Presented at WM2008 Conference February 24-28, 2008 Phoenix, AZ USA

ABSTRACT

During the decommissioning of structures and the remediation of soil, material that is known or suspected to be contaminated with radioactivity must be assayed. For many, if not most, of the cases, the radioactive material concentration can be determined or inferred by gamma spectroscopy. In most, if not all, of these cases, the radioactivity is not uniformly distributed. For these situations, is it better to perform in-toto measurements of the entire item, or is it better to extract a sample and analyze that sample? For the in-toto gamma spectroscopy measurement, the results will not be accurate unless the calibration method perfectly represents the measured item. And since the calibrations normally assume uniformly distributed radioactivity, any non-uniform distribution of radioactivity in the measured material will cause an error in the results. A mathematical program using probabilistic methods has been developed to estimate the uncertainty in in-toto gamma spectroscopy measurements. This program has the capability of determining the efficiency of various types of non-uniform radioactivity distributions and to compare it to the uniform distribution efficiency. Repeating this for a large number of random tries allows the estimation of the measurement uncertainty. A new feature has been added to the program to simulate the extraction of a sample from these same non-uniform distributions, to compute the average concentration in this extracted sample, and to compare it to the average concentration in the original item to be measured. Repeating this for a large number of random tries allows the estimation of the sampling uncertainty. These two computations were run on several different "typical" contamination measurement scenarios [200 liter drums, 27 cubic meter "truck" containers, 0.01-10 liter "laboratory" containers, and 3600 square meter soil "in-situ" plots]. In all cases examined, the sampling uncertainty was larger than the in-toto gamma measurement uncertainty - typically about an order of magnitude larger. This means that in-toto measurements are likely better for situations where the activity is not homogeneous, and where gamma spectroscopy is appropriate. If sampling is to be done, then the best results will be when the largest mass of sample is extracted and when it is made up from the largest number of discrete sub-samples. The more extreme the non-uniformity, the more important this is.