Digital Twins and Artificial Intelligence for Intelligent CO₂ Storage Management

Abstract

As geological CO₂ storage (GCS) advances from pilot demonstrations toward gigatonne-scale commercial deployment, the industry increasingly requires management systems capable of continuously integrating real-time monitoring data with predictive subsurface models to support safe, adaptive, and cost-effective operations. Digital twins—dynamic, uncertainty-aware virtual replicas of subsurface storage systems that assimilate real-time data through advanced artificial intelligence (AI) techniques—have emerged as a unifying framework for achieving this vision, bridging the historically siloed domains of geophysics, reservoir engineering, and operational decision-making. This review examines the state of digital twin and AI-enabled intelligent CO₂ storage management, synthesizing recent advances in generative AI-based data assimilation, graph neural network surrogate modeling, uncertainty-aware forecasting, and reinforcement-learning-based reservoir control. It draws on case studies from heterogeneous depleted gas reservoir optimization, geothermal-analogue reservoir characterization, and geomechanically informed reservoir assessment, and further considers methodological parallels with large-scale, real-time AI monitoring and predictive analytics frameworks developed for national healthcare systems. The review concludes by identifying persistent challenges in the transition from uncertainty-aware digital shadows to fully decision-capable digital twins, computational scalability, and multiphysics integration, and proposes a research agenda for advancing digital twin technology toward routine, trustworthy use in commercial-scale CO₂ storage management in the years immediately following its projected 2026 maturation milestones.