ABSTRACT
In this paper, an combined approach based on response surface sampling and importance sampling is developed to efficiently assess the structural reliability of a corroding concrete bridge girder. The approach employs cubic surface sampling with smart parameter updating and directed importance sampling to limit the computational cost of the reliability assessment. The case study considers a post-tensioned concrete bridge girder undergoing severe corrosion. Multiple limit states and various stochastic parameters related to the material and damage properties are considered in a finite element based reliability assessment. The combined approach limits the required number of simulations as much as possible and yields relatively quick convergence. The results show that the corrosion level and compressive strength play a significant role in the structural behavior. Given the high post-tensioning loads, the girder is limited by its ultimate capacity rather than serviceability constraints.
