An Improved Time-dependent Reliability Model for Aging RC Bridges Based on GPD Load Distribution

To accurately assess the time dependent reliability of a bridge structure, this work modelled the load using the Generalized Pareto distribution (GPD) and improved the traditional time dependent reliability model. The improved model was validated by a numerical case study with the Monte Carlo simulation (MCS). In addition, the effects of load distributions, load growth rates, live-to-constant load ratios, and resistance degradation functions on the accuracy of the reliability probability were analyzed. Finally, the improved time dependent reliability calculation model was applied to analyze the behaviors of a simply-supported beam bridge. The results confirmed that the improved time dependent reliability model was more accurate than the Generalized Extreme Value (GEV) model, as the GPD model focused on the tail loads that were crucial for structural failure beyond the threshold. The improved time dependent reliability model reduced computational errors caused by uncertainties in load distribution since the computation results were not affected by the overall load distribution. Compared with the GEV model, the improved time dependent reliability model was more sensitive to the load growth rate, and it overcame the shortcomings of the conservative structural safety evaluation. It was worth noting that the accuracy of the improved time dependent model was related to factors such as load growth rate and evaluation base period, but was not affected by the live-to-constant load ratio. To ensure accuracy of structural assessment, we recommend that the base period T should not exceed 10 years. The improved time dependent reliability model demonstrated its ability to accurately capture the time dependent reliability, which can provide valuable insights for safety operation and maintenance strategies of bridges.