Download as pdf or txt
Download as pdf or txt
You are on page 1of 18

Applied Acoustics 45 (1995) 193-210

Copyright 0 1995 Elsevier Science Limited


Printed in Great Britain. All rights reserved
0003-682X/95/%9.50
0003-682X(94)00045-X

Analysis of Interior Acoustic Fields Using the Finite


Element Method and the Boundary Element Method
S. Kopuz
Mechanical Engineering Department, Middle East Technical University,
0653 1 Ankara, Turkey

N. Lalor*
Institute of Sound and Vibration Research, University of Southampton,
Southampton, UK, SO9 5NH
(Received 16 August 1994; revised version received 2 December 1994;
accepted 9 December 1994)

ABSTRACT
For theJirst part of this study, the finite element method (FEM) and the
boundary element method (BEM) are both used to predict the interior
acoustic jield of a rectangular closed cavity, due to the vibration of one
wall. The direct (i.e. inversion of the dynamic stiffness matrix) and modal
superposition solution techniques are compared for the FEM, and the
collocation and variational techniques are compared for the BEM. It is
found that there are no significant diferences between the collocation and
variational results. Identical results are also obtained from both the direct
and modal superposition techniques, provided that suficient modes are
included in the latter case.
In order to investigate the influence of the boot (luggage) compartment
on the acoustic response of the passenger compartment of a car, a second
(smaller) rectangular cavity is joined to the one described above. The
acoustic response of the combination is computed by the FEM when the
two cavities are separated by dtflerent types of partition. It is found that
addition of the boot cavity, as well as the characteristics of the partition,
have a very marked effect on the main cavity response.
*To whom correspondence

should be addressed.
193

194

S. Kopuz,

N. Lalor

1 INTRODUCTION
The acoustic field inside a passenger car compartment
is affected by many
factors. One of these is the effect of the boot (luggage) compartment.
There are various types of partition between the main compartment
and
the boot compartment
of a car. There may be a permeable membrane or
there may be some holes in a basically rigid panel. Tt is important
to be
able to predict the effect of the boot compartment
on the sound pressure
levels at critical points such as the drivers ear position.
The two numerical
methods widely used in the analysis of interior
acoustic fields are the finite element method (FEM) and the boundary
element method (BEM). In this paper, these methods are applied to the
solution of the problems mentioned above using the SYSNOISE program,
and comparisons
are made wherever possible.
In order to perform this analysis FEM and BEM models of the passenger and boot compartments
were made. For simplicity these models
were idealized to rectangular
parallelepipeds
having the same overall
dimensions as an actual car.
First, two different solution techniques were used with a FEM model to
predict the acoustic field inside a single rectangular
cavity due to the
vibration of one wall, namely direct and modal superposition.
Then, the
same analysis was performed
using a BEM model which again uses two
different solution methods, namely the collocation
(direct BEM) and the
variational
(indirect BEM). The results obtained
from these different
methods are compared
with each other and also with the closed form
solution.
Having obtained some indication of the relative merits of these various
techniques, an acoustic cavity model consisting of both compartments,
as
described above, was set up. The prediction of drivers ear sound pressure
due to vibration of the front wall of the main cavity was made using both
the FEM and the BEM techniques.
Initially the acoustic field was analysed with no partition between the
two compartments.
In this case both the FEM and the BEM methods
were used. Finally, the same double compartment
FEM model has been
analysed with different types of partition between the two cavities, i.e. a
permeable membrane, and openings in an otherwise rigid partition.
This analysis could have been carried out on a BEM or FEM model
which accurately reproduced the internal shape of an actual car. However,
since the assessment of the two BEM and FEM methods was carried out
on a rectangular box (so that the results could be compared with the exact
solution) it was decided to continue using this simple configuration.
Thus,
the effect of an additional
(boot) cavity could be directly determined.

Analysis of interior acousticjields

195

Fig. 1. Geometry of the physical problem.

Nevertheless, the findings must be regarded as qualitative rather than


quantitative, although it is hoped that they will provide some basis
towards further research on the subject.

2 THEORY
The geometry of the problem is presented in Fig. 1. An ideal, homogeneous, inviscid fluid fills an interior volume V of the cavity surrounded
by a surface S. A general field point is denoted by A, a surface point by B
and the positive unit normal n is directed from the point on S as shown in
the figure, for BEM analysis. The problem to be considered here is the
acoustical behavior within cavities due to a vibrating boundary surface.
The acoustic medium inside the cavity is governed by the threedimensional wave equation. For most practical problems a sinusoidal time
dependence can be assumed so that the problem is greatly simplified.
Then, the wave equation reduces to the well known Helmholtz equation:
V2p+k2p=0,

k=w/c

(1)

where p is the acoustic pressure, k is the wave number, w is the angular


frequency, and c is the speed of sound in the medium. The above equation
may be subject to Dirichlet (known pressure), Neumann (known velocity)
or mixed (known impedance) boundary conditions specified on the surface
of the vibrating body.
The solution of the Helmholtz equation can be obtained by separation
of variables.2 This method involves the series expansion of the solutions in
terms of the normal modes of the system and can only be used with special
coordinate systems and boundary conditions. For arbitrarily shaped
bodies, solutions can only be obtained by numerical methods such as
FEM and BEM.

196

S. Kopuz, N. Lalor

2.1 The finite element method

The basic equations and the suitable boundary conditions are first
formulated and can be found in some references.,3 In the FEM approach,
the related functional is set up and its discretization by finite elements is
then performed. The final form of the equation in terms of pressure, p, can
be given as:

where [fl is the stiffness matrix, [w is the mass matrix [Cl is the damping
matrix of the fluid and v is the normal velocity on the surface. There are
basically two methods of solving eqn (2) namely direct (i.e. inversion of
the dynamic stiffness matrix), and modal superposition. Proportional
damping is used for the modal superposition technique. The reader is
referred to Ref. 4-8 for detailed information on these.
2.2 The boundary element method (BEM)
Applying Greens theorem together with the definition of Greens function
to the Helmholtz equation, one can obtain the following integral equation:
CWPW

G(4

B) q

-p(B)

BG(,An B)]

dS(B)

(3)

s[

where
e-MA,
G(A,

B) =

B)

4% B)

is the free-space Greens function due to a point source at point A, r is the


distance between points A and B, and c(A) is a constant that depends on
the location of point A as follows:
47r
c(A) =

27r

{ 0

for field points inside the body


for field points on the surface of the body
for field points outside the body

There are again basically two methods used for BEM analysis. The first
one is the direct BEM, known as collocation, and the second one is the
indirect BEM, known as variational. The basic difference between these
two methods is that one uses the pressure (velocity), but the other uses the

Analysis of interior acoustic fields

197

jump of pressure (jump of velocity) as the primary unknowns. Equation (3)


is the governing equation used by the BEM collocation method. The
readers can refer to Refs 9-20 for a detailed explanation of the solution
techniques.
2.3 Numerical solution technique
In the previous sections, the final form of the differential and the integral
equations which describe the interior acoustic field of an arbitrarily
shaped body is given. The numerical procedure for the solution of these
equations has been explained in detail in Refs 1 and 12 and generally
consists of the three main steps given below:
(1) Discretization of the boundary surface into finite or boundary
elements.
(2) Numerical integration to get an algebraic system of equations.
(3) Solution of the system of equations to obtain the unknown boundary
surface or field (interior) variables.
The two-dimensional surface of the body is discretized in the BEM
rather than the three-dimensional field inside the body as in the FEM.
Thus, the dimension of the problem is reduced by one. Another advantage
of the BEM is that the input data required needs less preparation effort
when compared to the FEM. However, there are also some advantages of
the FEM over BEM. Complex boundary conditions can be more easily
treated with FEM, and the total CPU times in solving the equations are
shorter for FEM for problems of relatively small size.
In order to represent both the geometry of the body and variation of the
acoustic variables within each element, quadratic isoparametric eiements
having four or eight nodes are used in this study. The eight-noded cubic
volumic element (HEXA8) is used in the FEM, whereas the four-noded
quadrilateral element (QUAD4) is used in the BEM.

3 CASE STUDIES

The numerical analyses performed were carried out using the SYSNOISE
program. This program is capable of applying both the FEM and the
BEM in the solution of interior acoustics problems. It is also possible to
treat complex boundary conditions in between the two cavities. In all the
numerical examples, a frequency range of 40-200 Hz is considered and the
frequency response of the sound pressure is observed at a point called the
drivers ear. Also the excitation is at the front wall (defined as the YZ

S. Kopuz, N. La/or

198

.A
X

1.6 m

Fig. 2. The plain box configuration.

plane through the origin) of the main compartment as a velocity boundary


condition with an amplitude of 1 mm/s while the other walls are treated
as rigid. The air density and speed of sound are taken as 1.2 kg/m3 and
340 m/s, respectively.
3.1 Case 1. Plain box (single compartment model)
A model of the plain box rectangular cavity is presented in Fig. 2 and has
224 HEXA finite elements with 365 nodes. The dimensions of the cavity
are taken as 1.6, 1.2 and O-8m in the x, y and z directions respectively. The
drivers ear point (point A) is located at 0.4, 0.8 and 0.6 m. The front wall
of the plain box model is excited harmonically as described above. The
frequency response of the sound pressure at point A is evaluated using the
FEM direct and modal superposition methods and the results are shown
in Figs 3, 4 and 5. It should be noted in these figures that, since the
excitation is in the x direction and is symmetrical with respect to the
cavity, only the x-wise resonance occurs.
As can be observed from the figures, the FEM direct modal superposition results agree well when 30 modes are considered in the superposition method. The frequencies given in the parentheses correspond to
the highest natural frequency found in the given number of modes.

Analysis of interior acoustic fields

199

The same problem is then solved using a BEM model which has the
same mesh density as the outer surface of the FEM model, and comprises
240 QUAD4 elements with 242 nodes. The different BEM approaches,
collocation and variational, produced almost identical results. Figure 6
shows that the results obtained from the BEM variational and the FEM
direct response are also very similar.

ModUSuperposlton(lOModes)

20
40

20

60

en

la,

120

140

1cO

180

200

220

FroqurnCYIW)

Fig. 3. Direct and modal superposition

(modes up to 282 Hz) response.

Direct

Wd

Superpo~ltk~

(22 Modes)

14)

20

40

60

a0

la,

la0
Fno*ney

Fig. 4. Direct and modal superposition

140

lco

180

200

(Hz)

(modes up to 402 Hz) response.

220

S. Kopuz, N. Lalor

200

The FEM and the BEM results were compared with the theoretical
results obtained from the closed form solution17 and found to be in very
good agreement (the maximum error was less than 1 dB). It is also of
interest that the total CPU times for one frequency of the FEM are
shorter than with the BEM.

20

40

60

80

1CCl

120

Fnqumy

Fig. 5. Direct and modal superposition

60

&xl

la,

140

SuPerposItion

160

(30 Mode9

l&I

ml

220

(W

(modes up to 500 Hz) response.

-FEM

40

Modd

EM

120
Fnqumc~

140

Ico

IW

Fig. 6. FEM direct and BEM variational

response

180

203

220

Analysis of interior acoustic fields

201

3.2 Case 2. Double compartment model


Having established the accuracy of the FEM and BEM results, the second
part of this study investigates the influence of the boot (luggage) compartment on the acoustic response of the passenger cavity of a car. A
model of the main compartment and the boot compartment made up of
rectangular cavities is represented in Fig. 7. The dimensions of the main
compartment cavity are the same as in the previous case and the dimensions of the boot compartment cavity are taken as O-4,0*8 and 0.8 m in the
x, y and z directions respectively. Analysis of the acoustic field was first
carried out when there was no partition in between the main and the boot
compartment.
The frequency response of the sound pressure at the
drivers ear was evaluated using the FEM direct and BEM variational
approaches. Two models having different mesh densities (coarse or fine)
were used for this analysis. The results are presented in Figs 8, 9 and 10.
The related mesh information and the first four natural frequencies of the
cavity obtained from the different methods are given in Table 1.
Figures 8, 9, and 10 show that the BEM and the FEM results are in
good agreement even in the case of a coarse mesh. The variation of the
FEM results with mesh density is slightly different than with the BEM
results, the latter being less sensitive to this. The first four natural frequencies of the cavity model in the frequency range of interest are also
shown in Table 1. It can be seen that all four models predict very similar
frequencies. It should be noted here that the presence of the extra modes is
because the excitation of the front wall of the main cavity is no longer
symmetrical with respect to the combined cavity.
TABLE 1

The modeling information

of the double compartment

Coarse mesh

Number of elements
Number of nodes
Element type
Mesh type
Natural frequencies (Hz)
1st mode
2nd mode
3rd mode
4th mode

cavity
Fine mesh

FEM

BEM

FEM

BEM

224
365
HEXA
Volume

240
242
QUAD4
Surface

448
695
HEXA
Volume

432
434
QUAD4
Surface

__-

89
143
163
192

88
142
162
190

89
143
162
191

88
141
161
190

S. Kopuz, N. Lalor

202

X
m
,

2.0m
I,,: 0.2m

I*

0.4m
-11
,

0.8m

2.0m
I,,: 0.2m

Z
(b)
Fig. 7. The mesh models used in the analysis:

(a) FEM; (b) BEM.

Analysis of interior acousticJields

203

3.3 Case 3. Double compartment model (with partitions)


For this analysis the previous double compartment cavity model was used
but with the addition of different types of partition in between the two
cavities. First, a simulation of a permeable membrane with flow resistance
is made. Secondly, a partition with different sizes of opening is analysed.

FEM

(FineMesh)

1Dl

1ZU

FEM (Coarse Mesh)

140

40

20
20

40

60

&I

-qrrmc~

140

140

180

200

220

ml

2M

Wz)

Fig. 8. FEM (fine mesh) and FEM (coarse mesh) response.

40

60

E!EM (FineMesh)

80

la,

- - - -

120
Fnqumcy

REM (Coarse Mesh)

143

160

l&l

Nb)

Fig. 9. BEM (fine mesh) and BEM (coarse mesh) response.

S. Kopuz, N. Lalor

204

All the analyses were carried out using the FEM direct response approach
as it was more suitable to simulate the required boundary conditions with
the FEM.
3.3.1 Permeable membrane
A permeable membrane is used to simulate the case when just the
back seat of a car forms the partition in between the two cavities. With
-

20

20

40

60

BEM (Fine Mesh) - - - - FEM (Fine Mesh)

80

103

120

140

163

180

200

220

Froquoncy (HZ)

Fig. 10. BEM (fine mesh) and FEM (fine mesh) response.

I...

No Partltlon

Permeable Membrane

140

120

loo

0
; 80
v)
&I
40
20

20

40

60

80

100

120

140

1cO

180

Fnquwwzy (Hz)

Fig. 11. FEM results for a permeable

membrane

partition.

200

220

Analysis of interior acoustic$elds

205

SYSNOISE it is possible to simulate a permeable membrane by means


of a transfer admittance. The membrane is discretized by elements
characterized by a transfer admittance value as given below:
At

=J-=

(4)

l.O-0.5j

AP

8116

4116

2116

l/16

Fig. 12. Details of openings in the partition (shaded elements are open).
-NoFc~rillbn

- 8/16~en

140

40

60

80

la,

120
Fnqrm~~

140

l&3

(W

Fig. 13. FEM results for an 8/16 opening ration.

180

MO

220

206

S. Kopuz, N. Lalor

where At is the transfer admittance. This factor is the ratio of the pressure
jump, Ap, induced by the shield to the normal velocity on the surface.3
Reference 19 shows typical values of acoustic impedance for the rear seat
of a car. These surface impedance values can be converted to transfer
admittance values by the method described in Ref. 21. Equation (4) gives
the complex value of the transfer admittance of the back seat obtained in

No PaMEm

- - - - 406 Open

140

40

20
xl

40

60

80

lal

140

120

Fnw.fw

Fig. 14. FEM results for a 4/16 opening

l&l

180

2fJl

220

l&l

2L-u

220

(W

ratio.

No Partition _ _ - - 2116Open

20

40

60

80

Ial

120
Fwwncy

140

lco

(W

Fig. 15. FEM results for a 2/16 opening

ratio.

201

Analysis of interior acousticJields

this way, averaged over the frequency range of interest. Figure 11 shows
the effect of the permeable membrane on the interior acoustic field when
compared to the case with no partition. It can be seen that the frequency
of the first mode is 97 Hz with the permeable membrane. This frequency is
quite close to the first natural frequency of the plain box compartment
model (see Fig. 6) which occurs at 106 Hz. Therefore, it appears that the
membrane reflects sound at low frequencies although there is also a
damping effect. It is interesting to note that at higher frequencies peak
sound pressure levels are significantly increased.
3.3.2 Partition with openings
The partition between the passenger and boot cavity often consists of a
quite stiff panel with openings on it. The effect of the openings on the
interior acoustics of the cavities is investigated in this part of the study.
The FEM direct method was used with a fine mesh discretization model.
The opening ratio is defined as the ratio of the surface area of the opening
to the surface area of the complete partition. Different opening ratios
ranging from 8/16 to l/16 were analyzed (see Fig. 12) and the results are
presented in terms of the first four modes in Table 2. Figures 13-16 also
show the frequency response of the sound pressure at the drivers ear for
the different opening ratios, compared with the case of no partition at all.
It can be seen that as the opening ratio is reduced there is an increasing
effect on the acoustic response. The first two opening ratios, 8/16 and
4/16, do not significantly change the natural frequencies of the first two

140

No Parlttion -

1116 Open

20

40

80

103

120

140

lti3

Frequency Hz)

Fig. 16. FEM results for a l/16 opening

ratio

180

200

220

S. Kopuz, N. Lalor

208

TABLE 2

The first four mode frequencies (Hz) for different opening ratios
No partition
Natural

1st mode
2nd mode
3rd mode
4th mode

89
143
163
192

8116 open
frequencies
88
143
161
189

4116 open

2116 open

l/16 open

82
136
145
181

77
127
144

(Hz)

87
143
154
181

modes, but change those of the last two modes, slightly in the case of the
S/16 opening ratio and more significantly in the case of the 4/16 opening
ratio. With the last two opening ratios, 2/16 and l/16, all natural frequencies are affected. In the latter case the drivers ear pressure is very
similar to that of a single main cavity but with the addition of a Helmholtz
resonance22*23 due to the small opening.

4 CONCLUSIONS
A study of the interior acoustic response of a rectangular closed cavity to
wall vibration has been carried out using both the finite element and
boundary element methods. It has been found that the collocation and
variational BEM solution techniques, and the direct and modal superposition FEM solution techniques all produce almost identical results,
provided that sufficient modes are introduced in the modal superposition.
For the example considered, this means in practice that modes up to two
and a half times the highest frequency of interest must be included.
An analysis of the effect of the boot (luggage) compartment on the
acoustic response of the passenger cavity has also been carried out. It
has been shown that the presence of the boot itself, together with the
characteristics of a partition between it and the main cavity, have a very
marked effect. For example, a permeable membrane can cause sound
reflections and increase peak sound pressure levels. It has also been found
that the presence of small openings in a rigid, non-porous partition can
give rise to marked Helmholtz resonance effects.

ACKNOWLEDGEMENTS
The Scientific and Technical Research Council of Turkey, Middle East
Technical University and University of Southampton Institute of Sound

Analysis of interior acoustic fields

209

and Vibration Research are gratefully acknowledged


for their support and
for the computational
facilities provided. Special thanks are also due to
Professor F. Fahy for his advice.

REFERENCES
1. SYSNOISE
Rev. 4.4A Theoretical
Manual. Numerical Integration Technologies, Belgium.
2. Morse, P. M. & Feshbach, H., Methods of Theoretical Physics. McGraw Hill,
New York, 1953.
3. SYSNOISE
Rev. 4.4A Training Material.
Numerical Integration Technologies, Belgium.
4. Zienkiewicz, 0. C., The Finite Element Method. McGraw Hill, New York,
1977.
5. ANSYS Rev. 4.4 Theoretical Manual. Swanson Analysis Systems, USA.
6. ANSYS Rev. 4.4 Reference Manual. Swanson Analysis Systems, USA.
7. Nefske, D. J., Wolf, J. A. & Howell, L. J., Structural acoustic finite element
analysis of the automobile passenger compartment: a review of current
practice. J. Sound Vib., 80(2) (1982) 247-66.
8. Nagy, L. I., Dede, M., Campbell, G. C. & Borders, S. G., Acoustic analysis
of a light truck cab. SAE Paper No. 880911, 1989, pp. 4.987-4.1001.
9. Chen, L. H. & Schweikert, D. G., Sound radiation from an arbitrary body. J.
Acoust.

Sot. Amer., 35(10 (1963) 1626-32.

10. Bell, W. A., Meyer, W. L. & Zinn, B. T., Predicting the acoustics of
arbitrarily shaped bodies using an integral approach. AZAA J. 15(6) (1977)
813-20.

11. Brebbia, C. A., The Boundary Element Method for Engineers. Pentech Press,
1980.
12. Brebbia, C. A. & Ciskowski, R. R., Boundary Element Methods in Acoustics.
Computational Mechanics Publications, 199 1.
13. Kipp, C. R. & Bernhard, R. J., Prediction of acoustical behavior in cavities
using an indirect boundary element method. J. Vib. AC. St. Rel. Des., 109
(1987) 2228.
14. Fyfe, K. R., Determination
of acoustic modal properties from boundary
element modelling. In Dynamic Engineering, Heverlee, Belgium, 1988.

15. Seybert, A. F., Cheng, C. Y. R. & Wu, T. W., The solution of coupled
interior/exterior acoustic-problem using the boundary element method. J.
Acout. Sot. Amer., W(3) (1990) 1612-18.
16. Seybert, A. F., Wu, T. W. & Wan, G. C., Recent applications

of boundary
element modelling in acoustics. Second International Congress on Recent
Developments in Air- and Structure-Borne Sound and Vibration, Auburn
University, USA, 1992, pp. 945-56.
17. Suzuki, S., Maruyama, S. & Ido, H., Boundary element analysis of cavity
noise problems with complicated boundary conditions. J. Sound Vib., 130( 1)
(1989) 79-91.
18. Ishiyama et al., The applications

predicting and reducing computer


pp. 4.976-4.986.

of ACOUST/BOOM
- a noise level
code. SAE Paper NO. 880910, 1989,

210

S.

Kopuz, N. Lalor

19. Utsuno, et al., Analysis of the sound field in an automobile


cabin by using
the boundary element method. SAE Paper No. 891153, 1990, pp. 1147-52.
20. Kopuz, S., Unlusoy, Y. S. & Caliskan, M., Boundary element method in the
development
of vehicle body structures
for better interior
acoustics.
In
Boundary Element Technology VIII, ed. C. A. Brebbia & H. Pina. Computational Mechanics Publications,
1993, pp. 46655.
21. Quinli,
W., Empirical
relations
between
acoustical
properties
and flow
resistivity of porous plastic open-cell foam. J. A&. Acoust., 25 (1988) 141~ 8.
22. Crighton,
D. G., Dowling,
A. P., Ffowcs Williams,
J. E., Heckl, M. &
Leppington,
F. G., Modern Methods in Analytic& Acoustics. Springer-Verlag,
London, 1992.
23. Kopuz, S., Unlusoy,
Y. S. & Caliskan,
M., Examination
of the interior
acoustic field in the presence of sound leakage through openings.
Bound.
Element Commun. J., 5(4) (1994) 168-~70.

You might also like