Understanding the performance of individual wind turbines at any modern wind farm is critical. The characterization of turbine efficiency is a necessary first step in detecting underperformance, evaluating turbine upgrades, refinancing or reselling a project, identifying degradation in performance over time, and, increasingly for wind farm owners, trying to decide whether to invest in contractual power performance testing within the first year of plant operations per the International Electrotechnical Commission’s (IEC) 61400-12-1:2005 standard.[adright zone=’190′]
It is well known that the IEC-12-1:2005 methodology requires the use of a hub height meteorological mast to measure the wind conditions upwind of the turbine(s) being tested. In the spirit of this current standard, wind farm owners in North America are increasingly performing operational (non-contractual) power curve tests using ground-based Doppler LiDAR alone. Indeed, certain Doppler LiDAR models provide a very similar input to hub height IEC masts for assessing wind turbine performance in simple terrain sites, mimicking the role of the mast but with a much higher degree of flexibility and at a much lower cost. Resultant uncertainties on power performance are slightly higher, but not significantly so, and the results of such a test give wind plant owners very similar benefits in understanding whether turbine performance meets, underwhelms or exceeds expectations and, just as importantly, whether far more costly and time-intensive actions such as formal IEC 61400-12-1 testing under the current standard will pay off.
Many readers are probably also aware that an update to IEC-12-1 is likely to be ratified and enacted in 2017 and that this update will incorporate the concept of rotor equivalent wind speed (REWS) as an alternative to hub height wind speed as input to the turbine power curve calculation. REWS is best derived with measurements across the vertical surface of the turbine rotor, and as a result, the IEC power performance test standard will, for the first time, allow the use of remote sensors as a complement to short (< hub height) met towers for contractual power curve tests. Data from the remote sensor will be used as input for the REWS calculation, while the short met tower serves as an on-site verification device for the remote sensor itself.[adleft zone=’190′]
Economics, practicality and uncertainty
A Doppler LiDAR-based operational power curve test is one of many such methods available to wind farm operators currently. Each available method strikes a unique balance in terms of economics, practicality and resulting annual energy production (AEP) uncertainty. We focus here specifically on ground-based wind measurement methods, which mimic the current IEC standard in terms of mast placement upwind of the turbine(s) being tested, acknowledging that a variety of different nacelle or hub-mounted instrumentation options are also being used as wind speed inputs to operational power curve tests. In general, we find that a ground-mounted remote sensor located upwind of the turbine is still the most comfortable alternative to conventional met masts for wind farm owners and investors in the U.S.
Ground-based measurement methods, such as those from an industry-standard SoDAR or Doppler LiDAR, will typically differ in price by a factor of 2 to 2.5 x when considering average monthly rental fees for each device, with both methods falling well below the cost of a hub height met tower, especially with the assumption of short (e.g., three-month) test campaign durations. A purchased remote sensor can be cost-amortized relatively rapidly when considering the low cost of relocations compared with repeated hub height met tower installation and decommissioning.
Preexisting nacelle anemometry is the low-cost wind speed input method for power curve testing, as well as the most practical one, but it suffers significantly from measurement uncertainty to the point where it cannot be utilized with any confidence for absolute evaluations of the turbine’s efficiency unless calibrated on-site by a reliable reference instrument. The remote sensor option is less practical in general than a nacelle anemometer would be by virtue of the need to install the device separately at a unique location upwind of the turbine(s) being evaluated, but it is far more practical than a met tower installation, given the typical lack of permit or soil disturbance required, as well as a far quicker and simpler installation and decommissioning process generally.[adright zone=’190′]
In considering various ground-based remote sensor options, it is important to note that industry-standard SoDARs measure with a significantly higher degree of uncertainty than industry-leading Doppler LiDARs due to their sensitivity to changes in wind shear. A study by independent consultancy Deutsche Wind Guard estimated the total hub height wind speed uncertainty from a typical SoDAR at greater than 5.0% versus 2.5% for the Doppler LiDAR (Figure 1). The SoDAR’s high sensitivity to shear implies that the accuracy at a given height will vary based on shear conditions far more than what the LiDAR has been shown to experience. In the case of the Doppler LiDAR, the uncertainty is driven more so by its verification to an already uncertain reference (a calibrated cup anemometer) than by any flaw in its measurement performance, such as shear-based sensitivity.
AEP uncertainties typical of mast-, Doppler LiDAR- and SoDAR-based power curve tests are illustrated in a relative sense, along with cost, in Figure 2. Based on its favorable nexus of cost, practicality and accuracy, a LiDAR-measured power curve will typically offer significantly more confidence on which to make actionable and defensible business decisions when compared with a nacelle anemometer or SoDAR-based test and still at a small cost compared with an IEC met mast.
A real-world example
A Doppler LiDAR was utilized for an operational power curve test in late 2014 at a utility-scale wind farm in the midwestern U.S. The wind farm owner’s objective in this case was to rapidly and economically assess the performance of a targeted turbine installed just a few months prior, both to confirm suspected underperformance and to determine whether to invest in more expensive, contractual testing as a basis for warranty claims and possible liquidated damages payments from the turbine original equipment manufacturer.
The turbine was tested because initial SCADA data analysis indicated an overestimation of the calculated energy produced based on the nacelle anemometer power curve when compared with the real energy produced (calculated > real production). This specific LiDAR was chosen to serve as the reference wind measurement device for the test due to its ease of installation, proven accuracy and wide industry acceptance. These data were combined with turbine SCADA data and ground-level atmospheric measurements to provide an accurate and meaningful understanding of the turbine’s performance relative to its reference power curve.
The turbine under evaluation was located on the outer edge of the wind farm and frequently experienced unwaked winds as a result. The LiDAR was placed 2.5 rotor diameters upwind of the turbine in a prevailing wind direction, beyond the turbine induction zone. Prior to its deployment, the LiDAR was tested and validated at company headquarters against a calibrated reference LiDAR.
Following the data collection period, a series of straightforward filtering steps to both LiDAR data and turbine SCADA data were applied by the wind farm owner’s internal performance analysts, and final calculation of turbine power curve was performed.[adleft zone=’190′]
The wind farm owner concluded that the wind turbine significantly underperformed its reference power curve.
A secondary result showed, not unexpectedly, that the performance deviated between unstable daytime conditions (low shear, high turbulence intensity in a convective boundary layer) and stable nighttime conditions (high shear, low turbulence intensity in a nocturnal boundary layer) (Figure 3 and Figure 4). Following this internal test, the wind farm owner was able to confidently proceed with a formal IEC power curve test while the LiDAR was moved to another location to analyze additional wind turbines. Background experience and industry research coupled with on-site measurements with high data availability and low scatter provided a high level of confidence on which to make this important business decision.
Evan G. Osler is senior technical lead for remote sensing at Renewable NRG Systems. He can be reached at email@example.com. Guillaume Coubard-Millet is founder of consultancy BCW Analysis. He can be reached at firstname.lastname@example.org.