Geostatistical realizations are often built at an arbitrary scale based on available data and computational resources. In certain settings, it may be necessary to downscale the realizations for flow simulation and local resource assessment. This is especially important in the Athabasca Oilsands where accurate flow simulation often requires numerical models with a very fine grid size. Flow simulation is undertaken for selected areas and realizations. It is intractable to construct the original geostatistical models at the fine scale. It is desirable to construct finer scale models that reproduce the original realizations exactly.

Approximate downscaling is always possible with geostatistical methods; however, it is of interest to create fine scale models that exactly reproduce the large scale models to ensure consistency and avoid potential biases. Direct block sequential simulation is developed to generate fine scale realizations that exactly reproduce block data.

A comprehensive case study is shown from the Athabasca Oilsands. Geostatistical realizations are constructed over 100s of square kilometers at a large scale. These realizations are locally downscaled to 20m by 2m by 2m for flow simulation around particular SAGD well pairs. The fine scale realizations are constructed such that they exactly match the initial coarse scale realizations. An approximate downscaling method is also used. The 3-D models and flow simulation results were compared to show the difference made by the exact downscaling method.

You can access this article if you purchase or spend a download.