Calibration of Multiscale Heterogeneities in Carbonate Reservoirs for CO₂ Storage Modelling

The study focuses on enhancing CO₂ storage modeling in carbonate reservoirs, which are crucial for hydrocarbon recovery and CO₂ sequestration due to their complex, multiscale pore structures. Traditional characterization methods struggle with the intricate heterogeneities of carbonate rocks, which feature both macroporosity and microporosity. To address this, the researchers developed an integrated multi-scale workflow combining micro-computed tomography (micro-CT), confocal microscopy, pore network modeling (PNM), and machine learning techniques, including generative adversarial networks (GANs). Using carbonate samples from the Arab D limestone in Saudi Arabia, the methodology involved resolving macropores with micro-CT and capturing microporosity through confocal microscopy. Machine learning generated synthetic 3D microporous structures, which were incorporated into the new Extensive Pore Modelling (XPM) software to simulate fluid flow at multiple length scales. Various modeling scenarios demonstrated that microporosity critically influences fluid connectivity and flow behavior, significantly affecting CO₂ injectivity and storage security. For instance, multiple microporous phases improved the realism of permeability curves and enhanced residual CO₂ trapping. The results underscore the necessity of accurately characterizing microporosity for reliable CO₂ storage predictions. The integrated approach allows for robust scenario testing and better reservoir characterisation, paving the way for improved strategies in CO₂ sequestration and long-term storage management.