Wind farms, a perennial source of green energy, already supply more than 10% of electricity in the United States. But since they are affected by the turbulent flows ubiquitous in the atmosphere, can future wind farms be made even more efficient?
Charles Meneveau, the Louis M. Sardella Professor in Mechanical Engineering, is analyzing datasets from computer models to improve the performance of large arrays of wind turbines. The datasets are part of the Johns Hopkins Turbulence Database, established with support from the Ralph O’Connor Sustainable Energy Institute and the Institute for Data-Intensive Engineering and Science, which Meneveau helps direct.
Researchers can take something invisible and give it color and shape, essentially “seeing” the air move. The result may look like abstract art, but it’s hard science and physics, and when illustrated, can reveal important trends and nuances.
“Our team simulated a wind farm for a full 24 hours to analyze what’s happening at different times of the day and night, under very different atmospheric conditions,” says Meneveau.
“We observed, for instance, that in the early morning, when the sun is rising but conditions are still nighttime-like, the back turbines generate more power than the ones in the front, which is usually the other way around,” says Meneveau. “The downstream turbines can generate more power at this time in the 24-hour period. Being able to predict such unexpected trends and their probability of occurrence is crucial for estimating power generation potential over time, which is important for planning and economic feasibility studies.”
“Our team simulated a wind farm for a full 24 hours to analyze what’s happening at different times of the day and night, under very different atmospheric conditions.”
— Charles Meneveau
The Johns Hopkins Turbulence Database, a massive repository of datasets from computer simulations of various types of turbulent flows that is publicly available, is one of 10 chosen by the National Science Foundation to be integrated into the National Artificial Intelligence Research Resource Pilot. A Johns Hopkins civil and systems database related to fracture mechanics was also selected for this program.
“It’s very complex data, but we put it out into the world, and researchers interested in a specific type of turbulence can find and use it to further their work. This effort is continuing to put Hopkins on the map in terms of generating trusted data for scientists and engineers, as well as for AI training,” says Meneveau.