Advanced Monte Carlo Methods for Eigenvalue Sensitivity Coefficient Calculations.

Abstract

The need to perform eigenvalue sensitivity and uncertainty analysis with greater fidelity for more complex systems has motivated the extension of the current multigroup sensitivity coefficient methodologies to continuous-energy applications. This research presents a survey of existing eigenvalue sensitivity coefficient methods, and develops two new approaches for sensitivity coefficient calculations. The newly-developed CLUTCH sensitivity method has demonstrated the ability to accurately and efficiently calculate sensitivity coefficients for a set of multigroup test problems, and has shown potential for future continuous-energy calculations. The accuracy, speed, efficiency, and memory requirements of the different sensitivity methods are quantified for a set of well-known criticality safety test problems, and the strengths and weaknesses of each method are assessed. The CLUTCH method requires a weighting function to tally the importance of neutronic reactions, and this research quantifies the level of accuracy needed by this weighting function to successfully calculate sensitivity coefficients. Lastly, general recommendations on the applicability of each sensitivity coefficient are given, and future approaches to improve the performance and of the methods are offered.