Preliminary Geostatistical Investigations Of Tar-Bearing Sands, Lease-13, Athabasca Tar Sands

Abstract

Abstract Preliminary geostatistical studies of the tar sand measures in the Lower Cretaceous McMurray formation have shown that variograms of net thickness and grade of tar sand are well formed. These show that the variogram ranges (radii of influence) of net thickness and grade are 4,000-5,000 feet, much wider than the sampling density. Fines, the clay and silt fraction within the orebody, were studied in conjunction with tar grade for one of the orebodies. The best estimation of fines was found to be from a combined, correlation of tar grade and natural radioactivity (gamma-ray emission). The variogram of fines in this orebody indicated that fines are randomly distributed. It was concluded that the sampling done on Lease-1S was sufficient for locating all major orebodies and for estimating the tar reservoirs accurately. The sampling was also sufficient for initial mine planning. One use of the geostatistical data, to estimate the 'Variance of tar grade and fines in the mill feed based on one mining method, is described. Introduction THE ATHABASCA TAR SANDS of Western Canada represent a large potential reserve of liquid hydrocarbon which is only now being developed. One commercial mine is in operation and a second is under construction. Tar, or bitumen, the objective mineral, is a black hydrocarbon so viscous that it Is immobile at normal in-situ rock temperatures. The tar occurs as pore filling in the sand portion of a sand shale series of Cretaceous age. The depositional environment of the formation is deltaic to near-shore where it lies close enough to the surface to be mined. In these areas, the tar sands are overlain mainly by a thin mantle of Pliestocene glacial till and underlain by Devonian carbonates. The material for this study comes from Shell Canada's Lease-13, located about 50 miles north of the town of Fort McMurray. Lease-13 is a block roughly 6 by 13 miles. One-Dimensional Variograms We decided to undertake statistical studies of the sands to give us a better predictive tool for ore reserves and for use in mine planning. At the time, we had information from 350 sample holes in which a foot-by-foot analysis of 40,000 feet had been made. In the analysis, overburden and other rejects had been identified, and a base to the ore picked. Further, within the ore, ore grade and fines had been determined. This material was available on a computer tape data file. The sampling pattern for the 350 holes is shown on Figure l. Our first approach, to see whether geostatistics would be useful, was to make a simple one-dimensional non-directional test of values. For this test, we determined for each hole a single value for each of the parameters we were to study - tar sand net pay, tar grade thickness, overburden thickness, total reject thickness and quality factor. Quality factor is an economic parameter relating overburden thickness, ore thickness and ore grade. The variograms for this study were developed using the onewdimensional form of Matheron's (1,2) formula, as follows: