Validation of in situ object counting system (ISOCS) mathematical efficiency calibration software

The ISOCS calibration method is a convenient tool for calibrating the detec tor efficiency as a function of energy for a wide variety of source geometr ies and activity distributions. The ISOCS method consists of a Canberra cha racterization of the detector, user input of source geometry data, and the ISOCS software which uses these to produce the efficiency calibration. Duri ng the characterization, an MCNP model of the detector is developed. The mo del is then independently validated using measurements with a NIST traceabl e source. Given the validated model, the response characteristics of the de tector are mapped out to cover any location inside a sphere of radius 50 m, centered on the detector, and over a photon energy range of 50 keV-7 MeV. The ISOCS software contains a series of mathematical models that can simula te a wide variety of sample shapes. The software divides each source region into a number of voxels. Inside each voxel, a point location is defined in a quasi-random fashion. At a given energy, the detector efficiency is calc ulated for each voxel, taking into account the attenuation due to absorbers both inside and outside the source. The efficiencies for all the voxels ar e summed up at the given energy. To determine the accuracy of this calibrat ion method, a large number of tests (about 109) were performed. In each of these tests, a reference efficiency calibration was compared to an ISOCS ef ficiency calibration at the same geometry. The reference calibration was ei ther from a full MCNP calculation, or from a multi-energy radioactive sourc e. The tests were categorized into three different counting geometries, nam ely, Field, Laboratory, and Collimated geometry. The data for each geometry were further divided into low energy (<150 keV) and intermediate to high-e nergy (>150 keV) groups. The mean ratio of ISOCS/True efficiencies was (i) 1.01 +/- 0.007 for the Field geometries, (ii) 0.97 +/- 0.007 for the Labora tory geometries, and (iii) 1.09 +/- 0.014 for the Collimated geometries. By analyzing the relative uncertainties in the True efficiencies, and the rel ative standard deviation in the ratios, the average relative standard devia tion due to ISOCS is estimated to be 6.5%, 5.4%, and 10.5%, for the Field, Laboratory, and Collimated geometries, respectively. Various sources of bia s affecting the data have been identified from this validation process. Imp rovements have been made in the characterization process and in the algorit hms, which will be implemented in future versions of the ISOCS efficiency c alibration software. (C) 1999 Elsevier Science B.V. All rights reserved.