Applicability of Monte Carlo Method with Different Sampling Techniques for Seismic Safety Evaluation of Nuclear Power Plants

Fault-tree and Event-tree analysis is widely utilized in safety engineering and is almost exclusively relied upon in Probabilistic Safety Assessment (PSA) for nuclear power plants (NPPs). In this approach, Binary Decision Diagram (BDD), which can obtain exact solutions in theory, was introduced as an alternative for a traditional approximation method, namely Minimal Cut Set (MCS) method. However, BDD method is still not always applicable without the truncation or simplification of PSA model. On the other hand, Monte Carlo (MC) method has unique advantages for seismic PSA: it enables the direct use of response data in risk computation and consideration of correlation by expert opinions. In this paper, the applicability of MC method for seismic PSA is examined by comparing the results of a seismic PSA study by MC method with that by MCS and BDD methods. The analysis of the results shows that MC method is an efficient method which can converge to the exact solution within reasonable time by both random and Latin hypercube sampling (LHS). Also, this paper proposes hazard proportional sampling (HPS) for MC method and confirms the faster convergence of the proposed sampling technique.