Resonance parameter adjustment in the resolved region based upon an Asymptotic Generalized Linear Least-Squares methodology in conjunction with the Monte Carlo method

Abstract In order to reduce methodological effects, a stochastic technique is proposed for adjusting basic ENDF parameters instead of preprocessed multi-group data as it is normally the case; the required sensitivity coefficients of the integral parameters to these parameters are recomputed by means of NJOY/Serpent within an iterative scheme based upon the Asymptotic Progressing Incremental nuclear data Adjustment (APIA) methodology. It is anticipated that the provided illustrative examples including selected benchmarks along with ENDF parameters to adjust are certainly limited, making the study too preliminary in order to reach meaningful conclusions on the quality of the posterior nuclear data. However, they were conceived for showing how the proposed novel methodology along with its main principles could be generalized and equally used in the framework of a future exhaustive adjustment, largely justifying this paper. More specifically dealt with in this study is the assimilation of the effective multiplication factor for low-enriched, U-solution, thermal systems by adjusting resonance widths for U-235 and U-238. It is shown that for the experiments part of the assimilation, the posterior ratios of computed to experimental central values are exactly equal to one, reflecting the general situation. This outcome may be expected from the theory, provided that convergence can be reached, which is always the case if the chosen experiments are not too strongly correlated among each other from the analytical viewpoint. Otherwise, differently from non-iterative methods, the procedure, since diverging, cannot provide useful data: in this case, some experiments of less relevance should be withdrawn. A similar approach could thus be used by considering a wide database in conjunction with a broader selection of basic parameters, to provide adjusted ENDF format files ensuring, by means of a reference stochastic tool, equality between computed and experimental values for a large variety of experimental data e.g. those available in the International Handbook of Evaluated Criticality Safety Benchmark Experiments (ICSBEP).