Validation of the sodium void reactivity effect prediction using JEFF-3.1 nuclear data

Near 60 Na void experiments performed in the zero power reactors MASURCA (CEA-Cadarache) and ZPPR (Argonne West – Idaho) have been analyzed using JEFF-3.1 nuclear data and the ERANOS-2.1 (deterministic) and TRIPOLI-4 (Monte-Carlo) codes. Some comparative calculations have been performed also using either JEFF-3.1, ENDF/B-VII.0 or JENDL-3.3 nuclear data for 23Na, as these three 23Na evaluations show marked differences. The Na void experiments have been selected to cover spectral conditions ranging from the relatively hard flux in the outer zone of a small fast reactor to the relatively soft flux in the inner zone of a large fast reactor. For in-fuel Na void patterns, there is a good agreement between ERANOS and TRIPOLI computations, while the deterministic calculations significantly underestimate the leakage component for Na void patterns in fertile regions. The agreement between ERANOS-2.1 + JEFF-3.1 predictions and experimental values is excellent for in-fuel Na void patterns in MASURCA experiments, but a significant underestimation of the leakage component occurs for in-fuel Na void patterns in ZPPR. For fertile Na void patterns, there is a clear underestimation of the leakage component, quantitatively different for MASURCA and ZPPR experiments. Variations in 23Na cross-section data also result in significant differences: ENDF/B-VII.0 and JENDL-3.3 nuclear data for 23Na increase noticeably the predicted Na void worth values with respect to JEFF-3.1 data. The three 23Na evaluations differ at high energy (>500 keV, and especially >2 MeV), and this stresses the need for accurate additional measurements in this energy range.