Sie sind auf Seite 1von 124

MATERIAL MODELLING OF

REINFORCED CONCRETE AT
ELEVATED TEMPERATURES

Master Thesis
February 2011

Fire Safety, Section for Building Design,
Department of Civil Engineering, the Technical University of Denmark

Josephine Voigt Carstensen, s052204


Material Modelling of Reinforced
Concrete at Elevated Temperatures

M.Sc. in Civil Engineering - Master Thesis credited with 30 ECTS points


Project Period: 2010.09.13-2011.02.11
Language: English

Fire Safety at the Section for Building Design


Department of Civil Engineering
Technical University of Denmark

In collaboration with:
BRE Centre for Fire Safety Engineering
The Univeristy of Edinburgh

Supervisor: External Supervisor:


Dr. Grunde Jomaas Dr. Pankaj Pankaj
Assistant Professor Senior Lecturer
Department of Civil Engineering School of Engineering
Technical University of Denmark The University of Edinburgh

Handed in 2011.02.11 by:

Josephine Voigt Carstensen, s052204

i
Abstract

Previous disasters have elucidated that accurate and realistic modelling of concrete behaviour
at elevated temperatures is fundamental for the safe design of, for example, nuclear and struc-
tures exposed to fire. However, when the same model is evaluated with different mesh sizes, the
existing models for the behaviour of concrete at elevated temperatures are subject to problems
with convergence of results in the Finite Element (FE) analysis. These problems arise as a result
of the problem of localization of deformations associated with the post-peak response of concrete.

This current research focuses on the modelling of the uniaxial behaviours of reinforced concrete
at elevated temperatures and in particular on the key issues associated with the post-peak be-
haviour.

It is generally recognized that in order to obtain mesh independent results of models of rein-
forced concrete in FE-analysis at ambient conditions, a fracture energy based material model
must be adopted. In tension, such models are widely used and in most FE-codes, for example
ABAQUS, it is possible to define the tensile post-peak behaviour in three ways; either through
an element size dependent stress-strain relation, through a stress-displacement formulation or
by giving the tensile fracture energy and letting ABAQUS define the behaviour. However, if
reinforced concrete is to be considered, the tensile definition must account for the tension stiffen-
ing effect that gradually shifts the load-bearing capacity from the concrete to the reinforcement
as the cracking progresses. This issue can be tackled by defining an element size dependent
interaction stress contribution that is combined with the concrete contribution for the definition
of the post-peak behaviour. In compression the fracture energy based behaviour models are less
used and the compressive fracture energy is, for example, not discussed in any current codes
and it is generally examined by very few. To apply a fracture energy based compressive model
in a FE-analysis, an element size dependent stress-strain formulation must be used.

In this current research, the existing models for the ambient condition have been extended to
elevated temperatures, largely by applying the material properties at a given elevated temper-
ature to the current formulation. Therefore, the existing models have been evaluated prior to
the extension and it has been found necessary to express limits for their application. It is well
established that a limit on the maximum element size exists. However, herein it has been found
that restrictions on the minimum element size and, if modelling the tension stiffening through
the definition of an interaction stress contribution, on the minimum level of reinforcement ad-
missible also apply.

As experimental data is currently not available on the evolution of the compressive and the
tensile fracture energy with temperature, the fracture energies inherent in the existing elevated
temperature models have been examined. It has been found that the tensile fracture energy
inherent in the currently available model follows the decay function for material strength. The

iii
compressive fracture energy has been based on the models by four current compressive models
where two considers solely the instantaneous stress-related strain and two includes the effects
of the LITS. It has been established that the current compressive elevated temperature models
does not agree on the post-peak behaviour and that the LITS does not seem to have an effect
on the post-peak response.

The limits of application are extended to elevated temperatures by expressing a validity range
for the element sizes and a minimum reinforcement ratio. It has been found that up to about
500◦ C, the maximum element size is typically governed by the tensile properties after which
the compressive parameters are governing. Once the compressive model becomes governing, it
only provides meaningful results within a very limited range of mesh-sizes. This range should
be considered the new validity domain of the model.

This novel model for the uniaxial behaviours of reinforced concrete at elevated temperatures can
readily be applied for FE analysis, for example in ABAQUS, and, if the modelling is performed
within the limits of application, it is possible to get mesh independent results of the analyses
with different mesh configurations.
Preface

This project is a M.Sc. thesis of 30 ECTS points created in the period September 13th 2010 to
February 11th 2011. A M.Sc. thesis is a compulsary project in order to fulfill the requirements
for the M.Sc. programme in Civil Engineering at the Technical University of Denmark, (DTU).

The project has been carried out for the Fire Safety Group at the Section for Building Design,
Department of Civil Engineering at the Technical University of Denmark in collaboration with
at the BRE Centre for Fire Safety at the University of Edinburgh.

The internal supervisor of the project has been Dr. Grunde Jomaas (Assistant Professor, DTU)
and the external supervisor has been Dr. Pankaj Pankaj (Senior Lecturer, Edinburgh).

The work presented in the thesis was conducted at the University of Edinburgh.

v
Acknowledgements

First, a great amount of appreciation must be given to the BRE Center for Fire Safety En-
gineering at the University of Edinburgh and especially to the students and staff in the John
Muir Building for creating a welcoming and inspiring research environment. My visit there has
proved to be a highly educative experience, thanks both to the academic and the non-academic
support received at the premises. A special expression of gratitude is given to Prof. José L.
Torero for setting up the practical framework, without which this project would not have been
accomplished.

A very special thanks is directed to Dr. Pankaj Pankaj for all his guidance and encouragement.
I have immensely appreciated that he has always taken time to patiently explain the arisen
problems - no matter the magnitude. His ability to make even the most complex problems
understandable is something I profoundly admired. On this note appreciation is also dedicated
to Prof. Kristian D. Hertz (DTU) and Dr. Martin Gillie for their clarifications of puzzling
definitions.

Will Kingston is to be deeply thanked for the helpful discussions and useful hints throughout
the project period. Especially his calm introduction to ABAQUS modelling at the project start
and his patient answers to emerging ABAQUS related questions have been beyond compare.
On this note Adam Ervine, Kate Andersson and Joanne Knox must also be recognized along
with Rory Hadden, Cristián Maluk, Nicolas Bal, Steffen Kahrmann and Dr. Francesco Colella.

Further, a particularly gratefulness is given to Dr. Grunde Jomaas (DTU) for his friendly ap-
proach and guidance. He must be recognized for creating the contact between the collaborators
of the project and for being an tremendous source of inspiration. His mentoring and guidance
through the project planning and execution, as well as through decision making about further
professional career, have had great effects on both the project at hand and on future choice of
occupation.

Lærke Mikkelsen (DTU) and Miki Kobayashi (DTU) are acknowledged for helping with retriev-
ing literature and Mads Mønster Jensen (DTU) for his clarification of the mysteries of concrete
technology.

Last but not least, gratitude is directed to the OTICON Foundation, Reinholdt W. Jorck’s
Foundation, KAB’s studielegat, the Department of Civil Engineering at the Technical Univer-
sity of Denmark and BRE Center for Fire Safety Engineering at the University of Edinburgh
for the received financial support.

vii
Contents

Abstract iii

Preface v

Acknowledgements vii

Nomenclature xiii

List of Figures xvii

List of Tables xxiii

1 Introduction 1
1.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Localization and Fracture Energy in Tension . . . . . . . . . . . . . . . . . . . . 2
1.3 Localization and Fracture Energy in Compression . . . . . . . . . . . . . . . . . . 3
1.4 Novelties and Milestones of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Finite Element modelling of Multiaxial Behaviour of Reinforced Concrete 7


2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 FE-Modelling of Concrete Cracking . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 Concrete Model in ABAQUS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3.1 Yield Surface Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3 Response of Reinforced Concrete to Fire Exposure 13


3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.2 Chemical and Physical Effects of Fire . . . . . . . . . . . . . . . . . . . . . . . . 13
3.2.1 Chemophysical Response of Concrete to Fire . . . . . . . . . . . . . . . . 14
3.2.2 Chemophysical Response of Reinforcing Steel to Fire . . . . . . . . . . . . 15
3.3 Typical Failures of Reinforced Members . . . . . . . . . . . . . . . . . . . . . . . 17
3.4 Choice of Analysis Type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.5 Overview of Concepts Involved in the Response of Reinforced Concrete to a Fire 18

4 Modelling of Uniaxial behaviour of Reinforced Concrete at Ambient Tem-


perature 19
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.2 Material Model of Reinforcement . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.3 Reinforced Concrete in Tension . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.3.1 Tension Stiffening Model as per the CEB-FIB Model Code . . . . . . . . 21
4.3.2 Tension Stiffening Model by Cervenka et al. . . . . . . . . . . . . . . . . . 22
4.3.3 Tension Stiffening Model by Feenstra and de Borst . . . . . . . . . . . . . 26
4.4 Compressive Behaviour of Concrete . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.4.1 Compression Model in CEB-FIB Model Code . . . . . . . . . . . . . . . . 30
4.4.2 Compressive Fracture Energy . . . . . . . . . . . . . . . . . . . . . . . . . 31

ix
4.4.3 Compression Model of Narakuma and Higai . . . . . . . . . . . . . . . . . 33
4.4.4 Compression Model by Feenstra and de Borst . . . . . . . . . . . . . . . . 34
4.4.5 Comparison of Compression Models . . . . . . . . . . . . . . . . . . . . . 36
4.5 Chosen Uniaxial Concrete Models . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.6 Numerical Test Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.6.1 Uniaxial Tension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.6.2 Uniaxial Compression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.6.3 Pure Shear . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

5 Existing Models of the Behaviour of Reinforced Concrete at Elevated Tem-


peratures 41
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5.2 Decay of Material Strength . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5.2.1 Compressive Strength of Concrete . . . . . . . . . . . . . . . . . . . . . . 41
5.2.2 Tensile Strength of Concrete . . . . . . . . . . . . . . . . . . . . . . . . . 44
5.2.3 Strength of Reinforcement . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.3 Uniaxial Compressive behaviour of Concrete at Elevated Temperatures . . . . . . 46
5.3.1 Strain Components at Elevated Temperatures . . . . . . . . . . . . . . . . 47
5.4 Uniaxial Tensile behaviour of Concrete at Elevated Temperatures . . . . . . . . . 51
5.5 Reinforcement Model at Elevated Temperatures . . . . . . . . . . . . . . . . . . . 52
5.6 Overview of Relevant Assumptions for the Formulation of the Fracture Energy
Based Material Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

6 Fracture Energy Based Uniaxial Material Models at Elevated Temperatures 55


6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
6.2 Fracture Energy Based Compressive behaviour Model for Concrete at Elevated
Temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
6.2.1 Compressive Fracture Energy at Elevated Temperatures . . . . . . . . . . 56
6.2.2 Application of the Elevated Temperature Model by Anderberg and The-
landersson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
6.2.3 Application of the Elevated Temperature Model by Lie and Lin . . . . . . 58
6.2.4 Compressive Fracture Energies at Elevated Temperatures for Models In-
cluding the Effect of the LITS . . . . . . . . . . . . . . . . . . . . . . . . . 59
6.2.5 Comparison of Compressive Fracture Energies at Elevated Temperatures . 60
6.3 Formulation of Fracture Energy Based Tensile Model for Concrete at Elevated
Temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
6.3.1 Tensile Fracture Energy at Elevated Temperatures . . . . . . . . . . . . . 62
6.3.2 Fracture Energy Based Tensile Model of Plain Concrete . . . . . . . . . . 63
6.3.3 Fracture Energy Based Tensile Model for Reinforced Concrete . . . . . . . 63
6.4 Limits of Fracture Energy Based Models at Elevated Temperatures . . . . . . . . 65
6.4.1 Limitations on the Element Size . . . . . . . . . . . . . . . . . . . . . . . 65
6.4.2 Minimum Reinforcement Ratio . . . . . . . . . . . . . . . . . . . . . . . . 66

7 Numerical Example of a Reinforced Concrete Slab at Elevated Temperatures 69


7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
7.2 Parameters for the Uniaxial Material Models . . . . . . . . . . . . . . . . . . . . 70
7.3 Material Properties for the Thermal Analysis . . . . . . . . . . . . . . . . . . . . 71
7.4 FE-Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
7.4.1 Element size h = 129 mm . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
7.4.2 Element size h = 73 mm . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

8 Conclusion 79
8.1 Remarks in Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
8.2 Suggestions for Future Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
Bibliography 84

A Detailed Description of Cracking and the Post-Peak Response of Concrete 85


A.1 Crack Propagation and Softening . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

B ABAQUS Functions for Definition of Uniaxial Behaviour, Embedment of


Reinforcement and Load Steps 87
B.1 Tension Stiffening and Compression Hardening Models . . . . . . . . . . . . . . . 87
B.2 Embedment of Reinforcement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
B.3 Load Step Definition for Static Analysis . . . . . . . . . . . . . . . . . . . . . . . 88

C ABAQUS Output from Pure Shear Example of Simple Plates with and
without Reinforcement 91
C.1 Simple Shear Example without Reinforcement . . . . . . . . . . . . . . . . . . . . 91
C.2 Simple Shear Example with Reinforcement . . . . . . . . . . . . . . . . . . . . . 93

D Uniaxial Material Models for the Numerical Example of a Reinforced Slab 95


Nomenclature

Ac,ef f Effective cross-sectional area of G(σ) Flow potential function in


the concrete. ABAQUS [8].
As Total area of the reinforcement. Gc Compressive fracture energy at
As,min Mimimum area of the reinforce- ambient temperature.
ment if the interaction contribu- (Gc /h)AT Compressive fracture energy di-
tion must be considered as a part vided by the corresponding ele-
of the tension stiffening. ment size inherent in the elevated
b Length of reinforced concrete temperature model by Anderberg
specimen. and Thelandersson [25].
c Cover layer of the reinforcement. (Gc /h)EC Compressive fracture energy di-
dmax Maximum aggregate size. vided by the corresponding ele-
Ec E-modulus of concrete at ambient ment size inherent in the elevated
temperature. temperature model in Eurocode 2
EciT Initial E-modulus at elevated [21].
temperatures. (Gc /h)model Compressive fracture energy
EciT,EC Initial E-modulus at elevated divided by the corresponding ele-
temperatures in Eurocode 2 [21]. ment size inherent in a given ele-
Ep∗ Slope of the descending branch in vated temperature model.
the elevated temperature model (Gc /h)LL Compressive fracture energy di-
by Li and Purkiss [23]. vided by the corresponding ele-
Es E-modulus of reinforcement at ment size inherent in the elevated
ambient temperature. temperature model by Lie and Lin
EsT E-modulus of reinforcement at el- [26].
evated temperatures. (Gc /h)LP Compressive fracture energy di-
fcm Compressive strength of concrete vided by the corresponding el-
at ambient temperature. ement size inherent in the ele-
fcT Compressive strength of concrete vated temperature model by Li
at elevated temperatures. and Purkiss [23].
fct,m Tensile strength of concrete at GcT Compressive fracture energy at el-
ambient temperature. evated temperatures.
fctT Tensile strength of concrete at el- GcT,AT Compressive fracture energy at el-
evated temperatures. evated temperatures as inherent
fy Yield strength of reinforcement at in the model by Anderberg and
ambient temperature. Thelandersson [25].
fyT Yield strength of reinforcement at GcT,EC Compressive fracture energy at el-
elevated temperatures. evated temperatures as inherent
F Yield function in ABAQUS [8]. in the model of Eurocode 2 [21].

xiii
Nomenclature

GcT,model Compressive fracture energy in- kp Parameter describing the stress-


herent i a given elevated temper- strain relationship suggested by Li
ature model. and Purkiss [23].
GcT,LL Compressive fracture energy at el- Kc Parameter determining the shape
evated temperatures as inherent of the yield surface in ABAQUS
in the model by Lie and Lin [26]. [8].
GcT,LP Compressive fracture energy at el- Kt Tangential stiffness.
evated temperatures as inherent LITS Load induced thermal strains.
in the model by Li and Purkiss P Load.
[23]. Pcr Load at which macrocracking of
Gf Tensile fracture energy at ambient concrete is initiated.
temperature. PE11 Output from ABAQUS of the
Grc
f Reinforced tensile fracture energy plastic strains in the x-direction.
at ambient temperature. ls Average crack spacing.
Gf T Tensile fracture energy at elevated p Pressure invariant in ABAQUS [8]
temperatures. s0 Minimum bond length.
h Element size. S11 Output from ABAQUS of stresses
hAT Element size corresponding to the in the x-direction.
compressive fracture energy in- t Thickness of reinforced concrete
herent in the elevated temper- specimen.
ature model by Anderberg and tF E Time in an FE-analysis.
Thelandersson [25]. ts Strength of the interaction con-
hEC Element size corresponding to the tribution as defined by Cervenka
compressive fracture energy in- et al. [17].
herent in the elevated tempera- T Temperature.
ture model of Eurocode 2 [21]. Ta Ambient temperature.
hef f Effective element size. w Displacement.
hmax Maximum element size at ambient wpeak Displacement at peak stress.
temperature. α Thermal expansion coefficient in
hmaxT Maximum element size at elevated ABAQUS [8].
temperatures. αconcrete Thermal expansion coefficient of
hmin Minimum element size at ambient concrete.
temperature. αsteel Thermal expansion coefficient of
hminT Minimum element size at elevated steel.
temperature. αts Strength level of the interaction
hmodel Element size corresponding to the contribution as defined by Feen-
compressive fracture energy in- stra and de Borst [18] (fraction of
herent i a given elevated temper- the tensile strength).
ature model. δ Displacement.
hLL Element size corresponding to the δp Plastic displacement.
compressive fracture energy in- ∆ Stress adjustment necessary
herent in the elevated tempera- in order to evaluate the com-
ture model by Lie and Lin [26]. bined concrete and interac-
hLP Element size corresponding to the tion contribution through the
compressive fracture energy in- *TENSION STIFFENING function
herent in the elevated tempera- in ABAQUS.
ture model by Li and Purkiss [23]. $ Flow potential eccentricity.
H Softening modulus. ε Strain.
k, T1 , T2 , T8 , T64 Constants describing the ε0T Strain at peak compressive stress
decay function. for concrete at elevated tempera-
tures.

xiv
Nomenclature

ε01 , ε02 , ε03 Parameters used in the compu- εs1 , εs2 Strain states used to compute the
tation of strain at peak compres- tension stiffening as per the CEB-
sive stress as defined by Terro [24]. FIB Model Code [16].
ε1 , σ1 Parameters describing the instan- εs,m Strain in the reinforcement with
taneous stress-related strain sug- tension stiffening as defined in the
gested by Anderberg and The- CEB-FIB Model Code [16].
landersson [25]. εσT Instantanious stress-related strain
εc Compressive strain. εt Tensile strain.
εin
c Inelastic strain in ABAQUS [8]. εck
t Cracked strain in ABAQUS [8].
pl
εpl
c , εt Hardening variables in ABAQUS εth Unrestrained thermal strain.
[8]. εpx Plastic strain in the x-direction.
εc0 Strain at peak tensile stress of εu Strain in the interaction contri-
concrete at ambient temperature. bution at which the yield stress
εc1 , εc,lim Constants used to define the of the reinforcement is reached at
compressive behavior as sug- ambient temperature.
gested by the CEB-FIB Model εy Strain at yield stress of reinforce-
Code [16]. ment at ambient temperature.
εc1t Strain at peak compressive stress κC , κT Internal parameters describing
as defined by Eurocode 2 [21]. the behavior at ambient tempera-
εctuT Ultimate tensile strain of concrete ture as suggested by Feenstra and
at elevated temperatures. de Borst [18].
εcu Ultimate strain of concrete at am- κe Equivalent strain corresponding
bient temperature. to peak compressive stress as sug-
ε∗cu Ultimate strain of concrete in the gested by Feenstra and de Borst
elevated temperature model sug- [18].
gested by Li and Purkiss [23]. κeT Equivalent strain corresponding
εcu1t Ultimate compressive strain as to peak compressive stress at el-
defined by Eurocode 2 [21]. evated temperatures.
εcuT Ultimate compressive strain of κuC Ultimate compressive concrete
concrete at elevated tempera- strain at ambient temperature
tures. as suggested by Feenstra and
εcuT,AT Ultimate compressive strain from de Borst [18].
the elevated temperature model κuCT Ultimate compressive concrete
by Anderberg and Thelandersson strain at elevated temperatures.
[25]. λL Initial compressive stress level.
εcuT,model Ultimate compressive strain µ Parameter for visco-plastic regu-
from a given elevated temperature larization of the concrete consti-
model. tutive equations in ABAQUS [8].
εcuT,LL Ultimate compressive strain from ν Poisson’s ratio.
the elevated temperature model ξ(T ) Decay function for material prop-
by Lie and Lin [26]. erties defined by Hertz [7].
εcuT,LP Ultimate compressive strain from ρp Reinforcement ratio in the direc-
the elevated temperature model tion of the load.
by Li and Purkiss [23]. ρq Reinforcement ratio in the direc-
εcT Compressive strain at elevated tion orthogonal to the loading.
temperatures. ρs Reinforcement ratio.
εe Elastic strain. ρs,ef f,min Minimum effective reinforce-
εp Peak strain the in compressive ment ratio for the interaction
material model by Nakamura and contribution defined by Cervenka
Higai [19] et al. [17] to be considered at
εp Plastic strain. ambient temperature.
εp0 Plastic strain corresponding to
peak compressive stress.

xv
Nomenclature

ρs,ef f,minT Minimum effective reinforce- σcu Ultimate compressive stress in


ment ratio for the interaction con- ABAQUS [8].
tribution to be considered at ele- σ̂max Maximum principle stress in
vated temperatures. ABAQUS [8].
σ Stress. σpeak Peack compressive stress.
σ1 , σ2 , σ3 Primary stress axis. σx Stress in the x-direction.
σ̂1 , σ̂2 Primary stress axis for plane σt Tensile stress.
stress. σt0 Uniaxial tensile peak stress used
σb0 /σc0 Ratio of the equibiaxial compres- for the definition of the tension
sive yield stress and the initial stiffening in ABAQUS [8].
uniaxial compressive yield stress φeq Equivalent reinforcement diame-
in ABAQUS [8]. ter.
σc0 Initial compressive yield stress φp Diameter of the reinforcement in
used in the *COMPRESSIVE the direction of the load.
HARDENING option in ABAQUS φq Diameter of the reinforcement in
[8]. the direction orthogonal to the
σcT Compressive stress. loading.
σcT Compressive stress at elevated ψ Dilation angle.
temperatures.

xvi
List of Figures

1.1 Examples of concrete subjected to elevated temperatures. . . . . . . . . . . . . . 1


1.2 Uniaxial tension test of pure concrete element with strain gauges at A, B and C,
(a), and the corresponding load-displacement diagrams, (b). Reproduced from
van Mier [5]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Stress-displacement curve in localized region, (a), to illustrate construction of
the load-plastic displacement diagram and of the fracture energy as the area
under the curve, (b). Reproduced from Pankaj [6]. . . . . . . . . . . . . . . . . . 2
1.4 Illustrations of a typical temperature variation caused by a fire, (a), of the tem-
peratures in the hot and cold phases of a fire, (b), and the strength ratio as a
function of the temperature, (c). . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.5 Descriptions of hot and cold phases of a fire, as defined by Hertz [7]. . . . . . . 4

2.1 Stress-strain relation for material undergoing hardening post-peak, (a), and ini-
tial and subsequent yield surfaces in deviatoric plane, (b). Reproduced from
Pankaj [9]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 Discrete, (a), and smeared crack, (b), approaches. Reproduced from Pankaj [6]. 8
2.3 Drucker-Prager yield criteria in the deviatoric plane for Kc = 2/3 and Kc = 1.0,
(a), and in three dimensions for Kc = 1, 0, (b). Reproduced from ABAQUS
Version 6.7 Documentation [8] and Pankaj [9], respectively. . . . . . . . . . . . . 10
2.4 Yield surface in plane stress. Reproduced from ABAQUS Version 6.7 Documen-
tation [8]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.5 Illustration of the plastic potential in relation to a yield surface. Reproduced
from Pankaj [9]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

3.1 Schematic overview of deterioration of plain concrete as the temperature is in-


creased. Based on Fletcher et al. [10] and Hertz [7]. . . . . . . . . . . . . . . . . 14
3.2 Dehydration of calcium hydroxide to calcium oxide and evaporable water caus-
ing shrinking, (a), and rehydration upon cooling of calcium oxide to calcium
hydroxide resulting in increased cracking, (b). Based on Hertz [7]. . . . . . . . . 15
3.3 Schematic overview of deterioration of reinforcement as the temperature is in-
creased. Based on Fletcher et al. [10]. . . . . . . . . . . . . . . . . . . . . . . . . 16

4.1 Illustrative load-displacement diagram explaining the concept of tension stiffen-


ing of reinforced concrete members. . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.2 Behaviour of reinforced concrete member using the CEB-FIB Model Code [16]. 21
4.3 Interaction contribution suggested by Cervenka et al. [17]. . . . . . . . . . . . . 22
4.4 Schematic plots of the stress-strain relation of pure concrete in tension, (a), and
the stress-plastic displacement diagram, (b), to illustrate the dependency of the
fracture energy on the element size, h. . . . . . . . . . . . . . . . . . . . . . . . 23

xvii
List of Figures

4.5 Combination of concrete and interaction stress contribution for different element
side lengths, h [mm]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4.6 The effect on the stress-strain curve of concrete, (a), and on the combined con-
crete and interaction contribution, (b), of snap-back of concrete for a model with
too large element side length, here h = 1000 mm, compared to model without
snap-back, h = 500 mm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.7 Stress-plastic strain diagram for concrete assuming linear softening to illustrate
the softening modulus, H. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.8 Combined concrete and interaction stress contribution for different different re-
inforced areas As [mm2 ] using the tension stiffening model by Cervenka et al.
[17]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.9 Combined concrete and interaction contribution for h = 50 mm as defined by
Cervenka et al. [17] and by Feenstra and de Borst [18]. . . . . . . . . . . . . . . 28
4.10 Examples of possible errors in the tension stiffening model by Feenstra and
de Borst [18], arising from the selection of a too large element side length, which
causes snap-back, (a), and a too low ratio of reinforcement, (b). . . . . . . . . . 28
4.11 Effect of changing the fraction of the ultimate tensile strength applied on the
interaction contribution of the tension stiffening model by Feenstra and de Borst
[18] on the combined concrete and interaction stress contribution, (a), and the
total stress-strain relation for the specimen, (b). . . . . . . . . . . . . . . . . . . 29
4.12 Compressive behaviour as defined by the CEB-FIB Model Code [16]. . . . . . . 31
4.13 Compressive post-peak fracture energies for different specimen geometries. Re-
produced from Vonk [20]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4.14 Illustrative stress-strain relationship for the compression model by Nakamura
and Higai [19]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.15 Compressive stress-strain relations for h = 100 mm, (a), and h = 500 mm,
(b), for a concrete defined by the variables in Table 4.3 model as suggested by
Nakamura and Higai [19]. The compressive fracture energies are computed based
on the compressive strength fcm , (4.10), and based on the tensile fracture energy
Gf = 0.095 N/mm, (4.11). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.16 Compressive behaviour model suggested by Feenstra and de Borst [18] plotted
using compressive fracture energies as defined by Vonk [20], Figure 4.13, and
Nakamura and Higai [19] based on the compressive strength fc , (4.10). A con-
crete grade C30 is considered and the material data of Table 4.3 are used. . . . 35
4.17 Stress-equivalent strain diagram for compression model by Feenstra and de Borst
[18] for concrete grade C30 with fracture energy by expression (4.10), element
side lengths h = 100 mm, h = 500 mm and h = 2000 mm. . . . . . . . . . . . . 36
4.18 Stress-strain diagrams for compressive behaviour as defined by the CEB-FIB
Model Code [16], Nakamura and Higai [19] and Feenstra and de Borst [18] for
concrete grade C30 with element side length h = 100 mm, (a), and h = 500 mm,
(b). The material data is taken from Table 4.3 and the compressive fracture
energy is computed based on the compressive strength as defined in (4.10). . . . 36
4.19 FE-configuration of the reinforced member considered for uniaxial load tests of
the tension stiffening and the compression model in ABAQUS. . . . . . . . . . . 38
4.20 The tension stiffening is defined in ABAQUS as the combination of the concrete
and interaction contributions and must be forced to constantly have a slope, by
subtracting ∆ from the stress at the input, defining εu . . . . . . . . . . . . . . . 38

xviii
List of Figures

4.21 ABAQUS output of load-displacement diagram in the y-direction on node 3


for the example plate subjected to uniaxial tension. The tension stiffening is
modelled as presented by Feenstra and de Borst [18], (a), and modified by ∆ =
0.01 MPa to ensure a constant presence of slope, (b). . . . . . . . . . . . . . . . 39
4.22 ABAQUS output of load-displacement diagram in the y-direction on node 3 for
the plate example subjected to uniaxial compression. The compressive properties
are modelled as presented by Feenstra and de Borst [18]. . . . . . . . . . . . . . 40
4.23 FE-configuration for numerical test element subjected to pure shear. . . . . . . 40

5.1 Comparison of the decay function for compressive strength presented by Hertz
[7] with the compressive decay function from Eurocode 2 [21] for a concrete with
siliceous, (a), and calcerous aggregates, (b). For computation of the decay of
strength as suggested by Hertz [7], equation (5.1) and the parameters of Table
5.1 are used and the reduction presented in Eurocode 2 [21] is given in Table 5.2. 43
5.2 Residual compressive strength of concrete after exposure to temperature level T ,
as presented by Eurocode 2 [21] and Hertz [7], for siliceous, (a), and calcerous,
(b), aggregates. The strength reduction presented by Hertz [7] is computed
by equation (5.1) with the parameters from Table 5.3 and the reduction from
Eurocode 2 [21] is given in Table 5.2. . . . . . . . . . . . . . . . . . . . . . . . . 44
5.3 Comparison of decay of tensile strength of concrete in the hot, (a), and the cold,
(b), phase of a fire from Eurocode 2 [21] and the method presented by Hertz [7]
with siliceous, main group and light weight aggregates. For the computations of
the strength by Hertz [7], equation (5.1) and the parameters of Table 5.1 and
Table 5.3 are used. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
5.4 Decay functions from the Eurocode [21] and Hertz [14] for hot-rolled, (a), and
cold-worked, (b), reinforcement bars when exposed to high temperatures. . . . . 45
5.5 Residual strength of cold-worked reinforcement steel after exposure to elevated
temperature level, T , as presented by Eurocode 2 [21] and Hertz [14]. . . . . . . 46
5.6 Instantaneous stress-related strain as presented by Anderberg and Thelandersson
[25] and by Lie and Lin [26] for temperatures of T = 20◦ C and T = 300◦ C, (a),
and T = 500◦ C and T = 700◦ C, (b). The ultimate stress is normalized by the
ultimate stress at ambient temperatures. . . . . . . . . . . . . . . . . . . . . . . 48
5.7 Illustration of the difference between the total strain when heated with and
without applied stress. Reproduced from Law and Gillie [27]. . . . . . . . . . . 49
5.8 Compressive stress-strain relations as defined by Li and Purkiss [23] and Eu-
rocode 2 [21] for siliceous concrete at T = 20◦ C and T = 300◦ C, (a), and
T = 500◦ C and T = 700◦ C, (b). . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.9 Tensile stress-strain relationship as suggested by Terro [24] for concrete at tem-
peratures of T = 20◦ C, T = 300◦ C, T = 500◦ C and T = 700◦ C. . . . . . . . . . 51
5.10 Example of reinforcement models at ambient and elevated temperatures for hot-
rolled reinforcement with the material characteristics of Table 4.1. . . . . . . . . 52

6.1 The compressive fracture energy is inherent in the existing elevated temperature
models for the compressive behaviour of concrete. . . . . . . . . . . . . . . . . . 56
6.2 Compressive material model by Anderberg and Thelandersson [25] and fracture
energy based formulation with an element size of h = 65 mm for concrete grade
C30 at ambient temperature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
6.3 Compressive material model by Lie and Lin [26] and fracture energy based for-
mulation with an element size of h = 300 mm for a concrete grade C30 at
ambient temperature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

xix
List of Figures

6.4 Comparison of the evolutions with temperature of the compressive fracture ener-
gies obtained when applying the methods of Anderberg and Thelandersson [25],
Lie and Lin [26], Li and Purkiss [23] and Eurocode 2 [21] to equation (6.6), for
the previously described example. . . . . . . . . . . . . . . . . . . . . . . . . . . 61
6.5 Illustration of how the tensile fracture energy changes due to the decrease of the
tensile strength, fctT , at an elevated temperature, T , compared to the strength
at the ambient temperature, fct,m . . . . . . . . . . . . . . . . . . . . . . . . . . 62
6.6 Comparison of fracture energy based tensile formulation of the tensile stress-
strain relationship of plane concrete to the model suggested by Terro [24]. An
element of size h = 16.5 mm is considered at temperatures of T = 20◦ C, T =
300◦ C and T = 500◦ C, (a), and T = 700◦ C, T = 900◦ C and T = 1100◦ C, (b). . 63
6.7 Combined concrete and interaction stress contributions for a concrete grade C30
with steel Grade 500 for a reinforced member with element size h = 100 mm. . 64
6.8 Evolution of the maximum element size, hmaxT , with temperature as defined by
equation (6.15) for an example with a reinforced concrete member of grade C30. 66
6.9 Illustration of how the modelling of the combined concrete and interaction stress
contributions at different temperatures yields unrealistic results if the reinforce-
ment ratio is too small. The temperature of the steel is assumed to be equal to
that of the concrete. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
6.10 Evolution of minimum reinforcement ratio for the example of a reinforced mem-
ber from Figure 6.9 as a function of the temperature. . . . . . . . . . . . . . . . 67

7.1 Illustration of the dimensions of the considered reinforced slab. . . . . . . . . . . 69


7.2 Illustration of the reinforced concrete slab considered in this example. . . . . . . 70
7.3 Temperature profile within the considered slab. . . . . . . . . . . . . . . . . . . 70
7.4 Overview of the time in the FE-analysis of the considered reinforced slab. . . . . 70
7.5 Thermal expansion coefficient for concrete, αconcrete , as a function of the tem-
perature for the considered example of a reinforced concrete slab. . . . . . . . . 72
7.6 Limits on the maximum and minimum element size, equation (6.15), as functions
of the temperature for the considered example of a reinforced slab. . . . . . . . 72
7.7 Verification of the requirement to the minimum level of reinforcement (equation
(6.16)) that can be considered for validity of the interaction stress contribution
of the tension stiffening for the considered example of a reinforcement slab with
element sizes of h = 73 mm, (a), and h = 129 mm, (b). . . . . . . . . . . . . . . 73
7.8 Material models for compression, (a), and tension, (b), for the reinforced slab
with an element size of h = 129 mm. . . . . . . . . . . . . . . . . . . . . . . . . 74
7.9 Position of the considered element for the post-processing of the contour plots
from ABAQUS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
7.10 Output from ABAQUS analysis of stress in the x-direction (S11) at various
times, where tF E = 1.00 coresponds to the onset of the temperature load. . . . . 75
7.11 Output from ABAQUS analysis of plastic strain in the x-direction (PE11) at
various times, where tF E = 1.00 coresponds to the onset of the temperature load 76
7.12 Position of element 2 and an indication of the location of the integration points
within it. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
7.13 Evolution of the stress and the plastic strain the x-direction in the integration
points of element 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
7.14 Stress in the x-direction through the thickness of the slab at the left fixed end
at times tF E = 1.00, (a), and tF E = 2.00, (b), for element configurations of
h = 129 mm and h = 73 mm, respectively. . . . . . . . . . . . . . . . . . . . . . 77

xx
List of Figures

A.1 Idealization of stresses around a single aggregate particle. Reproduced from


Mindess et al. [11]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
A.2 Characteristic nominal stress-deformation relation of a loaded specimen in com-
pression under displacement controlled test. Reproduced from Mindess et al.
[11]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

B.1 Definition of cracking and inelastic strain. Reproduced from the ABAQUS Ver-
sion 6.7 Documentation [8]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

D.1 Compressive concrete model, (a), and the combined concrete and interaction
stress contribution in tension, (b), for T = 20◦ C. . . . . . . . . . . . . . . . . . . 95
D.2 Compressive concrete model, (a), and the combined concrete and interaction
stress contribution in tension, (b), for T = 100◦ C. . . . . . . . . . . . . . . . . . 96
D.3 Compressive concrete model, (a), and the combined concrete and interaction
stress contribution in tension, (b), for T = 200◦ C. . . . . . . . . . . . . . . . . . 96
D.4 Compressive concrete model, (a), and the combined concrete and interaction
stress contribution in tension, (b), for T = 300◦ C. . . . . . . . . . . . . . . . . . 96
D.5 Compressive concrete model, (a), and the combined concrete and interaction
stress contribution in tension, (b), for T = 400◦ C. . . . . . . . . . . . . . . . . . 97
D.6 Compressive concrete model, (a), and the combined concrete and interaction
stress contribution in tension, (b), for T = 500◦ C. . . . . . . . . . . . . . . . . . 97
D.7 Compressive concrete model, (a), and the combined concrete and interaction
stress contribution in tension, (b), for T = 600◦ C. . . . . . . . . . . . . . . . . . 97
D.8 Compressive concrete model, (a), and the combined concrete and interaction
stress contribution in tension, (b), for T = 700◦ C. . . . . . . . . . . . . . . . . . 98
D.9 Compressive concrete model, (a), and the combined concrete and interaction
stress contribution in tension, (b), for T = 715◦ C. . . . . . . . . . . . . . . . . . 98

xxi
List of Figures

xxii
List of Tables

2.1 Input parameters used for *CONCRETE DAMAGED PLASTICITY in ABAQUS. . . . 12

3.1 Overview of the response of of the concrete and the reinforcement in reinforced
members upon exposure to a fire. . . . . . . . . . . . . . . . . . . . . . . . . . . 18

4.1 Material properties for reinforcing steel Grade 500 using the simplified material
model from the CEB-FIB Model Code [16]. . . . . . . . . . . . . . . . . . . . . . 20
4.2 Tensile material parameters for concrete grade C30 with maximum aggregate
size dmax = 32 mm [16]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.3 Parameters used for the compression model from the CEB-FIB Model Code [16]
for concrete grade C30. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.4 Compressive fracture energies in N/mm for a reinforced members of height 100
mm and 500 mm, fcm = 38 MPa and Gf = 0.095 N/mm, obtained using the
methods presented by Vonk [20] (Figure 4.13) and Nakamura and Higai [19]
(4.10 and 4.11). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

5.1 Parameters describing decay functions for concrete in the hot phase of a fire as
presented by Hertz [14]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5.2 Parameters describing the compressive behaviour of concrete at temperature T ,
(a), as defined by Eurocode 2 [21] for siliceous, (b), and calcerous aggregates, (c). 42
5.3 Parameters describing decay functions for concrete in the cold phase of a fire as
presented by Hertz [7]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5.4 Parameters describing decay functions for reinforcement in the hot phase of a
fire as presented by Hertz [14]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.5 Parameters describing decay functions for reinforcement in the cold phase of fire
as presented by Hertz [14]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

6.1 Element sizes obtained corresponding to the compressive fracture eneregies in-
herent in the elevated temperature models by Anderberg and Thelandersson [25],
hAT , Lie and Lin [26], hLL , Li and Purkiss [23], hLP , and Eurocode 2 [21], hEC ,
for the considered example. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

7.1 Parameters at ambient temperature used for the thermal analysis of concrete as
recommended by Teknisk Ståbi [29]. . . . . . . . . . . . . . . . . . . . . . . . . . 71

C.1 Output from ABAQUS for a simple shear example without reinforcement at
time increments 7, 19, 22 and 410. . . . . . . . . . . . . . . . . . . . . . . . . . . 91
C.2 Output from ABAQUS for a simple shear example with reinforcement at time
increments 7, 19, 22 and 410. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

xxiii
List of Tables

xxiv
Introduction

1.1 General

Concrete is a widely used construction material and has therefore been subjected to abundant
research. Much of this is aimed at developing accurate formulations for computer models, which
are becoming an increasingly utilized tool in the design phase of structures. A commonly used
numerical modelling method is the Finite Element (FE) analysis where the considered member
or structure is divided into smaller elements in which the response to a given load is evaluated.
The FE-model is evaluated for the tri-axial stress state by a defined yield criterion, where the
uniaxial tensile and compressive stress-strain relations determines the evolution of the criterion.
It is generally recognized that concrete is subject to localization of stresses due to the formation
of cracks which means that continued deformation upon crack initiation localizes in the formed
crack. This means that the uniaxial material models must be defined based on the size of the
elements in order to obtain convergence of the model response for different meshes.

(a) Diablo Canyon Nuclear Power Plant in San (b) The fire in the Mont Blanc Tunnel in France/I-
Luis Obispo County, California [1]. taly 1999 [2].

Figure 1.1: Examples of concrete subjected to elevated temperatures.

Concrete subjected to high temperatures are, for example, present in nuclear facilities, such as
the Diablo Canyon Nuclear Power Plant, illustrated in Figure 1.1a, or when fire occurs like the
in Mont Blanc Tunnel shown in Figure 1.1b. Hence, concrete failure at elevated temperatures
is potentially strongly related to public safety. In light of previous disasters, for example the
Chernobyl nuclear disaster (1986) [3] and more recently the Mont Blanc Tunnel fire (1999) [4]
that resulted in more than 50 and 41 fatalities, respectively, the understanding and accurate
modelling of the behaviour at elevated temperatures have gained importance. However, in spite
of the potentially large risks at stake, the knowledge base for concrete behaviour at elevated
temperatures is very limited and the existing models are subject to convergence problems in the
FE-analysis, when the same model is evaluated with different mesh sizes. Therefore, it is relevant
to take a material model formulation that at ambient temperatures is generally considered to
yield converging results and expand this to elevated temperatures.

1
Chapter 1: Introduction

1.2 Localization and Fracture Energy in Tension

Studies at ambient temperatures, among others by van Mier [5] and Pankaj [6], have shown
that as a result of the complex and highly heterogenous nature of concrete, it can be established
that the stresses and deformations occurring in concrete localizes in the formation of cracks.
A detailed describtion of crack initiation and progression at microlevel is given in appendix A.
Continued loading localizes the stresses in the formed cracks, which grow until failure occurs.
It is argued by van Mier [5] and Pankaj [6] that since the descending branch of a stress-strain
curve for concrete in tension is due to localized cracking (or strain localization), its slope cannot
be a local material property such as the E-modulus. In fact the slope of the softening branch is
a function of the specimen size. This can be illustrated by the means of a simple uniaxial tensile
test of a plain concrete member with three strain gauges; at A, B and C, as seen in Figure 1.2a.
The gauge at A measures the average strain in the region with a localized crack, whereas B
measures the strains in the uncracked part of the specimen. The gauge at C measures the strain
over the entire specimen. In this case, gauge A will indicate strain softening, gauge B unloading
and gauge C an average, as shown in Figure 1.2b.

P P

B B
C C
A A
B C B AC A

! !
(a) Uniaxial tension test (b) Load-displacement curve for gauge A, B and C.

Figure 1.2: Uniaxial tension test of pure concrete element with strain gauges at A, B and C,
(a), and the corresponding load-displacement diagrams, (b). Reproduced from van
Mier [5].

The size of the elements in a FE-model will have an effect on the material definition because the
load-displacement diagram for concrete depends on where the gauge is placed and the size of it.
As a result, the stress-strain relation cannot be taken as a material property for concrete.

"" ""

ffctm
ctm ffctm
ctm

""11 ""11

G
Gff

!! !!
pp pp
!! 11 !!uu !! 11 !!uu
(a) Stress-displacement diagram in localized (b) Stress-plastic displacement diagram in lo-
region calized region

Figure 1.3: Stress-displacement curve in localized region, (a), to illustrate construction of the
load-plastic displacement diagram and of the fracture energy as the area under the
curve, (b). Reproduced from Pankaj [6].

2
Chapter 1: Introduction

It has, however, been accepted that the stress-displacement curve can be taken as a material
invariant, see Figure 1.3a. Here, the stress-displacement relation is plotted for a concrete speci-
men with the tensile strength, fctm , and the ultimate displacement, δu . The considered stress,
σ1 , corresponds to the plastic displacement, δ1p . If plotting the plastic displacement, δ p , as a
function of the stress, σ, then the fracture energy, Gf , is defined as the area under the curve, as
indicated in Figure 1.3b. The fracture energy is the specific energy required for fracture growth
in an infinitely large specimen and, hence, the energy required to form a new fully separated
crack surface. In FE-modelling of concrete in tension, Gf is taken as the material property which
in turn implies that the decending branch of the stress-strain curve is a function of element size.
It has been shown by van Mier [5] that modelling based on Gf leads to mesh independent
results.

1.3 Localization and Fracture Energy in Compression

The considerations described above are also valid for concrete in compression, as compressive
crushing, like tensile cracking, is occuring in a localized region. However, the compressive
fracture energy based models are rarely implemented, in part because very few have investigated
or discussed the compressive fracture energy, Gc . It is, for example, not included in any of the
current codes.

1.4 Novelties and Milestones of the Thesis

Currently, reinforced concrete models are not fracture energy based at elevated temperatures.
In fact, even at ambient temperature, the existing compressive fracture energy models are rarely
implemented. As the underlaying assumption for structural modelling is that the modelled ma-
terial behaviour predicts the actual behaviour, it is evident that if this is not the case, the output
of an analysis will have little or no value. Therefore, the novelty of the current work lies in in-
vestigating the existing fracture energy based models, especially in compression, and expanding
these to elevated temperatures. While doing so, it is possibile to examine the evolution of both
the compressive and the tensile fracture energy with an increase in temperature.

Further, the limits of application imposed by the fracture energy based models at ambient tem-
perature are reviewed. As these are not currently defined, formulations of the limits are made
herein. This makes it possible to investigate how these limits evolve as functions of the temper-
ature, which is crucial to keep in mind, to ensure that they are not violated when the elevated
temperature model is applied.

Prior to extending a material model formulation to elevated temperatures, it is essential to


have knowledge about both the modelling of the behaviour of reinforced concrete at ambient
temperature and the physiochemical reactions caused by the temperature variation. Herein,
normal strength concrete is considered and for brevity, the elevated temperature caused by a
fire will be simplified into a hot and a cold phase. A typical fire course consists of a heating
phase to a certain temperature peak, followed by a cooling phase until the ambient conditions
are reached again as schematically illustrated in time-temperature plot in Figure 1.4a. Figure
1.4b shows how the hot phase refers to the reinforced concrete behaviour during exposure to
the maximum temperature of the fire and the cold phase refers to the residual behaviour after
exposure. The effect that the temperature elevation has on the strength of a considered material,
for example in the hot phase, is illustrated in Figure 1.4c, where three possible decay curves
are given; one where the strength at elevated temperatures remains as at ambient, one where

3
Chapter 1: Introduction

it decays rapidly and one intermediate. The rate of the decay depends on the physiochemical
response of the considered material to the temperature elevation.

T T T T T T f cT / f cmf cT /f fcTcm/ f cm
No decay
No No
decay
of strength
decay
of strength
ofwith
strength
Twithwith
T T
hot phase
hot hot
phase
phase

RapidRapid
decay
Rapid
decay
of decay
of of
strength
strength
with
strength
Twithwith
T T
cold phase
coldcold
phase
phase

t t t t t t T T T
heating heating cooling
heating cooling
cooling
(a) Schematic temperature vari- (b) Temperatures in the hot and (c) Schematic stength ratios as func-
ation in a typical fire course. the cold phases of a fire. tions of the temperature.

Figure 1.4: Illustrations of a typical temperature variation caused by a fire, (a), of the tem-
peratures in the hot and cold phases of a fire, (b), and the strength ratio as a
function of the temperature, (c).

As the hot phase and cold phase will be referd to in the following, they are scematically illustrated
in Figure 1.5.

Hot Phase Cold Phase


Properties of materials when a Residual material properties of a
structure or a member is exposed structure or a member after ex-
to elevated temperatures posure to elevated temperatures

Figure 1.5: Descriptions of hot and cold phases of a fire, as defined by Hertz [7].

As a result of the above, this thesis comprises the following:


• A discussion on the damaged plasticity formulation in ABAQUS used for the multiaxial
analysis of concrete (Chapter 2).
• A literary study of the physiochemical response of reinforced concrete exposed to fire
(Chapter 3).
• A study of the existing uniaxial fracture energy based behaviour models for the ambient
condition including formulations of the on the limits of application (Chapter 4). Further, a
choice of the material model formulations to expand to elevated temperatures is made and
numerical benchmark test are conducted to ensure correlation of the ABAQUS analysis
with the expected response.
• A literary study of the existing models for concrete behaviour at elevated temperatures
(Chapter 5). This includes the decay of strengths and a discussion on the formulations of
the stress-strain relationship.
• Formulations of fracture energy based uniaxial material models for reinforced concrete at
elevated temperatures (Chapter 6). This includes an investigation of the modifications of
the compressive and tensile fracture energies caused by temperature elevation. Further,
the evolution of the limits of application is studied.
• Numerical examples where the fracture energy based elevated temperature models are
implemented (Chapter 7).

4
Chapter 1: Introduction

• Concluesion and recommendation for future work (Chapter 8).

5
Chapter 1: Introduction

6
Finite Element modelling of
Multiaxial Behaviour of
Reinforced Concrete

2.1 Introduction

In FE-analysis the triaxial states of stress are evaluated and a yield criterion is used to de-
termine whether the deformation occuring in an element should be considered to be elastic or
plastic. It is generally accepted that concrete is a pressure sensitive material, which causes for
conical yield criterion in three dimensions. However, a variety of criterions exists, some more
complicated than others. Typically, a criterion that is very specific depends on several param-
eters and as each parameter to be defined is associated with a degree of uncertianty, this is
likely to accumulate. Herein, the FE-code ABAQUS Version 6.7 [8] is used for all finite element
computations and therefore this chapter commences with a description of how concrete cracking
can be considered in FE-computations, followed by a description of the model ABAQUS utilizes.

It is possible for the yield surface to change in size and shape as the plastic deformations evolve.
This is a necessity in order to account for hardening or softening behaviour in a model as
illustrated in Figure 2.1, where the uniaxial stress-strain relation is given in Figure 2.1a and the
yield surface of the initial yield point as well as a subsequent indicated yield point, is given in
in the deviatoric plane in Figure 2.1b.

Concrete has distinct strength assymetry, meaning that the uniaxial tensile and compressive
behaviours differ and, even at the ambient condition, there is still a great level of uncertainty
associated with material modelling of the uniaxial behaviours. The uniaxial tensile and com-
pressive behaviours of reinforced concrete will therefore be discussed in chapter 4.

A brief discription of how to define the uniaxial input parameters in ABAQUS is provided in
appendix B, along with explaniations of some of the ABAQUS functions used for the FE-models.

The derivation of the FE-equations will not be given and a detailed description of concrete
plasticity is also omitted as both are out of the current scope.

7
Chapter 2: Finite Element modelling of Multiaxial Behaviour of Reinforced Concrete

!1 !1
! ! Initial yield
Initialsurface
yield surface

" " !2 !2 !3 !3

Subsequent
Subsequent
yield surface
yield surface
(a) Stress-strain relation (b) Deviatoric plane

Figure 2.1: Stress-strain relation for material undergoing hardening post-peak, (a), and initial
and subsequent yield surfaces in deviatoric plane, (b). Reproduced from Pankaj
[9].

2.2 FE-Modelling of Concrete Cracking

Generally, there exist two distinctly different ways of modelling cracking in FE analysis; the
discrete and the smeared approach. The discrete approach models cracking as seperation of
elements, whereas the smeared approach models the solid cracked continuum, as described by
Pankaj [6].

In the discrete crack approach, Figure 2.2a, the nodes are separated during propagation of a
crack and each crack is therefore considered separately. The smeared crack model, illustrated
in Figure 2.2b, is a damage or plasticity model where the damage zone coincides with the
dimensions of the elements. The cracking of the concrete is therefore modelled by adjusting the
material properties in the regions of cracking or strain localisation. This can be adopted as the
cracking is assumed to consist of a set of densely populated or smeared cracks and is simulated
by altering the constitutive relation in the damaged region.

(a) Discrete crack model (b) Smeared crack model

Figure 2.2: Discrete, (a), and smeared crack, (b), approaches. Reproduced from Pankaj [6].

It is not possible to determine which type of crack modelling method that is best suited without
considering the context it is to be employed in. For example, the discrete crack approach is
difficult to use on large scale arbitrary structures as it requires a very fine mesh because the
separation takes place around the elements. This can be circumvented by redefining the original
mesh, but either way, the discrete crack approach imposes a large CPU-demand. This means
that the model will demand a lot of computer power due to the large number of computations re-

8
Chapter 2: Finite Element modelling of Multiaxial Behaviour of Reinforced Concrete

quired, and this may not be cost-effective when considering the level of accuracy that the model
predicts. For large arbitrary structures it is therefore often better suited to use the smeared
crack approach where it is possible to obtain mesh insensitive results, granted that the local
material softening law is made mesh dependent based on the fracture energy.

The concrete damage plasticity model in ABAQUS [8] is a smeared crack model in the sense that
it does not track individual macro cracks. Constitutive calculations are performed independently
at each integration point of the FE-model and the presence of cracks enters the calculations by
affecting the stress and material stiffness associated with the integration point.

2.3 Concrete Model in ABAQUS

In ABAQUS [8] it is assumed that the main two failure mechanisms are tensile cracking and
compressive crushing. When using the *CONCRETE DAMAGED PLASTICITY option the yield crite-
rion is defined and it is required to define the suboptions *CONCRETE TENSION STIFFENING and
*CONCRETE COMPRESSION HARDENING and through these, the evolution of the yield surface with
continued plastic loading.

In uniaxial tension the stress-strain relation is assumed to be linear until the failure stress, σt0 ,
which corresponds to the onset of macrocracking, is reached. This is most often followed by
softening which induces strain localization. In uniaxial compression it is also assumed that the
response is linear until the initial yield stress, σc0 , after which a plastic regime follows, typically
characterized by strain hardening until the ultimate stress, σcu , and thereafter softening. The
definition of the tension stiffening and compressive behaviour in ABAQUS is described in ap-
pendix B.

The damage model in ABAQUS [8] is based on the assumption that the uniaxial stress-strain
relations can be converted into stress-equivalent plastic strain curves and this is automatically
done from the user-provided inelastic strain data. The effective tensile and compressive cohesion
stresses are then computed to determine the current state of the yield surface that is used to
analyze multiaxial load cases.

2.3.1 Yield Surface Definition

A yield surface is a surface in the stress space enclosing the volume of the elastic region. This
means that the state of stress inside the surface is elastic, while stress states on the surface have
reached the yield point and have become plastic. Further deformation causes the stress state
to remain on the surface, as the states that lie outside are non-permissible in rate-independent
plasticity.

Several formulations of yield surface criterions exist and the Drucker-Prager yield criterion [9]
is used for concrete in ABAQUS [8], because it makes it possible to determine failure both by
normal and shear stress. It is a pressure dependent criterion based on the two stress invariants
of the effective stress tensor; the hydrostatic pressure, p, and the Mises equivalent stress, q.

It is possible for the user to somewhat determine the shape of the yield surface, by the input
parameter Kc in the *CONCRETE DAMAGED PLASTICITY function. Kc is the ratio of the second
stress invariant on the tensile median to that on the compressive median at initial yield for any

9
Chapter 2: Finite Element modelling of Multiaxial Behaviour of Reinforced Concrete

given value of the pressure invariant, p, such that the maximum principal stress is negative,
σ̂max < 0. It must be fulfilled that 0.5 < Kc ≤ 1.0 and the factor is per default 2/3, making the
yield criterion approach Rankine’s formulation [9].

!2 !2

Kc = 2/3

Kc = 1.0 !2

!3 !3

!1 !3 !1 !1 !3

(a) Yield surface in deviatoric plane (b) Yield surface in three dimen-
sions for Kc = 1.0

Figure 2.3: Drucker-Prager yield criteria in the deviatoric plane for Kc = 2/3 and Kc = 1.0,
(a), and in three dimensions for Kc = 1, 0, (b). Reproduced from ABAQUS Version
6.7 Documentation [8] and Pankaj [9], respectively.

The difference of the yield surfaces in the deviatoric plane, i.e. where σ1 + σ2 + σ3 = constant,
for Kc = 2/3 and Kc = 1.0 is shown in Figure 2.3a. For comparison, the Rankine criterion is
usually triangular whereas the Drucker-Prager criterion is circular in the deviatoric plane. Here,
Kc is set to unity, which corresponds to using the traditional Drucker-Prager yield criterion,
where the yield surface is cone shaped in the three-dimensional space as illustrated in Figure 2.3b.

Yield Function in ABAQUS

In order to account for the different evolution of strength under tension and compression, Fenve’s
modification of Lubliner’s yield function is used in ABAQUS [8]:
! $
1 " pl #
F = q̄ − 3αp̄ + β ε #σ̄ ˆmax $ − γ#−σ̄
ˆmax $ − σ̂c (εpl
c )=0 (2.1a)
1−α

where
(σb0 /σc0 ) − 1
α= for 0 ≤ α ≤ 0.5 (2.1b)
2(σb0 /σc0 ) − 1
σ̄c (εpl
c )
β= (1 − α) − (1 + α) (2.1c)
σ̄t (εpl
t )
3(1 − Kc )
γ= (2.1d)
2Kc − 1

ˆmax is the maximum principal effective stress and σb0 /σc0 is the user specified ratio of
In this, σ̄
the equibiaxial compressive yield stress and the initial uniaxial compressive yield stress, which
per default is set to 1.16.

It is seen from the expressions above, (2.1a-2.1d), that the evolution of the yield surface is
controlled by the hardening variables εpl
t and εc . The tensile and compressive stresses corre-
pl

sponding to these are computed from the input given by the tension stiffening and compression

10
Chapter 2: Finite Element modelling of Multiaxial Behaviour of Reinforced Concrete

hardening definitions.

The yield surface in plane stress is illustrated in Figure 2.4, where the enclosed area of the figure
represents the elastic states of stress. If a given member is loaded in tension in both the σ̂1 and
the σ̂2 directions, the stress state is in the first quadrant of the coordinate system. Likewise, if
it is loaded in compression in both directions, the stress state is in the third quadrant. For load
cases where a combination of tensile and compressive forces are applied (e.g. shear), the stress
states will be either in the second or the fourth quadrant.

1 ^
( q ! 3 " p + # !2 ) = !c 0
1 !"
^
! 2
COMPRESSION TENSION
TENSION TENSION
!t 0
Uniaxial tension
^
! 1
Biaxial tension
Uniaxial compression

(!b0 , !b0 ) Biaxial compression


!c 0

COMPRESSION TENSION
COMPRESSION COMPRESSION

1 1 ^
( q ! 3 " p ) = !c 0 ( q ! 3 " p + # !1 ) = !c 0
1 !" 1 !"

Figure 2.4: Yield surface in plane stress. Reproduced from ABAQUS Version 6.7 Documen-
tation [8].

Flow Potential Function in ABAQUS

Infinitely small strain increments can be divided into an elastic and plastic part, dε = dεp + dεe ,
and experimental results suggest that the plastic strain increment is normal to the yield surface
[9]. Sometimes plastic strain increments are assumed to be normal to a surface other than the
yield surface and this surface is referred to as the plastic potential and is illustrated in Figure 2.5.

The flow potential function, G(σ), used in ABAQUS [8] is the Drucker-Prager hyperbolic func-
tion given by:

%
G(σ) = ($σt0 tan ψ)2 + q̄ 2 − p̄ tan ψ (2.2)

Here, ψ is the dilatation angle, σt0 is the uniaxial tensile stress at failure from the tension stiffen-
ing definition and $ is the eccentricity that defines the rate at which the function approaches the

11
Chapter 2: Finite Element modelling of Multiaxial Behaviour of Reinforced Concrete

asymptote. Both ψ and $ are given as input parameters in the *CONCRETE DAMAGED PLASTICITY
function.
"2

d !p

Yield surface Plastic potential


F (" , Y ) = 0 G( ") = 0

"3 "1

Figure 2.5: Illustration of the plastic potential in relation to a yield surface. Reproduced from
Pankaj [9].

Input Parameters

The flow potential, yield surface and viscosity parameters for the concrete damaged plasticity
model are, as described, defined through the *CONCRETE DAMAGED PLASTICITY input.

*CONCRETE DAMAGED PLASTICITY


ψ, $, σb0 /σc0 , Kc , µ

Herein, the parameters in Table 2.1 are used for all FE-computations.

Table 2.1: Input parameters used for *CONCRETE DAMAGED PLASTICITY in ABAQUS.

ψ $ σb0 /σc0 Kc µ
31.0 0.1 1.16 1.0 0.0

The dilation angle, ψ, controls the amount of plastic volumetric strain developed during plastic
shearing and is assumed constant during plastic yielding. Typically, for normal strength con-
crete a dilation angle of ψ = 31◦ is used, and this is therefore also chosen herein.

The flow potential eccentricity is per default $ = 0.1, meaning that the material has almost the
same dilatation angle over a wide range of configuring pressure stress values.

The ratio σb0 /σc0 is set to the default value of 1.16 and it is chosen that Kc = 1.0 so that the
yield surface has a perfect cone shape in the three dimensional space, as previously described.

The viscosity parameter, µ, is used for the visco-plastic regularization of the concrete constitutive
equations. The default value is 0.0 which means that a rate-independent analysis is carried
out.

12
Response of Reinforced Concrete
to Fire Exposure

3.1 Introduction

In relation to temperature, a fire typically means an increase to high levels followed by a decay
until the amient conditions are reached again. The rate at which the elevation and decay is oc-
curring can vary considerably and depends on a number of factors, such as the type and amount
of fuel and the availability of oxygen. Herein, it is mainly the exposure to the temperature
elevation that is of concern and a detailed definition of the fire phases will therefore not be given.

The changes that reinforced concrete members undergo during fires are occuring at a micro-level
and are associated with the separate responses of the concrete and the steel reinforcement. As a
result of the different chemical composition of the two components, the response at micro-level
causes different thermal properties at macro-level. Therefore internal stresses are generated,
resulting in formation of cracks and potentially failure of the bond between the concrete and
the reinforcement. This effectively means that the material properties of the concrete and the
steel are reduced by the physiochemicak processes induced by temperature elevation, as Fletcher
et al. [10] describes.

In this chapter the chemical and physical responses to a fire of reinforced concrete members to
a fire are described. The effects of a fire on the concrete and on the reinforcement are explained
separately and the choice of analysis for the fracture energy based material model at elevated
temperatures (described in chapter 6) is discussed.

3.2 Chemical and Physical Effects of Fire

Concrete is, as described in appendix A, a heterogeneous material consisting of cement paste,


aggregate and, for reinforced concrete, steel. The response to thermal exposure of each of
these components is different in itself, and the behaviour at elevated temperatures is therefore
neither easy to define nor to model. This difficulty arises from the fact that the difference in
response of the components also affects the overall response. Fletcher et al. [10] explaines how,
for example, the thermal response of the aggregate may be straight forward, but in context it
can be substantially different.

13
Chapter 3: Response of Reinforced Concrete to Fire Exposure

3.2.1 Chemophysical Response of Concrete to Fire

According to Mindess et al. [11] the cement paste consists of a range of different chemical com-
pounds, which all react differently to exposure to high temperatures. A detailed description of
the chemical composition of the cement paste is beyond the current scope, and only the most
important compounds in the context of elevated temperatures will be considered; namely hy-
drated calcium silicate (commonly refered to as C-S-H) which typically makes up 50-60% of the
paste and calcium hydroxide (Ca(OH)2 but herein refered to as CH in accordance with common
terminology), which usually accounts for 20-25% of the solid paste volume. Finally, the water
contents of the concrete is of importance for the deterioration process at elevated temperatures.
The water contents in a given concrete member depends on the w/c (water/cement) ratio used
in the mixture of the concrete and the conditions of the curing process. The water is held in
the pores of the concrete and generally two different kinds of pores are defined; capillary pores
and gel pores. The two types are distinguished by their sizes, the size of the capillary pores
vary from 10-104 nm, whereas the gel pores are less than or equal to 10 nm in size. The water
held in the two kind of pores is also differentiated; the water in the capillary pores is considered
to be evaporable water, whereas the water held in the gel pores is regarded as a part of the C-S-H.

Due to of the complexity of the concrete composition, an exact temperature for a given chemo-
physical change in any concrete cannot be given. However, the general response of concrete to
elevated temperatures has described by various researchers, for example by Fletcher et al. [10]
and by Hertz [7].

Water starts Calcium hydroxid (CH) Feldspar melts and the minerals of the
to vaporise begins to dehydrate cement paste turn into a glass phase

100 140 400 600 1150


T [ C]
Ta 150 575 800

Chemically bound water in Aggregates starts to


hydrated calcium silicate (C!S!H) increase in volume and
initiates vaporisation and to decompose

Figure 3.1: Schematic overview of deterioration of plain concrete as the temperature is in-
creased. Based on Fletcher et al. [10] and Hertz [7].

In Figure 3.1 an overview of the most important chemical processes occurring in concrete due
to temperature rise is given and in the following these processes are elaborated upon:
• When concrete reaches 100-140◦ C, the water begins to evaporate, usually causing a buildup
of pressure within the concrete.
• Once the temperature reaches 150◦ C the chemically bound water is released from the C-S-
H and hence the cement matrix begins to dehydrate and shrink. This process has a local
peak at 270◦ C and internal stresses arise. From above 300◦ C these stresses will result in
micro cracking and hence irreversible deformations are initiated.
• At about 400-600◦ C, the CH (chemically denoted Ca(OH)2 ) in the cement begins to dehy-
drate, generating calcium oxide (CaO) and more vapor (H2 O). This dehydration process
causes the strength to decrease significantly.
• At 575-800◦ C strength loss due chemical changes of the aggregate is initiated. For quartz-
based aggregates a mineral transformation at 575◦ C causes the aggregates to increase in

14
Chapter 3: Response of Reinforced Concrete to Fire Exposure

volume and for limestone aggregates decomposition is commenced at 800◦ C; most often
causing the concrete to be crumbled to gravel.
• Above 1150◦ C feldspar melts and the remaining minerals of the cement paste turn into a
glass phase yielding high brittleness and almost no strength.

It must be noted that not only the composition of the concrete has an effect on the response, but
also environmental factors influence the chemical processes occurring at elevated temperatures.
The above process overview is for unsealed concrete, whereas the behaviour of externally moist
sealed concretes at temperatures above 100◦ C differs significantly, as Khoury [12] explains. This
is caused by the fact that the chemophysical response in unsealed conditions is dominated by
the loss of various kinds of water, whereas the process is dominated by hydrothermal chemical
reactions in sealed concrete.

Further, if the concrete is loaded in compression during heating, the loading compacts the
concrete and inhibits the development of cracks. Khoury [12] describes how this can decrease
the reduction of both the elastic modulus and the compressive strength due to temperature
effects significantly. Hertz [7] further explains that this is due to the fact that the compressive
stresses in the concrete must be unloaded before any tensile stresses can be established, and
hence before microcracking can be initiated. The strain contributions is called the load induced
thermal strains, LITS, and they are only occurring during first the heating cycle.

CaO
CaO C!S!H
C!S!H

CaO
CaO
+ H+2O
H 2O
CaO
CaO
C!S!H
C!S!H

(a) Hot phase of a fire, minimum 400- (b) Cooling phase of a fire
600◦ C

Figure 3.2: Dehydration of calcium hydroxide to calcium oxide and evaporable water causing
shrinking, (a), and rehydration upon cooling of calcium oxide to calcium hydroxide
resulting in increased cracking, (b). Based on Hertz [7].

It is necessary to emphasize that an important chemical reaction occurs after the exposure
to elevated temperatures, i.e. in the cooling phase of a fire. Figure 3.2 shows, and Hertz [7]
describes, how the calcium oxide expands during the cooling phase, as it absorbs water from the
ambient air. This process can reduce the compressive strength by another 20% after exposure
to elevated temperatures and the importance of considering the concrete strength in the cold
phase of a fire (for design purposes) must therefore be stressed.

3.2.2 Chemophysical Response of Reinforcing Steel to Fire

Nielsen [13] explains how iron is a crystalline solid, in which plastic deformations are caused
by mechanical distortions of the crystal lattice. Essentially, steel is iron with small quantities

15
Chapter 3: Response of Reinforced Concrete to Fire Exposure

of carbon added to it and the carbon atoms creates small irregularities in the lattice. The ir-
regularities inhibits the movements of the lattice by acting as anchors for the dislocations, and
thereby increasing the strength of the material but also making it more brittle.

As the temperature is related to the movement of the atoms, a temperature increase reduces
the external energy necessary to move the dislocations. Hertz [14] describes how this effectively
means that the yield stress (or the 0.2% stress) will decrease with an increase in temperature.
Moreover, within the first 200-300◦ C, a temperature increase also means that more new dislo-
cations can be formed if stress is applied. Cold-working of steel utilizes this effect to increase
the ultimate strength.

Two different types of reinforcement bars generally exists; hot-rolled and cold-worked. The dif-
ference between the two is, as Nielsen [13] explains, that cold-worked reinforcement has been
twisted, stretched or a combination of the two to obtain a more chaotic system of dislocations
and many sources for formations of new dislocations. Cold-working will therefore increase the
yield strength of the steel at ambient temperatures and the material becomes less ductile. The
effect of cold-working is permanently lost if the reinforcement is exposed to temperatures beyond
400◦ C as this is above the temperature of recrystallization for steel.

It should be noted that also pre-stressed reinforcement exists, however this is out of the current
scope and will therefore not be discussed.

Significantly larger thermal


expansion than concrete

400
T [ C]
Ta 250 300 700

Blue brittleness if Load!bearing capacity reduced


low carbon contents to about 20% of design value

Figure 3.3: Schematic overview of deterioration of reinforcement as the temperature is in-


creased. Based on Fletcher et al. [10].

The performance of steel during a fire is generally better understood than that of concrete and
has, for example, been described by Fletcher et al. [10] and Hertz [14]. An overview of the most
significant processes occurring in reinforcement as a result of increase in temperature is given in
Figure 3.3 and a brief summary is given below:

• At temperatures of 200-300◦ C, steels with low carbon contents show blue brittleness. The
steel takes a blue color and the strength of the material is increased, but the material
also loses its ductility and becomes very brittle. To avoid this, it is therefore generally
recommended that the reinforcement is protected from temperatures higher than 250-
300◦ C.

• Up until about 400◦ C the thermal expansions of steel and concrete are fairly similar, but
at higher temperatures the expansion of steel is significantly larger than the expansion of
concrete. This causes increased interface stresses and hence a great risk of bond failure.

• At temperatures in the order of 700◦ C the load-bearing capacity of steel reinforcement


will be reduced to about 20% of its design value.

16
Chapter 3: Response of Reinforced Concrete to Fire Exposure

3.3 Typical Failures of Reinforced Members

It is evident by the discussion of response to temperature changes that the concrete remains
strong at high temperatures where the steel is weak and that the properties of the steel are
regained upon cooling whereas the concrete strength is further reduced. Hertz [7] suggests to
define a hot and a cold phase of a fire for design purposes, which both must be investigated for
concrete members to determine the which of the states that is most likely to induce failure.

Fletcher et al. [10] states that structural failure in the hot phase of a fire often only occurs
due to bond failure or when the effective tensile strength of any of the steel reinforcement is
lost. However, concrete has low thermal conductivity and as a result the steel reinforcement is
effectively protected from exposure to the highest levels of temperature. It is therefore crucial
that the concrete keeps its integrity which can be lost by two mechanisms; either by extensive
cracking of the outer layer of the concrete or by spalling of the concrete surface. Khoury [12]
explains that spalling is a phenomenon involving ejections of chunks of concrete from the surface
of the material and is generated by the thermal stresses and the increased pore water-pressure
in a concrete member. It may occur under a variety of conditions where strong temperature
gradients are present. The presence of reinforcement enhances the risk of spalling, as it has a
large effect on the transport of water within a member because the water is forced around the
bars, increasing the pore pressure in some regions of the concrete. However, as normal strength
concrete is considered herein and it is assumed that microsilica is not used to densify it, it is
safe to ignore spalling, provided that the moisture content is low [15].

It is further noted by Fletcher et al. [10] that compressive failures often are associated with
temperature-related loss of compressive strength of the concrete in the compressive zone. This
type of failure is therefore most likely to arise in the cold phase of a fire.

3.4 Choice of Analysis Type

For concrete structures it is necessary to analyse the response of the entire exposed member,
as the structural effectiveness of a member is not lost until it reaches the critical temperature
where the material strength is deteriorated excessively. This due is to the fact that concrete has
low thermal conductivity and therefore strong temperature gradients are generally generated
within fire exposed concrete.

Khoury [12] argues that it is necessary to perform a thermal analysis that computes the tem-
perature distribution within the considered member for all types of analysis involving exposure
to elevated temperatures. In a simplified limit state analysis the 500◦ C isotherm, obtained by
the thermal analysis, is used to reduce the cross-section. Hereafter the load-bearing capacity is
carried out with the mechanical properties at ambient temperatures. A more accurate method
is the thermomechnical finite element analysis, where the thermal analysis is carried out for
the entire duration of the fire and then the a mechanical analysis is performed. However, as
the hydral problem of the deterioration process is simplified out of this analysis type, an exact
prediction for all types of structures cannot be obtained. Therefore, a comprehensive thermo-
hydromechanical finite element analysis has been developed, that includes a thermal, a hydral
and a mechanical analysis in a fully integrated and interactive model.

The use of the thermomechanical finite element analysis does, according to Khoury [12], predict
the response to heating and loading with reasonable accuracy for the type of concrete members

17
Chapter 3: Response of Reinforced Concrete to Fire Exposure

considered herein. The implication of a fire on a given member will therefore be modelled by the
deterioration of the material properties caused by the temperature variation. This means that
the physiochemical changes in the concrete will be simplified into deterioration of macroscopic
mechanical properties. As a results, some effects cannot be considered by the model. For
example, the determination of explosive spalling is governed by the pore pressure and because
this is not computed, spalling will not be detected. H owever, as specified in section 3.3, it is
assumed that spalling safely can be neglected.

3.5 Overview of Concepts Involved in the Response of Re-


inforced Concrete to a Fire

In Table 3.1 an overview of some of the physical concepts involved when reinforced concrete
is exposed to a fire are provided. The properties and the response of the concrete and of the
reinforcement steel is considered seperately to emphasise the significant difference between the
two, which contributes to the deterioation upon exposure.

Table 3.1: Overview of the response of of the concrete and the reinforcement in reinforced
members upon exposure to a fire.

Concrete Reinforcement

Conductivity Low High

Temperature
Must be considered due the Can be ignored because of the
gradient within the
low conductivity relatively high conductivity
material

Load type typically


Compression Tension and flexure
causing failure

In the hot phase of a fire,


In the cold phase of a fire as the strength is signifi-
Phase of fire where
as the strength is further de- cantly reduced when exposed
failure is most
creased after exposure to ele- to high temperature levels
likely to occur
vated temperatures and the strength is regained
upon cooling

Main mechanisms
causing failure of Spalling and cracking result- Strength reduction as a result
reinforced members ing in exposure of the rein- of exposure to high temper-
at high forcement tures
temperatures

18
Modelling of Uniaxial behaviour
of Reinforced Concrete at
Ambient Temperature

4.1 Introduction

As described in chapter 2, the uniaxial behaviour defines the evolution of the yield criterion in
a FE-analysis. However, the uniaxial behaviour of plain and reinforced concrete is associated
with a great degree of uncertainty at ambient temperatures.

At ambient conditions, it is generally recognized that in order to obtain mesh independent re-
sults of models of reinforced concrete in FE-analysis, a fracture energy based material model
must be adopted. In tension, such models are widely used and in most FE-codes, ABAQUS
for example, it is possible to define the tensile post-peak behaviour in three ways; through an
element size dependent stress-strain relation, through a stress-displacement formulation or by
giving the tensile fracture energy and letting ABAQUS define the behaviour. In compression
the fracture energy based behaviour models are less used and the compressive fracture energy
is, for example, not discussed in any current codes and it is generally examined by very few. To
apply a fracture energy based compressive model in a FE-analysis, an element size dependent
stress-strain formulation must be used.

As the scope herein is to extend the current formulations for the ambient condition to ele-
vated temperatures the existing models are examined. The evaluation of the models includes
investigating their limits of application, which poses demands on the following:
• The minimum element size.
• The maximum element size.
• The minimum amount of reinforcement that can account for the interaction contribution
for the modelling of the tension stiffening.
Of the above points, the limitation on the maximum element size is widely recognized whereas
the demand for a minimum element size and a minimum reinforcement ratio is novel research.

A discussion of how the strength level of the tension stiffening contribution influences the overall
behaviour of a reinforced member is given and, to conclude the chapter, a brief summary of the
chosen concrete model is provided. This is followed by three numerical benchmark examples
of a simple plate configuration subjected to uniaxial tension, uniaxial compression and pure
shear.

19
Chapter 4: Modelling of Uniaxial behaviour of Reinforced Concrete at Ambient Temperature

4.2 Material Model of Reinforcement

As the material model used for the reinforcement affects the tensile behaviour of reinforeced
concrete, a brief discription of the reiforcement model considered herein is given prior to the
discussion on the tensile behaviour of reinforced concrete.

The reinforment model used herein is based on the CEB-FIB Model Code [16] as this aims
at combining research and technical finding and translating these to practical purposes. A
simple bi-linear elasto-plastic model is therefore employed with steel Grade 500, with elastic
performance until yield sets in at fy , after which perfectly plastic deformations occur. Further,
it is assumed that the behaviour of the reinforcement is equal in tension and in compression and
that all reinforcement is in the form of either rods or grids of steel.

Table 4.1: Material properties for reinforcing steel Grade 500 using the simplified material
model from the CEB-FIB Model Code [16].

fy Es ν
500 MPa 200 GPa 0.3

4.3 Reinforced Concrete in Tension

The behaviour of reinforced concrete in tension depends on the state of the concrete; cracked or
not cracked. Once cracking has commenced, the load-bearing ability of the structure is shifted
from the concrete to the reinforcement. However, this does not occur abruptly as all the concrete
is not cracked simultaneously. Due to the presence of reinforcement the uncracked concrete con-
tinues to carry additional forces and, as a result, there is thus no reduction of the load carrying
capacity in the composite. The contribution to the stiffness of the structure by the uncracked
part of the concrete is known as tension stiffening and it decays as the stress increases due to the
continued formation of cracks. Some tension stiffening is always present until the reinforcement
starts to yield.

The load-displacement response of a reinforced concrete specimen in uniaxial tension is shown in


Figure 4.1. The figure displays the response with and without tension stiffening and illustrates
how the strength level of the tension stiffening effect influences the load-displacement relation.

The tension stiffening effect can be modelled in several ways, often using either a linear or
an exponential model. In the following some of the procedures recommended in literature are
discussed.

20
Chapter 4: Modelling of Uniaxial behaviour of Reinforced Concrete at Ambient Temperature

P Concrete does
not crack

Pservice Tensile stiffening, !"

Concrete is assumed
to be fully plastic
Pcr Concrete has no
tensile strength after
cracking
Pure reinforcement

Figure 4.1: Illustrative load-displacement diagram explaining the concept of tension stiffening
of reinforced concrete members.

4.3.1 Tension Stiffening Model as per the CEB-FIB Model Code

In this thesis, a concrete of grade C30 is considered for the computations. The material param-
eters for such a concrete in tension, suggested in the CEB-FIB Model Code [16], are given in
Table 4.2 for a maximum aggregate size of dmax = 32 mm.

Table 4.2: Tensile material parameters for concrete grade C30 with maximum aggregate size
dmax = 32 mm [16].

fct,m Ec Gf
2.9 MPa 33.5 GPa 0.095 N/mm

The tension stiffening model in the CEB-FIB Model Code [16] defines the strain in the reinforce-
ment, εs,m , as a piecewise linear function. Two states are defined; State I and State II-Naked,
corresponding to the behaviour of the uncracked section and the behaviour when concrete has
no contribution in tension, and the strain is computed on the basis of these two. The strains of
the two states are denoted εs1 and εs2 , respectively.

600
.*-/01-%/.0!!*-0-2
500 *!!*&.%0-&#"+*+, ! s,m

400
" [MPa]

300
"-&'(&)*+, ! !"##$%&'(&)*+,% !
s1 s2
200
!
s,m
100 !
s1
!
s2
0
0 0.5 1 1.5 2 2.5
! !3
x 10

Figure 4.2: Behaviour of reinforced concrete member using the CEB-FIB Model Code [16].

21
Chapter 4: Modelling of Uniaxial behaviour of Reinforced Concrete at Ambient Temperature

Figure 4.2 shows the stress-modified steel strain diagram for a reinforced concrete member in
tension. Here, a plate of length and width b = 1000 mm and thickness t = 10 mm is considered,
with a total area of the reinforcement of As = 300 mm2 and hence a reinforcement ratio of
ρs = 0.03. The material parameters from Table 4.1 and Table 4.2 are used for the reinforcement
and the concrete, respectively. It is seen that the stiffness of the reinforced member is gradually
shifted from the concrete to the reinforcement as cracking progresses.

In a FE-model, the tension stiffening input is given in the concrete definition and this is done
through a stress-strain diagram, as described in appendix B.1. However, as discussed in sec-
tion 1.2, a stress-strain diagram cannot be defined as a material property for concrete as it is
dependent upon the chosen element size. It is therefore necessary to look at possible ways of
incorporating the tension stiffening based on fracture energy.

4.3.2 Tension Stiffening Model by Cervenka et al.

Cervenka et al. [17] proposed that the tension stiffening effect should not be considered as a part
of the concrete constitutive law, but rather as a separate interaction contribution. This means
that the total stress in a member becomes a sum of the three stress contributions from the
concrete, the reinforcement and the interaction. This makes it possible to define the concrete
stress contribution, as well as the tension stiffening effect, by the tensile fracture energy and the
element side length, thus making the FE-model mesh insensitive.

The interaction contribution was defined as a trilinear function, where the first incline corre-
sponds to the decreasing branch of the tension softening of the concrete. The descending part
starts when yielding of the reinforcement begins and in between these two, a constant part equal
to a fraction of the tensile strength, ts , is given. It is convenient to define the strength of the
interaction contribution as a fraction of the tensile strength, i.e.. ts = αts fct,m where αts can
vary between 0 and 1. Cervenka et al. [17] use a constant value of αts = 0.4.

#
#
Tension stiffening Steel
Tension stiffening Steel
Concrete
Concrete

ts
ts

"
(a) Stress-strain diagram for reinforced concrete member
"

#
#
t s = !ts f ct,m
t s = !ts f ct,m

"c0 " cu "u "y "


"c0 " cu "u "y "
(b) Interaction contribution

Figure 4.3: Interaction contribution suggested by Cervenka et al. [17].

22
Chapter 4: Modelling of Uniaxial behaviour of Reinforced Concrete at Ambient Temperature

In Figure 4.3a a schematic stress-strain diagram for a reinforced concrete member is given, and
in Figure 4.3b the interaction contribution is illustrated. Here, εc0 is the strain at which the
concrete reaches its maximum tensile strength, εcu is the ultimate concrete strain, i.e. when the
concrete contribution is zero, εu is the strain at which the yield stress for the reinforcement is
reached in the member and εy is the strain at which the reinforcement starts to yield. The four
strain components are defined by the following: The strain at peak stress is given by:

fct,m
εc0 = (4.1a)
Ec

The ultimate concrete strain, εcu , can be determined if considering the tension stiffening effect
from the softening of the plain concrete as a function of the fracture energy, Gf , and element
side length, h.

"" ""
f ct,m
f ct,m f ct,m
f ct,m

GGf f

! !c 0c 0 ! !cucu !! ! !cucuhh ##p


(a) Stress-strain diagram (b) Stress-plastic displacement dia-
gram

Figure 4.4: Schematic plots of the stress-strain relation of pure concrete in tension, (a), and
the stress-plastic displacement diagram, (b), to illustrate the dependency of the
fracture energy on the element size, h.

In Figure 4.4a a schematic stress-strain diagram for plain concrete in tension is given, assuming
a linear post-peak softening. This can be transferred into a stress-plastic displacement diagram,
as done in Figure 4.4b, simply by multiplying the ultimate concrete strain, εcu , with the length
of the considered element, h. Beacause the fracture energy is defined as the area under the
stress-plastic displacement diagram, which is seen in Figure 4.4b to be Gf = 1/2εcu hfct,m , an
expression for εcu can easily be obtained as:

Gf
εcu = 2 (4.1b)
hfct,m

Cervenka et al. [17] defines the ultimate strain as:

αts fct,m
εu = ε y − (4.1c)
ρs,ef f Es

The yield strain of the reinforcement is given by:

fy
εy = (4.1d)
Es

Herein, it has become evident the that model suggested by Cervenka et al. [17] is prone to some
limitations on the element size and the minimum amount of reinforcement that must be present
for the interaction contribution to be considered. In the following, expressions for the validity
range are formulated.

23
Chapter 4: Modelling of Uniaxial behaviour of Reinforced Concrete at Ambient Temperature

Minimum Element Size

A clear dependency of the ultimate concrete strain on the element length is seen in equation
(4.1b). This means that the slope of the softening branch of the stress-strain curve decreases
with decreasing h and the ultimate concrete strain is increased. It has been found herein, that
this imposes a requirement on the minimum element side length as the ultimate concrete strain
cannot be more than the ultimate strain of the reinforced member, εu .

"" ""
h h= =500
500 h h= =150
150
2.5
2.5 h h= =250
250 2.5
2.5 h h= =5050
h h= =150
150
22 22
" [MPa]
" [MPa]

" [MPa]
" [MPa]
1.5
1.5 1.5
1.5

11 11

0.5
0.5 0.5
0.5

00 00
00 0.5
0.5 11 1.5
1.5 22 2.5
2.5 00 0.5
0.5 11 1.5
1.5 22 2.5
2.5
!! !"!" !! !"!"
x x1010 x x1010
(a) εcu < εu for h = 500 mm, h = 250 mm and (b) εcu > εu for h = 50 mm
h = 150 mm

Figure 4.5: Combination of concrete and interaction stress contribution for different element
side lengths, h [mm].

In Figure 4.5 examples of the combined concrete and interaction contributions are given for
different values of h for a specimen with height and side length b = 1000 mm, thickness t = 100
mm and reinforced in one direction with a total area of As = 300 mm2 . Further, the material
properties from Table 4.2 are used. Figure 4.5a shows, that by decreasing h, the ultimate
concrete strain is increased and Figure 4.5b illustrates the situation where εcu > εu . As it is not
possible for the ultimate concrete strain to exceed the ultimate strain, it is necessary to define a
lower limit for the element side length, h. Herein, this is done by equating the ultimate concrete
strain (4.1b) with the ultimate strain (4.1c), yielding:

2Gf
hmin = ! $ (4.2)
αts fct,m
fct,m εy −
ρs,ef f Es

For the above example, equation (4.2) yields a minimum element side length of hmin = 116
mm, which is violated by choosing an element side length of h = 50 mm as shown in Figure
4.5b.

Maximum Element Size

It is also possible to choose an element size that is too large, a situation that will cause a snap-
back on the plain concrete stress-strain diagram as illustrated in Figure 4.6a. Exactly the same
parameters are used for the example in Figure 4.6 as in the previous example, apart from the
element side length which is increased to h = 1000 mm.

24
Chapter 4: Modelling of Uniaxial behaviour of Reinforced Concrete at Ambient Temperature

++ ++
h)1)*!!
h)1)*!! h)1)*!!
h)1)*!!
#"*#"* h)1)'!!!
h)1)'!!! #"*#"* h)1)'!!!
h)1)'!!!

## ##
" ),-./0
" ),-./0

" ),-./0
" ),-./0
'"*'"* '"*'"*

'' ''

!"*!"* !"*!"*

!! !!
!! !"#!"# !"$!"$ !"%!"% !"&!"& '' '"#'"# !! !"*!"* '' '"*'"* ## #"*#"*
!! !$!$ !! !+!+
()'!
()'! ()'!
()'!
(a) Concrete (b) Concrete and interaction

Figure 4.6: The effect on the stress-strain curve of concrete, (a), and on the combined concrete
and interaction contribution, (b), of snap-back of concrete for a model with too
large element side length, here h = 1000 mm, compared to model without snap-
back, h = 500 mm.

The excessively large choice of h causes snap-back and this is a widely recognized problem. It
can be avoided by controlling the softening modulus, H, as suggested by Pankaj [6].

"
fct,m

! cu !p

Figure 4.7: Stress-plastic strain diagram for concrete assuming linear softening to illustrate
the softening modulus, H.

If the softening is assumed to be linear, as illustrated in Figure 4.7, H can be expressed as


a function of the tensile strength, fct,m , and the ultimate concrete strain, εcu . By using the
definition of εcu from equation (4.1b), the following expression for the softening modulus can be
obtained.

2
hfct,m
H=− (4.3)
2Gf

To avoid snap-back for softening plasticity, the softening modulus must be limited by the elastic
modulus of the concrete as it is ultimately the difference of the cracking strain of concrete, εc0 ,
and the ultimate concrete strain, εcu , that needs to be limited to ensure that εcu − εc0 ≥ 0. In
terms of H, this means that the following inequality must be fulfilled.

− Ec ≤ H ≤ 0 (4.4)

Thus, by rewriting (4.3) and the left part of (4.4) an expression for the maximum element side
length is found.

25
Chapter 4: Modelling of Uniaxial behaviour of Reinforced Concrete at Ambient Temperature

2Ec Gf
hmax = 2 (4.5)
fct,m

For the example in Figure 4.6, equation (4.5) yields a maximum element side length of hmax =
757.

Minimum Reinforcement Ratio

An equally critical situation arises if the ratio of reinforcement is so low that the ultimate concrete
strain is higher than the ultimate strain, εcu > εu . If this is the case, the interaction contribution
cannot be accounted for in the tension stiffening definition. The previously described example is
used to illustrate the phenomenon in Figure 4.8, where the element side length is held constant
at h = 100 mm.
"
A = 400
s
2.5 A #$#"%&
s
2 A = 250
s
" [MPa]

1.5

0.5

0
0 0.5 1 1.5 2 2.5
! !"
x 10

Figure 4.8: Combined concrete and interaction stress contribution for different different rein-
forced areas As [mm2 ] using the tension stiffening model by Cervenka et al. [17].

A minimum effective reinforcement ratio is defined as a function of h by equating expressions


(4.1b) and (4.1c):

αts fct,m
ρs,ef f,min = ! $ (4.6)
2Gf
εsy − Es
hfct,m

For the current example (h = 100 mm), it would mean that a minimum effective reinforcement
ratio of ρs,ef f = 3.15 · 10−3 , and hence As,min = 315 mm2 , would be required.

As a result of the above discussion, it has been established herein, that the tension stiffening
model by Cervenka et al. [17] is only valid for a reinforced concrete model that fulfills the three
described requirements. To summarize, these are the previously recognised limit on the maxi-
mum element size, (4.5), and the two limitations formulated herein; the limit on the minimum
element size, (4.2), and the limit on the minimum reinforcement ratio, (4.6).

4.3.3 Tension Stiffening Model by Feenstra and de Borst

The tension stiffening model by Feenstra and de Borst [18] is also based on the fundamental
assumption of the tension stiffening being defined as an interaction contribution. The issue of

26
Chapter 4: Modelling of Uniaxial behaviour of Reinforced Concrete at Ambient Temperature

the limit on the minimum length of the element size has however been tackled by defining a
modified fracture energy for reinforced concrete. Further, a specific value for the stress level of
the interaction contribution is not given but it is left as a variable, αts .

Fracture Energy for Reinforced Concrete

In reinforced concrete, the cracking process stabilizes after a number of cracks are developed
and at this instant the crack spacing is determined by the amount of reinforcement. In the
study by Feenstra and de Borst [18], it is taken into account that the average crack spacing, ls ,
is generally much larger than the element side length, h, by introducing a reinforced fracture
energy, Grc
f , released over the equivalent length, h.

! $
h
Grc
f = min Gf , Gf (4.7)
ls

The average crack spacing is defined as a function of the reinforcement ratio, position and the
diameters of the rebars used, but is independent of the equivalent length, h. It is determined
by initially computing the equivalent reinforced diameter for a member that is reinforced in the
orthogonal directions p and q:
φp ρp + φq ρq
φeq = (4.8a)
ρp + ρq
The effective element size can then be evaluated by the following:
 ! $
 φ
2.5 c + eq
hef f = min 2 (4.8b)

 t
2
In this, c is the cover layer of the reinforcement. Once the effective element size has been
determined, the effective reinforcement ratio can be found.
As As
ρs,ef f = = (4.8c)
Ac,ef f bhef f

Finally, the average crack spacing is determined by:


! $
2 φs
ls = 2s0 + (4.8d)
3 ρs,ef f

Here, s0 is the minimum bond length which usually is taken as 25 mm. This value is therefore
also chosen herein.

In the following examples the validity range for the model by Feenstra and de Borst [18] is ex-
plored for a plate that is reinforced in the p direction, which is the direction the load is applied
in. The same total steel area is used as in the examples of the model by Cervenka et al. [17],
As = 300 mm2 , and it is used that φp = 1.95 mm, yielding a cover layer thickness of c = 499.5
mm. This yields an average crack spacing of ls = 250.5 mm.

If using the same example as done for the model by Cervenka et al. [17], and taking into con-
sideration the layer of reinforcement, the combined concrete and interaction stress contribution
is obtained, as shown in Figure 4.9.

27
1 1

0.5 0.5

0 0
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5
Chapter 4: Modelling of Uniaxial behaviour of Reinforced
! Concrete
x 10
!" at Ambient Temperature ! !"
x 10

"
Feenstra
2.5 Cervenka

" [MPa]
1.5

0.5

0
0 0.5 1 1.5 2 2.5
! !"
x 10

Figure 4.9: Combined concrete and interaction contribution for h = 50 mm as defined by


Cervenka et al. [17] and by Feenstra and de Borst [18].

From Figure 4.9 it is seen how the minimum limit on side length is increased. In fact it is found
that by the introduction of the reinforced fracture energy, the requirement for the minimum
element side length is no longer needed.

However, with regards to the limit on element side length, the modification of the fracture en-
ergy only tackles the problem of small element sizes, and does not take into consideration the
errors created by choosing an excessive value of h. Therefore, it is still necessary to implement
the herein formulated equation (4.5).

Further, it is still possible to have a reinforcement ratio that is too low. The reinforced fracture
energy complicates the definition of a minimum reinforcement ratio as a function of the element
side length as it makes the expression iterative. Therefore, it is much more convenient for
practical purposes simply to check that the following inequality is fulfilled for the chosen element
size, h:
εcu ≤ εu (4.9)

+" +"
h)1)*!! A = 400 A = 400
h)1)*!!
s s
2.5
#"* h)1)'!!! A = 200 2.5
#"* h)1)'!!!
A = 250
s s
#2 A = 85 #2 A = 120
s s
" ),-./0

" ),-./0
" [MPa]

" [MPa]

1.5
'"* 1.5
'"*

'1 '1

0.5
!"* 0.5
!"*

!0 !0
!0 !"# 0.5 !"$ 1!"% 1.5
!"& ' 2 '"# 2.5 !0 0.5
!"* '1 1.5
'"* #2 2.5
#"*
!! !$!" !! !+!"
x 10
()'! x 10
()'!
(a) Concrete stress contribution for constant (b) Combined concrete and interaction stress
As"= 300 mm2 with h = 500 mm and 1000 contribution for h = 100 mm and As =
mm, respectively. de Borst {120; 250; 400} mm2 .
2.5 Cervenka
Figure 4.10: Examples of possible errors in the tension stiffening model by Feenstra and
2
de Borst [18], arising from the selection of a too large element side length, which
" [MPa]

1.5 causes snap-back, (a), and a too low ratio of reinforcement, (b).

1
28
0.5

0
0 0.5 1 1.5 2 2.5 "
! !"
x 10
Chapter 4: Modelling of Uniaxial behaviour of Reinforced Concrete at Ambient Temperature

Figure 4.10 shows examples of the concrete stress contribution if an excessive element side
length is chosen and the combined concrete and interaction contribution for different levels of
reinforcement for h = 100 mm. It is seen, by comparing Figure 4.10a to Figure 4.6a, that the
problem for large values of the element side length is unchanged, and hence that the requirement
for the maximum element side length is still given by (4.5). Figure 4.10b illuatrates the problem
of a reinforcement ratio that is too small. However, by the introduction of the reinforced fracture
energy, the limitation of the application is lowered. For the previously discussed example a
minimum steel area of As = 125 mm2 is found by iteration. It is used with the reinforced
member length b = 1000 mm, thickness t = 100 mm, and an element side length h = 100 mm.
The material properties of Table 4.2, a minimum bond length of s0 = 25 mm are used and the
cover layer is taken as a function of the diameter of the reinforcement c = (t − φp )/2.

Influence of Fraction on Strength of Interaction Contribution

The level of stress defined in the interaction contribution can be varied by changing the fraction
of the tensile concrete strength, αts . In Figure 4.11 the effect of increasing αts from 0.0 to 1.0
is shown for a specimen with length b = 100 mm, thickness t = 10 mm, As = 7 mm2 , thus
ρs = 0.007 and a constant element side length of h = 100 mm.

33
500
500
2.52.5
400
400
22
σ [MPa]
σ [MPa]

σ [MPa]
σ [MPa]

300
300
1.51.5

11 200
200

0.50.5 100
100

00 00
00 0.50.5 11 1.51.5 22 2.52.5 00 0.50.5 11 1.51.5 22 2.52.5
ε ε −3 −3 ε ε −3 −3
x 10
x 10 x 10
x 10
α α= 0.0
= 0.0 α α= 0.7
= 0.7 α α= 0.0
= 0.0 α α= 0.7
= 0.7
ts ts ts ts ts ts ts ts
αtsα=ts 0.4
= 0.4 αtsα=ts 1.0
= 1.0 αtsα=ts 0.4
= 0.4 αtsα=ts 1.0
= 1.0

(a) Combined concrete and interaction stress con- (b) Stress-strain diagram for combined concrete,
tribution interaction and steel

Figure 4.11: Effect of changing the fraction of the ultimate tensile strength applied on the
interaction contribution of the tension stiffening model by Feenstra and de Borst
[18] on the combined concrete and interaction stress contribution, (a), and the
total stress-strain relation for the specimen, (b).

Figure 4.11a shows that αts controls the strength level of the interaction contribution and how it
is increased for increasing values of αts . As expected, and seen in Figure 4.11b, the stress-strain
relation for the specimen is tri-linear for αts = 1.0, which is the level suggested by Feenstra
and de Borst [18]. However, this is considered to be an unreasonable model of the tension
stiffening effect, as the concrete maintains at full strength after onset of cracking. Further, on
the stress-strain curve for the reinforced member with αts = 0.4, which was the value suggested
by Cervenka et al. [17], it is seen that the shift from the concrete contribution to the interaction
contribution causes a slight decrease in strength. An intermediate value of αts = 0.7 is taken in
the following for the level of strength on the interaction contribution.

29
Chapter 4: Modelling of Uniaxial behaviour of Reinforced Concrete at Ambient Temperature

4.4 Compressive Behaviour of Concrete

As there is no concept equivalent to tension stiffening in compression, the compressive behaviour


of reinforced concrete is simply a combination of the stress contributions of the plain concrete
and the contribution of reinforcement. As a result, this section contains compressive models for
plain concrete, which for the reinforced case simply must be added to the steel contribution.
The compression model from the CEB-FIB Model Code [16] is described in section 4.4.1 and is
followed by a overview of the concept of compressive fracture energy in section 4.4.2. Further,
the compressive fracture energy based models suggested by Nakamura and Higai [19] and by
Feenstra and de Borst [18] are discussed in sections 4.4.3 and 4.4.4, respectively. The three
compression models are compared in section 4.4.5.

According to usual sign convention, compressive stresses and strains are negative, but in this
section, they are considered to be positive. This is chosen to avoid confusion, as the compressive
properties must be entered as positives in the FE-code utilized herein, as described in appendix
2.

4.4.1 Compression Model in CEB-FIB Model Code

In the CEB-FIB Model Code [16], the material data for a range of different concrete grades in
compression is given. Herein, it is chosen to focus on a concrete of grade C30 and the relevant
parameters for the compression model are given in Table 4.3.

Table 4.3: Parameters used for the compression model from the CEB-FIB Model Code [16] for
concrete grade C30.

fcm Eci εc1 εc,lim


38 MPa 33.5 GPa 2.2·10 −3
3.3·10−3

The compression model in the CEB-FIB Model Code [16] is described by two functions, one
prior to and after the compressive strain, εc,lim , which has no significance for anything other
than computational purposes.

Figure 4.12 shows that the CEB-FIB Model Code [16] models the compressive behaviour as
somewhat linear until a certain yield point, after which hardening is evident until the peak
strength is reached. This peak is followed by a clear softening branch.

30
Chapter 4: Modelling of Uniaxial behaviour of Reinforced Concrete at Ambient Temperature

'! '!
Gc -+./01231fc
Gc -+./01231Gf
&! &!

" ()*+,

" ()*+,
%! %!

$! $!

! !
! !"!!# !"!$ !"!$# !"!% !"!%# !"!& ! !"!!# !"!$ !"!$# !"!% !"!%# !"!
! !

Figure 4.12: Compressive


'!
behaviour as defined by the CEB-FIB Model
'!
Code [16].
Gc 452617238 9:;!<=;
>+5+8?6+1@fcA
However, the compression model &! in the CEB-FIB Model Codec
[16] is based
Gc -+./01231f
&! on a stress-strain 0/1;25.B1@fcA
relation, and therefore the problems with localizationGcof-+./01231G
deformations,
f
described in section 1.2,
are bound to arise, resulting in a mesh sensitive FE model. As a result, it is necessary to
" ()*+,

" ()*+,
investigate the possibility of using
%! a compressive fracture energy based material %! model, which
is element-size dependent.
$! $!

4.4.2 Compressive Fracture


!
Energy !
! !"!!# !"!$ !"!$# !"!% !"!%# !"!& ! !"!!# !"!$ !"!$# !"!% !"!%# !"!
! !
In tension the fracture energy, Gf , and the equivalent length has widely been used for FE
models. In compression, on the other hand, the use of the compressive fracture energy, Gc , and
the equivalent length, h, is less established. In this section an overview of two distinctly different
definitions of the compressive fracture energy is given, proposed by Vonk [20] and Nakamura
and Higai [19], respectively. Common for the two, is that the compressive fracture energy is
considered to be constant for a given reinforced member.

Compressive Fracture Energy by Vonk

It was suggested by Vonk [20] that the compressive fracture energy consists of two contributions;
a local and a continuum. The local fracture energy is considered to be constant, whereas the
continuum accounts for the size effects that were found to be significant in compression. The
continuum is considered to be linearly increasing with the specimen height as shown in Figure
4.13. This approximation is arrived at by linear regression of test results with three different
specimen heights.

If considering a specimen of height 100 mm, Figure 4.13 yields a compressive fracture energy of
Gc = 15 N/mm and for a specimen height of 500 mm a value of Gc = 41 Nmm/mm2 is obtained
by extrapolating the regression line. However, it must be stressed that as the spread of the test
results evidently was increasing for increased specimen heights, and that no tests were carried
out for members taller than 200 mm, an extrapolation is very crude and a great uncertainty is
associated with compressive fracture energies for larger specimens.

However, it should be stressed that the defintion of the fracture energy requires the term to be
a material constant. Therefore, it can be questioned whether the specimen size-dependency of
the compressive fracture energy induces that the expression by Vonk [20] is really not a fracture
energy at all.

31
Chapter 4: Modelling of Uniaxial behaviour of Reinforced Concrete at Ambient Temperature

30

25

Gc [Nmm/mm2]
20

15
Continuum Fracture Energy
10

5 Local Fracture Energy

0
0 50 100 150 200 250
Specimen height [mm]

Figure 4.13: Compressive post-peak fracture energies for different specimen geometries. Re-
produced from Vonk [20].

Compressive Fracture Energy as per Narakuma and Hiagai

In the study by Nakamura and Higai [19], experiments were carried out to investigate the na-
ture of the compressive fracture energy for plain concrete, i.e. its dependency on the specimen
size and shape, aggregate size and grading and compressive strength. It was found that the
compressive fracture energy is independent of the size and shape of the test specimens when
the aggregate grading is the same, and hence that when the aggregate grading is constant, the
compressive fracture energy can be assumed as a material property. This is distinctly different
from conclusion of Vonk [20], where the specimen height is stated to have an effect on the com-
pressive fracture energy.

It was further concluded by Nakamura and Higai [19] that the compressive fracture energy
is a function of the compressive strength of the concrete and they provide the following rela-
tion:
%
Gc = 8.8 fcm (4.10)

Here, Gc is the compressive fracture energy and fcm the compressive strength (in MPa).

Further, a linear relation between the compressive and tensile fracture energies is suggested
as:
Gc = 250Gf (4.11)

For a given reinforced member with concrete grade C30, the compressive fracture energies ob-
tained from Figure 4.13, expressions (4.10) and (4.11) are compared for member heights of 100
mm and 500 mm in Table 4.4.

Table 4.4: Compressive fracture energies in N/mm for a reinforced members of height 100 mm
and 500 mm, fcm = 38 MPa and Gf = 0.095 N/mm, obtained using the methods
presented by Vonk [20] (Figure 4.13) and Nakamura and Higai [19] (4.10 and 4.11).

Member height Figure 4.13 Eq. (4.10) Eq. (4.11)


100 mm 15 54 24
500 mm 41 54 24

32
Chapter 4: Modelling of Uniaxial behaviour of Reinforced Concrete at Ambient Temperature

From Table 4.4 it is seen that the two expressions from Nakamura and Higai [19] are not
consistent for this grade of concrete. As expression (4.11) is arrived at by comparing the obtained
compressive fracture energies from expression (4.10) to the tensile fracture energies of the test
specimens, expression (4.10) is here taken for the computations of the compressive fracture
energy. It is further seen, that the compressive fracture energies obtained by Vonk [20], which
herein are argued not to be fracture energies at all, are much smaller than the ones observed by
Nakamura and Higai [19], using (4.10).

4.4.3 Compression Model of Narakuma and Higai

In addition to the methods for computing the compressive fracture energy, Nakamura and Higai
[19] also suggested a compressive behaviour model for concrete.

"
f cm

Gc / h

Ec Ec

!p !

Figure 4.14: Illustrative stress-strain relationship for the compression model by Nakamura and
Higai [19].

This model consists of a parabola prior to peak stress and a linear decrease after as indicated
schematically in Figure 4.14. The compressive fracture energy is considered to be the area under
the stress-plastic displacement curve, and the indicated area in Figure 4.14 is therefore equal to
Gc /h.

'! '! '!


Gc -+./01231fc Gc -+./01231fc
Gc -+./01231Gf Gc -+./01231Gf
&! &! &!
" ()*+,

" ()*+,
" ()*+,

%! %! %!

$! $! $!

! ! !
$ !"!$# !"!% !"!%# !"!& ! ! !"!!#
!"!!# !"!$
!"!$ !"!$#
!"!$# !"!%
!"!% !"!%#
!"!%# !"!&
!"!& ! !"!!# !"!$ !"!$# !"!% !"!%# !"!&
! ! ! !
(a) h = 100 mm (b) h = 500 mm

'! '! '!


Gc 452617238 Gc 452617238relations for h = 100 mm, (a), and
Figure 4.15: Compressive stress-strain
9:;!<=;
h = 500 mm,
9:;!<=;
Gc -+./01231fc
(b), for a concrete defined
>+5+8?6+1@f A
Gc -+./01231f
by the variables in Table 4.3 model as suggested
>+5+8?6+1@f c
A by
c c
&! &! Nakamura and0/1;25.B1@f
Higai [19].
A The compressive
&! fracture energies are computed
0/1;25.B1@fcA based
c
on the compressive strength fcm , (4.10), and based on the tensile fracture energy
" ()*+,

" ()*+,
" ()*+,

%! %! Gf = 0.095 N/mm, (4.11). %!

In Figure 4.15 the compressive material model suggested by Nakamura and Higai [19] is plotted
$! $! $!
for concrete grade C30, using the parameters from the CEB-FIB Model Code [16] in Table 4.3
for element lengths of h = 100 mm and h = 500 mm. From Figure 4.15 it is apparent that the
! ! !
$ !"!$# !"!% !"!%# !"!& ! ! !"!!#
!"!!# !"!$
!"!$ !"!$#
!"!$# !"!%
!"!% !"!%#
!"!%# !"!&
!"!& ! !"!!# !"!$ !"!$# !"!% !"!%# !"!&
! ! ! 33 !
Chapter 4: Modelling of Uniaxial behaviour of Reinforced Concrete at Ambient Temperature

post peak behaviour in compression can be significantly different depending on the model used
for the compressive fracture energy. It is also seen, by comparing the two, that the behaviour
differs significantly with varying element side lengths.

Maximum Element Size

Herein, it has been found that a limitation must be imposed on the maximum element size.
Figure 4.15b illustrates the problem that occurs for large values of h. It is seen that when com-
puting the compressive fracture energy based on the tensile fracture energy, Gf , using (4.11),
a snap-back effect occurs in the stress-strain diagram. It must be understood that this is a
possible problem that can occur no matter which expression is used for the compressive fracture
energy, if the element side length is chosen to be of excessive magnitude.

As the snap-back is clearly not a realistic compressive behaviour, it is necessary to limit the
element side length. This will be different from previously defined expression for the maximum
element size in tension, (4.5), as both the strengths and fracture energies differ in tension and
in compression. Herein, the expression in compression is arrived at by equating the ultimate
concrete strain, εcu , with the peak strain, εp , which gives:

2 Gc Ec
hmax = (4.12)
3 fcm
2

By applying expression (4.12) to the material values from Table 4.3 and computing the compres-
sive fracture energy by (4.11), a maximum element side length of hmax = 371 mm is obtained.
The choice of h = 500 mm in Figure 4.15b is therefore clearly in violation of the validity range,
which explains the snap-back behaviour.

4.4.4 Compression Model by Feenstra and de Borst

As compressive failure is initiated by a combination of shear and tensile stresses (described in


detail in appendix A.1) Feenstra and de Borst [18] define the material behaviour by a damage
model. The concrete behaviour is described by two internal parameters; κT in tension and κC in
compression. These are related to the released energy per unit damaged area by an equivalent
length, h. Because the equivalent parameters are dependent on h, the material model is linked
to the element size and the ultimate parameters, κuC and κuT , are assumed to be constant
element-related material parameters as they can be calculated from the material properties and
the equivalent length, h; the latter related to the element area. In a monotonic loading cycle,
it is not possible to distinguish the roles of damage and plasticity (or cracking strains and plas-
tic strains), which only become apparent durring unloading. Here, the damage parameters, or
equivalent cracking strains, are therefore regarded as plastic strains.

The material strength in compression is defined by two functions; pre- and post-peak. The
equivalent strain, κe , corresponding to peak is expressed as:
4fcm
κe = (4.13)
3Ec

The ultimate compressive concrete strain is defined as the following and is seen to be dependent
upon the compressive fracture energy.
Gc
κuC = 1.5 (4.14)
hfcm

34
'! '!'! '!
Gc -+./01231fc
Gc -+./01231Gf
&! &!&! &!
Chapter 4: Modelling of Uniaxial behaviour of Reinforced Concrete at Ambient Temperature

" ()*+,

" ()*+,
" ()*+,

" ()*+,
%! %!%! %!
In Figure 4.16 the compressive material behaviour by Feenstra and de Borst [18] is shown for
a concrete
$! grade C30, defined by the parameters in $!$!Table 4.3. Two different values for the $!
compressive fracture energy are taken; the one obtained by Vonk [20] (see Figure 4.13) and
the chosen methods defined by Nakamura and Higai [19] computed by the compressive strength
!
using (4.10).
! The compressive
!"!!# behaviour
!"!$ !"!$# !"!% !"!%# is shown for! !!element
!"!& !"!!#side
! !"!!# lengths
!"!$
!"!$ !"!$# of
!"!$# h =!"!%#
!"!%
!"!% 100 mm
!"!%# !"!&and
!"!&
!
! !"!!# !"!
h = 500 mm. ! ! !

'! '!'! '!


Gc 452617238 Gc 452617238
9:;!<=;
Gc -+./01231fc >+5+8?6+1@f
Gc -+./01231f A
c c
&! &!&! 0/1;25.B1@fcA
&!

" ()*+,

" ()*+,
" ()*+,

" ()*+,
%! %!%! %!

$! $!$! $!

! !! !
! !"!!# !"!$ !"!$# !"!% !"!%# !"!& !! !"!!# !"!$
!"!!# !"!$ !"!$#
!"!$# !"!%
!"!% !"!%#
!"!%# !"!&
!"!& ! !"!!# !"!
! !!
(a) h = 100 mm (b) h = 500 mm

Figure 4.16: Compressive behaviour model suggested by Feenstra and de Borst [18] plotted
using compressive fracture energies as defined by Vonk [20], Figure 4.13, and
Nakamura and Higai [19] based on the compressive strength fc , (4.10). A concrete
grade C30 is considered and the material data of Table 4.3 are used.

It is clearly seen by Figure 4.16, that the chosen value for the compressive fracture energy has
great significance for the model, especially for low values of h. Further, it is illustrated that the
value obtained by Vonk [20] is much lower than the one obtained using the approach suggested by
Nakamura and Higai [19]. In the following it is decided to take the value obtained by expression
(4.10).

Maximum Element Size

Herein, it has been established that this compression model also has problems with excessive
values of the element side length, h. This is illustrated in Figure 4.17, where it is seen that the
concrete does not reach its full strength level for h = 2000 mm.

An expression for the maximum allowable element side length, can be arrived at by equating
the equivalent strain at peak κe , obtained using (4.13), with the ultimate equivalent strain κuC ,
obtained using (4.14).

9 G c Ec
hmax = (4.15)
8 fcm
2

In Figure 4.17, this imposes a maximum limit on the element side length of 1409 mm.

By comparing expression (4.15) to the obtained maximum element side length in the compression
model by Nakamura and Higai [19], obtained using (4.12), it is seen that the maximum limit is
increased by 69%, and thus, that the model by Feenstra and de Borst [18] allows considerably
larger element side lengths.

35
Chapter 4: Modelling of Uniaxial behaviour of Reinforced Concrete at Ambient Temperature

40
h = 100 mm
h = 500 mm
30 h = 2000 mm

" [MPa]
20

10

0
0 0.005 0.01 0.015 0.02 0.025
!
4040 C 40
Gc based on fc Gc based on fc
Figure 4.17: Stress-equivalent strain
G based
c
on diagram
G
f
for compression model by Feenstra
G basedand
on Gde Borst
3030 30 c f
[18] for concrete grade C30 with fracture energy by expression (4.10), element
side lengths h = 100 mm, h = 500 mm and h = 2000 mm.
" [MPa]
" [MPa]

" [MPa]
2020 20

4.4.5 10 Comparison of Compression Models


10 10

The agreement with the CEB-FIB Model Code [16] is evident for both the compressions model
presented
0 0 by Nakamura and Higai [19] and by Feenstra0 and de Borst [18]. The three are plotted
5 0.02 0.025 0.03 0 0 0.005
0.005 0.01
0.01 0.015
0.015 0.02
0.02 0.025
0.025 0.03
0.03 0 0.005 0.01 0.015 0.02 0.025 0.03
for the same material parameters!! in Figure 4.18. !

4040 40
G from Vonk GCEB!FIB
from Vonk CEB!FIB
c c
Narakuma (f ) Narakuma (fc)
G based on f G based on fc
c c 3030 c
de Borst (fc) c 30 de Borst (fc)
" [MPa]
" [MPa]

" [MPa]

2020 20

1010 10

00 0
5 0.02 0.025 0.03 00 0.005
0.005 0.01
0.01 0.015
0.015 0.02
0.02 0.025
0.025 0.03
0.03 0 0.005 0.01 0.015 0.02 0.025 0.03
!! !
(a) h = 100 mm (b) h = 500 mm

Figure 4.18: Stress-strain diagrams for compressive behaviour as defined by the CEB-FIB
Model Code [16], Nakamura and Higai [19] and Feenstra and de Borst [18] for
concrete grade C30 with element side length h = 100 mm, (a), and h = 500
mm, (b). The material data is taken from Table 4.3 and the compressive fracture
energy is computed based on the compressive strength as defined in (4.10).

It is easily seen by Figure 4.18b, that good agreement with the CEB-FIB Model Code [16] com-
pression model is found for large values of element side length with both the model by Nakamura
and Higai [19] and the model by Feenstra and de Borst [18]. However, it is clearly evident that
both fracture energy based models ignore the last segment of the softening branch where the
stress slowly decreases. Further, Figure 4.18a illustrates the necessity for a compressive fracture
energy based material model when using small element side lengths.

As expected, the most significant differences in the softening branches are found in the com-
pression models by Nakamura and Higai [19] and by Feenstra and de Borst [18]. Considering

36
Chapter 4: Modelling of Uniaxial behaviour of Reinforced Concrete at Ambient Temperature

the uncertainty associated with the value of the compressive fracture energy, the difference of
the employed model is regarded as being quite insignificant for reinforced concrete. Herein,
it is chosen to implement the compression model by Feenstra and de Borst [18], as its curved
softening resembles the shape of the compression model from the CEB-FIB Model Code [16] the
most, and it allows for the use of greater element side lengths.

4.5 Chosen Uniaxial Concrete Models

Herein, the considered concrete material is of Grade C30 and the material properties of the
CEB-FIB Model Code [16] are used. These are given in Table 4.2 and Table 4.3 for tension and
compression, respectively.

Because it has previously been found that the stress-strain relation cannot be taken as a material
property for concrete, it is here chosen to base the material model on the concept of fracture
energy. In both tension and compression the material models presented by Feenstra and de Borst
[18] are used as these were found to have the largest span in choice of the element side length,
h. It must still be ensured that the chocie of h falls within the acceptable range, where the
maximum limit is expressed in tension by equation (4.5) and in compression by equation (4.15).
Combining these two requirements yields:


 Ec Grf

2
f
for ls ≤ hmax
 f2
ct,m
hmax = min (4.16)

 9 Ec G c



8 fcm
2

If the interaction contribution is to be defined as a part of the tension stiffening, it must moreover
be ensured that the reinforcement ratio is sufficient. This is most practically done by ensuring
that inequality (4.9) is fulfilled.

A value of αts = 0.7 is taken for the strength level of the interaction contribution for the tension
stiffening definition.

The compressive fracture energy is computed by expression (4.10) presented by Nakamura and
Higai [19].

4.6 Numerical Test Examples

In this section the chosen uniaxial material models are applied to three benchmarck test exam-
ples in ABAQUS. This is done to verify that the expected results of the FE-analysis correlates
with the obtained output. Therefore, a simple reinforced concrete plate is considered in uniaxial
tension and in uniaxial compression. Further, a pure shear analysis is carried out.

The plate considered is square with side length b = 100 mm and thickness t = 10 mm and is
modelled by a single plane stress element. The plate is reinforced by a one dimensional rod
that has an area of As = 30 mm2 and hence ρp = 0.03. The member consists of concrete grade
C30 and steel Grade 500 and the material properties of Table 4.1 (steel), Table 4.2 (concrete
in tension) and Table 4.3 (concrete in compression) are used. Further, the fraction of strength

37
Chapter 4: Modelling of Uniaxial behaviour of Reinforced Concrete at Ambient Temperature

of the interaction contribution is taken as αts = 0.7 and the compressive fracture energy is
computed by equation (4.10).

2 3

p A s = 30 mm 2
100 mm
t = 10 mm
y
1 4

x q

100 mm

Figure 4.19: FE-configuration of the reinforced member considered for uniaxial load tests of
the tension stiffening and the compression model in ABAQUS.

If applying the used parameters to equation (4.16) it is found that the maximum element size is
756 mm and hence that h = 100 mm is admissible. Further, it is found that that the considered
level of reinforcemenet yields an ultimate strain of εu = 2.16 · 10−3 which is greather than the
ultimate concrete strain, εcu = 5.37 · 10−4 . Therefore, the tension stiffening can be modelled by
the interaction contribution as inequality (4.9) is fulfilled.

For the FE-analysis, the concrete plasticity damage model described in chapter 2 is applied,
using the input parameters of Table 2.1.

4.6.1 Uniaxial Tension

The FE-configuration of the considered reinforced member subjected to uniaxial loads can be
seen in Figure 4.19. The performed analysis is displacement controlled and the load is therefore
introduced by positive displacements of node 2 and 3 in the y-direction when considering uniaxial
tension.

$t

f ct,m

" ts f ct,m
" ts f ct,m ! #

! c 0 ! tck1 ! cu ! tck2 ! u !y !t

Figure 4.20: The tension stiffening is defined in ABAQUS as the combination of the concrete
and interaction contributions and must be forced to constantly have a slope, by
subtracting ∆ from the stress at the input, defining εu .

When attempting to define the tension stiffening by the combined concrete and interaction

38
Chapter 4: Modelling of Uniaxial behaviour of Reinforced Concrete at Ambient Temperature

contribution, it is found that an error occurs on the output load-displacement diagram. It is


established that the error does not occur when there is a slope between the points in the tension
stiffening input. Therefore, a small stress contribution, ∆, is subtracted from the last of the two
points, forcing a slope. The modified combined concrete and interaction contribution used as the
tension stiffening input in the ABAQUS model, can schematically be seen in Figure 4.20.

8000
8000 8000
8000

6000
6000 6000
6000
P [N]
P [N]

P [N]
P [N]
4000
4000 4000
4000

2000
2000 2000
2000
Expected
Expected Expected
Expected
ABAQUS
ABAQUSoutput
output ABAQUS
ABAQUSoutput
output
0 0 0 0
0 0 0.050.05 0.1 0.1 0.150.15 0.2 0.2 0.250.25 0 0 0.050.05 0.1 0.1 0.150.15 0.2 0.2 0.250.25
! [mm]
! [mm] ! [mm]
! [mm]
(a) *CONCRETE 4 4TENSION STIFFENING without mod- (b) *CONCRETE TENSION STIFFENING modified by
x 10x 10
ification ∆ =0 .01 MPa
3 3

Figure2.54.21:
2.5 ABAQUS output of load-displacement diagram in the y-direction on node 3 for
the example plate subjected to uniaxial tension. The tension stiffening is mod-
2 2 elled as presented by Feenstra and de Borst [18], (a), and modified by ∆ = 0.01
MPa to ensure a constant presence of slope, (b).
P [N]
P [N]

1.5 1.5

The output
1 1 of the ABAQUS analysis for the uniaxial tension test can be seen in Figure 4.21 for
both the tension stiffening defined with and without a modification of ∆ = 0.01 MPa. As the
Expected
Expected
0.5 0.5
output coincides with the exptected
ABAQUS
ABAQUStheoretical
output
output results, it is seen that the modification does not
interfere0 with
0 the expected theoretical results for this value of ∆.
0 0 0.5 0.5 1 1 1.5 1.5 2 2
! [mm]
! [mm]
4.6.2 Uniaxial Compression

The compressive model of Feenstra and de Borst [18] is used in a similar test example for uniaxial
compression. The compressive behaviour is defined by the *COMPRESSION HARDENING function
that is described in appendix B. According to the ABAQUS Version 6.7 Documentation [8] both
the in-elastic strains and their corresponding stresses must be given as positives and therefore
the parameters from Table 4.3 are used to compute the stress-strain relation for the material
model. The inelastic strains for the point-wise input are then found. In this example, the in-
elastic behaviour is defined by 15 points to check the agreement of the ABAQUS output with
the expected theoretical results.

The load is, as for uniaxial tension, introduced as displacements of node 2 and 3 in the y-direction
(Figure 4.19), but this time as negatives.

39
4000 4000

P
2000 2000
Expected Expected
ABAQUS output ABAQUS output
0 0
Chapter 4: Modelling of Uniaxial
0 behaviour
0.05 0.1of Reinforced
0.15 0.2Concrete
0.25 at Ambient
0 Temperature
0.05 0.1 0.15 0.2 0.25
! [mm] ! [mm]
4
x 10
3

2.5

P [N]
1.5

0.5 Expected
ABAQUS output
0
0 0.5 1 1.5 2
! [mm]

Figure 4.22: ABAQUS output of load-displacement diagram in the y-direction on node 3 for
the plate example subjected to uniaxial compression. The compressive properties
are modelled as presented by Feenstra and de Borst [18].

In Figure 4.22 the considered output is the reaction force and the displacement in the y-direction
of node 3. It is clearly seen that the output and the expected load-displacement diagram are in
good agreement.

4.6.3 Pure Shear

A similar plate as above is considered subjected to pure shear. The shear load is applied by
displacing all four nodes as illustrated in Figure 4.23. Two conditions are evaluated; when
the plate consists of plain concrete and when the plate is equally reinforced in the x and y-
directions.
2 3

A s = 30 mm 2
100 mm p
t = 10 mm
y

x q
1 4
100 mm

Figure 4.23: FE-configuration for numerical test element subjected to pure shear.

The outputs from the two analyses in form of the stresses and strains obtained in the integration
points and the forces and displacements in the nodes, are for some time increments enclosed in
appendix C for the plate example with and without reinforcement. By comparison of the two,
it is found that the output obtained with and without reinforcement are equal and thus the
reinforcement parallel to the plate edges has no effect in pure shear.

40
Existing Models of the Behaviour
of Reinforced Concrete at
Elevated Temperatures

5.1 Introduction
This chapter is concerned with the existing models of reinforced concrete at elevated tempera-
tures and is therefore commenced with a brief description of the decay of material strength in
these models. This is followed by a discussion of some of the existing models for the behaviour
of concrete in compression and in tension at elevated temperatures. Finally, the material model
for the reinforcement at elevated temperatures is described.

5.2 Decay of Material Strength

As described in chapter 3, the material strengths of both the plain concrete and the reinforcement
decrease when the temperature is increased. One existing model for the change in material
properties is suggested by Hertz [7], who defines a reduction parameter which is a S-shaped
function of the temperature. It is developed so that the same function can be used for the decay
of all material properties of a given material:
1−k
ξ(T ) = k + ! $2 ! $8 ! $64 (5.1)
T T T T
1+ + + +
T1 T2 T8 T64

The parameter ξ(T ) is the ratio of the material property at ambient temperatures to that
at elevated temperatures and the constants k, T1 , T2 , T8 and T64 depends on the considered
material.

5.2.1 Compressive Strength of Concrete

For computations of the decay of the material properties of concrete, using the method suggested
by Hertz [7], the parameters depend on the type of aggregate used. In the hot phase of a fire
these are specified in Table 5.1 for siliceous, main group and light aggregate concretes.

41
Chapter 5: Existing Models of the Behaviour of Reinforced Concrete at Elevated Temperatures

Table 5.1: Parameters describing decay functions for concrete in the hot phase of a fire as
presented by Hertz [14].

k T1 T2 T8 T64
Siliceous concrete 0.00 15000 800 570 100000
Main group concrete 0.00 100000 1080 690 1000
Light aggregate concrete 0.00 100000 1100 800 940

The decay of compressive strength as a function of temperature is tabulated in Eurocode 2 [21].


The code distinguishes between two different types of aggregates; siliceous and calcerous. In
Table 5.2 the reductions of compressive strength, fcT /fcm , recommended by Eurocode 2 [21],
are given for temperatures ranging from 20◦ C to 1200◦ C. Further, the two parameters describing
the compressive stress-strain relationship, εc1T and εcu1T , are given.

Table 5.2: Parameters describing the compressive behaviour of concrete at temperature T , (a),
as defined by Eurocode 2 [21] for siliceous, (b), and calcerous aggregates, (c).

(a) (b) Siliceous Aggregates (c) Calcerous Aggregates


T fcT /fcm εc1T εcu1T fcT /fcm εc1T εcu1T
20◦ C 1.00 0.0025 0.0200 1.00 0.0025 0.0200
100◦ C 1.00 0.0040 0.0225 1.00 0.0040 0.0225
200◦ C 0.95 0.0055 0.0250 0.97 0.0055 0.0250
300◦ C 0.85 0.0070 0.0275 0.91 0.0070 0.0275
400◦ C 0.75 0.0100 0.0300 0.85 0.0100 0.0300
500◦ C 0.60 0.0150 0.0325 0.74 0.0150 0.0325
600◦ C 0.45 0.0250 0.0350 0.60 0.0250 0.0350
700◦ C 0.30 0.0250 0.0375 0.43 0.0250 0.0375
800◦ C 0.15 0.0250 0.0400 0.27 0.0250 0.0400
900◦ C 0.08 0.0250 0.0425 0.15 0.0250 0.0425
1000◦ C 0.04 0.0250 0.0450 0.06 0.0250 0.0450
1100◦ C 0.01 0.0250 0.0475 0.02 0.0250 0.0475
1200◦ C 0.00 - - 0.00 - -

In Figure 5.1 the reduction factor suggested by Hertz [7], computed by (5.1), using the param-
eters of Table 5.1, are compared with the strength ratio of Eurocode 2 [21], Table 5.2.

It is seen by Figure 5.1 that the approaches of Eurocode 2 [21] and of Hertz [7] generally correlate.
However, Figure 5.1a shows that for siliceous aggregates exposed to temperatures above 500◦ C,
the compressive strengths computed by Hertz [7] are lower than those recommended by Eurocode
2 [21]. This is due to the fact that the strengths in Eurocode 2 [21] are based on transient strain
tests where the strain is held constant in the concrete and the temperature is varied. This type
of test is, according to Hertz [7], known to yield strengths of up to 25% greater magnitudes than
if data is collected by holding the temperature constant and varying the strain.

42
Chapter 5: Existing Models of the Behaviour of Reinforced Concrete at Elevated Temperatures

1 1 1 1
Eurocode
Eurocode
2 2 Eurocode
Eurocode
2 2
HertzHertz HertzHertz
0.8 0.8 0.8 0.8
fcT / fcm
fcT / fcm

fcT / fcm
fcT / fcm
0.6 0.6 0.6 0.6

0.4 0.4 0.4 0.4

0.2 0.2 0.2 0.2

0 0 0 0
200 200400 400600 600800 800
10001000
12001200 200 200400 400600 600800 800
10001000
12001200
T [oC]
T [oC] T [oC]
T [oC]
(a) Siliceous aggregates (b) Calcerous aggregates

1 1
Figure 5.1: Comparison of
Eurocodethe
2 decay
Eurocode 2 function for compressive strength presented by Hertz
0.8 0.8
[7] with the compressive
Hertz,
Hertz,
Sciliceous decay
Sciliceous function from Eurocode 2 [21] for a concrete with
siliceous, Hertz,
(a),Hertz,
MainMain
and groupgroup
calcerous aggregates, (b). For computation of the decay of
Hertz,
Hertz,
LightLight
aggregates
aggregates
0.6 0.6 strength as suggested by Hertz [7], equation (5.1) and the parameters of Table 5.1
fctT / fct,m
fctT / fct,m

are used and the reduction presented in Eurocode 2 [21] is given in Table 5.2.
0.4 0.4
In Eurocode 2 [21] the residual properties after exposure to a temperature T is considered to be
equal0.2 0.2 properties at high temperatures. The parameters used to compute the strength of
to the
concrete after temperature exposure, as suggested by Hertz [7], using equation (5.1), are given
0 0
in Table 5.3.200 200400 400600 600800 80010001000
12001200
T [oC]
T [oC]

Table 5.3: Parameters describing decay functions for concrete in the cold phase of a fire as
presented by Hertz [7].

k T1 T2 T8 T64
Siliceous concrete 0.00 3500 600 480 680
Main group concrete 0.00 10000 780 490 100000
Light aggregate concrete 0.00 4000 650 830 930

As described in section 3.2.1, the strength of concrete is further reduced upon cooling after
exposure to a temperature elevation. The strengths computed by Eurocode 2 [21] are therefore
bound to be higher than those modelled by Hertz [7], which also is evident in Figure 5.2.

43
Chapter 5: Existing Models of the Behaviour of Reinforced Concrete at Elevated Temperatures

1 1 1 1
Eurocode
Eurocode
2 2 Eurocode
Eurocode
2 2
HertzHertz HertzHertz
0.8 0.8 0.8 0.8
fcT / fcm
fcT / fcm

fcT / fcm
fcT / fcm
0.6 0.6 0.6 0.6

0.4 0.4 0.4 0.4

0.2 0.2 0.2 0.2

0 0 0 0
200 200400 400600 600800 80010001000
12001200 200 200400 400600 600800 80010001000
12001200
T [oC]
T [oC] T [oC]
T [oC]
(a) Siliceous aggregates (b) Calcerous aggregates

1 1
Figure 5.2: Residual compressive
Eurocode 2 2 strength of concrete after exposure to temperature level T ,
Eurocode

0.8 0.8
as presented by
Hertz, Eurocode
Hertz, Sciliceous 2 [21] and Hertz [7], for siliceous, (a), and calcerous, (b),
Sciliceous
aggregates.Hertz,
Hertz,
The Main Main
group
strength group
reduction presented by Hertz [7] is computed by equation
Hertz,
Hertz,
LightLight
aggregates
aggregates
(5.1) with the parameters from Table 5.3
1 1and the reduction from Eurocode 2 [21]
fctT / fct,m
fctT / fct,m

1
0.6 0.6 1
Eurocode 2 Eurocode 2 2
Eurocode
is given in Table 5.2.
Hertz HertzHertz
0.8 0.4
0.4 0.8 0.8 0.8
cm

cm

cm

cm
0.6 0.2
0.2 0.6 0.6 0.6
5.2.2 Tensile Strength of Concrete
f /f

f /f
f /f

f /f
cT

cT

cT

cT
0.4
0 0 0.4 0.4 0.4
200 200400 400600 600800 80010001000
12001200
The same considerations areo largely valid for the reduction of the tensile strength of concrete
T [oC]T [ C]
after exposure to elevated temperatures as for the 0.2
0.2 compressive
0.2 strength. However, as Eurocode 0.2
2 [21] does not generally recommend that the tensile properties of concrete are taken into
0 0 0 0
consideration,
200a simple
400 bi-linear
600 800 function for the reduction
1000 1200 of 200
200 tensile strength
400 400600 600800is800
given, as shown
10001000
1200 1200 200 400 60
in Figure 5.3. o
T [ C] o
T [ C] o
T [ C] T[

1 Eurocode 2 1 Eurocode 2
Hertz, Sciliceous Hertz, Sciliceous
0.8 Hertz, Main 0.8 Hertz, Main
Hertz, Light Hertz, Light
ct,m

ct,m

0.6 0.6
/f

/f
ctT

ctT

0.4
f

0.4
f

0.2 0.2

0 0
200 400 600 800 1000 1200 200 400 600 800 1000 1200
o
T [ C] T [oC]
(a) Hot phase of fire (b) Cold phase of fire

Figure 5.3: Comparison of decay of tensile strength of concrete in the hot, (a), and the cold,
(b), phase of a fire from Eurocode 2 [21] and the method presented by Hertz [7]
with siliceous, main group and light weight aggregates. For the computations of
the strength by Hertz [7], equation (5.1) and the parameters of Table 5.1 and Table
5.3 are used.

It is seen by Figure 5.3 that because Eurocode 2 [21] does not consider the aggregate type used, it
largely yields conservative tensile strengths in comparrison to the strength levels obtained using
the expression presented by Hertz [7] in both the hot and the cold phase of a fire. However, the

44
Chapter 5: Existing Models of the Behaviour of Reinforced Concrete at Elevated Temperatures

tensile strength of concrete by Eurocode 2 [21] is seen to be insensitive to temperatures below


200◦ C.

5.2.3 Strength of Reinforcement

The parameters for equation (5.1) in Table 5.4 defines the reduction of strength in the hot phase
for hot-rolled and cold-worked reinforcement, respectively, as suggested by Hertz [14].

Table 5.4: Parameters describing decay functions for reinforcement in the hot phase of a fire
as presented by Hertz [14].

k T1 T2 T8 T64
Hot-rolled bars, 0.2% stress 0.00 6000 620 565 1100
Hot-rolled bars, 2.0% stress 0.00 100000 100000 593 100000
Cold-worked bars, 0.2% stress 0.00 100000 900 555 100000
Cold-worked bars, 2.0% stress 0.00 100000 5000 560 100000

The parameters for computations of the strength reductions are given in Table 5.4 for both
0.2% stress and for 2.0% stress. Here, 0.2% corresponds to the yield stress of the steel and
2.0% corresponds to the ultimate strength of steel. These are both defined as it was found by
Hertz [14] that the tabular values for strength reductions recommended by Eurocode 2 [21] were
associated with the ultimate stress and not the yield stress as the code specifies.

1 1 Eurocode
Eurocode
2 2 1 1 Eurocode
Eurocode
2 2
Hertz,
Hertz,
0.2%
0.2%
stress
stress Hertz,
Hertz,
0.2%
0.2%
stress
stress
0.8 0.8 Hertz,
Hertz,
2.0%
2.0%
stress
stress 0.8 0.8 Hertz,
Hertz,
2.0%
2.0%
stress
stress
syT sy
sy

syT sy
sy

0.6 0.6 0.6 0.6


f /f
f /f

f /f
f /f
syT

syT

0.4 0.4 0.4 0.4

0.2 0.2 0.2 0.2

0 0 0 0
200200 400400 600600 800800 1000
10001200
1200 200200 400400 600600 800800 1000
10001200
1200
T [oTC][oC] T [oTC][oC]
(a) Hot-rolled (b) Cold-worked
500500
Figure 5.4: Decay functions from
T =T20=the
o o Eurocode [21] and Hertz [14] for hot-rolled, (a), and
C
20 C
400400 cold-worked, (b), reinforcement
T = T
300
= o o
300
C C bars when exposed to high temperatures.
o o
T =T500
= 500
C C
f [MPa]
f [MPa]

o o
In Figure 5.4 the decay of strength of C
300300 T = T600
= hot-rolled
600 C and cold-worked reinforcement exposed to ele-
o o
vated temperatures are shown as functions
T = T700
= 700
C C of the temperature, computed by the tabular values
syT
syT

200200 2 [21] and based on o o


of Eurocode T =the
T1100 method
= 1100
C C by Hertz [14]. The latter makes use of the pa-
rameters in Table 5.4 and is plotted for both the yield stress (0.2% stress) and the ultimate
100100
stress (2.0% stress). The problem discovered by Hertz [14] is clearly illustrated, and therefore
it is chosen
0 0
herein to compute the reduction of the strength of the reinforcement by equation
(5.1), making
0 0 use of
0.05the parameters
0.05 0.1 0.1 specified
0.150.15 for 0.2% stress to ensure that the strength of the
0.2 0.2
T [oTC][oC]
45
Chapter 5: Existing Models of the Behaviour of Reinforced Concrete at Elevated Temperatures

reinforcement is not overestimated.

The parameters used to compute the reduction of strength in the cold phase of a fire by the
method presented by Hertz [14], are given in Table 5.5.

Table 5.5: Parameters describing decay functions for reinforcement in the cold phase of fire as
presented by Hertz [14].

k T1 T2 T8 T64
Hot-rolled bars, 0.2% stress 1.00 100000 100000 100000 100000
Hot-rolled bars, 2.0% stress 1.00 100000 100000 100000 100000
Cold-worked bars, 0.2% stress 0.58 100000 5000 590 730
Cold-worked bars, 2.0% stress 0.52 100000 1500 580 650

The hot-rolled reinforcement regains it strength post-fire and the parameters of Table 5.5 there-
fore yield a constant value of ξ(T ) =1 for all T , when applied to equation (5.1). As described in
section 3.2.2, cold-worked reinforcement does not regain full strength after exposure to temper-
atures above 400◦ C and the decay of the additional strength post-fire is illustrated in Figure 5.5.
It is seen that the in method by Hertz [14] 50% of the yield strength (0.2% stress) is remaining
at T = 700◦ C and above and 60% strength for the ultimate strength (2.0% stress) at 800◦ C and
above. The strength reduction of the cold-worked steel is not considered by Eurocode 2 [21],
which therefore overestimates the post-fire strength of the reinforcement for high temperatures.
After exposure to elevated temperatures, the lower strength of the reinforcement can therefore
cause failure if the member is assumed to have regained its previous strength.

1 1
Eurocode 2
Hertz, 0.2% stress
0.8 Hertz, 2.0% stress 0.8
sy

fsyT / fsy

0.6 0.6
syT

0.4 0.4
Eurocode 2
0.2 0.2 Hertz, 0.2% stress
Hertz, 2.0% stress
0 0
200 400 600 800 1000 1200 200 400 600 800 1000 1200
T [oC] T [oC]

Figure 5.5: Residual strength of cold-worked reinforcement steel after exposure to elevated
temperature level, T , as presented by Eurocode 2 [21] and Hertz [14].

5.3 Uniaxial Compressive behaviour of Concrete at Ele-


vated Temperatures

Youssef and Moftah [22] provide a review of the existing models for uniaxial compressive be-
haviour of concrete and the choice of models to be reviewed herein is largely based on their
recommendations. Common for most models is that the total strain of the mechanical stress-

46
Chapter 5: Existing Models of the Behaviour of Reinforced Concrete at Elevated Temperatures

strain relationship is considered to consist of a number of components. Li and Purkiss [23]


considers three strain components and the total strain is therefore given by:

εtot = εσT + εth + LITS (5.2)

Here, εσT is the instantaneous stress-related strain from the applied load and εth is the unre-
strained thermal strain arising from the expansion caused by the temperature elevation. The
LITS are the load induced thermal strains which is the effect on the thermal expansion caused
by the presence of loads during first time heating.

5.3.1 Strain Components at Elevated Temperatures

Instantaneous Stress-Related Strain

The instantaneous stress-related strain, εσT , is a function of the applied stress and the temper-
ature. It has its peak value at ε0T and the initial modulus of elasticity, EciT , defines the shape
of the stress-strain curve. Expressions for these two parameters will therefore be given, before
the stress-strain relationship is elaborated upon.

The strain value for the peak stress at a elevated temperatures, ε0T , can account for different
levels of applied compressive stress prior to heating if expressed by the formulation of Terro [24]:

ε0T = (50λ2L + 15λL + 1) · ε01 + 20 · (λL − 5λ2L ) · ε02 + 5 · (10λ2L − λL ) · ε03 (5.3a)

Here, λL is the initial compressive stress level and ε01 , ε02 and ε03 are expressed by:

ε01 = 2.05 · 10−3 + 3.08 · 10−6 · T + 6.17 · 10−9 · T 2 + 6.58 · 10−12 · T 3 (5.3b)
ε02 = 2.03 · 10 −3
+ 1.27 · 10 −6
· T + 2.17 · 10 −9
· T + 1.64 · 10
2 −12
·T 3
(5.3c)
ε03 = 0.002 (5.3d)

The initial modulus of elasticity was proposed by Anderberg and Thelandersson [25] to be given
by:

2fcT
EciT = (5.4)
ε0T

Equation (5.4) enables to for the initial E-modulus to implicitly account for the type of aggregate,
when implementing the reduced strength proposed by Hertz [7]. As a result of using expressions
(5.3) and (5.4), it is possible to account for the effect of the initial compressive load and the ag-
gregate, as well as the temperature, in the computation of the instantaneous stress-related strain.

There exists sereval models for the stress-strain relationship for the instantaneous stress-related
strain. In Youssef and Moftah [22] the models by Anderberg and Thelandersson [25], Lie and
Lin [26] and Schneider are described. However, as the model proposed by Scheneider does not
consider the post-peak behaviour, only the models by Anderberg and Thelandersson [25] and
Lie and Lin [26], will be evaluated herein.

The compressive stress-strain relationship for the instantaneous stress-related strain is by An-
derberg and Thelandersson [25] modelled by a parabola for the ascending branch and assumes

47
Chapter 5: Existing Models of the Behaviour of Reinforced Concrete at Elevated Temperatures

linear softening:
! $
ε2
σcT = EciT · εσT − σT for εσT ≤ ε1 (5.5a)
2ε0T
σcT = σ1 − 880 · (εσT − ε1 ) for εσT ≥ ε1 (5.5b)

In (5.5b) the parameter σ1 must be entered in MPa and the parameters ε1 and σ1 are given by
the following, where the initial elastic modulus, EciT , also must be defined in MPa:
! $ ! $
880 ε2
ε1 = ε0T · 1 − and σ1 = EciT · ε1 − 1 (5.5c)
EciT 2ε0T

The model for the stress-strain relationship for the instantaneous stress-related strain developed
by Lie and Lin [26] assumes parabolic functions for both the ascending and descending branches:

* ! $2 +
ε0T − εσT
σcT = fcT · 1− for εσT ≤ ε0T (5.6a)
ε0T
* ! $2 +
εσT − ε0T
σcT = fcT · 1− for εσT ≥ ε0T (5.6b)
3ε0T

The two considered models for the instantaneous stress-related strains are plotted in Figure
5.6 for a siliceous concrete at ambient temperature and elevated temperatures of T = 300◦ C,
T = 500◦ C and T = 700◦ C.

1 1 1 1
Lie and
Lie and
Lin Lin
o o
T = 20
T =C20 C Anderberg
Anderberg
and and
Thelandersson
Thelandersson
0.8 0.8 0.8 0.8
T = o500
T = 500 C oC
cT cm

cT cm

cT cm

cT cm

0.6 0.6 0.6 0.6


T = o300
T = 300 C oC
" /f
" /f

" /f
" /f

0.4 0.4 0.4 0.4

T = o700
T = 700 C oC
0.2 0.2 0.2 0.2

0 0 0 0
0 0 0.010.01 0.020.02 0.030.03 0.040.04 0 0 0.010.01 0.020.02 0.030.03 0.040.04
! ! ! !
"T "T "T "T

(a) T = 20◦ C and T = 300◦ C (b) T = 500◦ Cand T = 700◦ C

Figure 5.6: Instantaneous stress-related strain as presented by Anderberg and Thelandersson


[25] and by Lie and Lin [26] for temperatures of T = 20◦ C and T = 300◦ C, (a),
and T = 500◦ C and T = 700◦ C, (b). The ultimate stress is normalized by the
ultimate stress at ambient temperatures.

From Figure 5.6, it is clear that the ultimate strains predicted by the models differ significantly.
However, due to the lack of expirimental investigations, it is, according to Youssef and Moftah
[22], not possible to determine which model that provides the most accurate results.

Unrestrained Thermal Strain

The unrestrained thermal strain results from the thermal expansion caused by the elevated
temperatures. This can therefore only contribute to the mechanical stress-strain relation if the

48
Chapter 5: Existing Models of the Behaviour of Reinforced Concrete at Elevated Temperatures

considered specimen is constrained against expansion.

A variety of models of the unrestrained thermal strain, εth , exists and Youssef and Moftah [22]
concludes that the model in Eurocode 2 [21] correlates the best with the expirimental observa-
tions. Herein, this model is therefore utilized.

Eurocode 2 [21] gives the following expression for concrete with siliceous aggregates:

εth = −1.8 · 10−4 + 9 · 10−6 · (T − 20◦ C) + 2.3 · 10−11 · (T − 20◦ C)3 ≤ 14 · 10−3 (5.7a)

For concrete with carbonate aggregates the expression is given by:

εth = −1.2 · 10−4 + 6 · 10−6 · (T − 20◦ C) + 1.4 · 10−11 · (T − 20◦ C)3 ≤ 12 · 10−3 (5.7b)

Load Induced Thermal Strains

Law and Gillie [27] explains how the LITS covers a number of different strain components
in heated concrete; the transitional thermal creep, the drying creep and the transient strain.
The transitional thermal creep develops irrecoverably during the first time heating of sealed
concrete under load and is the largest component of the LITS. The drying creep is the shrinkage
experienced by the material due to the evaporation of water, whereas the transient strain refers
to the sum of the transitional thermal strain and the drying creep, where the drying creep
is most often omitted because it is very small comparred to transitional thermal strain. It is
therefore chosen herein as well.

! tot
Free thermal expansion
Net thermal expansion under pre!stress

LITS

T [o C]
100 200 300 400 500 600

Figure 5.7: Illustration of the difference between the total strain when heated with and without
applied stress. Reproduced from Law and Gillie [27].

Figure 5.7 shows how the LITS can have a significant effect on the total strain of a concrete
member at elevated temperatures. The total strain is plotted as a function of the temperature
for an unloaded specimen that thus experiences free thermal expansion and of a specimen that
is pre-loaded. The difference between the two curves illustrates the effect of the LITS.

As a result of the fact that the LITS only occurs during first time heating, and because it is
irrecoverable upon cooling, Law and Gillie [27] argue that it is necessary to define the contri-
bution to the strain by the LITS as a plastic deformation. However, this is not the case in the
few existing models for the mechanical stress-strain relationship where the LITS are included.
According to Law et al. [28] there is no way to determine which of the existing methods that
provides the most accurate results and therefore it is chosen herein to consider the two most

49
Chapter 5: Existing Models of the Behaviour of Reinforced Concrete at Elevated Temperatures

simple models; the model by Li and Purkiss [23] and the model presented in Eurocode 2 [21].

The model by Li and Purkiss [23] is a simplification of the Anderberg and Thelandersson model
including the transient strain effect. The initial load is taken as 30% of the compressive strength
and the model does not allow other levels to be considered. The compressive concrete behaviour
is modelled by a simple bi-linear relationship and an empirical formula is developed for the
strain at peak stress, ε∗cu :

2fcm
ε∗cu = + 0.21 · 10−4 · (T − 20◦ C) − 0.9 · 10−8 · (T − 20◦ C)2 (5.8)
Ec

The tangent modulus in the descending branch at temperature T is expressed by:


C)2.15
Ep∗ = −880 · ekp (T −20 (5.9)

where the parameter kp should be taken as 10−6 .

Eurocode 2 [21] does not distinguish between any strain component, but it vaguely states that
transient effects are included to some extent. The stress-strain relation prior to peak stress is
given by the following, after which linear softening is assumed:

3 · εcT fcT
σcT = * ! $3 + (5.10)
εcT
εc1T · 2+
εc1T

In expression (5.10), εc1T is the strain at peak stress. This and the ultimate strain, εcu1t , is
given for a variety of temperatures in Table 5.2.

1 1 1 1
o o Li and
Li and
Purkiss
Purkiss
T = T20= C
20 C
Eurocode
Eurocode
2 2
0.8 0.8 0.8 0.8
= o500
T = T500 C oC
= o300
C oC
cT cm

cT cm

cT cm

cT cm

0.6 0.6 T = T300 0.6 0.6


" /f
" /f

" /f
" /f

0.4 0.4 0.4 0.4

0.2 0.2 0.2 0.2 = o700


T = T700 C oC

0 0 0 0
0 0 0.010.01 0.020.02 0.030.03 0.040.04 0 0 0.010.01 0.020.02 0.030.03 0.040.04
! ! ! !
"T "T "T "T

(a) T = 20◦ C and T = 300◦ C (b) T = 500◦ C and T = 700◦ C

Figure 5.8: Compressive stress-strain relations as defined by Li and Purkiss [23] and Eurocode
2 [21] for siliceous concrete at T = 20◦ C and T = 300◦ C, (a), and T = 500◦ C and
T = 700◦ C, (b).

In Figure 5.8 the compressive model in Eurocode 2 [21] is compared to the model by Li and
Purkiss [23] for a range of temperatures. It is seen that for T = 500◦ C and above, both the
strain at peak stress and the ultimate strain is significantly higher in the Eurocode 2 [21] model
than in the model by Li and Purkiss [23]. However, common for both models is that the strain

50
Chapter 5: Existing Models of the Behaviour of Reinforced Concrete at Elevated Temperatures

at peak stress is increased as the temperature rises.

It is widely recognized that the LITS changes the E-modulus in the direction where the load is
present, but it has been suggested by Law et al. [28] that the unloaded direction has a different
constitutive relation. However, there is a severe lack of information on the response in the di-
rections that the load is not applied in, as it has not been experimentally investigated.

For simplicity it is chosen herein to ignore the LITS in the formulation of a fracture energy
based model at elevated temperatures. If initial load is present this assumption will result in
a different distribution of the stresses modelled within the considered member to the stresses
actually occurring. However, this can prove to be a safe assumption in a structural context, if,
for example, considering a line of columns where one column is more heavily loaded then the
others. The presence of the LITS will then result in a smaller expansion of the heavily loaded
column, which ultimately redistribes the loads to the other columns in the line. As the material
strength decays, ignoring the LITS can in this case be considered a safer scenario.

5.4 Uniaxial Tensile behaviour of Concrete at Elevated Tem-


peratures

The knowledge on the tensile performance of concrete at elevated temperatures is limited. Terro
[24] suggests modelling the concrete behaviour as linear post-peak with an ultimate strain of
εctuT 0.004. This will yield the stress-strain relation illustrated in Figure 5.9 for a range of
temperatures.

1
T = 20oC
0.8 T = 300oC
T = 500oC
T = 700oC
ctT ctm

0.6
" /f

0.4

0.2

0
0 1 2 3 4
! !"
"T
x 10

Figure 5.9: Tensile stress-strain relationship as suggested by Terro [24] for concrete at tem-
peratures of T = 20◦ C, T = 300◦ C, T = 500◦ C and T = 700◦ C.

As noted in section 3.2.2, the bond strength between the concrete and the reinforcement de-
creases when the temperature is increased. Khoury [12] describes that the reduction of the bond
strength most often is modelled as a reduction of the tensile strength. However, according to
Youssef and Moftah [22] few models exist and there is a large scatter in the available experi-
mental results due to a lack of uniformity in the test procedures. Due to this uncertainty, and
for simplicity, the reduction of the bond strength is not considered herein.

51
Chapter 5: Existing Models of the Behaviour of Reinforced Concrete at Elevated Temperatures

5.5 Reinforcement Model at Elevated Temperatures

It has been shown by Hertz [14], and is it also implemented in the stress-strain relationship
defined in Eurocode 2 [21], that the yield strain of steel is the same at all temperatures. This
means that the elastic modulus at elevated temperatures can be expressed as:

fyT
EsT = = ξ(T ) · Es (5.11)
fy /Es

The reinforcement is still modelled as a bi-linear relationship at elevated temperatures as il-


lustrated in Figure 5.10. The strength reduction is modelled using the procedure presented by
Hertz [14], and the model can therefore both consider hot-rolled and cold-worked steel bars. To
compute the reduced yield strength, the parameters for hot-rolled or cold-worked steel stress
0.2% of Table 5.4 are used.

500

400

300
T = 20oC
"

200 T = 200oC
T = 300oC
100 T = 400oC
T = 500oC
0
0 0.002 0.004 0.006 0.008 0.01 0.012
!

Figure 5.10: Example of reinforcement models at ambient and elevated temperatures for hot-
rolled reinforcement with the material characteristics of Table 4.1.

5.6 Overview of Relevant Assumptions for the Formulation


of the Fracture Energy Based Material Models

In the subsequent chapter (chapter 6) the fracture energy based material models of reinforced
concrete at elevated temperatures is formulated. The formulation is based on the existing ma-
terial models at elevated temperatures and therefore it is subject to a number of assumptions.
The assumptions include:

• The strength reductions of both concrete and reinforcement caused by temperature rise
are computed using equation (5.1) as suggested by Hertz [7] and Hertz [14].
• The strain at peak compressive stress is evaluated by using equation (5.3) as recommended
by Terro [24].
• The initial E-modulus at elevated temperatures of concrete is modelled by equation (5.4)
as per Anderberg and Thelandersson [25].
• The unrestrained thermal expansion is modelled as defined by Eurocode 2 [21] in equation
(5.7).
• The effect on the mechanical stress-strain relation caused by the LITS is ignored.

52
Chapter 5: Existing Models of the Behaviour of Reinforced Concrete at Elevated Temperatures

• The strain at yield stress of the reinforcement is assumed to be constant for all tempera-
tures.
• The reduction of the bond strength between the concrete and the reinforcement is not
incorporated in the tensile concrete model formulation.

53
Chapter 5: Existing Models of the Behaviour of Reinforced Concrete at Elevated Temperatures

54
Fracture Energy Based Uniaxial
Material Models at Elevated
Temperatures

6.1 Introduction

In this chapter the fracture energy based material models at elevated temperatures are formu-
lated and examined. The examination includes an investigation of the evolution of the tensile
and the compressive fracture energies at elevated temperatures as well as of the limits of appli-
cation that were developed in chapter 4.

The formulations made in this chapter are based on models for the concrete behaviour at elevated
temperatures and therefore the models are only valid for concrete exposed to high temperatures.
Although out of the scope of the current investigation, it should be possible to extend the formu-
lations to the residual stress-strain relationship, if existing models for such are applied.

6.2 Fracture Energy Based Compressive behaviour Model


for Concrete at Elevated Temperatures

A compressive fracture energy based model for concrete behaviour at elevated temperatures
is developed based on the model for ambient temperature by Feenstra and de Borst [18]. As
currently no experimental data is available on the evolution of the fracture energy with temper-
ature, the existing models for concrete behaviour at high temperatures are used to determine
the compressive fracture energy.

As concluded in chapter 5, the fracture energy based compressive behaviour model at elevated
temperatures will consist solely of the instantaneous stress-related strain as the effects of the
LITS are ignored. However, it is still interesting to investigate the effect of the LITS on the
compressive fracture energy at elevated temperatures. Therefore, expressions for the compressive
fracture energy are developed for all the four previously described models (chapter 5). Herein,
the two exiting models that solely consists of the instantaneous stress-related strain are:
• The model by Anderberg and Thelandersson [25] expressed by equation (5.5).
• The model suggested by Lie and Lin [26] given by equation (5.6).

55
Chapter 6: Fracture Energy Based Uniaxial Material Models at Elevated Temperatures

The existing models that includes the effects of the LITS are:
• The formulation by Li and Purkiss [23] expressed by equations (5.8-5.9).
• The Eurocode 2 [21] formulation given in equation (5.10).

6.2.1 Compressive Fracture Energy at Elevated Temperatures

The fracture energy based model for ambient temperature by Feenstra and de Borst [18] can
readily be extended to elevated temperatures as the equivalent strain at peak stress is defined
by the E-modulus and the strength by equation (4.13). Extending this to elevated temperatures
yields:
4 fcT
κeT = (6.1)
3 EciT

If not considering the effects of the LITS on the compressive behaviour at elevated tempera-
tures, this is found to correlate well with the strain at peak stress from the existing elevated
temperature models for all temperatures.

The ultimate strain in the model by Feenstra and de Borst [18] is defined by equation (4.14).
Herein, this equation is extended to elevated temperatures, so that the ultimate strain at a given
temperature, κuCT , is a function of the compressive fracture energy at the same temperature,
GcT :
GcT
κuCT = 1.5 (6.2)
hfcm

As a result, it is necessary to use the compressive fracture energy at elevated temperatures.


In lack of experimental evidence on this subject, the compressive fracture energy at elevated
temperatures is estimated by computing the compressive fracture energies that are intrinsic in
the existing elevated temperature models.

!cT

f cm

(G c / h )model
"cT
" p0 " 0T "cuT,model

Figure 6.1: The compressive fracture energy is inherent in the existing elevated temperature
models for the compressive behaviour of concrete.

Figure 6.1 illustrates the stress-strain relation for a given considered existing elevated tempera-
ture model where linear softening is assumed. If wanting to describe the model in terms of the
compressive fracture energy, it has been established in section 4.4, that the indicated grey area is
equal to the compressive fracture energy divided by a corresponding element size, (GcT /h)model .
To determine the size of the area, the plastic strain at peak stress, εp0 , must be found. As the
line that connects the points (εp0 , 0) and (ε0T , fcT ) has the slope of the initial E-modulus, it
can be expressed as:
σcT = EciT εcT − EciT εp0 (6.3a)

56
Chapter 6: Fracture Energy Based Uniaxial Material Models at Elevated Temperatures

By insertion of the point (ε0T , fcT ) into equation (6.3a), the plastic strain, εp0 , can be estab-
lished:
fcT
εp0 = ε0T − (6.3b)
EciT
The ultimate strain, εcuT,model , is determined from the given considered elevated temperature
model. When this is found, the compressive fracture energy divided by the corresponding
element size can be arrived at. For a model that considers linear softening, the enclosed area
can be expressed as:
! $
1 1 fcT
(Gc /h)model = fcT (εcuT,model − εp0 ) = fcT εcuT,model − ε0T + (6.4)
2 2 EciT

For the considered elevated temperature model, the corresponding element size can be deter-
mined by considering the ambient condition. Herein, the compressive fracture energy is com-
puted by equation (4.10), and therefore the following expression is used:

Gc 8.8 fcm
hmodel = = (6.5)
(GcT /h)model (GcT /h)model

When the corresponding element size is found, the fracture energy at elevated temperatures
is:

GcT,model = (GcT /h)model hmodel (6.6)

6.2.2 Application of the Elevated Temperature Model by Anderberg


and Thelandersson

As described in section 5.3, the material model by Anderberg and Thelandersson [25] for the
compressive behaviour of concrete at elevated temperatures is defined by equation (5.5). The
ultimate strain in the model by Anderberg and Thelandersson [25], εcuT,AT , must be found for
ε1 ≤ εσT and, hence, by equation (5.5b). An expression can therefore be arrived at by letting
σcT = 0 and substituting εσT = εcuT,AT :
σ1
εcuT,AT = + ε1 (6.7a)
880

In this, σ1 and ε1 are still expressed by (5.5c), where the considered temperature is implicitly
accounted for.

As the model by Anderberg and Thelandersson [25] assumes linear softening, the compressive
fracture energy divided by the corresponding element size can be found by combining equations
(6.7a) and (6.4): ! $
fcT σ1 fcT
(GcT /h)AT = ε1 − ε0T + + (6.7b)
2 880 EciT

To determine the corresponding element size, a concrete grade C30 is considered with the ma-
terial parameters at ambient temperature taken from Table 4.3. The strain at maximum stress,
ε0T , is calculated by equation (5.3), where no initial load is considered, i.e. λL = 0, and the ini-
tial E-modulus is found by equation (5.4). As a result, the following element size corresponding
to the compressive fracture energy inherent in the model by Anderberg and Thelandersson [25]
(denoted with the subscript AT ) is found to be
hAT = 65 mm.

57
Chapter 6: Fracture Energy Based Uniaxial Material Models at Elevated Temperatures

Figure 6.2 illustrates the fracture energy based formulation for the compressive behaviour of
concrete at ambient temperature for the h = 65 mm element. A comparison with the original
model by Anderberg and Thelandersson [25] shows that the areas corresponding to (GcT /h) are
equal, as they should be.

0.8

cT cm
f /f 0.6

0.4

0.2

0
0 2 4 6 8 0 0.01 0.02 #$#" 0.04
! !" !
x 10
Anderberg and Thelandersson
Lie and Lin
Fracture Energy Based
Fracture Energy Based

Figure 6.2: Compressive material model by Anderberg and Thelandersson [25] and fracture
energy based formulation with an element size of h = 65 mm for concrete grade
C30 at ambient temperature.

6.2.3 Application of the Elevated Temperature Model by Lie and


Lin

In section 5.3 it is described that the compressive behaviour model of concrete at elevated
temperatures as suggested by Lie and Lin [26] is expressed by equation (5.6). As the descending
branch is not linear in this model, the indicated area in Figure 6.1 cannot be defined by equation
(6.4). Instead, it is found as:
, * ! $2 +
1 εcuT ,LL
εσT − ε0T
(GcT /h)LL = (ε0T − εp0 ) + fcT 1 − dεσT (6.8a)
2 ε0T 3ε0T

The ultimate strain will occur in the descending branch, and hence equation (5.6b) will be used
to derive an expression for the ultimate strain based on the model by Lie and Lin [26], εcuT,LL .
Substituting σcT = 0 and εσT = εcuT,LL and solving for εcuT,LL yields:

εcuT,LL = 4ε0T (6.8b)

When carrying out the integration in equation (6.8a), and inserting the ultimate strain from
equation (6.8b) along with the plastic strain from equation (6.3b), the compressive fracture
energy divided by the corresponding element size of the Lie and Lin [26] model can be written
as:
1 fcT
2
(GcT /h)LL = + 2fcT ε0T (6.8c)
2 EciT

For the previously considered example, it is found by equation (6.5), using (6.8c), that the
element size corresponding to the compressive fracture energy inherent in the model suggested
by Lie and Lin [26] is:
hLL = 300 mm

58
Chapter 6: Fracture Energy Based Uniaxial Material Models at Elevated Temperatures

In Figure 6.3, the material model by Feenstra and de Borst [18] is compared to the model by
Lie and Lin [26] for an element size of h = 300 mm at ambient temperature. The comparison is
made using the same material properties as for the model based on Anderberg and Thelandersson
[25] in Figure 6.2. The areas corresponding to the compressive fracture energy divided by the
element size are indicated for both formulations and it is, as expected, seen that they are
equivalent.

1 1

0.8 0.8

0.6 0.6
cT cm

cT cm
f /f

f /f
0.4 0.4

0.2 0.2

0 0
0 2 4 6 8 0 0.01 0.02 #$#"
! !" !
x 10
Anderberg and Thelande
Lie and Lin
Fracture Energy Based
Fracture Energy Based

Figure 6.3: Compressive material model by Lie and Lin [26] and fracture energy based formu-
lation with an element size of h = 300 mm for a concrete grade C30 at ambient
temperature.

6.2.4 Compressive Fracture Energies at Elevated Temperatures for


Models Including the Effect of the LITS

As the models by Li and Purkiss [23] and Eurocode 2 [21] include the effects of the LITS, it is
not possible to directly modify them into a stress-strain relation in the form of the model by
Feenstra and de Borst [18]. However, it is still possible to define the plastic strain at peak stress,
εp0 , and thus to find an expression for the compressive fracture energy.

As described in section 5.3.1, the model by Li and Purkiss [23] is of triangular shape. As a
result, the plastic strain will equal zero, εp0 = 0, when the initial E-modulus reaches the peak
stress. The compressive fracture energies will therefore simply be the areas under the triangular
shaped curves illustrated in Figure 5.8.

The strain at peak stress, ε∗cu , is defined by equation (5.8) and the slope of the descending
branch, Ep∗ , is given in equation (5.9). As a result, the descending branch can be described by
the line:
σcT = (εcT − ε∗cu )Ep∗ + fcT (6.9a)

The ultimate strain of the elevated temperature model by Li and Purkiss [23] can thus be found
by substituting εcT = εcuT,LL and σcT = 0 into equation (6.9a), yielding:
fcT
εcuT,LP = ε∗cu − (6.9b)
Ep∗

Combining equations (6.9b) and (6.4) for a situation where εp0 = 0 yields the following expres-
sion for the compressive fracture energy divided by the corresponding element size for the model

59
Chapter 6: Fracture Energy Based Uniaxial Material Models at Elevated Temperatures

by Li and Purkiss [23]:


! $
1 fcT
(GcT /h)LP = · fcT εcu − ∗

(6.9c)
2 Ep

Considering the same example as previously, it is found by equations (6.5) and (6.9c) that the
element size corresponding to the compressive fracture energy inherent in the model by Li and
Purkiss [23] is:
hLP = 63 mm

The model suggested in Eurocode 2 [21] is, as described in section 5.3.1, defined by the tabulated
values for the strain at peak stress, εc1t , and the ultimate strain, εcu1t , given in Table 5.2. The
ascending branch is described by equation (5.10) and linear softening is assumed between the
strain peak stress, εc1t , and the ultimate strain, εcu1t . As the initial E-modulus is not included
in the formulation, it is necessary to develop an expression for it. This is done by computing
the tangent to equation (5.10) at the origin of the stress-strain relation:
3 fcT
EciT,EC = (6.10a)
2 εc1t

By substituting ε0T = εc1t and EciT = EciT,EC in equation (6.4), the compressive fracture energy
divided by the corresponding element size, for the model in Eurocode 2 [21] becomes:
! $
1 fcT
(GcT /h)EC = εcu1t − εc1t + (6.10b)
2 EciT,EC

For the previously considered example, equation (6.5) and equation (6.10b) yields an element
size, corresponding to the inherent fracture energy of

hEC = 150 mm

6.2.5 Comparison of Compressive Fracture Energies at Elevated Tem-


peratures

From the equations obtained for the compressive fracture energy divided by the corresponding
element size for each of the four considered existing models, (GcT /h)model , and the found corre-
sponding element sizes, hmodel , the compressive fracture energies at elevated temperatures are
found by equation (6.6). The element sizes obtained for the considered example are given in
Table 6.1.
Table 6.1: Element sizes obtained corresponding to the compressive fracture eneregies inherent
in the elevated temperature models by Anderberg and Thelandersson [25], hAT , Lie
and Lin [26], hLL , Li and Purkiss [23], hLP , and Eurocode 2 [21], hEC , for the
considered example.

hmodel
hAT 65 mm
hLL 300 mm
hLP 63 mm
hEC 150 mm

From Table 6.1 it is seen that the magnitudes of the elements sizes that corresponds to the
inherent fracture energies of the four discussed models differ significantly. This highligts how

60
Chapter 6: Fracture Energy Based Uniaxial Material Models at Elevated Temperatures

the current knowledge about the post-peak compressive behaviour is insufficient and supports
the necessecity of an experimental study of the matter.

In Figure 6.4 the obtained compressive fracture energies for the considered example of a con-
crete grade C30, computed from the four models, are plotted as functions of the temperature.
The strength reduction caused by temperature elevation is computed using equation (5.1), as
suggested by Hertz [7], with the reduction parameters of Table 5.1 for a siliceous concrete.

Figure 6.4 shows that the compressive fracture energy obtained using the method suggested
by Li and Purkiss [23], yields results that are very similar to those obtained by the model of
Anderberg and Thelandersson [25]. However, this was expected as the model by Li and Purkiss
[23] is based on the model by Anderberg and Thelandersson [25].

Further, it is seen that in the models by Lie and Lin [26] and Eurocode 2 [21] the compressive
fracture energy increases until a certain temperature, after which a steep descent is visible. For
the model by Lie and Lin [26], it is not until the temperature reaches 600◦ C that the compressive
fracture energy becomes lower than the one at ambient temperature, whereas for the Eurocode 2
[21] model, this is seen to occur at about 500◦ C. The model by Anderberg and Thelandersson [25]
does not incorporate the increase at all, as it is evident that the compressive fracture energy is
reduced for increasing temperatures. This again illustrates how there is thus a clear discrepancy
between the assumptions of the softening behaviour in the existing models. As there currently is
no experimental evidence suggesting that one existing model is better than another, it is highly
encouraged that some experiments are made to investigate this further.

1.5
1500

1
[mm]

1000
c
G /G
cT
maxT
h

500 0.5

0 0
200 400 600 800 1000 200 400 600 800 1000
o
T [ C] T [oC]
Lie and Lin Li and Purkiss
Anderberg and Thelandersson Eurocode 2

Figure 6.4: Comparison of the evolutions with temperature of the compressive fracture energies
obtained when applying the methods of Anderberg and Thelandersson [25], Lie
and Lin [26], Li and Purkiss [23] and Eurocode 2 [21] to equation (6.6), for the
previously described example.

Furthermore, it is interesting that the presence of the LITS does not seem to have an effect
on the compressive fracture energy, as the models including the effects of the LITS, the Li and
Purkiss [23] and the Eurocode 2 [21] models, cannot be said to deviate significantly from the
models that consider solely the instantaneous stress-related strain. Therefore, this is also an
issue it would be interesting to see researched experimentally.

61
Chapter 6: Fracture Energy Based Uniaxial Material Models at Elevated Temperatures

6.3 Formulation of Fracture Energy Based Tensile Model


for Concrete at Elevated Temperatures

The fracture energy based model of the tensile behaviour of concrete at elevated temperatures is
based on the material model by Terro [24] for plain concrete. This model has a linear softening
branch and assumes an ultimate tensile strain of εctuT = 0.004 for all temperatures.

6.3.1 Tensile Fracture Energy at Elevated Temperatures

As described in section 1.2, the tensile fracture energy for plain concrete members is the area
beneath the stress-plastic displacement curve. As the plastic displacement of an element is
related to the plastic strain by δp = εp h, the area beneath the stress-plastic strain curve is equal
to Gf T /h, as illustrated in Figure 6.5.

f ct,m
f ctT

G fT / h
p
ctuT = 0.004

Figure 6.5: Illustration of how the tensile fracture energy changes due to the decrease of the
tensile strength, fctT , at an elevated temperature, T , compared to the strength at
the ambient temperature, fct,m .

Figure 6.5 clearly indicates that for an element of constant size h, the tensile fracture energy
decays as the tensile strength of the concrete decreases. By the simple geometric relation for
the area of a triangle, the tensile fracture energy at elevated temperatures can be expressed
as:
1
Gf T = fctT · εctu h (6.11)
2

The tensile strength of concrete at an elevated temperature is given by the tensile strength at
ambient temperatures multiplied by the decay function, fctT = ξ(T )fct,m . This means that
equation (6.11) can be written as:

1
Gf T = ξ(T )fct,m εctu h (6.12)
2

Equation (6.12) can now be expressed in terms of the tensile fracture energy at ambient tem-
perature, Gf , as this is given by Gf = 1/2fct,m εctu h. Rewriting yields:

1
Gf T = ξ(T ) fct,m εctu h = ξ(T )Gf (6.13)
2

From equation (6.13) it is clearly seen that the evolution of the tensile fracture energy is described
by the same S-shaped function that describes the decay of the strength.

62
Chapter 6: Fracture Energy Based Uniaxial Material Models at Elevated Temperatures

6.3.2 Fracture Energy Based Tensile Model of Plain Concrete

For plain concrete at ambient temperature it is found, using equation (4.1b) for the ultimate
strain of concrete, that for an element size of h = 16.5 mm, the material model by Terro [24]
yields the same result as the fracture energy based model by Feenstra and de Borst [18].

1 1 Fracture
Fracture
Energy
Energy
Based
Based 0.20.2 Fracture
Fracture
Energy
Energy
Based
Based
Terro
Terro Terro
Terro
0.80.8 = o20
T =T20 CoC 0.15
0.15
o o
o o T =T700
= 700
CC
ctT ct,m
ctT ct,m

ctT ct,m
ctT ct,m
0.60.6 T =T300
= 300
CC
" /f
" /f

" /f
" /f
o o 0.10.1
T =T500 CC
= 500 o o
0.40.4 T =T900
= 900
CC
o o
0.05
0.05 T =T1100
= 1100
CC
0.20.2

0 0 0 0
0 0 1 1 2 2 3 3 4 4 0 0 1 1 2 2 3 3 4 4
! ! !" !" ! ! !" !"
x 10
x 10 x 10
x 10
(a) T = 20◦ C, T = 300◦ C and T = 500◦ C (b) T = 700◦ C, T = 900◦ C and T = 1100◦ C

Figure 6.6: Comparison of fracture energy based tensile formulation of the tensile stress-strain
relationship of plane concrete to the model suggested by Terro [24]. An element
of size h = 16.5 mm is considered at temperatures of T = 20◦ C, T = 300◦ C and
T = 500◦ C, (a), and T = 700◦ C, T = 900◦ C and T = 1100◦ C, (b).

In Figure 6.6 an element of plain concrete of grade C30 with size h = 16.5 mm is considered for
a range of temperatures. The material parameters at ambient temperature are given by Table
4.2 and the decay of strength with temperature elevation is computed using equation (5.1), as
suggested by Hertz [7]. Further, the initial E-modulus is found by equation (5.4), as Terro [24]
recommends, where the strain at compressive peak stress is found by equation (5.3), with no
initial load.

It is seen that the fracture energy based formulation correlates well with the model by Terro
[24] for all temperatures.

6.3.3 Fracture Energy Based Tensile Model for Reinforced Concrete

As a result of the fact that the reinforced fracture energy complicates the expressions for the
validity range of the fracture energy based models, it is herein chosen for simplicity not to ex-
tend the formulation to elevated temperatures. This effectively means that the validity range
of the fracture energy based elevated temperature model can be described by a lower and an
upper limitation on the element size and a mimimum reinforcement ratio requirement for the
consideration of the interaction contribution of the tension stiffening.

The tensile formulation is now expanded to include the interaction contribution described in
section 4.2 as per Cervenka et al. [17]. The interaction contribution is still considered to be a
tri-linear function (Figure 4.3b) and is at elevated temperatures defined by the four strains; the
strain at peak stress, ε0tT , the ultimate concrete strain, εcuT , the ultimate strain, εuT , and the
yield strain of the reinforcement, εyT . These four are simply computed using the formulations
at ambient temperatures, with the material properties of the concrete and the reinforcement at
a given elevated temperature. This means that they can be expressed as:

63
Chapter 6: Fracture Energy Based Uniaxial Material Models at Elevated Temperatures

• The strain at peak tensile stress:


fctT
ε0tT = (6.14a)
EciT

• The ultimate concrete strain (extension of equation (4.1b)):


2Gf T
εcuT = (6.14b)
hfctT

• The ultimate strain (extension of equation (4.1c)):


αts fctT
εuT = εyT − (6.14c)
ρs,ef f EsT

• The yield strain of the reinforcement:


fyT
εyT = (6.14d)
EsT

It is apparent from equation (6.14), that the interaction contribution depends on both the
temperature of the concrete as well as the temperature of the reinforcement. As explained in
section 3.3, the temperature of the concrete is not necessarily uniform due to the low thermal
conductivity of concrete. Ultimately, this means than a thermal analysis must be conducted
to determine the temperature of the reinforcement, before the interaction contribution can be
defined.

1 1

0.8 0.8
ctT ct,m

ctT ct,m

0.6 0.6
" /f

" /f

0.4 0.4

0.2 0.2

0 0
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5
! !" ! !"
x 10 x 10
o
o
T = 20 C T = 500 C o
T = 20 C T = 500oC
#$%$"&&oC T = 700oC #$%$"&&oC T = 700oC

Figure 6.7: Combined concrete and interaction stress contributions for a concrete grade C30
with steel Grade 500 for a reinforced member with element size h = 100 mm.

Figure 6.7 illustrates the combined concrete and interaction contribution to the stress when the
steel and the concrete both are at temperature T . A reinforced member of concrete of grade
C30 is considered, with the material properties at ambient temperatures given in Table 4.2,
with reinforcement consisting of steel Grade 500, Table 4.1. The considered member has the
dimensions of b = 100 mm and t = 10 mm, and is assumed to be reinforced in one direction
with an reinforcement area of As = 7 mm2 . This yields a reinforcement ratio of ρs = 0.007.
The strength level of the interaction contribution is taken as αts = 0.7 and the element size
is chosen as h = 100 mm. The strength reductions at elevated temperatures are computed
as suggested by Hertz [7], utilizing the parameters from Table 4.1 and Table 4.2, respectively,
for a siliceous concrete and hot-rolled reinforcement with 0.2% stress. The initial E-modulus

64
Chapter 6: Fracture Energy Based Uniaxial Material Models at Elevated Temperatures

is computed using equation (5.4), in which the strain at maximum compressive stress is found
from equation (5.3), assuming no initial load, i.e. λL = 0.

It should be noted that the loss of bond strength between the concrete the reinforcement is
beyond the current objective and therefore is not accounted for herein. If it had been, the
strength level of the interaction contribution, αts , would most likely have been affected by
assuming a lower value for higher temperatures.

6.4 Limits of Fracture Energy Based Models at Elevated


Temperatures

The ambient temperature validity range of the material models by Cervenka et al. [17] and
Feenstra and de Borst [18] at ambient temperatures was developed in chapter 4. At elevated
temperatures, the fracture energy based model is based on the models by Cervenka et al. [17]
and Feenstra and de Borst [18] and, as a result, it is very relevant to examine the evolution of
the limits caused by an increase in the temperature.

6.4.1 Limitations on the Element Size

As it is chosen herein not to considered the reinforced fracture energy at elevated temperatures,
the limitation on the minimum element size is found by extending equation (4.2). The limitaition
of the minimum and the maximum element size imposed by the uniaxial tensile model is found
simply by extension of equation (4.16) and the limit imposed by the uniaxial compressive model
is redefined by equating expressions (6.1) and (6.2) and solving for h. Combining the limitations
yields the following:

 EciT Gf T


 2 2
2Gf T fctT
! $ ≤ h ≤ min (6.15)
αts fctT 
 9 E G
fctT εyT − 

ciT cT
ρs,ef f EsT 8 ξ(T )fcm
2

In Figure 6.8, the limits on the maximum element size imposed by both the tensile (Figure 6.8a)
and the compressive (Figure 6.8a) models are illustrated for the previously described example.
The compressive limits are illustrated for the formulations of the compressive fracture energy at
elevated temperature based on the four existing models considered.

65
Chapter 6: Fracture Energy Based Uniaxial Material Models at Elevated Temperatures

1000 1.5
1500 1500 Anderberg
Lie and Li
800
Li and Pur
1 Eurocode

[mm]

[mm]
[mm]

1000 1000
600

c
G /G
cT
maxT

maxT
maxT

400

h
h

500 0.5
500
200

0 0 0 0
200 400 600 800 1000 200 400200 600400 800600 1000800 1000 200 400 6
T [oC] T [oC] T [oC] T [o
Lie and Lin Lie and Lin Li and PurkissLi and Purkiss
h l
maxT s Anderberg and Eurocode 2
Anderberg
Thelandersson Eurocode 2
and Thelandersson
EciT Gf T 9 EciT GcT
(a) hmaxT = 2 2
, where ls ≤ hmax (b) hmaxT = 2
fctT 8 ξ(T )fcm

Figure 6.8: Evolution of the maximum element size, hmaxT , with temperature as defined by
equation (6.15) for an example with a reinforced concrete member of grade C30.

From Figure 6.8 it is seen that, for the considered reinforced concrete example, the maximum
element size is governed by the tensile model until about 500◦ C as the lowest value of hmaxT is
found in Figure 6.8a, provided that the compressive fracture energy is computed based on the
Lie and Lin [26] or the Eurocode 2 [21] formulation. It is evident that for this investigated sam-
ple it will be valid to use an element of size h = 300 mm, which seems reasonable for practical
modelling purposes of structures.

However, it is evident for all the found compressive fracture energies that the models are only
practically applicable until a certain temperature, after which a very fine element configuration
is required. It is seen that the models by Anderberg and Thelandersson [25] and Li and Purkiss
[23] only can be applied for temperatures up to about 450◦ C using reasonable element sizes.
Only very small elements can be applied in the range 450-600◦ C, after which it is clearly seen
that the model will no longer be able to produce meaningful results. When computing the
compressive fracture energies inherent in the Eurocode 2 [21] model or the model suggested by
Lie and Lin [26], it is seen that the validity range is increased. However very small element sizes
are still required for temperatures above 600◦ C.

6.4.2 Minimum Reinforcement Ratio

As at ambient temperatures, there are restrictions on the minimum level of reinforcement that
can be considered for the interaction formulation. Problems arise when the ultimate strain is
larger than the ultimate concrete strain. The minimum reinforcement ratio at elevated temper-
atures can be arrived at by extending equation (4.6):

αts fctT
ρs,ef f,minT = ! $ (6.16)
Gf T
εsyT −2 EsT
hfctT

Figure 6.9 illustrates the combined concrete and interaction stress contribution from the previ-
ously desrcibed example of a reinforced concrete member at elevated temperatures where the
temperature of the concrete and the steel are assumed to be equal. The member in the figure is

66
Chapter 6: Fracture Energy Based Uniaxial Material Models at Elevated Temperatures

reinforced in the direction of the applied load, the p-direction, and two reinforcement ratios are
considered, ρp = 0.007 and ρp = 0.0025.

1 1 1 1
T = T20=oC
20oC

0.8 0.8 0.8 0.8


o o
T = T500 C C
= 500
0.6 0.6 0.6 0.6
ct,m
ct,m

ct,m
ct,m
"/f
"/f

"/f
"/f
0.4 0.4 0.4 0.4
o o
T = T700 C C
= 700
0.2 0.2 T#$#"%% o
C C
T#$#"%% o 0.2 0.2

0 0 0 0
0 0 0.5 0.5 1 1 1.5 1.5 2 2 2.5 2.5 0 0 0.5 0.5 1 1 1.5 1.5 2 2 2.5 2.5
! ! !" !" ! ! !" !"
x 10x 10 x 10x 10

T = T20=oC
20oC
: : # =
p
# 0.0025
= 0.0025
p
#p =
#p0.007
= 0.007 T = T500o o
C:C:
= 500 # =
# 0.0025
p
= 0.0025
p
#p =
#p0.007
= 0.007
o o # =
# 0.0025
= 0.0025 # =
# 0.007
= 0.007 o o # =
# 0.0025
= 0.0025 # =
# 0.007
= 0.007
T#$#"%% C:C:
T#$#"%% p p p p T = T700 C:C:
= 700 p p p p
1 1
ρ = 0.0025 ρ = 0.0025
s,eff s,eff
Figure 6.9: Illustration of how the modelling of theρcombined
= 0.007 concrete and interaction stress ρ = 0.007
0.8 s,eff 0.8 s,eff
contributions at different temperatures ρyields unrealistic
= 0.0025 results if the reinforce- ρs,eff = 0.0025
s,eff
ment ratio is too0.6 small. The temperature of the steel is assumed
0.6 to be equal to
ρ = 0.007 = 0.007
ct,m

ct,m
ρ
that of the concrete. s,eff s,eff
σ/f

σ/f
0.4 0.4
From Figure 6.9 it is clear that the reinforcement ratio of ρp = 0.0025 is too low for all tem-
0.2
peratures. Further, the ultimate strain, εuT , is reduced as the temperature0.2 is increased and
hence the problem evolves as the temperature increases. It is therefore of interest to plot the
0
restriction of the minimum reinforcement ratio as 1.5
a function of 2.5 0
the temperature. This is 1done 1.5
0 0.5 1 2 0 0.5 2 2.5
in Figure 6.10. ε −3 ε −3
x 10 x 10
−3
x 10

14

12

10
s,eff
ρ

4
200 400 600 800 1000 1200
T [oC]
ρ
s,eff,minT
ρ = 0.0025
p
ρ = 0.007
p

Figure 6.10: Evolution of minimum reinforcement ratio for the example of a reinforced member
from Figure 6.9 as a function of the temperature.

A plot like the one presented in Figure 6.10 efficiently checks whether the requirement of equation
of (6.16) is fulfilled as it can readily be detected if the used reinforcement ratio, ρs,ef f , is lower
than the minimum reinforcement ratio, ρs,ef f,minT , or not and hence whether the interaction

67
Chapter 6: Fracture Energy Based Uniaxial Material Models at Elevated Temperatures

contribution can be considered or not. From Figure 6.10 it is seen that the considered reinforce-
ment ratio of ρp = 0.0025 is too low for all temperatures, whereas the ratio of ρp = 0.007 can
readily be applied. It is interesting to note the distinct increase in the minimum reinforcement
ratio that occurs for large values of T . This tendency was also seen for other values of ρs , not
included herein. This illustrates the necessity to check if the requirement of equation (6.16) is
fulfilled for all considered temperatures as a case where a reinforcement ratio would be sufficient
at most temperatures can easily be imagined.

68
Numerical Example of a
Reinforced Concrete Slab at
Elevated Temperatures

7.1 Introduction

In this chapter, the fracture energy based material formulations at elevated temperatures are
applied to an example of a reinforced slab. The slab (Figure 7.1) spans over L = 4 m, has a with
of W = 1 m and a depth of D = 150 mm. It is reinforced in the top of the cross-section with
longitudinal bars with the diameter φp = 8 mm and in the bottom with a diameter of φp = 6
mm. Both at top and bottom, a spacing of s = 150 mm is used, yielding a reinforcement ratio
of ρp = 0.0071. The cover layer of all the reinforcement is c = 25 mm.
q = 10.7 kN/m

W=1m

c = 25 mm
! p = 8 mm
D = 150 mm
! p = 6 mm
c = 25 mm
y
50 mm s = 150 mm
x
L=4m
L =4m

Figure 7.1: Illustration of the dimensions of the considered reinforced slab.

The considered slab is fixed in both ends and loaded with a uniformly distributed load of
magnitude q = 10.7 kH/m as illustrated in Figure 7.2.

69
Chapter 7: Numerical Example of a Reinforced Concrete Slab at Elevated Temperatures

q = 10.7 kN/m

W=1m

! p = 8 mm

! p = 6 mm

50 mm s = 150 mm

L=4m
L =4m

Figure 7.2: Illustration of the reinforced concrete slab considered in this example.

Further, the slab is exposed to a elevated temperature from its bottom surface which yields the
temperature distribution in Figure 7.3 through the depth of the slab. The source of the temper-
ature elevation is assumed to be purely convective and radiation is hence not considered.

150
50 oC
72 oC

100 101 oC
D [mm]

137 oC

50 196 oC

388 oC
715 oC
0
0 200 400 600 800
T [oC]

Figure 7.3: Temperature profile within the considered slab.

The analysis is essentially divided into two parts; one where the mechanical load is applied
and one where the thermal load is applied. Both are applied linearly and over a time step of
tF E = 1.00, which means that the total length of the analysis is tF E = 2.00. An overview of the
analysis time is given in Figure 7.4.

Linear application of the


mechanical load Constant mechanical load

t FE
0.00 1.00 Linear application of the 2.00
temperature profile

Figure 7.4: Overview of the time in the FE-analysis of the considered reinforced slab.

7.2 Parameters for the Uniaxial Material Models

The slab is considered to consist of concrete grade C30 and therefore the material properties
given in Table 4.2 and Table 4.3 are used for the ambient condition. The reinforcement consists
of steel Grade 500 and at hence the material properties of Table 4.1 are describing the behaviour

70
Chapter 7: Numerical Example of a Reinforced Concrete Slab at Elevated Temperatures

at ambient temperature.

The compressive concrete model is computed by the extension of the formulation suggested by
Feenstra and de Borst [18], where the equivalent plastic strain at peak, κeT , is given by equation
(6.1) and the ultimate strain, κuCT , by equation (6.2). The tensile model is also based on the
found extension of the model by Feenstra and de Borst [18], given by equations (6.14a-6.14d)
and the reinforcement is assumed to be bi-linear in both tension and compression.

At elevated temperatures, the decay of the tensile and compressive strength of concrete are
computed as suggested by Hertz [7], using the decay parameters of Table 5.1 for a siliceous
concrete in equation (5.1). The initial E-modulus for concrete at elevated temperatures is found
as suggested by Anderberg and Thelandersson [25] by equation (5.4), where the strain at peak
compressive stress is found by the formulation of Terro [24] using equation (5.3).

The compressive fracture energy is computed as inherent in the Eurocode 2 [21] model for com-
pressive concrete behaviour at elevated temperatures using equations (6.6) and (6.10b). The
tensile fracture energy is found by equation (6.13).

The reduction of the yield stress and the E-modulus of the reinforcement are found by equation
(5.1), as suggested by Hertz [14], using the parameters for 0.2% stress of Table 5.4 for hot-rolled
reinforcement.

7.3 Material Properties for the Thermal Analysis

A thermal analysis must be conducted to account for the stresses that arises within the slab
as a result of the restriction of thermal expansion imposed by the fixed ends. As a results, it
is necessary to define the thermal properties of the materials. As per common practice, the
conductivity and the density of the concrete is taken as defined in Table 7.1.

Table 7.1: Parameters at ambient temperature used for the thermal analysis of concrete as
recommended by Teknisk Ståbi [29].

Density Conductivity Specific Heat


2400 kg/m 3
1.7 W/(m C) ◦
900 J/(kg◦ C)

The thermal expansion is in ABAQUS computed by the function *EXPANSION which uses the
user-defined expansion coefficient, α. According to Youssef and Moftah [22] the thermal expan-
sion of concrete is typically found by the following linear function using the thermal expansion
coefficient, αconcrete :

εth = αconcrete · (T − 20◦ C) (7.1a)

However, as it here is desired to evaluate the thermal expansion, εth , by equation (5.7a), as
described in section 5.3.1, an expression for the thermal expansion coefficient is arrived at by
rewriting equation (7.1a):

εth
αconcrete = (7.1b)
(T − 20◦ C)

71
Chapter 7: Numerical Example of a Reinforced Concrete Slab at Elevated Temperatures

For the concrete considered herein, the thermal expansion coefficient as a function of the tem-
perature is presented in Figure 7.5.

!"
x 10

!"!concrete [oC!#] 4

200 400 600 800 1000 1200


T [oC]

Figure 7.5: Thermal expansion coefficient for concrete, αconcrete , as a function of the temper-
ature for the considered example of a reinforced concrete slab.

The expansion of the steel is typically assumed to be linear with an expansion coefficient of
αsteel = 1.248 · 10−5 at T = 100◦ C [29].

7.4 FE-Analysis

In the depth of the slab it is necessary to make a division into six elements in order to apply
the temperature profile with reasonable accuracy. However, the magnitude of the element depth
has no effect on the definition of the material models, as this is not the direction where the
localization of deformation is taking place. Nevertheless, it is necessary to investigate the validity
range for the element sizes of the material models in order to make an appropriate element
division along the span on the slab. By application of equation (6.15) as a functions of the
temperature, the validity range can be illustrated as the grey area indicated in Figure 7.6.

1500

1000
h [mm]

500

0
200 400 600 800 1000
T [oC]
h , Compressive Parameters h
maxT minT
h , Tensile Parameters Admissible values of h
maxT

Figure 7.6: Limits on the maximum and minimum element size, equation (6.15), as functions
of the temperature for the considered example of a reinforced slab.

72
Chapter 7: Numerical Example of a Reinforced Concrete Slab at Elevated Temperatures

From Figure 7.6 it is seen that it is not possible to perform analysis for slab temperatures above
800◦ C. As the maximum temperature slab regarded in this example is 715◦ C, it is possible to
obtain meaningful results. However, it is evident from Figure 7.6, that only very small elements
can be evaluated for temperatures exceeding 600◦ C. For the temperature considered herein,
715◦ C, it is found that the element size must be chosen within the range;

72.5 mm ≤ h ≤ 129.6 mm

Further, it is necessary to verify that the interaction stress contribution can be evaluated with
the level of reinforcement that is present in the slab. This is done in Figure 7.7 by plotting
equation (6.16) as functions of the temperature for the two extreme element sizes of h = 73 mm
and h = 129 mm.

0.025 0.025 0.025


!s,eff,minT !s,eff,minT !s,eff,minT
!p = 0.0071 0.02 !p = 0.0071 0.02 !p = 0.0071 0.02

0.015 0.015 0.015


!s,eff

!s,eff

!s,eff
0.01 0.01 0.01

0.005 0.005 0.005

0 0 0
400 600 800 1000 200 400 600 800 1000 200 400 600 800 1000 200
T [oC] T [oC] T [oC]
(a) h = 73 mm (b) h = 129 mm

Figure 7.7: Verification of the requirement to the minimum level of reinforcement (equation
(6.16)) that can be considered for validity of the interaction stress contribution
of the tension stiffening for the considered example of a reinforcement slab with
element sizes of h = 73 mm, (a), and h = 129 mm, (b).

From Figure 7.7 it is clearly seen that it is valid to consider the interaction stress contribution
as a part of the tension stiffening as the effective reinforcement ratio in the slabs is higher than
the minimum requirement for both the extreme element sizes.

7.4.1 Element size h = 129 mm

The least CPU-heavy FE-analysis is made where the length of the slab is divided into 31 ele-
ments, yielding an element size of h = 129 mm.

The uniaxial material models are defined for temperatures up to 715◦ C and are given in full
in appendix D. As examples, the compressive stress-strain relation and the tensile combined
concrete and interaction stress contributions used for the FE-analysis are illustrated in Figure
7.8 for T = 20◦ C, T = 300◦ C, T = 500◦ C and T = 715◦ C.

73
Chapter 7: Numerical Example of a Reinforced Concrete Slab at Elevated Temperatures

40
40 ""

2.5
2.5
"#
"#
22

"ctT [MPa]
[MPa]
[MPa]
" [MPa]

20
20 1.5
1.5

ctT
cT
cT
"

"
11
10
10
0.5
0.5

00 00
00 0.005
0.005 0.01
0.01 0.015
0.015 0.02
0.02 00 0.5
0.5 11 1.5
1.5 22 2.5
2.5
!cT
!cT !cT
!cT !"!"
xx10
10
oo oo
500oC
oo o
TT==20
20CC TT==500
500CC TT==20
20CC TT==500 C
T$%$"##oC
T$%$"## o
C 715oC
TT==715 o
C T$%$"##oC
T$%$"## o
C TT==715 oo
715CC
(a) Compression (b) Tension

Figure 7.8: Material models for compression, (a), and tension, (b), for the reinforced slab with
an element size of h = 129 mm.

The analysis is performed as a plane strain analysis where the slab is modelled using beam
elements and the *REBAR function is used to define both layers of the reinforcement within the
concrete elements.

As the largest stresses in the slab will occur at the fixed ends, the elements in this region are the
focus of the post-processing. For the analysis of the contour plots obtained from the FE-analysis
in ABAQUS, the three rows of elements neer the right hand sided fixed end, shown in Figure
7.9, are considered.

Figure 7.9: Position of the considered element for the post-processing of the contour plots from
ABAQUS.

In Figure 7.10 the obtained stresses in the x-direction for the elements in Figure 7.9 are shown
at the analysis times where the temperature load is applied (tF E = 1.00), an intermediate time
(tF E = 1.50) and when the temperature load is fully applied (tF E = 2.00).

74
Chapter 7: Numerical Example of a Reinforced Concrete Slab at Elevated Temperatures

S, S11
(Avg: 75%)
S, S11-"#)),'-&,
(Avg:-"#+)$'-&,
75%)
-"#".('-&,
-"#)),'-&,
-.#.*$'-&*
-"#+)$'-&,
S, S11-*#.$"'-&*
-"#".('-&,
(Avg:-%#$%('-&*
75%)
-.#.*$'-&*
!+#*(+'-&(
-*#.$"'-&*
-"#)),'-&,
!(#&"('-&*
-%#$%('-&*
-"#+)$'-&,
!*#$."'-&*
!+#*(+'-&(
-"#".('-&,
!.#$+$'-&*
!(#&"('-&*
-.#.*$'-&*
!"#"$%'-&,
!*#$."'-&*
-*#.$"'-&*
!"#+.$'-&,
!.#$+$'-&*
-%#$%('-&*
!"#).*'-&,
!"#"$%'-&,
!+#*(+'-&(
!"#+.$'-&,
!(#&"('-&*
!"#).*'-&,
!*#$."'-&*
!.#$+$'-&*
!"#"$%'-&,
(a) tF E = 1.00
!"#+.$'-&,
S, S11!"#).*'-&,
(Avg: 75%)
S, S11-$#)+)'-&*
(Avg:!+#..('-&,
75%)
!"#&)+'-&)
-$#)+)'-&*
!"#,,&'-&)
!+#..('-&,
S, S11!%#%+,'-&)
!"#&)+'-&)
(Avg:!"#,,&'-&)
75%)
!%#.("'-&)
!(#+")'-&)
!%#%+,'-&)
-$#)+)'-&*
!+#&&('-&)
!%#.("'-&)
!+#..('-&,
!+#*.$'-&)
!(#+")'-&)
!"#&)+'-&)
!*#")*'-&)
!+#&&('-&)
!"#,,&'-&)
!*#),&'-&)
!+#*.$'-&)
!%#%+,'-&)
!,#(+,'-&)
!*#")*'-&)
!%#.("'-&)
!,#$(%'-&)
!*#),&'-&)
!(#+")'-&)
!,#(+,'-&)
!+#&&('-&)
!,#$(%'-&)
!+#*.$'-&)
!*#")*'-&)
S, S11!*#),&'-&)
(Avg: !,#(+,'-&)
75%)
!,#$(%'-&)
(b) tF E = 1.50
S, S11
!(#"*.'-&,
(Avg:!)#+**'-&,
75%)
!"#")*'-&)
!(#"*.'-&,
!"#,&*'-&)
!)#+**'-&,
S, S11
!%#&(*'-&)
!"#")*'-&)
(Avg:!"#,&*'-&)
75%)
!%#+,+'-&)
!%#.$+'-&)
!%#&(*'-&)
!(#"*.'-&,
!(#(%+'-&)
!%#+,+'-&)
!)#+**'-&,
!(#)*('-&)
!%#.$+'-&)
!"#")*'-&)
!+#".('-&)
!(#(%+'-&)
!"#,&*'-&)
!+#,"('-&)
!(#)*('-&)
!%#&(*'-&)
!*#&+('-&)
!+#".('-&)
!%#+,+'-&)
!*#+)%'-&)
!+#,"('-&)
!%#.$+'-&)
!*#&+('-&)
!(#(%+'-&)
!*#+)%'-&)
!(#)*('-&)
!+#".('-&)
!+#,"('-&)
!*#&+('-&)
!*#+)%'-&)

(c) tF E = 2.00

Figure 7.10: Output from ABAQUS analysis of stress in the x-direction (S11) at various times,
where tF E = 1.00 coresponds to the onset of the temperature load.

From the stress contours in Figure 7.10 it is evident that the stress in x-direction (S11) does
not continuously increase with the application of the elevated temperatures. It is seen that it
decreases after a certain peak, because the deformations occurring are in the plastic regime and
thus that irreversible macro-cracking is occuring.

Figure 7.11 shows the obtained plastic strains in the x-direction for the elements in Figure
7.9 are shown at the analysis times where the temperature load is applied (tF E = 1.00), an
intermediate time (tF E = 1.50) and when the temperature load is fully applied (tF E = 2.00).
When considering contours of the plastic strains in the x-direction (PE11), the evolution of the
irriversible deformations are seen to be progressing as the temperature profile is applied.

75
Chapter 7: Numerical Example of a Reinforced Concrete Slab at Elevated Temperatures

PE, PE11
(Avg: 75%)
+0.000e+00
PE, PE11
+0.000e+00
(Avg:+0.000e+00
75%)
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
PE, PE11
+0.000e+00
+0.000e+00
(Avg:+0.000e+00
75%)
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
+0.000e+00
(a) tF E = 1.00 +0.000e+00
+0.000e+00
+0.000e+00
PE, PE11
+0.000e+00
(Avg:+0.000e+00
75%)
+0.000e+00
-+#),*'!&+
PE, PE11
-%#),.'!&+
(Avg:-)#)"%'!&*
75%)
!"#%%,'!&+
-+#),*'!&+
!(#%%('!&+
-%#),.'!&+
!*#%%&'!&+
PE, PE11
-)#)"%'!&*
(Avg:!)#%",'!&+
75%)
!"#%%,'!&+
!$#%"('!&+
!(#%%('!&+
-+#),*'!&+
!"#"%"'!&(
!*#%%&'!&+
-%#),.'!&+
!"#(%"'!&(
!)#%",'!&+
-)#)"%'!&*
!"#*%&'!&(
!$#%"('!&+
!"#%%,'!&+
!"#)%&'!&(
!"#"%"'!&(
!(#%%('!&+
!"#$%&'!&(
!"#(%"'!&(
!*#%%&'!&+
!"#*%&'!&(
!)#%",'!&+
!"#)%&'!&(
!$#%"('!&+
!"#$%&'!&(
!"#"%"'!&(
PE, PE11
!"#(%"'!&(
(Avg:!"#*%&'!&(
75%)
(b) tF E = 1.50 -$#,$"'!&+
!"#)%&'!&(
!"#$%&'!&(
PE, PE11
-*#"$"'!&+
(Avg:-,#$%+'!&*
75%)
!(#.&)'!&+
-$#,$"'!&+
!.#(&,'!&+
-*#"$"'!&+
!"#%.&'!&(
PE, PE11
-,#$%+'!&*
!"#)(&'!&(
(Avg:!(#.&)'!&+
75%)
!%#".&'!&(
!.#(&,'!&+
-$#,$"'!&+
!%#,(&'!&(
!"#%.&'!&(
-*#"$"'!&+
!(#&.&'!&(
!"#)(&'!&(
-,#$%+'!&*
!(#*(&'!&(
!%#".&'!&(
!(#.&)'!&+
!(#$.&'!&(
!%#,(&'!&(
!.#(&,'!&+
!+#+(&'!&(
!(#&.&'!&(
!"#%.&'!&(
!(#*(&'!&(
!"#)(&'!&(
!(#$.&'!&(
!%#".&'!&(
!+#+(&'!&(
!%#,(&'!&(
!(#&.&'!&(
!(#*(&'!&(
!(#$.&'!&(
!+#+(&'!&(

(c) tF E = 2.00

Figure 7.11: Output from ABAQUS analysis of plastic strain in the x-direction (PE11) at
various times, where tF E = 1.00 coresponds to the onset of the temperature load

The evolution of the plastic deformations can also be illustrated by considering the output of a
single element. As an example, element 2 is considered, the position of which is illustrated in
Figure 7.12 along with the locations of its integration points.

P3 P4

P1 P2

Figure 7.12: Position of element 2 and an indication of the location of the integration points
within it.

The obtained stresses in the x-direction, σx , and plastic strains, εpx , in element 2 from the
FE-analysis are given in Figure 7.13 as functions of the analysis time, tF E .

It is seen that during the application of the temperature load at tF E , the deformations becomes
increasingly plastic. Furthermore, it is, as expected, seen that the position of the element and
the integration points within it, has an effect on the plastic deformations as the peak compressive
stress is observed to be reached at a later time for integration points deeper into the slab.

76
Chapter 7: Numerical Example of a Reinforced Concrete Slab at Elevated Temperatures

7 7 !# !#
x 10x 10 x 10x 10

0 0 0 0

!% !%
!% !%
!$ !$
!x
!x

"p
"p
x

x
!$ !$
P1 P1 !# !# P1 P1
P2 P2 P2 P2
!# !#
&# &# !" !" &# &#
&" &" &" &"
!" !" !' !'
0 0 0.5 0.5 1 1 1.5 1.5 2 2 0 0 0.5 0.5 1 1 1.5 1.5 2 2
t t t t
FE FE FE FE

(a) Stress in the x-direction (b) Plastic strain in the x-direction

Figure 7.13: Evolution of the stress and the plastic strain the x-direction in the integration
points of element 2.

7.4.2 Element size h = 73 mm

The underlying assumption that states that the fracture energy based material model is mesh
independent is briefly investigated by analyzing the slab using a finer mesh. The span is therefore
divided into 55 elements yielding an element size of h = 73 mm. The material models are
developed for this element configuration can be seen in full in appendix D.

"'#
"'# "'#
"'#
h%+%"!,%))
h%+%"!,%))
h%+%-.%))
h%+%-.%))

"##
"## "##
"##
D ())*
())*

D ())*
())*
D

'#
'# '#
'#

h%+%"!,%))
h%+%"!,%))
h%+%-.%))
h%+%-.%))
## ##
!!
!! !"
!" ## "" !! !&
!& !'
!' !/
!/ !.
!. !!
!! !"
!" ##
!!x x && !!x x --
$%"#
$%"# $%"#
$%"#
(a) tF E = 1.00 (b) tF E = 2.00

Figure 7.14: Stress in the x-direction through the thickness of the slab at the left fixed end at
times tF E = 1.00, (a), and tF E = 2.00, (b), for element configurations of h = 129
mm and h = 73 mm, respectively.

Figure 7.14 shows the output from the two FE-analyses using h = 129 mm and h = 71 mm,
respectively. The considered output is the stresses in x-direction through the depth of the slab
at the left fixed end. Generally, except the dosagreement at the top of the slab at analysis time
tF E = 2.00, the outputs of the two analyses are seen to be alike and better than the one obtained
was not to be expected.

77
Chapter 7: Numerical Example of a Reinforced Concrete Slab at Elevated Temperatures

78
Conclusion

8.1 Remarks in Conclusion

Herein, the scope has been to extend the existing fracture energy based models for tension
and compression from ambient conditions to elevated temperatures. The existing models were
evaluated prior to the extension and it was found that it was necessary to define limits of
application in order to ensure meaningful results of a FE-analysis. A restriction on the admissible
element size has therefore been developed by the formulation of a maximum element size, given
by equation (4.16). Further, if modelling the tension stiffening through the definition of an
interaction stress contribution, it has been established that a sufficient level of reinforcement
must be ensured. This is most efficiently verified by making sure that inequality (4.9) is fulfilled.
The two requirements have been computed for a concrete grade C30 to examine the practical
implications and it has been found that:

• For a concrete grade C30 a maximum element size of 756 mm is admissible which seems
reasonable for practical application.

• For a considered example of a reinforced plate with a reinforcement ratio of ρp = 0.03


(typically a reinforcement level of 2-3% is used in practice), the requirement on the mini-
mum level of reinforcement is found to be fulfilled when the element size is 100 mm.

The extension of the fracture energy based models to elevated temperatures is largely done by
considering the ambient models with the material properties of a given elevated temperature.
The compressive concrete model is given by equations (6.1-6.2) and the tensile model for the
concrete and the interaction stress contribution of the tension stiffening effect is expressed by
equation (6.14). As experimental data on this is not currently available on the evolution of
the tensile and compressive fracture energies at elevated temperatures, the fracture energies
inherent in the existing elevated temperature models have been computed. The compressive
fracture energy has been found based on the models by Anderberg and Thelandersson [25], Lie
and Lin [26], Li and Purkiss [23] and Eurocode 2 [21]. Of these models, the two first considers
solely the instantaneous stress-related strain whereas the two latter includes the effects of the
LITS. The most important findings upon comparison of the four investigated compressive models
are:

• A significant spread in the post-peak behaviours of the existing models has been found.

• It has been seen that the LITS cannot be said to have an effect on the compressive fracture
energy.

• Further, it is seen that the compressive fracture energy inherent in the models by Eurocode
2 [21] and Lie and Lin [26] operates with an increase until a certain temperature in the
range of 500-600◦ C after which the fracture energy will be lower than at ambient condition.

79
Chapter 8: Conclusion

• The evolutions of the compressive fracture energy computed using the models by Ander-
berg and Thelandersson [25] and the Li and Purkiss [23] are seen not to incorporate any
increase, but to decrease continuously following a S-shaped curve. The two have been
seen to be very closely related, however this is not accredited any special significance
as the model by Li and Purkiss [23] is based on the Anderberg and Thelandersson [25]
formulation.
When investigating the current models for tensile behaviour of concrete at elevated temperatures
it is found that the knowledge base is extremely limited. One material formulation has been
identified; the one suggested by Terro [24] and when computing the inherent fracture energy it
is found that:
• The tensile fracture energy follows the decay function for the material strength and thus
can be described by equation (6.13).

The validity range for the element sizes of the fracture energy based elevated temperature model
has been formulated and is given by expression (6.15). To illustrate the impact of a temperature
variation on the validity range, the evolution of the admissible element sizes with temperature
have been investigated for an example of concrete grade C30.
• It is found that up to about 500◦ C the maximum element size is governed by the tensile
properties after which the compressive parameters are governing;
• as a result, the maximum element size above 500◦ C depends greatly upon the chosen
method for the computation of the compressive fracture energy. When applying the models
by Anderberg and Thelandersson [25] and by Li and Purkiss [23], a very fine mesh is
required for analysis of temperatures in the range 500-700◦ C, and analysis beyond 800◦ C
is not possible. However, the same material can be evaluated for temperatures up to
1100◦ C, when the compressive fracture energy is computed based on the models suggested
by Eurocode 2 [21] and by Lie and Lin [26], but a fine mesh requirement is still demanded
for temperatures above 600◦ C.
At elevated temperatures, a requirement on the minium reinforcement ratio that has to be
present if the tension stiffening must include the interaction contribution has been defined by
equation (6.16). For an example of a reinforced slab with ρp = 0.0071, this is found to be fulfilled
for all temperatures.

It is found that the model developed herein can readily be applied for a FE-analysis of reinforced
concrete at elevated temperatures, for example in ABAQUS, and, if the modelling is performed
within the limits of application, it is possible to get mesh independent results of the analysis
with different mesh configurations.

8.2 Suggestions for Future Research

In order to ensure proper use of the developed fracture energy based models at elevated tem-
peratures, experimental studies of the evolution of the fracture energies with temperature are
recommended as future research:
• An experimental study of the compressive fracture energy at ambient and at elevated
temperatures, including the effect of the LITS; and
• An experimental study of the tensile fracture energy at elevated temperatures.

80
Chapter 8: Conclusion

Further, it should be noted, that the fracture energy based model developed herein is subject
to some simplifications. An expansion of the model at elevated temperatures to include some
of the phenomena that is not considered herein is therefore also suggested as scope of future
investigations:
• The effect of the LITS on the compressive behaviour; both on the behaviour prior and
post-peak, and considering both the direction where the load is present as well as the
response of the unloaded directions.
• The loss of bond strength between the concrete and the reinforcement caused by a tem-
perature elevation. This could perhaps be modelled through a decay of the stress level of
the interaction contribution.
• Expanding the model to the cold phase of a fire.
• Spalling of the concrete surface. This would demand for the model to be extended to a
comprehensive thermo-hydro-mechanical finite element analysis in order to include calcu-
lations on the pore water pressure.

81
Chapter 8: Conclusion

82
References

[1] http://www.alphabetics.info/international/?p=3031 (Accessed 2011.01.07).


[2] fhm.fhsd.k12.mo.us (Accessed 2011.01.07).
[3] Hoffman, M. and Fleming, M. Chernobyl: The True Scale of the Accident. Press re-
lease, International Atomic Energy Agency, World Health Organization and United Nations
Development Programme (2005).
[4] http://www.atmb.net/atmb/en/tunnel/25/the-mont-blanc-tunnel/the-
history-of-the-tunnel/the-fire-of 1999.html (Accessed 2011.01.07).
[5] van Mier, J. G. M. Strain-Softening of Concrete Under Multiaxial Loading Conditions.
Ph.D. thesis, University of Technology Eindhoven (1984).
[6] Pankaj, P. Finite Element Analysis in Strain Softening and Localisation Problems. Ph.D.
thesis, University of Wales (1990).
[7] Hertz, K. D. Concrete Strength for Fire Safety Design. Magazine of Concrete Reseach,
57(8), (2005), 445–453.
[8] SIMULIA. ABAQUS Version 6.7 Documentation. Dassault Systemes (2010). URL http:
//abaqus.civil.uwa.edu.au:2080/v6.7/.
[9] Pankaj, P. Real Structural Behavior and Its Analysis - Material Nonlinearity. Lecture Note
(Accessed 2010.11.22). URL www.see.ed.ac.uk/~pankaj/Real_Structural_Behavior/.
[10] Fletcher, I. A., Welch, S., Torero, J. L., Carvel, R. O., and Usmani, A. Behaviour
of Concrete Structures in Fire. Thermal Science, 11(2), (2007), 37–52.
[11] Mindess, S., Young, J., and Darwin, D. Concrete. Prentice Hall, Pearson Education,
Inc. (2003).
[12] Khoury, G. A. Effect of Fire on Concrete and Concrete Structures. Progress in Structural
Engineering and Materials, 2(4), (2000), 429–447.
[13] Nielsen, A. Bygningsmaterialer - Metallære. Polyteknisk Forlag (1998).
[14] Hertz, K. D. Reinforcement Data for Fire Safety Design. Magazine of Concrete Reseach,
56(8), (2004), 453–459.
[15] Hertz, K. D. Limits of Spalling of Fire-Exposed Concrete. Fire Safety Journal, 38(2),
(2003), 103–116.
[16] CEB. CEB-FIB Model Code. Thomas Telford (1993).
[17] Cervenka, V., Pukl, R., and Eligehausen, R. Computer Simulation of Anchoring Tech-
nique in Reinforced Concrete Beams. In Bićanić, N. and Mang, H., editors, Computer
Aided Analysis and Deign of Concrete Structures, volume 1. Pineridge Press (1990), 1–21.

83
Appendix 8: Post-Peak Response of Concrete

[18] Feenstra, P. and de Borst, R. Constitutive Model for Reinforced Concrete. Journal of
Engineering Mechanics, ASCE, 121, (1995), 587–595.
[19] Nakamura, H. and Higai, T. Compressive Fracture Energy and Fracture Zone Length
of Concrete. In Shing, P. B. and Tanabe, T.-A., editors, Modeling of Inelastic Behavior
of RC Structures under Seismic Loads. ASCE (2001), 471–487.
[20] Vonk, R. Softening of Concrete Loaded in Compression. Ph.D. thesis, Eindhoven Univer-
sity of Technology (1992).
[21] CEN. Eurocode 2: Design of Concrete structures - Part 1-2: General rules - Structural
fire design, BS EN 1992-1-2:2004. Brithish Standard (2004).
[22] Youssef, M. and Moftah, M. General Stess-Strain Relationship for Concrete at Elevated
Temperatures. Engineering Structures, 29(10), (2007), 2618–2634.
[23] Li, L.-Y. and Purkiss, J. Stress-Strain Constitutive Equations of Concrete Material at
Elevated Temperatures. Fire Safety Journal, 40(7), (2005), 669–686.
[24] Terro, M. J. Numerical Modeling of the Behavior of Concrete Structures in Fire. ACI
Structural Journal, 95(2), (1998), 183–193.
[25] Anderberg, Y. and Thelandersson, S. Stress and Deformation Characteristics of Con-
crete at High Temperatures - 2. Expirimental Investigation and Material Behaviour Model.
Lund Inistitute of Technology (1976).
[26] Lie, T. T. and Lin, T. D. Fire Performance of Reinforced Concrete Columns. Fire Safety:
Science and Engineering, ASTM STP, 882, (1985), 176–205.
[27] Law, A. and Gillie, M. Load Induced Thermal Strain: Implications for Structual Be-
haviour. SiF, Singapore (2008).
[28] Law, A., Gillie, M., and Pankaj, P. Incorporation of Load Induced Thermal Strain in
Finite Element Models. Application of Structural Fire Engineering, Prauge, Czech Republic
(2009).
[29] Teknisk Ståbi. Nyt Teknisk Forlag, 19 edition (2007).

84
Appendix A:
Detailed Description of Cracking
and the Post-Peak Response of
Concrete

This appendix contains a detailed overview of the crack propagation causing the softening of the
post-peak response and the failure of the concrete. The softening arises due to the initiation of
cracks and their propagation and the post-peak response is the reduced load carrying capacity
with increasing deformation.

A.1 Crack Propagation and Softening

As described by Mindess et al. [11], concrete is a composite material with a microstructure


consisting of aggregate and cement paste separated by a interfacial transition zone. The het-
erogeneous nature causes the behaviour to differ significantly in tension and in compression as
the localized stresses in the material become very different from the nominal stresses. In most
cases, the interfacial transition zone acts as a weak link when stresses are applied. The strength
of this zone is lower than the strength of the cement paste and bond failure can arise as a con-
sequence of differences in elastic moduli between hardened paste and aggregate, but also due to
different coefficients of thermal expansion and different responses to change in moisture content;
ultimately causing microcracking.

Compressive force
Bond failure
due to tension Potential shear plane

Intact bond
Aggregate
particle
Tensile strain
Bond failure
due to tension
and shear
Cement matrix

Figure A.1: Idealization of stresses around a single aggregate particle. Reproduced from Min-
dess et al. [11].

85
Appendix A: Post-Peak Response of Concrete

Crack propagation is more prone to occur under tensile loads than compressive as there are
low frictional forces, and hence the tensile strength is lower than the compressive. Vonk [20]
describes how one of the prevailing beliefs is that failure in compression is really a secondary
tensile failure induced by the application indirect tension. This is illustrated in Figure A.1,
where compressive failure on the microscopic level is illustrated. Once microscopic cracking has
commenced, internal crack growth continues with force application to the element or structure.

On a macroscopic level the internal crack growth will result in reduced load carrying capacity
often termed as strain softening. This is evident in displacement-controlled test specimens,
where it is clearly detected as the post-peak decrease of mechanical resistance.

! Formation of macrocracks

!peak

Microcracking

Kt < 0

Kt

Pre Peak Post Peak

wpeak w
Figure A.2: Characteristic nominal stress-deformation relation of a loaded specimen in com-
pression under displacement controlled test. Reproduced from Mindess et al. [11].

In Figure A.2 a characteristic nominal stress-deformation relation of a concrete test specimen


undergoing a displacement controlled test is shown. The tangential stiffness, Kt , continuously
decreases in the pre-peak regime as the load is increased until it reaches zero at the peak load.
The continuous post-peak decrease of the mechanical resistance caused by the continued increase
of the deformations is called softening and it is characterized by the descending branch of the
stress-deformation curve, which has a negative tangential stiffness. The softening at macroscopic
level is caused by the crack growth as the material is gradually weakened when the internal bonds
are broken at microscopic level. Initially the cracking starts as distributed microcracking. This
is a stable process, which means that the cracking only grows when the load is increased. Around
peak load the formation of macrocracks begins. These cracks are unstable, which means that
the load has to decrease to avoid unstable growth.

86
Appendix B:
ABAQUS Functions for
Definition of Uniaxial Behaviour,
Embedment of Reinforcement
and Load Steps

This appendix contains a brief description of the *TENSION STIFFENING and the *COMPRESSION
HARDENING functions that has to be defined in order to utilize the ABAQUS concrete damaged
plasticity model. Further, the functions used to embed the reinforcement in the concrete surfaces
and the to define the load steps are described.

B.1 Tension Stiffening and Compression Hardening Mod-


els

When creating a FE-model in ABAQUS, the softening of concrete is defined through *CONCRETE
TENSION STIFFENING for tension and in compression by the use of *CONCRETE COMPRESSION
HARDENING. The behaviour in both tension and compression are defined by equivalent damaged
strain parameters; the cracking strain, εck
t , in tension and the inelastic strain, εc , in compression.
in

These are defined as the total strain at the considered point minus the corresponding elastic
strain for the undamaged material, and are thus described by the following relations [8]:

t = εt − σt /Ec
εck (B.1)
εin
c = εc − σc /Ec (B.2)

The cracking strain and the in-elastic strain for the ABAQUS definitions of tension stiffening
and compression hardening are illustrated in Figure B.1.

87
Appendix B: ABAQUS Functions

"t "c

"t 0 " cu

" c0
Ec
Ec
Ec Ec

!t !c
! tck ! cin
(a) Cracking strain for definition of tension stiffen- (b) Inelastic strain for definition of compression
ing hardening

Figure B.1: Definition of cracking and inelastic strain. Reproduced from the ABAQUS Version
6.7 Documentation [8].

For tension stiffening, ABAQUS has the possibility of a fracture energy definition via specifying
TYPE=GFI, where the fracture energy is defined as a constant and hence the material tensile
definition does not need to be altered for changes in the element side length. However, as the area
underneath the stress-plastic strain curve is deemed to be triangular, the interaction contribution
of the tension stiffening effect defined by Feenstra and de Borst [18] (section 4.3.3) cannot be
included in the model because the tension stiffening input must be the combined concrete and
interaction contribution (tri-linear). It is therefore necessary to utilize the TYPE=STRAIN option,
where the stress-strain diagram for the post-peak relation is point-wise defined.

B.2 Embedment of Reinforcement

As previously described, the reinforcement is modelled as rods embedded in the concrete sur-
faces. This means that the end nodes of the steel rods are considered to be slave nodes to the
concrete master nodes, and thus, that the steel nodes follow the deformations of the concrete
nodes.

The function *TIE is used to define the slave and the master surfaces.

*TIE
-Slave Surface-, -Master Surface-

B.3 Load Step Definition for Static Analysis

A load step in ABAQUS is created by *STEP and *END STEP, and in between these, the type of
analysis is defined and the relevant load is applied.
Herein, only static analyses are performed, and all loads are therefore defined using *STATIC.

Newton’s method is used in ABAQUS to solve nonlinear static problems [8]. This means that
each defined step is divided into increments in which a force is applied. When a force is applied,
iterations are performed within it until the residual forces equals the applied force with a defined

88
Appendix B: ABAQUS Functions

margin of error. The stiffness correlation of the member is updated for each iteration.

The function *STATIC makes it possible for the user to control the chosen incrementation some-
what.
*STATIC
-Initial Increment-, -Period-, -Min. Increment-, -Max. Increment-

The parameters to be user-defined controls the following:

-Initial Increment- the initial increment


-Period- the period of the load step
-Min. Increment- the minimum allowed increment
-Max. Increment- the maximum allowed increment

It is possible to select either an automatic or a direct increment division when defining a step.
The automatic choice is generally quicker and less CPU demanding and as a lot of experience
is required to implement the direct increment division, the automatic incrementation division is
chosen herein.

The specification of the minimum and maximum time increments allowed are only used if auto-
matic time incrementation is used. If nothing is specified for the maximum time increment, no
upper limit will be imposed.

89
Appendix B: ABAQUS Output from Pure Shear Example of Simple Plates

90
Appendix C:
ABAQUS Output from Pure
Shear Example of Simple Plates
with and without Reinforcement

This appendix contains the output from the ABAQUS run of the simple shear example from
section 4.6.3.

By examination of the outputs from the two runs, it can be seen that they are exactly alike
and thus that the reinforcement has no effect for the case which the plate is subjected to pure
shear.

C.1 Simple Shear Example without Reinforcement

Table C.1: Output from ABAQUS for a simple shear example without reinforcement at time
increments 7, 19, 22 and 410.

Node Integration Point Direction δ ε σ P


[mm] [-] [MPa] [N]
Increment no 7
11 -0.00375 -3.44·10−23 -0.00277 -1251.02
1 1 22 -0.00375 0 -0.00277 -1251.02
12 0.00015 2.5048
11 0.00375 -3.44·10−23 -0.00277 -1253.79
2 3 22 -0.00375 0 -0.00277 -1253.79
12 0.00015 2.5048
11 0.00375 3.44·10−23 -0.00277 1251.02
3 4 22 0.00375 2.12·10−22 -0.00277 1251.02
12 0.00015 2.5048
11 -0.00375 3.44·10−23 -0.00277 -1253.79
4 2 22 0.00375 2.12·10−22 -0.00277 1253.79
12 0.00015 2.5048

91
Appendix C: ABAQUS Output from Pure Shear Example of Simple Plates

Increment no 19
11 -0.52104 -5.39·10−20 -9.82075 -375.896
1 1 22 -0.52104 -9.11·10−21 -9.82075 -375.896
12 0.02084 10.5725
11 0.52104 5.39·10−20 -9.82075 10196.6
2 3 22 -0.52104 0 -9.82075 -10196.6
12 0.02084 10.5725
11 0.52104 -5.39·10−20 -9.82075 375.896
3 4 22 0.52104 6.69·10−20 -9.82075 375.896
12 0.02084 10.5725
11 -0.52104 -5.39·10−20 -9.82075 -10196.6
4 2 22 0.52104 3.52·10−20 -9.82075 10196.6
12 0.02084 10.5725

Increment no 22
11 -1.02605 -9.04·10−19 -3.54567 222.111
1 1 22 -1.02605 1.97·10−19 -3.54567 222.111
12 0.041042 3.10145
11 1.02605 -9.04·10−19 -3.54567 3323.56
2 3 22 -1.02605 -1.84·10−19 -3.54567 -3323.56
12 0.041042 3.10145
11 1.02605 9.04·10−19 -3.54567 -222.111
3 4 22 1.02605 -2.21·10−20 -3.54567 222.111
12 0.041042 3.10145
11 -1.02605 9.04·10−19 -3.54567 -3323.56
4 2 22 1.02605 1.60·10−19 -3.54567 3323.56
12 0.041042 3.10145

Increment no 410
11 -2 -6.70·10−18 -0.122829 41.35
1 1 22 -2 3.50·10−18 -0.122829 41.35
12 0.08 0.040115
11 2 -6.70·10−18 -0.122829 81.47
2 3 22 -2 -3.45·10−18 -0.122829 -81.47
12 0.08 0.040115
11 2 6.70·10−18 -0.122829 -41.35
3 4 22 2 -3.49·10−18 -0.122829 -41.35
12 0.08 0.040115
11 -2 6.70·10−18 -0.122829 -81.47
4 2 22 2 3.46·10−18 -0.122829 81.47
12 0.08 0.040115

92
Appendix C: ABAQUS Output from Pure Shear Example of Simple Plates

C.2 Simple Shear Example with Reinforcement

Table C.2: Output from ABAQUS for a simple shear example with reinforcement at time
increments 7, 19, 22 and 410.

Node Integration Point Direction δ ε σ P


[mm] [-] [MPa] [N]
Increment no 7
11 -0.00375 -3.44·10−23 -0.00277 -1251.02
1 1 22 -0.00375 0 -0.00277 -1251.02
12 0.00015 2.5048
11 0.00375 -3.44·10−23 -0.00277 -1253.79
2 3 22 -0.00375 0 -0.00277 -1253.79
12 0.00015 2.5048
11 0.00375 3.44·10−23 -0.00277 1251.02
3 4 22 0.00375 2.12·10−22 -0.00277 1251.02
12 0.00015 2.5048
11 -0.00375 3.44·10−23 -0.00277 -1253.79
4 2 22 0.00375 2.12·10−22 -0.00277 1253.79
12 0.00015 2.5048

Increment no 19
11 -0.52104 -5.39·10−20 -9.82075 -375.896
1 1 22 -0.52104 -9.11·10−21 -9.82075 -375.896
12 0.02084 10.5725
11 0.52104 5.39·10−20 -9.82075 10196.6
2 3 22 -0.52104 0 -9.82075 -10196.6
12 0.02084 10.5725
11 0.52104 -5.39·10−20 -9.82075 375.896
3 4 22 0.52104 6.69·10−20 -9.82075 375.896
12 0.02084 10.5725
11 -0.52104 -5.39·10−20 -9.82075 -10196.6
4 2 22 0.52104 3.52·10−20 -9.82075 10196.6
12 0.02084 10.5725

93
Appendix C: ABAQUS Output from Pure Shear Example of Simple Plates

Increment no 22
11 -1.02605 -9.04·10−19 -3.54567 222.111
1 1 22 -1.02605 1.97·10−19 -3.54567 222.111
12 0.041042 3.10145
11 1.02605 -9.04·10−19 -3.54567 3323.56
2 3 22 -1.02605 -1.84·10−19 -3.54567 -3323.56
12 0.041042 3.10145
11 1.02605 9.04·10−19 -3.54567 -222.111
3 4 22 1.02605 -2.21·10−20 -3.54567 222.111
12 0.041042 3.10145
11 -1.02605 9.04·10−19 -3.54567 -3323.56
4 2 22 1.02605 1.60·10−19 -3.54567 3323.56
12 0.041042 3.10145

Increment no 410
11 -2 -6.70·10−18 -0.122829 41.35
1 1 22 -2 3.50·10−18 -0.122829 41.35
12 0.08 0.040115
11 2 -6.70·10−18 -0.122829 81.47
2 3 22 -2 -3.45·10−18 -0.122829 -81.47
12 0.08 0.040115
11 2 6.70·10−18 -0.122829 -41.35
3 4 22 2 -3.49·10−18 -0.122829 -41.35
12 0.08 0.040115
11 -2 6.70·10−18 -0.122829 -81.47
4 2 22 2 3.46·10−18 -0.122829 81.47
12 0.08 0.040115

94
Appendix D:
Uniaxial Material Models for the
Numerical Example of a
Reinforced Slab

This appendix contains the unixial material models of the concrete used for the numerical anal-
ysis of the reinforced concrete slab described in chapter 7. For each of the given temperatures,
inputs are made in ABAQUS; the compressive behaviour illustrated in the plot is defined through
the *COMPRESSION HARDENING function and the combined concrete and interaction contribution
is given by the *TENSION STIFFENING function.

The uniaxial material models are illustrated for element sizes of both h = 129 mm and h = 73
mm.

2!
2! ((
h'.'$%/'00
h'.'$%/'00 h'.'$%/'00
h'.'$%/'00
h'.'1('00
h'.'1('00 %"#
%"# h'.'1('00
h'.'1('00
(!
(!
%%
"ctT )*+,-
"cT )*+,-

%!
%! $"#
$"#
ctT
cT

$$
$!
$!
!"#
!"#

!! !!
!! !"!$
!"!$ !"!%
!"!% !"!(
!"!( !! !"#
!"# $$ $"#
$"# %% %"#
%"#
!!cT !!cT !(
!(
cT cT
&'$!
&'$!
(a) Compression (b) Tension
2!
2! ((
Figure D.1: Compressive concrete model, (a), and the combined concrete andh'.'$%/'00
h'.'$%/'00
h'.'$%/'00 interaction stress
h'.'$%/'00
%"#
%"#
(!
(!
contribution in tension, (b),
h'.'1('00
h'.'1('00 for T = 20 ◦
C. h'.'1('00
h'.'1('00

%%
"ctT )*+,-
"cT )*+,-

%!
%! $"#
$"#
ctT
cT

$$
$!
$!
!"#
!"#

!!
!! !"!$
!"!$ !"!%
!"!% !"!(
!"!(
95 !!
!! !"#
!"# $$ $"#
$"# %% %"#
%"#
!!cT !!cT !(
!(
cT cT
&'$!
&'$!
%!
%! $"#
$"#

"ctT )
"cT )
$$
$!
$!
!"#
!"#

!! !!
!! Appendix
!"!$
!"!$ D:!"!%
!"!% Material
!"!( Models for the !!Reinforced
!"!( !"#
!"# Slab
$$ Example
$"#
$"# %% %"#
%"#
!!cT !!cT !(
!(
cT cT
&'$!
&'$!

2!
2! ((
h'.'$%/'00
h'.'$%/'00 h'.'$%/'00
h'.'$%/'00
h'.'1('00
h'.'1('00 %"#
%"# h'.'1('00
h'.'1('00
(!
(!
%%

"ctT )*+,-
"cT )*+,-

%!
%! $"#
$"#

$$
$!
$!
!"#
!"#

!! !!
!! !"!$
!"!$ !"!%
!"!% !"!(
!"!( !! !"#
!"# $$ $"#
$"# %% %"#
%"#
!!cT !!cT !(
!(
cT cT
&'$!
&'$!
(a) Compression (b) Tension

Figure D.2: Compressive concrete model, (a), and the combined concrete and interaction stress
contribution in tension, (b), for T = 100◦ C.

2!
2! ((
h'.'$%/'00
h'.'$%/'00 h'.'$%/'00
h'.'$%/'00
h'.'1('00
h'.'1('00 %"#
%"# h'.'1('00
h'.'1('00
(!
(!
%%
)*+,-
)*+,-
)*+,-
)*+,-

2!
2!
%!
%! ((
$"#
$"#
"ctT
ctT
"cT
cT

h'.'$%/'00
h'.'$%/'00 h'.'$%/'00
h'.'$%/'00
"

"

h'.'1('00
h'.'1('00 %"#
%"#
$$ h'.'1('00
h'.'1('00
(!
(!
$!
$!
%%
!"#
!"#
)*+,-
)*+,-
)*+,-
)*+,-

%!
%!
!! $"#
$"#
!!
!! !"!$
!"!$ !"!%
!"!% !"!(
!"!( !! !"#
!"# $$ $"#
$"# %% %"#
%"#
"ctT
ctT
"cT
cT
"

!!cTcT !!cTcT
"

$$ &'$!
!(
&'$!
!(
$!
$!
(a) Compression !"#
!"# (b) Tension
2!
2! ((
Figure !D.3:
! Compressive concrete model, (a), and the
h'.'$%/'00
h'.'$%/'00 !! combined concrete andh'.'$%/'00
interaction stress
h'.'$%/'00
!! !"!$
!"!$ !"!%
!"!% h'.'1('00
!"!(
!"!( %"#
%"#!◦! !"#
!"# $$ $"#
$"# h'.'1('00
%% %"#
%"#
(!
(!
contribution
!! in tension, (b),
h'.'1('00 for T = 200 C. !!
h'.'1('00
cT
cT cT
cT !(
!(
%% &'$!
&'$!
)*+,-
)*+,-
)*+,-
)*+,-

2!
2!
%!
%! ((
$"#
$"#
"ctT
ctT
"cT
cT

h'.'$%/'00
h'.'$%/'00 h'.'$%/'00
h'.'$%/'00
"

"

h'.'1('00
h'.'1('00 %"#
%"#
$$ h'.'1('00
h'.'1('00
(!
(!
$!
$!
%%
!"#
!"#
)*+,-
)*+,-
)*+,-
)*+,-

%!
%!
!! $"#
$"#
!!
!! !"!$
!"!$ !"!%
!"!% !"!(
!"!( !! !"#
!"# $$ $"#
$"# %% %"#
%"#
"ctT
ctT
"cT
cT
"

!!cTcT !!cTcT
"

$$ &'$!
!(
&'$!
!(
$!
$!
!"#
!"#

!! !!
!! !"!$
!"!$ !"!%
!"!% !"!(
!"!( !! !"#
!"# $$ $"#
$"# %% %"#
%"#
!!cTcT !!cTcT !(
!(
&'$!
&'$!
(a) Compression (b) Tension

Figure D.4: Compressive concrete model, (a), and the combined concrete and interaction stress
contribution in tension, (b), for T = 300◦ C.

96
Appendix D: Material Models for the Reinforced Slab Example

2!
2! ((
h'.'$%/'00
h'.'$%/'00 h'.'$%/'00
h'.'$%/'00
h'.'1('00
h'.'1('00 %"#
%"# h'.'1('00
h'.'1('00
(!
(!
%%

)*+,-
)*+,-
)*+,-
)*+,-

%!
%!
2!
2! $"#
$"#
((

"ctT
ctT
"cT
cT

h'.'$%/'00
h'.'$%/'00 h'.'$%/'00
h'.'$%/'00
"

"
h'.'1('00
h'.'1('00 $$
%"#
%"# h'.'1('00
h'.'1('00
$!
$!
(!
(!
!"#
!"#
%%

)*+,-
)*+,-
)*+,-
)*+,-

!!
%!
%! !!
$"#
$"#
!! !"!$
!"!$ !"!%
!"!% !"!(
!"!( !! !"#
!"# $$ $"#
$"# %% %"#
%"#

"ctT
ctT
"cT
cT

!!cTcT !!cTcT
"

"
!(
!(
$$ &'$!
&'$!
$!
$!
(a) Compression (b) Tension
!"#
!"#
2!
2! ((
Figure !D.5:
! Compressive concrete model, (a), and the
h'.'$%/'00
h'.'$%/'00 !! combined concrete andh'.'$%/'00
interaction stress
h'.'$%/'00
%"#
%"#!◦!
(!
(!
!! !"!$
!"!$
contribution
!
!"!%
!
!"!%
in !"!(
!"!(
h'.'1('00
h'.'1('00
tension, (b), for T = 400 C. !"#
!"# $$
!!
$"#
$"# %%
h'.'1('00
h'.'1('00 %"#
%"#
cT
cT cT
cT !(
!(
%% &'$!
&'$!

)*+,-
)*+,-
)*+,-
)*+,-

%!
%!
2!
2! "ctT $"#
$"#
((
ctT
"cT
cT

h'.'$%/'00
h'.'$%/'00 h'.'$%/'00
h'.'$%/'00
"

"

h'.'1('00
h'.'1('00 $$
%"#
%"# h'.'1('00
h'.'1('00
$!
$!
(!
(!
!"#
!"#
%%
)*+,-
)*+,-
)*+,-
)*+,-

!!
%!
%! !!
$"#
$"#
!! !"!$
!"!$ !"!%
!"!% !"!(
!"!( !! !"#
!"# $$ $"#
$"# %% %"#
%"#
"ctT
ctT
"cT
cT

!!cTcT !!cTcT
"

"

!(
!(
$$ &'$!
&'$!
$!
$!
!"#
!"#

!! !!
!! !"!$
!"!$ !"!%
!"!% !"!(
!"!( !! !"#
!"# $$ $"#
$"# %% %"#
%"#
!!cTcT !!cTcT !(
!(
&'$!
&'$!
(a) Compression (b) Tension

Figure D.6: Compressive concrete model, (a), and the combined concrete and interaction stress
contribution in tension, (b), for T = 500◦ C.

2!
2! ((
h'.'$%/'00
h'.'$%/'00 h'.'$%/'00
h'.'$%/'00
h'.'1('00
h'.'1('00 %"#
%"# h'.'1('00
h'.'1('00
(!
(!
%%
)*+,-
)*+,-
)*+,-
)*+,-

%!
%! $"#
$"#
"ctT
ctT
"cT
cT
"

"

$$
$!
$!
!"#
!"#

!! !!
!! !"!$
!"!$ !"!%
!"!% !"!(
!"!( !! !"#
!"# $$ $"#
$"# %% %"#
%"#
!!cTcT !!cTcT !(
!(
&'$!
&'$!
(a) Compression (b) Tension
2!
2! ((
Figure D.7: Compressive concrete model, (a), and the combined concrete andh'.'$%/'00
h'.'$%/'00
h'.'$%/'00 interaction stress
h'.'$%/'00
%"#
%"#◦
(!
(!
contribution in tension,
h'.'1('00(b),
h'.'1('00 for T = 600 C. h'.'1('00
h'.'1('00

%%
)*+,-
)*+,-
)*+,-
)*+,-

%!
%! 97 $"#
$"#
"ctT
ctT
"cT
cT
"

"

$$
$!
$!
!"#
!"#

!! !!
!! !"!$
!"!$ !"!%
!"!% !"!(
!"!( !! !"#
!"# $$ $"#
$"# %% %"#
%"#
!!cTcT !!cTcT !(
!(
&'$!
&'$!
h'.'$%/'00
h'.'$%/'00 h'.'$%/'00
h'.'$%/'00
h'.'1('00
h'.'1('00 %"#
%"# h'.'1('00
h'.'1('00
(!
(!
%%

)*+,-
)*+,-
)*+,-
)*+,-
%!
%! $"#
$"#
Appendix D: Material Models for the Reinforced Slab Example

"ctT
ctT
"cT
cT
"

"
$$
$!
$!
!"#
!"#

!! !!
!! !"!$
!"!$ !"!%
!"!% !"!(
!"!( !! !"#
!"# $$ $"#
$"# %% %"#
%"#
!!cTcT !!cTcT !(
!(
&'$!
&'$!

2!
2! ((
h'.'$%/'00
h'.'$%/'00 h'.'$%/'00
h'.'$%/'00
h'.'1('00
h'.'1('00 %"#
%"# h'.'1('00
h'.'1('00
(!
(!
%%

)*+,-
)*+,-
)*+,-
)*+,-

%!
%! $"#
$"#

"ctT
ctT
"cT
cT
"

"
$$
$!
$!
!"#
!"#

!! !!
!! !"!$
!"!$ !"!%
!"!% !"!(
!"!( !! !"#
!"# $$ $"#
$"# %% %"#
%"#
!!cTcT !!cTcT !(
!(
&'$!
&'$!
(a) Compression (b) Tension

2!
2! ((
Figure D.8: Compressive concrete model, (a), and the combined concrete and interaction stress
h'.'$%/'00
h'.'$%/'00 h'.'$%/'00
h'.'$%/'00
contribution in tension, (b), for T = 700
h'.'1('00
h'.'1('00 %"# C.
%"#

h'.'1('00
h'.'1('00
(!
(!
%%
)*+,-
)*+,-
)*+,-
)*+,-

%!
%! $"#
$"#
"ctT
ctT
"cT
cT
"

"

$$
$!
$!
!"#
!"#

!! !!
!! !"!$
!"!$ !"!%
!"!% !"!(
!"!( !! !"#
!"# $$ $"#
$"# %% %"#
%"#
!!cTcT !!cTcT !(!(
&'$!
&'$!

2!
2! ((
h'.'$%/'00
h'.'$%/'00 h'.'$%/'00
h'.'$%/'00
h'.'1('00
h'.'1('00 %"#
%"# h'.'1('00
h'.'1('00
(!
(!
%%
)*+,-
)*+,-
)*+,-
)*+,-

%!
%! $"#
$"#
"ctT
ctT
"cT
cT
"

"

$$
$!
$!
!"#
!"#

!! !!
!! !"!$
!"!$ !"!%
!"!% !"!(
!"!( !! !"#
!"# $$ $"#
$"# %% %"#
%"#
!!cTcT !!cTcT !(!(
&'$!
&'$!
(a) Compression (b) Tension

Figure D.9: Compressive concrete model, (a), and the combined concrete and interaction stress
contribution in tension, (b), for T = 715◦ C.

98

Das könnte Ihnen auch gefallen