Estimating Passive System Reliability and Integration into Probabilistic Safety Assessment

Passive safety systems are being increasingly deployed into advanced designs of nuclear power plants (NPPs) with an objective of enhancing the safety. Passive systems are considered to be more reliable than the active systems as mechanical failures of active component and failure due to human errors are not contributing towards the system failure probability. The introduction of passive systems into NPPs on the one hand improves the reliability, and it poses challenges to the estimation of its reliability and integration into probabilistic safety assessment (PSA). The active system reliability can be estimated using the classic fault tree analysis technique. For estimating the passive system reliability, a different approach is required due to the presence of phenomenological failures apart from the mechanical component failures. In this paper, the approach is presented to estimate the passive system reliability of a typical isolation condenser system using artificial neural network (ANN)-based response surface method. The integration of passive system reliability into PSA is also demonstrated using the accident sequence analysis for a typical operational transient, which requires NPP to shut down from the full power and successfully remove the residual heat from the reactor core through isolation condenser system. The thermal-hydraulic behaviour of the isolation condenser system is analysed using thermal-hydraulic computer code RELAP 5/MOD3.2 for different system configurations depending upon the initial conditions of the process parameters. The variability of the process parameters is represented by discrete probability distribution within the given operating range. The reliability of the reactor protection system (RPS) is estimated using the fault tree analysis method using Risk Spectrum computer code. The core damage frequency induced through the operational transient is estimated using the classical event tree analysis approach. The open issues are also identified for integration of passive system reliability into PSA for future work.