Development of a multi-unit seismic conditional core damage probability model with uncertainty analysis

Various studies about multi-unit probabilistic safety assessment (PSA) have been conducted following the Fukushima accident. In this paper, an intuitive and simple model is developed to supplement the disadvantages of huge models that integrate all units in the same site. Through this framework, pre-assessment can be made for multi-unit events, the results of which can assist risk-informed decision making in a relatively short time. The model deducts multi-unit seismic conditional core damage probability (MUSCCDP) with a Bayesian belief network, and by including seismic inter-unit correlation factors, the model can also represent the extent of simultaneous and identical core damage in multiple units at the same site due to an earthquake. The outcome of the model confirms the impact that seismic intensity and seismic correlation have on MUSCCDP. Moreover, it provides an importance analysis of the components in the units and represents which components are the most vulnerable in multiple units. As the uncertainty analysis of the input data for the model is conducted by changing the distribution of the variables, the results raise the necessity for further uncertainty analyses in multi-unit seismic PSA before accepting the PSA results.