Abstract
The necessity of effective method to estimate low failure probability is increased from the standpoint of life cycle cost, risk and probabilistic safety assessment (PSA). We proposed an effective Monte Carlo Simulation method combining with the progressive subspace concept which searches distributed failure event regions step by step. Samples to search the failure region are generated in each subspace by using MCMC (Markov Chain Monte Carlo). Efficiency of MCMC is improved by simplifying the sampling procedure based on a uniform distribution from which it is very easy to transform into arbitrary distributions such as normal and log-normal distributions. Effectiveness of the proposed method is verified through numerical examples with simple limit state functions, then fragility curves of a bridge pier and a slope are calculated by the proposed method.
