Spent-fuel composition: a comparison of predicted and measured data

The uncertainty in predictions of the nuclear materials content of spent light-water reactor fuel was investigated to obtain guidelines for nondestructive spent-fuel verification and assay. Values predicted by the reactor operator were compared with measured values from fuel reprocessors for six reactors (three PWR and three BWR). The study indicates that total uranium, total plutonium, fissile uranium, fissile plutonium, and total fissile content can be predicted with biases ranging from 1 to 6% and variabilities (1-sigma) ranging from 2 to 7%. The higher values generally are associated with BWRs. Based on the results of this study, nondestructive assay measurements that are accurate and precise to 5 to 10% (1sigma) or better should be useful for quantitative analyses of typical spent fuel