Sie sind auf Seite 1von 9

418 ACI Structural Journal/July-August 2002

ACI Structural Journal, V. 99, No. 4, July-August 2002.


MS No. 01-275 received August 21, 2001, and reviewed under Institute publication
policies. Copyright 2002, American Concrete Institute. All rights reserved, includ-
ing the making of copies unless permission is obtained from the copyright proprietors.
Pertinent discussion will be published in the May-June 2003 ACI Structural Journal if
received by January 1, 2003.
ACI STRUCTURAL JOURNAL TECHNICAL PAPER
This paper presents a nonlinear finite element procedure for the
analysis of reinforced concrete shells using the four-node quadri-
lateral flat-shell element with drilling rotational stiffness. A layered
approach is used to discretize, through the thickness, the behavior of
the concrete and the behavior of the reinforcement. The analysis
takes into account material nonlinearity by incorporating tensile,
compressive, and shear models of cracked concrete, in addition to
a model for the reinforcing steel. Using the smeared-crack method,
the cracked concrete is treated as an orthotropic nonlinear material.
The steel reinforcement is assumed to be in a uniaxial stress state and
to be smeared in a layer. The constitutive models, which cover the
loading, unloading, and reloading paths, and the developed finite
element procedure predicts with reasonable accuracy the behavior of
reinforced concrete shells subjected to different types of loading. The
proposed numerical method for nonlinear analysis of reinforced con-
crete shells is verified by comparison with reliable experimental results.
Keywords: load; reinforced concrete; strength.
INTRODUCTION
In recent years, reinforced concrete shells have been
widely applied to underground tanks, nuclear waste containers,
and offshore structures. Finite element analysis of such
structures has become increasingly important because it is
generally not possible to determine deformation and failure
behavior by conventional procedures, and experimental
studies on these structures are very expensive.
Considerable effort has been made to develop suitable
finite element analysis for application to reinforced con-
crete shells (ASCE 1993; Hinton and Owen 1984). Much of
this research has focused on developing specialized element
and efficient solution algorithms, with insufficient attention
given to the implementation of realistic constitutive models
that accurately predict the behavior of concrete shells (Polak
and Vecchio 1993).
In the present study, models were developed to address
material nonlinearity by incorporating tensile, compressive,
and shear models for cracked concrete in addition to a model
for the reinforcing steel, which uses the smeared-crack
approach. The impetus for developing these models was to
model cyclic behavior by proper theoretical representation
of the material parameters.
The primary objective of this study was to develop a new
finite element formulation for the nonlinear analysis of shell
structures. To analyze reinforced concrete shells with highly
nonlinear behavior, the layer method was introduced, assuming
that several thin-plane stress elements are layered in the
direction of thickness. The cross section of reinforced
concrete is divided into concrete and steel layers. Each layer
consists of four-node flat-shell elements. The flat-shell element
was developed by combining a membrane element with a
drilling degree of freedom and a plate-bending element.
Thus, the developed element possesses six degrees of freedom
(DOF) per node, allowing for an easy connection to other types
of finite elements with six DOFs per node and a three-dimen-
sional beam-column element, and providing a much improved,
more robust analysis procedure than currently exists.
RESEARCH SIGNIFICANCE
A four-node flat-shell element was developed for the
nonlinear analysis of reinforced concrete shells. Through
layered formulation, the model successfully utilizes consti-
tutive models of cracked concrete by considering both the
concrete tension stiffening/compression softening and the
reinforcing bar under a one-dimensional stress condition.
This element was incorporated into an existing general-
purpose finite element analysis program. A distinctive
characteristic of the element is its capability to simulate the
behavior of shells under cyclic loading and drilling rotational
stiffness. Consequently, the response of reinforced concrete
shell structures can be predicted accurately using the proposed
nonlinear finite element procedure.
FINITE ELEMENT FORMULATION
Presented herein is a finite element formulation for the
analysis of reinforced concrete shells using a four-node
quadrilateral thin flat-shell finite element with six DOFs per
node. The sixth DOF is obtained by combining a membrane
element with a normal rotation
z
, the so-called drilling
degree of freedom, and a discrete Kirchhoff plate element.
In small displacement models of flat-shell elements, the
effects of membrane and bending strain are not coupled in
the energy expression within the elements. The coupling
between in-plane membrane (plane stress) and out-of-
plane flexural response using the thin-shell application is
complex. Coupling occurs only on the interelement boundary;
therefore, a flat-shell element that combines a plane-stress
element and a plate-bending element is suggested. (It
should be noted that the shell element should not be applied
to thick shells influenced by punching shear because it ignores
the out-of-plane shear effects.) These modeling assumptions
are shown in Fig. 1.
Flat-shell finite elements may be formulated through the
use of a variational formulation that includes an independent
rotation field for the drilling degree of freedom. The drilling
degree of freedom may be interpreted physically as a true
rotation of the vertex bisecting the angle between adjacent
Title no. 99-S43
Nonlinear Analysis of Reinforced Concrete Shells using
Layered Elements with Drilling Degree of Freedom
by Tae-Hoon Kim, Kwang-Myong Lee, and Hyun Mock Shin
419 ACI Structural Journal/July-August 2002
edges of the finite element. A schematic of the angle bisector
and associated partial derivatives in element displacement is
shown in Fig. 2. The drilling degree of freedom is defined as
(1)
where u and v = in-plane displacements along the x- and y-
axes, respectively; and = drilling degree of freedom.
The variational formulation suggested by Hughes (1989) is
(2)
where u and = trial displacements and rotations on the
region , respectively; f = general external forces; and
= penalty. The corresponding variational formulation is
0 = D

(u, )(u, ) = (3)


The matrix counterpart of the variational equation (Eq. (3))
for one element may be expressed as
[K
m
][q] = [f ] (4)
where f = general external forces; q = nodal parameter vector;
and K
m
= final element membrane stiffness matrix with the
drilling degree of freedom (Lanheng 1994).
The plate-bending component of the shell element corre-
sponds to the 12-DOF discrete Kirchhoff quadrilateral (DKQ)
plate element, which was derived in detail using the discrete
Kirchhoff technique (Lanheng 1994). The DKQ element
formulation is based on the discretization of the strain energy.
The model neglects the transverse shear strain energy, that is
(5)

1
2
-- -
v
x
-----
u
y
------
,
_
=

u , ( )
1
2
-- - symm u ( ) C symm u ( ) d

=
1
2
-- - skew u ( )
2
u f d

d

+
symm u ( ) C symm u ( ) d

skew u ( )
T
skew u ( ) u f d

d

+
U U
b
e
e
n

=
with
(6)
where U
e
b
= element strain energy due to bending; x = curvature;
D
b
= moment-curvature relation; and A
e
= element area.
The element needed to be based on a three-dimensional
elastic formulation. To analyze reinforced concrete shells
with nonlinear behavior, the layer method was used, assuming
that several thin-plane stress elements were layered in the
direction of thickness.
In the layered-element formulation, the shell was divided
into several paneled layers, and two-dimensional constitutive
models were applied to take into account material nonlin-
earities. The constitutive matrix can be rotated from the
local axes to the global axes and added to the constitutive
U
b
e 1
2
--- [ ]
T
D
b
[ ] [ ] A
e
d
A
e

=
Tae-Hoon Kim is a PhD candidate in the Department of Civil Engineering,
Sungkyunkwan University, Suwon, Korea. He received his MSc in civil engineering
from Sungkyunkwan University. His research interests include nonlinear analysis and
design of concrete structures, constitutive modeling, and damage assessment.
ACI member Kwang-Myong Lee is an associate professor in the Department of Civil
Engineering, Sungkyunkwan University. He received his PhD in structural engineering
from the Massachusetts Institute of Technology, Cambridge, Mass. His research
interests include fracture mechanics of concrete, nonlinear analysis of concrete structures,
and high-performance concrete.
Hyun Mock Shin is a professor in the Department of Civil Engineering, Sungkyunkwan
University. He received his PhD in structural engineering from the University of
Tokyo, Tokyo, Japan. His research interests include nonlinear analysis and design of
concrete structures, constitutive modeling, and damage assessment.
Fig. 1Flat-shell element subjected to plane-membrane
and bending action: (a) plane-membrane actions and defor-
mations; and (b) bending actions and deformations.
Fig. 2Physical interpretation of drilling degree of freedom.
420 ACI Structural Journal/July-August 2002
matrix for the concrete, or it can be used to define the prop-
erties as overlaying elements, adding stiffness to the con-
nected nodes. This technique allows for the addition of any
number of additional layers of steel with different orienta-
tions to be added to the finite element model.
Figure 3 illustrates the layered element and the forces acting
on the shells. In-plane strain in each layer is written as
(7)
where
i
x
,
i
y
, and
i
xy
= in-plane strains in the x-y plane in layer
i;
x
,
y
, and
xy
= in-plane strains in x-y plane in the midsurface
of element thickness;
x
,
y
, and
xy
= bending and twisting
curvatures, respectively; and z = distance from the midsurface
of element thickness to the midsurface of layer i.
In-plane stresses in every layer can be obtained from the
in-plane strains in each layer through the application of two-
dimensional constitutive models of reinforced concrete and
are generally written as
(8)
where [C
ij
] is the constitutive equation.
Integration through the element thickness was achieved
through a layered-element formulation. In the depth of its
midsurface, one integration point was used for each layer of
panel. Each layer was classified as either a plain concrete
layer or a reinforced concrete layer, where the reinforcing

x
i

x
z
x

y
i
, +
y
z
y

xy
i
, +
xy
z
xy
+ = = =

x
i

y
i

xy
i


' ;


C
11
C
12
C
13
C
21
C
22
C
23
C
31
C
32
C
33

x
i

y
i

xy
i


' ;


=
bars were smeared in the layer as shown in Fig. 3. Constitutive
equations were applied through the integration point and, as a
result, in-plane stresses of the concrete and the reinforcement
were obtained separately based on the smeared reinforcement
along the element thickness. Internal forces were calculated
by integrating the corresponding stresses from each layer
over the element thickness. Through the process of integration,
the stress of reinforcement was calculated based on the real
location of the steel reinforcement.
By integrating in-plane stresses in each layer, membrane
forces N
i
and bending moments M
i
were obtained as
(9)
where n = number of layer through the thickness; and h
i
=
thickness of layer i.
NONLINEAR MATERIAL MODEL FOR
REINFORCED CONCRETE
The nonlinear material model for reinforced concrete is com-
posed of models to characterize the behavior of the concrete, in
addition to a model for characterizing the reinforcing bars.
Models for concrete may be divided into those for cracked con-
crete and those for uncracked concrete. For cracked con-
crete, the three models depict the behavior of concrete in the
direction normal to the crack plane, in the direction of the
crack plane, and in the shear direction at the crack plane
(Kim and Shin 2001).
The elastoplastic and fracture model for the biaxial state of
stress proposed by Maekawa and Okamura (1983) is used as
the constitutive equation for uncracked concrete. For
cracked concrete, the constitutive relations for reinforced
concrete may be expressed through the average stress and
strain based on the smeared-crack concept that regards the
tributary area with multiple cracks and reinforcement as a
finite continuum element (CEB 1996).
The smeared-crack concept may be applied assuming
either the fixed- or the rotating-crack method (Gupta and
Akbar 1984). The rotating-crack method assumes that each
direction of the first and second cracks varies continuously
to coincide with the direction of the principal strain, providing a
reasonable result for the localized crack at arbitrary loading.
This approach, however, is impractical for cyclic loads
where the stress history needs to be restored. The fixed-crack
N
x

x
z d
h 2
h 2


x
i
h
i
N
y

y
z d
h 2
h 2


y
i
h
i
,
1
n

= = ,
1
n

= =
N
xy

xy
z d
h 2
h 2


xy
i
h
i
,
1
n

= =
M
x

x
z z d
h 2
h 2


x
i
z
i
h
i
1
n

= =
M
y

y
z z d
h 2
h 2


y
i
z
i
h
i
,
1
n

= =
M
xy

xy
z z d
h 2
h 2


xy
i
z
i
h
i
1
n

= =
Fig. 3Shell element: (a) forces acting on reinforced concrete
shells; and (b) layered element.
ACI Structural Journal/July-August 2002 421
approach may be divided into the orthogonal and the nonor-
thogonal method. The orthogonal fixed-crack method over-
estimates the concrete stiffness, yielding a stiffer response
when the principle stress exceeds a cracking stress prior to
crack detection, as the direction of a new crack is limited to
be perpendicular to the first crack.
The basic model adopted for crack representation is a
nonorthogonal fixed-crack method of the smeared-crack
concept, which is widely known to be a robust model for
crack representation. The initiation of a crack is assumed to
start when the tensile stress reaches the fracture envelope.
After the initiation of a crack in concrete, anisotropy becomes
significant, and the stress-strain relationship is expressed
in orthogonal anisotropy in the direction normal and parallel
to the crack. Each constitutive model for the cracked re-
inforced-concrete element is formulated in the direction
of orthogonal anisotropy.
Model for uncracked concrete
For uncracked concrete, the nonlinearity and anisotropy of
concrete are expressed independent of the loading history,
including the strain-softening effects. The equivalent stress-
strain relation is given by
S = E
o
K
o
(
t
eq

p
eq
) (10)
where S = equivalent stress; E
o
= initial stiffness of concrete;
K
o
= fracture parameter for uncracked concrete;
t
eq
= equiva-
lent total strain; and
p
eq
= equivalent plastic strain. Then the
equivalent elastic strain
e
eq
is given by

e
eq
=
t
eq

p
eq
(11)
In terms of the maximum experienced equivalent strain

m
ex
, the expressions for the equivalent plastic strain and
the fracture parameter are
(12)
(13)
Model for cracked concrete
Model for cracked concrete in the direction normal to the
crack planeBecause of the bond effect between the concrete
and the reinforcing bars, cracked concrete may resist a cer-
tain amount of tensile stress normal to the cracked plane. To
obtain a more accurate tension-stiffness model, the tensile
stresses of concrete are transformed into the components in
the direction normal to the crack. This is especially true
when the reinforcing ratios in the orthogonal directions are
significantly different, and when the reinforcing bars are
distributed only in one direction.
The bond model was applied to each direction of the re-
inforcing bar. For the tension-stiffness model for unloading
and reloading, the model proposed by Shima, Chou, and
Okamura (1987) is used (Shima and Tamai 1987). This model,
based on an empirical set of data from reversed cyclic-loading
tests, relates the stresses in concrete as the sum of the stresses
from the bond action, with the reinforcing bars and with the
contact surface at the crack plane.

p
eq

m
ex
20
7
----- - 1 0.35
m
ex
( ) exp [ ] =
K
o
0.73
m
ex
1 1.25
m
ex
( ) exp [ ] { } exp =
Model for cracked concrete in the direction of the crack
planeA modified elastoplastic fracture model (Okamura,
Maekawa, and Izumo 1987) is used to describe the behavior
of concrete in the direction of the crack plane. The model de-
scribes the degradation in compressive stiffness by modify-
ing the fracture parameter in terms of the strain perpendicular
to the crack plane.
The compressive model for cracked concrete is given by

y
= E
o
K
cr
(
y

p
) (14)
where
y
= stress parallel to the crack plane; E
o
= initial stiffness
of concrete; K
cr
= fracture parameter for the cracked concrete;

y
= strain parallel to the crack plane; and
p
= plastic strain
parallel to the crack plane.
The cyclic load causes damage of the inner concrete, and
energy is dissipated during unloading and reloading processes.
Model for concrete in shear direction at crack planeThe
shear-transfer model based on the Contact Surface Density
Function (Li and Maekawa 1988) is used to consider the effect
of shear stress transfer due to the aggregate interlock at the
crack surface. As shown in Fig. 4, the model defines the form
of crack surface in terms of three parameters where: w = crack
gap at contact surface; = shear displacement at contact sur-
face; and
s
= angle defining direction of contact surface.
The contact surface is assumed to respond elastoplastically,
and the model is applicable to any arbitrary loading history.
This model defines the shear and compressive stress at the
crack surface as
(15)
(16)
where
cr
= shear stress at crack plane;
cr
= compressive
stress at crack plane; and Z = contact density function.
The shear stiffness shows a sudden increase as the crack
closes under reversed cyclic loading. This effect is modeled
for two reasons: to ensure the continuity of shear stiffness
evaluation through the opening and closing of the crack during
the reversed loading; and to obtain the equilibrium solution by
numerical analysis. Introducing the average strain concept of
the smeared-crack model used in the monotonic loading case,
Fig. 5 shows the shear transfer model for an unloading and re-
loading curve that uses a third-degree polynomial.

cr
Z w
s
, , ( )
s
sin [ ]
s
d
2
2

cr
Z w
s
, , ( )
s
cos [ ]
s
d
2
2

=
Fig. 4Shear-stress transfer through crack surface in concrete.
422 ACI Structural Journal/July-August 2002
Model for reinforcing bars in concrete
The stress acting on the reinforcing bar embedded in
concrete is not uniform, and the value is at maximum at
locations where the bar is exposed to a crack plane. The
constitutive equations for the bare bar may be used if the
stress-strain relation remains in the elastic range. The elastic
relationship between the average stress and strain of the
reinforcing bar is not valid when the bar yields at the
crack plane, even if other sections do not yield. The average
steel stress at that location is lower than the yield strength
(Tamai et al. 1987).
The postyield constitutive law for the reinforcing bar in
concrete considers the bond characteristics. The model is a
bilinear model given by
(17)
where
s
av
= average steel stress;
o
sh
= offset stress point
for initiation of strain hardening of bar;
s
av
= average steel
strain;
sh
av
= average steel strain for offset point for initiation
of strain hardening; and E
sh
= strain hardening rates of the
bar embedded in concrete.
The average stress-strain relationship of the reinforcing
bar for unloading and reloading is determined after the stress
distribution of the bar between the cracks and the stress-

s
av

sh
o
E
sh

s
av

sh
av
( ) + =
strain relation for the bare bar are calculated for that mode of
loading. Katos model (1979) for the bare bar under reversed
cyclic loading and the assumption of stress distribution
denoted by a cosine curve were used to derive the mechanical
behavior of reinforcing bars in concrete under reversed cyclic
loading. A modified version of the model for the postyield
steel behavior for the unloading and reloading branches is
shown in Fig. 6.
NUMERICAL EXAMPLES
A general-purpose finite element analysis program imple-
menting the reinforced concrete shell element with rotational
stiffness was developed. The program is built around the
finite element analysis program shell named FEAP, developed
by Taylor (2000). Custom elements and nonlinear routines are
easily adaptable in FEAP. Using the developed reinforced
concrete shell element, the element- and structural-level
specimens are analyzed. From numerical analysis of the
experimental specimens, it is found that the layering technique
employed is sufficiently accurate for the analysis of reinforced
concrete shells.
Elements subjected to biaxial bending and
in-plane loads
A series of tests by Polak (1992) was used to verify the
behavior of shell elements subjected to biaxial bending
and in-plane loads. The specimens were the same size: 316 mm
thick and 1625 x 1625 mm long, with effective test dimensions
of 1524 x 1524 mm. All specimens contained reinforcement
placed in two layers in each of the two orthogonal directions.
Of particular interest was the influence of tension-stiffening
mechanisms; therefore, the reinforcement ratio in one direc-
tion was much higher than in the other direction. The specimen
details and loading conditions are given in Table 1.
For finite element analysis, the specimens were modeled
using a mesh of nine equal-sized elements. Ten layers per ele-
ment were used for the integration through the depth.
Figure 7 compares the analytical moment-curvature relation-
ships with the experimental results. In general, the results from
the analysis were fairly accurate in predicting the behavior of
specimens in terms of yield and ultimate moments. An ex-
ception was the case of Specimen SM4, where the prediction
of the ultimate moment was higher than the experimental
result. Testing on this specimen was stopped because large de-
formations occurred in most of the panels before reaching final
failure. Although the predicted failure mode could not be
directly compared with the observed one, the failure mode
predicted by the analytical model was similar to the large defor-
mation and flexural failure observed during the test.
Fig. 6Reinforcement model for reversed cyclic loading.
Fig. 5Shear-transfer model for concrete.
Table 1SM series test specimens (Polak 1992)
Specimen
Concrete Reinforcement Applied
loading

M1:M2:P
f
c
, MPa
, deg

x
, %
*
f
yx
, MPa
y
, %
*
f
yy
, MPa
SM1 47 0 1.25 425 0.42 430 1:0:0
SM2 62 0 1.25 425 0.42 430 0.25m:0:1
SM3 56 0 1.25 425 0.42 430 3.2:1:0
SM4 64 45 1.32 425 0.44 430 0.25m:0:1
*
Per layer.

See Fig. 7.
Note: f
c
= compressive strength of concrete; = angle of orientation of reinforcing
bars to x direction of specimen;
x
,
y
= reinforcement ratio in x and y directions,
respectively; and f
yx
, f
yy
= yield strength of steel in x and y directions, respectively.
ACI Structural Journal/July-August 2002 423
Elements subjected to in-plane reversed
cyclic shear
To test the ability of the numerical procedure to predict the
behavior of concrete panels subjected to in-plane reversed
cyclic shear, the procedure was compared to experimental
results obtained by Stevens, Uzumeri, and Collins (1991).
The specimens were 285 mm thick and 1524 x 1524 mm
long, with two layers of deformed reinforcement in each of
the two orthogonal directions. One of the specimens had an
isotropic arrangement of reinforcement (SE9), and the others
had an anisotropic arrangement (SE8 and SE10). Figure 8
illustrates the reinforcement layout and the application of
stresses. Reversed cyclic shear loads were applied along the
reinforcement directions by applying equal tension and com-
pression forces in the two orthogonal directions at 45 degrees to
the reinforcement. The details of the material properties can be
found in Table 2.
Because the force distribution is uniform across the element,
only one finite element with two layers per element was used to
predict the response of the specimen. The comparison be-
tween the analytical and the experimental results is shown
in Fig. 9(a) to (c). The analytical results show good agree-
ment with the experimental results, not only in regards to the
Fig. 7Comparison of experimental and analytical responses for Specimens: (a) SM1; (b)
SM2; (c) SM3; and (d) SM4.
Fig. 8Reinforcement layout and application of stresses of
specimens (Stevens, Uzumeri, and Collins 1991).
424 ACI Structural Journal/July-August 2002
loading conditions, but also in unloading and reloading con-
ditions. It is suggested that the model be modified to rep-
resent the progressive-damage hysteresis.
Boumas reinforced concrete shell
To verify the present analysis in a complete system made
up of shell elements, the shells tested by Bouma et al. (1961)
were used as the basis for comparison. These shells, simply
supported at the two ends and constructed of reinforced
concrete, were 1/8th-scaled models of actual full-scaled
structures. As shown in Fig. 10, these scaled models had
identical cross sections and edge beams, but different span
lengths and amounts of reinforcement. In this test series, the
load was proportionally increased until failure occurred.
The reinforcement details are shown in Fig. 11. Because
the material data for the concrete and steel show substantial
scattering in the tests, the mean values were adopted in this
comparison study. The material properties are summarized
as follows:
(1) Concrete: f
c
= 28.4 MPa, and f
t
= 4.9 MPa; and
(2) Steel: E
s
= 206,153.3 MPa, and f
y
= 294.4 MPa.
Because of symmetry, the shell was modeled by 20 shell
elements and five beam elements. The shell element was di-
vided into eight layers through the thickness. The resulting
load-displacement response for the shell is presented in Fig. 12,
together with the test results given by Bouma et al. (1961) and
the analytical results of two other studies (Arnesen, Sorensen,
and Bergan 1980; Chan 1983). The agreement between
the present analytical results and Boumas test results are
satisfactory in terms of predicted strength and ductility.
Slabs subjected to cyclic transverse loads
Irawan and Maekawa (1997) conducted a series of tests
consisting of two slabs under cyclic and reversed cyclic
transverse-point load at the center of the slabs. The results
obtained from this experiment were used to verify the accuracy
of the analytical model. Two slabs, one with the reinforcing
bars arranged isotropically and the other with the reinforcing
bars arranged anisotropically, were prepared to check the
Fig. 9Comparison of experimental and analytical responses for Specimens: (a) SE8; (b) SE9; and (c) SE10.
Fig. 10Geometry for Boumas shell (Bouma et al. 1961).
Fig. 11Reinforcement layout for Boumas shell (Bouma
et al. 1961): (a) reinforcement in shell proper; and (b)
reinforcement in edge beam.
Table 2SE series test specimens (Stevens,
Uzumeri, and Collins 1991)
Specimen
Concrete Reinforcement
f
c
, MPa
, deg

x
,
*
% f
yx
, MPa
y
,
*
% f
yy
, MPa
SE8 37.0 0 1.465 492 0.490 479
SE9 44.2 0 1.465 422 1.465 422
SE10 34.0 0 1.465 422 0.490 479
*
Per layer.
Note: f
c
= compressive strength of concrete; = angle of orientation of reinforcing
bars to x direction of specimen;
x
,
y
= reinforcement ratio in x and y directions,
respectively; and f
yx
, f
yy
= yield strength of steel in x and y directions, respectively.
ACI Structural Journal/July-August 2002 425
capability of the finite element formulation to predict the
behavior under cyclic loading. The specimens were 1800 x
1800 mm long. In the isotropic slab (IS1), the reinforcement
ratio in the x- and y-directions was equal, while in the aniso-
tropic slab (IS2), the reinforcement ratio in the x-direction
was twice that in the y-direction. All specimens arranged the
reinforcement in two layers in each of the two orthogonal
directions. The material properties of the specimens are
given in Table 3.
Because both slabs were simply supported around the
perimeter of the central part of the slabs, only the 1400 x
1400 mm part was analyzed. The support consisted of two
plates, 25 mm thick and 100 mm wide, and a 40 mm-diameter
steel rod placed between the two bearing plates to avoid the
local damage around the supports. A transverse cyclic load
was applied at the center of the slab through a 240 mm-
diameter loading plate.
To facilitate the symmetry of the slab in the y-direction for
all cases of loading, this analysis modeled the specimens using
mesh of 6 x 3 elements, as shown in Fig. 13.
Comparisons of analyses and experiments for top-face
loading of both the isotropic and anisotropic slabs under
cyclic loads are given in Fig. 14(a) and (b), respectively.
The analysis predicted well the load-deflection envelope curve
for both the isotropic and anisotropic arrangements of the
reinforcing bars.
CONCLUSIONS
A nonlinear finite element procedure for the analysis of re-
inforced concrete shells was developed based on the four-
node quadrilateral flat-shell element with drilling rotational
stiffness. To analyze the reinforced concrete shell with non-
linear behavior, this study introduce the layer method, which
assumes that several thin plane stress elements are layered in
the direction of thickness. The analysis takes into account
material nonlinearity by incorporating tensile, compressive,
and shear models of cracked concrete, in addition to a model
for the reinforcing steel. The constitutive models cover the
loading, unloading, and reloading paths. Thus, the developed
Fig. 12Load-deflection curves at midspan of edge beams
for Boumas shell.
Fig. 13Element mesh for IS specimens.
Fig. 14Comparison of experimental and analytical responses for Specimens: (a) ISI;
and (b) ISI2.
Table 3Slab series test specimens (Irawan and
Maekawa 1997)
Specimen
Concrete Reinforcement
f
c
, MPa
x
,
*
% x
1
, mm x
2
, mm
y
,
*
% y
1
, mm y
2
, mm f
y
, MPa
IS1 37.0 0.78 25 75 0.78 15 85 380
IS2 37.0 0.78 25 75 0.39 15 85 380
*
Per layer.
Note: x
1
= distance between center of first reinforcing bar layer in x-direction to
bottom surface of specimen; x
2
= distance between center of second reinforcing
bar layer in x-direction to bottom surface of specimen; y
1
= distance between center
of first reinforcing bar layer in y-direction to bottom surface of specimen; and y
2
=
distance between center of second reinforcing bar layer in y-direction to bottom surface
of specimen.
426 ACI Structural Journal/July-August 2002
element predicts with reasonable accuracy the behavior of
reinforced concrete shells subjected to different types of loading.
A comparison with test data confirms that good predic-
tions were obtained in regard to load capacities, failure
modes, and load-deformation responses of reinforced
concrete shells. Hence, nonlinear finite element analysis
would be a useful tool to investigate design details or the
load-deflection response of reinforced concrete shells.
Because the general shell element has five DOFs, it was
necessary to introduce a DOF in the direction of member
rotation to combine with the three-dimensional beam-column
element with six DOFs. Thus, the developed element in this
study is promising for broader application in the analysis of
general structures consisting of slabs, shearwalls, and beam-
columns. Future work by the authors will include a formulation
of out-of-plane shear, accumulated damage, and geometrical
nonlinearity for reinforced concrete shell elements.
ACKNOWLEDGMENTS
The study described in this paper was supported by the Korea Science and
Engineering Foundation (KOSEF) through the Korea Earthquake Engineering
Research Center (KEERC) under Grant No. 2000G0203. The authors wish to
express their gratitude for the support received.
NOTATION
A
e
= element area
[C
ij
] = constitutive equations
[D
b
] = moment-curvature relation
E
o
= initial stiffness of concrete
E
sh
= strain-hardening rates of bar embedded in concrete
[ f ] = general external forces
K
cr
= fracture parameter for cracked concrete
[K
m
] = element membrane stiffness matrix with drilling degree of
freedom
K
o
= fracture parameter for uncracked concrete
M
i
= bending moments
N
i
= membrane forces
[q] = nodal parameter vector
S = equivalent stress
u = trial displacements
U
b
e
= element strain energy due to bending
w = crack gap at contact surface
Z = contact density function
z = distance from midsurface of element thickness to mid-
surface of layer i
= shear displacement at contact surface

s
av
= average steel strain

sh
av
= average steel strain for offset point for initiation of strain
hardening

e
eq
= equivalent elastic strain

p
eq
= equivalent plastic strain

eq
t
= equivalent total strain

m
ex
= maximum experienced equivalent strain

i
x
,
i
y
,
i
xy
= in-plane strains in xy plane in layer i

x
,
y
,
xy
= in-plane strains in xy plane in the midsurface of element
thickness

p
= plastic strain parallel to crack plane

y
= strain parallel to crack plane

s
= angle defining contact surface
[] = curvature
= penalty

cr
= compressive stress at crack plane

s
av
= average steel stress

sh
o
= offset stress point for initiation of strain hardening of bar

y
= stress parallel to crack plane

cr
= shear stress at crack plane
= drilling degree of freedom
= trial rotations

x
,
y
,
xy
= bending and twisting curvature
REFERENCES
Arnesen, A.; Soresen, S. I.; and Bergan, P. G., 1980, Nonlinear Analysis of
Reinforced Concrete, Computers and Structures, V. 12, pp. 571-579.
ASCE, 1993, Finite Element Analysis of Reinforced Concrete
Structures, Proceedings of the International Workshop, 717 pp.
Bouma, A. L.; Van Riel, A. C.; Van Koten, H.; and Beranek, W. J., 1961,
Investigations on Models of Eleven Cylindrical Shells Made of Reinforced
and Prestressed Concrete, Symposium on Shell Research, Delft.
Chan, E. C., 1983, Nonlinear Geometric, Material and Time-Dependent
Analysis of Reinforced Concrete Shells with Edge Beams, PhD dissertation,
Division of Structural Engineering and Structural Mechanics, University of
California, Berkeley, Calif., UC-SESM Report No. 82-8.
Comit Euro-International du Bton, 1996, RC Elements Under Cyclic
Loading, State of the Art Report, CEB, 190 pp.
Gupta, A. K., and Akbar, H., 1984, Cracking in Reinforced Concrete
Analysis, Journal of Structural Engineering, ASCE, V. 110, No. 8,
pp. 1735-1746.
Hinton, E., and Owen, D. R. J., 1984, Finite Element Software for Plates
and Shells, Pineridge Press Ltd., Swansea, UK, 403 pp.
Hughes, T. J. R., 1989, On Drilling Degrees of Freedom, Computer
Methods in Applied Mechanics and Engineering, V. 72, pp. 105-121.
Irawan, P., and Maekawa, K., 1997, Path-Dependent Nonlinear Analysis
of Reinforced Concrete Shells, Proceedings, JSCE, No. 557, pp. 121-134.
Kato, B., 1979, Mechanical Properties of Steel under Load Cycles
Idealizing Seismic Action, CEB Bulletin D'Information, No. 131, pp. 7-27.
Kim, T. H., and Shin, H. M., 2001, Analytical Approach to Evaluate the
Inelastic Behaviors of Reinforced Concrete Structures under Seismic
Loads, Journal of the Earthquake Engineering Society of Korea, EESK, V. 5,
No. 2, pp. 113-124.
Lanheng J., 1994, Analysis and Evaluation of a Shell Finite Element
with Drilling Degree of Freedom, Masters thesis, University of Maryland,
College Park, Md.
Li, B., and Maekawa, K., 1988, Contact Density Model for Stress
Transfer across Cracks in Concrete, Concrete Engineering, JCI, V. 26,
No. 1, pp. 123-137.
Maekawa, K., and Okamura, H., 1983, The Deformational Behavior
and Constitutive Equation of Concrete using Elasto-Plastic and Fracture
Model, Journal of the Faculty of Engineering, University of Tokyo, V. 37,
No. 2, pp. 253-328.
Okamura, H.; Maekawa, K.; and Izumo, J., 1987, RC Plate Element
Subjected to Cyclic Loading, IABSE Colloquium Delft, V. 54, pp. 575-590.
Polak, M. A., 1992, Nonlinear Analysis of Reinforced Concrete
Shells, PhD thesis, University of Toronto, Toronto, Canada.
Polak, M. A., and Vecchio, F. J., 1993, Nonlinear Analysis of Reinforced
Concrete Shells, Journal of Structural Engineering, ASCE, V. 119, No. 12,
pp. 3439-3462.
Shima, H.; Chou, L.; and Okamura, H., 1987, Micro and Macro Models
for Bond Behavior in Reinforced Concrete, Journal of the Faculty of
Engineering, University of Tokyo (B), V. 39, No. 2, pp. 133-194.
Shima, H., and Tamai, S., 1987, Tension Stiffness Model under
Reversed Loading Including Post Yield Range, IABSE Colloquium Delft,
pp. 547-556.
Stevens, N. J.; Uzumeri, S. M.; and Collins, M. P., 1991, Reinforced
Concrete Subjected to Reversed Cyclic ShearExperiments and Constitu-
tive Model, ACI Structural Journal, V. 88, No. 2, Mar.-Apr., pp. 135-146.
Tamai, S.; Shima, H.; Izumo, J.; and Okamura, H., 1987, Average
Stress-Strain Relationship of Steel in Uniaxial Tension Member in Post-
Yield Range, Proceedings, JSCE, No. 378, pp. 239-247.
Taylor, R. L., 2000, FEAPA Finite Element Analysis Program, Version
7.2 Users Manual, V. 1 and 2.

Das könnte Ihnen auch gefallen