Towards autonomous subsea processing: A review of integrated production systems, modeling, and multiobjective optimization

In response to rising energy demand and the depletion of conventional resources, offshore oil and gas is increasingly pursuing subsea production architectures that leverage advanced sensing, control, robotics, and data-driven decision support. This review provides a comprehensive, metrics-first analysis of autonomous subsea processing systems that shift critical functions from topside facilities to the seabed – separation, boosting/pumping, compression, heat exchange, and control – to improve efficiency, reduce lifecycle costs, and mitigate environmental impacts under harsh deepwater conditions. Synthesizing the open literature and selected industry evidence, we consolidate the sustainability metrics most relevant to autonomy-enabled subsea systems: energy intensity (e.g., kWh per produced barrel of oil equivalent), operational greenhouse-gas intensity (kgCO 2 e per boe, including power supply and flaring where reported), and embodied emissions associated with subsea equipment and power/control infrastructure. We then explain how autonomy changes these outcomes through distinct levers: closed-loop set-point optimization of pumps and compressors; improved flow assurance via multi-physics modeling and real-time condition monitoring; reduced chemical use, downtime, and shutdowns through predictive maintenance; and fewer offshore interventions and logistics emissions enabled by remote operations and subsea robotics. The article's core contribution is an integrated framework linking: (i) subsea hardware choices (heat exchangers, gas–liquid separation, pumps, compressors, and control systems) with comparative analysis against topside systems as a baseline; (ii) AI-enabled digital twins for behavior prediction in dynamic environments; and (iii) multiobjective optimization methods that balance economic performance, environmental sustainability, and operational reliability. Finally, we highlight gaps in consistent CO 2 e and embodied-carbon boundaries, scarcity of public field datasets for validation, and the need for safety- and cybersecurity-aware benchmarks for AI-in-the-loop control, outlining priorities for progressing toward resilient, fully autonomous subsea processing.