Using three-dimensional reconstructed microstructures for predicting intrinsic permeability of reservoir rocks based on a Boolean lattice gas method

Abstract

This paper presents a method for predicting the intrinsic permeability of porous media based on the integration of the local velocity field. Three-dimensional representations of the porous structure are reconstructed from two-dimensional binary images, after segmentation of digital images acquired from thin plates, commonly used in microscopy. Velocity field is calculated on these three-dimensional representations using a Boolean lattice gas method (LGA). Reconstruction is based on a Gaussian stochastic simulation. Mercury-intrusion results furnish auxiliary data that are used for the estimation of a critical percolation diameter and to establish a necessary condition for the binary source images to give accurate predictions of permeability, considering the intrinsic limitations of the reconstruction process. Reconstruction method and connection loss, resolution factor, adherence conditions and the effects of Boolean noise in the calculation of permeability are fully discussed. The method is used to simulate flows through several petroleum reservoir rocks, leading to intrinsic permeability prediction. Simulation is compared with experimental results. Considered as an intrinsic permeability prediction method based on the geometrical information that is possible to recovery from microscopy thin plates, three-dimensional reconstruction appears to be the most critical step in present simulation scheme.