Imaging Mass Transfer in Nanoporous Shale Using AI Tools for Improving X-ray CT Scan Resolution

Abstract:

Characterization of mass transfer mechanisms in geological systems is relevant to hydrocarbon production, geothermal energy, waste storage, and carbon sequestration. In conventional systems with appreciable permeability, fluid transport is typically convective and the fluids behave as continua. Shale systems, however, are more complex and are characterized by very small pore sizes, porosities, and permeabilities. In nm-sized pores characteristic of impermeable shale matrix, the continuum approximation progressively breaks down. Therefore, fluid transport in shale exhibits different transport mechanisms depending on the Knudsen number (ratio of mean free path to pore diameter) and the pore pressure. As pore size get smaller, fluid transport is dominated by diffusion. Diffusion models (similar to Fick’s First Law) have been developed to express flow rate in different regimes and are based on diffusion coefficients. This project characterizes unfractured Eagle Ford shale cores (2.5 cm diameter by 8 cm long) using X-ray Computed Tomography (CT) in order to quantify in situ porosity distribution and the major heterogeneities. Liquid-liquid and gas-gas diffusion coefficients are measured using X-ray CT imaging of the progress of mass transfer and cross-validated by comparison with Nuclear Magnetic Resonance (NMR) measurements. The samples utilized have average porosities between 5 to 10%. Importantly, a deep Convolutional Neural Network (CNN) was implemented for improving the X-ray CT scan resolution of conventional CT images of unfractured shale samples using high-resolution microCT images.