Benchmarks critically benchmarked [nuclear reactor computations]

The nuclear community needs benchmarks as a tool to check computational procedures or to check data. In order to be relevant, the definition of a benchmark has to be accurate and detailed enough to enable an unambiguous calculation. Errors or inaccuracies in cross sections or in calculational procedures lead to a calculation of a benchmark that disagrees with the experimentally measured benchmark. In this paper we analyze the recommended mode of calculation for critical benchmarks that, with proper data, allows a calculation of the measured k_eff within the experimental uncertainty. We found that the mode of calculation recommended in some publications on benchmarks is too coarse, and therefore deviations between calculation and experiment cannot be explained by the inaccuracy of the cross sections alone.