A Bayesian approach for condition assessment and damage alarm of bridge expansion joints using long-term structural health monitoring data

Abstract

Premature failure of bridge expansion joints has been increasingly observed in recent years, and nowadays it becomes a major concern of bridge owners. A better understanding of their performance in service is highly desired. Deterministic linear regression models between bridge temperature and expansion joint displacement have widely been adopted to characterize the in-service performance of bridge expansion joints. When such a regression pattern is elicited using real-time monitoring data, the deterministic models fail to account for uncertainty inherent in the monitoring data and interpret the model error. In this study, a probabilistic approach for characterization of the regression pattern between bridge temperature and expansion joint displacement by use of Structural Health Monitoring (SHM) data and for SHM-based condition assessment and damage alarm of bridge expansion joints is developed in the Bayesian context. The proposed approach enables to account for the uncertainty contained in the monitoring data and quantify the model error and the prediction uncertainty. By combining the Bayesian regression model and reliability theory, an anomaly index is formulated to evaluate the health condition of the expansion joint when newly collected monitoring data are available and to provide damage alarm once the probability of damage exceeds a certain threshold. In the case study, real-world monitoring data acquired from a cable-stayed bridge are used to illustrate the proposed approach, including examining the appropriateness of the design values of expansion joint displacements under extreme temperatures in serviceability limit state.