CMM-2005 Computer Methods in Mechanics

June 21-24, 2005, Czestochowa, Poland

Large deection analysis of moderately thick composite plates

Hakan Tanrver and Erol Senocak Mechanical Engineering Department, Istanbul Technical University Gmsuyu, 34437 Istanbul, Turkey s e-mails: hakant@itu.edu.tr and senocak@itu.edu.tr

Abstract Large deection analysis of laminated composite plates is considered. The Galerkin method along with NewtonRaphson method is applied to large deection analysis of laminated composite plates with various edge conditions. First order shear deformation theory and von Krmn type nonlinearity are utilized and the governing differential equations are solved by choosing suitable polynomials as trial functions to approximate the plate displacement functions. The solutions are compared to that of Chebyshev polynomials and nite elements. A very close agreement has been observed with these approximating methods. In the solution process, analytical computation has been done wherever it is possible, and analytical-numerical type approach has been made for all problems. Keywords: Galerkin method, large deection, composite plates

1.

Introduction

Different numerical techniques were employed to investigate the geometrically nonlinear behavior of thick plates. Pica et al. solved the large deection problem of isotropic plates using nite element technique [1]. Reddy and Chao applied the nite element method (FEM) to large deection and large amplitude free vibration problems of laminates [2]. Turvey and Osman performed the large deection analysis of isotropic plates using Dynamic Relaxation technique [3]. Liu et al. solved the large deflection problem of elliptical plates using Galerkin method [4]. Shukla and Nath presented a Chebyshev polynomials (CP) solution for geometrically nonlinear problem of moderately thick laminates [5]. Liew et al. proposed a mesh-free kp-Ritz method for the large deection exural analysis of laminated composite plates [6]. First order shear deformation theory (FSDT) based on Mindlins hypothesis and von Krmn type geometric nonlinearity were utilized in the aforementioned works. Among the method of weighted residuals, the Galerkin method (GM) is a powerful numerical solution technique to differential equations. The Galerkin technique has found a research area for a particular case of boundary conditions and trial functions for the large deection analysis [4, 7, 8, 9]. This paper concerns the effect of shear deformations in the large deection analysis of composite plates. The FSDT based on Mindlins hypothesis is imposed using von Krmn type geometric nonlinearity. Governing nonlinear equations are solved by employing GM with Newton-Raphson technique. In the solution process, computations have been carried out analytically wherever it is possible and analytical-numerical type approach has been made for all cases. 2. Governing equations

formed plate (see Fig. 1). The plate is assumed to be subjected uniform transverse pressure qo , and it is constructed of nite homogenous orthotropic layers perfectly bonded together.
y (a/2,b/2)

b (0,0) x

qo

Figure 1: Plate geometry and loading. Under the assumptions of rst order shear deformation theory based on Mindlins hypothesis; let u, v, w denote the displacements at an arbitrary point of the plate in the x, y, z directions and u0 (x, y), v 0 (x, y), w 0 (x, y) are the displacements at a corresponding point of the midplane of the plate in the x, y and z directions respectively. Then the displacement eld of the rst order theory is of the form [10]: u(x, y, z, t) = u0 (x, y, t) + x (x, y, t)z, v(x, y, z, t) = v 0 (x, y, t) + y (x, y, t)z, w(x, y, z, t) = w 0 (x, y, t), (1)

where x and y are the rotations of a transverse normal about the y and x axes respectively. The corresponding total strains could be expressed as follows: x = 0 + x z, x y = 0 + y z, y 0 xy = xy + xy z, 0 0 xz = xz , yz = yz .

(2)

Consider a rectangular laminated plate with dimensions a, b and uniform thickness h. The origin of the coordinate system is chosen to coincide with the center of the midplane of the unde-

0.7 center deflection 0.6 0.5 0.4 0.3 0.2 0.1 0 20 40 60 GM CS 1 CP CS 1 GM CS 2 CP CS 2 80 100 120 140 160 180 200 load

Considering von Krmn type geometric nonlinearity [11], the strain displacement relations can be written as
2 0 = u0 + 1 w,x , ,x x 2 0 0 1 2 y = v,y + 2 w,y , 0 0 xy = u0 + v,x + w,x w,y , ,y 0 xz = w,x + x , 0 yz = w,y + y ,

(3)

where differentiations are denoted by comma. Midplane curvatures and twist of the plate are the following: x = x,x , y = y,y , xy = x,y + y,x . (4) For a plate with an arbitrary number of layers, the constitutive relations are

Figure 2: Load (qo a4 /E2 h4 ) versus central deection (w/h) of [0 /90 /90 /0 ] laminate.

N M
3 2.5 moment 2 1.5 1 0.5 0 20 40 60 GM CS 1 CP CS 1 GM CS 2 CP CS 2 80 100 120 140 160 180 200 load

A B

B D A44 A45

3.5

where N and M are the resultant forces and moments conjugate to 0 and respectively. Qx and Qy are transverse forces and the parameter K is shear correction factor [12]. Aij , Bij and Dij are symmetric matrices dened as follows:
h/2

(Aij , Bij , Dij ) =

(1, z, z 2 ) Qij dz,

h/2

(a)
5 4.5 4 3.5 moment 3 2.5 2 1.5 1 0.5 0 20 40 60 GM CS 1 CP CS 1 GM CS 2 CP CS 2 80 100 120 140 160 180 200 load

where Qij is the corresponding reduced stiffness coefcients. Five governing equations of motion for the plate can be written as follows in the general form [13]: R1 = Nx,x + Nxy,y I0 u0 I1 x,tt = 0, ,tt 0 R2 = Nxy,x + Ny,y I0 v,tt I1 y,tt = 0, R3 = Qx,x + Qy,y + (w,x Nx + w,y Nxy ),x +(w,x Nxy + w,y Ny ),y + qo I0 w,tt = 0, R4 = Mx,x + Mxy,y Qx I2 x,tt I1 u0 = 0, ,tt R5 = Mxy,x + My,y Qy I2 y,tt 0 I1 v,tt = 0, where I0 , I1 and I2 are mass moments of inertia dened as =

(I0 , I1 , I2 ) =

Figure 3: Load (qo a4 /E2 h4 ) versus Mx a2 /D11 h (a) and My a2 /D22 h (b) at the center of [0 /90 /90 /0 ] laminate.

2 center deflection

(i) being the material density of the ith layer. The equations of motion (8) in terms of displacements can be obtained by making use of constitutive (5, 6), strain-displacement (3) and curvaturedisplacement (4) relations (these equations are not given here; see [13] for details). For the Galerkin approach, the normalized displacements of the plate are approximated in the form shown below:
M N

1.5

u0 =

amn ( t )Umn (x, y),

m=0 n=0 M N

1 GM CRS GM ANG ANSYS CRS ANSYS ANG 0 1 2 3 4 5 6 load 7 8 9 10

v0 = w=

bmn ( t )Vmn (x, y),

m=0 n=0 M N

0.5

cmn ( t )Wmn (x, y),

m=0 n=0 N M

x =
m=0 n=0 M N

dmn ( t )Smn (x, y),

Figure 4: Load (102 qo a4 /E2 h4 ) versus central deection (w/h) of cross-ply [0 /90 ] (CRS) and angle-ply [45 / 45 ] (ANG) laminates. 2

y =
m=0 n=0

emn ( t )Tmn (x, y),

(b)

h/2 (1, z, z 2 ) dz h/2 zi+1 (1, z, z 2 ) (i) i zi

dz,

,
0 yz 0 xz

(5) , (6)

Qy Qx

=K

A45 A55

(7)

(8)

(9)

(10)

CC

Table 1: Boundary conditions and corresponding weight functions Nx = Nxy = w = x = y = 0 at x = a/2, Ny = Nxy = w = x = y = 0 at y = b/2. i = 1 (i = 1, 2), i = (x2 a2 /4)(y 2 b2 /4) (i = 3, . . . 5). u0 = v 0 = w = 0 at x = a/2 and y = b/2, x = 0 at x = a/2 and y = b/2, Mx = 0 at x = +a/2, My = 0 at y = b/2, y = 0 at x = a/2. i = (x2 a2 /4)(y 2 b2 /4) (i = 1, . . . 3), 4 = (x + a/2)(y 2 b2 /4), 5 = (x2 a2 /4). u0 = v 0 = w = x = 0 at x = a/2 and y = b/2, My = 0 at y = b/2, y = 0 at x = a/2. i = (x2 a2 /4)(y 2 b2 /4) (i = 1, . . . 4), 5 = (x2 a2 /4).

CS1

CS2

where amn , bmn , cmn , dmn and emn are unknown functions of time, Umn , Vmn , Wmn , Smn and Tmn are the trial functions, and M and N are the number of terms in x and y directions respectively. In general, M and N may take different values for each displacement function. Herein, polynomials are used as trial functions, which are chosen to satisfy the geometric boundary conditions, where as natural boundary conditions are not satised. In this case, simultaneous approximation is made to the solutions of differential equations and to the boundary conditions. Since this is a static analysis amn , bmn , cmn , dmn and emn are not time dependent functions (they are taken as unknown constants) and time derivative terms in the equations of motion are all cancelled. Substituting Eq. (10) into nonlinear plate equilibrium equations and the boundary terms yields the residuals in the domain of the plate and at the boundaries of the plate. Forcing these residuals to be orthogonal to each member of a set of trial functions yields the following Galerkin equations.

where i (i = 1, . . . 5) denote the weight functions. Substituting Eq. (10) into Eqs. (11), nonlinear equations in terms of unknown coefcients amn , bmn , cmn , dmn and emn are obtained. These equations are solved by employing the Newton-Raphson methodology. 4. Boundary Conditions Three different boundary conditions are considered and shown in Table 1. The Galerkin integrals in Eqs. (11) are in the general form and must be modied according to relevant boundary conditions. Note that whole plate models are analyzed in all cases presented here. 5. Results Large deection of a symmetric cross-ply laminate and unsymmetric angle-ply and cross-ply laminates are chosen as numerical examples for the application of the GM. Comparisons with results of the other solution techniques such as Chebyshev polynomials and nite elements are given. In the application of the GM the displacement (trial) functions are approximated by polynomials expressed in Eq. (12). The proper choice of the trial functions considering symmetry of the problem can reduce the computation time [9]. The systematic choice of the trial functions is explained in detail for each case. The convergence study of the proposed solution technique is carried out and it has been determined that taking M and N as ve is appropriate. Hence, for all the GM applications given here, M and N are taken as ve. 5.1. Symmetric Cross-Ply Laminate

+b/2 b/2

3.

Solution Procedure

and for CS2 type boundary condition

Umn = 1 x y , Vmn = 2 xm y n , Wmn = 3 xm y n , Smn = 4 xm y n , Tmn = 5 xm y n ,

m n

(12)

m = odd n = even m = even n = odd m = even n = even

Umn , Smn , Vmn , Tmn , Wmn . (14)

In the application of the Galerkin method the geometrical boundary conditions are satised by choosing appropriate trial functions. In the Galerkin method used here, the evaluations of integrals are symbolically computed by using a commercial computer math code M athematicaT M [14]. Trial functions are weighted polynomials given as follows:

+b/2 b/2

+b/2 b/2 +b/2 b/2

+b/2 b/2

+a/2 U R dxdy a/2 mn 1 +b/2 U N | dy b/2 mn x x=a/2 +a/2 Umn Nxy |y=b/2 dx = 0, a/2 +a/2 V R dxdy a/2 mn 2 +a/2 V N | dx a/2 mn y y=b/2 +b/2 Vmn Nxy |x=a/2 dy = 0, b/2 +a/2 Wmn R3 dxdy = 0, a/2 +a/2 S R dxdy a/2 mn 4 +b/2 S Mx |x=a/2 dy = 0, b/2 mn +a/2 T R dxdy a/2 mn 5 +a/2 T My |y=b/2 dx a/2 mn

(11)

= 0.

The large deection of a symmetric cross-ply [0 /90 /90 /0 ] laminate under various boundary conditions is analyzed. The powers m and n are taken as in the following manner for CS1 m = 0, 1, 2, . . . 5 Umn , Wmn , Smn , n = even (13) m = 0, 1, 2, . . . 5 Vmn , Tmn , n = odd

The material and geometry constants of the plate are taken from [5] and given as follows: E1 = 175.78 GP a, E2 = E1 /25, G12 /E2 = G13 /E2 = 0.5, G23 /E2 = 0.2, 12 = 0.25, a = b, a/h = 10, K = 5/6. Normalized center deection and center moment values of the plate under CS1 and CS2 boundary conditions for GM and CP (values are read from [5]) are given in Figs. 2 and 3. 5.2. Unsymmetric Laminates

[2] Reddy, J.N. and Chao, W.C., Large-deection and largeamplitude free vibrations of laminated composite-material plates, Comput. Struct. 13, pp. 341-7, 1981. [3] Turvey, G.J. and Osman, M.Y., Elastic large deection analysis of isotropic rectangular Mindlin plates, Int. J. Mech. Sci., 32 (4), pp. 315-28, 1990. [4] Liu, R.-H., Xu, J.-C. and Zhai, S.-Z., Large-deection bending of symmetrically laminated rectilinearly orthotropic elliptical plates including transverse shear, Archive of Applied Mechanics, 67, pp. 507520, 1997. [5] Shukla, K.K. and Nath, Y., Nonlinear analysis of moderately thick laminated rectangular plates, Journal of Engineering Mechanics, 126 (8), pp. 8318, 2000. [6] Liew, K.M., Wang, J., Tan, M.J. and Rajendran, S., Nonlinear analysis of laminated composite plates using the mesh-free kp-Ritz method based on FSDT, Computer Methods in Applied Mechanics and Engineerig, 193, pp. 4763-79, 2004. [7] Savithiri, S. and Varadan, T.K., Large deection analysis of laminated composite plates, Int. J. Non-Linear Mechanics, 28 (1), pp. 112, 1993. [8] Dennis, S.T., A Galerkin solution to geometrically nonlinear laminated shallow shell equations, Comput. Struct., 63 (5), pp. 85974, 1997. [9] Tanrver, H. and Senocak, E., Large deection analysis of unsymmetrically laminated composite plates: analyticalnumerical type approach, Int. J. Non-Linear Mechanics, 39 (8), pp. 138592, 2004. [10] Mindlin, R.D., Inuence of rotatory inertia and shear on exural motions of isotropic, elastic plates, Journal of Applied Mechanics, 18 (March), pp. 318, 1951. [11] Brush, D.O. and Almroth, B.O., Buckling of Bars, Plates and Shells, McGraw-Hill, 1975. [12] Whitney, J.M., Structural Analysis of Laminated Anizotropic Plates, Technomic Publ. Co., 1987. [13] Reddy, J.N., Mechanics of Laminated Composite Plates, CRC Press, New York, 1997. [14] Wolfram, S., M athematicaT M : A System for Doing Mathematics by Computer, Addison-Wesley, Redwood City, CA, 1988. [15] Chia, C.Y. and Prabhakara, M.K., Large deection of unsymmetric cross-ply and angle-ply plates, Journal Mechanical Engineering Science, 18 (4), pp. 17983, 1976. [16] ANSYS Reference Manual, ANSYS Inc., 1999.

Unsymmetric angle-ply and cross-ply laminates under CC type boundary condition are considered for the large deection analysis. In the nonlinear analysis of the cross-ply laminate the powers m and n are taken as in the same manner given in Eq. (14). For the nonlinear analysis of angle-ply laminate the powers m and n are chosen in such a way that (m + n) is odd for the in-plane displacement functions and rotations and even for the transverse displacement function. Material and geometrical properties of the laminates are taken from [2] and given below. E2 = E1 /40, G12 /E2 = 0.6, G13 /E2 = 0.5, G23 = G13 , 12 = 0.25, a = b, a/h = 40, K = 5/6. Dimensionless center deection-load curves of these laminates are shown in Fig. 4. FEM [2] and perturbation technique [15] results (they are not shown here) for these laminates are found to be in good agreement with the present results. Results of the commercial FEM program ANSYS [16] can be seen in Fig. 4. Shell91 type elements including shear deformations with a 10 10 mesh are used in ANSYS program. 6. Conclusions

Geometrically nonlinear analysis of thick composite plates based on FSDT is performed by using Galerkin approach along with Newton-Raphson technique. The choice of trial functions is crucial to approximate the two dimensional displacement eld. The trial functions must be chosen in a way that essential boundary conditions are satised. The present solution methodology may be used to solve large deection analysis of the moderately thick laminates in an easy and effective way with the help of a symbolic math package. The method is found to determine closely the displacements with a few number of terms. The results are compared to that of known other approximating methods (Chebyshev polynomials and nite elements), and commercial FEM code ANSYS. A very good agreement is observed. The convergence of the Galerkin method is found to be quite fast. References [1] Pica, A., Wood, R.D. and Hinton, E., Finite element analysis of geometrically nonlinear plate behavior using a Midlin formulation, Comput. Struct., 11, pp. 203-15, 1980.