ARTICLE IN PRESS

Journal of Sound and Vibration 328 (2009) 607–617

Contents lists available at ScienceDirect

Journal of Sound and Vibration

journal homepage: www.elsevier.com/locate/jsvi

Acoustic design optimization for vibration of cylinder container

coupled with interior acoustic field
Zhang Jun a,b,, Cheng Geng-dong a, Zhao Wen-zhong b, Zhao Guo-zhong a
a
State Key Laboratory of Structural Analysis for Industrial Equipment, Dalian University of Technology, 2 Linggong Road, Dalian 116085, PR China
b
College of Traffic and Transportation Engineering, Dalian JiaoTong University, 794 Huanghe Road, DaLian 116028, PR China

a r t i c l e in fo abstract

Article history: Cylinder container is a thin cylindrical shell sealed by a top and bottom plate. The
Received 6 June 2007 internal noise reduction of the cylinder container is very important for vehicle, ship and
Received in revised form other projects. A finite element model of the cylinder container vibration coupled with
3 August 2009
internal acoustic field is set up and the acoustic response feature is determined. Then,
Accepted 26 August 2009
two optimization models are adopted: (1) the acoustic response is taken as the object
Handling Editor: L.G. Tham
Available online 2 October 2009 function and the structure weight is taken as constraints; (2) the structure weight is
taken as the object function and the acoustic response is taken as constraints. The
Keywords: design variables are thickness distribution for shell, bottom plate and top plate. The
Optimization
optimization problem is solved by the feasible direction method and a 30 dB reduction
Cylinder container
of the pressure level is achieved for optimum designs. The results show the effectiveness
Acoustic pressure
Coupled structure-acoustic of the method.
Weight & 2009 Elsevier Ltd. All rights reserved.

1. Introduction

Cylinder container is widely applied in many industries, such as the cab of the automobile, the cabin of the ship and that
of the aeroplane. Vibration of this kind of structure is coupled with interior acoustic field and becomes a major noise source
of the traffic facility. The structural vibration produces noise and the noise generates a reaction to the structure because of
the acoustic pressure. Finite element method is a major method of solving this kind of problem. The acoustic numerical
value method of the coupled structure-acoustic system based on FEM is a foundation of the design of the noise reduction. It
can forecast the vibration and noise at an early design stage and be widely used in the noise reduction research of car and
vehicle [1–3]. The traditional design of noise reduction requires time-consuming and tedious processes of successive
analyses involving sequential design modifications until the desired criteria are satisfied. However, when the design
sensitivity and optimization is used in design process, engineers can design the best suitable structure for the noise
reduction without both tedious and time-consuming sequential design modifications. The size optimization design of the
structure-acoustic system is to minimize the acoustic pressure or acoustic power by modification structure parameter.
Three steps must be done in this optimization process: the first one is to carry out the acoustic response, the second is to
derive the acoustic sensitivity with respect to the structure parameter, and the last step is to conduct the acoustic
optimization.

 Corresponding author at: State Key Laboratory of Structural Analysis for Industrial Equipment, Dalian University of Technology, 2 Linggong Road,
Dalian 116085, PR China.
E-mail address: [email protected] (Z. Jun).

0022-460X/$ - see front matter & 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jsv.2009.08.026
 ARTICLE IN PRESS
608 Z. Jun et al. / Journal of Sound and Vibration 328 (2009) 607–617

Several papers that deal with structural-acoustic optimization indicate significant improvements of the acoustic
properties inside or outside the structure. Noise reductions of 10–50 dB have been reported. In optimization process, the
sensitivity calculations of acoustic pressure or acoustic power with respect to the structure parameters are necessary when
classical basing-gradient methods are used. Ma and Hagiwara [4,5] performed the sensitivity analysis of both modal and
response acoustic pressure of coupled acoustic-structure system with respect to the structure parameters using modal
method and the direct method, and the pressure sensitivities had been successfully used in the design of reducing noise
inside the vehicle. Hambric [6,7] reported the reduction of the emitted sound power of a ribbed cylindrical shell with
hemispherical ends over a large frequency range by about 10 dB using location of ribs, shell thickness and loss factors as
parameters. Similarly, Lamancusa [8,9] indicated 8–15 dB reduction in a frequency range of 100–1000 Hz only by
optimizing the local distribution of plate thickness at a constant mass. Pal and Hagiwara [10] performed a optimization of
the structure response and the sound pressure within a box with the structure synthesis method and the pseudo-inverse
method and a reduction of about 50 dB is achieved. S. Marburg and H.-J. Beer et al. [11] researched a steel box of with an
external beam structure welded on three surface plates, their investigation included experimental modal analysis and
experimental measurements of certain noise transfer functions (sound pressure at points inside the box due to force
excitation at beam structure), and the results showed validity of the acoustic optimization. Jianghui Luo and Hae Chang Gea
[12] used modal method to evaluate the interior sound level for coupled structural-acoustic system, then performed a
topology optimization of stiffener with Microstructure-based Design Domain Method (MDDM) and a pressure level
reduction of about 20 dB is achieved. Erik Bangtsson et al. [13] optimized the shape of an acoustic horn to provide
impedance matching to the surrounding air using a BFGS quasi-Newton algorithm and the resulting horn has good
impedance properties throughout the entire frequency band of interest. Renato Barbieri and Nilson Barbieri [14] presented
a shape optimization of muffler design with the transmission loss being maximized in the frequency range of interest.
Eddie Wadbro and Martin Berggren [15] presented a method for topology optimization of an acoustic horn with the aim of
radiating sound as efficiently as possible, the Helmholtz equation modeling the wave propagation is solved using the finite
element method. In the above mentioned references, the optimization objects are all the acoustic parameters such as
acoustic pressure, acoustic power and acoustic feature coefficient, etc, the structure weight have not been taken as the
object.
In general, the weight reduction will induces a noise addition, reversely, a weight addition will result to the noise
reduction, the acoustic characteristic has a significant relation with the weight of structure. In this paper, the acoustic
response and the acoustic sensitivity have been carried out by the use of the direct FEM, the acoustic optimization have
been performed by the feasible direction method. The two optimization model have been set up: one is that the response
acoustic pressure of nodes is taken as object function and the structure weight is taken as the restriction, the other one is
that the structure weight is taken as object function and the response acoustic pressure of nodes is taken as restriction. The
acoustic optimization of the cylinder container have been performed as an example. The example results show that the
design optimization of cylinder container is an effective methods for simultaneously reducing response acoustic pressure
level and structure weight.

2. The acoustic response and its sensitivity of the coupled structure-acoustic system

The radiated steady-state acoustic pressure P of structure when acted a harmonic load satisfy Helmholtz differential
equation
2
r P þ k2 P ¼ 0 ð1Þ

where P is the acoustic pressure amplitude, k=o/c is the wavenumber, c is the speed of sound in the medium, o is the
frequency;

q2 q2 q2
r2 ¼ 2
þ 2þ 2
qx qy qz

is the Laplacian operator. The acoustic pressure amplitude P is independent of time, but relates directly to the structure
characteristic, acoustic field position and o.
The Eq. (1) indicates the distribution of the acoustic pressure P in space. The acoustic pressure P can be solved if the
boundary condition is knew well. If the vibration boundary condition of a structure is a small vibration amplitude on the
flexibility surface, it can be expressed as follows
qP
¼ nro2 US ¼ jnroVS ð2Þ
qn

where n is the unit normal vector on the structure surface; US, VS is the normal displacement amplitudepand ffiffiffiffiffiffiffi normal
velocity amplitude of the structure vibration, respectively, r is the density of the acoustic medium, and j ¼ 1.
 ARTICLE IN PRESS
Z. Jun et al. / Journal of Sound and Vibration 328 (2009) 607–617 609

The Eq. (1) having the boundary condition described by Eq. (2) can be solved using the numerical method such as FEM
(finite element method) and BEM (boundary element method), the FEM is more applied to the interior acoustic field
enveloped by the thin shell structure.
When the FEM is used to solve the acoustic problem radiated by a structure vibration, The acoustic field space described
by Eq. (1) with the boundary condition of Eq. (2) is partitioned into the finite elements while the structure should also be
partitioned into the finite elements, the acoustic pressure vector of acoustic media nodes is expressed as {P}, the vibration
displacement vector of the structure nodes is expressed as {Us}. Combining the Eq. (1) with the Eq. (2) and using the
weighted residual formulation and the Green’s function, the FEM equation used to solve the acoustic pressure {P} can be
formulated as follows
ð½Ka   o2 ½Ma ÞfPg ¼ o2 ½SfUs g ð3Þ

where [Ka], [Ma] and [S] are, respectively, stiffness, mass of acoustic field and coupled matrices between the acoustic and
the structure, which are directly related to the property of acoustic media, the geometry shape of the acoustic field, the
stimulated frequency and the coupling characteristic. If only the acoustic pressure {P} is required to solve, then {Us} should
be knew well. If the reaction on the structure loading by the acoustic pressure {P} can not be neglected, then this kind of
problem is regarded as the problem of the coupled acoustic-structure. In the problem of the coupled acoustic-structure, the
integrative FEM equation of the acoustic field and the structure is common established, then {P} and {Us} can be solved
simultaneously.
If the reaction force on the structure loading by the acoustic pressure {P} is considered, then the vibration FEM equation
of the non-damping structure can be expressed as follows:
ð½KS   o2 ½MS ÞfUs g ¼ fFa g þ fFS g ð4Þ

where fFa g ¼ ½ST fPg is the reaction force on the structure loading by the acoustic pressure {P}, [KS], [MS] are the stiffness
and mass matrices of the structure, respectively, {FS} is the outer stimulating load acting on the structure.
By a combination of Eq. (3) and (4), the coupled integrative FEM equation can be formulated as follows
½ZfUg ¼ fFg ð5Þ
" #
½KS   o2 ½MS  ½ST
½Z ¼ ð6Þ
o2 ½S ½Ka   o2 ½Ma 

where [Z] is called as impedance matrix, fUg ¼ ffUs g fPggT is unknown vector including the vibration displacement
amplitude vector of structure node and response acoustic pressure amplitude vector of fluid node, fFg ¼ ffFS g 0gT is the
outer load vector. If outer load {F} acting on the structure is knew well, then {U} can be solved from the Eq. (5).
The dynamics response of coupled structure-acoustic system can be obtained from Eq. (5) using the direct method
fUg ¼ ½Z1 fFg ð7Þ

We denote the design variable as X ¼ fx1 ; x2 ; . . . ; xn gT , differentiate Eq. (5) with respect to xj ðj ¼ 1; 2; . . . ; nÞ, we can have the
dynamics response sensitivity fUg0 with respect to xj, which include the structure displacement sensitivity fUs g0 and the
acoustic pressure sensitivity fPg0, this sensitivity indicate the direction of the structure modification for the design
optimization, and it is major part of the reduction noise design. The outer load {F} is not related to the design variables,
then differentiating Eq. (5) with respect to design variable X, we have
½Z0 fUg þ ½ZfUg0 ¼ 0 ð8Þ

from Eq. (8) we can get the sensitivity fUg0 as follows

½U0 ¼ ½Z1 ½Z0 ½U ð9Þ

where ½Z0 is the sensitivity of the impedance matrix [Z].

According to Eq. (7) the acoustic pressure {P} can be carried out, then the pressure level can be expressed as follows:
fLg ¼ 20 log10 ðfpg=U0 Þ ð10Þ

where U0=2  105 pa is the reference acoustic pressure.

From Eq. (10) the sensitivity of the pressure level can be formulated as follows:
fLg ¼ 20fpg0 ðfpgln 10Þ ð11Þ

3. The optimization design model

The structure-acoustic optimization depends on both the dynamics response and the sensitivity of the coupled
structure-acoustic system, the shell thickness is taken as the design variable in this research, the optimization model is
 ARTICLE IN PRESS
610 Z. Jun et al. / Journal of Sound and Vibration 328 (2009) 607–617

formulated as follows

(1) Minimizing the acoustic response

The object function
X
l X
m
min y ¼ ½ p2k;j ðx; oj Þ1=2 ; j ¼ 1; 2; . . . ; m ð12Þ
k¼1 j¼1

The constraining condition

(
G1 oGðXÞoG2
ð13Þ
xid oxi oxiu ; i ¼ 1; 2; . . . n

(2) Minimizing the structure weight

The object function
X
n
min G ¼ xi Ai r; i ¼ 1; 2; . . . ; n ð14Þ
i¼1

The constraining condition

(
pk;j ðX; oj Þrpu ; j ¼ 1; 2; . . . ; m
ð15Þ
xid oxi oxiu ; i ¼ 1; 2; . . . n

In Eqs. (12)–(15), the design variable X is the local shell thickness; G, G1 and G2 is the structure weight, the lower bounds
and the upper bounds of the structure weight, respectively; xid , xiu is the lower bounds and upper bounds of the design
variable xi, respectively; m is the number of the interest frequency considered; n is the number of the design variables; l is
the number of the considered interest fluid nodes; oj is the jth simulating frequency; pk;j is the response acoustic pressure
of the node k in the frequency oj.
The above optimization models belong to the constrained optimum problems, the feasible direction method is
commonly used and is an effective and main method for solving this kind of problems. The fundamental principle is that a
feasible preliminary point X0 is selected firstly, then a feasible direction d is established by the sensitivity information of
the object function with respect to the design variables and a step length a is figured out with one dimension search
method, and the iterative formulation can be written as follows
Xkþ1 ¼ Xk þ adðk ¼ 1; 2; . . . ; Þ ð16Þ

According to Eq. (16), we will achieve an optimum design variable X that it make the object function have a least value.
In optimization process, the sensitivities of both the object functions and the constrain conditions with respect to the
design variables are required, this sensitivities requires that the acoustic pressure sensitivities with respect to the design
variables are calculated.

4. Numerical results and the validation of the method

4.1. The FEM model and its acoustic pressure response

The FEM model of the cylinder container coupled with interior acoustic field is shown in Fig. 1 as an example for
validating the mentioned method. The cylinder height is 1.5 m and the diameter is 1 m, the interior is filled with air, the nodes
on circumference of lower circle plane are hinged, an excitation force with amplitude 50 N and the frequency range from 1 to
200 Hz is applied to center point on upper circle plane, which pointed to the z direction. The cylinder shell is meshed into
394 elements with both 294 quadratic shell elements, 100 triangle shell elements and 496 nodes, the interior acoustic field is
meshed into 600 hexahedron solid elements and 200 wedge-shaped solid elements. The shell material property is follows:
the elasticity modulus E=2  1011 Pa/m2, Poisson ratio m=0.3, the mass density rs=7800 kg/m3; the acoustic feature of air is
follows: the bulk modulus B=2.1 105 Pa/m3, the density ra=1.25 kg/m3, the acoustic speed C=340 m/s.
The node number on the axis of the cylinder acoustic field are shown in Fig. 2, the numbers in the bracket are the
coordinate parameters. The acoustic response of the 8 nodes on the center line of the cylinder acoustic field are shown in
Fig. 3, the node 8 is the upper end and the node 1 is the lower end, the 8 nodes are distributed uniformly on the center line
of the cylinder acoustic field from node 1 to node 8. Fig. 3 shows that the acoustic pressure peak are produced at the
frequency of 45, 135 and 175 Hz, the maximal pressure value is produced at node 1 at 175 Hz, the value is 251 Pa.
 ARTICLE IN PRESS
Z. Jun et al. / Journal of Sound and Vibration 328 (2009) 607–617 611

Fig. 1. The FEM model of the cylinder coupled with the interior acoustic field.

Fig. 2. The interesting optimization design nodes.

The pressure distribution contour of the cylinder acoustic field at frequency of the peak pressure value are shown in
Figs. 4–6, the Figs. 4 and 5 show that the acoustic pressure in cylinder acoustic field only change along the axis direction at
both 45 and 135 Hz. Fig. 6 shows that the acoustic pressure in cylinder acoustic field change not only along the axis
direction but also along the radial direction at 175 Hz, the change along the radial direction can be found on the down
end plane of acoustic field, it shows that its central region pressure is maximal, the pressure value reaches to the
424.8 Pa.

4.2. The acoustic pressure sensitivity

The shell of cylinder container is partitioned into 13 parts, the thickness of each part is taken as design variables, and it
is shown in Fig. 7. The upper circle plane and the lower circle plane are partitioned into 4 parts, respectively. Both part 1
and part 13 are a circle plane with the diameter 0.4 m on the lower and upper circle plane, and their thicknesses is denoted
as the design variable 1 and 13, respectively; the part 2 and 12, 3 and 11, 4 and 10 are the circle ring with the outer diameter
 ARTICLE IN PRESS
612 Z. Jun et al. / Journal of Sound and Vibration 328 (2009) 607–617

Fig. 3. The response acoustic pressure before optimization.

Fig. 4. The pressure contour at 45 Hz before optimization.

Fig. 5. The pressure contour at 135 Hz before optimization.

0.6, 0.8, and 1.0 m, respectively, and their thicknesses are denoted as the design variable 2 and 12, 3 and 11, 4 and 10,
respectively; the cylinder surface is partitioned into 5 parts with the uniform height 0.3 m, which are noted as 5, 6, 7, 8 and
9, and their thicknesses are denoted as the design variable 5, 6, 7, 8 and 9. The acoustic pressure level (SPL) sensitivity of
node 1–8 in Fig. 2 at 100 Hz for all design variables are shown in Fig. 8. The pressure level sensitivity of node 4 (1612) at
each frequency are shown in Fig. 9. Figs. 8 and 9 show that the frequencies of the peak of SPL sensitivities are same as the
frequencies of the peak of the response acoustic pressures.
 ARTICLE IN PRESS
Z. Jun et al. / Journal of Sound and Vibration 328 (2009) 607–617 613

Fig. 6. The pressure contour at 175 Hz before optimization.

Fig. 7. The design variable distribution.

Fig. 8. The pressure level sensitivity of all nodes at 100 Hz.

4.3. The structure-acoustic optimization design

4.3.1. Minimizing acoustic pressure with the restriction of weight

The optimization design of minimizing acoustic pressure is performed according to the optimization model (1)
formulated by Eqs. (12) and (13). The acoustic fluid nodes 1–8 in Fig. 2 are chosen as the design nodes; the design variables
are allowed to change from 3 to 7 mm; the structure weight is allowed to change from 240.92 to 242.92 kg, the initial
weight is 241.92 kg, which indicates that the structure weight is relatively non-changed in the optimization process. The
optimization frequencies are a series of values with an interval of 5 Hz starting from 5 Hz in 5–200 Hz. The actual
 ARTICLE IN PRESS
614 Z. Jun et al. / Journal of Sound and Vibration 328 (2009) 607–617

Fig. 9. The pressure level sensitivity of node 4 (1612) at each frequency.

Fig. 10. The optimized acoustic pressure by minimizing acoustic pressure.

Table 1
The optimized design variable values by minimizing acoustic pressure.

Design variable 1 2 3 4
Variable value (mm) 7.0 7.0 3.0 7.0

Design variable 5 6 7 8 9
Variable value (mm) 7.0 3.0 3.0 3.0 7.0

Design variable 10 11 12 13
Variable value (mm) 7.0 7.0 7.0 7.0

optimization model is follows

8
>
> X 8 X 20
>
> min y ¼ ½ p2k;j ðX; oj Þ1=2 ; oj ¼ 5 Hz; 10 Hz; . . . ; 200 Hz; k ¼ 1; 2; . . . ; 8
>
>
< k¼1 j¼1
ð17Þ
>
>
>
> 240:92 kgoGðXÞo242:92 kg
>
>
: 3 mmoxi o7 mm; i ¼ 1; 2; . . . 13
 ARTICLE IN PRESS
Z. Jun et al. / Journal of Sound and Vibration 328 (2009) 607–617 615

The optimized acoustic pressures of nodes 1–8 are shown in Fig. 10. Obvious reductions of the response acoustic pressures
can be seen in Fig. 10 compared with the Fig. 3, and the peak acoustic pressure at 175 Hz has been reduced to a little value;
the peak of acoustic pressure at 135 Hz has been also reduced from 150 to 80 Pa, and the peak of acoustic pressure at 45 Hz
has been reduced from 200 to 151 Pa at 60 Hz, this may be thought that the peak frequency is shifted from 45 to 60 Hz
and the pressure peak is decreased. The optimized design variable values are shown in Table 1, and the optimized
design variable values are all gotten the maximal value 7 mm or the least value 3 mm. The optimized structure weight is
242.6 kg.
The acoustic pressure will get more reduction if the structure weight is allowed to add DG kg in optimization process,
then the structure weight will achieve a maximal value DGþG kg for reducing the noise after optimization because of
optimization effect, and the optimization program will search for a optimum addition weight location. If DG is equal to
20 kg, then the upper restriction bounds of structure weight is Gr241.92þ20 kg (241.92 is the original weight), which is
that upper bound of the weight restriction is 261.92 kg. According to the optimization model (1), the optimization has been
performed again, the acoustic pressures of design nodes have been gotten the more reductions, the optimized response
acoustic pressures are shown in Fig. 11, the optimized design variable values are listed in Table 2. The Table 2 indicates that
the thickness of down circle plane of cylinder container not changed, the thickness of middle part of cylinder surface
(design variable 6, 7 and 8) is reduced to 4.7 mm while the thickness of upper part of cylinder surface (design variable 9) is
increased to 7 mm, and the thickness of upper circle plane is all increased to 7 mm that is the upper bounds value of design
variables. These clarify that an added weight of 20 kg is appended to the upside of the close cylinder container shell. Fig. 11
indicates that the response acoustic pressures are farther decreased from 151 to 120 Pa at 45 Hz when compared with
Fig. 10, but it can be found that the acoustic pressure at 135 Hz has been rather increased from 80 to 110 Pa, this fact shows
that an addition of 20 kg weight is chiefly used to reduce the lower frequency noise.

4.3.2. Minimizing the structure weight with the restriction of acoustic pressure
According to the optimization model (2) formulated with Eqs. (14) and (15), the optimization of minimizing weight has
been achieved with the restriction of the acoustic pressures at the nodes of the cylinder container center, the restriction

Fig. 11. The optimized acoustic pressure by minimizing acoustic pressure when having a added weight of 20 kg.

Table 2
The optimized design variable values when having a added weight of 20 kg.

Design variable 1 2 3 4
Variable value (mm) 5.0 5.0 5.0 5.0

Design variable 5 6 7 8 9
Variable value (mm) 5.0 4.7 4.7 4.7 7.0

Design variable 10 11 12 13
Variable value (mm) 7.0 7.0 7.0 7.0
 ARTICLE IN PRESS
616 Z. Jun et al. / Journal of Sound and Vibration 328 (2009) 607–617

equation Eq. (15) is actually formulated as follows

8
>
> pk;j ðX; oj Þr40 Pa; j ¼ 1; 2; . . . ; 40
>
>
< k ¼ 1; 2; . . . ; 8
ð18Þ
>
>
>
oj ¼ 5  j
>
: 0:003 mox o0:007 m;
i i ¼ 1; 2; . . . ; 13

Eq. (18) indicates that the node number of the acoustic pressure restrictions is 8, the stimulating frequency number is 40,
they are 5, 10, 15,y,195, and 200 Hz, the restrictions of acoustic pressures are 40 Pa, the design variable number is 13, the
upper bounds and the lower bounds of design variables is 0.003 and 0.007 m, respectively.
The optimized design variable values are listed in Table 3, it can be found that the thickness of the lower circle plane is
about all decreased except the design variable 2, the thickness of the upper circle plane is about all increased except the
design variable 10 and the thickness of the cylinder surface is gradually increased from lower end to upper end. The
optimized structure weight is 240.47 kg, it is that a reduction of 1.45 kg is achieved compared with the original structure
weight.
The optimized acoustic pressures are shown in Fig. 12, it can be found that the all peaks of acoustic pressures are small
than 40 Pa, which is the restriction value of acoustic pressure. Fig. 12 is compared with Fig. 3, it can be found that the
original peaks of acoustic pressures at 175 Hz have been decreased to about 10 Pa from original 251 Pa, a 30 dB reduction of
the pressure level is achieved for optimum designs, the optimized design has gotten a remarkable effect on reducing noise,
and the optimized structure weight has also gotten a reduction of 1.45 kg than the original structure weight, but optimized
acoustic field has been more complexity, the 4 pressure peaks appear in the Fig. 12 compared with the Fig. 11 that have
2 pressure peaks. Fig. 12 is compared with the Fig. 10, we can get the conclusion as follows: the weight optimization with
acoustic pressure restriction has a better effect on reducing noise than the acoustic pressure optimization with weight
restriction. Whether the weight optimization or the acoustic pressure optimization, the acoustic pressure reductions are
more great within higher frequency range.
The value of acoustic pressure used as restriction should be chosen in reason basing on both the response acoustic
pressure values and the demand of noise reduction, the optimization design may not be gotten if the restriction acoustic

Table 3
The optimized design variable values by minimizing structure weight.

Design variable 1 2 3 4
Variable value (mm) 3.3 6.0 3.6 3.0

Design variable 5 6 7 8 9
Variable value (mm) 4.0 5.2 5.2 5.2 5.8

Design variable 10 11 12 13
Variable value (mm) 3.4 6.9 6.9 7.0

Fig. 12. The optimized acoustic pressure by minimizing structure weight.

 ARTICLE IN PRESS
Z. Jun et al. / Journal of Sound and Vibration 328 (2009) 607–617 617

pressure is too lower. The more higher the restriction acoustic pressure is, the more smaller the optimized structure weight
is; if the restriction acoustic pressure is 160 Pa, then the optimized structure weight is decreased to 234.8 kg while the
maximal value of the acoustic pressure of optimization design is within 160 Pa.

5. Conclusion

The optimization designs of cylinder container structure for reducing interior noise are achieved basing on the finite
element model of coupled acoustic-structure system in this paper. The two optimization models are set up. One is that the
sum of the square of the response acoustic pressures at the design nodes is taken as the objective function with the
restriction of the structure weight, this optimization can decrease the acoustic pressure by redistributing of shell thickness
while the structure weight is almost unchanged; the other one is that the weight of cylinder container structure is taken as
the objective function with the restriction of the acoustic pressure at the design nodes, this optimization can
simultaneously decrease the response acoustic pressures and the structure weight if the restriction acoustic pressure is
chosen in reason. The weight distribution of optimized cylinder container shell is not uniform, generally, the shell thickness
is gradually become to large from down end to upper end. The optimization of minimizing weight with the restriction of
the acoustic pressure has been achieved a better effect on the reducing noise than the optimization of minimizing acoustic
pressure, and it can simultaneously reduce the structure weight. Whether or not, the acoustic pressure reductions are more
great within higher frequency. The optimization model of weight minimum with the restriction of acoustic pressure has
achieved a more better effect than the other. As a result of optimization, the reason of noise reduction is that the coupled
modal frequency is changed, the coupled modal frequency is shifted to apart from the stimulating frequency, because of the
acoustic field shape is not changed, so we also think that the structure modal frequency is shifted to apart from the
stimulating frequency and results the noise reduction.

Acknowledgements

We kindly acknowledge the financial support by Education Office of LiaoNing province government from China.
(No. 20060182, No. 2008102).

References

[1] S.H. Kim, J.M. Lee, M.H. Sung, Structural—acoustic modal coupling analysis and application to noise reduction in a vehicle passenger compartment,
Journal of Sound and Vibration 225 (5) (1999) 989–999.
[2] D.J. Nefske, J.A. Wolf Jr., L.J. Howell, Structural—acoustic finite element analysis of the automobile passenger compartment: a review of current
practice, Journal of Sound and Vibration 80 (2) (1982) 247–266.
[3] S.H. Sung, D.J. Nefske, A coupled structural—acoustic finite element model for vehicle interior noise analysis, Journal of Vibration, Acoustics, Stress and
Reliability 106 (9) (1984) 314–318.
[4] Z.-D. Ma, I. Hagiwara, Sensitivity analysis methods for coupled acoustic-structural system. Part I: modal sensitivity, AIAA Journal 29 (11) (1991)
1787–1795.
[5] Z.-D. Ma, I. Hagiwara, Sensitivity analysis methods for coupled acoustic-structural system. Part II: direct frequency response and its sensitivities, AIAA
Journal 29 (11) (1991) 1796–1801.
[6] S.A. Hambric, Approximation techniques for broad-band acoustic radiated noise design optimization problems, Journal of Vibration and Acoustics 117
(1) (1995) 136–144.
[7] S.A. Hambric, Sensitivity calculations for broad-band acoustic radiated noise design optimization problems, Journal of Vibration and Acoustics 118 (7)
(1996) 529–532.
[8] J.S. Lamancusa, Numerical optimization techniques for structural-acoustic design of rectangular panels, Computers and Structures 48 (4) (1993)
661–675.
[9] J.S. Lamancusa, H.A. Eschenauer, Design optimization methods for rectangular panels with minimal sound radiation, American Institute of Aeronautics
and Astronautics Journal 32 (3) (1994) 472–479.
[10] C. Pal, I. Hagiwara, Dynamic analysis of a coupled structural-acoustic problem: simultaneous multi-modal reduction of vehicle interior noise level by
combined optimization, Finite Elements in Analysis and Design 14 (5) (1993) 225–234.
[11] S. Marburg, H.-J. Beer, J. Gier, H.-J. Hardtke, Experimental verification of structural-acoustic modelling and design optimization, Journal of Sound and
Vibration 252 (4) (2002) 591–615.
[12] J. Luo, H. Chang Gea, Optimal stiffener design for interior sound reduction using a topology optimization basing approach, Journal of Vibration and
Acoustics 125 (7) (2003) 267–273.
[13] E. Bangtsson, D. Noreland, M. Berggren, Shape optimization of an acoustic horn, Computer Methods in Applied Mechanics and Engineering 192 (2003)
1533–1571.
[14] R. Barbieri, N. Barbieri, Finite element acoustic simulation based shape optimization of a muffler, Applied Acoustics 67 (2006) 346–357.
[15] E. Wadbro, M. Berggren, Topology optimization of an acoustic horn, Computer Methods in Applied Mechanics and Engineering 89 (2006) 89–106.