A method for propagating uncertainty in probabilistic risk assessment

A method is developed to propagate uncertainties in the basic event unavailabilities through a logic model to obtain the transient overpower event unavailability. The method consists of combining probability distributions in the discrete form without performing any sampling. The results are shown to be sufficiently accurate and contain no sampling errors; the computation time is considerably less compared to Monte Carlo simulation and histogram propagation. Uncertainty propagation methods are found to be sensitive to the spread of the basic event unavailability distributions; the proposed method produces results less conservative compared to those from propagation of moments or Monte Carlo simulation