experimental study of broadband trailing edge noise of a ...experimental study of broadband trailing...

PhD Defense

Experimental study of broadband trailing edge noise

of a linear cascade and its reduction

with passive devices

Arthur Finez

LMFA/Ecole Centrale de Lyon

Thursday 10th May 2012

A. Finez (LMFA/ECL) PhD Defense 10/05/2012 1 / 37

Introduction

IntroductionNoise reduction

Why aircraft noise is to be reduced ?

Orly airport noise disturbance map.: LDEN>70, : LDEN>65, : LDEN>55. [ACNUSA]

Noise problems nearby airports.


Introduction

IntroductionNoise reduction

Why aircraft noise is to be reduced ?

Orly airport noise disturbance map.: LDEN>70, : LDEN>65, : LDEN>55. [ACNUSA]

Noise problems nearby airports.

Fan noise contribution

Engine of an A380 aircraft

Take-off and landing :⋍ 40%.


Introduction

IntroductionDiscriminating fan noise sources

Fan noise composition

Tonal noise, significantly reduced in the past decades,

Broadband noise, still to be reduced.

Many broadband noise sources near the rotor

Figure: Axial flow compressor rotorflow phenomena [AGARD-AG-328].


Introduction






- Turbulence - Leading Edge (LE)interaction noise,



Introduction







- Turbulent Boundary Layer (TBL) -Trailing Edge (TE) interaction noise,



Introduction








- Blade Tip noise,



Introduction








- Blade Tip noise,

- Turbulence - Shock Surfaceinteraction noise,



Introduction








- Blade Tip noise,


- Stall noise ...



Introduction








- Blade Tip noise,


- Stall noise ...Figure: Axial flow compressor rotorflow phenomena [AGARD-AG-328].


Introduction

IntroductionCascade Effect

Blades are not isolated in a fan !

- Reflections,

- Resonances,

- Duct propagation ...


Introduction



- Reflections,

- Resonances,


Literature : few studies on cascade trailing edge noise

- Analytical modeling : Howe 92 ; Glegg 98 ;

- Experiments : Parker 66, Sabah & Roger 01.7

6

5

4

3

2

1

β1

α1χ

M

θU1

U2


Introduction



- Reflections,

- Resonances,


Literature : few studies on cascade trailing edge noise

- Analytical modeling : Howe 92 ; Glegg 98 ;

- Experiments : Parker 66, Sabah & Roger 01.7

6

5

4

3

2

1

β1

α1χ

M

θU1

U2

Need for cascade noise experiment & model validation.


Introduction

IntroductionObjectives

This study aims at

- Designing an aeroacoustic cascade set-up,

- Measuring cascade TBL-TE noise with various U1, α1...

- Assessing the extent of the cascade effect,

- Validating existing cascade noise models,

- Reducing cascade noise.


Introduction

Outline

1 Cascade ExperimentSet-upAerodynamic LoadingCascade Acoustics

2 Analytical Model ValidationInput Data for ModelsAmiet’s Isolated Airfoil Noise ModelGlegg’s Cascade Noise Model

3 Cascade Noise ReductionSet-upAcoustic ResultsPIV resultsNoise reduction mechanisms


Cascade Experiment Set-up

Outline






ExperimentSet-up of a unique linear cascade

Linear cascade of 7 NACA 6512-10 airfoils, studied by Emery 1.

7

6

5

4

3

2

1

β1

α1χ

M

θU1

U2

Chord c 100 mm σ = c/s 1.43Pitch s 70 mmUpstream velocity U1 80 m/s M 0.23Span L 200 mm Re 5.3 × 105

1. Systematic two-dimensional cascade tests of NACA 65-series compressor blades at lowspeeds, NACA report 1368, 1957



ExperimentImprovements from Sabah’s set-up

2001 : Sabah’s PhD thesis

- Aim : measurement of cascade leading edge and trailing edge noise,

- High background noise level.







Some improvements

- Side panels,

- Brushes at the end of woodboards,

- No boundary layer exhaust.







Some improvements

- Side panels,

- Brushes at the end of woodboards,

- No boundary layer exhaust.

Downstream far-field acousticspectra

100 1000 1000030

40

50

60

70

80

frequency, Hz

PS

D,

dB

Sabah ;Kevlar walls ;Kevlar walls without BL exhaust ;Metal walls without BL exhaust

⇒ downstream measurements only.


Cascade Experiment Aerodynamic Loading

Outline






Aerodynamic LoadingValidating the cascade design point

Reference point near maximum efficiency

α1 = 15◦, β1 = 35◦, U1 = 80 m/s. (σ = 1.43)





α1 = 15◦, β1 = 35◦, U1 = 80 m/s. (σ = 1.43)

0 20 40 60 80 100−0.6

−0.4

−0.2

0

0.2

0.4

0.6

0.8

x/c, %

−C

p

Control on center blade.Measurement,Bock RANS computation,Emery & al.,(α1 = 14.1◦ , β1 = 30◦, σ = 1.5)Sabah.





α1 = 15◦, β1 = 35◦, U1 = 80 m/s. (σ = 1.43)

0 20 40 60 80 100−0.6

−0.4

−0.2

0

0.2

0.4

0.6

0.8

x/c, %

−C

p

Control on center blade.Measurement,Bock RANS computation,Emery & al.,(α1 = 14.1◦ , β1 = 30◦, σ = 1.5)Sabah.

1 2 3 4 5 6 7

−0.4

−0.2

0

0.2

0.4

0.6

numero de pale

−C

p

Control on other blades.At 50% chord,

Suction Side,Pressure Side.


Cascade Experiment Cascade Acoustics

Outline






Velocity DependenceSpecific frequency scaling

Raw PSD

100 1000 1000020

30

40

50

60

frequency, Hz

PS

D (

dB/H

z)

100m/s

80m/s

60m/s

Scaled PSD

100 1000 1000020

30

40

50

60

frequency, Hz

DS

P (

dB/H

z)Far-field PSD at r=2 m

Scaling to U1 =80 m/s using PSD ∝ U 61 ,

Frequency axis unchanged ⇒ He= fL

c0rather than St= f δ∗

U1.



Acoustic ResonancesNear-field / far-field coherence

Remote Surface Unsteady Pressure Probes

Near the center blade trailing edge, midspan plane

A B

C

D

E

F

G

U

10mm

- To measure entry data of analytical models (boundary layer statistics),

- To investigate near-field/far-field coherence.






A B

C

D

E

F

G

U

10mm



1000 10000−0.1

0

0.1

0.2

0.3

0.4

βα

γγ2

frequency, Hz

θ < 0◦,θ > 0◦.

ParkerResonances : mode β

plate nodes






A B

C

D

E

F

G

U

10mm



1000 10000−0.1

0

0.1

0.2

0.3

0.4

βα

γγ2

frequency, Hz

θ < 0◦,θ > 0◦.

ParkerResonances : mode α

plate nodes






A B

C

D

E

F

G

U

10mm



1000 10000−0.1

0

0.1

0.2

0.3

0.4

βα

γγ2

frequency, Hz

θ < 0◦,θ > 0◦.

ParkerResonances : mode γ



Acoustic InterferenceVisible in far-field measurements

Blade-to-blade reflections

100 1000 1000020

30

40

50

60

70

PS

D,

dB

frequency, Hz





100 1000 1000020

30

40

50

60

70

PS

D,

dB

frequency, Hz

A B

δ1 δ2

|1 − eik(δ1+δ2)|





100 1000 1000020

30

40

50

60

70

PS

D,

dB

frequency, Hz

A B

δ1 δ2

|1 − eik(δ1+δ2)|

Most obvious cascade effects have been observed

To go further, need for analytical modeling : computation of the cascadetransfer function.


Analytical Model Validation Input Data for Models

Outline






Input Data Measurement : BL Statistics

# Φpp(ω)

100 1000 1000060

65

70

75

80

85

90

95

100

frequency, Hz

PS

D,

dB

Cascade,

Isolated airfoil,

On center blade,U1 ≃ 70 m/s,α1 = 35◦ ,β = 45◦ .

A B

C

D

E

F

G

U

10mm




# Φpp(ω)

100 1000 1000060

65

70

75

80

85

90

95

100

frequency, Hz

PS

D,

dB

Cascade,

Isolated airfoil,

# lz =∫ +∞

0γ(ω, η) cos(Kzη)dη

100 1000 100000

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

γ

frequency, Hz

η (mm) : 1, 2.5, 3.5, 5.


A B

C

D

E

F

G

U

10mm




# Φpp(ω)

100 1000 1000060

65

70

75

80

85

90

95

100

frequency, Hz

PS

D,

dB

Cascade,

Isolated airfoil,

# Uc

0 50 100 150 200 250 3000

0.2

0.4

0.6

0.8

1

Uc/U

1

St= 1 × f/U1

U1, m/s60,70,80,100.

# lz =∫ +∞

0γ(ω, η) cos(Kzη)dη

100 1000 100000

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

γ

frequency, Hz

η (mm) : 1, 2.5, 3.5, 5.


A B

C

D

E

F

G

U

10mm


Analytical Model Validation Amiet’s Isolated Airfoil Noise Model

Outline





Analytical Model Validation Amiet’s Isolated Airfoil Noise Model

Analytical ModelsAmiet’s isolated airfoil TE noise model

Isolated airfoil formulation :

S (n)pp (x , y, z , ω) = 2L

(

ωyc

4πc0S20

)2 ∣

∣

∣

∣

I

(

ω

Uc

, 0

)∣

∣

∣

∣

2

Φpp(ω)lz

(

kz

S0= 0, ω

)

7 independant airfoils, observer in the midspan plane, infinite span, shear layer refraction.

n=1

n=4 θ4

h

x

y

M

Measurement,Prediction,

200 1000 1000020

30

40

50

60

frequency, Hz

PS

D(d

B/

Hz)


Analytical Model Validation Glegg’s Cascade Noise Model

Outline






Glegg’s Analytical ModelOriginal formulation

Linear cascade model- Fan azimuthal periodicity ⇒ periodic sources in linear cascade

- Small numbers of propagating modes,

- BL information in Q,

- 3D : K = (Kx ,Kz ).

ϑ

x

K

ps =−2πi

Bh

m=+∞∑

m=−∞

QHm(Kx)e−iωt+iKz z K −

m e−iγ−

m (x−yd/h)−2iπmy/Bh

(2π)2i(γ−

m + Kx)J(m)+ (−Kx)J

(m)−

(γ−

m)







- 3D : K = (Kx ,Kz ).

ϑ

x

K

ps =−2πi

Bh

m=+∞∑

m=−∞


m e−iγ−


(2π)2i(γ−


(m)−

(γ−

m)

0 1 2 3−1

−0.5

0

0.5

1

−0.05

0

0.05

x,m

y,m

ℜ(ps), Pa

ω = 1 ⇒ 1 propagative mode.







- 3D : K = (Kx ,Kz ).

ϑ

x

K

ps =−2πi

Bh

m=+∞∑

m=−∞


m e−iγ−


(2π)2i(γ−


(m)−

(γ−

m)

0 1 2 3−1

−0.5

0

0.5

1

−0.05

0

0.05

x,m

y,m

ℜ(ps), Pa

ω = 1 ⇒ 1 propagative mode.

0 1 2 3−1

−0.5

0

0.5

1

−0.06

−0.04

−0.02

0

0.02

0.04

0.06

x,m

y,m

ℜ(ps), Pa

ω = 3 ⇒ 5 propagative modes.



Glegg’s Analytical ModelChoice of Periodicity Parameter B

Choosing B with the aim of isolating a single source contribution.

0 1 2 3−1

−0.5

0

0.5

1

−0.06

−0.04

−0.02

0

0.02

0.04

0.06

x,m

y,m

ℜ(ps),

Pa

B=30





0 1 2 3−1

−0.5

0

0.5

1

−0.06

−0.04

−0.02

0

0.02

0.04

0.06

x,m

y,m

ℜ(ps),

Pa

B=30

0 1 2 3−1

−0.5

0

0.5

1

−0.06

−0.04

−0.02

0

0.02

0.04

0.06

x,m

y,m

B=100





0 1 2 3−1

−0.5

0

0.5

1

−0.06

−0.04

−0.02

0

0.02

0.04

0.06

x,m

y,m

ℜ(ps),

Pa

B=30

0 1 2 3−1

−0.5

0

0.5

1

−0.06

−0.04

−0.02

0

0.02

0.04

0.06

x,m

y,m

B=1000

0 1 2 3−1

−0.5

0

0.5

1

−0.06

−0.04

−0.02

0

0.02

0.04

0.06

x,m

y,m

B=100





0 1 2 3−1

−0.5

0

0.5

1

−0.06

−0.04

−0.02

0

0.02

0.04

0.06

x,m

y,m

ℜ(ps),

Pa

B=30

0 1 2 3−1

−0.5

0

0.5

1

−0.06

−0.04

−0.02

0

0.02

0.04

0.06

x,m

y,m

B=1000

0 1 2 3−1

−0.5

0

0.5

1

−0.06

−0.04

−0.02

0

0.02

0.04

0.06

x,m

y,m

B=100

0 1 2 3−1

−0.5

0

0.5

1

−0.06

−0.04

−0.02

0

0.02

0.04

0.06

x,m

y,m

B=10000



Glegg’s Analytical ModelComparison with acoustic measurements

Last Steps :

- Expressing Q in terms of Φpp, lz , Uc,

- Integrating on Kz ,

- Summing over independant blade contributions.



Glegg’s Analytical ModelComparison with acoustic measurements

Last Steps :

- Expressing Q in terms of Φpp, lz , Uc,

- Integrating on Kz ,

- Summing over independant blade contributions.

Results : Suction side θ = 40◦

Slight modulation of singleairfoil results (3 dB).

Measurement,Glegg’s model,Amiet’s model.

200 1000 10000

30

40

50

60

frequency, Hz

PS

D,

dB



Analytical ModelsConclusions

Three noise models have been adapted to the test case :

- Amiet’s isolated airfoil TE noise model,

- Glegg’s cascade TE noise model,

- Howe’s cascade TE noise model (not shown).








Results show that :

- Cascade effect ≃ 3 dB,

- Experiment variation close to model discrepancies,

- Cascade CPU (10min) ≫ Isolated Airfoil CPU (1 sec).








Results show that :

- Cascade effect ≃ 3 dB,

- Experiment variation close to model discrepancies,

- Cascade CPU (10min) ≫ Isolated Airfoil CPU (1 sec).

Cascade TE noise measured and may be reduced.


Cascade Noise Reduction Set-up

Outline






IntroductionCascade TBL-TE Noise Reduction

Successful isolated airfoil noise reduction

- Many passive devices tested in the flocon eu project,

- te serrations gave best results & were down-selected.







⇒ Does the noise reduction also occur in a cascade ?







⇒ Does the noise reduction also occur in a cascade ?

Serrations sketches

10

13

20

- Straight edge

- Short serrationsλc = 2 mm, 2h = 13 mm

- Long serrationsλc = 2 mm, 2h = 20 mm,

2hc

λc

e


Cascade Noise Reduction Acoustic Results

Outline





Cascade Noise Reduction Acoustic Results

Serration Acoustic ResultsFar field noise reduction, suction side direction θ = 40◦

100 1000 100000

10

20

30

40

50

60

straight TElong serrationsshort serrations

40m/s

60m/s

80m/s

100m/s

Noiseincrease

Noise reduction

St0

frequency, Hz

Fa

rfi

eld

PS

D,

dB

/H

z

- Very similar results to single airfoil noise reduction,

- Single airfoil St0=1.18 ; Cascade St0=1.21.


Cascade Noise Reduction PIV results

Outline






Near Wake DynamicsMean Velocity U x

Straight Edge Serrated Edge

−10 −5 0 5 10

0

5

10

15

0

10

20

30

40

x location, mm

ylo

ca

tio

n,

mm

Ux ,

m/s

−10 −5 0 5 10

0

5

10

15

0

10

20

30

40

x location, mm

ylo

ca

tio

n,

mm

Ux ,

m/s

Observations

- Serration wake larger and less deep,

- Serrations : through flow with a 10◦ angle.



Near Wake DynamicsFluctuating velocity u

′x

Straight Edge Serrated Edge

−10 −5 0 5 10

0

5

10

15

0

1

2

3

4

5

x location, mm

ylo

ca

tio

n,

mm

u′

x ,

m/s

−10 −5 0 5 10

0

5

10

15

0

1

2

3

4

5

x location, mm

ylo

ca

tio

n,

mm

u′

x ,

m/s

Observations

- BL removed from the airfoil surface,

- Smaller turbulent area in the pressure sidewake.


Cascade Noise Reduction Noise reduction mechanisms

Outline






SerrationsInvestigation of noise reduction mechanisms

Serrations may reduce noise through :

- BL ejection,

- Spanwise decorrelation,

- Local sweep angle ? → Analytical investigations.




Serrations may reduce noise through :

- BL ejection,

- Spanwise decorrelation,

- Local sweep angle ? → Analytical investigations.

Modified Amiet model for sweep angle ϕ :

Step 1 : Airfoil response function - compressible unsteady aerodynamics

−→U

Ux

Uz

airfoil

ϕ −→x

−→y

−→z

−→K

Kx

Kz

ψ

ϕ = 0◦,

ϕ = 15◦ ,

ϕ = 60◦ ,

ϕ = 85◦.

−2 −1.5 −1 −0.5 00

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

2x/c

|P|




Modified Amiet model for sweep angle ϕ :

Step 2 : Far-field scattering - Curle’s theory

M

airfoilϕ

−→x

−→x ′−→y

−→z ′ −→z

L

ω=1,

ω=5,

ω=10,

ω=20.

0 20 40 60 8028

30

32

34

36

38

40

42

44

46

ϕ,◦

Wp

p,

dB

Conclusions :- Serration details cannot be analytically reproduced ⇒ only a hint,

- This new model could be useful for fan noise prediction,

- Decrease of noise power in cos ϕ3.


Conclusions

Conclusions

Cascade Experiment

Cascade rig significantly improved,

Cascade trailing edge noise measured,

Specific cascade effects observed and analysed.

Analytical models

Input data measured in cascade,

Amiet’s model give reasonable prediction ±3 dB above 200 Hz,

Glegg’s cascade model adapted to the experiment,

It differs from Amiet’s model in level by ±3 dB,⇒ need for increasing cascade effect via σ.


Conclusions

Conclusions

Cascade noise reduction

Serration inserted in cascade,

Reduction at low frequencies,

No influence of cascade effect on noise reduction,

Candidate mechanism to noise reduction :

BL removal,Spanwise decorrelation,Local sweep angle,

Amiet’s model modified for sweep angle.


Conclusions

Thank you for your attention !


experimental study of broadband trailing edge noise of a ...experimental study of broadband trailing...

Documents