International Journal of Science & Technology

Volume 2, No 1, 33-40, 2007

Geometrically Non-linear Analysis of Axisymmetric Plates and
Shells
Cengiz POLAT and Zlf nar ULUCAN
Frat University, Faculty of Engineering, Department of Civil Engineering,
23279, Elazig, TURKIYE
cpolat@firat.edu.tr
(Received: 27.02.2007; Accepted: 19.03.2007)
Abstract: A geometrically nonlinear formulation is presented for the axisymmetric plate and shell structures.
The formulation is based on the total Lagrangian approach. The nonlinear equilibrium equations are solved using
the Newton-Raphson method. Different numerical examples are performed to obtain the geometrically non-
linear behaviour of axisymmetric plates and shells.
Keywords: Axisymmetric plate and shell, geometric nonlinearity, Newton-Raphson method.
Eksenel Simetrik Plak ve Kabuklarn Geometrik Bakmdan Lineer
Olmayan Analizi
zet: Plak ve kabuk yaplar iin geometrik bakmdan lineer olmayan bir formlasyon verilmektedir.
Formlasyonda toplam Lagrange yaklam esas alnmaktadr. Lineer olmayan denge denklemleri Newton-
Raphson metodu kullanlarak zlmektedir. Eksenel simetrik plak ve kabuklarn geometrik bakmdan lineer
olmayan davrann elde etmek iin deiik nmerik rnekler gerekletirilmitir.
Anahtar kelimeler: Eksenel simetrik plak ve kabuk, geometrik bakmdan lineer olmayan davran, Newton-Raphson
metodu.


1. Introduction

In the linear analysis, the displacements
and strains developed in the structure are small.
That is, the geometry of the structure assumed
remains unchanged during the loading process
and linear strain approximations can be used.
However, the geometry of the structure changes
continuously during the loading process, and this
fact is taken into account in the geometrically
nonlinear analysis. In the linear analysis, the load
carrying capacity of the structure cannot be
predicted correctly. Therefore, it is necessary to
use the non-linear equilibrium equations to
describe the structural behaviour [1,2].
In this paper, the geometrically nonlinear
behavior of axisymmetric plates and shells
subjected to different axisymmetric loads for
various shell parameters is studied. The total
Lagrangian approach, in which all static and
kinematic variables at the current state are
referred to the initial configuration, is adopted.
The nodal displacements are nonlinear functions
of nodal rotations in the axisymmetric shell




formulation, thus restrictions of small nodal
rotations during deformation process is removed.
In the numerical application, the isoparametric
axisymmetric shell elements are used. For such
elements the element geometry is described in
terms of the coordinates of middle surface nodes
and the mid-surface nodal point normals. The
incremental equations of equilibrium are solved
by using the Newton-Raphson method. The
nonlinear analysis of axisymmetric plates and
shells is performed by a computer program
which is written by the authors in MATLAB
software.

2. Formulation

2.1. Kinematics of Deformation

For the isoparametric axisymmetric shell
elements, mid-surface is defined by only two
coordinates and . Figure 1 shows the nodal
variables and the definitions of angles for a
C. Polat and Z. . Ulucan
34
typical node k in the element. The element
geometry is defined by


= =

+ =
n
1 k
n
1 k
k k k k k
cos t N
2
x N x (1)


= =

+ =
n
1 k
n
1 k
k k k k k
sin t N
2
y N y
(2)

where
k
t is the element thickness at the node k,
and n is the number of nodes in the element. For
the finite element model, displacement field is
approximated by using nodal displacements
k
u
and
k
v , nodal rotation
k
, and the shape
function
k
N at the node k as follows;

= =

+ =
n
1 k
n
1 k
kx k k k k
F t N
2
u N u (3)



= =

+ =
n
1 k
n
1 k
ky k k k k
F t N
2
v N v
(4)

and
k
F is defined by the following expression
[3]

)
`

+

=
)
`

=
k k k k
k k k k
ky
kx
k
sin cos ) 1 (cos sin
sin sin ) 1 (cos cos
F
F
F
(5)

) , ( v
k
v
) , ( p
) , ( u
k
u
k

k
k

x
y

x
y

Figure 1. Definitions of angles and local axes of axisymmetric shell element

2.2. Nonlinear Finite Element Formulation

In the total Lagrangian approach, second
Piola-Kirchhoff stresses and corresponding
Green-Lagrange strains are considered [4]. It is
assumed that shell thickness does not change
during deformation that is,
x
and
x
can be
neglected. For a linear elastic material, stresses
and strains in the local coordinate system ) z y x (
can be written as

= == = D (6)

[ ]
T
z y x y x
= (7)





[ ]
T
z y x y x


= (8)

where D is the elasticity matrix in the local
coordinate system and it can be obtained using
these two assumptions 0
x
= == =

and 0
x
= == =

.
Similarly in the global coordinate system ) xyz (
stresses and strains are defined as

= == = D (9)

[ [[ [ ] ]] ]
T
z xy y x
= == = (10)

[ [[ [ ] ]] ]
T
z xy y x
= == = (11)
Geometrically Non-linear Analysis of Axisymmetric Plates and Shells
35
where D is the elasticity matrix in the global
coordinate system and can be defined by

T
T D T D = == = (12)

where T is the transformation matrix which is
defined as

( (( (
( (( (
( (( (
( (( (
( (( (

( (( (









= == =
1 0 0 0
0 s c sc sc
0 2sc c s
0 2sc s c
2 2
2 2
2 2
T (13)

where ) a cos( c = == = and ) a sin( s = == = ; a is the angle
between the global x axis and the local x axis
in the undeformed configuration.
Green-Lagrange strains in the global
coordinate system can be written as

( (( ( ) )) ) ( (( ( ) )) ) [ [[ [ ] ]] ]
( (( ( ) )) ) ( (( ( ) )) ) [ [[ [ ] ]] ]
[ [[ [ ] ]] ]




) )) )




` `` `












( (( (
( (( (

( (( (




| || |

| || |

\ \\ \
| || |
+ ++ +
+ ++ +
+ ++ +
+ ++ +


) )) )


` `` `








+ ++ +
= == =
+ ++ + = == =
2
y x y x
2
y
2
y
2
x
2
x
x y
y
x
L 0
x
u
2
1
v v u u
v u
2
1
v u
x
u
v u
v
u
2
1
(14)

where
0
and
L
are the linear and nonlinear
strains, respectively. If we define

[ [[ [ ] ]] ]
T
y y x x
x / u v u v u = == = (15)

thus the linear and nonlinear strains of Eq. (14)
can be written conveniently as

= == =
( (( (
( (( (
( (( (
( (( (

( (( (






= == = H
1 0 0 0 0
0 0 1 1 0
0 1 0 0 0
0 0 0 0 1
0
(16)

= == =
( (( (
( (( (
( (( (
( (( (

( (( (






= == = A
2
1
x / u 0 0 0 0
0 v u v u
0 v u 0 0
0 0 0 v u
x x y y
y y
x x
L
(17)

Taking the variation of the linear strain

= == = d d
0
H (18)

and let

u Gd d = == = (19)

then, we can obtain the linear strain-
displacement matrix
0
B as

u B u G H d d d
0 0
= == = = == = ; G H B = == =
0
(20)

where the variation of the nodal parameters du
is given by

[ [[ [ ] ]] ]
T
k k k 1 1 1
d dv du d dv du K = == = du

1,2,..n k = == = (21)

Taking the variation of the nonlinear strain using
Eq. (17)

+ ++ + = == = d
2
1
d
2
1
d
L
A A (22)

and we have

= == = d d A A (23)

then

= == = d d
L
A (24)

If Eq. (24) is rewritten using Eq. (19), the
nonlinear strain-displacement matrix
L
B can be
obtained as

u B u G A d d d
L L
= == = = == = ; G A B = == =
L
(25)

Thus using Eq. (20) and Eq. (25), we can write

C. Polat and Z. . Ulucan
36
[ [[ [ ] ]] ] u G A H u B B u B d d ) ( d d
L 0
+ ++ + = == = + ++ + = == = = == = (26)

where B is called as the strain-displacement
matrix [5,6].
The equilibrium equation of the
nonlinear system can be written as

P F P B u R
) i (
V
T
dV ) ( = == = = == =

(27)

where R, F and P represent the out-of-balance
force vector, the internal force vector and the
externally applied load vector respectively, and
) i (
is the load-level parameter.
Taking the variation of Eq. (27) we have


= == = + ++ + = == =
V V
T
T T
d dV d dV d d u K B B R (28)

and from Eq. (9) we can write

u B D D d d d = == = = == = (29)

then Eq. (28) becomes

u K B R d dV d d
V
T
+ ++ + = == =

(30)

The matrix B is defined by the Eq. (26), and the
matrix K is defined as


+ ++ + = == = = == =
V
L 0
T
dV K K B D B K (31)

Substituting
0
B and
L
B matrices into the first
part in the right hand of the Eq. (30) instead of
B, that part can be written as



+ ++ + + ++ + = == =
+ ++ + = == =
V
T T T T T T
V
T
L
T
0
V
T
dV ) d d (d
dV ) d (d dV d
A G A G H G
B B B
(32)

or

u K
u K K K B
d
d ) ( dV d
1 3 2
V
T

= == =
+ ++ + + ++ + = == =

(33)
Hence, the Eq. (30) can be given as

( (( ( ) )) ) u K u K K
u K u K R
d d
d d d
T
= == = + ++ + = == =
+ ++ + = == =

(34)

where
T
K is the tangent stiffness matrix and the
explicit definition of this matrix can be found in
the thesis of Polat [7].

2.3. Solution Method

In order to compute the nodal
displacements, it is necessary to solve the system
of nonlinear equilibrium equations using an
incremental/iterative method. The load
controlled NewtonRaphson method is the
earliest method that is used to trace the
equilibrium path [8]. This method is based on the
linearization of the equilibrium equations at a
prescribed load level, that is,
) i (
in Eq.(27) is
kept constant during iterations. The iterations are
performed until the residual is smaller than a
prescribed tolerance. This method can trace the
loaddisplacement curve before the occurrence
of a limit point, but generally it will fail to
converge beyond this point.
We adopted two convergence criterions
together in the algorithm. One of them is the
displacement-based convergence criterion and it
can be written as

u u
u
< << < (35)

where u are the iterative displacement
changes, u the total displacements and
u

prescribed tolerance. The norm of the iterative
displacement change can be very small while the
out-of-balance force norm is very large. To use
only this criterion may result in some errors.
Therefore, we used an additional convergence
criterion known as energy-based convergence
criterion of the form [9]

) (
) i ( T
r
T
P u R u < << < (36)





Geometrically Non-linear Analysis of Axisymmetric Plates and Shells
37
2.4. Numerical Examples

2.4.1. Geometric non-linear analysis of a
clamped circular plate subjected to a
uniformly distributed load: The large
deformation analysis of a clamped circular plate
subjected to a uniformly distributed load q with
Poissons ratio 3 . 0 = == = , Youngs modulus
MPa 10 1 E
7
= == = , radius , mm 100 R = == = and
thickness , mm 2 t = is considered in this
example. Linear (L2), quadratic (L3) and cubic
(L4) elements are used to model the circular
plate (Fig 2.). The element matrices and the load
vectors are formed using Gauss quadrature. 2x2,
3x2 and 4x2 integration rules are employed for
the linear, quadratic and cubic elements,
respectively. Table I shows the center deflection
of the plate for different elements and the
analytical solution. In addition, in the Figure 3,
the L3 element and the analytical solution results
are compared. The results of the higher order
elements are very close to the analytical solution.

) a ( ) b (
) c (

Figure 2. Meshes of a circular plate under uniform loading; a) eight linear element, b) four quadratic element, c)
three cubic element.

Table I. Non-dimensional central deflection (v/t) of the circular plate with clamped boundaries under uniform load.

4 4
Et / qR
L2 L3 L4 Analytical [10]
1 0.1500 0.1675 0.1675
0.1690

2 0.2906 0.3206 0.3206 0.3230
3 0.4176 0.4549 0.4550 0.4570
6 0.7210 0.7632 0.7632
0.7610

10 1.0055 1.0438 1.0438 1.0350
15 1.2615 1.2937 1.2937 1.2790



Figure 3. Central deflection of the circular plate.

C. Polat and Z. . Ulucan
38
2.4.2. Geometric non-linear analysis of a
clamped circular plate subjected to a
concentrated load at the centre of the plate:
The circular plate with the same geometric,
material property and the same meshes as the
one in Section 2.4.1 is subjected to a
concentrated load at the centre. The non-
dimensional central deflection of the circular
plate from the present elements by total
Lagrangian formulation is compared with the
analytical solution by Ref. [10] in Table II. The
non-linear relationships between load and
deflection are also shown in Figure 4. Also, the
solution of the higher order elements is very
close to the analytical solution for this problem.

Table II. Non-dimensional central deflection (v/t) of the circular plate with clamped boundaries under
concentrated load.

Yk ) N 10 (
4

L2 L3 L4 Analytical [10]
1.6 0.1955 0.2129 0.2129 0.2129
3.2 0.3749 0.4035 0.4036 0.4049
4.8 0.5331 0.5674 0.5675 0.5695
6.4 0.6717 0.7084 0.7086 0.7098
8.0 0.7939 0.8314 0.8316 0.8309
9.6 0.9030 0.9405 0.9406 0.9372



Figure 4. Central deflection of the circular plate

2.4.3. Geometric non-linear analysis of a
clamped spherical cap under uniform
external pressure: In Figure 5, the material and
geometric properties of the spherical cap under
uniform external pressure p are given. In the
figure, is the shell parameter and
0
p the
classical buckling pressure. Cubic shell elements
are used to model the clamped spherical cap. The



element matrices and the load vectors are formed
using Gauss quadrature with 4x2 integration
rule. For different values of , dimensionless
axisymmetric snap-through pressures
0
p / p of
clamped spherical cap was investigated. Results
were shown in Figure 6 with a reference
solution. The obtained results are very close to
the analytical solution.
International Journal of Science & Technology
Volume 2, No 1, 33-40, 2007

R
p
H
2 1 4 1
t 1 ( 3 [ 2
/ / 2
) / ( )] =
2 / 1 2 2
2
0
)] 1 ( 3 [ R
t E 2
p

=
3 / 1 MPa 10 2 E
mm 1 t mm 923 . 494 R
5
= =
= =

Figure 5. Geometric and material properties of the axisymmetric shell under uniform pressure.



Figure 6. Comparison of present values of p/p
o
with those of Luo and Teng [11].


3. Conclusion

A geometrically nonlinear formulation based on
total Lagrangian approach is given for the
axisymmetric shell elements. Geometrically non-
linear behaviour of axisymmetric plates
subjected to different loads are investigated. In






addition, axisymmetric snap-through pressures
of clamped spherical caps also are studied under
uniform pressure. The obtained results are
similar to the analytical solution. Especially,
those of the higher order elements give more
accurate results.
References

1 Zienkiewicz, O. C. (1977). The Finite-Element
Method. 3rd edition, McGraw-Hill, London.
2 Parente, E., Vaz., L. E. (2003). On Evaluation of
Shape Sensitivities of Non-Linear Critical Loads.
International Journal for Numerical Methods in
Engineering, 56(6), 809-846.
3 Surana, K. S. (1982). Geometrically Nonlinear
Formulation for the Axisymmetric Shell Elements.


International Journal for Numerical Methods in
Engineering, 18, 477-502.
4 Reddy, J. N. (1997). Mechanics of Laminated
Composite Plates and Shells: Theory and Analysis.
2nd edition, CRC Press, New York.
5 Polat, C., Calayr, Y. (2003). Sabit Ds Basnca
Maruz Eksenel Simetrik Kabuklarn Geometrik
C. Polat and Z. . Ulucan
40
Olarak Lineer Olmayan Analizi. XIII. Ulusal
Mekanik Kongresi, 657-665, Gaziantep.
6 Polat, C., Calayir, Y., Ulucan, Z. . ( 2006). Post-
buckling Behavior of Geometrically Non-linear
Axisymmetric Shells, Seventh International Congress
on Advances in Civil Engineering, October 11-13,
Yildiz Technical University, Istanbul, Turkey
7 Polat, C. (2006). Geometrik Bakmdan Lineer
Olmayan Kabuk Yaplarn Statik Ve Dinamik
Davran, Doktora Tezi.
8 Memon, B. A., X. Z. Su. (2004). Arc-Length
Technique for Nonlinear Finite Element Analysis.
Journal of Zhejiang University, 5(5), 618-628.
9 Warren, J.E. (1997). Nonlinear Stability Analysis of
Frame-Type Structures with Random Geometric
Imperfections Using a Total-Lagrangian Finite
Element Formulation, Ph. D. Thesis, Faculty of the
Virginia Polytechnic Institute and State University,
Virginia, 211p.
10 Zhang, Y. X., Cheung, Y. K. (2003). A Refined
Non-Linear Non-Conforming Triangular Plate/Shell
Element, International Journal for Numerical
Methods in Engineering, 56, 2387-2408.
11 Luo, Y. F., Teng, J.G. (1998). Stability Analysis of
Shells of Revolution on Nonlinear Elastic
Foundations. Computers and Structures, 69, 499-511.