Pore Network Modeling: Analysis of Pore Size Distribution of Arabian Core Samples

We use X-ray microtomography (micro-CT) to image rock cuttings of poorly consolidated sandstone and vuggy carbonate from Saudi Arabian oil and gas fields. The cuttings are a few mm across and are imaged to a resolution between 3 and 12 microns. The details of the three-dimensional pore space can be clearly seen. A maximal ball algorithm is used to extract a topologically equivalent pore network: the largest inscribed spheres in the pore space represent pores, with throats representing the connections between them. The results are validated through comparison with networks derived by a different method from idealized sphere packings and Fontainebleau sandstone. The aim of this work is to input the models into pore-scale network models to predict macroscopic properties such as relative permeability and capillary pressure. This acts as a valuable complement to special core analysis, enabling predictions of properties – such as three-phase relative permeabilities and the impact of wettability trends – outside the range of parameters probed experimentally. Furthermore, using microtomography, rock cuttings can be analyzed that are too small for conventional core flood experiments. Introduction Pore-scale network modeling can now be used to predict multiphase flow properties as a complementary tool to special core analysis. However, the pore structure of the rock and its wettability needs to be determined. It is possible to image the three-dimensional pore space directly using micro-CT scanning that has a resolution of a few microns. However, it is difficult to simulate quasi-static multiphase flow directly through these images. Instead a toplogically equivalent network of pores and throats is extracted through which flow is computed. This paper uses a novel method for this network extraction using maximal balls, validates it against other methods and applies it to a series of Saudi Arabian reservoir rocks. At present the networks that are used for predictions are derived from an analysis of two-dimensional thin section images of sandstones . From these the grain size distribution is determined. Then packings of these grains with subsequent compaction and diagenesis is simulated. Since the locations of the grain centers is known it is possible to find pores and throats in the void space and from this to extract a topologically equivalent network. 9 However, this processbased method is restricted to granular media and does not make use of three-dimensional images if they are available. Micro-CT imaging We extract pore networks from micro-CT images. A micro-CT scanner at Imperial College London, Fig. 1, and a synchrotron tomographic scanner at ELETTRA in Italy have been used to image sandstone and carbonate samples. The best spatial resolution obtained so far in our series of experiments is 2.9 μm on a carbonate sample with a diameter of 2 mm. Fig. 1. The left picture shows the micro-CT scanner at Imperial College London. The right picture shows the X-ray tube in this scanner and the sample stage. The micro-CT scanners output three-dimensional (3D) arrays of reconstructed linear X-ray attenuation coefficient values (CT numbers), which can be viewed as gray scales in image processing software. The raw images are filtered to smooth the image, reduce noise and improve the contrast between grain and void. We use the median filter which replaces the gray scale value of a voxel by the median value of the nearest 26 surrounding cells. Then a threshold value is chosen to binarize or segment the gray scales into two phases: solid and void. The effect of image processing is seen in