Leak frequency analysis for hydrogen-based technology using bayesian and frequentist methods

Dealing with hazardous environments such as hydrogen poses considerable risks to property, people, and the environment. Leak frequency analysis is a method of understanding the characteristics of risks at hydrogen refueling stations (HRSs). This paper proposes leak rate estimation using time-based evaluation methods that utilize historical HRS accident information. In addition, leak frequency estimates from another two methods (non-parametric and leak-hole-size) were examined. In the non-parametric approach, the leak frequency is estimated based on a Bayesian update. The results from these three approaches are summarized to understand the trend of leak rate data. The leak rate data from the time-based method displays a similar trend to the leak size based method. However, the non-parametric method tends to be conservative due to high failure observations (new evidences) during the Bayesian update. Finally, the unrevealed leak time was calculated as a function of the leak frequency. The quantitative insights of this study can be used to set performance standards for the availability and reliability in the operation and maintenance of HRSs.