An improved understanding of flow characteristics and imbibition processes in the highly complex natural fracture network of the Spraberry reservoir has been realized through detailed reservoir simulation. Hydrocarbon storage in the Spraberry is primarily in the low permeability (<0.5 md) matrix, while fractures provide nearly all flow capacity. Matrix imbibition processes play a critical role in waterflood performance of the naturally-fractured Spraberry reservoir. Although complex fracture and matrix systems can be represented in high resolution static geological models and discrete fracture networks, fine-scale dynamic flow behavior is typically lost in reservoir simulators due to grid coarsening required to manage simulation run times. For this study, grid resolution included simulation grid blocks (1x1 ft) smaller than average fracture spacing (3 ft). This resulted in pure matrix cells and cells containing both matrix and fractures. Explicit modeling of the fracture network, requiring over a million grid blocks to model a 5-acre area, would be impractical to use in most reservoir simulators. In this study, solutions were obtained with manageable run-times using a reservoir simulator optimized for processing on multi-thread, multi-core CPUs.
High resolution reservoir simulation of a small 5-acre area of the Spraberry provides improved understanding of flow characteristics and imbibition processes involved with waterflooding the low permeability, complexly fractured Spraberry reservoir. It also provides insight to impacts of variations in rock-fluid properties and parameters used to characterize the fracture system. The simulation model improves accuracy of calculating upscaled effective properties for use in coarsely-gridded dual-porosity models. This work can lead to increased efficiencies of Spraberry waterflood performance and allow for improved accuracy of reservoir simulation modeling of naturally fractured reservoirs.
Complex systems involving both matrix and natural fractures, like that seen in the Spraberry1,2, can be represented in high resolution static geological models. Modeling their unique contributions to flow at a fine scale in a dynamic reservoir simulator is important because water-imbibition displacement takes place at the interface between these two systems and can be the key driver to recovery in fractured reservoirs3,4,5. However, fine-scale modeling in a dynamic reservoir simulator becomes problematic due to lengthy run-times. Therefore, the user is typically forced to coarsen the dynamic model grid and upscale properties combining contributions from both matrix and natural fracture systems into effective properties. A side-effect of averaging these two systems over a large area is that their fine-scale interaction is then less preserved, understood, visualized in the simulator. This also leads to the danger of using incorrect upscaled properties in the coarse model. Upscaled properties can be estimated from many different statistical or flow-based techniques6, all of which may be inaccurate if not calibrated.