Severe accident sensitivity and uncertainty estimation using MELCOR and RAVEN

Abstract

Today, using a best-estimate approach is a key factor in the simulation and prediction of thermal-hydraulic and other multi-physics phenomena occurring during nuclear severe accidents. The best-estimate approach requires the quantification of both epistemic and stochastic uncertainties of safety codes to be effective. This safety assessment approach, named Best Estimate Plus Uncertainty (BEPU), is being pursued as an alternative to traditional deterministic analyses that are intrinsically conservative with not clearly defined safety margins. While in a conservative approach, the results are expressed in terms of a set of calculated conservative values of parameters, in a best-estimate methodology, the results are expressed in terms of uncertainty ranges for the calculated parameters. The International Technical Nuclear Community has made great efforts to develop methods and tools for uncertainty and sensitivity analyses of severe accident codes. In this framework, Sapienza University of Rome has developed a new Python script to use RAVEN as a tool for the application of the BEPU methods within the safety analyses performed with MELCOR. The aim of this paper is to show the capabilities of the coupling between RAVEN and MELCOR by performing a statistical analysis to estimate the range of evolution of a severe accident scenario.