Application of Sensitivity/Uncertainty Methods to Burnup Credit Criticality Validation

The responsible use of calculational methods in nuclear criticality safety includes a determination of bias and bias uncertainty that may exist between the calculated results and reality. Such biases exist due to approximations used to model the real world, uncertainties in nuclear data, and approximations associated with the calculational method (e.g., Monte Carlo method). The bias and bias uncertainty are typically determined by using the modeling approximations, nuclear data, and calculational method to model well-known, usually critical, systems. Unfortunately, the bias and bias uncertainty determined in this manner can vary significantly depending on the characteristics of the known “benchmark” systems. The most accurate determination of bias and bias uncertainty is obtained by using benchmark systems that are very similar to the real-world operational configuration, the subcriticality of which must be safely ensured. Historically, similarity has typically been determined using comparisons of gross integral parameters (e.g., lethargy of average energy of neutrons causing fission and hydrogen to fissile nuclide ratio) and on qualitative comparisons of the geometry and materials present in the benchmark and application systems. The development of sensitivity/uncertainty methods permits detailed quantitative comparison of these systems. The work presented in this paper is a sensitivity/uncertainty-based study of the similarity or applicability of many critical experiment models to a model of a high-capacity transportation cask that is loaded with spent commercial nuclear fuel and flooded with water. This paper includes descriptions of the sensitivity/uncertainty methods used, the operational configuration of interest, benchmark critical configurations used for comparisons, and discussion of the results from the sensitivity/uncertainty analyses.