Buffeting reliability assessment of long-span bridge with multi-uncertainty coupling based on probabilistic response surface method

Buffeting reliability analysis plays an important role in the safety assessment of long-span bridges. A buffeting reliability assessment approach considering multi-uncertainty coupling is proposed for the long-span bridge in this paper. The time domain buffeting analysis is adopted to capturing the stochastic nature of the wind force and the structure nonlinearities. A novel probabilistic response surface method (P-RSM) is proposed to capture the structural uncertainty and the epistemic uncertainty (arising from the assessment process) by combining the traditional response surface model and probabilistic modeling technique. The proposed algorithm is applied to the buffeting reliability analysis of the example long-span bridge. The nested probabilistic models are generated for the displacement standard deviation, the velocity standard deviation, and the static wind displacement. The buffeting reliability with multi-uncertainty coupling is obtained with the nested probabilistic models and compared with the buffeting reliability under deterministic structural parameters. The results show that the generated P-RSM models not only have good model performance and accuracy, but also well and precisely capture the most sensitive model influencing factors. The introducing of multi-uncertainty coupling not only decreases the buffeting reliability level, but also turns it into a random variable. Considering the multi-uncertainty coupling effect plays an important role and is quite necessary in the dynamic buffeting reliability analysis for long-span bridges.