Ayman Ababneh PHD Thesis

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/342563852
The Coupled Effect of Moisture Diffusion, Chloride Penetration, and Freezing-

Thawing on Concrete Durability
Thesis · December 2002
CITATIONS READS
15 457
1 author:
Ayman N. Ababneh
Jordan University of Science and Technology
72 PUBLICATIONS 1,142 CITATIONS
SEE PROFILE
All content following this page was uploaded by Ayman N. Ababneh on 01 July 2020.
The user has requested enhancement of the downloaded file.

THE COUPLED EFFECT OF MOISTURE DIFFUSION CHLORIDE
PENETRATION AND FREEZING-THAWING ON CONCRETE
DURABILITY
By
Ayman Nureddin Ababneh
B.Sc., Jordan University of Science and Technology, 1991
M.Sc., Jordan University of Science and Technology, 1994
Irbid – Jordan
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment of the requirement for the degree of
Doctor of Philosophy
Department of Civil, Environmental and Architectural Engineering
2002
ii
This dissertation entitled:
The Coupled Effect of Moisture Diffusion, Chloride Penetration and Freezing-Thawing

On Concrete Durability
written by Ayman Nureddin Ababneh
has been approved for the Department of
Civil, Environmental and Architectural Engineering
________________________________
Yunping Xi
________________________________
Kaspar Willam
________________________________
Dan Frangopol
________________________________
Benson Shing
________________________________
Kevin Rens
Date_______________________
The final copy of this thesis has been examined by the signatories, and we find that both
the content and the form meet acceptable presentation standards of scholarly work in the
above mentioned discipline.
HRC protocol # ___________________

iii
Ababneh, Ayman Nureddin (Ph.D., Civil, Environmental, and Architectural Engineering)
The Coupled Effect of Moisture Diffusion, Chloride Penetration and Freezing-Thawing
on Concrete Durability.
Thesis directed by Associate Professor Yunping Xi
In this thesis, both theoretical and experimental studies were conducted on some
of the processes related to concrete durability. Those processes include moisture
diffusion, chloride penetration, freezing-thawing and their coupled effects.
The first part of the study includes theoretical modeling of the coupled effect of
moisture diffusion and drying shrinkage and modeling of chloride penetration in partially
saturated concrete. This part encompasses the formulation of the governing equations and
boundary conditions, identification and modeling of the material parameters and the
numerical solution. The governing equations of moisture diffusion and chloride
penetration in concrete were characterized by the nonlinear diffusion equation derived
based on Ficks’ laws. The material parameters involved in the governing equations and
the drying shrinkage of concrete were modeled based on a multiscale method. In the
multiscale method, different theoretical models are used at different scale levels.
Diffusion mechanisms and shrinkage mechanisms are considered at the nanoscale; a
composite mechanics method (generalized self-consistent method, in particular) is used at
the microscale and mesoscale; and continuum approach is applied at the macroscale. The
numerical models were solved by the Alternating-Direction Implicit (ADI) finite
difference method. Several numerical examples were solved by the proposed models and
iv
the results obtained predicted well the basic trends and the effects of concrete material
parameters. The models can be used in the durability studies of concrete structures to
predict concrete deterioration and the onset of steel corrosion and thus the service life and
the time for rehabilitation.
The second part of this study includes experiments on the coupled effect of freezing-
thawing and chloride penetration, effect of concrete material parameters on chloride
penetration, the coupling effect of moisture diffusion and chloride penetration and the
effect of shrinkage-induced cracking on chloride penetration. The experimental study
presented the effect of water-cement ratio and aggregate volume fraction on chloride
penetration, the two-way coupling between moisture diffusion and chloride penetration
and the effect of the damage induced by drying shrinkage and freezing-thawing on
chloride penetration. The experimental results obtained in this study can be used to
validate and demonstrate the accuracy of the numerical models developed in this study
and the future studies.

v
DEDICATION
To my father, mother, wife, bothers and sisters and to my whole family who made
this work possible.

vi
ACKNOWLEDGMENT
I would like to express my deep appreciation and thanks to my advisor Professor
Yunping Xi for his support, guidance and patience during the course of my study. His
fruitful suggestions and valuable ideas have been encouraging and subsidizing this work.
Grateful acknowledgment is made to Professors Kaspar Willam, Dan Frangopol,
Benson Shing and Kevin Rens for reviewing this work and for their valuable suggestions
and comments.
I would like to extend my appreciation to my colleagues Ammar Nakhi, Zhaohui
Xie, A. Suito, Renee Cusson, Andi Asiz and our former colleague Dr. Young-Sook Roh
for their cooperation, help and encouragement. Also my sincere thanks are extended to
my friends Hassan Al-sukhni, Ahmad Ababneh, Mahmoud Shaheen, Fadi Mansour and
my sister in law Rula Abu-Dalo for their support and encouragement during my study.
I shall always be grateful to my parents, my wife and my family for their love and
encouragement without which this work has never been accomplished.
Finally the financial support under NSF grant CMS-9872379 to University of
Colorado at Boulder is gratefully acknowledged.

vii
TABLE OF CONTENTS
CHAPTER Page
1 INTRODUCTION
1.1 Durability of Concrete Structures …….………………………………1

1.2 Objectives of the Study …………………...……………………….….5
1.3 Thesis Organization …………..……………..………...………...……6
2 BACKGROUND INFORMATION AND LITERATURE REVIEW
2.1 Transport Properties of Concrete …..…………………..……………10

2.1.1 Concrete permeability ………...……..………………………12
2.1.2 Concrete diffusivity ……..…..…………..…………………..15
2.1.3 Concrete sorptivity …..…….….………………..…...……….16
2.2 Chloride Ions Penetration into Concrete ……...………….….………18
2.2.1 Mechanism of the reinforcement corrosion …….……….…..20
2.2.2 Onset of the steel corrosion ……………………………….…21
2.3 Heat Transfer in Concrete ….....…………………………………….22
3 COUPLED MOISTURE DIFUSSION AND DRYING SHRINKAGE
3.1 Introduction ……….. ….………………………………......……..… 25

3.2 Modeling Moisture Diffusion in Concrete ……………………....…..27
3.2.1 Basic formulation of the governing equation ….…………….27
3.2.2 Initial and boundary conditions …………………………..…29
3.2.3 Material parameters ………………………………………....30
3.3 Numerical Solution …………...…………………….…….................32
3.3.1 Finite element method …………………...……………….... 32
3.3.2 Finite difference method …...…………….…...…………......35
3.4 Multiscale Modeling of Drying Shrinkage ….……………....……...37
3.5 The Interaction of Drying Shrinkage and Moisture Diffusion.............44
3.5.1 Effect of shrinkage-induced damage on diffusion coefficient.45
3.5.2 Effect of drying shrinkage on moisture capacity……...…......49
3.6 Numerical Applications ..…………………………………................53
viii
3.6.1 Effect of damage ………………………………………….. 61

3.6.2 Effect of water-cement ratio ……………………………… 63
3.6.3 Effect of aggregate volume fraction ………………………. 65
3.7 Conclusion ………………………………………………………… 67
4 COUPLED MOISTURE AND CHLORIDE DIFFUSIONS
4.1 Introduction ………..……………………………………….……… 69

4.2 Basic Formulation ……………………………………………….… 71
4.3 Material Parameters ……………………………………………..… 74
4.3.1 Chloride binding capacity ……….………………………… 75
4.3.2 Chloride diffusion coefficient…………………………........ 80
4.4 Numerical Solution ………………………………………...……… 83
4.5 Numerical Applications and Discussions ………………….……… 85
4.5.1 Numerical example of chloride penetration ….…………… 85
4.5.2 Comparison with test data………… ..………………...…… 88
4.5.3 Effect of the moisture diffusion on chloride penetration…... 88
4.5.4 The Effect of the concrete mix parameters………………….91
4.6 Conclusions ………………………………………...……………… 96
5 THE COUPLED EFFECT OF FREEZING-THAWING AND CHLORIDE

PENETRATION IN CONCRETE
5.1 Introduction ………..……………………………………….……… 98

5.2 Ultrasonic Testing of Materials……………………..…….……….101
5.3 Experimental program …………………………………...………...104
5.3.1 Preparation of test specimens……………………………....104
5.3.2 Cyclic freezing-thawing and chloride penetration………....105
5.3.3 Measurement of Transit time and resonant frequency .…....106
5.3.4 Chemical analysis for total chloride concentration………...107
5.4 Test results and Discussion………………………………………...108
5.4.1 Cracking patterns and visual inspection……………………108
5.4.2 Ultrasonic test results……………………………………….111
5.4.3 Chloride penetration test results…………………………….113
5.5 Conclusion......……………………………………….……………..118
ix
6 EXPERIMENTAL STUDY ON THE EFFECT OF CONCRETE MATERIAL

PARAMETERS ON CHLORIDE PENETRATION
6.1 Introduction ………..……………………………………….……...120

6.2 Specimen Preparation and Experimental Parameters.…….………. 121
6.3 Test Procedure……………………………………………………...123
6.4 Test Results and Discussion……….…………………… ………... 124
6.4.1 Chloride penetration profile ………………………...……...124
6.4.2 Effect of water-cement ratio ………………………...……..124
6.4.3 Effect of aggregate volume fraction………………………..129
6.5 Conclusion………………………………………………………….134
7 EXPERIMENTAL STUDY ON THE COUPLING EFFECT OF MOISTURE

DIFFUSION AND CHLORIDE PENETRATION IN CONCRETE
7.1 Introduction…………………………………………………………135
7.2 Specimens Preparation……………………………………………...136
7.3 Effect of Moisture Diffusion on Chloride Penetration ……...……..137
7.3.1 Experimental procedure…………………………………….137
7.3.2 Experimental results and discussion………………………..139
7.4 The Effect of Chloride Penetration on Moisture Diffusion………...144
7.4.1 Experimental setup and procedure………………........…….144
7.4.2 Results and discussion……………………………………...146
7.5 Conclusions…………………………………………………………149
8 EXPERIMENTAL STUDY ON THE EFFECT OF SHRINKAGE-INDUCED

DAMAGE ON CHLORIDE PENETRATION
8.1 Introduction…………………………………………………………151
8.2 Experimental Program……………………………………………...153
8.2.1 Material……………………………………………………..153
8.2.2 Test specimens preparation………………………………....154
8.2.3 Experimental procedure…………………………………….155
8.3 Experimental results and discussion………………………….…….157
x
8.3.1 Shrinkage strain………………………………………….....157

8.3.2 Restrained shrinkage cracking………………………….…..159
8.3.3 Chloride profiles……………………………………….…...161
8.4 Conclusions…………………………………………………….…..163
9 CONCLUSIONS AND FUTURE RESEARCH RECOMMENDATIONS
9.1 Conclusion…………………………………………………….……165
9.2 Future Research Recommendations………………………….…….167
BIBLIOGRAPHY ………………..………………………………......…..….. 169
APPENDIX A: ADSORPTION ISOTHERMS AND MOISTURE CAPACITIES OF

CEMENT PASTE AND AGGREGATE…......………….178
APPENDIX B: VOLUME PERCENTAGES OF DIFFERENT HYDRATION

PRODUCTS.............................................................…………181
APPENDIX C: MIX DESIGN FOR THE EXPERIMENTAL

PROGRAM………………......................…….…………..…..185
APPENDIX D: COMPRESSION TEST RESULTS ……….......…………..…193
APPENDIX E: EXPERIMENTAL PROCEDURE FOR OBTAINING CHLORIDE

PROFILES..........….…………………………....195
xi
LIST OF FIGURES
Figure Page
1.1 Bridge deck…..………………………………………………………...........4
1.2 Crack on the bridge deck……….. ……………….............………………....4
1.3 Cracks on the shoulder (splash zone) of the bridge deck.………….………..5
2.1 A simple setup for measuring permeability of concrete…………………...13
2.2 Setup for measuring sorptivity of concrete………………………………...18
2.3 Concrete deterioration due to reinforcement corrosion……………………19
2.4 Mechanism of induced-chloride reinforcement corrosion…………………21
3.1 Finite difference mesh of concrete slab……………………………………36
3.2 Different scale levels of concrete…………………………………………..39
3.3 Composite action of concrete………………………………………………40
3.4 Microstructral features of cement paste……………………………............40
3.5 Three-phase effective media model…………………………………..........41
3.6 Four-phase composite model ……………………………….……………..43
3.7 Plane stress problem……………………………………………………….46
3.8 Tensile stress strain diagram ………………………………………………54
3.9 Variation of relative humidity with the depth ……………………………..56

xii
3.10 Variation of shrinkage strain with the depth ………………………….…...56
3.11 Variation of scalar damage parameter with the depth …………….………57
3.12 Variation of hygral stresses with the depth ………………………………..57
3.13 Variation of relative humidity with time of exposure ……….…………….59
3.14 Variation of shrinkage strain with time of exposure ………………………59
3.15 Variation of scalar damage parmater with time of exposure ……………...60
3.16 Variation of hygral stresses with time of exposure ………………………..60
3.17 Effect damage on drying of concrete…………………………….………...62
3.18 Effect of damage on shrinkage of concrete…………………………….......63
3.19 Effect of w/c on drying of concrete……………………………….……....64
3.20 Effect of w/c on shrinkage of concrete……………………………….........64
3.21 Effect of aggregate volume fraction on drying of concrete…………….....66
3.22 Effect of aggregate volume fraction on shrinkage of concrete……………66
4.1 Concrete slab used in the numerical solution……………………………...84
4.2 Depth profile of free chloride concentration………………………….........86
4.3 Time profile of free chloride concentration……………….……………….87
4.4 Comparison with the experimental results…………………………………89
4.5 Effect of the moisture diffusion on free chloride penetration……...………90
4.6 Moisture diffusion in non-saturated concrete……………………………...91

xiii
4.7 Effect of water-cement ratio on chloride penetration ……………………..92
4.8 Effect of water cement ratio on the chloride diffusion coefficient….……..93
4.9 Effect of water-cement ratio on humidity diffusion coefficient…………...93
4.10 Effect of aggregate volume fraction on the chloride penetration ……........94
4.11 Effect of aggregate volume fraction on the chloride diffusion coefficient...95
4.12 Effect of aggregate volume fraction on humidity diffusion coefficient…...95
5.1 Specimen for coupled freezing-thawing and chloride penetration….........104
5.2 Placing of concrete samples in the environmental chamber………….......105
5.3 The freezing and thawing temperature cycle………………….……….....106
5.4 Transit time measurement setup (V-meter)……………….………...........107
5.5 Setup for measurement of fundamental resonant frequency (E-meter)…..107
5.6 The concrete sample (46) after it was exposed to the freezing-thawing….109
5.7 The concrete sample (56) after it was exposed to the freezing-thawing.…109
5.8 The concrete sample (66) after it was exposed to the freezing-thawing.....110
5.9 The concrete sample (55) after it was exposed to the freezing-thawing… 111
5.10 The concrete sample (57) after it was exposed to the freezing-thawing….111
5.11 Chloride profiles for the concrete specimens made of mix 46…………...116
xiv
6.1 Concrete sample…………………………………………………………..122
6.2 Ponding with chloride solution…………………………………………...123
6.3 Chloride profiles for different (w/c) after 50 days exposure time………..126
6.4 Chloride profiles for different (w/c) after 100 days exposure time….…...127
6.5 Chloride profiles for different (w/c) after 200 days exposure time………128
6.6 Segregation of aggregate………………………………………………….128
6.7 Chloride profiles for different (gi) after 50 days of exposure……………131
6.8 Chloride profiles for different (gi) after 100 days of exposure…………..132
6.9 Chloride profiles for different (gi) after 200 days of exposure…………..133
7.1 Sample for coupled moisture diffusion and chloride penetration………...137
7.2 Setup for concrete humidity measurement…………………….…..……..138
7.3 Immersion bath of the concrete samples……………...…………….…….139
7.4 Chloride profiles after 30 days of exposure………...………....……….…142
7.5 Chloride profiles after 90 days of exposure …………………….…....…. 143
7.6 Chloride profiles after 180 days of exposure…………….…..………….. 144
7.7 Experimental setup for the effect of chloride diffusion on moisture

diffusion……………………………………………………………….….146
xv
7.8 The weight of accumulated solution in the center hole for the three different
solutions………………………………………………………………..…148
8.1 Restrained and free shrinkage concrete specimens…………………….…154
8.2 Stresses in the restrained shrinkage concrete sample………………....… 155
8.3 Data acquisition system and personal computer……………………….…156
8.4 Ponding with sodium chloride solution……………………………….….157
8.5 Shrinkage strains development in restrained and free shrinkage concrete.159
8.6 One of the microcracks in the restrained shrinkage ring……….………...160
8.7 Microcrack on the surface of the restrained ring………………………....160
8.8 Microcracking of the surface of the restrainedring…………………….....161
8.9 Chloride profiles after 50 days of exposure…...……………………….…163
E.1 Drilling of concrete sample………………………………………………195
E.2 Chloride concentration measurement kit…………………………..……..196
E.3 Sampling of 1.5 gram of concrete dust and the extraction liquid vial…....197
E.4 Calibration curve for obtaining chloride concentrations………………....199
E.5 Calibration curves for obtaining chloride profiles………………………..203
E.6 Calibration curves for obtaining chloride concentrations………….……..206
E.7 Calibration curve for obtaining chloride concentrations………………....207

xvi
LIST OF TABLES
Table Page
5.1 Ultrasonic test results………………………………..……………………113
5.2 Total chloride concentration as percentage of weight of concrete……….114
5.3 Summary of the experimental results…………………………………….118
6.1 Chloride concentration for different (w/c) after 50 days exposure time….125
6.2 Chloride concentration for different (w/c) after 100 days exposure time...126
6.3 Chloride concentration for different (w/c) after 200 days exposure time...127
6.4 Chloride concentration for different (gi) after 50 days exposure time…....131
6.5 Chloride concentration for different (gi) after 100 days exposure time......132
6.6 Chloride concentration for different (gi) after 200 days exposure time…..133
7.1 Total chloride concentration after 30 days of exposure…………………..141
7.3 Total chloride concentration after 180 days of exposure………………....143
7.4 Weight of the accumulated solution……………………………………...147
8.1 Measured shrinkage strains (in microstrains)……………………………158

xvii
B.1 Parameters related to the hydration of different compounds of cement…183
C.1 Concrete mix design proportions………………………………………....190
C.2 Mix proportions used in the experimental study of Chapters 5, 6 and 7…192
C.3 Concrete mix proportions used in experimental study of the effect of moisture
diffusion on chloride penetration………………..…..………….192
C.4 Proportions of the concrete mix used in the experiment of the effect of induced-
shrinkage cracking on chloride penetration….………………….192
D.1 Compression test results for the concrete mixes used in the experimental study
of Chapters 5, 6 and 7…………………………..……………..…....193
D.2 Compression test results for the concrete used in the experiment of the effect of
moisture diffusion on chloride penetration …………………...……….194
D.3 Compression test results for the concrete used in the experiment of the effect
induced-shrinkage cracking on chloride penetration…………..................194
E.1 The mV readings after 400 freezing-thawing cycles………………......…198
E.2 The mV readings for the standard solutions………………………….…..198
E.3 The mV readings after 50 days of exposure……………………….……..200
E.4 The mV readings after 100 days of exposure……………………….……201
E.5 The mV readings after 200 days of exposure……………………….……202
E.6 The mV readings for the standard solutions………………….…….…….203
E.7 The mV readings after 30 days of exposure……………………….……..204

xviii
E.8 The mV readings after 90 days of exposure…………………….………..204
E.9 The mV readings after 180 days of exposure………………….…………205
E.10 The mV readings for the standard solutions…………………….………..205
E.11 The mV readings after 50 days of exposure…………………….………..206
E.12 The mV readings for the standard solutions…………………….………..207

CHAPTER 1
INTRODUCTION
1.1 Durability of Concrete Structures
Durability of concrete structures can be defined as its ability to maintain a
certain strength and serviceability requirements during its designed service life. Thus,
durability design of concrete shall concern with enhancing its capability to resist
excessively applied mechanical and environmental loads, without significant
damages. Lack of durability in terms of concrete deterioration and reinforcing steel
bars corrosion can cause cracking, spalling, delamination and loss of strength of the
structural members affecting its integrity and ultimately leading to major cracking or
failure. As a result, the durability of concrete has received an increasing attention in
recent years and the developments in cement and concrete technologies have focused
on producing not only stronger but also more durable concrete.
Concrete deteriorations can be classified based on the factors attributing to the
deteriorations, such as physical, chemical, biological and reinforcement corrosion.
Some of the major deterioration mechanisms of concrete will be introduced briefly in
this chapter. Also, some common features of the deterioration mechanisms will be
identified.
Physical deteriorations can be attributed to physical actions such as, wetting

2
and drying cycles, freezing and thawing, temperature changes and other physical
actions such as waves, movement of solid particles in water or air, wear and tear
actions. Wetting and drying cycles may result in cracking due to excessive shrinkage.
Freezing and thawing can produce internal stresses inside concrete causing spalling
and cracking. Physical actions such as waves, movement of solid particles in water or
air and/or wear and tear can cause abrasion of concrete. Temperature changes can
produce thermal stresses causing damage and loss of strength of concrete members.
Chemical deteriorations can be attributed to chemical attacks from the
environment such as acid attack and sulfate attack or chemical reactions between the
constituents of the concrete such as alkali silica reaction and alkali carbonate reaction.
Acids can attack concrete by reacting with calcium silicate hydrate (C-S-H) gel
producing water-soluble calcium compounds, which may leach out causing surface
efflorescence. Acids also can attack reinforcing steel bars causing corrosion, which
may ultimately lead to cracking, spalling and delamination of concrete. Sulfates can
attack concrete reacting with calcium hydroxide (CH) to form gypsum. Gypsum
reacts with hydrated calcium aluminates to form ettringite. Both reactions cause
volume expansion and produce stresses causing damage and cracking of concrete.
Alkali silica reaction between the active silica in aggregates and alkalis in cement
results in the formation of alkali-silica complex, which absorbs water causing volume
expansion and producing internal stresses which may lead to map or pattern cracks.
Alkali-carbonate reaction occurs between some dolomitic limestone aggregates and
alkalis in cement. The products of the reaction have less volume than the original
materials and susceptible to swelling and volume changes. Those volume changes
3
result in internal stresses and deterioration of concrete.
Biological deterioration (Kawai, 1996) can be attributed to aerobic and
anaerobic sulfur bacteria and fungi, which metabolize organic and carbonic acids.
Those acids react with hydration products of cement causing leaching and volume
changes of concrete.
Reinforcement corrosion can be attributed to chloride attack or carbonation. In
chloride-induced corrosion, chloride ions breakdown the passive oxide film on the
steel bars and allow the corrosion process to start. In carbonation-induced corrosion,
carbonation process lowers the value of the pH by forming carbonic acid and induces
the steel corrosion.
One example on the lack of durability of concrete structures is the
deterioration of bridge deck shown in Fig. (1.1). The repeated overloading of the
bridge deck in terms of cyclic mechanical loading, wetting-drying, freezing-thawing
and chloride ions ingress from deicing salts causes fatigue cracks as shown in Figs.
(1.2) and (1.3), which result in two types of damage. The first one, which occurs in
most cases, is the reduction of the durability of concrete in a way that permeability of
the concrete is drastically increased. There has not been much research done in this
area. The other one is the immediate failure of the structure, after certain cycles of
loading, due to unstable propagation of the major fatigue crack. In the latter case,
fatigue fracture has been studied extensively by experiments (Nordby 1958, Hilsdorf
et al. 1966, Baluch et al. 1989).

4
Fig. 1.1: bridge deck (part of highway US-285 over North Turkey
Creek Road) near Evergreen, Colorado.
Fig. 1.2: Crack on the bridge deck.

5
Fig. 1.3: Cracks on the shoulder (splash zone) of the bridge deck.
As one can see, diffusion processes such as, temperature, moisture and
chemicals control all physical and chemical deteriorations of concrete. The common
feature of these concrete deteriorations is that they are all diffusion-controlled
processes. Therefore, accurate characterization of the diffusion processes becomes
very important.
1.2 Objectives of the Study
The main objective of this study is to quantify some of the different processes
that affect the durability of concrete structures. These processes include moisture
diffusion, chloride penetration, freezing-thawing and their coupling effects. The study
includes theoretical modeling and experimental study. The theoretical modeling

6
encompasses the formulation of the governing equations and the boundary conditions,
identification of the material parameters involved, solution of the governing equations
using one of the numerical techniques such as finite element method or finite
difference method, application of the numerical solution in a computer program and
performing a parametric study to analyze the effects of different influential
parameters. The experimental study include the effect of the concrete material
parameters on chloride penetration, the coupling effect of moisture diffusion and
drying shrinkage, the coupling effect of moisture diffusion and chloride penetration
and the coupling effect of freezing-thawing and chloride penetration in concrete.
1.3 Thesis Organization
This thesis has nine chapters. This chapter includes introductory information
about the durability of concrete structures, motivation and the objectives of the study.
Chapter 2 presents detailed background information about some issues
related to the durability of concrete structures such as its diffusion properties, chloride
ions penetration and heat conduction.
Chapter 3 discusses the interaction effect between the drying shrinkage and
the moisture diffusion process in concrete. This chapter contains seven sections. The
first section presents introductory information about the two processes, drying
shrinkage and moisture diffusion in concrete and their interaction. The second section
discusses various issues of the moisture diffusion in concrete such as the basic
formulation of the governing partial differential equation, the material parameters

7
involved in the governing equation. The third section discusses the finite element and
the finite difference numerical solutions. The fourth section includes the modeling of
drying shrinkage of concrete. The fifth section discusses the interaction of drying and
shrinkage in concrete. The sixth section presents some numerical applications on the
different influential parameters and the conclusions are in the last section.
Chapter 4 discusses the coupling effect of moisture and chloride diffusion
processes in concrete. This chapter contains six sections. The first section is an
introductory about the interaction effect of the two processes. The second section
presents the basic formulation of the governing partial differential equation. The third
section presents the material parameters involved in the governing equation. The
fourth section presents the finite difference numerical solutions. The fifth section
discusses some numerical applications on the different influential parameters and
section six presents the conclusions.
Chapter 5 discusses the coupling effect of freezing-thawing and chloride
penetration in concrete. This chapter contains five sections. The first section
introduces the coupling effect of the two processes. The second section presents a
brief review about the ultrasonic testing of materials. The third section presents the
experimental program, which includes the preparation of test specimens and the
description of the testing methods and techniques. The fourth section present the
experimental results, which include the visual inspection of the cracking of the test
specimens, the damage factors resulted from the ultrasonic testing and the chloride
profiles resulted from the electrochemical analysis. The fifth section presents
8
summary of the experimental results and conclusions.
Chapter 6 presents an experimental study on the effect of concrete material
parameters on chloride penetration. This chapter contains five sections. The first
section presents introductory information about the effect of different material
parameters on chloride penetration in concrete. The second section describes the
specimen and parameters of the experimental study. The third section describes the
test procedure. The fourth section presents the experimental results for the effect of
water-cement ratio and aggregate volume fraction and the fifth section contains the
conclusions.
Chapter 7 describes an experimental study on the coupled effect of moisture
diffusion and chloride penetration in concrete. This chapter contains five sections.
The first section introduces some information about the coupling effect of the two
processes. The second section describes the concrete specimen used in this study. The
experimental program, experimental results and discussions are presented in section
three for the effect of moisture diffusion on chloride penetration. Section four is for
the effect of chloride penetration on moisture diffusion, while section five contains
the conclusions.
Chapter 8 describes an experimental study of the effect of shrinkage-induced
damage on chloride penetration in concrete. This chapter contains four sections. The
first section contains an introductory information and literature review on the effect of
cracking on chloride penetration in concrete. The second section describes the
materials, test specimens and experimental procedures for measurement of shrinkage

9
strains and the chloride profiles. The third section contains the test results. The
conclusions are presented in section four.
Chapter 9 contains conclusions and recommendations for future research.

CHAPTER 2
BACKGROUND INFORMATION AND LITERATURE REVIEW
2.1 Transport Properties of Concrete
Durability of concrete structures is mainly affected by the transport of fluids
and some deleterious chemicals into or through concrete. There are many fluids (in a
broad sense) that can flow in concrete and affect its durability. Those fluids include
gases such as carbon dioxide (CO2) and oxygen (O2) and liquids such as pure water or
water carrying some deleterious chemicals. The deleterious chemicals include acids
and aggressive ions such as sulfates and chlorides. Freezing and thawing is mainly
affected by the diffusion of moisture in concrete. Corrosion of embedded steel bars is
affected collectively by gas diffusion (CO2 and O2), moisture diffusion and chloride
ions penetration in concrete. Chemical attacks such as acids; sulfate and chloride ions
are mainly affected by the diffusion of these chemicals in concrete, as well as the
subsequent chemical reactions between the chemicals and some components in
concrete.
Transport properties of concrete depend primarily on its pore structure, which
can be characterized by its porosity, pore size distribution, and morphology of the
pore structure (such as pore shape, pore connectivity and pore tortuosity). High
porosity does not always correspond to high permeability unless the pores are
11
continuous and effective with respect to flow. Discontinuous pores, long and tortuous
pore systems, pores contain adsorbed water and pores have narrow entrances are
ineffective with respect to the flow of fluids. Concrete is a particular composite
material composed of aggregates and cement paste. Aggregates are porous materials,
but their contribution to the diffusion of fluids is minor because of the discontinuity in
their pore structure and also, they are enveloped by the cement paste. Hydrated
cement is composed of a very poorly crystalline material called calcium silicate
hydrate (C-S-H) gel, unhydrated cement particles, calcium hydroxide (CH) crystals,
ettringite crystals and pores. Those pores include air voids and their size ranges
100µm−10mm, water-filled spaces called capillary pores and their size ranges
0.05−100µm, and gel pores and their size is smaller than 50 nm. Transport of fluids in
capillary pores is much easier than in gel pores. As a result, the permeability of
hardened cement paste is mainly controlled by its capillary porosity.
Transport of mass in concrete, differing from fracture failure and strength
failure of a concrete structure, involves not only conventional mechanics methods but
also material theories at different scale levels. For instance, the strength of concrete
can be correlated to the size of defects as small as 50 micrometers. But, the fine
pores with smaller sizes have almost no effect on strength of the materials. Mass
transport, on the other hand, is controlled by the internal pores with a very broad
range of size distribution from millimeter down to nanometer levels. Therefore, the
prediction models for mass transport must be developed based on the influential
parameters and mechanisms on multiple scale levels.

12
Transport of fluids and deleterious chemicals into or through concrete is
governed by several mechanisms and each mechanism is related to specific transport
property. Fluids and deleterious chemicals can flow into or through concrete under
pressure gradient, concentration gradient, capillary suction and potential gradient and
the relevant diffusion properties are the permeability, diffusivity, sorptivity and ionic
diffusivity.
2.1.1 Concrete permeability
Permeability of concrete refers to the ease of flow of fluids into or through
concrete under pressure gradient. In this work we concern about the moisture
diffusion in concrete and so water permeability of concrete will be discussed.
The flow of water through saturated concrete and under drop of a hydraulic
head (∆h) is governed by Darcy’s law as follows:
dq 1 ∆h
=K (2.1)
dt A L
where: dq/dt = flow rate of water in m3/sec,
A = cross-sectional area of flow in m2,
K = coefficient of permeability of concrete in m/sec,
∆h = drop in hydraulic head in m,
L = depth of flow in m.
13
The coefficient of permeability of concrete can be measured using the simple
setup shown in Fig. (2.1), in which water is allowed to flow through saturated
concrete under pressure gradient.
Inflow
Sealant
L Concrete
Sample
Outflow
Fig. 2.1: A simple setup for measuring permeability of concrete.
The intrinsic permeability of concrete, which is independent of the type of the
liquid and temperature, can be calculated using the following formula:
Q ηL
K′ = (2.2)
A (P1 − P2 )
where, K ′ = intrinsic permeability of concrete in m2,
Q = measured flow rate of water in m3/sec,
A = cross-sectional area of the concrete sample in m2,
η = dynamic viscosity of water in N . sec/m2,

14
L = depth of concrete sample in m,
P1 = upstream pressure in N/m2,
P2 = downstream pressure in N/m2.
Then the coefficient of permeability of concrete can be calculated using the
following formula:
ρg
K = K′ (2.3)
η
where, K = coefficient of permeability in m/sec,
ρ = water density in kg/m3,
g = gravitational acceleration in m/sec2,
η = dynamic viscosity of water in N . sec/m2,
The permeability of concrete is mainly affected by water-cement ratio, age,
temperature, curing condition, curing period, cement properties, cement composition
and the volume fraction of aggregates. The water-cement ratio affects the
permeability of concrete by introducing more capillary pores. As the water-cement
ratio increases, the capillary porosity increases resulting in an increase in the
permeability of concrete. The age of concrete affects its permeability due to the
progress of the hydration process. As the hydration process continues more calcium
silicate hydrate (C-S-H) gel is produced which fill the available capillary pores
15
resulting in a rapid decrease of the permeability of concrete (Neville 1996). The
temperature affects the permeability of concrete by changing the dynamic viscosity of
the flowing fluid. Also temperature changes produce volume changes in the concrete
body, when restrained by either internal or external restraints they produce stresses
and micro-crack the concrete increasing its permeability. Moist curing of concrete
will reduce its permeability by increasing the rate of cement hydration process and
decreasing the capillary porosity. Curing of concrete in dry condition will result in
moisture losses. Those moisture losses will produce internal strains in the concrete
body and if restrained, stresses will develop, causing cracking and thus increasing the
permeability of concrete. Cement properties such as degree of fineness can affect the
permeability of concrete by decreasing its porosity and thus decreasing its
permeability. Cement composition affects the permeability of concrete by increasing
the rate of cement hydration. And as a result it will not affect the ultimate
permeability of concrete. The volume fraction of aggregates in concrete affects its
permeability by changing the volume of the cement paste, which is the effective
media for the flow of fluids. Also aggregates block the flow paths in concrete making
them longer and more tortuous. And as a result an increase in the volume fraction of
aggregates will result in a decrease in the permeability of concrete.
2.1.2 Concrete diffusivity
Diffusivity of concrete is related to the diffusion of mass into or through
concrete under concentration gradient. In this work we concern about the moisture
and chloride ion diffusion processes in concrete and so moisture and chloride ion
16
diffusivities of concrete will be discussed. The diffusion of moisture and chloride ions
into or through concrete under concentration gradient is governed by Fick’s first law
as follows:
J = − D ∇C (2.4)
where: J = mass diffusion rate in kg/m2,
D = diffusivity of concrete in m2/sec,
∇C = concentration gradient in kg/m4.
The moisture diffusivity of concrete is affected by many factors such as pore
structures, humidity level and temperature. The chloride diffusivity of concrete
affected by many factors such as features of pore structures, level of chloride
concentration, humidity level and temperature. The pore structure of concrete is
affected by many factors such as water-cement ratio, age, temperature, curing
condition and period, cement properties and cement composition and the volume
fraction of aggregates.
2.1.3 Concrete sorptivity
Sorptivity of concrete is related to the diffusion of fluids into or through
concrete due to capillary forces in the pores. Sorption of water into concrete can be
expressed by the following formula:
M (t ) = M o + ρAS t (2.5)
17
where: M(t) = mass of fluid gained by the concrete at time t,
M0 = initial mass gain in kg,
ρ = water density in kg/m3,
A = cross-sectional area of sorption in m2,
S = sorptivity of concrete in m/sec0.5,
t = time in seconds.
Sorptivity of concrete can be estimated using the setup shown in Fig. (2.2). In
which a concrete specimen is exposed to water at one of the faces while the other
faces are sealed. The gain in the mass of concrete sample and the time of exposure are
reported. Also, the initial gain in the mass of concrete sample should be reported. The
sorptivity of concrete can be estimated using Eq. (2.5) and knowing the cross-
sectional area and the density of the water. It is important to make sure that the
concrete sample has uniform moisture content before exposing to water. This can be
done either by keeping the sample in oven for 7 days at 50 oC followed by another 7
days in sealed container which will produce a uniform moisture content for 100 mm
sample or by oven drying the sample at 105 oC until there is less than 0.1% change in
the mass after 24 hours of drying.

18
Sealed
Surfaces
Exposed Concrete
Surface Sample
Water
Fig. 2.2: Setup for measuring sorptivity of concrete.
2.2 Chloride Ions Penetration into Concrete
Chloride ions penetration is one of the major problems that affect the
durability of reinforced concrete structures. Chloride solution causes corrosion of
embedded steel bars, which in turn causes cracking, spalling and delimination of
concrete as shown in Fig. (2.3). It may also reduce the load carrying capacity of the
structural members by reducing the cross-sectional area of the steel bars.
There are three stages in the deterioration process of reinforced concrete
structures due to steel bars corrosion. The first stage is dominated by the penetration
of chloride ions through concrete towards the reinforcing bars. Once the chloride
concentration reaches a critical value, corrosion of steel starts. The second stage is
controlled by rust expansion, during which the rust forms in the interface transition
zone between aggregate and cement paste. The formation of rust is associated with
large volume expansion.

19
Steel
Bars
Steel
Bar
(a) (b)
Steel Bars
(c)
Fig. 2.3: Concrete deterioration due to reinforcement corrosion (a) cracking
(b) spalling and (c) delimination.
The third stage is concrete cracking. The volume expansion of the rust causes
the volumetric mismatch, which leads to cracking, spalling and delamination of
concrete. Among the three stages, the first one may be the longest one, which could
last for 15 to 20 years. The second stage is shorter than the first one, depending on the
porosity in the interface transition zone. The third stage is the shortest one among the
three stages, since the typical thickness of concrete cover is 5 cm, and crack
propagate through the concrete cover very fast once the crack is formed around the
steel bars.
20
2.2.1 Mechanism of the reinforcement corrosion
During the hydration process of the cement a thin layer of oxide forms on the
surface of the embedded steel bars called passivity layer. This layer protects the steel
from reacting with oxygen and moisture and thus prevents the rust formation and
corrosion of steel bars. Once the chloride ions concentration reaches a certain level, it
destroys the thin passive film around the reinforcing steel bar and the electrochemical
cell will setup as shown in Fig. (2.4). At the anode the oxidation of iron will occur,
and the ferrous ions (Fe++) will produce:
Fe ⇒ Fe + + + 2e − (2.6)
The ferrous ions will pass into the pore solution and the electrons will flow
through the steel bar to the cathode. At the cathode the electrons will react with
moisture and oxygen to form the hydroxyl ions (OH-).
2 H 2 O + O2 + 4e − ⇒ 4OH − (2.7)
Then the hydroxyl ions will travel through the pore solution and combine with
the ferrous ions to form ferrous hydroxide (Fe(OH)2 ):
Fe + + + 2OH − ⇒ Fe(OH ) 2 (2.8)
Further oxidation of ferrous hydroxide will result into the ferric hydroxide,
which is the rust:
4 Fe (OH ) 2 + 2 H 2O + O2 ⇒ 4 Fe (OH ) 3 (2.9)

21
Chloride Moisture Oxygen
Concrete
Cover
Cl- Ions -
Fe++ + 2(OH) Fe(OH)
2
Fe(OH)3 (Rust)
- Steel Bar
Fe++ OH
Anode Electrons ( e- ) Cathode
- -
Fe Fe++ + 2e 2H2O + O2 + 4e- 4(OH)
Passivity
Layer
Fig. 2.4: Mechanism of induced-chloride reinforcement corrosion.
2.2.2 Onset of the steel corrosion
The chloride concentration at which, corrosion of steel in concrete starts, is
called critical chloride concentration. Research results on the critical value of chloride
content showed some variations and also, they are not in the same unit. We first
convert different critical values from literature into a unified unit, i.e. gram of
chloride per gram of concrete, (g/g), then select a specific critical value to be used in
this study. This unit is considered to be convenient because the test results of
chloride content by AASHTO T259 and AASHTO T260 are in this unit.
Berke (1986) obtained the value of 0.9 - 1.0 kg of chloride per cubic meter of
concrete. Using 2300 kg/m3 as the density of concrete, the critical content in gram of
chloride per gram of concrete will be 0.039% to 0.043% (g/g). Browne (1982)
showed the value of 0.4% by weight of cement used in the concrete. Assuming 300
22
Kg of cement for a cubic meter of concrete, 0.4% of 300 Kg cement is 1.2 Kg, and
then using 2400 Kg/m3 as the density of concrete, the critical value will be 0.05%
(g/g). Cady and Weyers (1992) gave the value of 0.025 - 0.05% by weight of
concrete. Therefore, the range of the critical content is from 0.025 – 0.05% in gram
of chloride per gram of concrete, which is a quite broad range. Coggins and French
(1990) determined the chloride profiles in three girders (two facial and one interior)
and the deck of a twenty-year-old prestressed concrete bridge. The bridge deck and
the girders were exposed to chloride penetration during its service life. Dust samples
were collected at ½ in. (13 mm) depth increments and tested for total chloride
concentration. It was found that total chloride concentration at reinforcement depth
does not exceed 120 ppm (by concrete weight) in girders and in the range of 400 to
1500 ppm in bridge deck. Although, the chloride concentration in bridge deck
exceeds the corrosion threshold value (200-450 ppm), but no evidence of corrosion
was reported.
2.3 Heat Transfer in Concrete
In practice, concrete is exposed to temperature variations during mixing,
placing, hardening and its service life. Temperature variations can be attributed
mainly to the heat transferred into or from concrete by surface convection and
radiation and in concrete by conduction, or to the heat produced during cement
hydration. Those temperature variations affect the properties of both fresh and
hardened concrete and thus affect its durability performance.
Reasonable rise in the curing temperature would speeds up the rate of cement
23
hydration and subsequently increases the early strength of concrete without harmful
effect on its later strength. On the other hand, higher temperature variations during
placing and hardening of concrete would result in poorly structured cement paste with
higher porosity. The poor structure of the hydrated cement paste is due to the non-
uniform distribution of the hydration products within the cement paste. This non-
uniform distribution of hydration products would reduce locally the strength of
concrete. The higher porosity and the microcracking due to the lower strength would
accelerate the transport of moisture and some deleterious chemicals, which affect
adversely the durability and reduce the service life of concrete structures.
Heat developed during hydration of cement may have harmful effects on the
durability of concrete structures. The heat evolved during hydration of cement and the
loss of heat at the surface may results in steep temperature gradients between the
interior and the exterior of concrete resulting in cracking. On the other hand, the heat
produced during hydration of cement may prevent the freezing of pore water resulting
in continuous hydration of cement and thus more gain in the strength of concrete and
reduction in the porosity will be achieved.
High temperature rise in the surface layer of the hardened concrete due to fire
or some industrial applications such as take-off of airplane could result in steep
temperature gradients between the outside and the inside layer of concrete. This
could results in cracking and spalling of concrete at joints or poorly compacted areas
or delemination of concrete at the concrete-steel interface.
Very low ambient temperatures (−200 oC - 0 oC) could result in the formation
24
of ice in the hydrated cement paste. The formation of ice during setting and hardening
of concrete could seize the hydration process of cement. Upon thawing the hydration
of cement continues and concrete need to be vibrated again. Lack of vibration might
result in setting and hardening of concrete in such a way that high porosity will be
created due to expansion of mixing water upon freezing. The high porosity will
decrease the strength and accelerate the diffusion of moisture and some deleterious
chemical into concrete. If ice formed after setting and before an appreciable strength
gained, concrete might undergoes larger deteriorations. Those deteriorations can be
attributed to the low strength of concrete and the availability of more water, which is
not consumed yet by the hydration of the cement. The formation of ice in the
capillary pores of the hardened cement paste during the service life of concrete results
in volume expansion of those pores. Upon thawing and refreezing more volume
expansions take place in the capillary pores. Subsequent freezing and thawing cycles
would result in progressive damage and thus more cracking of concrete would occur.
The cracking will accelerate the diffusion of the moisture and chloride ions into
concrete and those affect adversely its durability.

CHAPTER 3
COUPLED MOISTURE DIFUSSION AND DRYING SHRINKAGE
3.1 Introduction
Drying shrinkage of concrete occurs due to the loss of moisture and thus, it is
controlled by moisture diffusion process. On the other hand, shrinkage causes
cracking of concrete and affects its moisture diffusion properties. And thus moisture
diffusion and drying shrinkage are two coupled processes and their interactive effect
is very important for durability of concrete structures, especially structures with large
surface area to volume ratio, such as highway pavements, bridge decks, parking
garages and industrial garages.
In recent years, many researchers have investigated the interactive effect of
the mass diffusion and the cracking resulted from moisture gradient, temperature
gradient and mechanical loading. All the research work indicated that the
microcraking produced during hardening, drying and heating of concrete and the
cracking resulted from the mechanical loading could increase the diffusion coefficient
and thus, accelerate the mass diffusion process. The diffusion properties of cracked
concrete were related to several damage indicators. One indicator is the critical crack
width (Aldea et al., 1999; Reinhardt et al., 1997; Wang et al., 1997; Bazant et al.,
1987). Another indicator is the critical monotonic or cyclic loading levels (Gontar et
26
al., 2000; Meziani and Skoczylas, 1999; Konin et al., 1997; Saito and Ishimori, 1995;
Kermani, 1991). Gerard and Marchand (2000) conducted a theoretical study on the
effect of both isotropic and anisotropic traversing crack networks on the steady state
diffusion properties of severely damaged concrete. The diffusion equation was
simplified by assuming that the cracks are uniform in size and distributed evenly on
one or two-dimensional grid. The study concluded that the cracking could accelerate
the diffusion process in concrete. Also, they proposed a simple model for the effect of
cracking on the diffusion coefficient of concrete based on the crack density and mean
crack aperture. Bazant et al. (1987) investigated the effect of crack width and crack
spacing on the permeability and diffusion coefficient of concrete. It was observed that
the permeability and diffusion coefficient of cracked concrete is proportional to the
crack width cubed and inversely proportional to the crack spacing. Bentz et al. (1995)
developed structural models for the properties of the calcium silicate hydrated (C-S-
H) gel at the nanometer level. In this model the calcium silicate hydrate (C-S-H) gel
particles are represented by hard core-soft shell concentric spheres. Each one of those
spheres can overlap with other sphere on the soft core as well as the hard core. This
nanostructural models can be used in the multi-scale modeling of the drying
shrinkage of concrete. Also, the developed models are validated by the images
obtained from the high-resolution transmission electron microscopy (TEM) and the
experimental data measured by sorption test.
In the following sections, the diffusion model and the shrinkage model are
briefly described first, and then the coupling of the model parameters is discussed in
detail. The moisture distribution in concrete was modeled using the nonlinear
27
moisture diffusion equation derived based on Ficks’ law (Bazant and Najjar 1972,
Bazant and Thonguthai, 1978, Xi et al. 1994). The nonlinear moisture diffusion
equation involves two diffusion parameters, the moisture capacity and humidity
diffusion coefficient. Those two parameters were coupled with the material
parameters related to drying shrinkage. Specifically, the moisture capacity is related
to the shrinkage coefficients of the constituents in concrete, and the moisture
diffusion coefficient is related to the damage parameter induced by the drying
shrinkage. Both of the diffusion model and the shrinkage model used in the study are
multiphase and multiscale models, in which the influential parameters at macro-,
meso-, and micro-scales are taken into account; and concrete design parameters are
included as model inputs. Finally, the coupled problem was solved using finite
difference method. The influential parameters such as water-cement ratio and
aggregate volume fraction were studied through several numerical applications.
3.2 Modeling Moisture Diffusion in Concrete (without considering shrinkage-
induced damage)
3.2.1 Basic formulation of the governing equation
The isothermal moisture flux in concrete can be formulated using Fick’s first
law by two different methods. In the first method, the moisture flux (J) is
proportional to the gradient of the water content and can be written as following:
J = − k grad (we ) (3.1)
In the second method, the moisture flux (J) is proportional to the gradient of the pore
28
relative humidity, and can be written as following:
J = − k grad (H ) (3.2)
where, we and H are the evaporable water content and the pore relative humidity,
respectively. Also k has two different physical meanings in above two formulations.
In Eq. (3.1), k is the moisture diffusion coefficient and in Eq. (3.2), k is the humidity
diffusion coefficient.
The isothermal rate of change of total water content, w, in concrete can be
expressed using the mass balance equation or Fick’s second law as follows:
∂w
= − div ( j ) (3.3)
∂t
in which, w=we+wne, where wne is the nonevaporable or chemically combined water
content, The substitution of the moisture flux (J), described in Eqs. (3.1) and (3.2)
into the mass balance equation, Eq. (3.3), will resulted in two formulations of the
nonlinear moisture diffusion equation as follows:
∂ (we + wne )
= div ( Dw grad we ) (3.4)
∂t
∂w ∂H
= div ( DH grad H ) (3.5)
∂H ∂t
where, H is the relative humidity, Dw and DH are the moisture diffusion coefficient
and the humidity diffusion coefficient, ∂w/∂H is the moisture capacity and t is the
29
time. The nonlinearity of the moisture diffusion equation, Eq. (3.4), resulted from the
dependence of the moisture diffusion coefficient, Dw, on the moisture content, w. And
the nonlinearity of the moisture diffusion equation, Eq. (3.5), resulted from the
dependence of the moisture capacity, ∂w/∂H and the humidity diffusion coefficient,
DH on the relative humidity, H.
In the present work the second formulation in terms of relative humidity H,
which described by Eqs. (3.2) and (3.5) will be used to describe the moisture
diffusion in concrete. This is due mainly to the fact that amounts of we and wne are
functions of time even if there is no moisture exchange with the surrounding
environment (i.e. total water, w, is a constant). Otherwise another model must be
included to characterize the effect of hydration of cement on we and wne.
3.2.2 Initial and boundary conditions
The initial condition can be written as follows:
H ( x, y, z, t = 0) = H o ( x, y, z ) (3.6)
where, Ho(x,y,z) is the initial distribution of relative humidity in the concrete body.
In this work only convective type boundary conditions will be considered.
This type of boundary conditions is prescribed at the surfaces of the body exposed to
the surrounding environment. In the case there is a difference in the relative humidity
between the inside of the exposed body and the surrounding environment and the
surfaces of the body are not sealed, an exchange of moisture will occur. If the
30
surrounding relative humidity is larger, the moisture will flux through the exposed
surfaces from the environment to the body and this case called wetting. On the other
hand, if the relative humidity inside the body is larger, the moisture will flux out of
the body to the surrounding environment and this case called drying. In both cases the
boundary conditions are imposed on the surface as follows:
q.n = 0 (3.7)
q.n = β (H − H env ) (3.8)
Where, q is the humidity flux, n is a unit vector normal to the exposed surface and
directed towards the exterior environment, β is the surface moisture diffusion
coefficient, H is the unknown surface humidity and Henv is the environmental relative
humidity. Eq. (3.7) represents the case of no moisture exchange between the body and
the environment on the surface of that body.
3.2.3 Material parameters
In order to find a numerical solution for the governing partial differential
equation of moisture diffusion in concrete, Eq. (3.5), the material parameters involved
should be determined.
Moisture capacity, ∂w/∂H
The moisture capacity of concrete can be calculated based on the multiphase
and multiscale model. Concrete is a two-phase material, in which the aggregates are
considered as the inclusions and represent one phase. The cement paste is the matrix
31
and represents the other phase. Then, without considering the effect of shrinkage of
concrete, the moisture capacity of concrete can be calculated as a certain average of
the moisture capacities of the two constituent phases:
⎛ ∂w ⎞ ⎛ ∂w ⎞ ⎛ ∂w ⎞
⎜ ⎟ = f agg ⎜ ⎟ + f cp ⎜ ⎟ (3.9)
⎝ ∂H ⎠ avg ⎝ ∂H ⎠ agg ⎝ ∂H ⎠ cp
where fagg and fcp are the weight percentages of the aggregates and cement paste. The
moisture capacities of the aggregate and cement paste in Eq. (3.9) can be evaluated by
using the model developed by Xi et al. (1994a, 1994b) and Xi (1995a, 1995b) as
shown in Appendix A.
Humidity diffusion coefficient, DH
Humidity diffusion coefficient depends strongly on the diffusion mechanisms,
which depend on the pore structure of the concrete body. There are three different
diffusion mechanisms of moisture into concrete; molecular or ordinary diffusion,
Knudsen diffusion and surface diffusion. Thus, the total diffusivity is a complex
property that often includes contribution from multiple mechanisms. The moisture
diffusion coefficient of concrete can be predicted by the composite model developed
by Christensen (1979):
⎛ gi ⎞
DH = DHcp ⎜1 + ⎟ (3.10)
⎜ [1 − g ] / 3 + 1/[(D / D ) − 1] ⎟
⎝ i Hagg Hcp ⎠
in which, gi is the aggregate volume fraction, DHcp is the humidity diffusion
coefficient of the cement paste and DHagg is the humidity diffusion coefficient of the
32
aggregates.
The humidity diffusion coefficient of aggregates in concrete is very small
comparing with the diffusion coefficient of concrete and can be neglected in Eq.
(3.10). The humidity diffusion coefficient of cement paste can be predicted by the
empirical model developed by Xi et al. (1995b).
γ h ( H −1)
D Hcp = α h + β h [1 − 2 −10 ] (3.11)
where, αh, βh and γh are coefficients to be calibrated from the test data and given as:
α h = 1.05 − 3.8 ( w c) + 3.56 ( w c) 2 (3.12)
β h = −14.4 + 50.4 ( w c) − 41.8 ( w c) 2 (3.13)
γ h = 31.3 − 136 ( w c ) + 162 ( w c ) 2 (3.14)
3.3 Numerical Solution
The nonlinearity of the partial differential equation, which governs the moisture
diffusion in concrete, makes the numerical solution favorable. The numerical solution
can be done either by finite element method or finite difference method.
3.3.1 Finite element method
The moisture diffusion equation, Eq. (3.5), can be solved using finite element
method. In the finite element solution the standard residual procedure may be applied
33
as follows:
⎡ ∂w ∂H ⎤
∫Ω
wi ⋅ ⎢
⎣ ∂H ∂t
− div ( D H grad H ) ⎥ d Ω = 0
⎦
(3.15)
By choosing standard Galerkian method in which, wi = Ni .
⎡ ∂w ∂H ∂H ⎤
∫Ω
NT ⋅ ⎢
⎣ ∂H ∂t
− div ( DH ) dΩ = 0
∂X ⎥⎦
(3.16)
Rearranging Eq. (3.16) results in:
∂w ∂H ∂H ∂N T ∂H
∫Ω
NT ⋅
∂H ∂t
dΩ − ∫ div ( N T ⋅ DH
Ω ∂X
) dΩ + ∫
Ω ∂X
DH
∂X
dΩ = 0 (3.17)
By using divergence theorem, Eq. (3.17) becomes:
∂w ∂H ∂H ∂N T ∂H
∫Ω
NT ⋅
∂H ∂t
dΩ − ∫ N T ⋅ DH
Γ ∂X
⋅ n dΓ + ∫
Ω ∂X
DH
∂X
dΩ = 0 (3.18)
The flux of moisture, q can be written as:
∂H
q = − DH (3.19)
∂X
Substituting Eq. (3.19) into Eq. (3.18) resulting in:
∂w ∂H ∂N T ∂H
∫Ω
NT ⋅
∂H ∂t
dΩ + ∫ N T ⋅ q ⋅ n dΓ + ∫
Γ Ω ∂X
DH
∂X
dΩ = 0 (3.20)
Substituting the boundary condition described in Eq. (3.8) into Eq. (3.20) resulting in:
34
∂w ∂H ∂N T ∂H
∫ NT ⋅ dΩ + ∫ N T ⋅ β (H − H env ) dΓ + ∫ DH dΩ = 0 (3.21)
Ω ∂H ∂t Γ Ω ∂X ∂X
The relative humidity at any point in an element can be determined using shape
functions, N as follows:
H ( X , t ) = ∑ N i ( X )H i (t ) (3.22)
By applying the finite element method with the shape functions Ni of the ith order we
have:
⎛ T ∂w ⎞ ∂H ⎛ ∂N T ∂N ⎞
⎜∫N N dΩ ⎟ + ⎜∫ D dΩ + ∫ N T
β N dΓ ⎟ H = ∫ N T β H env dΓ (3.23)
⎜ ∂H ⎟ ∂t ⎜ ∂X H
∂X ⎟
⎝Ω ⎠ ⎝Ω Γ ⎠ Γ
The moisture diffusion in concrete is described by the following general expression:
∂H
C + KH = F H (3.24)
∂t
Where, C is the moisture capacity matrix, K is the moisture diffusion coefficient
matrix and FH is the hygral load vector and can be written as follows:
∂w
C = ∫ NT N dΩ (3.25)
Ω
∂H
T
∂N ∂N
K =∫ DH dΩ + ∫ N T β N dΓ (3.26)
Ω
∂X ∂X Γ
FH = ∫ N T βH env dΓ (3.27)
Γ
35
Eq. ( 3.24) can be rewritten as follows:
∂H
+ AH = P (3.28)
∂t
where,
A = C −1 K (3.29)
P = C −1 FH (3.30)
Eq. (3.28) can be solved using one the step-by-step integration schemes as
follows (Argyris et al. 1977):
[I + α∆tA]H i +1 = [I − (1 − α )∆tA]H i + ∆t (1 − α ) Pi + ∆tαPi +1 (3.31)
The value of α should lies in the interval [0,1]; for α = 0, we have Forward Euler
Method (Explicit scheme), for α = 1, 0.5, and 2/3, we have Backward Euler Method,
Trapezoidal and Galerkian Methods, respectively (Implicit schemes).
3.3.2 Finite difference method
The Alternating-Direction Implicit (ADI) finite difference method developed
by Peaceman and Rachford (Carnahan et al. 1969) can be employed to solve the
moisture diffusion problem described in Eq. (3.5). A two-dimensional finite
difference mesh as shown in Fig. (3.1) can be used to descretize this problem. The
pore relative humidity, H(i,j,n), at different points of the mesh at different time
intervals need to be estimated.

36
∆X
Slab
∆Y Depth
Slab Width
Fig. 3.1: Finite difference mesh of concrete slab
For the two dimensional problem, see Fig. (3.1), Eq. (3.5) can be rewritten as
follows:
∂w dH ∂ ⎛ ∂H ⎞ ∂ ⎛ ∂H ⎞
= ⎜D ⎟+ ⎜D ⎟⎟ (3.32)
∂H dt ∂x ⎝ ∂x ⎠ ∂y ⎜⎝ ∂y ⎠
where, the moisture capacity, ∂w/∂H and the humidity diffusion coefficient, D
(replace DH) are functions of H, x, y and t. And, the relative humidity, H is function of
x, y and t.
Further simplification of Eq. (3.32) results in:
∂w dH ∂ 2 H ∂D ∂H ∂ 2 H ∂D ∂H
=D 2 + +D 2 + (3.33)
∂H dt ∂x ∂x ∂x ∂y ∂y ∂y
The principle of Alternating-Direction Implicit (ADI) method can be applied
to solve Eq. (3.33) by using two difference equations over two successive time-steps
each of duration ∆t/2:

37
n
H i*, j − H in, j
∆t / 2
⎛ ∂H ⎞
=⎜
∂ w
n
[
⎟ * Di , jδ x H i , j + δ x Di , jδ x H i , j + Di , jδ y H i , j + δ y Di , jδ y H i , j
2 * n * n 2 n n n
] (3.34)
⎝ ⎠i , j
H in, +j 1 − H i*, j n
∆t / 2
⎛ ∂H ⎞
=⎜
∂w
n
[
⎟ * Di , jδ x H i , j + δ x Di , jδ x H i , j + Di , jδ y H i , j + δ y Di , jδ y H i , j
2 * n * n 2 n +1 n n +1
] (3.35)
⎝ ⎠i , j
Where, δx, δ2x and δy, δ2y are the first and second order central-difference operators in
x and y directions, respectively. Eq. (3.34) is implicit only in x-direction and used to
solve for Hi,j* (intermediate value at the end of the first time-step) by knowing Hi,jn
starting from the initial condition. Eq. (3.35) is implicit only in y-direction and used
to solve for Hi,jn+1 (final value at the end of the whole time step) after calculating Hi,j*
in the first step.
The stability of the Alternating-Direction Implicit (ADI) method was
discussed by von Neumann method (Carnahan et al. 1969). It was shown that this
method is unconditionally stable and the solution converges within a discretization
error of O[(∆t)2+(∆x)2+(∆y)2].
3.4 Multiscale Modeling of Drying Shrinkage
Shrinkage of concrete has been evaluated in two different ways. It can be
estimated for structural members, and thus represents the average strain over a large
scale. In this case, the size and shape of the structure, properties of the concrete, and
the effect of the environment have to be taken into account. As a result, the prediction
has been based on some empirical equations.

38
On the other hand, the shrinkage can be evaluated for a small representative
volume of concrete. In this case, the predicting model is a constitutive relationship
among the shrinkage, the humidity, and related material parameters. The structural
configuration is not included. The first type of models (for structural members) is
used in structural analysis, and the second type (constitutive laws) is mainly used for
numerical analysis such as finite element method and finite difference method.
One of the constitutive models to predict free shrinkage of concrete is
suggested by Bazant and Chern (1985). In this model the rate of change of the local
shrinkage strain is related to the rate of change of the relative humidity as follows:
∆ε sh = β sh ∆H (3.36)
in which, βsh is shrinkage coefficient of concrete, which can be characterized by a
multiscale model developed by Xi and Jennings (1997).
Concrete is a multiscale material and can be partitioned into four different scale
levels ranging from nanometer to meter level as shown in Fig. (3.2). The first scale
level is the meter level where most of the civil engineering structures are controlled
by theories and laws applicable on this level. The second is the millimeter level where
most of the physical and mechanical properties of concrete are obtained from samples
and testing procedures applicable on this level. The third is the micrometer level
where the pore structure of cement paste can be measured at this level. Also, the
microstructural features of the cement paste such unhydrated cement particles,
calcium hydroxide (CH) crystals, calcium silicate hydrate (C-S-H) gel and other
39
hydration products are observed at this scale level. The fourth is the nanometer level
where many models for the structure of calcium silicate hydrate (C-S-H) are
developed at this level (Bentz et al. 1995).
Meter Level
(Real Structure)
Millimeter Level
(Concrete)
Micrometer Level
(Cement Paste)
X1000
Nanometer Level
(Calcium Silicate Hydrate)
X1000
X1000
Fig. 3.2: Different scale levels of concrete.
Multiscale mathematical modeling can be used to predict the drying shrinkage
of cement paste and concrete. Concrete as shown in Fig. (3.3) is a particulate
composite material in which the aggregates are the inclusions, and the cement paste is
the matrix. Also, cement paste as shown in Fig. (3.4) is a particulate composite
material composed of unhydrated cement particles; calcium hydroxide (CH) and
ettringite crystals, and calcium silicate hydrate (C-S-H) gel.

40
Fig. 3.3: Composite action of concrete.
Fig. 3.4: Microstructral features of cement paste.
The three-phase model developed by Christensen (1979) to determine
effective elastic properties of a two-phase composite can be used in the multiscale
modeling for concrete and cement paste. This model was originally developed for
elastic properties only, but many researchers (Herve and Zaoui 1990) proved that this
41
model could be extended to nonlinear materials.
The shrinkage of concrete at the millimeter level, see Figs. (3.2) and (3.3), is
equivalent to the effective shrinkage of the three- phase model shown in Fig. (3.5),
where the inclusions (aggregates) represent phase 1 and the matrix (cement paste)
represents phase 2.
Phase 1 (Aggregate)
Phase 2 (Cement Paste)
Effective homogeneous medium (Concrete)
Fig. 3.5: Three-phase effective media model.
The shrinkage coefficient and bulk modulus of concrete (at mesoscale level),
βsh and Kc, can be evaluated (Xi and Jennings 1997) for the composite system:
K1`ε 1sh c12 (3K 2 + 4G2 ) + K 2ε 2sh (1 − c12 )(3K1 + 4G2 )

β sh = (3.37)
K 2 (3K1 + 4G2 ) − 4c12 G2 (K 2 − K1 )
K 2 (3K1 + 4G2 ) − 4c12 G2 (K 2 − K1 )

Kc = (3.38)
(3K1 + 4G2 ) + 3c12 (K 2 − K1 )
where, ε 1sh and ε 2sh , K1 and K2, and G1 and G2 are the shrinkage, bulk modulus and
42
shear modulus of phase 1 (aggregates) and phase 2 (cement paste), respectively; c12 is
the volume fraction of phase 1 (aggregates). The shrinkage of concrete is mainly due
to the shrinkage of cement paste, where the shrinkage of the aggregates can be
neglected ( ε 1sh = 0) and so Eq. (3.37) becomes:
⎧ ⎫
⎪⎪ c12 (K1 / K 2 ) ⎪⎪
β sh = ε 2sh ⎨1 − ⎬ (3.39)
⎪ 1 + (K1 / K 2 − 1) 3 + 4c12 G2 / K 2 ⎪
⎩⎪ 3 + 4G2 / K 2 ⎭⎪
The shrinkage strain of concrete can be calculated using Eq. (3.39) and knowing the
bulk modulus K1, shear modulus G1 and volume fraction of aggregates, shrinkage of
cement paste ( ε 2sh ), bulk modulus K2, shear modulus G2 of cement paste.
Applying the same concept at the micrometer level, see Figs. (3.2) and (3.4),
the elastic properties (K and G) and the shrinkage strain of cement paste can be
calculated using the four-phase composite model shown in Fig. (3.6). In this model
phase 1 is the solid core, which includes the unhydrated cement particles, calcium
hydroxide (CH) and other crystalline products, phase 2 is inner hydration products,
phase 3 is the outer hydration products. Both the inner product and the outer product
are C-S-H (Jennings and Tennis 1994) but the density of the inner product is higher
than that of the outer product (Diamond and Bonen 1993), and the shrinkage of C-S-
H is mainly from the outer product. The bulk modulus and shrinkage strain of cement
paste is equivalent to the bulk modulus and shrinkage strain of the effective
homogeneous four-phase model as follows:

43
K eff
12
(ε effsh )12 ( f 1 + f 2 )(3K 3 + 4G3 ) + K 3ε 3sh (1 − f 1 − f 2 )(3K eff
12
+ 4G3 )
ε sh
eff = (3.40)
K 3 (3K eff
12
+ 4G3 ) − 4( f 1 + f 2 )( K 3 − K eff
12
)G3
( f 1 + f 2 )( K eff
12
− K3 )
K eff = K 3 + (3.41)
1 + (1 − f 1 − f 2 )(( K eff
12
− K 3 ) /( K 3 + 4G3 / 3))
where,
K 1ε 1sh ( f 1 /( f 1 + f 2 ))(3K 2 + 4G 2 ) + K 2 ε 2sh (1 − f 1 /( f 1 + f 2 ))(3K 1 + 4G 2 )

(ε effsh ) 12 = (3.42)
K 2 (3K 1 + 4G 2 ) − 4( f 1 /( f 1 + f 2 ))( K 2 − K 1 )G 2
( f 1 /( f 1 + f 2 ))( K 1 − K 2 )
K eff
12
= K2 + (3.43)
1 + (1 − f 1 /( f 1 + f 2 ))((K 1 − K 2 ) /( K 2 + 4G 2 / 3))
in which, f1, f2 and f3 are the volume fraction of phases 1, 2 and 3, K1, K2 and K3 are
the bulk modulus of phases 1, 2 and 3, G1, G2 and K3 are the shear modulus of phases
1, 2 and 3 and ε 1sh , ε 2sh and ε 3sh are the shrinkage of phases 1, 2 and 3.
Phase 1 (Unhydrated cement and crystalline phases)

Phase 2 (Inner products of cement hydration)
Phase 3 (Outer products of cement hydration)
Effective homogeneous medium (Cement paste)
Fig. 3.6: Four-phase composite model.

44
From test results using the environmental scanning electron microscopy
(Neubauer et al. 1997), it was found that the shrinkage of the phase 1 (unhydrated
cement particles, calcium hydroxide crystals and other crystalline products) is very
small and can be neglected ( ε 1sh =0). Also, It was found that the shrinkage of phase 2
(inner hydration products) in the range of relative humidity between 30% and 100%
is very small and can be neglected ( ε 2sh =0). As a results Eq. (3.42) vanishes and Eq.
(3.40) becomes:
K 3 ε 3sh (1 − f 1 − f 2 )(3K eff

12
+ 4G3 )
ε sh
eff = (3.44)
K 3 (3K eff
12
+ 4G3 ) − 4( f 1 + f 2 )( K 3 − K eff
12
)G3
The volume fraction of the three phases f1 (solid core), f2 (inner hydration
products) and f3 (outer hydration products) can be calculated using Jenning and
Tennis model (1994) as described in Appendix B.
Some of the elastic properties (bulk modulus, K and shear modulus, G) for the
constituent phases of the multiscale shrinkage model are difficult to evaluate,
especially at the microscale level. An inverse method was developed by Xi and
Jennings (1997), in which the elastic properties were evaluated based on the elastic
properties of concrete and cement paste at the millimeter level.
3.5 The Interaction of Drying Shrinkage and Moisture Diffusion
The interactive effect of drying shrinkage and moisture diffusion can be
studied using one of the two alternatives. The first alternative is to consider that the
stresses or strains induced by drying shrinkage are one of the driving forces, and thus,
45
there will be an additional term in the diffusion equation, which corresponds to the
effect of stresses and strains (Majorana and Mazars 1997). The second alternative is
to consider the effect of the drying shrinkage on the two material parameters of the
diffusion equation, Eq. (3.5), i.e. moisture capacity and moisture diffusion coefficient.
In this case, the stress or strain is not appeared explicitly in the diffusion equation, but
as an internal parameter. We applied the second alternative in the present study.
The effects of drying shrinkage on the two material parameters are
incorporated into the diffusion equation by two different methods. For the moisture
diffusion coefficient, we consider that the effect of the damage can be treated by a
way similar to the degradation in the secant modulus of elasticity, where the
degradation of the secant modulus can be characterize by the scalar damage
parameter. For the moisture capacity, we consider that the effect of the damage can
be treated by a way similar to the degradation in the heat capacity of thermoelasticity,
where one of the influential factors to the heat capacity is the coefficient of thermal
expansion.
3.5.1 The effect of shrinkage-induced damage on moisture diffusion coefficient
Let us first consider the drying shrinkage of a concrete layer, which can be
simplified as a plane stress problem, as shown in Fig. (3.7), in which x direction is
perpendicular to the plane, and the stresses in y and z direction are equal and can be
written as:
σ x = τ xy = τ xz = 0 (3.45)
46
γ xy = γ xz = 0 (3.46)
z z
y x
x
Fig. 3.7: Plane stress problem.
Since shrinkage strain is related to volumetric change only,
τ yz = 0 and γ yz = 0 (3.47)
The total strain is the summation of the mechanical strain and the shrinkage strain:
ε xt = ε xm + ε sh
ε ty = ε ym + ε sh (3.48)
ε = ε + ε sh
t
z
m
z
The shrinkage strain can be written in a similar form to the thermal strain as follows:
∆ε sh = β sh ∆H (3.49)
47
where, βsh is shrinkage coefficient, similar to thermal expansion, coefficient and ∆Η
is the change of relative humidity.
Since the total strains vanishes in y and z directions:
ε ym = −ε sh
(3.50)
ε zm = −ε sh
The stresses in y and z direction are equal and can be written as:
Es
σ y =σz = ε sh (3.51)
1 −υ
where, E s is the secant elastic modulus and υ is the Poissons’ ratio of concrete. The
secant elastic modulus can be written in terms of the scalar damage parameter, d as
follows:
E s = (1 − d )E o (3.52)
where, E o is the initial tangent elastic modulus of concrete? Combining Eq. (3.51)
and (3.52)
(1 − d )E o
σy =σz = ε sh (3.53)
1−υ
Since drying shrinkage produce tensile volumetric strains (εx=εy=εz=εsh), the scalar
damage parameter, d can be written in terms of the shrinkage strain (εsh) using a
tensile stress-strain relationship of concrete, and so Eq. (3.53) can be rewritten as:
48
σy =σz =
(1 − d (ε sh ) )Eo ε
1−υ
sh
(3.54)
The incremental stress can be written as:
Eo ⎛ ∂d ⎞
dσ y = dσ z = ⎜⎜ [1 − d (ε sh ) ] − ε sh ⎟⎟ dε sh (3.55)
1−υ ⎝ ∂ε sh ⎠
Starting from a certain time step, n where the stress and the strain are known, the
stress and the strain at the time n+1 can be written as:
ε shn +1 = ε shn + dε sh
σ yn +1 = σ yn + dσ y (3.56)
σ yn +1 = σ yn + dσ y
The scalar damage model, was first introduced by Kachanov (1958), based on
the assumption that the area (or the volume) with damage cannot hold any load, so,
there is a reduction in the effective cross section area as loading level increase, which
can be generally expressed by a reduction in the modulus of elasticity. Following the
same concept, we can assume that, for moisture diffusion in concrete, the damaged
area (or volume) cannot resist the penetration of moisture. So, there is also a
reduction in the effective cross section area, which can be generally expressed as an
increase in the volume of pathway for the moisture to diffuse in concrete.
Therefore, with the increase of damage, the concrete diffusion coefficient
increases. Similar to the effect of damage on modulus of elasticity, the effect of

49
damage on the moisture diffusion coefficient, DH, can be expressed in terms of the
scalar damage parameter, d:
D H (H )
D H (H , d ) = (3.57)
1− d
in which, DH(H) is the diffusion coefficient of intact of intact concrete, and DH(H,d)
is the diffusion coefficient with damage.
3.5.2 The effect of drying shrinkage on moisture capacity
First, we compare the similarities between the effect of thermal stress on heat
capacity of composite and the effect of drying shrinkage stress on moisture capacity.
Eq. (3.5) for moisture diffusion has the same form as the equation for heat
conduction, Eq. (3.58)
∂T
Ct = div[ K T grad (T )] (3.58)
∂t
where, T = temperature; KT = thermal conductivity, Ct = heat capacity. Moreover, the
constitutive equations for thermal stress, Eq. (3.59), and the constitutive equations for
shrinkage stress, Eq. (3.60), also have the same form
(
σ ij = Cijkl ε ij − α ijθ ) (3.59)
(
σ ij = Cijkl εij − βij η ) (3.60)
in which σij = stress tensor; εij = strain tensor; Cijkl = elastic tensor; αij = coefficients
50
of thermal expansion; β = coefficients of drying shrinkage; θ is the change of

ij
0
temperature in Kelvin from the reference state T , T = T + θ ; η is the change of
0
pressure from the reference state p0, p=p0+η, where p0 = reference pressure in Pa.
The similarities of the corresponding coefficients for heat conduction and
moisture transfer are obvious: αij verses βij; KT verses DH; and Ct verses ∂w/∂H. The
idea of the comparative study is that the composite models developed for heat
conduction may possibly be borrowed for evaluating the effective parameters of
moisture transfer. Based on Eqs. (3.58) and (3.59), the effect of thermal strain on
heat capacity has been determined by Rosen and Hashin (1970) using a generalized
method based on extreme energy principle. Rosen and Hashin model turned out to be
the only model for the effect of thermal strain on effective heat capacity. Similarly,
using Eqs. (3.5) and (3.60) together with the extreme energy principle, Xi (1995a and
1995b) derived an analytical model for the effect of shrinkage strain on moisture
capacity for two-phase composite materials
2
⎛ ⎞ ⎡ _ ⎤
⎜ ⎟ ⎢⎛⎜ 1 ⎞⎟ 1 ⎥
−
p ⎜ βi − βm ⎟
0
n, p = n, p + 9
p p
⎢⎜ ⎟− ⎥
RT 0 ⎜ 1 1 ⎟ ⎢⎜ K ⎟ K ⎥
(3.61)
⎜ − ⎟
⎝ Km Ki ⎠ ⎢⎣⎝ ⎠ ⎥⎦
p0
n −n =
v
,p
p
,p C ijkl β ij β kl (3.62)
RT 0
with,
51
⎛ _ ⎞
⎜ 1 ⎟ gm gi
_
n = gmn
p
,p
p
, pm + gin p
, pi and ⎜⎜ K ⎟⎟ = K + K (3.63)
m i
⎝ ⎠
p v
in which n, p is the moisture capacity at constant stress in moles/m3, n, p is the
moisture capacity at constant deformation in moles/m3; the subscript p means that the
moisture capacity is in terms of pressure p; p = p0+ η; n= moles of water adsorbed
per volume of the composite material; subscript m is for matrix and i for inclusion;
n,ppm and n,ppi are the moisture capacities at constant stress for matrix and inclusion,
respectively; K = bulk modulus in Pa; R = gas constant in Joules/mole/K; β =
shrinkage coefficient; where β i and β m are in terms of η; and g = volume fraction.
Eqs. (3.61) through (3.63) are general expressions for moisture capacities of two
phase composites. In order to apply these equations to concrete, some modifications
are needed.
First of all, the distinction between the moisture capacities at constant stress
and constant deformation has not been directly identified by experiments, especially
for cement paste and aggregate. In current engineering practice, experiments of
equilibrium isotherm are carried out in a condition of neither constant deformation
nor constant stress. After a detailed analysis, Xi (1995a) concluded that the
v p
difference between n, p and n, p are negligible, which means n,vp ≈ n,pp = n, p .
Superscripts v and p may be dropped. Furthermore, all adsorption test data and
shrinkage coefficients for concrete have been obtained in terms of relative humidity H
rather than pore pressure p, hence, Eq. (3.61) must be expressed in terms of H, too. H
52
= p/ps, where p s is saturation pressure and depends on temperature. At isotherm
condition, ps = constant, so, n, p may be replaced by ∂w/∂H, and p by H . Then, the

0 0
final form of the moisture capacity suitable for concrete is
⎧ ⎛ ⎞
2
⎡ _ ⎤⎫
⎪ ⎜ ⎟ ⎢⎜ 1 ⎟ 1 ⎥ ⎪⎪
⎛ ⎞
∂w ⎛ ∂w ⎞
=⎜ ⎟
⎪ H0
+ ψ ⎨9 ⎜ βi − βm ⎟ ⎢⎜ ⎟− ⎥⎬
∂H ⎝ ∂H ⎠ avg ⎜ 1 1 ⎟ ⎢⎜ K ⎟ K ⎥ ⎪ (3.64)
⎪ RT
0
⎜ − ⎟
⎪ ⎝ Km Ki ⎠ ⎢⎣⎝ ⎠ ⎥⎦ ⎪
⎩ ⎭
in which (∂w/∂H)avg is the moisture capacity of concrete without considering
the effect of shrinkage as shown in Eq. (3.9); β i and β m are in terms of η, which
now stands for the change of relative humidity H, and ψ is a factor used to convert
the units of the second term in the right hand side of Eq. (3.61) from mole/m3 to
gram/gram. In Eq. (3.64), all material parameters are for an individual phase (either
cement paste or aggregate) except K, which is the effective bulk modulus of concrete.
K may be related to the properties of cement paste and aggregate by a proper
composite model. In the present study, we used the so-called three-phase model
(Christensen 1979)
g i (K i − K m )
K = Km +
Ki − Km (3.65)
1+ gm
4
K m + Gm
3
where Gm = shear modulus of the matrix (cement paste) in Pa. As one can see
from Eq. (3.64), the moisture capacity of concrete is not just the weighted average of
53
the moisture capacities of the matrix and the inclusion, it depends strongly on the
shrinkage coefficients of the two constituents (see the second term).
In the case that the shrinkages of the inclusion and matrix are the
same, the second term in Eq. (3.64) vanishes, which means that there is no coupling
effect. Otherwise, the coupling effect must be taken into account. For concrete,
β m ≥ β i and K i ≥ K m (the matrix shrinks more than the inclusion and the inclusion is
stronger than the matrix), the additional term depends strongly on shrinkage values of
the constituent phases. With increasing β m β i the value of the additional term
increases drastically. Numerical analyses by Xi (1995a and 1995b) showed that the
influence of the shrinkage ratio is more significant than that of the ratio of bulk
modulus, and furthermore, the second term in the right hand side of Eq. (3.64) is
always positive when β m ≥ β i and K i ≥ K m , which means that the differential
shrinkage between the two constituent phases increases the moisture capacity of the
composite.
3.6 Numerical Applications
The coupled problem of moisture diffusion and drying shrinkage discussed in
the previous sections was implemented and solved in a computer program. The
moisture diffusion was solved using the Alternating-Direction Implicit (ADI) finite
difference method discussed in section 3.3.2. The drying shrinkage was modeled
according to the model shown in Eq. (3.36). As a numerical example, a 10 cm depth
concrete slab similar to the slab shown in Fig. (3.1) was analyzed. The concrete has a
54
water-cement ratio (w/c) of 0.55, aggregate volume fraction (gi) of 0.65, average
compressive strength (28 days) of 34 MPa and moist cured for 28 days before drying.
The slab, which is initially saturated (Hini=100%), is exposed to drying on the top
surface (Henv= 50 %) after curing. The effect of micro-cracking due to drying
shrinkage is taken into account through the scalar damage parameter. The hygral
stresses are calculated based on the assumption of plane stress for concrete. The
tensile stress strain curve used in the example is the one suggested by Zhen-hai and
Xiu-qin (1987) and it is shown in Fig. (3.8). The graph shows that the strain-strain
relationship is composed of two parts. The ascending (hardening) part is described by
high order polynomial. The descending (softening) part is described by an equation
with α = 0.312 f t ' 2 and β = 1.7, where f t ' is the ultimate tensile stress in (MPa).
f ' = 0.5 f'

t c
(MPa)
f ' [1.2 (εt /εtu) - 0.2 (εt /εtu)6 ] ε [ ε

t t tu
f =
t
Tensile stress, f
t f ' (εt /εtu)

t
β
(εt /εtu) + α[(εt /εtu) −1] εt > εtu
Eo = 4700 f'
c
ε tu = 44x10 -6 ft'
Tensile strain, ε t
Fig. 3.8: Tensile stress strain diagram.

55
The variation of the pore relative humidity (H), shrinkage strain (εsh), scalar
damage parameter (d) and hygral stresses (σy or σz) with the depth from the exposure
surface (top surface) are shown in Figs. (3.9) through (3.12). From Fig. (3.9), we can
see that the drying process becomes slower for concrete at deeper locations from the
exposed surface. The shrinkage is related to the moisture loss, and as a result the
exterior part of concrete slab has larger strain as shown in Fig. (3.10). The build up of
strains near the exposed surface results in more degradation of the elastic modulus
and thus more damage is expected in the exterior skin of concrete and the damage
become less as going deeper from the exposed surface as shown in Fig. (3.11). The
hygral stresses are tensile stresses produced by the restraining of the drying shrinkage
by interior concrete, which shrinks less. Fig. (3.12), shows that stresses are larger in
the outer skin of the concrete and they vanish as going deeper. Therefore, the drying
process, drying shrinkage, the local damage and hygral stresses in the concrete
structural members are related to their surface area-volume ratio. Shallow structural
members such as slabs, decks and walls have larger surface area with respect to their
volume, and thus are subjected to more moisture loss and more deterioration due to
drying shrinkage. On the other hand, other structural members of less surface area
with respect to their volume have less drying shrinkage and the damage is limited
only to surface layer.

56
0.95 t = 1 month
t = 2 months
0.9 t = 4 months
Pore relative humidity, H

0.85
0.8
0.75
0.7
0.65
0.6
0.55
0.5
0 1 2 3 4 5 6 7 8 9 10
Depth (cm)
Fig. 3.9: Variation of relative humidity with the depth.
450
400 t = 1 month
Shrinkage strain, ε sh (microstrain)
t = 2 months
350 t = 4 months
300
250
200
150
100
50
0
0 1 2 3 4 5 6 7 8 9 10
Depth (cm)
Fig. 3.10: Variation of shrinkage strain with the depth.

57
0.8
0.7 t = 1 month
t = 2 months
Scalar damage parameter, d

0.6 t = 4 months
0.5
0.4
0.3
0.2
0.1
0
0 1 2 3 4 5 6 7 8 9 10
Depth (cm)
Fig. 3.11: Variation of scalar damage parameter with the depth.
3.5
Hygral stress, σy or σz (MPa)
2.5
1.5
1
t = 1 month
0.5 t = 2 months
t = 4 months
0
0 1 2 3 4 5 6 7 8 9 10
Depth (cm)
Fig. 3.12: Variation of hygral stresses with the depth.

58
The variation of pore relative humidity, the shrinkage strain, the scalar
damage parameter and the hygral stresses with time of exposure are shown in Figs.
(3.13) through (3.16). The time of exposure is started after the curing period of
concrete. Fig. (3.13) shows the drying processes at different depths in the concrete.
One can see that the rate of drying is higher near the exposed surface. A 20% drop in
the relative humidity occurred approximately within the first 5 days of exposure at 1
cm depth, 25 days of exposure at 3 cm depth and 50 days of exposure at 5 cm depth.
Fig. (3.14) shows that the variations of shrinkage in the concrete specimen at different
depths with increasing time of exposure. It shows that, the rate of the strain
accumulation increases with an increase of exposure time. The strain reaches 0.0001
approximately within 10 days at 1 cm depth, 50 days at 3 cm depth, and 85 days at 5
cm depth. The difference in the shrinkage strain at different depths (i.e. local strain
gradient) results in shrinkage stress and the local damage. Fig. (3.15) shows the
increase of the scalar damage parameter due to the shrinkage stress. From the figure
we can see the damage starts approximately after 5 days at 1 cm depth, 40 days at 3
cm depth and 70 days at 5 cm depth. Also the figure shows that a 40% loss in the
initial elastic modulus occurs approximately after 20 days of exposure at 1 cm depth,
95 days of exposure at 3 cm depth and 125 days of exposure at 5 cm depth. Fig.
(3.16), shows the stress history at different depths in the concrete. The stress in
concrete reaches the maximum tensile strength after approximately 25 days at 1 cm
depth, 105 days at 3 cm depth and 135 days at 5 cm depth. Based on these numerical
results, we can see that long period of drying should be avoided. This is especially
important for shallow structures and early age concrete, where the effect of drying
59
becomes very severe and the damage due to drying reduces the service life of the
concrete structures.
Depth = 1.0 cm
0.9 Depth = 3.0 cm
Depth = 5.0 cm
0.8
0.7
0.6
0.5
0 30 60 90 120 150 180
Time (days)
Fig. 3.13: Variation of relative humidity with time of exposure.
350
Depth = 1.0 cm
Depth = 3.0 cm
300
Shrinkage strain, ε sh (microstrain)
Depth = 5.0 cm
250
200
150
100
50
0
0 30 60 90 120 150 180
Time (days)
Fig. 3.14: Variation of shrinkage strain with time of exposure.

60
0.7
0.6
Scalar damage parameter, d

0.5
0.4
0.3
0.2
0.1 Depth = 1.0 cm

Depth = 3.0 cm
Depth = 5.0 cm
0
0 30 60 90 120 150 180
Time (days)
Fig. 3.15: Variation of scalar damage parameter with time of exposure.
3.5
Hygral stress, σy or σz (MPa)
2.5
1.5
1 Depth = 1.0 cm
Depth = 3.0 cm
0.5 Depth = 5.0 cm
0
0 30 60 90 120 150 180
Time (days)
Fig. 3.16: Variation of hygral stresses with time of exposure.

61
Another important observation is that strain softening induced by shrinkage
reaches 5 cm or more in the concrete, which is the thickness of concrete cover in most
of concrete structures. Of course, the depth of softening front depends on the concrete
quality. This numerical example gives us an important estimation on the damage in
concrete due to drying shrinkage.
3.6.1 Effect of damage
The drying shrinkage creates local damage in concrete, which is taken into
account by the scalar damage parameter. The effect of the damage on the diffusion of
moisture is taken into account by the two models developed in the present study for
the moisture diffusion coefficient and the moisture capacity. Figs. (3.17) and (3.18)
show the coupling effect: profiles of the relative humidity and shrinkage strain in the
concrete at different exposure times with and without considering the effect of the
damage. The starting point of the profiles is from the exposed surface. In Fig. (3.17),
all curves eventually reach the equilibrium distribution of 50%, which is the
environmental relative humidity. One can see that the humidity profiles including the
shrinkage-induced damage approach the equilibrium distribution in a faster rate,
which indicates that the shrinkage-induced damage accelerates the diffusion process
and increases the rate of drying. In turn, as shown in Fig. (3.18), the accelerated
drying process increases the shrinkage strain. For example, comparing the two curves
exposed after three months with and without the damage, the curve with the damage
is much higher. One can also see that, at a fixed depth (say 4 cm), the incremental
shrinkage strain with the damage (e.g. the difference between one month and three
months) is much higher than the corresponding increment without the damage, which
62
indicates that the rate of drying shrinkage is increased due to the damage.
Figs. (3.17) and (3.18) show very clear that there is an active interplay between
the drying process and the resulting shrinkage. Basically the moisture loss causes
drying shrinkage, which generates the damage. In turn, the damage increases
diffusion coefficient of concrete, and thus accelerates the moisture transfer.
0.9
0.8
0.7
0.6
0.5
1 month (Damage Considerd)
0.4 3 months
6 months
1 month (Damage Not Considerd)
0.3 3 months
6 months
0.2
0 2 4 6 8 10
Depth (cm)
Fig. 3.17: Effect damage on drying of concrete.

63
600
1 month (Damage Considerd)

500 3 months
6 months
1 month (Damage Not Considerd)
3 months
ε sh (microstrains)
400
6 months
300
200
100
0
0 2 4 6 8 10
Depth (cm)
Fig. 3.18: Effect of damage on shrinkage of concrete.
3.6.2 Effect of water-cement ratio
The water-cement ratio affects both moisture diffusion and drying shrinkage of
concrete. The effect of water-cement ratio on drying and shrinkage of concrete is
shown by Figs. (3.19) and (3.20). The figures show the time variation of the internal
relative humidity and shrinkage strains at two different depths from the exposed
surface (1 cm and 3 cm) for two different water-cement ratios (0.45 and 0.65). Both
figures show that as the water-cement ratio increases the rate and the magnitude of
drying increases, thus resulting in larger shrinkage strains and higher rate of
shrinkage. By increasing water-cement ratio, the porosity of concrete increases
which, accelerates the moisture diffusion and thus increases the rate of drying.
64
w/c = 0.45
0.9
w/c = 0.65
Pore relative humidity 0.8
at 3 cm depth
0.7
0.6
at 1 cm depth
0.5
0 10 20 30 40 50 60 70 80 90
Time (days)
Fig. 3.19: Effect of w/c on drying of concrete.
450
w/c = 0.45
400
w/c = 0.65
350
300
ε sh (microstrains)
250
at 1 cm depth
200
150
at 3 cm depth
100
50
0
0 10 20 30 40 50 60 70 80 90
Time (days)
Fig. 3.20: Effect of w/c on shrinkage of concrete.

65
3.6.3 Effect of aggregate volume fraction
The aggregate volume fraction in concrete affects both moisture diffusion and
drying shrinkage processes of concrete. The effect of aggregate volume fraction on
drying of concrete is shown by Fig. (3.21). The figure shows the variation of relative
humidity with time at two different depths from the exposed surface (1 cm and 3 cm)
for two different aggregate volume fractions (0.55 and 0.75). From the figure, one
can see that adding more aggregate to concrete decreases the rate and the magnitude
of the loss of relative humidity. In general, aggregate has lower diffusion coefficient
than cement paste, which is mainly due to the discontinuity in pore system of
aggregate. Therefore, the addition of more aggregates to concrete reduces its
moisture diffusion coefficient. At the same time, more aggregate in concrete makes
the flow paths longer and more tortuous, which also help to reduce the diffusion
coefficient of concrete.
The effect of aggregate volume fraction on shrinkage of concrete is shown in
Fig. (3.22). The figure shows the time variation of shrinkage strain at two different
depths from the exposed surface (1 and 3 cm) for two different aggregate volume
fractions (0.55 and 0.75). The graph shows that addition of more aggregate to
concrete decreases the rate and the magnitude of the shrinkage of concrete. This can
be attributed to two reasons. One is the reduction of diffusion coefficient of the
concrete, and the other is that aggregates do not shrink.

66
gi = 0.55
0.9 gi = 0.75
Pore relative humidity 0.8

at 3 cm depth
0.7
0.6 at 1 cm depth
0.5
0 10 20 30 40 50 60 70 80 90
Time (days)
Fig. 3.21: Effect of aggregate volume fraction on drying of concrete.
800
gi = 0.55
700 gi = 0.75
600
ε sh (microstains)
500
400
300
at 1 cm depth
200
at 3 cm depth
100
0
0 10 20 30 40 50 60 70 80 90
Time (days)
Fig. 3.22: Effect of aggregate volume fraction on shrinkage of concrete.

67
3.7 Conclusions
1. The interactive effect of moisture diffusion and drying shrinkage in concrete was
studied systemically in this chapter. Multiscale modeling (at micro-, meso- and
macro-scale) was used for modeling the moisture diffusion and drying shrinkage
processes of concrete. The diffusion equation and the drying shrinkage of
concrete are established for a representative volume of element at the macro-
scale. At the meso-scale, both models consider concrete as a two-phase
composite with the aggregate as inclusion and the cement paste as matrix. At the
micro-scale, the influential parameters on microstructure of cement paste, such as
water-cement ratio and curing time, are included.
2. The effect of shrinkage-induced damage is characterized by modifying the two
material parameters in the moisture diffusion equation, i.e. the moisture capacity
and the humidity diffusion coefficient. In this manner, the shrinkage-induced
stress or strain is not shown in the diffusion equation as an explicit driving force,
but as an implicit parameter.
3. Since the microcracks induced by drying shrinkage are randomly distributed
without a preferential orientation, the shrinkage-induced damage may be
considered as isotropic and simply evaluated by the isotropic scalar damage
parameter, d. The effect of d on humidity diffusion coefficient is characterized by
a similar manner as the effect of d on the secant modulus proposed by Kachanov.
4. The effect of drying shrinkage on the moisture capacity is included by an
analytical model based on non-equilibrium thermodynamics and minimum
potential energy principle for a two-phase composite. The influence of the drying
68
shrinkage depends mainly on the difference between the shrinkage coefficients of
the two constituent phases: the aggregate and the cement paste. When the two
shrinkage coefficients are the same, the moisture capacity of the concrete is
simply the volumetric average of the moisture capacities of the two components;
when the two shrinkage coefficients are different, the coupling effect must be
taken into account. With increasing differential shrinkage, the moisture capacity
of concrete increases.
5. A numerical solution for the moisture diffusion equation, taken into account the
coupling effect of the drying shrinkage is developed by using Alternating-
Direction Implicit (ADI) finite difference method. The solution is stable and
convergent.
6. The present model can predict the basic trend for the coupling effect between the
drying shrinkage of concrete and the moisture diffusion in concrete. The
shrinkage-induced damage increases the humidity diffusion coefficient; the
differential shrinkage between the aggregate and the cement paste increases the
moisture capacity of concrete. As a result, the moisture diffusion process in
concrete is accelerated by the drying shrinkage. In turn, the accelerated drying
process increases the shrinkage of concrete.
7. The present model can predict the effects of water-cement ratio and aggregate
volume fraction on the drying shrinkage of concrete, the moisture diffusion and
the coupling effect between the two. Both water-cement ratio and aggregate
volume fraction have major impact on the internal humidity distribution and
shrinkage strain development of concrete.

CHAPTER 4
COUPLED MOISTURE AND CHLORIDE DIFFUSIONS
4.1 Introduction
There are mainly three driving forces of the ingress of chloride ions in non-
saturated concrete. The first driving force is the non-uniform distribution of chloride
ions, which is important for both saturated and non-saturated concrete. The second
driving force is the diffusion of moisture, which is important in the case of non-
saturated concrete. The third driving force is the ionic migration driven by an electric
potential gradient, which is important only in some special cases. The rapid chloride
permeability test in ASTM C1202 and AASHTO T277 involves the effect of the
second and the third driving forces, but not the first one. AASHTO T259-80 (90-day
ponding test) involves the first and the second driving forces, but not the third one. In
reality, the dominant driving forces for chloride penetration in the concrete of
highway pavements and bridge decks are the first and the second ones, i.e. moisture
diffusion and non-uniform chloride distribution.
Many researchers have investigated the diffusion of chloride ions into
concrete. The review of the available literature on this topic shows that there have
been two ways for formulating the chloride penetration in concrete. One is based on
the conventional diffusion theory at the macroscopic level, that is, Fick's first law and
mass conservation (Saetta et al., 1993; Frey et al., 1994; Wee et al., 1997;
70
Swaddiwudhipong et al., 2000). The other is based on the basic laws of
electrochemistry at the nanometer level, i.e. Nernst-Plank equation and Nernst-
Einstein equation (Andrade 1993 and Chatterji 1994). A new chloride penetration
model has recently been developed for saturated concrete (Xi and Bazant, 1999; Xi et
al., 2000) using the multiscale modeling approach (Xi and Jennings, 1997). In this
model, the conventional diffusion theory was used to formulate the diffusion equation
at the macroscopic level (for the so-called representative volume element); and the
composite theory was used at the mesoscopic level to incorporate the effect of
aggregate on chloride diffusion coefficient; and furthermore, the diffusion
mechanisms of chloride ions were taken into account at the microscopic level. The
comprehensive model can be used to simulate and to predict the chloride diffusion in
saturated concrete satisfactorily.
In reality, concrete is often found in a non-saturated condition rather than the
saturated condition. Swaddiwudhipong et al. (2000) developed a mathematical model
to predict the effect of moisture transport on chloride ingress in partially saturated
concrete. Their model includes the effect of moisture transport due to concentration
gradient (diffusion), capillary action (sorption) and pressure gradient (permeability).
Costa and Appleton (1999) presented simplified models for predicting diffusion
coefficient for chloride ions, surface chloride concentration and their dependence on
the exposure time, exposure conditions and concrete properties. Saetta et al. (1993)
analyzed the diffusion of chloride ions in non-saturated concrete. The non-linear and
coupled partial differential equations were solved using the finite element method.
The solution took into account several factors involved in the partial differential
71
equations and their dependence on the concrete mix parameters and the
environmental conditions.
In this chapter, a comprehensive model for chloride penetration in non-
saturated concrete was developed. This work is a continuation and further extension
of the research work of Xi and Bazant (1999) for the chloride penetration in saturated
concrete. The framework of the model is along the line of the multiscale modeling
(Xi et al., 2000). The coupled moisture and chloride diffusion is a complex problem;
it is not uncommon to find mathematical models with too many empirical parameters
to be determined by curve fitting, whereas the real physical meanings of the
parameters are lost. Our strategy is to establish the material models based on
analytical results first, and if not possible, empirical models are introduced based on
dominant physical or chemical mechanism(s) at different scale levels, and then each
individual empirical model is calibrated by related test data. A new numerical
method is used to solve for the coupled partial differential equations of chloride
penetration and moisture diffusion. The prediction of the theoretical model is
compared with available experimental results. Finally, the effect of moisture
diffusion on the chloride penetration is analyzed by numerical examples.
4.2 Basic Formulation
First of all, the governing equations representing the three driving forces for the
chloride diffusion must be established. When chloride ions enter concrete, some of
the chlorides react chemically with the cement paste components (like the reaction
with C3A and C4AF to form Friedels salt) and become chemically bound and some
72
become physically adsorbed to the cement gel or physically trapped in gel pores or
capillary pores. These chlorides are called bound chloride, while the others, called
free chloride, diffuse around in concrete. The sum of the two types of chloride is the
total chloride content in the concrete. Apparently, the steel corrosion is related only
to the free chloride content not to the total chloride content, because the bound
chloride is not free to move and never reaches the surface of embedded steel bars.
Therefore, the proposed governing equation for chloride diffusion is formulated in
terms of the free chloride concentration. Similarly, for the second driving force of the
diffusion process (i.e. the moisture diffusion), the moisture flux can only carry the
free chloride ions not the bound chloride.
The flux of chloride ions (J) through a unit area of porous media in a unit time
can be expressed using Fick’s first law:
J = − Dcl grad (C f ) − Dele grad (φ ) (4.1)
where, Dcl = chloride diffusion coefficient (cm2/day), Cf = free chloride concentration
(in gram of free chloride per gram of concrete, g/g), Dele = ionic diffusion coefficient
and φ = electric potential.
The mass balance of chloride ions can be expressed using Fick’s second law
as follows:
dCt dw
= − div ( J ) + µC f (4.2)
dt dt
73
in which Ct = total chloride concentration (in gram of total chloride per gram of
concrete, g/g); w = moisture content of concrete (in gram of moisture per gram of
concrete, g/g) and µ is a unit conversion factor for converting the unit of the second
term in the right hand side of Eq. (4.2) to gram of chloride per gram of concrete.
µ = 1/(ρsol βsol), where ρsol is the density of pore solution in gram of pore solution per
liter, and βsol is the ratio of the volume of pore solution to the weight of concrete. βsol
will be explained in detail in later sections. By substituting Eq. (4.1) into Eq. (4.2)
and decoupling the free chloride concentration (Cf) from the total chloride
concentration (Ct), the Partial Differential Equation (PDE), that governs the diffusion
of chloride ions in partially saturated concrete, can be written as:
∂Ct ∂Ct ∂C f ∂w
= = div [ DCl grad (C f )] + µ C f + div [ Dele grad (φ )] (4.3)
∂t ∂C f ∂t ∂t
in which, one can see clearly that the first term in the right hand side represents the
diffusion of chloride ions from high concentration to low concentration
(concentration gradient); the second term is for the diffusion of moisture, which
serves as a carrier of the chloride ions; and the third term describes the migration of
chloride ions driven by an electric potential gradient. In non-saturated concrete,
∂w/∂t in the second term of the governing PDE involves the diffusion of moisture,
which is described as following:
∂w ∂w ∂H
= = div [ DH grad ( H )] (4.4)
∂t ∂H ∂t
74
in which, H=Pore relative humidity; ∂w/∂H=Moisture capacity; DH=humidity
diffusion coefficient.
It is important to note that the third term in Eq. (4.3) becomes important only
when a strong external electric potential is imposed on the concrete member, as in the
case of cathodic protection. This term will not be investigated in the present work
and will be dropped from the governing equation, Eq. (4.3). Thus the penetration of
chloride ions in a non-saturated concrete is governed by the coupled system of PDEs:
∂w ∂w ∂H
= = div [ DH grad ( H )] (4.5)
∂t ∂H ∂t
∂C t ∂C t ∂C f ∂w
= = div [ DCl grad (C f )] + µ Cf (4.6)
∂t ∂C f ∂t ∂t
4.3 Material Parameters
In order to develop numerical solutions for the governing equations, Eqs. (4.5)
and (4.6), the material parameters involved in the equations must be determined first.
The material parameters are the moisture capacity (∂w/∂H) and the humidity diffusion
coefficient (DH) in Eq. (4.5), chloride binding capacity (∂Cf/∂Ct) and the chloride
diffusion coefficient (Dcl) in Eq. (4.5). The moisture capacity and the humidity
diffusion coefficient are discussed in the previous chapter (chapter 3). The chloride
binding capacity and the chloride diffusion coefficient will be discussed in this
chapter.
75
4.3.1 Chloride binding capacity
The total chloride concentration, Ct, is the summation of the free chloride
concentration, Cf, and the bound chloride concentration, Cb
Ct = C f + Cb (4.7)
The chloride binding capacity is the ratio between the change of free chloride
concentration and the total chloride concentration. The chloride binding capacity can
be expressed as follows:
dC f 1
= (4.8)
dCt dCb
1+
dC f
The ratio dCb/dCf can be obtained experimentally. Based on available test data, a
prediction model for the binding capacity has been developed based on Freundlich
isotherm (Xi and Bazant, 1999):
dC 1
=
f
A −1
(4.9)
dC t A 10 β C − S − H ⎛
B
Cf ⎞
1+ ⎜ ⎟
35450 β sol ⎜⎝ 35 .45 β sol ⎟
⎠
where, A and B are two material constants related to chloride adsorption and equal to
0.3788 and 1.14, respectively (Tang and Nilson 1993). The binding capacity depends
on the two parameters, βsol and βC-S-H (Xi and Bazant, 1999).
The parameter βsol, as explained for Eq. (4.2), is the ratio of pore solution to
76
concrete (L/g). This parameter reflects the effect of the structure of hydration
products and can be written as:
Vsol wsol n( H , T )
β sol = = = (4.10)
wconc ρ sol wconc ρ sol
in which Vsol and wsol are the volume and the weight of pore solution, wconc is
the weight of concrete, ρsol = density of the pore solution in (g/L) and it depends on
the chloride concentration. For simplicity, ρsol can be used as the density of the pore
water. The weight ratio of pore solution to concrete (wsol/wcoc) is the so-called
chloride adsorption isotherm, which depends on relative humidity H, temperature T
and the pore structure of concrete. Because of the lack of experimental results on the
chloride adsorption isotherms for concrete, the water adsorption isotherm is used
instead. Thus, n(H,T) can be expressed as follows:
n( H , T ) = f cp n cp ( H , T ) + f agg n agg ( H , T ) (4.11)
in which, fcp and fagg are weight percentages of cement paste and aggregates and
ncp(H,T) and nagg(H,T) are the water adsorption isotherms of cement paste and
aggregate. The adsorption isotherm of aggregate and cement paste can be calculated
as shown in Appendix A.
The parameter βC-S-H is the weight ratio of C-S-H gel to concrete (g/g). This
parameter determines the effect of the cement composition and age on the volume
fraction of C-S-H gel and can be written as:

77
wC − S − H
β C −S −H = (4.12)
wtotal
in which wC-S-H and wtotal are the weight of C-S-H gel and the total weight of concrete.
Since there is no difference in the densities of the C-S-H gel and the cement paste, the
parameter βC-S-H is approximately equal to fC-S-H and can be calculated using Eq. (B.2)
of Appendix B.
In Eq. (4.9), one can see that, since the parameter A < 1, ∂Cf/∂Ct = 0 when the
free chloride concentration Cf is zero. For saturated concrete, the second term
vanishes in Eq. (4.6), it is apparent that ∂Cf/∂Ct = 0 leads to ∂Cf/∂ t = 0. This means
that Cf is a constant at all time steps, and it equals to the initial free chloride
concentration, which is zero. This means that the chloride diffusion will never start.
This is the problem when Freundlich isotherm is used when the free chloride
concentration is low. According to Tang and Nilsson (1993), Freundlich isotherm is
valid only when Cf is large (> 0.01 mol/L), and Langmuir isotherm should be used
when Cf is very small (< 0.05 mol/L). Therefore, in the present study, the binding
capacity based on Langmuir isotherm is used for the initial condition with Cf = 0, and
Eq. (4.9) based on Freundlich isotherm is used for Cf > 0. The following is the
derivation of the binding capacity based on Langmuir isotherm.
First of all, Langmuir isotherm is of the form
1 1 1 1
= + (4.13)
Cb′ k ′Cbm C ′f Cbm
78
in which k’ is an adsorption constant, and Cbm is the bound chloride content at
saturated monolayer adsorption (Tang and Nilsson 1993). k’ and Cbm can be
determined by curve fitting based on adsorption test data; C ′b and C ′f are the bound
and free chloride concentrations, respectively, but in a unit system differing from Cb
and Cf. C ′b is in milligrams of bound chloride per gram of calcium silicate hydrate
(C-S-H) gel (mg/g); C ′f is in moles of free chloride per liter of pore solution (mol/L).
The advantage of using such a unit system in Eq. (4.13) is that the resulting Langmuir
isotherm is independent of the water-cement ratio and the aggregate content of
concrete, because it is based on the amounts of C-S-H gel and pore solution.
However, for practical engineering problems, it is necessary to use the unit system
defined for Cb and Cf, because the amounts of C-S-H gel and the pore solution vary
with the hydration process and are not easy to measure in conventional laboratories.
Using the two factors, βsol and βC-S-H, as defined in Eqs. (4.10) and (4.12), the two
different unit systems can be correlated
Cf
C ′f = (4.14)
35.45β sol
1000Cb
C′b = (4.15)
β C −S −H
Substituting Eqs. (4.14) and (4.15) into Eq. (4.13), we obtain
1 1000 ⎡ 35.45β sol 1 1 ⎤

= ⎢ + ⎥ (4.16)
Cb β C − S − H ⎣⎢ k ′Cbm C f Cbm ⎥⎦
79
Eq. (4.16) can be re-written in a simple form as
1
Cb = (4.17)
1
β+
αC f
in which,
k ′Cbm β C − S − H
α= (4.18)
35450 β sol
1000
β= (4.19)
β C − S − H Cbm
Derivative of Eq. (4.17) with respect to Cf gives
dCb 1
= (4.20)
dC f α (C ) 2 ( β + 1 ) 2
αC f
f
then, by substituting Eq. (4.20) into Eq. (4.8), we obtain the binding capacity based
on Langmuir isotherm:
dC f 1 1
= = (4.21)
dCt dCb 1
1+ 1+
dC f 1
α ( βC f + ) 2
α
when the free chloride concentration Cf approaches zero, the binding capacity
approaches to 1/(1+α). α can be calculated from Eq. (4.18). Although α is not a
constant but depends on many parameters, it is definitely a non-zero number.
80
According to the test data of Tang and Nilsson (1993), 1/ Cbm = 0.1849, and 1/(k’Cbm)
= 0.002438, and thus, k’ = 75.841 and Cbm = 5.4083 are used in the numerical model.
4.3.2 Chloride diffusion coefficient
The diffusion coefficient of chloride ions in concrete can be estimated using the
multifactor method as follows:
Dcl = f cl f 2 ( g i ) f 3 ( H ) f 4 (T ) f 5 (C f ) (4.22)
in which f1 is a factor that accounts for the effect of water-cement ratio and curing
time of concrete (t0) on chloride diffusion coefficient. An expression for f1 was
suggested (Xi and Bazant, 1999):
28 − t0 ⎛ 1 ( 28 − t0 ) ⎞⎛ w ⎞
6.55
f cl = +⎜ + ⎟⎜ ⎟ (4.23)
62500 ⎝ 4 300 ⎠⎝ c ⎠
The second factor, f2(gi) is to account for the effect of composite action of the
aggregates and the cement paste on the diffusion coefficient of concrete. This factor
can be formulated using the three phase composite model developed by Christensen
(1979):
⎛ gi ⎞
f1 ( g i ) = Dcp ⎜1 + ⎟ (4.24)
⎜ [1 − g ] / 3 + 1 /[( D / D ) − 1] ⎟
⎝ i agg cp ⎠
in which, gi is the volume fraction of aggregates in the concrete, Dagg and Dcp are the
diffusivities of aggregates and cement paste, respectively. Both Dagg and Dcp can be
calculated using the general model proposed by Martys (1994) for porous media:
81
D=
(
2 1 − (V p − V pc ) ) (V − V pc )
4.2
(4.25)
2 p
S
where, Vp is the porosity, S is the surface area and V pc is the critical porosity (the
porosity at which the pore space is first percolated). When Eq. (4.25) is used for the
diffusion coefficient of cement paste Dcp, Vp, S, and V pc in the equation should be the
parameters for cement paste. The critical porosity V pc may be taken as 3% for cement
paste (Martys 1994). The porosity of cement paste Vp can be approximately
estimated by using the water content at saturation, that is, the value of the adsorption
isotherm at H = 100%, ncp(H=1,T), which can be calculated from Eq. (A.1) and
Eqs.(A3) through (A.5) in Appendix A. One must keep in mind that the water
content such calculated is a weight fraction, which must then be converted into the
volume fraction, Vp = BSGcp*ncp(H=1,T), where BSGcp is the bulk specific gravity of
the cement paste (the estimated value for BSGcp ≈ 2.5). The surface area S of cement
paste is proportional to the monolayer capacity Vm and it can be estimated from Eq
(A.3) of Appendix A. The diffusion coefficient of aggregates Dagg can also be
calculated by Eq. (4.28) if the values for the porosity Vp, surface area S and critical
porosity V pc are known; or Dagg can be simply taken as a constant, and a suggested
value is 1x10-12 cm2/sec. The diffusion coefficient of aggregates Dagg can be
calculated using equation (4.25) and knowing the values for the porosity Vp, surface
area S and critical porosity V pc or simply may be taken as constant, a suggested value
is 1x10-12 cm/sec.
82
The third factor, f3(H) is to account for the effect of relative humidity on the
chloride diffusion coefficient. A model proposed by Bazant and Najjar (1972) can be
used, which was developed initially for moisture diffusion. In this work assuming a
full analogy between the moisture and chloride diffusion, the model is used:
−1
⎛ (1 − H ) 4 ⎞
f3 ( H ) = ⎜⎜1 + ⎟
4 ⎟
(4.26)
⎝ (1 − H C ) ⎠
in which Hc is the critical humidity level at which the diffusion coefficient drops
halfway between its maximum and minimum values (Hc = 0.75). Eq. (4.26) is very
important in terms of characterizing the coupling between the moisture diffusion and
the chloride penetration. Numerical examples will be shown later for detailed
analyses.
The fourth factor, f4(T) is to account for effect of temperature on the diffusion
coefficient of concrete. The temperature effect can be incorporated using Arrhenius’
law
⎡U ⎛ 1 1 ⎞ ⎤
f 4 (T ) = exp ⎢ ⎜⎜ − ⎟⎟ ⎥ (4.27)
⎣⎢ R ⎝ T0 T ⎠ ⎦⎥
in which U = the activation energy of the diffusion process, R = gas constant (8.314 J
mol-1K-1), T and T0 are the current and reference temperatures, respectively, in Kelvin
(T0 = 296 K). It was found that the activation energy of the diffusion process
depends on the porosity and cement type (Page et al., 1981). The values of the
activation energy, U, for a cement paste made of ordinary Portland cement are
41.8±4.0 (Kj/mol) for w/c of 0.4, 44.6±4.3 (Kj/mol) for w/c of 0.5 and
83
32.0±2.4 (Kj/mol) for w/c of 0.6.
The fifth factor, f5(Cf) is to account for the dependence of the chloride
diffusion coefficient on the free chloride concentration. For the diffusion of ions,
especially diffusion of chloride ions in the present study, the movement of ions is
restricted by the electrostatic field induced by the other ions present in the solution
f 5 (C f ) = 1 − k ion (C f ) m (4.28)
in which kion and m are two constants. kion and m were calibrated by Xi and Bazant
(1999) kion = 8.333 and m = 0.5.
4.4 Numerical Solution
The non-linearity of the material parameters involved and the coupling between
the chloride diffusion and the moisture diffusion in the governing equation, Eq. (4.6),
make the numerical solution favorable. In the present study the Alternating-Direction
Implicit (ADI) finite difference method developed by Peaceman and Rachford was
employed for solving the moisture and chloride diffusion problem (Carnahan et al.
1969). A two-dimensional finite difference mesh as shown in Fig. (4.1) was used in
this study. The free chloride concentration, Cf(i,j,n), at different points of the mesh at
different time intervals are estimated.
In general, the following partial differential equation was solved in two
consecutive steps.
84
df ( x, y, t )
dt
[( ) ]
= a( f , x, y, t ) ∇ D( f , x, y, t ) ∇ f ( x, y, t ) + µb( x, y, t ) f ( x, y, t ) (4.29)
Exposed Surface Chloride Solution

Boundary Condition
Initial Condition
Hini and Cfini 10cm
Width=30cm
Boundary Conditions
Henv and Cfenv
Fig. 4.1: Concrete slab used in the numerical solution.
As the first step, Eq. (4.29) is solved for the moisture diffusion problem, Eq.
(4.5), with a(f,x,y,t)=dH/dw, D(f,x,y,t) = DH, and b(x,y,t) =0. In the second step, Eq.
(4.29) is solved for the chloride penetration problem, Eq. (4.6), with f(x,y,t)=Cf(x,y,t),
a(x,y,t)=dCf/dCt,, D(f,x,y,t)=Dcl and b(x,y,t)=dw/dt.
For a two-dimensional finite difference mesh, see Fig. (4.1), Eq. (4.29), can be
rewritten as follows:
df ⎡ ∂ 2 f ∂D ∂f ∂ 2 f ∂D ∂f ⎤
= a ∗ ⎢D ∗ 2 + ∗ + D∗ 2 + ∗ + µb ∗ f ⎥ (4.30)
dt ⎣ ∂x ∂x ∂x ∂y ∂y ∂y ⎦
The principle of Alternating-Direction Implicit (ADI) Method can be applied to solve
Eq. (4.30) by using two difference equations over two successive time-steps each of
duration ∆t/2:
85
fi*,j − fi ,nj
∆t / 2
[
= ain, j * Din, jδ x2 fi*,j + δ x Din, jδ x fi*,j + Din, jδ y2 fi ,nj + δ y Din, jδ y fi ,nj + bin, j fi ,nj ] (4.31)
fi ,nj+1 − fi*,j
∆t / 2
[
= ain, j * Din, jδ x2 fi *,j + δ x Din, jδ x fi *,j + Din, jδ y2 fi ,nj+1 + δ y Din, jδ y fi ,nj+1 + bin, j fi *,j ] (4.32)
in which, δx, δ2x and δy, δ2y are the first and second order central-difference operators
in x and y directions, respectively. Eq. (4.31) is implicit only in x-direction and used
to solve for f i *,j (intermediate value at the end of the first time-step) by knowing
f i ,nj starting from the initial condition. Eq. (4.32) is implicit only in y-direction and
used to solve for f i ,nj+1 (final value at the end of the whole time step) after calculating
f i *,j in the first step.
4.5 Numerical Applications and Discussions
4.5.1 Numerical example of chloride penetration in non-saturated concrete
The numerical solution algorithm described in the previous section solving for
the governing partial differential equation is implemented in a computer program,
which is used to solve for the diffusion of chloride ions in partially saturated concrete
slabs. As a numerical example, a 10 cm depth concrete slab similar to the slab shown
in Fig. (4.1) was exposed to 5% chloride solution (8.24% sodium chloride solution)
and 100% relative humidity on the upper. The water-cement ratio of the concrete is
0.6 and the curing period is 28 days. The initial free chloride concentration in the
concrete slab Cfini = 0 and the initial relative humidity in the concrete Hini = 50%.
86
The variation of free chloride concentration with the depth at different times
of exposure is shown in Fig. (4.2). The depth of penetration is measured from the
exposed surface of the concrete slab. Fig. (4.2), shows that the free chloride
concentration decreases with the increases of the depth from the exposed surface.
The variation of free chloride concentration with the time of exposure at different
depths is shown in Fig. (4.3). The time of exposure starts right after the curing period.
Fig. (4.3) shows that at a fixed depth inside of the concrete slab, the free chloride
concentration increases with the increase of the time of exposure.
-3
x 10
7
Time = 1 month
6
= 3 months
Free chloride concentration, C f (g/g)
= 6 months
= 1 year
5
= 2 years
= 3 years
4
0
0 1 2 3 4 5 6 7 8 9 10
Depth from exposed surface (cm)
Fig. 4.2: Depth profile of free chloride concentration.

87
-3
x 10
7
Depth = 1 cm
= 2 cm
6
= 3 cm
Free chloride concentration, C f , (g/g)

= 4 cm
5 = 5 cm
0
0 200 400 600 800 1000 1200
Exposure time (days)
Fig. 4.3: Time profile of free chloride concentration.
In practice, penetration front of free chloride has often been used, which is
defined as the depth in the concrete where the free chloride concentration equals to a
predetermined value. The location of the penetration front can be plotted against the
time of exposure. In this way, the movement of the penetration front can be used as
an indicator (similar to a tracer), that follows the movement of the chloride from the
exposed surface to inner part of the concrete. Either Fig. (4.2) or Fig. (4.3) can be
used to predict the location of penetration front for a fixed free chloride
concentration. As an example, a penetration front of 0.06% (g/g) free chloride
concentration is reached approximately to the depth of 2 cm after one month of
exposure; 2.9 cm after three months, 3.7 cm after six months, 4.8 cm after 1 year, 6.4
88
cm after 2 years, and 8 cm after 3 years of exposure. If a reinforcing steel bar is
embedded at the depth of 4 cm, and 0.06% free chloride concentration is the critical
chloride concentration that triggers the onset of steel corrosion, then the relationship
(among the penetration front, time, and depth) can be used to predict the starting point
of the steel corrosion as well as the development of the corrosion process. Therefore,
it provides very useful information about the service life of a concrete structure and
the time needed for rehabilitation.
4.5.2 Comparison with test data
The experimental results obtained from the 90-day ponding test (Andrade and
Whiting 1996) for a concrete slab similar to the concrete slab shown in Fig. (4.1), is
compared with the numerical results obtained in the present study, see Fig. (4.4).
Two types of concrete with different water-cement ratios of 0.4 and 0.6 are compared.
One can see from Fig. (4.4) that the predicted results by the present model agree very
well with the test data for the chloride penetration into partially saturated concrete.
4.5.3 Effect of the moisture diffusion on chloride penetration
The effect of moisture diffusion becomes very important in partially saturated
concrete. To study the effect of the moisture diffusion on the diffusion of chloride
ions, a concrete slab of 10 cm thick, as shown in Fig. (4.1), was exposed to 5%
chloride solution (8.24% sodium chloride solution) on the top surface. The chloride
diffusion in the concrete slab is analyzed for two different moisture conditions. In
one of the cases, the moisture boundary condition H = 100% on the top surface, the
89
0.007
0.006 w/c = 0.4 (Experimental)
t (g/g)
w/c = 0.6
Total chloride concentration, C 0.005 w/c = 0.4 (Present)
w/c = 0.6
0.004
0.003
0.002
0.001
0
0 1 2 3 4 5
Fig. 4.4: Comparison with the experimental results.
initial pore relative humidity inside the slab is 50%, and therefore, the moisture
diffuses from the top of the slab into the concrete, the same direction as the chloride
diffusion. In the other case, the concrete is fully saturated, and thus, there is only
chloride diffusion due to the concentration gradient, and there is no moisture
diffusion occurs. The plots of depth versus free chloride concentration after 1 month,
6 months and 6 years of exposure for the two cases are shown in Fig. (4.5). The plots
of depth versus pore relative humidity after 1 month, 6 months and 6 years of
exposure for the two cases are shown in Fig. (4.6). One can see from Fig. (4.5) that, at
any fixed depth, the free chloride concentration is greater in the partially saturated
concrete (with 50% initial humidity condition). This is due to the coupled moisture
diffusion and chloride penetration, which is shown in the second term of the right
90
hand side of Eq. (4.6). So, the coupling effect contributes significantly to the chloride
penetration in the partially saturated concrete. The basic trend of the coupling effect
predicted by the present numerical model agrees very well with the test data of
chapter 7. Fig. (4.6) also shows that the contribution of the moisture diffusion to the
chloride penetration becomes less significant as the time of exposure increases. This
is due to that the fact that, the concrete becomes fully saturated after long-term
exposure and so there is no moisture exchange inside the concrete, see Fig. (4.6).
One can see from Fig. (4.6) that, after the long-term exposure of six year, the partially
saturated concrete becomes saturated, and the free chloride concentrations are the
same for both cases, see Fig. (4.5).
-3
x 10
7
P artially saturated, H in i = 50%

6
Fully saturated, H = 100%
4
6 years
3
1 month
2
6 months
0
0 1 2 3 4 5 6 7 8 9 10
Fig. 4.5: Effect of the moisture diffusion on free chloride penetration.

91
0.95
0.9
0.85
P ore relative humidity, H
0.8
0.75
0.7
0.65
0.6 Time = 1 month

= 6 months
0.55 = 6 years
0.5
0 1 2 3 4 5 6 7 8 9 10
Fig. 4.6: Moisture diffusion in non-saturated concrete.
4.5.4 The Effect of concrete mix parameters
The material models developed for chloride and humidity diffusion
coefficients can handle the effect of concrete mix parameters, such as water-cement
ratio and volume fraction of aggregate. Fig. (4.7) shows the profiles of the free
chloride concentration after 1 year of exposure for a concrete slab made with 70%
aggregate volume fraction and three different water-cement ratios, 0.5, 0.55 and 0.6.
The concrete slab is 10 cm thick and exposed to 5% chloride solution from the top
surface as shown in Fig. (4.1). Fig. (4.7) shows that the higher the water-cement ratio
the higher the free chloride concentration (at a fixed depth and fixed time). This is
due to the fact that both chloride and humidity diffusion coefficients are increased as
water-cement ratio increases.

92
-3
x 10
7
6
w/c = 0.50
= 0.55
= 0.60
5
0
0 1 2 3 4 5 6 7
Fig. 4.7: Effect of water-cement ratio on chloride penetration.
Figs. (4.8) and (4.9) show the chloride and the humidity diffusion coefficients
as functions of water-cement ratio, the higher the water-cement ratio, the higher the
diffusion coefficients. This is because the porosity of cement paste is higher with
higher water-cement ratios, and the connectivity of the pore system is also increased
which results in a reduction in the resistance to the diffusions of both moisture and
chloride ions.
93
-1 0
x 10
1
0.9 w/c = 0.50
Chloride diffusion coefficent, D cl (m /sec)

= 0.55
0.8 = 0.60
2
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
Fig. 4.8: Effect of water cement ratio on the chloride diffusion coefficient.
-1 0
x 10
1.2
w/c = 0.4
Humidity diffusion coefficent, D H (m /sec)
1 = 0.5
= 0.6
2
0.8
0.6
0.4
0.2
0
0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
Fig. 4.9: Effect of water-cement ratio on humidity diffusion coefficient.

94
The effect of the volume fraction of aggregates on the free chloride
concentration is shown in Fig. (4.10). The same concrete slab in the previous example
was used, but the water-cement ratio is fixed as 0.6 and the volume fraction of
aggregate varies, 0.5, 0.6 and 0.7. The effects of the volume fraction of aggregate on
the chloride and humidity diffusion coefficients are shown in Figs. (4.11) and (4.12),
in which both chloride and humidity diffusion coefficients decrease with the increase
of aggregate volume fraction. This is resulted from the low diffusivity of aggregate.
0.01
0.009 gi = 0.50
= 0.60
0.008
= 0.70
0.007
0.006
0.005
0.004
0.003
0.002
0.001
0
0 1 2 3 4 5 6 7 8 9 10
Fig. 4.10: Effect of aggregate volume fraction on the chloride penetration.

95
-1 1
x 10
5
4.5 gi = 0.50
= 0.60
Chloride diffusion coefficent, D cl (m /sec)

4 = 0.70
2
3.5
2.5
1.5
0.5
0
0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
Fig. 4.11: Effect of aggregate volume fraction on the chloride diffusion coefficient.
-1 0
x 10
2.4
2.1
Humidity diffusion coefficent, D H (m /sec)
gi = 0.5
2
1.8 = 0.6
= 0.7
1.5
1.2
0.9
0.6
0.3
0
0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
Fig. 4.12: Effect of aggregate volume fraction on humidity diffusion coefficient.

96
4.6 Conclusions
1. The multiscale modeling of material parameters involved in the coupled moisture
and chloride diffusion equations results in a robust mathematical model for
characterizing chloride ion ingress into non-saturated concrete. There are four
materials parameters in the model. Moisture diffusion coefficient and moisture
capacity are involved in the moisture diffusion equation, and chloride diffusion
coefficient and binding capacity are involved in the chloride penetration equation.
Four material models are developed for the four material parameters based on
different diffusion mechanisms and their dependence on the environmental
conditions and concrete properties. Each material model is calibrated based on
related test data, which ensure that each model parameter has distinct physical
meaning, not just a results from curve fitting.
2. The model simulates satisfactorily the chloride profiles obtained by the 90-day
ponding test (AASHTO T259–90). The model predictions agree with available
test data very well.
3. The model predicts very well the effect of moisture diffusion on the chloride ion
penetration in partially saturated concrete. The numerical results show that the
moisture diffusion accelerates the chloride penetration.
4. The material models incorporate the effects of various concrete mix parameters
(such as volume fraction of aggregate and w/c ratio) on the diffusion of chloride
ions into partially saturated concrete.
5. The model can be used to predict the onset of steel corrosion in reinforced
concrete structure and thus the service life and the time for rehabilitation.
97
6. The Alternating-Direction Implicit (ADI) finite difference method is used as an
effective tool for solving the coupled partial differential equations. A two-
dimensional finite difference program is developed.

CHAPTER 5
THE COUPLED EFFECT OF FREEZING-THAWING AND
CHLORIDE PENETRATION IN CONCRETE
5.1 Introduction
One of the major effects of freeze-thaw cycles on the deterioration of concrete
is the volume expansion associated with ice formation in the pores of concrete
subjected to freezing and thawing cycles (Philleo 1986). The ice formed in the pores
reduces the effective diffusion coefficient of the concrete, however, excessive volume
expansion due to ice formation results in progressive damage and cracking in
concrete, and the cracks accelerate the diffusion of moisture and chloride ions in
concrete.
The effect of freezing and thawing on concrete durability has been studied
theoretically and experimentally. Mohamed et al. (2000) conducted an experimental
study on the effect of cement type and air-entrainment on freezing-thawing durability
of concrete. Also, they studied the effect of damage caused by freezing and thawing
cycles through the degradation in the compressive strength of concrete. In their study,
they used the benchmark test setup prepared by the Portland Cement Association in
1940. In this test 28 different types of cement were used in a parapet wall panels as
part of Green Mountain Dam (Colorado, USA).

99
Zuber et al. (2000) conducted a theoretical and experimental study on the ice
formation mechanisms in normal and high performance concrete. Theoretically, they
modeled the ice formation mechanisms in the pores of concrete based on the
thermodynamic equilibrium (Gibbs-Duhem equation) between the solid and liquid
phases. Experimentally, the low-temperature calorimetry technique used to
investigate the ice formation in concrete mixes with different water-to-cement ratios.
The test results showed that reducing the water-to-cement ratio resulted in the
refinement of the pore structure and thus reduce the ice formation. Both numerical
modeling and experiments showed that ice formation mechanisms were controlled by
ice front propagation. Also, they concluded that it is difficult to totally impede the ice
formation by enhancing the performance of concrete.
Scherer (1999) studied the mechanism of crystallization of ice and the stresses
development on the walls of the pore network of concrete. Those stresses are
controlled by the pore size distribution, the interfacial energy between the pore wall
and the growing crystal and the yield stress or buckling strength of the crystal. The
crystallization in the pores of concrete was studied based on the principles of
thermodynamics. One of the main conclusions was that cracking is not due to
crystallization in a single pore rather it is due to crystallization in a region large
enough compared to a size of strength-controlling flaws.
Bazant et al. (1988) developed a complete mathematical model for freezing-
thawing durability of concrete. The model included the formulation of sorption
isotherm for concrete below 0 oC based on the well known sorption isotherm for
100
concrete above 0 oC; the formulation of the water diffusion equation in concrete in the
presence of air-entrained bubbles; the calculation of temperature variations in
concrete taking into consideration the latent heat of freezing, and prediction of the
stress produced in concrete due to freezing of pore water.
Saetta et al. (1998) studied the coupled effect of moisture, heat and pollutant
(e.g., Cl-, (SO4)-2 and CO2) flows through reinforced concrete structures. They
introduced two damage parameters. One is the mechanical damage parameter and it is
related to the degradation of the material due to mechanical actions such as
deformations, applied loading, fatigue, drying shrinkage and thermal expansion. And
the other is the chemical damage parameter and it is related to the degradation of the
mechanical strength of the material due to the chemical reaction between the
attacking chemical and the cementitious material, and the reinforcement corrosion.
In this chapter, an experimental study on the coupled effect of freezing–thawing
and chloride penetration on concrete was conducted. Two sets of concrete mixes were
used in this study. In one of the sets, the effect of water-cement ratio was studied,
while the effect of aggregate volume fraction was studied in the other set. The
experimental study resulted in the damage values and chloride profiles for especially
designed concrete specimen exposed to simultaneous action of cyclic freezing-
thawing and continuous chloride penetration. The obtained experimental results can
be used for validation and calibration of numerical results obtained by the
mathematical models.
101
5.2 Ultrasonic Testing of Materials
The experimental program in this chapter involved the assessment of damage in
concrete under the coupled action of freezing-thawing and chloride penetration. The
damage was evaluated using two non-destructive testing techniques, the pulse
velocity and the forced resonance methods. In this section, a brief review about the
ultrasonic testing of materials will be provided.
Ultrasonic testing methods are widely used in the quality assessment of
various materials used in civil engineering structures, including metallic, plastics,
rubber, wood, concrete and cement-based materials. The ultrasonic methods have
many advantages over the traditional testing methods. In addition to their non-
destructive nature, ultrasonic methods can be applied to in-service inspections and for
early detection of internal flaws and defects inside the material, which can be used as
an indicator for routine maintenance, retrofitting or rehabilitation of the structures.
Some characteristics of the ultrasonic pulse waves depend on the density,
elastic properties and the existence of air voids and cracks inside the material. Those
characteristics include the pulse velocity, resonant frequency, pulse amplitude and
frequency content. One of the main drawbacks on the pulse velocity method is the
small variation in the pulse velocity with the variation of concrete material
parameters, such as water-cement ratio and aggregate volume fraction, which may
lead to incorrect prediction of the concrete strength.
In this study, two indicators were estimated for concrete damage resulted from
102
the degradation of the initial dynamic modulus of elasticity due to cyclic freezing and
thawing. One was calculated based on the velocity of the compressive pulse wave
traveling through the concrete sample. The other was calculated based on the
fundamental resonant frequency of longitudinal wave. The damage factor based on
the degradation of the initial dynamic modulus of elasticity can be written as follows:
E ddam
d = 1− (5.1)
E do
where, E do = initial or undamaged dynamic elastic modulus and E ddam = dynamic elastic
modulus of the damaged concrete.
The dynamic modulus of elasticity of concrete can be calculated in terms of the
transmitted ultrasonic pulse velocity, V of compression wave as follows:
(1 + µ )(1 − 2 µ )
E d = ρV 2 (5.2)
(1 − µ )
where, ρ is the material density and µ is the dynamic Poisson’s ratio.
On the other hand, the fundamental resonant frequency (fL in Hz) of longitudinal wave
traveling through a concrete sample of prismatic cross-sectional area can be used to
estimate the dynamic modulus of elasticity (in Pa) as follows:
E d = 4 ρLf L
2
(5.3)
where, ρ is the material density in (Kg/m3) and L is the length of specimen in (m).
103
By substituting Eq. (5.2) in Eq. (5.1) and neglecting the effect of damage on the
Poisson’s ratio (µ) and material density (ρ), the damage factor based on the ultrasonic
pulse velocity can be written as:
2
⎛ V dam ⎞
d = 1− ⎜ o ⎟ (5.4)
⎜ V ⎟
⎝ ⎠
Since, the V-meter (Pulse Velocity Tester) measures the transit time (t in
microseconds), and the pulse velocity, V can be written in terms of transit time (t) and
the length of the sample (L) as follows:
L
V = (5.5)
t
The damage factor can be written in terms of the transit time by substituting Eq. (5.5)
into Eq. (5.4) as follows:
2
⎛ to ⎞
d = 1 − ⎜ dam ⎟ (5.6)
⎜t ⎟
⎝ ⎠
where, to is the initial transit time and tdam is the transit time of the pulse wave
measured after damage.
The damage factor based on the fundamental resonant frequency of longitudinal pulse
wave (fL in Hz) by substituting Eq. (5.3) in Eq. (5.1) as follows:

104
2
⎛ f dam ⎞
d = 1 − ⎜⎜ L 0 ⎟⎟ (5.7)
⎝ fL ⎠
5.3 Experimental Program

5.3.1 Preparation of test specimens
In order to study the effect of freezing and thawing on the chloride penetration,
a specially designed concrete specimen was used for this purpose. As shown in Fig.
(5.1), the concrete specimen is 40x13x13 cm in dimension. The dimensions of the
specimen were modified from the dimensions of the specimen used in the previous
experiment (30x15x15 cm). The reason for the modification was to maintain the
length to maximum width ratio between 3 and 5. Specimens having the length to
maximum width ratio outside this range are difficult to excite for the fundamental
mode of vibration.
Fig. 5.1: Specimen for coupled freezing-thawing and chloride penetration.
Two sets of specimens were used in this experiment. One set has an aggregate
105
volume fraction of 0.65 and three different water-cement ratios of 0.45, 0.55 and
0.65. The other set has a water-cement ratio of 0.55, and three different aggregate
volume fractions of 0.55, 0.65 and 0.75. The same concrete mixes used in the
experimental study of chapter 6 (see section 6.2) were used in the current study. The
mix proportions and the compression test results are shown in Table (C.2) of
Appendix C and Table (D.1) of Appendix D. The specimens were casted, remolded
after 24 hours and stored in the moisture room for 28 days. No air-entraining agent
was used in the concrete mixes.
5.3.2 Cyclic freezing-thawing and chloride penetration
After curing, all specimens were exposed to 8% sodium chloride (NaCl)
solution on the top surface and placed in an environmental chamber with
programmable freezing and thawing cycles as shown in Fig. (5.2).
Fig. 5.2: Placing of concrete samples in the environmental chamber.

106
25
15
Temperature ( C )
o
-5 0 1 2 3 4 5 6
-15
-25
Time (hour)
Fig. 5.3: The freezing and thawing temperature cycle.
5.3.3 Measurement of Transit time and fundamental resonant frequency
The setup for the V-meter (Pulse Velocity Tester) and the concrete sample is
shown in Fig. (5.4). This setup was used to measure the time required for the earliest
part of longitudinal (compression wave) pulse traveling through the concrete sample
from transmitting transducer to receiving transducer. This time is called the transit
time and it is measured in microseconds.
The other method was by measuring the fundamental resonant frequency (in
Hz) of a longitudinal wave traveling through concrete sent by a vibrator on one side
of the specimen and received by the pick-up on the other side. The E-meter (Resonant
Frequency Tester) and the concrete sample setup are shown in Fig. (5.5).
107
Fig. 5.4: Transit time of measurement setup (V-meter).
Fig. 5.5: Setup for measurement of resonant frequency (E-meter)
5.3.4 Chemical analysis for total chloride concentration
After exposed to freezing-thawing and chloride penetration, the total chloride
profiles for concrete specimens were obtained following the experimental procedure
described in Appendix E. The m-Volt readings resulted from the electrochemical
analysis are shown in Table (E.1) of Appendix E. The mV readings of the known
chloride concentration solutions (0.005%, 0.02%, 0.05% and 0.5%) were taken and
108
shown in Table (E.2). The calibration curve for standard concentration solutions is
shown in Fig. (E.4).
5.4 Test Results and Discussion
5.4.1 Cracking patterns and visual inspection
The cracking patterns for the concrete samples after exposed to 400 temperature
cycles of freezing and thawing are shown in Figs. (5.6) through (5.10). Fig. (5.6)
shows the cracking pattern for the top, bottom and the two sides of the concrete
specimen made of mix 46 (0.45 water-cement ratio and 0.65 aggregate volume
fraction). From visual inspection, large number of cracks was noticed but the width
was very small (0.4 mm). There were no cracks on the top surface of the sample
(underneath the chloride solution).
The cracking patterns of the sample made of the concrete mix 56 (have a 0.55
water-cement ratio and 0.65 aggregate volume fraction) are shown in Fig. (5.7). Few
cracks were noticed on the surfaces of the specimen, but their width was larger when
compared to the previous sample (made of mix 46). The average width of the crack
was 1.2 mm and there were no cracks on the top surface of the sample (surface
exposed to chloride solution).

109
(a) (b)
(c) (d)
Fig. 5.6: The concrete sample (46) after it was exposed to the freezing-thawing, a) top
of the sample b) bottom of the sample and c and d) the two sides of the sample.
(a) (b)
(c) (d)
Deep cracks were noticed in the concrete sample prepared from mix 66
(water-cement ratio is 0.65 and aggregate volume fraction is 0.65) as shown in Fig.
(5.8). There were a lot of cracks on the top surface of the sample (surface exposed to
110
chloride solution). The average width of the cracks was 2 mm.
The concrete sample shown in Fig. (5.9) was made of mix 55 in which the
water-cement ratio is 0.55 and the aggregate volume fraction is 0.55. From the
observation of the crack patterns, it was noticed that the cracks are deep and had an
average width of 1.4 mm. Also, large number of cracks was observed on the top
surface of the concrete sample.
As shown in Fig. (5.10), few but deeper and wider cracks were noticed on the
surfaces of concrete sample made of mix 57 (aggregate volume fraction=0.75 and
water-cement ratio=0.55). The average width of the cracks was a 2.6 mm. Few
number of cracks were noticed on the top surface underneath the chloride solution.
(a) (b)
(c) (d)
111
(a) (b)
(c) (d)
(a) (b)
(c) (d)
Fig. 5.10: The concrete sample (57) after it was exposed to the freezing-thawing, a)
top of the sample b) bottom of the sample and c and d) the two sides of the sample.
5.4.2 Ultrasonic test results
The measured transit times of the compression pulse wave and fundamental
112
resonant frequencies of longitudinal wave before and after the samples were exposed
to the freezing and thawing cycles are presented in Table (5.1). The damage factor
can be calculated either by Eq. (5.6) using the measured transit times or by Eq. (5.7)
using the measured fundamental resonant frequencies. It is clear that 400 cycles of
freezing and thawing coupled with chloride penetration resulted in a significant
damage in the concrete, and the damage evaluated by the two methods showed
consistent results. From the comparison of damage factors, as shown in Table (5.1),
the concrete samples prepared from mixes 66, 55 and 57 have the most severe
damage among the five samples. Also, during freezing and thawing of the specimens
while monitoring the test and adding chloride solution, it was noticed the cracking
started earlier and propagated faster in those three samples (samples 66, 55 and 57).
In mix 66, the water-cement ratio was 0.65 and it was the highest among the
three mixes 46, 56 and 66 (see Table C.2 in Appendix C). Increasing water-cement
ratio will introduce more capillary pore and coarsen the pore structure of concrete
which increase the amount of ice formed and accelerate the propagation of ice front.
This effect when combined with the decrease in the strength of concrete resulted in
larger damage as shown by the test results.
The effect of aggregate can be evaluated using the present test data. In general,
aggregate has lower permeability than cement paste, therefore, more aggregate in
concrete reduces overall permeability of the concrete. On the other hand, aggregate
has much higher freezing-thawing resistance than cement paste, therefore, more
aggregate in concrete enhances freezing-thawing resistance of the concrete. With

113
these two general criteria in mind, we can see from the present test data that
decreasing the aggregate volume fraction (mix 55 compared to mix 56) resulted in
high permeability; and increase in aggregate volume fraction (mix 57 compared by
mix 56) resulted in a decrease in the permeability.
There is another effect caused by the aggregate that may play a role in the
permeability of concrete. With more aggregate added in the concrete mix, there is a
significant increase in interface transition zone (ITZ) between the cement paste and
aggregate. ITZ has more porosity than the bulk cement paste. So, the increased
porosity (leading to high permeability) may counterbalance the effect of low
permeability of the aggregate.
Table 5.1: Ultrasonic test results.
V-meter E-meter
Mix Transit time (microsec.) Resonant frequency (Hz)
damage damage
before after (%) before after (%)
46 82.4 144.5 67 5116 2537 75

56 86.6 118.3 46 4831 2636 70
66 86.3 214.8 84 4782 983 96
55 84.7 216.8 85 4758 451 99
57 81.7 188.6 81 5026 469 99
5.4.3 Chloride penetration test results
The calibration curve shown in Fig. (E.4) in Appendix E was used to convert
the mVolt readings to total chloride concentration as shown in Table (5.2). The table
presents the values of the total chloride concentration (as a percentage of concrete
114
weight) at different depth ranges (from the exposed surface) for concrete samples
made of different concrete mixes exposed to 400 cycles of freezing and thawing.
Each cycle continues for 6 hours and so the total exposure time is 100 days.
Table 5.2: Total chloride concentration as percentage of weight of concrete.
Depth range (mm)

Mix Sample #
0-6 6-13 13-19 19-25 25-32 32-38 38-45
1 2.013 0.533 0.388 0.519 0.444 0.386 0.275
2 1.732 0.525 0.375 0.499 0.422 0.384 0.275
46 3 1.712 0.522 0.361 0.493 0.420 0.380 0.254
average 1.819 0.527 0.375 0.504 0.428 0.383 0.268
1 2.024 1.165 0.834 0.652 0.542 0.505 0.493
2 1.904 1.127 0.824 0.634 0.531 0.485 0.483
56 3 1.904 1.078 0.820 0.627 0.525 0.480 0.472
average 1.944 1.123 0.826 0.638 0.533 0.490 0.483
1 2.141 0.948 0.705 0.539 - - -
2 2.082 0.896 0.701 0.488 - - -
66 3 1.958 0.862 0.645 0.472 - - -
average 2.060 0.902 0.684 0.500 - - -
1 2.201 1.036 0.857 0.758 0.701 0.536 -
2 2.165 1.013 0.793 0.733 0.656 0.525 -
55 3 2.070 1.002 0.775 0.690 0.610 0.522 -
average 2.145 1.017 0.808 0.727 0.656 0.528 -
1 1.893 0.754 0.390 - - - -
2 1.821 0.610 0.386 - - - -
57 3 1.751 0.593 0.375 - - - -
average 1.822 0.652 0.384 - - - -
The chloride profiles obtained in this study are compared with the profiles
obtained in the experimental study of chapter 6 which was conducted at room
temperature and for the same exposure period (100 days). The comparisons are
115
shown in Figs. (5.10) through (5.20). One can see that the chloride ions in the
concrete experienced freeze-thaw cycles penetrate in a higher rate, which means that
the progressive damage produced by the cyclic freezing and thawing increases the
chloride penetration. The chloride profiles for damaged (due to cyclic freezing-
thawing) and undamaged specimens made of mixes 46, 56 and 55 shown in Figs.
(5.11) through (5.14) present that the concentrations in the damaged concrete is
approximately doubled in the depth range of 0-19 mm and the concentrations are
much higher in the depth range from 19-44 mm. Fig. (5.15), shows that even though
specimen 57 experienced severe damage, the chloride concentration did not increase
significantly. This is due to the high content of aggregate, which decreases diffusivity
and thus impedes chloride penetration.

116
2.00
Chloride conc. (% by concrete weight)

1.75 At room temperature
1.50 After freezing and thawing
1.25
1.00
0.75
0.50
0.25
0.00
0-6 6-13 13-19 19-25 25-32 32-38 38-45
Depth range (mm)
Fig. 5.11: Chloride profiles for the concrete specimens made of mix 46
2.00
1.25
1.00
0.75
0.50
0.25
0.00
0-6 6-13 13-19 19-25 25-32 32-38 38-45
Depth range (mm)
117
2.25


1.50
1.25
1.00
0.75
0.50
0.25
0.00
0-6 6-13 13-19 19-25
Depth range (mm)
2.25

1.50
1.25
1.00
0.75
0.50
0.25
0.00
0-6 6-13 13-19 19-25 25-32 32-38
Depth range (mm)
118
2.00

1.25
1.00
0.75
0.50
0.25
0.00
0-6 6-13 13-19
Depth range (mm)
5.5 Conclusions
The experimental results obtained by the ultrasonic testing, chloride profiles
and visual analysis of cracking are summarized in Table (5.3).
Table 5.3: Summary of the experimental results.
*
Damage (%) C dam
t /C undam
t
crack
Mix based on at depth range
pulse resonant 0-19 19-44 width
depth density connectivity
velocity frequency mm mm (mm)
46 67 75 1.5 11 0.4 small high high
56 46 70 2.2 7 1.2 small low low
66 84 96 2.1 - 2 large low low
55 85 99 1.6 19 1.4 large high high
57 81 99 1.3 - 2.6 large low high
* Ratio of total chloride concentrations obtained from damaged and undamaged

concrete specimens.
119
The following can be concluded from the analysis of the experimental results:
1. The exposure to cyclic freezing and thawing resulted in a substantial internal
damage and cracking of concrete samples. The progressive damage resulted in
new cracks, widening and extending of existing micocracks and interconnecting
of the crack network. The chloride penetration is not only related to the level of
damage and crack width, but also to the crack density and connectivity.
2. The damage factors of concrete was measured by the pulse velocity (V-meter) and
the resonant frequency (E-meter). After 400 temperature cycles, the damage
factors obtained by the two methods are consistent.
3. The damage and cracking increase the permeability and accelerate the chloride
penetration into concrete.
4. The damage and cracking resulted in a higher chloride penetration at larger depth
from exposed surface. This effect becomes very important in bridge decks, where
chloride penetration towards the reinforcing bars would increase the concentration
and thus exceeds the threshold value causing corrosion of steel bars.
5. The concrete material parameters such as aggregate volume fraction and water-
cement ratio have significant effect on the resistance to freezing-thawing damage
and chloride penetration and thus durability of concrete structures.

CHAPTER 6
EXPERIMENTAL STUDY ON THE EFFECT OF CONCRETE MATERIAL

PARAMETERS ON CHLORIDE PENETRATION
6.1 Introduction
The resistance of concrete to chloride penetration is determined by its internal
structure, which can characterized by the morphology of the pore structure, including
the total porosity, pore size distribution, spatial distribution of pores, shape,
connectivity and tortuosity of the pores network. These characteristics are controlled
by many factors including concrete material parameters, construction practices,
environmental conditions and the existence of cracking.
In this chapter, an experimental study on the effect of the concrete material
parameters; the water-cement ratio and the aggregate volume fraction, on the chloride
penetration was conducted.
The effect of water-cement ratio on chloride penetration in concrete was studied
by many researchers (Swaddiwudhipong et al., 2000; Suryavanshi et al., 1998;
Mangat and Molloy, 1995; Saetta et all., 1993; Collepardi et. al., 1972; Arya et al.
1990; Midgley and Illston, 1986). All studies show that increasing water-cement ratio
increases the capillary porosity of concrete and thus accelerates moisture diffusion
and chloride penetration. Swaddiwudhipong et al. (2000) found that lowering water-
121
cement ratio resulted in a microstructural densification of the material, which lowers
the moisture and chloride diffusion coefficients and improving the resistance to
chloride ingress in concrete. Suryavanshi et al. (1998) found that concrete mixes with
high water-cement ratio (0.75) have lower resistance to chloride penetration. While
concrete with lower water-cement ratio (0.45 and 0.60) have higher resistance to
chloride penetration. Mangat and Molloy (1995) reported an increase in the amount
of bound chloride with the increase of water-cement ratio.
The volume fraction of aggregate affects the chloride diffusion properties by
adding a low diffusion phase (aggregate) and also makes the flow paths longer and
tortuous (Bentz et al., 1998; Bentz, 2000). Thus, increasing volume fraction of
aggregate in concrete mixes decreases moisture diffusion and chloride penetration.
6.2 Specimen Preparation and Experimental Parameters
In this chapter, a 30x15x15 cm concrete slabs as shown in Fig. (6.1) were
prepared. Two sets of specimens were prepared. One set had the same aggregate
volume fraction (0.65), three different water-cement ratios (0.45, 0.55 and 0.65) and
identified as mixes 46, 56 and 66, respectively. The other set had three different
aggregate volume fractions (0.55, 0.65 and 0.75), the same water-cement ratio (0.55)
and identified as mixes 55, 56 and 57, respectively. Basically, the first digit of the
identification number represents water-cement ratio, and the second aggregate
volume fraction.
Since mix 56 (with 0.55 water-cement ratio and 0.65 aggregate volume
122
fraction) is a common mix in the two sets, five different concrete mixes were
prepared. The proportions of the five mixes are shown in Table (C.1) of Appendix C.
In order to keep a constant volume fraction of aggregate for mixes 46, 56 and 66, the
increase in the weight of mixing water was compensated by a decrease in the weight
of cement. In mixes 55, 56 and 57, the increase in volume fraction of aggregate was
compensated by a decrease in the weights of mixing water and cement, in order to
keep the water-cement ratio constant. As a result, the effect of another material
parameter, i.e. cement content was included in this experimental study. Three
specimens of each mix were casted, demolded after 24 hours and stored in a moist
room for 28 days. The total number of concrete specimens was 15.
Insulation
(four sides) 15c
m
8% NaCl
solution
15cm
Insulation
(two sides)
3 0c m
Fig. 6.1: Concrete sample
The compression test results for the five concrete mixes is shown in Table (D.1)
in Appendix D. The table shows that the compressive strength decreases with an
increase in the water-cement ratio. It also shows a small variation in the strength with
the variation of the aggregate volume fraction.

123
6.3 Testing Procedure
After 28 days of moist curing, the samples were insulated on all surfaces but the
top surface, as shown in Fig. (6.1). Then exposed to 8% sodium chloride (NaCl)
solution on the top surface and stored in a controlled temperature and humidity
conditions for 50, 100 and 200 days as shown in Fig. (6.2). Chloride solution level on
the top surface was monitored and kept constant during the ponding period.
The chloride penetration profiles for the concrete specimens exposed to
scheduled chloride ponding (50, 100 and 200 days) were obtained following the
experimental procedure described in Appendix E. The m-Volt readings resulted from
the electrochemical analysis after 50, 100 and 200 days of exposure are shown in
Tables (E.3, E.4 and E.5) of Appendix E, respectively. The mV readings of the
known chloride concentration solutions (0.005%, 0.02%, 0.05% and 0.5%) were
taken and shown in Table (E.6). The calibration curves for the standard concentration
solutions is shown in Fig. (E.5).
Fig. 6.2: Ponding with chloride solution.

124
6.4 Test Results and Discussion

6.4.1 Chloride penetration profiles
The calibration curves shown in Fig. (E.5) were used to convert the mV
readings to total chloride concentration as shown in Tables (6.1) through (6.6).
6.4.2 Effect of water-cement ratio
The chloride profiles after 50, 100 and 200 days of exposure for different water-
cement ratios 0.45, 0.55 and 0.65 (mixes 46, 56 and 66) are shown in Figs. (6.1)
through (6.3). The comparisons of chloride penetration profiles for mixes 46 and 56
after 50, 100 and 200 of exposure are shown in Figs. (6.3), (6.4) and (6.5),
respectively. One can see that increasing water-cement ratio from 0.45 to 0.55
resulted in a large increase in chloride concentration and this is due to the increase in
the capillary porosity and thus chloride penetration. This is correct except for the
chloride profile obtained from the specimen made of concrete mix 46 after 200 days
of exposure (see Fig. 6.5), where poor compaction of the concrete mix resulted in
high porosity and thus increase the chloride penetration. The poor compaction of mix
46 is a consequence of lowering the amount of mixing water. In the practice,
lowering mixing water should be compensated by adding high range water reducers
for better workability of the mix. But, in the present study, we decided that no
additives are used in all concrete mixes.
Also, the comparison of chloride profiles obtained from specimens made of
mixes 46 and 56 with the specimen made of mix 66 show that for all exposure periods
(50, 100 and 200 days), the chloride penetration in mix 66 is larger. This is true
125
expect in the depth range from 0 to 19 mm in which the specimens were exposed to
chloride solution for 50 and 100 days as shown by Figs. (6.3) and (6.4). In which,
segregation of aggregates away from the exposed surface, as shown in Fig. (6.6),
resulted in high porosity of the concrete in this depth range. The segregation of
aggregates in mix 66 is due to the higher amount of water in the concrete mix.
Table 6.1: Chloride concentration for different (w/c) after 50 days exposure time.
Depth range (mm)

w/c Sample #
0-6 6-13 13-19 19-25 25-32 32-38
1 1.191 0.392 0.159 0.110 0.021 -
2 1.127 0.574 0.222 0.083 0.016 -
0.45 3 0.986 0.521 0.310 0.083 0.065 -
Average 1.101 0.496 0.231 0.092 0.034 -
1 1.153 0.593 0.267 0.074 0.057 -
2 1.169 0.576 0.277 0.086 0.051 -
0.55 3 1.158 0.582 0.305 0.070 0.048 -
Average 1.160 0.584 0.283 0.077 0.052 -
1 1.076 0.471 0.296 0.153 0.134 0.104
2 1.137 0.505 0.202 0.108 0.106 0.116
0.65 3 1.033 0.507 0.166 0.096 0.089 0.103
Average 1.082 0.494 0.222 0.119 0.110 0.107
126
1.4
w/c = 0.45

1.2
w/c = 0.55
1.0 w/c = 0.65
0.8
0.6
0.4
0.2
0.0
0-6 6-13 13-19 19-25 25-32
Depth range (mm)
Fig. 6.3: Chloride profiles for different (w/c) after 50 days exposure time.
Depth range (mm)

w/c sample #
0-6 6-13 13-19 19-25 25-32 32-38 38-45 45-51
1 1.329 0.473 0.196 0.094 0.037 0.043 0.019 0.028
2 1.298 0.470 0.187 0.086 0.033 0.041 0.017 0.025
0.45 3 1.184 0.468 0.206 0.081 0.028 0.041 0.016 0.025
average 1.270 0.470 0.196 0.087 0.033 0.042 0.017 0.026
1 1.081 0.578 0.299 0.188 0.099 0.072 0.048 0.047
2 1.050 0.616 0.291 0.181 0.099 0.069 0.043 0.040
0.55 3 1.035 0.543 0.282 0.176 0.096 0.063 0.036 0.035
average 1.055 0.579 0.291 0.182 0.098 0.068 0.042 0.040
1 1.273 0.528 0.238 0.170 0.149 0.091 0.131 0.122
2 1.255 0.520 0.238 0.166 0.126 0.089 0.126 0.120
0.65 3 1.225 0.510 0.230 0.156 0.122 0.087 0.125 0.114
average 1.251 0.519 0.235 0.164 0.132 0.089 0.127 0.118
127
1.4
Chloride conc. (% by concrete weight) w/c = 0.45
1.2
w/c = 0.55
w/c = 0.65
1.0
0.8
0.6
0.4
0.2
0.0
0-6 6-13 13-19 19-25 25-32 32-38 38-45 45-51
Depth range (mm)
Depth range (mm)

w/c Sample #
0-6 6-13 13-19 19-25 25-32 32-38 38-45 45-51
1 1.540 0.758 0.577 0.378 0.249 0.133 0.028 0.020
2 1.439 0.758 0.536 0.376 0.215 0.099 0.021 0.019
0.45 3 1.371 0.652 0.509 0.353 0.206 0.090 0.020 0.016
average 1.450 0.722 0.541 0.369 0.223 0.107 0.023 0.018
1 1.846 0.622 0.304 0.137 0.101 0.104 0.125 -
2 1.608 0.591 0.296 0.081 0.099 0.097 0.124 -
0.55 3 1.600 0.564 0.286 0.134 0.093 0.094 0.119 -
average 1.685 0.592 0.295 0.117 0.098 0.098 0.122 -
1 1.855 0.871 0.685 0.570 0.460 0.424 0.405 -
2 1.827 0.866 0.651 0.561 0.426 0.377 0.401 -
0.65 3 1.818 0.828 0.579 0.558 0.422 0.367 0.371 -
average 1.833 0.855 0.638 0.563 0.436 0.389 0.392 -
128
2.0
1.8 w/c = 0.45
Chloride conc. (% by concrete weight) w/c = 0.55
1.6
w/c = 0.65
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0-6 6-13 13-19 19-25 25-32 32-38 38-45
Depth range (mm)
8% NaCl
Solution
150mm
300mm
Fig. 6.6: Segregation of aggregate.

129
As shown in Table (C.1) of Appendix C, the water-cement ratio and cement
content for mix 46 are 0.45 and 430 (Kg/m3); for mix 56 are 0.55 and 380 (Kg/m3);
and for mix 66 are 0.65 and 341 (Kg/m3), respectively. So, the increase in the water-
cement ratio is associated with a decrease in cement content. The review of literature
on the effect of cement content on chloride penetration revealed that chloride binding
capacity increases with the increase of the cement content (Mangat and Molloy,
1995). This was proven indirectly in this experimental study, where the increase in
the chloride penetration due to the increase in water-cement ratio was
counterbalanced by the decrease in the bound chlorides due to decrease in cement
content. This evidence becomes apparent after 200 days of exposure, where chloride
penetration is increased with the increase of water-cement ratio, as shown in Fig.
(6.5).
6.4.3 Effect of aggregate volume fraction
The chloride profiles for different aggregate volume fractions 0.55, 0.65 and
0.75 (mixes 55, 56 and 57) after 50, 100 and 200 days of exposure are shown in Figs.
(6.7) through (6.9). The figures show that the chloride concentration in a specimen
made of mix 55 is larger than the concentration in specimen made of mix 56 in the
depth range from 0-19 mm and smaller in the depth range from 19-45 mm. Due to
large particle size and high specific gravity aggregate tends to segregate from the
concrete mix and settles down in the lower portion of the specimen. The proportions
of mix 55 (see Table C.1 of Appendix C) shows that the mix has a large amount of
mixing water and cement. This resulted in segregation of aggregates and decreasing
the local aggregate volume fraction near the exposed surface and increasing the
130
aggregate content away form the exposed surface as shown in Fig. (6.6). The non-
uniform distribution of aggregates resulted in high diffusivity near the exposed
surface and low diffusivity away from the exposed surface. This results in higher
chloride concentration near the exposed surface (0-19 mm), and lower concentration
away from the exposed surface (19-45 mm).
On the other hand, mix 57 (see Table C.1 of Appendix C) has high aggregate
content, which results in poor compaction of the concrete and thus increasing the
porosity and air voids. The increase of the porosity resulted in high penetration rates
of chloride into concrete.
Thus the increase in the volume fraction of aggregate resulted in a decrease of
chloride penetration if the construction practices and the concrete mix are properly
designed to prevent the segregation and honeycombing. Otherwise, the segregation
of aggregate will have significant impact on the chloride resistance of the concrete at
different depths.
131
Table 6.4: Chloride concentration for different (gi) after 50 days of exposure.
Depth range (mm)

gi Sample #
0-6 6-13 13-19 19-25 25-32 32-38
1 1.808 0.709 0.129 0.053 0.038 0.023
2 2.113 0.825 0.183 0.030 0.081 0.036
0.55 3 1.901 0.791 0.176 0.020 0.017 0.040
average 1.941 0.775 0.163 0.035 0.045 0.033
1 1.153 0.593 0.267 0.074 0.057 -
2 1.169 0.576 0.277 0.086 0.051 -
0.65 3 1.158 0.582 0.305 0.070 0.048 -
average 1.160 0.584 0.283 0.077 0.052
1 1.252 0.784 0.699 0.770 0.394 0.322
2 1.132 0.844 0.732 0.471 0.390 0.334
0.75 3 1.293 0.766 0.719 0.454 0.399 0.335
average 1.226 0.798 0.716 0.565 0.394 0.330
2.5
gi = 0.55
2.0 gi = 0.65
gi = 0.75
1.5
1.0
0.5
0.0
0-6 6-13 13-19 19-25 25-32
Depth range (mm)
Fig. 6.7: Chloride profiles for different (gi) after 50 days of exposure.
132
Sample Depth range (mm)

gi
# 0-6 6-13 13-19 19-25 25-32 32-38 38-45 45-51
1 2.213 0.858 0.331 0.082 0.039 0.018 0.013 0.038
2 1.835 0.846 0.323 0.081 0.033 0.018 0.012 0.024
0.55 3 1.748 0.764 0.317 0.075 0.032 0.017 0.012 0.020
Average 1.932 0.823 0.324 0.079 0.035 0.017 0.012 0.027
1 1.081 0.578 0.299 0.188 0.099 0.072 0.048 0.047
2 1.050 0.616 0.291 0.181 0.099 0.069 0.043 0.040
0.65 3 1.035 0.543 0.282 0.176 0.096 0.063 0.036 0.035
Average 1.055 0.579 0.291 0.182 0.098 0.068 0.042 0.040
1 1.179 0.634 0.373 0.284 0.154 0.145 0.062 -
2 1.151 0.601 0.366 0.272 0.149 0.125 0.061 -
0.75
3 1.151 0.595 0.354 0.269 0.143 0.121 0.059 -
Average 1.160 0.610 0.365 0.275 0.149 0.131 0.061 -
2.5
gi = 0.55
2.0 gi = 0.65
gi = 0.75
1.5
1.0
0.5
0.0
0-6 6-13 13-19 19-25 25-32 32-38 38-45
Depth range (mm)
133
Depth range (mm)

gi Sample #
0-6 6-13 13-19 19-25 25-32 32-38 38-45
1 2.074 0.779 0.370 0.118 0.046 0.116 0.046
2 1.832 0.775 0.340 0.107 0.043 0.071 0.045
0.55 3 1.666 0.772 0.349 0.118 0.048 0.078 0.051
Average 1.857 0.775 0.353 0.115 0.046 0.089 0.047
1 1.846 0.622 0.304 0.137 0.101 0.104 0.125
2 1.608 0.591 0.296 0.081 0.099 0.097 0.124
0.65 3 1.600 0.564 0.286 0.134 0.093 0.094 0.119
Average 1.685 0.592 0.295 0.117 0.098 0.098 0.122
1 1.359 0.639 0.472 0.383 0.318 0.233 0.195
2 1.306 0.581 0.469 0.377 0.302 0.225 0.194
0.75 3 1.287 0.572 0.453 0.373 0.296 0.222 0.187
Average 1.318 0.597 0.465 0.378 0.305 0.226 0.192
2.0
1.8 gi = 0.55
1.6 gi = 0.65
gi = 0.75
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0-6 6-13 13-19 19-25 25-32 32-38 38-45
Depth range (mm)
134
6.5 Conclusions
1. Chloride penetration in concrete increases with the increase of water-cement ratio,
which is due mainly to the increase of the capillary porosity.
2. Chloride penetration in concrete decreases with the increase of aggregate volume
fraction. The additional aggregates decrease the diffusion coefficient of concrete
and thus impede chloride penetration in concrete.
3. Chloride binding in concrete increases with the increase of the cement content.
4. The experimental results of this study emphasized the importance of proper
construction practices and concrete mix design, where segregation of aggregates
and poor compaction of concrete would significantly increase chloride
penetration.
CHAPTER 7
EXPERIMENTAL STUDY ON THE COUPLING EFFECT OF MOISTURE

DIFFUSION AND CHLORIDE PENETRATION IN CONCRETE
7.1 Introduction
The interaction between moisture diffusion and chloride penetration becomes
more important in concrete structures subjected to repeated wetting and drying cycles.
Those structures include bridge decks, parking garage floors and marine structures in
the splash and tidal zones. Chloride diffusion due to concentration gradient is the
dominant penetration mechanism in fully saturated concrete, where pores inside
concrete become completely filled with solution and water movement ceased. On the
other hand, chloride penetration in partially saturated concrete involved many
mechanisms besides diffusion due to concentration gradient. Those mechanisms
include water movement (carrying chloride ions) due to diffusion, sorption and
permeability.
The review of the available literature showed that very few experimental
studies were conducted on the coupled effect of moisture and chloride penetration.
Swaddiwudhipong et al. (2000) investigated the effect of moisture transport due to
concentration gradient, capillary suction and pressure gradient on chloride ingress in
partially saturated concrete. Their experimental results showed that moisture transport
has no effect on chloride penetration for long term exposure period and chloride
136
concentration gradient is the most dominant mechanism for chloride penetration in
concrete. Buenfeld et al. (1995) studied the effect of wick action on chloride
penetration in concrete. Wick action can be defined as the transport of water or
solution through concrete from a wet front (face in contact with solution) to dry front
(face in contact with atmosphere).
The experimental study in this chapter concentrated on the mutually coupled
effects between moisture diffusion and chloride penetration. Two series of
experiments were conducted. One is for the effect of moisture diffusion on chloride
penetration, while the other for the effect of chloride penetration on moisture
diffusion. The effect of moisture diffusion on chloride was conducted by a
comparative study on saturated and non-saturated concrete. The inverse problem was
studied utilizing wick action.
7.2 Specimens Preparation
The specimen shown in Fig. (7.1) was used to study the coupled effect of
moisture diffusion and chloride penetration in concrete. The specimens were designed
as concrete cubes (20x20x15 cm) with a central hole, 10cm in diameter and extends
to the depth of 11.4 cm. The samples were casted, demolded after 24 hours and
stored in the moist room for 28 days.

137
Fig. 7.1: Concrete sample for coupled moisture diffusion and chloride penetration
7.3 Effect of Moisture Diffusion on Chloride Penetration

7.3.1 Experimental procedure
The mix proportions of the concrete used in this experiment are shown in
Table (C.3) of Appendix C. After 28 days of moist curing, 8 samples were taken from
the moisture room. Four of them were exposed to wetting by placing them in a tank
containing pure water for one month. The other four samples were exposed to drying
in the laboratory environment (temperature=27 oC and relative humidity=40%) for
the same period of time. The compression test for concrete cylinders of 10x20cm
(4”x8”) made of the same mix was performed as described in Appendix D. The
compression test results are shown in Table (D.2) of Appendix D. The table shows
that the average 28-day compressive strength of the concrete cylinders is 23.4 MPa.
After one month of exposure to drying (four samples) and wetting (the other
four samples), one sample of each set was tested for the internal relative humidity by
using the commercial HM44 concrete humidity measurement system manufactured

138
by Vaisala, Inc. This measurement system is mainly composed of relative humidity
probe connected to electronic indicator. The relative humidity can be measured by
boring a hole to the required depth as shown in Fig. (7.2). After cleaning the hole, a
plastic sleeve should be inserted in the hole. At this point, the probe can be pushed
into the sleeve and sealed. The concrete at the bottom of the hole release moisture
into the space around the probe until equilibrium is reached. The probe is then
connected to the electronic indicator, which gives the reading of the relative humidity.
The setup, which was used in the measurement of relative humidity, is shown in Fig.
(7.2).
Figure 7.2: Setup for concrete humidity measurement.
The rest of the specimens (6 specimens) were immersed in a bath containing
sodium chloride solution with the concentration of 5% NaCl (by weight) as shown in
139
Fig. (7.3). The specimens were exposed to chloride solution from outside, and the
initially fully saturated specimens were exposed to pure water in the central hole.
Fig. 7.3: Immersion bath of the concrete samples.
After certain exposure periods (30 days, 90 days and 180 days), samples of
concrete powder were collected at various depths from the outside surface of the wall
of specimens. The powder samples were analyzed for total chloride content by using
the rapid chloride test instrument manufactured by Germann Instruments, Inc. as
described in Appendix E.
7.3.2 Experimental results and discussion
The inside relative humidity, measured by concrete humidity measurement
system, in the set of specimens which was exposed to drying found to be 40%. In the
140
other set, which was exposed to wetting found to be 100%.
The results of the electrochemical analysis (mille Volt readings) of the two sets of
the specimens at different depths from the exposed surface and different time of
exposure (30, 90, and 180 days) are presented in Tables (E.7) through (E.9) in
Appendix E. The mV readings of the known chloride concentration solutions
(0.005%, 0.02%, 0.05% and 0.5%) are shown in Table (E.10). The total chloride
concentrations converted from mV-readings using the calibration curves shown in
Fig. (E.6) are presented in Tables (7.1) through (7.3).
The chloride penetration profiles obtained from the two different sets of
specimens after 30, 90 and 180 days of exposure are shown in Figs. (7.4) through
(7.6). One can see clearly from the figures that the chloride concentrations are higher
in the partially saturated concrete than the fully saturated concrete, which means that
the moisture diffusion accelerates the chloride diffusion when the two diffusion
processes occur in the same direction. Mathematically, the chloride diffusion in
saturated concrete can be expressed by a standard partial differential equation, Eq.
(7.2), based on Fick’s law, Eq. (7.1). However, due to the simultaneous occurrence
of the moisture diffusion and chloride diffusion in non-saturated concrete, an
additional term in the Fick’s law must be included to take into account the effect of
moisture diffusion, as shown in Eq. (7.3). As a result, the governing equation
becomes Eq. (7.4)
J Cl = − DCl grad ( C f ) (7.1)

141
∂ Ct
= − div ( J Cl ) = div[ DCl grad (C f )] (7.2)
∂t
J Cl = − DCl grad (C f ) − DCl − H grad ( H ) (7.3)
∂ Ct
= − div ( J Cl ) = div[ DCl grad ( C f )] + div[ DCl − H grad ( H )] (7.4)
∂t
in which, JCl = chloride flux, Ct = total chloride concentration, Cf = free chloride
concentration, H = pore relative humidity in concrete, DCl = chloride diffusivity, DCl-
H = chloride diffusivity due to moisture diffusion, t = time. The first term in the right
hand side of Eqs. (7.3) and (7.4) represents the driving force of chloride concentration
gradient, and the second term reflects the effect of moisture diffusion on chloride
penetration.
Table 7.1: Total chloride concentration after 30 days of exposure.
Depth Initially partially saturated Initially fully saturated

range sample number sample number
(mm) average average
1 2 3 1 2 3
0-6 0.241 0.248 0.256 0.248 0.217 0.203 0.155 0.192
6-13 0.161 0.146 0.162 0.156 0.109 0.076 0.076 0.087
13-19 0.102 0.081 0.092 0.091 0.036 0.035 0.020 0.031
19-25 0.024 0.034 0.025 0.028 0.017 0.012 0.011 0.014
25-32 0.008 0.011 0.008 0.009 0.012 0.011 0.006 0.010
142
0.30
Initially Fully Saturated
chloride conc. (% by concrete weight)

0.25
Initially Partially Saturated
0.20
0.15
0.10
0.05
0.00
0-6 6-13 13-19 19-25 25-32
Depth range (mm)
Fig. 7.4: chloride profiles after 30 days of exposure.

1 2 3 1 2 3
0-6 0.319 0.294 0.289 0.301 0.277 0.258 0.226 0.253
6-13 0.203 0.197 0.191 0.197 0.131 0.130 0.130 0.130
13-19 0.116 0.101 0.100 0.106 0.066 0.060 0.056 0.061
19-25 0.053 0.053 0.051 0.052 0.017 0.016 0.016 0.016
25-32 0.041 0.039 0.036 0.039 0.009 0.009 0.009 0.009
143
0.35

0.30

0.25
0.20
0.15
0.10
0.05
0.00
0-6 6-13 13-19 19-25 25-32
Depth range (mm)
Fig. 7.5: chloride profilest after 90 days of exposure.

1 2 3 1 2 3
0-6 0.363 0.360 0.353 0.359 0.294 0.277 0.269 0.280
6-13 0.251 0.238 0.228 0.239 0.183 0.165 0.165 0.171
13-19 0.164 0.157 0.140 0.153 0.099 0.095 0.094 0.096
19-25 0.126 0.122 0.117 0.122 0.046 0.046 0.045 0.046
25-32 0.111 0.085 0.082 0.093 0.024 0.022 0.020 0.022
144
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
0-6 6-13 13-19 19-25 25-32
Depth range (mm)

Fig. 7.6: chloride profiles after 180 days of exposure.
7.4 The Effect of Chloride Penetration on Moisture Diffusion
As shown in the previous section, the moisture diffusion accelerates chloride
penetration. Another interesting issue is the effect of chloride penetration on
moisture diffusion. This is the inverse problem to the first one. The two problems
are equally important, but the inverse problem has not been investigated before. A
seepage test was designed specifically to study the inverse effect.
7.4.1 Experimental setup and procedures
Concrete mix 56 was used in this experiment. This mix had a water-cement
ratio of 0.55, aggregate volume fraction of 0.65 and average 28-day compressive
strength of 34.1 MPa. The mix proportions and compression test results (see mix 56)
145
are shown in Table (C.2) of Appendix C, and Table (D.1) of Appendix D.
The experimental setup shown in Fig. (7.7) was used to study the effect of
chloride diffusion on moisture diffusion. In this experiment, three sets of concrete
specimens were immersed in three different bathes containing three different
solutions: pure water in Bath 1, sodium chloride (NaCl) solutions with concentrations
of 0.5M in Bath 2 and 1.0M in Bath 3, respectively. Each set had three concrete
specimens. The concrete specimens were initially saturated by water (Relative
Humidity =100%). The external surface of the center hole was exposed to the
atmosphere (40% relative humidity) during the test. In Bathes 2 and 3, chloride ions
penetrate into concrete specimens by diffusion due to concentration gradient and
permeation due to pressure gradient. Water (including chloride in specimens
immersed in bathes 2 and 3) accumulated in the center hole due to flow and wicking
action. The accumulated solution was collected and used as an indicator for the effect
of chloride penetration on moisture transfer. The accumulated solution in the center
holes was collected after 10, 20, 60 and 90 days of ponding. Except at the time of
collection for the accumulated solution, the ponding containers were covered all the
time to reduce the evaporation of the accumulated solution in the hole.

146
Fig. 7.7: Experimental setup for the effect of chloride diffusion on moisture
diffusion.
7.4.2 Results and discussion
The test results shown in Table (7.4) and Fig. (7.8), represent the weight of
the accumulated solution (in grams) in the center hole after 10, 20, 30 and 90 days.
One can see that the amount of the accumulated solution increases with the increase
of chloride concentration in the ponding solution. This means that chloride
penetration accelerates water flow and wicking.
Two important observations can be made from the test data shown in Fig.
(7.8). The first one is that the seepage test can obtain meaningful data in a rather short
time period, less than 30 days. In general, concrete has been considered as a solid
material with very low permeability, thus, direct seepage test is not viable for testing
concrete permeability. But, one can see from Fig. (7.8), that the test data of 10 days
147
shows the same trend as of 90 days, which implies that 10-day testing period may be
sufficient for studying the coupling effect. The second observation is that the
chloride penetration accelerates the moisture diffusion greatly! The amount of
seepage water is about tripled from zero chloride concentration to 0.5M, and then
doubled from 0.5M to 1.0M. It is very clear that the effect of chloride penetration on
moisture diffusion cannot be neglected.
Table 7.4: Weight of the accumulated solution.
Elapsed time (days)

Solution type Sample #
10 20 30 60 90
1 8.252 16.868 27.059 54.017 72.506
2 3.536 5.744 8.746 17.090 21.758
Pure water
3 0.287 0.444 0.797 2.023 2.923
Average 4.025 7.685 12.201 24.377 32.396
1 6.484 10.779 15.197 24.344 30.595
2 31.646 53.315 72.309 116.535 148.817
0.5M NaCl
3 11.873 21.498 30.559 55.062 73.459
Average 16.668 28.531 39.355 65.314 84.290
1 37.350 57.358 86.288 157.521 209.078
2 15.760 43.526 70.457 131.205 175.961

1.0M NaCl
3 32.340 44.796 59.071 95.566 120.817
Average 28.483 48.560 71.939 128.097 168.619

148
180
160 exposed to water

0.5M NaCl solution
weight of collected solution (g)

140 1.0M NaCl solution
120
100
80
60
40
20
0
10 20 30 60 90
elapsed time (days)
Fig. 7.8: The weight of accumulated solution in the center hole for the three
different solutions.
Similar to Eqs. (7.1) and (7.2) for the chloride penetration in saturated
concrete, the standard partial differential equation for moisture diffusion without
chloride ions is Eq. (7.6), which is based on Fick’s law as shown in Eq. (7.5). To take
into account for the effect of chloride penetration, an additional term in the Fick’s law
must be included, as shown in Eq. (7.7). As a result, the governing equation can be
modified as Eq. (7.8)
J w = − DH grad ( H ) (7.5)
∂w
= − div ( J w ) = div[ DH grad ( H )] (7.6)
∂t
149
J w = − DH grad ( H ) − DH − Cl grad ( C f ) (7.7)
∂w
= − div ( J w ) = div[ DH grad ( H )] + div[ DH − Cl grad (C f )] (7.8)
∂t
where Jw = moisture flux, H = pore relative humidity in concrete, w = moisture
content of concrete, DH = moisture diffusivity, DH-Cl = moisture diffusivity due to
chloride diffusion. The two governing equations for chloride penetration and
moisture diffusion, equations (7.4) and (7.8), are two fully coupled equations (two-
way coupled). The results of the seepage test showed that the second term in the right
hand side of equation (7.8) plays a very important role in the moisture diffusion in
non-saturated concrete, which has not be carefully evaluated so far.
7.5 Conclusions
1) In this chapter an experimental study was conducted on the coupled effect of
moisture diffusion and chloride penetration in concrete.
2) The experimental results show that the moisture diffusion accelerated the chloride
penetration (when the two processes are in the same direction). On the other hand,
the chloride penetration accelerated the moisture diffusion.
3) The effect of chloride diffusion on moisture diffusion is substantial and has not
been investigated systematically.
4) The moisture diffusion and chloride penetration are two mutually coupled
processes (two-way coupled), and the coupling effects should be considered in

150
theoretical modeling on the chloride penetration in non-saturated concrete.
5) The seepage test proposed in the study can be used to estimate the chloride
permeability of concrete in a relatively short period of time.

CHAPTER 8
EXPERIMENTAL STUDY ON THE EFFECT OF SHRINKAGE-

INDUCED DAMAGE ON CHLORIDE PENETRATION
8.1 Introduction
Drying shrinkage is one of the major problems that affect the durability of
concrete structures. The contraction of concrete due to moisture loss is usually
hindered by either internal or external restraints and thus tensile hygral stresses are
induced. Those stresses may exceed locally the tensile strength of concrete and thus,
generates microcracks (Bissonnette et al. 1999; Beddoe and Lippok, 1999;
Ramamurthy and Narayanan, 2000). The microcracks change transport properties of
concrete adversely, accelerating the moisture and chloride diffusion processes and
reduce service life of concrete structures. The external restraints include joints,
supports, connections and the bond between the newly cast concrete and the old
concrete in the case of overlays on top of existing concrete pavement. The internal
restraints include the differential shrinkage resulted from the non-uniform drying
along the depth of the concrete from the exposed surface and the non-shrinking
phases such as aggregates, unhydrated cement particles and calcium hydroxide
crystals (CH).
Pioneering experimental work by Powers et al. (1954) studied the effect of
drying on permeability of cement paste. Their results showed that the resistance to
152
water flow through the intact specimens was 70 times greater than those of
companion specimens that underwent a cycle of drying to 79% relative humidity and
resaturation. Since then, many experimental studies and theoretical analyses have
conducted (Aldea et al., 2000; Aldea et al., 1999; Gerard et al., 1996; Daian and
Saliba 1993; Francois and Maso 1988; Bazant et al., 1987). All results showed that
the concrete permeability is considerably increased with the increase of internal
damage in concrete. Bazant et al. (1987) studied the effect of cracking resulted from
drying on the permeability and diffusivity of concrete. Two types of C-shaped
reinforced concrete beams (cracked and uncracked) were exposed to controlled
drying environment and the loss in the weight of the specimen was reported. It was
observed that the cracked specimens had more weight loss, which means that the
drying process is faster in the cracked specimens. Samaha and Hover (1992)
investigated the effect of air-drying (similar to moderate environmental drying
condition) and oven-drying on the transport properties of concrete. They observed
that in air-dried concrete the shrinkage cracks are limited to a small depth from the
exposed surface. From this observation, they concluded that those cracks did not
affect greatly the transport properties of concrete. On the other hand, they noticed a
severe cracking in the oven dried concrete samples. Those cracks are uniformly
distributed through the sample and resulted in a great internal damage. And thus, they
contributed largely to the deterioration of transport properties of concrete. Gerard et
al. (1996) investigated the effect of mechanically induced cracking on the
permeability of normal and high strength concrete. The cracks were produced by
uniaxial tension loading. The water permeability was measured in a direction

153
perpendicular to the axis of loading. Two series of specimens having an average
strength of 32 and 50 (MPa) were tested to better understand the effect of the strength
on the crack pattern and thus the diffusion properties of concrete. It was found that
the damage is controlled by the length and aperture of the cracks. Based on the
observations from the cracking patterns a simple micro-macro model was suggested.
This model can be used to predict the effect of the damage evolution on the diffusion
properties of concrete.
In this chapter, a comparative study on the chloride penetration in cracked (due
to restrained drying shrinkage) and uncracked concrete was conducted. The concrete
was exposed to air drying for a certain period of time then to chloride penetration.
During the drying period the development of the shrinkage strains and cracking were
monitored. Also, chloride profiles for the cracked and uncracked concrete were
obtained.
8.2 Experimental Program
8.2.1 Material
The concrete mix used in this experiment has a water-cement ratio of 0.55 and
aggregate volume fraction of 0.55. The mix proportions are shown in Table (C.4) of
Appendix C. One can see from the proportions, that the mix has high water-cement
ratio, high cement content and low aggregate volume fraction. This was done in order
to have high shrinkage strains and thus significant damage in the concrete specimens.
The average 28-day compressive strength of concrete cylinders prepared from the
same mix is 34 (MPa). The compression test results are shown in Table (D.3) of
154
Appendix D.
8.2.2 Test specimens preparation
Two sets of concrete specimens were used in this experiment. One set was a
full concrete ring restrained by inner steel ring and the other set was half ring free to
shrink as shown in Fig. (8.1). The concrete rings had an outside diameter of 457 mm
(18”), inside diameter of 305 mm (12”) and height of 152 mm (6”). The full ring was
restrained by an inner steel ring having a thickness of 25.4 mm (1”). Total of four
specimens (two of each set) were casted, demolded after 24 hours and kept in the
laboratory environment for another day to dry the exterior surfaces for the purpose of
application of strain gages.
Steel ring
152mm
Side view
45° 45°
45° 45°
Extensometers
254 mm
305 mm 305 mm
457 mm 457 mm
a) Restrained ring b) Unrestrained half ring
Fig. 8.1: Restrained and free shrinkage concrete specimens.
The full ring is restrained against movement in the radial direction, which
develops internal pressure, p when the ring shrinks. This pressure creates tensile
155
stresses in the circumferential direction, σθ and compression stresses in the radial
directions, σr as shown in Fig. (8.2). Cracks in the concrete ring are initiated from the
inner boundaries because of the coexistence of the maximum tensile stress and the
biaxial compression-tension condition (Shah et al., 1998).
σr
internal
σθ pressure
p
σθ σr
σr σθ
Fig. 8.2: Stresses in the restrained shrinkage concrete sample.
8.2.3 Experimental procedure

Measurement of the shrinkage strains
After drying of the surface, a total of 6 strain gages were attached to the
exterior surface of the specimens in order to monitor the development of shrinkage
strains. Three gages were applied on one of the restrained rings while the others were
applied on one of the free shrinkage half rings as shown by Fig. (8.1). All the strain
gages were connected to a data acquisition system as shown in Fig. (8.3). The data
acquisition system was connected to personal computer used to display and process
156
the obtained strains in microstrains. The measurement of the shrinkage strains started
after two days of concrete casting. One day in the moist room before demolding and
the other day to dry the surface of the specimens for installation of the strain gages.
Fig. 8.3: Data acquisition system and personal computer.
Ponding by chloride solution
After 107 days of casting, two specimens (one restrained full ring and one free
half ring) were placed in 8% sodium chloride (NaCl) solution bath as shown in Fig.
(8.4). Before the placement, the specimens were insulated by aluminum foil on all
surfaces expect the exterior surface facing the chloride solution. After 50 days of
continuous exposure to chloride solution the chloride profiles are obtained for both
specimens (restrained and free shrinkage) following the procedure described in
Appendix E. The m-Volt readings resulted from the electrochemical analysis are
shown in Table (E.11) of Appendix E. The mV readings of the known chloride

157
concentration solutions (0.005%, 0.02%, 0.05% and 0.5%) were taken and shown in
Table (E.12). The calibration curve for standard concentration solutions is shown in
Fig. (E.7).
Fig. 8.4: Ponding with sodium chloride solution.
8.3 Experimental Results and Discussion
8.3.1 Shrinkage strain
The shrinkage strains due to drying from initially saturated state were measured
and recorded from four out of the six strain gages placed on the two specimens. This
is mainly due to the limitation by the data acquisition system. The readings of the
shrinkage strain from the restrained and free shrinkage specimens are shown in Table
(8.1). As shown in the table, five readings were taken in the first day, then one
reading per day up to 30 days, and finally weekly reading up to 15 weeks.

158
Table 8.1: Measured shrinkage strains (in microstrains).
Elapsed* Restrained shrinkage Free shrinkage

Time (days) strain gage # 1 strain gage # 2 strain gage # 3 strain gage # 4
0.000 0 0 0 0
0.563 7 18 32 23
0.604 10 18 31 22
0.615 12 22 33 26
0.780 10 22 42 30
1.481 22 43 63 47
1.892 25 47 72 58
3.940 66 92 128 111
4.857 86 110 152 133
5.504 98 123 168 147
5.961 115 128 178 155
6.531 122 132 187 160
7.504 145 153 204 175
10.594 182 198 253 221
10.760 185 200 255 224
11.459 194 204 258 229
12.439 204 217 270 240
13.453 215 229 281 253
14.465 232 239 289 263
17.565 261 280 322 295
18.682 271 288 330 300
19.583 276 297 338 314
20.565 282 303 343 318
21.504 291 313 355 329
24.555 314 337 375 347
25.562 321 344 383 353
26.569 330 352 392 358
27.767 339 359 398 366
31.674 356 379 418 384
33.510 373 394 430 393
39.628 401 420 456 415
45.511 416 439 474 430
46.944 419 445 477 433
52.558 427 442 493 443
56.649 425 454 497 446
62.055 432 466 507 453
66.680 439 473 514 461
75.496 443 479 524 469
79.817 450 487 531 478
83.667 452 481 522 472
94.522 465 505 545 498
105.000 470 505 544 499
* Elapsed time measured after 2 day of concrete casting.
159
Fig. (8.4) shows the development of the shrinkage strains with time, obtained
from the free and the restrained rings. The figure shows that the shrinkage in both free
and restrained specimens developed at higher rate in the first ten days of drying, then
at slower rate up to 60 days and almost leveled after that.
600
Shrinkage strain (microstains)
500
400
300
Restrained (gage #1)

200
Restrained (gage #2)
Free (gage #3)
100
Free (gage #4)
0
0 10 20 30 40 50 60 70 80 90 100 110
Elapsed time (days)
Fig. 8.5: Shrinkage strains development in restrained and free shrinkage concrete
8.3.2 Restrained shrinkage cracking
Microcracks as shown in Figs. (8.6) through (8.8), became visible on the outer
circumferential surface of restrained ring after 90 days of exposure. No cracks were
noticed on the free shrinkage specimens. Most of the time, the cracks started from the
edge (stress concentration location) and extend to the interior on the surface of the
specimens (see Fig. 8.8).

160
Fig. 8.6: One of the microcracks in the restrained shrinkage ring.
Fig. 8.7: Microcrack on the surface of the restrained ring.

161
Fig. 8.8: Microcracking of the surface of the restrained ring.
8.3.3 Chloride profiles
The calibration curve shown in Fig. (E.7) of Appendix E, was used to convert
the mVolt readings to total chloride concentration as shown in Table (8.2). The table
presents a comparison between the chloride concentrations at different depth ranges
for restrained and unrestrained specimens. The specimens were exposed to 105 days
of air drying, then 50 days to chloride penetration. The chloride profiles for damaged
(due to air drying) and undamaged specimens are shown in Fig. (8.9). The figure
shows that chloride penetrate restrained concrete ring in a higher rate compared to the
unrestrained half ring, which means that the shrinkage-induced damage increases the
chloride penetration. One can see that the concentration of chloride in the free
shrinkage specimen is higher in the depth range of 0-13 mm. Since cracks were
162
initiated and propagated from the inner part of the restrained ring, the figure shows
that chloride concentration was higher for restrained ring in depth range from 13-38
mm, but the difference is not significant. This is partly due to nonlinear inelastic
behavior due to the degradation of the initial elastic modulus and the unrecoverable
part of concrete creep.
Restrained shrinkage Free shrinkage

Depth
1 2 3 1 2 3
0-6 0.871 0.784 0.767 0.807 0.952 0.847 0.750 0.850
6-13 0.489 0.261 0.230 0.327 0.440 0.412 0.373 0.408
13-19 0.316 0.309 0.204 0.276 0.276 0.258 0.212 0.249
19-25 0.199 0.198 0.197 0.198 0.154 0.150 0.149 0.151
25-32 0.131 0.116 0.101 0.116 0.113 0.109 0.073 0.099
32-38 0.119 0.089 0.070 0.093 0.089 0.082 0.073 0.081

163
0.90
0.80 Free Shrinkage

Restrained Shrinkage
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0-6 6-13 13-19 19-25 25-32 32-38
Depth range (mm)
Fig. 8.9: chloride profiles for restrained and free shrinkage concrete samples after 50
days of exposure.
8.4 Conclusions
1) Concrete drying resulted in high shrinkage strains in both free and restrained
shrinkage specimens. The free shrinkage strain was 450-500x10-6 after 105 days
of drying. In restrained shrinkage it was 500-550x10-6 after 105 days of drying.
2) Drying shrinkage resulted in cracking and substantial damage when the concrete
was restrained.
3) The surface cracks resulted from the shrinkage-induced damage increase the
permeability of concrete and thus increase the moisture diffusion and chloride
penetration. However, the effect of shrinkage-induced damage on chloride
penetration is not very strong comparing with other effects (such as mechanical
loading).
CHAPTER 9
CONCLUSIONS AND FUTURE RESEARCH RECOMMENDATIONS
The main goal of this thesis was to investigate some of the processes related to
durability of concrete structures by means of theoretical modeling and experimental
study. The processes include moisture diffusion, chloride penetration, freezing-
thawing and the coupling effects between them. The objectives were achieved by
three consecutive steps. The first step was achieved by reviewing the available
literature and following up the new publications on the researched topics. This step
was necessary for strengthening the background information and understanding the
different mechanisms for each process involved in this study. Also, it paved the road
to the following steps. The second step was the theoretical modeling, which involved
the formulation of the governing equations and boundary conditions, identification of
the material parameters and the numerical solutions. This step resulted in numerical
models for the coupled effect of moisture diffusion and drying shrinkage and the
effect of moisture diffusion on chloride penetration. A systematic description of the
proposed models and parametric studies on the effects of different influential
parameters are given in chapters three and four. The third or the final step was the
experimental study, and it covered four series of experiments. The first series was
designed to study the coupling effect of freezing-thawing and chloride penetration,
which is described in chapter five. The second series was focused on the effect of
165
concrete material parameters on chloride penetration, which is presented in chapter
six. The third series was concentrated on the coupling effect of moisture diffusion
and chloride penetration, which is presented in chapter seven. And the last series was
for the effect of shrinkage-induced cracking on chloride penetration, which is
described in chapter eight. The experimental results obtained in this study can be used
to validate and demonstrate the accuracy of the numerical models developed in this
study and the future studies.
9.1 Conclusions
1) Concrete deteriorations mechanisms are attributed to several factors including
physical, chemical and biological attacks. The common feature of those factors is
that they are all diffusion-controlled processes. Those processes include moisture
diffusion, chloride penetration and freezing-thawing.
2) The diffusion properties of concrete are related to its permeability, which is
affected by many factors including material parameters and proportions,
construction practices, exposure conditions and the features of cracking and air
voids.
3) Durability of concrete structures can be enhanced by improving their resistance to
moisture diffusion, chloride penetration and frost action. This can be achieved by
lowering the permeability, lowering the water-cement ratio, appropriate selection
of materials, good compaction, prevent premature damage and cracking from
shrinkage and thermal effects, and good curing conditions.
4) The Multiscale modeling of the coupled effect of moisture diffusion and drying
shrinkage predicted well the basic trends and the effects of concrete material
166
parameters, such as the water-cement ratio and the aggregate volume fraction. In
this model, the effect of damage on moisture diffusion was characterized by the
concept of continuum damage mechanics using the isotropic damage parameter
proposed by Kachanov (1958). This is an acceptable approach, since microcracks
induced by drying shrinkage are randomly distributed without a preferential
orientation. The effect of drying shrinkage on the moisture capacity was modeled
based thermodynamics laws and minimum potential energy principle.
5) The numerical study of chapter four resulted in a rigorous model for chloride
penetration into non-saturated concrete. The model predicts very well the effect of
moisture diffusion on the chloride ion penetration in partially saturated concrete.
The numerical results show that the moisture diffusion accelerates the chloride
penetration. Also, it incorporates the effects of concrete mix parameters; such as
volume fraction of aggregate and water-cement ratio.
6) The numerical study also presented the stability and convergence of the
Alternating-Direction Implicit method for solving the governing equations of
moisture diffusion and chloride penetration.
7) The model can be used in the durability design of reinforced concrete structures to
predict the onset of steel corrosion, which is important for predicting the service
life of concrete structures and for scheduling inspection and rehabilitation of the
structures.
8) The experimental study of chapter 5 emphasized the following:
o The large deteriorating action of alternating freezing and thawing on
durability of concrete. The progressive damage and cracking of concrete cover

167
increases the permeability and accelerate the rate of chloride penetration
towards the steel bars, which increases the chloride concentration in concrete
and triggers the corrosion process.
o The consistency of the pulse velocity and the resonant frequency for
predicting the concrete damage.
o The permeability is not only related to the level of damage and crack width,
but also to the crack density and connectivity.
9) The experimental study of chapter six, presented the effect of concrete material
parameters as well as the construction practices on the resistance to chloride
penetration.
10) The experimental work in chapter seven emphasized the mutually effect of
moisture diffusion and chloride penetration in concrete. The experimental results
show that the moisture diffusion accelerated chloride penetration (when the two
processes are in the same direction). On the other hand, chloride penetration
accelerated the moisture diffusion.
11) The experimental study of chapter eight presented the effect shrinkage-induced
damage on the chloride penetration. In which, the surface cracks increase the
permeability of concrete and thus increase the moisture diffusion and chloride
penetration in concrete.
9.2 Future Research Recommendations
1) This study can be further extended to include the effect of the mechanical loading
168
on the diffusion processes.
2) Since the theoretical modeling was done before the experimental study, a lot of
new findings have not been included in the current theoretical models. i) the
effect of chloride penetration on moisture diffusion needs to be included in the
model for moisture diffusion, and in this way, the moisture diffusion and chloride
penetration become a fully two-way coupled processes; ii) a thermal conduction
model must be established in order to evaluate the damage due to the freezing-
thawing cycles, and furthermore to evaluate the chloride penetration under the
coupled thermal effect; iii) in the material models, such as binding capacity and
chloride permeability, one more material parameter, cement content, must be
incorporated; and finally iv) once the coupled diffusion models and material
models are established, they must be validated by using the experimental data
from Chapters 5, 6, 7, and 8.
3) Further experimental and numerical studies are needed on the effect of mineral
additives on chloride penetration, such silica fume (SF), ground blast furnace slag
(GBFS) and fly ash (FA) on the chloride penetration.

169
BIBLIOGRAPHY
Ababneh, A., Xi, Y., and Willam, K. (2001). “Multiscale Modeling of the Coupled
Moisture Diffusion and Drying Shrinkage of Concrete.” Proc. of 6th International
Conference on Creep, Shrinkage & Durability Mechanics of Concrete and other
Quasi-Brittle Materials, Aug. 20-22, MIT, Cambridge, USA, pp. 159-164.
Aldea, C.M., Shah S.P., and Karr, A. (1999), “Effect of Cracking on Water and
Chloride Permeability of Concrete.” Journal of Materials in Civil Engineering,
11(3), 181-187.
Aldea, C.M., Shah, S.P., and Karr, A. (2000). “Effect of Microcracking on Durability
of High-Strength Concrete.” Transportation Research Record, (1668),
Transportation Research Board, Washington DC, 86-90.
Al-Khaja, W.A. (1997). “Influence of temperature, cement type and level of concrete
consolidation on chloride ingress in conventional and high-strength concrete.”
Construction and Building Materials, 11(1), 9-13.
Altoubat, S.A. and Lange, D.A. (1997). “Early Age Shrinkage and Creep of Fiber
Reinforced Concrete For Airfield Pavement.” Proc. Of Aircraft/Pavement
Technology: In the Midst of Change, ASCE, Seatle, WA, 229-243.
Amey, S.L., Johnson, D.A., Miltenberger, M.A., and Farzam, H. (1998). “Predicting
the service life of concrete marine structures: an environmental methodology.”
ACI Structural Journal, 95(2), 205-214.
Andrade, C. (1993). “Calculation of Chloride Diffusion Coefficients in Concrete from

Ionic Migration Measurement.” Cement and Concrete Research,23,724-742.
Andrade, C., and Whiting, D. (1996). “A comparison of chloride ion diffusion

coefficients derived from concentration gradients and non-steady state
accelerated ionic migration.” Material and Structures, 29, 476-484.
Argyris, J.H., Vaz, L.E., and Willam, K.J. (1977). “Higher Order Methods for
Transient Diffusion Analysis.” Computer Methods in Applied Mechanics and
Engineering, 12, 243-278.
Arya, C., Buenfeld, N.R., and Newman, J.B. (1990). “Factors influencing chloride-
binding in concrete.” Cement and Concrete Research, 20(2), 291-300.
Baluch, M.H., Qureshy, A.B., and Azad, A.K. (1989). “Fatigue Crack Propagation in
Plain Concrete.” Proc. Of Fracture of Concrete and Rock, Ed. Shah, S.P., and
Swartz, S.E., Springer-Verlag, New York, 80-87.
170
Bazant, Z.P. and X., Y. (1994). “Drying creep of concrete: constitutive model and
new experiments seprating its mechanisms.” Materials and Structures, 27, 3-14.
Bazant, Z.P., and Chern, J.C. (1985). “Concrete creep at variable humidity:
Constitutive law and mechanism.” Materials and Structures (RILEM), 18, 1-20.
Bazant, Z.P., and Najjar, L.J. (1972). “Nonlinear water diffusion of nonsaturated
concrete.” Matériaux et Constructions, 5(25).
Bazant, Z.P., and Raftshol, W.J. (1982) “Effect of Cracking in Drying and Shrinkage
Specimens.”Cement and Concrete Research, 12, 209-226.
Bazant, Z.P., and Thonguthai, W. (1978). “Pore Pressure and Drying of Concrete at
High Tempearture.” Journal of Engineering Mechanics-ASCE, 104, 1058-1080.
Bazant, Z.P., Chern, J.C., Rosenberg, A. M., and Gaidis, J. M. (1988). “Mathematical
Model for Freeze-Thaw Durability of Concrete.” J. of Ame. Ceram. Soc., 71(9),
776-783.
Bazant, Z.P., Sener, S., and Kim J.K. (1987). “Effect of Cracking on Drying
Permeability and Diffusivity of Concrete.” ACI Materials Journal, 84, 351-358.
Beddoe, R.E., and Lippok, R. (1999). “Hygral stress in hrdenend cement paste.”
Materials and Structures, 32, 627-634.
Bentz, D.P., Clifton, J.R., Ferraris, C.F., and Garboczi, E.J. (1999). “Diffusion
Properties and Durability of Concrete: Literature Review and Research Plan.”
Building and Fire Research Laboratory, NIST, Gaithersburg, Maryland.
Bentz, D.P., Garboczi, E.J. and Lagergren E.S. (1998). “Multi-Scale Microstructural
Modeling of Concrete Diffusivity: Identification of Significant Variables.”
Journal of Cement, Concrete & Aggregates, 20(1), 129-139.
Bentz, D.P., Quenard, D.A., Veronique, B.B., Garboczi, E.J. and Jennings, H.M.
(1995). “Modelling during shrinkage of cement paste and mortar, Part 1.
Structral models from nanometers to millimeters.” Materials and
Stuctures,28,450-458.
Bentz, D.P.,(2000). “Influence of silica fume on diffusivity in cement-based materials

II. Multi-scale modeling of concrete diffusivity.” Cement and Concrete
Research, 30, 1121-1129
Berke, N.S. (1986). “Corrosion Rates of Steel in Concrete.” ASTM Standardization

News, 14(3), 57-61.
Bissonnette, B., Pierre, P., and Pigeon, M. (1999). “Influence of Key Parameters on
171
Drying Shrinkage of Cementitious Materials.” Cement and Concrete Research,

29(10), 1655-1662.
Bouny, V.B., Perrin, B., and Chemloul, L. (1997). “Experimental Determination of

Moisture Properties of Hardened Cement Pastes, Showing Hysteresis Effects.”
Materials and Structures (RELEM), 30, July, 340-348.
Browne, R. (1982). “Design Prediction of the Life for Reinforced Concrete in a

Marine and Other Chloride Environment.” Durability of Building Materials,
1(2), 113-125.
Brunauer S., Skalny J., and Bodor E.E. (1969). “Adsorption on nonporous solids.”
Journal of Colloid and Interface Science, 30 (4), 546-552
Buenfeld, N.R., Shurfa-Daoudi, M.T., and McLoughlin, I.M. (1995). “Chloride

Transport Due to Wick Action in Concrete.” Proc. Of International RILEM
workshop on chloride penetration in concrete, France, 315-324.
Cady, P.D., and Weyers, R.E. (1992). “Predicting Service Life of Concrete Bridge
Decks Subject to Reinforcement Corrosion.” Corrosion Forms & Control for
Infrastructure, San Diego, CA.
Carnahan, B., Luther, H.A. and Wilkes, J.O. (1969). Applied numerical methods.
John Wiley & Sons, Inc., New York.
Chatterji, S. (1994). “Transportation of Ions through Cement Based Materials, Part I:

Fundamental Equations and Basic Measurement Techniques.” Cement and
Concrete Research, 24(5), 907-912.
Christensen, R.M. (1979). Mechanics of composite materials. Wiley Interscience,

New York.
Coggins, F.B., and French, C.W. (1990). “Chloride Ion Penetration in Twenty-Year-
Old Prestressed Bridge Girders.” ACI Materials Journal,87(5),479-488.
Collepardi, M., Marciall, A., and Turriziani, R. (1972). “Penetration of chloride ions
in cement pastes and concrete.” Journal of American Ceramic Society, 55(10),
534-535.
Costa, A., and Appleton, J. (1999). “Chloride penetration into concrete in marine
environment–Part II: Prediction of long term chloride penetration.” Material and
Structures (RILEM), 32, 354-359.
Daian, J.F., Saliba, J. (1993). “Transient Moisture Transport in a Cracked Porous

Medium.” Transport in Porous Media, 13, 239-260.
172
Derucher, K. N., and Heins, C. P. (1981). Materials for Civil and Highway
Engineers. Prentice-Hall, New Jersey.
Diamond, S., and Bonen, D. (1993). “Microstructure of Hardened Cement Paste—A

New Interpretation.” J. Am. Ceram. Soc., 76(12), 2993-2999.
Digler, W.H., Ghali, A., Chan, M., Cheung, M.S., and Maes, M. A. (1983). “
Temparature-Induced Stresses in Composite Box-Girder Bridges.” Journal of
Structural Engineering, ASCE, 109(6), 1460-1478.
Elbadry, M., and Ghali, A. (1983). “Temperature Variations in Concrete Bridges”,

Journal of Structural Engineering, 109(10), 2355-2374.
Emanuel, J. H., and Hulsey, J. L. (1978). “Temperature Distribution in Composite

Bridges.” Journal of Structural Division, ASCE, 104, 65-78.
Fagerlund, G. (1992). “Effect of the Freezing Rate on the Frost Resistance f

Concrete.” Nordic Concrete Research, Publication No. 11, Oslo.
Fagerlund, G. (1994). “Predicting the Service Life of Concrete Exposed to rust

Action Through a Modeling of the Water Absorption Process in the Air Pore
syste.” Proc. of NATO/RILEM workshop on the Modeling of Microstructure and
its Potential for Studying Transport Properties and Durability, St. Remy-les-
Chevreuse, France, July 10-13.
Francois, R., Maso, J.C. (1988), "Effect of Damage in Reinforced Concrete on

Carbonation or Chloride Penetration", Cement and Concrete Research, 18, 961-
970.
Frey, R., Balogh, T., and Balazs, G. L. (1994). “Kinetic Method to Analyze Chloride
Diffusion in Various Concretes.” Cement and Concrete Research, 24(5), 863-
873.
Gerard, B., and Marchand, J. (2000). ”Influence of cracking on the diffusion

properties of cement-based materials, Part I: Influence of continuous cracks on
the steady-state regime.” Cement and Concrete Research, 30, 37-43.
Gerard, B., Breysse, D., Ammouche, A., Houdusse, O., and Didry, O. (1996).
“Cracking and permeability of concrete under tension.” Materials and Structures
(RILEM), 29, 141-151.
Gerard, B., Breysse, D., Ammouche, A., Houdusse, O., and Didry, O. (1996) “
Cracking and permeability of concrete under tension”, Materials and Structures,
29, 141-151.
Gontar, W.A., Martin, J.P., Popovics, J.S. (2000), “ Effect of Cyclic Loading on
173
Chloride Permeability of Plain Concrete”, Special Publication of ASCE

“Condition Monitoring of Materials and Structures”, F. Ansari, Editor, 84-94.
Hale, D.K. (1976). “Review the Physical Properties of Composite Materials.” J. of

Materials Science, 11, 2105-2141.
Hashin, Z. (1983). “Analysis of Composite Materials - A Survey.”, Journal of

Applied Mechanics, 50, 481-505.
Helmuth, R. A. (1960). “Capillary Size Restrictions on Ice Formation in Hardened

Portland Cement Pastes.” Proc. of the 4th Int. Sym. on the Chemistry of Cement,
Washington, D.C., Paper VI-S2, 855-869.
Herve, E., and Zaoui, A. (1990). “Modeling of Effective Behavior of Nonlinear

Matrix-Inclusion Composites.” Eur. J. Mech. And Solids, 9(6), 505-515.
Hilsdorf, H.K., and Kesler, C.E. (1966). “Fatigue Strength of Concrete under Varying
Flexural Stresses.” ACI Journal , 63(10), 1059-1075.
Jennings, H. M., and Tennis, P. D. (1994). “A model for the Developing

Microstructure in Portland Cement Paste.” Journal of American Ceramic Society,
77(2), 3161-3172.
Jennings, H.M., and Xi, Y. (1993) “Microstructurally based Mechanisms for

Modeling Shrinkage of Cement Paste at Multiple Levels.” Proc. of the 5th Int.
Symp. on Creep and Shrinkage of Concrete, Barcelona, Spain.
Kachanov, L. M. (1958). “Time rupture process under creep conditions.” Izv. Akad.
Nauk, USSR, Otd. Tech. Nauk. (8), 26-31.
Kachanov, L. M. (1986). Introduction to Continuum Damage Mechanics. Martinus

Nijhoff Dordrecht, The Netherlands.
Kawai, K., Kohno, K., Morinaga, T., and Tazawa, E. (1996). “Effects of Organic and
Carbonic Acids on Concrete Deterioration Caused by Aerobic Microorganisms.”
Journal of Materials, Concrete Structures and Pavements, 550, 105-115.
Kermani, A. (1991), “Stressed Concrete: Permeability of Stressed Concrete”,

Building Research and Information, 19(6), 360-366.
Kong, J., Ababneh, A., Frangopol, D.M., and Xi, Y. (2001). “Reliability Analysis of
Chloride Penetration in Saturated Concrete.” submitted to Probabilistic
Engineering Mechanics.
Konin, A., francois, R., Arliguie, G. (1998), “Penetration of Chloride in Relation to

the Microcracking State into Reinforced ordinary and High Strength Concrete”,
174
Materials and Structures (RILEM), 31, 310-316.
Li, Z., Qi, M., Li, Z., and Ma, B. (1999). “Crack Width of High-Perfromance
Concrete Due to Restrained Shrinkage.” Journal of Materials in Civil
Engineering, 11(3), 214-223.
Litvan, G.G. (1973). ”Pore Structure and Frost Susceptibility of Building Materials."
Proc., RILEM/IUPAC Int. Sym. on Pore Structure and Properties of Materials,
Sept. 18-21, Prague, F17-F30.
Luping, T., and Nillsson, L.O. (1993). “Chloride binding capacity and binding
isotherms of OPC pastes and mortars.” Cement and Concrete Research, 23(2),
247-253.
Majorana, C., and Mazars, J. (1997). “Thermohygrometric and Mechanical Behavior

of Concrete Using Damage Models.” Materials and Structures (RILEM), 30,
349-354.
Mangat, P.S., and Molloy, B.T. (1995). “Chloride binding in concrete containing
PFA, GBS or silica fume under sea water exposure.” Magazine of Concrete
Research, 47(171), 129-141.
Martys, N. S. (1995). Survey of Concrete Diffusion Properties and Their

Measurement. Building and Fire Research Laboratory, NIST, Gaithersburg,
Maryland.
Martys, N. S., Torquato, S., and Bentz, D. P. (1994). “Universal Scaling of Fluid
Permeability for Sphere Packing.” Phys. Rev., E, 50(1), 403-408.
Mehta, P. K. (1991). “Durability of Concrete -- Fifty Years of Progress?” Proc.,

Second International Conference on Durability of Concrete, V. M. Malhorta, ed.,
Montreal, Canada, v. 1, 1-31.
Mehta, P.K., and Monteiro, P.J.M. (1993). Concrete, Structures, Properties, and
Materials. Prentice Hall, Englewood Cliffs, New Jersey, p.120.
Meziani, H., Skoczylas, F. (1999), “An Experimental Study of the Mechanical

Behavior of a Mortar and of its Permeability under deviatoric Loading”,
Materials and Structures (RILEM), 32, 403-409.
Mier, J. G. van., (1997). Fracture processes of concrete: assessment of material

parameters for fracture models. CRC Press, Inc., Boca Raton, Florida.
Mohamed, O. A., Rens, K. L. and Stalnaker, J. J. (2000). “Factors Affecting

Resistance of Concrete to Freezing and Thawing Damage.” Journal of Materials
in Civil Engineering,12(1),26-32.
175
Neubauer, C. M., Bergstrom, T. B., Sujata, K., Xi, Y., Garboczi, E.J., and Jennings,
H.M. (1997). “Drying Shrinkage of Cement Paste as Measured in an ESEM and
Comparison with Microstructural Models.” Journal of Materials Science, 32,
6415-6427.
Neville, A. M. (1996). Properties of Concrete, John Wiley & Sons, Inc., New York.
Nordby, G.M. (1958). “Fatigue of Concrete-A Review of Research.” Journal of the

ACI Journal, 30(2), 191-219.
Olsen, M.P.J. (1984). “Mathematical Modeling of the Freezing Process of Concrete

and Aggregates", Cement and Concrete Research, 14, 113-122.
Page, C.L., Short, N.R., and El Tarras, A. (1981). “Diffusion of Chloride Ions in
Hardened Cement Paste”. Cement and Concrete Research, 11(3), 395-406.
Philleo, R.E. (1986) “Freezing and Thawing Resistance of High-Strength Concrete”,

NCHRP Synthesis 129, Transportation Research Board, NRC, Washington D.C.
Popovics, S. (1987). “A Classification of the Deterioration of Concrete Based on

Mechanism.” Proc., Katharine and Bryant Mather International Conference on
Concrete Durability, Atlanta, Georgia, 1, 131-142.
Powers, T.C., Copeland, L.E., Hayes, J.C., and Mann, H.M. (1954). “Permeability of
Portland Cement Paste.” ACI Journal, 51,285-298.
Ramamurthy, K. and Narayanan, N. (2000). “Influence of composition and curing on

drying shrinkage of aerated concrete.” Material and Structures, 33, 243-250.
Reinhardt, H.W., Sosoro, M., Zhu, X.-F. (1996), “Cracked and Repaired Concrete
Subject to Fluid Penetration”, Materials and Structures (RILEM), 31, 74-83.
Rosen, B. W., and Hashin, Z. (1970). “Effective Thermal Expansion Coefficients and
Specific Heats of Composite Materials.” Int. J. Engng. Sci., 8, 157-173.
Saetta, A.V., Scotta, R. V., and Vitaliani, R. V. (1993). “Analysis of Chloride

Diffusion into Partially Saturated Concrete.” ACI Materials Journal, 90(5), 441-
451.
Saito M., Ishimori, H. (1995) “Chloride Permeability of Concrete Under Static and
Repeated Loading”, Cement and Concrete Research, 25(5), 803-808.
Samaha, H., and Hover, K. C. (1992). “Influence of Microcracking on the Mass

Transport Properties of Concrete.” ACI Materials Journal, 89(4), 416-424.
Scherer, G. W. (1999). “Crystallization in Pores.” Cement and Concrete Research,

176
29,1347-1358.
Selleck, S. F., Landis, E. N., Peterson, M. L., Shah, S. P., and Achenbach J. D.
(1998). “Ultrasonic Investigation of Concrete with Distributed Damage.” ACI
Materials Journal, 95(1), 27-36.
Shah, S.P., Ouyang, C., Marikunte, S., Yang, W. and Becq-Giraduon, E. (1998). “A
Method to Predict Shrinkage Cracking of Concrete.” ACI Materials Journal,
95(4), 339-346.
Shah, S.P., Weiss, W.J., and Yang, W. (1997). “Shrinkage Cracking in High
Performance Concrete.” Proc. of the PCI/FHWA International Symposium on
High Performance Concrete, New Qrleans, LA, 148-158.
Sugiyama, T., Bremner, T.W., Holm, T.A. (1996), “Effect of Stress on Gas
Permeability in Concrete”, ACI Materials Journal, 93(5), 443-50.
Suryavanshi, A.K., Swamy, R.N., and McHugh, S. (1998). “Chloride penetration into
reinfrorced concrete slabs.” Canadian Journal of Civil Engineering, 25, 87-95.
Swaddiwudhinpong, S., Wong, S.F., Wee, T.H., and Lee, S.L. (2000). “Chloride
ingress in partially and fully saturated concretes.” Concrete Science and
Engineering, 2, 17-31.
Tang, L., and Nilson, L.O. (1993). “Chloride binding capacity and binding isotherms
of OPC pastes and mortars.” Cement and Concrete Research, 23, 247-253.
Taylor, H. F. W. (1987). “A method for Predicting Alkali Ion Concentration in

Cement Pore Solutions.” Advanced Cement Research, 1(1), 5.
Tritthart, J. (1989). “Chloride binding in cement. II. The influence of hydroxide

concentration in the pore solution of hardned cement paste on chloride binding.”
Cement and Concrete Research, 19, 683-691.
Troxell, G. E., Davis, H. E. and Kelly, J. W. (1968). Composition and Properties of

Concrete. McGraw-Hill, Inc., Second Edition.
Wang, K, Jansen, D.C., Shah, S.P. (1997), “Permeability Study of Cracked

Concrete”, Cement and Concrete Research, 27(3), 381-393.
Wee, T.H., Wong, S.F., Swadiwudhipong, S., and Lee, S.L. (1997). “A Prediction
Method for Long-Term Chloride Concentration Profiles in Hardened Cement
Matrix Materials.” ACI Materials Journal, 94(6), 565-576.
Xi, Y. (1991) Analysis of Concrete Creep, Shrinkage and Fracture by Deterministic

and Probabilistic Methods. Ph.D. Dissertation, Northwestern University,
177
Evanston, Il.
Xi, Y. (1995a). “A model for moisture capacities of composite materials, Part I:

formulation.” Computational Materials Science, 4, 65-77.
Xi, Y. (1995b). “A model for moisture capacities of composite materials, Part II:
application to concrete.” Computational Materials Science, 4, 78-92.
Xi, Y. and Ababneh, A. (2000) “Prediction of the Onset of Steel Corrosion in

Concrete by Multiscale Chloride Diffusion.” Proc. of the International
Symposium on High Performance Concrete: Workability, Strength and
Durability, December 10-15, Hong Kong & Shenzhen, China, pp. 181-186.
Xi, Y. and Jennings, M. (1997). “Shrinkage of Cement Paste and Concrete Modeled
by a Multi-scale Effective Homogeneous Theory.” Materials and Structures
(RILEM), 30, 329-334.
Xi, Y., and Bazant, Z.P. (1999). “Modeling Chloride Penetration in Saturated
Concrete.” Journal of Materials in Civil Engineering, 11(1), 58-65.
Xi, Y., and Jennings, H. M. (1992). “Relationships Between Microstructure and

Creep and Shrinkage of Cement Paste.” Material Science of Concrete, III, Ed. J.
Skalny, The Amer. Cer. Soc., Westerville, OH, 37-69.
Xi, Y., Bazant, Z.P., Molina, L. and Jennings, H.M. (1994b). “Moisture Diffusion in
Cementitious Materials: Moisture Capacity and Diffusivity.” Advanced Cement
Based Materials, 1, 258-266.
Xi, Y., Bazant, Z.P., Molina, L., and Jennings, H.M. (1994a). “Moisture Diffusion in
Cementitious Materials: Adsorption isotherm.” Advanced Cement Based
Materials, 1, 248-257.
Xi, Y., Willam, K., and Frangopol, D. M. (2000). “Multiscale Modeling of Interactive
Diffusion Processes in Concrete.” Journal of Engineering Mechanics, ASCE,
126(3), 258-265.
Zhen-hai, G. and Xiu-qin, Z. (1987). “Investigation of complete stress-deformation

curves for concrete in tension.” ACI Materials Journal, 84(4), 278-285.
Zuber, B., Marchand, J., Delagrave, A. and Bournazel, J. P. (2000). “Ice Formation
Mechanisms in Normal and High-Perfrormanc Concrete Mixtures.” Journal of
Materials in Civil Engineering,12(1),16-23.
APPENDIX A
ADSORPTION ISOTHERMS AND MOISTURE CAPACITIES OF
CEMENT PASTE AND AGGREGATE
The adsorption isotherm of porous media (the relationship between moisture
content and relative humidity at a constant temperature) can be described by the BSB
theory (Brunauer et al. 1969) as follows:
C k Vm H
w = (A.1)
(1 − kH )[1 + (C − 1 )kH ]
The moisture capacity can be obtained by deriving the adsorption isotherm with
respect to H as follows:
dw C k V m + wk [1 + (C − 1 )kH ] − wk [(C − 1 )(1 − kH ) ]

= (A.2)
dH (1 − kH )[1 + (C − 1 )kH ]
where, Vm is the monolayer capacity which is the mass of adsorbate required to cover
the adsorbent with a single molecular layer, C is the net heat of adsorption, which is a
function of temperature, and k determines the degree of saturation of the pores. The
three material parameters Vm, C and k can be evaluated by using adsorption test data
of the material.
179
Adsorption Isotherm and Moisture Capacity of Cement Paste
The adsorption isotherm and moisture capacity of cement paste can be
calculated using Eq. (A.1) and (A.2) and the following formulas for the Vm, C and k
parameters (Xi 1995b):
⎛ 0.22 ⎞⎛ w⎞ w
Vm = ⎜ 0.068 − ⎟⎜ 0.85 + 0.45 ⎟Vct t > 5 days, 0.3 < < 0.7 (A.3)
⎝ t ⎠⎝ c⎠ c
in which t is the time in days; w/c is the water-cement ratio; and Vct is for the type of
cement, Vct = 0.9, 1, 0.85 and 0.6 for type I, II, III and IV cements, respectively. The
model for parameter C is
⎛C ⎞
C = exp⎜ 0 ⎟ (A.4)
⎝T ⎠
in which T is the temperature in Kelvin; and C0 = 855. The model for parameter k is
(1 − 1 / n) C − 1
k= (A.5)
C −1
in which n is the number of the adsorbed layers in the pore and it is a function of t (in
days), water cement ratio, w/c, and type of cement (Nct = 1.1, 1.0, 1.15 and 1.5 for
type I, II, III and IV cements, respectively):
⎛ 15 ⎞⎛ w⎞ w
n = ⎜ 2.5 + ⎟⎜ 0.33 + 2.2 ⎟ N ct t > 5 days, 0.3 < < 0.7 (A.6)
⎝ t ⎠⎝ c⎠ c
180
Adsorption Isotherm and Moisture Capacity of Aggregate
The adsorption isotherm and moisture capacity of aggregates can be
calculated using the same BSB theory, and thus, Eqs. (A.1), (A.2), (A.4) and (A.5)
are still valid, but the monolayer capacity, Vm and the number of the adsorbed layers,
n, are different. The following formulas may be used for parameters Vm and n of
aggregate (Xi 1995b):
Vm = 0.0647Vagg (A.7)
n = 4.063nagg (A.8)
in which Vagg and nagg depend on the pore structures of the aggregates. For dense
aggregates, Vagg ranges from 0.05 to 0.1 and nagg from 1.0 to 1.5. For porous
aggregates, Vagg ranges from 0.01 to 0.04 and nagg from 1.7 to 2.0. It should be noted
that dense aggregates have larger Vagg and smaller nagg because of their small size
pores and high surface area. Dolomite, granite, crushed limestone and river gravel
are examples of dense aggregates. Lightweight aggregates and river sand are
examples of porous aggregates.

APPENDIX B
VOLUME PERCENTAGES OF DIFFERENT HYDRATION PRODUCTS
Jenning and Tennis model (1994) considers five different phases of cement
paste: anhydrous cement particles, inner hydration products, outer hydration products,
capillary pores and calcium hydroxide, CH (include AFm phase). The volume of C-S-
H gel is the sum of the volumes of inner and outer hydration products.
VC − S − H = Vinner + Vouter (B.1)
in which Vinner and Vouter are the volumes of inner and outer hydration products. The
volume fraction of C-S-H gel can be written as:
Vtotal − V1 − V pores
f C −S −H = = 1 − f1 − f pores (B.2)
Vtotal
in which Vpores and fpores are the volume and volume fraction of capillary pores and V1
and f1 are the volume and volume fraction of anhydrous cores of cement particles and
all the crystalline phases such as calcium hydroxide (CH) and ettringite.
The volume fraction of capillary pores can be written as follows:
f pores = ((1 − c) − cΩ ) p (B.3)
in which, c and p parameters depend on the water-cement ratio, w/c and the densities
182
of cement, ρc and water, ρw as follows:
1
c= (B.4)
1 + ( w / c)
(1 + (w / c))ρ c
p= (B.5)
1 + ( w / c) * ( ρ c / ρ w )
And Ω is a parameter depends on the degree of hydration of cement and can be
written as follows:
Ω = 0.437α C 3 S W C 3 S + 0.503α C 2 S W C 2 S + 0.397α C 3 AW C 3 A + 0.136α C 4 AF W C 4 AF (B.6)
in which WC3S, WC2S, WC3A, and WC4AF are the initial weight percentages of the
compounds of the cement. Also, αC3S, αC2S, αC3A, and αC4AF are the degree of
hydration of the compounds of the cement. The initial weight percentages of the
cement compounds, Wi (i = C3S, C2S, C3A and C4AF) depends on the type of the
cement. The degree of hydration of the cement compounds, αi (i = C3S, C2S, C3A and
C4AF) can be determined as follows:
αi = 1 − e − a ( t − b )
ci
i i
(B.7)
in which ai, bi and ci (i = C3S, C2S, C3A and C4AF) are constants determined
experimentally (Taylor et. al 1987) and shown in Table (B.1).

183
Table (B.1): Parameters related to the hydration of different compounds of cement.
Compound ai bi ci
C3S 0.25 0.90 0.70
C2S 0.46 0.00 0.12
C3A 0.28 0.9 0.77
C4AF 0.26 0.90 0.55
Also, f1 is the volume percentage of the anhydrous cores of cement particles, calcium
hydroxide (CH) crystals and ettringite as follows:
f 1 = f core + f CH + f AFm (B.8)
The volume fraction of the anhydrous cores of cement particles can be determined by
the following formula:
⎛ 1 −ψ ⎞
f core = c⎜⎜ ⎟⎟ p (B.9)
⎝ ρc ⎠
in which c and p are determined by Eqs. (B.4) and (B.5), ρc is the density of cement
and ψ is a factor depends on the degree of hydration of the compounds of the cement
and can calculated based weight average as follows:
ψ = α C 3SWC 3S + α C 2 SWC 2 S + α C 3 AWC 3 A + α C 4 AFWC 4 AF (B.10)
The volume fraction of CH crystals and AFm phase can be determined as follows:
f CH = c(0.189α C 3 S WC 3 S + 0.058α C 2 S WC 2 S ) p (B.11)

184
f AFm = c(0.849α C 3 AWC 3 A + 0.472α C 4 AF WC 4 AF ) p (B.12)
The volume fraction of the other twp phases can be calculated as follows:
f 3 = [0.706( w / c) + 0.131](1 − f1 − f pores ) + f pores (B.13)
f 2 = 1 − f1 − f 3 (B.14)
APPENDIX C
MIX DESIGN FOR THE EXPERIMENTAL PROGRAM
The description of the materials used in the concrete mixes, are as follows:
Coarse Aggregate (CA) or Gravel
Size, No. 4 to ¾ in., angular aggregate.
Absorption (%) = 0.384
SG (bulk dry) = 2.8
SG (bulk saturated surface dry) = 2.811
Moisture content (%) = 0.242
Dry-rodded unit weight (lb/ft3) = 96.81 [1550 Kg/m3]
Fine Aggregate (FA) or Sand
Fineness modulus (FM) = 2.93
Absorption (%) = 0.8
SG (bulk dry) = 2.153
SG (bulk saturated surface dry) = 2.305
Moisture content (%) = 0.351
Concrete Mix Design
This mix was designed according to Goldbeck and Gray Method (see the book
Materials for Civil and Highway Engineers, Kenneth N. Derucher and Conrad P.
Heins, 1981, pp. 153).
b: solid volume of coarse aggregate in a unit volume of concrete.
bo: solid volume of coarse aggregate in a unit volume of dry-rodded coarse aggregate.
186
Solid weight of CA = SG (bulk dry) *unit weight of water
= 2.8*62.4 = 174.72 lb/ft3 [2800 Kg/m3]
So,
dry − rodded unit weight 96.81

b0 = = = 0.554
solid weight 174.72
• From table 6-4 in page 155 (Derucher et al. 1981) and by using the size of coarse
aggregates (No. 4 to ¾ in) and fineness modulus of fine aggregates (2.93), the
value of b/ bo is 0.61.
• b can be calculated as follows:
b
b= * bo = 0.61 * 0.554 = 0.338
bo
• From table 6-5 page 156 in the above reference, and by using the size of CA
aggregate (No. 4 to ¾ in), angular in shape, slump of 3” and 28-day compressive
strength, ( f c' ) of 4333 psi [30 MPa],
¾ The required mixing water = 40 gal/yd3 (≈333lb/yd3) [198 Kg/m3]
333
Volume of mixing water = = 5.337 ft3/yd3
62.4
¾ The required cement = 6.4 sacks/ yd3 (≈604 lb/yd3) [358 Kg/m3]
604
Volume of cement = = 3.073 ft3/yd3
3.15 * 62.4
Volume of the entrapped air ≈ 2 % *27 = 0.54 ft3/yd3
333
water-cement ratio, w / c = ≅ 0.55
604
The volume of the coarse aggregate (CA) = 0.338*27 = 9.126 ft3/yd3

187
¾ The weight of the coarse aggregate (CA) = 9.126*174.72 ≈ 1595 lb/yd3
[946 Kg/m3]
The volume of fine aggregate (FA) = 27-volume of cement-volume of water-
volume of air-volume of coarse aggregate
= 27-6.4*0.48-40/7.5-0.02*27-9.126
= 8.929 ft3/yd3
¾ The weight of the fine aggregate (FA) = 8.929*2.153*62.4
≈ 1200 lb/yd3 [712 Kg/m3]
9.126 + 8.929
The aggregate volume fraction, g i = = 0.67
27
Design for certain water-cement ratio and different aggregate volume fractions
(w/c = 0.55 and gi = 0.55, 0.65 and 0.75)
The total volume of concrete, VT:
VT = Vw + Vc + VCA + VFA + Vair

123 1424 3 {
Vcp Vagg 2%
VT = Vcp + Vi + V air
dividing by VT :
1 = g cp + g i + 0.02
for gi = 0.55,
g cp = 0.43
Also, the volume of cement paste, Vcp :
Vcp = g cp *VT = 0.43 * 27 = 11.61 ft3/yd3

188
The volume of cement paste, Vcp:
Ww Wc W *w/c Wc W 1
Vcp = V w + Vc = + = c + = c (w / c + )
ρw 3.15 ρ w ρw 3.15 ρ w ρ w 3.15
And so the weight of cement, Wc:
Vcp * ρ w
Wc =
( w / c + 1 / 3.15)
For w/c = 0.55, the weight of cement:
Wc = 71.934 * Vcp
For gi = 0.55, the weight of cement :
¾ Wc = 71.934 *11.61 = 835 lb/yd3 [495 Kg/m3]
The volume of cement:
Wc 835
Vc = = = 4.248 ft3/yd3
3.15 * ρ w 3.15 * 62.4
The weight of water:
¾ Ww = w / c *Wc = 0.55 * 835 = 459 lb/yd3 [272 Kg/m3]
The volume of water:
Ww 459
Vw = = = 7.356 ft3/yd3
ρw 62.4
The volume of aggregate (Coarse and fine):
Vagg = 27 − Vcp − Vair = 27 − 11.61 − 0.02 * 27 = 14.85 ft3/yd3
Assuming that the VCA = VFA = 14.85/2 = 7.425 ft3/yd3:
The weight of gravel:
¾ WCA = VCA * 2.8 * 62.4 = 1297 lb/yd3 [769 Kg/m3]

189
The weight of sand:
¾ W FA = V FA * 2.153 * 62.4 = 998 lb/yd3 [592 Kg/m3]
By following the same procedure mentioned previously the mix design for w/c= 0.55,
gi= 0.65 and 0.75 can be obtained. The results of the mix design for w/c = 0.55, gi=
0.55, 0.65 and 0.75 are shown in the Table (C.1) below.
Design for certain aggregate volume fraction and different water-cement ratios
(gi = 0.65 and w/c = 0.45, 0.55 and 0.65)
for gi = 0.65:
g cp = 0.33
Also, the volume of cement paste, Vcp :
Vcp = g cp * VT = 0.33 * 27 = 8.91 ft3/yd3
The volume of aggregate (Coarse and fine):
V agg = 27 − V cp − V air = 27 − 8.91 − 0.02 * 27 = 17.55 ft3/yd3
Assuming that the VCA = VFA = 17.55/2 = 8.775 ft3/yd3:
The weight of gravel:
¾ WCA = VCA * 2.8 * 62.4 = 1533 lb/yd3 [909 Kg/m3]
The weight of sand:
¾ W FA = V FA * 2.153 * 62.4 = 1179 lb/yd3 [699 Kg/m3]
And the weight of cement, Wc:

190
Vcp * ρ w Vcp * 62.4

Wc = =
1 1
(w / c + ) (w / c + )
3.15 3.15
The weight of cement for w/c = 0.55 :
8.91 * 62.4
Wc = = 641 lb/yd3 [380 Kg/m3]
(0.55 + 0.317)
The weight of water for w/c = 0.55 :
Ww = w / c * Wc = 0.55 * 641 = 353 lb/yd3 [209 Kg/m3]
By following the same procedure mentioned previously the mix design for gi = 0.65,
w/c = 0.45 and 0.65 can be obtained. The results of the mix design for gi = 0.65, w/c
= 0.45, 0.55 and 0.65 are shown in Table (C.1) below.
Table C.1: Concrete mix design proportions.
Mix 46 56 66 55 57
water-cement ratio (w/c) 0.45 0.55 0.65 0.55 0.55
aggregate volume fraction (gi) 0.65 0.65 0.65 0.55 0.75
weight of water (Kg/m3) 193* 209* 222* 272* 146*
weight of cement (Kg/m3) 430 380 341 495 265
weight of sand (Kg/m3) 699* 699* 699* 592* 806*
weight of gravel (Kg/m3) 909* 909* 909* 769* 1049*
* Values need to be corrected to account for the moisture condition of aggregates.
Correction for moisture content of aggregates
The proportions of water, coarse and fine aggregates are based on their dry
weights. In reality aggregates are not dry (unless dried) and have certain moisture
191
content. The amount of water calculated based on the mix design should be
completely available for cement hydration during the hardening of concrete. Thus, the
amount of water absorbed by aggregates after the deduction of moisture content must
be added. This is correct if the moisture content of aggregate is less the amount of
water required to reach saturated surface dry condition (SSD). If the aggregates are
wet the amount of mixing water need to be decreased by the amount of free additional
water. Also, the proportions of coarse and fine aggregates must be adjusted to
compensate for the water content.
As an example for mix 46 (w/c = 0.45 and gi = 0.55) the additional amount of
water need to be added is:
From Coarse Aggregates (CA) = (0.00384 − 0.00242 ) * 909 = 1.25 Kg
From the Fine Aggregate (FA) = (0.008 − 0.00351) * 699 = 3.14 Kg
The amounts of aggregates need to be added are:
Coarse Aggregate (CA) = 0.00242*909 = 2.2 Kg
Fine Aggregate (FA) = 0.00351*699 = 2.5 Kg
For mix 46, the new mix proportions are, (193+1.25+3.14≈197 Kg/m3) for the
water, (699+2.5≈702 Kg/m3) for sand (FA) and (909+2.2≈911 Kg/m3) for gravel
(CA). All mix proportions for the five mixes are corrected and tabulated again in
Table (C.2).
192
Table C.2: Mix proportions used in the experimental study of Chapters 5, 6 and 7.
Mix 46 56 66 55 57
water-cement ratio (w/c) 0.45 0.55 0.65 0.55 0.55
aggregate volume fraction (gi) 0.65 0.65 0.65 0.55 0.75
weight of water (Kg/m3) 197 213 226 276 151
weight of cement (Kg/m3) 430 380 341 495 265
weight of sand (Kg/m3) 702 702 702 594 809
weight of gravel (Kg/m3) 911 911 911 771 1052
Table C.3: Concrete mix proportions used in experimental study of the effect of
moisture diffusion on chloride penetration (Chapter 7)
water-cement ratio (w/c) 0.65

3
weight of water (Kg/m ) 232
weight of cement (Kg/m3) 356
weight of sand (Kg/m3) 847
weight of gravel (Kg/m3) 1031
Table C.4: Proportions of the concrete mix used in the experiment of the effect of
induced-shrinkage cracking on chloride penetration (Chapter 8)
water-cement ratio (w/c) 0.55

aggregate volume fraction (gi) 0.55
weight of water (Kg/m3) 244
weight of cement (Kg/m3) 448
weight of sand (Kg/m3) 593
weight of gravel (Kg/m3) 771
APPENDIX D
COMPRESSION TEST RESULTS
Concrete cylinders of 10x20cm (4”x8”) were casted, remolded after 24 hours
and then cured in a moisture room for 28 days. After curing, all specimens were
capped using a suitable capping material. The compression test of the concrete
cylinders was performed using a universal MTS loading machine, with maximum
capacity of 484 KN (110 Kips). In this test was, a compressive axial load was applied
on the concrete cylinder at rate of 2 (Kn/sec) until failure. The maximum loads were
recorded and the compressive strength was calculated by dividing the average of
maximum loads by the cross-sectional area of the specimen.
The following are the compression test results of concrete specimens prepared
from the mixes used in the experimental studies of Chapters 5 through 8.
Table D.1: Compression test results for the concrete mixes used in the experimental
study of Chapters 5, 6 and 7.
Sample # Maximum load (KN)

Mix-46 Mix-56 Mix-66 Mix-55 Mix-57
1 303 265 226 226 245
2 307 267 233 268 265
3 315 280 235 276 279
4 323 283 248 287 286
5 325 297 255 310 306
Average 314.6 278.7 239.5 273.5 276.4
f c' (MPa) 38.5 34.1 29.3 33.5 33.8
194
Table D.2: Compression test results for the concrete used in the experiment of the
effect of moisture diffusion on chloride penetration (Chapter 7).
Sample number Maximum load (KN)

1 156
2 178
3 200
4 200
5 214
Average 189.6
'
f (MPa)
c
23.40
Table D.3: Compression test results for the concrete used in the experiment of the
effect induced-shrinkage cracking on chloride penetration (Chapter 8).
Sample number Maximum load (KN)

1 267
2 298
3 276
4 260
Average 275
' 34
f (MPa)
c
APPENDIX E
EXPERIMENTAL PROCEDURE FOR OBTAINING CHLORID PROFILES
After certain exposure period, the specimens were cleaned by removing the
insulation material and salts and then dried at room conditions for one day. After
drying, the surfaces of specimens were cleaned from salt deposits by using a wire
brush.
Collecting Dust Samples
Concrete dust samples were collected from certain depth ranges by using the
drill press machine shown in Fig. (E.1). Drilling started from the side exposed to
chloride penetration and extends at 1/4” ( ≈ 6 mm) depth increments.
Fig. E.1: Drilling of concrete sample.

196
Electro-chemical Analysis
The electro-chemical analysis of powder samples collected from the concrete
specimens resulted in a profile of total chloride concentration as weight percentage of
concrete weight. The test kit for electro-chemical analysis (manufactured by
Germann Instruments, Inc.) is shown in Fig. (E.2). It mainly consists of chloride
electrode and electrometer for mV (mille Volt), calibration liquids 0.005%, 0.02%,
0.05% and 0.50% Cl-, and chloride ions extraction liquid.
Fig. E.2: Chloride concentration measurement kit.
The electro-chemical analysis method procedure is described as follows:
1. A 1.5 gram of the collected dust sample was added to a vial containing 10
ml of chloride extraction liquid as shown in Fig. (E.3).

197
Fig. E.3: Sampling of 1.5 gram of concrete dust and the extraction liquid vial.
2. The vials containing the concrete dust and the extraction liquid were
secured by a lid and shaken for 5 minutes.
3. The electrode was submerged into a vial containing calibration solution
(know chloride concentration solution) and the mille Volt (mV) reading
displayed by the electrometer was recorded.
4. Four solutions of 0.005%, 0.02%, 0.05% and 0.5% chloride concentration
were used to plot the calibration curve.
5. The electrode need to be cleaned by spraying with distilled water and
wiped after each use.
6. The electrode was submerged into the vials containing the concrete dust
and extraction liquid and the measured mV-readings were recorded.
7. The measured mV-readings were converted to total chloride concentration
by using the calibration curve.

198
The following are the results of the electro-chemical analysis of the
experimental studies in chapters 5, 6, 7 and 8.
Table E.1: The mV readings after 400 freezing-thawing cycles.
Mix Sample # Depth range (mm)

0-6 6-13 13-19 19-25 25-32 32-38 38-45
1 -28.8 -5 0.7 -4.5 -1.7 0.8 6.9
2 -26.1 -4.7 1.3 -3.8 -0.8 0.9 6.9
46
3 -25.9 -4.6 2 -3.6 -0.7 1.1 8.3
1 -28.9 -19 -13 -8.6 -5.3 -4 -3.6
2 -27.8 -18.4 -12.8 -8.1 -4.9 -3.3 -3.2
56
3 -27.8 -17.6 -12.7 -7.9 -4.7 -3.1 -2.8
1 -29.9 -15.3 -10 -5.2 - - -
2 -29.4 -14.3 -9.9 -3.4 - - -
66
3 -28.3 -13.6 -8.4 -2.8 - - -
1 -30.4 -16.9 -13.5 -11.3 -9.9 -5.1 -
2 -30.1 -16.5 -12.1 -10.7 -8.7 -4.7 -
55
3 -29.3 -16.3 -11.7 -9.6 -7.4 -4.6 -
1 -27.7 -11.2 0.6 - - - -
2 -27 -7.4 0.8 - - - -
57
3 -26.3 -6.9 1.3 - - - -
Table E.2: The mV readings for the standard solutions.
% Cl- mV reading
0.005 77.4
0.02 51.3
0.05 42.5
0.5 -5.2
199
1
Chloride conc. (% by weight)
0.1
0.01
0.001
-10 0 10 20 30 40 50 60 70 80 90 100
mV
Fig. E.4: Calibration curve for obtaining chloride concentrations.

200
Table E.3: The mV readings after 50 days of exposure.
Depth range (mm)

Mix Sample #
0-6 6-13 13-19 19-25 25-32 32-38
1 -21.4 2.8 22.4 30.4 66.1 -
46 2 -20.2 -5.5 15.2 36.6 72.1 -
3 -17.3 -3.4 7.9 36.7 41.9 -

1 -20.7 -6.2 11.2 39 45 -
56 2 -21 -5.6 10.4 35.9 47.4 -
3 -20.8 -5.8 8.3 40.3 48.5 -

1 -19.2 -1.2 8.9 23.3 26.2 31.8
66 2 -20.4 -2.7 17.2 30.9 31.3 29.3
3 -18.3 -2.8 21.5 33.4 35 32

1 -30.5 -10.1 27.1 46.2 53.9 64.3
55 2 -33.9 -13.4 19.4 58.5 37.1 54.9
3 -31.6 -12.5 20.2 67.6 70.7 52.8

1 -22.5 -12.3 -9.8 -11.9 2.7 7.1
57 2 -20.3 -13.9 -10.8 -1.2 2.9 6.3
3 -23.2 -11.8 -10.4 -0.4 2.4 6.2

201
Sample Depth range (mm)

Mix
#
0-6 6-13 13-19 19-25 25-32 32-38 38-45 45-51
1 -22.8 -1.3 17 32.2 51.6 48.5 65.3 57.7
46 2 -22.3 -1.2 18 34.1 54.2 49.4 68.4 59.8
3 -20.4 -1.1 16 35.3 57.3 49.6 68.5 59.9
1 -18.5 -5.5 8.2 17.9 31.2 37.9 46.1 46.8
56 2 -17.9 -6.8 8.8 18.6 31.3 38.8 48.5 50.1
3 -17.6 -4.2 9.4 19.2 31.9 40.5 52.4 53
1 -21.9 -3.6 13 20 22.7 33 25.3 26.9
66 2 -21.6 -3.3 13 20.4 26.1 33.4 26.2 27.2
3 -21.1 -2.9 13.7 21.8 26.9 33.8 26.4 28.3
1 -33.4 -13.7 6.1 35.2 50.5 66.7 73.3 51.3
55 2 -29.5 -13.4 6.6 35.4 54 67 74.5 61
3 -28.5 -11.3 7 37 54.7 68 75.7 64.5
1 -20.3 -7.4 3.6 9.3 22 23.2 41 -
57 2 -19.8 -6.3 4 10.2 22.7 26.3 41.2 -
3 -19.8 -6.1 4.7 10.4 23.6 27 41.8 -

202
Depth range (mm)

Mix Sample #
0-6 6-13 13-19 19-25 25-32 32-38 38-45 45-51
1 -25.7 -11.1 -5.5 3.2 11.8 24.7 56.6 63.3

46
2 -24.3 -11.1 -4 3.3 14.8 30.7 62.6 65.2
3 -23.3 -8 -2.9 4.6 15.7 32.8 64.1 68.2
1 -31.3 -10 4 19.6 25.5 25 21.4 -

56
2 -28.6 -9 4.5 30 26 26.4 21.6 -
3 -28.5 -8.1 5.2 20.1 27.1 26.9 22.4 -
1 -31.4 -16.6 -11.9 -8.3 -4.1 -2.5 -1.6 -

66
2 -31.1 -16.5 -10.9 -8 -2.6 -0.2 -1.4 -
3 -31 -15.6 -8.6 -7.9 -2.4 0.3 0.1 -
1 -32.3 -12.6 2.4 25.3 44.4 25.7 44.3 -

55
2 -29.8 -12.5 4.1 27.3 45.7 35.5 44.8 -
3 -27.5 -11.5 5 27.5 46.2 36.1 44.9 -
1 -23.8 -8.6 -2.5 1.7 5.4 11.7 15.3 -

57
2 -23 -6.7 -2.4 2 6.5 12.4 15.4 -
3 -22.7 -6.4 -1.7 2.2 6.9 12.7 16.1 -

203
mV reading
% Cl-
50-days 100-days 200-days
0.005 99.4 91.3 82.5
0.02 60.3 60.8 54.4
0.05 53.3 52.8 45.85
0.5 -2.2 -4.3 -7.5
1
0.1
50 days
100 days
200 days
0.01 200 days
100 days
50 days
0.001
-10 0 10 20 30 40 50 60 70 80 90 100
mV
Fig. E.5: Calibration curves for obtaining chloride profiles.

204
Initially dry Initially saturated

Depth range
(mm) sample number sample number
1 2 3 1 2 3
0-6 15.7 15.1 14.3 18.2 19.7 25.9
6-13 25.1 27.3 24.9 34.1 42.3 42.4
13-19 35.7 41.1 38 59.6 60.2 73.3
19-25 69.4 60.8 68.5 77.5 84.5 86.2
25-32 94.4 87.6 93.3 84.6 87.2 102.7

Depth range
1 2 3 1 2 3
0-6 14.9 16.7 17.1 18.1 19.7 22.7
6-13 25.1 25.8 26.4 35 35.2 35.2
13-19 37.7 40.8 41 50.5 52.7 54.1
19-25 55.4 55.6 56.4 80.7 82.6 82.7
25-32 61.4 62.1 64.1 94.3 95.9 96.5

205

Depth range
1 2 3 1 2 3
0-6 2.9 3.1 3.5 9.4 10.6 11.2
6-13 10.3 11.4 12.3 18.8 20.8 20.8
13-19 18.9 19.8 22.1 31 31.8 32.1
19-25 24.1 24.8 25.6 46.1 46.1 46.4
25-32 26.8 32 32.7 59.4 60.8 62.3
mV reading
% Cl-
30-days 90-days 180-days
0.005 103.4 104.2 87.1
0.02 76 81.2 57.6
0.05 52.5 59.3 50.1
0.5 -2 2.7 -5.3
206
0.1
30 days
90 days
180 days
0.01
180 days
90 days
30 days
0.001
-10 0 10 20 30 40 50 60 70 80 90 100
mV
Fig. E.6: Calibration curves for obtaining chloride concentrations.
Depth range Restrained shrinkage Free shrinkage

1 2 3 1 2 3
0-6 -10.1 -8.2 -7.8 -11.7 -9.6 -7.4
6-13 0.3 11.6 13.9 2.2 3.4 5.2
13-19 8.2 8.6 16.1 10.6 11.8 15.4
19-25 16.5 16.6 16.7 21.1 21.6 21.7
25-32 24.1 26.3 28.8 26.7 27.3 34.5
32-38 25.7 31.1 35.3 31.1 32.4 34.5
207
% Cl- mV reading
0.005 71.9
0.02 60.3
0.05 53.4
0.5 -3.4
1
0.1
0.01
0.001
-10 0 10 20 30 40 50 60 70 80 90 100
mV
Fig. E.7: Calibration curve for obtaining chloride concentrations.
View publication stats

Ayman Ababneh PHD Thesis

Uploaded by

Copyright:

Available Formats

Ayman Ababneh PHD Thesis

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ayman Ababneh PHD Thesis

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

The Coupled Effect of Moisture Diffusion, Chloride Penetration, and Freezing-

Thesis · December 2002

The user has requested enhancement of the downloaded file.

Ayman Nureddin Ababneh

B.Sc., Jordan University of Science and Technology, 1991

M.Sc., Jordan University of Science and Technology, 1994

A thesis submitted to the

Faculty of the Graduate School of the

University of Colorado in partial fulfillment of the requirement for the degree of

Department of Civil, Environmental and Architectural Engineering

This dissertation entitled:

The Coupled Effect of Moisture Diffusion, Chloride Penetration and Freezing-Thawing

written by Ayman Nureddin Ababneh

has been approved for the Department of

Civil, Environmental and Architectural Engineering

HRC protocol # ___________________

Ababneh, Ayman Nureddin (Ph.D., Civil, Environmental, and Architectural Engineering)

The Coupled Effect of Moisture Diffusion, Chloride Penetration and Freezing-Thawing

Thesis directed by Associate Professor Yunping Xi

of the processes related to concrete durability. Those processes include moisture

diffusion, chloride penetration, freezing-thawing and their coupled effects.

numerical solution. The governing equations of moisture diffusion and chloride

penetration in concrete were characterized by the nonlinear diffusion equation derived

Diffusion mechanisms and shrinkage mechanisms are considered at the nanoscale; a

composite mechanics method (generalized self-consistent method, in particular) is used at

numerical models were solved by the Alternating-Direction Implicit (ADI) finite

the time for rehabilitation.

thawing and chloride penetration, effect of concrete material parameters on chloride

effect of shrinkage-induced cracking on chloride penetration. The experimental study

and the future studies.

this work possible.

I would like to express my deep appreciation and thanks to my advisor Professor

Grateful acknowledgment is made to Professors Kaspar Willam, Dan Frangopol,

I would like to extend my appreciation to my colleagues Ammar Nakhi, Zhaohui

encouragement without which this work has never been accomplished.

Finally the financial support under NSF grant CMS-9872379 to University of

Colorado at Boulder is gratefully acknowledged.

1.1 Durability of Concrete Structures …….………………………………1

2 BACKGROUND INFORMATION AND LITERATURE REVIEW

2.1 Transport Properties of Concrete …..…………………..……………10

3 COUPLED MOISTURE DIFUSSION AND DRYING SHRINKAGE

3.1 Introduction ……….. ….………………………………......……..… 25

3.6.1 Effect of damage ………………………………………….. 61

4 COUPLED MOISTURE AND CHLORIDE DIFFUSIONS

4.1 Introduction ………..……………………………………….……… 69

5 THE COUPLED EFFECT OF FREEZING-THAWING AND CHLORIDE

5.1 Introduction ………..……………………………………….……… 98

6 EXPERIMENTAL STUDY ON THE EFFECT OF CONCRETE MATERIAL

6.1 Introduction ………..……………………………………….……...120

7 EXPERIMENTAL STUDY ON THE COUPLING EFFECT OF MOISTURE

8 EXPERIMENTAL STUDY ON THE EFFECT OF SHRINKAGE-INDUCED

8.3.1 Shrinkage strain………………………………………….....157

9 CONCLUSIONS AND FUTURE RESEARCH RECOMMENDATIONS

BIBLIOGRAPHY ………………..………………………………......…..….. 169

APPENDIX A: ADSORPTION ISOTHERMS AND MOISTURE CAPACITIES OF

APPENDIX B: VOLUME PERCENTAGES OF DIFFERENT HYDRATION

APPENDIX C: MIX DESIGN FOR THE EXPERIMENTAL

APPENDIX D: COMPRESSION TEST RESULTS ……….......…………..…193