Escolar Documentos
Profissional Documentos
Cultura Documentos
www.elsevier.com/locate/engstruct
Department of Mechanical Engineering, School of Engineering, Shiraz University, Shiraz 71345, Iran
Department of Mechanical Engineering and Applied Mechanics, North Dakota State University, Dolve 201, PO box 5285,
Fargo, ND 58105, USA
Received 7 January 2003; received in revised form 25 August 2003; accepted 18 September 2003
Abstract
A three-dimensional elasticity based layer-wise nite element method (FEM) is employed to study the static, free vibration and
buckling responses of general laminated thick composite plates. Various mixed boundary conditions and free edge conditions are
conveniently and accurately implemented. The methodology is employed in modeling the laminated composite plates resting on
Winkler and Pasternak types of elastic foundations. Elastic line and point supports are also successfully incorporated for thick
plates. Employing this methodology, the eects of shear deformation may easily be studied for moderately thick and thick plates.
A computer code, named LW3D, has been developed. Several examples are presented to verify the accuracy of the algorithm and
the coding.
# 2003 Elsevier Ltd. All rights reserved.
Keywords: Composite laminates; 3-D Elasticity; Layer-wise; Finite elements; Buckling; Vibration; Elastic foundation
1. Introduction
Based on dierent assumptions for displacement
elds, dierent theories for plate analysis have been
devised. These theories can be divided into three major
categories, the individual layer theories, the equivalent
single-layer (ESL) theories, and the three-dimensional
elasticity solution procedures. These categories are further divided into sub-theories by the introduction of
dierent assumptions. For example, the second category includes the classical laminated plate theory
(CLPT), the rst-order and higher-order shear deformation theories (FSDT and HSDT).
In the individual layer laminate theories, each layer
is considered as a separate plate. Since the displacement elds and equilibrium equations are written for
each layer, adjacent layers must be matched at each
Corresponding author. Tel.: +1-701-231-5859; fax: +1-701-2318913.
E-mail address: g.karami@ndsu.nodak.edu (G. Karami).
0141-0296/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2003.09.009
212
2. Numerical formulation
As shown in Fig. 1, consider a at rectangular laminate resting on an elastic foundation with ber orientation angle of h with respect to x-axis. The plate has a
length a, width b, constant thickness h and density q. It
is assumed subject to uniformly compressive dis y bN0 and
x aN0 , N
tributed in-plane forces of N
N xy cN0 . For buckling analysis, these forces stand
for negative values of in-plane force resultants Nxx, Nyy
and Nxy, respectively. In Fig. 1, ks is the Winkler foundation stiness and Gb is the shear modulus of the subgrade which is distributed uniformly under the plate.
2.1. Three-dimensional layer-wise plate formulation
The internal virtual work in matrix notation is as
follows,
h=2 a b
Wi
fdegT frgdV
1
h=2 0 0
213
u1 x;y;z;t
2
u2 x;y;z;t
h=2 0 0
D66
4
u3 x;y;z;t
p
n P
P
uij xij ;yij ;t wi x;yUj z
j1i1
p
n P
P
j1i1
p
n P
P
j1i1
vij xij ;yij ;t wi x;yUj z
wij xij ;yij ;t wi x;yUj z
where wi(x,y) and Uj(z) are the 2-D and 1-D Lagrangian shape functions and uij, vij and wij are the displacement components of the ith node corresponding to the
jth plane in the x, y and z directions, respectively. p is
the number of nodes in the xy plane, and n is the
number of nodes along the normal direction to the xy
plane.
Each ply is discretized into several two-dimensional
elements in the xy plane. Also, depending on the
accuracy required each ply is divided into one or several layers along the thickness direction which are
named as numerical plies. Reddy and Barbero [1] used
global shape functions along the z-direction, while in
the present work, all shape functions are dened in
local coordinates to make the procedure more systematic. A brief derivation of the three-dimensional elasticity layer-wise descretization of Eq. (2) using the
displacement eld Eq. (9) is provided in Appendix A.
214
h=2 0 0
ab
qx;ydu3 dxdy
12
0 0
ks du3 u3 Gb du3; x u3; x
0 0
du3; y u3; y dA
We
13
In the presence of elastic line (either in x or y directions) and point supports, Eq. (13) is replaced by
Eqs. (14) and (15).
8 b
>
>
ks du3 u3 Gb du3; x u3; x
>
>
>
0
>
>
>
< du3; y
u3; y
dy
xcte
14
We a
>
>
>
ks du3 u3 Gb du3; x u3; x
>
>
> 0
>
>
:
du3; y u3; y ycte dx
We ks du3 u3 Gb du3; x u3; x
15
du3; y u3; y xcte; ycte
where ks is the Winkler foundation stiness and Gb is
the shear modulus of the sub-grade. If ks and Gb were
given values other than 0, the foundation model is
known as Pasternak. For Winkler-type foundation
model, Gb is 0.
g ki
ke
fDg
qm
fD
N0
P
fDg fFg
h
16
fDg f0g:
qm
fD
17
The natural frequencies are found by linking the computer code to the MATLAB software [14].
For buckling analyses, where the virtual work due to
in-plane loads has to be taken into account, the nal
system of discretized equations becomes
N0
P
fDg f0g
18
h
An important point is the coupling between the inplane displacements (uij, vij) and the displacement in
the z direction (wij) in which we are capable of nding
all modes in the three directions. After the assemblage,
Eq. (18) has to be uncoupled by rewriting it in terms of
(wij), followed by the application of the essential
boundary conditions.
ki
ke
fDg
3. Numerical results
Several examples are solved to demonstrate the accuracy and eciency of the method. In the examples considered, both symmetric and antisymmetric cross-ply
thick rectangular laminates are considered and the following material properties are assumed,
Material type 1: E 1 =E 2 40, E 2 E 3 ; G 12 G13
0:6E 2 , G23 0:5E 2 ; m12 m13 0:25, m23 0.
Material type 2: E 1 25 106 psi, E 2 E 3 106 psi;
G 12 G 13 0:5 106 psi, G 23 0:2 106 psi; m12 m13
m23 0:25.
Layers are assumed to have identical thickness and
material properties. The plates were modeled using 9node quadratic elements on the xy plane and quadratic 3-node elements along the thickness (z-direction).
For mesh designation, a 2 3 3 mesh means two
divisions in x, three divisions in y direction and three
10
20
8.17
7.9194
3.07
8.1408
0.36
5.0747
4.0710
2.82
2.7979
0.78
2.8191
0.03
2.2039
1.8004
0.919
0.9181
0.10
0.9188
0.02
0.8412
0.7704
0.610
0.6094
0.10
0.6095
0.08
0.5744
0.5404
215
Both symmetric and antisymmetric cross-ply composite laminates, which are subjected to a sinusoidal
transverse load (qx;y q0 sinpx=asinpy=b) distributed on the upper surface of plate, are considered.
Table 1 presents nondimensionalized center mid-plane
a=2; b=2; 0 for a rectangular simply supdeection w
ported (0/90/0) laminated plate. Due to symmetry of
the problem, only one quadrant of the plate is modeled
using two types of meshing. The essential boundary
conditions for this example are set to,
u1 a=2;y;z u1 x;b;z 0; u2 x;b=2;z u2 a;y;z 0;
u3 a;y;z u3 x;b;z 0
The materials of the laminates are of type 2. This
particular problem has an exact 3-D elasticity solution
([2]) employed here to verify the solutions. The results
show good agreement even for a thick composite plate
(a=h 2), with a coarse mesh and with a highly anisotropic nature. The eects of elastic foundation on center mid-plane deection are also investigated. As the
Pasternak model takes into account shear eects of the
sub-grade, the predicted deection is less than what the
Winkler model forecasts. As one moves towards thinner plates, the deections become larger.
Table 2 contains dimensionless center mid-plane
a=2; b=2; 0 of a square (0/90) laminated
deection w
plate with dierent boundary conditions. The numerical values are compared with rst-order and higherorder shear deformation plate theories obtained by an
analytical Levy-type solution [13]. The accuracy of the
present method is clear, keeping in mind that in this
approach we are not limited to modeling special combinations of boundary conditions as in Levy-type
method. For SFSF, SSSS and SCSC boundary con-
Table 2
of square antisymmetric two cross-ply laminates with mixed boundary conditions (material 1)
Nondimensional center deection w
a/h
Type of solution
SFSF
SFSS
SCSF
SSSS
SCSS
SCSC
10
HSDT [13]
FSDT [13]
LW3D (present)
LW3D (Winkler)a
LW3D (Pasternak)b
HSDT [13]
FSDT [13]
LW3D (present)
LW3D (Winkler)a
LW3D (Pasternak)b
1.992
2.028
1.9947
0.5809
0.3204
2.624
2.777
2.6441
0.6843
0.3903
1.658
1.687
1.6584
0.5678
0.2943
2.211
2.335
2.2172
0.6747
0.3638
1.184
1.223
1.1918
0.4846
0.2700
1.733
1.897
1.7748
0.6220
0.3471
1.216
1.237
1.2162
0.5525
0.2679
1.667
1.758
1.6671
0.6639
0.3374
0.848
0.883
0.8568
0.4677
0.2449
1.333
1.477
1.3710
0.6129
0.3225
0.617
0.656
0.6297
0.3964
0.2240
1.088
1.257
1.1487
0.5657
0.3079
a
b
216
E 1 =E 2 40
E 1 =E 2 30
E 1 =E 2 20
Noor [4]
HSDT [5]
FDST [5]
CLPT [5]
Present a
Present b
Present c
Present d
0.34250
0.36348
0.35333
0.42884
0.34960
0.34776
0.37542
0.42129
0.32705
0.34020
0.33284
0.39335
0.33254
0.33061
0.35948
0.40701
0.30698
0.31284
0.30824
0.35422
0.30866
0.30836
0.33481
0.38905
a
b
c
d
Mesh: 2 2 2.
Mesh: 3 3 2.
Winkler model: dk 100, dg 0, mesh: 2 2 2.
Pasternak model: dk 100, dg 10, mesh: 2 2 2.
Elastic
support
Mesh
10dk
10dg
X (dk 1)
CFFC
Point C
FFFC
Edge BC
2 2 2
3 3 2
4 4 2
5 5 2
6 6 2
2 2 2
3 3 2
4 4 2
5 5 2
6 6 2
0.5
0.5
0.5
10
10
10
0.05
0.05
0.05
1
1
1
2.7232
2.7187
2.7174
2.3911
2.3843
2.3829
3.1202
2.9527
2.8005
2.7911
2.7886
2.7546
2.7330
2.7220
2.7155
2.7143
217
Table 5
Fundamental frequency parameter X for square, thick (0/90) laminates with point supports at the corners and mixed boundary conditions (material 1)
Boundary
condition
Point
support
h/b 10dk
10dg
X (dk 1)
CFFC
Point C
SFFC
Point C
SFFS
Point C
FFFC
Point C
0.2
0.3
0.4
0.2
0.3
0.4
0.2
0.3
0.4
0.2
0.3
0.4
0.1
0.05
0.05
0.5
0.1
0.05
0.5
0.05
0.05
0.1
0.01
0.01
3.6419
3.1163
2.7174
3.2377
2.3806
2.0893
2.1315
1.3037
1.2405
2.3667
2.0534
1.8102
4.3331
3.3347
2.7886
3.6904
2.7491
2.2492
2.6726
1.8487
1.4832
2.5087
2.1513
1.8608
1
0.5
0.5
5
1
0.5
5
0.5
0.5
1
0.1
0.1
Line
support
SFFC
Edge BC
FFFC
FFFF
FFFF
h/b dk
0.2
0.3
0.4
Edge BC 0.2
0.3
0.4
Edge AB 0.2
0.3
0.4
Edges AB, 0.2
CD
0.3
0.4
5
1
1
5
1
1
0.5
0.1
0.1
5
5
1
dg
X (dk 1)
0.5
0.1
0.1
0.5
0.1
0.1
0.05
0.01
0.01
0.5
0.5
0.1
4.7242
2.9117
2.5735
4.3402
2.6469
2.3829
0.9394
0.4347
0.4133
3.8788
2.7219
1.3250
5.8295
3.8623
2.8773
5.5603
3.6606
2.7155
2.1106
1.7512
1.3818
5.3252
3.8498
2.6586
218
Table 7
Uniaxial, biaxial and shear buckling load factor K for a simply supported, symmetric three cross-ply laminate on elastic foundation
(a 1, b=h 10, material 1)
a/b dk
dg
0 0
100 0
100 10
0 0
100 0
100 10
b 0,
c 0
b 1,
c 0
b 1,
c 1
LW3D
22.2347
32.2356
49.2260
16.4247
32.3538
49.0397
22.3151
32.4472
50.7515
16.4340
32.4472
49.2666
10.2024
12.2288
22.2288
3.2868
9.5904
19.5904
9.9424
11.9233
21.8666
3.2694
9.3452
19.1402
8.8184
10.2648
17.2125
3.1317
8.2492
14.9427
including a plate with completely free edges was considered. Line supports were set individually and simultaneously at edge BC, AB, and CD to study dierent
types of boundary conditions. Two meshes (5 5 2
and 6 6 2) were used for the laminates with completely free edges. For other types of boundary conditions 4 4 2 and 5 5 2 meshes were used. The
ner mesh is related to rigid elastic line supports and
the coarser mesh is associated with Pasternak type line
supports. The results show the eect of thickness-toside length ratio on the natural frequencies. As this
ratio is increased, the frequency decreases.
4. Conclusions
Table 8
Comparison of dimensionless critical buckling loads K for square antisymmetric two cross-ply thick laminates with various boundary conditions
(a 0, b 1, material 1)
h/a
Type of solution
SFSF
SFSS
SFSC
SSSS
SSSC
SCSC
0.1
HSDT [5]
FSDT [5]
CLPT [5]
LW3D (present)
LW3D (Winkler)a
LW3D (Pasternak)b
LW3D (present)
LW3D (Winkler)a
LW3D (Pasternak)b
LW3D (present)
LW3D (Winkler)a
LW3D (Pasternak)b
LW3D (present)
LW3D (Winkler)a
LW3D (Pasternak)b
4.940
4.851
5.425
4.7662
14.8336
24.7203
3.4867
6.2206
9.9070
2.4279
2.9576
3.6722
1.5239
1.7272
1.8916
5.442
5.351
6.003
5.2781
15.3456
27.4407
3.8900
7.1285
10.6712
2.7430
3.3563
4.0222
1.7617
1.9444
2.0907
6.274
6.166
6.968
6.0917
16.1608
27.7657
4.4636
7.1423
10.6731
3.1121
3.3583
4.0225
1.7627
1.9448
2.0908
11.562
11.353
12.957
11.2560
19.2940
31.3206
8.0732
9.1339
12.2270
4.0104
4.2103
4.7505
2.2692
2.3999
2.5072
17.133
16.437
21.116
16.1480
20.4570
32.6087
8.8744
9.1648
12.2333
4.0171
4.2147
4.7515
2.2713
2.4007
2.5075
21.464
20.067
31.280
19.5762
22.0212
34.2170
8.9584
9.1968
12.2398
4.0240
4.2192
4.7525
2.2734
2.4014
2.5078
0.2
0.3
0.4
a
b
E 1 =E 2 40
E 1 =E 2 30
E 1 =E 2 20
Noor [3]
HSDT [5]
FDST [5]
CLPT [5]
Present a
Present b
Present c
Present d
10.817
11.563
11.353
12.957
11.2560
11.2382
19.3401
31.5981
9.3746
9.8695
9.7347
10.891
9.7156
9.6995
17.5249
29.7616
7.8196
8.1151
8.0423
8.8158
8.0696
8.0455
15.3245
27.5347
a
b
c
d
G r
6
6.
6 ..
4
Z0
T
r x;y
2 T
w
6 Z
T
6 0
Gz
6 .
4 ..
Mesh: 3 3 2.
Mesh: 4 4 2.
Winkler model: dk 100, dg 0, mesh: 3 3 2.
Pasternak model: dk 100, dg 10, mesh: 3 3 2.
219
Z0
T
Z 0
T
T
w;r
..
.
Z 0
Z0
T
7
7
Z0
T 7
7
7
..
7
.
5
T
w;r
nnp
A:4
Z0
w
T
..
.
Z0
T
Z 0
Z 0
T 7
7
7
..
5
.
w
T
A:5
2 T
U Gx
6 UT G
6h
y
6 T i
6 U Gz
6 ;z
6
6 G0
6
6
G
6 G0
6
6 G0
6
6
6 G0
6
6 G0
4
G0
Appendix A.
If H represents the partial derivatives of displacement eld with respect to the global coordinates and D
as the vector of displacement components as dened in
Section 2.3, then,
T
fHg u1; x u1; y u1; z u2; x u2; y u2; z u3; x u3; y u3; z
A:1
G 0
G 0
nnp
G0
G0
G 0
T
U G
T x
hU Gy i
UT;z Gz
G 0
G 0
G 0
7
7
7
7
G0
7
7
7
G0
7
7
G0
7
7
7
G0
T 7
7
U G
T x 7
7
7
hU Gy i 5
T
U;z Gz
A:6
feb g Hb
fHg and fes g Hs
fHg
2
A:7
3
1 0
0 0
60 0
6
Hb
6
40 0
0
0
0
0
1 0
0 0
0
0
0
0
07
7
7
15
0 1 0 1 0 0 0 0 0
0 0 0 0 0 1 0 1 0
Hs
0 0 1 0 0 0 1 0 0
A:8
T
fDg u11
upn v11
vpn w11
wpn
13np
A:2
Employing the displacement eld (9), and after the
evaluation of the partial derivatives with respect to global coordinates, the matrix H can be written as the
product of two matrices [G] and D
, i.e.
fHg91 G
93np D
3np1
A:9
A:3
fdeb gT fdDgT Bb
T
A:10
220
Kb
Bb
Cb
Bb
dV fdDg
T
Bb
Cb
Bb
dV
A:11
2 11
K
6 12 T
K
4 K
13 T
K
K 12
K 22
K
23 T
K 13
3
7
K 23 5
33
K
3np3np
A:12
References
[1] Reddy JN, Barbero EJ. A plate bending element based on a generalized laminate plate theory. Int J Numer Meth Eng
1989;28:227592.
[2] Pagano NJ. Exact solution for rectangular bi-directional composites and sandwich plates. J Composite Mater 1970;4:2034.
[3] Noor AK. Stability of multilayered composite plates. Fiber Sci
Tech 1975;8(2):819.
[4] Noor AK. Free vibration of multilayered composite plates.
AIAA J 1973;11:10389.
[5] Reddy JN, Khdeir AA. Buckling and vibration of laminated
composite plates using various plate theories. AIAA J
1989;12:180817.
[6] Gbadeyan JA. Lateral vibration analysis of an initially stressed
rectangular thin plate on Pasternak elastic foundation. Model,
Simul Control B: Mech Thermal Eng, Mat Resour Chem
1990;27(3):935.
[7] Takahashi K, Sonoda T. Dynamic stability of a rectangular
plate on Pasternak foundation subjected to sinusoidally timevarying in-plane load. Theor Appl Mech, Univ Tokyo Press
1992;41:5562.
[8] Xiang Y, Kitipornchai S, Liew KM. Buckling and vibration of
thick laminates on Pasternak foundation. J Eng Mech, ASCE
1996;122(1):5463.
[9] Wang CM, Kitipornchai S, Xiang Y. Relationships between
buckling loads of Kirchho, Mindlin, and Reddy polygonal
plates on Pasternak foundation. J Eng Mech, ASCE
1997;123(11):11347.
[10] Aiello MA, Ombres L. Buckling and vibration of unsymmetric
laminates resting on elastic foundations under in-plane and shear
forces. Composite Struct 1999;44:3141.
[11] Dogruoglu AN, Omurtag MH. Stability analysis of compositeplate foundation interaction by mixed FEM. J Eng Mech, ASCE
2000;126(9):92836.
[12] Huang MH, Thambiratnam DP. Analysis of plate resting on
elastic supports and elastic foundation by nite strip method.
Comput Struct 2001;79:254757.
[13] Reddy JN. Mechanics of laminated composite plates, theory and
analysis. 1997.
[14] Hunt B, Lipsman R, Rosenberg J. A guide to MATLAB. Cambridge University press; 2001.
[15] Mizusawa T, Kajita T. Vibration of skew plates resting on point
supports. J Sound Vibration 1987;115:24351.