Nuclear spent fuel parameter determination using multivariate analysis of fission product gamma spectra. (December 2017)

• Record Type: Journal Article
• Authors: Hellesen, C.; Grape, S.; Jansson, P.; Jacobsson Svärd, S.; Åberg Lindell, M.; Andersson, P.
• Is Part Of: Annals of nuclear energy. Volume 110(2017:Dec.)
• Page Start: 886
• Page End: 895
• Publication Date: 2017-12
• Subjects: Multivariate analysis -- Principal component analysis -- Partial least squares regression -- Gamma ray -- Nuclear fuel -- Safeguards; Nuclear energy -- Periodicals; Nuclear engineering -- Periodicals; 621.4805
• DOI: 10.1016/j.anucene.2017.07.035
• Languages: English
• ISSNs: 0306-4549

Abstract

Highlights: A multivariate analysis model for determination of spent nuclear fuel parameters is presented. Cooling time, burnup and initial enrichment can be obtained from one gamma ray spectrum. The multivariate model is made from simulated data. The multivariate model is also successfully applied to real measurement data. Abstract: In this paper, we investigate the application of multivariate data analysis methods to the analysis of gamma spectroscopy measurements of spent nuclear fuel (SNF). Using a simulated irradiation and cooling of nuclear fuel over a wide range of cooling times (CT), total burnup at discharge (BU) and initial enrichments (IE) we investigate the possibilities of using a multivariate data analysis of the gamma ray emission signatures from the fuel to determine these fuel parameters. This is accomplished by training a multivariate analysis method on simulated data and then applying the method to simulated, but perturbed, data. We find that for SNF with CT less than about 20 years, a single gamma spectrum from a high resolution gamma spectrometer, such as a high-purity germanium spectrometer, allows for the determination of the above mentioned fuel parameters. Further, using measured gamma spectra from real SNF from Swedish pressurized light water reactors we were able to confirm the operator declared fuel parameters. In this case, a multivariate analysis trained on simulated data and applied to real data was used.