Methodology

The current effort has focused on upscaling the detailed 0.5-m by 0.5-m by 6-inch geostatistical models to reflect selective remediation units of more reasonable size for excavation by mechanical equipment. Accordingly, we have adopted an admittedly simplistic approach of averaging the detailed grid using first a 4-by-4 horizontal compositing scheme, which creates remediation panels 2-m by 2-m in size. Additionally, averaging has included several different depth increments. First, we consider retaining the 6-inch depth discretization (yielding a 4x4x1 pattern), then we consider a 1-ft depth increment (yielding a 4x4x2 pattern), and finally we consider taking the entire (baseline) 2-ft depth increment at a single pass (yielding a 4x4x4 upscaling pattern). The same exercise has been repeated for an 8-by-8 horizontal compositing scheme (remediation panels 4-m by 4-m in plan); both 8x8x2 and 8x8x4 schemes are presented. Upscaling is computed on a simulation-by-simulation basis so that the reasonably anticipated internal heterogeneity of ¹³⁷Cs is captured.

Note that because of a relatively simplistic implementation of the averaging process, the “excavation boundaries” provided originally by the baseline remediation plan may move outward from the original boundaries. This may include migration to include regions outside the physical limits of contamination (e.g., into a paved parking area). The initial geostatistical grid cells are simply processed four at a time (or eight at a time) in both the x- and y-coordinate directions (and by variable multiples in the vertical z-direction). Thus, if the original boundary fell within a new upscaled grid panel, the activities of all originally modeled cells were included in estimating the new, upscaled activity, even if some of those original cells were below the threshold value of 23 pCi/g, and even if some original cells were not modeled. This latter impact of cells not originally modeled influences the area/volume estimates; however, the average activity is based only on the values of those original cells that were modeled. Although a more sophisticated upscaling algorithm could be developed, the changes in contaminated volumes with progressive upscaling are sufficiently large that the major conclusions of this exercise are unlikely to change. “Spreading” of the excavation boundaries is even more pronounced for the 8x8 upscaling patterns than for the 4x4 patterns.

Click here to go to the next section

Contact: Chris Rautman