Autonomous AI System for Real-Time CO2 Integrity Monitoring and Dynamic Optimization in the CCUS Ecosystem

This paper presents an AI-powered digital solution for real-time monitoring and optimization of the carbon capture, utilization, and storage (CCUS) value chain. It integrates machine learning with physics-based statistical models to detect anomalies and leaks, ensuring the integrity of CO2 capture, transport, and storage systems. The objective is to enhance operational safety, responsiveness, and sustainability through intelligent alert propagation and root-cause diagnostics.

The system continuously ingests real-time operational data such as pressure, temperature, flow rate, and volume from all major CCUS components. An AI-powered alert management engine monitors these parameters and triggers alarms when anomalies are detected. Machine learning algorithms then correlate alerts across assets, identifying potential upstream or downstream causes. For example, a pressure anomaly at a well is automatically cross-referenced with pipeline data to determine if the root cause lies elsewhere. The system then generates a comprehensive, actionable alert, notifying relevant personnel with contextual insights. This closed-loop process ensures timely, accurate, and coordinated responses across the CCUS network.

The integration of AI with physics-based statistical models has significantly improved the accuracy and timeliness of anomaly detection across the CCUS value chain. The system successfully identifies deviations from normal operating conditions using real-time Z-score analysis and pressure-flow correlations. For example, a pressure drop in a storage well is automatically cross-referenced with upstream pipeline data to determine if the anomaly is due to a leak or upstream fluctuation. This correlation is achieved through machine learning models trained on historical and synthetic data, enabling the system to distinguish between benign and critical anomalies. Engineers receive enriched alerts that include both the immediate issue and its likely origin, reducing false positives and improving response times. Field tests have shown a 30–40% reduction in downtime and a measurable increase in asset reliability. The system also supports predictive maintenance by identifying early warning signs of equipment degradation. Overall, this approach enhances the resilience, safety, and efficiency of CCUS operations, supporting global decarbonization efforts.

This solution uniquely combines AI-driven alert propagation, real-time data correlation, and machine learning-based diagnostics within a unified digital platform tailored for CCUS. Unlike traditional monitoring systems, it autonomously identifies cross-asset dependencies and root causes, enabling predictive and coordinated responses. Its adaptive intelligence and system-wide visibility mark a significant advancement in digital infrastructure for carbon management.