Survey of Dynamic PSA Methodologies

Event Tree(ET)/Fault Tree(FT) are significant methodology in Probabilistic Safety Assessment(PSA) for Nuclear Power Plants(NPPs). ET/FT methodology has the advantage for users to be able to easily learn and model. It enables better communication between engineers engaged in the same field. However, conventional methodologies are difficult to cope with the dynamic behavior (e.g. operation mode changes or sequence-dependent failure) and integrated situation of mechanical failure and human errors. Meanwhile, new possibilities are coming for the improved PSA by virtue of the dramatic development on digital hardware, software, information technology, and data analysis.. More specifically, the computing environment has been greatly improved with being compared to the past, so we are able to conduct risk analysis with the large amount of data actually available. One method which can take the technological advantages aforementioned should be the dynamic PSA such that conventional ET/FT can have time- and condition-dependent behaviors in accident scenarios. In this paper, we investigated the various enabling techniques for the dynamic PSA. Even though its history and academic achievement was great, it seems less interesting from industrial and regulatory viewpoint. Authors expect this can contribute to better understanding of dynamic PSA in terms of algorithm, practice, and applicability. In paper, the overview for the dynamic PSA was conducted. Most of methodologies share similar concepts. Among them, DDET seems a backbone for most of methodologies since it can be applied to large problems. The common characteristics sharing the concept of DDET are as follows: • Both deterministic and stochastic approaches • Improves the identification of PSA success criteria • Helps to limit detrimental effects of sequence binning (normally adopted in PSA) • Helps to avoid defining non-optimal success criteria that may distort the risk • Framework for comprehensively considering uncertainties and variability • Benefits for risk informed decision-making • As it demands robust PSA results • In addition to risk measures, decision makers want information on uncertainties Nevertheless, each of the methods has advantages and disadvantages, so we, first, need to find the nature of the problems that we are interested in. As a next step of this study, authors are going to investigate the specific problems or issues that require the support of the dynamic PSA, and develop the detailed algorithms