Aramco and Yokogawa deploy AI control at Saudi gas plant

Yokogawa Electric Corporation and Aramco have achieved a landmark in industrial automation with the successful commissioning of multiple autonomous control AI agents at the Fadhili Gas Plant in Saudi Arabia. The deployment marks one of the world’s first implementations of reinforcement learning-based AI for direct, autonomous control of large-scale process operations.

Developed by Yokogawa, the AI system uses coordinated agents built on the Factorial Kernel Dynamic Policy Programming (FKDPP) algorithm to autonomously optimize the plant’s acid gas removal (AGR) process. The solution rolled out in stages. First, it was trained in a virtual plant simulator. Then, it was added to Yokogawa’s CENTUM VP control system. This ensured it worked well with current safety protocols.

Early results from the Fadhili facility show big efficiency gains. There are 10–15% cuts in amine and steam use. Power consumption is about 5% lower. Also, process stability has gotten better, resulting in fewer manual interventions. AI-driven automation can boost sustainable and self-improving industrial operations.

Also Read: Asuene acquires 2nd US Firm and closes ¥11.5B Series C2

Yokogawa and Aramco see this achievement as a key step for future autonomous plant operations. It aligns with worldwide energy goals for digital transformation and carbon efficiency.