Depositional Space: Construction and Applications to Facies and Petrophysical Property Simulations

Laurent Souche, François Lepage, T. Laverne, C. Buchholz

International Petroleum Technology Conference · 2015 · 12 citations · 13 references

Concepts

Abstract

Abstract In the absence of strong diagenetic or compaction trends, the anisotropy and distribution of facies and petrophysical properties are essentially the result of the composition of the paleoenvironment at the time of deposition and of geological phenomena such as erosion, faulting, and folding, which subsequently altered the geometry of the rocks. Understanding and isolating these two parameters (depositional and post-depositional effects) is key to the accurate interpolation and extrapolation of subsurface properties. This requires establishing a mathematical transform between the present-day geometry of rocks and their geometry at the time of deposition. The most common solution consists in relying on the layering and indexing of 3D grids to approximate paleodepositional coordinates. However, it is shown that for both pillar and stair-step grids in presence of tight folds or low-dipping or reverse faults, the variogram ranges, orientations, and facies proportions are significantly distorted when geostatistics are based on cell indexing. Four essential properties that a rigorous depositional space should honor are then described and the shortcomings of the existing solutions are analyzed. A novel approach in which the representation of a depositional space is independent of a 3D reservoir grid is then introduced, and it is shown that solid mechanics provide a robust and practical way of building such a computational space. The use of depositional coordinates for performing variogram analysis and property interpolation is also exposed. The proposed approach yields facies distributions that accurately honor the model parameters independently from the structural configuration, while accounting for complex stratigraphic and depositional patterns (erosions, onlaps, etc.) and integrating nondepositional trends and overprints. We demonstrate that, in a variety of geological settings, better results are obtained when property interpolation and simulation do not rely on grid geometry and indexing.

References

13