1. Can wind lidars measure turbulence? A. Sathe J. Mann J. Gottschall M. Courtney Acknowledgements – 1.EU FP6 Upwind Project 2.EU FP7 SafeWind Project

2. Risø DTU, Danmarks Tekniske Universitet What do we mean by turbulence? Estimation of Reynolds stress tensor, Why should we measure turbulence? 1.Design of wind turbines 2.Power curve of a wind turbine 3.Validation of wind profile models at greater heights, e.g. using momentum flux measurement The aim is to reduce the overall cost of wind energy! Many other applications in fields other than wind energy, e.g. Bridge design

3. Risø DTU, Danmarks Tekniske Universitet How can we measure turbulence? Sonic anemometers (current industry standard) Cup anemometers Hot wire anemometers So, what is the problem? Remote sensors that do not need a tower 1.Wind lidars 2.Sodars 3.Any other remote sensing instrument All these instruments need a meteorological mast They are very expensive, both onshore and particularly offshore They have a fixed structure to the ground, and hence, cannot be moved easily What is the alternative?

4. Risø DTU, Danmarks Tekniske Universitet What are the challenges for remote sensors? They should measure exactly what they are supposed to measure – Mean wind speeds, Turbulence etc. They should be cheaper than or at least as expensive as the current standard measurement systems What are wind lidars?

5. Risø DTU, Danmarks Tekniske Universitet What are the challenges for the measurement of turbulence? Which wind lidars are currently available? Windcube – a pulsed lidar developed by Leosphere ZephIR – a continuous-wave (CW) lidar developed by Qinetiq, Natural power

6. Risø DTU, Danmarks Tekniske Universitet How are the wind speeds retrieved? WindcubeZephIR

7. Risø DTU, Danmarks Tekniske Universitet How accurate is the turbulence measured by lidars? ZephIR

8. Risø DTU, Danmarks Tekniske Universitet Can we understand why lidars do not measure turbulence precisely? Point measurements

9. Risø DTU, Danmarks Tekniske Universitet Can we understand why lidars do not measure turbulence precisely? Assumptions in modelling The terrain is horizontally homogeneous The spatial structure of turbulence flow is well described by the spectral tens or model of Mann(1994)

10. Risø DTU, Danmarks Tekniske Universitet Can we understand why lidars do not measure turbulence precisely? Only neutral conditions are shown The model compares reasonably well with the measurements from both lidars Large systematic errors are observed for both lidars in turbulence measurements Model Measurements Ideal case Variation with wind direction ZephIRWindcube

11. Risø DTU, Danmarks Tekniske Universitet How can we measure turbulence using lidars? Directly using the variances of line-of-sight velocities Avoids averaging due to conical scanning Averaging only along the beam direction In principle, only 6 beams are required to measure Reynolds stress tensor

12. Risø DTU, Danmarks Tekniske Universitet How can we measure turbulence using lidars? How did we arrive at this configuration? Assumptions in modelling The terrain is horizontally homogeneous The spatial structure of turbulence flow is well described by the spectral tensor model of Mann(1994 )

13. Risø DTU, Danmarks Tekniske Universitet Comparing the 6 beam configuration with the conical scanning method For ZephIR, the improvement is not significant except for w variance The line-of-sight averaging effect is quite significant, especially for the ZephIR. For the Windcube, the variances using 6 beam is quite different from the conical scanning method Better results with the Windcube for the conical scanning method is due to luck (contribution of the cross components). The errors are quite large for different stability conditions using the conical scanning method as compared to the 6-beam method. Conical scanning 6-beam Conical scanning 6-beam Ideal case ZephIRWindcube

14. Risø DTU, Danmarks Tekniske Universitet Conclusions Using the conical scanning method, turbulence cannot be measured precisely due to large systematic errors arising from spatial averaging In principle, the 6 beam configuration reduces systematic errors due to conical averaging. Windcube errors are reduced significantly We need to somehow negate the line-of-sight averaging effect from both lidars, especially the ZephIR Future work 6-beam experiment using a short-range windscanner developed at Risø DTU Mann et.al(2010) developed a spatial filter model for the ZephIR to negate the line-of-sight averaging effect, but needs measurements. Averaged doppler spectra will be used instead of using the radial velocities derived from the individual doppler spectra.

15. Risø DTU, Danmarks Tekniske Universitet Thank you