ICONE23-1293 APPLICATION OF SUBSET SIMULATION METHODS TO DYNAMIC FAULT TREE ANALYSIS

Meng-Yun Liu,Ding She,Jing-Quan Liu

DOI: https://doi.org/10.1299/jsmeicone.2015.23._icone23-1_139

2015-01-01

Abstract:Although fault tree analysis has been implemented in the nuclear safety field over the past few decades, it was recently criticized for the inability to model the time-dependent behaviors. Several methods are proposed to overcome this disadvantage, and dynamic fault tree (DFT) has become one of the research highlights. By introducing additional dynamic gates, DFT is able to describe the dynamic behaviors like the replacement of spare components or the priority of failure events. Using Monte Carlo simulation (MCS) approach to solve DFT has obtained rising attention, because it can model the authentic behaviors of systems and avoid the limitations in the analytical method. In this paper, it provides an overview and MCS information for DFT analysis, including the sampling of basic events and the propagation rule for logic gates. When calculating rare-event probability, large amount of simulations in standard MCS are required. To improve the weakness, subset simulation (SS) approach is applied. Using the concept of conditional probability and Markov Chain Monte Carlo (MCMC) technique, the SS method is able to accelerate the efficiency of exploring the failure region. Two cases are tested to illustrate the performance of SS approach, and the numerical results suggest that it gives high efficiency when calculating complicated systems with small failure probabilities.