Although composite modeling tools have existed since the 1950s, it’s only recently that they’ve been able to capture the complexity of a composite.

What’s changed is computing power. “You have more computing power in your phone today then we had at General Dynamics when we were trying to build airplanes in the 1970s,” says Pipes. “Now that we have this power at our fingertips, we finally have the simulation tools that we need to model every possible material combination or meso structure.”

But looking at the combination of materials is just the start. “Since the properties of the materials change by the way you make the material, you have to model and simulate the manufacturing process as well,” he says. In injection molding, for example, the way the material flows into the mold changes the fibers’ orientation, which means that one part of the composite will have different properties than another section that’s just two inches away. “So within this same manufactured part you have an array of materials; you do not have a single set of properties,” Pipes says.

NIAR---sample-composite-radome-bird-strike

Using simulations of bird strikes, researchers are able to get much better information about what is happening at the moment of impact than they can get by looking at films or measurement gauges. Photo Credit: Wichita State University

The properties of the material also change depending on the way the material fits in the mold. Take a flat sheet of woven fabric and drape it over a complex shape like a car part, and the angle between the fibers moves from being perpendicular to one another to some different angle to one another. So every point in the composite part has a different set of orientations and thus different mechanical properties.

To determine all of a composite part’s properties, then, requires modeling not only the materials that went into its manufacture, but also the physics of the manufacturing process.

Bringing It All Together

When modeling composites, Purdue researchers work with 10 or 12 different computer codes (which are the base of the computer models) from several different companies. Purdue researchers are knitting all of those existing codes together to provide a single program that includes all the information in those various codes. In cases where there’s a gap, Pipes and his team are working on developing interfaces between existing commercial codes to fill them.

The researchers are already teaching OEMs and other companies how to assemble the codes and build these connecting pieces. “When they’re done, they’ll be able to model from the beginning of the manufacturing process all the way through the performance of the product,” says Pipes.