Design of a neutron spectrometer and simulations of neutron multiplicity experiments with nuclear data perturbations

Master of ScienceDepartment of Mechanical and Nuclear EngineeringJ. Kenneth ShultisSimulations were performed using MCNP5 to optimize the geometry of a neutron spectrometer. The cylindrical device utilizes micro-structured neutron detectors encased in polyethylene moderator to identify sources based on energy spectrum. Sources are identifi ed by comparison of measured detector responses to predetermined detector response templates that are unique to each neutron source. The design of a shadow shield to account for room scattered neutrons was investigated as well. For sufficient source strength in a void, the optimal geometric design was able to detect all sources in 1000 trials, where each trial consists of simulated detector responses from 11 unique sources. When room scatter from a concrete floor was considered, the shadow shield corrected responses were capable of correctly identifying 96.4% of the simulated sources in 1000 trials using the same templates. In addition to spectrometer simulations, a set of neutron multiplicity experiments from a plutonium sphere with various reflector thicknesses were simulated. Perturbations to nuclear data were made to correct a known discrepancy between multiplicity distributions generated from MCNP simulations and experimental data. Energy-dependent perturbations to the total number of mean neutrons per fission [average velocity] of [superscript]2[superscript]3[superscript]9Pu ENDF/B-VII.1 data were analyzed. Perturbations were made using random samples, correlated with corresponding covariance data. Out of 500 unique samples, the best-case [average velocity] data reduced the average deviation in the mean of multiplicity distributions between simulation and experiment to 4.32% from 6.73% for the original data; the average deviation in the second moment was reduced from 13.87% to 8.74%. The best-case [average velocity] data preserved k[subscript]e[subscript]f[subscript]f with a root-mean-square deviation (RMSD) of 0.51% for the 36 Pu cases in the MCNP validation suite, which is comparable to the 0.49% RMSD produced using the original nuclear data. Fractional shifts to microscopic cross sections were performed and multiplicity and criticality results compared. A 1.5% decrease in fission cross section was able to correct the discrepancy in multiplicity distributions greater than the [average velocity] perturbations but without preserving k[subscript]e[subscript]f[subscript]f