With renewable energy sources like solar and wind power set to become much more cost-effective over the next ten years according to recent studies out of Finland and the EU, interest in the space is building.
As experts and pundits address the respective merits of the renewables, scientists continue to study and advance the enabling technologies with high-performance computing power undergirding much of the research. One of the latest such projects relied on advanced computer simulations to depict the path of air through large wind farms. The research, led by a team at Johns Hopkins University (JHU), has the promise to inform the physical structure and layout of a coming generation of wind farms with the potential to greatly increase efficiency and output.
The effort relied on high-resolution simulations, a visualization of which is depicted below, and the resources of the San Diego Supercomputer Center, specifically the Trestles supercomputer, one of the leading science gateway platforms on the XSEDE circuit.
The study, which appears in the Journal of Renewable and Sustainable Energy, generated some interesting results. Unlike the conventional understanding that links maximum power output with turbines situated in a perfectly-staggered checkerboard pattern, the simulations suggest a lateral placement of turbines that minimizes wake effects from turbines in several upstream rows delivers optimal power output.
“The observed trends have implications for wind farm designs, especially in sites with a well-defined prevailing wind direction,” the authors write.
“As wind energy is becoming more important around the globe as a source for clean, renewable power, we’re finding that understanding the effect of spacing and relative positioning of the turbines on the wind-farm is crucial for a good wind-farm design,” said Richard Stevens, who along with Charles Meneveau and Dennice Gayme at JHU, developed the simulations and deployed them on Trestles. “Wind-farm designers typically rely on simple computer models that predict the wake effects caused by the turbines. These models work well for smaller wind-farms, but become less accurate for larger wind-farms, where the wakes interact with one another as well as with the atmospheric wind.”
This study addresses a more complex scenario, involving the interaction between the wind-farm and the atmospheric wind, and as such it required more compute power to solve. That’s where supercomputers like Trestles come into play. With 100 teraflops of peak performance and 10,368 AMD cores, it was designed by Appro and SDSC “to enable modest-scale and gateway researchers to be as computationally productive as possible.”
The team is calling its new model the “coupled wake boundary layer model.” Compared to current industry standards, it was found to more effectively predict wind-farm performance.
“The use of Trestles and the support provided by SDSC’s expert staff was of paramount importance for our research, which we hope will lead to more efficient and better thought-out wind-farm designs,” notes JHU’s Meneveau.
Long recognized as the leading science gateway workhorse in the NSF/XSEDE portfolio, Trestles is set to be replaced by Comet when the successor comes online this summer.
“Comet will have all of the features that made Trestles popular with users, with much greater capacity, while providing ease-of-access and minimal wait times to appeal to a broader base of researchers,” reports SDSC Deputy Director Richard Moore, a co-PI of the Comet project.
Funding for this project was provided by a “Fellowship for Young Energy Scientists” awarded by the Foundation for Fundamental Research on Matter in the Netherlands, and the National Science Foundation (NSF), including the WINDINSPIRE project.