Engineering the future of manufacturing
Photo by Evan Krape February 07, 2022
COE’s Marianthi Ierapetritou leads $3 million National Science Foundation effort
Solving the climate crisis isn’t just about everyone driving electric vehicles and installing solar panels on our homes. It’s about redesigning the way we live, including sweeping changes to the way we produce everyday products.
As researchers race to find the latest and greatest technologies the world needs to be more resilient and sustainable, one team of educators at the University of Delaware is aiming to create a blueprint for a more renewable manufacturing future with a $3 million grant from the National Science Foundation.
“It’s important to educate the new generation of engineers to try to change the mentality of how we’re utilizing the limited resources we have,” said Marianthi Ierapetritou, UD’s Bob and Jane Gore Centennial Chair of Chemical and Biomolecular Engineering. She will lead the project as she works with Department of Chemical and Biomolecular Engineering Professors Dionisios Vlachos and Raul Lobo, Department of Electrical and Computer Engineering Associate Professor Hui Fang and Joseph R. Biden, Jr. School of Public Policy and Administration Assistant Professor Kalim Shah to launch the future manufacturing project in 2022.
The goal is to thoroughly examine existing literature around renewable products and processes in manufacturing, which will help researchers synthesize existing data and identify gaps in knowledge. From there, researchers can develop a framework for examining the potential economic, environmental and market impacts of alternative products and processes, while also evaluating the realistic probability of introducing new “green” solutions into existing supply chains and consumer markets.
“The big idea here is how to better utilize available information,” Ierapetritou said. “It’s a collaboration between chemical engineering, computer science and public policy.”
Several students at the undergraduate and graduate levels in both the College of Engineering and the Biden School will participate in some of the computational work, research and design while collaborating with real-world chemical companies and with the American Institute of Chemical Engineers’s Rapid Advancement in Process Intensification Deployment (RAPID) Institute as a manufacturing partner. The project’s funding is expected to span four years.
By using computers to mine for innovations in existing studies — a task that would take multiple graduate students months or years to complete — these researchers can extract the information needed to better understand what it will take to change the way we produce and consume products.
“We’re kind of a supporting team, while the chemical engineering team needs to use the information we are extracting,” Fang said. “It’s like we’re collecting all of the available recipes so we can enable the chef to create some new dishes.”
Since many of the products used every day are created from petrochemicals (fossil fuels), researchers are looking for ways to create more renewable products that require less energy and produce less waste. A consensus of scientists around the world say that greenhouse gas emissions must reach zero globally to avoid a level of global warming expected to result in more catastrophic climate disasters than the deadly floods, fires and storms seen in recent years worldwide.
But finding renewable and realistic replacements to the way societies manufacture products means getting innovative at the molecular level, explained Vlachos, Unidel Dan Rich Chair in Energy Professor of Chemical and Biomolecular Engineering, director of the Catalysis Center for Energy Innovation and director of the Delaware Energy Institute.
That task can be tackled much more efficiently when computer intelligence gets involved. Instead of using expensive laboratory equipment, chemicals, molecules and catalysts, researchers can use chemistry-informed and data science-informed computer programs to point them in the right direction.
“You don’t want to build a $20 billion plant and then it fails. That would be a disaster,” Vlachos said. “I want my computer programs to make better predictions. So, how do you build the new chemical route to go from here to there? We need the computer to tell us.”
That also means teaching computers chemistry, which means the models can only be as good as they’re trained to be. Still, these programs will be able to process information exponentially faster than a group of human researchers engaged in trial and error experiments, while also eliminating the need for physical resources.
For example, computer programs could extract everything from existing scientific literature about how a molecule like the sugar glucose reacts. With that information, the model could then explore how different combinations of different molecules act, and what new outcomes varying combinations might have, like how combinations of sugar and salt might impact a cake mix.
“We’re transforming this whole process and using computer science and computer simulations to understand options and give you the best alternative without even going into the lab,” said Ierapetritou. “That’s why it’s called ‘future manufacturing.’ It doesn’t address the current needs of manufacturing, but rather the future needs and where we’d like to go.”
Beyond searching for the ideal molecules and processes to build more sustainable products, the project will also look at the feasibility of producing those items. If, for example, the raw materials to create something are only available seasonally at certain locations, like useable waste from corn after the fall harvest, would it be more efficient and cost-effective to have modular manufacturing units that can be relocated at certain times of the year instead of building one huge manufacturing plant that would require additional transportation of raw items?
“I think it might revolutionize the way we’re thinking about supply chain,” Ierapetritou said. “Especially now that we’re all paying the price of not optimizing supply chains.”
This same interdisciplinary team also is working on a similar project funded by NSF and in collaboration with the University of Kansas and Pittsburg State University to find sustainable replacements for plastics.
To possibly work, these solutions also need to be based in reality — meaning those options also need to consider the logistics and costs of production and processing, real-world markets, consumer attitudes and potential environmental impacts.
“The second piece is the market piece,” said Shah, assistant professor with the Biden School. “How do we market this green or eco-friendly approach to industry?”
The modeling Shah will work with through this project — called “agent-based modeling” — will allow researchers to simulate real-world circumstances to explore whether a certain product would work at a scaled-up level, he explained. But human behavior isn’t exactly easy to model.
“We’re not going to assume we know how different kinds of actors are going to act,” Shah said. “We’re going to do behavioral surveys of people, communities, businesses, and use the behavioral and physical principles and ideas to try to translate what we get from the surveys into rules that we can program.”
This project focuses on optimizing future processes that will be needed to develop new products that could offer climate-related benefits, either from the way that foundational materials are harvested, how those base chemicals are processed to how the products are actually produced. That includes the supply chain processes from start to finish, as well as the role that marketing a new “green” solution will play.
By using agent-based modeling, researchers can simulate how a product and its related processing would fit into particular sectors, or where their proposed idea might hit unexpected roadblocks.
“Think of it like whatever goes into the programming of the Sims game,” Shah said, noting that their model outputs will be a collection of numbers, diagrams and statistics, not nearly as aesthetic as a multi-million dollar virtual game.
As scientists try to communicate a dire need for swift changes to address the climate crisis, simulations like the one Shah plans to develop for this project could be applied to other projects, as well. The computational and modeling work led by Fang could also be applied to other similar projects.
“There are models at multiple scales and multiple levels, and eventually we want to bring that all together,” Vlachos said. “Then we need to bring in society and decision-making, not just the science itself, and see where we go.”
“A science-based, fact-based, data-based model can help the country move in the right direction with the right science. Now we need to deliver. We will deliver.”