AI and Advanced Technologies Pave the Way for Improved Plastic Recycling

AI-assisted sorting, other new technologies could improve plastic recycling

Just 9% of plastic worldwide is recycled. Due to waste mismanagement, nearly three-quarters of it ends up in landfills or the environment. So how can plastic recycling be more efficient? A review article by University at Buffalo researchers summarizes the latest technologies and methods guided by process systems engineering approaches, from chemical solvents that can dissolve specific plastics to automated plastic sorting aided by artificial intelligence. Selected as the cover article for the July 9 issue of Industrial & Engineering Chemistry Research, a journal of the American Chemical Society, the article concludes that solvent-based recycling is both a sustainable and economical option, but that replacing fossil-based plastics with bio-based plastics remains a challenge. "More research and technology development are necessary to attain sustainability in plastics management," says the study's corresponding author, Aurora del Carmen Munguía-López, Ph.D., assistant professor in the Department of Chemical and Biological Engineering, within the UB School of Engineering and Applied Sciences. "We not only need holistic and comprehensive approaches, but to consider the pros and cons of those approaches throughout their entire life cycle." The article's co-authors are postdoctoral researcher Xate Sanchez-Zarco and Ph.D. student Alan Owusu-Boateng. Plastics still irreplaceable Improper disposal of plastics can cause plastic waste to accumulate in both the environment -- it's estimated there's 150 million tons of it in the ocean -- and the human body. Exposure to plastics has been linked to cancer, respiratory problems, fertility issues and developmental delays. However, plastics are also a crucial part of modern life, being used in packaging, electronics, buildings and textiles. They have plenty of environmental benefits, too, from reducing food waste to increasing the fuel efficiency of vehicles. "Eliminating the use of plastics is not currently a viable option," Munguía-López says. "We need alternatives to the current unsustainable management of plastics." Solvents, AI could offer recycling solutions One of the alternatives may be solvent-based recycling, which can recycle complex materials that cannot be recycled by traditional means. Solvents can dissolve high-purity polymers within a plastic waste stream, thereby removing them from unwanted contaminants. The article highlights a recent University of Wisconsin-Madison-led study, co-authored by Munguía-López, that found that solvent-based recycling was the most economical option for recycling complex, multilayer plastic film used in coffee grounds packaging. While solvent-based recycling has relatively low greenhouse gas emissions, variations in the process can significantly increase emissions. Solvent-based recycling should use a cooling method to reform the dissolved polymers from the solution, various studies have shown, as opposed to a heating method that generates more emissions. "Either way, solvent-based recycling does produce higher emissions than traditional recycling, so the best approach is likely combining both solvent-based and traditional recycling," Munguía-López says. The article also summarized research about the role of AI and machine learning in plastics management. One sorting model developed by University of Wisconsin-Madison researchers called PlasticNet achieved a classification accuracy of over 87% and even 100% on some specific plastics. Other teams have used AI to better study recycling technologies, developing natural language processing models to extract relevant data from the literature. "AI models will also be needed to address demands at the supply chain level, like improving transportation planning, coordinating stakeholders, and evaluating different policy scenarios," Munguía-López says. Can we switch to bio-based plastics? The feasibility of bio-based plastics, which are made from agricultural crops like sugar cane and corn and could be disposed of through composting, is less clear. While bio-based plastics have lower emissions, they also require lots of water and land and directly compete with the food supply. Implementing bio-based plastics would also require more composting facilities and equipping the public with ways to separate them from traditional plastics. "We can't validate bio-based plastics until we consider the impact of their entire life cycle, from raw materials extraction and production to disposal and sorting," Munguía-López says. "Future work in plastics management should include systems-level analyses to address this multiscale and multidimension problem."

AI-assisted sorting, other new technologies could improve plastic recycling

BUFFALO, N.Y. -- Just 9% of plastic worldwide is recycled. Due to waste mismanagement, nearly three-quarters of it ends up in landfills or the environment. So how can plastic recycling be more efficient? A review article by University at Buffalo researchers summarizes the latest technologies and methods guided by process systems engineering approaches, from chemical solvents that can dissolve specific plastics to automated plastic sorting aided by artificial intelligence. Selected as the cover article for the July 9 issue of Industrial & Engineering Chemistry Research, a journal of the American Chemical Society (ACS), the article concludes that solvent-based recycling is both a sustainable and economical option, but that replacing fossil-based plastics with biobased plastics remains a challenge. "More research and technology development are necessary to attain sustainability in plastics management," says the study's corresponding author, Aurora del Carmen Munguía-López, PhD, assistant professor in the Department of Chemical and Biological Engineering, within the UB School of Engineering and Applied Sciences. "We not only need holistic and comprehensive approaches, but to consider the pros and cons of those approaches throughout their entire life cycle." The article's co-authors are postdoctoral researcher Xate Sanchez-Zarco and PhD student Alan Owusu-Boateng.