Common cause analysis : a review and extension of existing methods

Abstract

The quantitative common cause analysis code, MOBB, is extended to include uncertainties arising from modelling uncertainties and data uncertainties. Two methods, Monte Carlo simulation and the Method-of-Moments are used to propagate uncertainties through the analysis. The two different capabilities of the code are then compared. When component failure rates are assumed lognormallv distributed, bounded lognormal (Sb) distributions are used to evaluate higher moment terms, as required by the Method-of-Moments, in order to minimize the effect of the tail of the lognormal. A code using the discrete probability distribution (DPD) method is developed for analyzing system unavailability due to common initiating events (internal and external). Sample problems demonstrating each approach are also presented.