University of Minnesota researcher Leonardo Chamorro adjusted a wind sensor that can measure fluctuations caused by wind turbine models in the U of M wind tunnel at St. Anthony Falls Laboratory.

Glen Stubbe, Star Tribune

Leonardo Chamorro adjusted a wind turbine model in the U of M wind tunnel at Saint Anthony Falls Laboratory.

Glen Stubbe, Star Tribune

Wind power: Tunnel vision at U

  • Article by: JIM DAWSON
  • Special to the Star Tribune
  • May 1, 2009 - 7:51 AM

In the long wind tunnel at the University of Minnesota's St. Anthony Falls Laboratory, miniature wind turbines stand waiting for the lab's scientists to turn on the flow of air. The dozen or so turbines, just several inches tall, are arranged in patterns similar to the layouts of the real 300-foot-tall wind energy turbines that populate wind farms across Minnesota's southern tier.

While the small turbines look like toys, they are really a critical part of a $1 million collaboration between the laboratory and two Minnesota companies -- WindLogics and Barr Engineering -- to improve efficiency at wind farms.

Wind turbines, with blades typically about 300 feet long, create wakes that disrupt the flow of air to other turbines.

Understanding the effects of this wake turbulence is critical because even a slight decrease in the power output of a turbine can cost a lot of money.

"If you look at a 1 percent reduction in output from a 100-megawatt wind farm, you are talking about the loss of $100,000," said Fotis Sotiropoulos, director of the St. Anthony Falls lab.

The goal of the collaboration, said U of M scientist Fernando Porte-Agél, is to develop high-resolution computer models that, by providing good information on wind fields and turbulence, "can guide the design of wind projects so we can know where to put the turbines."

Loss of efficiency isn't just an occasional problem. On wind farms across the country, "the output of turbines is a few percentage points below what the models predict they should be," said John Wachtler, an environmental engineer at Barr Engineering in Minneapolis.

"Nobody is sure why, and when you spend $3 or $4 million for a turbine you don't want to see a 3 or 4 percent decrease in output from what you expected," he said. "One possibility [for the lower power output] is that we're not able to model short bursts and other changes in the wind on the turbines."

Wind energy is of growing economic importance in Minnesota. The state gets more of its energy from wind -- on a percentage basis -- than any other state in the country. It ranks fourth for total output -- 1,725 megawatts of wind energy -- behind Texas, Iowa and California.

The power generated by wind farms along Buffalo Ridge in southwestern Minnesota has outstripped the ability of the grid to carry the power, so the newer farms are being constructed in the southeast part of the state, Wachtler said.

Model turbines

To develop a realistic computer model of the wind patterns that will affect not only a particular wind farm, but an individual turbine in the middle of that farm, researchers are doing "multiscale modeling," Porte-Agél said.

They start with existing models that see the atmosphere on a very large scale. Then, using data from that model, they move to models that see wind patterns and atmospheric turbulence on an ever-smaller scale, until they wind up with an accurate prediction for the exact site where a wind turbine will be placed.

At least, that is the goal. "The models that industry uses now are rather coarse," Sotiropoulos said. "So nobody really knows the effect of atmospheric turbulence on a wind turbine."

To make a useful model, he said, "we want to go from a regional scale, which is on the order of kilometers, to a local scale, which is on the order of 100 meters, down to 1 meter.

The current models are good at the kilometer scale, he said, "approximate" at the wind farm scale, and essentially nonexistent at the 1-meter scale that would be critical in deciding exactly how to place turbines.

After including atmospheric data, which is notoriously complex, modelers must include local topography. Is the wind farm on flat prairie or it near hills or a valley? Is the surrounding land smooth, or is it rough? Is there a forest nearby?

Once the models are refined enough to look at what is happening on a wind farm, then understanding the effect on a turbine of turbulence being created by other turbines is critical. "When you are extracting energy from the wind, you have to slow the wind down and it creates wake turbulence and that has an impact on the turbines downwind," said Bob Conzemius, a senior scientist at the WindLogics research center in Grand Rapids, Minn.

Which is why the "toy" turbines are arrayed in the wind tunnel. By testing different turbine layouts, scientists can observe which patterns result in the least interference.

"We don't understand how the wakes interact, and this will help with that," Conzemius said.

The data from the wind tunnel will be fed into the models and then compared with very precise measurements being taken at a wind farm near Prairie Island. By combining data from the existing models with data from the wind tunnel and testing it against what is happening on the real wind farm, scientists hope to find out what can be done to make wind farms as efficient as they should be.

This story is provided by the Inside Science News Service. Dawson is editor of the Washington, D.C.-based news service, which is supported by the not-for-profit American Institute of Physics, a publisher of scientific journals. He was a reporter at the Star Tribune for 20 years and a science writer for 13 of those years. His e-mail is

To see an interactive guide to Minnesota's wind power, go to

© 2018 Star Tribune