Year
2000
Abstract
Recently, the Nuclear Materials Identification System (NMIS) was employed to verify both the mass and enrichment of a number of high-enriched uranium (HEU) metal items stored at the Oak Ridge Y-12 Plant. NMIS was applied for this measurement as an active neutron interrogation system; the measurement performed was similar in nature to a pulsed neutron measurement in that the distribution of detector counts following source emission was accumulated and subjected to analysis. In order to develop a calibration “surface” versus mass and enrichment, Monte Carlo models of the item geometry were employed to calculate the distribution for a matrix of independently varying masses and enrichments spanning the nominal declared mass and enrichment of the actual items. Each calculated distribution was then decomposed into its moments, and an empirical model of each moment as a function of mass and enrichment was developed. These models were then simultaneously analytically inverted to yield a nonlinear calibration surface that predicted mass and enrichment given the low-order moments of a measured distribution. The uncertainty in the calibration was estimated using standard squared-error minimization and propagation-of-error techniques, and the bias (and its uncertainty) in the calibration was evaluated using measurements of several randomly selected samples. No single mass or enrichment verified using this method deviated from declaration by more than 5.0%. The typical, i.e., root-mean-squared, error in mass was only 1.6% of the declaration, and the typical error in enrichment was only 1.5% of the declaration. As a result, a finding previously issued for this storage area by the Department of Energy, Oak Ridge Operations, (DOE-ORO) was resolved.