Validation of the Fast-Neutron Multiplicity Expressions for Fissile Mass Estimation

Year
2017
Author(s)
D.L. Chichester - Idaho National Laboratory
S. D. Clarke - Department of Nuclear Engineering & Radiological Sciences, University of Michigan
S. A. Pozzi - Department of Nuclear Engineering & Radiological Sciences, University of Michigan
M.J. Marcath - Department of Nuclear Engineering and Radiological Sciences, University of Michigan
T. H. Shin - Department of Nuclear Engineering & Radiological Sciences, University of Michigan
M. Y. Hua - Department of Nuclear Engineering & Radiological Sciences, University of Michigan
Abstract
Neutron multiplicity analysis is a non-destructive assay technique that involves detection of the multiplets of neutrons emitted by an unknown sample to estimate its fissile mass. Typically, multiplicity analysis is conducted by means of a neutron multiplicity counting (NMC) system that uses capture-based thermal detectors (e.g. 3He-gas detectors). To relieve the reliance on such detectors, we have developed a fast-neutron multiplicity counting (FNMC) system that uses scatter-based, organic scintillators to directly detect the fast fission neutrons emitted from the sample. FNMC systems, however, are prone to effects of neutron cross-talk, where a single neutron may cause more than one count. This leads to an overestimate of the fission rate (F) and inaccurate estimation of other sample parameters such as the leakage multiplication (ML) and the alpha-ratio (a) when using the traditional capture-based NMC equations. We have formulated fully-generalized multiplicity expressions that account for cross-talk effects for any arbitrary order N and these expressions were verified with MCNPX-PoliMi simulations. This paper will present validation of the proposed generalized multiplicity expressions with experimental data. Pu metal plates were measured using the FNMC system and the data was used for neutron multiplicity analysis to estimate sample parameters F, ML, and a. We demonstrate that accounting for neutron cross-talk effects of order N = 2 improves the accuracy of the estimated sample parameters.