Phased Array Technology Applied to Aeroacoustics 15th AIAA/CEAS Aeroacoustics Conference (30th AIAA Aeroacoustics Conference) 13 May 2009 Miami, Florida Bob Dougherty President, OptiNav, Inc.
Outline Problems History Beamforming algorithms Airframe Noise Fan Noise Engine Noise Jet Noise Dipole Study from the OptiNav Aeroacoustic Facility Conclusions Postscript (added after the presentation)
Problems: Locate and isolate the sources Understand them Reduce them Predict the results Slat gap noise Slat edge noise Inlet noise: fan, low pressure compressor Flap side-edge noise Gear noise Aft fan, turbine, combustor and jet noise
History
The Telescope Hans Lippershey 1608 also Galileo Galilei 1609 source: Wikipedia
The Radio Telescope Karl Guthe Jansky 1931 source: Wikipedia
Astronomical Interferometry Ryle, M. & Vonberg, D., 1946 VLA: inaugurated in 1980 source: Wikipedia
The Acoustic Telescope J. Billingsley and R. Kinns, The acoustic telescope Journal of Sound and Vibration, 48, (4) 485-510, 1976. Computer: CA1 LSI-2, 48 kilobytes
AIAA JOURNAL VOL. 14, NO. 4, 489-497, APRIL 1976
Laufer, Schlinker, and Kaplan
Jet engine noise source location: The polar correlation technique M.J. Fisher, M. Harper-Bourne, and S.A.L. Glegg JSV 51(1), 23-54, 8 March 1977.
Fisher, Harper-Bourne and Glegg
Polar Correlation Technique in Context J. Billingsley, A comparison of the source location techniques of the acoustic telescope and polar correlation, JSV 61(3), 419-425, 1978.
More History Linear arrays - Soderman and Nobel, 1974 - Billingsley and Kinns, 1976 Directional mirror microphones, 1976 - Grosche, et al, Kendall, Schlinkler, Polar correlation technique - Fisher, Harper-Bourne, and Glegg, 1977 Advanced algorithms - CLEAN: Högbom, 1974 - Maximum likelihood: El-Behery & MacPhie, 1978 - MUSIC: Schmidt, 1986 - Robust adaptive beamforming: Cox, et al, 1987, Gramann & Mocio, 1993 - Making cross arrays work: Elias, 1995 - DAMAS: Brooks & Humphreys, 2004 - CLEAN-SC: Sijtsma, 2007 Flyover testing -Michel, et al, 1997 Wind tunnel techniques (spiral arrays, diagonal deletion, ignoring reflections) - Dougherty & Underbrink, 1994
Beamforming Algorithms
Optical Beamforming
Acoustics Beamforming
Time Domain Color contour plot noise p n (t) = time-domain pressure at microphone n, n = 1,, N. (real) Integrate square Array sum Delay Loop through grid points Data acquisition system 1 2... N
Case of one source at! r j n n t Delay t + " n ( )! r j Array data Array data in emission time
Frequency domain r u t ( ) = u n (t) = ( ) ( ) ( )! u 1 t # # u 2 t # u 3 t #... # "# u N t ( ) Digital filter " / 2 p n # (t + t ' )e j $ t' dt'!" / 2 $ & & & & & %& noise narrowband complex pressure Data acquisition system τ = block length 1-100 ms 1/τ = bandwidth 1 2... N
Cross-spectral matrix A = u v u v time average f (t) = 1 T T f (t)dt! 0 T 20 seconds (limited by disk storage or source motion)
Frequency domain data model v u ( t) = M v! C S m m( t) + n r ( t) = C S v (t) + n r ( t) m=1 and integration is long enough, then If A= Q = CQC where +! 2 n I S v (t) S v (t)
Frequency domain beamforming b m C v AC v (! ) = m! m! Array design goal v C v m! C m " # m! m ideally!)= b ( m Q m! m!
Nature of a Beamform Map
Nature of a Beamform Map 7 db dynamic range 20 db dynamic range
7 db dynamic range 20 db dynamic range 1/3 Octave Bands (Too wide for narrowband beamforming) 20 db dynamic range
7 db dynamic range 20 db dynamic range 1/3 Octave Bands (Too wide for narrowband beamforming) 7 db dynamic range
Problem Cases: Two Sources 7 db dynamic range 20 db dynamic range Weaker Source (10 db down) Coherent Source
Problem Case: Extended Source 7 db dynamic range 20 db dynamic range Weaker Source Coherent Source
DAMAS, Eigenvalues, CLEAN-SC and TIDY Goal: explain these buttons
Dirty map (initially regular beamforming) CLEAN Idea (Högbom, 1974) Clean map (initially blank) Find peak Move it to clean map Remove contribution from data and dirty map (removes sidelobes too) Final dirty map (empty) Final clean map (result) Iterate until dirty map is empty Clean map now contains real sources - No sidelobes - No peak spreading
Mutually Incoherent Sources M r r A = " Q m C m C m + # 2 n I m=1
Point Spread Function Consider the steering vector as a function of the beamform grid point r C n r " ( ) r x 1 r x 2 r r " 1 " 2 r x n r C n r " ( ) Array r x N r " Beamform Grid r " M map
Point Spread Function Beamform map: b ( r " )= 1 N 2 v C r " ( ) A v (" ) C r Assume incoherent sources b ( r M " )= Q m psf # m=1 r ", r ( " m ) where r psf (", r " m )= v C r " ( ) r C m 2
DAMAS (Brooks and Humphreys) r y = Ax v " $ $ r x = $ $ $ # $ Q 1... Q M map % ' ' ' ' ' &' # % % r y = % % % % $ b( r " ) 1... b r (" ) M map & ( ( ( ( ( ( ' A mm' = psf r (, r m m' )= v C r m ( ) r C m' 2 (A is not the CSM on this slide )
About DAMAS Important advancement in beamforming Assumes incoherent sources Narrow band Requires good estimate of psf Very slow in its pure form Closely related to other deconvolution methods used in image processing
Effect of DAMAS: Incoherent Source Conventional 12 db dynamic range DAMAS2 12 db dynamic range Weaker source Weaker source
Effect of DAMAS: Coherent Sources Conventional DAMAS2 Coherent source Coherent source
Eigenvalues (Schmidt, 1986) A = M r r " Q m C m C m + # 2 n I m=1 A = N # n=1 " n r V n r V n
Eigenvalues: Incoherent, First EV Conventional First EV Weaker source Weaker source
Eigenvalues: Incoherent, Second EV Conventional Second EV Weaker source Weaker source
Eigenvalues: Incoherent, Third EV Conventional Thrid EV Weaker source Weaker source
Eigenvalues: Coherent, First EV Conventional First EV Coherent source Coherent source Second EV is 27 db
Move Away from the Assumption of Mutually Incoherent Sources
CLEAN-SC (Sijtsma) "CSM = Q max hh h is determined so that C j AC max = C j GC max for all steering vectors C j.
TIDY (Dougherty 2009) Similar to CLEAN-SC, but operates in the time domain using the cross correlation matrix instead of the cross spectral matrix. Wide band. b FD ( ) = 1 N 2 r x m TD, FD relationship for beamforming C ( x m )AC r x m N#1 N#1 $ $ b TD ( x m ) = 1 s " N 2 im i= 0 k= 0 N"1 N"1 ## ( ) = 1 N 2 C i * x m i= 0 k= 0 ( )R ik ( t)s #" km ( ) ( )A ik C k x m t= 0 ( ) Shift operator s (") f ( t) = f ( t # ") R ik (! ) = p ( t) p ( t "! ) i k
Effect of TIDY: Incoherent Source Conventional 12 db dynamic range TIDY 12 db dynamic range Weaker source Weaker source
Effect of TIDY: Coherent Sources Conventional TIDY Coherent source Coherent source
Effect of TIDY: Extended Source Conventional TIDY Weaker Source Coherent Source Bandwidth too large for DAMAS
Other Deconvolution Algorithms Conventional TIDY Richardson-Lucy and NNLS (image processing) See Ehrenfried & Koop AIAA 2006-2711 DAMAS-C (Brooks & Humphreys) Coherent Source LORE (Ravetta) Generalized inverse beam-forming (Suzuki) Bandwidth too large for DAMAS
Airframe Noise
Wind tunnel testing: setup for a closed jet test phased array noise phased arrays flow noise phased array flow
RWS
13th AIAA/CEAS Aeroacoustics Conference (28th AIAA Aeroacoustics Conference) AIAA 2007-3448 A Preliminary Study of Landing Gear Noise in Low-Speed Wind Tunnel Hiroki URA, Takeshi ITO, Toshimi FUJITA, Akihito IWASAKI, Norihisa ANDO and Jun SATO
URA, ITO, FUJITA,IWASAKI, ANDO and SATO 6.3 khz 7 db range
Wind tunnel testing: setup for an open jet test free jet nozzle flow flow noise phased array free jet nozzle noise phased array
Michel, U., Barsikow, B., Helbig, J., Hellmig, M., and Schüttpelz, M., Flyover noise measurements on landing aircraft with a microphone array," AIAA Paper 98-2336, 1998,
Michel, U., Barsikow, B., Helbig, J., Hellmig, M., and Schüttpelz, M., Flyover noise measurements on landing aircraft with a microphone array," AIAA Paper 98-2336, 1998,
AIAA Paper 98-2336, 1998
Doughery, R. P., F. W. Wang, E. R. Booth, M. E. Watts, N. Fenichel, and R. E. D Errico, Aircraft wake vortex measurements at Denver International Airport, AIAA Paper. 2004-2880, May, 2004. 767 737
Fan Noise
Boeing/GE LSAF Test 1993
Boeing/Rolls Royce
Boeing/Pratt&Whitney ICD Array, 1999
Co-rotating mode Counter-rotating mode 90 2800 Hz 76 2708 Hz 85 74 80 72 75 70 70 68-40 -30-20 -10 0 10 20 30 40-40 -30-20 -10 0 10 20 30 40 Spinning order, m Spinning order, m
13th AIAA/CEAS Aeroacoustics Conference (28th AIAA Aeroacoustics Conference) AIAA 2007-3696 Feasibility of In-Duct Beamforming Pieter Sijtsma
Sijtsma
Sijtsma
Virtual Rotating Microphone Imaging of Broadband Fan Noise Robert P. Dougherty and Bruce E. Walker AIAA-2009-3121
Dougherty and Walker
Dougherty and Walker
Dougherty and Walker Baseline Modified
Phased Array Noise Source Localization Measurements Made on a Williams International FJ44 Engine Gary G. Podboy NASA Glenn Research Center Cleveland, Ohio U.S.A. Csaba Horvath ASRC Aerospace Corporation Cleveland, Ohio U.S.A. Presented by Daniel L. Sutliff at 15th AIAA/CEAS Aeroacoustics Conference Miami, Florida May 12, 2009
Podboy and Horvath The Williams FJ44 Engine 3000 lbf (12500 N) thrust class 16 fan blades 2 spools LP: fan, 3-stage axial comp, 2-stage axial turbine HP: single-stage centrifugal comp, single-stage axial turbine
Podboy and Horvath OptiNav Array 48 Phased Array System 48 flush-mounted electret microphones 1m x 1m Al plate Log spiral arrangement Data reduction options -conventional beamforming -deconvolution methods
Podboy and Horvath OptiNav Array 48 Phased Array System Software overlays acoustic source location data on top of photo taken with the camera.
Podboy and Horvath 10 khz 1/3rd Octave Beamform Map for Engine at 100% Speed Peak (db) Peak 7
Podboy and Horvath 1/3rd Octave Beamform Maps for 3 Engine Speeds
Engine Noise
Noise Source Analysis of an Aeroengine with a New Inverse Method SODIX AIAA-2008-2860 Ulf Michel and Stefan Funke
Michel and Funke
46th AIAA Aerospace Sciences Meeting and Exhibit 7-10 January 2008, Reno, Nevada AIAA 2008-51 Phased Array Beamforming with 100-foot Polar Arc Microphones in a Static Engine Noise Test Robert P. Dougherty and Jeff M. Mendoza NASA/Honeywell EVNERT 2006
Dougherty and Mendoza
Dougherty and Mendoza 60% Power, 5 High Frequency Array db
Jet Noise
Boeing/Honeywell Cage Array, 1999
Cage Array, Baseline Configuration 125 Hz 250 Hz 400 Hz 500 Hz 630 Hz 800 Hz
11th AIAA/CEAS Aeroacoustics Conference (26th AIAA Aeroacoustics Conference) 23-25 May 2005, Monterey, California AIAA 2005-2842 Phased-array Measurements of Single Flow Hot Jets Sang Soo Lee and James Bridges
12th AIAA/CEAS Aeroacoustics Conference (27th AIAA Aeroacoustics Conference) 8-10 May 2006, Cambridge, Massachusetts AIAA 2006-2644 LOCALIZATION OF MULTIPLE TYPES OF JET NOISE SOURCES Dimitri Papamoschou and Ali Dadvar
Papamoschou and Dadvar
Improved Phased Array Imaging of a Model Jet Robert Supersonic P. Dougherty and With Gary G. Shocks Podboy AIAA-2009-3186
90 Condition 2 Mj = 0.89, Ma = 1.46 40 St = 2.30-4.60 St = 1.15-2.30 St = 0.57-1.15 St 0.29-0.57 St = 0.14-0.29 St = 0. 071-0.14
Condition 2 Mj = 0.89, Ma = 1.46 90 CF (khz) BF Peak (db) Array Ave. (db) TIDY Integ. (db) Source Loc (diam.) 1 100.0 101.1 101.4 9.2 2 103.5 106.1 105.6 6.6 4 103.1 108.3 107.5 5.4 8 98.9 108.0 106.8 3.5 16 91.3 106.0 104.2 1.8 32 78.2 98.3 92.7 0.8 OASPL 113.6 112.6 40 CF (khz) BF Peak (db) Array Ave. (db) TIDY Integ. (db) Source Loc (diam.) 1 119.4 118.5 119.9 8.4 2 122.4 122.3 122.8 5.3 4 122.1 122.5 122.7 3.6 8 115.3 117.5 117.0 2.2 16 105.7 110.9 109.0 1.1 32 90.4 100.7 97.2 0.1 OASPL 126.9 127.3
90 Robert P. Dougherty and Gary G. Podboy Mj = 1.81, Ts = Tamb 40 St = 1.86-3.72 St =. 93-1.86 St = 46-.93 St =.23-.46 St =.12-.23 St =.058-.12
Robert P. Dougherty and Gary G. Podboy Underexpanded, Supersonic Case Conventional TIDY Conventional TIDY
Dipole Noise Study OptiNav Aeroacoustic Facility May, 2009
OptiNav Aeroacoustic Facility
Vortex Shedding TIDY Shop vac noise Vortex shedding Shear layer turbulence
Vortex Shedding: Dipole Sound
Vortex Shedding: Dipole Parallel to Array Conventional TIDY
Dipole Beamforming Option in Beamform Interactive Conventional, dipole TIDY, dipole See also: papers by Liu, Quayle and Dowling (and Sijtsma); Suzuki
Shear Layer Noise
String Trimmer Noise N 1 = 127 Hz (7620 RPM) d string = 1.7 mm r Hub = 4 cm r Tip = 20 cm M tip = 0.46
String Speed v s 159.6 m/s Frequencies Vortex Shedding Frequency f 0 = St v s d 19,715 Hz Doppler Shifted Frequency f = c c + v s f 0 13,467 Hz Shaft order 1 2 34 14,518 Hz Receding string 31.9 m/s 3940 Hz 3605 Hz Approaching string - 31.9 m/s - 159.6 m/s 3940 Hz 19,715 Hz 4342 Hz 36,779 Hz Assume c = 344 m/s, St = 0.21
Conventional
TIDY
X-Dipole
TIDY X-Dipole
Component Models: TIDY Integral Freq1 Freq2 Time1 Time2 z(m) Mic_Med. Peak_BF Integral xpeak ypeak 6120.6 6491 0 8.3 0.5 71.6 60.9 57.8 98 82 6491 6890.4 0 8.3 0.5 72.6 64.1 63.1 98 78 6890.4 7314.5 0 8.3 0.5 73.7 64.8 64.8 95 75 7314.5 7764.6 0 8.3 0.5 75.4 68.6 68 97 67 7764.6 8226.3 0 8.3 0.5 77.5 73.5 72.9 97 67 8226.3 8698.2 0 8.3 0.5 78.7 72.5 72.2 96 63
In-Plane String Trimmer Noise (String Components, 1Hz levels) Receding, Dipole Approaching, Dipole Approaching, Monopole
Conclusion Phased Array Technology is Well suited to aeroacoustics Nearly indispensable Continuously evolving Quite practical at this point
Postscript Notes added after the presentation Brian Tester and Ulf Michel have both noted that flyover measurements were reported in: G.P. Howell, A.J. Bradley, M.A. McCormick, and J.D. Brown, De-Dopplerization and acoustic imaging of aircraft flyover noise measurements, Journal of Sound and Vibration 105(1) 151-167, 1986. Another source of good examples is: L. Brusniak, J.R. Underbrink, and R.W. Stoker, Acoustic imaging of aircraft noise sources using large aperture phased arrays, AIAA 2006-2715