Stochastic shale permeability matching: Three-dimensional characterization and modeling

Abstract

The recently-discovered promise of shale-gas reservoirs necessitates their characterization. As a first step, two-dimensional (2D) scanning electron microscope (SEM) imaging can provide excellent view of the complexity associated with such reservoirs, and has been used in recent years to study organic materials, clays, minerals, and the pores in shales. Other important information, such as the connectivity of the pores in 3D and such macroscopic properties as the permeability are difficult to infer from 2D SEM images. Newer techniques, such as focused ion-beam SEM (FIB-SEM) have been utilized to overcome the shortcomings. The extremely small sample size and the costs associated with the FIB-SEM technique limit, however, wide use of the FIB-SEM in characterization of shale samples. An alternative approach is to use the recently developed advanced algorithms for 3D reconstruction that use one or a few 2D samples. Such methods may not, however, be able to fully reproduce the shale permeability. To accurately reproduce the permeability, we propose a new method based on a combination of the Metropolis-Hastings and the genetic algorithms. The new method learns from its own previously generated realizations of the shale and produces models that match the existing permeability data. The method is validated with the measured permeability for an actual 3D shale sample. It generates an ensemble of stochastic realizations that honor the permeability data, which may then be used for more accurate characterization of shale gas reservoir and analysis of their pore network.