UPSCALING AND MODELING IN A FRACTURED CARBONATE

The objective is to describe the integration of a fracture characterization study with fracture modeling, flow-property upscaling, flow simulation and well test matching for a fractured tight carbonate. The test area is a non-economic hydrocarbon reservoir at depths of more than 150 m located in the Abadia Valley, central Portugal. A conceptual geological model for the Abadia test volume is constructed that includes a small-scale fracture network and a number of major faults. Three wells have been drilled in the test volume with well spacing of 100-200 m. One of the major faults intersects all wells. The modeling is guided by a characterization study using the available data from a petrophysical log correlation of the three test wells, lithological core descriptions, core and log based fracture analysis, geochemical analyses of oil samples, and hydraulic well test results. Fracture and fault geometry and properties are reflected in the simplified discrete fracture network (DFN) model which includes only the features believed to be the main contributors to flow. Laterally varying properties of fault planes have not been included as heterogeneity.

The upscaling from the DFN model is performed with a combined gridding and upscaling tool. Because of the large difference in scale between the features to be included in the model, a step-wise upscaling procedure has been chosen. The grid cell permeabilities are calculated and upscaled with a 1-phase pressure solver. The upscaling can both calculate the bulk permeability 3D tensor for the full volume, and upscale into a new gridded model where a coarse grid can be specified.

From the combined fracture modeling and upscaling procedure are derived both some bulk flow properties and a coarse-grid reservoir model by a 2-step upscaling procedure. A comparison between reservoir test data and the model with upscaled flow properties is performed with flow simulation in the reservoir model. This investigation has shown the success of a consistent procedure for the characterization, modeling, and upscaling, in that a match can be obtained for some of the tests. However, it becomes apparent that some of the modeling tools are not yet capable of accepting information about important heterogeneities, e.g. the variable fault zone permeability.