Non-Resonant Sound Transmission Through Double Walls Using Statistical Energy Analysis

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

Non-resonant sound transmission through

double walls using statistical energy analysis


Robert J.M. Craik*
Department of Building Engineering and Surveying, Heriot-Watt University, Edinburgh EH10 5LH, UK
Received 10 October 2001; received in revised form 24 June 2002; accepted 26 June 2002
Abstract
Although SEA is a suitable framework for predicting sound transmission through double
walls it has been found that the standard method of computing the non- resonant coupling
loss factor from a room to cavity underestimates the coupling. A revised model for computing
this coupling loss factor is presented which gives much better agreement with measured data. This
allows better predictions to be made of sound transmission through lightweight double walls.
# 2003 Elsevier Science Ltd. All rights reserved.
Keywords: Non-resonant sound transmission; Double walls; Statistical energy analysis
1. Introduction
In a recent paper [1] sound transmission through a series of lightweight double
walls was examined. This study showed that, above the mass-spring-mass resonance
of the two layers of the wall, the double wall acted as a series of independent com-
ponents comprising two layers of sheet material (typically of plasterboard), a frame
(typically of timber) and a cavity. It was shown that statistical energy analysis (SEA)
could be used to model these walls with each of the elements being modelled as a
subsystem. The frames were either modelled as subsystems or as plate elements
connecting the two plasterboard sheets depending on the way in which the plaster-
board was attached to the frame and in particular the nails spacing.
A typical system can be seen in Fig. 1 which shows the plan of a stud wall together
with the SEA model used in this paper. For simplicity, it is assumed that there is no
structural coupling between the two layers of plasterboard and no structural frame.
Applied Acoustics 64 (2003) 325341
www.elsevier.com/locate/apacoust
0003-682X/03/$ - see front matter # 2003 Elsevier Science Ltd. All rights reserved.
PI I : S0003- 682X( 02) 00051- 8
* Tel.: +44-131-4495111; fax: +44-131-4513161.
When this model was used to study sound transmission through real walls it was
found that non-resonant airborne transmission into the cavity and non-resonant
transmission out of the cavity was not modelled accurately [1].
Non-resonant transmission between two rooms through a panel has been long
understood and a variety of equations are available for computing this transmission
ranging from simple mass law equation [2] to more sophisticated theories taking into
account the nite nature of the plate [3,4]. When modelling sound transmission
through a double wall, sound is transmitted non-resonantly through each panel.
However, the transmission is either from or to a cavity that cannot be considered as
a semi-innite free space. This cavity is generally narrow compared to a wavelength
and therefore only supports modes with particle motion parallel to the wall.
The conventional method of calculating the coupling between a room and cavity,
given by Price and Crocker [5], has been to assume that transmission from a room to
cavity is the same as transmission from a room to a room. This gives the coupling
loss factor from a room to cavity (or second room), Z
13
, using the subsystem num-
bering of Fig. 1, as
j
13

c
0
S
2
t
2
8

fV
1
1
where S is the wall area, V is the room volume and t
2
is the non-resonant trans-
mission coecient of wall 2 [which can also be expressed as a sound reduction index,
Fig. 1. Two rooms separated by a double leaf wall with no structural connection together with an SEA
model of the system.
326 R.J.M. Craik / Applied Acoustics 64 (2003) 325341
R=10 log(1/t)]. The simplest expressions for R are based on the simple normal inci-
dence mass law as [2]
R
0
20log
o,
s
2,
0
c
0
_ _
2
where ,
s
is the surface density. A better, but still very simple estimate, is based on
eld incidence [2] as
R 10log
o,
s
2,
0
c
0
_ _
2
1

_ _
3
where the 1/ term provides a 5 dB correction to account for the dierence between
normal and eld incidence.
The coupling in the reverse direction from the cavity to room (j
35
and j
31
) can be
found from the consistency relationship that relates coupling between two sub-
systems in each direction to the ratio of the modal densities.
The consistency relationship states that
n
i
j
ij
n
j
j
ji
4
where the n is the modal density. The modal densities of the room and cavity are
respectively
n
room

4f
2
V
room
c
3
0
. n
cavity

2fl
x
l
y
c
2
0
5
where V is the room volume and l
x,y
are the cavity dimensions [6]. This gives the
coupling loss factor from the cavity to room as
j
31

t
2
4
6
assuming the cavity and wall have the same area.
It is found that results predicted by these equations do not agree well with mea-
sured data, particularly when the depth of the cavity is small. In such cases the
sound transmitted is greater than predicted.
This paper shows how a simple model can be used to derive an expression for the
coupling between a cavity and a room directly which gives much better agreement
with measured data.
2. Theory
Although the best models for non-resonant sound transmission through a wall are
based on the behaviour of nite plates the simplest models assume that the plate is
R.J.M. Craik / Applied Acoustics 64 (2003) 325341 327
innite, leading to the well known mass law equation for transmission through a
partition. This theory generally works well and so similar assumptions about the
innite size of the plate are assumed here.
Consider a system as shown in Fig. 2 in which a sound wave with acoustic pres-
sure p is travelling along a narrow cavity of depth d. One side of the cavity is
bounded by a plate (2) with surface density, ,
s
and on the other side of the plate is
free space [or a large room (1)].
As the wave is assumed to be travelling parallel to the plate there is no power
incident on the plate and therefore the conventional denition of sound transmission
through the wall based on the transmission coecient, which is the ratio of power
transmitted though a wall to the power incident on it, has no meaning.
The acoustic pressure in the cavity will act on the wall causing it to move so that
the wall acceleration, a, can be given by
p ,
s
a io,
s
v 7
Therefore the mean squared velocity of the wall can be given by
- v
2
>
- p
2
>
o
2
,
2
s
8
Strictly, the displacement of the wall will change the wavespeed of the acoustic
waves in the cavity but assuming that this change in wavespeed is small then the
(forced) waves in the wall will travel at the same speed as the waves in air, c
0
. The
revised wavespeed can be found by solving the wave equation that includes a term to
allow for movement of the enclosing wall [7, 8]. In addition this equation ignores the
boundary eect present in nite plates where in a rigid cavity there will be a max-
imum pressure at the boundary yet the plate velocity is constrained to be zero at a
supported edge.
These structural waves will radiate sound into the space above the plate. The
power radiated by a plate with a mean square velocity -v
2
> is given by
Fig. 2. Wave motion in a cavity is parallel to the wall surface and therefore no power is incident on the
wall. However, the pressure changes in the cavity cause the wall to move with a velocity, v.
328 R.J.M. Craik / Applied Acoustics 64 (2003) 325341
W
rad
,
0
c
0
- v
2
> S
2
o
0

,
0
c
0
- p
2
> S
2
o
0
o
2
,
2
s
9
The term o
0
is the radiation eciency. Since the waves in the plate are travelling at
the same speed as those in the cavity and room, coincidence will occur and so the
equation for the radiation eciency is that for the case where the frequency is equal
to the critical frequency, f
c
.
In SEA notation the power transmitted from the cavity into the room is by de-
nition
W
rad
E
3
oj
31
10
and the energy in the cavity is given by [6]
E
3

- p
2
3
> V
3
,
0
c
2
0
11
Equating Eqs. (9) and (10) gives the coupling loss factor as
j
31

,
2
0
c
3
0
S
2
o
0
o
3
,
2
s
V
3
12
which can be simplied, if the cavity has the same area as the plate to,
j
31

,
2
0
c
3
0
o
0
o
3
,
2
s
d
3
13
The coupling loss factor from the room to the cavity can be found by using the
consistency relationship [Eq. (4)] to give the coupling loss factor as
j
13

,
2
0
c
4
0
S
3
o
0
o
4
,
2
s
d
3
V
1
14
where S
3
is the area of the cavity (equal to l
x
l
y
).
A number of dierent equations are available to calculate the radiation eciency
at the critical frequency [9,10]. However, in view of the number of assumptions
already made it is probably sucient to use the simplest given by Cremer, Heckl and
Ungar [11] as
o 0.45

U
2
f
c
c
0
_
15
where U is the perimeter length of the plate. Since the wavespeed in the plate is always
equal to that in air the frequency of interest is always the critical frequency and so
R.J.M. Craik / Applied Acoustics 64 (2003) 325341 329
o
0
0.45

U
2
f
c
0
_
16
Alternative expressions for o
0
all have the same frequency dependance though the
value of the constant will change slightly.
The ratio of the new to old coupling loss factor from either room to cavity or
cavity to room is
j
new
j
old

,
2
0
c
4
0
o
0
S
3
,
2
s
o
4
d
3
V
1
8fV
1
c
0
S
2
t
17
which can be simplied by assuming S
2
=S
3
, taking t from Eq. (3) and o
0
as Eq. (16)
to give
j
new
j
old

c
0
o
0
od
3

c
0
U
2
f
_
0.45
2d
3
18
It can be seen that the two expressions for the CLF have dierent frequency
dependance which leads to dierent slopes of any level dierence calculations based
on these theories.
A more detailed derivation of the coupling can be carried out taking into account
the stiness and damping of the plate (in addition to the mass), which is important
above the critical frequency, and also taking into account other possible angles of
incidence to allow the transition from a 2 dimensional cavity to a 3 dimensional
cavity (or room) to be determined.
Following the general notation of Fahy [12] the wall impedance can be given by
Z
w
io,
s
1
f
2
f
2
c
sin
4

_ _
,
s
o
f
2
f
2
c
sin4 j 19
where is the angle of incidence (0 is normal incidence), f
c
is the critical frequency
and j is the loss factor of the panel. The uid impedance seen by the plate is
Z
f
,
0
c
0
o 20
For an innite plate o=1/cos() so that for grazing incidence the uid impedance
is innite and no power is radiated. The total impedance seen by the wave is there-
fore the impedance of the wall plus the uid load on either side
Z j j
2
Z
w
2Z
f
j j
2
21
and t can be given by t 4 Z
f

2
, Z j j
2
.
The walls considered in this paper are nite and hence even at grazing incidence
(where f=f
c
) o and hence | Z
f
| is nite. For the structures considered here the uid
loading is not important and | Z |%| Z
w
|. The wall velocity is then
330 R.J.M. Craik / Applied Acoustics 64 (2003) 325341
- v
2
> [
- p
2
>
Z j j
2
22
and is directly comparable to Eq. (8). If the incident wave is a grazing wave when [
is 1 as in the simpler derivation. If the incident wave is not grazing then there will be
a reected wave and the mean squared surface pressure will be double the mean
squared average pressure so that [ will be 2. If f -f
c
then | Z | is approximately o,
s
giving Eq. (8). If f > f
c
then | Z | increases with frequency reducing the wall vibration
and transmitted sound.
The coupling loss factor can be determined in exactly the same way as before
except that if the waves are not grazing then the radiation eciency is 1/cos(). The
coupling loss factor (if all the energy is contained in one mode at one angle of inci-
dence) is
j
31

[,
2
0
c
3
0
S
2
o
Z j j
2
V
3
o
23
The value of [ will be either 1 or 2 depending on whether the waves are grazing or
not and o will be a function of angle having a maximum value [given by Eq. (15)] at
grazing incidence.
If a cavity mode, given by
f
r.s.t

c
0
2
r
2
l
2
x

s
2
l
2
y

t
2
d
2
_ _
1,2
24
where r, s, and t are integers, lies within a frequency band of interest then the angle
of incidence will be
cos
1
tc
do
_ _
25
and the band average coupling loss factor (assuming equipartition of modal energy)
of n such modes will be
j
31

1
n

n
[,
2
0
c
3
0
S
2
o
Z j j
2
V
3
o
26
If only grazing incidence occurs then Eq. (26) is the same as Eq. (14) if | Z | is
replaced by o,
s
. In the limit as the sound eld becomes diuse, the contribution of
the grazing modes becomes insignicant and the average summation can be replaced
by an integral. The coupling then becomes the equation for coupling between two
rooms, Eq. (1).
Coupling from the room to cavity can be found using the consistency relationship
[Eq. (4)] as before.
R.J.M. Craik / Applied Acoustics 64 (2003) 325341 331
3. Discussion
An example of transmission from a cavity to a room can be seen in Fig. 3 using
dierent calculation methods. The cavity is 3 4 m with a depth of 150 mm. This
gives the rst cross mode (f
x
) at 1140 Hz. The wall separating the cavity from the
room has a surface density of 10 kg/m
2
, a critical frequency of 5000 Hz and a loss
factor of 0.01.
At low frequencies it can be seen that there is excellent agreement between the
simple Eq. (12) and the calculation based on the average contribution of individual
modes [Eq. (26)] conrming that the eects of stiness and damping can be ignored.
This approximation is valid up about 1 octave below f
c
where the error is about 3
dB. This equation has a dierent slope from the Price and Crocker equation and in
this case the two curves cross at 250 Hz.
Above the rst cross mode in the cavity the slope of the Price and Crocker equa-
tion changes as the cavity is now modelled as a room and the room equation is used
to calculate the modal density of the cavity so that the coupling loss factor is the
same as the standard room to room Eq. (1). Although oblique modes are now pre-
sent in the cavity their number is initially relatively small and the band average
coupling is still dominated by the grazing modes. In this example, below the critical
Fig. 3. Coupling loss factor from a cavity to a room (f
x
=1140 Hz, f
c
=5000 Hz). , approximation of
Eq. (12); - - - -, Price and Crocker method; &, coupling from the sum of all modes [Eq. (26)];
. . . .
, contribution of modes with grazing incidence.
332 R.J.M. Craik / Applied Acoustics 64 (2003) 325341
frequency the grazing modes have stronger coupling than the oblique modes and there-
fore including the oblique modes in the summation reduces the band average value.
Above the critical frequency the coupling reduces sharply particularly for the
grazing incidence waves.
In these results no attempt has been made to separate transmission into resonant
and non-resonant transmission. Above the critical frequency resonant transmission
will dominate and the non-resonant contribution (which is the path of interest) can
be ignored. It is concluded that Eq. (12) is a reliable estimator of the coupling below
the critical frequency but that if the rst cross mode lies below the critical frequency
there will be a transition from Eq. (12) to Eq. (1) which could extend over 2 octaves.
It is interesting to note that if radiation into a cavity is the same as radiation into a
room then the equations determined above [such as Eq. (12)] would also give the
coupling from one cavity to another in a multi-layer wall such as a triple layer wall.
However, as the CLF is a function of cavity depth and the modal density is not so
the consistence relationship of Eq. (4) would not be satised. It is not clear if this is
because of some approximation in the calculations of the wall response or of the
assumption of equality of the radiation eciency.
Although the theory described above was developed for transmission into and out
of a cavity there are other cases where grazing incidence occurs where this theory
may be applied. One important case is in transportation noise where noise is generated
under a car or a train and then passes up the side of the vehicle and is transmitted
through the windows due to sound grazing over the glass. When using SEA to
model this type of problem it is usually assumed that the sound pressure level that is
measured exists as a diuse sound eld since this then allows the standard room to
room Eq. (1) to be used to compute the sound transmitted through the window. The
theory given in this paper should allow more accurate modelling of that transmission
by accounting for the fact that the sound waves are parallel to the window with the
coupling computed from Eq. (12). Eq. (12) has a dierent frequency dependance and
if t is taken from Eq. (3) then the ratio of Eq. (12) to Eq. (1) is o
0
/ .
4. Sound transmission through a double wall
Sound transmission through a complete double wall by the non-resonant acoustic
path (room 1 to cavity 3 to room 5) can be computed if transmission into and out of
the cavity is known. In SEA notation [8] the airborne level dierence between the
two rooms can be given by
D
135
10log
j
3
j
5
V
5
j
13
j
35
V
1
_ _
27
which gives the sound reduction index as
R
135
D
135
10log
S
A
5
_ _
10log
j
3
j
5
V
5
S
j
13
j
35
V
1
A
5
_ _
28
R.J.M. Craik / Applied Acoustics 64 (2003) 325341 333
Substituting the expressions for the two coupling loss factors [Eqs. (12) and (14)]
and noting that the total loss factor of the room (5) is [8]
j
5

c
0
A
5
8fV
5
29
gives the sound reduction index as
R 10log
16
5
f
6
,
4
s
d
2
j
3
c
6
0
,
4
0
o
02
_ _
30
This can be written in terms of the eld incidence SRI of the individual panels, R
2
and R
4
[dened in Eq. (3)] as
R R
2
R
4
10log
16
3
f
2
d
2
j
3
c
2
0
o
02
_ _
31
Sharp [13] gives a semi-empirical expression for the SRI of a wall above the mass-
springmass frequency (below which he models the wall as a single leaf wall with a
single equivalent mass) and below the rst cross mode in the cavity (above which the
cavity behaves as a room) as
R R
2
R
4
10log f
2
d
2
_ _
29 32
which has the same depth eect (10log d
2
). It also has the same frequency depen-
dance (10log f
2
) but this is less important as both j and o
0
will be dependant on
frequency.
Using the equations of Price and Crocker [5] to compute the coupling into and
out of the cavity the equivalent expression for the SRI of the wall can be given by
[8,14]
R R
2
R
4
10log
2ln10
6
T
3
f
_ _
33
This equation is independent of the depth of the cavity and is therefore incon-
sistent with classical models and some experimental data.
5. Comparison with experimental data
In order to verify the theory developed for transmission into and out of a cavity a
comparison was made with experimental data measured by Craik and Smith [1]. In
that paper 2 sets of results were presented (Figs. 9 and 10 in the paper) for
334 R.J.M. Craik / Applied Acoustics 64 (2003) 325341
transmission into a cavity. One set of results is for transmission into a specially built
variable depth cavity that was 150, 100, 50 and 25 mm deep through a sheet of
plasterboard with dimensions 1.8 0.9 m. The cavity was formed from concrete
blocks built on top of a concrete oor and the plasterboard was attached to the
blockwork by a layer of plaster. The other results were for transmission into the
cavity of a real stud wall with cavity depths of 150, 100 and 50 mm through plas-
terboard with dimensions 3 0.35 m which was screwed to the timber studs. Both
sets of results have been reproduced here as Figs. 4 and 6 and each shows the mea-
sured data together with the original prediction and the revised prediction.
The SEA models used included resonant transmission through each panel and the
structural paths for the real stud walls. The non-resonant coupling Eqs. (12) and
(14) are only valid up to the rst cross mode (where d=l/2) but for simplicity they
have been used up to f
c
. The rst cross mode varies from 3.4 kHz (50 mm) to 1.1
kHz (150 mm) but as this is close to the critical frequency resonant transmission
becomes more important and the approximation causes little error.
The original results in Fig. 4 for the custom built cavity agreed well with the
measured data, particularly for the 100 mm deep cavity, and therefore, as the revised
theory has a dierent frequency dependance, the agreement with the revised theory
is not quite as good but is still reasonable. Whereas the original theory tended to
give a level dierence that was too high at mid to high frequency the revised theory
tends to be too low at low frequencies. The agreement is particularly poor at low
frequencies where negative results are predicted (and measured). Results at low fre-
quencies are likely to be aected by the relatively low mode count for the cavity
which has a mode count of less than 1 below 250 Hz and a modal overlap less than 1
below about 500 Hz. This low mode count and modal overlap will tend to decrease
coupling [15] and increase the measured level dierence and may explain why the
measured results agree better with the original theory at low frequencies.
In addition to the measurements of transmission into the specially built cavity [1]
measurements were made of the airborne level dierence for transmission out of the
cavity into the room. The low mode count in the cavity means that the energy is less
well distributed making it dicult to measure the mean energy accurately. If the
coupling (and hence level dierence) is determined mainly by the modal properties
of the receiving subsystem [15] then the low mode count and modal overlap would
not be expected to signicantly inuence the results. The measured and predicted
results are shown in Fig. 5. The measured level dierence shows peaks and
dips particularly at low frequencies which are presumed to be associated with
the modal behaviour of the cavity. However, it can be clearly seen that overall
the revised theory for transmission out of the cavity agrees much better with
the measured data than the original theory and in particular has the correct
slope.
The results for transmission into the cavity of a double wall (Fig. 10 of Ref. [1])
are shown in Fig. 6. The results show that the revised theory agrees much better with
the measured data. The predictions are limited at low frequencies as in this fre-
quency region the predicted coupling loss factor exceeds the measured total loss
factor giving invalid results. The good agreement with the revised theory occurs
R.J.M. Craik / Applied Acoustics 64 (2003) 325341 335
Fig. 4. Airborne level dierence for transmission into rigidly bounded test cavities with depths of 25, 50,
100 and 150 mm: , measured; - - - -, predicted with new theory;
. . . .
, predicted with old theory.
336 R.J.M. Craik / Applied Acoustics 64 (2003) 325341
over the entire frequency range and there does not appear to be a dierence at low
frequencies as there was with the results of Fig. 4 although the modal overlap and
modal density values are very similar. The consistent over prediction of the original
theory no longer occurs and there is good agreement with the measured data over
the entire frequency range.
Fig. 5. Airborne level dierence for transmission out of rigidly bounded test cavities with depths of 50,
100 and 150 mm: , measured; - - - -, predicted with new theory;
. . . .
, predicted with old theory.
R.J.M. Craik / Applied Acoustics 64 (2003) 325341 337
Sound transmission out of these cavities was also measured on two of the same
test walls (but dierent cavities within it) and the results are shown in Fig. 7. The
measured results agree well with the revised theory and show very poor agreement
with the original theory.
When the two sets of results (sound transmission into and out of the cavity) are com-
bined the sound reduction index of a double wall can be computed. As transmission
Fig. 6. Airborne level dierence for transmission into cavities with depth 50, 100 and 150 mm that form
part of a double wall with no added absorption: , measured; - - - - predicted with new theory;
. . . .
, pre-
dicted with old theory.
338 R.J.M. Craik / Applied Acoustics 64 (2003) 325341
through a wall includes both transmission into a cavity and out of the cavity any
error in the theory will be doubled and therefore the ability to predict overall sound
transmission is a more demanding test.
Results given by Craik and Smith [1] are presented in Fig. 8 and show the mea-
sured results together with predictions based on the original theory and the revised
theory. It can be seen that the revised theory gives much better agreement for all
three walls tested, although the dierence is small for the 150 mm wall as other
transmission mechanisms (including structure-borne transmission) are also included.
A comparison of the theory with other published data is very dicult as in most
cases there is insucient data to allow a prediction to be made (for example, most
published data does not include cavity damping). The original results of Price and
Crocker [5] gave good agreement with their theory and therefore the use of the
revised theory presented here will make the agreement worse giving results similar to
Fig. 4. This is also true for the results of Craik and Nightingale [14]. Unfortunately
test results which do not agree well with predicted results tend not to be published so
a comparison with published data must be carried out with care.
Fig. 7. Airborne level dierence for transmission out of cavities with depth 50 and 100 mm that form part
of a double wall with no added absorption: , measured; - - - -, predicted with new theory;
. . . .
, predicted
with old theory.
R.J.M. Craik / Applied Acoustics 64 (2003) 325341 339
6. Conclusions
The clear conclusion from the 16 sets of measured data presented is that the
revised theory for coupling between a cavity and a room gives better agreement with
measured data than the original theory. In particular, the slope of the level dier-
ence error is correctly predicted. There are cases where the original theory ts better
but the balance is clearly in favour of the revised theory.
The theory presented used an innite plate model to determine the forced motion
of the plate together with a nite plate radiation eciency. A more sophisticated
Fig. 8. Sound reduction index for transmission through a 50, 100 and 150 mm double wall with no added
absorption: , measured; - - - -, predicted with new theory;
. . . .
, predicted with old theory; &, predicted at
low frequencies.
340 R.J.M. Craik / Applied Acoustics 64 (2003) 325341
model could be developed incorporating the nite dimensions of the plate. However,
the eort in making such improvements probably cannot be justied as the low
mode count and modal density in most cavities may lead to substantial uctuations
and would probably not lead to signicantly better agreement between measured
and predicted results.
Acknowledgements
This work was co-funded by the Engineering and Physical Science Research
Council (EPSRC) and Qinetic Group (formerly DERA) of the UK. The experi-
mental work was carried out by Dr. R.S. Smith.
References
[1] Craik RJM, Smith RS. Sound transmission through double leaf lightweight partitions. Part I: Airborne
sound. Applied Acoustics 2000;61:22345.
[2] Beranek LL Noise and vibration control. McGraw Hill; 1971.
[3] Sewell EC. Transmission of reverberant sound through a single leaf partition surrounded by an
innite rigid bae. Journal of Sound and Vibration 1970;12:2132.
[4] Leppington FG, Heron KH, Broadbent EG, Mead SM. Resonant and non- resonant acoustic
properties of elastic panels. II. The transmission problem. Proceedings of the Royal Society of London
1987;A412:30937.
[5] Price AJ, Crocker MJ. Sound transmission through double panels using statistical energy analysis. J
Acoust Soc Am 1970;47:68393.
[6] Craik RJM, Sound transmission through buildings using statistical energy analysis. Gower, 1996.
[7] Morse PM, Ingard KU. Theoretical acoustics. McGraw Hill; 1968.
[8] Cummings A. Low frequency acoustic transmission through the walls of rectangular ducts. Journal
of Sound and Vibration 1978;61:32745.
[9] Maidanik G. Response of ribbed panels to reverberant acoustic elds. J Acoust Soc Am 1962;34:
80926.
[10] Leppington FG, Broadbent EG, Heron KH. The acoustic radiation eciency of rectangular panels.
Proceedings of the Royal Society of London 1982;A382:24571.
[11] Cremer L, Heckl M, Ungar EE. Structure-borne sound. Springer-Verlag; 1998.
[12] Fahy FJ. Foundations of engineering acoustics. Academic Press; 2001.
[13] Sharp BH. Prediction methods for the sound transmission of building elements. Noise Control
Engineering 1978;11:5363.
[14] Craik RJM, Nightingale TRT, Steel JA. Sound transmission through a double leaf partition with
edge anking. J Acoust Soc Am 1997;101:9649.
[15] Craik RJM, Steel JA, Evans DI. Statistical energy analysis of structure-borne sound transmission at
low frequencies. Journal of Sound and Vibration 1991;144:95107.
R.J.M. Craik / Applied Acoustics 64 (2003) 325341 341

You might also like