July 2006, Issue 6

Creating Mathematical Models of Chemical Processes

In 2004 Christos Georgakis, PhD, came to Tufts from Polytechnic University of New York to chair the Department of Chemical and Biological Engineering. His research centers on complex chemical reactions that occur in various systems, from a pharmaceutical production plant to the chemical plant that is the human body. He creates mathematical models of the chemical processes in these systems. In collaboration with other researchers, he uses the models to develop computer programs and devices that improve the operation of chemical and biomedical processes and advance our understanding of metabolic pathways.

Since earning his PhD from the University of Minnesota, Georgakis has held positions at Massachusetts Institute of Technology, the University of Thessaloniki (Greece), Lehigh University, Rhône-Poulenc (France), and Delft University of Technology (the Netherlands). Georgakis has initiated research institutes throughout his career, including the Chemical Process Engineering Institute while at the University of Thessaloniki and the Chemical Process Modeling and Control Research Center while at Lehigh University. Soon after coming to Tufts, Georgakis founded and became director of the Tufts University Systems Research Institute (SRI) for Chemical and Biological Processes, which focuses on fostering research initiatives, cultivating interdisciplinary academic collaborations, and developing strong industrial partnerships.

Georgakis works in three broad research areas: (1) improving the development, design, and operation of batch pharmaceutical and chemical processes; (2) modeling and optimizing cellular systems for the production of biopharmaceuticals; and (3) understanding and modeling the chemical processes of physiological systems.
 
“The first area has to do primarily with how to design and operate pharmaceutical processes so that you can, if not reduce the cost of production, at least ensure repeatability, reliability, and strict quality control of the product,” says Georgakis. Pharmaceutical companies spend many years, and often millions of dollars, bringing a new product to market. “At a given time, a company might have ten molecules that they’re pursuing,” says Georgakis. “Not all will make it to the marketplace.” The problem that Georgakis sees is that in order to get successful products to market as soon as possible after clinical trials are finished, the company must begin designing production plants while the trials are still going on. Plant designs for unsuccessful molecules are a wasted effort. A second problem is that plant designs for successful molecules often resemble expanded arrays of chemists’ test tubes, usually not an efficient process for mass production.

Georgakis envisions a better way – designing a plant more precisely and more quickly by using mathematical models that incorporate an understanding of the chemistry and a quantification of the process operations. “I think there needs to be more sophisticated modeling and design tools to enable optimal design in a shorter time,” says Georgakis. “You need to reduce the time required to design and build the plant. Then you can start later and you won’t need to design plants for molecules that are not successful. That’s the key [financial] impact.”

The mathematical models are complex: they must take into account the stoichiometries (ratios of molecules) in reactions, the interaction effects when several steps are performed in the same reactor, and the purity of the yield, which decreases as the process is scaled up. The models also must incorporate coordination of feedback control, which may be simple if only one measurement is involved, such as from a thermostat, but very complex in a chemical plant where feedback is coming from 5,000 measurements.

Georgakis’s group tests some of their models in a lab in the Science and Technology Center on the Medford/Somerville campus. A new robotic system with ten small reactors is being used to develop online infrared techniques to measure composition versus time in various components, in order to figure out optimal reaction stoichiometries and kinetics. “We don’t want to focus on a specific pharmaceutical product,” says Georgakis. “We want to develop the modeling methodology. How many experiments do you do? How do you discover all the possible stoichiometries? How do you iterate from the wrong model to the improved model?”

The research group is currently working on organic synthesis of small molecules. Their plans include modeling production for biopharmaceuticals, which are large molecules produced by cells. This cellular engineering is Georgakis’s second area of focus. To model and improve the use of cells as chemical plants, he plans to collaborate with colleagues who have expertise in metabolic, molecular, and cellular engineering.

The third area Georgakis plans to explore is systems physiology, by which he means looking at the human body as a chemical plant and trying to understand and model its chemical and biological processes. One project Georgakis has in mind is to create a mathematical model of the information contained in a patient’s medical file, such as age, sex, weight, condition of arteries, and so on. Such a model could be used to personalize and improve patient care. It could also be used in conjunction with an anesthesia model to provide personalized on-line monitoring and control of anesthesia. Georgakis is discussing these areas with researchers in the School of Medicine’s Department of Pharmacology and Experimental Therapeutics.

“We want to look at certain things as a system – not just the chemical side, or the heat-transfer side, or the mixing side,” says Georgakis. “We want to look at how the overall system behaves and how its components impact the behavior of the system – whether it is a population of cells in a bioreactor, or it is a human body, or the different production units in a big pharmaceutical plant. The idea of the Systems Research Institute is to build the seed from which other things can grow.”

For more information, please see http://ase.tufts.edu/chemical/facGeorgakis.htm.