A Systematic Approach for Severe Accident Uncertainty Analysis

Uncertainty exists in every modeling process especially in those areas with complexity of the calculations like severe accident (SA) code which cover a broad range of physical and chemical phenomena. A systematic framework is proposed here for effective uncertainty assessment of SA computations by efficient use of available data and information. Available methodologies are either input-based or output based. The proposed methodology takes the advantages of both approaches and introduces an integrated one which quantifies the uncertainty of code input parameters (parameter uncertainty), code internal structure (model uncertainty) and code outputs (output uncertainty). The proposed methodology is comprisd of a hybrid qualitative and quantitative approach for identification of uncertainty sources. Using a Bayesian ensemble of sensitivity measures, identified severe accident phenomena are ranked according to their effect on the figure of merit. The other feature of the proposed methodology is the consideration of the SA code structural uncertainties (generally known as model uncertainty) explicitly by treating internal sub-model uncertainties and by propagating such model uncertainties in the code calculations, including uncertainties about input parameters. The code output is further updated through additional Bayesian updating with available experimental data from the integrated test facilities. In this paper, the key elements are discussed for the uncertainty analysis methodology and its application is demonstrated on the LP-FP2 experiment of LOFT test facility.