Вы находитесь на странице: 1из 22

A transfer-matrix approach for estimating the characteristic

impedance and wave numbers of limp and rigid porous materials


Bryan H. Song and J. Stuart Boltona)
1077 Ray W. Herrick Laboratories, School of Mechanical Engineering, Purdue University, West Lafayette,
Indiana 47907-1077

Received 10 February 1999; accepted for publication 10 November 1999


A method for evaluating the acoustical properties of homogeneous and isotropic porous materials
that may be modeled as fluids having complex properties is described here. To implement the
procedure, a conventional, two-microphone standing wave tube was modified to include: a new
sample holder; a section downstream of the sample holder that accommodated a second pair of
microphone holders and an approximately anechoic termination. Sound-pressure measurements at
two upstream and two downstream locations were then used to estimate the two-by-two transfer
matrix of porous material samples. The experimental transfer matrix method has been most widely
used in the past to measure the acoustical properties of silencer system components. That procedure
was made more efficient here by taking advantage of the reciprocal nature of sound transmission
through homogeneous and isotropic porous layers. The transfer matrix of a homogeneous and
isotropic, rigid or limp porous layer can easily be used to identify the materials characteristic
impedance and wave number, from which other acoustical quantities of interest can be calculated.
The procedure has been used to estimate the acoustical properties of a glass fiber material: good
agreement was found between the estimated acoustical properties and those predicted by using the
formulas of Delany and Bazley. 2000 Acoustical Society of America. S0001-49660004302-2
PACS numbers: 43.20.Jr, 43.20.Mv, 43.55.Ev, 43.58.Bh DEC
INTRODUCTION

Many types of sound-absorbing materials are used in


noise-control applications. These materials include, for example: glass fiber, polymeric fibrous materials, and various
types of foams. It is of interest to be able to predict the
noise-control impact of these materials, whether in sound
absorption or barrier applications, at the design stage: finite
element, boundary element, and statistical energy analysis
programs make that possible in principle. However, to take
full advantage of these software capabilities, it is necessary
to have detailed knowledge of the acoustical properties of the
noise-control materials. In particular, it is usually necessary
to know their characteristic impedance and wave number. A
procedure addressing that need is presented here.
We will consider, in particular, porous materials that are
homogeneous i.e., materials whose properties are spatially
uniform and isotropic i.e., materials whose properties are
independent of direction, and whose expanded solid phase
may be assumed to be either perfectly rigid or perfectly limp.
In general, most noise-control materials are poroelastic: i.e.,
they can support two compressional waves and a transverse
wave. However, in certain frequency or material property
regimes, certain porous materials behave as though they
were rigid, i.e., their solid phase motion is negligible compared to that of the fluid phase owing to weak coupling or
the relatively high density or stiffness of the solid phase. In
other regimes, certain materials such as low density, high
flow resistivity, unreinforced glass fibers may be approximated as limp: i.e., the in vacuo stiffness of their bulk solid
phase is negligible compared to that of the saturating fluid.
a

Author to whom correspondence should be addressed. Electronic mail:


bolton@ecn.purdue.edu

1131

J. Acoust. Soc. Am. 107 (3), March 2000

For practical purposes, many fibrous materials and foams fall


into these two categories in specific frequency ranges of interest. Air-saturated, homogeneous, and isotropic porous materials that can be approximated as either rigid or limp may
be considered to support only a single dilational wave type
and may thus be modeled as dissipative fluids having complex physical properties: e.g., characteristic impedance and
wave number.1 The acoustical properties of such a medium
are completely specified when the latter two quantities, or
two independent properties derived from them e.g., complex
density and complex sound speed are known. Thus, it is of
interest to be able to determine experimentally the wave
number and the characteristic impedance of a homogeneous
and isotropic porous material.
The current effort is an extension of work reported earlier by Bolton et al.2 In that work, a four-microphone, standing wave tube measurement of the normal incidence reflection and transmission coefficients of automotive sealant
materials was described. The success of that procedure depended on the provision of an anechoic termination for the
standing wave tube. Here, it is shown that the same measured data may be used to determine the transfer matrix of
the material under test even though no conditions are imposed on the tube termination properties. The transfer matrix
elements may then be related in closed form to the porous
materials characteristic impedance and wave number. Note
that the latter property, in particular, yields both the phase
speed and the spatial rate of energy dissipation within the
material. Note also that when the transfer matrix of a layer of
acoustical material is known, it may be combined with transfer matrices of other acoustical elements e.g., barrier or resistive layers to form the complete transfer matrix of a
multilayer noise control treatment.3 The latter matrix may

0001-4966/2000/107(3)/1131/22/$17.00

2000 Acoustical Society of America

1131

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

then be used to predict and optimize the absorption and barrier properties of the treatment.
The transfer matrix method has been widely used in the
past to analyze and to measure the acoustical properties of
flow system elements e.g., automotive mufflers. The particular implementation of the transfer matrix method described here was made possible by taking advantage of the
reciprocal nature of sound transmission through typical
acoustical materials: that reciprocity places conditions on the
structure of the transfer matrices of these materials. The resulting mathematical constraints make it possible to determine the acoustical properties of homogeneous and isotropic
porous materials by using fewer measurements than are typically necessary when using a transfer matrix approach.
The procedure is demonstrated here through measurements of the acoustical properties of aviation-grade glass fibers. Good agreement was found between the measured
acoustical properties of these materials and predictions made
using well-established semiempirical formulas. Note, however, that the procedure applied here cannot be applied directly in its present form to materials that must be modeled
as poroelastic: i.e., those materials in which two compressional waves and a transverse wave contribute significantly
to the materials acoustical properties.
I. BACKGROUND

In 1946, Scott described a technique for directly measuring the propagation constant and for inferring the characteristic impedance of a porous material.4 The propagation
constant, or wave number, as defined by Scott comprised a
real component, the propagation factor, governing the phase
change per unit length and an imaginary component, the attenuation factor, governing the exponential decay of sound
pressure within the porous material. The attenuation factor at
a single frequency was determined by passing a microphone
probe through a deep sample and measuring the decay of
mean-square sound pressure with position. The propagation
factor was determined at the same time by measuring the
sound fields phase change with distance. The characteristic
impedance was inferred by measuring the normal specific
acoustic impedance of a sample sufficiently deep that reflection from the tube termination was negligible. The same approach was also described by Beranek.5 The disturbance of
the acoustical material caused by the microphone probe has
proven to be a problem with Scotts technique in application
to some types of materials. More recently, a nonintrusive
version of Scotts wave number measurement was described
by Lambert and Tesar.6 In their procedure, a long sample of
porous material was placed in a standing wave tube as in
Scotts approach. However, the single-frequency sound field
within the porous material was sampled at 1-cm intervals
along its length by means of a probe microphone inserted
through the tube wall. The real and imaginary parts of the
wave number can easily be calculated from the latter data.
Following Scotts work, Beranek7 used an elastic porous
material theory similar to that of Zwikker and Kosten8 to
relate the wave number and characteristic impedance of a
porous material to its macroscopic physical properties: i.e.,
flow resistance, structure factor as then defined, porosity,
1132

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

density, and the volume stiffness coefficients for the air and
structure. He suggested that the theory could be used as an
alternative to direct measurement of the wave numbers and
characteristic impedances of either limp or rigid porous material.
Ferrero and Sacerdote subsequently proposed a second
style of material property measurement based on two measurements of surface-normal impedance.9 In particular, they
showed that it was possible to determine the characteristic
impedance and wave number of a porous material graphically and analytically by measuring at a single frequency the
surface-normal specific acoustic impedance of two different
thicknesses of the same porous material, one thickness being
double the other. Their particular procedure has since become known as the two-thickness method. When using this
technique, however, it is sometimes difficult to mount a second sample without disturbing the first, and the procedure
becomes inaccurate when the product of the attenuation factor and the sample depth is sufficiently large that the difference between the surface impedances of the two thicknesses
of material is relatively small. The latter problem is particularly significant for fibrous materials having high flow resistivities.
Yaniv eliminated problems associated with the use of
two thicknesses of material by introducing the two-cavity,
single-frequency method: during the first surface impedance
measurement the sample was backed by a rigid wall, while in
the second measurement the sample was backed by a onequarter-wavelength deep cavity terminated by a rigid wall.10
The wave numbers and characteristic impedances of a number of materials obtained by using this procedure were compared both with predictions made using the Beranek theory
and with results obtained by using Scotts method. The twocavity method was shown to give results in good agreement
with Beraneks theory over the frequency range considered:
Scotts method was found to be less accurate for the materials that Yaniv considered. Like the two-thickness method,
however, the two-cavity method can become inaccurate
when the sample material is highly dissipative: i.e., when the
surface-normal impedance is relatively insensitive to backing
conditions.
Smith and Parott have also evaluated the two-thickness
method, the two-cavity method, and Scotts method.11 They
found that the significant advantage of the two-cavity
method compared with the two-thickness method lies in the
reduction of both measurement times and the variation associated with mounting test materials that is made possible by
using the former method.
Utsuno et al. subsequently described a cross-spectral
implementation of the two-cavity method based on the twomicrophone method for measuring surface-normal
impedance.12,13 In their procedure, the wave number and
characteristic impedance were estimated on the basis of two
measurements of the surface impedance of a sample insonified by broadband noise: first when backed by a hard termination and then by an arbitrarily deep air space. Their approach allows results to be obtained over a broad range of
frequencies simultaneously and eliminates the need for a different backing depth at each frequency. However, in comB. H. Song and S. Bolton: Porous material properties

1132

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

mon with other techniques based on surface-normal impedance measurement, this approach can be inaccurate when
applied to highly dissipative materials.
To address the latter concern, a third general type of
estimation procedure combining surface impedance information with measurements of transfer properties across the
sample has been developed by a number of investigators.
Such a procedure was implemented by Bordone-Sacerdote
and Sacerdote.14 In their procedure, transfer functions were
measured in two directions across a porous sample that divided a short tube terminated at both ends by acoustic drivers. By taking advantage of the reciprocal nature of sound
transmission through the sample, they were able to determine
parameters related directly to the wave number and characteristic impedance of the material under test. Although the
authors intention was to develop a procedure specifically
useful at very low frequencies, and which could also be used
to measure the materials dynamic flow resistivity, there appears to be no reason that this procedure could not work at
any frequency in the plane wave range. Note, however, that
the acoustic drivers which terminate the tube are required to
present a zero-velocity boundary condition when not in use:
this requirement may be difficult to satisfy when using conventional loudspeakers.
Some time later, Ingard and Dear described a simplified
two-microphone procedure for measuring the dynamic
flow impedance of a sample thin enough so that the acoustic
particle velocities on its two surfaces are negligibly
different.15,16 The latter requirement restricts the use of their
procedure to low frequencies. In this procedure a sample is
placed in a standing wave tube at a distance L from a rigid
termination, and the sound pressure is measured both at the
termination and at the front surface of the sample. At frequencies such that L is an odd number of quarter wavelengths, the normalized flow impedance of the sample is directly related to a ratio of the two sound pressures. This
procedure is of interest since the flow impedance as defined
by the authors may be related to the materials characteristic
impedance. However, owing to the requirement that the
sample depth be very small compared to a wavelength, this
procedure cannot be used to estimate a materials wave number, and as noted, it may only be applied at a discrete set of
frequencies determined by the length of the terminating air
space.
McIntosh et al. subsequently described a more general
apparatus and procedure for measuring the low frequency,
dynamic flow impedance of a sample.17 Their apparatus was
similar to that of Ingard and Dear except that a linear array of
microphones positioned in front of the incident surface of the
sample was used to estimate the complex pressure and particle velocity on the incident face of the sample at each frequency of interest. The sound pressure measured at a microphone mounted flush with the rigid termination was used to
estimate the same information at the transmission face of the
sample without restriction on frequency. The pressure and
particle velocity information could then be used to estimate
the dynamic flow resistivity at low frequencies: i.e., when
the sample thickness was small compared to a wavelength.
However, McIntosh et al. noted that the same information
1133

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

could be combined with a single-wave model for sound


propagation within the porous layer to determine the latters
characteristic impedance and wave number at any frequency
in the plane wave region by means of an iterative, numerical
solution. Measured results for an open cell foam were in
generally good agreement with appropriate theoretical predictions.
Champoux and Stinson have described a procedure similar in some respects to that of McIntosh et al. that entails
measurement of both the surface impedance of a sample
backed by a fixed-depth air space and the pressure transfer
function across the sample.18 The latter quantity was estimated by using a reference microphone at the front surface
of the sample and a cavity microphone behind the sample.
The shallow cavity behind the sample was terminated by a
rigid backing. The procedure implemented by Champoux
and Stinson was based on the use of a traversing probe tube
to estimate the surface-normal impedance of the sample, and
so measurements were made one frequency at a time. The
surface impedance and the transfer function across the
sample together provide the information necessary to calculate the characteristic impedance and the wave number of the
material in closed form. The direct measurement of the transfer function across the sample makes it possible for this
method to yield accurate estimates of the characteristics of
dissipative materials having high flow resistivities. The latter
property is shared with all the other procedures described
here that involve measurements on both sides of the
sample.1417 Note that a procedure essentially identical to
that of Champoux and Stinson, save for the elimination of
the air space behind the sample, has recently been described
by Iwase et al.19 Note finally that the Ingard and Dear, McIntosh et al., and Champoux and Stinson procedures are similar in some respects to one- and two-microphone techniques
that were developed for measuring the impedance of duct
lining materials: see, for example, Ref. 20.
All of the procedures described above entail the use of a
sample placed in sealed standing wave tube. More recently,
Amedin et al. developed a measurement procedure based on
sound transmission from a standing wave tube to a free
field.21 Their method involved a two-microphone measurement inside an impedance tube, and one sound-pressure measurement immediately above a plug of porous material at the
tube termination: the tube termination was itself mounted
flush with a large baffle. The authors claimed that the direct
measurement of the sound pressure above the sample material resulted in high sensitivity and accurate results.
The procedure to be described here is of the same family
as those described in Refs. 1421 in the sense that measurements are made on both sides of a cylindrical plug of material placed in a standing wave tube. However, the present
procedure differs from earlier approaches since knowledge
of the tube termination condition is not required and because
it is based more closely on techniques previously used to
measure the acoustical performance of flow system components: i.e., transfer matrix methods.
Transfer matrix approaches have found frequent application in acoustics, and they have also been used in hydraulics applications where this approach is sometimes known as
B. H. Song and S. Bolton: Porous material properties

1133

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

plex pressures, P 1 to P 4 , comprise various superpositions of


positive- and negative-going plane waves in the up- and
downstream segments of the standing wave tube; the complex amplitudes of those waves are represented by the coefficients A to D. Equations 1a1d yield four equations for
the coefficients A to D in terms of the four measured sound
pressures, i.e.,
j P 1 e jkx 2 P 2 e jkx 1
,
2 sin k x 1 x 2

2a

j P 2 e jkx 1 P 1 e jkx 2
,
B
2 sin k x 1 x 2

2b

FIG. 1. Schematic diagram of the standing wave tube.

the two-port or four-pole method.22 As a result, there is a


large body of literature related to the measurement of the
properties of acoustical elements based on transfer matrix, or
two-port, representations.14,2328 The transfer matrix method
has most often been applied to the characterization of air
duct system components, exhaust silencers, etc. The transfer
matrix method can also be applied as an adjunct to the
boundary element method or the finite element method in
noise-control design procedures.29,30
An important observation related to the application of
the transfer matrix method to acoustical elements was made
by Pierce, who noted the constraints placed on the transfer
matrix elements by the requirements of symmetry when the
acoustical element was reciprocal.31 Those constraints had
been used earlier, without comment, by Bordone-Sacerdote
and Sacerdote.14
In the present study, an experimental transfer matrix approach of the type previously used to measure the performance of flow system elements was combined with the constraints imposed by sample symmetry and reciprocity to
make possible the efficient measurement of porous material
properties by using a suitably modified standing wave tube.

j P 3 e jkx 4 P 4 e jkx 3
,
2 sin k x 3 x 4

2c

j P 4 e jkx 3 P 3 e jkx 4
.
2 sin k x 3 x 4

2d

The latter coefficients provide the input data for subsequent


transfer matrix calculations. Note that the wave number, k,
should be made complex to account for the effects of viscous
and thermal dissipation in the oscillatory, thermoviscous
boundary layer that forms on the inner surface of the duct. In
the present work, formulas appropriate for wide ducts given
by Pierce were used to calculate the real and imaginary parts
of the wave number.31

B. Transfer matrix formulation

Here, the transfer matrix is used to relate the sound pressures and normal acoustic particle velocities on the two faces
of a porous layer extending from x0 to xd as in Fig. 1,
i.e.,

II. THEORY OF THE TRANSFER MATRIX METHOD


A. Sound-field representation

In the approach considered here, a loudspeaker was used


to generate a plane wave field in a standing wave tube, and a
single microphone was used to measure the transfer functions between the signal provided to the loudspeaker and the
sound pressure at the four locations shown in Fig. 1. Those
transfer functions are denoted P 1 to P 4 . In subsequent calculations, the quantities P 1 to P 4 always appear in ratios,
with the result that the spectral characteristic of the input
signal is canceled. Thus, for present purposes P 1 to P 4 may
be considered to represent the complex sound pressure at the
four measurement locations x 1 to x 4 , i.e.,
P 1 Ae jkx 1 Be jkx 1 e j t ,
P 2 Ae

jkx 2

P 3 Ce

jkx 3

Be

jt

1a

1b

jt

1c

P 4 Ce jkx 4 De jkx 4 e j t .

1d

jkx 2

De

jkx 3

Here, k represents the wave number in the ambient fluid and


an e j t sign convention has been adopted. The four com1134

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

x0

T 11

T 12 P

T 21 T 22 V

xd

In Eq. 3, P is the exterior sound pressure and V is the


exterior normal acoustic particle velocity. The pressures and
particle velocities on the two surfaces of the porous layer
may easily be expressed in terms of the positive- and
negative-going plane wave component amplitudes, i.e.,
P x0 AB,

4a

AB
,
0c

4b

P xd Ce jkd De jkd ,

4c

Ce jkd De jkd
,
0c

4d

V x0

V xd

where 0 is the ambient fluid density and c is the ambient


sound speed. Thus, when the plane wave components are
known based on measurements of the complex pressures at
four locations, the pressures and the normal particle velocities at the two surfaces of the porous layer can be determined.
It is then of interest to determine the elements of the
transfer matrix since, as will be shown below, the elements
B. H. Song and S. Bolton: Porous material properties

1134

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

of that matrix may be directly related to the properties of the


porous layer. However, first note that Eq. 3 represents two
equations in four unknowns: T 11 , T 12 , T 21 , and T 22 . Thus,
two additional equations are required in order to be able to
solve for the transfer matrix elements. Those equations may
be generated, for example, by making a second measurement
at the four microphone positions after changing the impedance terminating the downstream section of the standing
wave tube; that approach is the basis of the so-called twoload method.25
However, instead of making a second set of measurements it is possible to take advantage of the reciprocal nature
of a layer of homogeneous and isotropic porous material to
generate two additional equations. Pierce noted that reciprocity requires that the determinant of the transfer matrix be
unity.31 Ingard has noted that the latter constraint is a general
property of passive, linear four-pole networks.16 Allard has
also shown that this condition follows directly from the requirement that the transmission coefficient of a planar, arbitrarily layered acoustical system be the same in both
directions.32 Further, Pierce notes that for symmetrical systems, T 11T 22 . It may easily be shown that the latter condition follows when the plane wave reflection coefficients
from the two surfaces of a planar, layered system are the
same see the Appendix. Thus, given reciprocity and symmetry, it follows that
T 11T 22 ,

5a

T 11T 22T 12T 211.

5b

The latter two constraint conditions complete the set of four


equations necessary to solve for the transfer matrix elements.
By combining Eqs. 3 and 5, the transfer matrix elements can be expressed directly in terms of the pressures and
velocities on the two surfaces of the porous layer, i.e.,
T 11
T 12
T 21

P xd V xd P x0 V x0
,
P x0 V xd P xd V x0
2
2
P xd
P x0

P x0 V xd P xd V x0
2
2
V x0
V xd

P x0 V xd P xd V x0

P xd V xd P x0 V x0
.
T 22
P x0 V xd P xd V x0

When the incident plane wave is assumed to have unit amplitude, the sound pressures and particle velocities on the
two surfaces of the porous layer become
P x0 1R,

7a

1R
,
0c

7b

V x0

P xd Te jkd ,

7c

Te jkd
,
0c

7d

V xd

where RB/A and TC/A are the plane wave reflection


and transmission coefficients, respectively. When Eqs. 7
are substituted into Eq. 3, the normal incidence pressure
transmission and reflection coefficients for the case of an
anechoic termination, T a and R a , respectively, can be expressed as
T a

2e jkd
,
T 11 T 12 / 0 c 0 cT 21T 22

R a

T 11 T 12 / 0 c 0 cT 21T 22
.
T 11 T 12 / 0 c 0 cT 21T 22

In contrast, when the porous layer of depth d is positioned


against a rigid backing, V xd 0. Then, when the latter condition and Eqs. 7a and 7b are substituted into Eq. 3, the
normal incidence reflection coefficient for the hard-backing
case, R h , is obtained as
R h

T 11 0 cT 21
.
T 11 0 cT 21

10

Similar expressions may easily be derived for the case of a


porous layer backed by an arbitrary impedance.

6a

D. Calculation of wave number and characteristic


impedance

6b

Next, note that the normal incidence transfer matrix for


a finite depth layer of a homogeneous, isotropic porous material that can be considered to be either limp or rigid is16,32

6c
6d

Once the transfer matrix elements are known, all of the other
acoustical properties of a porous layer, e.g., the reflection
and transmission coefficients, can be calculated, as will be
demonstrated next.

T 11

T 12

T 21 T 22

cos k p d

j p c p sin k p d

j sin k p d/ p c p

cos k p d

11

In Eq. 11, k p is the wave number in the acoustical material,


d is the layer thickness, and p c p is the characteristic impedance of the material. Thus, the four transfer matrix elements
may be directly associated with the acoustical properties of
the porous material. In particular, the wave number can be
evaluated either as
1
k p cos1 T 11 ,
d

12a

1
k p sin1 T 12T 21,
d

12b

or
C. Calculation of reflection and transmission
coefficients

For example, consider a porous layer of depth d backed


by an anechoic termination, so that it can be assumed that D
is negligible compared to C in the downstream tube section.
1135

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

and the characteristic impedance of the acoustical material


can be calculated most directly as
B. H. Song and S. Bolton: Porous material properties

1135

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

pc p

T 12
.
T 21

13

Quantities such as the complex sound speed, c p /k p , and


complex density, p p c p /c p , can easily be determined
when k p and p c p are known.
III. EXPERIMENTAL PROCEDURE

The experimental procedure used in the present work


was based very closely on the earlier work by Bolton et al.,2
and will be summarized only briefly here. The measurements
were made using a modified standing wave tube the Bruel &
Kjr two-microphone impedance measurement tube type
4206. A loudspeaker at one end of the tube was used to
generate a broadband random signal over the frequency
range 100 to 6400 Hz and the transfer function between the
input signal to the loudspeaker and the microphone output at
each of the four measurement positions was measured over
that band by using a Bruel and Kjr 2032 frequency analyzer. Note that the Bruel and Kjr two-microphone impedance measurement tube type 4206 is suitable for both lowfrequency 100 to 1600 Hz and high frequency 500 to 6400
Hz measurements. In the work presented here, only the
2.9-cm inner diameter, high-frequency tube was used, so that
the results presented, with one exception, are confined to the
high-frequency range. Note also that a single-microphone
approach was adopted here as a matter of convenience: i.e., a
two-channel analyzer could be used to make the measurements and there was no need to phase-calibrate the microphone. Given access to four phase-matched microphones and
a multichannel data acquisition procedure, it would not be
difficult to implement a multimicrophone version of the
present procedure. Note that when using the singlemicrophone approach, a greater number of averages should
be used when estimating the transfer functions than might
normally be the case in a four-microphone measurement.2,33
In addition, it is necessary to compensate for the propagation
delay between the loudspeaker and the various microphones
to avoid the introduction of a time-delay bias error.34 Note
finally that it may be desirable in future implementations to
base the single-microphone procedure on the measurement
of transfer functions between a fixed microphone placed near
the sample and the roving microphone placed at each of the
four measurement positions in turn.34 The latter approach
ensures that loudspeaker nonlinearities do not affect the results.
In the present application, the standing wave tube was
modified by the addition of a second high-frequency microphone holder and the first section of a high-frequency sample
holder to create a new downstream section: see Fig. 2. The
downstream section was separated from the upstream section
by a sample holder of the same internal diameter as the upand downstream sections. The two measurement positions in
the up- and downstream tube sections were separated by a
distance of 2 cm; measurement position 2 was in all cases
21.2 cm from the front surface of the sample, and measurement position 3 was 13.6 cm downstream of the rear surface
of the sample when 7.5-cm-deep samples were used; this
distance increased to 16.1 and 18.6 cm, respectively, when
1136

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

FIG. 2. Schematic of the measurement setup.

5.0-cm and 2.5-cm-thick samples were used. Note that recent


work suggests that there is a frequency-dependent, optimum
microphone separation for multimicrophone measurements
of the type reported here.35 The latter effect should be investigated in the future to ensure the selection of an appropriate
microphone separation given the frequency range of interest.
A subsequent structural modal analysis of the sample
holder used in the present work revealed that its first two
natural frequencies fell within the frequency range of interest: at approximately 4.6 and 4.9 kHz. The first mode involved a longitudinally uniform ovalling of the sample
holder; in the second mode the ovalling was out of phase at
the two ends of the sample holder. Both of these motions can
presumably couple well with the first, higher-order acoustical circumferential mode within the sample contained within
the holder. It is believed that this effect accounted for some
of the anomalies in the measured results that are visible between 4 and 5 kHz; see Sec. IV. In future work the sample
holder will be stiffened to raise its first natural frequencies
above the frequency range of interest.
The downstream section was terminated by an approximately anechoic termination that was created by loosely
packing the standard sample holder with 66 cm of 3M Thinsulate sound-absorbing material. The absorption coefficient
of this termination was greater than 0.97 in the frequency
range of interest 500 to 6400 Hz.2 Note that an anechoic
termination is not required by the measurement procedure
used here; in principle, the termination impedance can be
arbitrary, and its value need not be known: see Sec. V B.
However, the presence of the anechoic termination causes
the sound field in the downstream section to be almost purely
propagational, thus minimizing the possibility of error in the
downstream transfer function estimates. Note also that when
the downstream termination impedance is known either a
priori e.g., if there is a rigid termination or by independent
measurement, the relation between the coefficients C and D
in Eqs. 1 can be determined. In that case only a single
downstream sound-pressure measurement would be required,
and the current procedure would become, in effect, a broadB. H. Song and S. Bolton: Porous material properties

1136

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

TABLE I. The physical properties of the glass fiber materials.

Sample

Bulk density
kg/m3

Flow resistivity
MKS Rayls/m

Std. dev. of flow resistivity


MKS Rayls/m

A
B

6.7
9.6

1.5104
2.4104

1.4103
2.5103

band implementation of the Champoux and Stinson technique described above.18


The material tested in the present work was aviationgrade glass fiber; the properties of the materials used are
listed in Table I. The sheets of glass fiber from which the
samples were cut were nominally 2.5 cm thick, and cylindrical samples of the lining material were carefully cut to fit
snugly inside the sample holder. The measurements reported
here were made using either three, two, or one layers of
lining material to give total sample thicknesses of 7.5, 5.0, or
2.5 cm, respectively. Lining materials were carefully inserted
into the sample holder so that the depth of the individual
layers comprising the complete sample was preserved. To
reduce the effect of mounting sample errors and material
variability, in each case ten different measurements were
performed using a total of 30, 2.5-cm-deep samples in the
case of 7.5-cm total depth, and correspondingly fewer pieces
for the smaller total sample depths; the results of the ten
individual tests were then averaged to give the results presented here.

proximation to the transmission loss since the magnitude of


the plane wave component reflected from the termination, D,
although small, is still finite.2 Note also that the results are
plotted here from 100 to 6400 Hz. Although the results are
not certain to be accurate below 500 Hz owing to the 2-cm
intermicrophone spacing, they are believed to be accurate in
character owing to the low reactivity of the sound field in
both the up- and downstream segments of the standing wave
tube.
While the transmission loss of both materials increases
with increasing frequency above 1 kHz as would be expected, note that the transmission loss also increases with
decreasing frequency below a minimum near 400 Hz
sample A and 500 Hz sample B. This behavior is typical
of the effect of sample-edge constraint on the normal incidence transmission loss of an elastic porous material,2,36,37 an
effect first noted by Beranek in the context of normal incidence absorption measurements.5 This effect is primarily significant at low frequencies owing to the strong viscous coupling between the solid and fluid phases of the porous
material in that region. In the present case, it was concluded
that the sample-edge constraint caused the measured results
to differ from those of a laterally infinite plane sheet at frequencies below 1 kHz. It is expected that the edge-constraint
effect would be mitigated by the use of larger diameter
samples; this suggestion will be investigated in future work.
B. Transfer matrix

IV. RESULTS AND DISCUSSION


A. Transmission loss

Consider first the normal incidence transmission loss of


7.5-cm depths of the two materials; see Fig. 3. This quantity
was calculated as TL20 logC/A and represents an ap-

The magnitudes and phases of the averaged transfer matrix elements for the two materials considered here, calculated using Eqs. 6, are plotted in Fig. 4. Each of the elements is significant in this case, and each has a particular
physical meaning. Note first that the elements T 11 and T 22
are identical owing to the constraint, Eq. 5a, imposed by

FIG. 3. Normal incidence transmission loss of lining materials A and B. Solid line: sample A; dashed line: sample B. Dotted lines indicate one standard
deviation.
1137

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

B. H. Song and S. Bolton: Porous material properties

1137

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

FIG. 4. Transfer matrix elements for samples A and B. Dotted lines indicate one standard deviation. a Magnitude, and b phase. Solid line: sample A;
dashed line: sample B.

symmetry. Further, the element T 11 is the ratio of the upstream and downstream pressures in the case of a zero velocity state at the downstream layer surface, and it is thus
dimensionless. The element T 12 is the ratio of the upstream
pressure and downstream velocity when a zero-pressure state
exists at the downstream surface of the sample, and thus it
has the units of impedance. Conversely, T 21 represents the
ratio of the upstream velocity and the downstream pressure
1138

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

when the downstream surface velocity state is zero; i.e., for


the case of a hard-backing boundary condition at the downstream surface of the sample. The element T 21 thus has the
units of admittance.
C. Wave number and characteristic impedance

The wave numbers and normalized characteristic impedances of the materials considered here are plotted in Figs. 5
B. H. Song and S. Bolton: Porous material properties

1138

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

FIG. 5. Wave number within lining material determined by using transfer matrix method and predicted using semiempirical formulas: a sample A; b
sample B.

and 6, respectively. Predictions made using the Delany and


Bazley semiempirical formulas38 for fibrous materials are
also plotted. It can be seen that the measured and predicted
wave numbers are in good agreement over the entire frequency range. The measured characteristic impedances agree
reasonably well with the predictions except at low frequencies; it is believed that the low-frequency discrepancy results
from the constraint of the sample around its edges, as discussed above.
1139

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

It is also of interest to plot the wave number components


in a more immediately meaningful form. For example, the
phase speed in the porous material, c ph , can be expressed in
normalized form as

c ph

,
c
cp

14

where p is the real part of k p . The normalized phase speeds


B. H. Song and S. Bolton: Porous material properties

1139

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

FIG. 6. Normalized characteristic impedance of lining material determined by using transfer matrix method and predicted using semiempirical formulas: a
sample A; b sample B.

are plotted in Fig. 7, where it can be seen that there is reasonable agreement between the measurements and the
Delany and Bazley prediction except at the lowest frequencies, for the reasons discussed above.
Note that, as expected, the phase speed is reduced in the
porous medium compared to that in air. As a result, higherorder duct modes may cut on within the porous material
at frequencies lower than they would in air, approximately
1140

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

by the ratio of the sound speeds although in a dissipative


medium there is, of course, no longer a strict distinction
between propagating and nonpropagating modes. In the
present instance, these modes may be closely coupled, in the
frequency range about 4 kHz, with the sample holder modes
that were noted in Sec. III. This effect may be responsible for
the feature visible in T 12 see Fig. 3 above 4 kHz for both
materials, as well as for the features in the real part of the
B. H. Song and S. Bolton: Porous material properties

1140

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

FIG. 7. Normalized phase speed and attenuation per wavelength determined by using transfer matrix method and predicted using semiempirical formulas: a
sample A; b sample B.

characteristic impedance in the same frequency range; see


Fig. 6.
The acoustical dissipation within a porous material may
be expressed in terms of the attenuation per wavelength,
which is also plotted in Fig. 7. This quantity is calculated as
p p where p is the imaginary part of the wave number in
the porous material, and p f /c ph where f is the frequency
is the wavelength within the material. The attenuation per
wavelength may thus be calculated as
1141

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

p p 2 p / p .

15

It may be seen that there is good agreement between the


measurements and predictions at frequencies above 1 kHz.

D. Complex density and sound speed

As mentioned in the Introduction, homogeneous and isotropic limp or rigid porous materials may be modeled as
B. H. Song and S. Bolton: Porous material properties

1141

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

FIG. 8. Normalized complex density determined by using transfer matrix method and predicted using semiempirical formulas: a sample A; b sample B.

fluids having complex properties. Thus, arbitrarily shaped


porous material domains may be modeled by using existing
finite and boundary element codes so long as provision has
been made for complex material properties; see Ref. 39 for a
recent example of this approach. The input data most often
required by these programs are complex density and the
complex sound speed within the porous material. The complex density can be calculated in normalized form as
1142

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

p pc p k p

,
0
0

16

and the normalized complex sound speed as

cp

.
c
k pc
B. H. Song and S. Bolton: Porous material properties

17
1142

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

FIG. 9. Normalized complex sound speed determined by using transfer matrix method and predicted using semiempirical formulas: a sample A; b
sample B.

The normalized complex densities of the present materials are shown in Fig. 8, where it can be seen that there is
good agreement with the theoretical prediction at frequencies
above 1 kHz. Note that the complex density used here is a
bulk density in contrast to the pore-based complex density
referred to in Ref. 32: the two quantities differ by a factor of
porosity. The normalized complex sound speeds are shown
1143

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

in Fig. 9, where again it can be seen that there is good general agreement with the theoretical predictions.
E. Prediction of reflection coefficient for the hardtermination case

Once the transfer matrix of a porous layer is known, the


performance of the latter may be predicted in a variety of
B. H. Song and S. Bolton: Porous material properties

1143

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

FIG. 10. Reflection coefficients in the hard-termination case: a sample A; b sample B.

applications. For example, the reflection coefficient of a porous layer positioned in front of either an anechoic or rigid
termination can be calculated based on a knowledge of the
transfer matrix elements by using Eqs. 9 and 10, respectively. In Fig. 10, the magnitudes of the directly measured
reflection coefficients for the two materials in the hardtermination configuration evaluated as R h B / A ) are
compared with the predicted ones based on the measured
transfer matrix elements.
The measured results shown are the average of ten indi1144

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

vidual measurements of 7.5-cm-deep layers assembled from


the same 30, 2.5-cm-deep samples used in the transfer matrix
measurements. The results were not expected to be particularly accurate below 1000 Hz owing to the sample edgeconstraint effect described above, nor above 4 kHz owing to
the effect of sample holder modes. Nonetheless, the predictions based on the measured transfer matrix elements faithfully capture the character of the directly measured results.
Note that once the wave number and characteristic impedance of a particular porous material are known, predicB. H. Song and S. Bolton: Porous material properties

1144

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

FIG. 11. Wave number within lining material determined by using transfer matrix method in combination with different sample thicknesses, as noted: a
sample A; b sample B.

tions of the reflection or transmission coefficients of an arbitrary thickness of that material may be made by using the
transfer matrix approach. In addition, the transfer matrix approach may be extended to non-normal incidence cases to
predict the oblique incidence behavior of sheets of porous
materials, either alone or in combination with other acoustical elements.3,16,32
1145

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

V. VERIFICATION STUDIES
A. Effect of sample depth

Finally, it was of interest to perform a sequence of experiments to verify the capabilities of the transfer matrix procedure described here. First, the effect of sample depth will
B. H. Song and S. Bolton: Porous material properties

1145

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

FIG. 12. Normalized characteristic impedance of lining material determined by using transfer matrix method in combination with different sample thickness:
a sample A; b sample B.

be demonstrated. The wave number and characteristic impedance are both inherent properties of a homogeneous fibrous material, and thus, if the proposed procedure works
properly, the estimated values of these quantities should not
depend on the depth of the sample under test. Here, measurements were made using ten 5-cm-deep samples and ten 2.5cm-deep samples of both materials A and B: those results
were compared with the estimates made using the 7.5-cm1146

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

deep samples that were presented earlier. Shown in Fig. 11


are the averaged wave numbers estimated using the three
sample depths. It may be seen that the propagation constants
are essentially identical in the 5.0- and 7.5-cm cases, but that
there is a slight discrepancy in the 2.5-cm case. It is possible
that the latter discrepancy resulted from leakage around the
edge of the sample; it was difficult to mount the relatively
thin and light 2.5-cm samples in the sample holder without
B. H. Song and S. Bolton: Porous material properties

1146

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

FIG. 13. Normal incidence transmission losses of lining material A estimated by using the transfer matrix method in combination with three different
termination conditions.

creating small gaps around the edge of the samples. Leakage


appeared not to be a problem with the deeper samples. Effects of sample inhomogeneity are also likely to be exaggerated when using thin samples since a smaller total amount
of material is involved in ten 2.5-cm-deep samples than in
ten 5.0-cm-deep samples, etc.. The characteristic impedance
results are shown in Fig. 12. It may be seen that there was
good agreement between the estimated values in all three
cases although, once again, the estimates based on the
2.5-cm samples deviate slightly from the other results. However, taken together, the wave number and characteristic impedance results do indicate that estimates of the fundamental
properties of the sample materials were independent of
sample depth, as expected, except, that is, when relatively
thin samples were used, in which case edge leakage and
intersample inhomogeneities may have affected the results.
B. Effect of tube-termination conditions

It was pointed out earlier that a knowledge of the tubetermination impedance is not required when using the
present approach: it was of interest to confirm this feature of
the transfer matrix method. Second, it was of interest to illustrate the errors introduced in transmission loss estimates
when no account was made of reflections from the termination. Measurements of the transmission loss of the 7.5-cm
depths of sample A made using the transfer matrix procedure
in combination with three termination conditions nearly
anechoic as described in Sec. III, rigid and open are shown
in Fig. 13. It may be seen that the differences between the
three estimates are small even when the termination conditions are dramatically different. Noticeable discrepancies occur below 500 Hz, which, however, is the lower limit of the
frequency range over which results are known to be accurate
1147

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

when using the present microphone spacing. In addition, in


the case of the hard termination, there is a small oscillatory
error in the transmission loss results at high frequencies, presumably resulting from the highly reactive standing wave
field in the downstream tube section in this case. The results
shown in Fig. 14 indicate that neither the estimates of the
wave number or characteristic impedance are significantly
affected by the tube termination conditions. Together, these
results support the claim that the transfer matrix procedure
results are essentially independent of termination conditions,
at least in application to fibrous material properties, of the
type considered here.
Shown in Fig. 15 are estimates of the transmission loss
of a 7.5-cm depth of sample A based simply on the use of
T C / A : i.e., no accounting was made of the reflection
from the tube termination. The results shown are the average
of ten individual measurements. It can be seen that the estimated transmission losses can be significantly in error in
these cases: only the anechoic termination case approaches
the results obtained using the transfer matrix method.

C. Comparison with two-load method

Finally, it was of interest to compare results obtained


using the proposed transfer matrix method and the two-load
method.25 The two-load method was implemented using the
same approach as for the transfer matrix method, except that
measurements at the four microphone locations were made
in sequence for two termination conditions: nearly anechoic
and rigid. Those data were then used in conjunction with Eq.
3, written twice, to solve for the transfer matrix elements.
In the two-load measurements, the 7.5-cm-deep samples of
material A were used. The results of ten individual measureB. H. Song and S. Bolton: Porous material properties

1147

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

FIG. 14. a Wave number and b normalized characteristic impedance estimated for material A by using the transfer matrix method in combination with
three different termination conditions.

ments were averaged as before. It may be seen from Fig. 16


that the two procedures yield essentially identical estimates
of the transmission loss except at frequencies below which
the present intermicrophone spacing is not guaranteed to
yield accurate results. Close agreement was also found when
estimating the wave number and characteristic impedances,
1148

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

as may be seen from Fig. 17. Note, however, that the twoload results are distorted by a small oscillatory error at high
frequencies that presumably results from the highly reactive
downstream sound field in the rigid termination case. Thus, it
may be concluded that the transfer matrix method presented
here yields results that are of the same or better quality than
B. H. Song and S. Bolton: Porous material properties

1148

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

FIG. 15. Normal incidence transmission losses of lining material A estimated using the ratio C / A for three different termination conditions.

those of the two-load method, even though only half the


number of measurements are required when using the former
procedure.
VI. CONCLUSIONS

The intention of the present work was to develop a quick


and convenient method for determining the fundamental
acoustical properties i.e., the wave number and characteristic impedance of commonly used porous materials. The procedure described here was based on well-known transfer ma-

trix methods, but the number of measurements required was


reduced compared to earlier methods by taking advantage of
the reciprocal nature of sound transmission through homogeneous and isotropic porous layers. The kind of information
that can be derived from the present method, e.g., complex
density and sound speed, is the information required to
model limp or rigid porous materials in finite or boundary
element procedures, and so the present procedure may prove
a useful adjunct to modern noise-control design procedures.
Note, finally, that elastic porous materials such as foams
can support two longitudinal waves and a single transverse

FIG. 16. Comparisons of the normal incidence transmission loss of sample A estimated by using the transfer matrix method and two-load method.
1149

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

B. H. Song and S. Bolton: Porous material properties

1149

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

FIG. 17. a Wave number, and b normalized characteristic impedance for material A estimated by using the transfer matrix method and two-load method.

wave.40 Although a single airborne wave is often dominant


in unfaced noise-control foams, in which case they can be
modeled as effective fluids see Ref. 39, for example, the
other two wave types may sometimes contribute very significantly, depending on the details of the foams surface boundary conditions.41 Thus, in principle, a transfer matrix representation of a foam layer is much more complicated than that
presented here as Eq. 10.16,32 The reader should therefore
be cautious when applying the present method to foams.
1150

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

ACKNOWLEDGMENTS

This paper was written while the authors were visiting


the Center for Noise and Vibration Control NOVIC in the
Department of Mechanical Engineering of the Korea Advanced Institute of Science and Technology. We are grateful
for the hospitality extended by the NOVIC faculty, students
and staff. The second author extends his thanks, in particular,
to Professor Yang-Hann Kim and Professor Jeong-Guon Ih
B. H. Song and S. Bolton: Porous material properties

1150

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

for arranging his visit and helping to make it productive, and


to the Korea Research Foundation for providing financial
support. We are also grateful to Joseph Pope, who suggested
the use of a four-microphone method to measure the transmission loss of porous materials and to David Apfel of L &
L Products of Romeo, Michigan, whose initial financial support helped to make this work possible. Finally, our thanks
go to Robert Schlinker of the United Technologies Research
Center, who arranged the financial support that made it possible to complete the present work.

APPENDIX A: IMPLICATIONS OF RECIPROCAL


SOUND TRANSMISSION AND REFLECTION

In this Appendix, normal incidence sound transmission


through, and sound reflection from, an arbitrarily layered
medium represented by a two-by-two transfer matrix are
considered. The approach followed is similar to that of Allard, who previously considered the implications of reciprocal sound transmission.32 The configurations considered are
shown in Fig. A1, where the two surfaces of the layered
system are labeled A and B. Note that the only difference
between the two configurations is the location of the
x-coordinate origin which is shifted in case b with respect
to case a as a matter of convenience. In case a, a unit
amplitude plane wave is assumed to be incident on surface
A, and the reflection and transmission coefficients are denoted R 1 and T 1 , respectively. In case b, a plane wave is
incident on surface B, and the reflection and transmission
coefficients are similarly denoted R 2 and T 2 .
In case a, the sound pressures and normal particle velocities at surfaces A and B are related by


P A

V A

T 12 P B

T 11

T 21 T 22 V B

FIG. A1. Sound transmission and reflection cases a and b.

V B

R 1

T 11 T 12 / 0 c T 21 0 cT 22
,
T 11 T 12 / 0 c T 21 0 cT 22

A11

T 1

2e jkd
,
T 11 T 12 / 0 c T 21 0 cT 22

A12

R 2

T 11 T 12 / 0 c T 21 0 cT 22
,
T 11 T 12 / 0 c T 21 0 cT 22

A13

T 2

2e jkd T 11T 22T 12T 21


.
T 11 T 12 / 0 c T 21 0 cT 22

A14

where
P A 1R 1 ,

A2

1R 1
,
0c

A3

V A

P B T 1 e

jkd

A4

T 1 e jkd
V B
.
0c

A5

In case b, the sound pressures and normal particle velocities at surfaces A and B are related by


P A

V A

T 11 T 12 P B
T 21

T 22 V B

A6

It can be seen by comparing Eqs. A11 and A13 that when


R 1 R 2 , i.e., when the layer is symmetrical in the x direction, it follows that
T 11T 22 .

T 11T 22T 12T 211.


A7

jkd

T 2e
V A
0c

P B 1R 2 ,
1151

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

A8
A9

A15

Similarly, by comparing Eqs. A12 and A14 it is clear that


when T 1 T 2 , i.e., when the sound transmission through the
layer is reciprocal, it follows that

where, in this case


P A T 2 e jkd ,

A10

When Eqs. A2A5 and A7A10 are substituted into


Eqs. A1 and A6, respectively, it is possible to solve for
R 1 , T 1 , R 2 , T 2 in terms of the transfer function elements,
i.e.,

A1

R 2 1
.
0c

A16

K. Attenborough, Acoustical characteristics of porous materials, Phys.


Rep. 82, 179227 1982.
2
J. S. Bolton, R. J. Yun, J. Pope, and D. Apfel, Development of a new
sound transmission test for automotive sealant materials, SAE Trans., J.
Pass. Cars 106, 26512658 1997.
3
H.-Y. Lai, S. Katragadda, J. S. Bolton, and J. H. Alexander, Layered
B. H. Song and S. Bolton: Porous material properties

1151

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

fibrous treatments for sound absorption and transmission, SAE Trans., J.


Pass. Cars 106, 33113318 1997.
4
R. A. Scott, The absorption of sound in a homogeneous porous medium, Proc. Phys. Soc. London 58, 358368 1946.
5
L. L. Beranek, Some notes on the measurement of acoustic impedance,
J. Acoust. Soc. Am. 19, 420427 1947.
6
R. F. Lambert and J. S. Tesar, Acoustic structure and propagation in
highly porous, layered, fibrous materials, J. Acoust. Soc. Am. 73, 1231
1237 1984.
7
L. L. Beranek, Acoustical properties of homogeneous, isotropic rigid
tiles and flexible blankets, J. Acoust. Soc. Am. 19, 556568 1947.
8
C. Zwikker and C. W. Kosten, Sound Absorbing Materials Elsevier, Amsterdam, 1949.
9
M. A. Ferrero and G. G. Sacerdote, Parameters of sound propagation in
granular absorption materials, Acustica 1, 135142 1951.
10
S. L. Yaniv, Impedance tube measurement of propagation constant and
characteristic impedance of porous acoustical material, J. Acoust. Soc.
Am. 54, 11381142 1973.
11
C. D. Smith and T. L. Parott, Comparison of three methods for measuring acoustic properties of bulk materials, J. Acoust. Soc. Am. 74, 1577
1582 1983.
12
Anonymous, Standard test method for impedance and absorption of
acoustical materials using a tube, two microphones, and a digital frequency analysis system, ASTM Standard E 1050-90 1990.
13
H. Utsuno, T. Tanaka, T. Fujikawa, and A. F. Seybert, Transfer function
method for measuring characteristic impedance and propagation constant
of porous materials, J. Acoust. Soc. Am. 86, 637643 1989.
14
C. Bordone-Sacerdote and G. G. Sacerdote, A method for measuring the
acoustic impedance of porous materials, Acustica 34, 7780 1975.
15
K. U. Ingard and T. A. Dear, Measurement of acoustic flow resistance,
J. Sound Vib. 103, 567572 1985.
16
K. U. Ingard, Notes on Sound Absorption Technology Noise Control
Foundation, Poughkeepsie, NY, 1994.
17
J. D. McIntosh, M. T. Zuroski, and R. F. Lambert, Standing wave apparatus for measuring fundamental properties of acoustic materials in
air, J. Acoust. Soc. Am. 88, 19291938 1990.
18
Y. Champoux and M. Stinson, Measurement of the characteristic impedance and propagation constant of materials having high flow resistivity,
J. Acoust. Soc. Am. 90, 21822191 1991.
19
T. Iwase, Y. Izumi, and R. Kawabata, A new measuring method for
sound propagation constant by using sound tube without any air spaces
back of a test material, in Proceedings INTER-NOISE 98 CD-ROM
edition, Causal Productions, Australia, 1998.
20
P. Dean, An in situ method of wall acoustic impedance measurement in
flow ducts, J. Sound Vib. 34, 97130 1974.
21
C. K. Amedin, Y. Champoux, and A. Berry, Acoustical characterization
of absorbing porous materials through transmission measurements in a
free field, J. Acoust. Soc. Am. 102, 19821994 1997.
22
D. N. Johnston, D. K. Longmore, and J. E. Drew, A technique for the
measurement of the transfer matrix characteristics of two-port hydraulic
components, Fluid Power Sys. Tech. 1, 2533 1994.

1152

J. Acoust. Soc. Am., Vol. 107, No. 3, March 2000

23

M. Fukuda, A study on the exhaust muffler of internal combustion engine, Bull. JSME 6, 255269 1963.
24
T. Y. Lung and A. G. Doige, A time-averaging transient testing method
for acoustic properties of piping systems and mufflers with flow, J.
Acoust. Soc. Am. 73, 867876 1983.
25
M. L. Munjal, Acoustics of Ducts and Mufflers Wiley-Interscience, New
York, 1987.
26
A. G. Doige, M. L. Munjal, and H. S. Alves, An improved experimental
method for determining transfer matrices of pipeline elements with flow,
in Proceedings of NOISE-CON 88, pp. 481486 Noise Control Foundation, Poughkeepsie, NY, 1988.
27
M. L. Munjal and A. G. Doige, Theory of two source-location method
for direct experimental evaluation of the four-pole parameters of an aeroacoustic element, J. Sound Vib. 141, 323333 1990.
28
K. S. Peat, A transfer matrix for an absorption silencer element, J.
Sound Vib. 146, 353360 1991.
29
T. Tanaka, T. Fujikawa, T. Abe, and H. Utsuno, A method for the
analytical prediction of insertion loss of a two-dimensional muffler model
based on the transfer matrix derived from the boundary element method,
Trans. ASME 107, 8692 1985.
30
U. R. Kristiansen and T. E. Vigran, On the design of resonant absorbers
using a slotted plate, Appl. Acoust. 43, 3948 1994.
31
A. D. Pierce, Acoustics: An Introduction to its Physical Principles and
Applications McGraw-Hill, New York, 1981.
32
J. F. Allard, Propagation of Sound in Porous Media Elsevier Applied
Science, London and New York, 1993.
33
M. Kompella, P. Davies, D. J. Ufford, and R. J. Bernhard, A technique
to identify the number of incoherent sources contributing to the response
of a system, Mech. Sys. Sig. Proc. 8, 363380 1994.
34
J. Kruger and M. Quickent, Determination of acoustic absorber parameters in impedance tube, Appl. Acoust. 50, 7989 1997.
35
S.-H. Jang and J.-G. Ih, On the multiple microphone method for measuring induct acoustic properties in the presence of mean flow, J. Acoust.
Soc. Am. 103, 15201526 1998.
36
Y. J. Kang and J. S. Bolton, A finite element model for sound transmission through foam-lined double panel structures, J. Acoust. Soc. Am. 99,
27552765 1996.
37
Y. J. Kang and J. S. Bolton, Sound transmission through elastic porous
wedges and foam layers having spatially graded properties, J. Acoust.
Soc. Am. 102, 33193332 1997.
38
M. E. Delany and E. N. Bazley, Acoustic properties of fibrous absorbent
materials, Appl. Acoust. 3, 105116 1971.
39
A. F. Seybert, R. A. Seman, and M. D. Lattuca, Boundary element
prediction of sound propagation in ducts containing bulk absorbing materials, Trans. ASME, J. Vib. Acoust. 120, 976981 1998.
40
J. S. Bolton and Y. J. Kang, Elastic porous materials for sound absorption and transmission control, SAE Trans., J. Pass. Car 106, 25762591
1997.
41
J. S. Bolton, N.-M. Shiau, and Y. J. Kang, Sound transmission through
multi-panel structures lined with elastic porous materials, J. Sound Vib.
191, 317347 1996.

B. H. Song and S. Bolton: Porous material properties

1152

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 103.21.125.84 On: Mon, 11 Jul 2016 09:03:43

Вам также может понравиться