UNCERTAINTY AND SENSITIVITY ANALYSIS OF A POST-EXPERIMENT CALCULATION IN THERMAL-HYDRAULICS

This uncertainty and sensitivity analysis accounts for modeling and parameter uncertainties originating from the experiment, the mathematical model, and the numerical solution algorithm. Quantitative uncertainty statements are derived for code results that are either single-value quantities, like the peak cladding temperature, or continuous-value quantities, like the cladding temperature at a given position within the rod bundle. They do not resort to simplifications, like fitted response surfaces, etc., nor to approximations, like fitted distributions, but are obtained directly from the original code results, employing well-established concepts and tools from probability calculus and statistics. These statements account for the combined influence of modeling and parameter uncertainties as well as for the possible impact of the sampling error that is due to the limited number of code runs performed. Sensitivity measures to single- and continuous-value code results are presented and the benefit from consulting so-called correlation ratios, in addition to the more conventional regression-based measures, is explained and demonstrated.