the effect of blade thickness and angle of attack on ...€¦ · • review of existing studies •...

1

The Effect of Blade Thickness and Angle of Attack on Broadband

Fan NoiseStewart Glegg

Florida Atlantic Universityand

William DevenportVirginia Tech

Work Supported by ONR, Prog. Manager: Dr. Ron Joslin

Authors would like to thank Ed Envia for providing additional slides

2

Sources of Broadband Fan Noise

Rotor Alone Rotor/Stator Interaction

• Trailing Edge Noise• Blade Tip Gap Noise• Duct Wall BL Inflow Noise

• Inflow Noise Rotor TurbulenceRotor Tip VorticesDuct Wall BL

• Self Noise

AIAA Papers No: 93-4402, 98-2316

3

Comparison of Predicted R/S Interaction Noise and Measurements

for the SDT Fan Rig (Envia et al, 2008)

Envia et al AIAA paper no 2008-2991

4

Prediction of R/S Interaction Noise with Rotor Alone Noise Added

for the SDT Fan Rig (Envia et al, 2008)

Envia et al AIAA paper no 2008-2991

5

Outline of Talk• Rotor Alone Noise• Predicting Trailing Edge Noise Sources • Blade Surface Pressure Spectra and AoA• Recent Developments• Rotor Stator Interaction Noise• Review of Existing Studies• Physics of Leading Edge Noise AoA and Thickness

Effects• Wind Tunnel Measurements Of Leading Edge Noise• Conclusions

6

Rotor Alone Noise

7

Sources of Rotor Alone Noise

• Tip Flows – may dominate for low speed fans with large tip gaps and low AR

• Wall BL Inflow Noise -typically BL is very thin and source is not important

• Trailing Edge Noise -assumed to dominate for moderate AR blades with small tip gaps

AIAA Paper No: 98-2316

8

Predicting TE Noise

9

Trailing Edge NoiseSingle Blade

BL Turbulence

Sharp Trailing Edge• Scattering of BL pressure fluctuations causes propagating acoustic waves•Kutta condition is assumed at the trailing edge

(Amiet (1977), Brooks(1984), Howe (1998)

Sound Waves

10

Trailing Edge NoiseBlade Row

BL Turbulence

Sound Waves

In a blade row there is acoustic scattering from multiple blades

AIAA Jnl Vol 36(9) 1998

11

TE Noise Prediction

• Blade is broken down into sections• Surface pressure wavenumber spectrum defined at each blade section• Coupled Duct/ 3D TE Scattering Response defined for each blade section • Spanwise wavenumber varies with radius• Assumes small spanwise correlation lengthscale

Blade Surface Pressure

Blade Row Scattering

Duct Mode Amplitudes


12

Properties of the Surface Pressure

Assume that the boundary layer on a single blade has the same surface pressure spectrum as a blade in a cascade if U and δ* are the same


13

Deducing the Surface Pressure Spectrum

for a given U and δ*


Blade Row Scattering

Duct Mode Amplitudes


TE Scattering

Far Field Sound

Single Blade

Blade Row

=*

* =AIAA Jnl Vol 36(9) 1998

14

0 5 10 15 20 25 30 35 4020

30

40

50

60

70

frequency kHz

Sour

ce P

ower

Spe

ctru

m d

B re

20e

-6Pa

/sqr

t(Hz)

blade radius

Source Spectra at Different Radii

Constant AoA of 7 deg


15

Source Spectra at Different Radii

0 5 10 15 20 25 30 35 4020

30

40

50

60

70

frequency kHz

Sour

ce P

ower

Spe

ctru

m d

B re

20e

-6Pa

/sqr

t(Hz)

blade radius

AoA is 8 deg at hub and 4 deg at tip


16

Effect of Increasing AoA on Sound Power Output

0 5 10 15 2050

55

60

65

70

75

80

85

90

frequency kHz

Soun

d Po

wer

Spe

ctru

m d

B re

1e-

12W

/Hz Increasing AoA9

8

765

AoA constant from hub to tip


17

Effect of AoA on Total Sound Power

-2 -1 0 1 2100

105

110

115

120

125

130

angle of attack re case A and B

Soun

d Po

wer

Out

put

Sound power increases as ~2.4 dB per degree


18

Recent Developments

In collaboration with Bruce Morin, Oliver Attasi and Ramon Reba

AIAA paper no 2008-2993

19

RANS

Using RANSto get surface pressure spectrum

Local LinearizedUnsteady Flow

Both models give tke

Mean FlowDissipation Scale

Surface pressurespectrum


20

102

103

104

90

95

100

105

110

115

F (Hz)

PWL

(dB)

SDT Test Case Nominal VAN

Dnstrm DataDnstrm Pred RMGDnstrm Pred Brooks

SDT Test ComparisonRotor Alone Noise

The downstream sound power radiated from the fan blade versus frequency. The blue curve corresponds to the measured data, the green curve is the prediction based on the current model and the red curve is the prediction using the Brooks et al (1989) self noise database.

Data

Brooks Input

RANS Input


21

Rotor Stator Interaction Noise

22

Rotor/Stator Interaction Noise• Hanson and Horan (AIAA 98-

2319)• Glegg and Walker (AIAA 99-

1888)• Evers and Peake (JFM v.43,

2002)• Nallasamy and Envia (JSV v.

283, 2005)• Cheong et al (JASA 119, 2006)• Attasi and Vinogradov (AIAA

2007-3691)

Uu.n

s

b

23

Rotor/Stator Interaction NoiseEffect of AoA

Evers and Peake(JFM v.43, 2002)

Gave a high frequency analysis of loaded cascade,

Showed effects of AoA for tonal noise from blade wakes was large ~10dB

However only a small effect ~2dB was found on broadband noise for a homogeneous turbulent inflow

Uu.n

s

b

24

Rotor/Stator Interaction NoiseFully Loaded Rotor

Attasi and Vinogradov (AIAA 2007-3691)

• Gave a complete analysis of loaded set of stator vanes in a circular duct

• Showed up to 3-4dB of real blade effects on broadband noise• Predictions were similar to flat plate results at high

frequencies• Verified predictions using experimental data

Concluded that real blade effects were small, but what is the physics??

25

Leading Edge Noise

26

Phase of a Vortical Gust


27

Streamwise Component of Goldstein’s Gust


28

Time Domain Methods

Works well for problems which can be reduced to 2D

Point vortex

• Calculate the unsteady loading from a blade vortex interaction and hence the noise

• Arbitrary gust is modeled by multiple vortices

Glegg and Devenport JSV 2008

29

Unsteady Loading from a BVI

-5 -4 -3 -2 -1 0 1 2 3 4 5-2

-1

0

1

210 deg

5 deg

0 deg

Ut/a

unst

eady

lift

-5 -4 -3 -2 -1 0 1 2 3 4 5-0.5

0

0.5

0 deg 5 deg

10 deg

Ut/a

unst

eady

dra

gUnsteady Lift

Unsteady Drag

Time


30

Response to Step Upwash Gust

Step Upwash Gust


U

F(t)

31

Unsteady Loading from a Step Gust for 0,5,10 deg AoA

-10 -8 -6 -4 -2 0 2 4 6 8 10

-5

0

5

Ut/a

unst

eady

lift

5

drag

Net Unsteady Loading

Uns

tead

y Lo

adin

g


Time

32

-10 -8 -6 -4 -2 0 2 4 6 8 10

-5

0

5

Ut/a

unst

eady

lift

5

agUnsteady Loading from a Step Gust

for thickness to chord ratios 0.001,0.06,0.15

Net Unsteady Loading

Uns

tead

y Lo

adin

g


Time

33-3 -2 -1 0-3

-2

-1

0

1

2

3

3

-3 -2 -1 0 1 2 3-3

-2

-1

0

1

2

3

(a) Flow at an angle of attack (b) Flow at zero a

Overlay of (b) onto (a) rotatedclockwise by 2α deg

(a) Streamlines at an AoA α deg (b) Streamlines at zero AoA

Conformal Mapping

For 2D incompressible flow the airfoil can be mapped onto a circle

Unsteady loading depends on the distance of the vortex from the singular point

Location of the singular point depends on airfoil thickness


34

Conformal Mapping

-3 -2 -1 0 1 2 3-3

-2

-1

0

1

2

3

-3 -2 -1 0 1 2 3-3

-2

-1

0

1

2

3

-3 -2 -1 0 1 2 3-3

-2

-1

0

1

2

3

(a) Flow at an angle of attack(b) Flow at zero angle of attack

Overlay of (b) onto (a) rotatedclockwise by 2α deg

(a) Streamlines at an AoA α deg (b) Streamlines at zero AoA


After rotation streamlines are overlaid

Increased levels from vortices in upper plane cancels reduced level from vortices in lower plane

35

Experimental Verification

Experiments carried out at Virginia Tech in collaboration with William

Devenport and Joshua Staubs

36

VT Stability Wind Tunnel

• 6'x6'x24' test section• 15' dia., 600 h.p. fan produces flow speeds to 80m/s (Re > 5000000 per meter)

RemoveableTest section

Air exchange tower

Control room

0.5MW Drive and Fan

Anechoic system installed in control room area

AIAA-2008-2911

37

From Inside

012345678

0 5000 10000 15000 20000 25000

Frequency (Hz)

Loss

(dB

)AIAA paper no 2008-3018

38

Turbulence Grid

0 2 4 6 8 10 12-0.2

0

0.2

0.4

0.6

0.8

1Transverse corrrelation functions, large grid, 30m/s

Separation (inches)

Ruu(y)Rw w (y)Ruu(z)Rvv(z)

von Karman

10-1

100

101

102

103

10-6

10-5

10-4

10-3

10-2

10-1

100

Longitudinal spectrum functions, large grid, 30m/s lf=1.28

k/ke

Euu Euu (single wire)

von Karman modified von Karman

L = 0.082m

• 5cm bars on 30cm centers• 69.4% open area ratio• In contraction (area 32% larger than test section)• 19.5 mesh sizes ahead of airfoils

u'/U = 3.8%v'/U = 3.9%w'/U = 4.0%


39

Measurement locations and

conditions

1.83

Foam transition

Kevlar acoustic window

1.8m

α

c α o

NACA 0012 0.2m -12,-8,-4,0,4,8,12

NACA 0015 0.6m 0,2,4,6,8,10,12

NACA 0012 0.9m 0,2,4,6,8,10,12

DU96 0.9m 0,2,4,6,8,10

S831 0.9m 0,2,4,6,8,10

• All measurements at 30m/s (additional at 40m/s not shown)


40

0.2-m NACA 0012Effects of Angle of Attack

L

102

103

20

30

40

50

60

70

80

90

SP

L 1/3

(dB

)

2.1 21Reduced Frequency ωr

0o

4o

8o

12o

Frequency (Hz)

300 400 500 60059

60

61

62

63

64

65

66

67

68

69

SP

L 1/3

(dB

)


WJD1

Diapositiva 40

WJD1 Sync animationchange Gershfeld to HoweWilliam Devenport; 06/05/2008

41

0.2-m NACA 0012

Reduced Frequency

Spp

(dB

)

Moreau et al. (2005)NACA 0012 h/c=1.3 (jet)L=6.6%c u'/U=5% U=40m/s

2.3 7.8 2320

30

40

50

Reduced Frequency

NACA 0012, 0.2-m h/c=9L=40%c u'/U=3.9% U=30m/s

1.8o

3.6o

5.4o

0oαeff

0o

4o

8o

12o

4 6 8 10 20 3025

35

45

55

SP

L (d

B)

12o0oαeff


421 10 100

30

35

40

45

50

55

60

65

70

75

80

SP

L 1/3

(dB

)

Noise Calculations

• Glegg, Devenport and Staubs (2008) Theory.

∫∞

∞−

−Ω= 2

2

3

22332

02

32

2),(ˆ

)0,,/()( dkR

kLkUcrURSpp

ωωπωω

Drift coordinates Wavenumber spaceω

k2

Von Karman

Sound from a flat plate

Reduced Frequency ωr

AmietAnalytical Solution

Panel Method

43

0.2-m NACA 0012Effects of Angle of Attack with predictions

L

102

103

20

30

40

50

60

70

80

90

SP

L 1/3

(dB

)


0o

4o

8o

12o

Frequency (Hz)

Flat plate


WJD2

Diapositiva 43

WJD2 Sync animationchange Gershfeld to HoweWilliam Devenport; 06/05/2008

44

0.6-m NACA 0015Effects of Angle of Attack

L

NACA 0015, 0.6-m h/c=3L=13%c u'/U=3.9% U=30m/s

102

103

20

30

40

50

60

70

80

90

SP

L 1/3

(dB

)

0o

2o

4o

6o

o8

10o

o12

Frequency (Hz)


0.203-m 0012, 12o

0.203-m 0012, 0o


45

0.6-m NACA 0015Effects of Angle of Attack with predictions

L

102

103

20

30

40

50

60

70

80

90

SP

L 1/3

(dB

)

0o

2o

4o

6o

o8

10o

o12

Frequency (Hz)



10 dB

Flat Plate

46

Conclusions• Rotor alone noise and rotor stator interaction noise

are equally important• Trailing edge noise can be predicted from

calculations of the blade surface pressure spectrum• Blade stall dramatically increases trailing edge noise• Angle of attack effects on blades in isotropic

turbulence are small (but not necessarily for anisotropic turbulence and asymmetric blades)

• Blade thickness can significantly reduce high frequency noise (but the details of the blade shape are important)

the effect of blade thickness and angle of attack on ...€¦ · • review of existing studies •...

Documents