The Findings from the OECD/NEA/CSNI UMS (Uncertainty Method Study)

Within licensing procedures there is the incentive to replace the conservative requirements for code application by a “best estimate” concept supplemented by an uncertainty analysis to account for predictive uncertainties of code results. Methods have been developed to quantify these uncertainties. The Uncertainty Methods Study (UMS) Group, following a mandate from CSNI (Committee on the Safety of Nuclear Installations) of OECD/NEA (Organization for Economic Cooperation and Development / Nuclear Energy Agency), has compared five methods for calculating the uncertainty in the predictions of advanced “best estimate” thermal-hydraulic codes. Most of the methods identify and combine input uncertainties. The major differences between the predictions of the methods came from the choice of uncertain parameters and the quantification of the input uncertainties, i.e. the wideness of the uncertainty ranges. Therefore, suitable experimental and analytical information has to be selected to specify these uncertainty ranges or distributions. After the closure of the Uncertainty Method Study (UMS) and after the report was issued comparison calculations of experiment LSTF-SB-CL-18 were performed by University of Pisa using different versions of the RELAP 5 code. It turned out that the version used by two of the participants calculated a 170 K higher peak clad temperature compared with other versions using the same input deck. This may contribute to the differences of the upper limit of the uncertainty ranges. A ‘bifurcation’ analysis was also performed by the same research group also providing another way of interpreting the high temperature peak calculated by two of the participants.