Optimization strategy for SAM in nuclear power plants based on NSGA-II

Abstract

Abstract The Severe Accident Management Guide (SAMG) is an important component of nuclear safety regulations. Many studies are being conducted to optimize severe accident management (SAM) strategies. To ensure the safety of nuclear power plants, decision makers need to monitor multiple parameters with security threats. Therefore, it is particularly important to search optimal SAM strategies under different numbers of mitigation targets. The Non-dominated Sorting Genetic Algorithm-II (NSGA-II) is an evolutionary algorithm that does not require derivative differentiation and is capable of population search. In this study, a nuclear power plant accident optimization strategy is developed using the Modular Accident Analysis Program (MAAP) in conjunction with NSGA-II. The strategy enables decision makers to consider multiple mitigation objectives in a complex decision environment. Focusing on the CPR1000, this study applies the optimization strategy to automatically search for optimal mitigation strategies for small break loss of coolant accident (SBLOCA) and station blackout hot leg creep rupture accidents (SBOHLCR). Comparing the optimization results with the basic accident sequence, it is found that the reactor pressure vessel (RPV) failure time is delayed from 72,702 s to 128,730 s under SBLOCA and from 23,828 s to 28,363 s under SBOHLCR. This study has also verified that the optimal SAM strategy obtained by the strategy through dual objective optimization has better mitigation effects than a strategy that only considers one objective. This optimization strategy has the potential to be applied to other types of severe accident management studies in the future.