Bayesian Updating of Seismic Probability

Abstract

For those nuclear power plants for which a seismic probabilistic risk assessment has been conducted, it is possible to infer a prediction for the recurrence rate of the safe shutdown earthquake, and then to compare it with the historic seismicity at the site. Using the Bayesian algorithm, it is then possible to update the prediction in such a way as to quantify the degree of conservatism. By using a sample of eight plants, and other assumptions that are reasonable but by no means unique, the conservatism is estimated to be a factor of the order of 2 or 3. The uncertainty is also reduced, though there are caveats on this point. This suggests that earthquakes are somewhat overrated as sources of risk for nuclear power plants.