Accuracy quantification metrics for CFD simulation of in-vessel flows

Abstract

Within more general frameworks dealing with uncertainty evaluation and code assessment, well assessed methodologies exist for the accuracy evaluation of system thermal hydraulic analysis results. As long as multi-dimensional transient data is involved, such as space and time distributions obtained from CFD calculations and experiments, the task of quantifying the accuracy and defining acceptance criteria becomes harder, and there seems to be a lack of systematic approaches available to that purpose.An example of such multi-dimensional data is represented by the time and space distribution of coolant properties (e.g. temperature, boron concentration) at the reactor core inlet, which is relevant to reactor safety as local perturbations potentially induce power excursions. An attempt is made here to tackle the accuracy quantification issue for that class of data, in a somewhat empirical way, i.e. by proposing a set of parameters that can be used, after proper qualitative analysis of measured data and code simulation results, to characterize the target variable time and space distributions, and to quantify the deviation of the simulation from the experimental data. The advantage of using such kind of metrics based on many parameters relies in the ability to cover different relevant features of the target variables and thus provide a more complete assessment and allow statements about the overall agreement.