Analysis of benchmark critical experiments with ENDF/B-VI data sets

Several clean critical experiments were analyzed with ENDF/B-VI data to assess the adequacy of the data for U235, U238 and oxygen. These experiments were (1) a set of homogeneous U235-H2O assemblies spanning a wide range of hydrogen/uranium ratio, and (2) TRX-1, a simple, H2O-moderated Bettis lattice of slightly-enriched uranium metal rods. The analyses used the Monte Carlo program RCP01, with explicit three-dimensional geometry and detailed representation of cross sections. For the homogeneous criticals, calculated kcrit values for large, thermal assemblies show good agreement with experiment. This supports the evaluated thermal criticality parameters for U235. However, for assemblies with smaller H/U ratios, kcrit values increase significantly with increasing leakage and flux-spectrum hardness. These trends suggest that leakage is underpredicted and that the resonance eta of the ENDF/B-VI U235 is too large. For TRX-1, reasonably good agreement is found with measured lattice parameters (reaction-rate ratios). Of primary interest is rho28, the ratio of above-thermal to thermal U238 capture. Calculated rho28 is 2.3 (± 1.7) % above measurement, suggesting that U238 resonance capture remains slightly overpredicted with ENDF/B-VI. However, agreement is better than observed with earlier versions of ENDF/B