You are on page 1of 14

Multi-directional modeling of crack pattern in 2D R/C members

R. Cerioni
*
, I. Iori, E. Michelini, P. Bernardi
Department of Civil and Environmental Engineering and Architecture, University of Parma, Parma 43100, Italy
Received 21 December 2006; received in revised form 3 April 2007
Available online 29 April 2007
Abstract
A macro-scale approach to R/C modeling is proposed in this paper by formulating a comprehensive model, that
describes R/C behavior in the uncracked stage (solid concrete) and in the cracked stage, the latter with either unidirec-
tional cracking (primary cracks) or with bidirectional cracking (primary and secondary cracks) or even multi-directional
cracking. The secant stiness matrix is formulated by means of a direct procedure, based on the assumption that the solid
concrete and the reinforcement work in parallel, while the solid concrete between the cracks and the cracks themselves
work in series. The resistant mechanisms active at the crack interface are introduced by means of their highly nonlinear
laws, that are taken from the literature and are based on well-documented tests. The reliability and accuracy of the
proposed model are checked against a few well-documented tests on 2D R/C members failing past the formation of
secondary cracks.
2007 Elsevier Ltd. All rights reserved.
Keywords: Reinforced concrete; Primary crack; Secondary crack; Concretesteel interaction; Crack surface interaction; Nonlinear
analysis
1. Introduction
In reinforced concrete (R/C) members subjected to plane stresses, mainly loaded in tension and shear, the
prediction of crack pattern and its evolution as loading increases is a very complex problem [16]. When the
rst (primary) cracks form, crack pattern shows few cracks with orientation depending on stress eld and on
spacing and arrangement of reinforcing bars. Normal and shear stresses are transferred across cracks through
complex phenomena, as aggregate interlock and connement actions, aggregate bridging eect, dowel action,
tension stiening and kinking eects of steel bars, etc. Material discontinuities due to cracks cause a marked
change of stress and strain elds in concrete and in reinforcing steel respect to that observed in the correspond-
ing uncracked phase. Therefore, as loading increases, new (secondary) cracks, oriented along dierent direc-
tions with respect to primary cracks, with a smaller spacing, can form. The re-orientation of cracks is strictly
connected to the anisotropic behaviour of R/C elements, above all when a markedly dierent steel ratio is
adopted for reinforcing steel bar layer.
0013-7944/$ - see front matter 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfracmech.2007.04.012
*
Corresponding author. Tel.: +39 0521 905928; fax: +39 0521 905924.
E-mail address: roberto.cerioni@unipr.it (R. Cerioni).
Available online at www.sciencedirect.com
Engineering Fracture Mechanics 75 (2008) 615628
www.elsevier.com/locate/engfracmech
In order to formulate an eective model which is able to describe the nonlinear response up to failure of
R/C members, the actual behavior of concrete and steel as well as the interaction phenomena that are
generated at the crack interface between steel and concrete and between the crack surfaces must be taken into
account through detailed and reliable laws formulated as a function of the crack kinematics.
In technical the literature several post-cracking R/C constitutive models have been proposed, which can be
divided into two dierent basic formulations: rotating and xed crack models. Models based on smeared and
rotating cracks [4,7,8] allow for a gradual re-orientation of cracks along principal stress or principal strain
directions and they describe the kinematics of crack through one parameter represented by the crack opening.
Models based on smeared-xed crack approach assume crack direction to remain xed in the direction of rst
cracking. These models, in order to satisfy the equilibrium conditions, have to add in their formulation the
shear stresses that develop on crack surfaces and the shear slip [2,9]. Other models [1015], more complex
but with a more realistic approach, have been recently formulated on the assumption that cracks were discrete
and xed and the global behavior of the R/C element was due both to the contribution of uncracked R/C
between two adjacent cracks and of local stress and strain at crack interface, as well as of bond performance
between concrete and reinforcement.
In this paper a macroscopic model [16], named 2D-PARC (two-dimensional physical approach for rein-
forced concrete), which represents an advanced version of a previous model [14], based on realistic semi-empir-
ical constitutive laws for concrete, for reinforcing steel and for their interaction at the crack interface, and
which is able to simulate the evolution of the crack pattern for plane stress R/C members, is proposed. By
assuming the behavior of uncracked concrete and reinforcement between contiguous cracks as that of two
structural elements working in parallel, while the behavior of uncracked R/C between cracks and the crack
as that of two structural elements working in series, secant stiness matrix is obtained in direct mode. Through
progressive, nonlinear analysis up to failure, the model takes into account the parameters inuencing primary
cracking, that is the stress eld and the orientation and spacing of reinforcing steel bars, and the parameters
that govern subsequent secondary cracking, that is, in addition to previous ones, bond between bars and con-
crete, dowel action, aggregate bridging, aggregate interlock, degradation of concrete between cracks.
In order to verify the reliability and capability of the proposed approach, a comparison with observations
of a well-documented experimental tests [4,5,12] is carried out.
2. Modeling of reinforced concrete
2.1. Basic hypotheses
The proposed model describes the 2D mechanical behavior of a reinforced concrete membrane element sub-
jected to general plane stresses (Fig. 1a). The theoretical formulation refers to a membrane element, with unit
Fig. 1. (a) Reinforced concrete membrane element: geometry and notation; (b) local co-ordinate system of the general ith steel bar layer
and (c) principal stress directions in the concrete.
616 R. Cerioni et al. / Engineering Fracture Mechanics 75 (2008) 615628
sides and thickness t, reinforced by ordinary steel bars arranged in n layers, each of them denoted by its axis x
i
forming an angle h
i
with respect to the global x-axis, by the cross-section area A
si
, the diameter /
i
, the spacing
s
i
, and smeared through the geometric steel ratio q
i
= A
si
/(s
i
t) (Fig. 1a and b).
By respecting the R/C physical conditions, the proposed model is able to simulate three dierent phases:
uncracked, unidirectional cracking and bidirectional or multi-directional cracking.
2.2. Uncracked stage
With reference to the global xy co-ordinate system (Fig. 1a), in the uncracked phase perfect bond is
assumed between steel and concrete; hence, concrete strain {e
c
} and steel strain {e
s
} are equal to the total
strain vector {e}:
fe
c
g fe
s
g feg: 1
Equilibrium conditions impose that the total stress {r} is the sum of the stresses in the concrete {r
c
} and
those in the steel {r
s
}. By equilibrium and compatibility conditions, as well as constitutive laws, it results:
frg fr
c
g fr
s
g D
c
fe
c
g D
s
fe
s
g D
c
D
s
feg Dfeg; 2
where [D] = [D
c
] + [D
s
] is the global uncracked R/C stiness matrix, [D
c
] is the concrete stiness matrix,
D
s


n
i1
D
si
; D
si
being the stiness matrix of the ith steel bar layer and n the total number of reinforcing
layers.
2.2.1. Concrete stiness matrix
The adopted constitutive law for uncracked concrete is based on [1719] and is dened in 12 local co-
ordinate system, whose axes are directed along the principal maximum and minimum stress directions, respec-
tively (Fig. 1c). Uncracked concrete is modeled as an orthotropic, nonlinear elastic material, being 12 the
orthotropic axes, with stiness matrix having the following expression:
D
1;2
c

1
1 m
2
E
c1
m

E
c1
E
c2
_
0
m

E
c1
E
c2
_
E
c2
0
0 0 1 m
2
G
c12
_

_
_

_; 3
E
c1
and E
c2
being the concrete secant elastic moduli in the two orthotropic directions, and G
c12
the concrete
shear modulus, assumed equal to:
G
c12

E
c1
E
c2
2m

E
c1
E
c2
_
41 m
2

:
E
c1
and E
c2
are evaluated by describing the actual biaxial state of stress with two equivalent uniaxial states of
stress. The uniaxial curves are reported in Fig. 2a, where the peak stress r
cmax
and its strain e
c
are computed in
function of the ratio a = r
1
/r
2
according to a biaxial strength envelope (Fig. 2b), [20].
In the global xy co-ordinate system the concrete stiness matrix results:
D
c
T
u

T
D
1;2
c
T
u
; 4
[T
u
] being the transformation matrix, function of the u angle between the x-axis and the 1-axis (Fig. 1c).
2.2.2. Steel stiness matrix
In the local x
i
y
i
co-ordinate system of each ith steel bar layer (Fig. 1b), the stiness matrix D
x
i
y
i

si
is eval-
uated by taking into account the axial and shear stinesses and strengths of steel bars, whose mechanical
behavior is described by an elastic-hardening constitutive law (Fig. 3). Steel stiness matrix takes the form:
D
x
i
y
i

si
q
si
E
si
0
0 G
si
_ _
; 5
R. Cerioni et al. / Engineering Fracture Mechanics 75 (2008) 615628 617
where: G
si

E
si
21m
. In xy co-ordinate system, the stiness matrix is evaluated by adding the contribution of
each ith layer of reinforcing steel bars:
D
s

n
i1
D
si

n
i1
T
si

T
D
x
i
y
i

si
T
si
; 6
[T
si
] being the transformation matrix, function of the h
i
angle.
2.3. Unidirectional cracking
Cracks are assumed to arise in concrete when the principal tensile stress (Fig. 1c) exceeds the tensile
strength. The crack is assumed to form at right angle with respect to principal tensile stress direction (that
is the current 1-axis described by the u angle, Fig. 1c). A local n
1
t
1
co-ordinate system of the crack, where
n
1
and t
1
are perpendicular and parallel to crack direction, respectively, is introduced; this co-ordinate system
is described by the w
1
angle (coincident with the current u angle) between n
1
and x-axes (Fig. 4a and b).
In the crack local n
1
t
1
co-ordinate system the mechanical quantities, associated with primary cracks, are
dened. Crack pattern is assumed as immediately fully developed with a crack spacing a
m1
which remains con-
stant during the loading process (Fig. 4a).
f
c
= -0.06
= -0.15
=1.0
=0.52
=0.0
f
ct
c
0 c
max c
0 c
c
cs
c
0 c
c
cs
c
c
2
E
E
1
E
E

c0

c0
0 c
max c
cs E

=
c1
c2

cmax
( ) ( ) ( )( ) ( ) [ ]
1 2
lim
1 - k 2 1 h 1 k 4 2 - h k 2 - h k -

+ + =
-1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 -1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
tension-tension
c1
c2

c1
=
c2
tension-compression
compression- compression

c1
=
lim

c2
( )
lim
c2,max c 2
c1,max c2,max
if 0
1 k
f
1

+
=
+
=
lim
c
c1,max
c2,max c1,max
c ct
if 1 1 0
f
h
f f ,
h 0.8 , k 3.65

=
+
=
=
= =
compression failure
tension
failure

c1

c2
compression-tension

c2

c1
a b
Fig. 2. Assumed (a) equivalent uniaxial curves for tension and compression and (b) biaxial strength envelope.
c

sy
f
sy
f
sy

sy

su

su
s
1
E
s
1
E
sp
1
s
E
1
E
s
1
E
sp
1
s
E
Fig. 3. Assumed elastic-hardening constitutive law for ordinary steel.
618 R. Cerioni et al. / Engineering Fracture Mechanics 75 (2008) 615628
After cracking, the total strain is constituted by two contributions, the rst one, {e
c
}, related to the
uncracked R/C, having a solid condition even if damaged, between two adjacent cracks and the second
one, {e
cr1
}, related to the kinematics developed in the crack. Therefore, it results:
feg fe
c
g fe
cr1
g
;
7
where {e
cr1
} is the crack strain, which is dened in the crack local co-ordinate system n
1
t
1
as
fe
n
1
t
1

cr1
g
w
1
a
m1
v
1
a
m1
_ _
; w
1
being the crack opening along the n
1
-axis and v
1
being the crack slip along t
1
-axis
(Fig. 4b). The local crack strain vector is then transferred to the global xy co-ordinate system through the
transformation matrix T
w1
: fe
cr1
g T
w1

1
fe
n
1
t
1

cr1
g.
In cracked concrete, crack opening and slip activate several resistant mechanisms which provide strength
and stiness. Some of these contributions are due to concrete, in particular to the aggregates acting upon
the crack surfaces (aggregate bridging and interlock), others are due to steel bars which cross the crack (ten-
sion stiening and dowel action). Hence, the crack stiness matrix [D
cr1
] is formed by adding the stiness
matrix due to concrete resistant contributions, [D
c,cr1
], to the stiness matrix due to reinforcement resistant
contributions, [D
s,cr1
], as follows:
D
cr1
D
c;cr1
D
s;cr1
: 8
With all the stiening contributions of the crack formulated as a function of w
1
and v
1
(Fig. 4b), the equilib-
rium in the crack, with reference to global xy co-ordinate system, yields to:
frg fr
cr1
g fr
c;cr1
g fr
s;cr1
g D
c;cr1
D
s;cr1
fe
cr1
g D
cr1
fe
cr1
g; 9
where {r
cr1
} is the stress in the crack, {r
c,cr1
} and {r
s,cr1
} are the stresses in the crack balanced by resistant
contributions due to concrete (as aggregate bridging and interlock) and due to steel bars crossing the crack
(as tension stiening and dowel action), respectively.
The equilibrium condition in the uncracked R/C between two adjacent cracks can be written as:
frg fr
c
g fr
s
g D
c
fe
c
g D
s
fe
s
g; 10
where [D
c
] and [D
s
] are the same stiness matrices dened for uncracked R/C, even if the terms of [D
c
] are
adequately softened by a damage coecient. As the average strain of steel between two contiguous cracks
{e
s
} has values little lower than the average strain {e} of element, here {e
s
} = {e} is assumed.
From Eqs. (9) and (10), the crack strain and the strain of concrete between two adjacent cracks are obtained
as follows:
fe
cr1
g D
cr1

1
frg
fe
c
g D
c

1
frg D
s
feg;
11
Fig. 4. (a) RC membrane element with unidirectional cracking: geometry and notation; (b) kinematical parameters of crack and (c)
principal stress direction in the concrete between two adjacent cracks.
R. Cerioni et al. / Engineering Fracture Mechanics 75 (2008) 615628 619
which, substituted into the compatibility Eq. (7), yield to:
feg D
c

1
frg D
s
feg D
cr1

1
frg D
c

1
D
cr1

1
frg D
c

1
D
s
feg;
and then:
frg D
c

1
D
cr1

1
I D
c

1
D
s
feg Dfeg; 12
where:
D D
c

1
D
cr1

1
I D
c

1
D
s
13
is the global R/C stiness matrix and [I] is the identity matrix.
2.3.1. Stiness matrix of the concrete between two adjacent cracks
The behavior of concrete between two adjacent cracks is similar to that assumed in uncracked stage, but a
degradation due to the damaged material near the crack and due to the irregularity of crack spacing is taken
into account. This is implemented through a damage coecient f [4,5,7], here assumed as f = (1 + 200
w/a
m1
)
1
, applied to the strength and to the stiness of concrete.
2.3.2. Stiness matrix of the steel reinforcement between two adjacent cracks
After cracking, in the R/C between two adjacent cracks steel bars retain axial stiness but their shear sti-
ness rapidly softens owing to the low bond and slip between steel bars and surrounding concrete.
2.3.3. Concrete contributions to crack stiness matrix: aggregate bridging and interlock
With reference to primary crack, the aggregate bridging is modeled by the following equation (Fig. 5a, [19]):
r
b1

r
ct max
1 w
1
=w
01

p
c
b1
w
1
a
m1
c
b1
e
1
; 14
where c
b1
is the bridging coecient, w
01
is the crack opening corresponding to r
b1
= 0.5r
ct max
and p is a coef-
cient dening the shape of the curve. The parameters w
01
and p are chosen according to the CEB-Model
Code 90 [21] bilinear law by imposing the same area under the curves, namely the fracture energy, in the range
from 0 to w
c1
, which is the crack opening corresponding to zero stress.
Aggregate interlock activates shear and normal stresses due to the slip between crack surfaces. This contri-
bution is formulated by:
r
a1
c
01
c
1
s
a1
c
a1
c
1
:
15
The c
a1
and c
01
coecients are dened as functions of the crack opening and slip [22], (Fig. 5b):
c
a1
s

2w
1
D
max

_ _
a
3
a
4
v
1
w
1

3
1 a
4
v
1
w
1
_ _
4
a
m
w
1
and c
01

a
1
a
2
w
2q
1
1
v
1
w
1
_ _
2
_ _
q
c
a1
; 16
where a
1
a
2
0:62; a
3

2:45
s

; a
4
2:44 1
4
s

_ _
; s

0:27f
c
and q = 0.25, D
max
being the maximum aggregate
size.
Fig. 5. (a) Aggregate bridging and (b) aggregate interlock actions.
620 R. Cerioni et al. / Engineering Fracture Mechanics 75 (2008) 615628
Finally, the concrete stiness matrix in the crack co-ordinate system is obtained:
D
n
1
t
1

c;cr1

c
b1
c
01
0 c
a1
_ _
: 17
2.3.4. Steel contributions to crack stiness matrix: tension stiening and dowel action
In the crack, the forces in the steel bars due to the axial stiness and to the dowel action are modeled in the
local steel co-ordinate system (Fig. 6a) and then smeared so as to obtain the static equivalent stresses (Fig. 6b).
Axial d
i1
and transversal g
i1
components, with respect to the bar axis (local x
i
y
i
co-ordinate system), of the
crack displacement vector s
1
are evaluated. Therefore, the matrix of stiening contributions for the ith steel
bar crossing the rst crack is:
D
x
i
y
i

si;cr1
q
si
E
cr1
si
g
i1
0
0 d
i1
_ _
; 18
where E
cr1
si
is the secant elastic modulus in correspondence of steel axial strain e
cr1
si
at the crack (evaluated by
tension stiening model proposed in the following), and g
i1

e
cr1
si
d
i1
l
si
is the stiening coecient, l
si

a
m1
cosh
i
w
1

is
the length of the ith bar between two adjacent cracks.
The dowel action contribution is modeled according to [23], obtaining the following:
d
i1
13:66f
0:38
c
/
0:25
i
l
si
d
i1
0:2g
0:64
i1
: 19
The stiening contribution due to steelconcrete bond between to contiguous cracks is accompanied by a non-
uniform distribution of the strain in the reinforcement. Through the equilibrium equations imposed for section
(Fig. 7a), for concrete (Fig. 7b), for steel bar (Fig. 7c), and the compatibility equation:
ds
dx
e
s
e
c
; 20
where s is the slip between steel and surrounding concrete along bar axis, e
s
and e
c
are the normal strain of steel
and concrete along bar axis, respectively, the solving equation is obtained as follows:
d
2
sx
dx
2

4
/E
s
1
E
s
E
c
q
_ _
ssx: 21
The bond-slip law of the Model Code 90 [21] (Fig. 7d) is adopted. The problem is numerically solved by
using the Finite Dierence Method with the boundary conditions s(0) = 0 and s(l
s
/2) = d/2 (Fig. 7e). Then,
the stress in the concrete are computed by integrating the concrete equilibrium equation, once the stresses
cr
s
i

cr
s

cr
s
i

cr
s

t
1
n
1
w
1

i1
cr
s

i1
cr
s

v
1
x
i y
i
cr
s

b
cr
si
N
a
w
1
v
1
S
di
l 2
a
m
si
N

i1

i1
x
i
y
i
t
1
n
1
si
N
Fig. 6. (a) Tension stiening and dowel action of bar and (b) their smeared equivalent eect on crack surfaces.
R. Cerioni et al. / Engineering Fracture Mechanics 75 (2008) 615628 621
in the concrete at crack are known. Finally, after computing concrete strains, from Eq. (20) the steel strain in
the bar is obtained by imposing the symmetry conditions in x = 0 and x = l
s
/2 and by imposing that the mean
value of steel strain computed between two adjacent cracks from the tension stiening formulation is equal to
the steel strain evaluated in the global procedure. When the tension stiening distribution satises the previous
conditions, the steel strain along the bar is obtained (Fig. 7e), where e
scri
is the steel strain of the ith reinforce-
ment layer at the crack.
Afterwards, all the matrices D
x
i
y
i

si;cr1
are transferred to the crack n
1
t
1
co-ordinate system and summed up,
obtaining the matrix D
n
1
t
1
s;cr1
. Then, the latter is added to D
n
1
t
1
c;cr1
obtaining D
n
1
t
1
cr1
that, nally, is transferred
to xy global co-ordinate system yielding [D
cr1
]:
D
cr1
T
w1

T
D
n
1
t
1
cr1
T
w1
: 22
Fig. 7. Interaction between steel bar and concrete: (a) equilibrium condition of reinforced concrete element under tension; (b) of concrete
alone; (c) of steel bar alone; (d) adopted bond relationship (Model Code 90) and (e) shear bond stress and non-uniform distribution of steel
bar tensile stress and strain.
622 R. Cerioni et al. / Engineering Fracture Mechanics 75 (2008) 615628
2.4. Bidirectional and multi-directional cracking
When in the concrete between contiguous cracks (Fig. 4c), considered in a solid condition even if damaged,
the maximum principal tensile stress exceeds the tensile strength of concrete, secondary cracks form (Fig. 8a
and b). The secondary crack orientation is assumed to form a right angle with respect to the current principal
tensile stress direction (current 1-axis described by the u angle, Fig. 4c). A local n
2
t
2
co-ordinate system of the
crack, where n
2
and t
2
are perpendicular and parallel to secondary crack direction, respectively, is introduced;
this co-ordinate system is described by the w
2
angle (coincident with the current u angle) between n
2
and
x-axes (Fig. 8c and d). The procedure by which stiness matrix is evaluated is similar to that already seen
for the singly cracked R/C. The strain is decomposed into three contributions, the rst one related to the con-
crete between two adjacent cracks and the others related to the kinematics of the two cracks (Fig. 8c and d):
feg fe
c
g fe
cr1
g fe
cr2
g: 23
Similarly to primary cracking stage (Eq. (13)), the global R/C stiness matrix is:
D D
c

1
D
cr1

1
D
cr2

1
_ _
1
I D
c

1
D
s

_ _
; 24
where [D
cr1
] and [D
cr2
] are the stiness matrixes containing the resistant contributions of the rst and second
crack system, respectively.
In the general case of multi-directional cracking stage of R/C, being N
c
the total number of occurred
cracks, the stiness matrix is given by:
D D
c

N
c
k1
D
crk

1
_ _
1
I D
c

1
D
s
: 25
3. Comparisons with experimental observations
The capability of the proposed model to describe the evolution of crack pattern in R/C membrane elements
subjected to a general plane stress state is highlighted through the analysis of two panels (named PV10 and
Fig. 8. Reinforced concrete membrane element with bidirectional cracking: (a) geometry and notation and (bd) kinematical parameters
of the primary and secondary cracks.
R. Cerioni et al. / Engineering Fracture Mechanics 75 (2008) 615628 623
PB21) tested in comprehensive and well-documented research projects [4,6,5,12]. In Table 1 the mechanical
characteristics of examined panels are reported.
Panel PV10, tested by Vecchio and Collins [4], was 890 mm square 70 mm thick, and it was reinforced by
bars at right angle, aligned with the panel sides, having dierent steel ratio, Table 1, and subjected to mono-
tonic pure shear, Fig. 9a.
Fig. 10 shows some images of the tested panel at various stages of loading: Fig. 10a shows the crack pattern
prior to yielding of transverse reinforcement (load stage 5) when there are only primary cracks oriented with
an angle little more than 45 with respect to y-direction. Fig. 10b shows the crack pattern corresponding to
ultimate load capacity (load stage 7) with the appearance of secondary cracks oriented with an angle of about
53 with respect to y-direction.
Fig. 11 shows some comparisons between experimental results and those predicted by the proposed model.
The trend of the applied shear stress is reported as a function of the shear strain (Fig. 11a), of the principal
strain angle (Fig. 11b), of the longitudinal strain (Fig. 11c) and nally of the transversal strain (Fig. 11d).
Table 1
Mechanical characteristics of examined panels
Panel Loading ratio s
xy
:r
x
:r
y
q
sx
f
sx
(MPa) q
sy
f
sy
(MPa) f
c
(MPa) f
ct
(MPa) e
c1
PV10 1:0.0:0.0 0.0179 276 0.0100 276 14.5 0.0027
PB21 1:3.1:0.0 0.0220 402 21.8 2.4 0.0018
70 mm
x
y
890 mm

yx
PV10
890
mm
70 mm
x
y

x
890
mm
PB21

yx

xy
xy

yx

xy
xy

x
890 mm
a b
Fig. 9. Geometrical characteristics of examined (a) PV10 [4] and (b) PB21 [5] panels.
Fig. 10. Crack pattern of panel PV10 at various stages of loading: (a) prior to yielding of transverse reinforcement (load stage 5) and (b) at
failure (load stage 7).
624 R. Cerioni et al. / Engineering Fracture Mechanics 75 (2008) 615628
Panel PB21 (Bhide and Collins, [5]) was 890 mm square 70 mm thick, containing only longitudinal rein-
forcement, subjected to combined shear and uniaxial tension with loading ratio 1:3.1, Table 1.
The panel showed a considerable load carrying capacity beyond the cracking load. The initial cracks formed
close to the direction of predicted principal stresses, at about 71 to the reinforcement (Fig. 12a). As load
increased, some cracks formed at about 50 and then others cracks formed at about 30 (Fig. 12b); the latter
cracks were characterized by a rapid widening that caused the failure of the panel. The predicted response, com-
pared with experimental observations, is shown in Fig. 13, in terms of shear stress against shear strain
Shear strain (10
-3
)
0 2.5 5.0 7.5 10.0 12.5 15
0
1.5
3.5
4.5
0.5
4.0
2.5
S
h
e
a
r

s
t
r
e
s
s

x
y

(
M
P
a
)

1.0
2.0
3.0
2D-Parc
Experimental
42.5 47.5 45 52.5 55 57.5
Direction of principal strain

(deg)
40 50 60
0
1.5
3.5
4.5
0.5
4.0
2.5
S
h
e
a
r

s
t
r
e
s
s

x
y

(
M
P
a
)

1.0
2.0
3.0
2D-Parc
Experimental
Observed cracking


Longitudinal strain
x
(10
-3
)
0 -0.2 0.2 0.4 0.6 0.8 1.0
0
1.5
3.5
4.5
0.5
4.0
2.5
S
h
e
a
r

s
t
r
e
s
s

x
y

(
M
P
a
)

1.0
2.0
3.0
2D-Parc
Experimental
1.2
Transversal strain
y
(10
-3
)
0.0 2.0 4.0 6.0 8.0 -2.0
0
1.5
3.5
4.5
0.5
4.0
2.5
S
h
e
a
r

s
t
r
e
s
s

x
y

(
M
P
a
)

1.0
2.0
3.0
2D-Parc
Experimental
a b
c d
Fig. 11. Comparisons between observed and predicted responses of panel PV10: plots between applied shear stress versus (a) shear strain;
(b) concrete principal strain angle; (c) longitudinal strain and (d) transverse strain.
Fig. 12. (a and b) Crack pattern of element PB21 subjected to combined shear and tension at dierent loading stages.
R. Cerioni et al. / Engineering Fracture Mechanics 75 (2008) 615628 625
(Fig. 13a), principal stress direction (Fig. 13b), longitudinal strain (Fig. 13c), and transversal strain (Fig. 13d).
On the whole, the proposed model ts well the test results.
4. Conclusions
A consistent constitutive model able to describe the progressive behavior up to failure of reinforced con-
crete plane stress elements under monotonic loading has been proposed. The model has been formulated in
terms of secant stiness matrix for the implementation into nite element codes and it is suitable for analyzing
the nonlinear response of reinforced concrete structures and for evaluating service and ultimate loading con-
ditions. The new characteristics of the extended model here presented, in comparison to those of the original
one [14], include: the use in the cracked phase of the same 2D constitutive law adopted for uncracked concrete,
a more detailed law for tension stiening, where a realistic nonlinear bond-slip law is adopted, a general for-
mulation of the stiness matrix for the cracked phase which allows a simple generalization for multi-direc-
tional cracking phase. As a result, through this formulation it has been possible to obtain the complete
strain and stress elds of the concrete and of the steel bars at the crack interface and between two adjacent
cracks, as well as the crack pattern, the slip and opening of the crack for any load increment.
The comparisons between experimental observations and numerical predictions have shown a good agree-
ment in terms of applied load versus shear strain, longitudinal and transversal strains, principal stresses and
strains in the concrete. In particular, the examined cases show that the proposed model is able to predict the
re-orientation of secondary cracks, thanks to the detailed description of interaction forces at the crack surface,
which cause a change of the stress and strain elds of concrete between adjacent cracks with respect to those of
primary cracks. This fact is well highlighted particularly in the analysis of PB21 specimen (which is subjected
to combined shear and uniaxial tension), because reinforcement is arranged only along one direction and shear
Shear strain (10
-3
)
0 2 4 6 8 10 12
0
0.50
1.00
1.50
0.25
1.25
0.75
S
h
e
a
r

s
t
r
e
s
s

x
y
(
M
P
a
)

10 20 30 40 50 60 70 80 90

0 0.5 1.0 1.5


Longitudinal strain
x
(10
-3
) Transversal strain
y
(10
-3
)
0
1.0 2.0 3.0 4.0 5.0 0
0
Direction of principal stress

(deg)
2D-Parc
Experimental
0
0.50
1.00
1.50
0.25
1.25
0.75
S
h
e
a
r

s
t
r
e
s
s

x
y
(
M
P
a
)
0.50
1.00
0.75
1.50
0.25
1.25
S
h
e
a
r

s
t
r
e
s
s

x
y
(
M
P
a
)
0
0.50
1.00
1.50
0.25
1.25
0.75
S
h
e
a
r

s
t
r
e
s
s

x
y
(
M
P
a
)
2D-Parc
Experimental
Observed cracking

2D-Parc
Experimental
2D-Parc
Experimental
a b
c d
Fig. 13. Comparisons between observed and predicted response of panel PB21: plots between applied shear stress versus (a) shear strain;
(b) concrete principal stress angle; (c) longitudinal strain and (d) transversal strain.
626 R. Cerioni et al. / Engineering Fracture Mechanics 75 (2008) 615628
stresses in the crack activate strong aggregate interlock and dowel action, producing a high slip of crack
surfaces.
The general approach here presented allows for some extensions which are currently under study. These
might include the case of non-monotonic and cyclic loadings (by introducing cyclic laws for describing the
constitutive relationships of materials and for the interactions between them), and the case of 3D modeling
of reinforced concrete.
Appendix A. Notation
A
si
cross-sectional area of the ith steel reinforcement layer
[D], [D
c
], [D
s
] reinforced concrete, concrete and steel stiness matrices, respectively, evaluated in xy co-ordi-
nate system
[D
crk
],[D
c,crk
],[D
s,crk
] stiness matrices of all the resistant contributions, of the resistant contributions due to
concrete, of the resistant contributions due to the steel of the kth crack, respectively
[D
si
] stiness matrix of the ith steel layer evaluated in xy co-ordinate system
D
x
i
y
i

s
i
cr
k
matrix of stiening contribution for ith steel bar crossing the kth crack in the local co-ordinate system
of the bar
E
c1
; E
c2
; G
c12
concrete secant longitudinal and shear elastic moduli in the two directions of orthotropy
E
si
; G
si
steel secant longitudinal and shear elastic moduli
E
crk
si
secant elastic modulus of the ith steel bar at the kth crack
f
c
uniaxial compressive strength
f
ct
uniaxial tensile strength
g
ik
tension stiening coecient
[I] identity matrix
l
si
length of ith bar between two adjacent cracks
n total number of reinforcement layers
s
i
spacing of bars of the ith layer
t thickness of membrane element
[T
si
], [T
u
], [T
wk
] transformation matrices function of the h
i
angle between local x
i
y
i
co-ordinate system of
steel bar ith layer and concrete principal stress co-ordinate system (Fig. 14b), function of the u angle
between concrete principal stress system and xy co-ordinate system (Fig. 14a), and function of the w
k
angle between crack local n
k
t
k
and global xy co-ordinate system (Fig. 14c and d), respectively
w
k
,v
k
opening and slip of the kth crack
{e},{e
c
},{e
s
} total strain, concrete strain and steel strain, respectively, evaluated in the global xy co-ordinate
system
fe
crk
g; fe
n
k
t
k

crk
g crack strain evaluated in the global and in the local crack co-ordinate systems, respectively
e
s
i
crk
steel axial strain at the kth crack
f damage coecient
q
i
geometric ratio of the ith steel reinforcement layer
Fig. 14. Local co-ordinate systems: (a) principal stress directions for concrete; (b) direction of ith layer of steel bars and (c and d) normal
and parallel directions of the primary and secondary cracks, respectively.
R. Cerioni et al. / Engineering Fracture Mechanics 75 (2008) 615628 627
{r},{r
c
},{r
s
} applied stress state, stress state in uncracked concrete or in concrete between two adjacent
cracks, stress state in the steel embedded in uncracked concrete or in concrete between two adjacent
cracks, respectively
{r
crk
},{r
c,crk
},{r
s,crk
} stresses in the crack interface, stresses in the crack interface balanced by resistant con-
tributions due to concrete and due to steel bars crossing the crack, respectively
r
bk
,r
ak
,s
ak
normal stress due to aggregate bridging eect, normal and shear stresses due to aggregate inter-
lock
/
i
diameter of the ith steel reinforcement layer
References
[1] Walraven JC, Reinhardt HW. Theory and experiments on the mechanical behaviour of cracks in plain and reinforced concrete
subjected to shear loading. HERON Concrete Mech Part A 1981:26.
[2] Iori I, Dei Poli S. Sulla torsione di elementi strutturali in cemento armato. LIndustria Italiana del Cemento 1985;2:1219 [in Italian].
[3] Collins MP, Vecchio FJ, Mehlorn G. An international competition to predict the response of reinforced concrete panels. Can J Civil
Engng 1985;12(3):62644.
[4] Vecchio F, Collins MP. The modied compression-eld theory for reinforced concrete elements subjected to shear. ACI J
1986;83(22):21931.
[5] Bhide SB, Collins MP. Inuence of axial tension on shear capacity of reinforced concrete members. ACI Struct J 1989;86-S55:57081.
[6] Pang XB, Hsu TTC. Behaviour of reinforced concrete membrane elements in shear. ACI Struct J 1995;92(6):66579.
[7] Hsu TTC. Softened truss model theory for shear and torsion. ACI Struct J 1988;65-S56:62435.
[8] Hu HT, Schnobrich C. Nonlinear nite element analysis of reinforced concrete plates and shells under monotonic loading. Comput
Struct 1991;38:63751.
[9] Pang XB, Hsu TTC. Fixed angle softened truss model for reinforced concrete. ACI Struct J 1996;93(2):197207.
[10] Kaufmann W, Marti P. Structural concrete: cracked membrane model. J Struct Engng 1998;24(12):146775.
[11] Vecchio FJ. Disturbed stress eld model for reinforced concrete: formulation. ASCE J Struct Engng 2000;126(9):10707.
[12] Vecchio FJ, Lai D, Shim W, Ng J. Disturbed stress eld model for reinforced concrete: validation. ASCE J Struct Engng
2001;127(4):3508.
[13] Belletti B, Cerioni R, Iori I. Shear tansfer in pre-cracked reinforced concrete elements. Studies and Researches, Milan University of
Technology 2001;22:126.
[14] Belletti B, Cerioni R, Iori I. A physical approach for reinforced concrete (PARC) membrane elements. ASCE J Struct Engng
2001;127(12):141226.
[15] Soltani M, An X, Maekawa K. Computational model for post cracking analysis of RC membrane element based on local-strain
characteristics. Engng Struct 2003;25:9931007.
[16] Cerioni R, Iori I, Michelini E, Bernardi P. Prediction of crack pattern in reinforced concrete members under plane stresses. In:
International Conference on Crack Paths CP. Parma 2006. Proc on CD.
[17] Darwin D, Pecknold DA. Non-linear biaxial stressstrain law for concrete. J Engng Mech Division: Proc ASCE
1977;103(EM2):22941.
[18] Elwi AA, Murray DW. A 3D hypoelastic concrete constitutive relationship. J Engng Mech Div: Proc ASCE 1979;105(EM4):62341.
[19] Belletti B, Bernardi P, Cerioni R, Iori I. On a ber reinforced concrete constitutive model for the NLFE analysis. Studies and
Researches, Milan University of Technology 2003;24:2350.
[20] Kupfer H, Hilsdorf HK, Rusch H. Behavior of concrete under biaxial stresses. Proc ACI J 1969;66(8):65666.
[21] CEB Bulletin dinformation No. 213/214. C.E.B.-F.I.P. Model Code 1990, 1993.
[22] Gambarova PG. Sulla trasmissione del taglio in elementi bidimensionali piani di c.a. fessurati. Proc Giornate AICAP, 2629 May,
1983, Bari (Italy); p. 14156 [in Italian].
[23] Walraven JC, Reinhardt HW. Theory and experiments on the mechanical behaviour of cracks in plain and reinforced concrete
subjected to shear loading. HERON 1981:26.
628 R. Cerioni et al. / Engineering Fracture Mechanics 75 (2008) 615628

You might also like