Digital core modeling based on multiple-point statistics

Key words:  digital core  multiple-point statistics  pore structure  variogram  pore network model

Abstract:

There are three commonly used digital core modeling methods:X-ray computed tomography scanning, simulated annealing and process-based method. Although the X-ray computed tomography scanning method can build an accurate 3D digital core model, it is expensive and time-consuming. The simulated annealing and process-based method can reduce the cost and improve the modeling efficiency, but the simulated annealing method cannot effectively characterize the long-range connectivity of the pore structure, and the process-based method is not suitable for simulating the rocks that have undergone complex diageneses. In order to solve these problems, a method of reconstructing digital core model based on multiple-point statistics was proposed. Taking as an example the Berea sandstone of the Waverly group in the Michigan Basin, two representative volume elements with the size of 150³ voxels were extracted from different locations of the Berea sandstone. One was set for the training image, the other for the real model. Two orthogonal two-dimensional slices were chosen from the real model as the conditioning data. The 3D digital rock models were reconstructed using multiple-point statistics. The variogram, pore structure parameters and percolation parameters of single phase flow experiment were applied as the evaluation indexes to validate the accuracy of the reconstructed models.The result shows that the reconstructed models agree well with the real model in the pore-throat geometry and topology features, which verifies the accuracy of the reconstructed models and feasibility of reconstructing digital core model using multiple-point statistics.