Mechanics of Advanced Materials and Structures

ISSN: (Print) (Online) Journal homepage: www.tandfonline.com/journals/umcm20

Damping characteristics and sound transmission

loss of finite composite plates with frequency
dependent constrained viscoelastic interlayer

Bo Wang, Hequn Min & Ting Qu

To cite this article: Bo Wang, Hequn Min & Ting Qu (08 Sep 2023): Damping characteristics
and sound transmission loss of finite composite plates with frequency dependent
constrained viscoelastic interlayer, Mechanics of Advanced Materials and Structures, DOI:
10.1080/15376494.2023.2253789

To link to this article: https://doi.org/10.1080/15376494.2023.2253789

Published online: 08 Sep 2023.

Submit your article to this journal

Article views: 152

View related articles

View Crossmark data

Citing articles: 2 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=umcm20
MECHANICS OF ADVANCED MATERIALS AND STRUCTURES
https://doi.org/10.1080/15376494.2023.2253789

ORIGINAL ARTICLE

Damping characteristics and sound transmission loss of finite composite plates

with frequency dependent constrained viscoelastic interlayer
Bo Wang , Hequn Min , and Ting Qu
Key Laboratory of Urban and Architectural Heritage Conservation, Ministry of Education, School of Architecture, Southeast University,
Nanjing, China

ABSTRACT ARTICLE HISTORY

This paper proposes a sublaminate layer-wise finite element modeling method to accurately evalu­ Received 17 July 2023
ate the damping and sound insulation characteristics of composite constrained layer damping Accepted 28 August 2023
plates. It incorporates the frequency-dependent property of viscoelastic interlayer into fully
KEYWORDS
coupled structure-acoustic interaction equations, and employs the generalized high-order displace­
Sound transmission loss;
ment hypothesis with tailored polynomial series expansion. The method is applicable to various layer-wise; constrained layer
boundary conditions and material layer combinations. Validation is conducted using numerical damping; viscoelastic
and experimental data. Parametric study identifies optimal thickness and stiffness for maximizing material; sound insulation;
modal loss factor. Additionally, sound insulation performance is examined in narrow frequency Multiphysics
bands, highlighting the damping enhancement of viscoelastic interlayer.

1. Introduction Zener model are three classical models based on different

series or parallel modes between the spring and the viscous
Composite structures have attained popular using in trans­
damper [16, 17]. These three models have limitations and
portation, aeronautics, and mechanical engineering due to
cannot accurately represent the creep behavior of VEM.
their excellent mechanical properties such as low density,
Hence, more accurate models have been conducted with
high specific modulus, and remarkable heat insulation [1–4].
fewer unknown coefficients and calculation conveniently,
However, the pure fiber reinforced polymer plates (e.g.
such as the Golla–Hughes–McTavish model [18], the gener­
CFRP, GFRP, etc.) tend to have low inherent damping and
are prone to induce strong resonant vibration, unnecessary alized Maxwell model [19], the Anelastic Displacement
noise, and premature failure. This severely limits their appli­ Fields (ADF) model [20], and the fractional derivative (FD)
cations in vibration and noise reduction fields. Constrained model [21, 22]. The last two models allow relatively few
layer damping (CLD) offers an effective way to realize high parameters to describe the materials’ mechanical behavior
structure damping with lightweight [5, 6], which achieves over a wide frequency band and are accordingly very suit­
higher damping than that of free layer damping by con­ able for implementation in the finite element (FE) solution
straining the viscoelastic material (VEM) interlayer in two procedure.
stiffer surface layers as pointed out by Kerwin et al. [7, 8] at The frequency dependent property of VEM makes the
first. The coupling of shear and stretching strain of the complex eigenvalue problem inherently nonlinear. Special
VEM interlayer can cause significant dissipation of vibration solution methods are usually adopted in the free response
energy. There have been many CLD applications in vibra­ analysis, including the complex eigenvalue method [23], the
tion control fields, and more details can be found in the modal strain energy (MSE) method [24], the direct fre­
review by Zhou et al. [9]. Recently, the concept of CLD is quency response method [25], the asymptotic method [26],
integrated into various composite structures, which can and the reduced order iteration technique [27]. For large
leverage both excellent characteristics of durability enhance­ problems, choosing a proper solution means accuracy and
ment and impact resistance extension [10–12]. significant time advantages. The iterative modal strain
The stress-strain relations of VEM are time delay with energy (iMSE) method is an extension of the MSE method
the influence of molecular chain structure. Then the mech­ in consideration of the frequency dependence of VEM prop­
anical properties, such as shear modulus, are frequency and erty [28]. The energy contributions of viscoelastic damping
temperature dependent, which means that it is not easy to are estimated by using the undamped mode shape to deter­
precisely characterize the dynamics of VEM [13–15]. mine the modal loss factor, which reduces the computa­
Previous studies have been conducted abundantly to develop tional cost greatly.
accurate rheological models to define complex constitutive Numerical solutions, such as the FE approach, are power­
relations. The Maxwell model, Kelvin-Voigt model, and ful to solve various refined plate and shell theories [29, 30],

CO-FIRST AUTHOR & CONTACT Hequn Min hqmin@seu.edu.cn School of Architecture, Southeast University, 2# Sipailou, Nanjing 210096, China.
� 2023 Taylor & Francis Group, LLC
2 B. WANG ET AL.

especially in dealing with dynamic responses of complex plates through the application of a high-order shear flexible
geometric structures. In the context of a FE model, the con­ model. The Zig-zag theory remains essentially a refined ESL
vergence is dependent on the degree of freedoms (DoFs), as approach by assuming a priori continuity of transverse-shear
well as a reasonable physical simplification. For the compos­ stress between successive layers. The dynamic problem of com­
ite CLD plates, a large interlayer stiffness difference exists posite CLD plates was addressed by Lewandowski et al. [48]
due to the insertion of the VEM interlayer. Hence, there are through the utilization of an enhanced Zig-zag theory, which
two additional difficulties: the first one is consideration of incorporates a piecewise linear function to enhance the in-
the frequency dependence of VEM property, and the second plane displacement without introducing HSDT. However, it is
involves accurately capturing various types of deformations. difficult to satisfy the Cz2 or Cz1 -continuity for the Zig-zag the­
Several well-known kinematic hypotheses, including the clas­ ory in FE solutions. The LW approach can better take into
sical laminate theory, first-order shear deformation theory, account the interlayer continuity of transverse stress and the
higher-order shear deformation theory (HSDT), and Zig-zag moderate to severe cross-section warping. This approach offers
theory have been detailly summarized by Sayyad et al. [31] the advantage of analyzing the localized behavior of individual
and Hu et al. [32]. Recently, through developing Reissner’s plies, which is a critical aspect for precisely evaluating the per­
mixed variational theorem as a natural extension of the vir­ formance of composite structures [49]. So, it seems very suit­
tual displacement principle, the framework of variable kine­ able to utilize the LW approach to investigate the dynamic
matics with condensed basic kernels has been formulated by characteristics of composite CLD structures, as noted by
Carrera et al. [30, 33], which was named Carrera Unified Carrera et al. [33, 50]. Ferreira et al. [37] used the LW
Formulation (CUF). It can keep the unknowns of the arbi­ approach for analyzing the free vibration of sandwich CLD
trarily selected displacement field hypothesis unchanged in plates with the focus on CUF. Then, this work was extended
compact formulas. The classical and higher-order theories by Liu et al. [51] through the quadrature hierarchical FE
can be realized in a unified way by selecting different basis model. Meanwhile, the DoFs of each node are larger and the
functions. The CUF has attracted a lot of attention and has calculation cost is higher for the case of more layers or higher
been applied to various problems such as statics, bending, expansion order of displacement field. The variable kinematics
and buckling [34–37]. As for the above kinematic hypothe­ framework such as CUF has been further extended by Demasi
ses, there are three FE modeling approaches for composite [39] and Yang et al. [52]. And the evaluation results of D
structures [38–43]: the equivalent single layer (ESL) ’Ottavio et al. ’s [53] showed that CUF can be implemented
approach, the layer-wise (LW) approach, and the sublami­ robustly in FE models with comparable solution precision to
nate layer-wise (sub-LW) approach that is the hybrid of the the 3D elasticity solution. Then, a more flexible sub-LW
former two. approach for composite structures was developed [40–43],
The ESL approach is on the basis of the classical laminate which allows the thinner ply or the layer with similar physical
theory or first-order shear deformation theory, and the dis­ properties to be combined into sublayers in the same model.
placement component is independent of the thickness coor­ Such method enables the mixed use of the ESL and LW
dinates. One of the key benefits of the ESL approach is approach simultaneously to balance the accuracy and the cost
simple implementation and low cost. However, the classical of DoFs. For example, the classical laminate theory is selected
laminate theory is under the Kirchhoff-Love assumption and for a thin surface layer, and the HSDT is for soft or thick core
neglects the transverse-shear strain so it is insufficient to layer such as foam core or VEM core. Jackstadt et al. [11]
describe thick structures. The first-order shear deformation applied the sub-LW approach to analyzing the free vibration
theory supposes vertically mid-plane during deformation, response of fiber-metal hybrid laminates, where the frequency
with which an appropriate shear correction factor should be dependent of VEM interlayer was considered. It is found that
introduced for thick plates, and even so, the Zig-zag effect that the third-order interpolation for in-plane displacement
or the interlayer continuity can be hardly captured. It is evi­ and second-order interpolation for out-plane displacement can
dent that the ESL approach can hardly address the issue of ensure good accuracy [11].
the structures like composite CLD plates with soft interlayer Many researchers have applied refined high-order theories
and large interlayer stiffness discrepancy. In fact, not only the and the FE solutions to investigate vibroacoustic problems of
transverse-shear deformation, but the transverse stretching composite CLD plates, mainly focusing on two aspects: the first
and compression deformation play a non-removable role in one is the response of plate-cavity coupling system [54, 55],
the dissipation mechanism of viscoelastic damping [44, 45]. and the second is on sound radiation at the aspect of the har­
This requires a higher-order or discrete layer description to monic point or line forces excitation [56–58]. From the litera­
take into account all kinds of deformations, as well as satisfy ture, it is noticed that there are few reports on the
the interlayer continuity condition of transverse stress. characteristics of sound transmission loss (STL) in composite
The HSDT expands the through-thickness displacement to CLD structures under acoustic excitation [59]. For accurate
the second or higher-order. Among these higher-order theo­ evaluation of the sound insulation, the structural coupling
ries, Reddy et al. [46] proposed a widely accepted third-order between the top and bottom surface layer shall be properly
shear deformation theory through the virtual displacement rule considered for evaluation of the incident and transmitted
for obtaining the equilibrium equation consistently. Mohamed sound fields, which is much more complex than the evaluation
et al. [47] observed that the damping performance of curved of the sound radiation under force excitation. Current research
fiber plates surpassed that of conventional composite laminate on the STL characteristics of CLD structures is often
 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES 3

characterized by certain limitations, such as the exclusive appli­

cation of classical plate theories or the focus on simplistic
sandwich plate configurations. Assaf et al. [60, 61] studied the
STL characteristics of symmetric and asymmetric finite iso­
tropic CLD plates with the modal superposition method.
Unfortunately, in their work, there is no narrowband informa­
tion on the STL spectrum, and the VEM layer’s frequency
dependency is not considered with the use of a constant com­
plex form. Wang and Dou [62] presented a LW approach
based on the generalized HSDT to explore the vibroacoustic
characteristics of finite sandwich CLD plates with a frequency-
dependent VEM core. The Rayleigh-Ritz solution of multilayer
panels with partial VEM patches was derived by Loredo et al.
[63] based on the LW approach utilizing the first-order shear
deformation theory, whose results demonstrate the effectiveness Figure 1. Sketch of the composite CLD plate under the excitation of plane
acoustic field, segregated by an infinite rigid baffle (z ¼ 0) in Cartesian coordi­
of appropriate VEM treatments for vibration suppression. nates. The azimuth angle is denoted as u 2 ½0, 2p�, and the elevation angle is
Larbi et al. [64, 65] conducted some FE models on the basis of denoted as h 2 ½0, p=2�:
the first-order shear deformation theory that fully couples the
acoustic-structure-piezoelectric field, which is not applicable for
composite CLD structures with more intermediate layers. The structure is segregated by an infinite rigid baffle parallel
In review of the aforementioned previous studies, there is to the xoy plane (z ¼ 0). All layers can be composed of
a great challenge still to efficiently and accurately evaluate homogeneous, isotropic, or orthotropic materials, as well as
the STL of complex composite CLD structures with numer­ the linear VEM with frequency dependency. Each ply is per­
ous layers and substantial interlayer dynamic discrepancies. fectly bounded and has no sliding at the interface.
In this paper, a fully coupled FE modeling method is pro­ Considering the fluid loading effect, the structure-acoustic
posed to accurately study the free vibration and STL charac­ interaction fully coupled governing equation can be derived.
teristics of composite CLD plates, utilizing the sub-LW And the immersed fluid field can be either air or water. The
approach as the foundation. The frequency dependent prop­ matrix form can be written after FE discretization as [66]
erty of VEM interlayers is considered, and the kinematic � � �� � � �
relations among composite CLD structures are established KS x2 MS CSA T u 0
¼ (1)
by the generalized HSDT with considering the frequency q0 x2 CSA KA x2 MA p FA
dependency of VEM interlayers. And the damping mecha­
where MS is the structural mass matrix. K�S ¼ KR þ iKI is
nisms of the VEM interlayer for sound insulation enhance­
the complex structural stiffness matrix with KI ¼ KI ðxÞ
ment are investigated additionally. The proposed model can
considering the frequency dependence. MA and KA are the
deal with composite structures with arbitrary layers and
thickness and is particularly suitable for the CLD applica­ acoustic field mass and stiffness matrix. FA is the equivalent
tions needing accurate calculation of the vibroacoustic and acoustic load at FE nodes. u and p are the vectors of the nor­
damping behaviors, such as aircraft soundproof fuselages, mal component of the structural displacement field and sound
lightweight building partitions, etc. pressure, respectively. The system coupling matrix is CSA :
In this paper, the theoretical formulation is outlined in At the interface, the normal velocity of the fluid and
Section 2. Then, in Section 3, the accuracy of the proposed structural particles is continuous, which can be written as
FE modeling method is verified with previously published X
nT q0 x2 u ¼ nT rp on (2)
numerical and experimental data in a free vibration case of SA
a five-layer laminated plate and an STL analysis case of a where the direction of unit normal vector n is unto the air,
complex triple-core composite plate. In Section 4, the para­ and the interaction interface is CSA :
metric effect of the VEM interlayer’s thickness and stiffness The acoustic wave equation describes the behavior of
on the free vibration of a composite CLD plate is explored. sound propagation through the distribution of sound pres­
And accordingly, the STL characteristics of the structure sure in the acoustic field, and can be expressed as
under the excitation of plane acoustic field and the sound
energy dissipation mechanism of the VEM interlayer are @2p
c2 r2 p ¼ 0 (3)
thoroughly discussed. @t2
where p and t represents the sound pressure and time,
2. Theoretical formulation respectively. c ¼ 343 m/s is the speed of sound. r2 is the
Laplacian operator, which quantifies the spatial variation of
2.1. Governing equation
sound pressure. Solving this equation helps analyze and pre­
Figure 1 presents the diagram of a composite CLD plate dict the behavior of sound waves in various environments,
under plane sound wave excitation with the variation of azi­ considering factors such as sound speed, spatial variations,
muth angle u 2 ½0, 2p� and elevation angle h 2 ½0, p=2�: and temporal changes.
4 B. WANG ET AL.

Substituting Eq. (2) into Eq. (3), the 3D discretized form approach can describe the displacement field of each layer
of the acoustic field is written as separately, the number of unknowns increasing with layers
� � needs to further reduce for lower computing costs. The pro­
P Р1 @2p e P �Рe
�
e e
XA c2 dp 2
dX A þ e e ðr � dpÞðrpÞdXA
XA
posed sub-LW approach can combine the advantages of the
@t
P �РP � previous two methods. It is very suitable to address the issue
þ e Pe q0 dpnT ud eSA
of high-precision STL prediction under the sound field load­
P �РP SAT Pe �
e
e n dpðrpÞd
AA ¼ 0
ing with a spatial and temporal variation. And the actual
AA
physical plies can be assembled flexibly into numerical layers
(4) or sublaminates, and the polynomial series expansion for the
Pe Pe
where AA , SA , and XeA
are the elemental acoustic field high-order displacement hypothesis can be chosen inde­
boundary, structure-acoustic interaction boundary, and the pendently for each sublaminate.
acoustic field domain, respectively. c and q0 ¼ 1.16 kg/m3 are As illustrated in Figure 2(a), assuming a composite CLD
the sound speed and density of the acoustic field, respectively. plate consists of p ¼ 1, 2, :::, Np seamlessly adhesion phys­
In this study, the coordinated nodes are used at the inter­ ical plies with the total thickness of H: In this paper, the
face surface [67]. Thus, the variables of u and p can be writ­ subscript p, k represent the physical plies and numerical
ten as layers, respectively. The xoy reference plane is located on
the structural mid-plane, and the layer index is counted
p¼NTP Pe and u ¼ NUe (5)
from the bottom layer. Some physical plies (p) can be arbi­
where NP and N are the global shape function matrices of trarily subdivided into a numerical sublaminate (k), which
fluid field and structure, respectively. Meanwhile, the elem­ indicates that the number of numerical sublaminates is
ent matrices in Eq. (1) can be written as always not higher than that of physical plies (Nk � Np ).
8 ð ð Thus, the physical domain (V) of the structure can be FE
>
> 1 e
� �T
>
>
>
e
MA ¼ 2 NP NP dXA , T e
KA ¼ rNTP rNTP dXeA discretized as
>
> c
>
> XeA XeA Ne Nk
Ne X
>
> X X
< Nk ð Nk ð V� Ve ¼ V e, k (7)
X X T
e, k ^ ð T e e, k
BðkÞ DðkÞ BðkÞ dXeS
> MS ¼ NM kÞN dXS , KS ¼
> e¼1 e¼1 k¼1
>
>
>
> k¼1 e
XS
k¼1 e
XS where the total finite elements number denotes Ne : V¼ X �
>
>
>
> ÐP P e H, and the reference surface X ¼ ½ a=2, a=2� � ½ b=2, b=2�
>
: CSA ¼ e N nT N d
T
FA ¼ CTSA p
P SA ,
SA with the global in-plane coordinates ðx, yÞ:
(6) In the global coordinate z 2 ½ H=2, H=2�, the thickness
ðkÞ ^ ð ðkÞ coordinates of each physical ply and numerical sublaminate
where D , M kÞ, and B are the generalized stiffness
(zp , zk ) are identified by their mid-plane coordinates (z0p ,
matrix, mass inertia matrix, and deformation matrix of the
z0k ), respectively.
k-th layer, respectively.
� �
hp hp
zp ¼ z z0p , zp 2 , and zk ¼ z z0k , zk
2.2. Kinematic relation 2 2
� �
As reviewed above, the ESL approach cannot capture the hk hk
2 , (8)
local deformation of the VEM interlayer. Although the LW 2 2

Figure 2. The k-th layer division and through-thickness kinematic assumptions of the composite CLD plate. (a) the k-th numerical layer is assemble of Np physical
plies. The superscript ðk, pÞ represents the p-th physical ply belonging to the k-th numerical layers. (b) the transverse deformation diagram with the cubic-order
expansion of in-plane displacement.
 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES 5

The thickness coordinates relation in the non-dimensional

coordinates is obtained as
hk 2
fp ¼ fk þ ðz0k z0p Þ (9)
hp hp
where fp , fk are the local thickness coordinates of physical
ply and numerical sublaminate, respectively.
The displacement field of k-th layer with generalized
high-order kinematical assumptions is expanded through
Taylor’s polynomial series
8
> k k
< uk ðx, y, zk Þ ¼ uk0 ðx, yÞ þ zk uk1 þ ðz2 Þ uk2 þ ::: þ ðzm Þ ukm
k k k k k 2 k k m k k (10)
v ðx, y, z Þ ¼ v0 ðx, yÞ þ z v1 þ ðz Þ v2 þ ::: þ ðz Þ vm
>
: k
w ðx, y, zk Þ ¼ wk0 ðx, yÞ þ zk wk1 þ ðz2 Þk wk2 þ ::: þ ðzn Þk wkn

where the coefficients uk0 , vk0 , and wk0 are the linear translations
along the x, y, and zk axes. The coefficients uk1 and vk1 are the
transverse-normal rotations about the y and x axes. wk1 is the
transverse-normal stretching and compress. The coefficients uk2 , Figure 3. The sequence of DoFs for the 8-node serendipity quadrilateral elem­
vk2 , and uk3 , vk3 are the parabolic stretching and cubic warping ent adopted in the present modeling method.
of the cross-section. The coefficient wk2 is linear translation of
the transverse-normal stretching (Figure 2(b)). The coefficients
ukj ðj ¼ 4, 5, 6, :::, mÞ, vkj ðj ¼ 4, 5, 6, :::, mÞ, and wkj ðj ¼ 3, 4, 5, :::, nÞ where i ¼ 1, 2, … , 8 represents the node index. n and g are
denote the higher expansion order of a polynomial and is called the element coordinate system can be mapped onto the glo­
generalized displacement field. All the coefficients would provide bal coordinate system’s x and y directions, respectively. Each
the node DoFs for the FE model. By combining Eq. (9) with edge length in the element coordinate system is standardized
Eqs. (6) � (8), the displacement field of each sublaminate can to 2, with the origin located at the center of each element.
be described separately. And following this definition of rela­ Thus, nodes are positioned at values of 1, 0, and þ1 along
tions, the fully coupled governing equation can be further the n and g directions, respectively.
derived, which is the core of the sub-LW approach. Indeed, the selection of a specific element type that
In the subsequent analysis, it assumes that the in-plane aligns with particular simulation requirements is regarded
displacement adopts the cubic-order expansion and the out-
as a pivotal decision in FE analysis, while ensuring an
of-plane displacement adopts the quadratic-order expansion
appropriate mesh density-wavelength relationship. This
for convenience [Figure 2(b)], denoted as fm, ng ¼ f3, 2g:
decision hinges upon factors such as the structural shape,
In fact, based on the extensive research on the adaptability
interpolation accuracy, and computational costs. Firstly,
of the refined plate/shell theory of Carrera et al. [33] and
Jackstadt et al. [11], this assumption can balance the DoFs second-order isoparametric elements, encompassing the 8-
and the calculation accurately well. node serendipity quadrilateral element and the 9-node
The generalized element displacements vector Ue, k of Lagrange quadrilateral element (with a central point), are
k-th layer is given by suitable for rectangular structures in Cartesian coordinates.
These elements mitigate the shortcomings of linear ele­
X
8
Ue, k ¼ Ni ðn, gÞUie, k (11) ments, like shear locking [68], allowing for higher-preci­
i¼1 sion approximations of complex transverse and in-plane
deformations. Specifically, the 8-node serendipity quadrilat­
where Uei , k ¼ ½u0 , v0 , w0 , u1 , v1 , w1 , u2 , v2 , w2 , u3 , v3 �T
eral element constructs shape functions through node pos­
is the displacement vector of k-th layer at a node i: Ni ðn, gÞ
are the two-dimensional shape functions of an 8-node C0 ser­ ition interpolation, exclusively suited for regularly shaped
endipity quadrilateral element. The detailed schematic of the quadrilateral structures. The shape functions of the 9-node
element is shown in Figure 3, and the expressions for the Lagrange quadrilateral element are founded upon Lagrange
shape functions Ni ðn, gÞ for the element can be written as interpolation polynomials, affording flexibility to approxi­
mate variable distributions at arbitrary positions, which
8 9 8 9 can accommodate structures with certain curvatures. For
> N 1 > >
> ð1 nÞð1 gÞð n g 1Þ > >
>
> >
> >
> > the plane CLD plates with regular shapes examined in this
>
> N 2 >
> >
> ð1 þ nÞð1 gÞðn g 1Þ > >
>
>
> >
> >
> >
> paper, the two element types achieve nearly identical accur­
>
> N3 >>
> >
> ð1 þ nÞð1 þ gÞðn þ g 1Þ >
>
>
< = 1 < ð1 nÞð1 þ gÞð n þ g 1Þ >
> = acy. Simultaneously, due to the simplified shape function
N4 �
¼ 2 (12) interpolation inherent to the 8-node serendipity quadrilat­
> N5 >
> > 4>> 2 1 n ð1 gÞ� >
>
>
> >
> >
> >
> eral element, the cost of element assembly can be signifi­
>
> N6 >> >
> 2ð1 þ nÞ �1 g2 >
>
>
> >
> >
> 2 >
> cantly reduced. Hence, the 8-node serendipity quadrilateral
>
> N7 >> >
> 2 1 n ð1 þ gÞ� >
>
: ; >
: 2
>
; element can be considered as a judicious and cost-effective
N8 2ð1 nÞ 1 g
choice for this study.
6 B. WANG ET AL.

Then, the mapping from local to global coordinates is From the application of Hamilton’s principle, the global
achieved using the 2 � 2 Jacobian matrix J and is given by weak form equilibrium equation is
8 9 2 38 9
>
> @Ni >
> @x @y > > @Ni >
>
Ne
X XNk ð
< = 6 7 < = fdUe gT U
€e ^ ð kÞNT dXeS
NM
@n ¼ 6 @n @n 7 @x (13) e¼1
k¼1
> @Ni > 4 @x @y 5> @Ni > XeS !!
> > > >
:
@g
;
@g @g
:
@y
; XNk ð ð Xe
€e ðkÞ ðkÞ T
þU k¼1
ðkÞ
B D B dXeS Pe NTP Pe d SA
¼0
SA
XeS

2.3. Constitutive relation and assembly (18)

where the definition of the element matrices is already given
Under the linear small deformation condition, the generalized
element strains vectors of k-th layer can be derived from the in Eq. (6).
nodal generalized displacement vectors in Eq. (11) as In this study, the 3 � 3 Gauss quadrature formula is uti­
8 lized to integrate the surface Xðx, yÞ to evaluate the global
>
> X8
matrices.
> e, k
Bebi, k Uei , k
< eb ¼
> Two constraint boundary conditions commonly employed
i¼1
(14) in structural dynamic analysis are herein delineated. The first
>
> X8
> e, k e, k e, k
Bsi Ui is the movable simply supported boundary condition, in
: cs ¼
>
i¼1 which the tangential displacement of the plate is constrained
h iT h iT along the boundaries, while the normal displacement remains
where ces , k ¼ ceyz, k , cezx, k , cexy, k and eeb, k ¼ eex, k , eey, k , eze, k unrestricted. The other is the clamped boundary condition,
are the transverse-shear strain vector and in-plane/trans­ simulating a scenario where the edges or boundaries of the
verse-normal strain vector. Besi, k and Bebi, k are the shear and plate are rigidly fixed and devoid of translational or deformable
bending parts of the strain-displacement matrix in the freedom in any direction. This implies that both tangential and
Cartesian coordinates. normal displacements, as well as rotations, are all constrained
In this model, each physical ply is allowed to use aniso­ at the edges. The expressions for these two boundary condi­
tropic materials with arbitrarily lay angles, and then tions are as follows:
assembled into a numerical sublaminate. The stress-strain
relations can be expressed with the Hooke law [69] i. Movable simply supported:
8 9ðk, pÞ 2 3 8 9ðkÞ v0 ¼ v1 ¼ v2 ¼ v3 ¼ 0, w0 ¼ w1 ¼ w2 ¼ 0 at x ¼ a=2; a=2
>
> rx >> � 11
C � 12
C � 13
C 0 0 � 16 ðk, pÞ >
C > ex >> ;
>
> >
> >
>e > > u0 ¼ u1 ¼ u2 ¼ u3 ¼ 0, w0 ¼ w1 ¼ w2 ¼ 0 at y ¼ b=2, b=2
>
> r y >
> 6 � 22
C � 23
C 0 0 � 26 7
C >
> y >
>
>
< > = 6 7 >
<e > =
rz 6 � 33
C 0 0 � 36 7
C z
(19)
¼6
6
7
>
> syz >
> 6
� 44
C � 45
C 0 7 7 >
> cyz >
> ii. Clamped:
>
> > > >
> szx >
> >
>
4 Sym: � 55
C 0 5 >
> czx >
> >
>
> >
: sxy ; � >
: >
; u0 ¼ u1 ¼ u2 ¼ u3 ¼ v0 ¼ v1 ¼ v2 ¼ v3 ¼ w0 ¼ w1 ¼ w2 ¼ 0
C 66 cxy
at x ¼ a=2; a=2 and y ¼ b=2, b=2
(15)
(20)
� ðk, pÞ
where C nm with n, m ¼ 1, 2, :::, 6 are the three-dimensional
stiffness coefficients for k-th numerical sublaminate and the
p-th physical ply in the global coordinate axes. 2.4. Solution procedure
By integrating the stress components through the k-th
layer’s thickness direction, the generalized element stress- The dynamic equilibrium equation can be derived through
strain relations can be obtained assembling the elemental mass and stiffness metrics to the
8 global coordinates, which can be written as
>
> X8
>
> reb, k ¼ Deb, k � eeb, k ¼
< ½DB�ebi, k Uei , k € þKS u ¼ FA
MS u (21)
i¼1
(16) where MS , KS , and FA are the same variables defined in
>
> X8
e, k e, k
> e, k e, k e, k ½DB�si Ui Eq. (1). u and u € are the structural displacement vector and
: rs ¼ Ds � cs ¼
>
i¼1 the second-order time derivative of displacement, respectively.
where Desi, k
and Debi, k
are the generalized transverse-shear It is suitable to describe through the complex modulus
stiffness matrix and membrane/bending/transverse-normal approach involving the VEM frequency dependency. So its
stiffness matrix. reb, k and res , k are the generalized stress vec­ shear modulus is given by
tors, and can be defined as G� ðxÞ ¼ G0 ðxÞ þ iG00 ðxÞ (22)
8
>
>
Ð hk � �e, k where the storage modulus and loss modulus are G0 ðxÞ and
< reb, k ¼ 2hk rx , ry , sxy , rz , zrx , zry , zsxy , zrz , z2 rx , z2 ry , z2 sxy , z3 rx , z3 ry , z3 sxy dzk
2 G00 ðxÞ: The damping loss factor is g ¼ G00 ðxÞ=G0 ðxÞ:
> Ð hk � �e, k
>
: res , k ¼ 2hk syz , szx , zsyz , zszx , z2 syz , z2 szx dzk A three-series form of the ADF model is adopted [20],
2
which characterized the viscoelasticity by elastic and anelas­
(17) tic parts. The stress is directly proportional to the elastic
 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES 7

part strain, and the anelastic part represents the relaxation. where the incident and transmitted power is represented as
The expression is as follows Wr and Wi :
! The incident power sound can be written as
Xn
Dj x
�
G ðxÞ ¼ G0 1 þ (23)
x iXj � �2
j¼1 �p � S cos h
i
Wi ðh, uÞ ¼ (29)
where G0 is the shear modulus in static regime. The strength 2q0 c
and relaxation time are Dj and 1=Xj :
Thus, the forced motion equation of the structure can be where q0 c and S are the air characteristic impedance and
deduced with the complex stiffness matrix as the structural surface area. The three-dimensional incidence
� � � plane sound waves form of pi is
KS ðxÞ x2 MS u� ¼ FA (24)
� ��
where K�S ðxÞ, MS and FA are the same variables as in Eq. ^ i exp
pi ðx, y, zÞ ¼ P i kx x þ ky y þ kz z (30)
(1). u� is the complex generalized displacement vector. The
eigenvalue solution for the free vibration analysis is nonlin­ The diffuse sound field is characterized by the sum of N
ear, and can be written as uncorrelated plane waves propagating in random directions.
� � �
KS ðxÞ k� MS U� ¼ 0 (25) Therefore, pi is

where the complex modal shape is U� , and k� is the associ­ X

N
� ��
ated complex eigenvalue with the expression as ^ i p1ffiffiffiffi
pi ðx, y, zÞ ¼ P exp i kxn x þ kyn y þ kzn z (31)
�
N n¼1
k ¼ kð1 þ igÞ (26)
pffiffiffiffiffiffiffiffiffiffi
where xn ¼ RðkÞ and g ¼ IðkÞ=RðkÞ is the damped In addition, the sound radiated power in the semi-infinite
modal eigenfrequency and corresponding modal damping fluid domain can be calculated as
factor, respectively. ð
1 X Se
As is shown in Table 1, An iterative procedure called the Wr ðh, uÞ ¼ P Rð^p r v�n Þd ¼ RhvH ð H Þ H
n Z Z vn i ¼ vn Rvn
iMSE method is adopted to obtain the solution. In this 2 SA 4
SA
scheme, a fixed-point iteration method is employed, starting (32)
from an initial guess for the real eigenproblem Kð0Þ, with a
pre-set convergence tolerance d: where the superscript � and H is the complex conjugate and
kxr xr 1 k Hermitian transpose. ^p r is sound pressure, and vn are the
�d (27) complex normal velocity amplitudes. Se is each element’s
xr 1
area. R is the real part of radiation impedance matrix Z,
where xr 1 and xr are the initial and updated values of the
named radiation resistance matrix. Moreover, the fluid load­
real part eigenfrequency for mode r:
ing effect can be indicated by the imaginary part of Z:
Finally, the structure-acoustic fully coupled governing equa­
2.5 Vibroacoustic indicators tion can be solved numerically by combining Eqs. (6)–(20) and
Eqs. (28)–(32), which constitute the sub-LW approach proposed
Finally, the STL represents the structural sound insulation
in our study and are the main contribution of this paper.
performance, and is given by
� �
Wi
STL ¼ 10 log (28) 3. Model validation
Wr
In this section, three cases with previously published numer­
ical and experimental data are selected for validation of the
Table 1. Algorithm of the iMSE method solving the nonlinear eigenvalue convergence and accuracy of the proposed sub-LW FE mod­
problem. eling method. The first two cases are on the free vibration
Algorithm: The iMSE procedure to determine the eigenfrequency and modal analysis and the last is on the STL prediction analysis,
loss factor of mode r.
1 Initial guess: xr ¼ x0r
respectively. For the first two cases on the free vibration
2 for j ¼ 1 : N do analysis, the natural frequency of a five-layer laminated plate
3 if dj � d do > used convergence: d ¼ 1e 6 comprising a thick soft frequency independent core and of a
4 Input the material properties of VEM at xjr 1 three-layer CLD plate with frequency dependent VEM core
5 Calculate: KS ðxÞ, MS
6 Solve ½RðK�S ðxÞ k� MS Þ�U ¼ 0 are investigated, respectively. And the study of mesh conver­
7 Update the eigenfrequency and modal shapes: xr ¼ xjr , Ur ¼Ujr gence is taken for analysis of the five-layer laminated plate.
8 Check convergence: dj ¼ kxrxrx1r 1 k � d For the third case on the STL prediction analysis, the STL
9 end if
10 The damped eigenfrequency: xr ¼ xjr
of a complex triple-core composite plate is evaluated and
Uj T IðK ðx ÞÞUj compared with the published analytical and experimental
The damped modal loss factor: gr ¼ Ujr T RðKS ðxr ÞÞUrj > converged solution
11 end for
r S r r results. The in-plane and out-of-plane displacement expan­
sion orders are set fm, ng ¼ f3, 2g for all cases.
8 B. WANG ET AL.

3.1. Free vibration analysis Figure 4 presents the results of the mesh convergence
analysis of the proposed FE modeling method. The mesh
3.1.1. Case 1: a five-layer laminated plate with a thick
density is set to 4 � 4, 8 � 8, 12 � 12, 16 � 16, 20 � 20,
soft core
24 � 24, and 28 � 28. It is clear that rapid convergence
The free vibration of a widely used 5-layer (0� /90� /core/0� /90� )
occurs as the mesh density increases. A satisfactory conver­
asymmetric square plate with four-edges simply supported
gence value is achieved for the mesh density of 24 � 24,
boundary condition [70–73] is investigated firstly. The study of
with an appropriate computational cost.
in-plane mesh convergence is also carried out. The surface
Table 2 lists the results compared with several other model­
layer is composed of anisotropic material, and the core layer is
ing methods in the literature, such as Vidal et al.’s LW
composed of softer isotropic material. It should be noted that
approach solved by the variable separation approach [70], the
in this case, the core’s frequency dependency is not considered,
Zig-zag method by Zhen et al. [71], the ESL approach by Kant
in order to maintain the consistency with the material parame­
et al. [72] and the exact solution by Rao et al. [73]. Excellent
ters in the literature. The thickness ratio of face layer to core
agreement is noticed between the results obtained from the
layer (tc =tf ) is 10, and the aspect ratio (a=h) is set as 10 and
proposed method and those reported by Vidal et al. [70] and
100, which is a good representation of the situation of thick
the exact solution [73], which validates the accuracy of the
and thin plates. The material parameters for the face layers
proposed method. The results obtained using the ESL
and core are listed as:
approach demonstrate significantly higher values compared to
Face layers: E1 ¼ 131 GPa, E2 ¼ E3 ¼ 10:34 GPa, G13 ¼
other models, and the percentage error for the thick plate
6:205 GPa, G12 ¼ G23 ¼6:895 GPa, � 12 ¼ � 13 ¼ 0:22, � 23 ¼
(a=h ¼ 10) is greater than that of the thin plate (a=h ¼ 100)
0:49, qf ¼ 1627 kg=m3 ;
in this situation. The main reason shall be attributed to the
Core: E1 ¼ E2 ¼ E3 ¼ 6:89 MPa, G12 ¼ G13 ¼ G23 ¼
large interlayer stiffness difference between the surface layers
3:45 MPa, � ¼ 0, qc ¼ 97 kg=m3 :
The first six non-dimensional fundamental frequency and the thick soft core, and the ESL approach fails to accur­
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ately capture local deformation such as compression and
(-mn ¼ xmn a2 =h ðq=E2 Þf ) is chosen to compare. stretching of the thick soft core with an overestimation of the
structural bending stiffness. This also demonstrates that the
LW approach is more suitable for composite CLD plates.

3.1.2 Case 2: a three-layer CLD plate with frequency

dependent VEM core
In this case, the accuracy of the adopted iMSE method in
addressing the complex eigenvalue is assessed by conducting
the analysis of the natural frequency and modal loss factor of
a three-layer CLD plate with a simply supported boundary
condition. The plate is composed of isotropic layers and sym­
metric, with a VEM core and surface layers having thicknesses
of 0.254 mm and 0.762 mm, respectively. The geometric
dimension of the plate is 0.348 m � 0.3048 m. Aluminum is
used for the surface layers (Young’s modulus E ¼ 68:9 GPa,
Poisson’s ratio � ¼ 0:3, and mass density q ¼ 2740 kg=m3 ).
A kind of VEM typed 3 M-ISD112 (Poisson’s ratio � ¼ 0:49,
Figure 4. Convergence of the (3,3) modal non-dimensional frequency of the and mass density q ¼ 1600 kg=m3 ) is used for the core with fre­
five-layer antisymmetric laminated plate with a=h ¼ 10:
quency dependency, whose shear modulus is fitted with the ADF

Table 2. The non-dimensional frequencies of the five-layer antisymmetric (0� /90� /core/0� /90� ) laminated plate with tc =tf ¼ 10: The “present” denotes the pro­
posed modeling method in this paper. The percentage error (%) is provided within parentheses when compared to the exact solution.
Mode
a=h Model (1,1) (1,2) (2,2) (1,3) (2,3) (3,3)
10
Present f3, 2g 1.8492 (0.1) 3.2226 (0.1) 4.2938 (0.1) 5.2331 (0.2) 6.1046 (0.2) 7.6919 (0.2)
Vidal et al. (VS-LD4) 1.8492 (0.1) 3.2227 (0.1) 4.2941 (0.1) 5.2341 (0.2) 6.1056 (0.2) 7.6933 (0.2)
Zig-zag (Zhen et al.) 1.9445 (5.2) 3.3796 (5.0) 4.5914 (7.0) 5.5268 (5.8) 6.5128 (6.9) 8.4311 (9.8)
ESL (Kant et al.) 4.8594 (162) 4.8594 (162) 10.2966 (140) 11.7381 (124) 13.4706 (121) 16.1320 (110)
Exact (Rao et al.) 1.8480 (0.0) 3.2196 (0.0) 4.2894 (0.0) 5.2236 (0.0) 6.0942 (0.0) 7.6762 (0.0)
100
Present f3, 2g 11.9458 (0.0) 23.4144 (0.1) 30.9605 (0.1) 36.1664 (0.1) 41.4734 (0.1) 49.0361 (1.5)
Vidal et al. (VS-LD4) 11.9469 (0.1) 23.4233 (0.1) 30.9723 (0.1) 36.1898 (0.1) 41.4961 (0.1) 49.0742 (1.5)
Zig-zag (Zhen et al.) 11.9840 (0.4) 23.3427 (0.3) 31.8651 (3.0) 36.5671 (1.2) 42.2904 (2.0) 50.0815 (0.6)
ESL (Kant et al.) 15.5093 (30) 39.0293 (67) 54.7618 (77) 72.7572 (101) 83.4412 (101) 105.3781(111)
Exact (Rao et al.) 11.9401(0.0) 23.4017 (0.0) 30.9432 (0.0) 36.1434 (0.0) 41.4475 (0.0) 49.7622 (0.0)
 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES 9

model in Eq. (23) at 27 � C, and the series terms ðDi , Xi Þ are pre­ On this triple-core composite plate, the geometry dimen­
sented in Table 3. Figure 5 shows the spectra of the shear modu­ sion is 0.840 m � 0.840 m, and the thickness is 21.68 mm.
lus and damping loss factor. This configuration is stacked symmetrically by 13 plies: four
Table 4 presents the first four-order natural frequency CFRP plies, a Nomex honeycomb core, and a melamine
and modal loss factor of the investigated CLD plate. Results foam with the glue layer (Figure 6). Detailed material
show good agreement between the proposed model and the parameters can be found in Ref. [25]. The damping loss fac­
previous models, and the predictions of the proposed model tors of all materials are considered to be frequency inde­
exhibit closer agreement with those of Ferreira et al. [37]. pendent in this case. Previous work on free vibration
This further demonstrates the validity of the proposed model. analysis [25, 41] shows that the through-thickness modes
due to weak melamine foam core appear at a frequency
higher than 620 Hz. In this case, the third-order in-plane
3.2. STL prediction displacement and the second-order through-thickness dis­
In the third case, STL predictions according to a triple-core placement are considered, in which the four CFRP surface
composite plate are conducted to validate the proposed sub- layers are modeled as one numerical sublaminate. Moreover,
LW modeling method, where the free vibration of the all the transverse-shear and stretching deformations of the
selected triple-core composite plate was initially examined melamine foam core are considered.
by Gorgeri et al. [41] and D’Ottavio et al. [25]. Additionally, Figure 7 presents comparisons among the STL spectra
the STL of the composite plate was studied by Yang et al. from 10 Hz to 2 kHz predicted with different models by
[74] through comparisons among the wave-based FE (WFE) Yang et al. [74] and the experimental data by Dozio et al.
method, the transfer matrix method (TMM), and the experi­ [75]. It is observed that the (1,1) order resonance frequency
mental results by Dozio et al. [75]. So, these published data of the plate is 19 Hz, and the coincidence frequency is
according to this triple-core composite plate can provide 618 Hz. The STL spectrum below the (1,1) order resonance
additional benchmark results for the validity of the proposed frequency is attributed to the stiffness control region, while
method. the frequency band between the (1,1) order resonance fre­
quency and the coincidence frequency is influenced by the
Table 3. The ADF series terms at 27 � C of the structural resonance modes, with the overall trend following
VEM typed 3 M-ISD112 [20]. the mass law. The frequency band above the coincidence
i Di Xi ðrad=sÞ frequency falls within the damping control region. It is clear
1 0.746 468.7 that all the three models demonstrate great agreement with
2 3.265 4742.4 the experimental data. The TMM solution is unable to pre­
3 43.284 71532.5
dict the effect of modal resonance. Excellent agreement is
observed between the present model and the WFE model
proposed by Yang et al. [74]. Although both models can
consider the influence of modal resonance on STL, there are
still subtle differences. This mainly arises from Yang et al.’s
model being semi-analytical and not accounting for the cou­
pling effect of the structure and sound field, which

Figure 5. Master curve of VEM typed 3 M-ISD 112 fitted with the ADF model at
27 � C [20]. Figure 6. Sketch of the stacking sequence of the triple-core composite plate.

Table 4. The first four-order natural frequency and modal loss factor for case 2.
Trindade et al. [20] Ferreira et al. [37] (12 �12 Q9) Proposed model f3, 2g
Mode (m, n) fn (Hz) g fn (Hz) g fn (Hz) g
1,1 53.77 0.213 53.66 0.235 53.67 0.238
2,1 110.31 0.272 107.84 0.291 107.73 0.298
1,2 126.72 0.283 123.17 0.297 124.06 0.290
2,2 176.97 0.289 171.81 0.309 171.40 0.259
10 B. WANG ET AL.

significantly affects the STL at resonances [76]. In all, the ±45� , which is commonly used in engineering to meet the
present modeling method is proven to be accurate for pre­ stiffness guarantee condition. Given that the four CFRP sur­
dicting STL. face layers have identical thickness and material properties,
and considering the proven accuracy of the present model
in Section 3, the ESL approach is employed to model the
4. Numerical examples four CFRP surface layers, while the LW approach is utilized
Since the accuracy of the present method on the dynamic for the other layers. The displacement field, denoted as
response of composite structures, as well as the iMSE fm, ng ¼ f3, 2g, is adopted for each numerical layer. The
method for solving the classical three-layer CLD plate has thickness and stacking sequence of the plate is provided in
been already demonstrated, the cases studied in this section Table 5. The plate has a geometric dimension of
are mainly concerned with composite CLD plates with fre­ 0.455 m � 0.386 m, and the four-edges simply supported
quency dependence. Firstly, a parametric study of the VEM boundary condition is applied. The VEM designated as 3 M-
interlayer’s thickness and stiffness based on the free vibra­ ISD112 in Section 3.1.2 is also selected for the VEM inter­
tion analysis is conducted in Section 4.1. Then, Section 4.2 layer. Table 6 lists the remaining material properties of the
concerns the STL characteristics under normal and oblique composite CLD plate. Poisson’s ratio is a mechanical prop­
plane sound wave excitation. erty of a material that represents the ratio of lateral strain to
axial strain when subjected to uniaxial stress, which can be
obtained from tensile tests. Determining the Poisson’s ratio
4.1. Parametric study of orthotropic materials (e.g. CFRP) can be relatively com­
In this section, the free vibration of a 13-layer composite plex due to the significant variations in material properties
CLD plate is analyzed. The surface layer consists of four across different directions. Here, Poisson’s ratio of CFRP is
CFRP layers arranged in the balanced lay-up method of derived from the test data of the material manufacturer
Hexcel and Liebig et al. [10]. Additionally, its transverse
shear modulus G23 is calculated using the rule of mixture,
assuming a Poisson’s ratio � 32 � 0.4.
Here, the influence of the VEM interlayer’s thickness
and stiffness variation on the natural frequency and corre­
sponding modal loss factor is discussed. The modal loss
factor quantitatively represents the structural energy dissi­
pation capacity. The results of parametric study can pro­
vide guidance for engineering practitioners in designing
the CLD plate configuration with the optimal mode loss
factor.
Figure 8 shows the changing ratio of the natural fre­
quency and modal loss factor with the variation of VEM
interlayer’s thickness and stiffness scaling factors. The nor­
malization is based on the specimen with the VEM inter­
layer’s thickness of 2.25 mm in Table 5. As depicted in
Figure 8(a), the natural frequency exhibits a gradual
decrease as the VEM interlayer’s thickness scaling factor (j)
varies from 0.01 to 10. The main reason is that the bending
Figure 7. Comparison of the STL spectra among the present model, the TMM
and WFE model of Yang et al. [74], and the experimental results of Dozio stiffness of the structure decreases with the increasing pro­
et al. [75]. portion of the softer VEM interlayer. However, when the

�
645
Table 5. Stacking sequence and thickness of the composite CLD plate specimen, which is nomenclature as ½ðC0:28 �
Þ2 =G0:1 =S0:65 =VEM 2:25 �s :
Physical ply No. Numerical layer No. Material Ply angle Ply thickness (mm)
1-4, 10-13 1, 7 CFRP 45� /-45� /45� /-45� 0.28
5, 8 2, 6 Glue – 0.1
6, 9 3, 5 Steel – 0.65
7 4 VEM – 2.25

Table 6. Material properties of the composite plate without VEM interlayer.

Young’s modulus (GPa) Shear modulus (GPa) Poisson ratio
Material q (kg/m3) g (%) E1 E2 E3 G12 G13 G23 � 12 � 13 � 23
CFRP 1496 0.1 113.7 7.75 7.75 3.76 3.76 2.75 0.34 0.34 0.4
Glue 1025 0.01 1.87 0.78 0.45
Steel 7800 0.1 210 – 0.3
 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES 11

scale factor is less than 1, the modal loss factor increases This local deformation can significantly increase the energy
firstly, then decreases, and finally increases almost exponen­ dissipation of the VEM interlayer.
tially at the scale factor greater than 1. There is an optimal From Figure 8(b), it can be observed that as the stiffness
thickness scaling factor that allows using a thinner VEM scaling factor (1) of the VEM interlayer varies in the range
interlayer to achieve a relatively higher modal loss factor. from 0.1 to 100, the natural frequency monotonically
For example, when the thickness scaling factor is 0.05, the increases. However, an optimal stiffness matching exists for
first modal loss factor can reach its maximum in the thick­ achieving higher modal loss factors. And specific stiffness
ness scaling range from 0.01 to 0.1. The cohesive force gen­ scaling factor differs for attaining the maximum loss factor
erated by the molecular structure is relatively weak with a value of different modes. For the maximum value of the first
thin VEM interlayer, and the constrained surface layer on mode, the stiffness scaling factor is 10. For the maximum
both sides leads to a greater degree of shear deformation. value of the 2nd and 3rd modes, stiffness scaling factor is
However, when the VEM interlayer is too thin or too thick, approximately 30. The stiffness of VEM would directly
the change of modal loss factor is not only caused by the impact its viscosity, thereby influencing the shear strain
deformation but mainly due to the proportion of VEM in energy. When the optimum stiffness is matched, the cohe­
the whole structure, that is to say, the appearance of high sive force of VEM is strong, enabling greater energy dissipa­
model loss factor is determined by the inherent damping of tion through shear deformation. Conversely, if the VEM
VEM. Meanwhile, another interesting phenomenon is that stiffness is excessively high, it becomes rigid and resistant to
the higher-order mode loss factor is more sensitive to the deformation. This leads to a reduction in the modal loss fac­
thickness scaling factor, mainly due to the higher-order tor, with the interlayer stiffness discrepancy becoming small
modes containing more bending and torsional deformation. and deviating from the underlying principle of CLD.

Figure 8. Parametric effects of the VEM interlayer’s thickness and stiffness scaling factor on the first six-order natural frequency and modal loss factor. The data is
normalized with the specimen having a VEM interlayer thickness of 2.25 mm. (a) the influence of thickness scaling factor (j). (b) the influence of stiffness scaling fac­
tor (1).
12 B. WANG ET AL.

4.2. STL characteristics and discussion Figures 9–12 show the narrow band STL spectra of the
CLD plate from 10 Hz to 2 kHz with a sweep interval of
The STL characteristics of the composite CLD plate are investi­
2 Hz under various incidence angles, respectively. The reson­
gated relying on the parametric study in Section 4.1. The influence
ant frequencies of the structure with obvious influence on
of the VEM interlayer’s thickness on STL is discussed. The config­
STL are also marked out. It can be found that low-order
uration used here is consistent with that in Section 4.1 with a
odd-odd modes, such as (1,1), (3,1), and (1,3), exert sub­
larger aspect ratio of 0.762 m � 0.483 m. The thickness scaling
stantial impacts on sound insulation performance, regardless
factor j of the VEM interlayer varies in four cases: 0, 0.01, 0.02,
of normal and oblique incidence. Moreover, this effect
and 0.3. The reference case corresponds to the absence of a
becomes more pronounced with higher elevation angles at
VEM interlayer (j ¼ 0). The results are presented for both nor­
oblique incidence. In practical scenarios where the max­
mal and several oblique incidence cases under plane sound
imum incidence angle (h ¼ 858) is typically encountered,
wave excitation. For the normal incidence case, the elevation
the dip in the STL spectrum attributed to the (1,1) mode is
and azimuth angle is fixed as h ¼ / ¼ 0o ; For the case of
even lower than 0 dB. This indicates that the coupling
oblique incidence, the azimuth angle is kept constant at / ¼45� ,
between the incident sound field and the structure induces a
while the elevation angle varies h ¼ 308 , 608 , and 858 :
more pronounced structural resonance.

Figure 11. STL spectra of the composite CLD plate with different VEM
Figure 9. STL spectra of the composite CLD plate with different VEM interlayer interlayer thickness scaling factor (j) under oblique plane sound excita­
thickness scaling factor (j) under normal plane sound excitation (h¼ 08 , /¼ 08 ). tion (h¼ 608 , /¼ 458 ).

Figure 10. STL spectra of the composite CLD plate with different VEM Figure 12. STL spectra of the composite CLD plate with different VEM
interlayer thickness scaling factor (j) under oblique plane sound excita­ interlayer thickness scaling factor (j) under oblique plane sound excita­
tion (h¼ 308 , /¼ 458 ). tion (h¼ 858 , /¼ 458 ).
 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES 13

Figure 13. Deflection bending mode shape of the composite CLD plate with different VEM interlayer’s thickness scaling factors under oblique plane sound excita­
tion (h ¼ 308 , / ¼ 458 ) under the influence of the (1,3) mode. The colormap in (a) and (b) is adjusted to the same range. (a) the thickness scaling factor j ¼ 0: (b)
the thickness scaling factor j ¼ 0:01: (c) the thickness scaling factor j ¼ 0:02:

For the situation of normal incidence, the sound pressure significant influence of odd-even and even-odd modes such
is evenly distributed on the surface layer of the structure, as (1,4) and (4,1), as well as even-even modes such as (2,4)
and only the odd-odd modes are excited. When the VEM and (2,6), on sound insulation.
interlayer’s thickness j ¼ 0:1, the influence of higher-order To further elucidate the VEM interlayer’s dissipative
odd-odd modes such as (5,1), (1,5), and (9,1) on STL cannot mechanism on sound energy, a comparison is made of the
be ignored (Figure 9). Moreover, the higher-order modes deflection bending mode controlled by the (1,3) mode with
tend to be denser and can exhibit degeneracy effects, leading different VEM interlayer’s thickness scaling factors (0, 0.01,
to a reduction in insulation performance at multiple con­ and 0.02) at a 30� oblique incidence. As is shown in Figure 13,
tinuous frequency points. For example, the merger between the forward direction of normal displacement (w > 0) faces
(5,7) and (11,1) produces an average reduction of about the incident acoustic domain. For j ¼ 0, the deflected mode
8 dB between 1250 Hz and 1360 Hz. Upon the addition of shape of the forced mode has no obvious change, and the dis­
VEM interlayer, the energy used for acoustic radiation is placement amplitudes of the incident direction and the trans­
significantly attenuated through shear deformation. Notably, mission direction are equal. For j ¼ 0:01 and 0.02, the
two significant improvements to STL are observed. Firstly, amplitudes of the two configurations are approximately two
the dip of STL spectrum at the low-order resonance fre­ magnitudes lower (10 2) compared to the case without a VEM
quencies is significantly mitigated, even with a thin VEM interlayer.
interlayer. Specifically, for the (1, 1) mode, the correspond­ Under the excitation of an incident sound field, the
ing modal loss factor for the four thickness scaling factor vibration energy is dissipated into internal energy by the
(ranging from low to high) are 0.000857, 0.06414, 0.10813, deformations of VEM, which significantly reduces the amp­
and 0.22694. When the thickness scaling factor is j ¼ 0:3, litude of the deflection mode shape. The range of color bars
the dip of STL spectrum caused by (1,1) mode can be in Figure 12(b) and (c) have been standardized. Notably,
greatly improved, effectively negating the influence of higher there are substantial changes in the amplitude distributions
modes. Although the increase of VEM interlayer thickness of the deflection mode shape. More amplitude distribution
leads to a decrease in overall bending stiffness, the STL in is observed pointing toward the incidence direction, aiding
the stiffness control region experiences a reduction of in the reflection of more sound waves. For j ¼ 0:01, the
approximately 5 dB. Secondly, it compensates for the wide- incidence direction has a larger displacement amplitude.
band STL reduction caused by higher-order modal degener­ Meanwhile, the configuration j ¼ 0:02 has a smaller ampli­
acy. Given that the amplitude of the higher-order resonance tude in the transmitted direction. This further indicates that
modes is generally smaller than that of the lower-order the addition of a VEM interlayer increases the modal loss
modes, a lower modal loss factor allows for a smoother nar­ factor and greatly reduces the vibration. Due to the strong
rowband STL spectral at the middle and high frequencies. It interlayer strain difference, when the energy of the incident
is feasible to achieve stable sound insulation performance plane wave is transferred between the layers, a large percent­
for real sound sources with different spectra. age is dissipated through the shear and the stretch-bending
Although the influence of azimuth angle on STL charac­ deformation of the VEM interlayer, which results in an evi­
teristics has been found to be negligible in oblique incidence dent alteration in the deflection mode shape. The deflected
situations, it should be noted that the azimuth angle is not bending modes, characterized by out-of-plane displacements,
0� here, which represents the typical scenario in a three- can exhibit strong coupling with the air surrounding the
dimensional sound field. Due to the occurrence of incident structural top and bottom surfaces. As a result, the STL
sound wave components in x and y direction, an expanded characteristics are influenced by these coupling effects.
range of resonant modes can be excited, such as odd-even,
even-odd order, and even-even order modes. The normal
5. Conclusions
component of acoustic pressure assigned to each element
node on the structure surface becomes smaller during In this paper, a refined FE modeling method based on the
oblique incidence, leading to an overall decrease in the STL sub-LW approach is proposed to analyze the vibroacoustic
spectrum. While different plate configurations exhibit simi­ behavior of composite CLD plates. The main advantage of
lar STL characteristics, it is crucial to consider the this approach lies in its flexibility to handle various plate
14 B. WANG ET AL.

configurations. To balance the computational cost generated Hequn Min http://orcid.org/0000-0002-8027-6204

by the DoFs and accuracy, the ESL and LW approaches can Ting Qu http://orcid.org/0000-0001-9062-8106
be combined into one model, as well as considering different
expansion orders of displacement fields for each sublaminate Data availability statement
freely. The VEM interlayer’s frequency dependency is charac­
terized by the ADF model. And the complex eigenvalue prob­ Data will be made available on request.
lem is solved by the iMSE method. The proposed method in
this paper is considerably suitable for high-precision predic­ References
tion of STL characteristics of composite CLD plates.
A parametric numerical study of free vibration is con­ 0[1] M.D. Rao, Recent applications of viscoelastic damping for noise
control in automobiles and commercial airplanes, J. Sound
ducted. Results show that the variation of VEM interlayer’s Vib., vol. 262, no. 3, pp. 457–474, 2003. DOI: 10.1016/s0022-
thickness and stiffness have significant effects on natural fre­ 460x(03)00106-8.
quency and modal damping loss factor. One can find that 0[2] R. Talebitooti, and M.R. Zarastvand, Vibroacoustic behavior of
there is an optimal thickness value within a relatively thin orthotropic aerospace composite structure in the subsonic flow
range for the VEM interlayer to obtain a relatively high considering the Third order Shear Deformation Theory,
Aerosp. Sci. Technol., vol. 75, pp. 227–236, 2018. DOI: 10.1016/
modal loss factor, as well as an optimal stiffness matching.
j.ast.2018.01.011.
Furthermore, numerical investigations are carried out on the 0[3] Y. Yan, and L. Zhang, Variable-kinematic finite elements for
narrow band STL characteristics of composite CLD plates the aero-thermo-elastic analysis of variable stiffness composite
under plane sound wave excitation. The primary focus of laminates, Mech. Adv. Mater. Struct., pp. 1–17, 2022. DOI: 10.
the investigations is to demonstrate the sound insulation 1080/15376494.2022.2119312.
enhancement mechanism of VEM interlayer. Results show 0[4] H. Kaidouchi, S. Kebdani, and S.A. Slimane, Vibro-acoustic
analysis of the sound transmission through aerospace composite
that not only a large amount of vibration power can be dis­
structures, Mech. Adv. Mater. Struct., vol. 30, no. 19, pp. 3912–
sipated through the VEM interlayer’s shear and stretching 3922, 2023. DOI: 10.1080/15376494.2022.2085348.
deformation, but deflection bending mode shapes of the 0[5] A. Treviso, B. Van Genechten, D. Mundo, and M. Tournour,
CLD plate can also be altered. Therefore, the dips of STL Damping in composite materials: properties and models,
spectrum can be enhanced or even eliminated at resonant Compos. Part B: Eng., vol. 78, pp. 144–152, 2015. DOI: 10.
frequencies through the two main damping characteristics of 1016/j.compositesb.2015.03.081.
0[6] A. Oulmane, and A. Ross, Effects of material parameters on the
the viscoelastic interlayer in the CLD plate. In the future, transient dynamics of an impacted plate with partial con­
the work in this paper is going to be extended for curved strained layer damping treatment, J. Acoust. Soc. Am., vol. 147,
panels, and also to be combined with multi-objective opti­ no. 3, pp. 1939–1952, 2020. DOI: 10.1121/10.0000890.
mization algorithms to identify simultaneously the optimal 0[7] E. M. Kerwin Jr, Damping of flexural waves by a constrained
parameters of the stack sequence and structural stiffness for viscoelastic layer, J. Acou. Soc. Am., vol. 31, pp. 952–962, 1959.
DOI: 10.1121/1.1907821.
sound insulation enhancement of CLD plates.
0[8] Eric E. Ungar, and Edward M. Kerwin, Jr., Loss factors of
viscoelastic systems in terms of energy concepts, J. Acou. Soc.
Am., vol. 34, no. 7, pp. 954–957, 1962. ; DOI: 10.1121/1.
Author contributions 1918227.
Conceptualization, Hequn Min; methodology, Hequn Min and Bo 0[9] X.Q. Zhou, D.Y. Yu, X.Y. Shao, S.Q. Zhang, and S. Wang,
Wang; software, Bo Wang; validation, Bo Wang, and Ting Qu; investi­ Research and applications of viscoelastic vibration damping
gation, Bo Wang and Hequn Min; resources, Hequn Min; writing-ori­ materials: a review, Compos. Struct., vol. 136, pp. 460–480,
ginal draft preparation, Bo Wang and Hequn Min; writing-review and 2016. DOI: 10.1016/j.compstruct.2015.10.014.
editing, Hequn Min; visualization, Bo Wang; supervision, Hequn Min; [10] W.V. Liebig, V. Sessner, K.A. Weidenmann, and L. K€arger,
project administration, Hequn Min; funding acquisition, Hequn Min. Numerical and experimental investigations of the damping
All authors have read and agreed to the published version of the behaviour of hybrid CFRP-elastomer-metal laminates, Compos.
manuscript. Struct., vol. 202, pp. 1109–1113, 2018. DOI: 10.1016/j.comp­
struct.2018.05.051.
[11] A. Jackstadt, W.V. Liebig, and L. K€arger, Analytical modeling
Disclosure statement and investigation of constrained layer damping in hybrid lami­
nates based on a unified plate formulation, Int. J. Mech. Sci.,
The authors declare that they have no known competing financial vol. 216, pp. 106964, 2022. DOI: 10.1016/j.ijmecsci.2021.106964.
interests or personal relationships that could have appeared to influ­ [12] C. Zheng, Y. Zhai, and S. Liang, Improving damping perform­
ence the work reported in this paper. ance of fiberglass-reinforced resin matrix composites by design­
ing viscoelastic sandwich layer, Mech. Adv. Mater. Struct.,
vol. 28, no. 21, pp. 2202–2209, 2021. DOI: 10.1080/15376494.
Funding 2020.1722873.
This work is supported by the Natural Science Foundation of China [13] Y. Koutsawa, W.L. Azoti, S. Belouettar, R. Martin, and E.
(Grant No. 51408113) and the Natural Science Foundation of Jiangsu Barkanov, Loss behavior of viscoelastic sandwich structures: a
Province, China (Grant No. BK20140632). statistical-continuum multi-scale approach, Compos. Struct.,
vol. 94, no. 4, pp. 1391–1397, 2012. DOI: 10.1016/j.compstruct.
2011.11.003.
ORCID [14] H. Daoud, A. El Mahi, J.-L. Rebi�ere, M. Taktak, and M.
Haddar, Characterization of the vibrational behaviour of flax
Bo Wang http://orcid.org/0000-0001-9157-8518 fibre reinforced composites with an interleaved natural
 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES 15

viscoelastic layer, Appl. Acoust., vol. 128, pp. 23–31, 2017. DOI: [31] A.S. Sayyad, and Y.M. Ghugal, On the free vibration analysis of
10.1016/j.apacoust.2016.12.005. laminated composite and sandwich plates: a review of recent lit­
[15] M. Hamdaoui, K. Ledi, G. Robin, and E.M. Daya, Identification erature with some numerical results, Compos. Struct., vol. 129,
of frequency-dependent viscoelastic damped structures using an pp. 177–201, 2015. DOI: 10.1016/j.compstruct.2015.04.007.
adjoint method, J. Sound Vib., vol. 453, pp. 237–252, 2019. [32] H. Hu, S. Belouettar, M. Potier-Ferry, and E.M. Daya, Review
DOI: 10.1016/j.jsv.2019.04.022. and assessment of various theories for modeling sandwich com­
[16] R. Lewandowski, and B. Chora˛z_ yczewski, Identification of the posites, Compos. Struct., vol. 84, no. 3, pp. 282–292, 2008.
parameters of the Kelvin–Voigt and the Maxwell fractional DOI: 10.1016/j.compstruct.2007.08.007.
models, used to modeling of viscoelastic dampers, Comput. [33] E. Carrera, F. Miglioretti, and M. Petrolo, Guidelines and rec­
Struct., vol. 88, no. 1-2, pp. 1–17, 2010. DOI: 10.1016/j.comp­ ommendations on the use of higher order finite elements for
struc.2009.09.001. bending analysis of plates, Int. J. Comput. Methods Eng. Sci.
[17] D.-L. Chen, P.-F. Yang, and Y.-S. Lai, A review of three-dimen­ Mech., vol. 12, no. 6, pp. 303–324, 2011. DOI: 10.1080/15502287.
sional viscoelastic models with an application to viscoelasticity 2011.615792.
characterization using nanoindentation, Microelectron. Reliab., [34] A. Alesadi, M. Galehdari, and S. Shojaee, Free vibration and
vol. 52, no. 3, pp. 541–558, 2012. DOI: 10.1016/j.microrel.2011. buckling analysis of composite laminated plates using layerwise
10.001. models based on isogeometric approach and Carrera unified for­
[18] Y. Wang, and D.J. Inman, Finite element analysis and experi­ mulation, Mech. Adv. Mater. Struct., vol. 25, no. 12, pp. 1018–
mental study on dynamic properties of a composite beam with 1032, 2018. DOI: 10.1080/15376494.2017.1342883.
viscoelastic damping, J. Sound Vib., vol. 332, no. 23, pp. 6177– [35] K. Koutoati, F. Mohri, E.M. Daya, and E. Carrera, A finite
6191, 2013. DOI: 10.1016/j.jsv.2013.06.016. element approach for the static and vibration analyses of func­
[19] F. Renaud, J.-L. Dion, G. Chevallier, I. Tawfiq, and R. Lemaire, tionally graded material viscoelastic sandwich beams with non­
A new identification method of viscoelastic behavior: applica­ linear material behavior, Compos. Struct., vol. 274, pp. 114315,
tion to the generalized Maxwell model, Mech. Syst. Sig. 2021. DOI: 10.1016/j.compstruct.2021.114315.
Process., vol. 25, no. 3, pp. 991–1010, 2011. DOI: 10.1016/j. [36] A.J.M. Ferreira, C.M.C. Roque, E. Carrera, M. Cinefra, and O.
ymssp.2010.09.002. Polit, Bending and vibration of laminated plates by a layerwise
[20] M.A. Trindade, A. Benjeddou, and R. Ohayon, Modeling of fre­ formulation and collocation with radial basis functions, Mech.
quency-dependent viscoelastic materials for active-passive vibra­ Adv. Mater. Struct., vol. 20, no. 8, pp. 624–637, 2013. DOI: 10.
tion damping, J. Vib. Acoust., vol. 122, no. 2, pp. 169–174, 1080/15376494.2011.643282.
2000. DOI: 10.1115/1.568429. [37] A.J.M. Ferreira, A.L. Ara� ujo, A.M.A. Neves, J.D. Rodrigues, E.
[21] M. Ziapkoff, L. Duigou, G. Robin, J.-M. Cadou, and E.M. Daya,
Carrera, M. Cinefra, C.M. Mota Soares, A finite element model
A high order Newton method to solve vibration problem of
using a unified formulation for the analysis of viscoelastic sand­
composite structures considering fractional derivative Zener
wich laminates, Compos. Part B: Eng., vol. 45, no. 1, pp. 1258–
model, Mech. Adv. Mater. Struct., pp. 1–11, 2023. DOI: 10.
1264, 2013. DOI: 10.1016/j.compositesb.2012.05.012.
1080/15376494.2022.2161115.
[38] Z. Zhou, M. Chen, and Y. Li, Application scope studies of ana­
[22] W. Cai, P. Wang, and Y. Zhang, Fractional-order model for
lytical methods based on the equivalent single-layer and layer­
temperature-dependent rheological behaviors of polymeric
wise theories for vibration analysis of symmetric sandwich
materials, Mech. Adv. Mater. Struct., pp. 1–12, 2023. DOI: 10.
plates, Acta Mech., vol. 234, no. 2, pp. 377–401, 2023. DOI: 10.
1080/15376494.2023.2190741.
1007/s00707-022-03392-3.
[23] M. Bilasse, E. Daya, and L. Azrar, Linear and nonlinear vibra­
[39] L. Demasi, G. Biagini, F. Vannucci, E. Santarpia, and R.
tions analysis of viscoelastic sandwich beams, J. Sound Vib.,
Cavallaro, Equivalent single layer, zig-zag, and layer wise theo­
vol. 329, no. 23, pp. 4950–4969, 2010. DOI: 10.1016/j.jsv.2010.
06.012. ries for variable angle tow composites based on the generalized
[24] J. Ro, and A. Baz, Optimum placement and control of active unified formulation, Compos. Struct., vol. 177, pp. 54–79, 2017.
constrained layer damping using modal strain energy approach, DOI: 10.1016/j.compstruct.2017.06.033.
J. Vib. Control., vol. 8, no. 6, pp. 861–876, 2002. DOI: 10.1177/ [40] M. D’Ottavio, A sublaminate generalized unified formulation for
107754602029204. the analysis of composite structures, Compos. Struct., vol. 142,
[25] M. D’Ottavio, A. Krasnobrizha, E. Valot, O. Polit, R. Vescovini, pp. 187–199, 2016. DOI: 10.1016/j.compstruct.2016.01.087.
and L. Dozio, Dynamic response of viscoelastic multiple-core [41] A. Gorgeri, R. Vescovini, and L. Dozio, Analysis of multiple-
sandwich structures, J. Sound Vib., vol. 491, pp. 115753, 2021. core sandwich cylindrical shells using a sublaminate formula­
DOI: 10.1016/j.jsv.2020.115753. tion, Compos. Struct., vol. 225, pp. 111067, 2019. DOI: 10.1016/
[26] F. Abdoun, L. Azrar, E. Daya, and M. Potier-Ferry, Forced har­ j.compstruct.2019.111067.
monic response of viscoelastic structures by an asymptotic [42] A. Gorgeri, R. Vescovini, and L. Dozio, Sublaminate variable
numerical method, Comput. Struct., vol. 87, no. 1-2, pp. 91–100, kinematics shell models for functionally graded sandwich pan­
2009. DOI: 10.1016/j.compstruc.2008.08.006. els: bending and free vibration response, Mech. Adv. Mater.
[27] J. Jith, and S. Sarkar, A model order reduction technique for Struct., vol. 29, no. 1, pp. 15–32, 2022. DOI: 10.1080/15376494.
systems with nonlinear frequency dependent damping, Appl. 2020.1749738.
Math. Modell., vol. 77, pp. 1662–1678, 2020. DOI: 10.1016/j. [43] L. Dozio, and L. Alimonti, Variable kinematic finite element
apm.2019.08.030. models of multilayered composite plates coupled with acoustic
[28] B. Sun, J. Wang, Z-l Li, Y. Qin, S-z Liu, and D. Tomker, fluid, Mech. Adv. Mater. Struct., vol. 23, no. 9, pp. 981–996,
Modified modal strain energy method for analyzing the 2016. DOI: 10.1080/15376494.2015.1121558.
dynamic damping behavior of constrained viscoelastic struc­ [44] S. Ren, G. Zhao, and S. Zhang, A layerwise finite element for­
tures, JESTR., vol. 10, no. 5, pp. 174–180, 2017. DOI: 10.25103/ mulation for vibration and damping analysis of sandwich plate
jestr.105.21. with moderately thick viscoelastic core, Mech. Adv. Mater.
[29] M.F. Caliri, A.J.M. Ferreira, and V. Tita, A review on plate and Struct., vol. 27, no. 14, pp. 1201–1212, 2020. DOI: 10.1080/
shell theories for laminated and sandwich structures highlighting 15376494.2018.1504360.
the finite element method, Compos. Struct., vol. 156, pp. 63–77, [45] Z. Xie, and W.S. Shepard, Jr, Analytical modeling and analysis
2016. DOI: 10.1016/j.compstruct.2016.02.036. of a constrained layer damped plate to examine the impact of
[30] E. Carrera, Theories and finite elements for multilayered, aniso­ the transverse compressional damping component, Mech. Adv.
tropic, composite plates and shells, ARCO., vol. 9, no. 2, pp. 87– Mater. Struct., vol. 21, no. 8, pp. 669–679, 2014. DOI: 10.1080/
140, 2002. DOI: 10.1007/BF02736649. 15376494.2012.707296.
16 B. WANG ET AL.

[46] J. Reddy, A general non-linear third-order theory of plates with [60] S. Assaf, M. Guerich, and P. Cuvelier, Vibration and acoustic
moderate thickness, Int. J. Non. Linear Mech., vol. 25, no. 6, response of damped sandwich plates immersed in a light or
pp. 677–686, 1990. DOI: 10.1016/0020-7462(90)90006-U. heavy fluid, Comput. Struct., vol. 88, no. 13-14, pp. 870–878,
[47] H. Mohamed, V. Gunasekaran, J. Pitchaimani, V. Rajamohan, 2010. DOI: 10.1016/j.compstruc.2010.04.006.
G. Kotriwar, and G. Manickam, A comprehensive damping [61] M. Guerich, and S. Assaf, Optimization of noise transmission
study of variable stiffness composite rectangular/skew laminates through sandwich structures, J. Vib. Acoust., vol. 135, no. 5,
reinforcement with curvilinear fibers by higher-order shear flex­ pp. 051010(1)- 051010(13), 2013. DOI: 10.1115/1.4024216.
ible model, Mech. Adv. Mater. Struct., vol. 30, no. 14, pp. 2806– [62] B. Wang, and T. Dou, A quasi-3D layer-wise finite element
2825, 2023. DOI: 10.1080/15376494.2022.2064014. model for vibroacoustic analysis of finite sandwich plates with
[48] R. Lewandowski, P. Litewka, and P. Wielentejczyk, Free vibra­ frequency-dependent viscoelastic core, Mech. Adv. Mater.
tions of laminate plates with viscoelastic layers using the refined Struct., vol. 30, no. 21, pp. 4419–4429, 2023. DOI: 10.1080/
zig-zag theory – Part 1. Theoretical background, Compos. 15376494.2022.2095064.
Struct., vol. 278, pp. 114547, 2021. DOI: 10.1016/j.compstruct. [63] A. Loredo, A. Plessy, A. El Hafidi, and N. Hamzaoui,
2021.114547. Numerical vibroacoustic analysis of plates with constrained-
[49] Q. Jin, Interlaminar stress analysis of functionally graded gra­ layer damping patches, J. Acoust. Soc. Am., vol. 129, no. 4,
phene reinforced composite laminated plates based on a refined pp. 1905–1918, 2011. DOI: 10.1121/1.3546096.
plate theory, Mech. Adv. Mater. Struct., vol. 29, no. 25, pp. 4138– [64] W. Larbi, J.F. De€ u, R. Ohayon, and R. Sampaio, Coupled
4150, 2022. DOI: 10.1080/15376494.2021.1919805. FEM/BEM for control of noise radiation and sound transmis­
[50] E. Carrera, S. Valvano, and M. Filippi, Classical, higher-order, sion using piezoelectric shunt damping, Appl. Acoust., vol. 86,
zig-zag and variable kinematic shell elements for the analysis of pp. 146–153, 2014. DOI: 10.1016/j.apacoust.2014.02.003.
composite multilayered structures, Eur. J. Mech. A Solids., vol. 72, [65] W. Larbi, J.F. De€ u, and R. Ohayon, Vibroacoustic analysis of
pp. 97–110, 2018. DOI: 10.1016/j.euromechsol.2018.04.015. double-wall sandwich panels with viscoelastic core, Comput.
[51] B. Liu, L. Zhao, A.J.M. Ferreira, Y.F. Xing, A.M.A. Neves, and Struct., vol. 174, pp. 92–103, 2016. DOI: 10.1016/j.compstruc.
J. Wang, Analysis of viscoelastic sandwich laminates using a 2015.09.012.
unified formulation and a differential quadrature hierarchical [66] M.C. Junger, and D. Feit, Sound, Structures, and Their
finite element method, Compos. Part B: Eng., vol. 110, pp. 185–
Interaction, vol. 225. Cambridge, MA: MIT Press, 1986.
192, 2017. DOI: 10.1016/j.compositesb.2016.11.028.
[67] S.K. Baydoun, and S. Marburg, Investigation of radiation damp­
[52] Y. Yan, E. Carrera, and A. Pagani, Free vibration analysis of
ing in sandwich structures using finite and boundary element
curved metallic and composite beam structures using a novel
methods and a nonlinear eigensolver, J. Acoust. Soc. Am., vol. 147,
variable-kinematic DQ method, Mech. Adv. Mater. Struct., vol. 29,
no. 3, pp. 2020–2034, 2020. DOI: 10.1121/10.0000947.
no. 25, pp. 3743–3762, 2022. DOI: 10.1080/15376494.2021.1909784.
[68] K.-J. Bathe, Finite Element Procedures, 2nd ed. Klaus-Jurgen
[53] M. d’Ottavio, D. Ballhause, T. Wallmersperger, and B. Kr€ oplin,
Bathe, Prentice Hall, New Jersey, 2006.
Considerations on higher-order finite elements for multilayered
[69] J.N. Reddy, Mechanics of Laminated Composite Plates and
plates based on a unified formulation, Comput. Struct., vol. 84,
Shells: Theory and Analysis. CRC Press, Florida, 2003.
no. 19-20, pp. 1222–1235, 2006. DOI: 10.1016/j.compstruc.2006.
[70] P. Vidal, L. Gallimard, and O. Polit, Free vibration analysis of
01.025.
composite plates based on a variable separation method,
[54] A. Tarkashvand, K. Daneshjou, A. Golmohammadi, and Z.
Daneshjoo, Viscoelastic-acoustic response of new polymeric Compos. Struct., vol. 230, pp. 111493, 2019. DOI: 10.1016/j.
matrix composite shells considering cavity resonances using a new compstruct.2019.111493.
approach: layered viscoelastic model, Compos. Struct., vol. 292, [71] W. Zhen, and C. Wanji, Free vibration of laminated composite
pp. 115673, 2022. DOI: 10.1016/j.compstruct.2022.115673. and sandwich plates using global–local higher-order theory, J.
[55] C. Nerse, S. Wang, and S. Goo, Effect of damping distribution Sound Vib., vol. 298, no. 1-2, pp. 333–349, 2006. DOI: 10.1016/
on coupling in panel–cavity systems: conditions for optimality j.jsv.2006.05.022.
through a modal approach, Int. J. Mech. Sci., vol. 187, pp. 105908, [72] T. Kant, and K. Swaminathan, Analytical solutions for free
2020. DOI: 10.1016/j.ijmecsci.2020.105908. vibration of laminated composite and sandwich plates based on
[56] N. Sharma, T.R. Mahapatra, S.K. Panda, and C.K. Hirwani, a higher-order refined theory, Compos. Struct., vol. 53, no. 1,
Acoustic radiation and frequency response of higher-order pp. 73–85, 2001. DOI: 10.1016/S0263-8223(00)00180-X.
shear deformable multilayered composite doubly curved shell [73] M. Rao, K. Scherbatiuk, Y. Desai, and A. Shah, Natural vibra­
panel – An experimental validation, Appl. Acoust., vol. 133, tions of laminated and sandwich plates, J. Eng. Mech., vol. 130,
pp. 38–51, 2018. DOI: 10.1016/j.apacoust.2017.12.013. no. 11, pp. 1268–1278, 2004. DOI: 10.1061/(ASCE)0733-9399
[57] C. Shen, F.X. Xin, L. Cheng, and T.J. Lu, Sound radiation of (2004)130:11(1268).
orthogonally stiffened laminated composite plates under air­ [74] Y. Yang, M.J. Kingan, and B.R. Mace, A wave and finite elem­
borne and structure borne excitations, Compos. Sci. Technol., ent method for calculating sound transmission through rect­
vol. 84, pp. 51–57, 2013. DOI: 10.1016/j.compscitech.2013.05.006. angular panels, Mech. Syst. Sig. Process., vol. 151, pp. 107357,
[58] C. Li, Z. Chen, T. Fu, and J. Li, Effect of truss core on sound 2021. DOI: 10.1016/j.ymssp.2020.107357.
radiation behavior of sandwich plate structures under structure [75] L. Dozio, R. Vescovini, and M. D’Ottavio, Structural and acous­
borne excitation, Mech. Adv. Mater. Struct., pp. 1–9, 2023. tic response of finite sandwich plates by using sublaminate
DOI: 10.1080/15376494.2023.2202670. models with variable kinematics, In Proceedings of the 12th
[59] C.W. Isaac, M. Pawelczyk, and S. Wrona, Comparative study of International Symposium on Vibrations of Continuous Systems
sound transmission losses of sandwich composite double panel (ISVCS12), 2019. pp. 1–3.
walls, Appl. Sci., vol. 10, no. 4, pp. 1543, 2020. DOI: 10.3390/ [76] F.J. Fahy, and P. Gardonio, Sound and Structural Vibration:
app10041543. Radiation, Transmission and Response. Elsevier, Oxford, 2007.