Why wind-farm developers should care about measuring atmospheric turbulence

By Sonia Wharton, Atmospheric Scientist, Deputy Group Leader, Energy Group
Lawrence Livermore National Laboratory

& Jennifer F. Newman, Postdoctoral Fellow
National Wind Technology Center, National Renewable Energy Laboratory

Article co-author Sonia Wharton (Lawrence Livermore Labs)

Article co-author Sonia Wharton sets up Lawrence Livermore National Laboratory’s Wind Cube v2 lidar at California’s Altamont Wind Farm. The lidar was used to measure turbulence and wind shear across wind-turbine blades.

The role of atmospheric turbulence in influencing wind-turbine power production remains an unsolved mystery despite a growing number of researchers who have attempted to make sense of this issue. Turbulence, a term for short-term deviations around the average wind speed, can cause fluctuations in turbine power production and structural loads.

While research strongly suggests that ignoring atmospheric turbulence can result in significant errors in power-curve measurements and annual energy production, it appears that there may be no universal relationship between turbulence and power production.

Typically when we think of a wind farm operating in a turbulent atmosphere, we picture a waked turbine, battered by vortex eddies (circular wind flow) shed from turbine blades upwind. However, turbulence is present nearly everywhere, and is constantly produced and diminished over a wide range of temporal and spatial scales. This article aims to unravel some of the complex factors that remain unsolved regarding turbulence and wind power.

Measuring turbulence
Many wind farms still rely on tall meteorological towers to measure the local wind resource. These towers are typically equipped with cup anemometers that provide the mean wind speed over a 10 or 15-minute averaging window.

However, wind is actually comprised of three components:

  • Mean wind, which is measured easily with a cup anemometer
  • Waves, which result from wind shear, wind flowing over obstacles, or the boundaries between layers of air with different densities (waves will not be discussed here), and
  • Turbulence, which is typically quantified with a parameter known as turbulence intensity (TI). TI is calculated as the horizontal wind-speed standard deviation divided by the mean wind speed over the same time period. Loosely translated, TI gives the percentage of the horizontal flow that is turbulent, and values of TI are used in turbine design standards and sometimes to stratify power curves. Values of TI measured by a cup anemometer fail to reflect turbulent motions in the vertical direction, which can be significant under daytime, convective conditions, or to indicate which scales of turbulent motion are prevalent in the atmosphere.

A more complete depiction of turbulence requires instruments with higher precision and faster sampling rates than provided by a cup anemometer. While sonic anemometry fulfills these requirements, high maintenance costs make it often impractical to operate long-term at wind farms.

Instead, remote-sensing instrumentation has provided a more practical solution given their mobility and reliability.

Remote sensing
Remote sensing is the science of obtaining information about objects or areas from a distance, typically from aircraft or satellites, but also from ground-based sensors. Sound Detection and Ranging (sodar) and, more recently, Light Detection and Ranging (lidar) instruments are increasingly deployed by wind-farm operators and are no longer seen as “neat tools” that only scientific researchers have access to.

Schematic of a Doppler (Lawrence Livermore Labs)

Schematic of a Doppler, vertical-profiling lidar. The instrument derives wind speed and direction by emitting beams of light forming a volumetric cone above the instrument.

Sodar and lidar are used as wind profilers. While sodar derives the wind speed and direction by measuring the scattering of sound waves, lidar uses the Doppler shift in back-scattered laser energy to estimate the wind flow. Companies such as Natural Power, Leosphere, and Avent offer wind industry-friendly lidars that can be deployed on the ground looking upward, or on a nacelle hub facing upwind.

Recently, ground-based profiling lidars, such as the ZephIR 300 and WindCube v2, were given an official “thumbs up” by the International Electrotechnical Commission (IEC) for use in a standard that describes wind-power performance testing and wind-resource assessments.

This new standard, IEC 61400-12-1:2017, is a major milestone for the lidar industry. Vertically profiling lidars can now replace traditional mast-mounted cup anemometers in simple, flat terrain.

Remote-sensing instruments come with their own set of challenges with measuring turbulence because they measure wind speeds averaged across probe volumes that are typically tens of meters in length. Also, they must collect data across a large scanning circle to deduce the three-dimensional components of the wind, which assumes homogeneity in the flow. Such measurement techniques are in stark contrast to the “point” measurements offered by a cup anemometer on a tower, and often result in different estimates of TI.

Power production
Regardless of the challenges inherent in measuring turbulence, researchers and wind developers are increasingly recognizing that turbulence estimates are vital to understanding the intricacies of turbine power production. Recent studies have used met towers, remote-sensing devices, atmospheric and turbine models, and machine learning tools to help understand the complex effects of turbulence on power production. Interestingly, an examination of the studies’ results offers no clear answer.

For example, some studies show that higher levels of turbulence lead to higher power generation. Other research shows this is only true during low wind-speed events. Still, other studies have found that turbulence hinders power generation. Here, the thought is that increased loads add fatigue to turbine components thereby reducing power output.

Some of the studies use TI to estimate the amount of turbulence in the air. Alternatively, turbulence kinetic energy (TKE), which employs all three components of variability in the wind rather than just the horizontal variability, is used as a metric in some studies. TKE is the mean kinetic energy per unit mass associated with eddies in turbulent flow. Nevertheless, even the use of TKE provides no clear universal relationship between turbulence and power production.

The future of turbulence measurements
So where do we go from here? One suggestion is to avoid thinking of turbulence as a “blanket” sum, such as in the calculation of TKE, or as a “blanket” ratio, such as in the calculation of TI. Atmospheric turbulence consists of a wide range of overlapping scales, unique to every point in space and time. By using parameters such as TKE or TI, calculations over-simplify the chaotic nature of the atmosphere. Researchers need better ways to characterize turbulence and relate turbulent motion to turbine power production.

Power generation is not a simple function (Lawrence Livermore Labs)

Power generation is not a simple function of hub-height wind speed as this plot of 10-minute data shows from a California wind farm. Much of the variability may be due to the variation in turbulence intensity at any one time period. Low turbulence intensities are most often indicative of stable, nighttime air, and high turbulence intensity is typically present during convective, daytime atmospheres.

One method for obtaining additional information is analysis of the spectral content of turbulence (i.e. the distribution of turbulence across different temporal and spatial scales). Although this approach is currently applied to design load calculations, it is not directly incorporated into power estimates.

Several recent studies have suggested that turbine power production is only sensitive to particular scales of turbulence, so the spectral content of turbulence is clearly important for power production. While a parameter such as TI cannot give an indication of the scales of turbulence present in the flow, variables such as a characteristic length scale can be derived from turbulence spectra and used to classify turbine performance.

In addition, data from remote-sensing devices can be used for more than just mean wind speed and TI calculations. Much work has been dedicated to adjusting TI measurements from remote-sensing devices to bring the values closer to what would be measured by a cup anemometer. In several ways, the remote-sensing device is providing more information than we give it credit for.

By collecting measurements at several points around a horizontal scanning circle, vertically profiling remote-sensing devices are gaining information about the spatial variability of the wind. This variability is directly related to the spatial distribution of turbulence in the atmosphere.

In summary, too much information is lost when turbulence is considered as a “blanket” sum or ratio of the standard deviation to the mean wind. The wind industry could obtain more detailed results if the full turbulence spectrum is accounted for when quantifying the influence of turbulence on power performance. While a couple methods for obtaining spatial information on turbulence from measurement devices were proposed here, many additional methods likely exist.

With the advancement of remote-sensing devices and the vast amount of field projects being conducted on operational wind farms, it will be worth the effort to unravel more of the mysteries of turbulence over the coming years.

The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States government or Lawrence Livermore National Security, LLC, and shall not be used for advertising or product endorsement purposes.

Speak Your Mind