Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

SCIENCE CHINA

Technological Sciences
Progress of Projects Supported by NSFC• November 2012 Vol.55 No.11: 2971–2980
doi: 10.1007/s11431-012-4983-6

A plastic damage model for concrete structure cracks with


two damage variables
ZHENG FuGang3, WU ZhongRu1, 2, 3*, GU ChongShi1, 2, 3, BAO TengFei1, 2, 3 & HU Jiang3
1
State Key Laboratory of Hydrology-Water Resources and Hydraulic Engineering, Hohai University, Nanjing 210098, China;
2
National Engineering Research Center of Water Resources Efficient Utilization and Engineering Safety, Hohai University,
Nanjing 210098, China;
3
College of Water-conservancy and Hydropower, Hohai University, Nanjing 210098, China

Received May 22, 2012; accepted June 4, 2012; published online August 12, 2012

Based on the concepts of continuum damage theory, a new plastic damage model for concrete crack failure is developed
through studying the basic damage mechanics. Two damage variables, tensile damage variable for tensile damage and shear
damage variable for compressive damage, are adopted to represent the influence of microscopic damage on material macro-
mechanics properties under tensile and compressive loadings. The yield criteria and flow rule determining the plasticity of
concrete are established in the effective stress space, which is convenient to decouple the damage process from the plastic pro-
cess and calibrate material parameters with experimental results. Meanwhile, the plastic part of the proposed model can be im-
plemented by back-Euler implicit algorithm, and the damage part is explicit. Consequently, there exist robust algorithms for
integrating the constitutive relations using finite element method. Comparison with several experimental results shows that the
model is capable of simulating the nonlinear performance of concrete under multiaxial stress state and can be applied to prac-
tical concrete structures.

concrete, cracks, plastic damage, constitutive relation, damage variable

Citation: Zheng F G, Wu Z R, Gu C S, et al. A plastic damage model for concrete structure cracks with two damage variables. Sci China Tech Sci, 2012, 55:
29712980, doi: 10.1007/s11431-012-4983-6

1 Introduction constitutive relationship of concrete [5–10], is word- widely


recognized. Though the plastic models are suitable for rep-
Concrete is widely used in various engineering fields due to resenting the strain-stress characteristics under multiaxial
its advantages of high compressive strength and strong states in monotonic loading, they fail to address the stiffness
adaptability. However, cracks are common in concrete degradation and the unilateral effect [13–15]. Fracture me-
structures, for instance more than 70% of concrete dams chanics puts emphasis on calculating the driving force on a
exist cracks [1–3]. These cracks directly affect the service crack, propagation rule of cracks and material’s resistance
life of concrete structures [4]. to fracture. It has been extensively used in engineering
Many valuable results concerning the concrete properties practice for fracture analysis. However, fracture models can
about cracks have been obtained using elastoplastic me- hardly represent the properties of concrete with many
chanics, fracture mechanics and damage mechanics [5–15]. smeared microcracks before the appearance of macrocracks.
Plasticity, which has been successfully used to analyze the Besides, it is difficult to get the precise criterions for con-
crete like fracture toughness, so the fracture mechanics has
*Corresponding author (email: zrwu@hhu.edu.cn)
limitation to model concrete [16–18]. Based on the theories

© Science China Press and Springer-Verlag Berlin Heidelberg 2012 tech.scichina.com www.springerlink.com
2972 Zheng F G, et al. Sci China Tech Sci November (2012) Vol.55 No.11

of continuum mechanics and irreversible thermodynamics, tions, including uniaxial tension, uniaxial compression,
the continuum damage mechanics considers microdefects as multiaxial compression and cyclic loading.
an internal state variable field (damage field), regards the
damage process as energy dissipation, and studies the in-
fluence of microdefect’s evolution on mechanical property 2 Framework for plastic damage model
by means of phenomenological method. It models the dam-
age by defining the relationship between nominal stress and According to the concept of continuum damage theory [19],
effective stress, and concerns with the evolution law of the stress-strain relation for damage models can be ex-
damage process on a phenomenological level [19, 20]. pressed as follows:
Since the continuum damage mechanics was first intro-
  (1  d )  (1  d ) E0 : (ε  ε p ), (1)
duced by Kachanov for creep-related problems in 1958 [21],
it has been applied successfully to the modeling of concrete where  is the nominal stress tensor; d is the scalar damage
[22]. However, the elastic damage models are not available variable describing the degradation of elastic stiffness; E0 is
to describe the inelastic deformations [23, 24]. the initial (undamaged) elastic stiffness matrix; ε is the total
Plastic damage, combined plasticity with isotropic or an- strain; εp is the plastic strain including the irreversible de-
isotropic damage to describe the nonlinear response of con- formation caused by microcracks;  is the effective stress
crete due to irreversible deformations and microcracks, is tensor. It can be concluded that plastic damage model con-
the appropriate method for modeling constitutive relation- tains two aspects: plastic process and damage process. They
ship. Anisotropic damage models for concrete are often too are discussed in the following sections respectively.
complex to apply [25–27]. On the other hand, isotropic
damage models are widely used with different types of
2.1 Plastic process
combinations with plasticity. One group of models relies on
plasticity formulated in the nominal stress space [28–30], Effective stress is more representative when damage occurs
and another group of models is based on plasticity formu- because it reflects the true stress on the effective stress area.
lated in the effective stress space [31–33]. Here, nominal On the other hand, according to the research result [33], the
stress is defined as the force divided by the total area, while damage model based on the nominal stress has a severe re-
effective stress is defined as the force divided by the un- striction to meet the local uniqueness condition than that
damaged part of the area. However, all of the aforemen- based on the effective stress. Therefore, the effective con-
tioned damage models have their application limitations. figuration is adopted to facilitate the plastic process. The
For instance, in spite of the satisfactory damage results un- plasticity of model is described by the yield function, the
der the cyclic loading, the model proposed by Lee [32] can flow rule, the evolution law for plastic hardening variable
hardly address the mechanical properties of concrete under and the loading-unloading conditions:
multiaxial loading with different levels of confinement; on
the contrary, the one-scalar damage models like the model F p ( ,  p )  f p ( )  k p ( p ),
proposed by Grassla [33] cannot represent the unilateral G( ,  p )
effect under cyclic loading. ε p   ,
 (2)
Aiming at these problems above, this paper presents a
κ p   H p ( , κ p , x p ),
new plastic damage model for concrete with two damage
variables, which satisfies the irreversible thermodynamics. F p0 , 0 ,  F p  0 ,
The two damage variables, a tensile damage variable for
where F p is the yield function; f p is the equivalent stress; kp
tensile damage and a shear damage variable for compressive
is the equivalent yield stress; p is the plastic hardening varia-
damage, are adopted to describe the damage state of the
concrete. Both the yield function and flow rule are estab- ble;  is the plastic multiplier; G is the plastic potential; Hp
lished in the effective stress space in order to calibrate ma- is the function relating the rate of plastic hardening variable
terial parameters with experimental results and separate the to the plastic multiplier; xp is the hardening ductility.
damage process from the plastic process advantageously. In the following subsequent sections, the basic compo-
Another advantage of this processing method is that the nents of plastic are specified in detail.
damage in stiffness degradation, which originally is coupled
with plasticity, can be separated from the constitutive rela- 2.1.1 Yield condition
tion conveniently. Therefore, the plastic part can be imple- A yield function suitable for multiaxial stress states is
mented implicitly, and the damage part can be implemented adopted, which was first proposed by Etse and Willam in
explicitly. Consequently, there exist robust algorithms for 1994 [8], and later exploited by Menétry and Willam [9],
integrating the constitutive relations, which is convenient Grassl and Lundgren [10]. The yield function is established
for finite element method. The model predictions agree well in terms of cylindrical coordinates in the principal effective
with the experimental results under complex stress condi- stress space (Haigh Westergaard coordinates). The three
Zheng F G, et al. Sci China Tech Sci November (2012) Vol.55 No.11 2973

coordinates: hydrostatic effective stress invariant  , devi-


atoric effective stress invariant  , and Lode angle  are
defined as
I1
  , I1  σ : δ   ij  ij ,
3
1 1
  2J2 , J2  S : S  Sij Sij, (3)
2 2
1 3 3 J  1 1
  arccos  3
3/ 2 

, J 3  S 3 : δ  Sij S jk S ki ,
3  2 J 2  3 3

where S  σ  δI1 / 3 indicates the deviatoric effective


stress tensor.
The yield function is given as
2
 3   r   2
F p
 m0 qh2     qh , (4)
 2 f   6 f c 3 f c 
 c 
Figure 1 The evolution of yield surface in deviatoric plane with different
where f c is the uniaxial compressive strength; m0 is the values of elliptic function.

friction parameter; r is the elliptic function; qh is the nor-


malized strength parameter. The shape of yield surface in
deviatoric section changes from triangular shapes at low
confinement to almost circular shapes at high confinement,
controlled by the elliptic function r ( , e) :

4(1  e2 ) cos2   (2e  1)2


r , (5)
2(1  e2 ) cos   (2e  1) 4(1  e2 ) cos2   5e2  4e

where e is the eccentricity parameter. The value of eccen-


tricity parameter e is required to be 0e1.In the case of
e  1, r ( ,1)  1 and the shape of yield surface in devia-
toric section becomes circular with the influence of the
Lode angle disappearing. For e  0.5, r ( ,1)  2cos ,
which forms a triangle in the deviatoric section as shown in Figure 2 The evolution of tensile and compressive meridians with dif-
Figure 1. Eccentricity parameter e can be evaluated by ferent values of friction parameter.

1 f t f b2  f c2
e ,   , (6) defined as
2  f b f c2  f t 2
qh  R( )qt  (1  R( ))qc , (8)
where f t is the uniaxial tensile strength; f b is the equibi-
axial compressive strength. where qt and qc are the equivalent tensile and compressive
The shape of yield surface in meridional section is para- strength invariants, respectively; R( ) is the stress weight
bolic, controlled by friction parameter m0, as shown in Fig- factor representing a measure of the ratio between tensile
ure 2. The value of m0 can be obtained by and compressive stresses, and can be evaluated as follows:

f c2  f t 2 e 0, if (ˆ  0),


m0  3 . (7)  3
fc f t 1  e 
R( )   
i
i 1
The normalized strength parameter qh controls the do-  3 , otherwise,
(9)
main of the yield surface, which means that qh determines   i
the evolution of the maximum size of the elastic domain in  i 1
deviatoric section and meridional section, as presented in 1
 i  ( i   i ),
Figure 3. Considering the multiaxial stress states, qh is 2
2974 Zheng F G, et al. Sci China Tech Sci November (2012) Vol.55 No.11

Figure 3 The evolution of yield surface during hardening. (a) Yield surfaces in meridional section; (b) yield surfaces in deviatoric section.

where  i is the principle stress; ̂ is the principle stress   / 3  ft / 3 


mQ  AQ BQ exp   , (11)
tensor.  BQ f c
 
Generally speaking, qt=1, on the assumption that con-
crete doesn’t exhibit yield hardening under tension, while where AQ and BQ are model parameters calibrated from ex-
f c 0 f c  qc 0qc1, where f c0 is the initial yield strength. periments. In the case of lack of experiment results, the
values of model parameters can be evaluated by the follow-
Note that under the uniaxial compression, which indi-
ing equations:
cates   0 and   π 3, we can obtain r  1,  
f t m0
 1.5 and    3 . Consequently, the yield func- AQ  3  ,
fc 2
tion is reduced as F p  ( / f c )2  qc2 , which means that (12)
(1  f t / f c ) / 3
qc describes the yield stress in uniaxial compression. Under BQ  ,
ln AQ  ln(2 Df  1)  ln(3  m0 / 2)  ln( Df  1)
the uniaxial tension, which indicates   0 and   0,
we can obtain r  1 e ,   1.5 and    3 . where Df is the ratio between lateral and axial plastic rates,
and generally chosen as Df =0.85.
Therefore, the yield function changes into
2
  f 2  ft2  2.1.3 Evolution of hardening variable
F    c
p
 1. It is obvious that concrete
The evolution law for the hardening variable under multi-
 fc  fc f t fc
axial stress is expressed as follows:
yields when   f t .
(1  R(σ ))  ˆmin
p
κ p  , (13)
2.1.2 Flow rule x p ( h )
In order to describe the inelastic deformation capacity under
where ˆmin
p
is the minimum eigenvalue of the plastic strain
multiaxial stress states, the plastic potential is based on a
volumetric modification of the yield function proposed by rate tensor i.e. the maximum principal compressive strain
Etse. The deviatoric component of the plastic strain rate rate; R(σ ) is the stress weight factor determined by eq. (7);
follows an associated flow rule, while the volumetric com- xp is the hardening ductility defining the peal value of
ponent obeys a nonassociated flow rule. The plastic poten- equivalent plastic strain when the failure envelope is
tial has the format as reached, which can be evaluated as a polynomial:
2 x p ( h )  Ah h2  Bh h  Ch , (14)
 3   m r 
G (σ , κ p )    qh2  0  mQ  . (10)
 2 f   6f  where Ah, Bh and Ch are model parameters calibrated from
 c   c 
compressive experiments under different levels of confine-
The flow direction is controlled by ment;  h describes the confinement level:
Zheng F G, et al. Sci China Tech Sci November (2012) Vol.55 No.11 2975

 max
  bles are adopted, one tensile damage variable for tension
h   , (15)
expressed as κ td , one shear damage variable for compres-
3 fc 3 fc
sion expressed as κcd . So the damage variables can be
 is the stress in the direction of maximum prin-
where  max
written as
cipal compressive strain rate.
Assuming that ˆmax
p
 ˆ1ˆ2ˆ3  ˆmin
p
, the following κ d  [ td  td ]T . (19)
form to obtain The damage part of the model is described by the dam-
age loading function, the evolution of damage variables, the
G(σˆ )
εˆ p  [ˆ1 ˆ2 ˆ3 ]T   . (16) evolution law of scalar damage variable, and the loading-
σˆ unloading conditions:
By substitution of eq. (16) into eq. (13), the expression of
F d (ε, κ d )  f d (ε)  k d (κ d ),
Hp in eq. (2) can be obtained as
κ d   H d (σ , κ d ),
 1  R(σ )  G(σˆ ) (20)
H p  0 0  . (17) d  d (d t , d c ),
 x p  σˆ
F d0, 0,  F d  0,
2.1.4 Isotropic hardening law where F d is the damage loading function; f d is the equiva-
The normalized strength parameter qh controls the size of lent strain; kd is the equivalent damage threshold; κ d is the
yield surface with the variable qc. Its equivalent compres- damage variable vector; Hd is the function relating the rate
sive strength invariant qc can be expressed as a monoton- of damage variables and plastic multiplier; dt and dc are the
ically increasing function of the hardening variable κp: scalar degradation variables of tension and compression,
q  (1  qc 0 ) p [( p )2  3 p  3], if  p  1, respectively. The detail information of damage variable and
qc =  c 0 (18) scalar damage variable are discussed as follows.
1, if  p1.
2.2.1 Evolution of damage variable
The evolution of hardening ductility measure xp and
According to the experimental results [13–15], microcracks
equivalent compressive strength invariant qc under different
usually occur in the direction of maximum plastic strain rate.
levels of confinement is show in Figure 4.
Therefore, it is reasonable that the rates of damage variables
are characterized by the maximum equivalent principle
2.2 Damage process plastic strain rates of tension and compression:
It is difficult to represent the damage states effectively for td  R(σ )ˆmax
p
,
concrete by one single damage variable because of the uni- (21)
lateral effect under cyclic loading and the different me- cd  (1  R(σ ))ˆmin
p
.
chanical responses under tensile damage and compressive
damage. In order to account for the different damage states By substituting eq. (16) into eq. (21), the evolution law
of concrete in tension and compression, two damage varia- of damage variables can be obtained in terms of

0, if qh  1,
κ d   d p (22)
 h  εˆ , if qh  1,

 R (σ ) 0 0 
where hd   . Combined with eq. (20),
 0 0 1  R(σ ) 
G(σˆ )
the expression of H d is H d  hd  .
σˆ

2.2.2 Evolution of scalar damage variable


Modified by the exponent format proposed by Mazars [34],
the evolution laws of tensile degradation variable and shear
degradation variable are defined as follows:
1  A A
Figure 4 The evolution of hardening ductility and compressive strength d  1   , (23)
under different levels of confinement. 1   /  exp( B d / f )
d f
2976 Zheng F G, et al. Sci China Tech Sci November (2012) Vol.55 No.11

where   t , c indicates the tensile and compressive part and damage part of the model can be implemented sep-
arately. Based on the simulating platform of ABAQUS, a
states, respectively; A and B are the damage parameters
user subroutine is developed using the redeveloping tech-
calibrated from experiments; f is the peak value of strain nique of UMAT. The implementation processes of plastic
when damage occurs. and damage are introduced as follows.
Under cyclic loading between tension and compression,
the degradation of the stiffness represents unilateral effect 3.1 Plastic part
because of the opening and closing of microcracks. Two
stiffness recovery coefficients are adopted to describe dif- Backward Euler implicit integration is used for plasticity
ferent stiffness recovery effects associated with stress re- implementation, which includes two aspects: elastic predic-
versals of tension and compression. The scalar damage var- tor and plastic correction. The unknown variables between
iable is assumed as loading steps n1 and n are updated as follows:
d  1  (1  sc d t )(1  st d c ), (24) G
ε(pn)  ε(pn 1)  ( n ) ,
σ ( n )
where st and sc represent stiffness recovery effects of ten-
sion and compression associated with stress reversals, re-  (pn )   (pn 1)  ( n) H p , (26)
spectively. They are defined as σ( n)  E0 : (ε( n)  ε(pn) ),
sc  1   c [1  R(ˆ1 )], st  1  t R(ˆ1 ), (25) F( pn )  F p (σ( n ) ,  (pn ) ).

where ςc and ςt are material parameters controlling the re- Eqs. (26) are a nonlinear function with multiple variables.
covery of the tensile and compressive stiffness upon load Therefore, Newton-Raphson iterative method is adopted for
reversal, and take values in [0,1]. R(ˆ )  ˆ ˆ when the integration by means of the following steps, and the
1 1 1
flow diagram is shown in Figure 6.
the first effective principle stress is not zero. The effluence 1) Initialization:
of stiffness recovery coefficient on stiffness changes where
the load changing from tension to compression is shown in k  0, ((0)
n )  0, ε( n )  ε( n 1) ,  ( n )   ( n 1) .
p(0) p p (0) p

Figure 5.
It is worthwhile to illustrate that the proposed plastic 2) Compute elastic predictor: σ(trn)  E0 : (ε( n)  ε(pn(0)
) ).

damage model for concrete satisfies the thermodynamic law. 3) Check the yield condition: F( p(tr
n)
)
 F p (σ (trn) ,  (pn(0)
) ).
The verification process can be referred to refs [32, 33], and
is not repeated here. If F( p(tr)
n)  TOLF, then calculation converges, indicating

3 Numerical implementation

According to the method proposed in this paper, the plastic

Figure 5 Effluence of the compression stiffness recovery parameter on


stiffness recovery. Figure 6 Flow diagram of numerical implementation.
Zheng F G, et al. Sci China Tech Sci November (2012) Vol.55 No.11 2977

that the stress is in the elastic domain; otherwise go to 4). that of the plastic part since  is known after the end of
4) Initial iteration: plastic analysis. So the calculation process of damage part
F p qc p between steps n1 and n can be treated efficiently as
n )  (1  R(σ ( n ) ))
H ((1) tr
H ,
qc  p κ (dn )  κ (dn 1)  ( n) H d ,
p(tr)
F d( n )  d (κ (dn ) ).
((1)
n) 
( n)
,
F p
G
E0  H ((1)
n) Finally, the nominal stress is computed as
σ (trn ) σ(trn )
G σ( n)  (1  d( n ) )σ ( n) . (27)
n )  σ ( n )  ( n ) E 0
σ((1) tr (1)
,
σ(trn )
 (p(1)
n )   ( n )  ( n ) H ,
p(0) (1) p
4 Model verification
G
n )  ε( n )  ( n )
(1)
ε(p(1) p(0)
.
σ (trn ) To verify the rationality and applicability of the proposed
model, the constitutive responses are compared to the ex-
5) Newton-Raphson iteration: perimental results of various loading conditions of concrete,
① Stress iteration: including uniaxial tension and compression, biaxial com-
G pression, triaxial compression, and cyclic loading. The cor-
r  σ ((nk))  σ (trn)  ((nk)) E0 , responding mechanical parameters are presented in Table 1
σ ((nk))
and the compared results are discussed respectively.
2 G
Q  I  ((nk)) E0 ,
(σ((nk)) )2 4.1 Uniaxial tensile test
F qc p p
The experimental results of uniaxial tensile test [13] are
H ((nk))  (1  R(σ((nk)) )) H ,
qc  p compared with the numerical predictions, as shown in Fig-
F p 1 ure 7. It can be observed that the predictions by the pro-
F( pn() k )  Q r posed model agree well with the experimental data. The
σ((nk))
((nk))  , model can represent the stress strain relation under uniaxial
F p 1 G
Q E0  H ((nk)) tension effectively.
σ (trn) σ(trn )
 2 G  4.2 Uniaxial compressive test
 σ((nk))  Q 1  r  ((nk)) E0  .
 (σ((nk)) )2
  The comparison between the numerical predictions and the
② Check the convergence conditions. If r  TOLR experimental results under uniaxial compression [14] is
presented in Figure 8. As shown in Figure 8, the numerical
and F( p(
n)
k)
 TOLF , go to 6); otherwise go to ③. simulation agrees well with the result of the experimental
③ Update internal variables: test, either in the hardening or in softening regimes.

((nk)1)  ((nk))  ((nk)) ,


4.3 Biaxial compressive test
σ ((nk)1)  σ((nk))   σ ((nk)) ,
The model ability under biaxial compression is validated by
G
ε(pn()k 1)  ε(pn()k )  ((nk)) , comparing the corresponding experiment reported by Kup-
σ((nk)) fer in 1969 [35]. The experimental results with stress ratios
 (pn()k 1)   (pn()k )  ((nk)) H p . of 1/0 and 1/1 are plotted against the numerical perdi-
tions in Figure 9. A good agreement with experimental re-
④ k=k+1, return to ①. sults is obtained as shown in Figure 9. Both strength en-
6) Update final results: hancement and lateral strain are well reproduced.
( n )  ((nk)) , σ( n )  σ((nk)) , ε(pn )  ε(pn()k ) ,  (pn )   (pn()k ) .
4.4 Triaxial compressive test
7) End.
The model response under triaxial compression must be
checked since concrete structures are usually exposed to
3.2 Damage part
triaxial loadings. Experiment study on the concrete response
The numerical implement of the damage part is easier than under triaxial loading was carried out by Imran with lateral
2978 Zheng F G, et al. Sci China Tech Sci November (2012) Vol.55 No.11

Table 1 Model parameters


Test kind E (GPa) v fc (MPa) fc0 (MPa) ft (MPa) Ah Bh Ch At Bt Ac Bc
Uniaxial tension 31 0.2 40  3.41    0.366 5.091  
Uniaxial compression 31 0.2 28 15.4 2.8   0.0012   1.995 0.553
Biaxial compression 31 0.15 32.8 22 3.3 0.0025 0.0026 0.00117   2.696 0.632
Triaxial compression 30 0.15 47.4 18 4.74 0.027 0.0813 0.00122   1.571 0.343
Cyclic tension 31 0.2 40  3.41    0.366 5.091  
Cyclic compression 31 0.2 28 15.4 2.8   0.0012   1.995 0.553
Cyclic tension-compression 39 0.18 47  3.2    0.533 0.088  

Figure 9 The model responses in biaxial compression compared to ex-


perimental results.

4.5 Cyclic loading tests


Figure 7 The model responses in uniaxial tension compared to experi-
mental results. Concrete structures often suffer the cyclic loading in their
service processes. Therefore, the cyclic uniaxial tensile test
[13], the cyclic compressive test [14] and cyclic tensile-
compressive test [37] were reproduced numerically to
demonstrate the capability of the proposed model by com-
parison, as shown in Figures 1113. The following conclu-
sions can be observed: 1) Experimental results including
softening envelop curve, stiffness degradation and devel-
opment of inelastic strain are well reflected by the proposed
model under both cyclic tension and compression. 2) There
exist certain discrepancies in modeling cyclic tensile-com-
pressive load condition. The main reason is that it is diffi-
cult to model such a discrete phenomenon by a continuum
damage model.

Figure 8 The model responses in uniaxial compression compared to


experimental results.

confinement varying from 043 MPa in 1996 [36]. The


comparison between model predictions and experiment re-
sults is shown in Figure 10. It can be concluded that not
only the variation of ultimate strength but also the ductility
under high confinement pressures are represented well in
this model. Meanwhile certain discrepancies are found in
the comparison with the lateral strain but accurate results
could be obtained by means of considering the variation of Figure 10 The model responses in triaxial compression compared to
Poisson’s ratio during the material damage process. experimental results.
Zheng F G, et al. Sci China Tech Sci November (2012) Vol.55 No.11 2979

mechanics, this paper presents a plastic damage model for


concrete structure cracks with two damage variables, which
can describe the nonlinear constitutive relation of concrete
reasonably. The plastic part is established in the effective
stress space, while softening is controlled by damage.
Therefore, the plastic part can be implemented by back-
Euler implicit algorithm and the damage part can be imple-
mented explicitly, which is convenient to numeral integra-
tion. The numerical predictions of the proposed model are
compared with several experimental results, and the fol-
lowing conclusions are obtained.
1) The plastic damage model for concrete structure
cracks proposed in this paper is able to represent the
stress-strain relation and nonlinear properties of concrete
Figure 11 The model responses in cyclic tension compared to experi- such as isotropic hardening, softening and stiffness degra-
mental results. dation under complex load conditions.
2) The adopted yield criterion established in the cylin-
drical coordinates can simulate the enhancement of ultimate
strength and ductility under biaxial or triaxial stress state.
3) The tensile damage variable for tensile damage and
the shear damage variable for compressive damage pro-
posed in this paper can effectively represent the unilateral
effect and stiffness recovery under cyclic loading. The evo-
lution of damage is derived from equivalent plastic strain
instead of plastic energy to avoid the problem of plastic
work decreasing in the case of kinematic hardening.
4) The numerical predictions are close to the experi-
mental results, indicating that the proposed model is appli-
cable. Moreover, the proposed model can be applied to non-
linear analysis of practical concrete structures.

Figure 12 The model responses in cyclic compression compared to ex-


perimental results. This work was supported by the National Natural Science Foundation of
China (Grant Nos. 51139001, 51179066, 51079046), the Program for New
Century Excellent Talents in University (Grant Nos. NCET-11-0628,
NCET-10-0359), the Special Fund of State Key Laboratory of China
(Grant Nos. 2009586012, 2009586912, 2010585212) and the Fundamental
Research Funds for the Central Universities (Grant Nos. 2010B20414,
2010B01414, 2010B14114).

1 Wu Z R, Li J, Gu C S, et al. Review on hidden trouble detection and


health diagnosis of hydraulic structures. Sci China Ser E-Tech Sci,
2007, 50: 34–50
2 Wang S S, Ren Q W. Dynamic response of gravity dam model with
crack and damage detection. Sci China Tech Sci, 2011, 54: 541–546
3 WU Z R, Gu C S. Trouble Detection and Health Diagnosis of Large
Hydraulic Concrete Structure (in Chinese). Beijing: Higher Education
Press, 2005
4 Wang W M, Ding J X, Wang G J, et al. Stability analysis of the tem-
perature cracks in Xiaowan arch dam. Sci China Tech Sci, 2011, 54:
547–555
Figure 13 The model responses in cyclic tension-compression compared 5 Chen A C, Chen W F. Constitutive relations for concrete. J Eng
to experimental results. Mech-ASCE, 1975, 101: 465–481
6 Pramono E, Willam K. Fracture energy-based plasticity formulation
of plain concrete. J Eng Mech-ASCE, 1989, 115: 1183–1203
5 Conclusions 7 Wu Z R, Peng Y, Li Z C, et al. Commentary of research situation and
innovation frontier in hydro-structure engineering science. Sci China
Tech Sci, 2011, 54: 767–780
Based on the thermodynamics law and continuum damage 8 Etse G, Willam K. Fracture energy formulation for inelastic behavior
2980 Zheng F G, et al. Sci China Tech Sci November (2012) Vol.55 No.11

of plain concrete. J Eng Mech-ASCE, 1994, 120: 1983–2011 23 Loland K E. Continuous damage model for load-response estimation
9 Menétrey Ph, Willam K J. Triaxial failure criterion for concrete and of concrete. Cement Concrete Res, 1980, 10: 395–402
its generalization. ACI Struct J, 1995, 92: 311–318 24 Mazars J. Continuous damage theory-application to concrete. J Eng
10 Grassl P, Lundgren K, Gylltoft K, et al. Concrete in compression: A Mech-ASCE, 1980, 115: 345–365
plasticity theory with a novel hardening law. Int J Solids Struct, 2002, 25 Krajcinovic D, Fonseka G U. The continuous damage theory of brit-
39: 5205–5223 tle materials, part 1: general theory. J Appl Mech-T ASME, 1981, 48:
11 Bao T F, Peng Y, Cong P J, et al. Analysis of crack propagation in 809–815
concrete structures with structural information entropy. Sci China 26 Voyiadijs G Z, Taqieddin Z N, Kattan P I, et al. Anisotropic dam-
Tech Sci, 2010, 53: 1943–1948
age-plasticity model for concrete. Int J Plasticity, 2008, 24: 1946–
12 Bao T F, Qin D, Zhou X W, et al. Abnormality monitoring model of
1965
cracks in concrete dams. Sci China Tech Sci, 2011, 54: 1914–1922
27 Du R Q, Zhang Q, Chen S H, et al. Safety evaluation of Dagangshan
13 Gopalaratnam V S, Shah S P. Softening response of plain concrete in
arch dam resisting strong earthquakes with a rate-dependency aniso-
direct tension. ACI J Proc, 1985, 82: 310–323
tropic damage model. Sci China Tech Sci, 2011, 54: 531–540
14 Karsan I D, Jirsa J O. Behavior of concrete under compressive load-
28 Lubliner J, Oliver J, Oller S, et al. A plastic-damage model for con-
ings. J Struct Div, 1969, 95: 2543–2563
crete. Int J Solids Struct, 1989, 25: 299–326
15 Mazars J. A model of unilateral elastic damageable material and its
29 Imran I, Pantazopoulou S J. Plasticity model for concrete under tri-
application to concrete. Wittmann F H, ed. Fracture Toughness and
axial conpressions. J Eng Mech-ASCE, 2001, 127: 281–290
Fracture Energy of Concrete, Amsterdam, Elsevier, 1986. 61–71
16 Xu S L, Zhao G F. Research on the Fracture Mechanics of Concrete 30 Krätzig W B, Pölling R. An elasto-plastic damage model for rein-
(in Chinese). Dalian: The Press of Dalian University of Technology, forced concrete with minimum number of material parameters.
1991 Comput Struct, 2004, 82: 1201–1215
17 Gu C S, Li Z C, Xu B, et al. Abnormality diagnosis of cracks in the 31 Ju J W. On energy-based coupled elastoplastic damage theories.
concrete dam based on dynamical structure mutation. Sci China Tech Constitutive modeling and computational aspects. Int J Solids Struct,
Sci, 2011, 54: 1930–1939 1989, 25: 803–833
18 Liu Y R, Chang Q, Yang Q, et al. Fracture analysis of rock mass 32 Lee J, Fenves G V. Plastic-damage model for cyclic loading of con-
based on 3-D nonlinear Finite Element Method. Sci China Tech Sci, crete structures. J Eng Mech-ASCE, 1998, 124: 892–900
2011, 54: 556–564 33 Grassl P, Jirásek M. Damage-plastic model for concrete failure. Int J
19 Yu T Q, Qian J C. Damage Theory and its Application (in Chinese). Solids Struct, 2006, 43: 7166–7196
Beijing: National defense Industry Press, 1993 34 Mazars J, Pijaudier-Cabot G. Continuum damage theory-application
20 Zhou H, Zhang K, Feng X T. A coupled elasto-plastic-damage me- to concrete. J Eng Mech-ASCE, 1989, 115: 345–365
chanical model for marble. Sci China Tech Sci, 2011, 54: 228–234 35 Kupfer H, Hilsdorf H K, Rusch H, et al. Behavior of concrete under
21 Kachanov L M. Time of the rapture process under creep conditions. biaxial stresses. ACI J Proc, 1969, 66: 656–666
Izv Akad Nauk SSR Otd Tech Nauk, 1958, 8: 26–31 36 Imran I, Pantazopoulou S J. Experimental study of plain concrete
22 Yazdani S, Schreyer H L. Combined plasticity and damage mechan- under triaxial stress. ACI Mater J, 1996, 93: 589–601
ics model for plain concrete. J Eng Mech-ASCE, 1990, 116: 37 Reinhardt H W. Fracture mechanics of an elastic softening material
1435–1450 like concrete. Heron, 1984, 29: 1–42

You might also like