Probabilistic fatigue assessment based on Bayesian learning for wind-excited long-span bridges installed with WASHMS

Abstract

For long-span bridges located in wind-prone regions, it is a trend to install in situ Wind and Structural Health Monitoring System (WASHMS) for long-term real-time performance assessment. One of the functions of the WASHMS is to provide information for the assessment of wind-induced fatigue damage. Considering the randomness of wind, it is more reasonable to describe wind-induced fatigue damage of bridge in a probabilistic way. This paper aims to establish a probabilistic fatigue model of fatigue damage based on Bayesian learning, and it is applied to a wind-excited long-span bridge installed with a WASHMS. Wind information recorded by the WASHMS is utilized to come up with the joint probability density function of wind speed and direction. A stochastic wind field and subsequently wind-induced forces are introduced into the health monitoring oriented finite element model (FEM) of the bridge to predict the statistics of stress responses in local bridge components. Bayesian learning approach is then applied to determine the probabilistic fatigue damage model. The Tsing Ma suspension bridge in Hong Kong and its WASHMS are finally utilized as a case study. It shows that the proposed approach is applicable for the probabilistic fatigue assessment of long-span bridges under random wind loadings. © 2013 Zhi-Wei Chen and Xiao-Ming Wang.