Sie sind auf Seite 1von 8

1

ECCM 2010
IV European Conference on Computational Mechanics
Palais des Congrs, Paris, France, May 16-21, 2010
Non-Linear Behaviour Modelling of RC
Panels Subjected to In-Plane Loads
R. Scotta
1
, D.A. Talledo
2
,
L. Tesser
3
, A. Saetta
4


1
Assistant Professor, DCT, University of Padua, Italy, scotta@dic.unipd.it
2
Graduate Student, DCT, University of Padua, Italy, talledo.diego@gmail.com
3
Ph.D., DCT, University of Padua, Italy, tesser@dic.unipd.it
5
Professor, DCA, IUAV, Venice, Italy, saetta@iuav.it
Introduction
Reinforced concrete panels find widespread use in many engineering structures and accurate
prediction of their structural behaviour is important in achieving a safe structural design. The shear
strengths of these panels depend strongly on the softening of concrete struts in the principal
compression direction due to the principal tension in the orthogonal direction. Intensive investigations
of the nonlinear structural behaviour of RC panels and shear walls by finite elements method have
been reported in the last decades. Despite commendable progress made in developing new
computational methods, accurate and efficient prediction of both the overall load-deflection and the
local stress-strain cyclic responses of RC panels is still challenging because of the complicated
nonlinear behaviour of these structures, especially in the case of coupled in-plane membrane-shear
nonlinear behaviours. The main issues are the development of proper finite elements and the
enhancement of effective constitutive laws for concrete, for reinforcement and for their interactions.
Many finite element models have been developed for the nonlinear analysis of RC elements and
generally there are three types of models: the discrete model, the smeared-crack model, and the
layered model. In the discrete approach [1]-[2], the concrete and steel reinforcement are modelled
separately by two different types of finite elements. The creation of discrete models can be quite
difficult especially for complex structures. Since a large number of degrees of freedom are generated
in the discrete model, it is significantly less efficient, which is of particular concerns in the nonlinear
analysis of these structures [2]. In the smeared-crack model [3]-[4], the cracking of concrete and the
degradation of its material properties are considered by using averaged stress-strain relationships, that
are established directly from full-scale biaxial tests. The resulting models turn out to have low
computational efficiency or even to cause numerical instability. The layered approach has been widely
used for FE analysis of RC structures, and it has been demonstrated to be effective, particularly in
predicting the cracking and the ultimate behaviour of RC panels and slabs in bending and shear [5]-
[6]. In this model, the element is formulated by assembling a finite number of concrete layers and
equivalent smeared steel layers. Each layer may have different material properties corresponding to its
particular material states, and the material properties of each layer are usually assumed to be constant
throughout the thickness of the layer. In this case the material constitutive laws for general stress states
can follow analytical approaches as the theory of fracture and the theory of continuum damage
mechanic. For out-of-plane loaded slabs, cracking and crushing of concrete and yielding of
reinforcement through the thickness of the cross-section can be monitored progressively using the
layered model, thereby providing an accurate and realistic representation of the structural behaviour
[6].
The aim of the work herein is the investigation of the nonlinear modelling of reinforced concrete
2
panels by means of a concrete constitutive law based on damage mechanics applied to a layered
quadrilateral element. The concrete constitutive law, that took its bases on the works of Faria [7], Lee
[8], Berto [9], is presented in its general formulation having the possibility to represent softening
isotropic and orthotropic material behaviour. The tensile branch takes into account the concrete energy
of fracture and the tension-stiffening effects. A particular effort has been made to improve the
convergence speed through the definition of an adequate secant material stiffness matrix. For what
concerns the reinforcing steel, in sake of simplicity, a simple elastic-plastic law has been used with
both kinematic and isotropic hardening. The material models have been implemented in the finite open
source code Opensees of the University of California, Berkeley [10]. The already implemented
quadrilateral layered element has been enhanced with the possibility of taking into account more than
one nonlinear material.
The validation of the proposed model has been made by comparison with entire experimental sets
such as Bhide and Collins [11] and Mansour and Hsu [12]. These test campaigns have been chosen for
representing a wide range of coupled membrane-shear nonlinear behaviours. In particular Bhide and
Collins [11] carried out 32 tests on square panels applying combined tension, compression and shear
stressed on their edge whereas Mansour and Hsu [12] presented 12 full-size reinforced concrete panel
tests investigating the behaviour of reinforced concrete membrane elements under reversed cyclic
shear stresses. These last set outlined the effects of the variation of angle of steel bar orientation with
respect to the applied principal vertical stress and different percentages of reinforcing steel in the
panels. The results of the numerical simulations are presented critically with the aim of showing the
achievements and the model drawbacks in order to clearly delineate the future developments. The
model showed its ability to interpret the experimental evidences especially in uniaxial stress states,
biaxial compression and biaxial tension both locally and discretely, but it demonstrated the need of
improvements on biaxial tension-compression due to its simplified definition of the damage limit
surface in these stress regions.
Concrete law
Readers can refer to work by Faria et al. [7] for a complete explanation of the model formulation.
In literature the effective stress in damaged material is assumed to follow the classic elastoplastic
behaviour as written by Ju [13] and Faria et al. [7]. The equation is:
( )
p e
C C = = : :
0 0

Where
0
C denote the usual fourth-order isotropic linear-elastic stiffness tensor, ,
e
,
p
are
rank.-two tensors denoting strain tensor, its elastic and plastic tensor components, respectively.
To account for the different nonlinear behaviour of concrete under tension and compression, a
decomposition of the effective stress tensor into positive and negative components is performed as
done by Faria et al. [7], Ju [13]:
+
+
=
=


p p
i
i i i
o

where
i
o denotes the ith principal stress of and
i
p represents the associated principal direction.
The equivalent stresses defined by Faria et al. [7] are used in the present work. Definition of these
stresses can be represented by following equations:
3
( )

+ + + +
s + =
s =
r K
r
oct oct
t o t
t
3
: :
1
0
D

where

oct
o and

oct
t are the octahedral normal stress and octahedral shear stress respectively and
K is a material property defined in [7] and the variables
+
r and

r denote current damage


thresholds. These variables control the size of the expanding damage surface.
The numerical applications, carried out by the same authors, showed that the original model
proposed by Faria et al. [7] agrees quite well with the experimental results for uniaxial, biaxial tension
and biaxial compression stress states. Whereas it has been demonstrated that improvements were
necessary to better fit the experimental evidences, because of its poor definition in biaxial tension-
compression states as the dashed lines in Fig. 1 clearly evidences.
To better fit experimental data, a modification of the original model is herein proposed in which
the two previously given disequations are grouped in the unique following expression in order to
consider the interaction between tensile and compressive stresses:
( ) 0 1 , , ,
2 2
s
|
|
.
|

\
|
+
|
|
.
|

\
|
=

+
+
+ +
r r
r r g
t t
t t
Geometric representation of the damage criteria in equivalent stresses space is reported in Figure
2.
Figure 1 shows a 2D representation of these surfaces at initial state (linear-elastic domain).
1.2 0.8 0.4 0.0
1.2
0.8
0.4
0.0
Presentmodel
Original Faria etal.
s
3 /f
0

s
1 /f
0


Figure 1: Initial 2D elastic domain
Evolution of internal variables
Evolution of current damage thresholds
+
n
r and

n
r is herein proposed. The equation of ellipse can
be write in a polar coordinates system ( ) 0 , :
4
( ) ( )
( ) ( ) 0 0

2
2
2
2
2 2
2
cos sin
+
+
+
=
n n
n n
r r
r r

Denoting with 1 > =
P Q s
o a coefficient that measures how much equivalent tension state is
outside of the current damage surface, we proposed the following updating for
+
n
r and

n
r :
+ + +
+

+
=
=
s n n
s n n
r r
r r
o
o
1
1

where

s
o and
+
s
o depends on actual position of point in equivalent stresses space. Following
formulation is assumed in present work:
( )
( )
( )
( )
( )
( )

> +

=
s +

+
+
+

L P P n
L n
s
s
L P n P
n L
s
s
r
r
r
r

o
o

o
o
if 1
1
if 1
1

where L represents the point of the ellipse in which the normal vector to its curve has equal
projections on the equivalent stress axes. A geometric representation is reported in Figure 2.
r
n
+
t
+
t
r
n

0
P
Q

Q
L
n

L

Figure 2: Geometric representation of damage surface in equivalent stresses space
Application of concrete law
A pure shear test is numerically reproduced to test the damage model behaviour under tension-
compression loads. In Figure 3 three different graphs are pictured. The evolution of damage surface in
the equivalent stress space is reported in Figure 3(a), while Figures 3(b) and 3(c) show the resulting
stress-strain paths in compression and in tension respectively. The bold blue line describes the Cauchy
stress, the dashed green line shows the effective stresses (hyper-elastic stresses) and finally the red and
magenta curves represent the evolution of compression damage and tension damage respectively. It is
clear how the compression strength along the other direction is reduced due to tension in the
orthogonal direction. In fact a compression damage occurs as can be noted in Figure 3(b) and 3(c).
5

(a)

(b)

(c)
Figure 3: Pure shear load condition; (a) Evolution of damage threshold in the equivalent stresses
space; (b) stress-strain path along the principal compression direction; (c) stress-strain path along the
principal tension direction
Validation by comparison with experimental data
Several validations by comparison with experimental data have been performed to test both
original and modified models and find eventual criticisms. The more relevant results are hereafter
presented.
Tension followed by compression
An experimental test reported in [11] have been simulated with OpenSees, where the original
model and also the modified one have been implemented. In the experiment a reinforce concrete
panel, equally reinforced along on both main directions, has been initially tensioned along x-direction
and then compressed along y-direction up to failure. Panel dimensions are 890x890 mm
2
and the
reinforcement ratio is 1.30% for both x- and y- directions. The maximum uniaxial compressive
strength of the concrete material is equal to 28.1 MPa at a deformation of -0.1715%.
In Figure 4 the result for the tension stage is reported. It can be noted that the new proposed
damage model and the original coincides in this case. In Figure 5 the result for the following
compression stage is reported. The proposed model correctly shows a reduced compression strength
respect the original model, even if still far from the experimental evidence. The new model is able to
foresee a reduction of the compression failure stress accordingly with the experimental test, while this
effect cannot be reproduced with the original model.
6

Figure 4: Tension stage response (the present model coincides with the original one)

Figure 5: Compression stage response
Tension, Compression and Shear
A new experimental test reported in [11] has been simulated. It considers a concrete panel
reinforced along x-direction only, simultaneously tensioned along x-direction, compressed along y-
direction and also subjected to shear. Panel was 890x890 mm wide and the reinforcement ratio was
1.085% along x-direction. The concrete has a compressive strength of 23.5 MPa reached at the
deformation of -0.178%.
In Figure 6 the graph of shear force (expressed per unit area, i.e. in MPa) vs. shear deformation is
reported. Both the original model and the modified model give a numerical response weaker than the
experimented. The new damage model shows an higher strength decay respect to the original one. The
reason of such a result is imputable to the fact that in the model implemented OpenSees no shear
retention effects are kept into account.
Previous research [13] have demonstrated that it is possible to correctly reproduce shear strength
of non-reinforced beams considering a residual frictional effect along open cracks. The original model
shows an lower strength decay just because it doesnt consider any tensile-compressive damage
interaction for the concrete material. The proposed model needs of considering a shear retention factor
as suggested in Scotta et al. [13].
Furthermore, in the proposed model, the results depend significantly on the stress history because
of the evolution of the damage surface. This fact can be shown if the same test is reproduced with
tension and compression solicitations applied simultaneously. Figure 5 shows both the numerical
analyses indicating the sequential application of loading as CAT and the simultaneous one as CCT.
7

Figure 6: Numerical results for a simultaneously tension, compression and shear load
Cyclic shear tests on reinforced concrete panels
Mansour and Hsu [12] presented 12 full-size reinforced concrete panel tests investigating the
behaviour of reinforced concrete membrane elements under reversed cyclic shear stresses. The test set
outlined the effects of the variation of angle of steel bar orientation with respect to the applied
principal vertical stress and different percentages of reinforcing steel in the panels.
The concrete numerical model is the one presented in this work whereas for the reinforcing steel
the model of Menegotto et al. [14] modified by Filippou et al. [15] has been used. The following
Figure 7 depicts the comparison between the experimental results and the one obtained by the present
plastic-damage model for a symmetrical 0.012 reinforcement ratio in the direction of the principal
stresses.
The ratio between the principal applied stresses of the experiment has been kept constant and
equal to 1 in both the experimental and numerical tests. This has led to a tensile dominated behaviour
of the panel in which four phases can be distinguished. The first cycles are dominated by the concrete
stiffness and its tensile strength. After the concrete cracking the principal tensile stresses are borne by
the reinforcing steel that acts first with the elastic stiffness and then with the hardening modulus. At
the load reversals the behaviour is dominated by the steel that is progressively brought from tension in
compression. It can be noticed how the present plastic damage model is able to effectively estimate the
initial stiffness, the specimen strength, the stiffness decay at the increasing displacement, and the panel
ductility.

Figure 7: Comparison between the experimental test on a panel subjected to cyclic shear stresses
(Mansour et al. [12]) and the present model numerical simulation
8
Conclusion
Two damage models were used to numerically predict the non-linear behavior of bi-dimensional
reinforced concrete elements subjected to in-plane load conditions. A model based on the original
damage formulation by Faria et al. [7] has been used first. The validation work has showed how this
model is able to reproduce non-linear behavior for uniaxial stress conditions and biaxial tension and
biaxial compression conditions only. When mixed or shear stress conditions have been considered, the
original model has showed its inadequacy to reproduce the experimental results.
To overcome this problem a modification of the original formulation has been here proposed
assuming an ellipse as damage surface in equivalent stresses space and also proposing a suitable rules
of its evolution. It has been showed that the modified model is now able to take in account the
compression strength reduction due to traction along the other direction, even if the results are quite
dependent on loading path. The enhancement of the ability of the constitutive model of considering
the traction-compression interaction produce the strength reduction of specimens subjected to shear
loadings. Therefore it has been demonstrated the need of introducing a shear retention factor in the
modified model. In spite of the limitations underlined critically, the new model shows commendable
ability in predicting the cyclic response of reinforced concrete panels subjected to shear stresses.
References
[1] Nonlinear analysis of reinforced concrete slabs by a discrete finite element approach, J. Jiang, F.A. Mirza,
Comput. Struct. 65 (4), 585592, 1997.
[2] Nonlinear finite element for reinforced concrete slabs, K. Phuvoravan, E.D. Sotelino, J. Struct. Eng., ASCE
13 (4), 643649, 2005.
[3] The modified compression field theory for reinforced concrete elements subjected to shear, F.J. Vecchio,
and M.P. Collins, ACI Journal, 83 (2), 219-231, 1986.
[4] Multiscale modeling of reinforced/prestressed concrete thin-walled structures, A. Laskar, J. Zhong, Y.L.
Mo, T.T.C. Hsu, Interaction and Multiscale Mechanics, 2 (1), 69-89, 2009.
[5] Cracking and punching shear failure analysis of RC flat plates, Y.C. Loo, H. Guan, J. Struct. Eng., ASCE
123 (10), 13211330, 1997.
[6] A layered shear-flexural plate/shell element using Timoshenko beam functions for nonlinear analysis of
reinforced concrete plates, Y.X. Zhang, M.A. Bradford, R.I. Gilbert, Finite Elements Analysis and Design,
Elsevier, 43, 888-900, 2007.
[7] A strain-based plastic viscous-damage model for massive concrete structures, R. Faria, J. Oliver, M.
Cervera, International Journal of Solid and Structures, 35, 1533-1558, 1998.
[8] Plastic-Damage Model for Cyclic Loading of Concrete Structures, J. Lee, L. Fenves, J. Eng. Mech., ASCE,
124 (8), 892-900, 1998.
[9] An orthotropic damage model for masonry structures, L. Berto, A. Saetta, R. Scotta, R. Vitaliani,
International Journal for Numerical Methods in Engineering, 55 (2), 127-157, 2002.
[10] Annual workshop on Open System for Earthquake Engineering Simulation, L. Fenves, Pacific Earthquake
Engineering Research Center, UC Berkeley, 2005.
[11] Reinforced concrete elements in shear and tension, S.B. Bhide, M.P. Collins, University of Toronto,
Publication n. 87-02, 1987.
[12] Behavior of Reinforced Concrete Elements under Cyclic Shear. I: Experiments, M. Mansour and T.T.C.
Hsu, J. Structural Engineering, ASCE, 131 (1), 44-53, 2005.
[13] A scalar damage model with a shear retention factor for the analysis of reinforced concrete structures:
theory and validation, R. Scotta, R. Vitaliani, A. Saetta, E. Oate, A. Hanganu, J. of Computers &
Structures, vol. 79 (7), 737-755, 2001.
[14] Methods of analysis for cyclically loaded R/C frames, M. Menegotto, P. Pinto, Proceedings of the
Symposium of Resistance and Ultimate Deformability of Structure Acted by Well Defined Repeated Load,
IABSE, Lisbon, Portugal, 1973.
[15] Effects of bond deterioration on hysteretic behavior of reinforced concrete joints, FC. Filippou, EP. Popov,
VV. Bertero, Technical Report EERC-83/19, Earthquake Engineering Research Center, University of
California, Berkeley, 1983.

Das könnte Ihnen auch gefallen