Long-term water flooding (LTWF) is an efficient way to improve oil recovery (EOR) in carbonate reservoirs in the Middle East. Due to the complex depositional environment and intricate pore-throat structures of carbonate reservoirs, the development characteristics differ significantly from those of conventional sandstone reservoirs. While the mechanisms of LTWF in carbonate reservoirs are well-documented, there remains a significant gap in understanding the microscopic pore-scale displacement characteristics and the dynamic evolution of residual oil. To address this, nuclear magnetic resonance (NMR) and computed tomography (CT) scanning techniques were employed to investigate the behavior of LTWF across various carbonate rock samples. Initially, NMR technology was utilized to elucidate the pore-throat displacement characteristics at the microscopic level for different core samples under LTWF. Subsequently, CT scanning was applied to explore the dynamic evolution of microscopic residual oil and to categorize the types of residual oil based on their formation mechanisms. We found that LTWF predominantly utilizes oil within microscale pores of 1–10 μm and > 10 μm. As the volumes of injected water increase, there is a noticeable improvement in oil displacement within submicron pores (0.1–1 μm). However, residual oil primarily accumulates in nanopores (< 0.1 μm) and submicron pores. The study identified five distinct types of microscopic residual oil: clustered, throat, droplet, corner adsorbed, and pore lining. Notably, the transformation of residual oil in dolomite cores generally shifts from clustered to throat forms, while in limestone cores, it transitions from clustered to predominantly corner adsorbed and pore lining configurations. This nuanced understanding of oil utilization and residual categories under LTWF offers valuable insights into optimizing EOR strategies in complex carbonate reservoirs.

Modeling of multiphase flow and reactive mass transport in porous media remains a pivotal challenge in the realm of subsurface energy storage, demanding a nuanced understanding across varying scales. This review paper presents a comprehensive overview of the latest advancements in multiscale modeling techniques that address the inherent complexity of these processes. Three cutting-edge approaches are presented: hybrid multiscale simulation, which leverages both continuum and discrete modeling frameworks to enhance model fidelity; approximated physics, which simplifies complex reactions and interactions to expedite computations without significantly sacrificing accuracy; and machine-learning-assisted multiscale simulation, which integrates predictive analytics to refine simulation outputs. Each method presents distinct advantages and hurdles, collectively advancing the precision and computational efficiency of subsurface modeling. Despite the substantial progress, we recognize the persistent challenges, such as the need for more robust coupling techniques, the balance between model complexity and computational feasibility, and effectively combining machine learning with traditional physical models. Promising directions for future work are discussed to address these challenges, aiming to push the boundaries of current multiscale modeling capabilities.

Deep learning has been widely recognized in the field of CO₂ geological utilization and storage applications. With the development of deep learning algorithms, intelligent models are gradually able to improve multi-source, multi-scale and multi-physicochemical mechanism barriers with high-fidelity solutions in practical applications. In this perspective, an overview of the traditional and state-of-the-art deep learning architectures involved in CO₂ geological utilization and storage is outlined in terms of evolutionary trajectories. Meanwhile, the favorable directions and application scenarios of different deep learning algorithms for geo-energy intelligence modeling are summarized. Moreover, further insights into the future direction of deep learning burgeoning architectures in this field are provided. The physics-guided deep learning, explainable artificial intelligence, and generative artificial intelligence are expected to deliver more accurate solutions for information extraction and decision support within the CO₂ geological utilization and storage communities.