Partial defect identification in PWR spent fuel using passive gamma spectroscopy

Passive Gamma Spectroscopy is a robust and relatively simple method to analyze spent nuclear fuel in nuclear safeguards. Most often it is used to verify initial enrichment, burnup, and cooling time of spent fuel assemblies. This work explores the possible application of multivariate data analysis methods to analyze Passive Gamma Spectroscopy results in order to identify partial defects of spent nuclear fuel assemblies. For the work, the spent fuel inventory was computed with Serpent-2 for several burnup and cooling time values. In order to compute the passive gamma spectra, the in-pool passive gamma measurement station at the central interim storage facility for spent nuclear fuel (Clab) in Oskarshamn (Sweden) was modeled in MCNPX. A set of simulated passive gamma spectra were created for various partial defect arrangements. Using the simulated spectra, the possibility of using multivariate data analysis to classify partial defects was investigated. This was accomplished by training and testing a classification method on simulated data. It is found that 30% partial defects can be identified if the outer rods of the assembly were replaced with dummy rods. In case the inner rods were replaced, the method can identify the partial defect if the cooling time of the assembly is not longer than 20-25 years. (Author)