The Leading Edge · 2018 · 11 citations · 16 references
EngineeringEarth SciencesGeologyPetrophysicsReservoir CharacterizationRock MechanicsPermeability CharacterizationPermeability EstimationPore Structure AnalysisRock PermeabilityRock PhysicsRock PropertiesClastic Rock PhysicsHigh-resolution ImagingIntrinsic PropertiesK-means ClusteringMicro-computed TomographyRock Physics AnalysisRock Physics InterpretationThin Section Analysis
Abstract Predicting rock elastic properties and permeability from high-resolution 2D thin sections has been a challenging problem in rock physics because the 2D thin sections reveal very little about how the microstructure connects in the third dimension. However, 2D thin sections are widely available and inexpensive because they are often produced as a part of the reservoir-quality workflow. Furthermore, they have much higher resolution and greater field of view than micro X-ray computed tomography images, which are commonly used for rock properties estimation. The 2D thin sections we studied are from various hydrocarbon-bearing clastic formations with a variety of provenances, depositional environments, and burial histories. The high-resolution 2D images were scanned from these physical 2D thin sections. K-means segmentation was then employed to identify different minerals and pores for creating 2D binary images. The focus of this study is to simulate 2D elastic properties and permeability from 2D thin sections and then to employ various empirical relations to transform these 2D simulation results to 3D intrinsic rock properties. We compared the rock properties from this process to those from core measurements and measured wireline logs and found that these 2D to 3D rock property transformations yield promising results, especially for elastic properties. The results show that 2D thin section images have high enough resolution to resolve grain contacts very well. Predicting the permeability from 2D thin sections is still challenging since the process requires fitting the physical equation in order to retrieve the fitting coefficient for prediction due to our lack of understanding of the difference between 2D and 3D pore size distribution.
16
Data clustering: 50 years beyond K-means
Anil K. Jain · Pattern Recognition Letters · 2009 · 9.2K citations
Digital rock physics benchmarks—Part I: Imaging and segmentation
Heiko Andrä, Nicolas Combaret, Jack Dvorkin et al. · Computers & Geosciences · 2012 · 698 citations
Engineering, Earth Sciences, Physics +12
Flow in simulated porous media
P. M. Adler, C.G. Jacquin, J. Quiblier · International Journal of Multiphase Flow · 1990 · 601 citations
Digital rock physics benchmarks—part II: Computing effective properties
Heiko Andrä, Nicolas Combaret, Jack Dvorkin et al. · Computers & Geosciences · 2012 · 582 citations