Analysis Of Fast Neutron Data From A Fresh Fuel Assembly

Understanding data better is always an unsolved problem. This is particularly true when employing neutron coincidence counting for the non-destructive assay of, e.g. 235 U in fresh fuel assemblies. For the infinite time limit Böhnel theory that is usually appropriate for 3 He based neutron detectors, we have developed an analysis that employs a sequential Bayesian sampling-importance-resampling particle filter to make non-parametric estimates of the posterior probability distributions for the theory’s parameters: mass of the isotope that undergoes spontaneous-fission, multiplication, detection efficiency, and neutron diffusion time. List-mode, time-tagged neutron multiplicity data provides opportunities to optimize one's ability to extract the underlying source information. For low-count rate systems, the likelihood function we use is based on the distribution of waiting times between neutrons. For high-count rate systems, counting distributions form the basis of our likelihood function. Our algorithm can assay highly-enriched uranium (HEU) in a few minutes with just 500 neutrons, is self-calibrating, and quantifies uncertainties. For fast neutron counting using liquid scintillators, e.g., the fast-neutron coincidence collar (FNCL), the likelihood function needs to be different because we must analyze the data against the full time-dependent statistical theory of fission chains. The time-of-flight to the detector, _TOF -1 , is usually longer than the time scale τ between fissions. For this reason, neutrons arrive at the detector spread out by the time scale λ _TOF -1 and neutrons from different fissions in the chain are interleaved in time. In the case of a fresh fuel assembly (FFA) for a reactor, the 235 U content is very small compared to 238 U so the neutrons scatter around for a long time before inducing a fission. The FFA is essentially a neutron scattering “pinball machine.” When λ _TOF -1 &lt&lt τ, it becomes necessary to add prompt neutron leakage to the theory. Multiple neutrons from a single induced fission can leak “instantaneously” compared to the time scale τ. This paper will discuss our modification to the statistical theory of fission chains and our statistically sophisticated data analysis methods.