The 13th International Conference on Structural Safety and Reliability (ICOSSAR 2021-2022)
Abstract
The probabilistic analysis is sometimes emphasized to explain an effect of probabilistic structural parameters, but it may be expensive especially when the finite element (FE) model is introduced. To this problem, the framework for probabilistic FE analysis of prestressed concrete (PSC) girder considering pit corrosion was proposed in this study, which consists of three steps: surrogate model construction, random sample generation, and goodness-of-fit test. Once an FE model is built, a cost-efficient surrogate model is constructed, and the random samples are generated using the Monte Carlo sampling method. Then, the best-fit probability distribution is selected by conducting a goodness-of-fit test so that the analysis results can be considered the random variables. The proposed framework was applied to the probabilistic analysis of corroded strands and PSC girders, and the best-fit probability distributions for the ultimate load and displacement were suggested.