SUMMARY

With the development of more powerful computational hardware and innovative transport solution routines, it is now feasible to quickly and accurately solve a wide variety of nuclear systems. These solutions, however, are governed by the overall accuracy of the reactor modeling and ultimately by the physics data. Improved nuclear data are vital to the continued safe operation and design of nuclear systems; small perturbations to an isotope's cross-sections can cause substantial changes in the neutron flux and k-effective. The need to ascertain the cross-section energy behavior aims at correctly calculating the cross section self-shielding effect that ultimately allows improved reactor calculations.

The different energy regimes of the cross section are determined by the physics of the interaction between the neutron and the target nucleus. In the resolved resonance region (RRR), the R-Matrix theory has been shown to represent the underlying reaction physics quite well. In the unresolved resonance region (URR), the same core physics apply, but a different approach has been used for cross section reconstruction. In contrast to the RRR, the URR cross sections are calculated using a statistical approach based on average parameters obtained from the resonance region. Subsequently, these average parameters are used to generate R-Matrix like resonance parameters. The issue with the URR approach is that a crude approximation to the R-Matrix, namely Single-Level Breit-Wigner (SLBW), is used to create the probability tables. While the underlying routine used in the URR to generate probability tables is sound, the accuracy of the results is driven by the choice of the cross-section formalism.

The proposed research will improve the probability tables generated for the URR. The SLBW formulae will be replaced by the robust R-Matrix limited formalism. The newly implemented formalism will then be verified in the RRR by reconstructing the cross sections directly from the resonance parameters provided in the ENDF/B-VII.1 libraries. Once the algorithm is verified, it will be implemented into a modified Monte Carlo pseudo-resonance ladder routine. Further, the new pseudo-resonance ladders will then be used to generate new ENDF libraries; the new libraries will then be used by NJOY to reconstruct cross sections for comparison with the new methodology. These new cross-section libraries will then be used in several well-documented benchmarking problems for verification.