Five key takeaways from the 2024 AFPM Annual Meeting that drive home the value of AI/ML applications in advanced process control.
By Allison Buenemann, Product Marketing Manager at Imubit
At this year’s AFPM Annual Meeting, Motiva Director of Technical Excellence Tom Chunat facilitated a panel session focused on the value that leveraging artificial intelligence and machine learning can bring to advanced process control. The panel included representatives from ExxonMobil, McKinsey & Company, Monroe Energy, and Imubit. Imubit CEO Gil Cohen and Monroe Energy COO Regan Howell joined forces to share their perspectives as a closed loop AI provider and an operating company that is realizing the value. Here were the main takeaways:
1. Technology driven efficiencies level the playing field for all company sizes.
You’ve probably heard of the singles tax as it pertains to the economic costs associated with living and traveling alone compared with a party of two. But have you heard of the single refinery tax? Single refineries, or more broadly the class of non-global multinational energy companies, miss out on the economies of scale advantage that comes from centralizing business functions like finance, G&A, R&D, and engineering. These higher operating costs can eat into profit margins.
Monroe Energy is a single-site refinery playing in a hydrocarbon space dominated by a small number of major players. So how do they stay competitive while maintaining margins? “One of the easiest ways to do that is to take advantage of new technology,” said Monroe COO Regan Howell. New uses of AI, particularly in closed loop, enable optimal, data-driven decision making to be done in real-time, without requiring operator action.
2. All you need to implement a closed loop data-driven model is… the data.
It sounds simple, but one of the biggest impediments to embarking on a process optimization project is the prerequisites. Whether it’s a robust process model or a lack of process control expertise that’s holding you back from achieving optimal operations, there will always be inertia to overcome.
Howell spoke of the benefits of using true AI models as a means to overcome this inertia by taking a data-first rather than first principles approach. “These models are created from OUR data. It’s not a model where we have to consider every potential variable programmed into some sort of matrix algebra.” The lack of matrix algebra also makes model building and interpretation accessible to all of their domain experts at the refinery, not just controls specialists.
3. Adoption comes by way of reducing pain.
Monroe acknowledged early on that gaining buy-in from different stakeholders would take different approaches. They knew that the board operators’ goal was to come in and have 11 hours and 45 minutes of relatively unexciting operation. That meant that they were passing a stable unit off to their relief, and hoping to have the favor returned in 12 hours. “Implementing these types of solutions eases their pain,” said Howell, by knowing the appropriate process handles are being turned in response to changing plant dynamics.
The board and executives at Monroe have different pain. They need compelling evidence that their investments are adding value to the business. To demonstrate value to these stakeholders, the Imubit platform provides a performance dashboard application that shows when control was engaged versus disengaged, and how the process performed during both conditions. “It’s very easy to point to a difference of achievement of your 90% target on your diesel streams in the time the model is engaged,” said Howell.
4. People have an inherent desire to learn and grow.
One of the recurring themes that’s cropped up on conference stages and in analyst reports the past few years is the use of new technologies to bring in new talent. “We can recruit people to come play with these kinds of [new technology] toys. It attracts and retains employees by helping them feel like they’re learning new things, like they’re progressing in their career,” said Howell. In fact, it’s not uncommon to see new technology adopters promoted in their organizations for this growth.
5. Choosing the right partners is always the key.
Or in the words of Imubit CEO, Gil Cohen, “the key is partnership that brings in domain expertise.” Changing the way you optimize and control your plant isn’t a one-time project where you can implement and leave. As plant conditions change, AI models should be monitored to detect drift the same way an instrument might be. Imubit’s Customer Success Managers are constantly monitoring Imubit models, identifying opportunities and taking proactive steps to improve the AI model.
As continuous model improvement leads to better optimization, measuring those incremental gains becomes more challenging due to the shifting baseline. The Imubit Economic Engineering team ensures that the business value added by the software is continuously measured and reported to relevant stakeholders. These are just a sample of the ways that the right technology partner can operate as an extension of the plant team.
Monroe Energy COO Regan Howell responds to a question from panel facilitator Tom Chunat. The panel also featured speakers from ExxonMobil and McKinsey & Company.
Connect with us
With a full event line-up this spring there is plenty of opportunity to connect with the Imubit team in your region! If you want to dive deeper, request a demo today.