USE OF AN INFLUENCE DIAGRAM AND FUZZY PROBABILITY FOR EVALUATING ACCIDENT MANAGEMENT IN A BOILING WATER-REACTOR

A new approach is presented for evaluating the uncertainties inherent in severe accident management strategies. At first, this analysis cons iders accident management as a decision problem (i. e., applying a str ategy compared with do nothing) and uses an influence diagram. To eval uate imprecise node probabilities in the influence diagram, the analys is introduces the concept of a fuzzy probability. When fuzzy logic is applied, fuzzy probabilities are easily propagated to obtain results. In addition, the results obtained provide not only information similar to the classical approach, which uses point-estimate values, but also additional information regarding the impact of using imprecise input data. As an illustrative example, the proposed methodology is applied to the evaluation of the drywell flooding strategy for a long-term sta tion blackout sequence at the Peach Bottom nuclear power plant. The re sults show that the drywell flooding strategy is beneficial for preven ting reactor vessel breach. It is also effective for reducing the prob ability of containment failure for both liner melt-through and late ov erpressurization Even though uncertainty exists in the results, floodi ng is preferred to do nothing when evaluated in terms of two risk meas ures: early and late fatalities.