Thesis PDF
Thesis PDF
Thesis PDF
INSTITUTE OF ENGINEERING
by
Himal Kafle
A THESIS
LALITPUR, NEPAL
NOVEMBER, 2014
COPYRIGHT©
The author has agreed that the library, Department of Civil Engineering, Institute of
Engineering, Central Campus, Pulchowk may make this thesis freely available for
inspection. Moreover, the author has agreed that permission for extensive copying of
this thesis for scholarly purpose may be granted by professor(s) who supervised the
thesis work recorded here in or, in their absence, by the Head of Department of
concerning M.Sc. Program coordinator or the Dean of the Institute wherein the thesis
work was done. It is understood that the recognition will be given to the author of this
thesis and to the Department Of Civil Engineering, Institute of Engineering, Central
Campus, Pulchowk in any use of the material of thesis. Copying or publication or the
other use of the thesis for financial gain without approval of Department Of Civil
Engineering, Institute of Engineering, Central Campus, Pulchowk and author’s
written permission is prohibited.
Request for permission to copy or to make any other use of material in this thesis in
whole or in part should be addressed to
Head
Institute of Engineering
Lalitpur, Nepal
2
TRIBHUVAN UNIVERSITY
INSTITUTE OF ENGINEERING
PULCHOWK CAMPUS
The undersigned certify that they have read, and recommended to the Institute of
Engineering for acceptance, a thesis entitled “Constitutive modeling of plain
concrete in multiaxial tensile loading utilizing damage mechanics theory”
submitted by Mr. Himal Kafle (069/MSS/105) in partial fulfillment of the
requirements for the degree of Master of Science in Structural Engineering.
…………………………….
Department of Civil
Engineering
Institute Of
Engineering
……………………………….
………………………………...
3
ACKNOWLEDGEMENT
I would also like to thank Prof. Dr. Hikmat Raj Joshie, Prof. Dr. Prem Nath Maskey,
Prof. Dr. Hariram Parajuli, Dr. Rajan Suwal, Er.Nabin Chandra Sharma, Er. Sidhharth
Sankar, Er. Alin Chandra Shakya for their valuable guidance during my course work.
Lastly all concerned staff of M.Sc. program in structural engineering also deserves
thanks for their help.
HIMAL KAFLE
069/MSS/105
4
ABSTRACT
5
TABLE OF CONTENTS
COPYRIGHT©………………………………………………………………………..2
CERTIFICATE………………………………………………………………………..3
ACKNOWLEGEMENT………………………………………………………………4
ABSTRACT…………………………………………………………………………...5
TABLE OF CONTENTS……………………………………………………………...6
LIST OF TABLES…………………………………………………………………….8
LIST OF FIGURES……………………………………………………………………9
LIST OF ABBREVIATIONS AND SYMBOLS…………………………………….10
CHAPTER 1 INTRODUCTION………………………………………………….12
1.1 Background ................................................................. ……………………..12
1.2 Rationale of study.......................................................................................... 13
1.3 Objectives of the present research ................................................................. 13
1.4 Methodology ................................................................................................. 13
1.5 Software to be used: ...................................................................................... 14
1.6 Organization of thesis.................................................................................... 14
CHAPTER 2 LITERATURE REVIEW…………………………………………15
2.1 Mechanical Behavior of Concrete ..................................................................... 15
2.1.1 Uniaxial Compression ................................................................................. 16
2.1.2. Uniaxial Tension ........................................................................................ 19
2.1.3 Biaxial Compression ................................................................................... 21
2.1.4 Triaxial Compression .................................................................................. 22
2.2 Constitutive Modeling of concrete .................................................................. 24
2.2.1 Michael Ortiz (1984, Brown University) .................................................... 24
2.2.2 V.S. Gopalratnam and Surendra P. Shah (1985, University of Missouri,
NorthwesternUniversity)...................................................................................... 24
2.2.3 W.A.M. Brekelmans P.J.M. Schreurs and J.H.P. de Vree (1991, Eindhoven
Netherlands). ........................................................................................................ 25
2.2.4 R. Raveendra Babu, Gurmail S. Benipal and Arbinda K. Singh ( IIT Delhi
,IITGuwahiti). ...................................................................................................... 25
2.2.5 K.R. Wu ,W.Yao (Tongji Univercity ) And Z.J. Li (Hongkong University)
.............................................................................................................................. 25
2.2.6 Graham Baker (2005, University of Southern Queensland) ....................... 25
2.2.7. Kondo,Djemedo and Welemane,Helene and Cormery Fabrice(2007) ...... 26
CHAPTER: 3 THERMODYNAMICS OF MATERIAL…………………...27
3.1 Thermodynamic potential and property: ............................................................ 27
6
3.2 Entropy:.............................................................................................................. 27
3.3 Enthalpy and Free Energy: ............................................................................... 28
3.4 First law of thermodynamics ............................................................................. 30
3.5 Second Law (Clausius-Duhem inequality) ........................................................ 31
7
LIST OF TABLES
Table 5.1 Material parameters are taken for the calculations: (S. Yazdani)…………40
Table B.2 Experimental data for uniaxial tension (Gopalratnam and Shah(1985)…..64
8
LIST OF FIGURES
Figure 2.2 Typical stress-volumetric strain curve for concrete in uniaxial compression
(data from Kupfer et al., 1969)………………………………………………………18
Figure 2.3 Typical tensile stress-strain curve for concrete (Hughes and Chapman,
1966)………………………………………………………………………………….19
Figure 2.4 Stress-strain relationships of concrete under biaxial compression(Kupfer et
al.,1969)………………………………………………………………………………22
Figure 2.5 Stress-strain curves of concrete in triaxial compression (Balmer, 1949)...23
Figure 2.6. Schematic representation of pressure-volumetric strain curve for
concrete………………………………………………………………………………24
Figure 4.1 Schematic representation of stress-strain behavior of a brittle solid in
tension………………………………………………………………………………..37
Figure 5.1 Theoretical Stress-strain curve of plain concrete under uniaxial
tension………………………………………………………………………………..41
Figure 5.2 Stress-strain curve for plain concrete under Biaxial tension.……………42
Figure 5.3 Stress-strain curve for plain concrete under Triaxial tension…………43
Figure 5.4 Stress-strain curve for plain concrete under Triaxial tension…………….44
Figure 5.5 Stress-strain curve for plain concrete under multiaxial tensile
loading………………………………………………………………………………..45
Figure 5.6 Comparison of theoretical and experimental axial stress-strain curve for
plain concrete under uniaxial tension………………………………………………...46
Figure 5.7 Comparision of proposed and experimental axial stress-strain curve for
plain concrete under uniaxial tension………………………………………………...47
9
LIST OF ABBREVIATIONS AND SYMBOLS
SYMBOLS DESCRIPTIONS
σ Stress tensor
ԑ Strain tensor
theoretical model
function
C Compliance tensor
s Specific entropy
10
u Specific internal energy
V Thermodynamic substate
𝜃 Temperature
𝜏𝑗 Thermodynamic Tensions
h Enthalpy
g Gibbs function
concrete
: Tensor contraction
⊗ Tensor multiplication
proposed model
µ* Damage parameter
11
CHAPTER 1
INTRODUCTION
1.1 Background
Concrete is most commonly used man made construction material. Concrete is getting
more popularity due to simple fact that except cement all ingredients of concrete are
commonly available local materials like aggregate and water. Therefore concrete is
being used as construction material from small pavements to run-ways and express-
ways, from small hutments to multi-storied buildings, from small culverts to long
multi-span bridges. Concrete is a non-homogeneous mixture of coarse aggregates,
sand and hydrated cement paste. The behavior of concrete under mechanical loads is
affected by the development, growth and coalescence of microcracks leading to the
formation and propagation of microcracks eventually to rupture. Classical fracture
mechanics is unable to predict crack initiation appropriately while crack propagation
can only be described when crack-path is known beforehand. Continuum damage
mechanics is the constitutive theory that describes the progressive loss of material
integrity due to the propagation and coalescence of microcracks, microvoids and
similar defects. These changes in the microstructure lead to a degradation of the
material stiffness observed on the macroscale.
The continuum damage approach, first elaborated by Kachanov (1958) using effective
stress concept for modeling creep rupture, has been used extensively by many authors
(Ortiz,1985; Simo and Ju 1987; Yazdani and Schreyer,1988; Yazdani,1993; Yazdani
and Karnawat,1997; Geers et al, 2000; Jirasek and Patzak,2002; Khan et al,2007 etc.)
to describe progressive deterioration of mechanical properties of brittle material due
to developments of cracks and microcracks. It has been argued that a successful
utilization of damage theories as an engineering tool for the analysis of complex
structural system would enhance our capability for a more reliable design.
12
theories, in this research it is used for constitutive modeling of plain concrete in
multiaxial tensile loading.
1.4 Methodology
Following procedures are adopted to accomplish the objectives of this thesis work:
14
CHAPTER 2
LITERATURE REVIEW
15
2.1.1 Uniaxial Compression
Stress-strain curves for concrete in monotonic and cyclic compressive loadings are
shown in Figure 1. The stress-strain curve is highly nonlinear in compression. The
reason behind this nonlinearity in stress-strain curve is the development and growth of
cracks during the application of compressive load. Based on the crack formation and
failure, this stress-strain curve is divided into four regions. These four major stages in
the development of microcracking and failure are briefly explained below.
Figure 2.1 Stress-strain behavior of concrete under monotonic and cyclic compressive
loading (Bahn and Hsu, 1998).
16
remain unchanged. In this stress range, the available internal energy is approximately
equal to the required crack release energy. The final lengths of cracks will remain
unchanged, if the stress level is maintained constant. In other words, under constant
applied stress, no further growth of cracks is observed. This means the crack
propagation is stable in this stress range.
3. Range of stable mortar cracks: This region lies between 50 to 75% of f c′in the
uniaxial compressive stress-strain curve. Bond cracks propagate through the mortar.
Moreover, some of the new bond cracks are also formed slowly. The propagation of
mortar cracks is approximately stable. Under constant applied load, cracks continue to
grow and slowly reach their final lengths with no further growth. In this range, the
load increment is fully utilized in the development and growth of cracks.
4. Range of unstable crack propagation: When the stress is above about 75% of f c′
,
cracks extended in the direction of applied compressive stress gradually become
wider and wider. The rate of crack propagation also increases. As the extended cracks
attain their critical lengths, the system becomes unstable. The available internal
energy is greater than the required crack release energy. After this stress range,
complete failure of concrete may take place even though the loading is maintained
constant.
In uniaxial compression tests of concrete, microcracks developed are approximately
parallel to the applied load. These extended cracks are mainly responsible for the
progressive failure of concrete. The nonlinear ascending portion of the stress-strain
curve is called the strain hardening regime, whereas the descending part of the curve
indicates the strain softening regime. The peak in the curve represents the maximum
compressive strength, f c′
.
The initial Poisson’s ratio of concrete, which is usually taken as 0.2 in uniaxial
compression, varies with the monotonically increasing compressive loads. Shah and
Chandra (1968) reported that concrete, under monotonically increasing uniaxial
compression, attains a certain stress level at which its Poisson’s ratio begins to
increase continuously and considerably. This stress level is called initiation stress.
Another phenomenological behavior of concrete in uniaxial compression test is shown
in Figure 2.This figure shows the relationship between axial stress and volumetric
strain in uniaxial compression test of concrete. Within stress level up to about 75 to
90% of f c′
, the volumetric strain (ε v ) of concrete is nearly linear. Maximum
17
compaction is attained at this stress level. As the concrete specimen is loaded above
this stress level, the lateral strain abruptly increases due to the widening of the
extended cracks, and the volume
Actually, cracks are not exactly parallel to the axis of compressive loading. They are
inclined at certain angle with the loading axis. The inclined bond cracks are mainly
due to the irregular shapes of the aggregates used in concrete. Due to the inclination
of these pre-existed bond cracks, shear stress components of the applied stress are
developed. The components of these shear stresses are mainly responsible for shear
sliding and the surface separation in two sides of the crack. The phenomena of shear
18
sliding and surface separation completely destroy the aggregate-mortar interface, and
the interface cracks start to penetrate through the mortar. With the addition of further
load in concrete, the propagation of cracks follows the preferred cleavage planes and
cracks become parallel to the axis of loading.
Figure2.3: Typical tensile stress-strain curve for concrete (Hughes and Chapman,
1966).
19
The stress level of about 60% of the uniaxial tensile strength, f t′
, up to which the
existing microcracks are passive, can be regarded as the limit of elasticity in tension.
Also, the new microcracks are not initiated in this stress range. Above 60% of stress
level, the growth of bond cracks is observed. From 60 to 75% of f t′
, the crack
propagation is stable. The interval of stable crack propagation in uniaxial tension is
found to be shorter than that in uniaxial compression because of its lower crack
arresting capacity. Unstable crack propagation begins, when the stress level is just
above 75% of f t′
. Due to the rapid propagation of unstable cracks, it becomes very
difficult to present the experimental data in the descending portion of the tensile
stress-strain curve.
Some other experimentally observed behaviors of concrete under loading are
the existence of permanent deformation, strain softening beyond the peak stress,
stiffness degradation, both in tension and in compression. The main causes of the
irreversible strains are the formation of microcracks and the flow of aggregates within
the concrete material. The descending part of the stress-strain curve represents the
strain softening behavior. Strain softening is regarded as a decrease in strength with
the gradual increase in strain after the peak strength value is reached. The stiffness
degradation, which is more pronounced after post-peak range, is mainly due to the
formation of cracks and microcracks (i.e. damage) in concrete. However, due to the
crack closure effect, stiffness of concrete degraded in tensile loading will be
recovered in reversed compressive loading. In this tension-compression load reversal,
cracks opened due to tensile loading will be closed by the reversed compressive
loading, and stiffness recovery happens.
21
1.4 σ/f'c
-σ1/-σ2 = -1/-0.52
1.2
1 -σ1/-σ2 = -1/-1
-σ1/-σ2 = -1/0
0.8
0.6
f'c = -4.757 Ksi
0.4 εc = -0.0022
0.2
Tensile strain Compressive strain
0
-2 -1 0 1 2
Strain, ε/εc
22
72
60
Pressure
48
Axial stress (Ksi)
20 Ksi
15 Ksi
36
7.5 Ksi
2.5 Ksi
24 1 Ksi
0 Ksi
Hydrostat
12
0
0 0.01 0.02 0.03 0.04 0.05 0.06
Axial strain
However, concrete under pure hydrostatic pressure possesses different behavior than
that observed in triaxial compression. The possibility of cracks to open is negligible in
hydrostatic compression, and thus completely obscures the process of crack
formation. Therefore, under hydrostatic compression, no damage is assumed to occur
in concrete due to the passive microcrack field. The above explanations are supported
by the experimental work of Resende (1987) where the unloading curve in pure
hydrostatic compression path is shown to be parallel to the initial modulus.
Another phenomenological behavior of concrete to be discussed is the shear enhanced
compaction. Hardening of concrete during loading is mainly due to pore compaction.
Void closure in concrete is caused both by shear stress and by hydrostatic pressure.
However, the shear stress present at the material point is more effective in closing the
voids than that of hydrostat. This means the shear stress governs the compaction
behavior of concrete. Figure 6 shows the schematic representation of pressure-
volumetric strain curve for concrete under hydrostatic pressure and uniaxial strain (in
the presence of shear stress). It is seen from the figure that pressure required to
compact concrete to a given volumetric strain in the presence of shear stress (i.e.,
23
uniaxial strain state) is lower than that in the absence of shear stress (i.e., hydrostatic
pressure). This behavior is also found in most geological materials.
HYDROSTATIC
COMPRESSION
UNIAXIAL
STRAIN
εv
Figure 2.6 Schematic representation of pressure-volumetric strain curve for concrete
Michael Ortiz proposed a general theory for the inelasticity of concrete, the main
constituents being new, rate independent model of distributed damage for mortar and
the application of mixture theories for the composite nature of concrete. The proposed
theory of damage was capable of accommodating fully anisotropic elastic degradation
both in tension and compression in a manner ideally suited for computation. Mixture
theories on the other hand are found to provide a simple yet effective for
characterizing the values of phase stresses that act on mortar and aggregate and which
derive damage and plastic flow.
24
made. An analytical expression was developed to describe the entire response of plain
concrete.
2.2.3 W.A.M. Brekelmans P.J.M. Schreurs and J.H.P. de Vree (1991, Eindhoven
Netherlands)
Continuum damage mechanics was used to model the failure behavior of brittle
material. In the constitutive relation a damage parameter was incorporated. A damage
criterion was postulated to show the difference between tension and compression
strength. A damage growth law was quantified based on experimental data of
concrete. For the elaboration of mathematical formulation finite element method.
Numerical results obtained for a plain strain example showed the merit of procedure.
2.2.4 R. Raveendra Babu, Gurmail S. Benipal and Arbinda K. Singh ( IIT Delhi
,IITGuwahiti)
R. Raveendra Babu, Gurmail S. Benipal and Arbinda K. Singh described the various
developments in this based on different approaches elasticity, plasticity, continuum
damage mechanics, plastic fracturing, indocronic theory, microplane models, etc. they
assumed that the material undergo small deformations. Only time dependent
constitutive models and issues related to their implementation are discussed.
25
2.2.7. Kondo,Djemedo and Welemane,Helene and Cormery Fabrice(2007)
Kondo, Djemedo and Welemane, Helene and Cormery Fabrice presented the basic
elements macroscopic modeling of damage. They recalled the general approach of
continuum damage mechanics based on the thermodynamic irreversible processes and
its application to isotropic damage modeling. The study of damage induced anisotropy
was treated by considering second order tensorial damage variable. Finally they
presented an original microscopic approach through which is addressed the question
of uniaxial lateral effects due to the microcracks closure.
26
CHAPTER: 3
THERMODYNAMICS OF MATERIAL
Thermodynamic potential are quantities from which all characteristics of the system
can be deduced. Energy is the obvious thermodynamic potentials. Specific form of
energy is most suitable form of thermodynamic potential for various uses.
In this very general expression, the derivatives like 𝜕Π 𝜕𝜒𝑗 are the thermodynamic
properties associated with independent variable 𝜒𝑗 .
3.2 Entropy:
Where dQ is the actually the heat entering the system across all boundaries in the
isothermal process, and 𝜃 is the absolute temperature. This leads simply to 𝑑𝑄 = 𝜃𝑑𝑆
for a reversible process. More accurately, we write equation (3.2) with closed
boundary integral
𝑡2 𝑑𝑞
𝑆 𝜀1 − 𝑆 𝜀2 = 𝑡1 𝑠 𝜃
𝑑𝑠𝑑𝑡, (3.3)
27
Entropy is thus associated with change in state, and plays a crucial role in
thermodynamics since it can distinguish between heat and work. That in itself leads to
the second law of thermodynamics. The second law recognizes that heat and work and
work are not equivalent because mechanical energy can be completely converted to
thermal energy while thermal can be partially converted to the mechanical energy
during any cyclic process. There must be then dissipation.
The latter term comes by comparison with equation (3.2), so that we deduce
𝜕𝐸
𝜃= , (3.5)
𝜕𝑆
The incremental increase in energy, 𝜕Π, involves the sum of many partial differential
terms, the first of which is the elementary heat exchange equations (3.2) and (3.4): the
remainder must be the change in work. Since the state variables are independent,
work is independent of entropy, and so entropy separates heat and work in energy
exchange. Furthermore, the rate of heat exchange is thus 𝑄 = 𝜃𝑆 , and so total heat
exchanged is then 𝜃𝑆.
𝐹 = 𝐸 − 𝜃𝑆 (3.6)
The consequence is that the temperature becomes independent variable for free
energy and entropy its thermodynamic property.
This form of energy is known as Helmholtz free energy is dominant in the study of
solids. For a solid internal energy is the function of entropy, strain and independent
variable specific to problem: 𝐸 = 𝐸 𝑆, 𝜀, … … … . . Thus the common stress-strain
laws are
28
𝜕𝐸 𝜕𝐹
𝜎= so that 𝜎 = 𝜕𝐸 . (3.8)
𝜕𝜀
Enthalpy is the concept more suited to gases, but it may play role in the study of
porous media, and partial phases of water and steam. When internal energy is function
of volume i.e.𝐸 = 𝐸 𝑆, V, … . . , and pressure is its thermodynamic property, then
𝑑𝐸
𝑝 = − 𝑑V . (3.9)
In case of solids the internal energy is taken as function of strain, not volume, for
which stress is the thermodynamic property. The relationship is then
𝜕𝐸
𝜎= (3.11)
𝜕𝜀
With is consistent with equation (3.9) because tensile stress is taken as positive where
as pressure is being compressive is taken as negative. Enthalpy for a unit volume is
then written
𝜕𝐻
𝐻 = 𝜎: 𝐸 − 𝐸 Such that 𝜀 = . (3.12)
𝜕𝜎
Free energy is effectively that which can be released to the system. It is the quantity
most often used to begin constitutive modeling. The most common form of energy
used for description is Gibbs and Helmholtz functional.
𝐺 = 𝐸 − 𝜃𝑆 + 𝑃𝑉 (3.13)
𝐺 = 𝜎: 𝜀 − 𝐸 + 𝜃𝑆 (3.14)
It is vivid from equations (3.6) and (3.14) that (F and G per unit volume) form
complementary energies
29
𝐹 + 𝐺 = 𝜎: 𝐸 (3.15)
Mathematically,
For the body of volume V , enclosed by surface S and subjected to traction t per unit
area and body force f per unit mass, the power input is given as
𝑃𝑖𝑛𝑝𝑢𝑡 = 𝑉
𝑝. 𝑓. 𝑣𝑑𝑉 + 𝑠
𝑡. 𝑣𝑑𝑆 (3.18)
With the use of t=𝜎. 𝑛 where 𝜎 denotes the Cauchy stress tensor and n the outer
normal vector, it follows that
𝑃𝑖𝑛 𝑝𝑢𝑡 = 𝑉
𝑝. 𝑓. 𝑣𝑑𝑉 + 𝑉
𝑣. 𝜎 . ∇𝑑𝑉
= 𝑉
𝑝. 𝑓 + 𝜎. ∇ . 𝑣𝑑𝑉 + 𝑉
𝑡𝑟 𝜎. ∇𝑉 𝑑𝑉
= 𝑉
𝑝. 𝑣 . 𝑣𝑑𝑉 + 𝑉
𝑡𝑟 𝜎. ∇𝑉 𝑑𝑉 (3.19a)
𝑑 1
𝑃𝑖𝑛𝑝𝑢𝑡 = 𝑑𝑡 𝑉 2
𝑝 𝑣. 𝑣 𝑑𝑉 + 𝑉
𝑡𝑟 𝜎. 𝐿𝑇 𝑑𝑉 (3.19b)
Where the divergence theorem and the equation of motion have been used in
obtaining equation (3.19) and L is the velocity gradient tensor.
𝑄𝑖𝑛𝑝𝑢𝑡 = 𝑉
𝑝. 𝑟 𝑑𝑉 + 𝑆
. 𝑛 𝑑𝑠 (3.20)
The total energy of the system will be considered as the sum of the kinetic energy and
the internal energy u. Then from the equation (3.17) together equations (3.19) and
(3.20), it follows that
30
𝑑 1 𝑑 1
= 𝑉 2
𝑝 𝑣. 𝑣 + 𝑝𝑢 𝑑𝑉 = 𝑑𝑡 𝑉 2
𝑝 𝑣. 𝑣 𝑑𝑉 + 𝑉
𝑡𝑟 𝜎. 𝐿𝑇 𝑑𝑉 + 𝑉
𝑝. 𝑟 𝑑𝑉 −
𝑑𝑡
𝑆
. 𝑛 𝑑𝑆 (3.21a)
Which reduces to
𝑑 1
𝑝𝑢𝑑𝑉 = 𝑡𝑟 𝜎. 𝐿𝑇 𝑑𝑉 + 𝑝. 𝑟 𝑑𝑉 − . 𝑛 𝑑𝑠
𝑑𝑡 𝑉 2 𝑉 𝑉 𝑆
= 𝑉
𝑡𝑟 𝜎. 𝐷 𝑑𝑉 + 𝑉
(𝑝. 𝑟) 𝑑𝑉 − 𝑉
∇. 𝑑𝑉 (3.21b)
Where D is the symmetric part of velocity gradient tensor L and is called the rate of
deformation tensor. The term 𝜎: 𝐷 is referred to as external power tensor.
𝑑𝑢
𝑉
𝑝 𝑑𝑡 − 𝑡𝑟 𝜎. 𝐷 − 𝑝𝑟 + ∇. h 𝑑𝑉 = 0 (3.22)
Equation (3.23) is the well known energy equation, and in the absence of any internal
heat source reduces to
𝑡𝑟 𝜎. 𝐷 + 𝑝𝑟 − ∇. 𝐻 (3.24)
The energy equation can be written in terms of HFE by using equation (3.17):
𝑃 𝐴 + 𝜃𝑠 + 𝑠𝜃 = 𝑡𝑟 𝜎. 𝐷 − ∇. (3.25)
The second law of thermodynamics is the postulate that the rate of entropy increases
is greater than or equal to entropy input rate, where the inequality implies an
irreversible process and the inequality holds for a reversible process.
The statement of the second law in the notion of continuum mechanics leads to the
integral form of the Clausius-Duhem inequality. Mathematically, the second law is
written as
𝑟 1
𝑉
𝑝. 𝑠 𝑑𝑉 ≥ 𝑉
𝑝. 𝑑𝑉 + 𝑆 𝜃
. 𝑛 𝑑𝑠 (3.26)
𝜃
Where 𝜃 the absolute temperature and n is is the unit outward normal as before.
31
In the absence of any internal heat source r, and the use of the divergence theorem in
the last term of equation (2.26), it follows that
𝑉
𝑝. 𝑠𝑑𝑉 ≥ − 𝑉
∇. 𝜃 𝑑𝑉
(3.27)
𝑑𝑠 1
For an arbitrary volume, Therefore, 𝑑𝑡 + ∇. 𝑝 ≥0 (3.28a)
𝜃
Or equivalently
∇. .∇𝜃
𝑝𝜂 = 𝑝𝑠 + − ≥0 (3.28b)
𝜃 𝜃2
Equations (3.28a) and (3.28b) are Clausius-Duhem inequality. Combining the later
equation with equations (3.27) and (3.28) leads to alternate expression in terms of the
internal energy function and the HFE.
𝑢 .∇𝜃
𝜂 = 𝑠 − 𝑝𝜃 𝑡𝑟 𝜎. 𝐷 − ≥0 (3.29a)
𝑝𝜃 2
1 .∇𝜃
𝜃𝜂 = −𝐴 − 𝑠𝜃 + 𝑝 𝑡𝑟 𝜎. 𝐷 − ≥0 (3.29b)
𝑝𝜃
i.e.,
𝜕𝑦 0
𝜉𝑖 = (3.30)
𝜕𝑥 𝑖 𝑥 1,𝑥 2 ,…………..𝑥 𝑚
Legendre transformation of total energy with respect to entropy gives Helmholtz free
energy.
32
CHAPTER 4
GENERAL FORMULATION
It will be assumed that a continuum approach can effectively be taken for the
modeling of a class of brittle solids whose microstructures are altered by the
nucleation and growth of distributed microcracks. With the assumption that a
neighboring equilibrium state exists for all irreversible processes, the formulation can
be cast within the general setting of internal variable theory of thermodynamics
(Kestin and Rice, 1970; Nemat-Nasser, 1976; Truesdell, 1984). For a stress space
formulation and for a non-polar continuum, the Gibbs free energy, G is given as
1
𝐺 = 𝐺 𝜎, 𝑘 = 2 𝜎: 𝐶 𝑘 : 𝜎 + 𝜎: 𝜀 𝑖 𝑘 − 𝐴𝑖 𝑘 (4.1)
Where the symbol “:” indicates tensor contraction. The Cauchy stress tensor is given
by σ, and the material compliance and total irrecoverable strain tensor are represented
by C and ԑi(k), respectively. The last term in equation (4.1) represents the inelastic
component of the Helmholtz free energy and is associated with the surface energy of
microcracks. The internal variable parameter k is a monotonically increasing scalar
and is used to reflect the total dissipated energy associated with material damage.
Assuming that only damage (cracking) contributes to the alteration of the elastic
properties ( Budiansky and O’Connell, 1976: Horii and Nemat- Nasser, 1983; Ortiz
and Popov, 1982; Ortiz, 1985), the following decomposition of the compliance tensor
is adopted
𝐶 𝑘 = 𝐶𝑜 + 𝐶𝑐 𝑘 (4.2)
Where Co and Cc( k) denote the initial and added flexibility tensors, respectively. The
dependence of C on k allows for the description of damage through the fourth order
material compliance tensor. For a purely mechanical state, the Clausius-Duhem
inequality must be satisfied. This inequality is given as
𝐺 − 𝜀: 𝜎 ≥ 0 (4.3)
Where the super dots reflect the rate forms. The substitution of the expression for G in
the above stated inequality would lead to the following alternate form of the
dissipation inequality
𝜕𝐺 𝜕𝐺 𝜕𝐺
− 𝜀 : 𝜎 + 𝜕𝐶 ∷ 𝐶 + 𝜕𝜀 𝑖 : 𝜀 𝑖 − 𝑔2 𝑘 ≥ 0 (4.4)
𝜕𝜎
33
Where the scalar function g2 is given by g2= 𝜕𝐴𝑖 𝜕𝑘 . Following the standard
arguments (Coleman and Gurtin, 1967; Lubliner, 1972) and assuming that unloading
is an elastic process, the following relations are established
𝜕𝐺
𝜀= = 𝐶 𝑘 : 𝜎 + 𝜀𝑖 𝑘 (4.5)
𝜕𝜎
An equation which indicates that, for a stress space formulation, The Gibbs free
energy is a potential for the infinitesimal strain tensor, 𝜀, and that
𝜕𝐺 𝜕𝐺
𝑑𝑠 = 𝜕𝐶 ∷ 𝐶 + 𝜕𝜀 𝑖 : 𝜀 𝑖 − 𝑔2 𝑘 ≥ 0 (4.6)
Where 𝑑𝑠 is the rate of dissipation indicating that the rate of work done by the
thermodynamic affinities, 𝜕𝐺 𝜕𝐶 , 𝜕𝐺 𝜕𝜀 𝑖 , and g2 , through their respective
fluxes 𝐶 , 𝜀 𝑖 , and 𝑘 , must be non- negative. The rate form of the total strain tensor in
terms of its components is expressed as
𝜀 = 𝜀 𝑒 + 𝜀 𝑑 + 𝜀 𝑖 = 𝐶: 𝜎 + 𝐶 𝑐 : 𝜎 + 𝜀 𝑖 (4.7)
𝐶 𝑐 = 𝑘𝑹,𝜀 𝑖 = 𝑘𝑴 (4. 8)
Where R and M are fourth and second order tensors that determine the direction of
elastic and inelastic damage process. If R is taken to be proportional to to the fourth
order identity tensor, then damage formulation would be isotropic. If R is chosen to
reflect directionality for damage, then an anisotropic damage formulation is obtained.
For a thermodynamically consistent formulation, R and M must be postulated such
that the dissipation inequality is satisfied. With the assumption that damage is
irreversible i.e., 𝑘 ≥ 0; equation (4.6) in terms of R and M becomes
1
𝑑 = 2 𝜎: 𝑅: 𝜎 + 𝜎: 𝑀 ≥ 0 (4.9)
34
will always be satisfied if R is a positive semi-definite tensor. Let 𝑑 = 𝑑1 + 𝑑2 where
𝑑1 = 1 2 𝜎: 𝑹: 𝜎 and 𝑑2 = 𝜎: 𝑀 for convenience. If R is positive semi-definite and
M is chosen such that 𝑑2 ≥ 0, equation (4.9) will always be satisfied. This approach
is taken in this thesis. Two other possible situations are (1) when 𝑑1 ≥ 0 and 𝑑2 ≤ 0,
and (2) when 𝑑1 ≤ 0 and 𝑑2 ≥ 0 with the constraint that 𝑑1 + 𝑑2 ≥ 0. Such
combinations are not considered here and can be found in Schreyer and Nielsen(
1991), a study of negative damage or healing in concrete.
𝜓 ≤ 0, 𝑘 ≥ 0, 𝑘𝜓 = 0 (4.11)
With the terms satisfying equation (4.11) simultaneously. When the energy is
dissipated, 𝑘 > 0, which requires that 𝜓 be zero. During elastic process, 𝜓 is less than
zero. For computational purposes, the rate of 𝑘 and tangent compliance tensor are
needed. 𝑘 is obtained from the consistency condition of the damage surface as
𝜕𝜓
𝜕𝜎
:𝜎 (𝜎 :𝑅+𝑀−𝑡𝑡 𝜎 )
𝑘 =− 𝜕𝜓 = :𝜎 (4.12)
𝑡𝑡 𝑘
𝜕𝑘
In which symmetric second order tensor 𝑡 𝜎 and the scalar function 𝑡𝑘 are defined as
𝑡 𝜎 = 𝜕𝑡 𝜕𝜎 and 𝑡𝑘 = 𝜕𝑡 𝜕𝑘 , respectively. The tangent compliance tensor is
obtained by considering the rate form of total strain tensor and substituting the
expression for 𝑘 from equation (4.12) as
𝜀 = 𝐶𝑇 : 𝜎 (4.13)
35
In which all terms has been defined previously. The symbol ⨂ indicates the tensor
product. It can be readily shown that if R and M are symmetric, the tangent
compliance tensor will be symmetric. If the damage function 𝑡 is taken as function of
damage parameter only, 𝑡 𝜎 will vanish and the original formulation of Ortiz (1985)
and Yazdani and Schreyer (1988) will be obtained. The dependence of 𝑡 on both state
variables is more general and allows for the description of an alternate damage
function, as will be presented later in this Thesis.
Two damage modes are usually identified in brittle solids. One is cleavage cracking
mode, where crack opening vector is parallel to the applied tensile stresses, and the
second is the compressive mode, which represents a combination of shear sliding and
crack opening. Spectral decomposition of the stress tensor is subsequently performed
in order to incorporate the two modes into the formulation. Denoting the positive and
negative cones of stress tensor as 𝜎 + and 𝜎 −, respectively, it follows that
𝜎 = 𝜎+ + 𝜎− (4.15)
With subscripts I and II referring to the two damage models described above. If R I and
RII are expressed as tensorial functions of 𝜎 + and 𝜎 − , respectively, the formulation
would imply decoupling between two damage modes. The validity of such
decomposition then becomes questionable, as has been explained by Krajcinovic
(1989), whose comments were based on micromechanics of shear cracks. The
decomposition, however, does provide a convenient, although perhaps an
approximate, approach to a fundamental and complicated problem abd thus will be
retained in the present model.
To progress further, particular forms of RI, RII, MI and MII must be specified. The
following discussions and representations are particularly relevant to concrete, with
potential application to other quasi-brittle solids such as mortar, rocks, and ceramics.
36
Mode I
For the cleavage cracking mode, Ortiz (1985) proposed the following expression for
RI
(𝜎 + ⨂𝜎 − )
RI= (4.17)
(𝜎 + : 𝜎 + )
Which was also used by Yazdani and Schreyer (1988) . This particular expression
leads to the prediction of anisotropy due to changes in Youngs’s moduli but cannot
model changes in the apparent poisson’s ratio, which is responsible for the nonlinear
behavior when the lateral strains are considered. To illustrate this point, consider
figure 1, where the material responses in both axial and lateral directions are shown
schematically with solid lines. This representation is consistent with the work of
Krajcinovic and Fonseka (1981). By using equation (4.17), the theoretical results
would predict that the material response would be along the broken line designated as
“OA”. To improve on the perceived shortcoming of the formulation, the following
modification is proposed. Let the maximum Eigen value of 𝜎 + be 𝜆1 and define
𝜙1 = 𝐻 𝜆1 for convenience where 𝐻 ∙ represents the Heaviside function. With
these, RI is postulated to be
(𝝈+ ⨂𝝈− )
𝑅𝐼 = + ΥΦ1 I − i⨂i (4.18)
(𝝈+ : 𝝈+ )
Where Υ is a material parameter. I is the fourth order intensity tensor, and i represents
the second order identity tensor.
37
For uniaxial tension equation (4.18) reduces to
1
𝜀1 = +𝑘 𝜎 (4.19)
𝐸
𝜐
𝜀2 = 𝜀3 = − + 𝑘𝛾 𝜎 (4.20)
𝐸
A particular form of t(k) was obtained from the uniaxial tensile tests of Smith and
Young(1955) was given by Ortiz (1985) as
ln
(1+𝐸𝑜 𝑘)
𝑡 𝑘 = 𝑓𝑡 𝑒 (4.21)
1+𝐸𝑜 𝑘
𝜐
𝜀3 = −2(𝑘 + 𝛾𝑘)𝜎 (4.24)
And, 𝑡 𝑘 = 2 1−𝛾 σ
𝑓𝑡 𝑒 𝑙𝑛
(1+𝐸𝑜 𝑘)
𝜎= (4.25)
2 1−𝛾 (1+𝐸𝑜 𝑘)
And 𝑡 𝑘 = 3 1 − 2𝛾 𝜎 (4.27)
𝑓𝑡 𝑒 ln
(1+𝐸𝑜 𝑘)
𝜎= (4.28)
3 1−2𝛾 1+𝐸𝑜 𝑘
The constitutive relations given by equations (4.24), (4.26) and (4.28) for uniaxial,
biaxial and triaxial tensile loading respectively shows more ductile nature of concrete
even after peak stress is achieved. The fact is that the theory assumes the uniform
propagation of crack in direction of stress throughout the volume of concrete. In fact
there is localized crack formation occur after peak stress causing brittle failure of
38
concrete. Considering localized crack formation we may modify the constitutive
relation for uniaxial tensile loading as
𝜎1 = 𝐸 𝑜 − 𝜇 𝜀1 (4.29)
𝜇 <𝜇 −𝜇 ∗ >
Where,𝜀1 = 𝜀𝑢 and 𝛼 = 1+
(𝐸 𝑜 −𝜇 )𝛼 𝐸𝑜
Where,𝜇 = 1/𝑘, 𝜇 ∗ is the value corresponding to peak stress and 𝜀𝑢 is the strain
corresponding to peak stress and the numerical values of 𝐸 𝑜 is taken as 31000MPa
same as the value taken by Gopalratnam and Shah(1985).
39
CHAPTER 5
CALCULATIONS, RESULTS AND DISCUSSIONS
CALCULATIONS:
For the calculation purpose equations (4.20) to (4.29) as derived in previous chapter
are used and Computations is done in MATLAB. Program codes (as shown in
appendix A) are developed. Different trial values of Eok are assumed and then
corresponding values of stress (σ), axial strain (ԑ1) and transverse strain (ԑ2, ԑ3) for
different magnitude of Eok are obtained from the equations are obtained. Stress- stain
curves for uniaxial, biaxial and triaxial tensions are plotted also the stress-strain curve
for proposed uniaxial tension model is plotted and compared with experimental
results obtained by Gopalratnam and Shah(1985)(as shown in appendix B and C)
from the equations mentioned above.
A plot between axial stress-axial strain and axial stress-transverse strains is made as
shown in the figure.
Table 5.1 Material parameters are taken for the calculations: (S. Yazdani)
Parameters Values
40
Figure 5.1 Stress-strain curves for plain concrete under Uniaxial tension.
41
Figure 5.2 Stress-strain curves for plain concrete under equal biaxial tension.
42
Figure 5.3 Stress-strain curves for plain concrete under equal triaxial tension
43
Figure 5.4 Theoretical Stress-strain curve of plain concrete under multiaxial tension
44
Figure 5.5 Multiaxial stress vs Eok curves for plain concrete
45
Figure 5.6 Comparison of theoretical and experimental axial stress-strain curve for
plain concrete under uniaxial tension
46
Figure 5.7 Comparison of proposed and experimental axial stress-strain curve for
plain concrete under uniaxial tension
47
Figure5.8 Variation of damage with damage parameter for proposed model
48
RESULTS, CONCLUSIONS AND RECOMMENDATIONS
Results:
From the analytical modeling of plain concrete under multiaxial tensile loading
following results are obtained:
1) The strength of plain concrete decreases successively when the tensile loading
changes from uniaxial to biaxial and triaxial i.e. damage increases which is
indicated by figure 5.4 and figure 5.5.
2) The damage mechanics deals with distributed cracks not with localized crack
which results more ductile failure of concrete as compared to experimental
results which is clarified by figure 5.6.
3) The proposed new damage model which is based on damage mechanics is
almost similar to that obtained from experimental work as seen in figure 5.7.
Conclusions:
Limitations of study:
50
REFERENCES
51
Kupfer, H., H.K. Hilsdorf and H. Rusch. 1969. “Behavior of Concrete under
Biaxial Stress”, J.Am. Concr. Inst., 66(8):656-666.
Ortiz M, Popov EP, “A physical model for the inelasticity of concrete,” Proc.
Roy. Soc. London, A383, pp. 101-125, 1982.
52
APPENDIX A
Program codes
Eo=32000;
ft=3.48;
gamma=0.2;
nu=0.2;
Eok=0:0.1:10;
sigma=ft*2.72*log(1.+Eok)./(1.+Eok);
e1=(1/Eo+Eok./Eo).*sigma;
e2=-(nu/Eo+gamma*Eok./Eo).*sigma;
plot(e2,sigma);
xlabel('strain');
ylabel('stress in MPa');
grid on
Eo=32000;
ft=3.48;
gamma=0.2;
nu=0.2;
Eok=0:0.1:10;
sigma=ft*2.72/sqrt(2*(1-gamma))*log(1+Eok)./(1+Eok);
53
epsilon1=(1/Eo+(1-gamma)*Eok./Eo-nu/Eo).*sigma;
epsilon2=-2*(nu/Eo+gamma*Eok./Eo).*sigma;
plot(epsilon2,sigma);
xlabel('strain');
ylabel('stress in MPa');
grid on
Eo=32000;
ft=3.48;
gamma=0.2;
nu=0.2;
Eok=0:0.1:10;
sigma=ft*2.72/sqrt(3*(1-2*gamma))*log(1+Eok)./(1+Eok);
epsilon1=(1/Eo+Eok./Eo-2*nu/Eo-2*gamma*Eok./Eo).*sigma;
epsilon2=-(1/Eo+Eok./Eo-2*nu/Eo-2*gamma*Eok./Eo).*sigma;
plot(epsilon2,sigma);
xlabel('strain');
ylabel('stress');
grid on
54
Program code for multiaxial tension
Eo=32000;
ft=3.48;
gamma=0.2;
nu=0.2;
Eok=0:0.1:10;
unisigma=ft*2.72*log(1.+Eok)./(1.+Eok);
bisigma=ft*2.72/sqrt(2*(1-gamma))*log(1+Eok)./(1+Eok);
trisigma=ft*2.72/sqrt(3*(1-2*gamma))*log(1+Eok)./(1+Eok);
unie1=(1/Eo+Eok./Eo).*unisigma;
unie2=-(nu/Eo+gamma*Eok./Eo).*unisigma;
bie1=(1/Eo+(1-gamma)*Eok./Eo-nu/Eo).*bisigma;
bie2=-2*(nu/Eo+gamma*Eok./Eo).*bisigma;
trie1=(1/Eo+Eok./Eo-2*nu/Eo-2*gamma*Eok./Eo).*trisigma;
trie2=-(1/Eo+Eok./Eo-2*nu/Eo-2*gamma*Eok./Eo).*trisigma;
plot(unie1,unisigma,unie2,unisigma,bie1,bisigma,bie2,bisigma,trie1,trisigma,trie2,tris
igma);
xlabel('strain(\epsilon)');
ylabel('stress(\sigma)in MPa');
grid on
55
Program code for sress vs Eok
Eo=32000;
ft=3.48;
gamma=0.2;
Eok=0:0.1:10;
sigmauniaxial= ft*2.72*log(1+Eok)./(1+Eok);
sigmabiaxial=ft*2.72/sqrt(2*(1-gamma))*log(1+Eok)./(1+Eok);
sigmatriaxial=ft*2.72/sqrt(3*(1-2*gamma))*log(1+Eok)./(1+Eok);
plot(Eok,sigmauniaxial,Eok,sigmabiaxial,Eok,sigmatriaxial);
title('stress(\sigma)vs Eok')
xlabel('Eok');
ylabel('stress(\sigma)in MPa');
legend('uniaxial','biaxial','triaxial',0);
grid on
ft=3.48;
gamma=0.2;
nu=0.2;
Eok=0:0.1:10;
sigma=ft*2.72*log(1.+Eok)./(1.+Eok);
epsilon1=(1/Eo+Eok./Eo).*sigma;
sigmaexperimental=[0
0.85633
1.69218
56
2.7114
3.2619
3.38376
2.19987
1.72922
1.4634
1.11483
0.868086
0.765232
0.599675
0.515898
0.41297
0.329636
0.224564];
epsilonexperimental=[0.00E+00
2.42E-05
5.07E-05
8.58E-05
0.000101259
0.000118697
0.000124762
0.000161537
0.000176652
0.000224343
0.000278596
0.000302469
57
0.000369797
0.000432808
0.000458855
0.000508823
0.000597913];
plot(epsilon1,sigma,epsilonexperimental,sigmaexperimental,'--');
xlabel('strain(\epsilon)');
ylabel('stress(\sigma)in MPa');
grid on
58
APENDIX B
(Calculation of stress and strains)
59
Table B.1 Sample Calculation for uniaxial tension
Sigma=2.72*ft* k= ԑr=-
Eok ln(1+Eok)/(1+Eok) Eok/Eo ԑ=Sigma*(1/Eo+k) (nu/Eo+k*gamma)
0 0 0 0 0
60
2.3 3.424604 7.19E-05 0.000353 -7.06325E-05
61
4.8 2.868824 0.00015 0.00052 -0.000103995
62
7.3 2.413449 0.000228 0.000626 -0.000125198
63
9.8 2.08554 0.000306 0.000704 -0.000140774
Strain Stress
0.00E+00 0
2.42E-05 0.85633
5.07E-05 1.69218
8.58E-05 2.7114
0.000101 3.2619
0.000119 3.38376
0.000125 2.19987
0.000162 1.72922
0.000177 1.4634
0.000224 1.11483
0.000279 0.868086
0.000302 0.765232
0.00037 0.599675
0.000433 0.515898
0.000459 0.41297
0.000509 0.329636
0.000598 0.224564
64