ELASTO-PLASTIC FINITE ELEMENT ANALYSIS

WITH SPECIAL REFERENCE TO REINFORCED CONCRETE

by

JOHN HON-SHING LAU

B . S c , National Taiwan University/ 1970

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF

THE REQUIREMENTS FOR THE DEGREE OF

MASTER OF APPLIED SCIENCE

In the Department

of

Civil Engineering

We accept this thesis as conforming to the

required standard

THE UNIVERSITY OF BRITISH COLUMBIA

November, 1973
In presenting this thesis in partial fulfilment of the requirements for

an advanced degree at the University of British Columbia, I agree that

the Library shall make i t freely available for reference and study.

I further agree that permission for extensive copying of this thesis

for scholarly purposes may be granted by the Head of my Department or

by his representatives. It is understood that copying or publication

of this thesis for financial gain shall not be allowed without my

written permission.

Department of Civil Engineering

The University of British Columbia

Vancouver 8, Canada

Date November 1 0 , 1973

 -1-

ABSTRACT

A numerical procedure is presented for the solution

of e l a s t o - p l a s t i c problems by means of the f i n i t e element

approach. The incremental constitutive relationship from

the Prandtl-Reuss equation is used in conjunction with the

Von Mises yield c r i t e r i o n , the continuum being divided

into triangular elements.

The Modified Newton-Raphson method is employed to

solve the nonlinear incremental equilibrium equation.

Numerical examples are studied and compared with the

experimental and theoretical results in the literature.

The f i n i t e element program is also extended to

nonlinear stress analysis of reinforced concrete structures.

 II

TABLE 0£ CONTENTS

Chapter Page

One Introductton 1

Two BasIc Def i n i t l o n s and Theories for Elasticity 5

and Plasti clty

I. Analysts of Stress 5

II. Analysis of Strain lk

III. Field Equations In Elasticity 18

IV. Field Equations In Plasticity 20

Three FlnIte Element Method 31

I. Variational Principles 32

II. Formulation of Finite Element Method by the 36

Principle of Minimum Potential Energy

Four Formulation of Elasto-plastlc Problems kO

I. General Considerations kO

II. Generalized Incremental Equilibrium Equation k2

III. Constitutive Equations: Elasto-Plastlc Matrix kk

IV. Solution of Plasticity Equations 51

V. Procedures of Calculation 56

VI. Examples 62

Five NonlInear Stress Analys1s of Relnforced Concrete 78

I. General Considerations 78

II. Structural Idealization 79

III. Fallure Conditlon 80

 111

CHAPTER Page

Five IV. Calculation Procedure 81

V. Numerical Example 8U

Six ConclusIons 98

Appendix 100

Bibliography 10U
 TABLE OF F I G U R E S

F IGURE

2-1 INTERNAL FORCES SYSTEM

2-2 S T R E S S V E C T O R S AND P O S I T I V E STRESS COMPONENTS

2-3 S T R E S S V E C T O R S ON A TETRAHEDRON

2-U ROTATION OF C O O R D I N A T E S

2-5 D I S P L A C E M E N T OF N E I G H B O R I N G POINTS

2-6 TRESCA Y I E L D SURFACE IN P R I N C I P A L STRESS SPACE

2-7 VON M I S E S YIELD SURFACE IN P R I N C I P A L STRESS SPACE

k-1 THE I N C R E M E N T A L LOADING PROCESS

k-Z A COMBINATION OF THE LOAD INCREMENT AND

NEWTON-RAPHSON (OR M O D I F I E D NEWTON-RAPHSON)

METHOD

k-3 THE I L L U S T R A T I O N S OF E Q S . U-25), U-26), U-27),

AND U-28)

k-ka NEWTON-RAPHSON METHOD

k-kb M O D I F I E D NEWTON-RAPHSON METHOD

«*-5 CONSTANT S T R A I N ELEMENT

U-6 UNIFORMLY LOADED PERFORATED STRIP

U-7 UNIAXIAL S T R E S S - S T R A I N CURVE

l*-8 END LOAD vs. MAX S T R A I N FOR P E R F O R A T E D STRIP

k-9a EXPERIMANTAL RESULTS

l»-9b FINITE ELEMENT R E S U L T S

 V

FIGURE

4-10 EQUIVALENT STRESS v s . EQUIVALENT STRAIN IN

THE E L E M E N T WHICH Y I E L D FIRST

4-11 SHEAR WALL S U B J E C T E D TO LATERAL LOAD

4-12 UNIAXIAL S T R E S S - S T R A I N CURVE

4-13 D E V E L O P M E N T OF PLASTIC ZONES FOR DIFFERENT

VALUES OF

4-14a NORMAL S T R E S S D I S T R I B U T I O N AT HIGHEST LOAD

4-14b SHEAR S T R E S S D I S T R I B U T I O N AT H I G H E S T LOAD

4-15a SHEAR S T R E S S D I S T R I B U T I O N AT H I G H E S T LOAD

4-16 RIGHT-ANGLE NOTCH: E L A S T I C - P E R F E C T L Y - P L A S T I C

4-17 END' LOAD v s . MAX. STRAIN FOR NOTCHED TENSION

SPECIMEN. ELASTIC-PERFECTLY-PLASTICITY

4-18a D. ALLEN &R. SOUTHWELL^ SOLUTION

4-18b FINITE ELEMENT SOLUTION

 v!

LIST OF TABLES

TABLE 2-1 MAXIMUM SHEAR S T R E S S AND PRINCIPAL DIRECTION

TABLE 2-2 TRESCA YIELD CONDITION

o
 v? I

Acknowledgement

In 1972 the writer received from his supervisor,

professor N.D. Nathan, two f u n d a m e n t a l courses In s o l Id

mechanics, namely, Energy Theorems and The Finite element

Method, which gave rise to the w r i t e r ' s interest In solid

mechanics. The writer wishes to acknowledge the help that

It has been: all of chapter three and most of chapter four

and the appendix are b a s e d on a study of his lectures. The

writer i s a l s o most grateful for his guidance, his reading

and checking of the entire thesis. Finally, the writer

wishes to register his gratitude to his wife, Teresa, for

typing the whole thesis.

 CHAPTER ONE

INTRODUCTION

There are two types of nonlinear Ity In solid mechanics,

namely, geometrical and physical nonllnearity (1). Geometrical

nonlinearity deals with finite deformations, while physical

nonllnearity deals with nonlinear stress-strain relations of

the body. These two types of nonlinearity are independent of

each other. Thus it follows that there are several types of

problems In solid mechanics:

(1) those linear both physically and geometrically (linear

problems, elasticity),

(2) those linear physically but nonlinear geometrically (non-

linear problems, elasticity),

(3) those nonlinear physically but linear geometrically (non-

linear problems, e.g. plastIclty wlth Infinitesimal defor-

mations),

(4) those nonlinear both physically and geometrically (non-

linear problems, e.g. p l a s t i c i t y with finite deformations). °

Nonlinear theories lead to nonlinear governing equations

which immediately render classical methods of analysis inappli-

cable. Despite recent efforts on nonlinear behavior, only a few

exact solutions to s p e c i f i c problems can be found; and these are

 - 2 -

the most simple geometric shapes and boundary conditions. On the

other hand, many n o n l i n e a r problems can be solved by numerical

techniques which lead to approximate solutions.

The finite element method is one o f these numerical tech-

niques and has been found to be a powerful approach to stress

analysis problems (2). Although early development of the method

was primarily concerned with linear systems, the method has been

extended to nonlinear problems by many researchers. The finite

element formulation of geometrically nonlinear problems of elas-

ticity has been generalized by N.D. Nathan (3). The study of

physically nonlinear problems by the finite element method Is

the major study of present work.

The finite element formulation of p l a s t i c i t y was first

presented by R.H. Gallagher and his colleagues (4). They followed

the suggestions given by Mendel son and Manson (5), who proposed

the so called Initial (or thermal) s t r a i n method which is based

on the Idea of modifying the equilibrium equations of elasticity

to compensate for the fact that the plastic strains do not cause

any change in stress. Further developments in this category have

been made by J.H. Argyris (6), L.H. Percy (7) and W.R. Jensen (8).

This approach, however, connot be u s e d for perfectly plastic

materials (9) or for a very small degree of hardening.

A few years later, G. Pope (10) Introduced a different

method known as the tangent modulus method, which is based on

the Incremental theory of p l a s t i c i t y , and solved the nonlinear

 -3-

problem as a series of piece-wise linear systems. Subsquent

contributions was made by T.L. Swedlow (11), S.F. Reys (12).

Two years later, P.V. Marcal and I.P. King introduced a part-

ial stiffness concept and make ah elasto-plastic analysis of plane

stress and plane strain problems as w e l l as problems of axi-

symmetrically loaded bodies of revolution (13). At the same

time, Y. Yamada (IU) obtained an explicit expression of the

incremental stress-strain relationship for the Prandtl-Reuss

equations associated with the Von M I s e s yield criterion.

Two years later, a general formulation of plasticity

was presented by O . C . Zienkiewic and his colleagues (15),

wherein an " initial stress " approach is proposed. This app-

r o a c h may be c o n s i d e r e d to save computing time, because a

constant linear elastic matrix is used throughout the process.

During the same year, Y. Yamada (16) extended his explicit

elasto-plastic matrix to orthrotropic materials on the basis

of Hill's constitutive equations (17). More recently, G.C. Nayak

and O.C. Zienkiewic (18) have generalized problems in plas-

ticity for various constitutive relations including strain

softening, and have renamed the " initial stress method " the

" residual force method ".

The present work, using an improved finite-element

representation, deals with a detailed study of two-dimensional

elasto-plastic stress analysis problems with infinitesimal

deformation. The computer programs for elastic analysis with

the well-known constant strain triangular element were

modified to include the elastoplastic analysis for the pre-

sent study.
 -5-

CHAPTER TWO

BASIC DEFINITIONS AND T H E O R I E S OF ELAST ICITY AND PLASTICITY

The theories of elasticity and p l a s t i c i t y deal with the

systematic study of the state variables, stress, strain, and

displacement in a solid body under the Influence of external

forces or prescribed displacements or both. In this chapter

the basic equations of these theories will be reviewed for refer-

ence purposes and to establtsh the notation to be used.

For convenience rectangular cartesian coordinates X ( K, Xa

; 4 Xj)

will be employed for defining the three-dimensional space con-

taining the body.

I. Analvs i s of Stress

When a s o l i d is under the a c t i o n of external forces, it

undergoes deformation and the effect of forces is transmitted

throughout the body by Interaction of material elements, giving

rise to an internal force-stress field.

a. Stress Vector

Inside the body, isolate a surface of area A containing a

point O . Fig.2-1. Due to external forces there will be some forces

distributed over area A . These distributed forces will have a

resultant force P and a resultant moment H a t the point to 0 .

 -6-

FIG.2-1 INTERNAL FORCES SYSTEM

After the concept of Love, the following limits are assumed

to exist:

Lim. _£L = 0 (2-la)

A-o A

, P
L\m _==^ — f (2-lb)
A-o A, 1 .

Where the limiting process is carried out so that 0 is always

inside A . t. is a stress vector — force per unit area.

 -7-

b. Stress Tensor

The stress vector t(*) o n an arbitrarily oriented sur-

face with unit normal r\ may be found once the stress vectors

on each of three mutually perpendicular planes are known. The

nine c a r t e s i a n components of those three stress vectors form the

stress tensor.(Fig.2-2):

Jt. = ( u s i n g summation c o n v e n t i o n
4
i ~) on r e p e a t e d i n d i c e s )

(2-2)
where
0",,

(2-3)
**
 -8-

Is called the stress tensor. The diagonal elements are called

normal stresses and the off-diagonal elements, shear stresses.

From (19) the stress tensor is symmetrical,I.e. o* . =

; <5j. .

FIG.2-3 STRESS V E C T O R S ON A TETRAHEDRON

By c o n s i d e r i n g Fig.2-3, we o b t a i n Cauchy's formula:

t(n) = jt. «;

or (2-5)

which completely characterizes the state of stress at a point

in terms of the nine cartesian stress components.

c. Trartsformation of Coordinates

Suppose that two c o o r d i n a t e systems are available, Fig.2-l».

A vector in s p a c e may be specified by giving tts coordinates in

either system, namely, K;(X, Xi

( or ^(x,', ). Then the coordinates

may be connected by the linear relations

(2-6a)

(2-6b)
FIG.2-4 R O T A T I O N OF COORDINATES
 -lO-

i n whi ch
On

w= —
(2-7)

>

are the direction cosines of the X^-axis w.r.t. the X,,-axls

as shown in Fig.2-4. Equations (2-6) define the transformation

of a tensor of order one (20).

d. Stress Transformation Laws

Since the stress vector is a tensor of first order,

Eq.(2-6) may be applied, I.e.

For arbi trary n (2-8a)

Similar! ly, 0"ij - a* ty* #

(2-8b)

e. EQUI1ibrlum Equations

With regard to figure 2-2, we w r i t e the equilibrium

equations In the X,, X 2 t Xj directions, and reduce to the

equation:

+ ? L - o (2-9)

which the stress tensor has to satisfy.

 -11-

f. Principal Stresses, Principal Directions. Stress Invariants

We now seek a plane with direction n. f o r which the stress

vector X.(.R) is parallel to the outward normal n. That is

or ( ftj -^fcp"} " °. (2-10)

where <£y is called the Kronecker delta defined by

(I c - ;

Such a direction nj is called a principal direction and the

corresponding stress value is a principal stress. The

charaterIstic equation for Eq.(2-10), which must be satisfied

if these homogeneous equations In nj are to have a solution, is

\< ij -<rStj \
r
= 0 ^ (2-11)

which leads to the values of the principal stresses. For each

value of the principal stress associated with njnj=l, a unit

normal vector & can be determined.

Expanding Eq.(2-ll), yields

-cT -!- 3
I,tf -r <r + I, = o
l
4 f

where
I. = <£.
(2-12a)

If the reference axes coincide with the principal axes, then

 -12-

l 2 = 07 <r t (r(r + <5](j;

z t 3
(2-12b)

where O", 0", <r, are the principal stresses, the solutions of Eq.(2-ll).

Ii , I 2 and 1^ a r e called the first, second and third invariants of

the stress tensor. It can be shown that one of the principal stress

is the algebraically largest and another the algebraically smallest

of those a c t i n g on any o f the planes passing through the point in

ques t i o n .

g» Shearing Stresses

The resultant shearing stress component at a plane with unit

normal vector ji is given by

or/ = i ( r i ) • tin) - ( i ( a ) - n ) 1
.

Let the principal axes be the coordinate axes, and:

where is the principal stress. To find the extreme shear stresses,

it is merely necessary to m a x i m i z e Eq.(2-13) subject to the constrain-

5 n g r e l a t J o n
n n. = 1 (2-n>
The results are summary in Table 2-1. One o f these values will, of

cause, be the largest shear stress a c t i n g on any plane passing through

the point In question.

TABLE 2-1 MAX. SHEAR STRESS AND PRINCIPAL D^ECTION

n n n c o r r e s p i ndi ng 0^

0 I | 'a -«3l
0
i M - M
1% 0 i 1 (Tii
 -13-

h. St ress Dev i a t i o n s . Devi a t o r 1 c Stress 1nvar i ants

The stress tensor mayb e s e p a r a t e d into t w op a r t s : o n e

is spherical stress tensor (hydrostatic stress) which causes

e l a s t i c delatation, and t h eother i s t h ed e v i a t o r i c stress

tensor 5^ (tangential stress) which causes e l a s t i c distortion

and p l a s t i c s t r a i n .

where n _ J^u, r (2-15)

(2-16)

The deviatoric stress invariants J} > J 3 a r ea n a l o g o u s to

I I # 1 2 / L 3 Thus

J, = Su = 0

^ = y C u Sjj - Sij Sij) = y Sij Stj

s

= i[s; + s;.^ + 2 (s^ 5,V5;.)J

+

= y ( S* + S + l
z ) ( 2 _ 1 7 )
 -14-

II. Analvsi s of Strain

When f o r c e s are applied to a body, the position of any

point of the body, In general, is changed. The displacement of

a point is defined as the vector distance from the initial to

the final location of the point. A body is considered to be

strained whenever the relative position of points in the con-

tinuous body Is altered.

a. Strain Tensor

FIG.2-5 DISPLACEMENT OF NEIGHBORING POINTS

In Fig.2-5, P.(x, Xi,Xj) i s a

material point in the unde-

formed body, and ?{*\ t

x
2,*p ' s t n e
location of the same par-

ticle of material after deformation. Q (,X, + d X | X ^ d X j

0 ( ,tyeUp 5 s a

neighboring point of f 9/ and its final position is Q(x +dx

1 )

X + dx
2 if Xj+dxj\ <4 I s the displacement vector.

For homogeneous deformation, we can write

(2-18a)
and

X; =• X-, (*.. *j) (2-18b)

 -15-

provlded theJacoblan J =J"^J d o e s not vanish.

The square of thedifferential element of length between

P 0 and Q 0 In t h e undeformed body i s

(dx) = dX • dX = dX^dX; #

SImilarily In t h e deformed body

( d x ) = dx. • dx. - dXidx^

Form (dx)*-(dx)*= d x d x . - d X d X
; m l n

Alternatively, write

( d x ) ' - ( d X ) = d x d x - olXrv, d X
; ; w

= x
*,m i , „ dX dX„ - dX^dX,,^
x
m

~ ( i,tn i,n
K X
^m* )^X^ dXn .

We d e f i n e thestrain tensors

= i t * ^ - ^ ) . ( 2
- 2 0 >

The strain tensor >s c a l l e d the Eulerlan finite strain tensor,

and J^j is called the Lagranglan finite strain tensor. Both o f

them are symmetric.

 -16-

The displacement components are defined by

M- = X< - Xi
t h en
ML
and X i
>j - i
Si

•J
Equations (2-19) and (2 -20)
i
become

(2-21)

(2-22)

For Infinitesimal deformation theory, neglect the high order

terms, and r e f e r to the initial coordinates, and t h e linear

strain Is then

(2-23)

r -ax,
or
r n oin

N=
SYMM
—-
OtUL.\ VIA*

£31
The diagonal elements are called extension strains and the off-

diagonal elements are c a l l e d shear strains.

b. StraIn DevIator Tensor

The strain tensor may b e s e p a r a t e d into two p a r t s . One

part is called the volumetric strain, and t h e o t h e r is called

the devlatoric strain. The latter Is associated with change of

shape, and Is a l s o known as d l s t o r t i o n a l strain defined by

 -17-

i lS. i2
- 2k)

c. c o m p a t i b i i i t v o_f_ strain Components

The strain components cannot be a r b i t r a r i l y prescribed

and certain relations, called the compatibility conditions,

must exist between them. The e q u a t i o n s , f o r small strain, are

given by ( 2 1 )

H<*'
e +
= £u,j t
+
^ U , (2-25)

III. Field Equations i n Elastic!tv

a. Generalized H o o k e ' s Law

For a three dimensional state of stress, Cauchy generalized

Hooke's law into the statement that each stress component is a

1inear function of a l l the strain components,

0" = h . F

ij K
* , ( 2 - 2 6a)
where E has
ijkl
the following porperties:

•~r(E;.v, ) = 0 f o r homogeneous materials,

r' _ r for isotropic materials,

(The primed value
being referred to a d i f f e r e n t set of coordineates)
 -18-

b. Constitutive Equations

For homogeneous isotropic linear elastic materials,

Eq.(2-26a) becomes ( 22)

or (2-26a)

{hi = 2 ^ , . .

Equation (2-26)can assume the alternative form

(2-26c)
where 1 2<T

modulus of elasticity
E = A, t<T

X Po i s s o n ' s ratio
i7 -
2(A. + Gr)

bulk modulus
K
" 3

Cjp and X are the Lame's constants

c. Field Equations in Elasticity

The solution of a given problem in linear elasticity

can proceed from the basic equations:

 -19-

1. Equilibrium Equations

% i + ft = 0 ; (2-9)

2. Strain-displacement Equations

(2-23)

3. Const i t u t i ve Equations

=
S'K*** E
'
(2 26)

These fifteen equations are the governing equations for the

fifteen unknown state variables o^. / £,j t w, . They must be satis-

fied at all point inside an elastic solid in static equilibrium,

Together with the boundary conditions:

^ ~ T. prescribed on 5^-

= M' K prescribed on $

The solution is possible and unique. (23)

 -20-

IV. Field Equations In Plasticity

a. Yield Theory

1. Yield Surface

The yield surface is defined as the surface in stress

space, with stress components as coordinates, within which

the stress vector may change without any plastic-strain in-

crement; stress Increments beginning from points in the sur

face, If directed toward the exterior, imply plastic strain

Increments.

2. Initial Yield Surface.

For isotropic plasticity, the initial yield surface

must be independent of the orientation of the reference

axes. By c h o o s i n g the axes of the principal stresses as

the reference axes, the' i n i t i a l yield surface may be

expressed in terms of the principal stresses and repre-

sented by a surface in a stress space with o", , (f t , (Tj

as coordinate axes. Thus the Initial yield function may

appear as

H <T , ovop =
X o.

Furthermore, experiment indicates that the hydro-

static pressure has no effect on the plastic deformation.

Hence the Initial yield condition may be expressed in

 -21-

terms of the deviatoric stress invariants in the form

.H J,, v = °.
Two simple yield conditions for the initial yield of

isotropic material which have provided highly useful descrip-

tions of many real materials are discussed in the following

section.

( 1 3 . .Tresca Yield Condi t i o n (Maximum S h e a r Theory)

This hypothesis is based on the assumption that yield

occurs when the maximum shear stress reaches a limiting value.

In view of the results given in Table 2-1/ a complete mathema-

tical description of the Tresca condition is given in Table 2-2,

and shown in Fig. 2-6.

TABLE 2-2

Stress State Tresca Condi t ion

*J <
<r < IT, «S = <5"o
t

°i < $ < IF.

< (Tj <<T, o; = <r 0

<r, <<r 3

S < <T, <(T 2 ^2- = <r 4

where ^=yield stress for simple tension.

 -22-

FIG. 2-6 TRESCA YIELD SURFACE IN PRINCIPAL STRESS SPACE

 -23-

The Tresca yield condition given by Table 2-2 can be repre-

sented by

where J J are the deviatoric stress invariants.

(2).^OJI Mjses Yield Condition (PistQrt'on Energy Theory?

Von M i s e s hypothesised that yielding o c c u r s when the

second deviatoric stress invariant attains a prescribed value

A, . For the uniaxial stress state -fc - — . Then

o

f= 7 - -4- = 0
(2-28)

Fig. 2-7 shows the geometrical meaning of the Von M i s e s con-

dition.

0~3

'< =o

FIG. 2-7. VON M I S E S YIELD SURFACE IN PRINCIPAL STRESS SPACE

 -2k-

3. S u b s e q u e n t Y i e l d S u r f a c e

Continued loading beyond the initial yield surface

leads to p l a s t i c deformation w h i c h may be accompanied by

changes in both size and shape of the yield surface. For

perfect plasticity the yield surface does not change

during plastic deformation and the initial yield surface

remains valid. For isotropic hardening, however, the size

of the yield surface Increases, but the shape remains the

same during loading. To take into account such changes It

is necessary to modify the initial yield surface and to

define the subsequent yield surface, also known as the

loading surface. A general form for the loading surface is

given by (24)

* (
V4> K ) =
°< (2
" 29)

Which depends not only upon the stresses 6\. , but also
P J

upon the plastic strains £,j and the work hardening charac-
teristics represented by the parameter K. Differentiating
f = oby the chain rule of calculus, we obtain

where dj- t d(Tj. } etc. represent time differentials.

 -25-

If f = o and dj-<0 a condition leading to an elastic state is

implied, and it must follow that di,j = dK.= 0. Thus

•f = 0 —do"-; <<• 0 is defined as "unloading"

J
/ 7>G\f J

j. = Q / ^f-dOlj = O is defined as " n e u t r a l loading",

J s i n c e it i m p l i e s that the stress-
point r e m a i n s on t h e i n i t i a l yield
su r f a c e .

I _ Q J^Ljir y Q is defined as " l o a d i n g " , since it

' J i m p l i e s that the s t r e s s - p o i n t is
'J moving outward from the current
yield surface.

For perfectly plastic materials plastic flow occurs

for

and the case

f ~ °, ~~^^"ii > ° does not exist.

b. C o n s t i t u t ?ve Equationr-Plastic-flow ru1e of i ncrementa1 type

Based on the following assumptions:

1. isotropic elasto-perfect1y plastic materials,

2. the principal axes of stress and plastic strain increment coincide,

3. dtij = dl'j * d£i' , (2-30)

5- <«t; = o ' < 2

" 3 2 )
 -26-

Reuss extended a proposal of Prandtl to say that the plastic

p

strain i n c r e m e n t s dijj a r e related to the stress deviator com-

ponents 5y by

d£ J
; = dA-Sg- ( 2
" 3 3

where d\ is the proportionality factor, w h i c h may change dur-

ing loading.

For work-hardening materials associated with the Von

Mises yield condition. It Is useful to define the equivalent

stress 0 and equivalent plastic strain increment d£ , P

as

(2-34)

and
(2-35)

Then the Prandt1-Reuss flow rule becomes

Q ' t j = =- Sjj (2-36)

J 2(J i ,

with

=
TF~ "
(2 37)

The p l a s t i c work increment is defined by (25)

 -27-

p d > W f
> j = S i
J d £
' J
in which has been introduced. Furthermore, if the

same m a t e r i a l follows the Prandtl-Reuss flow rule, the plastic

work increment may be e x p r e s s e d as

i

and the Prandtl- Reuss equation becomes

By u s i n g the work-hardening hypothesis discussed in (26),

the Prandtl-Reuss equation may be w r i t t e n as

AC
p
- 3da c
(2-38)

where

(2-39)
P

corresponds to the slope of the equivalent stress 0 versus

equivalent plastic strain J<j£P c u r v e . Substituting E<j. ( 2 - 3 s)

Into E<|. (2-3o) , In view of Ec|. , leads to

d£i
j • ~if +
—T— J T Si +
iffH' J; s, d < T ?
°
(2-40)
 -28-

For unloading where the e l a s t i c s t r a i n is the only

applicable term and

d£.. - i l l +
(
' " 2 0 )
L i 5 « (2-ui)
a t ,
J ZGr P J
3

writing (7^.(2-4.0) 1° decomposite form w i l l 1 ead to Hill's

complete s t r e s s - s t r a i n equation (27)

Ac - C<5?
J u. - j A L s

(2-42)

p
in which d S ^ o has been Introduced.

A general equation for determing the p l a s t i c strain-

stress relation f o r any y i e l d c o n d i t i o n was p r o p o s e d by

D.C. Drucker. Based on h i s d e f i n t i o n of work-hardening materials

(28), Drucker jumped one s t e p beyond the c l a s s i c a l treatment,

and showed that, for these m a t e r i a l s , the p l a s t i c s t r a i n in-

crement vector must be normal to the y i e l d or loading surface

at a smooth point on t h a t surface, and must lie between

adjacent normals at a corner point. I.e. (29)

Eq.(2-43) Is a l s o called the normality principle of plasticity.

 -29-

c. Field Equations In Plastic!tv

1. Eoui1ibrium Equations

CJCTjj.j + dF, = O ; (2-9)

2. Strain-displacement Equations

d £
' J = + * i*\
U
(2-23)

d£,j = de*. + d£y ; (2-30)

3. C o n s t ! t u t i ve Equations

(2-43)

^ = -^. <'-''Vit, +
«-»>
where f ((T;j, fcf.-, l c ) - 0 def ines the y i e l d surface.

4. Boundary Condi tIons

d(T,jlij =r d T i P r e s c r i b e d on 5 ff

J if i

dm; = C(M- prescribed on S

 -30-

It is seen that the pi a s t i c i t y problem is defined

in a manner similar to an e l a s t i c i t y p r o b l e m of the

infinitesimal deformation theory, except for the stress-

strain relations. Thus, once the problem in flow theory

has been formulated, problems of p l a s t i c i t y can be analysed

by integrating the resulting relations along the prescribed

loading path.
 -31-

CHAPTER THREE

FlNITE ELEMENT METHOD

The basis of and the formulation of the finite element

procedure will be reviewed here to establish the approach and

the notation used in these studies. This will be done for the

linear elastic case, and the extension to the plasticity pro-

blem w i l l be made in the next chapter.

It seems redundant to mention that the finite element

method has been found to be a p o w e r f u l approach to stress

analysis of boundary value problems. Part of its advantage

stems from its ability to take care of the irregular shapes

of boundary and mixed boundary conditions.

The basic concept of the finite element method is that

a continuum c a n be decomposed into a finite number of regions

(elements), in each of which the behavior is represented by

a seperate field. In order to guarantee the convergence of the

finite element solution to the exact values, these fields are

so c h o s e n that they satisfy the completeness requirement and,

if possible, inter-element boundaries are made compatible (the

" c o m f o rm i ty'.' c o n d i t i o n ) . Oliveira proved that, convergence of

the finite element solution requires only the preservation of

 -32-

correct rigid body and constant strain modes; ie. that compa-

tibility could be relaxed if the completeness condition is

satisfied (30). This remarkable result makes a p p l i c a t i o n of the

finite element method to plate bending problems, for example,

much simpler.

The finite element method may be interpreted as a

particular case of Ritz's method associated with a variation-

al principle in continuum mechanics. From the different varia-

tional principles in s o l i d mechanics, it is possible to derive

numerous finite element models w h i c h may lead to either a stiff-

ness method (principle of minimum p o t e n t i a l energy), a flexi-

bility method (principle of minimum c o m p l i m e n t a r y energy) or

a mixed method (31).

I. Variational Principle?

Variational principles have made a great contribution to

the development of stress analysis by the finite element

method: for example, in deriving a finite element formulation,

in providing the physical interpretation of the approximate

governing equations thus derived, and most notably, in establi-

shing convergence proofs and bounds formulae.

 - 3 3 -

a. Principle of M ? n imum Pot e n t i a 1 Energy

Consider a continuous and s u f f i c i e n t l y differentiab1e

displacement f i e l d ju* such that

a.

— prescribed displacements on S u

Note that the stress f i e l d d"- is not necessarily in e q u i l i -

0
b r i urn, i . e .

%i +
Fi = o ,

and 0-* n . — "T. o n

c

'J I i1
O f f may n o t b y true.

Form

1 Tj'(u.'- M.)d5 + [ /=• ( u j - u . ) d v = I ^ ( t , * - M-)dS + f F.Cu.'-M^dv

(equilibrium requirement at surface)

(divergence theorem)

= I(^, F ) ( u ; - V ^ } / y K j - » , j ) d y
J+ i + l

(3-1)
 -34-

Notlng that (for equil ibrlum)

%] +
F, = o ,

and that

%S (
"S 5 =
-5 ^ % ( +
*v - T ^ ^ <5i v )
)
( +

(dummy subscripts)

(moment equilibrium)

(strain-displacement relations)
thus Eq.(3-1) may by w r i t t e n as

N o t i ng that (3_2 )

since E4^ jj
K is positive definite, and that

because all the subscripts are dummy, Eq(3-2) becomes

or

(3-3)

Thus the minimum p o t e n t i a l energy principle states that: Of

all the kinematica11y admissible displacement fields, the

actual solution i.e. the one which not only satisfies the

compatibility conditions but which also corresponds to a

stress field which satisfies the equilibrium condition, is

distinguished by the stationary value of the corresponding

total potential energy. The solution can be shown to be unique

for 1inear systems.

 -36-

I I . Fonnulat ion ojF_ Finite E l ernent Method bv

The Principle o± M i n i mum P o t e n t i a l Energy

In formulating the finite element method, the displace-

ment field is represented by interpolation functions together

with generalized displacements at a finite number of nodal

points in each element. To fulfill the general completeness

requirement, the field components and all their derivatives,

of order not higher than the highest derivative entering the

energy density expression must be able to take up any constant

(non-zero) value within the element. It follows that if p is

the order of the highest derivative appearing in the energy

expression (for example, p=2 for plate bending problems),

then it is only necessary that the interpolation function

must contain a complete polynomial up to the degree,

all the terms of which have independent arbitrary coeffici-

ents. The terms of higher degree may be allowed to vanish

whatever values are taken by those coefficients. For example,

O.C. Z i e n k i e w i c z has omitted some fouth-order terms from a

complete fouth-order polynomial for a rectangular plate bending

element. The interpolation functions he uses satisfy the above

requirements for the plate bending problem and can yield cons-

tant values for deflection, slopes and curvatures if the nodal

variables are suitably prescribed. In other words, the

 -37-

parameters of the polynomial are entirely defined by the general-

ized displacements. In m a t r i x form the assumed displacement

may be w r I t t e n as
tx = As
— - - (3-4)

where = displacement column matrix,

A = nodal displacement column matrix,

A » shape function matrix.

The corresponding strain column matrix Is

I = U (3-5)

where JL Is the differential operators matrix.

The corresponding stress column matrix is

6 = VI
~ (3-6)
where t) i s t h e e l a s t i c constant matrix.
The t o t a l p o t e n t i a l energy functional for any elastic
solid is given by

If the body is divided into a finite number of discrete ele-

ments V n , then

(3-7a)

or in matrix form

v = i i T
I d $ - j u fdvt J i r t d v ]
T
( 3 . 7 b )

s "v
 -38-

Substi tutlng equations (3-4), (3-5), and ( 3 - 6 ) i n t o (3-7b),

leads to

= i f - • £ { ] ' - f i d ' s + $ A £ d v Y + | 1 {] (u) £(LA)4V

T
T T

(3-8)

where

i = j ( LA) D(LA)dV
T

V» (3-9)

— member stiffness matrix

S= J / I d s+ J
_
/ F
dV (3-io)

— nodal force column matrix

The nodal displacements S for different elements a r e

not completely independent; A transformation i s needed to

relate the element nodal displacements to t h e independent

generalized global displacements. Then the compatibility equa-

tions of the assembled structure arewritten in matrix form as

 -39-

5 = ^1 (3-11)

where = compatibility transformation matrix,

X = global displacement column matrix.

Substituting Eq.(3-ll) into (3-8), yields

V p = - i T
( i « T
5 J + ir (2a 4a)i T T

(3-12)
where

= global stiffness matrix

R - 2. aS
T
(3-14)
n —

= global force column matrix.

The principle o f minimum potential energy requires that

7^- = (- r
i +
is r ) S r = 0.

For arbltrary 8 r

=
& . (3-15)

Thus, f o r a given force vector R the corresponding displacement

vector £ c a n be o b t a i n e d by t h e i n v e r s i o n o f J$ a s

r = Kf'fc (3-16)

and the nodal displacements, corresponding strain and s t r e s s

within each element may be f o u n d by e q u a t i o n s (3-11),(3-5)

and (3-6).
 -40-

CHAPTER FOUR

FORMULATION 0£ ELASTO-PLASTIC PROBLEMS

I. Genera I Considerations

Plastic theory is often associated with large deforma-

tions, but a p l a s t i c deformation analysis involving large

deformations Is complicated both by geometric nonlInearities

and by a n i s t r o p l c hardening. Moreover, the lack of experiment-

al support for e l a s t o - p l a s t l c computational results in the

range of infinitesimal deformation makes an excursion Into

the field of large deformation p l a s t i c i t y a precarious enter-

prise. Hence, In the present study attention will be focused on

p l a s t i c Ity with infinitesimal deformations.

The finite element displacement method Is preferred, even

though the force method might be better in a philosophical

sense (32).

The method of Incremental loading Is frequently employed

in the analysis of a path-dependent process such as plastic

deformation. By this method the complete nonlinear response of

the solid body Is generated as a sequence of piece-wise linear

systems, Fig.4-1. If AR (a typical load Increment) is taken

sufficiently small, then the approximate state is reasonably

acceptable, even though some error £, will Inevitably enter

each step of the process. This error, however, can be reduced

by means of a numerical Iteration procedure such as the well-

known Newton-Raphson method or the m o d i f i e d Newton-Raphson

method, Fig.4-2.
 -41-

LOADS

APPROXIMATE STATE

REAL STATE

RESPONSE

FIG.4-1 THE INCREMENTAL LOADING PROCESS

LOADS

— RESPONSE

° - incremental loading
* - 1 c o r r e c t e d by Newton-Raphson method
* - 2 c o r r e c t e d by M o d i f i e d N e w t o n - R a p h s o n method

FIG.4-2 A C O M B I N A T I O N OF THE LOAD INCREMENT AND

NEWTON-RAPHSON (OR MAD I F I ED NEWTON-RAPHSON)
METHOD
 -I»2-

I I. General 1zed 1ncremental E o u i 1 i b r i urq F o u a t i o n

During each load increment/ the equilibrium equations

c a n be o b t a i n e d by following the procedures given in chapter

three.

L e t
A_U = A *1 , U-l)

A I - _L U-2)

where &u = a column vector of displacement increments

AS s= a c o l u m n vector of nodal displacement increments

&l - a column vector of strain increments.

For isotropic materials we a s s u m e d the incremental stress

and strain are related by Hooke's law in the elastic region and

by the Prandt1-Reuss (or Drucker) definition during loading in

the plastic region,!.e.

— = 2L*f A i , (U-3)

The matrix J)" P contains only the material properties of an elastic

element/ I.e.

T> - D

but in the plastic region it depends/ in addition, on the current

state of s t r e s s and hardening of the element, i.e.

D** = D* (see next section).

— * P — « / >

A variational formulation for the m a t e r i a l s of elasto-

 -43-

plastlc behavior assumes a stationary value of functional

(see appendix)

4)
" \ v„ \ J
'

where AT = a c o l u m n vector of surface traction increments,

AE=a column vector of body force increments.

By introducting the "global stiffness matrix":

1 = J ( LA) D;*.(
T
M)dV a (4-5)

the "global force vector":

(4-6)

and the compatibility equations

AS = a ir (4-7)
then Eq.(4-4) becomes

(4-8)

The necessary condition for the functional V to assume

a stationary value

J (4-9)

gives the incremental equilibrium equations

K *r = Aft
(4-10)

for the unknown displacement vector AT . Fq.(4-10) will be

solved in section IV.

 -44-

III. Const!tut!ve E q u a t i o n s : - E 1 a s t o - p l a s t i c Ma t r i x

The elasto-plastic matrix based upon the Prandtl-Reuss

theory and t h e Von M i s e s yield criterion was f i r s t presented

by Y. Yamada (14). A few years later O.C. Zienkiewiz (15) and

his colleagues presented a matrix of t h e same f o r m w h i c h was

good for any y i e l d criterion. The b a s i c part of Zienkiewiz's

development is presented here for completeness and t h e matrix

on which the current p r o g r a m was b a s e d is also developed.

Rewrite equations (2-30), (2-31), and (2-43) in matrix

form

(4-11)
<fU] = d U ] , + d [ t ] p ,
(4-12)
dUl,= [o]"dW.

(4-13)

and thus

where

i % 0 o 0

% t % 0 0 0

% \ t
0 0 0
(4-15)
1
0 0 0 0 0
2

0 0 1
o 0 o
2
I
0 0 0 0 0 t

In w h i c h x, I- f
i-2\r
 -45-

When p l a s t i c flow I s o c c u r r i n g t h e s t r e s s e s a r e on t h e

yield surface, J = o a n d 6j-=0. Thus, from Eq.(2-29),

or

dOJ - Ad\ = 0 (4-16)

where

(4-17)
dA
Multl. E q . ( 4 - 1 4 ) by 1eads t o

(4-18)

Sub. Eq.(4-16) Into E q . ( 4 - 1 8 ) , yields

(4-19)

A + [iL] T
[oj ri£1

Sub. Eq.(4-19) into Eq.(4-14), leads to

 -46-

or i

where |Q{ _ [ ]
0
1 0 1 J J L 1 J J
(4-21)

Thus In plastic flow analysis, the elasto-plastic matrix

replaces the e l a s t i c matrix [o] . [ ] n

a f may be interpreted

physically as the required correction to the elastic stress-

strain relation which keeps the stress increment on the expand-

ing yield surface (or tangential to the yield surface in the

case of elastIc-perfect1y plastic materials). It is symmetric,

positive definite and is valid whether f\ is zero or not.

For isotropic plasticity, it is reasonable to assume

that (33) .

Thus, Eq.(4-17) becomes

" die dK

For isotropic perfect p l a s t i c i t y , where K.= c o n s t a n t , dK = 0 /

h s 0
 -47-

For Isotropic work-hardening plasticity, the plastic work

done i s d e f i ned by

fjKL •= dV. = 5.. d£ - P

but
-ad;

and 1
^

If the Von M I s e d yield criterion

Is employed. Then

die = adi f

d<f do" I H
1K dK 6~

8f =

D0~; 2
0=
and

noting that

A = H (4-23)

Where o" , dt'' / and H are defined in chapter two.

 -1*8-

Noting that
[ D ] T
= [ D ] ;

Then the numerator of Eq.(U-21) becomes

3f 3_i
L?MJ
S Sj
*1 jx
x 5

s, SYMM
S, S«j
S.S 3 s, s }

5 ^

ss ss t

S 5 5
7i S
j* S
J *
 -49-

and the denominator

5
1

4 +feW ^1 H'+ 4=[S 5 5, 2 5„ 2 S

J =s

^ 3*J CT

L V

2?

= 3(p + H

Thus Eq.(4-21) c a n be w r i t t e n as

SYMM
s*

3<r
*.
St, $*,,
 -50-

For two d i m e n s i o n a l plane stress problems, the elasto-

plastic matrix becomes

5, • 2
1 i SYMM

[of = JL 5 Scj + a u - ^
x U-24)

I + IT 2 (1+ U )

where

•Itir q £
 -51-

IV. Solution o_f_ P l a s t i c i t y Equations

The global stiffness matrix I n Eq.(i»-5) Is different

from the linear elastic case, since for elasto-plastic pro-

blems the matrix [D] Is itself a function of current state

of stress. The i n c r e m e n t a l equilibrium equation, Eq.(U-lO), is

therefore nonlinear and cannot be s o l v e d by a s i m p l e matrix

inversion. Two w e l l - k n o w n methods, Newton-Raphson and modified

Newton-Raphson, a r e used to solve the nonlinear incremental

equilibrium equations. The p r o c e d u r e s a r e now b r i e f l y outlined.

a. Newtpn-Raphson, Method

Consider a system of nonlinear equations, for example,

Eq.U-25)
t*R3 - [F([*r))]

we m a y w r I t e

where is the v a r i a b l e and Is a known load increment.

At t h e "c o r r e c t " v a l u e of the value which solves Eq.(i»-25)-

we have

o Eq.U-26)

Let us expand as a Taylor series about some test value

C*irJ e such that

 -52-

Then

[$ ] + h.o.t. in Sr
[ H M ) ] = [?(M„)1 +
^
r

= [ o ]
by Eq.(4-26)

But 3W from Eq.(4-2i>)

3[*r]

where [ K J IS t h e " t a n g e n t i a l stiffness matrix" as defined by

Eq.(lt-lO). Thus

tjl. = W . M

= K H - W W 1 ,

Now [K] i s the Jacobian m a t i r x of the function [Fj and is there-

fore invertable, so

(4-28)

These arguments are illustrated diagramaticially in Fig.4-3.

This value of i s , of course, only approximate, but it can

be u s e d as a s t a r t i n g value for another step in the iterative

process. The a l g o r i t h m is simply;

(4-29)
FIG.l»-3 THE ILLUSTRQTIONS OF EQS. ( U - 2 5 ) ( U - 2 6 ) ( U - 2 7 )
/ /

AND (U-28)
 -54-

Thls process Is repeated until the residual force [^] M or

the increments of displacements \[ & r

) ) become zero,
lit

or within a prescribed criterion,

b. MQdified Newton-Raphson Method

The Newton-Raphson method could indeed be u s e d , b u t

it is generally inconvenient since the tangential stiffness

matrix depends on t h e c u r r e n t stress l e v e l s and has t o be

calculated at each step of the i t e r a t i v e process. Moreover,

we h a v e to invert a very large matrix everytlme. This d i f f i -

culty, however, c a n be overcome by an a l t e r n a t i v e process of

iteration, known a s t h e m o d i f i e d Newton-Raphson method, in

which only the i n i t i a l stiffness matrix i s used. Then, instead

of Eq.(l»-29), we h a v e

[>r] = O] _ [ ]"' K [oi u . )

3 0

These two methods a r e shown dtagramatical1y ( f o r one v a r i a b l e )

in Figs.4-4a and 4 - 4 b . In f a c t , it is generally found t o be

adequate t o compute [ K
] 0 f° r t n e
first load increment-the

elastic case- and to use t h i s value f o r a l l subsequent load

increments.

Obviously, t h e number o f iterations required for con-

vergence in the modified method is greater than In t h e s t a n d r

ard method; however, the actual c a l c u l a t i o n a n d amount o f com-

puting time required i s , in fact, less than that for the stand-

ard Newton-Raphson method because it i s not necessary to invert

a new s t i f f n e s s m a t r i x a t each c y c l e . In t h e p r e s e n t study,

the modified Newton-Raphson i t e r a t i v e method Is preferred.

-55-
 -56-

V. Calculation Procedure

It has been m e n t i o n e d that a combination of the method

of Incremental loading and the m o d i f i e d Newton-Raphson method

generates the complete nonlinear response by a sequence of

piece-wise linear steps. Thus the total load vector [fi] to

be a p p l i e d to the system ts divided into suitable small in-

crements [*1
4
and an available computer program developed

for the linear elastic problem has been adjusted to the elasto-

plastic problem by m i n o r modifications.

Since the Inverse of the tangential stiffness matrix [K,^

Is used throughout the process, It Is computed at the very

beginning for convenience. [K] o depends on the elastic constants

and the structural geometry only. The following indices are used:

n = number o f load increment

m = number o f iteration

a. Procedure:

A. Starting with a trial solution for the n** increment

and the load increment compute the residual force

til". - -KT.
B. Iterative loops:

1. Compute the approximate solution

 -57-

ML. = K - [ a ' h ] l t

( m = 0, I, 2 , 3, — )
ne
compute the e l a s t i c increment of strain A
£. and
m*i
nC r T^
1

stress 4 0" corresponding to UrJ i.e.

& i n e
= L/l a [*r]"

-w+i L
*tl y

add A £ to the current stress existing at the start

of the iteration 0"" to obtain

— m

if f ( /)«»only elastic strains occur, and the state

variables for the s t a r t of the next load increment are

c r v H
P " r ne
± = I + Ai

Go t o A and s t a r t the next load increment.

If f f C / U o a n d f(ff^ ,K)5o then

({) compute the a c t u a l stress &(T n<K

due to the nonlinear

stress-strain relation using the e l a s t o - p l a s t i c matrix

i.e.
 -58-

where * * *«

(li) compute the residual stress

Af = A £ -i f f

6. i f i(2"m*,/)*
0
and { C C , * ) * 0
then increase the value of

ff" to jf"' by i n t e r p o l a t i o n such that o then

compute
A tc*** ir n e
*••'»*

a £ « p A (T
—
rvt+» — —

and ^nr „x- n e

* >

7. compute the current stress and s t r a i n

c n
~n .ne

8. compute the residual force (see section b)

and t h e new K . In t h e c u r r e n t work, a constant strain

element is used, A15"*" is constant, and t h e intergration

is linear.
 -59-

i f
lUjj" II^IIULl) l # e
* l f s o m e s e l e c t e d
criterion for
*' r i *

the smallness of [9j„, H ' s

satisfied, then let

1 = 6
— c — m+l
i

and go t o A t o s t a r t the next load increment. Otherwise,

go to B and begin the next iteration.

 -60-

Resldual Force Vector

(i) Consider a typical triangular element (Fig.4-5),

I - X, u,

FIG.4-5 CONSTANT STRAIN ELEMENT

and w r i t e the p r i n c i p l e of virtual work equation:

i(iS-) AS" = [
T
H'£")V c|V n

— V.
J
t

but n n
= LA JS n

J w .

i
 -61-

For arbitrary we have

a S " = J ( L A ) 4 f f " dV
T
.

Since i s the actual s t r e s s due to we c a n write

the above e q u a t i o n as '

'y. »ti

(it) The residual force vector is given by

Noting that

dV

where we h a v e implied the integration over the whole volume.

 -62-

IV. Example of Numer i c a 1 Resu1ts

a. Perforated Strip: Strain Hardening Material

The problem of determining the elastic-plastic strain

and stress distribution that occurs in thin perforated

strips of a strain hardening material when the applied

stress is increased monotonical1y from the elastic region of

loading to values producing an impending plastic flow, was

investigated by Theocarls and M a r k e t o s . In (3U) they report

results of an experimental strain analysis. Using photoelas-

tic coating and e l e c t r i c a l analogy techniques total strains

were determined. Stresses were estimated by applying the

Prandtl-Reuss incremental plasticity relations Eq. (2-U2). In

this fashion a measure of the elastic strains and, hence of

the plastic strains, could be o b t a i n e d by Eq.(2-30).

The finite element Idealization of the geometry under

consideration Is shown in Fig. i*-6. The idealized uniaxial str

ess-strain diagram of the strip material, alumunium alloy

57S, Is reproduced in Fig. k-7. In Fig. k-S the development of

maximum strain in the longitudinal direction as o b t a i n e d from

finite element results is shown, together with similar results

published by TheocarIs and Marketos, for different values of

the applied load. In Fig. U-9, a comparison can be made

 -63-

between the development of p l a s t i c zones as reported in (34)

and as computed by finite elements. The load factor \ is

defined as

A p p l i ed load
K =
Load at which yield begins by closed form solution

Finally/ Fig. 4-10 shows the equivalent stress v.s. strain,

equations (2-34) and (2-35), in the element which yielded

first. The curve shows that the finite element solution Is

following the hardening rule specified.

b. PfirforafrpH Plate: Elastlc-perfectlv-Dlastlc Material

The problem solved in (a) was repeated/ but with

elastic-perfectly-plastic material/ ie. H'=0/ Eq. (2-39).

In o r d e r to compare the result with the strain hardening case,

the development of t h e maximum strain in the longitudinal

direction is shown In Fig. 4-8. It is interesting to note

that there is only a little difference at higher loads.

c. Deep C a n t i l e v e r Beam.:
The dimensions and loading conditions are shown in

Fig. 4-11/ in which the finite element net is presented.

 -64

The m a t e r i a l used is shown in Fig. 4-12. In Fig. 4-13 the

ratio K Is defined and the development of the plastic

zones for different values of ^ is shown. The region

enclosed by the dashed line was not calculated by the pro-

gram, b e c a u s e when the highest load was a p p l i e d ( A. * I )/

no convergence could be o b t a i n e d . This problem was solved

in reference (35) and a similar experience was noted. Figs.

4-14 and 4-15 show the distribution of normal and shear

stress acting across the sections AA' and BB' respectively.

d. Right-angle Notch: El a s t i c - o e r f e c t 1 v - p l a s t i c M a t e r i a i

The Right-angle Notch w i t h a notch to half-width

ratio of 1 to 2 is shown in Fig. 4-15/ wherein the finite

element net Is presented. An e l a s t i c - p e r f e c t l y - p l a s t i c

material was assumed: uniaxial yield stress ^ - K

l/mm

The development of t h e maximum strain in the longitu-

dinal direction for loading is shown in Fig. 4-17/ wherein

the curve for a circular notch is also shown for comparison.

Fig. 4-18a shows the development of the plastic enclaves

obtained by D. Allen and R. Southwell (36). They assumed that

there Is no volume change in either elastic or plastic

deformation and solved the stress function by a relaxation

method. Fig. 4-18b shows the finite element results for

different loading paths. These two p l a s t i c zones are of the

same g e n e r a l shape.
-65-
 -66-

28

0 1 2 3 l» 5 6 7 8 9 10 11 12 xlO
STRAIN

yield stress in pure tension <£ = 2k, "5 J/mm*

K

slope of work-hardening curve H » 225.0 K

3/*,*, a

modulus of elasticity E = 7000. 0

FIG.l»-7 UNIAXIAL STRESS-STRAIN CURVE

 -67-

1.25

r^fr _v^
_
— i
1.00 —A

0 . 75

0.50

0.25

0
0 .5 1. 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

MAX. STRAIN IN THE LONG I T U D I U A L DIRECTION

5
experimental from Theocaris & Marketos
work hardening
o—o f i n i t e element solution

*—* f i n i t e element solution for e l a s t i c perfectly plastic material

FIG.4-8 END LOAD vs. MAX STRAIN FOR PERFORATED STRIP

 -68

FIG. 4-9a

EXPERIMENTAL RESULTS

A. = tM5 14* 1-5 «

FIG. U-9b

FINITE ELEMENT RESULTS

A = Itf 1.VS 2.o2 |.J 13] |.1S

 -69-

FIG.U-10 EQUIVALENT STRESS v s . EQUIVALENT STRAIN IN

THE E L E M E N T WHICH Y I E L D F I R S T
 -70-

Finite Element Idealization:

196 elements
120 nodes

FIG.U-11 SHEAR WALL S U B J E C T E D TO LATERAL LOAD

 -71-

STRESS

TENSION

STRAIN

COMPRESSION

yield stress = 36x10 psl.

modulus of elasticity = 30x10 psi.

FIG.l»-12 UNIAXIAL STRESS-STRAIN CURVE

 -72-

APPLIED LOAD

LIMIT LOAD COMPUTED FROM PLASTIC BEAM THEORY

4 L

2L

X - 1 . 0.94 .75 0.625

FIG.4-13 D E V E L O P M E N T OF P L A S T I C ZONES FOR DIFFERENT

V A L U E S OF ^
 -73-

FIG.U-Htb SHEAR S T R E S S D I S T R I B U T I O N AT H I G H E S T LOAD

 -7k-

FIG.«*-15b SHEAR S T R E S S D I S T R I B U T I O N AT HIGHEST LOAD

FIG.4-16 RIGHT-ANGLE NOTCH: ELASTIC-PERFECTLY-PLASTIC
 -76-

1.50

1.25

1.00

E \^° 0.75

0.50

0.25

0 . 5 1. 1.5 2.0 2. 5 3.0 3. 5 4.0 4.5 5.0

MAX. STRAIN IN THE LONG I T U D I UAL D I R E C T I O N ~

finite element solution for right angle notch

finite element solution for circular notch

FIG.4-17 END LOAD v s . M A X . S T R A I N FOR NOTCHED T E N S I O N

SPECIMEN. ELASTIC-PERFECTLY-PLASTICITY
-77-
 -78-

CHAPTER FIVE

NONLINEAR S T R E S S A N A L Y S I S ££ REINFORCED CONCRETE

I. General Considerations

Studies of the nonlinear response to load of reinforced

concrete structures have occupied the attention of many resear-

chers during recent times. Because of the several complexities

involved^ however, the state variables of the concrete and

steel are difficult to analyze. Some o f the problems are stated

below:

1. The nonhomogenous nature of the system.

2. The nonlinear behavior of the concrete and steel.

3. The structural system has a continuously changing

character due to the influence of progressive

cracking of the concrete under increasing load.

1*. T h e p o s s i b i l i t y of relative movement between

concrete and the steel reinforcement.

5. The deformations of concrete are influenced by

creep and shrinkage and are time dependent.

The stress analysis of any solid b o d y may be approached

on a one, two o r three-dimensional basis, and this is , of

course,.true for reinforced concrete. For example, if the

 -79-

member is long compared with its depth and w i d t h , and

if It remains uncracked, then beam theory is applicable,

ie.the usual one-dimensional analysis is satisfactory.

After cracking, however, the one-dimensional analysis

model fails completely. Thus, a realistic stress analysis

of reinforced c o n c r e t e must be b a s e d on at least a two-

dimensional model. If the loads and the support reactions

act only in a vertical plane, then the principal stress

normal to this plane is considered to be very small com-

pared with the others and a two-dimensional model is satis-

factory.

Two-dimensional elasticity and plasticity could,

indeed, be u s e d to formulate the problem, but even then

mathematical difficulties are encountered. The difficulties,

however, can be o v e r c o m e by the finite element method.

I I. Structural I d e a l Iza t ion

In the application of the finite element method to

reinforced concrete, the structure is decomposed into a

number of regions, in each of which the behavior is repre-

sented by a separate field. Thus the nonhomogenous nature

of the construction can be accounted for. In this study, the

well-known constant strain triangular elements have been

used for the concrete. Perfect bond is assumed and the time

dependent effect Is also neglected. A displacement method

of analysis under plane stress conditions has been used.

 -80-

N l . Fa 11 u r e Cr 1 ter I a

Concrete:

1. Tensile crack: if the principal stress in any direction

within an element exceeds the limiting

tensile strength of concrete under uni-

axial test, then the element is considered

to have cracked perpendicularly to that

principal direction.

2. Compressive yield: The Von Mises yield criterion has

been used to define yield regions.

Steel: The Von Mises yield criterion has been used.

 -81-

IV. Calculation Procedure

The subscripts and notations used here are the same as

those in chapter four.

Procedure:

1. Compute

2. Compute [*R]

3, Compute & 6 n e

h, If the element has not cracked In previous steps, go to 6,

S, If the element has already cracked, then

a. Compute

where

" <0i*« (9 o

f s J - o
0 0 1
b. Compute —

— w 4 4
JL'
c. If the element cracked in direction (i . e . CT = 0 ),

then put

<Ti = o*i new v a l u e of the principal stress,

residual principal stress(to be removed),

3 = <
residual principal stress(to be removed),

and go to 12,
 -82-

d. if the element cracked in direction (i.e. G~ ~ o )/

then put

<T =
2 (f* new v a l u e of the principal stress

c5j" = fir* residual principal stress (to be removed)

~d = t 0 residual principal stress (to be removed)

and go to 12/

6. Compute rj n e
/
—— mt I
7. Compute the p r i n c i p a l stresses (T and (p and the principal
I i

direction 6 of iF-^-h '

8. If both <^ and fi- are less than the prescribed limiting

tensile strength fir then go to 11 to check yield.

9.. If (T ? £
x then put

<?; = 07 '

and go to 12/

10. If (T 1 rf then put

* it

c ; -= 0 >

?, « o '
and go to 12/

11. Same procedure as given in chapter four:

a. if the element has yielded, then compute the residual

stress A tf" hr
and go to 13/

b. if not/ then go to 1 starting with the next load

Increment.
 -83-

12. compute the residual stress

where

o 0\ 0
o

and [<*p^] i s given in step 5,

13. compute the residual force [^]^

i f
I I ^ C l l ^ j C j ] : ^ then let

m+i
and go to 1 starting with next load increment,

15. if not, then compute the current s t r e s s and strain and

go to 2 starting with next iteration.

 -84

IV. N'umerleal Examole

C a n t 11 e v e r Beam

The structure that was studied Is shown In Fig.5-1,

In which the finite element net Is presented. The material

properties are shown In Fig.5-2.

An e x a c t solution for this problem, within the assumption

of plane sections remaining plane, could be obtained by

manipulation of the equilibrium equations based on Fig.5-3,

and the results are shown In Fig.5-4.

Stages of the finite element solution are shown in Fig.

5-5 to 5-11. For a given moment, the extreme fibre strain of the

concrete and the neutral axis depth for the "exact" solution

can be obtained from Fig.5-4, and the corresponding stress

distributions are shown on Fig.5-5 to 5-8 for comparsion.

It will be seen that the results are In reasonable

agreement up to this stage, but that the finite element solution

shows the load Increasing well beyond the actual maximum.

Fig.5-4 shows that the load c a p a c i t y of the structure has

actually levelled off at a moment o f 170.6 Kft., and the

program is clearly failing to converge properly under these

conditions.
 -85-

F!g.5-12 shows the cracking pattern and the plastic

region which develops In the beam. These are believed to

representative of the truth.

FIG.5-1 REINFORCED CONCRETE CANTILEVER BEAM
 -87-

STEEL CONCRETE

40. O O K S l .

3. oo ksi.

6
It
= 0.42 ksi.

Z<\<\o Ksi.

3.1 * I o* ksi.

o.oo3

FIG.5-2 UNIAXIAL STRESS-STRAIN CURVES

 -88-

3loo £ k s i
t

• kl x

-2^00 0 £ Ksi ^ 40KSI

SWin Diagram

3 tcsl 3KSI*

3.1

9./

2<^ooo £ Ksi *z»o j Ks

Stress Pi^gra^ ( £ t = o. oo3)

5TEEL FORCE = T = 136GO 2

' ' B
X * £ C > IOI.G kips

CONCRETE- FOR.CE - C - ( I 8 & o o £ x - 3. Um- e * id ) k ;

fS £ $.ooo
6

= (36x-M^xl0 y -SMUtt — ) Kips , £ >. 0 0 0 ^

3 o . o g X Kips £ - .003
c

FI<r. 5 * - 3 .
 MAX. CONCRETE STRAIN
FI6r. 5 - 4
 STEEL FORCES Cio^lb) HIGH OF BEAM (in.)

o ro -t cs o > 5 -

7f oO to
% ?
n
h m
n
c
c 70
CU5 o
(» 3
in <£>oi
I i
CJ1 m

m
2
<0

II

era
m
Oo

7J?
 - 9 1 -

P = 13 kips.

I I | I
O I 2 3 KSI-

24

2o • — y

12
\ •
i
\

\ I
\\ A\ 1
:

.8 r

o 2 4 6 8 ' °

DISTANCE FROM FIXED END op SEAM (FTO

IZ 2 «f 8 IO

10

2.

TheoreticeJ 0cxlw.«5

Fig. 5-G
 p = 15 tan
I I I I
0 1 2 3 K5i.

( 1 » 1

1 * i

1
<

V
: 4

\
r--

\
k >

O 1 14- (, g 10

DISTANCE FROM FIXED END of BEAM (FT.)

o 2 >f b 8 10
 -93-

p zz II kZPS.

r m
0 1 2 3 «CS».

i 1
i
> • >

20
r »

>

>

> •

CO

\
•

\ \ A
»

M
O >

o 2 f 6 8 '°

DISTANCfc FROM FIXED END OF 6EAM (FT.)

( 1 0 2 <+ fc 8 JO
 -94-

P = 20

III!
0 1 23 «1.

—>

DISTANCE FROM FIXED END oF- BEAM CFT-)

2 A 6 8 «0

-£
*o

IU
o
m
o

Ul
Ul
I—
 P = ZM- «rs.

Fl<T. 5-10
 -96-

P = 2 6 kips.

I II I
0 1 2 3 Ksi.
 REINFORCEMENT

PUSTIC REGION

FIG.5-H PLASTIC REGION, CRACKED REGION

 -98-

CHA'PTER 1 1 1

Conclusions

A nonlinear finite element program has been developed

to o b t a i n the e l a s t o - p l a s t i c s t r e s s and strain distribution

In a plane stress continuum composed o f a perfectly plastic

and a linear strain-hardening material.

Since the program Is based on D.C. Drucker's definition,

it is very general and good for any c r i t e r i o n or for different

forms of hardening rule.

A combination of the load increment technique and the

modified Newton-Raphson method was u s e d to solve the nonlinear

equilibrium equation and shown to be p r a c t i c a l and economical

In computer time. For collapse Investigations, this procedure is

non-convergent or converges very slowly; this is clearly a

result of the iterative technique. Furthermore, It is known that

when a large proportion of the region becomes plastic the

c o n v e r g e n c e may be slow.
 99-

The p r o g r a m was e x t e n d e d to deal with reinforced

concrete members; for the concrete., a bilinear elastlc-

perfectly p l a s t i c curve was assumed for comparison and

an e l a s t i c curve up to cracking In tension; a bilinear

elastIc-perfectly plastic curve was used for the steel.

It was found that stress and crack distribution could

be predicted with reasonable accuracy until the ultimate

moment was a p p r o a c h e d . It appeared that the Iterative

method failed to converge s u c c e s s f u l l y at the yield

moment. It is possible that an e l a s t i c - s t r a i n hardening

curve for the concrete might have extended the usefulness

of these results.
 -100-

APPENPIX

XfclL P R I N C I P L E 0_F MINIMUM P O T E N T I A L ENERGY F f i R THE I N C R E M E N T A L FORMULATION

A brief mention i s made here concerning the p r i n c i p l e of

stationary potential energy applied to nonlinear problems in

solid mechanics. As stated earlier, the incremental theory is fre-

quently employed in the a n a l y s i s of nonlinear problems, thus it

is necessary t o expand the p r i n c i p l e to permit the determination

of an incremental state superimposed upon a body w i t h an existing

equilibrium stress state.

Let us c o n s i d e r that the loads on t h e body are increased

monotonlcal l y , and denote the s t a t e s of the body as Gr, G r . - q - ^ fi-^. •

we a s s u m e that the <T-state has been found a n d we w a n t to have

an incremental formulation (incremental equilibrium equation) for

the analysis of the £ M t | -state. This increment from ^ m to Cr W H

is small enough to allow linearization of all the f i e l d equations

w.r.t. the incremental quantities. Far this case, we may write

the potential energy functional for the Gr mn state as follows

V
P=V + V
f' = ?[-J^Tr^T )(«r^« ods i l

(A-1)
 -101-

where the internal energy function W is a function of

stresses or of strains. For the present, let us take

w
= w
^ y ) . (A-2)

T h e n
j i.

One form of the first law o f thermodynamics for a

continuous medium, where heat transfer terms are neglected

and displacement gradients are assumed small, may be written:

and Eq.(A-3) may be w r i t t e n as

( <r~ - —)
Noti ng that Eq.(A-5) must be an identity for any arbitrary dSy ;

we d e d u c e that the coefficient of must vanish, i.e.

n i} 'J » ( A
" 6
>

By T a y l o r ' s Theorem, the Increment of the internal

energy function AW c o r r e s p i nd I n g to increments of the

C|j Is
 -102-

Substituting Eqs.(A-6) and ( A - 7 )into (A-8), gives

aw = ifjoij + ( A . 9 )

Expanding Eq.(A-l), i n view of Eq.(A-9), leads to

Since the 5^-state i s i n equi1ibrlum, theprinciple of (A-10)

virtual displacements leads t o :

I [- f T f ^ i <*S " J n^^dv] +J j O J J ^ & I dV =0

and we have

1
' n V
n
 -103-

Taklng variations of Eq.(A-ll) w.r.t. a l l admissable incremental

displacement fields leads to the v a r i a t i o n a l principle:

SV = S [ 2 [ - {
p s 4T 4Ms-
L j>F;*Mv ij^(r^£ijdvj][ -
+
o

or in matrix form (A-12a)

(A-12b)

It is interesting to note that Eq.(A-12) h a s t h e same

form as the l i n e a r elastic case, Eq.(3-7), except that a l l

the state variables T , u , F , 0~, £ a r e now i n i n c r e m e n t a l forms.

 o

-104-

BlBLIOGRAPHY

1. V.V. Novozhllov: Foundations of Nonlinear Theory of Elasticity

Graylock P r e s s , Rochester, N.Y., 1953. P.128

2. O.C. Z i e n k i e w i c z : The F i n i t e Element Method in Engineering

Science, McGraw-Hill, London, 1971.

3. N.D. Nathan: F i n i t e Element Formulation of G e o m e t r i c a l l y Non-

l i n e a r Problems of E l a s t i c i t y , J a p a n - U . S . Seminar on
M a t r i x Methods o f S t r u c t u r a l A n a l y s i s and D e s i g n , Tokyo
1969, pp. 415-437.

4. R.H. Gallagher: Stress A n a l y s i s of Heated Complex Shapes, J . Am.

R o c k e t S o c . 3 2 , 1 9 6 2 . pp 7 0 0 - 7 0 7

5. A. Mendelson: P r a c t i c a l s o l u t i o n of P l a s t i c Deformation Problems

in the E l a s t i c - p l a s t i c R a n g e , NASA T R R 2 8 , 1959.

6. J/H» A r g y r i s : M a t r i x Methods of S t r u c t u r a l A n a l y s i s . A P r e c i s
of Recent Developments Proceedings, Fourteenth Meeting
o f S t r u c t u r e s a n d M a t e r i a l s P a n e l , AGARD, 1963.

7. J.H. P e r c y : A Study of M a t r i x A n a l y s i s Method f o r Inelastic

S t r u c t u r e s , RTD-TDRr63-4032, October 1963.

8. W.R. J e n s e n : M a t r i x A n a l y s i s methods f o r A n i s o t r o p i c Inelastic

Structures, AFFDL-TR-65-220, 1966.

9. P.V. Marcal: Comparative Study of Numerical Methods of Elastic-

Plastic Analysis, AIAA J . V o l . 6, N o . l , 1967, pp.157-58.

10. G. P o p e : The a p p l i c a t i o n o f t h e m a t r i x d i s p l a c e m e n t m e t h o d i n
plane e l a s t o - p l a s t i c problems, Proc. Conf. Matrix Meth.
S t r u c t . Mech. Wrigh-Patterson A i r Force Base, Ohio, 1965
PP.635-54.

11. T.L. Swedlow: E l a s t o - p l a s t i c s t r e s s e s in C r a c k e d P l a t e s , C a l c i t ,

R e p o r t SM.. 6 5 - 1 9 , C a l i f o r n i s I n s t i t u t e o f T e c h n o l o g y , 1965.

12. S.F. Reys: E l a s t o - p l a s t i c A n a l y s i s of Underground Openings by

the F i n i t e Element Method, P r o c . 1 s t . Int. Congr.
Rock M e c h a n i c s , 1 1 , 4 7 7 - 8 6 , Lisbon 1966.

13. P.V. M a r c e l : E l a s t i c - p l a s t i c a n a l y s i s of two dimensional Stress

S y s t e m s by t h e F i n i t e E l e m e n t M e t h o d , Int. L. Mech ScI.,
9, 143-55, 1967.

14. Y. Yamada: S t i f f n e s s M a t r i x in P l a s t i c - e l a s t i c Problems of

C o n t i n u a , S e i s a n Kenkyn ( m o n t h l y J . o f I n s t i t u t e of
Industrial S c i e n c e ) , V o l . 19, N o . 3 , 1967, pp. 75-76.
 -105-

15. O.C. Z i e n k i e w i c : E l a s t o - p l a s t i c Solutions of Engineering Problems

" i n i t i a l s t r e s s " , F i n i t e Element Approach. Int. J . Nun.
M e t h . E n g i n e e r i n g , 1, 7 5 - 1 0 0 , 1 9 6 9

16. Y. Yamada: Recent J a p a n e s e Developments in M a t r i x Displacement

Method of E l a s t i c - p l a s t i c p r o b l e m s , J a p a n - U . S . Seminar
on M a t r i x M e t h o d s o f S t r u c t u r a l A n a l y s i s and D e s i g n ,
Tokyo, 1969.

17. R. H i l l : The Mathematical Theory of P l a s t i c i t y , Oxford, Univ.

Press, 1 9 5 0 . P. 3 1 7 .

18. G.C. Nayak: E l a s t o - p l a s t i c S t r e s s A n a l y s i s . A G e n e r a l i z a t i o n

for various Constitutive relations including strain
S o f t e n i n g , I n t . J . Num.Meth. Engng. 5, 113-35,1972.

19. S. Tlmoshenko- and J . N . Goodier: Theory of Elasticity, MacGraw-

HIU Book Co. lnc.,N.Y., 1951,p.4.

20. H. Jeffreys: Cartesian Tensors, Cambridge, Univ. Press, 1931.

21. Y.C. Fung: Foundations of Solid Mechanics, Prentice-Hall, Inc.,

N . J . , 1965, p.100.
22. I.S. S o k o l n i k o f f : Mathematical Theory of Elasticity, McGraw-Hill
Book C o . I n c . , N . Y . 1956.p.66.
23. C.E. Pearson: T h e o r e t i c a l E l a s t i c i t y , Cambridge, Mass. Harvard
U n i v . P r e s s , 1 9 5 9 . p. 9 7 .

24. Y.C. Fung: Foundations o f Solid Mechanics, Prentice-Hall, Inc.,

N.J.,1965p. 137.

25. W. Johnson: P l a s t i c i t y for Mechanical Engineering, VAN Nostrand

C o . , L t d . London 1962. p.62.

26. A. Mendelson: P l a s t i c i t y - T h e o r y and Application, Macmillan, NVY".

1968. p.104.

27. R. H i l l : The M a t h e m a t i c a l Theory of P l a s t i c i t y , Oxford, Press,

1 9 5 0 . p. 39.

28. D.C. D r u c k e r : Some I m p l i c a t i o n o f Work H a r d e n i n g and Ideal

P l a s t i c i t y , Quart. Appl. M a t h . , 7, 1 9 5 0 , pp.411-18.

29. D.C. Trucker: Handbook of Engineering Mechanics, Chapter 46,

1962.

30. E.R. O l i v e i r a : T h e o r e t i c a l Foundations of the F i n i t e Element

Method, I n t e r n a l J . S o l i d s S t r u c t u r e s . 1968. Vol.4
pp.929-952.
 106-

31. T.H. P l a n and P. T o n g : B a s i c o f Finite Element Methods

for Solid Continua, Int. J . for Num. M e t h . in
Eng. V o l . 1, 3 - 2 8 , 1969.

32. R.H. G a l l a g h e r : An O v e r v i e w a n d Some P r o j e c t i o n s , Japan-

U.S. Seminar on M a t r i x Methods of S t r u c t u r a l An-
a l y s i s and D e s i g n , T o k y o , 1969, pp. 3-22.

33. D.R. B l a n d : The A s s o c i a t e d F l o w Rule of Plasticity,J. Mech.

Phys. S o l i d s , 6,71-78.

34. P.S. T h e o c a r i s and E. M a r k e t o s : E l a s t i c - p l a s t i c A n a l y s i s

of Perforated thin s t r i p s of a strain-hardening
M a t e r i a l , d. Mechanical Physical S o l i d s , 12,1964,p.
377-90.

35. O.C. Z i e n k i e w i c z : E l a s t o p l a s t i c S o l u t i o n s of Engineering

Problems-Inltlal S t r e s s F i n i t e Element Approach, Int.
J . f o r Num. M e t h o d i n E n g . 1 , 1 9 6 9 . p p . 75-100.

36. D.N. A l l e n a n d R. V. S o u t h w e l l , Phil. Trans. R. Soc.

A242, 379, 1950.