Post on 01-Jan-2017
transcript
DESIGN, FABRICATION, AND CHARACTERIZATION OF AN ANECHOIC WIND
TUNNEL FACILITY
By
JOSE MATHEW
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
2006
ii
ACKNOWLEDGMENTS
I would like to acknowledge the financial support of the University of Florida’s
College of Engineering and Department of Mechanical and Aerospace Engineering and
the financial support of a NASA Langley Research Center Grant NAG1-03044,
monitored by Dr. Mehdi Khorrami.
I would like to thank my advisor, Dr. Louis N. Cattafesta III, for his continual
guidance and motivation in making this work possible. I also would like to express my
heartfelt gratitude to Dr. Mark Sheplak, Dr. Bruce Carroll, Dr. Toshi Nishida and Dr.
Siddharth Thakur for their ideas and encouragement. I am also deeply indebted to Mr.
Chris Bahr for his help and support.
I would also like to thank numerous individuals for their invaluable contributions to
this project, including Cesar Moreno, Michael Sytsma, Nik Zawodny, Ryan Holman,
Todd Schultz, Ed Duell, Dragos Vieru, Raj Vaidyanathan, David Weiner, Jared Lee, Ron
Brown, Wayne Willis, and Grant Pettit.
My parents and sisters deserve special credit for giving me moral support and
motivating me through the course of my research work. Finally, I thank God for giving
me an opportunity to enjoy life as a successful doctoral student.
iii
TABLE OF CONTENTS Page ACKNOWLEDGMENTS .................................................................................................. ii
LIST OF TABLES............................................................................................................. vi
LIST OF FIGURES .......................................................................................................... vii
ABSTRACT.......................................................................................................................xv
CHAPTER 1 INTRODUCTION .......................................................................................................1
Background .................................................................................................................... 2 Airframe Noise ............................................................................................................... 3 Existing Anechoic Wind Tunnels................................................................................... 8 Motivation .................................................................................................................... 11 Thesis Objectives ......................................................................................................... 12 Technical Approach ..................................................................................................... 14 Thesis Organization...................................................................................................... 14
2 ANECHOIC CHAMBER ..........................................................................................16
Facility Description ...................................................................................................... 16 Free Field Characterization .......................................................................................... 20 Jet Noise Characterization............................................................................................ 23
3 DESIGN OF THE ANECHOIC WIND TUNNEL ...................................................33
Design Criteria ............................................................................................................. 33 Overall Layout.............................................................................................................. 35 Settling Duct/Honeycombs/Screens ............................................................................. 39 Contraction ................................................................................................................... 42 Test Section .................................................................................................................. 47 Diffuser......................................................................................................................... 48 Corner/Turning Vanes.................................................................................................. 56 Vibration Isolator ......................................................................................................... 60 Transition...................................................................................................................... 61 Fan ............................................................................................................................. 63
iv
Acoustic Treatment ...................................................................................................... 66 4 FABRICATION OF THE WIND TUNNEL COMPONENTS.................................70
Inlet ............................................................................................................................. 70 Diffuser......................................................................................................................... 71 Corner/Turning Vanes.................................................................................................. 74 Vibration Isolator ......................................................................................................... 76 Transition...................................................................................................................... 77 Fan ............................................................................................................................. 78 Acoustic Treatment ...................................................................................................... 80
5 EXPERIMENTAL METHODS.................................................................................83
Chamber Deflection and Wall Loading........................................................................ 83 Tunnel Circuit Static Pressure...................................................................................... 85 Flow Uniformity........................................................................................................... 87 Shear Layer Growth ..................................................................................................... 88 Freestream Turbulence Measurements......................................................................... 91 Background Noise ........................................................................................................ 93 Fan Noise Attenuation.................................................................................................. 95 Background Noise Source Identification ..................................................................... 98 Vibration Measurements ............................................................................................ 100 Acoustic Liner Absorption Coefficient Measurement Setup ..................................... 102
6 FACILITY CHARACTERIZATION......................................................................104
Chamber Deflection and Wall Loading...................................................................... 104 Inlet Wall Pressure ..................................................................................................... 106 Diffuser Wall Pressure ............................................................................................... 110 Flow Uniformity......................................................................................................... 112 Shear Layer Behavior................................................................................................. 115 Freestream Turbulence ............................................................................................... 120 Background Noise Measurements.............................................................................. 126 Fan Noise Decay ........................................................................................................ 132 Background Noise Source Identification ................................................................... 135 Vibration Measurements ............................................................................................ 143 Acoustic Liner Absorption Coefficient Estimation.................................................... 150
7 CONCLUSIONS AND FUTURE WORK ..............................................................152
REFERENCES ................................................................................................................156
APPENDIX
v
A SCHEMATICS OF THE WIND TUNNEL ............................................................162
B DERIVATION OF THE INLET SHAPE POLYNOMIAL ....................................164
C INLET OPTIMIZATION STUDY ..........................................................................169
D DIFFUSER OPTIMIZATION STUDY...................................................................174
E FAN LOSS CALCULATION .................................................................................179
F EFFECT OF LEAKAGE ON WALL PRESSURE .................................................185
G RESULTS OF FREE FIELD CHARACTERIZATION..........................................192
BIOGRAPHICAL SKETCH ...........................................................................................201
vi
LIST OF TABLES
Table page 1-1. Details of the existing anechoic wind tunnels. ..........................................................11
2-1. SPL deviation errors for northeast directions, before and after treatment.................23
3-1. Summary of the wind tunnel design. .........................................................................38
3-2. Test section details.....................................................................................................47
3-3. Turning vane coordinates. .........................................................................................60
3-4. Results of the wind tunnel circuit loss calculation. ...................................................63
6-1. Axial location of the inlet pressure taps. ...................................................................86
5-1. Location of the diffuser microphones .......................................................................98
6- 2. Spectral error estimates. .........................................................................................125
6-3. Free Stream Turbulence Intensity............................................................................126
6-4. Error estimates for the background noise spectra....................................................128
6-5. Nomenclature for input and output microphones. ...................................................137
C-1. Results of Inlet optimization study. ........................................................................172
D-1. Final dimensions of the tunnel obtained from the optimization study. ..................178
vii
LIST OF FIGURES
Figure page 1-1. Aircraft noise sources. .................................................................................................3
1-2. Propulsive noise reduction through the ages. ..............................................................4
1-3. Relative magnitudes of the various aircraft noise components during landing...........5
1-4. Various airframe noise sources....................................................................................6
2-1. Schematic of the original UF anechoic chamber.......................................................18
2-2. Schematic of the wall wedges. ..................................................................................19
2-3. Cross sectional view of the chamber wall panel........................................................19
2-4. Measurement array paths in the anechoic chamber...................................................21
2-5. Deviation of pressure measurements from free field from chamber center towards bell-mouth. ...............................................................................................................22
2-6. Deviation of pressure measurements from free field from chamber center towards Northeast corner of room with double door. ............................................................22
2-7. Side view schematic of the jet noise measurements..................................................25
2- 8. Top view schematic of the jet noise measurements. ................................................26
2-9. Schematic of the jet nozzle........................................................................................26
2-10. Cold jet noise data measured at 90o to the jet axis at 83.5 jet diameters for various jet Mach numbers. Exhaust fan is off. Compressor on. ..........................................28
2-11. Cold jet noise data measured at 90o and 140o to the jet axis at 83.5 and 114.5 jet diameters, respectively, at M=0.9. Exhaust fan is off. F and G are the large- and fine-scale similarity third-octave band spectra ........................................................28
2-12. Cold jet noise data measured at 90o to the jet axis at 83.5 jet diameters for various jet Mach numbers. Exhaust fan is operating at max speed (~6000 CFM). .............30
viii
2-13. Comparison between blowdown (compressor off) and continuous (compressor on) operating conditions for approximately identical flow conditions (M=0.7±0.01) (measured at 90o to the jet axis). ..............................................................................30
3-1. Wind tunnel design flow chart...................................................................................35
3-2. Plan view of the wind tunnel. ....................................................................................37
3-3. Schematic of the honeycomb section. .......................................................................40
3-4. Schematic of the screen design..................................................................................42
3-5. pC distribution along corner for a contraction..........................................................43
3-6. Schematic of the contraction shape polynomial. .......................................................44
3-7. Contours of x velocity along the half mid-plane for the contraction.........................46
3-8. Schematic of the collector. ........................................................................................49
3-9. Schematic of the 2D diffuser. ....................................................................................49
3-10. 2-D Diffuser Design Curves. ...................................................................................51
3-11. Comparison of local pressure coefficient with Stratford’s separation pressure coefficient for diffuser 1...........................................................................................54
3-12. Comparison of local pressure coefficient with Stratford’s separation pressure coefficient for diffuser 2...........................................................................................54
3-13. Centre plane x velocity profile along diffuser 1. .....................................................55
3-14. Schematic of the Turning Vanes. ............................................................................58
3-15. Results from turning vane simulation for a test section speed of 76 /m s ..............58
3- 16. Results from turning vane simulation for a test section speed of 18 /m s . ............59
3-17. Schematic of the rectangular to round transition section ( 2.22 eH m= , 1.2 W m= , 1.95 D m= )..............................................................................................................62
3-18. Results from transition flow simulation for a test section speed of 76 /m s ..........62
3-19. Fan Load curve. .......................................................................................................65
3-20. Estimated pressure drop along the wind tunnel circuit for a test section velocity of 76 /m s .....................................................................................................................65
ix
3-21. Details of the wind tunnel duct walls. .....................................................................69
4-1. Photograph of the inlet contraction. ..........................................................................71
4-2. Photograph of diffuser 1. ...........................................................................................72
4-3. Diffuser 1 internal skeletal view................................................................................72
4-4. Cross sectional view of the ‘I’ beam. ........................................................................73
4-5. Structural reinforcement using polyurethane foam. ..................................................73
4-6. Structural reinforcement using semi cylindrical hollow fiberglass sheets. ...............74
4-7. Photograph of the turning vane rack..........................................................................75
4-8. Side view of the cross plate. ......................................................................................75
4-9. Photograph of vane mold...........................................................................................76
4-10. Photograph of the vibration isolator section............................................................77
4-11. Photograph of the transition piece. ..........................................................................78
4-12. Front view of the fan. ..............................................................................................79
4-13. Back view of the fan................................................................................................79
4-14. View of the fan base. ...............................................................................................80
4-15. Chamber traverse acoustic treatment.......................................................................81
4-16. Garage door acoustic treatment. ..............................................................................82
4-17. Photograph of the flow silencer...............................................................................82
5-2. Schematic of the chamber wall loading measurement setup.....................................84
5-3. Schematic of the inlet static pressure taps. ................................................................86
5-4. Photograph of the inlet static pressure taps. ..............................................................87
5-5. Schematic of the flow uniformity measurement setup. .............................................88
5-6. Schematic of the shear layer measurement setup. .....................................................90
5-7. Photograph of the shear layer measurement setup. ...................................................90
5-8. Hotwire measurement block diagram........................................................................93
x
5-9. Photograph of the background noise measurement setup. ........................................94
5-10. Coherent power measurement. ................................................................................95
5-11. Schematic of the fan noise measurement microphone holder. ................................97
5-12. Setup for measurement of fan noise decay..............................................................97
5-14. Photograph of the fan vibration measurement test arrangement. ..........................101
5-15. Photograph of the vibration isolator vibration test arrangement. ..........................102
5-16. Schematic of the Impedance tube setup. ...............................................................103
6-1. Wall loading as a function of test section speed......................................................105
6-2. Variation of effective velocity with the test section velocity. .................................105
6-3. Wall deflection vs test section velocity. ..................................................................106
6-4. Contraction pC distributions versus length for the (a) sidewall, (b) base, and (c) corner for 17 /TSU m s= . .......................................................................................107
6-5. Contraction pC distributions versus length for the (a) sidewall, (b) base, and (c) corner for 30 /TSU m s= . .......................................................................................108
6-6. Contraction pC distributions versus length for the (a) sidewall, (b) base, and (c) corner for 42 /TSU m s= ........................................................................................108
6-7. Comparison of the pressure drop across the inlet and flow conditioner section for a test section speed of a) 18 /m s b) 37 /m s ..........................................................109
6-8. Pressure recovery across the diffuser. .....................................................................111
6-9. Photograph showing the waviness of the inner surface of diffuser 2. .....................111
6-10. Comparison of the pressure recovery across the diffuser duct work section for a test section speed of a) 18 /m s b) 37 /m s ...........................................................112
6-11. Normalized stagnation pressure contours (max =1 w/ 0.1 interval) at the test section entrance. eH and eW are the height and the width at the diffuser 1 entrance for a test section speed of 17 /m s . ........................................................................113
6-12. Normalized stagnation pressure contours (max =1 w/ 0.1 interval) at the diffuser entrance for a test section speed of 17 /m s ...........................................................114
xi
6-13. Test section centerline velocity profile development along the test section length for a test section speed of 17 /m s . ........................................................................114
6-14. Normalized velocity profile in the yz plane in the y direction for 18 /TSU m s= ( 1 7.8 mmθ = ). ........................................................................................................116
6-15. Normalized velocity profile in the yz plane in the y direction for 30 /TSU m s= ( 1 6.9 mmθ = ). ........................................................................................................116
6-16. Normalized velocity profile in the zy plane in the y direction for 37 /TSU m s= ( 1 7.5 mmθ = ). ..................................................................117
6-17. Normalized velocity profile in the yz plane in the z direction for 18 /TSU m s= ( 1 7.1 mmθ = ). ........................................................................................................117
6-18. Normalized velocity profile in the yz plane in the z direction for 30 /TSU m s= ( 1 7.2 mmθ = ). ........................................................................................................118
6-19. Normalized velocity profile in the yz plane in the z direction for 37 /TSU m s= ( 1 7.2 mmθ = ). ........................................................................................................118
6-20. Variation of y momentum thickness with test section length for a) 18 /TSU m s= b) 30 /TSU m s= c) 37 /TSU m s= .............................................................................119
6-21. Variation of z momentum thickness with test section length for a) 18 /TSU m s= b) 30 /TSU m s= c) 37 /TSU m s= .............................................................................119
6-22. Variation of the potential core velocity along the test section length for a) 18 /TSU m s= b) 30 /TSU m s= c) 37 /TSU m s= .................................................120
6-23. Locations of the hotwire measurement..................................................................121
6-24. Calibration curve showing the plot of mean velocity vs. mean voltage................122
6-25. Calibration curve corrected for flow temperature. ................................................122
6-26. Cubic fit to the calibration curve. ..........................................................................123
6-27. Turbulence spectra at location A. ..........................................................................123
6-28. Turbulence spectra at location B. ..........................................................................124
6-29. Turbulence spectra at location C. ..........................................................................124
xii
6-30. Narrow-band Spectra.............................................................................................127
6-31. 1/3rd Octave Band Spectra. ....................................................................................127
6-32. OASPL vs test section velocity .............................................................................128
6-33. Comparison of UF and Notre Dame tunnel background noise. ............................129
6-34. Narrow band inflow spectra. .................................................................................131
6-35. The influence of inflow microphone on the outflow spectra.................................131
6-36. Facility comparison of A-weighted in flow noise levels. ......................................132
6-37. Total power measured by the diffuser 2 microphones for a) 18 /TSU m s= b) 30 /TSU m s= c) 42 /TSU m s= . ...........................................................................133
6-38. Total power measured by the diffuser1 microphones for a) 18 /TSU m s= b) 30 /TSU m s= c) 42 /TSU m s= . ...........................................................................133
6-39. Total coherent power measured by the diffuser2 microphones for a) 18 /TSU m s= b) 30 /TSU m s= c) 42 /TSU m s= . .......................................................................134
6-40. Total coherent power measured by the diffuser1 microphones for a) 18 /TSU m s= b) 30 /TSU m s= c) 42 /TSU m s= . .......................................................................134
6-41. Schematic of the MISO model. .............................................................................136
6-42. Autospectra of the input and output microphones for 18 /TSU m s= . ..................138
6-43. Autospectra of the input and output microphones for 30 /TSU m s= . ..................138
6-44. Autospectra of the input and output microphones for 42 /TSU m s= ...................139
6-45. Ordinary coherence between the input microphones and output microphone for 18 /TSU m s= . ........................................................................................................139
6-46. Ordinary coherence between the input microphones and output microphone for 30 /TSU m s= . ........................................................................................................140
6-47. Ordinary coherence between the input microphones and output microphone for 42 /TSU m s= .........................................................................................................140
6-48. Comparison of the MISO model to the measured spectra for 18 /TSU m s= . ......141
xiii
6-49. Comparison of the MISO model to the measured spectra for 30 /TSU m s= . ......141
6-50. Comparison of the MISO model to the measured spectra for 42 /TSU m s= .......142
6-51. Total power for model and measured output for 18 /TSU m s= . ..........................142
6-52. Total power for model and measured output for 30 /TSU m s= . ..........................143
6- 53. Total power for model and measured output for 42 /TSU m s= ..........................143
6-54. Autospectra of the accelerometers attached to a) Fan slab b) Retainer wall for 18 /TSU m s= . ........................................................................................................144
6-55. Autospectra of the accelerometers attached to a) Fan slab b) Retainer wall for 30 /TSU m s= . ........................................................................................................145
6-56. Autospectra of the accelerometers attached to a) Fan slab b) Retainer wall for 42 /TSU m s= .........................................................................................................145
6-57. Transmission loss across the fan base for the x axis accelerometer for a) 18 /TSU m s= b) 30 /TSU m s= c) 42 /TSU m s= . ...............................................146
6-58. Transmission loss across the fan base and the building floor for the x axis accelerometer for a) 18 /TSU m s= b) 30 /TSU m s= c) 42 /TSU m s= . ..............146
6-59. Autospectra of the accelerometers attached to the vibration isolator for 18 /TSU m s= . ........................................................................................................148
6-60. Autospectra of the accelerometers attached to the vibration isolator for 30 /TSU m s= . ........................................................................................................148
6-61. Autospectra of the accelerometers attached to the vibration isolator for 42 /TSU m s= .........................................................................................................149
6-62. Transmission loss across the vibration isolator for the x axis accelerometer for a) 18 /TSU m s= b) 30 /TSU m s= c) 42 /TSU m s= . ...............................................149
6-63. Time response of a turning vane due to an impulsive impact. ..............................150
6-64. Normal incidence absorption coefficient for the acoustic liner.............................150
A-1. Plan view of the wind tunnel. .................................................................................162
A-2. Cross-section view of the wind tunnel....................................................................163
xiv
A-3. North elevation of the wind tunnel. ........................................................................163
B-1. Schematic of the inlet shape polynomial. ...............................................................164
B-2. Plot of contraction shape polynomial in the x-y plane. ..........................................168
B-3. Plot of contraction shape polynomial in the x-z plane............................................168
C-1. Velocity vector at the inlet exit plane. ....................................................................169
D-1. Wind tunnel flow path. ...........................................................................................175
D-2. Location of diffuser 1 and 2 designs on the Kline’s flat diffuser curves................178
F-1. Schematic of the chamber. ......................................................................................185
F-2. Equivalent electric circuit representation of the chamber flow...............................187
F-3. Variation of leakage ratio with the leakage flow resistance....................................190
F-4. Variation of the wall pressure differential with leakage area ratio for various leakage resistance ratios. .....................................................................................................191
F-5. Variation of the wall force with leakage area ratio for various leakage resistance ratios. ......................................................................................................................191
G-6. Deviation of pressure measurements from free field from chamber center in the Northeast direction. ................................................................................................193
G- 7. Deviation of pressure measurements from free field from chamber center in the West direction. .......................................................................................................194
G-8. Deviation of pressure measurements from free field from chamber center in the Northwest direction. ...............................................................................................195
G-9. Deviation of pressure measurements from free field from chamber center in the North direction. ......................................................................................................196
G-10. Deviation of pressure measurements from free field from chamber center in the Southwest direction. ...............................................................................................197
G-11. Deviation of pressure measurements from free field from chamber center in the South direction. ......................................................................................................198
G-12. Deviation of pressure measurements from free field from chamber center in the Southeast direction. ................................................................................................199
G-13. Deviation of pressure measurements from free field from chamber center in the East direction..........................................................................................................200
xv
BSTRACT
Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy
DESIGN, FABRICATION, AND CHARACTERIZATION OF AN ANECHOIC WIND TUNNEL FACILITY
By
Jose Mathew
May 2006
Chair: Louis Cattafesta Major Department: Mechanical and Aerospace Engineering
The design, fabrication, and characterization of an anechoic wind tunnel facility at
the University of Florida are presented. The objective of this research is to develop and
rigorously characterize an anechoic wind tunnel suitable for detailed aerodynamic and
aeroacoustic research. A complete tunnel design methodology is developed to optimize
the design of the individual components of the wind tunnel circuit, and modern analysis
tools, such as computational fluid dynamics and structural finite element analyses, are
used to validate the design.
The wind tunnel design is an “L-shaped”open circuit with an open jet test section
driven by a 300 HP centrifugal fan. Airflow enters the wind tunnel through a settling
duct with a honeycomb section and a set of four screens. An optimized, minimum length
(3.05 m) 8:1 contraction accelerates the flow into a rectangular test section that measures
0.74 m by 1.12 m by 1.83 m. Mach number similarity dictates the maximum velocity
xvi
attainable in the test section to be 76 m/s; thus the maximum Reynolds number based on
chord (chord=2/3 span) attainable is in the 3-4 million range. The flow leaving the test
section enters an acoustically treated and 2D diffuser that simultaneously provides static
pressure recovery and attenuates fan noise. The flow then turns a 90° corner with turning
vanes and enters a second diffuser. The flow leaving the second diffuser enters the fan
through a transition section.
The wind tunnel was characterized rigorously at speeds up to 43 m/s to ensure the
quality of the future aerodynamic and aeroacoustic measurements. The overall SPL from
100 Hz – 20 kHz ranges from 54.8 dB at 18 /m s to 75.7 dB at 43 /m s . The
freestream turbulence level has a value of 0.035 %, and the flow non uniformity in the
test section was found to be < 0.7 % for a test section speed of 17 /m s .
The outcome of this work is an anechoic wind tunnel with excellent flow quality,
low background noise, and the largest Reynolds number capability among university-
scale anechoic facilities in the US.
1
CHAPTER 1 INTRODUCTION
The goal of this research is to design, fabricate, and characterize an anechoic wind
tunnel facility at the University of Florida. An existing anechoic facility at the University
of Florida (Jansson et al. 2002) has been upgraded to an anechoic wind tunnel. The
purpose of this endeavor is to permit high quality fluid dynamic and aeroacoustic
experiments on airframe noise, with an initial focus on trailing edge noise. A research
flow facility with low turbulence levels, good flow uniformity, and low background noise
levels that can achieve high chord Reynolds numbers in the test section is essential in this
regard.
Strict regulations imposed by the Federal Aviation Authority (FAA, 2004) on the
noise from commercial aircraft have increased the emphasis on airframe noise, which is a
significant portion of the aircraft noise during approach and landing. A reduction in
aircraft noise will require attenuation of airframe noise produced by specific aircraft
components, such as airfoil trailing edges, landing gear, airfoil flaps and slats, and wing
tips. A fundamental understanding of the noise generation mechanisms will provide the
ability to model and predict the emitted noise and may enable researchers to devise
effective schemes to ultimately reduce airframe noise. However, appropriate experiments
conducted in an anechoic wind tunnel are required to achieve significant advances in this
regard. This chapter presents an overview of airframe noise and its various components,
a survey of other existing anechoic flow facilities, the motivation for this research, the
technical objectives and approach, and also provides an outline of this thesis.
2
Background
Commercial air traffic has been growing at such a fast pace that aircraft noise has
increasingly become an annoyance to the communities in close proximity to airports, and
the importance of aircraft noise reduction is now being realized by the international
community (Willshire 2001). The detrimental effects of aircraft noise include sleep
deprivation, irritability, reduced land resale value, and the delay in the growth of civil
aviation. Realizing these drawbacks, NASA has proposed a plan as a part of the Quiet
Aircraft Technology (QAT) program “to reduce the perceived noise levels of future
aircraft by 10 (decibels) dB from today’s subsonic aircraft and by 20 dB within twenty
five years” (Goldin 1997).
Figure 1-1 shows the various sources of aircraft noise. Aircraft noise consists
mainly of airframe noise and power plant noise. Power plant noise includes jet noise,
turbomachinery noise, and combustion noise, while airframe noise consists of noise due
to flaps, slats, landing gear, wing and tail, etc. Considerable research over the past three
decades has focused on the reduction of aircraft jet noise. Figure 1-2 (www.aia-
aerospace.org, 2004) shows the progress achieved in the development of quieter civilian
aircraft over the past 50 years. The turbofan engines of the present day are at least 20 dB
quieter than the turbojet engines of the early sixties. The use of ultra-high bypass ratio
turbofan engines and the successful implementation of liner technology has helped
mitigate jet noise to such an extent that airframe noise or non-propulsive noise has now
become a significant source of aircraft noise (Crighton 1995), especially during approach
and landing. The relative magnitudes of the various aircraft noise components during
approach are shown in Figure 1-3.
3
AIRCRAFT NOISE
AIRFRAMENOISE
POWERPLANT NOISE
Flap Slat LandingGear
Wing/Tail
JetNoise
TurbomachineryNoise
CombustionNoise
Figure 1-1. Aircraft noise sources.
During takeoff, the aircraft engine is operating at maximum thrust, and therefore jet
noise and fan exhaust noise dominates over other noise sources. However during
approach the aircraft engine is flying at low power and all the high lift devices and the
landing gear are fully extended, resulting in a greater contribution of airframe noise to the
total aircraft noise spectrum. Note that during approach, airframe noise is comparable to
the fan inlet noise, making them the primary noise sources during approach. In order to
design quieter airplanes, the physics behind the various airframe noise generation
mechanisms must be thoroughly understood.
Airframe Noise
Airframe noise is defined as the total aircraft noise minus the noise from the engine
and noise from engine-airframe interference (Lockard and Lilley 2004). Various sources
of airframe noise are annotated in Figure 1-4 (Golub et al. 2004).
4
1960 1965 1970 1975 1980 1985 1990 1995 2000
TurbojetsFirst Generation TurbofansSecond Generation Turbofans
20 dB
Entry into Service Date
Late
ral N
oise
Lev
elC
orre
cted
for A
ircra
ft Th
rust
Figure 1-2. Propulsive noise reduction through the ages.
The main sources of airframe noise are the flaps, slats and landing gear. Noise also
emanates from fuselage, wing, tail, landing gear cavities, etc.
A ‘clean’ full scale airframe with flaps, slats, and landing gear retracted generates
mainly broadband noise with the broadband peak located in the vicinity of several
hundred Hz (Smith 1989). An aircraft during approach has its flaps, slats and landing
gear extended, increasing the overall airframe noise levels by approximately 10 dB. The
landing gear noise is omnidirectional and has spectral characteristics higher in frequency
than the clean airframe (Smith 1989). Landing gear noise is caused in part by vortex
shedding over bluff bodies like wheels, axles, struts, shafts, etc. (Crighton 1995). At the
typical shedding frequency, the sound radiated has a 6U dependence on velocity, where
5
U is a typical velocity in the flow field. High-lift devices like wing flaps and slats
modify the spectral content of the clean airframe, tending to lower its characteristic
frequency as a result of extending the chord of the wing and inducing larger turbulence in
the wing wake (Smith 1989). Noise from high lift devices has been shown to exhibit a
5U dependence.
60 70 80 90 100 110
Total Aircraft Noise
Total Airframe
Jet
Turbine
Combustor
Aftfan
Inlet
EPNdB
P&W ADP Engine P&W 1992 Technology Engine
Figure 1-3. Relative magnitudes of the various aircraft noise components during landing.
There is a large body of literature available on the theory of trailing edge noise
from two-dimensional airfoils. Howe (1978) categorized trailing edge noise theories into
three categories based on a) Lighthill’s acoustic analogy, b) linearized hydrodynamic
equations, and c) ad-hoc models. All models predict a 5U dependence of the radiated
sound on the freestream velocity U . When turbulent boundary layer eddies convect past
6
the trailing edge of an airfoil ( chord, c λ< ), the acoustic scattering produces broadband
radiation to the farfield (Ffowcs Williams and Hall 1970; Crighton and Leppington
1971). However, if coherent vortex shedding is present (e.g., due to blunt trailing edges
at high angles of attack), tonal or narrowband noise is also present. Khorrami et al.
(2000) used highly resolved unsteady Reynolds Averaged Navier Stokes (URANS)
computations of a blunt trailing edge flow to reveal strong vortex shedding and
corresponding acoustic wave propagation from the trailing edge region. Blake and
Gershfeld (1989) earlier explained how tonal noise is generated when periodic vorticity
results due to an instability in the trailing edge wake. Furthermore, they identified the
broadband noise spectral component, occurring due to the convection of turbulence past
the trailing edge.
Nose LandingGear
Slats
Main LandingGear
Flaps
Vertical Tail
Horizontal Tail
Figure 1-4. Various airframe noise sources.
7
Several researchers have performed experimental studies of trailing edge noise via
unsteady surface pressure measurements and directional acoustic arrays. Brooks and
Hodgson (1981) used measured surface pressures near the trailing edge to arrive at a
correlation for the noise generated. Brooks and Marcolini (1985) used a cross-spectral
technique to determine noise sources from trailing edges and used the resulting data to
formulate scaling laws for trailing edge noise. More recently, Hutcheson and Brooks
(2002) have compared directional array measurements to a cross spectral method that
uses a pair of microphones on opposite sides of the airfoil. Macaraeg (1998) described a
fundamental investigation of airframe noise using extensive flow visualization, velocity
and noise measurements on a small-scale, part-span flap model. Kunze et al. (2002) have
measured trailing edge noise from flat plate geometries and devised a procedure to
distinguish the trailing edge noise component from the background noise.
All of these theoretical and experimental results have demonstrated the importance
of the following primary nondimensional parameters: Mach number, M∞ , chord
Reynolds number, cRe , angle of attack, c θ , t θ , (i.e., ratio of airfoil chord length, c , or
trailing edge thickness, t , to local boundary layer momentum thickness, θ ). In addition,
the structure of the turbulent boundary layer in the vicinity of the trailing edge (e.g.,
shape factor and nondimensional pressure gradient) is also important. Furthermore,
practical aircraft configurations have variable wing sweep, which leads to cross flow and
the development of three-dimensional boundary layers on the wing. The freestream
turbulence intensity and the farfield boundary conditions are also significant parameters
in experimental studies.
8
Since most previous research has focused on noise from flow over two-dimensional
airfoils, much less is known about the trailing edge noise from swept wings that exist on
modern commercial aircraft. It is anticipated that trailing edge sweep will have a
profound effect on the sound generation and directivity patterns due to the three-
dimensional nature of the boundary layer. Our goal is to develop a anechoic wind tunnel
facility that simulates a free field acoustic flight environment. The facility will enable us
to conduct future benchmark experiments to eventually understand the relation among
wing sweep, noise generation mechanisms, and trailing edge noise radiation patterns.
Typical experiments include the measurement of surface pressure fluctuations on the
trailing edge using flush mounted pressure sensors. Since wing sweep leads to the
development of a three dimensional boundary layer on the airfoil, crossflow
measurements of the three-dimensional boundary layer must also be made. Surveys in
the wake region must be conducted to estimate the amplitude and the frequency of vortex
shedding off the trailing edge. Acoustic measurements will include the measurement of
amplitude and directivity of the far field radiated noise from the trailing edge.
As a first step towards our goal, a high quality anechoic wind tunnel facility with
low turbulence and low background noise levels must be fabricated and characterized.
There are not many anechoic wind tunnels in the US where airframe measurements can
be conducted (Duell et al. 2002). A survey of existing anechoic wind tunnels in the
world was conducted prior to designing our wind tunnel and is summarized below.
Existing Anechoic Wind Tunnels
Anechoic wind tunnels are used extensively by both the automotive and aerospace
industries for scaled model testing. While the aerospace industry has focused on
improved aeroacoustic measurements with secondary emphasis on new anechoic wind
9
tunnels, the automotive industry has focused more on developing quieter anechoic wind
tunnels. The existing trend in the wind tunnel technology is to construct larger facilities
with lower background noise levels (Duell et al. 2002). The larger facilities can provide
higher Reynolds numbers and lower turbulence intensities in the test section, since the
turbulence intensities drops with increasing the contraction ratio of the inlet. However,
the cost of building these facilities is enormous, and also their power consumption is high
because the power required to run the wind tunnel fan scales with the area of the test
section and the third power of test section velocity (Pope & Harper 1966). The
maintenance of large facilities is also difficult. Larger facilities also require a larger fan
to operate, which in turn increases the background noise in the test section. As an
example consider the Daimler Chrysler wind tunnel facility located in Detroit, which is
used for automotive testing (Walter et al. 2003). The wind tunnel inlet has an entrance
area of 29 m2 and the facility itself occupies an area that spans 31,000 2ft . The fan
required to drive the facility requires a 6343 HP motor and the facility cost 37.5 million
dollars to build. The characteristics of existing anechoic wind tunnels are summarized in
Table 1-1. The facilities shown in the table include industrial tunnels, government
tunnels and university scale tunnels. These tunnels are used for automotive component
testing or aircraft component testing. The wind tunnel could be of the blower, blowdown
or the suction type. There are pros and cons for each design. The wind tunnel can also
be of the open circuit or the closed circuit type.
For a blower tunnel the fan is located upstream of the test section, and blows high
speed flow into the test section through an inlet contraction. Although the pressure in the
test section is atmospheric, the flow quality is generally not optimal. The blower tunnels
10
also suffer from a low frequency pumping effect due to spillage from the collector. This
low frequency phenomenon can potentially match the resonance frequency of the wind
tunnel structure and even damage the structure. A blowdown tunnel is a very simple
design which uses a nozzle fed by a compressed air storage tank to accelerate flow in the
test section. However to achieve steady flow over a large run time, the tank size is
prohibitive. A very common and efficient design is the suction tunnel, which uses a
downstream fan to pull the flow through the wind tunnel circuit. Although the pressure
drops to sub-atmospheric in the test section, the flow quality is generally higher than that
for a blower tunnel, for equivalent amounts of flow conditioning. A closed circuit tunnel
circulates the same air through the tunnel circuit, thereby turbulence can be reduced, as
opposed to an open circuit tunnel, where the scales of incoming atmospheric turbulence
are much larger. However, the additional duct work and space required in a closed
circuit design renders the cost much higher than that for an open circuit tunnel.
The quantities of most interest are the maximum Reynolds number, flow
uniformity, turbulence intensity, ( ./ 100u U′ ), where u′ is the root mean square value of
the axial component of the turbulent fluctuations, and the background noise level. The
maximum Reynolds numbers for the facilities are based on the test section hydraulic
diameter. A good facility should provide high Reynolds numbers, good flow uniformity,
low turbulence intensities, and low background noise levels. The background noise
levels have to be very low (preferably at least 10 dB below typical levels of
measurement) in an anechoic wind tunnel to make good quality acoustic measurements
(Duell et al. 2002). More details of this will be given in Chapter 3.
11
Table 1-1. Details of the existing anechoic wind tunnels.
Facility Circuit Type Drive Test Section
Type
Test Section Size (m)
Max Speed (m/s)
Max Re #
(million)
Flow Uniformity
Freestream Turbulence
Intensity
Background Noise Level
Langley QFF Open Pressure/
Vacuum Open 0.61 x 0.91 58 2.8 33 dB (1
kHz) (@ 18 m/s)
Boeing LSAF Open Blower Open 2.74 x 3.66 85 17 0.15 %
65 dBA (Outflow)
(@ 35 m/s) ONERA CEPRA
19, France
Open Fan Open 2 diam 130 17
NLR, Holland Open Fan Open 0.38 x 0.51 75 2.1
Closed 9.5 x 9.5 62 38 Closed 8 x 6 116 51 Closed 6 x 6 152 59 DNW,
Holland Closed Fan
Open 9.5 x 9.5 85 52 0.5% 80 dBA (@ 80 m/s)
NUWC, Newport Open Fan Open 1.22 diam 61 4.8
IVK, Stuttgart Closed Fan Open 5.8 x 3.87 80 24 0.3 %
(velocity) 71 dBA (@ 41.7 m/s)
Notre Dame Open Fan Open/Closed 0.61 x 0.61 28 1.1 0.04%
45 dB(1 kHz, third octave) (@
20 m/s) Audi,
Germany Closed Fan Open 3.94 x 2.8 83 17 0.3 % (velocity) 0.3 % 60 dBA
(@ 44.4 m/s) Open 5.56 x 3.34 54 14 0.4 % (SP) 0.34 % 63.7 dBA DTF
WT8, Detroit
Closed Fan Open 4.15 x 2.54 67 13.6
Open 4 x 7 53 17 0.13 % 66 dBA (@ 27.8 m/s) Nissan,
Japan Closed Fan Open 3 x 5 75 18 0.15 %
Daimler Chrysler AAWT, Detroit
Closed Circuit Fan Open 6.9 x 4.0 71 23.4 0.25 %
(SP) 0.16 %(@ 62.5 m/s)
62.3 dBA (@ 28 m/s)
Open 3 x 2.5 83 14.6 0.7 % (@ 90 m/s)
0.2 % (@ 100
m/s)
75.6 dBA (@ 83.3 m/s) RTRI,
Japan Closed Fan
Closed 5 x 3 111 27 0.5 % (@ 55 m/s)
0.2 % (@ 55 m/s)
Ford, Germany Closed Fan Open 20 m2 53 0.5 %
(velocity) 72 dBA (@ 38.9 m/s)
Virginia Tech Closed Fan Closed 1.83 x 1.83 80 9.4
75 dBA (Inflow)
(@ 32 m/s)
NASA Ames Open Fan Closed 24.2 x 12.1 154 160
86 dBA (Inflow)
(@ 52 m/s)
NASA Glenn Closed Fan Closed 4.6 x 2.74 68 15
65 dBA (Inflow)
(@ 22 m/s)
Motivation
Wind tunnels facilities are ubiquitous, whereas few anechoic wind tunnel facilities
exist in the US or rest of the world. At a university level, currently only two such facility
12
exists in the US. One of them is the subsonic, low turbulence anechoic wind tunnel at the
University of Notre Dame, for which there is established data (Mueller et al. 1992).
There is also a partially anechoic wind tunnel facility at Virginia Tech, which is currently
being upgraded from an aerodynamic stability tunnel to an anechoic facility (Smith et al.
2005). As such, the Notre Dame facility represents the benchmark for comparison. In
particular, the maximum velocity attainable in the Notre Dame wind tunnel is 28 m/s, and
hence the maximum Reynolds number based on the test section hydraulic diameter is
1.1.106. The test section of the Notre Dame facility measures 0.6 m by 0.6 m (24” by
24”). The inlet contraction has a contraction ratio of 20 and it is 4.26 m long (14 ft), and
is therefore rather large and expensive, but does provide freestream turbulence levels of
0.04%.
For the industrial and governmental wind tunnel facilities, airframe noise
measurements are often limited by scheduling and budgetary constraints. Our goal is to
build a facility that offers high quality airframe noise and flow measurements at relatively
low cost (< $200,000) to enable detailed research studies. Current
theoretical/computational studies require uncontaminated farfield noise spectra and
surface and flow field measurements. The goal of the facility will is to provide these
measurements.
Thesis Objectives
The Reynolds number based on wing chord for commercial aircraft vary anywhere
from 1.107-3.107 . Typical chord Reynolds numbers for the experimental trailing edge
noise measurements are in the 1–3 million range (Yu & Joshi 1979; Brooks & Hodgson
1981; Brooks & Humphreys 2003). Our objective is to develop a university scale,
anechoic flow facility that facilitates aeroacoustics research at Reynolds numbers higher
13
than what is currently attainable. Therefore for our facility, the maximum chord (chord
=2/3 span) Reynolds numbers attainable in the test section should be in the 3.106-4.106
range. The corresponding maximum velocity of the flow desired in the test section
corresponds to typical aircraft approach speeds and is approximately 76 m/s (250 ft/s,
M=0.22) to achieve Mach number similarity. The flow uniformity in the test section
should be high (flow non-uniformity < 1%). The turbulence levels for the flow in the test
section should be less than approximately 0.08%. This is to ensure that the trailing edge
noise spectra will not be contaminated by noise generated due to the impingement of
freestream turbulence on the leading edge of an airfoil model. This is also helpful while
studying trailing edge noise due to laminar boundary layers. The propagation of external
noise (e.g., due to the fan) into the test section has to be minimized to maintain low
background acoustic noise levels in the test section to accurately simulate an acoustic free
field. As mentioned above, the background noise should be at least 10 dB below
anticipated noise spectral levels. The facility should have an open jet test section, so as to
enable measurements of far field noise spectra and to minimize boundary layer noise. An
open jet test section also allows easy access to the airframe models in the flow. The
disadvantage of an open jet test section are the deflection of the shear layer (Brooks et al.
1984), leading to a difference between the geometric and effective angle of attack, and
the refraction of sound by the shear layer (Amiet, 1978). The vibrations from the fan
have to be isolated from the facility. This is to ensure that the noise due to the fan
induced vibrations do not contaminate the background noise spectra inside the test
section.
14
Our objective is to design and fabricate an anechoic wind tunnel facility that meets
these requirements. A tunnel design methodology has to be developed to optimize the
design of the wind tunnel circuit to meet these constraints within the allotted space and
budget. Various modern computational tools (e.g., CFD and FEA) will be used to
facilitate and validate the design of wind tunnel components, such as the inlet
contraction, diffuser, turning vanes and transition section. The wind tunnel has to be
fabricated and then rigorously characterized by making aerodynamic and acoustic
measurements. Beyond the expected outcome of a state-of-the-art university-scale
anechoic flow facility, the validated design procedure is expected to be of general interest
for wind tunnel designers.
Technical Approach
• Upgrade the existing UF anechoic chamber (Jansson et al. 2002) to an anechoic
wind tunnel that can provide unique flow capabilities.
• Validate the individual component design using CFD.
• Fabricate the individual wind tunnel components.
• Characterize the anechoic wind tunnel by making aerodynamic and acoustic
measurements that include
o Flow uniformity measurements
o Shear layer growth measurements
o Static pressure measurements in the tunnel circuit
o Turbulence intensity measurements
o Background noise measurements
o Vibration measurements
Thesis Organization
The remaining chapters of this thesis are organized as follows. Chapter 2 describes
the details of the existing anechoic chamber at UF that will be upgraded to the anechoic
15
wind tunnel facility. It also deals with the experimental characterization of the anechoic
chamber. Chapter 3 describes the design procedure adopted for the individual wind
tunnel components, starting from the settling chamber all the way to the drive fan
selection and duct acoustic treatment. Chapter 4 discusses the fabrication of the wind
tunnel facility. Chapter 5 deals with the setup and procedure for the experiments
undertaken to characterize the facility. Chapter 6 deals with the experimental results and
discussions concerning the facility characterization. Chapter 7 provides a summary, key
conclusions, and offers suggestions for future work.
16
CHAPTER 2 ANECHOIC CHAMBER
This chapter discusses the details of the UF anechoic chamber that will house the
wind tunnel. The purpose is to report on the experimental validation of the anechoic
chamber. Therefore, both free field and jet noise characterization studies of the facility
were conducted in order to validate the anechoic behavior of the chamber from 100-
100,000 Hz. Since the anechoic chamber will be upgraded to the anechoic wind tunnel
facility, it is important to validate the performance of the anechoic chamber to ensure that
good quality aeroacoustic measurements can be made in the wind tunnel. The details of
the facility as well as its characterization are given in the sections below.
Facility Description
The construction of the anechoic chamber was completed in the fall of 2001 by
Eckel Industries Inc. The US Air Force Office of Scientific Research provided the
financial support for the work. The initial objectives of the anechoic chamber were to
enable research in the areas of aeroacoustics, structural acoustics and industrial
noise/vibration control. The chamber was also equipped with a cold air jet, which
facilitated scaled aeroacoustic testing in an anechoic environment. A schematic of the
facility is shown in Figure 2-1 (Jansson et al. 2002; Sydhoff 2003). The University of
Florida anechoic chamber is a room contained within a noise enclosure to minimize
disturbances due to ambient noise and vibration. The inner dimensions from wedge tip to
wedge tip of the anechoic chamber are 5.5 m long by 5.0 m wide by 2.3 m high. The
outer dimensions are 6.92 m wide by 9.96 m long by 4.26 m high. The wedges are
17
constructed from fiberglass with cloth covers. The individual wedges are 0.81 m high by
0.61 m wide, and they have a maximum thickness of 0.35 m at the base. The wedges are
made of fiberglass enclosed in a woven fiberglass cover, enclosed by a steel mesh screen.
A schematic of the wedges along the north wall of the chamber is shown in Figure 2-2.
The floor wedges are housed in carts with removable metal grates along the top to allow
walk-in access. With the floor wedge carts removed, the semi-anechoic height is 3.3 m.
The wedges are designed to achieve a low-frequency cut-off of 100 Hz, which is the
frequency at which the energy absorption coefficient drops below 99% or the pressure
reflection coefficient exceeds 10%. The anechoic zone parallel to the floor at the
designed cut-off frequency of 100 Hz is 3.76 m by 3.3 m. The wall panels of the
chamber are also acoustically treated, aiding in the reduction of background noise inside
the chamber. A cross sectional view of the wall panel is shown in Figure 2-3. The wall
panels are made of 0.1 m thick fiberglass section covered in a woven fiberglass cloth
enclosed within a perforated metal sheet of 2.5 mm thickness.
For jet aeroacoustic applications, the chamber has intake and exhaust plenums on
opposite ends. The wedges along the plenum walls have openings to allow entrained
flow to pass through the chamber. Each plenum is itself a noise enclosure with flow
silencers to suppress ambient noise. The intake plenum also has adjustable flow restrictor
panels to assist in the even distribution of the entrained flow. A variable-speed fan (6000
cfm), downstream of the exhaust plenum silencers, assists in pulling the entrained flow
through the chamber. Equation Chapter 2 Section 1
18
100 Hz Anechoic zone
3.3 m
3.76 m
7.4
m
6.9 m
Figure 2-1. Schematic of the original UF anechoic chamber.
The intake plenum also houses the intermediate jet reservoir that is plumbed into a
compressor facility outside the building through two control valves. The dual-screw
compressor, rated for 28.3 m3/s at 1.4 MPa can continuously feed a perfectly expanded
2.54 cm diameter Mach 2 jet.
19
Floor Grates
Traverse
‘Cut out’ forentrained air
Figure 2-2. Schematic of the wall wedges.
10.1 cm
Fiberglass
Front Back
2.5 mm
Figure 2-3. Cross sectional view of the chamber wall panel.
20
Two 17 m3 storage tanks permit blowdown testing. Nozzle flow rates are set with
PC-controllable pneumatic valves. Individual nozzles connect to the intake plenum
reservoir via a 15 cm diameter supply pipe. The supply pipe diameter was chosen to
minimize pressure losses and assure a large reservoir to nozzle exit area ratio. Within the
pipe and upstream of the nozzle, flow conditioning honeycomb and screens are installed.
The jet exhaust is captured by a 1 m x 1 m acoustically-treated bell-mouth and a silencer
that extends through the exhaust plenum.
Free Field Characterization
Free field characterization of the chamber was performed in accordance with the
ISO 3745 standard (ISO 3745, 1977) from 100-20000 Hz. Two different noise sources
were located at the center of the anechoic chamber and detailed free field measurements
were made using 1/8 in. condenser microphones (B&K Type 4138). A B&K Omnisource
was used as the source for characterizing the chamber in the 125 Hz-4000 Hz frequency
range. For higher frequencies in the 5 kHz-20 kHz range, a JBL 2426H compression
driver with a pipe mount to simulate an omnidirectional monopole source. A total of 24
measurement surveys were chosen with between 28 and 32 measurement points in each
array depending on the length form source to wedge tip in that direction (Sydhoff, 2003).
All eight directions around the source were covered (N, NW, W, SW, S, SE, E, NE) and
in each direction three heights were covered, form the center of the source towards the
ceiling, middle and floor (up, mid, low), using a custom-designed B&K 5-DOF traverse
(see Figure 2-4).
The resulting SPL levels were compared to the theoretical free field decay, which
has a value of 6 dB per doubling of distance (Blackstock 2000). Figure 2-5 and Figure 2-
6 summarize the measurements for various frequency bands along the N and NE
21
directions. For the data shown, nearly all of the measurements are within tolerance (see
Appendix G for measurements along other directions). However, in the NE direction, the
250, 4000, and 5000 Hz fall out of tolerance near the double doors. Possible causes for
the tolerance violations near the corners at 250 Hz and near the NE corner at 4000 and
5000 are exposed metallic components, such as door handles, door hinge posts, the
traversing system structure, etc.
North Wall
South Wall
Figure 2-4. Measurement array paths in the anechoic chamber.
A later set of measurements were made (Sydhoff 2003), where the exposed metallic
parts, including the 5-DOF traverse, were covered with an acoustic absorbent material
called Nomex (Tex Tech Industries). The results, summarized in Table 2-1, revealed that
the discrepancies between the theoretical and experimental free field decay have now
been brought within the desired tolerance (< 1 dB) of the ISO 3745 standard. The
location up, mid, down, stands for the fact that the measurements are made from the
center of the chamber to the top most point, midpoint, and the bottom most point of the
chamber walls. Furthermore, our background noise floor measurements (Jansson et al.
22
2002) achieve the noise floor specifications of the B&K 4138 microphones (less that 30
dB).
Direction: North
0
5
10
15
20
25
30
35
40
45
0.0 1.0 2.0x (m)
Δ S
PL (d
B) 2000
5000
4000
6300
500
1000
250
125
100
OctaveBand
Δ 5dB
0 dB
{
2.74
Figure 2-5. Deviation of pressure measurements from free field from chamber center towards bell-mouth.
Direction: NorthEast
0
5
10
15
20
25
30
35
40
45
0.0 1.0 2.0 3.0x (m)
S
PL (d
B) 2000
5000
4000
6300
500
1000
250
125
100
OctaveBand
Δ 5dB
0 dB
{
3.72
`̀̀̀
Figure 2-6. Deviation of pressure measurements from free field from chamber center
towards Northeast corner of room with double door.
23
Table 2-1. SPL deviation errors for northeast directions, before and after treatment. Location Octave Band [Hz] Without Wrapping [dB] Traverse Partially Wrapped [dB]
1000 0.33 0.23 2000 1.11 0.65
Up 4000 1.03 0.84
1000 0.47 0.40 2000 0.98 0.43
Mid
4000 0.61 0.40
1000 0.70 0.73 2000 0.46 0.71
Dow
n
4000 0.62 0.30
Jet Noise Characterization
The purpose of jet noise characterization is to measure jet noise over a wider
frequency range (greater than could be achieved with the sound source) suitable for
scaled aeroacoustic testing. The results are then compared with well established subsonic
jet noise scaling to determine if the chamber operates as expected. The importance of
obtaining high quality data and the issues associated with it are discussed by
Vishwanathan (2002). This method exploits the universal fine-scale similarity noise
spectrum associated with acoustic radiation at 90o with respect to the incoming axis of a
subsonic axisymmetric turbulent jet. Tam et al. (1996, 2000) discuss the two self-similar
components of turbulent mixing noise, namely the fine-scale spectrum alluded to above
and the large-scale component that is dominant in the downstream quadrant close to the
jet axis. By measuring the sound produced by a subsonic axisymmetric turbulent jet and
comparing the results with the universal similarity spectrum, unwanted facility noise
sources can be identified and reduced.
24
Four B&K 1/4 in. Type 4939-A011 free-field microphones with B&K Type 2633
preamplifiers and B&K Type 2804 power supplies were used for measuring the jet noise
spectra. The microphones possess a specified frequency range of 4 Hz to 100 kHz. The
dynamic range of the microphones was 28 to 167 dB. Prior to taking jet noise
measurements, the microphones were calibrated using a B&K Type 4228 pistonphone
that provided a nominal amplitude of 124.7 dB at a frequency of 251.2 Hz.
Both third-octave band and narrow band noise spectra of the jet were measured.
Data acquisition was carried out using an HP E1433A VXI system. An ANSI S1.11-
1986 compliant LabVIEW® Third Octave Analyzer was used to measure the third-octave
band jet noise spectra. The frequency span for the third octave measurement was set at
76.8 kHz, which permitted third-octave bands up to 63 kHz. Jet noise data from the
microphones were acquired and processed in a real-time continuous mode using linear
averaging of 8192 samples in each block. Approximately 90 averages were performed
before filling the data acquisition system buffer. These acquisition and processing
parameters resulted in statistically converged spectra.
The schematic of the experimental test setup is shown in Figure 2-7 and Figure 2-
8. The four microphones were aligned at azimuthal angles of 90°, 110°, 130°, and 140°
with respect to the incoming axis of the jet. The radial distances from the nozzle exit
plane to the microphones are indicated in Figure 2-7. The microphones were attached to
the traverse rail, which runs along the West wall ceiling of the anechoic chamber. The
microphones were pointed towards the exit plane of the nozzle using a laser pointer.
Correct alignment of the microphones in this manner is consistent with free field
measurements from a point source located at the nozzle exit plane. In order to minimize
25
the effects of scattering, the protective grids of the microphones were removed prior to
any measurements. 7'8"
18'
FloorFloor
CeilingCeiling
SIDE VIEW
2.75"72.13"
140 degMic
130 degMic
110 degMIc
90 degMic
r/D=70.85
r/D=73.70r/D=84.90
r/D=97.13
3.125"
CIrcular Nozzle
JETTo Stgnchamber
traverse rail
11.25"13.38"
Figure 2-7. Side view schematic of the jet noise measurements.
Acoustic noise measurements were performed for seven different jet Mach numbers
ranging from approximately 0.3 to 0.9. An axisymmetric jet nozzle of inner diameter of
35.6 cm and wall thickness of 3.2 mm was used (Figure 2-9). The stagnation pressure
inside the jet reservoir and the static pressure inside the anechoic chamber were measured
using a static pressure ring to compute the nozzle pressure ratio and jet exit Mach
number. The actual Mach number was subsequently verified by mounting a pitot probe
at the exit of the nozzle, and measuring the pressure ratio between nozzle exit stagnation
pressure and the chamber static pressure. The jet pressure ratio was maintained via
software control of the supply valves using LabVIEW®.
26
44.75"
88.45"
120.89
16'6"
92"
TOP VIEW
140 deg
130 deg
110 deg
90 deg
6.5"
To Stagnchamber
West Wall
East Wall
traverse rail
JET
Figure 2- 8. Top view schematic of the jet noise measurements.
To stagnationchamber
Top View Front View
0.28 m3.2 mm
Ø 35.6 mm
Figure 2-9. Schematic of the jet nozzle.
Various configurations were tested in an effort to determine the influence of the
following: floor grating and wedge configurations, ambient noise, traverse mechanism
reflections, entrainment fan noise contamination, and compressor noise and vibration.
This section summarizes the most pertinent results.
27
The chamber contains removable floor wedges to simulate a semi-anechoic test
environment. The floor wedges are installed in carts with casters. Each cart has a
removable floor grating. Not surprisingly, the optimum configuration required removal
of the wedges from the carts, thereby maximizing the distance from the jet axis to the
wedge tips on the chamber floor (0.44 m). However, a row of carts on the outer rim of
the chamber were subsequently reinstalled to permit access to the chamber and the
microphones. Experiments showed that the peripheral row of carts did not affect the
results.
Figure 2-10 contains the measured noise floor of the 1/4 in. microphone at 90o with
respect to the jet axis and the third-octave band B&K noise-floor specification at 1 kHz.
The measured noise floor is within 1 dB of the B&K specification. This indicates that
sufficient suppression of ambient noise has been achieved for aeroacoustic applications.
The noise floor is below 36 dB from 100 Hz to 63 kHz. The data at the lowest Mach
number = 0.3 are clearly limited at the lowest and highest frequencies by the noise floor
of the microphone measurement.
Figure 2-10 also shows the results obtained for the microphone located at 90o as a
function of the exit jet Mach number. The nozzle pressure ratio was varied to achieve
Mach numbers from 0.3 to 0.9 in steps of 0.1. Superimposed on the plot is the fine-scale
noise similarity spectrum, ( )pG f f , presented in Tam and Zaman (2000), corresponding
to an peak frequency pf ≈ 3.2 kHz using a narrow-band spectrum analyzer. As
previously noted, this also corresponds to a measured resonant frequency of the untreated
traverse support rails (Sydhoff, 2003). Note that the similarity spectrum was integrated
over third-octave bands and plotted to compare with the experimental data.
28
102
103
104
105
−10
0
10
20
30
40
50
60
70
80
90
Frequency (Hz)
dB r
e 20
μP
a
M=0.3M=0.4M=0.5M=0.6M=0.7M=0.8M=0.9G
noise floor B&K noise−floor specification
Figure 2-10. Cold jet noise data measured at 90o to the jet axis at 83.5 jet diameters for
various jet Mach numbers. Exhaust fan is off. Compressor on.
102
103
104
105
40
45
50
55
60
65
70
75
80
85
90
Frequency (Hz)
dB r
e 20
μP
a
90o
140o
GF
Figure 2-11. Cold jet noise data measured at 90o and 140o to the jet axis at 83.5 and
114.5 jet diameters, respectively, at M=0.9. Exhaust fan is off. F and G are the large- and fine-scale similarity third-octave band spectra .
29
Several conclusions can be drawn from Figure 2-10. First, the data at the lower
Mach numbers are clearly corrupted, particularly at high frequencies. The low Mach
number data reveal high frequency noise contamination, the source of which was not
identified. Second, the data in the mid-frequency (2-10 kHz) range are characterized by
amplitude ripple above the experimental uncertainty. Various contaminating noise
sources were systematically identified and eliminated, leading to Figure 2-10. However,
acoustic absorbing material was not yet available at the time of these tests to wrap the
traverse mechanism, which served as the mount for the microphone holders. We suspect
that reflections from the traverse rails are the primary cause of the observed ripple in the
data.
Despite these issues, the data exhibit some expected trends. The data follow the
fine-scale self-similar spectral shape. It should be noted that these free-field microphone
data have not been corrected for atmospheric attenuation, which is significant at high
frequency. Preliminary estimates of corrections based on the work of Shields and Bass
(1977) indicates adjustments ranging from ~1 dB near 20 kHz to ~7 dB at 63 kHz.
Figure 2-11 compares the results between the 90o and 140o microphones for M=0.9.
Superimposed on the figure are the large-scale (F) and fine-scale (G) noise similarity
spectra. As expected, the results show that the 140o location in the downstream quadrant
is significantly influenced by the large-scale noise, while the opposite is true for the 90o
data.
30
102
103
104
105
30
40
50
60
70
80
90
Frequency (Hz)
dB re
20
μPa
M=0.3M=0.4M=0.5M=0.6M=0.7M=0.8M=0.9G
Figure 2-12. Cold jet noise data measured at 90o to the jet axis at 83.5 jet diameters for
various jet Mach numbers. Exhaust fan is operating at max speed (~6000 CFM).
102
103
104
105
30
35
40
45
50
55
60
65
70
75
80
Frequency (Hz)
dB r
e 20
μP
a
blowdowncontinuous
Figure 2-13. Comparison between blowdown (compressor off) and continuous
(compressor on) operating conditions for approximately identical flow conditions (M=0.7±0.01) (measured at 90o to the jet axis).
Figure 2-12 shows the effects of the entrainment fan operating at maximum volume
flow capacity. By comparing directly with Figure 2-10, the noise introduced by the fan is
significant at both very low and very high frequencies, as well as approximately 3.2 kHz,
corresponding to a resonance frequency of a traverse support beam. This demonstrates
31
the need for better acoustic/vibration control for the fan and traverse. However in the
anechoic wind tunnel facility, both traverse and the exhaust fan will be removed and/or
modified as required.
A common approach to minimizing contamination due to flow noise is to run in a
blowdown mode, in which the storage tanks are filled and the compressor is either
isolated or (as in the case here) shut off completely. Figure 2-13 shows a comparison
between blowdown (compressor off) and continuous operation (compressor on) modes.
The two test conditions are approximately the same (M=0.7±0.01, 90o) as are the
measured noise spectra, although the blowdown run introduces some very low frequency
(< 100 Hz) variations that are not visible in the plot. Otherwise, the data are nearly
identical. The operation of the compressor may have little effect on the acoustic data due
to the large distance between the compressor, the large pipe diameters, and the jet flow
conditioning. In any case, this result suggests that the compressor need not be turned off
or isolated during acoustic testing.
Future noise and aerodynamic experiments in the new acoustic wind tunnel stand to
benefit from the low cutoff frequency of 100 Hz and the low background noise levels in
the chamber. The results from the free field and jet noise characterization establish the
validity of the use of UF anechoic chamber for experiments involving acoustical
measurements. Free field measurements inside the chamber were shown to be compliant
with free field. The resonance of the untreated traverse was shown to contaminate the
free field and jet noise measurements in the 4- 5 kHz range. The operation of the exhaust
fan was shown to affect measurements in both low and high frequency ranges. These
32
components that affect the acoustic measurements inside the chamber will be removed
and the chamber will be re-evaluated after the final assembly of the wind tunnel.
33
CHAPTER 3 DESIGN OF THE ANECHOIC WIND TUNNEL
This chapter deals with the details of the design of the various wind tunnel
components. It begins with the settling chamber and follows the tunnel circuit all the
way to the fan that drives the tunnel. Equation Chapter 3 Section 3
Design Criteria
The anechoic wind tunnel will facilitate the measurement of good quality
aerodynamic and airframe noise data. An open jet test section will be well suited for this
purpose as far field noise measurements can be made. The disadvantage of an open jet
facility is that it has greater static pressure losses than a closed jet and the measurements
have to be corrected for the refraction of sound in the free shear layer. The background
noise in an open jet facility is much lower than a closed jet facility. This is due to the fact
that measurements in a closed jet facility suffer from noise reflection from the walls of
the test section and also there is the contribution of noise from the boundary layer
developing on the test section walls.
The main requirement for our facility is that Reynolds number based on chord
should be in the 3-4 million range. Mach number similarity must be achieved in the test
section. We have chosen a mach number of 0.22M = and a corresponding test section
velocity of 76 /m s . A knowledge of the Reynolds number (based on chord) and mach
number helps to establish the test section size. The maximum allowable chord size is
2/3rd the width of the test section to account for the developing shear layer from the edges
of the contraction. The flow non-uniformity in the test section has to be 1%< . The
34
turbulence intensity in the test section has to be low ( 0.08 %TI < ). This is to ensure
that the trailing edge noise spectra is not contaminated by the noise generated due to the
impingement of freestream turbulence on the airfoil leading edge. This would also help
us to make noise measurements from laminar boundary layer over the trailing edge. The
background noise levels in the chamber have to be at least 10 dB below the prominent
noise level.
A flow chart for the wind tunnel design is given in Figure 3-1. The two main
constraints for this design are the total budget for the project and also the dimensions of
the existing anechoic chamber that will be upgraded to the anechoic wind tunnel. The
test section size is fixed by the Reynolds number and mach number limit. We then
choose the largest contraction ratio for the inlet contraction to design a contraction that
can fit within the anechoic room. Care must be taken to ensure that no flow separation
occurs inside the inlet. The inlet size decides the size of the settling duct that houses the
honey comb and screens. The honeycomb affects the flow uniformity and the screens
dictate the turbulence intensity in the test section. A knowledge of the static pressure
drop in the test section helps to design the diffuser for pressure recovery. The room
dimensions constraints and the large pressure drop requires us to design 2 diffusers for
this tunnel, joined together by a 90° corner. Turning vanes have to be installed in the
corner to prevent separation. A rectangular to round transition section also needs to be
designed to attach diffuser 2 to the fan. The fan selected provides a volume flow rate and
corresponding static pressure loss compensation. The fan that drives the tunnel is usually
the costliest item and we want the largest fan possible that can fit the budget. Care must
be taken to ensure that flow separation is avoided in the duct work, which includes the
35
two diffusers, corner and the transition piece. The duct work also must be lined on the
inside with acoustic absorbent material to minimize the test section background noise.
U=76 m/s, Re=3-4 million
Fan
Contraction Diffuser 1
Corner/Vanes
TARGET
Settling DuctHoney Comb
Screen
Transition
Budget
TestSection
Size?
Separation Acoustic
Treatment
Diffuser 2
Vol flow rateSG
TI<.08%,Non-uniformity<1%
Figure 3-1. Wind tunnel design flow chart.
Overall Layout
The different views of the wind tunnel are shown in appendix A. The plan view of
the wind tunnel is shown in Figure 3-2.
36
Table 3-1 summarizes the details of the various tunnel components. The overall
dimensions of the room that houses the wind tunnel measure 16 m long by 8.7 m wide
by 4.3 m high. At the entrance to the wind tunnel is a 1.67 m long settling chamber that
houses the honeycombs and the screens. The flow enters the settling chamber through an
inlet bellmouth. The entrance of the settling chamber measures 2.57 m by 2.57 m
(8.42 ' by 8.42 ' ) square. A 9.5 mm cell, 95 mm long (L/D=10) honeycomb sits at the
entrance to the settling chamber. The honeycomb section is held in place by a 24 mesh
per inch (67% open area) screen on either side. This is followed by 4 screens with 67%,
62%, 62%, and 60% open area and no of mesh per inch of 24, 32, 46, and 56,
respectively.
The inlet follows the settling chamber and it accelerates the flow entering the test
section. A matched 3rd-order/8th-order polynomial forms the wall shape of the inlet. The
inlet is 3.05 m long and it has a contraction ratio of 8. The exit of the inlet section
measures 0.74 m by 1.12 m . The test section is of the open jet type and it has a total
length of 1.83 m . The test section measures 0.74 m by 1.12 m and the maximum
velocity attainable is 76 /m s ( 250 ft / s ). The flow exiting the test section enters the jet
collector. The ratio of jet collector area to the area of the test section is 1.174.
There are two diffusers in this design to achieve the required pressure recovery.
Both diffusers are 2D in shape. Diffuser 1 immediately follows the collector section and
has entry dimensions of 0.8 m by 1.2 m and a total length of 3.58 m . The exit of
diffuser 1 measures 1.49 m by 1.2 m . The total included angle, θ for the diffuser is
10.94º. The corner section that immediately follows diffuser 1 turns the flow 90º, leading
to diffuser 2. A set of 20 turning vanes is installed in the corner region to ensure smooth,
37
separation free flow. The turning vanes have a chord length of .15 m . Diffuser 2 has a
total length of 5 m and has exit dimensions of 2.2 m by 1.2 m . The included angle for
this diffuser is 11º. Diffuser 2 exits to a rectangular to round transition section that is
1.57 m long. The transition section joins the exit of diffuser 2 to the fan. The fan has a
circular inlet with a diameter of 1.95 m . A Twin City, single width acoustafoil fan that
can deliver a maximum flow rate of 147000 cfm ( 369.4 /m s ) against a static pressure
recovery of 28" H O (1992 Pa ) is used to drive the tunnel.
Figure 3-2. Plan view of the wind tunnel.
38
Table 3-1. Summary of the wind tunnel design. Dimension Comments Design goals
Settling Chamber 1.67 m long Cross section:
2.57 m by 2.57 m
9.5 mm cell, 95 mm long HC
6 screens: 3 with 24M 67% OA 1 with 32M 62% OA 1 with 46M 62% OA 1 with 56M 60% OA
Inlet Entrance Size 2.57 m by 2.57 m
L=3.05 m CR=8
Test Section 0.74 m by 1.12 m
L= 1.83 m Open Jet
Diffuser 1 Length 3.58 m θ =10.94º
AR=1.86 L/H=4.49
Diffuser 1 Exit Size 1.49 m by 1.2 m
2D, flat, No separation
Diffuser 1 Liner thickness
30.5 cm
Turning Vane Chord 15 cm No Separation
Diffuser 2 Length 5 m θ =8.12º
AR=1.49 L/H=2.59
Diffuser 2 Exit Size 2.2 m by 1.2 m
2D, flat, No separation Dh=1.57 m
Diffuser 1 Liner thickness
30.5 cm
Fan Inlet Diameter 1.95 m Twin City Blower 147000 cfm SG
28" H O Transition Section
Length 1.57 m L=1 Dh
AF=1.11
Re= 3-4 million
TSU = 76 /m s ( 0.22M = )
0.08 %TI <
Flow Non-uniformity<1 %
Background
Noise= 10 dB below
The detailed design of the individual components of the wind tunnel is described in
the sections below.
39
Settling Duct/Honeycombs/Screens
The settling duct has a total length of 1.67 m . The settling section serves the
purpose of straightening the flow as well as attenuating some sound disturbances in the
incoming flow. The settling section houses the honeycomb and screens. These
components aid in increasing the quality of the flow. The settling duct should be long
enough for the incoming turbulence to dissipate, while minimizing the boundary layer
growth.
Honeycomb is installed to straighten the flow as well as to attenuate some high
frequency noise. Honeycomb removes swirl from the incoming flow and minimizes the
lateral variations in mean velocity (Mehta & Bradshaw 1979). It breaks up the large
scale eddies in the incoming flow and also aids in reducing the magnitude of the lateral
turbulent velocity fluctuations. The yaw angle for the incoming flow should be less than
10° to avoid stalling of the honeycomb cells.
Honeycomb comes in different shapes, including circular, square, hexagonal etc.
Among these, hexagonal is usually the cross-sectional shape of choice, as it has the
lowest pressure drop coefficient (Pope & Harper 1966). The ideal length to hydraulic
diameter ratio of the honeycomb cells should be between 7 and 10. We have chosen a
value of 10 for the length to diameter ratio in our design. A longer honeycomb section
would lead to a larger boundary layer growth and hence leads to more pressure drop.
Mehta & Bradshaw (1979) also states that the cell size should be smaller than the
smallest lateral wavelength of the velocity variation, which is roughly equivalent to 150
cells per settling chamber diameter. The dimensions of each cell are given in Figure 3-3
(9.5 mm (3/8") width is a typical value of honeycomb width for similar tunnels). The
40
total length of the section is 95 mm ( 3.75") and the wall thickness of the honeycomb is
0.4 mm . The maximum Reynolds number based on the cell hydraulic diameter is 5800,
indicating that the flow is turbulent. The flow at the exit of the honeycomb section is not
fully developed, as the length of the honeycomb section is smaller than 25 D, which is a
requirement for fully developed flow in a turbulent pipe of diameter D. The honeycomb
section is constructed of aluminum, for the sake of structural rigidity. The honeycombs
have to be cleaned periodically to prevent dust from clogging the cells.
3/8"
3.75"
Figure 3-3. Schematic of the honeycomb section.
Stainless steel screens are also placed in the settling duct for the reduction of
turbulence levels of the incoming flow. Screens break up the large scale turbulent eddies
into a number of small scale eddies that decay rapidly. Schubauer et al. (1950) states that
the Reynolds number based on the screen wire diameter should be less than 60 to prevent
additional turbulence generation due to vortex shedding. The schematic of the screen
design chosen is shown in Figure 3-4. Shown in the figure is a square section with N
mesh/inch. The diameter if the individual screen wires is d . The solidity S , of the
screen is the flow area blocked by the wires of the screen and is defined by
41
2 22 .S Nd N d= − (3.1) Once the solidity is known, the percent open area for the flow through the screens
is given by the formula
Open Area (1 )100.S= − (3.2) The Reynolds number based on the wire diameter is given by the formula
/Re ,TSscreen
U CR dυ
= (3.3)
where TSU is the velocity in the test section, CR is the contraction ratio, and υ is the
kinematic viscosity of air. Schubauer et al. (1950) relates the critical Reynolds number to
the solidity, where the critical Reynolds number is defined as the Reynolds number at
which eddies are shed from the wires. By choosing N and d we can arrive at a screen
design iteratively. Based on Watmuff’s (1998) design, we have selected four screens in
the current design with open area ratios of 67%, 62%, 62%, and 60% and number of
mesh per inch of 24, 32, 46, 56, respectively. There are two additional screens (67% OA,
# mesh per inch=24) before and after the honeycomb that hold it in place. The spacing
between the screens should be of the order of the large energy containing eddies (Mehta
and Bradshaw 1979). The screens have to be at least 500 wire diameters apart (Mehta,
1977). In the current design we have chosen a screen separation distance of 95 mm
which corresponds to 500 wire diameters. The distance between the last screen and the
contraction is 1.07 m . The NASA Langley 2 foot by 3 foot low speed boundary layer
tunnel (King 2000), has a somewhat similar arrangement of screens as our facility. There
are two 24 by 24 mesh per inch screens and two 40 by 40 mesh per inch screens,
arranged in the settling chamber that gives a turbulence intensity (defined as the
turbulence fluctuation velocity in the x direction to the mean velocity in the x direction)
of less than 0.1 % for a speed of 45 /m s .
42
1"
1"
d
N mesh/in
N mesh/in
Figure 3-4. Schematic of the screen design.
Contraction
The contraction accelerates and aligns the flow into the test section. The design of
the tunnel inlet contraction is important for maintaining good flow quality in the test
section. The size and shape of the contraction also dictates the final turbulence intensity
levels in the test section (Derbunovich et al. 1987). The length of the contraction should
be small to minimize the boundary layer growth. The flow leaving the contraction should
be uniform and steady. The separation of flow, due to streamline curvature in the
contraction, has to be avoided at any cost. For a finite-length inlet contraction, an
example wall pressure distribution is shown in Figure 3-5. The is a maxima and a
minima for the wall static pressure distribution along the wall at two locations near the
entrance and exit respectively, resulting in regions of adverse pressure gradients.
Separation can occur in the regions of adverse pressure gradient, so the contraction must
be designed to minimize the possibility of separation.
43
For the design of the contraction, a method of matched cubic polynomials was
used, as suggested by Morel (1975). The schematic of the contraction shape polynomial
is shown in Figure 3-6. The entrance height of the contraction is iH and the exit height
is eH . The total length of the contraction is L and the two cubic polynomials are
matched at a specified location mx x= . A series of contraction ratios, test section sizes,
and match points were selected. The 3-D potential flow equation for the flow was solved
to obtain the velocity field at the exit of the contraction for the various test cases.
0 1 2 3 4 5 6 7 8-180
-160
-140
-120
-100
-80
-60
-40
-20
0
x (ft)
Cp
Regions ofpossible
separation
Figure 3-5. pC distribution along corner for a contraction.
44
He/2
Match Point
xmx
Hi/2
L
y
Symmetry Line
Figure 3-6. Schematic of the contraction shape polynomial.
To check the quality of a contraction design, two tests were performed in these
simulations. First, the flow non-uniformity at the exit plane was calculated from the
computed velocity field. Second, Stratford’s criterion (Stratford 1959) for separation of a
two-dimensional boundary layer was applied. While it is unknown if Stratford’s
Criterion is valid in a three-dimensional flow, it has been used in the past (Morel 1975),
and it serves a good qualitative analysis tool. It was found that having the match point at
the mid point of the contraction favorably affects the pressure maxima and minima of the
pressure distribution. Larger contraction ratios tend to increase pressure gradients within
the contraction, but have a favorable effect on flow uniformity, turbulence intensity
reduction, and system losses.
After this initial set of calculations was run, alternative contraction designs were
studied. The primary reference for contraction designs was found to be Su (1991). Su
recommended matching 3rd order polynomials at the contraction entrance with higher
order polynomials at the contraction exit. This approach has little effect on the region of
45
adverse pressure gradient, but it shifts the second adverse pressure gradient region (see
Figure 3-5) closer to the match point and intensifies the pressure gradient. While this can
lead to separation, in most contraction contours the chance of separation is much higher
at the first section of adverse pressure gradient, since the magnitude of second region is
smaller than the first. Su also mentioned crossflow as a major design concern, due to 3-D
flow in the contraction, and suggested varying the contraction aspect ratio, as this has a
favorable effect on the design variables.
The next stage of contraction design involved the derivation of a 3rd-8th matched
contour, and the testing of its performance. The details of the derivation are given in
Appendix B. The contraction was found to have similar separation characteristics of a
matched cubic with otherwise identical design parameters, but vastly improved the flow
uniformity values. It was decided to use a 3rd-8th matched contour, where position, slope,
and curvature are matched at the match point. Once the contraction shape was
established an optimization study was conducted to improve the contraction flow quality
by optimizing the contraction design parameters. The four contraction design parameters
used for this study are the total length L , the contraction ratio CR , the aspect ratio at the
entrance AR (i.e., entrance height to width ratio) , and the nondimensional match point
of the wall shape polynomials, X . The aim of the optimization study is to arrive at a
combination of the design parameters that gives the best flow uniformity and minimal
flow angularity. The details of the optimization study are given in Appendix C. The
resulting contraction was 3.05 m (10 ' ) long with a square entrance measuring 2.57 m by
2.57 m (8.42 ' by 8.42 ' ) and a contraction ratio of 8. The contraction shaped
46
polynomials were matched at the midpoint and the exit plane of the contraction measured
0.74 m by 1.12 m ( 29" by 44" ).
Figure 3-7. Contours of x velocity along the half mid-plane for the contraction.
Once the design is established it has to be validated. Conventional method
(Mueller, 1992) solves the 3-D potential flow equation inside the contraction. The
pressure distributions along the corner of the contraction walls are extracted and
Stratford’s criterion is used to check for flow separation. We have gone a step further to
validate our design by solving the steady, turbulent, 3-D, Navier Stokes equation inside
the contraction. An unstructured grid with over 350,000 hexahedral elements was used to
mesh a quarter section of the contraction. A uniform inlet velocity of 9.5 /m s was
47
specified at the contraction entrance and pressure boundary condition was applied at the
exit. The solid walls used no slip while the two other walls used symmetry boundary
condition. A numerical simulation in Fluent® was done where a k-omega turbulence
model was used to analyze the flow through the contraction. A plot of the x velocity
variation along the axis, for the half mid-plane (xy plane) is shown in Figure 3-7. The
results predicted no separation.
Test Section
The test section measure 0.74 m ( 29" ) in height by 1.12 m ( 44" ) in width by
1.83 m ( 6 ' ) in length. The test section is of the open jet type. The maximum velocity of
the flow in the test section is set at 76 /m s . The formation of boundary layer on the
leading edge of a model placed in the test section restricts the chord length not to exceed
two-thirds the test section width. The maximum Reynolds number calculated based on
the chord is . 63.6710 . It is also essential to have low values for the blockage ratio ( B ),
which is the fraction of the test section frontal area of the model, and is given by the
formula
( )sin
TS
cB
Hα
= (3.4)
where c is the chord length, α is the angle of attack (AOA) and TSH is the height of the
test section. Table 3-2 gives the details of the test section design.
Table 3-2. Test section details. Test section (H x W) 0.74 m x 1.12 m
Velocity 76 /m s Length 1.83 m
Re (based on c = W/2) . 62.7510 Re (based on c = 2W/3) . 63.6710 B (AOA = 20º, c=W/2) 0.26 B (AOA = 15º, c=W/2) 0.19
48
Diffuser
The purpose of the diffuser is to recover the static pressure drop that occurs from
the entrance of the contraction to the exit of the test section. The area of the diffuser
should increase gradually along its axis, so as to prevent the flow from separating. The
diffuser should not be too long either, as this would add to the total cost and also affect
the room dimension constraints. A longer diffuser also leads to a larger static pressure
loss (due to viscous effect) that the fan has to overcome.
The diffuser starts with a collector section at its entrance. The schematic of the
collector is shown in Figure 3-8. The collector is semi cylindrical in shape such that that
relatively smooth flow enters diffuser 1. Our current design is a based on the collector
design from a similar anechoic wind tunnel at Notre Dame (Mueller et al. 1992). The
collector is filled with acoustic insulation for the purpose of noise attenuation. The
collector feeds the flow from the test section into the diffuser. The area of the collector
entrance is made 1.174 times larger than the test section area to minimize the curvature of
the streamlines of the flow entering the collector.
Pope and Harper (1966) states that for a conical diffuser, the divergence half angle
of the diffuser walls should be less than 3.5° for a “good” design. Mehta (1977) states
that the diffuser included angle for a conical diffuser should be between 5° (for best flow
steadiness) and 10° (for best pressure recovery). The anechoic chamber dimensional
constraint prevents us from having a conical or 3D diffuser (see Appendix D). The only
other option is to have a 2D flat diffuser, with a fixed width and variation only in the
height. A 2D diffuser also allows expansion to a larger angle for a given area ratio than a
conical diffuser. The schematic of the diffuser design is shown in Figure 3-9. The
49
diffuser angle θ , is the included angle between the diverging walls. The total length is
L and the width of the diffuser is W and it is same as the test section width. The inlet
height is iH and the exit height is eH .
Diffuser 1Cross Section
Collector 0.3 m
1.2 m
0.8
m
R 0.15 m
Figure 3-8. Schematic of the collector.
/ 2θ
L W
eHiH
SIDE VIEW FRONT VIEW
Figure 3-9. Schematic of the 2D diffuser.
50
There are two diffusers in this design to achieve the required pressure recovery, as
shown in Figure 3-2. The design of the diffusers was made based on Kline’s flat diffuser
curves (Runstadler et al. 1975). The schematic of the diffuser design curves is shoen in
Figure 3-10. Plotted on the y axis is the area ratio between the exit and entrance of the
diffuser, /e iAR H H= and x axis represents the ratio of length to the entrance height of
the diffuser, / iL H . Shown on the plot is an envelope of the separation and no separation
region. The entrance height of diffuser 1 is fixed by the size of the test section. The
design of the diffuser is done by selecting a length for the diffuser that is within the
chamber dimension constraints. Once / iL H is known the corresponding value of AR is
selected from the region of no stall. Although a greater pressure recovery can be
achieved by operating outside the region of no appreciable stall, but still below the line of
appreciable stall, this would not guarantee the steadiness of the flow that can contribute
to unwanted noise. The idea is to achieve maximum expansion of the flow within the
shortest possible length of the diffuser, at the same time ensuring that the flow is steady
and devoid of separation. An optimization study was conducted to minimize the diffuser
divergence angles subject to constraints imposed by the chamber dimensions. Appendix
D gives the details of the optimization study. Both diffuser 1 and diffuser 2 were
designed based on the optimization scheme. Diffuser 1 immediately follows the test
section and has entry dimensions of 0.8 m by 1.2 m and a total length of 3.58 m . The
included angle for diffuser 1 is 10.94º. The exit dimensions of diffuser 1 are 1.49 m by
1.2 m which is also the inlet dimensions for diffuser 2. Diffuser 2 is 5 m long and the
included angle is 11º. The exit dimensions of diffuser 2 measure 2.2 m by 1.2 m .
51
0 5 10 15 20 25 30 35 401.5
2
2.5
3
3.5
4
L/Hi
AR
Stall Unsteadyflow
No appreciablestall
Line ofappreciable
stall
Figure 3-10. 2-D Diffuser Design Curves.
The objective of the diffuser design is to recover the dynamic pressure of the high
velocity fluid in the test section. The flow field within the diffuser is highly influenced
by the nature of the flow leaving the test section. The orientation and size (blockage) of
the airfoil models, and entrainment of air (assuming leakage) are some of the factors that
affect the diffuser incoming flow. In our design we have neglected all these effects for
the sake of simplicity. The flow entering the diffuser is assumed to be steady, uniform
and hence free of vorticity. The axial pressure gradient inside the diffuser can be
calculated by combining the continuity and momentum equations while assuming
inviscid, incompressible, steady, Newtonian flow with constant properties. Based on
these assumptions, the continuity equation takes the form
,diffuserA U const= (3.5)
52
where ( )U x is the axial velocity in the diffuser at any location x . Taking derivatives
on both sides of Eq. (3.5) with respect to flow direction x , we obtain
0.diffuserdiffuser
dAdUA Udx dx
+ = (3.6)
The inviscid, momentum equation is the steady Euler’s equation, which can be written in
the form
,dU dPUdx dx
ρ −= (3.7)
where ρ is the fluid density and ( )P x is the static pressure at any location x along the
axis of the diffuser. Combining Eq’s. (3.7) and (3.6) we obtain an expression for the
axial pressure gradient
2
3 ,diffuser
diffuser
dAdP mdx A dxρ
= (3.8)
where the mass flow rate m is given by
.TS TSm U Aρ= (3.9) where TSU and TSA are the velocity and area of the test section, respectively.
The total pressure recovery over the length of the two diffusers is calculated by
integrating Eq. (3.8).
2
30 0
.L L
diffuserdiffuser
diffuser
dAdP mP dx dxdx A dxρ
⎛ ⎞⎛ ⎞Δ = = ⎜ ⎟⎜ ⎟ ⎜ ⎟⎝ ⎠ ⎝ ⎠∫ ∫ (3.10)
Integrating Equation (3.10), we obtain
( ) ( )
2
2 21 1
2 0diffuser
diffuser diffuser
mPA A Lρ
⎧ ⎫⎪ ⎪Δ = −⎨ ⎬⎪ ⎪⎩ ⎭
(3.11)
The pressure recovery in a diffuser thus depends only on the mass flow rate,
density and the area at the entrance and exit of the diffuser, and it is independent of the
diffuser shape. The total recovered pressure is the sum of the recovered pressures from
53
the two diffusers and has a value of 3228 Pa . The static pressure drop in the test section
that needs to be overcome is given by
21 ,2TS TSP UρΔ = (3.12)
which has a value of 3556.4 Pa , assuming a density of 31.225 /kg mρ = and a test
section velocity of 76 /TSU m s= . Therefore the net un-recovered pressure of 330 Pa
(1.33" H2O) has to be provided by the fan. We have used Stratford’s separation criterion
to predict whether a turbulent boundary that originates at the diffuser entrance would
separate anywhere inside the diffuser. Assuming 1D, inviscid, steady flow inside the
diffuser, the pressure co-efficient was calculated by the formula
2
2
( ) ( )( ) 1 ,1/ 2
TSp
TS TS
P x P U xc xU Uρ
⎛ ⎞−= = − ⎜ ⎟
⎝ ⎠ (3.13)
where ( )U x is the velocity at any axial location x along the diffuser, ( )P x is the
corresponding value of static pressure, and TSP is the static pressure in the test section.
This value was compared with the pressure co-efficient value given by Stratford (1959).
The mathematical form of the Stratford’s criteria is shown in the equation below.
( ) ( )0.10.5 6/ 0.39 10p pc xdc dx R−= (3.14)
where pc is the pressure coefficient and R is the Reynolds number.
The results are shown in Figure 3-11and Figure 3-12. As seen from the figure the local
value of the pressure coefficient is always below the Stratford’s pressure coefficient.
Therefore, separation never occurs in our diffuser.
54
0 0.5 1 1.5 2 2.5 3 3.5 40
0.1
0.2
0.3
0.4
0.5
0 0.5 1 1.5 2 2.5 3 3.5 40
1
2
3
4
5
6
x (m)
Cp*√(xdCp/dx)0.35*(10-6Rex)0.1
CpCpstratford
Figure 3-11. Comparison of local pressure coefficient with Stratford’s separation
pressure coefficient for diffuser 1.
0 0.5 1 1.5 2 2.5 3 3.5 40
0.1
0.2
0.3
0.4
0.5
0 0.5 1 1.5 2 2.5 3 3.5 40
2
4
6
8
x (m)
Cp*√(xdCp/dx)0.35*(10-6Rex)0.1
CpCpstratford
Figure 3-12. Comparison of local pressure coefficient with Stratford’s separation
pressure coefficient for diffuser 2.
55
A numerical simulation of the flow through diffuser 1 was conducted using
Fluent®. The computational domain consists of the whole diffuser 1. A uniform inlet
velocity was specified at the entrance of the domain, and outflow boundary condition was
applied at the domain exit. The walls of the diffuser had no slip boundary condition. The
steady state, 3-D Navier Stokes equation was solved using the k-omega turbulence model
for a maximum test section velocity of 76 /TSU m s= . An unstructured mesh was used
for the domain meshing, and there were a total of 661,770 mixed elements in the
computational domain. The x velocity profile along the diffuser 1 centre plane is shown
in Figure 3-13. The result indicates that flow does not separate inside diffuser 1. A
similar analysis was done for diffuser 2 and it was found that diffuser 2 was also devoid
of any regions of separation.
X (meters)
Y(m
eter
s)
0 1 2 3 4
0
1
2
3
X Velocity (m/s)
65605550454035302520151050
Figure 3-13. Centre plane x velocity profile along diffuser 1.
56
Corner/Turning Vanes
The corner section connects the two diffusers in the wind tunnel circuit. Without
turning vanes, flow separation occurs, resulting in large losses and flow noise. Installing
an array of turning vanes can minimize this. Most available literature on turning vanes
are over 50 years old (Collar 1937; Salter 1946) and the data presented lack sufficient
analytical and experimental details. The turning vanes for the current design are taken
from Gelder et al. (1986). The dimensions of the turning vanes in our wind tunnel are
scaled based on the hydraulic diameter of the duct enclosing the vanes. The schematic of
the vanes is shown in Figure 3-14. The vanes of this type have a very low loss
coefficient (skin friction loss= 4 % of entrance dynamic pressure at M=0.35). The vane
chord length c has a value of 15 cm and the maximum vane chord thickness ratio is
0.196. The vanes will be fabricated from fiberglass, having an outer shell of 4-5 layer
thickness, and the inside will be filled with compliant rubber for the purpose of vibration
damping. The coordinates of one of the vanes, starting from the trailing edge and moving
in a clockwise direction around the vane and back to the trailing edge, is listed in
57
Table 3-3. The inner and outer wall of the corner section is made to match with the
bottom and top surface of the vanes respectively. A numerical simulation of the flow
over the vanes was conducted using Fluent®. The computational domain consists of the
region in between the top and bottom surfaces of adjacent vanes. The domain was
extended forward and aft of the vane edges in a straight section. A uniform inlet velocity
was specified at the entrance of the domain, and pressure boundary condition was applied
at the domain exit. The top and bottom surfaces of the vanes have no slip and periodic
boundary condition was applied on the walls of the straight section. The steady state, 2-
D Navier Stokes equation was solved using the k-omega turbulence model for two
different Reynolds numbers. An unstructured mesh was used for the domain meshing,
and there were a total of 97,544 mixed elements in the computational domain. The
maximum Reynolds number based on the vane chord is . 53.510 ( 76 /TSU m s= ) and the
minimum Reynolds number considered is . 50.810 ( 18 /TSU m s= ). Results from the
turning vane simulation for the maximum and minimum Reynolds number are shown in
Figure 3-15 and Figure 3- 16. Shown are the contours of x component of the velocity in
the flow domain. A careful examination of the velocity profiles in the domain indicated a
small separation region near the trailing edge, on the suction surface of the vane for both
the flow cases.
58
Figure 3-14. Schematic of the Turning Vanes.
Periodic
Inlet
PressureoutletPeriodic
Periodic
Periodic
No slip
No slip
Figure 3-15. Results from turning vane simulation for a test section speed of 76 /m s .
59
1.22e+011.15e+011.08e+011.01e+019.43e+008.75e+008.07e+007.39e+006.71e+006.03e+005.34e+004.66e+003.98e+003.30e+002.62e+001.94e+001.26e+005.74e-01-1.08e-01-7.89e-01-1.47e+00
Figure 3- 16. Results from turning vane simulation for a test section speed of 18 /m s .
60
Table 3-3. Turning vane coordinates. X (m) Y (m) X (m) Y (m) X (m) Y (m)
0.987 -0.512 0.107 -0.226 0.477 -0.026 0.987 -0.512 0.088 -0.232 0.491 -0.032 0.985 -0.509 0.071 -0.239 0.504 -0.038 0.981 -0.505 0.056 -0.244 0.518 -0.044 0.975 -0.499 0.042 -0.248 0.531 -0.052 0.967 -0.491 0.030 -0.251 0.545 -0.059 0.959 -0.482 0.020 -0.253 0.558 -0.067 0.949 -0.472 0.012 -0.251 0.571 -0.076 0.938 -0.461 0.006 -0.247 0.585 -0.085 0.926 -0.449 0.002 -0.241 0.598 -0.094 0.914 -0.436 0.000 -0.231 0.612 -0.104 0.900 -0.423 0.000 -0.220 0.625 -0.114 0.886 -0.410 0.003 -0.206 0.638 -0.125 0.871 -0.396 0.009 -0.192 0.652 -0.136 0.855 -0.382 0.017 -0.176 0.666 -0.148 0.839 -0.368 0.028 -0.160 0.679 -0.161 0.821 -0.354 0.040 -0.143 0.694 -0.174 0.803 -0.340 0.054 -0.127 0.708 -0.187 0.785 -0.326 0.068 -0.112 0.722 -0.202 0.765 -0.312 0.084 -0.098 0.737 -0.217 0.745 -0.299 0.099 -0.085 0.755 -0.232 0.724 -0.286 0.115 -0.074 0.769 -0.249 0.702 -0.273 0.131 -0.063 0.786 -0.267 0.680 -0.261 0.146 -0.053 0.803 -0.285 0.657 -0.249 0.161 -0.045 0.821 -0.305 0.633 -0.239 0.177 -0.037 0.840 -0.325 0.609 -0.228 0.192 -0.030 0.859 -0.346 0.584 -0.219 0.207 -0.024 0.879 -0.368 0.559 -0.210 0.222 -0.019 0.899 -0.390 0.533 -0.203 0.237 -0.014 0.919 -0.411 0.506 -0.196 0.252 -0.011 0.938 -0.432 0.480 -0.190 0.266 -0.008 0.955 -0.450 0.453 -0.186 0.281 -0.005 0.970 -0.466 0.425 -0.182 0.295 -0.003 0.982 -0.479 0.398 -0.180 0.310 -0.001 0.991 -0.488 0.371 -0.179 0.324 0.000 0.997 -0.494 0.344 -0.179 0.338 0.000 1.000 -0.497 0.317 -0.180 0.352 0.000 1.000 -0.497 0.290 -0.182 0.367 -0.001 0.264 -0.185 0.381 -0.002 0.239 -0.189 0.395 -0.004 0.215 -0.194 0.409 -0.006 0.191 -0.200 0.422 -0.009 0.168 -0.206 0.436 -0.013 0.147 -0.213 0.450 -0.017 0.126 -0.219 0.464 -0.021
Vibration Isolator
Mechanical vibrations from the drive fan must be isolated from the main body of
the tunnel. The fan is attached firmly to a large concrete slab, which rests on a sand bed.
The slab of concrete is sized based on the rule of thumb that its mass should be at least 10
61
times the rotational mass of the fan. The slab measures 6 4.4 .46 m m m× ×
( 20 14.3 1.5 ft ft ft× × ) and has a mass of 42.13 10 kg× ( 47,000 lbs ). The fan and its
base are surrounded by a retainer wall. A part of the vibrations caused by the fan is
transmitted through the slab to the sand layer and it is isolated from the building.
Transition
A schematic of the rectangular to round transition section is shown in Figure 3-17.
The rectangular-to-round transition joins the diffuser 2 exit to the circular entrance of the
fan which has a diameter of 1.95 m ( 76.75"). This is a slowly diverging section with an
area ratio of 1.11 between the exit and entry cross sections. The length of the transition is
selected to be same as the hydraulic diameter at the exit of diffuser 2 and it has a value of
1.57 m ( 5.14 ft ). A numerical simulation of the flow through the transition piece was
conducted using Fluent®. The computational domain consists of a quadrant of the
transition piece. A uniform inlet velocity was specified at the entrance of the domain,
and pressure boundary condition was applied at the domain exit. The walls of the
transition had no slip and symmetry boundary conditions. The steady state, 3-D Navier
Stokes equation was solved using the k-omega turbulence model for a maximum test
section velocity of 76 /TSU m s= . An unstructured mesh was used for the domain
meshing, and there were a total of 68,9905 mixed elements in the computational domain.
The x velocity profile along the transition centre plane is shown in Figure 3-18. The
result indicates that flow does not separate inside the transition.
62
He
W
D/2
Side View FrontView
Top View
D
W
Figure 3-17. Schematic of the rectangular to round transition section ( 2.22 eH m= , 1.2 W m= , 1.95 D m= ).
2.75e+012.61e+012.47e+012.34e+012.20e+012.06e+011.92e+011.79e+011.65e+011.51e+011.37e+011.24e+011.10e+019.62e+008.24e+006.87e+005.49e+004.12e+002.75e+001.37e+000.00e+00
ZY
X
Figure 3-18. Results from transition flow simulation for a test section speed of 76 /m s
63
Fan
The fan drives the wind tunnel by compensating the losses that occurs in the
various tunnel components. Fans are rated by the volume flow rate they can provide, as
well as the static pressure drop they can overcome. The procedure for estimation of the
losses is described in Appendix E. The results from the fan loss calculation are given in
Table 3-4. From the table, K is the pressure loss coefficient and PΔ is the pressure loss
across the individual wind tunnel component. They are related by the expression
PKq
Δ= (3.15)
where q is the dynamic pressure in the test section.
Table 3-4. Results of the wind tunnel circuit loss calculation. Component K PΔ (in H20) % Total Loss
Settling Duct
0.00017 0.0024 0.03
Honeycomb 0.0029 0.04 0.6 Screen 0.094 1.34 19.2 Inlet 0.022 0.32 4.6
Test Section 0.16 2.35 33.6 Diffuser 1 0.069 0.98 14
Corner 0.03 0.43 6.2 Diffuser 2 0.008 0.11 1.6 Transition 0.006 0.08 1.14
Unrecovered 0.093 1.33 19 Total 100
It can be observed from the table that the maximum percent loss occurs in the test
section since it is of the open jet type. Other notable losses occur across the inlet screens
and as unrecovered pressure across the diffusers. The total loss factor is the sum of the
individual loss coefficients, and has a value of 0.49 that translates to a value of 27" H O
gage pressure that the fan has to overcome.
64
A Twin City, single-width acoustafoil fan that can deliver a flow rate of
147000 cfm against a static pressure recovery of 28" H O is selected to be used to drive
the tunnel. Figure 3-19 shows the load curves for the selected fan versus the wind tunnel
loss curve. The fan can be operated several rpm settings using a variable frequency
drive, four of which are shown in the figure. For each rpm setting of the fan there is a
fixed static pressure loss curve and an efficiency curve. Ideally, we would want to
operate the fan in or around the maximum efficiency region. The wind tunnel
performance curve is obtained from the fan pressure loss calculations by estimating the
static pressure loss for various values of volume flow rates that include a 10 % leakage of
external air into the chamber. The points where the pressure load curves intersect the
tunnel performance curve determine the operating points of the wind tunnel. The fan rpm
loss and the efficiency curves are provided by the fan manufacturer. It can be seen from
the figure that at the four tunnel operating points shown, the efficiencies are over 50 %,
which is very reasonable. Figure 3-20 shows the pressure drop along the tunnel circuit
for the maximum value of test section velocity of 76 /m s . It can be observed that the
maximum static pressure drop occurs in the test section and the maximum pressure
recovery occurs inside diffuser 1. The unrecovered pressure is accounted for by the fan.
65
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
x 105
0
10
20
Stat
ic P
ress
ure
Loss
( in
H20)
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
x 105
0
50
100
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
x 105
0
50
100
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
x 105
0
50
100
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
x 105
0
50
100
220 rpm
Flow Rate (CFM)
220 rpm
439 rpm
439 rpm
658 rpm
658 rpm
884 rpm
884 rpm
Tunnel Performance Curve
Effic
ienc
y (%
)
Figure 3-19. Fan Load curve.
0 2 4 6 8 10 12 14 16 180
2
4
6
8
10
12
14
16
18
20
Axial distance along the Wind Tunnel circuit (m)
Pres
sure
dro
p (in
H2o)
W /T entrance Honeycomb
Screen Contraction
Fan entrance
Transition
Diffuser 2
Corner
Diffuser 1
Test section
Figure 3-20. Estimated pressure drop along the wind tunnel circuit for a test section
velocity of 76 /m s .
66
Acoustic Treatment
Acoustic treatment of the wind tunnel components is essential to minimize
disturbance noise from contaminating the acoustic measurements performed in the
chamber. The background noise levels in the test section have to be low enough to
ensure the quality of the acoustical measurements. The wind tunnel will be primarily
used for airframe noise studies, specifically trailing edge noise. Ffowcs, Williams, and
Hall (1970) gives the mean-square far field pressure field generated by airfoil trailing
edge in the absence of convection effects and vortex shedding as
( )2 2 2 2 2 30 0 2 sin sin / 2 cosLp u V M
Rυδρ α θ β⎛ ⎞≈ ⎜ ⎟
⎝ ⎠ (3.16)
where 0ρ is the ambient density, 0u is the rms turbulence velocity, V is a typical
velocity over the trailing edge, Mυ is the turbulence convection mach number, L is the
span-wise distance, δ is the turbulence length scale, R is the observer distance, β is the
sweep angle, and θ and α are the observer angles. It can be observed that the trailing
edge noise has a 5V dependence. Equation (3.16) can used to scale the typical trailing
edge noise spectra and predict what the anticipated levels of trailing edge noise will be in
our wind tunnel. The background noise in the chamber has to be well below this level.
In order to estimate how low the background noise has to be, consider a
microphone located in the chamber that measures the far field spectra from a trailing edge
configuration. The microphone signal micp , has contributions from both trailing edge
noise TEp , and the background noise, bp , and can be expressed in the mathematical form
as
2 2
mic TE bp p p= + (3.17)
67
We want to estimate how low bp has to be below TEp so that the contribution of bp can
be neglected from micp . Converting the pressures to sound pressure levels using a
reference pressure refp , we obtain
1020 log TETE
ref
pSPLp
⎛ ⎞= ⎜ ⎟⎜ ⎟
⎝ ⎠ (3.18)
1020 log bb
ref
pSPLp
⎛ ⎞= ⎜ ⎟⎜ ⎟
⎝ ⎠ (3.19)
The sound pressure level measured by the microphone can be expressed in terms of
TESPL and bSPL as
( )/10/101010log 10 10 bTE SPLSPL
micSPL = + (3.20) Typical trailing edge noise spectra are in the 80 dB range. If we assume that the
background noise levels are 70 dB , micSPL has a value of 80.4 dB . If the background
noise is of the order of 65 dB , the microphone measures 80.1 dB . Thus the background
noise levels have to be at least 10 15 dB− below the main noise source levels.
The background noise in the test section is both broadband and tonal in nature
(Duell et al. 2002). The main sources of broadband noise are the broadband fan noise,
flow over turning vanes, screens, boundary layer noise, turbulent shear layer noise etc.
Tonal noises mainly emanate from the blade passage of the fan, flow over perforated
ducts, vortex shedding off measurement probes etc. Duell et al. (2004) gives a range of
frequencies where each component of the background noise is important. An
understanding of the sources of background noise and its frequency range of influence is
essential in developing acoustic treatment for the wind tunnel components.
One of the main components of background noise comes from the fan. The blade
passage frequency (BPF) and its harmonics appear as discrete tones in the test section
68
background noise spectra. The blade passage frequency, in Hz, of the fan is calculated
based on the formula
60
BPF NΩ= (3.21)
where Ω is the fan speed in rpm and N is the number of fan blades. The maximum
speed of the fan is 884 rpm and it has 10 blades. The wall of the ductwork leading to the
fan has to be lined with an acoustical absorber to minimize upstream propagating fan
noise. Schultz (1986) showed that a 15 cm ( 6" ) layer of fiberglass can attenuate 99% of
the incident sound energy up to frequencies as low as 125 Hz . A thicker liner can extend
the attenuation to lower frequencies. The room dimension constraints prevent us from
having a liner thicker than 30 cm . Therefore this is the value of the thickness of the
fiberglass absorber we have used to line the walls of the ductwork, that includes diffuser
1, corner, diffuser 2, and the transition section. The outer walls of the ducts are made of
13 mm thick outer steel shell that carries the pressure loads. The inner wall of the
ductwork is made of a wire cloth mesh. The bulk fiberglass absorber is sandwiched
between the inner and outer walls of the duct. There is a thin layer of woven fiberglass
sheet located between the inner wall and the fiberglass absorber that prevent fiberglass
fibers from contaminating the flow through the ducts. The schematic of the duct wall
cross section is shown in Figure 3-21. The impedance of a 30 cm long sample of the
liner consisting of the fiberglass, woven cloth, and the screen will be measured in an
impedance tube.
69
DiffuserCross section
Inner wall (Steel wire cloth, 20 Mmesh)
Acoustic absorber (Fiber glass)
Outer wall (Plywood, 13 mm thick)
30 cm
Cover sheet (Woven fiber glass)
DiffuserCross-section
Figure 3-21. Details of the wind tunnel duct walls.
70
CHAPTER 4 FABRICATION OF THE WIND TUNNEL COMPONENTS
An anechoic chamber built by Eckel industries in 2001 was modified to incorporate
an anechoic wind tunnel. This chapter discusses the fabrication of the various wind
tunnel components. Except for the flow conditioner section and the fan, all other
components of the tunnel were built at the University of Florida. A brief discussion
about the acoustic treatment of the wind tunnel is also provided.
Inlet
A photograph of the inlet contraction is shown in Figure 4-1. The inlet is 3.04 m
long and it measures 2.57 m by 2.57 m at the entrance and 0.74 m by 1.12 m at the
exit. The contraction was fabricated using fiberglass reinforced wood with 0.127 m
( 0.5") core construction, similar to technology used in boat construction. The core is
embedded in isothalic milled fiber putty, and covered with two layers of 0.0425 kg
(1.5 oz ) mat. The walls are laminated with 0.76 mm ( 0.03") isothalic gel coat and
covered with 20.458 /kg m ( 21.5 /oz ft ) chopped strand mat alternated with 0.68 kg
( 24 oz ), 90 degree biaxial roving totaling approximately 4.45 mm ( 0.175"). The
outside of the structure is reinforced with longitudinal conformal strakes of 0.019 m
( 0.075") luan ply and fiberglass on all four sides.
71
Figure 4-1. Photograph of the inlet contraction.
Diffuser
A photograph of diffuser 1 is shown in Figure 4-2. The fabrication procedure for
both diffuser 1 and 2 is the same. An internal skeletal view of diffuser 1 is shown in
Figure 4-3. An internal frame made of aluminum is initially built to support the diffuser
structure. The inner surface of the frame is then covered with McNichols Stainless Steel
wire mesh (20 M mesh). A thin layer of woven fiberglass cloth is then spread over the
wire mesh. Both fiberglass cloth and the wire mesh are attached to the internal frame
using aluminum rivets. A 0.3 m (12") thick bulk fiberglass absorber (Owens Corning
PINK fiberglass insulation, Type R-19) is then spread over the surface. The woven
fiberglass cloth prevent fibers from contaminating the flow through the ducts. An outer
wooden cover sheet made of plywood (Norbord OSB sheet, 13 mm ) is built to enclose
the diffuser. The collector section that precedes diffuser 1 is made of a framework of
0.3 m diameter semicircular wooden ribs attached to the diffuser 1 structure. Bulk
fiberglass is used to fill the gaps between the ribs and a layer of 0.013 m thick, Tex Tech
industries Nomex sound absorber sheet is stretched across the collector section.
72
Collector
Plywood ‘I’ beam
Figure 4-2. Photograph of diffuser 1.
Wire mesh
Diffusercross
support rod
Semicircularribs
Figure 4-3. Diffuser 1 internal skeletal view.
Plywood ‘I’ beams 0.2 m high are attached to the surfaces of both diffusers for
structural support. A view of the ‘I’ beam cross section is shown in Figure 4-4.
73
Figure 4-5 shows the additional structural reinforcements applied to diffuser 2. A
PolyGard Hi-Bond Polyurethane foam is used to fill the gap between the diffuser outer
wall and the chamber wall as well as the diffuser outer wall and the wall of the building
that houses the anechoic chamber. Plywood (2 by 4) reinforcement rods further stiffen
the diffuser structure. Semi-cylindrical fiberglass sheets (Figure 4-6) are also attached to
the surface of diffuser 2 for structural rigidity.
v
21.6 cm
8.9 cm
3.8 cm
Figure 4-4. Cross sectional view of the ‘I’ beam.
Figure 4-5. Structural reinforcement using polyurethane foam.
74
Diffuser 2
Structuralreinforcement
Figure 4-6. Structural reinforcement using semi cylindrical hollow fiberglass sheets.
Corner/Turning Vanes
The fabrication of the corner section is similar to the diffuser, except that the inner
surface of the corner is made of perforate metal sheet, instead of wire mesh. The
perforate metal is made of steel and it is 0.76 mm thick. The corner section houses the
vane assembly that is shown in Figure 4-7.
The vane chord length c has a value of 17.1 cm ( 6.75" ), and the maximum vane
thickness ratio is 0.196 . The individual vanes are 0.48 m (19" ) in span. The vane
assembly consists of three rows of 20 vanes attached to 2 cross plates to provide
necessary structural rigidity. The side view of the cross plate is shown in Figure 4-8.
The cross plate is 13 mm thick and it consists on elliptical leading edge and a triangular
shaped trailing edge, both made of aluminum and attached to a wooden mid plate.
75
Cross plate
17.1 cm
0.48 m
Figure 4-7. Photograph of the turning vane rack
2.5 cm 20.3 cm 7.6 cm
Mid plateLeadingedge
Trailingedge
Figure 4-8. Side view of the cross plate.
The mold used for the turning vane fabrication is shown in Figure 4-9. The vane
outer structure was made of four layers of woven fiberglass ( 2 mm thick) mixed with
PTM&W Aeropoxy type PH3660 harder and type PR2032 resin. Each layer of fiberglass
was soaked in the epoxy and spread over the mold. The mold was released after a day
and the vanes were cut to the necessary size. A 13 mm long triangular , having a base
76
width of 3.8 mm and made of US Composites™ Easy Flo-60 low viscosity polyurethane
casting resin was used as an additional trailing edge attachment A trailing edge mold
made of U.S. Composites™ type 74-30 RTV silicone mold rubber was filled with the
casting resin and the trailing edge of the vane was inserted into the resin. The mold was
released once the resin was set. Later the vane was filled with U.S. Composites™ type
74-30 RTV liquid urethane mold rubber, to attenuate any vibrations of the vane structure
by increasing the structural damping coefficient.
Figure 4-9. Photograph of vane mold.
Vibration Isolator
The vibration isolator forms a sandwich between diffuser 2 and the transition piece,
as shown in Figure 4-10. The vibration isolator consists of an aluminum structure
attached to a 4 cm thick flex piece that minimizes the propagation of vibration induced
by the fan. The vibration isolation section is rated for 28" H O . ‘L’ brackets are attached
to the side of the vibration isolator for structural rigidity. There is an access door
( 0.61 m by 0.61 m ) in the bottom of the vibration isolator that enables entry into the duct
work.
77
Flex piece
Diffuser 2
Transition
‘L’ brackets
VibrationIsolator
Flow
Figure 4-10. Photograph of the vibration isolator section.
Transition
The transition section connects the vibration isolator to the fan. A photograph of
the transition piece is shown in Figure 4-11. A flange was constructed to match the fan
inlet with a 5 cm bore-axis flange extending toward the flow direction. The flange was
placed face down on the floor. Four isosceles triangles were constructed with the base
dimensions respectively matching the sides and top/bottom of the large end of diffuser 2,
and the height equal to the desired length of the transition section. These were suspended
above the flange and the peaks of the triangles attached to the flange at 0º, 90º, 180º and
270º positions. Four very thin flexible sheets of fiberglass ( 0.76 mm thick), triangular in
shape were constructed, large enough to overlap the remaining open areas of the
transition. These were temporarily fitted to the existing assembly with a hot melt glue
gun. When the desired conforming curvature was attained, the entire assembly received
several layers of additional fiberglass lamination to achieve the necessary strength and
78
rigidity. External braces were added to the rectangular end to assist with maintaining
squareness of the corners.
A vertical web ( 0.15 m wide) that runs all around the circumference of the
transition near its entrance and several semi cylindrical fiberglass brackets are attached to
the various surfaces of the transition provide structural rigidity.
Structuralreinforcements
Fan
Transition
VibrationIsolator
Figure 4-11. Photograph of the transition piece.
Fan
A Twin City, single-width acoustafoil centrifugal fan that can deliver a flow rate of
369 /m s (147,000 cfm ) with a static pressure recovery of 1993 Pa ( 28" H O ) was
selected to be used to drive the tunnel. The fan can be operated at many rpm settings
using a Toshiba Model E3-430K variable frequency drive. A 224 kW (300 HP ) motor
that operates at a maximum speed of 884 rpm is used to drive the tunnel. The front and
back views of the fan are shown in Figure 4-12 and Figure 4-13.
79
Motor
Fan
Fan stand(filled withconcrete)
Slab
Figure 4-12. Front view of the fan.
Fan Exhaust
Transition
Figure 4-13. Back view of the fan.
80
Fan stand
Concrete slab
Retainer wallSand bed
Figure 4-14. View of the fan base.
Mechanical vibrations from the drive fan must be isolated from the main body of
the tunnel. The fan is attached firmly to a huge concrete slab, which rests on a sand bed
that is isolated from the surrounding foundation, as shown in Figure 4-14. The rotational
mass of the fan is approximately 680 kg (1500 lbs ). The slab measures approximately
6 4.4 0.46 m m m× × ( 20 14.3 1.5 ft ft ft× × ) and has a mass of 42.13 10 kg×
( 47,000 lbs ). The fan is also mounted on a concrete-filled stand, which is then mounted
on the concrete base that is surrounded by a retainer wall. Vibrations caused by the fan
are transmitted through the slab to the sand layer, which is 9 cm and is thus partially
isolated from the building.
Acoustic Treatment
In addition to the acoustic wedges attached to the chamber inner surface and the
acoustic lining on the diffuser duct work, further acoustic treatment had to be used in
order to reduce the background acoustic noise floor inside the chamber. The traverse
inside the anechoic chamber, used for microphone attachment, is wrapped with 13 mm
81
thick Nomex sheet (Figure 4-15). This reduces acoustic reflections from the traverse (see
chapter 2). To minimize the noise from the motor reaching the chamber, the garage door
area is partially covered with rack mounted acoustic flow wedges as shown in Figure 4-
16. Flow silencers that measure 0.61 0.76 1.2 m m m× × and used as chamber vent
channels are located on either side of the inlet contraction as shown in Figure 4-17.
These silencers incorporate adjustable louvers to regulate the vent flow. All
measurements in this thesis were obtained with the louvers completely closed.
All gaps in the chamber wall were filled with a 3M insulating expanding foam
called ‘Great Stuff’. This helps in eliminating flow leakage that causes a drop in test
section velocity, as well as the noise associated with the leaks.
Traversewrapped with
Nomex
Figure 4-15. Chamber traverse acoustic treatment.
82
100 Hzfiberglass
wedge
3.7 m
3.2
m
Figure 4-16. Garage door acoustic treatment.
Flow Silencer
Inlet
Figure 4-17. Photograph of the flow silencer.
83
CHAPTER 5 EXPERIMENTAL METHODS
This chapter describes the experimental setups and procedures for the various
experiments undertaken to characterize the wind tunnel. The experiments include
chamber deflection and wall loading measurements, static pressure measurements, flow
uniformity measurements, shear layer growth measurements, freestream turbulence
measurements, background noise measurements, fan noise decay measurements, and
vibration measurements. Equation Chapter 5 Section 1
Chamber Deflection and Wall Loading
While the tunnel is in operation, the static pressure is reduced inside the anechoic
chamber in accordance with the dynamic pressure of the freestream flow in the test
section. This causes a differential pressure load to act on the walls of the chamber, since
the pressure outside the chamber is atmospheric and the pressure inside, chamberP , is below
atmospheric pressure. As a result, the chamber walls deflect inwards. The walls of the
anechoic chamber were not originally designed to withstand large pressure loads. The
setup shown in Figure 5-1 was designed to measure the chamber wall deflection. A
Mitutoyo Vernier caliper having an accuracy of .25.4 mμ is attached to a support
structure made of plywood. The support structure was rigidly attached to the wall of the
building that houses the chamber. The tunnel was then run at several speeds and for each
speed the wall deflection is measured. The measurement was made in the center of the
largest wall panel of the chamber, which measures 3.05 m by 1.2 m (120" by 48" ).
84
Wall Panel
Vernier Caliper
Building Wall
Caliper Support
Figure 5-1. Photograph of the chamber deflection measurement setup.
Heise Unit
PchamberPatm
Anechoicchamber wall
Pressure tap
Model HQS-1
Figure 5-2. Schematic of the chamber wall loading measurement setup.
The experimental setup used to measure the corresponding pressure load acting on
the chamber walls is shown in Figure 5-2. A pressure tap made of copper tubing, 0.3 m
long and having an outer diameter of 6 mm was flush mounted inside the wall of the
chamber. The pressure tap connects to the low pressure input on a Heise unit
( 20 50 H O′′− , Model HQS-1) that measures differential pressure. The high pressure
input on the Heise unit was open to atmosphere. The tunnel was operated at different
speeds and 500 differential pressure measurements were taken for each speed and
85
averaged over a 4 minute period. The measurements were made on both the north and
south walls of the chamber.
Tunnel Circuit Static Pressure
Measurements of static pressure are made along the length of the wind tunnel
circuit to estimate the pressure drop or recovery in each component. The wall static
pressure measurements were made both in the inlet contract as well as in diffusers 1 and
2.
Twenty-four pressure taps were installed in the contraction section to measure the
wall pressure distribution. The pressure taps are 0.5 m ( 2" ) long and they have an outer
diameter of 1.6 mm ( 0.063"). Eight taps each were installed along the contraction base,
corner, and sidewall, as shown in Figure 5-3, and compared to the potential flow analysis
used to design the inlet. A photograph of the pressure tap attachment is shown in Figure
5-4. The axial location of the pressure taps are given in Table 6-1. The origin for the
measurements is located at the mid point of the inlet entrance. The pressure taps were
flush mounted inside the inlet and silicone caulk was used to attach the taps to holes
drilled into the walls of the inlet contraction.
All pressure data was collected using a Heise unit ( 20 50 H O′′− , Model HQS-1)
that measures the differential pressure. A pitot probe was located near the exit plane of
the inlet contraction. The heise unit was used the measure the differential pressure
between the wall static ports and the test section static pressure, in addition to the test
section dynamic pressure. The reported measurements, along with the conditions in the
test section, consist of an average of 30 measurements.
86
INLET
Staticpressure
ports
Flowdirection
1.12 m
0.74
m
3 m
Figure 5-3. Schematic of the inlet static pressure taps.
Table 6-1. Axial location of the inlet pressure taps.
Tap Sidewall
(m)
Base
(m)
Corner
(m)
1 2.9 3.0 2.9
2 2.4 2.5 2.4
3 2.2 2.3 2.3
4 2.0 2.1 2.1
5 1.8 1.8 1.8
6 1.4 1.4 1.4
7 0.8 0.9 0.8
8 0.2 0.3 0.2
87
Static taps
Inlet contraction
Figure 5-4. Photograph of the inlet static pressure taps.
Static pressure recovery measurements were also made in the diffuser. Four static
pressure taps, similar to the ones used for the wall loading measurements were flush
mounted inside diffuser 1 and five taps were flush mounted inside diffuser 2 and the
static pressure recovery was measured for various tunnel operating speeds. The
schematic of the test arrangement is similar to the one shown in Figure 5-11. The Heise
unit was used to measure the pressure data. A pitot probe was located near the entrance
plane of diffuser 1. The heise unit was used the measure the differential pressure
between the wall static ports and the test section static pressure, in addition to the test
section dynamic pressure.
Flow Uniformity
Flow uniformity measurements were made using a pitot rake with 32 ports
separated by 0 25.4 mm , as shown in Figure 5-5. The pitot rake measures the total
pressure at each port. The measurements were made at 3 different planes along the
length of the test section: near the inlet exit plane (3% TS length), near the test section
mid plane (43% TS length) and near the diffuser entrance plane (83% TS length). At
88
each plane, the pitot rake was traversed across the test section span and measurements
were made for several test section velocities.
The measured data was acquired using the Pressure Systems, Inc. pressure scanner
(Model 9116). Two 16-channel systems were used: one with a full-scale range of
2491 Pa ( 210 H O′′ ), and the other with a full-scale range of 6895 Pa (1 psi ). The
reported measurements, along with the conditions in the test section, consist of an
average of 1000 measurements taken over an 8 minute period.
Inlet
Diffuser 1
Flow
Pitot rakeInterrogation
points
Diffuser entranceplane
Test sectionmid plane
Inlet exitplane
0.8 m
1.2 m
1.12 m
0.74 m2.5 mm
Figure 5-5. Schematic of the flow uniformity measurement setup.
Shear Layer Growth
Streamwise velocity measurements were also made in the shear layer of the test
section jet to estimate its growth rate. A schematic of the measurement setup is shown in
Figure 5-6. A pitot tube, attached to a Velmex traverse was used to obtain the velocity in
89
the shear layer. A pitot tube that is connected to a Heise unit ( 20 50 H O′′− , Model HQS-
1), measures the differential pressure in the test section, that corresponds to the dynamic
pressure of the flow. The chamber temperature is measured using a Platinum RTD (Type
DIN-43760). The density is calculated from the flow temperature and the ambient
pressure outside the chamber, measured using a Multi-function Pressure Indicator (Druck
Model DPI-145). The velocity is calculated from the relation
2 h cP RTVP∞
Δ= (5.1)
where V is the velocity, hPΔ is the differential pressure measured by the Heise unit, R is
the universal gas constant, cT is the chamber temperature and P∞ is the ambient pressure.
Thirty values for hPΔ and 10 values for cT are acquired and averaged for each
interrogation point. During the course of the run the ambient pressure change was
measured and it was found to be less than 0.1 % . The pitot probe was traversed across
the shear layer, starting from a point inside the potential core to a point outside the shear
layer. The measurements were made for three different speeds, at 10 locations along the
length of the test section. The measurement coordinate system was located at a point in
the shear layer where the nondimensional velocity, ( ) ( )min max minV V V V− − has a value of
0.5. The measurements were made both in the y direction as well as the z− direction.
Coarser measurements were made in the potential core (every 25.4 mm ) and finer
measurements were made in the shear layer (every 2.5 12.7 mm− ). During the course of
the experiment, care was taken to ensure that the traverse system was located outside of
the jet.
90
Flow
Diffuser
Inlet
Potential core
Shear layer
Location 1
Location 9
Location 10
Interrogationpoint
x
y
Pitot probe
1.12 m
1.2 m
Figure 5-6. Schematic of the shear layer measurement setup.
Traverse
Pitot probe
Figure 5-7. Photograph of the shear layer measurement setup.
91
Freestream Turbulence Measurements
Turbulence measurements were made using a Constant Temperature Anemometer
(CTA) system. The CTA anemometer works on the principle of convective heat transfer
from a heated element to a surrounding fluid, which is related to the fluid velocity
(Jorgensen, 2002). The advantages of the CTA system is that it has the requisite flat high
frequency response required for turbulence measurements. Heat loss due to conduction
and radiation are assumed to be minimal. In addition, the velocity and the flow
properties are assumed to be constant across the length of the measurement probe.
A Dantec (Streamline CTA module 90C10) was used to measure the velocity
fluctuations in the test section entrance. The block diagram for the measurement is
shown in Figure 5-8. The measurement probe is a straight 1-D Tungsten probe (Type
55P11) and it is 5 mμ in diameter and 1.2 mm long. The probe connects to a probe
support (Type 55H20) which is connected to the CTA module via a 4 m long BNC cable
(TypeA1863). The CTA module contains a Wheatstone bridge, one arm of which is the
measurement probe. The resistance and hence the temperature of the probe are held
constant by a servo amplifier, by adjusting the current through the bridge circuit.
The hotwire probe was statically calibrated in situ before turbulence measurements
are made. The output from the CTA module was fed to an Integrating Voltmeter (Type
HP 34970A), which measures the dc component of the voltage averaged over 2 power
line cycles to suppress 60 Hz cycle noise. Fifty samples were taken and the mean value
was calculated, for each value of the test section velocity. The test section velocity was
measured using a pitot probe that was connected to a Heise unit ( 20 50 H O′′− , Model
HQS-1). The pitot tube was located ~ 0.0254 m from the hotwire probe The flow
92
temperature is measured using a Platinum RTD (Type DIN-43760) located very close to
the other probes. The density was calculated from the flow temperature and the ambient
pressure outside the chamber, measured using a Multi-function Pressure Indicator (Druck
Model DPI-145). The pitot tube measures the test section dynamic pressure, and
therefore the velocity can be estimated from Eq. (5.1). Fifty values of the dynamic
pressure and 10 values of temperature were averaged and used to calculate the mean
velocity in the test section. The change in ambient pressure was found to be less than
0.1 % . A plot was made between the mean velocity and the dc voltage from which the
calibration factor was estimated. More details of this procedure will be given in Chapter
6.
The overheat ratio for the turbulence measurements is set at 0.8. A square wave
test is done prior to measurements to estimate the bandwidth for the measurement setup.
The bandwidth of the system, estimated via the square wave test, was over 20 kHz for all
measurements. The continuous voltage signal from the CTA module was also sent to a
B&K Type 2827-002 Pulse data acquisition system to measure the autospectra of the
fluctuating part of the test section velocity. The pulse system has a frequency range of
dc- 25 kHz . The high pass filter for the pulse was set at 0.7 Hz . The frequency span
was set at 12.8 kHz with a frequency resolution of 1 Hz , and 500 overlapped (75%),
Hanning-windowed blocks were averaged. These acquisition and processing parameters
resulted in statistically converged and repeatable spectra.
93
CTA Module
B&K Pulse Unit
U+u’
Flow
Probe Probe support
Computer
Cable
IntegratingVoltmeter
V (DC)
v’ (AC)
Heise Unit
U
TempDAQ
Pitot probePt
RTD
T
Figure 5-8. Hotwire measurement block diagram.
Background Noise
A Brüel & Kjær (B&K) 1/ 4" Type 4939-A011 free-field microphone with a B&K
Type 2633 preamplifier was used to measure the background noise levels inside the
anechoic chamber with an empty test section. The microphone was statically calibrated
using a B&K Type 4228 pistonphone that provided a nominal amplitude of 124.7 (re
20 Paμ ) at a frequency of 251.2 Hz .
Both outflow noise and inflow noise were measured. A photograph of the
measurement setup is shown in Figure 5-9. The outflow microphone was attached to the
traverse wrapped with Tex Tech Industries Nomex acoustic absorber. The outflow
microphone was located at a radial distance of 1.9 m ( ~ 3.3 λ ) from the center of the test
section. This corresponds to the acoustic far field in future airframe noise measurements
in the facility. The outflow microphone was pointed directly at the center of the test
94
section. For the measurement of inflow noise, a B&K (Model UA-0385) nose cone was
attached to the microphone cartridge after the protective grid was removed. The inflow
microphone was attached to a rectangular plate support structure (see Figure 5-9), which
in turn was attached to a tripod. The nose cone was located inside the test section at a
distance of 0.89 m from the inlet exit plane in the axial direction, 0.11 m to the east of
the midpoint of the inlet exit plane in the span wise direction, and 0.23 m below the axis
of the test section. The outflow measurements were made both with and without the
inflow microphone. The measurements were repeated at several tunnel operating speeds.
Both third-octave and narrow-band noise spectra were measured. Data acquisition
was carried out using a B&K Type 2827-002 Pulse system with a frequency range of dc-
25 kHz . The frequency span for the third octave measurement ranged from 100 Hz −
20 kHz . For the narrow-band spectra, the frequency span was set at 12.8 kHz with a
frequency resolution of 2 Hz , and 500 overlapped (75%), Hanning-windowed blocks.
These acquisition and processing parameters resulted in statistically converged and
repeatable spectra.
Diffuser 1 Inlet
Outflow mic
Inflow mic
Figure 5-9. Photograph of the background noise measurement setup.
95
Fan Noise Attenuation
Acoustic measurements were made along the length of the diffuser to estimate the
decay of fan noise as it propagates through the acoustically treated duct work, into the
tests section. Coherent power measurements are made to estimate the fan noise
attenuation. The schematic of the coherent power measurement is shown in Figure 5-10.
Acoustic Liner
Outer wall
1/4 " microphone
outer tube
Flow Direction Sound propagation
Turbulent boundary layer
δ
micL
0.3
m
Figure 5-10. Coherent power measurement.
Microphones are flush mounted to the inner wall of the duct, using an outer steel
tube that runs all the way from the outer wall of the duct to the surface of the acoustic
liner. A schematic of the microphone holder is shown in Figure 5-11. The steel tube has
an outer diameter of 7.8 mm and, at the exit of the tube, there is a small notch where the
inner diameter shrinks to 6.3 mm , providing a snug fit for the 1/ 4" microphone.
The mean flow causes the development of a turbulent boundary layer along the
surface of the liner. The flush mounted microphones measure pressure that has
contributions from both the hydrodynamic pressure fluctuations in the turbulent boundary
96
layer as well the acoustic pressure fluctuations from the fan. In order to eliminate the
boundary layer pressure fluctuations, a cross spectrum of the signals from two
microphones separated by a distance ( micL ) significantly greater than the local turbulent
boundary layer thickness (δ ), is calculated. Autospectra of the two individual
microphones are also calculated from which the ordinary coherence function, 2γ , is
obtained. The coherent output power spectral density at the desired microphone location
is given by the formula (Bendat & Piersol, pg 566, 2000)
( ) ( ) ( )2xxCOP f f G fγ= (5.2)
where COP is the coherent output power spectral density, which is a function of
frequency f measured at the location of interest, ( )2 fγ is the coherence between the
microphone at the location of interest and a reference microphone, and ( )xxG f is the
autospectral density of the microphone at the location of interest. The total coherent
output power is obtained by integrating the coherent output power spectral density
function over the measurement frequency range.
A schematic of the measurement arrangement is shown in Figure 5-12. There are
nine positions in all flush mounted inside the diffusers. Four are located inside diffuser 1
and the five remaining locations are inside diffuser 2. The first microphone is mounted at
the entrance of the diffuser 1, just behind the collector, and the reference microphone is
mounted near the exit of diffuser 2, 1.83 m from the fan. The coordinate system origin
is located at the reference microphone position. The coherent power of the eight other
microphones with respect to the reference microphone is estimated (one at a time) and
then plotted as a function of distance.
97
Data acquisition was carried out using a B&K Type 2827-002 Pulse system with a
frequency range of dc- 25 kHz . The narrow-band spectra were measured, for which the
frequency span was set at 12.8 kHz with a frequency resolution of 4 Hz , and 500
overlapped (75%), Hanning windowed blocks were averaged. The measurements were
made for three different tunnel operating speeds.
Microphone
Supporttube
0.306"
0.25"
Figure 5-11. Schematic of the fan noise measurement microphone holder.
Fan
Transition
VibrationIsolator
Diffuser 2
CornerDiffuser 1
InletFlow
ProbeLocation
Figure 5-12. Setup for measurement of fan noise decay.
98
Table 5-1. Location of the diffuser microphones . Mic
Location
Distance
from
Reference
mic (m)
1 0.61
2 2.5
3 3.1
4 3.7
5 8.5
6 9.1
7 9.7
8 10.3
Background Noise Source Identification
The spectra measured by the outflow microphone located in the acoustic far field
have contributions from several sources. The loudest source is the noise from the test
section jet. There are also other secondary sources, the contributions of which to the
outflow spectra have to be estimated. We have followed a conditional spectral analysis
technique using the Multiple Input Single Output model (MISO) to estimate the noise
sources (Bendat and Piersol, pg 240, 2000). An output microphone is located in the far
field of the jet inside the anechoic chamber. Five input microphones are located close to
various noise sources as shown in Figure 5-13. One of the main sources of secondary
noise is the fan motor winding noise that enters the chamber primarily through the garage
door entrance. The microphone labeled ‘Input Mic 1’ in the figure is located close
(1.2 m ) to the motor. Another import source is the noise from the fan variable frequency
driver (VFD) that propagates into the chamber.
99
MotorVFD
CollectorFan Exhaust
Scrubbing
Inside view (micflush mounted with
diffuser 1 )
Input Mic 1Input Mic 2
Input Mic 3 Input Mic 4
Input Mic 5
Output Mic
A B
C DE
F
Figure 5-13. Location of the microphones for the source identification procedure.
‘Input Mic 2’ is therefore placed close to the VFD unit. ‘Input Mic 3’ is place
behind the collector, and it measures the unsteady pressure fluctuations generated by flow
impinging on the collector. The protective screen is removed and the nosecone is
100
attached to this microphone. Scrubbing noise from the shear flow over the diffuser wire
mesh is picked up by ‘Input Mic 4’ that is flush mounted inside diffuser 1. Finally the
noise from the fan is acquired by ‘Input Mic 5’ that is attached to the exit of the fan
exhaust. Unfortunately, each of these input microphones also measure noise from some
or all of the other sources and are thus, in general, correlated. A conditional spectral
technique is used to separate out the sources (Bendat & Piersol, 2000).
Four 1/ 4" (B&K Type 4939-A-011) microphones and two 1/ 8" (B&K Type
4138-A-015) microphones were used to measure the spectra. Data acquisition was
carried out using a B&K Type 3032-A Pulse system with a frequency range of dc-
25 kHz . The narrow-band spectra was measured, for which the frequency span was set
at 10 kHz with a frequency resolution of 6.25 Hz , and 1000 overlapped (75%), Hanning
windowed blocks were averaged. The measurements were made for three different
tunnel operating speeds ( 18 / , 30 / , 42 /TSU m s m s m s= ).
Vibration Measurements
The vibrations generated by the fan can propagate into the anechoic chamber
through the wind tunnel duct work as well as the foundation. A vibration isolation
section is installed between the transition section and diffuser 2 to minimize the
propagation of vibrations. Also the fan rests on a concrete base separated from the
building that houses the chamber by a retainer wall filled with sand. Measurements are
made to estimate the effectiveness of these vibration isolation techniques. Vibration
measurements are made by placing two accelerometers (PCB Model Number 356A16),
one before and one after the vibration isolator. The accelerometers were calibrated using
101
a PCB Model 394C06 hand held shaker that generates an acceleration of 1 g at a
frequency of 159.2 Hz .
The transmission loss, ( )10 1 220 logTL A A= across the vibration isolator is
calculated, where 1A and 2A are the rms acceleration measured by the two transducers.
Measurements are made across the fan base (Figure 5-14) and the duct work vibration
isolation section (Figure 5-15). Measurements are also made for an arrangement where
one accelerometer is attached to the fan base and the other one to the floor of the
chamber. Both in plane ( x ) and out of plane ( , y z ) components of the acceleration are
measured.
Data acquisition was carried out using a B&K Type 2827-002 Pulse system with a
frequency range of dc- 25 kHz . The narrow-band spectra was measured, for which the
frequency span was set at 6.4 kHz with a frequency resolution of 1 Hz , and 500
overlapped (75%), Hanning-windowed blocks were averaged. The measurements were
made for three different tunnel operating speeds.
Fan Slab
Retainer wallAccel 1
Accel 2
yx
z
Figure 5-14. Photograph of the fan vibration measurement test arrangement.
102
Accel 2 Accel 1
Flow
xy
z
Figure 5-15. Photograph of the vibration isolator vibration test arrangement.
Acoustic Liner Absorption Coefficient Measurement Setup
A Bruel and Kjaer Type 4206 impedance tube (Bruel and Kjaer user manual, 2004)
is used for the measurement of the normal incident absorption coefficient of the acoustic
liner sample. The schematic of the impedance tube is shown in Figure 5-16. A speaker
located at the entrance to the tube generates plane waves that propagate down the tube
( iP ) and impinges on the test sample. The test sample is held against a rigid backing
piston. A reflected plane wave component ( rP ) which depends on the acoustic properties
of the test sample is generated and it propagates upstream. The impedance tube acoustic
propagation is assumed to be locally reacting; where by the impedance of the sample is
independent of the angle of incidence. The absorption coefficient of the test sample is
obtained using the Two Microphone Method (ASTM-E1050-98 1998). Two
microphones (B&K Type 4939, ¼”) are flush mounted on the wall. The microphone
closest to the sample is located a distance of l from the sample, and the two microphones
are separated by a distance s .
103
Speaker
iP
rP
Acoustic linersample
Rigid pistonbacking
s l
Microphones
Figure 5-16. Schematic of the Impedance tube setup.
The complex reflection coefficient can be calculated form the expression
( )
( )( )212
12
jksjk l s
jks
H eR ee H
−+−
=−
(5.3)
where 12H is the frequency response function between the two microphones and k is the
wavenumber. The absorption coefficient can be calculated from the expression
21 Rα = − (5.4) The speaker generates a pseudo random signal in the frequency range 50 Hz -1.6
kHz . The B&K Pulse system was used for the data acquisition. The frequency
resolution used for the data acquisition was 1.25 Hz . One hundred averages of the data
were taken and a box car window function was used.
In the next chapter, the results from all the experiments listed in this chapter will be
presented and discussed in detail.
104
CHAPTER 6 FACILITY CHARACTERIZATION
This chapter discusses the measurements conducted in the anechoic wind tunnel for
the purpose of aerodynamic and acoustic characterization. A preliminary
characterization of the facility has been completed for speeds up to 43 /TSU m s= .
Chamber wall loading and deflection measurements, inlet contraction wall pressure
measurements, test section pitot surveys, shear layer measurements, hot-wire freestream
turbulence measurements, background noise measurements, and vibration measurements
have been performed. The detail results of the characterization experiments are described
in the sections below. Equation Chapter 6 Section 1
Chamber Deflection and Wall Loading
The drop in static pressure inside the test section due to the flow acceleration
causes a differential pressure load to act on the chamber. This loading in turn causes the
chamber wall panels to deflect.
Differential wall pressures were measured at two locations inside the anechoic
chamber. Figure 6-1 shows the wall loading as a function of test section velocity,
measured at both the north wall and the south wall of the chamber. The wall loading
shows a nonlinear dependence on velocity, which becomes critical as the flow speed
increases. Also shown in the plot is the dynamic pressure in the test section. It can be
observed that the wall loading is actually more than the dynamic pressure in the chamber.
This is probably due to the local acceleration of the flow along the test section length.
105
Figure 6-2 shows the variation of the ratio of effective velocity to the test section velocity
with the test section speed.
15 20 25 30 35 40 450
200
400
600
800
1000
1200
1400
UTS (m/s)
Wal
l Loa
ding
(Pa)
North WallSouth wallDynamic Pressure
Figure 6-1. Wall loading as a function of test section speed.
15 20 25 30 35 40 451.08
1.09
1.1
1.11
1.12
1.13
1.14
UTS (m/s)
U eff/U
TS
Figure 6-2. Variation of effective velocity with the test section velocity.
106
200 400 600 800 1000 1200 14000
2
4
6
8
10
12
14
Wall loading (Pa)
Wal
l def
lect
ion
(mm
)
Figure 6-3. Wall deflection vs test section velocity.
The effective velocity, effU is defined as the speed in the test section, for which the
corresponding dynamic pressure would match the wall loading shown in Figure 6-1. The
effective speed has to be 8-13 % higher than the test section speed to match the wall
loading. Figure 6-3 shows the wall deflection on one of the largest chamber panels
(3.02 m by 1.2 m ) due to the differential pressure loading. The walls of the chamber are
not designed for significant pressure loading. The defection increases gradually with test
section speed until a speed of about 43 /TSU m s= , beyond which the deflection becomes
highly nonlinear. The wall caves inward under the pressure load. It is not safe to operate
the tunnel beyond this point without further reinforcements (which is part of planned
future work).
Inlet Wall Pressure
Static wall pressure measurements were made along the length of the inlet
contraction to ensure its effectiveness. Eight pressure taps each were installed along the
107
contraction base, corner, and sidewall and compared to the potential flow analysis used to
design the inlet. The measured pressure coefficient distribution, ( ) 2- 0.5p TS TSC p p Uρ= ,
where ‘TS’ refers to the ‘test section’, is compared to that obtained from a potential flow
calculation and is shown in Figure 6-4 - Figure 6-6. In the graph the axial location x is
normalized by the length of the inlet contraction L . The measurements are made for TSU
values of 17 /m s , 30 /m s , and 42 /m s . The measurement uncertainties are very small.
On the corner and the sidewall, the pressure undershoot is larger than that predicted by
the potential flow, while good agreement is obtained along the base. The overall curve
shape of the pressure distribution combined with Stratford’s criterion indicates that the
flow does not separate along the walls. Attached flow was also verified via tuft flow
visualization in the contraction.
0 0.5 1-0.2
0
0.2
0.4
0.6
0.8
1
1.2a)
c p
0 0.5 1-0.2
0
0.2
0.4
0.6
0.8
1
1.2b)
x/L0 0.5 1
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1.2c)
ExperimentPotential flow
Figure 6-4. Contraction pC distributions versus length for the (a) sidewall, (b) base, and
(c) corner for 17 /TSU m s= .
108
0 0.5 1-0.2
0
0.2
0.4
0.6
0.8
1
1.2a)
c p
0 0.5 1-0.2
0
0.2
0.4
0.6
0.8
1
1.2b)
x/L0 0.5 1
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1.2c)
ExperimentPotential flow
Figure 6-5. Contraction pC distributions versus length for the (a) sidewall, (b) base, and
(c) corner for 30 /TSU m s= .
0 0.5 1-0.2
0
0.2
0.4
0.6
0.8
1
1.2a)
c p
0 0.5 1-0.2
0
0.2
0.4
0.6
0.8
1
1.2b)
x/L0 0.5 1
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1.2c)
ExperimentPotential flow
Figure 6-6. Contraction pC distributions versus length for the (a) sidewall, (b) base, and (c) corner for 42 /TSU m s= .
109
Figure 6-7 gives the comparison between the pressure drop measured across the
inlet section that includes the flow conditioners and contraction, and the theoretical
predictions from the fan loss calculations (Appendix E). The measurements are shown
for two different tunnel operating speeds. The results indicate that the agreement is
reasonable
The measurement of pressure drop across the test section was also measured.
Although the theory predicts a large drop in total pressure, measurements indicated that
the total pressure remains constant across the length of the test section. This is due to the
fact that the pitot probe was located inside the potential core, where there is hardly any
pressure drop.
1 20
10
20
30
40
50
Pres
sure
dro
p (P
a)
1 20
20
40
60
80
100
Pres
sure
dro
p (P
a)
expt
expttheory
theory
a)
b)
Figure 6-7. Comparison of the pressure drop across the inlet and flow conditioner section
for a test section speed of a) 18 /m s b) 37 /m s .
110
Diffuser Wall Pressure
In order ensure that the flow does not separate inside the diffusers, pressure taps
were mounted along diffuser 1 (four) and along diffuser 2 (four), as described in Figure
xxx. Static pressure measurements were made for test section velocity ranging from
18 /m s to 37 /m s . The results are shown in Figure 6-8. A pressure coefficient defined
by the expression ( ) ( )1wall w TSP P q− , where wP is the wall pressure at each diffuser tap,
1wP is the static pressure at the first port located nearest to the diffuser 1 entrance and TSq
is the test section dynamic pressure, is plotted along the y axis. The axial distance x is
normalized by the total length of both diffusers and the corner section, dL , measured
along the centerline of the wind tunnel circuit starting from diffuser 1 entrance. Also
shown in the figure is the theoretically computed pressure coefficient using the euler
equation. It can be observed that most of the pressure recovery occurs inside diffuser 1.
The static pressure is almost constant across diffuser 2 for all test section speeds. The
difference between the theoretical and experimental data is small inside diffuser 1,
whereas in diffuser 2 the static pressure is nearly constant, which is not the case predicted
by the theory. This is because most of the pressure recovery is occurring inside diffuser
1, and the entire length of diffuser 2 is not required for adequate pressure recovery.
The difference between the theoretical and experimental results can be attributed in
part due to the additional pressure recovery that occur in the liner section and also due to
the waviness of the wall and the non-flush mount of the pressure taps, as shown in Figure
6-9. Due to the waviness of the surface a flush mount of the pressure taps is not
achieved, and therefore the measurements made by the taps are not accurate. Future
work is required to examine the flow in the diffuser more closely.
111
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
x/Ld
(Pw
all-P
w1)/(
q TS)
Diffuser 1 Diffuser 2
VTS=18 m/s (expt)
VTS=30 m/s (expt)
VTS=37 m/s (expt)
Theory
Figure 6-8. Pressure recovery across the diffuser.
Pressure/Microphone tap
Wall waviness
Corner
Figure 6-9. Photograph showing the waviness of the inner surface of diffuser 2.
112
1 20
10
20
30
40
50
Pres
sure
reco
very
(Pa)
1 20
20
40
60
80
100
Pres
sure
reco
very
(Pa)
theory
theory
expt
expt
b)
a)
Figure 6-10. Comparison of the pressure recovery across the diffuser duct work section
for a test section speed of a) 18 /m s b) 37 /m s .
Figure 6-10 gives the comparison between the pressure drop measured across the
diffuser duct work that includes diffuser 1, corner/turning vanes, diffuser 2, and the
transition section. The measurements are shown for two different tunnel operating
speeds. There is larger drop in pressure across the duct work than what is predicted.
This is probably due to the fact that the waviness of the diffuser inner surface can be
theoretically modeled.
Flow Uniformity
Flow uniformity measurements were made at three streamwise locations along the
test section length. All measurements were made for a test section speed of 17 /m s .
Figure 6-11 and Figure 6-12 show contours of the normalized total pressure distribution,
,maxt tP P , at the entrance to the test section and diffuser, respectively. The uniform core
and the growth of the bounding shear layer are evident. Figure 6-13 shows the mid-plane
113
velocity profile development along the length of the test section. The test section flow
non-uniformity is 2.5% based on the difference between the minimum and the maximum
inlet velocity outside the boundary layer. Alternatively, by taking the difference between
the contraction corner velocity (outside the boundary layer) and the contraction centerline
velocity, the flow non-uniformity is approximately 1%. A final statistical measure of
flow non-uniformity is the normalized rms variation of the axial velocity in the potential
core, which is 0.7%.
x/We
y/H
e
-0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Figure 6-11. Normalized stagnation pressure contours (max =1 w/ 0.1 interval) at the test
section entrance. eH and eW are the height and the width at the diffuser 1 entrance for a test section speed of 17 /m s .
114
x/We
y/H
e
-0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Figure 6-12. Normalized stagnation pressure contours (max =1 w/ 0.1 interval) at the
diffuser entrance for a test section speed of 17 /m s .
0 0.2 0.4 0.6 0.8 1 1.2 1.4-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
U/UTS
y/H
e
3% Test Section Length43% Test Section Length83% Test Section Length
Figure 6-13. Test section centerline velocity profile development along the test section
length for a test section speed of 17 /m s .
115
Shear Layer Behavior
Normalized shear layer profiles are shown in Figure 6-14 - Figure 6-19.
Measurements are made for three different speeds in the yz plane in both the y direction
as well as the x direction. The y and z axis locations are normalized by the local
measured momentum thickness, ( )xθ . The origin of the local coordinate system is
located at the point in the shear layer where the normalized velocity profile,
( ) ( )min max minV V V V− − has a value of 0.5. The measurements are done at ten streamwise
locations along the test section length. In the figures, the streamwise locations are
normalized by the momentum thickness value at the first streamwise location closest to
the inlet exit plane. It can be observed that the normalization collapses all the shear layer
velocity profiles to a self similar velocity profile. It can also be observed that the error
bars in the velocity profile are larger towards the edge of the shear layer than those near
the entrance of the shear layer.
The variation of the momentum thickness with normalized axial distance is shown
in Figure 6-20 and Figure 6-21. It is clear that the shear layers grow almost in a linear
fashion, from the exit of the inlet to the entrance of the diffuser. A linear least square fit
is also made to the data as shown in the figures. It can be observed that the slope of the
curve (jet growth rate) has a value around 0.03 in the y direction and a value of 0.04 in
the z direction. These values are lower when compared to the typical value for the
growth of a free shear layer, which has a value of 0.045 (Ho & Huerre 1984). This is due
to the fact that the test section jet is a confined jet with limited amount of air for
entrainment, and also the downstream boundary condition is set by the fan, unlike in a
free jet.
116
-5 -4 -3 -2 -1 0 1 2 3 4 50
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
y/θ
(V-V
min
)/(V
max
-Vm
in)
x/θ1=20.6
x/θ1=32.2
x/θ1=44.2
x/θ1=56.4
x/θ1=68.6
x/θ1=86.9
x/θ1=105.3
x/θ1=124.5
x/θ1=142.2
x/θ1=154.2
Figure 6-14. Normalized velocity profile in the yz plane in the y direction for
18 /TSU m s= ( 1 7.8 mmθ = ).
-5 -4 -3 -2 -1 0 1 2 3 4 50
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
y/θ
(V-V
min
)/(V m
ax-V
min
)
x/θ1=23.2
x/θ1=36.3
x/θ1=49.8
x/θ1=63.6
x/θ1=77.4
x/θ1=98
x/θ1=118.6
x/θ1=140.2
x/θ1=160.2
x/θ1=173.7
Figure 6-15. Normalized velocity profile in the yz plane in the y direction for
30 /TSU m s= ( 1 6.9 mmθ = ).
117
-5 -4 -3 -2 -1 0 1 2 3 4 50
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
y/θ
(V-V
min
)/(V
max
-Vm
in)
x/θ1=21.4
x/θ1=33.4
x/θ1=45.9
x/θ1=58.6
x/θ1=71.3
x/θ1=90
x/θ1=109.3
x/θ1=129.2
x/θ1=147.6
x/θ1=160.1
Figure 6-16. Normalized velocity profile in the zy plane in the y direction for
37 /TSU m s= ( 1 7.5 mmθ = ).
-5 -4 -3 -2 -1 0 1 2 3 4 50
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
z/θ
(V-V
min
)/(V m
ax-V
min
)
x/θ1=22.7
x/θ1=35.5
x/θ1=48.8
x/θ1=62.2
x/θ1=75.7
x/θ1=95.9
x/θ1=116.1
x/θ1=137.3
x/θ1=156.8
x/θ1=170
Figure 6-17. Normalized velocity profile in the yz plane in the z direction for
18 /TSU m s= ( 1 7.1 mmθ = ).
118
-5 -4 -3 -2 -1 0 1 2 3 4 50
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
z/θ
(V-V
min
)/(V m
ax-V
min
)
x/θ1=22.1
x/θ1=34.6
x/θ1=47.6
x/θ1=60.7
x/θ1=73.9
x/θ1=93.6
x/θ1=113.3
x/θ1=133.9
x/θ1=153
x/θ1=165.9
Figure 6-18. Normalized velocity profile in the yz plane in the z direction for
30 /TSU m s= ( 1 7.2 mmθ = ).
-5 -4 -3 -2 -1 0 1 2 3 4 50
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
z/θ
(V-V
min
)/(V m
ax-V
min
)
x/θ1=22.4
x/θ1=35.0
x/θ1=48.1
x/θ1=61.4
x/θ1=74.7
x/θ1=94.7
x/θ1=114.6
x/θ1=135.5
x/θ1=154.7
x/θ1=167.8
Figure 6-19. Normalized velocity profile in the yz plane in the z direction for
37 /TSU m s= ( 1 7.2 mmθ = ).
119
0 0.1 0.2 0.3 0.4 0.5 0.6 0.70
0.02
0.04
0.06b)
θ (m
) y = 0.056*x + 0.0023
data linear fit
0 0.1 0.2 0.3 0.4 0.5 0.6 0.70
0.02
0.04
0.06c)
x/LTS
y = 0.056*x + 0.0025
data linear fit
0 0.1 0.2 0.3 0.4 0.5 0.6 0.70
0.02
0.04
0.06a)
y = 0.064*x + 0.0018
data linear fit
Figure 6-20. Variation of y momentum thickness with test section length for a)
18 /TSU m s= b) 30 /TSU m s= c) 37 /TSU m s= .
0 0.1 0.2 0.3 0.4 0.5 0.6 0.70
0.02
0.04
0.06a)
y = 0.078*x - 0.0023
data linear fit
0 0.1 0.2 0.3 0.4 0.5 0.6 0.70
0.02
0.04
0.06b)
θ (m
)
y = 0.079*x - 0.0019
data linear fit
0 0.1 0.2 0.3 0.4 0.5 0.6 0.70
0.02
0.04
0.06c)
x/LTS
y = 0.074*x - 0.001
data linear fit
Figure 6-21. Variation of z momentum thickness with test section length for a)
18 /TSU m s= b) 30 /TSU m s= c) 37 /TSU m s= .
120
0 0.1 0.2 0.3 0.4 0.5 0.6 0.718
18.5
19
0 0.1 0.2 0.3 0.4 0.5 0.6 0.730.5
31
31.5U
TS (
m/s
)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.736
37
38c)
x/LTS
b)
a)
Figure 6-22. Variation of the potential core velocity along the test section length for a)
18 /TSU m s= b) 30 /TSU m s= c) 37 /TSU m s= .
Figure 6-22 shows the variation of the potential core flow velocity ( )( )maxV x with
axial distance, for the three different speeds ( 18 / , 30 / , 37 /TSU m s m s m s= ). It can be
observed that the maximum variation in the test section speed is < 1%.
Freestream Turbulence
Freestream turbulence measurements are obtained at three locations at the exit
plane of the inlet contraction. Location A is at the center of the exit plane, Location B is
0.28 m (25 % span) to the right of Location A and Location C is 0.28 m to the left of
Location A. Measurements are made for three different operating speeds of the tunnel
( 17 / , 28 / , 38 /TSU m s m s m s= ).
Figure 6-24 shows the static calibration curve, which is a plot of test section
velocity versus measured dc voltage. Measurements were made as the tunnel was
121
increases as well as when the tunnel speed was decreased. The figure shows that both
cases are close to each other and free of hysteresis. In order to account for the influence
of flow temperature ( fT ) on the measurements, the calibration curve was re-plotted with
the x axis replaced by ( )2 / w fV T T− , where V is the dc voltage and wT is the temperature
of the hot wire (Jorgensen 2005). This lead to a better collapse of the two different
measurements (‘up’ and ‘down’). A cubic polynomial fit of the calibration data is shown
in Figure 6-26. The slope of the calibration curve at the desired location ( TSU ) gives the
calibration factor.
Figure 6-27 - Figure 6-29 show the power spectra for the three locations. It can be
observed that beyond approximately 1 kHz , the energy content of the turbulent
fluctuations is negligible. This was verified by increasing the frequency span and
comparing the results. Most of the energy is at very low frequencies as expected. Table
6- 2 gives the uncertainty estimates for the hot wire spectra (Bendat and Piersol 2000).
A
Inlet exit plane
B Cy
z
0.74 m
1.12 m
0.28 m 0.28 m
Figure 6-23. Locations of the hotwire measurement.
122
2.1 2.15 2.2 2.25 2.3 2.35 2.4 2.4515
20
25
30
35
40
Voltage (V)
UTS
(m/s
)
UpDown
Figure 6-24. Calibration curve showing the plot of mean velocity vs. mean voltage.
0.02 0.021 0.022 0.023 0.024 0.025 0.026 0.02715
20
25
30
35
40
V2/(Tw-Tf)
U TS (m
/s)
UpDown
Figure 6-25. Calibration curve corrected for flow temperature.
123
15 20 25 30 35 400.019
0.02
0.021
0.022
0.023
0.024
0.025
0.026
0.027
V2 /(T
w-T
f)
UTS (m/s)
y = 5.3e-008*x3 - 7.4e-006*x2 + 0.00054*x + 0.013
Up cubicDown
Figure 6-26. Cubic fit to the calibration curve.
100 101 102 103 10410-12
10-10
10-8
10-6
10-4
10-2
100
Frequency (Hz)
Pow
er S
pect
ra (m
/s)2
Vts=17 m/sVts=28 m/sVts=39 m/s
Figure 6-27. Turbulence spectra at location A.
124
100 101 102 103 10410-12
10-10
10-8
10-6
10-4
10-2
Frequency (Hz)
Pow
er S
pect
ra (m
/s)2
UTS=17 m/s
UTS=28 m/s
UTS=38 m/s
Figure 6-28. Turbulence spectra at location B.
100 101 102 103 10410-12
10-10
10-8
10-6
10-4
10-2
Frequency (Hz)
Pow
er S
pect
ra (m
/s)2
UTS=17 m/s
UTS=28 m/s
UTS=38 m/s
Figure 6-29. Turbulence spectra at location C.
125
Table 6- 2. Spectral error estimates.
Location Speed
( /m s )
Mean error
( )%
Rms error
( )%
A 17 2.5 4.7
A 28 2.1 3.3
A 38 1.8 1.8
B 17 2.9 10
B 28 2.0 3.0
B 38 1.8 1.8
C 17 2.9 11
C 28 2.0 3.5
C 38 1.8 1.8
From the hot wire spectra, the freestream turbulence intensity (TI) was calculated
for each location and speed and is summarized in Table 6-3. The lower cut off frequency
is chosen based on the fact that the largest turbulence fluctuations have to be the order of
the test section size. Anything larger than this is considered as flow unsteadiness (Wind
Tunnel Design 2005). For the lower frequency cutoff is 5 Hz , TI at location A has a
value of 0.035% for a test section speed of 18 /m s . This value increases to 0.057% at a
speed of 38 /m s . A similar trend is found for the two other locations. For each location
and speed, two different experimental runs were made. As shown in the table, the
experiments were found to be very repeatable. The uncertainty in the freestream
turbulences values were less than 0.0005%.
126
Table 6-3. Free Stream Turbulence Intensity. Freestream Turbulence Intensity (%)
Location A Location B Location C Speed (m/s)
Cut off
frequency
(Hz) Run 1 Run 2 Run 1 Run 2 Run 1 Run 2
17 5 0.035 0.035 0.038 0.038 0.038 0.038
28 8 0.036 0.036 0.055 0.055 0.042 0.042
38 10 0.057 0.057 0.066 0.066 0.061 0.061
Background Noise Measurements
The narrow-band and third octave outflow spectra are shown in Figure 6-30 and
Figure 6-31. The microphone was located at a distance of 1.9 m from the center of the
test section in the far field (~3.3 wavelengths) based on the lowest frequency of interest
(600 Hz ) for an estimated trailing edge noise spectrum in the anechoic chamber. Figure
6-30 shows the narrow-band spectra for various speeds, including a noise measurement
with no flow. The low frequency peak at 282 Hz , which is visible at a test section
velocity of 18.1 /m s , is due to the noise propagating into the anechoic chamber from a
cooling fan used to cool the variable frequency drive located just outside the chamber
adjacent to the tunnel inlet. At higher velocities this peak is buried in the background
noise. Similarly, Figure 6-31 shows the third octave-band spectra. The overall sound
pres sure levels (OASPL) inside the chamber in the frequency range from 100 Hz − 20
kHz range from 40.8 dB with no flow to 74.7 dB at =43 /TSU m s . At this speed, the
narrow-band SPL at 600 Hz is 40 dB and 18.4 dB at 10 kHz . These levels are
significantly below projected levels associated with typical airframe noise components,
including trailing edge noise, thus ensuring high signal-to-noise ratio measurements.
127
Various spectral uncertainty estimates for the narrow-band spectra are given in Table 6-4
(Bendat & Piersol, 2000).
102 103 104-10
0
10
20
30
40
50
60
70
80
Frequency (Hz)
SPL
(dB)
No FlowV=18.1 m/sV=24.4 m/sV=30.6 m/sV=36.8 m/sV=43 m/s
Figure 6-30. Narrow-band Spectra.
102 103 1040
10
20
30
40
50
60
70
Frequency (Hz)
SPL
(dB
)
No FlowV=18.1 m/sV=24.4 m/sV=30.6 m/sV=36.8 m/sV=43 m/s
Figure 6-31. 1/3rd Octave Band Spectra.
128
Table 6-4. Error estimates for the background noise spectra. Speed
( /m s )
Mean error
dB
Rms error
dB
18 0.14 0.16
24 0.14 0.15
30 0.13 0.14
37 0.13 0.14
43 0.13 0.13
101
102
50
55
60
65
70
75
80
OA
SPL
(dB)
U (m/s)
data
prms2 ~U5.6
Figure 6-32. OASPL vs test section velocity
129
102 103 1040
10
20
30
40
50
60
70
80
Frequency (Hz)
SPL
(dB)
UF (no flow)ND (no flow)UF (V=20 m/s)ND (V=20 m/s)UF (V=25 m/s)ND (V=25 m/s)
Figure 6-33. Comparison of UF and Notre Dame tunnel background noise.
From these spectra, the overall sound pressure level as a function of tunnel velocity
can be calculated and converted to the overall mean-square pressure 2rmsP . Figure 6-32
shows the result; 2rmsP (expressed as OASPL, and not A weighted) exhibits a 5.6U
dependence on the test section velocity.
Figure 6-33 compares the third octave band spectra measured in the UF wind
tunnel to that measured in a similar tunnel at the University of Notre Dame (Mueller et al.
1992). The far field microphone was located at a distance of 1.5 m from the center of
inlet exit plane. For the two different speeds, it can be seen that the background noise
levels for the UF tunnel is ( )10O dB lower than the Notre Dame tunnel. It should also
be noted that the Notre Dame tunnel has a smaller test section ( 0.61 m by 0.61 m ) than
the UF tunnel ( 0.74 m by 1.12 m ).
130
Figure 6-34 shows the narrow-band inflow spectra measured inside the test section.
It can be observed that the inflow noise levels are significantly higher than the
corresponding outflow noise levels (Figure 6-30). However, the measurements were
made with a B&K UA0385 nose cone attached to the microphones. An aerodynamic
strut attachment for the inflow microphone, typically used for inflow acoustic
measurements (Mueller, 2002) was not available when the measurements were made.
Outflow measurements were also made in the far field while the inflow microphone was
installed in the test section. The difference between the outflow spectra and the outflow
spectra in the presence of the inflow microphone ( SPLδ ) is shown in Figure 6-35. At a
frequency of 1 kHz , the difference between the two measurements is as high as 8 dB for
a test section speed of 43 /m s . The difference between the two measurements can be
attributed to the boundary layer noise emanating from the nose cone surface, cavity noise
generated due to the flow over the nose cone mesh, and also the noise generated by the
interaction of the flow with the inflow microphone support structure.
By examining the third octave spectrum at the highest tested speeds thus far and
applying A-weighting, a qualitative comparison can be drawn to other facilities using
their previously published data. This A-weighted OASPL vs. test section velocity
comparison is shown in Figure 6-36 (adapted from Duell et al. 2002). While little data
about microphone type, placement, or other experimental setup details is available, these
data demonstrate a qualitative success in terms of achieving an acoustic flow facility with
low background noise.
131
102 103 104-10
0
10
20
30
40
50
60
70
80
90
100
Frequency (Hz)
SP
L (d
B)
No FlowV=18 m/sV=24 m/sV=30 m/sV=36 m/sV=42 m/s
Figure 6-34. Narrow band inflow spectra.
102
103
104
-4
-2
0
2
4
6
8
10
12
14
Δ S
PL (d
B)
Frequency (Hz)
V=18 m/sV=31 m/sV=43 m/s
Figure 6-35. The influence of inflow microphone on the outflow spectra.
132
UFUF
Figure 6-36. Facility comparison of A-weighted in flow noise levels.
Fan Noise Decay
As described in Chapter 5, coherent output power measurements were made to
estimate the decay of fan noise as it propagates through the acoustically treated duct work
into the test section. A reference microphone was flush mounted inside diffuser 2, near
its exit plane and 1.83 m from the fan. A second microphone was sequentially moved to
one of eight other locations. Power spectra and the coherent power spectra with respect
to the reference microphone for each of the eight microphone locations were measured
for three different test section speeds ( 18 /TSU m s= , 30 /TSU m s= , 42 /TSU m s= ).
The integrated total power for the microphones mounted inside diffuser 2 and diffuser 1
is shown in Figure 6-37 and Figure 6-38, respectively. The axial location x, measured
from the reference microphone is non dimensionalized by dL (10.3 m ), the distance from
the reference microphone through the tunnel circuit until the collector entrance where the
last microphone is located. The integrated coherent power for the microphones mounted
inside diffuser 2 and diffuser 1 is shown in Figure 6-39 and Figure 6-40, respectively.
133
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.40
1
2a)
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.40
20
40b)
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.40
100
200c)
x/Ld
Tota
l Pow
er (P
a)2
Figure 6-37. Total power measured by the diffuser 2 microphones for a) 18 /TSU m s=
b) 30 /TSU m s= c) 42 /TSU m s= .
0.82 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 10
50
100a)
0.82 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 10
1000
2000b)
0.82 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 10
5000
10000c)
x/Ld
Tota
l Pow
er (P
a)2
Figure 6-38. Total power measured by the diffuser1 microphones for a) 18 /TSU m s= b)
30 /TSU m s= c) 42 /TSU m s= .
134
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.41
2
3x 10
-3 a)
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.40
0.05
0.1b)
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.40
0.2
0.4c)
x/Ld
Tota
l Coh
eren
t Pow
er (P
a)2
Figure 6-39. Total coherent power measured by the diffuser2 microphones for a)
18 /TSU m s= b) 30 /TSU m s= c) 42 /TSU m s= .
0.8 0.82 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 10
0.2
0.4a)
0.8 0.82 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 10
2
4b)
0.8 0.82 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 10
10
20c)
x/Ld
Tota
l Coh
eren
t Pow
er (P
a)2
Figure 6-40. Total coherent power measured by the diffuser1 microphones for a)
18 /TSU m s= b) 30 /TSU m s= c) 42 /TSU m s= .
135
The coherent power, indicative of acoustic energy, for the microphones attached to
diffuser 2 decreases with increasing distance from the reference microphone, as expected.
However, for the microphones mounted inside diffuser 1, the coherent power increases
with increasing distance, although the fan acoustic power continues to decrease as one
moves away from the fan. This observed increase is likely due to the proximity of
diffuser 1 to the test section jet collector. The jet impingement may serve as a strong
acoustic source that decreases as one moves away from the collector.
Background Noise Source Identification
A conditional spectral analysis technique was used to identify the various sources
that contribute to the test section background noise (Bendat &Piersol, 2000). The
schematic of the setup is shown in Figure 5-13. As described in Chapter 5 and
summarized in Table 6-5, the output microphone labeled ‘output’ was located in the
acoustic far field of the test section jet. ‘Input 1’ microphone was located close to the fan
motor, ‘Input 2’ microphone was located close to the exhaust of the fan, ‘Input 3’
microphone was located near to the VFD, ‘Input 4’ microphone was flush mounted inside
diffuser 1 to measure the scrubbing noise and ‘Input 5’ microphone was located behind
the collector.
The schematic of the MISO model is shown in Figure 6-41 In the model 1X - 5X
are the five different inputs, 1H - 5H are the frequency response functions, Y is the
measured output, and N is the noise in the system. For our case, the noise N stands for
the jet noise in the test section devoid of any influences from the input. The noise spectra
is obtained from the relation
nn yy vvG G G= − (6.1)
136
where yyG is the measured output spectra and vvG is the model for the contribution of the
inputs alone to the output spectra. The model spectra is obtained from the relation,
5
*
1ivv i x y
iG H G
=
= ∑ (6.2)
where *iH is the complex conjugate of the transfer function and
ix yG is the cross spectra
between the inputs and the output. The transfer function is obtained from solving the
matrix equation
1 1 1 2 1 3 1 4 1 51
2 1 2 2 2 3 2 4 2 5
3 3 3 4 3 5
4 4 4 5
5 5 1 5 2 5 5
1
5
... . ..
.. . . .
. .
x x x x x x x x x xx y
x x x x x x x x x x
x x x x x x
x x x x
x y x x x x x x
G G G G GG HG G G G G
G G G
G GHG G G G
⎡ ⎤⎡ ⎤ ⎡ ⎤⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥= ⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎣ ⎦⎢ ⎥⎣ ⎦ ⎣ ⎦
(6.3)
1X
2X
3X
4X
5X
1H
2H
3H
4H
5H
1Y
2Y
3Y
4Y
5Y
N
YV
Figure 6-41. Schematic of the MISO model.
137
Table 6-5. Nomenclature for input and output microphones. Signal Microphone
Input 1 Motor
Input 2 Fan
Input 3 VFD
Input 4 Scrubbing
Input 5 Collector
Output Outflow
The measured power spectra are shown in Figure 6-42 - Figure 6-44, for various
test section speeds. The ordinary coherence functions between the input microphones
and the output microphone are shown in Figure 6-45 - Figure 6-47.
It can be observed that motor noise is reasonably correlated with the output at
approximately 4.2 kHz. The VFD noise is correlated at lower speeds of the tunnel. The
conditional spectral technique fits a model spectra ( vvG ) to the measured output
microphone spectra ( yyG ). It then removes the contribution of all the input sources to the
output spectra and estimates the resulting residual noise spectra ( nnG ). In this case, the
resulting nnG is the acoustic spectra of the test section jet in which all contaminating
sources have been removed. The results are shown in Figure 6-48 - Figure 6-50. It can
be observed that the residual noise spectra closely match the output spectra for all speeds.
There is slight discrepancy between the two in the lower frequency range (<100 Hz),
which is below the cut off range for the tunnel. The total contribution of the input
sources to the output power is: 6.9 % for 18 /TSU m s= (Figure 6-51), 9.6 % for
30 /TSU m s= (Figure 6-52), and 14% for 42 /TSU m s= (Figure 6- 53). This implies
138
that that the contribution from the extraneous sources are small and we indeed obtain a
relatively output spectra from the outflow microphone measurement.
100 101 102 103 1040
20
40
60
80
100
120
Frequency (Hz)
Aut
ospe
ctra
(dB
re 2
0 μ
Pa)
Input 1Input 2Input 3Input 4Input 5Output
Figure 6-42. Autospectra of the input and output microphones for 18 /TSU m s= .
100 101 102 103 1040
20
40
60
80
100
120
Frequency (Hz)
Auto
spec
tra (d
B re
20
μPa
)
Input 1Input 2Input 3Input 4Input 5Output
Figure 6-43. Autospectra of the input and output microphones for 30 /TSU m s= .
139
100 101 102 103 10420
40
60
80
100
120
140
Frequency (Hz)
Aut
ospe
ctra
(dB
re 2
0 μ
Pa)
Input 1Input 2Input 3Input 4Input 5Output
Figure 6-44. Autospectra of the input and output microphones for 42 /TSU m s= .
100 101 102 103 1040
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Frequency (Hz)
γ2
γ216
γ226
γ236
γ246
γ256
Figure 6-45. Ordinary coherence between the input microphones and output microphone
for 18 /TSU m s= .
140
100 101 102 103 1040
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
Frequency (Hz)
γ2
γ216
γ226
γ236
γ246
γ256
Figure 6-46. Ordinary coherence between the input microphones and output microphone
for 30 /TSU m s= .
100 101 102 103 1040
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
Frequency (Hz)
γ2
γ216
γ226
γ236
γ246
γ256
Figure 6-47. Ordinary coherence between the input microphones and output microphone
for 42 /TSU m s= .
141
100 101 102 103 104-40
-20
0
20
40
60
80
Frequency (Hz)
Aut
ospe
ctra
(dB
re
20μ
Pa)
Gyy
GvvGnn
Figure 6-48. Comparison of the MISO model to the measured spectra for 18 /TSU m s= .
100 101 102 103 104-20
0
20
40
60
80
100
Frequency (Hz)
Aut
ospe
ctra
(dB
re 2
0μ
Pa)
Gyy
GvvGnn
Figure 6-49. Comparison of the MISO model to the measured spectra for 30 /TSU m s= .
142
100 101 102 103 104-10
0
10
20
30
40
50
60
70
80
90
Frequency (Hz)
Aut
ospe
ctra
(dB
re 2
0 μP
a)
Gyy
GvvGnn
Figure 6-50. Comparison of the MISO model to the measured spectra for 42 /TSU m s= .
100 101 102 103 1040
0.01
0.02
0.03
0.04
0.05
0.06
0.07
Frequency (Hz)
Tota
l Pow
er (P
a2 )
Gyy
GvvGnn
Figure 6-51. Total power for model and measured output for 18 /TSU m s= .
143
100 101 102 103 1040
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
Frequency (Hz)
Tota
l Pow
er (P
a2 )
Gyy
GvvGnn
Figure 6-52. Total power for model and measured output for 30 /TSU m s= .
100 101 102 103 1040
0.5
1
1.5
Frequency (Hz)
Tota
l Pow
er (P
a2 )
Gyy
GvvGnn
Figure 6- 53. Total power for model and measured output for 42 /TSU m s= .
Vibration Measurements
The schematic of the fan vibration measurement setup is shown in Figure 5-14.
Figure 6-54 - Figure 6-56 shows the autospectra of the accelerometers attached to the fan
144
base (accelerometer 1) and the retainer wall (accelerometer 2). All three components of
the acceleration are measured for three different speeds
( 18 / , 30 / , 42 /TSU m s m s m s= ). It can be observed that the out-of-plane or normal
acceleration (x) is higher than the in-plane components. Therefore the transmission loss
across the fan base was calculated for the x axis. The results are shown in Figure 6-57.
Over the frequency range of interest the there is significant attenuation of vibration due to
the vibration isolation base. The transmission loss values are over 15 dB from 500 Hz
to 5 kHz range for all velocities and for velocities 30 /TSU m s= and higher, the
transmission loss increases to over 30 dB for frequencies above 2500 Hz . A similar
trend was observed across the fan base and the chamber floor (Figure 6-58).
100 101 102 103 104-120
-100
-80
-60
-40a)
Acc
eler
atio
n (d
B re
1 g
) Accelerometer 1 (x)Accelerometer 1 (y)Accelerometer 1 (z)
100 101 102 103 104-150
-100
-50b)
Frequency (Hz)
Acc
eler
atio
n (d
B re
1 g
) Accelerometer 2 (x)Accelerometer 2 (y)Accelerometer 2 (z)
Figure 6-54. Autospectra of the accelerometers attached to a) Fan slab b) Retainer wall
for 18 /TSU m s= .
145
100 101 102 103 104-120
-100
-80
-60
-40a)
Acc
eler
atio
n (d
B r
e 1
g) Accelerometer 1 (x)Accelerometer 1 (y)Accelerometer 1 (z)
100 101 102 103 104-150
-100
-50b)
Frequency (Hz)
Acce
lera
tion
(dB
re 1
g) Accelerometer 2 (x)
Accelerometer 2 (y)Accelerometer 2 (z)
Figure 6-55. Autospectra of the accelerometers attached to a) Fan slab b) Retainer wall
for 30 /TSU m s= .
100
101
102
103
104
-120
-100
-80
-60
-40
-20a)
Acc
eler
atio
n (d
B r
e 1
g) Accelerometer 1 (x)Accelerometer 1 (y)Accelerometer 1 (z)
100
101
102
103
104
-150
-100
-50b)
Frequency (Hz)
Acc
eler
atio
n (d
B re
1 g
)
Accelerometer 2 (x)Accelerometer 2 (y)Accelerometer 2 (z)
Figure 6-56. Autospectra of the accelerometers attached to a) Fan slab b) Retainer wall
for 42 /TSU m s= .
146
0 500 1000 1500 2000 2500 3000 3500 4000 4500 50000
15
30
4555
a)
0 500 1000 1500 2000 2500 3000 3500 4000 4500 50000
15
30
4555
b)Tr
ansm
issi
on L
oss
(dB
)
0 500 1000 1500 2000 2500 3000 3500 4000 4500 50000
15
30
455555
c)
Frequency (Hz)
Figure 6-57. Transmission loss across the fan base for the x axis accelerometer for a) 18 /TSU m s= b) 30 /TSU m s= c) 42 /TSU m s= .
0 500 1000 1500 2000 2500 3000 3500 4000 4500 50000
25
50
70a)
0 500 1000 1500 2000 2500 3000 3500 4000 4500 50000
25
50
70b)
Tran
smis
sion
Los
s (d
B)
0 500 1000 1500 2000 2500 3000 3500 4000 4500 50000
25
50
70c)
Frequency (Hz)
Figure 6-58. Transmission loss across the fan base and the building floor for the x axis accelerometer for a) 18 /TSU m s= b) 30 /TSU m s= c) 42 /TSU m s= .
147
The schematic of the vibration isolator vibration measurement setup is shown in
Figure 5-15. Figure 6-59 - Figure 6-61 show the autospectra of the accelerometers
attached to the vibration isolator section. Here again the out-of-plane acceleration (x)
dominates. Figure 6-9 shows the transmission loss across the vibration isolator for
various tunnel operating speeds. Although the transmission loss at the lower frequencies
is not very high (and is even negative at some frequencies), there is significant
attenuation of vibrations above 600 Hz .
Measurements were also made to estimate the natural frequency of the turning vane
as well as its damping coefficient. The accelerometer was attached to one of the vanes in
the turning vane rack and an impact test was conducted. The time signal measured by the
accelerometer is shown in Figure 6-63. The damped natural frequency is the inverse of
the time period of the decaying signal dT and it has a value of 170.5 Hz . This is higher
than the primary blade passage frequency (147.3 Hz ) of the fan at the highest rpm.
The damping ratio, ζ was estimated using the log decrement method (Craig 1981).
For small damping ( 0.2ζ < ), the damping ratio was estimated using the equation
1 1ln2 2
aa
ζπ
⎛ ⎞= ⎜ ⎟⎝ ⎠
(6.4)
where 1a and 2a are the accelerations at two peak time instances separated by a
time period. Following the above expression, the damping coefficient for the turning
vanes has a value of 0.025ζ ≈ . From the damping coefficient and the damped natural
frequency, the un damped natural frequency was estimated to be 171 Hz .
148
100
101
102
103
104
-120
-100
-80
-60
-40
-20a)
Acce
lera
tion
(dB
re 1
g) Accelerometer 1 (x)
Accelerometer 1 (y)Accelerometer 1 (z)
100
101
102
103
104
-150
-100
-50
0b)
Frequency (Hz)
Acce
lera
tion
(dB
re 1
g) Accelerometer 2 (x)
Accelerometer 2 (y)Accelerometer 2 (z)
Figure 6-59. Autospectra of the accelerometers attached to the vibration isolator for
18 /TSU m s= .
100 101 102 103 104-120
-100
-80
-60
-40
-20a)
Acc
eler
atio
n (d
B r
e 1
g)Ac
cele
ratio
n (d
B re
1 g
)
Accelerometer 1 (x)Accelerometer 1 (y)Accelerometer 1 (z)
100 101 102 103 104-150
-100
-50
0b)
Frequency (Hz)
Accelerometer 2 (x)Accelerometer 2 (y)Accelerometer 2 (z)
Figure 6-60. Autospectra of the accelerometers attached to the vibration isolator for
30 /TSU m s= .
149
100
101
102
103
104
-100
-80
-60
-40
-20
0a)
Acc
eler
atio
n (d
B r
e 1
g) Accelerometer 1 (x)Accelerometer 1 (y)Accelerometer 1 (z)
100
101
102
103
104
-150
-100
-50
0b)
Frequency (Hz)
Acce
lera
tion
(dB
re
1 g)
Accelerometer 2 (x)Accelerometer 2 (y)Accelerometer 2 (z)
Figure 6-61. Autospectra of the accelerometers attached to the vibration isolator for
42 /TSU m s= .
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
0
10
20
30a)
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
0102030
b)
Tran
smis
sion
Los
s (d
B)
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
0102030
c)
Frequency (Hz)
Figure 6-62. Transmission loss across the vibration isolator for the x axis accelerometer for a) 18 /TSU m s= b) 30 /TSU m s= c) 42 /TSU m s= .
150
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08-60
-40
-20
0
20
40
60
Time (s)
Acce
lera
tion
(m/s
2 )
a1 a2
Td
Figure 6-63. Time response of a turning vane due to an impulsive impact.
Acoustic Liner Absorption Coefficient Estimation
200 400 600 800 1000 1200 1400 16000.5
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
1
α
Frequency (Hz)
OC 703OC PINK
Figure 6-64. Normal incidence absorption coefficient for the acoustic liner.
151
Figure 6-64 shows the absorption coefficient (α ) measured using the two
microphone method. Two commonly used samples of acoustic liner (10.1 cm diameter,
30.5 cm long) were tested. OC 703 is the Owens Corning yellow fiberglass absorber and
OC PINK is the Owens Corning PINK (Type R19) absorber. The OC 703 is more rigid
than the OC PINK and it is mainly used for applications that require more structural
rigidity. The OC PINK is mainly used for heat insulation purposes, but it also has good
acoustic properties. As seen form the figure the OC PINK has a better normal incidence
absorption characteristic than the OC 703 absorber. The absorption coefficient for the
OC PINK is higher than 0.9 for frequencies as low as 160 Hz , whereas for the low
frequency absorption coefficient is not very good for the OC 703 liner. For this reason,
the OC PINK is used as the liner material for the wind tunnel duct work.
152
CHAPTER 7 CONCLUSIONS AND FUTURE WORK
The design, fabrication, and preliminary characterization of an anechoic wind
tunnel facility at the University of Florida are presented. A previously existing and ISO
3745 validated 100 Hz anechoic chamber has been modified to incorporate an open-jet
anechoic wind tunnel facility suitable for flow-induced noise studies. The wind tunnel is
driven by a 224 kW ( 300 HP ), 369 /m s (147,000 cfm ) centrifugal fan controlled by a
variable frequency drive. Airflow enters the wind tunnel through a settling duct with a
honeycomb section and a set of four screens for the purposes of flow straightening and
turbulence reduction, respectively. An optimized, minimum length, 3-D contraction
designed using various computational methods accelerates the flow into a rectangular test
section that measures 0.74 m ( 29" ) by 1.12 m ( 44" ) by 1.83 m ( 72 '' ). The contraction
shape consists of matched polynomials and is constructed using 19 mm ( 0.75") thick
reinforced fiberglass. A novel approach was followed for the inlet contraction design
where the results of the contraction optimization study conducted using potential flow
simulations were verified using 3-D turbulent Navier-Stokes simulations. Static pressure
measurements along the sidewall, base, and corners of the contraction validate the design
procedure to generate a minimum-length contraction devoid of flow separation. The
estimated maximum velocity attainable in the test section is 76 /m s ( 250 /ft s ); thus the
maximum chord Reynolds number based on a model with a 1.5 span c= is
6Re 3 4 10c = − × . The flow leaving the test section enters a 2-D diffuser, turns a 90°
153
corner using composite, rubber-filled turning vanes, and then enters a second 2-D
diffuser. The flow leaving the second diffuser then enters the fan through a vibration-
isolated rectangular-to-round transition section. The two diffusers and the corner sections
are lined with a metal screen or perforate bounding 30 cm (12") thick bulk fiberglass to
attenuate fan noise. The fan and its silencer rest on a vibration-isolated concrete mass
base located outside the building to minimize vibrations and the resulting noise that
propagates to the chamber. All components of the wind tunnel, except the flow
conditioner and the fan, were fabricated ‘in-house.’
The facility was rigorously characterized to ensure the quality of the future
aerodynamic and acoustic measurements. The characterization experiments were
performed up to a test section speed of 43 /TSU m s= , beyond which the deflection of the
chamber bounding walls necessitates further future structural reinforcements. Flow
uniformity measurements done at 17 /TSU m s= indicate the rms flow non-uniformity at
the nozzle exit are <0.7 %. The freestream turbulence level has a value of 0.035 % at the
test section exit for a tunnel operating speed of 17 /TSU m s= . These values are
comparable to or better than the corresponding values for the benchmark Notre Dame
facility. The Notre Dame tunnel inlet contraction is also longer ( 4.3 m ) and has a larger
contraction ratio (1:20) than the UF tunnel. Shear layer measurements were made to
quantify the growth of the test section jet. The time-averaged velocity profiles in the
shear layer were found to be self similar. Pressure recovery measurements made inside
the diffuser indicates that flow separation is avoided in the diffusers.
Background noise level measurements inside the anechoic facility with an empty
test section (up to 43 /TSU m s= ) inside the anechoic facility reveal OASPL from 100
154
Hz – 20 kHz of 75.7 dB , with a peak 1/3rd octave-band level of 71.3 dB at 100 Hz that
decreases to 51.6 dB at 1 kHz . These levels increase substantially with increasing flow
speed with 2 5.6rms TSP U∝ . Nevertheless, these results are encouraging and indicate the
quality of the anechoic flow facility relative to other existing facilities. The background
noise levels obtained in the tunnel are the lowest among all existing anechoic wind
tunnels documented in the literature. Coherent power measurements made along the
diffuser wall establish the effectiveness of the acoustic lining in attenuating the fan noise.
A conditional spectral technique was implemented to remove unwanted sources from the
test section background noise spectra. The results indicate almost negligible
contamination from unwanted external noise sources, the primary offenders being the fan
motor winding noise at 4-5 kHz and the cooling fan noise from the VFD at 282 Hz .
Acceleration measurements made across the fan base and the vibration isolation section
indicate that the overwhelming majority of fan induced vibrations are isolated from the
chamber.
Future research will continue to characterize the facility at still higher flow speeds,
after structural reinforcements to the anechoic chamber walls are made. An existing (but
currently uninstalled) silencer will also be attached to the fan exhaust if fan exhaust noise
becomes problematic at higher speeds. The wire mesh that lines the inner surface of the
two diffusers was found to be susceptible to wall waviness and will be replaced with a
lining of Nomex acoustic absorber to eliminate the waviness and corresponding boundary
layer scrubbing noise. Additional flow non-uniformity, turbulence, and background
noise measurements will be performed after these enhancements are completed. Laser
Doppler Velocimetry (LDV) will be used to quantify the boundary layer profile at the
155
exit plane of the inlet contraction. Furthermore, measurements of the fan noise
attenuation will be compared with an appropriate theoretical and computational models
that predict noise attenuation for a developing shear flow through an acoustically lined,
diverging duct.
The final product of this research is a unique, university-scale, CFD validated,
rigorously characterized anechoic wind tunnel facility that facilitates simultaneous
research in the fields of aerodynamics and aeroacoustics. The facility was built within a
very reasonable budget (<$400,000), and the acoustic/aerodynamic flow properties are
comparable to any existing anechoic facility in the world. The facility is expected to be
particularly relevant in research targeted at the understanding and, ultimately, the
reduction of aerodynamic noise due to airframe and wind turbine noise.
156
REFERENCES
Aerospace Industries Association , www.aia-aerospace.org, August, 2004.
“Acoustics ─ Determination of Sound Power Levels of Noise Sources ─ Precision Methods for Anechoic and Semi-anechoic Rooms,” International Standard ISO 3745, 1977.
Allen, R. M., and Reed, D. H., “Development of the Boeing Low Speed Aeroacoustic Facility (LSAF),” DLR/AIAA 92-02-030, DLR/AIAA 14th Aeroacoustics Conference, Aachen, Germany, May 1992.
Amiet. R, K., “Refraction of Sound by a Shear Layer.” Journal of Sound and Vibration, Vol. 58, pp. 467-482, 1978.
ASTM-E1050-98, “Impedance and Absorption of Acoustic Materials Using a Tube, Two Microphones and Digital Frequency Analysis Systems,” ASTM International, 1998.
Barlow, J. B., Rae, W. H. Jr., and Pope, A., Low-Speed Wind Tunnel Testing, 3rd Edn, John Wiley and Sons, New York, 1999.
Bendat, J. S., and Piersol, A. G., Random Data Analysis and Measurement Procedures, John Wiley and Sons Inc, New York, 2000.
Blackstock, D. T., Fundamentals of Physical Acoustics, John Wiley and Sons Inc, New York, 2000.
Blake, W. K., and Gershfeld, J. L., “The Aeroacoustics of Trailing Edges,” in Frontiers in Experimental Fluid Mechanics, Chap. 10 (Gad-el-Hak, M., ed.), Springer-Verlag, 1988.
Brooks, T. F., and Hodgson, T. H., “Trailing Edge Noise Prediction From Measured Surface Pressures,” Journal of Sound and Vibration, Vol. 78, No.1, pp 69-117, 1981.
Brooks, T. F., and Humphreys, W. M., “Flap-Edge Aeroacoustic Measurements and Predictions,” Journal of Sound and Vibration, Vol. 261, pp 31-74, 2003.
157
Brooks, T. F., and Marcolini, M. A., “Airfoil Trailing Edge Flow Measurements and Comparison with Theory Incorporating Open Wind Tunnel Corrections,” AIAA 84-2266, AIAA/NASA 9th Aeroacoustics Conference, Williamsburg, Virginia, October 1984.
Brooks, T. F., and Marcolini, M. A., “Airfoil Trailing-Edge Flow Measurements,” AIAA Journal, Vol. 24, No. 8, 1245-1251, August 1986.
Cogotti, A., “Aeroacoustics Development at Pininfarina,” SAE Paper 970402, Warrendale, PA, 1997.
Collar, A. R., “Some Experiments with Cascades of Airfoils,” ARC R&M 1768, UK, 1937.
Craig, R. R., Structural Mechanics-An Introduction to Computer Methods, John Wiley and Sons Inc, New York, 1981.
Crighton, D. G., “Airframe Noise,” in Aeroacoustics of Flight Vehicles: Theory and Practice, Volume 1: Noise Sources, ed. By H. H. Hubbard, Acoustic Society of America, Woodbury, NY, 1995.
Crighton, D. G., and Leppington, F. G., “On the Scattering of Aerodynamic Noise,” Journal of Fluid Mechanics, Vol. 46, pp. 577-597, 1971.
Davy, R., and Remy, H., “Airframe Noise Characteristics of a 1/11 Scale Airbus Model,” AIAA-98-2335, 1998.
Derbunivich, G. I., Zemskaya A. S., Repik, E. U., and Sosedko, Y. P., “Effect of Flow Contraction on the Level of Turbulence,” Izvestiya Akademii Nauk SSSr, Mekkhanika Zhidkosti I Gaza, No. 2, pp. 146-152, March-April 1987.
Duell, E. G., “Recent Aerodynamic Improvements at the Lockheed Martin Low Speed Wind Tunnel,” SAE Paper 960903, International Congress & Exposition, Detroit, Michigan, 1996.
Duell, E., Walter, J., Arnette, S., and Yen, J., “Recent Advances in Large-Scale Aeroacoustic Wind Tunnels,” 8th AIAA/CEAS Aeroacoustics Conference and Exhibit, Breckenridge, Colorado, June 2002.
Duell, E., Yen, J., Walter, J., and Arnette, S., “Boundary Layer noise in Aeroacoustic Wind Tunnels,” 42nd AIAA Aerospace Sciences Conference, Reno, Nevada, January 2004.
Federal Aviation Administration, http://www.aee.faa.gov/noise/, August, 2004.
Ffowcs Williams, J. E., and Hall, L. H., “Aerodynamic Sound Generation by Turbulent Flow in The Vicinity of a Scattering Half Plane,” Journal of Fluid Mechanics, Vol. 40, pp. 657-670, 1970.
158
Fox, R. W., and Mcdonald, A. T., Introduction to Fluid Mechanics, John Wiley & Sons, 5th edn, New York, 1998.
Gelder, T. F., Moore, R. D., Sanz, J. M., and Mcfarland, E. R., “Wind Tunnel Turning Vanes of Modern Design,” AIAA Paper-86-0044, Reno, Nevada, January 1986.
Goldin, D. S., “Turning Goals to Reality,” World Aviation Congress, Los Angeles, California, October 1997.
Golub, R. A., Rawls, J. W., and Russel, J. W., “Evaluation of the Advanced Subsonic Technology Program Noise Reduction Benefits,” NASA/TM-2004-212144, August, 2004.
Haalan, S. E., “Simple and Explicit Formulas for the Friction Factor in Turbulent Pipe Flow,” Journal of Fluids Engineering, Vol. 105, pp. 89-90, 1983.
Ho, C. M., and Huerre, P., “Perturbed Free Shear Layers,” Annual Review of Fluid Mechanics, Vol. 16, pp 365-424, 1984.
Howe, M. S., “A Review of the Theory of Trailing Edge Noise,” Journal of Sound and Vibration, Vol. 61, No. 3, pp. 437-465, 1978.
Hubbard, H. H., and Manning, J. C., “Aeroacoustic Research Facilities at NASA Langley Research Center,” NASA TM 84585, 1983.
Hutcheson, F. V., and Brooks, T. F., “Measurement of Trailing Edge Noise Using Directional Array and Coherent Output Power Methods,” AIAA Paper 2002-2472, June 2002.
Ingard, K. U., Notes on Sound Absorbers, Massachusetts Institute of Technology, Cambridge, Massachusetts, 1999.
International Standard ISO 3745, “Acoustics ─ Determination of Sound Power Levels of Noise Sources ─ Precision Methods for Anechoic and semi-anechoic Rooms,” Geneva, 1977.
Jansson, D., Mathew, J., Hubner, J. P., Sheplak, M., and Cattafesta, L. N. III., “Design and Validation of an Aeroacoustic Anechoic Test Facility,” AIAA Paper 2002-2466, June 2002.
Jorgensen, F. E., “How to Measure Turbulence with Hot-Wire Anemometers-a practical guide,” Dantec Dynamics, Denmark, 2005.
Khorrami, M. R., Berkman, M. E., and Choudhari, M., “Unsteady Flow Computations of a Slat with a Blunt Trailing Edge,” AIAA Journal, Vol. 38, No. 11, November 2000.
159
King, R. D., “Receptivity and Growth of Two- and Three-Dimensional Disturbances in a Blasius Boundary Layer,” Phd Thesis, MIT, Cambridge, MA, 2000.
Kunstner, R., Potthoff, J., and Essers U., “The Aero-Acoustic Wind Tunnel of Stuttgart University,” SAE Paper 950625, 1995.
Kunze, C., Lynch, D. A. III., and Mueller, T. J., “Effect of Trailing Edge Geometry on Vortex Shedding and Acoustic Radiation,” AIAA Paper 2002-2435, June 2002.
Lockard, D. P., and Lilley, G. M., “The Airframe Noise Reduction Challenge,” NASA/TM-2004-213013, 2004.
Macaraeg, M. G., “Fundamental Investigations of Airframe Noise,” AIAA Paper 98-2224, 1998.
Mehta, R. D., “The Aerodynamic Design of Blower Tunnels with Wide-Angle Diffusers,” Prog. Aerospace Sci, Vol. 18, pp. 59-120, 1977.
Mehta, R. D., and Bradshaw, P., “Design Rules for small low speed wind tunnels,” Aeronautical Journal, Paper No. 718, November 1979.
Morel, T., “Comprehensive Design of Axisymmetric Wind Tunnel Contractions,” Journal of Fluids Engineering, Transactions of the ASME, Series I, Vol. 97, pp. 225 – 233, June 1975.
Mueller, T. J, Scharpf, D. F., Batill, S. M., Strebinger, R. B., Sullivan, C. J., and Subramanian, S., “The Design of a Subsonic Low-Noise Low-Turbulence Wind Tunnel for Acoustic Measurements,” AIAA Paper 92-3883, July 1992.
Mueller, T. J., Aeroacoustic Measurements, Springer-Verlag, New York, 2002.
Oerlemans, S., “Aeroacoustic Tests of Candidate Wind Turbine Airfoils,” NLR Memorandum, AK-2002, 2002.
Oerlemans, S., Schepers, J. G., Guidati, G., and Wagner, S., “Experimental Demonstration of Wind Turbine Noise Reduction Through Optimized Airfoil Shape and Trailing Edge Serrations,” European Wind Energy Conference and Exhibition, Copenhagen, July 2001.
Oerlemans, S., and Sitjsma, P., “Effects of Wind Tunnel Side-Plates on Airframe Noise Measurements with Phased Arrays,” AIAA Paper 2000-1938, 2000.
Ogata, N., Iida, N., and Fujii, Y., “Nissan’s Low-Noise Full-Scale wind Tunnel,” SAE Paper 870250, 1988.
Panton, R. L., Incompressible Flow, 2nd edition, John Wiley & Sons, New York, 1996.
160
Piet, J., and Elias, G., “Airframe Noise Source Localization Using a Microphone Array,” AIAA-97-1643-CP, 1997.
Pope, A., and Harper, J. J., Low Speed Wind Tunnel Testing, John Wiley and Sons, New York, 1966.
Reshotko, E., Saric, W, S., and Nagib, H, M., “Flow Quality Issues for Large Wind Tunnels,” AIAA Paper 97-0225, 35th AIAA Aerospace Sciences Meeting & Exhibit, Reno, Nevada, January 1997.
Runstadler, P. W., Dolan, F. X., and Dean, R. C., Diffuser Data Book, Creare Inc, Hanover, 1975.
Salter, C., “Experiments on Thin Turning Vanes,” ARC R&M 2469, October 1946.
Schubauer, G. B., Spangenberg, W. G., and Klebanoff, P. S., “Aerodynamic Characteristics of Damping Screens,” NACA TN 2001, 1950.
Schultz, T. J., Acoustical Uses for Perforated Materials, Industrial Perforators Association, Boca Raton, Florida,1986.
Shields, F. D., and Bass, H. E., “Atmospheric Absorption of High Frequency Noise and Application to Fractional-Octave Bands,” NASA CR-2760, June 1977.
Smith, M. J. T., Aircraft Noise, Cambridge University Press, Cambridge, 1989.
Soderman, P. T., and Hoglund, L. E., “Wind-Tunnel Fan Noise Reduction Including Effects of turning Vanes on Noise Propagation,” AIAA Paper 79-0642, 5th AIAA Aeroacoustics Conference, Seattle, Washington, March 1979.
Stratford, B. S., “The Prediction of Separation of the Turbulent Boundary Layer,” Journal of Fluid Mechanics, Vol. 5, pp. 1-16, 1958.
Su, Y., “Flow Analysis and Design of Three-Dimensional Wind Tunnel Contractions,” AIAA Journal, Vol. 29, No. 11, 1991, pp. 1912 – 1920.
Sydhoff, F., “Verification of an Anechoic Chamber and related Experimental Techniques,” Master’s Thesis, Royal Institute of Technology, Stockholm, Sweden, 2003.
Tam, C. K. W., Golebiowski, M., and Seiner, J. M., “On the Two Components of Turbulent Mixing Noise from Supersonic Jets,” AIAA Paper 96-1716, May 1996.
Tam, C. K. W., and Zaman, K. B. M. Q., “Subsonic Jet Noise from Nonaxisymmetric and Tabbed Nozzles,” AIAA Journal, Vol. 38, No. 4, 2000, pp. 592 – 599.
161
Van Ditshuizen, J. C. A., Courage, G. D., Ross, R., and Schultz, K. J., “Acoustic Capabilities of the Greman-Dutch Wind Tunnel DNW,” AIAA-83-0146, AIAA 21st Aerospace Sciences Meeting, Reno, Nevada, January 1983.
Viswanathan, K., “Quality of Jet Noise Data: Issues, Implications, and Needs,” AIAA Paper 2002-0365, Jan. 2002.
Walter, J., Duell, E., Martindale, B., Arnette, S., and Geierman, R., “The DaimlerChrysler Full-Scale Aeroacoustic Wind Tunnel,” SAE Paper 2003-01-0426, 2003.
Walter, J., Duell, E., Martindale, B., Arnette, S., Nagle, P. A., Gulker, W., Wallis, S., and Williams J., “The Driveability Test Facility Wind Tunnel No.8,” SAE Paper 2002-01-0252, 2001.
Watmuff, J. H., “Detrimental Effects of Almost Immeasurably Small Freestream Nonuniformities Generated by Wind-Tunnel Screen,” AIAA Journal, Vol. 36, No. 3, March 1998.
Wickern, G., and Lindener, N., “The Audi Aeroacoustic Wind Tunnel: Final Design and First Operational Experience,” SAE Paper 2000-01-0868, 2000.
Wieghardt, L. G., “On the Resistance of Screens,” Aeronautical Quarterly, 4,186, 1953.
Willshire, B., “Progress Toward Quieter Aircraft,” Airport Noise Symposium, Berkeley, CA, February 2001.
Wind Tunnel Design, http://vonkarman.stanford.edu/tsd/pbstuff/tunnel/contraction.html, December, 2005
Yu, J. C., and Joshi, M. C, “On Sound Radiation from the Trailing Edge of an Isolated Airfoil in a Uniform Flow,” AIAA Paper 79-0603, 5th AIAA Aeroacoustics Conference, Seattle, Washington, March 1979.
162
APPENDIX A A SCHEMATICS OF THE WIND TUNNEL
Figure A-1. Plan view of the wind tunnel.
163
Equation Section 1
Figure A-2. Cross-section view of the wind tunnel.
Figure A-3. North elevation of the wind tunnel.
164
APPENDIX B B DERIVATION OF THE INLET SHAPE POLYNOMIAL
He/2
Match Point
xmx
Hi/2
L
y
Symmetry Line
Figure B-1. Schematic of the inlet shape polynomial.
The schematic of the inlet shape polynomial, also referred to in Chapter 3 is shown
in Figure B-1. The height at the entrance of the contraction is iH and the exit height of
the contraction is eH . The total length of the contraction is L and the contraction ratio is
denoted by CR . The shape consists of a 3rd-order polynomial matched with an 8th-order
polynomial at a distance of mx x= from the contraction entrance. The general equations
for a 3rd order polynomial and an 8th order polynomial are as follows, Equation Section 2
3 2
1 31 21 11 018 7 6 5 4 3 2
2 82 72 62 52 42 32 22 12 02
.
.
y a x a x a x a
y a x a x a x a x a x a x a x a x a
= + + +
= + + + + + + + + (B.1)
165
Taking the 8th-order origin at the exit plane of the contraction, and changing the equations
to non-dimensional form by normalizing by the exit height eH , gives
3 21 31 21 11 01
8 7 6 5
2 82 72 62 52
4 3 2
42 32 22 12 02
.
.
e e e e
e e e e
a a a a
L L L La a a aH H H H
L L L La a a a aH H H H
η ξ ξ ξ
η ξ ξ ξ ξ
ξ ξ ξ ξ
= + + +
⎛ ⎞ ⎛ ⎞ ⎛ ⎞ ⎛ ⎞= − + − + − + − +⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠
⎛ ⎞ ⎛ ⎞ ⎛ ⎞ ⎛ ⎞− + − + − + − +⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠
(B.2)
where
.
.
e
e
yHx
H
η
ξ
=
= (B.3)
The system has 13 unknown constants. The position is set at each end with non-
dimensional parameters based on the exit height eH and the contraction ratio CR , the
first two derivatives (slope and curvature) of the cubic are set to zero at the entrance, and
the first 6 derivatives of the 8th-order are set to zero at the exit.
166
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
1
1
21
2
2
2
22
2
32
3
42
4
52
5
62
6
0 .
0 0.
0 0.
1.
0.
0.
0.
0.
0.
0.
CRdddd
Ld Ldd Ldd Ldd Ldd Ldd Ld
ηηξηξ
ηηξηξηξηξηξηξ
=
=
=
=
=
=
=
=
=
=
(B.4)
This eliminates ten of the constants, leaving
31 31
8 7
2 82 72
.
1.e e
a CR
L La aH H
η ξ
η ξ ξ
= +
⎛ ⎞ ⎛ ⎞= − + − +⎜ ⎟ ⎜ ⎟
⎝ ⎠ ⎝ ⎠
(B.5)
To solve for the final three constants, the position, slope, and curvature of the two
curves are set equal to each other at the match point, mx . We can define a non-
dimensional match point X , as
.mxXL
= (B.6)
The equations reduce to
3 8 7
31 82 72 1e e e e e
L L L L La X CR a X a XH H H H H
⎛ ⎞ ⎛ ⎞ ⎛ ⎞+ = − + − +⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎝ ⎠ ⎝ ⎠ ⎝ ⎠. (B.7)
2 7 6
31 82 723 8 7e e e e e
L L L L La X a X a XH H H H H
⎛ ⎞ ⎛ ⎞ ⎛ ⎞= − − − −⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎝ ⎠ ⎝ ⎠ ⎝ ⎠. (B.8)
167
6 5
31 82 726 56 42e e e e e
L L L L La X a X a XH H H H H
⎛ ⎞ ⎛ ⎞= − + −⎜ ⎟ ⎜ ⎟
⎝ ⎠ ⎝ ⎠. (B.9)
Rearranging the equations in matrix form gives
( ) ( )
( ) ( )
( ) ( )
3 8 78 73
2 7 6 317 62
82
726 56 5
1 1
13 8 1 7 1 0 .
0
6 56 1 42 1
e e e
e e e
e e e
L L LX X XH H H
a CRL L LX X X a
H H Ha
L L LX X XH H H
⎡ ⎤⎛ ⎞ ⎛ ⎞ ⎛ ⎞⎢ ⎥− − − −⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎢ ⎥⎝ ⎠ ⎝ ⎠ ⎝ ⎠
⎧ ⎫⎢ ⎥ −⎧ ⎫⎛ ⎞ ⎛ ⎞ ⎛ ⎞ ⎪ ⎪⎢ ⎥ ⎪ ⎪− − =⎨ ⎬ ⎨ ⎬⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎢ ⎥⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎪ ⎪ ⎪ ⎪⎢ ⎥ ⎩ ⎭ ⎩ ⎭⎢ ⎥⎛ ⎞ ⎛ ⎞ ⎛ ⎞⎢ ⎥− − − −⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎢ ⎥⎝ ⎠ ⎝ ⎠ ⎝ ⎠⎣ ⎦
(B.10)
From here, values for the three remaining coefficients can be obtained, allowing for
the construction of the matched 3rd-8th inlet contour. For the contraction shape we have
chosen, the match point, X has a value of 0.5, the contraction ratio, CR is 8, and the
total length of the inlet section L , is 3.05 m (10 ft), giving a length to height ratio at the
contraction exit of / 4.14eL H = . The plots of the contraction shape polynomial in the x-
y plane and the x-z plane are given in Figure B-2, and Figure B-3, respectively.
168
0 1 2 3 4 5 6 7 8 9 101
1.5
2
2.5
3
3.5
4
4.5
x (ft)
y (ft
)
Match Point
Figure B-2. Plot of contraction shape polynomial in the x-y plane.
0 1 2 3 4 5 6 7 8 9 101.5
2
2.5
3
3.5
4
4.5
x (ft)
z (ft
) Match Point
Figure B-3. Plot of contraction shape polynomial in the x-z plane.
169
APPENDIX C C INLET OPTIMIZATION STUDY
The schematic of the velocity vectors at the inlet exit plane is shown in Figure C-1.
In the figure, V is the total velocity vector and u , v , and w are the various components
in the x , y , and z directions. Ideally, we want uniform, 1-D flow in the test section,
along the x axis. Due to the 3-D nature of the inlet contraction and the subsequent
streamline curvature, both v and w components will also be present. The flow
streamline curvature also causes non-uniformity of u velocity at the inlet exit plane.
Equation Section 3Equation Section 3
INLET
x
y
z
v
w
V
u
Figure C-1. Velocity vector at the inlet exit plane.
170
To achieve good flow quality in the test section, the non-uniformities in the u
velocity and also the magnitudes of both v and w has to be minimized. To meet these
requirements, an optimization study of the inlet contraction size and shape has to be
conducted.
The four inlet design parameters used for this study are the total length L , the
contraction ratio CR , the aspect ratio at the entrance AR , and the match point of the wall
shape polynomials, X . The inlet design parameters have to be optimized in order to
achieve good flow quality in the test section. A test matrix for the optimization study is
made where each of these design parameters are individually varied (Table C-1). The
flow quality in the test section can be quantified using the following three parameters,
wθ , vθ , and u that form the cost functions for this optimization study. wθ is the angle
that the flow streamlines makes with the z axis due to the wall curvature, and is given by
the expression
1tan ,wavg
wu
θ −⎛ ⎞
= ⎜ ⎟⎜ ⎟⎝ ⎠
(C.1)
where w is the velocity in the z direction at the exit of the inlet and avgu is the average
axial velocity at the exit. Similarly, vθ is the angle made by the flow streamlines with the
y axis and is given by the formula
1tan ,vavg
vu
θ −⎛ ⎞
= ⎜ ⎟⎜ ⎟⎝ ⎠
(C.2)
where v is the y velocity at the inlet exit. The peak flow non-uniformity, u at the inlet
exit is given by
max min .avg
u uuu
⎛ ⎞−= ⎜ ⎟⎜ ⎟
⎝ ⎠ (C.3)
171
where maxu and minu are the maximum and minimum values of the x velocity at the inlet
exit plane. The constraint for this optimization study is that the flow should not separate
at any location inside the inlet contraction. The 3-D, potential flow equation is solved
numerically using Abaqus® for each entry in the test matrix. For the boundary
conditions, a uniform velocity of 9.5 m/s is specified at the entrance to the inlet
contraction, the walls of the contraction have slip but the normal velocity across the walls
is zero, and an outflow boundary condition is specified at the exit. The Abaqus®
simulation outputs the velocity at each node point of the inlet mesh. Knowledge of the
nodal velocity is used to calculate the nodal pressure coefficients. The flow field is
checked for separation by extracting the pressure coefficients along the corner and using
the Stratford’s criteria (Stratford 1959) to ensure that these values are less than the
pressure coefficient values predicted by Stratford which correspond to the limit of
separation.
The results of the inlet optimization study are given in
Table C-1. For the match point of the wall shape polynomials, values of 0.4X =
and 0.5X = give good flow quality. A square inlet ( AR =1) is preferred over a
rectangular inlet for reasons of better flow quality. Increasing the contraction ratio for a
given inlet length L , leads to flow separation; however the reduction in turbulence
intensity scales with the square root of the contraction ratio (Derbunovich et al. 1987). A
large CR also requires larger entrance dimensions which can affect the room dimension
constraints. A longer inlet provides better quality flow and it also enables the decay of
the incoming turbulence. However, a longer inlet leads to increased losses due to greater
boundary layer growth and increased cost of manufacture. From the design table the 29”
172
by 44” inlet with X =0.5, AR =1, L =10’, and CR =8 is selected. This design fits the
dimension constraints, meets the budget and also provides good flow quality with no
separation.
Table C-1. Results of Inlet optimization study. Inlet L (ft) CR
AR
X wθ (º) vθ (º) u avgu Separation
24 X 36 14 16 1.5 0.5 0.0007 0.001 0.001 249.99 No
24 X 36 14 16 1.5 0.4 0.0001 0.0001 0.0002 250.00 No
24 X 36 14 16 1.5 0.6 0.0009 0.0007 0.0013 250.00 No
24 X 36 14 16 1.5 0.3 0.002 0.0009 0.0002 250.00 No
24 X 36 14 16 1.5 0.7 0.0048 0.0064 0.0042 250.00 No
24 X 36 6 10 1 0.5 Yes
24 X 36 7 10 1 0.5 Yes
24 X 36 8 10 1 0.5 .0017 .0154 .0062 250.00 No
24 X 36 8 16 1 0.5 Yes
24 X 36 8 12 1 0.5 .024 .008 .0078 250.18 No
24 X 36 8 10 1.5 0.5 .062 .045 .0125 249.87 No
24 X 36 10 12 1 0.5 .0008 .0009 .0022 249.7 No
24 X 36 10 12 1 0.4 .0004 .0004 .001 249.95 No
33 X 48 8 10 1 0.5 Yes 33 X 48 10 10 1 0.5 .0098 .0048 .008 249.97 No
33 X 48 12 12 1 0.5 .0025 .0037 .0043 250.02 No
33 X 48 11 12 1 0.5 .004 .008 .0068 250.04 No
33 X 48 12 12 1 0.4 .0005 .0007 .0019 250.00 No
29 X 44 11 12 1 0.5 .0022 .0043 .0041 250.02 No 29 X 44 12 8 1 0.5 .0006 .0012 .002 250.08 No
29 X 44 10 8 1 0.5 .0007 .0015 .005 249.99 No
We also studied the effects of corner flow on the pressure gradients in the flow
field. The cases considered were a square corner, a filleted corner with a small radius of
curvature, and a corner with a small chamfer. It was seen that the resulting flow field
173
was not very different for each of the above cases. However in the actual design there
will be a small curvature along the corners.
.
174
APPENDIX D D DIFFUSER OPTIMIZATION STUDY
Kline’s flat diffuser design charts used for the diffuser selection assumes that the
flow entering the diffuser is uniform. However this is not the case in reality as the flow
leaving the test section is perturbed by the airfoil model and a non-uniform flow enters
the diffuser. An optimization study in Matlab® was done to minimize the included angles
of both diffuser 1 and diffuser 2, as obtained from the flat diffuser chart (Kline’s flat
diffuser charts, Runstadler et al. 1975). The plan view of the wind tunnel with the
adequate dimensions is shown in Figure D-1. The total length from the inlet entrance to
the south wall is 1tL . The length of the inlet is iL and the test section length is tsL . The
test section measures tsH by tsW leading to a collector of size 1H by 1W . We define the
collector area ratio, ARc , as Equation Section 4
1 1AR
ts ts
H WcH W
= (D.1)
Diffuser 1 has a length of 1L and it is attached to the corner via a connection sleeve
of length sL . The inlet and outlet of the corner have equal area measuring 1H by 1W and
it is 1 1cL W+ long in the North-South direction and 2 1cL W+ long in the East-West
direction. The lengths 1cL and 2cL are dictated by the chord dimension of the turning
vanes. Another connection sleeve of length sL joins the corner to diffuser 2, which is 2L
in length. The distance between the wedges and the south wall is denoted by wwL . 2dL
is the length of the section of diffuser 2 that extends outside the building. The total
175
length from the corner wall to the west wall of the building is 2tL . The gap for the
installation of the acoustic liner has a thickness of gL . We also define quantities 1SF and
2SF that are the safety factors used in the design of diffuser 1 and diffuser 2. The safety
factors multiply the area ratios ( AR ) for a given ratio of length to height ( /L H )
obtained from Kline’s flat diffusers curves (Runstadler et al. 1975), such that the
resulting design is well below the separation region.
Li Lts L1
Ls
dL2
W1
W1
Lg Lg
Lt1
Wts
LsLc1
Lc2
L2
Lt2
Anechoic Zone
WEDGES
INLET
/ 4λ
SOU
TH W
ALL
WEST WALL
E
S
W
N
CORNER
DIFFUSER 1
DIFFUSER 2
SLEEVE
Lww
Garage Door
Liner
Figure D-1. Wind tunnel flow path.
In order to setup the optimization scheme a cost function needs to be identified.
The sum of the included angles for the two diffusers is the cost function in this case and
can be expressed as
1 2θ θ θ= + (D.2)
176
where 1θ (see Figure 2-5) is the included angle for diffuser 1, and 2θ is the included
angle for diffuser 2. The cone angles 1θ and 2θ can be expressed as follows,
1 2 11
1
2 tan2
H HL
θ − ⎛ ⎞−= ⎜ ⎟
⎝ ⎠ (D.3)
1 3 22
2
2 tan2
H HL
θ − ⎛ ⎞−= ⎜ ⎟
⎝ ⎠ (D.4)
where 1H and 2H are the height at the entrance and exit of diffuser 1 and 1L is the length
of diffuser 1. Similarly 2H and 3H are the height at the entrance and exit of diffuser 2
and 2L is its length. The height at the entrance of diffuser 1 is calculated from Equation
(D.1) as
1 AR tsH c H= (D.5) The dimensions 2H and 3H are calculated using the digitized diffuser tables and the
safety factors. The variables selected for this optimization scheme are the length of the
inlet iL , area ratio of collector to test section ARc , the safety factors for diffuser 1 and
diffuser 2 1SF and 2SF , and the length of diffuser 2 extending outside the building 2dL .
The length of both diffuser 1 and diffuser 2 can be defined in terms of these variables as
follows,
1 1 1 1t i ts s c gL L L L L L W L= − − − − − − (D.6)
2 2 2 2 1t s cL L dL L L W= + − − − (D.7) where the width of the diffuser 1 can be expressed as
1 AR tsW c W= (D.8) and the length of the gap region for the acoustic liner can be expressed as
1
2ww
gL WL −⎛ ⎞= ⎜ ⎟
⎝ ⎠ (D.9)
177
The lower bound for the variable iL is 10 ft and its upper bound is set at 12 ft.
This is done to ensure that the test section falls in the anechoic zone, more than / 4λ (@
f=100 Hz) away from the wedge tip. The range for ARc is between 1 and 1.174, where
1.174 is the collector area ratio for a similar facility at Notre Dame (Mueller et al. 1991).
The safety factors are set to vary between 0.8 and 1. Finally, the limits on 2dL are 0 and
1.5 ft. A very long diffuser 2 is not preferred as this would increase the fabrication cost
and also increase the losses in the tunnel circuit. The constraints for this optimization
problem can be expressed as follows,
18 f 12 ftt L≤ ≤ (D.10) 1 11θ ≤ (D.11) 2 11θ ≤ (D.12) 10 /12 ft gL≤ (D.13)
3 1
1 1.11fanAH W
≤ ≤ (D.14)
In words, the diffuser 1 length should fall between 8ft and 12 ft, the diffuser angles
should not exceed 11 , the space for liner should be at least 10” thick and finally the ratio
of fan area ( fanA ) to the exit area of diffuser 2 should be between 1 and 1.11 (i.e., to
facilitate an almost a constant area transition from diffuser 2 to the fan inlet). The cost
function as well as some constraints is nonlinear, therefore the ‘fmincon’ function from
the Matlab® Optimization tool box is used to solve for the results. The final dimensions
of the tunnel, obtained from the results of the optimization study are given in Table D-1.
The diffusers thus designed satisfy the room dimension constraint. It can also be seen
from the table that there is almost a foot of space for the acoustic liner. Figure D-2 gives
the location of the two diffusers thus designed on the Kline’s flat diffuser curves. It can
be observed that both diffusers fall in the no stall region.
178
Table D-1. Final dimensions of the tunnel obtained from the optimization study. iL 10’
tsH 29”
tsW 44”
ARc 1.174
1H 31.4”
1W 47.7”
1L 11.74’
1θ 10.94º
sL 1”
2H 58.4”
2L 16.5’
2θ 8.12º
3H 87.4”
2dL 18”
gL 11.8”
( )3 1/fanA H W 1.11
0 5 10 15 20 25 30 35 401
1.5
2
2.5
3
3.5
4
L/Hi
AR
No appreciable stall
Line of appreciable stall
Diffuser 1
Diffuser 2
Figure D-2. Location of diffuser 1 and 2 designs on the Kline’s flat diffuser curves.
179
APPENDIX E E FAN LOSS CALCULATION
Static pressure drop in the wind tunnel circuit is caused mainly by frictional losses.
The procedure followed here is adopted from Pope and Harper (1966). The sections
below describe the procedure for the calculation of losses in the various tunnel
components that include the inlet contraction, honeycomb, screens, open jet test section,
diffuser, corner, and transition. An estimate of the total pressure loss for various tunnel
operating speeds helps in selecting a fan to drive the tunnel. The calculation example
done in the sections below are for the highest velocity (76.2 m/s) case.
Settling Duct The losses occurring in the settling duct are mainly due to friction and the loss
coefficient is given as (Pope and Harper, 1966) Equation Section 5
4
SD SD TSSD
SD SD
f L DKD D
⎛ ⎞= ⎜ ⎟
⎝ ⎠ (E.1)
where SDf is the friction factor, SDL is the total length of the settling duct, SDD is the
hydraulic diameter and TSD is the hydraulic diameter of the test section. The friction
factor for a fully turbulent flow f , is given by Haaland’s (1983) formula.
1.11
101 6.9 /1.8log .
Re 3.7D
k Df
⎡ ⎤⎛ ⎞= − +⎢ ⎥⎜ ⎟⎝ ⎠⎢ ⎥⎣ ⎦
(E.2)
where ReD is the Reynolds number based on the hydraulic diameter and k is the surface
roughness. In all our calculations we have assumed a conservative value of
0.0016 k m= (corresponding to a fairly rough surface), as the surface roughness of the
180
individual components cannot be known beforehand. Usually settling chamber losses are
a very small fraction of the total loss.
The velocity at the entrance to the test section is 9.5 m/s (for the max tunnel speed
of 76.2 m/s ) and the hydraulic diameter is 2.6 m giving a Reynolds number of 1.6.106.
Assuming a value for 0.0016 k m= and substituting into Equation (E.2) we obtain
0.018SDf = . The length of the settling duct is 1.68 m and the test section hydraulic
diameter is 0.89 m. Substituting these values in Equation (E.1) yields a value of
. 41.6710SDK −= .
Screen The losses across a flow screen are given by Weighardt (1953) and can be
expressed as
( ) 41/3
5/3
16.5 screen screen TS
screeninlet
U d DKD
ββ υ
−− ⎛ ⎞⎛ ⎞= ⎜ ⎟⎜ ⎟⎝ ⎠ ⎝ ⎠
per screen, (E.3)
where β is the fraction of open area, screenU and screend are the velocity normal to the
screen and screen diameter respectively, and inletD is the hydraulic diameter at the inlet
entrance which is same as that for the settling chamber. This expression is valid for
60 600screen screenU dβυ
< < and 1.2 12 /screenU m s< < . The velocity encountering the screen,
screenU is given by the formula
/screen TSU U CR= (E.4) where CR is the contraction ratio of the inlet and TSU is the velocity in the test section.
We have used four screens in the current design with open area ratios of 67%, 62%, 62%,
and 60% and no of mesh per inch of 24,32,46,56 respectively as given in Watmuff
(1998). The coefficient of kinematic viscosity, υ has a value of 1.55.10-5 for air at 25° C.
181
The diameters of the four screens are 192 µm, 169 µm, 117 µm, and 102 µm. The total
value of K is thus 0.094. The value ( ) /screen screenU d βυ for each screen is 176, 167, 116,
and 104, respectively. Also the velocity through each screen is 9.5 m/s which satisfy the
other constraint on the screen velocity.
Honeycomb The loss across the honeycomb section is given by (Pope and Harper, 1966).
4
0.2 .TShc
inlet
DKD
⎛ ⎞= ⎜ ⎟
⎝ ⎠ (E.5)
This expression is valid for hexagonal honeycomb cells of equal area and with a length to
diameter ratio of at least 6. It was also found that the flow through each honeycomb cell
was turbulent based on the Reynolds number of the flow. For a turbulent flow the
velocity profile is fuller and therefore the losses due to entrance and exit effects are
neglected. The evaluation of Equation (E.5) yields a value of 0.0029 for hcK .
Contraction The losses occurring in the contraction are predominantly due to frictional effects.
The inlet loss coefficient can be expressed as (Wattendorf, 1938)
0.32 avg inlet
inletTS
f LK
D= (E.6)
where avgf is the average value of the friction factors at the entrance and exit of the inlet,
and inletL is the total inlet length. We have assumed that flow through the contraction is
similar to flow through rough pipes. The friction factor at the entrance to the contraction
is calculated using Equation (E.2), where ReD is the Reynolds number based on the
entrance hydraulic diameter and the incoming velocity which is same as screenU . The
value of ReD for the maximum tunnel operating speed of 76.2 m/s is 1.6.106, and for a
182
minimum tunnel operating speed of say one tenth of the max speed, ReD has a value of
1.6.105. Thus the flow through the contraction is always turbulent. The friction factor at
the contraction exit is calculated similarly using Equation (E.2), using the Reynolds
number based on the test section hydraulic diameter and the test section velocity, TSU .
The contraction length is 3.05 m. The velocity at the contraction exit and the
Reynolds number based on the hydraulic diameter is 76.2 m/s and 4.4.106, respectively.
The friction factor at the contraction exit is 0.023, and therefore the average friction
factor for the contraction is 0.02. Substituting these values into Equation (E.6) yields a
value of 0.022 for inletK .
Test Section Frictional losses predominate in the open jet test section due to the jet shear layer,
and the loss coefficient is given by (Poe and Harper 1966)
TS TSTS
TS
f LKD
= (E.7)
where TSL and TSD are the length and hydraulic diameter of the test section respectively.
The friction factor TSf has a value of 0.08 for a cylindrical open jet test section (Pope and
Harper 1966). The length of the test section is 1.83 m and therefore the loss factor is
0.16.
Diffuser Both frictional and flow expansion losses occur in the diffuser. The loss coefficient
is given by (Pope and Harper, 1966)
( ) ( )
4 4
4 40 .6 tan / 2 18 tan / 2
d i TSdiffuser
e i
f D DKD D
αα
⎛ ⎞ ⎛ ⎞= + −⎜ ⎟ ⎜ ⎟
⎝ ⎠⎝ ⎠ (E.8)
183
where df is the effective diffuser friction coefficient which is the average of the friction
factor at the entrance and exit of the diffuser, iD and eD are the hydraulic diameters at
the diffuser inlet and exit respectively and α is the included angle between the walls.
The friction factor at the diffuser exit is calculated similarly using the Reynolds number
based on the diffuser exit hydraulic diameter, eD and the diffuser exit velocity via
Equation (E.2).
The inlet and exit hydraulic diameters for diffuser 1 are 0.96 m and 1.33 m,
respectively. The Reynolds number at the entrance and exit is therefore 4.03.106 and
3.53.106 respectively. The friction factors calculated via Equation (E.2) are 0.023 and
0.02, giving a value of 0.0215 for df . Substituting these values into Equation (E.8) leads
to a value of 0.069 for the loss coefficient. The second diffuser has an inlet and exit
hydraulic diameter of 1.33 m and 1.57 m respectively. The Reynolds number at the inlet
and exit of diffuser 2 are therefore 3.53.106 and 2.77.106 respectively. The friction factors
at the entrance and exit of diffuser 2 are 0.021 and 0.019, respectively, and the average
value of the friction factor is therefore 0.02. Substituting the average friction factor and
the diffuser dimensions into Equation (E.8), we obtain a value of 0.008 for the diffuser 2
loss coefficient.
Corner Corner losses occur due to friction as well as flow rotation. The loss coefficient is
given by (Pope and Harper, 1966)
( )
4
2.5810
4.550.1log
TSeffective
Cec
DKDR
⎛ ⎞⎛ ⎞= +⎜ ⎟⎜ ⎟⎜ ⎟⎝ ⎠⎝ ⎠
(E.9)
184
where ecR is the Reynolds number based on the vane chord and CD is the local hydraulic
diameter of the corner, which in this case is same as the hydraulic diameter at the exit of
diffuser 1. The vane chord has a value of 0.16 m and the Reynolds number based on
vane chord is therefore 0.35.106. Substituting these values into Equation (E.9) yields a
value of 0.03 for the corner loss coefficient.
Transition A direct formula for the loss in a rectangular-round section is not given in
literature. Since in our case the flow through the transition section is similar to the flow
through a diffuser the formula for the diffuser loss, Equation (E.9), is used here. In this
case df is the average of the friction coefficients at the entrance and exit of the transition
section, and iD and eD are the hydraulic diameter at the transition inlet and fan exit
diameter, respectively. Since an expansion angle α is not defined for a rectangular to
round transition, we have calculated the expansion angle by the approximate formula
-1 -=2 tan2e i
t
D DL
α⎛ ⎞⎜ ⎟⎝ ⎠
(E.10)
where tL is the length of the transition section.
The length of the transition section is 1.57 m and the inlet and exit diameters are
1.57 m, and 1.95 m, respectively. This gives a divergence angle of 13.9α °= . The
Reynolds numbers at the inlet and exit of the transition section are 2.4.106 and 2.6.106
respectively. The average friction factor is 0.018 and hence the loss factor is 0.0056.
185
APPENDIX F F EFFECT OF LEAKAGE ON WALL PRESSURE
``
``
Patm
Patm
PTS Inlet
Orifice
Entrained flow
qleak
qinlet
Pwall
1
2
3 4qfan
Figure F-1. Schematic of the chamber.
The schematic of the wind tunnel test section is shown in Figure F-1. The volume
flow rate provided by the fan is denoted by fanq . If the anechoic chamber were
completely sealed, the entire flow would come through the inlet. This would lead to a
large static pressure drop inside the chamber and a large pressure force would compress
the walls of the chamber. To prevent this, the walls have to be made ‘porous’. The
porosity of the walls reduces the pressure differential acting on the walls by letting in
ambient air into the chamber. This idea is utilized in the Langley Quiet Flow Facility
(QFF), where a set of air vents on the chamber wall helps stabilize the pressure inside
186
(Hubbard and Manning 1983). The aim of this analysis is to minimize the loading on the
chamber walls due to the reduction of static pressure inside the chamber. This can be
attained by entraining external air into the main jet flow through openings on the wall
panels. The procedure for obtaining an approximate analytical expression for the wall
pressure loading is given below. Equation Section 6
The pressure outside the chamber is ambient and is denoted by atmP . The static
pressure drops to a value of wallP just inside the chamber and further to tsP inside the test
section. If the chamber were sealed completely the pressure everywhere inside the
chamber would be tsP . This is due to the fact that the main flow has nowhere to entrain
flow from resulting in a flow equilibration inside the chamber. The orifices drilled on the
walls as shown in Figure F-1 denote the leaks in our chamber. The flow rate provided by
the fan has two different components, one coming through the inlet, inletq and the other
coming through the orifices, leakq . The conservation of mass for steady, incompressible
flow inside the chamber thus takes the following form
.fan inlet leakq q q= + (F.1) An electrical equivalent for the chamber flow can be used to better understand the
effect of leakage on the differential wall pressure. Since the volume flow rate provided
by the fan is split into two components in the test section as given by Equation (F.1), the
electrical analog becomes a current divider circuit as shown in Figure F-2. In the circuit,
pressures replace the potentials and the volume flow rates replace currents.
187
fanqtsq
leakq
inR hcRscR
leakR
tsPatmP
Figure F-2. Equivalent electric circuit representation of the chamber flow.
The dynamic pressure in the test section is denoted by 21/ 2 tsUρ whereas that at the
orifice is denoted by 21/ 2 wallUρ . The velocity in the test section is tsU and that at the
wall is wallU . The resistances across the inlet, honeycomb, screen, and the orifice are
denoted by inR , hcR , scR , and leakR , respectively. The total resistance across the inlet
section is given by
.i in hc scR R R R= + + (F.2) Consider four points marked 1-4 in Figure F-1. Applying the energy equation
between points 1 and 2 yields
21 ,
2atm wall wall leak leakP P U q Rρ= + + (F.3)
Assuming ideal flow inside the chamber and applying the Bernoulli’s equation between
points 2 and 3 gives
2 21 1 ,
2 2ts ts wall wallP U P Uρ ρ+ = + (F.4)
Finally the application of energy equation between the points 3 and 4 leads to
( )21 .
2atm ts ts inlet hc sc inP P U q R R Rρ= + + + + (F.5)
The same results as above can be obtained by applying the loop laws around the
circuit. Combining Equations (F.2), (F.3), and (F.4), we obtain
188
,leak leak inlet iR q q R= (F.6)
( ) ,leak leak fan leak iR q q q R= − (F.7)
( ) .leak i leak fan iR R q q R+ = (F.8) Substituting for inletq from Equation (F.1) in Equation (F.8), we obtain
.fan ileak
leak i
q Rq
R R=
+ (F.9)
Define a non-dimensional resistance ratio R as
,leak
i
RRR
= (F.10)
Equation (F.9) can now be expressed as
.1
fanleak
R=
+ (F.11)
The velocity at the wall can be expressed as
,leakwall
leak
qUA
= (F.12)
Substituting Equation (F.12) in Equation (F.3) we obtain
2
2
1 .2
leakatm wall leak leak
leak
qP P P q RA
ρΔ = − = + (F.13)
Define a non-dimensional area ratio A as
,leak
ts
AAA
= (F.14)
Dividing Equation (F.13) by a dynamic pressure equivalent ( )21/ 2 /fan tsq Aρ we obtain
( ) ( )
2
2
2 2
12 ,
1/ 2 / 1/ 2 /
leakleak leak
leak
fan ts fan ts
q q RAP
q A q A
ρ
ρ ρ
+Δ
= (F.15)
Substituting for area ratio A and resistance ratio R in Equation (F.15) and simplifying,
we obtain the expression for the non-dimensional wall pressure loading coefficient.
( )
2
2 2
1 1 .1 11/ 2 /wallp
fan ts
P RCR A Rq Aρ
⎛ ⎞Δ ⎛ ⎞= = + Π⎜ ⎟ ⎜ ⎟+ +⎝ ⎠ ⎝ ⎠ (F.16)
189
where the dimensionless parameter Π is given by the expression
22 .ts i
fan
A Rqρ
Π = (F.17)
The pressure drop across the various inlet components including the settling duct, honey
combs, screens and the contraction was estimated from the fan loss calculations
(Appendix E). The volume flow rate through each component was calculated from the
continuity equation and the resistance across each component was calculated as the ratio
of pressure drop to the volume flow rate. The resistances of each inlet components were
summed up to estimate the total resistance of the inlet section. For the current inlet
design the total flow resistance along the inlet section was calculated to be 56.35 . /N s m .
The Π parameter thus has a value of 0.1013 .
It can be seen that the pressure differential acting on the walls is a function of the
leakage resistance ratio R , leakage area ratio A , in addition to the Π parameter. A plot
of the leakage flow rate ratio (Equation (F.11)) variation with the leakage resistance ratio
R is given in Figure F-3. It can be observed that in order to achieve low values of
leakage ( / 10%leak fanq q < ) R has to be in the order of 10. Figure F-4 shows the variation
of the wall pressure differential, PΔ as a function of the leakage area ratio A for various
values of R . It can be observed for the plot that increasing A (i.e., the area of the leaks)
reduces the pressure differential and for higher values of A the pressure differential is
almost constant. Decreasing R also reduces the pressure differential. The total force
acting on a wall panel (120” by 48”) was calculated from the wall pressure distribution
and the results are plotted in Figure F-5. The trend is similar to that of the wall pressure
distribution. For a leakage resistance ratio value of 15 ( 595.25 /leakR Ns m= ) the leakage
190
flow rate ratio is 6.3%, which is reasonable. The existing silencers of the anechoic
chamber have a total area of 21.4 m ( 1.67A = ). Assuming the leakage occurs only
through the exhaust silencers, the value of the wall force for 15R = , is 350 lbs. This wall
force is not large enough to cause an implosion of the chamber while the wind tunnel is
in operation at full speed.
0 2 4 6 8 10 12 14 160
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Rleak/Rinlet
q leak
/qfa
n
Figure F-3. Variation of leakage ratio with the leakage flow resistance.
191
1 2 3 4 5 6 7 8 9 10350
400
450
500
Aleak/Ats
P atm
-Pw
all (N
/m2 )
R=5R=10R=15R=20
Figure F-4. Variation of the wall pressure differential with leakage area ratio for various
leakage resistance ratios.
0 2 4 6 8 10 12 14300
350
400
450
500
550
600
Aleak/Ats
Wal
l For
ce (l
bs)
R=5R=10R=15R=20
Figure F-5. Variation of the wall force with leakage area ratio for various leakage
resistance ratios.
192
APPENDIX G G RESULTS OF FREE FIELD CHARACTERIZATION
The results of the free field characterization of the anechoic chamber are
summarized in this section. The results are adapted from Sydhoff (2003). The difference
between the measured SPL decay and theoretical free field decay as a function of
distance from the center of the sound source for various 1/3rd octave bands is shown.
Equation Section 7
193
0 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
2 6300 Hz
0 2
-2
0
28000 Hz
0 2 -2
0
210000 Hz
0 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
2 16000 Hz
Array Length [m]0 2
-2
0
220000 Hz
Array Length [m]
Direction Northeast Up
0 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
2 6300 Hz
0 2
-2
0
28000 Hz
0 2 -2
0
210000 Hz
0 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
2 16000 Hz
Array Length [m]0 2
-2
0
220000 Hz
Array Length [m]
Direction Northeast Mid
0 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
2 6300 Hz
0 2
-2
0
28000 Hz
0 2 -2
0
210000 Hz
0 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
2 16000 Hz
Array Length [m]0 2
-2
0
220000 Hz
Array Length [m]
Direction Northeast Down
Figure G-6. Deviation of pressure measurements from free field from chamber center in
the Northeast direction.
194
0 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL [d
B]
0 2
-2
0
26300 Hz
0 2
-2
0
28000 Hz
0 2 -2
0
210000 Hz
0 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
216000 Hz
Array Length [m]0 2
-2
0
220000 Hz
Array Length [m]
Direction West Up
0 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
26300 Hz
0 2
-2
0
28000 Hz
0 2 -2
0
210000 Hz
0 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
216000 Hz
Array Length [m]0 2
-2
0
220000 Hz
Array Length [m]
Direction West Mid
0 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
26300 Hz
0 2
-2
0
28000 Hz
0 2 -2
0
210000 Hz
0 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
216000 Hz
Array Length [m]0 2
-2
0
220000 Hz
Array Length [m]
Direction West Down
Figure G- 7. Deviation of pressure measurements from free field from chamber center in
the West direction.
195
0 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
] 0 2
-2
0
26300 Hz
0 2
-2
0
28000 Hz
0 2 -2
0
210000 Hz
0 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
216000 Hz
Array Length [m]0 2
-2
0
220000 Hz
Array Length [m]
Direction Northwest Up
0 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
26300 Hz
0 2
-2
0
28000 Hz
0 2 -2
0
210000 Hz
0 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
216000 Hz
Array Length [m]0 2
-2
0
220000 Hz
Array Length [m]
Direction Northwest Mid
0 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
26300 Hz
0 2
-2
0
28000 Hz
0 2 -2
0
210000 Hz
0 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
216000 Hz
Array Length [m]0 2
-2
0
220000 Hz
Array Length [m]
Direction Northwest Down
Figure G-8. Deviation of pressure measurements from free field from chamber center in the Northwest direction.
196
0 1 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
] 0 1 2
-2
0
26300 Hz
0 1 2
-2
0
28000 Hz
0 1 2 -2
0
210000 Hz
0 1 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 1 2
-2
0
216000 Hz
Array Length [m]0 1 2
-2
0
220000 Hz
Array Length [m]
Direction North Up
0 1 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
]
0 1 2
-2
0
26300 Hz
0 1 2
-2
0
28000 Hz
0 1 2 -2
0
210000 Hz
0 1 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 1 2
-2
0
216000 Hz
Array Length [m]0 1 2
-2
0
220000 Hz
Array Length [m]
Direction North Mid
0 1 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
]
0 1 2
-2
0
26300 Hz
0 1 2
-2
0
28000 Hz
0 1 2 -2
0
210000 Hz
0 1 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 1 2
-2
0
216000 Hz
Array Length [m]0 1 2
-2
0
220000 Hz
Array Length [m]
Direction North Down
Figure G-9. Deviation of pressure measurements from free field from chamber center in
the North direction.
197
0 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
] 0 2
-2
0
26300 Hz
0 2
-2
0
28000 Hz
0 2 -2
0
210000 Hz
0 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
216000 Hz
Array Length [m]0 2
-2
0
220000 Hz
Array Length [m]
Direction Southwest Up
0 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
26300 Hz
0 2
-2
0
28000 Hz
0 2 -2
0
210000 Hz
0 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
216000 Hz
Array Length [m]0 2
-2
0
220000 Hz
Array Length [m]
Direction Southwest Mid
0 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
26300 Hz
0 2
-2
0
28000 Hz
0 2 -2
0
210000 Hz
0 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
216000 Hz
Array Length [m]0 2
-2
0
220000 Hz
Array Length [m]
Direction Southwest Down
Figure G-10. Deviation of pressure measurements from free field from chamber center in the Southwest direction.
198
0 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
] 0 2
-2
0
2 6300 Hz
0 2
-2
0
28000 Hz
0 2 -2
0
210000 Hz
0 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
2 16000 Hz
Array Length [m]0 2
-2
0
220000 Hz
Array Length [m]
Direction South Up
0 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
2 6300 Hz
0 2
-2
0
28000 Hz
0 2 -2
0
210000 Hz
0 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
2 16000 Hz
Array Length [m]0 2
-2
0
220000 Hz
Array Length [m]
Direction South Mid
0 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
2 6300 Hz
0 2
-2
0
28000 Hz
0 2 -2
0
210000 Hz
0 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
2 16000 Hz
Array Length [m]0 2
-2
0
220000 Hz
Array Length [m]
Direction South Down
Figure G-11. Deviation of pressure measurements from free field from chamber center in the South direction.
199
0 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
] 0 2
-2
0
2 6300 Hz
0 2
-2
0
28000 Hz
0 2 -2
0
210000 Hz
0 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
2 16000 Hz
Array Length [m]0 2
-2
0
220000 Hz
Array Length [m]
Direction Southeast Up
0 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
2 6300 Hz
0 2
-2
0
28000 Hz
0 2 -2
0
210000 Hz
0 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
2 16000 Hz
Array Length [m]0 2
-2
0
220000 Hz
Array Length [m]
Direction Southeast Mid
0 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
]
0 2
-2
0
2 6300 Hz
0 2
-2
0
28000 Hz
0 2 -2
0
210000 Hz
0 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB]
0 2
-2
0
2 16000 Hz
Array Length [m]0 2
-2
0
220000 Hz
Array Length [m]
Direction Southeast Down
Figure G-12. Deviation of pressure measurements from free field from chamber center in the Southeast direction.
200
0 1 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
] 0 1 2
-2
0
26300 Hz
0 1 2
-2
0
28000 Hz
0 1 2 -2
0
210000 Hz
0 1 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 1 2
-2
0
216000 Hz
Array Length [m]0 1 2
-2
0
220000 Hz
Array Length [m]
Direction East Up
0 1 2 -2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
]
0 1 2
-2
0
26300 Hz
0 1 2
-2
0
28000 Hz
0 1 2 -2
0
210000 Hz
0 1 2 -2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL
[dB
]
0 1 2
-2
0
216000 Hz
Array Length [m]0 1 2
-2
0
220000 Hz
Array Length[m]
Direction East Mid
0 1 2
-2
0
2 5000 Hz
Diff
eren
ce S
PL
[dB
]
0 1 2
-2
0
26300 Hz
0 1 2
-2
0
28000 Hz
0 1 2
-2
0
2 10000 Hz
0 1 2
-2
0
2 12500 Hz
Array Length [m]
Diff
eren
ce S
PL [d
B]
0 1 2
-2
0
216000 Hz
Array Length [m]0 1 2
-2
0
220000 Hz
Array Length [m]
Direction East Down
Figure G-13. Deviation of pressure measurements from free field from chamber center in the East direction.
201
BIOGRAPHICAL SKETCH
Jose Mathew was born on May 17, 1978, in a town called Kottayam of Kerala
state, located in the southern part of India. After finishing schooling at Loyola,
Trivandrum, in 1995, he pursued his bachelor’s degree in aerospace engineering from the
Indian Institute of Technology, Madras. He started pursuing his M.S. degree in
aerospace engineering at the University of Florida in the fall of 1999 under the guidance
of Dr. Louis N. Cattafesta III. The thesis topic concerned the modeling and design of
piezoelectric actuators for fluid flow control. After graduating in 2002, he continued his
education by pursuing a doctoral degree concerning the design, fabrication, and
characterization of an anechoic wind tunnel facility. His areas of interest include
aeroacoustics, fluid mechanics and gas dynamics.