New Benchmarks Versus Old Benchmarks: A Comparison Of The Icsbep And Csewg Benchmarks

This paper presents results of validation calculations for a series of fast critical assemblies based on the traditional CSEWG benchmark models and the more current ICSBEP benchmark models. Benchmark results, based on continuous-energy Monte Carlo calculations using both the CSEWG and ICSBEP benchmark representations with various evaluated nuclear data libraries, are presented for 18 critical assemblies. Results are put on a comparable basis by applying corrections associated with each of the benchmark specifications allowing consistent comparison of the validation results from each model. For some of the assemblies, significant differences are observed in the C/E values. It is also observed that the bias between these models is dependent on the evaluated data libraries. More importantly, the comparisons extend beyond the calculated results and attempt to identify the extent of the differences in the benchmark models themselves, identifying differences in masses, compositions, geometry, and dimensions. Overwhelmingly, these small model differences occurred with the introduction of necessary modeling simplifications into the older models. The ICSBEP models, though limited to testing of criticality (k(eff)), often contain more details of the experiment (and therefore introduce fewer simplifications into the models) and generally include more detailed assessment of the benchmark uncertainties, making these benchmarks of higher value for data validation.