Workflow to Reconstruct 3D Pore Space from 2D CT-Scan Image of Berea Sandstone

Physical properties (e.g. porosity, permeability) prediction can be accurately estimated using numerical simulations of fluid flow through 3D pore space image. A digital volume can be obtained from CT-scan. However, for practical useful in massive experimentation the device is still prohibitively expensive and the scanning time is too long. An alternative is to reconstruct 3D volume from 2D image (e.g. thin section) information. These thin sections are relatively easy and cheap to prepare and often done at the drill site.
In this study, instead of using thin section, we used 2D image from CT-scan for training image. This image can be used as an input, similarly like thin section image. The density between pores and grains can be differentiated clearly from CT-scan measurement. The validation of the workflow can precisely validate since the 3D image is already there. 2D image was obtained from 3D x-ray tomography image of Berea sandstone. This data was used as the benchmark data for the test study of stochastic reconstruction. Berea sandstone generally serves as a basis for many rock physics experiments due to its fine-grained and well sorted characteristics. This sandstone also occurs in the oil and gas producing formation in the Michigan Basin. The goal of this study is to predict permeability using 2D image as input and validate the stochastically reconstructed pore space with actual porous media that obtain from CT-scan. LBM was used to simulate fluid flow in porous media to estimate permeability.
A geostatistics method (SIS) to reconstruct 3D porous media using 2D image was validated in this paper. Berea sandstone was used for this study because of their pore size and simplicity. Isotropy is assumed in the reconstruction; however more conditioning data, such as several images can be used as constrains and multi-orientation images can provide more statistics when anisotropy is expected. If detailed geological history is available, Process-based reconstruction (Oren and Bakke, 2003) is promising methods to reconstruct 3D porous media of rock. MPS is more general and it can generate realistic pore space image without knowledge of geological processes by which the rock was forms. However, this method still time consuming and take hours even to reconstruct one porous media (Okabe, 2004).
This method can be solution for simple, fast and robust prediction of porosity and absolute permeability from 2D image (e.g. thin section). Physical properties prediction; porosity and absolute permeability were compared with permeability estimation from CT-scan image and the results have good agreement with CT-scan image which is directly acquired 3D image from the CT-scan equipment.
Key Words — 3D pore space image, Stochastic reconstruction, Permeability, LBM, CT-Scan.