Effects of weight percent fines have been considered in the past as important factors in controlling the recoveries of bitumen from tar sands using batch extractors. However, in the marine sands of the Athabasca deposit, fines break down as a predictor of batch recovery; in some places even low fines tar sands do not yield acceptable batch bitumen recoveries. It was predicted that a sand unit high in rough-surfaced, diagenetically altered grains would yield low primary bitumen recovery. Scanning electron microscopy of the solids from batch extraction tests revealed that quartz grains with very rough, pitted, and overgrown surfaces retain bitumen. (These solids appear to be transported preferentially with the bitumen to froth and secondary tailings.) Grains with smooth sur aces were found to be dominant in primary tailings.

Fourier grain-shape analysis was employed to identify tar sand shape types since scanning electron microscopy inspections are time-consuming and expensive. Five grain shape families were so identified and verified by SEM.

Two types of grains unaltered by diagenesis are very fresh, rounded, sub-aerially abraided grains, and fresh, angular grains with concoidal fractures. Three other grain shape families represent particles whose surfaces were strongly modified by diagenesis. The diagenesis was found to be primarily overgrowth: (1) rough silica plastering with a sponge-like appearance, producing high surface areas; (2) silica plastering with multiple crystals; and (3) complete simple overgrowth.

Fourier grain-shape analysis yields the proportion of high surface area grains in 900 grain samples taken from marine sand intervals of 3 cores. That proportion, plotted versus weight percent primary recovery by batch extraction, indicates that when the proportion of high energy grains exceeds 40%, primary recoveries were less than 80 wt. %. (Primary recovery, in contrast, shows no relationship to weight percent fines for these marine sands and cannot be used as a batch recovery indicator.)

The results suggest the feasibility of predicting recovery in advance of mining, thus permitting adjustments to mining/extraction strategies. It must be kept in mind, however, that extrapolation of bitumen recovery results from bench-scale batch tests to large-scale continuous units is not straightforward.

End_of_Article - Last_Page 549------------