Oct 23, 2012
Analysing wind-turbine performance
Wind farms are becoming ever more prevalent: last year, the total world power capacity of turbines went up by about 20%. Nonetheless, it is difficult to predict exactly how well new turbines will perform in different atmospheric regimes. Manufacturers provide a "power curve", which describes how power output should vary with wind speed, but this does not take into account irregular wind patterns such as wind shear and turbulence.
At some wind farms, particularly test sites, scientists place wind instruments – LIDAR and SODAR (laser and sound versions of radar) – to accurately measure how such wind patterns affect performance. But these instruments are expensive to install, and this leaves scientists with a shortage of data. The result is that it is difficult to know where to site, and how to arrange, future wind farms for the best power generation.
Now atmospheric scientists Julie Lundquist and Brian Vanderwende at the University of Colorado at Boulder have shown that, in the absence of LIDAR and SODAR data, measurements taken at the nacelles of wind turbines are a good alternative.
Nacelle wind measurements are available at all wind farms, and have been considered before as promising data sources. However, Lundquist and Vanderwende are the first to show that nacelle measurements are statistically robust. "This approach opens up the possibility of executing turbine performance analysis at many new locations, not just at academic wind-energy testing sites," said Lundquist.
Lundquist and Vanderwende took available data – including wind direction, turbine power output, atmospheric stability and nacelle wind speeds – from central North American wind farms. They then removed data points for any turbines that were downwind of others, classified time periods as different atmospheric regimes, and generated their own regime-dependent power curves.
The researchers found that the turbines tended to under-perform during stable regimes and over-perform in "convective" regimes, in which winds are highly turbulent but well mixed. More importantly, though, a computational analysis showed that these results were statistically robust, despite the various errors inherent in nacelle wind measurements.
Lundquist stressed that their performance results for stable and convective regimes are probably not generalizable, since other site-specific factors – such as changes of wind speed and direction with height – are important. Only cumulative studies will reveal trends that can help with effective wind-farm distribution. "We suspect that there is no simple rule for how atmospheric stability will impact power performance," said Lundquist.
The team believes the next step is to investigate how the errors in nacelle wind-speed measurements vary with atmospheric stability. "Beyond that, we would suggest repeating this analysis in a location with more comprehensive meteorological profiles so that we could quantify the error induced by nacelle winds directly, instead of estimating it with the [computational] method," said Lundquist.
The study is published in Environmental Research Letters.
About the author
Jon Cartwright is a freelance journalist based in Bristol, UK.