Documente Academic
Documente Profesional
Documente Cultură
Advisor(s) Lo, SH
URL http://hdl.handle.net/10722/192857
YAO Zhong
(姚钟)
M.PHIL. THESIS
by
YAO Zhong
(姚钟)
April 2013
Abstract of thesis entitled
Submitted by
YAO Zhong
i
A nonlinear finite element program was developed, in which the above-
mentioned nonlinear effects have all been included in modeling the reinforced
concrete structures. The nonlinear equations of equilibrium are solved using an
incremental-iterative technique performed under displacement control. The
validity of the model including the confinement effect of secondary
reinforcements has been examined by analyzing three reinforced concrete beams.
The performance of the numerical model was assessed by comparing results with
those from available experimental data.
(253 words)
ii
DECLARATION
I declare that this thesis entitled “Nonlinear finite element analysis of reinforced
concrete beams” represents my own work, except where due acknowledgement is
made, and that it has not been previously included in a thesis, dissertation or
report submitted to this University or to any other institution for a degree, diploma
or other qualification.
Signed ______________________
YAO Zhong
iii
ACKNOWLEDGEMENT
Special thanks are also due to my dearest friends Mr. L. L. Gang, Mr. F.
Fu, and Ms. M. L. Chen for their valuable support and encouragement during my
depressing time.
The last but not the least, words cannot express the author’s immense
gratitude to his beloved parents. They have provided great support over the whole
period of my study. I gained from them a lot of courage and energy which are
always the sources of my progress.
iv
TABLE OF CONTENTS
ABSTRACT i
DECLARATION iii
ACKNOWLEDGEMENT iv
TABLE OF CONTENTS v
LIST OF FIGURES ix
LIST OF TABLES xi
CHAPTER 1 INTRODUCTION 1
1.1 General 1
2.1 Introduction 5
3.1 Introduction 46
3.5 Summary 51
4.1 Introduction 52
4.5 Summary 66
5.1 Introduction 67
5.3.1 Introduction 73
5.3.2 Formulation of displacement control method 74
5.6 Summary 80
REFERENCES 107
viii
List of Figures
ix
Figure 5.1 Full Newton-Raphson method 72
Figure 5.2 Modified Newton-Raphson method 72
Figure 5.3 Initial stress method 73
Figure 5.4 Typical nonlinear structural response 76
Figure 5.5 Flowchat for displacement control technique 77
Figure 6.1 Details of beam OA-1 (Bresler and Scordelis, 1963) 83
Figure 6.2 Details of beam A-1 (Bresler and Scordelis, 1963) 83
Figure 6.3 Details of beam J-4 (Burns and Siess, 1962) 84
Figure 6.4 Finite element mesh used 86
Figure 6.5 Load-deflection curves of beam OA-1 and A-1 87
Figure 6.6 Crack pattern of beam OA-1 89
Figure 6.7 Crack pattern of beam A-1 90
Figure 6.8 The load-deflection curves of beam OA-1 and beam A-1 92
Figure 6.9 Concrete finite element mesh for beam OA-1 94
Figure 6.10 Comparison of load-deflection curves of beam OA-1 (fine and coarse
mesh) 95
Figure 6.11 Comparison of load-deflection curves of beam J-4 to study
convergence tolerance 99
Figure 6.12 Comparison of load-deflection curves of beam J-4 to study tension
stiffening 100
Figure 6.13 Load-deflection curves of beam J-4 under different conditions 101
x
List of Tables
Table 6.2 Ultimate loads and the corresponding deflections of beam OA-1 and
A-1 86
Table 6.3 The ultimate loads of beam J-4 under different combinations 98
xi
CHAPTER 1
INTRODUCTION
1.1 General
1. Realistic material models for concrete and reinforcing steel and their
interactions.
2. Efficient and reliable solution technique (the method used to solve the
nonlinear problem).
3. Reasonable selection of numerical parameters in nonlinear solution
process.
1
Disparities in experimental results are often observed due to difficulties in
obtaining consistent test procedures and test specimens. Furthermore, this is also
due to the natural variability of concrete itself. Concrete is actually a composite
material composed of mortar, aggregates, and voids; therefore its properties vary
widely depending on the mixing conditions, placing and curing. It exhibits highly
nonlinear properties even under relatively low loads due to various factors
including cracking, environmental effects.
The finite element method has made possible the inclusion of highly
complex material behavior of reinforced concrete structures into analytical
solutions. Thus the finite element method has become a powerful computational
tool for the reinforced concrete structures. For examples cracking, softening and
crushing of concrete, multiaxial stress response of concrete, yielding of
reinforcement, bond and dowel action between steel and concrete, can be taken
into account by using the finite element method to analyze reinforced concrete
structures. Displacement- based finite element method was adopted in this thesis.
2
The performance of the numerical model was assessed by comparing
results with those from available experimental data. In this thesis, the behaviors of
different reinforced concrete beams have been studied.
1. To understand the material models for concrete and reinforcing steel and
the interactions between them.
There are seven chapters in this thesis and they are organized as follows:
3
In Chapter 2, a detailed survey of available literature on the experimental
behavior of concrete is presented and a discussion of the material models used for
concrete under biaxial states of stress is given. The modeling of cracking and
tension stiffening effect are also introduced.
4
CHAPTER 2
MODELING OF CONCRETE
2.1 Introduction
It is the purpose of this chapter to present the concrete material model used
in this research for nonlinear finite element analysis of concrete structures under
general biaxial stress states. First, the mechanical behavior of concrete under
different short term loading conditions is discussed in Section 2.2. A brief review
of concrete models is presented in Section 2.3. Section 2.4 introduces the concrete
material model adopted in this study. Last, a brief summary is given in Section 2.5.
5
water react to produce a cementing gel which bonds the fine and coarse
aggregates into a rock-like material. As a result, the strength of concrete is a
function of the strength of the cement, the aggregates, and the interaction between
the components.
6
2.2.1 Uniaxial loading behavior
Between about 30% and 50% of the ultimate stress, the cracks begin to
propagate at a very slow rate. It is observed that most of the crack growth
concentrates in the interfacial region between the aggregate and mortar, which is
the weakest link in normal strength concrete (Scrivener and Pratt 1987). At this
stage, the crack propagation is stable because the available internal energy is
almost balanced by the required crack release energy. The stress-strain behavior
of concrete begins to show increasing nonlinear properties.
Once the stress exceeds about 50% of the ultimate stress, the propagation
of cracks is not only remaining in the interfacial region, but also begins to extend
to the cement matrix. The newly generated cracks in the matrix begin to connect
the originally isolated bond cracks and to form a much more continuous crack
system.
7
At peak stress, the mesoscopic crack patterns finally grow into
macroscopic cracks which dominate the behavior after peak stress (Kotsovos
1983). A softening post-failure range follows the peak stress and finally ends with
a complete crushing curve.
8
Figure 2.2 Stress-strain behavior of concrete under uniaxial compression
9
2.2.2 Biaxial loading behavior
3. The strength under biaxial tension is almost the same as under uniaxial
tension.
10
According to Kupfer et al. (1969), a failure surface of concrete under
biaxial states of stress is presented in Figure 2.7. A few conclusions drawn from
this envelop can be summarized as follows:
11
Figure 2.4 Stress-strain relationships of concrete under biaxial compression
(Kupfer 1969)
12
Figure 2.6 Stress-strain relationships of concrete under biaxial tension (Kupfer
1969)
13
2.3 Review of Numerical Models
The elasticity based models are those models whose constitutive relations
are deduced from the elasticity theory. Early constitutive models were based on
isotropic elasticity to represent concrete behavior (Ngo and Scordelis 1967). The
stress-strain relationship is linear. For Cauchy elastic models, the current state of
stress is uniquely expressed as a function of the current state of strain as
σ = F (ε ) (2.1)
where F is the elastic-response function of material and σ and ε are stress and
strain tensors, respectively.
14
Nonlinear elastic models are needed to describe a material exhibits nonlinear
properties like concrete.
∂W
σ= = S ε (ε )ε (2.2)
∂ε
dσ = S (σ )dε (2.3)
where the variable tangent stiffness, S, is a function of the current state of stress.
15
Various special, simplified forms of the hypo-elastic formulation have
been developed and utilized for modeling concrete behavior. Some of these
models assume the behavior of concrete to be incrementally isotropic, i.e. the
effect of stress-induced anisotropy is neglected (Gerstle 1981). Other models
include a particular type of stress-induced anisotropy, since concrete is
represented as an orthotropic material. One of the most widely used orthotropic
hypo-elastic models is that proposed by Darwin and Pecknold (1977) which has
been used to represent the behavior of concrete under cyclic loading. The model is
based on the concept of "equivalent uniaxial strain", whereby the effect of biaxial
stresses on the internal damage in concrete is represented by equivalent stress-
strain curves for each of the principal stress axes.
The strong points of elasticity based models are their reasonably good
representation of nonlinear concrete behavior and their conceptual simplicity in
comparison with other existing models.
16
Many plasticity based models have been developed, e.g. Chen and Chen
(1975), Han and Chen (1987) and Chan et al. (1994). These models may
accurately predict inelastic deformation, dilatancy, and hydrostatic pressure
sensitivity, but material softening, and, degradation of stiffness and hysteretic
behavior of concrete under cyclic loading may not be represented satisfactorily.
Since a decrease of all components of stress is impossible for strictly plastic
behavior satisfying Drucker's stability postulate, models which combine plasticity
theories and fracturing concepts have been proposed to account for softening.
Bazant and his coworkers (1976) have extended the endochronic theory to
concrete structures with great success. The concept of the endochronic theory
consists in characterizing the inelastic strain accumulation by a certain scalar
parameter z, called intrinsic time, whose increment is a function of the
deformation strain increments and the real time increment. The endochronic
model appears to be capable of modeling aspects of the behavior of concrete
material, which were difficult to model by other constitutive models, such as, (a)
the hydrostatic pressure sensitivity of inelastic strain, (b) the inelastic dilatancy
due to shear straining, (c) the strain softening tendency at high stress. By virtue of
the last extension (c), the theory at the same time provides the failure criterion, in
which dependence on strain and stress histories is automatically accounted for.
The presence of real and intrinsic time in the constitutive relations also allows
simultaneous consideration of strain rate effects and nonlinear long-time creep.
17
2.4 Concrete Model Used in This Study
σ1 E1 υ E1 E 2 0 ε 1
1 υ E E ε
σ 2 =
E2 0
2
(2.4)
σ 1 − υ
2 1 2
12 0
0
1
4
( )
β E1 + E 2 − 2υ E1 E 2 ε 12
where σ 1 and σ 2 are stresses along two orthotropic axis 1 and 2 respectively; ε 1
and ε 2 are strains along two orthotropic axis 1 and 2 respectively; σ 12 and ε 12
are engineering shear stress and strain respectively; E1 and E 2 are the secant
stiffness, along the two orthotropic axis 1 and 2 respectively; υ is Poisson’s ratio
assumed as a constant. β is the parameter accounting for aggregate interlock,
termed of shear retention factor, will be discussed in detail later. It should be kept
in mind that E1 and E 2 are depending on the state of stress.
It was assumed for the model that stresses in the principal directions could
be calculated independently of each other, based on uniaxial stress-strain
18
relationships. The biaxial effect was assumed to be due to the interaction of the
two principal directions through the Poisson's ratio effect. Therefore, the concept
of “equivalent uniaxial strain” should be introduced in order to separate the
Poisson effect from the cumulative strain. In addition, equivalent uniaxial strain
can made a contribution to keep track of the degradation of stiffness and strength
of plain concrete and to allow actual biaxial stress-strain curves to be derived from
“uniaxial” curves.
ε iu E 0
σi = (2.5)
E ε ε
1 + ( 0 − 2) iu + ( iu ) 2
ES ε ic ε ic
i; σ ic and ε ic are the peak stress and strain at peak stress respectively. It should be
noted that σ ic and ε ic are functions of the state of stress, have to be determined.
19
In tension, a complete stress-strain curve of concrete can be treated
similarly to one in compression. Two equations continued at peak are suggested to
describe the ascending and descending branches, respectively (Guo and Zhang
1987). In this investigation, the Saenz’s model, which is used to describe the
concrete behavior under compression, is used to describe the ascending branch of
the stress-strain curve under tension. The post-crack behavior of concrete under
tension will be discussed in detail later.
strains ε ic are functions of the biaxial principal stress ratio, α . Darvin and
Pecknold (1977) gave expressions for the peak stresses in terms of α . For the
corresponding strains ε ic , expressions were obtained from the limiting tensile
strain criterion.
The biaxial strength envelope is divided into three regions, which depends
on the state of stresses, as represented by the biaxial principal stress ratio α . The
expressions for the peak stresses of the three regions of the strength envelope (T-T
zone, T-C zone, and C-C zone) are summarized as follows:
For biaxial compression, Kupfer and Gerstle found that the strength
envelope was closely approximated by the equation
σ1 σ2 σ2 σ1
( + )2 − − 3.65 =0 (2.6)
fc fc fc fc
21
Equation (2.6) can be rewritten to give the maximum compressive strength
of the concrete σ 2c , in the form of α and f c .
1 + 3.65α
σ 2c = fc (2.7)
(1 + α ) 2
σ 1c = α σ 2c (2.8)
The values of σ 1c and σ 2c are used to define the shape of the equivalent
where ε 0 c is the strain corresponding to the peak stress at the stress-strain curve
under uniaxial compression.
22
σ2
σ 1t = (1 - 0.8 ) ft (2.11)
fc
αf c f t
σ 1t = (2.12)
αf − 0.8 f t
1 + 3.28α
σ 2c = fc (2.13)
(1 + α ) 2
1+ α
ε 1c = A + B × 10 −3 (2.14)
0 .8 − m α
where
0.5136 m
A= (2.15)
0 .8 + m
0.6336 + 0.15m
B= (2.16)
0 .8 + m
− fc
m= (2.17)
ft
23
(iii) Tension-Tension region
σ 1t = f t (2.18)
σ 2t = σ 1t / α (2.19)
24
Several mathematical models have been proposed for the descending curve
of the uniaxial stress-strain relationship of concrete in compression by different
researchers (Hognestad 1951; Saenz 1964; CEB-FIP 1993). A simple approach to
include the falling branch was adopted in the present work. Beyond the peak
compressive stress, the stress-strain relationship is assumed linear, proposed by
Kent and Park (1971), as shown in Figure 2.9.
In this model, Kent and Park suggest the slope of the descending part, u ,
can be calculated as follows,
0.50
u= (2.20)
3 + 0.29 | f c |
− ε 0c
145 | f c | −1000
Figure 2.9 The stress-strain relationship beyond the peak compressive stress
25
2.4.4 Modeling of concrete cracking
The maximum stress and strain theories are frequently used to determine
whether tensile cracking has occurred in the concrete (Chen 1982). If the
maximum principal stress or strain in a point of the structure reaches the uniaxial
tensile strength or tensile strain limit, cracking is assumed to form perpendicular
to the direction of the maximum tensile stress or strain. The stress in that direction
is subsequently reduced to zero.
The limiting tensile stress value which causes the crack is not a well-
defined quantity. For specimens cast from the same concrete, the flexural tensile
strength determined from a modulus of rupture test is higher than the tensile
strength of a split cylinder, which is in turn higher than the tensile strength
obtained from a direct tension test. Furthermore, for each type of test there is
26
significant scatter in the results. For concrete structures subjected to rapid loading,
the maximum strain criterion is more realistic, since uniaxial dynamic tensile tests
(Hatano 1960) indicate that an almost constant failure strain is observed
irrespective of the strain rate or loading rate. The limiting tensile strain criterion
has been employed with success to represent the tensile cracking of concrete
under static (Chen and Chang 1980) loading.
The detailed explanation of limiting tensile strain criterion has been given
in previous section.
Cracks are represented in the finite element analysis in two different ways:
the discrete and smeared crack representations. The choice between these two
models depends on the purpose of the study. If bond slip, dowel shear effects and
aggregate interlocking across cracks, giving realistic crack patterns are of main
interest of the research, the discrete approach is convenient, since it can account
better for these. However, in cases where the overall behavior of a structure is
desired or where many cracks have formed, the smeared crack concept is
preferable.
27
interlock and dowel action. The discrete crack approach was further improved and
partially automated by Al Mahaidi (1979) who used a predefined crack utilizing
two nodes at one point connected by a linkage element.
Despite all these improvements, this approach has the disadvantage that it
involves changes in the topology of the finite element mesh following the
formation of a crack and the lack of generality in possible crack direction. Due to
these difficulties and restrictions the discrete crack models are not favored in
finite element analysis for general structural application.
The smeared crack approach which was proposed by Rashid (1968), offers
a more practical alternative for crack representation while using the finite element
method.
However, the smeared crack models also have appeared some arguments.
For example, smeared cracking representation is very sensitive to mesh
discretization, and also to the order of integration rule adopted, since the element
28
stiffness is determined from the contributions associated with each cracked or
uncracked integration point.
After a crack has formed its direction can be taken as fixed or can be free
to rotate. In the fixed crack approach the crack direction is defined by the
orientation of the initial crack and is then held fixed regardless of the change in
the principal directions with the changing loading.
The rotating crack approach was originally proposed by Cope et al. (1980).
This model is based on the assumption that the crack direction is always normal to
the direction of the maximum tensile principal stress. Milford and Schnobrich
(1985) stated that the cracks defined by the rotating crack model are not cracks in
the strict sense, but rather notational cracks defining the average crack rotation.
29
Figure 2.10 Formation of secondary crack in singly cracked concrete
30
(iii) Closing of cracks
In conclusion, concrete can crack along one or two directions, and cracks
are allowed to open or close in the present research. Figure 2.11 shows different
possible crack configurations in this research.
Many different shaped curves have been proposed. Scanlon (1971) used a
stepped piecewise linear unloading relation. A similar approach was used by Lin
and Scordelis (1975), with a smooth unloading curve. Figure 2.12(a) and (b)
shows these two alternatives stress-strain relations for concrete in tension.
31
Figure 2.11 Possible crack configurations
32
(a) Scanlon’s stepped unloading curve (Scanlon 1971)
(c) Modified stress-strain diagram for reinforcing steel (Gilbert and Warner 1978)
33
Another approach to modeling tension stiffening effect is to ignore the
concrete after cracking and increase the reinforcing steel stiffness fictitiously
(Gilbert and Warner 1978), as shown in Figure 2.12 (c). The concrete is assumed
to carry no stress normal to a crack, but an additional stress will be added to the
reinforcing steel. This approach implies that the tension stiffening effect is
concentrated at the reinforcing steel level, whereas experimental observations
showed that this effect is distributed throughout the depth of the tension zone.
The inclusion of tension stiffening effects does not only simulate the real
behavior of concrete, but it also improves the stability of the numerical analysis.
However, there are still some arguments with this tension stiffening effect. For
example, this effect, if explained in terms of bond interaction with the reinforcing
steel, cannot be applied to plain concrete structures, or to concrete located at a
certain distance from the reinforcement. But it is well known that plain concrete is
not a perfectly brittle material that it has some residual load carrying capacity after
reaching the tensile strength. Therefore, the concept of “tension softening” is
introduced. This can be accomplished by adding a descending branch to the stress-
strain curve.
ε
ft
εt
σ= (2.21)
ε ε
0.312 f t ( − 1)1.7 +
2
εt εt
where σ and ε is the stress and strain in concrete respectively; f t is the tensile
strength of concrete and ε t is the strain of concrete at tensile strength. And this
model is shown in Fig. 2.13.
34
To summarize, the post-peak stress-strain relationship of cracked concrete
is dependent on the tension softening behavior of plain concrete and the bond
force transferred from the reinforcement. Therefore, in this research, a
modification of the model of tension softening effect is made to take the tension
stiffening effect into account. This will be discussed in detail in Chapter 4.
Figure 2.13 Tension softening model of concrete (Guo and Zhang, 1987)
When the crack is formed and subjected to shear forces, the crack
interfaces tend to override each other as a result of the shear slip. This tendency
may be constrained either by surrounding a portion of concrete or by the
reinforcing bars crossing the cracks. It has been shown experimentally that the
contact area between aggregate particles and cement paste plays a prominent role
in the shear transfer mechanism of aggregate interlock, as shown in Figure 2.14.
35
That is to say, increasing crack width will lead to reduction of the contact area,
which results in smaller shear stiffness.
A reduced shear modulus which depends on the crack width seems more
realistic and was adopted in this research. A simple parabolic expression to
calculate the shear retention factor β , is adopted as the follows (He 1999):
2
ε - ε 1c
β = 0.2 1 - ≤ 0.2 (2.22)
ε 0c
a measure of the crack width. β decreases as the crack width increases and is
always larger than 0.
36
Figure 2.14 Shear transferring by aggregate interlock
Figure 2.15 Comparison of different expressions proposed for the reduced shear
stiffness of concrete
37
2.4.5 The computational procedure of this concrete model
2. Apply a load increment of ∆{P}n from which the incremental strains ∆{ε }n
and stresses ∆{σ }n are calculated.
{ε }n = {ε }n -1 + ∆{ε }n (2.23)
{σ }n = {σ }n -1 + ∆{σ }n (2.24)
1 2τ xy
αc = tan −1 ( ) (2.25)
2 σ xx − σ yy
This angle is shown in Figure 2.16, and it varies from -45° to +45°.
(b) Calculate the total principal stresses {σ p }n and total principal strain {ε p }n
{ε p }n = [T ] {ε }n (2.26)
[ ]
{σ p }n = T −1 {σ }n
T
(2.27)
38
in which [T ] is a transformation matrix defines by:
principal strain. Fix angle α c as the permanent angle of crack; and calculate the
new stresses in the direction perpendicular to the crack on the principal axis using
the tension stiffening equation, which will be introduced in Chapter 4. As a result,
ε i and σ i represents the strain and stress in the direction perpendicular to the
crack can be obtained.
(d) Calculate the stress in the direction parallel to the crack σ j and the assumed
σ j = E jε j (2.29)
and
τ = β Gγ (2.30)
where j refers to the direction parallel to the crack and γ is the calculated shear
strain value. The updated stresses can then be written as:
σi
{σ c }n = σ j (2.31)
τ
39
and
σ j
{σ c }n = σ i (2.32)
τ
{σ }n = [T ] {σ c }n
T
(2.33)
(f) Steps (c) to (e) are also used to check whether two cracks have occurred at the
same load level and if so stresses will be updated accordingly in both directions.
(g) For the stiffness calculation in subsequent iterations [Dc ] is set equal to:
0 0 0
[Dc ] = 0 E2 0 (2.34)
0 0 βG
when a crack occurs perpendicular to axis (1) in Figure 2.16, or is set equal to:
E1 0 0
[Dc ] = 0 0 0 (2.35)
0 0 β G
perpendicular to each other. The angle α c is fixed as before and the new stiffness
matrix is updated to:
40
0 0 0
[Dc ] = 0 0 0 (2.36)
0 0 β G
E1 and E2 are the secant modulus of elasticity of concrete in the direction parallel
to the single crack, and β is a shear factor defining the shear resistance parallel to
the crack due to aggregate interlocking. The value of β can be determined by
Equation (2.22).
41
5. If cracks already exist then:
(a) Use the fixed angle of crack α c and transform the Cartesian strains and
stresses into the crack direction:
{ε c }n = [T ] {ε }n (2.37)
[ ]
{σ c }n = T −1 {σ }n
T
(2.38)
(b) If one crack exists then check whether the strain perpendicular to the crack is
changed from the descending to the ascending part of the stress-strain curve, then
update the stress on the principal angle from
{σ c }n = [Dc ] {ε c }n (2.39)
where
E1 0 0
[Dc ] = 0 E2 0 (2.40)
0 0 βG
{σ }n = [T ] {σ c }n
T
(2.41)
0 0 0
[Dc ] = 0 E2 0 (2.42)
0 0 βG
42
when the change of strain occurs perpendicular to axis (1) in Figure 2.16, or is set
equal to:
E1 0 0
[Dc ] = 0 0 0 (2.43)
0 0 β G
when the change of strain occurs perpendicular to axis (2) in Figure 2.16.
This criterion is used instead of using a criterion for crack closing, and
when the strains become negative then the secant modulus for the compression
zone is used.
(c) To check whether a crack exists in second direction the same procedure for
checking a crack in the first direction is used except that [Dc ] for the stiffness
calculation is set up so that:
0 0 0
[Dc ] = 0 0 0 (2.44)
0 0 β G
If a second crack already exists, then check if the strain in any single
direction or both directions changes from the descending to the ascending part of
the curve. Then calculate the stresses using:
{σ c }n = [Dc ] {ε c }n (2.45)
where
0 0 0
[Dc ] = 0 E2 0 (2.46)
0 0 βG
if the strain changes on axis (1) in Figure 2.16, or is set equal to:
43
E1 0 0
[Dc ] = 0 0 0 (2.47)
0 0 β G
E1 0 0
[Dc ] = 0 E2 0 (2.48)
0 0 βG
{σ }n = [T ] {σ c }n
T
(2.49)
2.5 Summary
2. The limiting tensile strain failure criterion defines concrete fails when the
tensile strain of concrete exceeds the limiting tensile strain. Cracking occur
perpendicular to the direction of the maximum tensile strain.
44
3. Doubly-crack model based on smear crack representation is incorporated
in this research. Concrete can crack along one or two directions. Moreover,
cracks are allowed to open or close.
4. The tension stiffening and shear retention are taken into consideration in
this model.
45
CHAPTER 3
3.1 Introduction
46
3.2 Mechanical Properties of Reinforcing Steel in Static
Loading Conditions
47
Figure 3.1 Typical stress-strain curves for reinforcing steel (Chen 1982)
If the modulus of the strain hardening, E sh , was set equal to zero, the
stress-strain relationship became an elastic-plastic one. Loading in tension and
compression was assumed to be elastic until the stress reaches the yield stress, σ y ,
beyond that yielding was assumed to have occurred. When the strain of the
reinforcing steel reached the ultimate strain ε u , failure was assumed to occur.
Then the reinforcing steel was assumed to rupture and carry zero stress.
49
Figure 3.3 Idealized stress-strain curve for reinforcing steel used in this study
3.5 Summary
In this chapter, a brief introduction of the mechanical properties of
reinforcing steel is given in the first place. Next in this chapter, the stress-strain
relation and the finite element representation of reinforcing steel adopted in this
thesis are presented.
51
CHAPTER 4
4.1 Introduction
The behavior of plain concrete and reinforcing steel bars have been
introduced and modeled separately in the previous chapters. However, reinforced
concrete structures are the structures that composed of these two materials, and
exhibit tremendous different properties due to the complex interactions between
the reinforcing steel and the surrounding concrete. These interactions can be
summarized as three aspects:
In this chapter, these three interactions between reinforcing steel and the
surrounding concrete will be discussed in detail.
52
4.2 Modeling Bond Action
4.2.1 Introduction
Another way to account for bond action is to use a gradual stiffening curve
for concrete after cracking has occurred, and is usually termed of "tension
stiffening". This approach is based on the fact that if the opening of a crack occurs
at the same time as bond failure, it will cause some movement between the bar
and the concrete. This will then cause the shear force at the contact surface
between the cracks to feed tension stresses into the concrete. The concrete,
attached to the bar will contribute to the overall stiffness of the system. This
53
contribution is accounted for by the gradual stiffening curve described in this
chapter. On the other hand, an alternative way of representing this stiffening effect
is to increase the steel stiffness and stress.
Ngo and Scordelis (1967) proposed the earliest work on modeling of bond
in finite element analysis of reinforced concrete. They introduced a linkage
element between concrete and steel to represent bond. According to Ngo and
Scordelis, the linkage element can be thought of conceptually as consisting of two
linear springs parallel to a set of orthogonal axes H and V, as shown in Figure 4.1.
Each of the springs is assigned a stiffness value from which the stiffness matrix
for the linkage element is obtained. If the springs in the H and V directions have
stiffness k h and k v respectively then the stress strain relation is given by:
σ h k h 0 ε h
σ = 0 k v ε v
(4.1)
v
in which ε h and ε v are the relative displacements between points I and J in the H
and V directions.
Nilson (1968) pointed out that the relationship between bond stress and
bond slip is strictly nonlinear. He introduced a bond-slip equation which is
derived indirectly from experiments reported by Bresler and Bertero (1966). A
third order degree polynomial relating local bond stress µ to local bond slip d is
given as:
54
µ = 3606 × 10 3 d − 5356 × 10 6 d 2 + 1986 × 10 9 d 3 (4.2)
Bond stress can be thought as the shearing stress between reinforcing bar
and the surrounding concrete. Bond stress-slip relationship is strictly nonlinear
and it varies with the position along the bar. Experimental results of bond stress-
slip relationship are presented based on average or local values. To include such a
relation in analysis by finite element method normally an average bond stress-slip
relationship is used.
CEB FIP Model Code (2000) suggests an analytical bond stress-slip model
under monotonic loading, which is largely based on the model developed by
Eligehausen et al. (1983), as shown in Figure 4.3. In Figure 4.3, τ mac is bond
strength, s1 , s 2 , s3 are the characteristic slip value related to bond slip behavior.
The shape of this curve is determined by the main influencing factors for bond-
slip behavior: roughness of the bar surface,confinement, bond condition, and
concrete strength.
55
Figure 4.1 Linkage Element to Represent Bond (Ngo and Scordelis 1967)
56
Figure 4.3 Analytical bond stress-slip relationship suggested in CEB FIP
57
Figure 4.4 shows the physical situation in the vicinity of a crack in a
reinforced concrete tension member. It indicates that at a crack the full load is
shared between steel and concrete. This ability of concrete between cracks to
share the tensile load with the reinforcement is called “tension stiffening”. As the
load is increased and the stress in the concrete between cracks reaches the ultimate
strength, the concrete will then rupture and a further crack forms between the
main cracks. Therefore, the average concrete stress over the cracked region will
progressively decrease with loading.
58
4.2.4.1 Model of tension stiffening by increasing the stiffness of reinforcing
steel after cracking
σ1
= (1 − β b )e −λ (ε −ε
t 1 1c )
+ βb (4.3)
σ 1c
ρ
β b = 75 s (4.4)
db
270
λt = (4.5)
βb
where σ 1c and ε 1c are concrete stress and strain when cracking, σ 1 and ε 1 are
stress and strain in the principal direction normal to the crack, λt is a parameter
controls the rate at which the post-peak stress-strain response decays to a limiting
59
value, ρ s is reinforcement ratio , d b is the diameter of steel bar, β b is a
parameter for the bond action of reinforcing bars crossing cracks, which reflects
the influence of amount and distribution of reinforcement.
Equation (4.3) shows the average tensile stress in the concrete reduces
exponentially from tensile stress at cracking σ 1c to a limiting value of β b σ 1c at
large tensile strains. In order to coincide with the tension softening equation
adopted in this thesis, according to He (1999), the following modification is made,
ε1
σ1 ε 1c
= (1 − β b ) + βb (4.6)
σ 1c 2 ε1 ε1
0.312 f t ( − 1) +
1.7
ε 1c ε 1c
60
However, He and Kwan (2001) argued that dowel action can play an important
role if the other contributions to the shear resisting mechanisms become
insignificant. This may happen, for example, in the case of a beam with a small
amount of web reinforcement, or during the post-peak loading stages when, due to
widening of cracks, the contribution of aggregate interlock may decrease rapidly.
Dowel action may contribute significantly to the post-peak resistance and hence
contribute to the shear ductility of concrete members.
3. Since the dowel action is usually more significant near peak load and
at the post-peak stage, experimental testing or theoretical analysis
extending into the post-peak range are needed to investigate the full
61
effects of the dowel action, but such testing and analysis are generally
quite difficult.
Though the modeling of the dowel action in the finite element analysis
with smeared cracks can be accomplished by making some assumptions, the
dowel action is simply neglected in this thesis.
62
Figure 4.6 Internal forces in a cracked beam
63
Clearly it is important to be able to quantify these effects of confinement on the
stress-strain behavior of concrete.
The stress-strain model of Kent and Park (1971) for concrete confined by
rectangular transverse reinforcement neglected the increase in concrete strength
but took into account the increase in ductility due to rectangular confining steel.
More recently, Scott et al. (1982) and Park et al. (1982) modified the Kent and
Park (1971) stress-strain equations to take into account the enhancement of both
the concrete strength and ductility due to confinement and the effect of strain rate.
Monotonic stress-strain equations for concrete confined by rectangular-shaped
transverse reinforcement include those proposed by Vellenas et al. (1977) and
Sheikh and Uzumeri (1980). Park and Leslie (1977), Ahmad and Shah (1982) also
proposed various stress-strain relationships for confined concrete under
monotonic loading.
f cc = K f c (4.7)
ε 0 cc = K 2 ε 0c (4.8)
fy
K = 1 + a( ρ w )b (4.9)
fc
b are the empirical coefficients which are determined by the hoop patterns. The
values of a and b suggested by Kappos are listed as follows,
65
4.5 Summary
66
CHAPTER 5
5.1 Introduction
67
Many researchers have devoted for the development of efficient solution
algorithms for nonlinear problems. Some powerful alternatives to the traditional
Newton-Raphson and Modified-Newton-Raphson methods have been suggested.
These include the Quasi-Newton algorithm proposed by Matthies and Strang
(1979), the Secant-Newton methods advocated by Crisfied (1982), and the Arc-
length methods introduced by Riks (1979).
This Chapter describes the method of analysis used in the present work
and briefly introduces some other related methods. First, a brief description of the
incremental and iterative methods which are usually adopted in nonlinear analyses
is given in Section 5.2. Then a displacement control technique used in this study
to ensure the solution can be advanced in the post-peak range is introduced in
Section 5.3. Next in this chapter (Section 5.4), a discussion of the convergence
problems is given and residual load criterion is adopted in this study. In Section
5.5, criteria to define the collapse of the structure are suggested. Last in Section
5.6, a brief summary is given.
where
{R} is the residual force vector,
{F }= ∫VB T σ dv is the internal force vector,
{P} is the applied load vector,
{δ } is the unknown displacement vector, which expressed as follows,
68
{δ } = [k (σ , ε )]−1 {P} (5.2)
The satisfaction of equilibrium at the nodes requires that the external load
vector equals to the internal load vector, that is to say, the residual force vector
should be zero.
69
The first step is to calculate an initial estimate of the incremental
displacement {∆δ }1 , then additional corrective displacements {∆δ }i are obtained
n n
by iterations. (Note that superscript n denotes the load increment number while
subscript i corresponds to the current iteration number.)
70
equations. To overcome this difficulty, a modification to the full Newton-Raphson
algorithm is made.
This method was first advocated by Zienkiewicz and his coworkers for the
solution of elastic-plastic problems. The method can be thought of as a
modification to the Newton-Raphson algorithm where the initial elastic stiffness
matrix is maintained during the entire analysis, see Figure 5.3. This has the
immediate advantage of significantly reducing the computing cost per iterations,
however it shows less efficient than the previous Newton-Raphson methods
because the convergence rate is too slow.
71
Figure 5.1 Full Newton-Raphson method
72
Figure 5.3 Initial stress method
5.3.1 Introduction
Consider the structural response shown in Figure 5.4, the load control
technique can be used to trace the ascending part of the response before the limit
point, and obviously fails to trace the falling part of the response. The arc length
73
approach and the displacement control techniques can be used to trace the
nonlinear response in the analysis of such problems. Although the arc length
method has been used by many investigators for nonlinear problems owing to its
ability to handle snap-through and snap-back problems, and its improved
convergence characteristics over the load control, it has been found to fail for
some problems involving the nonlinear analysis of reinforced concrete structures.
This is likely due to the fact that the arc length method may pick up unstable
positions along the falling part of the response.
Instead of making the problem more difficult, the arc length method is not
incorporated into this work. A displacement control technique similar to that
proposed by Batoz and Dhatt (1979) was adopted in the analysis. In this technique
an incremental displacement component is specified and the corresponding load
becomes one of the unknowns.
in which the applied force vector {P} can be expressed in terms of an load factor
λ and a generalized force vector {P}* due to a unit load, such that
where
and
∆δ ib = ki−1 {P}
*
(5.11b)
∆δ iq (λi + ∆λi ) is zero, and the qth row of Equation (5.10) can be written as
λi +1 = λi + ∆λi (5.14)
75
The main operations performed in the displacement control method are
shown in Figure 5.5. At the beginning of each increment a load factor λ is set
equal to zero and the incremental displacement component is assigned a preset
value. The incremental load factor was obtained from Equation (5.13) and hence
the incremental displacement vector can be determined from Equation (5.10). At
the end of each iterations, the load factor λ was updated by using Equation (5.14),
also the nodal displacements were then updated using Equation (5.4) and the
process was repeated until convergence occurred.
76
Form load vector due to unit load {P}
*
Convergence
Yes
No
where, during the iterative process, the progress of the approximate solution is
referred to specified convergence criteria. In the solution strategy, the solution
obtained at the end of each iterations is checked to see whether it has converged
within a preset tolerance. The criteria used are set so that the iterative process is
terminated when it is considered that additional iterations would not improve the
accuracy of the solution significantly. Thus, it is important to include reliable
criteria which will terminate the iterative process when convergence to the desired
accuracy has been achieved.
The convergence criteria, usually used for nonlinear structural analysis, are
based on (i) displacements, (ii) out-of balance forces, and (iii) internal energy. In
this thesis, a force convergence criterion is adopted.
It is difficult and expensive to check the decay of residual forces for every
degree of freedom thus some overall evaluation is preferable. This is achieved by
using norms. The residual load convergence criterion is that
R i
× 100% ≤ FTL (5.17)
λP *
i
where i
is the norm at iteration i, and R i = {R}Ti {R}i
78
and FTL is a specific convergence tolerance. The convergence tolerance must be
realistic. If the convergence tolerance is too loose, inaccurate results are obtained
and if the tolerance is too tight, much expensive effort is spent to obtain needless
accuracy.
In this work, the program is stopped when any of the following occur:
3. The strain of reinforcing steel bars exceeds the ultimate strain permitted.
79
5.6 Summary
This chapter introduces the nonlinear solution techniques adopted in this
thesis. They can be summarized as follows:
3. The convergence of the iterative schemes can be checked using the force
convergence criteria, the selection of the convergence tolerance will be
discussed in Chapter 6.
80
CHAPTER 6
6.1 Introduction
In the previous chapters efforts have been made to discuss various models
for the finite element analysis of reinforced concrete structures. In this chapter the
computer program developed in this study is critically examined by carrying out
the numerical solution of a number of reinforced concrete structures. The results
of the analysis are compared with results obtained experimentally.
81
6.2 Classic Beam Specimens
82
Figure 6.1 Details of beam OA-1 (Bresler and Scordelis, 1963)
83
Figure 6.3 Details of beam J-4 (Burns and Siess, 1962)
84
6.3 Analysis of Beams
Beam OA-1 and beam A-1 are regarded as a classic test series, and have
been widely used as benchmark data for verifying the finite element models for
reinforced concrete. The details of these two beams are discussed in Section 6.2.
The finite element mesh boundary condition, and loading used in the analysis are
shown in Figure 6.4. Because of symmetry, only half of the beam will be
considered in this study.
The load-deflection curves of these two beams derived from the nonlinear
finite element analysis are shown in Figure 6.5. A good agreement with the
experimental results is achieved throughout the entire load-deflection range.
The ultimate loads and the corresponding mid-span deflections of the two
beams are listed in Table 6.2. The failure load of beam OA-1 tested by Bresler and
Scordelis is 333.6 kN with the corresponding mid-span deflection of 6.6 mm. The
numerical prediction of failure load in this study is 309 kN with the corresponding
mid-span deflection of 6.9 mm. And the failure load of beam A-1 tested by
Bresler and Scordelis is 467 kN with the corresponding mid-span deflection of
13.5 mm. The numerical prediction of failure load in this study is 502 kN with the
corresponding mid-span deflection of 14.7 mm. All the numerical results obtained
show a good agreement with that of the experimental study.
85
Figure 6.4 Finite element mesh used
Table 6.2 Ultimate loads and the corresponding deflections of beam OA-1 and A-
1
86
400
350
300
250
Load (kN)
Experimental result
200
Analytical result
150
100
50
0
0 2 4 6 8
Mid-span deflection (mm)
600
500
400
Load (kN)
Experimental result
300
Analytical resullt
200
100
0
0 5 10 15 20
Mid-span deflection (mm)
87
The finite element simulation can also successfully model the development
of cracks and explain the failure pattern of the reinforced concrete beams. The
predicted development of cracks of beams OA-1 and A-1 are shown in Figure 6.6
and Figure 6.7, respectively. These load levels represent the most important stages
of behavior. First at a relatively low load level, for example 100kN for beam OA-
1, the typical flexural cracks would develop at the bottom of the beam running
towards the loading point. Then the diagonal tension cracks would appear, usually
in the middle third of the overall beam depth and at various sections along the
span. These diagonal cracks extended both upwards and downwards with further
increase in load and form the “critical diagonal tension crack.” The critical cracks
formed at a load of 267kN for both beams OA-1 and A-1 according to the
experimental data. The predicted crack patterns of beam OA-1 and A-1 at this
load level are shown in Figure 6.6 (b) and Figure 6.7 (b), respectively.
Beam OA-1, which has no transverse reinforcement, fails shortly after the
formation of the critical diagonal crack. The failures occurred as a result of
longitudinal splitting in the compression zone near the load point, and also by
horizontal splitting along the tensile reinforcement near the end of the beam. The
crack pattern at failure load is shown in Figure 6.6 (c).
88
(a) Crack pattern of beam OA-1 at load= 100kN
89
(a) Crack pattern of beam A-1 at load= 180kN
90
6.4 Parametric Study
Scordelis (1972), Gerstle (1981), Nilson et al. (1978) and Crisfield (1982)
made a comprehensive review of various finite element methods for analyzing
reinforced concrete structures. In their work, many numerical parameters such as
aggregate interlocking, tension stiffening, and dowel action have been studied.
Although much progress has been made in the analysis of reinforced concrete
structures, a lot of the numerical parameters have still not been studied in
sufficient detail.
91
600
500
400
Load (kN)
Beam OA-1
300
Beam A-1
200
100
0
0 5 10 15 20
Mid-span deflection (mm)
Figure 6.8 The load-deflection curves of beam OA-1 and beam A-1
Comparison of the deflections of beam OA-1 with beam A-1 indicates the
effect of transverse reinforcement on the deflections. It is obvious that the
ascending parts of these two beams are almost identical for low levels of load; this
is because for low levels of strain in concrete, the stress state in the transverse
reinforcements is very small and the concrete is basically not confined.
Conversely, when the stress in the concrete gets close to its uniaxial compression
strength, progressive internal cracking occurs and the transverse strain increases
remarkably. At this stage, transverse reinforcement produces a confinement action
which opposes the expansion of the concrete core. Thus, the specimen with
transverse reinforcement can continue to take external loads to a relatively high
level, and shows a much ductile load-deflection performance.
92
6.4.2 Comparison study on finite element mesh size effect
Finite element mesh size is one of the solution parameters, which can
significantly influence the finite element solution process. The effect of finite
element mesh size is analyzed based on beam OA-1. In order to study the effects
of finite element mesh size on the analytical results, two different mesh
configurations were investigated, as shown in Figure 6.9.
The analytical results with both fine mesh and coarse mesh and the
experimental results are shown in Figure 6.10. The numerical prediction of the
failure load of beam OA-1 with fine mesh has been discussed in section 6.3. It
shows a good agreement with the experimental results. However, for the coarse
alignment, the predicted failure load is 34.7% higher than that of the experimental
data. This is because when the element size is sufficiently small, the tensile strains
and stresses in the adjacent element just ahead of the crack tip could become
sufficiently large, and the crack could propagate under a sufficiently small load.
Consequently, the smaller the element size, the lower the predicted structural
capacity would be; the larger the element size, the higher the predicted structural
capacity would be.
93
(a) Fine mesh (48 elements)
94
500
450
400
350
Experimental result
300
Load (kN)
100
50
0
0 2 4 6 8
Mid-span deflection (mm)
Figure 6.10 Comparison of load-deflection curves of beam OA-1 (fine and coarse
mesh)
95
Beam J-4 is an under-reinforced beam with longitudinal steel ratio equals
to 0.99%, which is used in this comparison study. And the following conditions
remain the same throughout the analysis:
First see Figure 6.11, load-deflection curves are compared for convergence
tolerances of 10% and 5%. It is observed that no matter tension stiffening effect is
considered or not, the predicted ultimate loads of beam J-4 with convergence
tolerance= 10% are higher than those with convergence tolerance= 5%. For
example, when tension stiffening is considered, the predicted ultimate load with
tolerance of 10% is 13.6% higher than that with tolerance of 5%. This is because
using a high convergence tolerance is a method of maintaining a certain
proportion of residual forces in the structure as this slows down the rate of
cracking. Thus the predicted ultimate load increases. It is also observed that when
tension stiffening is used, a high convergence tolerance of 10% overestimate the
ultimate load, and a low convergence tolerance of 5% underestimate the ultimate
load. This shows again that convergence tolerance can influence the numerical
results significantly.
Then see Figure 6.12, when the convergence tolerance set to be identical,
the predicted ultimate loads with tension stiffening effect are higher than those
96
without tension stiffening effect. For example, when the convergence tolerance is
set to be 5%, the predicted ultimate load with tension stiffening effect is 12.5%
higher than that when neglecting tension stiffening effect. This phenomenon is
quite similar with the influence of convergence tolerance. That is to say, a high
convergence tolerance and consideration of tension stiffening can both lead to an
overestimation of the stiffness and the ultimate load. Therefore, a conclusion that
tension stiffening has the same effect of retaining residual forces which can lead
to an overestimation of the stiffness and the ultimate load can be drawn.
97
Ultimate load (kN)
Analytical Experimental
Table 6.3 The ultimate loads of beam J-4 under different combinations
98
180
160
140
120
Load (kN)
60
40
20
0
0 2 4 6 8 10 12
Mid-span deflection (m m )
200
180
160
140
120
Load (kN)
TS, FTL=10%
100 TS, FTL=5%
Experimental
80
60
40
20
0
0 2 4 6 8 10 12
Mid-span deflection (mm)
99
200
180
160
140
120
Load (kN)
No TS, FTL=10%
100 TS, FTL=10%
Experimental
80
60
40
20
0
0 2 4 6 8 10 12
Mid-span deflection (mm)
180
160
140
120
Load (kN)
100 No TS,FTL=5%
TS, FTL=5%
80 Experimental
60
40
20
0
0 2 4 6 8 10 12
Mid-span deflection (mm)
100
200
180
160
140
Experimental
120
Load (kN)
No TS, FTL=10%
100 No TS,FTL=5%
TS, FTL=5%
80
TS, FTL=10%
60
40
20
0
0 5 10 15
Mid-span deflection (mm)
6.5 Summary
In this chapter, the models discussed in the previous chapters, are validated
by analyzing three reinforced concrete beams. The good agreement between
analytical and experimental results demonstrates that the developed finite element
code can provide reliable predictions of the load-deflection response and the crack
pattern of reinforced concrete structures. Then a parametric study is conducted in
order to understand how confinement, finite element mesh size, tension stiffening,
and convergence tolerance influence the numerical results.
101
The findings from parametric study can be summarized as follows:
102
CHAPTER 7
7.1 Summary
103
The nonlinear equations of equilibrium are solved by using incremental-
iterative techniques. The iterative technique used in this study is modified
Newton-Raphson method, and displacement control increment method is used.
7.2 Conclusion
Based on the results and discussions in the previous chapter, the following
conclusions can be made:
2. Comparison study on the finite element mesh size effects was carried out.
Two different mesh configurations were investigated, one is a fine mesh
and the other is a coarse mesh. The comparison of the analytical results
obtained from these two meshes demonstrates that finite element mesh
size is one of the most important parameters which can influence the finite
element solution results. The fine mesh is capable of obtaining a much
better solution than the coarse mesh.
104
be found. Satisfactory results can be obtained when the tension stiffening
effects were incorporated in the computational model, however it is very
interrelated to the convergence tolerance used in the study.
105
5. The adopted computational model is only suitable for monotonic loading.
Further improvement should be made to extend this model to include the
analysis of reinforced concrete structures under cyclic loading.
106
REFERENCES
Ahmad, S. H., and Shah, S. P. (1982). “Complete Triaxial Stress-Strain Curves for
Concrete.” Journal of the Structural Division, ASCE, 108(4), 728-742.
Andenaes, E., Girstie, K., and Ko, H. Y. (1977). “Response of Mortar and
Concrete to Biaxial Compression.” Journal of Engineering Mechanics Division,
ASCE, 103(4), 515-526.
107
Beshara, F. B.A. (1991). “Nonlinear Finite Element Analysis Reinforced Concrete
Structures Subjected to Blast Loading.” PhD thesis, City University, London, UK.
Carino, N. J., and Slate, F. O. (1976). “Limiting Tensile Strain Criterion for
Failure of Concrete.” ACI Journal, 73(3), 160-165.
CEB-FIP. (1993). CEB-FIP Model Code 1990: CEB Bulletin d'Information 213-
214, Thomas Telford Service Ltd., London, England.
Cedolin, L., and Dei Poli, S. (1977). “Finite Element Studies of Shear Critical
Reinforced Concrete Beams.” Journal of the Engineering Mechanics Division,
ASCE, 103(3), 395-410.
108
Chen, A. C. T., and Chen, W. F. (1975). “Constitutive Relations for Concrete.”
Journal of Engineering Mechanics Division, ASCE, 101(4), 465-481.
Chen, W. F., Suzuki, H., and Chang, T. Y. P. (1980). “End Effects of Pressure
Resistant Concrete Shells.” Journal of the Structural Division, ASCE, 106(4),
751-771.
Darwin, D., and Pecknold, D. A. (1977). “Nonlinear Biaxial Stress-Strain Law for
Concrete.” Journal of Engineering Mechanics Division, ASCE, 103(EM2), 229-
241.
Eligehausen, R., Popov, E. P., and Bertero, V. V. (1983). “Local Bond Stress-Slip
Relationships of Deformed Bars under Generalized Excitations.” UCB/ EERC-83/
23, University of California, Berkeley, USA.
109
Evans, R. J., and Pister, K. S. (1966). “Constitutive Equations for a Class of
Nonlinear Elastic Solids.” Solids and Structures, 23(6), 751-767.
Gilbert, R.I. and Warner, R.F. (1978). “Tension Stiffening in Reinforced Concrete
Slabs.” Journal of the structural division, 104(12), 1885–1900.
Hoshikuma, J., Kawashima, K., Nagaya, K., and Taylor, W. (1997). “Stress-strain
Model for Confined Reinforced Concrete in Bridge Piers.” Journal of Structural
Engineering, 123(5), 624-633.
111
Hughes, B. P., and Chapman, G. P. (1966). “The Deformation of Concrete and
Microconcrete in Compression and Tension with Particular Reference to
Aggregate Size.” Magazine of Concrete Research, 18(54), 19-24.
Kent, D. C., and Park, R. (1971). “Flexural Members with Confined Concrete.”
Journal of the Structural Division, ASCE, 97(7), 1969-1989.
Kupfer, H., Hilsdorf, H. K., and Rusch H. (1969). “Behavior of Concrete under
Biaxial Stresses.” ACI Journal, 66(8), 656-666.
Kwan, A. K. H., Ho, J. C. M., and Pam, H. J. (2002). “Flexural Strength and
Ductility of Reinforced Concrete Beams.” Structures and Buildings, 152(4), 361-
369.
112
Mainstone, R. J. (1975). “Properties of Materials at High Rates of Straining or
Loading.” Material and Structures (Paris), 8(44), 102-116.
113
Park, R., Priestley, M. J. N., and Gill, W. D. (1982). “Ductility of Square-
Confined Concrete Columns.” Journal of the Structural Division, ASCE, 108(4),
929-950.
Richart, F. E., Brandtzaeg, A., and Brown, R. L. (1928). “A Study of the Failure
of Concrete under Combined Compressive Stresses.” University of Illinois
Engineering Experimental Station, Bulletin No. 185, Illinois, USA.
114
Scanlon, A., (1971). “Time Dependent Deflections of Reinforced Concrete
Slabs.” PhD thesis, University of Alberta, Edmonton, Canada.
115
Vellenas, J., Bertero, V. V., and Popov, E. P. (1977). “Concrete Confined by
Rectangular Hoops Subjected to Axial Loads.” Report 77/13, Earthquake
Engineering Research Center, University of California, Berkeley, USA.
Yin, W. S., Su, E. C. M., Mansur, M. A., and Hsu, T. T. C. (1989). “ Biaxial Tests
of Plain and Fiber Concrete.” ACI Material Journal, 86(3), 236-243.
Zhu, R. R., Hsu, T. T. and Lee, J. Y. (2001). “Rational Shear Modulus for
Smeared-crack Analysis of Reinforced Concrete.” ACI Structural Journal, 98(4),
443-450.
116