Precision of the Accidentals Rate in Neutron Coincidence Counting

The quantification of Pu present, for example, as process hold-up in glove boxes or in waste containers may be accomplished by passive neutron coincidence counting (PNCC) to determine the Pu-240 effective mass which may then be related to the total Pu mass or Pu-239 mass through the isotopic composition. When multiplication can reasonably be taken as slight, assay results are normally based on the Reals coincidence rate since in contrast the specific Totals rate is sensitive to (α,n) production which may be difficult to estimate accurately. The precision of the net Reals rate and hence the quality of the assay depends on how well the subtraction of the Accidental coincidence rate can be made. Three approaches are readily available with current standard shift register modules, these are the signal triggered (or measured) Accidentals rate, the rate calculated from the Totals event rate, and the rate estimated according to the so called fast accidentals sampling method. The NDA literature contains practically no systematic information on when and how calculated Accidentals and fast Accidentals should be used to complement or replace the conventional signal triggered (measured) estimate. This work aims to redress this by making a systematic study, in the low efficiency domain typical of glove boxes and basic large volume waste assay systems. We discuss how good measurement control can be attained with shorter measurement times or lower efficiency when the best Accidentals treatment is used. PNCC is widely used for waste assay and MC&A throughout the nuclear fuel cycle. Achieving highly accurate results in a timely and cost effective way is an important operational goal, especially in the planning of new facilities and in the context of He-3 gas shortage. This work looks at how data analysis techniques can contribute to this end. We report a systematic study performed using Cf-252 sources measured between a pair of thermal neutron slab counters configured to achieve a detection efficiency ranging from about 2.5% to about 9.6%. Various coincidence gating structures and rates were examined. We discuss the findings in terms of practical performance advantage and, in the case of Totals and Reals compare the observed precisions to simple theoretical estimates.