DYNAMIC EVENT TREES IN ACCIDENT SEQUENCE-ANALYSIS - APPLICATION TO STEAM-GENERATOR TUBE RUPTURE

A dynamic event tree method for analyzing the risk associated with dynamic nuclear power plant accident sequences is presented. The method provides a framework for treating stochastic variations in operating crew states (defined by substates characterizing the accident diagnosis, the planned actions, and the crew quality) as well as stochastic variations in hardware states. Plant process variables are treated deterministically; they are used when determining the likelihood of stochastic branchings. The method is used in an analysis of a steam generator tube rupture (SGTR) accident; it is shown that a number of important operator behavior patterns can be reasonably represented, and that, comparing with conventional event trees, sources of dependencies between failure events can be better defined.