Quantification and uncertainties of common cause failure rates and probabilities

摘要

Simultaneous failures of multiple components due to common causes at random times are modelled for standby safety systems by multiple-failure rates rather than probabilities per demand. Extensions and improvements are made to estimating such rates from single- or multiple-plant event data with uncertainties, and to determine basic common cause event probabilities for explicit fault tree models in terms of the rates, test intervals, test schedules and repair policies. Such models are needed especially in standby safety system reliability analysis, testing optimisation and in risk-informed applications. The probabilities are derived such that the correct time-average system unavailability can be obtained with a single fault tree quantification. Improvements are made in the rale estimation with generalised impact vectors, and the probabilities now include higher-order terms and are independent of the system logic (success criterion).