A benchmark of dynamic reliability methods for probabilistic safety assessment

Various dynamic reliability methods have been developed during the last several decades, such as Monte Carlo (MC) simulation, Dynamic Event Tree (DET) and Monte Carlo Dynamic Event Tree (MCDET) methods, to name a few. This paper benchmarks these particular methods by applying them to a classic level control dynamic system and to a realistic Emergency and Standby Power System (ESPS). The analysis is done with respect to different component aging dynamics and transition rate state dependences. Efficiency and computational cost are evaluated. The results show that: (1) the dynamic methods indeed are capable of capturing the effects of the dynamics in the process; (2) DET can model the possible accidental sequences, but at a large computational cost; (3) accurate modelling of sequences with low probabilities of occurrence can be achieved by the MCDET method.