A Bayesian Network-based system framework for seismic resilience assessment of nuclear power plants

This paper proposes a Bayesian network (BN)-based framework for assessing the seismic resilience and loss of nuclear power plant (NPP) systems. The study emphasizes the importance of considering both internal components and external infrastructures, such as power and water supply networks, in a system-of-systems approach. A multi-state damage model is adopted to evaluate functional states of facilities, non-structural components, and equipment under seismic loads. The paper introduces a novel overall resilience metric that integrates resistance, absorption, and recovery capabilities, providing a comprehensive evaluation of NPP performance. Through extensive parametric analysis, the effects of recovery periods, earthquake magnitudes, source-to-site distances, soil conditions, and loss of off-site power on resilience are investigated. Additionally, the concept of “iso-resilience contours” is first proposed to support rapid resilience prediction and site selection for NPPs. The results demonstrate the practicality of the BN framework in enhancing seismic resilience assessment and decision-making for nuclear facilities.