West Lafayette, Ind.—A new computer model should help engineers design better "thermal barrier" coatings that will enable jet engines to run at higher temperatures, last longer, and perform better.

The model, developed by researchers at Purdue University, accurately predicts how well certain coatings will work before they are created, said Klod Kokini, a professor of mechanical engineering and an assistant dean of the Schools of Engineering at Purdue.

Thermal barrier coatings primarily see use in turbine engines in jet aircraft, but researchers are trying to develop coatings for diesel engines in cars and trucks. The coatings are a mixture of ceramic and metal applied to engine parts in a process known as plasma spray technology. You first heat powdered metal and ceramic with a torch and then spray it onto metal parts.

"It's like spray painting except that it occurs at very high temperatures," Kokini said. "The coatings extend an engine's lifetime, and they enable you to run engines at higher temperatures, which improves performance."

The coatings are "graded," meaning they are applied in layers that have varying concentrations of metal and ceramic. The outermost layer contains mostly heat-resistant ceramic, and underlying layers contain increasingly higher concentrations of metal.

Over time, as the high operating temperatures inside an engine affect the parts, the coating deforms. Then, after a user turns off the engine and it begins to cool, cracks form in the coating, eventually causing pieces to break off.

Some coatings resist cracking better than others. But the only way for engineers to rate the performance of different coating mixtures and various layer combinations has been to actually create and test them in a laboratory.

"We know the behavior of the metal alone, and we know the behavior of the ceramic alone," Kokini said. "But what happens when we mix them in, say, 20% metal and 80% ceramic, or 50% ceramic and 50% metal?

"We now have a comprehensive computational model that allows a designer to say, 'I'm mixing this and this together. What kind of properties can I expect?' It's a very powerful tool. No one had ever developed a methodology to actually do these predictions before."

The model, which is more than 90% accurate, promises to save time and money by ruling out mixtures that will not work well, Kokini said.