Constitutive Model for Concrete with Embedded Sets

of Reinforcement
S. Pietruszczak1 and A. Winnicki2

Abstract: In this paper a constitutive relation is developed for concrete reinforced with two orthogonal sets of steel bars. The formu-
lation incorporates a homogeneous deformation mode, prior to cracking, as well as a localized mode associated with formation of
macrocracks. In the latter case, the representative volume comprises the reinforced fractured zone and the ‘‘intact’’ material. The stiffness
of the reinforcing network is evaluated by considering the individual steel bars to be rigidly embedded in the adjacent intact material. An
extensive numerical analysis is conducted examining the performance of the proposed framework in pure shear and axial tension for
different reinforcement intensities and orientations. The results are compared with the available experimental data.
DOI: 10.1061/共ASCE兲0733-9399共2003兲129:7共725兲
CE Database subject headings: Concrete, reinforced; Plasticity; Localization; Constitutive models.

Introduction smeared crack is determined at the onset of cracking and it re-

mains fixed during the subsequent deformation history, e.g.,
One of the main components in research on reinforced concrete is Cervenka 共1985兲. In the latter approach, it is assumed that the
that involving formulation of constitutive relations, which can stress and strain tensors remain coaxial in the course of the load-
adequately describe the deformation process. In general, the pres- ing process. Hence, the direction of the smeared crack evolves
ence of reinforcement in large concrete structures, such as hy- 共i.e., the crack rotates兲 following the change in the direction of the
draulic or nuclear ones, cannot be modeled in a discrete way, as maximum principal strain 共e.g., Gupta 1984兲. The main argument
that would be beyond the capabilities of even most modern com- here is that such an approach is more appropriate for nonpropor-
puters. Thus, the reinforced concrete should be considered as a tional strain paths, as these may involve formation of new fami-
composite medium comprising the concrete matrix and a set of lies of cracks 共Vecchio and Collins 1982兲. A comprehensive com-
families of reinforcement. The overall macroscopic behavior can parison of both approaches is given by Crisfield and Wills 共1989兲.
then be defined by employing suitable homogenization/averaging The main drawback of the smeared crack models is the lack of
procedures. distinction between the properties within the fractured zone and
A majority of the existing constitutive models incorporate those of the intact material in between the cracks. Hence, in these
some explicit static and/or kinematic constraints between the con- models no internal length parameter is introduced, so that for a
crete and steel reinforcement in order to define the stress–strain low percentage of reinforcement the results are strongly mesh
relation for the composite medium. Such models appeared already dependent 共e.g., de Borst 2001兲. The simplest remedy is to adjust
in the early 1970s 共e.g., Scordelis 1972; Suidan and the value of the softening modulus for the cracked concrete, de-
Schnobrich 1973兲, following a rapid progress in the formulation pending on the dimensions of a given finite element mesh. In this
of the finite element method. A comprehensive review of the early way a constant value of the fracture energy can be preserved and
developments is provided in the ASCE report 共1982兲. Most of the the mesh objectivity restored 共Bazant and Oh 1983; de Borst
more recent approaches fall into the category of the so-called 2001兲. However, the smeared crack approach, even when aug-
smeared crack models. The primary assumption embedded there mented by the fracture energy considerations, still cannot prop-
is that the macrocracks in concrete are smeared over the volume erly predict the local displacement pattern, i.e., the width of the
associated with a given integration domain, and moreover, that a fracture zone is sensitive to discretization and it tends to zero
full bond between the concrete and reinforcement exists. Origi- when the finite element mesh is being refined.
nally, the smeared crack models were formulated following a so- An alternative approach involves a class of so-called ‘‘embed-
called ‘‘fixed crack approach;’’ later, the ‘‘rotating crack ap- ded discontinuities models.’’ Several such formulations have been
proach’’ was introduced. In the former case, the direction of the introduced for the analysis of both plain concrete and geomateri-
als 共Simo et al. 1983; Ortiz et al. 1987; Larsson et al. 1993; de
1
Professor, McMaster Univ., Hamilton, ON, Canada. E-mail: Borst 2001兲, ensuring the mesh objectivity and rendering the
pietrusz@[Link] ‘‘scale effect.’’ A similar methodology, applied at the level of
2
Assistant Professor, Cracow Univ. of Technology, Cracow, Poland. constitutive relation, was also pursued in Pietruszczak and Xu
Note. Associate Editor: Franz-Josef Ulm. Discussion open until De- 共1995兲 and Pietruszczak and Niu 共1993兲.
cember 1, 2003. Separate discussions must be submitted for individual
Apart from research on the development of constitutive mod-
papers. To extend the closing date by one month, a written request must
be filed with the ASCE Managing Editor. The manuscript for this paper
els, as mentioned above, there is also a large volume of a related
was submitted for review and possible publication on July 15, 2002; work dealing with the limit state design of reinforced concrete
approved on October 29, 2002. This paper is part of the Journal of panels. The simplest approach here is to assume that in the limit
Engineering Mechanics, Vol. 129, No. 7, July 1, 2003. ©ASCE, ISSN state the reinforcement can carry only the longitudinal tensile
0733-9399/2003/7-725–738/$18.00. stress, while the stress state in concrete is reduced to uniaxial

JOURNAL OF ENGINEERING MECHANICS © ASCE / JULY 2003 / 725

compression parallel to the direction of the crack. For the case of
two families of reinforcement, the problem becomes statically
determinate and the solution has been discussed by Leitz 共1923兲
and Baumann 共1972兲. However, the determination of the direction
of compressive stress in concrete is still the subject of a lasting
controversy and different additional assumptions in this respect
have been introduced 共Gupta 1981; Lourenco and Figueiras
1995兲. While the approach mentioned above is too simplistic for a
general purpose, the estimates of limit load are quite useful in the
context of verification of the results of numerical simulations.
Finally, as far as the experimental testing is concerned, the
available data are mostly restricted to the reinforced concrete pan-
els loaded in shear or tension and shear 共e.g., Vecchio and Collins
1982; Pang and Hsu 1995; Bhide and Collins 1989兲. The experi-
mental results concerning other modes of loading such as, for
example, uniaxial or biaxial tension with reinforcement placed at
different angles, are rather scarce 共Peter 1966; Aoyagi and Ya-
mada 1979兲.
The primary objective of this paper is to develop a compre-
hensive model for heavily reinforced concrete, which addresses
all stages of the deformation process, i.e., homogeneous deforma- Fig. 1. Schematic representation of composite medium 共phase I兲:
tion, formation of macrocracks, and subsequent inhomogeneous local 共x*兲 and global 共x兲 coordinate systems
mode leading to yielding of the reinforcement. Given the com-
plexity of the problem, the approach pursued here incorporates
simple averaging procedures rather than a rigorous formalism of ume fraction of the composite. Refer the geometry of the problem
classical homogenization schemes. The paper is written in the to the frame of reference x*, such that x * 2 and x * 3 axes are along
following sequence. First, a general three-dimensional formula- the reinforcement, Fig. 1, and denote by ␴ ˙*
i , ␧˙ *
i (i⫽1,2,m) the
tion is provided. Next, the details pertaining to specification of volume averages of stress/strain rates in the constituents involved.
material characteristics in the fractured zone are discussed. The Both these fields are considered here to be homogeneous within
paper is concluded by presenting the results of numerical analy- themselves.
ses. The latter include the simulations of pure shear and axial Given the assumptions stated above, the average macroscopic
tension tests at different intensities and orientations of reinforce- stress/strain rates for the composite body can be defined through
ment. The numerical predictions are compared with the available integration over the volume of the representative element 共Hill
experimental data. 1963兲, i.e.,
␴˙ * ⫽␩ 1 ␴
˙*1 ⫹␩ 2 ␴*
˙ 2 ⫹ 共 1⫺␩ 1 ⫺␩ 2 兲 ␴˙m
*;

General Formulation of Problem ␧˙ * ⫽␩ 1 ␧˙ *

1 ⫹␩ 2 ␧*
˙ 2 ⫹ 共 1⫺␩ 1 ⫺␩ 2 兲 ␧˙ m
* (1)
where ␩ 1 and ␩ 2 ⫽respective volume fractions of the reinforce-
In the formulation presented here, the reinforced concrete is ment sets. Assume, at this point, that the reinforcing steel is an
treated as a composite material comprising the concrete matrix isotropic linearly elastic material, so that
and two or more families of reinforcement. The overall macro-
scopic behavior is defined in terms of the properties of constituent ␧˙ *
1 ⫽关C*
1 兴 ␴*
˙1 ; ␧˙ *
2 ⫽关C*
2 兴 ␴*
˙2 (2)
materials and the respective volume fractions. In the stage prior to where 关C兴’s are elastic compliances. Moreover, consider the con-
cracking 共phase I兲 the constitutive relation is derived by imposing crete matrix as an elastoplastic strain-hardening material for
a set of explicit kinematic and static constraints between the which
stress/strain rate averages in both constituents. After cracking of
␧˙ m
* ⫽ 关 C m* 兴 ␴˙ m* (3)
concrete 共phase II兲 it is assumed that the deformation is localized
along a distinct interface and the representative volume is com- The specific form of the constitutive relation 共3兲 adopted further
posed of the fractured zone and the ‘‘intact’’ material, both rein- in this work is that based on a plasticity framework described in
forced with steel bars. The resulting averaging scheme incorpo- the article by Pietruszczak and Xu 共1995兲.
rates the ‘‘size effect,’’ which is pertinent to the description of In order to complete the formulation assume now, after Hill
localized deformation. In what follows, the mathematical formu- 共1963兲, that the local stress rate averages can be related to the
lation for phases I and II is outlined. For the clarity of presenta- overall macroscopic measure ␴ ˙ *, i.e.,
tion, several details pertaining to this formulation are amended
␴
˙*1 ⫽ 关 B 1 兴 ␴* ;
˙ ␴
˙*2 ⫽ 关 B 2 兴 ␴* ;
˙ ␴
˙m* ⫽ 关 B m 兴 ␴˙ * (4)
into appendices.
where 关B兴’s are structural matrices 共also known as ‘‘concentration
tensors’’兲, which depend on the volume fractions and phase
Phase I: Prior to Cracking
moduli. In the elastic range, the operators 关B兴 may be estimated
Consider a representative volume of the material, which com- from the solution of an inclusion problem. Several such microme-
prises the concrete matrix 共m兲 and two orthogonal sets of rein- chanical approaches have been discussed in the literature, e.g.,
forcement 共sets 1 and 2, respectively兲. Assume that the diameter Eshelby 共1957兲 and Benveniste 共1985兲. In this work, a simplified
of the reinforcement is negligible as compared to the spacing; at procedure is adopted based on incorporating a set of explicit ki-
the same time, however, the reinforcement occupies a finite vol- nematic and static constraints

726 / JOURNAL OF ENGINEERING MECHANICS © ASCE / JULY 2003

 s i共 ␴
˙ * ,␴
˙* ˙ 2 ,␴
1 ,␴*
˙m* 兲 ⫽0; k j 共 ␧˙ * ,␧˙ * * 兲 ⫽0
˙ 2 ,␧˙ m
1 ,␧* (5)
where k and s are scalar-valued functions and i and j specify the
number of constraints. Eqs. 共1兲–共3兲, together with Eq. 共5兲, form a
set of equations which can be solved to define the structural ma-
trices 关B兴. It should be emphasized again that this approach is
justified provided the reinforcement has a vanishing diameter. In
this case, Eqs. 共5兲 furnish some explicit kinematic constraints on
the axial deformation of both phases 共i.e., reinforcement and con-
crete兲, while the interaction in the transverse direction is ne-
glected. As a result, the local stress/strain fields remain uniform
共cf. Wung and Dvorak 1985兲. The details on the specification of
the operators 关B兴 are discussed in Appendix A.
Substitution of Eqs. 共2兲–共4兲 into the second equation in Eq. 共1兲
gives the average constitutive relation for the considered compos-
ite medium in the form
␧˙ * ⫽ 关 C * 兴 ␴
˙ *; Fig. 2. Schematic representation of composite medium 共phase II兲:
local base x⬘ associated with fractured zone
关 C * 兴 ⫽ 兵 ␩ 1关 C *
1 兴关 B 1 兴 ⫹␩ 2 关 C * * 兴关 B m 兴 其
2 兴关 B 2 兴 ⫹␩ m 关 C m (6)
In the above equation, the stress/strain rates are referred to the
local system x* attached to the reinforcement. Both these mea-
sures can now be transformed to the global Cartesian system x fact, the presence of reinforcement in the neighborhood of the
共Fig. 1兲, i.e., crack triggers a further local damage in the adjacent concrete
␴
˙ *⫽ 关 T 兴 ␴
˙; ␧˙ ⫽ 关 T 兴 T ␧˙ * (7) matrix, thereby resulting in widening of the crack on the macro-
scale. Thus, h is interpreted here as the thickness of the damaged
where 关T兴⫽transformation matrix. Using the above transforma- concrete zone, which in case of reinforced concrete is of the same
tion rules, the constitutive relation 共6兲 can be expressed in an order of magnitude as the diameter of the reinforcement 共e.g.,
equivalent form CEB/FIP code 1993兲.
˙ ⫽ 关 T 兴 T 兵 ␩ 1 关 C 1* 兴关 B 1 兴 ⫹␩ 2 关 C 2* 兴关 B 2 兴
␧˙ ⫽ 关 C 兴 ␴ In order to define the macroscopic response, an averaging pro-
cedure is employed which is conceptually similar to that devel-
* 兴关 B m 兴 其 关 T 兴 ␴˙
⫹␩ m 关 C m (8) oped in Pietruszczak 共1999兲 to describe the strain localization in
saturated geomaterials. The formulation incorporates again the
The constitutive relation 共8兲 governs the response of the com- stress/strain rate decomposition based on volume averaging
posite prior to the onset of a localized deformation mode, which
is associated with formation of macrocracks in the concrete ma- ␴
˙ ⫽␮ i ␴
˙ i ⫹␮ f ␴
˙f; ␧˙ ⫽␮ i ␧˙ i ⫹␮ f ␧˙ f (10)
trix. In the compression regime, the inception of strain localiza- Here, i refers to the intact material outside the localization zone, f
tion may be considered as a bifurcation problem, i.e., the loss of denotes the material in the fractured zone, and ␮⫽corresponding
stability of the constitutive relation governing the homogeneous volume fractions. All quantities are referred to the global coordi-
deformation. The conditions under which localized modes arise nate system. It is evident from Eq. 共10兲 that the macroscopic
have been investigated by a number of researchers, e.g., Rudnicki framework employed here incorporates two primary constituents
and Rice 共1975兲, Ortiz 共1987兲, etc. In reinforced concrete struc- which, in an average sense, are again considered as homogeneous
tures, the formation of macrocraks is typically associated with the within themselves.
tensile stress regime. In this case, the onset of localization is The strain rate in the fractured zone may be conveniently de-
commonly defined in terms of a tensile fracture criterion, while fined in terms of velocity discontinuities ġ, as a symmetric part of
the localization plane is assumed to be orthogonal to the direction a dyadic product
of the major tensile stress ␴ I . Thus,
1 1
兵 关 ␴ m 兴 ⫺ 共 ␴ I 兲 m 关 I 兴 其 n⫽0 (9) 关 ␧˙ f 兴 ⫽ 共 nġT ⫹ġnT 兲 ; ⇒␧˙ f ⫽ 关 N 兴 ġ (11)
2h h
where 关I兴⫽identity matrix and the components of n⫽direction
where n⫽unit vector normal to the fractured zone, Eq. 共9兲, and

冋 册
cosines of the localization plane. In this work, the conditions at
failure, in both compression and tension regimes, are defined by a n1 0 0 n2 0 n3
path-independent criterion introduced originally in the article by 关N兴 ⫽ 0
T n2 0 n1 n3 0 (12)
Pietruszczak et al. 共1988兲.
0 0 n3 0 n2 n1
The equilibrium requires that the traction along the discontinuity
Phase II: After Formation of Macrocrack
plane remains continuous. Thus,
Consider now a representative volume of the material as shown in
ṫ⫽ 关 N 兴 T ␴
˙i (13)
Fig. 2, which comprises the ‘‘intact’’ reinforced concrete inter-
cepted by a macrocrack of a given orientation. The latter repre- Assume now the constitutive relations in the general form
sents a composite medium within itself as it consists of a zone of
␧˙ i ⫽ 关 C 兴 ␴
˙i; ġ⫽ 关 K 兴 ⫺1 ṫ (14)
fractured concrete reinforced with steel bars. Let the thickness of
the fractured zone be h. It should be emphasized that this dimen- where 关C兴 is defined by Eq. 共8兲. Substituting the second relation
sion cannot be directly identified with the visible crack width. In in Eq. 共11兲 and taking into account Eq. 共13兲, leads to

JOURNAL OF ENGINEERING MECHANICS © ASCE / JULY 2003 / 727

 1 1 Specification of Mechanical Characteristics
␧˙ f ⫽ 关 N 兴 ġ⫽ 关 N 兴关 K 兴 ⫺1 关 N 兴 T ␴
˙i (15) in Fractured Zone
h h

Thus, in view of the strain decomposition 共10兲 Stiffness Matrix for Reinforcement Network
˙ i ⫹␮ 关 N 兴关 K 兴 ⫺1 关 N 兴 T ␴
␧˙ ⫽␮ i 关 C 兴 ␴ ˙i (16) The properties of the reinforcement network can be described by
adopting either a strictly phenomenological approach or by em-
where ␮⫽␮ f /h represents the ratio of the area of the fractured ploying some idealizations on the micromechanical level. In what
zone to the volume of the sample. Thus, the parameter ␮ is, in follows, the latter approach is implemented, whereby the stiffness
fact, independent of h. Noting now that ␮ f Ⰶ␮ i while the stress of the network is evaluated by considering individual reinforcing
rates in both constituents are of the same order of magnitude, the bars to be rigidly embedded in the adjacent intact material.
stress decomposition 共10兲 simplifies to ␴˙ i 䊐␮ i ␴
˙ i 䊐␴
˙ i . Therefore, Consider the reinforcement family # 1 and refer the problem to
Eq. 共16兲 can be approximated by the local coordinate system x* introduced in the preceding sec-
tion, Fig. 1. Denote by ei* (i⫽1,2,3) the corresponding base vec-
␧˙ ⫽ 共关 C 兴 ⫹␮ 关 N 兴关 K 兴 ⫺1 关 N 兴 T 兲 ␴
˙ (17)
tors, and assume that e3* is along the axis of the reinforcement.
which provides the global macroscopic constitutive relation for With reference to this coordinate system, the constitutive relation
the considered composite medium. can be written in a general form
Let us focus now on the specification of the operator 关K兴 ap-
1 ⫽关K*
ṫ* 1 兴 ġ* (25)
pearing in Eq. 共17兲. In general, in order to establish the averaged
constitutive relation for the fractured zone, it is convenient to where ṫ1* ⫽traction on the plane perpendicular to the reinforce-
choose a local base x⬘, Fig. 2, such that x ⬘1 and x ⬘2 axes are con- ment axis. Following now the Timoshenko’s beam theory 共cf.,
fined to the localization plane and x ⬘3 is normal to it. With refer- Przemieniecki 1968兲, the elastic stiffness operator 关 K *
1 兴 can be
ence to this base, Eqs. 共14兲 and 共15兲 take the form defined as

ṫ⬘ ⫽ 关 K ⬘ 兴 ġ⬘ ;
1
␧˙ ⬘f ⫽ 关 N ⬘ 兴关 K ⬘ 兴 ⫺1 关 N ⬘ 兴 T ␴
h

It should be noted that since the vector n, Eq. 共11兲, is now aligned
˙ i⬘ (18)
1 兴⫽
关K* 0
0
冋 kt
kt
0

0
0
0 ⫽
kn
E
l册 冋 册
␥
0
0
0
␥
0
0
E
0 ⫽ 关 S*兴
1
l
(26)

with the x ⬘3 axis, there is n⬘ ⫽ 兵 0,0,1 其 T , so that the operator 关 N ⬘ 兴 , where

Eq. 共12兲, is composed of 0 and 1. Denoting by 关Q兴 the transfor-
mation matrix from x to x⬘ system, one can write ␥⫽
␾
2 共 1⫹␾ 兲共 1⫹␯ 兲
; ␾⫽ 23 共 1⫹␯ 兲 冉冊d
l
2
(27)

˙ i⬘ ⫽ 关 Q 兴 ␴
␴ ˙i; ␧˙ f ⫽ 关 Q 兴 T ␧˙ ⬘f (19) In the above equations, E and ␯⫽Young’s modulus and Poisson’s
ratio of steel; d⫽diameter of an individual member; and
Therefore, the second equation in Eq. 共18兲 becomes l⫽length. The latter is a function of the width of the fractured
zone, h, i.e.,
1
␧˙ f ⫽ 关 M 兴关 K ⬘ 兴 ⫺1 关 M 兴 T ␴
˙i; 关 M 兴 ⫽ 关 Q 兴 T关 N ⬘ 兴 (20) ⫺1
h 3兲
l⫽h 共 nT e* (28)
where n⫽unit normal to the fracture plane, Eq. 共9兲. The details on
Thus, comparing the representations 共15兲 and 共20兲, it is evident
the derivation of Eqs. 共26兲 and 共27兲 are provided in Appendix B.
that
In order to define the material characteristics in the local co-
关 N 兴关 K 兴 ⫺1 关 N 兴 T ⫽ 关 M 兴关 K ⬘ 兴 ⫺1 关 M 兴 T (21) ordinate system x⬘ associated with the fractured zone, the relation
共25兲 needs to be transformed from the bases e* i to e⬘i (i⫽1,2,3).
so that the constitutive relation 共17兲 may be expressed in an For the velocity discontinuity ġ*, the transformation rule takes the
equivalent form form
␧˙ ⫽ 共关 C 兴 ⫹␮ 关 M 兴关 K ⬘ 兴 ⫺1 关 M 兴 T 兲 ␴
˙ (22) ġ⬘ ⫽ 关 T̂ 兴 ġ* (29)

As mentioned earlier, the fractured zone is a composite me- where 关 T̂ 兴 ⫽3⫻3 matrix, whose components are the direction co-
dium within itself. The primary constituents here are the damaged sines of the base vectors e⬘i relative to the triad e*
i associated with
concrete c and the two families of reinforcement 1 and 2. The the considered reinforcement family. Applying now the stress
traction vector ṫ⬘ may be expressed in terms of respective aver- transformation rule, it can be shown that the traction vector on the
ages as localization plane, which is defined by the unit normal n, may be
evaluated as
ṫ⬘ ⫽␩ 1 ṫ⬘1 ⫹␩ 2 ṫ⬘2 ⫹ 共 1⫺␩ 1 ⫺␩ 2 兲 ṫ⬘c (23)
ṫ⬘1 ⫽ 共 nT e*
3 兲关 T̂ 兴 ṫ*
1 (30)
Assuming now that the velocity discontinuity ġ⬘ is the same in the Substituting now Eqs. 共29兲 and 共30兲 into Eq. 共25兲, and utilizing
primary constituents, one obtains Eqs. 共26兲 and 共28兲, one obtains after some algebraic manipula-
ṫ⬘ ⫽ 关 K ⬘ 兴 ġ⬘ ; 关 K ⬘ 兴 ⫽␩ 1 关 K ⬘1 兴 ⫹␩ 2 关 K ⬘2 兴 ⫹ 共 1⫺␩ 1 ⫺␩ 2 兲关 K ⬘c 兴 tions
(24) E T
ṫ⬘1 ⫽ 关 K ⬘1 兴 ġ⬘ ; 关 K ⬘1 兴 ⫽ 3 兲 关 T̂ 兴关 S * 兴关 T̂ 兴
共 n e* 2 T
(31)
Thus, the evaluation of the stiffness operator for the fractured h
zone requires the assessment of the stiffness for both families of The above expression defines the stiffness operator 关 K ⬘1 兴 , ap-
reinforcement as well as that for the degraded concrete. pearing in Eq. 共24兲, for the first family of reinforcement. Obvi-

728 / JOURNAL OF ENGINEERING MECHANICS © ASCE / JULY 2003

ously, the expression for the stiffness matrix 关 K ⬘2 兴 has exactly the the action of the normal stress. The yielding associated with the
same form as that for 关 K ⬘1 兴 with the understanding that the coor- shear force V is evaluated independently through the criterion
dinate system ei* is now associated with the second family of 共Horne 1979兲
reinforcement. In fact, the above procedure can easily be ex-
tended for an arbitrary number of reinforcement families. ␴Y
V⭐V p ; V p ⫽A s (40)
Finally, as mentioned earlier, the present approach is discrete 冑3
in nature, i.e., the reinforcing bars are idealized as beams of a
It is noted that an alternative formulation of the problem may
characteristic length l, rigidly embedded in the adjacent intact
involve a macroscopic criterion, which again would incorporate a
material. In general, the stiffness operator for the reinforcement
set of distribution functions defining the onset of yielding in the
network can also be evaluated using a purely phenomenological
reinforcement network as a function of its orientation relative to
approach. In particular, one could introduce a set of distribution
the loading direction. A general approach to formulation of such
functions, which describe the stiffness characteristics as a func-
criteria, in the context of frictional materials, is outlined in the
tion of the orientation of reinforcement relative to the loading
article by Pietruszczak and Mroz 共2001兲.
direction. These characteristics could then be implicitly identified
from the available experimental data. Such a simplified approach,
although not pursued further in this work, may be considered as a Stiffness Operator for Damaged Concrete
feasible alternative here. The residual stiffness of concrete in the fractured zone may be
evaluated by adopting, once again, either a phenomenological or a
micromechanical approach. Several different formulations have
Yield Criterion for Reinforcement Network
been proposed in the literature, ranging from simple nonlinear
This criterion is formulated again in a discrete sense by consid- approximations to more elaborated concepts taking into account a
ering the bending characteristics of the structural system exam- progressive evolution of idealized geometries of the microstruc-
ined in the preceding section. The starting point here is the speci- ture 共e.g., Dowding et al. 1991; Pietruszczak and Xu 1995兲. In the
fication of the stress resultants in the reinforcement network. formulation outlined here, the primary emphasis is on simplicity,
Given Eqs. 共25兲 and 共29兲, the average traction on a plane perpen- so that the number of material parameters involved is restricted.
dicular to the reinforcement axis can be evaluated as The constitutive model is derived within the framework of strain-
softening plasticity. In addition, the elastic stiffness degradation is
ṫ1* ⫽ 关 K 1* 兴关 T̂ 兴 T ġ⬘ (32) accounted for by introducing a scalar damage parameter.
The problem is referred to the local frame x⬘ associated with
At the same time, in view of Eqs. 共18兲, 共19兲, and 共20兲 the fractured zone. Assuming the additivity of elastic and plastic
ġ⬘ ⫽ 关 K ⬘ 兴 ⫺1 ṫ⬘ ⫽ 关 K ⬘ 兴 ⫺1 关 N ⬘ 兴 T ␴
˙ ⬘ ⫽ 关 K ⬘ 兴 ⫺1 关 M 兴 T ␴
˙ (33) deformation, i.e., ġ⬘ ⫽ġ⬘ e ⫹ġ⬘ p , the constitutive relation in the
elastic range is expressed as
so that
t⬘c ⫽ 共 1⫺␻ 兲关 K ⬘e 兴 g⬘ e (41)
⫺1
1 兴关 T̂ 兴 关 K ⬘ 兴
1 ⫽关K* 关 M 兴 T␴
T ˙
ṫ* (34)
or in a rate form
which defines the local traction in terms of an arbitrary stress rate.
The internal forces in individual members can now be deter- ṫ⬘c ⫽ 共 1⫺␻ 兲关 K ⬘e 兴 ġ⬘ e ⫺␻˙ 关 K ⬘e 兴 g⬘ e (42)
mined as In the equations above, 关 K e⬘ 兴 ⫽elastic stiffness and ␻⫽damage

F* 共 ␶ 兲 ⫽ 冕 0
␶
Ḟ* d␶; Ḟ* ⫽A s ṫ*
1 (35)
parameter. The matrix 关 K e⬘ 兴 is a diagonal operator, analogous to
that appearing in Eq. 共26兲.
In the plastic range, the flow rule is defined as
where the integration is performed over the entire loading history
⳵f
and A s ⫽cross-sectional area of the member. For the given bound- ġ⬘ p ⫽␭˙ (43)
ary conditions, the stress resultant system at the fixed support can ⳵t⬘c
be evaluated as
Focusing the attention here on the tensile regime, the yield func-
V⫽ 冑 F 21 ⫹F 22 ; M ⫽ 21 Vl; N⫽F 3 (36) tion is assumed in the form

where V⫽shear force; M⫽bending moment; and N⫽axial force. f 共 t⬘c ,␬兲 ⫽ 共 t ⬘1 兲 2 ⫹ 共 t ⬘2 兲 2 ⫹ 共 t ⬘3 兲 2 ⫺C 2 共 ␬ 兲 (44)
Following now the classical approach, as outlined by Horne where t 1⬘ through t 3⬘ represent the individual components of t⬘c and
共1979兲, the yield criterion may be expressed in a parabolic form t ⬘3 ⭓0. The parameter C, which according to Eq. 共44兲 represents

冏 冏冉 冊
M
Mp
⫹
N
Np
2
⭐1.0 (37)
the magnitude of the traction vector t⬘c , is assumed here to be a
decreasing function of a measure of accumulated plastic deforma-
tion ␬. The strain softening law is postulated in a simple expo-
where nential form
N p ⫽A s ␴ Y ; M p ⫽Z p ␴ Y (38) C⫽C 0 e ⫺␣␬ (45)
In Eq. 共38兲, ␴ Y ⫽yield stress in axial tension, whereas Z p ⫽plastic where ␣⫽material constant; and C 0 ⫽magnitude of the normal
section modulus. For a circular cross section with a diameter d, traction t 3⬘ at the onset of localization. The latter stems from the
one has fact that the fracture plane is assumed to form in the direction
normal to n, which is the direction of the major tensile stress in
A s ⫽␲d 2 /4; Z p ⫽d 3 /6 (39)
concrete, so that at the inception of localization t⬘c is colinear with
The criterion 共37兲 governs the yielding of a cross section under n.

JOURNAL OF ENGINEERING MECHANICS © ASCE / JULY 2003 / 729

 The softening effects, as well as the growth of the damage Numerical Analysis
parameter, are both attributed to the evolution of the normal com-
The constitutive model presented in the preceding sections incor-
ponent of velocity discontinuity g 3⬘ p . The damage function is as-
porates a number of material parameters which define the prop-
sumed in a simple hyperbolic form
erties of the constituents, i.e., the concrete matrix, reinforcing
steel as well as the characteristics of the fractured zone. Most of
␬ ⳵f
␻⫽ ; ␬˙ ⬅ġ ⬘3 p ⫽␭˙ (46) these parameters can be identified from standard tests on constitu-
B⫹␬ ⳵t ⬘3 ent materials. The properties of concrete have been described here
by employing a plasticity framework outlined in the article by
where B⫽material parameter with dimension of 关length兴. It is Pietruszczak et al. 共1988兲. The primary material parameters in-
evident that ␬⫽0 gives ␻⫽0, whereas ␬→⬁ ⇒ ␻→1, so that clude those governing the elastic response (E c ,␯ c) and the speci-
␻⑀关0,1兲. fication of the conditions at failure. The latter are explicitly de-
Following now the standard plasticity procedure, the consis- fined in terms of f c 共compressive strength兲 and f t 共tensile
tency condition may be written as strength兲. The behavior of the reinforcing steel is governed by
(E 1 ,␯ 1 ) and (E 2 ,␯ 2 ), which specify the elastic properties of both
⳵f ⳵f
mT ṫ ⬘c ⫹ ␬˙ ⫽0; m⫽ (47) families, as well as ␴ Y and d, which represent the yield stress and
⳵␬ ⳵t⬘c the diameter of the bars, respectively. Finally, the properties of the
fractured zone are defined in terms of the parameter ␣, with di-
Expressing the elastic constitutive relation 共42兲 in the form mension 关1/length兴, which specifies the rate of degradation of the
cracked concrete, Eq. 共45兲; the parameter B 关length兴, which gov-
⳵␻
ṫ⬘c ⫽ 共 1⫺␻ 兲关 K ⬘e 兴共 ġ⬘ ⫺ġ⬘ p 兲 ⫺ ␬˙ 关 K ⬘e 兴共 g⬘ ⫺g⬘ p 兲 (48) erns the evolution of the damage variable, Eq. 共46兲; and the thick-
⳵␬ ness h of the fractured zone. In addition, the ‘‘characteristic di-
and substituting Eq. 共48兲, together with Eq. 共43兲, into Eq. 共47兲, the mension’’ ␮ 共关1/length兴兲 is required, Eq. 共16兲, which is defined as
the ratio of the area of the fracture plane to the volume of the
plastic multiplier ␭˙ may be evaluated as
adjacent intact zone.
1 The numerical simulations discussed below pertain to a set of
␭˙ ⫽ 共共 1⫺␻ 兲 mT 关 K ⬘e 兴 兲 ġ⬘ ; H⫽H e ⫹H p ⫹H d (49) experimental tests conducted by Vecchio and Collins 共1982兲, and
H
Pang and Hsu 共1995兲. The experiments involved reinforced con-
where crete 共RC兲 panels tested in pure shear with various reinforcement
ratios, ␩ 1 , ␩ 2 , and different concrete strengths. In general, the
⳵f information provided in both references is sufficient to identify
H e ⫽ 共 1⫺␻ 兲 mT 关 K ⬘e 兴 m; H p ⫽⫺ m ;
⳵␬ 3 the basic properties of concrete and the reinforcing steel. For
(50) concrete, the values of f c and f t have been provided, together
⳵␻ with estimates of elastic properties. Note that the formulation by
H d⫽ m mT 关 K e⬘ 兴 g⬘ e Pietruszczak et al. 共1988兲 employs some additional material pa-
⳵␬ 3
rameters which define the behavior in the elastoplastic range, i.e.,
Substituting now Eq. 共49兲 in Eq. 共48兲, the following constitutive those appearing in the hardening and plastic potential functions.
relation is obtained after some algebraic transformations: The values of these parameters have been chosen following the
specific suggestions provided in the original reference. It should
ṫ⬘c ⫽ 关 K ⬘c 兴 ġ⬘ be emphasized, however, that for the loading histories considered
here, the behavior of the concrete matrix is predominantly elastic-
1 brittle, so that these parameters have virtually no influence on the
关 K ⬘c 兴 ⫽ 共 1⫺␻ 兲关 K ⬘e 兴 ⫺ 共 1⫺␻ 兲 2 关 K ⬘e 兴 mmT 关 K ⬘e 兴 (51)
H mechanical characteristics. For the panels tested by Pang and Hsu
共1995兲 the compressive strength f c was of the order of 40 MPa,
1 ⳵␻ whereas for those tested by Vecchio and Collins 共1982兲 f c was in
⫺ 共 1⫺␻ 兲 m 关 K ⬘ 兴 g⬘ e mT 关 K ⬘e 兴
H ⳵␬ 3 e the range of 15–30 MPa. The specific values of both the strengths
and elastic parameters, for each test, are provided in the indi-
The above expression defines the stiffness operator 关 K ⬘c 兴 , which vidual figure captions.
appears in Eq. 共24兲, thereby completing the general formulation For the reinforcing steel, the basic material/geometric proper-
of the problem as outlined in the preceding section. It is evident ties were as follows:
that for ␻⫽0, i.e., when the elastic stiffness degradation is not
accounted for, there is H d ⫽0, Eq. 共50兲, and the last term in Eq. E 1 ⫽E 2 ⫽210 GPa; ␯ 1 ⫽␯ 2 ⫽0.33; d 1 ⫽d 2 ⫽0.025 m
共51兲 vanishes. In this case, the representation 共51兲 reduces to the The volume fractions of reinforcement and the yield strength ␴ Y
standard form associated with elastoplasticity. were different for each set of tests.
Finally, it should be noted again that the functional form of the For the fractured zone, the main difficulty arises in the context
yield surface employed here, viz. Eq. 共44兲, is associated with the of specification of the thickness h, as well as the parameter ␣,
tensile regime. Indeed, in reinforced concrete, the localization of which controls the rate of softening of the damaged concrete, Eq.
deformation commonly involves t ⬘3 ⭓0, i.e., tension. In order to 共45兲. In general, h is of the same order of magnitude as d, while ␣
apply the constitutive relation 共51兲 in the context of macrocracks should be determined from mechanical characteristics in a direct
forming in the compression regime, the criterion 共44兲 needs to be tension test carried out on a plain concrete specimen. Given the
extended to include a Coulomb type of envelope for t 3⬘ ⭐0. In this fact that no such data are available here, the numerical simula-
range, the softening effects can be attributed to the evolution of tions presented below were conducted assuming a fixed value of
the tangential component of the velocity discontinuity 共cf. h⫽0.04 m, while the corresponding value of ␣ was chosen by a
Pietruszczak and Xu 1995兲. trial and error procedure. In particular, one test from each group

730 / JOURNAL OF ENGINEERING MECHANICS © ASCE / JULY 2003

was selected at random 共panel A2 for Pang and Hsu and panel
PV16 for Vecchio and Collins兲 and the value of ␣ was estimated
so that to obtain a close approximation to the actual global stiff-
ness characteristic. This resulted in the specific values of
␣⫽20,000 m⫺1 共for tests of Pang and Hsu兲 and ␣⫽25,000 m⫺1
共for tests of Vecchio and Collins兲, which were subsequently used
for the validation purpose, i.e., prediction of the remaining tests.
It should be noted that the value of ␣ affects primarily the stiff-
ness of the composite immediately after the initiation of cracking
and has a marginal effect on the ultimate strength. This is evident
from the results of the parametric studies, which are presented
later in this section. For all simulations involving monotonic load-
ing histories, the degradation of elastic properties in the fractured
zone was neglected (B→⬁⇒␻→0 in Eq. 共46兲兲. The effect of the
parameter B was later examined in the context of cyclic loading,
as reported in the section on parametric studies. Finally, for both
series of tests considered here, the fracture pattern, as reported in
the original references, involved a set of multiple cracks with
approximately the same spacing of s⫽0.053 m. Based on this
observation, the ‘‘characteristic dimension’’ ␮ was taken as
␮⫽1/s⫽18.7 m⫺1 . The values of the key material parameters are
provided in the individual figure captions. In all figures, the nu-
merical predictions are shown by a solid line, whereas the dots Fig. 3. Numerical simulation of pure shear test on panel A1 共Pang
represent the experimental data. and Hsu 1995兲; f c ⫽42.2 MPa, E c ⫽40 GPa, f t ⫽1.10 MPa,
␣⫽20,000 m⫺1, ␴ Y ⫽444 MPa, and ␩ 1 ⫽␩ 2 ⫽0.00596

Tests in Pure Shear

In this section, the numerical simulations of the experimental tests not yet been attained. The predictions are, again, in a quite rea-
conducted by Pang and Hsu 共1995兲 are presented first. The ex- sonable agreement with the experimental data.
periments were performed on the panels with dimensions 140 Finally, Figs. 9–11 present the results of numerical simulations
⫻140 cm and the thickness of 17.8 cm. For the purpose of com- for a group of panels tested with different reinforcement ratios in
parison, three panels were chosen, A1 through A3, which were both directions. Here, only the panels for which the actual failure
loaded in pure shear with the same percentage of reinforcement in mode was associated with yielding of steel were selected, i.e.,
both directions. PV10, PV11, and PV27. For these panels, the volume fraction of
The results of the numerical simulations, corresponding to the
reinforcement ratios in the range from 0.6 to about 1.8%, are
presented in Figs. 3–5. The numerical curves clearly depict the
three different stages of the deformation process. The linear
branch corresponds to a homogeneous mode associated with
phase I, i.e., prior to cracking. The subsequent nonlinear branch
describes the deformation process associated with strain localiza-
tion and a progressive degradation of interface properties. Finally,
the branch which displays a plateau, corresponds to the yielding
of steel. The predictions are, in general, fairly consistent with the
experimental data.
The next set of simulations reported here pertains to the ex-
perimental tests carried out by Vecchio and Collins 共1982兲. The
tests were performed on panels with dimensions 89⫻89 cm and
the thickness of 7 cm. The simulations shown in Figs. 6 – 8 have
been carried out for the group of panels PV4, PV6, and PV16,
which were reinforced with the same intensity in both directions.
Here, the reinforcement ratios were in the range from 0.7 to about
1.8%, which is similar to that for tests of Pang and Hsu. This
time, however, the concrete strength was much lower and differed
considerably from panel to panel. Also, the yield stress for steel
was significantly lower. Despite the differences in key properties,
the general trends, as depicted in these figures, remain the same as
before, i.e., three different stages of the deformation process can
be clearly distinguished. It is noted that in the last phase, associ-
ated with yielding of steel, an unstable 共i.e., descending兲 branch is Fig. 4. Numerical simulation of pure shear test on panel A2 共Pang
and Hsu 1995兲; f c ⫽41.2 MPa, E c ⫽39 GPa, f t ⫽1.50 MPa,
sometimes visible prior to reaching a plateau. This is due to the
␣⫽20,000 m⫺1, ␴ Y ⫽462 MPa, and ␩ 1 ⫽␩ 2 ⫽0.01193
fact that the residual strength of concrete in the damaged zone has

JOURNAL OF ENGINEERING MECHANICS © ASCE / JULY 2003 / 731

Fig. 5. Numerical simulation of pure shear test on panel A3 共Pang Fig. 7. Numerical simulation of pure shear test on panel PV6 共Vec-
and Hsu 1995兲; f c ⫽41.6 MPa, E c ⫽43 GPa, f t ⫽1.50 MPa, chio and Collins 1982兲; f c ⫽29.8 MPa, E c ⫽24 GPa, f t ⫽1.60 MPa,
␣⫽20,000 m⫺1, ␴ Y ⫽446 MPa, and ␩ 1 ⫽␩ 2 ⫽0.01789 ␣⫽25,000 m⫺1, ␴ Y ⫽266 MPa, and ␩ 1 ⫽␩ 2 ⫽0.01785

the reinforcement for the fist family was kept constant, before and after yielding of the reinforcement family No. 2 共as
␩ 1 ⫽0.01785, while ␩ 2 was varied. The dimensions of the panels ␩ 2 ⬍␩ 1 ). The ultimate load is reached at the stage when the first
and the loading conditions were the same as those for the previ- family yields.
ous group. For these tests, the yield stress for steel was in the
range from 235 to 442 MPa, while the compressive strength of
Parametric Studies
concrete was relatively low 共14.5–20.5 MPa兲. Examining the pre-
dicted characteristics, it is evident here that the branch associated In this part, the results of a set of parametric studies are presented
with strain localization consists now of two distinct parts, i.e., examining the influence of some material parameters on the over-

Fig. 6. Numerical simulation of pure shear test on panel PV4 Fig. 8. Numerical simulation of pure shear test on panel PV16
共Vecchio and Collins 1982兲; f c ⫽26.6 MPa, E c ⫽21 GPa, f t ⫽1.20 共Vecchio and Collins 1982兲; f c ⫽21.7 MPa, E c ⫽22 GPa, f t ⫽0.80
MPa, ␣⫽25,000 m⫺1, ␴ Y ⫽242 MPa, and ␩ 1 ⫽␩ 2 ⫽0.01056 MPa, ␣⫽25,000 m⫺1, ␴ Y ⫽255 MPa, and ␩ 1 ⫽␩ 2 ⫽0.00740

732 / JOURNAL OF ENGINEERING MECHANICS © ASCE / JULY 2003

 Fig. 11. Numerical simulation of pure shear test on panel PV27
Fig. 9. Numerical simulation of pure shear test on panel PV10 共Vecchio and Collins 1982兲; f c ⫽20.5 MPa, E c ⫽22 GPa,
共Vecchio and Collins 1982兲; f c ⫽14.5 MPa, E c ⫽11 GPa, f t ⫽1.26 f t ⫽1.49 MPa, ␣⫽25,000 m⫺1, ␴ Y ⫽442 MPa, and ␩ 1 ⫽␩ 2
MPa, ␣⫽25,000 m⫺1, ␴ Y ⫽276 MPa, ␩ 1 ⫽0.01785, and ␩ 2 ⫽0.01785
⫽0.00999

all behavior of the composite. Figs. 12–14 present the results out assuming the same set of basic material parameters as before,
corresponding to the case of axial tension in the direction of the i.e., same properties of reinforcement, thickness of the damaged
first family of reinforcement. The main variables here are: the zone, etc. The values of the remaining constants are, once again,
parameter ␣, which specifies the rate of degradation of the dam- provided in the respective figure captions.
aged concrete, Eq. 共45兲, the tensile strength f t and the percentage Fig. 12 shows the results corresponding to different values of
of reinforcement. The numerical simulations have been carried the parameter ␣, ranging from 30,000 to 120,000 m⫺1. It is evi-
dent that an increase in the rate of degradation of the damaged
concrete leads eventually to a local instability developing at the
inception of phase II 共descending branch in Fig. 12兲. However,

Fig. 10. Numerical simulation of pure shear test on panel PV11

共Vecchio and Collins 1982兲; f c ⫽15.6 MPa, E c ⫽12 GPa, Fig. 12. Numerical simulations of uniaxial tension test; influence of
f t ⫽1.30 MPa, ␣⫽25,000 m⫺1, ␴ Y ⫽235 MPa, ␩ 1 ⫽0.01785, and softening parameter ␣; f c ⫽30 MPa, E c ⫽25 GPa, f t ⫽3.0 MPa, ␴ Y
␩ 2 ⫽0.01306 ⫽400 MPa, and ␩ 1 ⫽␩ 2 ⫽0.05

JOURNAL OF ENGINEERING MECHANICS © ASCE / JULY 2003 / 733

Fig. 13. Numerical simulations of uniaxial tension test; influence of Fig. 15. Numerical predictions of uniaxial tension tests at different
tensile strength f t ; f c ⫽30 MPa, E c ⫽25 GPa, ␣⫽60,000 m⫺1, ␴ Y orientation of reinforcement 共␤兲; f c ⫽30 MPa, E c ⫽25 GPa,
⫽400 MPa, and ␩ 1 ⫽␩ 2 ⫽0.05 f t ⫽3.0 MPa, ␣⫽60,000 m⫺1, ␴ Y ⫽400 MPa, and ␩ 1 ⫽␩ 2 ⫽0.05

the ultimate load, which corresponds to the yielding of reinforce- eter. Finally, Fig. 14 shows the predictions for different intensities
ment, is virtually independent of the value of ␣. It should be of reinforcement. As expected, the ultimate strength strongly de-
mentioned here that qualitatively similar trends have also been pends on the percentage of reinforcement. For plane concrete the
obtained for tests involving pure shear. strength is equal to f t ⫽3 MPa, while at ␩ 1 ⫽␩ 2 ⫽0.05, the ulti-
Fig. 13 shows the results of a parametric study examining the mate load is in the range of 20 MPa.
influence of the tensile strength f t on the global mechanical char- The next aspect of the parametric studies reported here is the
acteristics. The simulations correspond to f t 傺 关 1.0,5.0兴 MPa and examination of mechanical characteristics in tension and pure
a fixed value of ␣⫽60,000 m⫺1. It is clear that the value of f t shear as a function of the orientation of the reinforcement. Fig. 15
affects primarily the load intensity at which macrocracks form. shows the results corresponding to axial tension, whereas Fig. 16
Again, the ultimate load is virtually independent of this param- presents those pertaining to pure shear. The angle ␤, as identified

Fig. 14. Numerical simulations of uniaxial tension test; influence of Fig. 16. Numerical predictions of pure shear tests at different orien-
reinforcement ratio; f c ⫽30 MPa, E c ⫽25 GPa, f t ⫽3.0 MPa, tation of reinforcement 共␤兲; f c ⫽30 MPa, E c ⫽25 GPa, f t ⫽3.0 MPa,
␣⫽60,000 m⫺1, and ␴ Y ⫽400 MPa ␣⫽60,000 m⫺1, ␴ Y ⫽400 MPa, and ␩ 1 ⫽␩ 2 ⫽0.05

734 / JOURNAL OF ENGINEERING MECHANICS © ASCE / JULY 2003

 Fig. 19. Numerical prediction of cyclic tension test; damage param-
Fig. 17. Distribution of yield load with orientation of reinforcement; eter: B⫽140⫻10⫺6 m; f c ⫽30 MPa, E c ⫽25 GPa, f t ⫽3.0 MPa,
axial tension; f c ⫽30 MPa, E c ⫽25 GPa, f t ⫽3.0 MPa, ␣⫽60,000 ␣⫽60,000 m⫺1, ␴ Y ⫽400 MPa, and ␩ 1 ⫽␩ 2 ⫽0.05
m⫺1, ␴ Y ⫽400 MPa, and ␩ 1 ⫽␩ 2 ⫽0.05

The distribution displays a symmetry with respect to ␤⫽45°,

which is due to the fact that the reinforcement ratios are the same
in these figures, refers to that between the x 3 and x 3* axes, as for both families, i.e., ␩ 1 ⫽␩ 2 .
shown in Fig. 1. Comparing the results in tension and shear, it is Finally, Figs. 19 and 20 present the results of the simulations
evident that the trends are essentially the same. For ␤ within the for axial tension involving a series of loading–unloading cycles at
range from 0° to 20°, the strength progressively decreases. The different stress intensities. The simulations have been carried out
maximum value corresponds to the orientations of ␤⫽0 and incorporating the degradation of elastic stiffness, as defined in
␤⫽45°. Figs. 17 and 18 show the distribution of the load intensity Eqs. 共41兲 and 共46兲. The results correspond to two different values
at which the yielding occurs in the first family of reinforcement. of the parameter B, which controls the rate of damage. It is evi-
dent here that a smaller value of B leads to a more pronounced

Fig. 18. Distribution of yield load with orientation of reinforcement; Fig. 20. Numerical prediction of cyclic tension test; damage param-
pure shear; f c ⫽30 MPa, E c ⫽25 GPa, f t ⫽3.0 MPa, eter: B⫽8⫻10⫺6 m; f c ⫽30 MPa, E c ⫽25 GPa, f t ⫽3.0 MPa,
␣⫽60,000 m⫺1, ␴ Y ⫽400 MPa, and ␩ 1 ⫽␩ 2 ⫽0.05 ␣⫽60,000 m⫺1, ␴ Y ⫽400 MPa, and ␩ 1 ⫽␩ 2 ⫽0.05

JOURNAL OF ENGINEERING MECHANICS © ASCE / JULY 2003 / 735

degradation of the elastic stiffness, which is clearly visible during ˙ 共 1 兲⫽ 关 B 共 1 兲兴 ␴
␴ ˙; ˙ 共 2 兲⫽ 关 B 共 2 兲兴 ␴
␴ ˙; ␴
˙ m⫽ 关 B m兴 ␴
˙ (52)
the unloading and reloading phase. Moreover, for a fixed value of
B, the stiffness on unloading/reloading decreases with an increase with the understanding that all operators are referred to the base
x* and ␴˙ ⫽ 兵 ␴˙ 11 ,␴˙ 22 ,␴˙ 33 ,␴˙ 12 ,␴˙ 13 ,␴˙ 23其 T .
in the external load.
The static equilibrium constraints, defined in Eq. 共5兲 are as-
sumed in the form
Final Remarks ␴˙ 共i 1j 兲 ⫽␴˙ 共i 2j 兲 ⫽␴˙ i j for i, j⫽1,2,3 i⫽ j
(53)
In this work, a constitutive model has been presented for concrete ␴˙ 共111 兲 ⫽␴˙ 共112 兲 ⫽␴˙ 11 ; ␴˙ m ˙ 共2兲
33⫽␴ 33 ; ␴˙ m ˙ 共1兲
22⫽␴ 22
strengthened with orthogonal sets of reinforcing steel bars. The
formulation addresses two main stages of the deformation pro- whereas the kinematic ones are taken as
cess, i.e., a homogeneous mode as well as that involving localized ␧˙ m ˙ 共1兲
33⫽␧ 33 ; ␧˙ m ˙ 共2兲
22⫽␧ 22 (54)
deformation, associated with formation of macrocracks. In the
former case, i.e., prior to cracking, the problem is formulated by As mentioned earlier, the above approach is similar to that pur-
invoking a simple volume averaging procedure. After the onset of sued by Wung and Dvorak 共1985兲. Thus, according to Eq. 共54兲,
localization, the representative volume incorporates the fractured the compatibility requirements are imposed on the axial deforma-
zone and the adjacent ‘‘intact’’ material, both reinforced with steel tion in the direction of reinforcement, while the interaction be-
bars. For the latter case, the stiffness of the reinforcement network tween the phases in the transverse direction is largely neglected.
and the criterion of yielding are both assessed in a discrete sense, Substituting now the constitutive relations 共2兲–共3兲, together
by considering the bending characteristics of a structural system with static constraints 共53兲 into the set of Eqs. 共54兲, the expres-
comprising individual bars, with a characteristic length l, rigidly sions defining individual components of the structural matrices
embedded in the adjacent intact material. Extensive numerical 关B兴 may be derived. For a special case, when the reinforcing steel
study has been conducted examining the behavior of RC panels in represents an elastic material with Young’s moduli E 1 ,E 2 and
pure shear and tension for various reinforcement ratios, different Poisson’s ratios ␯ 1 ,␯ 2 , the following form of the matrix 关 B m 兴 is
concrete strengths as well as different orientations of the rein- obtained:
forcement network. The results clearly indicate that both qualita-
i j ⫽0
Bm for i, j⫽1,4,5,6 and i⫽ j
tive and quantitative aspects of the numerical predictions are
fairly consistent with the available experimental data. B iim ⫽1 for i⫽1,4,5,6 (55)
It is noted that, within the general framework outlined here, an
alternative approach may also be pursued. In particular, if the c
21⫽ a 关 ⫺ 共 E 2 C 21⫹␯ 2 兲 ⫹b 共 E 1 C 31⫹␯ 1 兲共 E 2 C 23⫹␯ 2 兲兴
Bm m m m
inelastic deformation of concrete prior to onset of localization is

冉 冊
neglected, then more rigorous estimates can be employed for the
c 1
structural matrices 关B兴, Eq. 共4兲, based on the existing analytical
22⫽
Bm
a ␩2
solutions 共e.g., Eshelby 1957兲. In this case, the transition to local-
ized mode may be defined by means of a macroscopic criterion, bc
which incorporates a microstructure-orientation tensor 23⫽⫺
Bm 共 E C m ⫹␯ 兲
a␩ 1 2 23 2
共Pietruszczak and Mroz 2001兲. After localization, the stiffness
characteristics of the reinforcement network can also be assessed c
24⫽ a 关 ⫺E 2 C 24⫹bE 1 C 34共 E 2 C 23⫹␯ 2 兲兴
Bm m m m
based on a phenomenological approach, whereby a set of distri-
bution functions is introduced describing the variation of elastic
moduli with the orientation of the reinforcement relative to the c
25⫽ a 关 ⫺E 2 C 25⫹bE 1 C 35共 E 2 C 23⫹␯ 2 兲兴
Bm m m m
macrocrack.

c
26⫽ a 关 ⫺E 2 C 26⫹bE 1 C 36共 E 2 C 23⫹␯ 2 兲兴
Bm m m m
Acknowledgments

The writers wish to acknowledge the financial support of Unité b

31⫽ a 关 ⫺ 共 E 1 C 31⫹␯ 1 兲 ⫹c 共 E 1 C 32⫹␯ 1 兲共 E 2 C 21⫹␯ 2 兲兴
Bm m m m
Mécanique, Hydro-Quebec, Montreal. The contribution of Profes-
sor G. N. Pande 共University of Swansea, U.K.兲, in terms of nu-
merous discussions in relation to this work, is also acknowledged. bc
32⫽⫺
Bm 共 E C m ⫹␯ 兲
a␩ 2 1 32 1

Appendix A
33⫽
Bm 冉 冊
b 1
a ␩1
In this appendix, the static and kinematic constraints, Eqs. 共5兲, are
specified and the components of structural matrices, appearing in b
34⫽ a 关 ⫺E 1 C 34⫹cE 2 C 24共 E 1 C 32⫹␯ 1 兲兴
Bm m m m
Eqs. 共4兲, are explicitly defined. The problem is formulated here in
the frame of reference x* attached to the reinforcement, Fig. 1.
b
The x *1 axis is assumed to be normal to the plane of reinforce- 35⫽ a 关 ⫺E 1 C 35⫹cE 2 C 25共 E 1 C 32⫹␯ 1 兲兴
Bm m m m
ment, whereas x 2* and x 3* are taken along the second and the first
families, respectively. In order to maintain the clarity of presen- b
36⫽ a 关 ⫺E 1 C 36⫹cE 2 C 26共 E 1 C 32⫹␯ 1 兲兴
Bm m m m
tation, the mathematical notation is slightly altered here to allow
an explicit reference to the individual components of stress/strain
rates. Thus, Eqs. 共4兲 are cast in the form where

736 / JOURNAL OF ENGINEERING MECHANICS © ASCE / JULY 2003

 32⫹␯ 1 兲共 E 2 C 23⫹␯ 2 兲
a⫽1⫺bc 共 E 1 C m m

␩1
b⫽ (56)
33⫺1 兲
1⫹␩ 1 共 E 1 C m

␩2
c⫽
22⫺1 兲
1⫹␩ 2 共 E 2 C m
The structural matrix for the first family of reinforcement, i.e.,
关 B (1) 兴 , takes the form
B 共i 1j 兲 ⫽0 for i, j⫽1,4,5,6 and i⫽ j Fig. 21. Geometry of problem
B 共ii1 兲 ⫽1 for i⫽1,4,5,6

B 共2i1 兲 ⫽B 2i
m
for i⫽1,2¯6

B 共311 兲 ⫽B m
21共 E 1 C 32⫹␯ 1 兲 ⫹B 31E 1 C 33⫹E 1 C 31⫹␯ 1
m m m m
12EI
K t⫽ (64)
B 共321 兲 ⫽B m l3
22共 E 1 C 32⫹␯ 1 兲 ⫹B 32E 1 C 33
m m m
(57)

B 共331 兲 ⫽B m
23共 E 1 C 32⫹␯ 1 兲 ⫹B 33E 1 C 33
m m m whereas for l→0 the shear stiffness of a solid body is recovered

B 共341 兲 ⫽B m GA
24共 E 1 C 32⫹␯ 1 兲 ⫹B 34E 1 C 33⫹E 1 C 34
m m m m
K t⫽ (65)
l
B 共351 兲 ⫽B m
25共 E 1 C 32⫹␯ 1 兲 ⫹B 35E 1 C 33⫹E 1 C 35
m m m m
Eq. 共63兲 can also be expressed in the form
B 共361 兲 ⫽B m
26共 E 1 C 32⫹␯ 1 兲 ⫹B 36E 1 C 33⫹E 1 C 36
m m m m

The matrix 关 B (2) 兴 can be defined by noting that the following

identity holds:
K t⫽
EA 1 12EI G EA
l 1⫹␾ GAl 2 E
⫽
␾ G
l 1⫹␾ E
⫽
EA
l冉␥ 冊 (66)

␩ 1 关 B 共 1 兲 兴 ⫹␩ 2 关 B 共 2 兲 兴 ⫹ 共 1⫺␩ 1 ⫺␩ 2 兲关 B m 兴 ⫽ 关 I 兴 (58) where

from which ␾ G ␾
␥⫽ ⫽ (67)
1 1⫹␾ E 2 共 1⫹␾ 兲共 1⫹␯ 兲
关 B共2兲 兴⫽ 关关 I 兴 ⫺␩ 1 关 B 共 1 兲 兴 ⫺ 共 1⫺␩ 1 ⫺␩ 2 兲关 B m 兴兴 (59)
␩2 For a circular cross section, with a diameter d, the parameter ␾ in
where 关I兴⫽identity operator. Eq. 共63兲, can be evaluated as

Appendix B ␾⫽ 23 共 1⫹␯ 兲 冉冊
d
l
2
(68)

Referring to Fig. 21, consider a member of length l, which is Thus, the shear stiffness, which appears in Eq. 共26兲, is defined as
rigidly embedded in the concrete matrix and is subjected to a
tangential displacement u. Using Timoshenko’s beam theory, the K t E␥
shear stiffness may be defined by decomposing the displacement k t⫽ ⫽ (69)
A l
into a part resulting from the action of the bending moment, u b ,
and that associated with the shear force, u v , i.e., while k n ⫽E/l.
u⫽u b ⫹u v (60)
Both these components are defined through a set of differential References
equations
du v V d 2u b M Aoyagi, Y., and Yamada, K. 共1979兲. ‘‘An experimental approach to the
⫽⫺ ; ⫽ (61) design of the network reinforcement against in-plane shear in rein-
dx GA dx 2 EI forced concrete containments.’’ Proc., 5th Int. Conf. Struct. Mech.
Reac. Techn. (SMiRT), Berlin, Div. J4/7, 1–10.
where G⫽shear modulus; and I⫽second moment of the cross-
ASCE Task Committee on Finite Element Analysis of Reinforced Con-
sectional area A. For the given boundary conditions, the solution crete Structures of the Structural Division Committee on Concrete and
takes the form 共cf. Przemieniecki 1968兲 Masonry Structures. 共1982兲. ASCE State-of-the-Art Rep. on Finite El-
V⫽K t u (62) ement Analysis of Reinforced Concrete, New York.
Baumann, T. 共1972兲. ‘‘Tragwirkung ortogonaler Bewehrungsnetze belie-
where V and u are evaluated at point A, Fig. 21, and biger Richtung in Flächentragwerken aus Stahlbeton.’’ DAfSB, Heft
217, Berlin.
12EI 12EI Bažant, Z. P., and Oh, B. H. 共1983兲. ‘‘Crack band theory for fracture of
K t⫽ ; ␾⫽ (63)
共 1⫹␾ 兲 l 3
GAl 2 concrete.’’ Mater. Struct., 16共93兲, 155–177.
Benveniste, Y. 共1985兲. ‘‘The effective mechanical behavior of composite
It should be noted that for l→⫹⬁ the shear stiffness K t reduces materials with imperfect contact between the constituents.’’ Mech.
to that corresponding to the Bernoulli–Euler’s beam theory Mater., 4, 197–208.

JOURNAL OF ENGINEERING MECHANICS © ASCE / JULY 2003 / 737

Bhide, S. B., and Collins, M. P. 共1989兲. ‘‘Influence of axial tension on the Ortiz, M., Leroy, Y., and Needleman, A. 共1987兲. ‘‘A finite element method
shear capacity of reinforced concrete members.’’ ACI Struct. J., 86共5兲, for localized failure analysis.’’ Comput. Methods Appl. Mech. Eng.,
570–581. 61, 189–214.
de Borst, R. 共2001兲. ‘‘Some recent issues in computational failure me- Pang, X.-B. D., and Hsu, T. T. C. 共1995兲. ‘‘Behavior of reinforced con-
chanics.’’ Int. J. Numer. Methods Eng., 52, 63–95. crete membrane elements in shear.’’ ACI Struct. J., 92共6兲, 665– 679.
CEB/FIP 共1993兲. ‘‘CEB–FIP Model Code 90 共Final version兲.’’ Comité Peter, J. 共1966兲. ‘‘Zur Bewehrung von Scheiben und Schalen für
EuroInternational du Béton, Bull. d’Information No. 213-214, Lau- Hauptspannungen schiefwinklig zur Bewehrungsrichtung.’’ Bautech-
sanne, Switzerland. nik, 6, 149–154; 7, 240–248.
Červenka, V. 共1985兲. ‘‘Constitutive model for cracked reinforced con- Pietruszczak, S. 共1999兲. ‘‘On homogeneous and localized deformation in
crete.’’ ACI J., 82, 877– 882. water-infiltrated soils.’’ Int. J. Damage Mech., 8, 233–253.
Crisfield, M. A., and Wills, J. 共1989兲. ‘‘Analysis of RC panels using Pietruszczak, S., Jiang, J., and Mirza, F. A. 共1988兲. ‘‘An elastoplastic
constitutive model for concrete.’’ Int. J. Solids Struct., 24共7兲, 705–
different concrete models.’’ J. Eng. Mech., 115共3兲, 578 –597.
722.
Dowding, C. H., Zubelewicz, A., O’Connor, K. M., and Belytschko, T. B.
Pietruszczak, S., and Mróz, Z. 共2001兲. ‘‘On failure criteria for anisotropic
共1991兲. ‘‘Explicit modeling of dilation, asperity degradation, and cy-
cohesive-frictional materials.’’ Int. J. Numer. Analyt. Meth. Geomech.,
clic seating of rock joints.’’ Comput. Geotech., 11, 209–228.
25, 509–524.
Eshelby, J. D. 共1957兲. ‘‘The determination of elastic field of an ellipsoidal
Pietruszczak, S., and Niu, X. 共1993兲. ‘‘On the description of localized
inclusion and related problems.’’ Proc. R. Soc. London, Ser. A, 241, deformation.’’ Int. J. Numer. Analyt. Meth. Geomech., 17, 791– 805.
376 –396. Pietruszczak, S., and Xu, G. 共1995兲. ‘‘Brittle response of concrete as a
Gupta, A. K. 共1981兲. ‘‘Membrane reinforcement in shells.’’ J. Struct. Div., localization problem.’’ Int. J. Solids Struct., 32, 1517–1533.
ASCE, 107共1兲, 41–56. Przemieniecki, J. S. 共1968兲. Theory of matrix structural analysis,
Gupta, A. K., and Akbar, H. 共1984兲. ‘‘Cracking in reinforced concrete McGraw-Hill, New York.
analysis.’’ J. Struct. Eng., 110共8兲, 1735–1746. Rudnicki, J. W., and Rice, J. R. 共1975兲. ‘‘Conditions for the localization
Hill, R. 共1963兲. ‘‘Elastic properties of reinforced solids: some theoretical of deformation in pressure sensitive dilatant materials.’’ J. Mech.
properties.’’ J. Mech. Phys. Solids, 11, 357–372. Phys. Solids, 23, 371–394.
Horne, M. R. 共1979兲. Plastic theory of structures, Pergamon, Oxford, Scordelis, A. C. 共1972兲. ‘‘Finite element analysis of reinforced concrete
U.K. structures.’’ Proc., Spec. Conf. on the FEM in Civil Eng., Montreal.
Larsson, R., Runesson, K., and Ottosen, N. S. 共1993兲. ‘‘Discontinuities Simo, J. C., Oliver, J., and Armero, F. 共1983兲. ‘‘An analysis of disconti-
displacement approximation for capturing plastic localization.’’ Int. J. nuities induced by strain-softening in rate-independent inelastic sol-
Numer. Methods Eng., 36, 2087–2105. ids.’’ Comput. Mech., 12, 277–296.
Leitz, H. 共1923兲. ‘‘Eisenbewehrte Platten bei allgemeinem Biegungs- Suidan, M., and Schnobrich, W. C. 共1973兲. ‘‘Finite element analysis of
zustand.’’ Bautechnik, 16 –17. reinforced concrete.’’ J. Struct. Div., ASCE, 99共10兲, 2109–2122.
Lourenço, P., and Figueiras, J. A. 共1995兲. ‘‘Solution for the design of Vecchio, F., and Collins M. P. 共1982兲. ‘‘The response of reinforced con-
reinforced concrete plates and shells.’’ J. Struct. Eng., 121共5兲, 815– crete to in-plane shear and normal stresses.’’ Publ. No. 82-03, Univ. of
823. Toronto, Dept. of Civ. Eng., Toronto.
Ortiz, M. 共1987兲. ‘‘An analytical study of the localized failure modes of Wung, C. J., and Dvorak, G. J. 共1985兲. ‘‘Strain-space plasticity analysis
concrete.’’ Mech. Mater., 6, 159–174. of fibrous composites.’’ Int. J. Plast., 125–139.

738 / JOURNAL OF ENGINEERING MECHANICS © ASCE / JULY 2003