Yokogawa and Saudi Aramco Achieve Breakthrough with Autonomous AI Control at Fadhili Gas Plant

A Milestone in Energy-Sector Automation

Yokogawa Electric Corporation (Tokyo: 6841), a well-known leader in process automation, has partnered with Saudi Aramco to achieve a major milestone: deploying multiple autonomous control AI agents at the Fadhili Gas Plant in Saudi Arabia. 
This deployment represents a strategic shift toward “industrial autonomy” (IA2IA) rather than merely automation—moving from human-guided systems to self-optimising AI-powered operations.

The Technology: Autonomous Control AI and FKDPP Algorithm

Yokogawa’s AI solution is built on the “Factorial Kernel Dynamic Policy Programming” (FKDPP) reinforcement-learning algorithm, jointly developed with the Nara Institute of Science and Technology (NAIST). 
The system uses multiple coordinated AI agents that can independently determine optimal control strategies—even for conditions they had not encountered before (as defined by Yokogawa’s concept of “autonomous control AI”). 
Implementation followed a three-phase approach: first simulation for training, then validation, and finally integration with the existing control system (Yokogawa’s CENTUM VP) to connect with existing safety infrastructure. 

Deployment Focus: AGR (Acid Gas Removal) Unit at Fadhili

The target of this advanced deployment was the acid-gas removal (AGR) process unit within the Fadhili facility. The AI agents were tasked with optimising the process of removing acidic gases using amines and steam while managing energy consumption. 
Because AGR is a high-complexity, high-cost segment of the plant’s operations, improvements here translate to meaningful cost and emissions savings.

Early Results: Efficiency Gains and Stability Improvement

Although the deployment is still under detailed evaluation, initial operational data from the Fadhili plant indicate the following improvements:

• A reduction of 10–15% in amine chemical and steam usage. 

• A reduction of approximately 5% in power consumption.

• Enhanced process stability and significantly reduced need for manual operator intervention despite changing ambient conditions. 
  These metrics suggest that the autonomous AI agents are not only optimising for efficiency, but maintaining operational resilience.

Strategic Implications for the Energy Industry

This deployment underscores two major strategic themes:

1. Operational Efficiency & Sustainability – By reducing chemical, steam and power consumption, plants lower operational costs and environmental footprint, aligning with energy-sector decarbonisation goals.

2. Shift Toward Autonomy – The transition from automation (pre-programmed control logic) to autonomy (AI agents making and executing control decisions) signals a new phase in industrial operations. Yokogawa explicitly frames this as moving from “industrial automation to industrial autonomy (IA2IA)”.
   For energy sector players, this means future competitiveness will depend not just on hardware or control systems, but on AI-enabled operational platforms and ecosystem partnerships.

What This Means for Control Systems and AI Integration

For engineers and automation professionals, several practical takeaways emerge:

• Simulation-First Approach: Training AI agents using plant simulation prior to live deployment helps ensure safety and reliability in critical processes.

• Seamless Integration: The AI agents were integrated with the existing control architecture (CENTUM VP) rather than replacing it, allowing operators to retain oversight and safety layers.

• Performance Monitoring: Tracking real-world KPIs (chemical consumption, energy use, manual interventions) is key to validating AI deployments in operational settings.

• Scalability & Coordination: Using multiple AI agents in coordination (rather than a single algorithm) enables optimisation across interlinked process segments.
  This deployment thus offers a blueprint for how AI can move from pilot trials into full-scale plant operations.

Looking Ahead: Toward the Autonomy Era in Industrial Energy

This milestone at the Fadhili Gas Plant is likely just the beginning. For companies in energy and heavy process industries:

• AI agents may expand into other high-value units (e.g., distillation, hydrogen production, upstream processing).

• Ecosystems of software, controllers, sensors and algorithms will become the battleground—hardware alone won’t suffice.

• Operational culture must evolve: from “control the plant” toward “guide the autonomous system”.

• Data infrastructure, model governance, safety and explainability of AI decisions will become central to deployment success.
  In short, the energy industry is shifting from automation to intelligence—and companies that embrace this shift early are likely to gain strategic advantage.