Hierarchical Bayesian Model Concerned Uncertainty in Number of Failure Events for Component Failure Rates

Components failure rates are important inputs of Probabilistic Risk Assessment (PRA). They are generally estimated from data observed in plant’s experiments. However, number of failure events derived from plant’s experiments might not represent the actual value. This means that true failure rates might have more uncertainty than one estimated from the actual experiments. Therefore it has been necessary to develop statistical Bayesian model which can reflect an uncertainty in data for components failure rates. In this study, the uncertainty in number of failure events is considered as a probability process and the hierarchical Bayesian model is applied to reflect the uncertainty for failure rates. As a result, more appropriate result of PRA is obtained with a state of knowledge about number of failure events for failure rates.