Hybrid Method for Eigenvalue Sensitivity Coefficient Calculations: Part II

The SCALE 6.3.2 code package contains two methods for determining the sensitivity to keff for each nuclide reaction-specific cross section in the TSUNAMI-3D sequence using continuous energy-the iterated fission probability method (IFP) and the Contributon-Linked eigenvalue sensitivity/Uncertainty estimation via Tracklength importance CHaracterization (CLUTCH). The CLUTCH method has been observed to have difficulties in generating accurate sensitivity coefficients for systems with large reflectors (HEU-MET-FAST-028-001) and hydrogenous materials, such as polyethene (PU-MET-MIXED-002). As a result, a new hybrid methodology within the CLUTCH sequence was developed to replace the existing F*(r) weighting function calculated during the inactive generations of the transport sequence with a deterministic adjoint importance solution. The results of the new hybrid method showed significant improvement in the accuracy of the generated sensitivity coefficients and was comparable to those generated with IFP and multigroup methods. This paper extends the hybrid methodology approach by examining the external adjoint solutions generated from the deterministic code, Denovo, with several additional systems using SCALE 7.0beta10. The hybrid method was tested against several critical benchmark experiments to provide a variety of fissile materials, fissile forms, and energies of average lethargy causing fission (neutron spectra). Results demonstrate that the hybrid method with an external adjoint solution provides comparable or even slight improvements over the additional sensitivity coefficient methods and direct perturbation results.