GHFGHF

INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI films
the text directly from the original or copy submitted. Thus, some thesis and
dissertation copies are in typewriter face, while others may be from any type of
computer printer.
The quality of this reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality illustrations
and photographs, print bleedthrough, substandard margins, and improper
alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete manuscript
and there are missing pages, these will be noted. Also, if unauthorized
copyright material had to be removed, a note will indicate the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and continuing
from left to right in equal sections with small overlaps.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6 x 9 black and white
photographic prints are available for any photographs or illustrations appearing
in this copy for an additional charge. Contact UMI directly to order.
ProQuest Information and Learning
300 North Zeeb Road, Ann Arbor, Ml 48106-1346 USA
800-521-0600
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
HYDRATION KINETICS AND
THERMOMECHANICS OF BLENDED
CEMENT SYSTEMS
by
Ivindra Pane
A dissertation submitted in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
(Civil Engineering )
in The University of Michigan
2001
Doctoral Committee:
Associate Professor W. Hansen, Chairperson
Associate Professor C. Borgnakke
Professor V.C. Li
Professor J. Pan
UMI Number: 3029410
___
UMI
UMI Microform 3029410
Copyright 2002 by Bell & Howell Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
Bell & Howell Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, Ml 48106-1346
Ivindra Pane 2001
All Rights Reserved
For my beloved parents
S. Zuhdi Pane and Irawati Siregar
u
ACKNOWLEDGEMENTS
The author would like to express his sincere gratitude to his Advisor, Prof. Will
Hansen, for providing support, encouragements, and suggestions during this research.
The author also wish to thank the members of his thesis committee: Prof. Claus
Borgnakke, Prof. Jwo Pan, and Prof. Victor Li, for spending their time evaluating
this thesis and providing comments.
The author is indebted to his parents, S.Zuhdi Pane and Irawati Siregar, for
their support, guidance, and encouragements. Finally, helpful discussions and help
ing hands from the staff of Civil and Environmental Engineering Department and
the authors close friends including Elin Jensen, Phil Mohr, and Endra Susila, are
gratefully acknowledged.
iii
TABLE OF CONTENTS
D E D I C A T I O N ............................................................................................................... ii
ACKNOWLEDGEMENTS.................................................................................... iii
LIST OF F I G U R E S ................... viii
LIST OF T A B L E S ............................................................................................................ xv
CHAPTER
I. INTRODUCTION...................................................................................... 1
1 . 1 Background.................................... 1
1 . 2 Objectives, Scope, and Research Significance........................... 4
1.2.1 Research Objectives and Scope..................................... 4
1.2.2 Research Significance...................................................... 7
1.3 Thesis Organization......................................................................... 9
II. REVIEW ON BEHAVIOR OF HYDRATING CONCRETE 11
2.1 Hydration............................................................................................. 11
2.1.1 Measurement M e t h o d s .................................................. 11
2.1.2 Influencing f a c t o r s ............................................................ 12
2.1.3 Hydration K i n e t i c s ........................................................ 13
2.2 Microstructure of Cement Paste .................................................. 15
2.2.1 Pore Ch a r acterizat i ons.................................................. 15
2.2.2 Microstructure Related Behavior................................. 16
2.3 Mechanical and Thermal P r o p e r t i e s ........................................... 17
2.3.1 Strengths and Elastic Moduli........................................ 17
2.3.2 Creep and Relaxation ........................................ 18
2.3.3 Measurement of Stresses due to Restrained Defor
mation................................................................................... 2 0
2.3.4 Early Age Stress P r e d i ct io n s........................................ 21
2.3.5 Coefficient of Thermal Dilation (CTD) .................... 23
2.4 Deformations in Hydrating C o n c r e t e .......................................... 24
2.4.1 Drying S h r i n k a g e .................................................... 24
iv
2.4.2 Chemical Shrinkage......................................................... 25
2.4.3 Autogeneous Deformations............................................ 26
I I I . E X P E R I M E N T A L M E T H O D S ......................................................... 32
3.1 I n t r o d u c t i o n ........................................................................................ 32
3.2 Materials and Mix Design................................................................ 33
3.3 Hydration............................................................................................... 34
3.3.1 Thermal Analysis (DTA/TGA) .................................. 34
3.3.2 Isothermal and Semi-adiabatic Calorimetry.............. 37
3.4 Mechanical P r o p e r t i e s ....................................................................... 38
3.4.1 Coefficient of Thermal Dilation (CTD) ...................... 38
3.4.2 Tensile Creep Compliance ............................................ 39
3.4.3 Strength and Elastic M o d u l i ........................................ 43
3.5 Autogeneous S h r i n k a g e ............................................................. 43
3.6 Early Age Stress Development.................................................. . 44
3.7 Pore Structure Me as ur em en t .................................................... 46
IV. H Y D R A T I O N AND MI CRO ST RUCT URE OF BLENDED
C E M E N T S Y S T E M S ................................................................................ 49
4.1 I n t r o d u c t i o n ................................................................................. 49
4.2 Heat of Hydration and Thermal Decomposition.................... 51
4.2.1 Heat of Hydration............................................................ 51
4.2.2 DT A/ T GA......................................................................... 52
4.2.3 Effect of Temperature..................................................... 56
4.2.4 Heat (Q) and Combined Water ( w b ) ............................ 61
4.3 Degree of Hydration of Blended Systems and Pozzolans . . . 6 6
4.3.1 Procedure for Determining the Degree of Hydration 6 6
4.3.2 The Degree of Hydration of Fly Ash, Slag, and Silica
F u m e ................................................................................... 69
4.4 H2 O and N2 Sorption Isotherms ................................................... 74
4.4.1 Characterization of Isotherms ................................. 76
4.4.2 Surface Area of S o l i d ..................................................... 77
4.4.3 Surface area of Hydration P r o d u c t s .......................... 93
4.4.4 Pore Characteristics........................................................ 96
V. MIRO ME CHAN I SM S OF HUMIDITY-DRIVEN S H R I N K
AGE I N PO RTL AN D CEMENT PASTE ..................................... 100
5.1 I n t r o d u c t i o n ................................................................................... 100
5.2 Microstructure E v o l u t i o n .............................................................. 101
5.2.1 Microstructure Model........................................................ 102
5.2.2 Area of Capillary Pore W a l l ........................................... 103
v
5.3 Humidity-Driven Shr i n k a g e ............................. 108
5.3.1 Gibbs-Bangham S t r e s s ......................................................114
5.3.2 Disjoining P r e s s u r e ............................................................ 120
5.3.3 Capillary S t r e s s ................................................................... 122
5.4 Application to Drying S h r i n k a g e .................................................125
5.5 Application to Autogeneous Shrinkage....................................... 127
5.5.1 Pore Relative Humidity During Hydration ................. 127
5.5.2 Predictions............................................................................. 133
VI. P R O P E R T I E S DE VE LOPMENT AND E F F E C T S OF MIN
ERAL A D D I T I V E S ...................................................................................... 140
6.1 I n t r o d u c t i o n ........................................................................................ 140
6 . 2 Method of Calculating the Degree of Hyd r a ti on ......................... 142
6.2.1 Phenomenological Hydration M o d e l .............................. 142
6.2.2 Hydration Under Non-isothermal Conditions . . . . 142
6.2.3 Verifications of the Method...............................................144
6.3 Strength and Youngs M o d u l u s ....................................................... 151
6.4 Concept of Hydration T i m e ..............................................................159
6.5 Autogeneous Deformations..............................................................163
6 . 6 Coefficient of Thermal Dilation ( C T D ) ..........................................172
6.7 Viscoelastic Properties........................................................................172
6.7.1 Creep Compliance and Relaxation Modulus..................172
6.7.2 Dirichlet Series....................................................................... 176
6.7.3 Representation of Aging by Heat of Hydration . . . 184
6.7.4 Effects of Mineral Additives................................................ 185
V I I . EARLY AGE STRESS ANALYSIS OF HYDRATING CON
CRETE .......................... 192
7.1 I n t r o d u c t i o n ........................................................................................192
7.2 Uniaxial Stress Calculations............................................................. 193
7.2.1 Important Material Parameters ......................................193
7.2.2 Stress Calculation................................................................195
7.2.3 Verification with E x p e r i m e n t s .........................................197
7.2.4 Effect of Mineral Additives............................................... 210
7.2.5 Effect of Aging...................................................................... 210
7.3 Application to Foundation or Pavement S l a b s ............................ 212
7.3.1 Elastic Stress ...................................................................... 212
7.3.2 Viscoelastic Stress with Through-thickness Proper
ties Variation ..................................................................... 218
7.3.3 Viscoelastic Stress without Through-thickness Prop
erties Variation ..................................................................219
7.3.4 Application ......................................................................... 222
vi
V I I I . CONCLUSIONS, AND R E C O M M E N D A T I O N S ....................... 225
8.1 Conclusions............................................................................................. 225
8.2 Recommendations................................................................................229
B I B L I O G R A P H Y ..........................................................................................................230
vii
LIST OF FIGURES
Figure
2.1 Influences of intrinsic factors on degree of hydration [Bru.2].................27
2.2 Effect of temperature on degree of hydration [Bru.2]............................... 27
2.3 Development of the compressive strength with time [Byf.l].................. 28
2.4 Development of Youngs modulus with time [Thi.l]................................ 28
2.5 Schematic development of compressive strength, tensile strength,
and Youngs modulus with degree of hydration (Q/Qoo) [Onk.l]. . . 29
2.6 Basic creep compliance of concrete at early ages [Emb.l]...................... 29
2.7 Relaxation modulus of concrete at early ages [Wes.l]............................. 30
2.8 The stress history of a uniaxial early age concrete specimen [Emb.l]. 30
2.9 The dependency of CTD of mortar and concrete specimens on ag
gregate contents [Min.l] ............................................................................. 31
2.10 Development of autogeneous shrinkage in cement paste of w/c=0.4
[Sel.l]................................................................................................................... 31
3.1 DTA/TGA data and the measured hydration quantities....................... 36
3.2 Rate of heat of hydration measured by the isothermal calorimeter. . 38
3.3 Variations of concrete stain with temperature........................................... 40
3.4 Schematic of tensile creep testing.................................................................. 42
3.5 Schematic of measurement of autogeneous deformation......................... 45
3.6 Test set up for measurement of the early age stress................................. 47
viii
4.1 Isothermal heat of hydration at 15C.......................................................... 53
4.2 Isothermal heat of hydration at 23C. ................................................... 54
4.3 Isothermal heat of hydration at 34C.......................................................... 55
4.4 Chemically bound water of mix 45-1........................................................... 57
4.12 Relations between heat of hydration and chemically bound water. . 67
4.13 Relations between wbj00 and temperature, w/b=0.45......................... 69
4.14 Degree of hydration of blended cement pastes..................................... 70
4.15 Q/Qoo of blended cement pastes from short term heat measurement. 70
4.16 Relation between CH;oss and wb for OPC.................................................. 72
4.17 Relation between CHioss and wb for 75% OPC+25% FA....................... 72
4.18 Relation between CH;oss and wb for 75% OPC+25% GGB................... 73
4.19 Relation between CH;oss and wb for 90% OPC+10% SF....................... 73
4.20 Degree of pozzolan reaction....................................................................... 75
4.21 Water sorption isotherms of mix 45-1.................... 78
4.22 Water sorption isotherms of mix 45-2..................................................... 79
ix
4.23 Water sorption isotherms of mix 45-3................................. 80
4.24 Water sorption isotherms of mix 45-4......................................................... 81
4.25 Water sorption isotherms of mix 35-1......................................................... 82
4.26 Nitrogen sorption isotherms of mix 45-1..................................................... 83
4.27 Nitrogen sorption isotherms of mix 35-1..................................................... 84
4.28 Comparison between V/Vm and t /t m for mix 45-1................................... 87
4.29 Comparison between V/Vm and t / t m for mix 45-2................................... 8 8
4.30 Comparison between V/Vm and t / t m for mix 45-3................................... 89
4.31 Comparison between V/Vm and t / t m for mix 45-4................................... 90
4.32 Comparison between VfVm and t / t m for mix 35-1................................... 91
4.33 Sads for various mixes............................................. 94
4.34 V/Vm for various mixes of mature paste (older than 190 days). . . . 99
5.1 Fit of pore size distribution data [Win.2], w/c=0.4.................................... 104
5.2 Fit of pore size distribution data [Win.2], w/c=0.6....................................104
5.3 Parameter b obtained t h e o r e t i c a l l y . ............................................................ 108
5.4 Parameter b calculated from experiments..................................................... 109
5.5 Capillary pore wall area obtained theoretically..........................................109
5.6 Capillary pore wall area calculated from experiment................................. 110
5.7 The schematic of paste microstructure at capillary pore level.................I l l
5.8 Drying shrinkage data for mature pastes [Aim. 1]........................................113
5.9 Relation between drying shrinkage and A7 ..................................................113
5.10 Variations of A7 d and A7 with h................................................................... 116
x
5.11 A unit cell used to calculate cgb.....................................................................117
5.12 The composite model used to calculate Eg...................................................119
5.13 Aas a function of degree of hydration............................................................ 119
5.14 The effect of age on A from experiment [Alm.l].......................................... 120
5.15 Variations of eSh gb with h.......................................................................... 1 2 2
5.16 A unit cell used to calculate ec.........................................................................124
5.17 Ssp/ S p calculated using data in [Win.2 ]......................................................... 126
5.18 Prediction of drying shrinkage: w/c=0.4 and age=3 days.........................128
5.19 Components of shrinkage strain: w/c=0.4 and age=3 days..................... 128
5.20 Prediction of drying shrinkage: vv/c=0.4 and age=7 days........................ 129
5.21 Prediction of drying shrinkage: \v/c=0.4 and age=28-30 days. . . . 129
5.22 Prediction of drying shrinkage: w/c=0.6 and age=7 days........................ 130
5.23 Prediction of drying shrinkage: w/c=0 . 6 and age=30 days...................... 130
5.24 Schematic illustration of the inclusion of chemical shrinkage................... 132
5.25 Idealization of capillary pore geometry. ...................................................... 134
5.26 The unit cell of capillary pore..........................................................................134
5.27 Autogeneous shrinkage as function of time...................................................136
5.28 Autogeneous relative humidity as function of a ..........................................137
5.29 Autogeneous shrinkage as function of a ........................................................ 137
5.30 Contributions of Gibbs-Bangham strain and capillary strain................... 138
5.31 The effect of w/c on autogeneous relative humidity. ..................... 138
5.32 The effect of w/c on autogeneous shrinkage................................................. 139
xi
6 . 1 Schematic of the inclusion of temperature effects........................................ 145
6 . 2 Degree of hydration data for portland cement paste [Abd.l].................. 146
6.3 / ( o ) and E ( a ) obtained by TPM.................................................................... 148
6.4 Temperature history for semi-adiabatic heat development........................ 150
6.5 Semi-adiabatic heat development due to Ti .................................................. 150
6 . 6 Semi-adiabatic heat development due to To.................................................. 151
6.7 Compressive strength development with at 5 isothermal conditions
[Abdel-J awad].......................................................................................................153
6 . 8 Degree of hydration (o) determined by means of chemically bound
water................................................................. 154
6.9 Youngs modulus development with a ............................................................ 154
6.10 Compressive strength development with a .................................................... 155
6 . 1 1 Split tensile strength development with a ..................................................... 155
6.12 Relations between Youngs modulus and heat of hydration......................156
6.13 Relations between compressive strength and heat of hydration. . . . 157
6.14 Relations betwreen split tensile strength and heat of hydration. . . . 157
6.15 Relations between Youngs modulus and heat of hydration......................158
6.16 Relations between compressive strength and heat of hydration. . . . 158
6.17 Relations between split tensile strength and heat of hydration. . . . 159
6.18 Relation between compressive strength and aging time, data from
[Abd.l]................................................................................................................... 167
6.19 Relation between compressive strength and aging time, data from
[Kje.2]..................................................................................................................... 168
6.20 Relation between fracture energy and aging time, data from [Yu.l]. 169
xii
6.23 Examples of the Arrhenius type equivalent time, data from [Kje.l]. . 171
6.24 Autogeneous deformation with time............................................................. 173
6.25 Autogeneous deformation with heat................................................................ 174
6.26 Creep data and curve fits, w/b=0.45.............................................................. 177
6.27 Creep data and curve fits, w/b=0.35.............................................................. 178
6.28 Calculated relaxation and curve fits, w/b=0.45........................................... 179
6.29 Calculated relaxation and curve fits, w/b=0.35........................................... 180
6.30 Dirichlet series representation of creep function (dashed lines indi
cate the log-power approximation) 182
6.31 Dirichlet series representation relaxation function (dashed lines...in
dicate the hyperbolic approximation)...................................................183
6.32 Variation of parameter with age and degree of hydration........... 186
6.33 Variation of parameter m2 with age and degree of hydration......... 187
6.34 Variations of C^ with age and heat of hydration.........................................188
6.35 Variations of E^ with age and heat of hydration.........................................189
6.36 Variations of E / t * + E / r (l with age................................................................190
6.37 The effect of additive on specific creep 0(t, t ' ) .............................................191
6.38 The effect of additive on normalized relaxation modulus R(t, t ' )/ E{t ' ). 191
7.1 Early age stress development in mix 45-1..................................................... 198
7.2 Early age stress development in mix 45-2....................................................199
7.3 Early age stress development in mix 45-3................................................... 200
xiii
7.4 Early age stress development in mix 45-4.................................................. 201
7.5 Early age stress development in mix 35-1................................................. 202
7.6 Early age stress development in mix 35-1, second test............................... 203
7.8 Early age stress development in mix 35-2, second test............................... 205
7.10 Early age stress development in mix 35-3, second test...............................207
7.12 Early age stress development in mix 35-4, second test.............................. 209
7.13 Effects of mineral additive on early age stress.............................................. 211
7.14 Temperature and stress history for studying effects of aging.................... 213
7.15 Stress developement with t t a........................................................................ 214
7.16 Time required to reach the critical level.........................................................214
7.17 Temperature history at different locations in the slab................................220
7.18 Temperature profiles at different time............................................................ 220
7.19 Stress on the top surface of the slab (Case C)..............................................221
7.20 Stress at the middle thickness of the slab (Case C)................................... 221
7.21 Stress at the bottom of the slab (Case C)..................................................... 222
7.22 Stress profiles accross the slab thickness........................................................ 223
xiv
LIST OF TABLES
Table
3.1 Mix design information..................................................................................... 34
3.2 Test and curing conditions . ..................................................................... 48
3.3 List of tests performed on m a t e r i a l s ........................................................... 48
4.1 TPE parameters for bound water data, w/b=0.45.................................. 62
4.2 TPE parameters for bound water data, w/b=0.35.................................. 63
4.3 TPE parameters for isothermal data, w/b=0.45...................................... 64
4.4 TPE parameters for isothermal heat data, w/b=0.35............................. 65
4.5 Values of maximum ultimate bound water and maximum ultimate
heat (per gr binder)......................................................................................... 71
4.6 Sg...for mix 45-1................................................................................................. 96
4.7 Sg...for mix 45-2................................................................................................. 97
4.8 Sg...for mix 45-3................................................................................................. 97
4.9 Sg for mix 45-4................................................................................................. 98
4.10 Sg for mix 35-1................................................................................................. 98
6.1 Optimized model parameters for Abdel-Jawad data [ A b d . l ] ...................147
6.2 Parameters At-, Bi, and Ct generated from the isothermal heat data. 149
6.3 Optimized model parameters for compressive strength data [Abd.l]. 164
6.4 Optimized model parameters for compressive strength data [Kje.2]. 165
xv
6.5 Optimized model parameters for fracture energy data [Yu.l] . . . . 166
6 . 6 CTD values for different mixes (10_6 / C) .................................................... 174
7.1 Material parameters for the relaxation modulus......................................... 194
xvi
CHAPTER I
INTRODUCTION
1.1 Background
Concrete undergoes unavoidable complex interactions with the surrounding en
vironment. Such interactions are manifested as physical and chemical changes in
concrete. Many characteristic features of concrete come from these physical and
chemical changes, including thermal deformation, drying and autogeneous shrink
age, age-dependent mechanical properties, and creep. These features make concrete
unique among other construction materials.
Thermal, drying, and autogeneous deformations are the primary sources of di
mensional changes in unloaded concrete. Early age cracking is often linked to such
deformations, especially when concrete is restrained. Under this condition, unloaded
concrete experiences stresses that often exceed its cracking limit. The risk of crack
ing is even higher in young concrete because significant autogenous shrinkage occurs
and the strength is still low. Once cracks form at early ages, many other problems
are stimulated. Cracks can propagate under service loads and results in structural
disintegration. The early age cracking can also lead to durability problems that may
shorten the service life of concrete structures. Deformation however, is not the only
factor that determines the potential for early age cracking. Factors like curing con-
1
ditions. creep properties, types of cement, and mix compositions are also important.
Several techniques have been introduced to minimize the risk of concrete cracking
at early ages. The installation of cooling pipes in concrete members has been able to
reduce the temperature change considerably [Bru.l]. Precooling of mix components
with ice during hot weather concreting has also been applied. These techniques,
however, could alter the nature of hydration process of the cement. In some cases, the
use of low heat cements or blended cements can be quite effective in reducing the risk
of cracking. The use of blended cements is also in accordance with the development of
high performance concrete (HPC) and fiber reinforced concrete (FRC). Mature HPC
is often used for high rise building for providing size-efficient compressive members.
FRC have been introduced to meet the requirements for ductile behavior of concrete
elements. The use of low water-cement ratio concrete together with silica fume,
fly ash, and superplasticizers enables us to produce high grade concrete adequate
for more extreme loading conditions. However, before concrete can possess such
superior properties it must survive the early age cracking problem. In addition, the
development of the properties must be able to be monitored.
In order to assess the risk of cracking at early ages several basic issues have to
be resolved. From the explanations above, these issues include finding the source
of aging, identifying and understanding the deformation components, characterizing
the relevant age dependent properties, calculation of early age stress, and effects of
blended cements. The primary source of aging, consistent with the statement on
the chemical changes given in the first paragraph, is hydration. The components
of deformation, as described before, are thermal, drying, and autogeneous deforma
tions. The relevant concrete properties include creep compliance, elastic modulus,
coefficient of thermal dilation, and strength.
3
Apart from the early age cracking, cement hydration alone is an important subject
of research. Hydration together with microstructural evolution provide the funda
mental explanation of aging. Unfortunately, hydration has not been studied exten
sively in blended cements. Furthermore, aging is often represented by time instead of
degree of hydration. This lack of interpretation is forgivable if effects of environmen
tal factors such as temperature or humidity are not considered. However, temper
ature or humidity in field concrete are rarely constant. Meanwhile, microstructure
is often shown to relate to mechanisms such as drying shrinkage, creep, and au
togeneous shrinkage. Our understanding of humidity-driven shrinkage mechanisms
can be improved by looking the problem at the micro level. Studies on concrete
microstructure also allow us to link aging concrete properties and hydration. It is
unfortunate to find out that there are not many microstructural studies conducted
on blended cement systems, although they have been widely used.
In view of concrete properties, creep compliance is probably the most important
property in the early age stress calculation. I t strongly influences the stress relaxation
behavior of early age concrete. Yet, there are not many creep compliance obtained
for blended systems. Further, since early age cracking occurs in tension, creep com
pliance in tension is more relevant. However, most of concrete creep compliance is
still obtained in compression. Many have reported that the tensile creep compliance
is quite different from the compressive creep compliance [Nev.2,Ill.l,,Bro.l]. There
fore, to assure accurate and reliable early age stress predictions, the tensile creep
compliance must be used.
All the issues elaborated above and gaps in knowledge associated with them are
the subjects to be investigated in this research.
4
1.2 Objectives, Scope, and Research Significance
1.2.1 Research Objectives and Scope
Based on the problems defined above, many complex mechanisms are found to
exist when cement or concrete hydrates. It is clear that one of such mechanisms
is hydration. Thus, the investigation of the hydration process together with its
influencing factors is mandatory in understanding the behavior of hydrating con
crete. Deformations that may lead to development of stresses including thermal
dilation/contraction, drying shrinkage, and autogeneous shrinkage, and how they
are affected by hydration and microstructure must be studied. To assess the risk of
cracking at early ages, thermal stresses developing during concrete hydration must
be investigated. Relations between concrete properties, that are important for the
early age stress, and hydration level have to be found. Therefore, hydration kinetics,
microstructure, and mechanical behavior must be investigated together. Based on
this knowledge, this research focuses on achieving the following objectives:
1. To analyze hydration kinetics of blended cement systems by means of exper
iments and analytical formulations, for isothermal and non-isothermal condi
tions.
2. To characterize the hydration products and the microstructure of blended ce
ment systems.
3. To develop a model for the humidity-driven shrinkage in cement paste based on
the microstructural characteristics and micromechanics, and apply it to drying
and autogeneous shrinkage problems.
4. To establish relations between degree of hydration and material properties such
as strength, Youngs modulus, creep compliance, relaxation modulus, and au
togeneous shrinkage.
5. To calculate the early age stress development in hydrating concrete using the
established relations between properties and hydration, and verify it experi
mentally.
6 . To investigate the effects of using blended cements on concrete properties and
in the early age stress development.
The scope of this research is confined to several major features of hydrating
cement based materials, including experiments and analysis on hydration kinetics,
microstructure, deformations, and the early stress development in hydrating blended
cement systems. The experimental part of this study is intended for obtaining im
portant hydration parameters, mechanical properties, and for simulating the stress
development. The analytical part consists of modeling of hydration, shrinkage be
havior, development of computational procedures for the stress analysis which uses
material parameters obtained experimentally.
To cover the scope explained above and to achieve the research objectives, the
following tasks must be completed:
Experimental determination and analysis of hydration. Two methods are em
ployed to investigate hydration of blended cements: isothermal calorimetry
and differential thermal analysis (DTA). Further, the measurement of hydra
tion must be performed at different isothermal conditions in order to study the
temperature effect. The data from isothermal calorimeter can distinguish hy
dration of different blended systems but cannot be used to quantify the degree
6
of pozzolanic reaction. However, isothermal calorimetry provide accurate in
formation on early hydration which cannot be obtained by DTA. On the other
hand, DTA can be used to quantify the degree of pozzolanic reaction.
Analysis of microstructure. In this research, the microstructure of pastes are
analyzed to characterize the hydration products and their pore space in blended
cement pastes, and to describe the mechanisms of shrinkage. The analysis is
based on water and nitrogen sorption isotherms.
Experiments on deformation components. As described before, three compo
nents of deformation are thermal, drying, and autogeneous deformations. Two
out of this three components, thermal and autogeneous deformations are mea
sured experimentally. The measurement of thermal deformation is done to
obtain the coefficient of thermal dilation (CTD) of concrete. Experimental
drying shrinkage is beyond the scope of this research.
Experimental determination of mechanical properties. Concrete properties
measured include the compressive strength, split tensile strength, Youngs mod
ulus, and creep compliance. Each property is measured at different ages. The
compressive strength, split tensile strength, and Youngs modulus are basic
properties that are often used in concrete design and are monitored for quality
control. The creep compliance is the most important property in the context
of early age stress development.
Establishing relations between mechanical properties and hydration. Results of
isothermal heat of hydration and mechanical properties need to be related for
several reasons. First, to bridge kinetics and mechanical behavior and second,
to be used as a general measure of aging.
(
Method to predict hydration. A realistic analysis of hydrating concrete requires
the incorporation of effects of temperature on the mechanical properties and
degree of hydration. If under any temperature history, the degree of hydration
is uniquely related to the mechanical properties, then one will only have to
develop a method to predict hydration. The effects of temperature on the me
chanical properties will be included through the degree of hydration. Further,
the method to predict hydration must be able to handle the non-isothermal
curing condition.
Prediction of early age stress. The procedure for the early age stress calculation
must be developed in order to assess the risk of cracking at early ages. The
procedure, if proven to be accurate, can also be used to investigate the benefits
of using blended cements.
Measurement of early age stress. The experimental determination of early age
stress is necessary to verify the early age stress calculation.
1.2.2 Research Significance
According to the problems stated at the end of the first subsection here, there are
many gaps separating our knowledge on early age cracking behavior in particular,
and on hydrating concrete behavior in general. Hydration kinetics of blended cement
systems have not been thoroughly studied. Techniques to understand and to quantify
blended cement hydration are also lacking. On the microstructural analysis, blended
cements are not well investigated like ordinary cement. As a result, we know very
little about hydration products, surface area, and pore characteristics of blended
cement pastes, although we know that these information allow us to understand
mechanisms like shrinkage or creep. On mechanical properties, relations have to be
8
established between the mechanical properties and hydration so that aging can be
quantified by a measure of hydration and not by time. Therefore, a method has to be
developed to first quantify degree of hydration of concrete subjected to general curing
conditions. On the early age stress calculation, most concrete creep compliance is
obtained in compression, while cracking occurs in tension. Tension creep compliance
is therefore, more relevant.
There are several important improvements in this research as compared to the
previous work. Hydration of blended cements is studied rigorously using two meth
ods: isothermal calorimetry and differential thermal analysis. Important information
about the pozzolanic reactions can be obtained using these methods. Further, based
on Powerss concept, the degree of hydration of blended systems and the degree of
pozzolanic reaction can be determined.
In this research, the characterization of the paste microstructure is done by ana
lyzing water sorption isotherms. Most studies utilized nitrogen sorption isotherms to
do such a task. Water is considered better for several reasons. First, its interactions
with solid paste plays important roles in mechanisms such as creep, drying shrinkage
and freeze thaw durability, and second, it is the fluid that originally occupies the
pore space. An important but rarely studied issue that is also covered in this re
search is the microstructure of blended cement paste. Based on the microstructural
analysis above and micromechanics, a very realistic humidity-driven shrinkage model
is proposed in this work.
A method to calculate the degree of hydration and properties development under
non-isothermal conditions is proposed in this work. Unlike the previous methods, the
Arrhenius concept is not used. Instead, since it is shown that the Arrhenius concept
has some weakness, a three parameter equation is used for the calculation. This
9
method together with the viscoelasticity theory is then used to develop a procedure
to calculate the early age stress development. The new feature in this procedure is
the direct use of degree of hydration as the measure of aging. This is a significant
improvement since the accelerating and deccelerating effects of temperature on vis
coelastic properties are incorporated through the degree of hydration. Furthermore,
the creep complaince used in the early age stress analysis is obtained under tension
and thus, according to the previous section, is more reliable.
1.3 Thesis Organization
This thesis is organized in a manner such that the micro aspects and basic be
havior are exploited first. Then macro or bulk behavior is studied next. The micro
aspects can be used to study hydration products and pore characteristics of blended
systems, and to describe the shrinkage mechanisms. This information together with
basic analysis on hydration are useful in describing macroscopic behavior like aging
properties and autogeneous shrinkage.
The second chapter of this thesis reviews the phenomenological features of hy
drating concrete. In this chapter, analytical and experimental studies, and some
recent development on the subject matter reported in the literature are brought to
attention.
In Chapter 3, experimental work conducted in this research is explained in detail.
This chapter covers methods used to determine degree of hydration, compressive
strength, split tensile strength, Youngs modulus, tensile creep, sorption isotherms,
coefficient of thermal dilation, and autogeneous shrinkage. Information on materials
such as mineral additives, aggregates and mix parameters are all provided in Chapter
3.
10
The analysis of results begin at Chapter 4. A significant portion about hydration
kinetics and the method to predict the degree of hydration of blended cements as
well as pozzolan are covered. The sorption-based analysis of hydration products and
the solid surface area is presented. Several important findings on how the hydration
products develop with degree of hydration are also included.
Chapter 5 covers the modeling effort in this work. It is shown in this chapter that
good prediction of autogeneous shrinkage can be achieved only by understanding
drying shrinkage mechanisms at micro level. Thus, a model is first develop for
the drying shrinkage then extended to the autogeneous shrinkage. The formulation
details the mechanisms involved, the microstructure evolution, and the driving forces
for shrinkage.
Discussions on how blended cements affect mechanical properties and methods to
calculate non-isothermal degree of hydration are presented in Chapter 6 . In addition,
a concept similar to the equivalent time concept is also given and applied. The
results generated in Chapter 6 are then used for predictions of the early age stress
development in Chapter 7. A calculation procedure is developed and the results of
prediction are verified in Chapter 7. In addition, the procedure is also extended to
predicting the stress history in field pavement or foundation slabs caused by nonlinear
temperature profiles.
Finally, a summary of important findings and conclusions are presented in Chap
ter 8 . Some recommendations on some future research directions pertaining to hy
drating concrete behavior are also given.
CHAPTER II
REVIEW ON BEHAVIOR OF HYDRATING
CONCRETE
2.1 Hydration
Numerous aspects in concrete behavior such as properties development, creep,
relaxation, shrinkage, heat transfer, and mass transport, are tied closely to the de
gree of hydration. This makes hydration the primary source of aging behavior in
concrete. Therefore, measurement methods become vital in obtaining information
on how different cements hydrate and on factors that affect hydration.
In order to estimate degree of hydration accurately, factors that affect hydration
must be determined. For evaluating degree of hydration of a concrete mix in the
field, environmental factors such as temperature may become very important.
2.1.1 Measurement Methods
Chemical and physical reactions that occur between cement compounds and water
are often called hydration. During hydration, the mass compositions of cement plus
water change. Phases in cement paste include unhydrated cement grains, hydration
products, water (liquid and vapor), and air. During hydration, the mass compositions
of these phases changes. Experimental methods are developed to quantify such
changes.
11
12
The degree of hydration has been measured in terms of different substances such
as chemically bound water [Tap.l], heat of hydration [Kon.l], and C-S-H compounds
[Kon.l,Ram.l]. Techniques to measure the degree of hydration include Thermogravi-
metric Analysis (TGA) [Ram.l], Differential Thermal Analysis (DTA) [Ram.l], Loss
of Ignition (LOI) [Tap.l], and Calorimeter Analysis (CA) [Par.l]. Not all of these
techniques are applicable to concrete. TGA, DTA, and LOI are only applicable to
cement paste. CA, although does not directly measure the mass or volume changes,
can be used for paste, mortar, and concrete. It has also been shown that the result
of CA on paste specimens agrees very well with test results produced by the other
techniques [Par.l].
2.1.2 Influencing factors
The hydration of cement is influenced by intrinsic factors and environmental
factors. Intrinsic factors include the physical and chemical characteristics of cement
such as [Bru.2]:
1 . Chemical compositions of cement or blended cement.
2. Fineness and particle size distribution.
3. The presence of admixture.
4. Water/cement ratio
The effects of the intrinsic factors are illustrated in Figure 2.1. Meanwhile, the
environmental factors that predominantly affect the hydration are:
1. Temperature
2. Humidity
1 3
The effect of temperature is illustrated in Figure 2.2.
Hydration products obtained from concrete and mortar appear to be quite differ
ent compared to those obtained from paste [Scr.l]. This indicates that the hydration
process in concrete and mortar differs from that in paste. Many investigations reveal
that the presence of aggregate somewhat retards the hydration. This is also sup
ported by the microstructural analysis of the interface between the aggregate and
cement paste [Scr.2]. In addition, the measured heat of hydration of concrete ap
pears to be lower than mortar, and the heat of hydration of mortar tends to be lower
than that of paste [Byf.l]. In this measurement, the heat measured in concrete and
mortar have been normalized by the cement content. Therefore, in addition to those
factors mentioned above apparent hydration of concrete and mortar are also affected
by:
1. Aggregate type.
2. Aggregate content or volume fraction.
3. Aggregate gradation.
2.1.3 Hydration Kinetics
Quantitatively, temperature plays an important role in accelerating hydration of
cement. This has been shown by different measurements methods [Bru.2,Kon.l]. A
commonly used approach is to apply the Arrhenius equation that makes the rate of
hydration (d) depends on the degree of hydration (a) and the temperature history
(T). The Arrhenius equation takes the form:
d ~ f ( a ) exp ( ~ ^ ) (2 -1 )
1 4
where E is the apparent activation energy, R is the ideal gas constant, and T is
temperature.
The incorporation of temperature in degree of hydration is often done using the
maturity concept [Abd.,Wan.l,Mae.l] similar to handling the temperature effect in
compressive strength [Car.2]. Maturity (M) is intended to include the effect of
temperature in time or age and can be calculated using the above expression relative
to a reference temperature Ta using [Car.2]:
The application of maturity calculated from the above equation turns out to be
limited. In [Abd.l], the activation energy was found to vary with the degree of
hydration and thus, M could not be calculated without knowing a.
The effect of humidity on the degree of hydration has also been studied by several
investigators [Par.2,Pow.l]. It is generally found that the humidity tends to reduce
the rate of hydration. An equation that incorporates the temperature and humidity
history in the form of rate of hydration has also been proposed [Ros.l], and is written
as:
d ~ / ( a ) exp ( ^ ) [l + (3-5 - 3.5/i)4] 1 (2.3)
where h is the relative humidity. The application of the above equation is rarely
done since there are not many studies conducted on cement hydration at reduced
relative humidity. A recent study by Jensen et.al. [Jen.2] however, revealed that the
effect of humidity can be clearly seen on the induction time of hydration.
A more theoretical approach to understanding the hydration kinetics is to use
the particle kinetics. Models have been proposed to explain hydration process using
particle kinetics [Bru.4,Nav.l,Kon.l,Koe.l]. The particle kinetics approach usually
1 5
consider the hydration process to be controlled by multiple mechanisms, including
nucleation and growth, and mass transport. The effect of temperature on each of
the mechanisms can also be incorporated [Bru.4]. The particle kinetics approach can
also incorporate effects of mix compositions like water-cement ratio and particle size
distribution. In a model by Navi and Pignat [Nav.l], the connectivity of capillary
space can also be predicted.
2.2 Microstructure of Cement Paste
2.2.1 Pore Characterizations
Pore characteristics of cement or concrete are usually obtained using techniques
like mercury intrusion porosimetry (MIP), gas pycnometry and gas sorption. Dif
ferent techniques often provide different quantities. For instance, MIP is used to
obtained pore size distribution and pore volume, while the gas sorption is often used
to determine the surface area and the distribution of pore smaller than those mea
surable by MIP. Pycnometry is not as frequently used as the two methods above,
but it has been used to measure the solid density and pore volume.
Numerous studies on microstructure of paste have been based on MIP and sorp
tion measurements. Studies by [Win.2,Dia.l,Aus.l] can be considered the earliest
work done on cement paste. Early sorption isotherm studies on cement paste dated
back to the work by [Pow.2] who used water sorption data as a basis for their mi
crostructure model.
Techniques such as MIP is often used to quantify the evolution of capillary pores.
These are the pores that are initially occupied by water. As hydration proceeds these
pores are replaced by hydration products or gel. The capillary porosity obtained from
MIP has been shown to uniquely relate to mechanical properties like compressive
1 6
strength [Ver.l] and Youngs modulus [Hel.l]. Unique relations between the capillary
porosity and the degree of hydration has also been illustrated [Abd.l,Win.2].
The gas sorption isotherm is normally limited for studying smaller pore systems
such as the gel microstructure. For cement paste, two types of gas are commonly
used: nitrogen (N2 ) and water (H20). Unfortunately, the results obtained using
the two gases differ from each other. In the past, the nitrogen was more preferred
because its data can be acquired relatively fast. However, the nitrogen sorption data
often estimates a surface area lower than that obtained by the water sorption. In
addition, the results of obtained from both gases are known to be sensitive to the
sample preparation methods. Litvan [Lit.l] showed that methods such as ice-drying,
vacuum-drying, oven-drying, and methanol exchange can significantly alter the result
of nitrogen sorption. In addition, it is known that oven-drying significantly reduces
the surface accessible for adsorption [Han.l].
The surface area obtained using water sorption usually almost doubles the value
obtained from the nitrogen sorption [Mik.l]. It is believed that nitrogen cannot
access the whole surface of solid in cement paste. The value obtained using water
sorption has been shown to be close to what is obtained using the other techniques
such as small-angle X-ray scattering (SAXS) [Vol.l].
2.2.2 Microstructure Related Behavior
Concrete behavior that strongly relates to microstructure include shrinkage, creep,
and mass transport. Bentur et.al. [Ben.l] reported the dependence of shrinkage,
basic creep, drying creep to microstructure. Mechanisms of drying shrinkage and
swelling in cement paste are often described based on microstructure properties
[Han.2,Baz.l]. Many investigations on shrinkage mechanisms reported the signifi-
1 7
cance of surface area in determining the magnitude of shrinkage strain [Han.2,AIm.l,Fer.l,Wit.l].
Microstructure has been shown to control the transport of fluid in concrete or
paste. It has been illustrated in [Xi.l,Xi.2] that modeling of drying process requires
transport properties that can be generated from water sorption. In [Odl.l], the major
transport properties like permeability, has been reported to depend strongly on the
capillary pore volume and size distribution.
2.3 Mechanical and Thermal Properties
2.3.1 Strengths and Elastic Moduli
Strength parameters normally used in practice are the compressive strength ( / ')
and split tensile strength ( f sp). The development of / ' with time is illustrated in
Figure 2.3. During the very early period of curing time, the strength is almost zero .
This period is sometimes called the setting period [Meh.l]. After the setting, period
the strength grows very fast, especially within the first three days.
A behavior similar to what is observed in strength growth is also found in the
Youngs modulus (E ) of concrete (Figure 2.4). However, the Poissons ratio changes
insignificantly with time. The Poissons ratio (i/) tends to slightly decrease as the age
increases. Stress-strain relationships of concrete appear to be different at different
ages, as reported in [Byf.l,Olu.l]. This finding is also directly correlated with the
changes of strengths and moduli with time.
It is believed that there are unique relationships between the strength and degree
of hydration. Many investigations [Gru. 1 ,Nev. 1 ,Pow.2] show that factors affecting
the degree of hydration are very similar to the ones affecting the development of
strengths. For cement paste, the cube strength is found to be uniquely related with
the degree of hydration [Pow.2]. The relationship between the cube strength and the
1 8
degree of hydration has been found to be almost linear except at the beginning. For
concrete, the strength is often plotted versus equivalent ages (te) or maturity time
(M). The main purpose of this was to include the effect of curing temperatures on
the strength development [Car.2 ,Emb.l]. However, this concept does not physically
tied with the hydration process. A more realistic measure of aging of concrete is the
heat of hydration (Q) or degree of hydration (a). When the ultimate value of heat
of hydration (Qoo) is known, the degree of hydration can be found from a = Q/Qoa-
Relationships between E-a, f sp-a, and /'-c* are shown schematically in Figure 2.5.
Concrete strength and fracture properties are also found to be rate dependent. In
general, the strength tends to decrease with decreasing rate of deformation [Rol.l].
The same trend applies to the three point bending test results used to determine
the flexural strength and the fracture energy of concrete (G f ) [Baz.3,Baz.4,Wit.2].
Several theories regarding the rate dependent fracture process have been proposed.
Some of these theories consider t h a t the rate of crack opening displacement is directly
influenced by the stress level [Baz.4|. The strengths of concrete under sustained was
also found to be lower than the so called the static strength.
2.3.2 Creep and Relaxation
The stress analysis of early age concrete members must take into account creep
and relaxation behavior. Creep behavior of early age concrete is quite complex since
aging or hydration occurs fast at early ages. Figure 2.6 illustrates how significant
concrete creep compliance changes with age. The relaxation modulus, which is de
fined as the inverse integral of creep compliance, has also been shown to be very
significant during early ages. Measurements to determine the relaxation modulus
at early ages are rarely done because it is difficult to separate the deformation in
1 9
duced by the applied load, from other deformations such as thermal and autogeneous
shrinkage. Normally, the relaxation modulus is obtained from the creep compliance
using the following relation [Baz.l]:
1 = J{t . t ' )dR{t ' ) (2.4)
Jo
where J and R are, respectively, the creep compliance and relaxation modulus. Fig
ure 2.7 illustrates the relaxation modulus reported in [Wes.l]. Due to creep or relax
ation of concrete, early age stress development is a strongly time dependent problem.
Creep of concrete is also influenced by the humidity condition. Simultaneous
creep and drying can result in deformations higher than the creep deformation under
sealed condition (basic creep). This behavior is often called the Picket effect [Pic.l].
Drying creep is essentially caused by the shrinkage-induced stress which develops
because of non-uniform distribution of shrinkage strain in the concrete specimen
[Baz.l]. In some cases, the shrinkage-induced stress can be quite significant to cause
surface cracking. This is because the concrete surface, which is drier than the interior,
experiences the highest tension stress.
In early age concrete, the humidity effect on creep is often assumed to be minor
because concrete members are normally covered by formworks. In some members
such as concrete pavements and foundation slabs however, the concrete surface is
exposed to drying at the very beginning hence, the effect humidity should not be
ignored. However, because drying is a very slow process [Baz.l], it is sometimes
assumed that the strain gradient due to temperature rise predominantly determines
the stress development in early age concrete.
Creep of concrete appears to be different for different state of stresses [Nev.2].
For the uniaxial compression-tension stress state, it is observed that the basic creep
20
compliance is higher for concrete under tension [111.1]. This phenomenon is some
times associated with the hydration process. It is believed that compressive stresses
enhance the rate of hydration through compaction or densification of hydrating ce
ment [Baz.5] while tension tends to decelerate the hydration. Early age concrete can
experience both tension and compression for long period of time prior to cracking.
Therefore, for better prediction early age cracking, the compression and tension creep
compliances needs to be estimated separately.
2.3.3 Measurement o f Stresses due to Restrained Deformation.
The potential for cracking due to restrained deformation has been recognized long
enough. Drying shrinkage was more of interest in the past. Several test methods
were then proposed to assess the potential for cracking due to drying shrinkage. One
of such methods is called the ring test, which consists of a thin specimen cast around
a rigid steel ring. This test has been used quite often due its simplicity. Among
other problems investigated using the ring test is the benefit of using different types
of fiber in controlling cracking of concrete and mortar [Grz.l,Swa.l,Sar.l], The only
draw back with this test is that the stress cannot be directly measured and must
be back calculated. Precaution must also be addressed to the tests involving drying
shrinkage. Since drying is a very slow process, the humidity profile in the tested
specimen may not be uniform until equilibrium is reached. As an illustration, thin
paste specimens of 1 mm thickness took about two months [Alm.l] to reach mass
equilibrium when dried at 1 1 % relative humidity.
Measurements of early age stresses due to restrained deformation dated back to as
early as 1960s. During those years, researchers in Germany developed a test method
to measure the stress in a restrained uniaxial member. The early generation of such
21
a test utilized a very stiff steel frame and a load cell without any moving crosshead.
The later test method used moveable crosshead that can be controlled to provide
100% restraint, and a temperature control system. More detail information about
this test can be found in [Spr.l].
Since the measurement of early age stress is not a standard test, many other re
searchers developed their own test methods. Emborg [Emb.l] developed a procedure
to test early age concrete vertically in a container filled with water. The procedure
allowed the specimen temperature to be controlled accurately. In such a test proce
dure, it was not clear whether autogeneous shrinkage was taken into account or not,
because autogeneous deformation is normally measured from a sealed specimen and
not from a saturated specimen. A typical stress history together with the temper
ature history of early age concrete member reported in [Emb.l] is shown in Figure
2. 8.
The idea of using adjustable cross head was utilized by several other researchers
including Bloom and Bentur [Blo.l], and Kovler [Kov.l]. Bloom and Bentur de
veloped a system to test a uniaxally restrained specimen in a horizontal position.
The specimen was gripped at one end fixed and the other end moveable. The grip
movement is controlled by feedback from a displacement gage. Kovler used a dif
ferent arrangement. Two identical specimens, one was restrained and the other was
free, were measured simultaneously. In his system, the stress was measured using a
closed-loop computerized controller.
2.3.4 Early Age Stress Predictions
The prediction of stress in concrete at early ages is very important for assessing
the risk of cracking at early ages. In general, the stress prediction must consider
22
viscoelastic behavior and several deformation components. Most of the prediction
methods reported in the literature assumes the linear viscoelastic behavior. When
the linear viscoelastic behavior is assumed for concrete, the following conditions must
be satisfied [Baz.l]:
The stress is less than about 0.40 of the strength.
The strain rate does not decrease markedly in magnitude (but the stress may),
e.g. violations are often observed at low-stress level and during creep recovery.
The specimen undergoes no significant drying, and hence, the drying creep is
negligible.
There is no major increase of stress magnitude long after initial loading. Oth
erwise, low stress non-linearity and/or plastic flow may occur.
Emborg [Emb.l] developed a procedure to calculate the early age stress incorpo
rating relaxation behavior. He formulated the procedure in two forms. First, in the
integral form and second, in the differential/rate form. The two formulations were
equivalent but required different expressions for the relaxation modulus. Emborg
also extended his formulation to include the nonlinear behavior due to tensile micro
cracking but did not consider the deformation component coming from autogeneous
shrinkage. Furthermore, he also ignored the effect of temperature on the aging vis
coelastic properties. Another formulation that also considers some non-liearity due
to cracking was also proposed by Huckfeldt et.al. [Huc.l]
In more recent work by Ulm and Coussy [Ulm.l], Sercombe et.al [Ser.l], and
Hellmich et.al. [Hel.3], temperature was not only numerically predicted but was
also included in the creep function. In their formulations, rigorous and fundamental
2 3
thermodynamic arguments were used. However, no verification from a simple test
like the uniaxial early age testing was given.
In view of mathematical formulations, some other predictions by different re
searchers are essentially the same [Ume.l,Koe.l,Onk.l]. The integral formulation is
essentially used. Because the integral formulation can accommodate more general
forms of relaxation modulus. The rate formulation, as shown in [Emb.l,Baz.l,Baz.2],
requires the relaxation modulus to be expressed in terms of Dirichlet series. Fitting
general relaxation modulus using the Dirichlet series can be troublesome and may
result in violating the thermodynamic restriction for the relaxation modulus [Car.4].
2.3.5 Coefficient o f Thermal Dilation (CTD)
In general, thermal deformation of solid depends on the microstructure. For
composite, in addition to the coefficient of thermal expansion of the individual phase,
thermal deformation also depends on the elastic moduli of each phase [Min.l]. The
presence of a rigid phase or inclusions can restrain the deformation of the matrix.
For concrete, the coefficient of thermal dilation (CTD) depends on the age, the
mix composition, moisture content, and the type of aggregate [Meh.l]. The effect
of moisture content (in terms of humidity) on CTD has been found to be significant
[Min.l]. CTD increases considerably for humidity ranging from 40 to 80%. This
dependency is sometimes referred to as hydrothermal expansion and is affected by
water to cement ratio (w/c) and age.
The coefficient of thermal expansion for common rocks varies with the mineral
composition which in turn affects the coefficient of thermal dilation of concrete con
taining different aggregates. For a given aggregate type, CTD also varies with the
volume fraction of aggregate. An experimental data showing the dependency of CTD
2 4
of mortar on sand content is given in Figure 2.9.
According to ACI Committee 517 [Aci.l], CTD of fresh, concrete is several times
higher than that of hardened concrete. However, no significant change in CTD has
been observed for concrete ages greater than 1 day. For normal weight concrete, CTD
varies from 7xl0- 6 to 14xlO-6/C depending on the aggregate type. In practice, it
is customary to use 10xl0_6/C as CTD of concrete [Aci.2].
2.4 Deformations in Hydrating Concrete
2.4.1 Drying Shrinkage
Drying shrinkage is defined as the reduction in dimensions of solid due to drop in
the relative humidity. Drying, which involves transports of multiphase fluids in the
pore system, is a very slow process. Thin paste specimens of 1 mm thickness takes
about two months [Alm.l] to reach mass equilibrium when dried at 11% relative
humidity. For hydrating concrete members which are large in dimensions, drying
towards the equilibrium may take months or years. On the other hand, hydration
occurs relatively fast in the first few weeks. It is also possible that only some portion
of the concrete members can fully hydrate while the other part stops hydrating
because of drying.
Major shrinkage mechanisms in cement paste have been reported by different in
vestigators [Han.2,Fer.l,Alm.l,Wit.l,You.2]. These include disjoining pressure, cap
illary stress, and surface tension. Although major shrinkage mechanisms have been
discovered for quite sometime, practically very little effort has been made to ex
plicitly consider them at the micro or pore level. Only few works were done on
developing the work by Bangham and Maggs [Ban.l] and Yates [Yat.l], who initially
proposed the expressions for surface energy-driven strains (Gibbs-Bangham strains).
2 5
A notable work on this area was also done by Scherer [Ref.], who proposed the ex
pression of the Gibbs-Bangham strain for a system of intersecting cylindrical pores
and applied it to porous glass. The other interesting work is due to Bentz [Ref.] who
developed a computer-based model to analyze shrinkage strains induced by surface
energy (Gibbs-Bangham stress) and capillary stress, in porous glass. The same type
of modeling however, has not been done on cement paste.
It very likely that the magnitude of drying shrinkage changes as a result hydra
tion or aging. As mentioned earlier, several studies have reported the dependency
of drying shrinkage to the surface area, which in turn depends on degree of hydra
tion [Han.2, Alm.l,Fer.l,Wit.l]. Different mechanisms have also been proposed to
describe drying shrinkage. Ferraris and Wittmann [Fer.l] tried to verify t h a t dry
ing shrinkage came from two sources, Gibbs-Bangham stress and disjoining pressure.
Hansen [Han.2] and Almudaiheem [Alm.l] applied the Gibbs-Bangham stress concept
to explain their experimental shrinkage data. Bazant [Baz.7] formulated some rela
tions to determine the disjoining pressure using some thermodynamics arguments.
The shrinkage mechanisms and its interactions with creep have also been reviewed
in [You.2].
2.4.2 Chemical Shrinkage
Chemical shrinkage, also named Le Chatelier contraction, is defined as the re
duction in volume of hydration product [Sor.l]. The deformation on the bulk solid
(specimen) due to chemical shrinkage is negligible as long as the pores remain sat
urated. The chemical shrinkage mainly increases the capillary porosity. In fact, the
measurement of chemical shrinkage is conducted in this manner, that is by measur
ing the amount of water necessary to replace the volume reduction. Since it alters
2 6
the capillary pores, chemical shrinkage is probably related to autogeneous shrinkage.
Such a relation has been formulated in [Koel] in attempt to model the autogeneous
shrinkage mechanism.
2.4.3 Autogeneous Deformations
Autogeneous shrinkage has always been defined as the reduction in bulk solid
(specimen) volume due to self-desiccation. The later is closely tied to chemical
shrinkage. According to [Koe.l], when the additional space created by chemical
shrinkage is not filled with water the pressure in the empty pores created pressure
difference that induces autogeneous deformation.
As reported in [Koe.l], the autogeneous shrinkage is always followed by reductions
in pore relative humidity. Therefore, it is believed that the driving forces for auto
geneous shrinkage is similar to drying shrinkage. Only few attempts were made to
quantitatively analyze the autogeneous shrinkage [Koe.l,Hua.l]. These studies also
suggest that modeling of autogeneous shrinkage must be based on drying shrinkage
mechanisms coupled with a model of evolving microstructure. The development of
autogenous shrinkage with time is illustrated in Figure 2.10
2 7
1-0
0.8
0.6
0.4
0.2
0
Figure 2.1: Influences of intrinsic factors on degree of hydration [Bru.2].
degree of hydration
effect of increasing:
fineness
- temperature
- w/e ratio
time
adiabatic temperature rise I K]
50
30
= 20 C
20
10
0
concrete age [h]
Figure 2.2: Effect of temperature on degree of hydration [Bru.2].
2 8
frr . N/inn^
7d 3d 28 d id
Age
Figure 2.3: Development of the compressive strength with time [Byf.l].
modulus of elasticity (kN/mm2)
30
20
0 12 36 24 48 60
time (h)
Figure 2.4: Development of Youngs modulus with time [Thi.l].
2 9
mechanical properties Ecc, 5;-<
tc *
0 . 6
0 . 2
*
:t( * 1)
0 . 8 0.4 0 0 . 6 1 . 0
degree of hydration a [-]
Figure 2.5: Schematic development of compressive strength, tensile strength, and
Youngs modulus with degree of hydration (Q/Qoo) [Onk.l].
<
o
SAIN MODEL
cu 1^*
C4
U
23:
0.001 o.ot t o o 1 3 0 0 1 0 0 0 0
Figure 2.6: Basic creep compliance of concrete ad early ages [Emb.l].
3 0
INPUT
MAXWELL CHAIN MODEL 4 0 -
23.
Mix So 4
.2.3
X
1 0 -
0 . 0 0 1 0 . 0 1 0 . 1 10C 1 0 0 0 10 000
Figure 2.7: Relaxation modulus of concrete at early ages [Wes.l].
riMC
Figure 2.8: The stress history of a uniaxial early age concrete specimen [Emb.l].
3 1
Silicaous and/quartz grvrci
ton*
Silii
20 40 SO
Sand in monar (vol. %i
80 100 0 100 20
Coarw aggregate in concrete (vol. %)
40 60 80
Figure 2.9: The dependency of CTD of mortar and concrete specimens on aggregate
contents [Min.l]
Total cnemrcai snnnxage o
>
2
75
ro
c
Formation of empty
pores ana unaerpressure i
u
Fcrmaocn cf a self-
supporting sKeietcn
ci
external cnemicai snnnnage
40
Time (hours)
Figure 2.10: Development of autogeneous shrinkage in cement paste of w/c=0.4
[Sel.l].
CHAPTER III
EXPERIMENTAL METHODS
3.1 Introduction
Experiments conducted for this study are intended to provide information on
hydration kinetics, microstructure, properties, deformations, and early age stress
relaxation of blended cement systems. The tests include: differential thermal anal
ysis and thermogravimetric analysis (DTA/TGA), isothermal and semi-adiabatic
calorimetry, water and nitrogen sorptions, determinations of properties: compres
sive strength, split tensile strength, Youngs modulus, coefficient of thermal dilation,
creep compliance, and measurements of autogeneous shrinkage and early age stress
relaxation.
DTA/TGA, isothermal and semi-adiabatic calorimetry are intended to study hy
dration behavior, to determine the degree of hydration, and to describe the effect
of temperature on hydration in blended cements. Water and nitrogen sorptions are
used to obtain physical characteristics of paste microstructure and their relations
with degree of hydration and humidity-driven shrinkage. Properties such as com
pressive strength, split tensile strength, Youngs modulus, coefficient of thermal dila
tion, creep compliance, are measured at different ages in order to quantify the effect
hydration on properties development. Autogeneous shrinkage and CTD are mea-
32
3 3
sured in order to distinguish different deformation components. The measurement of
early age stress relaxation provide the experimental verification for the calculation
procedure.
Two summary tables on the tests conducted together with the test and curing
conditions are given at the end of this chapter.
3.2 Materials and Mix Design
Materials included in the investigation are: cement paste and plain concrete. The
mix designs for concrete are chosen from those commonly used in practice. Only
type I ordinary portland cement (OPC type I) is used. Two water to binder ratios
(w/b) are considered: 0.35 and 0.45. These two values are chosen since they are
quite common in practice; concrete with w/b=0.35 is often used for bridge deck and
w/b=0.45 is widely used for concrete pavement slabs. Three mineral additives are
considered for investigating the behavior of blended cements: fly ash class F (FA),
ground granulated blast furnace slag (GGB), and silica fume (SF). Their fraction
relative to the weight of binder are kept fixed. The blended cement mixes contain
25% FA, 25% GGBF, and 10% SF. These proportions can be considered moderate
and common in practice for users of FA and GGBF. The use of 10% SF is considerably
high and is quite typical for high strength concrete. Natural glacial gravel (gradation
6 AA according to Michigan-DOT classification) and class II sand (Michigan-DOT
classification) are used as coarse and fine aggregates. For concrete, the binder content
is kept constant at 350 kg/m3. Each mix is labeled according to its w/b and is listed
in Table 3.1.
3 4
Table 3.1: Mix design infonnation.
Mix no. % OPC I % Additives w/b
45-1 1 0 0 0 0.45
45-2 75 25 (FA) 0.45
45-3 75 25 (GGBF) 0.45
45-4 90 10 (SF) 0.45
35-1 1 0 0 0 0.35
35-2 75 25 (FA) 0.35
35-3 75 25 (GGBF) 0.35
35-4 90 10 (SF) 0.35
3.3 Hydration
3.3.1 Thermal Analysis (DTA/TGA)
DTA locates the temperature peaks due to thermal decompositions of different
phases in paste, while TGA simultaneously measures the weight loss corresponding
due to thermal decompositions. To study the effect of temperature on hydration (ki
netics) , the measurements are conducted for paste specimens cured at three isother
mal conditions: 9, 23, and 34C. For all mixes, the ages of testing are: 0.25, 0.5, 1,
2, 3, 7, 14, 28, and 180-210 days.
The cement paste is cast in the form of slab specimens of approximately 100 mm
thin. After two sequences of mixing, the slab specimens are cured at the designated
temperature and are turned upside down every 30 minutes for a period of 6 hours to
prevent bleeding. Prior to testing, these slab specimens are cured in lime saturated
water until the designated age of testing. Small pieces are first obtained by crushing
the slab. Then, the pieces are ground using mortar and pestle until particles mea
3 5
suring about 0.5 mm in diameter are obtained. To stop hydration, exchange water,
and prevent carbonation, the samples are soaked in methanol and stored there for
one week. The methanol is replaced at least three times during the storage period.
The final stage of sample preparation prior to testing is vacuum drying for about one
week. The thermal analyzer used for this study was manufactured by Netzch^-^.
All specimens are decomposed at a temperature range between 21 to 1100C at the
rate of 10 C per minute.
Two important quantities are obtained from DTA/TGA. First, the chemically
bound water, defined as the mass loss due to decomposition between the boiling
temperature and 1100C (uib), which can also be understood as the amount of water
needed for cement to react The chemically bound water does not include free evap-
orable water. In our test system, free evaporable water starts to boil at 100C and
turns completely to gas at around 140C. So Wf, is calculated within a temperature
range of 140-1100C. Our initial test on unhydrated cement however, indicate some
mass loss within a temperature range of 600-780C. This mass loss is labeled COioss
and is illustrated in Figure 3.1. Since the paste sample consists of unhydrated cement
as well, a correction on the calculation of Wb has to be made. Such a correction is
done on cement paste by subtracting the mass loss between within a temperature
range of 600-780C from Wb- Such a mass loss is also labeled CO[oss in the second
figure in Figure 3.1.
The second important quantity is the mass loss corresponding to the decomposi
tion of Ca(OH) 2 (labeled as CHi oss in Figure 3.1) which occurs around 420 to 460C.
Variations of wb and CH[oss with time are reported after being normalized by the
ignited weight of the sample.
3 6
0.5 100
DTA
0.4 99.5
0.3
a>
CO. 0.1
loss
97.5
TGA
Unhydrated cement type I
- 0.1 J97
1200 200 400 600
T(C)
800 1000
0.5 1 0 0
DTA
o>
90
CH,
lost
CO,
loss
TGA
Mix 45-1, 7-day, cured at 3^C
-0.5 80
1200 200 400 600
T(C)
800 1000
JZ
CT>
5
co
o
S'
<
o
K
Figure 3.1: DTA/TGA data and the measured hydration quantities.
8
3 7
3.3.2 Isothermal and Semi-adiabatic Calorimetry
Isothermal heat of hydration is measured only on paste specimens. The lowest
allowable temperature for the measurement is 15C. Due to this limitation and in
order to compare between hydration data obtained using DTA/TGA and isothermal
heat, the measurements are done at 15, 23, and 34C.
Cement paste is cast at room temperature. The mixing water is cooled or heated
prior to mixing, while the cement and mineral additives are kept at room temper
ature. This is done in order to produce mixes whose temperatures are close to the
measurement temperatures (15, 23, and 34C). Since the specimen mass is relatively
small, mixing is performed manually by stirring the cement, additive, and water with
a wooden stick. Once the mix is ready, the neat paste is cast directly in the sample
container. A total of 8 samples can be measured simultaneously. The apparatus was
manufactured by Thermometric A M ^ , which allows a maximum weight of about
10 gr per container. The maximum duration that allows for an accurate measure
ment is about three weeks. The data measured is the rate of heat and, after being
normalized by the solid weight and calibrated, is expressed in J/ h / gr . Figure 3.2
shows an example of the rate of heat measured by the isothermal calorimeter. The
value is then reported as the cumulative heat which is the integral of the rate with
time.
The semi-adiabatic heat is measured using several drum calorimeters. It is ob
tained only for concrete. The specimen shape is cylindrical measures 6 inch, in
diameter and 12 inch. high. The apparatus used was manufactured by Digital Site
System Inc. [Dig.l]. Unlike isothermal calorimeter, semi-adiabatic calorimeter mea
sures heat under non-isothermal condition. In this study, the semi-adiabatic heat
data are only used for procedure verifications and not to study hydration behavior.
3 8
3.5
Mix 45-1, cured at 23 C
1 mV/gr = 1.258 mW/gr
3
2.5
p>
2
tr
1
0.5
0
20 0 40 60 1 0 0 80
Time (h)
Figure 3.2: Rate of heat of hydration measured by the isothermal calorimeter.
3.4 Mechanical Properties
3.4.1 Coefficient o f Thermal Dilation (CTD)
The coefficient of thermal dilation of concrete is determined using a specially
designed system consisting of a frame made of invar steel, a circulating water bath,
and a heating-cooling controller. The specimens can be rectangular or circular in
cross section. The maximum height allowed for the specimen is 12 inch, and the
maximum diameter or cross sectional length allowed is 6 inch..
Two specimens can be tested simultaneously in the water bath. The value of
CTD is obtained for wet concrete since the specimens are wet cured prior to mea
surement. To investigate the effect of aging, the specimens are tested at ages 1, 3,
and 7 days. For determining CTD, the water temperature is raised between 20 to
35C. Displacements are then measured using high precision LVDTs. Each speci
3 9
men uses one LVDT. The data acquisition system allows a low noise measurements
of temperature and displacements. Before testing, the a stainless steel bar is used
to calibrate the frame, and hence, the measurement can be corrected from errors
induced by the expansion or contraction of the frame.
Specimens to be tested must have, at least, two thermocouples within each of
them. One thermocouple must be placed as close as possible to the center of each
specimen. At least two thermocouples must be used to measure the water tempera
ture.
To examine the system performance, two concrete specimens from the present
research were tested. Each test was performed on a concrete specimen and a reference
steel bar simultaneously. Two identical concrete specimens were tested this way. The
purpose was to check the accuracy of the LVDT and to find out if there was any
additional deformations caused by the frame. The temperature varied from about 22
to 34C. The plots of the strain with temperature are shown in Figure 3.3 for the two
identical specimens. As can be seen the strains vary very linearly with temperature
and each plot seems to be similar to each other. The average value of CTD were
found to be 12.6xlO-6/C for concrete and 17.4xlO~6/C for stainless steel. The
value for concrete is within the range mentioned in Chapter 2 . Furthermore, CTE
for steel is very close to what is found in the literature proving the high accuracy of
the test system.
3.4.2 Tensile Creep Compliance
As mentioned earlier, only basic creep is investigated in this study. This means
the specimens have to be sealed so that moisture loss can be minimized. Therefore,
the specimens are covered with thin polyethylene sheets during testing. The specimen
4 0
1.5
co
0.5
20 34 22 26 28 24 30 32
T(C)
Figure 3.3: Variations of concrete stain with temperature.
4 1
is rectangular in cross section with dimensions of 4x4x16 in2. The tensile creep test
is only performed on concrete.
The tensile creep frame is a lever type frame with 1:10 ratio. Each end of the
specimen is attached to the loading frame by threaded bars bolted to the steel plates
that grip the specimen. Four threaded rods which are connected to a steel plate are
embedded at each end of the specimen during casting. To assure proper alignment,
a specially design mold is used for casting. Dead weights are used for loading the
specimens. The frame is equipped with a pneumatic loading unit that enabled one
to control the rate of loading. In addition, a self-guiding apparatus was mounted
to the frame to minimize the eccentricity. These features are shown schematically
in Figure 3.4. To assure the linear viscoelastic behavior, the maximum load applied
was below 0.3 f sp.
Loading is applied mainly at four ages: 1 , 3, 7, and 14 days. The load durations
are the time lags between each age of load application. Two specimens are measured
simultaneously at each age of testing. At the same time, two free or unloaded
specimens are also measured simultaneously so the strains due to temperature change
in loaded concrete can be corrected. Just like the loaded specimens, the unloaded
specimens are also covered with thin polyethylene sheets to minimize drying. Two
linear variable displacement transducers (LVDTs) are used for each loaded specimen
(4 inch, gage length) and one is used for each of the unloaded specimen (12 inch,
gage length). The LVDTs used are high precision ones with the accuracy (for 4 in.
gage length) of 1 micron strain.
4 2
1:10 lever ratio
guiding apparatus
LVDT
loading plates
j l pneumatic pump
Figure 3.4: Schematic of tensile creep testing.
4 3
3.4.3 Strength and Elastic Moduli
The compressive strength, split tensile strength, and Youngs modulus were de
termined using a servo-hydraulic compression testing machine at seven different ages:
0.5-0.8 , 1 , 2, 3, 7, 28, and 90-120 days, for two specimens. All properties are de
termined for concrete only. The specimens tested are all cylindrical, measuring 100
mm (4 in.) in diameter and 200 mm ( 8 in.) in length. The testing machine and
extensometers were manufactured by M T S ^ ^ corporation. The elastic displace
ments in the determination of Youngs modulus are measured using two 50 mm
gage-extensometers attached to each specimen. All specimens are cured in lime-
saturated water at 22.5 to 23C starting from 1 2 hours after casting and remained
in the curing tank until they were ready for testing.
3.5 Autogeneous Shrinkage
A frame and a water bath similar to the ones used for measuring the thermal
coefficient of expansion, is used to measure the autogeneous shrinkage of concrete.
No measurement is made for paste. The autogeneous shrinkage mold is illustrated
in Figure 3.5. It is equipped with closed channels on both vertical sides. The
channels are used to flow water that maintains the test condition isothermal. Water
temperature is maintained constant by a water bath equipped with a heating-cooling
controller.
Autogeneous shrinkage is measured simultaneously on minimum two specimens
and the data reported is the average value obtained from both specimens. The
measurement begins immediately after casting. Each specimen measures 1000 mm
in length, 100 mm in width, and 60 mm in thickness. The bottom of the shrinkage
mold is covered by a very compliant rubber pad in order to minimize friction. During
4 4
measurement, each specimen is covered by thin polyethylene sheets to minimize
drying. This effort has been proven to be effective as one measurement showed that
only about 0.5% out of the total water was lost. In most cases, the test is terminated
after one week.
3.6 Early Age Stress Development
The stress relaxation of concrete specimen at early ages is measured by a hori
zontal frame specially built for the purpose. The frame can fix the position of the
actuator or it can also be controlled to produce a zero displacement on the specimen.
However, installing LVDTs on hardening concrete specimen is a very difficult task,
therefore only the first method will be performed for this study. The measurement
assumes a uniaxial condition.
The specimen of 820 mm long with 100 mm by 100 mm cross section is cast
directly onto the insulated mold held by the frame. The schematic of the early age
testing is illustrated in Figure 3.6. The mold is equipped with cooper pipes t hat flow
water from a heating-cooling control bath. The measurement can start immediately
after casting. The temperature field in the specimen is measured by at least three
thermocouples. Due to the small cross sectional dimensions, the specimen can have
a quite uniform temperature distribution at all time. In general, it is often not
possible to always apply the same temperature history to the specimen. Because
the initial temperature of concrete after casting cannot be made the same. The end
temperature and the duration of cooling are easier to control. In this study, two cases
are considered. In both cases the end temperature is about 4C and the concrete is
slowly heated up to about 35C prior to cooling. For most cases, the time required
to heat up the concrete is about 12 to 18 hours. In the first case, the period required
Figure 3.5: Schematic of measurement of autogeneous deformation.
4 6
to cool the concrete down to 4C is 48 hours, while in the second case the period is
24 hours.
3.7 Pore Structure Measurement
In this study, sorption isotherms are used to investigate the microstructure and
hydration products oh hydrating blended cement pastes. Two gases used are water
vapor and nitrogen. Water sorption isotherms are obtained for five mixes: 45-1, 45-2,
45-3, 45-4, and 35-1, while nitrogen sorption isotherms are obtained for two mixes:
45-1 and 35-1. For water sorption, the ages of testing are 0.5, 1, 2, 3, 7, 28, and
180-210 days (except for mix 45-4, age 28 days is replaced by 14 days). Meanwhile
for nitrogen sorption, the ages of testing are 0.25, 0.5, 1 , 2, 3, 7, and 28 days. Water
sorption is arguably better than nitrogen sorption because water can access almost
all parts of the solid phase and therefore, is applied more. Both water and nitrogen
sorptions are measured using the machine manufactured by Quantachrome TM.
The mass remaining from the DTA/TGA test is used as the sample, for both
water and nitrogen sorption. Therefore, the sample has been prepared similarly to
the sample used in DTA/TGA test. However, only samples cured at 23C are used
to obtained sorption isotherms. Prior to testing, each sample is outgassed for at least
48 hours to achieve the proper dry condition. The nitrogen isotherms are measured
at 77K, while water isotherms are measured at 15C. On average, each test lasts in
about 2 days.
4 7
Actuator
100 mm x 200 mm cross section
Load cell
/77777777777T7777
Figure 3.6: Test set up for measurement of the early age stress.
4 8
Table 3.2: Test and curing conditions
Test Curing condition Age of testing (days)
/ ' , f sp and E wet 0.5-0.8, 1, 2, 3, 7, 28, <,90
Creep sealed 1, 3, 7, 14
Autogeneous deformation sealed after casting to 7 days
Heat of hydration sealed after casting to 2 1 days
Table 3.3: List of tests performed on materials
Test Materials
Thermal analysis (DTA/TGA) Paste
Isothermal calorimeter Paste
Water and nitrogen sorptions Paste
Semi-adiabatic calorimeter Concrete
Thermal expansion Concrete
Stress relaxation Concrete
Creep Concrete
Compressive strength Concrete
Split tensile strength Concrete
Elastic moduli Concrete
Autogeneous shrinkage Concrete
CHAPTER IV
HYDRATION AND MICROSTRUCTURE OF
BLENDED CEMENT SYSTEMS
4.1 Introduction
Two important aspects in cement and concrete science are hydration and mi
crostructure. Hydration, representing the mass conversion of different phases in the
cement paste, is tied closely to the microstructure development. Meanwhile, the
paste microstructure is known to control the mechanical performance and durability
of cementitious materials. To be able to correlate the performance and durability to
microstructure it is necessary to understand the characteristics and the development
of paste microstructure. This is especially important for blended cement systems
which have been widely used in our modern infrastructure.
Unlike the hydration process in ordinary portland cement, the hydration process
in blended cements is more complex. It involves reactions of mineral additives or
pozzolanic materials in addition to the reaction of portland cement. One quantity
that unambiguously represents hydration of complex systems like blended cements
is heat of hydration. Since almost all chemical reactions are followed by the heat
generation the overall hydration can be measured. For cement hydration, the most
convenient condition for measuring the heat is at an isothermal condition. The other
49
5 0
conditions like the adiabatic and semi-adiabatic often require more involved data
analysis.
Unfortunately, heat is a measure of the combined portland cement-water reaction
and pozzolanic reaction. In order to optimize the use of blended cements we must
understand and be able to quantify the pozzolanic reaction. A method called dif
ferential thermal analysis (DTA) and differential thermogravimetric analysis (TGA)
can be useful for this purpose. There are other means of measuring the degree of
pozzolanic reaction [Abo.l]. However, DTA/TGA has some advantages since it can
give information about the composition of the hydration product and is relatively
quick.
Since cement paste is a porous material, the analysis of the paste microstructure is
usually done to obtain pore characteristics like pore size distribution, pore volume,
and solid surface area. In this study, the focus is on the microstructure of the
hydration product/gel. This is considered more relevant to the global objective in
this study. Many of the gel characteristics are strongly related to major phenomena
like the drying shrinkage, autogeneous deformation, and development of mechanical
properties. The measurement is done to obtain sorption curves for blended cement
pastes.
One stimulating concept of hydration and microstructure development in cement
pastes has been proposed by Powers [Pow.2]. The specific objectives of the inves
tigation on the hydration of blended cements are to develop an understanding on
hydration proceeds and to quantitatively analyze the product compositions based
on Powerss idea. The investigation of microstructure is intended to provide some
insights on gel pore and solid surface area development, and their relationships with
degree of hydration.
51
4.2 Heat of Hydration and Thermal Decomposition
4.2.1 Heat o f Hydration
Plots of heat of hydration (Q) of all mixes at different temperatures are given in
Figures 4.1 to 4.3. Explanations about mix numbers indicated in these figures can
be found in Table 3.1, Chapter 3. Each figure has two sets of curves corresponding to
mixes having two different water-binder ratios (w/b), 0.45 for the first set and 0.35 for
the second set. At the first glance, it is clear that systems with w/b=0.35 produce
heat lower than the ones with w/b=0.45. This is to be expected since the space
available for the growth of hydration products is smaller for the lower w/b systems.
Thus, hydration is slightly retarded in systems with low w/b. The other unique
feature is that higher hydration temperature produce higher heat. However, this is
not necessarily true for the ultimate heat of hydration. Many studies [Kje.l,Abd.l]
reported that at higher temperatures hydration develops faster but ended at lower
ultimate values, while at lower temperatures the ultimate degree of hydration is often
higher.
The pozzolanic reactions can be seen by observing the differences in heat between
ordinary portland cement (OPC) systems (mixes 45-1 and 35-1) and blended cement
systems. Such differences may be more pronounced when heat of hydration is high.
Therefore, will observe only the hydration data at 23C and 34 when discussing
the pozzolanic reactions. There is a slightly different trend in the difference of heat
between the lower w/b systems and the higher w/b systems. However, the overall
trend is still the same. For the pastes made with fly ash, the heat appears to be lower
than the pastes made of OPC, for the entire measurement period. This is not what is
observed in pastes made with ground granulated blast furnace slag (GGB). Although,
the GGB pastes does not generate higher heat initially, they seem to generate heat
5 2
higher than the OPC pastes later on. Such a difference may be attributed to the
slag reaction. The silica fume (SF) pastes exhibit this behavior even earlier than the
GGB pastes. Although this is not the case for pastes with higher w/b. This behavior
can inform us when the pozzolanic reactions start. As depicted from Figures 4.1 and
4.3, the silica fume and slag reactions appear to start earlier than fly ash reaction.
It is important to remember t hat higher heat of hydration does not necessarily
mean higher degree of hydration. The degree of hydration of different mixes must be
calculated by normalizing the heat with its ultimate value. Results presented here
suggest that mixes containing GGB or SF reaches higher ultimate heat of hydration
than the mixes containing OPC. However, heat measurement is not sufficient to
accurately determine the ultimate value, since it is measured only within a short
period of time (around 21 days). The data from DTA/TGA can be used to determine
the ultimate quantity, as results were obtained up to 180 days.
4.2.2 DTA /TG A
The differential thermal analysis (DTA) and the differential thermogravimetric
analysis (TGA) give the information on what type of material and how much it looses
weight when it is decomposed at a certain temperature. The two most important
quantities are the amount of chemically bound water (wb) and the weight loss cor
responding to the decomposition of calcium hydroxide (CH{oss). The procedure for
obtaining both quantities have been described in Chapter 3. For type I OPC, the
maximum value of wb varies from 0.23 to 0.25 per gr ignited sample [Sor.l, You.l,
Min.l]. According to the analysis in [Cop.l], the value of 0.23 is accurate enough
and therefore, wb/ 0.23 can be used to describe the degree of hydration of type I OPC
pastes.
5 3
350
o 45-1
+ 45-2
x 45-3
* 45-4
300
250
" 200
pi
150
1 100
50
1 0
t(h)
300
o 35-1
250
x 35-3
* 35-4
200
100
1 0 1 0 10
t(h)
Figure 4.1: Isothermal heat of hydration at 15C.
5 4
400
350
+ 4 5 - 2
x 4 5 - 3
* 4 5 - 4
300
c 250
150
100
50
t(h)
350
o 35-1
+ 3 5- 2
x 35-3
* 3 5- 4
300
250
? 200
w
O)
O 150
1 0 0
6 +
50
,2 ,3 1
10 1 0 1 0
t(h)
5 5
400
- o 45-1
+ 45-2
. x 45-3
* 45-4
350
300
X | *
c 250
^ 150
1 0 0
50
0
,o ,2 1 .3
1 0 1 0
t(h)
350
^ 0
.+++^+
o 35-1
+ 35-2
x 35-3
* 35-4
300
250
9 x*+
A x+
1 0 0
50
10 1 0
t(h)
56
In Figures 4.4 to 4.11, variations of u/6/0.23 are plotted with time for different
mixes. The pastes hydrated isothermally at 9, 23, and 34. Notice that two of
these curing temperatures are similar to the temperatures where heat hydration
Wf,/0.23. However, the use of the amount chemically bound water as a measure of
hydration in blended cement systems is yet to be justified. Results presented here
in terms of normalized chemically bound water (wb/0.23).
From Figures 4.4 to 4.11, it is seen that Wb/0.23 in the blended systems is, for some
curing time, less than that for OPC. In the early hydration period, this difference
pastes, and close to 10% in SF pastes. This finding is associated to the fact that in
the beginning only the OPC portion of the total binder (OPC and pozzolan) hydrates
with water. In other words, the difference is proportional to the OPC fraction in
each blended system, i.e. OPC is 25% less in FA and GGB pastes, and 10% less in
SF pastes.
4.2.3 Effect of Temperature
The hydration data can fitted accurately by an equation similar to the one used
in [Han.4] for describing the strength development in concrete with respect to the
maturity. This equation is written as:
where a can be used to describe degree of hydration, heat of hydration or chemically
bound water. All three parameters in TPE, Oqo, r , and a can be used to describe
the effect of temperature. Fits of experimental hydration data are shown as solid
are measured. This allows direct comparisons between heat of hydration (Q) and
are used only to distinguish the differences in hydration process of blended systems
in amount of chemically bound water is very close to 25% for FA pastes and GGB
(4.1)
5 7
Mix 45-1
o 9C
+ 23C
* 34C
0.9
0.8
0.7
0.6
m
cm
0.5
0.4
0.3
0.2
A-
0.1 -
1 0
time (h)
Figure 4.4: Chemically bound water of mix 45-1.
Mix 4 5- 2
o 9C
+ 23C
* 34C
0.9
0.8
0.7
0.6
m
CM
o 0.5
0.4
0.3
+s
O
0.2
0.1
1 0
time (h)
1 0
5 8
Mix 45-3
o 9C
+ 23C
* 34C
0.9
0.8
0.7
0.6
co
CM
o' 0.5
0.4
0.3
O'
0.2
0.1
10
time (h)
Mix 45-4
o 9C
+ 23C
* 34C
0.9
0.8
0.7
0.6
co
CM
o 0.5
0.4
' O-
+ ' O / 0.3
0.2
0.1
O
1 0 10 10
time (h)
5 9
Mix 3 5 -
o 9C
+ 23C
* 34C
0.9
0.8
0.7
0.6
CO
CM
o 0.5
-0
3
0.4
V
0.3
0.2
0.1
10 1 0
time (h)
1 0 1 0
Mix 35- 2
o 9C
+ 23C
* 34C
0.9
0.8
0.7
0.6
CO
CM
O 0.5
n
3
0.4
>o
0.3
O'
O'
0.2
0.1
1 0 1 0 1 0
time (h)
6 0
Mix 35-3
o 9C
+ 23C
* 34C
0.9
0.8
0.7
0.6
C*5
CM
o 0-5
a
3
0.4
O'
0.3 O
O '
0.2
0.1
1 0
time (h)
Mix 35-4
o 9C
+ 23C
* 34C
0.9
0.8
0.7
0.6
cn
CM
o 0.5
.O
0.4
0.3
0.2
0.1
1 0
time (h)
6 1
lines in Figures 4.4 to 4.11. As has been discussed, the heat rate is generally higher
when hydration takes place at higher temperature. The same phenomenon is also
observed on the development of u/j,/0.23 versus time which suggests that w\,j0.23 may
represent hydration of blended cement systems, although it is only used to represent
the degree of hydration of OPC so far.
As has been confirmed in many studies [Abd.l, Kje.l], ty6/0.23 in OPC is seen to
reach higher ultimate values at lower temperature. This has been shown in Tables
4.1 and 4.2 where the parameters of TPE for the normalized bound water are listed.
The same behavior is also observed in the parameters of TPE obtained for heat of
hydration data, which are shown in Tables 4.3 and 4.4 The systems with blended ce
ments also exhibit the same behavior. This comes with no surprise since the blended
systems are still dominated by OPC. This further implies that Wb can very likely
represent the hydration of blended systems. However, we still need to confirm this
by comparing Wb with heat of hydration. Also, the maximum amount of chemically
bound water needs to be determined for the blended systems in order for the degree
of hydration (a) to be calculated, i.e. a. = Wb/wboQ-
4.2.4 H e a t ( Q) a n d Combined W a t e r (wb)
The plots of heat as a function of combined water are shown in Figure 4.12. The
data points in each plot correspond to measurements at two temperatures, 23 and
34C, and at two water-binder ratios (w/b). If unique relations exist between Q
and Wb and they both describe the amount of hydrated cement then factors such
as temperature and w/b should not affect their relationship. The only factor that
differentiate each mix is the type of binder. Indeed, this is what we observe in
Figure 4.12. In each figure, all data points seems to merge into a single curve. More
6 2
Table 4.1: TPE parameters for bound water data, w/b=0.45.
Mix 45-1
T(C) wb, oc/0.23 r (hours) a
9 0.982 84.043 0.484
23 0.949 25.399 0.5895
34 0.947 14.152 0.4704
Mix 45-2
T(C) Wb, oc/0.23 r (hours) a
9 0.899 142.14 0.402
23 0.832 36.914 0.482
34 0.769 15.271 0.545
Mix 45-3
9 1-387 459.99 0.2607
23 1.153 82.955 0.3575
34 0.978 19.267 0.4946
Mix 45-4
T(C)
W6,oo/0.23
r (hours) a
9 1.034 126.05 0.3576
23 0.975 43.172 0.3592
34 0.842 11.483 0.5304
63
Table 4.2: TPE parameters for bound water data, w/b=0.35.
Mix 35-1
T(C) W6,oo/0.23 r (hours) a
9 0.860 68.08 0.523
23 0.877 23.439 0.593
34 0.845 9.328 0.5102
Mix 35-2
T(C) W6,oo/0-23 r (hours) a
9 0.892 135.16 0.388
23 0.848 40.062 0.3806
34 0.756 15.038 0.497
Mix 35-3
9 1 . 1 1 2 220.84 0.2866
23 0.976 51.405 0.4147
34 0.983 23.205 0.4381
Mix 35-4
T(C) ^6,oo/0.23 r (hours) a
9 0.929 94.778 0.354
23 0.840 28.092 0.434
34 0.796 9.7606 0.5115
64
Table 4.3: TPE parameters for isothermal data, w/b=0.45.
Mix 45-1
T(C)
Qoo (J/gr)
r (hours) a
15 373.95 24.783 0.76238
23 367.40 14.395 0.87068
34 355.14 8.3019 1.0044
Mix 45-2
T(C)
Qoo (J/gr)
r (hours) a
15 350.94 32.094 0.65141
23 347.62 17.359 0.73667
34 342.95 10.655 0.80537
Mix 45-3
T(C)
Qoo (J/gr)
r (hours) a
15 421.83 41.286 0.53383
23 411.05 21.589 0.65199
34 380.64 11.743 0.80618
Mix 45-4
T(C)
Qoo (J/gr)
r (hours) a
15 370.66 30.25 0.59273
23 360.19 16.372 0.69006
34 346.81 9.9344 0.7341
65
Table 4.4: TPE parameters for isothermal heat data, w/b=0.35.
Mix 35-1
T(C)
Qoo (J/gr)
t (hours) a
15 308.27 19.037 0 . 8 6 8
23 306.88 10.923 0.995
34 302.72 6.828 1.048
Mix 35-2
T(C)
Qoo (J/gr)
r (hours) a
15 308.55 25.957 0.721
23 300.52 14.326 0.845
34 288.91 8.876 0 . 8 8 6
Mix 35-3
T(C)
Qoo (J/gr)
r (hours) a
15 348.10 29.777 0.620
23 338.99 15.605 0.776
34 325.12 9.2938 0.906
Mix 35-4
T(C)
Qoo (J/gr)
t (hours) a
15 313.08 20.709 0.728
23 339.44 11.975 0.790
34 344.12 7.775 0.773
66
presentation of such a feature will also be given in Chapter 6, where the analysis
of the effect of temperature on hydration is related the effect of temperature on
mechanical properties.
A slightly nonlinear relation can be deduced from each figure showing Q-Wb plots.
In fact, it is not important as long as Q and Wb relate to each other uniquely and
both are monotonically increasing, as hydration products always increase with time
and therefore so do Q and Wb- Hence, just like Q, Wb also measures the degree of
hydration of blended cement pastes. The remaining question now is how to quantify
the degree of hydration of the pozzolan.
4.3 Degree of Hydration of Blended Systems and Pozzolans
4.3.1 Procedure for Determining the Degree of Hydration
In general, the degree of hydration is defined as Q/Qoo or Wb/wôo, which requires
Qoo or iL>b,oo to be determined. Unfortunately, the heat data are not measured long
enough (up to 21 days) to allow for accurate determination of Qoo- The chemically
bound water is however, measured on mature paste specimens (up to more than
2 0 0 -day old) and hence, the predicted values of wôo based on the three-parameter
equation (TPE) can be used. iU6j00 used must be obtained for mixes having w/b=0.45
because in these mixes, hydration is allowed to reach the higher values, while for
mixes having w/b=0.35 hydration is limited by the capillary space.
From Tables 4.1 and 4.2, it can be said that the ultimate bound water increases
as temperature decreases. The plot of the ultimate chemically bound water (wb,oo)
versus temperature is given in Figure 4.13. A linear relation can be seen in this
figure. The maximum value of Wbt00 can be found by extrapolation at the lowest
possible hydration temperature, which in this study was chosen to be 0.5C. The
Q
(
J
/
g
r
-
b
i
n
d
e
r
)

Q
(
J
/
g
r
-
b
i
n
d
e
r
)
6 7
400
300
200
1 0 0
0
0 0.05 0.1 0.15 0.2
wb (per gr binder)
100% OPC
x
6n n + D
HO
o+
O w/b=0.45,23C
x w/b=0.45,34C
+ w/b=0.35,23C
-b
w/b=0.35,34C
350
75% OPC+25% FA
300
O +
v 1
250
200
50
0.05 0.1
wh (per 9r binder)
0.15 0.2
400
75% OPC+25% GGB
300
a
jp
200
1 0 0
0.15 0.2 0.05 0.1
w. (per gr binder)
f
90% OPC+10% SF
o 250
5? 150
O 1 0 0
0.05 0.1 0.15
wb (per gr binder)
Figure 4.12: Relations between heat of hydration and chemically bound water.
68
degree of hydration of each paste mix is therefore,
= Wb,oo(T 0.5C) (4' 2)
The extrapolated values of w6iOCis listed in Table 4.5.
Plots of a with time are shown in Figure 4.14. Surprisingly, the lowest degree of
hydration (around 6 6 % at more than 200 days) is exhibited by GGB paste. Mean
while, the degree of hydration of FA paste and SF paste are relatively high (around
80% after more than 200 days). Another interesting finding is that for OPC, wôo is
very close to 0.23 gr/gr ignited solid, just as as suggested in the literature. This sup
ports the validity of the procedure used. The same idea can also be applied to heat
of hydration. However, the maximum values of heat of hydration determined using
the above procedure correspond to the short term heat measurement (Q0>oo)- The
plots of Q{t)/Q0,oo(T 0.5C) are shown in Figure 4.15. It is clear that Q(t)/Q0i00
exhibits a trend similar to Wb(t)/wb,oo- This finding suggests that Wb represents the
overall degree of hydration and supports the validity of the above procedure.
A more accurate prediction of the maximum ultimate heat (Qoo) can be done
with the help of the relations between Q and Wb shown in Figure 4.12. By fitting
such relations, values of Qoo corresponding to u6j00 can be found. The values of both
Q0,00 and Qoo obtained using the above method are listed in Table 4.5. Notice that
Qoo in Table 4.5 represents the characteristic heat of hydration per unit weight of
binder. Values below 500 J / g r have been reported in the literature [Hel.2,Min.l] for
OPC type I, which are consistent with the present finding.
An attempt to use Wb/wb,oo as the degree of hydration was made by [Mik.2] for
slag-OPC systems. However, they did not determine Wbj00 at different temperatures,
only at room temperature. Because they used a lower Wbi00>their degree of hydration
6 9
o Mix 45-1-
+ Mix 45-2
x Mix 45-3.
* Mix 45-4
0.34
0.32
0.3
0.28
1-0-26
0.24
0.22
O'
0.2
0.18 -
0.16
2 0
Temperature f c )
1 0 15 25 30 35
Figure 4.13: Relations between wbi00 and temperature, w/b=0.45.
is slightly higher than what reported here.
4.3.2 The Degree o f Hydration o f Fly Ash, Slag, and Silica Fume
Pozzolans can react with hydration products (calcium silicate hydrate and cal
cium hydroxide) of OPC and water. Calcium hydroxide (CH) is known to react with
FA and SF, and GGB and water [Abo.l,Shi-1,You.l]. This becomes more evident
as the amount of CH is plotted against the specimen age. For OPC pastes CH
increases as the age increases indicating a proportional relation with the degree of
hydration or wb. In pastes containing FA, GGB, and SF, CH initially increases then
drops as it is consumed more and more by pozzolans. How specifically FA, GGB,
and SF react with water and CH is not the subject of this investigation. However,
experimental data provide an evidence that CH produced by blended cement hydra
tion (containing FA, GGB, and SF) is less than CH produced by OPC hydration
7 0
1
o Mix 45-1
+ Mix 45-2
x Mix 45-3
* Mix 45-4
0.9
0.8
0.7
t 6
~d0.5
s 0.4
'/-'
7 X '
/
0.3
0.2
0.1
1 0
time (h)
Figure 4.14: Degree of hydration of blended cement pastes.
o 45-1
+ 45-2
x 45-3
* 45-4
0.9
O +*^xx
0.8
0.7
0.6
5 0.5
0.4
0.3
0.2
0.1
1 0
t(h)
Figure 4.15: Q/Qoo of blended cement pastes from short term heat measurement.
71
Table 4.5: Values of maximum ultimate bound water and maximum ultimate heat
(per gr binder).
Mix
^ 6,oo
QoUJ) QooiJ)
100 % OPC 0.228 388.44 438.74
75% OPC+25% FA 0.217 356.84 453.40
75% OPC+25% GGB 0.318 455.19 582.96
90% OPC+10% SF 0.256 388.03 513.85
[Abo.l,Shi.l,Han.5,Mos.l]. Based on this argument, the variation of the amount of
CH with time can be used as a measure of pozzolan reaction. In this work, only
the weight loss associated with the decomposition of CH (CHjoss) can be measured.
However, as shown in [Mid.l] this weight loss is very close to the amount of water in
CH and therefore, is proportional to the amount of CH.
Variations of CH;oss with wb are shown in Figures 4.16 to 4.19. All data points
were generated from measurements at three temperatures and for two w/b. In corre
lating CHioss to wb, it was assumed that CH was produced only by OPC, not by OPC
and pozzolan. Therefore, CHioss to wb were normalized by the weight of OPC in each
mix, which was estimated from the proportion between OPC and pozzolan, e.g. 75%
OPC+25% FA. This means that (% OPC)iy&/gr-OPC=u;&/gr-binder. CH[oss versus
wb in each system in seems to follow one unique trend line. For OPC pastes, CHioss
changes almost linearly with wb. This comes with no surprise since the hydration
products, which is proportional to wb, include CH. However, for pastes containing
pozzolans the amount of CH is less, especially in FA pastes. Such a trend has been
experimentally confirmed [Hel.2].
In order to estimate the degree of hydration of pozzolans, a reference CH/oss-u;b
plot is needed. Clearly, this would be the plot for OPC. The difference in CH*oss-u;6
72
0.07
0.06
- 0.04
O 4 5 - 1 , 9C
+ 4 5 - 1 , 23C
45- 1, 3 4 C
35-1, 9C
35-1, 23C
35-1, 34C
fit
0.03
0.05 0.1 0.15
w (per gr OPC)
0.2 0.25
Figure 4.16: Relation between CH;oss and Wb for OPC.
0.07
90.03
o
45-2, 9C
+
45-2, 23C
X
45-2, 34C
*-
35-2, 9C
35-2, 23C
O
35-2, 34C
fit OPC
-
fitOPC+FA
(per gr OPC)
Figure 4.17: Relation between CH;G5S and wb for 75% OPC+25% FA.
7 3
0.08
0.07
0.06
o
0.05
o>
0.04
o
45- 3, 9C
+
45- 3, 23C
X
45- 3, 34C
*
35- 3, 9C
35-3, 23C
0
35- 3, 34C
fit OPC
-----
fit OPC+GGB
<b * 0
0.1 0.15 0.2
w (per gr OPC)
0.25 0.3
Figure 4.18: Relation between CH;oss and Wb for 75% OPC+25% GGB.
0.07
0.06
o
Q.
O
- 0 . 0 4
O)
w
0)
Q.
O
45-4, 9C
+
45-4, 23C
X
45-4, 34C
*
35-4, 9C
35-4, 23C
0
35-4, 34C
fit OPC
-----
fit OPC+SF
0.2 0.25
w (per gr OPC)
Figure 4.19: Relation between CH/oss and Wb for 90% OPC+10% SF.
7 4
plot obtained for each blended cement system relative to CHioss-Wb plot for OPC
normalized by its ultimate value is the degree of pozzolan reaction (ap). To calculate
such a difference, the data points need to be curve-fitted. Results of fitting the data
are shown in Figures 4.16 to 4.19. In each figure, the CHiOss-wb plot for OPC is
included and indicated by the solid line. The degree of pozzolan reaction is calculated
from:
and from the inversion of Eq.(4.1) the time can be expressed in terms of
so OLp {t ) can be found. The degree of pozzolan reaction determined by the above
calculation is shown in Figure 4.20. As seen, SF starts reacting earlier than FA
or GGB, while, GGB appears to react earlier than FA. These finding confirm the
observation on heat of hydration (Figures 4.1 to 4.3), that the pozzolan reactions in
pastes containing SF and GGB occur early. Figure 4.20 also provides the information
on how reactive different types of mineral additive are.
4.4 H2 O and N2 Sorption Isotherms
Sorption isotherms of cement pastes can be categorized as type IV according to
the classification of Brunauer, Deming, Deming, and Teller (BDDT) [Gre.l]. This
is the typical case of adsorption in mesoporous solids. One characteristic of type IV
isotherms is the occurrence of hysteretic loops. In mesoporous solids, such hysteretic
loops are associated to the capillary condensation. In cement paste, as will also
define the reduction of Ca(OH) 2 due to the pozzolanic reaction. These difference
Or,
C H spsc (wb) - CHloss{wb)
(4.3)
p C f i g ^ K o o ) - CHloss(wb,oo)
_ A CH[oss(wb(t))
pl A C i W a O
(4.4)
(4.5)
7 5
0.7
o 75% OPC+25% FA
+ 75% OPC+25% GGB
x 90% OPC+10% SF
0.6
0.5
0.4
a
P
0.3
0.2
0.1
I Q CHOIQ-O
1 0 1 1 0 - 1 0
t(h)
Figure 4.20: Degree of pozzolan reaction.
illustrated later, another source of hysteretic is due to the closing and opening of the
interlayer space [Ram.2].
In the past, sorption isotherms of cement paste were commonly obtained using
N2 because it was believed that purely physical interaction (without chemisorption)
could be expected from using N2 [Ram.2 ]. Measurements using H20 had also be done
in the past [Mik.l,Pow.2]. However, physical properties like the surface area and pore
size distribution obtained using H20 were always significantly different from those
obtained using N2. In particular the H20 isotherms always gave higher surface area.
This has brought some controversy among cement chemist. However, many applied
problems involving the use of cement or concrete are tied closely to the cement-water
interaction, for example drying shrinkage, creep, and strength behavior. We will see
that this argument holds in this study and more consistent results can be obtained
76
from H20 isotherms.
4.4.1 Characterization o f Isotherms
H20 sorption isotherms are shown in Figures 4.21 to 4.25 and N2 sorption isotherms
are shown in Figures 4.26 and 4.27. The purpose of doing measurements only on
selected mixes has been described in Chapter 3. Some important intrinsic behav
ior of H20 sorption is the occurrence of hysteretic loops. A close observation on
a adsorption-desorption curves reveals that the loop consists of two features. The
first feature is an almost uniform separation between the adsorption and desorption
curves. The second feature is the downward movement of the desorption curve near
the relative humidity (h) of 0.25. Yet, the desorption curve never closes at h=0.25.
It is very likely that these features describe the combined effect of capillary con
densation and movement of interlayer water. The capillary condensation makes the
desorption curve to follow the adsorption curve after the passing the lowest h that
allows the condensation. But the movement of interlayer water causes the specimen
to deform. What likely happens is that upon drying and outgassing the interlayer
space is emptied and collapses. As adsorption begins, multilayer water molecules
start filling in the space and forcing it to open. Then when desorption starts, wa
ter does not take the same path as it entered the interlayer space and some of its
molecules may be trapped in this very small space. Keep in mind that interlayer
space measures around 1 to 2 molecular diameter in thickness [Ram.2].
What is just described above cannot be clearly seen in N2 sorption isotherms.
The other characteristic feature, the hysteretic due to capillary condensation, is still
noticeable in N2 sorption isotherms. The shape of N2 isotherms however, are quite
different from H20 isotherms indicating differences in adsorbate-adsorbent interac-
7 7
tions.
The classical adsorption equation by Brunauer, Emmet, and Teller, called the
BET equation, has been widely used in the past to fit sorption data. This equation
is written as [Gre.l]:
M = cMrnh (4 6 )
where the parameter c relates to the net heat of adsorption ( ql) is given by:
q i - q L = R T l n . c (4.7)
In the above expression, M is the adsorbed mass, Mm is the monolayer adsorbed
mass, and h is the relative humidity. The BET equation was developed based on
multilayer adsorption theory and was not for mesoporous solids in which the capil
lary condensation may occur. In this study, as has been mentioned above and will
also verified further, the capillary condensation occurs between h=0.25 to h=0.3.
Therefore, the BET equation can only be used between h= 0 to h=0.25. A further
modification of BET equation, called the modified BET equation, was proposed by
[Gre.l]. Unlike the BET equation, the modified BET equation can fit sorption data
accurately even up to relatively high h. The modified BET equation reads:
M = ---------- --------------------- (4 8 )
( 1 kh)[l -t- (c l)fc/i] 1 '
Notice that the modification is done by multiplying h with a constant k that must
always be less than one. This modification assures that M will be finite at h=1 . The
accuracy of Eq.(4.8) in fitting adsorption data has been demonstrated in [Xi.l,Xi.2].
4.4.2 Surface Area of Solid
The surface area of solid in cement paste plays an important role in helping us
understanding the physical nature of the hydration products. Commonly, the surface
R
e
p
r
o
d
u
c
e
d

w
i
t
h

p
e
r
m
i
s
s
i
o
n

o
f

t
h
e

c
o
p
y
r
i
g
h
t

o
w
n
e
r
.

F
u
r
t
h
e
r

r
e
p
r
o
d
u
c
t
i
o
n

p
r
o
h
i
b
i
t
e
d

w
i
t
h
o
u
t

p
e
r
m
i
s
s
i
o
n
.
*4
t
(g
f-i
n>
to
rt-
fD
i-l
Cfl
O
O
P
M
o
e*
tr
n>
R
at
i
t=0.5d t=1d
o
o
>
0.25 0.5 0.75 1
Relative humidity (h)
0.25 0.5 0.75
t=2d
O)
o
o
o
>
0.25 0.5 0.75
100
t=3d
o
>
0.25 0.5 0.75
150 150
t=28d t=7d
o 100
p
o
>
0.25 0.5 0.75
0.25 0.5 0.75
'l
00
150
t=191d
10 0
p>
o
o
o
>
0.25 0.5 0.75
R
e
p
r
o
d
u
c
e
d

w
i
t
h

p
e
r
m
i
s
s
i
o
n

o
f

t
h
e

c
o
p
y
r
i
g
h
t

o
w
n
e
r
.

F
u
r
t
h
e
r

r
e
p
r
o
d
u
c
t
i
o
n

p
r
o
h
i
b
i
t
e
d

w
i
t
h
o
u
t

p
e
r
m
i
s
s
i
o
n
.
*1
H
03
I
(T>
to
to
t
ct-
a>
C/3
O
n
*d
cr
*-
O
P
w
o
r+
PT-
fD
B
CA
O
S
O t
i
to
O)
o
o
>
40
t=0.5d
30
2 0
10
0
0 0.25 0.5 0.75 1
1 00
t=3d
k_
O)
o
o
o
>
0.25 0.5 0.75
150
t=191d
o
>
0 0.25 0.5 0.75 1
V
o
l
.

(
c
c
/
g
r
)
80
t=1d
60
40
2 0
0
0 0.25 0.5 0.75 1
150
t=7d
1 0 0
o
>
0.25 0.5 0.75
0.25 0.5 0.75
150
t=28d
1 0 0
o
>
0.25 0.5 0.75
R
e
p
r
o
d
u
c
e
d

w
i
t
h

p
e
r
m
i
s
s
i
o
n

o
f

t
h
e

c
o
p
y
r
i
g
h
t

o
w
n
e
r
.

F
u
r
t
h
e
r

r
e
p
r
o
d
u
c
t
i
o
n

p
r
o
h
i
b
i
t
e
d

w
i
t
h
o
u
t

p
e
r
m
i
s
s
i
o
n
.
*n
t
<2
l-s
io
CO
I
e+
(Si
O
rt
o
p
M*
C/)
o
r r
ro
3
Cfl
O
f-n
R'
Ol
t
CO
50 80
t=0,5d t=1d
60
k.
p>
o
o
CD
40
o
>
2 0
0
0.25 0.5 0.75 0 0.25 0.5 0.75 1
1 0 0 150
t=3d
t=7d
100
O)
0.25 0.5 0.75
0.25 0.5 0.75
100
t=2d
o
>
0.25 0.5 0.75
150
t=28d
1 0 0
o
o
0.25 0.5 0.75
O)
o
o
o
>
150
t=191d
100
50
0 <
0 0.25 0.5 0.75 1
R
e
p
r
o
d
u
c
e
d

w
i
t
h

p
e
r
m
i
s
s
i
o
n

o
f

t
h
e

c
o
p
y
r
i
g
h
t

o
w
n
e
r
.

F
u
r
t
h
e
r

r
e
p
r
o
d
u
c
t
i
o
n

p
r
o
h
i
b
i
t
e
d

w
i
t
h
o
u
t

p
e
r
m
i
s
s
i
o
n
.
*8
i
(D
io
CT
a>
n
m
O
*-*
O
rr
H*
o
p
co
O
e-h
cr
-1
S'
Ol
I
60
&40
t5
o
2 0
o
>
t=0.5d
150
g 100
o
o
50
0.25 0.5 0.75
t=3d
0.25 0.5 0.75
2 0 0
-S>
o
o
o
>
t=208d
0.25 0.5 0.75
V
o
l
.

(
c
c
/
g
r
)

V
o
l
.

(
c
c
/
g
r
)
100
t=1d
0.25 0.5 0.75
1 0 0
o
o
o
>
150
t=16d
100
0.25 0.5 0.75
0.25 0.5 0.75
150
t=28d
1 0 0
8
o
>
0.25 0.5 0.75
R
e
p
r
o
d
u
c
e
d

w
i
t
h

p
e
r
m
i
s
s
i
o
n

o
f

t
h
e

c
o
p
y
r
i
g
h
t

o
w
n
e
r
.

F
u
r
t
h
e
r

r
e
p
r
o
d
u
c
t
i
o
n

p
r
o
h
i
b
i
t
e
d

w
i
t
h
o
u
t

p
e
r
m
i
s
s
i
o
n
.
+*
%
-
to
01
3
t
ct
a>
<1
V)
o
o
P
35
0
e-t-
t=r
a>
3
V)
8 ,
3
Pi-
CO
01
I
t=0.5d
O)
c5
o. 30
> 2 0
0.25 0.5 0.75
t=3d
0.25 0.5 0.75
O )
o
o
o
>
150
t=204d
100
V-
8
o
>
0 0.25 0.5 0.75 1
0.25 0.5 0.75
0.25 0.5 0.75
0.25 0.5 0.75
10 0
t=28d
o
o
>
R
e
p
r
o
d
u
c
e
d

w
i
t
h

p
e
r
m
i
s
s
i
o
n

o
f

t
h
e

c
o
p
y
r
i
g
h
t

o
w
n
e
r
.

F
u
r
t
h
e
r

r
e
p
r
o
d
u
c
t
i
o
n

p
r
o
h
i
b
i
t
e
d

w
i
t
h
o
u
t

p
e
r
m
i
s
s
i
o
n
.
*1
i
<8
i-i
to
05
et
O
fD
P
CO
O
3
rt-
o
p
CO
o
et-
tr
3
CO
o
S'
Ol
i
t=0.25d
o
>
0.25 0.5 0.75
0.25 0,5 0.75
S 1 00
o
o
o
>
t=28d
0.25 0.5 0.75
V
o
l
.

(
c
c
/
g
r
)

V
o
l
.

(
c
c
/
g
r
)
100
150
t=1d
t=0.5d
0.25 0.5 0.75 1
0.25 0.5 0.75
150
t=3d
10 0
0.25 0.5 0.75
150
t=7d
1 0 0
k_
-!?
o
o
0.25 0.5 0.75
1
R
e
p
r
o
d
u
c
e
d

w
i
t
h

p
e
r
m
i
s
s
i
o
n

o
f

t
h
e

c
o
p
y
r
i
g
h
t

o
w
n
e
r
.

F
u
r
t
h
e
r

r
e
p
r
o
d
u
c
t
i
o
n

p
r
o
h
i
b
i
t
e
d

w
i
t
h
o
u
t

p
e
r
m
i
s
s
i
o
n
.
<9
if
to
'21
rt-
O
cD
P
W
O
o
p
*
c/>
o
ct-
tT
a>
CO
0
l-h
3
*
X
w
at
1
150 100 150
t=1d
t=0.25d t=0.5d
100
O) O)
o
>
0 0.25 0.5 0.75 1
0.25 0.5 0.75
0.25 0.5 0.75
150 150
100
t=2d
t=3d t=7d
100 w
P)
O
O
O)
o
>
o
>
0.25 0.5 0.75 1
0.25 0.5 0.75
0.25 0.5 0.75
150
t=28d
100
o
o
o
>
0 0.25 0.5 0.75 1
00
85
area of paste is calculated from H2 O or N2 sorption isotherms. Unfortunately, the
results do not agree with each other. This debatable issue has for long time been a
subject of controversy. However, with a proper treatment (like solvent replacement)
on the specimen the results obtained using H20 and N2 can become very close to
each other [Lit.l]. As explained in Chapter 3, this solvent replacement was done to
all specimens tested. For the reason explained earlier, we will carefully analyze the
surface area generated from H20 sorption then, compare it with the one obtained
from N2 sorption.
There two methods to calculate of the surface area: using the BET equation and
using the statistical thickness curve. The first method, normally for the adsorption
data, can be done by using the following relation:
h _ 1 , c 1 1, rAn\
M( 1 - h) ~ Mmc + (
This is a linear relation between h / M ( 1 h) and h with the slope (c 1 ) / M mc and
the intercept 1 / M mc at h = 0. Once h / M ( l h) vs. h from the sorption data is fitted
using a linear equation, the mass of the monomolecular layer (Mm) can be obtained.
The surface area can be calculated from Mm [Gre.l]. The second method requires
some reference curves of volume versus h generated from a nonporous material.
Examples of the applications of this method can be found in [Han.lj. It has been
reported in [Bru.3] that the second method agrees very well with the BET method
as long as the constant c of the reference nonporous material and the constant c of
the measured material match each other.
In this study, the surface area will be calculated using the BET equation. A
different approach is taken to estimate the true surface area of solid pastes. First,
Eq.(4.9) is applied to the adsorption data. Since the data is expressed as a volume
86
quantity, the values of the volume of monolayer Vm are obtained instead of Mm. The
conversion from VTn to Mm can be easily done assuming ideal gas law at the standard
condition (STP). Then, the sorption data is normalized by Vm. The purpose is to
express the sorption data as number of layers instead of volume. Next, the normalized
data (V/Vm) are compared with the statistical thickness curve. Two curves are widely
used for studying cementitious materials. The first one is proposed by [Set.2 ] and
In order to use Eq.(4.10) and (4.11) together with VfVm, the thickness (t) must
be normalized by a single molecule thickness (tm) which is estimated to be around
3.2 Angstrom [Ram.2,Koe.l]. The plot of t / t m with h describes the number of
layer forming during adsorption or desorption of water on nonporous solids. The
comparison between VfVm and t / t m will allow us to first, determine the value of
h associated with the monolayer thickness and second, find out the volume of gel
space.
Comparisons can now be made between V/Vm and t / t m in terms of number of
layers. Plots of both V/Vm and t / t m for different mixes at several ages are shown
in Figures 4.28 to 4.32. Each figure gives plots of V/Vm with h for both adsorption
and desorption. As mentioned earlier, the adsorption curve is likely to represent
the paste whose interlayer space collapses. On the other hand, the desorption curve
is considered to have an open microstructure since at the desorption branch the
interlayer space opens. By looking at V/Vm and t / t m curves the monolayer capacity
reads
(4.10)
and the second one by [Bad.l] reads (in Angstrom):
r = t = 3.95 1.89 ln(In h) (4.11)
8 7
I
E 3
o 0.5d
X
1d
2d
+ 3d
> 7d
* 28d
0 191d
----- Eq.(4.11)
Eq.(4.10)
4f*> 9
Mix 45-1 adsorption
0.4 0.6
6
5
4
E
evi
2
* 0 0 0 0 0 0 0 0 0
1
Mix 45-1 desorption
0
0.8 0.6 1 0 0.2 0.4
Figure 4.28: Comparison between V/Vm and t / t m for mix 45-1.
88
6
O 0.5d
5
> 7d
* 28d
O 191d
Eq.(4.11)
- Eq.(4.10)
4
3
2
1
Mix 4 5 - 2 adsorption
0
0.8 0 0.6 1 0.2 0.4
6
5
4
e
CVJ
Ojt
2
o o
Mix 4 5 - 2 desorption
0
0 0.6 0.8 1 0.2 0.4
Figure 4.29: Comparison between V/V-m and t / t m for mix 45-2.
89
O 0.5d
> 7d
* 28d
O 191d
Eq.(4.11)
- Eq.(4.10)
E 3
1 - -
Mix 45 - 3 adsorption
0.2 0.4
Relative
0.6
0.8
6
5
4
E
3
oj
2
1
Mix 4 5 - 3 desorption
0
0.2 0 0.4 0.8 0.6 1
Figure 4.30: Comparison, between V/Vm and t / t m for mix 45-3.
90
6
O 0.5d
5
D> 28d
* 208d
Eq.(4.11)
- Eq.(4.10)
4
E 3
2
1
Mix 45- 4 adsorption
0
0 0.2 0.4 0.6 1 0.8
6
5
4
E
oj

o Ot t* B*
2
1
Mix 45- 4 desorption
0
0.2 0 0.4 0.6 1 0.8
91
o 0.5d
X
1d
o 2d
+ 3d
> 7d
28d
0 204d
Eq.()
-
Eq.()
* > # *
&
f t
Mix 35-1 adsorption
>
cy
E 3 -
0.4 0.6
Mix 35-1 desorption
0.4 0.6 0.8
9 2
of the pastes during the desorption brach can be found. Therefore, the surface area
can be calculated based on the desorption curves.
From Figures 4.28 to 4.32, the monolayer filling (V/Vm = 1 and t / t m = 1 ) along
the adsorption branch can be seen to occur around h=0.21. This is determined
from the intersection between the set of data points and the two curves based on
Eq.(4.10) and (4.11). Notice that the limit h=0.21 is very close to the limit we the
BET equation was applied to find the monolayer volume Vm. It can be seen that, for
the range of h=0 to h0.21, V/Vm curves on the adsorption branch are very close
to each other although they are obtained at different ages. Clearly, Vm represent the
surface area of the solid and therefore, changes with age. A similar feature can also
be seen in the desorption branch but oniy after the volume is normalized by 2Vm
(Vm is always calculated from adosprtion). During the desorption, the interlayer
microstructure can be considered more open and more accessible by water. This
condition is arguably closer to the state of microstructure prior to drying. According
to Figures 4.28 to 4.32, the monoloyer volume during the desorption branch is about
2Vm. The difference in the monoloyer volume between the adosorption and desorption
may be attributable to the interlayer space, which is about 2 V^ Vm = Vm.
The surface area calculated from Vm are shown in Figure 4.33 as it varies with
wt,/0.23. The true solid surface area is twice of that shown since it is calculated from
2Vm or (only for H20 sorption):
Sdes = 2Sads (4.12)
The increasing trend is to be expected since adsorbed water occupies the surface and
space of the hydration products. Since the hydration products increase with Wb/0.23
then the area per unit mass of solid increases as well. Also shown in Figure 4.33 is
9 3
the surface area obtained from N2 sorption. A similar trend between the surface area
and is also observed. However, the surface area generated from N2 sorption
is considerably smaller than the one obtained from H20 sorption. The reason is
apparent, N2 molecules cannot penetrate into some space like the interlayer space
like H20 molecules do. This also means that the surface area calculated using N2
sorption will be the same regardless whether it is calculated for the adsorption or
desorption branches.
4.4.3 Surface area o f Hydration Products
In the previous discussion, the surface area is not the intrinsic surface area of
the hydration products/gel because the pastes have not fully hydrated. To calculate
the specific surface area of the hydration products some quantitative volume-mass
relations for blended cements are needed. Powers [Pow.2] developed a phase-volume
model which is based on water sorption isotherms. VVe shall apply the Powerss
concept to calculate the surface area of gel. First of all however, it needs to be shown
that the solid surface measured belongs to the gel. In Figure 4.33, the surface area
changes proportionally with hydration (wi,) and therefore is directly proportional to
the amount of gel in the paste. But the surface area has been calculated based on
the dry weight of the sample not the weight of the gel. When a equals 1, the surface
area of the sample becomes the gel surface area.
To calculate the gel surface area let us consider a mass of solid in cement paste,
Ms = C ( l - a ) + C a + j3aC = ( l +/ 3a) C (4.13)
where C is the amount of cement and /3 is the maximum amount of chemically bound
water (see also Table 4.5), if the amount of chemically bound water was used for the
9 4
O)
HI
m
CO
200
150
100
50
0
200
Mix 45-1 (H20 )
_ 150
Mix 45-2 (H20 )
O O
o
o
O O
o>
CtJ
^ 100
UJ
CD
CO
50
O
o
o
o
0
o
0.25 0.5 0.75 1
W./0.23
D
200
Mix 4 5 - 3 (H O )
150
O)
1 0 0
CO co
50
0.25 0.5 0.75
W./0.23
200
Mix 35-1 (H O )
150
cn
CM
1 0 0
O O
50
0.25 0.5 0.75
W./0.23
0.25 0.5
w./0.23
D
0.75
200
Mix 4 5 - 4 ( H 0 )
150
Oi
eg
100
o o
CO
50
0.25 0.5 0.75
W./0.23
200
4 5 - 1 (N2 )
35-1 (NJ
150
05
eg
1 0 0
UJ
CO
50
0.25 0.5 0.75
w./0.23
Figure 4.33: Sads for various mixes.
9 5
degree of hydration. The mass of hydration products or gel is given by:
Mg = Ca + ,3aC = ( 1 + 0)aC ( 4 . 1 4 )
The surface area (Sdes) is calculated based on the dry weight of the sample which is
Ms. The gel surface area is calculated using:
The result of applying the above equations is given in Table 4.6 to 4.10. It can
be seen that Sg is essentially the same regardless the age of the specimen. This
supports the fact t hat the surface where water adsorbed on belongs to the gel/product
and not to the other solid phase like the unhydrated cement. Interestingly, the
maximum average value of Sg is given by GGB paste (400 m2 /gr) then is followed
SF paste (386 m2/gr). The value of Sg for FA paste is very close to that of OPC
paste. This indicates the significant contribution of pozzolans on altering the paste
microstructure. It can be expected that many microstructure-related behavior like
shrinkage and creep will also be greatly affected by the presence of pozzolans.
Notice that for OPC, Sg is around 260 m2/ g r which is very close to the value
reported in [Sor.l] who used the same technique to measure Sg. A close agreement
with the values reported in [Dai.l] is also obtained. In [Dai.l], the measurement was
done for C3 S. However since the OPC used belongs to the type I with high content
of C3S and CoS, its hydration products should be similar to the hydration products
of C3 S. The other technique such as the small-angle X-ray scattering (SAXS) by
[Vol.l] essentially gave a confirming result, Sg ~ 220 m2/gr. In their investigation
OPC pastes were 2 -year old and assumed fully hydrated. The other SAXS studies
by [Win.l] reported a value around 600 m2/ g r for Sg hence, did not confirm the
result obtain in [Volk et.al.]. Since the SAXS technique requires some theoretical
(4.15)
96
Table 4.6: Sg for mix 45-1.
Age (hours) a Sg (m2 /gr)
1 2 0.181 262.69
24 0.363 264.42
48 0.479 217.05
72 0.550 229.91
172 0.684 293.92
675 0.811 285.09
4584 0.930 259.77
Average Sff=260 m2/gr
calculations, [Win.l] could have used calculation procedures different from the ones
used in [Vol.l]. The SAXS measurement for D-dried samples in [Win.l] however,
agrees with what obtained in this study.
4.4.4 Pore Characteristics
The first thing to look at here is what is the volume of the gel pore. One advantage
of plotting sorption data the way shown in Figures 4.28 and 4.32 is that one can see
whether the the sorption is accompanied with capillary condensation or not. It is
worth mentioning that only the H2 O desorption data can be used in the pore analysis,
since during the desorption the paste has an open (true) microstructure. As seen,
most data from all mixes follow the multilayer curve (t /t m) despite the scatter caused
by the age of the paste. Now, the monolayer volume is understood as twice of the
monolayer volume obtained from the adsorption branch or:
Vm4es = 2 Vm (4.16)
97
1 2 0.136 291.14
24 0.283 259.96
48 0.374 249.23
73 0.419 233.02
177 0.544 243.58
679 0.702 245.86
4584 0.824 277.74
Average Sg= 257 m2/gr
1 2 0.103 471.96
24 0.197 413.62
48 0.264 381.76
73 0.296 359.51
177 0.351 417.99
679 0.541 386.48
4584 0.662 366.79
Average S9=400 m2/gr
98
1 2 0.168 356.00
24 0.271 435.98
48 0.339 353.47
72 0.387 409.57
384 0.562 365.04
672 0.596 385.63
4992 0.769 392.21
Average Sg=386 m2/gr
1 2 0 . 2 1 1 329.96
24 0.343 284.86
48 0.437 302.21
72 0.519 269.53
172 0.671 241.71
675 0.740 256.36
4584 0 . 8 6 8 240.09
Average Sg= 275 m2/gr
9 9
o45- 1 191d
x 45-2191 d
> 45-3191 d
+ 45-4 208d
* 35-1 204d
3.5
2.5
1.5
0.5
0.2 0.4 0.6 0.8
RH
Figure 4.34: V/ Vm for various mixes of mature paste (older than 190 days).
An important finding that can be obtained is that the capillary condensation is
negligible here. Rather, the gel seems to consist of a relatively uniform pore size.
The complete multilayer filling of the gel pore occurs at /i~ 0.95. At this point,
V/2Vm and t / t m coincides with each other and therefore, the pore size can found to
be 2 molecular thickness. The volume of the gel pore corresponding to this size is
about 2 to 2.5 Vm,des In the case of OPC paste having w/b=0.35 the volume is a
little less than 2 Vm,des- A similar finding can also be obtained from the adsorption
curve at 0.95. In Figure 4.34, the maximum volume of adsorbed water is seen
around 4Vm.
The pore size distribution is not presented here because it is apparent that the
paste microstructure consists of a relatively uniform pore size. The results of No
sorption cannot be used since they do not represent the true microstructure.
CHAPTER V
MIROMECHANISMS OF HUMIDITY-DRIVEN
SHRINKAGE IN PORTLAND CEMENT PASTE
5.1 Introduction
Porous materials such as cement paste and concrete shrink upon removal of water
and swell upon addition of water. The effects of deformations induced by humid
ity, also when combined with temperature, can sometimes be detrimental. Stresses
developed when such deformations are restrained can result in cracking. Examples
of shrinkage related deterioration have been observed in concrete pavements, bridge
girders and slabs, structural walls, and highrise buildings.
One attempt to avoid the destructive effects of humidity on cementitious ma
terials is to understand the mechanisms of humidity-driven shrinkage. A sound
understanding on the shrinkage mechanisms will enable us to also understand the
mechanisms of autogeneous shrinkage. The latter has also been found to be signifi
cant in magnitude and relate to some early age problems in concrete structures. One
important feature of autogeneous shrinkage is that the magnitude strongly increases
as the water-binder ratio of concrete or paste decreases. This implies that the risk
of early age cracking due to autogeneous shrinkage is greater in structures made of
low water-binder ratio concrete.
100
101
Previous investigations [Han.2,Fer.l,Alm.l,Wit.l] reported three major driving
forces of humidity-driven shrinkage: disjoining pressure, capillary stress, and surface
tension. It is believed that each force acts on different parts of the solid. Out of these
three forces, the disjoining pressure is less undertood. Although it has been discov
ered long enough [Der.l], the inclusion of the disjoining pressure for paste shrinkage
is rarely done. Only few studies [Fer.l,Baz.7,Wit.l] were conducted in attempt to
understand how the disjoining pressure operated in cement paste. Meanwhile, con
cepts of capillary stress and surface tension are relatively well understood. What
remains difficult to understand is where and how each of these driving forces act si
multaneously on the paste microstructure. Further complications also come from the
fact the paste microstructure changes with age in the case of autogeneous shrinkage.
The goal of this chapter is to define the mechanisms of humidity-driven shrinkage
based on drying shrinkage experiments, micromechanics, a microstructure model,
and microstructural characteristics obtained in Chapter 4. Once the mechanisms
are well defined, then model predictions and validations are performed for drying
shrinkage and autogeneous shrinkage.
5.2 Microstructure Evolution
Cement paste consists of porous hydration products (gel) occupying the capillary
space. This space is originally field with water and is gradually reduced as hydra
tion continues [Pow.2,Sor.l]. In order to correctly understand shrinkage behavior, a
quantitative model of evolving paste microstructure has to be developed. This model
should provide information on the pore size distribution, surface area, and on how
these quantities evolve with time. This information are necessary to determine the
action of relative humidity at capillary pore level. At this level, capillary strain is
102
believed to be active. The consideration, on the gel-pore level will be made later,
when each shrinkage mechanism is discussed.
5.2.1 Microstructure Model
To describe the evolving microstructure, the pore volume and the pore size dis
tribution must be first defined. The cumulative volume of the pore can be found
using Powerss concept [Pow.2]. The total pore volume in ordinary portland cement
(OPC) paste can be calculated from [Sor.l,Pow.2 ] (cm3 per gr cement):
Vp = uj 0.367c* (5.1)
where ui is the water to cement ratio and a is the degree of hydration. The pore
microstructure can be described in terms of a pore size distribution function (PSD).
In this study, the distribution function, v, is defined as:
e - 4 >/ b
v = (5.2)
For pore volume Vp:
r^Pm.a.r 7T /*0max g fi/b
Vp(a) {uj 0.367a;) I rd(f>= (ui 0.367a) / -----d(f> (5.3)
If cylindrical pore geometry is chosen, then <f>will become the pore diameter and b
will be the pore diameter corresponding to the maxima of PSD in the logarithmic
scale, x is a correction factor such that:
1 /*0max f0
I dv = I dv 1 (5.4)
X tprnin J OO
X = e-<t>min/b _ e-0mx/6
Notice that the parameter b still needs to be determined. This is given in the next
subsection when a separate model is used to determine the area of the capillary pore
wall.
1 0 3
In order to use Eq.(5.3) for different pore geometries, a general or shapeless
measure of a pore must be used. The hydraulic radius, which is defined as:
can be used for this purpose. In Eq.(5.6), V is the pore volume and S is the surface
area. Notice t h at this idea is not causing any changes in total volume or area.
Because the volume and area are basically calculated from the definition of the
hydraulic radius [Mae.l].
To illustrate the accuracy of Eq.(5.3) above, let us consider the cummulative
pore volume obtained experimentally by Winslow and Diamond [Win.2] using the
mercury intrusion method (MIP). The experimental data together with the fits using
Eq.(5.3) are shown in Figures 5.1 and 5.2 for two values of water-cement ratio (w/c).
The data was obtained assuming the pore geometry to be cylindrical. The fits by a
simple equation like Eq.(5.3) appear to be very accurate.
5.2.2 Area o f Capillary Pore Wall
In order to complete the model above, it is necessary to take into account the
evolution of pore geometry with degree of hydration. The purpose is to find the
relation between the parameter b and the degree of hydration. We follow the work
by Bhatia and Perlmutter [Bha.l] who developed an analytical model for calculating
the area of the capillary pore wall in an evolving microstructure.
Initially, cylindrical pore geometry having a length distribution f ( r ) is chosen.
The cylinder can be characterized by a length Le and a radius r. The subscript e
indicates that volumetric changes of the cylinder which are not influenced by the
presence of others. In other words, no interaction between pores takes place. In
[Bha.l], d.Le/ dt was shown to be zero and dr / dt was assumed to be constant. The
1 0 4
OPC I
w/c=0.4
1d o
0.25
2d + +
3d 0.2
7d
28(
0.15
320d.
0.1
0.05
-2
1 0 10
Pore diameter (pm)
Figure 5.1: Fit of pore size distribution data [Win.2], w/c=0.4.
0.4
OPC I
w/c=0.6
0.35
0.3
7d,
28d
0.25
o>
8 0.2
320(
0.15
0.1
0.05
- 2 - 3
10 1
Pore diameter (pm)
1 0
Figure 5.2: Fit of pore size distribution data [Win.2], w/c=0.6.
1 0 5
length, the surface area and the volume of the cylindrical pores are:
Le = [ f ( r ) d r (5.7)
Jo
Se = 2tt f f ( r ) r d r (5.8)
Jo
Ve = 7Tf f (r)r2dr (5.9)
Jo
Then, dLe/ dt = 0 and dr/ dt = k can be used to derive the following equations:
dSP d r 0 0 , ,
S f = 2* d t i / ( r ) r d r
= 27T JjT rdLe + J r(0)j = 2-KkLeQ (5.10)
dVe
= r f ( r) T*dr
r roc roo
= 2rfdL + j 0 r ()
dt
r roo roo
= k Se (5-11)
Solving Eq.(5.10) and (5.11) above gives:
Se = Se o 2irkLe0t (5-12)
Ve = Veo - kSeQt + irk2LcQt 2 = Ko + S\ (5.13)
47rLe0
Se = yjSeQ-I- A-KLeQ{Ve Ve0) (5-14)
Now, the remaining task is to correlate the surface area and the degree of hydration.
This can be done using an approach developed by Avrami [Avr.l] who suggested
that the incremental volume growth of an overlapping system, dF, is proportional
to the incremental growth of a non-overlapping system, dFe by a factor ( 1 F). To
apply this approach, V and Ve must be normalized so that the maximum value is
one. The space available for hydration is similar to Eq.(5.1). The normalized volume
after including the initial condition (o0) becomes:
T> (w - 0.367ao) - 0.367(a - a 0) 1 ^
v = -------------------------- = 1 V (o.lo)
co 0.367q!o
1 0 6
dV = ( I - V)dVe ( 5 . 1 6 )
Solving the above equation results in:
V = 1 e (5.17)
Each term containing Ve and VeQcan now be substituted with Eq.(5.17) to give:
V e - V e o = (u j - 0 . 3 6 7 a 0) ( K - K o )
(a; 0.367oro) 0.367(a <*o)
= (a; 0 . 3 6 7 a o ) I n
uj 0 . 3 6 7 a o
(5.18)
For the overlapping system, the rate of change of pore volume with time is related
to the surface area by an equation similar to Eq.(5.11),
dV = kSdt (5.19)
Then 5 and Se can be related by using Eq.(5.17):
dV dV S_
Se dVe dVr
S = Se(l - V)
After rearranging, Eq.(5.21) can be written in the following form:
~ _ (cj - Q.367a0) - 0.367(a - a 0) a
uj 0.367ao e
^ 47rLe0 (a! 0.367q!o) jn (^ _ 0.367ao) 0.367(a a 0)
5g0 uj 0.367ao
(5.20)
(5.21)
0.5
_ uj - 0.367a
uj 0.367a0 e
ATrLeo(uj 0.367a0) w ~ 0.367a
1 ----------------------------In
0.5
(5.22)
(5.23)
Sf0 * uo 0.367ao
As suggested by Khinast et.al. [Khi.l], Eq.(5.23) can be modified to better fit
experimental data. The modified form they suggested is:
5 =
uj 0.367a "
.uj 0.367ao.
->e0
^ 4irLe0(uj 0.367a0) ^ uj 0.367a
S I
e0
uj 0.367a0
( 5 . 2 4 )
1 0 7
In the above equation, all terms are dimensionless except Seo- An interesting feature
of this model is that Seo can be used to represent the initial surface area of the
unhydrated cement, which in turn, can be correlated to the cement fineness or blaine.
By defining the hydraulic radius, f as:
the surface area can also be calculated from Eq.(5.3) by using:
The parameter b in Eq.(5.3) can be calculated by solving Sp = S numerically.
The calculated values of the parameters b as functions of a are illustrated in
Figure 5.3. As a comparison, the values determined by fitting the experimental
in [Win.2] using Eq.(5.3) are also plotted with a in Figure 5.4. The plots in both
figures look to be very similar in shape. The plots of the surface area calculated using
Eq.(5.24) are given in Figure 5.5. These plots appear to be very similar to the values
calculated using the data in [Win.2], as shown in Figure 5.6. The capillary wall area
initially increases as degree of hydration increases. Then, the area decreases after
reaching a maximum. Initially, there is no hydration products that form the capillary
wall and thus, the capillary wall belong to the unhydrated cement. The surface area
of the unhydrated cement is very small and is not shown, i.e. for a < 0.05. Because
the gel are very fine the pore wall area increases as the gel covers the unhydrated
cement. However, only the external part the gel contributes to the capillary wall.
The internal part that does not intersect with the capillary space does not contribute.
That is why the capillary area is smaller than the gel surface area (see Sg in the next
'mm
(5.26)
1 0 8
0.25
Mod. parameters: S =0.4, L =80, a =0.02
o w/c=0.25
x w/c=0.3
+ w/c=0.4
* w/c=0.5
0.2
0.15
X I
0.1
0.05
0.2 0.4 0.6 0.8
a
Figure 5.3: Parameter b obtained theoretically.
subsection). Eventually, as the pore volume becomes filled with the gel, the capillary
area decreases and goes to zero when all pore volume is occupied. It is clear from
Eq.(5.24) and Figure 5.5 that this feature is satisfied by the present model.
5.3 Humidity-Driven Shrinkage
Let us first attempt to illustrate how drying proceeds in the paste microstructure.
A sketch that depicts the paste microstructure at the capillary pore level is given
in Figure 5.7. As shown, the microstructure consists of capillary pore and the solid
gel/hydration product. The capillary wall is essentially made of the external surface
of the gel. At saturated condition, all capillary pores are still filled with water.
When relative humidity (h) drops the capillary condensation represented by the
Kelvin equation (see Eq.(5.52)) starts occurring. Then, the condensed water in the
1 0 9
0.7
w/c=0.6
0.6
0.5
0.4
w/c=0.4
-0.3
0.2
0.1
0.4 0.5 0.6 0.3 0.7 0.8 0.9 1
a
Figure 5.4: Parameter b calculated from experiments.
35
Mod. parameters: S =0.4, L =80, a =0.02 o w/c=0.25
x w/c=0.3
+ w/c=0.4
* w/c=0.5
-5.25
CO
10
0.4 0.2 0 0.6
a
Figure 5.5: Capillary pore wall area obtained theoretically.
110
35
w/c=0.6
30
CM
E.
Q_
E 25
eo
CO
- -
20
CO
<D
O
CO
tr
w/c=0.4
3 15
co 13
co
Q.
co i n
O lw
0.6 0.7 0.9 0.3 0.4 0.5 0.8
a
Figure 5.6: Capillary pore wall area calculated from experiment.
larger pores disappears leaving only the multilayer adsorbed water on the capillary
pore wall. The thickness of this adsorbed water can be described by [Ben.l] (in m):
r (h) = [0.395 - 0.189 In [ - \n(h)]] xlO- 9 (5.27)
As h drops further, the movement of water by multilayer desorption, also possibly
by other mechanisms, occur at the gel pore level.
According to the process above, there are several possibilities where humidity can
act as driving forces for shrinkage deformation. First, a change in the thickness of
the adsorbed water can cause a change in surface energy of the gel [Ben.l,Kav.l],
which in turn can act as a driving force [Alm.l,Ban.l,Han.2j. Second, the capillary
pressure acts a driving force on the capillary wall [Ben. 1 ,Mae. 1]. Third, the disjoining
pressure may act on some portion of the gel where free adsorption is hindered [Baz.7].
In order to investigate the action of each stress component during shrinkage,
I l l
Gel product
Adsorbed water
Figure 5.7: The schematic of paste microstructure at capillary pore level.
112
consider first the variations of shrinkage strain with h and with the change of surface
energy (Ay). The change in surface energy of the solid in the paste is calculated
using the thickness of adsorbed water given above and is given as [Ben.l] (in N/m):
10~9RT
Ay(fi) = y(/i) 7 0 = ------ ---- [0.584ln(/i) 0.189 ln(/i) ln(ln(/i))] (5.28)
For water, y0=0.07 N/m2 is the liquid surface tension and V^,=1.8xl0~ 5 is the molec
ular volume, while Z?=8.314 J/mol/K is the ideal gas constant and T (in K) is
temperature. The purpose is to find some correlations between the shrinkage strain
and Ay. The strain components driven by the capillary stress and disjoining pressure
can be found by subtracting from the total strain the strain induced by Ay. Figures
5.8 and 5.9 give the shrinkage variations esh with, respectively, h and the Ay. In
each plot in Figure 5.9, a straight line has been drawn to identify a linear relation
between es/l and Ay. This relation is also called the Gibbs-Bangham strain. The
slope of each of the straight line is named A which can be determined using two pairs
of points, esh and Ay, corresponding to h = 0.99 and h 0.11. Beyond h 0.11,
it is argued that the water in the interlayer space starts to disappear19 resulting in
additional deformation.
It is apparent that if the linear strain or es/l(Ay) is the GB-strain (cgb), then it
becomes the most significant contributor to the overall shrinkage. The magnitude of
GB-strain however, depends on Aor,
gb = AAy (5.29)
Therefore, if the relation between Ay and h can be justified and Acan be determined,
then the most significant contributor to the shrinkage strain can predicted.
The change in surface energy requires the change in the thickness of the adsorbed
water. Intuitively, this can only happen if all surface is exposed to changes in the
1 1 3
0.2
0.4
o>
0.6
0.8
Age=240d
o w/c=0.4
+ w/c=0.6
* w/c=0.75
1.2
0.2 0.6
0.4
Relative
0.8
Figure 5.8: Drying shrinkage data for mature pastes [Alm.l].
1.2
$5 0.8
O)
co
- 0.6
x:
co
CO
0)
5 0.4
Age=240d
o w/c=0.4
+ w/c=0.6
* w/c=0.75
0.2
0.05 0.1
Surface energy (N/m)
0.15 0.2 0.25
(N/m) energy
Figure 5.9: Relation between drying shrinkage and A7 .
114
relative humidity. This condition requires all capillary pores not to be filled with
condensed water or no capillary pore exists. The first situation is only possible
below a certain value of h (remember that capillary pore size can be related to
humidity by the Kelvin equation) and the second one is applicable to hydrated low
0.75 a linear relation between esh and A y is still apparent. The paste with w/c=0.75
should have some capillary pores even if it is fully hydrated (see Eq.(5.1)). Thus,
capillary condensation cannot be avoided in fully hydrated paste having w/c=0 .7 5 .
5.3.1 Gibbs-Bangham Stress
First, it needs to be justified that A y varies with h over the entire range of relative
humidity. Based on the illustration on the drying process, one might think that since
only a portion the capillary surface is exposed when the relative humidity drops, then
the change in the surface energy (Ay) only works there. This is not true, because
as has been shown in [Baz.7] that, under the assumption of static equilibrium, the
chemical potential of the adsorbed water on the capillary wall (fia) has to be equal
to the chemical potential of the hindered adsorbed water (fid), i.e.:
Otherwise, transport of adsorbed water driven by the gradient of chemical potential
may take place.
B3/ definition, the surface energy of the solid where adsorption is hindered is
calculated from [Kav.l]:
w/c systems. However, as shown by Figure 5.9, even in the paste with w/c as high as
fid fia ; dfid = dfia = RTd ln(h) (5.30)
(5.31)
A t d = - r d- la(h)
*0
(5.32)
1 1 5
In this equation A y d was calculated assuming the thickness of the hindered water,
to be constant. can be estimated from the finding in Chapter 4 and [Baz.7]
to be about 2 molecular thickness. A value of 3.4 Angstrom has been used as one
molecular thickness in Chapter 4. Using Tj = 6 . 8 Angstrom, A y d is plotted in Figure
5.10 together with the change of solid surface energy ( Ay) defined by Eq.(5.28). It
must be remembered that Eq.(5.32) is valid only up to the hindered adsorption and
capillary condensed water diminish, after that it switches to Eq.(5.28). The value of
h corresponding to end of capillary condensation is estimated using sorption data.
From the analysis given in Chapter 4, this is located at h 0.35.
From Figure 5.10, one can see that A y d and A y do not differ significantly up
to h ^ 0.35. Therefore, Eq.(5.28) can be safely used for the entire range of relative
humidity. Also, this finding implies that the locations where A y d and A y act are
the same, the gel surface area.
The next task is to calculate A. In order to do this the geometry of gel microstruc
ture must be idealized. An intersecting rectangular rods is chosen. This idealization
is very similar to the one used in [Sch.l] but is modified for rectangular rods instead
of cylindrical rods. The reason for choosing such a shape is because the gel phase in
cement paste acts as a continuous solid binder. Other geometrical representations are
also possible, however they may not yield closed form solutions for A. To calculate
strains, consider a unit cell of solid consisting of intersecting rectangular rods shown
in Figure 5.11. The method of calculation follows the one given in [Sch.l] for a unit
cell of intersecting cylindrical rods. The stress components acting in one of the rods
are essentially the same as the components in the other rods but rotated 90. From
the force equilibrium, the stress components due to surface tension are found to be:
2 A y A y
116
0.25
0.2
"E 0.15
0.05
0.2 0.4
Relative
0.6
0.8
Figure 5.10: Variations of Ayd and A y with h.
8Ay 2Ay
cr,z = (5.33)
At t
The strain components can be found from Hooks law [Tim.l] for isotropic materials
and are found to be:
A-v At
(5.34)
A7 dt
fxx fjry ~E~t ~ '
2 A y dr
e,~ = ( 1 vc) =
Est K s) r
(5.35)
where E s and u are the Youngs modulus and Poissons ratio of the solid skeleton,
respectively. The linear strain is defined as:
d(t + r) A y 1 3vs + 2y(l vs)
gb
(5.36)
t + r Est (14- y)
A can be found from the above expression. However, it more convenient to describe
A in terms of the specific surface area and the solid density. To do this, the solid
1 1 7
Figure 5.11: A unit cell used to calculate gb-
hydraulic radius must be obtained,
_ AVS 24rt2 + 8 3 t ( y / 2 + 1/6)
= a s : = 4s<r = ; ( 0 - 3 7 )
The specific surface area is essentially the surface area per unit mass of solid or:
A Ss 1
5 =
A VsPs TsPs
thus using Eq.(5.37) and Eq.(5.38), qb can be written as:
PsS [1 3i/s -f 2y(l vs)\ ( l / 2 + y/ 6) psS
(5.38)
eGB = A7 :
E
Hence, A is found to be:
2 / ( 1 + y)
x PsS .
= ~E~t9
= ~E^g(y, vs) A t
(5.39)
(5.40)
The solid subjected to surface tension consists of unhydrated cement grains and
gel/hydration product. It has been shown in Chapter 4 that the gel surface area,
Sg, is about 260 m2/gr-gel. The surface area per unit mass of combined gel and
unhydrated grain is:
q - q Mi. - q ^ +
9 Ms 9 1 +(3a
( 5 . 4 1 )
1 1 8
In addition, the density must also be calculated for the solid consisting of unhydrated
cement grains and gel, which is found from:
C + (3aC _ 1 4- (3a.
Ps Vs C/ pc + (3aC/pw l / p c -\- (3a/pw
where C is the mass of cement, pc is the density of cement (3.15 gr/cm3), pw
( 1 gr/cm3) is the density of water, and (3 is, as explained in Chapter 4, the maxi
mum chemically bound water. For ordinary portland cement paste, (3 is about 0.23
[Cop.l,Sor.l].
Eg has to be calculated for a composite made of unhydrated cement grains and
gel. To do this a simple parallel-series model as illustrated in Figure 5.12 is chosen.
For such a simple arrangement the composite modulus is given by:
- i
; s = 1 y / l - a (5.43) Es =
1 s s
+
Ec(l s) + EgS Eg
where Ec is the Youngs modulus of unhydrated cement grain and Eg is the Youngs
modulus of the gel/product. The value of vs is approximated as a constant and to
be close to the Poissons ratio of OPC paste {ys 0.2). Finally, Acan be written as:
( 1 +/3)a
A(a) =
1 / Pc + 0 o > / p v
y/1 a 1 y / l a
Ecy/ l a + Eg( 1 \ / l <*) E,
9
9(y, vs)
(5.44)
Variations of Acalculated using Eq.(5.44) are illustrated in Figure 5.13. with material
parameters included in the figure. In this calculation, the value g(y, us) is 2, which
is obtained for y=0.04 and vs =0.2. From Figure 5.13, Adoes not appear to increase
linearly with a and its rate of change becomes much slower as a approaches one. A
supporting experimental data on how Achanges with age is given by [Alm.l] and is
shown in Figure 5.14.
1 1 9
1-s s
Figure 5.12: The composite model used to calculate Eg.
0.06
E /E =
0.05
0.04
z
1 0.03
<<
0.02
0.01
0.2 0.6 0.8 0.4
a
Figure 5.13: Aas a function of degree of hydration.
120
1.2
+w/c=0.6, 7d
* w/c=0.6, 30d
o w/c=0.6, 240d
0 0.05 0.1 0.15
Surface energy (N/m)
0.2 0.25
Figure 5.14: The effect of age on A from experiment [Alm.l].
5.3.2 Disjoining Pressure
According to [Baz.7], the disjoining pressure II, is defined as the pressure acting
normal to the pore wall where the hindered adsorbed water is located. The size of
this pore is about two molecular thickness, so it most likely belong to the gel phase.
The disjoining pressure is exerted by the adsorbed phase in order to maintain an
equilibrium with the free adsorbed phase located on the boundary between the gel
space and the capillary space. Bazant has derived an expression for the disjoining
pressure and he has obtained:
C12/ C 22 has been extimated to be 0.6, while hj and 11/ are the values of h and II
just after the pore of two molecular thickness becomes filled.
The humidity range where Eq.(5.45) is applicable is between h = 1 to hj . In
(5.45)
121
[Baz.7], hf was estimated to be 0.12. Further, Bazant also illustrated the magnitude
of the relative change of the disjoining pressure (11 11/) to be 1720 atm at T = 25C
due to a drop in relative humidity from h = 1 to h = hf = 0.12. This is indeed
a large magnitude and is significant enough to cause deformation, probably within
the order of shrinkage/swelling strain. Results in Chapter 4 however, indicate that
the complete monolayer adsorption in ordinary portland cement paste (OPC) occurs
at h = 0 .2 1 , which also corresponds to one molecular thickness of adsorbed water.
Therefore according the above results, hf must be around 0.21 instead of 0.12.
It is clear that II is analogous to the capillary stress in the hindered layer. An
expression analogous to Eq.(5.45) is also possible for the capillary pore. In fact,
this has been shown in [Baz.7]. It is also clear that because both the disjoining
pressure and capillary stress are very similar in nature, both cannot coexist at the
same time. Both types of stress however, act at different parts of the solid phase in
cement paste. The disjoining pressure is felt in the gel pore where the adsorption is
hindered. Meanwhile, the capillary stress acts on the capillary wall which is made of
the gel.
It has been shown that egb relates linearly with A7 through the slope A which
turns out to be realistic and follows the age dependent trend shown in Figure 5.14.
Therefore, strains due to the disjoining pressure or capillary stress can be found by
subtracting eqb from the shrinkage strain. The results are shown in Figure 5.15. It
appears that II does not follow the trend in Figure 5.15. However, as will be shown
later, the capillary stress exhibits this trend.
122
0.05
0.1
0.2
Age=240d
o w/c=0.4
+ w/c=0.6
* w/c=0.75
0.25
0 0.2 0.4 0.8 0.6 1
Figure 5.15: Variations of eSh gb with h.
5.3.3 Capillary Stress
The capillary stress is defined as the stress exerted by the pressurized liquid
existing in the pores. When a continuous pore system is subjected to some change
in the relative humidity, the liquid pressure drops as the pore water disappears
due to condensation or evaporation. As a result, the stress exerted by the liquid
pressure on the pore wall also drops relative to a certain reference condition. This
liquid pressure drop is found upon relating the Young-Laplace equation to the Kelvin
equation [Mae. 1,Ben.l]:
RT
A Pc = - ln(/i) (5.46)
During drying, the pores that are initially filled with water are emptied gradually
as h is reduced. According to the Kelvin equation this means that during drying,
only some pores of a certain sizes are emptied. Liquid pressure drop is experienced
1 2 3
by the pores filled with condensed water. The pores that become empty no longer
felt the pressure drop as h decreases. At a certain value of h, all pores will become
empty and the capillary stress vanishes.
Clearly the location where the capillary stress acts changes with h. As a con
sequence of Kelvin capillary condensation theory, the area and volume where Apc
acts always become smaller as h drops. Therefore, an effective stress must be used
instead of Eq.(5.46). In this study, this effective stress is calculated based on the
effective area. The reason for this will be justified subsequently.
To find the effective capillary stress and the resulting shrinkage strain, consider
a geometry similar to the used for calculating cqb (intersecting rectangular rods in
Figure 5.11). The only difference here is that now the intersecting solid geometry is
used to describe the paste microstructure at the capillary pore level. This also means
that the capillary wall is made gel whose geometry is intersecting rectangular rods.
The stress components can be easily found from the equilibrium of force in a unit
cell made of intersecting rectangular rods sketched in Figure 5.16. Due to symmetry,
only one continuous rectangular rod needs to be considered. The stress components
are calculated from:
&yy ^Pc
2 A pcR ApcR
2(6 + R) ~ 6 + R
(5.47)
and the strain components are calculated using Hooks law,
(5.48)
( 5 . 4 9 )
25
25.
Figure 5.16: A unit cell used to calculate ec.
The capillary strain is found from:
e, =
d{6 + R) Apc
S R Eca 5 -F R
_ Apc 1 + x uca{ 1 + 2x) + x[x 2vca(l + x)]
E ca . ( 1 H- x ) 2
Apc
Ec
-m(x, U c a ) (5.50)
where Eca is the Youngs modulus of the solid phase of the capillary pore, uca is the
corresponding Poissons ratio, and x = R/5.
The effective capillary stress is derived considering that it acts only on the fraction
of the total pore wall. Hence, the average stress acting on the capillary pore wall is:
Ss
f - ' a s , = * P ' i r
*^p J rm ^n
(5.51)
In the above equation, f f is the largest capillary hydraulic radius that is filled with
water and Ssp is the area of the water filled pores, f f can be calculated using the
Kelvin equation:
7 7 (A) -
To Vo
R T ln(h)
( 5 . 5 2 )
1 2 5
For water, 70=0.07 N/m2 is the liquid surface tension and V^=1.8xl0_o is the molec
ular volume, while fi=8.314 J/mol/K is the ideal gas constant and T (in 0 K) is the
temperature. Sp can be found in two ways: predicted using Eq.(5.24) given in the
beginning or calculated using Eq.(5.26) when the capillary pore size distribution is
given. Using the capillary stress, the capillary strain is rewritten as:
. . . ApJh) Ssp(h)
ec(h) = ------- - m(x, uca) (o.53)
H'ca
Based on the previous analysis, ec strain can be calculated by subtracting from the
total shrinkage strain the Gibbs-Bangham strain (Figure 5.16). As seen, ec initially
increases with h then drops to zero around h = 0.1. A logical explanation of this can
be obtained by looking at Figure 5.17 which shows the variations of Ssp/ S p with h,
calculated using Eq.(5.26) for the data by [Win.2]. As shown in Figure 5.17, Ssp/ S p
decreases as h decreases. By a careful observation, one can see that the product
of Apc(h) (described by Eq.(5.46)) and Ssp/ S p must always lead to the behavior
shown in Figure 5.16. Therefore, the analytical procedure developed above has been
confirmed experimentally.
5.4 Application to Drying Shrinkage
The above formulation will be verified in this section for the case of equilibrium
drying shrinkage in ordinary portland cement pastes (OPC). The equilibrium drying
shrinkage is defined as the final deformation measured at each relative humidity.
Since drying is a slow process, the time required to reach the final deformation at
each relative humidity is quite long. This also means that at the final deformation,
the humidity of the specimen is the same as the humidity it is subjected to and is
uniform within the specimen. The experimental data obtained for OPC pastes in
1 2 6
* 28-day
+ 7-day
o 3-day
w/c=0.4
0.9
0.8
0.7
0.6
co
c0.5
co
0.4
0.3
0.2
0.1
0 0.2 0.4 0.6 0.8 1
Relative humidity
Figure 5.17: Ssp/ S p calculated using data in [Win.2 ].
[Alm.l] and [Fer.l] are used. The shrinkage strain is calculated from:
esh(h) = Gb (h) + ec{h) = A(Es, S, ps, g(y, us) ) Aj ( h) + Ssp^ m(x, uca)
&ca ^p
(5.54)
For the drying shrinkage prediction, it is assumed that hydration stops as soon as the
specimens start to dry. Therefore as shown in the above equation, Ais independent
of a. The values of material parameters used in the prediction are: Ec = 15000 MPa,
Eg = l. oEc, Eca = 20000 MPa, vs = vca = 0.2, pc = 3.15 gr/cm3, pw = 1 gr/cm3,
3 = 0.23. The value for Ec or Eg is estimated based on the Youngs modulus values
of mature OPC paste obtained in [Hel.l], while us = 0.2 is considered typical in
OPC pastes. The geometric factor, g, embedded in Ais chosen to be 2. This value
corresponds to y = 0.04. Meanwhile, the geometric factor, m, embedded in ec is
chosen to be 0.75, which corresponds to x = 0.04.
1 2 7
Comparisons between the predicted results and experiments are shown in Fig
ures 5.18 to 5.23. In calculating ec, the data in [Win.2] for the capillary pore size
distribution are used to obtain variations of Ssc/ S c with h. In general, very good
agreement is found between the predictions and experimental data. This is true even
for both old and young OPC pastes, and even for different water-cement ratios. As
has been repeated in this chapter, the shrinkage include the interlayer deformation
which cannot be predicted here. For drying shrinkage, this interlayer deformation
becomes obvious for h lower than 0.2. That is why shrinkage between h = 0 to
h = 0 . 2 is often underestimated.
Each strain component, the Gibbs-Bangham strain and the capillary strain, is
shown in Figure 5.19. As shown, the expression for the capillary strain gives a
closing shape similar to the shape obtained from the experimental data in Figure
5.15. Figure 5.19 also shows that the Gibbs-Bangham strain dominates over the
entire relative humidity range and is the major contributor to the overall shrinkage
strain.
5.5 Application to Autogeneous Shrinkage
5.5.1 Pore Relative Humidity During Hydration
Unlike drying shrinkage, the relative humidity in the case of autogeneous shrink
age must be calculated from the volume balance between the water adsorbed on the
capillary wall and the change in amount of water in the capillary space due to hy
dration and chemical shrinkage. There are not many studies conducted to develop
models to calculate the drop in the relative humidity due to hydration. One quite
interesting model has been developed by [Koe.l]. In this study, a model that fol
lows the concept used in [Koe.l] is presented. The fundamental difference from the
1 2 8
0.1
0.2
'S'0.3
O)
c 0 . 4
0.5
0.6
Experiment [Alm.1] (w/c=0.4, 3 - d a y ) .
Prediction
0.7
0.2 0.6
0.4
Relative
0.8
Figure 5.18: Prediction of drying shrinkage: w/c=0.4 and age=3 days.
1C* o cr<37 ---------1--------------------------1------------------------- 1----------------------- j
o o o o o o o o O
x /
X x
y _
s '
X ^
X ^
_ X
X ^
X x ^
X
X ^ -
X.
x'
*
/ -
X
1
1
w/c=0.4, 3 - d a y
1
o Capillary strain
I
x Gibbs-Bangham strain
I
i . _.
----- Total strain
j . -i i .
0.1
0.2
oT0.3
cn
CO
i 0.4
. e
CO
I 0.5
0.6
0.7
0.8
0.2 0.4 0.6
0.8
Figure 5.19: Components of shrinkage strain: \v/c=0.4 and age=3 days.
129
0.1
0.2
0 . 4
0.5
0 . 6
0.7 Experiment [Fer.1] (w/c=0.4, 7-day)
Prediction
0.8
0.2 0.4
0.6 0.8
humidity
(h)
0.1
0.2
gp.4
co
0.5
m0.6
0.7
0 . 8
o Experiment [Fer.1] (w/c=0.4, 28-day)
+- Experiment [Aim.1] (w/c=0.4, 30-day)
Prediction
0 0 . 2 0.4 0 . 6 0 . 8 1
Figure 5.21: Prediction of drying shrinkage: w/c=0.4 and age=28-30 days.
1 3 0
0.1
0 . 2
0.3
O)
S 0.4
<50.5
0.7
Experiment [Alm.1] (w/c=0.6, 7 - d a y ) -
- Prediction
0.9
0 . 2 0 . 6
0.4 0 . 8
Relative
0 . 2
0.4
o>
CQ
0 . 6
- 0 . 8
Experiment [Aim.1] (w/c=0.6, 30-day)
Prediction
1 . 2
0 . 2 0 0.4 0 . 6 0 . 8 1
1 3 1
previous model lies in the calculation of autogeneous strain. The stress mechanisms
developed earlier will applied here.
Due to Le Chatelier shrinkage the gel phase produced does not equal the volume of
hydrated cement plus chemically bound water. This behavior is illustrated in Figure
5.24. Le Chatelier shrinkage is approximately 25% of the chemically bound water18.
As a result of hydration and Le Chatelier shrinkage, the pore volume (V^10) decreases
leaving some empty space. When the cement paste is sealed, the creation of such
empty space is followed by reduction in pore relative humidity. Such a phenomenon
is called the autogeneous relative humidity change. To calculate the autogeneous
relative humidity, the largest pore diameter that remains filled with water needs to
be determined. The volume of pores filled with water, V), can be found from the
capillary pore volume, Vc. The latter includes the water adsorbed on the pore wall.
The volume of capillary is defined as the volume of pore (Vp) without Le Chatelier
shrinkage [Sor.l,Koe.l] and is given as (cm3 per gr cement):
Vc = u - 0.367a - 0.25(0.23a) = u - 0.424a (5.55)
It follows from Figure 5.24 that:
min
(5.56)
uj0 0.367a rfr
rm.in
ojq 0.367a
X*>
u> 0.367a
max rT
e~Tl d f - /
tin Jff
(5.57)
Using Eq.(5.1), (5.55), and (5.57) we obtain:
u) 0.367a
exp - p - = 0.0575a 4- Vads(hza)
_6 (a)_
(5.58)
X
Capillary pore Capillary pore
Without chemical
shrinkage
With chemical
shrinkage
Figure 5.24: Schematic illustration of the inclusion of chemical shrinkage.
r / 6 (a) In
0.0575a Vads(h, a)
-Vo7o
RTl n( h)
(5.59)
(a; - 0 . 3 6 7 a ) / *
where the right hand side of Eq.(5.59) is obtained from Kelvins capillary condensa
tion theory [Gre.l,Kav.l].
The adsorbed volume has to be determined assuming a certain pore geometry.
The capillary pores has been idealized as intersecting rectangular pores shown in
Figure 5.25. Notice that this was the same geometry used to idealize the gel mi
crostructure. At the capillary pore level however, the solid phase is constructed from
the gel whose microstructure is idealized also as intersecting rectangular microsolids.
For autogeneous shrinkage, the drop in h is relatively small (down to 70%). Because
of this, the gel space is always filled with water and only the water adsorbed on the
capillary wall changes with h. For the intersecting rectangular pore, Vads can ob
tained from a unit cell whose dimensions are represented by R and 5 (Figure 5.26).
133
The volume, surface area and hydraulic radius of the unit cell are given by:
dVp = 24R2S 4- 8 R 3 (5.60)
dSp = 48ito (5.61)
dVc R R x
f = i s r 2 + i r (562)
where x = R / 5. The adsorbed volume is calculated from:
dVads = 6 J(8 f 2T 4 r 2) 4- 4T2 (10i2 6 r )
6 ^(8 f 2r - 4 r 2) 4- 4T2 (10fl - 6 T)
dVI
24R25 4 - 8 R 3 p
/ F r 2 ^ r 2 r 3 \
= + ( 5 ' 6 3 )
Vp is given in Eq.(5.3)and T is the statistical thickness of the adsorbed water given
in Eq.(5.27). Vads is obtained by integrating Eq.(5.63) and defining 0 = R/ r ,
uj 0.367 rf M/ T 5T2x - 3 T 2 T3x \
= # L ( ? + 6 flf2 + W ^ ) e ' dT
ST2x - 3T2 u - 0.367 /* e~f!b ,
= r 5 P + ------^ ------------- a----- /
60 x 6 r2
r 3x u>0.367 rf Me~f Ib
~ W Jrm ~1*~ r
q ( r - 5 T2 x ~ 3r2 -
PV 6 9b 2d2b2)
CJ - 0.367 / s r 2x - 3r 2 T3:r \ e~f!b
Xb V + 262b2 ) f |rm'r"
r 3x ui 0.367 e~f !b
+ 40 Xb f 2 'rm,rv/ ( ^
5.5.2 P r e d i c t i o n s
The above formulation will be verified in this section for the case of drying shrink
age in portland cement pastes. The experimental data in [Koe.l] will used. The strain
1 3 4
A
8
V
I
2R
Figure 5.25: Idealization of capillary pore geometry.
2R
2R
Figure 5.26: The unit cell of capillary pore.
1 3 5
is calculated from the same equation used in the case of drying shrinkage. However,
the value of A and S sc/ S c have to be updated with a. So Eq.(5.54) now becomes
eSh(h) = eCB(h) + ec(h) = A(Es(a), S(a), p(oc), g(y, i/s))A7 (/i(q))
&pc( h( a) ) Ssp(h(a))
H r u 5T7-1 Uca) (0. 65)
Sp{cx)
To calculate A, g, and m, the same material parameters used in the prediction of
drying shrinkage are used. Ssp and Sp on the other hand, are calculated using the
microstructural model (Eq.(5.24)).
The comparisons between the predicted equilibrium shrinkage and experiments
are shown in Figures 5.27 to 5.32. The autogeneous strain predicted in Figures 5.27
and 5.29 seems to be very accurate despite the fact that there is the initial expansion
that cannot be predicted. As can be seen in Figure 5.28, h changes dramatically with
a. Although the relative humidity change is not verified experimentally here, such a
behavior is to be expected. Because according to the model developed above, as a
increases and the capillary pore volume decreases the maximum capillary pore size
that is remained filled becomes smaller and smaller. The contribution of each type
of strain towards autogeneous shrinkage can be seen in Figure 5.30. eqb appears to
be much more significant than ec.
One important thing observed from this analysis is the fact that the magnitude
of autogeneous shrinkage is not necessarily higher than the magnitude of drying
shrinkage for the same relative humidity range. However, the autogeneous shrinkage
behavior is more sensitive to the water-cement ratio (w/c). Figure 5.32 shows how
the autogeneous shrinkage is affected by w/c. The model developed, in addition to
showing fair agreement with experimental results, also picks up the effect of w/c.
Further, it can be seen from Figure 5.31 that pastes with different w/c undergo
0
800 - Experiment [Koe.1]
(w/c=0.3, blaine=300 cm2/gr)
Prediction
100f ----------- --------
1 0
,3
1 0
.4
time (h)
Figure 5.27: Autogeneous shrinkage as function of time.
different variations of h. At the same value of a, the paste with lower w/c paste
appears to have lower value of h. This is confirmed by the microstructure model
presented in the beginning of this chapter, the model to calculate h versus a, and
the experiments by [Win.2]. As described by Eq.(5.59), h versus a depends on the
microstructure paramater b in Eq.(5.3) and the capillary surface area Sc. Results
shown in Figures 5.3 to 5.6 (Sp and b vs. a), either obtained from experiments or
predicted, indicate that b and Sp are strongly affected by w/c.
The effect of cement fineness, as described by blaine value, on autogeneous shrink
age can also be seen by comparing Figure 5.28 and Figure 5.32. For w/c=0.3, the
paste with higher blaine gives higher autogeneous shrinkage. In the present model
prediction, this effect is taken onto account through parameters Eq.(5.24), (5.39),
and (5.53) .
1 3 7
0.95
g 0.9
3 0.85
0 . 8
0.75
w/c=0.3, blaine=300cm /g
0.7
0.1 0 . 2 0.3 0.4
Degree of hydration (a)
0.5 0.6 0.7
Figure 5.28: Autogeneous relative humidity as function of a.
, , , , ,-------- 0 0 =T-
1 0 0 -
200
300 -
\
\
c O
S 400 - x
500(-
,6001- \
o
< 7001- O \
\
\
800 h x
o Experiment [Koe.1] \
9 0 0 1- (w/c=0.3, blaine=300 cm2/gr) x
Prediction v
1000
0.1 0.2 0.3 0.4 0.5 0.6 0.7
Figure 5.29: Autogeneous shrinkage as function of a.
1 3 8
w/c=0.3, blaine=300cm /g
100
300
CO
Capillary strain'
400
CD
500
600
Gibbs-Bangham straii
i - ------
700
0.2 0.1 0.3 0.4
0.5 0.7 0 . 6
Figure 5.30: Contributions of Gibbs-Bangham strain and capillary strain.
0.95
g 0.9
30.85
0 . 8
0.75
o w/c=0.3, blaine=420cm /g
+ w/c=0.4, blaine=420cm2/g
0.7
0 . 2 0.3 0.5 0 . 6 0 0 . 1 0.4 0 . 8 0.7
Figure 5.31: The effect of w/c on autogeneous relative humidity.
o
3
<
1200
1400
1600
\
Experiment [Koe. 1]
(w/c=0.3, blaine=420 cm2/gr) x
Experiment [Koe.1 ] x
(w/=0.4, blaine=420 cm2/gr) v
Prediction
1800
0.1 0.2 0.3 0.4 0.5
0 . 6 0.7 0.8
Figure 5.32: The effect of w/c on autogeneous shrinkage.
CHAPTER VI
PROPERTIES DEVELOPMENT AND
EFFECTS OF MINERAL ADDITIVES
6.1 Introduction
Temperature is one of the most influencing factors on the hydration process.
Its effects have been observed on so many concrete properties especially mechanical
properties such as compressive strength which is often used as a quality control of
concrete in the field. For the field quality check in the laboratory, concrete is usually
cured isothermal. However, the temperature history experienced by concrete in the
field is rarely isothermal and therefore, does not match the laboratory condition.
The impact is even more important when the risk of early age cracking is concerned.
Early age stresses develop with time and involve time dependent properties like
creep while other mechanical properties are strongly influenced by hydration. Several
investigations have been dedicated to studying creep of concrete [Baz.l,Baz.2,Nev.2,Ill.l].
Yet, only few investigations have been conducted to obtain early age creep proper
ties [Emb.l,Kha.l]. In addition, most of creep compliance data were obtained under
compression. Despite the argument that tells creep of concrete is the same in tension
and compression, there have been several studies reporting the difference between
them [Bro.l,Dav.l,Ill.l]. Analyzing the risk of cracking at early ages favors the use
140
1 4 1
of tensile creep since concrete cracks under tension early ages.
The prediction of the temperature effects on the property development is often
done using the maturity concept together with the activation energy. Unfortunately,
the activation energy is not constant. It varies with the degree of hydration [Abd.l].
A more fundamental quantity that relates unambiguously to mechanical properties of
concrete is degree of hydration. Because all age dependent properties must strongly
tie with the microstructure development, which in turn depends on the degree of
hydration [Min.l,Meh.l,Gra.l].
In the early age cracking problem, restraints to hydration-induced deformations
cause the stress to develop. Sources of hydration-induced deformation which are
temperature change and self-desiccation are known to be affected by the cement type
[Koe.l,Jen.l]. Blended cements are often thought to be beneficial in reducing the
risk of early age cracking. Because some blended systems can reduce the autogeneous
shrinkage [Tan.l,Hor.l]. In order to effectively use blended cement and to optimize
its composition, behavior such as like hydration, properties development, and early
age stress development, must be analyzed and the relations among them must be
understood.
In this chapter, the specific goals are first, to develop a general procedure to
calculate degree of hydration and mechanical properties development under non-
isothermal conditions and to establish relations between properties and hydration,
and second, to study the benefit of using blended cements. This chapter is the first
part of an effort to bridge kinetics and mechanical behavior involving the use of
blended cements. The second part of this effort will be presented in Chapter 7.
1 4 2
6.2 Method of Calculating the Degree of Hydration
6.2.1 Phenomenological Hydration Model
Based on previous work [Bor.l], the relation between degree hydration and time
can be accurately modeled using a three-parameter equation (TPE). This equation
was first used in [Han.5] and is written as:
o = a 0 0 e x p [ - Q 1 (6 .1 )
with the derivative
The rate has a maximum at
j , - 1-H* H I-f-q
da. a f l + a \ * a _ i t 2. / ' l + a ' \ ~
m a x & m a x I J = C a I I (6.3)
at t V a ) tol oo \ a J
In the above equations, a and a <*, are the degree of hydration and the ultimate
degree of hydration, respectively, while r and a are model parameters. From Eq.(6 .1 )
and (6 .2 ), it can be seen that the three-parameter model can be converted into a
differential equation. It can be used to solve the degree of hydration as long as
the relationships between parameters O o o , r , and a can be identified as functions
of factors that influence hydration, such as temperature. Writing Eq.(6.2) as a
differential equation, one obtains:
da a (
d t = 7 V n i r j (6-4)
The accuracy and capability of this model is very high and have been demostrated
and compared with other models in [Bor.l].
6.2.2 Hydration Under Non-isothermal Conditions
The incorporation of temperature effects into degree of hydration using the three-
parameter model (TPM) has been done previously [Han.l]. It has been found that
1 4 3
Qfoo and r strongly depend on temperature (T) whereas the parameter a is nearly
independent of T. In general, these parameters can be written as functions of tem
perature in the form:
Ooo = -4q A] T (6-5)
r = B0e
a = C0(l - Ci T)
(6 .6 )
(6.7)
The rate equation can now be written as:
( t w , . . a o o e m 5 ? 1
= ?m0 v n^H
(6 .8 )
Under non-isothermal condition T becomes T(t ). It is apparent from Eq.(2) and (1 ),
that the differential equation above can only be solved for constant T. However,
the temperature variation can be approximated as constant during a small time
increment. Such an approximation is illustrated in Figure 6.1. As time changes, the
rate also changes and by assuming constant T at each time increment the degree of
hydration can be integrated numerically using, for example, the trapezoidal rule,
dt
rt da
T =
? ( l ) 0xp [ ( l )
A U
(6-9)
(6 .1 0 )
where subscribed T means that the quantity is evaluated at constant temperature.
The classical approach using the Arrhenius equation writes the rate in the form:
^ = 1(a) exp
E( a)
(6 . 1 1 )
RT( t )
in which E{ a ) is the apparent activation energy and R is the universal gas constant.
For general forms of f ( a ) and E( a), Eq.(6.11) is a nonlinear ordinary differential
1 4 4
equation and its solution cannot be found in closed form even for constant tempera
ture. On the other hand, TPM can be used to calculate / ( a ) and E( a). It can also
determine a constant apparent activation energy provided t h at a is constant and a
is normalized by Ooq.
To show the equivalence between the formulation using TPM and the formulation
using the Arrhenius-type approach, consider a data set in [Abd.l] for hydration of
0.4 and 0.6 water to cement ratio (w/c) pastes made of ordinary portland cement
type I (OPC paste). Degree of hydration was determined by means of chemically
bound water and the paste specimens were cured five different isothermal conditions:
3, 13, 23, 33, and 43C. The experimental data is shown in Figure 6.2 and the fitting
parameters are given in Table 6.1. To determine f ( a ) and E( a) , the rate can be
written as:
to ( ) = ht / ( )1" a r ( 612>
Then for each temperature, the rate is calculated at several values of degree of
hydration using Eq.(6.4) and the parameters in Table 6.1. The result is a set plots of
da/ dt vs. 1/T. Next, the result is fitted linearly. The slope of these linear equations
is E(a) and the intercept at 1/T = 0 is f(oc). The plots of / ( a ) and E(a) with a are
shown in Figure 6.3. The result looks very similar to that reported in the literature
[Abd.l,Kje.l] and thus, confirms the equivalence between TPM formulation and the
Arrhenius formulation.
6.2.3 Verifications of the Method
To verify the above procedure, semi-adiabatic heat obtained for concrete is used.
It is necessary to first determine parameters Ai, Bi, and Ct in Eq.(6.5), (6 .6 ), and
(6.7) for the mix to be verified. The parameters Ai , 2?,-, and Ci can be found with
1 4 5
da/dt
A
A
Figure 6.1: Schematic of the inclusion of temperature effects.
1 4 6
1
0.9
0.8
Is 0.7
-J= 0 . 6
Co
a
.c 0.5
"o
CD
<0
0.4
o>
at
Q 0.3
0 . 2
0.1
0
1 0 u
1
0.9
0 . 8
S 07
I O '6
O
j= 0.5
CD
CD
0.4
CD
CD
Q 0.3
0.2
0.1
a
*
S
X
*
O
+
*
t
X
S
+
3C
*
x o o
13C
X
23C
X
O +
*
33C
O
+
43C
w/c=0.40
1 0 1 0
Time (hours)
1 0 1 0
*
x
*
X
O
+
9
X
O
+
9
x
O
+
*
x
O
*
+
3C
x
o
13C
X
23C
O +
*
33C
43C
w/c=0.60
1 0 1 0 1 0
Time (hours)
10 1 0
Figure 6.2: Degree of hydration data for portland cement paste [Abd.l]
1 4 7
Table 6 .1 : Optimized model parameters for Abdel-Jawad data [Abd.l]
w/c=0.4
T (C) Qt oo T a
3 1.064 57.564 0.526
13 0.937 28.007 0.544
23 0.912 13.635 0.566
33 0.860 6.75 0.593
43 0.852 4.2 0.6075
w/c=0 . 6
T (C)
oo
r a
3 1.04 56.22 0.544
13 1.018 31.38 0.498
23 0.953 16.39 0.549
33 0.910 9.84 0.647
43 0.875 5.16 0.677
1 4 8
16
o w/b=0.4
1 4 |- O O 0 + w/b=0.6
O
o + + +
1 2 ? + 9 j.
+ O
+
10 h 0 4-
t n +
|L 8
6
4
2
4-
O
) 0.2 0.4 0.6 0.8
a
50
o w/b=0.4
4 5 ! ' ^ o O O O O o 0 + w/b=0-6
4 + + + - + * 9 * s +
3 5 L + +
4-
_ 3 0
o
| 25
iu go
15
1 0
5
0.2 0.4 0.6 0.8
a
Figure 6.3: / ( a ) and E ( a ) obtained by TPM.
149
Table 6.2: Parameters Aj, Bj , and C, generated from the isothermal heat data.
Mix Ao
^ 1 B0 Co
45-1 389.44 0.997 930.41 1.336 0.574 0.013
45-2 357.26 0.421 1223.1 1.349 0.539 0.008
45-3 457.39 2.207 2652 1.536 0.320 0.014
45-4 389.28 1.253 1196.7 1.362 0.499 0.007
35-1 313.09 0.299 564.13 1.254 0.750 0.009
35-2 324.16 1.033 898.03 1.313 0.616 0.008
35-3 366.51 1.214 1392.3 1.424 0.411 0.015
35-4 294.75 -0.156 525.66 1.199 0.712 0 . 0 0 2
the help of the isothermal heat data obtained for paste. Although semi-adiabatic
heat is measured on concrete but the temperature dependent behavior of concrete
hydration should be the same as the paste. TPE parameters (Qoo, t , and a) for
the isothermal heat data are listed in Tables 4.3 and 4.4. The parameters Ai, Bi,
and C'i found for the isothermal heat are given in Table 6.2. The values of Ai, Bi,
and Ci given in Table 6.2 are then optimized to fit the semi-adiabatic heat data
shown in Figure 6.5. The temperature history corresponding to the semi-adiabatic
heat development (Ti and T2 ) are given in Figure 6.4. Only semi-adiabatic heat
corresponding to Ti are used for the optimization. The heat data corresponding to
To are used to verify the accuracy of the procedure above. The prediction result
for semi-adiabatic heat of hydration due to T2 is shown in Figure 6 . 6 for mixes 45-3
and 35-1. The agreement between the prediction and measurement is seen to be
very good. Overall, the cumulative error expressed as \Qpred,i ~ Qmeas,i\/Qmeas,i
never exceeds 3.5%.
1 5 0
60
Mix 4 5 - 3 T(t)
55
Mix 4 5 - 3 T (t)
50
45
40
\
Mix 35-1 T.(t) v
30
Mix 35-1 T,(t)
25
2 0
100
Time (hours)
200 250 150
Figure 6.4: Temperature history for semi-adiabatic heat development.
350
Mix 4 5 - 3 T (t)
300
250
Mix 35-1 T (t)
0150
1 0 0
calculated
measured
50 150 250 300 1 0 0 200 0
Time (hours)
Figure 6.5: Semi-adiabatic heat development due to 7\.
1 5 1
350
Mix 4 5 - 3 T (t)
300
250
Mix 35-1 T (t)
3 2 0 0
100
calculated
measured
50
50 100
Time (hours)
150 200 250
Figure 6 .6 : Semi-adiabatic heat development due to T2 .
6.3 Strength and Youngs Modulus
It is very important to first show that unique relations between E, / ' , and f sp,
and degree of hydration exist. Because ideally, we would like to determine how the
properties develop under general curing conditions such as non-isothermal curing
condition. Since effects of temperature are observed on both degree of hydration and
the mechanical properties, we may suspect that temperature affect the mechanical
properties through degree of hydration. To prove this, let us examine the compressive
strength variations shown in Figure 6.7. The data is given [Abd.l] for both / ' and a.
In [Abd.l], a was determined from measuring the amount of chemically bound water
(wb). As seen in Figure 6.7, the effect of temperature on / ' is clearly significant,
while the effect of temperature on a for the same system can be seen from Figure
6.2. The relation between f'c and a however, appears to be unique regardless the
1 5 2
temperature conditions. Therefore for a given mix, it is only necessary to establish
the relation between / ' and a at one temperature condition. The relations between
other properties such as E and f sp and a can found using hydration data presented
previously. Figures 6 . 8 to 6.11 shows the hydration curves and how E, / ' , and f sp
obtained in this study vary with a. There appear to be some unique trends between
E, / ' , and f sp and a .
As discussed in Chapter 4, heat of hydration is indeed a measure of degree of
hydration and thus, relate uniquely to E , and f sp. Variations of E, / ' , and f sp with
isothermal heat of hydration are given in Figures 6.12 to 6.17. The heat data have
been given in Chapter 4 and have been measured up to about 21 days. Meanwhile,
E , / ' , and f sp are measured up to about 100 days. For the purpose of correlating
heat and the properties, values of heat beyond 2 1 days are obtained by extrapolating
heat data using the three-parameter equation (TPE). The TPE parameters used for
the extrapolation are listed in Table 4.3 and 4.4. These parameters, as presented
later, will also become useful in calculations of early age stress.
As also discussed in Chapter 4, each mix may not have the same ultimate heat
of hydration. Therefore in order to compare the properties of different mixes, the
isothermal heat has been normalized by the corresponding ultimate value. Strictly
speaking, this normalized heat of hydration cannot be interpreted as degree of hy
dration rather, as relative hydration. By observing Figures 6 . 1 2 to 6.17, it is clear
that E , / ' , and f sp are generally higher in mixes having w/b=0.35 (mixes 35) which
is to be expected. For the mixes having the same w/b, variations of E , / ' , and f sp
with Q/Qoo appear to be very similar. The effect of adding pozzolans is not very
clear from Figures 6.12 to 6.17. However, it can still be sen that fly ash (FA) concrete
shows slightly lower values of E, / ' , and f sp compared to the other mixes, expecially
1 5 3
50
OPC concrete w/c=0.4
45
30
20
o 3C
+ 13C
x 23C
* 33C
> 43C
1 0
time (hours)
OPC concrete w/c=0.4
45
35
20
o 3C
+ 13C
x 23C
* 33C
> 43C
0.6 0.2 0.4 0.8
a
Figure 6.7: Compressive strength development with at 5 isothermal conditions
[ Abdel-Jawad].
1 5 4
Mix 45-1
Mix 35-1
0.9
0.8
0.5
0.4
0 3
<D
O 0.3
0.2
0.1
1 0
Time (hours)
10
Figure 6 .8 : Degree of hydration (a) determined by means of chemically bound water.
45
Mix 45-1
Mix 35-1
40
35
30
l u 2 0
0
0.4 0.2 0.6 0.8
a
Figure 6.9: Youngs modulus development with a.
1 5 5
70
60
50
Q_
5
-30
20
10
o Mix 45-1 +
+ Mix 35-1
*
+ 8
+
' 4 0 l - +
* 8
+
* o
8
0.2 0.4 0.6 0.8
a
Figure 6.10: Compressive strength development with a.
+ O
o Mix 45-1
+ Mix 35-1
4.5
040
40
3.5
O
O
0.5
0&
0 0.4 0.2 0.6 0.8 1
a
Figure 6 .1 1 : Split tensile strength development with a.
1 5 6
35
o 4 5 - 1
x 4 5 - 2
+ 4 5 - 3
* 4 5 - 4
25
fit 45-1
w 15
0.2 0.4 0.8 0.6
Q/Q
oo
Figure 6.12: Relations between Youngs modulus and heat of hydration.
for low Q/Qoo-
Each property seems to vary differently with Q/Qoo- The variation of E with
Q/Qoo follows a power type function with decreasing rate (exponent less than 1 ). / '
and / spappear to follow power type functions with increasing rates (exponents larger
than 1 ). Similar behavior can also be observed from Figures 6.12 to 6.17 which shows
properties variations with a. The dependencies of E , / ' , and f sp and a suggest the
following form:
L{Q) = kiQk- (6.13)
where L can be / ' or E or f sp.
1 5 7
60
50
40
2 30
20
10
0
0 0.2 0.4 0.6 0.8 1
Q/Q
oo
Figure 6.13: Relations between compressive strength and heat of hydration.
5
4.5
4
3.5
_ 3
CO
Q.
2 2 . 5
Q.
</)
~ 2
1.5
1
0.5
0
0 0.2 0.4 0.6 0.8 1
Q / Q
Figure 6.14: Relations between split tensile strength and heat of hydration.
o 45-1
x 45-2
+ 45-3
* 45-4
fit 45-1
o45-1
x 45-2
+ 45-3
* 45-4
# C
fit 45-1
1 5 8
45
o 3 5 - 1
x 35 - 2
+ 35- 3
* 3 5 - 4
40
35
o' 25
o.
fit 35-1
15
10
0.2 0.6 0.4 0.8
Q/Q
oo
Figure 6.15: Relations between Youngs modulus and heat of hydration.
70
o 3 5 - 1
x 3 5 - 2
+ 3 5- 3
* 3 5 - 4
60
50
40
30
+ x
fit 35-1
20
10
0
0.6 0 0.2 0.8 0.4 1
Q/Q
oo
Figure 6.16: Relations between compressive strength and heat of hydration.
1 5 9
6
o 3 5 - 1
x 3 5 - 2
+ 3 5 - 3
* 3 5 - 4
5
4
3
fit 35-1
2
1
0
0.2 0.8 0.6 0.4
Q/Q
oo
Figure 6.17: Relations between split tensile strength and heat of hydration.
6.4 Concept of Hydration Time
The maturity concept was originally used to incorporate the effect of temperature
on the development of mechanical properties [Car.2] and on heat of hydration [Ras.l].
A maturity relation is usually derived from the rate equation, i.e. Eq.(6.4):
^ = / M s C T ) (6.14)
under the assumption that term containing a and that containing T can be separated.
In the above equation, a can be used to describe mechanical properties such as
compressive strength, tensile strength or young's modulus. To calculate maturity, a
reference rate is needed. The rate at any temperature can calculated relative to the
reference rate, usually determined at a fixed temperature. Rewriting Eq.(6.4) in this
1 6 0
form one obtains:
a T
/ ( a ) | r 9(T)
( 6 . 1 5 )
&\to f(a.)\To g(To)
Assuming that a and T terms separable, the first term on the right hand side of
Eq.(6.15) cancels out because f { a ) is independent of T. Then, maturity (M) can be
calculated as:
An equivalent quantity called the equivalent time (t e) is related to maturity by:
The only criticism addressed to the above concept is the fact that a and T cannot be
separated according to the analysis given previously. In what follows, it will shown
that the above maturity concept can be improved by using TPE.
It has been shown in the previous section that concrete mechanical properties
are uniquely tied to a. This is not surprising since there have been many studies
t hat also report the strong correlations between microstructural characteristics of
Portland cement paste and degree of hydration [Win.l,Win.2,Dia.l,Abd.l]. Since
mechanical properties should be dictated by the microstructure, their unique relation
with degree of hydration are to be expected. This argument basically suggests that
no other quantity except degree of hydration can uniquely describe the development
of properties with time. In this sense, M and t e should be described in terms of a
which in turn needs to be calculated.
The experiment needed to determine a may not be as easy as obtaining the com
pressive strength or split tensile strength. However, M and t e can still be calculated
from the development of mechanical properties with time. Since E, / ' . and f sp vs. a
3(T(t))
(6 . 1 6 )
( 6 . 1 7 )
'o 9(T(t))
(6.18)
161
are unique, factors affecting degree of hydration such as temperature will also affect
mechanical properties. For example, one can obtain the compressive strength devel
opment from specimens cured at several isothermal temperatures. The data can be
fitted using TPE and the model parameters can be found. For the isothermal case,
/ ' can be written as:
It is easy to see from Eq.(6.19) that a normalized expression exists between f ' J f ' c =
a and t / f = t and is unique for each set of compressive strength data as long as the
parameter a does not vary significantly with T,
This suggests an expression for a generalized hydration time or aging time t a by
inverting the above expression:
where F is an arbitrary constant. In the above equation, a0 is essentially equal to a.
However, it can be assumed constant or can be estimated from the average value of
a. Notice that unlike the traditional equivalent time, ta is not expressed in terms of
any reference temperature. Also, the purpose of generalizing / ' is now apparent, to
include f coo in t a. Therefore, now t a contains all temperature dependent parameters.
(6.19)
a = exp ( t ) (6.20)
- 1 [a
(6.21)
We then generalize the expression for / ' as:
(6.22)
and substitute this equation into Eq.(6.21),
(6.23)
1 6 2
The general expression of ta for the non-isothermal condition can be found by using
a similar argument used to calculate the non-isothermal degree of hydration,
Obviously, the calculation of M is no longer needed once t e is obtained- In fact,
using a similar procedure for calculating degree of hydration one can now calculate
the development of properties with time for the non-isothermal condition,
The parameters necessary for these calculations can be obtained by determining the
property development (E ( t ), and / sp(t)) at several isothermals.
Examples of applying the proposed method can be seen in Figures 6.18 to 6.22.
The experimental data were obtained from literature. For the compressive strength,
the experimental data by [Abd.l] and [Kje.2] were used. TPM parameters corre
sponding to each data set are given in Tables 6.3 and 6.4. The results shown suggest
that the proposed method can successfully coalesce the compressive strength curves
at different isothermal curing temperatures into one curve. This is essentially the
purpose of applying the maturity concept. Unlike the previously developed relations
[Car. 1, Car. 2], no normalization with respect to the ultimate values is necessary. As
a comparison, let us apply the Arrhenius type maturity or equivalent time relation.
dta
(6.24)
(6.25)
1 6 3
The activation energy was first obtained using the method given in [Han.4]. Then
t e was calculated using Eq.(6.18). As can be seen in Figure 6.23, the Arrhenius
type relation does not work well beyond t e 72 hours. This is the reason why the
activation energy needs to be dependent on a , i.e. see Figure 6.3.
Using TPM parameters the temperature effect can be generalized in the form of
Eq.(6.5) to (6.7). Essentially, similar temperature dependencies of o ^ , r , a are also
observed in f Ci00, r , and a.
The present method has also been applied to the fracture energy of concrete
(Gf ( t )). Three sets of data were obtained by [Yu.l]. Each data set corresponded
to a different beam size. The result is illustrated in Figures 6.20 to 6.22 and TPE
parameters are given in Table 6.5. As seen, the method is capable of coalescing each
set of Gf curves. This finding is especially important when the risk of cracking of
concrete members at early ages is to investigated. Cracking at early ages can occur
when the tensile stress developed reaches the strength yet it is Gj that controls the
extend of crack propagation.
6.5 Autogeneous Deformations
Autogeneous strains of different concrete mixes as functions of time are illus
trated in Figure 6.24. The magnitudes of strain are seen higher for mixes with lower
w/b. This behavior also confirms the analytical model developed in Chapter 5. The
deformation history observed in Figure 6.24 indicates that during the early period,
some concrete mixes undergo expansions. It is rather difficult to explain such a
behavior. Possibly, such expansions are caused by early formation of expansive hy
dration products. As reported in the literature [Law.l] formation ettringite results
in expansion of paste. Ettringite is also known to form during the early period of
1 6 4
Table 6.3: Optimized model parameters for compressive strength data [Abd.l].
\v/c=0.4 Ea=42 kJ/moi i r =45 a0 =0.5
T (C)
J c, OO
T a
3 50.11 114.63 0.673
13 53.55 65.5 0.586
23 52.07 32.88 0.515
33 42.46 14.11 0.617
43 46.53 9.1 0.437
w/c=0 . 6 Ea=49 kJ/mol
I
I
C
O
oao=0.5
T (C)
J c,OO
T a
3 46.2 248.09 0.557
13 34.12 73.41 0.719
23 37.22 51.55 0.648
33 33.99 28.46 0.554
43 28.93 14.61 0 . 6 6
1 6 5
Table 6.4: Optimized model parameters for compressive strength data [Kje.2].
w/c=0.43 Ea= 39 kJ/mol F=6o a0 =0.5
T (C)
fc, CO
T a
5 74.7 145.93 0.822
1 2 70.31 90.27 0.771
23 67.61 51.77 0.756
32 58.83 30.35 0.813
43 57.69 17.96 0.752
w/c=0.5 .E'a=44 kJ/mol F=6o ao=0.5
T (C)
f
J c,oo
T a
5 62.79 84.91 0.822
12.5 65.44 51.09 0.727
2 0 58.82 28.46 0.785
35 52.26 11.3 0.875
50 48.2 6.04 0.721
1 6 6
Table 6.5: Optimized model parameters for fracture energy data [Yu.l]
Beam A
T (C)
Gf,ao
T a
Fa=51 kJ/mol 14 1 0 0 . 8 29.46 1.348
jF = 1 0 2 23 102.44 13.18 0.945
ao=0.5 35 96.99 7.76 1.496
Beam B
T (C)
Gf,oo
T a
^=48 kJ/mol 14 95.99 27.11 1.434
=104 23 104.21 13.65 0.813
a0 =0.5 35 96.04 6.46 1.292
Beam C
T (C)
Gf,oo
r a
Ea= 58 kJ/mol 14 99.38 28.04 1.152
F = 1 0 0 23 97.13 12.76 1.044
ao=0.535 93.91 6.29 1.343
1 6 7
(0
Q_
10
45
40
35
30
25
5
~20
15
10
5
0
1 0
>---- - ............. ...... .
o 3C
- + 13C
x 23C
* 33C
> 43C
-
>
......... 'M .............. . .

. X
& *
*
X
X
*
X
O
' +
OPC w/c=0.6
O-1 10 10 t 10z
a
103 10
o 3C
+ 13C
x 23C
* 33C
- > 43C
*
* x >
>
*
x +
X
+
>
>
X
_
+o >
* .
X
o
-
-P
. . . .. * - - *......... 1------ * . . .
OPC w/c=0.4
-1
1 0 1 0 ' 1 0 1 0
Figure 6.18: Relation between compressive strength and aging time, data from
[Abd.l],
1 6 8
70
60
50
i 40
Q_
5
*~30
2 0
10
10 " 1 1 0 1 0 1 1 0 2 1 0 3
t.
70
60
50
nT 40
Q.
5
- 3 0
2 0
10
10"2 10 102 104
Figure 6.19: Relation between compressive strength and aging time, data from
[Kje.2],
o 5C
---------------- T---
O
. + 12C
(h
x 23C
X
n x
* 32C
> 43C
-
p >
&
F
-
... . i
OPC Mortar w/c=0.43
- -
..................... . 1
o 5C
. +12.5C
+
x 20C
x O
* 35C
* *
_ >
- > 50C
o
X
O
* >
>
+
*
Oo
o
- ........1
OPC Mortar w/c=0.5
- - .............. - -
1 6 9
110
100
90
80
70
E 60
~3
CT 50
40
30
20
1 0
io_1 io t 1 0 1 1 0 2 1 0 3
a
Figure 6.20: Relation between fracture energy and aging time, data from [Yu.l].
110
1 0 0
90
80
70
E 60
~2
CT 50
40
30
2 0
10
1 0 ' 1 io t 1 0 1 1 0 2 1 0 3
a
o
........................ .... - .................................. I--------
14C
X +
+ 23C
o
X 35C
x
O x +
X T
-
-
o +
-
-
X
-
-
+
-
-
<5
-
Beam B
_ ___ . . . -----------1------1.... ................ _____
OPC w/c=0.5
i * ................ ______ ___ i__._
......... ....................... .......... - .... ...........
. o 14C
................. 1 ........................
O + .
+ 23C
x +
- x 35C
x +
O
9-
X ~
x +
-
O
P
.
Beam A
............ ............................
OPC w/c=0.5
1 7 0
100
90
80
70
^ 60
CM
e
3 50
CD~
40
30
2 0
1 0
?o_1 10 101 102 103
hydration [Ben.3]. From Figure 6.24, it is seen that expansion is more pronounced in
the mixes having w/b=0.45. It is very likely due to the higher capillary space that
is provided by the mixes having w/b=0.45.
The expansive behavior is not necessarily beneficial in the context of early age
stress development. Because the stress generated from restrained autogneous defor
mation depends on the viscoelastic properties and hydration. It is very possible that
the stress starts to develop after the expansion has ceased and thus, leaving only the
contribution of shrinkage deformation.
Relations between autogeneous strains and concrete hydration, as represented by
Q, are illustrated in Figure 6.25. In showing such relations, the heat of hydration
of paste has been assumed to be the same as the heat of hydration of concrete.
Although in Chapter 2 the literature reported that the heat in concrete might be
.......... . ................. .......................
o 14C
( * +
+ 23C ~
x 35C
P
+
-
o
-
X +
-
-
o
-
Beam C
-
OPC w/c=0.5
......... - ........................... - - -
1 7 1
70
60
50
40
CL
5
30
2 0
1 0
o 1 0 1 1 0 2 1 0 3 1 0 4
t .
70
60
50
40
Q.
2
30
2 0
10
10 101 102 103 104
Figure 6.23: Examples of the Arrhenius type equivalent time, data from [Kje.l].
o T=5C
* ................... ............. 1 --- . ..
O
o +
- + T=12C X
x T=23C
o x .
Jk. * ^
- *T=32PC
* %
x
> T=43C
*
9- >
I
-
-
-
-
-
-
-
-
A
*
A
X
X
'
-
-
-
p
-
-
-
-
A
OPC mortar w/c=0.43
-----.--- . . >>>- i---------.---- ... ........... ............. ............. .
" ................................1
o T=5C
' 1I .................... 1 ................
+
o X
- + T=12.5?C
x T=20C
+ * *
- * T=35C
s.
> T=50C O >
X
-
O
+ * >
-
>
+
*
>
> o
o
>
+
X *
. . . .
OPC mortar w/c=0.5
1 7 2
different from, the heat in paste yet, the time dependent variation of heat should be
relatively the same. In general, a similar trend observed in Chapter 5 for paste is
also observed here for concrete. This again, confirms the validity of the autogeneous
shrinkage model developed earlier. For the stress calculation, relations in Figure
6.25 are considered unique just like the relations between the strength or Youngs
modulus and heat. Also for the stress calculations, polynomials of order three or
higher can be used to describe autogeneous strains as functions of heat of hydration.
6.6 Coefficient of Thermal Dilation (CTD)
Values of CTD obtained for different mixes are listed in Table 6 .6 . Each value
is obtained from two identical specimens. There seems to be no clear trend found
for CTD values measured at various ages, although in general CTD tends to be
higher in younger specimens. A somewhat similar finding was reported by [Bjo.l]
who also showed that CTD changed dramatically between the age of 6 to 20 hours.
Bjontengaards results also indicated that CTD did vary significantly after 24 hours.
From Table 6 .6 , the effect of w/c is also found to be minor. Also, the average
CTD does not vary significantly from one mix to another. Based on the results
presented CTD can be approximated to be age independent and only the average
value is used for each mix.
6.7 Viscoelastic Properties
6.7.1 Creep Compliance and Relaxation Modulus
The creep compliance of concrete is often written in the form [Nev.2]:
= + (6.29)
A
u
t
o
g
e
n
e
o
u
s

s
t
r
a
i
n

(
m
i
c
r
o
n
)

A
u
t
o
g
e
n
e
o
u
s

s
t
r
a
i
n

(
m
i
c
r
o
n
)
1 7 3
80
60
40 45-2
45-1
20
45-3
-20
-40
45-4
60
160 140 180 1 2 0 100 80 40 20
t ( h)
40
20 35-3
35-2 35-1
35-4
-20
-40
-60
-80
-100
-120
180 140 160 1 2 0 80 100 40 60 20
t ( h)
Figure 6.24: Autogeneous deformation with time.
1 7 4
40
35-3
2 0
35-1
3 -20
C
<0
-40
CO
3
o
g -60
O)
o
-80
35-2
-100
35-4
-120
-140
1 0 0 150
Q (J/gr-binder)
200 250 50 300
Figure 6.25: Autogeneous deformation with heat.
Table 6 .6 : CTD values for different mixes (10~6 /C).
Age (h) 45-1 45-2 45-3 45-4
24 10.07 8.41 12.51 1 0 . 8 6
72 12.60 11.14 9.14 7.92
168 9.11 7.72 9.56 8.56
Average 10.59 9.09 10.40 9.11
Age (h) 35-1 35-2 35-3 35-4
24 1 1 . 1 1 12.09 10.06 9.21
72 9.49 9.16 9.17 9.87
168 8.63 9.13 8.85 8.23
Average 9.74 1 0 . 1 2 9.36 9.10
1 7 5
where <(,') = E(t, t' )C(t, t') is called the specific creep and C(t, t') is called the
creep coefficient. Creep data plotted as creep coefficient for various mixes are plotted
in Figures 6.26 and 6.27. The particular aging creep compliance chosen in this study
is written as:
and is called the log-power creep law by [Baz.l,Baz.8 ]. The log-power creep law was
shown to be very simple yet accurate. As opposed to the double-power law, the log-
power creep law produces non-negative relaxation. Results of fitting are indicated
as solid lines in Figures 6.26 and 6.27. As shown, the creep function chosen is very
accurate. For each mix, parameters qi and n can be fixed while the parameter q2
varies with the age.
In order to obtain the relaxation modulus, the age-dependent parameter q2 must
be interpolated by a smooth function. In this work, the following power equation
was chosen:
The relaxation modulus was obtained after solving the volterra integral equation:
Results obtained numerically together the results of curve fitting using Eq.(6.31) are
shown in Figures 6.28 and 6.29 (solid lines indicate the curve fit). In the above
equation, it suffices to set m2 as the aging parameter since m x can be determined
J (t , tr) --- + C ( t , i ' ) = g - j 4- qi In [14- q2(')(t - t')n] (6.30)
q2(t') = Cit' C2 (6.31)
(6.32)
numerically using the trapezoidal integration scheme [Baz.l]. The calculated relax
ation data had been fitted by a hyperbolic function:
1 7 6
from the Young's modulus data. The variation of m2 with time can also be accurately
fitted by a power equation:
The other option is to fit the relaxation data with an exponential type function
rate as the hyperbolic function above, more age-dependent parameters are needed.
Another reason for avoiding such an exponential type function will be explained sub
sequently, when the thermodynamic argument is used to check the validity of the
chosen relaxation functions.
6.7.2 Dirichlet Series
Another alternative expression for creep compliance is to use the Dirichlet or
Prony series. The Dirichlet series are often use for the calculation of stress/strain
in general viscoelastic bodies. It allows for solving the viscoelastic problem in the
form of differential equations. When the creep compliance is expressed as Eq.(6.28),
the stress/strain history must be integrated incrementally and the aging coefficients
must be updated in each increment. Such a method requires the storage of all stress
history from the beginning of the application of load. Reviews on the use of the
Dirichlet series for concrete can be found in [Baz.l].
According to [Baz.l], the creep compliance can be represented by:
i 2 ( 0 = dit'dz (6.34)
similar to the one used for the heat of hydration. However, for it to be as accu-
(6.35)
and the relaxation modulus can be represented by:
k
W O = E ^ ( O e (t- ,)/
(6.36)
n=i
1 7 7
x 10"
t=20
Mix 45-1
t=46
t=101 11
o 2
t=191 h
x 10"
Mix 4 5 - 2 t=26
t=168
t=336
t - f (h)
x 10"
t=24 h
Mix 4 5 - 3
t=72 h
CO
CL
2
f =168 l
o
xx*
1 0 1 0 '
t - f (h)
x 10"
t=24 h
Mix 4 5 - 4
t=72 h
t=1 ^ Bh
f =336 11
1 0 '
t - f (h)
Figure 6.26: Creep data and curve fits, w/b=0.45.
C
(
t
,
f
)

(
1
/
M
P
a
)

C
(
t
,
t
'
)

(
1
/
M
P
a
)
1 7 8
x 10
x 10
1.2
Mix 35-1
0.8
0.6
t=72 h
0.4
0.2
1 0
t - f (h)
x 1 0 "
t=24 h
Mix 35-3
0.8
0.6
0.4 t=72 h
t'=168
0.2
'=360
1 0
Mix 35-2 f=24 h
0.8
2 0.6
f =72 h
:-0.4
t=168
0.2
=336
1 0
t - f (h)
x 10
- 3
1.5
Mix 35-4
t=24 h
o_
t=72 h
0.5
t=336
1 0
t - f (h)
Figure 6.27: Creep data and curve fits, w/b=0.35.
R
(
t
,
f
)

(
M
P
a
)

R
(
t
,
t
)
(
M
P
a
)
1 7 9
x 10
3.5
t=24, 48, 72,168, and 6
w Mix 45-1
2 h
2.5:
1.51
0.5
1 0 '
t-t* (h)
x 10
3.5
t=24, 48, 72, 168, and 672 h
___ Mix 45- 2
2.5
:-i.5
cc
0.5
1 0 ' 1 0 '
t - f (h)
x 10
3.5
t=24, 48, 72,168, and 6;
^ Mix 45-3
2 h
2.5:
1.5
0.5
1 0
t - f (h)
x 10
3.5
t=24, 48, 72, 168, and 6 r2h
Mix 45- 4
2.5
0.5
t - f (h)
Figure 6.28: Calculated relaxation and curve fits, w/b=0.45.
R
(
t
,
f
)

(
M
P
a
)

R
(
t
,
f
)

(
M
P
a
)
1 8 0
x 10
^24, 48, 72,168, and 6
Mix 35-1
'2 h
x 10
f=24, 48, 72,168, and 67 2 h
aw Mix 35-2
Q-
t-r (h)
x 10
t=24, 48, 72, 168, and 6
Mix 35-3
2 h
1 0 '
t - f (h)
x 10
f=24, 48, 72,168, and 6 r2h
Mix 35-4
CO
o.
2
cc
1 0
t - f (h)
Figure 6.29: Calculated relaxation and curve fits, w/b=0.35.
1 8 1
and are the aging moduli of respectively, creep compliance and relaxation
modulus, and is the retardation time. As suggested in [Baz.2], can be equally
spaced in the logarithmic axis and thus, can be determined using the following form:
rM= l o ^ n (6.37)
where t\ can be found from the point where the creep curve plotted in log(t-t) scale
begins to rise. The instantaneous or elastic modulus is calculated by fitting Eq.(6.33)
to the age-dependent Youngs modulus after setting r ^ - 0 several magnitude smaller
than 7-^=1, e.g. r 0 = 1 .0 3t i .
The aging moduli D^ and E^ are usually obtained using the least quare minimiza
tion. However, such a method often leads to negative moduli and as a consequence,
the thermodynamic restriction on creep compliance can be violated [Car.4]. In this
study, another method based on the asymptotic series expansion of laplace trans
formation is employed. This approximation has been discussed and used in [Tsc.l,
Baz.8 ]. The objective is to calculate the continuous form of and E^ in terms of
continuous retardation spectrum L(r). The derivation of this method can be found
in [Baz.8 ]. For a given creep compliance function or creep coefficient the
continuous retardation spectrum is:
= C6-38>
where r is now the continuous retardation time, C^ is the kth. derivative of function
C, and k must be larger than one. The discrete value of and E^ are calculated
from
- J - = L{r) In 10A(log t) (6.39)
E^ = L(t ) In 10A(logrM) (6.40)
1 8 2
1.2
Mix 35-1
t=24 h
0.8
07/
t'=72 h -
0.6
0.4
t=168
0.2
f =336 h
1 0 ' 10
t - f (h)
Figure 6.30: Dirichlet series representation of creep function (dashed lines indicate
the log-power approximation).
The interval A(logr/1) can be safely set as log 10 = 1. The above method can be
used to determine the age-dependent moduli by calculating the moduli at different
ages and interpolating the values using some smooth functions.
As in the case of q<i in the log-power creep, D^{t') can be accurately fitted by a
power equation:
D^ t ' ) = e i ,e2 (6.41)
for each value of retardation time. On the contrary, can not as easily be fitted
by a certain function except by the approximation above.
The Dirichlet series representations of creep data, as indicated by solid lines, are
shown in Figure 6.30 together with the log-power creep law. In this figure, the series
consist of five terms. The result of the asymptotic approximation looks very satisfying
1 8 3
3.5
r=24, 48, 72, and 168 h Mix 35-1
t ^ . l O ^ O ^ I O ^ I O M O ^ I O '
2.5
re
CL
2
1.5
cr
0.5
1 0 ' 1 0 1 0
t - f (h)
Figure 6.31: Dirichlet series representation relaxation function (dashed lines indicate
the hyperbolic approximation).
1 8 4
and accurate. The result of the approximation for the relaxation modulus is shown
in Figure 6.31. As indicated by solid lines, the Dirichlet series representations are not
as good as in the case of creep. Seven terms are needed in the series. It appears that
due to significant relaxation the number of terms needed to produce quite accurate
series increases. The order of the asymptotic expansion (index k) are set to three,
for either the creep (Eq.(6.33)) or the relaxation (Eq.(6.34)) function.
6.7.3 Representation o f Aging by Heat o f Hydration
The source of aging is undoubtedly hydration. A measure of hydration that
is suitable in our case is heat of hydration. However, finding the most accurate
functions that can fit age-dependent material parameters so that the age (') can
be replaced by heat of hydration (Q), is not trivial. The most logical approach is
to invert the time variable in the heat equation (Eq.(6 .1 )) and substitute it with t ' .
Since all aging parameters in Eq.(6.3l), (6.34), and (6.41) are power type equation,
after some generalization, this leads to:
x ( Q ) a (ln (6-42)
where x can be <72 or m2 or D^.
In Figures 6.32 to 6.35, the creep and relaxation parameters q2, m2, D^, and
are plotted against age and heat of hydration. The fits for each parameter obtained
using the general form Eq.(6.40) are also shown as solid lines. The relations between
creep and relaxation parameters, and heat of hydration can be assumed unique, just
like the relations between other mechanical properties and heat described above.
In this study, a direct determination of D^ and by the least square method
does not produce completely non-negative values. On the other hand, the values
obtained using the asymptotic approximation may. However, this may not apply for
1 8 5
all relaxation functions. For example, an exponential type function,
R( t , t ') = m i (t')
U (
\ m2 (?)'
pw
1 exp I -
[ t - f )_
(6.43)
is able to produce a very good fit of the relaxation data. Yet in this study, it has
been found to produce negative values of E^. The thermodynamic argument of
non-diverging relaxation modulus requires that [Car.4]
dt dt
which obviously requires
d2R_ *
/ Zt
n=i
E,
> 0
(6.44)
(6.45)
Some negative values of may not always result in the violation of Eq.(6.43), but it
is not realistic according to the solidification theory [Car.4]. Figure 6.36 showing the
plots of Eq.(6.43) suggests that the result of the asymptotic approximation above is
satisfactory. As for the creep compliance the non-divergence argument reads [Baz.l]:
dt dt '
> 0 (6.46)
which will always be satisfied by guarantying the non-negativeness of D^. The plots
of Dfj. against t' shown in Figure 6.36 satisfy this requirement.
6.7.4 Effects of Mineral Additives.
Effects of mineral additive on creep are best seen in terms of the specific creep
(f>{t, t'). Because it describes the magnitude of the creep strain relative to the elastic
strain per a unit step load. By plotting 4>(t, t') not only the creep strain is included
but also the elastic strain. Figure 6.37 illustrates how 4>{t,t') varies with the loading
time. Since all concrete is made of the same aggregates and with the same binder
content, any variations of creep from one mix to the others must be caused solely
1 8 6
10
Mix 35-1
1 0
,-2
1 0
1 0
200 300 400 100 500 600 700
Age (h)
Mix 35-1
1 0 '
-1
,-2
,-3
150 200 1 0 0 0 50 250 300 350
Q (J/gr-binder)
Figure 6.32: Variation of parameter q<i with age and degree of hydration.
1 8 7
Mix 35-1
CM
- 1
-2
, -3
200 400 600 0 800 1000
Age (h)
Mix 35-1
-2
50 1 0 0 150 200
Q (J/gr-binder)
250 300 350
Figure 6.33: Variation of parameter m2 with age and degree of hydration.
1 8 8
10'
: Mix 35-1
1 0
2 1 0
1 0 '
10 1 0 ' 1 0 10
Age (h)
Mix 35-1
1 0
1 0 '
1 0
1 0 0 250 50 150 200
Q (J/gr-binder)
300 350
Figure 6.34: Variations of with age and heat of hydration.
1 8 9
Mix 35-1
1.8
1 . 6
1.4
1.2
Q.
LLI
0.8
0.6
0.4
0.2
200 400 600 800 1000 1200
Age (h)
18000
Mix 35-1
16000
14000
12000
1 0 0 0 0
6000
4000
2000
250 300 150 200 1 0 0 350
Q (J/gr-binder)
Figure 6.35: Variations of E^ with age and heat of hydration.
1 9 0
-----r
T = 1
LU
/ J O
,-5
LU
,-10
1 0
- 15
Mix 35-1
0 200 400 600 1200 800 1000
t(h)
Figure 6.36: Variations of E / t* 4 - E/ r ^ with age.
by the binder type and the additive type. It appears that the presence of SF seems
to induce more creep, while the presence of GGB tends to lower the magnitude of
creep.
Effects of additive on the relaxation modulus can be seen in Figure 6.38. As in the
case of creep, the relaxation modulus must be first normalized by the elastic moduli.
Such a normalization allows us to see how the stress in each mix relaxes from the
same initial value. As shown by Figure 6.38, concrete containing SF appears to relax
more rapidly than any other mixes. Relaxation of concrete containing FA and GGB
do not seem to significantly differ from OPC concrete.
1 9 1
25
35- 4
f =72 h
2 0
35-2 -
I, 35-1
15
35- 3
1 0
5
1 0 1 0 1 0 '
t - f (h)
Figure 6.37: The effect of additive on specific creep (f>{t, t').
0.9
t=72 h
0.8
35-1
0.7
.35-2
0.6
0.5
3 5 - 3
0.4
0.3
0.2
3 5 - 4
N.
0.1
X . .
1 0 ' 1 0
t - f (h)
1 0 1 0
Figure 6.38: The effect of additive on normalized relaxation modulus R( t , t ' ) / E( t ' ) .
CHAPTER VII
EARLY AGE STRESS ANALYSIS OF
HYDRATING CONCRETE
7.1 Introduction
Several fundamental considerations involved in the early age stress analysis of
concrete. Among others, these considerations can be reduced to only four. First,
deformation components needs to be determined. Second, relevant mechanical prop
erties must be obtained as inputs. Then, a general measure of aging must be defined
and its relation to the deformation components and the relevant properties must be
established. The last one, some procedure incorporating the considerations above
must be developed, as a computational tool.
A number of study has been focused on developing procedures for predicting the
early age stress development. These studies were pioneered by researchers in Europe
[Emb.l,Ros.2,Onk.l] however, the problem has now become recognized all over the
world. Due to the complexity involved, only few rigorous experimental and analytical
works have been done. In [Emb.l], numerical methods based on integral and rate
formulation were presented and validated with experiments. This report excluded
the autogeneous deformation component and did not use degree of hydration as the
aging variable. More recent investigations [Ser.l, Gue.l] tried to included maturity or
192
1 9 3
equivalent time in their computational procedures. More involved numerical studies
involved the use of finite element method for solving the heat transfer and stress
development in hydrating concrete. Other studies focused on more specific issues
like the mechanism of autogeneous deformation [Koe.l,Jen.l,Hua.lj and maturity
concept [Yu.l]. Some of these studies considered silica fume systems especially those
concerning the autogeneous deformation [Jen.l]. However, very little work has been
done on other cement binders involving fly ash and slag.
Considered in this chapter is the early age stress behavior as affected by hydration
process in blended cement systems. The work presented here aims at establishing
connections between hydration-induced mechanical behavior and kinetic behavior in
blended cement systems, and is a continuation from Chapter 6 . The outcome of
Chapter 6 enable to use degree or heat of hydration instead of age (?) as a measure
of aging, to incorporate temperature effects, and provide inputs for calculating the
of early age stress.
7.2 Uniaxial Stress Calculations
7.2.1 Important Material Parameters
From Chapter 6 , several constitutive functions and parameters have been pre
sented. Inputs needed for the calculation of early age stress can be summarized
as:
1. Heat of hydration (Q).
2. Autogeneous deformation (e.4 ).
3. Temperature history.
4. Coefficient of thermal dilation (o:t).
194
Table 7.1: Material parameters for the relaxation modulus.
Mix ki
k2 P
a
0 7
45-1 1035.3 0.765 0.7 0.197 362.93 1.752
45-2 106.91 1.333 0.56 1 . 0 0 326.64 1.268
45-3 1 0 1 1 0.773 0 . 6 0.036 453.37 5.375
45-4 590.55 0.938 0 . 6 0 . 1 0 337.76 1.596
35-1 150.97 1.308 0.7 0.27 303.38 1.314
35-2 170.87 1.295 0.7 0.215 290.99 1.195
35-3 810.97 0.893 0.56 0 . 1 0 2 321.67 1.194
35-4 840.89 0.879 0.7 0.044 329.89 1.583
5. Relaxation modulus (R).
6 . Elastic moduli (E and v).
Almost all material parameters listed above relate to hydration. The coefficient of
thermal dilation that is relatively constant. The other material parameter that can
be assumed constant is the Poissons ratio (u) [Nev.2,Baz.l]. For the relaxation
modulus expressed by Eq.(6.33), six material parameters are needed: k u ko, p. a,
/?, and 7 . These parameters are given in Table 1 for all mixes. The inputs for
the calculation of heat of hydration, thermal (coefficient fo thermal dilation) and
autogeneous deformations have been given in Chapter 6 .
In this study, only two deformation components are considered: thermal and
autogeneous deformations. The other deformation component such as drying shrink
age is not considered as this property is developing much more slowly over time.
Inclusion of drying shrinkage requires more experimental data and a more rigorous
analytical treatment.
1 9 5
7.2.2 Stress Calculation
Two methods that are widely used to perform stress calculations are based on
the superposition integral (integral formulation) and the Dirichlet series expansion
(rate type formulation) [Baz.l]. Using the integral formulation, the calculation of
stress due to restrained thermal and autogeneous deformations can be performed in
two ways. First by a direct use of the creep compliance, which leads to solving for
a(t) in:
e(t) eT(t) eA(t) = f (7.1)
J o
and second, by using the relaxation modulus which requires solving for cr(t) in:
a(t ) = [ R(t , - eT(fO - <u(')) (7.2)
J o
In both equations above, the strain components have been split into three terms: the
load induced strain (e), the thermal strain (eT), and the autogeneous strain (e.4 ). For
the early age problem at hand, the load induced strain is zero. The stress calculation
based on the rate formulation can be done using the dirichlet series representation
of the relaxation modulus. In the rate formulation, the series representation of
relaxation modulus allows the integral formulation to be converted into a set of first
order ordinary differential equations of the form:
^ ^ = E(t)(e(t) - eT(t) - <u(i)) (7.3)
at Tp
a(t) = ^ 2 a ^ t ) (7.4)
(i= 1
The rate type formulation essentially allows for more efficient computation. Un
fortunately, in this study, the series representation of relaxation modulus cannot be
obtained in closed form and is too complicated to model (Chapter 6 ). As discussed
previously, the Laplace transformation method used was very accurate and produced
1 9 6
thermodynamically admissible dirichlet series, but age dependent parameters gener
ated by this procedure became so complicated- Thus, the integral formulation was
chosen instead. Further, the solution was obtained using Eq.(7.2) instead of Eq.(7.4).
The stress calculation incorporating the effect of temperature on aging is done
by replacing the age (?) with the heat of hydration (Q). Due to the non-isothermal
temperature history, the heat of hydration has to be calculated numerically using
the method given in Chapter 6 . Then the early age stress history is calculated using
Eq.(7.2) above. As has been shown earlier, the coefficient of thermal dilation coeffi
cient (at) can be considered constant, so the thermal strain is linearly proportional
to T. The stress is calculated numerically using the following form:
a(t ) = f R( t , Q( ?) ) d( - atT ( ? ) - e A(Q(?))]
J o
= a t f R(t, Q(?))dT(?) - [ ' R(t , Q(?))deA(Q(?)) (7.5)
J o J o
where R is expressed by Eq.(6.31). The numerical integration chosen here is the
trapezoidal rule which is considered accurate [Nev.2]. According to the numerical
scheme given in [Nev.2], the stress increment can be calculated using:
&cr(tj+1) = ------- ^ --- [R(tj+Ut j - i ) -+- i2(tj+i,ty+1)] (7.6)
At time ti+1, the stress increment due to AT} and AeAj is:
A<r(ii+1) = - ATl + AeAJ {R(tM + fl(ti+1, i , +l)l (7.7)
The total stress at time tt-+i is the sum of the stresses due to increments of AT} and
AeAj applied during all the previous intervals:
i
<*{u+1) = 5 Z (7-8)
j i
1 9 7
7.2.3 Verification with Experiments
The above procedure is now used to predict the stress development measured in
the early age stress testing. The results are shown in Figures 7.1 to 7.12. In each
figure, both temperature and stress history are shown. In addition, the development
of split tensile strength (f sp) is also shown. Notice that 0.85 f sp is plotted instead
of f sp. This is done in order to take into account the effect of deformation rate
on strength. The strength measured at very slow deformation rates, such as in our
case, has been reported to be around 85% of the value measured at a medium rate
[Baz.l,Baz.6 ]. Plotting f sp in each figure also allows us to check the assumption made
in formulating Eq.(7.2), which is linear superposition of stress. In most mixes, the
stresses measured do not exceed 70% of 0. 85f sp. In has been argued in [Baz.l] that
microcracking of concrete does not begin below about 80% of the ultimate stress.
Two stress predictions are made, first, including the effect of temperature on
aging and second, without including the effect of temperature. These prediction
are indicated as, respectively, pred. 1 (solid lines) and pred. 2 (dashed lines) in
Figures 7.1 to 7.12. Overall, the predictions including the temperature effect are
very accurate and more superior compared to the prediction without the temperature
effect. This is to be expected since at early ages, temperature significantly accelerates
the development of viscoelastic properties.
The effect of different temperature history can be seen in Figures 7.5 to 7.12.
The response of concrete subjected to two different temperature history is very obvi
ous, since concrete properties develop differently under different temperature history.
As before, the predictions including the temperature effect appear to be very accu
rate. This suggests that a significant improvement from the ordinary method can be
achieved by the proposed procedure.
S
t
r
e
s
s
/
T
e
n
s
i
l
e

S
t
r
e
n
g
t
h

(
M
P
a
)

T
e
m
p
e
r
a
t
u
r
e

f
C
)
1 9 8
40
Mix 45-1
35
30
25
2 0
15
2 0 1 0 30 40
t(h)
50 60 70 80
3.5
0.85f
sp
2.5
pred. 1_
pred. 2
0.5
-0.5
2 0 40 1 0 30 50 0 60 70 80
t (h)
Figure 7.1: E a r l y age stress development in mix 45-1.
S
t
r
e
s
s
/
T
e
n
s
i
l
e

s
t
r
e
n
g
t
h

(
M
P
a
)

T
e
m
p
e
r
a
t
u
r
e

^
C
)
1 9 9
35
Mix 45- 2
30
25
2 0
1 0
70 60 40 2 0 30
t (h)
2.5
0.85f "
pred. 1
pred. 2
-0.5
70 60 40 2 0 30
t(h)
Figure 7.2: Early age stress development in mix 45-2.
S
t
r
e
s
s
/
T
e
n
s
i
l
e

S
t
r
e
n
g
t
h

(
M
P
a
)
200
40
Mix 4 5 - 3 .
35
25
S. 2 0
I
15
2 0 30 40 70 50 60
t(h)
- 0.85f
2.5
pred.1
pred.2
1.5
0.5
-0.5
2 0 30 40 70 50 60
t(h)
201
45
Mix 4 5 - 4 - 40
35
O
30
25 CO
Q.
E
CD
I
2 0
1 0
2 0 30 1 0 50 40
t(h)
60 80
3.5
0.85f
pred. z
05
pred. 1
- 0 . 5
0 2 0 30 1 0 50 60 40 70 80
t(h)
S
t
r
e
s
s
/
T
e
n
s
i
l
e

s
t
r
e
n
g
t
h

(
M
P
a
)

T
e
m
p
e
r
a
t
u
r
e

(
C
)
202
40
Mix 35-1
30
25
2 0
2 0 30 50 60 70
t(h)
3.5 0 - 8 5 f -
2.5
pred. 1
pred. 2
0.5
-0.5
2 0 30 40 60 50 70
t (h)
2 0 3
45
mix 35-1
40
35
25
2 0
2 0 40 50 60
t(h)
pred. 1
2.5
CL
pred. 2 .
0.5
-0.5
2 0 30
t(h)
40 50 60
Figure 7.6: Early age stress development in mix 35-1, second test.
S
t
r
e
s
s
/
T
e
n
s
i
l
e

s
t
r
e
n
g
t
h

(
M
P
a
)

T
e
m
p
e
r
a
t
u
r
e

f
c
)
2 0 4
40
Mix 35-2 - 35
25
2 0
15
1 0
2 0 30 1 0 40
t (h)
60 70 80
0.85f
3.5
pred. T
pred. 2
-0.5
0 2 0 30 1 0 40 50 60 70 80
t (h)
2 0 5
50
45
------- T 1 1------------------------
mix 3 5 - 2 .
40
/ \
-
o 35
/ \
CD
w
3
30
_ \
25
CD
CL
25
>S.
E
a>
H
20
- N .
15
- N .
10
-
5
------------------1---------------------- 1 I . 1 _ i
0 10 20 30 40 50 60
t ( h )
1.6
pred. 1
1.4
1.2
0.8
pred. 2
0.6
co
0.4
0.2
- 0.2
2 0 30
t ( h )
40 50 60
S
t
r
e
s
s
/
T
e
n
s
i
l
e

s
t
r
e
n
g
t
h

(
M
P
a
)

T
e
m
p
e
r
a
t
u
r
e

f
C
)
206
40
Mix 3 5 - 3 - 35
30
25
2 0
15
2 0 30 40 50 70 60
t (h)
3.5
0.85f,
2.5
s>
pred. t
0.5
- 0 . 5
40 2 0 30 50 60 70
t (h)
2 0 7
45
--------------------------!------ --------------
-----------,------- - ---- 1------------------------- 1-------------------------
40 / \
mix 35-3
o
35 -
3
30
2
CD
Q.
25
E
CD
1 -
20 \
15
-
10
5
1 -----1--------------
---------- 1------------------------- 1------------- ------------ --------------------
0 10 20 30 40 50 60
t(h)
1.5
Pred- 1
^' pred. 2
CO
Q.
2
CO
CO
Q>
0.5
co
-0.5
1 0 2 0 0 30 40 50 60
t(h)
S
t
r
e
s
s
/
T
e
n
s
i
l
e

s
t
r
e
n
g
t
h

(
M
P
a
)

T
e
m
p
e
r
a
t
u
r
e

^
C
)
2 0 8
45
40
Mix 35-4 -
35
30
25
2 0
15
1 0 *
2 0 30 40
t(h)
60 70
3.5
0.85f
2.5
pred. 1
23DQ5BD552D pred. 2
0.5
-0.5
1 0 2 0 30 40
t(h)
50 60 70 80
2 0 9
40
mix 3 5 - 4
30
25
2 0
2 0 50 60 40
t(h)
0.7
0.6
0.5
pred. 1
0.4
to
Q_
pred. 2
0.3
g 0.2
0.1
- 0.1
CO
- 0.2
1 0 2 0 30 0 60 40 50
t (h)
210
7.2.4 Effect o f Mineral Additives
To study the early age stress of concrete containing different additives, one tem
perature history has to be chosen as a reference. Figure 7.13 shows the results of
stress development in different concrete mixes subjected to the temperature history
that belongs to mix 45-1. Essentially, the presence of additives reduces the magnitude
of stress. Such a reduction is even more apparent in mixes having lower water-binder
ratio (w/b). The stress in mixes containing silica fume (45-4 and 35-4) appears to be
lower than the stress in any other mixes. This also confirms the finding observed in
Chapter 6 , when the normalized relaxation modulus of different mixes are compared.
From the above finding, it is conclusive t h a t the use of additives can be beneficial
in reducing the risk of early age cracking.
7.2.5 Effect o f Aging
In previous discussions, the effect of temperature on aging has been shown to be
significant. To further elaborate this effect the stress development is predicted for
one concrete mix subjected to a set of unique temperature history. After temperature
is held constant for certain periods of time, t 0, it is dropped at the rate of 5C per
hour. The purpose is to see how the stress develops in concrete that is allowed to
reached different levels of hydration before subjected to a temperature drop. For this
purpose, mix 35-1 is chosen as an example.
The temperature history and the predicted stress are illustrated in Figure 7.14.
As expected, the temperature drop causes significant tensile stress development. In
order to clearly see the effect of aging during the period where temperature drops, it
is necessary to plot the stress with the relative time t t a. In Figure 7.15, the stress
normalized by 0.85f sp is plotted against t t Q. Also given in Figure 7.15 is the limit
211
40
35
<. 30
3 25
oo
20
a.
E
I
10
20 30 40
t (h)
50 70 80
2.5
^ 45-1 -
' " 4 5 - 2
. . . 4 5 - 3
4 5 - 4
00 0.5
-0. 5
20 30 40
t (h)
50 70 80
2.5
35-1
to 1.5 3 5 - 2
3 5 - 3 .
CO
3 5 - 4
55 0.5
- 0.5
40
t ( h )
20 30 70 80
Figure 7.13: Effects of mineral additive on early age stress.
212
when cracking occurs. It can be seen that larger ta results in more rapid increase
in tensile stress. Further, as illustrated in Figure 7.16, larger t 0 causes the cracking
stress to be reached earlier. Of course the stress close to the cracking stress cannot
be predicted using the superposition principle because the constitutive behavior is
no longer linear viscoelastic. However, assessment made using the linear assumption
can still be useful, at least up to 85% of the ultimate value. In view of this argument,
it can be said that larger t a increased the risk of early age cracking.
7.3 Application to Foundation or Pavement Slabs
Considered in this section is calculations of early age stresses in thin pavement
or foundation slabs due to general temperature profiles. The formulation first begins
with the thermoelastic case then proceeds to the thermo-viscoelastic case by apply
ing the elastic-viscoelastic analogy. The solutions are derived for two cases: without
through-thickness properties variation and with through-thickness properties varia
tion. The solution procedure for the foundation or pavement slabs will be presented
for plates satisfying the following conditions:
1. Plane sections remain plane.
2. Temperature varies only through the slab thilccness, T(x, y, z) T(z).
3. The slab rests on very stiff foundations or subgrade.
4. Drying shrinkage is excluded.
7.3.1 E l a s t i c Stress
The strain components can be expressed as [Bol.l]:
_ du d2w _ dv d2w
6xx dx dx2 yy dy 2 dy2
S
t
r
e
s
s

(
M
P
a
)

T
e
m
p
e
r
a
t
u
r
e

(
?
C
)
2 1 3
40
Mix 35-1 35
30
25
20
10
t =12 h t =24 h
t=36 h L=48h
20 40 80 1 0 0 60 140 1 2 0
t(h)
7
to=36 h 48 h t =72h.
6
5
4
3
2
1
0
1
0 20 40 1 0 0 60 80 140 1 2 0
Figure 7.14: Temperature and stress history for studying effects of aging.
2 1 4
Mix 35-1
1.8
t =72 h
1.6
1.4
t =48 h
. 1.2
in
ao
t =36 h
O. -
CO
CO
2
55 0.8
t =24 h
t =12 h
0.6
0.4
0.2
20 60 50 40
Figure 7.15: Stress developement with t t Q.
50
Mix 35-1
45
40
35
30
25
20
20 30 80 10 70 50 60
Figure 7.16: Time required to reach the critical level.
2 1 5
du dv d2w .
7x y ~ d ^ ^ d ^ ~ d^dy ^ )
From the elastic stress-strain relationship (Hooks law), stress components are:
E
&xx j" ~^\(xx "F V^-yy (1 ~F
<7yy 1 2 i^yy "F ^ ex x ( 1 -F v)cx.tT\
E
1 - 1/2
^ = 2 ( 1 + is)7xy (7-10)
where a t is the coefficient of thermal dilation (CTD). The force components can be
obtained by integrating the stress and can be written in terms of u, v, and w such
as:
__ Eh ( d u d v \ Nt
N* ~ rr ^ \ d i + , ' w - rr ^
__ Eh ( d v d u \ Nt
1 y ~ 1 v2 \ dy + V~dx) ~
. . Eh ( du d v \ .
* " 2 ( 1 + 7 ) i d ^ + " s i ) ( u )
. . d2w d2w \ Mt
1 ~ Vdx2 + U~d^2) ~ r ^
r _ f d 2w d2w \ Mt
y Vdy2 V dx 2 ) 1 v
M,
lxy
dxdy
d3w d3w
v* = ~ D a.? ; ^ = ~ DW (713)
where D = Eh? j 12(1 v2) is the flexural rigidity and quantities Nt and Mt are:
rh/2 rh/ 2
Nt =act ETd z ; Mt = oct I ETz d z (7-14)
J-h/2 J-h/2
Notice that in defining iVr and Mt , E is kept inside the integrand to allow the
formulation involving the variation of mechanical properties through the thickness.
2 1 6
Meanwhile, a t is kept outside the integrand since it has been found to be relatively
constant (see also Chapter 6 ).
After substitution involving Eq.(7.10) to (7.13), the stress components can be
expressed in terms of the forces:
crxx = ^ ^ | a.tET -+- ^[(1 v)Nx + Nt ]-I- u)Mx + Afr]|
= j 4 i ; { - atET + i [ ( l - V)N, + ATr ] + ^ [ ( 1 - v)M, + Mr]}
ayy
<r*y = j^Nzy - ^ r Mxy (7.15)
In the case of the unrestrained plate with all edges traction free, the elastic stress
solution can be obtained by satisfying the equilibrium condition at the free edges:
fh/2 rh/2
/
n / z rn/ 2
axxdz = / avvdz = 0
-A/2 J-h/2
rh/2 rh/2
/ axxzdz = / (TyyZdz = 0 (7.16)
J-h/2 J-h/2
The boundary conditions for shear forces are not considered since the thermal strain
does not contribute to shear deformations and thus, the shear forces are always zero
everywhere. The same argument also holds to the twisting moment. It is easy to see
from Eq.(7.11), (7.12) and (7.15) that Nx, Ny, Mx, and My must be zero everywhere.
Therefore the elastic stress components now become:
1 f Np 1 2 Afj-z 1
= r ^ a _ a , E T + + p - }
1 f N t 12 M t z I
w - r r ^ t ~ ET + I T +
(Try = 0 (7-17)
The strain components and the curvature can be found from Eq.(7.11) and (7.12),
du dv rh/ 2
du dv r
dx dv at J - dy J-h/ 2
d2w d2w 1 2 rh/ 2
Tdz
12 rni z
= ~TTa t / Tz dz (7-18)
h3 J-h/2 dx2 dy2 h3 J-h/ 2
2 1 7
When a slab rests on a very stiff foundation, any axial and bending deformations
produced by temperature can be completely restrained. The stress components asso
ciated with the restrained slab are obtained by superimposing the stress components
from the unrestrained condition with the stress required to restraint the free strains
and curvatures above. In this study three cases are considered:
1 . Case A, the bending deformation is restrained.
2 . Case B, the axial deformation is restrained.
3. Case C, both bending and axial deformations are restrained.
The stress components needed to restrained the deformations can be obtained by
applying boundary forces that cancel such deformations. These forces are calculated
using Eq.(7.11) and (7.12). It is obvious that such forces are described by the second
terms in Eq.(7.11) and (7.12). Therefore, the restraining normal and bending forces
are:
NT
N = N =
i y x . r l v t / , r -
i i/
Mt
MXir = My^r = ^ (7-19)
The total stresses can be obtained from Eq.(7.15) by substituting Nx, Ny, Mx, and
My with, repectively, iVI)7., iVy>r, Mx<r, and My<r. The total stress components are
found to be:
1 f Nt \
oxx = cryy = ^ _ f octE T + J ; for case A (7.20)
1 f Y2M t z \
oxx = oyy -- - - ( octE T + -I---- J ; for case B (7-21)
ottE T
Oxx = Oyy = _ ^ ; for case C (7.22)
2 1 8
7.3.2 Viscoelastic Stress with Through-thickness Properties Variation
The thermo-viscoelastic stress calculation can be done by applying the elastic-
viscoelastic analogy. According to [Baz.6 ] the viscolastic stress can be calculated
using:
<7-23)
In the expression above, the Poissons ratio is assumed to be time independent,
which is valid for most concrete materials [Baz.l]. Eq.(7.23) can be generalized to
the problem involving through-thickness variations of E and R. This is done by
keeping E and R in the integrand (see Eq.(7.14)). Thus, the viscoelastic stress
components can be written as:
crxx(z, t) = ayy(z, t) = *' 1 z )d Ti t>) + TF? - - "T
1 v J o 1 u )
rh/2 rt
N } ( t ) = a t l I R( t , t ' : z) dT( t t, z) dz- , i ovcaseA (7-24)
J-h/ 2 Jo
o-xx(z, t ) = ayy(z, t ) = f0 t l z)dT(t ' , z) 4-
fh/2 rt
M^ ( t , z ) = a tz / R(t , t', z) dT{t \ z)zdz; for case B (7-25)
J-h/ 2 Jo
axx(z, t) = ayy(z, t) = f R(t , ', z)dT{t' , z ); for case C (7.26)
1 u J o
The general expressions when aging is expressed in terms of heat, Q, and the
autogeneous strain is included are:
<Txx(z,t) = z ~ f R( t , Q( t \ z ) ) [ a tdT( t ' , z ) +de A{Q(t'))]
1 V JO
, m t )
h( 1 - u)
rh/2 rt
N*( t ) = / / R( t , Q( t \ z ) ) [ at dT( t ' , z ) +d e A(Q(t' ))]dz (7.27)
J-h/2 JO
2 1 9
for case A.
crxx(z, t ) = ^ f R{t , Q{t ' , z) ) [atdT{ t \ z ) + deA{Q{t'))}
1 v J o
1 2 M(, z)
+ h3( 1 - u )
rh/2 rt
M^{t , z) = z / R(t , Q(t' , z))[atdT(t' , z) + deA{Q{t'))]zdz (7.28)
J-h/2 Jo
for case B, and
<7 rx(-z, t) = ^ f R(t , t \ z)[atdT(t ' , z) + deA{Q(t'))} (7.29)
1 v J o
for case C.
7.3.3 Viscoelastic Stress without Through-thickness Properties Varia
tion
The case without through-thickness properties variation is essentially a simplifica
tion of the previous case. The only difference is that the spatial and time integrations
can now exchange places. So the viscoelastic stress can be expressed as:
1 r l
<7xx(z,t) -------- f R(t,t'))[cLtdT( t ' , z) -hdeA(t') +dNT(t')/h]
1 u J o
rh/2
dNri t ' ) = / [atdT(t ' , z) + deA(t')]dz (7.30)
Jh/2
(z , t ) = - f R(t , t ' ) ) [atdT(t' , z) -f- deA(t') 4- 12dMT(t', z ) / h 3]
1 v Jo
for case A.
J _ f t
rh/2
dMri t ' ^z) = z f [a.tdT{t ' , z) + deA(t')]zdz (7.31)
J h/2
for case B, and
crXx(z, t) = ^ f R(t, t' )[atdT(t ' , z) -+- deA(t')] (7.32)
1 v Jo
for case C.
220
40
Bottom
35
Middle
g 25
Top
\
20
Ambient
20 80 40 60
time (h)
100 1 2 0
Figure 7.17: Temperature history at different locations in the slab.
40
t=24h
35
t=6h
oe-
t=59h
E25
20
t=108h t=71-25h t=82.5h
-10 - 5 15
z (cm)
Figure 7.18: Temperature profiles at different time.
221
Top of slab
0.75
With property variation
0.5
CO
CL
2
r 0.25
CO
2
55
- 0 . 2 5
Without property variatidh
- 0 . 5
20 40 1 2 0 1 0 0 80
time (h)
Figure 7.19: Stress on the top surface of the slab (Case C).
0.6
Middle of slab
With property variation
0.5
0.4
0.3
0.2
CO
CO _ .
ffl 0.1
55
Without property variation
- 0.1
- 0.2
- 0 . 3
20 40 1 0 0 1 2 0
time (h)
Figure 7.20: Stress at the middle thickness of the slab (Case C).
222
0.5
Bottom ot slab
With property varial
0.4
0.3
~ 0.2
1 Without property variation
- 0.1
- 0.2
- 0 . 3
40 1 2 0 20 60
time (h)
80 1 0 0
Figure 7.21: Stress at the bottom of the slab (Case C).
7.3.4 Application
The procedure presented above is now applied to concrete having real temperature
history. The experimental work has been detailed in [Moh.l]. Essentially, Mohr
et.al. measured the temperature profiles of a slab resting on granular sub-base under
a certain history of ambient temperature. The data were collected for a 26.7 cm
thick slab at every 15 min. increment. The measured temperature history is shown
in Figure 7.17 for the ambient and for three locations in the slab. The temperature
profiles at different time are shown in Figure 7.18.
To calculate the stress development, material properties corresponding to mix
35-1 are used. The first prediction is made for case C in Figures 7.19 to 7.21.
As seen from Figures 7.19 to 7.21, the difference in temperature history in each
location results in different stress development. Furthermore, it can be seen that
2 2 3
0.4
0.2
CO
CL
co
CO
0)
L_
55
- 0.2
- 0 . 4
c a s e A t=24h
iL
N
0.4
0.2
0
- 0.2
- 0 . 4
c a s e B
v . o L
t=24h
-k v
4-
0.4
0.2
- 0.2
- 0 . 4
case C
t=24h
- 1 5 - 1 0 - 5 0 5 10 15 - 1 5 - 1 0 - 5 0 5 10 15 - 1 5 - 1 0 - 5 0 5 10 15
z (cm) z (cm) z (cm)
0.6
0.3
co
CL
CO
CO
s
55
- 0. 3
- 0.6
t=71.25h
\
0.6
0.3
- 0 .3
- 0.6
Vv
t=71.25h
4-
0.6
0.3
0
- 0 . 3
- 0.6
t=71.25h
\
- 1 5 - 1 0 - 5 0 5 10 15 - 1 5 - 1 0 - 5 0 5 10 15 - 1 5 - 1 0 - 5 0 5 10 15
CO
CL
CO
CO
CD
55
1
0.8
0.6
0.4
0.2
0
- 0.2
t=82.5h
1
0.8
0.6
0.4
0.2
0
- 0.2
t=82.5h
V
1
0.8
0.6
0.4
0.2
0
- 0.2
t=82.5h
.T, n -----J
- 1 5 - 1 0 - 5 0 5 10 15 - 1 5 - 1 0 - 5 0 5 10 15 - 1 5 - 1 0 - 5 0 5 10 15
F i g u r e 7.22: S t re ss p r o fi le s a c cr o ss t h e s l a b t h i c k n e s s .
2 2 4
the calculation considering the variation of properties through the slab thickness is
significantly different from the one without considering the variation of properties.
In this study, the maximum stress is found to be on top of the slab. In general,
the location of the maximum stress depends strongly on the temperature history.
However, in field concrete, the top surface is also exposed to drying which causes
shrinkage in addition to thermal deformation. Under this circumstances the top
surface of the slab possibly experiences the greatest stress.
Stress profiles due to the temperature history are given in Figure 7.22 for all three
cases. In each figure, stresses are calculated with without considering the through-
thickness variation (or effect of temperature). Both calculations seem to be very
similar in case A but not in case B or C. The stress profiles are notably different
in different cases. In case A, where only bending deformation is restrained, the
tensile stress appears to be low. On the other hand, the tensile stress is seen to be
much higher in case B. Meanwhile the stress profiles case C, where both bending
and axial deformations are restrained, tend to follow the stress profiles in case B.
This suggests that a significant contribution to tensile stress is given by restrained
axial deformation. Under general temperature history however, there certainly is a
possibility that restrained bending deformations can also contribute significantly to
tensile stress.
CHAPTER VIII
CONCLUSIONS, AND RECOMMENDATIONS
8.1 Conclusions
The research conducted here encompasses a broad spectrum of concrete behavior.
Although the scope is confined to areas such as hydration kinetics, microstructure,
deformations, and early age stress, the outcome from of each area also provide infor
mation and implications about concrete behavior in general.
It has been shown in this research that hydration kinetics can be considered
as the most definite source of aging behavior. Almost all time dependent concrete
properties are strongly tied to hydration. A new method is proposed in this research
to determine degree of hydration and mechanical property development under non-
isothermal conditions. Applying more than one experimental techniques has been
proven to be important in explaining hydration mechanisms especially in blended
cement systems. Based on the DTA/TGA analysis degree of hydration of blended
cement systems and degree of hydration of pozzolans can be determined. When hy
dration kinetics are combined together with microstructural analysis and microme
chanics, some fundamental mechanisms like drying and autogeneous shrinkage can
be well described. Microstructural-based modeling developed here turns to be a pow
erful tool in explaining such basic concrete behavior. Experiments on early age creep
225
2 2 6
and stress development and on hydration kinetics have given insight on the thermo
mechanical response of hydrating concrete. Very valuable material parameters are
obtained from these experiments. A more improved procedure for predicting early
age stress has been developed making use of the aforementioned material parameters.
Findings gathered from all chapters in this dissertation must be divided according
to the area of investigation. The conclusions drawn this way tend to be less general
but are more definite. The conclusions are:
1. Hydration Kinetics
The effect of temperature on hydration is very important and can be well
captured by the three-parameter equation (TPE). Such an effect can be
invariably seen in all mixes, including in blended cement systems.
For the blended systems considered here, hydration can be measured by
means of chemically bound water. Using TPE, values of maximum ulti
mate bound water and the overall degree of hydration of blended cements
can be quantified. The result indicates that hydration is slowrer in blended
systems than in OPC.
A method based on measurements of the decomposition of Ca(OH) 2 to
gether with TPE can be used to determine the relative amount of hydrated
pozzolans. The result obtained by this method suggests that reactions of
ground granulated slag (GGB) and fly ash (FA) in cement pastes proceed
slowly, while the other pozzolan, silica fume (SF), hydrates faster than
FA or GGB.
2 . Microstructure
H20 sorption accurately measures the gel pore but not the capillary pore.
2 2 7
Accurate estimates on the surface area of hydration products can be made
using H20 desorption curves. For OPC, the surface area is about 260 m2
per gram product.
Surface area of hydration product in different pastes containing 25% FA,
25% GGB, and 10% SF are estimated to be, respectively, 257, 400, and
386 m2 per gram product.
In all mixes, the gel space determined using H20 sorption is about 2-2.5
times the corresponding monolayer volume.
N2 sorption does not measure the whole surface area of solid in cement
paste.
3. Mechanics of humidity-driven shrinkage
A model for developing capillary pore microstructure proposed in this
study can accurately capture the important experimental features.
Experimental evidence on drying shrinkage together with micomechan-
ics suggest that the major contributor to drying shrinkage is the Gibbs-
Bangham strain which operates over the entire range of relative humidity.
A minor contribution is also given by capillary strain but only on a limited
range of relative humidity.
Autogeneous shrinkage can be accurately predicted using the same mech
anisms applied during drying shrinkage together with the model for de
veloping microstructure. Essential features such as effects of water-binder
ratio and cement fineness are also captured.
4. Non-isothermal hydration and property development
2 2 8
Mechanical properties including compressive strength, tensile strength,
and Youngs modulus relate uniquely to degree of hydration. Influenc
ing factors such as temperature affects the properties through degree of
hydration.
Degree of hydration, heat and development of properties such as com
pressive strength, tensile strength, Youngs modulus, and fracture energy
under non-isothermal conditions can be calculated using the procedure
formulated based on TPE.
The temperature effect on aging can be incorporated by using heat or
degree of hydration instead of age.
The use of blended cements changes the creep property and autogeneous
strain.
5. Early age stress
Significant improvement on early age stress predictions are obtained by
including the temperature effect through degree of hydration.
The use of blended cements appear to be beneficial in reducing the risk
of early age cracking.
Input parameters such as temperature history strongly alters aging char
acteristics of relaxation modulus and thus, the stress development with
time.
Because the predicted results between the ordinary procedure and the im
proved procedure are significantly different, accurate stress predictions in
concrete slab having variations of temperature history (thus also property
variation) through the thickness must done using the improved procedure.
2 2 9
8.2 Recommendations
Based on the previous presentation, there are numerous aspects related to hydrat
ing concrete that needs more investigations. Recommended areas to be investigated
include:
1 . Microstructure-type analysis of creep and shrinkage of mortar and concrete.
2 . Effects of hydration and microstructure on transport properties such as per
meability, diffusivity, and on thermal properties such as heat conductivity and
specific heat.
3. Effects of reinforcement or aggregates and cement types on autogeneous shrink
age.
4. Development of testing procedure to determine early age stress history in lab
oratory or in the field.
5. Development of a more general procedure capable of calculating temperature
and humidity distributions, and stresses/strains due thermal, autogeneous, and
drying/swelling deformations.
6 . Creep fracture behavior of hydrating concrete.
BIBLIOGRAPHY
230
2 3 1
[Abd.l]
[Abo.l]
[Aci.l]
[Aci.2]
[Alm.l]
[Aus.l]
[Avr.l]
[Bad.l]
[Ban.l]
[Baz.l]
[Baz.2]
[Baz.3]
BIBLIOGRAPHY
Abdel-Jawad, Y.A., The Relationships of Cement Hydration and Con
crete Compressive Strength to Maturity, Ph.D Thesis, Univ. of Michigan,
1988.
Abo-Elenein, S., Daimon, M., Ohsawa, S., and Kondo, R., Hydration of
Low Porosity Slag-Lime Pastes, Cem. Con. Res. 4, pp.299, 1974.
ACI Committee 517, Accelerated Curing of Concrete at Atmopheric
Pressure State of the Art Rep., ACI Man. Cone. Prac. Part 5, 1988.
American Concrete Institute (ACI), Manual of Concrete Practice, Part
I, 1987.
Almudaiheem, J.A., The prediction Drying Shrinkage of Portland Ce
ment and Concrete, Ph.D Thesis, Univ. of Michigan, 1986.
Auskem, A., and Horn, W., ASTM J. Test Eval. 1, pp.74, 1973.
Avrami, M., Kinetics of phase change I, J. Chem. Phys. 7, pp.1103,
1939.
Badmann, R., Stockhausen, N\, and Setzer, M.J., Statistical Thickness
and the Chemical Potential of Adsorbed Water Films, J. Coll. Interface
Sci. 82, pp.534, 1981.
Bangham, D.H., and Maggs, F.A., The Strength and Elastic Constant of
Coals in Relation to Their Ultra-Fine Structure, Conf. Ultrafine Struc.
of Coals and Cokes, British Coal Util. Res. Assoc., London, pp.118, 1944.
Bazant, Z.P. (Editor), Material Models for Structural Creep Analysis,
RILEM Proc., Mathematical Modeling of Creep and Shrinkage of Con
crete, John Wiley &c Son, 1988.
Bazant, Z.P., and Wu, S.T., Dirichlet Series Creep Function for Aging
Concrete, ASCE J. Eng. Mech. EM2, pp.367, 1973.
Bazant, Z.P., Current Status and Advances in The Theory of Creep and
Interaction with Fracture, 5th Int. RILEM Symp. Creep and Shrink.
Cone. (Ed. Z. P. Bazant and I. Carol), E &: FN Spon, London, 1993.
2 3 2
[Baz.4]
[Baz.5]
[Baz.6 ]
[Baz.7]
[Baz.8 ]
[Ben.l]
[Ben.2]
[Ben.3]
[Bha.l]
[Bjo.l]
[Bio.l]
[Bol.l]
[Bor.l]
[Bro.l]
Bazant, Z.P., and Gettu, R., Rate Effects and Load Relaxation: Static
Fracture of Concrete, ACI Mat. J., Vol. 89, No. 5, 1992.
Bazant, Z.P., and Kim, S., Nonlinear Creep of Concrete - Adaptation
and Flow, ASCE J. of Eng. Mech. 105 (EM3), 1979.
Bazant, Z.P., and Planas, J., Fracture and Size Effect in Concrete and
Other Quasibrittle Materials, CRC Press, Boca Raton, 1998.
Bazant, Z.P., Thermodynamics of Interacting Continua with Surfaces
and Creep Analysis of Concrete Structures, Nuclear Eng. Des. 2 0 , pp.477,
1972.
Bazant, Z.P., and Xi., Y., Continuous Retardation Spectrum for Solidi
fication Theory of Concrete Creep, ASCE J. Eng. Mec. 121 (2), pp.281,
1995.
Bentz, D.P., Garboczi, E.J., and Quenard, D.A., Modeling of Drying
Shrinkage in Reconstructed Porous Materials: Application to Porous Vi-
cour Glass, Mod. Simul. Mat. Sci. Eng. 6 , pp.211, 1998.
Bentur, A., Berger, L., Lawrence, F.V., Milestone, N.V., and Young, J.F.,
Creep and Drying Shrinkage of Tricalcium Silicate, Cem. Con. Res. 8 ,
pp.721, 1978.
Bensted, J., Special Cements, Leas Chem. Cem. Cone. 4th ed. (Peter
C. Hewlett, ed.), pp.779, 1998.
Bhatia, S.K., and Perlmutter, D.D., A Random Pore Model for Fluid-
Solid Reactions: Isothermal, Kinetic Control, AIChE Journal 26 (3),
pp.379, 1980.
Bjontegaard, O., Thermal Dilation and Autogeneous Deformation as
Driving Forces to Self-Induced Stresses in High Peformance Concrete,
Doctoral Thesis, Norwegian Univ. Sci. and Tec., Trondheim, 1999.
Bloom, R., and Bentur, A., Free and Restrained Shrinkage of Normal
and High Strength Concrete, ACI Mat. J. 92 (2 ), pp.2 1 1 , 1995.
Boley, B., and Weiner, J.H., Theory of Thermal Stress, Dover Inc.,
1997.
Borgnakke, C., Hansen, W., Pane, I., and Abdel-Jawad, Y.A., Phe
nomenological Kinetics of Portland Cement Hydration: I. Evaluation of
Mathematical Models, unpublished manuscript.
Brooks, J.J., and Neville, A.M., A Comparison of Creep, Elasticity, and
Strength of Concrete in Tension and in Compression, Mag. Cone. Res.,
29 (100), pp.131, 1977.
2 3 3
[Bru.l]
[Bra.2]
[Bra.3]
[Bru.4]
[Byf.l]
[Car.l]
[Car. 2]
[Car.3]
[Car.4]
[Cop.l]
[Dai.l]
[Dav.l]
[Der.l]
[Dia.l]
[Dig-1]
van Bruegel, K., Artificial Cooling of Hardening Concrete, Delf Univer
sity of Technology, Research Report No. 5-80-9, Delft, 1980.
van Bruegel, K., Prediction of Temperature Development in Hardening
Concrete, RILEM Rep. No. 15, Prev. Thermal Crac. Cone, at Early Ages,
E & FN Spon, 1998.
Brunauer, S., Statistical Thickness Curve, J. Coll. Interface Sci.
van Bruegel, K., Simulation of Hydration and Formation of Structure
in Hardening Cement-Based Materials, Delft University Press, 2nd Ed.,
1997.
Byfors, J., Plain Concrete a t Early Ages, Swedish Cem. Cone. Inst.,
1980.
Carino, N.J., Temperature Effects on the Strength-Maturity Relation of
Mortar, Nat. Bureau Stdrd., NBSIR 81-2244, Washington DC, pp.98,
1981.
Carino, N.J., The maturity method: Theory and application, Cem.
Cone, and Agg. 6 (2), pp.61, 1984.
Carino, N.J., Lew, H.S., and Volz, C.K., Early Age Temperature Effects
on Concrete Strength Prediction by Maturity Method, ACI J. 80 (2),
pp.93, 1983.
Carol, I., and Bazant, Z. P., Viscoelasticity with Aging Caused by Solid
ification of Nonaging Constituent, ASCE J. Eng. Mech. 119 (11), 1993.
Copeland, L.E., Kantro, D.L., and Verbeck, G.J., Chemistry of Hydra
tion of Portland Cement, Proc. Symp. Cem. Chemistry Vol.I, Washing
ton, pp.429, 1960.
Daimon, M., Abo-Elenein, S., Hosaka, G., Goto, S., and Kondo, R., Pore
Structure of Calcium Silicate Hydrate in Hydrated Tricalcium Silicate,
J. Am. Cer. Soc. 60 (3-4), pp. 110, 1977.
Davis, R.E., Davis, H.E., and Brown, E.H., Plastic Flow and Volume
Changes of Concrete, Proc. ASTM 37 Part 2, pp.317, 1937.
Derjaguin, B.V., On the Repulsive Forces Between Charged Colloid Par
ticle and the Theory of Slow Coagulation and Stability of Lyophole Sols,
Trans. Faraday Soc. 36, pp.203, 1940.
Diamond, S., and Dolch, W., J. Colloid Interface. Sci. 38, pp.234, 1972.
Digital Site System Inc., Quadrel User manual, Pennsylvania, 1996.
234
[Emb.l]
[Emb.2]
[Fer.l]
[Gra.l]
[Gre.l]
[Gru.l]
[Grz.l]
[Gue.l]
[Han.l]
[Han. 2]
[Han.3]
[Han.4]
[Han.5]
[Hel.l]
Emborg, M., Thermal Stresses in Concrete Structures at Early Ages,
Doctoral Thesis, Lulea Univ. of Tech., 1989.
Emborg, M., Models and Methods for Computation of Thermal Stress,
RILEM Rep. No. 15, Prev. Thermal Crac. Cone, at Early Ages, E &: FN
Spon, 1998.
Ferraris, C.F., and Wittmann, F.H., Shrinkage Mechanisms of Portland
Cement Paste, Cem. Con. Res. 17, pp.453, 1987.
Granju, J.L., and Grandet, J., Relation Between the Hydration state and
the Compressive Strength of Hardened Portland Cement Pastes, Cem.
Cone. Res. 19, pp.579, 1989
Gregg, S.J., and Sing, K.S., Adsorption, Surface Area, and Porosity,
2 nd. ed., Academic Press, 1991.
Grudemo, A., Development of Strength Properties of Hydrating Cement
Pastes and Their Relation to Structural Features, Proc. Symp. Some
Recent Res. Cem. Hydration, Denmark, 1975.
Grzybowski, M., and Shah, S.P., Shrinkage Cracking of Fiber Reinforced
Concrete, ACI Mat. J. 87 (2), pp. 138, 1990.
Guenot, I., Torrenti, J.M., and Laplante, P., Stresses in Early-age Con
crete: Comparison of Different Creep Models, ACI Mat. J. 93 (3), pp.254,
1996.
Hansen, W., and Almudaiheem, J., Pore Structure of Portland Cement
Measured by Nitrogen Sorption and Mercury Intrusion Porosymetry,
Proc. MRS Symp., pp.105, 1987.
Hansen, W., Drying Shrinkage Mechanisms in Portland Cement Paste,
J. Am. Cer. Soc. 70 (5), pp.323, 1987.
Hansen, W., Borgnakke, C., and Pane, I., Phenomenological Kinetics of
Portland Cement Hydration: II. Temperature Effects, Activation Energy,
and Other Influencing Factors, unpulished manuscript.
Hansen, P.F., and Pedersen, E.J., Maleinstrument til kontrol of betons
haerdning, Nordisk Betong (1 ), pp. 21, 1977.
Hanehara, S., Tomosawa F., Kobayakawa, M., and Hwang, K.R., Effects
of Water/Powder Ratio, Mixing Ratio of Fly Ash, and Curing Tempera
ture on Pozzolanic Reaction of Fly Ash in Cement Paste, Cem Con. Res.
31 (2), pp.31., 2001.
Helmuth, R.A., and Turk, D.H., Elastic Moduli of Hardened Portland
Cement and Tricalcium Silicate Pastes: Effect of Porosity, PCA Res.
Bull., 1966.
2 3 5
[Hel.2]
[Hel.3]
[Hor.l]
[Hua.l]
[Huc.l]
[Ml ]
[Jen.l]
[Jen.l]
[Kav.l]
[Kha.l]
[Khi.l]
[Kje-l]
[Kje.2]
Helmuth, R.A., Fly Ash in Cement and Concrete, PCA Res. Dev. Lab.,
1987.
Hellmich, C., Sercombe, J., Ulm, F-J, and Mang, H., Modeling of Early-
age Creep of Shotcrete. II: Application to Tunneling, ASCE J. Eng. Mech.
126 (3), pp.292, 2000.
Hori, A., Morioka, M., Sakai, E., and Daimon, M., Influence of Expan
sive Additives on Autogeneous Shrinkage, Proc. JCI Workshop Autog.
Shrink, of Cone., Japan, E & FN Spon, pp.187, 1999.
Hua, C., Acker, P., and Ehrlacher, A., Analysis and Models of the Au
togeneous Shrinkage of Hardening Cement Paste, Cem. Con. Res. 25,
pp.1457, 1995.
Huckfeldt, J., Duddeck, H., and Ahrens, H., Numerical Simulation of
Crack Avoiding Measure, RILEM Proc. No. 25, Thermal Crack. Cone,
at Early Ages, E & FN Spon, pp.255, 1995.
Illston, J.M., The Creep of Concrete Under Uniaxial Tension, Mag. of
Concrete Res. 17 (100), 1965.
Jensen, O.M., and Hansen, P.F., Autogenous Deformation and Change
of the Relative Humidity in Silica Fume-Modified Cement Paste, ACI
Mat. J. 93 (6 ), pp.539, 1996.
Jensen, O.M., Hansen, P.F., Lachowski, E.E., and Glasser, F.P., Clinker
Mineral Hydration at Reduced Relative Humidities, Cem. Cone. Res. 29
(9), pp.1505, 1999.
Kaviany, M., Principles of Heat Transfer in Porous Media, 2nd Ed.,
Springer-Verlag, 1995.
Khan, A.A., Cook, W.D., and Mitchell, D., Creep, Shrinkage, and Ther
mal Strains in Normal, Medium, and High-strength Concretes During
Hydration, ACI Mat. J. 94 (2), pp.156, 1997.
Khinast, J., Krammer, G.F., Brunner., Ch., and Staudinger, G., De
composition of Limestone: the Influence of CO2 and Particle Size on the
Reaction Rate, Chem. Eng. Sci. 51 (4), 1996.
Kjellsen, K.O., and Detwiler, R.J., Reaction Kinetics of Portland Cement
Mortar Hydrated at Different Temperatures, Cem. Cone. Res. 2 2 (1 ), pp.
112, 1990.
Kjellsen, K.O., and Detwiler, R.J., Later-age Strength Prediction by a
Modified Maturity Model, ACI Mat. J. 90, pp.220, 1993.
236
[Koe.l]
[Kon.l]
[Kov.l]
[Law.l]
[Lit.l]
[Mae.l]
[Meh.l]
[Mid.l]
[Mik.l]
[Mik.2]
[Min.l]
[Moh.l]
[Mos.l]
[Nav.l]
Koenders, E.A., Simulation of Volume Changes in Hardening Cement-
Based Materials, Ph.D Thesis, Delft Univ. Tech., 1997.
Kondo, R., and Ueda, S., Kinetics and Mechanism of Hydration of Ce
ments, Proc., 5th Int. Symp. on Chemistry of Cement, Vol. 2, Tokyo,
1969.
Kovler, L., Testing System for Determining the Mechanical Behavior of
Early Age Concrete Under Restrained and Free Uniaxial Shrinkage, Mat.
and Struc. 27, pp.324, 1994.
Lawrence, D., The Constitution and Specification of Portland Cements,
Leas Chem. Cem. Cone. 4th ed. (Peter C. Hewlett, ed.), pp.131, 1998.
Litvan, G.G., Variability of Nitrogen Surface Area of Hydrated Cement
Paste, Cem. Con. Res. 6 , pp.139, 1976.
Maekawa, K., Chaube, R-, and Kishi, T., Modeling of Concrete Per
formance: Hydration, Microstructure Formation, and Mass Transport,
London, Routledge, 1999.
Mehta, P.K., and Monteiro, P.J., Concrete, 2nded., Prentice-Hall, 1993.
Midgley, H.G., The Determination of Calcium Hydroxide in Set Portland
Cements, Cem. Con. Res. 9 ( 1 ), pp.77, 1979.
Mikhail, R.S., and Abo-Elenein, S., Studies on Water and Nitrogen Ad
sorption on Hardened Cement Pastes: I. Development of Surface Area in
Low Porosity Pastes, Cem. Con. Res. 2, pp.401, 1972.
Mikhail, R.S., Abo-Elenein, S., and Gabr, N.A., Hardened Slag-Cement
Pastes of Various Porosities: I. Compressive Strength, Degree of Hydra
tion, and Total Porosity, J. Appl. Chem. Biotec. 234, pp.735, 1974.
Mindess, S., and Young, J.F., Concrete, Prentice-Hall, Englewood
Cliffs, N. J., 1981.
Mohr, P., Hansen, W., and Mohammed, A.R., Thermal Stresses in Con
crete Pavements at Early Ages, Proc. Canadian Soc. Civ. Eng. Int. Conf.
Eng. Mat., Ottawa, 1997.
Mostafa, N.Y., El-Hemaly, S.A., Al-Wakeel, E.I., and El-Korashy, S.A.,
and Brown, P.W., Characterization and Evaluation of the Pozzolanic Ac
tivity of Egyptian Industrial By-products: I: Silica Fume and Dealumi-
nated Kaolin, Cem. Con. Res. 31 (3), pp.467, 2001.
Navi, P., and Pignat, C., Simulation of Cement Hydration and the Con
nectivity of Capillary Pore Space, Adv. Cem. Based Mat. 4, pp.58, 1996.
237
[Nev.l]
[Nev.2 ]
[Odl.l]
[Olu.l]
[Onk.l]
[Par.l]
[Par.2]
[Pic.l]
[Pow.l]
[Pow.2]
[Ram.l]
[Ram.2]
[Ras.l]
[Rol.l]
Neville, . M., Properties of Concrete, 3rd ed., Pitman Books Ltd., Lon
don, 1983.
Neville, A.M., Dilger, W.H., and Brooks, J.J., Creep of Plain and Struc
tural Concrete, Construction Press, New York, 1983.
Odler, I., and Koster, H., Investigation of the Structure of Fully Hydrated
Portland Cement and Tricalcium Silica Pastes. III. Specific Surface Area
and Permeability, Cem. Con. Res. 21, pp.975, 1991.
Oluokun, F.A., Burdette, E.G., and Deatherage, J.H., Elastic Modulus,
Poissons Ratio, and Compressive Strength Relationships at Early Ages
ACI Mat. J., Jan.-Feb., 1991.
Onken, P., and Rostasy, F.S., A Practical Planning Tool for the Simula
tion of Thermal Stresses and for the Prediction of Early Thermal Cracks
in Massive Concrete Structures, RILEM Proc. No. 25 Thermal Crack.
Cone, at Early Ages, E & FN Spon, 1995.
Parrot, L.J., Geiker, M., Gutteridge, W.A., and Killoh, D., Monitoring
Portland Cement Hydration: Comparison of Different Methods, Cem.
Cone. Res., Vol. 20, 1990.
Parrot, L.J., Killoh, D.C., and Patel, R.G., Cement Hydration Under
Partially Saturated Curing Conditions, 8 t h Int. Cong, on Chemistry of
Cement, Rio de Jeneiro, 1986.
Pickett, G., The Effect of Change in Moisture Content on The Creep of
Concrete Under a Sustained Load, ACI Journal 38, 1942.
Powers, T.C., A Discussion of Cement Hydration in Relation to the Cur
ing of Concrete, Proc. Highway Res. Board, No. 27, 1947.
Powers, T.C., and Brownyard, T. L., Studies of Physical Properties of
Hardened Portland Cement, PCA Res. Bull. No. 22, 1948.
Ramachandran, V.S., Applications of Differential Thermal Analysis,
Chemical Publishing Company Inc., New York, 1969.
Ramachandran, S.V., Feldman, R.F., and Beaudoin, J.J., Concrete Sci
ence, Treatise on Current Research, Heyden & Son Ltd., 1981.
Rastrup, E., Heat of hydration of concrete, Mag. Cone. Res., Sept. 1954,
pp.79, 1954.
Roll, R., Long Time Creep Recovery of Highly Stressed Concrete Cylin
ders ACI SP-9, Symposium on Creep, Detroit, 1964.
238
[Ros.l]
[Ros.2 ]
[Sar.l]
[Sch.l]
[Scr.l]
[Scr.2 ]
[Sel.l]
[Ser.l]
[Set.l]
[Set. 2]
[Shi.l]
[Sor.l]
[Spr.l]
Rostasy, F.S., Session D: Determination and Modelling of Mechanical
Properties, RILEM Comtt. 119 Tech. Recomm. on Avoidance of Thermal
Crac. Early Age Cone. (Draft), 1994.
Rostasy, F.S., Tanabe, T., and Laube, M., Assessment of External Re
straint, RILEM Rep. No. 15, Prev. Thermal Crac. Cone, at Early Ages,
E & FN Spon, 1998.
Sarigaphuti, M., Shah, S.P., Vinson, K.D., Shrinkage Cracking and Dura
bility Characteristics of Cellulose Fiber Reinforced Concrete, ACI Mat.
J., 90 (4), pp.309, 1993.
Scherer, G.W., Dilatation of Porous Glass, J. Am. Cer. Soc. 69 (6 ),
pp.473, 1986.
Scrivener, K.L., The Microstructure of Concrete, Mat. Sci. of Cone. I,
American Ceram. Soc. (Ed. J. Skalny), 1989.
Scrivener, K.L., and Gartner, E.M., Microstructural Gradients in Ce
ment Paste Around Aggregate in Particles, MRS Proc.: Bonding in Cem.
Composites (Ed. S. Mindess and S.P. Shah), Pittsburgh., No. 114, 1988.
Sellevold, E., and Bjontegaard, O., High Performance Concrete: Early
Volume Change and Cracking Tendency, RILEM Proc. No. 25, Thermal
Crack. Cone, at Early Ages, E & FN Spon, 1995.
Sercombe, J., Hellmich, C., Ulm, F. J, and Mang, H., Modeling of Early-
age Creep of Shotcrete. I: Model and Model Parameters, ASCE J. Eng.
Mech. 126 (3), pp.284, 2000.
Setzer, M.J., Interaction of Water with Hardened Cement Paste, Ce
ramic Transaction 16, S. Mindess (Ed.), Am. Cer. Soc., 1991.
Setzer, M.J., Einfluss des Wassergehalts auf die Eigenschaften des er-
hateten Betons, Deutscher Ausschuss Fur Stahlbeton, Heft 280, pp.43,
1978.
Shi, C., Pozzolanic Reaction and Microstructure of Chemical Activated
Lime-fly Ash Pastes, ACI Mat. J. 95 (5), pp.537, 1998.
Soroka, I., Portland Cement Paste and Concrete, MacMillan Press,
1979.
Springenschmid, R., Breitenbucher, R., and Mangold, M., Development
of Thermal Cracking Frame and the Temparature Testing Machine, In
Thermal Crac. Early Ages, Proc. RILEM Symp., R. Springenschmid Ed.,
Munich, pp.137, 1994.
239
[Svva.l]
[Tan.l]
[Tap.lj
[Tap.2 ]
[Thi.ll
[Tim.l]
[Tsc.l]
[Ulm.l]
[Ume.l]
[Ver.l]
[Vol.l]
[Wan.l]
[Wes.l]
Swamy, R.N., and Stavrides, H., Influence of Fiber Reinforcement on
Restrained Shrinkage and Cracking, ACI Journal 76, pp.443, 1979.
Tangtermsirikul, S., Effect of Chemical Composition and Particle Size of
Fly Ash on Autogeneous Shrinkage of Paste, Proc. JCI Workshop Autog.
Shrink, of Cone., Japan, E & FN Spon, pp.175, 1999.
Taplin, J.H., A Method for Following the Hydration Reaction in Portland
Cement Paste, Australian J. of App. Science, Vol. 10, 1959.
Taplin, J.H., The Temperature Dependence of the Hydration Rate of
Portland Cement Paste, Australian J. of App. Science, Vol. 13, No. 2,
1962.
Thielen, G., and Hintzen, W., Investigation of Concrete Behavior Under
Restraint with a Temperature-Stress Test Machine, RILEM Proc. No. 25
Thermal Crack. Cone, at Early Ages, E FN Spon, 1995.
Timoshenko, S.P., and Goodier, J.N., Theory of Elasticity. 3rd ed.,
McGraw-Hill, New York, 1970.
Tschoegl, N., The Phenomenological Theory of Linear Viscoelastic Be
havior : an Introduction, Sprigner-Verlag, 1989.
Ulm, F-J, and Coussy, O., Couplings in Early-age Concrete: from Ma
terial Modeling to Structural Design, Int. J. Solid Struc. 35 (31-32),
pp.4295, 1998.
Umehara, H., Uehara T., Iisaka, T., and Sugiyama, A., Effects of Creep
in Concrete at Early Ages on Thermal Stress, RILEM Proc. No. 25,
Thermal Crack. Cone, at Early Ages, E & FN Spon, pp.79, 1995.
Verbeck, G., and Helmuth, R.A., Structures and Physical Properties of
Cement Pastes, Proc. 5th Int. Symp. Chem. Cement Vol.III, Tokyo, pp.l,
1968.
Volkl, J.J., Beddoe, R.E., and Setzer, M.J., The Specific Surface of Hard
ened Cement Paste by Small-Angle X-Ray Scattering, Effect of Moisture
Content and Chloride, Cem Con. Res. 17, pp.81, 1987.
Wang, C., and Dilger, W.H., Prediction of Temperature Distribution in
Hardening Concrete, RILEM Proc. No. 25, Thermal Crack. Cone, at
Early Ages, E FN Spon, pp.2 1 , 1995.
Westman, G., Basic Creep and Relaxation of Young Concrete, RILEM
Proc. No. 25, Thermal Crack. Cone, at Early Ages, E & FN Spon, pp.87,
1995.
240
[Win.l]
[Win. 2]
[Wit.l]
[Wit.2]
[XL1]
[Xi-2]
[Yat.l]
[You.l]
[You.2]
[Yu.l]
Winslow, D.N., and Diamond, S., Specific Surface of Hardened Portland
Cement Paste as Determined by Small-Angle X-Ray Scattering, J. Am.
Cer. Soc. 57 (5), pp.193, 1974.
Winslow, D.N., and Diamond, S., A Mercury Porosimetry Study of the
Evolution of Porosity in Portland Cement, ASTM J. Mat. 5 (3), pp.564,
1970.
Wittmann, F.H., Interaction of Hardened Cement Paste and Water, J.
Am. Cer. Soc. 56 (8), pp.409, 1973.
Whittman, F. H., Roelfstra, P. E., Mihashi, H., Huang, Y. Y., Zhang, X.,
and Nomura, N., Influence of Age of Loading, Water-Cement Ratio, and
Rate of Loading on Fracture Energy of Concrete, Mat. Struc. 20 (116),
1987.
Xi, Y., Bazant, Z.P., and Jennings, H.M., Moisture Diffusion in Cemen-
titious Materials: Adsorption Isotherms, J. Adv. Cem. Based Mat. 1,
pp.248, 1994.
Xi, Y., Bazant, Z.P., and Jennings, H.M., Moisture Diffusion in Cementi-
tious Materials: Moisture Capacity and Diffusivity, J. Adv. Cem. Based
Mat. 1, pp.248, 1994.
Yates, D.J., The Expansion of Porous Glass on the Adsorption of Non-
Polar Gases, Proc. Royal Soc. London Ser. A (A244), pp.526, 1954.
Young, J.F., and Hansen, W., Volume Relationships for C-S-H Formation
Based on Hydration Stoichiometry, Mat. Res. Soc. Symp. Proc. 85, pp.313,
1987.
Young, J.F, Physical Mechanisms and their Mathematical Descriptions,
in Math. Modeling of Creep and Shrinkage, Z.P. Bazant Ed.. John Wiley
& Son, New York, 1988.
Yu, B-J., and Ansari, F., Method and theory for nondestructive deter
mination of fracture energy in concrete structure, ACI Mat. J. 93 (5),
pp. 602, 1996.

GHFGHF

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

GHFGHF

Uploaded by

Copyright:

Available Formats

INFORMATION TO USERS

......... 'M .............. . .

You might also like