Nuclear data uncertainty quantification analysis at Studsvik Scandpower

Studsvik Scandpower has recently added nuclear data uncertainty quantification to several of its tools, with the hope to provide user-friendly access to these methods to existing users of HELIOS2, CASMO5, and SNF. The lattice physics codes HELIOS2 and CASMO5 have been extended to perform the propagation of nuclear data cross section uncertainty using statistical sampling. HELIOS2 utilizes existing uncertainty data based on the SCALE/XSUSA 44-group covariance library. In contrast, for CASMO5, Studsvik generated a custom covariance library based primarily on ENDF/B-VIII using the cross section processing code NJOY/ERRORR. This paper presents the methods for constructing and applying these perturbations. The perturbations applied to the cross sections have been verified by comparing HELIOS2 and CASMO5 uncertainties using identical perturbation vectors. Furthermore, the UAM-1 benchmarks were applied for the BWR, PWR, and VVER pincells; in these cases, significant differences were observed due to the different covariance libraries. The uncertainties from HELIOS2 and CASMO5 are passed to SNF, which uses the isotopic uncertainties for the burned fuel to compute uncertainties on various backend quantities, such as decay heat and activity. The SNF UQ features are applied to the Clab decay heat measurements, where the predicted uncertainty of 1.8 % was close to the measurement-to-computed error with bias of 0.72 % and standard deviation of 1.49 %. Finally, the effect of the decay heat uncertainty is shown for a disposal canister loading scenario for more than 8000 BWR fuel assemblies from Olkiluoto 1 and 2. The decay heat uncertainty imposes a similar to 1 % decay heat power safety margin in the loading of the disposal canister.