Vous êtes sur la page 1sur 13

International Journal of Solids and Structures 94–95 (2016) 125–137

Contents lists available at ScienceDirect

International Journal of Solids and Structures


journal homepage: www.elsevier.com/locate/ijsolstr

3D elastoplastic damage model for concrete based on novel


decomposition of stress
Ji Zhang a,b, Jie Li b,∗, J. Woody Ju b,c
a
School of Civil Engineering and Mechanics, Huazhong University of Science and Technology, Wuhan 430074, China
b
School of Civil Engineering, Tongji University, Shanghai 200092, China
c
Department of Civil and Environmental Engineering, University of California, Los Angeles, CA 90095, USA

a r t i c l e i n f o a b s t r a c t

Article history: This study aims to develop a versatile constitutive framework for concrete that can be used in various
Received 5 October 2015 stress states and under reversed cyclic loading. A multi-axial compression stress state is decomposed into
Revised 20 March 2016
a hydrostatic confining part and a biaxial net part to distinguish concrete behavior in triaxial and biaxial
Available online 6 May 2016
compression while preserving their similarity. This novel decomposition is combined with the positive–
Keywords: negative decomposition of stress, and on its basis is developed a 3D elastoplastic damage model. The
Stress decomposition suppression on damage of the net part of stress from the presence of the confining part is introduced
Confining effect through an increase in microscopic fracture strains, and confining effects in ductility and lateral defor-
Unilateral effect mation are introduced, respectively, by hardening slowdown and dilation reduction. In the meantime,
Damage independent damage/plasticity evolution in tension and compression is incorporated in the proposed 3D
Plasticity model with two damage variables and two hardening variables. An elastic–plastic–damage operator split
Concrete
integration algorithm with the backward Euler return mapping is derived, and a series of numerical sim-
ulations are carried out in comparison with tests under uniaxial, biaxial, pseudo-triaxial, and true triaxial
loading with satisfactory agreement.
© 2016 Published by Elsevier Ltd.

1. Introduction pure plasticity and fail to describe stiffness degradation (Imran and
Pantazopoulou, 2001; Grassl et al., 2002; Papanikolaou and Kap-
Nonlinear analysis of concrete structures has drawn broad at- pos, 2007; Carrazedo et al., 2013; Chi et al., 2014), which are also
tention and extensive efforts in the civil engineering community dedicated to multiaxial compression and are not intended for use
for half a century, in which constitutive modeling remains a criti- in tension; some are pure damage mechanics and cannot represent
cal issue owing to the diverse and complex mechanical behavior of irrecoverable deformation (Mazars et al., 2014); some employ dam-
concrete materials under various loading scenarios. age mechanics for tension and plasticity for compression but not
Research on constitutive relations is of everlasting interest in simultaneously (Cervenka and Papanikolaou, 2008; Sánchez et al.,
solid mechanics. As for concrete, different types of theories have 2012); and some really combine plasticity and damage simulta-
been proposed, among which plasticity models (e.g. Lubliner et al., neously and take tension into account, but do not reflect the in-
1989; Lee and Fenves, 1998; Grassl et al., 2002; Papanikolaou and dependent hardening (Luccioni and Rougier, 2005; Grassl and Ji-
Kappos, 2007; Carrazedo et al., 2013) and damage models (e.g. rasek, 2006; Jason et al., 2006; Caner and Bažant, 2013a; Grassl
Ju 1989; Faria et al., 1998; Wu et al., 2006; Cicekli et al., 2007; et al., 2013; Valentini and Hofstetter, 2013) or damage (Luccioni
Abu Al-Rub and Kim, 2010; Grassl et al., 2013; Mazars et al., 2014; and Rougier, 2005; Grassl and Jirasek, 2006; Jason et al., 2006;
Zhang and Li, 2014a) are the most fully developed. Valentini and Hofstetter, 2013) in tension and compression (under
While plasticity/damage models have achieved more and more multiaxial tension–compression or reversed cyclic loading), among
success in their characterization of concrete behavior under biaxial which the powerful microplane models (Caner and Bažant, 2013a)
loading (and triaxial loading with tensile stress components), the are formulated in terms of stress and strain vectors rather than
situations are much more difficult for nonlinear analysis involving tensors, leading to much greater computational cost.
triaxial compression. There exist plasticity and/or damage models In particular, Caner and Bažant (2013a) and Grassl et al.
developed with special attention to triaxial compression: some are (2013) possess a wide scope of applicability since not only are tri-
axial confining effects successfully characterized, but the stiffness

recovery during transition between tension and compression can
Corresponding author. Tel.: +86 21 65983526; fax: +86 21 65983944.
E-mail addresses: zhang_ji@hust.edu.cn (J. Zhang), lijie@tongji.edu.cn (J. Li),
be captured with independent formulation for tensile and com-
juj@ucla.edu (J.W. Ju). pressive damage. In these theories, however, the independency

http://dx.doi.org/10.1016/j.ijsolstr.2016.04.038
0020-7683/© 2016 Published by Elsevier Ltd.
126 J. Zhang et al. / International Journal of Solids and Structures 94–95 (2016) 125–137

between tensile and compressive hardening is not included, which,


we believe, is also necessary for a realistic characterization of uni-
lateral effects and situations of mixed tensile–compressive stress
states.
It is the purpose of the present study to develop (within the
classical tensor-based context) a more versatile constitutive frame-
work that is able to meet the manifold demands in nonlinear anal-
ysis of concrete structures. It is intended that the distinct behavior
of concrete in tension, uni-/bi-axial compression, and triaxial com-
pression (except for hydrostatic compression) is treated appropri-
ately, which is required for various stress states, and at the same
time the independent plasticity/damage evolution in tension and
compression is allowed, which is important for multiaxial tension–
compression (shear and torsion) as well as reversed cyclic loading
(seismic analysis).
Fig. 1. Stress decomposition in terms of principal values. (a) σ̄1 ≥ 0 (Ladeveze,
2. Stress decomposition 1983). (b) σ̄1 < 0 .

The positive–negative decomposition of stress (Ladeveze, 1983)


peak value around σ 2 : σ 3 = −0.5 : −1, then decreases, and re-
has proved effective in distinguishing tension and compression be-
mains still higher than the starting point (σ 1 = σ 2 ) until σ 2 :
havior (e.g. Wu et al., 2006; Cicekli et al., 2007; Grassl et al., 2013).
σ 3 = −1 : −1. This characteristic of concrete in triaxial compres-
Traditionally, the spherical–deviatoric decomposition is performed
sion can be found directly from some targeted experimental study
upon a symmetric second-order tensor, and frequently used for
(e.g. Shang and Ji, 2014), and indirectly from the geometric prop-
stress and/or strain in constitutive modeling. However, a spheri-
erties of failure surfaces based on early test data.
cal part of stress is present also in stress states other than triaxial
Based on this similarity, the multi-axial compression stress − σ̄
compression and, therefore, is not a good indicator of triaxial con-
is further decomposed into a hydrostatic confining part σ̄ ≡ and a
fining effects. This study attempts to decompose stress in a new
biaxial net part σ̄ − as
way and, by doing so, distinguish triaxial and biaxial compression
behavior. −
σ̄ = σ̄ ≡ + σ̄ − , (4)
where
2.1. Positive–negative decomposition

To describe the distinct damage behavior of concrete in tension (5)


and compression, the nominal stress σ / effective stress σ̄ is de-
composed into its positive part + σ /+ σ̄ and negative part − σ /− σ̄ ,
with I being the second-order identity tensor, and
σ = +σ + −σ, σ̄ = + σ̄ + − σ̄ , (1)
as first advocated by Ladeveze (1983) and Ortiz (1985), and nowa- (6)
days inherited in a lot of successful damage models for concrete
(e.g. Wu et al., 2006; Cicekli et al., 2007; Abu Al-Rub and Kim, This confining–net decomposition is illustrated in Fig. 1(b) with
2010; Grassl et al., 2013; Zhang and Li, 2014a). the principal stresses, where the dashed area represents the con-
The meaning of this popular positive–negative decomposition fining part, and the blank area the net part.
can be clearly shown from the dyadic forms of these stress tensors Combining Eqs. (1) and (4) gives complete decomposition of
in the principal coordinate system, stress
+
σ̄ + σ̄ − , σ̄1 > 0
σ̄ = σ̄ − , σ̄1 = 0 . (7)
σ̄ ≡ + σ̄ − , σ̄1 < 0
(2)
With the definitions in Eqs. (2)b, (5), and (6) and + σ̄ = σ̄ + ,
where σ̄i are the principal effective stresses, the orthonormal ba- Eq. (7) can be rephrased in a unified form
sis along the principal directions, and the Mcauley brackets  are σ̄ = σ̄ + + σ̄ − + σ̄ ≡ . (8)
defined as
In the present decomposition, σ̄ +
stands for a uni-/bi-/tri-axial
1
x = (|x| + x ). (3) tension state, σ̄ − a uni-/bi-axial compression state, and σ̄ ≡ an
2 equi-triaxial compression state.
This positive–negative decomposition is illustrated in Fig. 1(a)
with the principal stresses, where the dotted area represents the 3. Damage framework
positive part, and the blank area the negative part.
3.1. Damage variables
2.2. Confining–net decomposition
On the basis of the positive–negative decomposition in Eq. (1),
In order to further describe the distinct behavior in triaxial the two-parameter damage expression by Faria et al. (1998)
compression, the authors have been seeking a proper way to make
further decomposition, and are inspired by an interesting similar-
σ = (1 − d+ )+ σ̄ + (1 − d− )− σ̄ , (9)
ity between concrete behavior in biaxial and triaxial compression: where d± is the tensile/compressive damage variable, provides a
with the magnitude of the intermediate principal stress |σ 2 | in- simple and effective platform for the construction of damage mod-
creasing, the multiaxial strength |σ 3 | first increases, reaching its els (e.g. Wu et al., 2006; Grassl et al., 2013; Zhang and Li, 2014a).
J. Zhang et al. / International Journal of Solids and Structures 94–95 (2016) 125–137 127

With the present decomposition in Eq. (8), it is natural to for- 3.4. Damage evolution laws
mulate a three-parameter damage expression as
Damage is driven by the thresholds as reflected in the ten-
σ = (1 − d+ )σ̄ + + (1 − d− )σ̄ − + (1 − d≡ )σ̄ ≡ , (10) sile/compressive damage evolution function
where d≡ is the damage variable in equi-triaxial compression. d ± = G± ( r ± ), (18)
Since hydrostatic compression tests of concrete (e.g. Green and
which is expected to meet the requirements for initial/final value
Swanson, 1973; Gabet et al., 2008) indicate that stiffness degrada-
and monotonicity
tion hardly occurs (despite considerable plastic deformation) un-
til the stress level reaches a very high value, d≡ is neglected (i.e. dG± ( r ± )
G± ( 0 ) = 0, lim G± (r ± ) = 1, ≥ 0. (19)
σ ≡ = σ̄ ≡ ) here and Eq. (10) is simplified to r ± →∞ dr ±

σ = (1 − d+ )σ̄ + + (1 − d− )σ̄ − + σ̄ ≡ . (11) 3.5. Loading/unloading conditions

Loading or unloading is expressed by the Karush–Kuhn–Tucker


3.2. Damage energy release rates conditions
g± ≤ 0, d˙± ≥ 0, d˙± g± = 0. (20)
Within the traditional thermodynamic framework, the ten-
sile/compressive damage energy release rate Y± is introduced as Moreover, the consistency condition
the conjugate force of d± with respect to the Helmholz free
d˙± g˙ ± = 0 (21)
energy ψ :
is always required.
± ∂ψ
Y = − ±. (12)
∂d 4. Damage evolution
The free energy ψ of concrete consists of an elastic part and ψe
While the expression of a damage state (Eq. (11)) treats a
a plastic part ψ p (Ju, 1989; Zhang and Li, 2014b): ψ e is the macro-
damaged material as a continuum, the evolution laws of damage
scopic (elastic) strain energy, and ψ p can be interpreted as the mi-
(Eq. (18)) are rooted in the behavior of individual microcracks.
croscopic additional strain energy locked by frictional microcracks.
Because the formulation or simulation of a discontinuum is ex-
Because tensile plastic deformation of concrete comes mainly
tremely complex and difficult, in this section a simplified microme-
from the microscopic stress-free opening rather than the frictional
chanical formulation of damage evolution by Li and Ren (2009) is
sliding and roughness-induced opening of microcracks, the plastic
used with an extension to triaxial compression in accordance with
free energy ψ p can be neglected in spite of considerable irrecov-
the present decomposition of stress (Eq. (8)).
erable deformation in tension. Hence, the tensile damage energy
release rate is simply 4.1. Microscopic parallel system
1 +
Y+ = σ̄ : εe , (13) Three kinds of basic stress states are selected to represent the
2
damage evolution of the three parts from the stress decomposition
where ɛe is the elastic strain. (Eq. (8)), respectively: the uniaxial tension σ + for the tension part
In the compression case, it is still a challenge to formulate σ̄ + , the uniaxial compression σ − for the net compression part σ̄ − ,
elastoplastic free energy from micromechanics. Here the macro- and the equi-triaxial compression σ ≡ for the confining part σ̄ ≡ .
scopic approximation of the compressive damage energy release For each basic stress state, a parallel system (bundle) of mi-
rate by Wu et al. (2006) crosprings (fibers) is built up to formulate damage evolution as ex-
  2 hibited in Fig. 2, where ν is the Poisson’s ratio and ε +,e , ε −,e , ɛ≡, e
C the elastic strains in the directions of σ + , σ − , σ ≡ , respectively. It
Y− ≈ α I¯1− + 3J¯2− (14)
Ē is assumed that all the fibers in a bundle have the same mod-
ulus of elasticity but random fracture strains l as displayed in
is adopted, where C is a dimensionless material constant, Ē the ef- Fig. 3, where σ l and Al are the stress and cross-sectional area of
fective (initial) Young’s modulus, I¯1− the first invariant of σ̄ − , J¯2− the the lth fiber, respectively. Since the damage variable d≡ for σ̄ ≡ in
second invariant of the deviatoric part of σ̄ − , and Eq. (10) has been neglected in the present study, no fracture occurs
in Fig. 2c, i.e. ≡ = ∞.
fbc / fc − 1 l
α= (15)
2 fbc / fc − 1 Remark 4.1. The rupture of fibers in Fig. 2a and c represents the
mode I fracture, mode II fracture, and no fracture of microcracks,
with fc being the uniaxial compressive strength and fbc being the respectively, in σ + , σ − , and σ ≡ .
equi-biaxial compressive strength.
4.2. Basic stress states
3.3. Damage criteria
According to the classical definition of damage in terms
of cross-sectional area, the tensile/compressive damage variable
The maximum of Y± throughout the loading history τ is de-
(Fig. 2a/b) can be calculated as
fined as the tensile/compressive damage threshold
  ±,e 
Al H ε − ±l
r = max Y (τ ),
± ± ± l
(16) d =  , (22)
τ ∈[0,t ] l Al

where t is the current instant. Y± and r± build up the ten- where H( ) is the Heaviside step function. As the fibers become
sile/compressive damage criterion infinitely thin, i.e. max {Al } → 0, the limit of Eq. (22) is
1
g± (Y ± , r ± ) = Y ± − r ± = 0. (17) d± = H[ε ±,e − ± (x )]dx, (23)
0
128 J. Zhang et al. / International Journal of Solids and Structures 94–95 (2016) 125–137

Fig. 2. Microspring bundles. (a) Uniaxial tension. (b) Uniaxial compression. (c) Equi-triaxial compression.

Fig. 3. Microscopic and macroscopic stress–strain curves.

where x is the normalized spatial coordinate of a fiber within a where


() and
( , ; r) are the 1D and 2D cumulative distribution
bundle. functions of the standard normal distribution, respectively,
± (x) constitute a microscopic random field, consequently d± μ± = E[ln ± ], (28)
determined from Eq. (23) is a random variable, whose mean value
and variance can be derived (Li and Ren, 2009) under the assump- and

tion of homogeneity (stationarity) to be V± = D[ln ± ]. (29)


E [d ] = F ( ε
± ± ±,e
) (24) The stochastic damage evolution is formulated above, neverthe-
less only the mean values (Eq. (26)) are to be used in the following
and
sections as available data from triaxial tests are very limited.
1
D[d± ] = 2 (1 − r )F ± (ε ±,e , ε ±,e ; r )dr − [F ± (ε ±,e )]2 (25) Remark 4.2. The formulation of uniaxial damage evolution can be
0
connected to the fracture energy Gf by
respectively, where F ± () = F ± (, x ) is the 1D cumulative distribu- +∞
Gf
tion function of ± (x), r = |x1 − x2 |, and F ± (, ; r ) = F ± (, x1 ; , x2 ) σ + dε + = , (30)
lch
the 2D cumulative distribution function of ± (x). 0

Assume that the microscopic fracture strains ± (x) are log- where lch is the characteristic length of an element, which is intro-
normally distributed similar to the fracture strength of concrete duced to achieve mesh objectivity (Oliver et al., 1990).
(Bazant and Becq-Giraudon, 2002), and then the general expres-
sions in Eqs. (24) and (25) can be specifically calculated as 4.3. Superimposed basic stress states
 ±,e 
ln ε − μ± The damage evolution in uniaxial tension (Fig. 2a) and that in
E[d ] =

±
±
(26)
V uniaxial compression (Fig. 2b) are tentatively taken to be indepen-
dent from each other for convenience as treated in many other
and theories (e.g. Wu et al., 2006; Cicekli et al., 2007; Caner and Ba-
 ±,e 
1
ln ε − μ± ln ε ±,e − μ± žant, 2013a; Grassl et al., 2013).
D[d± ] = 2 ( 1 − r )
, ; r dr The presence of the third kind of basic stress state, equi-triaxial
0 V± V±
 ±,e  compression (Fig. 2c), suppresses the damage evolution in uniaxial
ln ε − μ± compression (Fig. 2b). To account for this interaction, the super-

2 ±
, (27)
V position of uniaxial compression and equi-triaxial compression is
J. Zhang et al. / International Journal of Solids and Structures 94–95 (2016) 125–137 129

considered, i.e. the pseudo-triaxial compression in the compressive Replacing ε −,e in Eqs. (32) and (33) with ε˜−,e (r − ) leads to
meridional plane (σ̄1 = σ̄2 ). 1
In the present bundle model, this suppression on damage is re- d − = G− ( r − ) = H[ε˜−,e (r − ) − −
≡ (x )]dx (42)
flected through an increase in the microscopic fracture strains un- 0

der confinement and


   
−σ̄1  ln ε˜−,e (r − ) − μ−
−≡ (x ) = − (x ) exp c , (31) E[d− ] =

. (43)
fc V≡−
where − ≡ (x ) are the fracture strains of fibers in the superimposed
5. Plasticity
σ − (Fig. 2b) and σ ≡ (Fig. 2c), and c the coefficient of fracture
enhancement.
For infinitesimal elastoplastic deformation, the strain ε is split
Replacing − (x ) in Eqs. (23)b, (28)b, and (29)b with − ≡ (x )
additively into an elastic part ɛe and a plastic part ɛp as
leads to
1 ε = εe + εp . (44)

d = H ε −,e − −
≡ ( x ) dx (32)
0 The hypothesis of elastic strain equivalence is taken here:
and
 −,e  σ = C : εe , σ̄ = C̄ : εe , (45)
ln ε − μ−

E[d ] =



, (33) where C and C̄ are the nominal and effective elastic stiffness, re-
V≡ spectively. From Eqs. (44) and (45)b, one gets the elastoplastic re-
lation in the effective stress space
where
−σ̄1  σ̄ = C̄:(ε − εp ). (46)
μ−≡ = E ln −≡ = μ− + c , (34)
fc In isotropic elasticity, Ē and ν are all that are needed to deter-
and mine C̄ and construct a linear elastic relation. A plastic relation

comprises several intertwined components and becomes rather
V≡− = D [ln − −
≡] = V . (35) complex when multiple hardening is introduced. Here the rel-
Remark 4.3. From a micromechanical view, this interaction be- atively simple Lubliner–Lee plasticity (Lubliner et al., 1989; Lee
tween the net part σ̄ − and confining part σ̄ ≡ comes from the sup- and Fenves, 1998) with tension–compression double hardening is
pression on shear sliding between crack surfaces (damage) by the adopted with some enrichment for triaxial compression.
normal pressure on them (confinement).
5.1. Yield criterion
4.4. General stress states
The yield function assumes the form

With the concepts of energy-driven thresholds (Eq. (16)) and


f p (σ̄ , κ +, κ − ) = 3J¯2 + α I¯1 + β (κ + , κ − )σ̄1 
thresholds-driven damage (Eq. (18)), one can extend the above
damage evolution functions to general stress states. − (1 − α ) f¯− (κ − ), (47)
In uniaxial tension, Eq. (13) is calculated to be where κ±
is the tensile/compressive hardening variable, I¯1 the first
1 invariant of σ̄ , J¯2 the second invariant of the deviatoric part of σ̄ ,
Y+ = Ē (ε +,e )2 . (36) and
2
From Eqs. (16)a and (35), an equivalent tensile elastic strain can f¯− (κ − )
be determined as
β (κ + , κ − ) = (1 − α ) − (1 + α ) (48)
f¯+ (κ + )

2r + with f¯± (κ ± ) being the subsequent effective yield stress (cohesion)
ε˜+,e (r+ ) = , (37)
Ē in uniaxial tension/compression.
which takes the place of ε +,e (Fig. 2a) in multiaxial tension. Remark 5.1. Since the yield surface described by Eq. (47), which
Replacing ε +,e in Eqs. (23)a and (26)a with ε˜+,e (r + ) leads to consists of two cones (Zhang et al., 2010; Zhang and Li, 2012), is
1 open in the negative direction of the hydrostatic axis, the present
d + = G+ ( r + ) = H[ε˜+,e (r + ) − + (x )]dx (38) 3D model cannot characterize concrete behavior in hydrostatic
0
compression.
and
 
ln ε˜+,e (r + ) − μ+ 5.2. Flow rule
E[d+ ] =
. (39)
V+
The non-associated flow rule is adopted to characterize the
In uniaxial compression, Eq. (14) is calculated to be plastic dilation of concrete
Y − = C (1 − α )2 Ē (ε −,e )2 . (40) ∂ gp
ε˙ p = λ˙ , (49)
From Eqs. (16)b and (39), an equivalent compressive elastic
∂ σ̄
strain can be determined as where λ˙ is the plastic consistency parameter, and gp the plastic
potential function. The simple Drucker–Prager type function
1 r−

ε˜−,e (r− ) = , (41)


1−α C Ē gp (σ̄ ) = 2J¯2 + η0 I¯1 , (50)

which takes the place of ε −,e (Fig. 2b) in general multiaxial com- where η0 is the parameter controlling dilation, is sufficient for bi-
pression. axial loading. To reflect the decrease in plastic dilation of concrete
130 J. Zhang et al. / International Journal of Solids and Structures 94–95 (2016) 125–137

in triaxial compression, η0 in Eq. (49) is reduced in the present


formulation, leading to

gp (σ̄ ) = 2J¯2 + ηI¯1 , (51)

where
 
−σ̄1 
η = η0 exp −cη (52)
fc

with cη being the coefficient of dilation reduction.

5.3. Hardening laws

The Ludwik power law is employed to formulate the strain


hardening as follows
±
f¯± (κ ± ) = f¯y± + K̄ ± (κ ± )n , (53)

where f¯y±
is the initial effective yield stress in uniaxial ten-
sion/compression, K̄ ± the effective tensile/compressive strength in-
dex, and n± the tensile/compressive hardening exponent.

5.4. Hardening variables

The rate evolution equations for the hardening variables are


originally

κ˙ + = wε˙ 1p , κ˙ − = −(1 − w )ε˙ 3p , (54)

where ε1p
and ε3p
are the major and minor principal plastic strains,
respectively, and the weight factor

σ̄i 
w = i . (55)
i |σ̄i |
In the present model, the increase in ductility of concrete in
triaxial compression is reflected by slowing down the evolution in
Eq. (54)b as
 
−σ̄1 
κ˙ = wε
+
˙ 1p , κ˙ = −(1 − w )ε exp −cκ

˙ 3p , (56)
fc

where cκ is the coefficient of hardening slowdown.


With the flow rule (49), Eq. (56) can be rephrased with λ˙ as

κ˙ ± = λ˙ h± , (57)
Fig. 4. Evolution of yield surface. (a) Tensile hardening. (b) Compressive hardening.
where the tensile and compressive hardening functions are, respec-
tively,
  It is noticed from Fig. 4b that the biaxial tensile strength de-
+ ∂ gp − ∂ gp −σ̄1  creases with compressive hardening. Detailed discussion on this is-
h =w , h = − (1 − w ) exp −cκ . (58)
∂ σ̄1 ∂ σ̄3 fc sue can be found in Zhang and Li (2012). For multiaxial tension,
tensile hardening is not independent in a strict sense from com-
5.5. Loading/unloading conditions pressive hardening.

Remark 5.3. From Eqs. (5), (31), (52), and (56), it can be seen
Finally with the Karush–Kuhn–Tucker loading/unloading condi-
that the present decomposition of stress uses −σ1 , the mini-
tions
mum principal compressive stress, to evaluate triaxial confinement,
f p ≤ 0, λ˙ ≥ 0, λ˙ f p = 0 (59) in contrast with the traditional spherical–deviatoric decomposi-
tion, which employs I1 . In the critical case of biaxial compression
and the consistency condition (σ1 = 0), there is no triaxial confinement at all according to the
λ˙ f˙ p = 0, (60) present theory.

the plastic relation is complete.


6. Algorithm
Remark 5.2. It is indicated by Eqs. (47), (48) and (53) that the size
and shape of the yield surface is controlled by the independent 6.1. Operator split
evolution of the two cohesions f¯+ (κ + ) and f¯− (κ − ). The expansion
and distortion of the yield surface during tensile/compressive hard- In accordance with the concept and methods of product for-
ening are illustrated in Fig. 4a/b, where f¯± (κ ± ) increases while mulas and operator split (Chorin et al., 1978; Ju, 1989; Simo
f¯∓ (κ + )= f¯y∓ remains unchanged. and Hughes, 1998), the constitutive equations presented in the
J. Zhang et al. / International Journal of Solids and Structures 94–95 (2016) 125–137 131

Box 1
Split of rate constitutive equations.

Total = Elastic predictor + Plastic corrector + Damage corrector

  2 
previous sections are replaced here with the elastic, plastic, and
∂ d+ 1 1 ln ε˜+e − μ+
damage parts in Box 1, where is the displacement. It can serve = √ exp − (66)
∂ r + 2 2π V + r + 2 V+
as a mathematical foundation for developing first-order accurate
algorithms. and
These three parts are taken as split problems in a series, which ⎧    2 ⎫
⎨ ln ε˜−e − μ− + c −fσ̄c 1  ⎬
add up to the original elastoplastic damage problem. The elas- ∂d −
1 1
= √ exp − ,
tic predictor problem and damage corrector problem are straight- ∂ r − 2 2π V − r − ⎩ 2 V− ⎭
forward and involve no iteration. For the plastic corrector prob-
lem, the spectral version (Lee and Fenves, 20 02; Wu, 20 04) of (67)
the closest-point projection return mapping algorithm (Simo and
Hughes, 1998) is employed. and for ∂ r± /∂ σ̄ one can use Eqs. (13)/(14) and (16) to get

∂ r+ 1 + −1 −1
= (σ̄ : C̄ + P+ : C̄ : σ̄ ) (68)
6.2. Consistent tangent stiffness ∂ σ̄ 2
and
The differentiation with respect to the strain increment εn+1
of the resulting σ̄n+1 and σn+1 give, respectively, the tangent stiff-
  


∂ r− 2C − − 3 s
ness consistent with the updating algorithms in elastic–plastic = α I 1 + 3J 2 αI + . (69)
∂σ E 2  s− 
steps and elastic–plastic–damage steps,

∂ σ̄n+1 ∂ σn+1 The differentiation of the three parts from stress decomposition
C̄n+1 = , Cn+1 = , (61) in Eq. (64) generates three fourth-order tensors: ∂ σ̄ ≡ /∂ σ̄ can be
∂ εn+1 ∂ εn+1
obtained from Eq. (5) to be
which lead to
∂ σ̄ ≡
∂ σn+1 P≡ = = H (−σ̄1 )I  N11 ; (70)
Cn+1 = : C̄n+1 . (62) ∂ σ̄ ]
∂ σ̄n+1
∂ σ̄ + /∂ σ̄ is derived in Faria et al. (20 0 0) as
In Eq. (62) C̄n+1 is already given as (Wu, 2004)
 −1 ∂ σ̄ +   σ̄i  − σ̄ j 
∂ gp ∂ ( λ ) ∂ 2 gp P+ = = H (σ̄i )Nii  Nii + 2 Ni j  Ni j ; (71)
C̄n+1 = C̄
−1
+  + λ (63)
∂ σ̄ i
]
i< j
σ̄i − σ̄ j
∂ σ̄n+1 ∂ σ̄n+1 ∂ σ̄n2+1
and the remaining ∂ σ̄ − /∂ σ̄ is
from conventional elastic–plastic algorithm, and ∂ σn+1 /∂ σ̄n+1
needs to be derived in accordance with the present damage for- ∂ σ̄ −
P− =
mulation. ∂ σ̄  
From Eq. (11) one can obtain   −σ̄ j  − −σ̄i 
= H (σ̄1 ) H (−σ̄i )Nii  Nii + 2 Ni j  Ni j
∂σ ∂ σ̄ + ∂ σ̄ − ∂ σ̄ ≡ ∂ d+ σ̄i − σ̄ j
− σ̄ + 
[ [
= (1 − d + ) + (1 − d − ) + i i< j
∂ σ̄ ∂ σ̄ ∂ σ̄ ∂ σ̄ ∂ σ̄
+ H (−σ̄1 )(I − I  N11 ), (72)
∂ d −
− σ̄ −  . (64) ]
∂ σ̄
In Eq. (64) ∂ d± /∂ σ̄ is broken down into where Ni j = I is the fourth order symmetric iden-
∂ d± ∂ d± ∂ r± tity tensor, and the left-/right-continuous Heaviside step function
= , (65) is defined as
∂ σ̄ ∂ r± ∂ σ̄  
where ∂ d± /∂ r ± can be acquired from Eqs. (39)/(43) and (37)/(40) 1, x>0 1, x≥0
H (x ) = 0, x≤0
, H (x ) = 0, x<0
. (73)
as ] [
132 J. Zhang et al. / International Journal of Solids and Structures 94–95 (2016) 125–137

Further, they constitute a partition of unity: Remark 7.1. Although plastic deformation of concrete in tension
is much smaller than that in compression, it is still quite consid-
I = P+ + P− + P≡ . (74)
erable compared to elastic deformation in tension. The reason to
Remark 6.1. Wu and Xu (2013) demonstrate that the derivative neglect tensile plastic free energy is that tensile plastic deforma-
expressed by Eq. (70) also serves as the thermodynamically con- tion does not involve much frictional sliding and energy locking as
sistent projection operator such that discussed before Eq. (13). Plastic flow in tension generated by the
present model is attributed to microcrack opening without promi-
σ̄ + = P+ : σ̄ . (75) nent frictional effects.
Apparently, the derivative expressed by Eq. (69) satisfies
7.1.2. Cyclic compression
σ̄ = P≡ : σ̄ .

(76) For cyclic uniaxial compression, the test by Karsan and Jirsa
And with Eq. (73), one has (1969) is simulated. The material parameters are calibrated as
listed in Table 3. The calculated stress–strain curve is shown along
σ̄ = P− : σ̄ .

(77) with the test data in Fig. 6.
Substituting Eqs. (74)–(76) into Eq. (10), comparing it to the
common representation of σ̄ with the damage effect tensor M 7.1.3. Reversed cyclic loading
For reversed cyclic uniaxial loading, appropriate test results are
σ̄ = M : σ , (78) not available. The material parameters listed in Table 4 in the next
and using Eq. (73), one gets section are used in this simulation, which are calibrated from the
tests by Kupfer et al. (1969). The calculated stress–strain curve is
M = (I − d+ P+ − d− P− − d≡ P≡ )−1 . (79) shown in Fig. 7.
Since d≡ is neglected in the present study, Eq. (78) is simplified In comparison with those successful 3D models (Caner and Ba-
to žant, 2013b; Grassl et al., 2013), which employ single hardening
plasticity, the present model (with tension–compression double
M = (I − d+ P+ − d− P− )−1 . (80) hardening) appears to produce more realistic behavior for concrete
under reversed cyclic loading. The performance of double harden-
7. Validations ing will be further examined later in Fig. 10 for biaxial loading.
It is seen that most of the basic characteristics of concrete un-
To validate the present constitutive theory, a series of typical der uniaxial loading, such as hardening, softening, stiffness degra-
mechanical tests of concrete specimens are simulated. For all these dation, irrecoverable deformation, and unilateral effects, can be
simulations, some of the material parameters are set fixed as listed represented appropriately with the proposed model. It is also
in Table 1, where ft is the uniaxial tensile strength. noted that the hysteresis between unloading and reloading fails
In the following simulations of uniaxial loading, the uniaxial to be characterized, which can be included through introduction
material parameters are just calibrated to reproduce the experi- of microsliders into the microspring systems in Fig. 2, leading to
mental results. As for multiaxial loading, part of the multiaxial test much more complex formulation.
data are set aside to calibrate the multiaxial material parameters,
which, together with the already calibrated uniaxial material pa- 7.2. Biaxial loading
rameters, are then used to blindly predict the other multiaxial test
data. For proportional biaxial loading, the tests by Kupfer et al.
(1969) are simulated. First the uniaxial material parameters Ē , ft , fc ,
7.1. Uniaxial loading

Table 3
7.1.1. Cyclic tension
Parameters for cyclic compression simulation.
For cyclic uniaxial tension, the test by Geopalaeratnam and
Shah (1985) is simulated. The material parameters are calibrated as Effective Young’s modulus Ē 34.7 × 103 N/mm2
Uniaxial compressive strength fc 27.4 N/mm2
listed in Table 2. The calculated stress–strain curve is shown along
Mean value of natural logarithm of μ− 7.33
with the test data in Fig. 5. compressive fracture strain
Standard deviation of natural logarithm V− 0.53
of compressive fracture strain
Table 1 Effective compressive strength index K− 0.045 Ē
Parameters for all simulations.

Poisson’s ratio ν 0.2 Table 4


Dilation parameter η0 0.3 Parameters for biaxial loading simulations.
+
Tensile hardening exponent n 0.5
Compressive hardening exponent n− 0.5 Effective Young’s modulus Ē 31.0 × 103 N/mm2
Initial effective yield stress in uniaxial tension f¯y+ ft Uniaxial tensile strength ft 2.89 N/mm2
Initial effective yield stress in uniaxial compression f¯− 0.5fc Uniaxial compressive strength fc 32.1 N/mm2
y
Mean value of natural logarithm of μ+ 4.75
tensile fracture strain
Table 2 Mean value of natural logarithm of μ− 7.56
Parameters for cyclic tension simulation. compressive fracture strain
Standard deviation of natural logarithm V+ 0.1
Effective Young’s modulus Ē 31.9 × 103 N/mm2
of tensile fracture strain
Uniaxial tensile strength ft 3.48 N/mm2 −
Standard deviation of natural logarithm V 0.5
Mean value of natural logarithm of tensile μ+ 4.85
of compressive fracture strain
fracture strain
Effective tensile strength index K+ 0.003 Ē
Standard deviation of natural logarithm of V+ 0.07
Effective compressive strength index K− 0.045 Ē
tensile fracture strain
Effective tensile strength index K+ 0.002 Ē Equi-biaxial compressive strength fbc 1.16fc
J. Zhang et al. / International Journal of Solids and Structures 94–95 (2016) 125–137 133

Fig. 5. Cyclic tension (Geopalaeratnam and Shah, 1985).

Fig. 6. Cyclic compression (Karsan and Jirsa, 1969).

μ+ , μ− , V + , V − , K + , and K − are calibrated to fit the experimental the x axis; and then the displacement in this direction is kept con-
uniaxial stress–strain curves; and then they are used to predict the stant (εx ≡ 0.4 × 10−3 ), and the specimen is subjected to compres-
biaxial behavior with fbc calibrated from the peak stresses. These sion along the y axis. The material parameters listed in Table 4 are
parameters are collected in Table 4. used. The calculated stress–strain curves are displayed in Fig. 10. A
simulation of free uniaxial compression (without lateral restraint)
7.2.1. Tension–compression is also presented in Fig. 10b as a reference. Figs. 7 and 10 demon-
The calculated stress–strain curves in biaxial tension– strate the ability of the present model to formulate independent
compression are displayed along with the test data in plasticity/damage evolution in tension and compression.
Fig. 8. It is observed that the variation in strength and ductility, the
lateral deformation, and the independent tensile and compressive
7.2.2. Compression–compression plasticity/damage evolution of concrete under biaxial loading can
The calculated stress–strain curves in biaxial compression are be represented appropriately with the proposed model.
displayed along with the test data in Fig. 9.
7.3. Triaxial loading
7.2.3. Successive tension and compression
For compression that follows preceding tensile failure in a per- 7.3.1. Pseudo-triaxial compression
pendicular direction, a special loading scenario is simulated: first For pseudo-triaxial compression, the tests by Candappa et al.
a specimen is loaded in tension to almost complete failure along (2001) are simulated. First the uniaxial material parameters Ē ,
134 J. Zhang et al. / International Journal of Solids and Structures 94–95 (2016) 125–137

Fig. 7. Reversed cyclic loading.

Fig. 8. Biaxial tension–compression (Kupfer et al., 1969).

fc , μ− , V − , and K − are calibrated to fit the experimental uniax- calculated stress–strain curves are exhibited along with the test
ial stress–strain curves, the triaxial material parameters c and data in Fig. 11.
cκ to fit the longitudinal stress–strain (σ 3 –ε 3 ) curve, and cη to This set of numerical examples demonstrate the ability of the
fit the longitudinal stress–lateral strain (σ 3 –ε 1 ) curve under the proposed model to reflect three basic confining effects: increase in
confining pressure σ 1 = σ 2 = −4 N/mm2 ; and then all these cali- strength, increase in ductility, and decrease in dilation.
brated parameters are used to predict the triaxial behavior under
σ 1 = σ 2 = −8 and −12 N/mm2 . They are collected in Table 5. The

Table 5
7.3.2. True triaxial compression
Parameters for pseudo-triaxial compression simulations. For true triaxial compression, the tests by van Mier (1984) are
simulated. First the uniaxial material parameters Ē , fc , μ− , V − , and
Effective Young’s modulus Ē 50.9 × 103 N/mm2
Uniaxial compressive strength fc 60.6 N/mm2
K − are set to produce the uniaxial stress–strain curve, and the tri-
Mean value of natural logarithm of μ− 7.65 axial material parameters c , cκ , and cη are calibrated to fit the
compressive fracture strain stress–strain curves in the loading path σ 1 : σ 2 : σ 3 = −0.05 :
Standard deviation of natural logarithm of V− 0.4 −0.1 : −1; and then all these calibrated parameters are used to
compressive fracture strain
predict the triaxial behavior in σ 1 : σ 2 : σ 3 = −0.05 : −0.33 : −1
Effective compressive strength index K− 0.035 Ē
(the cylindrical strength of the specimen is a bit lower than that
Coefficient of fracture enhancement c 3 of the one for calibration). They are collected in Table 6. The cal-
Coefficient of hardening slowdown cκ 5
Coefficient of dilation reduction cη 10.5
culated stress–strain curves are exhibited along with the test data
in Fig. 12.
J. Zhang et al. / International Journal of Solids and Structures 94–95 (2016) 125–137 135

Fig. 9. Biaxial compression (Kupfer et al., 1969).

Fig. 10. Longitudinal compression after transversal tension. (a) Transversal stress. (b) Longitudinal stress.
136 J. Zhang et al. / International Journal of Solids and Structures 94–95 (2016) 125–137

Fig. 11. Pseudo-triaxial compression (Candappa et al., 2001).

Fig. 12. True triaxial compression (van Mier, 1984).

Table 6 tential surface. More delicate plastic potential function is required


Parameters for true triaxial compression simulations.
to improve this aspect, which, however, would make the implicit
Effective Young’s modulus Ē 38.0 × 103 N/mm2 return-mapping algorithm very difficult especially when distortion
Uniaxial compressive strength fc 39.9 N/mm2 of the yield surface (Fig. 4) resulting from tension–compression
Mean value of natural logarithm of μ− 7.47
double hardening (Eq. (47)) is present at the same time.
compressive fracture strain
Standard deviation of natural logarithm of V− 0.35 Remark 7.3. It is expected that with adequate cyclic and multiaxial
compressive fracture strain
− experimental stress–strain curves, many of the material parameters
Effective compressive strength index K 0.035 Ē
in the present model can be determined in a statistical sense from
Coefficient of fracture enhancement c 5 the strength and type of a certain concrete material.
Coefficient of hardening slowdown cκ 12
Coefficient of dilation reduction cη 36
8. Conclusions
The novel confining–net decomposition of stress in the pre-
This set of numerical examples preliminarily demonstrate the sented work is effective in distinguishing concrete behavior in tri-
ability of the proposed model to reflect the effects of the interme- axial compression from that in biaxial compression and at the
diate principal stress. same time preserving their similarity. In contrast with the tradi-
tional spherical–deviatoric decomposition, the confining part out
Remark 7.2. It can be found that the ability of the present of the proposed decomposition, which is also a spherical tensor,
model to calculate lateral deformation in true triaxial compression is present only in triaxial compression, thus serving as an indi-
(Fig. 12) as well as in asymmetric biaxial loading (Figs. 8 and 9) cator of triaxial confining effects in both damage and plasticity
is limited by the simple circular deviatoric trace of the plastic po- formulation.
J. Zhang et al. / International Journal of Solids and Structures 94–95 (2016) 125–137 137

This confining–net decomposition can be combined naturally Grassl, P., Xenos, D., Nyström, U., Rempling, R., Gylltoft, K., 2013. CDPM2: A dam-
with the positive–negative decomposition, which deals with uni- age-plasticity approach to modelling the failure of concrete. Int. J. Solids Struct.
50, 3805–3816.
lateral effects successfully. Consequently, the 3D damage model Green, S.J., Swanson, S.R., 1973. Static Constitutive Relations for Concrete Report No.
developed on the basis of the novel decomposition of stress, in AFWL-TR-72-2. Air Force Weapons Laboratory, Kirtland Air Force Base, Albu-
conjunction with the tension–compression double hardening plas- querque, NM.
Imran, I., Pantazopoulou, S.J., 2001. Plasticity model for concrete under triaxial com-
ticity, is also applicable under multiaxial tension–compression or pression. J. Eng. Mech. ASCE 127, 281–290.
reversed cyclic loading. Jason, L., Huerta, A., Pijaudier-Cabot, G., Ghavamian, S., 2006. An elastic plastic dam-
More reliable experimental results in true triaxial compression age formulation for concrete: Application to elementary tests and comparison
with an isotropic damage model. Comp. Meth. Appl. Mech. Eng. 195, 7077–
are warranted to further examine and improve the proposed the-
7092.
ory and to calibrate stochastic damage evolution. Ju, J.W., 1989. On energy-based coupled elastoplastic damage theories: Constitutive
Finally, with the neglected third damage variable restored and modeling and computational aspects. Int. J. Solids Struct. 25, 803–833.
Karsan, I.D., Jirsa, J.O., 1969. Behaviour of concrete under compressive loadings. J.
a closed yield surface employed, the presented framework can be
Struct. Eng. ASCE 95, 2543–2563.
extended to a more complex form for use under very high confine- Kupfer, H., Hilsdorf, H.K., Rusch, H., 1969. Behavior of concrete under biaxial
ment and hydrostatic compression. stresses. ACI J. Proc. 66, 656–666.
Ladeveze, P., 1983. Sur une Theorie de l’endommagement Anisotrope. Laboratoire de
Mecanique et Technologie, Cachan, France.
Acknowledgments Lee, J., Fenves, G.L., 1998. Plastic-damage model for cyclic loading of concrete struc-
tures. J. Eng. Mech. ASCE 124, 892–900.
The financial support from the National Science Foundation of Lee, J., Fenves, G.L., 2002. A return-mapping algorithm for plastic-damage models:
3D and plane stress formulation. Int. J. Numer. Meth. Eng. 50, 487–506.
China (NSFC) Projects (Grant nos. 51378377 and 51108336) and Li, J., Ren, X.D., 2009. Stochastic damage model for concrete based on energy equiv-
Fundamental Research Funds for the Central Universities (Grant no. alent strain. Int. J. Solids Struct. 46, 2406–2419.
2016YXMS094) is greatly appreciated. The third co-author (J.W. Ju) Lubliner, J., Oliver, J., Oller, S., Oñate, E., 1989. A plastic-damage model for concrete.
Int. J. Solids Struct. 25, 229–326.
acknowledges the financial support from the 10 0 0 Talents Program Luccioni, B.M., Rougier, V.C., 2005. A plastic damage approach for confined concrete.
of Tongji University and the Central Organization Department. Comput. Struct. 83, 2238–2256.
Mazars, J., Hamon, F., Grange, S., 2014. A new 3D damage model for concrete under
References monotonic, cyclic and dynamic loadings. Mater. Struct. 1–15.
Oliver, J., Cervera, M., Oller, S., Lubliner, J., 1990. Isotropic damage models and
smeared crack analysis of concrete. In: Proceedings of the Second International
Abu Al-Rub, R.K., Kim, S., 2010. Computational applications of a coupled plastici-
Conference on Computer Aided Analysis Design Concrete Structure. Zell am See,
ty-damage constitutive model for simulating plain concrete fracture. Eng. Fract.
pp. 945–957.
Mech. 77, 1577–1603.
Ortiz, M., 1985. A constitutive theory for inelastic behaviour of concrete. Mech.
Bazant, Z.P., Becq-Giraudon, E., 2002. Statistical prediction of fracture parameters of
Mater. 4, 67–93.
concrete and implications for choice of testing standard. Cem. Concr. Res. 32,
Papanikolaou, V.K., Kappos, A.J., 2007. Confinement-sensitive plasticity constitutive
529–556.
model for concrete in triaxial compression. Int. J. Solids Struct. 44, 7021–7048.
Candappa, D.C., Sanjayan, J.G., Setunge, S., 2001. Complete triaxial stress–strain
Sánchez, P.J., Huespe, A.E., Oliver, J., Diaz, G., Sonzogni, V.E., 2012. A macroscopic
curves of high-strength concrete. J. Mater. Civ. Eng. ASCE 13, 209–215.
damage-plastic constitutive law for modeling quasi-brittle fracture and ductile
Caner, F.C., Bažant, Z.P., 2013. Microplane model M7 for plain concrete. I: Formula-
behavior of concrete. Int. J. Numer. Anal. Meth. Geomech. 36, 546–573.
tion. J. Eng. Mech. ASCE 139, 1714–1723.
Shang, H.S., Ji, G.J., 2014. Mechanical behaviour of different types of concrete under
Caner, F.C., Bažant, Z.P., 2013. Microplane model M7 for plain concrete. II: Calibra-
multiaxial compression. Mag. Concr. Res. 66, 870–876.
tion and verification. J. Eng. Mech. ASCE 139, 1724–1735.
Simo, J.C., Hughes, T.J.R., 1998. Computational Inelasticity. Springer-Verlag, New
Carrazedo, R., Mirmiran, A., Hanai, J.B.D., 2013. Plasticity based stress–strain model
York, NY.
for concrete confinement. Eng. Struct. 48, 645–657.
Valentini, B., Hofstetter, G., 2013. Review and enhancement of 3D concrete models
Cervenka, J., Papanikolaou, V.K., 2008. Three dimensional combined fracture-plastic
for large-scale numerical simulations of concrete structures. Int. J. Numer. Anal.
material model for concrete. Int. J. Plast. 24, 2192–2220.
Meth. Geomech. 37, 221–246.
Chi, Y., Xu, L., Yu, H., 2014. Constitutive modeling of steel-polypropylene hybrid fiber
Van Mier, J.G.M., 1984. Strain-Softening of Concrete under Multiaxial Loading Con-
reinforced concrete using a non-associated plasticity and its numerical imple-
ditions Ph.D. thesis. Eindhoven University Technology, Eindhoven, Netherlands.
mentation. Compos. Struct. 111, 497–509.
Wu, J.Y., 2004. Damage Energy Release Rate-based Elastoplastic Damage Constitutive
Chorin, A.J., Hughes, T.J., McCracken, M.F., Marsden, J.E., 1978. Product formulas and
Model for Concrete and its Application to Nonlinear Analysis of Structures Ph.D.
numerical algorithms. Commun. Pure Appl. Math. 31, 205–256.
thesis. Tongji University, Shanghai, China.
Cicekli, U., Voyiadjis, G.Z., Abu Al-Rub, R.K., 2007. A plasticity and anisotropic dam-
Wu, J.Y., Li, J., Faria, R., 2006. An energy release rate-based plastic-damage model
age model for plain concrete. Int. J. Plast. 23, 1874–1900.
for concrete. Int. J. Solids Struct. 43, 583–612.
Faria, R., Oliver, J., Cervera, M., 1998. A strain-based plastic viscous-damage model
Wu, J.Y., Xu, S.L., 2013. Reconsideration on the elastic damage/degradation theory
for massive concrete structures. Int. J. Solids Struct. 35, 1533–1558.
for the modeling of microcrack closure-reopening (MCR) effects. Int. J. Solids
Faria, R., Oliver, J., Cervera, M., 20 0 0. On Isotropic Scalar Damage Models for the Nu-
Struct. 50, 795–805.
merical Analysis of Concrete Structures. CIMNE Monograph, No.198, Barcelona,
Zhang, J., Li, J., 2012. Investigation into Lubliner yield criterion of concrete for 3D
Spain.
simulation. Eng. Struct. 44, 122–127.
Gabet, T., Malécot, Y., Daudeville, L., 2008. Triaxial behaviour of concrete under high
Zhang, J., Li, J., 2014. Elastoplastic damage model for concrete based on consistent
stresses: influence of the loading path on compaction and limit states. Cem.
free energy potential. Sci. China Technol. Sci. 57, 2278–2286.
Concr. Res. 38, 403–412.
Zhang, J., Li, J., 2014. Microelement formulation of free energy for quasi-brittle ma-
Geopalaeratnam, V.S., Shah, S.P., 1985. Softening response of plain concrete in direct
terials. J. Eng. Mech. ASCE 140, 06014008.
tension. ACI J. Proc. 82, 310–323.
Zhang, J., Zhang, Z.X., Chen, C.Y., 2010. Yield criterion in plastic-damage models for
Grassl, P., Jirasek, M., 2006. Damage-plastic model for concrete failure. Int. J. Solids
concrete. Acta Mech. Solida Sin. 23, 220–230.
Struct. 43, 7166–7196.
Grassl, P., Lundgren, K., Gylltoft, K., 2002. Concrete in compression: A plasticity the-
ory with a novel hardening law. Int. J. Solids Struct. 39, 5205–5223.

Vous aimerez peut-être aussi