Simulation-based level 2 multi-unit PRA using RAVEN and a simplified thermal-hydraulic code

The paper introduces a simulation-based Level 2 probabilistic risk assessment (PRA) of a multi-unit nuclear power plant, which consists of four boiling water reactors. We propose the methodology by quantifying risk metrics for a station-blackout accident scenario, initialized by a loss-of-offsite-power event. Contrary to classical PRA that generally applies static models such as event-tree/fault-tree, the analysis is seamlessly integrated with mechanistic simulation and PRA models, including: (1) a simplified thermal-hydraulic code for simulating behaviors of four reactor cores; (2) a Markovian model for the degradation and recovery processes for decay-heat-removal systems, to investigate the interaction between mechanistic simulation and reliability analysis; and (3) classical containment event trees for evaluating containment performances and hydrogen-explosion risk under severe accident conditions. All dynamic and static models, including plant dependencies, are unified within the RAVEN computational framework, applying RAVEN components, External Model, Ensemble Model, and PRA Plugins. The study demonstrates an integrated assessment of plant risks by considering accident progression and inter-unit system interactions, both time dependent. Statistical data analysis is used to quantifying risk metrics, including core damage frequencies, large early release frequencies and plant damage status. The methodology pertains to modern risk-analysis methodologies such as risk-informed safety margin characterization (RISMC) and dynamic PRA.