Bayesian‐network‐based risk modeling and inference for structures under a sequence of main and aftershocks

• Published in: Earthquake engineering & structural dynamics 2022-04, Vol.51 (5), p.1058-1075
• Main Authors: Mun, Changuk, Song, Junho
• Format: Article
• Language: English
• Subjects: aftershock; Aftershocks; Bayesian analysis; Bayesian network; Bayesian theory; Earthquake damage; Earthquakes; Fragility; mainshock; Mathematical models; Modelling; Probabilistic inference; Probability theory; Risk management; Seismic activity; Sequencing; sequential earthquakes; Stochasticity; Structural damage; Structures
• ISSN: 0098-8847; 1096-9845
• DOI: 10.1002/eqe.3604

A sequence of main and aftershocks is a complex natural hazard, which may introduce severe damage to critical infrastructure systems. Managing the risk caused by sequential earthquake events is challenging due to the uncertainties of aftershocks and their dependency on the preceding mainshock. Moreover, the dynamic behavior of a structural system under aftershocks must be evaluated with consideration of possible structural damage caused by the mainshock. To overcome these difficulties, this paper proposes a new framework based on a Bayesian network (BN) for risk modelling and inference for structures under a sequence of main and aftershocks. The stochastic mechanisms of sequential earthquakes and the hysteretic behavior of structures are described by BN models, which are then integrated for comprehensive modeling of the risk. The developed BN model facilitates probabilistic inference to obtain fragility curves given various circumstances regarding sequential earthquake events and observations. Numerical examples demonstrate fragility assessment and inference with the proposed BN‐based framework.