Treating support system dependencies among safety functions in dynamic event tree quantification

Abstract

Dependencies are present among the safety systems in a nuclear power plant; for example, shared support systems such as power supply, service water, component cooling water, etc. The unavailability of the support systems contributes to the failure of the frontline systems. The solution and quantification of the event trees and fault trees that make up PSA models account for these dependencies. These dependencies must also be treated in dynamic risk assessments using the dynamic event tree framework, in which plant simulations are coupled to the stochastic models. The approaches from classical PSA for dealing with this issue can be adapted to the dynamic event tree work. While the principles are clear, the implementation is complex because of the need to couple the Boolean (logical) algorithms and probabilistic computations with the dynamic progression of the analysis. As an alternative to a fully integrated (online) treatment of the issue within the DET algorithm, this paper presents and demonstrates a post-simulation treatment based on conditional probability calculations. The obtained results are analyzed and compared to a reference solution based on a classical PSA approach.