Non-destructive assay of drum package radioactive wastes utilizing tomographic gamma scanning

A methodology for non-destructive assay (NDA) of drum packaged radioactive waste materials is investigated using Emission Computed Tomography (ECT) procedures. This method is suitable for mapping the activity distribution of point sources encased within typical 208 liter drums containing a primarily homogeneous medium. A requirement of this method is accurate gamma attenuation correction. This is accomplished by the use of a constant density distribution for the drum content, thereby requiring the need for a homogeneous medium. The current predominant NDA technique is the use of the Segmented Gamma Scanner (SGS). Tomographic Gamma Scanning (TGS) improves upon this method by providing a low-resolution, three-dimensional image of the source distribution, yielding both spatial and activity information. Reconstruction of the source distribution is accomplished by utilization of algebraic techniques with a nine (9) by six (6) voxel model with detector information gathered over scanning intevals of ninety (90) degrees. Construction of a linear system to describe the scenario was accomplished using a point-source response function (PSRF) methodology, where a 54x120 matrix contained the projected detector responses for each source-detector geometry. Entries in this matrix were calculated using the point-kernal shielding code QAD-CGGP. Validation was performed using the MCNP photon transport code. Solutions to the linear system were determined using two distinct least squares procedures, the Non-Negative Least Squares (NNLS) algorithm and the LSMOD algorithm. Both function as subroutines, each called from a user-written FORTRAN program. A series of four (4) scans were performed, each reconstructing the source distribution of a mock-up waste package containing a single 73 mCi ¹³⁷Cs point source. For each scan, the source was located in a different location. Results of the reconstruction routines accurately predict the location and activity of the source. The range of activity calculated using the NNLS routine is 0.2681 mCi with an average value of 77.7995 mCi. The range of values calcuated using LSMOD is 5.1843 mCi with an average of 72.8018 mCi.