Probabilistic risk model of digital reactor protection system

This report presents a probabilistic risk assessment (PRA) model of a nuclear power plant focusing on digital I&C in the reactor protection system (RPS). The model is prepared for an international benchmark study in WGRISK project DIGMAP. The model contains one event tree representing loss of main feed-water accident in a fictive boiling water reactor plant. The model is very simplified. Only the RPS is modelled in detailed, because that is the focus of the benchmark study. The selected modelling approach is close to the previous model of the DIGREL project employing small fault trees as building blocks. I&C component failures have been divided into detected failures and undetected failures. Significant portion of the contribution of the RPS related risk comes from application software failures, along with undetected hardware failures. On the other hand, detected hardware failures in the RPS have insignificant contribution to the core damage risk, likely because spurious actuations have not been analysed. The importance of automatic testing and periodic testing as fault tolerant techniques to reduce the risk of undetected hardware failures was recognized in the sensitivity studies. Selection of common cause failure groups and parameters, and application software basic events are expected to be major issues in the benchmark study.