You are on page 1of 16

Thin-Walled Structures 101 (2016) 141–156

Contents lists available at ScienceDirect

Thin-Walled Structures
journal homepage: www.elsevier.com/locate/tws

Full length article

NURBS-based isogeometric analysis of buckling and free vibration


problems for laminated composites plates with complicated cutouts
using a new simple FSDT theory and level set method
Tiantang Yu a,n, Shuohui Yin a, Tinh Quoc Bui b,n, Shifeng Xia a,c, Satoyuki Tanaka d,
Sohichi Hirose b
a
Department of Engineering Mechanics, Hohai University, Nanjing 210098, PR China
b
Department of Mechanical and Environmental Informatics, Tokyo Institute of Technology, 2-12-1-W8-22, Ookayama, Meguro-ku, Tokyo 152-8552, Japan
c
College of Energy and Electrical Engineering, Hohai University, Nanjing 210098, PR China
d
Graduate School of Engineering, Hiroshima University, Japan

art ic l e i nf o a b s t r a c t

Article history: We combine the isogeometric analysis, the level set and a simple first-order shear deformation theory (S-
Received 31 January 2015 FSDT) to form a new effective and accurate approach for simulating free vibration and buckling problems
Received in revised form of laminated composite plates with cutouts. The problem domain is discretized using non-uniform ra-
4 December 2015
tional B-spline (NURBS) basis functions, which are utilized for geometry and field variables approx-
Accepted 8 December 2015
Available online 14 January 2016
imation. In S-FSDT, the shear locking effect is no longer valid and more interestingly, low computational
cost is saved because of having fewer unknowns. The trimmed NURBS surface often used in existing
Keywords: approaches, which has some disadvantages in describing the geometrical structure with internal cutouts
Isogeometric analysis due to the tensor product of the NURBS basis functions, is no longer required in the present formulation,
NURBS
and the level set method (LSM) is employed to describe the cutouts instead. More interestingly, the
First-order shear deformation theory
numerical integration in the developed method is performed only inside the physical domain. All those
Laminated composite plates
Cutouts features make the method effective in modeling cutouts with complicated shapes. Numerical examples
Level set method are presented to show the performance of the method. The results obtained are validated against re-
Free vibration ference solutions showing a good agreement and performance.
Buckling & 2015 Elsevier Ltd. All rights reserved.

1. Introduction cutouts for practical problems often requires advanced numerical


methods rather than analytical approaches. In this line, experi-
Laminated composite materials and structures have been mental methods are of course very useful but their costs are
widely used in a variety of engineering applications due to their usually high. Many numerical methods have introduced to the
excellent properties such as high stiffness and strength-to-weight analysis of vibration and buckling problems of laminated compo-
ratios, long fatigue life, and wear resistance, etc. Composite plates site plates with cutouts including the finite element method (FEM)
are one of the key and common structural elements found in civil, [1–5], the Rayleigh–Ritz method [6], the mesh-free method [7,8],
marine, mechanical, and aerospace etc. engineering fields. Differ- the finite strip method [9–12], and the extended finite element
method (XFEM) [13].
ent shaped cutouts are often created in the plate elements as many
In the numerical analysis of laminated composite plates, the
practical requirements. The presence of cutouts may drastically
approach used is another key issue besides the numerical method.
affect the mechanical behavior of structure in whole or part, thus a
One of the popular approaches to model laminated composite
thorough understanding of dynamic and buckling characteristics plates is to rely on equivalent single-layer models such as classical
of composite plates with cutouts is essential and important to the laminated plate theory [13–16], first-order shear deformation the-
design of engineering applications. ory (FSDT) [17–19], and high-order shear deformation theory
Due to the inherent anisotropic and heterogeneous behaviors (HSDT) [20–22]. The FSDT is simple to implement and offers better
of materials in nature, analysis of laminated composite plates with results than the classical laminated plate theory because the gen-
eralized displacement field takes into account the transverse shear
n
Corresponding authors.
strains. In addition, the computational cost based on the FSDT is
E-mail addresses: tiantangyu@hhu.edu.cn (T. Yu), relatively cheaper than that employing the HSDT. The FSDT only
tinh.buiquoc@gmail.com (T.Q. Bui). requires C0-continuity of generalized displacements. However, it

http://dx.doi.org/10.1016/j.tws.2015.12.008
0263-8231/& 2015 Elsevier Ltd. All rights reserved.
142 T. Yu et al. / Thin-Walled Structures 101 (2016) 141–156

requires the so-called shear correction factors to take into account completely removed. In addition, the C1 continuity requirement of
the non-linear distribution of shear stress terms. Recently, a simple the S-FSDT is easily handled by the present formulation without
FSDT (S-FSDT) was formulated in [23,24] by partitioning the additional efforts due to the high-order continuity of NURBS basis
transverse deflection into the bending and shear components. functions. The accuracy, applicability and performance of the
Compared with the traditional FSDT, one unknown can be saved proposed approach are demonstrated through a number of re-
and the shear locking is free in the S-FSDT. However, the S-FSDT
presentative numerical examples. Numerical results for free vi-
requires C1 continuity of the generalized displacements leading to
bration and buckling of isotropic and laminated composite plates
the second-order derivative of the stiffness formulation which is
with simple to complicated cutouts subjected to in-plane stresses
awkward in conventional finite element analysis based on the C0
continuity. are analyzed, validated against the reference solutions available in
Isogeometric analysis (IGA) [25] is a new numerical method the literature.
that shares the same splines basis function for representing the The rest of this paper is structured as follows: basic equations
geometry in design and solution approximations in analysis. The for laminated composite plates are presented in Section 2. In
IGA originally devotes to a cost-effective approach that is to fill the Section 3, IGA formulation for laminated composite plates is de-
gap between geometric modeling and numerical simulation, and rived. Numerical examples with complicated shapes for the free
may eliminate the meshing difficulty. IGA is based on CAD splines vibration and buckling problems of laminated composite plates
(e.g. NURBS), and proved to be an efficient, highly accurate, robust with cutouts are presented in Section 4 to verify the accuracy and
and higher order continuity approach. The inherently desirable effectiveness of the present method. Section 5 concludes.
characteristics of IGA makes it superior to classical finite element
method in many aspects and has successfully applied to many
engineering problems, including steel–concrete beams [26],
structural problems [27–30], fluid mechanics [31], fluid–structure 2. Basic equations for laminated composite plates
interaction [32], fracture mechanics [33], contact mechanics [34],
unsaturated flow problem in porous media [35], and structural 2.1. The displacements, strains, and stresses of plates
shape optimization [36], etc.
For complex geometry like structures with cutout, the trimmed The xy-plane is the mid-plane of the plate, and the positive
NURBS surface is often used in IGA. However, many geometrical z-axis is upward from the mid-plane. In the S-FSDT, the displace-
models are still hard to be solved by using the trimmed NURBS
ment field (u, v, w ) can be expressed in terms of four unknown
surface. Shojaee et al. [37,38] artificially divided the square plate
variables as follows [23,30,43]:
with a hole of complicated shape into several NURBS patches and
then the bending strip method applied to maintain C1 continuity
between patches. However, this method is in general ineffective u (x, y , z ) = u0 (x, y) − z ∂wb (x, y)/∂x
and not straightforward to unify design and analysis. To deal with v (x, y , z ) = v0 (x, y) − z ∂wb (x, y)/∂y
the trimmed NURBS surfaces, Schmidt et al. [39] developed a local w (x, y , z ) = wb (x, y) + ws (x, y) (1)
reconstruction technique to rebuild trimmed elements with se-
parate patches, in which the geometric errors and loss of the where u0 and v0 represent the displacements at the mid-plane of a
higher-order continuity along the interior and limited to single plate in the x, and y directions respectively; the transverse dis-
trimming curves are introduced. Numerical treatment of the
placement at the mid-plane in the z direction is divided into
trimmed objects such as internal cutouts in terms of NURBS-based
bending component wb and shear component ws .
IGA owns some drawbacks primarily due to the tensor product of
Because the rotations are obtained by using the partial deri-
the NURBS basis functions. Numerical approaches like the XFEM
vatives of the bending component wb conforming discretizations
[40] and Finite Cell Method (FCM) [41,42] are associated with level
set method (LSM) to describe the geometry of the computational of the S-FSDT are inherently free from the issue of shear-locking.
domain, and the numerical integration is used only inside the By making small strain assumptions, the strain–displacement
physical domain. By this way, it is thus easy to deal with complex relations are expressed as
geometry models.
Recently, the S-FSDT theory has successfully integrated into the ⎧ ∂u0 ∂ 2w ⎫
⎪ − z 2b ⎪
IGA by the authors to the analysis of linear and geometrically non- ⎪ ∂x ∂x ⎪
linear problems of functionally graded plates [30,43]. We have ⎪ 2 ⎪
∂v0 ∂ w
shown that the new approach based on the IGA and the S-FSDT is ⎧ εx ⎫ ⎪ − z 2b ⎪
very effective and highly accurate in modeling structural problems ⎪ εy ⎪ ⎪ ∂y ∂y ⎪
⎪γ ⎪ ⎪ ⎪ 2


like plates. In order to effectively treat cutouts in plates, especially ⎨ xy ⎬ = ⎨ ∂u0 ∂v0 ∂ wb ⎬
⎪ γxz ⎪ ⎪ ∂y + − 2 z
laminated composite plates, the present works developed by the ∂x ∂x∂y ⎪
⎪ ⎪ ⎪ ⎪
authors need further study and development. Therefore, the main γ
⎩ yz ⎭ ⎪ ∂ws ⎪
objective of our work is to perform the arisen task as we aim to ⎪ ∂x ⎪
improve the approach presented in [30,43] by integrating the LSM ⎪ ⎪
⎪ ∂ ws ⎪
to accurately simulating cutouts in complicated laminated com- ⎪
⎩ ∂y ⎪
⎭ (2)
posites plates. In other words, we develop a new numerical model
in which the S-FSDT and LSM are simultaneously integrated into
Eq. (2) can be written in matrix form:
the IGA for modeling buckling and free vibration problems of la-
minated composites plates with cutouts. In the new approach, the
εp ε0 − zκ
LSM is mainly to describe the cutouts, while numerical integration
is used only inside the physical domain. As a result the trimmed
ε= {} {
εs
=
γ } (3)
NURBS surface is no longer required to describe the geometrical
structure with cutouts. By employing the S-FSDT, the shear locking where ε p and εs represent the in-plane strain vector and the
problem often encountered in the conventional plate theories is transverse shear strain vector, respectively.
T. Yu et al. / Thin-Walled Structures 101 (2016) 141–156 143

⎧ ∂ 2w ⎫ ⎡ε ⎤ ⎡A B ⎤
⎧ ∂u0 ⎫ ⎪ b
⎪ εm = ⎢⎣ 0 ⎥⎦, D = ⎢ ⎥
⎪ ⎪ ⎪ ∂x2 ⎪ ⎧ ∂ws ⎫ κ ⎣ B Db⎦ (9)
⎪ ∂x ⎪
⎪ ∂v0 ⎪ ⎪
⎪ ∂ 2wb ⎪⎪ ⎪
⎪ ∂x ⎪⎪
ε0 = ⎨ ⎬; κ = ⎨ ⎬; γ = ⎨ ⎬ in which ρ is density; A , B , Db , and Ds are matrices of extensional
⎪ ∂y ⎪ ⎪ ∂y ⎪
2 ∂w
⎪ s⎪
⎪ stiffness, bending-extensional coupling stiffness, bending stiffness,
⎪ ∂u0 ∂v ⎪ ⎪ 2 ⎪ ⎩ ∂y ⎪⎭
⎪ + 0⎪ ⎪ 2 ∂ wb ⎪ and transverse shearing stiffness, respectively, defined as [44]:
⎩ ∂y ∂x ⎭ ⎪ ⎪
⎩ ∂x∂y ⎭ (4) h /2
(Aij , Bij , Dijb ) = ∫−h /2 Q¯ ij (1, z, z2) dz (i, j = 1, 2, 6) (10a)
The laminate is usually made of several orthotropic layers in
which the stress–strain relation for the kth orthotropic lamina
with the arbitrary fiber orientation compared with the reference h /2
axes is given by Dijs = ∫−h /2 Q¯ ij dz (i, j = 4, 5) (10b)
⎡ Q¯ Q¯ 12 Q¯ 16 0 0 ⎤ ⎧ εx ⎫(k)
(k )
⎧ σx ⎫(k) ⎢ 11 ⎥ where h is plate thickness.
⎪ σy ⎪ ⎢ Q¯ 21 Q¯ 22 Q¯ 26 0 0 ⎥ ⎪ εy ⎪ To account for the plates subjected to in-plane stresses σx0 , σy0
⎪ ⎪ ⎢ ⎥ ⎪ ⎪
⎨ τxy ⎬ 0 ⎥ ⎨ γxy ⎬
0
= ⎢ Q¯ 61 Q¯ 62 Q¯ 66 0 and τxy , the in-plane stresses virtual work should be included in
⎪ τxz ⎪ ⎪γ ⎪
⎪τ ⎪ ⎢ 0 0 0 Q¯ 55 Q¯ 54 ⎥ ⎪ xz ⎪ the model, which is the work done by the initial stresses acting on
⎩ yz ⎭ ⎢ ⎥ ⎩ γyz ⎭ the non-linear strain parts [45,46]:
⎣ 0 0 0 Q¯ 45 Q¯ 44 ⎦ (5)
⎧ 2⎫ ⎧ 1 ⎛ ∂wb ∂ws ⎞
2 ⎫
⎪ 1 ⎛ ∂w 0 ⎞ ⎪ ⎪ ⎜ ⎟

⎪ 2 ⎜⎝ ∂x ⎟⎠ ⎪ ⎪
+ ⎪
where ⎪ ⎪ ⎪
2 ⎝ ∂x ∂ x ⎠

⎪ 2 ⎪ ⎪⎪ 2 ⎪

εnl = ⎨ 1 ⎛ ∂w 0 ⎞ ⎬ = ⎨ 1 ⎛ ∂wb ∂ws ⎞ ⎬
Q¯ 11 = Q 11 cos4 θ + 2 ( Q 12 + 2Q 66 ) sin2 θ cos2 θ + Q 22 sin4 θ (6a) ⎪ 2 ⎜ ∂y ⎟ ⎪ ⎪ ⎜
2 ⎝ ∂y
+
∂y ⎠
⎟ ⎪
⎪ ⎝ ⎠ ⎪ ⎪ ⎪
⎪ ∂w ∂w ⎪ ⎪ ⎛ ∂w w ⎞ ⎛ ∂wb w ⎞⎪
⎪ 0 0 ⎪ ⎪⎜ b ∂ s ∂ s
+ ⎟⎜ + ⎟⎪
⎩ ∂x ∂y ⎭ ⎪ ⎩ ⎝ ∂x ∂x ⎠ ⎝ ∂y ∂y ⎠ ⎪⎭ (11)

(
Q¯ 12 = ( Q 11 + Q 22 − 4Q 66 ) sin2 θ cos2 θ + Q 12 sin4 θ + cos4 θ ) (6b) To the end, the virtual work of the initial stresses yields
⎧ 2 2 ⎫
⎪ ⎛ δwb δws ⎞⎟ 1 ⎛ δw δws ⎞ ⎛ ⎞
0 ⎜⎛ δwb + δws ⎟⎞ ⎜ δwb + δws ⎟ ⎪
δW = ∫Ω ⎨⎪ 21 σx0 ⎝⎜ ∂x
+
∂x ⎠
+ σ y0 ⎜ b +
2 ⎝ ∂y
⎟ + τxy
∂y ⎠ ⎝ ∂x ∂x ⎠ ⎝ ∂y
⎬ dΩ
∂y ⎠ ⎪
⎩ ⎭ (12)
Q¯ 16 = ( Q 11 − Q 12 − 2Q 66 ) sin θ cos3 θ
In the case of in-plane buckling analyses for pre-buckling
+ ( Q 12 − Q 22 + 2Q 66 ) sin3 θ cos θ (6c) σx0 τxy
0
stresses σ0 = [ ], the weak form can be expressed as
0
τxy σy0

Q¯ 22 = Q 11 sin4 θ + 2 ( Q 12 + 2Q 66 ) sin2 θ cos2 θ + Q 22 cos4 θ (6d)


∫Ω δ εTm Dεm dΩ + ∫Ω δ εTs Ds εs dΩ
+h ∫Ω ∇T δ ( wb + ws ) σ0∇ ( wb + ws ) dΩ = 0 (13)
Q¯ 26 = ( Q 11 − Q 12 − 2Q 66 ) sin3 θ cos θ
T
+ ( Q 12 − Q 22 + 2Q 66 ) sin θ cos3 θ (6e) where ∇ = [∂/∂x ∂/∂y ] is the gradient operator.

Q¯ 66 = ( Q 11 + Q 22 − 2Q 12 − 2Q 66 ) sin2 θ cos2 θ 3. IGA formulation for laminated composite plates

(
+ Q 66 sin4 θ + cos4 θ ) (6f) 3.1. The NURBS basis functions

with The non-uniform rational B-spline (NURBS) is a parametric


E1 v12 E2 E2 function of rational polynomials B-spline. The NURBS basis func-
Q 11 = ; Q 12 = ; Q 22 = ; tions are created via the B-splines that are a linear combination of
1 − v12 v21 1 − v12 v21 1 − v12 v21
B-spline basis functions. In the parametric space ξ ∈ [0, 1], a knot
Q 66 = G12; Q 55 = G13; Q 44 = G23 (7)
vector k (ξ ) is a set of non-decreasing numbers which are between
in which, θ is the fiber orientation; E1 and E2 are the Young zero and one, i.e., k (ξ ) = {ξ1 = 0, … , ξi, … , ξn + p + 1 = 1}, where i is
modulus in the x and y directions, respectively; v12 and v21 are the the knot index, ξi is the ith knot, n is the number of basis functions,
Poisson’s ratios; and G12 , G23, and G13 are the shear modulus in the and p is the order of the polynomial. A knot vector is called an
open knot vector if the knots are repeated p + 1 times at the start
xy, yz, and zx planes, respectively.
and end of the vector. Basis functions that are defined with an
open knot vector are interpolatory at the beginning and end of the
2.2. Weak form parametric space interval; thus, the open knot vector is used in the
analysis.
The weak form of free vibration analysis of the laminated Given a knot vector k (ξ ), the ith B-spline basis function of de-
gree p, written as Ni, p (ξ ), can be constructed recursively following
composite plate can be derived from the principle of virtual work
the Cox-de Boor formulation [25,33]:
as
⎧1 if ξi ≤ ξ < ξi + 1
∫Ω δ εTm Dεm dΩ + ∫Ω δ εTs Ds εs dΩ = ∫Ω δ uT ρu¨ dΩ Ni,0 (ξ ) = ⎨ for p = 0
(8) ⎩0 otherwise (14)

with and
144 T. Yu et al. / Thin-Walled Structures 101 (2016) 141–156

follows:
Ni, p (ξ ) Nj, q (η) wi, j
Rip, j, q (ξ , η) = n m
∑i = 1 ∑ j = 1 Ni, p (ξ ) Nj, q (η) wi, j (17)

where wi, j represents the 2D weight; Ni, p (ξ ) and Nj, q (η) are the
B-spline basis functions of order p in the ξ direction and order q in
the η direction, respectively; Nj, q (η) follows the recursive formula
shown in Eqs. (14) and (15) with knot vector k (η). The definition of
k (η) is similar to that of k (ξ ).
By using the NURBS basis functions, a NURBS surface of order p
in the ξ direction and order q in the η direction can be constructed
as follows:
n m
S (ξ , η) = ∑ ∑ Rip, j, q (ξ, η) Bi, j
i=1 j=1 (18)

where B i, j represents the coordinates of control points in two


dimensions.
Fig. 1. Quadratic B-spline basis functions.
3.2. Discrete equations

The displacement field is approximated based on the NUBRS


basis functions as
NP
u h0 = ∑ RI uI
I=1 (19)

with
T
u h0 = ⎡⎣ u0h v0h wbh wsh ⎤⎦ (20a)

uI = [uI vI wbI wsI ]T (20b)

where NP = (p + 1)(q + 1) is the number of control points per


element, RI and uI denote the shape function and the unknown
displacement vector at control point I , respectively.
By substituting Eq. (19) into Eq. (4), one can obtain the fol-
lowing:
NP NP NP
ε0 = ∑ BmI uI , κ = ∑ BbI uI , γ = ∑ BsI uI
I=1 I=1 I=1 (21)

with
⎡ RI, x 0 0 0⎤ ⎡ 0 0 RI, xx 0⎤
Fig. 2. A square plate with a circular hole. ⎢ ⎥ ⎢ ⎥
Bm
I = ⎢ 0 RI, y 0 0⎥, B I = ⎢ 0 0 RI, yy 0⎥, B sI
b

⎢ ⎥ ⎢ ⎥
⎣ RI, y RI, x 0 0⎦ ⎣ 0 0 2RI, xy 0⎦
⎡ 0 0 0 RI, x ⎤
ξ − ξi ξi + p + 1 − ξ =⎢ ⎥
Ni, p (ξ ) = Ni, p − 1 (ξ ) + Ni + 1, p − 1 (ξ ) for p ≥ 1 ⎢⎣ 0 0 0 RI, y ⎥⎦
ξi + p − ξi ξi + p + 1 − ξi + 1 (15) (22)

Fig. 1 illustrates a set of one-dimensional quadratic B-spline Substituting Eqs. (19) and (21) into Eqs. (8), (12) and (13), the
basis functions for open uniform knot vector formulations of the free vibration and buckling problems are re-
k (ξ ) = {0, 0, 0, 0.2, 0.4, 0.6, 0.8, 1, 1, 1} . It can be observed that a written in the following form:
B-Spline basis function is C1 continuous inside a knot span, i.e.,
between two distinct knots, and Cp  1 continuous at a single knot.
(K + K σ
− ω2M U = 0 ) (23)

The NURBS basis function Ri, p(ξ ) in the framework of partition


of unity is constructed by a weighted average of the B-spline basis
functions [25]: (K − N cr K
G
)U = 0 (24)

Ni, p (ξ )wi where ω is the natural frequency, Ncr is the critical buckling load, U
Ri, p(ξ ) = n
∑ j = 1 Nj, p (ξ ) wj is the eigenvector {w1, w2, ... , wNP } consisted of deflection at all
(16)
control points, K , K σ , M and K G are the global stiffness matrix,
where wi is the ith weight, and 0 < wi ≤ 1. initial stresses stiffness matrix, mass matrix and geometrical
Similarly, 2D NURBS basis functions can be constructed by stiffness matrix.
taking the tensor product of two 1D B-spline basis functions as The global stiffness matrix K is given by the following
T. Yu et al. / Thin-Walled Structures 101 (2016) 141–156 145

(a) 15×15 control points (b) 27×27 control points


Fig. 3. NURBS meshes for the square plate, green circles denote control points. (For interpretation of the references to color in this figure legend, the reader is referred to the
web version of this article.)

Table 1 Table 2
Normalized frequency parameters of clamped square plate with a circular hole for Normalized frequencies of simply supported square plate with a centre square hole.
different meshes.
Mode Present FEM Liu et al. [7] Huang et al. [48] Ali et al. [49]
Mode 9  9 15  15 21  21 27  27 33  33 Huang et al. Liu et al.
[48] [7] 1 4.9648 4.9615 4.9217 4.839 4.936
2 6.5476 6.5011 6.4810 6.435 6.502
1 6.217 6.182 6.182 6.181 6.180 6.240 6.149 3 6.5499 6.5019 6.4821 6.440 6.502
2 8.784 8.741 8.663 8.649 8.650 8.457 8.577 4 8.6613 8.6507 8.5509 8.492 8.525
3 8.784 8.741 8.663 8.649 8.650 8.462 8.634 5 8.9269 8.7976 8.8656 8.875 8.813
4 10.558 10.512 10.505 10.501 10.503 10.233 10.422 6 10.8968 10.8325 10.720 10.805 –
5 11.635 11.554 11.542 11.545 11.546 11.719 11.414 7 10.8982 10.8341 10.767 10.831 –
6 12.064 12.044 12.043 12.041 12.034 12.299 11.838 8 12.3188 12.3144 12.045 12.291 –
7 13.168 13.076 13.012 13.004 13.004 13.037 12.829 9 13.5544 13.4599 13.370 13.534 –
8 13.168 13.076 13.012 13.004 13.004 13.041 12.842 10 14.4913 14.4912 14.180 14.108 –

(a) geometry (b) mesh with 21×21 control points


Fig. 4. Geometry and mesh of square plate with a square cutout. (a) Geometry. (b) Mesh with 21  21 control points.
146 T. Yu et al. / Thin-Walled Structures 101 (2016) 141–156

Table 3
Normalized frequencies of simply supported square plate with a centre square hole for different thickness ratio h/a.

Mode h/a ¼10000 h/a ¼1000 h/a ¼ 100 h/a ¼10

IGA FEM IGA FEM IGA FEM IGA FEM

1 4.9657 4.9615 4.9657 4.9615 4.9648 4.9615 4.8742 4.9615


2 6.5498 6.5011 6.5498 6.5011 6.5476 6.5011 6.3455 6.5011
3 6.5522 6.5019 6.5521 6.5019 6.5499 6.5019 6.3478 6.5019
4 8.6667 8.6507 8.6667 8.6507 8.6613 8.6507 6.7131 6.7107
5 8.9335 8.7976 8.9334 8.7976 8.9269 8.7976 8.2073 8.2923
6 10.9085 10.8325 10.9083 10.8325 10.8968 10.8325 8.3493 8.2925
7 10.9099 10.8341 10.9097 10.8341 10.8982 10.8341 8.3539 8.6507
8 12.3351 12.3144 12.3350 12.3144 12.3188 12.3144 8.3884 8.7976
9 13.5778 13.4599 13.5776 13.4599 13.5544 13.4599 10.0136 10.0748
10 14.5183 14.4912 14.5180 14.4912 14.4913 14.4912 10.0160 10.3483

⎧ m ⎫T ⎡
A B ⎤ ⎧ Bm ⎫
∫Ω ⎨⎩ Bb ⎬⎭
T
K= ⎢ ⎥ ⎨ ⎬ dΩ + ∫Ω ( Bs) D sBsdΩ
Table 4
B ⎣ B D b ⎦ ⎩ Bb ⎭ (25)
Normalized frequencies of square plate with a centre square hole for different
boundary conditions.
and the initial stresses stiffness matrix K σ is given by
Mode FCFC SCSC CCCC
Kσ = ∫Ω ( Bσ)T σ0BσdΩ (26)
IGA FEM IGA FEM IGA FEM

1 5.1200 5.1133 6.6838 6.665428 8.2752 8.2781 with


2 5.1886 5.1765 7.0753 7.020709 8.9793 8.9502
3 7.9392 7.9308 8.6121 8.592881 8.9832 8.9509 ⎡ 0 0 RI, x RI, x ⎤
4 8.0338 8.0228 9.6940 9.685091 10.5368 10.4024 BσI = ⎢ ⎥
⎢⎣ 0 0 RI, y RI, y ⎥⎦ (27)
5 8.2006 8.1911 9.9963 9.880878 10.5840 10.5779
6 8.6985 8.6351 11.5112 11.42871 12.4148 12.3376
7 9.6227 9.5793 11.9847 11.92509 12.4191 12.3397 and the global mass matrix M is expressed as
8 10.3201 10.1812 13.2058 13.19959 13.9450 13.9358
T
9 11.1443 11.1128 14.2295 14.11501 14.8372 14.6856 ⎧ 1⎫ ⎧ 1⎫
10 11.3980 11.3812 14.9985 14.99941 16.1732 16.1357 M= ∫Ω ⎨⎩ N2⎬⎭ m ⎨⎩ N2⎬⎭ dΩ
Ν Ν (28)

Fig. 5. Mode shapes 1–6 of a CCCC square plate with a central square hole.
T. Yu et al. / Thin-Walled Structures 101 (2016) 141–156 147

Table 5
Critical buckling coefficient k for a simply supported square plate with a square cutout.

Mode 99 16  16 21  21 28  28 33  33 IGA with CPT FSM [10] FSM [50] FEM [51]

1 2.062 2.017 1.999 1.992 1.988 1.999 1.9837 2.000 2.090


2 4.806 3.917 3.810 3.747 3.726 3.811
3 12.025 10.053 9.978 9.904 9.890 9.986
4 13.492 12.315 12.184 12.122 12.100 12.193
5 16.880 15.704 15.642 15.509 15.495 15.651
6 19.084 18.287 18.268 18.138 18.100 18.272
7 19.826 18.521 18.304 18.243 18.236 18.318
8 23.986 20.793 20.363 20.130 20.059 20.376

Fig. 6. The first eight mode shapes of a square plate with a square cutout.

with
⎡ RI 0 0 0 ⎤ ⎡ 0 0 RI, x 0⎤
⎡h 0 ⎤ 1 ⎢ ⎥ ⎢ ⎥
m=⎢ 3
⎥, NI = ⎢ 0 RI 0 0 ⎥ , N2
I = ⎢ 0 0 RI, y 0⎥
⎣ 0 z ρ /12⎦ ⎢⎣ 0 0 RI RI ⎥⎦ ⎢⎣ ⎥
0 0 0 0⎦ (29)
G
and the geometric stiffness matrix K is
T
KG = ∫Ω ( B g) σ0B gdΩ (30)

with
⎡ 0 0 RI, x RI, x ⎤
B Ig = ⎢ ⎥
⎢⎣ 0 0 RI, y RI, y ⎥⎦ (31)

3.3. Treatment of cutout in IGA

A geometric model with cutout is tackled via Boolean opera-


tions, i.e., union, intersection and difference, which hence results in
a trimmed NURBS object. Trimmed NURBS object in the CAD
system is only visually trimmed, the parameterization remains
unchanged and the cropped part is not visualized [39]. This im-
plies that the geometric information of the structural object has
Fig. 7. Geometry of a square plate with a complicated cutout (unit:m). changed, nevertheless the NURBS topology has not altered and the
entire construction process is being stored. The method described
148 T. Yu et al. / Thin-Walled Structures 101 (2016) 141–156

Table 6 circular cutout, the minimum signed distance to construct the


Normalized frequencies of a SSSS laminated square plate with complicated cutout level set function is as follows:
for various orientations.
φ (x, 0) = ‖x − x c ‖ − rc (34)
Angle Method Mode
ply where x is any query point, x c denotes the center coordinates of the
1 2 3 4 5 6 circle and rc is the radius of the circle. The level set function will have
(0°, 0°, Present 18.190 30.928 36.073 56.390 61.986 82.894
a positive value outside the circle and a negative distance value for
0°) IGA with 18.194 30.932 35.678 55.383 62.164 81.795 any point inside the circle. The points that lies on the circle will be
8-patches [37] represented as a zero value of the function. Similarly, the level set
EFG [52] 18.226 31.127 36.237 56.874 62.390 83.565 function for a cutout with any shape can be constructed.
MKI [52] 18.169 30.303 36.581 57.429 64.145 85.656
(15°, Present 19.097 32.141 36.449 57.541 63.320 84.702
We use the displacement field described in Eq. (32), and use
 15°, IGA with 18.912 32.045 36.004 56.345 63.370 83.629 the LSM to represent the location of cutout, thus the cutout is not
15°) 8-patches [37] considered in generating the computational mesh for IGA, in other
EFG [52] 19.177 32.445 37.238 58.716 63.994 86.500 words, the Boolean operation on the NURBS geometry to describe
MKI [52] 18.323 31.472 37.617 63.077 66.538 86.486
(30°, Present 20.603 33.987 37.600 59.762 65.646 88.731
cutout can be avoided. Hence, the trimmed NURBS functions are
 30°, IGA with 20.316 33.933 37.074 58.484 65.895 87.973 not be used in discretization of the computation domain. In the
30°) 8-patches [37] analysis, the location of an integration point is determined with
EFG [52] 20.926 34.915 39.101 62.222 67.054 92.715 the level set function value, i.e., integration point x p locates out-
MKI [52] 20.310 33.987 39.898 58.111 69.699 92.099
side the physical domain for φ (x p) < 0, while integration point x p
(45°, Present 21.310 34.791 38.278 60.859 66.841 91.527
 45°, IGA with 20.982 34.848 37.559 59.325 67.518 91.220 locates inside the physical domain for φ (x p) > 0.
45°) 8-patches [37] In practice, the displacement function is not faithfully im-
EFG [52] 21.736 36.079 39.975 63.897 68.525 96.767 plemented as described in Eq. (32). We can implement the dis-
MKI [52] 20.987 34.897 39.269 63.375 69.017 96.588
placement function similar to that in conventional IGA and merely
(0°, 90°, Present 18.199 31.075 36.087 56.443 62.487 83.593
0°) IGA with 18.190 31.087 35.655 55.452 62.582 82.383 remove the integral on the void part in calculating the element
8-patches [37] contribution to K , K σ , M and K G , which are calculated as follows:
EFG [52] 18.278 32.264 36.134 57.151 65.853 90.678
⎧ m ⎫T ⎡
MKI [52] 18.027 32.506 37.268 57.698 70.768 92.998 A B ⎤ ⎧ Bm ⎫
∫Ω H (x) ⎨⎩ Bb ⎬⎭
T
K= ⎢ ⎥ ⎨ ⎬ dΩ + ∫Ω H (x) ( Bs) D sBsdΩ
B ⎣ B D b ⎦ ⎩ Bb ⎭ (35)
in Ref. [39] for treating trimmed NURBS surface is generally not
effective and limited to single trimming curves. Kσ = ∫Ω H (x) ( Bσ)T σ0BσdΩ (36)
In XFEM, the modeling of voids is carried out using an en-
richment function for nodes that intersect the boundary of the
void, and the discretization of the boundary of the hole is thus ⎧ N1 ⎫T ⎧ N1 ⎫
M= ∫ H (x) ⎨ ⎬ m ⎨ ⎬ dΩ
independent of the mesh. The level set method (LSM) [47] is used Ω ⎩ Ν2 ⎭ ⎩ Ν2 ⎭ (37)
to represent the voids, and all geometric computations that per-
tain to the void are carried out using the level set function. The T
displacement field of plates with cutout is approximated by KG = ∫Ω H (x) ( B g) σ0B gdΩ (38)
NP
u 0h = ∑ H (x) RI uI
I=1 (32)
4. Numerical results and discussion
where
⎧1 x ∈ Ω In this section, we present three examples for isotropic plates,
H (x) = ⎨ two examples for free vibration analysis of composite plates, and
⎩0 x ∉ Ω (33)
two examples for buckling analysis of composite plates to show
Generally, the LSM is a numerical technique for representing the accuracy and effectiveness of the developed method. In all
the interfaces at any time t with a zero level set function, i.e. numerical calculations, a cubic NURBS basis function is adopted, a
φ (x, t ) = 0, where φ (x, t ) is the level set function. The initial 4  4 Gaussian quadrature scheme is used in normal element, and
condition is usually taken as the signed distance function (SDF) a triangular sub-domain technique is adopted in the elements cut
such that the level set function has positive values on one side of by internal cutout. The boundaries of the plate are denoted as
the interface and negative values at the other side of the interface simply supported (S), clamped (C), and free (F). As for clamped
and the interface is identified by the zero level set function. For a boundary conditions, the rotations are obtained from the

Table 7
Normalized frequencies of a CCCC square plate with complicated cutout for various orientations.

Angle ply Mode

1 2 3 4 5 6 7 8

(0°, 0°, 0°) 44.239 71.348 81.520 96.969 101.449 119.014 134.868 147.100
(15°,  15°, 15°) 45.052 71.720 81.148 96.742 101.823 119.294 133.722 143.736
(30°,  30°, 30°) 45.704 73.315 79.179 96.413 101.720 121.293 130.533 140.611
(45°,  45°, 45°) 45.965 75.993 76.171 96.262 101.674 124.967 126.013 139.724
(0°, 90°, 0°) 44.398 72.159 81.690 97.047 101.682 119.491 134.825 147.280
T. Yu et al. / Thin-Walled Structures 101 (2016) 141–156 149

Table 8
Normalized frequencies of a SCSC laminated square plate with complicated cutout for various orientations.

Angle ply Mode

1 2 3 4 5 6 7 8

(0°, 0°, 0°) 28.747 39.321 54.863 68.725 79.861 99.721 107.478 121.378
(15°,  15°, 15°) 29.901 39.805 56.119 70.061 81.686 102.139 107.527 119.660
(30°,  30°, 30°) 31.887 41.441 59.435 73.567 85.237 107.163 107.804 120.077
(45°,  45°, 45°) 33.850 43.212 63.817 77.524 89.028 106.188 114.701 122.988
(0°, 90°, 0°) 29.071 39.549 55.693 69.312 80.960 100.885 107.952 122.001

Table 9
The first four mode shapes of three-layer symmetric laminated composite square plate with complicated cutout for different boundary conditions.

Mode
B.C.
1 2 3 4

SSSS

SCSC

CCCC

derivatives of bending deflection component, thus, the constraint depicted in Fig. 2. The geometry and material parameters are:
on the rotations is imposed by fixing the transverse defection with a¼ 10 m, thickness h¼0.1 m, radius R ¼1 m, Young’s modulus
two rows of control points as described in Ref. [30]. E¼ 2.0  1011 N/m2, Poisson’s ratio ν ¼0.3, density ρ ¼8000 kg/m2.
The typical boundary conditions can be rewritten as Five different meshes with 9  9, 15  15, 21  21, 27  27 and
Simply supported boundary condition: 33  33 control points are used. Fig. 3 shows two different meshes
v0 = wb = wb, y = ws = ws, y = 0, on x − edge (39a) of 15  15 and 27  27 control points.
A normalized frequency parameter is defined by
ω˜ = [ρhω2a4 /D (1 − v 2)]1/4 with D = Eh3/12 (1 − v 2). Numerical re-
u0 = wb = wb, x = ws = ws, x = 0, on y − edge (39b)
sults of the first eight frequency parameters calculated by the
Clamped boundary condition: present method are presented in Table 1 for different meshes. The
present results are compared with the reference solutions derived
u0 = v0 = wb = wb, x = wb, y = ws = ws, x = ws, y = 0 (40)
from the meshfree conforming radial point interpolation method
(RPIM) [7] and the discrete Green function method (DGF) [48]. It
can be seen in the table that the present method offers good
4.1. Isotropic plates convergence and agreement with other approaches. The obtained
results are very interesting and one can also observe in the results
4.1.1. Free vibration analysis: a square plate with a circular hole that the developed IGA could offer acceptable natural frequencies
We consider a clamped square plate with a circular hole as even using a coarse mesh e.g., 9  9 control points.
150 T. Yu et al. / Thin-Walled Structures 101 (2016) 141–156

Table 11
Normalized frequencies of a SSSS laminated square plate with a hole of complicated
cutout for various orientations.

Angle ply Method Mode

1 2 3 4 5 6

(0°, 0°, Present 8.339 12.287 16.710 22.002 35.120 39.857


0°, 0°) IGA with 8.339 12.287 16.710 21.999 35.121 39.859
CPT
FEM 8.322 12.276 16.609 21.939 34.731 39.890
(0°, 30°, Present 10.174 14.780 20.531 28.372 38.555 44.631
30°, 0°) IGA with 10.174 14.781 20.531 28.370 38.557 44.633
CPT
FEM 10.144 14.752 20.426 28.301 38.514 44.704
(0°, 45°, Present 10.671 15.515 22.275 32.711 39.187 47.335
45°, 0°) IGA with 10.671 15.515 22.275 32.709 39.189 47.338
CPT
FEM 10.660 15.509 22.180 32.599 38.858 47.111
(0°, 60°, Present 10.188 15.106 22.793 35.715 38.431 48.577
60°, 0°) IGA with 10.188 15.106 22.794 35.712 38.433 48.580
CPT
FEM 10.179 15.096 22.696 35.419 38.170 48.486
(0°, 90°, Present 8.613 13.620 22.284 36.354 37.614 48.408
90°, 0°) IGA with 8.613 13.620 22.285 36.355 37.610 48.411
CPT
FEM 8.600 13.627 22.149 36.167 37.460 48.336

Table 12
Normalized frequencies ω˜ = ωa 2 (1 + ν ) ρ /E of a fully clamped isotropic square
Fig. 8. The geometry of square plate with complicated cutout (unit:m).
plate subjected to in-plane stresses.

4.1.2. Free vibration analysis: a square plate with a centre square k Method Mode
hole
1 2 3 4 5 6
Next, a simply supported square plate with a centre square hole
is considered. The geometry and its control points are shown in  5 Trapezoidal 0.2417 0.4407 0.6177 0.7338 0.7358 0.8993
Fig. 4. In the analysis, a ¼10 m, c/a ¼0.5, thickness ration h/a ¼0.01, p-element [46]
Young’s modulus E ¼2.0  1011 N/m2, Poisson’s ratio ν ¼0.3, den- Rayleigh–Ritz 0.2417 0.4407 0.6177 0.7337 0.7357 0.8992
method [53]
sity ρ ¼8000 kg/m2. The present results of normalized frequencies Rayleigh–Ritz 0.2419 0.4412 0.6193 0.7344 0.7363 0.9017
ω̃ are compared with solutions given by Liu et al. [7], Huang et al. method [45]
[48], Ali et al. [49], and the FEM (ANSYS) with a very fine mesh, FEMQ4 0.2422 0.4434 0.6221 0.7448 0.7468 0.9111
Present 0.2418 0.4408 0.6180 0.7341 0.7359 0.8999
tabulated in Table 2.
5 Trapezoidal 0.0421 0.2448 0.4142 0.5286 0.5341 0.6899
Table 3 presents the results of normalized frequencies with p-element [46]
different thickness ration h/a obtained by the present method and Rayleigh–Ritz 0.0421 0.2448 0.4141 0.5286 0.5340 0.6899
the FEM (ANSYS), and a very good agreement can be found. More method [53]
Rayleigh–Ritz 0.0421 0.2449 0.4144 0.5287 0.5341 0.6907
importantly, the numerical results also reveal that the locking can
method [45]
be avoided. Table 4 reports the results of normalized frequencies FEMQ4 0.0445 0.2489 0.4196 0.5417 0.5472 0.7033
with different boundary conditions and thickness ration h/a ¼ 100 Present 0.0422 0.2450 0.4145 0.5290 0.5342 0.6905
obtained by the present IGA method and the FEM (ANSYS). As

Table 10
Normalized frequencies of a CCCC laminated square plate with a hole of complicated cutout for various orientations.

Angle ply Method Mode

1 2 3 4 5 6

(0°, 0°, 0°, 0°) Present 29.620 32.570 46.995 51.321 67.041 69.432
IGA with CPT 29.623 32.573 47.009 51.321 67.057 69.474
FEM 29.573 32.565 46.767 51.237 66.712 69.463
(0°, 30°, 30°, 0°) Present 31.177 34.796 48.252 56.116 70.978 71.319
IGA with CPT 31.182 34.800 48.267 56.120 71.022 71.337
FEM 31.155 34.731 48.143 56.052 70.838 71.182
(0°, 45°, 45°, 0°) Present 32.963 37.690 49.244 61.770 72.515 75.511
IGA with CPT 32.968 37.695 49.259 61.773 72.562 75.529
FEM 32.909 37.826 49.174 61.554 72.558 75.309
(0°, 60°, 60°, 0°) Present 34.948 41.273 50.026 67.271 74.237 80.013
IGA with CPT 34.955 41.280 50.044 67.271 74.287 80.032
FEM 34.731 41.265 49.862 67.056 73.933 79.779
(0°, 90°, 90°, 0°) Present 37.094 45.645 50.604 72.563 76.223 84.378
IGA with CPT 37.103 45.655 50.623 72.562 76.279 84.402
FEM 37.023 45.661 50.416 72.187 76.053 84.189
T. Yu et al. / Thin-Walled Structures 101 (2016) 141–156 151

Table 13 Table 15
Normalized frequencies of a fully clamped three-ply laminated square plate sub- Normalized frequencies of a fully clamped three-ply laminated square plate with a
jected to in-plane stresses. complicated cutout subjected to in-plane compression or tension.

k Method Mode k Method Mode

1 2 3 4 5 6 1 2 3 4 5 6

3 Present 14.4107 17.1063 23.8157 34.7468 38.9377 40.4354 3 Present 17.3079 23.3413 28.7154 41.3071 42.6269 52.6315
FEMQ4 14.4677 17.2125 24.1717 35.7609 39.3510 40.8612 FEMQ4 17.1607 22.8514 28.6624 41.2440 41.7111 52.5397
2 Present 14.4967 17.2779 24.0509 35.0136 39.0129 40.5496 2 Present 17.4819 23.5807 28.9484 41.4849 42.7765 52.8423
FEMQ4 14.5534 17.3835 24.4054 36.0244 39.4256 40.9747 FEMQ4 17.3360 23.1023 28.8982 41.4202 41.8709 52.7410
1 Present 14.5820 17.4475 24.2836 35.2783 39.0879 40.6634 1 Present 17.6535 23.8164 29.1787 41.6617 42.9252 53.0507
FEMQ4 14.6383 17.5527 24.6367 36.2859 39.5000 41.0878 FEMQ4 17.5089 23.3490 29.1311 41.5950 42.0300 52.9405
 0.5 Present 14.6244 17.5317 24.3991 35.4100 39.1253 40.7202 0 Present 17.8228 24.0487 29.4062 41.8375 43.0730 53.2569
FEMQ4 14.6806 17.6366 24.7516 36.4160 39.5372 41.1442 FEMQ4 17.6794 23.5919 29.3613 41.7685 42.1885 53.1380
0 Present 14.6666 17.6153 24.5141 35.5411 39.1627 40.7769 1 Present 17.9898 24.2776 29.6311 42.0122 43.2200 53.4607
FEMQ4 14.7226 17.7200 24.8659 36.5456 39.5743 41.2005 FEMQ4 17.8477 23.8310 29.5886 41.9406 42.3463 53.3336
p-Ritz 14.6655 17.6138 24.5114 35.5318 39.1572 40.7685 2 Present 18.1546 24.5032 29.8535 42.1859 43.3663 53.6621
method FEMQ4 18.0137 24.0665 29.8134 42.1113 42.5036 53.5272
[54] 3 Present 18.3174 24.7257 30.0733 42.3586 43.5117 53.8611
0.5 Present 14.7086 17.6985 24.6285 35.6717 39.2000 40.8334 FEMQ4 18.1776 24.2985 30.0355 42.2806 42.6605 53.7188
FEMQ4 14.7645 17.8030 24.9796 36.6747 39.6113 41.2567
1 Present 14.7505 17.7813 24.7423 35.8019 39.2373 40.8899
FEMQ4 14.8063 17.8855 25.0929 36.8034 39.6484 41.3128
Table 16
2 Present 14.8338 17.9456 24.9684 36.0607 39.3117 41.0025
Normalized frequencies of a fully simply supported three-ply laminated square
FEMQ4 14.8893 18.0494 25.3177 37.0593 39.7222 41.4248
plate with a complicated cutout subjected to in-plane compression or tension.
3 Present 14.9165 18.1081 25.1923 36.3177 39.3860 41.1148
FEMQ4 14.9716 18.2115 25.5405 37.3135 39.7960 41.5363
k Method Mode

1 2 3 4 5 6
Table 14
Normalized frequencies of a fully simply supported three-ply laminated square 3 Present 2.8474 6.6759 9.4576 14.3940 19.0715 25.4221
plate subjected to in-plane stresses. FEMQ4 2.7522 6.6580 9.0463 14.3671 18.6995 25.3962
2 Present 3.7474 7.1945 9.9498 14.7868 19.3603 25.6715
k Method Mode FEMQ4 3.6746 7.1771 9.5536 14.7599 18.9944 25.6554
1 Present 4.4508 7.6840 10.4174 15.1687 19.6438 25.9174
1 2 3 4 5 6 FEMQ4 4.3889 7.6669 10.0340 15.1418 19.2839 25.9104
0 Present 5.0424 8.1489 10.8634 15.5404 19.9224 26.1604
3 Present 6.1559 8.6168 15.2522 24.8151 25.6873 26.0430 FEMQ4 4.9873 8.1320 10.4911 15.5136 19.5683 26.1616
FEMQ4 6.1663 8.6536 15.4439 24.9720 26.2078 26.3151 1 Present 5.5594 8.5925 11.2904 15.9027 20.1963 26.4007
2 Present 6.3162 8.9022 15.5766 24.9156 26.0161 26.1962 FEMQ4 5.5090 8.5757 10.9279 15.8758 19.8479 26.4094
FEMQ4 6.3266 8.9388 15.7670 25.0722 26.3606 26.6412 2 Present 6.0223 9.0173 11.7005 16.2561 20.4658 26.6385
1 Present 6.4726 9.1787 15.8943 25.0158 26.3408 26.3484 FEMQ4 5.9754 9.0006 11.3466 16.2293 20.1229 26.6541
FEMQ4 6.4829 9.2152 16.0837 25.1721 26.5125 26.9634 3 Present 6.4440 9.4253 12.0955 16.6013 20.7310 26.8740
 0.5 Present 6.5494 9.3139 16.0508 25.0657 26.4242 26.5017 FEMQ4 6.3998 9.4087 11.7493 16.5744 20.3935 26.8960
FEMQ4 6.5596 9.3504 16.2397 25.2219 26.5882 27.1231
0 Present 6.6253 9.4471 16.2058 25.1155 26.4998 26.6615
FEMQ4 6.6355 9.4836 16.3943 25.2716 26.6636 27.2818 Table 17
p-Ritz meth- 6.6252 9.4470 16.2051 25.1146 26.4982 26.6572 Normalized frequencies of a fully clamped three-ply laminated square plate with a
od [54] complicated cutout subjected to in-plane shear stresses.
0.5 Present 6.7003 9.5786 16.3594 25.1652 26.5751 26.8205
FEMQ4 6.7105 9.6150 16.5474 25.3212 26.7388 27.4396 k Method Mode
1 Present 6.7745 9.7082 16.5115 25.2148 26.6503 26.9785
FEMQ4 6.7847 9.7446 16.6990 25.3707 26.8138 27.5965 1 2 3 4 5 6
2 Present 6.9206 9.9624 16.8116 25.3138 26.8000 27.2917
FEMQ4 6.9307 9.9987 16.9984 25.4694 26.9632 27.9077 3 Present 17.8653 24.0595 29.4357 41.8394 43.0847 53.2447
3 Present 7.0636 10.2102 17.1064 25.4124 26.9488 27.6014 FEMQ4 17.7290 23.6073 29.3897 41.7650 42.1952 53.1221
FEMQ4 7.0736 10.2466 17.2925 25.5677 27.1118 28.2154 2 Present 17.8521 24.0563 29.4268 41.8394 43.0813 53.2491
FEMQ4 17.7136 23.6026 29.3811 41.7668 42.1935 53.1277
1 Present 17.8380 24.0527 29.4169 41.8388 43.0774 53.2532
FEMQ4 17.6970 23.5975 29.3716 41.7679 42.1913 53.1330
expected, the normalized frequency increases from FCFC, SCSC to
0 Present 17.8228 24.0487 29.4062 41.8375 43.0730 53.2569
CCCC. The first six shape modes of the clamped boundary condi- FEMQ4 17.6794 23.5919 29.3613 41.7685 42.1885 53.1380
tion are given in Fig. 5. 1 Present 17.8065 24.0441 29.3946 41.8355 43.0681 53.2603
FEMQ4 17.6607 23.5858 29.3501 41.7685 42.1851 53.1427
2 Present 17.7892 24.0392 29.3822 41.8329 43.0627 53.2633
4.1.3. Buckling analysis: a simply supported square plate with a FEMQ4 17.6409 23.5793 29.3380 41.7678 42.1813 53.1471
central square cutout 3 Present 17.7709 24.0337 29.3688 41.8297 43.0568 53.2659
The same simply supported square plate with a central square FEMQ4 17.6200 23.5723 29.3251 41.7665 42.1769 53.1511

cutout shown in Fig. 4 is considered. However, the geometry and


material properties are assumed as follows: a ¼1 m, c/a ¼0.5,
thickness h¼ 0.01 m, Young’s modulus E ¼7.0  1010 N/m2, Pois- points are used. Normalized critical buckling coefficients are pre-
son’s ratio ν ¼0.3. The uniaxial in-plane compressive load is ap- sented in Table 5. The results of the present method are close to
plied in the vertical direction. A normalized critical buckling those from the NURBS-based IGA using the classical plate theory
coefficient is defined by k = 12 (1 − v 2)(a/h)2Ncr /π 2E . Five different (CPT) and the published solutions. Once again, it is interesting to
meshes of 9  9, 16  16, 21  21, 28  28 and 33  33 control see that our IGA using a coarse mesh can offer acceptable results.
152 T. Yu et al. / Thin-Walled Structures 101 (2016) 141–156

Table 18 Table 20
Normalized frequencies of a fully simply supported three-ply laminated square Buckling loads of a CCCC plate with a complicated cutout for various orientations.
plate with a complicated cutout subjected to in-plane shear stresses.
Angle ply Method Mode
k Method Mode
1 2 3 4 5 6
1 2 3 4 5 6
(0°, 0°, 0°) Present 11.235 11.431 11.670 11.890 22.611 23.050
3 Present 5.1511 8.1299 10.8704 15.5718 19.8832 26.1843 IGA with 11.240 11.436 11.674 11.894 22.627 23.065
FEMQ4 5.1032 8.1161 10.5010 15.5453 19.5275 26.1866 CPT
2 Present 5.1184 8.1376 10.8684 15.5622 19.8970 26.1776 (15°,  15°, 15°) Present 11.048 11.179 11.786 11.946 21.208 21.519
FEMQ4 5.0682 8.1228 10.4979 15.5356 19.5418 26.1794 IGA with 11.053 11.184 11.791 11.951 21.223 21.534
1 Present 5.0822 8.1439 10.8660 15.5518 19.9100 26.1697 CPT
FEMQ4 5.0296 8.1280 10.4947 15.5250 19.5554 26.1711 (30°,  30°, 30°) Present 10.794 10.921 11.720 11.890 18.299 18.524
0 Present 5.0424 8.1490 10.8634 15.5404 19.9224 26.1604 IGA with 10.798 10.925 11.725 11.895 18.311 18.537
FEMQ4 4.9873 8.1320 10.4911 15.5136 19.5683 26.1616 CPT
1 Present 4.9987 8.1527 10.8604 15.5282 19.9341 26.1499 (45°,  45°, 45°) Present 10.509 10.605 11.546 11.693 15.638 15.808
FEMQ4 4.9410 8.1346 10.4873 15.5012 19.5805 26.1509 IGA with 10.513 10.610 11.551 11.698 15.649 15.818
2 Present 4.9511 8.1551 10.8571 15.5150 19.9450 26.1382 CPT
FEMQ4 4.8906 8.1360 10.4832 15.4880 19.5920 26.1392 (0°, 90°, 0°) Present 11.290 11.507 11.684 11.925 22.296 22.696
3 Present 4.8993 8.1562 10.8534 15.5009 19.9553 26.1252 IGA with 11.294 11.512 11.688 11.929 22.311 22.711
FEMQ4 4.8358 8.1360 10.4788 15.4739 19.6028 26.1263 CPT

Table 21
Buckling loads of a SCSC plate with a complicated cutout for various orientations.

Angle ply Method Mode

1 2 3 4 5 6

(0°, 0°, 0°) Present 2.860 2.925 7.595 7.784 13.822 14.353
IGA with CPT 2.861 2.925 7.597 7.786 13.828 14.359
(15°,  15°, 15°) Present 2.971 3.032 7.921 8.111 13.566 14.070
IGA with CPT 2.971 3.032 7.923 8.113 13.572 14.076
(30°,  30°, 30°) Present 3.233 3.279 8.752 8.912 12.985 13.354
IGA with CPT 3.234 3.279 8.755 8.914 12.991 13.360
(45°,  45°, 45°) Present 3.476 3.497 9.682 9.760 12.340 12.506
IGA with CPT 3.476 3.497 9.684 9.763 12.345 12.512
(0°, 90°, 0°) Present 2.902 2.960 7.757 7.926 13.868 14.358
IGA with CPT 2.902 2.960 7.759 7.928 13.874 14.365

Table 22
Comparison of critical buckling load for isotropic plate under in-plane shear
loading.

Theoretical [55] FEM [55] Present

8.442 8.400 8.423

The first eight buckling shape modes of this example are visua-
lized in Fig. 6.
Fig. 9. Geometry of square plate with a complicated cutout (unit:m).

4.2. Laminated composite plates

Table 19
4.2.1. Free vibration analysis: a laminated composite square plate
Buckling loads of a SSSS plate with a complicated cutout for various orientations.
with a complicated cutout
Angle ply Method Mode A three-layer symmetric laminated composite square plate
with a complicated cutout is considered. The geometry of the plate
1 2 3 4 5 6
and its dimensions is the same as Fig. 7. The material and geo-
(0°, 0°, 0°) Present 1.360 2.319 3.448 5.689 8.972 12.163 metrical parameters are considered as follows: ratio of elastic
IGA with CPT 1.360 2.319 3.448 5.689 8.974 12.166 constants E1/E2 ¼2.45 and G13 ¼G12 ¼ 0.48 E2, G23 ¼0.2 E2, Pois-
(15°,  15°, 15°) Present 1.479 2.402 3.706 5.962 8.883 12.030 son’s ratio ν12 ¼0.23, mass density ρ ¼8000 kg/m3 and thickness
IGA with CPT 1.479 2.402 3.706 5.963 8.886 12.034
h¼0.06 m. The frequency is normalized by ω˜ = [ρhω2a4 /D0.1]1/2,
(30°,  30°, 30°) Present 1.715 2.563 4.221 6.506 8.506 11.412
IGA with CPT 1.715 2.563 4.221 6.507 8.508 11.416 with a ¼10 m and D0.1 = E1h3/12 (1 − v12 v21). The normalized first
(45°,  45°, 45°) Present 1.827 2.603 4.494 6.765 7.788 10.335 six frequencies of the three-ply laminated plate with SSSS
IGA with CPT 1.827 2.603 4.495 6.766 7.790 10.338 boundary conditions for various fiber orientations are presented in
(0°, 90°, 0°) Present 1.359 2.317 3.474 5.710 8.957 12.127
Table 6. The present solutions are compared with the results ob-
IGA with CPT 1.359 2.317 3.474 5.710 8.959 12.131
tained by IGA and Kirchhoff theory with eight patches [37], EFG
T. Yu et al. / Thin-Walled Structures 101 (2016) 141–156 153

ABAQUS. The results of the present method are in good agreement


with those from the NURBS-based IGA using the classical plate
theory (CPT) and the FEM (ABAQUS).

4.2.2. Free vibration analysis: laminated composite plates subjected


to in-plane stresses
The vibration characteristics of laminated plates subjected to
in-plane stresses are numerically studied. To validate the accuracy
of the present approach, a fully clamped isotropic square plate
with its material parameters of the Young’s modulus
E¼ 2.0  1011 N/m2, Poisson’s ratio ν ¼0.3, density ρ ¼8000 kg/m2
is considered. In Table 12, the first six non-dimensional fre-
quencies computed by using the developed method are tabulated
and compared with those derived from the trapezoidal p-element
[46], the Rayleigh–Ritz method [45,53], and the finite element
method with 4-node element (FEMQ4) based on Mindlin theory.
Note that the reduced integration is used in the FEMQ4 to avoid
the locking effect problem. In this example, k = σx0 h/π 2D with D
being the flexural rigidity and σx0 = σy0 , τxy
0
is assumed to be zero. As
expected, good agreement between them can be observed.
Then, a three-ply laminated square plate (0°, 90°, 0°) subjected to
in-plane stresses k = σN h/π 2D is considered. The thickness ratios h/a
Fig. 10. A square plate with a complicated cutout subjected to in-plane shear ¼0.001, the in-plane stresses are σx0 = σy0 = σN (1 + 2z /h) and τxy
0
=0
loading (unit:m). are taken. The material properties for all layers of the laminates are
set to be identical, i.e., E1/E2 ¼40 and G13 ¼G12 ¼ 0.6 E2, G23 ¼0.5 E2,
and MKI methods [52]. It clearly observes that the present results Poisson’s ratio ν12 ¼ 0.25, mass density ρ ¼8000 kg/m3. The results of
are in good agreement with those published solutions for all the first six dimensionless frequencies ω˜ = ωa2/π 2 ρh/D0 with
considered orientations. Tables 7 and 8 tabulate the first ten nor- D0 = E2 h3/12 (1 − v12 v21) for fully clamped and simply supported
malized frequencies of CCCC and SCSC boundary conditions. The boundary conditions are listed in Tables 13 and 14, respectively. The
first four mode shapes of SSSS, SCSC and CCCC with fiber or- numerical results obtained by our developed method without in-
ientations (15°, 15°, 15°) are also presented in Table 9. plane stresses (k¼ 0) are in excellent agreement with those given by
Next, a four-layer square laminated plate with a complicated Liew et al. [54], and the numerical results for other cases of in-plane
cutout as shown in Fig. 8 is analyzed as the last numerical example stresses are well agreed with the FEMQ4 solutions. Compression and
of this subsection for free vibration analysis. The material prop- tension will decrease and increase the natural frequencies with in-
erties are set up to be as follows: ratio of elastic constants creasing the in-plane stresses, respectively.
E1/E2 ¼ 40 and G13 ¼0.5 E2, G13 ¼G12 ¼ 0.6 E2, Poisson’s ratio Next, a three-ply laminated square plate (0°, 90°, 0°) with a
ν12 ¼ 0.25, mass density ρ ¼8000 kg/m3 and thickness h¼0.01 m. complicated cutout as shown in Fig. 7 is analyzed as the last nu-
The first six normalized frequencies ω˜ = [ρhω2a4 /D0.1]1/2of various merical example. The same material parameters in the former ex-
fiber orientations are listed in Tables 10 and 11 for the two dif- ample are used here and the length a¼10 m and the thickness ratios
ferent boundary conditions, respectively. In Tables 10 and 11, the h/a ¼0.01 are used. The influence of the in-plane compression,
results of FEM are obtained with the commercial software tension ( σx0 = σy0 = σN (1 + 2z /h), τxy
0
= 0) and the shear loading
Table 23
Buckling loads of a SSSS plate with a complicated cutout for various orientations under shear loading.

Angle ply ξ Method Mode

1 2 3 4 5 6

(0°, 0°, 0°) þ1 Present 7.252 8.039 10.585 12.532 13.705 19.083
FEMQ4 7.342 8.119 10.628 12.742 13.949 19.490
1 Present 3.447 10.227 12.236 15.147 15.747 18.450
FEMQ4 3.451 10.192 12.433 15.361 15.881 19.067
(15°,  15°, 15°) þ1 Present 8.800 9.692 12.757 15.009 16.388 22.964
FEMQ4 8.902 9.777 12.794 15.196 16.664 23.354
1 Present 3.225 9.429 10.664 13.321 13.781 15.968
FEMQ4 3.232 9.431 10.775 13.520 13.848 16.617
(30°,  30°, 30°) þ1 Present 10.755 11.502 14.840 17.589 18.720 26.555
FEMQ4 10.863 11.596 14.871 17.762 19.005 26.937
1 Present 3.257 9.434 9.882 12.595 12.992 14.752
FEMQ4 3.268 9.458 9.930 12.782 13.016 15.499
(45°,  45°, 45°) þ1 Present 11.758 12.227 15.705 18.721 19.608 28.005
FEMQ4 11.859 12.330 15.737 18.893 19.887 28.386
1 Present 3.300 9.568 9.635 12.461 12.809 14.359
FEMQ4 3.314 9.595 9.668 12.667 12.786 15.165
(0°, 90°, 0°) þ1 Present 7.318 8.124 10.580 12.607 13.765 19.191
FEMQ4 7.407 8.208 10.624 12.831 14.009 19.614
1 Present 3.468 10.368 12.273 15.263 15.879 18.554
FEMQ4 3.474 10.345 12.458 15.488 16.006 19.225
154 T. Yu et al. / Thin-Walled Structures 101 (2016) 141–156

Table 24
Buckling loads of a CCCC plate with a complicated cutout for various orientations under shear loading.

Angle ply ξ Method Mode

1 2 3 4 5 6

(0°, 0°, 0°) þ1 Present 13.865 16.803 18.418 21.391 22.543 28.385
FEMQ4 14.113 17.057 18.941 22.159 22.727 29.234
1 Present 6.024 20.892 22.322 24.353 25.027 30.333
FEMQ4 6.055 21.812 22.312 24.601 26.692 31.427
(15°,  15°, 15°) þ1 Present 16.068 19.585 20.806 24.584 25.683 33.070
FEMQ4 16.348 19.847 21.298 25.326 25.804 33.799
1 Present 5.528 19.394 20.679 21.783 22.236 25.849
FEMQ4 5.555 20.206 20.791 22.302 23.471 26.882
(30°,  30°, 30°) þ1 Present 18.535 22.258 23.593 26.416 29.049 37.269
FEMQ4 18.817 22.484 24.062 27.053 29.072 38.095
1 Present 5.237 18.859 19.807 20.588 21.203 22.817
FEMQ4 5.272 19.525 20.359 21.313 21.932 23.730
(45°,  45°, 45°) þ1 Present 19.741 23.384 25.795 25.873 30.633 39.295
FEMQ4 20.008 23.584 26.255 26.447 30.597 40.324
1 Present 5.145 18.863 19.355 20.570 20.695 21.728
FEMQ4 5.187 19.368 20.306 21.233 21.459 22.446
(0°, 90°, 0°) þ1 Present 14.092 16.923 18.685 21.415 22.882 28.579
FEMQ4 14.347 17.177 19.244 22.189 23.070 29.502
1 Present 6.061 21.365 22.867 24.671 25.244 30.273
FEMQ4 6.096 22.314 22.905 24.990 26.902 31.375

( σx0 = σy0 = 0, τxy


0
= σN (1 + 2z /h)) on the non-dimensional frequencies 4.2.4. Buckling analysis: laminated composite plates under in-plane
ω˜ = ωa2/π 2 ρh/D0 characters are explored. The numerical results shear loading
computed by the present method and the FEMQ4 are reported in The first example of buckling analysis for in-plane shear loading
Tables 15–18. It can be observed in Tables 15 and 16 that in both deals with an isotropic square plate to further validate the applic-
clamped and simply supported boundary conditions, compression ability of the proposed method. The geometry and material proper-
ties used for the computations are assumed as follows: with a¼2 m,
and tension will decrease and increase the natural frequencies with
thickness h¼0.02 m, Young’s modulus E¼2.0  1011 N/m2, Poisson’s
increasing the in-plane stresses, respectively. The in-plane shear
ratio ν ¼0.3. The critical buckling coefficient k = Ncr a2/Eh3 results are
stresses case presented in Tables 17 and 18, it is found that the in-
compared to theoretical and numerical values [55], which are shown
fluences of the shear stresses on the frequencies are very small.
in Table 22 yielding a very good agreement.
The second example takes a square laminated plate with a
4.2.3. Buckling analysis: laminated composite plates under uniaxial
complicated cutout subjected to in-plane shear loading shown in
loading
A three-layer symmetric laminated composite plate with a com- Fig.10. The buckling load coefficient is defined by k = Ncr a2/π 2D0.1,
plicated cutout is considered for buckling analysis. The geometry of with D0.1 = E1h3/12 (1 − v12 v21). The obtained results are compared
any layer plate is the same as depicted in Fig. 9. A uniaxial in-plane with the FEM solutions (4-node element with 880 nodes and 800
compressive load is applied in the horizontal direction. The material elements) based on Mindlin theory integrated the reduced in-
tegration. The effects of boundary conditions and fiber orienta-
and geometrical parameters are considered as follows: ratio of elastic
tions and directions of in-plane shear loading on the buckling load
constants E1/E2 ¼2.45 and G13 ¼G12 ¼ 0.48 E2, G23 ¼ 0.2 E2, Poisson’s
coefficient are investigated and their results are then reported in
ratio ν12 ¼0.23 and thickness h¼0.06 m. The buckling load coeffi-
Tables 23 and 24.
cient is defined by k = Ncr a2/π 2D0.1, with D0.1 = E1h3/12 (1 − v12 v21).
Different from the uniaxial compressive case mentioned above,
The effects of different boundary conditions and fiber orientations on
we observe in the numerical values that when the shear loading is
the buckling load coefficient are investigated and their results are
positive, the maximum buckling load coefficient takes place at
then reported in Tables 19–21, respectively. θ ¼45° for all the specified boundary conditions. When the shear
For the SSSS boundary condition, we observe in the numerical loading changes from positive to negative, the critical buckling
values that maximum buckling load coefficient of the first, second, loads decrease with the angle ply (15°,  15°, 15°), (30°,  30°, 30°)
third and fourth modes occur at θ ¼45°; the fifth, sixth, seventh and and (45°,  45°, 45°). But, when the angle ply (0°, 0°, 0°) and (0°,
eighth modes occur at θ ¼ 0°; the ninth and tenth modes occur at 90°, 0°), the first and sixth critical buckling loads decrease once the
θ ¼ 30°. Similar to the CCCC boundary condition, however, maximum shear loading varying from positive to negative, while the second,
buckling load coefficient of the first and second, modes occur at (0°, third and fourth critical buckling loads increase. Similar to the
90°, 0°); the third and fourth modes occur at θ ¼15°; the others uniaxial compressive, however, the critical buckling loads of the
modes occur at θ ¼0°. For the SCSC boundary condition, maximum laminates with clamped boundaries are generally higher than
buckling load coefficient of the first, second, third and fourth modes those with simply supported boundaries.
occur at θ ¼ 45°; the fifth and sixth modes occur at (0°, 90°, 0°); the
seventh and eighth modes occur at θ ¼0°, the ninth and tenth modes
occur at θ ¼30°. It is important to note that the buckling loads of the 5. Conclusions
laminates with clamped boundaries are generally higher than those
with simply supported boundaries. All of the results match very well We present in this work a development of an effective NURBS-
with those derived from the NURBS-based IGA using the classical based isogeometric analysis using a simple first-order shear de-
plate theory. formation theory (S-FSDT) and the LSM for free vibration and
T. Yu et al. / Thin-Walled Structures 101 (2016) 141–156 155

buckling analysis of laminated composite structures with compli- using spline finite strip method, Thin-Walled Struct. 97 (2015) 207–214.
cated cutouts. Owing to the higher-order continuity of the NURBS [13] S. Natarajan, P.S. Deogekar, G. Manickam, S. Belouettar, Hygrothermal effects
on the free vibration and buckling of laminated composites with cutouts,
basis functions, the C1 consistency requirement of S-FSDT is Compos. Struct. 108 (2014) 848–855.
straightforward to construct. The LSM is used to describe the [14] Y. Stavsky, Bending and stretching of laminated aeolotropic plates, J. Eng.
cutouts and the numerical integration is performed only inside the Mech. (ASCE) 87 (1961) 31–56.
[15] P.C. Yang, C.H. Norris, Y. Stavsky, Elastic wave propagation in heterogeneous
physical domain, thus the trimmed NURBS surface is no longer plates, Int. J. Solids Struct. 2 (4) (1966) 665–684.
required to describe the geometrical structure with cutouts. The [16] J.M. Whitney, A.W. Leissa, Analysis of heterogeneous anisotropic plates, J.
shear-locking effect is also removed due to the inherent char- Appl. Mech. 36 (2) (1969) 261–266.
[17] P. Malekzadeh, A.R. Zarei, Free vibration of quadrilateral laminated plates with
acteristics of the S-FSDT. Numerical results for free vibration and carbon nanotube reinforced composite layers, Thin-Walled Struct. 82 (2014)
buckling of homogeneous isotropic and complicated laminated 221–232.
composites plates with cutouts are presented and investigated in [18] K.M. Liew, Y.Q. Huang, J.N. Reddy, Vibration analysis of symmetrically lami-
nated plates based on FSDT using the moving least squares differential
detail. Different complicated geometries and boundary conditions
quadrature method, Comput. Methods Appl. Mech. Eng. 192 (19) (2003)
for the laminated composites plates are examined. Compared with 2203–2222.
reference solutions, it can be concluded that the accuracy of the [19] K.M. Liew, J.B. Han, Z.M. Xiao, Differential quadrature method for thick sym-
metric cross-ply laminates with first-order shear flexibility, Int. J. Solids Struct.
developed method is high and its efficiency is obvious as many
33 (18) (1996) 2647–2658.
aspects have been treated as already mentioned above. Obviously, [20] J.N. Reddy, N.D. Phan, Stability and vibration of isotropic, orthotropic and la-
the main objective of the present work devoted for treating cut- minated plates according to a higher order shear deformation theory, J. Sound.
outs in laminated plates has been reached as it reveals that the Vib. 98 (2) (1985) 157–170.
[21] A.A. Khdeir, L. Librescu, Analysis of symmetric cross-ply elastic plates using a
proposed isogeometric analysis approach with S-FSDT and LSM is higher-order theory: Part II–Buckling and free vibration, Compos. Struct. 9 (4)
effective and accurate in modeling dynamic and buckling analysis (1988) 259–277.
of laminates with cutouts. [22] A. Chakrabarti, A.H. Sheikh, Buckling of laminated composite plates by a new
element based on higher order shear deformation theory, Mech. Compos.
Mater. St. 10 (4) (2003) 303–317.
[23] H.T. Thai, D.H. Choi, A simple first-order shear deformation theory for the
Acknowledgments bending and free vibration analysis of functionally graded plates, Compos.
Struct. 101 (2013) 332–340.
[24] H.T. Thai, D.H. Choi, A simple first-order shear deformation theory for lami-
This work was supported by the National Sci-Tech Support Plan nated composite plates, Compos. Struct. 106 (2013) 754–763.
of China (Grant No. 2015BAB07B10), the National Natural Science [25] T.J.R. Hughes, J.A. Cottrell, Y. Bazilevs, Isogeometric analysis: CAD, finite ele-
ments, NURBS, exact geometry and mesh refinement, Comput. Methods Appl.
Foundation of China (Grant No. 51179063), the scholarship from Mech. Eng. 194 (39-41) (2005) 4135–4195.
China Scholarship Council (CSC), Jiangsu Province Graduate Stu- [26] M. Lezgy-Nazargah, An isogeometric approach for the analysis of composite
dents Research and Innovation Plan (Grant No. CXZZ13_0235), and steel–concrete beams, Thin-Walled Struct. 84 (2014) 406–415.
[27] N. Valizadeh, S. Natarajan, O.A. Gonzalez-Estrada, T. Rabczuk, T.Q. Bui, S.P.
the Grant-in-Aid for Scientific Research (No. 26-04055)-Japan So- A. Bordas, NURBS-based finite element analysis of functionally graded plates:
ciety for the Promotion of Sciences (JSPS, ID No. P14055). The fi- static bending, vibration, buckling and flutter, Compos. Struct. 99 (2013)
nancial supports are gratefully acknowledged. 309–326.
[28] N. Valizadeh, T.Q. Bui, T.V. Vu, H.T. Thai, N.M. Nguyen, Isogeometric simulation
for buckling, free and forced vibration of orthotropic plates, Inter J. Appl.
Mech. 5 (2) (2013) 1350017.
References [29] S.H. Yin, T.T. Yu, T.Q. Bui, M.N. Nguyen, Geometrically nonlinear analysis of
functionally graded plates using isogeometric analysis, Eng. Comput. 32 (2015)
519–558.
[1] G.B. Chai, Free vibration of laminated composite plates with a central circular [30] S.H. Yin, J.S. Hale, T.T. Yu, T.Q. Bui, S.P.A. Bordas, Isogeometric locking-free plate
hole, Compos. Struct. 35 (4) (1996) 357–368. element: A simple first order shear deformation theory for functionally graded
[2] K. Sivakumar, N.G.R. Iyengar, K. Deb, Free vibration of laminated composite plates, Compos. Struct. 118 (2014) 121–138.
plates with cutout, J. Sound. Vib. 221 (3) (1999) 443–470. [31] Y. Bazilevs, V.M. Calo, J.A. Cottrell, T.J.R. Hughes, A. Reali, G. Scovazzi, Varia-
[3] M. Aydin Komur, Faruk Sen, Akın Ataş, Nurettin Arslan, Buckling analysis of tional multiscale residual-based turbulence modeling for large eddy simula-
laminated composite plates with an elliptical/circular cutout using FEM, Adv. tion of incompressible flows, Comput. Methods Appl. Mech. Eng. 197 (1–4)
Eng. Softw. 41 (2) (2010) 161–164. (2007) 173–201.
[4] D. Kumar, S.B. Singh, Effects of boundary conditions on buckling and post- [32] Y. Bazilevs, V.M. Calo, T.J.R. Hughes, Y. Zhang, Isogeometric fluid–structure
buckling responses of composite laminate with various shaped cutouts, interaction: theory, algorithms, and computations, Comput. Mech. 43 (1)
Compos. Struct. 92 (3) (2010) 769–779. (2008) 3–37.
[5] L.R. Kumar, P.K. Datta, D.L. Prabhakara, Tension buckling and parametric in- [33] T.Q. Bui, Extended isogeometric dynamic and static fracture analysis for cracks
stability characteristics of doubly curved panels with circular cutout subjected in piezoelectric materials using NURBS, Comput. Methods Appl. Mech. Eng.
to nonuniform tensile edge loading, Thin-Walled Struct. 42 (7) (2004) 295 (2015) 470–509.
947–962. [34] 200 J. Lu, Isogeometric contact analysis: geometric basis and formulation for
[6] C.C. Chen, S. Kitiporanchai, C.W. Lim, K.M. Liew, Free vibration of symme- frictionless contact, Comput. Methods Appl. Mech. Eng. 200 (5–8) (2011)
trically laminated thick perforated plates, J. Sound. Vib. 230 (1) (2000) 726–741.
111–132. [35] M.N. Nguyen, T.Q. Bui, T.T. Yu, S. Hirose, Isogeometric analysis for unsaturated
[7] G.R. Liu, X. Zhao, K.Y. Dai, Z.H. Zhong, G.Y. Li, X. Han, Static and free vibration flow problems, Comput. Geotech. 62 (2014) 257–267.
analysis of laminated composite plates using the conforming radial point in- [36] W.A. Wall, M.A. Frenzel, C. Cyron, Isogeometric structural shape optimization,
terpolation method, Compos. Sci. Technol. 68 (2) (2008) 354–366. Comput. Methods Appl. Mech. Eng. 197 (33–40) (2008) 2976–2988.
[8] A.J.M. Ferreira, C.M.C. Roque, A.M.A. Neves, R.M.N. Jorge, C.M.M. Soares, J. [37] S. Shojaee, N. Valizadeh, E. Izadpanah, T.Q. Bui, T.V. Vu, Free vibration and
N. Reddy, Buckling analysis of isotropic and laminated plates by radial basis buckling analysis of laminated composite plates using the NURBS-based iso-
functions according to a higher-order shear deformation theory, Thin-Walled geometric finite element method, Compos. Struct. 94 (5) (2012) 1677–1693.
Struct. 49 (7) (2011) 804–811. [38] S. Shojaee, E. Izadpanah, N. Valizadeh, J. Kiendl, Free vibration analysis of thin
[9] G. Eccher, K.J.R. Rasmussen, R. Zandonini, Elastic buckling analysis of perfo- plates by using a NURBS-based isogeometric approach, Finite Elem. Anal. Des.
rated thin-walled structures by the isoparametric spline finite strip method, 61 (2012) 23–34.
Thin-Walled Struct. 46 (2) (2008) 165–191. [39] R. Schmidt, R. Wüchner, K.U. Bletzinger, Isogeometric analysis of trimmed
[10] H.R. Ovesy, J. Fazilati, Buckling and free vibration finite strip analysis of NURBS geometries, Comput. Methods Appl. Mech. Eng. 241–244 (2012)
composite plates with cutout based on two different modeling approaches, 93–111.
Compos. Struct. 94 (3) (2012) 1250–1258. [40] T.Q. Bui, Ch. Zhang, Analysis of generalized dynamic intensity factors of
[11] Y.S. Zhang, S. Wang, J. Loughlan, Free vibration analysis of rectangular com- cracked magnetoelectroelastic solids by X-FEM, Finite Elem. Anal. Des. 69
posite laminates using a layerwise cubic B-spline finite strip method, Thin- (2013) 19–36.
Walled Struct. 44 (6) (2006) 601–622. [41] E. Rank, M. Ruess, S. Kollmannsberger, D. Schillinger, A. Düster, Geometric
[12] H. Assaee, H. Hasani, Forced vibration analysis of composite cylindrical shells modeling, isogeometric analysis and the finite cell method, Comput. Methods
156 T. Yu et al. / Thin-Walled Structures 101 (2016) 141–156

Appl. Mech. Eng. 249–252 (2012) 104–115. variously-shaped holes, J. Sound. Vib. 226 (4) (1999) 769–786.
[42] D. Schillinger, L. Dede, M.A. Scott, J.A. Evans, M.J. Borden, E. Rank, T.J. [49] R. Ali, S.J. Atwal, Prediction of natural frequencies of vibration of rectangular
R. Hughes, An isogeometric design-through-analysis method based on adap- plates with rectangular cutout, Comput. Struct. 12 (1980) 819–823.
tive hierarchical refinement of NURBS, immersed boundary methods, and [50] Y.K. Cheung, J. Kong, Linear elastic stability analysis of shear-deformable plates
T-spline CAD surfaces, Comput. Methods Appl. Mech. Eng. 249 (S1) (2012) using a modified spline finite strip method, Compos. Struct. 41 (2) (1993)
S116–S150. 189–192.
[43] T.T. Yu, S.H. Yin, T.Q. Bui, S. Hirose, A simple FSDT-based isogeometric analysis [51] L.G. Tham, A.H.C. Chan, Y.K. Cheung, Free vibration and buckling analysis of
for geometrically nonlinear analysis of functionally graded plates, Fin. Elem. plates by the negative stiffness method, Compos. Struct. 22 (4) (1986)
Anal. Des. 96 (2015) 1–10. 687–692.
[44] J.N. Reddy, Mechanics of Laminated Composite Plates and Shells Theory and [52] T.Q. Bui, N.M. Nguyen, Ch Zhang, An efficient meshfree method for vibration
Analysis, 2nd ed, CRC Press, New York, 2004. analysis of laminated composite plates, Comput. Mech. 48 (2) (2011) 175–193.
[45] O.L. Roufaeil, D.J. Dawe, Raleigh–Ritz vibration analysis of rectangular Mindlin [53] K.M. Liew, Y. Xiang, S. Kitipornchai, Transverse vibration of thick rectangular
plates subjected to membrane stresses, J. Sound. Vib. 85 (2) (1982) 263–275. plates––IV. Influence of isotropic in-plane stresses, Comput. Struct. 49 (1)
[46] A.Y.T. Leung, C. Xiao, B. Zhu, S. Yuan, Free vibration of laminated composite (1993) 69–78.
plates subjected to in-plane stresses using trapezoidal p-element, Compos. [54] K.M. Liew, Solving the vibration of thick symmetric laminates by Reissner/
Struct. 68 (2) (2005) 167–175. Mindlin plate theory and the p-Ritz method, J. Sound. Vib. 198 (3) (1996)
[47] S. Osher, J. Sethian, Fronts propagating with curvature-dependent speed: al- 343–360.
gorithms based on Hamilton–Jacobian formulations, J. Comput. Phys. 79 (1) [55] H.A. Qablan, H. Katkhuda, H. Dwairi, Assessment of the buckling behavior of
(1988) 12–49. square composite plates with circular cutout subjected to in-plane shear,
[48] M. Huang, T. Sakiyama, Free vibration analysis of rectangular plates with Jordan J. Civ. Eng. 3 (2) (2009) 184–195.

You might also like