A multiscale reconstructing method for shale based on SEM image and experiment data

Owing to the presence of multiscale pore structures, characterization of laminated shales is both extremely difficult and substantially different from that of conventional reservoirs, and defies conventional methodologies. In this paper, a multiscale reconstructing method for shale is proposed to generate 3D layer representative elementary volume (lREV)-scale digital-experimental models to characterize the multiscale pore structure of the shale by means of the combination of a large area SEM image, nitrogen adsorption and pressure pulse decay experiment result. In this method an improved multiscale superposition algorithm is introduced to integrate the reconstructed complex models from nanoscale to mesoscale together, and it can preserve the details and main features enormously of each typical component (nanoscale organic pores in organic matter and pyrites, micro-nano inorganic pores and micro slits) in the shale. Especially, to accurately reproduce the realistic morphology for shale, the proposed method uses the experimental pore size distribution and permeability as constrain conditions to adjust and optimize the lREV-scale digital-experimental model. Our proposed method was tested on Longmaxi and Wufeng shale samples, and the reconstructed lREV-scale digital-experimental model are proved to accurately describe the representative structure of the complex multiscale pore space of the typical layer of the shale. The success of this method provides a promising way for reconstructing more realistic model to continuously and systematically characterize the pore (slits) structure from the nanopore-scale to the lREV-scale. It can advance the understanding of the various gas transport mechanisms at different scales and will be helpful for understanding the quality of the shale reservoir.