Uncertainty bands in CCDF risk curves; their importance in decision making processes

In the present study, examples of uncertainty bands of early fatalities CCDF risk curves are shown for nuclear power plants located in two densely populated sites, A and B, which differ from each other mainly by the population distribution patterns and the meteorological conditions. The REMAND computer code was used for assessing the consequences of various severe accidents in a nuclear power plant. The accidents analyzed were the WASH-1400 types PWR-1, PWR-2 up to PWR-5 and the annual probabilities of occurrence therein were also adopted, for each accident. The uncertainty bands were obtained by using two sets of parameters in the consequence modelling part of the PRA, namely a conservative and best estimate set. Based on sensitivity studies it is believed that the conservative and the best estimate curves represent the 90%-95% and the 50% curves, respectively. In other words, it is believed that there is only a few percent chance of getting a complementary cumulative distribution function curve higher than the conservative one, and that the best estimate curve lies within a range of no more than +/-10% from the real 50% curve