NASA SP-5108

TECHNOLOGY

HANDBOOK FOR INDUSTRIAL

NOISE CONTROL

NATIONAL AERONAUTICS AND SPACE ADMINISTRATION

 NASA SP-5108

HANDBOOK FOR INDUSTRIAL

NOISE CONTROL

Submitted to
National Aeronautics and Space Administration
Langley Research Center
Hampton, Virginia

Prepared under Contract NAS1-15721

by
The Bionetics Corporation
Hampton, Virginia

Technology Transfer Dwistbn 1981

NATIONAL AERONAUTICS AND SPACE ADMINISTRATION

V/ashtngton, D. C.
For sale by the National Technical Information Service
Springfield, Virginia 22161
Price -- $7.50
 Preface

This handbook gives basic and comprehensive information on the understanding, measurement, and
control of noise in industrial environments. It is intended for engineers with or without acoustical experi-
ence; to this end, it presents sections on noise problem analysis, instrumentation, fundamental methods of
noise control, and properties of acoustical materials.
The National Aeronautics and Space Administration has been one of the chief instigators of many
important research and development studies within the field of aeroacoustics. These have been conducted
in-house, by major aerospace contractors to NASA, and by academic institutions under the aegis of the
extensive NASA research grant program. The material included in this book is limited to that which is
clearly applicable to nonaerospace industrial noise control problems. The later chapters, in particular,
include more advanced and source-specific noise control technology. Emphasis has been placed on fan
noise reduction, noise transmission control techniques, and jet noise suppression.
An extensive bibliography and reference list of books and articles has been assembled. Many of these
resulted from NASA sponsorship, and, to a lesser extent, that of other federal agencies.
The author makes no claim to originality regarding the basic factual content of this work. In blending
some of the more esoteric NASA-sponsored research with fundamental, well-established tenets of noise
control, a careful selectivity must be exercised. In preparing this work, many NASA formal publications
and journal articles were reviewed. The author is solely responsible for the material included here, and he
apologizes to those who may feel that their work has been overlooked.

W. Graham Orr
Hampton, Virginia

iii
 Acknowledgment

This document was prepared by The Bionetics Corporation for NASA under contract No.
NAS1-15721. The author thanks Kevin P. Shepherd, Ph.D., Project Manager, for his patient assistance in
proofreading the text and for his technical counsel. He also extends his sincere appreciation to l_ester J.
Rose of NASA Langley Research Center, Technical Representative of the Contracting Officer, for advice
and coordination of review.
Too numerous to mention are the many individuals and companies who responded to requests for
information by providing updates of research work in progress, published reports, and technical specifica-
tions of equipment and materials. Source credits are given next to all photographs that appear in the text.
Finally, the support of the staff at Bionetics is acknowledged: Alan J. Rosing for his judicious tech-
nical editing; Sandy Barnes for her splendid artwork; and Frankie Freeze, Michael Platt, and Nancy
Walton for the exceptional care taken in typing the manuscript.

iv
 Contents

Preface .................................................................................. iii

Acknowledgment .......................................................................... iv

Chapter 1. Introduction ................................................................... 1

Chapter 2. Basic Physics of Sound .......................................................... 3

Chapter 3. Measurement Techniques ........................................................ 13

Chapter 4. Instrumentation for Sound and Vibration Measurement .............................. 3l

Chapter 5. Noise and Vibration Control Materials ............................................ 43

Chapter 6. Noise Control Procedures ....................................................... 71

Chapter 7. Specific Noise Sources and Solutions .............................................. 93

Appendix. OSHA Noise Regulations and Assessment of Hearing Impairment ..................... 127

Glossary ................................................................................ 129

Bibliography ............................................................................ 135

 CHAPTER 1

Introduction

Noise is commonly described as unwanted cost of $8 billion would be required to achieve

sound. This definition implies a strong subjective engineering compliance with a standard of 85
element in any assessment of the effects of noise, dB(A).
since what may be tolerated by one person may be An aspect of industrial noise that is not
intolerable to another. Although noise is by no considered in detail here is the effect of noise
means a recent phenomenon, it is only with the levels outside the work place on the community
rapid industrialization of the last century and the and environment. The Noise Control Act of 1972
concomitant accelerated development of powerful governs such noise levels and provides for regional
high-speed machinery that it has become a pollu- enforcement through state and local noise ordi-
tant of major concern. The effects of noise expo- nances. 5 One of the main provisions of the Noise
sure on humans range from disturbance or annoy- Control Act is the development of Federal noise
ance to temporary or even irreversible deafness. emission standards for major noise sources in the
Conventionally, industrial noise control is categories of construction, transportation, motors
aimed at alleviating conditions likely to lead to and engines, and electrical or electronic equipment
hearing impairment. In 1969 the first federal regu- in commerce. The EPA is responsible for enforc-
lation limiting civilian occupational noise exposure ing this act and is required to develop standards
was promulgated under the Walsh-Healey Public and suitable labeling of these noise sources, as well
Contracts Act of 1936. Initially these limits ap- as for such noise control devices as hearing pro-
plied only to certain government suppliers, but tectors. The act, together with the extensive provi-
they were soon extended to all industries by regu- sions of the Quiet Communities Act of 1978, gives
lations under the Occupational Safety and Health EPA a charter to conduct and finance noise con-
Act of 1970. These standards remain the most per- trol research. The bibliography at the end of this
tinent to industrial occupational noise control and book includes several important reports published
are given in detail in the appendix. as a result of such work. The EPA is also em-
Currently, 90 dB(A) is the maximum allowa- powered to act as the chief coordinator of all
ble continuous noise level that can occur through- Federal agency noise control programs. The EPA
out an 8-hour day. At higher levels some form of has proposed that community noise levels be
ameliorating action must be taken. There has been reduced to Ldn = 65 dB(A) as soon as possible 2
considerable pressure from organized labor, and will enlist the help of state and local govern-
NIOSH, I and the Environmental Protection ments toward this objective.
Agency 2 to reduce this limit to 85 dB(A) or less, Partly as a result of research authorized by
but serious questions concerning the benefits and the Noise Control Act, increased knowledge
costs of such a reduction are unresolved. concerning the nonauditory effects of noise is
It is estimated that some 3.5 million Ameri- being sought. These effects may arise from non-
cans now work in environments in which the adaptive physiological responses to noise as a non-
OSHA noise standards are exceeded, and that 13.5 specific biological stressor. 6, 7 Some of the more
million risk some form of hearing loss. Even with important responses may be increased cardiovas-
a stricter standard than that currently in force, a cular disease, elevated blood cholesterol levels,
proportion of workers will be at risk because of increased adrenal hormone production, blood
variations in individual susceptibility to hearing vessel constriction, and accelerated heartbeat. 7
damage. One study 4 estimates that a total capital Although no firm scientific consensus has
 emerged on thepossiblelink between noiseand Althoughit is not explicitlystated,manyof
theseconditions,the potentialbenefitsof noise thereferences,
especiallythoselistedat theendof
reductionarecertainlynot confinedto thereduc- thelaterchapters,resulted
fromNASA-sponsored
tionof hearingdamage. research.Thisenablessuchmaterialto beplaced
Chapters2-4area comprehensive introduc- withina relevantandpropercontext.
tion to the basicphysicalprinciples,measuring
techniques, and instrumentation associatedwith REFERENCES
generalpurposenoisecontrol.Chapter3 includes
anoutlineshowinghowbesttoidentifyandchar- 1. NIOSH, Criteria for a Recommended Stand-
acterizea noiseproblemsothatsubsequent work ard--Occupational Exposure to Noise. HSM
mayprovidethemostefficientandcost-effective 73-11001, 1972.
solution.A detailedmethodology for choosing 2. Office of Noise Abatement and Control, EPA,
appropriatenoisecontrolmaterials andtheproper Toward a National Strategy for Noise Control,
implementation of controlprocedures beginsin April 1977.
chapter5.ThemostsignificantNASA-sponsored 3. Office of Noise Abatement and Control, EPA,
contributions to the state-of-the-art
development Noise Technology Research Needs and the Rela-
of optimumnoisecontroltechnologies arecon- tive Roles of the Government and the Private
tainedin the finalchapter.Of particularinterest Sector, EPA 550/9-79-311, May 1979.
are casesin which aeroacoustics and related 4. Bruce, R. D., et al., Economic Impact Analysis
research haveshedsomelightonwaysof reducing of Proposed Noise Control Regulation, BBN
noisegeneration at itssource. Report No. 3246, April 1976.
Equationsand associated mathematics are 5. Haag, F. G., State Noise Restrictions: 1979,
presented in asimple,comprehensible manner. To Sound and Vibration, December 1979, p. 16.
aidthereader,examples arepresentedatpertinent 6. Elkins, C., Noise: The Invisible Pollutant,
pointsin thetext.In addition,thereisa substan- EPA Journal, October 1979.
tial categorization
of thetextanda largeselection 7. Center for Policy Alternatives, MIT (prepared
of figuresillustratingparticularitemsof equip- for EPA), Some Considerations in Choosing
ment, nomograms,and plots of performance an Occupational Noise Exposure Regulation.
characteristics. EPA 550/9-76-007, February 1976.
 CHAPTER 2

Basic Physics of Sound

Sound is a mechanical disturbance of a wave As an illustration of the magnitude of the

nature that propagates through any elastic medium fluctuations involved, atmospheric pressure is
at a speed characteristic of that medium. When normally 1.012 x 10 5 N/m 2 at sea level; a barely
detected by the human ear, the presence of sound audible signal might fluctuate as little as 10 5
waves is often described by the subjective attribute N/m 2 from this value, and a painfully loud contin-
of loudness, and this quality is usually quantified uous noise only about 10 N/m 2, that is, only
by averaging over a range of people (see section 0.01°70 of the atmospheric value. (The unit of
3.1.3). Loudness is related to the amplitude of newtons per square meter, N/m 2, is sometimes
movement in the transmitting medium next to the replaced by its equivalent, the pascal, Pa.)
eardrum.
Sound may be assessed after measurement of
an appropriate physical quantity that varies about 2.1 SIMPLE WAVE NATURE OF SOUND
an equilibrium position as the sound wave or
vibration passes the measuring point. Sound pres- 2.1.1 Introduction to Longitudinal Propagation
sure is the macroscopic physical quantity of most
interest in fluids and gases. It depends on the Wave motion or vibration is made possible by
amplitude of the disturbance in a material and inertia and elasticity in a material. Because of iner-
most adequately represents the effect of acoustic tia, matter remains at rest or in uniform motion in
waves on human response. Figure 2.1 illustrates the absence of external forces; this permits the
how an acoustic pressure disturbance might vary transfer of momentum between neighboring parti-
about the ambient value for a typical sound wave. cles or elements in a medium. Elasticity designates
Atmospheric pressure is taken as the base line for the tendency of displaced elements to return to
airborne sound propagation. their original or equilibrium position after the pas-
sage of a disturbance. Air, water, glass, and such
metals as steel and iron are relatively elastic.
Wave propagation may be demonstrated by
the use of a long helical spring fixed at one end. If
a disturbance is imparted at the free end, the
motion is carried through the adjacent coils, while
at the same time the initially displaced coils tend to
return to their former positions. The net effect is
O-
to produce two distinct regions that travel down
the coil structure: in one, a series of neighboring
coils are pushed close together because of the ini-
tial action of the disturbance (compression); in the
Time, t is) other, the relative spacing of coils is enlarged be-
cause of the removal of the disturbing force and
Figure 2.1 - Example of acoustic pressure fluc- subsequent overcompensation by the elastic coils
tuations about mean ambient pressure. Po is the (rarefaction).
ambient atmospheric pressure," p is the acoustic The wave nature of sound in air may also be
pressure; and PT is the total pressure (equal to demonstrated by the device presented in figure
Po + PT)- 2.2.
O ..... r Equivalent of closely approximate longitudinal plane progressive
sciilarlng . l

piston \ FComp resslOn /infinitely long waves.

\ / ! termination
Wavelength
\ / V_ _ , ;x /(noreflect,ons)
2.1.2 Frequency

The representation of pressure change at a

_- Rarefaction point in space with time, p(t), or at a fixed instant
otating disc
of time throughout space, p(x), as shown in figure
Figure 2.2 - Generation of longitudinal progres- 2.3, is described by the equations
sive waves in a tube.
p(t) = Po sin 27rft
The action of the piston in figure 2.2 alter- p(x) = Po sin 2rfx
nately pushes the air molecules together or pulls
them apart. It is the inherent elasticity of air, com- where
bined with inertia, that allows these successive dis- Po = equilibrium or ambient sound pressure
turbances to travel down the tube as the displaced (N/m z)
molecules try to regain their former equilibrium. f = frequency of the disturbance (Hz)
The net effect on the pressure distribution is maxi- x = distance (m)
mum pressure amplitude at points of maximum t = time (s)
compression and minimum pressure at points of
maximum rarefaction. If the disc rotates at a con- A complete cycle of sound pressure oscillation
stant speed, then the wave form, as represented by is normally represented by 360 ° or 27r radians. The
the pressure, will be sinusoidal. Figure 2.3 shows symbol commonly used to denote frequency is f,
some of the characteristics of this simple wave which is the number of complete cycles in a second.
form, termed a longitudinal plane progressive wave The fundamental unit of measurement is cycles
because the elements of the medium are displaced per second or Hertz (Hz). Sometimes the term
in the direction of propagation and the wave is not "period" is invoked to describe the nature of the
stationary in space. Waves that propagate in many wave. This is simply the time taken to complete
duct noise control problems or at large distances one cycle, T; it is equal to the reciprocal of the
in free space from a spherically radiating source frequency.
The normal frequency range of adult hearing
extends from about 20 to 20 000 Hz. Because of its
Pa Wavelength, ;x innate characteristics, the human ear is most sensi-
+ f_,----' P,ms .,_ ",
tive to sound around 3 kHz, especially at the thres-
0/ Y""/ \ F hold of audibility. The subjective attribute known
Q--I _ _j/ Distance, x as pitch, which is often used to describe musical
{a) Spatial-varying characteristic quality, is highly correlated with frequency. The
sound waves normally associated with noise in in-
i Period, T
dustry are very rarely single frequencies in isola-
tion (also known as pure tones). In general, there
are two classes of sounds: those which repeat at
x._J 'oo' regular intervals (periodic)
of a variety of related
and most often consist
frequencies occurring
(b) Time-varying characteristic
together, and those which fluctuate randomly and
I i [ I I do not repeat (aperiodic). The latter sounds may
0° 180 ° 360 ° 180 ° 3600 be considered part of an infinitely long period.
A mathematical technique known as Fourier
where Prms = root mean square sound pressure
analysis allows any periodic signal to be partitioned
Pay = average sound pressure during positive
half-cycle into a series of harmonically related pure tone
Pa = maximum sound pressure amplitude signals. This enables any complex but regular
waveform to be described in terms of a number of
Figure2.3- Characteristics of simple acoustic discrete frequencies, each with a particular
sine waves. amplitude. A pictorial representation of this is
givenin figure2.4a,showinghowthreesuperim- spectrum simplydescribes therelativeamplitude
posedsinusoidal waveforms
canbeusedto pro- of eachsinglefrequencycomponent;theseare
duceamorecomplex waveform.
Thepresence ofa shownin figure2.4b,corresponding tothevarious
harmonic relationship
meansthatthefrequencyof waveforms in figure2.4a.
eachwaveisrelatedto thefundamental or lowest In contrast,aperiodicsoundscanonly be
frequencyby an integermultiple.A frequency treatedasconsistingof aninfinitelylargenumber

3
-2 Pn

3p.

=-- _r. 3ft

=_ ---- =

0_

Pa P2(t) = 3p a sin 2r • ft

Time, t (s)

VV VVV VkF
P3 It) = Pa sin 2_r • 9 ft

Time, t(s)

(a)

" I I I i
0.

Frequency, f(Hz)

Frequency, f(Hz)

(b)

Figure 2.4 - Complex periodic waves. (a) Formation from simple pure tone components of a complex
wave. (b) Corresponding frequency spectra. Pa/s the ordinate scale division of sound pressure amplitude,
of pure tones spaced apart only infinitesimally and r is the constant dependent on gas involved
with differing amplitudes. The spectrum in this and equal to the ratio of the gas constant
case is usually considered to be broadband. and molecular weight of the gas, J/kg.K
Examples of periodic sources and broadband T is the absolute temperature, K.
sources are, respectively, rotating machinery such
as turbines, fans, etc., and aerodynamic noise The velocity is directly proportional to the
associated with jets or flow past a solid surface. In square root of the absolute temperature, the other
fact, the two types often occur together with a terms being constant. Pressure changes and non-
series of pure tone peaks above a broadband spec- uniformity of gas composition are of much less
trum. An example is where a large machine with significance in affecting the velocity. 1 For air at
several random sources has a strong periodic mode atmospheric pressure, the relationship simplifies to
of operation. For more specific detail on the fre-
quency spectra and appropriate analysis required c = 20.05x/T m/s
for different kinds of sources see section 3.3.3.
More complex relationships exist for liquids and
2.1.3 Wavelength solids 1:

A wavelength is defined as the distance be-

tween analogous points on successive waves of a Liquids: c -"4 00
pure tone signal; it is most easily assessed by meas-
uring the distance
minima
between
on the pressure
adjacent maxima
spatial distribution
or
(see (large cross
Solids:

section) c
_/
xt
Bb + _Bs

_0
figure 2.3a, above). Wavelength is denoted by the
symbol X and is numerically equal to the ratio of Solids: c = B/Bv
the speed of sound in a medium to the frequency: (small cross section) xte0

_. = elf m where
Bbt is the isothermal bulk modulus, N/m 2
where
0o is the density, kg/m 3
c is the speed of sound, m/s B b is the bulk modulus, N/m 2
f is the frequency, Hz.
B s is the shear modulus, N/m 2
By is the stiffness (Young's) modulus, N/m 2.
2.1.4 Speed

These equations apply for longitudinal waves

For very high amplitude waves, which occur only. For transverse waves in solid bars and
either when the ambient speed of sound is exceeded plates, where flexural motion involves the dis-
(see section 7.1.1) or with strong source radiation placement of the structural elements perpen-
as in certain severe impact situations, shock forma- dicular to the direction of propagation, the
tion occurs, and the waveform is distorted into non- velocity is also dependent on the frequency of the
linear behavior. Essentially, this means that sounds waves and is therefore not simply described. This
of different frequencies propagate at different gives rise to the phenomenon of coincidence,
speeds through the transmitting medium. How- important for its deleterious effect on sound
ever, for most commonplace situations in industry transmission properties (see section 5.2.1.2). Note
this will not be so, and the speed of sound can be that all the above equations indicate an inversely
regarded as being the same for all frequencies. proportional relationship of velocity to density.
For propagation in a gas, the general expres- Velocity often increases with density because for
sion for the velocity of sound waves, c, is given by many materials it is accompanied by a concom-
itant increase in elasticity, represented by the
c = _fTrT m/s various moduli concerned.
Table 2.1 gives examples of sound velocity in
where several materials and transmission media fre-
7 is the ratio of specific heat of gas at constant quently encountered in noise control. All values
pressure to that at constant volume are quoted for 20 ° C unless otherwise noted.
Table2.1- Soundvelocities
for somematerials of the sound pressure. It is achieved by squaring
commonin noisecontrol. the instantaneous pressure values over a specified
period of time and then taking the square root. 2
Material Additional Sound The process of squaring converts all the negative
factors velocity
(m/s) pressure values into positive quantities, which can
Air -- 343 then be summed with all the squared positive
Steam 100°C 405 values. An additional advantage is that energy is
Water -- 1408 related directly to the square of pressure. The
Aluminum Bar 5150 process of finding the root mean square pressure
Concrete Bulk 3100
Steel Bar 5050 Prms is represented by
Steel Bulk 6100
Lead
Glass
Wood (soft)
Wood (hard)
Bulk
Bar
Bulk
Bulk
2050
5200
3500
4000 where
Prms =
[ 0J"1" p(t) 2 dt N/m 2

p(t) is the instantaneous sound pressure

fdt is the integration over infinitesimal inter-
2.1.5 Spherical Wave Propagation vals of time, dt
T is the time period concerned
Consider a balloon expanding and contracting
in phase at all points on its surface about some For simple sinusoidal waveforms (see figure
equilibrium size. The speed of propagation of 2.3), the RMS pressure is related to the maximum
sound waves away from the surface is the same as peak pressure Pa by
for a plane wave in a tube. In the latter situation
the sound pressure along the tube is independent Pa
of distance (assuming no frictional losses), whereas Prms = --;w N/m 2
x/2
in the case of a balloon it is dependent on the radial
distance, r, of the wavefront from the center of
the balloon. This is because the area of the wave- It is important to note that an RMS measure
front increases with r and so, as shown in section retains a constant value regardless of the relative
2.2.3, the intensity must decrease (see also figure phases of constituent pure tones in a complex
3.10). wave. All future references in the text to pressure
will implicitly refer to RMS sound pressure unless
otherwise stated.
2.2 PHYSICAL QUANTITIES AND UNITS
2.2.2 Sound Intensity
2.2.1 Root Mean Square (RMS) Sound Pressure
Intensity is defined as the average rate of flow
For sounds which are not pure tone, it is not ot" energy through a unit area normal to the direc-
sufficient to characterize the associated pressure tion of propagation; it is normally measured in
fluctuations with a simple measurement of the watts per square meter, W/m 2. This quantity is
maximum amplitude attained. This is because the often used in acoustics, since many problems are
waveform is not simple, and maxima and minima associated with identifying sound or vibrational
are not necessarily of equal amplitude. Since most energy transfer between component spaces and
common sounds consist of a series of rapid positive materials within a system. The principle of energy
and negative pressure displacements about the conservation allows the paths of transmission and
equilibrium pressure value, straightforward com- dissipation to be tracked. For free plane waves,
putation of the arithmetic mean value would lie the average intensity measured is equal to the time-
very close to zero. averaged product of pressure, p, and particle
Ideally, a measurement of sound pressure velocity, U, as measured in the direction of
should take account of the magnitude of excur- interest:
sions from the equilibrium value on both sides,
p2 s
especially for irregular sounds. This measurement
I = pU - W/m 2
should also represent some kind of time-average pc
2.2.3 Sound Power whose phase relationship with reference to any
particular point is always changing. This can occur
The rate of flow of energy that is transmitted in ducts and at low frequencies in enclosed spaces.
through a medium surrounding a source generat- The total energy density consists of two parts: one
ing sound or vibrational energy is termed power. associated with the kinetic energy of particles con-
When there are no losses in the medium itself, tained within a unit volume, and another associ-
then any surface which completely encloses the ated with the potential energy of the gas within
source must have all the power passing through it. that volume. The expression for energy density at
An analogy with heat may be drawn here. Although a particular point is
a radiator or flame may have a certain thermal
p2
energy measured in watts which characterizes that D(x,t) = QU 2 +-- W.s/m 3
particular heat source, the perception of the pc 2
effects of energy radiating from that source is in
terms of temperature, which is partly a function of
where
the surrounding medium. So it is with sound, ex-
U is the rms particle velocity, m/s
cept that its effects are characterized in terms of
p is the rms acoustic pressure, N/m 2
pressure or intensity.
0c is the characteristic impedance of air
As the hypothetical surface enclosing a sound
(=415 N.s/m 3)
power source is made larger, less power per unit
c is the velocity of sound in air ( = 343 m/s).
area or intensity passes through any portion of the
surface (see figure 3.10). The total sound power
measured in watts (W) is equal to the product of
the intensity and the area of the surface; this is When the pressure and velocity are space-averaged
this reduces to
assessed by summation of these products over
optimal-sized elements of the surface chosen. _2
(Refer to sections 3.4.2 and 3.4.4 for a discussion D (t) - W.s/m 3 (or J/m 3)
Qc2
of the approach used for directive sources.) Thus

W = fl s dS W where
S d p is the space-averaged rms acoustic pressure,
where N/m 2.
S is the area of enclosing surface, m2
I s is the intensity through each elemental area
dS, W/m 2.
2.2.5 Acoustic Impedances

With spherical waves in free space, the area

Acoustic impedances are usually expressed as
of any enclosing sphere is 47rr 2, and hence intensity
the complex ratio of sound pressure to particle
is inversely proportional to the distance from the
velocity at a particular location, usually the sur-
source, r, according to
face of an acoustic absorber. Thus

W = 1-47rr 2 W (for omnidirectional P

Z =-- N.s/m 3 (mks rayls)
sources only) U

In the far field of a spherically radiating

where
source (see figure 3.11), the relationship of sound
p is the complex acoustic pressure, N/m 2
intensity and pressure is the same as that given in
section 2.2.2. U is the complex particle velocity, m/s
Z is the specific acoustic impedance,
N.s/m 3.
2.2.4 Energy Density
Complex impedances may also be represented
Energy density is significant where the sound in the form
field is composed mainly of standing waves (see
section 2.3.3), rather than free progressive waves, Z=R+iX
where waves experience a change in their direction of
R is the resistance (or real) component propagation after encountering an obstacle; it is
X is the reactance (or imaginary) component equivalent to a bending effect. For most general
i is the x/l. noise control purposes, if one or more major
dimensions of an intervening object are of the
same order of magnitude or larger than the wave-
Impedances are of particular importance be-
lengths associated with the frequencies of interest,
cause they relate directly to the ease with which a
then the object will act as a barrier to the sound.
sound wave or vibration may be transferred be-
Otherwise, diffraction will occur around the edges
tween two media. The resistance represents the
of the object as illustrated in figure 2.5. The effect
relative magnitude of an acoustic wave which
typically means that high-frequency waves tend to
passes through a medium interface, whereas the
be much more directional in character; their path
reactance takes into account any phase changes
of propagation can often be successfully traced
which occur. When dealing with porous acoustic
using a geometric ray approximation. Diffraction
materials, the above definition is refined some-
is particularly important when examining the in-
what because of the variety of angles which the
fluence of microphones and barriers on sound
alternating fluid particle velocity may assume)
fields (see sections 4.1.4 and 6.2.1.1, respectively).
The normal specific acoustic impedance, Z n (see
section 5.1.1), is defined using the normal compo-
nent (perpendicular to air-material boundary) of
the particle velocity, Un; thus

For free progressive

waves at a large distance
tionship between
Zn -

sound
P
Un
N- s/m 3

plane waves, or spherical

from the source, the rela-
pressure and particle
i Acoustic barrier

velocity is expressed as (a)

P
Qc =_- N.s/m 3

where

.//)
e is the density of air, kg/m 3 _ Diffraction
c is the speed of sound propagation in air,
m/s. around barrier
due to effective
radiation from
The term Qc is sometimes known as the line source on
characteristic impedance; it occurs frequently in Source
top of barrier
acoustic equations. Under normal atmospheric Acoustic barrier

temperature and pressure, 20 ° C and 1.013 × 105

N/m 2, its value is 415 N. s/m 3. (b)

Figure 2.5- Diffraction effects of a solid ob-

stacle. (a) High frequency. (b) Low frequency.
2.3 ADDITIONAL FEATURES ASSOCIATED
WITH THE WAVE NATURE OF SOUND
2.3.2 Superposition
2.3.1 Diffraction
The principle of superposition, which applies
Sound waves radiating from a source are re- to waveforms with amplitudes typical of industrial
flected or diffracted by objects in their path in situations, states that the net result of adding two
much the same way as waves on the surface of waves is to sum their effects arithmetically. In
water. Diffraction is the process by which sound figure 2.6 this is illustrated for two sets of waves
 09

t3-

Time, t
or distance, x

Net amplitude effectively Complete cancellation of

doubled at all points net sound wave amplitude

(a) (b)

Figure 2.6 -Addition of two identical sine waves, with (a) 0 ° and (b) 180 ° phase differences.

which are identical in all respects except that their (antinodes). At intermediate points, the amplitude
phase differences are respectively 0 ° and 180 °. varies sinusoidally with a fluctuating range
This result flows from the addition of n linear described by the envelope in figure 2.7 at a rate
equations: equal to the frequency of the contributing waves.
Because of the fixed pattern of amplitudes in
p(x,t) = pl(x,t) + p2(x,t) + • .. + pn(x,t) space, the resultant waveform is called a standing
wave (figure 2.7).
2.3.3 Standing Waves or Resonances Standing waves most commonly occur in any
regular enclosure which possesses at least one set
When a free progressive wave encounters a of parallel walls. The criterion for the frequencies
sharp change in impedance such as at an air-hard- at which they may become established is related to
wall boundary, and if the angle of incidence is the separation distance between parallel surfaces
normal, then the reflected wave travels back along by
the original path of propagation with a 360 °
nc
change in phase. If there is another boundary Hz
parallel to the first one, then the reflection process 2d
occurs again, and so on ad infinitum, until all the
acoustic energy has dissipated by absorption at the where
boundaries and by propagation losses in the inter- n is the 1,2,3...
vening medium. Standing waves or resonances c is the velocity of sound propagation, m/s
occur when the distance between the boundaries d is the separation distance, m
coincides with an integer multiple of the half- fn is the nth harmonic frequency, Hz.
wavelength; they are more likely to appear when
the original complex waveform has strong discrete Note that the term n is included because
components. harmonics of standing waves may occur at integer
By superposition, two waves 360 ° out of multiples of the fundamental.
phase add constructively to produce a waveform This phenomenon is of particular importance
greater in amplitude than the original constituents. in the transmission of vibrational energy through
In this unique case, however, because a complete plates and bars, as well as in airborne sound trans-
reversal of direction has also taken place, fixed mission through limited spaces of regular geome-
positions are created of zero net pressure ampli- try. In both situations, reflection relatively free of
tude (nodes) and of maximum pressure amplitude energy loss at the metal or room boundaries may

10
 Oscillation within waveform An acoustical source having a sound power of
envelope at rate equal to W watts is therefore said to have a sound power
Antinode
fundamental frequency
level PWL given by

/__[1_ N°de -_ PWL = 10 log 14";ref dB

• /

HJJUUuu
., ' ul
where
Wre f is the reference power value normally
kt4
taken as 10- 12 W, according to interna-
X tional convention.

Figure 2.7 - Standing wave between two parallel

surfaces one wavelength apart. For example, if W = 1.3 x 10-5 W, then
PWL = 101og(l.3 x 10-5/10 12)
= 10 log 1.3 x 107
occur, with a phase change of approximately 360 °.
= 10 x 7.11 = 71.1 dBre
In practice, therefore, particular structures and
10- ]2 W
spaces may tend to amplify acoustic energy at
those frequencies where standing waves occur.
Because we have taken a ratio, decibels are
Therefore, when designing noise control proce-
expressed in dimensionless units. It is important,
dures, for example, enclosures, particular atten-
however, always to state clearly the associated ref-
tion must be paid to preventing standing wave for-
erence to avoid confusion with possible alterna-
mation at frequencies given by the above formula,
tives. Sound intensity level IL is defined in a way
to prevent undue degradation in predicted attenu-
identical to sound power level using a referencce
ation performance. An alternative description
intensity of 10 12 W/m 2. Thus
used for resonance in solid materials is normal
modes of vibration.
IL = 10 log l/lre f dB

Sound pressure level, SPL, is defined by the

2.4 DECIBELS
relation

The parameters typically used to describe

sound have very wide ranges. Frequencies may
change by an order of 10 4, sound power or inten-
,o,o
sity by 1016 , and sound pressure by 10 8. To sim-
where
plify the graphical and numerical expression of
these quantities, they are represented by means of Pref is the reference pressure value, taken as
2 × 10- 5 N/m 2
a logarithmic transformation. In the case of sound
power, intensity, and pressure, this is achieved
or
using the decibel scale.
P
Essentially, decibels are calculated by taking SPL : 20 log_--
the logarithm of the ratio of the value of the Pref
parameter of interest to a fixed reference value of
that parameter, and multiplying it by 10. The Since most noise control problems involve a
advantages of this are threefold: it avoids large calculation of airborne sound propagation, the
exponents in representing the quantities; the reference pressure, Pref' iS optimized so that sound
human ear judges sounds according to the ratio of pressure level and sound intensity level are numer-
intensities, so that the subjective effect is reasona- ically equal for measurements made in the atmos-
bly approximated; and electrical engineers use phere. We have already presented an equation
decibels in evaluating electrical power and voltage, relating sound intensity to sound pressure under
so that conversion of quantities and units at elec- certain circumstances (see section 2.2.2):
troacoustic transducers is simplified. By conven-
p2
tion, parameters are taken as "levels" when their I = - W/m 2
units are expressed in decibels. Oc

11
and, Sound power level PWL re 10-12 W (dB)

160 -- Jet engine

10logI/Ire f = 10 log Pref

Since the reference intensity value equals

10- 12 W/m 2, therefore
140

(Pref) 2 = 10-12 × Oc Threshold of pare

= 10-[2 × 415
(0c=415 N-s/m3; see section 2.2.5) 120 Chipping hammer

Hence Punch press

Pref = 2.04 × 10 -5 N/m 2 IO0 Ventilating fan (centrifugal)

To standardize, this is rounded to give Ventilating fan (axial)

Printing plant

Pref = 2 × 10 -5 N/m 2 80

Note that this unique equivalence between

sound pressure level and sound intensity level ob-
6O Voice -- normal speech
tains only for airborne sound; in other fluids or
solids different reference values are used. For
techniques of combining quantities expressed in
decibel values refer to section 3.2. The relationship
4O
of sound pressure and sound intensity to decibels
for airborne propagation is shown in table 2.2.
Voice -- whisper
Figure 2.8 illustrates sound power levels of
some typical noise sources.
20

Table 2.2 - Relationship between decibels, sound

Threshold of hearing
intensity, and sound pressure for atmospheric
propagation.
Figure 2.8 - Sound powers for a variety of sources.
11 re 10-12 W/m2
Sound intensity, Sound pressure, (dB) or
I (W/m E) p (N/m2) SPL re 2 x 10- 5 REFERENCES
N/mE (dB)

104 2 × 103 160 1. Kinsler, L. E., and A. R. Frey, Fundamentals

l02 2 x 102 140 of Acoustics, 2d ed., John Wiley & Sons, 1962.
10° 2 × 10 120 2. Maling, George C., Jr., and Williams W. Lang,
10 -2 2 × 10° 100
Theory into Practice: A Physicist's Helpful
10 -4 2 × 10 -1 80
10 -6 2 x 10 -2 60 View of Noise Phenomena, Noise Con. Eng.
10 -8 2 × 10 -3 40 vol. 1, p. 46, 1973.
10 lo 2 x 10 -4 20 3. Beranek, Leo L., Noise and Vibration Control,
McGraw-Hill Book Co., 1971.

12
 CHAPTER 3

Measurement Techniques

Sound has temporal, spectral, amplitude, and 8-10 dB, across the frequency range of most inter-
spatial characteristics; because the process by which est in noise control, 300 Hz to 4 kHz. This allows
it interacts with human physiology and behavior is the generalization that a reduction of approxi-
by no means well specified, different indices and mately 10 dB is needed to reduce loudness by one-
scales exist. This chapter initially treats in some half, since a change of 10 phons represents a doub-
detail the most important parameters and measure- ling or halving of loudness. In typical working
ment approaches used for general-purpose indus- conditions, however, an observer is rarely able to
trial noise control. Section 3.4 contains a full dis- detect a change of less than 3 dB in levels. Espe-
cussion of alternative means of identifying noise cially prominent is the decrease in sensitivity at all
sources, and section 3.5 presents the overall ap- levels for sound at the lower frequencies. This
proach for a noise survey. These sections are pre- means, for example, that a sound of 80 dB at 30 Hz
cursors to chapter 6, which examines general noise produces the same sensation of loudness as a tone
control procedures that are applied after the prob- of 42 dB at 4 kHz.
lem is defined. A more analytical approach has been taken
by both Zwicker and Stevens, 2 whose models are
predicted on the observation that the ear appears
3.1 ACOUSTIC PARAMETERS OF SPECIAL to act as a set of filters normally around one-third
RELEVANCE octave in bandwidth (see section 3.3.1 for defini-
tion). Their basic contention is that subjective
3.1.1 Loudness Evaluation loudness should be obtainable by summing the
sound pressure in these bands after correcting for
It has been indicated in section 2.0 that loud- the sensitivity of the ear. This varies with frequency
ness is an important subjective perceptual charac- and level. The Zwicker and Stevens MK. VI tech-
teristic of interest. There are several approaches to
quantifying loudness; usually the results are aver-
aged over a number of people, since individual Loudness level (phons)
140
responses can vary considerably.
Figure 3.1 shows equal loudness contours, 120
which represent the variation in actual sound pres- _,_ ._ _ 110 ="'_//1 "_
loo
sure level with frequency necessary to produce an
equivalent sensation of judged loudness by the lis- _ 80
tener. These curves are based on pure tone stimuli
x 60
and are characterized by assigning them a phon =SOi
....... "--.._/'%,
number, which corresponds to their sound pressure = 40

level at 1 kHz. Thus, for example, a sound of

m 20
103 dB at I00 Hz represents a loudness value of
of hearing-- " -_.:
100 phons.
[ I [ "-
Several interesting features appear in these 20 63 250 1000 4000 16 000
curves. The dip around 3-4 kHz corresponds to 31.5 125 500 2000 8000

an increase in sensitivity resulting mainly from a

Frequency (Hz)
resonance effect in the ear canal. Also, the spacing
between adjacent curves is roughly constant, at Figure 3.1 - Equal loudness contours (reference 1).

13
niqueshavefoundwidespread useandhavebeen 2O

adopted as international standards. 2 More

recently,Howeshasproposed a model3forcalcu- 10

latingtheloudness of soundsthatarereasonably
steadyovera periodof at leasta secondor so. ._ 0
Thismodelusespowersummation overthecritical o lo
bandsin the earand the conceptof frequency ×

interaction.Computation techniquesfor Zwicker -20

andStevens' phonsmaybecompared byexamin- m -30

ingreference 4, andHoweshasdeveloped a com-
puterprogram 5 for generaldissemination which ¢o
- 40
requiresas inputonly the soundspectrumand rr

modeof listening. - 50
20 50 100 200 500 lk 2k 5k 1Ok 20k
Apart from the common-sense assumption
thata loudsoundtendsto bemoredamaging in Frequency (Hz)

regardto hearing loss,it isalsoreasonable to sup- Figure 3.2 - Frequency response of weighting
posethat,otherthingsbeing equal, a louder sound scales. The straight line through 0 dB represents
tends to be more disturbing. Various researchers the unweighted response normally used when as-
have tried to characterize sound in terms of annoy- sessing linear overall sound pressure level,
ance or noisiness ratings, which presumably repre- OASPL.
sent a more comprehensive subjective evaluation
than loudness per se. Reference 6 contains an early Table 3.1 - A- and C-weighting corrections for
description of such a procedure as well as a com- octave band data.
parison with loudness calculation techniques.
Octave Relative Relative
3.1.2 Weighting Scales band center A-weighted C-weighted
frequency (Hz) response (dB) response (dB)

In order to assess human response, it would 31.5 -39.5 -3.0

be convenient to use a single numerical criterion 63 - 26.0 - 1.0
which can be easily measured. Initially, it was pro- 125 - 16.0 0
posed that this be achieved by inverting the 40-, 250 - 8.5 0
500 - 3.0 0
70-, and 90-phon curves in figure 3.1 and provid-
tO00 0 0
ing electronic filter characteristics in sound level 2000 + l.O 0
meters and other devices which closely approxi- 4000 + 1.0 -1.0
mate these shapes. This permitted integration of 8000 - 1.0 -3.0
sound pressure over the range of interest, provid- 16 000 - 6.5 - 8.5
ing a measure of subjective response to low,
medium, and high level sounds, respectively. The
corresponding filter networks are designated A (40 3.1.3 Other Indices and Scales
phon), B (70 phon), and C (90 phon). The relative
responses of these are illustrated in figure 3.2 As has already been implied in section 3.1.1,
Subsequently, it was found in practice that, sounds may be identified as noise resulting from a
although anomalous, the use of the A-weighting variety of factors. Several techniques for assessing
scale gives results which correlate best with human the effects of noise in specific situations have
response in regard to noise-induced deafness, the evolved, and it is worthwhile to mention a few that
acceptability of noise from transportation and can be relevant to industrial noise problems. A
other machines, and communications interference. fuller description of these and others, including
Measurements on the various scales are denoted applications and calculation procedures, may be
by the symbols dB(A), dB(C), etc.; of course, the found in reference 4.
reference value should be specified where neces- A technique widely used in the United States
sary. Table 3.1 shows the approximate numerical for assessing the relative intrusiveness of back-
value of the two most popular weightings relative ground noise is based on a set of contours known
to linear response, applied in octave bands (see as Noise Criteria curves (see figure 3.3). These are
section 3.3.1). similar to equal loudness curves and have been

14
 Not the least of the reasons for this is the belief
that the amount of sound energy received at the
eardrum is the most critical cause of noise-induced
hearing loss. Leq takes the value of the constant
sound pressure level which contains the same sound
energy as the actual time-varying level of interest.
It is always calculated with respect to a particular
,_ 50 time interval, generally an hour or an 8-hour
period; normally, the sound level is A-weighted.
With digital and microprocessor technology, a
variety of portable instruments can accurately
compute Leq (see section 4.4.1).
The mathematical expression for the compu-
tation of Leq over a specified time interval T is

20 i l fT_
Leq(T) = 10 log T Jo pr2ef dt dB(A)
10
Thre h
for continuousnoise where £ T
0 6J3 I
125 I
250 500 lk 2k 4k 8k dt represents the integration
tion of squared instantaneousor summa-
sound
Octaveband centerfrequencies(Hz) pressure values over the time interval
T
Figure 3.3 - Noise criteria curves (NC).
p is the A-weighted instantaneous
sound pressure

developed after considerable practical experience. Pref is the reference sound pressure.
They have essentially been optimized to provide
an assessment of the effects of noise at particular Community noise can often be a problem
frequencies on speech communications, with some when external noise levels are high. Consideration
regard to the loudness of the sound as well. should be given to this especially when planning a
A Noise Criteria (NC) rating is found by plot- new industrial development or implementing noise
ting the actual octave band sound pressure levels control technology in a situation where community
and identifying the highest NC curve which pene- aggrievance has already been established. A very
trates the sound spectrum from figure 3.3. Certain concise introduction to the subject may be found
criteria, shown in table 3.2, have been recom- in reference 8.
mended for particular environments. 7 One of the indices most frequently used in
The equivalent continuous sound level, Leq, is this connection 9 includes the day-night level, Ldn,
increasingly used in a wide variety of situations. which represents a long-term average 24-hour Leq,
with sound energy received between 10:00 p.m.
and 7:00 a.m. being weighted by a factor of
Table 3.2 - Recommended noise criteria values + 10 dB. The community noise equivalent level,
for several work environments. CNEL, performs a similar function except that a
three-way subdivision between day, evening, and
Probable
Work environment acceptable night is used. Statistical measures such as Lg0, L50,
NC levels and L1, illustrated in figure 4.10, are also widely
used. Lg0, or the level exceeded for 90°70 of the
Factories and workshops (no hear- 60-75 sampling time is often used as a measure of am-
ing damage risk) bient noise.
Factories and workshops (communi- 50-60
cation required)
Kitchens and light workshops 45-55
Mechanized offices 40-50 3.2 LOGARITHMIC ADDITION OF DECIBELS
General offices 35-45
Private offices 30-40 It is frequently necessary to combine several
Cafeterias and canteens 35-45
sound levels in noise control problems; this is done

15
bytakingantilogarithms andsumming withrespect where
to meansquaresoundpressure. Themostimpor- SPL T is the total sound pressure level at the
tantinitialstepistodeterminewhether thesources measuring point.
concerned radiatesoundintensitycoherently, be-
Two levels may be most simply combined by
cause puretonesof thesamefrequency caninter-
use of figure 3.4. The numerical difference between
fereconstructively, asdescribedin section2.3.2,
the levels concerned is first found, and then the
andit is necessary to accountfor thephaserela-
corresponding position on the curve identified.
tionshipbetween the sources,
whichis fixedfor
From this, after alignment with the relevant value
coherent radiation.Theappropriate equationfor
on either the abscissa or ordinate, the total level is
calculating theRMSsoundpressure in thiscaseis
computed by straightforward addition.
Two propositions should be noted regarding
p = xfplpl2 -+ 1_ 2 + 2pip2 COS (01-02) N/m 2
figure 3.4:
where 1. For two levels with the same value, the

Pl, P2 is the RMS sound pressure associated numerical increase between the total and the origi-
with the two sources nal values is 3 dB.

(01-02) is the relative phase difference between 2. For levels more than 10 dB apart, the con-
sources at measuring point. tribution of the lower level is negligible.
Figure 3.4 is plotted according to the relation-
Note that for completely in-phase signals of ship
equal amplitude this has the effect of doubling the
resultant pressure. The total sound pressure level SPL r = SPL l + 10 log (10(SPL2 - SPL_)/10 + 1)
in this case would be raised 6 dB with respect to the
original pressure: where
SPL T is the total level
p -- +12 + pl 2 + 2/912 N/m2 SPLI, SPL 2 indicates that two levels are being
added
= 2Pl

Therefore, For example: If we have a series of levels 46,

2Pl 56, 61, and 61 dB, figure 3.4 allows computation
ASPL = 20 log-- = 6dB of the total as 64.7 dB. Thus
Pj
46 56 61 61
I I I I
For most industrial situations, it is reasonable
56.4 64.0
to assume that the sources radiate sound randomly
I J
with respect to each other (i.e., are incoherent), !

64.7
especially when only weak pure tones are present.
Under these conditions, phase relationships be-
SPL T = 64.7 dB
come unimportant and the squares of the RMS
pressures of the n contributing sources at the point
0
of interest may be added linearly:

p = xfp12 + p22 + p32 + . .- + pn 2 N/m2 i 2 3 two levels being added [dB)

The squares of pressures in the above equa- 1 Numerical

g__ 6 difference between
tions are taken before addition because it is impor-
_ 7 8
tant to observe the principle of conservation of
energy, and energy or intensity is proportional to
the square of sound pressure. In terms of sound
pressure levels this may be expressed as

3 4 5 6 7 8 9 10 11 12 13 14
SPL T = l0 log [10SPLJ l0 + 10SPL2/10 +
Numericaldifference between total and smaller levels(dB)
• '' + 10SPLn/10] dB Figure 3.4 - Combination of two decibel levels.

16
 °l
A chart for the addition of equal levels is pre- Octave band center frequency (Hz)
63 125 250 500 1000 2000
sented in figure 3.5. This is an exercise frequently
SPL, dB 65 75 77 80 83 81 79
required, especially when assessing the contribu- -26 -16 -8.5 -3 0 +1 +1
A-weighting, dB
tions of several identical machines to a reverberant 39 59 68.5 77 83 82 80
SPL (dB(A))
field.
69 84 84
I I

10 = I I I
f
Increase in sound pressure /
(dB or dB(A)) ##
I I / Therefore total sound pressure level is given
A 8
by
_5
/
..-c
/ SPLT(63-4000 Hz) = 87 dB(A)
"o 6
=-
"10 / Reference 10 provides mathematical tech-

/
..J
tl
niques for combining decibel levels of either octave
4
.c_ bands or constant bandwidths (see section 3.3) into
8
an overall spectrum level.

_c 2
/
/ 2 3 4 5 6 7 8910
3.3

3.3.1
FREQUENCY

Description of Octave,
ANALYSIS

One-Third Octave,
Number of independent sources and Narrow Bands

Figure 3.5 - Combination of identical levels.

In order to divide the frequency spectrum con-
(From Lewis H. Bell, Fundamentals of Industrial
veniently into bands which can be modeled by
Noise Control, 1973. Used with the permission of
electronic filters and analyzed independently of
Harmony Pubfications.)
each other, the concept of octaves is utilized. From
an acoustician's point of view, the octave is signif-
Average levels are computed according to the icant because it represents a doubling of frequency.
principle used above; that is, the sum of the asso- Examination of table 3.3 reveals that the preferred
ciated energy quantities is taken, usually pressure octave and one-third octave bands in use are based
squared, before taking the average value and con- on geometric mean center frequencies, fc; thus
verting back into decibels. Thus an average sound
pressure level SPL would be computed for n sepa- L = V7_. xf.+ I nz
rate levels according to
where
fn and fn + I are the lower and upper band
SPL = 10 log I [10SPLt/10 + 10SPL2/10 + limits, respectively.
n

• .. + 10SPL./10] dB
The octave bands of most interest for general-
purpose use extend from 63 to 8000 Hz; any over-
where all sound pressure level computed from measure-
nth sound pressure level to be ments made in a series of contiguous bands should
SPL n is the
added. properly have the range of bands quoted alongside.
One-third octave band filters are the second
Another computation frequently required is most common variety in widespread use; they in-
the addition of sound pressure levels in contiguous crease the degree of resolution obtainable in the
frequency bands to produce an overall sound pres- field without making the process of data collection
sure level. In the example below, the A-weighted too cumbersome (see section 3.3.2 for discussion
overall level is calculated from linear octave band of appropriate filter selection). Because the energy
data using the numerical weightings taken from contained in each one-third octave band is propor-
table 3.1. tionally less than that in the octave band which

17
 Table 3.3 - Octave and one-third octave center frequencies and band limits (reference 11).

Octave band One-third octave band

Lower Center Upper Lower Center Upper

band frequency band band frequency band
limit* (Hz) limit* limit* (Hz) limit*

18 20 22
22 25 28
23 31.5 45 28 31.5 35
35 4O 45
45 50 56
45 63 90 56 63 71
71 80 90
90 100 1t2
90 125 180 112 125 141
141 160 178
178 200 225
180 250 355 225 250 282
282 315 355
355 400 450
355 500 710 450 500 560
560 630 710
710 800 890
710 1000 1400 890 1000 1120
1120 1250 1400
1400 1600 1780
1400 200O 2800 1780 2000 2240
2240 2500 2800
2800 3150 3550
2800 4000 5600 3550 4000 4470
4470 5000 5600
5600 63O0 7080
5600 8000 11 200 7080 8000 8900
8900 10000 11 200
11 200 t2 500 14 130
11 200 16 000 22 400 14 130 16000 17 780
17 780 20000 22400

*Rounded off slightly.

covers that particular range, one-third octave lyzers suitable for deployment in the field, narrow
band spectra always have values below the corre- band analysis is more practical. Nevertheless, the
sponding octave band spectra when plotted on the total spectrum bandwidth may have to be limited
same graph (see figure 3.6). Reference 10 provides to provide the degree of resolution required at low
information on converting acoustical information frequencies particularly, since typically these
from one octave set into another, and computing instruments only have 400 bands over their fre-
one-third octave band levels knowing the octave quency range (see section 4.6.2).
level and the decibel-per-octave slope of the Sometimes it is convenient to measure a
spectrum. parameter known as the power spectral density,
In contrast to octave and fractional octave PSD, which is plotted as a function of signal
bands, which increase in bandwidth by a fixed per- power in a frequency bandwidth divided by that
centage as the center frequency increases, narrow bandwidth, against frequency. When the PSD is
band filters have a constant bandwidth through- computed using a sufficiently high resolution fre-
out the entire spectrum. They may be tuned to quency analysis, then the difference in PSD func-
measure several different bands, typically 10, 5, 2, tions between broadband and discrete spectra
or 1 Hz in width. With the advent of real time ana- becomes sharply evident, as the former have a

18
 80-

70-
v

X 60-

a_

50-
m

40
63 125 250 500 1000 2000 4000 8000
Frequency(Hz)

Figure 3.6 - Comparison of octave, one-third octave, and narrow band spectra for the same sound field
measurement.

characteristically smooth and monotonic nature tion. More detail on the instrumentation available
whereas the latter are very discontinuous. is given in section 4.6.
There are several quick procedures for obtain- The traditional mode used in the field is the
ing approximate spectral content. It is clear from serial filter, where the signal is processed by each
figure 3.2 that the C-weighting curve is effectively bandwidth filter, one at a time, usually over a con-
linear over much of the frequency range of interest tiguous frequency range. In the laboratory, a simi-
in general noise control. Consequently, by taking lar process may be carried out using a sweep filter,
both A- and C-weighted readings and comparing where the bandwidth of the filter expands auto-
them, a rough assessment can be made according matically as the center frequency moves continu-
to the following rules: ously up through the spectrum. The latter technique
involves no discrete filter bands. Either of these
dB(C) > dB(A)- indicates low frequency serial approaches is only strictly applicable when
mainly the input signal is relatively constant in level and
dB(A) > dB(C) -- indicates high frequency frequency content over the period of time taken to
mainly complete the measurement.
dB(A) = dB(C) -- indicates that it is located Less frequently used is the process of feeding
around the 1-2 kHz re- a signal to a set of contiguous and parallel analog
gion, i.e. mid-frequency. filters and then, either separately or in combina-
tion, recording the output of each. This approach
Many frequency filters can now be used in does not usually have the same constraints on sig-
conjunction with weighting networks to provide a nal uniformity as the serial process.
direct readout, in dB(A) for example, in specific Finally, in digital filtering the analog signal is
frequency bands. quantized into discrete portions and subjected to a
computation process which very efficiently per-
3.3.2 Description of Filtering Techniques forms effective frequency filtering. This can be
accomplished quickly and accurately. Reference
Various modes of filtering may be employed 12 provides a brief introduction to the problems in
for either the fixed percentage bandwidth or fixed digital analysis of one-third octave bands with
bandwidth filters described in the previous sec- similar accuracy across the spectrum of interest.

19
In this paper,two differentsamplingratesare frequencies, fn, most likely to dominate according
usedto achieve a higherresolutionin thelower to the relation
frequencybandswithout wastingcomputation
timein higherfrequencybands,therebyachieving fn = nNK Hz
a moreuniformconfidence level.
where
3.3.3 Spectral Character of Typical Industrial N is the shaft rotational speed measured in
Sounds and Analysis Procedure revolutions per second
K is the number of blades, teeth, etc., attached
Only rarely do sounds consist of pure tones in to the rotating component
isolation. Generally, it is necessary to conduct a n is the integer corresponding to the harmonic
spectral analysis of the acoustic or vibration frequencies, i.e., 1, 2 ....
parameter of interest over a series of adjoining
frequency bands using a bandwidth and mode of Because the frequencies of interest are very
analysis appropriate to the problem at hand. The distinct, it is usual to conduct initially at least one-
bandwidth should be chosen to be only as narrow third octave band analysis followed by further
as necessary for the sound spectrum concerned. refinement of the frequency resolution as
This section outlines the general nature of com- required.
mon industrial sounds and the practical methodol-
3.3.3.2 Broadband frequency sources
ogy for conducting routine analyses.
Broadband frequency sources are usually
3.3.3.1 Discrete frequency sources
associated with high velocity gases or fluids vent-
Fans, compressors, gear and mechanical trans- ing to atmosphere. They induce a shearing action
misson elements, and transformers, give rise to in the structure of the ambient air, and the result-
spectra with very prominent discrete frequency ant turbulence generates random pressure fluctua-
tones, as illustrated in figure 3.7a. These usually tions. The spectrum is continuous, lacking any
present a number of harmonic components which particular discrete quality, and with no fixed rela-
may even be larger in magnitude than the funda- tionship between either amplitude or phase com-
mental frequency. This may occur because the ponents. This is not to say, however, that the
sound spectrum reaching the measuring position sound may not be characterized easily with refer-
can depend on intervening structures (such as ence to its frequency distribution. For example, in
pressed metal panels around the machine) which heating ducts the nature of sound at some distance
radiate noise differentially according to the from the air handling equipment is mainly low
inherent sound transmission and damping charac- frequency; it is audible as a rumble. By contrast,
teristics of the complete system. This kind of high speed exhaust vents generate a spectrum
spectrum often emanates from rotating equipment which is predominantly high frequency in charac-
and it is possible to predict the harmonic ter and is best described as a kind of hiss. An

"o

09
_0

I I I | I I I I I | I I
101 10 2 10 3 10 4 10 _ 10 2 10 3 10 4 101 10 2 10 3 10 4

Frequency (Hz) Frequency (Hz) Frequency (Hz)

la) {b) (c)

Figure 3.7 - Illustrations of various kinds of frequency spectra. (a) Discrete. (b) Broadband. (c) Impact.

2O
exampleof this is shownin figure3.7b.Octave particularly in the case of large industrial plants,
bandanalysiswill normallysufficewhenconduct- where several ancillary units such as hydraulic
inganalysesof thistypeof noise. pumps and generators may be switched in con-
currently.
3.3.3.3 Impact sources A second procedure, which is rather lengthy
Impact sources include sources such as punch but nonetheless of considerable value because of
presses, hammering operations, and very short the amount of information which may be extracted,
duration venting operations to atmosphere. The is to conduct a survey of sound pressure levels
character of these may be regarded as aperiodic; around the affected area. This is especially rele-
consequently they have a continuous spectrum vant in the case of shop floors where many sources
(see section 2.1.2). The distinguishing features are operate simultaneously and where, perhaps for
determined primarily by the duration of impact reasons of assembly-line production or constraints
and the elasticity and damping properties of the on the time during which machines may be shut
materials. Basically, the shorter the duration and down, it is necessary to conduct an in situ assess-
the harder the surfaces involve.d, the more likely it ment. Figure 3.8 is an example of such a survey in
is that the resultant sound will have a sharp ring a poultry processing plant. If the contour lines of
because of the high frequency content. This is equal sound pressure level are sufficiently well
illustrated in figure 3.7c. Because of the very short defined, it is possible to estimate at least relative
durations involved in impact noise, which can be sound power levels of contributing sources by
of the order of milliseconds, the meter RMS char- calculating the area around each machine which is
acteristic may not respond fully to the maximum above a particular noise level. The sound power
instantaneous amplitude. For this reason a peak level is proportional to this area (see section 3.4.4).
hold characteristic with a very short integration The survey is of importance, too, in making a quick
time has been built into some modern sound level assessment of the degree of compliance with
meters to gauge the peak levels involved (see sec- OSHA noise exposure regulations (see the
tion 4.4.1). Differences between this meter reading Appendix).
and the normal meter response can be up to 30 dB; In addition, the directivity of the source may
the differences tend to increase as the distance to be estimated if the measurements are made in a
the source is shortened, especially for high fre- direct field (see section 3.4.3). Occasionally, this
quency components. can be done by making measurements very close
to the radiating surfaces or duct openings, 13
although there are often theoretical objections to
3.4 GENERAL ANALYSIS this; however, it is always wise to make follow-up
accelerometer measurements of vibration on the
3.4.1 Source Identification radiating surfaces themselves. These may then be
matched with the sound pressure level measure-
Noise control problems are usually examined ments to verify their value in measuring sound
by separate consideration of the source, propaga- power and directivity.
tion path, and receiver; an optimal solution based NASA programs on jet noise sources and
on an overall assessment of the results of this tri- aerodynamic noise generation in wind tunnels
partite analysis is then designed (see section 6.0). have yielded several devices designed to determine
Of these three considerations, correct identifica- source position, especially in turbulent flow condi-
tion of the source and its acoustical nature is the tions. One of these 14 uses a linear array of micro-
most important; otherwise, unnecessary expendi- phones with a digital time delay applied to each
ture may be incurred because of misdirected con- one in the series according to the acoustic fre-
trol procedures. quency of interest. Unwanted noise containing
One of the simplest and quickest ways of reverberation, background, and wind noise com-
accurately defining a source is to run all the noisy ponents of the order of 3-10 dB is rejected in the
machines connected with a location of interest range 63 Hz to 10 kHz. Another instrument, 15
independently, i.e., one at a time. Individual noise which permits easy analysis in two dimensions
contributions can then be compared in order to without the need for large numbers of micro-
assess the most dominant or troublesome noise phones in two separate array configurations, has a
sources. Several sources may interact, however, directional acuity such that high frequency sources

21
 Gizzard Lung
Circulating-_ /- USDA trim F machine F- gun
/

fan \L_j k /are a /area

Picking
/ , Luochroom I
t.---
area I__" _ i__ / area

) ,--4(....
' _ _ | ', ] _92 '94-- :" , }
Exhaust
fan

,N'- : 'xl ,_"_ is- i i -- P_,.,, •

Conveyor _
portals

,z _"_3_,-"--L_lf
/q
,__d :::::
"'i
/i
_ Pre-chiller
Pre-chiller
IL '
_"_

--, -I1 'cedumPJ

Packing
i;
area
I

Conference
room

Figure 3.8 - Example of sound pressure level contour survey. All data are A-weighted. (Taken from
Cassanova, R.E., et al., Study of Poultry Plant Noise Characteristics and Potential Noise Control Tech-
niques, Progress Report, Oct. 1978-Mar. 1979, NASA Res. Grant NSG 3228.)

can be specified within 1 cm. This is achieved interest. An excellent account of the development
using an ellipsoidal acoustic mirror with a micro- of these and other techniques, as well as their
phone placed at one focus; the intervening space application to industrial noise, is given in refer-
may be filled with a gas heavier than air, to de- ence 13. Coherence and cross-correlation tech-
crease the speed and hence wavelength of sound niques are of particular value in situations where
waves, so that the resolution power of the device is only in situ measurements can be made with all
enhanced. 16 By mounting the mirror and micro- equipment running, and for pinpointing particu-
phone on an automated drive assembly, complete larly troublesome areas on very large sources.
two-dimensional plots may be made of an area, Basically, microphones or vibration transducers
and the frequency-analyzed output highlights sig- are placed at the receiving point of interest and at
nificant noise sources, iv the various suspected contributing sources, and
Another technique for determining important the signals from each compared.
sources is to examine frequency plots made for Cross-correlation is essentially a complete
measurements close to the suspected noise con- comparison of the amplitude of two signals at
tributors, and to compare them with the composite every frequency, as well as a measure of the time
level received at the position of interest. This ap- or phase difference between them. The common
proach is of particular value when prominent dis- parts of the signals combine and the unrelated
crete frquencies are present, although narrow band parts cancel out. By finding the delay time be-
analysis may be necessary to discriminate and tween the signals where the cross-correlation is a
match exact values. maximum, it is possible to specify noise sources
More recently, many data processing tech- and their relative contributions to the sound field
niques have been developed to extract maximum by tracing possible transmission paths. Although
information from signals and to discard extra- reverberant conditions tend to militate against the
neous components. These can be broadly classi- effective use of cross-correlation, Ahtye and co-
fied into either coherence or cross-correlation workers 18,19 have successfully used the technique
techniques, and their value is increasing with the to identify sources as low as 18 dB below the over-
increasing availability of sophisticated instrumen- all level. Furthermore, a measurement device
tation for direct print-out of the parameters of known as a sound separation probe, 2° using simi-

22
lar principles, hasbeendeveloped to separateout This is the same as taking the ratio of the
soundwaves fromturbulentflowpressure fluctua- associated sound pressure levels; in fact, the loga-
tionsin a duct. rithmic equivalent of the directivity factor, known
Coherence measures theextentto whichre- as the directivity index, DI, is defined as
ceivedpoweris dependent on a particulartrans-
mittedor inputpower.It canbeusedto assess the DI = 10 log Qo dB
noisesource characteristics inreverberant
spaces,13
as well as in situationswherecross-correlation = SPL o - SPL s

mightnormallybeutilized.Anexample of thesep- where

arationof contributionsfrom a fan andaero-
SPL 0 is the sound pressure level at angle 0
dynamicturbulence in a NASAwindtunnelusing
coherence techniques isgivenin reference 21.The SPL s is the space-averaged sound pressure
level at the same distance from the
sources involvedshouldbeindependent of each
source (equivalent to assuming omni-
other,22,23andit isevenpossible to assess
relative
directional radiation).
source contributions usinga pairof closelyspaced
receiving transducers at somedistancefromany
source,whenit isnotpossible to mountthemany Where complete spherical radiation is not
closerbecause of environmental or geometrical possible because of the presence of nearby reflect-
constraints.23 ing surfaces, figure 3.9 gives the values of Q and
DI. The directivity index increases in value by 3 dB
3.4.2 Source Directivity for every added reflecting plane. This is a natural
corollary of the same sound energy flowing through
Directivity measures the extent to which the half the surrounding surface area with the addition
sound pressure level in any particular direction of each extra plane, i.e., a doubling of intensity.
deviates from the level which would be present if It may sometimes be necessary to measure the
the source radiated in all directions equally. Direc- directivity of a source simultaneously using a com-
tivity of a source is normally measured in the far plete 360 ° array of microphones. This is often the
field and in the absence of any intervening obsta- case with random sources such as aerodynamic
cles or reflecting surfaces other than those nor- noise generated by fan or compressor blades, or
mally associated with the source itself. It is repre- an exhaust vent where a jet of high velocity gas
sented by the directivity factor Q, which is defined causes turbulence in the surrounding medium.
as the ratio of the actual intensity at angle 0, 10, to Normally, a spacing of transducers at 10 ° or 15 °
the intensity at that point if the source were radi- increments enables the source to be completely
ating omnidirectionally, I s, i.e., defined. Extensive work on aircraft jet engine
noise in static tests has led to the development of a
to computer program for quick handling of one-third
Qo - is octave band data. 24.25 This program includes

Spherical Hemispherical
radiation radiation
Q=I O =2 Q =4 O =8
DI : 0 dB DI = 3dB DI = 6dB DI = 9 dB

Figure 3.9 - Directivity effects resulting from reflecting planes.

23
standardizing andextrapolatingthedataaswellas \ \
computingthe soundpowerspectrumleveland \ \ \

directivityindex.Directivityeffectsbecomepar-
ticularlyimportantat higherfrequencies because
at low frequenciesdiffractioneffectsreduceany
directivecomponents. Point I I
source /// __
3.4.3 Sound Field Characteristics / / I' t_
i /
It is important when making sound pressure 0 r 2r
level measurements to consider the portion of the Relative distance, r
sound field likely to be measured. The direct or
free field is that part where the sound level decays Figure 3. IO - Increase in area through which unit
in a constant manner depending on the geometry sound intensity passes because of spherical
of the source and its surroundings. For example, divergence. S, unit area; I, constant intensity
for a relatively small source radiating freely, i.e., through section considered.

with no reflecting surfaces nearby, the level SPL r

at a distance r from the source is given by

from, W = 4_rr2 I0 = 4_rr2 I W

Free field
thus, IL 0 - IL r = SPL 0 - SPL r I .... ff_ Reverberant field
akss_pe I
-20 log (r/ro) M_o. I_'_ er d°ublingl 3-dB rise in reverberant tevel
o_ _a_ __ f r '
dB ii_ld _ -'_._ r I caused by reflecting surface
therefore, SPL r = SPL o - 20 log r/r o
..... LFar
'--,,e,d
where SPL 0 is the level at distance r 0 from the
source in decibels. rc log r

Figure 3. ]] - Sound field characteristics in semi-

This relation holds because the radiation pat- reverberant conditions.
tern is spherical and the area through which unit
intensity passes is proportional to r 2 (see section
2.2.3). The above relationship is referred to as the cerned. The reverberant field is said to be diffuse,
inverse square law and is illustrated in figures 3.10 since sound waves arrive at any point from many
and 3.11. The net result is a 6 dB reduction in level different directions and, except for points very
with each doubling of distance from the source. close to hard surfaces, the sound level tends to be
For line sources such as the radiating surface of a constant throughout the field.
length of pipe, the radiation pattern is cylindrical The distinction between near field and far
and a 3 dB reduction in level with doubling of dis- field is somewhat more subtle; basically, in the
tance is found. near field the sound pressure level does not have a
However, for any space confined by solid simple relationship with distance, because of the
boundaries such as partitions or factory walls, the proximity of vibrating surfaces. In this region
sound waves propagating freely from the reflected levels may remain constant or even increase with
component contribute to an effective increase in distance because of the strong dependence on
the sound pressure level in regions relatively near nearby radiating surfaces.
the reflecting surface. The distance from the With respect to noise control, increased ab-
source at which this so-called reverberant field sorption on surfaces has some effect on the rever-
begins to dominate over the direct field is labeled berant field, but none on the direct sound field,
r c in figure 3.11; that figure also represents a sum- which has encountered no reflecting surfaces. For
mary of sound field characteristics for a single the latter, either direct source modification or the
machine. The value of rc depends on the acoustic erection of some kind of partition represents a
power radiated and its directivity, and the absorp- highly effective way of reducing the noise level.
tion characteristics of the principal boundary sur- Chapter 6 discusses in more detail the approaches
faces, as well as the dimensions of the space con- to solving these problems.

24
3.4.4 Sound Power Level Estimation assumed. For the former case, all the above rela-
tionships apply, and it is usual to use a regular
Sound power is important because it repre- array of microphones equidistant from the source
sents a unique characteristic of a particular source and at least k/4 from any wall of the room, where
which, when properly specified, is independent of X is the wavelength of the lowest frequency of
the particular environment concerned. This meas- interest. Figure 3.12 shows a six-point array over a
urement is useful in comparing the relative acous- hypothetical hemisphere surrounding a source.
tic properties of various machines and the effects The sound power level, PWL, of the source,
of changing the acoustic conditions on the sound assuming omnidirectional radiation through equal
pressure. Knowing the sound power level of a areas associated with each measuring point, is
source in appropriate frequency bands, and by given by
making intelligent assumptions about the nature
of the sound field generated at points of interest . l%6/p, \2
(see section 3.4.3), it is possible to predict accu- PWL = 101ogz-z._l_-I + 201ogr + 8 dB
v i \Yref/
rately the sound pressure level at these points.
The general relationship between the sound
or
power level of a spherically radiating source and
the sound pressure level at distance r is derived
PWL =-SPL + 20 logr + 8 dB
from the following equation (see section 2.2.3):

W = I(47rr 2) W

Taking logarithms and using the sound pressure

level and intensity level equivalence principle (see
(6)
section 2.4),

PWL = SPL + 101og47rr 2 --

= SPL + 20 logr + 101og47r
= SPL + 20 logr + 11 dB

where r is the distance from source, in meters.

-=y
For a directional source the general equation
is given by

PWL = SPL + 20 logr + 11 - DI dB

where DI is defined as in section 3.4.2.

-- Noisesource
In ducts, assuming no losses at the peripheral
surfaces, the relationship becomes Coordinatesof unit hemisphere

PWL = SPL D + 10logS D dB Microphone

X Y Z
position
where
SPL D is the sound pressure level measured just 1 0.89 0 0.45
off center line, dB 2 0.28 0.85 0.45
3 - 0.72 0.53 0.45
S D is the cross section of duct, m 2.
4 -0.72 - 0.53 0.45
5 0.28 - 0.85 0.45
Ideally, the measurement of sound power 6 0 0 1
level should be made either in free-field conditions
as in an anechoic chamber, or in a highly reflective Figure 3.12 - Six-point hemispherical micro-
reverberant room where a diffuse field may be phone array for measuring sound power.

25
 whereSPLis the logarithmicaverageof sound where
pressure levels(space-averaged).
Note SPL is the space-averaged sound pressure
that the directivityindexof hemi- level, dB
spherical radiationabovethereflect- Vis the room volume minus source vol-
ingplaneis +3 dB(fromfigure3.9), ume, m 3
so that PWL = SPL + 20logr + TR is the Sabine reverberation time defined
11-3. in section 6.2.2, s
S is the area of all room boundaries, m2
Typicallyanevennumberof microphones is X is the wavelength at frequency of inter-
usedforthis,forexample, 4,6, 10,or 12,depend- est, m.
ingon thesourcesizeandspaceavailable.With
this method,directivityeffectsare easilyesti- A comparative technique may be employed to
mated,althoughif thesourceis highlydirective determine the sound power level of a test source
thena largernumber,up to 24or so,shouldbe where the acoustic and geometric characteristics
used.26,27
If a sufficientnumberof measurements are similar to those of a standard reference source. 33
of thesoundpressure levelaretakenon thesur- In this case, the procedure is first to find average
faceof ahypothetical sphere or otherregulargeo- sound pressure levels in the appropriate frequency
metricshapesurroundingthe source,then the bands for both the reference and test sources. The
errorimplicitin failuretosample thelevelatevery next stage is to calculate the sound power level of
point can be minimized. 2s Normally,measure- the test source according to
mentsaremadein octavebandsandthe sound
powerlevelfor eachof thesemaybesummed to PWL t = SPL t + PWLre f - SPLre f
giveanoverallpowerlevel.
In a reverberation room, the numberof where
microphones needed in anarray,AIM,isgenerally PWLre f, PWL t are the sound power levels of
lessthanfor a free field andis givenapproxi- reference and test sources,
matelyby respectively
SPLre f, SPL t are the space-averaged sound
pressure levels of reference
and test sources.

where In many situations, it is not possible either to

V is the room volume, m3 obtain prior information about machinery sound
is the longest wavelength of interest, m. power levels or to remove such machinery for lab-
oratory assessment of sound power. In these cases
it is necessary to make some kind of in situ meas-
Note: The microphones should be placed at urement, and this is most frequently performed
least _./2 apart and at least twice the maximum under semireverberant conditions such as exist
source dimension from the source. Their positions inside most work places. A thorough review of the
are not as critical as for free field assessment, since different techniques which may be used is given in
a diffuse field is being measured. chapter 4 of reference 26 (see also section 6.3.1).
There is an inherent inaccuracy in assessing As an example, we outline here the two-surface
pure tone radiation by this method29; this error is method developed by Diehl, 26 which is quite suit-
exacerbated when considering low frequency radi- able for tests on large machinery.
ation from large sources. 30 The equation for mak- If a 6 dB decrease in sound pressure level is
ing an absolute determination under reverberant obtainable in a semireverberant field, then even
conditions is31, 32 though the distance needed to achieve this is
usually greater than it would be in a free field,
approximate free field conditions may be assumed.
PWL = SPI_ + lOlog V- lOlog TR Corrections can be applied to the estimated sound
+ 10log 1 + - 13.5 dB power level in each frequency band to give an
approximate free field result. Measurements are

26
madeatlocationscorresponding to specificloca- machine is being run, since conditions may change
tionson the surfaceof a hypothetical
parallele- considerably during the course of operation. Also,
pipedenclosingthe machine,with total surface in some situations, it may not be easy to make far
field measurements and the microphones must be
areaS1.A second seriesof measurements is then
madeona largerparallelepiped
surfaceS 2, where placed close to the source. In this case, the meas-
urement surface should conform closely to the
the major dimensions have the same proportional
source shape, and microphones should be equally
relationship as for S 1. The average sound pressure
sensitive to all incident sound over the front 180 °
levels for each surface, SPL I and SPL 2, respec-
arc. 28
tively, are then calculated.
It can be shown that the actual sound power
is given by
3.5 SURVEY APPROACH
PWL = SPL t + 10logS 1 - C dB
3.5.1 Initial Preparation
where the correction factor C is
The proper definition of a noise problem is of
primary importance; this definition is best achieved
c= 101ogkl - 1 by first determining the need for noise control at
all. There may be a variety of factors which, either
where E = 10 (SPL_ - SPL2)/10 and S 1, S 2 are sur- singly or in combination, are responsible for the
face areas, in square meters. need for a noise analysis. The most common ones
include 33 potential hearing damage to employees
C is readily found from the family of curves because of adverse noise working conditions;
shown in figure 3.13. determination of whether equipment or vehicles
Sound power levels must always be quoted are in compliance with test codes or specifications;
for the particular conditions under which the community annoyance resulting from external

10=
9 _
8

1
.015 .02 .025 .03 .04 .05 ,06 .08 .15 .2 .25 .3 .4 .5 .6 .8 1.0
0.01 0.1

Area ratio, SJS 2

Figure 3.13 - Correction factors for two-surface method of sound power level determination (refer-
ence 26). (From George M. DieM, Machinery Acoustics, 1973. Used with the permission of John Wiley &
Sons, Inc.)

27
noiselevels;andnoiselevelcontoursurveysfor changes in excess of 5 dB should be sampled every
generalplanningpurposes. Theappropriatecri- 15 seconds or so and the arithmetic average taken
teriafor controlmaythenbeselected, for exam- after a few minutes. In this case it is also necessary
ple,OSHAnoiseexposure limits(seetheAppen- to specify the maximum range of excursions from
dix),community noiseordinances, orsoundpower the mean. Special meter characteristics are re-
levelmeasurement standards. Wherepossible,a quired to give an accurate representation of
groundplanandevaluation of thebuildingor area impulsive-type sounds.
concerned shouldbe obtained,as this becomes Initial measurements should be relatively few
invaluablein planningsurveysand identifying in number, provided the level does not vary greatly
particularlysensitive
areas. over large areas. It is usual to obtain representa-
Informationis alsorequiredon thetypeof tive measurements of a sound field by avoiding any
noisesources involved,theirdimensions, operat- reflecting surfaces by at least 1 m. When assessing
ingcharacteristics,directionality,andlong-term personal exposure conditions for individuals, the
influences suchas patternof usethroughouta microphone should be located at the normal posi-
workweek.Particularlyimportantis theneedto tion of the head (or 1 m away for in situ measure-
gainsomeideaof thelevelandspectral contentof ment). This is taken to be 1.5 m in height for
thenoise.In thisrespect,it isemphasized thatan standing positions and i. 1 m for seated workers. 34
empiricalon-siteinvestigation evenwithnothing When making measurements in a reverberant
more than aural inspectionby a well-trained field, it is normal to take a space-average level,
acoustician,andcarefulexamination of theabove especially for low frequencies.
factors,isof considerablevalueinhelpingtoselect In general, the measurements are made in
thetypeof equipment requiredandthemeasure- dB(A) before further specific analysis into octave,
mentsnecessary. Forexample, rotatingmachinery one-third octave bands, etc., is performed. It is
components suchaslathesandsawsgiveriseto always valuable to record a background or ambient
periodicwaveforms, whichtendto possess strong level of the sound field when the source of interest
discretefrequencycomponents requiringrelatively is not generating. This permits the effective source
detailedfrequency analysis. contribution to the total noise environment to be
estimated and is especially important in dealing
3.5.2 Measurement of Acoustic Quantities with community noise problems. Vibration meas-
urements may also be necessary, particularly when
Accuracy of measurements is essential. It is source modifications are being examined or when
imperative to ensure that equipment is functioning the relative importance of several structure-borne
correctly and to calibrate it using one of the transmission paths are being considered. This is
devices described in section 4.9. Internal reference discussed further in chapter 6. All operating
calibration signals should always be cross checked modes of the machines concerned should be ac-
with external calibrators. Other considerations counted for in the survey and analysis.
include the correct selection and deployment of a
microphone for the sound field involved, i.e., 3.5.3 Data Presentation
random incidence, perpendicular incidence, or
grazing incidence; and whether avoidance of inter- Generally, four major areas should be cov-
ference effects when performing precision meas- ered: sound field and source description; acoustic
urements at high frequencies requires the separate and physical description of the environment in
mounting of the microphone or instrument at least which propagation occurs; classification of instru-
1 m from the observer. mentation; and operator/receiver positions and
The possibility of instrument overload is exposure conditions.
reduced by ensuring the utilization of the lowest
3.5.3.1 Sound field and source description
available measuring range for which the complete
range of fluctuations in level still registers on the Sound field and source description may in-
instrument scale. For sound level meters the clude the specification of sound power levels as
"slow" response setting (see section 4.4.1) is used well as full identification of the location using
frequently unless the sound has a sharply changing diagrams, sketches, or photographs. Operating
nonperiodic amplitude, in which case the "fast" conditions should also be noted, including length
response setting is used. Cyclical noises which have of running time, cyclical or steady mode, work-

28
loadrequired,andsoon. Recorded levelsshould 5. Howes, W. L., Overall Loudness Calculation
befullycategorized
accordingto measuringposi- Procedure for Steady Sounds, Cosmic Pro-
tion, microphoneorientation,weightingsused, gram Abstract, LEW-12914, 1979.
frequencyband,fluctuations,
andtime. 6. Kryter, K. K., Review of Research and Meth-
ods for Measuring the Loudness and Noisi-
3.5.3.2 Environment description
ness of Complex Sounds, NASA CR-422,
A full description of the physical dimensions 1966.
of the surrounding walls, surfaces, and neighbor- 7. Beranek, Leo L., Noise and Vibration Con-
ing equipment is necessary, together with an anno- trol, McGraw-Hill Book Co., 1971.
tation of their probable acoustic qualities, e.g., 8. Fidell, S., Community Response to Noise in
absorptive or reflective, massive or lightweight. Handbook of Noise and Control, Cyril M.
3.5.3.3 lnstrumentalion description Harris, ed., Second ed., Chapter 36, McGraw-
Hill Book Co., 1979.
Instrumentation description should include a
9. Beranek, Leo k., Noise Measurements, Noise
complete logging of all equipment in terms of
Con. Eng., vol. 9, no. 3, 1977, p. 100.
manufacturer's type identifications and serial
10. Hayes, C. D., and M. D. Lamers, Octave and
numbers. Frequency bandwidths and signal-to-
One-Third Octave Acoustic Noise Spectrum
noise ratios may be specified where relevant. Cali-
Analysis (JPL TR 32-1052), NASA CR-81251,
bration results before and after the survey should
1967.
also be included, as well as the instrument settings
used for the measurements. 11. ANSI, American National Standard Specifi-
cation for Octave, Half-Octave, and Third-
3.5.3.4 Receiver positions and conditions Octave Band Filter Sets, SI.11-1966 (R1976).
Description of receiver positions and condi- 12. Shipley, J. W., and R. A. Slusser, A Digital
tions should include a complete description of Technique for Determining Third-Octave
levels received at the positions of critical exposure, Sound Pressure Levels with a More Uniform
as well as some account of both temporal and Confidence Level. JPL TM 33-422, 1969.
frequency characteristics. Background levels may 13. Ebbing, C. E., and T. H. Hodgson, Diag-
be included for comparison. Also, a record of nostic Tests for Locating Noise Sources:
machine operator movements throughout a work- Classical Techniques (Part I), Signal Process-
ing period should be provided. Where a workman ing Techniques (Part I1), Noise Con. Eng.,
is performing a variety of tasks throughout the vol. 3, no. 1, July-August 1974, p. 30.
day during which his noise environment changes 14. Soderman, P. T., and S. C. Noble, Direc-
substantially, then it is more relevant to use
tional M.icrophone Array for Acoustic Studies
dosimeters in assessing his personal exposure and of Wind Tunnel Models, J. Air., vol. 12,
to monitor closely his daily routine. no. 3, 1975, p. 168.
15. Kendall, J. M., Airframe Noise Measure-
ments by Acoustic Imaging, AIAA Paper
REFERENCES
77-55, 1977.
16. Kendall, J. M., Resolution Enhanced Sound
1. International Organization for Standardiza-
Detecting Apparatus, U.S. Patent 4 149 034,
tion, Normal Equal Loudness Contours for
April 1979.
Pure Tones and Normal Threshold of Hear-
ing under Free-Field Listening Conditions, 17. Kendall, J. M., Acoustic Imaging System,
ISO Recommendation R226, 1961. NASA Tech Brief, Spring 1977, p. 56.
2. International Organization for Standardiza- 18. Ahtye, W. F., W. R. Miller, and W. C. Mee-
tion, Method for Calculating Loudness Level, cham, Wing and Flap Noise Measured by
ISO Recommendation R532, 1966. Near- and Far-Field Cross-Correlation Tech-
3. Howes, W. L., Overall Loudness of Steady niques, AIAA Paper 79-0667, 1979.
Sounds According to Theory and Experiment, 19. Ahtye, W. F., and G. K. Kojima, Correlation
NASA Reference Publication 1001, 1979. Microphone for Measuring Airframe Noise in
4. Pearsons, K. S., and R. Bennett, Handbook Large-Scale Wind Tunnels, AIAA Paper
of Noise Ratings, NASA CR-2376, 1974. 76-553, 1976.

29
20. Moore,M. T., andE. B.Smith,SoundSepa- Pressure (prepared by National Bureau of
rationProbe,NASATech Brief B75-10286, Standards for EPA), EPA-550/8-76-001,
1975. 1975. (Available from NTIS as COM-75-
21. Wilby, J. F., ed., Coherence and Phase Tech- 11399.)
niques Applied to Noise Source Diagnosis in
29. Doak, P.E., Fluctuations of the Sound Pres-
NASA Ames 7 x 10 Foot Wind Tunnel
sure Level in Rooms when the Receiver Posi-
No. 1, NASA CR-152039, 1977.
tion is Varied, Acoustica, vol. 9, 1959, p. 1.
22. Brown, D. L., and W. G. Halvorsen, Appli-
cation of the Coherence Function to Acoustic 30. Smith, W. F., and J. R. Bailey, Investigation
Noise Measurements, lnternoise 72 Proceed- of the Statistics of Sound-Power Injection
ings, 1972, p. 417. from Low-Frequency Finite-Size Sources in a
23. Piersol, A. G., Use of Coherence and Phase Reverberant Room, J. Acoust. Soc. Amer.,
Data Between Two Receivers in Evaluation of vol. 54, no. 4, 1973, p. 950.
Noise Environments, J. Sound Vib., vol. 56, 31. Beranek, Leo L., The Measurement of Power
no. 2, 1978, p. 215. Levels and Directivity Patterns of Noise
24. Montegani, F. J., Some Propulsion System Sources in Noise and Vibration Control, Leo
Noise Data Handling Conventions and Com- L. Beranek, ed., Chapter 6, McGraw-Hill
puter Programs used at the Lewis Research Book Co., 1971.
Center, NASA TM X-3013, 1974.
32. ANSI, American National Standard Methods
25. Montegani, F. J., Some Propulsion System
for the Determination of Sound Power Levels
Noise Data Handling Conventions and Com-
of Small Sources in Reverberation Rooms,
puter Programs used at the Lewis Research
S1.21-1972.
Center, Cosmic Program Abstract, LEW-
12285, 1975. 33. International Organization for Standardiza-
26. Diehi, G. M., Machinery Acoustics, John tion, Determination of Sound Power Levels
Wiley & Sons, 1973. of Noise Sources-Methods Using a Reference
27. ANSI, American National Standard Method Sound Source, International Standard 3747,
for the Physical Measurement of Sound, 1975.
S1.2-1962 (1971) (partially revised by S1.13- 34. Wells, Richard J., Noise Measurements:
1971 and by S1.21-1972). Methods in Handbook of Noise Control,
28. Holmer, C. I., Procedures for Estimating Cyril M. Harris, ed., Second ed., Chapter 6,
Sound Power from Measurements of Sound McGraw-Hill Book Co., 1979.

3O
 CHAPTER 4

Instrumentation for Sound and Vibration Measurement

To transform the variations in sound pressure tion of the relatively weak transducer signal; proc-
associated with audio frequencies (20 Hz to 20 essing according to predetermined parameters of
kHz), or the mechanical excitation caused by interest; and eventual display on some form of
vibrations in a structure, into an electrical analog recorder, meter, or oscilloscope for measurement
signal suitable for processing and subsequent and assessment by the observer. The schematic
measurement of the original stimulus, a device diagram in figure 4.1 illustrates typical stages in
known as a transducer is required. Those used for the measuring chain. Frequently, vibration and
measuring sound pressure are known as micro- sound pressure analyses may be made using the
phones, of which there are several commonly same equipment.
available types. When making vibration measure-
ments, a transducer known as an accelerometer is 4.1 MICROPHONES
most often used.
In both instances, the general processing and There are currently three types of micro-
analysis system is similar, involving preamplifica- phones used for acoustic measurements; table 4.1

Tape
recorder "1

Data storage
E r-t Real time
analyzer

Sound
level
meter

Oscilloscope

___ analyzer
Frequency

Chart
Pre amp
recorder

I Accelerometer

Microphone meter
._ Vibration ?

Meter

analyzer
Statistical

I I I I
Transducers Signal conditioning and analysis Display

Figure 4.1 - Schematic diagram of typical measuring combinations of sound and vibratton equipment.

31
 Incident
Diaphrag m ___'_ sound waves

.nc.de
sound.n tu.ato
waves

Backplate/
_

V _
-- --

J/_Hous,ng
Metalized

etectret
---_

film electrode
Polarized -_
L'_.
_

duuuuuuu
_

_
_kd
-!
I Diaphragm

i
Metal_plate j I
electrode
Hole for static -_

pressure equalization

(a) (b)

Figure 4.2 - Schematic diagrams of condenser microphones. (a) Air-condenser. (t5) Electret-condenser.

Table 4.1 - Characteristic features of microphone rigid backplate, with an air gap between them (see
types used for acoustic measurements. figure 4.2a). A polarizing voltage is applied across
these, thus forming a capacitor which is highly
Piezo- Air- Electret-
sensitive to changes in the air gap caused by rela-
electric condenser condenser
tive movement of the diaphragm resulting from
Sensitivity Reasonable Good Good fluctuating air pressure. The change in capaci-
Frequency response Reasonable Good Good tance of the device generates an alternating voltage
External polarization
proportional to the sound pressure fluctuations
voltage None Yes None
Long-term stability Reasonable Excellent Good over a large frequency range. There is a small vent
Influence of humidity Reasonable Poor Good which connects the air gap to the ambient atmos-
Ruggedness Good Poor Good
phere, so that the diaphragm remains unaffected
Relative cost Low High Medium
by changes in the ambient pressure.
Although renowned for their excellent stabil-
highlights some of the principal features of each. ity and low internal noise, conditions of high
Figure 4.2 illustrates the basic construction details humidity adversely affect the output of air-con-
for the more sophisticated condenser types. denser microphones.

4.1.1 Piezoeleclric Microphone 4.1.3 Electret-Condenser Microphone

This design incorporates a piezoelectric Electret-condenser microphones have become

ceramic crystal connected to a diaphragm. This increasingly popular in recent years and have an
nonconducting crystal produces an electric charge advantage over the air-condenser type by having a
when disturbed by the movement of the diaphragm built-in polarizing voltage. This voltage is achieved
caused by incident sound pressure. by incorporating a plastic diaphragm with a metal-
Piezoelectric microphones do not have as ized sensing surface in front of a prepolarized
linear a frequency response as the other types dis- polymer material (see figure 4.2b). The backplate
cussed below. However, they do have the advan- has a large number of small protuberances, each
tages of lower cost, robustness, and no require- representing an individual sensing cell. These cells
ment for a polarization voltage. act in concert to provide a composite output pro-
portional to the incident sound pressure. The
4.1.2 Air-Condenser Microphone design is inherently less sensitive to humidity,
although at the present time they are not generally
This type of microphone has a very thin believed to possess quite the same long-term sta-
stretched-metal diaphragm positioned close to a bility as their cousins, the air condensers.

32
4.1.4 Directionality and Sensitivity strated by plotting frequency versus relative re-
sponse in decibels for varying angles of incidence
The sensitivity of a microphone is measured in a free field (see section 3.4.3). This character-
by the size of its output voltage for a given strength istic may be optimized such that linearity through
of acoustic signal at the diaphragm. Microphones high frequencies is obtained for specified angles of
are extremely sensitive pressure sensors, since for incidence.
typical industrial situations they need to be able to Three classes of microphone are normally
detect sound pressure in the range 6.3 × 10 -4 available: free field or perpendicular-incidence
N/m 2 to 6.3 × 102 N/m 2 (30-150dB). In order to microphones, which have the best response when
measure low sound pressure levels they need to the diaphragm is perpendicularly oriented with
have a reasonably large diaphragm. Unfortu- respect to the sound source; random-incidence or
nately, this conflicts with another prerequisite, pressure microphones, which have the flattest fre-
which is that the body of the microphone itself quency response when the sound field is diffuse
should not interfere with the sound field through and sound waves randomly impinge on the dia-
diffraction or reflection effects, such that the phragm at all angles (this roughly corresponds to
reading obtained is different from one obtained if an average angle of incidence of 70 ° with respect
the microphone were infinitely small. This inter- to the microphone longitudinal axis); and grazing-
ference phenomenon predominates at high fre- incidence microphones, which are optimized for
quencies, where the wavelength is of the same best response at 90 ° angles of incidence. Figure
order or less than the microphone diameter. The 4.3 illustrates the frequency response character-
net result is to make the output dependent on the istics of two classes of microphones and figure 4.4
angle of incidence of the sound waves. It follows shows the effect of size on frequency response.
that the smaller the microphone, the more omni- Random incidence correctors are available for
directional its behavior will be; however, this is attaching to free field microphones to change their
accompanied by a reduction in sensitivity. As a free field characteristic such that they give good
guide, i-inch microphones are directional above linear response in diffuse sound fields. The Amer-
about 3 kHz, 1/2-inch above 6 kHz, and 1/4-inch ican National Standard 2 sets design criteria for
above 12 kHz. z random-incidence microphones only.
The directionality characteristic, otherwise
termed the free field characteristic, is demon-

90 o

I
M ic rophone"_ -- 0°

/-- 900 ,/-900

6

2 J 6 / ,/
J
"-- -2
2
rr _ -6
\ -6
20 50 100 500 lk 5k 10k 15k 20k 20 50 100 500 lk 5k 10k 15k 20k

Frequency (Hz) Frequency (Hz)

(a) (b)

Figure 4.3- Frequency response characteristics of typical 1-inch microphones. (a) Grazing-incidence
microphone. (b) Perpendicular-incidence microphone.

33
 0° 0o

I i
Relative Relative
microphone microphone
response response

OdB OdB

_[_ 90 °

4kHz -/ _l _--5 kHz

1800 1800

(a) (b)

Figure 4.4 - Directivity plots at different frequencies for two sizes of microphones. (a) l. 2 7-cm (0.5 -inch)
microphone. (b) 2.54-cm (1-inch) microphone.

4.2 ACCELEROMETERS AND OTHER Table 4.2 - General categorization of

VIBRATION TRANSDUCERS vibration transducers.

Ideally, the location of a vibration transducer Generating Nongenerating

should not influence its output; it should perform
Seismic Piezoelectric Strain gauge acceler-
its function equally well whether it is connected to accelerometer ometer (thin wire or
a structure on the ground, a moving vehicle, or a piezoresistive semicon-
complex machine. The quantities used to describe ductor construction)
vibration movements are either displacement,
Nonseismic Electro-dynamic Noncontacting dis-
velocity, or acceleration. The output of the trans-
velocity placement transducers
ducer should be directly proportional to the phys- transducer (utilizing changes in
ical quantity of interest and normally it is attached capacitance, induct-
directly to the vibrating body. If this is not feas- ance, or eddy currents)
ible, it may be attached to a fixed structure, and
measurements made relative to the movement of cause of its high output relative to size and its rela-
the body of interest. tively uniform frequency response. Generally, it
Table 4.2 lists the main vibration transducers consists of a relatively large mass clamped firmly
commonly used. These can be broadly classified against two piezoelectric discs which act as a
into generating or nongenerating types, according spring between the mass and a solid base. When
to whether an electrical supply voltage is needed; subjected to vibration, the mass exerts on the discs
and seismic or nonseismic, depending on whether a force proportional to its acceleration. In turn,
the device is fixed to the vibrating body or meas- the discs generate a voltage proportional to this
ures relative motion only. force and therefore to the acceleration to which
the whole unit is exposed.
4.2.1 Piezoelectric Accelerometer The sensitivity of the unit depends on its size;
however, the larger the unit, the lower its resonant
Of the devices categorized in table 4.2, the frequency. This latter parameter is important in
piezoelectric accelerometer is the most popular be- determining the high frequency performance of

34
accelerometers.In addition,because of itsmass,a Voltage and charge amplifiers are two types
largerunit will tend to interactmorewith the of preamplifiers in common use. Condenser
response of a system. microphones or piezoelectric transducers may be
Research conducted by NASAandUSAFon regarded as having a constant charge sensitivity.
propellersandfan bladeshasplacedsevere con- This means that a long length of cable between
straintsonthesizeandweightof accelerometers. microphone and preamplifier, which increases the
Asa majorconsequence of thiswork,a series of capacitance of the system, reduces the output
miniaturepiezoelectric accelerometers with good voltage. Hence voltage preamps must always be
response characteristics
has beendeveloped by used close to the transducer. Some preamplifiers
Bolt, Beranek, andNewman,Inc. Thesepossess provide a polarization voltage for use with air-
relativelyhighsensitivity,built-in preamplifiers, condenser microphones.
andhavea typicalsensing diameter of only0.2cm. On the other hand, charge amplifiers, which
Because of their smallsizetheyhavealsobeen are sensitive only to charge, may be located at any
usedto makemeasurements withinthinboundary distance from the transducer. Charge amplifiers
layerswith minimumdisturbance to the flow are, however, more expensive than voltage pre-
structure.Suchminiatureaccelerometers areof amplifiers.
considerable valuein anysituationwherethereare
severespatialor massconstraints.
4.4 SOUND LEVEL METERS
4.2.2 Integrators
Sound level meters represent a complete port-
Electronic integrators, inserted in series be- able measuring system in one unit. They are espe-
tween a transducer and the preamplifier (see sec- cially convenient for field use and can be adapted
tion 4.3), make the output of an accelerometer to take input from an accelerometer so that vibra-
frequency sensitive. By switching in the appropri- tion measurements may be made directly. There is
ate frequency slope characteristic, an output pro- a great variety of instruments available on the
portional to displacement, velocity, or acceleration market and a wide variety of applications to which
can be produced. These quantities can therefore they may be put, from making sound power meas-
be measured directly, extending the versatility of urement surveys to monitoring of factory ambient
piezoelectric devices. noise levels.

4.4.1 Electrical and Operational Characteristics

4.3 PREAMPLIFIERS
The basic components of a sound level meter
Preamplifiers are used to provide impedance are shown in figure 4.5. Because of the large range
matching and to boost the relatively low output of sound levels (up to 30-150 dB) which the instru-
signal from the transducer before it is routed ment may be required to measure, representing as
through signal conditioning systems. much as 106 change in pressure and hence signal

networks I--1

(A,B,C,D)I _

Amplifier Amplifier rectifier (I,Meter

F, 5')
IMicrophone H Pre amp I_ RMS
!
_-¢'I Weighting /}___I'-[ I_ _ I
I External
I filter
I

L power
I t Stabilized
supply

Figure 4.5 - Schematic diagram showing component parts of a sound level meter. Response characteristics
of meter needle with time constants in parentheses may be either I, impulse (35 ms); F, fast (125 ms); or S,
slow (1000 ms).

35
magnitude, the amplifier inputs are controlled by Readings may be taken by eye-averaging
fixed attenuators, usually in steps of 10 dB. More where the total range of fluctuation is within _+3
recently, with the advent of digital processing dB or so, and the total range of variation is also
techniques, it has become possible to extend the recorded. For steady sounds, the "slow" response
scale over which a meter can measure without mode may be used, but for short-term variations,
changing the attenuation. such as the cycle of a cutting machine, the "fast"
Electronic weighting scales, discussed in the response is more appropriate. In order to take even
previous chapter, are incorporated to assess the longer-term averages, Leq readings may be made
effect of noise on people. Usually the A-weighted (see section 3.1.3), since these can be integrated
scale is employed (see section 3.1.2). Many instru- over any time period. AC and DC outputs are
ments also have facilities for switching in a fre- often provided so that tape recordings of the
quency filter which may be either octave or one- measured sound may be made or amplitude anal-
third octave band. ysis in either the frequency or time domains
The meter needle response is determined by performed.
preselection of a fast, slow, impulse, or peak hold
characteristic. These characteristics determine the 4.4.2 Standards and Specifications
time over which the RMS signal from the rectifier
is integrated to provide an averaged reading. The The principal discriminating feature in terms
fast and slow modes incorporate 1/8-s and l-s of quality and accuracy of different sound level
time constants, respectively. The peak mode has a meters is the type of microphone used and its char-
very short time constant, less than 50 _, to assess acteristics. There are now three international
short duration sounds such as punch press or standards, 4-6 which are soon to be replaced and
impact noise. It has a rapid response to increasing updated by a single document, and a single na-
sound levels and a slower response to decreasing tional one, 2 which is also scheduled to undergo
levels. The proposed new OSHA regulations (see revision. The amended standards will recategorize
the Appendix) specify the requirements for this. instruments in terms of four classifications. Even
The impulse mode with a 35-ms time constant is though the design goals for all the standards are
more commonly used outside of the United States. identical in regard to frequency response, the tol-
Recently, the detection-averaging characteristic of erances allowed vary considerably at low and high
sound level meters has been significantly im- frequencies. The American standard specifies
proved, especially with the recognition of danger greatest accuracy for random incidence, while IEC
to hearing from impulsive sounds, which have a requirements are for optimum performance at
high instantaneous peak level relative to the RMS either 0 ° or 90 ° incidence.
level. The crest factor, which is the ratio of the The three major classes of sound level meters
pressure associated with these two measurements, in current use are discussed in more detail below
has been extended in many impulse precision with an indication of their use.
meters to a factor of 20 dB or even more, so that The Type 3-Survey Meter is rarely used ex-
when taking measurements of impulsive sounds cept as a quick check on noise levels. Accuracy 2
there is less danger of the meter's failing to give a may only be + 3 dB at 1 kHz, when A-weighted,
true indication of the sound level. Additionally, but it does offer the advantages of economy and
some meters have an overload indication to give a compactness.
warning when the signal reading may be in error. Type 2-General Purpose Meters often use
Differential sound level meters can provide a condenser microphones and have an accuracy of
direct indication of small differences in sound at least + 2 dB at 1 kHz. The A, B, and C weight-
pressure between two sites. This may be of advan- ing networks (see section 3.1.2) are usually incor-
tage when estimating insertion loss characteristics porated and measurements may be made over a
across partitions or spatial variations in level wider range of sound levels than with Type 3
within a sound field. One instrument whose devel- meters. Special meters known as Type S2A offer
opment was sponsored by NASA 3 is especially only A-weighted measurement capability and are
suitable for measuring high intensity sounds; it relatively inexpensive. Occasionally, octave-band
features automatic gain control in order to ensure analyzers are also incorporated as well as detach-
the continuous matching of gains on both chan- able microphones. Examples of Type 2 are illus-
nels as the signals change. trated in figure 4.6a.

36
 m

Sound Level Melof

Type 2219

(a) {b)

Figure 4.6 - Sound level meters. (a) General-purpose sound level meters (Type 2) (by courtesy of Bruel &
Kjaer and Quest Electronics). (b) Impulse and precision sound level meters (Type 1) (by courtesy of
GenRad and Bruel & Kjaer).

Type 1-Precision Meters are used only in trial situations. They integrate noise exposure over
carefully controlled field environments or in labo- time periods of several hours or more, so as to
ratory work. They have accurate microphones of make a comparison with OSHA specified limits
either the air- or electret-condenser type, which (see the Appendix). Generally, they read out re-
may be mounted remotely in order to minimize the suits as a fraction or percentage of total allowable
effect of the observer and instrument case on the dosage; sometimes a separate readout unit is re-
incident sound field. Type 1 meters measure over a quired, to prevent manipulation of the received
larger range than the other two types and have an dose by the wearer. Typically, dosimeters have a
accuracy of at least +1 dB at ! kHz when A- restricted range over which they register, often
weighted. 90-115 riB(A), although a detector may store any
Impulse precision meters are governed by IEC signals in excess of this range. OSHA limits are
179A. 6 These normally have all the facilities of a exceeded if the noise exposure index obtained
Type 1, with an additional impulse and sometimes from summation of levels over an 8-hour day is
peak hold characteristic. They can measure signals greater than 1.0.
with a high crest factor in any mode and possess Noise dosimeters provide a more reasonable
an overload signal indicator, which may be impor- estimate of an individual's complete noise expo-
tant when dealing with complex signals. sure than representative measurements taken with
a sound level meter. This is particularly true when
4.5 NOISE DOSIMETERS
the operator's position may change during the day
Noise dosimeters are commonly used to deter- or when the operator is exposed to varying noise
mine an individual's exposure to noise in indus- levels. Dosimeters have been designed as compact

37
 Figure 4.7 - Noise dosimeters (by courtesy of Bruel & Kjaer and Quest Electronics).

and lightweight units so that they can be comfort- criteria (see section 3.3.1). A full discussion of fre-
ably worn during a working day. Figure 4.7 shows quency analysis techniques is given in section 3.3.
some typical units.
A limited measuring range 7 is not the only 4.6.1 Serial Analysis Instruments
shortcoming in dosimeters. In addition, their
operating crest factor may not be very high, limit- Portable serial analysis instruments (see sec-
ing sensitivity to impulsive-type sounds. In addi- tion 3.3.2) in the form of fixed percentage band-
tion, their accuracy in measuring any particular width frequency analyzers complementary to
level may be less than that of a sound level meter. sound level meters are now commonly available.
Also, the microphone must be positioned correctly These are usually octave band or one-third octave
on either the shoulder lapel or breast pocket, ac- band units, they enable analyses to be made con-
cording to the type of sound field involved, 8 and veniently on site. Fixed bandwidth instruments,
the wearer must be instructed appropriately. Ac- such as 20 Hz, 10 Hz, and 5 Hz, have until recently
curate calibration is essential, because of the been too bulky to use in the field because they
lengthy measurement periods usually involved (see employ many electronic filters; and it has often
section 4.9.1), and the complete instrument must been necessary to tape record the signal for future
be protected from adverse environments such as laboratory analysis. Real time analyzers (see sec-
dust and heat. tion 4.6.2) have largely supplanted such fixed
bandwidth analog instruments.
4.6 FREQUENCY ANALYZERS A complete analysis with a serial frequency
analyzer is a lengthy process, and the results are
Frequency analysis is an important feature of invalid if the signal spectrum or level changes
noise control engineering, especially when design- during the analysis. Figure 4.8 shows a typical
ing enclosures, optimizing absorption linings, frequency analyzer of this type.
calculating sound power levels of machinery, or
identifying noise sources within a complex sound 4.6.2 Real Time Analyzers
field. All of these procedures, as well as noise
control materials, exhibit a strong frequency Within the last decade or so, the development
dependence. A single number assessment of the of high-speed analog-to-digital converters, inte-
frequency spectrum, such as comparing A, C, and grated circuitry, and increased digital memory
linear-weighted readings to give an approximate capacity has enabled frequency analyses to be
assessment of low or high frequency content, is completed in a matter of seconds. This technique
not sufficient for establishing design and control is called real time analysis, and it is clearly an

38
 In On 08 Figure 4.8 - Serial frequency analyzer (by courtesy of Bruel & Kjaer).

_0_ W.,eheng S,_¢h

1000 2O0O

(a) (b)

Figure 4.9 - Portable real time spectrum analyzers. (a) By courtesy of GenRad. (b) By courtesy of Nicolet
Scientific Corporation.

advantage over the serial process of manually instrument is determined by the number of lines
switching from one filter to another, since essen- on the display screen, representing the number of
tially all frequency bands are examined at the constant bandwidths available. Fixed percentage
same time. The rate at which digital sampling of bandwidth capabilities may also be provided, and
the analog signal occurs predetermines the real the analyzers are especially useful for examining
time frequency range, although it may be consid- transient sounds.
erably below the highest useful frequency range of Clearly, then, these instruments represent a
the instrument. For example, an instrument capa- very useful and versatile tool to the noise control
ble of analyzing from 0 to 20 kHz may only have engineer. Although they are becoming lighter,
true real time capability from 0 to 2 kHz. cheaper, and easier to use, the serial type filter will
Real time analyzers can perform various probably remain in popular usage in the field for a
functions, 9 and the popular fast Fourier trans- considerable time for reasons of economy and
form (FFT) type computes power spectral densi- simplicity. Figure 4.9 shows some typical real time
ties, correlation functions, and the like (see sec- analyzers.
tions 3.3.1 and 3.4.1). The resolution of the

39
 20

rt_

I00 1 ' __
z an

o
^1 o 1o
e_

X ,,i ¥V - "'--._
L 9o
1-
6O 0 I I i
g 60 80 100 dg(,4)

Sample probability distribution

Time trace

Figure 4.10 - Typical traces and parameters obtainable from amplitude distribution analyzers.

4.7 AMPLITUDE AND STATISTICAl, a recording pen which marks on the paper the
ANALYZERS signal parameter of interest. This measure is
usually either amplitude against time or frequency
Many signals analyzed in noise control have a analysis in percentage bandwidths. Specific appli-
fluctuating level which may be described in a cations of these instruments include measurement
number of different ways. An energy average of sound level with time, reverberation times,
taken over a specified period of time constitutes sound transmission loss values for insulating par-
an Leq measure, and the fast or slow responses on titions, and frequency calibration of instruments.
sound level meters constitute short-term RMS By varying both pen and chart speed, the equiva-
averaging. The time-varying facet of a noise envi- lent response of a sound level meter needle may be
ronment may also be more precisely described by modeled. Graphic recorders are usually laboratory
recording the proportion of time during which the based, although portable recorders which may be
sound occurs at each of a series of contiguous
used in conjunction with sound level meters have
sound level intervals within the range of interest. become available. These recorders allow the
With recent developments in analysis instrumenta-
analyst to obtain a quick and clear subjective
tion, portable digital units have become available. impression of the parameter changes of interest.
These can quickly give a wide range of statistical
parameters, from probability distributions to the 4.8.2 Magnetic Tape Recorders
sound level that is exceeded X% of the total sam-
pled time (usually designated Lx). Magnetic tape recorders store a magnetic
A typical selection of statistical parameters analog of sound pressure or mechanical vibration
and their relationships is illustrated in figure 4.10. response so that subsequent analysis may be per-
Although the total number of samples which may formed at a more convenient time and often with
be taken is limited, the integrating period of the
greater sophistication than at the recording site.
analyzer may be reduced to permit finer categori- Moreover, the capability of recording several sig-
zation of a highly fluctuating signal, or lengthened nals simultaneously can help in making sound
to allow a larger total measuring time. power level and survey-type measurements. Also,
by changing the tape speed, signals with wide
4.8 RECORDERS
bandwidths may be brought within the scope of
Recorders retain a convenient record of the the analysis equipment. Digital signal processing
analysis signal which can be conveniently trans- increasingly obviates the need for tape recorders in
posed, reexamined, or referenced at a later date. many of these roles.
Figure 4.11 shows examples of both graphic and Frequently, a recorder is the weak link in a
magnetic tape recorders. measurement chain; direct on-line measurement is
generally preferred if possible. When selecting
4.8.1 Graphic Recorders recorders, consideration must be given to input
Graphic recorders are generally of the strip monitoring, distortions imparted to the original
chart variety: a continuous roll of paper is fed past signal such as phase-shift, adequate signal-to-

40
 [a) Ib)

Figure 4./1 - Recorders used in noise and vibration analysis. (a) Graphic/eve/recorder (by courtesy of
Bruel & Kjaer). (b) Magnetic tape recorder (by courtesy of Nagra Kudelski).

noise ratio, calibration of the taped signal, and 4.9.1 Sound Pressure Calibrators
compatibility of the tape with the recorder. Pistonphones, electrically generated sources,
Magnetic tape recorders are either direct or
and falling ball calibrators are all used for cali-
frequency-modulated. Within the former, profes-
brating sound pressure readings in measurement
sional hi-fi standard is normally satisfactory using
equipment. The first two are more accurate be-
one-quarter inch tape, but instrumentation class cause the calibrator is fitted securely around the
equipment with a larger channel capacity may also
microphone neck with a known void in between.
be used. These are adequate within the audio fre- Examples of both are shown in figure 4.12. There
quency range of 40 Hz to 14 kHz. FM recorders are also electronic calibrators that produce known
are usually used for vibration studies, or for situa- sound pressure levels at several different frequen-
tions where low frequencies from 50 Hz down to cies, thus allowing the frequency and weighting
DC must be recorded, or where good phase repre- network responses to be checked. Pistonphones
sentation is required, as with impulse waveforms. have a cam-driven diaphragm which produces a
known sound level at a single low frequency. Fall-
ing ball calibrators rely on a collection of ball
4.9 CALIBRATION bearings which pass through an orifice and strike a
taut skin diaphragm. The constriction prevents all
the balls from falling at once. These units only
In order to get maximum accuracy from a give an approximate calibration; over a period of
measuring system the measurement must be com- time they degrade because the diaphragm stretches.
pared against a calibrated signal. This ensures the Dosimeters are calibrated with a stop watch
reproducibility of the measurement and, if prop- in conjunction with a standard sound level meter
erly carried out, gives correct absolute values. A calibrator suitably coupled to the dosimeter. The
system can be accurately calibrated by applying a level used must be higher than 90 dB to register on
reference source, which produces a known level at the dosimeter, since these are specifically designed
a specified frequency, to the sound or vibration to assess compliance with OSHA limits, which
transducer and adjusting the sensitivity until the currently apply only to levels in excess of this
RMS value indicated by the meter agrees with the value.
reference. This calibration should be performed
immediately before and after measurement on 4.9.2 Vibration Calibrators
location.
It is also possible to calibrate by feeding an Vibration calibrators subject the transducer
internal reference signal to the preamplifier; how- to a known vibration level and the measuring
ever, the accuracy of this method depends on the equipment sensitivity is then suitably adjusted.
absolute calibration of the internal reference The value used most commonly is a peak accelera-
signal. tion of 1 g at a single frequency, usually 80 or

41
Figure 4.12 - Field sound level calibrators and adaptor rings for various microphone sizes (by courtesy of
GenRad and Bruel & Kjaer).

100 Hz (g is unit acceleration from gravity, or 3. Zuckerwar, A. J., and R. E. Jones, A Differ-
9.81 N/m2). ential Sound Pressure Level Meter, Third
lnteragency Symposium on University Re-
4.10 MISCELLANEOUS ANCILLARY search in Transportation Noise, November
EQUIPMENT 1975, p. 105.
4. International Electro-technical Commission,
The most important items of ancillary equip- Recommendations for Sound Level Meters,
ment are those which improve the performance of Publication 123, 1961.
microphones, which are very sensitive to adverse 5. International Electro-technical Commission,
environments, l° For example, dehumidifiers are Precision Sound Level Meters, Publication
available as a screw-on connection to air-condenser 179, 1973.
microphone cartridges for high humidity environ- 6. International Electro-technical Commission,
ments. Push-on windscreens can be used for out- Additional Characteristics for the Measure-
door noise measurements and are effective for ment of Impulsive Sounds, Publication 179A
wind speeds of up to around 4 m/s. Nose cones (First Supplement to Publication 179), 1973.
are used for reducing turbulence noise in high 7. Leasure, William A., Performance Evalua-
speed, relatively uniform flow air streams, such as tion of Personal Noise Exposure Meters,
in ventilation or exhaust ducts. Because the acous- Sound and Vibration, March 1974, p. 36.
tic signal must be separated from within the highly 8. Seiler, John P., Noise Dosimeters, Sound and
turbulent air flow in wind tunnels and ducts, Vibration, March 1977, p. 18.
NASA and other organizations have developed a 9. Kamperman, George W., and Mary Moore,
variety of microphone attachments, the best of Real Time Frequency Analyzers for Noise
which suppress turbulent noise by 12-15 dB rela- Control, Noise Control Engineering, vol. 9,
tive to nose cone performance. I1,12 These may no. 3, Nov.-Dec. 1977, p. 131.
take the form of porous surface airfoils and 10. Magrab, E. B., Environmental Effects on
porous or slit tubes, with a well-defined direc- Microphones and Type II Sound Level Meters,
tionality characteristic. Tripods and extension National Bureau of Standards, Technical Note
cables also reduce the interaction between the 931, 1976.
acoustic field and the instrument case and observer. 11. Noiseux, D. E., Porous Surface Microphone
for Measuring Acoustic Signals in Turbulent
REFERENCES Windstreams, NASA Tech. Brief B73-10490,
1973.
1. Schneider, Anthony, Instrumentation Micro- 12. Crocker, M. J., R. Cohen, and J. S. Wang,
phones, Sound and Vibration, March 1977, Recent Developments in the Design of Tubu-
p. 6. lar Microphone Windscreens for In-Duct Fan
2. ANSI, American National Specification for Sound Power Measurements, Inter-Noise 73
Sound Level Meters, ANSI SI.4-1971. Proceedings, 1973, p. 594.

42
 CHAPTER 5

Noise and Vibration Control Materials

This chapter describes classes of noise and tion (total intensity 11). The remainder is re-
vibration control materials, tabulates their radiated at the air-structure boundary of the
specifications, and outlines their general partition with intensity 14. The degree of re-
application. In theory, any material has some radiated energy on the receiver side of the
effect on the propagation of acoustic energy, but structure is characterized by the transmission
we only review here those commonly used for coefficient, 7".r is numerically equal to the ratio of
modifying the acoustic environment to improve transmitted intensity, 14, to the incident intensity,
the exposure conditions of people at their place of 11. Because energy is always conserved, these
work. intensities are related thus:
Such materials work by redirecting acoustical
and vibrational energy or converting such energy 4
to another form, typically thermal energy. The
various modes of operation involved and the rela- The coefficients ¢x and z, numerically defined
tionship of air- or fluid-borne acoustic energy to in figure 5.1, describe the ability of a material to
absorb and transmit, respectively, air- or fluid-
structure-borne energy are shown in figure 5.1.
The difference between the incident and re- borne acoustical energy, and each has a maximum
value of unity. The loss factor, rl, is used to quan-
flected acoustical energy (11 -Iz), divided by the
tify a material's ability to attenuate structural
incident intensity, I l, is a measure of the total de-
vibration by internal damping. Certain materials
gree of absorption, c_, which takes place at that
are frequently applied to vibrating surfaces in
boundary. The residual energy intensity (I l -12),
which is not reflected, may either be dissipated order to reduce the radiated sound energy. This is
distinct from vibration isolation, which is the
within the structure itself by friction, viscous, and
process by which a material is used to inhibit the
hysteretic losses (total intensity I3), or transmitted
transmission of vibrational energy from one struc-
away from the partition by structure-borne vibra-
ture to another.
The four major categories of acoustic mate-
_b"sor b e_d"_
Incident _/ _intensity _J/I rials are absorbers, insulators, damping materials,
and vibration isolators. Absorption materials are
(r) =. Transmitted
intensity /1 -_. normally lightweight, and fibrous or highly porous,
14 w intensity
Reflected (_)__ whereas effective barrier or insulating materials
tensity t _._,_//_
are of high density and often quite limp. Hence, in
intensity /2 _._
O_ _ • - many noise control applications, absorption and
Absorber and /1
Source insulation materials often are used together, since
solid structure
r =- Receiver their distinctly different functions and modes of
tl
operation make it difficult to attenuate effectively
Figure5.1- Relationship of acoustic energy the sound at a receiver by the use of either material
intensities for propagation across a solid struc- on its own. Damping materials usually consist of
ture. The acoustic coefficients with most influence viscoelastic materials such as epoxy-polymer
on relevant intensities are shown in parentheses. mixes or rubbers, which are inserted or adhered
Strictly speaking, this conceptual illustration is onto machine parts, panels, and so on, in a num-
only relevant for normal propagation to the parti- ber of possible configurations. There is a wide
tion. variety of vibration isolators in use; they typically

43
consistoflightlydamped steel springs, elastomeric composition such as Kevlar and Scottfoam. 2,3 A
rubber compounds, or even pneumatic devices, reasonable prediction of the normal surface im-
which must all possess sufficient resilience to sup- pedances (see section 2.2.5) and normal absorp-
port the loads required. tion coefficients was achieved by identifying the
The following sections outline the means of relevant structure factor and introducing a new
optimizing the specification for each of these four parameter called the quality factor. The particular
categories of materials and the basic quantities effect of all these parameters on the absorption
which affect their performance. properties of a material are not discussed further
here since the relationships are often complex and
do not always match empirical data well. The user
5.1 ABSORPTION MATERIALS is directed to chapter 10 of reference 4 for further
information on the subject.
5.1.1 Physical Theory The absorption coefficient, c_, is the single
most important macroscopic quantity of interest;
The mechanism of absorption involves con- it is defined variously according to the mode of
version of the kinetic and potential energy of inci- measurement used. It is nearly always highly
dent acoustic waves to thermal energy. This may dependent on the frequency of the incident sound
occur either via viscous losses due to oscillatory wave. A standing wave tube is used to measure the
motion of air molecules in absorption materials normal incidence absorption coefficient, %. This
such as porous foams, or via vibration of the entire measurement is made with a single frequency sinu-
material fabric, which may be the solid skeleton of soidal wave input by a loudspeaker, with a disc of
closed-cell foams, or impermeable panels at reso- the material under test placed at right angles to the
nance, as in wall panel absorbers. Fibrous mate- direction of sound propagation at the other end of
rials mainly absorb energy through the effect of the tube. Although this technique is suitable for
bending and friction between the fibers themselves. comparing the properties of absorbers quickly, it
For effective absorption, the pores and voids does not provide values which adequately match
in cellular materials must be highly interconnected. the typical performance in practical installations.
This is commonly referred to as the degree of To give a practical measure, a panel is inserted
reticulation. The specific parameter normally into a reverberation chamber and the absorption
measured in connection with this quality is known coefficient calculated as a function of the change
as the flow resistance, of which a crude assessment in the reverberation time (see section 6.2.2). This
may be made by judging the ease with which one case deals with a diffuse sound field where the
can blow air through the material. If the resistance sound waves are incident from approximately all
is too low, little energy conversion can take place angles at the material surface. This approach
within the material; too high a resistance limits air enables a statistical absorption coefficient, C_stat,
motion and hence reduces oscillatory and friction to be calculated. The noise field is usually gener-
losses. An interrelated quantity also of interest is ated by a random broad-spectrum source; quoted
the degree of porosity in the material, usually sim- specifications must correspond to measurement by
ply related to the density. A third quantity, also of this technique as specified in the standard testing
importance in describing the behavior of absorbers, procedure. 5 Unless otherwise indicated, all future
is the structure factor, which accounts for detailed references to absorption coefficients will implicitly
inner conformation and describes the effective denote use of Otstat. Table 5.1 gives some typical
change in fluid density which occurs within the absorption coefficients for a selection of noise
absorber voids. control materials.
Characteristically, it has been rather difficult Since the absorption coefficient is a measure
to predict the performance of the so-called bulk of the fraction of acoustical energy which enters
absorber materials without relying heavily on em- the material and is not reflected, a perfect absorber
pirical techniques. The recent interest of the air- would have _ = 1. Any practical absorber should
craft industry and NASA in the deployment of have a coefficient of at least 0.60, corresponding
bulk absorbers in jet engine nacelles has led to a to 60°70 absorption. This represents a reduction in
refinement in the analytical prediction of absorber acoustic intensity at the boundary of only 4 dB
behavior, l especially as a result of experiments (10 log (0.6/1) = 4 dB). Normally, absorption
with highly porous materials with a very uniform coefficients are quoted for octave bands between

44
 Table 5.1 - Sound absorption coefficients c_ for some typical noise control materials.

Octave band center frequency, Hz

Material
125 250 500 1000 2000 4000

Brick, unglazed 0.03 0.03 0.03 0.04 0.04 0.05

Brick, painted 0.01 0.01 0.02 0.02 0.02 0.02
Concrete 0.01 0.01 0.02 0.02 0.02 0.03
Concrete block, painted 0.10 0.05 0.06 0.07 0.09 0.08
Linoleum, rubber, or
cork on concrete 0.02 0.03 0.03 0.03 0.03 0.02
Wood 0.15 0.11 0.10 0.07 0.06 0.07
Glass
< 4 mm thick 0.35 0.25 0.20 0.10 0.05 0.05
6-mm plate 0.15 0.06 0.04 0.03 0.02 0.02
Plaster
On solid surface 0.01 0.02 0.02 0.03 0.04 0.05
On lath 0.10 0.10 0.06 0.05 0.04 0.03
Plywood, 10 mm thick 0.30 0.20 0.17 0.10 0.10 0.11
Gypsum, 12.7 mm thick
on 51 × 102-ram studs
410 mm apart 0.29 0.10 0.05 0.04 0.07 0.09
Fiber glass
25.4 mm thick, 24-48 kg/m3 0.08 0.25 0.65 0.85 0.80 0.75
51 mm thick, 24-48 kg/m3 0.17 0.55 0.80 0.90 0.85 0.80
25.4 mm thick, 24-48 kg/m3,
25.4-mm air space 0.15 0.55 0.80 0.90 0.85 0.80
51 mm thick, plastic
covering with perforated metal
facing 0.33 0.79 0.99 0.91 0.76 0.64
Polyurethane foam
6.3 mm thick 0.05 0.07 0.10 0.20 0.45 0.81
12.7 mm thick 0.05 0.12 0.25 0.57 0.89 0.98
25.4 mm thick 0.14 0.30 0.63 0.91 0.98 0.91
51 mm thick 0.35 0.51 0.82 0.98 0.97 0.95
Mineral fiber
12.7 mm thick 0.05 0.15 0.45 0.70 0.80 0.80
25.4 mm thick 0.10 0.30 0.60 0.90 0.90 0.95
12.6 mm thick on lath,
25.4-mm air space 0.25 0.50 0.80 0.90 0.90 0.85
Drape
0.35 kg/m2, hung straight
against the wall 0.03 0.04 0.1 ! 0.17 0.24 0.35
0.61 kg/m2, draped to
half area 0.14 0.35 0.55 0.72 0.70 0.65
Carpet
Haircord on felt 0.05 0.05 0.10 0.20 0.45 0.65
Thick pile on sponge
rubber 0.15 0.25 0.50 0.60 0.70 0.70

125 and 4000 Hz. A single-number rating, known Theoretically, maximum absorption occurs
as the noise reduction coefficient (NRC), is often when the absorber thickness is about one-fourth
taken; the NRC is the arithmetic average of the co- the wavelength of the lowest frequency of interest.
efficient values in the octave bands 250, 500, 1000, Then, throughout a complete cycle, the maximum
and 2000 Hz. This is not useful at extreme ends of air particle velocity always occurs within the mate-
the frequency range or for situations where a high rial. In practice, however, a gap may be left be-
degree of optimization at a particular frequency is tween the backing surface and the absorption
called for. material, as illustrated in figure 5.2a. Also, if the

45
 Acoustic particle 1.0
velocitymagnitude

placed against t-

wall (Cab
s = c) J
_E _-_6mm
o.5
(ii) Absorber spaced
.__/_
f_ _"_ (i) Thick absorber
approximately
X/4 from wall L18m 'j / J
J
.Q
<
placedagainst o
J Ji
-_ _- XI4 wall (Cab s < C) 125 250 500 1ooo 2000 4000
Frequency(Hz)

(a) (b)

Figure 5.2 - (a) Choice of nominal positions of bulk absorber materials from a hard surf ace for optimum
absorption, where Cabs is the velocity of sound propagation in the absorbent material and c is the ambient
velocity of sound propagation. (b) Typical effect of material thickness on sound absorption (reference 6)
(used with permission of Acoustical Publications).

speed of propagation through the absorber is sub- vibration. Figure 5.3 shows some typical foams
stantially less than in free air, the effective wave- used for noise reduction.
length is contracted and so the thickness of mate-
5.1.2.2 Fibrous materials
rial needed is reduced. The influence of increased
thickness on absorption, especially at low frequen- Fibrous materials are usually constructed of
cies, is shown clearly in figure 5.2b. 6 glass fiber or mineral wool; their acoustical prop-
erties are varied by modifying the diameter of the
5.1.2 Description of Common Absorbers fibers, packing density, and thickness of the
finished product. Usually the fibers are cemented
The values of absorption coefficients depend together at contact points with a suitable resin,
on the physical mounting of the absorber, surface although unbonded glass fiber blankets are also
finish, and the immediate surroundings, as well as available. A major advantage of fiber blankets is
on their own innate characteristics such as thick- that they have good inherent fire-resistant proper-
ness and porosity. Consequently, the values culled ties. However, they have to be used with great care
from tables should be regarded as approximate and in situations with high aerodynamic flows or vibra-
may need to be changed depending on the situ- tion levels, since their delicate nature, especially at
ation and configuration used. exposed edges, can lead to a great deal of fiber
shedding.
5.1.2.1 Foam
To alleviate shedding, and to improve their
Foam is a very versatile material, especially attenuation characteristics, particularly in the
because its acoustic properties are easily controlled lower part of the frequency range, a perforated
and reproduced by appropriate modification of impermeable outer cover is sometimes placed on
the manufacturing process. Polyester or polyether top of the blanket. Figure 5.4 shows the typical
formulations are common, although the former acoustic effect of using such a protective covering
permits a more close specification of cell size and on top of any porous layer. This essentially makes
reticulation. Cell sizes of between 5 and 50 pores the material behave like a series of Helmholtz
per centimeter are readily achievable. Foam may resonators (see section 5.1.2.3). The effect of the
be compressed under heat and pressure to a per- perforations is especially noticeable at low fre-
manent set; moreover, it is easy to change the quencies, when the total open area is less than
thickness and density. Furthermore, because foam around 25 percent. By varying the porosity, hole
represents a completely integrated material, no size, and hole spacing, a higher and increasingly
fiber shedding or material decomposition is likely sharper absorption peak may be obtained in con-
to occur from the effects of sustained air flow or trast to the broadband absorption characteristics

46
 = .

Figure 5.3 - Acoustical foams. Top, by courtesy of Specialty Composites Corp. Bottom left, Eckousta-
Foam being applied (Eckel Industries, Inc.). Bottom right, Eckoustic-Klip fastener for installing sound
control materials (Eekel Industries, Inc.).

Figure 5.4 - Effect of placing a perforated im-

permeable cover on a porous acoustical material.

0f
Porous
,7"2"/-/"/Z7 layer
N
common to most acoustical foams and fibrous '='¢==_ Porous
0.5 /
layer +
materials. This must be traded off against the
perforated
deterioration in high frequency performance be-
cover
cause these components cannot diffract through .a 100 1000 10000
the holes. Frequency (Hz)

47
 5.1.2.3 Helmholtz resonators where

Helmholtz resonators are cavities connected c is the speed of sound, m/s

S is the area of orifice, m2
to the ambient atmosphere by holes or slots in a
well-defined geometrical pattern, as indicated in L is the length of orifice neck, m
V is the volume of cavity, m 3.
figure 5.5. When the dimensions of the cavity are
small relative to the wavelength of the sound, then
Normally, the height and width of the slots
the air in the "neck" or connecting orifice oscil-
lates like a plug between the atmosphere and the into the cavities are varied to achieve optimum
absorption at a nominal frequency. Resonators
air in the cavity, resulting in high viscous losses. A
very narrow absorption spectrum is associated are principally of value for attenuating prominent

with Helmholtz resonators, and they can be tuned low frequencies below around 400 Hz. However,

for maximum absorption at a particular frequency by filling the neck or cavity with a porous or
fibrous absorber, it is possible to broaden the
by use of the following relationship4:
effective absorption range of these, albeit with
some penalty in terms of the peak absorption, as
shown in figure 5.5. The absorption range may
also be broadened by a series of resonators
fres = _ L + 0.8 "4'S) V Hz
coupled together laterally; a parallel configuration

Orifice area, S
Cavity neck
length, L

"///
/
tgL_ /
/
/ /

I I I
/ 100 1000 10000
"////////,
(a) 1.0

.__
o

/
/
/ 0.5
/
/
/
/ , I 1 I
/ i

/ /////////', 100 1000 1o0130

(bl

/
jjjjjj jjj(c)
0.5

100 1000
I
10000

Frequency (Hz)

Figure 5.5- Typical spectra for several Helmholtz resonator configurations. (a) Simple resonator.
(b) Absorbent lined resonator. (c) Parallel-coupled resonator pair.

48
is particularly effective for sound entering at a
tangent. 7 Reference 7 also gives details of
optimum geometries for designing coupled reso-
nators. An increase in energy attenuated of up to
50 percent as well as a significant increase in effec-
tive bandwidth may be gained. Figure 5.6 shows
some examples of single resonator units used in
industry.
Helmholtz resonators, mounted either as
single units or as part of a series with complete
access between cavities, such as perforated plate
coverings (see section 5.1.2.2), are sometimes
referred to as single degree of freedom systems be-
cause acoustic intensity at the resonator orifice
can only propagate normally to the surface. It is
also possible to stack several layers of coupled
resonators, with different controlling parameters,
on top of each other.
Extensive analytical work on the use of perfo-
rated plate liners in aircraft jet and fan noise
reduction programs has highlighted several inter-
esting phenomena of particular interest for duct
J
noise control problems, where grazing incidence at
the absorption liner boundary predominates. One
study 8 showed that for the same porosity, flow
resistance decreases as the number of holes in the
plate is increased, with a concomitant decrease in
hole size to retain the same total open area. It has
also been shown that the nonlinear acoustic im-
pedance effects at the absorber-air boundary,
which is critical to the value of the absorption co-
efficient, may be treated as a high amplitude
acoustic spectrum without grazing flow. 9
The extreme environment of jet aircraft
engines has resulted in considerable develop-
ment of the broadband resistive resonator
concept, l°,JI where some kind of metal fibrous
layer is placed over a resonating absorber, often a
metal honeycomb structure cut to vary cell depth
across the panel. Such structures usually have a
significantly broader attenuation bandwidth than
the best perforate
compares
plate-faced systems;
the merits of the two systems. Figure 5.8
figure 5.7
!
shows some honeycomb absorbers developed by
NASA.
Devices for more general application based
on similar principles have been developedl2;
although these are often more expensive than
traditional absorbers, they are of use in rigorous
industrial environments where, for example, high
temperatures and contamination would detract Figure 5.6 - Illustrations of Helmholtz resonator
considerably from the performance and suitability configurations (by courtesy of The Proudfoot
of foams and fibrous materials. Another design Company, Inc.).

49
 Narrow-band
resonator

Felt, fiberglass,
Broad-band
etc. .o
homogeneous
absorber
iliii
!!!jiiiiiiii!! !!!!!ii!i!
Broad-band Honeycomb --_ /_ Porous
resistive
cells ..._---.-; ............ ..//,layer
resonator

i1'1iii
Frequency (Hz)

Figure 5.7 - Comparison of absorption spectra for perforated plate and broadband resistive resonators,
and a homogeneous blanket absorber (reference 10).

Lining materials and construction Multilayer wall construction

Perforated plate-honeycomb Porous layer-honeycomb

Bulk absorber tc_.c_._L_¢_

Double layer

Figure 5.8 - Examples of honeycomb and perforate plate-faced absorber linings (by courtesy of NASA
Lewis Research Center).

developed by NASA to help suppress screech in 5.1.2.4 Functional absorbers

rocket motors utilizes longitudinal slots cut into "Functional absorbers'" is a generic term
mild steel liners with a backing cavity depth of used for highly efficient sound absorbing panels
around 2.3 cm; these slots essentially act as suspended from a ceiling. They offer greater
coupled Helmholtz resonators. 13 Combinations of absorption than that possible with absorption on
slot widths were used to achieve the broadband wails and are often used because of limitations on
performance needed in long and narrow com- affixing panels directly to the room surfaces
bustion chambers of this type. because of lighting, maintenance access

50
requirements,andsoon(seesection 6.2.2).Their Functional absorbers are used to control the
performanceis determinednot only by the reverberant sound field. Appreciable absorption
intrinsicabsorptionpropertiesof the material can often be added to an industrial shop space
fromwhichtheyareconstructed, butalsobytheir only in the ceiling area, and the use of functional
relativespacing
and,to alesser
extent,theirshape. absorbers often conveniently avoids interfering
with permanent fixtures. They are usually
5.1.3 Applications of Sound Absorbers suspended vertically, and their individual
performance improves substantially as the sep-
Sound absorbers are especially useful in aration distance between them is increased. 14
helping to control the reverberant sound field, Figure 5.10 shows some typical configurations and
although the degree of additional absorption their associated sound absorption values. Stag-
needed to effect a reduction follows a logarithmic gering the panels horizontally and vertically can
law of sharply diminishing returns (see section also enhance the total absorption.
6.2.2). Frequently, absorbers are used to line
enclosures or partial barriers to prevent the
Absorption on ceiling
particle build-up of reverberant sound caused by
and walls
components reflected from the inner surfaces. _-", ,'_l_. _ ....... 1
This is illustrated in figure 5.9.
Resonators can be used successfully where N
prominent low frequency tones are present, such \ I '_\ J. "ll
\ j Reverberant ">--/" "_
as electrical hum from main transformers or
acoustic radiation from a machine with a low \ / field .-'_" \\ ," ]
,/// //_/Z/ /_ ...... "/ ,_////;'//. ;/////)
shaft rotational speed. They may be built into the
- Local noise
structure of the building at the design stage, fitted control barrier
source
as side-branch elements to ducts (see section
7.3.4), or added retroactively as independent units Figure 5.9 - Illustration of absorber use in help-
suspended in space (see figure 5.6). ing to control direct and reverberant fields.

Plan configuration

1 2 3

0.6 m .--, _ Microphone

Source
1_ 4.6 rn

Frequency Insertion loss (dB)

(Hz) 63 125 250 500 1000 2000 4000 8000

Configuration 1, 2.7 4.3 4.8 5.0 4.6 4.1 4.1 3.9

0.5 m spacing
Configuration 2,
1.7 3.2 3.9 5.0 4.6 4.1 4.0 2.8
0.9 m spacing
Configuration 3,
2.6 3.0 3.9 4.9 4.6 4.1 4.2 3.1
0.9 rn spacing

Figure 5.10 - Typical functional absorber arrays and associated sound absorption values (reference 14).
Measurements made in reverberation room; panel dimensions are 1.2 x 0.6 x 0.04 m. (Used with the
permission of the Noise Control Foundation and J. D. Moreland.)

51
5.2 INSULATING MATERIALS only a relative quantity, a partition providing a net
30 dB of noise reduction reduces incident sound
5.2.1 Physical Theory levels of 90 and 75 dB to 60 and 45 dB, respec-
tively, at the other side.
Insulating materials reduce the amount of
sound transmitted from one space to another by 5.2.1.1 Mass control
maximizing the impedance mismatch. This Figure 5.11 shows an idealized plot of trans-
mismatch reduces the amount of acoustic energy mission loss changes with frequency, illustrating
entering and leaving the wall or partition con- the dominant features and controlling parameters.
cerned. At the same time it is often useful to have
The most uniform portion of the curve is con-
a high internal damping factor (see section trolled solely by the mass of the partition; for a
5.3.1.1), which attenuates the acoustic energy single panel the curve rises at the rate of 6 dB per
traveling through the partition by converting it doubling of mass and 6 dB per octave increase and
into heat. Consequently, many noise insulators
hence is linear. In practice a number of expres-
are as massive as possible and quite limp, where sions relate transmission loss to the frequency and
stiffness is not a structural prerequisite. to the density of the material. The normal-
The acoustical performance of any insulating
incidence mass law is an analytical expression
material is described by a parameter known as the
derived by assuming a sound field only incident at
transmission loss, R, which is highly dependent on angles normal to the surface, i.e., at 0°:
frequency. This is measured in decibels and is re-
lated to r, the transmission coefficient defined in R n = 20 Iog fM- 48 dB
section 5.0, by
where
R = 10 log (l/r) dB R n is the normal incidence transmission loss,
dB
To compute the transmission loss it is
f is the frequency, Hz
necessary to make measurements in a specially
designed facility. 15 This is performed by mounting M is the superficial density of insulator,
kg/m 2.
a sample of the material under examination in an
aperture between two reverberation rooms. The
completely sealed aperture, and the massive Note: When dealing with insulating materials
construction of both rooms, ensures that the only it is usual to speak of mass in terms of superficial
density, which accounts for the total mass
significant path of transmission is through the test
panel. Measurements are taken of the sound throughout a unit surface cross-sectional area of
the material.
pressure level in both rooms, in sixteen one-third
octave bands, after activation of a broadband By making a further assumption that the
noise source in one of them. The transmission loss sound field is diffuse and incident at all angles
is computed as from 0 ° to 90 °, then the random-incidence mass
law is defined as
R = NR + 10log(S/A) dB
Rrandom = Rn - 10 log 0.23R n dB
where
NR is the SPL source minus SPL received in where Rrandom is the random incidence trans-
mission loss, dB.
each one-third octave frequency band,
dB The field-incidence mass law is a variant of

S is the total area of exposed test panel, m2 this, again assuming a diffuse field, but only
accounting for incident angles between 0 ° and
A is the total absorption in receiving room,
metric sabins. 78 ° . In practice it has been found to be the most
accurate and maintains a 6 dB increase with
In practical situations, possible discrepancies doubling of either mass or frequency, as con-
from this idealized measurement procedure may trasted to the 5 dB increase of the random-
occur as a result of alternative paths of trans- incidence law. It is given by
mission and inhomogeneities in the separating
Rfiel d = R n - 5 dB
partition; these are dealt with in section 5.2.3. A
full range of typical sound insulation values is where Rfiel d is the field incidence transmission
shown in table 5.2. Because sound insulation is loss, dB.

52
 Table 5.2 - Transmission loss values R for some typical noise control materials.

Material Octave band center frequency (Hz)

Superficial
Description density 125 250 500 1000 2000 4000 8000
(kg/m2)

Lead
1.5 mm thick 17 28 31 27 32 32 33 36
3 mm thick 34 30 32 33 38 44 33 38
Lead vinyl 2.5 11 12 15 20 26 32 37
5 15 17 21 28 33 37 43
Aluminum
20 gauge, stiffened 2.5 11 10 10 18 23 25 30
Steel
22 gauge 6 8 14 20 23 26 27 35
20 gauge 7 8 14 20 26 32 38 40
18 gauge 10 13 20 24 29 33 39 44
16 gauge 13 14 21 27 32 37 43 42
Plywood
7 mm thick 3 17 15 20 24 28 27 25
20 mm thick 10 24 22 27 28 25 27 35
Sheet metal
Viscoelastic core 10 15 25 28 32 39 42 47
Plexiglass
6 mm thick 7 16 17 22 28 33 35 35
13 mm thick 14 21 23 26 32 32 37 37
25 mm thick 28 25 28 32 32 34 46 46
Glass
6 mm thick 15 11 24 28 32 27 35 39
9 mm thick 22.5 22 26 31 30 32 39 43
25 mm thick 62.5 27 31 30 33 43 48 53
Double glazing
2 x 6-ram panels,
12-mm gap 30 23 24 24 27 28 30 36
2 x 6-mm panels,
188-mm gap 30 30 35 41 48 50 56 56
2 x 6-ram panels,
188-mm gap with
absorbent in reveals 30 33 39 42 48 50 57 60
Double walls
2 x 280-mm brick,
56-mm gap
2 x 12-mm plaster 380 30 38 47 61 75 80 81
2 × 12-mm insulating board
50 × lO0-mm studs 19 16 22 28 38 50 52 55
Sandwich partitions
2 x 5-mm plywood,
1.5-mm lead 25 25 30 34 38 42 44 47
2 x 18-gauge steel,
9-mm asbestos board 27 22 27 31 27 37 44 48
Masonry, brick, 125 mm thick
Plastered both sides 240 36 37 40 46 54 57 59
Brick, 255 mm thick
plastered both sides 480 41 45 48 56 65 69 72
Concrete block, 152 mm thick 176 33 34J 35 38 46 52 55
Concrete, 103 mm thick 234 29 35 37 43 44 50 55

53
 v

z00'° : , ///
t- 0 I o 0,_- ." Rn, Normal incidence (0°), //i/

t
o . I o I oo L-< ." /
Rfie_d, Field i
_ _, _ :_ -. x 40 -- -----_ .... "
°
o4
.c S
I .J_/ _ Coinodence ¢: 30 //,//¢]- Rrandom, Random incidence

20 /._'/ I ,0to_, i

.... Effect of damping

,_o
Panel resonance frequencies

i
dip
-
.-g
.__
_ o
10--

Critical c 100 1000 10000 100000 1 000000

E
frequency
fp = frequency x surface density (Hz x kg/m 2)
Frequency (Hz)

Figure 5.12- Transmission loss curves valid for

• Figure 5.11 - Idealized transmission loss curve. mess-controlled limp panels (reference 4). (From
Leo L. Beranek, Noise and Vibration Control,
A rapid calculation of any of these quantities Copyright © 1971 by McGraw-Hill, Inc. Used with
may be made using figure 5.12. All three relation- thepermission of the McGraw-Hill Book Company.)
ships are valid only below the critical frequency of
the panel of interest.
Approaching sound
5.2.1.2 Coincidence effects

The critical frequency occurs when the speed

wave_ X = sound
of sound in air coincides with the propagation wavelength
speed of free bending waves in the partition.
Above this frequency, the free bending wave speed ka = bending
increases still further as a result of dispersive wavelength
effects (see section 2.1.4), thus allowing for a
range of frequencies at which the wavelength of
airborne sound waves projected onto a panel
coincides with the wavelength of free bending or
flexural waves for that particular panel, as shown
in figure 5.13. At the critical frequency this occurs
for grazing-incidence airborne sound waves. At i Vibrating partition

Coincidence occurs when

higher frequencies the effect occurs at angles
increasingly less acute to the surface since the _1p xa = --_-x
sin 0
wavelength of flexural waves decreases less than Direction of
that of corresponding airborne sound waves. The bending wave
net result is a marked reduction in transmission
Figure 5.13 - illustration of the coincidence
loss because of easily induced vibration. This is
effect. (From /an Sharland, Woods Practical
shown in figure 5.11; the coincidence dip in the
Guide to Noise Control, 1972; used with the per-
transmission loss curve usually extends over an
mission of Woods Acoustics.)
octave or so above the critical frequency.
Table 5.3 lists critical frequencies for some
typical insulating materials. The size of the coinci-
dence dip in the transmission loss spectrum nounced dips. In addition, denser materials tend
depends mainly on the homogeneity of the to have a lower critical frequency. Consequently,
material and the presence of damping. Broad and coincidence effects may be mitigated by the use of
shallow dips tend to occur with plastic and heavy additional damping, or by changing the thickness
masonry materials, while glass, most metals, and and hence superficial density of the panel, so that
lining materials such as plasterboard and the dip occurs outside the frequency range of
hardboards exhibit rather sharper and more pro- concern.

54
 Table 5.3 - Critical frequencies and surface excellent stiffness properties, 21 and the ultimate
densities of some common materials. choice is determined by the necessity for other
structural properties. Honeycomb materials are
Critical frequency Surface density particularly valuable in that they possess a high
× surface per unit stiffness/mass ratio; figure 5.14 shows the effect
Material
density thickness of applying a paper honeycomb onto an alumi-
(Hz.kgm-Z) (kg m 2mm-l)
num panel with a constraining layer on top. Stiff-
ening with ribs, stringers, and so on increases the
Lead 600 000 11.2 number of radiating areas operating indepen-
Partition board 124 000 1.6
dently. Such stiffening increases the acoustic radi-
Steel 97 700 8.1
Reinforced concrete 44 000 2.3 ation efficiency of the panel in the mass-controlled
Brick 42 000 1.9 region, thereby reducing available insulation,
Glass 39 000 2.5 while at the same time raising the fundamental re-
Plexiglass 35 500 1.15 sonance frequencies and improving the low fre-
Asbestos cement 33 600 1.9
quency insulation performance. Figure 5.15 illu-
Aluminum 32 200 2.7
Hardboard 30 600 0.81 strates this result.
Plasterboard 32 000 0.75
Flaxboard 13 200 0.39 5O
Plywood 13 000 0.58
40
From lan Sharland, Woods Practical Guide to Noise z
Control, 1972. Used with the permission of Woods
Acoustics. O 30
\.
X
\.
5.2.1.3 Stiffness control 20

The dips in transmission loss at the low o 10

C l_,e'_ 7 Bale (047 kgl
frequency end of the spectrum (see figure 5.11) .o_ .... Sliffened with 13 crn paper
result from the fundamental resonances in the 0 -- honeycomb 1070 kg)
----- Stiffened and 002 cm
panel itself. The position and amplitude of these ¢- constrmn_ng layer (082 kg)
-10 I
resonances are governed by the inherent stiffness 2O 1O0 1000 5000
of the panels as well as by their inherent damping Frequency (Hz)
and size.
Figure 5.14- Effect of paper honeycomb as a
The effect of stiffening on these resonances stiffener on an aluminum panel (reference 20).
has been studied by NASA and other organiza- Panel dimensions are 45.7 x 45.7 x O.1 cm; pure
tions in the aerospace field, 16-19 to elucidate tone testing at normal incidence only to panel.
techniques for increasing the low frequency trans-
mission loss of aircraft fuselage panels. Stiffening
Mass law for 0.8 mm aluminum
increases the fundamental resonance frequencies

.,..,ou.
and essentially shifts the transmission loss curve to
the right on figure 5.11, making it difficult for any
major low frequency exciting frequencies to 20 2 V with single
induce resonance conditions. However, increasing 0[- _,] aluminum stiffeners
/ _1 I I I I
the density or net mass of the insulating panel i 102 103 10 4 1_- 103 104
conversely reduces the fundamental resonance ._ One-third octave
frequencies. Thus achieving extra stiffness by
i I- fl = 227 Hz / frequency (Hz)
increasing the panel thickness is not the most
effective way to improve panel behavior in this z 20}_ " with double
area. 2° Ideally, stiffening or damping (discussed
0 I- , stiffeners
in section 5.3.1) should be incorporated in the 102 103 104
One-third octave
areas where the largest panel deflections occur;
frequency (Hzl
these are normally the result of the first funda-
mental resonance mode of the panel. Figure 5.15- Effect of rib stiffeners on an alu-
Honeycomb lattices, fiberglass, and so-called minum panel (reference 16). f] is the fundamental
"noiseless" or highly damped steel, all have panel resonance.

55
 The aircraft noiseprogramsalsoestablish
thataclosedcavitybehindtheinsulating panelcan
be disadvantageous, because strongacousticre-
sonances maycouplewiththepanelresonances to
produce a verylowtransmission loss.17,22In fact, Z

resonant modes in a receiving

roomareoneof the o

principal causes of frequency-dependent ×

variationsin low frequencysoundinsulation, 23 ¢.q

40 dB
andpredictionmethods havebeendeveloped for
the overallresponse of a panel-room or panel- o
e-
cavity system. 16,23Structuralgeometryis of ._o
----STC 45 contour
specialconcernheresinceit affectsbothacoustic 30
and structural resonances.For instance, _TL
with
curve
STC
for
=
partition
45
cylindricalandsphericalpanelsare"membrane I---

controlled,"unlike flat rectangular or circular

panelswhere"flexuralcontrol" is present.24, 25 20
125 250 500 1000 2000 4000
Theformeraremuchmoreresistantto volume One-third octave frequency (Hz)
displacements, anddesign chartshavebeendrawn
for optimalgeometries for panelswith similar Figure 5.16 - Calculation of STC ratings.
dimensions to theincidentwavelength inthestiff-
ness-controlledregion. As a rough guide, handle noises with prominent spectra in the
"membrane control"is achieved whenthelength middle and high frequency ranges and, apart from
of a cylinderexceeds thegeometric meanof the obscuring serious dips in the transmission loss
sumof thecylinderradiusandwallthickness, and spectrum, they are not strictly relevant to many
whenthedepthof a spherical segment exceeds the spectra found in industry, such as those with
radiusof curvature. 24
strong discrete tones or prominent low frequency
Analyticalworkindicatesthatclamped-edge components.
platesgive better noise reductionbelow the
Generally, for good acoustical performance
fundamental frequencies, 17althoughthis is not throughout the average industrial noise spectrum,
trueof theresonantfrequencies themselves since a barrier should be as limp and massive as possi-
anunclamped platehashigherinherentdamping. ble. It is also important that any weak paths, such
A slopeof -6 dB/octaveincrease normallypre- as air gaps or poor insulators such as windows and
vails in the stiffness-controlled region,except doors, be reduced or eliminated since their effect
wheresignificantcavity-coupling effectsoccur, is disproportionate to their total exposed area (see
whichmayflattenthe transmission curvesome-
section 5.2.3). Hence for brick or lightweight walls
what.
with a relatively high porosity, it is advantageous
to use a plaster or alternative impervious finish to
512.2 Description and Rating of Common Sound
Insulation Materials keep the transmission loss as high as possible, even
though this may not add substantially to the mass.
As for absorption materials, it is often Lead is frequently used as an insulating
convenient to describe insulation in terms of a material because of its high density, limpness, and
single number rating. This rating is known as the good internal damping, which combine to produce
sound transmission class, STC, and is found by a response which follows the mass-controlled
matching the transmission loss data against a transmission loss curve fairly uniformly without
standard transmission loss curve in a rigorously any serious coincidence or resonance dips (see
defined manner.26 The STC rating is found from figure 5.11). It may provide additional mass by in-
the value of the standard curve at 500 Hz. cluding pulverized or sheet lead to form composite
Basically, the measured values should be aligned materials such as shot-loaded vinyls and other
such that they fall below the standard curve no plastics. Sometimes these are fabricated in curtain
more than 2 dB on average in each frequency form so that ease of access to a machine is main-
band, and no greater than 8 dB in any one. An ex- tained.
ample of a typical calculation is shown in figure Where spatial restraints are not severe, dou-
5.16. STC ratings were developed specifically to ble panels are a very convenient way of increasing

56
the availableinsulationwithoutaddingto the air-mass resonance occurs when the air between
overallmass.Sincemosttransmissionlossis ob- the leaves acts as a spring and the whole system
tainedby impedance mismatching,
subdividing behaves as a single highly resonant unit. The fre-
thetotalweightinto twoseparate
partitions,thus quency at which this occurs, fm a m, is given by
increasingthe numberof air-insulatorboun-
daries, gives considerable advantage, particularly
6OO
in the mass-controlled region. To ensure max- fm-a-m = Hz
imum benefits, the gap between the two leaves _/ mlln2d
should be as large as possible with at least one of m i +"_
the partitions resiliently mounted to prevent any where
structure-borne vibration from prejudicing the in- ml, m 2 are the superficial densities of two
dependent action of each panel. A comparison of leaves respectively, kg/m 2
double and single panels with the same total d is the separation distance, cm.
equivalent weight is shown in figure 5.17. 27
Ideally, the slope of the double leaf transmis- This effect reduces the overall performance
sion loss curve should be 12 dB/octave and the ab- of the double panel to below that for an equivalent
solute value equal to the arithmetic sum of the two single one unless the separation distance d is large
elements. However, in practice this does not occur enough. The adverse effects of standing waves in
for several reasons. At low frequencies a mass- the air gap may be reduced by introducing absorp-
tion into the cavity; for double-glazing only the re-
veals may be treated. In addition, mechanical con-
nections, if not resiliently isolated, can bridge the
cavity and permit direct transfer of high frequency
vibration across the structure. Staggering the studs
e° E f "_ to which the leaves are fastened is of value in pre-

50 // venting such transfer.

eral of these points.
Figure 5.18 illustrates sev-

Z
Serious coincidence problems are avoided by
"7
O making the two leaves of different materials or
X thicknesses. In one of the most comprehensive ex-
perimental and analytical studies of sound insula-
,o - / tion elements to date, 28 a series of expressions for
o
t-- f 2/ '-..-''\ ./ predicting transmission losses were derived. Most
._o
.! _,,./.........-;.i-- ,._,. of these
provide
are for multiple-leaf
the most cost-effective
partitions,
means of gaining
which

high insulation. Examples of double panels in-

I--- 2o-- t./ clude double-glazing, gypsum board on stud
-/
_- Flexible Absorption
lo I,,I,,I,, ll,llll, _- in reveals
125 250 500 1000 2000 4000 steel

One-thirdoctavefrequency(Hz) spring
Resilientseal
Figure 5.17 - Comparison of single and double (neoprene
gasket,etc.)
leaf partitions of same equivalent weight
(reference 27). A, curve for single-leaf 12-mm gyp-
sum board; B, curve for two 12-mm leaves rigidly
bonded; C, curve for two 12-mm leaves on either
side of 90-mm steel channel studs; D, curve for Staggeredstud Resilientl Resiliently
configuration as in C, but with 50-mm fiber glass partitions mountedstud sealeddouble-
partition glazedpartition
in stud space. (From Cyril M. Harris, Handbook with absorption
of Noise Control, Second ed., Copyright © 1979
by McGraw-Hill, Inc. Used with the permission of Figure 5.18 - Illustrative designs for double parti-
the McGraw-Hill Book Company.) tions.

57
walls,andwood-or expanded metal-lathceiling the coincidence frequency, often into an area of
andwallsystems. critical importance. In consequence, it would be
Fornoisetransmission througha pipeor duct desirable to have a material exhibiting high stiff-
wallit isoftenbesttoaddasecond skintotheout- ness at low frequencies to provide insulation at
sideof thewallratherthanto increaseitsmassor frequencies below the panel fundamental, and a
to adddamping.Normallytheouterskinis con- lower stiffness at high frequencies. One way of
structed of leador cement
screed,
of around10 to achieving this is by the use of laminated sandwich
20 kg/m 2 superficial density; a sufficiently thick panels. One study 29 has considered several op-
resilient layer between the screed and the inner timization procedures in terms of types of core
wall prevents undue coupling and allows the struc- and panel materials, or changes in the thicknesses
ture to act as a double-walled construction. This of these, and concluded that two approaches may
treatment must be applied over the entire length of improve the overall insulation performance. One
a duct or pipe where high surface vibration exists, approach, which is sometimes impractical, uses
including side-branch turns, valves, and so on. less stiff and more massive panels effective over a
The most important characteristic in deter- greater frequency range; the other increases the
mining the overall transmission loss of a panel is core stiffness. At high frequencies, the shearing
its mass, which may be increased by increasing effect in the core reduces the total stiffness of the
either the density or the thickness. However, high combination to something approaching that of the
thickness, and hence stiffness inevitably reduces constituent materials.

Limit for area ratio

of apertures to maintain
stated insulation
1:1000000

1:500000
I I I I
1:250000

1 : 125000

1:64000
\ \ N "\ _ $1 "--

TL 2 TL_
1:32000

1:16000
\\N\\
1:8000

1:4000

1:2000
\\,\ \ x
1:1000 _, __ _?\_.6O DifferenceTL, - TL2
1:500

1:250

1:125

1:64
• --),.
\\
:::
1:2
,\\ \ \
\\ \10 \\ \ _ \ \ _ \ \ \
4:1
8:1
1 \ \\\\\\\\\
\ \ % \ \ \i \ _
\
\
0 5 10 15 20 25 30 35 40 45 50 55 6O
Loss of insulation (subtract from higher insulation, TL 1) (dB)

Figure 5.19 - Nomogram for computing the sound reduction index of composite partitions.

58
 5.2.3 Installation of Sound Insulating Elements Note that if the air gap under the door
threshold is ignored, an overall net transmission
It is imperative to make at least two calcula- loss of 45 dB would be indicated, which shows the
tions before actually installing sound insulation. deleterious effect of small air leaks and other poor
The first calculation gives the effective transmis- insulating elements on the overall performance.
sion loss resulting from inhomogeneities within The calculation is performed across the entire fre-
and around the installed panel such as air leaks, quency range, usually in octave bands.
glazing, service conduits, and the like. The The second calculation of importance in-
average transmission loss R is readily found accor- volves an estimation of the actual change in sound
ding to the relations pressure level at a point after the harrier or in-
sulating element has been installed. This is known
r|S 1 + T2S 2 + • • • + 7nS n as the insertion loss; it should take account of the
transmission loss, absorption of the reverberant
Sr
sound field within enclosures, etc., and any flank-
where ing paths which may dominate the overall noise
S T is the total area of partition, m2 reduction after the insulation of the installed par-
Sn,T n are the area and transmission coeffi- tition reaches a certain level. The use of absorp-
cient of the nth insulating element, tion when constructing barriers or enclosures
respectively prevents the buildup of reverberant sound fields
_-is the average transmission coefficient which would demand a higher insulation than in-
for the entire partition; itially planned. Examples of flanking transmission
and the means of assessing and coping with it are
= 10 log (I/Y) dB discussed in section 6.2.3. Because the building
and construction techniques used in the field may
For convenience, a nomogram can be used be quite different from those utilized when mak-
for quickly calculating the overall transmission ing laboratory measurements, it is often wise to
loss (figure 5.19). Two elements are taken at a make conservative performance predictions by
time and their net transmission loss computed subtracting 3-5 dB from the quoted transmission
before incorporating a third and so on. losses. Common causes of this degradation in per-
Example: For a wall with specifications shown formance are structural bridging caused by incor-
as rect placement or construction of ties between
walls, rubble in-filling, etc., or poor sealing of
cracks around doors, windows, and conduit
Wall: 18 m 2 [

i
50 dB penetrations.
Door: 1.8 m 2
37 dB

5.3 VIBRATION CONTROL MATERIALS

Air leak: 1.8 x 10 3m 2
0 dB Vibration control materials can redirect or
dissipate as heat acoustic energy entering a struc-
Net R for door + air leak: ture and appearing as structural vibrations. This is
area ratio of door : air leak = l:1000 important, because once energy enters a solid
Hence, from figure 5.19, material, it is possible for propagation to proceed
high R - low R difference = 37 dB, loss of along several different paths which may be dif-
insulation = 8 dB ficult to predict, later to be re-radiated as sound
Therefore, waves in the atmosphere (see figure 6.13).
net R, door + air leak = 29 dB
Net R for corrected door + wall:
area ratio of door : wall = 1:10 5.3.1 Vibration Damping
Hence
highR - low R difference = 50 - 29 5.3.1.1 Physical principles of damping
= 21 dB, loss of insulation = 9 dB
Therefore, Damping materials dissipate energy as heat
net R, for entire partition = 41 dB somewhere within the structure to which they are

59
applied,through friction, or less frequently, the following ratio, _', is often used as a
viscouseffects, as well as the well-known comparative measure between alternative
hysteresis effect. Hysteresis losses occur because, materials:
after deformation caused by an applied force,
C
many elastic materials do not regenerate the same Damping ratio, _" =
amount of applied vibrational energy when Cc
returning to their original equilibrium position.
where
Damping is typically most effective when
C is the damping coefficient of material
conditions of large displacement amplitudes are
under consideration
found, and this is most common when resonance
conditions prevail. For this reason they are usually Cc is the critical damping coefficient.
used to complement insulator performance in the
critical frequency and fundamental panel A variant on this is the loss factor, rt, which is
related to the damping ratio by
resonance ranges.
Damping materials may be used to control 1
the radiation of sound from panels at the surface -q_
of a machine, where the resonant frequencies of 2r Wo
the panels may be excited by either a broadband
spectrum or discrete frequencies close to a 2C
r_ - or 2_"
resonant frequency. Damping also reduces impact Cc
noise. The addition of surface damping has a
twofold effect in this regard: it considerably where
modifies the time history in which the largest D O is the total energy dissipated by damping
frequency-transposed elements are well away from during one cycle
the lowest natural frequencies of the structure. W 0 is the total vibrational energy of system.
Figure 5.20 shows how this may be achieved to
produce a small force acting over a long time
Decay rates may range from 5 dB/s to over 80
period as opposed to a large force acting for a
dB/s and damping ratios from 0.5 to 20 percent of
short time.
critical damping. Quite often, extra damping of
Damping is measured in several different
significance can only be added when the damping
ways. The decay rate in decibels per second ratio is less than 10070.
expresses the reduction in radiation from vibrating Until very recently there have been no really
surfaces, once the exciting force is removed, with
well-defined approaches to choosing damping
respect to a standard plate and shaking rig.
materials for a particular situation. Successful ap-
Critical damping measures the damping necessary
plication of damping control depends on a high
to just prevent oscillation and when expressed in
degree of optimization after careful consideration
of the following factors: temperature, frequency,
static or dynamic loading, uncertainties in the
precise nature of the incident spectrum,
knowledge of the inherent damping of structures,
hard surface _>
and structural behavior under impulsive loading,
== i.e., linear or non-linear. However, owing in large
part to pioneering work on acoustic fatigue and
2 _
other vibration related problems in the U.S.
Same impact 1o aerospace industry, pioneered particularly by
on resilient
researchers at the Wright-Patterson Air Force
surface
Base, a number of important advances have been
Time, t Frequency (Hz) made.30,31
Figure 5.20 - Change in frequency spectrum for A temperature-frequency equivalence princi-
an impact due to damping. The area under the ple has been elucidated to permit the reduction of
force-time curve in each case is the same and must several highly specific parametric variations to a
always remain so when the total momentum in- standardized and simple nomographic representa-
volved in the interaction is constant. tion. 31 The performance specifications of several

60
dissimilarmaterials,from siliconeelastomersto surface or about 15-20 percent of its weight are
vitreousenamels,areadequately representedby often recommended. The operating principle of
thismeans. Figure5.21provides atypicalexample constrained-layer damping is that shearing action,
of thisconcept.
Thepointof intersection between as opposed to fiexural or extensional motion, is
frequency andtemperature islocated,X, and the considerably more efficient at damping vibra-
appropriate value of the reduced frequency, fay, tions. The constraining layer needs to be firmly at-
read on the lower abscissa scale at D. Knowing tached to the viscoelastic core and important
this value then allows the loss factor T/ and the parameters to consider are as follows32:
modulus of elasticity E to be found from the (a) Base structure: thickness, inherent
curves plotted for the particular material con- damping, frequency, density, and
cerned at B and ,4, respectively. Other design cri- modulus of elasticity.
teria for damping control are extant, 32,33 incorpo- (b) Viscoelastic material: thickness, densi-
rating configurational changes for viscoelastic ty, modulus of elasticity, and damp-
sandwich panels, and taking account of loss fac- ing loss factor.
tors and structural resonant frequencies. (c) Constraining layer: thickness, density,
and modulus of elasticity.
The improvement in performance for a free
Temperature (K)
layer of damping is demonstrated in figure 5.22,
7"2 11 TO r_l 1-2
104 10 3 which shows that approximately five times the ef-
o (10.0) fectiveness is gained by doubling the thickness of
tu
the viscoelastic layer.
10 2
Analytical work 34 has shown that low fre-
.. (1.o) quency performance can be enhanced in
&-- u
E
E viscoelastic sandwich panels by using a relatively
z 102 10 o-
hard core; this may provide up to 50 dB more
Io.1) =1
reduction below 200 Hz. The operating range of
=

o
10-210 -11 10 102 103 104 106 106
Reducedfrequency fa T(Hz)

"te2=-2=,o
</////;
2t /i i
10 2[ , . . _ =
C- 10-'_
"-_ St- /
, /
/. 10-2/ //
3x10-3/
J 2
/ /
6 6 ' I/:10 -a' / /
T_ 2 To T2

Figure 5.21 - Temperature-frequency nomogram

2t / / ' / V/130 ;4 ;0 /5
for damping materials (reference 30). (Used with
the permission of Bunhill Publications, Ltd. and --I 10-2 - / / 10 -5

D.I.G. Jones.)

5.3.1.2 Types of damping materials

lO 2 468 2 46_ _ _
Damping is effective only when intimate con- 0.1 1.0 10 100

tact with the vibrating surface is obtained. To this Relative thickness of damping layer (h 2 = H2/H 1)
end, secure bonding with an adhesive is essential.
The two most popular types of damping are free Figure 5.22 - Loss factor dependence on thick-
layer and constrained layer. In general, effective ness for a free-layer damper (reference 4). n, loss
free-layer materials are quite stiff and should be factor of viscoelastic material," n 2, total loss factor
applied in thick layers, while constrained-layer of panel; E l , E 2, elastic moduli of basic panel and
materials are relatively pliable and often used in viscoelastic layer, respectively. (From Leo L.
very thin layers. Free-layer materials normally can Beranek, Noise and Vibration Control. Copyright
be trowelled, sprayed, or glued on to a surface, © 1971 by McGraw-Hill, Inc. Used with the per-
and coatings at least as thick as the vibrating metal mission of McGraw-Hill Book Company.)

61
 dampingmaterials, in bothtemperature andfre- Viscoelastic
quencydomains,may be improvedby using lJ material
multilayertreatments or by a judiciousblendof
dampingcompounds. It hasrecentlybeenshown
Structure
thatsomeenamels haveusefuldamping properties
in thehightemperature range800° to 2000 ° F.35 Free layer
Multiple constrained layer
Tuneddampers operatewithina relatively narrow F

range,beingdesigned primarilytocopewithjusta
few vibrationmodeswell separated from any
others.Figure5.23 illustratesa few of these Sandwich panel
Single constrained layer
typicaltreatment configurations.Reference 36in- Mass of
cludesa detaileddiscussion of optimizationfor optimized weight
eachof these,includingtheeffectsof stiffness and
geometrical characteristics.
Table5.4 showsthe
typical loss factorsfor a rangeof structural
materials. Tuned damper

Muchresearch andimplementation remains Figure 5.23 - Typical damping treatment con-

to bedoneregardingdampingcontrolwithinthe figurations.
fieldof practicalnoiseandvibrationengineering.
Moreover, in mostextended metalstructures such Table 5.4 - Typical loss factors of various
asair-handling ducts,awidespectrum of resonant structural materials.
frequencies prevail,sothattheeffectof damping
maybefeltthroughouta largefrequency range.36 Material Loss factor,
Furthermore, in the mass-controlled region,in-
Aluminum 10 4
creasingthemasscertainlyimproves thetransmis- Brass, bronze 10 3
sionloss,but doesnot necessarily decrease the Brick 1 to2 x 10-2
vibrationamplitudein an equallyunambiguous Concrete
manner.Damping,on the other hand,affects Light 1.5 x 10-2
both without entailing the fairly extensive Porous 1.5 x 10-2
Dense 1 to5 x 10-2
modifications involvedwithstiffeningor massim- Copper 2 x 10 3
provements. Cork 0.13 to 0.17
Glass 0.6 to 2 x 10 3
Gypsum board 0.6 to 3 x 10 -2
5.3.2 Vibration Isolation
Lead 0.5 to 2 x 10 3
Magnesium 10 4
5.3.2.1 Physical principles of vibration isolation
Masonry blocks 5 to 7 x 10 3
Oak, fir 0.8 to l x 10 2
Vibration isolators operate as low impedance
Plaster 5 x 10-3
materials inserted between high impedance
Plexiglass, lucite 2 to 4 x 10 2
elements in order to interrupt transmission paths. Plywood I to 1.3 x 10-2
The mode of operation usually requires that a high Sand, dry 0.6 to 0.12
proportion of the incident energy be absorbed and Steel, iron I to 6 x 10 4
then returned before the end of each frequency cy- Tin 2 x 10-3
Wood fiberboard 1 to 3 x 10-2
cle. Thus, even if the isolator materials are highly
Zinc 3 x 10 4
elastic, the vibrations cannot be transferred fast
enough across it to produce effective transmis- Notes: (1) Normal room temperatures.
sion. (2) Typical audio frequency range.
Notwithstanding the fundamental difference From Leo L. Beranek, Noise and Vibration Control.
in the mode of operation and function between Copyright © 1971 by McGraw-Hill, Inc. Used with the
damping and vibration isolation, the two must be permission of McGraw-Hill Book Co.
incorporated together for effective isolation,
because large amplitude displacements occur when
the loaded isolator exhibits resonant behavior at dous rocking movements may damage the
its fundamental frequency. Apart from amplify- mounted machine. Damping inhibits the degree of
ing the amount of transmitted vibration, hazar- movement possible and concurrently reduces the

62
transmission of vibrationat thisfrequency. Con-
versely, toomuchdamping limitstheeffectiveper-
formance of theisolators,especially athigherfre-
quencies. Thisis illustratedin figure5.24,which where
showsa setof generalcurvesfor thetransmissi- k is the spring stiffness, N/m
bilityof a vibratingsystem. Transmissibility refers m is the mass loading, kg.
totheease withwhichphysical parameters suchas
force,velocity,or acceleration canbetransferred For many materials, however, stiffness
acrossa system. changes nonlinearly with dynamic loading and it is
Isolatorsoperateeffectivelyas suchonly therefore usually necessary to use the manufac-
above2fl sincebelowthis valuethe isolatorac- turer's data to obtain the correct deflections and
tuallyactsasanamplifierespecially at thereso- corresponding resonance frequencies. 37,38 The
nantfrequency, fl" The resonant frequency can be specifications on inherent damping provided in
calculated by assessing the static deflection of the the material should also be closely examined.
system involved when the isolator is under load, As a general rule, 10 percent (0.1) trans-
according to missibility is a useful design criterion to meet. It is
5 first necessary to determine the lowest forcing fre-
fl- quency likely to be encountered, which may be a
shaft rotational frequency, for example. The
amount of damping used rarely exceeds 10 per-
where/_ is the static deflection, cm. cent, and after selection of this and an appropriate
Static deflection usually changes linearly with system resonant frequency, normally set at one-
loading for steel springs because the stiffness con- third or less of the value of the major lowest forc-
stant k in the following relationship remains ing frequency, it is possible to calculate the deflec-
constant: tion which the chosen isolators must provide.

5.3.2.2 Types of vibration isolators

Solid, stiff pads of rubber may not be effec-

Damping ratio, _" tive as isolators because their displacement is small
10
and nonlinear relative to the load, making predic-
6 _-q -
tion of isolation performance difficult. Steel
4 - - -w
LI/]I_
o.Io" __
springs are more useful, especially for low fre-
quency isolation, while high frequency perfor-
2 -- -÷
¢L_ or_- -_ mance may be improved considerably with the use
-I .0
of pads of elastomeric material at either end of the

0.e
i

+
_t!- __ isolator. With regard to mounting, elastomers
e_ always perform better under shear loads, whereas
0.4-
.... L _ _l _ _ - cork, felt, plastics, and coils are best used in com-

0.2 _ , - -_1.o pression. Elastomeric isolators are fairly common

p- i \ in noise control; examples include natural,
0.1 -_ - silicone, and butyl rubbers, as well as neoprene
++ = . X ._6
0.OE- - _ 12I
,
:i _i
and combinations of all these. Their utility is ex-
O.2O tended because of their inherent damping and
0.04 +- "5
0.03-- relatively low specific acoustic impedance.
Reference 37 provides a good introduction to most
0.02 --
- " --''I - -';o.o6 isolators in common use and figure 5.25 depicts
\
o
0.01
0'1
1
0.2
I
0.4016
; I _ _ m
,j_-2 4 6 lO some typical units.
Pneumatic vibration-isolator systems offer
Forcing frequency f
several unique advantages relative to conventional
Undamped natural frequency fl
isolators, which are discussed at length in
Figure 5.24 - Transmissibility of a vibrating reference 39. They are particularly good for in-
system, fl is the undamped natural frequency of sulating at very low frequencies and do not incur
loaded system. the same lateral instabilities as the lengthy steel

63
 springs needed to perform the same task. The air
! pressure may be adjusted to support a higher load
or to provide a constant level and height regardless
of load changes. Typically they are rubber-bellows
type units reinforced with nylon, with an internal
pressure of around 7 x 10 5 N/m 2.
As a guide to the suitable selection of isolator
mounts, generic types are grouped in table 5.5
with their appropriate ranges of normal static
deflections.

Table 5.5 - Design ranges for types of

vibration isolators.

Type Static deflection (mm)

Steel springs 12.5 - 50

Rubber (in shear) < 12.6
Cork, felt, highly-damped
elastometers <6
Pneumatic isolators No deflection require-
ments since internal
pressure may be varied
from atmospheric at
105 N/m 2 up to
20 × 105N/m2

5.3.3 Application of Vibration Control Materials

As has already been intimated in section

5.3.1.1, damping can be particularly convenient as
part of a noise control solution, because it is often
easily applied as a free-layer treatment by trowel-
ing or spray techniques, and constrained-layer
treatments are especially valuable in situations
where spatial or structural restraints inhibit the
addition of extra mass or changes in stiffness.
Where impacting elements are involved, for exam-
ple, a gravity chute used to channel products along
a line, or drop-forge heads, or engine panels such
as sump and valve head covers, either type of
damping treatment may be applied. However,
when some kind of elastomer is placed between
the impacting surfaces involved, it is possible to
modify the vibration spectrum imparted as well as
to control the subsequent acoustic radiation.
Although the basic principles of isolator
choice have already been presented, the
methodology was based on the idealized situation
of an infinitely heavy and rigid base mounting. In
practice, this is seldom the case and unless the
Figure 5.25- Vibration isolation units (by cour- natural frequency of the isolator under load is
tesy of Vlier Engineering and Machinery Mount- about half that for the floor, then even low
ings, Inc.). amplitude, low frequency forces will propagate

64
throughfromthemachine andmayexcitethefun- theisolatorsequallyaboutthecenterof gravity.
damental frequency of theflooritself.Asa rough Thelatterapproach is oftenaccomplished using
guide,the staticdeflectionof theisolatorshould aninertiablockwithamasslargerthanthatofthe
beatleastthreetofourtimesthatofthefloor,and machine.Thisinvariablyresultsin a loweringof
approximately 90 percentisolationis produced thecomplete unitcenterof gravitywithrespect Io
whenthis is ten timesthe floor deflection.An theisolators,
whichimpartsextrastability.Figure
estimate of thelatterquantityis usuallyhardto 5.27showstwoexamples of inertiablockmount-
make,but 1/300thof the total floor spanis ing.Pneumatic isolatorsareveryflexiblein this
sometimes assumed. Figure5.26illustratesthe respect,in that they may be designedto ac-
correctiveeffectswith the proper choiceof complishmaximum isolationandtomaintaincon-
isolator. stantheightregardless of theloadconditions. 39
Another important considerationwhen Figure5.27bhighlightsanimportantfeature
choosing isolatorsis to ensure
equalstaticdeflec- whenisolatingunits, namely,the avoidance of
tion for eachmountunderthe machine;other- bridgingbya rigidmechanical connection between
wise,unduepitchingandexcitationof vibration the machineandfoundation.Oftensuchshort-
modeshigherthanthe simpleverticalor axial circuitsareoverlooked in theprocess of designing
modesmayoccur.Thisis achieved eitherby op- isolators;commonly, serviceconduits,ducts,and
timizingthe stiffnessfor eachseparatemount, hosesshouldbeflexiblyconnected to surrounding
knowingtheproportionate load,or by disposing structuralsurfaces.

1
Amplified I Isolated Amplified I Isolated
vibration I vibration v brat on vibration
forces I forces forces forces

Machine
mass Major machine Major machine
forcing frequency forcing frequency

Vibration
,
Machine
mount
resilience
curve of
mount
"_ 1.0
c

I---
I
I
I
\ \
I
I
I
I
I
Components Components I
transmitted transmitted to
Effective
to floor floor II
floor
mass _----t I

I
Effective _- Vibrationof 23 1.0 1.0 _
curve '_ I
floor _ floor "_:=
I I
resilience _-

///k /// I Frequency (Hz) Frequency (Hz)

Reference
Floor excited at Isolator attenuates components
foundation
resonance at resonance frequency of floor

(a) Mount selection Ib) Mount selection

considering machine considering machine
only and floor

Figure 5.26 - Comparison of isolator performance for flexible and stiff floor mounts. (From lan
Sharland, Woods Practical Guide to Noise Control, 1972. Used with the permission of Woods Acoustics.)

65
 Flexible ---7

connection//

_ J _ octco,,ar
Flexible

Ceoter
o, :::::::::::::::::::::::::::::
....
Fan __ Inertia
grav ty of _ i:::_iii:i
i:::i!ii::::_)_:i:i!l
inertia block U_,t/A'_::::' ............::-_:::j
motor _ __'_1___j,_,_i::iii
::!::i
:::...........
............. block
Vibration isolators

(a) (b)

Figure 5.27 - Examples of various mounting installations for isolators.

Finally, it is important to ensure that lateral bulk absorbers into engine duct linings has
stiffness is no greater than vertical stiffness, in resulted in research examining the net effect of us-
order to preclude serious sideways resonances. ing water repellants. One such study 41 has shown
One way of trading off these two stiffness com- that for a 1 percent decrease in flow resistance and
ponents is to orient the isolators so that their main a density increase of 3 percent after treatment
axis is in direct alignment with the center of with a repellant, some small changes in acoustic
gravity of the machine, which may be some properties took place over the range 0.5-3.5 kHz.
distance away. An excellent systematic approach The attenuation constant, reflecting the absorp-
to vibration isolation is presented in reference 40, tion coefficient c_, displayed no significant
which also contains formulas for calculating rock- diminution in value except at the higher frequen-
ing mode resonances. cies, where it dropped about 10 percent.
Machine enclosures often have absorption
placed on the inside which brings the material into
5.4 ENVIRONMENTAL AND LEGISLATIVE
proximity with running equipment such as pumps,
CONSIDERATIONS
generators, saw blades, chucks, and the like.
The class of materials most sensitive to en- There is therefore a potential problem due to
vironmental constraints are traditionally the wicking of the fibers, etc., as a result of over-
acoustic absorbers, especially those with a highly lubrication procedures, or to inadequately con-
porous nature and consequent fine structure. Ex- trolled air blow-off or cooling lubricant flows.
posure to humidity, water spray, oil, grease, dirt, The provision of good lubrication seals and
vibration, temperature extremes, and high fluid deflecting shields may be of help; otherwise the
flow rates may all have a deleterious effect on absorbent may be modified by using imperforate
acoustic performance and can influence the en- thin sheeting as before. Normally it is only the
vironmental suitability of the absorption material. high frequency end of the absorption spectrum
Protection against water and high humidity which is slightly affected by these techniques, but
conditions can be achieved by placing a thin im- to prevent further degradation in this, as well as to
pervious plastic polyester sheet loosely over the comply with sanitary and fire requirements, it is
face of the absorber or by incorporating a water- necessary to clean these surfaces regularly. This
repellant treatment to stop waterlogging of the requires a strong surface, capable of withstanding
pores with concomitant loss of performance and repeated rubbing and cleaning with warm
increase in weight. Porous ceramics, open-cell detergent solutions. 42,43
polyurethane foams, and glass wool fibers all For high temperature conditions, stainless
withstand humidity well. steel or copper "metal wools" may be used as ab-
The renewed interest of the aeronautics in- sorbers, and both are also of benefit where cor-
dustry in evaluating the utility of incorporating rosive fumes exist.

66
 In high fluid flow conditions,a retaining Finally, the manufacturer's recommenda-
screen of perforatedmetalprevents fibershedding tions regarding noise control material specifica-
fromfragilematerials, althoughit is importantto tions and installation must be considered in im-
securethe correctopen-to-total-area ratio for plementing practical engineering measures to
suitableacoustical performance, as discussed in alleviate noise. Other factors to be considered in-
section5.1.2.2.Evendense foamsmusthavetheir clude ease of maintenance after installation of
rawedges sealed
by thinplasticor a suitablepaint noise control materials, and the effect on produc-
surface. tion rates both during and following installation
Theextensive useof honeycomb perforates
in of the designed measures.
theextreme temperatureandflowenvironment of
jet engineshasresultedin severaloptimization
procedures for the acoustical performance of per-
forates, as discussed in the section dealing with REFERENCES
ducts in chapter 7 (see section 7.3.2). One broad-
band sound absorber based on the honeycomb 1. Kraft, R. E., R. E. Motsinger, W. H. Gau-
perforate principlO 2 can be supplied with a special den, and J. F. Link, Analysis, Design, and Test
aluminum corrosion-resistant treatment. In addi- of Acoustic Treatment in a Laboratory Inlet
tion to lending itself to ease of cleaning, it can be Duct, NASA CR 3161, 1979.
made self-draining, and possesses good thermal 2. Hersh, A. S., and B. Walker, Acoustic Behav-
conductivity, which is of value where heat build- ior of a Fibrous Bulk Material, AIAA Paper
up near machinery housings can occur. 79-0599, March 1979.
Barriers are usually constructed according to 3. Lambert, R. F., Acoustical Properties of Highly
structural and acoustical requirements, but Porous Fibrous Materials, NASA TM 80135,
sometimes, when dealing with localized noise 1979.
sources, it is necessary to retain some visibility of 4. Beranek, Leo L., ed., Noise and Vibration
the work site. This can be accomplished by the use Control, McGraw-Hill Book Co., 1971.
of shot-loaded transparent plastic barriers, for ex- 5. ASTM, Standard Method of Test for Sound
ample, to shield conveyor systems. Although their Absorption of Acoustical Materials in Rever-
performance may be diminished because of essen- beration Rooms, ASTM C423-66, 1966.
tial maintenance and access apertures, reductions 6. Purcell, W. E., Materials for Noise and Vi-
of up to 10 dB(A) are achievable for a well- bration Control, Sound and Vib., July 1976,
installed system. Flexible barriers made of loaded p. 6.
rubber or vinyl are useful where regular access 7. Dean, L. W., Coupling of Helmholtz Reso-
between different parts of a shop floor is required, nators to Improve Acoustic Liners for Turbo-
or for screening of small noise sources on a tem- fan Engines at Low Frequency, NASA CR-
porary basis. 134912, 1976.
Regulatory restrictions 44 have most impact 8. Budoff, M., and W. E. Zorumski, Flow Re-
within the food industry, where only glass, monel, sistance of Perforated Plates in Tangential
or stainless steel are permitted direct contact with Flow, NASA TM X-2361, 1971.
foods. Lead-bearing materials are forbidden, but 9. Rice, E. J., A Model for the Pressure Excita-
barium can be used as a suitable alternative. There tion Spectrum and Acoustic Impedance of
are also stringent cleaning requirements, which Sound Absorbers in the Presence of Grazing
usually require that surfaces be resistant not only Flow, AIAA Paper 73-995, October 1973.
to high pressure steam and a certain amount of 10. Marsh, A. H., 1. Elias, J. C. Hoehne, and R.
abrasion, but also to bacterial or fungal growth. L. Frasca, A Study of Turbofan-Engine
Firebreak requirements demand a detailed con- Compressor-Noise-Suppression Techniques,
sideration of ducting routes and the nonflam- NASA CR 1056, 1968.
mability of partitions and associated absorbing 11. Mangiarotty, R. A., Acoustic-Lining Concepts
surfaces. Fiber dissipation may be a problem, not and Materials for Engine Ducts, J. Acoust.
just from the point of view of acoustical degrada- Soc. Amer., vol. 48, no. 3, 1969, p. 783.
tion, but also because of the possible damaging ef- 12. Warnaka, G. E., N. D. Wonch, and J. M. Za-
fect of fibers on machine performance and their las, A New Broadband Sound Absorber, Inter-
threat as a health hazard. noise 76 Proceedings, 1976, p. 283.

67
13. Vincent,D. W., B. Phillips,andJ. P. Wan- 28. Wyle Laboratories, A Study of Techniques
hainen,Experimental Investigationof Acous- to Increase the Sound Insulation of Building
ticLinersto SuppressScreech in StorablePro- Elements (prepared for HUD), available from
pellantRocketMotors,NASATN D-4442, NTIS as PB-222 829, 1973.
1968. 29. Lang, M. A., and C. L. Dym, Optimal Acous-
14. Moreland,J. D., ControllingIndustrialNoise tic Design of Sandwich Panels, J. Acoust. Soc.
by RoomBoundaryAbsorption,NoiseCon. Amer., vol. 57, no. 6, 1975, p. 1481.
Eng.,vol.7,no.3,November-December 1976. 30. Jones, D. 1. G., Two Decades of Progress in
15.ASTM,StandardRecommended Practicefor Damping Technology, Air. Eng., Jan. 1979,
LaboratoryMeasurement of AirborneSound p. 9.
Transmission of Building Partitions, ASTM 31. Jones, D. 1. G., Temperature-Frequency
E90-70, 1970. Dependence of Dynamic Properties of Damp-
16. Barton, C. K., Experimental Investigation on ing Materials, J. Sound Vib., vol. 33, no. 4,
Sound Transmission Through Cavity-Backed 1974, p. 451.
Panels, NASA TM X-73939, 1977. 32. Nashif, A. D., and W. G. Halvorsen, Design
17. McDonald, W. B., R. Vaicaitis, and M. K. Evaluation of Layered Viscoelastic Damping
Myers, Noise Transmission through Plates into Treatments, Sound and Vib., July 1978, p.
an Enclosure, NASA TP 1173, 1978. 12.
18. Getline, G. L., Airframe Design for Reducing 33. Derby, T. F., and J. E. Ruzicka, Loss Factor
Cabin Noise, NASA Tech Brief, Summer 1978, and Resonant Frequency of Viscoelastic
p. 265. Shear-Damped Structural Composites, NASA
19. Sengupta, G., Low Frequency Cabin Noise CR-1269, 1969.
Reduction Based on the Intrinsic Structural 34. Vaicaitis, R., Noise Transmission by Visco-
Tuning Concept, NASA CR-145262, 1977. elastic Sandwich Panels, NASA TN D-8516,
20. Roskam, J., C. vanDam, D. Durenberger, and t977.
F. Grosveld, Some Sound Transmission Loss 35. Graves, G., et. al., On Tailoring a Family
Characteristics of Typical General Aviation of Enamels for High Temperatures Vibra-
Structural Materials, AIAA Paper 78-1480, tion Control, Proc. 15th Annual Meeting,
August 1978. Society of Engineering Science (SES),
21. Roskam, J., F. Grosveld, and J. vanAken, Gainesville, FL, Dec. 1978.
Summary of Noise Reduction Characteristics 36. Jones, D. 1. G., Damping Treatments for
of Typical General Aviation Materials, SAE Noise and Vibration Control, Sound and
Tech. Paper 790627, 1979. Vib., July 1972, p. 25.
22. Barton, C. K., and E. F. Daniels, Noise 37. Hochheiser, R. M., How to Select Vibration
Transmission through Flat Rectangular Pan- Isolators for OEM Machinery and Equip-
els into a Closed Cavity, NASA Tech. Paper ment, Sound and Vib., August 1974, p. 14.
1321, 1978. 38. V_r, I. L., Measurement of Dynamic Stiffness
23. Mulholland, K. A., and R. H. Lyon, Sound and Loss Factor of Plastic Mounts as a Func-
Insulation at Low Frequencies, J. Acoust. Soc. tion of Frequency and Static Load, Noise
Amer., vol. 54, no. 4, 1973, p. 867. Con. Eng., vol. 3, no. 3, November-
24. Lyon, R. H., C. W. Dietrich, E. E. Ungar, R. December 1974, p. 37.
W. Pyle, Jr., and R. E. Apfei, Low Frequency 39. Adair, R., The Design and Application of
Noise Reduction of Spacecraft Structures, Pneumatic Vibration-Isolators, Sound and
NASA CR 589, 1966. Vib., August 1974, p. 24.
25. Bailey, J. R., and F. D. Hart, Noise Reduc- 40. Rivin, R. E., Vibration Isolation of Industrial
tion Shape Factors in the Low Frequency Range, Machinery--Basic Considerations, Sound
NASA CR 1155, 1968. and Vib., November 1978, p. 14.
26. ASTM, Tentative Classification for Determi- 41. Smith, C. D., and T. L. Parrott, An Ex-
nation of Sound Transmission Class, ASTM perimental Study of the Effects of Water
413-70T, 1970. Repellant Treatment on the Acoustics Proper-
27. Harris, Cyril M., ed., Handbook of Noise ties of Kevlar, NASA TM 78654, 1978.
Control, Second ed., McGraw-Hill Book Co., 42. Waggoner, S. A., J. F. Shackelford, F. F.
1979. Robbins, Jr., and T. H. Burkhardt, Materials

68
 for NoiseReduction in FoodProcessing En- 44. IndustrialNoise Services,Inc., Industrial
vironments,App. Ac., vol. 11,1978,p. 1. Noise Control Manual (prepared for
43. Miller,R.K., AcousticMaterialsfortheFood NIOSH),availablefrom NTIS as DHEW
Process Industry,Noise-Con 73Proceedings, Pub.No. (NIOSH)75-183,1975.
1973,p. 519.

69
 CHAPTER 6

Noise Control Procedures

Before implementing engineering or ad- representative industries, dispels any misconcep-

ministrative measures to control noise, it is tion that there is a "typical" industrial noise en-
necessary to define the problem in regard to the vironment. Approximately 11 percent of the large
persons affected by the exposure, the type and sample in the latter study was considered to con-
location of noise sources (see section 3.4.1), and tain pure tones.
the appropriate criteria for assessing the severity Unfortunately, control at the noise source is
of the situation (see chapters 1 and 3 and the Ap- not always practical for machinery already in-
pendix). Although every noise control problem stalled, and in such cases retrofit techniques must
must be examined individually, there are three be applied at some stage along the path of propa-
separate components which should always be con- gation. These may involve partial or complete
sidered, namely the source, the propagation enclosure of the noise source, using a combination
path(s), and the receiver(s). Acoustical treatment of insulation and absorption techniques; absorp-
may be applied to any or all of these components tion at room boundaries to control the reverberant
with the general order of precedence being as field; or vibration isolation where there is signifi-
listed; i.e., the most satisfactory solution usually cant structure-borne transmission.
results from noise reduction at or near the source. Alleviation of the noise field at the receiver is
There may be several different units or normally considered as a last resort or as a tem-
machines operating as sources, or perhaps a single porary expedient during the lead time required to
one with several distinctly different components implement some suitable engineering solution at
contributing to the noise field. In either case the source or along the propagation path. Usually
source identification and characterization are the most critical receiver is the operator of a
necessary to the proper design of control machine or perhaps the occupant of the nearest
measures. For example, if a machine has several dwelling where there is a community noise prob-
modes of operation, it is important to attribute lem. The three alternatives are the provision of
sound pressure levels correctly to each distinct personal hearing protection, administrative con-
task. This may be accomplished, for example, by trol of the exposure duration, or complete
using high speed photography synchronized to the enclosure of personnel in a well-insulated booth.
sound level measuring equipmen0 or some of the The wide variety of techniques which may be
other techniques discussed in section 3.4.1. applied at either the source, path, or receiver
Control of the source noise emission levels should be evaluated in terms of cost, time needed
normally implies modifications to the structure, to complete satisfactory noise control, availability
operational changes, or redesign of the physical of new components or quieter machines, and the
configuration of the unit so as to relocate the effect on operations from the point of view of
noisiest components away from the most sensitive health and safety as well as productivity. The EPA
areas. Up to two percent of the energy input to a has published a substantial survey of industrial
machine may be converted to acoustic energy and noise 4 which includes source identification, in-
the particular characteristics of this may vary plant levels, community noise impact, and ap-
widely. An EPA report 2 lists noisy machines and plication of suitable abatement technology.
typical levels for a wide variety of industries, and The economic impact of noise control is not
reference 3, which is essentially an examination of necessarily amenable to straightforward accoun-
the spectral content of noise produced by several ting, since there may be long-term benefits of im-

71
provedproductivityin alessstressful
environment impulsive forces. Vibration may be caused by ec-
and better personnelrelations and morale. centric rotation, rapid changes in translational or
However,asaroughguide,onestudy5definedthe rotational velocity, or impact on other machine
impactincrementonproduction costs,A, as parts. The most fundamental solution is to reduce
the amplitude of vibration and hence the energy
Cnc available for acoustical radiation. Reducing the
A- vibrating area altenuates the total sound power
B.Y
radiated, and changing the dominant frequency of
vibration can alleviate the forcing conditions on
where
the structure or move the spectrum to a region that
Cnc is the cost of noise control is less sensitive with respect to the human ear.
Y is the life of machines, years
Amplitude reduction can be accomplished by
B is the annual productivity.
keeping all contact surfaces smooth and well
lubricated, which requires a thorough and regular
This study essentially examined retrofitting of maintenance schedule. Where possible, all
quieter components to a specific type of machine,
rotating parts should be dynamically balanced,
but a more general discussion of the economics of
and their mass reduced as much as possible. This
engineering design for noise control may be found
is particularly important for fans. Radiation from
in reference 6. Net costs may be reduced
machine panels may be controlled by vibration
somewhat by eliminating any hearing conserva-
isolation of the particular component. It is nor-
tion program if the control measures reduce ex-
mally possible to detect the crucial radiating com-
posure conditions to below the relevant standard.
ponents by making accelerometer measurements
This chapter discusses each part of the total
and considering the panel area involved. Vibration
noise problem in some detail, and a detailed
isolation may be achieved as described in section
discussion of some common noise sources is given
5.3.2. The acoustic power PWL radiated from a
in chapter 7.
panel surface may be calculated as 7

6.1 NOISE REDUCTION AT SOURCE PWL = tr < v2 > ocS W

Some typical steps which may be taken to

where
achieve source noise reduction include substituting
0 is the radiation efficiency of the
a different mode of operation such as welding for
panel, usually approximately 1
riveting; using entirely new, quieter equipment
< v2> is the space-averaged mean square
which may also have a direct production benefit;
and modifications to existing equipment such as velocity of panel surface, m/s
balancing rotating parts, improving gears, using Qc is the specific acoustic impedance
of air, N • sec/m 3
higher-powered motors with a consequent lower
S is the radiating area of panel, m 2.
speed, and relocating service components of a
large machine, such as generators and fans. The
major nonacoustic considerations which enter into Impact noise can be suitably reduced with the
the final choice are the cost of alternative noise addition of some form of damping in order to ap-
reduction schemes which achieve the same goal ply the force more gradually and continuously (see
and productivity constraints due to post-control section 5.3.1.1 and figure 5.20). This may also be
restricted access or machinery speed reductions. achieved by canting blades and edges on die cut-
The two general categories of noise source are ters, punch presses, and stampers. For blade cut-
vibratory and aerodynamic; table 6.1 shows some ting operations, increasing the number of blades
noise level ranges for a variety of these sources. reduces the impact load on each one, and shifts
the generated spectrum toward higher frequencies
6.1.1 Vibratory and Impact Sources where the noise field is often easier to control.
A good example of source noise control, fur-
Vibration usually results from the excitation nished from an extensive series of studies on
of resonances in machine components by cyclic or woodworking noise conducted at North Carolina

72
 Table6.1- Noiselevelsat theoperatorpositionfor a rangeof industrialmachinery
andprocesses
(reference4).
Noiselevels
(dB(A))

80 85 90 95 100 105 110 115 120

Pneumatic power tools

(grinders, chippers, etc.)

Molding machines
(I.S., blow molding, etc.)

Air blow-down devices

(painting, cleaning, etc.)

Blowers
(forced, induced, fan, etc.)

Air compressor
(reciprocating, centrifugal)

Metal forming
(punch, shearing, etc.)

Combustion
(furnaces, flare stacks)
6.1 m (20 ft)

Turbogenerators
(steam 1.8 m (6 ft))

Pumps
(water, hydraulic, etc.)

Industrial trucks
(LP gas)

Transformers

[
State University 8-1° and for which partial support _- Straight knife cutterhead
came from NASA, is the redesign of wood
planers. The most interesting innovation in this
case was the successful deployment of a helical l.
blade cutterhead,8 which introduces point impact
and reduces force fluctuations on the work piece
(see figure 6.1).
Damping
propriate
solutions
when a large number
are particularly
of resonances are
ap- t,o
_LLnC
5001000 2000 3000 4000 5000
excited. An example of damping control is fur- Frequency (Hz)
nished in figure 6.2, where a vibratory parts feeder
used to orientate work pieces was quieted. The Figure 6.1 - Comparison of sound spectra for
noise sources were chiefly the impact of the parts helical and straight knife cutterheads (reference
with each other and with the bowl itself. An 8). Board width, 30.5 cm; thickness, 2.54 cm; cut
acoustical hood was also placed over the bowl to depth, 0.32 cm. (Used with the permission of the
inhibit airborne radiation. Noise Control Foundation and J. S. Stewart.)

73
 Metal Sprayed urethane
lOO
9O

bowl /[-- boatilng inside d

03 80

_ 7o
I

_ 6o
×

_ 50

0
40 _

I 3C
63 125 250 500 lk 2k 4k 8k 16k

Frequency (Hz)

O No acoustical treatment [100 dB(A), 104 dB (lin)l

[] Damping on bowl; hood over bowl [92 dB(AL 94.5 dB (lin)}

Figure 6.2 - Acoustical treatment of vibratory parts feeder and attendant reduction in noise spectrum
(reference 11).

6.1.2 Aerodynamic Sources sidering it to be composed of cells with boun-

daries, appropriately defined as building walls,
Aerodynamic sources usually generate broad- machinery, pipe arrays, etc. 12 Statistical energy
band noise because of turbulent, vortical motions analysis, SEA, is being increasingly used to in-
in gases. This category includes the noise pro- vestigate sound propagation through extended,
duced by flames, jets, valves, fans, and relatively complex structures. The uses and possible applica-
high speed fluid flows. The noise level is usually tions (see section 6.2.4) of SEA are presented here
related to the turbulence intensity, which in turn is because the technique helps to solve complex
a function of the velocity of the medium con- acoustic propagation problems through aircraft
cerned. Thus it is often possible to make substan- and space vehicle fuselages.
tial reductions in the noise by increasing cross- The following sections describe specific
sectional areas and hence decreasing net flow mathematical relationships and design criteria to
velocities, and by eliminating right-angled bends be used when considering control of structure-
or constrictions in the ducting. There is a whole borne or air-borne sound between the source and
gamut of silencers which can be deployed to at- receiver positions.
tenuate air jets and exhausts venting to at-
mosphere; these are dealt with in section 7.1.2.
6.2.1 Acoustic Insulation Using Barriers and
Other subsections in section 7.1 deal with valve, Enclosures
combustion, and fluid flow problems, while sec-
tion 7.2.2 is concerned with fan noise reduction
The basic principle involved in insulating
techniques.
against acoustic energy propagation is to provide
impedance mismatching, as discussed in section
6.2 PROPAGATION PATH AND
5.2.1. The most common forms of construction
REVERBERATION FIELD CONTROL
used are close-in acoustic hoods, solid barriers,
and complete enclosures. These are most effective
There are few models, other than highly when used to limit the direct or free field of the
idealized mathematical models, that describe the source, and thus it is important to ensure their
behavior of sound propagating through complex placement within this area. In order to achieve
industrial environments. However, an attempt has this, the following series of relationships may be
been made to model a complex noise field by con- used to predict the total sound pressure level at a

74
 particular point resulting from contributions from Room to room transmission:
both the direct and reverberant field.
From section 3.4.4, the direct field intensity is SPL 2 = SPL l - R + 10logS T- 101ogA dB
governed by
where
p2 WQ SPL l, SPL 2 are the incident and received

Oc 47rr 2 sound pressure levels respec-

tively, dB
where A is the total absorption in receiv-
ing room (= $6_, metric Sabins)
0c is the characteristic acoustic impedance,
N. sec/m 3 S T is the total area of common par-
tition between rooms, m 2
Po is the sound pressure at angle 0, N/m 2
R is the composite transmission
Q is the directivity factor at angle 0
r is the distance from source, m loss of partition, dB.

W is the source acoustic power, W.

Notes:

The reverberant field intensity is governed by (1) A _- R c when d < 0.3.

(2) Reverberant field excitation only is
p2 4W assumed; the machine or source should
Oc Rc be located as far as possible from the
common room boundary to prevent di-
rect field excitation.
where (3) Flanking paths via the floor, wall, or
R c is the room constant given by Sod l- o_, ceiling structures may be present, thus
m2 significantly increasing the received
S is the area of room with average absorp- level, SPL 2.
tion coefficient c_, m 2.
Room to free space transmission:
Combining these,

SPL 2 = SPL 1 - R - 6 dB
P2 _W Q + (Close to wall)
pc
where

In logarithmic form, using the intensity level- SPL 1, SPL 2 are the incident level and level at
sound pressure level equivalence principle, external wall surface, respec-
tively;

SPL 2 = SPL 1 - R + 10logS

sPL0= PwL+ l0log Q+ 4] - 20 logr + DI
(At a distance from wall)

where
The leveling off in the decay rate from 6 dB
per doubling of distance when traversing from the r is the distance from wall, m
direct field into the reverberant field is governed S is the area of transmitting surface, m2
by the relative magnitude of r and Re, and the DI is the directivity index as defined in sec-
tion 3.4.2.
transition point r c, between the two fields is shown
in figure 3.11 (see also section 6.2.2). The above
Notes:
relationship applies only where a spherical radia-
tion pattern may be assumed as from a point (1) Free-field conditions external to the
source (see section 3.4.3). boundary are assumed.
For transmission across a solid boundary, the (2) Shape and proportions of wall deter-
following relationships apply: mine its directivity characteristics.

75
6.2.1.1 Barriers The major limitation on barrier attenuation
results from the finite size of the barrier. This may
Barriers generally offer little noise reduction
be overcome to some extent by making use of any
potential except in the acoustic shadow behind the
local features such as storage bins or control units,
barrier, where mid to high frequency sound can-
which effectively extend the dimensions of the
not easily diffract over the top or around the bar-
barrier, or by curving the ends to increase the path
rier edges. The placement of the barrier close to
length difference. An example of the prediction of
the source or receiver is important, as this in-
creases the path length difference defined in figure
Example: Find the noise reduction due to a
6.3, which is the most critical parameter in deter-
barrier of effective height 2 m, placed 2 m from a
mining its performance. Normally, a broadband
source when the receiver location is at 100 m from
sound spectrum is assumed in making barrier
this.
calculations, but if prominent pure tones are
present and the source or receiver are some A//"_" "_......... B Receiver
distance from the ground, up to three additional , IEI2 ..........
Source _'-_ _-]j • ..... :o
propagation routes may be created; these routes
////////////////////////////////////////////
can exhibit significant destructive or constructive
interference. ÷2 m_,* 98 m

Barriers are usually floor standing and should

A = %/22 + 22 = 2.8m
be wider than they are high. They need not be par-
ticularly massive, since 20 dB(A) is about the B = ,_,[2_ + 982 = 98.0 m
maximum net attenuation attainable in practice
D = 100m
and the attenuation for propagation directly
6 = 100.8 - 100 = 0.8 m
through the barrier itself need not be more than
10 dB above the overall attenuation desired.13 The
same criteria apply to acoustic hoods used to
Octave band center frequency(Hz)
shield only a relatively small area of a machine,
except that their smaller dimensions tend to make 63 125 250 500 1000 2000 4000 8000
them worthwhile only when dealing with high fre-
quency sounds such as air ejectors. Normally they PWt, ofmachine 88 97 110 108 102 96 90 85
source (Hz)
are used in conjunction with well-placed ab-
from SPL=PWL 20 logr+Dl-ll
sorbers, as indicated in figure 5.9, to prevent
flanking circuits resulting from reflected, and substituting r= 100 m; and for a
ultimately reverberant, sound build-up at the hemisphere, DI = + 3 dB
receiver.
SPl.i(_l,n PWI. 20log 100+3-11
According to the Fresnel theory of diffrac-
= PWL - 48 dB
tion, the performance of barriers may be predicted
as a function of the value of the Fresnel number, SPL at 100 m 40 49 62 60 54 48 42 37
N, calculated as shown in figure 6.3. without barrier
(db)

Wavelength, X 5.4 2.7 1.4 0.69 0.34 0.17 0.086 0.043

N = -0.2 (in)
x
from N- 26tX
Bright Transition N = 0
substituting, 6 - 0.8 m
zone /. zone f,._i" 'Path length difference
--' D
Fresnelnumber, 0.3 0.6 1.1 2.3 4.7 9.3 18.6 37.2
_ .fZ'i"', )..... ' Fresne number
',bnaoow zone . ' N
A -''l _:"_'::, :' ', \ ^,_ ±z6
." i V, ' L<';_ " ' ' i*- --
Source.-" I IZ " '_ _ ,_,,.Z'_, X Attenuation due 10 12 13 17 19 23 24 24
......... I _-Fx,_,_'(_C._
T_-'_ Rece ver to barrier
(from figure
6.4) (dB)
Figure 6.3 - Barrier Fresnel zones and calculation
of path length differences. Sign for N chosen plus 30 37 49 43 35 25 18 13
SPL at 100 m
or minus, depending whether receiver is in the with barrier
shadow or bright zone, respectively. (dB)

76
 ] ! 6.2.1.2 Total enclosures
30 I
I Complete enclosure is often the simplest and
most effective procedure for solving a noise con-
25: 1
trol problem when adequate source control is not
"7 i attenuation possible, in particular, enclosures can serve a
20 I
limit
x I multipurpose role by providing a safety barrier,
I controlling lubricant and heat emission from a
15--
I machine, and controlling the noise. It is usually
/
t-
O I
-I easy to meet acoustic requirements this way,
I0 --- --T---
especially if the enclosure is vibration isolated
from the machine and not too closely positioned
to it. This minimizes the possibility of strong
0,/ L I i l I J 1 i | 1

-0.3 -.01 .01 0.3 2 4 6 20 40 60 coupling effects caused by air resonances in the in-
-0.1 0 0.1 1 10 100 tervening space.
Fresnel number, N Typically, enclosure panels are specially
Bright = I_ _1 Shadow
I - I - I designed for the situation; they often consist of a
zone i zone
Transi- carefully balanced combination of absorbing,
tion damping, and insulating materials incorporated
zone into modular units which can be bolted to an
Figure 6.4 - Attenuation values for a barrier angle-iron frame. The provision of absorption is
against Fresnel number (reference 7). Strictly for important because it prevents high reverberant
point sources. (From Leo L. Beranek, Noise and sound levels within the enclosure walls from
Vibration Control. Copyright © 1971 by degrading the overall insulating properties. In
McGraw-Hill, Inc. Used with the permission of order to preclude flanking transmission caused by
McGraw-Hill Book Company.) direct vibration of the enclosure walls, the
machine source should be mounted on vibration
isolators. Other factors affecting the acoustic per-
a barrier's performance calculated for each octave
formance include the potential weakening effect
band is presented below. The attenuation values
are taken from figure 6.4, which displays barrier of apertures made to carry through essential ser-
attenuation as a function of Fresnel number, N. vices or for general access purposes. Where possi-
ble, service conduits should be grouped together
Other nomograms are available 13 which
and appropriate seals, such as lead-loaded vinyl,
enable the effect of 3, the subtended angle of the
used to minimize the chance of serious leaks.
barrier width at the source, to be taken into ac-
Several examples of sealing techniques are shown
count. In addition, the angle of the barrier apex
can be important, and double diffraction may in figures 6.5 and 6.6.
even occur for particularly thick barriers. 14 One Double-glazed windows may be necessary
where more than 20 dB(A) of attenuation is re-
particularly detailed study involving considerable
experimental work 15 considered the diffracted quired; the general principles governing their per-
sound intensity around the top and around each formance are given in sections 5.2.2 and 5.2.3.
edge and summed them to find the net barrier in- Doors should be as massive as possible, with tight-
sertion loss, lnL, according to: fitting positive pressure seals around all edges.
Adequate ventilation may be ensured by using

InL = 10log(l/D) specially integrated attenuating blower units or by

and having larger apertures with absorbent-lined baf-
fles and bends as shown in figure 6.7. The
cumulative effect of the heterogeneity introduced
with all these elements is taken into account by
D = Z I/(3 + 10Ni)
i computing the composite transmission loss, R, ac-
cording to the method in section 5.2.3.
where N i is the Fresnel number for the ith edge.
Enclosing a machine produces a reverberant
More complex expressions can be derived
field inside with an average sound pressure level,
which include the separate effects of absorption
on each side of the barrier, as well as the extent of SPLE, given by
the gap between the barrier and the ceiling. SPL E = PWL + 101og4/RcE dB

77
 "_ Peel back
segments

l | _X_ Cut run to

Segments cut (1) (2)
at point of bottom of sheet
pipe penetration

Acoustic

absorption
Fit snugly
Stainless

around pipe
steel clamp
/

3
_[_ead-vinyl
barrier
Acoustic -_

_J,A_ sheet

(4) _- Adhesive tape

for binding
(a) segments

conduits which interleave

to form a tight-fitting seal.

(1) (2)

Figure 6.5 - Use of lead-vinyl sheet to seal acoustic leaks. (a) Changes of seal installation for a single large-
bore pipe. (b) Installation of multi-seal for a series of small-bore pipes. (Derived from NASA Langley
Research Center architectural drawings.)

78
 /_ Acoustic lining

3 112 inch (8.9 cm)_

fiberglass batt

Internal _ External

I
I 4k

I
Fan unit I _.

Figure 6.6 - Pipe multi-seal using lead vinyl sheet

as insulation (by courtesy of NASA Langley I
)<
Research Center). /-
1/4 inch 10.6 cm)_ "__ 2 × 5/8 inch
(2 × 1.6 cm)
perforated
hardboard sheetrock insulating
where boards

PWL is the sound power level of machine

source, dB Figure 6.7- Example of an assisted ventilation
enclosure aperture. (Derived from NASA Langley
RCE is the room constant of enclosure in-
Research Center architectural drawings.)
terior (= SERE1 --WE) , m 2.

The level just outside the enclosure wall,

SPLEx T, is given by
where A is the total absorption within the room
(S_), assumed approximately equal to R C.
SPLEx T = SPL E - R E - 6 dB
Note: This relationship assumes a reverberant
forcing field inside the enclosure, omnidirectional
where R E is the average transmission loss of source radiation, and a common value of R for all
enclosure walls, dB. surfaces.
The net effect is to replace the machine source Figure 6.8 shows before and after photo-
with an enclosure of total sound power level graphs of an enclosure around an air handling
PWL E given by unit, highlighting the complexities of placing an
enclosure around such a unit with a large array of
PWL E = SPLEx T + 10log SE dB pipe and service connections. An example of cal-
culating the performance of an enclosure follows:
where S E is the radiating surface of enclosure, m 2.
From this the general reverberant level Example: A generator with known sound
SPLRE v in a room resulting from an enclosed power levels is enclosed within a structure whose
machine noise source is given by dimensions are 3 × 3 × 2.5 m. The average trans-
mission loss of the enclosure walls, R, the mean
absorption coefficient within the enclosure, _E,
SPLRE v = PWL E + 101og4/R C dB
and total room absorption, A, are given in octave
bands.
where R C is the room constant of the room, m 2.
These expressions may be summarized in the It is assumed that the generator inside the
enclosure does not affect _'E, either by obscuring
relationship below, which is analogous to that
or adding to the existing absorption, and that all
presented in section 6.2.1:
enclosure surfaces radiate omnidirectionally.
The results are presented as a comparison of
SPLRE V = SPL E - R E + 10 log S E A-weighted reverberant levels within the room,
- 10 log A dB with and without the enclosure.

79
 Totalinternalsurfacearea,S= 48 m 2
Total radiating surface area, SE= 39 m 2

The example highlights the importance of us-

inf,sufficient absorption inside the enclosure,
especially when the low frequency power level is
relatively high. In the calculation it may be noted
that the enclosure resulted in very little attenua-
tion' at 63 Hz.
(a)
Octa_c band center Irequcnc', (Hz)

63 125 250 51X) 10IX) 2(XX) 4000

I!nclosure mean 0.14 0.20 0.3() 0.63 0.91 0.91 0.88

absorption
coctlicicnt, 7_t: From RCE = S_E/I 6E
and subslituting S 48 m 2, and,el:

Enclosurerotvo 7.8 12.0 20.6 81 7 485.3 485.3 352.0

¢.'OllS1_lllt, R( I (11121

IOIog4 R(I 2,9 4.8 7.1 13.1 20.8 2().8 19.4

(dill

PV_'[ OI generator IO9 I(i9 11)9 11)3 I()11 92 84

(dB)

lrom _,l>I.t: = PV_'I + I0 log 4 R(I. (b)

_ t)1./, in_,ide 11)6.1 11)4.2 I(ii.9 89.9 79.2 71.2 64.6 Figure 6.8 - Example of enclosure fitted around
enclosure (dB)
an air handling unit. (a) Unit without enclosure.
A,,erage I rall <,lllis_,it)ll 8 13 18 23 28 33 35 (b) Similar unit after enclosure. (By courtesy of
1o_ ol enclosure NASA Langley Research Center.)
_all_, Rt-(Bill

total absorption ol 311 31) 3(I 60 6(t 61) 9(1

The predicted insertion loss values of the
room, .-I IUll)
enclosure walls, i.e., the difference between the
From %PIRI % SPI l RI: + I0 log .S/:
10 log A room overall sound pressure levels with and with-
and _,ubsliluting .SL 39 nl 2, and for SP[ 1_, R#., out the enclosure, must represent something close
a nd .,1
to the actual values obtained in a practical installa-
_Fq RI_ ' in room 99.2 92.3 85.0 65.11 49.3 36.3 26.0 tion. For instance, it has been shown that some
v, iih enclosure (dB)
foam absorbers can have a pronounced dilata-
l:rom 5;PI.Rt _, PWL + 11) log 4'A
tional resonance, 16 which can seriously detract
SPI RI\ ill loon1 I(X1.2 I(XI.2 I(X).2 91.2 88,2 80,2 70.5 from the performance of the overall partition.
v, il hotll enlclo',li rc
This particular problem can be solved by having
(dB)
the foam loosely attached to the insulating panel
A-_eighling 26 16 8.5 3 0 , I 1 in only a few places. For particularly critical in-
stallations, it may be of some advantage to use a
SPI 14t_ _ith 73.2 76.3 76.5 62 49,3 37.3 +25
eirclosurc Id Bt,,I)) computer program developed under NASA spon-
sorship Iv which predicts panel response to
%PI.RI _, _,_ithoui 74.2 842 92.7 88.2 88.2 81.2 69.5
reverberant acoustic fields. The required program
enclosure (d B(,4 ))
input parameters are the power spectral density of
Hence, o_erall sound pressure level in room _ill'_ enclosure = 80 dB(.4) the sound field, panel modes of concern and the
o_erall ,-ound pressure level ill roon'_ _ilhoul CllC]OMlrt' : 95 d|lJ(,,l)
associated damping values, as well as the defini-

8O
tion of the panelboundaryconditions.Thepro- methods. The convenience of avoiding a deter-
gramthenfurnishesthe fundamental resonance mination of each separate natural mode when us-
frequencies andassociated modeshapes; it maybe ing SEA is pointed out, as well as the need for less
easilyadaptedto handleothertypesof acoustic detailed modelling of acoustic modes when there is
fieldsaswell. appreciable absorption present.
Whenenclosures arebuilt verycloseto the 6.2.1.3 Partial enclosures
sourceof noise,suchthattheseparation distance
is lessthanhalfthewavelength of thelowestfre- Partial enclosures may simply be treated as
quencyof interest,thentheeffectof theinterven- enclosures with holes or missing panels. The in-
ingair spacebecomes important.Resonances in terior absorption coefficient u and transmission
thecavityspace mayexhibita strongcouplingef- coefficient r should both be set equal to unity
fect with panel resonances; furthermore,the where there are open sides. Because the assump-
assumptionof a reverberantfield insidethe tion of a reverberant field may often still be made,
enclosure is questionable whenthe machineoc- normal enclosure prediction formulas as given in
cupiesa largefractionof theenclosure volume. section 6.2.1.2 remain valid with the above coeffi-
Thisleadsto substantial deviations from simple cient substitutions. Partial enclosure perfor-
masslaw predictions,whichare not easilyac- mance, however, is obviously limited and can be
countedfor.18,19 difficult to predict because of serious directivity
Recentlythe problemof predictingthe effects at the open sides. If major noise sources
acousticenvironmentwithin the spaceshuttle are kept well within the enclosure and their spec-
payloadbayhasresultedin thedevelopment of an trum is mainly high frequency, then a reduction of
analyticalmodel2° to predict space-averaged levels as much as 10 dB in level may be gained outside
within the empty or partly filled bay. This incor- the open end. This is more readily achieved by the
use of internal absorption, especially with
porates the use of statistical energy analysis, SEA,
to be used when the excitation field is diffuse and judicious positioning of this near the open areas.
there are large numbers of acoustic and structural Figure 6.9 presents some partial barrier configura-
modes present. 21 At low frequencies, account tions and their anticipated performance.
must be taken of each individual mode, whereas Partial enclosures can be particularly useful
where there are severe access constraints to a
SEA utilizes the concept of an average modal den-
sity in frequencies (see section 6.2.4). The model machine because of supply/product feed or ven-
computes net power flow from the exterior to the tilation requirements. Partial enclosures may also
interior and between subvolumes surrounding the be helpful in reducing localized noise sources
payload. The potential value of this model with which are integral parts of a larger machine sur-
face.
respect to other applications is in accounting for
variations in sound pressure level produced when
6.2.2 Use of Acoustic Absorption in
large-scale structures are introduced into a cavity:
Reverberation Control
these are particularly evident at low frequencies.
Some of these theoretical results have been
validated using physical scale models. 2°,22 The Absorption has limited value within the direct
field of a source unless used in conjunction with
most difficult task in this approach is to define
insulating materials in enclosures. Figure 5.9
adequately a series of subvolumes, each of which
shows the appropriate placement of material for
must be considered homogeneous in terms of the
different propagation paths. As a general rule, at
acoustic field. A finite element approach can also
be used. 23 less than 3 m from machine surfaces, no signifi-
A comprehensive review 24 of all the recent ac- cant reduction may be achieved with absorption
alone. The transition from a direct to a
complishments in the prediction of interior noise
reverberant field occurs at an approximate
for transportation vehicles and cavities of various
sorts helps to summarize the work of NASA in distance r c (see figure 3.11), which may be com-
this field. It is considered therein that the increas- puted according to

ing use of modal representations of walls and rc = 0.2 x/S,x m

cavities provides a useful conceptual and com-
putational framework with which to improve ex- It is sometimes possible to detect this
isting structural noise transmission prediction changeover aurally by cupping the hands behind

81
 -13 dB with or without

_with no absorptive

- 9 dB with absorptive lining

j.,._ lininglining + 1 dB with no absorptive -12 dB with lining
- 6 dB with no lining lining
-3 dB with lining
-1 dB with absorptive lining
+ 3 dB with no lining (b)
(a)
- 5 dB with no lining on
screen or enclosure
- 8 dB with lining on
enclosure but not
on screen
-10 dB lining on screen
and enclosure

-4 dB with no lining on
screen and enclosure
- 6 dB with lining on
enclosure but not
-4 dB with no lining on on screen
screen or enclosure -10 dB lining on screen
-10 dB with lining on enclosure and enclosure
but not on screen
-10 dB with lining on screen
and enclosure
(c)

Figure 6.9 - Partial enclosure approximate attenuation performances. Values specify average change in
SPL in the frequency range 500-4000 Hz. (a) Enclosure with one side open. (b) Tunnel. (c) Open-sided
enclosure and moveable screen. (From lan Sharland, Woods Practical Guide to Noise Control, 1972. Used
with the permission of Woods Acoustics.)

the ears and finding the position for which any formance of insulating materials. It is commonly
detectable source directionality is seriously accepted that the improvement from small reduc-
diminished. Normally the noise exposure of a tions in the reverberant level are subjectively
machine's operator is dominated by the direct rather greater than would be anticipated merely
sound field. However, in some instances, noisy from consideration of the changes in the overall
machines may produce reverberant sound levels sound pressure level. It should be cautioned,
which exceed the direct field levels received by however, that only rarely are sound fields truly
operators of quieter machines. Some examples of diffuse; consequently there is often considerable
this situation and the effects of possible uncertainty as to the exact reduction which is
treatments illustrating the balance between achievable at a particular point by means of extra
reverberant and direct fields are shown in figure absorption. This is especially so where the room
6.10. When sources or machines are scattered over shape deviates from regular geometry as, for ex-
the whole floor area, it is unusual for very few ample, in very long, low-ceilinged buildings.
positions to be outside the dominant influence of As noted in section 6.2.1, the reverberation
direct fields. level is related to room characteristics by
The usual objectives of reverberant field con-
trol are to reduce the noise exposure of personnel SPL = PWL + 101og4/R c dB
below the statutory or desired limits, to improve
speech communications and the detection of aural where R c is the room constant (= S6dl -_), m 2.
warning signals, and to reduce net incident sound The quantity S_, which is a measure of the
pressure level, thereby alleviating the required per- total absorption in the room, is found by assessing

82
 Action
Before Required After

+10 -- 3dB
+1o-I
Reduction in total
SPL of 3 dB
0
0 ' --4--m-- _.- _--

OdB Reduction in
-10 --- reverberant SPL
by 20 dB
-10 t

-20 -- -20

+10 -

Z
+10 1
_7
o 0 | | "

°t
X
SPL of 10 dB
0 dB Reduction in I Reduction in total
0)
direct SPL -10 ....
rt -10
t.,O by 20 dB

- 20 -- -20
tr

+10 -- +10 -

0
A
0---- 7 Reduction in
reverberant SPL I Reduction
SPL of 7 dB
in total
0dB
of 10 dB
T
-10 -10

-20 -- -20 -- _ Total SPL

Direct-field SPL

.... Reverberant-field SPL

Figure 6.10 - Contributions of direct and reverberant fields to the total sound pressure level.

the reverberation time as defined by Sabine (see where

below). This is the time taken for a steady-state V is the room volume, m 3
reverberant sound field to decay to 60 dB below its S_is the total absorption for i areas with
initial level after removing the noise source input. separate absorption characteristics
This is usually measured using a starting pistol or (= ESioti) , metric sabins, m 2.
a high level broadband source, and the decay rate
is measured in each octave band. The following Note: A separate term dependent on room volume
equation may be used except where the room is ex- and frequency is added to the denominator to ac-
tremely large, in which case air absorption of high count for air absorption when necessary.
frequencies becomes important: There may be several difficulties even in mak-
ing this simple estimate, because of uncertainties
0.16V
Reverberation time, TR - s in estimating the total volume V and the internal
Sa surface area S of many industrial spaces. 25 Also,

83
high ambient noise levels tend to make suitable where
measurements of the decay curve difficult and the Rci is the room constant before absorption
inherent assumption of a diffuse field is often in- treatment
valid. Figure 6.11a shows a typical change in Rc2 is the room constant after absorption
reverberation time for varying amounts of absorp- treatment.
tion added to the main doors, ceilings, and upper
half of the walls of a large machine shop. In this A ready account of the attenuation may be
case, the precaution was taken of averaging the made with reference to figure 6.12, from which it
results at three separate locations to account for may be seen that there is a decreasing benefit in in-
spatial variations. The accompanying inset table creasing the absorption without limit. As a rough
shows the actual performance against the design guide, a factor of 5 change in the average absorp-
goal. A photograph of the absorption treatment, tion coefficient _ produces a change of 10 dB or so
which was spray-installed using a cellulose base of in the sound pressure level.
average depth 3.0 in. (7.6 cm), is shown in figure It is often possible to add appreciable absorp-
6.1lb. tion only in the ceiling area. This may conven-
The reduction in sound level ASPL which can iently be achieved by the use of functional ab-
be expected from the addition of absorption sorbers (see section 5.1.3 and figures 5.9 and
material is related to the room constant by 5.10). Other useful techniques include standard
modular absorbing panels, using a spray-on
ASPL = 10 log Rc2/Rcl dB fibrous absorbent which can often be applied

C)_ Before treatment

O--- - Door only treated

(_..-.. m Door and walls treated

Predicted Achieved
"7
*.a
e"
._o

RT 500 Hz
e_
2.0 1.9
Is)
ew

Total RT
reduction 75 49
(%)
0 I 1 I 1 I t I
Pink 63 125 250 500 1k 2k 4k

Noise Frequency (Hz)

lal

{b)

Figure 6.11 - Reverberation field control in a

large industrial space. (a) Reverberation time
changes. (b) Photographs of installed absorption.
(By courtesy of NASA Langley Research Center.)

84
easilyandeconomically,andresonatorunitsfor 6.2.3 Flanking Path Analysis and Identification
handlingdiscretefrequency
problems(seefigure
5.6). Energy not transmitted from the source to the
receiver along the most direct path is termed
flanking or indirect energy. This may occur as a
result of mechanical excitation of the structure
0 and subsequent re-radiation at points distant from
the original source; break-out of energy from
v --5
ducts or ceiling voids into neighboring spaces; and
airborne sound transmission through weak points
z -10
in existing insulating barriers. Several of these
0
_ -15 routes are illustrated in figure 6.13.
×

IN It is not easy to determine the extent of flank-

-20 ing transmission in many situations. As a general
.J
_ -25 rule, flanking becomes significant where insula-
tion above 35 dB is desired, and becomes par-
_ - 30
ticularly critical where 50 dB of insulation and
rr
-35 over is required. 26 To achieve insulation above
0.12 0.30 0.6 1.21.8 3 6 12 18 30
this, it is usually necessary to build a completely
Distance from acoustic source center (m) isolated inner room. There are several techniques
Figure 6.12 - Attenuation in reverberation field which can be used to estimate the extent of the
with change in room constant and the transition problem. A substantial difference between the
effect from a direct field regime. predicted and observed values for the transmission

f ceiling
Structural

: .... . .: . :.....

Re-entry of acoustic
energy into structure

Figure 6.13 - Flanking and indirect transmission paths.

85
 lossR, of the order of 5 dB or more, often in- SPL N = PWL - 10logS s + 10log
dicates the amount of sound propagation via
unaccounted paths. In this case the slope of the [(Co + 4/R 0) (C 1 + 4/R l) ... (C_, + 4/Rx) ]
octave band plot for the partition insulation is
often less than predicted. Directional microphones + 10 tog [(r0)(rl)(r2) ... (rN_l) ] dB
are sometimes useful for locating areas radiating
into a room when the acoustic space concerned is where
relatively small and anechoic. Even more effective PWL is the source power level, dB
is the direct assessment of sound power radiated s, is the source area, m 2
from neighboring walls relative to the insulating R N is the room constant of the Nth cavity,
partition of interest, using accelerometers. The ap- m2
propriate equation for estimating this is is the correction factor of the Nth
cavity, dependent on source shape and
distance and orientation of source with
W = cr<_2>ocS W
respect to receiver.
_N-I is the transmission coefficient associ-
where ated with the partition between the
is the radiation
(3" efficiency of the N- l th and Nth cavities.
panel, usually approximately 1
<_2> is the space averaged mean square Comparing the relative magnitude of CN with
velocity of panel surface, m/s 4/R N indicates whether the direct or reverberant
oc is the specific acoustic impedance field dominates. As the cavity walls become less
of air, N. s/m effective as barriers, then the product of r's tends
S is the radiating area of panel, m 2. to unity, and consequently zero net barrier attenu-
ation results, since log (1) = 0.
Note: This equation is strictly applicable only For the case of structural radiation by N sur-
above the panel critical frequency fc' faces into a single cavity, the resultant sound pres-
Other techniques employed for this purpose sure level is given by
include correlation analysis (see section 3.4.1),
and the use of a vibration isolated noise source SPL = PWL l - 10 log S 1 + 10 log
(e.g., a loudspeaker) in place of a nonisolated
source. The latter procedure enables a comparison r2(C 2 + 4/R)
+ 4/R) +
to be made between the attenuation obtained with rl(Cl fl2/_2

the substitute source and that obtained with the

actual source. Any disparity can be taken as an rN(C N + 4/R)-
dB
estimate of structure-borne transmission. flN/EN

One particular approach used to account

comprehensively for flanking paths stemmed from where
the development of a series of equations for cop- PWL 1 is the sound power level at wall 1
ing with multiple path propagation in a helicopter S 1 is the area of wall 1
internal noise study. 27 This may be suitable for /3N is the area ratio SN/S j
well-defined noise control problems. The general eN is the power ratio WN/W I.
procedure is applicable to other internal noise
prediction cases where structural geometries, noise Note: eN/flN is the ratio of the intensity at
source strengths, and material acoustic properties wall N to the intensity at wall 1.
are defined. Because of the importance of The study 27 elucidates an integrated step-by-
minimizing the weight penalty associated with step procedure utilizing the above equations to
acoustic treatment in this particular situation, the obtain the total sound pressure level at a particular
trade-off between insulation and absorption was space after accounting for structural and geomet-
emphasized. For the case of sound propagation rical factors, major source strengths and paths,
through N-1 cavities sharing common boun- transmission loss and absorption data, correction
daries with cavity N, the sound pressure level in factors from radiating surface sizes, and distances
the latter is given by to the receiver position.

86
 Someof theprocedures to reduceweakened 6.2.4 Statistical Energ} Analysis (SEA)
transmissionpathshavealreadybeendealtwithin In the early 1960s statistical energy analysis
thesectionon enclosures (section6.2.1.2).Good evolved mainly as a result of NASA and USAF
caulkingat paneledgeandapertures with bitu- funding of studies regarding complex noise and
mastic-type compounds,positivepressureseals vibration problems in high speed aircraft and
arounddoors,doublepanelsresilientlyisolated, spacecraft. The three main areas of application
resilientmountingof ductsandserviceconnec- have been structural vibration excited by airborne
tions, andvibrationisolationof machinesand noise, generation of a sound field within an elastic
enclosurescanallhelpachieve therequisite insula- container caused by external excitation by noise,
tionperformance withoutdegradation fromunac- and the flux of vibratory energy through intercon-
countedpropagation routes. nected elastic structures. 29 Many of the original
Forthehelicopter problema widevarietyof principles were developed from room acoustics
remedialprocedures wereusedto reducethe modal analysis.
interiorlevelby some29riB(A).28Theseincluded Reference 30 summarizes the early work on
applyingdampingtape to structuralmembers, acoustic excitation of structures and provides a
mountingtheinteriortrim onantivibrationisola- useful introduction to the techniques invoh, ed.
tors to precludestrongradiationfromthe sides Basically, SEA relies on estimating the average
and ceiling,acoustically sealingthe doorsand number of resonances in a moderately broad band
ducts,installinga raisedplywoodfloor2cmthick, of frequencies and the spatial distribution of
andfilling thevolumebetween trim andfuselage response amplitude for a "typical" mode of reso-
wallswith16-kg/m 3densityfiberglass or twolead nance. By computing the average response of this
layersseparated with absorbentfoam of total typical mode to sound waves from many angles, it
thickness5 cmandtotalsurface density7.3kg/m2. is possible to estimate the average space-time
Thesearesummarized in figure6.14,whichshows response in that frequency band, after multiplica-
the acoustictreatmentconfigurations aswell as tion with the appropriate modal density. The
majornoisesources. benefit of such an approach is that, even if all the
individual modes could be exactly characterized,
the computation of the separate effects of each
mode would be impossibly cumbersome.
IVl ...... Main gear box 57 Normally at least two of three distinct phases
aln cr ve .¢n_ Tail rotor _ are involved in conducting SEA analysis,
Engine___'X /drive_/ namely, 31 the development of a power-flow model
O' -- -

C___2_#_ _ I _ /Intermediate with component areas of homogeneous energy dis-

_. ...... i _ _ gear box tribution; evaluation of parameters such as modal
densities, structural toss factors, and coupling loss
Acoustically _ ,I,
treated 1 factors; and the calculation of equilibrium energy

passenger (a) distribution and the corresponding statistical

cabin values of variables such as plate velocity and
Interior trim panels acoustic pressure, which are of direct concern to
Double wall (lead -. ',_-_mounted on isolators
the engineer. It is important to recognize that SEA
andfoam) _ : . _ _ Bagged fiberglass
= \ methods are inherently inappropriate at low fre-
/ , Damping tape on quencies, where statistical averaging cannot be
i ', skin panels applied because modal density is insufficient and
| ,_ 122cm
198cmi r_-_f--_ ,, ucm where the coherent addition of modes (see section
: I I/ f Single wall (lead 3.2) and mode-to-mode coupling can become
! : I H and foaml important.
Carpet-covered _ '
A typical representation of an SEA model
p, ood.oo, (r
/
t J applied to sound transmission across a single
I
i panel is shown in figure 6.15. The loss factor T/i
(b)
and the coupling loss factor 0d, both dimension-
less quantities, are defined as 32, 33
Figure 6.14 - Illustrative sketches of integrated
acoustic treatment in a helicopter noise problem (ref- lli,diss

erence 28). (a) Side view. (b) Internal cross section. r/i = wEstored

87
where Although SEA has been used mainly as a
Hi,dissis thepowerdissipated
in theith ele- research tool, these techniques are used increas-
ment, W ingly in building acoustics, marine engineering
o: is the angular frequency (= 27rf), Hz and shipbuilding, and nuclear reactor design, as
Estored is the stored energy, J; well as in industrial noise problems. An example
of the predictive accuracy with SEA appears in
figure 6.16, which is taken from a study on
acoustic excitation of a spacecraft shroud and the
subsequent effect on the interior space. 34 The
where calculated noise reduction is consistently lower at
Flij is the power flow between systems i low frequencies, probably as a result of the
and j, W reduction in internal acoustic modal density.
El, Ej is the total energy of systems i and j, Fortunately, some relevant parameters, such
respectively, J. as acoustical radiation impedance, which provides
a measure of the coupling between sound fields,
The total modal count in the ith system, Ni, is and modal densities, can be calculated using basic
related to the modal density n i by acoustical theory. 35

N i = niA f
4O
,-.q
where Af is the frequency bandwidth of interest. d_
J
co _ 30

A series of simultaneous equations is then Q

%
written and subsequently solved: z 20

o o o_
_ x I Theoretical
]-Ii,in = I-ll,diss + l-Ii2 + 1-[13
prediction

.__
O
l]2,in = 1-I2,diss - 1-I12 + H23
z
-10
1-13,in = I-I3,diss - 1-I13 - I-I23 63 125 250 500 1000 2000 4000 8000

One-third octave frequency (Hz)

I theoretical prediction
or alternatively, 0 experimental values

Figure 6.16 - Comparison of experimental and

theoretical (SEA) noise reduction for a noise
I_l'in = C°_lEl + c°fll2 nl nil + transmission problem (reference 34).

e,c.
6.3 ADMINISTRATIVE AND HEARING
PROTECTION PROGRAMS FOR
RECEIVERS
1] 2, in 113, in I
ll , in
After suitable assessment of noise exposure
parameters, the risk of hearing hazard should be
Resonant
evaluated. In critical work positions, it is usually
path
best to use noise dosimeters to check on the total
i]12
exposure received by each employee. To comply
1113 t Nonresonant path
1-I1, diss 112, diss 133, diss with OSHA regulations, the allowed exposure
dose for each 8-hour period may not exceed unity,
Figure 6.15 - Single-panel sea model. which is equivalent to a 90 dB(A) continuous level

88
(seetheAppendix).If possible, theexposure dose termediate varieties as well. The former are often
shouldnotexceed 0.5,to obviatetheneedforany prefabricated in several sizes, although some may
formof hearingprotection.As shownin figure be custom-shaped in situ for the wearer. Ear plugs
A. 1in theAppendix,levelsover85dB(A)carrya made of glass down or waxed cotton wool,
seriousrisk of eventualpermanent hearingdam- designed to be used only once, are also available.
age. It is importantthereforeto developand The attenuation afforded by these types varies
implement sourceandpathengineering solutions widely, depending on the manufacturer, wearer,
to noisecontrol.Controlat thereceiver position and quality of fit. Consequently, both the average
shouldnot beconsidered unlesssourceandpath attenuation and the standard deviation over a
engineering solutionsrequirea substantiallead range of people are usua!ly quoted to provide a
time for completeinstallationor unlesstheycan better estimate of their potential value. Earmuffs
provideonlya limitedattenuation. Thefollowing consist of circumaural cups, usually built of
sectionsset forth the long-termlimitationsof plastic shells with fluid- or foam-filled seals, fit-
administrativeor personalhearingprotection ting completely around each ear and connected by
programs, fromthestandpoint of economicsand a properly tensioned headband. Photographs of
of satisfactoryacousticattenuation. the various types of hearing protectors are shown
in figure 6.17.
6.3.1 Personal Hearing Protection The performance of ear plugs is more variable
than that of muffs because plugs require an indi-
There are two common types of protectors, vidual optimization for each wearer moreover,
namely, ear plugs and ear muffs, with several in- workers often insert plugs incompletely because

Figure 6.17- Examples of common types of hearing protectors. (By courtesy of Flents Products Co.,
Bilsom International, Inc., Adco Hearing Conservation, Inc., and E.A.R. Corp.)

89
theymaybeuncomfortable. Theperformance of A hearing conservation strategy should also
earmuffsdepends oncorrecttensionin thehead- include a continual education program with such
band,propermaintenance of thesealsaroundthe features as labelling of hazardous areas with warn-
cups,andtheabsence of longhairor spectacles, ing signs, personal instruction for fitting hearing
whichwill detractfromtheircorrectfitting.Two protectors, informative films and pamphlets, and
standardprocedures for testingthe performance personal example by management and union offi-
of eithertype are now available, 36,37and the cials. All new employees should be given instruc-
Environmental Protection Agencyhasproposed a tion on hearing conservation as part of their
singlenumberrating,basedin parton thetesting formal induction schedule.
proceduresetforth in reference 37, calledthe Administrative noise control procedures are
noisereductionrating(NRR).38TheNRRis sub- based on the principle that exposure is a function
tractedfromadB(C)estimate of thenoiseclimate both of time and level. According to the OSHA
togivethenetlevelreceived attheearin dB(A).It standard (see the Appendix), the trade-off be-
is intended to introducethisprocedure for rating tween these is a 5 dB increase in permissible level
andlabellingall futurehearingprotectors.The for each halving of the exposure time. For exam-
NationalInstitutefor Occupational Safetyand ple, employees who steadily experience 90 dB(A)
Health(NIOSH)haspublisheda comprehensive should never be exposed to higher levels, while
list of attenuationdataandstandarddeviations those who do have exposure above 90 dB(A)
for proprietaryhearingprotectormodels. 39A should be removed from the noise source after
recentreport4°hasalsoestablished thatthenoise expiration of the relevant OSHA time limit and
attenuation providedby earplugsin actualuseis should spend the balance of the day in areas at
muchlessthanthatpredicted by laboratorytest- under 90 dB(A). This approach obviously requires
ing.Thisappears to betheresultof improperfit- careful allocation and scheduling of work duties;
ting in the workplace,underscoring the impor- in practice it has met with little success. 41 It works
tanceof carefulselection andskilledsupervision best where it is feasible to split shifts, so that an
when ear plugs are initially introducedto individual's work time is divided between
employees. operations at various noise levels, or by using a
particularly noisy machine on several different
days rather than on one occasional full day.
6.3.2 Monitoring Audiometry and Administrative
Control Procedures

In conjunction with the use of hearing protec- REFERENCES

tors, a monitoring audiometry program should be
implemented, with semiannual hearing tests for 1. Eckert, W. L., E. T. Booth, P. D. Emerson
employees who risk hearing damage from a high and J. R. Bailey, Fly-Shuttle Loom Noise
level of noise, i.e. above 90 dB(A). This expensive Source Identification, ASME 76-DE-39, April
testing, which may disrupt production, is needed 1976.
because noise limits and attenuation data for pro- 2. Bergmann, E. P., Severe Sources of Industrial
tectors are both statistical quantities and can there- Machinery Noise (EPA Contract 68-01-2234,
fore only provide protection for a percentage of liT Research Institute), Task Report J 6331,
an exposed group. In addition, since protectors November 1974.
are often removed, and since noise levels may fluc- 3. Royster, L. H., and J. R. Stephenson, Char-
tuate more than anticipated, a monitoring pro- acteristics of Several Industrial Noise Envir-
gram provides the only check on the efficacy of onments, J. Sound Vib., vol. 47, no. 3, 1976,
the overall noise control program. A substantial p. 313.
survey of companies conducted by NIOSH 41 indi- 4. L. S. Goodfriend Associates, Noise from
cates that around 25 percent of those who Industrial Plants (prepared for EPA), availa-
responded had hearing_ conservation programs; ble from NTIS as NTID300.2, 1971.
however, the same study also showed severe defi- 5. Emerson, P. D., J. R. Bailey, and W. D.
ciencies in audiometric testing equipment and in Cooper, Economic Impact of a 90 dB(A)
the training and qualifications of personnel who Noise Standard on Textile Spinning Opera-
operate them. tions, ASME 77-RC-15, May 1977.

90
 6. Miller,T. D., Industrial Noise Control: Put- Con. Eng., voi. 10, no. 2, March-April 1978,
ting it all Together, Noise Con. Eng., vol. 9, p. 74.
no. 1, July-August 1977, p. 24. 20. Wilby, J. F., and L. D. Pope, Prediction of
7. Beranek, Leo L., ed., Noise and Vibration the Acoustic Environment in the Space
Control, McGraw-Hill Book Co., 1971. Shuttle Payload Bay, AIAA Paper 79-0643,
8. Stewart, J. S., and F. D. Hart, Control of In- March 1979.
dustrial Wood Planer Noise through Improved 21. Pope, L. D., and J. F. Wilby, Band-Limited
Cutterhead Design, Noise Con. Eng., vol. 7, Power-Flow into Enclosures, J. Acoust. Soc.
no. 1, July-August 1976, p. 4. Amer., vol. 62, no. 4, 1977, p. 906.
9. Stewart, J. S., A Theoretical and Experimental 22. Piersol, A. G., and P. E. Rentz, Experimental
Study of Wood Planer Noise and its Control, Studies of the Space Shuttle Payload Acoustic
PhD. Thesis, North Carolina State Univ., Environment, SAE Paper 770973, November
1972, NASA CR-I12314, 1972. 1977.

10. Brooks, T. F., and J. R. Bailey, Mechanisms 23. Unruh, J. F., A Finite Element Subvolume
of Aerodynamic Noise Generation in Idling Technique for Structural-Borne Interior Noise
Woodworking Machinery, ASME 75-DET-47, Prediction, AIAA Paper 79-0585, March
1979.
September 1975.
11. Hart, F. D., D. L. Neal, and F. O. Smetana, 24. Dowell, E. H., Master Plan for Prediction of
Industrial Noise Control: Some Case His- Vehicle Interior Noise, AIAA Paper 79-0582,
March 1979.
tories, Volume 1, NASA CR-142314, 1974.
(Primary source--Warnaka, (3. E., H. T. 25. Moreland, J. D., Controlling Industrial Noise
Miller, and J. M. Zalos, Structural Damping by Means of Room Boundary Absorption,
as a Technique for Industrial Noise Control, Noise Con. Eng., vol. 7, no. 3, Novem-
AIHA J., January 1972, p. 1.) ber-December 1976, p. 148.
26. Lang, J., Differences Between Acoustical
12. Davies, H. (3., and R. H. Lyon, Noise Propa-
Insulation Properties Measured in the Labor-
gation in Cellular Urban and Industrial Spaces,
atory and the Results of Measurements in
J. Acoust. Soc. Amer., vol. 54, no. 6, 1973, p.
situ, App. Ac., vol. 5, no. 1, 1972, p. 21.
1565.
27. Levine, L. S., and J. J. DeFelice, Civil Heli-
13. Simpson, Myles A., Noise Barrier Design
copter Research Aircraft Internal Noise Pre-
Handbook (prepared for FHA by Bolt
diction, NASA CR-145146, 1977.
Beranek and Newman), available from NTIS
28. Howlett, J. T., S. A. Clevenson, J. A. Rupf,
as PB-266 378, 1976.
and W. J. Snyder, Interior Noise Reduction in
14. Kurze, U. J., Noise Reduction by Barriers, J.
a Large Civil Helicopter, NASA TN D-8477,
Acoust. Soc. Amer., vol. 55, no. 3, 1974, p. 1977.
504.
29. Smith, P. W., Concepts and Applications of
15. Moreland, J. B., and R. S. Musa, The Per- SEA, J. Acoust. Soc. Amer., vol. 65, supp. 1,
formance of Acoustic Barriers, Noise Con. 1979, p. $97.
Eng., vol. 1, no. 2, Autumn 1973, p. 98. 30. Smith, P. W. and R. H. Lyon, Sound and
16. Nordby, K. S., Measurement and Evaluation Structural Vibration, NASA CR 160, 1965.
of the Insertion Loss of Panels, Noise Con. 31. Lyon, R. H., What Good is SEA Anyway?
Eng., vol. 10, no. 1, January-February 1978, Shock and Vibration Digest, vol. 2, no. 6,
p. 22. 1970, p. 2.
17. Lockheed Missiles and Space Co., Response 32. Ghering, W. L., Reference Data for Acoustic
of a Panel to Reverberant Acoustic Excita- Noise Control, Ann Arbor Science Publishers,
tion, Cosmic Program Abstract, MFS-21774, 1978.
1972. 33. Lyon, R. H., Statistical Energy Analysis of
18. Dowell, E. H., Acoustoelasticity (Princeton Dynamical Systems: Theory and Applications,
Univ. Report AMS 1280, May 1976; NASA MIT Press, Cambridge, MA, 1975.
Grant NSG-1253), NASA CR-145110, 1977. 34. Manning, J. E., and N. Koronaios, Experi-
19. Tweed, L. W., and D. R. Tree, Three mental Study of Sound and Vibration Trans-
Methods for Predicting the Insertion Loss of mission to a Shroud-Enclosed Spacecraft,
Close-Fitting Acoustical Enclosures, Noise NASA CR-96144, 1968.

91
35. Lyon,R. H., andG. Maidanik,Reviewof 39. Kroes, P., R. Fleming, and B. Lempert, List
SomeRecentResearch on NoiseandStruc- of Personal Hearing Protectors and Attenua-
turalVibration,NASATN D-2266, 1964. tion Data, available from NTIS as HEW Pub.
No. (NIOSH) 76-120, 1975.
36. ANSI, American National Standard Method
for the Measurement of the Real-Ear Attenu- 40. Edwards, R. G., et. al., A Field Investigation
ation of Ear Protectors at Threshold, ANSI of Noise Reduction Afforded by Insert-Type
Z24.22-1957. Hearing Protectors (prepared for NIOSH by
Watkins and Associates, Inc.), available from
37. ANSI, American National Standard Method NTIS as DHEW (NIOSH) Pub. No. 79-115,
for the Measurement of Real-Ear Protection 1979.
of Hearing Protectors and Physical Attenua- 41. Schmidek, M. E., M. A. Layne, B. L.
tion of Earmuffs, ANSI $3.19-1974.
Lempert, and R. M. Fleming, Survey of
38. Anon. Noise Labeling Requirements, Hearing Hearing Conservation Programs in Industry,
Protectors, Federal Register, vol. 42, p. 120, available from NTIS as HEW Pub. No.
June 22, 1977. (NIOSH) 75-178, 1975.

92
 CHAPTER 7

Specific Noise Sources and Solutions

This chapter examines in detail several major "/.1.1 Jet Noise Theory
categories of industrial noise sources and tech-
niques for the alleviation of the noise problems Jet exhausts create noise in fundamentally the
they present. The selection and proportional cov- same manner regardless of whether they are hot or
erage of material used here was governed primarily cold, located in aircraft engines, or form part of
by the amount of NASA and related aerospace an industrial process. The high velocity gas inter-
research in areas important to nonaerospace noise acts with the ambient air at rest, producing severe
control engineers, namely, turbulence and rotor- shearing stresses. These in turn generate a mixing
stator interactions within flow ducts, blade config- region of high turbulence where individual eddies
urations of fans and compressors, combustion act as separate sources of noise. The resultant
noise, and jet exhaust noise. There is much theo- spectrum is typically broadband with high fre-
retical and experimental analysis of these topics; quency contributions from the shearing region
where prior work is innovative and relevant to and low frequency components from the region of
general purpose noise control, references are larger scale turbulence downstream of the jet (see
given. There are, however, several outstanding figure 7.1).
examples of NASA-sponsored technology that The total sound power generated by the mix-
have begun to show immediate practical value; for ing of the jet alone, W, is normally taken as
others, prototype testing has indicated a strong dependent on the eighth power of the exit velocity
potential noise reduction benefit. The most salient after Lighthill's theory.l The full expression is
features of these are discussed in sufficient detail
to provide a reasonable idea of their mode of k sjv
W- W
operation and construction, as well as their appro- c5
priateness for certain physical and acoustical
where
environments.
k is the constant of proportionality (depend-
ent on nozzle geometry concerned)
7.1 JET AND TURBULENCE NOISE 0 is the jet air density, kg/m 3
Sj is the fully expanded jet area, m2
This section discusses the theory of jet noise
and some of the more practical examples of
Mi_;g region of, \ f-- Boundary with free air
suppression devices from the many techniques
proposed. Valve noise is essentially caused by the
-scale lurDulence_ _ /_
generation of turbulence inside pipes after
formation of a jet at a restricted orifice, and thus
has many properties similar to free jets. Finally,
the characterization of flame-burning profiles and
Nozzle orifice J _/_ _ _ _ _ _ _ \
their concomitant turbulent effect on the ambient
atmosphere, whether in a lined combustor or in L_ Laminar core flow _ _'/

free air, remains imprecise, although aerospace • Fully mixed region ofl
investigators and commercial manufacturers of large-scale turbulence
burners have made increasing efforts to improve
prediction and suppression techniques. Figure 7.1 - Diagram of free jet.

93
 vj is the jet velocity, m/s
c is the speed of sound in ambient air, m/s. _.-r-&'_
8
-- tj

The frequency at which peak sound pressure

level occurs is related to the diameter of the jet
nozzle and the nozzle exit velocity by the simple
"I3
relation ,- +

Jet exhaust
t- o 0=o o _ h_-18o o
+ -10
9' =17_0° < 0' <- 110 °

where "_ o 1
_ -30
vj is the jet core velocity, m/s + °0, 1
D is the jet nozzle diameter, m g o -40
"_ _j
fp is the peak frequency, Hz. --50b
<
O --60 L • L - L _ L
0
As the ratio between the jet exit pressure and I --1.6 --1.2--0.8--0.4 0 0.4 0.8 1.2 1.6
z
the ambient air pressure is increased, then there is _-
(/3
Nondimensional frequency parameter, log F

a corresponding increase in the velocity of the dis-

Figure 7.2 - Prediction curves for shock-free jet
charge. This continues until the flow becomes
noise sound pressure level spectra (reference 2).
sonic at a pressure ratio of around 2, when a
choked condition is said to exist with little further
increase in flow velocity. The parameter used to
where
describe this is the Mach number, which is equal
f is the frequency, Hz
to the ratio of flow velocity to the ambient speed
of sound propagation; sonic flow is said to exist Tj is the jet temperature, K
when this ratio is equal to unity. It is not uncom- To is the ambient temperature, K
mon to find values of the Mach number just above D e is the equivalent circular nozzle diameter,
x/4S/Tr, m
this in industry when supersonic conditions are
said to prevail and shock waves may be generated. vj is the jet exit velocity, m/s
S is the nozzle area, m 2
Air used to pressurize industrial shop air lines can
often be at a pressure more than 3.1 × 105 N/m 2
(45 psi), compared to the ambient air pressure of and where the effective angle 0' is given by
1.0 x 105 N/m 2 (15 psi). Shock waves develop be-
cause the acoustic pressure disturbances can no
longer radiate away from the source region faster
than they are generated, thereby producing high
0: °'
amplitude sound waves with a characteristic saw-
where
tooth-type time history.
A prediction method has been developed by 0 is the actual angle from inlet, deg
NASA for free jet noise incorporating the results c is the velocity of sound, m/s.
of both cold- and hot-flow jets, using analytical as
The ordinate is defined as
well as empirical data. 2 A directionality pattern
resulting from convection and refraction effects is
always established; figure 7.2 shows a distillation
SPL - OASPL9oo + 30 log [1 + Mc
of directionality effects predicted using the above
method for subsonic jets. The abscissa is defined (1 + Mc5) - 1/5 cos 0]
according to

where
SPL is the level in one third octave
F =--fOe band frequency at 0
(Tj_ ) °'4(l +c°s °')
_,j 1"0 OASPL9oo is the overall level at 0 = 90 °.

94
 Note: The convection Mach number M c is 7.1.2.1 Velocit) _ reduction

defined as approximately 60 percent of the con-

Because the exponent of the velocity term in
ventional Mach number, to take account of the
the acoustic power relation for jets is of the eighth
average jet velocity, i.e., M c = 0.6vj/c. order, any velocity reduction has a markedly bene-
Figure 7.2 includes a family of curves, each of
ficial effect. Frequently, shop air pressure in com-
which applies to a particular value of 0'. The effect
pressed air lines is much higher than necessary to
of temperature with directional angle, as indicated achieve the requisite force at the output; a reduc-
by the abscissa scale, can be related to refraction tion in pressure immediately attenuates the gener-
effects: higher frequencies are more strongly ated sound pressure.
affected. Consequently, the high frequency con- One way of reducing velocity without directly
tent increases as the angle with the jet axis in- reducing pressure is by stepping up the pipe diam-
creases. At 0' _< 110 ° a simple model is suffi- eter some distance from the orifice. To compute
ciently accurate that only one curve is necessary;
the total emitted sound power it is then necessary
in any event, at these angles the noise levels are
to add sound powers from the new free jet with
relatively low.
the expanded diameter and lower velocity, as well
In simple terms, for velocities up to the speed
as from the source within the pipe for which no
of sound, overall peak noise levels are of the fol- distance attenuation should be assumed. If this in-
lowing relative magnitudes: + 5 dB at 150 ° from
crease in pipe diameter does not directly reduce
axis; 0dBat 120°; and -5 dB to - 10dBat 90 ° to
the radiated noise, it may nevertheless allow the
0 °. Where shock waves are generated, direction- addition of in-line silencers. A similar approach
ality is normally assumed to be uniform at all
has recently been evolved at the NASA Lewis
angles.
Research Center for supersonic jet flow. The basic
One study 3 indicates that the far-field
scheme is illustrated in figure 7.3, which shows the
acoustic intensity may be satisfactorily predicted flow phenomena within the can-type suppressor. 6
without taking detailed measurements along the
jet axis, if measurements are made near the conical
outer boundary of the jet exhaust flow.
If the exhaust stream is interrupted by any Low pressure area

kind of solid boundary, then a new source of noise Vent pipe _ ':_::::::::::_ i :: ::::::_ :::::_:
is generated, the intensity of which is dependent
on the location of this boundary, the geometry of
the surface involved, and the acoustic nature of
the surface. Subsonic jets impinging on a hard
reflector may produce an increase of about 7
dB(A), and if the surface is slotted in some way p re,jure
4
then the increase may be an extra 10 dB(A) over
Suppressor -J L Jet streamlines
this. 4

Figure 7.3- Noise suppressor for vented high-

pressure gas (reference 6).
7.1.2 Jet Noise Reduction

The technique basically uses overexpansion

Common industrial jet noise sources include of a supersonic jet to create a series of strong
pneumatic discharge systems, steam vents, and shock waves. The space between the jet stream-
blow-off nozzles for machine tool surfaces. Many lines and pipe enclosure is partially evacuated by
practical procedures can attenuate jet noise out- the flow, thus causing the overexpansion. The gas
put. These procedures may be categorized as core therefore achieves a high Mach number before
velocity reduction techniques, dissipative or reaching the pipe wall, which, together with the
absorptive devices, and the introduction of shield- forced turning of flow at the wall, creates shock
ing secondary air flows. Reference 5 provides waves in the flow. In consequence, large total
comprehensive performance specifications for pressure losses are incurred, resulting in a decrease
common pneumatic exhaust and ejection in jet velocity and a reduced sound pressure level
silencers. at the pipe exit. Acoustic tests on scale models

95
gavemaximum noisesuppressionupto21dB.The also shifts the spectrum to high frequencies and
mostcriticalfeatureof thedesign
isto ensure
that may be accompanied by some net velocity reduc-
the overexpanded flow reachesthe suppressor tion. Shrouds should usually be sufficiently long
wallsbeforethe exit. Approximatepipedimen- to prevent direct radiation at angles greater than
sionsarederivedfrom 10 ° from the jet axis, and reductions of up to 20
dB are achievable. The principal defect of these
devices is their propensity to generate back pres-
L /Rsp ) sure; also, severe size constraints often apply in
RZ = 11.4 _Rvv-1
practical working situations.
where
R v is the vent pipe radius
Rsn is the suppressor radius
L is the length of suppressor.

The device is considered to be useful only

where the loss of jet thrust is not important, such
as in excess steam venting at power plants. Fur-
thermore, in general only low frequency noise is
suppressed. 7

7.1.2.2 Dissipative techniques

Absorptive shrouds are usually the most ef-

fective form of jet noise suppression, because they
shield against direct radiation from the source
area and absorb acoustic energy as a result of
multiple reflections inside the shroud. Several
examples of absorptive devices are shown in figure
7.4. The introduction of diffusive elements such as
"pepperpots" or grid mesh screens within the
shroud breaks up the flow into many smaller jets
with a lower net total acoustic power output. This

Figure 7.4 - Examples of several kinds of jet noise silencers (by courtesy of G. IV. Norris Co. and Vlier
Engineering).

96
 Manyhighpressure jet exhausts fromcom- settling chambers with pressures nearly equal to
pressed air machineactuatorsandtoolshavea the ambient level, cancelling abrupt positive and
smallsize;because of their sizeandfunctional negative pressure gradients within the jet stream.
design,they may be unsuitedto absorptive In addition, because of the geometrical configura-
shrouds.In consequence, it is oftenmostconven- tion of the plug, some noise reduction of the
ientsimplytopipetheexhaust awayfromtheim- shearing action of the jet stream with the ambient
mediate workingareausinga thick-walled flexible air is observed at all subsonic and supersonic flow
tube. conditions.
Onedevice recently
developed in experimental The most important parameters are plug
formby NASA8,9usesa centerbodyporousplug length and porosity; apparently, the ratio of the
insertedalongthelongitudinal axisof thejet atthe plug to nozzle diameters is less significant. Super-
orificeas illustratedin figure7.5.This is most sonic flow may be present at the end of the shorter
effectivefor noisesuppression ofsupersonic flows plugs tested, so that a longer configuration might
andprovides a shock-free flowovera widerange be more effective at high pressure ratios. Figure
of pressure ratiosfrom 1.14to 3.72.Shocksand 7.6 shows the extent of attenuation possible with
the resultantnoiseare eliminatedby pressure these devices at lengths of 25 cm and 69 cm.
equalization aroundthejet axis.The poresbe- Porous plug suppressors concentrate noise
tweenthejet flowandinteriorplugcavityactlike sources in the flow, leading to a more uniform
directivity pattern, and reduce the size of the
large-scale structure of the flow. These results
tend to enhance the performance of any acoustic
baffle at the jet, further reducing noise.

Pressure ratio = 3.72

Aspect ratio (diameter plug/diameter nozzle) = 0.55

Porosity = 1.2%
120

"10

_ 110

100
(69 cm)

_r'
o 90
03
O-

I i
o'.3 Ito 3to 3o
Freouencv (kHz)

Figure 7.6 - Power spectral density of far field

pressure at 90 ° from jet axis with porous plugs
(reference 9).

7.1.2.3 Shielding by secondary air flows

Shielding by secondary air flows requires the

introduction of a secondary air flow which acts as
a lubricant and thereby reduces the sharpness of
the shearing velocity gradient between the ambient
air and jet core. Many of the techniques for
achieving this have been adopted from pioneering
research on aircraft jet engines.
Figure 7.5 - Porous plug jet noise suppressor (by By using a series of smaller nozzles at the exit
courtesy of NASA Langley Research Center). or a suitable eductor tube, ambient air becomes

97
 /-- Secondary or outer air flow or to provide emergency pressure relief systems
designed to operate either fully open or fully
_-'_---- =-2 -_-................. closed. The flow velocity of fluids and gases in
pipes is usually low, but where valves are used to
regulate high pressure or high velocity flows, then
the "choking" of flow increases the velocity to the
--_r_-" - " - "- II III
eair flow ----'-- -_-----____ speed of ambient sound, i.e., Mach 1 (see section
7.1.1). Although there are generally no increases
Figure 7. 7 - Jet noise suppression by entrainment in velocity beyond the critical pressure ratio, the
of ambient air flow. Noise producing regions: L density does increase, and a mismatch in flow con-
core-outer flow mixing; H, outer flow - ambient ditions at the vena contracta can produce down-
mixing; HI, merged jets - ambient/nixing. stream shockwaves (see figure 7.9).

Aoutef
Area ratio: Valve stem seals
Valve body
Acore

0:r 0.5

f,ow
/ I " J L,--r_
' r i
-----

:!i
8,1
LT_-rJ r
L__J
Propagation
of shock
waves
Propagation
acoustic
higher
of
plane and
order waves

Figure 7.9 - Schematic representation of valve

noise generation and propagation.

< -12
O
-14 1 t I 1 J
0 0.2 0.4 0.6 0.8 1.0 The turbulence produced by any effective
valve as a by-product of the required pressure
Voute_

Velocity ratio, drop often leads to the generation of intense noise,

Vcore with sound power levels of up to 150 dB. Valve
noise is similar to jet noise, in that sound power is
Figure 7.8 - Coannular noise reduction for jets
closely proportional to the eighth power of the ori-
(adapted from reference 10).
fice velocity in subcritical pressure ratio regions.
In an extensive review of valve noise prediction
schemes, 12 it was pointed out that for subcritical
entrained in the air flow before the jet core is fully
pressure ratios, the location of the most intense
developed, as shown in figure 7.7. The major
noise generation is about three pipe diameters
noise sources are the core-outer flow mixing
downstream of the valve orifice. Above the critical
region, the outer flow-ambient air mixing region,
pressure ratio, shock waves are present and there
and the merged-jets-ambient mixing region. Over-
appears to be no simple means of accounting for
all reductions of between 5 and 10 dB(A) can be
their contribution to total acoustic power.
achieved using this approach, and figure 7.8 shows
As a result of the production of shock waves
a family of curves yielding attenuation predictions
and other high acoustic pressure disturbances
for different ratios between the outer and inner
downstream, pipe ringing may occur. In high pres-
nozzle areas.
sure systems, the pipe walls are probably already
Jet noise may also be reduced by imparting a
stiff and therefore have a number of well-defined
rotary component to the flow by the introduction
resonances easily excited by the broadband excita-
of radial vane structures 11; this may be particu-
tion forces inside the pipe. For this reason, valve
larly useful in hot gas exhausts.
noise often contains strong discrete frequency
components in addition to the broad frequency
7.1.3 Valve Noise
spectrum normally associated with jet noise.
Valves are used either to control flows by cre- Noise reduction may be achieved by modify-
ating a pressure drop, which is known as trimming, ing the valve itself, silencing the sound propagated

98
upstream or downstream of thevalvewithin-line to use a series of staged pressure reductions or to
units,or increasing thetransmission lossof pipe utilize a cage with diametrically opposed holes in
wallsasdiscussed in section5.2.2. the valve trim. The flow through opposite pairs of
Quiet valvesare producedwhich operate these holes meets in the center of the cage and
according to eitheror bothof twoprinciples: (a) creates a continuous fluid cushion which prevents
thedivisionof thetotalpressure dropintoa series cavitating fluid from reaching the valve plug and
of smallersuccessive pressure drop stages,each seat line. Reductions of up to 6 dB are possible
belowthecriticalpressure ratio,and(b)theuseof using this approach.
multipleflow restrictions in parallel,with a high For path control of valve noise, in-line
length-to-diameter ratio, whichcausehigh fric- silencers are most effective when situated about
tionalandviscouslosses. Examples of bothkinds two to three pipe diameters from the valve. The
aregivenin figure7.10;reductions of upto 15-20 length of pipe between the valve and silencer
dBarepossible. Diffusersmaybeplacedsomedis- should be heavily wrapped in order to reduce
tancedownstream of thevalveorificetoattenuate sound transmission through the pipe wall. Note
thenoisefurther.Reasonable predictions ofsound that the outlet velocity of the silencer is controlled
powerlevelcan be madefor valveconfigura- by the configuration of downstream pipework.
tions,12,13althoughspectralcontentdepends For vent attenuators where gas is exhausted to
heavilyonindividualcharacteristics. open air, the diameter should be large enough to
For hydrodynamicflow with fluids, the keep the merging gas velocity low (less than 50
majorproblemisto prevent cavitationatthevena m/s); otherwise a secondary jet noise source could
contracta.Cavitation occurs because, as the fluid be generated at the outlet.
passes through the control valve orifice, its in-
creased velocity causes the pressure to drop below 7.1.4 Fluid Flow Problems
that of the fluid vapor pressure, and hence bub-
bles are formed. As the flow moves downstream, Hydraulic systems may present noise prob-
the velocity decreases again with resultant pressure lems, especially where motor-pump attachments
recovery. Implosion of the bubbles occurs when are located. The general rank order for the level of
the static pressure becomes greater than the liquid noise generated by such pumps is screw-drive,
vapor pressure and consequently extremely high piston-drive, and gear-drive, the latter being the
pressure shock waves are produced. These have worst case. ]4 Noise generated in the fluid line itself
serious consequences, not only regarding noise, occurs because of sharp bends, restrictions, and
but also because of the erosion of valve parts. The internal pipe fittings, which cause cavitation and
most effective way of combating erosion is either turbulence. A series of early NASA studies 15-18
examining acoustic propagation in a liquid flow
line looked at the separate effects of 90 ° bends,
closed-end side branches, dynamic response of the
line, and pure tones superimposed on laminar
ve plug
flows of liquids with varying viscosity. It was
found that the amount of energy imparted by a
massive fluid to a pipe wall because of turbulent
flow at internal asymmetries could be satisfacto-
(a)
rily predicted. Some of the results showed that
there was no substantial evidence for acoustic
reflection effects at elbows or pipe terminations
per se.
I'L I'V_ SectionA-A The interrelationship of the parameters influ-
encing fluid turbulence in pipes is conveniently
(b) (c) (d)
shown using the Reynolds number, NR:

Figure 7.10- Examples of quiet valve profiles Dvo

(adapted from reference 13). (From P. N. NR-
#
Cheremisinoff and P. P. Cheremisinoff, Industrial
Noise Control Handbook, 1977. Used with the where
permission of Ann Arbor Science Publishers.) D is the pipe diameter, m

99
 v is the fluid velocity, m/s of the sound power in burners is difficult because
Q is the fluid density, kg/m 3 of the wide range of variables involved, not the
# is the coefficient of viscosity, N.s/m 2. least of which is the flame geometry.
An examination of the characteristics of a
When the fluid particles move parallel to each can-type combustor with a large by-pass ratio of
other and to the general direction of flow, laminar draft to consumed air produced the following pro-
flow is considered to prevail. This normally occurs portional relationship for sound power level,
at Reynolds numbers of less than 1200. At higher PWL20:
Reynolds numbers turbulent flow becomes estab-
lished, but the level of sound generated is not PWL ¢x mfm 2
necessarily proportional to the value of the Rey-
nolds number concerned. where
Dissipative mufflers or tuned reactive units m f is the fuel flow rate
(see section 7.3.4) may be employed where appro- m a is the air flow rate through the can.
priate. When a pipe size for nonturbulent flow
conditions is not feasible, then acoustic energy Oscillations and combustion roar are often
may be attenuated using flexible hoses, surge both produced; typical frequency spectra for these
chambers, or vibration-isolated supporting struts; are shown in figure 7.11. 21 The former is
in addition, the transmission loss of the pipe walls characterized by a discrete frequency spectrum
may be enhanced by wrapping or increasing the and high levels and is not always present, whereas
wall thickness. the latter has a relatively broadband spectrum of
moderate levels and is associated with any turbu-
7.1.5 Furnace and Combustion Noise lent combustion process. Oscillations in particular
need to be suppressed, since they represent large-
The noise from natural and forced draft fur- scale turbulence which can cause severe vibration
naces is predominantly low to mid frequency, and and even fatigue problems in the surrounding
one analytical study 19 confirms that combustion structure.
noise is inherently broadband with peak frequen- A general discussion of combustion noise in
cies typically in the 400-1000 Hz range. There are reference 20 indicates that the generation of a
also indications that the amount of heat released roar, often exacerbated by amplification at
per unit mass of fuel is the parameter that controls natural frequencies of the furnace, depends on the
the sound level at the peak frequency. Prediction square of the turbulence intensity. It is therefore

130

120 80

o 110 70

1 O0 60
_x

90 50

1 I 1 I 40 I I 1 I I I
80
5 10 20 30 40 100 200 400 600 1000 2000 5000

Frequency (Hz) Frequency (Hz)

(a) (b)

Figure 7.11 - Noise spectra for combustion phenomenon (reference 21). (a) Combustion-driven oscillation.
(b) Combustion roar. (Used by permission of the Noise Control Foundation and A. A. Putnam.)

100
importantto reduceturbulenceintensitywhile configuration.
22Otherpossible techniques include
maintainingsatisfactorycombustionperform- ductedcombustors whichadjustthefuel/airmix-
ance.This canbe achieved by reducingburner ture to giveshorterflameswith inherentlyless
flowrates,reducing excessiveswirl,andmatching large-scaleoscillations,and the use of rods
flowdirections to thegeometry of thecombustion insertedinto the flame region. 23 Combustion
vessel.Flame-holder geometriesare especially oscillations
areoftenbesteliminated orattenuated
importantin influencingtheefficiency andrough- by tunedresonator cavities.
nessof combustion. Figure7.12shows thebenefi- It is sometimesexpedientto construct
cial effectof usinga plane-slotted flameholder externalacoustically
treatedplenums orenclosures
with two counter-swirling streamsovera simple andleaveasidethemoreunpredictable approach
perforatedcanholder.Apart from reducingthe of controllingthesourcenoise.Thesecanbebuilt
total powerspectraldensityby some15dB, the arounda completegroupof furnaces,andit is
moststablecombustion is alsoachieved withthis sometimes necessaryto incorporatefansto pro-
videsufficientsupplemental airto feedthem.
0

7.2 FAN AND COMPRESSOR NOISE

- 20 In a recent review of fans used in industry, 24

v
it was pointed out that more than half of 70 identi-
fiable noise sources were the result of either supply
-40 and extract fans or compressor systems. These
fans and compressors are in wide use for moving
- 60
large quantities of air for cooling, drying, or oil-
demisting operations within most manufacturing
and processing operations, as well as for conven-

_80!
tional heating, ventilation, and cooling systems
for human comfort. Fortunately, relatively simple
acoustic solutions may be applied to alleviate
I
- 100 I i i I
200 400 600 800 1000 noise problems. Conventional industrial fans have
Frequency (Hz) lower blade loadings and lower speeds than air-
craft engine fans and compressors and so they
(a)

- 2O
oI
- 20

"2 - 40
-40

-60 _-60

0
a_ --80
o -80

-100 ,
200 ,
400 ,
600 ,
800 ,
1000 _lOO
o 200 400 -- ±
8OO lOOO,
Frequency (Hz) Frequency (Hz)

(b) (c)

Figure 7.12 - Power spectral densities of various flame holder configurations (reference 22). (a) Plane,
slotted flame holder with swirl vanes. (b) Plane, slotted flame holder. (c) Perforated can flame holder.

101
eliminate many of the more complex sound-gener-
ating mechanisms associated with the latter.

7.2.1 Fan Types and Noise Prediction Techniques

There are many types of fans because they

have many different functions. However, once a
Outlet
fan has been selected for aerodynamic efficiency,
load requirements, and so on, little can be done to (a)
improve source noise characteristics; noise reduc-
tion must be achieved by absorption lining and by
ensuring flow stability within the vicinity of the
fan unit.

7.2.1.1 Axial fans

Axial fans resemble a propeller or rotating

vanes, set such that the airflow is parallel to the
axis of rotation. They produce relatively little _x,_ _ CUtcutawaysh_ingdetaJ_
....
static pressure, and energy is transferred to the air
_'Inlet of fan wheel
by imparting a twisting motion to it. According to
a detailed study, 25 the dominant noise mecha- (b) (c)
nisms of low-tip-speed axial-flow fans can be
separated into nonrotational and rotational noise. Figure 7.13- Major fan types. (a) Tubeaxial. (b)
The former includes laminar boundary layer Centrifugal tangential. (c) Centrifugal radial.
vortex shedding, the formation of vortices accord-
ing to the clearance between blade tips and the fan
housing, and blade stall, all of which generate 7.2.1.2 Centrifugal fans
broadband noise. Rotational noise, on the other Centrifugal fans create a pressure differential
hand, results from blade interactions with inflow by a combination of centrifugal forces and rotat-
distortion and turbulence, as well as with the re- ing velocity. The noise generated arises from dis-
flected pressure field from nearby structures. This crete tones at the blade pass frequency and from
component is characterized by a series of har- aerodynamic noise associated with the shearing
monically related pure tones determined by the action of the blades on the air flow and the result-
blade pass frequency according to ant turbulence. There are two major types of cen-
trifugal fans, both of which are illustrated in
nKN
figure 7.13. The radial fan usually consists of
fn- 60 Hz fairly heavy, flat blades built around a central
hub; it is of particular value in heavy-duty applica-
where tions where foreign material, such as woodchip-
N is the revolutions per minute, rpm pings or heavy particulate matter, passes through
K is the number of blades the unit. The blade pass frequency component is
n is the integer equal to 1, 2, 3, etc. especially prominent and is calculated in the same
fn is the nth harmonic frequency, Hz. way as for axial fans. The second type is some-
times known as the tangential fan; it consists of a
Axial fans are used extensively throughout indus- circumferential wheel, within which are mounted
try, and a wide range of pressure ratios and air a series of blades aligned parallel to the central
volume handling capacities are possible. The sev- inlet axis, and often with a forward- or rearward-
eral configurations commonly available may be curved characteristic with respect to the direction
categorized as shrouded with guide vanes (vane- of rotation of the wheel. These fans tend to pro-
axial); shrouded without guide vanes (tubeaxial); duce a fairly uniform spectrum, in contrast to the
and freely turning without a shroud (propeller). radial type, which have dominant low frequency
An example of a tubeaxial fan is shown in figure components. They are used extensively in heating
7.13. and ventilation systems, and the backward-curved

102
type,in particular,hasthebestefficiency of allthe measure of fan performance. At the appropriate
centrifugalfansandcorrespondingly the lowest flow coefficient, Fan A operates at peak effi-
soundpowerlevels. ciency, whereas for the same value of the flow'
Twoparameters areof particularimportance coefficient Fan B is not operating at optimum effi-
in influencingthenoisecharacteristics of circum- ciency and may well be noisier in consequence.
ferentialfans.Thescrollarounda centrifugalfan
is sometimes curvedsothatastheangletraversed 3.18 x 105 • Q
Flow coefficient,
by thebladesincreases, sodoesthedistancebe-
tweenthe bladetips andthe scrollcasing.This
helpsto ensurea constantmeanvelocityof flow
aroundthefan circumference, whichis desirable 1000. AP
Pressure coefficient, ¢ -
for minimumnoiseproduction.Additionally,the
_v7
clearance distancebetween the outletscrolledge
or cut-offandthebladetipsshouldbearound5to
10percentof the fan wheeldiameter.Smaller Q00 AP
clearances thanthisincrease theamountof noise Efficiency,
60w; Q
produced.
7.2.1.3 Fan noise prediclion where
Q is the air flow rate, mS/rain
After the fan type is selected there are several
N is the fan speed, rpm
procedures for optimizing the fan parameters to
D t is the tip-tip diameter, cm
achieve maximum operating efficiency. 26, 27 How-
is the pressure change, kPa
ever, this does not necessarily ensure the quietest
is the air density, kg/m 3
running profile since blade pitch, blade curvature,
vt is the fan blade tip speed, m/s
number of blades, and tip solidity all act as miti-
_0 is the 1.2015 kg/m 3
gating factors, in which changes resulting in im-
W z is the fan input power, kW.
proved efficiency do not unambiguously decrease
noise levels. 28 For geometrically similar fans, a
Moreover, for a given type of fan and at the
flow coefficient, as defined below, exists at which
same value of the flow coefficient, effects on the
a given air pumping task can be performed more
sound power generated resulting from diameter
quietly. There is evidence to suggest that the flow
and velocity changes are given by
coefficient at which minimum noise occurs is
unique for a particular type of fan, and it is logical
W 2 = {DT2_2 and W._" [/vt2/x\
therefore to compare sound power levels at these
points to gain some appraisal of the value of dif- w, : \v,,/
ferent fan designs. Figure 7.14 illustrates this ap-
where
proach using the pressure coefficient as a direct
WI, W2 is the initial and final sound powers
respectively, W
Efficiency, x - - - X is the fan velocity exponent.
Pressure coefficient, q/----

Peak efficiency Peak efficiency Note: The velocity exponent may only be
i
_L Fan B determined experimentally by measuring sound
___. . FanA
i I \ power at a large variety of velocities. It normally
lies in the range 5.5-6.5 and is a function of both
perating the flow coefficient and fan design.
_-.,Operating
I _ point oint Several empirical equations give a rough esti-
I
Flow coefficient d_ mate of fan noise in regard to such parameters as
Flow coefficient _b
static pressure developed by the unit, volume flow
Figure 7.14- Comparison of optimum fan system rate, and horsepower capacity (see, for example,
operating points (reference 26). (Used with the reference 29). However, a more precise prediction
permission of the Noise Control Foundation and scheme has recently taken into account in more
R. C. Mellin.) detail the fan type and fundamental blade pass

103
frequency. 3° The basic empirical data have been The blade pass frequency component of fan
reduced to specific sound power levels expressed noise is included by adding the appropriate blade
in octave bands, as shown in table 7.1; these are frequency increment, BFI (see table 7.1) to the
the sound power levels generated by a fan type specific sound power level of the octave band
when the flow rate is 1 m3/sec and the static pres- which includes the blade pass frequency, as calcu-
sure 1 kilopascal. lated from
To obtain the actual sound power level,
PWL, the specific sound power levels for the rele- KN
f- Hz
vant type of fan are chosen, PWksp, and adjust- 60
ments made as necessary according to the follow-
ing relations: where
K is the number of blades
APWL = 101ogQ + 101ogP dB N is the revolutions per minute, rpm.
PWL = PWLsp + A
Finally, for estimating the values of propa-
where gating sound pressure levels, this procedure gives
Q is the air flow rate, m3/s the total sound power generated and it is normal
P is the total pressure, kPa. to subtract 3 dB in each octave band to obtain the

Table 7.1 - Specific sound power levels for various fan types (reference 30).

Fan type Wheel size Octave band center frequency (Hz)

BFI
(m) 63 125 250 500 1000 2000 4000 8000

Centrifugal, radial > 1 98 94 90 87 83 78 75 74 7

(material handling) < 1 110 106 100 91 88 85 80 86 7

Centrifugal, radial > 1 93 87 90 87 85 80 78 77 8

(pressure blowing) >0.5, < 1 103 96 96 93 93 88 86 85 8
<0.5 111 105 106 98 92 87 86 81 8

Centrifugal, tangential >0.75 80 80 79 77 76 71 63 55 3

(airfoil, backward <0.75 84 86 84 82 81 76 68 60 3
curved, or backward
inclined blades)

Centrifugal, tangential All 95 91 86 81 76 73 71 68 2

(forward curved)

Vaneaxial > 1 87 84 86 87 85 82 80 70 6
< 1 85 87 91 91 91 89 86 80 6

Tubeaxial > 1 89 87 91 89 87 85 82 75 7
< 1 88 89 95 94 92 91 85 83 7

Propeller All 96 93 94 92 90 90 88 86 5

BRI refers to blade frequency increment, which measures the relative magnitude of the discrete tone at the blade pass
frequency.

Power levels expressed in decibels re 10 12 W.

104
soundpowerradiatedfrom eitherthe inlet or on the low pressure side of the blades, which re-
outE_t
alone. suits in a very narrow frequency band of noise.
This may be reduced considerably by changing the
Example: Calculate the sound power levels boundary layer from laminar to turbulent flow
propagating upstream in octave bands for a six- before reaching the blade trailing edge using serra-
bladed axial fan of tip-tip diameter !.5 m, and tions which create small-scale vortices. 33,34 The
turning at 1000 rpm, capable of producing an air serrations break up the blade tip vortices and
flow rate of 4 m3/s and a total pressure of 2 kPa. change the character of the fan wake vortex noise
from periodic to broadband. With tip speeds less
Octave Band Center Frequency (Hz) than 135 m/s, attaching serrated brass strips to the
leading edge of a rotor gave an overall sound pres-
63 125 250 500 1000 2000 4000 8000
sure level reduction of 4-8 dB on a 1.5-m-diameter
from table 7.1 rotor. 33 With proper design of the serrations, one
study showed that potential reductions of up to
PWLsp for 85 87 91 91 91 89 86 80 37 dB in generated sound power were possible. 34
vaneaxial
from A = 10log Q + 20 IogP
fan < 1.0 m Figure 7.15 shows some typical reductions in noise
and substituting, Q = 4 m3/s
level with the use of various lengths of serration
and P = 2 kPa
'_ - lOIog4 + 20 log2 12dB strips.

Therefore actual sound power level,

PWL = PWL_p + 12 Propeller speed = 2000 rpm
Equivalent air flow = 12.2 m/s
PWL under 97 99 103 103 103 101 98 92 80

operating
conditions
from fn = KN/60 '
i ' i i i !
and substituting K - 6 and N = 1000 rpm
70 -- Unmodified propeller
fl = 100 Hz
Serration strip 1116th span i
Serration strip 1/8th span
Blade pass 0 6 0 0 0 0 0 0
60
frequency
-og
component, r"
BFI
50
oX
Total corrected ..........
_, Serration!
] strip] 1/4thl spanli "_: i._l[i_
r-

PWL 97 105 103 103 103 101 98 92 ffl 40

63 125 250 500 1000 2000 4000 8000
Upstream PWL 94 102 100 100 100 98 95 89
One-third octave band center frequency (Hz)

Several researchers working on comprehen- Figure 7.15- Vortex noise reduction using ser-
sive fan noise prediction techniques 31,32 have indi- rated strip on propeller (reference 35). (Used with
cated that where the inflow is isotropic and the permission of the Noise Control Foundation
homogeneous, adequate procedures already exist, and R. E. Hayden.)
but in the more general case of disturbed or turbu-
lent nonisotropic flow there is a serious dearth of
Serrations on the trailing or leading edges of
knowledge.
low speed rotors can reduce the noise of small-
scale ventilation fans, as is illustrated in figure
7.2.2 Fan Noise Reduction 7.16. This design gives a reduction of 9 dB at a
characteristic 2-kHz pure tone using serrated
After choosing the most suitable fan for a trailing edges without sacrificing fan pressure
particular application from both utility and acous- performance.
tic points of view, little can be done further to An advanced blade design concept has recently
reduce source-generated noise. However, blade been tested by NASA in conjunction with several
modification techniques may reduce vortex shed- commercial contractors. 36 By sweeping the blade
ding and subsequent noise generation. leading or trailing edges, fluctuating forces caused
Vortex shedding from fan blades occurs as a by a distorted inflow may to some extent be can-
result of instabilities in the laminar boundary layer celled and shock formation resulting from super-

105
 / \ / \ / \
I \ I \ / \_
la) (b) (c)

Figure 7.17 - Forward view of several swept blade

configurations with subsonic leading edges (refer-
ence 36). (a) Swept back. (b) Swept forward. (c)
Compound sweep.

IL, PWL BVr41 dB

where

B is the turbulent energy

is the relative flow velocity
Vre I

L x is the axial distance from fan center.

To ensure minimum turbulence and fairly uni-

form flow conditions at a fan inlet, there should be
a distance of at least one fan diameter between a
Figure 7.16- Example of small ventilation fan
fan and any bends or upstream elements such as
with serrated trailing edge (by courtesy of Rotron,
reduced cross sections, side branch take-offs, and
Inc.).
so on, to allow some settling of the air flow. In this
connection, all flexible connectors should be as
tight as possible and tapered inlets and cross-sec-
sonic rotating speeds prevented. Sound power is tional area change sections should be used. These
proportional to the velocity component normal to general principles are illustrated in figure 7.18; they
the blade leading or trailing edge, fn, according to a apply to centrifugal fans as well.
factor of (Vn)6. By progressively sweeping a blade For ground static testing of aircraft engines
along its span (see figure 7.17), this component is with the kind of turbulence-free inflow typical of
gradually reduced. Thus by appropriately designing an in-flight situation, NASA has developed a
the sweep for the particular aerodynamic environ- number of acoustically transparent configurations
ment in which a fan will operate, a considerable which are fitted around the engine duct opening. 38
reduction in vortex shedding noise may be achieved These inflow control devices reduce turbulence
using one of the basic configurations shown in before the air reaches the fan. They are con-
figure 7.17. Reference 36 gives a comprehensive structed of a honeycomb structure, sometimes
account of design procedures for sweeping rotor with perforated plate panels as well. All of them
blades and stator vanes with regard to aerothermo- were successful at reducing blade pass frequency
dynamic design parameters, as well as geometric tones and conceivably may be of value in reducing
and acoustical considerations. inflow turbulence in more common duct noise
The effect of entry turbulence on fan blade problems, although there may be an associated
loading is particularly important, especially because pressure loss penalty.
the greatest inflow turbulence occurs in the boun- Having taken all practical measures to reduce
dary layer right next to the duct wall, where the fan fan noise at the source, absorptive parallel or co-
blade speed is highest. 31,37 The indications are that annular baffled silencers may provide good high
blade root and tip regions are both primary sources frequency attenuation with minimum pressure
of sound, and that boundary layers are often pres- loss. It is particularly important to minimize pres-
ent at both. One relationship evolved after a com- sure losses, since many industrial fans are of the
prehensive study of turbulence noise 31 showed that low pressure type and are extremely susceptible to
both the sound intensity level and sound power blockages in the flow on either side. This can be
level may be related to the turbulence energy as accomplished by using silencers with a cross-

106
 i;iiii',ii,i
= 11."---I F_--_------_--'--_

[a) Before

/ /I

Plain inlet Impeller immediately Slack flexible

after bend connector

[-- Duct diameter min.

I I _1

.'...%._

(b) After
_//i_._ _..: ----

-//,,,;r -r

Coned or bellmouth Settling length Taut flexible

inlet before impeller connector

Figure 7.18 - Methods for reducing turbulent flow conditions at fans. (From lan Sharland, Woods Prac-
tical Guide to Noise Control, 1971. Used by permission of Woods Acoustic.)

sectional area of 1.25 to 1.5 times that of the fan generally their dominant frequencies are much
duct, and by using turning vanes at bends. Plenums higher. Sliding vane rotary pumps are noisier than
are frequently used to good effect in low pressure screw types.
air heating and ventilation systems, and their Compressor noise is best treated at or near
attenuation properties are considerably enhanced the source, although, as in many other situations,
by sound absorptive lining and baffles openings. it is often more straightforward and economical to
All of these techniques are discussed in more detail use an appropriately designed acoustic enclosure.
in section 7.3.4. For reciprocating machines, which produce domi-
nant low frequency sound, reactive silencers are
7.2.3 Compressors and Rotor-Stator Interactions best because dissipative silencers would need to be
inordinately large to achieve the same results; con-
There are a multitude of compressor configu- versely, for rotary machines absorptive units are
rations and power ratings. Most machines driven usually satisfactory (further details of silencer
by reciprocating units have rotational speeds rang- design are given in section 7.3.4). Noise radiated
ing from around one hundred revolutions per from the machine casing may be reduced by
minute up to thousands of revolutions per minute. damping, by adding mass, or by stiffening, de-
In industry, where compressors usually feed com- pending on panel characteristics; wrapping the
pressed air to power tools and blow-off jets, fairly inlet and outlet ducts also reduces noise (see sec-
slow speed compressors are used; these produce tion 5.2.2). Vibration isolation of the pump, pos-
predominantly low frequency spectra with major sibly using an inertia block, may be necessary if
contributions at harmonics of the compressor there is substantial structure-borne propagation
rotational speed. Larger units produce a faster rise through the mountings (see figure 5.27, above).
in the work capacity than in acoustic output, and Mechanical noise is usually caused by worn
the use of larger but slower machines can provide bearings, and in this kind of application it may be
quieter running. The noise radiated from the best to replace ball or roller bearings with sleeve-
machine casing becomes important only when the type fittings. Suitable bearing lubrication is for
inlet and outlet ducts are effectively silenced, but maximum benefit; the general order of increasing
the level is governed chiefly by the piston speed noise level is oil-flooded rotary pumps, oil-lubri-
and not the mechanical power produced by the cated bearings, and carbon composite bearings
machine. 39 Rotary positive displacement type (oil-free).
blowers suffer from acoustic problems similar to The major source of aerodynamic noise in
those of their reciprocating cousins, although compressor units is from rotor-stator interaction.

107
 Compressors usually function by drawing air past influence of the obstacle. This interaction does not
inlet guide vanes which act to steady the flow, and produce any new tones, but simply enhances the
then compressing the flow drawn past the driven noise level output already present at the funda-
rotor against a series of static vanes or stators, mental blade pass frequency and associated har-
which have a fixed and regular geometric relation- monics. This is illustrated in figure 7.20, which
ship with respect to the rotor configuration. A shows the generation of a one-lobed spinning pres-
schematic of a single stage unit is shown in figure sure pattern because of the interaction of a four-
7.19; multistage units have additional rows of bladed rotor with three flow guides a, b, and c.
rotors and stators and can thereby achieve greater For each one-quarter turn of the rotor, a complete
power efficiency and generate high pressures. cycle of the interaction pressure pattern occurs;
As a free rotor rotates there is no change in therefore, for a complete rotor cycle four interac-
the blade pressure loading, but with the presence tion cycles occur, thereby enhancing the funda-
of a nearby regular obstruction such as stator mental blade pass frequency (at n = 1) and har-
vanes, pressure fluctuations are experienced be- monics thereof. Multilobed spinning pressure pat-
cause of the distorted flow. These are not abrupt terns excite higher harmonics of the blade pass
as in a "slap"; rather, they arise from velocity frequency.
gradients, which change the angle of attack at the The parameters of interest are the number of
blades and cause a corresponding flux in the lift lobes, m, and phase rotational speed of the inter-
forces. The magnitude of such fluctuations de- action patterns, Nm, calculated as 41
pends on the size, proximity, and aerodynamic
m = nF + kH

where
Drive motor Stator --_ F Rotor
n is the integer number
F is the number of fan blades
H is the number of stator vanes
k is the positive or negative integer values,
i.e. +_1, _+2....

_ _4,1 Iolet
and

-- 71g / nFN
m

Figure 7.19 - Schematic diagram of single stage where N is the rotor rotational speed, rpm.
compressor.
The particular blade pass frequency harmonic
excited by a particular lobe pattern m is given by
Pressure lobe due to
interaction pattern

Rotor_ _P" \
Stator.-_
C

(1) (2) (3) (4)

Figure 7.20 - Generation of an interaction pressure pattern (reference 40). (From Lewis H. Bell, Funda-
mentals of Industrial Noise Control, 1973. Used by permission of Harmony Publications.)

108
 nFN several rotor chord lengths. Figure 7.21 shows the
fn- 60 Hz effect of increasing rotor-to-stator spacings for
several experimental blade configurations. Gen-
where n is the integer number 1, 2, 3 .... erally, this axial separation should be of the order
of at least 1.5 rotor chord widths, 41 to make a
Example." In figure 7.20, the number of fan worthwhile reduction in the strength of the inter-
blades is 4, and the number of stator vanes is 3, action. Another study 43 indicates that for com-
thus m is calculated as parable geometries and spacings, inlet guide
vane-rotor interactions were much higher in level
m = n4 + k3 than rotor-downstream-stator interactions for
single stage axial flow compressors.
Therefore, for the fundamental (n = 1),

when k = -1 m = 4 - 3 = 1
k = +1 m = 4 + 3 = 7
-1

The phase rotational speed of the interaction -2

pattern is given by
-3
m m ,.-. Scatter limits on

4N experimental data
_, -4
for m = 1 Nm - ! (n = 1) -5 - \
\
i.e., 4 times shaft speed as illustrated in figure 7.21; 6 \
o -7 - \
4N "o

g -8- \
form = 7 Nm = 7 (n = 1) 8 \
\
i.e., 4/7 times shaft speed or 1/7 phase speed of
-I0 -- \
one-lobed pattern because seven times as many t-
tJ
lobes generate the same frequency. --II--

In designing for reduced interaction noise, it --12j J t _t J J \ J

is important to avoid combinations of vane and 1.5 2 3 4 5 10

blade configurations which produce a phase rota- Spacing/rotor chord, XIC.

tional speed for the pressure interaction pattern

Figure 7.21 - Changes in rotor-stator interaction
which is many times the original drive shaft speed.
sound pressure levels due to rotor-stator spacing
This is because as the speed of the pressure fluc-
(adapted from reference 42).
tuations approaches that of ambient sound propa-
gation, i.e., Mach 1, there is a rapid increase to
peak noise level and maximum acoustic power is
radiated. Axial and centrifugal fans as well as 7.3 DUCT-BORNE NOISE
compressors display these effects, and one of the
best ways of mitigating them is to optimize the Duct work is present in most industrial
number of blades and vanes so that a high lobe plants, either as part of the ventilation and heating
number m is generated, thus reducing the radia- system, or as an integral part of the production
tion efficiency of the interaction because of the process, such as structures around inlet chutes and
resultant lower phase speed, Nm.24,40,41 Small conveyors carrying components to and from the
changes in the relative twist angle between rotor processing area. This section examines most of the
and stator blades do not apparently affect the aeroacoustics work performed on acoustic propa-
noise level, although large changes do have an gation in ducts in the presence of flow; the in-
influence at close rotor-stator spacings. There has fluence of various kinds of duct liner and means
also been some indication that increasing the of optimizing these; and, finally, the effect of duct
stator chord length is not necessarily a useful in-line elements such as bends, open-ends, internal
expedient unless the rotor-to-stator spacing is at vanes, and mufflers.

109
 7.3.1 General Propagation This relationship indicates that maximizing
the ratio of perimeter length to cross-sectional
Unlike most other enclosures in acoustics, area gives the best attenuation performance. 44
ducts do not generally contain diffuse sound fields Circular and square cross sections give the same
since most of the acoustic energy typically propa- order of attenuation except at low frequencies,
gates parallel to the duct axis. The most prominent where the extra stiffness of the circular duct
sound waves are plane waves, which corkscrew reduces the degree of acoustic energy absorption
around the duct cross section with a general direc- and hence attenuation from resonance of the duct
tion of propagation along the duct axis. For the walls, compared to the square duct. By using rec-
most part, spinning modes are only important tangular ducts and making one of the lateral
when a driving fan or compressor, located within dimensions much larger than the other, for exam-
the duct cross section, is operated at very high ple, height greater than width, the potential
speeds such that the blade tips exceed the speed of attenuation is greatly increased. The advantages
sound, and when the duct section under considera- of this configuration, however, must be balanced
tion is located fairly close to the generating source. against the disadvantages of the higher pressure
In an acoustically lined duct, energy is con- drop involved with rectangular sections. Figure
tinually being extracted from the edges of plane 7.23 shows the typical benefit from varying the
waves, especially at the mid to lower frequencies. relative duct wall separation in this way.
Because the effective speed of sound in porous
materials is less at lower frequencies (see figure
5.2a, above), the effect is rather to bend the waves __L ..... II11
as shown in figure 7.22. The limitation on the ex-
tent of this attenuation at low frequencies is deter-
4O
mined by the available thickness of material. At
high frequencies, where the wavelength is of the "= 30
same order or less than the duct width, most of the e- 50
energy propagates down the center of the duct and 2o
so the lining has a diminished attenuation per-
13..

formance. 44 The amount of energy absorbed per

I I I I J J
unit length of a lined duct is given approximately 0 5 10 15 20 25 30
by h (cm)

Lp Figure 7.23 - Attenuation in ducts by varying the

Attenuation, dB/m = o_L.4 × _-
height (reference 46).
where

Lp is the perimeter length, m

S is the cross-sectional area, m 2 A variety of analytical procedures describe
o_ is the statistical absorption coefficient of sound in ducts, including modal theory, purely
lining. numerical techniques, and geometric or ray trac-
ing. Modal theory normally requires foreknowl-
Highly directional edge of the modal structure and considerable
radiation from duct computation, and so is largely precluded from
termination for
practical noise control use. Numerical techniques
suffer from providing no direct notion of the
_._gh frequencies
Duct wall Acoustic lining mechanisms of propagation and attenuation. Ray
tracing, however, although only valid at high fre-
quencies, does at least allow the relative effects of
// / /Omnidirectional refraction, diffraction, absorption, and reflection
High frequency Low frequency _/ radiation from to be considered.
/
propagation plane waves bent _ duct termination
down duct as energy absorbed for low
7.3.2 Internal Duct Linings
center by lining frequencies

Figure 7.22 - Acoustic energy propagation in a As has already been suggested in section
lined duct. 5.1.2.3, much attention has been given to lining

110
configurationsfor therelativelyextremeenviron- Figure 7.24 illustrates the effect of filling
mentof aircraftjet engines.Mostof thisresearch 2.54-cm (l.0-in.) cavities with compressed poly-
hasfocusedon broadbandresonatorabsorbers, urethane foam with varying overall flow resist-
andparameters suchasacoustic treatmentlength ance. The relative increase in attenuation with the
anddepth,core-celldimensions, acoustic
imped- number of duct wall surfaces treated is shown in
ance,channelwidth, numberof wallslined,air figure 7.25. It is clear that this attenuation de-
flowvelocity,andnumberof layersin thelinings, pends not only on the additional area covered, but
haveallbeenexamined, n5Amongthemostimpor- also on the separation between these surfaces, as
tantfindingsarethatboththemaximumattenua- indicated by the poor improvement in perform-
tion andthe frequencyat whichthis occursare ance obtained by treating walls C and D, which
stronglydependent ontheductwallseparation46; are spaced further apart than walls A and B.
splitterelements
aremosteffectivewithanimper- Another study examining the behavior of single
viouscentralcore46;and low flow resistance orifice Helmholtz resonators in a grazing flow
absorbentmaterialplacedin cavitiesbehinda environment 48 showed that at high flow speeds,
porousfibermetalsurfacegivesthe bestoverall such as in many ducts, acoustic resistance is nearly
absorption.47 linearly proportional to grazing flow speed and

1.0

0.8

(a) E
0.6

E} 25-rayls
oo
0.4 /_ 40-rayts

60-rayls
t-,
<
0.2

o
1ooo 1250 1600 2000 2500 3150 4000 5000 6300 8O0O
Frequency(Hz)

1.0

0.8

(b)

._o 0.6

E3 25-rayls
c
o
.= 0.4 40-rayls
Q.

(_ 60-rayls
e_
<
0.2

0
1000 1250 1600 2000 2500 3150 4000 5000 6300 8000

Frequency (Hz)

Figure 7.24 - Normal incidence absorption coefficients of fibermetal surfaces with air- and foam-filled
cavities for various resistances (reference 4 7). (a) 2.54-cm cavity, foam filled. (b) 2.54-cm cavity, air filled.

111
 A,
optimum resistance, so that the peak attenuation
"0 c D may be calculated by
4O
B
Z R
/,.._._..._(A + B + C + D)
for R < Rmp
o 30 M = 0.2 /f_ _......._(A + B + C, AdBpa = Rm p × AdBmp a
X

Rmp
¢- AdBpa _- _ × AdBmp a for R > Rmp
.o_

<
,o
0 i I I
where
500 1000 2000 5000 10000 R is the specific acoustic resistance

Frequency(Hz) Rmp is the value of optimum specific

acoustic resistance at frequency of
Figure 7.25 - Effect of lining various combina- peak attenuation
tions of duct walls (reference 46). Attenuation
AdBpa is the peak value of sound power
values taken from duct with net flow rate of Mach attenuation, dB
number, M = 0.2.
AdBmpa is the maximum possible attenuation
at frequency for which optimum
impedance is specified.
almost independent of incident sound pressure. In
fact, as the ratio of plate thickness to orifice diam- Thus for liner resistances greater than the
eter is increased, both acoustic resistance and reac- optimum, peak attenuation is inversely propor-
tance became increasingly less sensitive to grazing tional to the resistance, and below the optimum
flow speeds. resistance peak attenuation is directly propor-
Where spinning modes are present to any tional to resistance.
large extent, then circular ducts and noise sup- The attenuation bandwidth is derived from
pressors become very sensitive to the particular
modes involved, whereas annular ducts and silenc-
fl/fp = fl/fp)Opt; fz/fp = fz/fp)Opt
ers are fairly insensitive. 49 With the latter it is
possible to provide reasonably effective designs for R < Rmp
for attenuating spinning modes using regular mode
theory in conjunction with experimental flow duct fl/fp _=_Rm_ x -)opt; f2/fp =-
tests; this is not the case for circular ducts, where f_ R
R Rmp
the best approach is probably to rely mainly on
for R > Rmp
empiricism. A technique for calculating the opti-
mum wall impedance for spinning mode attenua- where
tion in soft wall ducts using as a fundamental
is the frequency at which peak attenua-
parameter the angle of propagation within the tion occurs, Hz
duct with respect to the angle of incidence at the
fl 'f2 are the lower and upper frequencies for
wall, is considered easier to use than conventional
one-half peak attenuation at fp, i.e.,
modal analysis techniques. 5°
AdB = 1½AdBpa, Hz
A generalized approximate equation for duct )opt is the frequency ratio when R = Rmp.
lining sound attenuation, intended for the use of
initial acoustic liner design, has been developed. 51 Thus attenuation bandwidth remains constant
The specification of only two parameters for a
for linear acoustic resistances, 0, below the opti-
particular acoustic mode, optimum impedance
mum, 0mp , and increases with acoustic resistance
and maximum possible attenuation at this imped-
when this is above the optimum. By the use of the
ance, AdBmp (calculated from reference 52), following relationships, several of the more com-
allows the determination of sound attenuation to
plex parameters above can be found directly:
be made for that acoustic mode at any selected
impedance. The peak and bandwidth of sound
1 + 0.564/2.
attenuation are represented by simple ratio func- fl/fp)opt =
1 + 1 " 3/2 '
tions of the actual wall acoustic resistance and the .OVp

112
 1 + 4.21@ 7.3.3 Effect of Lined Bends, Duct Terminations,
f 2/ f p)Opt = 2 and Splitter Elements
1 + 3.05Vp

where This section and section 7.3.4 deal with in-line

attenuation elements for which only general infor-
Upis the f_/c at the peak attenuation
frequency mation is provided. For more specific information
D is the duct diameter, m on attenuation and insertion loss calculations, and
¢ is the speed of sound, m/s. specifications for individual units, the user is
directed to reference 59.
These formulas may apply to more than just Bends attenuate primarily because they reflect
circular ducts, and although the equations are propagating sound waves back towards the source;
approximate, the unknown complete modal input this becomes important when the duct width is
and errors in analytical values of the wall imped- greater than the wavelength concerned. Therefore,
ance render superfluous any more exact analysis. low frequency sound in general is little affected by
Recently, a computer program for the design bends. A secondary phenomenon, which affects
of annular duct liners using perforated plate over frequencies which are diffracted or reflected
a honeycomb structure has been developed. 53 This around the corner, is a scattering effect which con-
program is based on experimental data so that the verts energy hitherto propagating as plane waves
typical effects of boundary layer turbulence, duct into higher order transverse acoustic modes which
terminations, and so on may be included. Prob- can be attenuated much more readily. For this
ably the greatest value of this program is in gener- reason, lining bends for a short distance down-
ating candidate designs which, for example, indi- stream can have a pronounced effect on the over-
cate the propitious effect of changing the number all noise reduction achieved. Figure 7.26 shows the
of splitter rings versus overall length of treatment. typical effect of lining a 90 ° bend both upstream
Investigation to extend the broadband per- and downstream for various duct widths. A reso-
formance of resonator configurations has exam- nance condition at the frequency where the wave-
ined multisection liners arranged either axially or length is twice the duct width for unlined ducts is
circumferentially with respect to each other. Apart responsible for the peak attenuation value associ-
from providing a composite absorbing structure ated with each curve; generally, as the internal
with several different peak attenuation frequen- width is reduced, the low frequency attenuation
cies, the technique appears to enhance the overall performance is also reduced.
attenuation performance by reflecting certain The use of sharp bends with associated ab-
modes at the intraliner boundaries and redistrib- sorption lining is limited by the amount of pres-
uting others into high-order modes which are sure drop permissible across the unit. Long radius
more rapidly attenuated. 54-56 Correct selection of bends or mitred bends (90 ° ) with long internal
the requisite liner impedances is imperative, and turning vanes are excellent for minimizing pres-
the approach seems to work best at intermediate sure losses but unfortunately do not provide much
frequency ranges. NASA has patented a device attenuation. In a wind tunnel noise study 60 with
based in part on this principle which has already high velocity flow rates, considerable attention
found commercial application 57 (see section was given to the influence of turning vanes. It was
7.3.4). Another approach has used two layers of noted that the short-chord vanes turned sound
honeycomb cells on top of each other, but this waves with wavelengths less than the vane chords
does not appear to broaden attenuation perform- through an angle dependent on the incident angle
ance as much as might be anticipated from a cur- at the vane row, as shown in figure 7.27. Figure
sory examination. 45 7.27 also shows the extent and configuration of
Bulk absorbers have recently undergone con- wall lining treatment used to absorb sound energy
siderable experimental and analytical reappraisal at the bend; at this particular corner, the bend
for deployment in aircraft nacelles. The indica- lining was rendered fairly ineffective at high fre-
tions are that conventional multisegmenting tech- quencies because the vanes turned the propagating
niques give no significant improvement, although sound past the lining. One way around this prob-
impedance considerations are as important for lem, other than increasing the lining length down-
optimizing suppression at discrete frequencies as stream, is to acoustically treat the vanes them-
they are in resonator or reactive systems. 58 selves, although this is usually an expensive

113
 2O

15

Z
m
i
o

x 10
- o00
:m
15 cm d = Open internal
width of bend
P,l _d = 120cm
d 60-120 cm Lined bend
g d 30-60 cm
_.,_.m,.,, Unlined bend

<
_ 7 d 15-30 cm

,."/,'" I, -.2":,.=-.......
63 125 250 500 1000 2000 4000 8000

Frequency (Hz)

Figure 7.26 - Typical effect of acoustically lining 90 ° bends (reference 59).

I I i I

Low frequenc es --
_
High f r-eCluenc-ie-s--
.... ._%_
-'_"_
- -_t'"'_'_ /
76-m m foam 24,,
22 ._I_.." L --

_ _' t ! l ..--- 79% perforated plate 16 \ I! ''r_ ml--

', _

I *-- 508mm
,q
T__ 406-mm spacing between _ 8
cavity baffles
High frequency I
sound reflected
past downstream
lining
0 .02 .04 .08 _ 2 , 6"8 '-_" 2 _ 6. -
Figure 7.27- Installation of corner vanes at an .1 1 10

acoustically lined bend (adapted from reference 60). Area of opening (m 2)

Figure 7.28 - Attenuation due to reflection at a

duct outlet. (From Ian Sharland, Woods Practical
expedient. The cavity baffles spaced at 406-mm Guide to Noise Control, 1971. Used by permission
intervals prevent sound from propagating along of Woods Acoustics.)
the sides behind the foam.
At duct terminations there is a large change
of acoustic impedance, because elements of air shows the degree of attenuation which occurs with
through which sound energy propagates move various frequencies at a duct opening flush with
against a much larger mass of air outside the duct. the surrounding surface. Low frequencies do not
The result is an end reflection dependent on the propagate well into an open space, although the
frequency and duct area concerned. Figure 7.28 energy that does so radiates fairly uniformly from

114
the duct opening compared to the more directional the requirements for a minimum pressure loss
high frequencies (see also figure 7.23). The effect achieved by increasing the height and total width
of grilles, diffusers, louvres, and so on at the ter- of the complete splitter unit. In addition, the lead-
mination must be separately computed. ing and trailing edges of the splitter modules
A much more sophisticated and exact predic- should be aerodynamically designed to reduce flow
tion technique has been developed for the radia- resistance. The manufacturer's data should specify
tion of sound from a rotating blade source within the acoustic performance for various rates of air
circular hard-wall ducts. 61 This may be of value in flow: in general, attenuation decreases as the flow
appraising the validity of other more approximate rate in the direction of sound propagation in-
approaches. creases. Annular modules in cylindrical ducts
Attenuation predictions at acoustic branches operate on the same principles. Illustrations of
which occur at angles relatively oblique to the both cylindrical and annular silencers are shown in
main flow direction are based on the approxima- figure 7.29, with attenuation graphs showing the
tion that the acoustic energy is divided according effect of different lining configurations. An exam-
to the relative cross-sectional areas: ple of a large scale annular silencer for use in
attenuating a 126 dB(A) level at a wind tunnel
Branch attenuation = ejector by some 26 dB(A) is shown in figure 7.30.
Rather than using a fibrous backing for the
10 log main duct acoustic baffles in the splitter units, it may be
[ branch duct area
area ; dB
more expedient to use a resonator-type design with
relatively large air spaces filled with absorber
7.3.4 Mufflers and Duct Silencers
behind a perforated plate. In a comprehensive
series of tests, 62 there were indications that a saw-
Mufflers and duct silencers use absorption or
tooth type design with a 0.3 percent perforate
the reflection and expansion of sound waves due
partition is particularly useful at improving low
to sharp impedance changes at cross-sectional area
frequency performance. Figure 7.31 shows the
changes and turns. The former technique relies on
configurations and performance data obtained for
dissipative mechanisms and the latter on reactive
several configurations. The use of a central parti-
mechanisms; in many silencer units both princi- tion inclined at an angle to the longitudinal axis of
ples operate to some extent. Mufflers or silencers the baffle is intended to reduce cavity whistling
are used to reduce noise transmission along ducts
caused by high speed air flows at grazing inci-
or pipework, and the noise may be flow-generated dence, by detuning the cavity because of its vary-
or may emanate from a machine source. In choos-
ing depth. Transverse partitions across the baffles
ing an appropriate device, particular regard should enhance absorption at the frequencies whose half-
be given to the allowable pressure drop, maximum
wavelength corresponds to the distance between
length needed, gas or fluid flow rate, and whether them. Reference 62 also includes a discussion of
acoustic propagation is downstream or upstream various techniques for reducing the self-generating
with respect to the general flow. noise of mufflers. Some of these are illustrated in
7.3.4.1 Resistive units figure 7.32. Resonant tones produced by single
orifice Helmholtz resonators because of the shear-
By splitting the duct width into smaller sec-
ing action of the grazing flow layer have been
tions with reduced width, low frequency perform-
shown to occur at a grazing flow speed, voo, given
ance may be significantly improved. The partitions
by 47
for this are usually of perforated plate backed by
an absorbent material such as fiber glass or foam. vo_ = fres D/0.26 m/s
It is usually most convenient to use proprietary
splitting units with subsection widths down to as where
low as 50 mm. These units can often be most satis- fres is the classical Helmholtz resonant fre-
factorily deployed at bends, where the total at- quency (see section 5.1.2.3), Hz
tenuation is given by the total equivalent straight D is the orifice diameter, m.
length of splitter plus a bend attenuation factor.
Attenuation may generally be increased by Other techniques for improving the perform-
reducing the airway width and increasing the total ance of lined baffles in splitter units include block-
splitter length, but these must be balanced against ing the direct line of sight through the unit and the

115
 mm
50

_ 200 mm
v
r
.£
4O

30
/
(a)

<
-I

/
20

0
63 125 250 500 1000 2000 4000 8000

Frequency (Hz)

3O
760 mm-

12 Omm
480 mm, ,,-" 300 mm

.-,,.!
20
(b)

_P 300mm

l /-- 480 mm
?= _y"_ ___ -_:

I / 760ram -''_

12_0 mm

63 125 250 500 1000 2.300 4000 8000

Frequency (Hz)

Figure 7.29 - Typical attenuation characteristics of rectangular and circular splitter modules. (a) Rec-
tangular module. (b) Circular modules. (Adapted by Ian Sharland, Woods Practical Guide to Noise Con-
trol, 1971. Used by permission of Woods Acoustics.)

116
 tion; their performance is enhanced by incorporat-
ing an absorbent-lined partial barrier which serves
to block the direct line of sight across the cham-
ber. The most important design characteristics are
the absorption of the lining and the distance be-
tween the inlet and outlet ducts. The sound fields
associated with plenums may be regarded as simi-
lar to those inside large acoustical enclosures.
Figure 7.33 shows a single chamber plenum instal-
lation, and the insertion loss of this after taking
account of reflection at the inlet duct is given by 29

F COS 0 R_pl
InL = 10 log Spk_ +
Figure 7.30 - Photograph of large annular duct
silencer (by courtesy of NASA Langley Research
Center).
where
Rcp is the room constant of the plenum,
use of different liners in each separate airway. 63 equal to Sp_p m2
The latter technique can introduce a relative phase (i - _p) '
lag between propagating components in each air- d is the directdistancebetwcen inletand
way with resultant reflection and interference outletopenings, m
effects at low frequencies. 0 is the slant angle between longitudinal
Acoustically lined plenum chambers repre- axes of inletand outletducts
sent a very efficient way of achieving noise reduc- Sp is the outlet cross-sectional area, m 2.

30

Baseline center partition

25 1

Ramp
z

Saw tooth
: -"
J 4

.o /,,.
Saw tooth with 0.3% 20 / l/] -
perforated partition °
.__
o 15
rll

z II
1£
125 250 500 1000 2000 4000

Frequency (Hz)

Figure 7.31 - Attenuation performance of various staggered central partitions in acoustic baffles (reference
60).

117
 27rL nTr
Flow [_I
Flow X 2
_mll_Absorbing layer
where n = 1, 3, 5 ....

No perforations near or
Flow _ on tail to trip flow so that peak attenuation occurs at L = nk/4.
These silencers are therefore often referred to
_ Slanted partition
as quarter-wave mufflers, and several of them can
Flow
be placed in series to provide a broader range of
Flow attenuation. Typically their performance is best at
-_d _-
Recessed holes low frequencies; a variation on this principle, the
Helmholtz resonator, may be used when it is more
Diamond shaped holes
convenient to have the muffler element as a side-
FlowJll_ ---_'
branch to prevent flow obstruction (see section
. d ,4-- Elongated holes to minimize 5.1.2.3).
Elevated hole leading edges edge length normal to flow There are many prediction techniques for
muffler design, several of which have been com-
Figure 7.32 - Candidate perforated plate designs puterized. One in particular 64 improves on stand-
for minimizing the generation of self-noise ard transmission line theory by accounting for the
(reference 60). effect of mean gas exhaust flow on the overall
attenuation properties of a three-chamber muffler.
This NASA program has been used to design a
lined baffle unit capable of providing up to 11 dB of extra
Inlet -- J 1 attenuation over standard mufflers in an opera-
tional helicopter, without any deterioration in
-_ _ __- Internal acoustica/_
AllYea, S engine performance from back pressure. The pro-
gram is based on empirical data for exhaust flows
d. -_ ._=t__Outle.t in the Mach number range of 0.05 to 0.15 and
t r-- expansion ratios less than 0.6.
'_ Internal surface: area, Sp Another example of the relevance of aero-
average absorption acoustics technology to general noise control
coefficient, c_
problems is the recent development to prototype
F@ure 7.33 - Acoustically lined plenum chamber. stage by a commercial organization of a multiple-
segmented acoustic liner in a standard truck
muffler geometry (see figure 7.34). The device has
7.3.4.2 Reactive unils a uniform inner diameter such that the direct
exhaust air flow is not blocked, and it utilizes a
The most simple kind of reactive silencer is unique combination of absorbing elements work-
designed by changing the cross-sectional area of a ing in conjunction with reflecting elements in order
short length of duct. The insertion loss is given by to provide maximum attenuation. 56 It achieves this
by forming an acoustic trap where the reflective
elements on the ends direct sound energy into the
lnL = 10log 1 + -_ sin -X dB
central sound dissipating units. This particular
where application uses three cylindrically stacked seg-
M is the $2/S 1, cross-sectional area ratio ments in the center, each of different porosity and
L is the muffler length, m of varying depth. It provides a noise attenuation
k is the wavelength of frequency of interest, of peak levels of twice that possible with more
m. conventional designs. 65
An example of a serious aerodynamic noise
Since the attenuation is obviously greatest problem is included here and the various ap-
when sin (2rL/k) equals 1, and this occurs at r/2, proaches taken toward its solution highlighted. 66
3rr/2, 5r/2 .... the length of silencer L corre- Figure 7.35 shows a wind tunnel configura-
sponding to maximum attenuation is given by tion complete with installed acoustical treatment.

118
 Figure 7.34 - Truck muffler utilizing multisegmented liner (by courtesy of Donaldson Company).

Exhaust 2. Incorporation of reactive resonator, tuned

Inline muffler _ 7 Inlet
to 250 Hz, for attenuating high frequency
Inline resonator.. _ _ _
sound.
(250 Hz) . _-{, _"
Side-branch -_.--, ,<J'_._-J/,. . ".\ //_, 3. Design of in-line dissipative muffler and
resonator _ _/)" Interior 5 ./_//" _ Test.
112.z, section the incorporation of diffuser cones to in-
Internal turning-,__/"_ _) cone // _. hibit the development of standing waves in
vanes at corner [T "-,L,' ..... // Diffuser the exhaust duct.
Upstream end-_l I //
of exhaust _ _)_'J 4. External wrapping of inlet duct with ab-
sorbent materials on high flow section and
duct _ internal lining in low flow section.
_- Pumps
Reactive techniques were mainly employed be-
Figure 7.35 - Acoustical treatment of wind tunnel cause the high pressure losses associated with dis-
(reference 66). sipative muffler units would have compromised
operating ability.

The total air pump handling rate is 500 m3/min, 7.4 TRANSMISSION AND GEAR NOISE
and the inlet and discharge lines are of 35-mm
steel, up to 1 m in diameter and bolted directly to
Gear noise requires careful analysis to identify
the pumps. The generated spectrum contained
major sources and to rank-order their levels. This
many harmonics with a fundamental frequency of
analysis generally requires narrow-band frequency
12 Hz. The noise and vibration generated were so
analysis and a detailed knowledge of the mecha-
great that the facility could only be run for a few
nism of the gear-train involved. The systematic
minutes. Measured exhaust levels were in excess of
approach adopted should include an appraisal of
120 dB with a maximum peak at around 12 Hz.
basic gear types, major noise generating mecha-
The following procedures were implemented in
nisms, and their spectral character, and a detailed
order to attenuate the noise:
comparison of noise reduction measures. Unfor-
1. Incorporation of two reactive Helmholtz tunately, it is rare for only single gear trains to be
resonators with internal lining, tuned to involved, access may be limited, and several of the
12 Hz. major noise mechanisms may exist simultaneously,

119
 makingit difficult to discovereachindividual width, along the mutual traverse of adjacent tooth
contribution. pitch circles.
Therearea numberof techniques for reduc- The computation of gear clash frequencies is
inggearnoisewithoutsource modification.
These simply made as
are discussedlater in section7.4.4andinclude
usingcompletely differentmeansof transferring nKN
andcontrollingmechanical power,aswellastra- fn- 60 Hz
ditionalpathnoisecontrolprocedures.
where
7.4.1Gear Train Parameters of Importance n is the integer representing the various har-
monics possible
Several basic components and gear wheel K is the number of gear teeth
parameters are shown in figure 7.36. The core of
the wheel is known as the hub and is an integral
Circular
part of the drive shaft. The web is the flange con-
pitch ,
centric with the shaft, and the rim is the section of
the flange which incorporates the entire gear tooth
profile. The pitch circle is one of the most funda-
mental gear characteristics, and this normally runs
through points midway between the top and bot-
tom of each tooth; pitch circles of meshing gears
are always tangential to each other. The circular Ir "l _

pitch is measured along the pitch circle and is the HR_im__ ( Hub I _ ;" - Pitch circle
distance between congruent points on adjacent
teeth, and the backlash is the difference between Figure 7.36 - Schematic diagram of major gear
engaging tooth width and receiving tooth space parts and important parameters.

• " gears

(b)

Straight bevel Spiral bevel Hypoid

Figure 7.37- Common gear configurations. (a) Parallel gears. (b) Nonparallel gears. (Adapted and used by
permission of Machine Design.)

120
 N is the shaft rotational speed, rpm on the relationship of excitation frequen-
fn is the nth harmonic frequency, Hz. cies to gear wheel natural frequencies.

A typical spectrum consists of a dominant

For systems of gears with a central sun gear
fundamental component corresponding to the
surrounded by a series of equally sized and equally
blade pass frequency and decreasing amplitudes
spaced planetary gears, the following relationship
for harmonics of this. Deviations from this pattern
is used4°:
suggest structural resonances of some sort. Other
features include side band tones generated around
nKs(Ns - Nps) Hz gear mesh frequencies because of modulation by
fn= 60
eccentricities arising with rotation of the shaft,
and harmonics at shaft rotational speeds resulting
where from chipped or missing teeth. Several of these
K s is the number of teeth on sun gear phenomena are illustrated in the spectrum shown
N s is the rotational speed of sun gear, rpm in figure 7.38.
Nps is the rotational speed of planetary
system, rpm. Side bandsat fundamental gear mesh
frequency fl due to modulation
Figure 7.37 shows common gear types. with shaft rotation frequency

f2 Amplification of higher
7.4.2 Gear Noise Mechanisms
sFhanf_:ot:tntoa_ i/] /_ gear mesh harmonic, f5'
frequency,NA2-'N | _1 due to structural resonance
There are three aspects of gear quality that Q.

radically affect subsequent noise levels regardless

of the precise mechanism involved. Errors in the
tooth profile shape and roughness of the machined
2
surfaces both tend to create high levels at the upper
frequencies, even at low peripheral speeds, while
Frequency (Hz)
basic pitch error is the third factor of impor-
tance. 67 Wheel quality is therefore a sensitive Figure 7.38 - Narrow band spectral analysis
variable. The most significant sources of gear- showing discrete tones.
meshing noise are as follows4°:

1. Engagement--The teeth of both engaging

7.4.3 Gear Noise Reduction
gears are deflected tangentially, then return
to their former positions after disengage-
ment. Although there are many possibilities for
noise attenuation at source, most procedures
2. Pitch-circle impulse--Deflections of teeth
result in very limited reduction, and an unsyste-
during engagement mean that a continual
matic ad hoc approach produces very poor return
rolling contact can never be ensured, and
for the gear redesign involved.
the friction from the resultant sliding
Generally, helical gears are quieter than iden-
action creates periodic noise.
tical spur gears because the loading is better
3. Air-oil jet noise--This occurs when the distributed and the impact forces reduced. 68
pressure between teeth compresses and Worm gears are even more effective, because the
expells air and oil; shock waves may form continuous sliding action minimizes impulses and
with high-speed gears because of the high corresponding tooth deflections. By the same
velocities involved. token, spiral bevel and hypoid gears can be some
4. Structural vibration--As a result of the 5 dB quieter than their straight-bevel analogs.
three processes outlined above, which all Aside from changing the gear type, better noise
directly create acoustical energy at the gear performance may be achieved by optimizing pitch-
mesh frequency, radiation of acoustical contact ratios, teeth profiles, and backlash. The
energy by the wheels themselves also use of a higher pitch contact ratio, that is, increas-
occurs; the extent of this radiation depends • ing the number of gear teeth, gives quieter running

121
because the impulse loading on each tooth is ing of the belt, lack of sufficient contact resist-
lower. For the same drive speed, however, this ance, and so on. This type of installation is there-
raises the frequency spectrum, which may offset fore generally unsuited to systems which require a
any potential benefit. Ideally, profiles should be high loading or exact speed transformation.
designed so that maximum tooth contact occurs Another well-known technique is to use trac-
during the phase when the point of contact is tion element drives, although this technique has
inside the pitch circle, and always so that a rolling serious constraints. Traction element drives offer
as opposed to a sliding action is favored. Backlash quiet and smooth power transmission, whereas
must be balanced between excess noise produced even with perfectly machined gear teeth there are
with a loose mesh and the mechanical drag and always torsional fluctuations caused by load
wear of a tight mesh. transfers between teeth. Recently, however, with
Other than careful redesign of gear param- the active participation of the inventor of a new
eters, a number of secondary techniques may be traction drive geometry, the NASA Lewis
used. Lubrication of all running parts and mesh- Research Center has developed and demonstrated
ing surfaces is essential and can provide as much considerable advantages in using this technology
noise reduction as the approaches already dis- in a much wider range of situations than was
cussed. By separating surfaces and providing a hitherto possible. 69
mechanical impedance mismatch, the oil reduces The new approach uses a number of planetary
impact forces directly. A high damping factor may rows of stepped rollers, normally two or three, to
be obtained with synthetic gear wheel materials achieve power transfer between concentric sun
such as nylon, subject to load and temperature and ring rollers on the inner and outer surfaces
constraints. These can provide as much as 10-20 respectively (see figure 7.39). By applying input to
dB attenuation and have found widespread appli- either the sun or ring roller, the unit may act as a
cation in machines ranging from spinning looms speed decreaser or increaser, respectively. Previous
to household washing machines. External damp- traction devices had used only one row of plane-
ing can be useful where the rim and web flange is tary rollers, which severely handicapped their
very narrow and therefore susceptible to induced speed and horsepower capacity. Reaction torque is
resonance because of its low stiffness. transferred to or from the ring by a pair of rolling-
Finally, but perhaps most importantly, the element bearings installed in the outer planetary
advantages of total gearbox enclosure should be rollers. This considerably reduces the total number
mentioned. This is a particularly good expedient of drive bearings needed, and, together with the
since it involves no costly analysis and lengthy self-correcting positioning of the rollers because
redesign programs. Adherence to the general prin- of their three symmetric contact points with adja-
ciples of enclosure construction outlined in section cent elements, ensures a quiet and almost vibra-
6.2.1.2 can yield excellent results. It is sufficient to tionless power transmission with no need for espe-
reiterate here that accessibility or visibility must be cially fine tolerances. This characteristic is further
maintained, removable panels should be well enhanced by the presence of a thin elastohydro-
sealed and should have transmission loss values of dynamic lubricant film between all contacting sur-
the same order as the main body of the enclosure, faces, which, along with the elastic compliance of
and resonance should be avoided by stiffening and the rollers, provides effective vibration damping.
damping techniques. An automatic loading mechanism is incorporated
as part of the latest design to ensure that the
7.4.4 Alternative Means of Transmission correct ratio of traction forces to normal contract
forces is maintained under all conditions. This

Conventional toothed-gear drives may be must be carefully optimized to provide minimum

replaced in several ways. Perhaps one of the most slip without unnecessary wear on the roller ele-
common is the use of roller-belt systems where a ments. In general, two planetary rows are suitable
series of parallel roller pairs with correctly sized for speed ratios less than 35 to 1, and three plane-
diameters are interconnected by flexible rubber or tary rows for speed ratios up to 150 to 1.
synthetic belts, which may be ribbed for a better The successful operation of a 15:1 speed ratio
grip. Unfortunately, substantial slip at high speeds test rig has recently been validated at speeds up to
or high torque ratios may occur because of stretch- 7300 rpm, power levels up to 170 horsepower, and

122
 First row of
Sun roller
REFERENCES
planet rollers
Second row of
planet r oilers

1. Lighthill, M. J., On Sound Generated Aero-

dynamically, II., Turbulence as a Source of
Sound, Proc. Roy. Soc. (London) ser. A, vol.
222, no. 1148, Feb. 1954, p. 1.
2. Stone, J. R., Interim Prediction Method for
Jet Noise, NASA TM X-71618, 1975.
3. Maestrello, L., On the Relationship between
Acoustic Energy Density Flux near the Jet
Axis and Far-Field Acoustic Intensity, NASA
TN D-7269, 1973.
4. Sahlin, S., and R. Langhe, Origins of Punch
Press and Air Nozzle Noise, Noise Con. Eng.,
vol. 3, no. 3, November-December 1974, p. 4.
R olling_eiement 5. Lord, H. W., H. A. Evensen, and R. J. Stein,
Ring rotler bearings
Pneumatic Silencer for Exhaust Valves and
• Automatic
Parts Ejectors, Sound and Vib., May 1977,
n loading
mechanism
p. 26.
6. Huff, R. H., A Simple Noise Suppressor De-
Figure 7.39 - Example of nasvytis traction drive sign for Vented High Pressure Gas, NASA
unit (by courtesy of NASA Lewis Research Tech. Brief, summer 1979, p. 278.
Center). 7. Huff, R. H., and D. E. Groesbeck, Cold-
Flow Acoustic Evaluation of a Small-Scale,
Divergent, Lobed Nozzle for Supersonic Jet
a peak efficiency of 95%. The overall speed effi- Noise Suppression, NASA TM X-3210, 1975.
ciency was 98.6%, which compares favorably with 8. Maestrello, L., Initial Results of a Porous
gear-driven systems. Based on other models al- Plug Nozzle for Supersonic Jet Noise Sup-
ready built, it is considered that power-to-weight pression, NASA TM 78802, 1978.
ratios of 4.5 horsepower/Ib for continuous opera- 9. Maestrello, L., An Experimental Study on
tion are achievable, and units of over 500 horse- Porous Plug Jet Noise Suppressor, AIAA
power capacity can be easily designed. It should Paper 79-0673, March 1979.
also be noted that the test models used a consider- 10. Gutierrez, O. A., and J. R. Stone, Develop-
able amount of knowledge gained from NASA re- ments in Aircraft Jet Noise Technology. Air-
search programs in traction fluids, rolling element craft Safety and Operating Problems, NASA
fatigue, bearing steels, and roller kinematics. SP-416, 1976, p. 497.
These devices have particular application in 11. Schwartz, I. R., Minimization of Jet and Core
such areas as the machine tool industry, where the Noise by Rotation of Flow, NASA Tech Brief
high speed rates would result in much better qual- B75-10131, 1975.
ity, and in high speed drive systems in gas turbines 12. Reethoff, G., Control Valve and Regulator
or turbomachinery generally. 7° Currently, their Noise Generation, Propagation, and Reduc-
utility in replacing noisy helicopter main rotor tion, Noise Con. Eng., vol. 9, no. 2, Septem-
transmissions and aircraft drive systems is being ber-October 1977, p. 74.
investigated. Although no formal acoustic or 13. Cheremisinoff, P. N., and P. P. Cheremisi-
vibration tests have yet been conducted on the noff, Industrial Noise Control Handbook, in
revised traction drive units, indications are that Hydrodynamic Control of Valve Noise,
sound pressure level reductions of at least 10 dB Chapter 14, Ann Arbor Science Publishers,
compared to conventional drives are possible. 1977.
Furthermore, the approximately white noise spec- 14. Industrial Noise Services, Inc., Industrial
trum which they generate is easier to control than Noise Control Manual (prepared for NIOSH),
the spectrum associated with toothed-drive available from NTIS as DHEW Pub. No.
systems. (NIOSH) 75-183, 1975.

123
 15.Regetz,
J. D., Jr., An Experimental Determi- 28. Mellin, R. C., Noise and Performance of
nation of the Dynamic Response of a Long Automotive Cooling Fans, SAE Tech Paper
Hydraulic Line, NASA TN D-576, 1960. 800031, Feb. 1980.
16. Blade, R. J., W. Lewis, and J. H. Goody- 29. Beranek, Leo L., ed., Noise and Vibration
koontz, Study of a Sinusoidally Perturbed Control, McGraw-Hill Book Co., 1971.
Flow in a Line Including a 90 ° Elbow with the 30. Graham, J. B., Fans and Blowers, in Hand-
Flexible Supports, NASA TN D-1216, 1962. book of Noise Control, Cyril M. Harris, ed.,
17. Lewis, W., R. J. Blade, and R. G. Dorsch, second ed., Chapter 27, McGraw-Hill Book
Study of the Effect of a Closed-End Side Co., 1979.
Branch on Sinusoidally Perturbed Flow of 31. Lakshminarayana, B., Influence of Turbu-
Liquid in a Line, NASA TN D-1876, 1963. lence on Fan Noise, in Workshop on Ventila-
18. Blade, R. J., and C. M. Holland, Attenuation tion System Cooling Fan Noise, Nov. 1975,
of Sinusoidal Perturbations in Nonviscous M. Sevik and J. Pierpoint, eds., Naval Sea
Liquid Flowing in a Long Line with Distrib- Systems Command, 1976, p. 61.
uted Leakage, NASA TN D-3563, 1966. 32. Heidmann, M. F., Interim Prediction Method
19. Stephenson, J., and H. A. Hassan, The Spec- for Fan and Compressor Source Noise, NASA
trum of Combustion-Generated Noise, J. TM X-71763, 1975.
Sound Vib., vol. 53, no. 2, 1977, p. 283. 33. Soderman, P. T., Leading Edge Serrations
20. Shivashankara, B. N., W. C. Strahle, and which Reduce the Noise of Low-Speed Rotors,
L. C. Cho, Combustion Generated Noise in NASA TN D-7371, 1973.
Turbopropulsion Systems, Second Inter- 34. Hersh, A. S., and R. E. Hayden, Aerody-
agency Symposium on University Research in namic Sound Radiation from Lifting Surfaces
Transportation Noise, Vol. II, June 1974, with and without Leading-Edge Serrations,
p. 714. NASA CR-114370, 1971.
21. Putnam, A. A., Combustion Noise in Indus- 35. Hayden, R. E., Some Advances in the Design
trial Burners, Noise Con. Eng., vol. 7, no. 1, Techniques for Low Noise Operation of Pro-
July-August 1976, p. 24. pellers and Fans, Noise-Con. 77, 1977, p. 381.
22. Plett, E. G., M. D. Leshner, and M. Summer- 36. Hayden, R. E., D. B. Bliss, B. S. Murray, et
field, Combustion Geometry and Combustion al., Analysis and Design of a High Tip Speed,
Roughness Relation to Noise Generation, Sec- Low Source Noise Aircraft Fan Incorporating
ond Interagency Symposium on University Swept Leading Edge Rotor and Stator Blades,
Research in Transportation Noise, Vol. II, NASA CR-135092, 1977.
June 1974, p. 723. 37. Moiseev, N., B. Lakshminarayana, and D. E.
23. Plett, E. G., M. D. Leshner, and M. Summer- Thompson, Noise due to Interaction of Boun-
field, Effect of Flame Distribution on Noise dary-Layer Turbulence with a Compressor
Generation, Third Interagency Symposium on Rotor, J. Aircraft, voi. 15, no. 1, 1978, p. 53.
University Research in Transportation Noise,
38. Goldstein, M. E., Highlights 1979, AIAA
November 1975, p. 207. Journal, vol. 17, no. 12, December 1979.
24. Mugridge, B. D., Noise Characteristics of
39. Webb, J. D., ed., Noise Control in Industry,
Axial and Centrifugal Fans as used in Indus-
Halstead Press, 1976.
try--A Review, Shock and Vibration Digest,
Feb. 1979, p. 17. 40. Bell, L. H., Fundamentals oflnclustrialNoise
25. Longhouse, R. E., Vortex Shedding Noise of Control, Second ed., Harmony Publications,
1973.
Low Tip Speed, Axial Flow Fans, J. Sound
Vib., vol. 53, no. 1, 1977, p. 25. 41. Tyler, J. M., and T. G. Safrin, Axial Flow
26. Mellin, R. C., Selection of Minimum Noise Compressor Noise Studies, SAE Trans. vol.
Fans for a Given Pumping Requirement, 70, 1962, pp. 309-332.
Noise Con. Eng., vol. 4, no. 1, January- 42. Lumsdaine, E., and J. Chang, Influence of
February 1975, p. 35. Blade Characteristics on Axial Flow Com-
27. Graham, J. B., How to Estimate Fan Noise, pressor Noise, AIAA Paper 76-570, July
Sound and Vibration, May 1972, p. 24. 1976.

124
43. Copeland,W., J. L. Crigler, and A. C. 56. Zorumski, W. E., Acoustic Theory of
Dibble, Jr., Contributionof Downstream Axisymetric Multisectioned Ducts, NASA TR
Stator to the Interaction Noise of a Single- R-419, 1974.
Stage Axial-Flow Compressor, NASA TN 57. Zorumski, W. E., Noise Suppressor, U.S.
D-3892, 1967. Patent No. 3 830 335, August 1974.
44. Posey, J. W., High Frequency Sound Attenua- 58. Kraft, R. E., R. E. Motsinger, W. H.
tion in Short-Flow Ducts, NASA TM 78708, Gauden, and J. F. Link, Analysis Design, and
1978. Test of Acoustic Treatment in a Laboratory
45. Atvars, J., and R. A. Mangiarotty, Para- Inlet Duct, NASA CR-3161, July 1979.
metric Studies of the Acoustic Behavior of 59. ASHRAE Handbook and Product Directory,
Duct-Lining Materials, J. Acoust. Soc. Systems Volume, Chapter 35, Amer. Soc.
Amer., vol. 48, no. 3, 1970, p. 815. Heat, Refrig., and Air-Con. Eng., New York,
46. Mangiarotty, R. A., Acoustic-Lining Con- 1976.
cepts and Materials for Engine Ducts, J. 60. Soderman, P. T., and L. E. Hoglund, Wind-
Acoust. Soc. Amer., vol. 48, no. 3, 1969, Tunnel Fan Noise Reduction including

p. 783. Effects of Turning Vanes on Noise

47. Marsh, A. H., I. Elias, J. C. Hoehne, and Propagation, AIAA Paper 79-0642, March
R. L. Frasca, A Study of Turbofan-Engine 1979.
Compressor-Noise-Suppression Techniques, 61. Lansing, D. L., Exact Solution for Radiation
NASA CR-1056, 1968. of Sound from a Semi-Infinite Circular Duct
48. Hersh, A. S., and B. Walker, Effect of Graz- with Application to Fan and Compressor
ing Flow on the Acoustic impedance of Helm- Noise. Analytic Methods in Aircraft
holtz Resonators Consisting of Single and Aerodynamics, NASA SP-228, October 1969,
Clustered Orifices, CR 3177, 1979. p. 323.
49. Rice, E. J., Spinning Mode Sound Propaga- 62. Scharton, T. D., and M. D. Sneddon,
tion in Ducts with Acoustic Treatment, Acoustical Properties of Materials and
NASA TN D-7913, 1975. Muffler Configurations for the 80 x 120 Foot
50. Rice, E. J., Modal Propagation Angles in Wind Tunnel, NASA CR-152065, 1977.
Ducts with Soft Walls and their Connection 63. Ingard, U., and W. P. Patrick, A New
with Suppressor Performance, AIAA Paper Approach to Acoustic Filtering with Lined
79-0624, March 1979. Ducts, Third Interagency Symposium on
51. Rice, E. J., Attenuation of Sound in Ducts University Research in Transportation Noise,
with Acoustic Treatment--A Generalized November 1975, p. 522.
Approximate Equation, NASA TM X-71830, 64. Parrott, T. L., An lmproved Method for the
1975. Design of Expansion-Chamber Mufflers with
52. Zorumski, W. E., and J. P. Mason, Multiple Application to an Operational Helicopter,
Eigenvalues of Sound-Absorbing Circular Cosmic Program Abstract, LAR-I 1548, 1974.
and Annular Ducts, J. Acoust. Soc. Amer., 65. Anyos, T., L. Christy, R. Lizak, and J.
vol. 55, 1974, p. 1158. Wilhelm, Technology Transfer-Transporta-
53. Minner, G. L., and E. J. Rice, Annular tion, (prepared for NASA by SR1 Inter-
Acoustic Liners for Turbofan Engines, NASA national), Annual Report, Contract
Tech. Brief, Spring 1979, p. 132. NAS2-9848, 1978, p. 39.
54. Kraft, R. W., J. E. Pass, and L. R. Clark, 66. Hart, F. D., D. L. Neal, and F. O. Smetana,
Effects of Multi-Element Acoustic Treatment Industrial Noise Control: Some Case
on Compressor Inlet Noise, AIAA Paper Histories, Volume 1, NASA CR-142314,
76-515, July 1976. 1974.

55. Chestnutt, D., and C. E. Feiler, Advanced 67. Giurgiuman-Negrea, H., M. Creto, and 1.
Inlet Duct Noise Reduction Concepts. Moraru, Aspects of Reducing Spur Gear
Aircraft Safety and Operating Problems, Noise, The Fourth National Conference on
NASA SP-416, 1976, p. 481. Acoustics, Vol 1A: Noise and Vibrations

125
 Control,NASATT F-15,385,June1974,p. Multiroller Traction Drive, NASA TP
124. 1378/AVRADCOM TR 78-36, 1978.
68. Schlegel, R.Y., R. J. King,andH. R.Mull, 70. NASA Activities, National Aeronautics and
Howto Reduce GearNoise,MachineDesign, Space Administration, Voi. 10, no. 10, 1979,
Feb.27, 1964,pp. 134-142. p. 8.
69. Loewenthal, S. H., N. E. Anderson, and A.
L. Nasvytis, Performance of a Nasvytis

126
 APPENDIX

OSHA Noise Regulations and Assessment

of Hearing Impairment

A. OSHA STANDARD "Exposure to impulsive or impact noise

should not exceed 140 dB peak sound pressure
Under the Occupational Safety and Health level."
Act of 1970, two sets of regulations were promul- To achieve compliance with the standard, it
gated in August 1971 pertaining to general indus- is necessary first to compute the total noise expo-
try I and to the construction field. 2 The terms of sure of each employee and then, if necessary,
these are essentially the same and provide as examine the permutations of noise level and expo-
follows: sure duration which will ensure a daily noise dose
"When employees are subjected to sound of less than unity. Even with levels greater than
levels exceeding those listed in Table G-16, feasi- 90 riB(A), the total exposure may be within the
ble administrative or engineering controls shall be standard. Conversely, even when compliance is
utilized. If such controls fail to reduce sound achieved for each individual level, the total noise
levels within the levels of Table G-16, personal exposure for an 8-hour period may be greater than
protective equipment shall be provided and used unity. Both of these points are illustrated in the
to reduce sound levels within the levels of the following example with the modified noise level
Table. series 1 and 2, respectively.
Duration per Day Sound Level dB (A)
lz npltlyee Noise lc_et (dB 14i) _Xctual Permitted dLJrallotl (hr)
(hours) Slow Response
c'..p,,r, urc Initial Modified Mo,.lificd duration, Initial Modified Modilicd
8 90 ,:ondilioz/ t 2 (" Ihr) I 2
I 95 83 8,_ 27 4O
6 92 2 102 94 95 ; 6 15 44 a.
4 95 3 99 88 _8 O2 2_
4 108 91 ItS) 15 (1 _ _9 2
3 97
2 100
1 '/2 102
I 105 Daily noise dose
tA i10
_Aor less 115 Modified Modified
Initial
I 2
TABLE G-16-Permissible Noise Exposure
0.7 -- --

2.4 0.8 0.9

"When the daily noise exposure is composed
0.1 -- --
of two or more periods of noise exposure of 2.1 0.2 0.8
different levels, their combined effect should be Totals 5.3 1.0 1.7
considered, rather than the individual effect of
each. If the sum of the following fractions: If the daily noise exposure exceeds unity, it is
C l / T 1 + C 2/ T 2 + ... + Cn T n exceeds necessary to calculate the noise reduction
unity, then the mixed exposure should be necessary to achieve compliance with the
considered to exceed the limit value. T n indicates standard. It has been established that the question
the total time of exposure permitted at the level n, of "feasibility" of noise control measures rests on
and C n is the actual time of exposure at that level. both economic and technical factors) In some

127
noisecontrolproblems it maynot bepossible to -- Range of probability

establish
a totalcostestimateduntilseveralstep- of hearing impairment

wisestagesof engineering controlhavebeen 100

implemented. Costestimatesshouldbecompared
with the generallevelof total expenditure per 50
employeeto meetOSHA compliancefor that
particularindustry.
4
In October1974,proposedchanges
OSHAnoisestandardwerepublished.
in the
5 Although
...... ___/4
ao..... 17
there was to be no alteration of the basic noise
exposure
protectors
limits, it was proposed
be worn in exposure
that hearing
conditions of
,o / /
above 85 dB(A) and that monitoring audiometric
testing programs be established. At present there
are no standard procedures for using audiometry 3---]
as part of a hearing conservation program.

21Ip'I /
B. HEARING IMPAIRMENT
80 90 100 110 120

There is considerable debate as to what Tim_weighted continuous sound level (dB(A)}

constitutes hearing impairment as reflected in

Figure A.1- Probability of workers having a
different state procedures for assessing compensa- hearing threshold of 25 dB or more, averaged over
tion, although the most common criterion invoked 500, 1000, and2000 Hz, after 2O years of exposure
is that of the American Academy of Ophthal- to workplace noise. (Used by permission of
mology and Otolaryngology (AAOO). This uses a Acoustical Publications.)
hearing threshold of 25 dB relative to audiometric
zero, averaged over 500, 1000, and 2000 Hz.
However, there is no question that considerable REFERENCES
noise-induced hearing loss occurs at 3000 and 4000
Hz, and there is strong evidence that hearing 1. Anon., 29 Code of Federal Regulations 1910.
ability at these frequencies is very important under 95(b)(1). Federal Register, vol. 36, May 29,
less than ideal listening conditions. Figure A.1 1971, p. 10466.
shows the range of hearing loss risks in workers 2. Anon., 29 Code of Federal Regulations 1926.
after 20 years of continuous employment. 6 The 52(b).
lower curve shows the risk of permanent impair- 3. Parashes, P. T., A Legal Overview of the
ment after all factors except workplace noise have OSHA Noise Standard at 29 CFR 1910.
been eliminated; the upper curve includes the 95(b)(1), Sound and Vibration, June 1978, p.
effects of disease, abnormalities, and intense noise 10.
experienced outside work. A monitoring audi- 4. Heggie, A. S., The OSHA Noise Stan-
ometry program consists of an initial aural exam- dard-How to Live With It, Sound And Vibra-
ination of all new company employees and their tion, June 1978, p. 20.
audiometric testing at regular intervals thereafter. 5. Occupational Noise Exposure--Proposed Re-
Apart from benefiting the health of employees, quirements and Procedures, Federal Register,
such a program also enables the company to check vol. 39, 207, p. 20.
on any noise-induced hearing loss caused by 6. Broce, R. D., et al., Workplace Noise Ex-
previous workplace exposure and to identify posure Control--What are the Costs and
individuals whose hearing is already seriously Benefits?, Sound and Vibration, September
diminished. 1976, p. 12.

128
 Glossary
Units

A Total room absorption, equal to the sum of the products of the area of
each ith surface element, Si, and the corresponding absorption coeffi-
cients, % defined mathematically as

A = ESia i metric
1
sabins,
m2

Stiffness (Young's) modulus, the ratio of stress to strain in a solid bar N/m 2

D Acoustic energy density contained in a sound field J/m 3

D Diameter m

1 Acoustic intensity, the average rate of flow of sound energy through a

unit area perpendicular to the direction under consideration W/m 2

/ref Reference sound intensity taken as 10- 12 watts per square meter W/m 2

IL Acoustic intensity level expressed in decibel notation as 10 times the

logarithm (base 10) of the ratio of a sound intensity to the reference
sound intensity, 10- 12 W/m 2 dB

lnL Insertion loss, the difference in sound pressure level at a particular loca-
tion before and after acoustic treatment is applied dB

K Number of blades, gear teeth, etc.

L Length (cavity, muffler, etc.) m

Ldn Day-night average sound level, the 24-hour equivalent continuous sound
level with a 10-riB weighting applied to sound levels between 10:00 p.m.
and 7:00 a.m. dB(A)

Leq Equivalent continuous sound level, the level of a steady sound which,
over the same time period, contains the same sound energy as the time-
varying sound dB(A)

L X x-Percentile exceeded sound level, the level (fast) equaled or exceeded by

a time-varying sound x percent of a given time period dB(A)

129
 Units

M Mach number, the ratio of the flow velocity to the ambient speed of
sound

N Fresnel number, defined mathematically as

26
N= +
X

where
6 = path length difference around barrier
X = wavelength of interest

N Shaft or fan rotational speed rpm, rps

NR Noise reduction, the difference between two sound pressure levels across
a partition dB

OASPL Overall sound pressure level, i.e., the sound pressure level over all fre-
quency bands in the audio frequency range (usually 31-16 000 Hz) dB

PT Total pressure N/m 2 (Pa)

Po Reference ambient atmospheric pressure, normally taken as equal to

1.012 x 102 newtons per square meter at sea level N/m 2

PSD Power spectral density, defined mathematically as

PSD lim Wi
Bi_O Bi

i.e., in the limit as the bandwidth

approaches zero, the PSD is the ratio
of power in a bandwidth W i, divided
by that bandwidth Bi W/Hz

PWL Sound power level expressed in decibel notation as 10 times the logarithm
(base 10) of the ratio of a sound power to the reference sound power,
10- 12 watts dB

Q Air flow rate m3/min

Q DirectMty factor, the ratio of sound intensity at angle 0 to the intensity

at the same location when the source is radiating omnidirectionally

SPL Sound pressure level expressed in decibel notation as 20 times the

logarithm (base 10) of the ratio of a sound pressure to the reference
sound pressure, 2 × 10 -5 N/m 2 dB

130
 Units

R Sound transmission loss of an insulating material, defined as 10 times the

logarithm (base 10) of the reciprocal of the transmission coefficient dB

g c Room constant, a convenient measure of room absorption; defined

mathematically as

where metric
S = total room internal area, m 2 sabins,
o_ = average absorption coefficient of room surfaces m2

RMS Root mean square

S Area (partition area, hypothetical enveloping surface, etc.) m2

T Absolute temperature K

T Period, equal to the time taken for a regular waveform to complete one
cycle of oscillation min, s

rR Reverberation time according to Sabine, defined mathematically as

0.161V
TR - A

where
V = room volume, m3
A = room absorption, metric sabins s

U Acoustic medium particle velocity in direction of propagation m/s

V Volume (room, cavity, etc.) m3

W Acoustic power, the total sound energy produced by a source in unit time W

Wref Reference sound power taken as 10- 12 watts W

Z Specific acoustic impedance, the ratio of complex acoustic pressure to Ns/m 3

complex particle velocity at the surface of a material (mks rayls)

Velocity of sound propagation in air, normally taken as 343 meters per

second at room temperature and pressure m/s

dB 10 times the logarithm (base 10) of the ratio of two physical quantities dB

131
dB(A) A-Weighted decibels, normally summed across the entire audio fre-
quency range dB(A)

dB(C) C-Weighted decibels, normally summed across the entire audio fre-
quency range dB(C)

d Separation distance of two surfaces cm, m

f Frequency, the number of complete oscillations of a periodic waveform

that occur in one second Hz

P Acoustic sound pressure, the difference between the actual total pressure
and the mean ambient pressure N/m 2

Pref Reference sound pressure taken as 2 x 10 -5 newtons per square meter N/m 2

Prms Root mean square of instantaneous sound pressure over a specified time
interval N/m 2

r Distance from noise source m

r_ Distance from source at which transition from direct to reverberant fields

begins to occur m

v Air or liquid flow velocity m/s

x Distance coordinate m

Absorption coefficient, taken usually as the fraction of random in-

cidence sound power absorbed by a material

Path length difference, the difference between direct path from source to
receiver and the shortest path around a barrier m

Damping ratio, the ratio of the damping coefficient of the material under
consideration to the critical damping coefficient

Damping loss factor, equal to twice the ratio of the material damping
coefficient to the critical damping coefficient of that material

Wavelength, equal to the distance between congruent points on suc-

cessive oscillations of a single frequency tone m

0 Density kg/m 3

Oc Characteristic acoustic impedance, the real part of the specific acoustic Ns/m
impedance Z, and equal to this only for free progressive plane waves (mks rayls)

Transmission coefficient, taken as the fraction of incident sound power

transmitted through a material

132
SUBSCRIPTS

E Enclosure variable

REV Reverberant field variable

S Component of variable passing through area S

T Total value

J Jet variable

Normal component of variable or nth harmonic component

o Equilibrium or ambient value of variable

P Peak value of variable

ref Reference value of variable

res Value of variable at resonance or at undamped natural fre-

quency of a loaded system

rms Root mean square

! Test value of variable

0 Directional component at angle 0 of variable

133
 Bibliography

GENERAL TEXTS Yerges, L. F., Sound, Noise, and Vibration Con-

trol, Second ed., Van Nostrand Reinhold Co.,
AIHA, Industrial Noise Manual, second ed., 1978.
American Industrial Hygiene Assoc., 1966.
Beranek, Leo L., ed., Noise and Vibration Con-
trol, McGraw-Hill Book Co., 1971. JOURNALS AND PERIODICALS
Berendt, R. D., E. L. R. Corliss, and M. S.
Ojalvo, Quieting: A Practical Guide to Noise Acustica: S. Hirze, Stuttgart 1, Birkenwaldstr. 44,
Control, National Bureau of Standards Hand- Postfach 347, Germany (monthly).
book 119, U.S. Dept. of Commerce, 1976. Applied Acoustics: Applied Science Publishers,
Broch, J. T., Acoustic Noise Measurements, sec- Ltd., Ripple Rd., Barking, Essex, England
ond ed., Bruel & Kjaer, 1971. (quarterly).
Bell, L. H., Fundamentals of Industrial Noise Journal of Sound and Vibration: Academic Press,
Control, second ed., Harmony Publications, Inc., 111 Fifth Avenue, New York, N.Y. 10003
1973. (bi-weekly).
Cheremisinoff, P. N. and P. P. Cheremisinoff, Noise Control Engineering: Institute of Noise
Industrial Noise Control Handbook, Ann Ar- Control Engineering, P.O. Box 3206, Arlington
bor Science Publishers, 1977. Branch, Poughkeepsie, N.Y. 12603 (bi-
Cremer, L., and M. Heckl (E. E. Ungar, monthly).
transl.), Structure-Borne Sound, Springler- Noise/News: Institute of Noise Control Engineer-
Verlag, 1973. ing, P.O. Box 3206, Poughkeepsie, N.Y. 12603
Crocker, M. J., ed., Reduction of Machinery (bi-monthly).
Noise, Revised ed., Purdue University, 1975. Sound and Vibration: Acoustical Publications,
Diehl, G. M., Machinery Acoustics, John Wiley & Inc., 27101 E. Oviatt Road, Bay Village, Ohio
Sons, 1973. 44140 (monthly).
Faulkner, L. L., ed., Handbook of Industrial The Journal of the Acoustical Society of America:
Noise Control, Industrial Press, 1976. Acoustical Society of America, 335 E. 45 Street,
Harris, C. M., ed., Handbook of Industrial Noise New York, N.Y. 10017 (monthly).
Control, Second ed., McGraw-Hill Book Co., The Shock and Vibration Digest: The Shock and
1979. Vibration Information Center, Code 8404
Lyon, R. H., Statistical Energy Analysis of Naval Research Laboratory, Washington, D.C.
Dynamical Systems: Theory and Applications, 20375 (monthly).
MIT Press, 1975.
Miller, R. K., and W. V. Montone, Handbook
of Acoustical Enclosures and Barriers, Fair- NOISE CONFERENCE PROCEEDINGS
mont Press, 1978.
Peterson, A. P. G., and E. E. Gross, Jr., Hand- Interagency Symposium on University Research in
book of Noise Measurement, Seventh ed., Transportation Noise: Proceedings of 1st at
General Radio, 1974. Stanford University, 1973; 2nd at North Caro-
Salmon, V., J. S. Mills, and A. C. Peterson, lina State University, 1974; 3rd at University of
Industrial Noise Control Manual (prepared for Utah, 1975.
NIOSH), Dept. of Health, Education, and Wel- International Congress on Acoustics: lst-9th
fare, 1975. Congress Proceedings, held every 3 years.
Webb, J. D., ed., Noise Control in Industry, Elsevier Pub. Co., and Haistead Press.
Sound Research Laboratories, Halstead Press, Inter-Noise: Proceedings of the International
1976. Conference on Noise Control Engineering.

135
 Noise Control Foundation, P.O. Box 3469, Arl- NIOSH: Criteria for a Recommended Standard
ington Branch, Poughkeepsie, N.Y. 12603, ... Occupational Exposure to Noise. HSM
1972. 73-11001. National Institute for Occupational
Noise and Vibration Control Engineering: Pro- Safety and Health, U.S. Dept. of Health,
ceedings of the Purdue Noise Control Con- Education, and Welfare, 1972.
ference. Purdue University, Lafayette, Indiana, Schemidek, M. E., et al., Survey of Hearing Con-
July 14-16, 1971. servation Programs in Industry. HEW Pub.
Noise-Con: Proceedings of the National Con- No. (NIOSH) 75-178. National Institute for Oc-
ference on Noise Control Engineering. Noise cupational Safety and Health, U.S. Dept. of
Control Foundation, P.O. Box 3469, Arlington Health, Education, and Welfare, 1975.
Branch, Poughkeepsie, N.Y. 12603. Wyle Laboratories: Community Noise. NTID
Noisexpo: Proceedings of the Technical Program: 300.3. U.S. Environmental Protection Agency,
National Noise and Vibration Control Con- Washington, D.C. 1971.
ference, Sound and Vibration, 27101 E. Oviatt
Road, Bay Village, Ohio, 44140. MANUALS
The Fourth National Conference on Acoustics
(Rumania): VoL L A. Noise and Vibration Emerson, P. D., J. R. Bailey, and F. D. Hart,
Control 1973. NASA TT F-15,375, 1974. Manual of Textile Industry Noise Control,
The Shock and Vibration Bulletin: Proceedings of North Carolina State University, 1978.
the Annual Shock and Vibration Symposium. Hart, F. D., C. L. Neal, and F. O. Smetana,
The Shock and Vibration Information Center, Industrial Noise Control." Some Case Histories,
Code 8404 Naval Research Laboratory, Wash- Vol. 1, NASA CR 142314, North Carolina
ington, D.C. 20375. Science and Technology Research Center, 1974.
University Noise Research: Proceedings of the Pearsons, K. S., and R. Bennett, Handbook of
EPA-University Noise Seminar. EPA 550/9-77 Noise Ratings, NASA CR-2376, 1974.
Office of Noise Abatement and Control, EPA, Purcell, W. E., and B. L. Lempert, Compendium
Washington, D.C. 20460, October 18-20, 1976. of Materials for Noise Control HEW Pub. No.
(NIOSH) 75-165, NIOSH, U.S. Dept. Health,
Education, and Welfare, 1975.
GENERAL POLICY AND INFORMATION Salmon, V., et al., Industrial Noise Control
DOCUMENTS Manual (revised), HEW Pub. No. (NIOSH)
79-117, prepared by Industrial Noise Services,
Center for Policy Alternatives, MIT: Some Con- Inc. for NIOSH, U.S. Dept. Health, Education,
siderations in Choosing an Occupational Noise and Welfare, 1979.
Exposure Regulation. EPA 550/9-76-007. Of- Simpson, M. A., Noise Barrier Design Handbook,
fice of Noise Abatement and Control, U.S. FHWA-RD-76-58, prepared by Bolt Beranek
Environmental Protection Agency, Washing- and Newman, Inc., for Federal Highway Ad-
ton, D.C., 1976. ministration, U.S. Dept. of Commerce, 1976.
EPA: Toward a National Strategy for Noise Con- Wyle Laboratories, Aerodynamically Quiet Cir-
trol. Office of Noise Abatement and Control, cular Saw Blades... A Design and Application
U.S. Environmental Protection Agency, Wash- Manual prepared by Wyle Labs., for Wood-
ington, D.C., 1977. working Machinery Manufacturers of America,
Kroes, P., R. Fleming, and B. Lempert, List of Philadelphia, Pa., 1978.
Personal Hearing Protectors and Attenuation
Data. HEW Pub. No. (NIOSH) 76-120. Na- BIBLIOGRAPHIES AND RESEARCH
tional Institute for Occupational Safety and REVIEWS
Health, U.S. Dept. of Health, Education, and
Welfare, 1975. EPA: Noise Technology Research Needs and The
L. S. Goodfriend Associates: Noise from In- Relative Roles of the Federal Government and
dustrial Plants. NTID 300.2. Office of Noise The Private Sector, EPA 550-/9-79-311, Office
Abatement and Control, U.S. Environmental of Noise Abatement and Control, U.S. Environ-
Protection Agency, Washington, D.C., 1971. mental Protection Agency, 1978.

136
Pusey,H. C., andR. H. Bolin,An International Hersh, A. S., and R. E. Hayden, Aerodynamic
Surveyof Shockand VibrationTechnology, Sound Radiation from Lifting Surfaces with
ShockandVibrationInformation Center,Naval and without Leading-Edge Serrations, NASA
Research Laboratory,Washington,D.C.,1979. CR-114370, 1971.
TheFederallnteragency
Machinery andConstruc- Howes, W. L., Overall Loudness of Steady Sounds
tion Noise Research Panel: Federal Research, According to Theory and Experiment, NASA
Development and Demonstration Programs in RP 1001, 1979.
Machinery and Construction Noise, EPA 550/9- Howlett, J. T., S. A. Cievenson, J. A. Rupf, and
78-306, Office of Noise Abatement and Control, W. J. Snyder, Interior Noise Reduction in a
U.S. Environmental Protection Agency, 1978. Large Civil Helicopter, NASA TN D-8477,
The Federal lnteragency Surface Transportation 1977.
Noise Research Panel: Federal Research, Devel- Lambert, R. F., Acoustical Properties of Highly
opment and Demonstration Programs in Sur- Porous Fibrous Materials, NASA TM 80135,
face Transportation Noise, EPA 550/9-78-305, 1979.
Office of Noise Abatement and Control, U.S. Levine, L. S., and J. J. DeFelice, Civil Helicopter
Environmental Protection Agency, 1978. Research Aircraft Internal Noise Prediction,
The Technology Application Center, Institute for NASA CR-145146, 1977.
Social Research and Development: Noise Pollu- Manning, J. E., and P. J. Remington, Statistical
tion Resources Compendium, TAC Bibliographic Energy Methods, NASA CR-118987, 1971.
Series No. 2, The University of New Mexico, McDonald, W. B., R. Vaicaitis, and M. K. Myers,
1972. Noise Transmission Through Plates into an En-
Shock and Vibration Information Center: An closure, NASA TP 1173, 1978.
International Survey of Shock and Vibration Montegani, F. J., Some Propulsion System Noise
Technology, Naval Research Laboratory, Wash- Data Handling Conventions and Computer
ington, D.C., 1979. Programs used at the Lewis Research Center,
NASA TM X-3013, 1974.
Piumblee, H. E., Jr., P. D. Dean, G. A. Wynne,
NASA TECHNICAL PAPERS and R. H. Burrin, Sound Propagation in and
Radiation from Acoustically Lined Flow Ducts:
Clemons, A., Quiet Clean Short-Haul Experi- A Comparison of Experiment and Theory,
mental Engine (QCSEE) Acoustic Treatment NASA CR-2306, 1973.
Development and Design, NASA CR-135266, Parrott, R. L., An Improved Method for the
1979. Design of Expansion-Chamber Mufflers with
Copeland, W. Latham, Inlet Noise Studies for an Application to an Operational Helicopter,
Axial-Flow Single-Stage Compressor, NASA NASA TN D-7309, 1973.
TN D-2615, 1965. Smith, P. W., and R. H. Lyon, Sound and Struc-
Getline, G. L., Low-Frequency Noise Reduction tural Vibration, NASA CR-160, 1965.
of Lightweight Airframe Structures, NASA Stone, J. R., Interim Prediction Method for Jet
CR-145104, 1976. Noise, NASA TM X-71618, 1975.
Goldstein, M. E., Aeroacoustics, NASA SP-346, Vaicaitis, R., Noise Transmission by Viscoelastic
1974. Sandwich Panels, NASA TN D-8516, 1977.
Goldstein, M. E., and W. L. Howes, New Aspects Zorumski, W. E., Acoustic Theory of Axisym-
of Subsonic Aerodynamic Noise Theory, NASA metric Multisectioned Ducts, NASA TR R-419,
TN D-7158, 1973. 1974.

137
 NASA Technology Utilization Information

PUBLICATIONS Technology Applications Center (TAC)

University of New Mexico
NASA Tech Briefs (quarterly) Albuquerque, New Mexico 87131
Director, Technology Utilization Office
P.O. Box 8757
Baltimore/Washington International Airport, Western Research Applications Center
MD 21240 (WESRAC)
University of Southern California
Cosmic Computer Programs University Park
Los Angeles, California 90007
Computer Software and Management
Information Center
Suite 112, Barrow Hall
Kerr Industrial Applications Center (KIAC)
Athens, GA 30602
Southeastern Oklahoma State University
Durant, Oklahoma 74701

INDUSTRIAL APPLICATIONS CENTERS

Aerospace Research Applications Center NASA TECHNOLOGY APPLICATION

(ARAC) TEAMS
1201 East 38th Street
Indianapolis, Indiana 46205 Public Technology Inc.
1140 Connecticut Avenue, N.W.
Knowledge Availability Systems Center (KASC) Washington, D.C. 20036
University of Pittsburgh
Pittsburgh, Pennsylvania 15260
SRI International
333 Ravenswood Avenue
New England Research Application Center
(NERAC) Menlo Park, California 94026
Mansfield Professional Park
Storrs, Connecticut 06268 liT Research Institute
10 West 35th Street
North Carolina Science and Technology Chicago, Illinois 60616
Research Center (NC/STRC)
Post Office Box 12235
Research Triangle Park, North Carolina 27709

139
NASA-Langley,
1981