A Rigorous Algebraic-Analytical Method for Pore Network Extraction from Micro-Tomography Images

Abstract Static and dynamic properties of porous media are highly dependent on its internal geometry. CT scan images are generally used to characterize porous media geometry. Direct simulation of fluid flow on CT scan images is possible but considerably time-consuming. In this study, a new method was developed for extracting a simplified representation known as “pore network model” by utilizing a rigorous algebraic-analytical method. By using a moving frame in the 3D matrix of the CT scan image and stepwise identifying-removing of image components, running time for a 4003 voxels sample in a typical computer system decreased to less than 350 seconds. The identification of throats was based on a new clustering method maximizing connectivity degree for an extracted model. For a sandstone sample, the developed model was consistent with the “maximum ball” model. The values obtained for average coordination number from the model proposed in this work and one proposed by Dong and Blunt using the maximum ball, were 4.13 and 3.98, respectively. Moreover, comparing the results of the proposed model with the ones obtained by Dong and Blunt using the maximum ball, average pore and average throat radii [PR,TR] were [26.43, 12.23] and [15.36, 7.01], respectively. The numbers of pores and throats [NP, NT] for the proposed model were [4921, 9494] and [6004, 12067] were reported by Dong and Blunt for the maximum ball model. In addition, an identification algorithm of curved throat was developed and validated with the help of a simple synthetic sample.