Coal seams are increasingly regarded as promising sites for CO₂ storage, with cleat networks serving as both the primary storage space and the key pathways controlling coal’s mechanical and fluid flow behavior. Although adsorption–desorption experiments and permeability models are well established, there is still very limited direct 3D imaging evidence of how CO₂ injection alters cleat and fracture structures in coal. Previous studies have captured bulk property changes (e.g., porosity, permeability) but missed the spatial microstructural evolution and the associated microscopic fluid dynamics.

X-ray microCT (µCT) offers a powerful means to directly visualize coal microstructures and fluid dynamics in three dimensions at micron-scale resolution. When coupled with advanced image processing, for instance machine learning–based segmentation, µCT can quantify changes in fracture intensity, connectivity, and fluids distribution during and after CO₂ injection—providing unprecedented insights into the fundamental mechanisms and challenges of CO₂ storage in coal seams.

Despite its potential, several challenges remain. Small-scale fractures near the resolution limit of µCT are difficult to segment reliably, leading to underestimation of fracture connectivity. Similarly, resolving phase boundaries between injected CO₂ and other fluid within fractures is complicated by grayscale overlap and image quality/resolution. These challenges highlight the need for improved segmentation algorithms and novel imaging and experimental set-ups to fully capture the complex mechanisms governing CO₂ storage in coal.

In this project the student will be working on the development of advanced image processing methods and the design of novel experimental set-ups that allow in-situ CO₂ injection under controlled conditions. The project will provide new insights into the fracture evolution, fluid distribution, and storage potential of coal seams.

Some knowledge of rock mechanics, flow in porous media and experience in image processing, programming.