ScholarWorks@중앙대학교

초록

This study compares an example of dynamic probabilistic safety assessment (D-PSA) for a small break loss of coolant accident with two refined conventional probabilistic safety assessment (PSA) models. In this example case, high-pressure safety injection (HPSI) and atmospheric dump valve (ADV) actions are divided into multiple time intervals, reflecting time-dependent behavior but significantly increasing the complexity. In this study, we developed two models. Model 1 includes these time-dependent interactions in a conventional PSA by applying fault tree logic to represent delayed HPSI and ADV responses without extensive branching. Model 2 refines the approach further by defining a continuous boundary function between HPSI and ADV times, then integrating each variable's probability density along that boundary to determine the failure probability. The derived probability calculates the core damage frequency, yielding approximate results comparable to those from the D-PSA. These findings demonstrate that, by modeling success criteria to reflect time-dependent behaviors, a conventional PSA can approximate dynamic outcomes without incurring excessive computational demands.

키워드