Understanding Flow Behavior in Carbonate Reservoirs from Facies-Based Earth Models*
Marjorie Levy1, William Milliken1, Paul M. (Mitch) Harris1, and Sebastien Strebelle1
Search and Discovery Article #40288 (2008)
Posted July 12, 2008
*Adapted from oral presentation at AAPG and AAPG European Region Energy Conference and Exhibition, Athens, Greece, November 18-21, 2007. See companion article, "Importance of Facies-Based Earth Models for Understanding Flow Behavior in Carbonate Reservoirs", Search and Discovery Article #40306 (2008).
Reservoir models attempt to mimic the distribution of reservoir properties in subsurface systems, and in carbonate reservoirs should capture geologically meaningful and realistic heterogeneity. In this study, we explore grainstone-dominated systems. On the basis of modern analogs from the Bahamas, grainstone shoals are modeled with linear or crescent-shaped bars, and include barcrest, barflank, and island facies.
A series of flow experiment studies investigate the impact of a variety of model input parameters on different measures of flow performance. The first study focuses on using three different modeling techniques: SGS, a non-facies-based method using a continuous variogram; SIS, a facies-based method using an indicator variogram, and MPS, a facies-based method using a training image and facies probability cube. The second study considers the impact of reservoir facies percent on flow behavior and how this changes with different modeling methods and different facies geobody shapes. The third study is a Plackett-Burman experimental design varying a) proportions of reservoir facies vs non-reservoir facies, b) proportions of barflank/barcrest reservoir facies, c) dimensions of facies geobodies, d) dissolution or cemented diagenetic zones, e) stratigraphic cyclicity f) the porosity histogram, g) the permeability transform, and h) spatial distribution of reservoir facies (distributed across platform vs. localized).
Each model was tested using reservoir simulation and considered different development scenarios. Models were compared on the basis of static measures of OOIP, reservoir connectivity, and permeability heterogeneity and on the basis of dynamic measures of recovery factor vs. time, recovery factor vs. pore volumes injected, net present oil, cumulative oil produced, and water breakthrough time.
Using experimental design, we examine the uncertainty in input parameters on flow performance using Multiple Point Statistics for a synthetic carbonate platform.
The objectives of this study are to:
Methodology includes using:
MPS is an innovative reservoir facies modeling technique that uses conceptual geological models as 3D training images to generate geologically realistic reservoir models:
The 3D training image is a rendering of the geological model that defines relative facies body dimensions and shapes, as well as associations between facies.
Givens for study: