3D reconstruction of porous media from a 2D section and comparisons of transport and elastic properties

High-resolution, 3D microstructure images of rocks can be used to compute the transport and elastic properties of those samples by use of a digital rock-physics approach. Those properties are complex functions of the pore-size distribution, geometry, and morphology, and necessitate the use of accurate 3D volumes. Because of the limited availability of 3D microstructure images of rocks, several attempts to construct 3D images from 2D images have been made. In this study, we propose a new stochastic method to reconstruct a 3D image of the rock by use of only a 2D section of the imagedrock sample. Our method is derived from a simple observation that the pores are gradually deformed from one section to the next. Therefore, the first step is to generate multiple independent realizations by performing multiple-point-statistics (MPS)-based stochastic simulations. These simulations represent independent 2D scans through the rock volume. Next, a succession of images is generated spanning two adjacent independent sections. These images consist of gradually morphed features from one section to those in the next independent section. Juxtaposing these 2D images results in the reconstructed 3D image. We implement the algorithm on a Berea sandstone rock for which the 3D high image was available for comparison. We calculate the spatial connectivity in the third direction, and confirmed that the proposed method can retrieve the connectivity in the third direction accurately. We also compute transport and elastic properties from the reconstructed image and from the original image to verify that this method reproduces the appropriate spatial statistics and pore-size distribution, geometry, and morphology. We also compare the numerical results with laboratory measurements performed on the sample. The results obtained by use of the reconstructed image reveal that the numerically calculated properties are similar to the measured values. We compare the mismatch of transport and elastic properties with the original measurements with that of the previous reconstruction algorithm. These comparisons show that the proposed simulation method has the same accuracy as previous ones. However, the proposed method is much more computationally efficient than the other algorithms, which are dependent on simulation of all 2D layers, mainly because of the faster MPS algorithm and the fact that the simulation is being performed only for the independent layers instead of all the layers. The proposed methodology is an accurate method to reconstruct a 3D representative sample of a rock given only one 2D thin section. The algorithm is computationally efficient and faster than the previously introduced algorithm, and can easily be used to characterize samples for which 3D images are difficult to obtain in terms of both time and expense.