Curved and Layer. Struct.
2017; 4:8–20
Research Article
Yu Wang, Guangyu Shi*, and Xiaodan Wang
Displacement and stress analysis of laminated
composite plates using an eight-node
quasi-conforming solid-shell element
DOI 10.1515/cls-2017-0002
Received Jun 28, 2016; accepted Sep 15, 2016
Abstract: This paper presents the efficient modeling and
analysis of laminated composite plates using an eight-
node quasi-conforming solid-shell element, named as
QCSS8. The present element QCSS8 is not only locking-
free, but highly computational efficiency as it possesses
the explicit element stiffness matrix. All the six compo-
nents of stresses can be evaluated directly by QCSS8 in
terms of the 3-D constitutive equations and the appropri-
ately assumed element strain field. Several typical numer-
ical examples of laminated plates are solved to validate
QCSS8, and the resulting values are compared with ana-
lytical solutions and the numerical results of solid/solid-
shell elements of commercial codes computed by the
present authors in which fine meshes were used. The nu-
merical results show that QCSS8 can give accurate dis-
placements and stresses of laminated composite plates
even with coarse meshes. Furthermore, QCSS8 yields also
accurate transverse normal strain which is very important
for the evaluation of interlaminar stresses in laminated
plates. Since each lamina of laminated composite plates
can be modeled naturally by one or a few layers of solid-
shell elements and a large aspect ratio of element edge to
thickness is allowed in solid-shell elements, the present
solid-shell element QCSS8 is extremely appropriate for the
modeling of laminated composite plates.
Keywords: Laminated composite plates; interlaminar
stress; transverse normal stress; solid-shell element;
quasi-conforming element technique; computational effi-
ciency
Yu Wang, Xiaodan Wang: Department of Mechanics, Tianjin Uni-
versity, 135 Yaguan Road, Tianjin 300354, China
*Corresponding Author: Guangyu Shi: Department of Mechanics,
Tianjin University, 135 Yaguan Road, Tianjin 300354, China; Email:
shi_guangyu@163.com
© 2017 Y. Wang et al., published by De Gruyter Open.
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivs 3.0 License.
Open Access
1 Introduction
Laminated composite materials are used in a wide range of
industries like aerospace, marine, transportation etc. due
to their main advantage in mechanical properties such as
high specific stiffness and high specific strength [1]. How-
ever, laminated composite structures are apt to produce
the delamination [2–6] at the interfaces between plies be-
cause of the interlaminar stresses caused by the dissimilar
material properties between the lamina [7], which is one of
the most significant failure modes of composite materials.
Thus, it is very important to accurately predict stresses of
the laminated composite structure.
Finite element models of laminated composite plates
for design and analysis have been commonly used duo
to the fast increase in computing capabilities. It is well
known that the conventional 3-D solid elements are ca-
pable of yielding accurate analysis of laminated compos-
ites [8]. Nevertheless, the discretization through the thick-
ness direction should be similar with the mesh density in
the in-plane direction due to the limitation of the element
aspect ratio, resulting in much higher computational cost
since the overall model composed of great many solid ele-
ments comprises million degrees of freedom. In contrast,
the shell elements are computationally efficient [9–11], but
most of shell elements are not able to accurately capture
the interlaminar transverse shear stress and the transverse
normal stress through the plate thickness direction, be-
cause these shell elements employ the common kinematic
assumptions of inextensibility in the thickness direction
or the zero transverse normal-stress condition etc.
As an improved approach for the conventional shell el-
ement, the stacked-shell model [12] based on FSDT is used
to simulate the laminated composite plates by a discrete
number of sublaminates, and these sublaminates are tied
together with solid cohesive elements or contact interface
elements. Although stacked-shell element can accurately
present the out-of-plane behavior of laminated plates, it
is not advisable in the computational efficiency since the
connection between shell elements or “contact elements”
Unauthenticated
Download Date | 8/27/18 2:17 PM
 Displacement and stress analysis of laminated composite plates | 9
introduced always need specific equations or restrictions
to establish their relationship.
The 3-D continuum element [13, 14] is formulated
using the 3-D field equations of elasticity, which has
been resorted to improve the computational efficiency
and to resolve the inaccurate evaluation of the interlam-
inar stresses occurred in the shell elements. The solid-
shell element was introduced based on the experience
of the derivation of the 3-D continuum element. In 1997,
Domissy [15] and Sze et al. [16] analyzed the plate/shell
problems using the the solid-shell element. Soon after,
Hauptmann et al. [17] also proposed the solid-shell formu-
lation for linear and non-linear analysis. The solid-shell
element has the same geometrical description as the con-
ventional solid element but can be discretized in space like
the degenerated shell element. Besides, the solid-shell el-
ement can also be used in the complicated 3-D plates/shell
structures combined with the solid element directly with-
out the transitional treatment. Thus, the solid-shell ele-
ment has been employed in the modeling of laminated
structures [18–22] due to its advantages.
Nevertheless, the solid-shell formulation based on the
assumed displacement field cannot calculate the interlam-
inar stresses accurately, because displacements are the
primary variables, and stresses are calculated in terms
of the constitutive law with respect to the strains which
are derived from the displacements at the Gauss points,
thereby introducing numerical errors [23]. Additionally,
the interlaminar stresses at the interface rather than at
the Gauss point need to be determined. Since 1964, the
hybrid [24] and the mixed element formulation [25, 26]
have been used for the solid-shell element. For instance,
Rah et al. [21] proposed the partial hybrid stress solid-shell
element based on the modified Hu–Washizu variational
principle, where displacements, strains and stresses can
be approximated independently from each other, there-
fore providing the highest possible flexibility in selecting
the trial functions for primary variables. And eight partial
stress modes are used to capture interlaminar stresses ac-
curately in the laminated composite plate. However, the
static condensation at the element level has been used
to achieve the pure displacement-based equations in the
global coordinate system, leading to complicated calcula-
tions. Whereafter, Rah et al. [22] presented a mixed sold-
shell element, both EAS and ANS methods are used to-
gether in the formulation to alleviate the locking problem.
This mixed solid-shell element could provide very accu-
rate interlaminar stresses through the element thickness
by imposing the interlaminar stresses as constraints in the
enhanced assumed strains formulation. While the mixed
formulation also bring a highly computational cost. Apart
from the FEM used in the modelling of laminated compos-
ite plates, the formulations based the generalized differ-
ential quadrature (GDQ) method [27] have been success-
fully employed in static and dynamic analysis of laminated
composite materials [28–31].
The quasi-conforming element technique (QCET) pro-
posed by Tang et al. [32] is employed to explore accurate
and efficient solid-shell element. It has been demonstrated
by many investigations that QCET is a powerful method
to formulate efficient and reliable membrane, beam and
plate/shell elements [33–37]. Then QCET is extended to
the derivation of the 3D solid-shell element in this paper.
According to the deformation feature, the 3-D strains in
a plate-like structure can be decomposed into the com-
ponents of in-plane strains and the components of out-
of-plane strains, which are interpolated independently to
prevent locking phenomena. Moreover, the solid-shell ele-
ment using QCET possesses explicit element stiffness ma-
trix, which can constitute a simple formulation and finally
devote to the balance between the accuracy and the com-
putational efficiency. Additionally, the real thickness and
the material properties of the layers can be used straight-
forwardly in the solid-shell element, supporting to the bet-
ter prediction of the interfacial damages [38, 39] for the
laminated composite plates.
Based on the quasi-conforming element technique,
accurate, reliable and efficient eight-node solid-shell ele-
ment named as QCSS8 is presented in this paper to model
the laminated composite plates. The detailed formulation
of the solid-shell element QCSS8 is given in the second
section. Then the accuracy and reliability of the resulting
element used in the analysis of the laminated composite
plates are examined by several typical numerical exam-
ples, which are shown in the third section. Conclusions are
summarized in the last section.
2 A brief review of the formulation
of the quasi-conforming
solid-shell element
The quasi-conforming element technique (QCET) is a gen-
eral assumed strain method. The locking problems could
be alleviated by using the assumed strain method when
the element strain field is given properly and reasonably.
Besides, the element strain field is interpolated indepen-
dently from the element displacement field, meanwhile
the relationship between displacements and strains of the
Unauthenticated
Download Date | 8/27/18 2:17 PM
10 | Y. Wang et al.

element is contained within the weak form of the compat-

ibility equation.

2.1 Theoretical basis for the

quasi-conforming element technique

Firstly, we consider a typical continuum in the boundary

value problem, V and S̄ denote, respectively, the domain
and boundary of the continuum; x i (i = 1, 2, 3) denotes the Figure 1: The geometry and coordinates of the solid-shell element
rectangular Cartesian coordinates; S̄ σ and S̄ u represent the QCSS8.
force boundary and the displacement boundary on S̄; t̄ i are
the given tractions on S̄ σ ; n j (j = 1, 2, 3) are the direction
which is the case in the element formulation of the quasi-
cosines of a point on S̄ σ , ∂V and ∂S signify, respectively, a
conforming elements, then Eq. (2) reduces to the following
subdomain and its surface in the continuum.
form as
The weak forms of the generalized equilibrium and the
Num
∑︁e ∫︁ ∫︁ ∫︁ [︂(︂ ∂u i ∂u j )︂
compatibility equation for a continuum discretized by the
]︂
+ − 2ε̃ ij δσ ij dv = 0 (3)
finite element method take the form as [40, 41] ∂x j ∂x i
e=1
∂V e
⎧ ⎫
⎨Num
⎪
∑︁e ∫︁ ∫︁ ∫︁ [︀ (︀ )︀ ]︀
∫︁ ∫︁ ⎪
⎬ If the assumed strains are approximated indepen-
δ U e ε̃ ij − f i u i dv − tu i ds = 0 (1) dently over each element, the weak form of the strain-
⎪
⎩ e=1 ⎪
⎭
∂V e Sσ displacement relation given in Eq. (2) can be enforced in
(i, j = 1, 2, 3) each element individually as
∫︁ ∫︁ ∫︁ [︂(︂ )︂ ]︂
∂u i ∂u j
+ − 2ε̃ ij δσ ij dv = 0 (4)
Num
∑︁e ∫︁ ∫︁ ∫︁ [︂(︂ )︂ ]︂ ∂x j ∂x i
∂u i ∂u j ∂V e
+ − 2ε̃ ij δσ ij dv (2)
∂x j ∂x i
e=1
∂v e Eq. (4) provides a theoretical basis for the deriva-
Num
∑︁b ∫︁ ∫︁ tion of the eight-node solid-shell element QCSS8 based on
u+i − u−i n i + u+j − u−j n j δσ ij ds
[︀(︀ )︀ (︀ )︀ ]︀
+ QCET [32].
eb=1 ∂S
eb
∫︁ ∫︁
[︀ (︀ )︀ ]︀
+ (u i − u i ) n i + u j − u j n j δσ ij ds = 0
2.2 The geometric description and
∂S u
coordinates
where Num e shown in Eq. (1) is the number of total ele-
ments in the continuum under consideration; ε̃ ij are the in- The eight-node solid-shell element QCSS8 considered in
dependent trial strains rather than the strains derived from this study is an arbitrary hexahedron as shown in figure 1,
the assumed displacement field; and U(ε̃ ij ) denotes the where orthogonal coordinate axes X, Y, Z are the global
strain energy density in terms of strains. In Eq. (2), Num b is coordinates and x, y, z constitute the element local coordi-
the number of common surfaces/boundaries between ele- nates, the x-y plane is the reference plane.
ments; δσ ij are the weighting functions; the superscripts The origin of the element local coordinates is set at the
“+” and “-” signify the displacements associated with dif- geometric center of element as shown in figure 1. Vectors
ferent elements along their common element boundary. ex , ey , ez are directions of coordinate axes (x-axis, y-axis
The physical meaning of Eq. (1) is similar to the mini- and z-axis) in the element local coordinates. The direction
−
⇀
mum potential principle. The physical meaning of Eq. (2) of the vector ex is parallel to the line 12, ez is along the
is that the strain-displacement relationship and the con- opposite direction of the normal on area (1234), ey is given
forming condition along the element boundaries are satis- in terms of the right-handed system, and vectors ex , ey , ez
fied in a weighted integral form instead of the point-wise take the forms as
−
⇀ −
⇀ −
⇀
form. When the displacement conforming condition be- 12 13 × 24
e x = ⃒− ⃒;e = ⃒ ⇀ − ⇀⃒ ; ey = ez × ex (5)
⃒ ⇀⃒ z ⃒−
⃒
tween the elements is set to be satisfied a priori and the ⃒12⃒ ⃒13 × 24⃒
displacement boundary conditions are enforced posteriori

Unauthenticated
Download Date | 8/27/18 2:17 PM
 Displacement and stress analysis of laminated composite plates | 11

then the unit base vector of the element local coordinates can be written as
[︁ ]︁T
e = ex ey ez (6)

Let E be the unit base vector of the global coordinates, then the relationship between the global coordinates and
the element local coordinates is established by the transformation matrix T as

e=T·E (7)
[︁ ]︁T
The transformation of the displacement components u i = u i v i w i at node i (i=1, 2,. . . , 8) from the element
local coordinate system to the global coordinate system are performed by
⎧ ⎫ ⎧ ⎫
⎨u⎪
⎪ ⎬ ⎨U⎪
⎪ ⎬
T
v =T V (8)
⎩w⎪
⎪ ⎭ ⎪
⎩W ⎪ ⎭
i i
[︁ ]︁T
where U i = U i Vi Wi are the displacement components of element node i (i=1, 2,. . . , 8) in the global coordinate
system.

2.3 The element assumed strain field

According to the deformation feature of the solid-shell element and elastic theory, the linear strain ε takes the following
form.
{︁ }︁T {︁ }︁
ε = ε x ε y 𝛾xy ε z 𝛾yz 𝛾zx = ∂u∂x
∂v
∂y
∂u ∂v
∂y + ∂x
∂w
∂z
∂w ∂v
∂y + ∂z
∂w ∂u
∂x + ∂z (9)

It is advantageous for element QCSS8 to be a locking-free element by interpolating the element strain field indepen-
dently from the element displacement field [32]. Then the element assumed strain ε̃ is approximated as

ε̃=Nα (10)

where N is the interpolation function matrix for the element assumed strain, α is the vector composed of the correspond-
ing undetermined generalized strain parameters. To prevent the spurious zero-strain energy mode, the number of the
undetermined generalized strain parameters should be not less than the difference between the total number of nodal
DOFs and the total number of the rigid body displacement.
The determination of the element assumed strain field for a solid-shell element should follow the basic principles
of finite element method, such as the constant strain, the coordinate invariance, the requirement of rank sufficiency
etc. The empirical approach based on deformation characteristic of thin plate is also helpful. Besides, the assumed
strain field of element QCSS8 need be interpolated as simply as possible to ensure the highly computational efficiency,
the factors to be free from locking should be taken into account. The element assumed strain of element QCSS8 is also
inspired by the idea in the development of the element assumed strain fields for the four-node C0 plate element and
the four-node quadrilateral membrane element proposed by Shi et al. [35, 41]. Thus, the assumed strain field of element
QCSS8 takes the form as

1 y z yz
⎡ ⎤ ⎡ ⎤
ε̃ x ⎧ ⎫
⎢ ε̃ y ⎥ ⎢
⎢ ⎥ ⎢ 1 x z xz 0 ⎥⎪
⎥⎪⎪ α1 ⎪⎪
⎪
⎥ α2
2
⎢𝛾˜ ⎥ ⎢ ⎨ ⎪
⎪
⎢ xy ⎥ ⎢ 1 z z ⎥ ⎬
ε̃ = ⎢ ⎥ = ⎢ . (11)
⎢ ε̃ z ⎥ ⎢ 1 x y .. ⎪
⎥
⎥⎪
⎢ ⎥ ⎢ ⎥⎪⎪ ⎪
⎪
⎣𝛾˜yz ⎦ ⎣ 0 1 z
⎪ ⎪
α18
⎦⎩ ⎭
𝛾˜zx 1 z

The element assumed strain field for element QCSS8 defined in the Eq. (11) can be completely reduced to the as-
sumed strain filed of a four-node quadrilateral membrane element [41].

Unauthenticated
Download Date | 8/27/18 2:17 PM
12 | Y. Wang et al.

2.4 The element stiffness matrix direction cosines on surface A j . And u j , v j , w j are the dis-
placement components on the surface A j (the subscript
If let the transpose of the interpolation matrix N defined j=1, 2,. . . , 6).
in Eq. (11) be a weighing function for δσ in Eq. (4), then The volume integration shown in Eq. (18) is reduced
Eqs. (9-10) yield the weak form of the compatibility equa- to the integration around the bounding surfaces of the ele-
tion for element QCSS8 as ment, and each surface of the solid-shell element contains
∫︁ ∫︁ ∫︁ four nodes which is similar to a four-noded quadrilateral
NT (ε − ε̃) dxdydz = 0 (12) element. Then the bilinear interpolation form can be used
Ωe to describe the displacement fields of the bounding sur-
faces. The evaluation of matrices C and A are the key work
By making use of Eq. (10), Eq. (12) can also be rewritten as
for deriving the element stiffness matrix of QCSS8.
Matrix A can be explicitly solved by substituting the
∫︁ ∫︁ ∫︁ ∫︁ ∫︁ ∫︁
T
N εdxdydz = NT Nαdxdydz (13)
interpolation matrix N given in Eq. (11) into Eq. (16). There-
Ωe Ωe
fore, vector α (the generalized strain parameter) is ex-
Eq. (13) can be written in a matrix for as pressed by matrices A, C and vector q in terms of Eqs. (14-
16) as
Cq = Aα (14) α = A−1 Cq (19)

where matrix C and matrix A are expressed in Eq. (15) and Then substituting Eq. (19) into Eq. (10), the element
Eq. (16), q is the nodal displacement vector of element assumed strain field is rewritten as
QCSS8 in the local coordinate system as defined in Eq. (17).
ε̃ = NA−1 Cq = Bq (20)
∫︁ ∫︁ ∫︁
Cq = NT εdxdydz (15) The corresponding strain energy for an element can be
Ωe expressed as
∫︁ ∫︁ ∫︁ ∫︁ ∫︁ ∫︁
A= NT Ndxdydz (16) 1 T *
U e (ε̃) = ε̃ Q ε̃dxdydz (21)
2
Ωe Ωe
⎛ ⎞
∫︁ ∫︁ ∫︁
1 *
= qT ⎝ BT Q Bdxdydz⎠ q
{︁ }︁
qT = u 1 v1 w1 u2 v2 w2
w8 ··· u8 v8 2
Ωe
(17)
Using integration by parts, the volume integration *
where Ke is the element stiffness matrix and Q is the off-
given in Eq. (15) and Eq. (16) could be degenerated into axis stiffness matrix of composite lamina attached in the
the integration around the bounding surfaces of the ele- following subsection in detailed.
ment. Matrix C is achieved correspondingly by substitut-
ing Eq. (9) and Eq. (11) into Eq. (15) and then separating
nodal displacement components. Taking ε x for example, 2.5 Eflcient analysis of the laminated
the left-hand side of Eq. (15) for an eight-node solid-shell
composites by element QCSS8
element is written in the following form.
∫︁ ∫︁ ∫︁ 6 ∫︁∫︁ The laminated composite plate under consideration is
T ∂u ∑︁
(N1 ) dxdydz = ○ (N1 )T u j n jx dA j (18) composed of a finite number of linear elastic orthotropic
∂x
j=1 A
Ωe j laminae. The constitutive equation [7] in the principal ma-
∫︁ ∫︁ ∫︁
∂ ( N1 ) T terial coordinates of the lamina take the form as
− u dxdydz
∂x ⎧ ⎫ ⎡
σ1 ⎪ Q11 Q12 Q13 0 0 0 ⎪
⎤⎧ ⎫
ε1 ⎪
Ωe ⎪
⎪ ⎪ ⎪ ⎪
⎪
σ Q Q Q 0 0 0 ⎪ ε2 ⎪
⎪ ⎪ ⎪ ⎢ ⎪
2 12 22 23
⎪ ⎥⎪ ⎪
6 ∫︁∫︁
∑︁
⎪
⎪ ⎪
⎪
⎪
⎪ ⎢ ⎥⎪⎪ ⎪
⎪
○ (N1 )T u j n jx dA j 0 0 0 ⎥ ε3 ⎬
⎨ σ ⎬ ⎢Q
= ⎢ 13 Q23 Q33
⎥ ⎨
3
=⎢
τ23 ⎪ ⎢ 0 0 0 Q44 0 0 ⎥⎪
⎥
j=1 A
j
⎪
⎪ ⎪ 𝛾23 ⎪
⎪ ⎪ ⎢ ⎥⎪⎪ ⎪
⎪
τ 0 0 0 0 Q 0 𝛾
⎪ ⎪ ⎪ ⎪
31 55 31
⎪
⎪ ⎪
⎪ ⎣ ⎦ ⎪
⎪ ⎪
⎪ ⎭ ⎪ ⎪ ⎪ ⎪
in which N1 is the non-zero term in the corresponding ⎩
τ12 0 0 0 0 0 Q66
⎩
𝛾12
⎭
row vector
[︁ of N shown
]︁ in Eq. (11) and takes the form as (22)
N1 = 1 y z yz . Parameter n jx , n jy and n jz are the

Unauthenticated
Download Date | 8/27/18 2:17 PM
 Displacement and stress analysis of laminated composite plates | 13

The stiffness components can be written in terms of the engineering constants as

1 − ν23 ν32 ν + ν13 ν32 ν + ν12 ν23 1 − ν13 ν31 ν + ν21 ν13
Q11 = ; Q12 = 12 ; Q13 = 13 ; Q22 = ; Q23 = 23 ; (23)
E2 E3 S E2 E3 S E2 E3 S E1 E3 S E1 E3 S
1 − ν12 ν21 1 − ν12 ν21 − ν23 ν32 − ν13 ν31 − 2ν12 ν23 ν31
Q33 = ; Q44 = G23 ; Q55 = G31 ; Q66 = G12 ; S =
E1 E2 S E1 E2 E3
in which, E i , ν ij and G ij (i, j = 1, 2, 3) are elastic modulus, Poisson ratio and shear modulus of the lamina in the principal
material coordinates respectively. When there is an orientation angle θ between the principal material axis of a lamina
and the reference coordinate x-axis (assumed that z-direction is always consistent), the stresses of the lamina in the
plate global coordinate and in the principal material coordinates have the following relationship.
⎧ 2
2
⎪σ x = σ1 cos θ + σ2 sin θ − τ12 2 sin θ cos θ
⎪
⎪
σ y = σ1 sin2 θ + σ2 cos2 θ + τ12 2 sin θ cos θ
⎪
⎪
⎪
⎪
⎪
⎪
⎨σ z = σ3
⎪
(24)
⎪τ zx = τ31 cos θ − τ32 sin θ
⎪
⎪
⎪
⎪τ zy = τ31 sin θ + τ32 cos θ
⎪
⎪
⎪
⎪ (︁ )︁
cos2 θ − sin2 θ
⎪
⎩τ =σ sin θ cos θ − σ sin θ cos θ + τ
xy 1 2 12

The off-axis stiffness matrix denoted as Q is defined to establish the stress-strain relationship in the reference coor-
dinate, the detailed expression is given as
⎧ ⎫ ⎡ ⎤⎧ ⎫
⎪
⎪ σx ⎪⎪ Q11 Q12 Q13 0 0 Q16 ⎪ ⎪ εx ⎪
⎪ ⎪
σ Q Q Q 0 0 Q ⎪ εy ⎪
⎪ ⎪ ⎪ ⎢ ⎪
y
⎪
⎪ ⎪ 12 22 23 26
⎥⎪⎪ ⎪
⎪ ⎪
⎪ ⎪ ⎢ ⎥⎪ ⎪
⎪
⎨ σ ⎬ ⎢Q
z ⎢ 13 Q23 Q33 0 0 Q36 ⎥ ε z ⎬
⎥ ⎨
=⎢ (25)
τ yz ⎪ ⎢ 0 0 0 Q44 Q45 0 ⎥⎪
⎥
⎪
⎪ ⎪ ⎪ 𝛾yz ⎪
⎪
⎪ ⎪ ⎢ ⎥ ⎪ ⎪
⎪ ⎣ 0 0 0 0 ⎦⎪
⎪ τ zx ⎪ Q45 Q55 ⎪𝛾zx ⎪
⎪
⎪ ⎪ ⎪ ⎪ ⎪
⎪ ⎪ ⎪ ⎪
0 0
⎩ ⎭ ⎩ ⎭
τ xy Q16 Q26 Q36 Q66 𝛾xy

The components Q ij (i, j = 1, 2, 3) for the off-axis stiffness matrix Q are given by the previous stiffness components Q ij
and angle θ as
⎧
⎪
⎪
⎪ Q11 = cos4 θ Q11 + 2sin2 θcos2 θ (Q12 + 2Q66 ) + sin4 θ Q22
Q12 = sin2 θcos2 θ(Q11 + Q22 − 4Q66 ) + (sin4 θ + cos4 θ)Q12
⎪
⎪
⎪
⎪
⎪
⎪
⎪Q13 = cos2 θQ13 + sin2 θQ23
⎪
⎪
⎪
⎪ (︁ (︁ )︁ )︁
2 2 2 2
⎪
Q = sin θ cos θ cos θQ − sin θQ + sin θ − cos θ (2Q + Q )
⎪
16 11 22 66 12
⎪
⎪
⎪
⎪
Q22 = sin4 θ Q11 + 2sin2 θcos2 θ (Q12 + 2Q66 ) + cos4 θ Q22
⎪
⎪
⎪
⎪
⎪
⎪
⎪Q23 = sin2 θ Q13 + cos2 θ Q23
⎪
⎪
⎪
⎨ (︁ (︁ )︁ )︁
Q26 = sin θ cos θ sin2 θ Q11 − cos2 θ Q22 + cos2 θ − sin2 θ (2Q66 + Q12 ) (26)
⎪
⎪
Q = Q33
⎪
⎪ 33
⎪
⎪
⎪
⎪Q36 = sin θ cos θ (Q13 − Q23 )
⎪
⎪
⎪
⎪
Q44 =cos2 θ Q44 + sin2 θ Q55
⎪
⎪
⎪
⎪
⎪
⎪
⎪Q45 = sin θ cos θ (Q44 + Q55 )
⎪
⎪
⎪
⎪
2 2
⎪
⎪Q55 = sin θ Q44 + cos θ Q55
⎪
⎪
⎪ (︁ )︁
⎪
⎩Q = sin2 θcos2 θ (Q + Q − 2Q ) + sin4 θ + cos4 θ − 2sin2 θcos2 θ Q
⎪
66 11 22 12 66

*
The matrix Q needs to make a little change to match with Eq. (9), then the new off-axis stiffness matrix Q and the
corresponding stiffness components are of the forms as
*
σ=Q ε (27)

Unauthenticated
Download Date | 8/27/18 2:17 PM
14 | Y. Wang et al.

3.1 A four-ply [0/90/90/0] laminated plate

under the uniformly distributed load
⎧ ⎫ ⎡ ⎤⎧ ⎫
⎪
⎪ σx ⎪ Q11 Q12 Q16 Q13 0 0 ⎪ εx ⎪
⎪ ⎪ ⎪
⎪ ⎪
σ ⎪ ⎢Q12 Q22 Q26 Q23 0 0 ⎥ εy ⎪
⎪ ⎢ ⎪
y
⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪ ⎥⎪⎪ ⎪
⎪
⎨τ ⎬ ⎢Q
xy Q26 Q66 Q36 0 0 ⎥ 𝛾xy ⎬
⎥ ⎨ The laminated plate with the simply-supported condition
= ⎢ 16
⎢
σz ⎪ ⎢Q13 Q23 Q36 Q33 0 0 ⎥⎪ εz ⎪ on the plate edges is investigated in this example. The
⎥
⎪
⎪ ⎪
⎪ ⎪ ⎢ ⎥⎪⎪ ⎪
⎪
⎪ τ yz ⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎣ 0 0 0 0 Q44 Q45 ⎪
⎦ ⎪
⎪ 𝛾
⎪ ⎪ yz
⎪
⎪
⎪ length-to-thickness aspect ratios of the laminated plate
0 0 0 0 are a/h = 10 and a/h = 100. Material properties are given
⎩ ⎭ ⎩ ⎭
τ zx Q45 Q55 𝛾zx
(28) as
According to Eqs. (20-21, 28), the element stiffness E1 = 174.6 GPa, E2 = E3 = 7 GPa, G12 = G13 = 3.5 GPa,
matrix Ke for element QCSS8 with domain Ω e used in
G23 = 1.4 GPa, ν12 = ν13 = ν23 = 0.25
the modeling of laminated composites takes the following
form. A quarter of the laminated plate is modeled due to
∫︁ ∫︁ ∫︁
* the symmetry, and each lamina of the laminated plate is
Ke = BT Q Bdxdydz (29)
Ωe
modeled by one layer or a few layers of QCSS8 elements.
∫︁ ∫︁ ∫︁
*
Both the deflection and the different stress components are
= CT A−1 NT Q NA−1 Cdxdydz computed in this example, in which the transverse shear
Ωe stress at the interface of the laminated plate τ ixz is also con-
Matrix A and C in the Eq. (29) can be taken out- sidered. The numerical results are normalized as
side from the integration because they are the matri- 100E2 (σ x , σ y , τ xy )
w=w× ; (σ x , σ y , τ xy ) = ; (31)
ces composed of constants. N is a very simple polyno- q0 hS3 q0 S2
mial interpolation matrix, the integration about matri- (σ z , τ yz , τ xz ) a
* (σ z , τ yz , τ zx ) = ;S=
ces multiplication NT Q N can also be derived straightfor- q0 S h
wardly in the element local Cartesian coordinates. Thus, The numerical results of the deflection and stresses
the element stiffness matrix Ke can be represented explic- given by element QCSS8 with different mesh densities and
itly as different aspect ratios (a/h) are listed in Table 1. The re-
⎛ ⎞
∫︁ ∫︁ ∫︁
*
sults obtained from elements Solid45 and Solsh190 in AN-
Ke = CT A−1 ⎝ NT Q Ndxdydz⎠ A−1 C (30) SYS are also given here for comparison. The reference so-
Ωe lutions are computed using the high-order solid element
It is worthwhile to point out that the element stiffness Solid186 in ANSYS [45] with a fine mesh 60×60×3 on each
matrix Ke defined in Eq. (30) can be evaluated explicitly as lamina. Deflections and stresses are evaluated at different
it only involves with simple polynomials. plate locations, as
(︂ )︂ (︂ )︂ (︂ )︂
a b a b a b 3h
w , , h , σx , , h , σy , , ,
2 2 2 2 2 2 4
3 Numerical examples and τ yz
(︂
a
, 0,
h
)︂ (︂
, τ cxz a, ,
b h
)︂ (︂
b 3h
, τ ixz a, ,
)︂

discussions
This section focuses the assessment of the performance
of the present solid-shell element QCSS8 used in anal-
yses of laminated composite plates, which includes the
convergence rate, the sensitivity to mesh, the influence of
the length-to-thickness aspect ratio and the influence of
the face-to-core stiffness ratio, etc. The numerical results
given by the present element will be compared with the
reference results reported in the literature, analytical solu-
tions based on Shi’s higher-order shear deformation plate
theory [42] accounting for interlaminar transverse shear
stress continuity given by the third author [43, 44] as well
as the numerical results given by the reliable elements in
ANSYS.
2 2 2 2 2 4
In Table 1, symbol ‘+/-’ is used to denote the location
of the top and bottom surfaces of the interlaminar inter-
face. The numerical results of the deflection and stresses
given by QCSS8 converge to the reference solutions both
in two aspect ratio cases. The present shear stresses are
calculated from the average value of the shared node of
the neighboring elements at the interlaminar interface. Al-
though the shear stress τ ixz given by the present element
are smaller than reference solutions, it can be seen from
the results in Table 1 that the present element QCSS8 yield
better results than elements Solid45 and Solsh190. The ac-
curacy of shear stress τ ixz given by QCSS8 can be improved
when more layers of elements are used through the plate
thickness.

Unauthenticated
Download Date | 8/27/18 2:17 PM
 Displacement and stress analysis of laminated composite plates | 15

Table 1: Normalized deflections and stresses in a [0/90/90/0] laminated plate under the uniformly distributed load.

a/h Elements & mesh w σx σ+y /σ−y τ yz τ cxz τ ixz

10 Solid45 (4×4×4) 1.1561 0.8143 0.0253/0.5771 0.0400 −0.5135 −0.4159
Solsh190 (4×4×4) 1.1568 0.8148 0.0253/0.5775 0.3037 −0.5142 −0.4172
QCSS8 (4×4×4) 1.1616 0.8251 0.0246/ 0.5751 0.3345 −0.5184 −0.4203
Solid45 (8×8×8) 1.1702 0.8355 0.0252/0.5758 0.4404 −0.5465 −0.5155
Solsh190 (8×8×8) 1.1704 0.8356 0.0252/0.5759 0.4407 −0.5467 −0.5157
QCSS8 (8×8×8) 1.1655 0.8363 0.0250/0.5700 0.4835 −0.5708 −0.5342
Solid186 (60×60×12) 1.1758 0.8403 0.0251/0.5767 0.5037 −0.5793 −0.5702
100 Solid45 (4×4×4) 0.6606 0.7788 0.0143/0.3589 0.1189 −0.5076 −0.3465
Solsh190 (4×4×4) 0.6930 0.8186 0.0148/0.3708 0.2512 −0.5951 −0.4606
QCSS8 (4×4×4) 0.6906 0.8237 0.0147/0.3705 0.2731 −0.6056 −0.4691
Solid45 (8×8×8) 0.6949 0.8242 0.0146/0.3646 0.3641 −0.6126 −0.5421
Solsh190 (8×8×8) 0.6971 0.8268 0.0146/0.3652 0.3898 −0.6363 −0.5658
QCSS8 (8×8×8) 0.6931 0.8287 0.0146/0.3627 0.4331 −0.6508 −0.6182
Solid186 (60×60×12) 0.6994 0.8307 0.0146/0.3635 0.4575 −0.6648 −0.6559

(a) (b)

Figure 2: The distribution of the normalized stresses σ x and σ y through the plate thickness for a simply-supported laminated plate
[0/90/90/0] subjected to the uniformly distributed load. (a/h = 10).

Since the stress analysis plays a crucial role in real en-
gineering problems, it is worthwhile to further examine
the accuracy of the stresses given by QCSS8. The trans-
verse normal stress is also considered in this example. The
distributions of two normal stresses and two transverse
stresses through the plate thickness given by different ele-
ments are plotted in figures 2-3.
As shown in figure 2, the numerical results of the in-
plane normal stresses through the plate thickness given
by different elements with two coarse meshes are nearly
identical. Figure 3(a) illustrates that the present element
QCSS8 is capable of accurately predicting the transverse
normal stress even with a coarse mesh. Figure 3(b) shows
that transverse shear stress obtained from QCSS8 is bet-
ter than the transverse shear stresses calculated by ele-
ments Solid45 and Solsh190. Figure 3(b) also shows that
the present element together with Solid45 and Solsh190
does not satisfy the zero traction conditions on the plate
surfaces. It stands to reason that most of lower-order solid
elements cannot exactly satisfy the zero traction condi-
tion on the top and bottom surfaces of the laminated plate.
Because the linear distribution of the interpolated trans-
verse shear strain/stress through the plate thickness in
these elements lead to an inability to accurately character-
ize the highly nonlinear transverse shear stress in various
plates [46]. But the computed transverse shear stresses can
be improved by using the equilibrium equations in terms
of stresses as the manner to evaluate the transverse shear

Unauthenticated
Download Date | 8/27/18 2:17 PM
16 | Y. Wang et al.

laminar stress through the plate thickness, since the trans-

verse shear stress on the plate surfaces has a neglected role
in the safety design of laminated structures.
In conclusion, the stress comparisons depicted in fig-
ures 2-3 show that the present element QCSS8 yields not
only very accurate in-plane normal stresses but good re-
sults for the transverse normal stress even with the coarse
mesh. Besides, QCSS8 can also give the accurate maxi-
mum and interlaminar shear stress through the plate thick-
ness. On the other hand, Solid45 and Solsh190 are able to
give accurate axial normal stresses, but fail to give reliable
transverse shear stresses.

(a) 3.2 A simply-supported sandwich-like plate

under the uniformly distributed load

A squared sandwich-like plate under the uniformly dis-

tributed load shown in figure 4 is investigated in this ex-
ample. The two face panels of the sandwich-like plate are
identical with thickness h f = 0.1h and the thickness of
the core h c = 0.8h. The length-to-thickness aspect ratio is
core
a/h = 10. The stiffness components Q ij of the sandwich
core are expressed in the dimensionless form as [44]
core core
Q11 = 0.999781 Q12 = 0.231192 (32)
core core core
Q22 = 0.524886 Q33 = 0.266810 Q44 = 0.266810
core core
Q55 = 0.159914 Q66 = 0.262931

In order to study the influence of the face-to-core stiff-

(b) ness ratio on the analysis of the sandwich plate, the me-
chanical properties of the face panel are defined in terms
Figure 3: The distribution of the normalized stresses σ z and τ xz
of the core’s mechanical properties and a multiplier R as
through the plate thickness for a simply-supported laminated plate
[0/90/90/0] subjected to the uniformly distributed load. (a/h =
Eq. (33) shown. The values of 5, 10 and 15 for R are consid-
10). ered.
face core
Q ij = RQ ij (33)

Each skin panel of the sandwich is discretized by a

mesh of 8×8×1 and the core is modeled by 8×8×8 mesh. The
numerical solutions obtained from the present element are
compared with exact solutions of Srinivas [47], the analyt-
ical solutions obtained from a refined sandwich plate the-
ory [44] and the numerical solutions calculated by element
Solsh190. All the results for comparison listed in Table 2
are normalized as
(︂ )︂
Figure 4: The geometry and coordinates of a sandwich-like plate. a b h 0.999781
w=w , , , (34)
2 2 2 hq
(︂ )︂
a b 1
stresses used in laminated plate theories [44]. Moreover, σbx = σ x , ,0 ,
2 2 q
the accuracy of the resulting transverse shear stresses are (︂
a b h 1
)︂ (︂
a b h 1
)︂
face core
measured by the predicted maximum stress and the inter- σx = σx , , , σx = σx , , ,
2 2 10 q 2 2 10 q

Unauthenticated
Download Date | 8/27/18 2:17 PM
 Displacement and stress analysis of laminated composite plates | 17

Table 2: Normalized deflections and stresses in the sandwich-like plate under the uniformly distributed load.

R Models/Elements w σ bx σ face
x σ core
x τ cxz τ ixz
5 Solsh190 (8×8×10) 265.733 60.426 48.238 9.641 4.416 3.726
QCSS8 (8×8×10) 261.879 60.469 46.720 9.345 4.255 3.493
Analytical solution [44] 259.232 60.279 46.599 9.320 4.321 3.401
Elasticity Solution [47] 258.970 60.353 46.623 9.340 4.364 3.268
10 Solsh190 (8×8×10) 166.522 67.292 51.914 5.193 4.196 3.889
QCSS8 (8×8×10) 163.614 67.095 50.368 5.037 4.056 3.630
Analytical solution [44] 159.550 65.205 48.840 4.884 4.062 3.656
Elasticity Solution [47] 159.380 65.332 48.857 4.903 4.272 3.508
15 Solsh190 (8×8×10) 128.151 70.171 52.374 3.495 4.067 3.902
QCSS8 (8×8×10) 125.594 69.719 50.858 3.391 3.930 3.651
Analytical solution [44] 121.856 66.621 48.293 3.220 3.936 3.718
Elasticity Solution [47] 121.720 66.787 48.299 3.238 3.694 3.577

(︂ )︂ (︂ )︂
b h b h 1
τ cxz = τ xz 0, , , τ ixz = τ xz 0, ,
2 2 2 10 q
Table 2 shows that the results given by QCSS8 agree
with elasticity solutions [47] and the analytical solutions
evaluated from a refined sandwich plate theory [44]. Ta-
ble 2 also indicates that the accuracy of QCSS8 deteriorates
as the face-to-core stiffness ratio increasing. Nevertheless,
the present solid-shell element QCSS8 possesses a higher Figure 5: Geometry, lamination, boundary and load conditions of
accuracy of deflections and stresses than the solid-shell el- the laminated strip.
ement Solsh190 in ANSYS in all the cases of face-to-core
stiffness ratios. For example, when R = 15, the error of the
ment, while the other lamina is modeled by one layer ele-
maximum shear stress given by QCSS8 with respect to the
ments only. Thus this strip is modeled by ten layers of ele-
elasticity solution is 6.4%, while the error of the maximum
ments though the thickness. Each lamina is discretized by
shear stress given by Solsh190 is 10.1%.
a 10×4 mesh. The deflection (w Cz ) and the bending stress
(σ Cx ) at point C as shown in figure 5, as well as the inter-
laminar shear stress (τ Dxz ) located at the interface of the
3.3 A laminated strip under three-point
strip edge (point D) are calculated by element QCSS8. The
bending resulting values are compared with numerical results ob-
tained from Solid45 and Solsh190 together with the results
The laminated strip subjected to three-point bending [48]
given by the stacked-shell model [12] which is denoted as
is used to verify the correct representation of trans-
SSA-n, herein ‘n’ is the number of sublaminates. Results
verse shear effects by the present solid-shell element
for the deflection, the bending stress and the shear stress
QCSS8. The lamination scheme of the laminated strip is
at the specified points of the strip are presented in Table 3.
[0/90/0/904 /0/90/0]. The dimensions of the strip are: the
Table 3 shows that both the deflection and the two
length L = 30 mm, the width W = 10 mm and the thick-
components of stresses calculated by the present element
ness T = 1 mm as depicted in figure 5.
QCSS8 agree well with reference solutions. Particularly,
The laminated strip is subjected to a line load of
QCSS8 gives more accurate stresses than the 3-D solid ele-
10 N/mm on the top surface of the laminated strip as
ment Solid45 and the solid-shell element Solsh190 in AN-
shown in figure 5. The material properties are
SYS when the same mesh is used. It can also be found from
E1 = 100 GPa, E2 = 5 GPa, G12 = 3 GPa, Table 3 that the stacked-shell model presented by Lam-
G13 = G23 = 2 GPa, ν12 = 0.4, ν13 = ν23 = 0.3 peas et al. used a much higher density of mesh to obtain
accurate results. Besides, sublaminates of the stacked-
Because the core of the laminated strip is composed
shell model are tied by solid cohesive elements or con-
of four laminae, the core is modeled by four layers of ele-

Unauthenticated
Download Date | 8/27/18 2:17 PM
18 | Y. Wang et al.

Table 3: The maximum displacements and the stresses on special point of a laminated strip.

SSA−1 [12] SSA−3 [12] SSA−7 [12] Solid45 Solsh190 QCSS8 Analytical
(100×20×1) (100×20×3) (100×20×7) (10×4×10) (10×4×10) (10×4×10) solution [48]
w Cz (mm) −1.065 −1.070 −1.077 −1.067 −1.048 −1.054 −1.060
σ Cx (MPa) 673.090 679.800 693.520 648.312 650.874 675.193 684.000
τ Dxz (MPa) 4.831 −3.837 −4.114 −4.204 −4.195 − 4.182 −4.100

tact interface elements, which leads to the higher com-

putational cost since the use of either cohesive elements
or contact interface elements consumes much more com-
putational efforts. In contract, one layer or few layers of
QCSS8 elements can be used to model each lamina of a
laminated plate directly without resorting to interface ele-
ments, which makes QCSS8 possess highly computational
efficiency.

3.4 An asymmetric cross-ply [0/90] plate

under the sinusoidally distributed load

The asymmetric cross-ply [0/90] plate with simply-

supported edges is investigated to examine the perfor- (a)
mance of element QCSS8 in this example. The length-to-
thickness aspect ratio of this squared plate is equal to 10.
The material properties of the lamina are the same as those
in the first example. The distributed load given in Eq. (35)
is applied on the top surface of the laminated plate.
(︁ πx )︁ (︁ πy )︁
q = q0 sin sin (35)
a b
The distribution of the normal stress and the trans-
verse shear stress obtained from element QCSS8 will be
compared with the finite element results calculated by
element Solsh190 and the reference solutions given in
Ref. [43].
The numerical solutions of the in-plane normal stress
given by QCSS8 are nearly identical to the reference result
shown in the figure 6(a) both in the fine mesh and in the
(b)
coarse mesh. Meanwhile, figure 6(b) represents the trans-
verse shear stresses through the thickness of the lami- Figure 6: The distribution of stress σ x and stress τ xz through the
nated plate and reflects that the present element possesses thickness of the asymmetric cross-ply [0/90] plate subjected to the
higher accuracy than the same-type element Solsh190 es- sinusoidally distributed load (a/h = 10).

pecially for the coarse mesh case. QCSS8 can also give the
accurate maximum and interlaminar shear stress through
the plate thickness in the analysis of the antisymmetric
laminated plates. However, as pointed out in Section 3.1
that the present QCSS8 still does not satisfy the zero trac-
tion condition on the top and bottom surfaces of the lami-
nated plate like most of lower order elements such as ele-
ment Solsh190. Since these lower order elements interpo-
lated the linear transverse shear strain through the plate
thickness, and the transverse shear stresses are derived
from the corresponding strain directly using the constitu-
tive law, which lead to an inability to accurately character-
ize the highly nonlinear transverse shear stress for these
elements. Therefore, there is some difference between the

Unauthenticated
Download Date | 8/27/18 2:17 PM
 Displacement and stress analysis of laminated composite plates | 19
analytical solution of the transverse shear stress and the
numerical solution of the transverse shear stress given
by the present element. The computed shear stresses can
be improved by using equilibrium equations in terms of
stresses as the manner to evaluate the transverse shear
stresses used in laminated plate theories. The fundamen-
tal improvement for this problem is going to further ex-
plore a more proper interpolation function matrix of the
assumed element strain which can well describe nonlinear
deformation and well solve the doubly-curved structures
etc.
4 Conclusions
A reliable and accurate eight-noded solid-shell element
based on the quasi-conforming element technique de-
noted as QCSS8 is used for the modeling and analysis of
laminated composite plates. Unlike the conventional fi-
nite elements, the solid-shell element QCSS8 presented in
this work is derived from the assumed element strain field
which ensures stresses evaluated straightforwardly and
accurately from the 3-D constitutive equation. The lami-
nated plates can be modeled by QCSS8 efficiently because
each lamina of a laminated plate is modeled by one layer
or few layers of QCSS8 elements and a large aspect ratio of
element edge to thickness is allowed.
Four typical examples of laminated composite plates
are used to demonstrate outstanding performance of
QCSS8. The numerical results of both deflections and
stresses given by QCSS8 agree well with corresponding
reference solutions. The numerical results show that the
present solid-shell element QCSS8 possesses higher accu-
racy than the solid-shell element Solsh190 in ANSYS that
has the same nodal degrees of freedom as QCSS8, particu-
larly in the evaluation of transverse shear stresses and in-
terlaminar stresses with the coarse meshes. The third ex-
ample indicates that the solid-shell element QCSS8 pos-
sesses the highly computational efficiency in the analy-
sis of the laminated composite plates than the stacked-
shell model since QCSS8 need neither solid cohesive ele-
ments nor contact interface elements which is used in the
stacked-shell model.
Acknowledgement: The financial support partially pro-
vided by the grants of NSFC-91230113 is thankfully ac-
knowledged.
References
[1] J. N. Reddy, Mechanics of Laminated Composite Plates and
Shells: Theory and Analysis, CRC Press Boca Raton, 2003
[2] E. D. Reedy, Modeling the initiation and growth of delamina-
tions in composite structures, Journal of Composite Materials,
207(53) (1993), 315-323
[3] Z. Zou, S. R. Reid, S. Li, P. D. Soden, Application of a delamina-
tion model to laminated composite structures, Composite Struc-
tures, 56(4) (2002), 375-389
[4] S. R. Finn, Y. F. He, G. S. Springer, Delaminations in compos-
ite plates under transverse static loads —experimental results,
Journal of Reinforced Plastics and Composites, 11(11) (1992),
1196-1238
[5] Z. Petrossian, M. R. Wisnom, Prediction of delamination initia-
tion and growth from discontinuous plies using interface ele-
ments, Parallel Computational Fluid Dynamics, 29(5-6) (1998),
503-515
[6] G. Wimmer, C. Schuecker, H. E. Pettermann, Numerical simu-
lation of delamination in laminated composite components—
A combination of a strength criterion and fracture mechanics,
Composites Part B: Engineering, 40(2) (2009), 158-165
[7] G. Shen, J. Hu, Mechanics of Composite Materials, 3rd ed, Ts-
inghua University Press, Beijing, 2010, 39-55 (in Chinese)
[8] S. Y. Lee, D. Y. Park, Buckling analysis of laminated composite
plates containing delaminations using the enhanced assumed
strain solid element, International Journal of Solids and Struc-
tures, 44(24) (2007), 8006-8027
[9] S. Brischetto, O. Polit, E. Carrera, Refined shell model for the lin-
ear analysis of isotropic and composite elastic structures, Euro-
pean Journal of Mechanics—A/Solids, 34(34) (2012), 102-119
[10] L. Vu-Quoc, I. K. Ebcioğlu, Multilayer shells: Geometrically-exact
formulation of equations of motion, International Journal of
Solids and Structures, 37(45) (2000), 6705-6737
[11] L. Vu-Quoc, H. Deng, X. G. Tan, Geometrically-exact sandwich
shells: The static case, Computer Methods in Applied Mechan-
ics and Engineering, 189(1) (2000), 167-203
[12] G. Lampeas, K. Fotopoulos, Interlaminar stresses calculation
using a stacked-shell finite element modeling approach, Inter-
national Journal of Applied Mechanics, 7(5) (2015), 1-22
[13] C. Liao, J. N. Reddy, Analysis of anisotropic, stiffened composite
laminates using a continuum-based shell element, Computers
and Structures, 34(6) (1990), 805-815
[14] S. Klinkel, F. Gruttmann, W. Wagner, A continuum based three-
dimensional shell element for laminated structures, Computers
and Structures, 71(1) (1999), 43-62
[15] E. Domissy, Formulation et évaluation d’éléments finis volu-
miques modifiés pour l’analyse linéaire et non linéaire des co-
ques, Université de Technologie de Compiègne, France, 1997 (in
French)
[16] K. Y. Sze, S. Yi, M. H. Tay, An explicit hybrid stabilized eighteen-
node solid element for thin shell analysis, International Journal
for Numerical Methods in Engineering, 40(10) (1997), 1839-1856
[17] R. Hauptmann, K. Schweizerhof, A systematic development
of ‘solid-shell’ element formulations for linear and non-linear
analyses employing only displacement degrees of freedom, In-
ternational Journal for Numerical Methods in Engineering, 42(1)
(1998), 49-69
[18] L. Vu-Quoc, X. G. Tan, Optimal solid shells for non-linear anal-
Unauthenticated
Download Date | 8/27/18 2:17 PM
20 | Y. Wang et al.
yses of multilayer composites. I. Statics, Computer Methods in
Applied Mechanics and Engineering, 192(9-10) (2003), 975-1016
[19] L. Vu-Quoc, X. G. Tan, Optimal solid shells for non-linear analy-
ses of multilayer composites. II. Dynamics, Computer Methods
in Applied Mechanics and Engineering, 192(9-10) (2003), 1017-
1059
[20] R. A. S. Moreira, R. J. Alves de Sousa, R. A. F. Valente, A solid-
shell layerwise finite element for non-linear geometric and ma-
terial analysis, Composite Structures, 92(6) (2010), 1517-1523
[21] K. Rah, W. Van Paepegem, A. M. Habraken, J. Degrieck, A partial
hybrid stress solid-shell element for the analysis of laminated
composites, Computer Methods in Applied Mechanics and En-
gineering, 200(49-52) (2011), 3526-3539
[22] K. Rah, W. Van Paepegem, A. M. Habraken, J. Degrieck, A mixed
solid-shell element for the analysis of laminated composites,
International Journal for Numerical Methods in Engineering,
89(7) (2012), 805-828
[23] H. Naceur, S. Shiri, D. Coutellier, J. L. Batoz, On the modeling and
design of composite multilayered structures using solid-shell fi-
nite element model, Finite Elements in Analysis and Design, 70-
71 (2013), 1-14
[24] T. H. H. Pian, Derivation of element stiffness matrices by as-
sumed stress distributions, AIAA Journal, 2(7) (1964), 1333-1336
[25] S. N. Atluri, R. H. Gallagher, O. C. Zienkiewicz, et al.,Hybrid and
Mixed Finite Element Models, John Wiley, New York, 1983
[26] S. V. Hoa, W. Feng, Hybrid Finite Element Method for Stress
Analysis of Laminated Composites, Kluwer Academic Publish-
ers, Massachusetts, 1998
[27] R. Bellman, J. Casti, Differential quadrature and long-term in-
tegration, Journal of Mathematical Analysis and Applications,
34(2) (1969), 235-238
[28] F. Tornabene, N. Fantuzzi, M. Bacciocchi, et al., Higher-order
theories for the free vibrations of doubly-curved laminated pan-
els with curvilinear reinforcing fibers by means of a local ver-
sion of the GDQ method, Composites Part B Engineering, 81(1)
(2015), 196-230
[29] F. Tornabene, General higher-order layer-wise theory for free vi-
brations of doubly-curved laminated composite shells and pan-
els, Mechanics of Advanced Materials and Structures, 23(9)
(2015), 1046-1067
[30] F. Tornabene, N. Fantuzzi, M. Bacciocchi, Higher-order struc-
tural theories for the static analysis of doubly-curved laminated
composite panels reinforced by curvilinear fibers, Thin-Walled
Structures, 102(1) (2016), 222-245
[31] F. Tornabene, N. Fantuzzi, M. Bacciocchi, The local GDQ method
for the natural frequencies of doubly-curved shells with variable
thickness: a general formulation, Composites Part B Engineer-
ing, 92(1) (2016), 265-289
[32] L. Tang, W. Chen, Y. Liu, The quasi-conforming element tech-
nique for finite element analysis, Journal of Dalian University of
Technology, 19(2) (1980), 19-35 (in Chinese)
[33] L. Tang, W. Chen, Isoparametric quasi-conforming element,
Journal of Dalian University of Technology, 20(1) (1981), 63-74
(in Chinese)
[34] H. Liu, L. Tang, H. Lv, The quasi-conforming plane elements
with rotational degree of freedom, Computational Structural
Mechanics and Applications, 7(4) (1990), 23-31 (in Chinese)
[35] G. Shi, G. Z. Voyiadjis, Eflcient and accurate four-node quadri-
lateral C0 plate bending element based on assumed strain
fields, International Journal for Numerical Methods in Engineer-
ing, 32(5) (1991), 1041-1055
[36] G. Shi, G. Z. Voyiadjis, Geometrically nonlinear analysis of
plates by assumed strain element with explicit tangent stiffness
matrix, Computers and Structures, 41(4) (1991), 757-763
[37] W. Chen, Y. Li, Refined non-conforming quadrilateral plane
isoparametric element with drilling degrees of freedom, Chi-
nese Journal of Computational Mechanics, 10(1) (1993), 22-29
(in Chinese)
[38] C. Wang, H. Zhang, G. Shi, 3-D finite element simulation of im-
pact damage of laminated plates using solid-shell interface el-
ements, Applied Mechanics and Materials, 130 (2011), 766-770
[39] G. Shi, H. Zhang, J. Wang, Z. Wang, Progressive delamination
simulation of laminated plates based on a solid-shell interface
element with damage-node model, Procedia Engineering, 31
(2012), 324-330
[40] G. Shi, L. Tang, Weak forms of generalized governing equations
in theory of elasticity, Interaction and multiscale mechanics,
1(3) (2008), 329-337
[41] G. Shi, Y. Liu, X. Wang, Accurate, eflcient and robust Q4-like
membrane elements formulated in Cartesian coordinates using
the quasi-conforming element technique, Mathematical Prob-
lems in Engineering, 2015 (2015), 1-14
[42] G. Shi, A new simple third-order shear deformation theory of
plates, International Journal of Solids and Structures, 44(13)
(2007), 4399-4417
[43] X. Wang, G. Shi, A refined laminated plate theory accounting
for the third-order shear deformation and interlaminar trans-
verse stress continuity, Applied Mathematical Modelling, 39(18)
(2015), 5659-5680
[44] X. Wang, Laminated composite plate theories accounting for in-
terlaminar continuity and its accurate, eflcient plate elements,
Ph.D. thesis, Tianjin University, Tianjin, China, 2015 (in Chinese)
[45] X. Wang, Y. Li and H. Xu, Structural Analysis and Application of
Elements in ANSYS, China Communications Press, Beijing, 2011,
239-244 (in Chinese)
[46] X. Wang, G Shi, Evaluation of various laminated plate theories
accounting for interlaminar transverse shear stress continuity,
Advanced Materials Research, 716 (2013), 119-126
[47] S. Srinivas, A refined analysis of composite laminates, Journal
of Sound and Vibration, 30(4) (1973), 495-507
[48] S. Hardy, Composite Benchmarks, NAFEMS, Glasgow, 2001
Unauthenticated
Download Date | 8/27/18 2:17 PM