Complex Successions of Rock Categories in Clastic Reservoirs: Extreme Geological Heterogeneity
Vargas-Guzman, J Antonio *1
(1) Saudi Aramco, Dhahran, Saudi Arabia.
Modeling complex spatial patterns of rock-bodies in the subsurface is critical to achieve realistic evaluations of hydrocarbon production scenarios in flow simulation models. Three or more spatial locations of the same rock type might belong to a single curvilinear rock-body in the subsurface (e.g., meandering or braided channels with highly permeable sandstones). While the characterization and modeling of “pairs” of correlated rock locations is commonly practiced in geostatistical models using covariances (i.e., variograms), three or more connected locations are not in practical use. A lot of petroleum geoscience effort has been placed on the use of proportions for connected rock locations (i.e., data events) extracted from numerical “training outcrop images” (i.e., the multipoint statistics approach). Analogue images of outcrops are used to compute numerical multipoint probability; however, the fundamental higher-order spatial relations have been elusive.
This paper revisits various findings made by the author regarding the multipoint phenomena. Volumetric evaluation is comparable to upscaling, where detailed heterogeneous reservoir parameters (e.g., net thickness, porosity and fluid saturation) are replaced by upscaled (reservoir average) numbers. The study starts by quantifying bias, which occurs when higher-order terms are ignored in the computation of hydrocarbon in place volumes. It is therefore important to search for meaningful ways to avoid biased proportions in the geological interpretation and spatial modeling of rock- bodies. This study introduces the application paradigm of higher-order cumulants, which yields a tool to characterize the probability of coarsening and fining upwards rock sequences. The higher-order multipoint probability estimation problem is resolved by proposing a novel and practical numerical model based on cumulants (Kappa model). This approach is used to characterize the occurrence of “multiple” rock locations conforming to connected curvilinear rock-bodies. In addition, the Kappa model is a predictor comparable to the Bayesian exercise used to count exhaustive rock occurrences in complex stratigraphic sequences.
Curvilinear sandstone rock bodies are targeted for enhanced oil recovery using maximum reservoir contact wells, and optimized completions. Therefore, the results of this study are paramount to modeling extreme clastic geological heterogeneity for reservoir development and hydrocarbon production.
AAPG Search and Discovery Article #90141©2012, GEO-2012, 10th Middle East Geosciences Conference and Exhibition, 4-7 March 2012, Manama, Bahrain