Halstead cluster is aeronautics professor’s 21st century wind tunnel
March 10, 2017
Jonathan Poggie's high-speed fluid mechanics research aims to make supersonic flight more efficient, but he doesn't need a physical wind tunnel to test his theories – because he has access to supercomputers like Purdue’s Halstead cluster.
Poggie, an associate professor of aeronautics and astronautics, studies fluid mechanics in the supersonic and hypersonic (greater than five times the speed of sound) regimes. In particular, he researches a phenomenon known as “separation unsteadiness,” in which air flows away from the aircraft surface and recirculates, creating oscillations that can lead to extra drag and to mechanical fatigue. He looks for ways to predict and control this flow, and analyzes devices that can be used to control flow.
"It’s a lot faster to try out different ideas for flow control on the computer than it is in the wind tunnel, and it’s a lot cheaper, especially in the high-speed regime,” he explains. “Running a hypersonic wind tunnel costs millions of dollars and it’s something we can prototype on the computer in just a few hours.”
But not on just any computer – high-performance computing is essential to Poggie’s research, because of the way fluid flow must be broken up into grid cells to be modeled. Using a larger number of smaller cells creates more accurate models that can capture turbulence on a smaller scale, but requires a supercomputer’s parallel processing capabilities to efficiently perform the calculations.
Poggie’s graduate student Akshay Deshpande is currently using Halstead to model a configuration representing the inlet of a supersonic jet engine. He is investigating how a vortex forms next to a sharp fin, and how the flow can be controlled by an electrical device known as a plasma actuator.
“This geometry contains around 17 million grid points, and the code I am using is parallel software,” says Deshpande. “So running it on multiple cores speeds up my calculations significantly.”
To get his codes running correctly on Halstead, Deshpande has taken advantage of support from ITaP Research Computing staff over email and at the regular Coffee Break Consultations, where researchers can seek help from ITaP staff in an informal setting.
Halstead is the ninth research supercomputer built in as many years through Purdue’s Community Cluster Program, which gives Purdue researchers the best collection of high-performance computing resources for use on a single campus in the country. Halstead features two 10-core Intel Xeon CPUs per node and 128 GB of RAM and is optimized for tightly-coupled science and engineering applications.
Poggie appreciates the hands-off nature of the Community Cluster Program. “I’ve been involved in maintaining computers before, and it’s extremely tedious,” he says. “We have to do a lot of things that we’re not specialists in to maintain our own computers, so having Halstead is great.”
Space in Halstead is still available, and researchers can buy capacity in it through ITaP Research Computing’s cluster orders website. To learn more about Halstead and the Community Cluster Program, contact Preston Smith, director of research services and support for ITaP, firstname.lastname@example.org or 49-49729.