Technologies for a U.S. manufacturing renaissance: A Q&A with Jeff Abell

Expanding the U.S. manufacturing sector has been a national priority since pandemic-era shortages exposed how fragile global supply chains can be. Major federal investment during the Biden administration and newer policies under President Trump aim to bolster domestic production and spur a U.S. manufacturing renaissance.

Academic institutions have an important role to play in these efforts, and they’ve always been critical contributors to the nation’s manufacturing leadership. A century ago, for example, research from University of Michigan Engineering helped lay the scientific foundations of modern manufacturing—from advancing machining processes like turning and milling to pioneering research on metal machinability. 

Today, Michigan Engineering is continuing to build on those foundational efforts, with a re-imagined manufacturing lab, the launch of the U-M Advanced Manufacturing Institute (UMAMI), and a series of faculty hires that started with Jeff Abell, former chief scientist of manufacturing at GM Global R&D, now serving as a professor of practice in mechanical engineering and co-director of UMAMI. 

In this Q&A, Abell offers an industry veteran’s perspective on the future of U.S. manufacturing, the technologies he believes are underleveraged, and how universities like U-M can bridge those gaps. 

After 40 years in the manufacturing industry, what brought you to Michigan Engineering?

It’s an exciting time to be in manufacturing with so many opportunities aligning both nationally and regionally. The federal government is recognizing manufacturing again—not only as an economic driver, but as a national strategic imperative. U-M is in a great position to lead in many areas, and I see tremendous opportunity here. 

Manufacturing, by its very nature, is cross-disciplinary, and that goes beyond just the technical aspects. U-M has excellent cross-disciplinary capabilities, with so many faculty, researchers, students and labs focused on addressing challenges across a spectrum of areas. We have deep expertise in engineering—including in materials development, materials discovery and robotics—and in other fields like business and medicine. There’s also huge medical manufacturing potential, not only regionally but nationally. 

There’s strength here in additive manufacturing, or 3D printing, which lets you make things no other process can. As a technical community, we’re just scratching the surface of designing products in ways that take advantage of additive and unlock performance that other processes can’t achieve. Michigan also has deep resources in advanced computing—AI, quantum computing and advanced data analytics. 

These are all things that can, and should, be applied in manufacturing, not only to design new processes, but to transform manufacturing itself. 

Additive manufacturing, or 3D printing, has been around for decades. Why are we still just scratching the surface of its potential?

Historically, additive manufacturing has been viewed purely from the manufacturing side as the ability to “make the unmakeable.” To unlock its potential and make it part of a U.S. manufacturing renaissance, we need to rethink the way we design in order to take advantage of 3D printing. One of the reasons it’s only taken hold in a few industries is because it’s difficult to scale. 3D printing machines are generally standalone stations—not integrated into the larger system—so it’s hard to make anything that scales easily. The aerospace industry uses additive manufacturing because you can’t make parts any other way. Scaling has been an issue in automotive, for example, and the business case hasn’t always been clear. But then, neither has the opportunity. I’ll give you an example.

At GM, 3D printing had long been used for prototyping. Then metal additive machines came out and people wondered if we could print production parts. But it was challenging to find the right opportunity to make it feasible. One example that surprised us was a supercharger vane. A supercharger injects air into the engine’s combustion chamber to boost performance. It has a screw-shaped rotor, or vane, that rotates at an incredibly high RPM to compress the air before it enters the engine. Normally, they’re machined out of steel. We used 3D printing to make a lightweight supercharger vane to reduce energy loss and increase the power. At first, we saw a nominal increase in power from the drop in vane mass, but then we also realized we could rotate the vane at higher speeds because it was so much lighter. The higher speed gave us a much higher output. That wasn’t what we initially expected.

The example shows that in order to get the most advantage from 3D printing, you have to really consider what design features will unlock game-changing performance and which can’t be made with any other process. It makes 3D printing genuinely compelling.

The University of Michigan has a strong design curriculum, and I think we can play a big part in demonstrating the value of this way of thinking. We can also leverage AI to help take sets of design requirements and envision new designs that take advantage of additive manufacturing.

You mentioned a U.S. manufacturing renaissance. How could universities like U-M help support it?

Universities play a number of important roles: Education, invention, transferring those technologies out into the world to benefit society, for example. That transfer process has a lot of room to improve, and we’ll be working on that through our new institute. On the technology side, I’ve already talked about some of the areas Michigan Engineering can lead in, like design-for-manufacturability, additive manufacturing and what I am calling “AI-native” manufacturing—where systems are designed from the outset with AI in mind, rather that adding AI in later on.

Another big opportunity is explainable AI—getting inside black-box models to understand how they make predictions or decisions. That’s key for advancing manufacturing science. Take machining, a process that we have used, studied, and largely understood since the late 1800s. You wouldn’t generally use AI to explain machining because the physics are well established, except possibly in very specialized cases. But ultrasonic welding? We know it works, but we can’t write down the differential equations for the process. Yet we have black-box models that can predict good versus bad welds. If we interrogate those models and extract the relationships between inputs and outputs, that becomes explainable AI—and it could lead to deeper scientific knowledge, letting us design entirely new processes or equipment. That’s the kind of contribution Michigan is positioned to make.

You’ve talked about quantum computing and “cognitive manufacturing.” How do they apply? 

Quantum computing has the potential to address challenges that classical computing alone simply can’t ever approach. Imagine the control system for an automotive assembly plant. These facilities have about 1,000 programmable logic controllers (PLC) in them. If these PLC’s were simple on/off switches (they’re far more complex) controlling the plant, you’d still be talking about 2 to the 1,000th power possible combinations—far larger than the number of stars in the universe. You can’t comprehend that or optimize that with classical computing. Quantum computing will eventually let us navigate and optimize those massive operating landscapes—controls, scheduling, logistics, processes and many others. 

People talk about quantum computing for material discovery. I want to use it for manufacturing, where we simultaneously discover materials and optimize processes. Paired with explainable AI, that can be genuinely transformative. Quantum isn’t here yet, but the hardware is coming, and we have the Quantum Research Institute bringing these initiatives together to put U-M ahead of the wave.

Cognitive manufacturing is the next frontier of AI and manufacturing. It’s the ability of an AI system to understand its situation and adapt to a changing environment, to improve, to dissolve disruptions and to fix problems on its own. Today’s AI is often fragile—great when the situation fits its training data, but struggling when something unexpected happens. 

The idea of cognitive AI is two-fold. First is human-AI learning—not replacement of humans but a real partnership between them and the AI. The other is adaptability—the ability to respond to genuinely new situations. Some might call this artificial general intelligence, but I don’t like to go that far because I think AI models and systems will be purpose-built. Manufacturing can be defined and constrained so that cognitive systems might actually be achievable within a decade.

Through the faculty expansion program, we’re now in the process of hiring three new faculty in the cognitive manufacturing space, spanning engineering and the supply chains through the Ross School of Business, and we hope to create collaborative research pathways in cognitive science through the College of Literature, Science and the Arts.

---

Related Stories