Applications of High-Resolution Imaging and High-Performance Parallel Computing in Unconventional Energy Recovery

Abstract

Two shale gas rock samples, from a Middle East shale gas play and the Eagle Ford shale play, respectively, were scanned using a nanometer-scale focused ion beam-scanning electron microscope (FIB-SEM). The geometrical properties were extracted and compared. The high-resolution image data were then processed and used as boundary conditions in the pore-scale GPU-accelerated lattice Boltzmann simulator (GALBS) for permeability simulation. The GALBS is based on the lattice Boltzmann (LB) method and optimized by graphics processing unit (GPU) parallel computing. Image processing showed that although the intrakerogen pores in the Eagle Ford sample had larger pore volumes compared to those in the Middle East sample, their morphologies were more laminar, which leads to higher friction to fluid flow and consequently gives rise to lower macroscopic permeability. GALBS simulations confirmed that the permeability was at the nanodarcy (nd) level in the Eagle Ford sample, while it was at the microdarcy (μd) level in the Middle East sample. Furthermore, anisotropy in the permeability tensor was observed in both shale samples. The computing speed of the GALBS is more than 1,000 times faster than the serial code and more than 10 times faster than the parallel code run on a standalone CPU, which suggests that many more samples can be analyzed given the same processing time. The combination of high-resolution imaging methods and high-performance parallel computing is a powerful tool for studying microscopic processes and upscaling. It provides for a more accurate estimation of the total stored gas and is helpful in the optimization of hydraulic fracturing treatments, which are aimed at connecting as many isolated intrakerogen pores as possible. The method presented in this study enables more accurate characterization of microscopic geometries and faster upscale transport properties, illustrating that unconventional energy recovery requires unconventional solutions.

Temporal Evolution of the Geometrical and Transport Properties of a Fracture/Proppant System Under Increasing Effective Stress

“…This resolution is capable of achieving a balance between sufficient field of view and computing accuracy (Chen et al 2013); it is suitable for resolving the microscopic fractures on the matrix surface, but too coarse to resolve the microscopic pores in the rock matrix. Therefore, the porosity and permeability of a rock matrix without microscopic fractures were determined with laboratory measurements independently.…”

“…The focus of this study is the interaction between proppants and fractures; thus, the resolution of 30 lm/pixel was selected to resolve the processes at this scale. To resolve individual pores in tight rocks and shale kerogen, an image resolution as high as 10 nm/pixel was used in our previous work (Chen et al 2013). However, this resolution led to an extremely small field of view and was thus unable to characterize millimeter-scale fractures.…”

“…Raw CT images were processed with a powerful quantitativetool package for CT data (Ketcham 2005;Chen et al 2013) to remove background noise and segment components. The 3D pore structure was reconstructed with a commercial software application for scientific and industrial data visualization and analysis and used as boundary conditions in the lattice Boltzmann model for pore-scale flow simulations.…”

“…The D3Q19 lattice structure (19 velocity vectors in 3D space; Fig. 3a) was used because it has a good balance between computational stability and efficiency (Chen et al 2013). The LB equation for fluid flow is written as…”

Investigation of Gas Transport in Shale Reservoirs Using Multiscale Modeling and Principal Component Analysis

“…Based on image data, quantitative image analysis and pore-scale direct simulation were performed to obtain the geometrical and transport properties of kerogen, such as porosity, tortuosity, specific surface area, and absolute permeability (Chen et al 2013). Parallel computing on central processing units (CPUs) and graphics processing units (GPUs) was applied to optimize pore-scale direct simulation to handle larger computational domains (Chen et al 2014).…”