by Peter Page

Photo by Frank Wojciechowski Professor Christodoulos Floudas and doctoral student Xiaoxia Lin were part of the team that developed a novel optimization approach and a graphic interface for optomizing plant scheduling.

Pure theory from Princeton is making a big change on the factory floor in distant Birdsboro, Pa., thanks to three years of work by chemical engineering professor Christodoulos Floudas and his Computer-Aided Systems Laboratory.

"It is a pleasure to see these theoretical ideas became algorithms that became graphic interfaces that now have been extensively beta-tested,'' he said.

In 1998 Professor Floudas was approached by Nikola Juhasz, a senior manager of ATOFINA North America's Birdsboro chemical plant. He was interested in Professor Floudas' work on the theory of scheduling processes.

Professor Floudas, in turn, was quickly intrigued by the opportunity to apply his theoretical work to the complicated scheduling problems of a facility straining to maximize production.

"It is a complex operation because of the number of units, the number of products, the recipes for the products, and the time horizon,'' he said. "This gave us the opportunity to work with actual industrial case studies that have many subtleties that we never see in academic case studies. That was of tremendous value.''

The plant produces Rilsan® Nylon 11 and 12 thermoplastic resins. That facility, and its sister plant in Serquigny, France, are the only producers of Nylon 11 in the world.

Professor Floudas noted that the plant is "sold out,'' meaning its entire output is purchased as quickly as it is produced, creating a compelling incentive to increase production. He assembled a team comprised of Xiaoxia Lin, a doctoral student; Marianthi Ierapatritou, a former postdoctoral student now on the faculty at Rutgers University; and Manos Chatzakis, a postdoctoral associate now a researcher at Merrill Lynch. They worked with Dr. Juhasz and ATOFINA engineers.

Data collection was challenging because dozens of operations occur simultaneously or in a closely choreographed sequence. Much of the needed information was not in files but in the experience of longtime plant employees, he said.

"The key to success is working with industrial partners who will put in the effort, not just provide funding,'' he said. The hard work has paid off handsomely for ATOFINA. Plant managers now have, based on the work of Professor Floudas' group, Windows software with a graphical interface to optimize plant scheduling. The result is a significant increase in production, achieved without any capital investment.

"The benefits of this theoretical and algorithmic work are very significant and that, of course, improves the bottom line,'' Professor Floudas said.

The bottom line for SEAS students will come as Professor Floudas incorporates the ATOFINA work into his teaching. He plans to bring the insights of this work to his undergraduate course, CHE 442: Design, Synthesis, and Optimization of Chemical Processes.