On predicting wind turbine noise and amplitude modulation using Amiet s theory

Size: px

Start display at page:

Download "On predicting wind turbine noise and amplitude modulation using Amiet s theory"

Adrian Chambers
5 years ago
Views:

1 On predicting wind turbine noise and amplitude modulation using Amiet s theory Samuel Sinayoko 1 Jeremy Hurault 2 1 University of Southampton, ISVR, UK 2 Vestas, Isle of Wight, UK Wind Turbine Noise 2015, Glasgow, 22 April / 20

2 Introduction Motivation Long term objective: Develop an efficient numerical tool for designing low noise wind turbine blades. 2 / 20

3 Introduction Strategy 1 Propagation: develop and validate noise prediction code based on Amiet s theory for: time averaged spectra amplitude modulation trailing edge noise and inflow noise 2 Source: assess existing semi-analytical models against experimental data 3 Blade optimization Current study focuses on first step 3 / 20

4 Paper key objectives Review Amiet s theory and hightlight key challenges Present our first comparison against measured data Showcase the key advantages of the theory 4 / 20

5 Outline 1 Amiet s theory and prediction methodology 2 Results 3 Conclusions 5 / 20

6 Outline 1 Amiet s theory and prediction methodology 2 Results 3 Conclusions 6 / 20

7 Amiet s theory for solated airfoils Amiet 1974, 1975, 1976 Leading edge noise S pp(ω, x) = B Ψ LE (K, k c ) 2 Φ ww (K) 7 / 20

8 Amiet s theory for solated airfoils Amiet 1974, 1975, 1976 Leading edge noise Trailing edge noise S pp(ω, x) = B Ψ LE (K, k c ) 2 Φ ww (K) S pp(ω, x) = A Ψ T E (K, k c ) 2 l S (K) S qq (ω) Key pitfalls 1 Common definitions of Ψ T E 2 become singular for certain observer angles. 2 Frequency spectrum S qq (ω) is wall pressure spectrum. 7 / 20

9 Amiet s theory for rotating blades Rotating blade element Split blade into blade elements of size l S. Observer x α γ Ω θ z Flow y 8 / 20

10 Amiet s theory for rotating blades Split blade into segments larger than l S For each segment, use isolated airfoil theory to estimate the instantaneous PSD S pp (ω, t, x) ωs pp (ω, t, x) = ω S pp(ω, t, x ) ω = D ω S pp (ω, t, x) = DS pp(ω, t, x ) Integrate over one period T = 2π/Ω T S pp (ω) = 1 S pp (ω, t)dt T 0 dt = Ddt S pp (ω, x) = 1 T T 0 D 2 S pp (ω, t, x )dt Assume blade in rectilinear motion to derive Doppler shift D: D = 1 M BO σ/(1 + M FB σ) 9 / 20

11 Key pitfalls and remarks 1 Amiet s theory provides both averaged PSD and instantaneous PSD. 2 Doppler shift exponent is 1 for instantaneous PSD and 2 for stationary PSD S pp (ω, t, x) = DS pp(ω, t, x ) S pp (ω, x) = 1 T T 0 D 2 S pp (ω, t, x )dt 3 Pitfall: in blade frame, observer position x is relative to the retarded source position x e = x s + M FO c 0 T e Validation and details in Sinayoko, Kingan and Agarwal (2014). 10 / 20

12 Aerodynamics Chordwise Mach number M ch and angle of attack AoA needed for each blade segment. 1 Use Blade Element Momentum code PYRO to get M ch and AoA given wind speed and rotation speed; 2 Use Xfoil or measurements to get lift and drag coefficients needed by PYRO; 3 Use Xfoil to get boundary layer thickness δ needed for S qq model. 11 / 20

13 Outline 1 Amiet s theory and prediction methodology 2 Results 3 Conclusions 12 / 20

14 Test case Blade geometry: DAN-AERO (Masden et al 2010) Tower: 58 m Rotor radius: 40 m Wind speed: 8.5 m/s Rotation speed: RPM Absorption coefficients: ISO with 10 C and 70% humidity 13 / 20

15 Results Sound Power Level SPWL [dba re ] 5 db Measured (102.5 dba) Trailing edge noise (101.2 dba) Third octave frequency bands [Hz] 14 / 20

16 Results Sound Power Level SPWL [dba re ] 5 db Measured (102.5 dba) Trailing edge noise (101.2 dba) Inflow noise {40m, 12.5%} (95.3 dba) Total (102.2 dba) Third octave frequency bands [Hz] 14 / 20

17 Effect of integral length scale Sound Power Level SPWL [dba re ] Measured (102.5 dba) Inflow noise {20m, 12.5%} (96.7 dba) Total noise {20m, 12.5%} (102.5 dba) Third octave frequency bands [Hz] 15 / 20

18 Effect of integral length scale Sound Power Level SPWL [dba re ] Measured (102.5 dba) Inflow noise {20m, 12.5%} (96.7 dba) Total noise {20m, 12.5%} (102.5 dba) Inflow noise {40m, 12.5%} (94.7 dba) Total noise {40m, 12.5%} (102.1 dba) Third octave frequency bands [Hz] 15 / 20

19 Effect of integral length scale Sound Power Level SPWL [dba re ] Measured (102.5 dba) Inflow noise {20m, 12.5%} (96.7 dba) Total noise {20m, 12.5%} (102.5 dba) Inflow noise {40m, 12.5%} (94.7 dba) Total noise {40m, 12.5%} (102.1 dba) Inflow noise {80m, 12.5%} (93.5 dba) Total noise {80m, 12.5%} (101.9 dba) Third octave frequency bands [Hz] 15 / 20

20 Effect of integral length scale Sound Power Level SPWL [dba re ] Measured (102.5 dba) Inflow noise {20m, 12.5%} (96.7 dba) Total noise {20m, 12.5%} (102.5 dba) Inflow noise {40m, 12.5%} (94.7 dba) Total noise {40m, 12.5%} (102.1 dba) Inflow noise {80m, 12.5%} (93.5 dba) Total noise {80m, 12.5%} (101.9 dba) Inflow noise {160m, 12.5%} (92.7 dba) Total noise {160m, 12.5%} (101.8 dba) Third octave frequency bands [Hz] 15 / 20

21 Effect of turbulence intensity scale Sound Power Level SPWL [dba re ] Measured (102.5 dba) Inflow noise {40m, 5.0%} (87.4 dba) Total noise {40m, 5.0%} (101.4 dba) Third octave frequency bands [Hz] 16 / 20

22 Effect of turbulence intensity scale Sound Power Level SPWL [dba re ] Measured (102.5 dba) Inflow noise {40m, 5.0%} (87.4 dba) Total noise {40m, 5.0%} (101.4 dba) Inflow noise {40m, 12.5%} (95.3 dba) Total noise {40m, 12.5%} (102.2 dba) Third octave frequency bands [Hz] 16 / 20

23 Effect of turbulence intensity scale Sound Power Level SPWL [dba re ] Measured (102.5 dba) Inflow noise {40m, 5.0%} (87.4 dba) Total noise {40m, 5.0%} (101.4 dba) Inflow noise {40m, 12.5%} (95.3 dba) Total noise {40m, 12.5%} (102.2 dba) Inflow noise {40m, 20.0%} (99.4 dba) Total noise {40m, 20.0%} (103.4 dba) Third octave frequency bands [Hz] 16 / 20

24 Amplitude modulation bladed rotor 42 Horizontal direction in rotor plane y/h Flow direction z/h / 20

25 Amplitude modulation bladed rotor 21 Horizontal direction in rotor plane y/h Flow direction z/h / 20

26 Outline 1 Amiet s theory and prediction methodology 2 Results 3 Conclusions 18 / 20

27 Conclusions Achievements 1 Reviewed Amiet s theory 2 Code for predicting wind turbine noise using Amiet s theory 3 Inflow noise dominates at low frequencies 4 Total noise is: Sensitive to axial turbulence intensity level (+2 dba from 5% to 20%) Insensititive to integral length scale (-0.7 dba from 20m to 160m) 19 / 20

28 Conclusions Achievements 1 Reviewed Amiet s theory 2 Code for predicting wind turbine noise using Amiet s theory 3 Inflow noise dominates at low frequencies 4 Total noise is: Sensitive to axial turbulence intensity level (+2 dba from 5% to 20%) Insensititive to integral length scale (-0.7 dba from 20m to 160m) Future work Investigate alternative trailing edge frequency spectra; Develop new source modelling techniques if needed; Improve AM predictions in rotor plane; Optimize a wind turbine blade for low noise and/or low AM. 19 / 20

29 Acknowledgements Also: Anurag Agarwal, Mitsubishi Heavy Industries, Phil Joseph. 20 / 20

30 Acknowledgements Also: Anurag Agarwal, Mitsubishi Heavy Industries, Phil Joseph. Further information 20 / 20

31 Acknowledgements Also: Anurag Agarwal, Mitsubishi Heavy Industries, Phil Joseph. Further information 20 / 20

Trailing edge noise prediction for rotating serrated blades

Trailing edge noise prediction for rotating serrated blades Samuel Sinayoko 1 Mahdi Azarpeyvand 2 Benshuai Lyu 3 1 University of Southampton, UK 2 University of Bristol, UK 3 University of Cambridge, UK