Abstract
Pore-level imaging and modelling were shown to be robust and useful techniques, at least if applied to conventional rocks such as sandstones. This type of modelling directly within the pore space of the imaged samples provides valuable insights into rock heterogeneity, estimates local single and multi-phase flow properties, and serves as a key tool for upscaling and parameterizing Darcian models. Yet, numerous problems are still to be solved related to rocks with complex and hierarchical structure, such as carbonates, shales and coals. These rocks possess pore sizes in a wide range of values which has to be imaged with different resolutions in order to capture all relevant pore scales. This is due to so-called sample size/imaging resolution trade-off. To develop a detailed 3D structure model, such rocks are imaged using different resolutions and even using different imaging techniques. The problem lies with combining all these multiscale images into a single 3D digital structure model. In this work the recently developed multiscale image fusion technique was tested on complex carbonate samples with hierarchical structure. For two samples we performed a detailed structural study on two different scales: 3D XCT scanning (2.7 µm resolution) and 2D SEM imaging (0.9 µm pixel size). These two scales were fused to represent carbonate rocks structure with the predefined resolution of 0.9 µm and volume of 15003 voxels combining structural features discernible on both XCT and SEM images. Fused 3D images were used as input data to a hybrid median axis/maximum inscribed ball pore-network technique with subsequent modelling of permeability. Resulting simulated values were compared with laboratory measurement on the cores with dimeter of 5 cm. For the Sample 1 micropores visible on XCT scan were not connected, thus, preventing any flow simulations. After fusion with SEM image simulated permeability agreed favourably with the measurements. For the Sample 2 micropore network was percolating, but simulated permeability was lower than the experimental one. Incorporating sub-resolution porosity in this sample by adding SEM finer porosity structure resulted in higher permeability value very close to the laboratory measurement. In this contribution we also discuss why simulated and measured permeability values do not agree perfectly, which is most likely due to the scale difference between the volumes of simulated and measurement domains. We also covered all major drawbacks of the multiscale image fusion techniques and discussed possible solutions. Current study clearly showed the potential of this novel approach to facilitate pore-level modelling of flow and transport in rocks with complex and hierarchical structure such as carbonates, shales and coals. We believe that after some improvements and rigorous testing multiscale fusion technique may become a core tool in imaging and pore-level modelling of flow properties for complex rocks with hierarchical structure.
