Fuzzy probability on reliability study of nuclear power plant probabilistic safety assessment: A review

• Based on fuzzy implementation, two types of fault tree analysis have been reviewed.
• The strengths and weaknesses of each type are discussed.
• Use fuzzy based FTA when uncertainties are the main focus of the analysis.
• Use fuzzy hybrid FTA when experts need to evaluate reliabilities of basic events.

Fault tree analysis (FTA) is a graphical model which has been widely used as a deductive tool for nuclear power plant (NPP) probabilistic safety assessment (PSA). The conventional one assumes that basic events of fault trees always have precise failure probabilities or failure rates. However, in real-world applications, this assumption is still arguable. For example, there is a case where an extremely hazardous accident has never happened or occurs infrequently. Therefore, reasonable historical failure data are unavailable or insufficient to be used for statistically estimating the reliability characteristics of their components. To deal with this problem, fuzzy probability approaches have been proposed and implemented. However, those existing approaches still have limitations, such as lack of fuzzy gate representations and incapability to generate probabilities greater than 1.0E-3. Therefore, a review on the current implementations of fuzzy probabilities in the NPP PSA is necessary. This study has categorized two types of fuzzy probability approaches, i.e. fuzzy based FTA and fuzzy hybrid FTA. This study also confirms that the fuzzy based FTA should be used when the uncertainties are the main focus of the FTA. Meanwhile, the fuzzy hybrid FTA should be used when the reliability of basic events of fault trees can only be expressed by qualitative linguistic terms rather than numerical values.