NEWCOV: A Nuclear Data Covariances Library in Support of Liquid-Metal Fast Reactor Design

Fast reactor systems benefit from less operational experience than traditional water-cooled reactors, making it essential to assess the reliability of computational tools and of input nuclear data for such systems. This assessment is conducted rigorously with uncertainty quantification as well as with Verification&Validation (V&V) studies, both of which rely on handling uncertainty in nuclear data systematically and consistently. This paper focuses on the generation of a library of nuclear data covariances and on its use in uncertainty and sensitivity analyses in the framework of fast reactors. The paper introduces a novel library of covariance matrices called NEWCOV, which is currently in use for the lead-cooled fast reactor of newcleo. NEWCOV contains covariances from ENDF/B-VIII.0, JEFF-4.0 and JENDL-4.0u collapsed on 33 energy groups of neutron multiplicities cross-sections, angular distributions and fission spectra of the main nuclides appearing in both the fresh and depleted fuel of fast reactor designs. Differently from other publicly available pre-computed covariance libraries, NEWCOV contains a complete set of covariances, including angular distributions of lead and delayed neutron multiplicities, making it the most suitable library for lead-cooled fast reactors. Covariances in NEWCOV are provided in the ERRORR and BOXER format, and they can furthermore be used with deterministic codes. The covariance matrices produced in this paper are combined with the sensitivity profiles obtained by Serpent to estimate the nuclear data-induced uncertainties featuring different fast systems. The results are then compared with those computed with existing covariance libraries, showing the improved performance of the NEWCOV covariances.