~OMPI.

rrER METHODS IN APPLIED MECHANICS AND ENGINEERING 27 (1981) 167-181

qORTH-HOLLAND PUBLISHING COMPANY

NONLINEAR FINITE ELEMENT ANALY3IS OF

S H E L L S - P A R T II. TWO-DIMENSIONAL SHELLS

Thomas J.R. HUGHES* and Wing Kam LIU**

Division of Engineering and Applied Science. California Institute of Technology, Pasadena. CA 91125,
USA
Received 9 October 1980

A general nonlinear finite element formulation is given for two-dimensional problems. The
formulation applies to the practically important cases of shells of revolution, tubes, rings, beams and
frames. The approach is deduced from a corresponding three-dimensional formulation [4] and this
enables a simplified implementation, especially with respect to constitutive software. Uniform reduced-
integration Lagrange elements are employed and shown to be very effective for the class of problems
considered.

1. Introduction

This paper is a sequel to [4] in which a general three-dimensional nonlinear finite element
shell formulation is given. Ref. [4] is referred to as *Part I" subsequently. Herein we consider
an analogous two-dimensional formulation which includes several special cases of practical
interest, namely, the axisymmetric case for shells of revolution, the plane strain case for long
tubes, and the plane stress case for rings and two-dimensional beam/frame structures. A novel
i~.~ture of the present formulation is that it is deduced from the three-dimensional for-
mulation. This represer~ts a conceptual economy and allows much of the three-dimensional
software to be also used for the two-dimensional cases. In particular, the entire three-
dimensional continuum constitutive library may be used to directly support all three-dimen-
sional and two-dimensional shell elements. This considerably simplifies the implementation
phase of development, but necessarily entails some computational overhead due to the degree
of generality attained in one element software package.
Much of the present formulation follows the three-dimensional formulation of Part I
closely that it was decided to omit all redundancy and make heavy reference to Part I. As a
result, a serious study of this work requires a preliminary reading of Part !. Only a small
number of references, which have direct bearing on the present developments, are cited
herein. A more extensive list is presented in Part I.
An outfine of the remainder of the paper is as follows: in section 2 we discuss the kinematic
and geometric descriptions of two-dimensional shell elements. The ideas here are virtually
identical to the three-dimensional case and only a very brief sketch is presented. In ~ction .~
we consider the numerical integration of a class of rate-type constitutive equations introduced
in Part I. Specific attention is paid to the modifications of kinematic quantities necessary tc
manifest the particular constraints of the various two-dimensional theories (e.g.. in an~

*Now at Division of Applied Mechanics, Stanford University, Stanford, CA 94305, USA.

**Now at Dept. of Mechanical and Nuclear Engineering, Northwestern University, Evanston, IL 60201. USA

D 045-7825/81/0000--0000/$02..50 © North-Holland
168 T.J.R. Hughes, W.K. Liu, Nonlinear F E M of shells - Part 1I. Two-dimensional shells

out-of-plane zero normal stress conditions). Once the kinematic quantities are appropriately
modified, use may be made of the general three-dimensional algorithm described in Part I.
This algorithm has some nice computational properties and the reader is referred to Part I for
a detailed description. In section 4 the construction of various element arrays and related
topics are dealt with. It is shown, among other things, how to modify the tangent stiffness
matrix to account for the constraints of the two-dimensional theories. In section 5 we present
the elements studied herein. These are the 'uniform reduced-integration Lagrange elements'
which are ideally suited to the class of problems considered. The effectiveness of the
formulation is demonstrated in section 6 by way of a number of nonlinear example problems.
Conclusions are presented in section 7.

2. Geometric and kinematic descriptions

The geometric and kinematic assumptions for the two-dimensional case follow directly from
the three-dimensional case treated in Part I by omitting the ~ dependence and ignoring the
third (out-of-plane) component of all vectors. In this case, the geometry of the shell is
represented by a two-dimensional quadrilateral section as shown in fig. 1. As in Part I, the
~-coordinate defines the fiber direction, and the ~-coordinate lines coincide with the shell
laminae.

physical shell
9
element domain

biunit square

Fig. 1.

Lamina coordinate systems

Lamina coordinate systems are defined by (see fig. 2)

;e~,i (2.1)
e~ = [eI2J--- y 'JI[y "~[['
@ 2/ -----e 1;1 @ 2 - el.2@1, (2.2)
•here

._{o} (2.3)
 T..LR. Hughes, W.K. Liu, Nonlinear FEM of shells - Part II. Two-dimensional shells 169

Fig. 2. Lamina coordinate systems along a fiber.

•h e global-lamina transformation is defined by

q - [el e2]' ' : global -* lamina. (2.4)

~ber coordinate systems
The fiber coordinate basis is defined as follows:

d = f, (2.5)
e( = ~ e , - ";',e~. (2.6)

"he fiber-lamina transformation is given by

r = [r0]: fiber--* lamina, (2.7)

,,, = e~ " d , (2.8)

nd the fiber-global transformation is

s - [e~ e~]: fiber-* global. (2.9)

,. C o n s t i t u t i v e e q u a t i o n s

The class of constitutive equations considered is the same as in Part !. The three-
imensional, numerical algorithm for integrating constitutive equations presented in Part I can
lso be used in the present circumstances if the incremental strain and rotation matrices (i.e., y
nd ~ ) ate a~oropriately modified for the two-dimensional cases. To simplify the presentation
~is helpful r~¢ this point to introduce [Link] notations for stress, strain and rotation
Luantities. These definitions are consistent with those introduced in Part I and are reproduced
ere for the convenience of the reader, viz.
YO [Link]. Hughes, W.K. Liu, Nonlinear F E M of shells - Part 11. Two-dimensional shells

0.1 0". ]
0.2 0"12
0.3 0"22 (3.1)
0"4
0"5 0"23
0"6 0"31

Yll
Y2 2y12
Y3 y22 0.2)
Y41 Y33
"/5 2y23
Y6 273]

0,2 = 2 ~023 (3.3)

033 ('031

Etis worthwhile to reiterate that, in two-dimensional cases, the above coordinate indices 1 and
represent the updated lamina coordinates, and index 3 is the out-of-plane coordinate.
In all two-dimensional cases we assume:

0"3= 0 (zero normal-stress condition), (3.4)

0"5=0"6=0, (3.5)

75 = 76 = 0 , 0.6)

(/)2 ~ 033--'-- 0 .
(3.7)

Note that for a fully-general coupling in the material constitutive equation, (3.5) and (3.6) will
not be consistent. Thus we restrict ourselves to cases in which there is no coupling between
these components. For most situations of practical interest this limitation is unimportant.
Likewise, for all two-dimensional cases, yl, Y2, and col are calculated from eqs. (3.9) and
(3.10) of Part I.
The additional assumptions particular to the individual two-dimensional cases are sum-
marized below:

Torsionless axisymmetric case

y4 = &/r, 0.81
where r denotes the radial coordinate, and 8, is the increment in radial displacement (cf, eq
(3,6) of Part I). The radial coordinate is assumed to coincide with direction 1 of the globa
reference frame.
 T.1.R. Hughes, W.K. Liu, Nonlinear FEM o.f shells- Part 1I. Two-dimensional shells 171

)lane strain

y,=O, (3.9a)

)lane stress

0"4= O. (3.9b)

In each case, 3,3 must be computed so that the zero normal-stress condition is satisfied. In
he plane stress case, both y3 and y4 must be determined so that a zero normal-stress condition
; satisfied both in and out-of-plane (i.e., 0" 3 ---" 0"4 = 0 ) . The calculation of these strain measures
; described as follows:

[xisymmetriclplane strain

(3.1o)
)lane stress

{I,:}- - re_-'_,,e_,,-r, e,,,,+ (~11)

tc,, c.] ~,,y,+ O,~y~j"
n (3.10) and (3.11), the c'~mponents of the matrix C" are determined by

Cu = ~jkt, (3.12)

~here ~'~k~denotes the given material response tensor (see eq. (3.1)of Part i).
All kinematic quantities having been defined above, direct use may now be made of the
lgorithm for stress computation presented in Part I, (3.14)-(3.17).

, Element arrays

:.1. Material tangent matrix

The material tangent matrix for the three-dimensional case. including initial stress effects,
as been defined in Part I (see section 4.1). The construction of the two-dimensional
ounterpart begins with the 9)< 9 matrix, D, of three-dimensional continuum theory. This
latrix is contracted via a transformation which removes unnecessary rows and columns, and
ccounts for zero normal-stress conditions. This may be written in the following way-

= ptDp ' (4. l)

there
172 TJ.R. Hughes, W.K. Liu, NonlinearFEM of shells- Partill. Two-dimensionalshells

1 0 0
0 1 0 0
Pn Plz P13 pt4
172, p22 0 pz~
p_. 0 0 0 0 PI4 = C310"2 (4.2)
0 0 0 0
0 0 0 1
0 0 0 0
0 0 0 0

and
Axisymmelric

(~11P12 P13) -" -- ~31(C31 C32 C34), P21 -" P22 = O, p23 --" 1. (4.3)
Plane swain

P2I = P22 = P23 = 0 . (4.4)

Plane stress

p,,1--- c_-:-I-' c_-q Pt3 = p~ = O. (4.5)(4.6)

21 P22J [.Co C44J [.C4I C42J'

l'he matrix b has dimension 4 x 4. In fact, in the plane stress and plane strain cases, only a 3 x 3
;ubmatrix is nontrivial. Due to the sparse, simple structure of P, efficient implementation of the
~atrix triple product, indicated in (4.1), may beeasily carried out.

4.2. Swain-displacement matrix

The nodal strain--displacement matrix is defined by

[s:l
B , = I.B~'J' a = 1, 2 , . . . , n~,, (4.7)

~¢here n,. is the number of element nodes, and

B'. =
I B~
B2
0
Bl
B. 0]
Bs B~ , (4.8)
quBs q21Bs qllB6 q21B6
B~ = (B2 - Bl i Bs - B4), (4.9)
Na, i , i = 1,2,
B, = NJr , i=3,
(N.z.) , /=4,5, (4.10)
NozJr , i-6.
 [Link]. Hughes, W.K. Liu, Nonlinear F E M of shells - Part 1I. Two-dimensional shells 17"

In (4.10), N, and zo denote the shape and 'thickness' functions associated with node a (sec
Part I, section 2 for further explanation). The derivatives in (4.10) are taken with respect to the
lamina coordinates at the point under consideration (e.g., N,~, = ONdOy~, where yt = qy).
Shear correction effects may be introduced by multiplying the second row of B~ by Kv2 (K i,.
Ihe shear correction factor and is typically taken equal to 5/6). Whenever B~ is employed, the
;hear correction effect should be consistently introduced. For further simplification on thL,
aspect, see Remark 2, section 4.2, Part I.

4.3. Transformation matrices

The global degrees-of-freedom in the two-dimensional case are the nodal displacement and
rotation increments. This results in three degrees-of-freedom per node. ~'lhe matrices required
to transform element arrays to global degrees-of-freedom are given as follows:

l., q,2i
q2, q -
""6" .... i/ ..... '
(4.11]
O*=Lo o
1 0 0 1
S. = .....°..I.
0 -1 I (4.12)
0 0 OA

4.4. Tangent stiffness matrix and internal force vector

The element stiffness and internal force are defined by

k= [k,,], j~' = {jf~,'}, (4.13)

fTf7 [Link] d~ d~?,

t t "
(4,14}

r-'-ITI7 O',,(BlYa.,,j d~"d~ (4.15)

where

plane stress/strain, (4. lh]

J= /]-" axisymmetric,

f ~ d©t[Y:.: Y',~] (4.17)

• YzcJ'

O'vec = 0"2 "

(4.181
0"4
174 [Link]. Hughes, W.K. Liu, Nonlinear FEM of shells- Part 11. Two-dimensional shells

REMARK 1. As in our three-dimensional work, the variation of Q. with ~" is assumed

nsignificant. Thus Q°(~)-Qo(~), where ~ defines the location of the reference surface (see
;ection 2.1, Part I, for further details), which facilitates the removal of 0= from the fiber
integral, viz.
g+l F t +I - •
,..... = j_. o:.tj_. B-.B..j d+'J0 d., (4.19)

.fi:t : f ~ ' 'Q: [f~+ (B:)to',,.=j d+] d+. (4.2o)

R E M A R K 2. When initial stress effects may be neglected in the formation of the stiffness, the
BtDBb product may be replaced by (B.'yL'B], where

= p'Cp, (4.21)

where p denotes the first four rows and columns of P, defined in (4.2). The matrix C has
:omponents defined by Cu = cs~tt (see Part I, section 4.1, and [3] for further details). The
dimension of L" is 4 × 4, although in the platte stress case only a 3 x 3 submatrix, is non-trivial.

4.5. External force vector

Body force
The body force vector is given by
~, = {,,~.,}, (4.22)

~" = s"
f 'fy
_ _ N'~bpojo d~ d~ (4.23)

where p0 is the mass density in the initial configuration, b is the prescribed body force vector
(per unit mass),

j0, plane stress/strain, (4.24)

jo = tro)'0, axisymmetric,

]'o = det[ x''" (4.25)

I.X2.n X2,¢J '
o + o] (4.26)
No= N,, ] Nozo '

md x~, x2 and ro denote coordinate functions in the initial geometry.

~urface force
The surface force vector is given by
 T.Y.R. Hughes, W.K. Liu, Nonlinear F E M of shells - Pare II. Two-dimensional shells 17~

url ._ f~art
f (4.27'
{+I top,
bottom, (4.281
/. = ~'~, plane stress/strain,
(4.291
tr], , axisymmetric,
/. -Ily,.ll, (4.30i

s,here li is the surface force vector (per unit surface area), which includes both normal
pressure and tangential shear as special cases, viz.
Pressures

= ( - 1)'pc2~ . (4.31)

In (4.31), p is the pressure.

Shear

--~,Del. (4.32)

In (4.32), ~,, is the tangential shear in the ~-direction.

4.6. SWess resultants

Let fit denote the quadrature points of the rule used in the ~-direction ('lamina rule'). The
;tresses are assumed stored at various points along the fibers through the ,j~'s. The stress
resultants are given as follows:
Moment

m.(fii) = _ o'.(fi,, ~')z(fih ~')d~'zz(fi,), (4.33)

z(fi,, D = N÷(~)z÷(~,)+N-(Dz-(~,), (4.34)

mdm
z'(,7,) ffi ]~ No(l~,)z:, (4.35)
a-I

zz(fi,) = [ z ' ( f i , ) - z-('~,)|12. (4..~)

Membrane forces

n.(fi,)= _ o'u(fiu~')d~'zz(fit), ii= 11 or 33 (no sum). (4.37)

Shear

q(lq,) = K '~ f÷' ,r,,(,i,, t3 d~z~(,i,) • (4.38)

J-I
176 TJ.R. Hughes, W.K. Liu, Nonlinear F E M of shells - Part 11. Two-dimensional shells

1133 ~ roll

rnll
Fig. 3. Sign conventions for stress resultants.

The fiber integrations may be calculated by a variety of quadrature rules. A discussion of

various possibifities is presented in section 4.6 of Part I.
The sign conventions for the stress resultants are illustrated in fig. 3.

4.7. Boundary conditions

The boundary conditions are essentially the same as for the three-dimensional case (see
section 4.8, Part I ) That is, the kinematic conditions are the displacements of the node and
unit vector which defines the fiber orientation, and the mechanical conditions are the edge
force and moments. These latter quantities are specified per unit depth in the plane
stress/plane strain cases, and per unit radian in the axisymmetric case.

$. Shell elements

The shell elements used in the present work employ one-dimensional Lagrangian inter-
polation in the ,1 direction, and uniform reduced quadrature (see fig. 4). It is essential to use
reduced integratio~a to avoid the shear 'locking' phenomenon which has been described at
length elsewhere (see Part I, section 4.2 for pertinent references). Rank deficiency problems,
which afflict many three-dimensional shell elements derived along similar lines are not present
at all in the two-dimensional uniform reduced integration Lagrange elements.
The 2-node element was first suggested in [5] and developed independently for linear
~pplications in [2] and [14]. In [14] a number of problems were solved, and the accuracy and
economy of the element were demonstrated. The 3-node element seems to have been
employed by several investigators and its origin is uncertain. It has been observed by H. AUik
Lhat the 3-node element yields the exact stiffness matrix of the technical theory of beams
~ccounting for shear deformations, Equivalence of these elements with elements derived via
;mixed methods' has been proven in [7].
A possible improvement in the 2-node element in axisymmetric applications might be made
Jsing a 'mean value' of the matrix B, (with respect to ~ integration) rather than the value at
r/= 0. This is in keeping with the approach originally presented for incompressibility by
~agtegaal et at. [10] and discussed more recently in [9].
 TJ.R. Hughes, W.K. Liu, Nonlinear FEM of shells - Part II. Two-dimensional shells 17:

shape
linear quadratic cubic
functions

quadrature
I- point :) - po int 3-point
rule

Fig. 4. Uniform reduced integration two-dimensional shell elements.

S. S a m p l e problems

All computations were performed in double precision (64 bits per floating point word) using
Ihe California Institute of Technology IBM 3032 computer. Linear elastic properties were used
Ihroughout and 2-point Gaussian fiber integration was employed.

5.1. T~e elasfica

The classical elastica [12] was analyzed with a mesh of twenty 2-node elements. Data
~mployed in the calculations are given as follows: Young's modulus = 4.8 × 10~: Poisson's
ratio=0, length (L) = 10, and thickness (h)= 0.01. Load-deflection results and deformed
~rofiles are shown in fig. 5.

3.0
P
| ¢
]~cr 2.5
• ¢ ¢ . , . , o- •
2.0 ~w

h5

1.0

0.5
I I I t
o .2 .4 .6 .8 hO .z .a .6 ~ ........ "
w/L wtL
load vs deflechon defor meO Drofdes
Fig. 5. The elastica.
178 TJ.R. Hughes, W.K. Liu, NonlinEar F E M of shells - Part II. [Link] sh~lls
,%
0.0
1.0
qp~
.8 " M
-2.0

-4.0
6
" M/2~T = .2
-6.0

20 2-node elements and

I0 3-node elements
-8.0
Fig. 6. Pure bending of a cantilever beam,

fl
|.0
./ I
0.8
hinge ; 2P

(16
T/=
IS.O
0.4
~ large deflection
12.0 ~" --- smOIIMflection
0.2 .~ --,-- lilwor theory
~" o o e~perimentol dote
jo)nt r 6 • present

9,0 Oo i
[Link]
I
O,IS
,.I
0.24
I
0,32 0.4
2v/L
t2P vertical •ontraction
6.0

1.0

~2 =
,./
0,8
3.0
0.0
9.0 12.0 15,0 18.0 21.0
0.6

3.8
-3.0
0.4
E = 3.0 x 10 7 psi

-6.0 =0.0
0.2
I = 1/12 in4 (h=[Link])

L = 15~"~ in
I I i ,
~2: pL2/EI ~00 aDS 0.24 0.32 ~4
2u/L
d e f l e c t e d shapes f o r t h i r t y horizontal elongation
2 - node e l e m e n t s
~g. 7. Large deflection of a diamond-shaped frame. Experimental data and analytical results are from J.A. Jenkins et
,t. [6].
 TJ.R. Hughes, W.K. Liu, Nonlinear FEM of shells- Part H. Two-dimensional shells 179

;.2. P u r e bending of a cantilevered beam

A thin cantilever beam was subjected to an end moment. The end moment was increased
inearly up to a final value of 21r. Two hundred load steps were employed and approximately two
Ierations per step were needed for convergence. The parameters of the beam were chosen
.ch that the exact final diameter to length ratio (D/L) equaled ~r-' = 0.3183. The results for
nenty 2-node elements and ten 3-node elements were found to be virtually indistinguishable.
)eformed profiles are shown in fig. 6. The approximate DIL equaled 0.3182.

;.3. Large deylection of a diamond-shaped frame

A diamond shaped frame was analyzed and results were compared with analytical and
xperimental data from [6] (see fig. 7). Seventy-six equal load steps were employed. The average

R : 4`76in
i
h : 0.01576 in
~. \ ~ 4 ~ - H = 0.0859 in
R e = 10.9 a
E = IOXlO s Ib/in z
v =0.3

•' 2 F P, o

0.0 ~ " 54 .?2 .90

IO0
-.12
deformed profiles

tO0
Bathe ./~
w -- Stricklin (thirty shell elements) ~ 4
• Mescoli (finite difference soluhon| /./I
80 - o present Study I 0 0 sleds x~./~
x present study 5 0 steps / I !
(twenty and thirty 2 - n o d e ~,,{ i J
e l e m e n t s , and ten ond "~/ I
f i f t e e n 3 - n o d e elements /if t
60 - yield the some results) x~//
JD
m

O.
I i neoy ~, T
40

20

I 1 , t l l
o.4 o.e 1.2 I.S 2.0
wo/H

Fig. 8. C~nped spherical cap subjected to concentrated load.

180 TJ.R. Hughes, W.K. Liu, Nonlinear FEM of shells - Part II. Two-dimensional shells

number of iterations per step over the first forty steps was less than three. Subsequently the
number of iterations necessary for convergence increased. During the last six steps, betweee
sixteen and thirty iterations per step were required.

6.4. Clamped spherical cap subjected to a concentrated load

In this problem a thin spherical cap is loaded by a concentrated force at the vertex. Results
were calculated for meshes consisting of twenty and thirty 2-nede elements and ten and fifteen
3-node elements. The different spatial discretizations did not produce discernible differences in
results. However, refining the load discretization from fifty to one hundred did bring the
results in closer agreement with results of other investigators [1, 8, 11]. (See fig. 8.)

6.5. Clamped circular plate subjected to uniform pressure

Calculations for a clamped circular plate subjected to uniform pressure (nonconservative}
are compared with analytical results [13] in fig. 9. Meshes of thirty-two and sixty-four 2-node
elements were employed and the load was applied in twenty-four equal steps.

qL' Jr Jr Jp a = lO.O(radius) edge moment, edge moment,

JP q h - 0.1 nonlinear linear
E E = iOe 7
q ~ Wo !: =1 = .25
6

2.0 v i ~ v ,/i 5
~ Timoshenko and / -center
1.6 - Woinowsky-Krieger/ moment,
4 linear
• present study, /Flinear
thirty-two and /
.,- I.."" sixty-four / 3
"~- J 2- node / _~r -center
.~0.81 e l e ~ nonlinear moment,
2
I I nonlinear
° , I / I
0 I I I I I 1 | I I
0 2 4 6 8 I0 12 0o 0.4 0.8 1.2 1.6 1.8
load qa4/(Eh 4) *o/h
displacement vs load bending moment vs displacement

Fig. 9. Clamped circular plate subjected to uniform pressure.

7. Conclusions

In the present paper a general nonlinear finite element formulation is given for two-
Jimensional shell analysis. The formulation is appficable to the practically important cases of
;hells of revolution, tubes, rings, beams and frames. The development herein is deduced
Jirectly from the three-dimensional shell formulation presented in Part I.
A unique feature of the approach is that nonlinear three-dimensional constitutive equations
 T.J.R. Hughes, W.K. Liu, Nonlinear F E M of shells - Part 11. Two-dimensional shells 18

ire employed in all two-dimensional cases. The necessary modifications for each case arc
incorporated directly in the numerical algorithms. This avoids developing special 'reduced
~onstitutive equations for each two-dimensional case, an inconvenient task in nonlineal
dtuations, and also entails an economy of computer implementational effort, in that much ol
Ihe software for three-dimensional continuum constitutive equations may be employed un.
~ltered.
The elements advocated herein are the uniform reduced-integration Lagrange elements
rhis class of elements possesses excellent properties for the type of problems considerec
[terein. Application of the 2- and 3-node members of the family to several nonfinear problem,
produced very good nur~erical results.

Acknowledgment

We wish to gratefully acknowledge the following individuals and organizations for theix
interest in and support of ~he research presented herein: J. Crawford, Civil Engineering
Laboratory, Port Hueneme, CA; the Electric Power Research Institute, Palo Alto, CA and
Ihe National Science Foundation.

References
[1] K.J. Bathe, E.L. Wilson and R.H. Iding, NONSAP-A structural analysis program for static and dynamic
response of nonlinear systems, UC SEM 74-3, Dept. of Civil Engrg,, Univ. of California, Berkeley. CA 94720
[2] G.L. Gouclreau, A computer module for one-slep dynamic response of an axisymmetric plane linear elastic thin
shell, Lawrence Livermore Laboratory Rep. No. UCID-17730 (19"/8)
[3] TJ.R. Hughes, On consistently derived tangent stillness matrices, in preparation.
[4] T.J.R. Hughes and W.K. Liu, Nonlinear finite element analysis of shells: Part 1, Three-dimensional shells,
Comput. MetEs. Appl. Mech. Enlprg. 26 (19'31) 331-362.
[5] T.J.R. Hughes, R.L. Taylor and W. Kanoknukulchai. A simple and elficient element for plate bending,
[nternat. J. Numer. Meths. Engrg. 11(10)(1977) 1529-1543.
[6] J.A. Jenkins, T.B. Seitz and J.S. Przemieniecki, Large deflections of diamond-shaped frames, lnternat..I,
Solids and Structures 2 (1966) 591-603.
[7] D.S. Malkus and TJ.R. Hughes, Mixed finite element methods-Reduced and selective integration tech-
niques: A unification of concepts, Comput. Meths. Appl. Mech. Engrg. 15(1)(1978)6.t-81.
[8] J.F. Mescall, Large dellection of spherical shells under concentrated loads, J. Appl Mech. 32 (196.';) 9~,--t~3~¢
[9] J.C. Nagtegaal and J.E. de Jong, Scene computational aspects of elastic-plastic large strain analysis, preprint
[10] J.C. Nagtegaal, D.M. Parks and J.R. Rice, On numerically accurate finite element solutions in the fully plastic
range, Comput. Meths. Appl. Mech. Engrg. 4 (1974) 153-177.
[11] J,A. Stricklin, Geometrically nonlinear states and dynamic analysis of shells of revolution. High Speed
Computing of elastic Structures, Proc. IUTAM Syrup., University of Liege (1970).
[12] S.P. Timoshenko and J.M. Gere, Theory of Elastic Stability (M~n'[Link], New York. 19~1).
[13] S.P. Timoshenko and S. Woinowsky-Krieger, Theory of Plates and Shells (McGraw-Hill. New York. 1959) 2nd
ed.
[14] O.C. Zienkiewicz, J. Bauer, K. Morgan and E. Onate, A simple element for axisymmetric shells with ~itear
deformation, Internat. J. Numer. Meths. Engrg. 11 (1977) 1545-1558.