Performance Model for the Lawrence Livermore National Laboratory Passive- Active Neutron Drum Shuffler

A performance model of the measurement performance of the Lawrence Livermore National Laboratory (LLNL) passive-active neutron (PAN) drum shuffler was developed for use in measurement of highly enriched uranium (HEU) oxide and the HEU content of mixed U-Pu oxide items. Earlier papers [1, 2] described the PAN shuffler calibration over a range of item properties by standards measurements and an extensive series of detailed simulation calculations, and provided an illustration of the model. With a single normalization factor, the simulations agree with the HEU oxide standards measurements to within ±1.2% at one standard deviation. Measurement errors on mixed oxide samples are in the ±2% to ±10% range, or ±20 g for the smaller items. The performance model should not add more than 1% to 3% uncertainty to the evaluated results. The simulations account for item mass, diameter, density, isotopic and atomic proportions, and measurement properties such as geometry and irradiation protocol, which affect the count rate per fissile atom in the item. The detailed simulations provide a wealth of intermediate information, e.g., the fissions produced in each of the isotopes present, and the count rate per delayed neutron emitted. A synthesis performance model based on the MCNP simulations allows interpolation between the simulation calculations, and a modest degree of extrapolation, and thus allows the calculation of an item’s HEU mass from a count rate. The model’s approach is to model sub-processes separately, and to use analytical equations consistent with the physical processes, and curve fitting and interpolation logic. An error analysis tabulates the net potential error from the simulations, curve fitting, and interpolation.