Year
2004
Abstract
The Segmented Gamma Scanner (SGS) is a long established workhorse nondestructive assay instrument for drummed gamma emitting waste and scrap. The technique combines quantitative high-resolution gamma spectrometry with an axial scan of the item as it is rotated to produce an estimate of the radioactivity content on a segment by segment basis. Matrix attenuation compensation factors are based on determining the diametrical transmission factor for each segment using an external transmission source. Within each layer the calibration assumes that the material composition and density is constant and also that the distribution of activity is uniform. Because the field of view of the collimator generally takes in segments above and below the one being targeted there is also an implicit assumption that the variation axially is smooth and gradual. In practice, when counting statistics are not limiting, the major source of measurement uncertainty when using the SGS is due to the unknown spatial variation in the activity distribution. Methods to partition the data taking as a function of rotation angle have been developed to identify non-uniform distributions of matrix and activity but the traditional approach does not implement them and most SGS systems currently in use follow the traditional approach. This being the case it is necessary to include a realistic and justifiable allowance for the uncertainty in the matrix correction factor due to potential non-uniformity of the source distribution in the overall evaluation of the Total Measurement Uncertainty (TMU) assigned to the assay. In this work we present a simple means of achieving this. The model is a based on the far field approximation and makes use of the fact that when cast in terms of diametrical transmission, the shape of the response curve as a function of radius within a layer follows a universal curve. We parameterize the minimum, volume averaged and maximum responses in terms of closed-form expressions in terms of transmission and use these to bound the response about the expected (uniform distribution) value. The true value should lie within ±s of the reported value for all practical purposes. Invoking the central limit theorem and assuming that real drums typically have a minimum of three equivalent concentrations (packets or ‘points’) of activity allows us to obtain an algebraic expression for the fractional uncertainty which is reasonable for all practical SGS assays. We illustrate the approach with calculations over the energy range 60keV to 1408keV representative of standard 208-liter waste drums containing matrices of densities 0.11g.cm-3 to 1.8g.cm-3