WPEC Subgroup 44 computational Inter-comparison exercise on correlations in nuclear data libraries

There is a long-standing controversy on nuclear data uncertainty assessment for general purpose nuclear data libraries. On the one hand, nuclear data users would like the libraries to predict uncertainties for selected integral quantities consistent with the integral experimental uncertainties, while on the other hand, doing so could make evaluations dependent on selected integral datasets breaking the general applicability of the library to any existing or future applications. This article studies the hypothesis that certain correlations between nuclear data, which come from the immutable nature of the reactor physics in the integral experiment used as benchmarks, and can be estimated almost independently of the choice of selected integral experiments, nuclear data library, or evaluation methodology. This article reports the findings of an international computational inter-comparison study carried out under the auspices of the Working Party on International Nuclear Data Evaluation Co-operation. The participants represented 5 different organizations, on three different continents and used different initial nuclear data libraries and different calculation methodologies. This study focused on estimating the correlation coefficients between fission, capture and nu-bar for Pu-239 which would arise in the final evaluated nuclear data library if a plutonium metal fast-neutron-spectrum critical experiment with typical integral-measurement uncertainty of 100 pcm was used in the validation and feedback to compile the nuclear data library. The additional knowledge of the correlation coefficients can effectively reduce the propagated uncertainty on criticality experiments from the extended library. This exercise helped to improve understanding the different approaches used, to identify weaknesses and provide indications where further work is required to develop a scientifically rigorous method. This article does not aim at recommending these methods as standards. It aims at honoring the effort of Massimo Salvatores, who was instrumental in pushing studies that should lead to development of uncertainty estimation and by participating in them.