Improvements to a level 3 PSA event tree model and case study

This report presents implemented and potential improvements to a level 3 PSA model and its analysis previously developed in the PRADA project. The model is an event tree model, where wind direction, wind speed, precipitation, success of evacuation and success of sheltering (if evacuation is unsuccessful) are the nodes. The case modelled was an alternative take on the Fukushima Daiichi nuclear accident: what adiological consequences would the accident have had if the population in nearby big cities had been in place and not dislocated due to the tsunami. The radiological consequences were found to be small even under rather conservative assumptions. The most important change to the event tree model introduced in this report is that wind speed is now drawn from a Weibull distribution in a Monte Carlo simulation. Also evacuation modelling was improved slightly. In the uncertainty analysis, the most important change concerns the handling of source term uncertainties. Population dose is composed from doses caused by each radionuclide, and thus the amounts of radionuclides in the release can be made random variables and subjected to Monte Carlo simulation. An uncertainty distribution was attached also to wind speed parameters and evacuation success probability in the uncertainty analysis. The radiological consequences to the general population were minor also in this study. This gives support to the hypothesis that the very small (according to UNSCEAR) radiological consequences to the general population in the Fukushima Daiichi nuclear accident were not a matter of good luck but rather something to be expected concerning the case.