QUANTIFICATION OF UNCERTAINTY OF REFLOOD-RELATED MODELS OF RELAP5 MOD3.3 CODE

摘要

Best-Estimate plus Uncertainty Evaluation methods are gaining increased interest in the licensing process. Besides, lessons learnt from the BEMUSE benchmark and others show that improvements of the present methods are necessary, especially developing methods to properly estimate input parameter uncertainties. A methodology has been proposed and developed at University of Pisa for quantification of variation ranges of input parameters through comparison of sensitivity calculations results of a selected system thermal-hydraulic code with experimental data utilizing the mathematical apparatus of Fast-Fourier Transform Based Method (FFTBM). Within the framework of OECD/NEA PREMIUM benchmark this methodology has been applied to the quantification of RELAP5 code models related to simulation of reflood phenomena. The calculations of an experimental test of FEBA facility (KIT, Germany) have been performed with RELAP5 Mod3.3 code and the uncertainties of reflood-influential models (e.g. wall-to-fluid heat transfer) were quantified. The obtained model uncertainties were verified by performing blind calculations of various tests from FEBA and PERICLES test facilities, performing the uncertainty analysis of obtained results and verifying that the uncertainty band envelopes the experimental data. Comparison of the RELAP5 calculated results with the FEBA and PERICLES test data shows that the uncertainty bands envelope the experimental data in the majority of the measurement points. This supports the validity of obtained ranges of reflood-related RELAP5 code models.