measurement and scaling of trailing-edge noisem. herr > 01.10..2009 measurement and scaling of...

M. Herr > 01.10..2009

Measurement and Scaling of Trailing-Edge Noise*

Michaela Herr

Institute of Aerodynamics and Flow TechnologyDLR Braunschweig

13th CEAS-ASC Workshop and 4th Workshop of X3-Noise, “Resolving Uncertainties in Airframe Noise Testing and CAA

Validation”, Bucharest, Romania, 1-2 October 2009

*rather an extensive look at the scaling of the related source quantities

CEAS-ASC/ X3-Noise Workshop 2009, Bucharest, (Romania) > M. Herr > 01.10.2009

2

BackgroundPrimary goals

Set-up of an experimental database which is suitable for parametric trailing-edge (TE) noise source studies and for CAA validationImprove DLR’s TE noise measurement and prediction methods

With reference to the upcoming “Workshop on Benchmark Problems for Airframe Noise Computations-I” to be held on June 10-11, 2010

ApproachDefinition of a simple generic test setup in the Acoustic Wind-Tunnel Braunschweig (AWB) with a reduced set of parameters, including realistic (HLD) Reynolds numbersComparison with theoretical approaches and with other available test data to provide

Validation of the chosen measurement data correctionsEstimates of systematic uncertainty contributions


3

Plate model0.15 mm

1 mm

Generic Test Setup - Acoustic Wind-Tunnel Braunschweig (AWB)

u

5 deglc

= 0.8–2.0 mu∞

= 40–60 m/sRe

= 2.1 Mio–7.9 Mio

= 0 deg

= 0 deg

LE tripping

Nozzle exit:0.8 m x 1.2 m

Exchangeable TE sections


4

Trailing-Edge Noise Data AssessmentMeasurement of related scaling parameters in the source region Measurement of the farfield TE noise; estimation of absolute levels

TBL mean velocity profiles and integral scales

Unsteady surface pressure point and cross spectra

Farfield TE noise spectra (re. unit distance and span)

Brooks & Hodgson (1981): )()1²(²2

)( 03

SMr

bMS

V

xV

00


5

Trailing-Edge Noise Data Assessment

TBL thickness wall friction velocity uw

/∞

, TBL edge velocity ue

Chase (1987):

Estimation of the surface pressure spectrum from TBL data

23

21 kkk

²21

212 k

ubVkk

|/²²|²

)(²)(²...................

...1|/²²|

²²

|/²²|)(

),(

22132

0

20

2

254

25

2

42/520

2

32

0

ckkkb

bkbkkbbb

ckkb

kckb

bkuP

k

Goody (2004), based on Howe (2000):

757.07.375.0

2

20

/1.15.0/

/3)/)((

ete

e

w

e

uRu

uuS

)//()/( 2 uuR et


6

Part I – Measured TBL Mean Velocity Profiles Hot-wire (CTA) measurement resultsue

/u∞

= 0.98 ±

0.02 (40:1)

u =40, 50, 60 m/s

lc = 0.8 mlc = 1.2 mlc = 1.6 mlc = 2.0 m

x2, mm

u/u e

0 10 20 30 400

1


7

u =40, 50, 60 m/s

lc = 0.8 mlc = 1.2 mlc = 1.6 mlc = 2.0 m

x2+

u+

100 101 102 103 1040

10

20

30

40

u+ = x2+

u+ = 1/0.41 ln x2+ + 5

Scaling based on inner scale, ∞/u

);( 2 xf

uuu

uxx 22

u =40, 50, 60 m/s

lc = 0.8 mlc = 1.2 mlc = 1.6 mlc = 2.0 m

x2, mm

u/u e

0 10 20 30 400

1

Part I – Measured TBL Mean Velocity Profiles


8

u =40, 50, 60 m/s

lc = 0.8 mlc = 1.2 mlc = 1.6 mlc = 2.0 m

x2/99

u/u e

10-3 10-2 10-1 1000

1

u/ue = 0.084 ln(x2/99) + 0.82

Scaling based on outer scale,

);( *2xF

uuue

2*

2xx

u =40, 50, 60 m/s

lc = 0.8 mlc = 1.2 mlc = 1.6 mlc = 2.0 m

x2, mm

u/u e

0 10 20 30 400

1

Part I – Measured TBL Mean Velocity Profiles


9

Part I – Derived ParametersReconstruction of the velocity profiles, Coles’ (1956) Law of the Wake

;)ln(1122

sin15.0)ln(1 22

B

uuxBxu e

x2, mm

u/u e

0 10 20 30 400

1

u =40 m/s50 m/s60 m/s

99

1

,, 2

, H12

uw

These are used in the following for the scaling of unsteady surface pressure and TE noise data.

/u


10


Part II – Unsteady Surface Pressure Data



Point PSD close to the TEStreamwise and spanwise coherence decayStreamwise and spanwise coherence lengthsConvection velocities


11

f, kHz

10lo

g(S

0(f)/

p2 ref)

=L p(

f=

1H

z),d

B

100 10160

65

70

75

80

85

90lc=0.8 mlc=1.2 mlc=1.6 mlc=2.0 m

u =40 m/s50 m/s60 m/s

Part II – Point PSDs close to the TE

TE

u∞


12

1/ue

10lo

g[S

0(

)ue/q

e2 1],

dB

10-1 100 101-65

-60

-55

-50

-45

-40

-35lc=0.8 mlc=1.2 mlc=1.6 mlc=2.0 m

u =40 m/s50 m/s60 m/s

f, kHz

10lo

g(S

0(f)/

p2 ref)

=L p(

f=

1H

z),d

B

100 10160

65

70

75

80

85

90lc=0.8 mlc=1.2 mlc=1.6 mlc=2.0 m

u =40 m/s50 m/s60 m/s

Part II – Point PSDs close to the TE Scaling based on outer scales:

Pressure scale: qe

= ½∞

ue2

Time scale: 1

/ue or 99

/ue


13

Scaling based on “mixed” scalesPressure scale: w

=∞

u2Time scale: 1

/u

, 99

/u

, 1

/ue

, 99

/ue

Scaling based on inner scalesPressure scale: wTime scale: ∞

/u2

Part II – Point PSDs close to the TE

99/u

10lo

g[S

0(

)u/

w2 99

],dB

101 102 103-35

-30

-25

-20

-15

-10

-5lc=0.8 mlc=1.2 mlc=1.6 mlc=2.0 m

u =40 m/s50 m/s60 m/s

-0.8

0.3

-1

/u2

10lo

g[S

0(

)u2 /

w2 ],

dB

10-2 10-1 1000

5

10

15

20

25

30lc=0.8 mlc=1.2 mlc=1.6 mlc=2.0 m

u =40 m/s50 m/s60 m/s

/u2 = 0.3


14

Part II – Current Prediction Models for point PSDs Chase (1987) model:

99/ue

10lo

g[S

0(

)ue/

w2 99

],dB

100 101 102-20

-15

-10

-5

0

5

10lc=0.8 mlc=1.2 mlc=1.6 mlc=2.0 m

u =40 m/s50 m/s60 m/s

Chase (1987)

99/ue

10lo

g[S

0(

)ue/

w2 99

],dB

100 101 102-20

-15

-10

-5

0

5

10lc=0.8 mlc=1.2 mlc=1.6 mlc=2.0 m

u =40 m/s50 m/s60 m/s


15

Part II – Current Prediction Models for point PSDs Chase (1987) model:

Empirical factors need a more detailed experimental assessment

99/ue

10lo

g[S

0(

)ue/

w2 99

],dB

100 101 102-20

-15

-10

-5

0

5

10lc=0.8 mlc=1.2 mlc=1.6 mlc=2.0 m

u =40 m/s50 m/s60 m/s

Chase (1987)

99/ue

10lo

g[S

0(

)ue/

w2 99

],dB

100 101 102-20

-15

-10

-5

0

5

10lc=0.8 mlc=1.2 mlc=1.6 mlc=2.0 m

u =40 m/s50 m/s60 m/s

coefficientschanged acc.to Schewe (1993)

99/ue

10lo

g[S

0(

)ue/

w2 99

],dB

100 101 102-20

-15

-10

-5

0

5

10lc=0.8 mlc=1.2 mlc=1.6 mlc=2.0 m

u =40 m/s50 m/s60 m/s


16

Part II – Current Prediction Models for point PSDs Goody (2004) model

RT = 123

RT = 51

Goody (2004)

99/ue

10lo

g[S

0(

)ue/

w2 99

],dB

100 101 102-20

-15

-10

-5

0

5

10lc=0.8 mlc=1.2 mlc=1.6 mlc=2.0 m

u =40 m/s50 m/s60 m/s


17


High frequency data corrections have to be taken with care!Existing deviations are due to application of high-frequency data corrections which seem to overvalue actual levels

RT = 123

RT = 51

Goody (2004)

99/ue

10lo

g[S

0(

)ue/

w2 99

],dB

100 101 102-20

-15

-10

-5

0

5

10lc=0.8 mlc=1.2 mlc=1.6 mlc=2.0 m

u =40 m/s50 m/s60 m/s

RT = 123

RT = 51

Goody (2004)

99/ue

10lo

g[S

0(

)ue/

w2 99

],dB

100 101 102-20

-15

-10

-5

0

5

10lc=0.8 mlc=1.2 mlc=1.6 mlc=2.0 m

u =40 m/s50 m/s60 m/s


18


High frequency data corrections have to be taken with care!Existing deviations are due to application of high-frequency data corrections which seem to overvalue actual levels

A combination of both models might apply.

RT = 123

RT = 51

Goody (2004)

99/ue

10lo

g[S

0(

)ue/

w2 99

],dB

100 101 102-20

-15

-10

-5

0

5

10lc=0.8 mlc=1.2 mlc=1.6 mlc=2.0 m

u =40 m/s50 m/s60 m/s

RT = 123

RT = 51

Goody (2004)

99/ue

10lo

g[S

0(

)ue/

w2 99

],dB

100 101 102-20

-15

-10

-5

0

5

10lc=0.8 mlc=1.2 mlc=1.6 mlc=2.0 m

u =40 m/s50 m/s60 m/s

RT = 123

RT = 51

combined approach

99/ue

10lo

g[S

0(

)ue/

w2 99

],dB

100 101 102-20

-15

-10

-5

0

5

10lc=0.8 mlc=1.2 mlc=1.6 mlc=2.0 m

u =40 m/s50 m/s60 m/s


19

Part II – Coherence Decay and Coherence LengthsCorcos’ (1963) similarity approach:

Streamwise

Spanwise

TE

u∞

3x

1x)()(

),( 3,1

ji

xijij SS

S

1111 ),(/),( xvxxxij kV

),(/ 1 xv Vk

Currently restricted to 2-m-plate only!

x1,3/V(x1, )

ij

0 5 10 15 200

0.2

0.4

0.6

0.8

1

x3 = 0.714

x1 = 0.16

3 mm 0.10

7 mm 0.2317,16 mm 0.56,0.52

|x1,3| = |x1,3|/99=

u = 60 m/s

20,23 mm 0.65,0.75

4 mm 0.13

),(/exp),(

),(/exp),(

1333

1111

xxxxij

xxxxij

V

V


20


Streamwise

Spanwise

TE

u∞

3x

1x)()(

),( 3,1

ji

xijij SS

S

1111 ),(/),( xvxxxij kV

),(/ 1 xv Vk


),(/exp),(

),(/exp),(

1333

1111

xxxxij

xxxxij

V

V

x1,3/V(x1, )

ij

0 5 10 15 200

0.2

0.4

0.6

0.8

1

x3 = 0.714x1 = 0.17

3 mm 0.10

7 mm 0.2217,16 mm 0.54,0.51

|x1,3| = |x1,3|/99=

u = 50 m/s

20,23 mm 0.64,0.73

4 mm 0.13


21


Streamwise

Spanwise

TE

u∞

3x

1x)()(

),( 3,1

ji

xijij SS

S

1111 ),(/),( xvxxxij kV

),(/ 1 xv Vk


),(/exp),(

),(/exp),(

1333

1111

xxxxij

xxxxij

V

V

x1,3/V(x1, )

ij

0 5 10 15 200

0.2

0.4

0.6

0.8

1

x3 = 0.714x1 = 0.19

3 mm 0.09

7 mm 0.2217,16 mm 0.53,0.50

|x1,3| = |x1,3|/99=

u = 40 m/s

20,23 mm 0.62,0.71

4 mm 0.12


22

Part II – Coherence Decay and Coherence Lengths

x1

x3

f(peak)

f, kHz

x1,

3/99

5 10 15

0.5

1

1.52

2.5u =40 m/s50 m/s60 m/s

333

111

//1//1

xvxx

xvxx

VkVk


Assuming a constant V

0.65 u∞

Streamwise

Spanwise


23


Part III – Farfield TE Noise Data



Absolute 1/3-octave band SPL re 1m span and 1m observer distance

Brooks & Hodgson (1981): )()1²(²2

)( 03

SMr

bMS

V

xV

00

¼``- Microphone


24

Estimation of absolute levels (re 1m span and 1m observer distance) requires extensive data corrections, as shown in the following…

Part III – Farfield TE Noise DataPrediction from the surface pressure data, assuming constant V

0.65 u∞


fm, kHz

L p(1/

3),d

B

2 4 6 8 1030

40

50

60

70

80

90

100

u =40 m/s50 m/s60 m/s

Surface Pressure Kulite Data

FF TE Noise Mirror Data

fm, kHz

L p(1/

3),d

B

2 4 6 8 1030

40

50

60

70

80

90

100

u =40 m/s50 m/s60 m/s


FF TE Noise PredictionFF TE Noise Mirror Data

fm, kHz

L p(1/

3),d

B

2 4 6 8 1030

40

50

60

70

80

90

100

u =40 m/s50 m/s60 m/s


Brooks & Hodgson (1981): )()1²(²2

)( 03

SMr

bMS

V

xV

00


25

Part III – Elliptic Mirror SetupElliptic mirror specifics:

Focal distance: 1.15 mReflector diameter D = 1.4 mResolution width Aperture: 63 deg

Data correction for:Background noise (S/N

3 dB)

Sound wave convectionFrequency response function (gain, resolution)Shear layer refraction/scattering

¼``- Microphone

x

y


26

Part III – Elliptic Mirror SetupCorrection for frequency response (Schlinker, 1977)

21 )(2)(

JH

DRcxfD

3

2/

2/ 3

22

)(b

b

M

dxH

bGp

p

fm, kHz

L p(1

/3)co

rrec

tion,

dB

5 10 15 20

-40

-30

-20

-10

0

10

gain correctionresolution correctiontotal correction


27

Part III – Effect of Data Corrections1.6-m plate model with blunt TE (h = 1 mm)Focusing measurement techniques must be used because TE noise is buried by the tunnel self-noise!

fm, kHz

L p(1

/3),

dB

5 10 15 2030

40

50

60

70

80

uncorrected

empty test section

u= 60 m/s

fm, kHz

L p(1

/3),

dB

5 10 15 2030

40

50

60

70

80

uncorrectedLp (1/3) M

empty test section

u= 60 m/s

fm, kHz

L p(1

/3),

dB

5 10 15 2030

40

50

60

70

80

uncorrectedLp (1/3) MLp (1/3) M, shear-layer corrected

empty test section

u= 60 m/s

fm, kHz

L p(1

/3),

dB

5 10 15 2030

40

50

60

70

80

uncorrectedLp (1/3) MLp (1/3) M, shear-layer correctedLp (1/3) TE (b = 1 m)

empty test section

u= 60 m/s

mirror focus at TE

fm, kHz

L p(1

/3),

dB

5 10 15 2030

40

50

60

70

80

farfield mic. plate airfoilfarfield mic. BGNmirror mic. plate airfoil TE

u= 60 m/sb = 1.2 mr = 1.15 m(origin at TE, midspan)


28

Part III – Effect of Data Corrections1.6-m plate model with blunt TE (h = 1 mm)Focusing measurement techniques must be used because TE noise is buried by the tunnel self-noise!“Validation” of the procedure by alternative measurement techniques: COP

fm, kHz

L p(1

/3),

dB

5 10 15 2030

40

50

60

70

80

uncorrected

empty test section

u= 60 m/s

fm, kHz

L p(1

/3),

dB

5 10 15 2030

40

50

60

70

80

uncorrectedLp (1/3) M

empty test section

u= 60 m/s

fm, kHz

L p(1

/3),

dB

5 10 15 2030

40

50

60

70

80

uncorrectedLp (1/3) MLp (1/3) M, shear-layer corrected

empty test section

u= 60 m/s

fm, kHz

L p(1

/3),

dB

5 10 15 2030

40

50

60

70

80

uncorrectedLp (1/3) MLp (1/3) M, shear-layer correctedLp (1/3) TE (b = 1 m)

empty test section

u= 60 m/s

mirror focus at TE

fm, kHz

L p(1

/3),

dB

5 10 15 2030

40

50

60

70

80

farfield mic. plate airfoilfarfield mic. BGNmirror mic. plate airfoil TE

u= 60 m/sb = 1.2 mr = 1.15 m(origin at TE, midspan)

fm, kHz

L p(1

/3),

dB

5 10 15 2020

30

40

50

60

70

directional microphoneCOP

u= 40 m/s

b = 1.2 m

u= 60 m/s


29

fm/u

L p(1/

3)no

rm,d

B

0.1 1100

110

120

130

AWB (h = 0.15 mm)

Lp(1/3)norm= Lp(1/3)- 10 log (M5b/ r2)

u=40 m/s50 m/s60 m/s

fm/u

L p(1/

3)no

rm,d

B

0.1 1100

110

120

130

AWB (h = 0.15 mm)

Lp(1/3)norm= Lp(1/3)- 10 log (M5b/ r2)

u=40 m/s50 m/s60 m/s

BPM, sharp TE

Part III – Comparisons with Published TE Noise Data NACA0012 data by Brooks et al. (1986): COP semi-empirical prediction method (BPM, NAFNoise)

18 deg

0.4-m 2D-”NACA0012” with varying TE thickness


30

fm/u

L p(1/

3)no

rm,d

B

0.1 1100

110

120

130

AWB (h = 0.15 mm)

Lp(1/3)norm= Lp(1/3)- 10 log (M5b/ r2)

u=40 m/s50 m/s60 m/s

fm/u

L p(1/

3)no

rm,d

B

0.1 1100

110

120

130

AWB (h = 0.15 mm)

Lp(1/3)norm= Lp(1/3)- 10 log (M5b/ r2)

u=40 m/s50 m/s60 m/s

BPM, sharp TE

Part III – Comparisons with Published TE Noise Data NACA0012 data by Herrig et al.(2008): CPV 18 deg

0.4-m 2D-”NACA0012” with varying TE thickness

fm/u

L p(1/

3)no

rm,d

B

0.1 1100

110

120

130

AWB (h = 0.15 mm)AWB (h = 1 mm)

Lp(1/3)norm= Lp(1/3)- 10 log (M5b/ r2)

u=40 m/s50 m/s60 m/s

fm/u

L p(1/

3)no

rm,d

B

0.1 1100

110

120

130

AWB (h = 0.15 mm)AWB (h = 1 mm)

Lp(1/3)norm= Lp(1/3)- 10 log (M5b/ r2)

u=40 m/s50 m/s60 m/sHerrig et al.

(h = 1 mm)


31

Conclusions Presentation of a parametric plate model TE noise data set which (with the limitation to nonzero angles-of-attack and =

= /2) could

help to validate current CAA approaches, maximum Re

of 7.9 MioExcellent agreement of the plate model data set with theoretical modelsFair agreement with published data sets as derived at comparable but not identical test conditions (e.g. zero-pressure gradient surface pressure data covered by the Goody model), estimation of absolute TE noise levels was cross-checked by comparisons of additional NACA0012 measurement results with available NACA0012 dataLiterature review: Considerable uncertainty with regard to the measurement of farfield TE noise and its related source quantities prevails even for very simple generic test configurations Need of benchmarks for TE noise measurement (with major focus on the necessary frequency response corrections and facility-related effects) to provide the necessary data quality for CAA validation


32

Outlook Conduct RANS/CAA based predictions (PIANO-RPM) for the presented plate model experiment

respective RANS/CAA based predictions for a NACA0012, (published in Ewert et al., AIAA 2009-3269) were promising


33

Outlook Conduct RANS/CAA based predictions (PIANO-RPM) for the presented plate model experiment

respective RANS/CAA based predictions for a NACA0012, (published in Ewert et al., AIAA 2009-3269) were promising

Detailed comparison of directional microphone data with corresponding microphone array data

NACA0012 data available, but not yet analysedNumerical simulation of the mirror system transfer function (including shear-layer effects)Still open questions: determination of the various empirical coefficients in existing surface pressure models, estimation of based on mean TBL velocity profiles

See you in June 10-11, 2010 at the “Workshop on Benchmark Problems for Airframe Noise Computations-I”?

),( 1 xV


34

Thank you for your attention!

[email protected]


35

Appendix


36

Part II – Convection velocity

TE

u∞

3x

1x

1111 ),(/),( xvxxxij kV ),(/ 1 xv Vk


(99/u)peak

Eq. 5.2

99/u

V(

x1,

)/u

200 400 600 800

0.4

0.6

0.8

1

1.2 u =40 m/s50 m/s60 m/s

(99/u)peak

99/u

V(

x1,

)/u

200 400 600 800

0.4

0.6

0.8

1

1.2

3 mm 0.17 mm 0.2

20 mm 0.6-0.7|x1| = |x1|/99=

u =40 m/s50 m/s60 m/s


37

Part III – Elliptic Mirror Setup

fm, kHz

SP

L 1/3

Cor

rect

ion,

dB

5 10 15 200

1

2

3

4

5

Position 2Position 3Position 4

Shear-layer effect at:

Shear-layer correction from comparative measurements at different x- positions, re Position 1

x

z

x

y1 2 3 4


38

Part III – Elliptic Mirror SetupMeasured “point source” frequency response function (Dobrzynski et al.,1998)

fm, kHz

w,1

0-3

m

10 20 30 40

20

40

60

80100

u∞=0 m/s40 m/s50 m/s60 m/s

theoretical prediction

��

�

� ��

��

��

fm, kHz

G,d

B

10 20 30 40

20

25

30

35

40

45u∞ =0 m/s40 m/s50 m/s60 m/s

theoretical prediction

η

H(η

),d

B

-2 0 2 4 6

-10

-5

0

fm = 2.5 kHzfm = 5 kHzfm = 10 kHz

Eq. 2.17�

� �

��

��

��21 )(2)(

JH

measurement and scaling of trailing-edge noisem. herr > 01.10..2009 measurement and scaling of...

Documents