A Bayesian Network (BN) methodology is developed for performing infrastructure seismic risk assessment and providing decision support with an emphasis on immediate post-earthquake applications. A BN is a probabilistic graphical model that represents a set of random variables and their probabilistic dependencies. The variables may represent demand or capacity values, or the states of components and systems. Decision and utility nodes may be added that encode various decision alternatives and associated costs, thus facilitating support for decision-making under uncertainty.
BNs have many capabilities that make them well suited for the proposed application. Most important among these is the ability to update the probabilistic states of the variables upon receiving relevant information. Evidence on one or more variables, e.g., measured component capacities or demands, or observed states of components, can be entered into the BN and this information propagates throughout the network to provide up-to-date probabilistic characterizations of the infrastructure components and system as well as optimal ordering of the decision alternatives. This can be done in near-real time and under the uncertain and evolving state of information that is characteristic of the post-event period. As is the case with most computational methods, BNs have their limitations: calculations can be highly demanding when the BN is densely connected, or when the infrastructure system is complex and large. This study addresses these challenges.
The proposed methodology consists of four major components: (1) a seismic demand model of ground motion intensity as a spatially distributed random field, accounting for multiple sources and including finite fault rupture and directivity effects, (2) a model for seismic performance of point-site and distributed components, (3) models of system performance as a function of component states, and (4) models of post-earthquake decision-making for inspection and operation or shutdown of components.
Two example applications demonstrate the proposed BN methodology. The second of these employs a hypothetical model of the proposed California high-speed rail system subjected to an earthquake.
Full List of PEER Reports: click here.
