Year
2016
Abstract
In nuclear fuel cycle safeguards, special nuclear material held up in pipes, ducts, and glove boxes causes significant uncertainty in material-unaccounted-for estimates. Quantitative imaging is a pro- posed nondestructive assay technique with potential to estimate the holdup mass more accurately and reliably than current techniques. However, uncertainty analysis for quantitative imaging remains a significant challenge. This work demonstrates an analysis approach for data acquired with a fast- neutron coded aperture imager. The work includes a calibrated forward model of the imager. Cross validation indicates that the forward model predicts the imager data typically within 23%; further improvements are forthcoming. A new algorithm based on the chi-squared goodness-of-fit metric then uses the forward model to calculate a holdup confidence interval. The new algorithm removes geometry approximations that previous methods require, making it a more reliable uncertainty esti- mator.