Answers are grounded in this paper and cite their sources.
BackgroundMultigroup nuclear data and their covariance matrices are essential inputs for sensitivity analysis, uncertainty quantification, and nuclear data adjustment in reactor physics calculations. The average number of neutrons released per fission is a key nuclear data because its uncertainty can propagate directly into criticality and other integral responses. In evaluated nuclear data libraries, the covariance data for the average number of neutrons released per fission are usually provided in continuous energy, whereas reactor physics and uncertainty analysis applications generally require multigroup covariance data under specified energy-group structures and weighting functions. Existing processing schemes may introduce additional approximations when the weighting function on a union energy grid is derived from a coarser user-defined group structure.PurposeThis study aims to construct a reproducible dataset of multigroup relative covariance matrix for the average number of neutrons released per fission and to evaluate the influence of weighting-function treatment on covariance processing.MethodsThe dataset was generated based on the ENDF/B-VIII.1 evaluated nuclear data library. Firstly, three uranium isotopes, 233U, 235U, and 238U, were selected, and three typical neutron energy-group structures, the epri-cpm 69-group, lanl 187-group, and xmas lwpc 172-group structures, were adopted. For each nuclide and energy group structure, the covariance data of total, delayed, and prompt average number of neutrons released per fission (corresponding to MT452, MT455, and MT456) were processed. Then, the evaluated covariance energy nodes in MF31 and the user-defined group boundaries were merged to establish a union energy grid. On this grid, the weighting function was obtained by integration directly over each union energy group, rather than by energy-width proportional allocation from the user-defined coarse group. Subsequently, multigroup average number of neutrons released per fission were collapsed according to fission neutron production-rate conservation, and the corresponding relative covariance matrices were obtained by covariance propagation and group collapsing. The generated matrices were compared with those processed by NJOY2016. Finally, the epri-cpm 69-group MT452-MT452 covariance matrices of 235U and 238U were used to generate perturbed nuclear data samples, and the VERA 1B benchmark was calculated with OpenMC to examine the propagation effect on effective multiplication factor uncertainty.ResultsComparison results show that the maximum relative deviation between the developed processing code GNAP and NJOY2016 is less than 0.1%, when the user-defined group structure is set to be identical to the union energy grid, confirming the numerical consistency of the covariance processing implementation. Under the practical epri-cpm 69-group structure, the different treatment of the weighting function on the union energy grid produces observable differences in the final relative covariance matrices. For the three typical group structures considered in this study, the largest average relative difference occurs in the MT452-MT455 covariance matrix of 233U, with an average value of 2.42%, while several local maximum relative deviations exceed 162%, which is attributed to the small absolute values of the reference matrix elements. In the VERA 1B benchmark, both the GNAP- and NJOY-based covariance samples successfully completed 100 valid criticality calculations. The sample standard deviations of the keff responses are 495 pcm and 486 pcm, respectively, and the noise-corrected standard deviations are 493 pcm and 485 pcm, corresponding to an approximately 1.8% difference in propagated standard deviation.ConclusionsThe results of this study indicate that direct integration of the weighting function on the union energy grid provides a physically clearer and more refined treatment for multigroup covariance processing for the average number of neutrons released per fission. The constructed dataset is reproducible, traceable, and suitable for supporting uncertainty analysis of reactor physics calculations. The comparison with NJOY2016 shows that the proposed method is consistent with the conventional processing route when the weighting-function approximation is removed, while the practical group-structure results demonstrate that the unionized-grid weighting treatment can affect both covariance matrix elements and propagated response uncertainties. The dataset can therefore serve as a reference basis for covariance processing method verification, nuclear data uncertainty propagation, and future extension to other evaluated nuclear data libraries.
No takes yet. Share an insight, caveat, or question.
DONG et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: