Application of an iterative methodology for cross-section and variance/covariance data adjustment to the analysis of fast spectrum systems accounting for non-linearity

作者：Pelloni Sandro^*

来源：Annals of Nuclear Energy, 2014, 72: 373-390.

DOI：10.1016/j.anucene.2014.06.002

• Fast spectrum systems
• Deterministic methods
• Sensitivity coefficients
• Cross-section adjustment
• Non-linearity
• Nuclear data uncertainty reduction

The data assimilation benchmark launched by the "Subgroup 33" on "Methods and issues for the combined use of integral experiments and covariance data" of the Working Party on Evaluation Cooperation (WPEC) of the OECD Nuclear Energy Agency Nuclear Science Committee is recalculated by means of a multistep adjustment procedure using the deterministic code system ERANOS in conjunction with a dedicated Generalized Linear Least-Squares approach based on the Bayesian parameter estimation method. Nuclear data in terms of multi-group cross-sections as well as their variances and covariances, are adjusted for 11 nuclides, namely B-10, O-16, Na-23, Fe-56, Cr-52, Ni-58, U-235, U-238, Pu-239, Pu-240 and Pu-241 and 6 nuclear reactions which are elastic and inelastic scattering, lumped (n,2n) and (n,3n), capture, fission and (nu) over bar. The adjustment is carried out by making use of experimental data for 19 integral parameters obtained in 7 different fast spectrum systems. In the determination of a posteriori values for these integral parameters including effective multiplication factors, spectral indices and void effects, along with their nuclear data uncertainty, the required adjusted data for these nuclides and reactions are generated in conjunction with pre-computed sensitivity coefficients of the analytical integral parameters to the nuclear data to adjust. The suggested multistep scheme aims at accounting for non-linear effects. Correspondingly, the sensitivity coefficients are recalculated within an iterative procedure on the basis of the a posteriori analytical values and adjusted cross-sections. The adjustment is thus repeated several times until convergence is reached for the analytical values and their uncertainties.
An important result of the study is that the asymptotic analytical values of the integral parameters are closer to the experimental values as compared to the standard first adjustment results. Moreover, the asymptotic analytical values seem rather independent of the specific a priori variance/covariance data used in the analysis, namely COMMARA-2.0 or BOLNA, despite different a priori analytical values respectively obtained with JEFF-3.1 or ENDF/B-VI.8 data. The asymptotic uncertainties obtained on the basis of the two libraries are also similar.

• 出版日期2014-10