Você está na página 1de 44

Non-linear Finite Element Analysis of

RC Shell Structures

M.Tech Project Report Phase - I

Submitted by
Aminoor Rahman Laskar
Roll No - 09410405

Supervisor - Dr. A. K. Singh

DEPARTMENT OF CIVIL ENGINEERING

INDIAN INSTITUTE OF TECHNOLOGY GUWAHATI

November, 2010
Abstract

The main objective of the present study is to develop a general, robust and ecient procedure
for non-linear nite element analysis of reinforced concrete (RC) shell structures considering
material non-linearity. Finite element formulation is based on so-called degenerated shell
element proposed by Ahmad and co-workers. Reduced and selective integration schemes
with various spurious mode stabilization techniques will be implemented to overcome the
problem of shear and membrane locking in case thin shell problems. Layered theory will be
implemented to model the material properties of concrete and steel separately. Formulation
for non-linear analysis has been presented where elastoplastic constitutive relations involving
nonlinear isotropic and kinematic hardening have been implemented to deal with material
nonlinear behavior of shells. von Mises yield criteria has been used where associated ow
rule has been employed.

In rst phase of the project, nite element formulation based degenerated shell element has
been implemented using C programming for linear analysis and some benchmark examples
has been solved. Discussions on the computed results and plan of future work are presented.
Contents

Abstract i

Contents ii

List of Figures iv

1 Introduction 1

1.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Literature review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.3 Scope of the Present Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.4 Layout of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2 FE formulation of shell using degenerate shell element 7

2.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2.2 Element geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.3 Displacement eld . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.4 Strain-displacement relationship . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.5 Stress-strain relationship . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2.6 Stiness matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.7 Consistent load vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.7.1 Gravity load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

ii
2.7.2 Uniform vertical load . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

2.7.3 Uniform normal load . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

2.8 Layered model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

3 Formulation for elastoplastic analysis of shells 19

3.1 Constitutive relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

3.1.1 Decomposition of elastic and plastic strains . . . . . . . . . . . . . . . 19

3.1.2 Elastic stress-strain relationship . . . . . . . . . . . . . . . . . . . . . 20

3.1.3 Initial yield function and subsequent yielding . . . . . . . . . . . . . . 20

3.1.4 Flow rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

3.1.5 Hardening rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

3.1.6 Loading-unloading criteria . . . . . . . . . . . . . . . . . . . . . . . . 21

3.2 Integration scheme: Return mapping algorithm . . . . . . . . . . . . . . . . 22

3.2.1 Steps in return mapping algorithm . . . . . . . . . . . . . . . . . . . 25

3.3 Consistent tangent modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

4 Results and discussions 27

4.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

4.2 Numerical examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

4.2.1 Example 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

4.2.2 Example 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

4.2.3 Results and discussions . . . . . . . . . . . . . . . . . . . . . . . . . . 30

4.3 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

Bibliography 39

iii
List of Figures

2.1 3-D solid element . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.2 Degenerate shell element . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.3 Typical degenerate shell element . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.4 Typical node i and thickness direction vector V3i . . . . . . . . . . . . . . . 10

2.5 Orthogonal vectors at node i and nodal dof . . . . . . . . . . . . . . . . . . . 10

2.6 Displacement of a point P on V3i due to nodal rotations i and i . . . . . . 10

4.1 Programming algorithm for FE formulation . . . . . . . . . . . . . . . . . . 28

4.2 Scordelis-Lo Roof . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

4.3 Vertical deection at midspan(4 4 mesh) . . . . . . . . . . . . . . . . . . . 29

4.4 Vertical deection at midspan(8 8 mesh) . . . . . . . . . . . . . . . . . . . 30

4.5 Parabolic hyperboloid shell . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

4.6 Vertical deection along y = 0(4 4 mesh) . . . . . . . . . . . . . . . . . . . 31

4.7 Vertical deection along y = 0(8 8 mesh) . . . . . . . . . . . . . . . . . . . 32

iv
Chapter 1

Introduction

1.1 Overview

A shell is a curved structure in space which thin in one direction compared to the other two
directions. Main advantage of shell structures is that they are thin and light, span over large
areas, hold applied loads very eectively and thus utilize minimum amount of materials.

Shells as structural elements occupy a leadership position in engineering and, in particu-


lar, in civil, mechanical, architectural, aeronautical, and marine engineering. Examples of
shell structures in civil and architectural engineering are large-span roofs, liquid-retaining
structures and water tanks, containment shells of nuclear power plants, and concrete arch
dams. In mechanical engineering, shell forms are used in piping systems, turbine disks, and
pressure vessels technology. Aircrafts, missiles, rockets, ships, and submarines are examples
of the use of shells in aeronautical and marine engineering.

In order to obtain an eective design, a detailed analysis of shell structures is required. The
diculty of analysis lies in that a shell resists the applied loads largely by its curvature. The
peculiarities of shell structural behavior and the wide use of shell structures have instigated
a large research eort in shell analysis. Various shell theories have been developed and
approximate solution techniques have been proposed. However, for shells having arbitrary
geometry, loading and support conditions, variation in material properties, openings and
many other aspects of practical design, analytical solutions are practical impossible. As such
the nite element method is well suited to structural analysis of shells. The development of
the nite element method for shell analysis is not without formidable diculties and pitfalls.

1
1.2 Literature review

Three distinct approaches to nite element analysis of shells have been pursued:

1. Flat plate elements.

2. Curved shell elements.

3. Degenerated shell elements.

In rst approach, a shell is modeled using at elements by superposition of membrane element


and plate bending element. Main features of this approach are easy formulation, capable
of modeling rigid body motion, capable of incorporating complex loading and boundary
condition. This approach gives acceptable results for many practical purposes. However,
a number of diculties like non-coupling between membrane and bending actions within
individual element, presence of discontinuity bending moments at element juncture line,
which do not exist in the continuously actual curved structure result in requirement of a large
number of elements to achieve satisfactory accuracy. Modeling of shells using at elements
based on displacement formulation can be found in the works of Clough and Johnson [22],
Dawe [27], Argyris, Dunne, Malejannakis and Schelke [3], Meek and Tan [58]. In addition
to displacement based at element, hybrid and mixed formulations were also used by many
authors like Dungar and Severn [31], Mau, Pian and Tong [57], Liu and To [85, 55]. In
non-linear analysis of shells, application of at elements can be found in the works of Bergan
and Horrigmoe [37], Hsiao [38], Liu and To [53, 54], Levy and Gal [52].

The second approach uses curved elements based on classical shell theory developed from
3-D eld equations by incorporating various simplifying assumptions appropriate to the
structural behavior. This approach gives better representation of element geometry than
at plate elements, but main diculties are choice of appropriate shell theory, complicated
mathematical formulation, inter-element incompatibility, incapability to represent rigid body
motion and convergence. In case of non-linear analysis, additional diculties arise in nite
element formulations due to non-availability of a general non-linear shell theory. The curved
shell element can be divided into singly and doubly curved shell elements. Connor and
Brebbia [23], Bonger, Fox and Schmit [13], Catin and Clough [16], Sabir and Lock [76],
Ashwell and Sabir [5], Catin [15], Dawe [26], Fonder and Clough [34] worked on displacement
based singly curved element, while Hanshell, Neale and Warburton [35], Tabarrok and Gass
[81], Edwards and Webster [33], Altmann and Iguti [2] developed various singly curved
elements based on hybrid formulations. Doubly curved shell elements can be found in the

2
works of Utku [86], Argyris and Scharpf [4], Pecknold and Schnobrich [70], Cowper, Lindberg
and Olson [25], Dupuis and Goel [32], Yang [91], Dawe [28], Idelsohn [43]. Hybrid and
mixed formulations for doubly curved have been used by Prato [72], Tahiani and Lachance
[82], Talaslidis and Wunderlich [83], Rigby [75] and Kang and Pian [45]. Curved shell
elements based on discrete Kirchho theory (DKT) were also developed by many authors
like Wempner, Oden and Kross [89], Dhatt [30], Batoz and Dhatt [10], Batoz, Cattopadhyay
and Dhatt [9], Bathe and Ho [8], Nagtegaal and Slater [61],Murthy and Gallegher [60], Dhatt,
Marcotte and Matte [29].

In third approach, the shell is discretized directly from 3-D continuum eld equation. The
basic element in this approach is the isoparametric, curved shell element proposed by Ahmad,
Zienkiewicz and Irons [1] known as degenerated shell element. The shell element is developed
by degenerating 3-D solid element by introduction of assumptions that normal to mid surface
remains straight after deformation and normal stress to the middle surface is zero. This
formulation incorporates the shear deformations and thus gives satisfactory results for thick
to moderately thick shells. This approach gives a better t of curved shell geometry and can
be easily formulated using computer programming. Although this type of element seemed
very promising as a general shell element, it was observed that the element was very inecient
in treating thin systems. The element tends to show shear locking and membrane locking
when applied to thin plates and shells.

Since the introduction of the degenerated shell element by Ahmad, there have been many
attempts to improve the performance of the shell element and various techniques have been
used to eliminate the problems of locking in case of thin shell systems. Pawsey and Clough
[69] developed a selective reduced integration scheme while Zienkiewicz, Taylor and Too
[62] proposed uniform reduced integration scheme to overcome the problems of locking in
Ahmad shell element. However, elements with underintegration subsequently suered from
element deciencies like rank deciencies, spurious zero energy modes or hour-glass modes.
Belytschko [11, 12] proposed some stabilization procedures for spurious mode control. Wil-
son, Taylor, Doherty and Ghaboussi [90] and Choi and Schnobrich [21] suggested addition
of nonconforming displacement modes to improve the performance. Konak-Nukulchai [46]
developed a bilinear degenerated shell element from eight node 3-D brick element based on
degeneration concept while Talha [84] proposed modied version of degenerated shell ele-
ment using he concept of strain denition of semiloof element developed by Irons. Based
on a mixed or hybrid variational principle in which displacements, strains and stresses are
taken as independent variables, many mixed or hybrid degenerated shell elements have been
developed by Saleeb, Chang and Graf [77], Cheung and Chen [19], Rhiu and Lee [74] and

3
Chieslar and Ghali [20]. Many authors tried to develop modied versions of the degenerated
shell element based on assumed strain concept. The elements developed based on assumed
strain concept showed signicant improvements and they generally do not exhibit locking
problems and are also free from spurious or hourglass modes. Some of these works include
MacNeal [56], Bathe and Dvorkin [7], Park and Stanely [68], Huang and Hinton [40], Jang
and Pinsky [44], Kailash and Lakshminayarana [50].

The nite element formulation based on degenerated shell element and its modied versions
has been extensively used in both material and geometric non-linear analysis of shell struc-
tures. Important contributions in this area are Cormeau [24], Paricsh [65, 66], Bathe and
Bolourchi [6], Hughes and Liu [41], Hughes and Carnoy [42], Ramm and Sattele [73], Surana
[80], Owen and Figueiras [63], Jang and Pinsky [71], Hsiao and Chen [39], Kebari and Cas-
sel [47], Hauptmann and Schweizerhof [36], Lee and Konak-Nukulchai [51], Voyiadjis and
Woelke [88].

In case of laminated shell, the material properties are dierent in each layer and the inte-
gration has to be carried out for each layer along thickness direction. However, numerical
integration along the thickness direction results in considerable increase in the computa-
tional time. Hence, in order to increase the computational eciency, explicit integration is
carried out along thickness direction. Various explicit through-thickness integration schemes
are proposed by many researchers like Panda and Natarajan [64], Chao and Reddy [17],
Milford and Schnobrich [59], Yunus, Kohnke and Saigal [92], Vlachoutsis [87], Premakumar
and Palninathan [48, 49]. In non-linear analysis of laminated shells, there are basically two
approaches. One is the layered approach or through-thickness integration approach while
the other is the non-layered approach or stress resultant approach. In non-layered approach,
constitutive relations are integrated through the thickness of shell and expressed in terms of
stress resultants. For the case of non-layered approach, a general and accurate shell theory
is of crucial importance. In layered approach, a shell is divided into layers where stresses
are calculated and the yield condition is checked for each layer separately. The forces and
moments are then calculated by integration through the thickness. Although this procedure
can give very accurate results, computational eciency is very low if numerical integration is
carried out through the thickness. Layered approach is implemented by Parisch [67], Hughes
and Liu [41] and Buchter, Ramm and Roehl [14], while stress resultant approach is used by
Simo and Kennedy [79].

4
1.3 Scope of the Present Work

From above literature review, we have seen that the rst approach i.e. at plate element
approach is easy to formulate, but needs a large number of elements to give satisfactory
results and thus not much reliable. The second approach of curved shell element based
on classical shell theory gives a better representation of curved geometry of the shells, but
is dicult to formulate specially in case of non-linear analysis as governing equations for
non-linear analysis are very complex and awkward to deal with. On the other hand, the
formulation of the degenerated or continuum based shell element approach is straightforward
and avoids the complexity of shell theory; it yields good results and can be easily implemented
to non-linear analysis of shell structures. In the present study, the degenerated shell element
has been adopted.

The main objective of the present work is to develop a general, robust and ecient proce-
dure for non-linear nite element analysis of reinforced concrete shell structures considering
material non-linearity. The main feature of the present work are outlined as follows:

1. Finite element formulation is based on degenerated shell element proposed by Ahmad


and co-workers [1].

2. Reduced and selective integration techniques along with various stabilization proce-
dures will be employed to overcome the problem of locking in case thin shell systems.

3. Layered theory will be implemented to model the material properties of concrete and
steel separately along the thickness direction. Explicit integration assuming constant
jacobian along the thickness direction will be carried out to improve the computational
eciency.

4. Elastoplastic constitutive relations involving associated ow rule and nonlinear isotropic


and kinematic hardening will be implemented to deal with material nonlinear behavior
of shells.

1.4 Layout of the Thesis

The thesis consists of four chapters. A brief description of each chapter is presented here.

Chapter 1 presents a brief overview of analysis of shell structures followed literature review
and scope of present work.

5
Chapter 2 provides the nite element formulation of shells based on so-called degenerated
shell element and formulation of the layered approach.

Chapter 3 deals with material nonlinear formulation of shells considering elastoplastic con-
stitutive relations.

Chapter 4 presents the results and discussion on present work and gives plans for future
work.

6
Chapter 2

FE formulation of shell using


degenerate shell element

2.1 General

Figure 2.1 shows a solid 3-D element based on a quadratic displacement eld and Figure 2.2
shows the corresponding degenerated shell element. The external faces are curved, while the
sections across the thickness are generated by straight lines.

Two main assumptions are used in the degeneration process from solid 3-D element to shell
element. They are:

1. Normals to the middle surface before deformation remain straight after deformation
but not necessarily normal to the deformed middle surface.

2. Strain energy associated with stresses perpendicular to middle surface is disregarded


i.e., the stress component normal to the middle surface is constrained to zero in the
constitutive equation.

The statement that after deformation normals do not necessarily remain normal to deformed
middle surface permits the shell to experience shear deformation - an important feature in
thick shell systems. Typically ve independent degrees of freedom are specied at a nodal
point viz. three displacements of node and two rotations of normal at the node.

7
Figure 2.1: 3-D solid element Figure 2.2: Degenerate shell element

2.2 Element geometry

Let us consider a typical degenerate shell element illustrated in Figure 2.3. The global
coordinates of pairs of points on the top and bottom surface at each node are usually input
to dene the element geometry. Let , be the two curvilinear coordinates in the mid-surface
of the shell and let be a linear coordinate in the thickness direction. Further, we assume
that , and vary between -1 and 1 on the respective faces of the element. Then, the
global coordinates of any point within the element can be obtained as follows

Figure 2.3: Typical degenerate shell element


x

x

x

n
1+
i

n
1
i

y = Ni (, ) yi + Ni (, ) yi (2.1)

i=1 2

2


z zi top i=1 zi bottom

where, Ni (, ) are shape functions for two dimensional element.

8

x
 n
i
x  n

y = Ni (, ) yi + Ni (, ) V3i (2.2)

i=1

i=1 2
z zi

or,



xi
1 xi

xi

yi
= yi + yi (2.3)
2

zi zi top zi bottom

where,



xi

xi

l3i

V3i = yi yi = ti m3i (2.4)







zi top zi bottom n3i



l3i
x
i x
i
1



m3i = yi yi (2.5)


ti




n3i zi top zi bottom

or,


x

x
 l3i

n i n
y = Ni (, ) yi + Ni (, ) ti m3i (2.6)

i=1

i=1 2

z zi n3i

2.3 Displacement eld

The displacement eld has now to be specied for the element. Taking into consideration
the two shell assumptions of the degeneration process, the displacement eld is described by
ve degrees of freedom of a normal viz. the three displacement components of its mid-node
and two rotations. The displacement of a point P on vector V3i consists of the displacement
of node i plus the displacement of P relative to i created by rotation of V3i through small
angles i and i . The relative displacement must be resolved into x, y and z components
before being added to displacement components of node i. For example, point P has the

9
x-direction displacement

   
ti ti
u p = u i i l2i + i l1i (2.7)
2 2

Hence, by shape function interpolation, displacements of an arbitrary point in the element


are


 
u 
n ui
t  
i i
v = Ni (, ) vi + v2i v1i (2.8)




2 i
i=1
w wi

where,



l1i

l1i

V1i V1i
v1i = = m1i and v1i = = m1i (2.9)
|V1i |
|V1i |

n1i n1i

Figure 2.4: Typical Figure 2.5: Orthogonal Figure 2.6: Displace-


node i and thickness vectors at node i and ment of a point P on
direction vector V3i nodal dof V3i due to nodal rota-
tions i and i

Vectors V1i and V2i are perpendicular to each other and to V3i . Thus V1i and V2i are tangent
to the element mid-surface at node i, but they are not required to have any perpendicular
orientation with respect to global coordinate directions. Here, V1i is dened normal to both
V3i and the global y-direction by writing the cross product V1i = j V3i , where j is unit
vector in y-direction. Then V2i = V3i V1i . The calculation V1i = j V3i fails if j and V3i are
parallel. In such case, the calculations are substituted as V2i = V3i i and V1i = V2i V3i .

10
2.4 Strain-displacement relationship

The strains in terms of displacement derivatives are expressed as follows



u,x



u,y







x
1 0 0 0 0 0 0 0 0


u,z







y
0 0 0 0 1 0 0 0 0

v,x

z 0 0 0 0 0 0 0 0 1
= v,y (2.10)





xy
0 1 0 1 0 0 0 0 0



v,z


yz
0 0 0 0 0 1 0 1 0







w,x

0 0 1 0 0 0 1 0 0


zx
w,y





w,z

or,

{}xyz = [H]{u}xyz (2.11)

Again, we have


u x y z u u




u














x



u
x
x y z u
= = [J] (2.12)

y y


u











x y z u u

z z
or,


u u










x
u u
1
= [J] (2.13)

y





u



u


z

where, [J] 1 is the inverse of jacobian matrix [J]. The jacobian matrix can be computed
from equation Eq.(2.6) as follows

11
n   n   n  
 ti Ni  ti Ni  ti Ni
xi + l3i yi + m3i zi + n3i
i=1 
2 i=1 2 i=1 2
  n   n  
n
ti Ni ti Ni ti Ni
[J] =
xi + l3i yi + m3i zi + n3i (2.14)

2 i=1 2 i=1 2
i=1
 ti
n  ti
n  ti
n

l3i Ni m3i Ni n3i Ni
i=1
2 i=1
2 i=1
2

The derivatives of displacements in the global x, y, z coordinates are related to the curvilinear
, , coordinates as follows




u,x
u,








u,y
u,









u,z




u,









v,x [J] 1
0 0


v,
1
v,y = 0 [J] 0 v, (2.15)







v,z 0 0 [J] 1
v,









w,x




w,









w,y
w,







w,z w,

or,

{u}xyz = [ J ]1 {u} (2.16)

From equation Eq.(2.8), we get



u,
Ni, 0 0 ti Ni, l2i /2 ti Ni, l1i /2




u,
Ni, 0 0 ti Ni, l2i /2 ti Ni, l1i /2





0

u,
0 0 ti Ni l2i /2 ti Ni l1i /2 ui







ti Ni, m1i /2
v, n 0 Ni, 0 ti Ni, m2i /2 vi

v, = 0 Ni, 0 ti Ni, m2i /2 ti Ni, m1i /2 wi (2.17)



i=1

v,
0 0 0 ti Ni m2i /2 ti Ni m1i /2 i









w,
0 0 Ni, ti Ni, n2i /2 ti Ni, n1i /2 i







w,
0 0 Ni, ti Ni, n2i /2 ti Ni, n1i /2



w, 0 0 0 ti Ni n2i /2 ti Ni n1i /2

12
or,


n
{u} = [Gi ]{di }e (2.18)
i=1

Combining the Eq.(2.11), Eq.(2.16) and Eq.(2.18), the strain-displacement relationship can
be written as follows


n
{}xyz = [B]{di }e = [Bi ]{di }e (2.19)
i=1

where,

[Bi ] = [H][ J ]1 [Gi ] (2.20)

2.5 Stress-strain relationship

The stress-strain relationship in the global x, y, z coordinates is given by

{}xyz = [D]{}xyz (2.21)

The stress-strain relationship in the local coordinates is given by

{}xyz = [D  ]{}xyz (2.22)

Matrix [D] is obtained from [D  ] by coordinate transformation as

[D] = [T ]T [D  ][T ] (2.23)

where,

13

l12 m21 n21 l1 m1 m1 n1 n1 l1

l22 m22 n22 l2 m2 m2 n2 n2 l2

l32 m23 n23
l3 m3 m3 n3 n3 l3
[T ] = (2.24)
2l1 l2 2m1 m2 2n1 n2 l1 m2 + l2 m1 m1 n2 + m2 n1 n1 l2 + n2 l1

2l l 2m m 2n n l m + l m m n + m n n l + n l
23 2 3 2 3 2 3 3 2 2 3 3 2 2 3 3 2
2l3 l1 2m3 m1 2n3 n1 l3 m1 + l1 m3 m3 n1 + m1 n3 n3 l1 + n1 l3

In numerical integration, this transformation must be carried out at each sampling point for
which and dier. Direction cosines needed in [T ] are direction cosines of vectors V1 , V2
and V3 at he sampling point. In turn, these vectors can be established by shape function
interpolation at nodal values


n 
n 
n
V1 = Ni V1i V2 = Ni V2i V3 = Ni V3i (2.25)
i=1 i=1 i=1

in which the Ni are evaluated at the sampling point in question.

2.6 Stiness matrix

The element stiness matrix with n numbers of nodes per element is given by

 1  1  1
[k] = [B]T [D][B]det[J]ddd (2.26)
1 1 1

Here, [k] is 5n 5n matrix, [B] is 6 5n matrix and [D] is 6 6 matrix. An alternate form
of the product [B]T [D][B] is [B  ]T [D  ][B  ], where [B  ] = [T ][B]. In this form one can omit
null terms associated with mid-surface normal strain 3 and write [E  ] as a 5 5 matrix and
[B  ] as a 5 5n matrix, thus making computations more ecient.

Eq.(2.25) is most economically integrated by assuming that the ratio of element thickness
to element span is suciently small that the dependence of [J] on can be ignored. Thus
[J] is evaluated on the mid-surface and [B] can be written as [B] = [B0 ] + [B1 ], where [B0 ]
and [B1 ] are independent of . Also let [D] be constant. Then, after explicit integration in
the thickness direction, Eq.(2.26) becomes

14
 1  
1 
T 2 T
[k] = 2[B0 ] [D][B0 ] + [B1 ] [D][B1 ] det[J]dd (2.27)
1 1 3

2.7 Consistent load vector

Consistent nodal load vector due to body force is given by


{Qi } = [Ni ]T {b}dV (2.28)
V

Consistent nodal load vector due to surface traction is given by



{Qi } = [Ni ]T {t}dS (2.29)
S

where, [Ni ] is shape function matrix, {b} is body force vector and {t} is surface traction
vector.


Ni 0 0 ti Ni l2i /2 ti Ni l1i /2

[Ni ] = 0 Ni 0 ti Ni m2i /2 ti Ni m1i /2 (2.30)
0 0 Ni ti Ni n2i /2 ti Ni n1i /2



bx

tx

{b} = by {t} = ty (2.31)





bz tz

2.7.1 Gravity load

Body force vector for gravity load is given by,




0

{b} = 0 (2.32)



15
where, is the weight density. Using Eq.(2.28), nodal load vector at node i is given by,


0




  
0

1 1 1
{Qi } = Ni det[J]ddd (2.33)
1 1 1



1
2 ti Ni n2i




1
t N n
2 i i 1i

Integrating explicitly along thickness direction,we get



0



 1 1 0


{Qi } = Ni det[J]dd (2.34)
1 1



0




0

2.7.2 Uniform vertical load

Surface traction vector for uniform vertical load is given by,




0
{t} = 0 (2.35)



pv

where, pv is the uniform vertical load. Using Eq.(2.29), nodal load vector at node i due to
uniform vertical load acting at the top surface = 1 is given by,


0




 
0

1 1
{Qi } = pv Ni J  dd (2.36)
1 1



1
2 ti Ni n2i




1
tNn
2 i i 1i

where,

J = (J12 J23 J22 J13 )2 + (J13 J21 J11 J23 )2 + (J11 J22 J21 J12 )2 (2.37)

where, Jij are elements of Jacobian matrix.

16
2.7.3 Uniform normal load

Surface traction vector for uniform normal load is given by,




lpn

{t} = mpn (2.38)



npn

where, pn is the uniform normal load. Using Eq.(2.29), nodal load vector at node i due to
uniform normal load acting at the top surface = 1 is given by,


0




 
0

1 1
{Qi } = pn Ni dd (2.39)
1 1



12 lti Ni l2i 12 mti Ni m2i 12 nti Ni n2i




1
lt N l + 12 mti Ni m1i + 12 nti Ni n1i
2 i i 1i

where,
l = J12 J23 J22 J13
m = J13 J21 J11 J23 (2.40)
n = J11 J22 J21 J12

2.8 Layered model

If the shell is built up from a series of layers having dierent materials, [D] is dierent from
layer to layer and is discontinuous function of . The integration has to be carried out by
splitting the limits through each layer for calculation of the stiness matrix. The layers
are numbered sequentially starting from bottom surface of the shell. For any kth layer, the
limits of integration will be between k1 and k , where k is given by,

2
NL
k = 1 + ti (2.41)
t i=1

where ti is thickness of individual layer, t is total thickness of the shell and NL is the total
number of layers.

Assuming the Jacobian [J] to remain constant along the thickness direction,[B] can be writ-
ten as [B] = [B0 ] + [B1 ], where [B0 ] and [B1 ] are independent of . The matrix [B]T [D][B]

17
can be written as

[B]T [D][B] = [B0 ]T [D][B0 ] + [B0 ]T [D][B1 ] + [B1]T [D][B0 ] + 2 [B1 ]T [D][B1 ] (2.42)

Now, using Eq.(2.41) and Eq.(2.42), Eq.(2.26) can be written as

NL 
   
1 1 k  
[k] = [B0 ]T [Dk ][B0 ] + [B0 ]T [Dk ][B1 ] + [B1 ]T [Dk ][B0 ]
k=1 1 1 k1

+ 2[B1 ]T [Dk ][B1 ]det[J] ddd (2.43)

where [Dk ] is the constitutive relation for individual layer.

Integrating Eq.(2.43) explicitly, we get

 1  1 
[k] = [B0 ]T [D1][B0 ] + [B0 ]T [D2][B1 ] + [B1 ]T [D2][B0 ] + [B1 ]T [D3][B1 ]det[J] dd
1 1
(2.44)

where


NL
[D1] = (k k1 )[Dk ] (2.45)
k=1


NL 2 2
k k1
[D2] = [Dk ] (2.46)
k=1
2


NL 3 3
k k1
[D3] = [Dk ] (2.47)
k=1
3

18
Chapter 3

Formulation for elastoplastic analysis


of shells

3.1 Constitutive relations

In the present study, a general constitutive relations are considered for elastoplatic analysis
of RC shell structures. Nonlinear isotropic and nonlinear kinematic hardening have been
employed which considers von-Mises yield criteria and associated ow rule. Deformation as
well as strain has been assumed to be small; hence linear strain displacement relations have
been used.

3.1.1 Decomposition of elastic and plastic strains

As strains are considered to be small, the total strain can be split into two parts viz. the
elastic part and the plastic part as follows

= e + p (3.1)

where e is the elastic part and p is the plastic part of strain tensor respectively.

19
3.1.2 Elastic stress-strain relationship

Elastic strain tensor e can be related to stress tensor as

= C : e (3.2)

where C is the elastic constitutive matrix given by

C = 2GI dev + Kg g (3.3)

where G and K are shear modulus and bulk modulus of undamaged materials, I dev fourth
order deviatoric identity tensor and g is second order identity tensor.

3.1.3 Initial yield function and subsequent yielding

The yield function for von-Mises material considering the mixed hardening rule can be ex-
pressed as !
3
f (S, D , r) = (S D ) : (S D ) y R(r) = 0 (3.4)
2
where, S D is called shifted or reduced stress, y is the yield stress, D and R(r) are the
kinematic and isotropic hardening parameters respectively.For a given material, satisfaction
of Eq.(3.4) represents the initial yield surface. With inelastic deformation, the hardening pa-
rameters change. Thus, the yield surface changes. These yield surfaces are called subsequent
yield surface.

3.1.4 Flow rule

Flow rule denes the evolution of plastic strain.In present study, the associative ow rule is
considered in which yield function is taken as the plastic potential.

f
p = = (3.5)

where f is the yield function, = f


is normal to the tangent plane of the yield surface
at a point on the surface and is a non-negative scalar factor of proportionality which is
nonzero only during plastic deformation.

20
3.1.5 Hardening rule

The hardening rule tells us the evolution of yield surfaces which are called subsequent yield
surfaces. Both isotropic and kinematic hardening has been taken into consideration in the
present study. This is called mixed hardening. In isotropic hardening, yield surface changes
its size, whereas yield surface translates for kinematic hardening.

Nonlinear isotropic hardening can be expressed as follows

R = hr n (3.6)

where h andn are isotropic hardening parameter and r is the eective plastic strain ep which
given by !
2 p p
p
r = e = : (3.7)
3
or
r = (3.8)

Nonlinear kinematic hardening can be expressed as follows

2
D = [ X D ] (3.9)
3

where X and are kinematic hardening parameters

3.1.6 Loading-unloading criteria

(a) Elastic loading :


f <0 (3.10)

(b) Elastic-plastic loading :


f = 0 and f > 0 (3.11)

(c) Neutral loading :


f = 0 and f = 0 (3.12)

(d) Elastic unloading :


f = 0 and f < 0 (3.13)

21
The loading-unloading criteria can be combined as Kuhn-Tucker optimality criteria as

f 0 0 f = 0 (3.14)

and the consistency condition


f = 0 (3.15)

3.2 Integration scheme: Return mapping algorithm

In this section, we shall discuss the integration of constitutive relations discussed in the pre-
vious section between time tn and tn+1 . Using Eq.(3.1), Eq.(3.2), Eq.(3.4),Eq.(3.5),Eq.(3.8)
and Eq.(3.9), we have the following relations

= C : ( p ) (3.16)
!
3
f (S, , r) =
D
(S D ) : (S D ) y R(r) = 0 (3.17)
2
p = (3.18)

r = (3.19)
2
D = [ X D ] (3.20)
3
We shall assume that all the quantities viz. n , en , pn , rn and D n are known at time tn and
strain eld n+1 = n + n is known at time tn+1 . Our objective is to nd all the quantities
viz. n+1 , en+1 , pn+1 , rn+1 and D
n+1 at time tn+1 taking into account of constitutive relations
given Eq.(3.16) to Eq.(3.20).

Now, using Backward Euler method, Eq.(3.16) to Eq.(3.20) are integrated numerically to
give
n+1 = C : (n+1 pn+1 ) (3.21)
!
3
f (S n+1 , D
n+1 , rn+1 ) = (S n+1 D
n+1 ) : (S n+1 n+1 ) y R(rn+1 ) = 0
D
(3.22)
2
pn+1 = pn + n+1 (3.23)

rn+1 = rn + (3.24)
2
n+1 = n + [ X n+1 n+1 ]
D D D
(3.25)
3

22
Using Eq.(3.21) to Eq.(3.23)

n+1 = C : (n+1 pn n+1 ) (3.26)

or
n+1 = C : (n + n+1 pn n+1 ) (3.27)

or
n+1 = C : (en + n+1 ) C : n+1 (3.28)

or
n+1 = Tn+1 C : n+1 (3.29)

where
Tn+1 = C : (en + n+1 ) = n + C : n+1 (3.30)

Eq.(3.29) and Eq.(3.30)can be written in deviatoric form as

S n+1 = S Tn+1 2G n+1 (3.31)

S Tn+1 = S n + 2GI D : n+1 (3.32)

By enforcing consistency at tn+1 , we can write from Eq.(3.22)


!
3
(S n+1 D
n+1 ) : (S n+1 n+1 ) = y + R(rn+1 )
D
(3.33)
2

Expression for n+1 can be written as

3 S n+1 D 3 S n+1 D
n+1 = " n+1
= n+1
(3.34)
2 3 (S D : (S n+1 D 2 y + R(rn+1 )
2 n+1 n+1 ) n+1 )

Substituting n+1 from Eq.(3.34) in Eq.(3.25)

2 3 S n+1 D
D
n+1 = D
n + [ X n+1
D
n+1 ] (3.35)
3 2 y + R(rn+1 )

Rearranging Eq.(3.35) can be written as

D D
n+1 = P1 ()n + P2 ()S n+1 (3.36)

23
where,
y + R(rn+1 )
P1 () = (3.37)
X + (1 + ){y + R(rn+1 )}
X
P2 () = (3.38)
X + (1 + ){y + R(rn+1 )}
Substituting n+1 from Eq.(3.34) in Eq.(3.31)

S n+1 D
S n+1 = S Tn+1 3G n+1
(3.39)
y + R(rn+1 )

Substituting D
n+1 from Eq.(3.36) in Eq.(3.39)

3G
S n+1 = S Tn+1 n + {1 P2 ()}S n+1 ]
[P1 ()D (3.40)
y + R(rn+1 )

Rearranging Eq.(3.40) can be written as

S n+1 = P3 ()S Tn+1 + P4 ()D


n (3.41)

where,
y + R(rn+1 )
P3 () = (3.42)
y + R(rn+1 ) + 3G(1 P2 )
3GP1
P4 () = (3.43)
y + R(rn+1 ) + 3G(1 P2 )
Now let us evaluate S n+1 D
n+1 by substituting n+1 from Eq.(3.36)
D

S n+1 D
n+1 = (1 P2 )S n+1 P1 n
D
(3.44)

Substituting S n+1 from Eq.(3.41) in Eq.(3.45)

S n+1 D
n+1 = (1 P2 )P3 S n+1 + {(1 P2 )P4 P1 }n
T D
(3.45)

or
S n+1 D
n+1 = {(1 P2 )P4 P1 }n + (1 P2 )P3 S n+1
D T
(3.46)

or
S n+1 D
n+1 = P5 n + P6 S n+1
D T
(3.47)

where
P5 = (1 P2 )P4 P1 (3.48)

24
P6 = (1 P2 )P3 (3.49)

Now substituting S n+1 D


n+1 in Eq.(3.22)

!
3
(P5 D
n + P6 S n+1 ) : (P5 n + P6 S n+1 y R(rn+1 ) = 0
T D T
(3.50)
2

The above expression is only a function of . Solving the equation by Newton-Raphson


method, we can get the value of .

3.2.1 Steps in return mapping algorithm

1. Calculate S Tn+1 = S n + 2GI D : n+1

2. If f (S Tn+1 , D
n , rn ) = 0, then

S n+1 = S n , n+1 = n , en+1 = en , pn+1 = pn , rn+1 = rn and D


n+1 = n
D

3. If f (S Tn+1 , D
n , rn ) > 0, then express fn+1 = f (S n+1 , n+1 , rn+1 ).
T D

The expression is only a function of . Solve the equation fn+1 = 0 using Newton-
Raphson method to get .

4. Calculate n+1 = tr
n+1 C : n+1

5. Calculate pn+1 = pn + n+1

6. Calculate rn+1 = rn +
2
n+1 = n + [ 3 X n+1 n+1 ]
7. Calculate D D D

3.3 Consistent tangent modulus

It is necessary to have consistency between the tangent modulus and the integration al-
gorithms employed in the solution of incremental problem to preserve the quadratic rate of
asymptotic convergence of the iterative solution schemes based on Newton-Raphson method.
It is called consistent tangent modulus which is derived imposing aforementioned consistency
requirement. The consistent tangent modulus C consist is given as

= C consist : (3.51)

25
can be split up into its deviatoric and spherical parts as

= S + K(g g) : (3.52)

The explicit expression for consistent tangent modulus in Eq.(3.51) will be derived by exact
linearization of the update equations described in Section 3.2. We can see that derivation of
consistent tangent modulus is simplied to obtain S n+1 in terms of n+1 .

Eq.(3.41) can be written as


1 P4
S Tn+1 = S n+1 D (3.53)
P3 P3 n
Dierentiating Eq.(3.53)

1 1 P3 1 P4 P3 D
S Tn+1 = S n+1 + [ 2 S n+1 2 {P3 P4 }n ] (3.54)
P3 P3 P3

or
M11 S n+1 + M12 = S Tn+1 (3.55)

where
1
M11 = (3.56)
P3
1 P3 1 P4 P3 D
M12 = [ 2
S n+1 2 {P3 P4 }n ] (3.57)
P3 P3
Substituting D
n+1 from Eq.(3.36) in Eq.(3.22)

!
3
((1 P2 )S n+1 P1 D
n ) : ((1 P2 )S n+1 P1 n ) y R(rn+1 ) = 0
D (3.58)
2

Dierentiating

3 (1 P2 )S n+1 + ( P 2
S n+1 P 1
D
n)
"
y R(rn+1 ) = 0 (3.59)
2 3 ((1 P )S P D ) : ((1 P )S P D)
2 2 n+1 1 n 2 n+1 1 n

26
Chapter 4

Results and discussions

4.1 General

In chapter 2, we have seen the nite element formulation of shells based on degenerated shell
element. The layered theory formulation for analysis of laminated shells is also presented. In
this chapter, the above formulation has been implemented using C programming for linear
analysis and some benchmark examples have been solved. The algorithm for programming
of FE formulation is presented in Figure 4.1.

4.2 Numerical examples

4.2.1 Example 1

Figure 4.2 shows a thin cylidrical shell which has no edge beam and is supported by a rigid
diaphragm. Theoritical solution of the problem is available in Sordelis and Lo [78]. The
problem is solved by 4 4 and 8 8 mesh using both 2 2 and 2 2 Gauss quadrature and
the computed results are shown in Figure 4.3 and Figure 4.4.

Thickness, t = 0.25, Radius, R = 25, Length, L = 50,

E = 4.32 108 , = 0, Self weight = 90 per unit area,

u = 0, w = 0, = 0, = 0 along curved edges

27
Figure 4.1: Programming algorithm for FE formulation

28
Figure 4.2: Scordelis-Lo Roof

Figure 4.3: Vertical deection at midspan(4 4 mesh)

29
Figure 4.4: Vertical deection at midspan(8 8 mesh)

4.2.2 Example 2

A parabolic hyperboloid shell clamped along all the boundary is shown in Figure 4.5. The-
oritical solution of the problem is available in Chetty and Tottenham [18]. 4 4 and 8 8
meshing is used to solve the problem considering both 2 2 and 2 2 Gauss quadrature and
the computed results are shown in Figure 4.6 and Figure 4.7.

Thickness, t = 0.25, a = 6.46, b = 6.46, c = 1.304,

E = 5 105 , = 0.39, Vertical loading = 1 per unit area,

Boundary condition : clamped along all the edges.

4.2.3 Results and discussions

In Example 1, when the shell is analyzed using 4 4 meshing and 3 point gauss quadrature,
the solution obtained showed locking. When reduced integration is done using 2 2 gauss
qudrature, the results obtained are in fairly good agreement with the theoretical solution.
At this point, we hope to get more accurate results as we go for ne meshing. However,
when 8 8 meshing is used, the results are coming very arbitrary in case of both full and
reduced integration. In case of Example 2, the similar problems are obtained. In case of
4 4 meshing, both full and reduced integration are giving fairly good results. However,
when 8 8 meshing is used, the results are again coming arbitrary.

30
Figure 4.5: Parabolic hyperboloid shell

Figure 4.6: Vertical deection along y = 0(4 4 mesh)

4.3 Future work

From above results and discussions, we have seen that the program is giving fairly good
results for coarse mesh. However, for ne meshing, it is giving arbitrary results. Thus, the
main task for the future is to improve the programming so that it works for n meshing also.

31
Figure 4.7: Vertical deection along y = 0(8 8 mesh)

Followings are the highlights of the future work to be carried out in the 2nd phase of the
project,

To make the program robust so that the results converges as we go for ner meshing
for linear analysis of shell.

To incorporate the reduced and selective integration schemes into the programming
along with spurious mode stabilization techniques to eliminate the problem of locking
in case of thin shells.

To implement the layered model into the programming for analysis of laminated shell
using explicit through-thickness integration schemes.

To implement the elastoplastic constitutive relations for both concrete and steel con-
sidering both isotropic and kinamatic hardening for non-linear analysis of RC shell
structures.

32
Bibliography

[1] S. Ahmad, B.M. Irons, and O.C. Zienkiewicz. Analysis of thick and thin shell structures
by curved nite elements. Int. J. Numer. Meth. Engng, 2:419451, 1970.

[2] W. Altmann and F. Iguti. A thin cylindrical shell nite element based on a mixed
formulation. Comput. Struct., 6:149155, 1976.

[3] J.H. Argyris, P.C. Dunne, G.A. Malejannakis, and E. Schelke. A simple triangular facet
shell element with application to linear and nonlinear equilibrium and elastic stability
problems. Comput. Meth. appl. Mech. Engng, 10:371403, 1977.

[4] J.H. Argyris and D. Scharpf. The sheba family of shell element for matrix displacement
method. Aeronaut. Jnl., 72:873883, 1968.

[5] DG. Ashwell and A.B. Sabir. A new cylindrical shell nite element based on simple
independent strain functions. Int. J. mech. Sci., 14:171183, 1972.

[6] K.J. Bathe and S. Bolourchi. A geometric and material nonlinear plate and shell element.
Comput. Struct., 11:2348, 1980.

[7] K.J. Bathe and E.N. Dvorkin. A formulation of general shell elements - the use of mixed
interpolation of tensorial components. Int. J. Numer. Meth. Engng, 22:697722, 1986.

[8] K.J. Bathe and L.W. Ho. A simple and eective element for analysis of general shell
structures. Comput. Struct., 13:673681, 1981.

[9] J.L. Batoz, A. Cattopadhyay, and G. Dhatt. Finite element large deection analysis of
shallow shells. Int. J. Numer. Meth. Engng, 10:3958, 1976.

[10] J.L. Batoz and G. Dhatt. Development of two simple shell elements. AIAA Jnl, 10:237
238, 1972.

33
[11] T. Belytschko, W.K. Liu, J.S.J Ong, and D. Lam. Implementation and application of a
9-node lagrange shell element with spurious mode control. Comput. Struct., 20:121128,
1985.

[12] T. Belytschko, B.L. Wong, and H. Stolarski. Assumed strain stabilization procedure for
the 9-node lagrangian shell elements. Int. J. Numer. Meth. Engng, 28:385414, 1989.

[13] F.K. Bonger, R.L. Fox, and L.A. Schmit. A cylindrical shell discrete element. AIAA
Jnl, 5:745750, 1967.

[14] N. Buchter, E. Ramm, and D. Roehl. Three-dimensional extension of non-linear shell


formulation based on the enhanced assumed strain concept. Int. J. Numer. Meth. Engng,
37:25512568, 1994.

[15] G. Cantin. Rigid body motions in curved nite elements. AIAA Jnl, 8:12521255, 1970.

[16] G. Cantin and R.W. Clough. A curved, cylindrical shell nite element. AIAA Jnl,
6:10571062, 1968.

[17] W.C. Chao and J.N. Reddy. Analysis of laminated composite shells using a degenerated
3-d element. Int. J. Numer. Meth. Engng, 20:19912007, 1984.

[18] S.M.K. Chetty and H. Tottenham. An investigation into the bending analysis of hyper-
bolic paraboloid shells. Ind. Concr. Jnl., 38:248258, 1964.

[19] Y.K. Cheung and W. Chen. Generalized hybrid degenerated elements for plates and
shells. Comput. Struct., 36:279290, 1990.

[20] J.D. Chieslar and A. Ghali. A hybrid strain technique for nite element anlysis of plates
and shells. Comput. Struct., 24:749765, 1986.

[21] C.K. Choi and W.C. Schnobrich. Use of nonconforming modes in nite element analysis
of shells. J. Engng Mech. Div., ASCE, 101:447464, 1975.

[22] R.W. Clough and C.P. Johnson. A nite element approximation for the analyis of thin
shells. Int. J. Solids Struct., 4:4360, 1968.

[23] J.J. Connor and C.A. Brebbia. A stiness matrix for a shallow rectangular shell element.
J. Engng Mech. Div., ASCE, 93:4365, 1967.

[24] I. Cormeau. Elastoplastic thick shell analysis by viscoplastic solid nite elements. Int.
J. Numer. Meth. Engng, 12:203227, 1978.

34
[25] G. Cowper, G.M. Lindberg, and M.D. Olson. A shallow shell nite element of triangular
shape. Int. J. Solids Struct., 6:11331156, 1970.

[26] D.J. Dawe. A nite deection analysis of shallow arches by the discrete element method.
Int. J. Numer. Meth. Engng, 3:529552, 1971.

[27] D.J. Dawe. Shell analyis using a facet element. J. Strain Anal., 6:266270, 1972.

[28] D.J. Dawe. High-order triangular nite element for shell analysis. Int. J. Solids Struct.,
11:10971110, 1975.

[29] G. Dhatt, L. Marcotte, and Y. Matte. A new triangular discrete kirchho plate/shell
element. Int. J. Numer. Meth. Engng, 23:453470, 1986.

[30] G.S. Dhatt. An ecient triangular shell element. AIAA Jnl, 8:21002102, 1970.

[31] R. Dunger and R.T Severn. Triangular nite elements of variable thickness and their
application to plate and shell problems. J. Strain Anal., 4:1021, 1969.

[32] G. Dupuis and J.J. Goel. A curved nite element for thin elastic shells. Int. J. Solids
Struct., 6:14131428, 1970.

[33] G. Edwards and J.J. Webster. Hybrid cylindrical shell nite elements. Finite Elements
for Thin Shells and Curved Members. In: R.H. Gallagher and D.G. Ashwell, Editor,
John Wiley, pages 171195, 1976.

[34] G.A. Fonder and R.W. Clough. Explicit addition of rigid body motions in curved nite
elements. AIAA Jnl, 11:305312, 1973.

[35] R.D. Hanshell, B.K. Neale, and G.B. Warburton. A new hybrid cylindrical shell element.
J. Sound Vibr., 16:519531, 1971.

[36] R. Hauptmann and K. Schweizerhof. A systematic development of solid-shell element


formulations for linear and non-linear analyses employing only displacement degrees of
freedom. Int. J. Numer. Meth. Engng, 42:4969, 1998.

[37] G. Horrigmoe and P.G. Bergan. Nonlinear analysis of free-form shells by at nite
elements. Comput. Meth. appl. Mech. Engng, 16:1135, 1978.

[38] K.M. Hsiao. Nonlinear analysis of general shell structures by at triangular shell ele-
ment. Comput. Struct., 25:665675, 1987.

35
[39] K.M. Hsiao and Y.R. Chen. Nonlinear analysis of shell structures by degenerated
isoparametric shell element. Comput. Struct., 31:427438, 1989.

[40] H.C. Huang and E.Hintan. A new nine node degenerated shell element with enhanced
membrane and shear interpolation. Int. J. Numer. Meth. Engng, 22:7392, 1986.

[41] T.J.R. Hughes and W.K. Liu. Nonlinear nite element analysis of shells: Part i. three-
dimensional shells. Comput. Meth. appl. Mech. Engng, 26:331362, 1981.

[42] T.R.J Hughes and E. Cornoy. Nonlinear nite element shell formulation accounting for
large membrane strains. Comput. Meth. appl. Mech. Engng, 39:6982, 1983.

[43] S. Idelsohn. On the use of deep, shallow or at shell nite elements for the analysis of
thin shell structures. Comput. Meth. appl. Mech. Engng, 26:321330, 1981.

[44] J. Jang and P.M. Pinsky. An assumed covariant strain based 9-node shell element. Int.
J. Numer. Meth. Engng, 24:23892411, 1987.

[45] D. Kang and T.H.H. Pian. A 20-d.o.f. hybrid stress general shell element. Comput.
Struct., 30:789794, 1988.

[46] W. Kanok-Nukulchai. A simple and ecient nite element for general shell analysis.
Int. J. Numer. Meth. Engng, 14:179200, 1979.

[47] H. Kebari and A.C. Cassel. A stabilized 9-node non-linear shell element. Int. J. Numer.
Meth. Engng, 35:3791, 1992.

[48] W.P. Prema Kumar and R. Palaninathan. Finite element analysis of laminated shells
with exact through-thickness integration. Comput. Struct., 63:173184, 1997.

[49] W.P. Prema Kumar and R. Palaninathan. Explicit through-thickness integration


schemes for geometric nonlinear analysis of laminated composite shells. Finite Elements
Anal. Des., 32:235256, 1999.

[50] H.V. Lakshinarayana and K. Kailash. A shear deformable curved shell element of
quadrilateral shape. Comput. Struct., 33:9871001, 1989.

[51] S.J. Lee and W. Kanok-Nukulchai. A nine-node assumed strain nite element for large-
deformation analysis of laminated shells. Int. J. Numer. Meth. Engng, 42:777798,
1998.

[52] R. Levy and E. Gal. Geometrically nonlinear three-noded at triangular element. Com-
put. Struct., 79:23492355, 2001.

36
[53] M.L. Liu and C.W.S. To. Hybrid strain based three node at triangular shell elements
-i. nonlinear theory and incremental formulation. Comput. Struct., 54:10311056, 1995.

[54] M.L. Liu and C.W.S. To. Hybrid strain based three node at triangular shell elements
-ii. numerical investigation of nonlinear problem. Comput. Struct., 54:10561076, 1995.

[55] M.L. Liu and C.W.S. To. A further study of hybrid strain-based three-node at trian-
gular shell elements. Finite Elements Anal. Des., 31:135152, 1998.

[56] R.H. MacNeal. Derivation of element stiness matrices by assumed strain distributions.
Nucl. Eng. Des., 70:312, 1982.

[57] S.T. Mau, T.H.H. Pian, and P.Tong. Vibration analysis of laminated paltes anf shells
by a hybrid stress element. AIAA Jnl., 11:14501452, 1973.

[58] J.L. Meek and H.S. Tan. A faceted shell element with loof nodes. Int. J. Numer. Meth.
Engng, 23:4967, 1986.

[59] R.V. Milford and W.C. Schnobrich. Degenerated isoparametric nite elements using
explicit integration. Int. J. Numer. Meth. Engng, 23:133154, 1986.

[60] S.S. Murthy and R.H. Gallagher. Anisotropic cylindrical shell element based on discrete
kirchho theory. Int. J. Numer. Meth. Engng, 19:18051823, 1983.

[61] J.C. Nagtegaal and J.G. Slater. A simple non-compatible thin shell element based on
discrete kirchho theory. Non-linear Finite Element Analysis of Plates and Shells. In:
R.J.R. Hughes, A. Pifko and A. Jay, Editor, ASME, AMD, 48:167192, 1981.

[62] J. Too O.C. Zienkiewicz and R.L. Taylor. Reduced integrat1on technique in general
analysis of plates and shells. Int. J. Numer. Meth. Engng, 3:275290, 1971.

[63] D.R.J. Owen and J.A. Figueiras. Anisotropic elasto-plastic nite element analysis of
plates and shells. Int. J. Numer. Meth. Engng, 19:541566, 1983.

[64] S. Panda and R. Natarajan. Analysis of laminated composite shell structures by nite
element method. Comput. Struct., 14:225230, 1981.

[65] H. Parisch. Geometrical nonlinear analysis of shells. Comput. Meth. appl. Mech. Engng,
14:159178, 1978.

[66] H. Parisch. Large displacements of shells including material nonlinearities. Comput.


Meth. appl. Mech. Engng, 27:183214, 1981.

37
[67] H. Parisch. An investigation of a nite rotation four-node shell element. Int. J. Numer.
Meth. Engng, 31:127150, 1991.

[68] K.C. Park and G.M. Stanely. A curved c0 shell element based on assumed natural-
coordinate strains. J. appl. Mech., ASME, 53:278290, 1986.

[69] S.F. Pawsey and R.W. Clough. Improved numerical integration of thick shell nite
elements. Int. J. Numer. Meth. Engng, 3:575586, 1971.

[70] D.A. Pecknold and W.C. Schnobrich. Finite element analysis of skewed shallow shells.
J. struct. Div., ASCE, 95:715744, 1969.

[71] P.M. Pinsky and J. Jang. Elastoplastic shell element based on assumed covariant strain
interpolations. J. Engrg. Mech., 114:10451062, 1988.

[72] C.A. Prato. Shell nite element method via reissners principle. Int. J. Solids Struct.,
5:11191133, 1969.

[73] E. Ramm and J.M. Sattele. Elasto-plastic large deformation shell analysis using de-
generated elements. Nonlinear Finite Element analysis of Plates and Shells. In: T.J.R.
Hughes, Editor, AMD, 48:265282, 1981.

[74] J.J. Rhiu and S.W. Lee. A new ecient mixed formulation for thin shell nite element
models. Int. J. Numer. Meth. Engng, 24:581604, 1987.

[75] F.N. Rigby, J.J. Webster, and R.D. Henshell. Hybrid and hellinger-reissner plate and
shell nite elements. Hybrid and Mixed Finite Element Methods. In: S.N. Atluri, R.H.
Gallagher and O.C. Zienkiewicz, Editor, John Wiley, pages 7392, 1983.

[76] A.B. Sabir and A.C. Lock. A curved cylindrical shell nite element. Int. J. mech. Sci.,
14:125135, 1972.

[77] A.F. Saleeb, T.Y. Chang, and W. Graf. A quadrilateral shell element using a mixed
formulation. Comput. Struct., 26:787803, 1987.

[78] A. C. Scordelis and K. S. Lo. Computer analysis of cylindrical shells. J. Amer. Concr.
Inst., 61:539562, 1964.

[79] J.C. Simo and J.G. Kennedy. On a stress resultant geometrically exact shell model.
part v. nonlinear plasticity: formulation and integration algorithm. Comput. Meth.
appl. Mech. Engng, 96:133171hughes2, 1992.

38
[80] K.S. Surana. Geometrically nonlinear formulation for the curved shell element. Int. J.
Numer. Meth. Engng, 19:581615, 1983.

[81] B. Tabarrok and N. Gass. Vibration of cylindrical shells by hybrid nite element method.
AIAA Jnl, 10:15531554, 1972.

[82] C. Tahiani and L. Lachance. Linear and nonlinear analysis of thin shallow shells by
mixed nite elements. Comput. Struct., 5:167177, 1975.

[83] D. Talaslidis and W. Wunderlich. Static and dynamic analysis of kirchho shells based
on mixed nite element formulation. Comput. Struct., 10:239249, 1979.

[84] M.A. Talha. A theoritically improved and easily implemented version of the ahmad
thick shell element. Int. J. Numer. Meth. Engng, 14:125142, 1979.

[85] C.W.S. To and M.L. Liu. Hybrid strain based three-node at triangular shell elements.
Finite Elements Anal. Des., 17:169203, 1994.

[86] S. Utku. Stiness matrices for thin triangular elements of non-zero gaussian curvature.
AIAA Jnl, 5:16591667, 1967.

[87] S. Vlachoutsis. Explicit integration for three-dimensional degenerated shell nite ele-
ments. Int. J. Numer. Meth. Engng, 29:861880, 1990.

[88] G.Z. Voyiadjis and Pawel Woelke. General non-linear nite element analysis oft thick
plates and shells. Int. J. Solids Struct., 43:22092242, 2006.

[89] G.A. Wempner, J.T. Oden, and D.A. Kross. Finite element analysis of thin shells. J.
Engng Mech. Div., ASCE, 94:12731294, 1968.

[90] E.L. Wilson, R.L. Taylor, W.P. Doherty, and J. Ghaboussi. Incompatible displacement
modes. Numerical and computer methods in structural mechanics. In: S.T. Fenves et
al., Editor, academic press, 1973.

[91] T.W. Yang. High order rectangular shallow nite element. J. Engng Mech. Div., ASCE,
99:157181, 1973.

[92] S.M. Yunus, P.C. Kohnke, and S. Saigal. An ecient through-thickness integration
scheme in an unlimited layer doubly curved isoparametric composite shell element. Int.
J. Numer. Meth. Engng, 28:27772793, 1989.

39

Você também pode gostar