Application of Multivariate Analysis to Gamma and Neutron Signatures from Spent Nuclear Fuel

High-resolution passive gamma spectroscopy of spent nuclear fuel (SNF) provides the intensity of gamma emission lines of isotopes present in the SNF. Typically, gamma spectroscopy analysis focuses on a limited number of these isotopes and their respective gamma emission lines. However, using multivariate analysis (MVA), a larger number of intensities can be studied simultaneously, providing more comprehensive information about the SNF. MVA is not limited to data analysis on the response from one single measurement technique; one of the benefits is that it allows for simultaneous analysis of data from different measurement techniques.In the first part of this paper, MVA techniques are used to study the ability to correctly classify spent nuclear fuel as being intact or suffering from partial defects on the 30% level. Intact and partial defect PWR 17x17 SNF are modelled using Serpent2. The gamma emission full peaks as measured by the high-purity germanium (HPGe) gamma spectroscopy station at the Swedish Central Interim Storage Facility for Spent Nuclear Fuel (Clab) are modelled. The results are analyzed using MVA classification algorithms, and show that classification is possible.In the second part of the presentation, the gamma spectroscopy data from intact PWR 17x17 fuel are combined with results from simulations of the prototype Differential Die-away Self-Interrogation (DDSI) instrument developed at Los Alamos National Laboratory. In this part, the aim is to determine the spent fuel parameters Initial Enrichment (IE), Burn-Up (BU) and Cooling Time (CT). Serpent2 is used for burnup calculations and the DDSI response is modelled using MCNP6. Results of the combined MVA is presented and show that, if CT is known, BU and IE can be determined with RMSE of 2.4 GWd/tU and 0.4% respectively.