Uncertainty Components for Two Approaches to Spent Fuel Assay

The Next Generation Safeguards Initiative is developing nondestructive assay (NDA) methods to assay Pu mass in spent fuel assemblies. Uncertainty quantification is an important task in most assay methods, and particularly for spent fuel assay. A computer model (MCNPX) was used to predict the isotope masses and the spatial distribution of masses in the spent fuel assemblies dependent on three inputs: initial fuel enrichment (IE), fuel utilization (burnup, BU), and cooling time (CT). A variety of virtual assemblies were created for a range of BU, IE, and CT and additional computer modeling was done to simulate the expected detector responses (DR) for any of various NDA measurement options such as differential dieaway or passive neutron albedo reactivity. The DR is given in terms of multiple outputs including the effective fissile content or the content of particular fissile isotopes such as 239 Pu and 240 Pu. Computer model uncertainty is therefore expected to be a large component of the total uncertainty in using measured IE, BU, and CT plus MCNPX and one or more measured DRs to predict fissile content and total Pu mass. This paper describes uncertainty components for each of two analysis approaches being followed. The first approach (?traditional?) attempts to correct for neutron absorbers that impact the relation between DRs and Pu mass for different values of IE, BU, and CT. The second approach (?code emulator?) interpolates MCNPX at various (IE, BU, CT) values that are weighted by agreement between measured and MCNPX-predicted DRs. This work is part of a larger effort sponsored by the Next Generation Safeguards Initiative to develop an integrated instrument, comprised of individual NDA techniques with complementary features, that is fully capable of determining Pu mass in spent fuel assemblies