Reliability analysis of bridge girders based on regular vine Gaussian copula model and monitored data

Abstract

For improving the reliability analysis of bridge girders, a consideration of the nonlinear dependence among multivariable random variables is essential. Thus, this study presents a new reliability analysis method for bridge girders, and the regular vine Gaussian copula model (RVGCM) combining the regular vine structure with multiple bivariate Gaussian copulas is established for modeling the nonlinear dependence between the failure modes at different control monitoring points. With the established RVGCM, the failure problem of the bridge girder can be simplified into a series of bivariate failure problems. The nonlinear dependence among different pairs of variables, which may affect the reliability evaluation, can be captured and described. The feasibility of the proposed method is demonstrated using a practical bridge girder with multiple control monitoring points. The reliability analysis results using the RVGCM, statistical independent method (statistical independence between the failure modes), and Importance Sampling (IS) simulation approach are compared. It is proven that the RVGCM is superior in modeling dependence between the failure modes for reliability analysis of bridge girder. The reliability analysis results of the RVGCM are closer to the benchmark results obtained from IS than the statistical independent method, and the calculation efficiency is better than IS. In addition, the results of the proposed method are consistent with the inspection results of actual bridges, which proves the practical application potential of the proposed method.