Optimizing power and efficiency in electric machines requires the testing of physical models during the design stage, which often requires significant time, knowledge and computational resources.
Led by Ali Davoudi, an assistant professor in the Electrical Engineering Department, the team will use a $285,000 grant from the National Science Foundation to develop hybrid software/hardware approaches and create highly accurate models that incorporate high-frequency effects and mimic the actual hardware prototype as closely as possible, all at very high speed on a desktop platform.
"Nearly all electricity in the world is produced through electric generators, and more than two-thirds of that is consumed by electric motors," Davoudi said. "Annual energy consumption by electric motors is expected to exceed 10 quadrillion watt-hours soon. Any improvement we can make in the process will have a huge impact in the way we generate, convert and store energy..
"The idea is to replicate the energy conversion devices and systems as closely as possible and run the tests as close to real time as possible. However, the more details added to the physical system, the slower it will run. We don't want to build a model, test it, fail, rebuild and repeat. If this research is successful, we will be able to design, optimize and run tests quickly and push out improvements in a much shorter timeframe."
To achieve this improvement, Davoudi and co-PIs Taylor Johnson and David Levine of the Computer Science and Engineering Department will begin with fundamental physics, then apply their findings to create order-reduction techniques that will speed up model simulations by as much as 1 million times. Finally, these models will be installed on hardware-centric simulation platforms to run parallel simulations and further improve the simulation speed.
Khosrow Behbehani, dean of the College of Engineering, underscored the importance of Davoudi's research to future breakthroughs.
"This transformative research will make it easier to create more efficient motors while greatly reducing the time and expense of modeling prototypes," Behbehani said. "With industry competing to increase efficiency and decrease costs, the potential return on Dr. Davoudi's team's work is noteworthy."
Highlights, News, UT Arlington