Uncertainty quantification and representativity analysis of LWR-PROTEUS Phase III experiments using SHARKX

• Pinpower uncertainty quantification is applied to fission rate measurements.
• The pinpower uncertainty due to nuclear data and technological parameters is in the range of 0.4–0.8%.
• Assessment of similarities between LWR-PROTEUS Phase III experiments and a BWR.
• The influence of the input uncertainty origin on the representativity is significant.

SHARKX is a set of Perl-based tools build around the lattice code CASMO-5, used to perform sensitivity analysis (SA), uncertainty quantification (UQ) and representativity analysis (RA) developed at the Paul Scherrer Institut. The paper describes the application of recent developments in the SHARKX methodology to zero-power reactor experiments, namely the LWR-PROTEUS Phase III (LWR-III) experiments.The SHARKX Stochastic Sampling (SS) pinpower UQ method is applied to the LWR-III campaign, which featured SVEA-96 Optima2 fresh BWR fuel assemblies, in an attempt to explain the small discrepancy previously observed between the fission rate measurements (E) and their predictions with CASMO-5 (C). The pinpower uncertainty due to nuclear data and technological parameters is in the range of 0.4–0.8%. This additional source of uncertainty, combined with the 0.6% measurement uncertainty, failed to fully explain the remaining few C/E discrepancies in the pinpower map. The source of nuclear data and technological parameters uncertainty is either underestimated, not well modeled or an additional source of uncertainty has not been taken into account.The relevance of the LWR-III experiments to the usage of SVEA-96 Optima2 assemblies in a BWR is rigorously assessed using the newly developed RA capability of SHARKX. The pinpower maps of the three different radial cross-sections of the SVEA-96 Optima2, featuring 96, 92 and 84 pins, are measured in turn in the LWR-III experiments; and various thermal–hydraulic conditions encountered during normal operation of a BWR are considered for the applications. The representativity coefficients ck between each radial map of the LWR-III experiments and the considered applications are always higher than 0.9. It shows that the LWR-III experiments are highly representative of the applications considered. The representativity decreases with increasing void fraction in the application configuration, which is expected as the amount of D2O in the PROTEUS test tank during the LWR-III experiment is designed to simulate moderation of a HFP BWR at 0% void. The influence of the Variance Covariance Matrix origin on the representativity results is responsible for up to 3.5% variation in ck coefficient for experiments and applications with lower ck values.