On the impact of non-gaussian wind statistics on wind turbines - an experimental approach

Transcription

1 On the impact of non-gaussian wind statistics on wind turbines - an experimental approach Jannik Schottler, N. Reinke, A. Hölling, J. Peinke, M. Hölling ForWind, Center for Wind Energy Research University of Oldenburg, Germany jannik.schottler@forwind.de 1

2 Motivation source: youtube.com 2

3 Motivation source: youtube.com wind turbines are subjected to atmospheric turbulence! potential impact on......power output: grid fluctuations...torque: drive train failure...loads: lifetime [ Burton et al., 2001] [Carrasco et al., 2006; Sørensen et al., 2007] [Musial et al., 2007; Feng et al., 2013] 2

4 Motivation source: youtube.com wind turbines are subjected to atmospheric turbulence! potential impact on......power output: grid fluctuations...torque: drive train failure...loads: lifetime [ Burton et al., 2001] [Carrasco et al., 2006; Sørensen et al., 2007] [Musial et al., 2007; Feng et al., 2013] $ - cost of energy 2

5 Motivation Field measurements expensive limited availability uncontrolled boundary conditions 3

6 Motivation Field measurements expensive limited availability uncontrolled boundary conditions Numerics turbulence models computational costs validation? 3

7 Motivation Field measurements expensive limited availability uncontrolled boundary conditions Numerics turbulence models computational costs validation? Experiments inexpensive controlled environment tunable boundary conditions upscaling? 3

8 Motivation Field measurements expensive limited availability uncontrolled boundary conditions Numerics turbulence models computational costs validation? validation Experiments inexpensive controlled environment tunable boundary conditions upscaling? 3

9 Motivation Field measurements expensive limited availability uncontrolled boundary conditions Numerics turbulence models computational costs validation? validation Experiments inexpensive controlled environment tunable boundary conditions upscaling? 3

10 Describing turbulence industry standard for wind field description: 10 min mean values, turbulence intensity TI = u /hui 4

11 Describing turbulence industry standard for wind field description: 10 min mean values, turbulence intensity TI = u /hui 14 Ti = 19.7% velocity [m/s] time [min] 4

12 Describing turbulence industry standard for wind field description: 10 min mean values, turbulence intensity TI = u /hui 14 Ti = 19.7% 14 Ti = 19.7% velocity [m/s] velocity [m/s] time [min] time [min] 4

13 Increment statistics time series 5

14 Increment statistics time series 5

15 Increment statistics time series velocity increment u := u(t + ) u(t) 5

16 Increment statistics time series velocity increment u := u(t + ) u(t) time series of increments 0.2 uτ [ms 1 ] t[s] 5

17 Increment statistics time series velocity increment u := u(t + ) u(t) increment PDF 10 0 uτ [ms 1 ] time series of increments P (uτ )(a.u.) t[s] u τ /σ τ 5

18 Industry standards IEC ED3, 2005 wind turbines, design requirements turbulence: Mann model (1998) / Kaimal model (1972) 6

19 Industry standards IEC ED3, 2005 wind turbines, design requirements turbulence: Mann model (1998) / Kaimal model (1972) Increment PDF according to IEC-Norm (TurbSim, Kaimal model) 10 0 τ = 3 s Gauss P (uτ )(a.u.) u τ /σ τ 6

20 Industry standards IEC ED3, 2005 wind turbines, design requirements turbulence: Mann model (1998) / Kaimal model (1972) =3s offshore wind data non-gaussian, intermittent increments underestimation of extreme events [Wächter et al. 2012] 7

21 Industry standards IEC ED3, 2005 wind turbines, design requirements turbulence: Mann model (1998) / Kaimal model (1972) =3s offshore wind data non-gaussian, intermittent increments underestimation of extreme events once a year every 5 minutes! [Wächter et al. 2012] 7

22 Field data vs model datasets nearly equal acc. to mean + TI 8

23 Field data vs model datasets nearly equal acc. to mean + TI FINO Kaimal Gauss s strongly different regarding increment PDF intermittency not reflected correctly by Kaimal model p(uτ )(a.u.) s s 30s u τ /σ τ 60s 8

24 Field data vs model datasets nearly equal acc. to mean + TI FINO Kaimal Gauss s strongly different regarding increment PDF intermittency not reflected correctly by Kaimal model p(uτ )(a.u.) s s s Impact on wind turbines? u τ /σ τ 60s 8

25 Setup [Schottler et al., 2017] 9

26 Turbulence generation 10

27 Turbulence generation 1 m 0.8m 10

28 Turbulence generation 16 axes w/ stepper motors individually tunable defined, turbulent flows reproducible: 1 m time series statistics 0.8m 10

29 Turbulence generation 16 axes w/ stepper motors individually tunable defined, turbulent flows reproducible: time series statistics 1m 0.8m 10

30 Setup model wind turbine D=58cm active load control hot wire measurements upstream of rotor TSR = 7 turbine data: thrust (load cell) torque (generator current) power (electric) 11

31 Setup model wind turbine 1 m D=58cm active load control hot wire measurements upstream of rotor TSR = 7 0.8m turbine data: thrust (load cell) torque (generator current) power (electric) 11

32 Main idea Inflow A) Inflow B) 12

33 Main idea Inflow A) Inflow B) equal according to mean+ TI 12

34 Main idea Inflow A) Inflow B) equal according to mean+ TI Gaussian increments 12

35 Main idea Inflow A) Inflow B) equal according to mean+ TI Gaussian increments intermittent flow 12

36 Main idea Inflow A) Inflow B) equal according to mean+ TI Gaussian increments intermittent flow Does the turbine,see the difference? 12

37 Inflow A B hot wire data measured at rotor plane no turbine installed [Schottler et al. 2017] 13

38 Inflow A B hot wire data measured at rotor plane no turbine installed [Schottler et al. 2017] 13

39 Inflow A B hot wire data measured at rotor plane no turbine installed [Schottler et al. 2017] 13

40 Inflow 10 0 Gauss A (Gaussian) B (intermittent) p(uτ )(a.u.) ms 67ms ms (~rotor diameter) u τ /σ τ 2s 14

41 Inflow 10 0 Gauss A (Gaussian) B (intermittent) p(uτ )(a.u.) ms 67ms ms (~rotor diameter) u τ /σ τ 2s discrepancy between Gaussian assumption and intermittency reproduced in the lab! effect of properties beyond mean + TI (intermittency) isolated 14

42 Turbine reaction - thrust 25ms (~blade length) 67ms 80ms (~rotor diameter) 2s 15

43 Turbine reaction - thrust 25ms (~blade length) 67ms 80ms (~rotor diameter) 2s Gaussian inflow! Gaussian thrust 15

44 Turbine reaction - thrust 25ms (~blade length) 67ms 80ms (~rotor diameter) 2s Gaussian inflow intermittent inflow! Gaussian thrust! intermittent thrust 15

45 Turbine reaction - thrust 25ms (~blade length) 67ms 80ms (~rotor diameter) 2s Gaussian inflow intermittent inflow! Gaussian thrust! intermittent thrust no filtering of intermittency by the turbine 15

46 Turbine reaction - all quantities Gauss inflow thrust power torque 10 5 p(xτ ) / [a.u.] ms 67ms 80ms (~rotor diameter) x τ / σ τ 2s 16

47 Turbine reaction - all quantities Gauss inflow thrust power torque 10 5 p(xτ ) / [a.u.] ms 67ms 80ms (~rotor diameter) x τ / σ τ 2s Intermittent characteristics remain present in turbine data! 16

48 Impact on wind turbine One second data, multi MW nearshore turbine [P. Milan] [Milan et al. 2013] 17

49 Impact on wind turbine One second data, multi MW nearshore turbine [P. Milan] [Milan et al. 2013] 17

50 Impact on wind turbine One second data, multi MW nearshore turbine [P. Milan] [Milan et al. 2013] 17

51 Thank you for your attention! Funded by the Reiner Lemoine Stiftung Further information: 18

52 Load Control TSR [-] u [m/s] w [Hz] P [W] cp [%] 19

53 Load Control TSR [-] u [m/s] w [Hz] P [W] cp [%] 19