Digital Exclusive: The biggest obstacle to the refinery of the future is a mindset shift
JOACHIM BOESE, AVEVA, Frankfurt, Germany
Autonomous operations in the oil and gas industry are no longer a figment of the imagination. Companies now have all the tools—and the business case—to take the industry into the future.1
So why, across the global refining industry, do the vast majority of plants still operate much the same as a decade ago? Barring a few exemplary cases, refineries continue to rely heavily on manual interventions, reactive maintenance and periodic optimization reviews, leaving enormous value trapped inside their own processes.
For operations leaders, these facts hide an uncomfortable truth: the biggest barrier to autonomous refining is not technology, but mindset.
Autonomous operations: The key to relieving pressure on oil and gas. As anyone in the industry knows, refineries face pressure from multiple directions. On one side, the business is squeezed by thin and cyclical margins, crude price volatility, shifting product demand and relentless competition. On the other, companies face a growing sustainability imperative: carbon pricing and emissions regulations, but also investors and customers demanding progress on decarbonization.
What makes autonomous operations so strategically compelling is that they address both pressures at once. Energy is both the largest single operating cost in refinery operations and the primary source of the industry’s direct carbon emissions. Every unit of energy saved is therefore simultaneously a margin improvement and an emissions reduction.
The most powerful tool to optimize energy use at modern refineries is autonomous operations (FIG. 1), with the potential to increase efficiency by at least 11%.2
FIG. 1. The broader the scope of autonomous control, the more critical it becomes to handle transient behavior reliably and safely across multiple interconnected units at the same time.
How autonomous refining is evolving. True autonomous operations go far beyond conventional process control and require the ability to look ahead and address challenges proactively. These capabilities are now enabled by the convergence of predictive analytics, process simulation and real-time optimization.
One of the most demanding tests is managing transient processes outside of steady-state operations - think startups, shutdowns or feedstock switches. The challenge becomes even more acute as refineries move toward higher levels of autonomous operations maturity. The broader the scope of autonomous control, the more critical it becomes to handle transient behavior reliably and safely across multiple interconnected units at the same time.
Evidence for the benefits of autonomous operations is mounting among early adopters. For example, ADNOC, one of the world's largest energy companies, has deployed Neuron 5, an advanced artificial intelligence (AI)-powered autonomous operations platform co-developed with the author’s company and AIQ, across its upstream and downstream facilities.
Initially tested at a crude field and gas compression plant, Neuron 5 not only autonomously monitors thousands of critical equipment assets (e.g., compressors, valves, generators), but also monitors and predicts process performance across ADNOC’s facilities.3 The results from the pilot phase are striking, promising to reduce unplanned shutdowns by 50% and extend planned maintenance intervals by 20%.
Separately, a new generation of technology is redefining what is possible: Deep reinforced learning (DRL) combines the predictive power of dynamic process simulation with the adaptive, experience-driven training of machine-learning.4 DRL systems learn optimal control strategies through continuous interaction with a simulated process environment, developing the ability to navigate complex and evolving operational conditions in ways that static models cannot.
The potential upside is impressive: early deployments demonstrate that DRL-driven control policies can stabilize industrial processes twice as fast as manual operators during large feed changes and process upsets—exactly the kind of transient conditions where traditional control falls short.
From operator to orchestrator: The mindset shift in autonomous refining. For these critical tools to find broader adoption, however, a fundamental mindset shift is required.
The refining industry has always attracted deeply skilled professionals who take pride in hard-won operational expertise. However, that same pride, combined with organizational inertia, is increasingly becoming the prime obstacle preventing refineries from adopting meaningful automation and achieving improvements across the two dimensions that define success in modern refining: margin performance and environmental responsibility.
Historically, each refinery shift had its own operational philosophy, a way of working that is upended by automatic, standardized processes. Operators and supervisors are also rightfully worried about taking responsibility for autonomous processes they do not understand, fearing they will bear the consequences if something goes wrong. Finally, the adoption of autonomous operations can appear to threaten their very livelihood—or, at the very least require a substantial rethink of their role.
Companies must be conscious of these concerns and seek to replace them with an alternative message: autonomous systems do not remove human control, but rather elevate it. After all, when AI monitors process control, or predictive systems anticipate future incidents, experienced operators are freed from reactive firefighting and can instead evolve to become empowered orchestrators.
Of course, this evolution relies on next-generation data capabilities since autonomous operations cannot function on fragmented information. Refinery orchestrators require supervisory platforms that provide a 360° operational view of a plant’s processes, assets, energy flows, cost drivers and emissions profile at any given moment.
Advanced analytics sit at the heart of this supervisory layer, extracting meaningful intelligence from the vast data streams modern refineries generate. Without this comprehensive picture, optimization remains partial—meaning margin opportunities are missed and environmental management remains reactive.
The cost of standing still on autonomous refinery operations. It is clear that the question is no longer whether autonomous operations work in refining, but whether operations leaders are ready to lead the transformation.
The refinery of the future is being built today by organizations that recognize autonomous operations as a genuine inflection point. As ADNOC and other industry leaders show, this is now a measurable operational reality delivering real value. Those who wait will find that the performance gap, both economically and environmentally, widens to a point where it becomes very difficult to close.
The good news is that succeeding in this transition only requires operations leaders to actively champion the change and effect a mindset shift among the skilled workers who will continue to be crucial to orchestrate and optimize performance.
These operations leaders must embed both economic and sustainability targets into their optimization frameworks, define governance that clearly delineates where autonomous control operates and where human judgment is required, and measure success by whether people and systems together deliver outcomes that neither could achieve alone.
The technology is here and the business case is clear. All the industry needs now is operations leadership with the vision to see the opportunity and the courage to seize it.
LITERATURE CITED
1 Rack, Y., “Paw patrol: How robots are changing oil and gas operations,” January 17, 2025, online: https://www.aveva.com/en/our-industrial-life/type/article/paw-patrol-how-robots-are-changing-oil-and-gas-operations/
2 DeVries, S., “Autonomous operations in refineries,” AVEVA, June 14, 2024, online: https://www.aveva.com/en/perspectives/blog/autonomous-operations-in-refineries/
3 “UAE’s ADNOC deploys Neuron 5 for AI-powered efficiency,” Gulf Business, August 27, 2024, online: https://gulfbusiness.com/en/2024/uae/adnoc-streamlines-operations-with-neuron-5/
4 Nvidia Developer, “Spotlight: Advancing autonomous operations with AVEVA dynamic simulation and NVIDIA Raptor,” November 21, 2024, online: https://developer.nvidia.com/blog/spotlight-advancing-autonomous-operations-with-aveva-dynamic-simulation-and-nvidia-raptor/
ABOUT THE AUTHOR
Joachim Boese is Industry Principal—Oil & Gas at AVEVA. He has worked for more than 35 yrs in the oil and gas industry with operational experience in international operating companies as well as consulting expertise with international solution suppliers. Today, Boese works with major international customers in the oil and gas space to identify and realize potential improvements in operations efficiency and sustainability. Before joining Invensys in 2007 and now as a part of AVEVA, Boese worked for ConocoPhillips in its Wilhelmshaven Refinery and in engineering, operations, and planning and scheduling for Deutsche Shell in its two German refineries in Hamburg and Cologne. Boese holds a degree as Diplom Ingenieur of Clausthal University in Germany.
Comments