1. Turbulence and Heterogeneous Wind
Conditions in the Field of Wind Energy
MSc Thesis Presentation
Robin Keus
Wednesday, 17 May 2017
Supervisors:
Dr. Ir. W.A.A.M. Bierbooms, TU Delft
Drs. J. P. Coelingh, Vattenfall

2. Conditions under which turbine performance is guaranteed should be improved
Power curve only guaranteed in 30% of the operating time
Need turbulence normalisation!
Source: Windpower

3. Content
Background
Methodology
Results
Conclusions

4. Background

5. Power Curve Verification
Monitoring wind turbine performance
Operator strives for best turbine performance
Source: Vestas

6. Monitoring wind conditions
IEC-certified met-mast
Lidar
Spinner anemometers

7. Met-mast Cup anemometers and wind vanes Limited in height Foundation
Building permits
Costs
Source: SgurrEnergy

8. Lidar Aerosols Doppler shift Flexible Range Nacelle-based (optional)
Five-beam Lidar:
Shear
Performance
Source: Unitte

9. Sonic spinner anemometers
Sound pulses
Mounted on the spinner
Undisturbed measurements
Calibrate spinner design
Source: Pedersen et al. [2015]
Source: iwrpressedienst

10. Innovative measurement devices
Still uncertain measurement accuracy
Turbulence measurements of Lidar
Different method
Volume averaging
Require more research!

11. Power curves depend on turbulence
TI effect on power curve
IEC standard turbulence normalisation procedure
Reduce power curve uncertainty
4 m/s: 200 kW -270 kW
5 m/s: 470 kW
6 m/s: 840 kW +340 kW
Vavg=5 m/s: 503 kW > 470 kW
𝜎=0.8 𝑚/𝑠
𝑇𝐼= 𝜎 𝑉 𝑎𝑣𝑔 =0.16
Power (kW)
Wind speed (m/s)
Source: Kaiser

12. Homogeneous wind conditions
LOS 2 experiences the same wind conditions as LOS 1
LOS 2
LOS 1

13. Heterogeneous wind conditions
LOS 2 experiences different wind conditions than LOS 1
LOS 2
LOS 1

14. Research Main topic: Objectives:
To what extent can a (nacelle-based) Lidar measure YM and TI accurately compared to other measurement devices and what is the effect of heterogeneity on its measurements, as well as turbulence normalisation on turbine power performance?
Objectives:
Difference in PCV between Lidar and met-mast?
Accuracy in measuring wind conditions and PCV of five-beam Lidar compared to other Lidar, spinner anemometers and met-mast?
Effect of turbulence normalisation on the turbine performance and AEP? Effect on the scatter around the power curve?
Can heterogeneity be quantified? Effect on Lidar measurements?

15. Methodology

16. Overview of research Prinses Alexia: Nørrekær Enge:
Experience and knowledge with Lidar measurements
Nørrekær Enge:
Comparison of met-mast, Lidars and spinner anemometers
Turbulence normalisation
Validation heterogeneity

17. Nørrekær Enge Flat site Nacelle-based Lidars Spinner anemometers
Avent 5-beam Lidar
ZephIR Lidar
Spinner anemometers
Met-mast close to turbine 4
13 2.3MW Siemens turbines

18. Simulations for turbulence norm
IEC procedure
PC uncertainty
Site-specific TI for PC
Steps:
Determine site TI
Determine zero TI power curve
Normalise power curve to site TI
Zero TI PC – Measured PC (%)

19. Simulations for heterogeneity
Synthetic wind field
Approach the heterogeneity effect on Lidar measurements
Possibility to correct for heterogeneity
Linear
Non-linear α x α x2 x1

20. Results

21. Wind speed comparison
Good comparisons between Lidars and spinner anemometers with met-mast
Slight overestimation of Avent Lidar
Avent Lidar
ZephIR Lidar
ROMO spinner

22. Yaw misalignment comparison
Spinner anemometer shows good comparison with met-mast
Lidars show poor accuracy
Avent Lidar
ZephIR Lidar
ROMO spinner

23. Turbulence intensity comparison
Unfiltered and filtered measurements
Avent Lidar compared best with met-mast
ZephIR underestimates TI
Spinner average TI is close to unity, but high scatter
Avent Lidar
ZephIR Lidar
ROMO spinner

24. PCV Nørrekær Enge Better than guarantees
Underestimation in AEP by Avent Lidar compared to met-mast (-3.5%)
Overestimation by spinner (+2%)
P – Pguaranteed(%)

25. Turbulence normalisation
Simulations
Effect of TI
Improvement in scatter
Zero TI PC – Measured PC (%)
σPower/Prated

26. Turbulence normalisation
Measurements
Effect of TI
No improvement in scatter
Other factors influence power output
Zero TI PC – Measured PC (%)
σPower/Prated

27. Heterogeneity Simulations: Measurements:
Only significant heterogeneity in wind field affects Lidar
Possible to quantify heterogeneity
In case of linear heterogeneity a correction is feasible
Measurements:
No significant impact of heterogeneity on Lidar measurements
LOS 2
LOS 1

28. Conclusions

29. Measurement devices Good accuracy in measuring wind speed
Poor accuracy of Lidars in YM
Spinner anemometer shows better correlation with met-mast regarding YM
Avent Lidar TI shows some correlation with met-mast
ZephIR and spinner anemometer show low accuracy in TI

30. PCV Underestimation of PC by Avent Lidar
Spinner anemometer overestimated PC

31. Turbulence normalisation
Simulations showed TI effects clearly
Improvement in scatter and uncertainty
Measurements susceptible to other factors besides TI
Possible to determine site-specific PC based on TI

32. Heterogeneity Possible to quantify heterogeneity
Only significant heterogeneous wind conditions lead to uncertainty in measurements
Corrections sensitive to wind field heterogeneity

33. Recommendations Perform measurements in complex terrain
Investigate turbulence normalisation by isolating effects of TI on power curve
Improve heterogeneity simulations
Change reference point wind speed to wind conditions experienced by a turbine rotor

34. Thank you for your attention!

35. Research objectives
1. What is the difference in PCV between a ground- or nacelle-based Lidar and a met-mast? 2. How does accuracy in measuring wind conditions and PCV of a five-beam Lidar compare to other nacelle-based Lidar, spinner anemometers and IEC-compliant met-masts? 3. What is the effect of turbulence normalisation on the turbine performance and AEP? And how does it affect the scatter around the power curve? 4. Can heterogeneity be quantified? And what is its effect on Lidar measurements?

36. Lidar measuring volumes

37. Lidar derivations

38. Site layout Prinses Alexia Ground-based Lidar
36 3.4MW Senvion turbines

39. Data synchronisation

40. Results wind speed validation
>3.5 m/s
Lidars vs spinner
Lidars

41. Results YM validation
Avent vs ZephIR
Lidars vs spinner

42. Results TI validation

43. Results PCV

44. Results inner-outer conditions
Prinses Alexia
Met-mast
Lidar

45. Turbulence normalisation

46. TI normalisation procedure
Avent ZephIR Spinner
Met-mast

47. TI normalisation zero TI
Avent ZephIR Spinner
Met-mast

48. Heterogeneity Non-linear adjustment for heterogeneity
Linear simulations
Non-linear simulations