Modeling common-cause failures using stochastic hybrid systems
Résumé
In this paper, we develop a novel Common-Cause Failure (CCF) model for reliability analysis based on component degradation information. A Stochastic Hybrid Systems (SHS) model is developed to describe the components and system degradation process by state dynamics, in presence of CCFs. A component failure is either caused by cumulative degradation (independent failure) or by destructive external events (shared root causes of CCFs, e.g. tornados or earthquakes). System reliability is estimated by the First Order Second Moment (FOSM) method using conditional moments of component degradation levels acquired from the SHS model. A case study is considered regarding the Auxiliary Feedwater Pumps (AFPs) of a Nuclear Power Plant (NPP) suffering internal flooding. Results show that the SHS-based method yields an accurate estimation of reliability compared with Monte Carlo simulation.