Anda di halaman 1dari 55

95

CHAPTER FIVE
MODELING OF REINFORCED CONCRETE FOR
FINITE ELEMENT ANALYSIS
5.1 Introduction
Present day design codes are mainly based on static design methods.
Established design and analysis methods are based on simplified
equivalent static forces or single degree of freedom systems. Dynamic
loads from earthquakes are usually considered by introducing an
equivalent dynamic load factor. The case also stands for design of
significant concrete structures where an impact factor is applied. For both
static and dynamic analysis the design of the structure is based on
working stress method or ultimate theory and plastic methods.
Linear elastic theory by assuming a linear response materials and
structures, has been employed in the design of structural members. On the
basis, a factor of safety relating working stresses with failure stresses is
defined to guarantee the safety of the analysis. Although this method has
generally resulted in safe designs, it contains inherent inconsistencies and
does not reflect the real behavior. Materials behave in an inelastic manner
and the structures often experience a profound stress distribution. Linear
theory is no longer valid in estimating the true stress distribution and thus
the factor of safety becomes meaningless. In addition the factor of safety
is the same for different types of structure and different boundary
conditions.
To overcome the drawback of the linear elastic theory, ultimate load
theories and plastic methods of design have bean developed. Using these
methods, the structure is analyzed when it is to fail, that is, when collapse
mechanism is formed or yield lines have been formed. To apply these
plastic methods of design however one has to know the stress distribution
near failure, as well as, the actual collapse mechanism. This is not a
straightforward problem to solve specially in the case of complex loading
96
and geometry. On the other hand, the upper bound solutions, which are
assumed to be given by the yield line theory, often underestimate the
collapse loads since inplane forces and geometric nonlinear behavior are
difficult to be taken into account. The occurrence of brittle failure in
reinforced concrete structure is also a serious drawback in applying these
plastic methods, which require the necessary deformation to develop the
collapse mechanism.
The objections previously pointed out to the traditional methods of
analysis, the new application of reinforced concrete, such as in nuclear
power plants for which the traditional analysis method are clearly
inadequate, and the development of powerful computers have created the
conditions for developing new methods. One of these methods, is the
finite element method, which can incorporate the material modeling and
thus predict the response of structures throughout the entire load range.
The complex behavior of reinforced concrete components and
structures, namely concrete cracking, tension stiffening, nonlinear
multiaxial material properties and complex interface behavior, previously
ignored or treated in a very approximate way, can presently be
incorporated more rationally into the analysis. Improved understanding of
reinforced concrete can be achieved by using the finite element method. It
provides an invaluable tool for research and interpreting experimental
results, and a firm basis for codes and specifications. Realistic and
complete structural response up to collapse load can be carried out using
the present method, assessing all safety aspects of analysis. Therefore the
finite element method is also an essential tool to be used directly in the
analysis and design of complex structures, such as long-span roofs, large
panel building systems, segmental box girders, cooling towers, and
nuclear power plants.
97
The finite element method is capable to simulate the complex
behavior of reinforced concrete materials and geometry. So it is important
to know the experimental behavior of such material so as to model this
behavior. Modeling of reinforced concrete in a finite element code needs
to introduce the mechanical behavior of concrete in uniaxial, biaxial, and,
triaxial state and for steel in uniaxial state.
In this chapter the mechanical properties of concrete under uniaxial,
biaxial, and triaxial states of stress and also some general stress-strain
characteristics of steel reinforcement are presented. Numerical models for
reinforced concrete members in tension and compression are presented.
Elastic-perfectly plastic and strain hardening plasticity approaches are
employed to model the compressive behavior of concrete. A dual criterion
for yielding and crushing in terms of stresses and strains is employed. In
tension modeling the tension stiffening is taken into account by tri-linear
model fitted to experimental results and by the concepts of fracture
energy. Aggregate interlock and sometimes the dowel action are taken
into account by employing the reduced shear modulus. Layered
formulation is adopted to simulate the progressive concrete cracking and
yielding through thickness. Nonlinear model for the curved beam through
layers with numerical treatment is adopted for the nonlinear analysis of
the reinforced beams and reinforced concrete stiffeners for the stiffened
shell when the stiffeners are simulated by curved beam elements.
The dynamic effect on the concrete is included for both
compression and tension situation. Strain rate effect is included for the
compression state and appropriate model for tension is included based on
the maximum strain value.
5.2 Mechanical Behavior of Reinforced Concrete
Reinforced concrete is a composite material consisting of steel
reinforcement and concrete, these two materials having vastly different
98
properties. The required mechanical properties of reinforcing steel are
generally known. However, those for concrete are more difficult to define
depending upon the particular condition of mixing, placing, curing,
nature, rate of loading and environmental influences.
Concrete contains a large number of micro-cracks, especially at
interfaces between coarse aggregate and mortar, even before any load has
been applied. This property is decisive for the mechanical behavior of
concrete. The propagation of these micro-cracks during loading
contributes to the nonlinear behavior of concrete at low stress level and
cause volume expansion at failure.
Many of these cracks are caused by segregation, shrinkage, or
thermal expansion in the mortar. Some micro-cracks may be developed
during loading because of difference in stiffness between aggregate and
mortar. The differences can result in a strain at interface zone several
times larger than the average strain. Since the aggregate-mortar interface
has a significantly lower tensile strength than the mortar, it constitutes the
weakest link in the composite system. This is the primary reason for low
tension strength of the concrete material. From the preceding discussion
one can expect that the size and texture of aggregate have a significant
effect on the mechanical behavior of concrete under various types of
loading.
5.3 Experimental Concrete Behavior
The mechanical properties of concrete under uniaxial, biaxial, and triaxial
states of stress and also some general stress-strain characteristics of
reinforcement are essential for a generalized development of
mathematical modeling of concrete and steel. These serve two major
purposes, firstly, to give guidance on the proper type of material behavior
to be developed in mathematical models and to provide data for the
determination of various material constants, which appear inmathematical
models.
99
5.3.1 Uniaxial Compression Behavior of Concrete
A typical stress-strain relationship for concrete subjected to uniaxial
compression is shown in Figure(5.1)[7]. The stress-strain curve has
nearly linear-elastic behavior up to 30 percent of its maximum
compressive strength. For stress above this point, the curve shows a
gradual increase in curvature up to a bout 0.75 f c , whereupon it bends
more sharply and approaches the peak point at f c . Beyond this peak, the
stress-strain curve has a descending part until crushing failure occurs at
some ultimate strain  u .
The shapes of stress-strain curves are closely associated with the
mechanism of internal progressive micro-cracking. For stress in the
region up to 30 percent of f c , the cracks exist in concrete before loading
remain nearly unchanged. The stress level of about 30 percent of f c has
been termed the onset of localized cracking and has been proposed as a
limit of elasticity. For stress between 30 to 50 percent of f c , the bond
crack start to extend due to stress concentration at crack tips. For stress

between 50 to 75 percent of f c , some cracks at nearby aggregate surface


start to bridge in the form of mortar cracks. At the same time the other
bond cracks continue to grow slowly. The progressive failure of concrete

near f c is primarily caused by micro-cracks through the mortar. These


micro-cracks join bond micro-cracks at the surface of nearby aggregate
and form micro-cracks zones. The shapes of stress-strain curves are
similar for concrete of low, normal, and high strength, as shown in
Figure(5.2) [7], all peak points are located close to the strain value of
0.002.
5.3.2 Uniaxial Tension Behavior of Concrete
Figure (5.3) shows the stress-strain curve in unaxial tension [7]. The
shape of the curves shows similarities to the uniaxial compression curves.
100
This is not surprising since the role of microcracking must be more
important for tensile state of stress.
For stress less than 60 percent of uniaxial tensile strength f t , the
creation of new cracks is negligible. Thus, this stress level will
correspond to a limit of elasticity.
101

The direct tensile strength is difficult to measure and it is usually

taken with respect to the compression uniaxial test range from .05 to 0.1.
fr
The modulus of rupture ( ) or the split cylinder strength is often used to

approximate the tensile strength of concrete. The value of the modulus of

rupture varies widely but it is normally taken as[7]

f r  0.62 f c ( N / mm 2 )
(5.1)

5.3.3 Biaxial Behavior of Concrete

Figure (5.4) shows typical experimental stress-strain curves of concrete

under biaxial compression. It is seen that the maximum compressive

strength increases for the biaxial compressive state. A maximum strength

increase of approximately 25 percent is achieved at the stress ratio of


 1 /  2  0.5
and it is reduced to about 16 percent at an equal
 1 /  2  1. 0
biaxialcompression state ( ). Under biaxial compression-

tension, the compressive strength decreases almost nearly as the applied

tensile stress is increased. Under biaxial tension, the strength is almost the

same as for uniaxial tensile strength. (see Kupfer et al 1969)Figure(5.5)

[19].

Figure(5.1) Typical plot of compressive stress vs. axial and lateral strain
102

Figure (5.2) Uniaxial compressive stress-strain curves for concrete [7]

Figure (5.3) Uniaxial tensile stress-strain curves for concrete [7]


103

Figure (5.4) Typical stress-strain relationships for concrete under biaxial


compression [19]

Figure (5.5) Biaxial strength envelopes proposed by Kupfer and Grestle


[18]
104

5.3.4 Triaxial Behavior

Figure (5.6) shows typical stress strain curves of concrete under triaxial

stress state [7]. In the figure, the effect of the confining stress on the

behavior of concrete is shown. Depending on confining stress, concrete

can act as a quasi-brittle, plastic softening, or plastic hardening. This is

because under higher confining stress the possibility of bond cracking is

greatly reduced and the failure mode shifts from cleavage to crushing of

the cement paste.

Assuming isotropic material behavior, the general shape of the

failure surface for plain concrete has the characteristics illustrated in

Figure (5.7).

In Figure (5.8), the characteristics of failure based on the

experimental results is shown [40] and it is discussed in the next section.


105

Figure (5.6) Typical stress-strain relationships for concrete under triaxial


compression [7]

Figure (5.7) General shape of the failure surface for plain concrete under
triaxial loading.

Figure (5.8) Trace of failure in the deviatoric planes and set of


corresponding experimental data [39]
106

Figure(5.7)General shape of the failure surface for plain concrete under


triaxial loading

Figure(5.8) Trace of failure in the deviatoric planes and set of


corresponding experimental data[40]
107

5.4 Failure Criterion for Concrete

The strength of concrete under multiaxial stresses is a function of the

state of stress and cannot be predicted by limitation of simple tensile,

compressive or shearing stress independently of each other. The general


I1 , J 2 J3
shape of failure surface is defined by the stress invariants and ,
I1 J2 J3
where is the first stress invariant and and are the second and

the third invariant of the deviatoric stress tensor. The shape of the failure

surface can also be represented in a stress coordinate system by two

vectors, the first pass through the hydostatic axis and the second is

perpendicular to the hydostatic axis, Figure (5.7). The latter vector could

be rotated by an angle to make a plane called the deviatoric plane, Figure

(5.7). In other words, the deviatoric plane is perpendicular to the

hydostatic axis. The general characteristics of the failure surface of

concrete are determined by experiments. The deviatoric plane passing

through the origin is called the II-plane.

The general shape of the failure surface in a three dimensional

stress space can best be described by its cross sectional shapes in the

deviatoric planes and its meridians in the meridian planes. The meridians

of the failure surface are the intersection curves between the failure

surface and a plane( the meridian plane) containing hydrostatic axis.


108

Based on experimental data on concrete specimens, the failure

curves in the deviatoric plane in general are smooth, convex, three-fold

symmetry, and nearly triangular for tensile and small compressive

stresses and bulged (more circular) for higher compressive stresses,

Figure (5.8).

Because the failure has surface threefold symmetry, it is sufficient

to explore only the sector   0 to   60 . The two extreme meridian

planes corresponding to this sector are called the tensile meridian and the

compressive meridian respectively.

From the experimental test the failure curves in the meridian plane

are in general smooth, curved, and convex. Another important property

for the yield surface for plain concrete is rt / rc  1.0 , where the indices, t

and c correspond to the tensile and compressive meridians, respectively.

Different yield criterions were proposed. The geometric

representation of these criterions will be represented by surfaces. These

surfaces will be proposed with parameters. The number of parameters

will depend on the degree of the surface in the meridian and deviatoric

planes. For example the conventional Drucker-Prager yield surface is

represented by a cone, in other words a straight line in the meridian plane

and a circle in the deviatoric plane. Two parameters are sufficient to

represent this surface one for the meridian plane and the second for the

deviatoric plane [7,30].


109

In the present work, four criterions will be described and used to

model the concrete in compression. Comparison between these criterions

is discussed through analyzing reinforced concrete members and

comparing with the published experimental results.


5.4.1 Willam Criterion
Willam and Warnke(1975) as cited in reference [7] suggested a

three-parameter failure surface for concrete in tension and low

compression regime Figure(5.9). This failure surface has a straight line in

the meridian planes, and noncircular cross section in the deviatoric plane.
1 m 1 m (5.2)
f ( m , m , )   1
 f f c r ( ) f c

2rc (rc 2  rt 2 ) cos r  rc (2rt  rc )[4(rc 2  rt 2 ) cos 2   5rt 2  5rt 2  4rt.rc ]0.5
r ( ) 
4( rc 2  rt 2 ) cos 2   (rc 2  2rt ) 2
(5.3)
1 2
where  m  I1 , m
2
 J2 and  f , rt , rc are material parameters.
3 5
The three parameters are identified by the three typical concrete

tests: the uniaxial tension test ( f t ),the uniaxial compression test( f c ), and

the equal-biaxial compresion test ( f bc . ). By substituting these strengths

into the failure condition, the three model parameters are readily obtained

[7]
f bc . f t
f  (5.4)
f bc .  f t.
1/ 2
5 f bc . f t
rt    (5.5)
6 2 f bc .  f t.
1/ 2
5 f bc . f t
rt    (5.6)
6 3 f bc . f t  f bc .  f t.
110
f bc . f t
where f bc .  and f t 
f c f c
In the present analysis f bc .  1.8 and f t  0.15 is adopted which give
good agreement when compared with Launay and Gachon[21].

Figure (5.9) Comparison of Willam criterion with triaxial data (a) hydrostatic
section (   0 ), f ab 1.8 , x= experiment ( Launay et. al. 1972); (b) deviatoric
section.
111
5.4.2.Generlized Willam Criterion
Menertrey and Willam (1995) refined the three-parameter Willam
criterion by adjusting the compression and tension meridians[25]. These
meridians are no longer straight lines. The adjustment also is made to the
deviatoric plane which depends on the eccentricity (e).

2
 f   f f 
f ( f ,  f , )   1.5   m r ( , e)   1  0 (5.7)
 f c   6 f c 3 f c 
4(1  e 2 ) cos 2   (2e  1) 2
r ( , e)  (5.8)
2(1  e 2 ) cos  (2e  1)[4(1  e 2 ) cos 2   5e 2  4e]0.5
1
f  I1 (5.9)
3
 f  2J 2 (5.10)
3 3 J3
cos 3  (5.11)
2 J 32
2

 f c 2   f t 2 e
m (5.12)
f c f t e 1
112

The eccenricity value is obtained form a figure which depends on

f bc .
the relation between the axial and biaxial compressive strength f c and

also on the relation of uniaxial compressive strength on and uniaxial

f t
tension strength relation f  [25]. For example e=0.52 is obtained by
c

f bc . f t
using  1.14 and  0.1 which give good agreement when
f c f c

compared with the experimental results of Kupfer [19], Figure(5.11) in

the deviatoric plane. The comparison of this criterion in the meridian

plane for compression and extension meridians gives good agreement

with triaxial test data by Chinn and Zemmerman[8] and Mill and

Zimmerman[26] Figure(5.10).
113

Figure (5.10) Comparison of three-parameter concrete criterion with


triaxial test data by Chinn and Zimmerman[8] and Mill and
Zimmerman[26]

Figure(5.11) Comparison of three-parameter concrete criterion with biaxial


test data by Kupfer et al [19].
114
5.4.3.Ottosen Criterion

Ottosen (1977) suggested a four-parameter criterion [29]. The failure

surface has curved meridians and noncircular cross sections. The failure

curves on the deviatoric plane change from nearly triangular to the nearly

circular with increasing hydostatic pressure and this criterion can be

represented by:

J2 J2 I
f ( I 1 , J 2 , cos 3 )  a  f  b 1 1 (5.13)
f c
2
f c f c
 1 
 f  k1 cos   cos 1 ( k 2 cos 3 ) for cos 3  0 (5.14)
3 3 
1 
 f  k1 cos  cos 1 (k 2 cos 3 ) for cos 3  0 (5.15)
3 
I1
:is the first stress invariant tensor
J2 : is the second invariant of stress deviatoric tensor.
k1 , k 2 , a , b
are material parameters
115

The four parameters in the failure criterion are determined on the


f c 
ft
basis of two typical uniaxial concrete tests ( and ) and two typical

biaxial and triaxial concrete data.

In the present study, the four-parameter criterion which agrees well

with experimental results [29] are adopted as a= 0.9218, b=2.5969,

k1=9.9110, and k2=0.9647.

Figure(5.12) illustrates the agreement between the Ottosen criterion

and the experimental data referred to the compressive and tensile

meridians. The ability of this criterion to represent the experimental

biaxial data of Kupfer et. al [19] is shown in Figure(5.13)

Figure (5.13) Biaxial representation for Ottosen criterion [29]


Figure (5.12) Correlation between Ottosen criterion and experimental data
[29]
116
5.4.4 Drucker-Prager Modified by Kupfer
This criterion is formulated in terms of the first two stress invariants and
only two material parameters are involved in its definition
[6,10,15,37,38]:

f ( I 1 , J 2 )   f (3J 2 )   f I 1  0.5
0 (5.16)
where f and  f are material parameters and   taken as the compressive
strength from uniaxial test( f c ). In terms of the principal stresses the
expression for yielding can be written as:
 
 f  12   22   32     1 2   1 3   3 2    f   1   2   3    02 (5.17)

The material parameters have been obtained by fitting biaxial test


results of Kupfer [19] as follows:
for uniaxial compression test
 1  f c , 2   3  0 (5.18)
for biaxial compression test
 1   2  1.16 f c (5.19)
 f  0.355 0 ,  f  1.355

Equation (5.16) can be written in terms of stress components as:


f ( )  [1.335[( X2   Y2   Y  X )  3( XY
2
  XZ
2
  YZ
2
)]
(5.20)
 0.355 0 ( X   Y )]1 / 2   o

in Figure (5.14) a comparison of this expression with experimental results


of Kupfer[19] is illustrated.

5.5 Compressive Behavior of Concrete


The following three conditions have to be considered in establishing the
nonlinear relationships based on flow theory of plasticity:
(i) The yield criterion
(ii) The flow and hardening rules
(iii) Crushing condition
117
(i) The Yield Criterion
The strength of concrete under multiaxial state of stress is a function of
the state of stress and can not be predicted by limitation of simple tensile,
compressive, and shearing stress independently of each other. Therefore,
a proper evaluation of plain concrete strength can be achieved by
considering interaction of various components of state of stress. This can
be accomplished by using a yield criterion or yield condition. This yield
criterion determines the stress level at which plastic deformation begins.
The yield criterion can be written in a general form as:
f  F ( )  h(k ) (5.21)
where F is some function that represents the yield function, and can be
represented by a surface called yield surface or loading surface and h is a
material parameter. The term h may be a function of the hardening
parameter. In equation (5.21) if f  0 this indicates that plastic state
begins, and f  0 indicates elastic stage. No significance is attached to the
case where f 0.

In the present study, the two models of elastic perfectly plastic and
strain hardening approach are used.
In elastic-perfectly plastic state the equivelent yield stress (  0 ) is
taken as the compressive strength ( f c ), so the Gauss point is still in the
elastic state if the effective stress is less than the compressive strength.
After that the plastic state is assumed until crushing occurs. In the strain
hardening approach the inelastic stage begins when the effective stress
exceeds 30% of the compressive strength, after that perfect plastic state
is considered again when the effective stress exceeds the compressive
strength until crushing is assumed to occur. Figure(5.15) illustrates the
one dimensional representation of both perfectly plastic and the strain
hardening model.
118
(ii-a)The Flow Rule
To construct the stress-strain relationship in the plastic stage, the
normality of plastic deformation rate vector to yield surface or the so
called flow rule is commonly assumed. The plastic strain increment is
then defined as [30]
f
d p  d (5.22)


where d  is a proportionality constant, which determines the magnitude


of the plastic strain increment. The current stress function f ( .) is the
yield condition, or the subsequent loading function in strain hardening
model.
The yield function derivative which defines the flow vector (a)
takes the following expression for the present yield surfaces:
 f f f f f 
aT   , , , ,  (5.23)
  x  y  xy  yz  xz 

The constant d can be obtained as follow [30]

d 
 a T  D  a . d
(5.24)
H    a T  D . a

where [D] is the elasticity matrix for elastic concrete and H  is the
hardening parameter which can be expressed as:
d
H  (5.25)
d P
119

Figure (5.14) Compressive yield and tensile cracking criterion for


concrete.

Figure (5.15) One dimensional representation of the concrete constitutive


model
120

The total strain increment  d  is the sum of elastic and plastic


components, so that
 d    d e    d p 
f
  D  d   d (5.26)


Equation(5.24) is substituted into Equation (5.26) to obtain the complete


elasto-plastic incremental stress-strain relation as:
d  Dep d (5.27)

Dep  D 
 D  a a  D
T

with
H    a  D  a
T

(ii-b) The Hardening Rule


The hardening rule defines the motion of the subsequent yield surface
during plastic deformation. If the subsequent yield surface is a uniform
expansion of the original yield curve, without translation, the strain
hardening model is said to be isotropic. On the other hand, if the
subsequent yield surface preserves its shape and orientation but translates
in stress space as a rigid body, kinematic hardening model is said to take
place. Mixed hardening model is achieved by combining the two previous
models, a uniform expansion of the original yield curve with translation
in any direction with the same orientation. In the present study, isotropic
hardening model is adopted.
The hardening rule determine the relation between the load
surfaces (or effective stress) and the accumulated plastic strain (or
effective plastic strain). The concept of effective stress and effective
plastic strain makes it possible to extrapolate from uniaxial test to
multiaxial situation. In the present study, the following relationship
between the effective stress and the effective plastic strain is adopted [10]
  
 0   f c   2 (5.28)
f   f 
121
2f
if E0  
c

E0 2
  0  E0  2 f
(5.29)

where  0 : effective stress


E0 the initial Young’s modulus
 is the current total strain
f is the strain at peak stress ( f c )

Knowing that the elastic strain  e 
E0 , substituting this in equation(5.29),

the described relationship is obtained as [10]:


 0  E 0  p   2 E 02 . f  p  0.3 f c <  0 < f c
0.5
(5.30)
where  p is the plastic strain component
Differentiating equation (5.28) gives the slope of the  0   p curve

which defines the hardening parameter [10,15,37,38].


d 0  f 
H   E0   1 (5.31)
d P   p 

This equation is adopted in the study.

(iii)The Crushing Condition of Concrete


After the compressive stress reaches the compressive strength of plain
concrete, a perfect plastic state is assumed until crushing occur. Then the
material is assumed to lose all its characteristics of strength and rigidity.
The crushing type of such material is controlled by strain phenomenon. A
simple way is to convert the yield criterion (described in terms of
stresses) into strains [10,15,37,38]. Thus
 f  3J 2    f I 1   u2 (5.32)

where I 1 and J 2 are strain invariants and  cu is the ultimate total strain

extrapolated from uniaxial test.


122
The crushing condition can be expressed in terms of strain
components as:
1.355[( x2   y2   x  y )  0.75( xy2   xz2   yz2 )]  0.355 u ( x   y )   u2 (5.33)
When the strain reaches the crushing surface the material is assumed to
lose all its characteristics of strength and rigidity.

5.6 Interactive Effects of Cracked Concrete


5.6.1 Bond Slip
After the formation of primary cracks, the total load applied is suddenly
transferred from the concrete to the steel bar at the cracked sections. This
causes a differential deformation of both materials with the consequent
sliding along the contact surface. The bond mechanism is then developed,
transferring tensile stresses from steel to concrete between primary
cracks. Firstly, adhesion and friction are the principal effects responsible
for bond forces, but with increasing differential deformation adhesion is
destroyed and the mechanical resistance of lugs and the frictional
resistance are then the responsible mechanism of stress transfer. But when
plasticity starts in the reinforcing bars the bond interaction is completely
extinct.
In a reinforced concrete structure cracked concrete can therefore
carry, between cracks, a certain amount of tensile load which contributes
to the overall stiffness of the structure. This effect is often known as
tension stiffening of cracked concrete. Figure (5.16) shows an assumed
tensile stress diagram in cracked concrete at cross sections increasingly
distant from a primary crack. Stress in reinforcing steel and average
stresses in the concrete between two cracks as well as bond stresses are
illustrated in Figure (5.17).
123
5.6.2 Shear Transfer
Experimental investigations have shown that the cracked concrete
contributes in shear carrying capacity of the overall reinforced concrete
members. Two mechanisms are present in that transfer: the dowel action
and the interface shear transfer.

5.6.2.1 Dowel Action:


Shear force transfer by the dowel action occurs when the reinforcing bars
cross the cracks in the reinforced concrete members. Figure (5.18)
illustrates this effect in a reinforced concrete beam with a diagonal crack.
The longitudinal splitting of concrete is regarded as a limitation to the
dowel action when stirrups are absent. This splitting is due to the vertical
tensile stress produced in concrete at the plane of reinforcement. The bar
diameter and bar arrangement are important variables in shear transfer by
dowel action.

5.6.2.2 Aggregate Interlock


The surface of cracked concrete is rough and irregular. These rough
surfaces can transfer shear stresses by aggregate interlock and friction as
shown in Figure (5.19). The interface shear capacity depends to a large
degree on the development of constraining forces normal to the crack
plane. The crack width was found to be the primary variable affecting the
aggregate interlock action. The aggregate size, reinforcement ratio and
concrete compressive strength are also important variables.

Figure(5.16)
Figure(5.17)concrete
Stress variation
stress diagrams
betweenintwo
axial
primary
tensioncracks
cracked members
124

Figure (5.16) Concrete stress diagrams in axial tension in cracked


members

Figure (5.17) Stress variation between two primary cracks


125

Figure (5.18) Dowel action mechanism of shear transfer across cracks

Figure (5.19) Aggregate interlock mechanism of shear transfer across


cracks.
126
5.7 Concrete in Tension:
In the finite element analysis, plain concrete is initially considered to be
an isotropic material. For thick shell analysis, the stress-strain relation
takes a simple form:

1  0 0 0
 x     x 
   1 0 0 0  
 y   1  y 

   E 0 0 0  
0
 xy   2 2  xy  (5.34)

 1    1  
 xz  0 0 0 0  
2r  xz 
 yz   1    yz 
  0 0 0 0  
 2r 
When the maximum tensile stress exceeds the tensile strength at any
Gauss point, a crack is assumed to occur perpendicular to the direction of
the tensile stress. This approach is called the maximum tensile stress
criterion. Maximum tensile strain criterion can also be used by the same
way. Both approaches are used in the present study. After cracking has
occurred the cracked concrete becomes an orthotropic material and a new
stress-strain relationship must be derived. This is accomplished by
modifying the stiffness in the direction perpendicular to crack direction.
127

0 0 0 0 0
 1  0 E 0 0 0 1 
    
 2  0 0 c 0 0  2 
 12    G12  12  (5.35)

  0 c  
0 0 G13 0   13
 13    
 23   5  23 
0 0 0 0 G
 6
where E: modulus of elasticity
Gc
12 and Gc
13 are the reduced shear moduli
When cracks occur in both directions 1 and 2, the stress-strain
relationship becomes:

0 0 0 0 0
 1    1 
  0 0 0 0 0  
2 
 2  0 0 c 0 0  
 12    G12  12  (5.36)

   c  
 13  0 0 0 G13 0  13 
 23   c  23 
0 0 0 0 G23 

This relation must be transformed to the local axes by using


transformation relations.
128

 
 1,2  T   xy (5.37)

 
 1,2  T   yx (5.38)
 cos 2  sin 2  2 sin  cos  0 0 
 
 sin  cos   2 sin  cos 
2 2
0 0 
 T    sin  cos 

cos sin  cos   sin 
2 2
0 0 

(5.39)
 0 0 0 cos sin  
 0 0 0 sin  cos 

 cos 2  sin 2  sin  cos 0 0 


 
 sin 2  cos  2
 sin  cos 0 0 
 
T    2 sin  cos

2 cos sin  cos 2   sin 2  0 0 

(5.40)
 0 0 0 cos sin  
 0 0 0 sin  cos 

where  is the angle of principal direction to the local axes system.

5.8 Modeling of Cracked Concrete


The main feature of plain concrete is its low tensile strength compared
with the failure stress in compression. In the finite element methods, two
main approaches have been used for cracked section representation, the
discrete cracked model, and the smeared crack model.

(a) Discrete Crack Model

This approach was first used by Ngo and Scordelis [27] to analyze a
simply supported reinforced concrete beam. In this approach, the cracks
are restricted to occur at the boundary of the elements by separation of
nodal points initially occupying the same position in space, Figure (5.20).
This means that when any crack occurs the topology of the mesh varies.
This makes the analysis expensive. These difficulties have resulted in a
very limited acceptance of this model in the general structural
applications.
129
(b) Smeared Crack Model
In this approach, the cracked concrete is assumed to remain a continuum,
the cracks are smeared out in a continuous fashion. It is assumed that the
concrete becomes orthotropic after first cracking has occurred. In the
smeared-cracking model, a crack is not discrete but implies an infinite
number of parallel fissures across that part of the finite element, Figure
(5.21). After cracking has occurred, the cracked concrete becomes an
orthotropic material and a new relationship must be derived. This is
accomplished by modifying the stiffness of the element, the modulus of
elasticity is reduced to zero in the direction normal to the crack. Further, a
reduced shear modulus is assumed on the cracked plane to account for the
aggregate interlocking.
In the smeared crack simulation, two different models are used for
defining the crack direction. The first is the fixed crack model. In this
model, the crack is fixed. A perpendicular crack plane is allowed if the
tensile stress in this direction exceeds the tensile strength. The direction
of crack is held and fixed at all subsequent time steps.
The second approach is the rotating crack model, in this approach
the cracks are permitted initially to be perpendicular to the principal
tensile stress direction when the stress reaches the specified limiting
value. With further increment of loading the principal stress changes, the
crack is assumed to rotate and orthotropic material axes are set in a new
crack direction. In the present study, the fixed smeared approach is
adopted.
130

( a) (b)
Figure (5.20) Cracking representation in discrete cracking modeling
approach (a) one directional cracking (b) two directional cracking

Figure (5.21) Idealized representation of single crack in the smeared


cracking modeling approach.
5.9 Strain Softening Rule
The first studies done on numerical analysis of reinforced concrete
structures assumed concrete to be an elastic brittle material in tension.
When cracking occurred, the stress normal to the crack direction was
immediately released and dropped to zero. This leads to great
convergence difficulties.
It was then argued that due to the bond forces, cracked concrete
carries between cracks a certain amount of tensile stress normal to the
cracked plane. The concrete adheres to the reinforcing bars and
contributes actively to the overall stiffness of the structure. This effect is
known as tension stiffening. It can be incorporated into the computational
model in two indirect ways:
131
(a) Assuming that the loss of tensile strength in concrete is gradual
after cracking
(b) Modifying the steel stress-strain curve
The first option is equivalent to consider the concrete as an elastic
strain softening material in tension and has been extensively used in
computational analysis of reinforced concrete structures.
One of the computational modeling of the tension stiffening is
based on fitting of the experimental results of bond interaction with
reinforcing steel [10]. Fracture mechanics models have been also used to
model the tension stiffening.
5.9.1 Fracture Models
Fracture mechanics models have been applied to the problem in plain
concrete [3,14]. These studies have shown the attractiveness of using
some fracture energy concepts in the material modeling of concrete. The
fracture energy is the energy needed to separate the two crack surfaces.
Based on these concept the composite damage models have been
proposed for cementitious materials such as concrete[3]. The main
concept borrowed from fracture mechanics to develop these models is the
assumption that the fracture energy release rate, Gf , a material property, is
used rather than the local stress-strain curve. The implementation of ‘G f
=constant’ concept leads to the important conclusion that the local strain
softening law depends on a characteristic length, lc , depending on the
finite element mesh.
The shape of the strain the softening curve is not important than the
assumption used to select the parameter defining such curve. The strain
softening curve must be related to fracture energy of concrete.
The fracture energy is defined

G f    ( w) dw (5.41)
132
Gf represents the energy needed to separate the two crack surfaces.
Typical values of the fracture energy for normal concrete are in the range
of 50 to 200 N/m, and w is the crack width.
The smeared approach does not represent individual cracks, so
crack width, Wc , must be smeared into equivalent crack strain,  c ,
related to the physical crack opening by a characteristic length lc .
Nilsson [28] derived this relation considering a control volume Vc ,

containing a crack with an area Sc.


Wc  (Vc / S c ) c l c  c (5.42)
An exponential function is used to simulate the strain softening
effect, so that:
  E 0  t e  (   t ) /  c (5.43)

E0
where is the elastic Young’s modulus
t is the tensile strain at cracking
c is the softening parameter
 is the nominal tensile strain in the cracked zone.

The softening parameter  c is determined by the evaluation of the


integral in equation (5.41) and by introduction the relation between the
crack opening Wc and fictitious crack strain  c of equation (5.42), this
yields
Gf
c  (5.44)
E0 t l c
133
In the context of finite element computation the control volume for
crack monitoring is the volume associated with a sampling point in a
given element. In the present work the characteristic length is computed
for each sampling point as:
lc  dA
(5.45)
where dA is the area of the concrete represented by the sampling point.
This model was used to analyze the reinforced concrete when using
the brick elements [15,37] and also used for shell elements in analyzing
reinforced concrete structures [6].

5.9.2 Trilinear Tension Stiffening


A trilinear tension stiffening is also adopted in the present work. This
model is produced from simulating the experimental results on reinforced
concrete specimens subjected to unixial tension[11].
A gradual release of the concrete stress component normal to the
crack plane is adopted in this work. Figure (5.23) shows the loading-
unloading process. Unloading and reloading of cracked concrete is
assumed to follow the linear behavior shown with a fictitious elasticity
modulus given by:
Ei   m f t (1   i /  m ) /  i for  t < i < m (5.46)
where (  m ,  m ) are tension stiffening parameters and  i is the maximum
value reached by the tensile strain at the point considered. The normal
stress  1 (and/or  2 ) is obtained by the following expression,
Figure(5.23).
   m f t(1   1 /  m ) t  i  m (5.47)
1
or by  1   i 1   i (5.48)
i
where  1 is the current tensile strain in direction 1
5.10 Shear Modulus of Cracked Concrete
134
The experimental results on cracked concrete indicate that a considerable
amount of shear stress is transferred across the surfaces of cracked
concrete due to the aggregate interlock and the dowel action.
These phenomena can be incorporated to the finite element code by
incorporating to the smeared cracked model an appropriate cracked shear
modulus G C . In the present work, cracked shear modulus is assumed to
be a function of the current tensile strain [10].
For concrete cracked in one direction

G12C  0.25G(1   1 / .004) ; G12C  0 if   .004 (5.49)


C
G13  G12
C
(5.50)
5
G23  G (5.51)
6

where G is the uncracked shear modulus and  1 is the tensile strain in the
1-direction. For concrete cracked in both directions,

G13C  0.25G(1   1 / .004) ; G13C  0 if   .004 (5.52)


C
G 23  0.25G (1   2 / .004) ; C
G23 0 if   .004 (5.53)
C
G12  0.5G13
C
; or C
G12  0.5G23
C
if C
G23  G13
C
(5.54)
If the crack closes the uncracked shear modulus is again assumed in
the corresponding direction.
135

f t

i

ε
c
i

Figure (5.22) Loading-unloading behavior of cracked concrete illustrating


tension softening behavior (Fracture energy model)

Stresses

0.5   m  0.7
 m f t
 m  .002

i Tension

Strains

Compression
t i m

Figure (5.23) Loading-unloading behavior of cracked concrete illustrating


tension stiffening behavior (Trilinear model)
136
5.11 Reinforcing Steel
Reinforcing steel is considered as a homogeneous material, and has a
similar stress-strain relationship in compression and in tension and its
properties are well defined. Steel reinforcement is comparatively thin
being generally assumed to be capable of transmitting axial force only.
Uniaxial stress-strain relationship is sufficient for general use in finite
element code.
A typical stress-strain curve for different qualities of reinforcement
is shown in Figure (5.24)[7].
A bi-linear stress-strain relationship allowing for strain hardening
and elastic-perfect plastic relation and allowing for elastic unloading is
adopted in the present work Figure (5.25).

Figure (5.24) Typical stress-strain curves of steel reinforcement [7]


137
Stress
f y E S
Tension

ES
Strain

Compression
f y

Figure (5.25) Bilinear stress-strain model for steel reinforcement

5.12 Nonlinear Modeling of the Curved Beam Element


(i) General Remarks
Nonlinear analysis has been carried out on reinforced concrete stiffened
shells [1,38]. Two different finite element idealizations were adopted. The
first modeled the stiffeners by the shell elements and the second modeled
the stiffeners by the curved beam elements. The second approach has
been proven to be efficient in simulating the behavior of the reinforced
concrete stiffened shells. In this approach no modification of geometry
was done, in addition, the obtained numerical results agreed better from
the elastic range up to failure [38].
Nonlinear modeling for the curved beam was developed by
Thannon[38] and it is adopted in the present work. Layered formulation
with uniaxial behavior to simulate concrete behavior in cracking, yielding
and crushing was developed. Numerical treatment for modeling the
curved beam was developed and checked by loading the beam in two load
cases applications. The first is by applying a force in the thickness
138
direction perpendicular to the layers direction and the second is by
applying the force in the width direction.
When this model of the curved beam was connected to the
degenerated shell element and applying the compatibility and nonlinear
layered formulation for nonlinear analysis of reinforced concrete stiffened
shells all drawbacks of analyzing of reinforced concrete stiffened shells
were vanished [38].

(ii) Numerical Treatment for Beam Element


The nonlinear model for uniaxial behavior in each layer will be treated
through adjustment of moment about the weak axis M Z and the torque T
by assuming [38]:
1. For uncracked layers, the M Z is retained without reduction
2. For cracked layers three cases can exist:
1. If the crack occurred in the current step, M Z , is reduced to zero for
the calculation of residuals. In the next evaluation of the tangential
stiffness matrix the moment of inertia about the z-axis is reduced to
zero.
2. If the layer has cracked during the previous step, the calculated M Z ,
is already zero and there is no effect on the calculation of residuals.
3. If the crack closes at any load step, the uncracked properties of the
concrete are retained.
An analogous procedure is used for the treatment of torque (T),
using the reduced shear modulus, which reduces the shear modulus for
the cracked concrete layers. It should be noted here that the calculated
value of T is reduced only as a function of the shear retention factor.
For concrete in compression, the same procedure is used
considering yielding in place of cracking. For steel layers, the same
procedure is adopted with respect to yielding of the steel.
139

5.13 Dynamic Constitutive Relations of Concrete


The rate of straining during the standard material tests is of order of
10 5 / sec [2], while during the severe seismic excitation the concrete in the
critical regions of typical reinforced concrete frame wall system is
subjected to compressive strain rate of the order of 10 2 / sec . (as measured
by Sorushian and Alhamad[33]). Much of the work has been summarized
as a state-of the art paper by Bischoff and Perry (1991)[5]. Strain rate and
their conceptual equivalent situation are as in Table(5.1) ( (Bischoff and
Perry 1991)[5]
Table (5.1) Strain rates with the associated situations.

Strain rate
Condition mm/mm (in/in) per sec
Creep 10 8 to 10 6
Static 10 6 to 10 4
Earthquake 10 3 to 10 2
Hard Impact 10 0 to 101 *
Blast 10 2 to 10 3 *
(*) the time application is a fraction of second.
This leads to a study of the effect of rate of loading on the concrete
and to study the concrete properties for these situations, which are
referred to dynamic loads.
The dynamic properties of concrete are measured with respect to
the static properties (strength, modulus of elasticity, strain at maximum
stress in compression, tensile strength, also Poisson’s ratio, or flexural
strength) in terms of strain rate. Dynamic compression tests on confined
and unconfined concrete have shown that the strain rate increases the
compressive strength, secant modulus of elasticity, and the slope of the
descending branch in the concrete stress-strain diagram [33]. The
maximum strain at failure decreases [9,16] while the strain at maximum
140
stress might decrease or increase depending on the rate of straining
[9,33]. Hatano [13] showed that for uniaxial test on concrete an almost
constant failure strain is obtained irrespective of the rate of straining.
The effect of strain rate on both compressive and tensile strength is
studied [20,31,35,36]. These tests showed that the compressive and
tensile strengths increase with increasing the strain rate, and the strain
rate sensitivity at higher strain rate is greater for tension than for the
compression.
Harris et al [11] studied the material properties for mass concrete
tested at strain rates that correspond to seismic (dynamic) and static
loading. Results of this study indicated that dynamic/static strength ratios
are greater than one for both the compressive and tensile strength tests.
The dynamic static ratios of modulus of elasticity and failure strain for
the compressive strength tests are generally less or equal to one.

5.14 Dynamic Modeling of Concrete.


5.14.1 Compression Modeling
Several approaches have been use in modeling the concrete for
dynamic analysis, but mainly two approaches are widely used.
1. Viscoplastic approach [4,15,23].
2. Elasto-plastic approach with strain rate effect [22,32].
In the present study, the second approach is adopted to model the
concrete in compression. The implementation of the strain rate is
followed as discussed below [22]
The compressive behavior of concrete is modeled using the flow
rule of plasticity. It is well known that the elasto/plastic models in their
basic formulation do not meet the requirements of dynamic state, so it
should be modified to incorporate some features of dynamic cases. Owen
and Liu[22] incorporated into the elasto/plastic models current strain rate
141
based on experimental tests, and so the plastic yielding is not only
governed by the plastic history (  P ) but also by the current strain rate (  
)as follows:
F ( ij , yd ( P ,   ))  0 (5.55)
This can be done by modifying the initial and subsequent static yielding
stress  ys ( ) to obtain the dynamic yielding  yd ( ,  ) as [22]:
P P 

 yd ( P ,   )   ys ( P )[1  (  )]

(5.56)
where  (  ) is the a rate function based on experimental data.
a1
  
 ( )  a 

 (5.57)
 s 

The material parameters a and a1 are material dependent. For a concrete


with static strength f c lower than 440 kg/cm2, a  0.02789 , a1  0.330123


and for fc =605 kg/cm2, a  0.011768 , a1  0.38533 .
  s : strain rate value below which no rate effects are evident,
approximately:
  s  10 5 sec 1 (5.58)

The solution algorithm and the essential steps of the


implementation of strain rate for the implicit Newmark algorithm are
described below, with emphasis being given to the operations which are
different from those of a standard elasto-plastic solution scheme [30].
Step (a). Evaluate the current strain rate i  n 1 , the effective current strain
rate i  n 1 and the strain increment  i  n 1 for the ith iteration of ( n  1)th

time step. The strain rate is related to the nodal velocities V n 1


i

according to
i  n 1  BiVn 1

(5.59)
142
where B is the usual strain matrix. The effective current strain rate i  n 1

is expressed as an invariant function of the current strain rate as follows:


1
 2 1 2 2
i  n 1    oct   oct  (5.60)
 4 

where  oct and  oct are, respectively, the octahedral normal strain and the
octahedral shear strain.
Step (b). Compute the incremental elastic stress changes
 i  ne1  D i  n 1 (5.61)
in which D is the usual elastic matrix.
Step (c). Accumulate the total stress for each element at Gauss point
i  ne1  i 1  n 1   i  ne1 (5.62)
and the effective stress, i  ne1 , according to the yield function chosen.
Step (d). Check at each Gauss point whether or not yielding takes place,
?
i  ne1  i  yd   ys ( i 1  np1 )(1   ( i  n 1 )) (5.63)
where the uniaxial dynamic stress i  yd is determined from the current
effective strain rate i  n 1 and the effective plastic strain exiting at the end

of the (i  1)th iteration. If  ne1  i  yd the current stress i  n 1 is set equal

to i  ne1 and the solution process proceeds to step (g).


Step (e). Check whether or not the effective current strain rate is greater
than the previous iteration. The case i  n 1  i 1  n 1 implies an expansion
of the dynamic yield surface. The stress which satisfies the yield
condition F ( i 1  nP1 , i  n 1 )  0 can be found as
i  npre
1  i 1  n 1  (1  R )  i  n 1
e

(5.64)
in which
i  ne1  i  yd
R (5.65)
i  ne1  i 1  n 1
143
if i  n 1  i 1  n 1 , shrinkage of the dynamic surface occurs. In order to find
the stress which satisfies the yielding condition at the current strain rate
i  n 1 , the stress i 1  n 1 is assumed to shrink , so that
i  npre
1  R i 1  n 1
*
(5.66)
where
i  yd
R*  (5.67)
i 1  n 1

Step (f). According to whether i  n 1 is larger or smaller than i 1  n 1 , the


remaining excess portion of stress is, respectively, R i  ne1 or
 i  ni 1  (1  R * ) i 1  n 1 which can be redistributed as described in reference
[30] until both equilibrium conditions and constitutive relations are
satisfied. The incremental elasto-plastic stress and total stress can be
written as
F ( i 1  np1 , i  n 1 )
 i nep1  D i  n 1  D (5.68)


i  n 1  i 1  pre n 1   i  nep1 (5.69)


The effective plastic strain increment at the end of the ith iteration for the
( n  1)th time step can be evaluated by equating the plastic work
increment under the multiaxial stress state to that occurring under
uniaxial conditions
T
 F 
  i  n 1
pre

   (5.70)
i  np1  i 1  np1 
i  yd

Step (g). Finally, evaluate the equivalent nodal forces from the element
stresses according to

i Pn 1   B T i n 1 dV
V

(5.71)
144

5.14.1 Tension Modeling


Experimental results showed the effect of strain rate on the splitting
tensile strength [20,31,35,36]. Splitting tensile strength is sensitive more
than the compressive strength due to the strain rate effect. This effect
appears clearly for high strain rates[36] for impact and blast loads,
Table(5.1).
Experimental results also showed that almost constant failure strain
is obtained irrespective of the rate of straining [11,13]. For static analysis,
the modulus of rupture equation(5.1) or direct tensile strength was used to
characterize the tensile strength for the finite element analysis [10,38].
Modulus of rupture is also used to characterize the tensile strength for the
finite element analysis without including the strain rate [39]. The splitting
tensile strength is also used with a reduction of 40% without including the
strain rate [17].
Strain criterion is used to distinguish the elastic behavior from the
tensile fracture in finite element code [4,15,22]. Such an assumption is
realistic, since uniaxial tests on concrete have shown that an almost
constant failure strain is obtained irrespective of the rate of
straining[11,13]. This approach is adopted in the present study.
145

Reference
1. Arnesen, A. (1979). "Analysis of reinforced concrete shells
considering material and geometric nonlinearities." Division of
Structural Mechanics, Norwegian Institute of Technology, University
of Trondheim, Norway, Report No. 79-1.

2. ASTM standards. (1982). "Standared test method for compressive


strength of cylindrical concrete specimens." ASTM, C 39-81.

3. Bazant, Z. P. and Cedolin, L. (1981). "Fracture mechanics of


reinforced concrete." ASCE, J. Mech. Div., 106, pp. 1207-1306.

4. Beshara, F. B. A., and Virdi, K. S. (1992). "Prediction of dynamic


response of blast loaded reinforced concrete structures." Computers
and Structures, Vol. 44, No ½, pp. 297-313.

5. Bischoff, P. H., and Perry S. H. (1991). "Compressive behavior of


concrete at high strain rate" Material and Structures, Vol. 24, pp. 425-
450.

6. Cervera, M., Abdel Rahman, H. H. and Hinton, E. (1984). "Material


and geometric nonlinear analysis of reinforced concrete plate and shell
system." in Computer Aided Analysis and Design of Concrete
Structures, (Damjanic et al., eds.) 547-563, Pinerdge Press, Swansea.

7. Chen W. F. (1982). "Plasticity in reinforced concrete." McGraw-Hill


Book Company.

8. Chinn, J. and Zimmerman, R. M. (1965). "Behavior of plain


concrete under various high triaxial compression loading conditions."
Technical Report, WLTR 64-163, New Mexico.

9. Dilger, W. H., Koch R., and Kowalczyk, R. (1984). "Ductility of


plain and confined concrete under different strain rates." ACI Journal,
Vol. 81, No. 1, pp. 73-81.
146

10. Figueiras, J. A. (1983). "Ultimate loads analysis of anisotropic and


reinforced concrete plates and shells." Ph.D. Thesis, c/p1/72/83,
Department of Civil Engineering, University Collage of Swansea,
U.K.

11. Harris, D. W., Mohorovic, C. E., Dolen, T. P. (2000). "Dynamic


properties of mass concrete obtained from dam core." ACI Material
Journal, Vol. 97, No. 3.

12. Harsh S., Shen Z., and Darwin, D. (1990). "Strain rate sensitive
behavior of cement paste and mortar in compression." ACI Material
Journal, Vol. 87, No.5.

13. Hatano, T. (1960). "Relation between strength of failure, strain


ability, elastic modulus and failure time of concrete." Central Research
Inst. of Electric Power Industry, C-6104, Tokyo.

14. Hilleborg A., Modeer H., and Petersson, P. E. (1976). "Analysis of


crack formation and crack growth in concrete by means of fracture
mechanics and finite elements." Cement and Concrete Research, 6,
773-781.

15. Hinton, E (1988). "Numerical methods and software for dynamic


analysis of plates and shells." Pineridge Press, Swansea U.K.

16. Hughes, B. P., and Watson, A. J. (1978). "Compressive strength and


ultimate strain of concrete under impact loading." Magazine of
Concrete Research, Vol. 30, No. 105.

17. Inoue, N., Yang, K., and Shibata, A. (1998). "Dynamic nonlinear
analysis of reinforced concrete shear wall by finite element methods
with explicit analytical procedure." Earthquake Engineering and
Structural Dynamics, Vol.26, pp. 967-986.
147
18. Kupfer, H. B., and Grestle, K.H. (1973). " Behaviour of concrete
under biaxial stresses. " ASCE Journal of Engineering Mechanics
Division, Vol. 99, No. EM4, 893-866.

19. Kupfer H., Hilsdorf, K. H., and Rusch H. (1969). "Behavior of


concrete under biaxial stresses." ACI Journal, Vol. 66, No.8, pp. 656-
666.

20. Labmert, D. E., and Ross, C. A. (2000) "Strain rate effect on


dynamic fracture and strength." International Journal of Impact
Engineering, Vol. 24, 985-998.

21. Launay, P, and Gachon, H. (1972). " Strain and ultimate strength of
concrete under triaxial stress." ACI, SP. 34, pap. 13.

22. Liu G.Q., and Owen, D.R.J (1986). "Ultimate load behavior of
reinforced concrete plates and shells under dynamic transient loading."
International Journal for Numerical Methods in Engineering, Vol. 22,
pp. 189-208.

23. Lopez Cela, J.J., Pegon, P., Casadei, F. (1997). "Fast transient
analysis of reinforced concrete structure with Drucker-Prager model
and viscoplastic regularization. " In: D.R.J. Owen, E. Onate, E. Hinton
(eds.), Proceeding of the fifth International Conference on
Computational Plasticity, Fundamental and Applications, Barcelona,
Spain, 17-20.

24. Malvar, L. J., Crawford, J. E., Wesvich, J. W. and Simsons, D.


(1997). "A plasticity concrete material model for DYNA3D"
International Journal for Impact Engineering, Vol. 19, Nos. 9-10, pp.
847-873.

25. Menertrey Ph., and Willam, K.J. (1995) "Triaxial failure criterion
for concrete and its generalization." ACI Journal, Vol. 92, No.3.
148
26. Mills, L. L., and Zimmerman, R. M. (1970)." Compressive strength
of plain concrete under multiaxial loading conditions. " ACI Journal,
Vol.67, No.10, pp. 802-807.

27. Ngo, D., and Scordelis, A. C. (1967). " Finite element analysis of
reinforced concrete beam" ACI Journal, Vol. 64, 152-163.

28. Nilsson, L. and Oldenburg, M. (1982)."Nonlinear wave propagation


in plastic fracturing materials–a constitutive modeling and finite
element analysis" IUTAM Symp. On "Nonlinear deformation waves".

29. Ottosen, N.S. (1977). "Failure criterion for concrete." ASCE, Vol.,
103, No. EM4.

30. Owen D.R.J. and E. Hinton, (1980) "Finite element in plasticity."


Pineridge Press, Swansea, U.K.

31. Ross C. A., Tedesco, J. W., and Kuennen, S.T. (1995). "Effect of
strain rate on concrete strength" ACI Journal, Vol.92, No.1, pp. 37-47.

32. Shimazaki, K. and Wada, A. (1998). "Dynamic analysis of a


reinforced concrete shear wall with strain rate effect." ACI structural
Journal, Vol. 95, No.5.

33. Soroushian, P. Choi K., Alhamad, A. (1986). "Dynamic constitutive


behavior of concrete. " ACI Journal, No.83, 251-259.

34. Steven, D. J., Krauthammer, T., and Chandra, D. (1991) " Analysis
of blast loaded buried RC arch response." Journal of Structural
Engineering, ASCE, 117 (1), 213-234.

35. Tedesco, J.W, Ross, C. A., and Brunair, R. M. (1989)." Numerical


analysis of dynamic split cylinder tests. " Computers and Structures,
Vol. 32, No ¾, pp.609-624.
149
36. Tedesco J. W., Ross C. A., and Kuennen S. T. (1993). "Experimantal
and numerical analysis of high strain rate splitting tensile tests" ACI
Material Journal, Vol. 90, No.2.

37. Thannon A. Y. (1993). "Nonlinear three dimensional finite element


analysis of reinforced concrete curved beams." Al-Rrafidain
Engineering, Vol.1, No.2.

38. Thannon A., Y, (1988). " Ultimate load analysis of reinforced


concrete stiffened shells and folded slabs used in architectural
structures." Ph.D. Thesis, University of Wales, Swansea, U.K.

39. Thannon, W. A. (1994) "Inelastic dynamic analysis of concrete


frames of material engineering under non-nuclear blast loading."
Ph.D. Thesis, University of Roorkee, India.

40. Wastiels, J. " (1981). "Failure criterion for concrete subjected to


multiaxial stress." Lecture held at University of Illinois at Chicago
circle Department of Material Engineering.

41. Willam, K. J. (1984)." Experimental and computational aspects of


concrete fracture." in "Computer Aided Analysis and Design of
Concrete Structures" (Damjanic et al., eds.) Pinerdge Press, Swansea,
U.K.

Anda mungkin juga menyukai