Seismic risk assessment of high-voltage transformers using Bayesian belief networks

Abstract

Large-magnitude earthquakes can damage high-voltage transformers, trigger power flow disruption and impact the economy and society. However, methods that enable large transformer vulnerability assessment in a practical and rigorous way are scarce. This paper proposes a probabilistic framework using Bayesian belief networks (BBNs) to predict the damage of high-voltage transformers subjected to seismic events. This framework incorporates major causes of transformer vulnerability at once, such as liquefaction, rocking response of the transformer, and interactions with interconnected equipment, which are otherwise commonly studied in isolation. To demonstrate the application of the framework, the paper elaborates on each step of the BBN framework, which is then validated with historical empirical data. Furthermore, the value of the proposed method is illustrated with high-voltage transformers in substations of the electric value BC Hydro in British Columbia, Canada. The paper also offers a sensitivity analysis that evaluates the effects of input variables on transformer damage. The proposed framework is simple to perform in practice, and the results are expected to support decisions on mitigation measures, seismic risk management, and to provide a step towards modelling the vulnerability of entire electrical substations.