A Bayesian Network-based decision framework for managing bridge scour risk

22 bridges fail every year on average in the USA because of scour, whereas in the UK soil erosion was the cause of 138 collapses of bridges in the last century. These digits fully explain why flood-induced scour is the leading cause of bridge failures worldwide. Scour assessments are currently based on visual inspections, which are expensive, and provide qualitative and subjective information. SHM offers the possibility to measure scour depth at any location of a bridge network; yet monitoring an entire infrastructure network is not economically feasible. In this paper, we propose a Decision Support System (DSS) for bridge scour management that achieves a more confined estimate of the scour risk for a bridge network through a probabilistic approach. A Bayesian Network (BN) is used to estimate, and update, the scour depth using real-time information from limited number of scour monitoring systems (SMSs) and river flow characteristics. Data collected by SMSs and BN's outcomes are then used to inform a decision model and thus support transport agencies' decision frameworks. The idea is to use this information to update the scour threshold after which bridges are closed. An infrastructure network, consisting of three bridges in Scotland, is built to test the functioning of the DSS. They cross the same river and Only one bridge is instrumented with a SMS. The BN is found to estimate the scour depth at unmonitored bridges and the decision model provides higher values of scour threshold compared to the ones implicitly chosen by transport agencies.