Proceedings of the Eighth International Conference on Probabilistic Safety Assessment & Management (PSAM)
Download citation file:
- Ris (Zotero)
- Reference Manager
Uncertainty analyses of probabilistic accident consequence codes provide large insight in the range of output of the models analysed. Particularly, when such codes are used to meet the requirements of safety and environmental laws, the robustness of the uncertainty analysis is of vital importance. Nuclear power plants are, for instance, subjected to environmental laws whereby iso-risk-contours are calculated to decide on emergency zoning distances around the plant. There is always one important decision to be made, namely with which output data will the iso-risk-contours be determined: median values or 95th percentile values or any other percentile values?
The COSYMA package developed by the European Commission is an example of a probabilistic accident consequence code applied to derive the risk of nuclear power plants to third parties. A full uncertainty analysis has been done on the COSYMA code. The analysis evaluated the uncertainty on 191 endpoints. The study evaluated the uncertainty for three different situations that are of relevance to regulatory processes in different European countries, considering consequences if countermeasures were imposed at levels based on international guidance, consequences if no countermeasures were considered but taking account of the shielding properties of buildings for normal living patterns, and individual doses and risks for potential outdoor exposure. The uncertainties were evaluated for three source terms: a design basis accident, a severe degraded core accident in which about 50% of the volatile material in the core would be released and a sequence giving a release of about 1% of the volatile material in the core. In this way the study encompassed the wide range of source terms and other situations that have been considered in safety assessments undertaken in the past, and that are likely to be considered in the future.
A set of default values is accomplished for the most important input variables in a PSA-3-analysis by determining a so-called reference curve in the uncertainty analysis. Moreover, the associated reference coefficients link the 95th percentile of the uncertainty distribution of an endpoint to the set of default values. This leads to a base how to determine where the 95th enveloppe in the uncertainty analysis exists.
How to handle uncertainties in a level-3 PSA by applying the results of the uncertainty study for other situations is a useful concept. If new situations are close to the source terms and location used in the current uncertainty analysis, the use of the study results is appropriate. If the location of the nuclear power plant is different, but the source terms are comparable, then the results of the study are quite useable (except maybe when different threshold values will be applied in the analysis). If the source terms are different from the ones used in the current analysis, one should be careful in using the study results. Examples of consequences for emergency zoning around nuclear power plants will be shown. Individual risks will be presented as iso-risk-contours. Group risk (or societal risk) will also be discussed. The uncertainties associated with early deaths are in the order of several 10's, the uncertainties associated with late deaths are in the order of 100's.