Boosting Industrial Autonomy with AI

Replacing human labor with machines has in general brought about tremendous gains in wealth and health for the world. It has given us all sorts of labor-saving, distance-shrinking, communication-enabling, and knowledge-expanding innovations which define modern existence. But it has also been the source of painful transformations of the status quo while generating many of the gravest challenges for human beings.

From the terrible working conditions and local pollution of the earliest industrial factories in England, once famously described as “dark satanic mills” by the poet William Blake in 1804, to broader structural challenges of urbanization, population growth, social inequality, and environmental degradation^*1, the relentless march of industrialization at global scale has generated new problems for each generation. Thankfully, many of these are being addressed by government action such as expanding labor and environmental regulations along with improved efficiency, safety, and sustainability of industrial processes.^*2

The Janus-faced evolution of our relationship with machines continues and is now entering a new phase.

After the revolutions of steam-powered mechanization, electric-powered mass production, and electronics and IT-driven automation, some argue we are now in a fourth industrial revolution of technologies which fuse physical, digital and biological spheres.^*3 Moreover, these changes in our economic production are generating diverse risks, not only to humanity in general, but also for industry itself.

In 2023, the World Economic Forum, in collaboration with the University of Cambridge’s Industrial Innovation Policy Group, and the United Nations Industrial Development Organization, identified five grand challenges which industries must address into the future.^*4

First and foremost are the existential challenges for the planet: “environmental degradation, resource shortages, and the need for a green transition.” Resource use and carbon emissions must be tracked and minimized across the production chain. ^*5

While achieving this vast task, industries must ready themselves for supply chain disruptions such as climate disasters, pandemics, geopolitical tensions, and cyberattacks. Third, adopting and scaling new technologies such as AI will entail building new frameworks to deal with governance, interoperability and security challenges. Cybersecurity challenges must be met, even as digital transformation needs to be accelerated. ^*6

Fourth, skills gaps, labor shortages, and reskilling needs in the workforce need to be met. Industries must also pass on the know-how of seasoned plant operators and veteran workers in the face of the “great crew change,” i.e. challenges of a shrinking skilled workforce due to retirement and lack of new workers.^*7 Increasingly complex, round-the-clock plant operations also threaten to overburden workers.

And finally, industries as businesses must balance bottom-line goals while promoting “social interests and equitable growth.”^*8 Socially just and responsible capitalism must account for the needs of all relevant parties, not just shareholders.

One path through this thicket of thorny challenges is to accelerate AI and automation in industrial processes. Ultimately plant assets and operations will achieve greater levels of autonomy, being able to learn, adapt, and respond with minimal human interaction with operators focusing on higher-level tasks of optimization and creativity.

The white paper “Industries in the Intelligent Age” argues that AI and automation in industry will help meet many of these issues, greatly lifting productivity and competitiveness, thereby improving efficiencies and minimizing resource use and waste, while freeing personnel to focus on creativity, oversight and decision-making.^*9

In the report’s survey, 89% of companies see AI as highly relevant to industrial operations and plan to adopt it into their production networks. In fact, AI is already boosting production efficiency, with early adopters reporting an average 14% savings on manufacturing costs, according to the report. Although so far actual AI adoption remains low in the industrial sector, the potential gains are particularly significant for industrial processing plants, argues McKinsey and Company.

“Operators of industrial processing plants are particularly well positioned to capture the benefits of AI. Many already rely heavily on data-driven decision-making using processing data,” states a report by McKinsey’s Global Energy & Materials Practice. “And most plants today have made significant investments in enablers of AI, such as network design, control systems, and historical data capture.”^*10

To accelerate and adopt AI for industrial automation, Yokogawa is pioneering important solutions.

In 1975, Yokogawa Electric released CENTUM, the world's first distributed control system. Last year, the company launched the latest version of CENTUM. The CENTUM VP Release 7 is positioned to accelerate the benefits of automation and AI for industry. It features expanded scope of control and monitoring, predictive monitoring and ability to reduce operator workload. It also has greater cybersecurity features and would be able to deliver AI-driven plant operations which leverage the knowhow of plant operators.

“In this era of greater volatility, uncertainty, complexity and ambiguity, the conditions required for manufacturing sites and management are becoming ever more complex,” says Mitsuhiro Yamamoto, Vice President of Yokogawa Electric. “By realizing stable operations and expanding the scope of autonomy, CENTUM VP Release 7 will help our customers achieve a more sustainable society and sustainable business growth.”

Yokogawa has an impressive track-record of evolving its groundbreaking control and automation technologies over the years to meet changing industry needs and broader technological trends: the beginnings of computerization and automation in the 70s; the rapid adoption of PCs, the internet, and digital technologies in the 80s and 90s; the rise of mobile and wireless technologies in the noughties; and now in the past decade, big data, cloud computing, and IoT solutions.

Throughout CENTUM has gained new features and functions to better integrate these trends and advances in technology so its customers can achieve greater “reliability, stability and continuity” in its industrial operations. Now into its 10th generation, some 30,000 CENTUM systems have been deployed in more than 100 countries around the world, making it a market leader for DCSs. Such systems are generally used in the oil and gas, petrochemicals, chemicals, steel, pulp and paper, electric power, and water treatment industries.

Looking forward, the company identifies two significant trends which its DCS will help navigate: advances in AI and machine learning which power predictive maintenance, quality control, and automated optimization, and the progress in using digital twins for simulation and optimization. All of these features will further the ultimate goal of “autonomous operations.”

Besides developing the AI-enabling DCS for plants, Yokogawa has already demonstrated in different domains how industrial AI can deliver.

In 2022, together with the former elastomers business unit of JSR Corporation (now part of ENEOS Materials Corporation), Yokogawa deployed the reinforcement learning-based Al algorithm called Factorial Kernel Dynamic Policy Programming (FKDPP), which was jointly developed by Yokogawa and the Nara Institute of Science and Technology (NAIST) in 2018, and was recognized at an IEEE International Conference on Automation Science and Engineering as being the first reinforcement learning-based AI in the world that can be utilized in plant management. In JSR's chemical plant, which produces a raw material used in synthetic rubber, Yokogawa achieved a world's first by integrating the AI with CENTUM to autonomously and safely control a plant for 35 days, stabilizing quality, achieving high yields, and saving energy.

Yokogawa's Al solution continued to control the operations at the industrial facility and after a year-long trial this reinforcement-learning-based Al was officially adopted, another world’s first. FKDPP is used to automate valve operations and has brought about decreased operator workload, eliminated the production of off-spec products, and reduced steam usage and CO2 emissions by 40%, thereby saving energy and reducing costs.

Building on this success, in 2025 Yokogawa deployed multiple, coordinated autonomous control AI agents of FKDPP at the Fadhili Gas Plant operated by Aramco, one of the world’s leading integrated energy and chemicals companies. The AI solution directly and autonomously controls and optimizes acid gas removal (AGR) operations at the plant to reduce energy and chemical usage.

Some two centuries since the first modern industrial factory began operations in England^*11 and 50 years since the first distributed control system was introduced, Yokogawa is working towards a future where industrial plants perform beyond human limits while further unlocking human potential.

---

References

*1 Britannica: https://www.britannica.com/story/the-rise-of-the-machines-pros-and-cons-of-the-industrial-revolution 
*2 Britannica: https://www.britannica.com/topic/industrialization 
*3 World Economic Forum: https://www.weforum.org/stories/2016/01/the-fourth-industrial-revolution-what-it-means-and-how-to-respond/ 
*4 World Economic Forum: https://www3.weforum.org/docs/WEF_The_Future_of_Industrial_Strategies_2023.pdf 
*5 World Economic Forum: https://www.weforum.org/publications/net-zero-industry-tracker-2024/ 
*6 World Economic Forum: https://initiatives.weforum.org/cyberresilienceindustries/activities 
*7 451 Alliance: https://blog.451alliance.com/smart-technologies-respond-to-industrys-great-crew-change/ 
*8 World Economic Forum: https://www3.weforum.org/docs/WEF_The_Future_of_Industrial_Strategies_2023.pdf 
*9 World Economic Forum: https://www.weforum.org/stories/2025/01/why-manufacturers-should-embrace-next-frontier-ai-agents/ 
*10 McKinsey & Company: https://www.mckinsey.com/industries/metals-and-mining/our-insights/ai-the-next-frontier-of-performance-in-industrial-processing-plants?cid=soc-web#/ 
*11 Historic England: https://historicengland.org.uk/whats-new/research/back-issues/when-soho-led-the-world/