Failure possibilities for nuclear safety assessment by fault tree analysis

Abstract

Fault tree analysis (FTA) is a deductive tool to assess the safety of nuclear power plants. This analysis can only be implemented if all basic events in the tree have their corresponding failure rates. Therefore, safety analysts have to provide those failure rates well in advance. However, it is often difficult to obtain those failure rates due to insufficient data, changing environment or new components. This paper proposes a failure possibility based FTA approach to overcome the limitation of the conventional FTA for nuclear safety assessment. It utilises the concept of failure possibilities to evaluate basic event failure without historical data, fuzzy numbers to map component failure possibilities into mathematical form and defuzzification algorithms to convert fuzzy numbers into component failure rates. A case study on evaluating a typical high pressure core spray system of a boiling water reactor illustrates the applicability of the proposed approach.