Real-Time Reservoir Management for CO₂ Enhanced Oil Recovery and Geological Sequestration: An Integrated Framework for Monitoring, Modeling, and Optimization in Offshore Fields

Abstract

Managing reservoirs for CO₂ injection in offshore fields calls for an approach that combines modern monitoring, simulation and optimization methods together. At the moment, oil firms have to deal with doubts in determining the structure of reservoirs, slow data processing and choosing between profits and caring for the environment. This research suggests using different fields and deep learning to keep an eye on conditions in real time, perform data-based reservoir simulations and perform optimization. A MobileNet v2 and Faster R-CNN model in the framework are used to identify oil leaks and monitor the reservoir with more than 90% accuracy at 28 frames per second, so that anomalies are quickly detected. At the same time, machine learning models assessed with field observations cut simulation time by a factor of 700 to 5000 compared to conventional methods, making it possible to provide ongoing estimates of both oil and CO₂ related factors. Operationally, the optimization module changes water-alternating-gas (WAG) ratios and well distances, trying to equalize the costs and emissions levels with smart strategies. When combined with intelligent well completions, tracer injections and phase-wise development, offshore applications such as Petrobras’ Lula field gain from better reservoir connectivity and less expensive work. Ensuring compliance with CO₂ floods, trapping of CO₂ by solubility and dynamics of mixture miscibility adds strength to the framework’s performance over the years. When we balance deep learning, reservoir simulations and economic-environmental trade-offs, the approach contributes to making offshore operations sustainable and supports carbon neutrality goals