A New Approach and Computational Algorithm for Sensitivity/Uncertainty Analysis for SED and SAD with Application to Beryllium Integral Experiments

Abstract

A new approach for treating the sensitivity and uncertainty in the secondary energy distribution (SED) and the secondary angular distribution (SAD) has been developed, and the existing two-dimensional sensitivity/uncertainty analysis code, FORSS, was expanded to incorporate the new approach. The calculational algorithm was applied to the ⁹Be(n,2n) cross section to study the effect of the current uncertainties in the SED and SAD of neutrons emitted from this reaction on the prediction accuracy of the tritium production rate from ⁶Li (T₆) and ⁷Li (T₇) in an engineering-oriented fusion integral experiment of the U.S. Department of Energy/Japan Atomic Energy Research Institute Collaborative Program on Fusion Neutronics in which beryllium was used as a neutron multiplier. In addition, the analysis was extended to include the uncertainties in the integrated smooth cross sections of beryllium and other materials that constituted the test assembly used in the experiment. This comprehensive two-dimensional cross-section sensitivity/uncertainty analysis aimed at identifying the sources of discrepancies between calculated and measured values for T₆ and T₇. Without considering the uncertainties in the SED and SAD of the ⁹Be(n,2n) cross section, the uncertainties in T₆ are ∼2 to 3% in the Li_2O breeding zone, whereas they are ∼10% when the uncertainties in the SED and SAD are included. The contribution from the uncertainties in the SAD was small (∼1%) compared with the contribution from the uncertainties in the SED. As for T₇, the uncertainties in the Li_2O zone with and without considering the SED and SAD results are 4 to 7 and 2 to 5.5%, respectively. The estimated uncertainties in T₆ and T₇ could partly cover the observed discrepancies between calculations and measurements, although other sources have been identified. Although the approach followed to complete the uncertainty analysis is not standard because of the absence of an existing file that contains the uncertainty information in the SED and SAD of the ⁹Be(n,2n) reaction, the results obtained by introducing approximations to these data clearly demonstrate the importance of accounting for the uncertainties in the SED and SAD when a complete cross-section sensitivity/uncertainty analysis is to be performed.