Improving Polymers Through Simulation
Wouldn't it be nice to know how a material would flow just by looking at its molecular structure?" asks Associate Professor of Mechanical Engineering A. Jeffrey Giacomin, director of the college's Rheology Research Center. Predicting rheological properties-the way materials flow and deform-is a cornerstone of the research programs for many of the center's 11 faculty members.
Partially through cooperation and partially through kismet, the work of some of the center's newest members dovetails to reveal the inner workings of rheological materials. Giacomin, for example, is developing ways to measure flow behavior of materials under extreme conditions. That work helps other researchers verify their simulations, providing powerful new tools for developing polymers. Ideally, he explains, an engineer who wanted to find the perfect polymer for a given product could select characteristics of different molecular combinations and would then be able to view on screen how the resulting plastic would behave for a given application and without the expense of producing the actual substance.
"That's what I'm trying to do: go all the way from atoms to the factory," says Juan J. de Pablo, an associate professor of Chemical and Biological Engineering. De Pablo is developing computer models that describe how individual polymer molecules, which are long chains of atoms, fit together and interact. As these models evolve, he hopes to link them with models that other center researchers are developing to provide turn-key simulation that could follow the manufacturing process all the way from the molecular level to final product.
Currently, de Pablo is concentrating on modeling thermophysical properties of fluids and solids. His first step is to simulate the interaction of as many molecules as possible of pure polymer. (Increasing the number of molecules makes the models more accurate, but can bog down the computer, and slow the simulation.) Once he has a good representation of how the material behaves, he can simulate conditions that would be difficult or impossible to measure in the real world. He can, for example, predict how materials flow at pressures of thousands of atmospheres or how well they lubricate in a space just a few millionths of an inch across. This software, Giacomin says, would let researchers "tinker with molecular structures until they figured out which one would give them the best flow behavior. It would really change the way new polymer products are invented."
Another application of de Pablo's simulations is predicting the effects of imperfections in manufactured materials. Manufactured polymers always have inconsistencies and contaminants that can alter the materials' behavior. And because these flaws usually appear in combinations and can't be entirely eliminated, it is difficult to gauge their effects. But after de Pablo has generated a computerized version of a pure material, he can follow its subtle changes as he adds individual flaws. "An advantage of the simulation is that I have perfect control over the molecule, which is difficult to achieve in real life," he says.
Just as important as predicting the behavior of polymer molecules is understanding how different combinations blend. Because creating polymers from scratch is expensive, most "new" plastic materials are blends of established polymers. And most of this mixing is in machines that were designed through trial and error almost a century ago. To improve the mixing process, researchers need to understand what exactly is happening inside these devices. "Now, through simulation, for the first time we can peek inside and see how things flow in these mixers," says Tim A. Osswald, an associate professor of Mechanical Engineering.
Osswald's simulations provide striking, real-time images of how different types of plastics combine synergistically to produce blends that out-perform their composite parts. To find the most efficient process, the models permit the researcher to vary virtual components-such as mixing pins and different types of mixing heads-and factors such as viscosity ratio, surface tension, temperature dependence and viscous heating to accurately project how the materials will interact.
The resulting patterns, however, are extraordinarily complex, and would normally tax even the fastest computer workstation. Osswald, however, accelerates and simplifies his models with mathematical strategies that use a sampling of data from the mixture's edges to portray the entire batch. "With that mathematical trick," he explains, "we eventually have a governing equation that only needs information on the boundaries, but is still representative of what's happening inside."
Actually seeing what's happening inside a mixing machine, however, isn't that easy. The center's battery of polymer processing equipment-which represents all of the usual manufacturing methods used in industry-is designed for production rather than observation. To check his simulations and watch how his polymers were mixing, Osswald's group constructed a single-screw extruder with a transparent shell. Based on his simulations and observations, companies such as 3M are expected to design a whole new generation of processing equipment.
Osswald is also focusing on a specific manufacturing process: how sheets of plastic shrink and warp when compressed in large molds. "This is a problem that has given people headaches for 50 years," Osswald says. But now auto-parts manufacturers are using simulation software that his group developed to fine-tune the production of plastic body panels, which are made by squeezing layers of plastic in hot molds. Stringent ISO 9000 quality standards require that these panels deviate by no more than a single millimeter, a difficult-to-reach goal that demands precise control of the production process. By simulating the patterns that the polymer fibers form, how the materials solidify and how heat travels through the material in the molds, then varying such factors as the mold's temperature and the placement of the plastic sheets, Osswald has been able to accurately predict how well the plastics will conform to the molds. "Our models can predict the filling and they agree quite well with experiments," he says.
As computer technology evolves and researchers continue to make breakthroughs, Giacomin expects that soon-perhaps in just five years-a single set of linked simulations will allow engineers to choose among molecules on a computer screen and immediately see how the resulting material would perform. "That is the cutting edge of polymer technology," he says. "It's been a longtime dream."