LAUSR

Abstract Numerical flow simulation in shale, especially in the nanoscale porous system of a shale matrix, is still challenging because no imaging device can effectively describe the nanoscale porous structure of shale to satisfy both resolution and field of view (FOV). The resolution of an X-ray computed tomography (µ-CT) image is too low to detect the nanoscale features of porous structure in shale. The FOV of a focused ion beam scanning electron microcopy (FIB-SEM) image, on the other hand, is too small to capture the heterogeneity of a shale matrix. Therefore, we propose a semi-quantitative, Multiscale reconstruction strategy to build an image based network model of a shale sample with nanoscale resolution covering three orders of magnitude of FIB-SEM image volume. In this study, shale is considered a Multiscale porous media consisting of microscale and nanoscale structures. The microscale structures reflect the morphological features of organic matter, clay minerals, intergranular pores and micro-cracks that can be detected by µ-CT. The nanoscale features focus on the inner porous structure of organic matter and clay minerals that can be characterised using a scanning electron microcopy (SEM) or a FIB-SEM. nanoscale. Multiscale means reconstruction work that is carried out at the microscale and nanoscale separately, using the multiple-point statistics (MPS) method. Microscale reconstruction aims to recover the connections that are undetected among microscale objects due to the low resolution of the µ-CT image. Within the organic and clay phases, a nanoscale reconstruction is then carried out to model the nanoscale porous structure. Semi-quantitative means the segmented µ-CT image is used as conditional data in microscale reconstruction to maintain the reality of microscale structures. To relieve the heavy burden of data storage, a network modelling procedure is simultaneously undertaken alongside nanoscale reconstruction to transfer the reconstructed structure to a network model. Based on the established network model, flow simulation is undertaken by applying an extended Beskok-Karniadakis (B–K) model considering continuum, non-continuum and surface diffusion. The results show that the permeability ratio is affected by pressure, the molecule diameter of gas and surface diffusion.