Sie sind auf Seite 1von 13

Finite Elements in Analysis and Design 7 (1990) 229-241

229

Elsevier

Use of incompatible displacement modes


for the calculation of element stiffnesses or stresses
Edward L. Wilson and Adnan Ibrahimbegovic
Department of Civil Engineering University of California, Berkeley, CA 94720, U.S.A.
Received March 1990
Revised May 1990
Abstract. The addition of incompatible displacement modes to lower-order displacement-based elements is
re-evaluated. Recent research has indicated that a simple numerical correction can be applied to the shape
functions in order that the constant-strain patch test is passed. In this paper a new method of stress recovery is
presented in which incompatible modes are introduced. A least-square approximation is used to calculate
element stresses which are in microscopic equilibrium and tend to be in global equilibrium with the applied
nodal loads on the finite element assemblage. Also, a consistent and robust method for the evaluation of thermal
stresses is presented. The numerical methods presented are general and can be applied to all displacement-based
finite elements. The basic formulation and examples which are presented in this paper are in three-dimensional
elasticity using the eight-node isoparametric elements. Accuracy of the element is illustrated and it is
demonstrated that both displacements and stresses are alsmost identical to those produced by Pian's hybrid
stress elements.

Introduction

Incompatible displacement modes were initially added to isoparametric elements in order to


avoid shear locking in the early 1970's [12]. The amplitude of the modes was calculated by the
minimization of the potential energy; therefore, the addition of modes improved microscopic
equilibrium within the elements. The use of the method for lower-order rectangular or
parallelogram elements in both two- and three-dimensional problems gives the most impresive
performance. However, for the irregularly shaped elements, the consistent stresses were very
inaccurate. Prior to the introduction of incompatible modes it was assumed that displacement
compatibility between elements was absolutely necessary if the solution was to converge to the
exact solution as the mesh is refined. Clearly, the behavior of rectangular elements with
incompatible modes eliminated the formal "displacement compatibility" restriction. In order
that the finite element method would have an appropriate mathematical basis, the restriction of
the " p a t c h test" was introduced as a "relaxed restriction" on displacement compatibility [2,7].
In 1976 an ad-hoc method [8] was presented which forces non-rectangular elements with
imcompatible modes to pass the patch test while maintaining their high accuracy for elements
with parallel sides. This approach has recently been re-evaluated within a more formal
variational framework. Incompatible modes can be interpreted either as an enhanced strain
field or as a well-known B-bar formulation [14, p. 334]. It has been found that the modified
incompatible elements which pass the patch test possess excellent accuracy.
One of the most justifiable criticisms of the compatible displacement-based finite elements is
that the calculated displacements and stresses, respectively, are always less or equal to the true
displacements and stresses. From a theoretical point of view this is "nice", since it gives a lower
bound to the exact solution. However, for the practical design engineer, the use of displace0168-874X/90/$03.50 1990 - Elsevier Science Publishers B.V.

E.L. Wilson,A. Ibrahimbegooic/ Incompatible displacementmodes

230

2~,~

I r
Z

Degrees of F r e e d o m
Fig. 1. Eight-node isoparametric solid element.

ments and stresses which are a lower bound of the true values is very dangerous. Also, for most
structures accurate stresses are far more important than accurate displacements. Unfortunately,
most recent theoretical comparisons of the performance of different finite element formulations
compare the accuracy of computed displacements, and tend to neglect the accuracy of stresses.
In this paper a new method of evaluation of element stresses is presented in which the
accuracy of the calculated stress is independent of the accuracy of the computed displacements.
For example, if a cantilever beam which is modeled by compatible rectangular finite elements is
loaded by an end moment, the end displacements may be a small fraction of the true
displacement due to shear locking phenomena. However, if the element stresses are calculated
directly from element nodal forces (obtained from element stiffness matrix and nodal displacements that are both incorrect), the exact bending stresses can be evaluated. The method of
stress evaluation presented in this paper is based on a direct equilibrium approach. It can be
used for both compatible displacements and incompatible displacements to produce values of
stresses which can improve the confidence level of the engineering design where stress accuracy
is of major importance.

Three-dimensional element with incompatible modes


The element displacements u e = ( u x, Uy, uz) x approximation for the eight-node isoparametric
element (Fig. 1) with nine incompatible modes are of the form
8

ue=(r,s,t)

= Ni(r,s,t)u,+(1-r2)al+(1-s2)a2+(1-t2)a3,

(1)

1=1

where u~ are nodal values of the displacements, the N i ( r , s, t) are the standard isoparametric
shape functions (see, e.g. [14]), while a j are incompatible modes interpolation parameters.
The element strains, ~, are then given as
8

= E n,,,,+ E
I=1

J=l

(2)

E.L, Wilson, A. Ibrahimbegovic / Incompatible displacement modes

231

where strain-displacement matrices B and G are obtained by differentiation of the displacement interpolation in expression (1) and ~ = (ex, ey, ez, ~xy, 7yz, Yzx)TThe strain energy of the element is given as
W= ffe

dr,

(3)

or, by utilizing (2),

w = f~TBu dV+ fvoTGa dV.

(4)

Similar to the method presented in [3] we impose the requirement that under the state of
constant stress, the strain energy associated with the incompatible modes vanishes (which is
equivalent to the patch test of Irons [2]):

OXfvG dVa= O.

(5)

For constant non-zero stress over the element o is factored out of the integration and (5) can
be simplified to

ffi

dV = 0.

(6)

Equation (6) can be satisfied by adding a constant correction matrix 17c to the matrix t7 (i.e.
6~ = 17 + Go) such that

fvd

dV=

fv(ao + a)

d V = O,

(7)

from where, by using the fact that Gc is a constant, it follows

o___afy
The correction matrix Gc is evaluated numerically before the element stiffness is calculated.
The same integration formula must be used to calculate Gc as the one used for element stiffness
computation, since the correction matrix is applied at each integration point.
Minimizing the potential energy for a single element, we can write

[E

=F,

(9)

where

fpTcJ~ dV,

(10)

/~ m fvGTc]~ dV,

(11)

It = f/Tct~, dV,

(12)

I~ =

in which C is the general anisotropic (6 6) stress-strain matrix. The (24 24) element
stiffness matrix is calculated from (9) by using the static condensation [10] to obtain

R= K - EH-1E T.

(13)

E.L. Wilson, A. Ibrahimbegooic / Incompatible displacement modes

232

Calculation of element stresses


The basic equilibrium equations of the finite element assemblage, as produced by the
application of the principle of minimum potential energy, can be written as a summation of
element contributions in the following form:
R=

E
elements

/(u=

F,

(14)

elements

w h e r e / f is an element stiffness, u are element nodal displacements and F are the generalized
element loads acting on the element nodes (see also equation (9)). The external nodal points
stress resultants, R, are the specified concentrated point loads, the body forces which are
integrated over the element volume, the consistent nodal loads associated with surface tractions
and thermal loads. Except for thermal loads, these nodal loads are exact or an exact statically
equivalent load system. Therefore, from (14) the element nodal forces are in exact equilibrium
with the external nodal points stress resultants R. After the nodal displacements u are
calculated, the element nodal forces F are calculated from
F = l~u.

(15)

For a statically determinate system, the element nodal forces are not a function of element
stiffness or nodal displacements. For highly indeterminate systems the element nodal force
distribution is a function of the relative stiffness of the elements. Therefore, the nodal forces
are not necessarily correct for any one element. However, the sum of the element nodal forces
will always be equal to the applied nodal forces.
The original development of the finite element method was presented as an extension of
structural analysis in which nodal point equilibrium was the fundamental starting point. Hence,
the accuracy of the element nodal forces was apparent. In most of the modern research work in
computational mechanics the use of the abstract variational methods have tended to make the
important equilibrium property of the nodal forces obscure. Element stresses can be calculated
directly from element nodal forces by the use of virtual work and the method of least squares.
In the case of three-dimensional elements the basic form of the stress distribution within the
element is assumed to be of the following form:
Txy = fll + f12 x + fl3Y + f14 z '

(16)

~'yz = f15 + f16 x + fl7Y + f18 z '

(17)

~x = f19 + fllo x + t i n Y + fl12 z,

(18)

x = fl13 - (f13 + f l , 2 ) x + fl14Y + B15Z,

(19)

e = B16 + #17 x - (B2 + B s ) y + #,8 z,

(20)

Oz = ill9 -t- 20 x -I- fl21Y -- (fllO + f17) z '

(21)

or, in a matrix form


O" = P ~ .

(22)

N o t e that the assumed stress distribution satisfies the microscopic three-dimensional equilibrium equations. The nodal forces can be expressed in terms of the assumed stresses by the
application of principle of virtual work in which the virtual displacements are of the same form

E.L. Wilson,A. Ibrahimbegovic / Incompatible displacement modes

233

as the basic displacement approximation (1). Including the incompatible virtual displacements,
these forces are
(23)
or

F = Qfl,

(24)

where

The (33 21) matrix Q is calculated using standard numerical integration.


The overdetermined system of equations (24) is then solved by the least squares method,
which involves the solution of
[ Q T Q I f l = QTF,

(26)

where [QTQ] is a 21 21 positive-definite symmetric matrix. The vector F contains 24 non-zero


terms and 9 zero terms associated with the incompatible modes. After fl is evaluated for various
load conditions, the stresses at any point (x, y, z) within the element can be directly calculated
from (22).
It is interesting to note that the method of stress evaluation is valid even if incompatible
modes are not used in the element stiffness calculation. Also, the method does not have
numerical problems when the nodal points are coalesced and the element degenerates into a
wedge or a tetrahedron.

Consistent thermal stresses

In a case of thermal loads the previous method of calculating stresses may not be valid since
the thermal stresses may be a function of the stiffness. Therefore, it is necessary to introduce a
special approach for the calculation of thermal stresses.
The accurate evaluation of thermal deformation and stresses for temperature fields, which
vary significantly in space, is a problem which is of important practical interest. Displacementbased elements can have an incorrect stiffness due to shear locking or element distortion. If
initial thermal stresses are treated in a formal theoretical manner (via theory of minimum
potential energy), large errors in stresses may result.
The total strain at a point, in the presence of a change of temperature field, is defined as
e =

C - a o + ATet,

(27)

where the purely mechanical deformation is C - l o and the temperature expansion strain is ATa.
Both the anisotropic material properties C-1 and the thermal expansion coefficients a can be
obtained from simple laboratory experiments. The form of the total stress distribution used in a
displacement-based finite element formulation is
o = Ce + A T C a .

(28)

Normally, the consistent nodal loads are calculated using the appropriate derivatives of the
shape functions times the initial thermal stresses, (ATCa). However, an alternative approach,
based on physical intuition, can be used to calculate nodal loads.

234

E.L Wilson, A. Ibrahimbegovic / Incompatible displacement modes

First, let us assume that the temperature change with respect to the local d e m e n t reference
frame has the following form:
A T = b 1 + b2x + b3y + b4z ,

(29)

For three-dimensional elements with more than 4 nodes the constants b i can be determined by
the least-squares method.
Second, for the unconstrained displacements within the element which are associated with
the temperature change it is assumed that no shearing deformations exist. For orthotropic
thermal coefficients ax, Oty, Otz, the resulting displacements are assumed to be of the following
form:
UxT = bletxx + b2ax x2 + b3etxXy + b4a~,xz - b2ctyy 2 - b2az z2,

(30)

Uy T =

blOtyy + b2etyxy + b30tyy 2 + b4etyyZ - b30tx x2 - baotzz 2,

(31)

UzT

baOtzZ + b2ctzxz + b3a~yz + b40tz z2 - b4ax X2 - b40tyy 2,

(32)

Third, using these equations the free expansion thermal displacements u T at all nodes of a
finite element can be evaluated for a specified temperature change. The thermal nodal forces,
consistent with the element stiffness, can now be calculated from
FT =/~U T .

(33)

Finally, a completely consistent approach to the evaluation of d e m e n t stresses is now


possible. The nodal forces to be used for calculation of element stresses will now be
/~ =

F - F T,

(34)

where F are the calculated element nodal forces using (15).


For the special case in which the temperature change within the element cannot be
approximated by (29), it may be necessary to evaluate the temperature change error at the point
where the stress is to be calculated. The initiated restrained stresses associated with this error
can be added directly to the stresses calculated by using element nodal forces in (34).
It is clear that for the case of free expansion, in which the temperature change is linear in
space, the internal stresses will be identically zero regardless of what approximations were used
to form the dement stiffness.
The method of calculating thermal stresses which is presented here is valid for either regular
or distorted elements, with or without incompatible modes. It is an approach which is
independent of the approximations which are used to form the element stiffness. For example,
if we have a distorted mesh idealization of a cantilever beam which is subjected to a linear
temperature change in the thickness direction, the exact displacements and stresses will be
automatically calculated if this method is used for thermal nodal loads computation. Whereas,
if the formal variational method is used, incorrect displacements and stresses will be calculated.

Numerical

examples

A series of numerical examples is presented in this section. All results are obtained by a
research version of computer program SAP80 [11]. The incompatible modes given in (1) are used
to supplement the linear displacement fields of the 8-node brick element.

E.L. Wilson,A. Ibrahimbegovic/ Incompatibledisplacementmodes


E=I.
v= 0.25

/ =6
h =3

F=l.5
P = 3.

235

load load
case case
1

mv
regular mesh

distorted mesh

B.C. for load


case 2 only

Fig. 2. Models for patch test.

The patch test


In order to demonstrate that the formulation given herein satisfies the patch test, the same
example as given in [8] is solved. The first load case is a uniform tension under axial load and
the second load case is a uniform bending under the end moment (the higher order patch test
[9]). For both meshes in Fig. 2 the patch test is satisfied (see Table 1). For the regular mesh
even the higher order patch test is satisfied.

Comparison with hybrid elements


The cantilever beam presented in Fig. 3 is used to evaluate the performance of the
two-dimensional hybrid element in [6] and three-dimensional hybrid element in [1]. In this
work the same example is used to demonstrate that the same accuracy can be obtained with the
least-squares stress recovery method. Two load cases, end moment and end shear, are considered. The same distorted mesh as shown in Fig. 3 is used as the one presented in [1,6]. The
d e m e n t performance is compared with several hybrid dements and presented in Table 2. In
addition, results obtained by the solid d e m e n t with incompatible modes given in [13] are
presented for comparison.
It is interesting to note that the present formulation gives approximately the same results as
the ones obtained by the hybrid element based on a linear variation of all stress components in
natural coordinates. The present formulation is superior to the incompatible modes solid
element of [13].

Accuracy ofstress recovery


The same cantilever beam shown in Fig. 3 is used to evaluate the performance of the element
when the mesh of trapezoidal elements is selected. It was shown in [4] that such a choice will

Table 1
Tip displacement--the patch test (Fig. 2)
Mesh
Regular
Distorted

Load
case
axial
axial

Horizontal
displacement
6
6

Load
case
moment
moment

Vertical
displacement
18
17.61

Exact solution

axial

moment

18

236

E.L Wilson,A. Ibrahimbegovic / Incompatible displacement modes

i'/2

h)ad case

load case

h
F
D,

[_ 2 _l_ 2 _L1 J 1 I_

_l
E = 1500
v = 0.25
P = 300
F = 11)00

i 1 i Ij _ 2 _13
_1_
r-- -1 - -1- _l)isl-~rted
._. ___. Mesh wI-

_1
-i

j_

_1

2 _11 i_

IJ_i_

_1 1 i_

J2.1.

1=10
h= 2
u nornlal sir.
~x vert. displ.

_1_ 2 _1

Trapeziodal Mesh
Fig. 3. Short cantilever beam.

c a u s e t h e l o c k i n g p h e n o m e n o n f o r a n y e l e m e n t w h i c h p a s s e s t h e p a t c h test. I n d e e d , f o r t h e
m e s h s h o w n i n Fig. 3 t h e t i p d i s p l a c e m e n t s f o r b o t h l o a d c a s e s a r e a p p r o x i m a t e l y 2 5 % s m a l l e r
t h a n t h e e x a c t s o l u t i o n . H o w e v e r , t h e c o m p u t e d s t r e s s e s a r e w i t h i n 5% o f t h e e x a c t v a l u e s .
Very similar conclusions can be drawn for the standard compatible eight-node brick. The
results of this analysis are presented to demonstrate that benefits of stress recovery introduced
h e r e i n e x t e n d t o c o m p a t i b l e e l e m e n t s a s well. I n t h i s c a s e a s e r i o u s s h e a r l o c k i n g r e d u c e s t h e
d i s p l a c e m e n t a c c u r a c y b y a p p r o x i m a t e l y 50%. H o w e v e r , t h e s t r e s s e s c o m p u t e d w i t h t h e m e t h o d
i n t r o d u c e d h e r e i n a r e w i t h i n 10% o f t h e e x a c t . F o r c o m p a r i s o n w e a l s o p r e s e n t t h e s t r e s s e s f o r
compatible eight node element which are computed from nodal displacements in a standard
m a n n e r . T h e e r r o r i n c o m p u t e d s t r e s s e s i n t h i s c a s e is o f t h e s a m e o r d e r as i n t h e c o m p u t e d
d i s p l a c e m e n t s . A l l r e s u l t s a r e p r e s e n t e d i n T a b l e 3.

Table 2
Short cantilever beam--distorted mesh (Fig. 3)
Element

Load
case

Pian [6]
Cheung [1]
Wu [13]
Present

moment
moment
moment
moment

Exact solution

moment

Pian [6]
Chenng [1]
Present

shear
shear
shear

Exact solution

shear

Vertical
displacement
96.18
92.2
93.7
95.8
100
98.19
94.05
97.9
102.6

Error
(%)

Stress

Error
(%)

3.82
7.80
6.30
4.20

0.47
0.22
17.2
0.50

3014
3006.6
2484
3015

- 3000

4.30
8.33
4.58

- 4137
- 4125.3
- 4138.5

2.15
1.85
2.19

- 4050

E.L, Wilson, A. Ibrahimbegovic / Incompatible displacement modes

237

-T

1
Regular Mesh

E=IO 7
r=0.3
a = 0.0001
/=6
tt = 0.2
b=0.1
T=5

Parallelogram Mesh

Trapezoidal Mesh

Fig. 4. Long cantilever beam.

Thermal stresses recovery

The

long cantilever

accuracy

beam

of finite element

shown

in Fig. 4 is used

formulation.

as a standard

In this numerical

example

problem

[5] t o t e s t t h e

we demonstrate

the accuracy

Error
(%)

Stress

Error
(%)

46.6
46.6
23.7

- 2720.5
- 1473
- 2883.5

9.3
50.9
3.9

Table 3
Short cantilever b e a m - - t r a p e z o i d a l mesh (Fig. 3)
Element

Load
case

Compatible
Compatible a
Present

moment
moment
moment

Exact solution

moment

Compatible
Compatible a
Present

shear
shear
shear

Exact solution

shear

Vertical
displacement
53.439
53.439
76.252
100
57.728
57.728
80.115
102.6

- 3000

43.7
43.7
21.9

- 3784
- 2051
- 3860

6.6
49.4
4.7

- 4050

Stresses evaluated from the displacements.

Table 4
Long cantilever beam (Fig. 4)
Mesh:

Regular

Parallegram

Trapezoidal

Dement

Displacement

Stress

Displacement

Stress

Displacement

Stress

Compatible
Compatible ~
Present

0.09
0.012
0.09

0
244.2
0

0.09
0.006
0.09

0
346.6
0

0.09
0.002
0.09

0
270.7
0

Exact solution

0.09

0.09

0.09

a Formal computation of temperature-induced nodal loads.

238

E.L Wilson, A. lbrahimbegooic / Incompatible displacement modes

of the thermal stress recovery method. Temperature is distributed linearly through the thickness
decreasing from T = 5 C at the bottom to T = - 5 C at the top. Three finite element models
used in the analysis are presented in Fig. 4. The results of the analysis are given in Table 4. The
exact solution for the vertical tip displacement is 0.09, and all stresses are zero (statically
determinate structure). The thermal stress recovery method introduced herein, when used with
either compatible or incompatible solid elements, is capable of reproducing the exact solution.
The analysis based on formal computation of thermal stresses, however, exhibits very serious
locking and spurious stresses occur.

Nearly incompressible material


The thick-walled cylinder (see Fig. 5) is used to demonstrate the advantage of using
incompatible modes for nearly incompressible materials. The cylinder is made of isotropic
elastic material with Young's modulus E = 10 6, inner radius r i = 3 and outer radius ro = 9. The
cylinder is subjected to unit internal pressure.
For Poisson's ratio i, = 0.49 (nearly incompressible material) and v = 0.3 (normal compressibility) the maximum radial displacement and stresses distribution through the thickness
are summarized in Table 5. The nine-node isoparametric plane strain element gives slightly
more accurate results than the present four-node element for normal materials. However, for
nearly incompressible materials, the four-node element with corrected incompatible modes
produces superior results. The nine-node compatible element exhibits locking and the errors in
the computed stresses are large. For Poisson's ratio greater than 0.49 the accuracy of the
nine-node compatible element sharply deteriorates.

LY
thcq[Sot
UNOEFORMEO
SHAPE

OPTIDNS
HIRE FRAME

'E

SAP?~
Fig. 5. Thick-walledcylinder.

E.L Wilson, A. Ibrahimbegovic / Incompatible displacement modes

239

Table 5
T h i c k - w a l l e d c y l i n d e r u n d e r i n t e r n a l p r e s s u r e (Fig. 5)
Stress

Radius

Compatible
9-node

Present

Exact
solution

(a) P o i s s o n ' s r a t i o p = 0.30


Radial

3.00
3.56
4.20
5.20
6.75
9.00

Hoop

3.00
3.50
4.20
5.20
6.75
9.00

1.26
0.96
0.70
0.50
0.35
0.25

1.23
0.95
0.69
0.50
0.34
0.24

1.25
0.95
0.70
0.50
0.35
0.25

Axial

3.00
3.50
4.20
5.20
6.75
9.00

0.075
0.075
0.075
0.075
0.075
0.075

0.073
0.076
0.076
0.075
0.075
0.074

0.075
0.075
0.075
0.075
0.075
0.075

3.00

0.458 1 0 - 5

0.458 1 0 - 5

0.458 x 10- 5

Radial
displacement

0.98
0.69
0.44
0.24
0.09
0.01

0.97
0.69
0.44
0.24
0.09
0.02

1.00
0.70
0.45
0.25
0.10
0.00

(b) P o i s s o n ' s r a t i o u = 0.49


Radial

3.00
3.50
4.20
5.20
6.75
9.00

Hoop

3.00
3.50
4.20
5.20
6.75
9.00

1.45
1.14
0.88
0.67
0.48
0.35

1.23
0.95
0.70
0.50
0.34
0.23

1.25
0.95
0.70
0.50
0.35
0.25

Axial

3.00
3.50
4.20
5.20
6.75
9.00

0.32
0.32
0.30
0.29
0.26
0.22

0.120
0.125
0.124
0.125
0.123
0.122

0.123
0.123
0.123
0.123
0.123
0.123

3.00

0.503 x 1 0 - 5

0.503 x 1 0 - 5

0.503 1 0 - 5

Radial
displacement

A standard
incompressible
strain

energy.

incompatible

method

0.79
0.51
0.26
0.07
0.05
0.10

of improving

materials
However,
modes

is to use
this

approach

the

performance

selective

approach
presented

reduced

cannot

be

0.97
0.69
0.44
0.24
0.09
0.02

of compatible
integration

used

in this paper

for
does

on

finite
the

anisostropic
not

have

1.00
0.70
0.45
0.25
0.10
0.00

elements

volumetric
material;

this limitation.

for nearly
part

of the

whereas,

the

240

E . L Wilson,A. Ibrahimbegovic / Incompatible displacement modes

q)

1.0

E
0.9
O
O_

~3~- 0.8
O_

~-- 0.7
-0

.N__0.6
0

E
o 0.5

Present
Pion (1984)

.......

~
Mesh

~
Distortion

~
Parameter

s
a

F i g . 6. M e s h d i s t o r t i o n s e n s i t i v i t y .

Mesh distortion sensitivity


The test for the mesh distortion sensitivity is performed on the cantilever beam under
uniform bending. Mesh distortion is proportional to the parameter a (see Fig. 6). Starting from
a regular 2-dement mesh, the central nodes are displaced by ___a until the elements take almost
triangular shapes. The results of the test are presented in Fig. 6, along with the results obtained
by the hybrid element of Pian and Sumihara [6]. We note that the element performance is
essentially the same as the hybrid element performance.

Final remarks
The numerical technique for correcting the strains associated with incompatible modes is
more general than the application shown in this paper. For example, cubic incompatible modes
can be added to the compatible nine-node plane element in which the exact results can be
calculated for a one d e m e n t model of a cantilever b e a m loaded with an end shear force.
It has been demonstrated that this approach produces elements with approximately the same
accuracy, of both displacements and stresses, as the results obtained from the well-known
hybrid dements.
The new nodal force equilibrium approach for evaluation of element stresses appears to
eliminate the major criticism that displacement based elements do not satisfy equilibrium.
There appears to be no reason why this approach cannot be extended to other types of
elements such as plates and shells. Also, the method may be of significant value for nonlinear
materials.
The strength of material method which was presented for the evaluation of thermal loads
eliminates the errors associated with mesh distortion. Also, it can be used for different types of
elements.

Acknowledgement
The research presented in this paper was partially supported by Computers and Structures,
Inc., Berkeley, California.

E.L Wilson, A. Ibrahimbegovic / Incompatible displacement modes

241

References
[1] CHEUNG,Y.K., and W. CHEN, "Isoparametric hybrid hexahedral elements for three dimensional stress analysis",
Int. J. Numer. Methods Eng. 26, pp. 677-693, 1988.
[2] IRONS, B.M., and A. RAZZAQUE,"Experience with the patch test for convergence of finite element method, Proc.
Mathematical Foundations of the Finite Element Method, edited by A.R. AZlZ, Academic Press, New York, 1972.
[3] JETTEUR, P., and F. FgEY, "A four node Marguerre element for non-linear shell analysis", Eng. Comput. 3, pp.
276-282, 1986.
[4] MACNEAL, R.H., "A theorem regarding the locking of tapered four-noded membrane elements", Int. J. Numer.
Methods Eng. 24, pp. 1793-1799, 1987.
[5] MAcNEAL, R.H., and R.L. HARDER, "A proposed standard set of problems to test finite element accuracy", Finite
Elements in Analysis and Design 1, pp. 3-20, 1985.
[6] PLAN,T.H.H., and R. SLrMIHARA,"Rational approach for assumed stress finite elements", Int. J. Numer. Methods,
Eng. 20, pp. 1685-1695, 1984.
[7] STRANG, G., and G.J. FIX, An Analysis of the Finite Element Method, Prentice-Hall, Englewood Cliffs, NJ, 1973.
[8] TAYLOR, R.L., P.J. BERESFORD and E.L. WILSON, "A non-conforming element for stress analysis", Int. J. Numer.
Methods Eng. 10, pp. 1211-1219, 1976.
[9] TAYLOR, R.L., J.C. Sh~tO, O.C. ZI~NKIEWICZ and A.C. CHAN, "The patch test: a condition for assessing finite
element convergence", Int. J. Numer. Methods Eng., 22, pp. 39-62, 1986.
[10] WILSON, E.L., "The static condensation algorithm", Int. J. Numer. Methods Eng. 8, pp. 199-203, 1974.
[11] WILSON, E.L., "SAPS0. Structural Analysis Program for small or large computer systems", Proc. CEPA Fall
Conference, 1980.
[12] WILSON, E.L., R.L. TAYLOR, W.P. DOHERTV and J. GHABOUSSI, "Incompatible displacement models, in:
Numerical and Computer Methods in Structural Mechanics, edited by S.J. FENVES, N. PERROIqE, A.R. ROnlNSON
and W.C. SCHNOnRICH, Academic Press, New York, 1973.
[13] Wu, C.C., M.G. HUANG and T.H.H. PLAN, "Consistency condition and convergence criteria of incompatible
elements: General formulation of incompatible functions and its application", J. Comput. Struct. 27, pp. 639-644,
1987.
[14] ZIENKIEWlCZ, O.C., and R.L. TAYLOR, The Finite Element Method: Basic Formulation and Linear Problems, Vol. I,
McGraw-Hill, New York, 1989.

Das könnte Ihnen auch gefallen