Experimental Fluid Mechanics R.

R.

J.

Engineering

Aeroacoustic Measurements

With 321 figures and 184 tables

, Springer

PROF. DR. THOMAS J. MUELLER University of Notre Dame

Aerospace and Mechanical Engineering 112 Hessert Center

Notre Dame, 46556-5684 IN USA

Authors

CHRISTOPHER S. ALLEN NASA Johnson Space Center MailCode SF22, 2101 NASA Road 1 Houston, 77058 Tx

USA

WILLIAM K. BLAKE David Taylor Model Basin Code 7051, NSWC Carderock Div.

Bethseda, 20084-5000 MD USA

Library of Congress Cataloging-in-Publication Data

ROBERT P. DOUGHERTY OptiNav Inc.

10914 NE 18^th Street Bellevue, 98004 WA USA

DENIS LYNCH

University of Notre Dame Hessert Center

Notre Dame, 46545 IN USA

PAUL T. SODERMAN

NASA Ames Research Center Mail Stop 247-2

Moffett Field, 94035-1000 CA USA

JAMES R. UNDERBRINK

Boeing Commercial Airplane Group P.O. Box 3707 M/C 1 W-03

Seattle, 98124 WA USA

Aeroacoustic measurements: with 184 tables I Thomas J. Mueller (ed.).

(Experimental fluid mechanics) (Engineering online library)

ISBN 978-3-642-07514-8 ISBN 978-3-662-05058-3 (eBook) DOI 10.1007/978-3-662-05058-3

This work is subject to copyright. All rights are reserved, whether the whole or part of the material is con- cerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from

Springer-Verlag Berlin Heidelberg GmbH.

Violations are liable for prosecution under the German Copyright Law.

http://www.springer.de

© Springer-Verlag Berlin Heidelberg 2002

Originally published by Springer-Verlag Berlin Heidelberg New York in 2002

The use of general descriptive names, registered names, trademarks, etc. in this publication does not imp- ly, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use.

Coverdesign: design & production, Heidelberg Typesetting: Fotosatz-Service Kohler GmbH, Wiirzburg

SPIN: 10831770 61/3020Wei Printed on acid-free paper -5432 10-

University of Illinois at Urbana-Champaign Dept. of Theoretical and Applied Mechanics 216 Talbot Laboratory

104 South Wright Street Urbana, IL 61801 USA

PROF. M. GHARIB

California Institute of Technology Graduate Aeronautical Laboratories 1200 E. California Blvd.

MC205-45 Pasadena, CA 91125 USA

PROF. DR. WOLFGANG MERZKIRCH

Universitat Essen

Lehrstuhl fiir Stromungslehre Schiitzenbahn 70

45141 Essen Germany

Lehigh University

Dept. of Mechanical Engineering and Mechanics

Packard Lab.

19 Memorial Drive West Bethlehem, PA 18015-3085 USA

PROF. J. H. WHITELAW

Imperial College

Dept. of Mechanical Engineering Exhibition Road

London SW7 2BX UK

During the past three decades, there has been a growing concern over the in- crease in noise pollution that comes as a direct result of the increased volume of automobile traffic, high-speed trains, and larger aircraft. Additional sources of noise are commonly found in air handling equipment (such as fans and pro- pellers) and a variety of machinery used in construction and manufacturing.

A vast majority of these noise sources are the result of a given system's aero- acoustic response, or sound generated by the interaction of a flow field with the given structure.

While barriers are commonly used to shield communities from highway and train noise, and absorption materials are used to shield machinery noise, there is no way to shield communities near major airports from the noise gen- erated by low-flying aircraft. Tens of millions of people worldwide are affected by this airport noise problem. In densely populated Europe, up to 15 % of the total population is strongly influenced by airport noise. Since the volume of air traffic will continue to grow, so too will the problem and the number of people involved. It is not surprising that many countries and communities have taken legal action to preserve the quality of life in these areas. As a result, the airlines, airports, manufacturers and governments are working together to set new standards for aircraft noise reduction. In order to establish realistic goals, the generation and propagation of acoustic sources must be better understood.

The goal of aero acoustic measurements is to provide a basis for assessing mechanisms of noise generation, and to develop methods of reducing noise to more acceptable levels. However, the measurements themselves are complex, and require a deep understanding of the experimental facility utilized (such as a wind tunnel), measurement instrumentation, and data analysis techniques.

This book contains descriptions of the state-of-the-art in aero acoustic mea- surements by recognized leaders in the field. In Chapter I, Paul Soderman and Christopher Allen describe techniques, corrections, and concerns involved in the setup, measurement, and data reduction involving microphone measure- ments within and outside of the airstream. Robert Dougherty develops beam- forming techniques with sparse wide-band phased arrays of microphones in Chapter 2. These techniques allow for non -intrusive measurements in acousti- cally untreated hard-wall wind tunnels. James Underbrink presents methods

for the design and use of phased-array microphones in low-speed wind tunnel testing in Chapter 3. In Chapter 4, William Blake and Denis Lynch develop cor- relation measurement techniques as they apply to the description of acoustic transmission paths. Finally, Chapter 5 describes the design of a low-speed anechoic wind-tunnel facility. This facility, housed at the Hessert Center for Aerospace Research, has been used extensively in aero acoustic studies in- volving propeller noise due to inflow distortion and turbulence, and some of the results of these studies are also presented in this chapter.

The topics covered were chosen because of their importance in current and future aero acoustic research. Each chapter represents a large amount of cru- cial information that is collected, organized and presented by experts in these areas. As editor, I would like to express my appreciation to the authors for tak- ing time out of their busy schedules to pass on their expertise to others inter- ested in this challenging field. I would also like to thank the staff of Springer- Verlag for their patience with us, and for making this Volume possible.

Thomas

J.

Preface . . . VII 1. Microphone Measurements In and Out of Airstream 1 Introduction . . . 1

Research Objectives (and What to Measure) 2

Wind Tunnel Background Noise Including Flow-Induced Noise 3sw Drive Fan . . . 4

Self-Noise from Strut-Mounted Microphones 6

Air Ducts ... . . . 11

Voice Communication in Wind 13

Strut Noise . . . 13

Wall-Mounted Microphones 18

Background Noise from Wall Boundary Layer 22

Open-Jet Background Noise 23

Microphone Placement 24

Directivity ... 24

Near-Field Effects . 25

Background Noise and Microphone Placement 25

Reverberant Field . . . 26

Reflections in a Semi-Anechoic Environment 26

Tonal Sound Interference .. . . . 27

Random Sound Interference ... 29

Source Identification by Signal Processing 30

Convection Effects and Doppler Shift ... 31

Sound Propagation Through Shear Layers in Open-Jet Wind Tunnels 34 Change in Acoustic Propagation Direction by Refraction .. 34 Change in Sound Pressure Level Caused by Refraction ... 38 Wave Absorption And Scattering by Shear Layer Turbulence 40 Procedure for Applying Shear Layer Corrections . . . 41

Microphone Corrections at High Frequency 42

Free-Field Response . 42

Pressure Response . . . 43

Free-Field Correction ... 45

Microphone Directional Response 47

Aerodynamic Microphone Forebody Frequency Response

and Directivity . . . 47

Scaling, Extrapolation and Flight Simulation 51

Removal of Test Day Effects . . . 51

Scaling From Small Scale to Full Scale 52

Flyover Simulation 53

References . . . 58

2. Beamforming in Acoustic Testing 62

Nomenclature 62

Introduction . 64

Analysis of Wind Tunnel Acoustic Propagation by Geometrical Optics 66 Uniform Flow . . . 66 Amplitude Approximations for Nonuniform Flow 69

Ray Tracing for Travel Time 69

Array Source-Receiver Model 72

Temporal Considerations 72

Distributed Sources .... 73

Beamforming . . . 75

Microphone Weight Vectors 75

Beamforming Expressions 76

Performance Analysis ... 76

The Point Spread Function and Sidelobes 77

Effect of Wind Tunnel Walls 79

Reflected Images . . . 80

Sidelobes from Reflected Images 81

Resolution Requirement 81

Removal of Flow Noise 83

Isolation of the Diagonal Elements of the Cross Spectral Matrix 83

Diagonal Elements Are Not Helpful 83

Diagonal Elements Are Harmful . . . 84 Beamforming without the Diagonal . . . 85 Determination of Quantitative Source Spectra by Integrating

the Beamform Map . . . . 86

Normalizing the Integral to Account for the Width of the Point Spread Function . . . 86

Rejecting Sidelobes with a Threshold . . . 87 Caution #1: The Sidelobes Are Controlled by the Loudest Source 88 Caution #2: The Threshold Excludes Some Real Noise . . 88 Caution #3: Array Performance May Be less than Optimal 88 Successful Integration . . . 88 Eigenvalue Classification for Quantitative Source Spectra

Relationship between Beamforming Weight Vectors and Cross Spectral Matrix Eigenvectors .

The Array-Centric Definition of Sources Classification . . . . Benefits . . . .

89 89 90 91 91

Coherent Sources and Virtual Microphones 92

Array Calibration Using a Speaker . 93

Setup . . . 93 The Diagonal Calibration Matrix . . . 94 The Effect of Speaker Calibration for Microphone Position Errors 94

Array Level Calibration 96

Conclusions 96

References . 97

3. Aeroacoustic Phased Array Testing in Low Speed Wind Tunnels 98 Introduction . . . 98 Justifying the Cost of Aeroacoustic Phased Array Testing 99 An Overview of Aeroacoustic Phased Array Deployment 100

Array Design . . . 100

Array Mount Design and Build . . . 100 Instrumentation Plan . . . 10 1

Data Acquisition System Configuration 101

Installation . . . . 10 1

Array Calibration 10 1

Testing ... 102

Tear Down .... 102

Non-intrusive Aeroacoustic Array Measurement 102

Broadband Array Design 103

Background . . . 103

Beamforming . . . 104

Evaluating Array Performance 108

Array Resolution . . . 109

Spatial Aliasing . . . Array Design Strategy . Random Array Theory

Aperiodic Array Design and The Co array Spiral Arrays . . . . Other Array Design Strategies .... . Designing Arrays for Existing Facilities Array Design Process ... . What to Do in Difficult Situations Array Construction and Installation .

Panel Strength Requirements ...

Simultaneous Measurement Considerations Sensor Mounting ...

Calibration Requirements Traverse Requirements Nested Arrays ... . Cable Strain Relief . . . .

Heating to Avoid Potential Condensation Problems Geometric Survey Considerations

Recessed Array Considerations Laminar Flow Control .... . Array Cover . . . . Sensor Location Identification Hole Location Accuracy Hole Plugs

Instrumentation Sensors ...

Signal Conditioning

Instrumentation Setup and Checkout Phased Array Data Acquisition . . . .

Data Acquisition System Requirements Data Acquisition System Architecture Acquiring the Data

Data Management . . Array Calibration ....

Calibration Enclosure

Array Calibration Source Requirements Calibration Source Evaluation ...

Locating the Calibration Source in the Wind Tunnel - Geometric Survey Techniques .... .

Calibration Data Acquisition ... . Evaluating Array Calibration Goodness

112 114 117 117 119 128 128 129 131 141 142 142 144 144 145 149 149 151 154 154 155 155 155 156 158 158 159 175 178 179 179 186 193 196 199 200 201 201 202 204 205

Known Sources . . . 205

Multiple Array Calibration 206

Phased Array Data Reduction 208

The cross-spectral matrix . 208

Use of Parallel Processing . 209

Special Considerations for Pressurized Wind Tunnel Testing 211 Pressurization of Instrumentation . . . 211 Special Instrumentation Configuration Requirements . . 211 Reference Microphone Electrostatic Response Calibration

Under Pressure . . . 212

Tunnel Operation 213

Acoustic Phased Array Testing in Conjunction with Traditional

Aerodynamic Test Techniques . . . 213 Beyond Aeroacoustic Phased Array Measurements in Low Speed

Wind Tunnels 214

References . . . 215 4. Source Characterization By Correlation Techniques 218

Nomenclature 218

Introduction . 219

Mathematical Definitions 220

Cross Spectral Analysis of Linear Systems 222

Application to Aeroacoustic Applications: The Correlation Volume,

Correlation Area, and Correlation Time . 225

Examples of Correlations . . . 228

Correlation of Fluid Motion Variables 228

Correlation of Acoustic Pressure 229

References . . . 256 5. An Anechoic Facility for Basic Aeroacoustic Research 258

Nomenclature 258

Introduction . 259

The Design of an Anechoic Wind Tunnel Facility . 260

The Anechoic Room . . . 260

Wind Tunnel Design Criteria 261

Inlet Section . . . 262

Diffuser Components 264

Diffuser Materials .. 266

Wind Tunnel Drive System 268

Analysis of the Anechoic Wind Tunnel Facility Performance 268

Acoustic Calibration . . . 269

Aerodynamic Calibration . . . 270 Design Summary. . . 274

Propeller Response to Inflow Distortions 275

Background . . . 275

Initial Measurements and Analysis . . 276

Development of Unsteady Surface Pressure Sensors

for Model Propellers . . . 278 Unsteady Pressure Measurements Using a Single, Thin Airfoil 279 Experimental Characterization of Grid-Generated Turbulence,

Including the Aeroacoustic Response of a Downstream Propeller 280 Unsteady Pressure Measurements of a Four-Bladed Propeller

Ingesting Turbulence 281

Summary . 306

References 306

Authors . . 309

Biographical Sketch of Authors 311

of Airstream

PAUL T. SODERMAN ¹ and CHRISTOPHER S. ALLEN²

1 NASA Ames Research Center, mail stop 247-2, Moffett Field, California 94035

2 NASA Johnson Space Center, mail code SF22, Houston, Texas 77058

Abstract

The wind tunnel has become an important research facility for the study of aircraft and automobile noise. In this chapter, the acoustic characteristics of wind tunnels are discussed along with methods for conducting research in such an environment. Microphone mea- surements require low background noise and minimal reflections for accurate results. Typ- ical sources of wind tunnel background noise are described including noise from the wind tunnel components, apparatus support struts, and microphones installed in the flow. In most cases, proper design of wind tunnel components and test apparatus are critical to suc- cessful aero acoustic measurements. And, it is often necessary to add silencers and acoustic treatment to the facility. Criteria for proper simulation of aero acoustic phenomena are dis- cussed along with necessary data manipulations to correct for propagation effects, scaling to the correct source size, and extrapolating to the desired flight or drive-by situation. Fi- nally, current methods are discussed for identification and analysis of noise sources using advanced signal analysis techniques.

Introduction

Aeroacoustic study of fluid-mechanically-generated sound (aerosound) has become an important research endeavor because of the growing need to con- trol aircraft and automobile noise. Research in aero acoustics has expanded with the development of special wind tunnels designed for experimental sim- ulations (Soderman 1999). Government, industry, and university researchers have developed new laboratories and modified existing facilities to study air- craft and automotive noise. Because both closed-test section and open-jet wind tunnels are used to simulate noise generating vehicles in motion, the op- portunity exists for acoustic measurements both in airstream and out of airstream.

This chapter discusses methods for using microphones in wind tunnels to measure sound. Acoustic measurements require low background noise and minimal acoustic reflections for accurate results, requirements that are diffi- cult in most wind tunnels. Typical sources of wind tunnel background noise are described including drive fans, apparatus support struts, shear layers, and microphones themselves. Techniques for minimizing microphone self-noise and analyzing acoustic data are described. Propagation effects in flow and

through open-jet shear layers are discussed along with methods to scale and extrapolate model data.

Related to the problem of microphone self-noise is the classic challenge of recording voice communications that are masked by wind noise. This chapter discusses methods to deal with this problem by using technologies developed to minimize flow noise on microphones in wind tunnels.

Research Objectives (and What to Measure)

Wind tunnels are often used for aircraft flyover simulation in which the prop- agating sound levels radiated by a vehicle or vehicle component in motion are usually important to the researcher. Aeroacoustic sources by their very nature generate sound because of unsteady pressures on and around a body moving through air. In the case of propulsion devices such as jets or propellers, the source mechanism itself is rooted in fluid mechanics, and the sound radiation is modified by the external flow.

Almost as important as absolute sound level is the change in sound level caused by a change in model configuration or operating condition. For exam- ple, measurement of noise reduction by a muffler is often paramount, whereas the absolute sound level may be of secondary importance. By recording only a change in sound level, the researcher can ignore certain complications such as simulating the entire vehicle to get the total acoustic environment.

Another important objective in aero acoustics is source directivity because what might be perceived as a noise reduction at one location could be a shift in acoustic energy from one direction to another. Consequently, acoustic mea- surements in airstream or out of airstream require multiple measurement lo- cations around the source so as to map the radiating acoustic field. This can be difficult to achieve if microphone access is limited or, as with phased arrays, the numerous microphones, cabling and support struts are not easily moved.

However, we can simplify the problem by ignoring directions that are less im- portant than others. Figure 1.1 illustrates an aircraft flyover simulation in the NASA Ames 40- by 80-Foot Wind Tunnel (40x80) that involved mapping the noise radiated below a jet simulator attached to a semispan wing mounted ver- tically. A few measurement points off the wing tip were recorded with ceiling- mounted microphones, but the noise radiated above the wing was considered unimportant. Some jet noise suppressors are in fact designed to redirect sound upward where it will not bother anyone.

In addition to the basic acoustic parameters discussed above, it is also nec- essary to document the test environment. All distances and angles between the model and microphones must be recorded using a suitable coordinate system.

Static temperature, static pressure, and humidity are needed to evaluate at- mospheric sound absorption and, if desired, correct the data to a standard day atmosphere (see section below). At a minimum, the flow field can be charac- terized by airspeed, but a more thorough documentation would include Mach

Fig. 1.1. Aircraft flyover noise simulation in the 40 x 80. Microphones mounted to travers- ing carriages map the engine noise radiated under the wing. Sound absorbent linings cover the test section walls and traverse fairing

number and dynamic pressure - parameters that are normally provided by wind tunnel operators. Other parameters such as body forces and local pressures also provide information relating the acoustic sources to fluid mechanics.

The aeroacoustic researcher should be aware of the many fluid mechanic measurements now available to document or visualize the flow field in a way that will shed light on the acoustic source generation. There are also other acoustic objectives and metrics that might be important to a specific simula- tion such as acoustic intensity, sound power, effective perceived noise level, and so on.

Wind Tunnel Background Noise Including Flow-Induced Noise

To measure aerosound from a powered or unpowered model in a wind tunnel, it is necessary to know the background noise levels in the facility. Such knowl-

edge along with estimates of model noise levels allows the researcher to esti- mate the signal-to-noise ratio for test planning and to identify any portions of the acquired data that were contaminated by background noise. Generally, the desired signal should be at least 10 dB greater than the background noise level in the frequency range of interest, though the signal can be recovered if the signal-to-noise ratio is on the order 6 dB, less than 6 dB and the data accuracy becomes poor. If part of the acoustic spectrum is contaminated by back- ground noise, other parts of the spectrum may have adequate signal-to-noise ratio and contain useful information.

Jacob et al. (2000) developed methods to measure aero acoustic sound below the ambient level using causality techniques. The techniques require at least two microphones - one near the source and one in the radiating field. The co- herence of the two signals is used to separate the source noise from back- ground noise.

The sources of wind tunnel background noise are the drive fan, wall bound- ary layer, test dependent hardware, and microphone self noise. The latter can be associated with the microphone boundary layer, screen or cavity perturba- tions, electronic noise, and free-stream turbulence.

Drive Fan

Most subsonic wind tunnels are powered by single-stage axial fans that gener- ate broadband (random) noise and tones at the fundamental blade passage rate (blade number x revolutions per second) and harmonics. Most axial fans are operated at low speed for efficiency, so the fan tone harmonics usually dominate the low frequencies of the background noise spectrum. Generally, the tones peak at the first or second harmonic, decay quickly and become masked by broadband noise at frequencies beyond the fourth or fifth har- monic. Although the strength of fan noise depends greatly on inflow distur- bances, an estimate of the drive-fan sound power level (tonal and broadband) in a third-octave band can be made using the following empirical equation (Soderman and Mort 1983):

Lw(f) = -58.2 - 10 10glO [1 + (4.4X)2] + 10 10glO!

+ 40 10glO N + 70 loglo Dt + 10 10glO Q + 10 loglo Fn + 0.3f3 (dB rei 10-¹² watts)

(1.1)

where x = QfIN, Q = 1 - (DHID_t)3, DR = hub diameter, m, Dt = tip diameter, m,!

=

=

=

=

Fan sound power can be related to test section sound levels by computing the acoustic power flow around the circuit (a non-trivial exercise) or by mea- surements of test section sound levels with a known sound power source in the

fan section (Wilby and Scharton 1975). Information on sound transmission around corners and through turning vanes is a key part of the power flow analysis. Figure 1.2 shows third -octave spectra in several sections of the NASA Ames 7-by lO-Foot Wind Tunnel (7x 10) at zero airspeed with a broadband noise source in the fan drive area. Similar curves were obtained for upstream propagation. Below 1000 Hz, the sound attenuates as it propagates toward the test section. Baffle 'f' was acoustically treated. Above 1000 Hz, the sound atten- uates much less because the corner vanes turn and radiate sound efficiently for wavelengths equal to or less than the vane chords (Soderman and Hoglund 1979). Thus, high frequency sound traveling along the duct axis turns the cor- ner and follows the next duct section with little energy loss. Low frequency sound waves diffract around the vane set and strike the walls. The data of Fig- ure 1.2 were obtained using a calibrated (ILG Industries) fan that generated third-octave sound power levels of approximately 76 dB re 10-¹² watts between 50 Hz and 4 kHz. In the 7 x 10, the difference between sound power input in the fan section and the test section sound pressure levels were as follows:

Lw (/) - Lp (f) '=' 37 dB 200 Hz

5,/5,

Lw (f) - Lp (f) '=' 28 dB 1000 Hz

< /5,

(1.2) (1.3) These results depend on the acoustical property of the wind tunnel circuit. In the 40 x 80, the decibel values in Equations 1.2 and 1.3 are greater because of the larger distances and volumes involved. At the German-Dutch (DNW) wind tunnel in Holland, the test section noise is reduced by acoustically treated turning vanes that attenuate fan noise by 6 dB at the first corner and 17 dB at the fourth corner, both at 2.5 kHz (Van Ditshuizen and Ross 1982).

Researchers often measure test section background noise prior to installa- tion of the model to be tested rather than use empirical estimates of back- ground noise. It is helpful to identify the various background noise sources so that steps can be taken to minimize them if possible. It is also important to know how the sound varies with airspeed. The noise variation with fan speed is somewhat obscure in Equation 1.1 because N appears in two parameters.

The NASA Ames 40 x 80 drive-fan tonal and broadband noise vary with rota- tional speed as (Soderman 1988, Jaeger et al.I995):

(1.4) This is typical for dipole-type sources related to blade loads, exponents be- tween 5 and 6 are often reported for axial-flow fan noise.

Equation 1.4 is valid for fixed blade-pitch operation. A few wind tunnels, such as the 40x80, have drive fans with variable-speed, variable-pitch blades that allow flexibility in minimizing power consumption and noise radiation.

The variation (ildB) of 40 x 80-fan noise with blade pitch angle (f3) at fixed rpm was determined experimentally to be (Soderman 1988):

60.---7---.

I ' \

'/, \ _ fan leg -g ...

1/ , \" .... __ ... _ "'-""" ,

-,

.' \end leg -f "

\ \, "r .... - __ ... ~~ .. ... _""', '

",....

...

x. ",.. ___ . ./'

"

/.-

-..

settlirig chamber - e ,

100 1000

. _•

10000 a frequency, 113 octave bands, Hz

b

LldB = 0.3 Llf3 LldB = 0.5 Llf3

Fig. 1.2. Sound distribution in the Ames 7 x 10 circuit with a noise source near the drive fan, zero air- speed

(1.5) (1.6) Variable blade-pitch operation results in considerably quieter fan operation.

For example, at 204 km/h test section airspeed the 40 x 80 drive fans at 135 rpm are 4 dB quieter and consume 2 MW less power than at 180 rpm. The cor- responding blade-pitch angles are 24° and 16°, respectively. At 124 km/h airspeed, fan operation at 60 rpm is up to 18 dB quieter than operation at 180 rpm.

Other fan noise control methods include sound absorbent walls and baffles (including acoustically treated turning vanes) and active sound cancellation.

In all cases, good aerodynamic design of fans is important for mechanical ef- ficiency and minimal noise generation.

Self-Noise from Strut-Mounted Microphones

The most common instrument for measuring aerosound in closed-jet wind tunnels is a condenser or electrostatic microphone sensor mounted in a screened cavity that is protected from the flow impact by an aerodynamic fore-

body such as those illustrated in Figure 1.3. The microphone with preamplifier threads into the back of the forebody, which is pointed into the wind. An aero- dynamic strut supports the assembly. Microphones can also be flush mounted in a panel or wind tunnel wall without forebodies, but the sensors are then in close contact with a (usually) turbulent, noisy boundary layer (see section on wall-mounted microphones). For phased microphone arrays, this is a common arrangement because signal processing mitigates flow noise. But for individual microphone measurements, strut-mounted probes are usually preferred.

The aerodynamic forebody exposes the sensor to static pressure fluctua- tions and isolates it from total pressure fluctuations. Thus, as the acoustic pres- sures oscillate about the airstream static pressure, the sensor is in the optimum location to detect sound - this will be discussed in more detail in the section on convection effects. Forebodies also protect the fragile microphone dia- phragms and reduce flow-induced noise. With a forebody, the microphone re- sponse becomes virtually omnidirectional at frequencies below 5 kHz. How- ever, at frequencies above 35 kHz, the forebody screens can attenuate sound by as much as 30 dB. Careful calibrations and corrections are required to account for the attenuation as described below. The Briiel and Kjaer (B&K) nose cone design has been marketed for years. Allen (1993) developed the FITE (Flow-In- duced Tone Eliminator) forebody at NASA Ames; Dassen, et al. (1996) at DNW (German-Dutch Wind Tunnel) developed similar devices independently. Even with the forebodies shown in Figure 1.3, Allen and Soderman (1993, 1998) found that microphone self-noise caused by turbulence in the airstream dom- inates the background noise in the 40 x 80 above 185 km/h from roughly 200 Hz to 5 kHz. Flow turbulence is related to wind tunnel design. Because many wind tunnels have similar, low turbulence flows by design, the back- ground noise levels of many facilities are remarkably similar in that frequency range (Soderman 1988).

The 6.4-mm diameter FITE forebody in Figure 1.3 was designed to elimi- nate high frequency cavity tones generated by the B&K commercial forebody.

B&K UA 0385 forebody

FITE forebody

(Flow-Induced Tone Eliminator)

Fig. 1.3. Two 6.4-mm diameter forebodies designed to protect in-flow microphones from wind

Figure 1.4 compares the 6.4-mm FITE and B&K forebody response at Mach 0.36 (445 km/h). Although the strong B&K cavity tones are beyond the audible range, they can be devastating to small-scale simulations that require high fre- quency data in order to model large-scale, low frequency acoustic phenomena.

Studies by Fields, et al. (1997) indicated that the shear layer over the B&K sen- sor cavity penetrates the screen and periodically impacts the back cavity wall to generate tones. The phenomenon seems to be aggravated by the streamwise pressure gradient and laminar boundary layer at the screened cavity, which is located close behind the forebody nose. The FITE forebody, which has more length from nose to sensor, dampens the shear-layer oscillations by creating a weakly turbulent boundary layer and a zero streamwise pressure gradient at the sensor screen.

As mentioned above, turbulence embedded in the free stream has a very strong effect on microphone noise in flow, but the mechanisms are complex and not well understood. To investigate turbulence-induced noise in a system- atic way, Allen and Soderman (1998) developed a very quiet, low-turbulence wind tunnel in which they could vary turbulence from low to high levels of in- tensity and show a direct cause and effect between flow turbulence and mi- crophone self noise. Figure 1.5 a shows response of the 6.4-mm B&K and FITE forebody equipped microphones to three flow fields having low, medium, and high turbulence intensities and a mean velocity of 185 km/h. The correspond- ing turbulence spectra acquired by hot wires are shown in Figure 1.5 b. As free- stream turbulence was increased, the microphone self noise increased accord- ingly. The streamwise and cross-stream components of turbulence were gen- erally of similar magnitude. It was found that the FITE forebody was less sensitive to streamwise turbulence than the B&K type UA 0385 forebody. Both probes are equally sensitive to cross-stream turbulence. Bauer (1994) devel- oped a cloverleaf cross-sectional shape probe that minimizes response to cross-stream turbulence. It is clear that a reduction of wind tunnel turbulence would reduce microphone self-noise.

110 100

'"

~ 90

0 N "

..

£@ 70

~ 60 50

0 20 40 60 80

fi:equellCY, kHz (M = 187 Hz) 100

Fig. 1.4. A comparison of 6.4-mm FITE and B&K UA 0385 flow noise narrow band spectra to 100 kHz, U= = 445 km/h (Allen and Soder- man 1993)

90Tr---,---,---~---,---.

- - 6.4-nun B&E! (UA 0385)

I .

^{~.4-nun FITEI}

" ... · ... ·."· .. I ... ,,· ... · .. ··,, ... l!lcrea smg turbulence "'"

; /1 .

80

i

60

SO~~~~~~~-r~~~~~~~~~~

o

a frequency, Hz (M = 16 Hz)

~ O~---y---~---,---~---,

S

~ -10 ... ;....

..1.

o ('t1

e

¢:l high turbulence

t medium turbulence

.. ~.

1000 2000 3000 4000 5000

frequency, Hz (M

=

Fig. 1.5. (a) In-flow noise of 6.4-mm FITE and B&K UA 0385 for three values of turbulence.

The mean velocity, U = = 185 km/h. The figure (b) shows the turbulence spectra measured b)

(..[iii

a hot cross-wire probe, amplitude is 20 loglo u'. The overall turbulence levels --- x 100 for the low, medium, and high turbulence spectra are 0.085%, 0.52%, and 1.10%, respec-U=

tively

10,---~---~---~----~---,

-cross-stream turbulence

til --streamwise turbulence

¢l 0 +" ... " .. " ... " .. ". "+,., .. " ... " ... " .. " ... "+ ... " .. " ... " .. " ... " .. " .. ·"· .. ····,,1"·"·"···"·· .. ··"···"··"··· .. ··"·+-· .... ·"··"···" .. · .. "·· .... ·"··,,···"··"1

a frequency, Hz (M

=

110.---,---,

80+---~--~~~~+_--~--~~~~+_--~

100 1000 10000

b frequency, lJ3 octave bands, Hz

Fig. 1.6. Response of various microphones to medium turbulence flow in the audible range, U= = 370 km/h

Generalizing the results of Figure 1.5 to other forebodies is not straightfor- ward. Consider a medium turbulence flow with turbulence spectrum shown in Figure 1.6 and mean velocity of 370 km/h. The overall turbulence intensity (ra- tio of root-mean-square velocity to mean velocity) was 0.52 %. Allen and So- derman (1998) installed four microphones in the flow as follows: a) 6.4-mm sensor with FITE forebody, b) 6.4-mm sensor with B&K VA 0385 forebody, c) 12.7-mm sensor with FITE forebody, and d) 12.7-mm sensor with B&K VA

0386 forebody. The 6.4-mm FITE forebody self-noise was the lowest of the lot being as much as 18 dB quieter than the 12.7-mm FITE forebody and was quite effective at rejecting turbulence-induced flow noise. The larger FITE forebody, which has twice the distance from nose to sensor, developed a thicker bound- ary layer that presumably generated excessive noise at the sensor. Similar trends can be noted on comparison of the 6.4-mm and 12.7-mm B&K fore- bodies. Hence, both shape and size are important to forebody design, the FITE shape minimizes response to streamwise turbulence, and small size is neces- sary to minimize boundary-layer noise. It is anticipated that future micro- phones will be even smaller as long as sensitivity is adequate. Research on Micro- Electro-Mechanical Systems (MEMS) as applied to microphones in airstream is underway at NASA and elsewhere (Sheplak et al. 1999).

Foam balls or windscreens can also protect microphones, but large round ones create turbulence and wake noise in any but light winds. Nakamura et al.

(1993), however, reported good results in Honda's low noise wind tunnel with small oval-shaped windscreens measuring 50 mm streamwise and 38 mm cross stream. Mounted over 12.7-mm microphones, the windscreens con- trolled wind noise in turbulent flow at 100 km/h.

Air Ducts

Related to the problem of measuring sound in wind tunnels is the measure- ment of sound in air conditioning ducts, mine ventilation ducts, engine ex- haust pipes or in other air moving systems, many of which are of relatively small cross section. In these cases, the sound propagating along the duct axis is of prime interest and is often embedded in fully developed turbulent pipe flow. Noise from the strong turbulence induced pressure fluctuations can mask acoustic signals recorded by a simple in-flow microphone even if protected by a forebody as described above. For this reason, the long slit-tube and porous pipe microphones (also called sampling-tube microphones) were developed (see Figure 1.7).

-

damping tube r--1.27 em microphone

oo:y~~

cone protective '

~1

Fig. 1.7. Slit-tube microphone (Brfiel & Kjaer type UA 0436)

These devices function as follows. The pressure fluctuations caused by flow turbulence convect along the tube's outer surface and, entering via the slit or pores, generate sound waves along the inner surface of the tube that propagate to the sensor diaphragm. Because the turbulence convection speed outside the tube and sound speed in the tube are very different, and because the sound originates from various locations along the tube wall, the flow-induced sound waves arrive at the sensor with random phase and do not add effectively. Con- versely, a plane acoustic wave in the outer duct will also enter the tube at many locations along its length, but if the wave speeds inside and outside the tube are nearly equal, the waves will remain in phase and add constructively at the diaphragm. The resulting difference between the propagating sound and tur- bulence-induced sound is quite substantial.

Neise and Stahl (1979) showed that a slit-tube microphone with a 400-mm long, I-mm wide slit suppressed up to 20 dB of turbulence-induced noise relative to a standard B&K nosecone type VA 0386 in a 20 mls flow with a streamwise turbulence intensity of 15% (u'IU= = 0.15). Brock (1986) showed, similarly, that the B&K type VA 0436 slit-tube microphone sensed lower flow- induced noise than the B&K nosecone VA 0387 for "medium and high turbu- lence" flows. For "no and low turbulence" flows, the slit-tube microphone had higher flow-induced self-noise than the nosecone, possibly because of bound- ary-Iayer noise or surface irregularities.

Von Heesen (1999) showed that oscillations in the frequency response of the B&K type VA 0436 slit-tube microphone caused inaccuracies on the order of

±2 dB in the measurement of in-duct fan tones. In an effort to overcome this shortcoming as well as reduce the self-noise problem, Peter Baade developed a new slit-tube microphone design, the detailed drawings of which are included in the pending reversion of the ISO standard for these types of measurements, ISO DIS 5136: 1999.¹

Because of its design, the slit-tube microphone must be pointed into the wind to avoid flow separation and excessive flow noise. The directional re- sponse and signal-to-noise ratio are maximum for sound propagating down- stream. Though very effective for acoustic measurements in small ducts with axially propagating sound, the device is seldom suitable for wind tunnel stud- ies where the noise sources are usually cross-stream. The effectiveness of the slit-tube is reduced in high-speed flows because the propagating wave speed inside the tube cannot match the wave speed outside, which is the vector sum of the airspeed and sound speed. This mismatch is especially destructive to high-frequency phase correlation along the tube. Finally, high order modes in the outer duct confuse the microphone response. Neise and Stahl (1979) and Arnold (1999) developed elaborate schemes to deal with flow and modal ef- fects.

1 Private conversation with Peter Baade.

Voice Communication in Wind

Communication or audio recording microphones used outdoors are often plagued by wind-induced noise. The situation is complicated by the non-aero- dynamic shape of most microphones, the proximity of the microphone to the body, direct exposure of the sensing element to wind, and the semi-random direction and magnitude of free wind. To combat this, most microphones can be fitted with foam coverings that provide separation between the sensing element and flow field to reduce wind noise. However, foam coverings are bulky and sometimes bothersome to the user, particularly when the micro- phone is attached to the person or is meant to be unobtrusive.

Unpublished studies at NASA Ames indicate that technology developed for the wind tunnel test environment may also benefit communication micro- phones. First, we found that proximity of the microphone to the body may actu- ally help by shielding the microphone from wind coming from directions other than in front of the user. The local flow field close to the body must follow the general shape of the body, so the important wind conditions can be predicted.

Screened forebodies based conceptually on those of Figure 1.3, but allowed to have novel shapes can be designed to protect the sensor from wind in a similar manner to foam, but with more robust, compact, and attractive shapes. Minia- ture brass frames shaped aerodynamically and wrapped with 100 mks rayl2 stainless-steel screen used for wind tunnel linings (Soderman et al. 2000) had success at reducing wind noise. Double layers of 20-mks rayl screen were some- what better. To optimize the forebody for a specific application, testing ofvari- ous shapes and screens in a controlled flow field is recommended.

Strut Noise

Much noise is caused by wind tunnel test hardware such as struts and cables used to support microphones, aerodynamic probes, and models. If the test hardware sheds vortex streets that induce periodic or quasi-periodic loads on the body, the resulting noise can be quite loud. The acoustic spectra can either be tonal (aeolian tones) or haystack-shaped with a peak at the vortex shedding frequency. For an airfoil strut or cylinder, the shedding frequency in Hertz is given by:

f= Sp=

I ^(1.7)

2 The term mks rayl comes from the definition of specific flow resistance of porous mate- rials: R = i1p/u, where i1p is the pressure differential across the material, and u is the air velocity through the material. Because specific flow resistance varies with velocity, it is customary to extrapolate measured values of R to that corresponding to a low value of u between 0 and 0.05 cm/sec. The common units of Rare Nsec/m3, also called mks rayl.

where St is the Strouhal number, U = is the free-stream airspeed, and I is a char- acteristic dimension of the body equal to the vortex row separation distance, which is also the cross-stream distance between the points where the vortices leave the body. Strouhal number varies with Reynolds number, but Roshko (1968) found experimentally that cylinders in flow with Reynolds numbers be- tween 400 and 10,000 have a Strouhal number of 0.21. For cables and cylinders, the characteristic dimension I in Equation 1.7 is the diameter.

For an airfoil, the dimension in Equation 1.7 is related to the point on the airfoil where the vortices shed. At that point, the sum of the local airfoil thick- ness plus the upper and lower boundary-layer thicknesses equals the dimen- sion, l. However, the shedding point is difficult to predict. It is easier to calcu- late conditions at the airfoil trailing edge, in which case the following equation can be used (Hersh et al. 1974):

(1.8) where Du and Dz are the upper and lower boundary-layer thicknesses at the air- foil trailing edge. As a last resort, the dimension I can be approximated as 26 % of the maximum airfoil thickness, tmax •

Vortex shedding is a phenomenon sensitive to Reynolds number. Bodies in flow with very low Reynolds numbers below approximately 60 will not shed vortices periodically. At high Reynolds numbers, the turbulent boundary layer will disrupt the periodic shedding process. Patterson et al. (1972) identi- fied vortex streets in an NACA 0012 airfoil wake up to a Reynolds number of 2.2 X 106•

Cables and cylinders can create very strong vortex shedding noise and should be avoided. Even airfoils can generate loud tones. Improper airfoil shape can exacerbate the problem. An NACA 0012 airfoil, for example, will generate aeolian tones unless the boundary layer is tripped to overcome the shedding process, as was demonstrated by Hersh et al. (1974) using leading- edge serrations. Oval shaped tubing is also prone to vortex shedding.

A McMasters-Henderson (M-H) airfoil (McMasters et al. 1981) illustrated in Figure 1.8 does not generate vortex tones. The M-H airfoil section (see Table 1.1) is similar to an NACA 0030 shape, but has been subtly modified to provide a well-attached boundary layer nearly to the trailing edge. An unpublished acoustic study in the Ames 7- by 10-ft Wind Tunnel of several airfoil shapes by the coauthor showed that the M-H airfoil and an NACA 65-021 airfoil did not exhibit the distinctive tone generated by an NACA 0012 airfoil (Figure 1.8).

The NACA 0012 tones were eliminated by thick, sticky tape attached to the air- foil leading edge. Of the three airfoils, the M-H airfoil is preferred because in addition to its low self- noise, it is the thickest (30 % thickness-to-chord ratio) and structurally strongest of the three airfoils tested. In most installations, side braces can be avoided by using an M-H airfoil strut, which eliminates the strut/brace junction - a region of spoiled flow, high drag, and noise (Jaeger et al. 1995).

0.4 f-_ •..• _ ...•.. _._ ••• _ ...•• _._._ .. _ .. +-_ ... _ .. _._._+ .... _. __ ._ ... + .... _ ... __ ._-\

McMasters-Henderson

O. 2 1-.•..•...•...•...•...•..• + ... .; ... ,'< •...•..• + ... .; ... _ ... -\

-0.2 1-•..•..•....•...•...•. + ... -1

NACA0030

!

-0.4 _'~-... v'··~'~-~~-·rv'.'¥.~-~~.-, ... ^v'.'- NACA 65-021 ~ .... ~· .. ¥'·'~· .. - .... ·-·~¥'··-·'-... ~'··~'··~· ..

i

! ! ;

o

a xlchord

100 , . - - - , 90

«I 80

§: 0012 airfoil with NACA 0012 airfoil

~ 70 60

40

trippc:d flow

_I

I ^{J ~ \ \}

MlHairfoil ?'

30 +-~~-+~~~~~~_r~~~~~~

o

b frequency, Hz (M= 1.25 Hz)

Fig. 1.8. (a) McMasters-Henderson airfoil section compared with three NACA airfoil sec- tions. Acoustic data from wind tunnel tests of three of the airfoils are shown in the figure (b). The unmodified NACA 0012 airfoil generated strong tones, which were eliminated by tripping the boundary layer with sticky tape at the leading edge (duct tape with adhesive side to the flow). Tones below 500 Hz came from the wind tunnel. U= = 23 mls

A smooth blending of an M-H strut and probe support developed by the coauthor at NASA Ames Research Center and identified as ARC Quiet Strut (Jaeger et. a11995) is shown in Figure 1.9a. The probe support minimizes strut junction noise and positions the microphone away from the strut to reduce the effect of strut noise and reflections. The strut junction has a smooth fillet to eliminate any horseshoe vortex structure (Dickinson 1990). The extension of the microphone holder tube downstream of the strut junction was eliminated

Table 1.1. McMasters-Henderson symmetrical airfoil coordinates. The chordwise dimen- sion is x, and the thickness distance from the centerline is y, both are normalized by the chord c

X/C y/c X/C y/c X/C y/c X/C y/c X/C y/c

0 0 0.0436 0.0876 0.2221 0.1423 0.5407 0.1169 0.8792 0.0343 0.0004 0.0047 0.0532 0.0951 0.2620 0.1440 0.5804 0.1081 0.8991 0.0294 0.0010 0.0121 0.0628 0.1014 0.3018 0.1438 0.6202 0.0986 0.9188 0.0247 0.0017 0.0195 0.0824 0.1115 0.3416 0.1423 0.6600 0.0889 0.9390 0.0194 0.0040 0.0294 0.1025 0.1194 0.3814 0.1395 0.7001 0.0788 0.9589 0.0140 0.0077 0.0390 0.1227 0.1257 0.4212 0.1357 0.7400 0.0687 0.9788 0.0083 0.0122 0.0484 0.1428 0.1308 0.4610 0.1309 0.7800 0.0587 0.9988 0.0021 0.0176 0.0581 0.1626 0.1349 0.4808 0.1279 0.8196 0.0488

0.0235 0.0668 0.1823 0.1382 0.5008 0.1246 0.8394 0.0440 0.0337 0.0782 0.2022 0.1406 0.5205 0.1210 0.8593 0.0392

in order to create a smaller and smoother wake flow behind the strut. Com- pared with the other designs evaluated by Woodward et al. (1995) at NASA Glenn Research Center (Figure 1.9b), the Quiet Strut generated lower back- ground noise (Figure 1.9c).

Microphone strut connections should mate smoothly, especially in close proximity to the microphone diaphragm. All discontinuities, cavities, or pro- trusions should be smoothed over with putty, wax or tape. Leading-edge fillets, if not built into the hardware should be added to any junction region using putty or wax. The transition from the microphone housing to the microphone holder tube is an especially sensitive area. A conical transition piece that pro- vides a snug friction fit on the microphone housing when inserted into the mi- crophone holder tube works well. This transition piece should be made from an electrically non-conductive material to provide electrical isolation from the wind tunnel structure and other microphones.

a

Fig. 1.9. (a) ARC Quiet Strut with 12.7-mm FITE forebody

110

100

& _:::t

~ 90

e

!Xl -0 _Q; 80

...l

70

60 100 b

NASA Glen 9- by IS-Foot LSWT Mach 0.2

ARC Quiet Stnlt .

~

1000 10000 100000

fh:qucncy, lf3 octave bands, Hz

Fig. 1.9. (continued) (b) Noise from the Ames Research Center Quiet Strut compared with other struts tested at Glenn Research Center (Woodward et al. 1995). The figure (c) shows several Quiet Struts installed in the Glenn 9 x 15-Ft Wind Tunnel

Fig. 1.10. A dual microphone mount for microphones in airstream

An alternate dual probe design is shown in Figure 1.10. That concept has the advantage of redundant data acquisition - if one microphone fails, the other will allow test continuation. Cross-spectra of the two microphone signals can also be helpful to attenuate uncorrelated self-noise and estimate the sound propagation direction (Bendat and Piersol 1980).

Strut noise in open-jet wind tunnels is aggravated by the need in most cases to penetrate the turbulent shear layer to support the model, microphone or other test hardware. The turbulence intensity in the shear layer can exceed 5%

of the mean centerline velocity (Allen and Reed 1992) depending on distance downstream from the nozzle. The unsteady loads and noise from a strut in the shear flow is strong. Boeing researchers minimize the noise intrusion by using, in some cases, an overhead mount system (Figure 1.11) and recording the noise below and to the side of the model in a region more or less outside the dipole radiation pattern perpendicular to the strut. An alternative is a bayonet mount cantilevered from a solid fixture downstream of the model. Mounting to the collector is possible, but collectors are subject to strong vibrations.

Schmitz³ used an external cuff to protect a microphone strut spanning the DNW open-jet shear layer. The cuff projected part way into the stream, and an inner strut extended to the microphone position. However, inner strut vibra- tions were not eliminated. He also attached polyurethane foam on the inner strut to minimize reflections. A very porous gauze held the foam in place.

Other mounting concepts include half-models attached to a reflection plane that extends streamwise from the nozzle.

Wall-Mounted Microphones

As mentioned above, microphones are often mounted in panels or walls when used as elements in an acoustic antenna (see Chapters 2 and 3 on phased microphone arrays). In open-jet wind tunnels, the microphones can be kept

3 Private conversation with Prof. Fredric H. Schmitz, Univ. of Maryland.

Fig. 1.11. An unpowered aircraft model installed in the Boeing LSAF open-jet wind tunnel (Herkes and Stoker 1998). Acoustic wedges line the chamber walls surrounding the test section

out of the airstream, but in closed test sections the microphones are exposed to boundary-layer turbulence, which can create high noise levels. Figure 1.12 shows a wall mounted array in the Ames 7- by 10-Foot Wind Tunnel that was used to measure noise radiating from a simulated wing high-lift system. One hundred microphone diaphragms were mounted flush with the test section wall. Phased array signal processing is designed to amplify the signal from the noise source under investigation and attenuate the response to sound from other directions. Similarly, sound that is not well correlated over the array sur- face such as boundary-layer noise is attenuated relative to the acoustic target.

Thus, boundary-layer noise from flush-mounted microphones is often accept- able if the target noise is strong or if the number of microphones and averag- ing time is sufficiently large (Horne and James 1999). However, if the target noise is weak, the signal-to-noise ratio must be increased by other methods such as reducing the boundary-layer turbulence intensity or recessing the microphones.

Recessing wall-mounted microphones is quite effective at reducing self- noise from the boundary layer. The strong pressure fluctuations at the wall impart evanescent waves that decay exponentially in the direction normal to the surface. Even a small displacement of the microphones into a recess be- hind the flow surface will result in significant reduction of measured bound- ary-layer noise. If the microphones are merely recessed in their individual mounting holes, however, the flow can interact with the open tubes to cause

Fig. 1.12. Wall mounted phased array of microphones in the Ames 7- by 10-Foot Wind Tun- nel. The microphone bare diaphragms are flush with the wall (Storms et al. 1999)

acoustic resonance. The tubes also act as wave-guides that prevent the rapid decay of evanescent waves. If the entire plate containing the microphones (see Figure 1.12) is recessed, the flow can still enter the large cavity and generate additional turbulence and noise. So, it is necessary to shield the microphones from the flow. A simple way to do that is to mount a porous membrane over the recessed array cavity that is flush with the flow surface. Ideally, the porous membrane will keep the flow out of the cavity and let the sound enter unim- peded.

Jaeger et al. (2000) describe a method for reducing flow-induced noise of wall-mounted microphones using a stretched Kevlar 124® cloth sheet (7.9 gm/cm2 ) over microphones recessed 12.7 mm behind the flow surface. In wind tunnel trials, the extreme strength and durability of Kevlar 124® pre- vented flutter and withstood flow-induced fatigue. Less durable materials such as fiberglass cloth and a fine mesh metal screen failed after a short time in wind. The insertion loss of the Kevlar 124® cloth varied from nearly zero at low frequencies to about 5 dB at 20 kHz. Figure 1.13 shows the recessed array geometry and microphone mounting details. The recess depth was limited to 12.7 mm because of hardware constraints. A theoretical analysis of the evanes- cent pressure wave decay from the boundary layer indicated that the noise at- tenuation in decibels doubles for every doubling of recess depth. Thus, deeper is better - though at some point acoustic shadowing from the edges would be- come detrimental.

Figure 1.14 shows the noise reduction achieved by the microphone reces- sion of Figure 1.13. The average auto-power spectra of the array microphones flush-mounted and recessed are plotted. No array processing was done, so the curves represent the noise reduction a single microphone would experience.

porous cloth

microphone holder

b

Fig. 1.13. (a) Recessed microphone array in wind tunnel wall with circular Kevlar 124®

cover and (b) microphone mounting details (Jaeger et al. 2000)

Also plotted is the theoretical estimate of boundary-layer noise for a 12.7-mm deep recess. Below 2 kHz, the measured noise agrees with the prediction.

Above 2 kHz, the noise reduction reaches 12 to 19 dB, which is much less than predicted because of other wind tunnel noise sources external to the micro- phone array.

110 .---~--,---~--.---~--,---~--,---~--, 0.... ctI _::!. 100

0 N 0,) .... ₉₀

C::l

"'0 0,)-

Cl 80

ctI ....

0,) ffi

c.. 70

...J

60 0

predicted recessed backgrolUldnoise

2000 4000

Mach = 0_22

measlU-ed recessed backgrolUulnoise

6000 8000

fi-eqllency, Hz (M= 150 Hz)

10000

Fig. 1.14. Average microphone sound pressure levels of flush-mounted and recessed microphones in the empty 7x 10 wind tunnel at M = 0.22. The microphones were recessed 12.7 mm below a taught Kevlar® cloth mounted flush with the wind tunnel wall. Just up- stream of the array, the wall boundary layer was turbulent with a depth of 152 mm (u/U=

= 0.99). The boundary-layer momentum and displacement thicknesses were 8.1 mm and 11.4 mm, respectively 0 aeger et al. 2000)

Background Noise from Wall Boundary Layer

Depending on wind tunnel wall roughness and distance from strut-mounted microphones, wall boundary layers can be a source of background noise.

Between roughly 5 kHz and 12 kHz, the 40 X 80-background noise was domi- nated by noise from the test section wall boundary layer before steps were taken to smooth the surface. This is evident in the acoustic spectra of Figure 1.15 measured before and after installation of a new test section lining (So- derman et al. 2000). Prior to the modification, a 152-mm deep sound ab- sorbent lining with a 40 %-open perforated metal face sheet covered the walls.

Below 5 kHz, the B&K and FITE microphone forebodies respond differently to in-airstream turbulence, the dominant source mechanism in that frequency range as already discussed. But the data collapse roughly to a common haystack shape between 4 kHz and 12 kHz suggesting responsibility from a noise source external to the microphone. This was confirmed by a recent facil- ity modification that included replacement of the 152-mm deep test section lining with a I-m deep lining. A fine-mesh screen is attached to the 68%-open perforated surface panels to create a much smoother surface. Figure 1.15 shows the resulting noise reduction over a broad frequency range. Below 5 kHz, the reduction can be attributed primarily to a reduction in flow turbulence. From 5 kHz to 12 kHz, the reduction is probably related to the thinner floor bound- ary layer and improved sound absorption at all walls.