Input uncertainties in uncertainty analyses of system codes: Quantification of physical model uncertainties on the basis of CET (combined effect tests)

Abstract

Quantification of the model uncertainties can be performed in different ways. It is generally accepted that for those phenomena for which separate effect tests exist, the quantification is performed by direct comparison of code simulations with experimental data for selected phenomena. For phenomena for which no separate effect tests exist, but there are available combined effect tests (CET), they can be applied for model uncertainties quantification. But inverted uncertainty methodologies applied for uncertainties quantification on the basis of such experiments are complex. To push forward such methods of physical model uncertainties quantification in the thermal-hydraulic codes OECD/NEA benchmark PREMIUM had been launched. GRS participation in NURESAFE project was also, attended to this problem. GRS participated in the benchmark using a simple “trial-and-error” method and in parallel an attempt has been performed to develop and apply a new method base on adjusting of samples of input uncertainties combinations. On the basis of FEBA reflood, experiment quantification of uncertainty of the models important for reflooding simulation has been performed. Afterwards, the quantified model uncertainties were verified and validated. The verification step was done performing uncertainty analyses of the test from the same FEBA experiment. For validation reflooding experiment PERICLES has been selected. The validation was done performing “blind” uncertainty analyses of several tests from PERICLES experiment. It can be stated that the selection of the experimental data as well as the applied quantification method was not fully satisfactory. The requirements concerning the qualification of the input data, experimental data, and accuracy of statistical methods for uncertainties quantification appear to be more rigorous than typical requirements for code validation procedure.