A Bayesian approach to unanticipated events frequency estimation in the decision making context of a nuclear research reactor facility

• The Bayes’ theorem is employed to support the decision making process in a research reactor.
• The intention is to calculate parameters related to unanticipated occurrence of events.
• Frequency, posterior distribution and confidence limits are calculated.
• The approach is demonstrated using two real-world numerical examples.
• The approach can be used even if no failures have been observed.

Research reactors are considered as multi-tasking environments having the multiple roles of commercial, research and training facilities. Yet, reactor managers have to make decisions, frequently with high economic impact, based on little available knowledge. A systematic approach employing the Bayes’ theorem is proposed to support the decision making process in a research reactor environment. This approach is characterized by low level complexity, appropriate for research reactor facilities. The methodology is demonstrated through the study of two characteristic events that lead to unanticipated system shutdown, namely the de-energization of the control rod magnet and the flapper valve opening. The results obtained demonstrate the suitability of the Bayesian approach in the decision making context when unanticipated events are considered.