Efficient sampling techniques for simulation-based life-cycle structural reliability and seismic fragility assessment

Life-cycle structural reliability assessment and risk analysis of deteriorating systems may involve the modeling of complex time-variant probabilistic processes. Although simulation methods are frequently the only viable tools to solve this kind of problems, they are time-consuming and might be computationally inefficient and unfeasible in practice if small probabilities of failure need to be estimated, particularly for large-scale reliability and risk analysis problems. To mitigate the computational effort of simulation methods in estimating the time-variant failure probability of deteriorating structures, a novel computational approach based on Importance Sampling and clustering-based data reduction techniques is proposed. The computational efficiency of the proposed methodology is demonstrated with practical applications to life-cycle reliability and seismic fragility of reinforced concrete structures exposed to corrosion.