nafnoise and fast noise module - wind.nrel.gov workshop... · • users guide • some validation...

1Operated for the U.S. Department of Energy Office of Energy Efficiency and Renewable Energy by Midwest Research Institute • Battelle

NAFNoise and FAST Noise Module

TurbSim & Design Codes Workshop

September 25, 2008

Pat Moriarty

2

Innovation for Our Energy Future National Renewable Energy Laboratory

Outline

• Noise mechanisms• Code Descriptions• Simple Models

– 2D Data Comparison– Full Turbine Comparison

• More complex Models– 2D data Comparison

• Inputs & Outputs• Future Work

3


Technical report

• NREL/TP 500-34478• http://www.nrel.gov/docs/fy04o

sti/34478.pdf • Theory manual• Users guide• Some validation• See also Moriarty 2004 ASME

and Moriarty 2005 CEAS for more model detail and validation

4


Noise Mechanisms Two-dimensional only

Turbulent boundary layer trailing-edge Blunt trailing edge

Laminar vortex sheddingTurbulent inflow

5


Code Descriptions• Mostly empirical models noise models

– Brooks, Pope, & Marcolini (1989) – airfoil self noise • Based on NACA 0012 measurements• Turbulent boundary layer trailing edge, separating flow, laminar boundary layer vortex

shedding, and trailing edge bluntness vortex shedding– Amiet (1975) – turbulent inflow noise

• Flat plate formulation• Leading edge noise source

• Models with more physics– Xfoil (Drela 2001)

• Calculates boundary layer properties directly• Any airfoil shape • Tripping included

– Full or simplified Guidati turbulent inflow noise (CEAS 2005)– TNO turbulent boundary layer trailing edge (CEAS 2005)

• NAFNoise– 2D airfoil only– Standalone code

• FAST Noise module– Full turbine– No Xfoil (only in alpha version)– No Guidati turbulent inflow noise module (simplified in alpha version)– No TNO trailing Edge noise module– BPM tip noise included (1989)– No mechanical or tower wake noise or atmospheric propagation

⎟⎟⎠

⎞⎜⎜⎝

⎛∝ 2

*5log10

e

sTBL r

bMSPL δ

6


Full Turbine Modeling

• Rotor broken into many different segments• Airfoil noise calculated for each segment

– Relative to observer position (directivity)– Two-dimensional– Quasi-steady

• Tip Noise only 3D model• Total rotor noise is the summation of all the

segments– Time and space dependent SPL– Fully averaged spectra (OASPL)

• NOT modeled– Mechanical noise – Tower influence– Atmospheric propagation

7


Tip Noise

NACA 0012 Airfoil, a=10.8º, U∞

=71.3 m/sBrooks Pope, & Marcolini 1989

NACA 0012

8


Aeroacoustic Measurements NLR Small Aeroacoustic Wind Tunnel

• 0.38 m x 0.51 m open jet (80 m/sec)• Inflow turbulence intensity

– 1% clean: 5-9 % with grid• 48-microphone phased array 0.6 m

from model• Seven different airfoils• Re ~ 200,000 – 1 million (chord =

0.23m)• α

= 0°- 8° (corrected for jet)• Tripped airfoils – 0.25 mm zigzag tape

@ 5% l.s. & 2% u.s.

9


Airfoil Validation

NACA 0012

⎟⎟⎠

⎞⎜⎜⎝

⎛∝ 2

5log10e

pLBL r

bMSPL

δ

α

= 4.4°Untripped Boundary Layer

10


Full Turbine Comparison

• CART Turbine at NWTC– 600 kW turbine– Constant speed (41.8 rpm)

• 95 m/s tip speed (M = 0.28)• Re ~ 5 million near tip

– mod LS(1)-0417 airfoils– TE thickness ~ 1% chord

• Upwind met measurements– 2.0 diameters west-northwest– Sonic anemometer at hub height

• 3 ground-level microphones– Downwind 58 m & 99 m– Rotor plane (south) 58 m– One-minute average spectra

LS(1)-0417

NACA 0012

11


Full Turbine Distance

~6 dB difference in measurement and prediction

12


Full Turbine Mean Wind Speed

58 m downwind, δ

= 3°

13


Xfoil Boundary Layer Prediction

• Sound intensity proportional to boundary layer thickness (δ* or δ)

• Brooks, Pope, and Marcolini – NACA 0012 measurements– empirical formulae

• Use Xfoil (Drela, 2001) instead– Allows for any airfoil shape

• Also used for TNO TBLTE model

⎟⎟⎠

⎞⎜⎜⎝

⎛∝ 2

*5log10

e

sTBL r

bMSPL δ

⎟⎟⎠

⎞⎜⎜⎝

⎛∝ 2

5log10e

pLBL r

bMSPL

δ

14


Turbulent Inflow Noise Guidati Model

• Based on method of Guidati et al. (1997)

• Produces correction to flat plate estimate

• Acoustic analogy of Möhring (1979)

– Steady, inviscid, and incompressible potential base flow

– Turbulent gusts modeled by discrete sheets

– Sheets carry vorticity– Vorticity determined from von

Karman spectrum

Grosse Wellenzahl

2Δζ2/λ ≈ 1 !!

LD/λ >> 1 ok

λ

λ

dBSPLSPLSPL GuidatiAmietInflow 10++=

( ) φ022 iHeMii eBHeB U×⋅−∇=+∇ ω

( ) ( )UU ×⋅−∇=∇⋅++∇ ωBiHeMBHeB 022 2

15



S822, α

= 0.0°, M=0.095, f = 2 kHz

16



M = 0.095 α

= 4.4°

5 min./ freq. (1.7 GHz) = Too Computationally Expensive !!!

17


Turbulent Inflow Noise Simplified Model – Straight Line Approximation

SD2030

fx63

S822

SG6043

S834

SH3055

Dec

ambe

red

airf

oils

Strouhal number 2πfC/U (-)

Diff

eren

cein

soun

dpo

wer

leve

lSP

L Airf

oil-S

PLFl

atpl

ate

(dB

)

0 25 50 75 100-30

-20

-10

0

Origin at (-5, 0)Slope is only unknown!

18


Turbulent Inflow Noise Simplified Model

Inflow turbulence noise indicator IT

Slop

epa

ram

eter

SL

0 0.1 0.20

0.1

0.2

0.3

0.4

0.5

0 0.1 0.2 0.3

Drel,1%Drel,10%

V10 0.25 0.5 0.75 1

%,%, yrelxrel DADIT ⋅+=

%10,%1, relrel DDIT +=

2317.5123.1 ITITSL ⋅+⋅=

2 5fCSPL SLUπ⎛ ⎞Δ = − +⎜ ⎟

⎝ ⎠

Optimization

19


Turbulent Inflow Noise Simplified Model – Further Validation

R A N D - 0 1 R A N D - 0 2 R A N D - 0 3 R A N D - 0 4 N A C A - 0 0 0 6T h i n p l a t e N A C A - 6 3 6 1 Strouhal number 2πfC/U (-)

Diff

eren

cein

soun

dpo

wer

leve

lSP

L Airf

oil-S

PLFl

atpl

ate

(dB

)

0 25 50 75 100-40

-30

-20

-10

0

Strouhal number 2πfC/U (-)

Diff

eren

cein

soun

dpo

wer

leve

lSP

L Airf

oil-S

PLFl

atpl

ate

(dB

)

0 25 50 75 100-40

-30

-20

-10

0

Airfoils with random combinations of thickness and leading edge radius

20


Turbulent Boundary Layer TE Noise TNO Model

• Originally developed by R. Parchen of TNO-TPD in Netherlands

• XFOIL for B. L. parameters (Cf , δ, and Uo )

• Coles’ law of the wall & law of the wake

• Prandtl’s mixing length hypothesis

• Empirical relations for mixing length and dissipation length scales

• 4 constants tuned to NLR data

( ) ( )∫∞

=0

1110

2 ,0,4

dkkPkcR

DS ωωπ

ω

( ) ( ) ( )∫∞

−−⎟⎟⎠

⎞⎜⎜⎝

⎛∂∂

+=

02

21,13122

2

2

12222

32

1

212

0312,,4,, dxekUkk

xUuL

kkkkkP xk

cm ωφωφρω

⎟⎟⎠

⎞⎜⎜⎝

⎛⎟⎟⎠

⎞⎜⎜⎝

⎛−⎟⎟

⎠

⎞⎜⎜⎝

⎛−++⎟⎟

⎠

⎞⎜⎜⎝

⎛= Bu

uUWBxuuU

νδ

κνκ

*

*02

**

1 ln121ln1

⎟⎠⎞

⎜⎝⎛=

δκδ085.0

tanh085.0 2xlm

32

2 1

1 2 1xc

L c x e δε δ

⎛ ⎞⎛ ⎞⎜ ⎟− −⎜ ⎟⎜ ⎟⎝ ⎠⎝ ⎠

⎛ ⎞⎜ ⎟= −⎜ ⎟⎜ ⎟⎝ ⎠

2

12

xUlmt ∂∂

=ν

21


Turbulent Boundary Layer TE Noise TNO vs. BPM

TNORe = 1.1 · 106

NACA 0012

BPMFrequency (Hz)

Soun

dpr

essu

rele

vel(

dB)

2000 4000 6000 800050

60

70

80

1000

0°, 4.5°, 9°, 12°, 16.5°

22


Turbulent Boundary Layer TE Noise TNO vs. BPM

Frequency (Hz)

Soun

dpr

essu

rele

vel(

dB)

2000 4000 6000 800050

60

70

80

1000

0°, 10°, 18°

TNORe = 1.0 · 106

BPM

23


Inputs (NAFNoise)

• Desired noise modules to use (e.g. TBL, LBL, etc..)• Boundary layer calculation (Xfoil vs. BPM)• TBLTE Noise calculation (BPM vs. TNO)• Turbulent Noise calculation (Amiet vs. full Guidati vs. simple Guidati)• Airfoil properties

– Airfoil shape (NACA automatically calculated)– Trip locations– Airfoil thickness (simple Guidati model)– Chord, Span– Trailing edge thickness & angle

• Incoming flow– AoA, Velocity– Turbulence intensity, length scale– Speed of sound, viscosity, and density

• Observer position

24


Inputs (FAST Noise Module)

• Desired noise modules to use (e.g. TBL, LBL, etc..)• Desired outputs

– Which noise source to use for OASPL• Speed of sound• Tip shape parameters

– Round or square– Spanwise lift distribution slope

• Observer position• Blade trailing edge thickness & angle• Many parameters passed automatically from FAST input files

25


Outputs - Spectra

NAFNoise - NACA 0012

M=0.207

FAST - AOC 15/50

Time Averaged

26


FAST Outputs – Full Turbine

Noise Footprint Rotor Plane – Doppler Amplification

AOC 15/50Time dependent

27


Code Observations

• Two-dimensional airfoil– Reasonable agreement with validation data depending on physics– Using XFOIL to calculate BL thickness improves TBL noise prediction– Using XFOIL to calculate BL thickness DOES NOT improve LBL-VS noise

prediction• Measurement and calculation sensitive to transition

– TNO adequate replacement for BPM• As accurate as BPM• More physical detail than BPM• A reasonable agreement with measured data• May improve with more sophisticated boundary layer information

– Guidati model accurate for turbulent inflow noise– Simplified turbulent inflow model as accurate as full boundary element

method• Full Turbine

– Absolute levels not accurately estimated– Some trends in SPL spectra predicted by the code, but not all (e.g. pitch

effect)

28


Future Work

• Incorporate more physical models into 3-D wind turbine code

• Validate against more full wind turbine test data• Improve turbulence length scales used in TNO• Improve BL calculation for TNO (CFD based)• Tower influence model• Atmospheric propagation

29Operated for the U.S. Department of Energy Office of Energy Efficiency and Renewable Energy by Midwest Research Institute • Battelle

Questions?

Pat Moriarty

+1 (303) 384 – [email protected]

nafnoise and fast noise module - wind.nrel.gov workshop... · • users guide • some validation...

Documents