Skip to main content
NREL Leads Wind Farm Modeling Research

This offshore wind farm, off the coast of Denmark, features ten 2-MW Bonus wind turbines.
Photo provided by HC Sorensen, Middelgrunden Wind Turbine Cooperative

NREL Leads Wind Farm Modeling Research

Researchers study the atmosphere surrounding large turbines to optimize performance.

Wind turbines can be greedy because individual machines are designed to gobble up as much wind energy as possible. Until recently, that was considered normal. Now, however, most turbines are built to be used in wind farms, and a single turbine's performance matters far less than the farm's overall efficiency. To address this new paradigm, the National Renewable Energy Laboratory (NREL) has created complex computer modeling tools to improve wind turbine design and overall wind farm performance, based on the best possible research data.

To that end, starting in 2011, NREL and the Renewable and Sustainable Energy Institute (RASEI) partnered with scientific institutions to conduct a complex study to understand atmospheric turbulence and turbine wake behavior. NREL and RASEI worked directly with:

  • National Oceanic and Atmospheric Administration
  • University of Colorado, Boulder
  • Lawrence Livermore National Laboratory.

A multi-organizational team of experts deployed precise instruments to create a detailed picture of the atmosphere surrounding large turbines. Among these instruments is the high-resolution Doppler LIDAR—a laser-based system that stands for "light detection and ranging." LIDAR scans can measure the wind speeds in a slice of air up to 1 kilometer (3,280 feet) from the ground and 6.9 kilometers (4.3 miles) long. In other scanning modes, it can focus on a single turbine and the area within 10 kilometers (6.2 miles) of it to observe the turbine's wake near the surface. For this study, researchers focused on a 2.3-megawatt turbine that rises almost 100 meters (328 feet) from its base to its hub. This research produced the first ever three-dimensional portrait of atmospheric activity surrounding a multi-megawatt wind turbine.

In addition to its work with Doppler LIDAR, the interagency team employed other high-resolution atmospheric instrumentation. NREL scientists gathered wind and turbulence data using commercial scientific platforms, including a specialized laser called a Windcube LIDAR and a sonic detection and ranging (SODAR) system, the Second Wind Triton. While LIDAR focuses on light bouncing off of particles in the atmosphere to measure wind speed, SODAR measures sound waves bouncing off of density fluctuations in the atmosphere. One beam of light or sound measures one direction, so three beams are needed to measure all three directions in space—up/down, forward/backward, and side/side. The researchers also installed high-frequency sonic anemometers on two new 440-foot meteorological towers to further supplement the data. Each of these instruments provided essential information to understand the dynamic inflow and turbine wake system. See the Spectrum of Clean Energy Innovation story, "NREL Studies Wind Farm Aerodynamics to Improve Siting".

Computing the Complex Nature of Wind Turbine Wakes

Under a clear blue sky, and constructed on treeless, grassy hills, a multitude of wind turbines, featuring various construction styles, cover the landscape.

The Tehachapi Pass Wind Farm, located in California, has nearly 5,000 turbines, and can generate 705-MW of wind energy.
Photo by David Hicks, NREL

By their nature, wakes are complicated. Fluctuations in air temperature throughout the day can affect wind turbine wakes, so it's important to observe them in detail and understand how to minimize their impacts, according to Julie Lundquist, professor of atmospheric and oceanic sciences at the University of Colorado and a joint appointee at NREL.

Drawing on Lundquist's findings, NREL researchers are now implementing a Dynamic Wake Meandering model to simulate airflow through a wind farm. They wanted to compare the results they were collecting from their high performance computing model to this lower fidelity model as well as to more standard, basic industry models.

Unlike most wind energy industry teams, NREL investigators have access to high performance computing, including RedMesa, NREL's most powerful high-performance computing system. This system has a peak computational capability of about 180 teraFLOPS, which means it can perform 180 trillion "FLOPS," or floating point operations, per second. In the past, this sort of high performance modeling has not typically been available for creating industry wind models.

"We've been able to see what the differences are, and ask, 'Do we really need more complicated models, or can we just improve more simple ones?'," said Pat Moriarty of the National Wind Technology Center, based at NREL.

For the multi-agency research project, this high performance computing is invaluable. In order to study offshore wind farms, NREL gathered wind plant data from Lillgrund Wind Farm, located off the coast of southern Sweden. Moriarty, along with NREL teammates Sang Lee and Matthew Churchfield, was able to simulate wind velocity, turbulence, kinetic energy, and time-averaged power output of a wind plant.

They have also successfully calculated the relative impact of wake and atmospheric turbulence on wind turbine structural loading. This was the basis for a computational model that will simulate wind turbine arrays. The group expects to conduct another such study next year for a wind farm off the coast of the Netherlands.

To ensure accuracy, their results must be compared with actual observations. To carry that out, NREL and the Centro Nacional de Energías Renovables (CENER) in Spain are leading an international collaboration through the International Energy Agency to gather data from researchers and operating wind farm owners, and to benchmark existing simulation tools.

Dual Purposes of the Research

A computer visualization showing wind velocities in the vicinity of a wind turbine. Colors indicated lower velocities in the wind turbine wake.

This high-resolution Doppler LIDAR scan shows radial wind velocities in the vicinity of a wind turbine, with cooler colors indicating lower velocities in the wind turbine wake.
Image provided by Kenny Gruchalla, NREL

Overall, this ongoing research has two basic purposes:

  • The first is to understand the physics of local airflow, including how airflow changes in a wind farm, and how turbines interact with each other and the atmosphere. This is key because as turbines grow in size—approximately doubling in height over the past five years alone—they present more complex problems to wind turbine designers and operators.
  • A second purpose, Moriarty said, is that new insights from models can improve wind farm layout and operations. Based on the findings in the study, a wind project designer may want wider spaces between turbines to enable the farm to reap more energy. Also, after building a wind farm, an operator can design and implement a wind farm control system to fine-tune efficiency gains. This is becoming a cutting-edge concept in the wind farm industry.

As Moriarty explained, control systems were traditionally designed around single turbines—such as the above-mentioned wind gluttons. These designs allowed each turbine to optimize its own energy capture and minimize any wind damage to its structure. Using wind farm control systems, turbines also "know" about surrounding turbines. This technology can coordinate turbines to allow the maximum efficiency for a renewable energy project. For example, a front row of turbines may be tilted a certain direction in order to let more wind pass through so that a stronger airflow hits downstream turbines. As a result, each lone turbine may not capture the most energy, but the wind farm as a whole is capturing more energy. That's the eventual goal of these sorts of optimization tools, Moriarty said.

Leading a Worldwide Effort in Airflow Simulation for Wind Farms

An illustration showing radial wind velocities in the vicinity of a wind turbine. The illustration shows lower speeds of wind turbine wake with cooler colors.

A computer model image of the downstream wakes on a wind farm.
Credit: Kenny Gruchala, NREL

The timing is right for this pioneering research. Today, even though industry is increasingly interested in the large potential payoff from using these new techniques, there are relatively few groups around the world looking into wake models and testing airflow in simulation. One of the more intriguing innovations is the "wake steering method," which uses a control system to reconfigure a turbine's orientation and thereby guide the downstream wakes. This technique works without reconfiguring the massive wind towers. "Nothing else is changing--and you're getting free energy because of that," Moriarty said.

Efficiency gains can make a big economic impact. Field studies in Europe have shown an estimated 2% efficiency improvement with the wake steering method. However, these techniques are so new, the studies haven't been fleshed out yet. Moriarty said that "probably the most you can gain in efficiency will be about 10%, but even 1% over the lifetime of an average wind farm is still $20 million."

NREL publicly released its latest wind farm modeling tool in January, and held a webinar in May which was attended by about 100 wind energy stakeholders worldwide. The new tool is already proving popular, and there have been several hundred downloads of the airflow simulation tool. Not only is this going to academia, but industry—manufacturers, wind farm developers, and consulting groups—are taking advantage of this offering as well.

And this acceptance is expected to grow. In November, NREL will host the International Wakebench meeting for International Energy Agency Task 31. Sixty groups from around the world are expected to attend, and NREL researchers will showcase their models and findings. "We are one of the world leaders in the topic right now—it's a big topic of interest in general," Moriarty said.

Learn more about NREL's wind research.

Learn more about NREL's Spectrum of Clean Energy Innovation and how the laboratory's capabilities emulate the nature of the innovation process.

The NREL Spectrum of Clean Energy Innovation

Issue 3

RSS Subscribe to the RSS feed.

Editorial Team

  • Kim Adams | Managing Editor
  • Bill Gillies | Creative Director
  • Dennis Schroeder | Photographer
  • Jennifer Josey | Editor
  • Michael Oakley | Web Development
  • Amy Glickson | Web Development
  • Email the editor