ABSTRACT
Safety assessment of aging bridges may require the modeling of complex deterioration processes, further exacerbating in time the uncertainties involved in the mechanical response of critical members. Simulation-based techniques applied to life-cycle reliability problems may rely on time-consuming structural analyses, leading to untenable computational efforts and impairing the accuracy of the numerical estimates. This paper discusses an efficient computational framework for time-variant failure probability based on Importance Sampling (IS) with Stationary Proposal (SP) distribution. This novel methodology exploits numerical approaches traditionally developed for time-invariant structural reliability problems and extends them into a life-cycle context. Specifically, the simulation density is adaptively selected by minimizing the Kullback–Leibler Cross-Entropy from a chosen family of parametric distributions. The numerical procedure is applied to lifetime structural reliability of a reinforced concrete bridge exposed to chloride-induced corrosion.
