Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Cracked Membrane Model: Finite Element Implementation

Stephen J. Foster1 and Peter Marti, F.ASCE2

Abstract: In this paper, the cracked membrane model is developed into a finite element formulation for the analysis of orthogonally
reinforced structural elements in plane stress 共such as deep beams and walls兲. The cracked membrane model falls into the category of
Downloaded from ascelibrary.org by UNIV OF STELLENBOSCH-PERIOD on 09/24/13. Copyright ASCE. For personal use only; all rights reserved.

‘‘compression field models’’ combining elemental components from the modified compression field approach with those of the tension
chord model. The tension stiffening effect is modeled using a stepped, rigid-plastic steel–concrete bond relationship. The tension
stiffening and tension softening components of the concrete tension resistance mechanism are decoupled. As equilibrium is fully satisfied
at the cracks and not expressed in terms of average stresses across the element, the link to limit analysis is maintained. The model has been
incorporated into a finite element program for the analysis of reinforced concrete structures with verification against experimental data
presented.
DOI: 10.1061/共ASCE兲0733-9445共2003兲129:9共1155兲
CE Database subject headings: Finite elements; Compression; Beams; Plane strain; Concrete, reinforced; Membranes.

Introduction Background

Since finite elements 共FEs兲 were first used to model concrete Taking equilibrium of forces on the cracked plane stress element
structures in the 1960’s, it has been recognized that the concrete shown in Fig. 1, it can be shown that
state, the steel state, and the concrete–steel interaction need to be ␴ x ⫽␴ cn cos2 ␪⫹␴ ct sin2 ␪⫹␶ cnt sin共 2␪ 兲 ⫹␳ x ␴ sx (1a)
considered, including any effects that may result as a function of
steel–concrete bond, aggregate interlock, and dowel effects. In ␴ y ⫽␴ cn sin2 ␪⫹␴ ct cos2 ␪⫺␶ cnt sin共 2␪ 兲 ⫹␳ y ␴ sy (1b)
this paper, the cracked membrane model 共CMM兲 of Kaufmann ␶ xy ⫽0.5共 ␴ cn⫺␴ ct兲 sin共 2␪ 兲 ⫺␶ cnt cos共 2␪ 兲 (1c)
and Marti 共1998兲 is developed into a FE formulation for the
analysis of orthogonally reinforced structural elements in plane where ␪⫽angle between a vector normal to the cracks and the
stress 共such as deep beams and walls兲. The cracked membrane global X axis 共⫺␲/2⬍␪⭐␲/2兲; ␴ x , ␴ y , and ␶ xy ⫽in-plane normal
model falls into the category of ‘‘compression field models’’ com- and shear in the global XY coordinate system, respectively; ␴ cn
bining elemental components from the modified compression and ␴ ct⫽concrete stresses normal and parallel to the direction of
cracking, respectively; ␶ cnt⫽corresponding shear stress; ␳ x and
field model of Vecchio and Collins 共1986兲 with those of the ten-
␳ y ⫽steel reinforcement ratios in the global X and Y directions;
sion chord model of Marti et al. 共1998兲. Tension stiffening is
and ␴ sx and ␴ sy ⫽stresses in the X and Y reinforcement, respec-
modelled using a stepped, rigid-perfectly plastic concrete–steel
tively.
bond slip relationship between the cracks with equilibrium main-
The underlying assumptions to the CMM are: 共i兲 that the
tained at the crack faces. As equilibrium is fully satisfied at the cracks are stress free and able to rotate; 共ii兲 the concrete principal
cracks and not expressed in terms of average stresses across the stresses and principal strains are coincident; and 共iii兲 tension stiff-
element, the link to limit analysis is maintained and the coupling ening occurs between the cracks as per the tension chord model
between principal stress and principal strain angles removed shown in Fig. 2. The implication of assumptions 共i兲 and 共ii兲 being
共Kaufmann and Marti 1998兲. The FE model developed is verified that
against experimental data from shear panel tests by Meyboom
共1987兲 and Zhang 共1992兲; a coupling beam test of Paulay 共1971兲; ␴ cn⫽0; ␶ cnt⫽0 (2)
and a two span deep beam test of Leonhardt and Walther 共1966兲. and that the n and t axes are coincident with the major and minor
principal stress and strain axes of the concrete, respectively. As-
sumption 共iii兲 gives the average tension stiffening between cracks
in the X and Y directions and we can write this as
1
Associate Professor, School of Civil and Environmental Engineering,
The Univ. of New South Wales, UNSW Sydney 2052, Australia.
␭ x f ct ␭ y f ct
2
␴ ctsx ⫽ • f 共 ␶b兲; ␴ ctsy ⫽ • f 共␶b兲 (3)
Professor, Institute of Structural Engineering, ETH, CH-8093 Zürich, 2 2
Switzerland.
where ␭ x f ct and ␭ y f ct⫽maximum stresses in the concrete due to
Note. Associate Editor: Dat Duthinh. Discussion open until February
1, 2004. Separate discussions must be submitted for individual papers. To tension stiffening in the X and Y directions, respectively;
extend the closing date by one month, a written request must be filed with f ct⫽tension strength of the concrete; and ␶ b ⫽bond shear stress.
the ASCE Managing Editor. The manuscript for this paper was submitted The tension stiffening factors are obtained from Fig. 2 and are
for review and possible publication on January 31, 2002; approved on given by
December 6, 2002. This paper is part of the Journal of Structural Engi-
neering, Vol. 129, No. 9, September 1, 2003. ©ASCE, ISSN 0733-9445/
⌬␴ cx s rm
␭ x⫽ ⫽
2003/9-1155–1163/$18.00. f ct s rmx0 cos兩 ␪ 兩
(4)
JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2003 / 1155

J. Struct. Eng. 2003.129:1155-1163.


Downloaded from ascelibrary.org by UNIV OF STELLENBOSCH-PERIOD on 09/24/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Orthogonally reinforced membrane subject to plane stress:


共a兲 applied stresses; 共b兲 axis notation; 共c兲 and 共d兲 stresses at crack

Fig. 3. Tension chord model: 共a兲 bond stress versus slip relationship
and 共b兲 tension stiffening stress components
⌬␴ cy s rm
␭y⫽ ⫽
f ct s rmy0 sin兩 ␪ 兩
where ⌬␴ cx and ⌬␴ cy ⫽X and Y component stresses of the ten-
sion stiffening stress 共shown in Fig. 2兲; s rm⫽crack spacing mea-
sured normal to the cracks; and s rmx0 and s rmy0 ⫽crack spacings forcement the bond capacity reduces to ␶ b1 . Applying equilib-
for uniaxial tension in the X and Y directions, respectively. rium of forces 关as shown in Fig. 3共b兲兴 gives
In Eq. 共4兲 and Fig. 2, the crack spacings are determined in ⌬␴ cx cos2 ␪ c ⫹⌬␴ cy sin2 ␪ c ⫽ f ct (6)
accordance with the tension chord model of Marti et al. 共1997,
1998兲. Derived on the premise that steel–concrete bond can be where ␪ c ⫽angle between the global X axis and the major princi-
modeled as a rigid-plastic phenomenon 关as shown in Fig. 3共a兲兴, pal stress measured midway between the cracks. Substitution of
the tension chord model gives the crack spacings for a reinforced Eq. 共4兲 into Eq. 共6兲 and setting ␪ c ⫽␪ yields the Vecchio and
concrete element in uniaxial tension as Collins 共1986兲 crack spacing equation
f ct 共 1⫺␳ 兲 1
s rm0 ⫽ (5) s rm⫽ (7)
2␶ b0 ␳ cos兩 ␪ c 兩 sin兩 ␪ c 兩

where ⫽diameter of the reinforcing bars; ␶ b0 ⫽plastic bond s rmx0 s rmy0
strength; and ␳⫽reinforcement ratio; after yielding of the rein-
In the development of Eq. 共7兲, the principal stress angle midway
between cracks is approximated as equal to the angle at the crack
共that is ␪ c ⬇␪). In a further development of the work of Kaufman
and Marti 共1998兲, Foster and Marti 共2002兲 showed that the crack
spacing is fully and rationally determined using Mohr’s failure
criterion and that Eq. 共7兲 is a reasonable approximation to the
exact solution. Substitution of Eqs. 共2兲 and 共3兲 into Eq. 共1兲, writ-
ing in terms of the principal 1–2-axis system, gives the average
stresses due to any set of applied boundary tractions as
␴ x ⫽␴ c2 sin2 ␪⫹␴ c1 cos2 ␪⫹␳ x ␴ sx ⫹␭ x f ct/2 (8a)

␴ y ⫽␴ c2 cos2 ␪⫹␴ c1 sin2 ␪⫹␳ y ␴ sy ⫹␭ y f ct/2 (8b)

␶ xy ⫽ 共 ␴ c1 ⫺␴ c2 兲 sin ␪ cos ␪ (8c)


In Eq. 共8兲 the concrete principal tension term (␴ c1 ) is maintained
whereas in the CMM ␴ c1 ⫽0. The concrete tension strength is
kept here to maintain the link with concrete fracture and for the
modeling of elements with low quantities of steel reinforcement
共discussed further below兲.
For fully developed cracks spaced at s rm across a continuum,
the crack width can be obtained from elasticity. Taking a priori
that the strain of the continuum normal to the crack direction (␧ 1 )
is a function of the crack width, the principal strains and the
residual strain due to tension stiffening, the crack width is written
Fig. 2. Tension stiffening stresses: 共a兲 in material axis directions and
as
共b兲 in orthogonal tension chords of cracked membrane w cr⫽s rm共 ␧ 1 ⫹␯ 12␧ 2 ⫺␭ f ct /2E c 兲 (9)

1156 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2003

J. Struct. Eng. 2003.129:1155-1163.


Downloaded from ascelibrary.org by UNIV OF STELLENBOSCH-PERIOD on 09/24/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 4. Concrete in tension: 共a兲 bilinear stress–strain model and 共b兲


unloading law
Fig. 5. Concrete in biaxial stress: 共a兲 strength envelope and 共b兲 stress
versus strain for concrete in compression
where ␧ 1 and ␧ 2 ⫽strains in the major and minor principal direc-
tions, respectively; ␯ 12⫽Poisson’s ratio for expansion in the 1
direction resulting from stress in the 2 direction; and E c ⫽initial tension reinforcement through concrete in compression weakens
elastic modulus for concrete. It is assumed in Eq. 共9兲 that there is the concrete. The concrete compression strength is given by
no transmission of normal stresses in the concrete across the
c ⫽␤ f cp
f* (12)
cracks.
where f cp⫽compression strength of the in situ concrete. For C-C
stresses the factor ␤ in Eq. 共12兲 is a confinement factor and is
Material Laws obtained from the biaxial strength envelope shown in Fig. 5共a兲.
For T-C stresses ␤ is a disturbance factor obtained from the
modified compression field model of Vecchio and Collins 共1982,
Concrete in Tension
1986兲.
Concrete is a quasibrittle material that softens both in compres- A number of models have been developed for compression
sion and tension leading to strain localization. For tension, a bi- softening and are discussed in detail by Vecchio and Collins
linear stress–strain model is used as shown in Fig. 4. For the 共1993兲; Belarbi and Hsu 共1995兲; Kaufmann 共1998兲; and Vecchio
bilinear softening model of Petersson 共1981兲, the tension soften- 共2000a兲. In the numerical examples that follow the writers have
ing parameters are given by opted for the Vecchio and Collins 共1986兲 model, that is

2 18 E c G f 1
␣ 1 ⫽1/3; ␣ 2 ⫽ ␣ 3 ⫹␣ 1 ; ␣ 3⫽ (10) ␤⫽ ⭐1.0 (13)
9 5 l ch f 2 0.8⫹0.34␧ 1 /␧ cp
ct
where ␧ 1 ⫽major principal strain normal to the direction of the
where E c ⫽initial elastic modulus of the concrete; G f ⫽fracture
compression field and ␧ cp⫽strain corresponding to the peak in
energy; and l ch⫽characteristic length of the finite element. For the
situ stress, f cp ( f cp and ␧ cp are positive兲.
linear softening law of Hillerborg et al. 共1976兲
The stress versus strain relationship adopted for concrete in
2E c G f compression is taken as a scale of the Thornfeldt et al. 共1987兲
␣ 1 ⫽0; ␣ 2 ⫽␣ 3 ⫽ 2
(11) uniaxial compression base curve, that is
l ch f ct
n␩
For reinforced concrete elements with well distributed cracks l ch ␴ c ⫽⫺ f cp (14)
⫽s rm while for fracture type problems l ch⫽length dimension of n⫺1⫹␩ nk
the finite element normal to the direction of the cracks.
where ␩⫽ 兩 ␧ c 兩 /␧ cp ; ␧ c ⫽concrete strain; n⫽E c /(E c ⫺E cp);
E c ⫽initial modulus of elasticity of the concrete; and E cp
Concrete in Compression ⫽ f cp /␧ cp and k⫽decay factor for the postpeak response. Collins
and Porasz 共1989兲 calibrated the decay factor for conventional
The behavior of concrete in compression is a function of the
and high strength concrete leading to
biaxial state of stress. For concrete in biaxial compression (C-C)
the crushing strength in the minor principal stress direction is 兩 ␧ 兩 ⭐␧ cp . . . k⫽1 (15a)
greater than its uniaxial strength. In biaxial tension compression
兩 ␧ 兩 ⬎␧ cp . . . k⫽0.67⫹ f cp/62⭓1.0 (15b)
(T-C) it has been shown by a number of researchers 共Robinson
and Demorieux 1977; Vecchio and Collins 1982, 1986; Miyakawa where f cp is in MPa. Fig. 5共b兲 shows the stress versus strain
et al. 1987; Belarbi and Hsu 1991; Pang and Hsu 1992, among relationship given by Eqs. 共14兲 and 共15兲 for various scaling fac-
others兲 that after cracking the disturbing effect of passing strained tors, ␤.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2003 / 1157

J. Struct. Eng. 2003.129:1155-1163.


Before yielding of the reinforcing steel the mean stress in the
concrete is

␶ b ␭ f ct
␴ cm⫽ (16)
␶ b0 2
In Eq. 共16兲, ␭ is replaced by ␭ x or ␭ y depending on the direction
being considered and ␶ b ⫽bond shear stress and is given by
␧ mE s
␶ b⫽ ⭐␶ b0 (17)
2s rm
Downloaded from ascelibrary.org by UNIV OF STELLENBOSCH-PERIOD on 09/24/13. Copyright ASCE. For personal use only; all rights reserved.

where ␧ m ⫽mean strain over the length of the element and


E s ⫽elastic modulus of the reinforcing steel corresponding to ␧ m .
Eqs. 共16兲 and 共17兲 are applicable provided that the steel has not
Fig. 6. Effects of steel–concrete bond on concrete and steel stresses
reached yield at any point within the element 关Fig. 6共a兲兴. After the
between cracks: 共a兲 before yielding and 共b兲 after yielding
steel has yielded along the full length of the element then in Eqs.
共16兲 and 共17兲 ␶ b0 is replaced by ␶ b1 and E s is replaced by the
For unloading of concrete in compression the modulus E d secant modulus E sec . The mean stress in the steel ␴ sm is related to
关Fig. 5共b兲兴 is taken as that given by Filippou et al. 共1983兲. the mean stress in the concrete via equilibrium and is given by
␴ sm⫽␴ sr⫺␴ cm共 1⫺␳ 兲 /␳ (18)
Tension Stiffening
where ␴ sr⫽stress in the steel at the cracks 共that is the maximum
Tension stiffening of reinforced concrete is a result of the steel– stress兲.
concrete bond that occurs between cracks. Adopting the stepped For the case of transition of the steel between linear elastic and
rigid-perfectly plastic bond–slip relationship as given in Fig. 3共a兲, fully yielded 共between cracks兲 the relationship between the bond,
the stresses in the steel and in the concrete can be determined for concrete and steel stresses is as shown in Fig. 6共b兲. For this case
any point within the element between cracks, as shown in Fig. 6. the stresses in the steel are 共Kaufmann 1998兲

␴ sm⫽ f sy ⫺
4␶ b1 s rm 冉
共 ␴ sr⫺ f sy 兲 2 ␶ b0
␶ b1 冊
⫺1 ⫹ 共 ␴ sr⫺ f sy 兲
␶ b0 ␶ b0 s rm
␶ b1


(19a)

␶ b0 s rm

⫺ 冑 共 f sy ⫺E s ␧ m 兲

␶ b1 s rm ␶ b0 E s

␶ b1 E w
⫹冊Es
Ew
␶ b0 ␶ b1 2

2
s rm

冉 冊
␴ sr⫽ f sy ⫹ (19b)
␶ b0 E s
0.5 ⫺
␶ b1 E w

and the mean concrete stress obtained from Eq. 共18兲. In applying stress–strain relationships for uncracked concrete in T-C and for
Eq. 共19兲 to the X and Y direction reinforcement s rm is replaced by concrete in C-C. The equivalent uniaxial strain concept was de-
s rmx and s rmy , respectively. veloped by Darwin and Pecknold 共1977兲 to calculate the multi-
Finally, the secant modulus of concrete tension stiffening is axial behavior of concrete by subtracting the Poisson effect and
obtained by relating the mean stress in the concrete, between allowing the use of uniaxial stress–strain base curves. That is, the
cracks, to the mean strain. That is, E cts⫽␴ cm /␧ m .

Reinforcing Steel
A trilinear stress–strain model is adopted to model the reinforcing
steel with the properties as defined in Fig. 7. The unloading
modulus for the steel is taken as equal to the initial elastic modu-
lus, that is E d ⫽E s .

Constitutive Relationships for Orthotropic


Membranes
Before obtaining stresses in a finite element under biaxial loading
the constitutive relationships require definition. This is compli-
cated by the Poisson effect. In this study a modified form of
Fig. 7. Trilinear stress–strain model for reinforcing steel
equivalent uniaxial strain model is used to obtain the biaxial

1158 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2003

J. Struct. Eng. 2003.129:1155-1163.


Downloaded from ascelibrary.org by UNIV OF STELLENBOSCH-PERIOD on 09/24/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 8. Comparison of analytical model with experimental data of Kupfer et al. 共1969兲: 共a兲 ␣⫽0, ␤⫽1.0; 共b兲 ␣⫽0.52, ␤⫽1.24; and 共c兲 ␣⫽1.0,
␤⫽1.15

equivalent uniaxial strains can be thought of as the strain that


would exist in one direction when the stress is zero in the other.
We write this as 关 D 兴 c12⫽
1
共 1⫺␯ 12␯ 21兲

E c1
21E c2冋 ␯ 12E c1
E c2
0
0
共 1⫺␯ 12␯ 21兲 G c12

再 冎冋 册再 冎
0 0
␧1 1 ⫺␯ 12 ␧ 1u (23)
␧ 2 ⫽ ⫺␯ 21 1 ␧ 2u (20)
The shear modulus G c12 used is that derived by Attard et al.
where ␧ 1 and ␧ 2 ⫽strains in the principal directions; ␧ 1u and 共1996兲 and is given by
␧ 2u ⫽equivalent uniaxial strains in the principal directions; and
共 1⫺␯ 12␯ 21兲 G c12⫽ 4 关 E c1 共 1⫺␯ 12兲 ⫹E c2 共 1⫺␯ 21兲兴
1
(24)
␯ 12 and ␯ 21⫽Poisson’s ratios. By inverting the coefficient matrix
of Eq. 共20兲 the equivalent uniaxial strains are obtained as func- It is not generally the case that ␯ 12E c1 ⫽␯ 21E c2 and thus Eq. 共23兲
tions of the principal strains, that is is nonsymmetric. In days past it was customary to approximate

再 冎
␧ 1u
␧ 2u ⫽
1 1

1⫺␯ 12␯ 21 ␯ 21 册再 冎
␯ 12 ␧ 1
1 ␧2
(21)
the nonzero off diagonal terms as equal in the material elasticity
matrix to reinstate symmetry. For example Darwin and Pecknold
共1977兲 used 冑␯ 21E c1 •␯ 12E c2 for the nonzero off diagonal terms in
The stress–strain relationship is now expressed as Eq. 共23兲. However, with the relatively higher power and capacity

再 冎冋 册再 冎
of today’s computers this is unnecessary and the constitutive
␴ c1 E c1 0 ␧ 1u equations are maintained as nonsymmetric and solved accord-
␴ c2 ⫽ 0 E c2 ␧ 2u . . . 共 i⫽1,2兲 (22) ingly.
The material elasticity matrix is transformed into the global
where E c1 and E c2 ⫽secant moduli in the principal 共1,2兲 stress XY coordinates by
directions and are determined from the appropriate uniaxial
stress–strain law. After cracking, it is taken that there is no trans- 关 D 兴 cxy ⫽ 关 T 兴 T␧ 关 D 兴 c12关 T 兴 ␧ (25)
mission of lateral tension strains across the cracks and thus for
cracking in the major principal direction ␯ 21⫽0. When cracking where 关 T 兴 ␧ ⫽strain transformation matrix.
occurs in the minor principal direction ␯ 12⫽␯ 21⫽0. Adding the contributions of the reinforcing steel and the con-
In Fig. 8, the model is compared to the experimental data of crete tension stiffening to the material elasticity matrix we obtain
Kupfer et al. 共1969兲 for prisms loaded in biaxial compression for
关 D 兴 xy ⫽ 关 D 兴 cxy ⫹ 关 D 兴 cts⫹ 关 D 兴 s (26)
three loading ratios ␣⫽␴ 1 /␴ 2 . The material parameters used are
those obtained for the uniaxial specimen: f cp⫽32 MPa, ␧ cp where 关 D 兴 cts⫽concrete tension stiffening component; and
⫽0.0022, E c ⫽30 GPa, ␯⫽0.2. The peak load factors are taken 关 D 兴 s ⫽reinforcing steel component. Grouping the tension stiffen-
from the experiment and are ␤⫽1.24 for the panel with ␣⫽0.52 ing and steel reinforcements together, we write

冋 册
and ␤⫽1.15 for the equibiaxially loaded specimen 共␣⫽1.0兲. For
the prisms loaded in biaxial compression the analytical model 共 E ctsx ⫹␳ x E sx 兲 0 0
correlates well with the experimental data. For the uniaxially 关 D 兴 cts⫹ 关 D 兴 s ⫽ 0 共 E ctsy ⫹␳ y E sy 兲 0 (27)
loaded prism the model reflects the compression response well
0 0 0
but underestimates the volumetric expansion of the prism beyond
the peak load. This will generally be the case for uncracked pan- Finally, the element stiffness matrix is obtained in the usual man-
els loaded such that one axis is in compression and ␣⬍␯. To ner
increase the volumetric expansion an apparent Poisson’s ratio
may be introduced as per Darwin and Pecknold 共1977兲 and others.
Relating the stresses and strains in the familiar manner of 关 k 兴 ⫽t 冕
A
关 B 兴 T 关 D 兴 xy 关 B 兴 dA (28)
兵 ␴ 其 ⫽ 关 D 兴 兵 ␧ 其 , the material elasticity matrix in the material 1–2
coordinate system is where t⫽element thickness; and 关B兴⫽strain displacement matrix.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2003 / 1159

J. Struct. Eng. 2003.129:1155-1163.


Table 1. Concrete Materials Properties
Specimen f cp (MPa) f ct (MPa) E c (GPa) ␧ cp
PP1 27.0 2.7 28.0 0.0021
HB3 60.1 3.1 35.0 0.0028
311 36.7 2.4 29.4 0.0020
DWT2 27.1 2.50 27.1 0.0020
Note: for all specimens ␯⫽0.2; ␭⫽1.0; ␶ b0 /2⫽␶ b1 ⫽ f ct ; G f ⫽75 N/m.

Comparisons with Test Data


Downloaded from ascelibrary.org by UNIV OF STELLENBOSCH-PERIOD on 09/24/13. Copyright ASCE. For personal use only; all rights reserved.

The model described in the previous sections has been incorpo-


rated into a finite element program for the analysis of reinforced
concrete 共Foster and Gilbert 1990兲. Four node isoparametric con-
crete membrane elements were developed using the material and
constitutive relationships discussed above.
Kaufmann 共1998兲 extensively verified the CMM for a single
element for many panels tested by a number of researchers and
good correlation was shown against the experimental data. In this
paper the model results are compared with two panels tested in
shear, panel PP1 of Meyboom 共1987兲 and panel HB3 of Zhang
共1992兲. These panels are selected as they were orthotropically
reinforced giving a significant variation between the angles of
Fig. 9. Results of finite element model of Meyboom 共1987兲 panel
principal stress and principal strain. As the panels were uniformly
PP1: 共a兲 geometry; 共b兲 shear stress versus shear strain; 共c兲 shear stress
stressed over their entirety they could be modelled using a single
versus strain angle; and 共d兲 shear stress versus maximum crack width
four-node finite element of unit in-plane dimensions. Details of
the material properties are given in Tables 1 and 2 and the geom-
etries are shown in Figs. 9共a兲 and 10共a兲 for panels PP1 and HB3,
respectively. not fully report the crack width details for panel HB3, however,
The results of the modelling of panels PP1 and HB3 are com- from the strain measurements combined with the reported crack
pared with the experimental data for shear stress versus shear patterns the average crack widths can be determined. These are
strain 关Figs. 9共b兲 and 10共b兲兴 and the shear stress versus the prin- presented in Fig. 10共d兲 together with the values calculated from
cipal strain angle 关Figs. 9共c兲 and 10共c兲兴. In tracking of the finite the finite element model 共FEM兲 共where for the FEM, for ␭⫽1, the
element solution, an arc length procedure was adopted 共refer to average crack width is three quarters of the maximum crack
Foster 1992兲 and, thus, immediately after cracking the load is width兲.
reduced while the cracks open. On stabilization of the cracks the Figs. 9 and 10 show that the finite element results correlate
load again increases. In the experiments, however, the control well with the test data. The peak calculated shear of 4.76 MPa for
adopted does not allow the capture of the decline in load as the panel PP1 compares to 4.95 MPa in the experiment and for panel
forces are gradually transferred from the concrete to the reinforce- HB3 a peak calculated shear stress of 4.49 MPa compares to a
ment. In Fig. 9共d兲 the maximum crack width calculated by Eq. 共9兲 stress of 4.89 MPa in the experiment. For panel PP1 the element
is plotted against the shear stress for panel PP1. Zhang 共1992兲 did failed by crushing of the concrete after yielding of the Y direction

Table 2. Reinforcing Steel Properties


Specimen Location ␳ 共mm兲 E s (GPa) E w (GPa) E u (GPa) ␧ sy ␧w ␧u
PP1 X steel 0.0194 19.5 200 0.0 2.47 0.0024 0.014 0.09
PP1 Y steel 0.0065 11.3 200 0.0 1.95 0.0024 0.088 0.091
HB3 X steel 0.0171 19.5 200 0.0 3.43 0.00223 0.01 0.05
HB3 Y steel 0.0057 11.3 200 0.0 3.23 0.00225 0.01 0.05
311 X steel 1a 0.0762 23.4 200 0.0 5.0 0.00157 0.01 0.1
311 Y steel 1a 0.0088 9.3 200 0.0 5.0 0.00193 0.01 0.1
311 X steel 2a 0.0 — — — — — — —
311 Y steel 2a 0.0088 9.3 200 0.0 5.0 0.00193 0.01 0.1
DWT2 X steel 1b 0.0195 8.0 206 5.0 0.0 0.002 0.05 0.05
DWT2 Y steel 1b 0.0020 5.0 206 5.0 0.0 0.002 0.05 0.05
DWT2 X steel 2b 0.0042 6.0 206 5.0 0.0 0.002 0.05 0.05
DWT2 Y steel 2b 0.0020 5.0 206 5.0 0.0 0.002 0.05 0.05
DWT2 X steel 3b 0.0019 5.0 206 5.0 0.0 0.002 0.05 0.05
DWT2 Y steel 3b 0.0020 5.0 206 5.0 0.0 0.002 0.05 0.05
a
Refer to Fig. 11 for location.
b
Refer to Fig. 13 for location.

1160 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2003

J. Struct. Eng. 2003.129:1155-1163.


Downloaded from ascelibrary.org by UNIV OF STELLENBOSCH-PERIOD on 09/24/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 12. Tension force distribution along bottom reinforcement of


Paulay coupling beam 311 at shear force of 525 kN

details given in Fig. 11共a兲. The beam was modeled using 350
nodes and 306 elements with the material properties as given in
Tables 1 and 2. The finite element mesh for the beam is shown in
Fig. 11共b兲.
For coupling beam 311, the finite element model gave a failure
shear of 647 kN—almost exactly that on the experimental beam
which failed at a shear of 650 kN. In Fig. 12, the force in the
Fig. 10. Results of finite element model of Zhang 共1992兲 panel HB3: bottom reinforcement obtained from the FE model is compared to
共a兲 geometry; 共b兲 shear stress versus shear strain; 共c兲 shear stress that measured in the experiment at a shear of 525 kN. The plot
versus strain angle; and 共d兲 shear stress versus average crack width shows a good comparison for the FE model result with that ob-
tained in the laboratory.
Leonhardt and Walther 共1966兲 tested nine deep beams to study
reinforcing steel but before yielding of the X direction steel. For nonflexural behavior of reinforced concrete members. In this
panel HB3 failure was by crushing of the concrete after yielding study, the two span deep-beam DWT2 共shown in Fig. 13兲 is ana-
of both the X and Y direction reinforcement, as observed in the lyzed using the CMM-FE formulation. The model used consisted
laboratory. The trends in the shear stress versus shear strain, strain of 172 elements and 201 nodes and was modeled as symmetric
angle, and crack widths obtained from the FE model all compare about the central support center line. The material properties are
well with the test observations. as given in Tables 1 and 2 with the thickening over the central
Paulay 共1971兲 tested nine beams to simulate the transfer of support and the support platens modeled as linear elastic.
loading between coupled shear walls under wind or seismic load- The results of the analysis of beam DWT2 are plotted in Fig.
ing. Beam 311 was analyzed using the CMM-FE model with the 14 for the load versus midspan displacement and it is seen that the
CMM-FE formulation 共with a residual concrete tension strength
after cracking of f rt⫽0) undercalculates both the stiffness and the
failure load of the member after cracking. The model failure load
was 985 kN, 80% of the experimental failure load of 1230 kN. At
a load of P⫽750 kN in the FE model the top reinforcement
yielded and at P⫽974 kN the bottom reinforcement yielded.
After yielding of the bottom reinforcement a failure mechanism
was formed. The collapse load for beam DWT2 is calculated as
953 kN using the failure mechanism shown in Fig. 14 and with
the steel strength taken as f sy . Thus, the failure load that is cal-

Fig. 11. Paulay 共1971兲 specimen 311: 共a兲 geometry and reinforcing
arrangements and 共b兲 finite element mesh 共material details 1 and 2 Fig. 13. Details of Leonhardt and Walther 共1966兲 beam DWT2 and
given in Table 2兲 finite element mesh 共material details 1, 2, and 3 given in Table 2兲

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2003 / 1161

J. Struct. Eng. 2003.129:1155-1163.


As equilibrium is fully satisfied at the cracks and not expressed in
terms of average stresses across the element, the link to limit
analysis is also maintained.
The FEM presented was tested against two orthotropically re-
inforced shear panels and two nonflexural members: a coupling
beam and a two-span deep beam. The results from the numerical
model were seen to compare favorably against the experimental
data for members with reasonable quantities of reinforcement. In
tests on the lightly reinforced deep beam of Leonhardt and
Walther, however, the peak load obtained from the finite element
model was 20% lower than that of the tests. In this model it was
Downloaded from ascelibrary.org by UNIV OF STELLENBOSCH-PERIOD on 09/24/13. Copyright ASCE. For personal use only; all rights reserved.

observed that after yielding of the top and bottom reinforcement


failure occurred via a shear slip in the layer of elements adjacent
to the support thickening through the depth of the member. The
comparisons of the finite element model calculations with the
Fig. 14. Leonhardt and Walther deep beam DWT2: finite element
experimental data were improved when a residual tension capac-
model load versus deflection results for residual tension across cracks
ity in the concrete was maintained through the cracks.
of 0.0f ct , 0.1f ct , and 0.2f ct compared with experimental data
Lacking from the current implementation of the CMM is the
allowance for shear transfer parallel to the crack faces via aggre-
gate interlock and dowel action. This leads to an underestimation
culated by the FE model for f rt⫽0 is consistent with the collapse- of the capacity of lightly reinforced members where the aggregate
load expectations. An inspection of the displaced shape after fail- interlock mechanism is significant in the establishment of the ul-
ure revealed that after yielding of the reinforcement a shear slip timate load path. This is a subject for further research.
occurs in the elements adjacent to the central support thickening.
In the current CMM formulation no shear strength exists parallel
to the cracked surfaces and hence no strength increase is possible Acknowledgments
due to the effect of aggregate interlock or dowel action.
The poor performance of rotating crack models for the analy- The work reported in this study was undertaken while the first
sis of lightly reinforced concrete members was recognized by writer was on special studies leave at the Institute of Structural
Vecchio 共2000b兲 who showed that the lack of a shear stress law Engineering, ETH, Zürich, Switzerland. The assistance of the
along the crack faces leads to an under calculation of strength and staff of the Institute, and the provision by the Institute of the
stiffness. Vecchio suggests taking a residual tension capacity resources necessary to complete this work are gratefully acknowl-
across the cracks of 0.10–0.15 f ct for beams having low shear edged.
reinforcement ratios. Using a rotating crack FE formulation based
on the modified compression field model of Vecchio and Collins
共1986兲, Foster and Gilbert 共1990兲 achieved a good correlation for References
beam DWT2 with the residual concrete tension strength f rt
Attard, M. M., Nguyen, D. M., and Foster, S. J. 共1996兲. ‘‘Finite element
⫽0.2f ct . In Fig. 14 the results of the FEM are shown for beam
analysis of out of plane buckling of reinforced concrete walls.’’ Int. J.
DWT2 for residual tension strengths of f rt⫽0.1f ct and f rt Comput. Struct., 61共6兲, 1037–1042.
⫽0.2f ct . The results show an improved correlation when a re- Belarbi, A., and Hsu, T. T. C. 共1991兲. ‘‘Constitutive laws of reinforced
sidual tension is maintained through the cracks. The maintenance concrete in biaxial tension compression.’’ Research Rep. No. UHCEE
of a residual strength across the cracks in excess of the physical 91-2, Dept. of Civil Engineering, Univ. of Houston, Houston.
strength of the concrete, however, violates the concept of ratio- Belarbi, A., and Hsu, T. T. C. 共1995兲. ‘‘Constitutive laws of softened
nally modeling 共an important aspect in the development of the concrete in biaxial tension compression.’’ ACI Struct. J., 92共5兲, 562–
CMM兲. A general CMM would include an aggregate interlock 573.
shear law in Eq. 共1兲 such that ␶ cnt⫽0 共Kaufmann and Marti Collins, M. P., and Porasz, A. 共1989兲. ‘‘Shear strength for high strength
1998兲. concrete.’’ Bull. No. 193, Design Aspects of High Strength Concrete,
Comité Euro-International du Béton, 75– 83.
Darwin, D., and Pecknold, D. A. 共1977兲. ‘‘Nonlinear biaxial stress-strain
law for concrete.’’ J. Eng. Mech. Div., Am. Soc. Civ. Eng., 103共2兲,
Conclusions 229–241.
Filippou, F. C., Popov, E. P., and Bertero, V. V. 共1983兲. ‘‘Effects of bond
In this paper a finite element formulation was presented based on deterioration on hysteretic behaviour of reinforced concrete joints.’’
the CMM. The CMM is a compression field model combining Rep. No. UCB/EERC.83/19, Earthquake Engineering Research Cen-
elemental components from the modified compression field model ter, Univ. of California, Berkeley, Calif.
of Vecchio and Collins 共1986兲 with those of the tension chord Foster, S. J. 共1992兲. ‘‘An application of the arc length method involving
model of Marti et al. 共1998兲. The tension stiffening effect is mod- concrete cracking.’’ Int. J. Numer. Methods Eng., 33共2兲, 269–285.
eled using a stepped, rigid-perfectly plastic steel–concrete bond Foster, S. J., and Gilbert, R. I. 共1990兲. ‘‘Non-linear finite element model
for reinforced concrete deep beams and panels.’’ UNICIV Rep. No.
relationship. The CMM has two advantages over previous mod-
R-275, School of Civil Engineering, Univ. of New South Wales, Kens-
els: 共1兲 the tension stiffening and tension softening components of ington, U.K.
the concrete are decoupled allowing for separate constitutive laws Foster, S. J., and Marti, P. 共2002兲. ‘‘FE modelling of RC membranes
to be implemented for the two tension effects; and 共2兲 the model using the CMM formulation.’’ Proc., 5th World Congress on Compu-
removes the constraint that the principal stress and principal strain tational Mechanics 共WCCM V兲, Vienna, Austria, H. A. Mang, F. G.
angles are aligned as is the case for other rotating crack models. Rammerstorfer, and J. Eberhardsteiner, eds., Vienna Univ. of Technol-

1162 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2003

J. Struct. Eng. 2003.129:1155-1163.


ogy, Austria 具http://wccm.tuwien.ac.at典 Paulay, T. 共1971兲. ‘‘Coupling beams of reinforced concrete shear walls.’’
Hillerborg, A., Modeer, M., and Petersson, P. E. 共1976兲. ‘‘Analysis of J. Struct. Div. ASCE, 97共3兲, 843– 862.
crack formation and crack growth in concrete by means of fracture Petersson, P. E. 共1981兲. ‘‘Crack growth and development of fracture zone
mechanics and finite elements.’’ Cem. Concr. Res., 6, 773–782. in plain concrete and similar materials.’’ Rep. No. TVBM-1006, Lund
Kaufmann, W. 共1998兲. ‘‘Strength and deformations of structural concrete Institute of Technology, Lund, Sweden.
subjected to in-plane shear and normal forces.’’ Rep. No. 234, Institute Robinson, J. R., and Demorieux, J.-M. 共1977兲. ‘‘Essais de modèles d’âme
of Structural Engineering, ETH, Zürich, Switzerland. de poutre en double té.’’ Annales de l’Institut Technique du Bâtiment
Kaufmann, W., and Marti, P. 共1998兲. ‘‘Structural concrete: Cracked mem- et des Traveaux Publics, No. 354, October, Serie: Beton No. 172,
brane model.’’ J. Struct. Eng., 124共12兲, 1467–1475. 77–95.
Kupfer, H., Hilsdorf, H. K., and Rüsch, H. 共1969兲. ‘‘Behaviour of con- Thorenfeldt, E., Tomaszewicz, A., and Jensen, J. J. 共1987兲. ‘‘Mechanical
crete under biaxial stresses.’’ ACI J., 66共8兲, 656 – 666. properties of high strength concrete and application in design.’’ Proc.,
Leonhardt, F., and Walther, R. 共1966兲. ‘‘Wandartige Träger.’’ Bull No. International Symposium on Utilization of High Strength Concrete,
Downloaded from ascelibrary.org by UNIV OF STELLENBOSCH-PERIOD on 09/24/13. Copyright ASCE. For personal use only; all rights reserved.

178, Wilhelm Ernst and Sohn, Berlin. Stavanger, Norway, 149–159.


Marti, P., Alvarez, M., Kaufmann, W., and Sigrist, V. 共1998兲. ‘‘Tension Vecchio, F. J. 共2000a兲. ‘‘Disturbed stress field model for reinforced con-
chord model for structural concrete.’’ Struct. Eng. Int. (IABSE, Zurich, crete: formulation.’’ J. Struct. Eng., 126共9兲, 1070–1077.
Switzerland), 4共98兲, 287–298. Vecchio, F. J. 共2000b兲. ‘‘Analysis of shear critical reinforced concrete
Marti, P., Sigrist, V., and Alvarez, M. 共1997兲. ‘‘Mindestbewehrung von beams.’’ ACI Struct. J., 97共1兲, 102–110.
Betonbauten.’’ Schlussbericht, Forschungsauftrag Nr. 82/95, des Vecchio, F. J., and Collins, M. P. 共1982兲. ‘‘The response of reinforced
Bundesamts für Strassenbau. concrete to in-plane shear and normal stresses.’’ Dept. of Civil Engi-
Meyboom, J. 共1987兲. ‘‘An experimental investigation of partially pre- neering, Univ. of Toronto, Toronto.
stressed, orthogonally reinforced concrete elements subjected to mem- Vecchio, F. J., and Collins, M. P. 共1986兲. ‘‘The modified compression
brane shear.’’ Dept. of Civil Engineering, Univ. of Toronto, Toronto. field theory for reinforced concrete elements subjected to shear.’’ ACI
Miyakawa, T., Kawakami, T., and Maekawa, K. 共1987兲. ‘‘Nonlinear be- J., 83共2兲, 219–231.
havior of cracked reinforced concrete plate element under uniaxial Vecchio, F. J., and Collins, M. P. 共1993兲. ‘‘Compression response of
compression.’’ Proc., JSCE, 378, 249–258. cracked reinforced concrete.’’ J. Struct. Eng., 119共12兲, 3590–3610.
Pang, X. B., and Hsu, T. T. C. 共1992兲. ‘‘Constitutive laws of reinforced Zhang, L. X. 共1992兲. ‘‘Constitutive laws of reinforced elements with
concrete in shear.’’ Research Rep. No. UHCEE92-1, Dept. of Civil medium-high strength concrete.’’ PhD thesis, Univ. of Houston,
Engineering, Univ. of Houston, Houston. Houston.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2003 / 1163

J. Struct. Eng. 2003.129:1155-1163.

You might also like