Post-earthquake assessment of bridges using Bayesian networks with continuous variables

This paper presents a framework for Bayesian Network-based assessment of the seismic vulnerability of bridges. This framework incorporates a fragility function and capacity and demand models. The fragility function and the capacity model are constructed using Hazus. The demand on the bridge is calculated using an attenuation function. The uncertainty terms are considered in both the capacity and the demand model. In order to apply the exact inference algorithm of continuous Bayesian Networks in this framework, all the variables in the framework are assumed to follow Gaussian distribution and the continuous variables are not allowed to have discrete child variables. Using this framework, the correlation between two bridges is analyzed, and the prior probabilities of the two bridges being in a damage state are calculated during the initialization procedures. When an earthquake occurs, if data such as magnitude and epicenter are known, or the damage state of one bridge is observed by on-site sensors, then the probability of another bridge being in a damage state can be predicted and updated. The framework is applied to the bridge networks in Trentino, Italy.