Wind spirals and swirls around wind turbines in complex, chaotic patterns that are difficult to predict. Accurately modeling these dynamics could significantly enhance the reliability and efficiency of wind farms.
With this goal, engineers at the University of Notre Dame and mathematicians at the University of Oxford are collaborating to develop accurate, computationally efficient models for wind turbine simulations and wind farm control. These models will be tested on Frontier, one of the fastest supercomputers in the world.
The team has been awarded access to one million node-hours on Frontier, hosted at the Oak Ridge Leadership Computing Facility in Tennessee, to develop numerical methods that improve wind farm efficiency. Each of Frontier’s 9,400 nodes—a node being the equivalent of one computer—can perform the same number of calculations as roughly seven top-of-the-line MacBook Pros.
The team’s research is made possible with an award from the Department of Energy Office of Science and its Innovative and Novel Computational Impact on Theory and Experiment (INCITE) program. The program aims to accelerate scientific discoveries and technological innovations by awarding time on supercomputers to researchers with large-scale, computationally intensive projects that address “grand challenges” in science and engineering.
“Wind farms are highly dynamical systems,” said Jonathan MacArt, assistant professor of aerospace and mechanical engineering at the University of Notre Dame and the project’s principal investigator. “Given the unsteady nature of the wind entering the farm, the turbines must be operated in a way that maximizes power output. The turbines also interact with each other via their wakes—similar to the turbulent air left behind a helicopter as it hovers or takes off. These disturbances, which also interact with the surrounding atmosphere, can upset the flow over individual turbines, reducing the farm’s overall efficiency.”
MacArt’s team combines strengths in deep learning and physics-based modeling to simulate turbulent flows over static and rotating turbine blades. To correct for a scarcity of detail-rich datasets at wind farm conditions, the team will combine the predictive capacity of machine learning with the powerful computational methods of supercomputers.
Team members at the University of Oxford’s Mathematical Institute will contribute numerical methods development as well as wind turbine expertise. Applied mathematician Professor Justin Sirignano, as well as postdoctoral researchers Tom Hickling and Daniel Dehtyriov, will serve as the project’s co-investigators.
In addition to improving wind turbines and farms, the project could provide engineering industries with better tools for designing and optimizing high-performance aerodynamic systems. “Our work on flows over rotating blades has broader relevance beyond wind turbines”, said Dehtyriov. “The insights gained from this project can improve our understanding and control of flow over propeller and rotor systems. This is of growing interest in aviation, particularly in the development of tiny, unmanned flying devices and Electric Vertical Takeoff and Landing (eVTOL) systems that can take off, hover, and land vertically.”
— Karla Cruise, Notre Dame Engineering