A Bayesian network-based probabilistic framework for updating aftershock risk of bridges

The evaluation of a bridge's structural damage state following a seismic event and the decision on whether or not to open it to traffic under the threat of aftershocks (ASs) can significantly benefit from information about the mainshock (MS) earthquake's intensity at the site, the bridge's structural response, and the resulting damage experienced by critical structural components. This paper illustrates a Bayesian network (BN)-based probabilistic framework for updating the AS risk of bridges, allowing integration of such information to reduce the uncertainty in evaluating the risk of bridge failure. Specifically, a BN is developed for describing the probabilistic relationship among various random variables (e.g., earthquake-induced ground-motion intensity, bridge response parameters, seismic damage, etc.) involved in the seismic damage assessment. This configuration allows users to leverage data observations from seismic stations, structural health monitoring (SHM) sensors and visual inspections (VIs). The framework is applied to a hypothetical bridge in Central Italy exposed to earthquake sequences. The uncertainty reduction in the estimate of the AS damage risk is evaluated by utilising various sources of information. It is shown that the information from accelerometers and VIs can significantly impact bridge damage estimates, thus affecting decision-making under the threat of future ASs