Uncertainty Evaluation in Best Estimate Accident Analysis of NPPs

Abstract

Recent trends in carrying out deterministic safety analysis for safety assessment are attracting more attention for the use of best estimate approach. However, conservative approaches are still used in licensing safety analysis. The best estimate approach provides more realistic information with respect to conservative approach for predictions of physical behaviour and provides information about the existing safety margins and between the results of calculation and regulatory acceptance criteria, whereas conservative approach also does not give any indication about actual plant behaviour, including time scale, for preparation of emergency operating procedures. Best estimate methodology results are affected by various sources of uncertainty like code or model uncertainty, representation uncertainty, scaling uncertainty and plant uncertainty. Therefore, uncertainty in the results due to unavoidable approximation in the modelling should be quantified. Various uncertainty analysis methodologies have been emerged after the development of code scaling, applicability and uncertainty (CSAU) by US NRC in 1989, like CIAU by Italy and GRS by Germany. This chapter deals with the sources of uncertainty and its quantifications by various methodologies in terms of confidence and probability. This chapter also deals with the application of a sampling-based uncertainty evaluation in the best estimate analysis of station blackout and small break LOCA in integral test facilities carried out under the framework of IAEA CRP. Uncertainty evaluation for TMI-II accident has been carried out using this methodology.