Creep Behavior of GP

Creep Behaviour of
Geopolymer Concrete
Chi Hou Un
Submitted in total fulfilment of the requirements of the degree of
Doctor of Philosophy
July 2017
Faculty of Science, Engineering and Technology
Swinburne University of Technology, Melbourne, Australia

ii
Abstract
Ordinary Portland cement concrete (OPCC) is widely utilised as a major construction
material over the past century. However, the manufacturing of ordinary Portland
cement (OPC) produces a great amount carbon dioxide (CO2) into the atmosphere,
which is one of greenhouse gases contributing to global warming. Geopolymer
concrete (GPC) has been researched during the past few decades as an alternative to
sustainable construction materials, which can minimise CO2 emission for its use of
industry by-products. GPC can be achieved by activation of the alumina- and silica-
containing solid precursors, such as blast furnace slag, fly ash and metakaolin, by an
alkali activator. Past research on GPC shows that GPC can be suitable for structural
application, with workable slump, and comparable grade of strength to OPCC.
However, there is little research on creep of GPC, which is essential to determine the
serviceability of concrete structures, as creep is a phenomenon that increases strain
under sustained load over time. Therefore, research on creep of GPC is the focus of
this thesis.
Standard creep test on a slag-based GPC and OPCC shows that GPC exhibited a
lower creep than the OPCC in the early age but the creep rate of GPC did not reduce
quickly and GPC yielded significant higher creep effect than OPCC by 80% in later
ages. However, such difference is reduced to 35% when drying creep effect is not
considered. This indicates that drying has a big impact on creep of GPC. It was also
found that the creep behaviour would be similar to OPCC as more mature and higher
strength GPC showed less creep potential. Based on these results, creep models of
GPC can be developed successfully.
The microstructure of GPC and OPCC was investigated to understand its influence
on creep. Analysis results show that GPC contained more amorphous phases, less
porosity and more pores in the mesopores range than OPCC. The fine pore network
slowed down the rate of micro-diffusion of GPC under creep load, and thus the early
age creep was low in GPC. Because the amorphous phases are thermodynamic
unstable, it increases the creep response of GPC. After the long period of creep test,
alternation of microstructure of GPC subjected to sustained stress was observed and
more pores shifted to below 2 nm, and this did not happen on OPCC. The alternation
iii
in microstructure would be related to the slippage of phases, and thus explain the
creep behaviour of GPC in later ages.
Two types of GPC beams of 4 m and 6.5 m each were tested under sustained loads
for long-term deflection, to stimulate real-life structures. The long-term deflection of
the GPC beams was higher than that of the OPCC beams from computer stimulation,
but it can be controlled by careful reinforcement detailing. However, it was found
that the GPC beams were affected by seasonal change as more creep deflection was
observed during summer period. This sensitivity of creep of GPC to temperature is
unique characteristic that was not reported previously. Therefore, further research
into the temperature sensitivity of creep in GPC was conducted.
A new accelerated test method was developed to investigate creep of paste and
mortar specimens of geopolymer and OPC under various conditions. The specimens
are of a thin plate, or called “mini-beam”. It was found that the creep of geopolymer
paste (GPP) increased more significantly with reducing relative humidity than OPC
paste (OPCP). Higher creep was observed on GPP tested at 50 °C than that tested at
23 °C. The effect of temperature on compressive creep of GPP was further
investigated up to 100 °C on the cylinder specimens. The 1-day creep of GPP
increases significantly at temperature above 80 °C. This indicates the temperature
dependency of geopolymer.
Therefore, the mechanism of creep of GPC is significantly different from the OPCC.
The sustained stress on GPC causes more slippage of phases, and alternation of
microstructure to denser phases, which results in more creep deformation than
OPCC. In addition, migration of water within the pore network due to drying
accelerates this process, causing more drying creep in GPC. GPC is also more
sensitive to temperature than OPCC, which is caused by increasing visco-flow
behaviour and further geopolymerisation subjected to high temperature.
iv
Declaration
I hereby certify that this thesis entitled “Creep Behaviour of Geopolymer Concrete”
contains no material which has been accepted for the award to the candidate of any
other degree or diploma, except where due reference is made in the text of the
examinable outcome. To the best of my knowledge, the thesis contains no material
previously published or written by another person except where due reference is
made in the text of the examinable outcome. Where the work is based on joint
research or publications, discloses the relative contributions of the respective workers
or authors.
_____________________________
Chi Hou Un
July 2017
v
Acknowledgement
This PhD study is done with a lot of effort and time, and especially guidance and
support from many people. I would like to express my sincere appreciation to all of
them.
First of all, I would like to express my sincere thanks to my research supervisor,

Prof. Jay Sanjayan, for providing the opportunity to PhD study, the guidance and
support throughout my PhD study. This thesis would not be achieved without the
invaluable guidance and advice from him.
Secondly, I would like to extend my gratitude to my research fellow, Dr. Rackel San
Nicolas from the University of Melbourne, for guidance and assistance in the field of
chemistry. Since I am from the structural engineering background, I would not
achieve the knowledge of chemistry without her help. I also appreciate that she
allows me to access her geopolymer laboratory and conduct experiments. I would
like to thank the laboratory technician, Laura Jukes, for the assistance in sample
preparation. I would like to thank Prof. Jannie van Deventer for his advice and
comment on our publications.
I would like to express my sincere thanks to the sponsors of my scholarship, Nan and
Elizabeth Brown, for their generous financial support. This allows me to concentrate
in the research without worry.
I am grateful for the support from the staff of the Smart Structural Laboratory in
Swinburne University of Technology, especially to the laboratory manager, Michael
Culton, and the test engineers Kia Rasekhi, Sanjeet Chandra and Kevin Nievaart. I
would like to acknowledge the members of geopolymer concrete research group in
Swinburne University of Technology, in particular Chandani Tennakoon, Ahmed
Graytee and Ghidak Al-Bayati, for the encouragement and discussion.
I would like to thank Aurora Construction Materials, for supplies of the large volume
of geopolymer concrete for the full-scale beam tests, and the aggregates materials for
the laboratory concrete mixes.
vi
Finally, I would like to thank my family for the encouragement and advice in making
the decision of PhD study and the ongoing support during my study. And of course
to my fiancée, Agatha Yii, for her support, encouragement and patient during my
lengthy study period.
vii
Publications
Published journal article:
Un, C.H., Sanjayan, J.G., San Nicolas, R. and van Deventer, J.S.J. (2015),
“Predictions of long-term deflection of geopolymer concrete beams”,
Construction and Building Materials, 94, pp. 10-19.
Journal articles under preparation:
Un, C.H., San Nicolas, R., and Sanjayan, J.G., “Relationship of microstructure and
creep behaviour of geopolymer concrete and ordinary Portland cement
concrete”
Un, C.H., Sanjayan, J.G., San Nicolas, R., “Comparison of creep behaviour of
geopolymer concrete and ordinary Portland cement concrete”
Un, C.H., Sanjayan, J.G., San Nicolas, R., “Creep deformation of geopolymer RC
beams”
Un, C.H., Sanjayan, J.G., San Nicolas, R., “Investigation of creep behaviour of
geopolymer and ordinary Portland cement pastes and mortars by mini-beam
test”
Un, C.H., San Nicolas R., Sanjayan, J.G., “Thermal effect on creep of geopolymer
paste up to 100 °C”
Published conference papers:
Un, C.H., Sanjayan, J.G., San Nicolas, R. and van Deventer, J.S.J. (2014),
“Investigation of long term behaviour of composite geopolymer concrete
beams under sustained loads”, 23rd Australasian Conference on the Mechanics
viii
of Structures and Materials (ACMSM23), Byron Bay, Australia, 9-12th
December 2014.
Un, C.H., Sanjayan, J.G., San Nicolas, R. and van Deventer, J.S.J. (2015), “Mini-
beam test for assessing the creep trend of paste, mortar and concrete”,
ConCreep 10 - Mechanics and physics of creep, shrinkage, and durability of
concrete and concrete structures, Vienna, Austria, 21-23rd September 2015.
Un, C.H., Sanjayan, J.G., San Nicolas, R. and van Deventer, J.S.J. (2015), “Creep
behaviour of geopolymer”, 10th International Conference on the Physical
Properties and Application of Advanced Materials (ICPMAT2015), Chiang
Mai, Thailand, 17-21st November 2015.
Un, C.H., Sanjayan, J.G., and San Nicolas, R. (2016), “Comparison of creep
behaviour of geopolymer concrete and OPC concrete”, 24th Australasian
Conference on the Mechanics of Structures and Materials (ACMSM24), Perth,
Australia, 6-9th December 2014.
ix
Table of Contents
Abstract ...................................................................................................................iii
Declaration ............................................................................................................... v
Acknowledgement ................................................................................................... vi
Publications ........................................................................................................... viii
List of figures ......................................................................................................... xv
List of tables .......................................................................................................... xxi
List of symbols ..................................................................................................... xxii
Abbreviation index ............................................................................................. xxvii
Chapter 1 : Introduction ............................................................................................ 1
1.1 Background .................................................................................................... 1
1.2 Research significants ...................................................................................... 2
1.3 Scope of thesis ................................................................................................ 3
1.4 Thesis organisation ......................................................................................... 4
Chapter 2 : Literature review .................................................................................... 6
2.1 Overview ........................................................................................................ 6
2.2 Background of ordinary Portland cement concrete and geopolymer concrete .. 7
2.2.1 Ordinary Portland cement concrete........................................................... 7
2.2.2 Geopolymer concrete ............................................................................... 9
2.2.3 Blended cement concrete........................................................................ 14
2.3 Microstructures ............................................................................................. 15
2.3.1 Hardened OPC ....................................................................................... 15
2.3.2 Hardened geopolymer ............................................................................ 17
2.3.3 Differences in microstructure between OPC and geopolymer ................. 18
2.4 Relationship between microstructure and properties of concrete .................... 21
2.4.1 Strength ................................................................................................. 21
2.4.2 Permeability and durability .................................................................... 23
2.4.3 Shrinkage ............................................................................................... 29
2.4.4 Analysis for pore characteristic .............................................................. 30
2.5 Creep of concretes ........................................................................................ 32
2.5.1 Creep ..................................................................................................... 32
2.5.2 Creep of OPCC ...................................................................................... 33
2.5.3 Creep of GPC ......................................................................................... 40
x
2.5.4 Creep of blended cement concrete ..........................................................42
2.5.5 Creep at high temperature .......................................................................44
2.6 Creep models and structural performance ......................................................46
2.6.1 Early development of creep model for OPCC .........................................46
2.6.2 Creep models in standards ......................................................................51
2.6.3 Structural design .....................................................................................53
2.7 Effect of creep on concrete structures ............................................................57
Chapter 3 : Methodology and experimental program ...............................................59
3.1 Introduction...................................................................................................59
3.2 Materials .......................................................................................................59
3.2.1 Binder materials .....................................................................................59
3.2.2 Alkali activator .......................................................................................61
3.2.3 Aggregates .............................................................................................61
3.3 Sample preparation ........................................................................................63
3.3.1 Mix design .............................................................................................63
3.3.2 Mixing and casting procedures ...............................................................64
3.3.3 Curing ....................................................................................................65
3.4 Material property tests ...................................................................................66
3.4.1 Compressive strength and elastic modulus tests ......................................66
3.4.2 Flexural tensile strength test ...................................................................68
3.4.3 Creep test ...............................................................................................69
3.4.4 Shrinkage tests .......................................................................................71
3.5 Tests for microstructure characteristic ...........................................................72
3.5.1 Nitrogen gas adsorption test ....................................................................72
3.5.2 Other analysis .........................................................................................74
3.6 Beam tests .....................................................................................................75
3.7 Thermal creep tests on paste specimens .........................................................76
Chapter 4 : Test results of GPC and OPCC and creep prediction of GPC ................77
4.1 Introduction...................................................................................................77
4.2 Fresh property of GPC and OPCC .................................................................79
4.3 Strengths of GPC and OPCC .........................................................................80
4.3.1 Compressive strength .............................................................................80
4.3.2 Flexural tensile strength ..........................................................................81
4.3.3 Elastic modulus and Poisson’s ratio ........................................................84
xi
4.4 Shrinkage of GPC and OPCC........................................................................ 85
4.5 Creep of GPC and OPCC .............................................................................. 89
4.5.1 Elastic modulus obtained from creep test ................................................ 90
4.5.2 Comparison of creep behaviour between GPC and OPCC ...................... 91
4.5.3 Influence of age of loading and strength on creep of GPC ...................... 94
4.5.4 Creep recovery ....................................................................................... 96
4.6 Creep models for GPC .................................................................................. 97
4.6.1 Prediction by standard creep models....................................................... 97
4.6.2 Recommended modification to standard creep models for GPC ............ 103
4.6.3 Comparison of the modified creep models ............................................ 110
4.7 Concluding remarks .................................................................................... 117
Chapter 5 : Microstructure and creep behaviour of GPC and OPCC ...................... 119
5.1 Introduction ................................................................................................ 119
5.2 Microstructure of GPC and OPCC .............................................................. 120
5.2.1 XRD .................................................................................................... 120
5.2.2 TGA..................................................................................................... 122
5.2.3 BSEM .................................................................................................. 123
5.2.4 Pore characteristic ................................................................................ 125
5.2.5 Difference between microstructure of GPC and OPCC ......................... 126
5.3 Change of microstructure after creep test .................................................... 126
5.3.1 XRD .................................................................................................... 126
5.3.2 TGA..................................................................................................... 128
5.3.3 BSEM .................................................................................................. 129
5.3.4 Pore characteristic ................................................................................ 131
5.4 Relationship between microstructure and creep behaviour........................... 134
Chapter 6 : Assessment of creep behaviour with mini-beam test and microstructural
study..................................................................................................................... 137
6.1 Introduction ................................................................................................ 137
6.2 Experimental program................................................................................. 138
6.2.1 Test details ........................................................................................... 138
6.2.2 Assessment of creep coefficient from mini-beam test ........................... 142
6.3 Experimental results of mini-beam test ....................................................... 143
6.3.1 Compressive strength ........................................................................... 143
6.3.2 Flexural capacity .................................................................................. 145
xii
6.3.3 Creep response of mini-beams of paste specimens ................................ 147
6.3.4 Creep response of mini-beams of mortar specimens .............................. 150
6.3.5 Thermal effect on mini-beam test ......................................................... 152
6.3.6 Shrinkage obtained from mini-beam test ............................................... 154
6.3.7 Autogenous shrinkage .......................................................................... 158
6.3.8 Relationship between the paste and mortar mini-beams ........................ 159
6.4 Analysis of pore structure of paste mini-beams ............................................164
6.4.1 Pore structure before mini-beam test ..................................................... 164
6.4.2 Change of pore structure after mini-beam test ....................................... 167
6.5 Development of creep model for mini-beam test.......................................... 172
6.5.1 Background .......................................................................................... 172
6.5.2 Constitutive Equations .......................................................................... 173
6.5.3 Continuity Equations ............................................................................175
6.5.4 Application of sustained loads on mini-beam ........................................176
6.5.5 Viscoelastic behaviour .......................................................................... 179
6.5.6 Estimation of permeability from pore size distribution .......................... 181
6.5.7 Validation of creep model..................................................................... 184
Chapter 7 : Long-term deflection of GPC beams ................................................... 197
7.1 Introduction................................................................................................. 197
7.2 Experimental program ................................................................................. 198
7.2.1 Beam type 1: Reinforced concrete beam ............................................... 198
7.2.2 Beam type 2: Composite steel-concrete beam ....................................... 199
7.2.3 Structural analysis ................................................................................ 202
7.3 Results of long-term beam tests ................................................................... 202
7.3.1 GPC beam type 1 .................................................................................. 202
7.3.2 GPC beam type 2 .................................................................................. 213
7.4 Prediction of long-term deflection of GPC beams ........................................218
7.4.1 GPC beam type 1 .................................................................................. 218
7.4.2 GPC beam type 2 .................................................................................. 227
Chapter 8 : Thermal effect on creep behaviour of GPP and OPCP ......................... 238
8.1 Introduction................................................................................................. 238
8.2 Experimental program ................................................................................. 239
xiii
8.2.1 Test setup ............................................................................................. 239
8.2.2 Hot strength ......................................................................................... 241
8.2.3 Thermal creep and shrinkage ................................................................ 241
8.2.4 Free thermal strain ............................................................................... 242
8.2.4 Transitional thermal creep .................................................................... 242
8.2.5 XRD .................................................................................................... 242
8.3 Test results.................................................................................................. 243
8.3.1 Compressive strength and hot strength ................................................. 243
8.3.2 Thermal creep ...................................................................................... 244
8.3.3 Transitional thermal creep and free thermal strain ................................ 249
8.3.4 XRD .................................................................................................... 251
8.4 Effect of temperature on creep behaviour of GPP and OPCP ....................... 253
8.5 Feasibility for accelerated creep test ............................................................ 255
Chapter 9 : Summary, conclusions and recommendations ..................................... 258
9.1 Summary .................................................................................................... 258
9.2 Conclusions ................................................................................................ 262
9.3 Recommendations for future research ......................................................... 263
References ............................................................................................................ 265
xiv
List of figures
Figure 2.1: Production of geopolymer .....................................................................10
Figure 2.2: Conceptual model for geopolymerisation (Source: Duxson et al. (2007)
[17])........................................................................................................................12
Figure 2.3: Diagrammatic model for hydrated OPC (Source: Mehta and Monteiro
(2006) [48]) ............................................................................................................16
Figure 2.4: 1.4 nm tobermorite structure of C-S-H (Adapted from Richardson (2008)
[50])........................................................................................................................17
Figure 2.5: Illustration of constraints in CSTM (Source: Myers et al. (2013) [51]) ..18
Figure 2.6: Ternary plot of binder gel compositions of geopolymer (noted as
AA100S AA50S/50FA) and OPC (noted as 100OPC) (Source: van Deventer, et al.
(2015) [56]) ............................................................................................................19
Figure 2.7: Pore size distribution of geopolymer and OPC (Source: Collins and
Sanjayan (2000) [60]) .............................................................................................20
Figure 2.8: Strength-porosity relationship in GPM and OPCM (PC denotes OPC, NS,
NC and NH denote geopolymer with different activators) (Source: Shi (1996) [67])
...............................................................................................................................22
Figure 2.9: Typical development of creep in OPCC (adapted from Bažant and Kim
(1978) [117]) ..........................................................................................................34
Figure 2.10: Transitional thermal creep: (a) GPP (b) OPCP (source: Pan et al. (2014)
[165] .......................................................................................................................45
Figure 2.11: Spring-dashpot models for viscoelastic materials: (a) Kelvin chain
model; and (b) Maxwell chain model ......................................................................46
Figure 3.1: Dry raw binder materials .......................................................................60
Figure 3.2: Alkali activator in this study (sodium metasilicate) ...............................61
Figure 3.3: Grading of fine aggregate ......................................................................62
Figure 3.4: Dry aggregates for concrete and mortar mixes .......................................62
Figure 3.5: Rotating pan mixer for concrete mixing ................................................64
Figure 3.6: Sealed specimens: (a) Application of MasterKure® CC 1315WB; (b)
resin coating and plastic layer .................................................................................66
Figure 3.7: Cylinder end grinder .............................................................................67
Figure 3.8: Tests for determination of elastic modulus: (a) Concrete (b) Paste .........68
xv
Figure 3.9: Test for determination of modulus of rupture ........................................ 68
Figure 3.10: Test instruments for creep test: (a) Spring supported load frame and
hydraulic jack; (b) demec gauge ............................................................................. 69
Figure 3.11: Companion specimens for shrinkage effect in creep test...................... 70
Figure 3.12: Measurement on creep specimen ......................................................... 71
Figure 3.13: Vertical comparator ............................................................................ 71
Figure 3.14: Test equipment for autogenous shrinkage test on pastes and mortar: (a)
Dilatometer bench with length gauge, (b) Vicat needle ........................................... 72
Figure 3.15: Test instruments for nitrogen gas adsorption test ................................. 73
Figure 4.1: Slump test carried out on GPC .............................................................. 79
Figure 4.2: Modulus of rupture of GPC at different ages (Bath cured: Bath cured at
all time; Drying room: Bath cured until 7 days then moved to drying condition of 23
°C and 50% RH; Sealed: Sealed after demould) ...................................................... 82
Figure 4.3: Typical failure in flexural tensile strength test of GPC specimens ......... 83
Figure 4.4: Stress vs. Vertical and lateral strains of GPC ........................................ 84
Figure 4.5: Prism specimens for shrinkage test ....................................................... 85
Figure 4.6: Shrinkage strain of GPC between batches ............................................. 87
Figure 4.7: Total shrinkage and autogenous shrinkage of GPC and OPCC .............. 87
Figure 4.8: The change of weight of shrinkage specimens of GPC and OPCC related
to before test condition ........................................................................................... 89
Figure 4.9: Creep coefficient of GPC and OPCC: (a) up to age of 500 days, (b) in
first three months of loading ................................................................................... 92
Figure 4.10: Total creep and basic creep coefficients of GPC and OPCC (Total creep
from specimens marked as solid lines, while basic creep from specimens marked as
broken lines.) .......................................................................................................... 94
Figure 4.11: Creep coefficient of GPC loaded at different ages ............................... 95
Figure 4.12: Creep recovery of GPC and OPCC ..................................................... 97
Figure 4.13: Prediction of creep coefficient with standard creep models ................. 99
Figure 4.14: Creep coefficient of GPC by modified creep models ......................... 110
Figure 4.15: Prediction of creep coefficient on GPC ............................................. 114
Figure 5.1: X-ray diffractograms of samples at age of 28 days before creep test.
Peaks marked are Ettringite (E), Portlandite (P), Hydrotalcite (H), Calcite (C),
Gismondine (G), Quartz (Q) and C-A-S-H (CS) ................................................... 121
xvi
Figure 5.2: Differential thermograms (DTG) of GPC and OPCC at age of 28 days
before creep test ....................................................................................................123
Figure 5.3: Backscattered electron imaging of GPC and OPCC at age of 28 days .. 124
Figure 5.4: Pore size distribution of GPC and OPCC at age of 28 days.................. 125
Figure 5.5: X-ray diffractograms of samples after creep test (Creep) and the reference
sample (Control) of GPC and OPCC. Peaks marked are Hydrotalcite (H), calcite (C),
Gismondine (G), Quartz (Q), Diopsite (D), and C-A-S-H (CS), Portlandite (P),
Ettringite (E), Margarite (M), Chabazite (Ch)........................................................ 127
Figure 5.6: Differential thermograms for GPC and OPCC samples after creep test
(Creep) and the reference sample (Control) ........................................................... 129
Figure 5.7: Backscattered electron imaging of GPC after creep test ....................... 130
Figure 5.8: Backscattered electron imaging of OPCC after creep test .................... 131
Figure 5.9: Pore size distribution of GPC samples from core of specimens ............132
Figure 5.10: Pore size distribution of OPCC samples from core of specimens ....... 132
Figure 5.11: Pore size distribution of GPC samples from skin of specimens .......... 133
Figure 5.12: Pore size distribution of OPCC samples from skin of specimens ....... 133
Figure 6.1: Mini-beam specimens in custom designed moulds .............................. 139
Figure 6.2: Setup of mini-beam test....................................................................... 140
Figure 6.3: Mini-beam test in the environmental chamber ..................................... 141
Figure 6.4: Compressive strength of geopolymer and OPC pastes and mortars ...... 143
Figure 6.5: Maximum tensile stress at failure for geopolymer and OPC pastes and
mortars ................................................................................................................. 146
Figure 6.6: Creep coefficient of paste mini-beams: (a) GPP; (b) OPCP ................. 148
Figure 6.7: Creep coefficient of GPP and OPCP mini-beams at early stage of loading
............................................................................................................................. 150
Figure 6.8: Creep coefficient of mortar mini-beams: (a) GPM; (b) OPCM ............151
Figure 6.9: Creep coefficient of paste mini-beams under various temperatures: (a)
GPP; (b) OPCP ..................................................................................................... 153
Figure 6.10: Shrinkage strain of GPP mini-beams: (a) Effects of age of loading and
drying; (b) Thermal effect ..................................................................................... 155
Figure 6.11: Shrinkage strain of OPCP mini-beams: (a) Effects of age of loading and
drying; (b) Thermal effect ..................................................................................... 156
Figure 6.12: Shrinkage strain of GPM and OPCM mini-beams ............................. 158
xvii
Figure 6.13: Test results of autogenous shrinkage of GPP, OPCP, GPM and OPCM
............................................................................................................................. 159
Figure 6.14: Predicted creep coefficient of mortar mini-beams based on test results of
paste mini-beams: (a) GPM-1d-23-50, (b) OPCM-1d-23-50, (c) GPM-28d-23-50, (d)
OPCM-28d-23-50, (e) GPM-28d-23-S, (f) OPCM-28d-23-S ................................ 162
Figure 6.15: Predicted Shrinkage strain of mortar mini-beams based on test results of
OPCM-28d-23-50, (e) GPM-28d-23-S, (f) OPCM-28d-23-S ................................ 163
Figure 6.16: Samples prepared for nitrogen adsorption test ................................... 164
Figure 6.17: Pore size distribution on GPP-1d by different analysis methods ........ 165
Figure 6.18: Pore size distribution of GPP and OPCP: (a) DH method; (b) MP
method ................................................................................................................. 166
Figure 6.19: Pore size distribution of GPP after mini-beam test: (a) DH method; (b)
MP method ........................................................................................................... 169
Figure 6.20: Pore size distribution of OPCP after mini-beam test: (a) DH method; (b)
MP method ........................................................................................................... 170
Figure 6.21: Cross sections of GPP mini-beams: (a) GPP-28d-23-S; (b) GPP-28-50-S
............................................................................................................................. 171
Figure 6.22: X-ray diffractograms of mini-beam samples tested at 23 °C and 50 °C.
Peaks marked are Hydrotalcite (H), Chabazite (Ch) and C-A-S-H (CS) ................ 171
Figure 6.23: Porous material under loading ........................................................... 173
Figure 6.24: Adsorption/Desorption isotherms from nitrogen gas adsorption test: (a)
GPP-28d; (b) OPCP-28d....................................................................................... 184
Figure 6.25: Pore size distribution by DH method: (a) GPP-28d; (b) OPCP-28d ... 186
Figure 6.26: AFM images of elastic modulus: (a) Geopolymer, (b) OPC .............. 189
Figure 6.27: Viscosity measurement of pore solution: (a) Pore solution from
expression, (b) Rheometer .................................................................................... 191
Figure 6.28: Creep coefficient by creep model for mini-beam: (a) GPP, (b) OPCP 193
Figure 6.29: Maximum pore pressure during loading period by creep model......... 194
Figure 7.1: Detail of GPC beams type 1: (a) test setup and (b) cross-section detail.
............................................................................................................................. 198
Figure 7.2: Detail of GPC beams type 2: (a) test setup and (b) cross-section detail 200
Figure 7.3: Protecting coats for strain gauge ......................................................... 201
xviii
Figure 7.4: Photo of long –term beam test on GPC beams type 1 .......................... 203
Figure 7.5: Deflection of reinforced GPC beams type 1.........................................203
Figure 7.6: Ratio of long-term deflection to immediate deflection ......................... 205
Figure 7.7: Finite element model for OPCC beams type 1: (a) GiD; (b) ATENA
Studio ................................................................................................................... 207
Figure 7.8: Comparison of deflection of tested GPC beams to modelled OPCC
beams: (a) GPRCB1 and RCB1; (b) GPRCB2 and RCB2; (c) GPRCB3 and RCB3;
(d) GPRCB4 and RCB4; (e) GPRCB1-A and RCB1-A .........................................208
Figure 7.9: Test setup of the GPC beam under four-points bending ....................... 210
Figure 7.10: Load vs. deflection of beam test on GPC beams ................................ 210
Figure 7.11: Deformed shape of GPC beams after testing: (a) GPRCB1; (b)
GPRCB2; (c) GPRCB4; (d) GPRCB1-A ............................................................... 211
Figure 7.12: Cracking pattern of GPC beam at failure ...........................................211
Figure 7.13: In-situ specimens for compression strength test: (a) coring; (b) core
specimens .............................................................................................................212
Figure 7.14: Long-term deflection of GPC beams type 2 ....................................... 213
Figure 7.15: Recorded strain from strain gauges on GPC beams type 2 ................. 214
Studio ................................................................................................................... 215
Figure 7.17: Deflection of GPC beams type 2 and modelled OPCC beams type 2 . 216
Figure 7.18: Assessment of carbonation on GPC with phenolphthalein method: (a)
right after spraying indicator solution; (b) a few minutes later ............................... 218
Figure 7.19: Prediction of long-term deflection of GPC beams type 1 ................... 219
Figure 7.20: Estimated shrinkage strain of GPC beams related to age of 7 days..... 221
Figure 7.21: Estimated creep coefficient GPC beams: (a) GPRCB1, GPRCB2,
GPRCB3 and GPRCB4, (b) GPRCB1-A ............................................................... 222
Figure 7.22: Creep change and temperature change in each month of loading ....... 224
Figure 7.23: Long-term deflection of GPC beams with updated creep coefficient and
shrinkage strain by using AEMM .......................................................................... 226
Figure 7.24: Calculated shrinkage strain of GPC beam type 2 ............................... 228
Figure 7.25: Curvature of GPC beams type 2 at first loaded at 14 days and at the end
of test by RCM: (a) GPCB1, (b) GPCB2 ............................................................... 229
xix
of test by EMM: (a) GPCB1, (b) GPCB2 .............................................................. 230
of test by AEMM: (a) GPCB1, (b) GPCB2 ........................................................... 231
Figure 7.28: Deflection of GPC beams type 2 by RCM, EMM and AEMM: (a)
GPCB1, and (b) GPCB2 ....................................................................................... 232
Figure 7.29: Deflection of GPC beams type 2 by AEMM with coefficient β = 1 for
first two months of loading period: (a) GPCB1; and (b) GPCB2 ........................... 235
Figure 8.1: Testing equipment for thermal tests .................................................... 239
Figure 8.2: Heating conditions for thermal tests on geopolymer and OPC pastes .. 240
Figure 8.3: Compressive strength of GPP and OPCP under different temperatures 243
Figure 8.4: Creep and shrinkage strains at ambient temperature: (a) GPP, and (b)
OPCP ................................................................................................................... 245
Figure 8.5: Creep and shrinkage strains at 50 °C: (a) GPP, and (b) OPCP ............. 246
Figure 8.6: Creep and shrinkage strains at 80 °C: (a) GPP, and (b) OPCP ............. 247
Figure 8.7: Creep and shrinkage strains at 100 °C: (a) GPP, and (b) OPCP ........... 248
Figure 8.8: Specific creep of GPP and OPCP tested at various temperatures ......... 249
Figure 8.9: Transitional thermal creep strain: (a) GPP, and (b) OPCP ................... 250
Figure 8.10: Transitional thermal creep strain per unit stress ................................. 251
Figure 8.11: X-ray diffractograms of samples after thermal creep test. Peaks marked
are Portlandite (P), Hydrotalcite (Ht), Calcite (C), Gismondine (G), Wollastonite
(W), Vaterite (V), Ettringite (E), Gypsum (Gs) and C-S-H, C-A-S-H (CS) ........... 252
xx
List of tables
Table 2.1: Main compounds of OPC ........................................................................ 8
Table 3.1: Composition of OPC, GGBFS and FA from X-ray fluorescence analysis
(LOI is loss on ignition at 1000 °C) .........................................................................60
Table 3.2: Physical properties of the source materials .............................................60
Table 3.3: Mix designs for geopolymer and OPC concretes, mortars and pastes ......63
Table 4.1: Details of concrete specimens for testing ................................................78
Table 4.2: Compressive strength of OPCC and GPC ...............................................81
Table 4.3: Details of concrete specimens under shrinkage test .................................85
Table 4.4: Details of concrete specimens under creep test .......................................90
Table 4.5: Instantaneous elastic modulus of GPC and OPCC in creep test ...............90
Table 4.6: GPC creep model based on AS3600-2009 ............................................103
Table 4.7: GPC creep model based on Eurocode 2-2004 ....................................... 105
Table 4.8: GPC creep model based on model B3 ................................................... 106
Table 4.9: GPC creep model based on ACI model ................................................. 108
Table 4.10: GPC creep model based on CEB-fib model code 2010 ....................... 109
Table 4.11: Root-mean-square error (r2) of the GPC creep models to test data....... 113
Table 4.12: Prediction of creep coefficient at 30 years by the modified creep models
............................................................................................................................. 117
Table 6.1: List of specimens in mini-beam test ...................................................... 141
Table 6.2: Specific surface area of GPP and OPCP ............................................... 165
Table 6.3: Specific surface area of GPP and OPCP after mini-beam test ............... 167
Table 6.4: Input parameters and estimated permeability of GPP-28d and OPCP-28d
............................................................................................................................. 187
Table 6.5: Elastic modulus based on mini-beam test on GPP-28d-23-S and OPCP-
28d-23-S ............................................................................................................... 188
Table 6.6: Parameters for viscoelastic modelling ................................................... 192
Table 7.1: Details of GPC beam type 1 ................................................................. 199
Table 7.2: Details of reinforcement in GPC beam type 2 ....................................... 200
Table 8.1: Weight of specimens after test relative to measurement before test ....... 254
xxi
List of symbols
A Cross-sectional area of pore
Ac Cross-sectional area of concrete only
Aid Idealised cross sectional area
Am Average area occupied by one molecule of adsorbate in the completed
monolayer
As Cross-sectional area of steel component
b Half of width
Bc First moment of area of concrete only
Bid First moment of area of concrete only and idealised section
C BET constant
Cd(t,t0,td) Additional compliance function due to simultaneous drying from model
B3
C0(t,t0) Compliance function for basic creep from model B3
d Half of depth
dc Critical pore diameter
Da Aggregate volume concentration
Df Fractal dimension
Dmax,eff Maximum effective pore diameters by truncating the pore volume pore
size distribution by 2.5% from upper end of the distribution
Dmean Mean pore diameter from the pore volume pore size distribution
Dmin,eff Minimum effective pore diameters by truncating the pore volume pore
size distribution by 2.5% from lower end of the distribution
E Elastic modulus
Ea Age-adjusted effective modulus
Eagg Elastic modulus of aggregates
Ee Effective modulus
Em Elastic modulus of mortar
Ep Elastic modulus of paste
EP Elastic modulus of the porous material
EPV(t) Elastic modulus of the viscoelastic porous material
Es Elastic modulus of steel
E0 Asymptotic modulus
E˝m Age-adjusted effective modulus of mortar
E˝p Age-adjusted effective modulus of paste
fc Compressive strength
fcm Mean concrete compressive strength
fcmi Mean value of the in-situ compressive strength
fc0 compressive strength at zero porosity
f’c Characteristic compressive strength
f’ct.f Modulus of rupture
xxii
G Shear modulus
GP Shear modulus of the porous material
GPV(t) Shear modulus of the viscoelastic porous material
h Thickness
H(t) Function for humidity
I Moment of inertia
Ic Moment of inertia of concrete only
Iid Idealised moment of inertia
J Flux
J(t,t0) Creep compliance function
Jm(t,t0) Creep compliance function of mortar
Jp(t,t0) Creep compliance function of paste
k Curvature
kH Experimental constant for porosity-strength equation
kL Langmuir constant
kr Experimental constant for porosity-strength equation
ks Experimental constant for porosity-strength equation
k2 Time-dependent factor from AS3600
k3 Factor for age of concrete from AS3600
k4 Environment factor from AS3600
k5 Modification factor of high strength concrete from AS3600
Ka Bulk modulus of aggregates
KL Bulk moduli of the liquid phases
Kp Bulk modulus of paste
KP Bulk modulus of the network of the porous material drained of liquid
KPV(t) Bulk modulus of the network of the viscoelastic porous material drained
of liquid
KS Bulk moduli of the solid phases
l Global length of the porous material
li Length of pore
L Length
M Bending moment
Mcr Bending moment combined with creep effect produced by initial stress
ML Mass of fluid
Ms Applied bending moment
n Experimental constant for porosity-strength equation
N Axial load
NA Avogadro constant
Ncr Axial load combined with creep effect produced by initial stress
Ns Pore length parameter
NS Applied axial load
p Stress in the liquid phase of the porous material
xxiii
P Pressure
Po Saturated vapour pressure
ΔP Pressure drop
q1 Heat of adsorption of the first layer
qL Heat of liquefaction of the liquid adsorption
Q Total flow
Q(t,t0) Function for effect of age of loading for model B3
Qi Flow corresponded to ith pore sizes
r Radius
re Equivalent pore radius
ri ith pore sizes
rK Radius of the pore in which condensation occurs
rm Mean pore size
R Gas constant
R(t,τ0) Relaxation function
RH Relative humidity
Rn Critical radius of pore corresponding to the smallest section
S Specific surface area
Sn Smallest pore section
t Time
T Temperature
td Age of concrete when drying commenced
th Hypothetical thickness
t0 Age of concrete when loaded
u Constant for the viscoelastic material
v Constant for the viscoelastic material
VE(t) Time-depended function for viscoelastic behaviour
Vi Pore volume corresponding to the pore radius ri
VL Volume of fluid
Vm Volume of gas adsorbed when the entire surface is covered by a
monomolecular layer
VM Molar volume
VS Volume of solid
W Applied load on mini-beam
x Horizontal distance from the centroid of the cross section
x(t,τ0) Age-adjusted coefficient
y Vertical distance from the centroid of the cross section
yid Distance between the steel component and the centroid of the idealised
cross section
β(fcm) Factor based on the mean compressive strength of concrete from
Eurocode 2
β(RH) Factor related to the relative humidity and size of concrete member from
fib model code 2010
xxiv
β(t0) Factor for the age of loading from Eurocode 2
βbc(fcm) Factor estimated from the mean compressive strength of concrete at age
of 28 days from fib model code 2010
βbc(t,t0) Ttime-dependent factor from fib model code 2010
βc(t,t0) Factor for development of creep with time from Eurocode 2
βdc(fcm) Factor estimated from the mean compressive strength of concrete at age
of 28 days from fib model code 2010
βdc(t0) Factor for the age of loading from fib model code 2010
βdc(t,t0) Time-dependent factor from fib model code 2010
βH Factor depending on the relative humidity from Eurocode 2
γfcm Strength of concrete factor from ACI model
γh Thickness factor from ACI model
γla Age of loading factor from ACI model
γs Size factor from ACI model
γsh Coefficients for sustained loads and shrinkage
γsus Coefficients for sustained loads
γλ Relative humidity factor from ACI model
γψ Temperature factor from ACI model
Γhydr Degree of hydration
δ Deflection
Δ Delfection
ε Strain
εcm Creep strain of mortar
εcp Creep strain of paste
εcr Creep strain
εe Instantaneous elastic strain
εk Linear strain
εm Elastic strain of mortar
εp Elastic strain of paste
εsh Shrinkage strain
εsh∞ Ultimate shrinkage strain
εsm Shrinkage strain of mortar
ε sp Shrinkage strain of paste
εth Free thermal strain
εTTC Transit thermal creep strain
εv Volumetric strain
ε*sh Final shrinkage strain
ζ Constrictivity factor
η Viscosity
θ Function of time
θv Function of time
xxv
θL Transition time
κ Permeability of the medium
λ Constant for GPC_ACI creep model
υ Poisson’s ratio
υm Poisson’s ratio of mortar
υp Poisson’s ratio of paste
υPV Poisson’s ratio of the viscoelastic porous material
ρ Density
ρL Density of fluid
ς Volume fraction of solid phase in the porous material
σ Stress
σc Sustained stress
σct (t) Change of stresses of concrete related to the initial stresses
σssus0 Initial stress of steel component under sustained loads
σst (t) Change of stresses of steel components related to the initial stresses
σ ̃k Stress on solid phase of the porous material
τ Tortuosity factor
ϕ Porosity
ϕ0S Porosity at zero strength
Φ(t,t’) Micro-compliance function
φ(t,t0) Creep coefficient
φbc(t,t0) Basic creep coefficient
φcc Design creep coefficient
φcc.b Basic value for creep coefficient
φdc(t,t0) Drying creep coefficient
φm Creep coefficient of mortar
φp Creep coefficient of paste
φRH Factor for the effect of relative humidity
φu Ultimate creep coefficient
φ0 Notional creep coefficient
ω Constant for GPC_ACI creep model
Ω Constant
xxvi
Abbreviation index
AAM Alkali-activated materials

AEMM Age-adjusted effective modulus method
AFM Atomic force microscopy
ASR Alkali silica reaction
BET Brunauer-Emmett-Teller theory
BFS Blast furnace slag
BJH Barrett-Joyner-Halenda method
BSE Backscattered electron
BSEM Backscattered scanning electron microscopy
C2S Dicalcium silicate
C3A Tricalcium aluminate
C3S Tricalcium silicate
C4AF Tetracalcium aluminoferrite
CI Cranston-Inkley method
CPSM Corrugated pore structure model
DH Dollimore-Heal method
DTG Differential thermogram
EMM Effective modulus method
FA Fly ash
FTS Free thermal strain
GGBFS Ground granulated blast furnace slag
GPC Geopolymer concrete
GPM Geopolymer mortar
GPP Geopolymer paste
HPC High-strength performance concrete
ITZ Interfacial transition zone
LITS Load-induced thermal strain
LOI Loss on ignition
LVDT Linear variable differential transformer
MIP Mercury intrusion porosimetry
MP Micropore method
OPC Ordinary Portland cement
OPCC Ordinary Portland cement concrete
OPCM Ordinary Portland cement mortar
OPCP Ordinary Portland cement paste
RCM Rate of creep method
RH Relative humidity
SCM Supplement cement materials
SSA Specific surface are
TGA Thermogravimetry analysis
xxvii
TTC Transitional thermal creep
w/b Water to binder ratio
w/c Water to cement ratio
XRD X-ray diffraction analysis
xxviii
Chapter 1 : Introduction
1.1 Background
Ordinary Portland cement (OPC) had been invented and widely utilised over a
century as the major construction material to form concrete and mortar. However,
production of one ton of Portland cement can result in up to one ton of carbon
dioxide (CO2) into the atmosphere [1]. Increasing amount of carbon dioxide is
considered as one of greenhouse gases, which contributes to global warming.
Geopolymer concrete (GPC) was introduced in past few decades to address the
problem with its possibility of reducing CO2 production up to 80% compared to
Portland cement [2]. This is because the industry by-productions or refinement of
natural source, such as blast furnace slag, fly ash and metakaolin, are used to produce
GPC through the activation of the above alumina- and silica-containing solid
precursors by an alkali source. Utilisation of these materials can relieve stress in
disposal and treatment.
Past research on GPC shows that GPC can achieve workable slump, and comparable
grade of strength to ordinary Portland cement concrete (OPCC) [3]. It is also shown
that GPC can perform well in durability. For example, some research shows that
GPC has a good resistance to chloride penetration [4, 5]. Despite the environmental
benefit that GPC provides in terms of low energy consumption in production and
utilisation of industrial by-product and the mechanical properties for practical use, it
is not widely used in the construction industry. One of the causes is lack of codes and
standards for structural design for GPC [2]. Therefore, it requires more investigation
on the performance of GPC in concrete structures. For concrete structures, the
critical design factor is not only the strength of concrete, but also the long-term
structural behaviours. However, these properties are not well known for GPC and
there is little research on creep of GPC. Creep of concrete is one of the major factors
to determine the suitability of concrete as main structural component. Creep is a
phenomenon that increases strain under sustained load over time, and it influences
the serviceability of concrete structures. Therefore, it is essential to understand the
creep behaviour of GPC, before its application on concrete structures.
1
The available test data for creep of GPC is little compared with the vast amount of
research on OPCC in the last century. It is not feasible to build a large database for
creep of GPC in the short period. Studying the microstructure can be a path to
understand the creep mechanism of GPC. Researchers proposed that creep of OPCC
can be caused by breakage or slipping of gels [6, 7], or change of distance of
interlayers due to pore solution diffusion [8]. It is expected some of these
mechanisms can also be applicable on GPC. However, due to the difference in
microstructure between GPC and OPCC, the relationship between microstructure and
creep of GPC should be investigated.
1.2 Research significants
The aim of the research is to compare the different creep behaviours between GPC
and OPCC, and to identify the cause of these differences through the experimental
results and microstructure analysis, and thus establish the creep model for GPC to
apply on structural application.
There are some objectives in this research:
1. To compare the different creep behaviours between GPC and OPCC, and to
identify the factors influencing the creep behaviours of GPC through various
experimental configurations.
2. To investigate the creep mechanism of GPC through the comparative study of
microstructure with OPCC.
3. To develop a new accelerated test method in creep analysis, which can
provide more test data in a short period.
4. To establish creep model for GPC.
5. To investigate the long-term performance of full-scale concrete beams made
of GPC.
2
1.3 Scope of thesis
This research was conducted through tests in a wide range of scales, which include
analysis of solid phases and pore size (nm), thin plates of paste specimens of 10 mm
thick, concrete cylinders of 150 mm diameter, and concrete beams of several metres.
Tests were done on a single formulation of GPC and OPCC each. The formulation of
GPC was chosen from the preliminary works due to its satisfactory strength property
and its use in full-scale application. Ground granulated blast furnace slag (GGBFS)
and fly ash (FA) were used as raw binder materials for GPC. GGBFS was obtained
from Independent Cement and Lime Pty Ltd without added Gypsum. FA was
obtained from Gladstone power station. Reference specimens of OPCC were made
by replacement of all amount of geopolymer binder materials with OPC. The OPC
used for the reference formulation was commercial type GP cement.
The shrinkage, elastic modulus and strength of GPC and OPCC were tested for the
material properties. The creep tests on concrete specimens were carried out on both
GPC and OPCC under standard drying and sealed conditions to investigate the effect
of drying on GPC and OPCC. Three more sets of GPC were tested for study of the
effect of strength and the effect of age of loading on creep of GPC. All of these creep
tests were undergoing for a period from 1 to 3.5 years. Creep models were proposed
for GPC based on these five sets of GPC creep data. The microstructure of the GPC
and OPCC was assessed on the concrete samples obtained before and after creep test.
Phase change and pore sizes alternation due to the sustained loads were investigated.
The effects of relative humidity, temperature and age on creep were further
investigated through a test developed on paste and mortar specimens of geopolymer
and OPC. The specimen size is of 50 mm wide × 10 mm deep × 350 mm long, and
subjected to long-term four-points bending test under sustained loads. Eight test
configurations, in the combinations of three relative humidity levels (50%, 70% and
sealed), three temperature levels (0, 23 and 50 °C) and two ages (1 and 28 days),
were set up for each type of specimens. A creep model was developed to study the
hydrodynamic and the viscoelastic behaviours on creep of geopolymer and OPC
specimens based on the microstructural analysis.
3
The long-term performance of GPC structures was studied based on two types of
concrete elements aimed for reinforced concrete beams of 4 m long and composite
steel floor slabs of 6.5 m long. These GPC beams were under sustained loads for
flexural actions in a period up to 2 years. Due to the restrained on space and
resource, beam of OPCC was not constructed. Virtual OPCC beams were made with
the aid of computer software to compare with GPC beams. The effects of
reinforcement detail, age of loading and the seasonal change of temperature on creep
of the GPC beams were investigated.
The effect of temperature on creep of geopolymer paste (GPP) and OPC paste
(OPCP) specimens was further investigated through sustained compression test on
cylinder of 50 mm diameter from 50 to 100 °C.
1.4 Thesis organisation
There are nine chapters in this thesis. Chapter 1 presented the background
information of this research, the motivation in conducting research on creep of GPC,
the aim of the research, the objectives of the research, the scope of the research and
brief details on the research topics in this thesis.
Chapter 2 provided the literature reviews on the research covering the difference in
the creep behaviour and the microstructure between GPC and OPCC, and the
relationship between microstructure and creep behaviour was discussed. In addition,
some creep models and calculation methods for creep deformation for OPCC were
presented, which can be applicable on GPC structural design. The importance of
accuracy in creep prediction for structural design was discussed.
Chapter 3 presented the information of materials used in this research and the
preparation of test specimens. The test methods and technologies were also
presented.
Chapter 4 presented the results of GPC and OPCC subjected to standard tests,
including compressive strength test, flexural tensile strength test, elastic modulus
4
test, shrinkage test and creep test, followed by discussion. The influence of age of
loading and strength on creep of GPC was further investigated. The creep recovery
of GPC and OPCC was compared and discussed. Creep models of GPC was
proposed based on the creep data of GPC.
Chapter 5 presented the analysis of microstructure of GPC and OPCC from Chapter
4 before and after creep test. The difference in microstructure between GPC and
OPCC was discussed. Change of microstructure due to sustained loads was
investigated. Finally, the relationship between the microstructure and the creep
behaviour of GPC and OPCC was presented.
Chapter 6 presented a new test method developed for flexural creep behaviour of
geopolymer and OPC pastes and mortars. Results of the test were presented and the
effect of relative humidity, age of loading and temperature on creep behaviour were
discussed. The change of pore size distribution of each specimen before and after the
test period was investigated. A creep model was developed to investigate the
hydrodynamic creep behaviour and the visco-elastic creep behaviour of the basic
creep of the GPP and OPCP specimens.
Chapter 7 presented the long-term deflection of two types of GPC beams; one aimed
for reinforced concrete beam and another aimed for composite floor slab. The test
results were compared with OPCC beams by computer stimulation. Prediction of
long-term deflection of these GPC beams were done by using rational methods with
material properties based on the test results from Chapter 4. The effect of
reinforcement detail and the seasonal change of temperature on the long-term
deflection of these GPC beams were discussed.
Chapter 8 presented the investigation of effect of temperature up to 100 °C on creep

behaviour of GPP and OPCP, which included the 1-day creep, transit thermal creep,
thermal expansion strain and hot strength. The different creep response of GPP and
OPCP was discussed.
Chapter 9 provided a summary of the research findings and the conclusions from the
research. Moreover, recommendations on future works were suggested.
5
Chapter 2 : Literature review
2.1 Overview
Ordinary Portland cement (OPC) had been invented and widely utilised over a
century as the major construction material in production of concrete and mortar. In
the design of concrete structures, the time-depended effects of creep and shrinkage
are critical for serviceability. Shrinkage occurs as a reduction of volume due to loss
of water in the drying process [9], whilst creep increases strain when the concrete
member is subjected to a sustained stress [10].
Creep causes additional deformation of the concrete under sustained load. The rate of
change of creep decreases over time since loaded [6, 10]. Creep is often responsible
for excessive deflections under service loads. For example, concrete beams and slabs
undergo long-term deflection by creep caused by their self-weight with or without
other sustained load. When such deflection exceeds the acceptable limit by the
authorities, these concrete structures do not satisfy the serviceability requirement.
The excessive deflections can affect and cause damage to the non-load bearing
components associated with the structure, e.g. window frame. Creep not only
increases the deflection, it also changes the maximum stress and width of crack in the
structures. This influences the design of some creep-sensitive structures, such as
bridges (prestressed box girder bridges, cable-stayed bridges, arch bridges, etc.),
shell structures, and tall buildings [11]. Underestimation in these effects can cause
the ultimate failure of the structures in worst case, such as bridge collapse [12].
Although creep in ordinary Portland cement concrete (OPCC) has been studied for a
long period, its full mechanism has not been fully understood. Furthermore, the
research on creep of geopolymer concrete (GPC) is little compared to OPCC. It is
believed that some of the proposed creep mechanisms of OPCC may assist in
exploring and explaining the creep phenomenon in GPC. As creep mainly occurs in
the paste portion in concrete, understanding the microstructure of paste would be a
key to study the mechanism. The significant of creep will be discussed in the
following sections.
6
2.2 Background of ordinary Portland cement concrete
and geopolymer concrete
2.2.1 Ordinary Portland cement concrete
OPCC is the most used construction materials in the world. The patent on Portland
cement was taken by Joseph Aspdin in 1824 [13] . Since then the use of OPCC has
been developed in a wide range of structures and spread over the world.
The manufacture of OPC requires mainly two raw materials: limestone (calcium
carbonate) and clay or argillaceous rock (silica source). First, the raw materials are
ground and blended to a constant composition, which ensures predictable properties
of the end product. Then the blended raw material is fed to the kiln, in which it is
heat up and allows chemical reactions to proceed. A mixture of fuel and air is
injected at the low end of the kiln and burned to provide heat. Inside the kiln, four
distinct processes take place in different temperature zones: dehydration zone,
calcination zone, clinkering zone and cooling zone. When the blended raw material
is fed to the dehydration zone, free water is lost by evaporation at a temperature up to
450 °C. The temperature increases gradually in the calcination zone up to 1350 °C.
The carbon dioxide (CO2) is lost from limestone and the raw materials are
transformed into a mixture of oxides that can enter into new chemical combinations
as below:
CaCO3 → CaO + CO2
2CaO + SiO2 → 2CaO·SiO2
2CaO·SiO2 + CaO → 3CaO·SiO2
3CaO·Al2O3 + CaO + Fe2O3 → 4CaO·Al2O3·Fe2 O3
Calcium silicates (2CaO·SiO2 and 3CaO·SiO2) and other chemical compounds are
formed in liquid phase in the clinkering zone up to 1600 °C. After it passes the
clinkering zone into the cooling zone, the liquid phase solidifies to produce the
clinker. The hot clinker is further cooled down and ground to the fine powder.
7
Usually gypsum is added during grinding to control the early setting property of OPC
[13].
The manufacture of OPC is an energy-intensive process, as fuel is burned to maintain

high temperature in the kiln. A large amount up to 1 tonne of CO 2 can be produced
by burning the fuel and heating the limestone in order to produce 1 tonne of OPC
[14]. This releases a lot of CO2, and CO2 is one of the main greenhouse gases. Such
gas emission can be reduced by using dry process instead of wet process in
manufacture method, as less heat is required to evaporate all free water from the raw
materials for the dry process [13]. The use of waste derived fuels can also reduce the
CO2 emission from manufacture of OPC [15].
For construction use, OPC is mixed with water and fine aggregates to produce mortar
and with additional coarse aggregates to produce concrete. After addition of water,
the loose mixture undergoes hydration to harden and gain strength as a solid mass:
2(3CaO·SiO2) + 7 H2O → 3CaO·2SiO2·4H2O + 3 Ca(OH)2
2(2CaO·SiO2) + 5 H2O → 3CaO·2SiO2·4H2O + Ca(OH)2
There are four main chemical compounds in OPC as shown in Table 2.1. Among
these compounds, the hydration products of C3S and C2S are believed to be the major
bonding agent in the system. This hydration products are 3CaO·2SiO 2·4H2O,
calcium silicate hydrates, or as C-S-H, and Ca(OH)2, calcium hydroxide or
portlandite[1].
Table 2.1: Main compounds of OPC
Name of compound Composition Abbreviation

Tricalcium silicate 3CaO·SiO2 C3S
Dicalcium silicate 2CaO·SiO2 C2S
Tetracalcium aluminoferrite 4CaO·Al2O3·Fe2O3 C4AF
Tricalcium aluminate 3CaO·Al2O3 C3A
8
OPCC is typically strong in compression resistance but weak against tension.
Reinforcement, usually steel, is required to strengthen OPCC. The construction
methods using OPCC include cast in-situ construction and precast construction. The
cast in-situ construction requires formwork for each element set up with
reinforcement on site, and then the fresh mix of OPCC is poured into the formwork,
and allowed to set and harden. The precast construction involves prefabrication of
OPCC elements in the factory and then transportation and erection on site [16].
2.2.2 Geopolymer concrete
There are a few different names when it comes to describe geopolymer, for
examples, inorganic polymer and alkali-activated materials (AAM), due to the wide
range of formulations. Some argued that geopolymer referred to alkali-activated
binders with precursors which had low calcium content, like low-calcium fly ash and
calcined clays [14, 17]. This would create confusion, and thus only one name,
geopolymer, is used in this research.
Geopolymer refer to the binder materials achieved by the reaction of an alkali source
when an alumina- and silica-containing solid precursor (see Figure 2.1). It was first
discovered by Kühl in 1908 [18]. Since then research on this technique have been
carried out and identified some potential solid precursors such as fly ash, metakaolin
and blast furnace slag (BFS) [3, 19-21]. In these potential solid precursors, fly ash is
captured by the electrostatic precipitators from the flue gases after coal combustion
in the power plant; blast furnace slag is the by-product in the iron production
process; and metakaolin can be produced from the calcination of kaolin-containing
clays at 600 to 900 °C [22].
9
Figure 2.1: Production of geopolymer
Besides the precursors, another major component of geopolymer is the alkali

activator. The common alkali activator can be sodium or potassium hydroxides,
silicates, carbonates or a mixture of those, which are soluble in water [4, 23, 24]. In
order to obtain good performance of the final product, the coupling of precursor and
alkali activator need to be taken into account. Some precursors will react
preferentially or work efficiently with a certain type of activator than the others, due
to the difference in chemistry. For example, sodium silicate solution would provide
better activation to ground granulated blast furnace slag (GGBFS) than the other
types of alkali activator; Experiment results showed that high early strength can be
obtained with sodium silicate solution of low modulus (mass ratio of SiO 2 to Na2O,
Ms = 0.75), maximum later age strength shifts toward higher modulus [25].
Moreover, sodium hydroxide was preferred in the alkali-activation of fly ash [26].
The series of reactions between the solid precursors and alkali sources can be called
as alkali activation or geopolymerisation. Duxson et al. [17] proposed a conceptual
model for this reaction as shown in Figure 2.2. Initially after the contact between the
precursor and the alkali activator, the amorphous components (aluminates and
silicates) of the precursor dissolve. Then the aluminates and silicates inter-react to
form an aluminosilicate gel. This gel is first formed as an aluminium-rich gel (Gel 1),
10
since the reactive aluminium dissolves more rapidly than the silicon. When more
silicon dissolves in the later stage, the gel structure is reorganised to form the zeolite
precursor gel (Gel 2). This gel is more stable than the previous form since Si-O
bonds is stronger than Al-O bonds. These reorganisation processes keep going and
result of some crystallised zeolite. Therefore, the gel bond together and form a solid
mass similar to the hydration of OPC.
Similar to OPCC, aggregates can be added into geopolymer to produce GPC.

Research on GPC has shown that it has equal or superior properties to OPCC. GPC
can achieve strength comparable to OPCC in normal strength grade [3, 27]. Also,
GPC is a good fire resistant material [28-30], and it has a better performance in
resistance to sulphate attack [31] and chloride [4, 32]. Further details about durability
will be presented in section 2.4.2.
Douglas et al. [3] conducted tests on five GPC mixtures activated by sodium silicate
and found that GPC can provide satisfactory workability and strength properties. The
mechanical strengths of GPC were influenced by a range of factors. Among them,
generally the order of the most significant effects on the development of mechanical
strengths was alkaline activator nature, activator concentration, curing temperature
and specific surface [24].
The workability and strength development can be influenced by the type and dosage
of activator. A study by Collins and Sanjayan [33] showed that slag-based
geopolymer mortar activated (GPM) by NaOH and Na2CO3 achieved high early
strength compactable to OPC mortar (OPCM), but the 28-day strength was lower.
Slag-based GPC activated by high concentration of liquid sodium silicates presented
considerably higher shrinkage than OPCC. It would be due to the Ca in the type of
C-S-H formed in the presence of high NaOH concentration was replaced by Na. The
Na-substituted form was less dense structure of N-C-S-H, which also has a low
strength to restraint shrinkage [25].
11
• Aluminosilicate source
• Alkali activator
Dissolution
• Aluminate
• Silicate
Speciation
equilibrium
• Aluminosilicate gel
Gelation
• Aluminium-rich gel (Gel 1)

Reorganisation
• Zeolite precursor gel (Gel 2)

Polymerisation and
hardening
• Geopolymer
Geopolymer
Figure 2.2: Conceptual model for geopolymerisation (Source: Duxson et al. (2007)
[17])
12
Bath cured slag-based GPC at ambient temperature showed higher early strength
than OPCC [34]. On the other hand, fly ash-based GPC showed low strength and
slow strength development when cured in ambient temperature. Heat curing is
applied to improve the properties. Long duration of heat curing and high temperature
can increase the strength of fly ash-based GPC [35-37]. Although heat curing can
accelerate strength gain for slag-based GPC in early age, it leaded to a lower strength
at later age than the samples cured at room temperature. This would be due to the fast
rate of reaction, which caused inhomogeneity of microstructure, localisation of
reaction product near slag grains and coarse pore structures. These reaction product
formed a barrier for further reaction, so it resulted in slow strength growth in later
ages [38]. Therefore, heat curing is not necessary for slag-based GPC. In addition, fly
ash-based geopolymers had highly stable structures than the metakaolin-based
geopolymers [29, 30].
Bakharev, Sanjayan and Cheng [39] assessed the performance of slag-based

geopolymer with various admixtures and found that the admixtures for OPCC may
have significant side effects, such as increase in shrinkage, strength loss, on
geopolymer. These effects would be due to the reaction between the admixtures and
activators.
GPC can provide a number of environmental advantages. The utilisation of GPC can
reduce CO2 emission up to 80% compared to OPCC and removed materials (BFS
and fly ash) from landfill, and thus leads to sustainable construction [2]. It was also
found that production of slag consumed less energy than production of OPC in same
amount [40]. Although GPC is a potential alternative as a sustainable construction
material, but the current application on structural components is limited due to lack
of codes and standards for structural design. More research on durability and long-
term behaviours is required for commercialisation of GPC [2].
13
2.2.3 Blended cement concrete
A substantial reduction of CO2 emission can be achieved by blended OPC with

natural pozzolans and industrial by-product [41]. When the supplement materials
(BFS and fly ash, etc.) are added into OPC, the reaction mechanism becomes a bit
different. Both hydration and pozzolanic reactions happen in blended cement
concrete. The OPC undergoes hydration in the beginning. It produces portlandite
beside C-S-H and other hydration products. Portlandite increases the pH of the pore
solution to a high level. It acts as similar role to alkali source and enables the
pozzolanic reaction of BFS or fly ash as shown below:
1.1Ca(OH)2 + SiO2 + H2O → 1.1Ca(OH)2·SiO2·2.1H2O
Therefore, more C-S-H is formed in the system [15]. Because of that, it is usually
used to produce high-strength performance concrete (HPC).
The research on blended cement would provide some information on the

performance of geopolymer, since the performance of blended cement depends on
the supplement cement materials (SCM) by their source. The reactivity would be
varied by the chemical composition, particle fineness, and particle size distribution
[42], which are similar to geopolymer.
The curing method has an impact on the performance of blended cement. Research
on strength development of cement mortar with GGBFS showed a higher early
strength but a lower later strength at higher curing temperature, which is due to the
rapid hydration at early age resulting in poor microstructure of hydrated cement paste
and restraining the pozzolanic reaction at later age. The time required to reach 50%
of ultimate strength increased with decreasing temperature, increasing water-binder
ratio and increasing GGBFS level. The apparent activation energy was higher for
higher GGBFS level. At 5 °C, the compressive strength of the specimens was lower
than that at 20 °C, and it decreased with increasing content of GGBFS. Using finer
GGBFS can minimised the influence by cold temperature curing. When air-curing
was applied, the strength development was greatly affected, since the water required
for the reaction of GGBFS no longer existed at the later age [43, 44].
14
For SCM of fly ash instead of GGBFS, Malhotra et al. [45] found that the HPC
incorporating 57% fly ash had a low early compressive strength (28 days) but the
strength gain was large at later age (up to 10 years) subjected to outdoor exposure
condition. Sometimes, silica fume is also used in blended cement. It formed a denser
microstructure that mitigated the loss of moisture from the concrete surface under
air-curing and allowed for continued strength development [46].
Collins and Sanjayan [47] compared the properties of concrete columns made with
GPC, OPCC, and 50% blended cement mix. They found that the GPC column
showed lower rate of temperature rise and lower peak temperature at the centre of the
column during hardening. The maximum temperature difference between the centre
and exterior of the column was 10.5 °C for GPC, whereas OPCC and 50% blended
cement concrete were 20 °C and 15 °C, respectively. The compressive strengths of
GPC and OPCC were similar at 28 and 91 days, with an increase towards the centre
of the columns. Both of them had higher compressive strength than the blended
cement concrete. The generally higher strength at the centre of the columns was
because the condition of the centre of the column was similar to the sealed condition,
where the effect of drying was minimised.
2.3 Microstructures
2.3.1 Hardened OPC
The microstructure of hydrated OPC consists of two parts; solid phases and voids.
The solid phases are the hydration products of OPC. The main hydration product,
which give strength to hydrated OPC, is C-S-H gel; it exists in variety of forms.
Generally, it is believed to have a layer structure, and the individual layers are well
crystallised. This structure is highly disordered and Van der Waal forces are formed
between these gels to give strength. It bonds the unhydrated cement particles and the
aggregates together as a whole. A model for hydrated OPC is illustrated in Figure
2.3. It can be seen that hydration products do not occupy all the space. Voids existing
in such space are of various sizes and some of them are filled with water. The pore
15
structure, which is of interest in study of creep and shrinkage in OPC, is related to
the capillary pores (less than 1 μm) and the gel pores (less than 2 nm or 3 nm). The
water in these pores can be divided into three types; capillary water which is the free
water in the capillary pore, physically adsorbed water which is the water molecules
adsorbed on the wall of the capillary, and interlayer water which exists between the
gels and its movement is highly constrained [1, 48, 49].
Figure 2.3: Diagrammatic model for hydrated OPC (Source: Mehta and Monteiro
(2006) [48])
The C-S-H is mainly amorphous phases. The layer structure of C-S-H involves
elements of tobermorite-like structure intermixed with jennite crystalline structure.
The layers are separated by an intermediate layer with H 2O, Ca2+, OH- and other
ions, forming the structures of 1.4 nm tobermorite and jennite of various chain length
(see Figure 2.4). When the layer structure lost the interlayer water under the electron
beam, the layer spacing reduced from 1.42 nm to 1.24 nm to 1.13 nm, and
decomposition of the tobermorite structure [50].
16
C-S-H
Intermediate layer
Figure 2.4: 1.4 nm tobermorite structure of C-S-H (Adapted from Richardson (2008)
[50])
2.3.2 Hardened geopolymer
Past researches showed that the chemistry of geopolymer is different to that of

hardened OPC. The microstructure of geopolymers varies with the type of precursors
used. The main reaction product of slag-based geopolymer is an aluminium-
substituted C-A-S-H type gel, with cross-linked and non-cross-linked tobermorite-
based structures [51], unlike the main hydration product of OPC is C-S-H gels with
non-cross-linked structures [50]. The cross-linked structures of geopolymer are
resulted from the geopolymeristaion. Myers et al. [51] proposed the cross-linked
substituted tobermorite model (CSTM) for C-A-S-H gel, which described the solid
phases in geopolymer as a mixture of 14, 11 and 9 Å tobermorite structures, which
included cross-linked and non-cross-linked tobermorites. The CSTM only allowed
for aluminium substitution in bridging sites and did not allow single vacancies in the
combined bridging sites. An illustration is shown in Figure 2.5.
17
Figure 2.5: Illustration of constraints in CSTM (Source: Myers et al. (2013) [51])
In fly-ash based geopolymer, the main reaction product is an alkaline silicoaluminate

gel containing silicon and aluminium tetrahedral randomly distributed and cross-
linked. When sodium hydroxide (NaOH) is used as activator to this type of
geopolymer, the produced gel is known as N-A-S-H gel [26]. Furthermore,
experiments done by Garcia-Lodeiro et al. [52] showed that in presence of Ca, N-A-
S-H would transform to C-A-S-H at high pH until available Ca was exhausted.
Criado et al. [53, 54] studied the nanostructure of N-A-S-H gel and proposed that it
was an amorphous gel consisting in a polymeric, cross-linked aluminosilicate
network similar to C-A-S-H gel. The N-A-S-H gel comprised two types of Si-O
bonds, which were the bridge and terminal bonds. They found that the nature of this
gel was influenced by the activator and curing method. When there was more soluble
silica in the system, more bridge bonds were formed. Moreover, when it was under
thermal curing, the structure became more orderly with a predominance of bridge
bonds over terminal bonds, and hence more stable.
2.3.3 Differences in microstructure between OPC and geopolymer
Because of the cross-linked feature mentioned above, hardened geopolymer has a

denser microstructure than hardened OPC. The C-A-S-H gel chains in geopolymer
system were found longer than the C-S-H gel chains in OPC system, due to the
substitution of Al3+ for Si4+ in bridging positions. C-A-S-H gels had higher Al/Si
18
ratio and lower Ca/Si ratio than C-S-H, and it was indicative of the co-existence of
tobermorite 1.4 nm, with a chain length of 11, and tobermorite 1.1 nm, with a chain
length of 14 tetrahedra. It was found that the Young’s modulus for perfect
tobermorite 1.1 nm and tobermorite 1.4 nm was 77.3 and 49.9 GPa, respectively.
This would be because of the higher interlayer cohesion provided by the bonding
between bridging tetrahedral of conservative layers [55].
Figure 2.6 shows a ternary plot of gel composition of geopolymer and OPC. When
comparing the gel compositions from geopolymer and OPC systems, the main
difference is that the C-S-H forming in OPC has a higher Ca and lower Al content
than the C-A-S-H forming in slag-based geopolymer. When fly ash is presented in
the system of geopolymer, N-(C)-A-S-H can be identified with low Ca content, but
pure N-A-S-H is only stable at pH lower than 12 [52, 56].
Figure 2.6: Ternary plot of binder gel compositions of geopolymer (noted as

AA100S AA50S/50FA) and OPC (noted as 100OPC) (Source: van Deventer, et al.
(2015) [56])
19
Microstructure analysis of pore characteristic with the techniques, such as mercury
intrusion porosimetry (MIP) and gas adsorption, revealed that the pore size
distribution of slag-based geopolymer showed more mesopores (pore size < 50 nm)
and also lower porosity than those of OPC. A comparison of pore size distribution of
slag-based geopolymer paste (GPP) and OPC paste (OPCP) is shown in Figure 2.7.
Most of the pores in OPCP were distributed in the range between 10 and 100 nm,
whilst most of the pores in GPP were below 20 nm [57-61]. It was also found that
when SCM, such as slag and silica fume was used in OPCC, the pozzolanic reaction
forms a denser microstructure, results in high strength and low permeability [40, 62].
Study in porosity can help in understanding the behaviours and properties of GPC
and OPC, since it gave indication on the density of the microstructure, presence of
microcracks, the rate of diffusion of pore solution [63].
Figure 2.7: Pore size distribution of geopolymer and OPC (Source: Collins and
Sanjayan (2000) [60])
20
2.4 Relationship between microstructure and
properties of concrete
It is believed that the microstructure has an influence on the properties and

behaviours of the materials. The gels are the main strength giving component in
geopolymer and OPC products [56]. Besides, it was found that the pore structure has
a certain effect on a range of properties for OPCC, such as strength and permeability.
As pores are formed within the microstructure of C-S-H, size, shape, bonding and
packing arrangement of the C-S-H would affect the pore microstructure. Prediction
models, which include microstructure, or even nanostructure, can improve prediction
of the properties of hydrated OPC, which including strength, shrinkage, creep,
permeability and diffusion [64, 65].
2.4.1 Strength
The pore structure of concrete has an important influence on the strength of hardened
paste and thus on concrete. Generally, strength decreases as total porosity increases.
The relationships between porosity and strength of OPCP were summarised by
Rößler and Odler [66] as follows:
Balshin’s equation: 𝑓𝑐 = 𝑓𝑐0 (1 − 𝜙)𝑛 (2.1)
Ryshkevitch and Duckworth’s equation: 𝑓𝑐 = 𝑓𝑐0 𝑒 −𝑘𝑟 𝜙 (2.2)
Schiller’s equation: 𝑓𝑐 = 𝑘𝑆 𝑙𝑛(𝜙0𝑆 /𝜙) (2.3)
Hasselman’s equation: 𝑓𝑐 = 𝑓𝑐0 − 𝑘𝐻 𝜙 (2.4)

where fc is the compressive strength at porosity ϕ, fc0 is the compressive strength at
zero porosity, ϕ0S is the porosity at zero strength, and n, kr, kS and kH are the
experimental constant.
Shi [67] used equations 2.1 to 2.4 to predict the strength-porosity relationship in slag-
based GPM, and found that it followed different trend to OPCM due to the different
nature of their pore structures. These equations yielded less accurate results for GPM
as they were derived for OPCM primarily. However, he suggested that equation 2.4
21
was the most accurate to predict the strength-porosity relationship in both GPM and
OPCM. The strength-porosity relationship in GPM and OPCM is shown in Figure
2.8.
Figure 2.8: Strength-porosity relationship in GPM and OPCM (PC denotes OPC, NS,
NC and NH denote geopolymer with different activators) (Source: Shi (1996) [67])
The compressive strength of slag-based GPM was found to be higher with increasing
modulus and solution concentration. However, the flexural tensile strength did not
improve due to increasing of microcracks in the matrix phase and matrix-aggregate
interface. Aydın and Baradan [68] assessed two types of heat treatment and found
that autoclaving would be more effective than steam curing in strength development
and shrinkage reduction in slag-based GPM. This would be due to the formation of a
finer pore size distribution, lower Ca/Si ratio of C-S-H, denser and compact structure
of matrix. However, autoclaving caused reduction in strength for OPCM. Under the
condition of high temperature and pressure, the chemistry of hydration had been
substantially altered. C-S-H form was converted to a crystalline structure, α-calcium
silicate hydrate (α-C-S-H), which caused an increase in porosity.
22
Jambunathan et al. [69] suggested that the slow conversion of the hydration products
(CAH10, C2AH8 and C4AH13) to a more stable structure (C3AH6) at room temperature
had an effect on the loss of compressive strength in slag-based GPP. Since the latter
structure has a greater density than the former structure, the conversion resulted in an
increase of porosity, and thus deterioration in strength.
Collins and Sanjayan [70] summarised the effect of curing on strength of slag-based
GPC from a range of studies. The curing method mainly influenced the pore
structure and it can result in microcracks. For the GPC under exposed curing, a
network of microcracks was indicated. These microcracks in the specimen increased
the total porosity. The longer exposed specimens containing more microcracks
showed a reduction in strength. Furthermore, the strength loss was larger in the
smaller GPC exposed specimen. The pore distribution showed coarser pore size
toward the outer surface in the large specimen (150 mm dia.), so the higher loss in
strength in the small specimen was due to larger portion of coarser pore size.
Brough and Atkinson [71] investigated the evolution of the paste/aggregate

interfacial transition zone (ITZ) in slag-based GPM. Initially the anhydrous material
was packed against the sand grains, leaving a narrow interfacial zone of about 20 μm
width containing less anhydrous slag than the regions of bulk paste. This zone was
filled up as hydration process showing an improved microstructure at ages after 14
days. Therefore, the slag-based GPM had much lower interfacial porosity than the
OPCM, and the strength of slag-based GPM was higher than that of OPCM. Similar
result was reported by San Nicolas and Provis [72].
2.4.2 Permeability and durability
Gas and water permeability are of interest in studying the durability of concrete,
since they are related to the mass transport within concrete. There are two types of
pore in the porous materials; open and closed. Open pores refer to pores with various
size connected to each other and have an end open at surface of the materials, but
closed pores do not. Some porosimetry analysis methods, such as MIP and gas
adsorption, can only be used in access to the open pores. Although there is such
23
limitation, the analysis on these connected open pores is critical to the study of the
permeability and diffusivity of concrete [63].
Durability of slag-based GPC has been studied in the past few decades, covering
topics on sulphate attack, acid attack, chloride resistance, carbonation, resistance to
corrosion of steel reinforcement, alkali silica reaction (ASR), high-temperature and
fire resistance, and freezing and thawing resistance. Immersing slag-based GPC in
sulphate solution, acetic acid and sea water showed little negative effect on GPC and
GPC exhibited better performance than OPCC [4, 5, 31, 73, 74]. Thus the
degradation of strength of GPC was lower and the resistance to corrosion of steel
reinforcement was higher for GPC [75, 76]. Bakharev, Sanjayan and Cheng [77] and
Byfores et al. [4] reported that slag-based GPC had lower resistance to carbonation
than that of OPCC. However, this would be related to the relative humidity (RH) in
where the specimens were exposed. High RH in the saturated slag-based GPC would
retard the carbonation in early stage, the carbonation is faster when slag-based GPC
when drying occurs and proceed from the surface [78]. The ASR in GPC was not
well known. Bakharev, Sanjayan and Cheng [79] found that GPC expanded more
than OPCC under ASR, but Fernández-Jiménez and Puertas [80] found the opposite
way.
At high temperature above 300 °C, OPC chemically dehydrates while geopolymer
remains chemically stable up to 800 °C. This makes geopolymer a potential option in
fire-proofing for construction purposes. However, spalling was report in some GPC
under elevated temperature, and this was related to the size of the specimens and the
size of aggregates [28, 81-83]. On the other hand, GPC has excellence freeze-thaw
durability but it is affected by the air entrainment in the concrete [4, 5, 84].
The rate of most of the above degradation processes are controlled by the rate of
diffusion of the species of the penetrating chemical and the viscosity of the solution
in the pores. Over years, studies have been carried on to investigate the relationship
between pore characteristic and the permeability. Powers, Mann and Copeland [85]
reported the relationship between the pore structure and the permeability of OPCP.
Only the evaporable free water moved in the pore network. The viscosity of fluid, as
stated in Darcy’s law, also influences the flow of fluid through the specimen. The
24
viscosity of the pore solution is higher than water due to the presence of alkali, and it
varied with the concentration and temperature. The more alkali exists in the pore
solution, the higher viscosity it is. They also pointed out that the creep of concrete
was a result of redistribution of evaporable water in the concrete member under
change of stress.
Pore characteristic of concrete can be obtained through a number of methods. The

most common methods are MIP and gas adsorption test. Damage would be induced
to the sample during MIP. Experiment on OPC samples showed a higher
connectivity, lower number of pores and greater average pore size after intrusion.
This would be due to the barriers of hydration product within the paste was destroyed
during intrusion due to the pressure difference [86]. The analysis can be also be done
using image analysis method, but the range of pore sizes is limited by the resolution
of the image [87, 88]
Kondraivendhan, Divsholi and Teng [89] conducted MIP on concrete broken chunks
from the centre of specimen crushed with hammer. The permeability was estimated
through the porosity and the mean pore size by using model by Pradhan, Nagesh and
Bhattacharjee [90]. Their model predicted the hydraulic diffusivity from pore size
distribution. It was applied on unsaturated concrete in which the hydraulic diffusivity
values were calculated for ideal continuous wetting and ideal continuous drying
representing the upper and lower bounds, respectively. Such range can be in 3
magnitude orders. All practical values of hydraulic diffusivity were likely to lie
within these bounds.
Nitrogen gas adsorption test has been used to examine the pore characteristic of
OPCP at pore size around 1 to 40 nm [91]. Unlike MIP, gas adsorption test is more
gentle to the sample when assessing the small pore sizes [63]. Therefore, gas
adsorption test can be a better method to investigate the gel pores. However, the
effectiveness of this method relies on the pretreatment method. It was pointed out
that oven drying at 105 °C would increase the chance of pore collapsing and results
in less surface area in measurement [91]. Solvent exchange with organic liquid is
also an alternative in removing water from the pore structure by counter-diffusion
process, but this process is slow and takes a long period [92].
25
It was found that the lower the porosity and the finer the pore structure, the lower the
permeability. The rate of the degradation processes can be lower [57, 67]. The
permeability can be determined based on Darcy’s law:
𝜅𝐴 ∆𝑃
𝑄= × (2.5)
𝜂 𝐿
where Q (m3/s) is the flow, κ (m2) is the permeability of the medium, A (m2) is the
cross-sectional area of the pore, 𝜂 (Pa∙s) is the viscosity of the fluid, and ΔP (Pa) is
the pressure drop over the length L (m).
Besides Darcy’s law, Hagen-Poiseuille law, as known as Poiseuille law, can also be
used to analyse the flow through porous medium. It assumes that the Newtonian fluid
flow through a cylindrical pipe of constant cross-section. The Hagen-Poiseuille law
is presented as the following equation:
𝜋𝑟 4∆𝑃
𝑄= (2.6)
8𝜂𝐿
where r (m) is radius of the cylindrical pipe. Hagen-Poiseuille law was used in
prediction of permeability of harden OPCP. However, the pores inside harden OPCP
were not constant, but distributed in a wide range. Hughes [93] developed a model to
investigate the relationship between pore size distribution and the permeability based
on Hagen-Poiseuille law. In order to account the influence of pore size, he divided
the pore size distribution into segments of log r = 0.05:
𝑟 2∆𝑃𝜙
𝑄 = 32𝜏2 𝜂ℎ (2.7)
where r expressed in Å, τ is the tortuosity factor and h (m) is the thickness of the
specimen. With the assumptions of the pores within each interval forming a discrete
package and no flow in pores size of less than 7.5 nm, the total flow was calculated
as sum of flow in each interval.
Holly, Hampton and Thomas [94] proposed a model of two dimensional network of
tubes which took pore size distributions in account for permeability of OPCP based
on Hagen-Poiseuille law. The network was of a square lattice arrangement.
8𝜂𝑏 𝑄𝑖
∆𝑃 = ∑𝐼𝑖=1 (2.8)
𝜋 𝑟𝑖2
26
where b is the proportionality constant of length to radius of the tube, Qi (m3/s) is the
flow corresponded to pore sizes of ri (m). They pointed out the importance of pore
size distribution in estimation of permeability, as it cannot be defined in term of an
average pore size and porosity. Different results were given by bimodal and
unimodal log normal distribution of pore size since flow through capillary pores was
favoured.
There are more methods in prediction of permeability by pore size distribution.

Garboczi [57] presented a review and discussed about a few methods, which include:
 Kozeny-Carman equation: It treats the pore structure as a collection of circular

cylindrical tubes that are arranged in parallel and have length, L, equal to the
sample thickness in the flow direction.
𝜋𝑟 4 𝜋𝑁<𝑟 4 > 𝜙3
𝜅 = ∑𝑁
𝑖=1 8𝐿2 =
𝑖
= 2𝑆 2 (2.9)
8𝐿2
where S is the specific surface area (SSA).
 Katz-Thompson permeability theory: It uses a critical pore diameter and a single

measurement of the conductivity of the fluid-saturated porous sample as inputs.
κ = c dc 2 / F (2.10)
where c is a constant = 1/226, dc is the critical pore diameter obtained by MIP

and F is the formation factor.
He suggested that a good pore structure and transport theory must be based on
experimental characterizations of the pore structure that are reproducible, simply
interpreted and directly relevant to transport properties. Moreover, it had to be able to
formulate the measured pore parameters into a prediction using mathematical
principles and be applicable to a single sample. Therefore, he suggested that Katz-
Thompson permeability theory fit the above criteria.
Aït-Mokhtar et al. [95] proposed a new model based on the use of a tridimensional
cubic pore network. The model was based on the assumption of bimodal pore size
distribution.
27
𝜁𝜙
𝜅 = 24𝜏2 𝑅2 2 (2.11)
where ζ is the constrictivity factor and R2 is the pore radius corresponding to the
minimal pore section. Since the cement-based materials can have a more complex
pore size distribution, this model was further developed for polymodal of pore size
distribution [96].
𝜁𝜙
𝜅 = 24𝜏2 𝑅𝑛 2 (2.12)
The equation is similar to equation 2.11, but the model contains n number of section.
Rn is the critical radius of pore corresponding to the smallest section Sn, and the
calculation of the constrictivity factor ζ is also different.
Atzeni, Pia and Sanna [97] developed a geometric fractal model based on a unit of
volume made from a rectangular base, with sides of a ratio of 2:1 from which 25
rectangular elements were replicated times along the z axis. In their work, 2 groups
of pore diameter were defined from the samples; one large and one small size. The
permeability of OPC pastes was calculated from the equation:
2−𝐷 𝜙
𝜅 = 32(4−𝐷𝑓 𝑅2 (2.13)
𝑓 ) 1−𝜙
where Df is the fractal dimension and ϕ was calculated based on the geometric fractal
model.
Besides, some researchers proposed the empirical expression to calculate the water
permeability coefficient with the pore structure parameters, including apparent and
total porosity, volume of micropores, pore surface area, total pore radius median and
micropore radius median [98-101]. Kumar and Bhattacharjee [101] found that using
mean distribution pore radius, rm, yielded a better correlation than using equivalent
pore radius, re. Defining re as a uni-sized pore radius of a concrete medium would
lead to the same permeability of concrete, when considered a medium having a series
of cylindrical parallel pores of varying radii. re can be calculated by:
28
𝑑𝑣
∫𝑑(𝑙𝑜𝑔𝑟)𝑟 2𝑑(𝑙𝑜𝑔𝑟)
𝑟𝑒2 = 𝑑𝑣 (2.14)
∫𝑑(𝑙𝑜𝑔𝑟) 𝑑(𝑙𝑜𝑔𝑟)
And the mean distribution pore radius, rm, is defined as the mean pore size and is the
weighted average, weighted with respect to the volume of pores at the corresponding
pore radius.
∑𝑛
𝑖=1 𝑉𝑖 𝑙𝑛𝑟𝑖
𝑙𝑛𝑟𝑚 = ∑𝑛
(2.15)
𝑖=1 𝑉𝑖
where Vi is the pore volume corresponding to the pore radius ri.
2.4.3 Shrinkage
The pore structure and the diffusivity also play an important role on shrinkage of
concrete. There are two major parts of shrinkage in concrete: autogenous shrinkage
and drying shrinkage. Autogenous shrinkage occurs due to a capillary shrinkage
related to water consumption through hydration, and drying shrinkage occurs as a
reduction of volume due to loss of water from the pores to the external environment
by the gradient of humidity [102, 103]. The size of the pores influenced the diffusion
process because the resistance was larger for smaller pores. Thus it takes longer time
for the drying process [104]. Collins and Sanjayan [27, 60] found that the drying
shrinkage of slag-based GPC was higher than that of OPCC. However, the water loss
in OPCC measured was higher than that in GPC after drying. They related this to the
difference of pore size distribution between OPCP and GPP in concrete. GPP has a
finer pore size distribution than OPCP and a much higher proportion of pore size
within the mesopore (1.25 to 25 nm) limits than OPCP. GPP was found to have the
larger capillary tensile forces set up at the meniscus, resulting higher the resulting
shrinkage. Similar to drying shrinkage, creep of concrete depended on the magnitude
of flux of the local microdiffusion of water between the capillary pores and the
adjacent gel pores under external pressure and drying process [105].
Drying general induced microcracking in concrete and reduced both compressive and
tensile strengths, but tensile creep reduced the risk of cracking in term of stress
relaxation [27]. de Sa et al. [106] suggested that the effect of creep, effect of
29
aggregate restraint, concrete mix and capillary pressure should be taken into account
for more accurate prediction of microcracking and of the residual mechanical
properties after drying.
The w/c ratio, paste volume, size effect and RH also influence the drying shrinkage
of concrete. The shrinkage strain is directly proportional to the paste volume in
concrete. Although the sample size does not have an effect on the ultimate drying
shrinkage strain, it influences the strain rate. The drying shrinkage is approximately
inversely proportional to the RH in the range from 48% to 100% [107].
2.4.4 Analysis for pore characteristic
Since the pore characteristic is crucial in the analysis of the material properties, it is
important to obtain the pore characteristic from the material. There are a number of
methods to analyse test result from gas adsorption test for pore characteristic. One of
the important parameters is specific surface area (SSA). In 1918, Langmuir [108]
proposed the Langmuir theory, which assumed that the adsorbate behaved as an ideal
gas and the amount of gas adsorbed on the solid surface was confined to a
monomolecular layer. This layer is uniform and there is no lateral interaction
between adsorbed molecules. Equations were derived as:
𝑃 1 𝑃
=𝑘 +𝑉 (2.16)
𝑉 𝐿 𝑉𝑚 𝑚
𝑉𝑚 𝑁𝐴 𝐴𝑚
and 𝑆= × 10−20 (2.17)
𝑉𝑀
where V is the volume of adsorbed vapour at pressure P, Vm is the volume of gas

adsorbed when the entire surface is covered by a monomolecular layer, kL is the
Langmuir constant, NA is the Avogadro constant, Am is the average area occupied by
one molecule of adsorbate in the completed monolayer, and VM is the molar volume.
After that, Brunauer et al. [109] extended the Langmuir’s monolayer theory to
multilayer adsorption. Their theory is called BET theory, which stands for the name
of three authors (Brunauer-Emmett-Teller). The BET theory assumes that the solid
30
surface is energetically homogeneous and there is no variation in properties of
adsorbed layers after the first layer. The BET equation is shown as follow.
𝑃
𝑉𝑚 𝐶( )
𝑃𝑜
𝑉= 𝑃 (2.18)
1+𝐶( )
𝑃𝑜
𝐶 = 𝑒 (𝑞1 −𝑞𝐿 )/(𝑅𝑇)
𝐶−1
and 𝑆=𝑉 (2.19)
𝑚𝐶
where Po is the saturated vapour pressure, C is the BET constant, q1 is the heat of
adsorption of the first layer, qL is the heat of liquefaction of the liquid adsorption, R
is the gas constant and T is the temperature.
Furthermore, the pore size distribution in mesopore sizes can be obtained through
methods such as Barrett-Joyner-Halenda (BJH) method, Cranston-Inkley (CI)
method and Dollimore-Heal (DH) method. These methods are based on the
assumption that the mesopores have cylinder shape and adopt the Kelvin equation,
which relates the size of a pore to the relative pressure [110]. For nitrogen as the
adsorbate, the Kelvin equation may be written as:
−9.593
𝑟𝐾 = 𝑃 × 10−10 (2.20)
𝑙𝑛( )
𝑃𝑜
where rK is the radius of the pore in which condensation occurs. Using the Kelvin
equation, BJH method divides the pore sizes into groups and assumes all pores in
each group of capillaries have an average radius. It considers the isotherm as a series
of steps of pressure change. After the pore size distribution is obtained, the pore
volume and the surface area can be calculated by adding values from the pore size
distribution [111]. However, the results varies with the use of adsorption and
desorption isotherm. Similar to BJH method, CI method also divides the pore sizes
into groups from 1 to 30 nm. It assumes that the pores have one end closed. The
adsorption isotherm is used in the analysis to give more precise result than BJH
method. The pore sizes are calculated when the pores are filled due to multilayer
adsorption and capillary condensation under increasing pressure [112]. DH method is
a simple method compared to BJH method and CI method. It also assumes that the
31
pores have one end closed. The analysis is done using the desorption isotherm, as the
critical radius of pores is calculated when the total volume of pores which are still
filled with adsorbate after each step of desorption decreases [113].
Since cementitious materials contain micropores, which sizes are below 2 nm, the
above methods are not applicable in these sizes. Micropore (MP) analysis method by
Mikhail et al. [114] can be used for analysis of pore volume and pore surface
distributions of micropores. Adsorption isotherm is used in this method. A t-plot is
required in order to use the MP method.
2.5 Creep of concretes
2.5.1 Creep
Creep in concrete is a phenomenon that increases strain when concrete is subjected to

a sustained stress. It is a time-dependent behaviour. The increase in strain causes
increase in deformation and curvature, losses of prestress and redistribution of
stresses and internal action, and thus leads to serviceability issue [10].
Similar to shrinkage, there are two types of creep: basic creep and drying creep.
When a specimen is loaded in hygral equilibrium with the ambient medium, the
time-depended deformation caused by stress is known as basic creep. Additional
creep occurs in drying specimens, known as drying creep [6, 10].
Total strain, ε(t), is usually measured from the creep test along the test period. It is a
combination of the creep strain, εcr(t), the instantaneous strain, εe, and the shrinkage
strain, εsh(t) [10]. Therefore, the calculation of creep strain is:
𝜀𝑐𝑟 (𝑡) = 𝜀 (𝑡) − 𝜀𝑒 (𝑡0 ) − 𝜀𝑠ℎ (𝑡) (2.21)
There are two common ways to express creep in concrete. One of them is known as
specific creep, which is the creep strain per unit of applied stress. This can be
expressed in a compliance function, J(t,t0), which is the proportionality factor
relating stress, σc(t0), to linear creep:
32
1 𝜀 (𝑡)
𝐽(𝑡, 𝑡0 ) = 𝐸(𝑡 ) + 𝜎𝑐𝑟(𝑡 (2.22)
0 𝑐 0)
Where E(t0) is the elastic modulus at time of loading, t0. Another one is the creep
coefficient, φ(t,t0), which is the ratio of creep strain to instantaneous strain:
𝜀𝑐𝑟(𝑡)
𝜑(𝑡, 𝑡0 ) = (2.23)
𝜀𝑒 (𝑡0 )
2.5.2 Creep of OPCC
Creep of OPCC has been studied for a long period. A lot of research has been carried
out to investigate the nature of creep in order to predict the time-dependent behaviour
accurately. The creep of concrete is influenced by a number of factors, including
stress/strength ratio, maturity of the concrete when loaded, cement content,
water/cement ratio, relative humidity, dimensions of member, and temperature [115].
The research on creep of OPCC showed that creep develops quickly in the early
period, then the rate of creep decreases until very small in 2-3 years (see Figure 2.9)
[10]. However, a final creep value is still yet to be observed [12].
Creep has a linear relationship to the sustained stress up to a certain stress level,
which is believed to be 40% of the ultimate strength. Above this stress level, non-
linear creep starts to develop. The non-linear creep increases the total creep response
in concrete. If the sustained stress is up to 80-90% of the strength, failure would
occur during the loading period and called “creep rupture”. Creep rupture would be
due to the development and growth of microcracking under high sustained load
[116].
33
Creep compliance
Creep compliance
Creep recovery Creep recovery
Creep compliance
Creep recovery
Figure 2.9: Typical development of creep in OPCC (adapted from Bažant and Kim
(1978) [117])
When OPCC is loaded at early age, it exhibits high basic creep and drying creep
[118]. This is the effect of age of loading. Bažant and Kim [117] pointed out that the
creep curves of concrete for different ages of loading usually first converge but after
a longer creep duration they start to diverge with the early-age loaded concrete
yielded higher creep. De Schutter [119] proposed a relationship between the degree
of hydration and the basic creep behaviour in a Kelvin chain model, in which the
input parameters of Young’s modulus (spring stiffness) and dumper viscosity were
related to the degree of hydration. It shows the final creep coefficient decreases as
degree of hydration increases.
The composition of OPC and the water to cement ratio (w/c) would influence the
hydration process. Since the hydration product has a less volume than the combined
volume of water and cement, the remaining volume became pores after hydration and
free water and air would fill in the pores. Experiments showed that higher w/c in
OPCC resulted in lower strength and higher creep [120].
34
Drying has an impact on the creep behaviour of concrete as it induces additional
creep. It is also called “Pickett effect” as it was first pointed out by Pickett in 1942
[121]. Wittmann and Roelfstra [122] explained this effect as a result of difference in
moisture distribution from the outer drying surfaces to the centre of the specimen.
Therefore, tensile stress develops at the outer shell of the drying specimen under
loaded, which would cause crack formation and extra compressive load in the centre
of the specimen. The moisture migration from the centre of the specimen to the outer
shell is related to the diffusion equation. In addition, Bažant and Chern [105]
suggested another mechanism, stress-induced shrinkage, which involved micro-
diffusion of free water locally between the capillary pores and the gel pores. This
transport affected the deformation rate of the solid phases of C-S-H gel. It was also
found that the diffusion coefficient for drying concrete decreased when passing pore
humidity from 90% to 60% [102]. To prove this, Bažant and Yunping [123]
proposed an experiment that axial compression was applied with an eccentricity on
the specimens. They separated the effect of drying creep by measuring curvatures
between the sealed and the two-sides drying specimens. They found that the stress-
induced shrinkage increased with time. Because of the diffusion process, size of the
specimen has an effect on drying creep. The effect of drying creep on a larger
specimen is less, because the halftime of the drying process increases with increasing
specimen size. Therefore, an infinitely thick specimen would be similar to a
specimen under sealed condition [124].
Alexander, Wardlaw and Ivanusec [125] found that when the concrete was cured at
low temperature, it took longer time to achieve the level of maturity as the concrete
cured at higher temperature. This low-temperature cured concrete exhibited higher
creep strain. These factors can be combined to cause a large range of creep variation.
Moreover, concrete cured at a higher temperature is usually stronger and has a higher
modulus of elasticity; it has less free water to lose because of a more complete
hydration, and therefore exhibit less shrinkage [62].
Vidal et al. [126] compared the basic creep of HPC at 20, 50 and 80 °C and showed
that the basic creep increased with increasing temperature. The rate of this increase
was not linear. This could be due to thermal damage caused by thermal deformation
gradient between OPCP and aggregates above 50 °C. They also suggested that
35
change of water viscosity in the pore solution with temperature through an Arrhenius
law would be one of the causes because of its probable effect on the slippage of C-S-
H. Similar result was reported by Nasser and Neville [127], but they found that creep
increased, from 70 ºF (= 21.1 ºC) to 160 ºF (= 71.1 ºC), but it decreased at 205 ºF (=
96.1 ºC). They explained that the adsorbed water began to evaporate at certain
temperature so that the rate of creep decreased. Further details of creep behaviour at
elevated temperature will be presented in section 2.5.5.
Using round gravel resulted in higher creep as the smooth surface caused a weaker
aggregate-paste interface. Greater shrinkage strain was also observed for HPC made
with round gravel for the same reason. However, the effect of the type of coarse
aggregate can be reduced by moist curing [62].
When the sustained load on concrete is removed, the instantaneous recovery of strain
occurs first, followed by an additional gradual recovery over a period. This is called
creep recovery. The sum of both recovery of strain was found not equal to the total
deformation caused by instantaneous load and creep, leaving an irreversible
deformation as residual deformation. Part of the reason is that the instantaneous
recovery corresponded to the modulus of elasticity at the time of load removal.
However, creep recovery cannot be related to the strength. It may be due to the
inward movement of water in the cement gel released from load. The creep recovery
is usually completed in several weeks, because this movement of water is rapid in
comparison with creep. The rigidity of the OPCP as a whole would be the governing
factor of creep recovery [128].
The mechanism of creep is not well known yet. Research has been done and theories,
such as seepage theory, rate-determining process and structure-deforming process,
were proposed. Feldman [129] conducted a series of experiments on OPCP samples
under rewetting-drying cycles and showed that each time the sample was exposed to
more than 50% relative humidity (RH), its properties altered. It possessed the
potential for irreversible shrinkage when d-dried. He suggested that the potential of
creep and the potential for irreversible shrinkage were manifestations of the same
change that had occurred in the solid. Furthermore, he pointed out that the hindered
adsorption or disjoining pressure was not connected to the creep process since little
36
of potential irreversible shrinkage had yet occurred at 10% RH and an increase in
creep rate was observed when the samples were being wetted from 50% to 100%
RH. He then concluded that creep was a combination of rate-determining process and
activating process. The former indicates adsorbed water seepage from the entrances
of interlayer spaces and areas of close proximity of two sheets control the initial
rapid creep rate while shear slippage, microcracking, bond breaking and reforming
are involved in longer-term creep. The latter suggests that the increasing creep rates
at specific regions of increasing shear or tensile stresses is created by compressive
stress or drying.
Bažant [6] described several mechanisms causing or influencing creep:
1. Plastic flow.
2. Consolidation theory.
3. Loading bearing hindered adsorbed water.
4. Bond breakage in slip and its reformation.
5. Nonlinear deformations and cracking as a contribution to Pickett effect
(drying creep).
6. Solidification theory for short-term aging.
7. Microprestress of creep sites in cement gel microstructure, causing the Pickett
effect and long-term aging.
Part of the creep effect was assumed to be due to the diffusion of water and other
molecules along the micropores towards the macropores in the cement gel
microstructure under applied stress based on thermodynamic in the early stage of
research. This involves the migration of water molecules and the dissolved adsorbent
(e.g. calcium ion) along the hindered adsorbed layers (less than about ten molecules,
26.3 Ǻ). The layer thickness decreases as these molecules moved out, which causes
creep. The rate of diffusion is related to the thickness and the distance of the
hindered adsorbed layer. It took the longest time for the longest or the thinnest
adsorbed layers. This would take more than 30 years in such layers but perhaps a
lapse for the short or thick layers [130]. Similarly, Acker and Ulm [8] explained that
creep of concrete involved two type of mechanisms. The first mechanism is due to
the stress-induced water movement towards the largest capillary pores, which
resulted in the short-term component of creep. The second mechanism is related to
37
viscous flow in the C-S-H sheets of hydrated cement paste for the long-term creep.
They also pointed out the importance role of water in creep, as the basic creep
reduced with lower evaporable water content. The effect of lower RH in the
environment on creep is due to greater water movement. When OPCC was
completely dried before subjected to loading, it would exhibit negligible creep.
Hope and Brown [131] proposed that the mechanism of creep of OPCC was related
to the increased interlayer volume under sustained stress. This was a result of the
physical distortion and general reduction in spacing of the solid sheets of cement
hydrate. Their experimental results showed higher interlayer porosity in loaded
specimens than in the unloaded specimens. Partial drying induced the movement of
the interlayer water and accelerated the rate of creep. Recently with advance
technology, Vandamme and Ulm [7] hypothesised that C-S-H creep was due to nano-
particle sliding. They conducted nanoindentation creep experiments on micrometer-
sized volume of materials and proposed that the sliding of gels led to a local increase
of the packing density toward the jammed state. The volume changes were small
with limit packing density when approaching the jammed state. This suggested the
dependence of creep rate on the packing density. From the experiments on three
distinct C-S-H phases, they found that the denser C-S-H phases showed the lower
creep rates.
When the concrete member is under restrained condition, cracking is developed as

the tensile stress induced by restrained shrinkage exceeds the ultimate tensile
strength. In this case, tensile creep plays an important role in stress relaxation, and
hence the crack tendency. The crack tendency can be assessed by casting the test
specimen around the outside of a steel cylinder (ring test) [132]. However, Altoubat
and Lange [133] developed an uniaxial restrained shrinkage test with specimens of
dog-bone shape. The test results indicated that the higher strength concrete failed
(cracking occurs) earlier than lower strength concrete, but the tensile stress at failure
was lower than the tensile strength indicated by splitting test (indirect tensile test)
and direct tensile test. This was due to the high stress at the early age caused the
permanent damage at the micro-level in the samples, especially for high strength
concrete which experienced high stress due to rapid strength development. Other
causes of the tensile strength reduction would be static fatigue and internal damage
38
accumulation under sustained loads from restrained shrinkage. They suggested that
tensile creep doubles the shrinkage cracking capacity of the restrained concrete,
which delayed the failure time, and the tensile creep progressed more slowly under
gradually increasing loads.
It is usually assumed that the tensile creep is nearly the same as compressive creep
[115]. In order to investigate tensile creep of concrete, custom designs of loading rigs
are required, due to the difficulties in secure the test samples in uniaxial tension and
load application [134-136]. Because of that, the test results varies. Brooks and
Neville [136] compared the compressive creep and the tensile creep of concrete and
found that the initial rates of basic creep in tension and in compression were similar
but the rate of tensile basic creep was higher than that of compressive basic creep in
later ages. Similar founding was observed in the rates of total creep in tension and in
compression. However, when the specimens were loaded after drying for 28 days,
the total tensile creep was less than the total compressive creep. On the other hand,
Rossi, Tailhan and Le Maou [135] found the basic compressive creep was higher
than the basic tensile creep in both young and old concrete specimens, but they
reported a larger amount of drying tensile creep than drying compressive creep. They
suggested that the density of microcrack was higher in tension than in compression,
due to load application, sustained load and drying, and the density of cracks from the
surface of the specimens induced large strain. The tensile creep measured from the
direct tensile creep test can be affected by damage from microcracking and
additional shrinkage. This leads to a contracting tendency if the shrinkage from the
companion specimens is used in calculation of drying tensile creep. One of the ways
to overcome is to obtain tensile creep from the flexural creep test, since it would give
a better result as the above effects in damage are less [137]. Ranaivomanana, Multon
and Turatsinze [138] suggested that the basic creep strains were the same in
compression and tension in the bending specimens under sealed condition.
Therefore, flexural creep tests have been carried out by a number of researchers for
the investigation of creep of concrete [139, 140].
39
2.5.3 Creep of GPC
Compared to OPCC and blended cement concrete, there are limited research on creep
of GPC. Hardjito et al. [141] carried out a series of tests on the properties of heat
cured fly ash-based GPC. They reported that both creep and drying shrinkage strains
were lower than OPCC. The creep coefficient of fly ash-based GPC in 6 weeks of
sustained loads was 0.3. Further works were carried out by Wallah and Rangan [142]
on four series of specimen with two mixture details and two heat curing methods
(oven dry curing or steam curing). All specimens showed creep coefficient below 1
in a year of loading period. Sagoe-Crentsil, Brown and Taylor [143] also found lower
basic creep in fly ash-based GPC after steam cured than OPCC.
A GPC from a mixture of fly ash and metakaolin showed low creep coefficient as
0.609 at 60 days under sustained loads [144]. A short-term creep test was done on
GPC with mixture of GGBFS and fly ash by Lee [145]. The test results of one-month
test period showed that the total specific creep increased with increasing strength,
although the basic specific creep was similar between the specimens. The reason was
not known since the mixture detail was not reported. Another GPC mixture of fly
ash, Kaolite high-performance ash and GGBFS was tested by Castel et al. [146]. The
proportions of the above materials in the mixture were 8.2, 2.2 and 1.8 %wt,
respectively. The specimens cured for three days at 40 °C showed much higher creep
than those cured for seven days at 80 °C. The creep coefficient of the latter was about
0.2. However, such low creep may be a problematic in concrete structures as it
reduces the ability in relieving concrete stress concentrations and imparts
deformability to concrete.
For slag-based GPC, Collins and Sanjayan [27] found that although slag-based GPC
showed higher strength than OPCC, it exhibited higher drying shrinkage and greater
creep than OPCC. However, combined with the lower elastic modulus of slag-based
GPC, the greater creep would minimise the risk of cracking.
The mechanism of creep of GPC has not been studied. No test results on creep of
GPC above exceeds 1 year. Therefore, more long-term tests on GPC is required to
understand the complete creep behaviour of GPC. However, compared with the creep
40
of GPC, more works on shrinkage of GPC have been done in a number of research.
Some of the findings can be useful in understanding creep of GPC.
It was found that a number of factors influenced the shrinkage of geopolymer.

Ridtirud, Chindaprasirt and Pimraksa [147] found that the shrinkage of fly ash-based
GPM decreased with increasing curing temperature. This would be due to fast
geopolymerisation process resulting low shrinkage. This agreed with the general
results found by others in heat-cured fly ash-based GPC [141-143, 148]. Similar to
the effect of w/c of OPCM, higher water to binder materials ratio (w/b) of the GPM
gave higher shrinkage, because of the higher porosity of the hardened fly ash-based
GPM [147].
The drying shrinkage of slag-based GPC was found higher than that of OPCC. When
air curing was used, the slag-based GPC demonstrated lower compressive strength
than OPCC. This would be due to the effect of drying. This can be improved by
replacement of aggregates to porous coarse BFS aggregate. This can be attributed to
the internal curing effect whereby water from the aggregate was gradually released
into the concrete to further hydrate the paste. This can also explain the lower drying
shrinkage compared to the GPC with normal aggregate [149]. Because of the low
drying shrinkage, high tensile strength and low elastic modulus of slag-based GPC
the cracking tendency can be significantly improved [150, 151].
Neto, Cincotto and Repette [152] conducted research on autogenous shrinkage and
drying shrinkage of slag-based GPM and found that both shrinkages increased with
increasing amount of silica in the activator. The GPM with higher content of sodium
silicate resulted in more intense hydration (C-S-H formation), reducing porosity.
Therefore, it increased both autogenous shrinkage and drying shrinkage, with the
autogenous shrinkage contributing a significant amount in total shrinkage. On the
other hand, the reabsorption of the free water in the mix would cause the expansion
in GPM with low sodium silicate content. This movement of water generates an
increase in the RH of the pores, diminishing the capillary tension that causes
shrinkage.
When the replacement level of slag into fly ash in the GPP increases, the paste had
fewer pores and more compacted structure. However, it also increases the chemical
41
and autogenous shrinkage amounts. The drying shrinkage of GPP depends on the w/b
and the concentration of the alkali activator. Compared with OPCP, this GPP has
higher mesopore volume and hence the autogenous shrinkage of GPP is higher, due
to the self-desiccation resulting from a decrease in the RH within the capillary pores
[153].
2.5.4 Creep of blended cement concrete
Because there is little research on creep of GPC, creep of blended cement concrete
with SCM is also studied to provide some insight on creep behaviour of GPC.
Alexander et al. [125] found that fly ash blended concrete exhibited lower creep than
OPCC under the same condition. This is similar to the experimental results by
Bilodeau and Malhotra [154] on high volume fly ash concrete with very low water
content (w/c = 0.275 to 0.397). Part of this would be because of the improved
mechanical properties by pozzolanic reaction. However, the low drying shrinkage
and creep strains would also be due to low water content of this concrete. The above
results show similar trend of creep to the flay-ash based GPC.
When high calcium fly ash was blended with OPC, the compressive strength and
flexural strength of mortar reduced with increasing replacement. It also exhibited
lower shrinkage than OPCC, but it would be due to the high SO3 content in the fly
ash, causing expansion [155]. In addition, OPCC with higher fly ash replacement
showed lower autogenous shrinkage, while concrete with higher GGBFS
replacement showed higher autogenous shrinkage [156].
Brooks and Megat Johari [157] reported that partially replacement of cement with
metakolin can reduce the total shrinkage and creep in concrete. In this case, the
acceleration in cement hydration, the filler effect and the pozzolanic reaction of
metakolin with Ca(OH)2, play an important role in creep improvement. This also
suggests that the fineness of the material can be a factor in refinement of pore
structure, which influences the movement of pore solution and initial creep
(metakolin with SSA = 15,000 m2/kg).
42
Creep experiment on concretes made with BFS and OPC by Chern and Chan [158]
showed that the basic creep decreases with increasing BFS replacement, but the total
creep increased under simultaneous loading and drying condition. They related the
effect on basic creep to the greater strength development by pozzolanic reaction of
the amount of BFS, and thus lower mean stress in the cement matrix. However, the
increase in total creep reflected the impact of drying to concretes made with BFS and
OPC. They explained that it would be due to the premature damage by loss of
moisture from hydration and pozzolanic reaction, microcracking caused by high
drying shrinkage and stronger “Pickett effect” than OPCC. Shariq, Prasad and Abbas
[159] also found similar result of high creep and drying shrinkage in concrete
containing GGBFS. They suggested that the results would be due to the slow gain of
strength for GGBFS concretes that leaded to more free water availability during the
period of sustained load, and more migration of water under external load caused
higher creep strain. They also found the creep recovery was lower for GGBFS
concrete than OPCC. This would be because of the growth of strength in GGBFS
concrete at later ages. However, Li and Yao [160] reported an opposite trend in creep
and shrinkage of HPC with BFS and silica fume. They found that both creep and
drying shrinkage were reduced. It would be due to the fast reaction with ultrafine
particles of BFS (SSA = 800 m2/kg) and silica fume (SSA = 20,000 m2/kg), which
increased the amount of AFt hydrates and C-S-H gel hydrates in early age, and fill
the small pores and voids.
The autogenous shrinkage of blended cement with GGBFS was influenced by the
mixture proportions and the fineness of GGBFS. Test results revealed that the 50%
GGBFS concrete had the highest autogenous shrinkage among 0 to 80% GGBFS.
This suggests that the 50% GGBFS replacement level can lead to the highest degree
of hydration, since when the replacement level was above 50%, there was inadequate
amount of Ca(OH)2 being produced by the hydration of OPC, which was needed for
the pozzolanic reaction in BFS. Increasing fineness also increases autogenous
shrinkage, as the rate of reaction is faster. The fine and rough textured GGBFS
particles are able to adsorb more water. With less free water in the system, self-
desiccation in the cementitious matrix results in greater capillary pore pressure,
which leads to a larger autogenous shrinkage [161].
43
2.5.5 Creep at high temperature
The total creep increases with increasing temperature [127]. It exceeds the expected
increase with temperature of basic creep and elastic strain. This additional strain is
termed transitional thermal creep (TTC), which results in relaxation and
redistribution of thermal stresses, rendering the elastic stress analysis inappropriate
for structures heated for the first time. The term “transient creep” is used to denote
the combined effects of TTC and drying creep. When the concrete specimen is heat
up, the rate of heating is critical as high rate would cause cracking and damage to the
specimen by thermal stresses due to the thermal gradient between the surface and the
inner part, so called structural effect. However, more moisture loss was observed in
slow heating rate [162].
Thermal creep strain can be explained and predicted in terms of the free thermal
strain (FTS) of unloaded concrete and the load-induced thermal strain (LITS) of
loaded concrete in first time heating. The LITS is measured by the difference
between the total thermal strain of loaded specimen and FTS. Khoury, Grainger and
Sullivan [163] suggested that a “master” LITS curve existed for the C-S-H based
concretes for temperatures up to 450 °C, provided that the stress/strength ratio,
preheating condition, heating rate and curing regime were similar. The various
materials behaved differently above 450 °C, due to the damage in concrete during
first heating. Following that, Dias, Khoury and Sullivan [164] investigated the basic
creep of OPCP under elevated temperature. They suggested that the basic creep at
first heating to a given temperature can only be obtained if the load is applied after
the achievement of thermal, hygral, chemical and dimensional stability at that
temperature. To achieve this, they heated up the specimens to the target temperature
and kept constant for a certain time. Creep test was started after the achievement of
dimensional stability, which was after the shrinkage phase. Power law was found to
be best-fit to the basic creep in this case. However, large increase in creep of OPCP
was observed above 600 °C, which set a temperature limit to OPCP independent of
w/c ratio. This may be due to atomic self-diffusion. They also found that preheating
the specimen at a higher temperature than the test temperature reduced the creep
response, since the TTC has been eliminated before load was applied.
44
The fly ash-based GPP shows a larger degree of transient creep than OPCP below
250 °C, then after that GPP did not show any pronounced increase but the transient
creep of OPCP increased significantly (see Figure 2.10). This may be due to the
molecular rearrangement associated with polysilicate formation and further
geopolymerisation below 250 °C. The absence of geopolymerisation above such
temperature leaded to no further increased in TTC. In the case for both concretes, the
strength of concrete reduced more than that of paste under elevated temperature, due
to the incompatibility of coefficient of expansion between the paste and aggregates.
Because the difference for geopolymer to aggregates was higher than that for OPCC,
GPC suffered more loss in strength even its strength and elastic modulus of GPP
increased more than OPCP at elevated temperature. This resulted in similar residual
strength in both OPCC and GPC [165].
Figure 2.10: Transitional thermal creep: (a) GPP (b) OPCP (source: Pan et al. (2014)
[165]
45
2.6 Creep models and structural performance
2.6.1 Early development of creep model for OPCC
For the development of creep function model for concrete, Bažant and Wu [166]
proposed that the creep curves in the log(t−t’)-scale can be a horizontal line segment,
followed by one or two inclined straight line segments. They fit the known data on
creep at constant temperature and water content by Dirichlet series. This
approximation is equivalent to the Kelvin chain model (see Figure 2.11a). The
Kelvin chain model consists a series of spring moduli and viscosity dampers in
parallel, which are age-dependent. Later on, they developed a rate-type creep law,
which can be interpreted by the Maxwell chain model (see Figure 2.11b). The
Maxwell chain model consists a parallel of a number of series of spring moduli and
viscosity dampers. In this case, the spring moduli and viscosities are age-dependent,
which indicates the correct form of the equations for time-variable springs and
dashpots. Therefore, the identification of material parameters from the test data was
simpler for Maxwell chain model [167].
η σ
E 1
1
E E E
1 2 3
E η
2 2
η η η
1 2 3
σ
σ
(a) (b)
Figure 2.11: Spring-dashpot models for viscoelastic materials: (a) Kelvin chain
model; and (b) Maxwell chain model
46
After that, Bažant and Osman [168] proposed the double power law for basic creep
of concrete and was described as below:
1 𝜙
𝐽(𝑡, 𝑡 ′ ) = 𝐸 + 𝐸 1 (𝑡 ′ −𝑚 + 𝛼)(𝑡 − 𝑡 ′ )𝑛 (2.24)
0 𝑜
where E0 is the asymptotic modulus, and n, m, α and ϕ1 are the material parameters.
Bažant and Chern [169] improved the double power law by transiting it to a straight
line in the logarithmic scale of creep duration at a certain transition time as they
found that the double power law overestimated the creep effect for longer duration of
loading. They called the proposed formula the double power logarithmic law:
𝑛𝜙𝐿 𝑡−𝑡 ′ 1+𝜙𝐿

𝐽(𝑡, 𝑡 ′ ) = ln + (2.25)
𝐸0 𝜃𝐿 𝐸0
𝜙𝐿 1/𝑛
for 𝜃𝐿 = ( ′ −𝑚 ) ≤ (𝑡 − 𝑡 ′ )
𝜙1(𝑡 +𝛼)
where ϕL is the material parameter, and θL is the transition time.
Bažant and Prasannan [170-172] then proposed the solidification theory which took
aging of concrete into account. This is because the laws of thermodynamics are only
for the systems of substances, whose properties do not vary in time, but the hydration
of cement continues after setting of OPC. This causes the aging of cement. The
solidification theory suggests that layers of solidified matter from a solution within
the pores gradually deposits on the surface of solidified matter forms previously (e.g.
on pore walls) at various time. It increases the effective load-bearing volumes
fraction of hydrated cement, but this growth terminates at about one month of age.
From this age onward, further bonds continue to form among the solid particles of
the hydrated cement. These change the microstructure and increase stiffness of
hydrated cement, thus the hydrated cement carries more loads. This also explains the
behaviour of creep recovery in hydrated OPC, in which some of the creep strain is
not recovered after the removal of load, due to the change of microstructure.
The solidification theory was proposed for the short-term aging in hydrated OPC
only, so microprestress-solidification theory was proposed for long-term aging. The
disjoining pressure and the crystal growth pressure in the micropores are carried by
47
the solid framework and the bridges or bonds between the opposite pore walls.
Therefore, these bridges or bonds are in large tension, and result a pretensioned state
to the microstructure. Creep in this case is described as shear slip of hindered
adsorbed layers. The interatomic bond breaks happen at the localised overstressed
creep sites and these bonds restore with the adjacent atoms. This results in relaxation
of shear stress. Such bond breakage and restoration redistribute stress toward
exhaustion of the overstressed creep sites, causing the declined creep rate under a
constant applied macrostress. It was also proposed that the local microdiffusion flux
of water molecules in the micropores accelerated the process of breakage of atomic
bonds. Therefore, when there is a change of humidity (e.g. drying to external
environment), there is additional creep caused by the rate change of relative humidity
in the capillary pores. This additional creep is known as drying creep [173, 174].
Based on these theories above and the data bank of creep of concrete, Bažant and
Murphy [11] developed Model B3 for creep and shrinkage prediction. This helps the
design of concrete structures for time-dependent effects. The calculation of the
compliance function contains three parts; the instantaneous strain due to unit stress
q1, the compliance function for basic creep C0(t,t0) and the additional compliance
function due to simultaneous drying Cd(t,t0,tc). Therefore, the creep effect under
various drying conditions can be calculated. Both compliance functions for basic
creep and drying creep were calculated separately as below.
𝑡
𝐶0 (𝑡, 𝑡0 ) = 𝑞2 𝑄(𝑡, 𝑡0 ) + 𝑞3 ln[1 + (𝑡, 𝑡0 )𝑛 ] + 𝑞4 ln⁡(𝑡 ) (2.26)
0
′ 1/2
𝐶𝑑 (𝑡, 𝑡0 , 𝑡𝑑 ) = 𝑞5 [𝑒 −8𝐻(𝑡) − 𝑒 −8𝐻(𝑡 0) ] (2.27)
where t0 is the age of concrete when loaded, td is the age of concrete when drying
commenced, the Q(t,t0) is a function for effect of age of loading, q2, q3, q4 are the
factors derived from the material properties including composition of mixture and
compressive strength, q5 is the factor considering compressive strength and
shrinkage, and H(t) and H(t’0) are the functions for humidity at loading period and
time of first drying, respectively.
48
Therefore,
𝐽(𝑡, 𝑡 ′ ) = 𝑞1 + 𝐶0 (𝑡, 𝑡 ′ ) + 𝐶𝑑 (𝑡, 𝑡 ′ , 𝑡0 ) (2.28)
and the creep coefficient is:
𝜑(𝑡, 𝑡 ′ ) = 𝐸 (𝑡 ′ )𝐽(𝑡, 𝑡 ′ ) − 1 (2.29)
The Model B3 provides a satisfactory prediction on creep of OPCC, but uncertainties

and variation in environmental conditions would affect the result. In this case,
improvement on the creep prediction can be done by the use of short-time test data,
and the factors q1, q2, q3, q4, and q5 are updated accordingly.
Recently, the Model B3 has been recalibrated to Model B4, aiming for prediction of
multi-decade of the structural response [175]. A large amount of creep and shrinkage
data from laboratory database and bridge structures were used in the calibration. The
main change in this model is on the separation of the predictions on autogenous
shrinkage and drying shrinkage. Further refinement on the model has been done on
the influence of shape factor on the drying process [176].
There were other researchers also working on prediction of creep of concrete. Brooks
and Neville [177] used the short-term result from testing to predict the long-term
creep and shrinkage. They suggested equations to predict basic creep, total creep and
shrinkage from 28-day values, but longer test duration would give more accurate
prediction. Gardner and Lockman [178] proposed a prediction model, identified as
GL2000, for shrinkage and creep of OPCC. In their model, the measured mean
concrete compressive strength, fcm28, is used to estimate the modulus of elasticity,
Ecmt, strength at different ages, fcmt, and shrinkage, εsh. The creep coefficient, φ28, is
calculated by the following equation:
(𝑡−𝑡0 )0.3 7 0.5 𝑡−𝑡0 0.5

𝜑28 = 𝛷(𝑡𝑑 ) [2 ((𝑡−𝑡 ) + (𝑡 ) (𝑡−𝑡 ) + 2.5(1 −
0 )0.3 +14 0 0 +7
0.5
2 𝑡−𝑡0
1.086𝑅𝐻 ) ( 𝑉 ) ] (2.30)
𝑡−𝑡0 +0.15( )2
𝑆
If t0 = td, Φ(tc) = 1
49
0.5 0.5
𝑡−𝑡0
When t0 > td, Φ(𝑡𝑐 ) = [1 − ( 𝑉 ) ]
𝑡−𝑡0 +0.15( )2
𝑆
where RH is the relative humidity as a decimal, and V/S is the volume-surface ratio.
They compared the results from their model to experimental results from RILEM
data bank and concluded that GL2000 showed a good agreement with the
experimental results. They suggested that improvement could be done by measuring
concrete strength development with time and modulus of elasticity, instead of using
the strength development equations.
Gawin, Pesavento and Schrefler [179, 180] developed a hygro-thermo-chemo-

mechanical model for shrinkage and creep of concrete based on the concept of
effective stress, which is often used in geomechanics. They modified the
microprestress-solidification theory by Bažant et al. [173, 174] and assumed that the
application of the external load acted as a kind of triggering mechanism which
activated the flow creep. The breakage of atomic bonds at microstructural level
initiates further shear slips in cement gel, and finally manifests at the macro-level as
creep strains. They considered aging in concrete as an increase in hydration degree
and coupled it with the material creep. Therefore, the creep compliance function can
be expressed as:
𝐽(𝑡, 𝑡 ′ ) = 𝑞 ′ 1 + 𝛷(𝑡, 𝑡 ′ ) + 𝑐𝑝𝑆 𝑝−1 (2.31)
1
𝑞′1 = 𝐸(𝛤 (2.32)
ℎ𝑦𝑑𝑟 )
𝜉
−
𝛷(𝑡, 𝑡 ′ ) = 𝛷 (𝜉 ) = ∑𝑁
𝜇 𝐴𝜇 (1 − 𝑒
𝜏𝜇
) (2.33)
1
𝑐𝑝𝑆 𝑝−1 = 𝜂(𝑆) (2.34)
where q’1 is the term related to elastic strains, in which the Young’s modulus E is
considered with the degree of hydration Γhydr, Φ(t,t’) is the micro-compliance
function expanded in Dirichlet series which corresponds to a Kelvin chain formed by
N non-aging units, Aμ = 1/Eμ, Eμ is the elastic modulus of μth Kelvin unit, ξ is the
load duration, S is the microprestress, c and p are material constants, and η is the
50
apparent macroscopic viscosity. Numerical simulation is required to compute the
creep strains. Their results showed that some additional autogenous strains, that
cannot be justified by the autogenous shrinkage phenomenon, can be explained as the
creep strains caused by the effective stresses.
Although there is a large amount of model for prediction of creep in OPCC, there is
yet any creep model for GPC. However, some work has been done for creep in
blended cement concrete. Some design models considered choices of input parameter
for fly ash or slag as SCM to concrete [175, 178]. Shariq, Prasad and Abbas [159]
suggested a modification factor to incorporate the influence of the content of GGBFS
on creep coefficient to the existing creep models. They proposed a new creep model,
which reflected the increasing trend in creep effect due to increasing amount of
GGBFS in concrete. However, the use of this model is not practical since it is based
on limited data of test results.
2.6.2 Creep models in standards
There are a number of creep model from different design standards and guidelines,
widely used over the world. In this section, four creep models will be presented,
including AS3600 by Standard Australia [181], Eurocode 2 by British Standards
Institution [182], ACI model by ACI Committee 209 [183], and fib model code 2010
by The International Federation for Structural Concrete [184], due to their usage in
practical field around the world.
2.6.2.1 AS3600
The creep of concrete can be estimated according to AS3600-2009, clause 3.1.8 for
the design of concrete structures in Australia. The design creep coefficient, φcc, is
calculated as the product of a number of factors (k2, k3, k4 and k5) and the basic value
for creep coefficient, φcc.b:
𝜑𝑐𝑐 = 𝑘2 𝑘3 𝑘4 𝑘5 𝜑𝑐𝑐.𝑏 (2.35)
51
φcc.b can be determined based on the characteristic strength of concrete. For the
factors, k2 is the time-dependent factor which is also influenced by the size of the
concrete member; k3 is related to the age of concrete at time of loading; k4 is the
environment factor and k5 is the modification factor of high strength concrete.
2.6.2.2 Eurocode 2
The details of calculation of creep of concrete in Eurocode 2 (BC EN 1992-1-1:2004)
refers to ANNEX B in the standard. Similar to AS3600, the design creep coefficient,
φ(t,t0), is calculated as the product of the notional creep coefficient, φ0, and the factor
which describes the development of creep with time, βc(t,t0):
𝜑(𝑡, 𝑡0 ) = 𝜑0 𝛽𝑐 (𝑡, 𝑡0 ) (2.36)
𝜑0 = 𝜑𝑅𝐻 𝛽(𝑓𝑐𝑚 )𝛽(𝑡0 )
(𝑡−𝑡0 ) 0.3
𝛽𝑐 (𝑡, 𝑡0 ) = [(𝛽 ]
𝐻 +𝑡−𝑡0 )
where φRH is the factor for the effect of relative humidity, β(fcm) is the factor based on
the mean compressive strength of concrete, fcm, β(t0) is the factor for the age of
concrete at time of loading, βH is the factor depending on the relative humidity.
2.6.2.3 ACI model

The calculation of creep of concrete, φ, by ACI model is done by an equation:
𝑡𝜓
𝜑 = 𝑑+𝑡 𝜓 𝜑𝑢 (2.37)
where d and ψ are determined by fitting the data obtained from the property tests,
and φu is the ultimate creep coefficient, also determined by fitting. Further
recommendations were given for standard conditions, in which d = 10 days and ψ =
0.6. φu can be estimated based on a range of correction factors, including age of
loading, relative humidity, size and shape, temperature, and concrete composition.
52
However, Bažant and Li [185] assessed a range of creep models by fitting to the
existing creep database and ranked the ACI model the lowest.
2.6.2.4 fib model code 2010

The fib model code 2010 is the latest creep model among these models. The
calculation of the creep coefficient, φ(t,t0), is separated into two components; φbc(t,t0)
for the basic creep coefficient and φdc(t,t0) for the drying creep coefficient. Both are
derived from different equations:
𝜑(𝑡, 𝑡0 ) = 𝜑𝑏𝑐 (𝑡, 𝑡0 ) + 𝜑𝑑𝑐 (𝑡, 𝑡0 ) (2.38)
𝜑𝑏𝑐 (𝑡, 𝑡0 ) = 𝛽𝑏𝑐 (𝑓𝑐𝑚 )𝛽𝑏𝑐 (𝑡, 𝑡0 ) (2.39)
𝜑𝑑𝑐 (𝑡, 𝑡0 ) = 𝛽𝑑𝑐 (𝑓𝑐𝑚 )𝛽 (𝑅𝐻)𝛽𝑑𝑐 (𝑡0 )𝛽𝑑𝑐 (𝑡, 𝑡0 ) (2.40)
where βbc(fcm) and βdc(fcm) are the factors estimated from the mean compressive
strength of concrete at age of 28 days, βbc(t,t0) and βdc(t,t0) are the time-dependent
factors, β(RH) is the factor related to the relative humidity and size of concrete
member, βdc(t0) is the factor for the age of loading.
2.6.3 Structural design
After the creep coefficient is calculated for a certain time according to the service life
of the concrete structure, the creep effect on the overall structure can be predicted.
There are a few rational methods for the prediction of deflection or displacement of
concrete structures due to creep.
2.6.3.1 Rate of creep method

The application of rate of creep method (RCM) on concrete structures was
introduced by Dischinger [186]. The assumption on RCM is the independence of age
of loading on the rate of change of creep, which mean that the creep curves of
53
concrete loaded at different ages are assumed to be in parallel. The Dischinger’s
equation is shown below:
𝜑(𝑡, 𝜏) = 𝜑(𝑡, 𝜏0 ) − 𝜑(𝜏, 𝜏0 ) (2.41)
𝜎 ̇ (𝑡) 𝜎 (𝑡) 𝜀∗
𝜀̇(𝑡, 𝜏) = E 𝑐(𝜏 ) + 𝜑̇ (𝑡, 𝜏0 ) [E 𝑐(𝜏 ) + 𝜑∗ (𝜏
𝑠ℎ
] (2.42)
𝑐 0 𝑐 0 ) 𝑜
where φ(t,τ) is the creep coefficient at time t loaded at time τ, 𝜀̇(𝑡, 𝜏) is the rate of
change of strain at time t loaded at time τ, 𝜎𝑐̇ (𝑡) is the rate of change of stress at time
t, 𝜑̇ (𝑡, 𝜏0 ) is the rate of change of creep coefficient at time t loaded at time τ0, Ec(τ0)
is the elastic modulus at time τ0, ε*sh is the final shrinkage strain and φ*(τ0) is the
final creep coefficient when loaded at time τ0.
Rüsch, Jungwirth and Hilsdorf [187] further developed the Dischinger’s equation and
applied on analysis of reinforced concrete flexural members, in which the change of
stresses of steel components and concrete related to the initial stresses can be
calculated using the following equations:
𝜎𝑠𝑡 (𝑡) = (𝛾𝑠𝑢𝑠 − 1)𝜎𝑠𝑠𝑢𝑠0 + 𝛾𝑠ℎ 𝜀𝑠ℎ (𝑡)𝐸𝑠 (1 − 𝛼) (2.43)
𝜎𝑐𝑡 (𝑡) = −𝛾𝑠ℎ 𝜀𝑠ℎ (𝑡)𝐸𝑐 𝛼 (2.44)
1−(1−𝛼)𝑒 −𝛼𝜑
𝛾𝑠𝑢𝑠 = α
𝛾𝑟 = 𝑒 −𝛼𝜑
1−𝑒 −𝛼𝜑
𝛾𝑠ℎ = αφ
𝐴 𝐴𝑖𝑑 2
𝛼 = n 𝐴 𝑠 (1 + 𝑦𝑖𝑑𝑠 ), n = Es/Ec
𝑖𝑑 𝐼𝑖𝑑
where σst(t) and σssus0 is the change of stress at time t and the initial stress of steel
component under sustained loads, respectively, σct is change of stress at time t of
concrete, εsh(t) is the shrinkage strain at time t, Es is the elastic modulus of steel, γsus,
and γsh are the coefficients for sustained loads and shrinkage, respectively, α is the
54
stiffness ratio, As and Aid are the cross-sectional area of steel component and
idealised cross section, Iid is the idealised moment of inertia, and yid is the distance
between the steel component and the centroid of the idealised cross section. The
strain of steel can be calculated with εs(t) = σs(t)/Es. Therefore, the strain distribution
and the curvature across the cross section can be computed.
2.6.3.2 Effective modulus method

The effective modulus method (EMM) is a simple method, which assumes the total
strain including instantaneous and creep effects is based on an effective modulus:
𝐸 (𝜏 )
𝑐 0
𝐸𝑒 (𝑡, 𝜏0 ) = 1+𝜑(𝑡,𝜏 (2.45)
0)
where Ee(t,τ0) is the effective modulus at time t loaded at time τ0.
The strain at the centroid, ε(t), and the curvature, k(t), of the cross section can be
calculated through the following equation:
𝜀(𝑡) 1 𝐼 −𝐵𝑖𝑑 𝑁𝑠 + 𝜀𝑠ℎ (𝑡)𝐸𝑒 (𝑡, 𝜏0 )𝐴𝑐

[ ] = 𝐸 (𝑡,𝜏 )(𝐴 𝐼 −𝐵 ) [ 𝑖𝑑 ][ ]
𝑘(𝑡) 𝑒 0 𝑖𝑑 𝑖𝑑 𝑖𝑑 −𝐵𝑖𝑑 𝐴𝑖𝑑 𝑀𝑠 + 𝜀𝑠ℎ (𝑡)𝐸𝑒 (𝑡, 𝜏0 )𝐵𝑐
(2.46)
where Ac and Aid are the cross-sectional area of concrete only and idealised section,
Bc and Bid are the first moment of area of concrete only and idealised section, and Ns
and Ms are the applied axial load and bending moment at the cross section,
respectively. The response from instantaneous actions when loads are applied can be
calculated through the same equation as Equation 2.46, with Ee(t,τ0) replaced by
Ec(τ0) and εsh(t) set as 0.
55
2.6.3.3 Age-adjusted effective modulus method
Furthermore, EMM had been refined by Trost [188] and Bažant [189] to take ageing
of concrete into account with the introduction of the age-adjusted coefficient, and the
improved method is called age-adjusted effective modulus method (AEMM). The
age-adjusted coefficient, x(t,τ0), was applied to the calculation of the age-adjusted
effective modulus Ea(t,τ0):
𝐸 (𝜏0)
𝐸𝑎 (𝑡, 𝜏0 ) = 1+𝑥(𝑡,𝜏𝑐 (2.47)
0 )𝜑(𝑡,𝜏0 )
𝐸𝑐 (𝜏0) 1
𝑥(𝑡, 𝜏0 ) = − (2.48)
𝐸𝑐 (𝜏0)−𝑅(𝑡,𝜏0 ) 𝜑(𝑡,𝜏0 )
Determination of x(t,τ0) required the relaxation function, R(t,τ0), which can be

estimated based on the creep compliance function [190]. Bažant, Hubler and Jirásek
[191] proposed an improved estimation of the relaxation function:
1 𝑐1 𝛼(𝑡,𝜏0)𝐽(𝑡,𝜏0 ) −𝑞
𝑅(𝑡, 𝜏0 ) = 𝐽(𝑡,𝜏 ) [1 + ] (2.49)
0 𝑞𝐽(𝑡,𝑡−𝜂)
where 𝑐1 = 0.0119𝑙𝑛𝜏0 + 0.08
𝐽(𝜏0+𝜉,𝜏0 )
𝛼(𝑡, 𝜏0 ) = −1
𝐽(𝑡,𝑡−𝜉)
𝑡−𝜏0
𝜉= , 𝜂 = 1⁡ and 𝑞 = 10
2
An equation similar to Equation 2.46 is used to calculate the strain at the centroid
and the curvature of the cross section. The equation is revised to include the age-
adjusted coefficient:
𝜀(𝑡) 1 𝐼 −𝐵𝑖𝑑 𝑁𝑐𝑟 + 𝜀𝑠ℎ (𝑡)𝐸𝑎 (𝑡, 𝜏0 )𝐴𝑐

[ ] = 𝐸 (𝑡,𝜏 )(𝐴 𝐼 −𝐵 ) [ 𝑖𝑑 ][ ]
𝑘(𝑡) 𝑎 0 𝑖𝑑 𝑖𝑑 𝑖𝑑 −𝐵 𝑖𝑑 𝐴𝑖𝑑 𝑀𝑐𝑟 + 𝜀𝑠ℎ (𝑡)𝐸𝑎 (𝑡, 𝜏0 )𝐵𝑐
(2.50)
𝑁𝑐𝑟 𝑁 𝜑(𝑡, 𝜏0 )[𝑥(𝑡, 𝜏0 ) − 1] 𝐴 𝐵𝑐 𝜀 (𝜏0 )

[ ] = [ 𝑠] − 𝐸𝑐 (𝜏0 ) [ 𝑐 ][ ]
𝑀𝑐𝑟 𝑀𝑠 [1 + 𝜑(𝑡, 𝜏0 )𝑥(𝑡, 𝜏0 )] 𝐵𝑐 𝐼𝑐 𝑘(𝜏0 )
(2.51)
56
where Ncr and Mcr are the axial load and bending moment combined with creep effect
produced by initial stress at the cross section and Ic is the moment of inertia of
concrete only.
2.7 Effect of creep on concrete structures
Excessive deflection and time-dependent cracking under long-term service life are
the issues affecting serviceability of concrete structures. These can cause aesthetic or
functional problems, damage to either structural or non-structural members, and
enhance dynamic effects causing discomfort to occupants. Wide open cracks can also
cause durability problem accelerating corrosion of the reinforcements and lead to
failure of structures. However, these can be avoided by careful structural design with
a reliable approach, which allows accurate prediction of long-term deformation of
structures, such as RC beams, composition concrete beams, prestressed beams, etc.
[9, 192]. Accurate time-dependent analysis is critical for bridge design. Loss of
prestress occurs due to combined effects of creep and shrinkage of concrete and
relaxation of prestress cable, resulting in increasing deflection [193]. This is crucial
for long span bridges, as a few percentages of error in design calculation would end
up more than 1 m of additional deflection to plan. Moreover, drying creep and
shrinkage would vary with different sizes of webs and slabs, and lead to further
displacement due to differential actions. In worst case, structure failure would occur
[194].
Creep also affects the integral structures, such as integral abutment bridges and
beam-column frames in buildings. In integral abutment bridges, the bridge deck and
the abutments are cast integrally. Because of that, creep and shrinkage in the deck
and girder concrete induce bending stresses in the piling that supports the abutments,
so that the structural should include the addition loads due to these long-term effects
[195]. For buildings, the load bearing would be different between the exterior and the
interior columns. Differential vertical deflections would be observed between the
exterior and the first interior columns, and amplified by creep. Since beams are built
to connect these columns, load transfer would occur among columns through the
57
beams. This adds extra stresses on beams and the beam-column joints, but this can be
controlled by the percentage of reinforcement in columns and the beam stiffness. The
greater the percentage of reinforcement in a column, the greater would be the stress
transfer from concrete to reinforcement, and the smaller the strain and deformations
and floor deflections. The load transfer in the frames with high beam stiffness are
higher than those in the frames with low beam stiffness [196]. Furthermore, the
reduction of creep deflections in RC beams can be achieved by the use of
compression reinforcement [197].
Composite steel-concrete beam is popular in modern construction. It is good in

bearing sagging moment rather than hogging moment, which would cause cracking
in concrete section. In the continuous supported composite beams, creep and
shrinkage cause a gradual redistribution of moment throughout any period during
sustained loads. This may give additional hogging moment at the intermediate
support, and cause cracking [198]. Shrinkage in the hogging moment region
increases the chance of cracking. Once this section is cracked, great increase in
deflection will be resulted in spans. This condition can be controlled by installing
more reinforcements in the concrete slab to restrict the occurring and propagation of
cracks [199]. Another way to prevent cracking in the hogging moment region is to
apply post-connected precast concrete slab, so that the effect of shrinkage is
minimised [200].
Shariq, Abbaas and Prasad [201] studied the long-term deflection of RC beams made
of blend cement with GGBFS, and they concluded that the creep deflection increased
with increasing amount of GGBFS contents in concrete. This would be due to the
delay in the process of hydration and low compressive strength for blended concrete
with GGBFS at the age of loading at 28 days, which is younger than the maturity age
of 56 to 90 days. Therefore, the creep deflection is expected to reduce if these beams
are loaded at later ages.
Liu, Lu and Peng [144] suggested fly ash and metakaolin-based GPC can be applied
to prestressed concrete structures, due to similar mechanical properties to OPCC and
low creep. However, the prestressing force and loading were applied only when the
GPC beams reached the age of 60 days for attaining sufficient strength.
58
Chapter 3 : Methodology and experimental
program
3.1 Introduction
In this chapter, the preparation of the test specimens, including geopolymer and OPC
concretes, mortars and pastes, from small to large scales will be presented. One
formulation of GPC was chosen from the preliminary works for its satisfactory
strength property and its use in full-scale application. Reference specimens of OPCC
were made with replacement of all amount of geopolymer binder materials with
OPC.
Both GPC and OPCC were cured under the same conditions and their performances
were compared among a range of property tests. Long-term tests were carried out to
obtain the creep and shrinkage behaviour of both types of specimens. Microstructure
analysis was done to reveal the pore structures and the chemistry of the binders. This
enables understanding the differences between GPC and OPCC, and the mechanism
of creep. In this way, the application of GPC on structural design can be evaluated.
3.2 Materials
3.2.1 Binder materials
Ground granulated blast furnace slag (GGBFS) and fly ash (FA) were used as raw
binder materials for GPC. GGBFS was obtained from Independent Cement and Lime
Pty Ltd without added Gypsum. FA was obtained from Gladstone power station.
The OPC used for the reference formulation was commercial type GP cement
according to Australian Standard AS 3972-2010 [202]. The same sources of these
materials were ensured throughout the whole research period. A picture of these
materials is shown in Figure 3.1. Their chemical composition and physical properties
are presented in Tables 3.1 and 3.2.
59
FA
GGBFS
OPC
Figure 3.1: Dry raw binder materials
Table 3.1: Composition of OPC, GGBFS and FA from X-ray fluorescence analysis
(LOI is loss on ignition at 1000 °C)
Oxides, SiO2 Al2O3 Fe2O3 CaO MgO SO3 Na2O K2O LOI
mass %
GGBFS 34.2 13.8 0.4 43.1 5.4 0.8 0.1 0.4 1.8
FA 48.8 28.5 12.5 4.4 0.3 0.3 0.3 0.4 2.7
OPC 20.3 4.5 4.6 62.9 1.2 2.6 0.3 0.3 3.3
Table 3.2: Physical properties of the source materials
Raw materials Specific gravity D50 (μm) Specific surface (m2/g)

GGBFS 2.8 15 2.4
FA 2.2 10 1.6
OPC 3.2 8 1.6
60
3.2.2 Alkali activator
Commercial sodium metasilicate with a chemical composition of 50.5 wt% Na2O

and 46.0 wt% SiO2 (balance H2O) was used as activator for producing geopolymer.
The activator is in bead form from the commercial package (see Figure 3.2). The size
of the bead ranges from 0.5 to 2000 μm. The anhydrous beads was applied directly in
the production of geopolymer; preparation into solution form was not required. The
purpose of using anhydrous beads of activator instead of solution was to allow longer
setting time of GPC, since the beads of activator dissolved slowly in the mixture
[203].
Figure 3.2: Alkali activator in this study (sodium metasilicate)
3.2.3 Aggregates
Two grades of alluvial siliceous aggregates of 20 mm and 14 mm maximum size

with specific gravity of 2.83 and water absorption of 1.10% were used as coarse
aggregates in concrete production. For fine aggregate, a type of silica sand with a
specific gravity of 2.55 and water absorption of 0.90% was used. The grading of this
type of sand is shown in Figure 3.3 according to ASTM C136/C136M-14 [204]. The
61
aggregates were prepared and kept in dry condition before use. A picture of the
aggregates used in this project is shown in Figure 3.4.
Figure 3.3: Grading of fine aggregate
20 mm aggregates
Fine sand 14 mm aggregates
Figure 3.4: Dry aggregates for concrete and mortar mixes
62
3.3 Sample preparation
3.3.1 Mix design
In this project, specimens made with concrete, mortar and paste of both geopolymer
and OPC were tested. The mix designs of these specimens are shown in Table 3.3.
Water-binder ratio (w/b) is used to describe the water content in all mixes, instead of
water-cement ratio (w/c). The mix design of GPC was chosen because it can provide
sufficient strength for structural purpose from the preliminary works. It was found
that using 5% FA and 95% GGBFS instead of 100% GGBFS helped with
workability and increased the short-term availability of Al compared with the higher
Ca available from the slag. As a result, it manages the short-term chemical balance to
get the best phase composition. In this case, OPCC was produced in similar
formulation with same amount of binder materials for comparison purpose. The mix
designs of mortars and pastes were decided by removing coarse aggregates and all
aggregates from the concrete mix designs, respectively.
Table 3.3: Mix designs for geopolymer and OPC concretes, mortars and pastes
Materials Concrete Mortar Paste

GPC OPCC GPM OPCM GPP OPCP
3
GGBFS (kg/m ) 380 0 722 0 1368 0
3
FA (kg/m ) 20 0 38 0 72 0
3
OPC (kg/m ) 0 400 0 760 0 1500
Water-binder 0.4 0.4 0.4 0.4 0.4 0.4
ratio, w/b
20 mm aggregates 450 450 - - - -
(kg/m3)
14 mm aggregates 700 700 - - - -
(kg/m3)
Fine aggregates 630 630 1197 1197 - -
(kg/m3)
Activator (wt.% 10 - 10 0 10 -
binder materials)
63
3.3.2 Mixing and casting procedures
There are two methods used in mixing concrete in this project, depended on the sizes
of the specimen. For small concrete specimens for standard material property tests, a
rotating pan mixer of 70 litres capacity was used in the laboratory (see Figure 3.5).
At first, the coarse and fine aggregates were mixed well in the mixer. Then the binder
materials were added to the mixture and mixed for 2 minutes further. Then water was
poured to the mixture gradually during mixing for further 2 minutes. Until this step,
the preparation of fresh OPCC was done, but for GPC, further step was required. The
alkali activator was added to the wet mixture gradually. Further mixing for 5 minutes
was allowed to ensure well concentration distribution of activator throughout the
mixture. Finally, the fresh mixture can be cast in the moulds. The setting time of
GPC used in this study is approximately 30 to 60 minutes.
Figure 3.5: Rotating pan mixer for concrete mixing
64
For full-scale beam tests in this project, concrete truck was used to provide large
volume of GPC. The mixing steps was similar to laboratory mix. Aggregates, binder
materials and water were loaded in the concrete mixer on the truck from the concrete
plant. They were well mixed during transportation to the site. Once it arrived at the
site, the activator was added gradually to the mixer and mixed for 5 minutes. The
fresh GPC was ready to be poured to the beam moulds.
Mortar mix followed similar way to laboratory mixing, except there was no coarse
aggregate used. Therefore, the mixing started from mixing binder materials with fine
aggregate in a mortar mixer of 5 litres capacity. For the paste mix, although there
was no aggregate, mixing of GGBFS and FA was required in the beginning, before
water and activator were added.
Before casting, the moulds were prepared with oil grease. It was found that GPC
would bond to metal surface of the mould, so a layer of plastic film was applied
between the steel components of the mould and the fresh mixture of geopolymer for
the ease of demoulding. The placing of specimens was done in layers according to
AS1012.8-2014 [205]. Each layer was well compacted by using vibrating table or
vibrating rod before application of next layer.
3.3.3 Curing
All specimens for property tests besides the GPC beams were demoulded after 1 day.
Then they were cured in a lime saturated bath at 23 °C. The duration of bath curing
was determined in different test purposes. For standard creep and drying shrinkage
tests according to AS1012.16-1996 [206] and AS1012.13-2015 [207], bath curing
was provided to the specimens until 7 days, then the specimens were moved into the
environmental chamber for air curing, in which the temperature was kept at 23°C and
the relative humidity (RH) at 50%.
For specimens under sealed condition, there are two methods applied on different
specimen types. MasterKure® CC 1315WB was used to seal the specimens for short-
term tests, such as strength tests, after demoulded. Two coats of this compound were
65
applied on all surfaces of the specimens as shown in Figure 3.6a. Then the specimens
were kept in the same environmental chamber above. For long-term tests of creep
and shrinkage, two layers of resin were applied to coat the surfaces of these
specimens after removed from bath, and then one layer of plastic tape was used to
seal the specimens finally (see Figure 3.6b specimens at bottom).
(a) (b)
Figure 3.6: Sealed specimens: (a) Application of MasterKure® CC 1315WB; (b)

resin coating and plastic layer
3.4 Material property tests
There was no specific test method for geopolymer specimens at the time when the
research project was conducted, so geopolymer specimens were tested according to
the standard test procedures for OPC concrete, mortar and paste.
3.4.1 Compressive strength and elastic modulus tests
Compressive strength and elastic modulus tests were conducted on concrete

specimens of 100 mm dia. × 200 mm long approximately. The ends of the specimens
were grinded flat and parallel before testing by using a cylinder end grinder – Marui
Triple Hi Kenma as shown in Figure 3.7. A compression machine of 3000 kN
66
capacity was used to apply the force. For compressive strength test according to
AS1012.9-2014 [208], the force was increased continuously at a rate equivalent to
20±2 MPa compressive stress per minute until failure of the specimen. The
compressive strength was reported from the average of two tests. For elastic modulus
tests according to AS1012.17.1997 [209], the force was increased up to 40% of the
compressive strength obtained previously. The deformation of the specimens in both
longitudinal and lateral directions was measured by the strain gauges attached to the
surfaces, two in each direction (see Figure 3.8a). Both elastic modulus and Poisson’s
ratio were reported from the average of two tests.
For mortar and paste specimens, compressive strength tests were done according to
ASTM 109C/109M [210] on a specimen size of 50 mm cubic. Elastic modulus of
paste specimens was obtained in a similar way to concrete specimens, but in a
smaller size of 50 mm dia. × 100 mm long. Due to its small size, a compressometer
and extensometer was used (see Figure 3.8b) to measure deformation during the test.
Average result of three tests was taken as test values for both tests.
Figure 3.7: Cylinder end grinder
67
(a) (b)
Figure 3.8: Tests for determination of elastic modulus: (a) Concrete (b) Paste
3.4.2 Flexural tensile strength test
Flexural tensile strength/modulus of rupture of concrete was determined on prism

specimens of 100 × 100 × 400 mm according to AS1020.11-2000 [211]. A four-
point-bending test was carried with a setting of 100 mm spacing between each
loading point as shown in Figure 3.9. The modulus of rupture was reported from the
average of three tests.
Figure 3.9: Test for determination of modulus of rupture
68
3.4.3 Creep test
Creep tests were conducted in the spring supported load frames specified in
AS1012.16-1996 [206]. A cylinder size of 150 mm dia. × 300 mm long
approximately was chosen for the concrete specimens. The ends of the specimens
were grinded flat and parallel before testing. Two concrete specimens of same type
were placed into a column in the load frame, and then sustained load was applied up
to 40% of compressive strength at the ages of loading with the aid of a hydraulic jack
as shown in Figure 3.10a. The specimens remained under sustained load in the creep
loading frames throughout the test period. Regular load adjustment was required
when the sustained load dropped more than 2% of the target load.
(a) (b)
Figure 3.10: Test instruments for creep test: (a) Spring supported load frame and
hydraulic jack; (b) demec gauge
69
Another two concrete cylinders were reserved as companion specimens to observe
effects without the applied load. Both end faces of the companion specimens were
painted with resin to prevent moist loss from these ends corresponding to the
condition of the loaded specimens. All specimens were stored in the environmental
chamber kept at 23°C and 50% RH during test period as shown in Figure 3.11.
Measurement of strain was done with a digital detachable mechanical gauge (demec
gauge) of 200 mm gauge length as shown in Figure 3.10b, with gauge points
attached on three sides in a triangle on the specimens in loading direction, and on
opposite sides on the companion specimens (see Figure 3.12). The accuracy of this
demec gauge is to 4 microstrains (×10 -6). The measurements were recorded before
and right after loading, then 2 hours, 6 hours, daily until 7 days, weekly until 28 days
and monthly afterward.
Figure 3.11: Companion specimens for shrinkage effect in creep test
70
Figure 3.12: Measurement on creep specimen
3.4.4 Shrinkage tests
Standard drying shrinkage test was

done on three prism specimens of 75 ×
75 × 280 mm according to AS1012.13-
2015 [207]. Measurement taken straight
after the bath curing at 7 days by the
use of vertical comparator (see Figure
3.13), with an accuracy to 0.01 mm.
Then continuous measurements were
done every week until 28 days, and
then every month. Extra specimens
were cast and sealed with the method
described for creep specimens for
measurement of autogenous shrinkage
strain.
Figure 3.13: Vertical comparator
71
Standard autogenous shrinkage test was done on paste and mortar specimens
according to ASTM C1698 [212]. Prior to this test, the final setting time of the paste
specimens were determined by Vicat needle test according to ASTM C191 [213].
The pictures of these test equipment are presented in Figure 3.14. The first
measurement in autogenous shrinkage test started at the final setting time. Then
continuous measurements were done every hour for 6 hours, then every day until 7
days, and weekly until two months.
(a) (b)
Figure 3.14: Test equipment for autogenous shrinkage test on pastes and mortar: (a)
Dilatometer bench with length gauge, (b) Vicat needle
3.5 Tests for microstructure characteristic
3.5.1 Nitrogen gas adsorption test
Because the gel pores are believed to have an influence on creep behaviour of the
materials, the pore sizes below 50 nm in both GPP and OPCP were assessed by using
nitrogen gas adsorption test. This test is capable to analyse pore size down to 1-2 nm
without causing damage to the specimen. The nitrogen adsorption test was carried on
the testing unit, Belsorp max, manufactured by BEL Japan, Inc.. A picture of the test
72
instrument and the pre-treatment unit is shown in Figure 3.15. This test instrument
measures the amount of nitrogen molecules condensing on the all surfaces (internal
and external) of the specimen from vacuum toward atmosphere pressure under the
condensation temperature provided by the vessel filled with liquid nitrogen. This is
the adsorption path. After that, the nitrogen molecules leave the specimen under
vacuuming process during the desorption path.
Figure 3.15: Test instruments for nitrogen gas adsorption test
Paste specimens were saw cut to a size of 1 mm thick approximately. Pre-treatment

was required to remove most free evaporate water in the specimens. Usually for the
OPC specimen, the pre-treatment would be heating to 105 °C in the oven under
simultaneous vacuum, but this heat treatment method would influence the
microstructure of GPC, e.g. pore collapse, and thus affect the accuracy of the result
[214, 215]. Therefore, solvent replacement was used to minimise the change to the
microstructures of the specimens. The specimens were immersed in the isopropanol
solvent, since the organic solvents tend to replace the pore solution in the paste
73
specimens by a simple counter-diffusion process [92]. However, this process is slow
and is affected by the concentration of the solvent. The solvent was renewed
regularly throughout a period of 1 month. After that, the specimens were vacuum-
dried in room temperature for 4 days to remove the solvent in the specimens. These
procedures were based on the method from Aligizaki [63], but the period was
extended to ensure removal of most water.
After the test, BET method [109] was used to calculate the surface area of the porous
material from the measured results of adsorption-desorption isotherm. BJH method
[111], CI method [112] and DH method [113] were used to analyse the mesopore
size distribution of the pastes, whilst MP analysis [114] was used to analyse the
micropore size distribution of the pastes.
3.5.2 Other analysis
Besides the above analysis, X-ray diffraction (XRD) analysis, thermogravimetry

analysis (TGA) and backscattered scanning electron microscopy (BSEM) was used
to characterise the microstructure.
XRD was used to identify the phases in the geopolymer and OPC samples before and
after creep test, and after shrinkage. The samples were crushed and ground manually
up to 70 μm prior to the analysis. XRD was conducted using a Bruker D8 advance
instrument with Ni-filtered Cu Kα radiation scanning from 5°-65° 2θ, with a 0.02°
step size and 2 s per step count time. The result was identified by comparison to the
International Centre for Diffraction Data PDF4 database of powder diffraction files.
TGA was used to identify and quantify different reaction products forming in the
samples. It was conducted using a PerkinElmer Diamond instrument. The samples
were crushed and ground manually and held in the instrument under isothermal
conditions for 60 minutes at 40 °C to equilibrate, and then heated up to 1000 °C at a
rate of 10 °C per minute in a nitrogen environment at a purge rate of 200 mL per
minute.
74
BSEM was used to observe the distribution of alkali-activated binders and unreacted
precursor particles in geopolymer samples. The samples were prepared by
sectioning, grinding and polishing for the analysis. It was conducted using an FEI
Quanta ESEM instrument, with energy dispersive X-ray analysis at an accelerating
voltage of 15 kV in a low vacuum environment with a pressure of 0.5 mbar and a
working distance of 10 mm.
3.6 Beam tests
The long-term structural behaviour of GPC was observed from concrete beam tests.
Beams constructed with GPC were set up in two types and stored in an indoor
environment:
1. RC beams of 200 mm wide  240 mm deep  4000 mm long.

2. Composite beam with steel formwork of 600 mm wide  240 mm deep 
6500 mm long for floor systems in practice.
These GPC beams were simply supported with sustain loads applied on top of the
beams for a long period. Long-term deflection was measured with the aid of dial
gauges, which had an accuracy to 0.01 mm, at mid-span below the beams before and
after applied loads, and at regular basis until the end of test.
Due to the constraints in test environment and resource, OPCC beam was not
constructed in this project. Instead, virtual concrete beams were made with the use of
a finite element modelling software, ATENA Studio, from the input of property
parameters of the reference OPCC used in this project. The same reinforcement
details as the GPC beams detail was applied to these OPCC beams. Comparison
between the actual GPC beams and the virtual OPCC beams was done.
Structural analysis on the GPC beams was done by using three rational methods,
RCM, EMM and AEMM. Properties of GPC, reinforcement details and load profile
for input parameters to the analysis methods were obtained from the experimental
75
results including elastic modulus, modulus of rupture, creep and shrinkage. The
prediction by these methods was then compared to the test results.
3.7 Thermal creep tests on paste specimens
The creep behaviour of both geopolymer and OPC specimens under thermal effect
was investigated through a series of test on paste specimens of 50 mm dia. × 100 mm
long. The tests were carried in an electrical furnace with a capacity of 1200 ºC
combined with an actuator of 250 kN compression capacity. Thermocouples were
used to measure the temperature on the specimens. Displacement of the specimens
was detected by the sensor from the actuator. Calibration with a steel specimen was
required. These results were recorded in three types of thermal test performed in this
study, which include:
1. Hot strength
2. 1-day thermal creep and free thermal expansion
3. Transit thermal creep
76
Chapter 4 : Test results of GPC and OPCC and
creep prediction of GPC
4.1 Introduction
In this chapter, the GPC and OPCC were tested for material properties. A
comparison study between GPC and OPCC is presented. Moreover, prediction of
creep of GPC is proposed.
In total, three batches of OPCC and ten batches of GPC were made in this study for a
number of tests. A single formulation was used in both types of concrete as specified.
Therefore, the ID of the specimens is established as the type of concrete with the date
of casting. The detail of the specimens is tabulated in Table 4.1.
The fresh property of GPC and OPCC are reported from slump test and discussed in
section 4.2.
The results of strength tests are presented in section 4.3. Compressive strength test
was done on all batches of concrete as the primary test. Specimens were tested at 7
days, 14 days and 28 days of ages and under different curing methods, including bath
curing, 7 days bath curing followed by air drying and sealed conditions. Further
testing on flexural tensile strength was done on three batches of GPC specimens.
In section 4.4, the results of shrinkage test on GPC and OPCC are presented. The
difference between them in autogenous shrinkage and drying shrinkage is discussed.
Section 4.5 presents the results of creep tests on GPC and OPCC. The different creep
behaviour between GPC and OPCC are discussed. The effect of age of loading and
the effect of drying on creep of GPC are investigated.
In section 4.6, the results of creep test are compared with the prediction by the
standard creep models. Necessary modifications are proposed to these creep model to
provide a better prediction on GPC.
77
In addition, GPC121113 and GPC040614 are also used in the construction of GPC
beams for long-term flexural test. Further details refer to chapter 7.
Table 4.1: Details of concrete specimens for testing

Specimens Compressive Flexural tensile Shrinkage test Creep test
strength test strength test
GPC010813  7 days
 28 days
GPC040913  7 days
 28 days
GPC221013  7 days  Drying  Loaded at 28
 28 days shrinkage days
GPC121113#  7 days  Drying  Loaded at 14
 14 days shrinkage days
 28 days  Loaded at 28
days
GPC260314  7 days  28 days bath
 14 days cured
 28 days  28 days drying
GPC160414  7 days  7 days bath cured
 14 days  14 days drying
 28 days
GPC040614#  7 days  Drying
 14 days shrinkage
 28 days
GPC190814  14 days  14 days sealed
 28 days  28 days sealed
GPC300615  7 days  Drying  Loaded at 28
 28 days shrinkage days under
 Shrinkage sealed
under sealed condition
condition
GPC150715  7 days  Loaded 7 days
 28 days
OPCC300713  7 days  Loaded at 28
 28 days days
OPCC010715  7 days  Loaded at 28
 28 days days
OPCC020715  7 days  Drying  Loaded at 28
 28 days shrinkage days under
 Shrinkage sealed
under sealed condition
condition
# Batches prepared for GPC beams from concrete plant
78
4.2 Fresh property of GPC and OPCC
Slump test was done on both GPC and OPCC according to AS1012.3.1-2014 [216]
as shown in Figure 4.1. The slump of the fresh GPC from all the batched ranged
from 70 to 150 mm. In some cases, the workability was low due to the water
adsorption by the dry aggregates. Because the superplasticiser for OPCC would
cause different effects on GPC despite of increasing workability, such as accelerated
setting and lower strength [39], it was not used in this research. Instead, extra water
was added when the workability was too low, in order to achieve suitable workability
for construction purpose. However, the final w/b should not exceed 0.5 w/b.
Similar amount of extra water was also added to OPCC for consistency. However,
since no superplasticiser was used, the slump of the fresh OPCC was 10 mm. Extra
effort in compacting the specimens was required during placing.
Figure 4.1: Slump test carried out on GPC
79
4.3 Strengths of GPC and OPCC
4.3.1 Compressive strength
Compressive strength test was carried out on three sets of OPCC specimens and ten
sets of GPC specimens. The unconfined compressive strengths, f’c, of the OPCC and
the GPC at 7 days, 14 days and 28 days are shown in Table 4.2. All the obtained
strength values are within ±10% of the average value. In these results, GPC010813
and GPC040913 were originally cast for creep test, but they were discarded due to
the low strength values. These low strength results would be due to poor handling in
the mixing and casting for the first time. For example, the activator was added in
during mixing rapidly, which affected setting and caused difficulty in compaction, so
that it resulted in inhomogeneous properties within the specimens.
The average compressive strength of GPC except GPC010813 and GPC040913 at

28-days is 52.6 MPa, which is generally comparable with that of OPCC, 57.6 MPa.
The average 7-day and 14-day compressive strengths of GPC are 82.2% and 87.9%
of its f’c, respectively. The average 7-day compressive strengths of OPCC was 74.3%
of its f’c. This indicates that GPC developed higher early strength than OPCC, and
this generally agrees with the result of past research on similar GPC mix [27].
The strength values above were obtained from the samples with standard moist
curing (bath curing) until tested. Further tests were done on GPC121113 and
GPC2603114, which were bath curing until 7 days and then moved into the drying
room, to observe the effect of drying. The compressive strength of these specimens is
generally lower than the same batch of specimens with bath curing at all time. The
strength development is minor from 14 days to 28 days on the specimens subjected
to drying. These reductions in strength development were hypothesized to be due to
microcracking induced by drying [70].
For GPC190814, compression test was carried on specimens sealed with

MasterKure® CC 1315WB at 14 days and 28 days. The compressive strength of
these specimens are 51.2 MPa and 57.2 MPa for 14 days and 28 days, respectively.
They are in similar in strength values to their bath cured counterparts.
80
Table 4.2: Compressive strength of OPCC and GPC
Specimens Compressive strength (MPa)

Bath cure Bath cure until 7 days
then air drying
7-day 14-day 28-day 14-day 28-day
GPC010813# 20.2 - 35.9 - -
GPC040913# 21.3 - 25.1 - -
GPC221013 37.6 - 49.0 - 46.5
GPC121113 42.6 42.9 46.7 40.3 41.8
GPC260314 46.3 44.4 53.3 48.4 48.7
GPC160414 47.8 48.1 55.7 - -
GPC040614 45.5 49.5 50.2 - -
GPC190814 - 53.6 54.0 - -
GPC300615 36.0 - 55.2 - -
GPC150715 46.7 - 58.2 - -
Average 43.2 47.7 52.8 44.4 45.7
OPCC300713 29.5 - 50.6 - -

OPCC010715 50.6 - 64.5 - -
OPCC020715 48.2 - 57.7 - -
Average 42.8 - 57.6 - -
# Excluded from the calculation of average values
4.3.2 Flexural tensile strength
GPC is expected to have higher tensile strain capacity and tensile strength than
OPCC [5, 47, 217]. Therefore, flexural tensile test was done on GPC260314 and
GPC160414. In total, six tests were tested for modulus of rupture, f’ct.f, under various
ages and drying conditions, with results shown in Figure 4.2. The typical failure
mode of the test specimens follows the acceptable failure mode from AS1012.11-
2000 [211], where crack should be within the 100 mm region in the middle of the
specimen (see Figure 4.3). The modulus of rupture of GPC was 6.8 MPa at 28 days,
81
while the modulus of rupture of OPCC was 4.6 MPa for the same grade based on an
equation from AS3600-2009 [181]:
𝑓′𝑐𝑡.𝑓 = 0.6√𝑓′𝑐 (4.1)
The modulus of rupture of GPC at 7 days is 92% of that at 28 days. This indicates
that GPC develops higher tensile strength at early age compared to OPCC, which can
be due to the bond strength from the cross-linked microstructure [51, 218].
Drying seems to affect the modulus of rupture of GPC. A reduction of 40% was
found in the modulus of rupture of GPC specimens stored in the drying room
compared with the well-cured specimens. This is more than the loss percentage in
compressive strength under drying, which was up to 10% approximately. This can be
explained by the similar findings in OPCC [9], in which the extra tensile stress at the
drying surface induced by differential drying shrinkage. Microcracking at the surface
can also occur due to drying [106], and it weakens the strength at surface. Therefore,
these contribute to the loss in modulus of rupture of GPC.
Figure 4.2: Modulus of rupture of GPC at different ages (Bath cured: Bath cured at
all time; Drying room: Bath cured until 7 days then moved to drying condition of 23
°C and 50% RH; Sealed: Sealed after demould)
82
Figure 4.3: Typical failure in flexural tensile strength test of GPC specimens
On the other hand, the modulus of rupture of the sealed specimens is 5.2 MPa and
5.0 MPa at 14 days and 28 days, respectively. Although there was no drying effect
on these specimens, they still suffered from loss in flexural strength compared to the
well-cured specimens. Therefore, it can be seen that proper curing method is vital for
strength development in GPC as in OPCC.
83
4.3.3 Elastic modulus and Poisson’s ratio
The elastic modulus, Ec, and the Poisson’s ratio, ν, of GPC was obtained from the 28
days bath cured specimen GPC160414 as 33,270 MPa and 0.23, respectively. A
graph of applied stress versus vertical and lateral strains of the test cylinder is shown
in Figure 4.4. The elastic modulus of OPCC can be estimated from the equation
according to AS3600-2009 [181]:
𝐸𝑐 = 𝜌1.5 (0.024√𝑓𝑐𝑚𝑖 + 0.12) (4.2)
where ρ is the density of concrete and fcmi is the mean value of the in-situ
compressive strength. In this case, when f’c = 57.6 MPa, fcmi = 60.6 MPa, and the
density was taken as 2400 kg/m3 approximately for OPCC, the elastic modulus of
OPCC was then calculated as 36,070 MPa. Therefore, it can be seen that the elastic
modulus of GPC is lower than that of OPCC. This agrees with the finding by Collins
and Sanjayan [27] on slag-based GPC and OPCC. Further results on elastic modulus
were obtained from creep test refer to section 4.5.1.
Figure 4.4: Stress vs. Vertical and lateral strains of GPC
84
4.4 Shrinkage of GPC and OPCC
There were seven sets of prism specimens made of GPC and OPCC for the shrinkage
test. The details of specimens are shown in Table 4.3. Two of them were from the
batches prepared for GPC beams, which was prepared from the concrete truck. A
picture of the concrete prism specimens is shown in Figure 4.5.
Table 4.3: Details of concrete specimens under shrinkage test
Specimen ID Batch details Test details

GPC221013-Sh Laboratory mix Drying shrinkage test
GPC121113-Sh Factory mix Drying shrinkage test
GPC040614-Sh Factory mix Drying shrinkage test
GPC300615-Sh Laboratory mix Drying shrinkage test
GPC300615-Sh-Seal Laboratory mix Shrinkage test on sealed specimens
OPCC020715-Sh Laboratory mix Drying shrinkage test
OPCC020715-Sh-Seal Laboratory mix Shrinkage test on sealed specimens
Figure 4.5: Prism specimens for shrinkage test
85
The original length (L) of the specimens excluding the steel studs at the ends was
measured right after the specimens were taken out from the bath at 7 days. After that,
the first measurement (L1) was done immediately by using the vertical comparator.
Continuous measurements (L2, L3, L4 … Li) were done in regular basis. The
shrinkage strain (εsh) is calculated as:
𝐿𝑖 −𝐿1
𝜀𝑠ℎ.𝑖 = (4.3)
𝐿
Three series of shrinkage results of GPC221013-Sh, GPC121113-Sh and

GPC040614-Sh are shown in Figure 4.6. All three series show a rapid increase in
shrinkage strain with 200 days, and the increase rate decreases over time. The
increase of shrinkage strain is minor after 600 days in general. Although the mix
designs are similar in these three mixes, the measured shrinkage strains are different
in this case. The final values of shrinkage strain at the end of the experiment for
GPC221013-Sh, GPC121113-Sh and GPC040614-Sh were 890, 1235 and 1140
microstrains, respectively. This shows up to 30% more shrinkage strain occurred in
GPC121113-Sh and GPC040614-Sh than GPC221013-Sh. There seems to be an
influence on the material properties by the manufactory methods. GPC121113-Sh
and GPC040614-Sh were prepared from the concrete plant in a practical method,
which composition would be less controlled than GPC221013-Sh made in the
laboratory environment. The proportion of aggregates to binder materials, w/b ratio,
and the amount of activator would not be exactly the same between the plant mix and
the laboratory mix. Moreover, it was noted that more water was added to the plant
mix on site to increase workability in this case. The increase in the w/b ratio can lead
to increase in shrinkage strain, which is similar to OPCC when more free water
content between the C-S-H gels causes higher shrinkage [1].
Further study was done to comparison the shrinkage of GPC and OPCC. In addition
to the procedures of standard drying shrinkage test, some shrinkage specimens were
sealed to assess both autogenous shrinkage and drying shrinkage. The test results of
up to 600 days are shown in Figure 4.7. The final shrinkage result of GPC300614-Sh
was 870 microstrains, which is similar to GPC221013-Sh. This indicates the
consistency between these two batches. However, it is higher than the shrinkage
strain of OPCC020714-Sh, which was 680 microstrains approximately at same age.
86
This agrees with the overall finding by Collins and Sanjayan [149], but the difference
between GPC and OPCC was not significant in this case.
Figure 4.6: Shrinkage strain of GPC between batches
Figure 4.7: Total shrinkage and autogenous shrinkage of GPC and OPCC
87
The autogenous shrinkage strains of GPC and OPCC at age of 600 days are 630 and
320 microstrains, respectively. The autogenous shrinkage of GPC is almost a double
of that of OPCC, but the total shrinkage of both GPC and OPCC is similar. In other
words, the drying shrinkage, which is the difference between the autogenous
shrinkage and the total shrinkage, is less in GPC than in OPCC. Therefore, the
autogenous shrinkage contributed approximately 70% of the total shrinkage for GPC,
while it was only about 45% for OPCC. This would be due to higher self-desiccation
and chemical shrinkage in the slag-based GPC, which also results in higher mesopore
volume and more compact pore structure than OPCC [153]. The activator
concentration also has an effect on autogenous shrinkage of GPC as test result by
Neto, Cincotto and Repette shows increase in autogenous shrinkage by
approximately 60% when the activator concentration is increase from 6.75% to
12.15% weight of binder materials, due to more intense reaction to form more
mesopore volume [152]. Therefore, the high activator concentration (10% weight of
binder materials) in GPC in this study can also contribute to high autogenous
shrinkage.
The weight of these shrinkage specimens was also measured as shown in Figure 4.8.
There is loss of weight observed in all specimens. Such loss of weight is due to loss
of evaporable free water from the specimens to the environment. Although
GPC300614-Sh exhibited more shrinkage strain than OPCC020714-Sh, it lost less
water than OPCC020714-Sh. This results agreed with the previous study by Collins
and Sanjayan [60], which they concluded the mechanism of high drying shrinkage of
GPC was not entirely linked to the amount of loss of water, and it would be related to
the finer pore size distribution in GPC than in OPCC. However, they did not conduct
experiment on sealed specimens to investigate the autogenous shrinkage and treated
the measured shrinkage strain as drying shrinkage strain. This ignores the high
autogenous shrinkage in GPC as it was observed from the results from this research.
Therefore, the actual drying shrinkage in GPC would be low. The finer pores in GPC
can slow down the rate of loss in evaporable free water and thus less water loss was
recorded. Although the sealed specimens still lost weight in this case, the losses are
significant less than those in the drying specimens. This suggests that the sealing
method used in the specimens can be considered effective.
88
Figure 4.8: The change of weight of shrinkage specimens of GPC and OPCC related
to before test condition
4.5 Creep of GPC and OPCC
There were five series of creep test data for GPC and three series for OPCC. The
result of the GPC specimens loaded at 28 days was compared with the OPCC
specimens loaded at same age. Since there was limited test data available on creep of
GPC subjected to long testing period by the time when this research was conducted,
the GPC specimens were tested for more than 1 year to investigate the creep
behaviour at later ages. The effect of age of loading on GPC was also investigated.
The details of the specimens under creep test are shown in Table 4.4.
89
Table 4.4: Details of concrete specimens under creep test
Specimen ID Test details Test period

GPC221013-28d Loaded at age of 28 days Unloaded at 1043 days
GPC121113-14d Loaded at age of 14 days
GPC300615-28dS Loaded at age of 28 days under
sealed condition
OPCC300713-28d Loaded at age of 28 days Unloaded at 400 days
OPCC010715-28d Loaded at age of 28 days
OPCC020715-28dS Loaded at age of 28 days under
sealed condition
4.5.1 Elastic modulus obtained from creep test
According to AS1012.16-1996 [206], the instantaneous elastic modulus can be

calculated in creep test by dividing the applied load by the average strain
immediately after loading. The results for each specimens are shown in Table 4.5.
Table 4.5: Instantaneous elastic modulus of GPC and OPCC in creep test
Specimen ID Applied load Average Instantaneous

(MPa) instantaneous elastic modulus
strain (×10-6) (MPa)
GPC221013-28d 17.9 663 26,970
GPC121113-14d 16.7 577 28,970
GPC121113-28d 15.5 555 27,970
GPC300615-28dS 19.6 727 26,940
GPC150715-7d 18.1 675 26,810
OPCC300713-28d 19.9 633 31,470
OPCC010715-28d 19.6 615 31,850
OPCC020715-28dS 19.6 602 32,520
90
The elastic modulus of GPC specimens found in this way was overall less than
29,000 MPa, which was lower than that of OPCC specimens (above 31,000 MPa).
This indicates that OPCC is stiffer than GPC under similar grade. This trend agreed
with the previous findings in section 4.3.3. However, it is found that the results
shown in Table 4.5 are generally less than the results from standard test for
determining elastic modulus. It would be due to the influence by the drying on the
creep specimens, because drying condition reduces the stiffness of concrete [219].
Another cause would be due to the delay of time when the strain was measured
immediately after loading. Since the load was applied by using a hand jack and this
process was slower than the load rate of 20 MPa/min specified by the standard test
for determining elastic modulus, small amount of creep effect would have been taken
place during loading [206]. This leads to larger measurement in instantaneous strain
and thus lower calculated elastic modulus.
4.5.2 Comparison of creep behaviour between GPC and OPCC
The creep behaviour of GPC is investigated through the comparison with OPCC
under standard creep test. Figure 4.9a shows the creep coefficient of GPC and OPCC
up to age of 650 days. All four series of creep data were loaded at age of 28 days.
The creep coefficients of two GPC specimens, GPC221013-28d and GPC121113-
28d, at 400 days of loading were 2.93 and 3.72, which are higher than that of two
OPCC specimens, OPCC300713-28d and OPCC0101715-28d, 2.17 and 1.77,
respectively. Besides, it is interesting to note that the creep behaviours are different
between GPC and OPCC. In general, the GPC specimens exhibits a lower creep than
the OPCC specimens in the early age but the creep rate of GPC does not reduce
quickly in the later ages (see Figure 4.9b). The creep trend of GPC shows a
significant growth even after one year of loading period. This is different to OPCC in
this study, which shows the typical creep behaviour of OPCC with creep rate
decreasing quickly in the first three months of loading. A finite ultimate creep
coefficient for GPC is not anticipated within this test period. Therefore, GPC would
yield significant higher creep effect than OPCC in later ages.
91
(a)
(b)
Figure 4.9: Creep coefficient of GPC and OPCC: (a) up to age of 500 days, (b) in
first three months of loading
92
The difference in creep behaviour can be due to the difference between the
microstructure of GPC and OPCC. The main hydration product of OPCC is calcium
silicate hydrates, or C-S-H gel [1], while the main reaction product of slag-based
GPC is an aluminium-substituted C-A-S-H type gel [51]. The microstructure of GPC
is the cross-linked and non-cross-linked tobermorite-based structures, but that of
OPCC is the non-cross-linked structures [50]. The response of these gels subjected to
sustain stress would be different. In addition, previous research had found that the
pore size distribution of GPC showed more mesopores (pore size < 50 nm) than that
of OPCC [60]. These would influence the visco-elastic creep behaviour and the
diffusion of free water within the pore network, and lead to difference creep
behaviours of GPC and OPCC. Further details of microstructural analysis of GPC
and OPCC are presented in Chapter 5.
The basic creep of GPC and OPCC was obtained through the sealed specimens under
creep test. The total creep coefficient and the basic creep coefficient of GPC and
OPCC are shown in Figure 4.10. All these specimens were loaded at 28 days. At 615
days, the total creep and basic creep coefficients of GPC are 3.53 and 2.07, observed
from GPC221013-28d and GPC300615-28dS, while the total creep and basic creep
coefficients of OPCC are 1.93 and 1.52, observed from OPCC010715-28d and
OPCC020715-28dS, respectively. Although the total creep coefficient of
GPC221013-28d is 80% higher than that of OPCC010715-28d, there is only a small
difference between the basic creep coefficients of GPC300615-28dS and
OPCC020715-28dS. However, the creep of GPC300615-28dS was still larger than
that of OPCC020715-28dS by 35% approximately. The drying creep can be derived
from the difference between the total creep and the basic creep. In this case, the
drying creep of GPC at 615 days would be 1.46 and that of OPCC would be 0.41.
This suggests that drying has a bigger impact on the creep behaviour of GPC than
OPCC, since the drying creep is almost three times higher in GPC compared to
OPCC. Therefore, it is suggested that sealing can be used to limit the creep
behaviour of GPC when the concrete structures are creep sensitive. At this stage, the
mechanism of drying creep in GPC is not fully understood yet. More works are
required on creep test for GPC subjected to various levels of humidity.
93
Figure 4.10: Total creep and basic creep coefficients of GPC and OPCC (Total creep
from specimens marked as solid lines, while basic creep from specimens marked as
broken lines.)
It was found that GPC221013-28d and GPC300615-28dS exhibited similar creep

behaviour in the first two weeks of loading. Such behaviour is different to OPCC,
which presents a creep difference between OPCC010715-28d and OPCC020715-
28dS from the early stage of loading. It can be explained by the fine pore sizes in the
pore network of GPC [60], which would increase the resistance of diffusion of pore
solution from the interior of the specimen to the atmosphere, since the flow rate is
low in the fine pores according to Darcy’s Law. Therefore, the interior of drying
specimen would be similar to sealed condition in the early stage of loading. This
delays the drying effect on GPC.
4.5.3 Influence of age of loading and strength on creep of GPC
Besides the standard creep test carried out at age of 28 days, the GPC specimens
were also tested at ages of 7 days and 14 days under same drying condition. The test
results of creep coefficient of these GPC specimens are shown in Figure 4.11. The
creep coefficient of GPC121113-14d is higher than that of GPC121113-28d. Both of
94
these GPC specimens are from the same batch of concrete mix, so the variation
between them is small. This result suggests that more mature GPC exhibits less creep
potential. This behaviour is similar to OPCC, as creep decreases with increasing
degree of hydration [119]. Similar mechanism would happen on GPC as the
microstructure refined over further geopolymerisation. Another explanation would be
less evaporable free water existing in the older concrete under drying, which
decreases the creep potential for drying creep [118].
Figure 4.11: Creep coefficient of GPC loaded at different ages
There is an exception that GPC150715-7d exhibited less creep than both

GPC121113-14d and GPC121113-28d. This would be due to the lower compressive
strength of GPC121113-14d and GPC121113-28d than that of GPC150715-7d as
shown in Table 4.2. In the comparison of GPC221013-28d and GPC121113-28d, the
former shows a higher compressive strength and lower creep than the latter. The
variation in strength would occur due to the quality of binder materials, the mixing
procedure (plant mix/laboratory mix), w/b, etc. The higher strength of GPC would
link to lower creep profile through finer pore structure. This is similar to reports of
lower creep strain in OPCC with higher compressive strength [220, 221]. OPCC with
95
lower strength has higher relative volume of pores and more water content occupying
these pores. The movement of this water can result in higher final creep [220].
Therefore, when taking the effect of strength into account, the effect of age of
loading can still be applicable on GPC150715-7d, which exhibited higher creep
coefficient than GPC221013-28d due to higher strength.
4.5.4 Creep recovery
Creep recovery is a common phenomenon in creep of OPCC when the sustained load
on concrete is removed, referring the gradual recovery of strain over a period of time
[128]. In this research, only one of each GPC and OPCC specimens were unloaded
and allowed for observation of creep recovery. Although these specimens were
unloaded at different ages, the creep recovery can be observed on both specimens.
The available results of creep recovery of GPC and OPCC are shown in Figure 4.12.
At 150 days after unloaded from creep test, both GPC and OPCC show similar creep
recovery strain of approximately 7 microstrains per MPa of applied stress. Besides,
irreversible deformation is observed since the sum of the immediate strain due to
unloading and the recovery of strain is not equal to the sum of the instantaneous
strain due to loading and the creep strain. The irreversible deformation in GPC is
significantly larger than that of OPCC. Part of the reason would be due to larger
creep strain of GPC under longer loading period than OPCC. The instantaneous
strain due to loading is 32 and 37 microstrains per MPa of applied stress for OPCC
and GPC, and the elastic strain due to unloading is 28 and 40 microstrains per MPa
for OPCC and GPC. Therefore, OPCC was getting stiffer over the test period, but
GPC was not.
96
Figure 4.12: Creep recovery of GPC and OPCC
4.6 Creep models for GPC
Creep model is essential in structural design for long-term performance. However,

the current creep models are developed for OPCC, based on the vast research in the
past. The research on creep behaviour of GPC is little compared to OPCC. Therefore,
there is not any standard design model for GPC at this stage. In this section, the
performance of the standard creep models on GPC is assessed. Suggestions are made
to improve the accuracy of prediction of creep behaviour for GPC.
4.6.1 Prediction by standard creep models
The creep models from a number of design standards and models were used to
estimate the creep response of the GPC and OPCC specimens; these creep models
included AS3600-2009 [181], Eurocode2 [182], Model B3 [11], ACI model [183]
and CEB fib-model code 2010 [184]. The prediction of creep curve was done by
using these creep models on the concrete specimens with the obtained material
properties, such as compressive strength and age of loading. There were a few
assumptions made for the creep estimation from these models in application on GPC:
97
1. Because there was no specification on sealed condition, the creep of sealed
specimens was calculated based on 100% RH, or as basic creep effect.
2. There was no detailed guideline in AS3600-2009 for sealed condition, in
which only environmental conditions such as inland and near-coastal were
described. The factor k4 was assumed to be 0.4 for sealed condition in this
case, since 0.7 was for RH of 50% and 0.5 was for RH of 70%.
3. The factor α1 accounting for type of cement in model B3 was assumed to be
1.1 for slag-based geopolymer, instead of 1.0 for general purpose OPC. This
is due to the fast early strength of slag-based GPC, similar to type III cement.
4. The OPC used in this case can be considered as cement strength class 42.5N
for its normal strength grade, so the coefficient for binder type in CEB fib-
model code 2010, α, would be assumed as 0 for OPCC. Due to the fast early
strength development in slag-based GPC, the cement strength class of GPC
was assumed to be similar to 42.5R. α = 1 was used for GPC.
The predictions of creep curve for each creep test up to approximately one year are
shown in Figure 4.13. It can be seen that none of these creep models can predict the
creep curves of all GPC specimens properly. Although AS3600-2009 gives the
closest estimation on GPC due to its highest estimated values among all models, the
predicted creep curves overestimates the creep effect at early age and underestimated
at later ages. This is because all creep models present a typical creep behaviour of
OPCC, in which most creep occurs in the first few months of loading period. These
models trend to approach approximately 90% of ultimate creep in 2 to 3 years of
loading period. Therefore, the test data of the OPCC specimens lie between the creep
estimations by the creep models. The results of creep test in GPC specimens show
that the creep rate is moderate in the early stage, but it does not reduce a lot in the
first few months of loading period. The creep curves of GPC specimens still increase
at a high creep rate after one year of loading. As a result, all creep models
overestimate the creep behaviour of GPC in early age and underestimate it in later
ages. The pattern of creep behaviour of GPC is not predicted well with these current
creep models. This would yield a large error in prediction of multi-decade creep
behaviour. This result is expected since all these creep models are designed for
98
OPCC originally. Therefore, these creep models have to be modified accordingly, in
order to predict the creep response of GPC.
(a) GPC150715-7d
(b) GPC121113-14d
Figure 4.13: Prediction of creep coefficient with standard creep models
99
(c) GPC221013-28d
(d) GPC121113-28d
Figure 4.13 (cont.): Prediction of creep coefficient with standard creep models
100
(e) GPC300615-28dS
(f) OPCC300713-28d
101
(g) OPCC010715-28d
(h) OPCC020715-28dS
102
4.6.2 Recommended modification to standard creep models for GPC
The modification to the standard creep models above to suit the creep behaviour of
GPC was done by fitting to the five series of 1-year period test data of GPC in this
project. A few factors were considered for creep behaviour based on this result data,
including the influence of age of loading, compressive strength and sealed/drying
conditions.
4.6.2.1 GPC_AS3600
The creep model in AS3600-2009 consists of a basic value for creep coefficient,
φcc.b, and four factors, k2, k3, k4 and k5. They are computed based on the properties of
concrete, age at loading, duration of loading, and the external environment. In the
calculation of design creep coefficient, φcc, during the loading period, k2 is
responsible for the development of the creep prediction, because it depends on time
after loading, t. These factors are suggested to be modified for GPC as shown in
Table 4.6.
Table 4.6: GPC creep model based on AS3600-2009
Model factors GPC_AS3600 AS3600-2009

Basic value for 1700 Table 3.1.8.2, Clause 3.1.8 [181]
𝜑𝑐𝑐.𝑏 = 1.5
creep 𝑓′𝑐
coefficient, φcc.b
Time- 𝛼2 𝑡 0.55 𝛼2 𝑡 0.8
𝑘2 = 𝑘2 =
dependent 𝑡 0.5 + 0.35𝑡ℎ 𝑡 0.8 + 0.15𝑡ℎ
factor, k2 𝛼2 = 0.5 + 1.4𝑒 (−0.004) 𝛼2 = 1 + 1.12𝑒 (−0.008𝑡ℎ)
th = 150 mm for sealed condition
Age of loading 2.2 2.7
𝑘3 = 𝑘3 =
factor, k3 0.7 + log⁡(𝜏) 1 + log⁡(𝜏)
Environment k4 = 0.7 for 50% RH k4 = 0.7 for 50% RH

factor, k4 k4 = 1 for sealed condition Not specified for sealed condition
High strength 𝑘5 = 1 when f’c ≤ 50MPa 𝑘5 = 1 when f’c ≤ 50MPa
concrete factor, 𝑘5 = (2−𝛼3)−0.02(1−𝛼3)f’c 𝑘5 = (2−𝛼3)−0.02(1−𝛼3)f’c
k5 when f’c > 50MPa when f’c > 50MPa
103
The basic creep coefficient, φcc.b, is calculated regarding the influence by
compressive strength. k2 is modified to suit the creep behaviour of GPC. The
hypothetical thickness, th, is 75 mm for all test specimens expect the sealed condition
in this project. When the specimen is sealed, th would become infinity since there is
no exposed surface according to the calculation from AS3600-2009 [181]. This
yields the value of k2 approaching zero and leads to zero for creep coefficient, which
does not agree with the observation from experiment. Therefore, the creep model
from AS3600-2009 is not suitable for creep estimation on sealed concrete, and such
condition is not specified as well. In this case, it is suggested to apply a maximum
limit value for th under sealed condition. For GPC, th = 150 mm is estimated for
sealed condition, while th = 400 mm is estimated for sealed OPCC. The difference in
size is due to the finer pore sizes in GPC, which increases the resistance for micro-
diffusion, so that more water would be kept in GPC than OPCC.
The age of loading factor is modified corresponded to the time age of loading, τ,
since it is found that GPC develops high strength in early age. Because no sealed
condition is specified in AS3600-2009, the environmental factor, k4, which is mainly
for drying effect, had no impact on sealed condition, and thus it is set as 1 for sealed
condition. The factor k5 remains unchanged, since GPC with higher strength is not
tested. Therefore, φcc at any time of loading for GPC can be taken as:
𝜑𝑐𝑐 = 𝑘2 𝑘3 𝑘4 𝑘5 𝜑𝑐𝑐.𝑏 (4.4)
4.6.2.2 GPC_Eurocode 2
The creep model of Eurocode 2 states the calculation of the design creep coefficient,
φ(t,t0), as the product of the notional creep coefficient, φ0, and the factor which
describes the development of creep with time, βc(t,t0). Adjustment to adopt this creep
model on GPC is mainly done to the factor βc(t0,t), because it is responsible for the
development of the creep prediction. Besides, the factors which allows the effect of
RH are required to be modified, since the original model underestimates the creep of
the specimens under sealed condition. These modified factors include φRH and βH.
The adjustment is done based on the same assumption of 100% RH and infinity for th
under the sealed condition. The factor β(fcm), for the effect of compressive strength of
104
concrete, and the age factors β(t0) are used in the calculation of the national creep
coefficient, φ0.
𝜑0 = 𝜑𝑅𝐻 𝛽 (𝑡0 )𝛽(𝑓𝑐𝑚 ) (4.5)
The modified factors are shown in Table 4.7. Because the GPC with low
compressive strength is not tested in this project, φRH and βH are not specified for
compressive strength lower than 40 MPa. α1, α2 and α3 are factors influenced by the
concrete strength, which remain unchanged for GPC creep model. Therefore, φ(t0,t)
is calculated as below.
⁡𝜑(𝑡0 , 𝑡) = 𝛽𝑐 (𝑡0 , 𝑡)𝜑0 (4.6)
Table 4.7: GPC creep model based on Eurocode 2-2004
Modified factors GPC_Eurocode 2 Eurocode 2-2004

Relative 𝜑𝑅𝐻 = (1 +
1−𝑅𝐻/100
𝛼1 ) 𝛼2 𝜑𝑅𝐻 = (1 +
1−𝑅𝐻/100
)
0.5 3√𝑡ℎ 0.1 3√𝑡ℎ
humidity factor,
φRH for fcm > 40 MPa for fcm ≤ 35 MPa
1−𝑅𝐻/100
𝜑𝑅𝐻 = (1 + 𝛼1 ) 𝛼2
0.1 3√𝑡ℎ
for fcm > 35 MPa

Age of loading 4.9 1
𝛽(𝑡0 ) = 𝛽(𝑡0 ) =
factor, β(t0) 0.6 + 𝑡0 0.3 0.1 + 𝑡0 0.2
Concrete 1500 16.8

𝛽(𝑓𝑐𝑚 ) = 𝛽(𝑓𝑐𝑚 ) =
strength factor, 𝑓𝑐𝑚 1.55 𝑓𝑐𝑚 0.5
β(fcm)
Factor of 𝑡 − 𝑡0 0.4
𝑡 − 𝑡0 0.3
𝛽𝑐 (𝑡0 , 𝑡) = [ ] 𝛽𝑐 (𝑡0 , 𝑡) = [ ]
development of 5𝛽𝐻 + 𝑡 − 𝑡0 𝛽𝐻 + 𝑡 − 𝑡0
creep, βc(t,t0) 𝛽𝐻 = 1.5[1 + (0.012𝑅𝐻)18 ]𝑡ℎ + 𝛽𝐻 = 1.5[1 + (0.012𝑅𝐻)18 ]𝑡ℎ +
250𝛼3 ≤ 750𝛼3⁡ 250⁡ ≤ 1500
for fcm > 40 MPa for fcm ≤ 35 MPa
𝛽𝐻 = 1.5[1 + (0.012𝑅𝐻)18 ]𝑡ℎ +
250𝛼3⁡ ≤ 1500𝛼3⁡
for fcm > 35 MPa
105
4.6.2.3 GPC_B3
There are two main components in model B3 for creep modelling. C0(t,t’) is the
compliance function for basic creep and Cd(t,t’,t0) is the additional compliance
function due to simultaneous drying. Both compliance functions are time-dependent.
In this case, modification is done on both functions for prediction of the creep
behaviour of GPC. Based on the test results from five series of GPC creep specimens
in this project, the following suggestions are given to adopt model B3 for creep of
GPC as shown in Table 4.8.
Table 4.8: GPC creep model based on model B3
Model components GPC_B3 Model B3

Compliance 𝐶0 (𝑡, 𝑡′) 𝐶0 (𝑡, 𝑡′)
function for basic = 𝑞2 𝑂(𝑡, 𝑡′)+𝑞3 ln⁡[1 = 𝑞2 𝑄(𝑡, 𝑡′)+𝑞3ln⁡[1
𝑡 𝑡
creep, C0(t,t’) + (𝑡 − 𝑡′)𝑛 ]+𝑞4𝑙𝑛 ( ) + (𝑡 − 𝑡′)𝑛 ]+𝑞4𝑙𝑛 ( )
𝑡′ 𝑡′
Related factors 𝑂(𝑡, 𝑡 ′ ) = (𝑡 ′ )−𝑚 {𝑙𝑛[1 𝑄(𝑡, 𝑡′) from Table 1.1 [11]
+ (𝑡 − 𝑡 ′ )𝑛 ]}2
𝑞2 = 185.4𝑐 0.5 𝑓𝑐−0.9
𝑞2 = 360𝑏 0.5 𝑓𝑐−1.65
𝑞3 = 0.29(𝑤/𝑐)4 𝑞2
4
𝑞3 = 0.29(𝑤/𝑏) 𝑞2
𝑞4 = 20.3(𝑎/𝑐)−0.7
−0.7
𝑞4 = 20.3(𝑎/𝑏)
m = 0.5, n = 0.1
m = 0.2, n = 0.5
Additional 𝐶𝑑 (𝑡, 𝑡 ′ , 𝑡𝑜 ) = 𝑞5 [𝑒 −8𝐻(𝑡) 𝐶𝑑 (𝑡, 𝑡 ′ , 𝑡𝑜 ) = 𝑞5 [𝑒−8𝐻(𝑡)
compliance − 𝑒 −8𝐻(𝑡
′ )
𝑜 ]
0.65
− 𝑒 −8𝐻(𝑡
′ ) 0.5
𝑜 ]
function due to
simultaneous
drying, Cd(t,t’,t0)
Drying factor, q5 𝑞5 = 13 × 106 𝑓𝑐−1.55 |𝜀𝑠ℎ∞ |−0.6 𝑞5 = 7.57 × 105 𝑓𝑐−1 |𝜀𝑠ℎ∞ |−0.6
It is noted that b was used to stand for the total amount of the binder materials
instead of c for cement, so w/b is the water to binder ratio and a/b is the aggregates to
binder ratio. The change due to variation in composition in mix design is not
examined, because only one mix design is tested. Therefore, q3 and q4 are not
modified. Changes to the creep model are mainly done on factors q2, m and n, and
O(t,t’) is used to replace Q(t,t’). For Cd(t,t’,t0), since it is reported that GPC exhibits
higher shrinkage than OPCC [60], it is suggested the factor α1 = 0.85 for GPC. It is
106
also found that GPC is affected by drying from the test results. Therefore, the
equations for calculation of Cd(t,t’,t0) is modified accordingly. Therefore, the
compliance function J(t,t’) and the creep coefficient φ(t,t’) can be calculated:
𝐽(𝑡, 𝑡′) = 𝑞1 + 𝐶0 (𝑡, 𝑡′) + 𝐶𝑑 (𝑡, 𝑡 ′ , 𝑡𝑜 ) (4.7)
𝜑(𝑡, 𝑡′) = 𝐸 (𝑡 ′ )𝐽(𝑡, 𝑡 ′ ) − 1 (4.8)
Because of the compliance functions of basic creep and drying creep, the creep of
concrete under sealed condition can be predicted by ignoring compliance function of
drying creep.
4.6.2.4 GPC_ACI
The creep model by ACI committee 209 includes a few constants, which are
determined by fitting data from the property tests. From the test data of GPC
samples, it is found that using a single value for constant, ψ, cannot predict the creep
behaviour of GPC properly. It is suggested to replace ψ with two constants, ω and λ.
The ultimate creep coefficient, φu, can be estimated from the correction factors of age
of loading, external condition, geometry and concrete composition. The proposed
creep model is presented in Table 4.9.
It was suggested ω = 0.6, λ = 0.45 for slag-based GPC in this case. The other
correction factors regarding the size and temperature (γh, γs and γψ) remain
unchanged, since other test conditions are not included. However, because this creep
model does not take the compressive strength of the concrete into account, the error
of predicted creep curve would be large. It is suggested to introduce a new factor,
γfcm, for the effect of strength of concrete.
107
Table 4.9: GPC creep model based on ACI model
Model components GPC_ACI ACI model

Creep coefficient, φt 𝜑 = 𝑡𝜔 𝑡𝜓
𝑡 𝜑
𝜆 𝑢 𝜑𝑡 = 𝜑
10 + 𝑡 𝑑 + 𝑡𝜓 𝑢
Ultimate creep 𝜑𝑢 = 2.35𝛾𝑐 𝜑𝑢 = 2.35𝛾𝑐
coefficient, φu 𝛾𝑐 = 𝛾𝑓𝑐𝑚 𝛾𝑙𝑎 𝛾𝜆 𝛾ℎ 𝛾𝑠 𝛾𝜓 𝛾𝑐 = 𝛾𝑙𝑎 𝛾𝜆 𝛾ℎ 𝛾𝑠 𝛾𝜓
Age of loading 𝛾𝑙𝑎 = 1.6𝑡𝑙𝑎 −0.2 𝛾𝑙𝑎 = 1.25𝑡𝑙𝑎 −0.118

factor, γla for moist cured concrete for moist cured concrete
𝛾𝑙𝑎 = 1.13𝑡𝑙𝑎 −0.094
for steam cured concrete
Relative humidity 𝛾𝜆 = 1.27 − 0.006𝑅𝐻 𝛾𝜆 = 1.27 − 0.67𝑅𝐻
factor, γλ for RH > 40%
Strength of 𝛾𝑓𝑐𝑚 = 400𝑓𝑐𝑚 −1.5 None
concrete factor, γfcm
4.6.2.5 GPC_CEB-fib
The creep model by CEB-fib model code 2010 is the latest creep model among all
these models, it consists of two components to predict basic creep and creep due to
drying. Modification of the creep model for creep behaviour of GPC is done to both
components; the basic creep coefficient, φbc(t,t0), and the drying creep coefficient,
φdc(t,t0). Therefore, the creep coefficient, φ(t,t0), is calculated as the sum of these two
components.
𝜑(𝑡, 𝑡0 ) = 𝜑𝑏𝑐 (𝑡, 𝑡0 ) + 𝜑𝑑𝑐 (𝑡, 𝑡0 ) (4.9)
The proposed creep model is presented in Table 4.10. The strength factors for both
basic and drying creep calculations, βbc(fcm) and βdc(fcm) are modified. The effect of
age of loading is associated in the factors of creep development, βbc(t,t0) and βdc(t,t0).
Since only one size of specimen is tested, the size factor, βh, is not modified in this
case. Similar to GPC_B3, the creep coefficient of sealed GPC specimen can be
obtained from φbc(t,t0) only.
108
Table 4.10: GPC creep model based on CEB-fib model code 2010
Model GPC_CEB-fib CEB-fid model code 2010

components
Basic creep 𝜑𝑏𝑐 (𝑡, 𝑡0 ) = 𝛽𝑏𝑐 (𝑓𝑐𝑚 )𝛽𝑏𝑐 (𝑡, 𝑡0 ) 𝜑𝑏𝑐 (𝑡, 𝑡0 ) = 𝛽𝑏𝑐 (𝑓𝑐𝑚 )𝛽𝑏𝑐 (𝑡, 𝑡0 )
coefficient,
φbc(t,t0)
Strength 35 1.8
𝛽𝑏𝑐 (𝑓𝑐𝑚 ) = 𝛽𝑏𝑐 (𝑓𝑐𝑚 ) =
factor, βbc(fcm) (𝑓𝑐𝑚 )1.6 (𝑓𝑐𝑚 )0.7
Basic creep 𝛽𝑏𝑐 (𝑡, 𝑡0 ) = 𝛽𝑏𝑐 (𝑡, 𝑡0 ) =

development 30
2
30
2
factor, βbc(t,t0) 2.3 {𝑙𝑛 [(

𝑡0,𝑎𝑑𝑗
+ 0.035) (𝑡 − 𝑡0 ) 𝑙𝑛 [(
𝑡0,𝑎𝑑𝑗
+ 0.035) (𝑡 − 𝑡0 ) + 1]
1.5
𝛼
9
+ 1]} 𝑡0,𝑎𝑑𝑗 = 𝑡0,𝑇 ( + 1) > 0.5 days
2+𝑡0,𝑇 1.2
9
𝛼 α = -1 for strength class 32.5 N
𝑡0,𝑎𝑑𝑗 = 𝑡0,𝑇 (2+𝑡 + 1)
0,𝑇
1.2
α = 0 for strength class 32.5 R, 42.5N
α = 1.5 for GPC α = 1 for strength class 42.5 R, 52.5 N,
52.5 R
Drying creep 𝜑𝑑𝑐 (𝑡, 𝑡0 ) 𝜑𝑑𝑐 (𝑡, 𝑡0 )

coefficient, = 𝛽𝑑𝑐 (𝑓𝑐𝑚 )𝛽(𝑅𝐻)𝛽𝑑𝑐 (𝑡0 )𝛽𝑑𝑐 (𝑡, 𝑡0 ) = 𝛽𝑑𝑐 (𝑓𝑐𝑚 )𝛽(𝑅𝐻)𝛽𝑑𝑐 (𝑡0 )𝛽𝑑𝑐 (𝑡, 𝑡0 )
φdc(t,t0)
Strength 650 412
𝛽𝑑𝑐 (𝑓𝑐𝑚 ) = 𝛽𝑑𝑐 (𝑓𝑐𝑚 ) =
factor, βdc(fcm) (𝑓𝑐𝑚 )1.4 (𝑓𝑐𝑚 )1.4
Humidity 1.4(1 − 𝑅𝐻) 1 − 𝑅𝐻

𝛽(𝑅𝐻) = 𝛽(𝑅𝐻) =
factor, β(RH) 0.1𝑡 0.1𝑡ℎ 1/3
(1000ℎ )1/3 (1000)
Drying creep (𝑡 − 𝑡0 )
𝑟(𝑡0 )
(𝑡 − 𝑡0 )
𝑟(𝑡0 )
𝛽𝑑𝑐 (𝑡, 𝑡0 ) = [ ] 𝛽𝑑𝑐 (𝑡, 𝑡0 ) = [ ]
development 𝛽ℎ + (𝑡 − 𝑡0 ) 𝛽ℎ + (𝑡 − 𝑡0 )
factor, βdc(t,t0) 35 0.5 35 0.5
𝛽ℎ = 1.5𝑡ℎ + 250 ( ) 𝛽ℎ = 1.5𝑡ℎ + 250 ( )
𝑓𝑐𝑚 𝑓𝑐𝑚
1 35 0.5
𝑟(𝑡0 ) = < 1500 ( )
0.5 + 𝑡0,𝑎𝑑𝑗 −0.5 𝑓𝑐𝑚
1
𝑟(𝑡0 ) =
2.3 + 3.5 × 𝑡0,𝑎𝑑𝑗 −0.5
109
4.6.3 Comparison of the modified creep models
The results of the improved creep models are shown in Figure 4.14. The root-mean-
square error of the improved creep models to the test data is shown in Table 4.11.
Overall, all modified creep models match well with the test data of creep of GPC.
The average values of r2 are above 0.95 in all modified models, although
GPC_AS3600 and GPC_Eurocode 2 show lower than 0.95 in some cases. Among
them, the creep models GPC_B3 and GPC_CEB-fib perform the best with the
average value of r2 above 0.975. The lowest average r2 value of 0.919 is found in
GPC_Eurocode 2. This is because GPC_B3 and GPC_CEB-fib estimate basic and
drying creep coefficients in separate calculations, and it allows more accurate
estimation on basic creep coefficient from the sealed GPC specimen than the other
models. Besides, the variation between the models is due to the contributing factors
in each models. This includes the creep development functions which are presented
in different forms among these creep models, as well as the factors of strength of
concrete, age of loading and humidity. Some factors would have more weight in the
creep model, and thus influence the performance.
(a) GPC150715-7d
Figure 4.14: Creep coefficient of GPC by modified creep models
110
(b) GPC121113-14d
(c) GPC221013-28d
Figure 4.14 (cont.): Creep coefficient of GPC by modified creep models
111
(d) GPC121113-28d
(e) GPC300615-28dS
Figure 4.14 (cont.): Creep coefficient of GPC by modified creep models
112
Table 4.11: Root-mean-square error (r2) of the GPC creep models to test data
GPC test data GPC_ GPC_ GPC_ GPC_ GPC_
AS3600 Eurocode2 B3 ACI CEB-fib
GPC150715-7d 0.981 0.993 0.968 0.973 0.996
GPC121113-14d 0.99 0.969 0.993 0.989 0.980
GPC221013-28d 0.958 0.959 0.986 0.970 0.993
GPC121113-28d 0.952 0.919 0.985 0.961 0.991
GPC300615-28dS 0.930 0.939 0.956 0.962 0.983
Average 0.962 0.956 0.977 0.971 0.989
The modified creep models are then used to predict the creep coefficient of
GPC121113-14d, GPC221013-28d and GPC121113-28d for period beyond one year
of loading period. The results are shown in Figure 4.15 together with the test data.
GPC_AS3600 gives the best prediction in GPC121113-28d, but it overestimates
GPC121113-14d and GPC221013-28d. GPC_Eurocode 2 overestimates the creep
coefficient in early age of all specimens. GPC_B3 underestimates the creep
behaviour of GPC121113-14d in later age. GPC_ACI and GPC_ CEB-fib match the
test data well in all cases. In general, the proposed creep models performed well in
creep prediction and thus they are applicable on the GPC.
113
(a) GPC_AS3600
(b) GPC_Eurocode 2
Figure 4.15: Prediction of creep coefficient on GPC
114
(c) GPC_ B3
(d) GPC_ACI
Figure 4.15 (cont.): Prediction of creep coefficient on GPC
115
(e) GPC_CEB-fib
Figure 4.15 (cont.): Prediction of creep coefficient on GPC
The 30-year creep coefficient, which is usually considered as ultimate creep for long-
term design in concrete structures, is predicted by the GPC creep models for the
above specimens. The results are calculated and listed in Table 4.12. Among these
modified creep models, GPC_Eurocode 2 gives the lowest prediction values, while
GPC_ACI gives the highest among all. These provide a possible range for the
ultimate creep of GPC. However, there is no creep test on GPC which has been
carried out for such long period, real test data can not be obtained to confirm. All
these predictions of ultimate creep coefficient of GPC are generally higher than that
of OPCC at similar strength grade, which would be generally in a range 2 to 3. This
indicates the high creep potential of GPC.
116
Table 4.12: Prediction of creep coefficient at 30 years by the modified creep models
GPC test data GPC_ GPC_ GPC_ GPC_ GPC_
AS3600 Eurocode2 B3 ACI CEB-fib
GPC121113-14d 9.88 8.44 8.51 10.77 8.94
GPC221013-28d 6.94 5.57 6.18 7.42 6.41
GPC121113-28d 8.50 7.14 7.67 9.08 7.89
The proposed GPC creep models are calibrated with only five series of creep data of
GPC in this project. In order to fully model the creep behaviour of GPC, further
creep test data are required to investigate the effects of humidity and temperature on
creep behaviour of GPC. Longer term test period, different loading levels and
different compositions would also be done, so that it can match the vast data of creep
of OPCC from tests in last century. When the creep behaviour of GPC can be
predicted accurately, it can be applicable in the practical field.
4.7 Concluding remarks
From the various tests on GPC and OPCC above, GPC showed some similarity and
yet some difference to OPCC. The compressive strength of GPC used in this research
was generally above 45 MPa, together with high tensile strength of 6.75 MPa. GPC
can achieve sufficient strength and workability for structural and construction
purposes. GPC exhibits higher shrinkage and creep coefficient than OPCC. A large
portion, approximately 70%, of total shrinkage of GPC is from autogenous
shrinkage, while compared to 45% in OPCC. This would be due to higher self-
desiccation and chemical shrinkage in the slag-based GPC, which also results in
higher mesopore volume and more compact pore structure than OPCC. Another
factor influencing the shrinkage of GPC would be the method of manufacturing.
Extra water added to the system on site application would increase the actual w/b
117
ratio in GPC from original formulation, so that the higher w/b ratio would result in
higher shrinkage.
GPC exhibits a lower creep than the OPCC in the early age of loading but the creep
rate of GPC does not reduce significantly like OPCC in the later ages. This yields a
higher creep effect in GPC in later ages. It is found that the drying creep of GPC is
almost three times higher than that of OPCC. Therefore, by sealing the GPC to
minimise drying effect, the creep coefficient can be reduced by up to 40% in the first
year of loading period compared to the creep specimens subjected to drying, because
the large portion of drying creep was excluded. Similar to OPCC, increase in strength
of GPC and loading at more mature GPC can also reduce the creep effect.
Because of these creep behaviours of GPC, the current creep models in the design
standards and design codes cannot be used to predict the creep of GPC as they are
designed for OPCC originally. Suggestions are given to improve these creep models
on prediction of five series of GPC creep data in this study and show good results,
but further test data is required to develop a reliable creep model for GPC under
various environmental conditions. The investigation of the mechanism of creep of
GPC would be important in achieving such creep model, but it was not well
understood yet. Since the creep of OPCC is influenced by the viscoelastic behaviour
of the hydrated gels and the pore structure, it is expected that the creep of GPC
would be influenced in similar ways. Study on microstructure in both GPC and
OPCC would provide the answer to that. After that, the improvement to long-term
behaviour of GPC would be achievable.
118
Chapter 5 : Microstructure and creep
behaviour of GPC and OPCC
5.1 Introduction
Research on the microstructure of GPC and OPCC shows that the main binding
phases in these materials are different. The main hydration product in OPCC is C-S-
H gel [1], while the main binding product of slag-based GPC is an aluminium-
substituted C-A-S-H gel [51]. This influences the microstructure of the materials
since the C-S-H gel is a non-cross-linked structures [50], and the C-A-S-H gel is
cross-linked and non-cross-linked tobermorite-based structures [51]. Furthermore,
the pore structure existing within these microstructures is also different between GPC
and OPCC. It was found that the pore size distribution of slag-based GPC showed
more mesopores (pore size < 50 nm) and also lower porosity than those of OPCC
[60].
These differences in microstructures of GPC and OPCC can be linked to the creep
behaviours. Several mechanisms of creep in OPCC were proposed related to the
microstructure. Acker and Ulm [8] explained that creep of OPCC involved the stress-
induced water migration from gel pores to capillary pores for short-term component
of creep, and the viscous flow in the C-S-H sheets for the long-term creep. This
viscous flow of C-S-H was due to sliding of gels toward the jammed state [7]. This
sliding of C-S-H gels resulted in bond breakage in slip and its reformation [6]. Since
the microstructures of GPC is different, the creep mechanism can be different from
that of OPCC.
In this chapter, test samples were taken from the core and the skin layer of the
concrete cylinders of GPC and OPCC for tests to determine the microstructure,
which include XRD, TGA, BSEM and nitrogen adsorption test. In order to study the
relationship between the microstructure and the creep behaviour, samples were
obtained from the test specimens of GPC221013 and OPCC300713 in Chapter 4 at
age of 28 days (before creep test) and creep and control specimens after unloaded
from creep test.
119
5.2 Microstructure of GPC and OPCC
The original microstructure of GPC and OPCC at age of 28 days before creep test
was investigated through the following tests. This can assist the study to understand
the mechanism of creep in GPC.
5.2.1 XRD
XRD was performed on samples obtained from the skin layer and the core of the
GPC and OPCC specimens at age of 28 days. The results are shown in Figure 5.1.
There seems to be no differences between the skin and the core of the GPC samples
observable. Except from the Quartz (SiO2, powder diffraction file (PDF)# 00-005-
0490) from the aggregates, the main crystalline phases identified in the GPC samples
are Gismondine (Ca2Al4Si4O16·9(H2O), PDF# 00-039-1373), the calcium-containing
zeolitic phases Hydrotalcite (Mg6Al2(CO3)(OH)16·4(H2O), PDF# 04-011-5899) and
other poorly ordered Tobermorite like C-A-S-H (Ca4·3Si5·5Al0.5O16(OH)2·4(H2O),
PDF# 00-019-0052) along with traces of Calcite (CaCO3, PDF#00-003-0612). It is
interesting to note the results are in good agreement with the expected
thermodynamic equilibration of a calcium-rich zeolite and a sodium-rich pore
solution over an extended period. These crystalline phases are accompanied by
residual X-ray amorphous binder gel material, visible as a broad ‘amorphous hump’
between 25° and 35° 2θ. Few of these high calcium-containing zeolite type gel has
previously been reported [25, 71, 222, 223] as the main reaction product in alkali-
activated slag. A low intensity peak at 10° 2θ is also observed, assigned to
Hydrotalcite (Mg6Al2(CO3) (OH)16·4(H2O), PDF# 00-041-1428)), which has been
identified in other alkali-activated slag systems [218, 224].
In the OPCC samples, except the quartz presents in the samples due to the
aggregates, the dominant binder phases are crystalline hydration products Portlandite
(Ca(OH)2; PDF# 00-044-1481), Ettringite (Ca6Al2(SO4)3(OH)12·25H2O; PDF # 00-
009-0414) and disorganised C-S-H. The reflections of all phases in this sample were
assigned considering the results reported by Lothenbach et al. [225]. In addition,
120
trace of Calcite is identified in the core OPCC sample after 28 days and a high
content is observable in the skin sample. Since the specimens were removed from
bath and subjected to drying from the age of 7 days, carbonation of the skin is
responsible of these changes along with the decrement of Portlandite due to the
following equation as previously observed in the literatures [226-228].
𝐶𝑂2 + 𝐶𝑎(𝑂𝐻)2 + 𝐻2 𝑂 → 𝐶𝑎𝐶𝑂3 + 𝐻2 𝑂
The little amount of Ettringite also observed in the core OPCC sample is not in the
skin sample, it can also be due to carbonation according to the following equation
[229, 230].
3𝐶𝑎𝑂. 𝐴𝑙2 𝑂3 . 3𝐶𝑎𝑆𝑂4 . 32𝐻2 𝑂 + 3𝐶𝑂2

→ 3𝐶𝑎𝐶𝑂3 + 3(𝐶𝑎𝑆𝑂4 . 2𝐻2 𝑂) + 𝐴𝑙2 𝑂3 . 𝑥𝐻2 𝑂 + (26 − 𝑥 )𝐻2 𝑂
Figure 5.1: X-ray diffractograms of samples at age of 28 days before creep test.
Peaks marked are Ettringite (E), Portlandite (P), Hydrotalcite (H), Calcite (C),
Gismondine (G), Quartz (Q) and C-A-S-H/C-S-H (CS)
121
5.2.2 TGA
TGA for GPC and OPCC samples at age of 28 days for the skin and the core parts
are presented in the Figure 5.2. The drop in approximately 100 to 120 °C in all
samples is related to the release physically bound water in the gel structure [215,
223, 231]. In GPC samples, the dehydration of the gel through loss of loosely bound
water occurred at around 130 °C can be attributed to Gismondine zeolite, as
identified in section 5.2.1. Gismondine reports mass loss as a double peak at around
90 °C and 160 °C [232], and also a minor mass loss around 700 °C consistent with
the minor peak especially in the GPC skin sample. The peaks located at around 270
°C and 340 °C in the GPC skin sample and 350 °C in the core sample have been
identified as dehydration of hydrotalcite according to the literature in the alkali-
activated slag sample [233]. This agrees with the finding in XRD analysis. The
calcium carbonate dehydroxylation can be identified at the small hump between 550
°C and 600 °C in GPC samples, which is related to the decomposition of calcite
[234]. This agrees with the traces of Calcite identified in XRD in both core and skin
samples. This suggests that the Calcite in these specimens is from the anhydrous
slag, since it would be slightly carbonated during storage. No Portlandite was
identified in the TGA results, which confirms the finding in XRD. This is expected
because the formation of C-A-S-H type gels is more favoured than the formation of
Portlandite in activated slag binders with high silica and alumina availability.
Contrary to GPC samples, Portlandite was identified in the TGA results of the OPCC
sample, where the sharp peak of Portlandite is highly visible in the skin and the core
samples at around 460 °C [225]. This agrees with the finding in previous section by
XRD analysis. The intensity of the peak of Calcite is slightly reduced in the OPCC
skin sample due to carbonation, which confirms the XRD analysis. The peak located
at around 650 °C up to 780 °C correspond to the decarbonation of Calcite from the
cement [225]. The peak is more intense in the skin than in the core due to the larger
amount of Calcite due to carbonation of the skin confirming XRD analysis.
122
Figure 5.2: Differential thermograms (DTG) of GPC and OPCC at age of 28 days
before creep test
5.2.3 BSEM
Backscattered electron (BSE) imaging of GPC and OPCC at different magnification

are shown in Figure 5.3. It shows well-distributed paste areas (darker grey regions),
with embedded unreacted anhydrous cement and slag particles (light grey). Coarse
aggregates have many different shade of grey as their composition is not pure. On the
other hand, the fine aggregate (sand) is purely quartz base, hence presents a uniform
shade of light grey. It is also observed that the amount of unreacted particle of slag in
GPC is in higher content in the binder after 28 days compared to the unhydrated
particle of cement in OPCC. Compared with the OPCC binder, it seems that the
binding paste in the GPC is less dense (see Figure 5.3 smallest magnification). They
both show heterogeneously distributed cavities (black) especially around the
interfacial transition zone [71].
123
GPC OPCC
Figure 5.3: Backscattered electron imaging of GPC and OPCC at age of 28 days
124
5.2.4 Pore characteristic
The pore characteristic of GPC and OPCC at age of 28 days was analysed through
the nitrogen adsorption test. The isotherms obtained from this test was used to
investigate the pore size distribution by using DH method [113]. The results of the
pore size distribution of both GPC and OPCC samples are shown in Figure 5.4. It can
be seen that the GPC samples had different pore size distribution from the OPCC
samples. At the core of concrete cylinders, GPC had less pores in the size above 2
nm, but more pores in the size below 2 nm than OPCC. However, GPC sample from
the skin layer showed more pores within a range of 15 to 40 nm than OPCC. It also
had more pores larger than 2 nm than the GPC sample from the core. It is noted that
the concrete cylinders were removed from the bath and subjected to drying at 50%
RH from age of 7 days according to AS1012.16-1996 [206]. Therefore, the large
pores detected from GPC sample from the skin layer was hypothesised to be due to
microcracks caused by drying from 7 days to 28 days. Comparison of the skin and
the core of OPCC samples shows slight more pores above 2 nm in the skin sample.
This would be due to carbonation at the skin of the OPCC [226].
Figure 5.4: Pore size distribution of GPC and OPCC at age of 28 days
125
5.2.5 Difference between microstructure of GPC and OPCC
There are some significant difference between the microstructure of GPC and OPCC.
The results of XRD and TGA show that GPC contains some crystalline phases like
Hydrotalcite and Gismondine, but OPCC contains Portlandite and Ettringite. In
addition, the main binding phase of GPC is C-A-S-H and that of OPCC is C-S-H. A
broad ‘amorphous hump’ between 25° and 35° 2θ of XRD of GPC suggests that
there are more amorphous phases existing in GPC. The amorphous phases consist of
disorganised product of geopolymerisation, and unreacted particle of slag, which also
shows amorphous hump between 25° and 35° 2θ of XRD [235]. Therefore, the creep
response of GPC and OPCC would be different due to the slippage of these phases
under load. More slippage of phases would happen in GPC, and this allows more
creep deformation.
On the other hand, more fine pores are presented in the GPC core sample than the
OPCC core sample. This influences the rate of diffusion of pore solution within the
pore network [85], and thus influences the creep phenomenon as the reducing
distance of the hindered adsorbed layers [130]. Therefore, the creep response of GPC
would be slowed down by the low rate of diffusion from the fine pore network.
5.3 Change of microstructure after creep test
5.3.1 XRD
The results of XRD on samples of GPC and OPCC after creep test are presented in
Figure 5.5. The samples were collected from the core of the specimens, since it is
believed that carbonation affected the skin of the specimens after long test period.
There are no obvious differences between the samples from OPCC creep specimen
and the control specimen (no load applied), which can imply there is no structural
changes of the binding phases presented in OPCC. The binder composition seems to
be equivalent to the one observed earlier in the samples after 28 days of curing and
as it has been observed in the literatures [236, 237].
126
Figure 5.5: X-ray diffractograms of samples after creep test (Creep) and the reference
sample (Control) of GPC and OPCC. Peaks marked are Hydrotalcite (H), calcite (C),
Gismondine (G), Quartz (Q), Diopsite (D), and C-A-S-H/C-S-H (CS), Portlandite
(P), Ettringite (E), Margarite (M), Chabazite (Ch)
For GPC, the Calcite peaks do not show differences in intensity in creep sample,
control sample and 28 days sample, suggesting that this compound in the GPC
samples is a residue of the anhydrous slag. Besides, only a few differences are
noticed between the creep sample and the control sample of GPC. A magnesium rich
variant of Diopside (Ca0.89Mg1.11Si2O6, #PDF 01-087-0698), Margarite (CaAl4Si2 O10,
#PDF 04-013-3004) and Chabazite (NaAl2Si2O6·3H2O, #PDF 00-019-1178) appear
to be in a larger content in the sample after creep test compared to the control
sample. The identification of these extra species in this sample suggests that the
sustained load can involve small changes in the microstructure. It is possible that the
sustained load leads to a higher mobility of the Magnesium ions which facilitate their
substitution with the C-S-H structure and can explain the increment of Mg-rich
Diopside [238].
127
5.3.2 TGA
Differential thermogravimetry data for GPC and OPCC samples after creep test and
the control samples are shown in Figure 5.6. These samples were collected from the
core of the specimens. For the GPC samples, the minor drop around 170 °C is
attributed to the first stage of the thermal dehydration of Hydrotalcite [239], which
also contributes to the mass loss between 350 and 400 °C. In the samples of GPC
after creep, it seems that there is a higher amount of this phases comparing to the
control samples. Besides, such drop at 170 °C can also be attributed to dehydration
of AFm and C-A-S-H phases [240]. The mass loss below 470 °C is consistent with
the release of the molecular water remaining in the pore structure. The mass loss in
the region 350 °C is related to the presence of hydrous calcium aluminate phases
[241, 242]. The progressive mass loss above 380- 480 °C is related to the continuing
dehydration of the C–S–H type gel, which also shows a sharp drop in the OPCC
samples. The dehydration of Margarite takes place at temperatures below 357 °C
[243], at which the GPC creep sample shows more loss of mass than the GPC control
sample. This is also different to the TGA from the 28 days core sample of GPC. This
suggests the larger amount of Margarite in the GPC creep sample than the GPC
control sample.
On the other hand, the peak locations of the creep and control samples of OPCC are
similar to the 28 days sample. This confirms the finding from XRD that there is no
major phase alternation in the OPCC specimens before and after creep test.
128
Figure 5.6: Differential thermograms for GPC and OPCC samples after creep test
(Creep) and the reference sample (Control)
5.3.3 BSEM
BSE imaging of creep sample and control sample of GPC and OPCC at different
magnification are shown in Figure 5.7 and Figure 5.8. There is no obvious difference
between the creep sample and control sample in both cases of GPC and OPCC
observed. However, compared with the 28 days samples of OPCC, the OPCC
samples after creep test shows less amount of unhydrated particle of cement. This
indicates further hydration in OPCC after long period of creep test. The amount of
unreacted particle of slag seems to only reduce less after the creep test. This would
be due to the reaction product of slag forming a barrier at the surface of the slag
particle, which limits further reaction [38].
129
GPC control GPC creep
Figure 5.7: Backscattered electron imaging of GPC after creep test
130
OPCC control OPCC creep
Figure 5.8: Backscattered electron imaging of OPCC after creep test
5.3.4 Pore characteristic
Figure 5.9 to Figure 5.12 show the pore size distribution of both GPC and OPCC
samples obtained from the core and the skin layer of creep and control specimens, in
comparison to the samples at 28 days before creep test. It can be seen in Figure 5.9
that the pore size distribution of GPC control specimen is similar to the sample
obtained at 28 days. This indicates only minor change in pore structure occurred
during the test period. However, a significant difference is observed between the
GPC creep specimen and control specimen. When subjected to sustained loads, a
great amount of pores above 2 nm disappeared in the GPC creep specimen, and more
pores below 2 nm were formed in comparison with the GPC control specimen. This
131
is different from the OPCC specimens, in which both creep and control specimens
show similar pore size distribution with pore refinement by further hydration [244]
(see Figure 5.10).
Figure 5.9: Pore size distribution of GPC samples from core of specimens
Figure 5.10: Pore size distribution of OPCC samples from core of specimens
132
Figure 5.11: Pore size distribution of GPC samples from skin of specimens
Figure 5.12: Pore size distribution of OPCC samples from skin of specimens
For the samples taken from the skin layer of the specimens, both GPC creep and
control specimens show similar pore size distribution. There is a shrift of pores to the
smaller sizes in comparison with the GPC specimen at 28 days as shown in Figure
5.11. This indicates that there was pore refinement within the test period, but the
sustained loads did not have effect on pore structure of GPC when the GPC was dry.
Pore refinement was also observed from the skin layer of OPCC specimens with
similar pore size distribution between creep and control specimens (see Figure 5.12).
133
5.4 Relationship between microstructure and creep
behaviour
The creep behaviour can be related to the microstructure in two aspects; the sliding
of gels and the microdiffusion within the pore network, based on the research on
OPCC [6, 8]. The original GPC sample in this case was found to contain more
amorphous phases. This agrees with the finding by White et al. [245], and she stated
that the GPC system is less thermodynamic stable than the OPC-based system. This
influences the phase dissolution and precipitation in long term, and thus influence
creep [246]. It is then hypothesised that the large creep effect in GPC would be due
to dissolution of large amount of amorphous phases. The slippage of these
amorphous phases then reprecipitated and formed more stable crystalline structure.
One example is the crystalline phase of Mg-rich Diopside identified only in the creep
specimen of GPC in this research.
From the initial test on GPC and OPCC, there are more fine pores presented in the
pore network of GPC then OPCC. This influences the microdiffusion and has an
effect on the short-term creep behaviours [8, 247]. Since the permeability is lower in
the finer pore structure [67], it slows down the seepage of pore solution under creep
loads. This explained the low creep rate observed in GPC in early loading period.
The larger slippage of gel phases in GPC would also influence the pore structure.
The phase dissolution and precipitation under sustained stress would result in a more
compact microstructure, since a great amount of pores above 2 nm had shifted to
pores below 2 nm in the GPC creep specimen. This suggests the less thermodynamic
stability in the GPC system, which allows more movement of phases when subjected
to sustained loads. Therefore, when the system contains more amorphous phases,
more creep effect would be anticipated.
134
Microstructural analysis was carried out to investigate the creep behaviour of GPC
and OPCC. Samples were collected from the concrete specimens before and after
creep test and prepared for XRD, TGA, BSEM and nitrogen adsorption test. Besides
the Quartz from the aggregates, XRD and TGA identified different phases between
GPC and OPCC at age of 28 days. The GPC contained Hydrotalcite, Gismondine and
C-A-S-H, while OPCC contained Portlandite, Ettringite and C-S-H. Besides, Calcite
also presented in both system, but the sources are different. Calcite in GPC was a
residue of the anhydrous slag, but it was also a carbonation product in OPCC, as
more Calcite was found in the skin layer of the OPCC specimen. Moreover, a broad
‘amorphous hump’ between 25° and 35° 2θ in the GPC sample indicates more
amorphous phases in GPC than in OPCC. After the long period of creep test, samples
were collected from the creep specimens and the control specimens of GPC and
OPCC. There was no phase alternation observed in OPCC, since the binder
composition seems to be equivalent to the samples before creep test. However, new
crystalline phases, Mg-rich Diopside, Margarite and Chabazite were identified only
in the creep GPC sample, while the control GPC sample showed little difference to
the GPC sample before creep test.
BSEM revealed that GPC had more amount of unreacted particle of slag in the
binder than the amount of unhydrated particle of cement in OPCC at 28 days. The
OPCC samples after creep test showed less amount of unhydrated particle of cement,
due to further hydration in OPCC after long period of creep test. However, there was
no obvious difference between the GPC samples before and after creep test. This
would be due to the reaction product of slag forming a barrier at the surface of the
slag particle for further reaction.
Pore size distribution was analysed from the results from nitrogen adsorption test.
For the core samples of GPC and OPCC at age of 28 days, GPC had less pores in the
size above 2 nm, but more pores in the size below 2 nm than OPCC. However, the
skin sample of GPC showed more pores within a range of 15 to 40 nm, due to
microcracks caused by drying. After the creep test, all samples generally showed a
pore refinement due to further geopolymerisation or hydration. However, a great
135
amount of pores above 2 nm disappeared in the GPC creep specimen, and more pores
below 2 nm were formed in comparison with the GPC control specimen.
The GPC system which contains more amorphous phases would be less
thermodynamic stable than the OPC-based system. The large creep effect in GPC
would be due to dissolution of large amount of amorphous phases subjected to
sustained loads, which caused more slippage of these amorphous phases then
reprecipitated and formed more stable crystalline structure, like the crystalline phase
of Mg-rich Diopside in GPC. This process would result in a more compact
microstructure. Therefore, the long-term creep of GPC is higher than that of OPCC
due to more amount of amorphous phases in GPC. In addition, the finer pore network
of GPC would slow down the microdiffusion of pore solution and cause low creep
rate, which influences the short-term creep behaviour of GPC.
136
Chapter 6 : Assessment of creep behaviour with
mini-beam test and microstructural study
6.1 Introduction
Creep testing for concrete specimens is usually carried out on concrete cylinders of
100 mm or 150 mm diameter in compression by using a creep loading rig specified
in the standards, such as ASTM C512 [248] and AS1012.16 [206]. Load adjustment
is required regularly to ensure constant sustained loads on the samples at all time
during long test period. On the other hand, tensile creep testing can only be done on
custom designed loading rigs [134, 135]. In such case, difficulty is experienced in
both the equipment setup and sample preparation. The tensile creep measured from
the direct tensile creep test can be affected by damage from microcracks and
additional shrinkage. This leads to a contracting tendency if the shrinkage from the
companion specimens is used. One of the ways to overcome this problem is to obtain
tensile creep from the flexural creep test, since it would give a better result as the
above effects are less [137]. Flexural creep tests have been carried by a number of
researchers for the investigation of creep of concrete [139, 140]. In addition,
Ranaivomanana et al. [138] suggested that the basic creep strains were the same in
compression and tension in the bending specimens under sealed condition.
The paste content in concrete and mortar, as the major binder materials, is mainly
responsible for effects of creep and shrinkage. This is due to the diffusion of water
and other molecules along pore networks in the gel microstructure of C-S-H [130]. A
study of paste samples would give the fundamental information on the long-term
behaviour of concrete and mortar. Therefore, a new type of flexural creep test has
been designed for paste samples of small size to accelerate the investigation of creep
of concrete. This flexural creep test was expanded from the test developed by
Scherer [249], which used the relaxation time to measure the elastic properties and
permeability of gels of porous materials. In his method, a rod of saturated porous
material is deformed to a certain deflection within the linear elastic range in the bath.
This induces extra pressures to the liquid in the rod. The liquid flows within the
137
network and between the network and the bath until the pressure equilibrated. When
the deflection is maintained, the applied force decayed with time. This involves two
processes: hydrodynamic relaxation owing to flow of liquid in the pores and
viscoelastic relaxation of the stress on the solid phase. The method can also be
applied on plate of cement paste [250-252]. Therefore, the new test, called mini-
beam test, was developed based on the concept in the reversed way of Scherer’s
method, which predicts the creep behaviour from the pore characteristic obtained
from nitrogen adsorption test.
In this chapter, the details of the mini-beam test and the specimen details of
geopolymer and OPC are presented in section 6.2. The results of the mini-beam test
from a range of experimental conditions are reported and discussed in section 6.3.
The effect of age of loading, effect of drying and effect of temperature on the creep
behaviour have been investigated.
Small samples from the mini-beams were undergone nitrogen gas adsorption test and
the results is presented in section 6.4. The change of pore characteristic before and
after mini-beam test was observed.
A creep model has been developed and presented in section 6.5, to describe the creep
behaviour of the paste specimens with the input of microstructural analysis. This
model was then validated with the experimental results.
6.2 Experimental program
6.2.1 Test details
Flexural creep test was done on thin plate specimens, or so called “mini-beam”, to
observe the flexural creep behaviour. The mini-beams were made of geopolymer or
OPC pastes and mortars, which have a dimension of 50 mm wide × 10 mm deep ×
350 mm long. The shape aspect ratio of width to depth is set as 1:5, followed the
suggestion by Valenza II and Scherer [252], in which the flow in horizontal direction
138
would be minor and thus it was effectively considered “infinitely wide”. The mini-
beams were cast in the custom design moulds as shown in Figure 6.1.
Figure 6.1: Mini-beam specimens in custom designed moulds
The mini-beam sample was supported over a span of 300 mm for the flexural creep
test. Loads were applied in the middle 100 mm apart as four-point-bending. Because
of the thin property of this mini-beam, it only requires small test area and light
sustained loads. An illustration of the test setup is shown in Figure 6.2. A dial gauge
with accuracy to 1 µm was used to measure the long-term mid-span deflection. For
each batch of test specimens, a minimum of six mini-beam samples were made
including two creep test samples, two companion samples and two samples for
destructive testing. The companion samples were made mainly for shrinkage effects
without load applied. Measurement of shrinkage was done by using a demec gauge
of 200 mm gauge length. Since the samples under creep test were subjected to drying
at all surfaces, the companion samples were stored with the short side standing to
allow drying at most surfaces and bending under self weight. Because it is important
139
to ensure the creep samples remained uncrack under load, the four-point-bending test
was done on the last two samples from the same batch to determine the failure load.
The sustained load on the creep samples was applied at less than 40% of the failure
load.
Figure 6.2: Setup of mini-beam test
Mini-beam test is designed to allow more investigation on the creep behaviour of

geopolymer and OPC pastes and mortars under various environmental condition.
Primary tests were done on paste specimens. A number of mini-beam tests were done
under a number of environments. The specimens are identified with the following
format:
Type of binder – Age of loading – Temperature– Relative humidity/Sealed
Types of binder include geopolymer paste (GPP), OPC paste (OPCP), geopolymer
mortar (GPM) and OPC mortar (OPCM). For example, GPP-1d-23-50 stands for
geopolymer paste loaded at 1 day at 23 °C and 50% RH. The sealed specimen can be
prepared by wrapping the specimen with two layers of plastic wrap. In total, eight
different test conditions were performed as shown in Table 6.1. A picture of the test
is shown in Figure 6.3.
140
Table 6.1: List of specimens in mini-beam test
GPP OPCP GPM OPCM

GPP-1d-23-50 OPCP -1d-23-50 GPM-1d-23-50 OPCM -1d-23-50
GPP-28d-23-50 OPCP -28d-23-50 GPM-28d-23-50 OPCM -28d-23-50
GPP-28d-23-70 OPCP -28d-23-70 - -
GPP-28d-23-S OPCP -28d-23-S GPM-28d-23-S OPCM -28d-23-S
GPP-28d-50-50 OPCP -28d-50-50 - -
GPP-28d-50-S OPCP -28d-50-S - -
GPP-28d-0-30 OPCP -28d-0-30 - -
GPP-28d-0-S OPCP -28d-0-S - -
Figure 6.3: Mini-beam test in the environmental chamber
141
6.2.2 Assessment of creep coefficient from mini-beam test
The creep of the specimen can be calculated from the deflection from mid-span.
Based on the uncrack condition for the whole length of the mini-beam, the deflection
at mid-span is:
𝑊⁄ ×𝐿⁄ 4 23𝑊𝐿3
∆(𝑡) = − 24𝐸2 (𝑡)𝐼3 (3𝐿2 − 9 𝐿2 ) = − 1296𝐸 (6.1)
𝑒 𝑒 (𝑡)𝐼
𝜎𝐼 𝜀(𝑡)𝐸𝑒 (𝑡)𝐼 𝑊𝐿
and 𝑀= = = (6.2)
𝑦 𝑦 6
Therefore,
23𝐿2
∆(𝑡) = − 216𝑦 𝜀(𝑡) (6.3)
23𝐿2
When t = 0, ∆(0) = − 216𝑦 𝜀(0)
∆(𝑡) 𝜀(𝑡)
Hence, = 𝜀(0) = 1 + 𝜑(𝑡) (6.4)
Δ(0)
where Δ(t) and Δ(0) are the mid-span deflection at time t and immediately after
loading, respectively, W is the total applied load in kN, L is the span length, Ee(t) is
the age-adjusted effective modulus, I is the moment of inertia, M is the maximum
bending moment, σ is the maximum compressive stress on the top surface, y is the
distance from the surface to the centroid of the cross section, ε(t) and ε(0) are the
strain at time t and immediately after loading, and φ(t) is the creep coefficient.
Because the mini-beam sample is thin, symmetric and not reinforced, the shrinkage
strain is assumed to be uniform without gradient. Therefore, it is assumed that the
effect of shrinkage does not affect the overall deflection of the specimen. The total
shrinkage strain and the autogenous shrinkage strain are measured from the
specimens subjected to drying and those in sealed condition, respectively. The drying
shrinkage strain is calculated as the difference between the autogenous shrinkage
strain and the total shrinkage strain.
142
6.3 Experimental results of mini-beam test
6.3.1 Compressive strength
Cubic specimens of 50 mm were cast together with the mini-beams for compressive
strength test according to ASTM C109/C109M [210]. This test was performed on the
same day when the mini-beams from the same batch were loaded. The test results are
shown in Figure 6.4.
Figure 6.4: Compressive strength of geopolymer and OPC pastes and mortars
Right after demoulded at 1 day, the compressive strength of GPP was higher than
that of OPCP, and the compressive strength of GPM was also higher than that of
OPCM. This indicates high early strength of geopolymer. After that, all specimens
were cured in the bath until 28 days. The compressive strength of both geopolymer
and OPC specimens was comparable at 28 days. Therefore, the further growth in
strength was smaller in the geopolymer specimens compared with the OPC
specimens, which compressive strength increased approximately by double
143
compared with the 1-day compressive strength. This was likely due to the fast
reaction at early age involved by the activator in slag-based geopolymer, resulting in
high early compressive strength [71].
After removed from the bath, the cubic specimens were stored in the same condition
as the test environment of mini-beams until 90 days. Compressive strength test was
then carried on these specimens. GPP continued the growth in strength at 23 °C and
50% RH by 20%. When the humidity was increased to 70%, the compressive
strength of GPP increased to 90.0 MPa. Further growth of the compressive strength
to 98.1 MPa was observed from the GPP under sealed condition. This indicates the
effect of drying and the importance of curing to strength development in geopolymer.
However, the growth in compressive strength of OPCP was not observed at 50% RH,
instead, it reduced by 20%. Higher compressive strength of OPCP of 86.0 MPa was
recorded at 70% RH, which was higher than the 28-day compressive strength by
25%. Therefore, the result of compressive strength of OPCP at 50% RH does not
follow the increasing trend, and thus it would be considered as an outliner. The
growth in compressive strength of OPCP under sealed condition was lower than that
of OPCP at 70% RH by 9%. The compressive strength of sealed OPCP was also
lower than that of GPP by 20%. For mortar specimens, tests were only carried out
on specimens under 50% RH and sealed condition. The strength development in
GPM is similar to that in GPP. The maximum compressive strength of GPM is found
in sealed specimens at 89.24 MPa. The growth in compressive strength was also
identified in OPCM, but it showed higher result for OPCM in 50% RH than OPCP in
50% RH. The sealed OPCM also had a lower compressive strength than OPCM in
50% RH by 5%.
The compressive strength of these paste specimens was tested for the thermal effect.
It was found that both GPP and OPCP stored at 50 °C gained more strength than
those stored at 23 °C. The highest compressive strength was recorded from GPP
under sealed condition, with a value of 99.9 MPa. This was higher than 84.8 MPa on
OPCP under same condition. This agrees with the results on the effect of heat curing
on strength of geopolymer by Altan and Erdoğan [253]. They found that strength
development was higher in the humidity condition than the drying condition for
GPM, since moisture retention was required for hydration. This explained the higher
144
compressive in GPP under sealed condition in this case. On the other hand, when the
GPP was stored at 0 °C, only minor growth of compressive strength was observed.
OPCP also showed lower compressive strength when stored at 0 °C. It seems that the
strength of GPP specimens stored at 0 °C after bath curing was retarded at low
temperature. This can be seen that low temperature not only caused retardation of
strength development in early age [71], it also affected the strength of mature
geopolymer. Another finding in this study was the reduction in strength in sealed
specimens compared to the drying specimens in both GPP and OPCP. It would be
due to the 8% increase in volumetric strain of moisture contents during
transformation of water to ice at freezing temperature, causing cracks [254]. Since
there was more moisture content in the sealed specimens than the drying specimens,
the expansion would cause more microcracks within the sealed specimens, leading to
a reduction in compressive strength.
6.3.2 Flexural capacity
Before the mini-beam test was carried out, the flexural capacity of the mini-beams
was determined through a four-point bending test, which had the same load
configuration as the mini-beam test. Since the mini-beams had a small variation
between their dimensions, tensile stress at bottom surface at failure was calculated
and compared instead of the failure load. This can be referred as the flexural capacity
of the mini-beams and the calculation is simply by the equation below:
𝑀𝑦 𝑊𝐿𝑦
𝜎= = (6.5)
𝐼 6𝐼
The results were calculated and shown in Figure 6.5. It was found that the GPP had a
lower flexural capacity than OPCP at 1 day. Although the flexural capacity of GPP
had increased after bath curing, it was still slightly lower than that of OPCP at 28
days. Furthermore, the similar trend was also observed on mortar specimens. It was
also found that there was a reduction in flexural capacity of OPCP from 1 day to 28
days. This was different to the compressive strength development at early age.
145
Figure 6.5: Maximum tensile stress at failure for geopolymer and OPC pastes and
mortars
The flexural capacity of both GPP and OPCP increased at 90 days after creep test at
23 °C and 50% RH, but the OPCP showed higher flexural capacity than GPP by
15%. However, this result is opposite to the results of sealed paste specimens at 23
°C. On the other hand, the growth in flexural strength was significant without drying,
especially in GPP, which showed an increase of 140%. This would be due to the
surface microcracking occurred when subjected to drying conditions [70], which
reduced the tensile capacity of the geopolymer samples and thus resulted in low
flexural strength. There was an exception for the paste specimens tested at 23 °C and
70% RH. Both GPP and OPCP showed lower flexural capacity than their
counterparts did at 50% RH. One of the possible causes would be that these
specimens were weakened by carbonation. It was reported by Chen and Ho [255]
that carbonation occurred more rapidly at a constant 70% RH than at lower or higher
RH in OPCC. Lower RH like 50% RH reduced the amount of water to form water
films for CO2 diffusion path in the microstructure. Similar result was found on slag-
based GPC, which showed slightly higher carbonation at 65% RH [78]. Since the
mini-beam is thin, carbonation from all exposed surfaces would have more impact of
strength on mini-beams than on cubic specimens for compressive strength test.
146
It was found that the flexural capacity of mortar specimens was generally higher than
that of the paste specimens. This can be explained that the bond strength of sand and
paste had more influence on the strength of mortar than the inherent tensile strength
of paste. Since the sand particle had higher tensile strength than paste, the failure
occurred at the surface around it. The total surface area under tension in mortar
specimen was larger than the projected cross-sectional area under tension in paste
specimen and thus the total bond strength was higher than the direct tensile strength
[256].
Paste specimens stored at 50 °C generally present a higher flexural capacity than

those stored at 23 °C. The sealed condition together with heat environment seems to
provide a better curing condition to the specimens and results in a higher flexural
capacity. On the other hand, storing the specimens at 0 °C caused retardation of the
flexural capacity. These results are similar to the thermal effect on compressive
strength.
6.3.3 Creep response of mini-beams of paste specimens
The creep coefficient of the mini-beams tested at 23 °C up to 56 days of loading

period was computed as shown in Figure 6.6. Test result for GPP-1d-23-50 could not
be obtained throughout the test period due to cracking. The mini-beam was loaded
less than 40% of the flexural capacity, so the crack would be caused by high drying
shrinkage at early age. The last measurement of GPP-1d-23-50 was at 6 hours after
loaded, and the calculated creep coefficient was 8.98. GPP-28d-23-50, which was
tested at age of 28 days, showed no crack during the test. It is believed that the bath
curing allows further strength development (e.g. higher flexural tensile strength) and
minimises the amount of shrinkage [47], so it can resist cracking. The creep
coefficients at the end of the test for GPP-28-23-50, GPP-28-23-70 and GPP-28-23-S
are 8.24, 6.75 and 1.33, respectively. This indicates that the creep of GPP reduced
with increasing humidity in the environment, when drying creep is minimised. When
drying is eliminated, GPP shows much lower creep effect in the sealed specimen,
compared with the drying specimens.
147
For the mini-beam test on OPCP, OPCP-1d-23-50 did not fail like GPC-1d-23-50
and its creep coefficient at the end of the test is 8.77. The creep coefficients at the
end of the test for OPCP-28-23-50, OPCP-28-23-70 and OPCP-28-23-S are 8.91, 7.4
and 3.47, respectively. This also follows the general creep behaviour of OPCC, in
which the drying creep increases with decreasing RH [105].
(a)
(b)
Figure 6.6: Creep coefficient of paste mini-beams: (a) GPP; (b) OPCP
148
The calculated creep coefficients of GPC mini-beams are generally lower than those
of OPCP mini-beams. However, it is noticed that there is an unusual reduction of
creep coefficient in GPP-28d-23-50 and GPP-28d-23-70 after the first day of
loading. This was caused by an upward deflection in the mini-beam, countering the
downward deflection from creep effect. This upward defection would be due to
differential shrinkage across the depth of the mini-beam. The drying shrinkage at the
bottom side of the mini-beam would be higher than that at the top side. The
additional drying shrinkage at the bottom side of the mini-beam would be caused by
microcracking under tensile stress [135], which allows drying proceeding further into
the specimen. Such effect may have been taken place in OPCP as well, but in a
smaller scale than the creep effect. Therefore, it was not discovered from the test on
OPCP. Since the pore structure of GPP is finer than that of OPCP, the effect of
drying shrinkage would be greater in GPP due to greater capillary tensile forces [60].
In order to minimise the effect of differential shrinkage on the analysis of creep

behaviour between GPP and OPCP, results of GPP and OPCP mini-beams in first
day of loading are compared as shown in Figure 6.7. It can be seen that the GPP
generally showed higher creep than OPCP under drying conditions. The largest
difference is found between GPP-1d-23-50 and OPCP-1d-23-50; the creep
coefficient of GPP-1d-23-50 is approximately 8 times larger than that of OPCP-1d-
23-50 at 6 hours after loaded. However, the basic creep coefficient from GPP-28d-
23-S is lower than that of OPCP-28d-23-S. Therefore, this study shows that the creep
coefficient of both GPP and OPCP increases under drying condition, but drying has a
larger impact on creep of GPP than OPCP. This result agrees with the previous
finding on the creep behaviour of GPC and OPCC in chapter 4.
149
Figure 6.7: Creep coefficient of GPP and OPCP mini-beams at early stage of loading
6.3.4 Creep response of mini-beams of mortar specimens
Figure 6.8 presents the test results for the mortar mini-beams. Unlike GPP-1d-23-50,
GPM-1d-23-50 did not crack during the test period because of the restraint by the
aggregates, which reduced the shrinkage in the specimen. Another reason would be
the higher flexural capacity of mortar mini-beam than paste mini-beam. Therefore,
GPM-1d-23-50 did not fail during mini-beam test, and it yielded a high creep
coefficient of 42.05 at the end of test. When GPM was loaded at later age of 28 days
under same condition, the creep coefficient reduced to 20.33 at the end of test. This
indicates that the effect of age of loading reduces the creep effect of geopolymer.
However, this value is significantly higher than the basic creep coefficient obtained
from GPM-28d-23-S, 3.61, at the end of test. The overall creep behaviour of GPM
agrees with the previous results of GPP. It is found that the GPM mini-beams are less
affected by the differential shrinkage observed from the GPP mini-beams. Upward
displacement of the mortar mini-beams was not noticeable. This would be due to the
effect of reduced shrinkage by the aggregates, which also minimises the differential
shrinkage.
150
(a)
(b)
Figure 6.8: Creep coefficient of mortar mini-beams: (a) GPM; (b) OPCM
For OPCM, the effect of age of loading was not strong as expected. OPCM-28d-23-
50 shows less creep than OPCM-1d-23-50 by 10%. OPCM-28d-23-S shows less
creep than OPCM-28d-23-50 by 20%, since drying creep was minimised similar to
GPM.
151
In the comparison between the GPM and OPCM mini-beams, it can be seen that
GPM exhibits higher creep than OPCM when subjected to drying condition, but less
creep under sealed condition. This generally agrees with the previous results on the
paste mini-beams, which suggests that drying has a large influence on creep of
geopolymer.
6.3.5 Thermal effect on mini-beam test
Paste mini-beams were also tested at 50 °C and 0 °C for the thermal effect on creep
behaviour. All of these tests were started at age of 28 days. The results are shown in
Figure 6.9. It can be seen that the creep coefficient of GPP-28d-50-50 is higher than
that of GPP-28d-23-50, but the creep coefficients of GPP-28d-50-S and GPP-28d-23-
S are similar. This indicates that the heat environment accelerates the evaporation of
water from the mini-beam to the environment, since the viscosity of water decreases
with increasing temperature [126]. Another cause would be the change of
microstructure of GPP under high temperature [257]. Although the test temperature
was below 100 °C, it would influence the viscoelastic behaviour in a long test period.
On the other hand, the creep coefficient of GPP-28d-0-30 is lower than that of GPP-
28d-23-50. This would be due to frozen of the water in the pore network, which
reduces the free water migration. One of the observations from these results is the
negative creep development in GPP-28d-0-S. Since this specimen was sealed, water
was not able to move to the environment. Such trapped water would be frozen in this
temperature and caused negative effect on strength as shown previous. The
expansion by frozen ice would cause uneven strain across the depth of mini-beam,
which would lead to upward or additional downward deflection of the mini-beam.
Similar to GPP, there was no significant difference between the creep coefficients of
OPCP-28d-50-S and OPCP-28d-23-S. OPCP-28d-50-50 exhibits less creep than that
OPCP-28d-23-50. The difference is not significant, but such behaviour is different to
GPP. It would be because the external heat accelerates the hydration of OPCP and
improves the microstructure [258]. As a result, the stiffness and strength of OPCP are
improved. Although the strength of GPP increases in high temperature due to further
152
geopolymerisation, such reaction also causes more transitional thermal creep [165].
Therefore, it indicates that creep of GPP is influenced more by high temperature than
OPCP. In other words, creep of geopolymer seems to be dominated more by
temperature dependent flow than by moisture diffusion under stress.
(a)
(b)
Figure 6.9: Creep coefficient of paste mini-beams under various temperatures: (a)
GPP; (b) OPCP
153
OPCP-28d-0-30 exhibits similar creep behaviour to OPCP-28d-23-50 at later ages,
but the creep coefficient of OPCP-28d-0-30 is less in the early stage of loading. This
indicates that the low temperature also influences the creep behaviour of OPCP.
Although the movement of water is slow at low temperature, the rate of drying of
OPCP-28d-0-30 can be faster than that of GPP-28d-0-30, due to the larger pore
network in OPCP. When OPCP mini-beam is sealed, it also shows negative creep
development like GPP-28d-0-S. This indicates that the GPP and OPCP experienced
similar effect from frozen water in this condition.
6.3.6 Shrinkage obtained from mini-beam test
The shrinkage strain of the mini-beams was measured from the companion
specimens without loading, relative to the measurement from the beginning of mini-
beam test. The test results for paste mini-beams are shown in Figure 6.10 and Figure
6.11.
The shrinkage strain of GPP-1d-23-50 reached 3900 microstrains in first day. The
measurement on the second day could not be obtained, since it was more than the
measurement range of the demec gauge. The shrinkage strain of GPP is significantly
reduced after bath curing; the shrinkage strain of GPP-28d-23-50 at first day is 980
microstrains. The high shrinkage strain of GPP-1d-23-50 would be contributed by
the autogenous shrinkage caused by geopolymerisation in the early age. The
shrinkage strain values of GPP-28d-23-50, GPP-28d-23-70, and GPP-28d-23-S at the
end of the test are 7400, 2890 and 950 microstrains, respectively. This indicates that
drying has a significant influence on shrinkage of GPP. Increasing the RH from 50%
to 70% decreases the shrinkage strain by more than half. The shrinkage strain is
further reduced without drying for the sealed specimen. The low shrinkage in the
sealed specimen would be due to the fact that most autogenous shrinkage had
occurred prior to 28 days.
154
(a)
(b)
Figure 6.10: Shrinkage strain of GPP mini-beams: (a) Effects of age of loading and
drying; (b) Thermal effect
155
(a)
(b)
Figure 6.11: Shrinkage strain of OPCP mini-beams: (a) Effects of age of loading and
drying; (b) Thermal effect
Comparison of GPP mini-beams under various temperature environments shows the

highest shrinkage strain observed in GPP-28d-23-50. Although the shrinkage strain
of GPP-28d-50-50 increased rapidly in the early stage, an expansion occurred after 4
156
days of heating period. Such expansion would be due to microcracking caused by the
effects of drying and heat. However, expansion did not happen on GPP-28d-50-S,
and it shrank more than GPP-28d-23-S. The difference in shrinkage of these sealed
specimens can be explained by accelerated geopolymerisation by heat curing [38].
When the GPP was stored at 0 °C, the shrinkage strain was less than that of GPP
stored at 23 °C. It would be because the free evaporable water in the pores was
frozen, which limited the loss of water and thus shrinkage.
Generally, OPCP shows similar effects of age and drying to GPP, but the magnitudes
of shrinkage strain of OPCP are significantly smaller. The shrinkage strain of OPCP-
28d-23-50 at the end of test is 2840 microstrains, which is only 38% of that of GPP-
28d-23-50. Comparing the difference between the drying and sealed specimens of
both GPP and OPCP shows that drying has more influence on GPP than on OPCP. It
can be explained by the finer pore network of GPP than OPCP, so the capillary
tensile force is higher at the meniscus and this causes higher drying shrinkage [60].
This also explains the high shrinkage observed in GPC due to high shrinkage in GPP.
For the thermal effect, OPCP-28d-50-50 shrank less than OPCP-28d-23-50 like
observation in GPP. This would be due to the same cause by microcracking. The
shrinkage strain of OPCP-28d-0-30 was 1410 microstrains at the end of test, and
higher than that of OPCP-28d-50-50 due to its expansion. Besides, all sealed
specimens show similar shrinkage strain in low range below 500 microstrains.
The shrinkage strain of both GPM and OPCM are shown in Figure 6.12. Unlike
GPP-1d-23-50, GPM-1d-23-50 did not crack in the whole test period, and its
shrinkage strain is 4950 microstrains at the end of the test. This would be due to the
restraint provided by the aggregates, which reduces the amount of shrinkage, and the
high creep behaviour, which relaxes the tensile stress [47]. Therefore, comparison
between geopolymer and OPC specimens at early age can be achieved. The
shrinkage strain of OPCM-1d-23-50 is approximately 23% of that of GPM-1d-23-50
at the end of the test. The shrinkage strain of OPCM-28d-23-50 is also less than that
of GPM-28d-23-50 at the end of the test. However, after bath curing, the difference
of shrinkage between GPM and OPCM was reduced. The largest reduction of
shrinkage strain is observed on GPM-28d-23-50, which decreases from 4950 to 2750
157
microstrains. It is found that the amounts of autogenous shrinkage of GPM and
OPCM are 35% and 26% of their total shrinkage values, respectively. However, the
autogenous shrinkage of GPM is still high after 28 days of bath curing. For
autogenous shrinkage at early age, further test is required.
Figure 6.12: Shrinkage strain of GPM and OPCM mini-beams
6.3.7 Autogenous shrinkage
Standard autogenous shrinkage test was done on paste and mortar specimens of
geopolymer and OPC for autogenous at early age according to ASTM C1698 [212].
The shrinkage measurement starts as soon as the final setting of the specimens
occurs. The times of final setting of GPP and OPCP are 327 and 317 minutes after
casting by Vicat needle test according to ASTM C191 [213], respectively.
The test result of autogenous shrinkage of GPP, OPCP, GPM and OPCM is shown in
Figure 6.13. It can be seen that the autogenous shrinkage strain of the geopolymer
specimens is significantly higher than that of the OPC specimens. The values of
autogenous shrinkage strain of GPP, OPCP, GPM and OPCM at 56 days are 7030,
910, 2670 and 390 microstrains, respectively. Most of the autogenous strain
158
happened within the age of first 14 days. The development of autogenous shrinkage
strain in these specimens in later ages also agrees with the previous observation from
the mini-beam test, where measurement was started at age of 28 days. The effect of
aggregate is significant in control of autogenous shrinkage in this case. The
autogenous shrinkage strain is reduced by approximately 60% in both mortar
specimens in comparison with their paste counterparts.
Figure 6.13: Test results of autogenous shrinkage of GPP, OPCP, GPM and OPCM
6.3.8 Relationship between the paste and mortar mini-beams
Comparison of paste and mortar mini-beams in both GPP and OPCP shows that the
creep and shrinkage of the paste mini-beams are generally higher than those of the
mortar mini-beams. This is because the portion of aggregates in the mortar
specimens does not exhibit creep effect, and it restrains the shrinkage effect.
Generally, more creep is expected in the binder-aggregates composite when there is
more paste proportion in it. The long-term behaviour of mortar or concrete can be
159
predicted based on the behaviour of paste. Hobbs [259] proposed some simple
equations on prediction of properties of composite upon aggregate volume
concentration,
𝜀𝑐𝑚(𝑡) 1−𝐷
= 1+𝐷𝑎 (6.6)
𝜀𝑐𝑝 (𝑡) 𝑎
𝜀𝑠𝑝 (𝑡)𝐷𝑎 ∙2𝑘𝑎

𝜀𝑠𝑚 (𝑡) = 𝜀𝑠𝑝 (𝑡) − 𝑘 (6.7)
𝑝 +𝑘𝑎 +𝐷𝑎 (𝑘𝑎 −𝑘𝑝 )
where εcm and εcp are the creep strain of mortar and paste, Da is the aggregate volume
concentration, εsm and εsp are the shrinkage strain of mortar and paste, and Ka and Kp
are the bulk modulus of aggregates and paste, respectively. Equation 6.6 develops a
linear relationship between εcm and εcp. Hobbs [259] also proposed the relationship of
elastic modulus between mortar and paste as below:
1−2𝜐 1−𝐷
𝐸𝑝 = (1−2𝜐 𝑝 ) (1+𝐷𝑎 ) 𝐸𝑚 (6.8)
𝑚 𝑎
where Em and Ep are the elastic modulus of mortar and paste, and υm and υp are the
Poisson’s ratio of mortar and paste, respectively. Usually υm and υp were close, so
that it can be simplified to,
𝐸𝑝 1−𝐷
= 1+𝐷𝑎
𝐸𝑚 𝑎
Hence, under same stress,
𝜀𝑚 𝐸 1−𝐷
= 𝐸 𝑝 = 1+𝐷𝑎 (6.9)
𝜀𝑝 𝑚 𝑎
where εm and εp are the elastic strain of mortar and paste, respectively. Therefore, the
relationship of the elastic strain between mortar and paste has the same factor as that
for creep strain. In other words, both mortar and paste would share the same creep
coefficient profile according to equation 6.6. This does not agree with the
observation from the laboratory tests, so this equation would not be applicable.
Another function was proposed by Bažant [260] based on the combined series-
parallel model for creep interaction of aggregates and mortar matrix. This was
originally developed for concrete but it can be applied to the creep prediction of
160
mortar, based on the creep of paste [120]. This model was applied with the age-
adjusted effective modulus method.
𝛽 𝐸”𝑝
𝐽𝑚 (𝑡, 𝑡0 ) = 𝛼𝐸 [1 + (1 − 𝛼 ) 𝜑𝑝 (𝑡, 𝑡0 )] + (1 − 𝛽 )𝐽𝑝 (𝑡, 𝑡0 )
𝑎𝑔𝑔 +(1−𝛼)𝐸𝑝 (𝑡0 ) 𝐸”𝑚
(6.10)
𝜑𝑚 (𝑡, 𝑡0 ) = 𝐸𝑚 (𝑡0 )𝐽𝑚 (𝑡, 𝑡0 ) − 1 (6.11)
where Jm and Jp are the creep compliance function of mortar and paste, α and β are
the section parameter and the length parameter (αβ = Da), Eagg is the elastic modulus
of aggregates, E˝p and E˝m are the age-adjusted effective modulus of paste and
mortar, and φp and φm are the creep coefficient of paste and mortar, respectively.
Therefore, equations 6.10 and 6.11 were used on the results of both geopolymer and
OPC mini-beams. With the Ea = 72 GPa for the aggregates, Da = 0.53, and an
assumption of Poisson’s ratio, ν = 0.2 for both paste and aggregates, the creep
coefficient of mortar mini-beams was predicted based on the results from the paste
mini-beams as shown in Figure 6.14, while the prediction for shrinkage strain of
mortar mini-beams is shown in Figure 6.15.
The prediction method by Bažant [260] only shows the good match for OPCM-1d-
23-50. Although it does not provide a good prediction in the other cases, it generally
shows better prediction on OPCM mini-beams. For GPM-1d-23-50, since these is not
enough test data, the effectiveness of the prediction method can not be assessed.
GPP-28d-23-50 was affected more significantly by drying than GPM-28d-23-50, so
the predicted results based on GPP-28d-23-50 is not accurate. Furthermore, the
method cannot predict the creep coefficient of the sealed mortar mini-beams, because
their creep curves are higher than those from the sealed paste mini-beams. This does
not match the expectation of creep trend from paste to mortar. It is noticed that the
surface of the mortar mini-beams is rougher than that of the paste mini-beams, so it
may confine more air under the seal. This would cause some degree of drying creep
associated with the test results of GPP-28d-23-S and OPCP-28d-23-S. On the other
hand, prediction on shrinkage strain by Hobbs [259] generally gives good prediction
on the drying mini-beams of both GPM and OPCM. However, it underestimated the
shrinkage strain of the mortar mini-beams under sealed condition.
161
(a) (b)
(c) (d)
(e) (f)
Figure 6.14: Predicted creep coefficient of mortar mini-beams based on test results of
OPCM-28d-23-50, (e) GPM-28d-23-S, (f) OPCM-28d-23-S
162
(a) (b)
(c) (d)
(e) (f)
Figure 6.15: Predicted Shrinkage strain of mortar mini-beams based on test results of
OPCM-28d-23-50, (e) GPM-28d-23-S, (f) OPCM-28d-23-S
163
6.4 Analysis of pore structure of paste mini-beams
6.4.1 Pore structure before mini-beam test
The pore characteristic of the paste specimens was obtained by using nitrogen gas
adsorption test. In order to assess the pore structure before mini-beam test, samples
were collected from the end of mini-beams after the strength test. These samples
were then saw cut into pieces with thickness of approximately 1 mm as shown in
Figure 6.16. Test was conducted on approximately 0.5 g of samples after the
pretreatment specified in Chapter 3. The average of results from two tests is reported.
Figure 6.16: Samples prepared for nitrogen adsorption test
The specific surface area of both GPP and OPCP was derived according to BET
equation [109]. Table 6.2 shows that both GPP and OPCP have similar specific
surface area at age of 1 day. The specific surface area of GPP reduces at age of 28
days, but that of OPCP increases instead. This suggested that GPP would have less
porosity than OPCP at 28 days. Further analysis was then carried out to analyse the
pore size distribution.
164
Table 6.2: Specific surface area of GPP and OPCP
Specimens Specific surface area, S

(m2/g)
GPP-1d 51.86
OPCP-1d 50.48
GPP-28d 30.98
OPCP-28d 73.01
DH method, BJH method and CI method were used to analysis the pore sizes in
moespore range. A comparison of these three methods on GPP-1d is shown in Figure
6.17. The results are very close between these methods. Therefore, only results
produced by DH method are presented for the following analysis. Besides DH
method, MP method was also used to analysis the pore sizes in micropore range (< 2
nm). The pore size distributions of GPP and OPCP at 1 day and 28 days are shown in
Figure 6.18.
Figure 6.17: Pore size distribution on GPP-1d by different analysis methods
165
(a)
(b)
Figure 6.18: Pore size distribution of GPP and OPCP: (a) DH method; (b) MP
method
166
The total cumulative pore volume at the smallest pore size measured from DH
method is reported as the indication of porosity. It can be seen that the porosity of
GPP is less than that of OPCP at both 1 day and 28 days. Besides, the overall pore
volume of GPP decreases from age of 1 day to 28 days in mesopore range, but there
is an increase in pore size below 1.5 nm. There would be the result by
geopolymerisation in refining the microstructure. For OPCP, similar trend is
observed, but more pores are formed below 2 nm at age of 28 days. It would be due
to more hydration over 28 days, since pore structure is refined with reduction of
porosity under hydration [261]. These results explains the higher specific surface
area calculated previously due to more small pores formed. Since the porosity of
GPP reduces significantly from 1 day to 28 days, the specific surface area decreases
as well. However, the porosity of OPCP does not reduce significantly and more
micropores are formed, so the specific surface area increases.
6.4.2 Change of pore structure after mini-beam test
The pore characteristic of GPP and OPCP after mini-beam test was investigated.
Samples were obtained from the mini-beams after the long-term tests and prepared
for nitrogen gas adsorption test. The results of specific surface area of these
specimens are shown in Table 6.3. The pore size distributions are shown in Figure
6.19 and Figure 6.20.
Table 6.3: Specific surface area of GPP and OPCP after mini-beam test
Geopolymer Specific surface OPC specimens Specific surface

specimens area, S (m2/g) area, S (m2/g)
GPP28d 30.98 OPCP28d 73.01
GPP-28d-23-50 18.56 OPCP-28d-23-50 22.72
GPP-28d-23-S 13.43 OPCP-28d-23-S 85.20
GPP-28d-23-70 30.56 OPCP-28d-23-70 72.89
GPP-28d-50-50 6.79 OPCP-28d-50-50 30.18
GPP-28d-50-S 3.74 OPCP-28d-50-S 42.02
GPP-28d-0-30 29.45 OPCP-28d-0-30 25.05
GPP-28d-0-S 28.78 OPCP-28d-0-S 29.23
167
Figure 6.19a shows that the porosity of the drying GPP specimens generally
increases compared to GPP-28d. However, this increase is not uniformly distributed
across the range of pore sizes. GPP-28d-50-50 gains porosity in pore sizes between
10 nm and 30 nm, whilst the GPP-28d-23-70, GPP-28d-23-50 and GPP-28d-0-30
gains in pore sizes between 3 nm and 30 nm. This increase in overall porosity can be
due to microcracking caused by drying [262]. It is also suggested that the
temperature has an effect on the pore structure, as shown in different pores increase
pattern in these drying specimens. The increase of porosity below 2 nm was minor in
GPP-28d-50-50. Further result from Figure 6.19b shows the negative porosity below
2 nm. This would be due to the low porosity in such range of pore sizes. On the other
hand, porosity in sealed geopolymer specimens except GPP-28d-0-S generally
decreases compared to GPP-28d. This reduction in porosity would be due to further
geopolymerisation. GPP-28d-0-S remains similar pore distribution to GPP-28d, so it
suggested that the geopolymerisation would be stopped, as microstructure did not
change much at 0 °C. The specific surface area of GPP-28d-0-S is also close to that
of GPP-28d.
Among these GPP specimens, GPP-28d-50-S has the lowest porosity and lowest
specific surface area. Therefore, it indicates that the heat curing during mini-beam
test provided after 28 days of age had a significant influence on the microstructure of
geopolymer. Rovnaník, Bayer and Rovnaníková [263] found the phases
transformation from amorphous to crystalline phases in a slag-based GPP after
heating above 710 °C for one hour. Although the temperature in this mini-beam test
was 50 °C, the long heating period of two months would also alternate the
microstructure. An observation in the cross sections of GPP-28d-23-S and GPP-28d-
50-S showed a smoother and darker fracture surface in GPP-28d-50-S (see Figure
6.21). This would be due to the crystalline phases in GPP-28d-50-S. In order to
assess the phases, XRD assessment was carried out on the samples of GPP after
tested at 23 °C and 50 °C as shown in Figure 6.22. Comparison between these
samples shows that GPP-28d-50-S contains more crystalline phases (Hydrotalcite
and C-A-S-H) than GPP-28d-23-S after the mini-beam test. Therefore, it suggests
that long exposure to high temperature can influence the microstructure of GPP as
further geopolymerisation occurred, and thus it would influence the creep behaviour.
168
(a)
(b)
Figure 6.19: Pore size distribution of GPP after mini-beam test: (a) DH method; (b)
MP method
169
(a)
(b)
Figure 6.20: Pore size distribution of OPCP after mini-beam test: (a) DH method; (b)
MP method
170
(a) (b)
Figure 6.21: Cross sections of GPP mini-beams: (a) GPP-28d-23-S; (b) GPP-28-50-S
Figure 6.22: X-ray diffractograms of mini-beam samples tested at 23 °C and 50 °C.

Peaks marked are Hydrotalcite (H), Chabazite (Ch) and C-A-S-H (CS)
Comparing the OPCP specimens tested at 23 °C, both OPCP-28d-23-70 and OPCP-
28d-23-S contained less total porosity than OPCP-28d generally, but there is an
increase of porosity in the pore sizes larger than 14 nm in OPCP-28d-23-50. This
would be due to drying, which caused microcracking to OPCP-28d-23-50. Moreover,
there was less drying effect on OPCP-28d-23-70 and OPCP-28d-23-S, and better
curing condition was provided to these specimens for further hydration.
171
Although it was found that the curing temperature can affect the microstructure of
OPC pastes, including the morphology of the portlandite crystal and pore structure
[264], the test results show that the curing temperature can also affect the mature
OPCP. However, the trend of the result is different to GPP. OPCP-28d-50-S shows
an increase of porosity in pore sizes above 2.4 nm, while OPCP-28d-50-50 shows a
sharp increase in pore sizes above 30 nm, followed by less pores below that pore
size. This increase in pores would be due to pore refinement from pore sizes above
100 nm, since it was reported that OPCP contains more pores larger than 100 nm
than GPP [60]. In such heat environment, the precipitation of denser inner C-S-H
and the decrease of ettringite content resulted in heterogeneous distribution of the
hydration production and coarser porosity. And on the other hand, the pore structure
of OPCP at low temperature would be affected by the formation of large ettringite
crystal [265], as an increase in pore volume was observed for pore size larger than 16
nm in OPCP specimens tested at 0 °C.
6.5 Development of creep model for mini-beam test
6.5.1 Background
Pores in both geopolymer and OPC specimens exist across a range of sizes. It can be
assumed that the gel pores (the smallest size pores) are filled with pore solution.
When the specimen is under loading, both solid phases and pores inside the specimen
are under stress. This exerts pressure on the liquid in the pores as illustrated in Figure
6.23. Since the capillary pores may not be fully filled with the pore solution, there is
less pressure in the capillary pores. This pressure gradient would create a flow
through the porous structure according to Darcy’s Law. Creep is assumed to be due
to the diffusion of water and other molecules along the micropores towards the
macropores in the cement gel microstructure under sustained stress, as the pores are
squeezed when these molecules moved out. The rate of diffusion is related to the
thickness and the distance of the hindered adsorbed layers [130].
172
Applied loads
Stress
on solid
Solid phase
Pore filled
with Pressure
liquid on liquid
Figure 6.23: Porous material under loading
Scherer [249, 266] proposed a method to characterise elastic properties and

permeability from the relaxation response in the porous materials. The relaxation
response is obtained from a three-points bending test. Since relaxation is related to
the creep behaviour, this can use to establish a relationship between the creep
behaviour and the permeability of the porous materials. From that, a new concept has
been developed; when the permeability can be predicted based on the pore
characteristic of the porous materials, the creep behaviour can also be predicted in
this way. Therefore, the mini-beam test is used to investigate the developed method
in this study.
6.5.2 Constitutive Equations
It is assumed that creep involved flow of liquid in the pores under pressure.
Therefore, the flow through porous materials obeys Darcy’s law:
𝜅
𝐽 = − 𝜂 ∇𝑝 (6.12)
173
where J is the flux, κ is the permeability, η is the viscosity and ∇p is the gradient of
pressure.
For an elastic porous material, Biot [267] showed the mechanical behaviour of a
porous material containing a compressible fluid:
𝑄̅ 2 𝑃𝑄̅
𝜎̃𝑘 = 2𝐺𝑃 𝜀𝑘 + (𝑃̅ − 2𝐺𝑃 − 𝑅̅ ) 𝜀𝑣 + 𝜙 𝑅̅ k = x, y, z (6.13)
where 𝜎̃𝑘 is the stress on solid phase of the porous material, GP is the shear modulus
of the porous material, εk is the linear strain, εv is the volumetric strain (εv = εx + εy +
εz), ϕ is the porosity, P is the stress in the liquid phase of the porous material, and 𝑃̅,
𝑄̅, 𝑅̅ are the constants, which can be adopted from Johnson [268]:
𝐾 𝐾 𝐾
4𝐺 𝜍(𝜌− 𝑃 )𝐾𝑠 +𝜙 𝑃 𝑆
𝑃̅ = 3𝑃 +
𝐾𝑆 𝐾𝐿
𝐾𝑃 𝐾𝑆 (6.14)
𝜍− +𝜙
𝐾𝑆 𝐾𝐿
𝐾
(𝜍− 𝑃 )𝜙𝐾𝑠
𝑄̅ =
𝐾𝑆
𝐾 𝐾 (6.15)
𝜍− 𝑃 +𝜙 𝑆
𝐾𝑆 𝐾𝐿
𝜙 2 𝐾𝑠
𝑅̅ = 𝐾𝑃 𝐾 (6.16)
𝜍− +𝜙 𝑆
𝐾𝑆 𝐾𝐿
where ς is the volume fraction of solid phase in the porous material (ς = 1 – ϕ), KS
and KL are the bulk moduli of the solid and liquid phases, respectively, and KP is the
bulk modulus of the network of the porous material drained of liquid. Substituted
equations 6.14 to 16 into equation 6.13:
2 𝐾𝑃
𝜎̃𝑘 = 2𝐺𝑃 𝜀𝑘 + (− 𝐺𝑃 + 𝐾𝑃 ) 𝜀 + (𝜍 − )𝑃 (6.17)
3 𝐾𝑆
Because the total stress, σk, is the sum of the stresses on solid and liquid phases,
2 𝐾
𝜎𝑘 = 𝜎̃𝑘 + 𝜙𝑃 = 2𝐺𝑃 𝜀𝑘 + (− 3 𝐺𝑃 + 𝐾𝑃 ) 𝜀 + (𝜍 − 𝐾𝑃 ) 𝑃 + 𝜙𝑃
𝑆
(6.18)
174
The constitutive equations are derived from equation 6.18:
1 𝑃 𝐾
𝜀𝑥 = 𝐸 [𝜎𝑥 − 𝜈𝑃 (𝜎𝑦 + 𝜎𝑧 )] − (3𝐾 )(1 − 𝐾𝑃 ) (6.19a)
𝑃 𝑃 𝑆
1 𝑃 𝐾
𝜀𝑦 = 𝐸 [𝜎𝑦 − 𝜈𝑃 (𝜎𝑥 + 𝜎𝑧 )] − (3𝐾 )(1 − 𝐾𝑃 ) (6.13b)
𝑃 𝑃 𝑆
1 𝑃 𝐾
𝜀𝑧 = 𝐸 [𝜎𝑧 − 𝜈𝑃 (𝜎𝑥 + 𝜎𝑦 )] − (3𝐾 )(1 − 𝐾𝑃 ) (6.13c)
𝑃 𝑃 𝑆
where EP is the elastic modulus of the porous material, EP = 2(1+υP)GP = 3(1-

2υP)KP, and υP is the Poisson’s ratio of the porous material.
6.5.3 Continuity Equations
Considering the mass of fluid, ML, flowing through a plane of area, A, into the region
of the gel between x and x + Δx, the mass after the time interval, Δt [269]:
𝑀𝐿 (𝑡 + ∆𝑡) = 𝑀𝐿 (𝑡) + [𝜌𝐿 (𝑡, 𝑥 )𝐽(𝑥 ) − 𝜌𝐿 (𝑡, 𝑥 + ∆𝑥 )𝐽(𝑥 + ∆𝑥 )]𝐴∆𝑡
(6.20)
Because 𝑀𝐿 = 𝜌𝐿 𝑉𝐿 and 𝑉𝐿 = 𝜙𝐴∆𝑥,
1 𝜕 1
(𝜌𝐿 𝑉𝐿 ) = − ∇(𝜌𝐿 𝐽) (6.21)
𝑉𝐿 𝜕𝑡 𝜙
where ρL is the density of fluid and VL is the volume of fluid. Using Darcy’s law:
𝑉̇ 1 𝜅
𝜌̇ 𝐿 + 𝜌𝐿 𝑉𝐿 = − 𝜙 ∇(𝜌𝐿 𝜂 ∇𝑃)
𝐿
Rearranging,
𝜌̇ 𝐿 𝑉̇ 𝜅
+ 𝑉𝐿 = − 𝜙𝜂 ∇2 𝑃 (6.22)
𝜌𝐿 𝐿
175
Since the total volumetric strain rate, 𝜀̇𝑣 , is related to the volumetric strain rates of
both solid and liquid phases in saturated condition as below:
𝑉̇ +𝑉̇
𝜀̇𝑣 = 𝑉𝑆 +𝑉𝐿
𝑆 𝐿
where VS is the volume of solid, then
𝑉̇𝐿 1 𝜍𝑉̇𝑆
= 𝜙 (𝜀̇𝑣 − ) (6.23)
𝑉𝐿 𝑉𝑆
Substitute equation 6.23 into equation 6.22:
𝜌̇ 𝜍𝑉̇𝑆 𝜅
𝜙 𝜌𝐿 + 𝜀̇𝑣 − = − ⁡𝜂 ∇2 𝑃 (6.24)
𝐿 𝑉𝑆
𝑉̇ 𝑃̇ 𝜌̇ 𝑃̇
For 𝑉𝑆 = 𝐾 and 𝜌𝐿 = − 𝐾 by the definition of the bulk moduli [249], where
𝑆 𝑆 𝐿 𝐿
𝑑𝑃 𝑑𝑃
𝐾𝑆 = 𝑉𝑆 𝑑𝑉 and 𝐾𝐿 = −𝜌𝐿 𝑑𝜌 (6.25)
𝑆 𝐿
which leads to:
𝜙 𝜍 𝜅
(𝐾 + 𝐾 ) 𝑃̇ − 𝜀̇𝑣 = 𝜂 ∇2 𝑃 (6.26)
𝐿 𝑆
6.5.4 Application of sustained loads on mini-beam
For the bending action on mini-beam, the stress is only resulted in z-direction
(longitudinal direction), so σx = σy = 0. Apply this to equations 6.19 and 6.26,
𝜙 𝜍 𝐸 1 1 𝐸 𝜅
[ + 𝐾 − (9𝐾𝑃 − 1) (𝐾 − 𝐾 )] 𝑃̇ − 3𝐾𝑃 𝜀̇𝑧 = 𝜂 ∇2 𝑃 (6.27)
𝐾𝐿 𝑠 𝑃 𝑃 𝑆 𝑃
or
𝜙 𝜙 1 𝜎̇ 𝜅
(𝐾 − 𝐾 + 𝐾 ) 𝑃̇ − 3𝐾𝑧 = 𝜂 ∇2 𝑃 (6.28)
𝐿 𝑠 𝑃 𝑝
At moment immediately after bending of the mini-beam by the applied loads, there is
no flow through the porous body yet. J = 0, equation 6.27 becomes,
176
𝜙 𝜍 𝐸 1 1 𝐸
[ − 𝐾 − (9𝐾𝑃 − 1) (𝐾 − 𝐾 )] 𝑃̇ − 3𝐾𝑃 𝜀̇𝑧 = 0
𝐾𝐿 𝑆 𝑃 𝑃 𝑆 𝑃
𝐸𝑃 𝜙 𝜍 𝐸 1 1
𝜀̇𝑧 = [𝐾 − 𝐾 − (9𝐾𝑃 − 1) (𝐾 − 𝐾 )] 𝑃̇
3𝐾𝑃 𝐿 𝑆 𝑃 𝑃 𝑆
Therefore,
𝐸𝑃 𝜙 𝜍 𝐸 1 1
𝜀 = [𝐾 − 𝐾 − (9𝐾𝑃 − 1) (𝐾 − 𝐾 )] 𝑃0 (6.29)
3𝐾𝑃 𝑧0 𝐿 𝑆 𝑃 𝑃 𝑆
The mini-beam is under four-point bending. In this case, the deflection, δ, and the
bending moment, M, at mid-span are:
23𝑊𝐿3
𝛿 = − 1296𝐸 (6.30)
𝑃𝐼
𝜎𝑧𝐼 𝜀𝑧 𝐸𝑃 𝐼 𝑊𝐿
𝑀=− =− = (6.31)
𝑦 𝑦 6
Therefore,
216𝛿𝑦
𝜀𝑧 = (6.32)
23𝐿2
where W is the sum of two equal point loads on the mini-beam, L is the span, I is the
second moment of area of the cross section, and y is vertical distance from the
𝜙 𝜍 𝐸 1 1
centroid of the mini-beam. Let 𝜆 = [𝐾 − 𝐾 − (9𝐾𝑃 − 1) (𝐾 − 𝐾 )] and with
𝐿 𝑆 𝑃 𝑃 𝑆
equation 6.32 and the initial condition, equation 6.29:
72𝐸 𝛿 𝑦
𝑃0 (𝑦) = 23𝐾𝑝 𝜆𝐿
0
2 (6.33)
𝑃
For anytime under sustained load, 𝜎̇𝑧 = 0. Therefore, equation 6.28 becomes:
𝜙 𝜙 1 𝜂
(𝐾 − 𝐾 + 𝐾 ) 𝜅 𝑃̇ = ∇2 𝑃 (6.34)
𝐿 𝑆 𝑃
For the rectangular cross section of mini-beam,
𝜕 2𝑃 𝜕 2𝑃
∇2 𝑃 = 𝜕𝑥 2 + 𝜕𝑦 2 (6.35)
177
Let x = d×i and y = b×j, where x and y are the vertical and horizontal distances from
the centroid of the cross-section, respectively, d and b are half of the depth and width
of the cross-section, respectively, and i and j are the vector from -1 to 1. It is
assumed when the aspect ratio (depth to width) of the beam is 1:5 or greater, it is
effectively “infinitely wide” [252], and there is negligible pressure difference at the
same horizontal level at the cross-section of the mini-beam. Therefore, when flow in
horizontal direction is neglected, equation 6.34 becomes,
𝜙 𝜙 1 𝜂 𝜕𝑃 1 𝜕 2𝑃
(𝐾 − 𝐾 + 𝐾 ) 𝜅 𝜕𝑡 = 𝑑2 𝜕𝑖 2 (6.36)
𝐿 𝑆 𝑃
which can be written in the form of partial differential equation (PDE),
𝜕𝑃 𝜕 2𝑃
= (6.37)
𝜕𝜃 𝜕𝑖 2
𝑡𝜅𝑑2
where 𝜃 = 𝜙 𝜙 1 as a function of time.
( − + )𝜂
𝐾𝐿 𝐾𝑆 𝐾𝑃
The mini-beams under sealed condition are used to investigate the basic creep of the
mini-beams. In this case, no flow exchange is happened between the specimens and
the environment. The boundary condition is considered as insulated. Therefore, the
calculation of this PDE has the initial condition and boundary conditions as:
72𝐸 𝛿 𝑑
𝑃(𝑖, 0) = 𝑃0 (𝑦) = (23𝐾𝑝 𝜆𝐿
0
I.C. 2) 𝑖
𝑃
B.C. 𝑃(0, 𝜃) = 0
𝜕𝑃(1,𝜃)
=0
𝜕𝜃
𝜕𝑃(−1,𝜃)
=0
𝜕𝜃
72𝐸 𝛿 𝑑
Let 𝛺 = 23𝐾𝑝 𝜆𝐿
0
2 . Solving this PDE gives the following result:
𝑃
∞ 1 2
(2𝑛 − 1)𝜋𝑖 (2𝑛 − 1)𝜋𝑖 −[(2𝑛−1)𝜋 ] 𝜃
𝑃(𝑖, 𝜃) = ∑ ∫ [𝛺𝑖 sin ] 𝑑𝑖 sin 𝑒 2
−1 2 2
𝑛=1
(6.38)
178
Using equations 6.29 and 6.38,
𝜎(𝑖) 𝑃(𝑖,𝜃) 𝐾 1 𝐾
𝛿(𝑡) 𝜀𝑧 (𝑡) −( )(1− 𝑃 ) (1− 𝑃 )[𝑃(𝑖,0)−𝑃(𝑖,𝜃)]
𝐸𝑃 3𝐾𝑃 𝐾𝑆 3𝐾𝑃 𝐾𝑆
=𝜀 = 𝜎(𝑖) 𝑃(𝑖,0) 𝐾𝑃 =1+ 𝜎(𝑖) 𝑃(𝑖,0) 𝐾
δ(0) 𝑧 (0) −( )(1− ) −( )(1− 𝑃 )
𝐸𝑃 3𝐾𝑃 𝐾𝑆 𝐸𝑃 3𝐾𝑃 𝐾𝑆
(6.39)
Therefore, the creep function is,
1 𝐾
(1− 𝑃 )[𝑃(𝑖,0)−𝑃(𝑖,𝜃)]
3𝐾𝑃 𝐾𝑆
𝜑 (𝑡 ) = 𝜎(𝑖) 𝑃(𝑖,0) 𝐾 (6.40)
−( )(1− 𝑃 )
𝐸𝑃 3𝐾𝑃 𝐾𝑆
This is the basic theory for the creep model for elastic porous materials. For
viscoelastic material, it requires further development.
6.5.5 Viscoelastic behaviour
However, Past research showed that OPC specimens exhibits viscoelastic behaviour
under sustained loads. This influences the creep behaviour, because the creep strain
is contributed by the viscoelastic strain and viscous strain [171]. Although
geopolymer sample was found to show the viscoelastic behaviour due to its viscous-
flow-type phenomenon at high temperature [270], there is no research on how
viscoelastic behaviour influences the creep property of geopolymer. At this stage, it
is assumed that viscoelastic behaviour of geopolymer would be the similar effect on
creep as it does on OPC specimens. In the research on OPC specimens, aging was
found to have an influence on the viscoelastic behaviour. Aging in hardened OPC
paste involves volume growth of the hydration products, which increases the
macroscopic stiffness and viscosity of the material [173].
The viscoelastic behaviour is modelled as a time-depended change in material

properties in this case. The following properties are defined for viscoelastic analysis:
 Elastic modulus, EPV(t)

 Shear modulus, GPV(t)
 Bulk modulus of porous material, KPV(t)
 Poisson’s ratio, υPV(t)
179
The viscoelastic change on elastic modulus can be assumed to be the following
equations:
𝐸𝑃𝑉 (𝑡) = 𝐸𝑃 × 𝑉𝐸(𝑡) (6.41)
1
𝑉𝐸 (𝑡) = 𝑡 (6.42)
1+ln[1+( )𝑣 ]
𝑢
where VE(t) is a time-depended function for viscoelastic behaviour, and u and v are
the constants for the viscoelastic material. Since the applied strain on the mini-beam
is too small to cause significant alternation on the microstructure of the mature
material, the Poisson’s ratio is assumed to be remained the same. The bulk moduli of
solid and liquid phase are also remain constant [249]. Therefore, GPV(t) and KPV(t)
are proportional to EPV(t).
𝐺𝑃𝑉 (𝑡) = 𝐺𝑃 × 𝑉𝐸(𝑡) (6.43)
𝐾𝑃𝑉 (𝑡) = 𝐾𝑃 × 𝑉𝐸(𝑡) (6.44)
Hence, equation 6.36 becomes,
𝜙 𝜙 1 𝜂 𝜕𝑃 1 𝜕 2𝑃
(𝐾 − 𝐾 + 𝐾 ) = 𝑑2 𝜕𝑖 2 (6.45)
𝐿 𝑆 𝑃𝑉 (𝑡) 𝜅 𝜕𝑡
𝑡𝜅𝑑2
Let 𝜃𝑉 = 𝜙 𝜙 1 , the PDE becomes,
( − + )𝜂
𝐾𝐿 𝐾𝑆 𝐾𝑃𝑉 (𝑡)
72𝐸 (𝑡)𝛿 𝑑 72𝐸 𝛿 𝑑

𝑃(𝑖, 0) = 𝑃0 (𝑦) = (23𝐾𝑃𝑉 (𝑡)𝜆𝐿0 2 ) 𝑖 = (23𝐾𝑝 𝜆𝐿
0
I.C. 2 ) 𝑖 = 𝛺𝑖
𝑃𝑉 𝑃
B.C. 𝑃(0, 𝜃𝑉 ) = 0
𝜕𝑃(1,𝜃𝑉 )
=0
𝜕𝜃
𝜕𝑃(−1,𝜃𝑉 )
=0
𝜕𝜃
Therefore,
∞ 1 2
(2𝑛 − 1)𝜋𝑖 (2𝑛 − 1)𝜋𝑖 −[(2𝑛−1)𝜋] 𝜃𝑉
𝑃(𝑖, 𝜃𝑉 ) = ∑ ∫ [𝛺𝑖 sin ] 𝑑𝑖 sin 𝑒 2
−1 2 2
𝑛=1
(6.46)
180
The linear strain in z-direction, εz(t), at time t is calculated as:
𝜎(𝑖) 𝑃(𝑖,𝜃𝑉 ) 𝐾𝑃𝑉 (𝑡)

𝜀𝑧 (𝑡) = 𝐸 − (3𝐾 )(1 − ) (6.47)
𝑃𝑉 (𝑡) 𝑃𝑉 (𝑡) 𝐾𝑆
Therefore,
𝜎(𝑖) 𝑃(𝑖, 𝜃𝑉 ) 𝐾𝑃𝑉 (𝑡)

𝛿(𝑡) 𝜀𝑧 (𝑡) 𝐸𝑃𝑉 (𝑡) − (3𝐾𝑃𝑉 (𝑡))(1 − 𝐾𝑆 )
= =
δ(0) 𝜀𝑧 (0) 𝜎(𝑖) 𝑃(𝑖, 0) 𝐾𝑃
𝐸𝑃 − ( 3𝐾𝑃 )(1 − 𝐾𝑆 )
=1
𝜎(𝑖) 𝑃(𝑖, 𝜃𝑉 ) 𝐾 (𝑡 ) 𝜎(𝑖) 𝑃(𝑖, 0) 𝐾
[𝐸 (𝑡) − ( ( ) ) (1 − 𝑃𝑉 )] − [ 𝐸 − ( 3𝐾 )(1 − 𝐾𝑃 )]
𝑃𝑉 3𝐾𝑃𝑉 𝑡 𝐾𝑆 𝑃 𝑃 𝑆
+
𝜎(𝑖) 𝑃(𝑖, 0) 𝐾𝑃
𝐸𝑃 − ( 3𝐾𝑃 )(1 − 𝐾𝑆 )
𝜎(𝑖) 𝑃(𝑖, 𝜃𝑉 ) 𝐾𝑃𝑉 (𝑡) 𝜎(𝑖) 𝑃(𝑖, 0) 𝐾𝑃

[𝐸 (𝑡) − ( ) (1 − )] − [ − ( )(1 −
𝑃𝑉 3𝐾𝑃𝑉 (𝑡) 𝐾𝑆 𝐸𝑃 3𝐾𝑃 𝐾𝑆 )]
𝜑 (𝑡 ) =
𝜎(𝑖) 𝑃(𝑖, 0) 𝐾𝑃
𝐸𝑃 − ( 3𝐾𝑃 )(1 − 𝐾𝑆 )
(6.48)
6.5.6 Estimation of permeability from pore size distribution
One of the major input parameters to this model is the permeability of the porous
material. It can be estimated by analysis on pore size distribution. However, the
range of sizes involved in the flow should be considered. For microdiffusion under
applied stress, it is assumed that the pore solution moves along the micropores
towards the macropores in the pore structure [130]. Usually the gel pores and the
small capillary pores (micropores to small mesopores) are filled with pore solution,
while the large capillary pores (macropores) contain both pore solution, vapour and
air. Such diffusion of pore solution can be indicated by an increase in internal RH
inside the concrete specimen under compressive stress [271]. Therefore, the pore
sizes below 50 nm are considered in this micro-diffusion process under saturated
condition in this study. Pore size distribution by the nitrogen gas adsorption test is
suitable in the estimation of permeability within this range of pore sizes.
181
A permeability estimation method by Amiri, Aït-Mokhtar and Sarhani [96] was
adopted. This method is based on polymodal of pore size distribution:
𝛿𝜙
𝜅 = 24𝜏2 𝑅𝑛 2 (6.49)
where ζ is the constrictivity factor, ϕ is the porosity, τ is the tortuosity factor and Rn is
the pore radius corresponding to the minimal pore section. The constrictivity factor
can be defined by,
2
1
(∑𝑛
1=1 )
𝑞𝑖𝑛
𝜁= 𝑚2
(6.50)
1 𝑖𝑛
(∑𝑛
1=1 )(∑𝑛
1=1 𝑞 )
𝑞𝑖𝑛𝑚𝑖𝑛 𝑖𝑛
𝑆𝑗 𝑙𝑗
𝑚𝑖𝑗 = and 𝑞𝑖𝑗 =
𝑆𝑖 𝑙𝑖
where the pore section Si > Sj if i < j, the li and lj are the corresponding pore length.
The pore length can be calculated from division of the pore volume by the pore
𝑉𝑖
section area, 𝑙𝑖 = .
𝑆𝑖
The tortuosity factor can be defined by,
∑𝑛
𝑖=1 𝑙𝑖
𝜏= (6.51)
𝑙
where l is the global length of the porous material. However, it is difficult to estimate
l, due to the complexity of the microstructures of geopolymer and OPC specimens.
An empirical expression for cement-based materials by Carniglia [272] was
suggested to calculate the tortuosity factor based on the porosity:
𝜏 = 2.23 − 1.13𝜙 (6.52)
However, Salmas and Androutsopoulos [273] argued the accuracy of Carniglia’s

model, since the empirical expression was based on results by MIP, which was not
reliable in tracing pore sizes in the lower mesopore range. They proposed the
corrugated pore structure model (CPSM) to estimate the tortuosity factor:
182
𝐷𝑚𝑎𝑥,𝑒𝑓𝑓 −𝐷𝑚𝑖𝑛,𝑒𝑓𝑓
𝜏 = 1 + 0.69 ( ) (𝑁𝑆 − 2)0.58 (6.53)
𝐷𝑚𝑒𝑎𝑛
where Dmax,eff and Dmin,eff are the maximum and minimum effective pore diameters by
truncating the pore volume pore size distribution by 2.5% from each end of the
distribution detected by the nitrogen gas adsorption test, Dmean is the mean pore
diameter from the pore volume pore size distribution, and Ns is the pore length
parameter.
The viscoelastic behaviour would influence the permeability, since the size of the
pores would be reduced or enlarged during the loading period. The change of
permeability through equation 6.49 can be expressed as below,
𝜅(𝑡) 𝑅 (𝑡)2
= 𝑅𝑛(0)2
𝜅(0) 𝑛
Assuming the creep strain is equal to the change of strain in pores, hence,
𝑅𝑛 (𝑡)−𝑅𝑛 (0)
= 𝜀𝑧 (𝑡) − 𝜀𝑧 (0) = 𝜀𝑧 (0)𝜑(𝑡)
𝑅𝑛 (0)
𝑅𝑛 (𝑡) = 𝑅𝑛 (0)[1 − 𝜀𝑧 (0)𝜑(𝑡)]
Therefore,
𝜅(𝑡) 𝑅 (𝑡)2
= 𝑅𝑛(0)2 = [1 − 𝜀𝑧 (0)𝜑(𝑡)]2 (6.54)
𝜅(0) 𝑛
The maximum initial strain resulted from the applied load on the mini-beam is less
than 50 microstrains. For example, with the creep coefficient φ = 4, the change of
permeability is approximately 0.04% only. Therefore, the effect of viscoelastic
behaviour on the permeability is negligible in this case. In order to simplify the
calculation, only the initial permeability estimated from equation 6.49 was used.
183
6.5.7 Validation of creep model
The proposed method was applied on the mini-beams, GPP-28d-23-S and OPCP-
28d-23-S, for the basic creep behaviour. The input parameters and results are
presented below.
6.5.7.1 Permeability
The permeability was estimated according to method described in section 6.5.6. The
adsorption/desorption isotherms of GPP-28d and OPCP-28d are shown in Figure
6.24. It can be seen that the hysteresis loop of GPP is wide, unlike that of OPCP,
which has a narrow end close to high pressure region.
Ns was estimated according to CPSM [274] by matching the isotherm characteristic.

The adsorption/desorption isotherms of GPP agrees with the hysteresis loop H 4-
CPSM, and the adsorption/desorption isotherms of OPCP agrees with the hysteresis
loop H3-CPSM. Therefore, Ns for GPP and OPCP were assumed to be 5 and 3
according to these types of hysteresis loop, respectively. The tortuosity factor was
then calculated.
(a) (b)
Figure 6.24: Adsorption/Desorption isotherms from nitrogen gas adsorption test: (a)
GPP-28d; (b) OPCP-28d
184
Based on the desorption isotherms, the pore size distributions of GPP-28d and
OPCP-28d were analysed and shown in Figure 6.25. Two samples were analysed for
both specimens. In general, GPP-28d showed a pore size distribution with most pores
below 2 nm approximately. This can be taken as two modes, with possible three to
four modes, since the distribution in small pores covered a range of pore sizes. On
the other hand, OPCP-28d showed a distinctive two-modes distribution; one at 1.8
nm and another one at larger pore size. A possible three to four modes can be
considered in this case.
The permeability was estimated using equation 6.49. The results are shown in Table
6.4. With these input parameters, the estimated permeability of GPP-28d and OPCP-
28d were 2.16×10-5 to 4.48×10-5 nm2 and 3.36×10-4 to 1.28×10-3 nm2, respectively.
There is no significant difference in the estimated permeability between the
considered modes for the same sample. However, the estimated permeability is
double between two samples of GPP-28d and approximate 4.7 times for OPCP-28d.
This would be due to variation in pore sizes within the specimen, and such variation
is considered acceptable.
In the calculation, the porosity was obtained from the total volume of pores within
the measurement range of the nitrogen gas adsorption test. Therefore, it did not
include the large capillary pores and air voids. The adopted porosity of GPP-28d was
0.073 to 0.078, while the adopted porosity of GPP-28d was 0.219 to 0.237. These
also contributed to the estimated permeability of GPP-28d and OPCP-28d, which
leaded one order difference in magnitude. This is because that GPP-28d has overall
lower porosity and most porosity at the small pore sizes. Hence, the diffusion process
in GPP is expected to take longer time than in OPCP.
Based on the above calculations, the estimated permeability for the creep model was
taken from the average of the estimated values for each specimens, which were
3.28×10-5 nm2 and 6.75×10-4 nm2 for GPP-28d and OPCP-28d, respectively.
185
(a)
(b)
Figure 6.25: Pore size distribution by DH method: (a) GPP-28d; (b) OPCP-28d
186
Table 6.4: Input parameters and estimated permeability of GPP-28d and OPCP-28d
Specimens Considered Pore Pore length Tortuosity Constrictivity Porosity, Permeability,
modes, n radius, rp parameter, factor, τ factor, δ φ κ (nm2)
(nm) Ns
GPP-28d
Sample 1 2 2.7075 5 22.37 0.983 0.078 2.20E-05
1.8527
3 22.072 5 22.37 0.967 0.078 2.16E-05
2.7075
1.8527
4 22.072 5 22.37 1.026 0.078 2.29E-05

2.7075
2.0972
1.8527
Sample 2 2 2.7075 5 15.40 0.980 0.073 4.33E-05
1.8527
3 4.6128 5 15.40 0.951 0.073 4.20E-05
2.7075
1.8527
4 4.6128 5 15.40 1.014 0.073 4.48E-05

2.7075
2.0972
1.8527
OPCP-28d
Sample 1 2 14.125 3 4.26 0.685 0.237 1.28E-03
1.8527
3 14.125 3 4.26 0.637 0.237 1.19E-03

10.651
1.8527
Sample 2 2 33.814 3 9.03 0.874 0.219 3.36E-04

1.8527
3 33.814 3 9.03 0.773 0.219 2.97E-04

14.125
1.8527
4 33.814 3 9.03 0.710 0.219 2.73E-04

14.125
9.2279
1.8527
6.5.7.2 Elastic modulus of porous material by mini-beam test

The elastic modulus can be determined through the immediate deflection of the mini-
beams at loading by using equation 6.36. The test details, the immediate deflection
and the resultant elastic modulus are shown in Table 6.5. The values of elastic
modulus of GPP-28d-23-S and OPCP-28d-23-S are 10.74 and 12.40 GPa,
respectively. It is found that the stiffness of GPP-28d-23-S is lower than that of
OPCP-28d-23-S. This agrees on the results of the previous tests on GPC and OPCC.
After the elastic modulus was determined, the shear modulus and the bulk modulus
of these materials can be calculated based on an assumption on the Poisson’s ratio of
187
0.2 (similar to Poisson’s ratio of concrete). Therefore, the values of shear modulus of
GPP-28d-23-S and OPCP-28d-23-S are 4.48 and 5.17 GPa, and the values of bulk
modulus of GPP-28d-23-S and OPCP-28d-23-S are 5.97 and 6.89 GPa, respectively.
Table 6.5: Elastic modulus based on mini-beam test on GPP-28d-23-S and OPCP-
28d-23-S
GPP-28d-23-S OPCP-28d-23-S
Load, W (g) 758 755.85
Span, L (mm) 300 300
Width, b (mm) 49.72 49.94
Depth, d (mm) 9.92 10.04
4
Second moment of area, I (mm ) 4044.90 4212.02
Deflection, δ0 (μm) -82 -68
Elastic modulus, EP (GPa) 10.74 12.40
6.5.7.3 Bulk moduli of solid and liquid phases

Besides the elastic modulus of the porous body, the bulk moduli of the solid and
liquid phases are required in the model. The bulk modulus of the solid phase was
decided from the major strength giving gels of hardened geopolymer and OPC,
which are C-A-S-H gel in slag-based geopolymer and C-S-H gel in OPC. The elastic
modulus of C-S-H can be measured by atomic force microscopy (AFM) [275, 276].
Test was done on the paste samples of GPP-28d and OPCP-28d and the resultant
mapping images are shown in Figure 6.26.
It can be seen that the image of geopolymer shows two large areas with over 120
GPa (shown bright in Figure 6.26). These are the unreacted grains of slag particle. In
this case they are treated similar to function of aggregates and thus has no effect on
creep. The rest can be divided into two groups: the calcium rich phases (C-A-S-H)
and the sodium rich ones (C-(N)-A-S-H. The former and the latter has an elastic
modulus of approximately 70 and 40 GPa, respectively. Therefore, an average value
of 52 GPa was determined by the proportion of these two phases from the AFM
analysis.
188
(a)
(b)
Figure 6.26: AFM images of elastic modulus: (a) Geopolymer, (b) OPC
189
Similar to geopolymer, unhydrated OPC particles are also presented in the hardened
OPC. Without these particles, the average elastic modulus of C-S-H gel was
estimated to be 56 GPa. After the estimated elastic modulus was obtained, the bulk
modulus of the solid phases can be calculated by assuming the Poisson’s ratio of the
solid phase similar to that of the porous network, which was 0.2. Therefore, the
values of bulk modulus of the solid phase in GPP and OPCP are calculated as 28.9
and 31.1 GPa, respectively.
The bulk modulus of the liquid phase was determined from the pore solution within
the pore network in the materials. It was assumed to be similar to the bulk modulus
of water, which is 2.15 GPa at room temperature.
6.4.7.4 Viscosity of pore solution

The viscosity of the pore solution was first assumed to be similar to water, which is
0.88 mPa.s. However, it was noted that the pore solution is not pure water; it contains
alkaline which results in high pH value. Such alkaline solutions have higher viscosity
than the pure water [277]. The pore solutions of hardened geopolymer and OPC were
extracted from the samples by using an expression device with high pressure over
350 MPa [278]. 3 to 5 ml of pore solution was obtained from each sample (see
Figure 6.27a). They were then tested for the viscosity by using the rheometer as
shown in Figure 6.27b. It was found that the average viscosity of the pore solution of
geopolymer was 1.31 mPa.s, which was higher than that of OPC, 1.01 mPa.s. This is
due to the rich alkaline contents in the pore solution of geopolymer.
190
(a) (b)
Figure 6.27: Viscosity measurement of pore solution: (a) Pore solution from
expression, (b) Rheometer
6.5.7.5 Results of creep model

The above parameters were input to the developed creep model for mini-beam.
Although the mini-beams were sealed, the test results suggests that there was some
effect of drying at the later age, which causing reduction in creep on GPP-28d-23-S
after 14 days. This would be due to the sealing method leaving a thin gap between
the plastic wrap and the specimen. A local environment was developed and caused
the drying effect. Therefore, it is suggested to apply the results from the early period
of loading, e.g. first seven days, on the model. For the drying specimens, GPP-28d-
23-50 and OPCP-28d-23-50, it is found that drying influences the results at early
loading period, giving an uncertainty to the creep behaviour. Since the proposed
creep model on mini-beam is developed based on isolated boundary condition, it
would not be applicable to the mini-beams under drying conditions.
The results of creep curve of the mini-beams are shown in Figure 6.28. It shows three
curves: the creep curve from mini-beam tests, the elastic creep function
191
(hydrodynamic) and the viscoelastic creep function. It can be seen that when there is
only the hydrodynamic behaviour (for elastic porous materials), the creep model
greatly underestimates the creep behaviour of both GPP and OPCP. This indicates
that the viscoelastic behaviour has a great influence on both GPP and OPCP, but the
hydrodynamic behaviour does influence the creep behaviour to some degrees. The
hydrodynamic behaviour is caused by the flow of the pore solution from high
pressure to low pressure within the pore network. Figure 6.29 shows the change of
maximum pore pressure in both specimens according to creep model. It can be seen
that GPP takes a longer time than OPCP in return to zero pore pressure (equilibrium).
This is due to the low permeability by fine pore network and the high viscosity of the
alkaline pore solution of geopolymer. Because the flow of pore solution continues
until the pore pressure back to zero, the hydrodynamic behaviour of GPP takes a
longer time to complete compared to OPCP. In this case, the creep response of GPP
is delayed due to this hydrodynamic behaviour. This can be observed from the results
by the creep model for elastic porous materials. GPP took 3 days to approach creep
coefficient of 0.18, while OPCP took less than a day in approaching creep coefficient
of 0.11. This also influences the viscoelastic creep function, which shows a slower
creep increase in GPP in first few days of loading.
By including the viscoelastic behaviour, the creep model matches the creep curves of
the test data of GPP and OPCP. The factors, u and v, for time-dependent behaviour
were determined for GPP-28d-23-S and OPCP-28d-23-S by fitting as shown in Table
6.6. This is only a small difference in the values of u and v between GPP and OPCP.
This suggests the viscoelastic behaviour of GPP and OPCP would be similar within
this short testing period.
Table 6.6: Parameters for viscoelastic modelling
Model parameters GPP-28d-23-S OPCP-28d-23-S

u 0.03 0.035
v 0.2 0.25
192
(a)
(b)
Figure 6.28: Creep coefficient by creep model for mini-beam: (a) GPP, (b) OPCP
193
Figure 6.29: Maximum pore pressure during loading period by creep model
Comparing the pore pressure calculated based on the elastic model and the
viscoelastic model, the pore pressure relaxation in the viscoelastic model is delayed
(see Figure 6.29). This is the result of the viscoelastic deformation of the solid
phases, which continuously applies pressure on the liquid phase. This forces the
liquid phase migration from high pressure area to low pressure area, until the
deformation of the solid phases becomes stable. This process can take a long period,
as it did not complete for GPP in comparison to the estimation from the elastic creep
function.
The creep behaviour of geopolymer and OPC was investigated through the mini-
beam test, which is designed for specimens made of geopolymer and OPC pastes and
mortars. The test is mainly carried on paste specimens because the paste portion in
the paste-aggregates composition of concrete or mortar is mainly responsible for
creep effect. The specimens are cast into a thin plate shape, so called mini-beam, and
two sustained point loads are applied on it for measurement of long-term deflection.
The test results are then used to calculate the creep coefficient.
194
Drying affected the geopolymer paste specimens and the high shrinkage strain
caused cracking on 1-day old specimen. It also greatly increased the creep response
of GPP specimens compared to the same specimens tested in the sealed condition.
This effect due to increasing drying creep with decreasing relative humidity. The
same effect also happened on OPCP specimens but the impact of drying effect was
less. In general, the GPP specimens exhibited high creep response than the OPCP
specimens under drying conditions, although both specimens showed similar creep
coefficient under sealed condition. Another observation on the drying GPP
specimens is a counter displacement caused by differential shrinkage across the
depth of the mini-beam. This is because the tension stress at the bottom side of the
mini-beam induces microcracks, which causes drying deeper into the section. As a
result, the drying shrinkage at the bottom side of the mini-beam higher than that at
the top side.
The creep response of GPP specimens increased when it was tested at 50 °C, but this
did not happen on OPCP specimens. It indicates that creep of GPP was influenced
more by high temperature than OPCP, and thus creep of geopolymer would be
dominated more by temperature dependent flow than by moisture diffusion under
stress. Such effect would be due to further geopolymerisation, since an increase of
porosity in pore sizes between 10 nm and 30 nm was observed, which formed more
crystalline phases and alternated the microstructure.
A creep model for the mini-beam was developed based on the pore characteristic and
the viscoelastic behaviour of GPP and OPCP. This estimated the creep behaviour
from the hydrodynamic behaviour and the viscoelastic behaviour. The hydrodynamic
behaviour is controlled by the flow of pore solution within the pore network under
stress. Therefore, the permeability is vital to estimate the hydrodynamic behaviour.
The estimation of permeability from the pore size distribution shows that the
permeability for GPP was approximately 20 times lower than that of OPCP. In
addition, the viscosity of the pore solution of geopolymer was 1.31 mPa.s, which was
higher than that of OPC, 1.01 mPa.s, owning to the rich alkaline contents in the pore
solution of geopolymer. Therefore, the resistance to the flow was higher in GPP, and
thus the creep model showed that the pore pressure caused by applied load took
longer time to disappear in GPP than OPCP. This explained the lower creep
195
development of GPP in the early age than OPCP. However, majority of the creep
response was due to the viscoelastic deformation of the solid phases, which would
involve ageing, sliding or breakage of the solid phases. The creep model showed that
the viscoelastic behaviour of GPP and OPCP was similar within the short testing
period. However, this can be difference in later age since the creep rate of GPC was
higher that of OPCC in the standard creep test. The viscoelastic deformation of the
solid phases continuously applies pressure on the liquid phase, and forces the flow
until deformation of the solid phases becomes stable. This would take a very long
period to complete. Although the creep model shows a good correlation between the
porosity characteristic and the creep behaviour of the GPP and OPCP specimens
under sealed condition, it is not suitable for drying specimens due to the unstable
movement caused by differential shrinkage.
196
Chapter 7 : Long-term deflection of GPC
beams
7.1 Introduction
There has been little research done on long-term mechanical properties on

geopolymer structural elements [144]. In order to investigate the creep behaviour on
GPC structures, two types of full-scale concrete beam were constructed with GPC
and tested under sustained loads for a long period. The profile of these GPC beams
are designed for two types of structural elements; beam type 1 for normal reinforced
concrete beam and beam type 2 for composite concrete floor. The dimensions of
these GPC beams are:
3. RC beams of 200 mm wide  240 mm deep  4000 mm long.

4. Composite beam with steel formwork of 600 mm wide  240 mm deep 
6500 mm long.
Data of long-term deflection over the test period was collected and analysed. Due to
the constraints in test environment and resource, OPCC beam was not constructed in
this project. Instead, virtual concrete beams were made with the use of a finite
element modelling software, ATENA Studio, based on the input of property
parameters of the reference OPCC used in this project. The same reinforcement
details as the GPC beams detail was applied on these OPCC beams. The different
structural behaviour during the long-term tests between GPC beams and OPCC
beams was discussed.
The results of creep tests on GPC from Chapter 4 were used to predict the long-term
deflection of the GPC beams by using the rational methods. The effectiveness of
these methods was discussed. Factors that influence the long-term behaviour of the
GPC beams were identified.
After the long-term test, these GPC beams were subjected to further tests on strength
and carbonation. These provide further information on long-term properties of GPC.
197
7.2.1 Beam type 1: Reinforced concrete beam
The RC beams were cast into the mould, and then it was demoulded at the age of 7
days. They were then placed on the supports and a sustained uniformly distributed
load (UDL) of 2.0 kN/m was applied on top of the beams with the use of concrete
blocks. Long-term deflection was measured by the dial gauges with accuracy of 0.01
mm at mid-span below the beams before and after applied loads, and at ages from 7
days to 1.5 years. The loads were kept on the beams throughout the test period. The
articulation of the beams was simply supported over the span of 3600 mm (see
Figure 7.1). Four beams with various reinforcement details were constructed to
investigate the effect of reinforcement on GPC beam. One extra beam with the same
reinforcement detail as GPRCB1 was made and loaded at later age of 28 days. Detail
of these beams is shown in Table 7.1.
(a)
(b)
Figure 7.1: Detail of GPC beams type 1: (a) test setup and (b) cross-section detail.
198
Table 7.1: Details of GPC beam type 1
Specimen ID Top Bottom Details

reinforcement reinforcement
GPRCB1 None 2N12 Loaded at age of 7 days
GPRCB2 2N12 2N12 Loaded at age of 7 days
GPRCB3 None 2N16 Loaded at age of 7 days
GPRCB4 2N16 2N16 Loaded at age of 7 days
GPRCB1-A None 2N12 Loaded at age of 28 days
# All reinforcements have a minimum yield strength of 500 MPa.
7.2.2 Beam type 2: Composite steel-concrete beam
Flexural creep testing was carried out on two composite GPC beams, which
replicated the suspended floor systems in practice in full size (span, depth and
reinforcements). Two identical formworks were set up with composite steel
formwork. The formworks were elevated from the ground and supported on two
permanent supports on both ends and two temporary supports spaced at 2 m. Both
beams were cast in place where they were to remain under sustained load inside the
laboratory to simulate cast in-situ construction. Detail of the beam setup is shown in
Figure 7.2. The reinforcement details of these two beams are listed in Table 7.2. All
reinforcements have a minimum yield stress of 500 MPa. Dial gauges with accuracy
of 0.01 mm were placed at mid-span below the beams for the measurement of
deflection throughout the test period. Six foil type strain gauges were attached to the
top reinforcement and covered with a number of coats for long-term protection on
each beam (see Figure 7.3). These strain gauges are used to investigate change of
strain in three different sections:
 Section A: Over the supports for uncrack section without creep effect;
 Section B: 1 m from supports for uncracked section under creep effects;
 Section C: Mid-span for fully cracked section under creep effects.
199
(a)
(b)
Figure 7.2: Detail of GPC beams type 2: (a) test setup and (b) cross-section detail
Table 7.2: Details of reinforcement in GPC beam type 2
Beam types GPCB1 GPCB2

Top reinforcement SL72 Mesh# SL92 Mesh#
Bottom reinforcement N12 at 200 mm centres* N12 at 200 mm centres*
Composite steel formwork KF57 1.00 BMT KF57 1.00 BMT
#
SL72 and SL92 are the code for square steel reinforcement mesh of 6.75 mm and 8.6 mm diameter wire, respectively. Both
minimum yield strength is 500 MPa and spaced at 200 mm in both longitudinal and transverse directions;
*
N12 is the code for 12 mm diameter deformed bar, minimum yield strength of 500 MPa;

KF57 1.00 BMT is the supplier’s code for composite steel formwork sheet of 1 mm thick. Further information can be found in
the supplier’s website. <http://www.fielders.com.au/aspx/kingflor.aspx>
200
Figure 7.3: Protecting coats for strain gauge
After the GPC beams were cast, they were covered with plastic sheets to minimise
moisture loss from the surface to the atmosphere. Water was applied regularly to
provide better curing. The formworks on the side and end faces were removed at 3
days after casting. Two layers of MasterKure® CC 1315WB curing compound were
then painted on the side and end surfaces. Therefore, only the top surface was
exposed to drying, which simulated the wide slabs where there is no moisture loss
from the sides. The temporary supports were removed at the age of 14 days, followed
by the application of uniformly distributed load of 1 kPa on top of the beams with the
use of 20 kg sandbags. These loads were maintained on the GPC beams throughout
the test period. The articulation of the beams was simply supported over a span of
6000 mm. Records of strain data from the strain gauges and measurement of
deflection from the dial gauges were obtained before and after casting of GPC, and at
ages from 3 days to 2 years.
201
7.2.3 Structural analysis
The GPC beams were under analysis for their creep behaviour. Research by Sarker
[279] shows that the analytical method for ultimate strength design of OPCC can be
applicable on GPC. Therefore, three rational methods, which were developed for
OPCC, were assessed in prediction of long-term deflection and change in strain of
these GPC beams:
1. RCM
2. EMM
3. AEMM
The results from these methods were compared with the test results. Properties of
GPC, reinforcement details and load profile for input parameters to the analysis
methods were obtained from the experimental results including elastic modulus,
modulus of rupture, creep and shrinkage from Chapter 4.
Based on the initial calculation, all GPC beams type 1 were designed to be uncrack
throughout the test period, and both GPC beams type 2 would be cracked within 4 m
in the middle of the beams under immediate applied loads. Further refinement was
done with the input of actual material properties, and then it was compared with the
observation from these beams.
7.3 Results of long-term beam tests
7.3.1 GPC beam type 1
7.3.1.1 Long-term deflection

The GPC beams were under self-weight and sustained loads for one and a half years
(see photo in Figure 7.4). The long-term deflection of these beams measured at mid-
span is shown in Figure 7.5. GPRCB3 shows the largest deflection among all beams,
with the total deflection of 10.86 mm and the creep deflection of 9.63 mm at the end
202
of the test, whilst GPRCB4 has the least deflection, with the total deflection of 4.89
mm and the creep deflection of 3.70 mm at the end of the test.
Figure 7.4: Photo of long –term beam test on GPC beams type 1
Figure 7.5: Deflection of reinforced GPC beams type 1
203
7.3.1.2 Effect of reinforcement detail on creep deformation
Since all these GPC beams carried similar amount of loads, the different responses in
deflection are due to the reinforcement details. The effect of reinforcement detail on
long-term deflection is significant in this test. Figure 7.5 shows that the deflection of
the beams can be identified into two groups: Large deflection (GPRCB1, GPRCB3
and GPRCB1-A) and small deflection (GPRCB2 and GPRCB4). This is because
there is no top reinforcement in the former group of GPC beams. Without the top
reinforcement in the compression zone of the beams, the compressive stress on the
GPC is higher and the stiffness of these beams is less, compared to another group of
beams with top reinforcement.
GPRCB4 shows the least long-term deflection among all beams, since it is heavily
reinforced by 2 numbers of 16 mm dia. steel reinforcing bar at top and bottom. When
smaller size of 12 mm dia. reinforcing bars at top and bottom are used, GPRCB2
deflects slightly more than GPRCB4. The larger compression reinforcement reduces
more amount of compressive stress in the GPC, so that the creep deformation in
GPRCB4 is less than that in GPRCB2.
Another time-dependent effect, shrinkage, also contributes the deflection of these

beams. Since there is no top reinforcement in GPRCB1, GPRCB1-A and GPRCB3,
they are not symmetric in reinforcement profile in horizontal. Shrinkage occurred in
these beams is restrained by the bottom reinforcements only. This causes the gradual
change in strain across the cross-section of the beams, and results in an additional
downward deflection to the creep deflection of these GPC beams. Since the GPRCB3
is reinforced heavier than GPRCB1, the total deflection of GPRCB3 is larger than
that of GPRCB1 due to more restraint by the reinforcement. Sometimes, the
deflection caused by shrinkage can be a big part in the long-term deflection, when
the shrinkage of the material is high [9].
To sum up the above findings, it can be seen that the design of reinforcement detail
can have a big impact on the long-term behaviour of GPC structures. A careful
design of reinforcement detail (e.g. choice of compression reinforcements) can
control the long-term deflection of the GPC beams.
204
7.3.1.3 Effect of age of loading on creep deformation
GPRCB1 and GPRCB1-A are identical in design details. The only difference is that
GPRCB1-A was loaded at later age of 28 days, instead of 7 days for GPRCB1. The
long-term deflection of GPRCB1-A is only slightly less than GPRCB1 as shown in
Figure 7.5. This would contradict the observation of the effect of age of loading from
the standard creep test on the GPC cylinders presented in Chapter 4. However, when
the amount of long-term deflection is compared with the immediate deflection as
shown in Figure 7.6, the ratio of GPRCB1-A is lower than that of GPRCB1. This is
due to the larger immediate deflection of GPRCB1-A, which is 1.55 mm, compared
to 1.20 mm for GPRCB1. This larger immediate deflection would be a result of
weakened stiffness due to drying [219].
This suggests that the effect of age of loading has an influence on the long-term
deformation of GPC structures. Therefore, when GPC structures are designed, the
details and construction method (curing method, time of construction load
application, etc.) should be planned carefully for serviceability requirement.
Figure 7.6: Ratio of long-term deflection to immediate deflection
205
7.3.1.4 Comparison to OPCC beams
OPCC beams were modelled with using computer software, GiD, and the long-term
response was analysed in computer software, ATENA Studio (see Figure 7.7). The
input parameters for the material properties were adopted from the reference
specimens of OPCC presented in Chapter 4. These OPCC beams were modelled with
the same reinforcement details and loading details as the GPC beams for comparison.
The OPCC beams are identified as RCB1, RCB2, RCB3, RCB4 and RCB1-A,
corresponded to the identity of the GPC beams in this case.
The predicted deflection of OPCC beams from the models is presented in Figure 7.8
together with the test results of GPC beams. It can be seen in Figure 7.8a~d that
GPRCB1 to GPRCB4 generally have the deflection double more than their OPCC
counterparts by the end of the test period, e.g., the final deflections of GPRCB1 and
RCB1 are 10.48 and 4.39 mm, respectively. This result generally agrees with the
previous finding on the standard creep test on GPC and OPCC, in which the creep
and shrinkage effects of GPC are higher than those of OPCC. Therefore, these affects
the long-term structural behaviour of GPC structures. Another contribution to the
large deflection of GPC beams is the low elastic modulus of GPC, so the GPC beams
have lower stiffness than OPCC beams. This causes large immediate and long-term
deflection. For example, the elastic modulus of OPCC at 28 days is higher than that
of GPC from previous study, the immediate deflection of RCB1-A is 0.86 mm and
this deflection is lower than that of GPRCB1-A, which is 1.55 mm. However, the
final deflection of GPRCB1-A, 10.44 mm, is almost four times larger than that of
RCB1-A, which is 2.76 mm.
The difference in deflection between the GPC beams and the OPCC beams is
significant. This can be contributed by high creep, high shrinkage and low elastic
modulus as stated above. However, there is also a major difference between the GPC
and OPCC beams in environment conditions. The OPCC beams were modelled
under constant condition of 23 °C and 50% RH, while the GPC beams were tested
subjected to the seasonal changes inside the laboratory environment. Such changes in
temperature and humidity would influence the creep response of GPC. Further
discussion of influence by external environment was presented in section 7.4.1.2.
206
(a)
(b)
Studio
207
(a) (b)
(c) (d)
(e)
Figure 7.8: Comparison of deflection of tested GPC beams to modelled OPCC

beams: (a) GPRCB1 and RCB1; (b) GPRCB2 and RCB2; (c) GPRCB3 and RCB3;
(d) GPRCB4 and RCB4; (e) GPRCB1-A and RCB1-A
208
7.3.1.5 Flexural capacity and compressive strength of GPC beams after creep test
After the sustained loads were removed from the GPC beams at the end of the creep
test, the GPC beams were tested under four-points bending for flexural capacity.
Unfortunately, GPRCB3 was damaged during transportation, so the tests were only
performed on the rest of the GPC beams. The beams were supported over a span of
3.6 m. Two equal point loads were exerted 900 mm apart at the middle of the beams.
Two LVDT were employed to record the deflection at the mid-span and under the
loading point of the beams. A picture of the test setup is shown in Figure 7.9.
The mid-span deflection under applied load for GPRCB1, GPRCB2, GPRCB4 and
GPRCB1-A is shown in Figure 7.10. The capacity of GPRCB1, GPRCB2 and
GPRCB1-A is around 30 kN, while that of GPRCB4 is 58 kN. This is due to the
larger bottom reinforcement in GPRCB4, which increases the flexural capacity.
Pictures of the deformed beams are shown in Figure 7.11. It is found that all GPC
beam were highly ductile during the test. Failure of GPRCB1 was not defined
because its deflection reached the limit of clearance below the beam. The maximum
measured deflection of GPRCB1 is 168 mm in this case. After that, the clearance
was increased for the rest of the beam tests. The maximum deflection of GPRCB2,
GPRCB4 and GPRCB1-A is 300 mm, 280 mm and 300 mm, respectively. The
failure of the GPC beam was only observed on GPRCB2 and GPRCB4, while
GPRCB1-A is anticipated to deflect more than 300 mm before failure occurs. The
typical failure mode of the GPC beams is shown in Figure 7.12. It can be seen that a
number of flexural cracks due to bending of the beam are well distributed within the
region of mid-span, which was subjected to maximum bending moment. All these
cracks are approximately in same spacing to each other. When the beams were close
to failure, crushing of concrete was observed at the top surface of the beams. This
result indicates the ductile performance of the GPC beams.
209
Figure 7.9: Test setup of the GPC beam under four-points bending
Figure 7.10: Load vs. deflection of beam test on GPC beams
210
(a) (b)
(c) (d)
Figure 7.11: Deformed shape of GPC beams after testing: (a) GPRCB1; (b)
GPRCB2; (c) GPRCB4; (d) GPRCB1-A
Concrete crushing
Flexural cracks
Figure 7.12: Cracking pattern of GPC beam at failure
211
After the beam test, concrete cores of 89 mm dia. was taken from the end of the GPC
beams as shown in Figure 7.13. These concrete cores were then prepared for
compressive strength test. Both ends of the cores were grinded flat and parallel. The
cores had a height to diameter of 2:1 approximately. The average compressive
strength of four GPC cores after the test period of creep test is 55.1 MPa, which is
higher than the 28-day compressive strength of GPC of same batch by approximately
10%. This indicates the continuous growth in compressive strength during the test
period by further geopolymerisation.
(a) (b)
Figure 7.13: In-situ specimens for compression strength test: (a) coring; (b) core
specimens
212
7.3.2.1 Long-term deflection

Figure 7.14 shows the change of deflection at mid-span for both beams relative to the
measurement before removal of temporary supports. Beams GPCB1 and GPCB2
deflect 8.61 and 8.79 mm under self-weight, respectively. The instantaneous
deflection due to the applied loads is 1.34 mm for GPCB1 and 1.09 mm for GPCB2.
The big difference between the immediate deflections caused by self-weight and
applied loads is due to large depth of the beams, which is equal to 6 kPa of load by
self-weight only. The total deflection at the end of the test is 33.06 mm and 32.09
mm for GPCB1 and GPCB2, respectively. This is because heavier reinforced beam
GPCB2 is stiffer than GPCB1. Deflection due to creep is calculated from the
difference between the total deflection and the immediate deflection. GPCB2
exhibits slightly less creep deflection than GPCB1 by 0.9 mm.
Figure 7.14: Long-term deflection of GPC beams type 2
The recorded data from the strain gauges for all three cross sections on the beams is
shown in Figure 7.15 as the average of two strain values. The strain values measured
from the reinforcements of GPCB2 is generally less than those of GPCB1. Since it
can assume that the strain of the steel reinforcement is the same as that of concrete at
213
the same horizontal level, this suggests that there is less compressive stress
experienced in the GPC in GPCB2. However, it shows that increasing the
compression reinforcement from SL72 to SL92 is not effective in reducing long-term
deflection in this case, although the increase is 62%. This would be because the
amount of top reinforcements is low, compared to the amount of bottom
reinforcements in both beams, which includes 3 numbers of N12 and the composite
steel sheet of 1 mm thick. More top reinforcements would be installed to reduce
long-term deflection.
Figure 7.15: Recorded strain from strain gauges on GPC beams type 2
7.3.2.2 Comparison to OPCC beams

OPCC composite beams were modelled with using computer software, GiD and
ATENA Studio as shown in Figure 7.16, similar to beam type 1. The OPCC
composite beams in this case are identified as B1 and B2.
The deflection of these OPCC composite beams from the models is shown in Figure
7.17, together with the test results of the GPC beams type 2. The immediate
deflections of the OPCC composite beams, B1 and B2, caused by loading are 3.45
and 3.43 mm, and the final deflections are 13.41 and 12.99 mm, respectively. The
214
deflection caused by loading is significant lower than that of GPC composite beams,
which are 9.95 and 9.88 mm for GPCB1 and GPCB2, respectively. This indicates the
lower stiffness of GPC beams than OPCC beams.
(a)
(b)
Studio
215
Figure 7.17: Deflection of GPC beams type 2 and modelled OPCC beams type 2
The deflections due to long-term effects are 9.95 and 9.55 mm for B1 and B2,
respectively. The effect of reinforcement is not effective for the choice of
reinforcement in these OPCC beams, which is similar case to GPCB1 and GPCB2.
However, these values of long-term deflection are lower than those of GPCB1 and
GPCB2 by 57% approximately. This agrees with the finding in the comparison of
GPC and OPCC beams type 1, but the difference is less for GPC and OPCC beams
type 2. Since it was found in the standard creep tests previously that the sealed GPC
cylinders exhibits significant lower creep deformation than the drying GPC
cylinders, isolation provided by the steel composite formwork at the bottom of the
beams would reduce the overall creep effect in GPC beams type 2.
7.3.2.3 Carbonation and compressive strength after creep test

The GPC beams had been stored in the indoor environment for more than 2 years at
the end of the long-term beam test. Concrete cores of 89 mm dia. were taken from
the end and the mid-span of the beams. Right after the cores were obtained, they
were cleaned with a dump cloth, and the carbonation of the GPC after the test period
was assessed. It was done by spraying the phenolphthalein indicator solution to the
216
fresh shaft surface of the cores. The indicator solution was prepared with 1 gram of
phenolphthalein in 70 ml of ethanol and 100 ml of water. The mechanism of the
phenolphthalein method is that it shows a pink/purple colour on the concrete surface
when its pH value exceeds 9, and colourless when its pH value is below 9. Since
carbonation causes a reduction in pH of concrete below 9, the colour change is used
as the indication of depth of carbonation [77]. All GPC cores were assessed, and it
was found that when the indicator solution was first sprayed to the cores, there were
two regions of colour change on the concrete surface. A light pink was shown close
to the exposed surface, followed by the darker colour at deeper section (Figure
7.18a). However, the light pink colour faded after a few minutes (Figure 7.18b). This
light pink region would be due to the spread of alkali contents from GPC into the
water used in coring process. Therefore, the carbonation depth was taken from the
GPC cores a few minutes after application of the indicator. The results show that the
carbonation depth of GPC is around 11 to 16 mm after 800 days of age. This is
higher than the test results by Sanjuán, Andrade and Cheyrezy [280] on OPCC
subjected to natural carbonation for 2 years, which was around 3 to 5 mm [280]. The
high carbonation of the GPC in this test generally agreed with the investigation by
Bakharev, Sanjayan and Cheng [77] on slag-based GPC.
After the carbonation test, the GPC cores obtained from the end of the beams were
then prepared for compressive strength test. Both ends of the cores were grinded flat
and parallel. The cores had a height to diameter of 2:1 approximately. The average
compressive strength of two GPC cores after the long-term test was 66.0 MPa. This
shows a significant increase in strength compared with the 28-day compressive
strength of 42.9 MPa. This increase is more than the strength growth in GPC beams
type 1. This would be due to better internal curing provided by the large dimension
of beam type 2 and the sealing condition from the bottom of the beam.
217
(a) (b)
Figure 7.18: Assessment of carbonation on GPC with phenolphthalein method: (a)

right after spraying indicator solution; (b) a few minutes later
7.4 Prediction of long-term deflection of GPC beams
7.4.1.1 Prediction of deflection

Estimating the creep response of the structures is one of the important parts in
serviceability design. In this case, three different methods, rate of creep method
(RCM), effective modulus method (EMM) and age-adjusted effective modulus
method (AEMM), were used to predict the long-term deflection of the GPC beams
type 1 with the test data from the standard creep tests and shrinkage test presented in
chapter 4. The elastic modulus of the GPC applied to these methods was 25,000 MPa
at 7 days and 27,000 MPa at 28 days, approximately obtained from the standard
creep tests. The density of the GPC was approximately 2460 kg/m3. The predictions
of the long-term deflection of these GPC beams are shown in Figure 7.19.
218
(a) GPRCB1 (b) GPRCB2
(c) GPRCB3 (d) GPRCB4
(e) GPRCB1-A
Figure 7.19: Prediction of long-term deflection of GPC beams type 1
219
RCM performs poorly in the prediction since it underestimates the deflection in
GPRCB1, GPRCB2 and GPRCB1-A, and overestimates the deflection of GPRCB3
and GPRCB4. The overestimation by RCM is due to an indication of cracking
caused by combined creep and shrinkage effects for GPRCB3 and GPRCB4. The
flexural cracks start from the mid-span of the beams and spread toward to supports.
The stiffness of the beams is reduced in the cracking region and thus the deflection
increases. However, it does not match the observation on the test beams, which
showed no sign of cracking throughout the test period.
EMM shows a close prediction of deflection in GPRCB1 and GPRCB3, but less
accurate in GPRCB2 and GPRCB4 than AEMM. It is noted that the former group of
beams has no top reinforcement installed, while the latter group has symmetric
reinforcement detail. This difference in reinforcement detail results in different creep
prediction from these two methods. Therefore, AEMM, as an improved version to
EMM, performs more consistent in the prediction of all beams.
The prediction of the deflection of GPRCB1-A is underestimated by all methods (see

Figure 7.19e). It is found that the predicted immediate deflection after loaded is
computed to be 1.06 mm, which is lower than the test result of 1.55 mm. This is
different to the results of the deflection calculation for the other GPC beams, which
predicted immediate deflection after loaded is close to the test results. This suggests
the actual elastic modulus on GPRCB1-A is lower than the input parameter of elastic
modulus, which is 27,000 MPa. GPRCB1-A would be weaken during the period
before tested. It is noticed that GPRCB1-A was supported on supports and two
temporary props to minimise creep due to self-weight before being loaded at 28 days.
However, shrinkage occurred within this period would induce a tensile stress at the
upper part of the beam due to the non-symmetric beam profile and the support
condition, since the downward displacement was restrained. Such tensile stress
would cause micro-cracking and thus reduce the stiffness of the beam. The elastic
modulus of GPRCB1-A was then estimated to be approximately 19,000 MPa via a
backward calculation in the elastic beam analysis.
220
7.4.1.2 Effect of environmental changes on GPC beams
The prediction of the long-term deflection of the GPC beams from the previous
section is generally lower than the test results. This is due to the effect of
environment. It is noted that the beams were stored in the laboratory under ambient
indoor condition, which temperature and humidity were not controlled. However, the
input parameters to the prediction were obtained from the standard tests on GPC
cylinder specimens, which were tested under standard condition in an environmental
chamber. The GPC beams were subjected to seasonal change of temperature and
humidity. Such environmental difference would influence the actual creep and
shrinkage effects experienced in the GPC beams. Therefore, application of the results
from the standard creep test and standard shrinkage test on the deflection prediction
of the GPC beams would be inappropriate, and in this case, it underestimates the
long-term deflection. In order to understand the effects of the environmental
condition, a profile of shrinkage and creep effects on these GPC beams was derived
by back calculation from the long-term deflection using AEMM as shown in Figure
7.20 and Figure 7.21.
Figure 7.20: Estimated shrinkage strain of GPC beams related to age of 7 days
221
(a)
(b)
Figure 7.21: Estimated creep coefficient GPC beams: (a) GPRCB1, GPRCB2,
GPRCB3 and GPRCB4, (b) GPRCB1-A
222
It can be seen that the creep effect on the GPC beams is higher than the creep result
from the standard creep test, but the shrinkage effect is lower in the GPC beams. The
estimated creep coefficient of the beams increases significantly in the first 300 days,
and then it turns into a slow rate of increase until a large increase occurs again at 400
days. Such behaviour is believed to be due to the seasonal change in the local area, as
temperature increases and decreases yearly. It is possible that higher temperature
results in higher creep potential in GPC, as it was observed in OPCC in the tests
conducted by Vidal, et al. [126]. Therefore, a plot of increment of creep increase in
each month of loading period is drawn against the maximum temperatures recorded
in the local area, obtained from the bureau of meteorology Australia [281], as shown
in Figure 7.22. It can be seen that there is an effect of seasonal changes on the creep
behaviour of the GPC beams. The creep rate increases between the 5th month and the
10th month is corresponded to the high temperature during the summer season, when
temperature was higher than 23 °C in some periods. In that period, the creep
increment of the GPC beams is significant higher than that of the standard creep test.
After that, the creep rate of the GPC beams decreases in 11 th to 15th month, and it
gets closer to that the standard creep test. However, it increases again in 16 th to 18th
month, when it enters the beginning of second summer season. Because of that, the
application of creep coefficient obtained from standard creep test at 23 °C
underestimates the long-term response of the GPC beams. This result indicates that
the effect of temperature has a large influence on the creep of GPC.
One of the explanation of the effect of temperature on creep behaviour is the change
of viscosity of water in the pore network of concrete [126]. The viscosity of water
decreases with increasing temperature; it would accelerate the flow rate of micro-
diffusion and thus lead to higher creep effect. However, the increase of creep of GPC
is large in this case. Even the temperature did not exceed 40 °C in the test period; the
estimated creep increase in the GPC beams is almost 3 times more than the result
from the standard creep test for some period. This increase is high in comparison
with the basic creep of high-strength concrete between 20 and 50°C from Vidal, et al
[126], which showed an increase by double approximately. The high temperature
allows further geopolymerisation in GPC [270], it would change the microstructure
of the geopolymer, and then influence the creep behaviour.
223
Figure 7.22: Creep change and temperature change in each month of loading
Figure 7.20 shows that the shrinkage strain of the GPC beams and the standard
drying shrinkage prism of GPC match each other in the beginning of the test.
However, the shrinkage strain in the GPC beams is lower than the result from the
standard drying shrinkage test in the later age. This would be due to the size effect
between the two specimen types; the GPC beams has a cross section of 240 × 200
mm, while the shrinkage prism specimens has a cross section of 75 × 75 mm.
Although it was suggested that the size had no effect on the ultimate shrinkage strain
in OPCC, it took longer time for the large specimens to reach ultimate shrinkage
strain than the small specimens [282]. This is due to the longer time required for the
longer distance of water movement from the internal to the environment, until the
equilibrium of RH is reached. Since it was reported previously that GPC had less
porosity and smaller pore sizes than OPCC, it is expected that the GPC beams would
take a much longer period until they reach the ultimate shrinkage strain.
The test results of the GPC indicate that high temperature can increase the creep
response of the concrete structures, this is critical in design of structures subject to
change due to seasonal variation, such as bridges [283]. Further investigation is done
224
to understand the creep behaviour of GPC under various temperature conditions in
Chapter 8.
7.4.1.3 Predicted deflection of GPC beams with effects of environmental changes

The calculated results of the deflection of the GPC beams with the updated creep and
shrinkage profile are shown in Figure 7.23. The deflection contributed by only
shrinkage effect was also calculated. This deflection is caused by differential
shrinkage due to the non-symmetric reinforcement profile of the GPC beams, so
there is no effect to the deflection of the GPC beams with symmetric reinforcement
profile, which are GPRCB2 and GPRCB4 in this case. The estimated deflections due
to shrinkage effect for GPRCB1, GPRCB3 and GPRCB1-A are 2.20 mm, 3.01 mm
and 1.28 mm, respectively. The deflection of GPRCB3 is 0.81 mm larger than
GPRCB1 because its heavier bottom reinforcements provides more restraint to
shrinkage effect. However, the total deflection of GPRCB3 is only 0.38 mm larger
than GPRCB1. This indicates that the deflection of GPRCB3 due to creep effect is
less than that of GPRCB1. This is because the stiffness of GPRCB3 is higher than
that of GPRCB1, and thus the deflection by loading and creep effect is less in
GPRCB3 than GPRCB1.
GPRCB1-A shows less deflection due to shrinkage effect than GPRCB1 by 40%.
This is because GPRCB1-A was loaded at later age than the other GPC beams.
Hence, some amount of shrinkage had happened before it was loaded.
Therefore, it shows that the long-term behaviour of the GPC structures is influenced
by a number of factors, including environmental factor, size factor, and
reinforcement details. However, the creep behaviour of GPC can be controlled by
careful detailing and construction sequence. When the creep mechanism of GPC is
fully understood and the creep and shrinkage models of GPC are developed for real
life environments, GPC structures can be designed accordingly for required
performance.
225
(a) GPRCB1 (b) GPRCB2
(c) GPRCB3 (d) GPRCB4
(e) GPRCB1-A
Figure 7.23: Long-term deflection of GPC beams with updated creep coefficient and
shrinkage strain by using AEMM
226
7.4.2.1 Prediction of deflection

Because the beams were loaded at 14 days, elastic modulus and creep coefficients
obtained from the creep tests on concrete cylinders at loading age of 14 days were
taken for the calculation. The elastic modulus was taken as 29,000 MPa in this case.
Drying shrinkage is influenced by the size of concrete in term of strain rate [107], so
the strain data from section A of the beams were taken to calculate the shrinkage
strain for the analysis of the other sections within the beams, instead of using the test
data from the standard drying shrinkage test on prism specimens. It can be done
simply by performing section analysis at this cross-section in a reverse calculation
from the strain of the top reinforcement to an equivalent shrinkage strain of concrete,
as the creep effect caused by sustained loads is excluded at such location. In the
prediction of deflection, the equivalent shrinkage strain is assumed to be uniform
across the section. Since the shrinkage strain is assumed to remain the same across
the beam, the use of equivalent shrinkage strain provides a simple and quick
calculation. The equivalent shrinkage strain for GPCB1 and GPCB2 by RCM, EMM,
and AEMM is shown in Figure 7.24. The results show that the calculated values of
equivalent shrinkage strain for EMM and AEMM are very close, but the value from
RCM is 10% higher than the values from the others. In general, the calculated
shrinkage strain of these GPC beams is approximately 1110 to 1240 × 10 -6.
In order to predict the deflection of the beams, the first step in the member analysis is
to identify the crack region of the beams immediately after loading and at the later
ages. Loads on the beams include self-weight (3.48 kN/m) and sustained loads (1 kPa
= 0.6 kN/m). The modulus of rupture of 5.21 MPa from the sealed condition at age
of 14 days was used to determine the location of cracks due to immediate loading,
and a value of 4.99 MPa was used for the ages later than 28 days. The modulus of
rupture between 14 days and 28 days was obtained by linear interpolation between
these two values. When the calculated maximum stress of concrete exceeds the
modulus of rupture, the beam is considered cracked from that cross-section. The
region of crack is then determined. The next step of the member analysis is to
calculate the curvature at cross sections of every 0.5 m according to the status of
227
cracking. Finally, the deflection can be estimated with the obtained values of the
curvature by double integration.
The graphs of curvature of beams GPCB1 and GPCB2 calculated from RCM, EMM,
and AEMM at first loaded and at the end of the test are shown in Figure 7.25 to
Figure 7.27. When there is a crack identified in the analysis, the calculated curvature
increases in such location due to the loss of stiffness. Moreover, crack status take
place from this point toward mid-span of the beam.
The resulting estimation of the deflection of the GPC beams were calculated
according to the curvature profiles of the GPC beams by each method as shown in
Figure 7.28. All methods show similar result overall. RCM gives the highest
estimation among all three methods in both GPC beams, whilst EMM gives the
lowest.
Figure 7.24: Calculated shrinkage strain of GPC beam type 2
228
(a)
(b)
of test by RCM: (a) GPCB1, (b) GPCB2
229
(a)
(b)
of test by EMM: (a) GPCB1, (b) GPCB2
230
(a)
(b)
of test by AEMM: (a) GPCB1, (b) GPCB2
231
r2 RCM = 0.88
r2 EMM = 0.94
r2 AEMM = 0.90
(a)
r2 RCM = 0.89
r2 EMM = 0.93
r2 AEMM = 0.91
(b)
Figure 7.28: Deflection of GPC beams type 2 by RCM, EMM and AEMM: (a)
GPCB1, and (b) GPCB2
232
7.4.2.2 Comparison of the prediction methods
Figure 7.25 shows that there is a large difference in the curvature of both GPC beams
predicted by RCM between first loaded and end of the test. This is because the
analysis by RCM shows that the beams does not crack under immediate loads, and
the first crack of both beams is estimated to appear at age of 35 days. Since then the
beams would crack gradually from mid-span to supports due to the creep effect at
later ages. However, it is difficult to determine the cracking status of the GPC beams
at early stage because when the beams were first loaded at the age of 14 days, no
visible crack was observed. The first flexural cracks were only observed after 35
days within the 2 m mid-span region. These cracks would be caused by the applied
loads but the crack width only opened wide enough to make it visible at later age.
Unlike RCM, results from both EMM and AEMM indicate that both beams crack
when first loaded as shown in Figure 7.26 and Figure 7.27. However, due to the high
creep coefficient of GPC, the values of the maximum tensile stress of concrete
estimated after age of 35 days were smaller than that when first loaded. The beam
would be indicated by the analysis as uncracked for later ages, and the uncracked
analysis approach would be used to estimate the deflection. Therefore, the long-term
deflection would be underestimated. This does not agree with the fact that the GPC
beams had already cracked under sustained loads. Therefore, cracked section
analysis should be applied in the cracked regions predicted in the early age for the
rest of the test period. The long-term deflection can be predicted properly, as the
cracked section remains cracked at all times. This indicates the importance of
analysis at a number of time intervals in GPC structural design.
The prediction of deflection was overestimated by RCM for GPCB1 and GPCB2,
with a root-mean-square of 0.88 and 0.89, respectively. The results from EMM and
AEMM show a better prediction on deflection compared to RCM, since their values
of root-mean-square are above 0.9. This result is expected because RCM does not
include delayed creep recovery, the prediction would be overestimated and thus, it
would be used to represented the upper bound of the prediction [115]. It is found that
the results from EMM and AEMM are very similar. The introduction of the age-
adjusted coefficient only has a minor effect on the calculations of stress and strain
across the sections in this case.
233
There is an overestimation in the first two months of test period in both EMM and
AEMM prediction. This is because the coefficient β in the calculation of effect of
tension stiffening by Eurocode 2-2004 [182] was chosen as 0.5 for long-term
analysis, e.g. age after first loaded at 14 days, and this underestimated the tension
stiffening effect at the early age. Since GPC shows a generally higher tensile strength
than OPCC [27] and GPC was reported to have higher bond strength to steel than
OPCC [284], it is expected the effect tension stiffening to be stronger in GPC. Since
the period considering short-term was not specified in Eurocode 2-2004, it is
suggested to apply β = 1 for the age up to 56 days in this analysis. In this case, the
prediction of deflection by AEMM is in close agreement with the test results with the
values of root mean square of 0.98 and 0.99 for GPCB1 and GPCB2, respectively
(see Figure 7.29).
The long-term deflection of these GPC beams was also influenced by seasonal
change in the local environment. Two significant increases in deflection were
observed from 300 days to 500 days and after 700 days. Therefore, the deflection
prediction by AEMM with revised β up to 56 days shows some overestimation at
some periods and yet some underestimation at other periods, but these differences are
small. This would be because using the actual shrinkage strain obtained from the end
of the GPC beams improves the overall accuracy. When the creep coefficient of GPC
can be predicted according to the environment condition, the prediction can be
improved further. Therefore, it is critical to develop a creep model for GPC prior to
practical design of GPC structures.
234
(a)
(b)
Figure 7.29: Deflection of GPC beams type 2 by AEMM with coefficient β = 1 for
first two months of loading period: (a) GPCB1; and (b) GPCB2
235
The long-term behaviour of GPC structures was investigated through the flexural
tests on GPC beams under sustained loads for a long period up to two years. Two
types of GPC beams were constructed for different structural purposes. Beam type 1
is for normal reinforced concrete beam and beam type 2 is for composite concrete
floor. Loads were kept on beam type 1 until 540 days, and on beam type 2 until 780
days. Virtual beams of OPCC were made with the used of the finite element
modelling software, ATENA, for comparison of performance with the GPC beams. It
is found that the long-term deflection of GPC beams is higher than that of the virtual
OPCC beams of similar reinforcement details. The larger deflection of GPC beams
than OPCC beams is contributed by lower stiffness, and higher creep and shrinkage
effects of GPC. Besides, it is also found that the creep behaviour of GPC can be
influenced by the temperature of the environment. Higher temperature can result in
higher creep response in GPC, which is similar trend from the observation on OPCC
in other study, but such impact on GPC can be higher. Higher temperature from the
environment would change the microstructure of the geopolymer, and then influence
the creep behaviour.
Although the long-term deflection of GPC beams is high, it can be controlled by

careful detailing and construction sequence. Compression reinforcement installed can
effectively reduce the deflection. Having a symmetric reinforcement detail can also
ease the deflection due to shrinkage effect. Large member size of GPC experiences
less effect of drying shrinkage than small member sizes. Since it was found in the
previous test that creep of GPC was low under sealed condition, the GPC structures
can be isolated with a protective layer for creep-sensitive design.
RCM, EMM and AEMM were used to predict the long-term deflection of the GPC
beams. The results show that EMM and AEMM provide better prediction than RCM,
since the latter overestimated in all GPC beams. This is because RCM does not
include delayed creep recovery, but the result from RCM can be used to represent the
upper bound of the prediction. The accuracy of these methods relies on the proper
input parameters, including elastic modulus, creep coefficient and shrinkage strain,
etc. With these material properties given close to the actual properties of the GPC
236
structures, it is suggested that the long-term performance of the GPC structures can
be predicted using the same rational methods for OPCC structures. Therefore, an
accurate creep model for GPC is critical.
After the long-term test, GPC beams type 1 were tested under four-points bending
for flexural capacity. It is found that these beams showed significant ductile
performance, with some beams achieved more than 300 mm deflection before
failure. The flexural cracks were well spread within the beams with similar width.
Further test was also done on concrete cores from these GPC beams. An increase in
compressive strength was generally achieved on these GPC cores after the test period
of creep test. The GPC cores also show the carbonation depth of GPC around 11 to
16 mm after 800 days of exposure in atmosphere. This indicates that GPC structures
would be prone to carbonation. Protecting coating would be used to improve the
durability.
237
Chapter 8 : Thermal effect on creep behaviour
of GPP and OPCP
8.1 Introduction
There are a number of factors influencing the creep of concrete. One of them is the
temperature at where the concrete is subjected to loading [127]. Research by Vidal et
al. [126] shows that the creep magnitude and creep rate of OPCC tested at 80 °C was
significantly influenced compared with that tested at 20 °C. In some cases, the
difference between these two test temperatures was up to a factor of 9. They
explained that this was due to the lower viscosity of water within the pore network at
higher temperature, which amplified the sliding of C-S-H, and the thermal damage,
which weakened the specimen. Besides, the creep increase in high temperature is
also contributed by transitional thermal creep (TTC). This additional creep occurs in
OPCC when it is heated for the first time [163]. Experiment by Pan, Sanjayan and
Collins [165] shows that fly ash-based GPP showed a larger degree of transient creep
than OPCP below 250 °C. However, no work has been done on thermal creep of
slag-based GPC.
Test results from Chapters 6 and 7 show that the creep of slag-based GPC and GPP
can be influenced significantly by temperature. Therefore, further works on thermal
creep of geopolymer were done. In this chapter, specimens of GPP and OPCP were
tested for thermal creep for 1 day under four target temperatures: ambient
temperature (20 to 23 ºC), 50 ºC, 80 ºC, and 100 ºC. These temperatures were chosen
for practical condition. Tests for TTC were also conducted on both GPP and OPCP.
The thermal creep behaviour of GPP was investigated through the comparison of
results of GPP and OPCP from these tests.
238
8.2.1 Test setup
The dimension of the specimens of GPP and OPCP for these tests was 50 mm dia. ×
100 mm long. These specimens were bath cured until the age of 28 days before
tested. The ends of these paste cylinders were prepared flat and parallel, and then the
actual length of the specimens was measured for strain calculation. The result was
reported as the average of two specimens of same type tested in each test condition.
The tests were carried in an electrical furnace with a capacity of 1200 ºC combined
with the actuator 250 kN compression capacity. Thermocouples were used to
measure the temperature on the specimens. Displacement of the specimens was
measured by sensor attached to the actuator. However, this measured displacement
included the displacement of compression shaft. The displacement on the specimens
was calculated after the calibration done on a steel sample of same size, which
thermal expansion coefficient is 11 × 10-6 K-1 approximately and elastic modulus is
200,000 MPa. This calibration was done in each target temperature of the thermal
test. A picture of the test setup is shown in Figure 8.1.
Figure 8.1: Testing equipment for thermal tests
239
There were four thermal tests performed in this study, which included:
4. Hot strength
5. Thermal creep and shrinkage
6. Free thermal expansion
7. Transitional thermal creep
Both geopolymer and OPC were tested at ambient temperature (20 to 23 ºC), 50 ºC,
80 ºC, and 100 ºC. The heating conditions are shown in Figure 8.2. The increase rate
of temperature in the furnace was set as 1 ºC/min until reaching the target
temperature. The temperature was then maintained for one hour to allow equal
distribution of temperature inside the specimens by the recommendation of RILEM
129-MHT [285]. Free thermal expansive strain and transitional thermal creep strain
were obtained during the heating period. Hot strength test was commenced at the end
of heating. The thermal creep and shrinkage tests were carried out at the end of
heating path, but with temperature maintained for one day further. After the tests, the
specimens were allowed to cool down naturally.
Figure 8.2: Heating conditions for thermal tests on geopolymer and OPC pastes
240
8.2.2 Hot strength
Compressive strength test was carried out at the target temperatures on the specimens
at time indicated in Figure 8.2. The load was applied at a rate of 20 MPa/min. The
failure load was reported as reference strength at ambient temperature and hot
strength at the higher temperatures.
8.2.3 Thermal creep and shrinkage
The specimens were preloaded at 0.01 MPa prior to heating. The specimens were
loaded to 30% of hot strength at the target temperature at one hour after the target
temperature was reached. The loading rate was 5 MPa/s. The load and the
temperature were kept constant for the test period of one day. Measurements of
displacement and temperature were done at time as recommended by RILEM 129-
MHT [286] as follows:
 Before and after load application

 Every one minute for first five minutes
 Every five minutes for first hour
 Every hour until 1 day
 After unloaded
The measurement of displacement was used to calculate the total strain, ε. However,
this includes the instantaneous strain, εe, the creep strain, εcr, and the shrinkage strain,
εsh. Extra tests were done with the applied load of 0.01 MPa for effect of shrinkage
only. The thermal creep strain was calculated as:
𝜀𝑐𝑟 = 𝜀 − 𝜀𝑒 − 𝜀𝑠ℎ (8.1)
The specific creep was derived from the calculated creep strain for comparison
purpose between GPP and OPCP.
241
8.2.4 Free thermal strain
According to recommendation by RILEM 129-MHT [285], the specimens were

preloaded at 0.01 MPa prior to heating. The measurements of thermal displacement
and temperature were done during heating at a time interval of every one minute. The
temperature of the specimens was monitored with the use of thermocouples. The free
thermal expansive strain, εth, was calculated from the actual displacement after
calibration, ΔLth, and the initial measured length, L, as below.
∆𝐿𝑡ℎ
𝜀𝑡ℎ = (8.2)
𝐿
8.2.4 Transitional thermal creep
According to recommendation to obtain TTC by RILEM 129-MHT [286], the

specimens were loaded at a rate of 5 MPa/s to no more than 30% of reference
strength at ambient temperature before heating. Once the load reached the target
load, the temperature was increased to target temperature according to heating
conditions under simultaneous sustained loads. The displacement and temperature
were measured during heating at a time interval of every one minute. The transitional
thermal creep strain, εTTC, was calculated as:
𝜀𝑇𝑇𝐶 = 𝜀 − 𝜀𝑒 − 𝜀𝐹𝑇𝐸𝑆 (8.3)
The TTC strain per unit stress was then derived for comparison purpose between
GPP and OPCP.
8.2.5 XRD
Samples were taken from each specimen after the thermal creep test for XRD test.
The effect of temperature on the microstructure was investigated for its influence on
the creep behaviour.
242
8.3 Test results
8.3.1 Compressive strength and hot strength
Both GPP and OPCP specimens were tested at 50, 80 and 100 °C for compressive
strength and compared with those tested at room temperature. The results are showed
in Figure 8.3. The compressive strength of GPP and OPCP at room temperature was
52.9 and 58.9 MPa, respectively. When the specimens were tested at higher
temperatures, the compressive strength of OPCP decreased to 57.3 MPa at 50 °C,
56.8 MPa at 80 °C, and 46.0 MPa at 100 °C. However, the decrease in compressive
strength was larger in GPP specimens; it decreased to 35.9 MPa at 50 °C, 29.9 MPa
at 80 °C, and 32.5 MPa at 100 °C. The compressive strength of GPP reduced
approximately 40% at 100 °C, while the loss of strength was only 22% for OPCP.
Figure 8.3: Compressive strength of GPP and OPCP under different temperatures
The loss in strength was linked the increase in porosity caused by hot environment
[287]. The increase of porosity was due to quick release of moisture contents from
the specimens [288]. This would cause thermal damage to the specimens. However,
243
this caused more impact on the strength of GPP, as the finer pores in GPP was
quickly replaced with the coarser pores. An example of change in pore sizes
subjected to high temperature can be seen in Chapter 6. The change in pore size
distribution was more significant in GPP than in OPCP at 50 °C. Another cause of
loss of strength is due to damage caused by differential shrinkage in hot environment
[289]. Cracks can be developed on GPP subjected to simultaneous drying and
heating.
8.3.2 Thermal creep
The results of 1-day thermal creep test on GPP and OPCP under various temperature
conditions are shown in Figure 8.4 to Figure 8.7. There are three series of data shown
in these figures, which include the total strain from the thermal creep test, the
shrinkage strain from the thermal test without load, and the creep strain derived from
these two tests.
After the test period of 1 day, both GPP and OPCP show similar shrinkage strain in
the cases at ambient temperature, 50 °C and 80 °C. The values of shrinkage strain of
GPP and OPCP are 890 and 880 microstrains at ambient temperature, 1170 and 1440
microstrains at 50 °C and 1600 and 1530 microstrains at 80 °C, respectively.
However, a significant difference is found in the case at 100 °C, in which the
shrinkage strains of GPP and OPCP are 3600 and 1680 microstrains, respectively.
This indicates that high temperature can have a large impact on the geometry
stability of GPP.
Since the GPP and OPCP specimens were tested under different amounts of load in
these creep tests, specific creep was calculated for each case and shown in Figure
8.8. The specific creep of GPP is generally larger than that of OPCP in each
temperature condition by 80% to 350%. In addition, the ratios of specific creep of
GPP at the end of the test in each temperature condition relative to the ambient
temperature are 2.3, 6.6 and 7.4 for 50 °C, 80 °C and 100 °C, and those of OPCP are
2.0, 2.6 and 3.8 for 50 °C, 80 °C and 100 °C, respectively. It shows that the specific
creep of GPP increases with increasing temperature, which is similar to the thermal
244
creep behaviour of OPCP. However, the increase in GPP is larger than that in OPCP,
especially at temperature above 50 °C.
(a)
(b)
Figure 8.4: Creep and shrinkage strains at ambient temperature: (a) GPP, and (b)
OPCP
245
(a)
(b)
Figure 8.5: Creep and shrinkage strains at 50 °C: (a) GPP, and (b) OPCP
246
(a)
(b)
247
(a)
(b)
248
Figure 8.8: Specific creep of GPP and OPCP tested at various temperatures
8.3.3 Transitional thermal creep and free thermal strain
TTC strain was calculated as the difference between the total thermal creep strain
from the TTC test and the free thermal strain from the thermal creep and shrinkage
tests at 100 °C during the heating path. The result is shown in Figure 8.9. It can be
seen that the TTC strain of both GPP and OPCP increases gradually under increasing
temperature, but TTC strain of GPP increases significantly after 80 °C. A
comparison of TTC strain per unit stress is shown in Figure 8.10. From 80 °C to 95
°C, the TTC strain per unit stress of GPP increases from 175 to 320 microstrains,
while that of OPCP increases from 95 to 125 microstrains.
Because the temperature was measured from the surface of the specimens, it was
expected that the core of the specimens did not reach the temperature shown in the
measurement. The significant increase of TTC strain of GPP above 80 °C as shown
in Figure 8.10 would suggest that the increase happened when the temperature of the
core of the specimens approached 80 °C. This indicates a slight delay on the results.
249
Therefore, when the core of the specimens reached the same temperature as the
surface above 90 °C, more TTC strain was measured. This generally agrees with the
thermal creep behaviour of GPP in previous section.
(a)
(b)
Figure 8.9: Transitional thermal creep strain: (a) GPP, and (b) OPCP
250
Figure 8.10: Transitional thermal creep strain per unit stress
8.3.4 XRD
The samples collected from GPP and OPCP after the thermal creep test were
analysed through XRD, and the results are shown in Figure 8.11. The main
crystalline phases identified in the GPP samples are the calcium-containing zeolitic
phases Hydrotalcite (Mg0.67Al0.33(CO3)0.165(OH)2(H2O)0.48, PDF# 04-011-5899),
Wollastonite (CaSiO3, PDF# 00-003-0626), Gismondine (CaAl2Si2O8·4(H2O), PDF#
00-020-0452) and other poorly ordered Tobermorite like C-A-S-H
(Ca5Si5AlO17(OH)·5(H2O), PDF# 00-019-0052). There is no obvious difference
between the GPP samples tested at 23 °C and 50 °C, expect there is a very small
trace of Vaterite (CaCO3, PDF# 00-060-0483) identified in GPP 50 °C sample. There
is not found or at very small amount in the other GPP samples, so its presence is
possible due to small cabonation of the sample over the conditioning prior to XRD
analysis. For the GPP samples tested at 80 °C and 100 °C, Wollastonite and
Gismondine were identified in the GPP sample of 80 °C, together with growth of C-
A-S-H. This would be due to thermal activation at 80 °C, which forms more C-A-S-
H gel [71]. It can be seen that there is a broad hemp between 15° and 30° 2θ in GPP
100 °C sample compared to GPP 80 °C sample. This indicates the higher content
251
amorphous phases, which are poor organised zeolite, C-A-S-H, in GPP 100 °C
sample.
(a) GPP
(b) OPCP
Figure 8.11: X-ray diffractograms of samples after thermal creep test. Peaks marked
are Portlandite (P), Hydrotalcite (Ht), Calcite (C), Gismondine (G), Wollastonite
(W), Vaterite (V), Ettringite (E), Gypsum (Gs) and C-S-H/C-A-S-H (CS)
252
On the other hand, the main crystalline phases identified in the OPCP samples are
crystalline hydration products Portlandite (Ca(OH) 2, PDF# 00-004-0733), Calcite
(CaCO3, PDF#00-003-0596), Ettringite (Ca6Al2(SO4)3(OH)12·25H2O, PDF# 00-009-
0414), Gypsum (CaSO4·2H2O, PDF# 00-003-0044) and disorganised C-S-H
(Ca5Si6O16(OH)2, PDF# 00-029-0329). No new crystalline phase is identified with
increasing temperature. The intensity of the peaks shows little difference among the
OPCP samples of 23 °C, 50 °C and 80 °C, except reduction of Ettringite in OPCP 80
°C sample. This is due to decomposition of Ettringite to meta-ettringite above 50 °C
[290]. After that, there is a reduction of intensity of the peak of Portlandite for in the
OPCP 100 °C sample. The amorphous hemp between 15° and 30° 2θ in OPCP 100
°C sample also increases similar to GPP 100 °C sample.
8.4 Effect of temperature on creep behaviour of GPP

and OPCP
The test results show that the creep behaviour of both GPP and OPCP is influenced
by temperature. This thermal effect can affect these specimens in a few ways. When
the specimens were heated, the temperature of the pore solution within the specimens
also increased. Since the viscosity of the water decreases with increasing
temperature, the viscosity of the pore solution also decreases. It accelerates the
slippage of the gel structure, which influences the creep phenomenon [126]. This can
be applied on both GPP and OPCP.
Because the specimens were tested subjected to environment in the furnace, drying
effect can influence the specimens. A measurement of weight loss after the tests
shows that generally more water loss occurs in tests at higher temperature (see Table
8.1). It indicates that more drying effect occurs at higher temperature. This agrees
with the observation of higher shrinkage at higher temperature in this study, which
can be due to higher drying shrinkage. Therefore, the drying creep would also be
influenced due to the drying effect. However, there is less water loss in GPP tested
above 80 °C than OPCP, but yet GPP exhibits higher creep increase than OPCP. This
253
would be due to the finer pore sizes in GPP which causes larger capillary tensile
force under drying condition [60].
Table 8.1: Weight of specimens after test relative to measurement before test
Temperature at test GPP OPCP

23 °C 98.8% 98.9%
50 °C 96.5% 96.5%
80 °C 93.9% 91.9%
100 °C 89.4% 87.2%
One of the contributors to higher creep in GPP would be due to damage at high
temperature. This can be seen in significant strength loss in GPP subjected to high
temperature reported from the test results, due to such thermal damage. The thermal
damage can be caused by quick release of water to the environment [288] and
differential shrinkage [289]. The higher damage in GPP would weaken the
specimens, so it leads to higher creep response.
The TTC also has an influence on the results from the thermal creep tests, since the
specimens for the thermal creep tests were heated for the first time. It can be seen
that there is a relationship between the TTC and thermal creep; both of them show
more creep increase in GPP at temperature above 80 °C. This would be due to
change of microstructure subjected to high temperature. Research on C-S-H gel
showed that heating removed the interlayer water, and the C-S-H gel structure
transformed from 1.4 nm tobermorite to 1.1 nm tobermorite near 100 °C. This
decreased the local structural order and increased the polymerisation of hydrated
silicates [291]. This explains that the broad amorphous hemp between 15° and 30° 2θ
in OPCP 100 °C sample. The polymerisation of hydrated silicates would be related to
TTC. Because of the change of microstructure after heat treatment, the creep and
shrinkage of OPCP was found reduced in the literature [292]. This was because that
the TTC has been eliminated after the heat treatment when load is applied [164].
Furthermore, White [293] suggested that drying at below 110 °C did minor
254
alternation to the C-S-H structure, since more phase alternation was observed at
temperature above 200 °C. This matches the observation on XRD analysis in this
research, in which there is minor phase alternation in OPCP from ambient
temperature to 100 °C. However, phase alternation was observed in GPP from 50 °C
to 100 °C. New crystalline phases were identified in GPP sample tested above 80 °C.
This would cause change of microstructure as more denser phases were form, and
thus the microstructure would be more compact under sustained stress, which leads
to more creep response. In addition, the increase of poor organised C-A-S-H
observed at 100 °C would be linked to the reduction of GPP sample geometry, which
could be seen in the significant increase in shrinkage strain at 100 °C. Therefore, the
rapid increase in thermal creep of GPP samples above 80 °C would be contributed by
the phase alternation of C-A-S-H.
8.5 Feasibility for accelerated creep test
From the results of the above tests, it can be seen that high temperature can increase
the creep of both GPP and OPCP. It is possible to develop an accelerated creep test
in a high temperature environment. It can shorten the test period from 1 year for the
standard creep test [206] to weeks. This is because high temperature decreases the
viscosity of pore solution, which accelerates drying and slippage of gel phases [126].
Furthermore, heat provided also accelerate hydration of OPC [287] and
geopolymerisation of geopolymer [38]. This suggests that ageing of these materials
would be accelerated.
It is critical to select a suitable temperature for such test. It is not recommend to carry
out this test at above 100 °C, due to the significant loss of water content by
evaporation. Moreover, both GPP and OPCP suffer from dehydration of major
phases [293, 294], and this would change the properties of these samples. Therefore,
it is suggested to choose the test temperature in the range between 50 to 80 °C. More
creep tests within this temperature range should be conducted to develop the
correlation to the creep test at 23 °C.
255
The thermal effect on creep behaviour of GPP and OPCP has been investigated
through the thermal creep test and TTC test on the specimens of 50 mm diameter and
100 mm long from ambient temperature to 100 ºC. Before carrying out the creep test,
the specimens were tested for compressive strength at ambient temperature, 50, 80
and 100 °C. It was found that the compressive strength generally decreased in both
specimens with increasing temperature, but the loss of strength in GPP was more
severe than OPCP. The compressive strength of GPP reduced approximately 40% at
100 °C, but it was only 22% for OPCP. Such loss of strength would be related to the
thermal damage, since the quick release of moisture contents altered the pore
structure. Cracks can also be propagated due to differential drying shrinkage between
the outer and core of the specimen.
After that, thermal creep test were carried on these specimens with the applied
loaded up to 30% of the compressive strength at test temperature for 1 day. In order
to obtain the creep behaviour only, further tests were done with the applied load of
0.01 MPa for the shrinkage effect. Then the specific creep was derived for both GPP
and OPCP. The test results show that the shrinkage strains of both GPP and OPCP
were similar at ambient temperature, 50 °C and 80 °C, with an increasing trend.
However, a significant increase was observed in GPP at 100 °C by more than twice,
compared to that at 80 °C. This would be due to dehydration of C-A-S-H gel at high
temperature, causing geometry unstable. The 1-day specific creep of GPP is larger
than OPCP in all thermal creep tests. Compared with the 1-day specific creep at
ambient temperature, the 1-day specific creep of OPCP increases by 2.0, 2.6 and 3.8
times for 50 °C, 80 °C and 100 °C. However, a significant increase in the 1-day
specific creep was observed in GPP by 6.6 and 7.4 times for 80 °C and 100 °C, while
it increases only by 2.3 times for 50 °C. This indicates the temperature dependency
in creep of GPP.
The increase in creep of GPP and OPCP at high temperature can be explained in a
number of factors. First, the viscosity of pore solution decreases with increasing
temperature, and it accelerates the slippage of the gel and increase the creep
response. Second, the quick release of water contents in high temperature also
256
increases the drying shrinkage and drying creep. Such process would also cause
thermal damage, which weakens the specimens. Finally, the thermal creep is also
contributed by the TTC, since the results from TTC tests show that the TTC strain of
GPP increases significantly from 80 °C to 95 °C by nearly 5 times compared to that
of OPCP. This is related to the change of microstructure as the XRD analysis
revealed that phase alternation of GPP occurred in the specimens tested at 80 °C and
100 °C, while there was no major alternation in OPCP. More crystalline phases were
observed in GPP samples tested above 80 °C. Such change of microstructure would
increase the thermal creep of GPP.
Because of the increase trend of creep in high temperature, it is possible to develop

an accelerated creep test, which shortens the long test period to weeks. However, it is
important to choose a suitable temperature for this accelerated creep test. Since
significant loss of water content by evaporation at 100 °C can affect the material
properties, it is recommended to conduct this test at 50 to 80 °C. Further works
should be carried out to develop the correlation between the creep tests at high
temperature and at 23 °C.
257
Chapter 9 : Summary, conclusions and
recommendations
9.1 Summary
This thesis presents a detailed investigation on creep behaviour of slag-based GPC

through a series of laboratory tests across a range of scales from microstructure to
full-scale structures.
Chapter 2 presents the literature reviews on the creep behaviour of GPC and OPCC.
It shows that there is difference between the microstructure of GPC and OPCC in gel
phases and pore size distribution. Past research on OPCC shows that the
microstructure has an influence on the material properties including strength,
durability, shrinkage and creep. However, there is some similarity in both GPC and
OPCC, so some of the creep mechanisms of OPCC, such as micro-diffusion, slippage
of gel phases and bond breakage and reform, would be applicable on GPC.
Chapter 3 presents the properties of the raw materials, the preparation of the test
specimens, and the test details. It shows that the GGBFS contains more silica and
alumina but less lime than OPC. Only one single formulation of GPC was chosen for
this research because of its satisfied strength properties for structural purpose from
the preliminary study. OPCC specimens were prepared in similar formulation with
same amount of geopolymer binder materials replaced by OPC for reference
purpose.
Chapter 4 presents the test results of GPC and OPCC on a number of standard tests
of property. It was found that the slump of the fresh GPC without superplasticiser
ranged from 70 to 150 mm, which is significantly higher than that of fresh OPCC, 10
mm. This slump value of the fresh GPC is suitable for construction purpose. The
average compressive strength of GPC at 28-days is 52.6 MPa, which is generally
comparable with that of OPCC, 57.6 MPa. GPC also develops higher early strength
than OPCC. The modulus of rupture of GPC is 6.8 MPa at 28 days, which is higher
than that of OPCC, 4.6 MPa, based on prediction by AS3600-2009 for the same
258
compressive strength grade. Drying has a negative effect on the modulus of rupture
of GPC. It was also found that GPC has lower elastic modulus than OPCC. GPC
shows higher shrinkage strain than OPCC by 30%, but 70% of the total shrinkage for
GPC is caused by autogenous shrinkage. This is not like OPCC, where autogenous
shrinkage contributes 45% of its total shrinkage. In general, GPC exhibits a lower
creep than the OPCC in the early age of loading but the creep rate of GPC does not
reduce significantly like OPCC in the later ages. Drying has a large impact on creep
of GPC, since the drying creep of GPC is almost three times higher than that of
OPCC. When drying is eliminated on the sealed specimens, the creep coefficient of
these specimens are still larger than that of OPCC by 35% approximately. However,
some creep behaviour of GPC is similar to OPCC, as higher strength and more
mature GPC shows less creep coefficient. Because of the difference in creep of GPC
and OPCC, the current creep models in the design standards and design codes for
OPCC cannot be used to predict the creep of GPC. Suggestions are given to adapt
these creep models on prediction of GPC through five series of creep data in this
study. The modified creep models, which consider factors of strength, drying and age
of loading, show good performance on prediction of creep of GPC.
Chapter 5 shows the microstructural analysis of GPC and OPCC, which includes the
gel structure and the pore structure. The initial analysis on samples of GPC and
OPCC collected at the age of 28 day before creep test shows that GPC contains more
amorphous phases and more mesopores of size below 2 nm than OPCC. In general,
the samples collected from the skin layer of the concrete specimens suffered
carbonation, since crystalline phase like Calcite was identified in the XRD and TGA
analysis. Besides carbonation, GPC also suffered from microcracking due to drying
at the skin layer. Further investigation was done on samples of GPC and OPCC
collected after creep test and shows that there is no phase alternation observed in
OPCC, but new crystalline phases, Mg-rich Diopside, Margarite and Chabazite were
identified only in the creep GPC sample rather than the control GPC. Although pore
refinement is found in both GPC and OPCC in general, GPC creep sample shows a
great replacement of pores above 2 nm to below 2 nm. This suggests that more creep
movement would be caused by the more amount of amorphous phases in GPC,
which causes high creep in long-term. Moreover, the finer pore network of GPC
259
would slow down the microdiffusion of pore solution and cause low creep rate in
short-term.
Chapter 6 presents a creep test method developed for flexural creep of geopolymer
and OPC pastes and mortars under various test conditions, such as temperature and
humidity. The test specimen is of a thin plate of 10 mm thick, 50 mm wide and 350
mm long, and is called “mini-beam”. It is supported over a span of 300 mm and two
equal point sustained loads are applied on the mini-beam at 100 mm apart in the
middle, as a configuration of four-point bending. The long-term deflection is
recorded to calculate the creep coefficient. Besides, cubic specimens of 50 mm was
cast for compressive strength and it is found that the geopolymer specimens show
higher strength at age of 1 day than OPC specimens, while their 28-day strengths are
similar. However, the flexural capacity of geopolymer specimens is lower than that
OPCP at ages of 1 day and 28 days. This would be due to geopolymer specimens
suffering more from the surface microcracking when subjected to drying conditions.
In general, the flexural capacity of mortar specimens is higher than that of the paste
specimens. This is because the bond strength of sand and paste had more influence
on the strength of mortar than the inherent tensile strength of paste. In the mini-beam
test, the GPP specimens exhibited high creep response than the OPCP specimens
under drying conditions, although both specimens showed similar creep coefficient
under sealed condition. This indicates drying has a significant impact of creep of
geopolymer. Drying also caused a counter displacement on the geopolymer mini-
beams due to differential shrinkage across the depth of the mini-beam. This would be
due to more drying shrinkage on the bottom side, caused by microcracks induced by
tensile stress. The increase in creep response of GPP specimens when tested at 50 °C
shows that geopolymer would be dominated more by temperature dependent flow
than by moisture diffusion under stress. A creep model for the mini-beam was
developed based on the pore characteristic and the viscoelastic behaviour of GPP and
OPCP. It shows that the fine pore network of geopolymer causes lower creep
development of GPP in the early age in term of hydrodynamic creep behaviour.
However, the viscoelastic creep behaviour of GPP allows more creep response in the
later ages.
260
Chapter 7 shows the test results of two types of full-scale GPC beams under long-
term sustained loads. The deflection was monitored for the investigation of long-term
performance. Beam type 1 is designed for normal reinforced concrete beam and
beam type 2 is designed for composite concrete floor. When compared with the
OPCC beams generated from the computer stimulation, the long-term deflection of
GPC beams is higher than that of the virtual OPCC beams of similar reinforcement
details. This result is due to lower stiffness, and higher creep and shrinkage effects of
GPC. Furthermore, it is found that the GPC was influenced by the seasonal change in
the local environment, which is not considered in the stimulation of virtual OPCC
beams. More increase in creep during the summer season confirms the effect of
temperature on creep of GPC. Although the long-term deflection of GPC beams is
high, it can be reduced by choice of compression reinforcement and loading at later
ages. The long-term deflection of the GPC beams can be predicted by using the
rational method for OPCC. Among them, EMM and AEMM provide better
prediction than RCM, because RCM does not include delayed creep recovery and
causes overestimation. However, the accuracy of these methods relies on the proper
input parameters of material property, including elastic modulus, creep coefficient
and shrinkage strain, etc.
Chapter 8 presents the investigation of thermal creep behaviour of GPP and OPCP up
to 100 °C. It was found that the compressive strength of both GPP and OPCP
decreased with increasing temperature, but the loss of strength was more in GPP. The
compressive strength of GPP and OPCP reduced approximately 40% and 22% at 100
°C relative to the strength at ambient temperature, respectively. This would be due to
thermal damage and cracks subjected to simultaneous drying and heating. The results
of 1-day thermal creep test show that the specific creep of GPP increased with
increasing temperature, but such increase was larger than the observation in OPCP
under similar condition, especially at temperatures of 80 °C and 100 °C. Compared
with specific creep at ambient temperature, the specific creep of GPP increased by
640% at 100 °C, while that of OPCP increased by 280% at 100 °C. This increase in
creep response is contributed by a number of factors involving change of viscosity of
pore solution, accelerated slippage of gels, thermal damage, increase in drying creep,
and transitional thermal creep (TTC). Further thermal test also shows that the TTC
261
strain per unit stress of GPP increased significantly by 145 microstrains from 80 °C
to 95 °C, but that of OPCP only increased by 30 microstrains. Such increases in TTC
and thermal creep would be due to change of microstructure simultaneous loading
and heating, as XRD analysis revealed that more crystalline phases were observed in
GPP samples tested above 80 °C, while there was no major phase alternation in
OPCP.
9.2 Conclusions
1. The slag-based GPC in this research shows suitable fresh and harden
properties for structural application. It is possible to apply GPC on concrete
structures.
2. The creep of slag-based GPC is higher than that of OPCC in similar mix
design. However, it can be reduced by isolation to eliminate drying effect,
and loading at more mature age. Therefore, the construction method and
sequences should be change for GPC structures.
3. The construction method of GPC can influence the long-term properties of
GPC, since the plant mix GPC shows higher creep than the laboratory mix
GPC. This would be due to the additional water in-situ which increases the
w/b. This results in change of properties. Therefore, the actual w/b should be
considered in the investigation of creep behaviour.
4. Creep models for GPC were proposed based on the current creep models for
OPCC. The factors of strength, age of loading and relative humidity are
considered for the success of these models. However, these creep model
require further improvement based on more creep data on GPC.
5. The GPC beam tests show that GPC structures is feasible. It can be adequate
in strength and excellence in ductility. Although the creep of GPC is high, the
deflection of GPC can be controlled by careful reinforcement detailing, and
protective layer.
6. The mini-beam test can be used to investigate the creep behaviour of paste
and mortar specimens under various test conditions. Results show that the
creep of geopolymer specimens are affected more by drying and high
262
temperature than that of OPC specimens. This would be due to microcracks
and microstructure alternation.
7. The GPC system exhibits more creep than OPC-based system because the
former contains more amorphous phases, and it would cause more visco-flow
behaviour. The finer pore network of GPC results in lower permeability,
which slows down the mircodiffusion of pore solution from the hindered
adsorbed layers. Therefore, GPC shows a lower creep rate in early stage of
loading than OPCC.
8. The creep model for mini-beam confirms the relationship of microstructure
(pore characteristic and gel phases) and creep behaviour of GPP and OPCP,
which is contributed by hydrodynamic creep function and viscoelastic creep
function. The hydrodynamic creep function affects the short-term creep
behaviour, while the viscoelastic creep function affects the creep behaviour at
later ages.
9. The significant increase in creep of GPP above 80 °C, which was not
observed in OPCP, indicates the temperature dependency in creep of GPC.
This would be due to the phase alternation of GPP subjected to high
temperature, which changes the microstructure and allow more visco-flow
behaviour in GPP.
9.3 Recommendations for future research
1. The author recommends more creep tests to be carried on different mix

design of GPC and various test conditions in order to expand the database of
creep of GPC.
2. GPC produced from the industry plant should be tested for long-term creep
and shrinkage properties, since it can be different from the results of the test
specimens produced in the laboratory.
3. Due to the finding of continuous creep development of GPC in later age,
creep test should be carried out for more than 1 year. This will provide
valuable data for development of creep model for GPC.
263
4. Mini-beam can be done in specimens of larger dimensions, which would
minimise the effect of differential shrinkage.
5. Creep test can be carried out on the synthetic system to identify the creep
behaviour of different phases in GPC and OPCC.
6. Accelerated creep test can be developed at temperature between 50 °C and 80
°C. Further works should be carried out to develop the correlation between
the creep tests at high temperature and at standard test temperature, 23 °C.
264
References
1. Neville, A.M. (1997), Properties of concrete, 4th ed, New York, John Wiley
& Sons, Inc.
2. Duxson, P., Provis, J.L., Lukey, G.C., and van Deventer, J.S.J. (2007), "The
role of inorganic polymer technology in the development of ‘green
concrete’", Cement and Concrete Research, 37(12), pp. 1590-1597.
3. Douglas, E., Bilodeau, A., Brandstetr, J., and Malhotra, V.M. (1991), "Alkali
activated ground granulated blast-furnace slag concrete: Preliminary
investigation", Cement and Concrete Research, 21(1), pp. 101-108.
4. Byfors, K., et al. (1989), "Durability of concrete made with alkali activated
slag", in 3rd International Conference Proceedings. Fly Ash, Silica Fume,
Slag, and Natural Pozzolans in Concrete, Malhotra, V.M., Editor, Trondheim,
Norway, pp. 1429-1466.
5. Douglas, E., Bilodeau, A., and Malhotra, V.M. (1992), "Properties and
durability of alkali-activated slag concrete", ACI Materials Journal, 89(5),
pp. 509-516.
6. Bažant, Z.P. (2001), "Prediction of concrete creep and shrinkage: past,
present and future", Nuclear Engineering and Design, 203(1), pp. 27-38.
7. Vandamme, M. and Ulm, F.J. (2009), "Nanogranular origin of concrete
creep", Proceedings of the National Academy of Sciences of the United States
of America, 106(26), pp. 10552-10557.
8. Acker, P. and Ulm, F.J. (2001), "Creep and shrinkage of concrete: Physical
origins and practical measurements", Nuclear Engineering and Design,
203(2-3), pp. 143-158.
9. Gilbert, R.I. (2001), "Shrinkage, cracking and decflection-the serviceability
of concrete structures", Electronic Journal of Structural Engineering, 1(1),
pp. 2-14.
10. Gilbert, R.I. and Ranzi, G. (2011), Time-Dependent Behaviour of Concrete
Structures, London, Spon Press.
11. Bažant, Z.P. and Murphy, W.P. (1995), "Creep and shrinkage prediction
model for analysis and design of concrete structures - model B3", Materiaux
et constructions, 28(180), pp. 357-365.
12. Bažant, Z.P., Hubler, M.H., and Yu, Q. (2011), "Pervasiveness of excessive
segmental bridge deflections: Wake-up call for creep", ACI Structural
Journal, 108(6), pp. 766-774.
13. Mindess, S., Young, J.F., and Darwin, D. (2003), Concrete, 2nd ed, Upper
Saddle River, Pearson Education, Inc.
14. Davidovits, J. (1991), "Geopolymers - Inorganic polymeric new materials",
Journal of Thermal Analysis, 37(8), pp. 1633-1656.
15. Bye, G. (1999), Portland cement, 2nd ed, ed. Livesey, P., London, ICE
Publishing.
16. Warner, R.F., Rangan, B.V., Hall, A.S., and Faulkes, K.A. (1999), Concrete
structures, South Melbourne, Addison Wesley Longman Australia Pty
Limited.
17. Duxson, P., et al. (2007), "Geopolymer technology: The current state of the
art", Journal of Materials Science, 42(9), pp. 2917-2933.
265
18. Kühl, H. (1908), "Slag cement and process of making the same", US Patents
900936.
19. Palomo, A., Grutzeck, M.W., and Blanco, M.T. (1999), "Alkali-activated fly
ashes: A cement for the future", Cement and Concrete Research, 29(8), pp.
1323-1329.
20. Roy, D.M. (1999), "Alkali-activated cements: Opportunities and challenges",
Cement and Concrete Research, 29(2), pp. 249-254.
21. Alonso, S. and Palomo, A. (2001), "Alkaline activation of metakaolin and
calcium hydroxide mixtures: influence of temperature, activator
concentration and solids ratio", Materials Letters, 47(1–2), pp. 55-62.
22. Rashad, A.M. (2013), "Metakaolin as cementitious material: History, scours,
production and composition – A comprehensive overview", Construction and
Building Materials, 41, pp. 303-318.
23. Shi, C. and Day, R.L. (1996), "Some factors affecting early hydration of
alkali-slag cements", Cement and Concrete Research, 26(3), pp. 439-447.
24. Fernández-Jiménez, A., Palomo, J.G., and Puertas, F. (1999), "Alkali-
activated slag mortars: Mechanical strength behaviour", Cement and
Concrete Research, 29(8), pp. 1313-1321.
25. Bakharev, T., Sanjayan, J.G., and Cheng, Y.-B. (1999), "Alkali activation of
Australian slag cements", Cement and Concrete Research, 29(1), pp. 113-
120.
26. Fernández-Jiménez, A. and Palomo, A. (2005), "Composition and
microstructure of alkali activated fly ash binder: Effect of the activator",
27. Collins, F.G. and Sanjayan, J.G. (1999), "Workability and mechanical
properties of alkali activated slag concrete", Cement and Concrete Research,
29(3), pp. 455-458.
28. Cheng, T.W. and Chiu, J.P. (2003), "Fire-resistant geopolymer produce by
granulated blast furnace slag", Minerals Engineering, 16(3), pp. 205-210.
29. Kong, D.L.Y., Sanjayan, J.G., and Sagoe-Crentsil, K. (2007), "Comparative
performance of geopolymers made with metakaolin and fly ash after
exposure to elevated temperatures", Cement and Concrete Research, 37(12),
pp. 1583-1589.
30. Kong, D.L.Y., Sanjayan, J.G., and Sagoe-Crentsil, K. (2008), "Factors
affecting the performance of metakaolin geopolymers exposed to elevated
temperatures", Journal of Materials Science, 43(3), pp. 824-831.
31. Bakharev, T., Sanjayan, J.G., and Cheng, Y.B. (2002), "Sulfate attack on
alkali-activated slag concrete", Cement and Concrete Research, 32(2), pp.
211-216.
32. Ismail, I., et al. (2013), "Influence of fly ash on the water and chloride
permeability of alkali-activated slag mortars and concretes", Construction
and Building Materials, 48, pp. 1187-1201.
33. Collins, F. and Sanjayan, J.G. (1998), "Early age strength and workability of
slag pastes activated by NaOH and Na2Co3", Cement and Concrete
Research, 28(5), pp. 655-664.
34. Collins, F. and Sanjayan, J.G. (2002), "Development of novel alkali activated
slag (AAS) binders to achieve high early strength concrete for construction
use", Australian Civil Engineering Transactions, 44, pp. 91-102.
266
35. Puertas, F., Martínez-Ramírez, S., Alonso, S., and Vázquez, T. (2000),
"Alkali-activated fly ash/slag cements. Strength behaviour and hydration
products", Cement and Concrete Research, 30(10), pp. 1625-1632.
36. Adam, A.A. and Horianto (2014), "The effect of temperature and duration of
curing on the strength of fly ash based geopolymer mortar", in Procedia
Engineering,
37. Abdulkareem, O.A., Mustafa Al Bakri, A.M., Kamarudin, H., and Khairul
Nizar, I. (2012), "The influence of curing periods on the compressive strength
of fly ash-based geopolymer at different aging times", in Advanced Materials
Research, pp. 512-516.
38. Bakharev, T., Sanjayan, J.G., and Cheng, Y.B. (1999), "Effect of elevated
temperature curing on properties of alkali-activated slag concrete", Cement
and Concrete Research, 29(10), pp. 1619-1625.
39. Bakharev, T., Sanjayan, J.G., and Cheng, Y.B. (2000), "Effect of admixtures
on properties of alkali-activated slag concrete", Cement and Concrete
Research, 30(9), pp. 1367-1374.
40. Roy, D.M. and Idorn, G.M. (1982), "Hydration, structure, and properties of
blast furnace slag cements, mortar, and concrete", J AM CONCR INST, V
79(N 6), pp. 444-457.
41. Huntzinger, D.N. and Eatmon, T.D. (2009), "A life-cycle assessment of
Portland cement manufacturing: comparing the traditional process with
alternative technologies", Journal of Cleaner Production, 17(7), pp. 668-675.
42. Bougara, A., Lynsdale, C., and Milestone, N.B. (2010), "Reactivity and
performance of blastfurnace slags of differing origin", Cement and Concrete
Composites, 32(4), pp. 319-324.
43. Barnett, S.J., Soutsos, M.N., Millard, S.G., and Bungey, J.H. (2006),
"Strength development of mortars containing ground granulated blast-furnace
slag: Effect of curing temperature and determination of apparent activation
energies", Cement and Concrete Research, 36(3), pp. 434-440.
44. Miura, T. and Iwaki, I. (2000), "Strength development of concrete
incorporating high levels of ground granulated blast-furnace slag at low
temperatures", ACI Materials Journal, 97(1), pp. 66-70.
45. Malhotra, V.M., Zhang, M.H., Read, P.H., and Ryell, J. (2000), "Long-term
mechanical properties and durability characteristics of high-strength/high-
performance concrete incorporating supplementary cementing materials
under outdoor exposure conditions", ACI Materials Journal, 97(5), pp. 518-
525.
46. Sanjayan, J.G. and Sioulas, B. (2000), "Strength of slag-cement concrete
cured in place and in other conditions", ACI Materials Journal, 97(5), pp.
603-611.
47. Collins, F. and Sanjayan, J.G. (1999), "Strength and shrinkage properties of
alkali-activated slag concrete placed into a large column", Cement and
48. Mehta, P.K. and Monteiro, P.J.M. (2006), Concrete : microstructure,
properties, and materials, New York, McGraw-Hill.
49. Feldman, R.F. and Sereda, P.J. (1970), "A new model for hydrated Portland
cement and its practical implications", Engineering Journal, 53(8/9), pp. 53-
59.
267
50. Richardson, I.G. (2008), "The calcium silicate hydrates", Cement and
51. Myers, R.J., Bernal, S.A., San Nicolas, R., and Provis, J.L. (2013),
"Generalized structural description of calcium-sodium aluminosilicate
hydrate gels: The cross-linked substituted tobermorite model", Langmuir,
29(17), pp. 5294-5306.
52. Garcia-Lodeiro, I., Palomo, A., Fernández-Jiménez, A., and Macphee, D.E.
(2011), "Compatibility studies between N-A-S-H and C-A-S-H gels. Study in
the ternary diagram Na2O–CaO–Al2O3–SiO2–H2O", Cement and Concrete
Research, 41(9), pp. 923-931.
53. Criado, M., Fernández-Jiménez, A., and Palomo, A. (2007), "Alkali
activation of fly ash: Effect of the SiO2/Na2O ratio. Part I: FTIR study",
Microporous and Mesoporous Materials, 106(1-3), pp. 180-191.
54. Criado, M., et al. (2008), "Effect of the SiO2/Na2O ratio on the alkali
activation of fly ash. Part II: 29Si MAS-NMR Survey", Microporous and
Mesoporous Materials, 109(1–3), pp. 525-534.
55. Puertas, F., et al. (2011), "A model for the C-A-S-H gel formed in alkali-
activated slag cements", Journal of the European Ceramic Society, 31(12),
pp. 2043-2056.
56. van Deventer, J.S., et al. (2015), "Microstructure and durability of alkali-
activated materials as key parameters for standardization", Journal of
Sustainable Cement-Based Materials, 4(2), pp. 116-128.
57. Garboczi, E.J. (1990), "Permeability, diffusivity, and microstructural
parameters: A critical review", Cement and Concrete Research, 20(4), pp.
591-601.
58. Lu, S., Landis, E.N., and Keane, D.T. (2006), "X-ray microtomographic
studies of pore structure and permeability in Portland cement concrete",
Materials and Structures/Materiaux et Constructions, 39(290), pp. 611-620.
59. Häkkinen, T. (1993), "The influence of slag content on the microstructure,
permeability and mechanical properties of concrete Part 1 Microstructural
studies and basic mechanical properties", Cement and Concrete Research,
23(2), pp. 407-421.
60. Collins, F. and Sanjayan, J.G. (2000), "Effect of pore size distribution on
drying shrinking of alkali-activated slag concrete", Cement and Concrete
Research, 30(9), pp. 1401-1406.
61. Häkkinen, T. (1993), "The influence of slag content on the microstructure,
permeability and mechanical properties of concrete. Part 2 technical
properties and theoretical examinations", Cement and Concrete Research,
23(3), pp. 518-530.
62. Mokhtarzadeh, A. and French, C. (2000), "Time-dependent properties of
high-strength concrete with consideration for precast applications", ACI
Materials Journal, 97(3), pp. 263-271.
63. Aligizaki, K.K. (2006), Pore structure of cement-based materials, New York,
Taylor & Franics.
64. Jennings, H.M., et al. (2008), "Characterization and modeling of pores and
surfaces in cement paste: Correlations to processing and properties", Journal
of Advanced Concrete Technology, 6(1), pp. 5-29.
268
65. Beaudoin, J.J., Feldman, R.F., and Tumidajski, P.J. (1994), "Pore structure of
hardened portland cement pastes and its influence on properties", Advanced
Cement Based Materials, 1(5), pp. 224-236.
66. Rößler, M. and Odler, I. (1985), "Investigations on the relationship between
porosity, structure and strength of hydrated portland cement pastes I. Effect
of porosity", Cement and Concrete Research, 15(2), pp. 320-330.
67. Shi, C. (1996), "Strength, pore structure and permeability of alkali-activated
slag mortars", Cement and Concrete Research, 26(12), pp. 1789-1799.
68. Aydın, S. and Baradan, B. (2012), "Mechanical and microstructural
properties of heat cured alkali-activated slag mortars", Materials & Design,
35(0), pp. 374-383.
69. Jambunathan, N., et al. (2013), "The role of alumina on performance of
alkali-activated slag paste exposed to 50°C", Cement and Concrete Research,
54(0), pp. 143-150.
70. Collins, F. and Sanjayan, J.G. (2001), "Microcracking and strength
development of alkali activated slag concrete", Cement and Concrete
Composites, 23(4–5), pp. 345-352.
71. Brough, A.R. and Atkinson, A. (2002), "Sodium silicate-based, alkali-
activated slag mortars: Part I. Strength, hydration and microstructure",
72. San Nicolas, R. and Provis, J.L. (2015), "The interfacial Transition Zone in
alkali-activated slag Mortars", Frontiers in Materials, 2, pp. 70.
73. Dellinghausen, L.M., Gastaldini, A.L.G., Vanzin, F.J., and Veiga, K.K.
(2012), "Total shrinkage, oxygen permeability, and chloride ion penetration
in concrete made with white Portland cement and blast-furnace slag",
Construction and Building Materials, 37(0), pp. 652-659.
74. Bakharev, T., Sanjayan, J.G., and Cheng, Y.B. (2003), "Resistance of alkali-
activated slag concrete to acid attack", Cement and Concrete Research,
33(10), pp. 1607-1611.
75. Wang, S.-D., Pu, X.-C., Scrivener, K.L., and Pratt, P.L. (1995), "Alkali-
activated slag cement and concrete: a review of properties and problems",
Advances in Cement Research, 7(27), pp. 93-102.
76. Roy, D.M., Jiang, W., and Silsbee, M.R. (2000), "Chloride diffusion in
ordinary, blended, and alkali-activated cement pastes and its relation to other
properties", Cement and Concrete Research, 30(12), pp. 1879-1884.
activated slag concrete to carbonation", Cement and Concrete Research,
31(9), pp. 1277-1283.
78. Bernal, S.A., et al. (2014), "Durability and testing – degradation via mass
transport", in Alkali activated materials, Provis, J.L. and Van Deventer, J.S.J.,
Editors, Springer, London.
activated slag concrete to alkali–aggregate reaction", Cement and Concrete
Research, 31(2), pp. 331-334.
80. Fernández-Jiménez, A. and Puertas, F. (2002), "The alkali–silica reaction in
alkali-activated granulated slag mortars with reactive aggregate", Cement and
269
81. Kong, D.L.Y. and Sanjayan, J.G. (2010), "Effect of elevated temperatures on
geopolymer paste, mortar and concrete", Cement and Concrete Research,
40(2), pp. 334-339.
82. Zhao, R. and Sanjayan, J.G. (2011), "Geopolymer and Portland cement
concretes in simulated fire", Magazine of Concrete Research, 63(3), pp. 163-
173.
83. Pan, Z., Sanjayan, J.G., and Kong, D.L.Y. (2012), "Effect of aggregate size
on spalling of geopolymer and Portland cement concretes subjected to
elevated temperatures", Construction and Building Materials, 36(0), pp. 365-
372.
84. Fu, Y., Cai, L., and Yonggen, W. (2011), "Freeze-thaw cycle test and damage
mechanics models of alkali-activated slag concrete", Construction and
Building Materials, 25(7), pp. 3144-3148.
85. Powers, T.C., Mann, H., and Copeland, L.E. (1958), "The flow of water in
hardened-portland cement paste", Highway Research Board Special Report,
(40).
86. Olson, R.A., Neubauer, C.M., and Jennings, H.M. (1997), "Damage to the
pore structure of hardened portland cement paste by mercury intrusion",
Journal of the American Ceramic Society, 80(9), pp. 2454-2458.
87. Neithalath, N., Sumanasooriya, M.S., and Deo, O. (2010), "Characterizing
pore volume, sizes, and connectivity in pervious concretes for permeability
prediction", Materials Characterization, 61(8), pp. 802-813.
88. Hu, J. and Stroeven, P. (2005), "Size characterisation of pore structure for
estimating transport properties of cement paste", Heron, 50(1), pp. 41-54.
89. Kondraivendhan, B., Divsholi, B.S., and Teng, S. (2013), "Estimation of
strength, permeability and hydraulic diffusivity of pozzolana blended
concrete through pore size distribution", Journal of Advanced Concrete
Technology, 11(9), pp. 230-237.
90. Pradhan, B., Nagesh, M., and Bhattacharjee, B. (2005), "Prediction of the
hydraulic diffusivity from pore size distribution of concrete", Cement and
91. Garci Juenger, M.C. and Jennings, H.M. (2001), "The use of nitrogen
adsorption to assess the microstructure of cement paste", Cement and
92. Feldman, R.F. (1987), "Diffusion measurements in cement paste by water
replacement using Propan-2-OL", Cement and Concrete Research, 17(4), pp.
602-612.
93. Hughes, D.C. (1985), "Pore structure and permeability of hardened cement
paste", Magazine of Concrete Research, 37(133), pp. 227-233.
94. Holly, J., Hampton, D., and Thomas, M.D.A. (1993), "Modelling
relationships between permeability and cement paste pore microstructures",
95. Aït-Mokhtar, A., Amiri, O., Dumargue, P., and Sammartino, S. (2002), "A
new model to calculate water permeability of cement-based materials from
MIP results", Advances in Cement Research, 14(2), pp. 43-49.
96. Amiri, O., Aït-Mokhtar, A., and Sarhani, M. (2005), "Tri-dimensional
modelling of cementitious materials permeability from polymodal pore size
distribution obtained by mercury intrusion porosimetry tests", Advances in
Cement Research, 17(1), pp. 39-45.
270
97. Atzeni, C., Pia, G., and Sanna, U. (2010), "A geometrical fractal model for
the porosity and permeability of hydraulic cement pastes", Construction and
Building Materials, 24(10), pp. 1843-1847.
98. Bágel, L. and Živica, V. (1997), "Relationship between pore structure and
permeability of hardened cement mortars: On the choice of effective pore
structure parameter", Cement and Concrete Research, 27(8), pp. 1225-1235.
99. Luping, T. and Nilsson, L.O. (1992), "A study of the quantitative relationship
between permeability and pore size distribution of hardened cement pastes",
100. Nyame, B.K. and Illston, J.M. (1981), "Relationships between permeability
and pore structure of hardened cement paste", Magazine of Concrete
Research, 33(116), pp. 139-146.
101. Kumar, R. and Bhattacharjee, B. (2004), "Assessment of permeation quality
of concrete through mercury intrusion porosimetry", Cement and Concrete
Research, 34(2), pp. 321-328.
102. Bažant, Z.P. and Najjar, L.J. (1971), "Drying of concrete as a nonlinear
diffusion problem", Cement and Concrete Research, 1(5), pp. 461-473.
103. Ulm, F.J. and Coussy, O. (1995), "Modeling of thermochemomechanical
couplings of concrete at early ages", Journal of Engineering Mechanics,
121(7), pp. 785-794.
104. Xi, Y., Bažant, Z.P., Molina, L., and Jennings, H.M. (1994), "Moisture
diffusion in cementitious materials Moisture capacity and diffusivity",
Advanced Cement Based Materials, 1(6), pp. 258-266.
105. Bažant, Z.P. and Chern, J.C. (1985), "Concrete creep at variable humidity:
constitutive law and mechanism", Materials and Structures, 18(1), pp. 1-20.
106. de Sa, C., Benboudjema, F., Thiery, M., and Sicard, J. (2008), "Analysis of
microcracking induced by differential drying shrinkage", Cement and
Concrete Composites, 30(10), pp. 947-956.
107. Bissonnette, B., Pierre, P., and Pigeon, M. (1999), "Influence of key
parameters on drying shrinkage of cementitious materials", Cement and
108. Langmuir, I. (1918), "The adsorption of gases on plane surfaces of glass,
mica and platinum", The Journal of the American Chemical Society, 40(9),
pp. 1361-1403.
109. Brunauer, S., Emmett, P.H., and Teller, E. (1938), "Adsorption of gases in
multimolecular layers", Journal of the American Chemical Society, 60(2), pp.
309-319.
110. Thomson, W.T. (1871), "On the equilibrium of vapour at a curved surface of
liquid ", Philosophical Magazine and Journal of Science, 42(282), pp. 448-
452.
111. Barrett, E.P., Joyner, L.G., and Halenda, P.P. (1951), "The determination of
pore volume and area distributions in porous substances. I. Computations
from nitrogen isotherms", Journal of the American Chemical Society, 73(1),
pp. 373-380.
112. Cranston, R.W. and Inkley, F.A. (1957), "The determination of pore
structures from nitrogen adsorption isotherms", in Advances in Catalysis, pp.
143-154.
271
113. Dollimore, D. and Heal, G.R. (1973), "Pore size distribution in a system
considered as an ordered packing of spherical particles", Journal of Colloid
And Interface Science, 42(2), pp. 233-249.
114. Mikhail, R.S., Brunauer, S., and Bodor, E.E. (1968), "Investigations of a
complete pore structure analysis. I. Analysis of micropores", Journal of
Colloid And Interface Science, 26(1), pp. 45-53.
115. Bažant, Z.P. (1975), "Theory of creep and shrinkage in concrete structures: A
precis of recent developments", Mechanics today, 2, pp. 1-93.
116. Ruiz, M.F., Muttoni, A., and Gambarova, P.G. (2007), "Relationship between
nonlinear creep and cracking of concrete under uniaxial compression",
Journal of Advanced Concrete Technology, 5(3), pp. 383-393.
117. Bažant, Z.P. and Kim, S.S. (1978), "Can the creep curves for different
loading ages diverge?", Cement and Concrete Research, 8(5), pp. 601-611.
118. Niyogi, A.K., Hsu, P., and Meyers, B.L. (1973), "The influence of age at time
of loading on basic and drying creep", Cement and Concrete Research, 3(5),
pp. 633-644.
119. De Schutter, G. (1999), "Degree of hydration based Kelvin model for the
basic creep of early age concrete", Materials and Structures/Materiaux et
Constructions, 32(218), pp. 260-265.
120. Granger, L.P. and Bažant, Z.P. (1995), "Effect of composition on basic creep
of concrete and cement paste", Journal of engineering mechanics, 121(11),
pp. 1261-1270.
121. Pickett, G. (1942), "The effect of change in moisture content on the creep of
concrete under a sustained load", ACI Journal Proceedings, 38, pp. 333-356.
122. Wittmann, F.H. and Roelfstra, P.E. (1980), "Total deformation of drying
loaded concrete", Cement and Concrete Research, 10(5), pp. 601-610.
123. Bažant, Z.P. and Yunping, X.I. (1994), "Drying creep of concrete:
constitutive model and new experiments separating its mechanisms",
Materials and Structures, 27(1), pp. 3-14.
124. Bažant, Z.P., Havlásek, P., and Jirásek, M. (2014), "Microprestress-
solidification theory: Modeling of size effect on drying creep", in
Computational Modelling of Concrete Structures - Proceedings of EURO-C
2014,
125. Alexander, K.M., Wardlaw, J., and Ivanusec, I. (1986), "A 4:1 range in
concrete creep when cement SO3 content, curing temperature and fly ash
content are varied", Cement and Concrete Research, 16(2), pp. 173-180.
126. Vidal, T., Sellier, A., Ladaoui, W., and Bourbon, X. (2015), "Effect of
Temperature on the Basic Creep of High-Performance Concretes Heated
between 20 and 80°C", Journal of Materials in Civil Engineering, 27(7), pp.
B4014002.
127. Nasser, K.W. and Neville, A.M. (1966), "Creep of concrete at temperatures
above normal", Nuclear Engineering and Design, 4(1), pp. 90-96.
128. Neville, A.M. (1960), "Recovery of creep and observations on the mechanism
of creep of concrete", Applied Scientific Research, 9(1), pp. 71-84.
129. Feldman, R.F. (1972), "Mechanism of creep of hydrated portland cement
paste", Cement and Concrete Research, 2(5), pp. 521-540.
130. Bažant, Z.P. (1972), "Thermodynamics of interacting continua with surfaces
and creep analysis of concrete structures", Nuclear Engineering and Design,
20(2), pp. 477-505.
272
131. Hope, B.B. and Brown, N.H. (1975), "A model for the creep of concrete",
132. Whiting, D.A., Detwiler, R.J., and Lagergren, E.S. (2000), "Cracking
tendency and drying shrinkage of silica fume concrete for bridge deck
applications", ACI Materials Journal, 97(1), pp. 71-77.
133. Altoubat, S.A. and Lange, D.A. (2001), "Creep, shrinkage, and cracking of
restrained concrete at early age", ACI Materials Journal, 98(4), pp. 323-331.
134. Bissonnette, B., Pigeon, M., and Vaysburd, A.M. (2007), "Tensile creep of
concrete: Study of its sensitivity to basic parameters", ACI Materials Journal,
104(4), pp. 360-368.
135. Rossi, P., Tailhan, J.-L., and Le Maou, F. (2013), "Comparison of concrete
creep in tension and in compression: Influence of concrete age at loading and
drying conditions", Cement and Concrete Research, 51(0), pp. 78-84.
136. Brooks, J.J. and Neville, A.M. (1977), "Comparison of creep, elasticity and
strength of concrete in tension and in compression", Magazine of Concrete
Research, 29(100), pp. 131-141.
137. Ranaivomanana, N., Multon, S., and Turatsinze, A. (2013a), "Tensile,
compressive and flexural basic creep of concrete at different stress levels",
138. Ranaivomanana, N., Multon, S., and Turatsinze, A. (2013b), "Basic creep of
concrete under compression, tension and bending", Construction and
139. Arango, S.E., Serna, P., Martí-Vargas, J.R., and García-Taengua, E. (2012),
"A test method to characterize flexural creep behaviour of pre-cracked FRC
specimens", Experimental Mechanics, 52(8), pp. 1067-1078.
140. Tailhan, J.L., et al. (2013), "Compressive, tensile and bending basic creep
behaviours related to the same concrete", Structural Concrete, 14(2), pp. 124-
130.
141. Hardjito, D., Wallah, S.E., Sumajouw, D.M.J., and Rangan, B.V. (2004), "On
the development of fly ash-based geopolymer concrete", ACI Materials
Journal, 101(6), pp. 467-472.
142. Wallah, S. and Rangan, B.V. (2006), "Low-calcium fly ash-based
geopolymer concrete: Long-term properties", Res. Report-GC2, Curtin
University, Australia. pp, pp. 76-80.
143. Sagoe-Crentsil, K., Brown, T., and Taylor, A. (2013), "Drying shrinkage and
creep performance of geopolymer concrete", Journal of Sustainable Cement-
Based Materials, 2(1), pp. 35-42.
144. Liu, H., Lu, Z., and Peng, Z. (2014), "Test research on prestressed beam of
inorganic polymer concrete", Materials and Structures.
145. Lee, N.P. (2007), "Creep and shrinkage of inorganic polymer concrete", in
BRANZ Study Report SR 175, BRANZ Ltd., Judgeford, New Zealand.
146. Castel, A., et al. (2016), "Creep and drying shrinkage of a blended slag and
low calcium fly ash geopolymer Concrete", Materials and
Structures/Materiaux et Constructions, 49(5), pp. 1619-1628.
147. Ridtirud, C., Chindaprasirt, P., and Pimraksa, K. (2011), "Factors affecting
the shrinkage of fly ash geopolymers", International Journal of Minerals
Metallurgy and Materials, 18(1), pp. 100-104.
148. Olivia, M. and Nikraz, H. (2012), "Properties of fly ash geopolymer concrete
designed by Taguchi method", Materials and Design, 36, pp. 191-198.
273
149. Collins, F. and Sanjayan, J.G. (1999), "Strength and shrinkage properties of
alkali-activated slag concrete containing porous coarse aggregate", Cement
150. Collins, F. and Sanjayan, J.G. (2000), "Cracking tendency of alkali-activated
slag concrete subjected to restrained shrinkage", Cement and Concrete
Research, 30(5), pp. 791-798.
151. Collins, F. and Sanjayan, J.G. (2000), "Numerical modeling of alkali-
activated slag concrete beams subjected to restrained shrinkage", ACI
152. Neto, A.A.M., Cincotto, M.A., and Repette, W. (2008), "Drying and
autogenous shrinkage of pastes and mortars with activated slag cement",
153. Lee, N.K., Jang, J.G., and Lee, H.K. (2014), "Shrinkage characteristics of
alkali-activated fly ash/slag paste and mortar at early ages", Cement and
Concrete Composites, 53, pp. 239-248.
154. Bilodeau, A. and Malhotra, V.M. (2000), "High-volume fly ash system:
Concrete solution for sustainable development", ACI Materials Journal,
97(1), pp. 41-48.
155. Atiş, C.D., Kiliç, A., and Sevim, U.K. (2004), "Strength and shrinkage
properties of mortar containing a nonstandard high-calcium fly ash", Cement
156. Yoo, S.-W., Koh, K.-T., Kwon, S.-J., and Park, S.G. (2013), "Analysis
technique for flexural behavior in RC beam considering autogenous
shrinkage effect", Construction and Building Materials, 47(0), pp. 560-568.
157. Brooks, J.J. and Megat Johari, M.A. (2001), "Effect of metakaolin on creep
and shrinkage of concrete", Cement and Concrete Composites, 23(6), pp.
495-502.
158. Chern, J.-C. and Chan, Y.-W. (1989), "Deformations of concretes made with
blast-furnace slag cement and ordinary portland cement", ACI Materials
Journal, 86(4), pp. 372-382.
159. Shariq, M., Prasad, J., and Abbas, H. (2016), "Creep and drying shrinkage of
concrete containing GGBFS", Cement and Concrete Composites, 68, pp. 35-
45.
160. Li, J. and Yao, Y. (2001), "A study on creep and drying shrinkage of high
performance concrete", Cement and Concrete Research, 31(8), pp. 1203-
1206.
161. Lim, S.N. and Wee, T.H. (2000), "Autogenous shrinkage of ground-
granulated blast-furnace slag concrete", ACI Materials Journal, 97(5), pp.
587-593.
162. Khoury, G.A., Grainger, B.N., and Sullivan, P.J.E. (1985), "Transient thermal
strain of concrete: literature review, conditions within specimen and
behaviour of individual constituents", Magazine of Concrete Research,
37(132), pp. 131-144.
163. Khoury, G.A., Grainger, B.N., and Sullivan, P.J.E. (1985), "Strain of concrete
dueing first heating to 600 degree C under load", Magazine of Concrete
Research, 37(133), pp. 195-215.
164. Dias, W.P.S., Khoury, G.A., and Sullivan, P.J.E. (1987), "Basic creep of
unsealed hardened cement paste at temperatures between 20 degree C and
725 degree C", Magazine of Concrete Research, 39(139), pp. 93-101.
274
165. Pan, Z., Sanjayan, J.G., and Collins, F. (2014), "Effect of transient creep on
compressive strength of geopolymer concrete for elevated temperature
exposure", Cement and Concrete Research, 56, pp. 182-189.
166. Bažant, Z.P. and Wu, S.T. (1973), "Dirichlet series creep function for aging
concrete", ASCE J Eng Mech Div, 99(EM2), pp. 367-387.
167. Bažant, Z.P. and Wu, S.T. (1974), "Rate-type creep law of aging concrete
based on maxwell chain", Matériaux et Constructions, 7(1), pp. 45-60.
168. Bažant, Z.P. and Osman, E. (1976), "Double power law for basic creep of
concrete", Matériaux et Constructions, 9(1), pp. 3-11.
169. Bažant, Z.P. and Chern, J.-C. (1984), "Double-power logarithmic law for
concrete creep", Cement and Concrete Research, 14(6), pp. 793-806.
170. Bažant, Z.P. and Prasannan, S. (1988), "Solidification theory for aging
creep", Cement and Concrete Research, 18(6), pp. 923-932.
171. Bažant, Z.P. and Prasannan, S. (1989), "Solidification theory for concrete
creep. I: Formulation", Journal of Engineering Mechanics, 115(8), pp. 1691-
1703.
172. Bažant, Z.P. and Prasannan, S. (1989), "Solidification theory for concrete
creep. II: Verification and application", Journal of Engineering Mechanics,
115(8), pp. 1704-1725.
173. Bažant, Z.P., Hauggaard, A.B., Baweja, S., and Ulm, F.J. (1997),
"Microprestress-solidification theory for concrete creep. I: Aging and drying
effects", Journal of Engineering Mechanics, 123(11), pp. 1188-1194.
174. Bažant, Z.P., Hauggaard, A.B., and Baweja, S. (1997), "Microprestress-
solidification theory for concrete creep. II: Algorithm and verification",
Journal of Engineering Mechanics, 123(11), pp. 1195-1201.
175. Wendner, R., Hubler, M.H., and Bažant, Z.P. (2013), "The B4 model for
multi-decade creep and shrinkage prediction", in 9th International
Conference on Creep, Shrinkage, and Durability Mechanics: A Tribute to
Zdenek P. Bazant, Cambridge, United States, 22-25 September 2013
176. Dönmez, A. and Bažant, Z.P. (2016), "Shape factors for concrete shrinkage
and drying creep in model B4 refined by nonlinear diffusion analysis",
Materials and Structures/Materiaux et Constructions, 49(11), pp. 4779-4784.
177. Brooks, J.J. and Neville, A.M. (1978), "Predicting long-term creep and
shrinkage from short-term tests", Magazine of Concrete Research, 30(103),
pp. 51-61.
178. Gardner, N.J. and Lockman, M.J. (2001), "Design provisions for drying
shrinkage and creep of normal-strength concrete", ACI Materials Journal,
98(2), pp. 159-167.
179. Gawin, D., Pesavento, F., and Schrefler, B.A. (2006), "Hygro-thermo-chemo-
mechanical modelling of concrete at early ages and beyond. Part II:
Shrinkage and creep of concrete", International Journal for Numerical
Methods in Engineering, 67(3), pp. 332-363.
180. Gawin, D., Pesavento, F., and Schrefler, B.A. (2007), "Modelling creep and
shrinkage of concrete by means of effective stresses", Materials and
Structures/Materiaux et Constructions, 40(6), pp. 579-591.
181. Standards Australia (2009), "AS3600-2009: Concrete structures", Sydney.
182. British Standards Institution (2004), "BS EN 1992-1-1:2004: Eurocode 2:
Design of concrete structures - part 1-1: General rules and rules for
buildings", London.
275
183. ACI Committee 209 (1994), Prediction of creep, shrinkage, and temperature
effects in concrete structures, American Concrete Institute.
184. The International Federation for Structural Concrete (2013), The fib Model
Code for Concrete Structures 2010, Berlin, Ernst & Sohn.
185. Bažant, Z.P. and Li, G.H. (2008), "Unbiased statistical comparison of creep
and shrinkage prediction models", ACI Materials Journal, 105(6), pp. 610-
621.
186. Dischinger, F. (1937), "Untersuchungen über die Knicksicherheit, die
elastische Verformung und das Kriechen des Betons bei Bogenbrücken", Der
Bauingenieur, 18, pp. 487-520, 539-552, 595-621.
187. Rüsch, H., Jungwirth, D., and Hilsdorf, H.K. (1983), Creep and shrinkage -
Their effect on the behavior of concrete structures, New York, Springer-
Verlag.
188. Trost, H. (1967), "Implications of the superposition principle in creep and
relaxation problems for concrete and prestressed concrete", Beton und
stahlbeton (Berlin–Wilmersdorf), 62(10).
189. Bažant, Z.P. (1972), "Prediction of concrete creep effects using age-adjusted
effective modulus method", Journal of the American Concrete Institute, 69,
pp. 212-217.
190. Bažant, Z.P. and Kim, S.S. (1979), "Approximate relaxation function for
concrete", Journal of the Structural Division-Asce, 105(12), pp. 2695-2705.
191. Bažant, Z.P., Hubler, M.H., and Jirásek, M. (2013), "Improved estimation of
long-term relaxation function from compliance function of aging concrete",
Journal of Engineering Mechanics, 139(2), pp. 146-152.
192. Gilbert, R.I. (2013), "Time-dependent stiffness of cracked reinforced and
composite concrete slabs", in 11th International Conference on Modern
Building Materials, structures and Techniques, Vilnius, Lithuania, 16-17
May 2013
193. Au, F.T.K. and Si, X.T. (2011), "Accurate time-dependent analysis of
concrete bridges considering concrete creep, concrete shrinkage and cable
relaxation", Engineering Structures, 33(1), pp. 118-126.
194. Bažant, Z.P., Yu, Q., and Li, G.H. (2012), "Excessive long-time deflections
of prestressed box girders. I: Record-span Bridge in Palau and other
paradigms", Journal of Structural Engineering (United States), 138(6), pp.
676-686.
195. Arockiasamy, M., Butrieng, N., and Sivakumar, M. (2004), "State-of-the-art
of integral abutment bridges: Design and practice", Journal of Bridge
Engineering, 9(5), pp. 497-506.
196. Sharma, R.K., Maru, S., and Nagpal, A.K. (2009), "Effect of beam stiffness-
column reinforcement on creep and shrinkage behaviour of R.C frames",
Structural Design of Tall and Special Buildings, 18(3), pp. 327-339.
197. Bažant, Z.P. and Oh, B.H. (1984), "Deformation of progressively cracking
reinforced concrete beams", Journal of the American Concrete Institute,
81(3), pp. 268-278.
198. Gilbert, R.I. and Bradford, M.A. (1995), "Time-dependent behavior of
continuous composite beams at service loads", Journal of structural
engineering New York, N.Y., 121(2), pp. 319-327.
276
199. Fan, J., Nie, J., Li, Q., and Wang, H. (2010), "Long-term behavior of
composite beams under positive and negative bending. I: Experimental
study", Journal of Structural Engineering, 136(7), pp. 849-857.
200. Dezi, L., Leoni, G., and Tarantino, A.M. (1995), "Modified AAEM method
for composite beams with post-connected slab", Journal of structural
engineering New York, N.Y., 121(11), pp. 1726-1729.
201. Shariq, M., Prasad, J., and Abbas, H. (2013), "Long-term deflection of RC
beams containing GGBFS", Magazine of Concrete Research, 65(24), pp.
1441-1462.
202. Standards Australia (2010), "AS3972-2010: General purpose and blended
cements", Sydney.
203. Tennakoon, C., Nicolas, R.S., Sanjayan, J.G., and Shayan, A. (2016),
"Thermal effects of activators on the setting time and rate of workability loss
of geopolymers", Ceramics International, 42(16), pp. 19257-19268.
204. American Society for Testing and Materials (2014), "ASTM C136/C136M:
Standard test method fro sieve analysis of fine and coarse aggreagtes", United
States.
205. Standards Australia (2014), "AS1012.8-2014: Methods of testing concrete -
Method for making and curing concrete ", Sydney.
Determination of creep of concrete cylinders in compression", Sydney.
Determination of the drying shrinkage of concrete for samples prepared in the
field or in the laboratory", Sydney.
Compressive strength tests - Concrete, mortar and grout specimens ", Sydney.
Determination of the static chord modulus of elasticity and Poisson's ratio of
concrete specimens", Sydney.
Standard test method for compressive strength of hydraulic cement mortars
(Using 2-in. or [50-mm] Cube Specimens) ", United States.
Determination of the modulus of rupture", Sydney.
212. Materials, A.S.f.T.a. (2014), "ASTM C1698-09: Standard test method for
autogenous strain of cement paste and mortar", United States.
213. Materials, A.S.f.T.a. (2013), "ASTM C191-13: Standard test methods for
time of setting of hydraulic cement by Vicat needle", United States.
214. Bakharev, T. (2006), "Thermal behaviour of geopolymers prepared using
class F fly ash and elevated temperature curing", Cement and Concrete
Research, 36(6), pp. 1134-1147.
215. Ismail, I., et al. (2013), "Drying-induced changes in the structure of alkali-
activated pastes", Journal of Materials Science, 48(9), pp. 3566-3577.
216. Standards Australia (2014), "AS1012.3.1-2014: Methods of testing concrete -
Method 3.1: Determination of properties related to the consistency of
concrete-Slump test", Sydney.
217. Sarker, P.K., Haque, R., and Ramgolam, K.V. (2013), "Fracture behaviour of
heat cured fly ash based geopolymer concrete", Materials & Design, 44(0),
pp. 580-586.
277
218. Bernal, S.A., et al. (2014), "MgO content of slag controls phase evolution and
structural changes induced by accelerated carbonation in alkali-activated
binders", Cement and Concrete Research, 57, pp. 33-43.
219. Hou, D.W., Zhang, J., Chen, H.Y., and Liu, W. (2012), "Development of
strength and elastic modulus of concrete under moisture and drying curing
conditions", Shuili Xuebao/Journal of Hydraulic Engineering, 43(2), pp. 198-
208.
220. Smadi, M.M., Slate, F.O., and Nilson, A.H. (1987), "Shrinkage and creep of
high-, medium-, and low-strength concretes, including overloads", ACI
221. Khan, A.A., Cook, W.D., and Mitchell, D. (1997), "Creep, shrinkage, and
thermal strains in normal, medium, and high-strength concretes during
hydration", ACI Materials Journal, 94(2), pp. 156-163.
222. Wang, S.-D. and Scrivener, K.L. (1995), "Hydration products of alkali
activated slag cement", Cement and Concrete Research, 25(3), pp. 561-571.
223. Ismail, I., et al. (2014), "Modification of phase evolution in alkali-activated
blast furnace slag by the incorporation of fly ash", Cement and Concrete
Composites, 45, pp. 125-135.
224. Oh, J.E., et al. (2010), "The evolution of strength and crystalline phases for
alkali-activated ground blast furnace slag and fly ash-based geopolymers",
225. Lothenbach, B., Le Saout, G., Gallucci, E., and Scrivener, K. (2008),
"Influence of limestone on the hydration of Portland cements", Cement and
226. Johannesson, B. and Utgenannt, P. (2001), "Microstructural changes caused
by carbonation of cement mortar", Cement and Concrete Research, 31(6), pp.
925-931.
227. Papadakis, V.G., Vayenas, C.G., and Fardis, M.N. (1991), "Fundamental
modeling and experimental investigation of concrete carbonation", ACI
228. Parrott, L.J. and Killoh, D.C. (1989), "Carbonation in a 36 year old, in-situ
concrete", Cement and Concrete Research, 19(4), pp. 649-656.
229. Nishikawa, T., et al. (1992), "Decomposition of synthesized ettringite by
carbonation", Cement and Concrete Research, 22(1), pp. 6-14.
230. Xiantuo, C., Ruizhen, Z., and Xiaorong, C. (1994), "Kinetic study of
ettringite carbonation reaction", Cement and Concrete Research, 24(7), pp.
1383-1389.
231. Alarcon-Ruiz, L., Platret, G., Massieu, E., and Ehrlacher, A. (2005), "The use
of thermal analysis in assessing the effect of temperature on a cement paste",
232. Majchrzak-Kucȩba, I. and Nowak, W. (2004), "Thermal analysis of fly ash-
based zeolites", Journal of Thermal Analysis and Calorimetry, 77(1), pp.
125-131.
233. Haha, M.B., Lothenbach, B., Le Saout, G., and Winnefeld, F. (2011),
"Influence of slag chemistry on the hydration of alkali-activated blast-furnace
slag - Part I: Effect of MgO", Cement and Concrete Research, 41(9), pp. 955-
963.
278
234. Villain, G., Thiery, M., and Platret, G. (2007), "Measurement methods of
carbonation profiles in concrete: Thermogravimetry, chemical analysis and
gammadensimetry", Cement and Concrete Research, 37(8), pp. 1182-1192.
235. van Deventer, J.S.J., et al. (2014), "Microstructure and durability of alkali-
activated materials as key parameters for standardization", Journal of
Sustainable Cement-Based Materials, 4(2), pp. 116-128.
236. Snellings, R., Salze, A., and Scrivener, K.L. (2014), "Use of X-ray diffraction
to quantify amorphous supplementary cementitious materials in anhydrous
and hydrated blended cements", Cement and Concrete Research, 64, pp. 89-
98.
237. Scrivener, K.L., et al. (2004), "Quantitative study of Portland cement
hydration by X-ray diffraction/Rietveld analysis and independent methods",
238. Elsayed, H., et al. (2017), "Bioactive glass-ceramic scaffolds from novel
'inorganic gel casting' and sinter-crystallization", Materials, 10(2).
239. Rey, F., Fornés, V., and Rojo, J.M. (1992), "Thermal decomposition of
hydrotalcites. An infrared and nuclear magnetic resonance spectroscopic
study", Journal of the Chemical Society, Faraday Transactions, 88(15), pp.
2233-2238.
240. Gruskovnjak, A., et al. (2006), "Hydration of alkali-activated slag:
comparison with ordinary Portland cement", Advances in Cement Research,
18(3), pp. 119-128.
241. Rivas-Mercury, J.M., Pena, P., de Aza, A.H., and Turrillas, X. (2008),
"Dehydration of Ca3Al2(SiO4)y(OH)4(3-y) (0 < y < 0.176) studied by
neutron thermodiffractometry", Journal of the European Ceramic Society,
28(9), pp. 1737-1748.
242. Ramachandran, V.S. and Chun-Mei, Z. (1986), "Thermal analysis of the
3CaO · Al2O3-CaSO4 · 2H2O-CaCO3-H2O system", Thermochimica Acta,
106(C), pp. 273-282.
243. Földvári, M. (2011), Handbook of thermogravimetric system of minerals and
its use in geological practice, Bubapest, Hungary, the Geological Institute of
Hungary.
244. Gawin, D., Pesavento, F., and Schrefler, B.A. (2006), "Hygro-thermo-chemo-
mechanical modelling of concrete at early ages and beyond. Part I: Hydration
and hygro-thermal phenomena", International Journal for Numerical
Methods in Engineering, 67(3), pp. 299-331.
245. White, C.E., Daemen, L.L., Hartl, M., and Page, K. (2015), "Intrinsic
differences in atomic ordering of calcium (alumino)silicate hydrates in
conventional and alkali-activated cements", Cement and Concrete Research,
67, pp. 66-73.
246. Bažant, Z.P. (1972), "Thermodynamics of hindered adsorption and its
implications for hardened cement paste and concrete", Cement and Concrete
Research, 2(1), pp. 1-16.
247. Powers, T.C., Copeland, L., Hayes, J., and Mann, H. (1954), "Permeablity of
Portland Cement Paste", in ACI Journal Proceedings, ACI,
Standard test method for creep of concrete in compression", United States.
279
249. Scherer, G.W. (2000), "Measuring permeability of rigid materials by a beam-
bending method: I, Theory", Journal of the American Ceramic Society, 83(9),
pp. 2231-2239.
250. Vichit-Vadakan, W. and Scherer, G.W. (2002), "Measuring permeability of
rigid materials by a beam-bending method: III, cement paste", Journal of the
American Ceramic Society, 85(6), pp. 1537-1544.
251. Scherer, G.W. (2004), "Measuring permeability of rigid materials by a beam-
bending method: IV, transversely isotropic plate", Journal of the American
Ceramic Society, 87(8), pp. 1517-1524.
252. Valenza II, J. and Scherer, G.W. (2004), "Measuring permeability of rigid
materials by a beam-bending method: V, isotropic rectangular plates of
cement paste", Journal of the American Ceramic Society, 87(10), pp. 1927-
1931.
253. Altan, E. and Erdoğan, S.T. (2012), "Alkali activation of a slag at ambient
and elevated temperatures", Cement and Concrete Composites, 34(2), pp.
131-139.
254. Wang, J.A., Lubliner, J., and Monteiro, P.J.M. (1988), "Effect of ice
formation on the elastic moduli of cement paste and mortar", Cement and
255. Chen, C.T. and Ho, C.W. (2013), "Influence of cyclic humidity on
carbonation of concrete", Journal of Materials in Civil Engineering, 25(12),
pp. 1929-1935.
256. Thomas, T.C.H. and Floyd, O.S. (1963), "Tensile bond strength between
aggregate and cement paste or mortar", ACI Journal proceedings, 60(4), pp.
465-486.
257. Türker, H.T., et al. (2016), "Microstructural alteration of alkali activated slag
mortars depend on exposed high temperature level", Construction and
258. Kjellsen, K.O. (1996), "Heat curing and post-heat curing regimes of high-
performance concrete: Influence on microstructure and C-S-H composition",
259. Hobbs, D.W. (1971), "The dependence of the bulk modulus, Young's
modulus, creep, shrinkage and thermal expansion of concrete upon aggregate
volume concentration", Matériaux et Constructions, 4(2), pp. 107-114.
260. Bažant, Z.P. (1993), "Current status and advances in the theory of creep and
interaction with fracture", in Proceedings of the Fifth International RILEM
Symposium, Barcelona, Spain, E & FN SPON,
261. Winslow, D. and Liu, D. (1990), "The pore structure of paste in concrete",
262. Wong, H.S., Zobel, M., Buenfeld, N.R., and Zimmerman, R.W. (2009),
"Influence of the interfacial transition zone and microcracking on the
diffusivity, permeability and sorptivity of cement-based materials after
drying", Magazine of Concrete Research, 61(8), pp. 571-589.
263. Rovnaník, P., Bayer, P., and Rovnaníková, P. (2013), "Characterization of
alkali activated slag paste after exposure to high temperatures", Construction
and Building Materials, 47, pp. 1479-1487.
264. Kjellsen, K.O., Detwiler, R.J., and Gjørv, O.E. (1991), "Development of
microstructures in plain cement pastes hydrated at different temperatures",
280
265. Lothenbach, B., et al. (2007), "Effect of temperature on the pore solution,
microstructure and hydration products of Portland cement pastes", Cement
266. Scherer, G.W. (1992), "Bending of gel beams: method for characterizing
elastic properties and permeability", Journal of Non-Crystalline Solids,
142(C), pp. 18-35.
267. Biot, M.A. (1956), "Theory of propagation of elastic waves in a fluid -
Saturated porous solid. I. Low-frequency range", Journal of the Acoustical
Society of America, 28(2), pp. 168-178.
268. Johnson, D.L. (1982), "Elastodynamics of gels", The Journal of Chemical
Physics, 77(3), pp. 1531-1539.
269. Scherer, G.W. (1994), "Stress in aerogel during depressurization of
autoclave: I. theory", Journal of Sol-Gel Science and Technology, 3(2), pp.
127-139.
270. Pan, Z. and Sanjayan, J.G. (2010), "Stress-strain behaviour and abrupt loss of
stiffness of geopolymer at elevated temperatures", Cement and Concrete
Composites, 32(9), pp. 657-664.
271. Wyrzykowski, M. and Lura, P. (2014), "The effect of external load on
internal relative humidity in concrete", Cement and Concrete Research,
65(0), pp. 58-63.
272. Carniglia, S.C. (1986), "Construction of the tortuosity factor from
porosimetry", Journal of Catalysis, 102(2), pp. 401-418.
273. Salmas, C.E. and Androutsopoulos, G.P. (2001), "A novel pore structure
tortuosity concept based on nitrogen sorption hysteresis data", Industrial and
Engineering Chemistry Research, 40(2), pp. 721-730.
274. Androutsopoulos, G.P. and Salmas, C.E. (2000), "A new model for capillary
condensation-evaporation hysteresis based on a random corrugated pore
structure concept: Prediction of intrinsic pore size distributions. 1. Model
formulation", Industrial and Engineering Chemistry Research, 39(10), pp.
3747-3763.
275. Trtik, P., Kaufmann, J., and Volz, U. (2012), "On the use of peak-force
tapping atomic force microscopy for quantification of the local elastic
modulus in hardened cement paste", Cement and Concrete Research, 42(1),
pp. 215-221.
276. Jones, C.A., Grasley, Z.C., and Ohlhausen, J.A. (2012), "Measurement of
elastic properties of calcium silicate hydrate with atomic force microscopy",
Cement and Concrete Composites, 34(4), pp. 468-477.
277. Hitchcock, L.B. and McIlhenny, J. (1935), "Viscosity and density of pure
alkaline solutions and their mixtures", Industrial & Engineering Chemistry,
27(4), pp. 461-466.
278. Barneyback Jr, R.S. and Diamond, S. (1981), "Expression and analysis of
pore fluids from hardened cement pastes and mortars", Cement and Concrete
Research, 11(2), pp. 279-285.
279. Sarker, P.K. (2009), "Analysis of geopolymer concrete columns", Materials
and Structures/Materiaux et Constructions, 42(6), pp. 715-724.
280. Sanjuán, M.A., Andrade, C., and Cheyrezy, M. (2003), "Concrete
carbonation tests in natural and accelerated conditions", Advances in Cement
Research, 15(4), pp. 171-180.
281
281. Bureau of Meteorology, Historical weather observations and statistics,
[cited 2016 22nd April]; Available from:
http://www.bom.gov.au/climate/data-services/station-data.shtml.
282. Almudaiheem, J.A. and Hansen, W. (1987), "Effect of specimen size and
shape on drying shrinkage of concrete", ACI Materials Journal, 84(2), pp.
130-135.
283. Saudi, M.S., Shields, J., O'Connor, D., and Hutchens, E. (1996), "Variation of
prestress force in a prestressed concrete bridge during the first 30 months",
PCI Journal, 41(5), pp. 66-72.
284. Rajamane, N.P., Nataraja, M.C., Lakshmanan, N., and Dattatreya, J.K.
(2012), "Pull-out tests for bond strengths of geopolymer concretes", Indian
Concrete Journal, 86(10), pp. 25-38.
285. RILEM 129-MHT (1997), "Recommendations: Part 6-thermal strain",
Materials and Structures, 30(1), pp. 17-21.
286. RILEM 129-MHT (2000), "Part 8: Steady-state creep and creep recovery for
service and accident conditions", Materials and Structures, 33(1), pp. 6-13.
287. Kayyali, O.A. (1986), "Effect of hot environment on the strength and porosity
of Portland cement paste", Durability of building materials, 4(2), pp. 113-
126.
288. Janotka, I. and Mojumdar, S.C. (2005), "Thermal analysis at the evaluation of
concrete damage by high temperatures", Journal of Thermal Analysis and
Calorimetry, 81(1), pp. 197-203.
289. Guerrieri, M. and Sanjayan, J. (2011), "Investigation of the Cause of
Disintegration of Alkali-Activated Slag at Temperature Exposure of 50°C",
Journal of Materials in Civil Engineering, 23(12), pp. 1589-1595.
290. Zhou, Q. and Glasser, F.P. (2001), "Thermal stability and decomposition
mechanisms of ettringite at <120°C", Cement and Concrete Research, 31(9),
pp. 1333-1339.
291. Cong, X. and Kirkpatrick, R.J. (1995), "Effects of the temperature and
relative humidity on the structure of CSH gel", Cement and Concrete
Research, 25(6), pp. 1237-1245.
292. Parrott, L.J. (1977), "Basic creep, drying creep and shrinkage of a mature
cement paste after a heat cycle", Cement and Concrete Research, 7(5), pp.
597-604.
293. White, C.E. (2016), "Effects of temperature on the atomic structure of
synthetic calcium-silicate-deuterate gels: A neutron pair distribution function
investigation", Cement and Concrete Research, 79, pp. 93-100.
294. Bernal, S.A., Rodríguez, E.D., De Gutiérrez, R.M., and Provis, J.L. (2015),
"Performance at high temperature of alkali-activated slag pastes produced
with silica fume and rice husk ash based activators", Materiales de
Construccion, 65(318).
282

Creep Behavior of GP

Uploaded by

Document Information

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

Creep Behavior of GP

Uploaded by

Copyright:

Available Formats

Creep Behaviour of

Submitted in total fulfilment of the requirements of the degree of

Faculty of Science, Engineering and Technology

Swinburne University of Technology, Melbourne, Australia

First of all, I would like to express my sincere thanks to my research supervisor,

Journal articles under preparation:

Published conference papers:

AAM Alkali-activated materials

1.2 Research significants

There are some objectives in this research:

1.4 Thesis organisation

Chapter 8 presented the investigation of effect of temperature up to 100 °C on creep

2.2.1 Ordinary Portland cement concrete

CaCO3 → CaO + CO2

2CaO + SiO2 → 2CaO·SiO2

2CaO·SiO2 + CaO → 3CaO·SiO2

3CaO·Al2O3 + CaO + Fe2O3 → 4CaO·Al2O3·Fe2 O3

The manufacture of OPC is an energy-intensive process, as fuel is burned to maintain

2(3CaO·SiO2) + 7 H2O → 3CaO·2SiO2·4H2O + 3 Ca(OH)2

2(2CaO·SiO2) + 5 H2O → 3CaO·2SiO2·4H2O + Ca(OH)2

Table 2.1: Main compounds of OPC

Name of compound Composition Abbreviation

2.2.2 Geopolymer concrete

Besides the precursors, another major component of geopolymer is the alkali

Similar to OPCC, aggregates can be added into geopolymer to produce GPC.

• Aluminium-rich gel (Gel 1)

• Zeolite precursor gel (Gel 2)

Bakharev, Sanjayan and Cheng [39] assessed the performance of slag-based

A substantial reduction of CO2 emission can be achieved by blended OPC with

1.1Ca(OH)2 + SiO2 + H2O → 1.1Ca(OH)2·SiO2·2.1H2O

The research on blended cement would provide some information on the

2.3.1 Hardened OPC

2.3.2 Hardened geopolymer

Past researches showed that the chemistry of geopolymer is different to that of

In fly-ash based geopolymer, the main reaction product is an alkaline silicoaluminate

2.3.3 Differences in microstructure between OPC and geopolymer

Because of the cross-linked feature mentioned above, hardened geopolymer has a

Figure 2.6: Ternary plot of binder gel compositions of geopolymer (noted as

It is believed that the microstructure has an influence on the properties and

Balshin’s equation: 𝑓𝑐 = 𝑓𝑐0 (1 − 𝜙)𝑛 (2.1)

Ryshkevitch and Duckworth’s equation: 𝑓𝑐 = 𝑓𝑐0 𝑒 −𝑘𝑟 𝜙 (2.2)

Schiller’s equation: 𝑓𝑐 = 𝑘𝑆 𝑙𝑛(𝜙0𝑆 /𝜙) (2.3)

Hasselman’s equation: 𝑓𝑐 = 𝑓𝑐0 − 𝑘𝐻 𝜙 (2.4)

Brough and Atkinson [71] investigated the evolution of the paste/aggregate

2.4.2 Permeability and durability

Pore characteristic of concrete can be obtained through a number of methods. The

There are more methods in prediction of permeability by pore size distribution.

 Kozeny-Carman equation: It treats the pore structure as a collection of circular

where S is the specific surface area (SSA).

 Katz-Thompson permeability theory: It uses a critical pore diameter and a single

where c is a constant = 1/226, dc is the critical pore diameter obtained by MIP

where Vi is the pore volume corresponding to the pore radius ri.

2.4.4 Analysis for pore characteristic

where V is the volume of adsorbed vapour at pressure P, Vm is the volume of gas

𝐶 = 𝑒 (𝑞1 −𝑞𝐿 )/(𝑅𝑇)

2.5 Creep of concretes

Creep in concrete is a phenomenon that increases strain when concrete is subjected to

𝜀𝑐𝑟 (𝑡) = 𝜀 (𝑡) − 𝜀𝑒 (𝑡0 ) − 𝜀𝑠ℎ (𝑡) (2.21)

2.5.2 Creep of OPCC

Creep recovery Creep recovery

Bažant [6] described several mechanisms causing or influencing creep:

When the concrete member is under restrained condition, cracking is developed as

It was found that a number of factors influenced the shrinkage of geopolymer.

2.5.4 Creep of blended cement concrete