 岡野 靖

Proceedings of 30th International Conference on Nuclear Engineering (ICONE30) (Internet), 10 Pages, 2023/05

Quantitative assessment of the effect of accident management on the various external hazards is essential in the nuclear safety analysis. This study aims to establish the dynamic probabilistic risk assessment methodology for sodium-cooled fast reactors that can consider the transient plant status under continuous external hazards with corresponding countermeasures operating stochastically. Specifically, the Continuous Markov chain Monte Carlo (CMMC) and Deterministic and Stochastic Petri Nets (DSPN) methods are newly applied to the severe accident analysis code, SPECTRA, which can conduct dynamic plant evaluation in the different severe accident conditions of nuclear reactors, to develop an evaluation methodology for typical external hazards. In the DSPN-CMMC-SPECTRA coupled frame, the latest safety functions of the plant components/systems can be stochastically determined by the DSPN-CMMC grounded on the current plant states under continuous hazard and the interaction between the multi-state components/systems; then, SPECTRA can evaluate the following plant state determined by the latest safety function of the components/systems. Therefore, the advantage of this newly developed DSPN-CMMC-SPECTRA frame is having the capability to quantitatively and stochastically evaluate the transient accident progressions that potentially lead to the core damage under the continuous external hazard scenario. As for the preliminary exam on the DSPN-CMMC-SPECTRA frame, one of the typical external hazards of continuous volcanic ashfall is selected in this research. In addition, the numerical investigation of alternative accident management' effects has also been carried out and quantitatively confirmed in this research.