Assessing the uncertainties in reactor accident consequences by means of RSM: the case of an ATWS in a PWR

A simulation study of a station black-out ATWS has been performed by applying Response Surface Methodology (RSM) on the data obtained by inspecting the ALMOD code.

The case under study has shown that the a priori information which alone could be inadequate, is optimally utilized if coupled with a preliminary sensitivity analysis through RSM techniques. In particular the engineering selection of the model variables and the rank order of the remaining ones had to be modified after an RSM preliminary sensitivity analysis.

Another qualifying feature of the exercise is the use of the randomization of the variables not included in the model in order to coherently exploit the methodology in its full efficiency. This procedure is able to give a figure of merit of the global importance of the neglected variables through the analysis of residuals. 

Results show that the proposed technique is an effective tool for selecting the most important accident variables and that the body of information gained is significant with respect to the number of observations performed.