Response of Concrete Pavements Under Moving Vehicular Loads and Environmental Effects

RESPONSE OF CONCRETE PAVEMENTS UNDER
MOVING VEHICULAR LOADS AND

ENVIRONMENTAL EFFECTS
by
Mostafa Yousefi Darestani
A dissertation submitted for the degree of
Doctor of Philosophy
Centre of Built Environment and Engineering Research
Queensland University of Technology
2007
i
To my parents and to my wife, Nazli, and
my sweet girl, Parastoo
ii
KEYWORDS
Dynamic Analysis, Dynamic amplification, Finite element analysis, Finite element
model, Static analysis, Axle group loads, Critical axle group configuration, Critical
position of axle groups, Tyre pavement interaction, Stress distribution, Fatigue
failure, Concrete pavement distresses, Corner cracking, Longitudinal cracking,
Transverse cracking, Top-down cracking, Bottom-up cracking, Boundary conditions,
Debonding layer, Temperature effects, Loss of moisture contents, Shrinkage effects,
Curling induced stress, Warping stress, Pavement curvature, Field test, Laboratory
text, Experimental study, Truck load, JPCP, JRCP, CRCP, SAST, SADT, TAST,
TADT, TRDT, QADT.
iii
ABSTRACT
The need for modern transportation systems together with the high demand for
sustainable pavements under applied loads have led to a great deal of research on
concrete pavements worldwide. Development of finite element techniques enabled
researchers to analyse the concrete pavement under a combination of axle group
loadings and environmental effects. Consequently, mechanistic approaches for
designing of concrete pavements were developed based on results of finite element
analyses. However, unpredictable failure modes of concrete pavements associated
with expensive maintenance and rehabilitation costs have led to the use of empirical-
mechanistic approach in concrete pavement design.
Despite progressive knowledge of concrete pavement behaviour under applied loads,

concrete pavements still suffer from deterioration due to crack initiation and
propagation, indicating the need for further research. Cracks can be related to fatigue
of the concrete and/or erosion of materials in sub-layers. Although longitudinal, mid-
edge and corner cracks are the most common damage modes in concrete pavements,
Austroads method for concrete pavement design was developed based on traditional
mid-edge bottom-up transverse cracking introduced by Packard and Tayabji (1985).
Research presented in this thesis aims to address the most common fatigue related
distresses in concrete pavements. It uses comprehensive finite element models and
analyses to determine the structural behaviour of concrete pavements under vehicular
loads and environmental effects. Results of this research are supported by laboratory
tests and an experimental field test.
Results of this research indicate that the induced tensile stresses within the concrete
pavement are significantly affected by vehicle speed, differential temperature
gradient and loss of moisture content. Subsequently, the interaction between the
above mentioned factors and concrete damage modes are discussed. Typical dynamic
amplifications of different axle groups are presented. A new fatigue test setup is also
developed to take into consideration effects of pavement curvature on fatigue life of
the concrete. Ultimately, results of the research presented in this thesis are employed
to develop a new guide for designing concrete pavements with zero maintenance of
fatigue damage.
iv
LIST OF PUBLICATIONS
International Refereed Journal Papers:

“Structural Response of Concrete Pavements under Moving Truck Loads”, Journal of
Transportation Engineering, ASCE (Accepted but not published yet).
International Refereed Conference Papers:

“Dynamic Response of concrete pavements under moving Vehicular loads”, IABSE
SYMPOSIUM on Response to Tomorrow’s Challenges in Structural Engineering,
Budapest, Hungary, 2006
“A Review of 2004 Austroads Rigid Pavement Design”, 22nd ARRB Conference –

Research into Practice, Canberra, Australia, 2006
“Experimental Study on Structural Response of Rigid Pavements under Moving Truck

Load”, 22nd ARRB Conference – Research into Practice, Canberra, Australia, 2006
Submitted Papers to International Refereed Journals:

“Influence of Vehicular Positions and Thermal Effects on Structural Behaviour of
Concrete Pavement”, Journal of Mechanics and Materials and Structures, USA (This
paper was review and needs some revision)
“Sensitivity Analysis of the 2004 Austroads Design Guide”, Concrete Institute of

Australia
Preparation of Journal Papers:

“A New Method for Laboratory Fatigue Test of the Concrete”
“A New Slab Thickness Design Guide for Zero Maintenance of Fatigue Damage”
“Structural dynamic analysis of unbonded JPCP Subjected to Moving Axle Group

Loads and Environmental Effects”
“Structural dynamic analysis of unbonded JPCP Subjected to Moving Axle Group

Loads and Environmental Effects”
v
Executive Research Summary
Sustainability of concrete pavements depends on the rate of pavement deteriorations.

Cracks, as the main reason behind deteriorations of concrete pavements, are initiated
at the top or the bottom surface layer of the concrete slab and then, propagated into
the depth of concrete slab due to fatigue of the concrete.
Corner, longitudinal and transverse cracks are the most common fatigue failure
modes of concrete pavements. Location and density of cracks can be predicted based
on distribution of induced tensile stresses within the concrete slab. The tensile
stresses in concrete pavements are induced by a combination of vehicular loads and
environmental effects.
Temperature fluctuation within the depth of concrete slab and loss of moisture content
are the most important environmental effects considered in concrete pavement
analysis. However, analysis of concrete pavements under vehicular loads and
environmental effects depends on a vast number of parameters and interrelationships
among them. To address the interrelationships among parameters and structural
behaviour of concrete pavements, two types of concrete pavements, namely JPCP and
JRCP, were studied.
In the first phase, a sensitivity analysis of the Austroads slab thickness design guide
(2004) was performed to clarify the interrelationships between design parameters and
calculated slab thickness. Results of the current study indicate that the Austroads
method (2004) has several shortcomings and needs to be improved. The significant
shortcomings can be summarised as follows:
- In contrast with what is commonly accepted as the fundamental concrete

characteristics, calculations performed using the 2004 Guide seem to suggest
that an increase in concrete compressive strength decreases the possibility of
fatigue damage such that erosion damage becomes critical.
- Increasing the subgrade CBR above 5 per cent has no effect on slab thickness
for design traffic in excess of 1×107 HVAGs.
vi
- The minimum recommended slab thickness is greater than the calculated
thickness provided that the pavement is dowelled and restrained by shoulder or
adjacent traffic lanes.
- Vehicular loads are considered as static loads although they are dynamic in
nature.
- To account for environmental factors, the Guide provides a variety of

minimum concrete slab thickness for different types of concrete pavements and
diverse range of design traffic. However, as shown in this thesis, behaviour of
concrete pavements under vehicular loads depends on the magnitude of
differential temperature and/or loss of moisture contents. Hence, consideration
of environmental effects as a constant value, i.e. a minimum slab thickness,
may result in other failure types in the concrete slabs that are not considered in
the method.
- Flexural fatigue damage was assumed to only occur at the bottom surface layer
of the concrete slab.
To address the above mentioned shortcomings, parametrical studies of the concrete

pavements were involved in the second phase. Since parametrical dynamic studies of
the concrete pavements are highly time consuming, these studies were performed using
static analyses. Diverse finite element programs for analysis of concrete pavements
were developed in the past. However, most of these programs were developed to study
behaviour of JPCP under vehicular loads and environmental effects. Moreover, other
finite element programs for analysis of JRCP or CRCP were not accessible.
Consequently, the EverFE, a 3D finite element program for analysis of JPCP whose
accuracy was previously examined by several researchers, was selected to be used in
parametrical studies.
Static analyses of diverse plain concrete pavements with different configuration were
performed to understand how debonding layer, axle group configurations, differential
temperature and position of axle groups upon the pavement affect the induced tensile
stress within the concrete slab.
vii
Although the main aim for placing a debonding layer between concrete slab and
subbase is to eliminate the early age cracking in the concrete slab, provision of this
layer plays a significant role in structural behaviour of concrete pavement during the
pavement life. Results of the current study show that the benefits offered by
consideration of the unbonded boundary condition between concrete slab and subbase
cease at a certain value of differential temperature. Hence, particular care needs to be
given to those pavement projects constructed in hot or cold weather, where high
differential temperature gradients may be produced in concrete depth.
Since axle group configuration varies among heavy vehicle manufacturer and across
countries, it is essential to determine the critical dimensions of a given axle group. If
the critical axle group configuration is not considered in the analysis, the results of the
FEA may be inadequate and lead to early cracking of the pavement in the real
condition. As a result, the critical dimensions of axle groups were determined.
Subsequently, some practical values for determining the critical axle group
configuration were provided.
Critical positions of different axle groups in uncurled and curled jointed concrete
pavement with different configurations were also studied. Results of the current study
indicate that AASHTO recommendation (2003) and results of Packard and Tayabji
(1985) are valid for an uncurled pavement with a fully unbonded boundary condition
between concrete slab and subbase. Results of the current study also show that
pavement performance under combinations of vehicular loads and differential
temperatures is significantly affected by the boundary condition between concrete slab
and subbase.
The reasons behind longitudinal, transverse and corner cracking were addressed.
Depending on differential temperature between the top and the bottom surface layers
of the concrete slab, corner, centre and mid-edge loadings can result in different types
of fatigue failure in concrete slab. For instance, corner loading may enhance corner
cracking, transverse cracking at the edge or mid-edge of the pavement and longitudinal
cracking, depending on the differential temperature considered in the analysis. In terms
of maximum induced tensile stress, results of the current study show that corner
loading is critical in the presence of bonded boundary condition between concrete slab
viii
and subbase. In an unbonded pavement, corner loading is also critical when a
separation due to environmental forces occurs between the unbonded concrete slab and
subbase.
Parametrical static analyses of the JPCP were also performed to define effects of
concrete slab thickness and modulus of subgrade reaction on concrete pavement
behaviour. Results show that an inverse relationship exists between induced tensile
stress and the thickness of the concrete slab. In the other words, an increase in the
thickness of concrete the slab decreases the magnitude of induced tensile stress.
However, this result is not valid when a combination of vehicular loads and high
differential temperature is considered. Consequently, a maximum slab thickness in the
presence of high differential temperature between the top and the bottom surface layer
of the concrete slab was defined. A certain dowel arrangement at the corner of the
concrete slab can also eliminate the aforementioned problem. Hence, the use of longer
dowel with greater size and shorter distance between dowels was recommended.
Depending on the boundary condition between the concrete slab and the subbase,
corner or mid-edge loading and daytime or nighttime differential temperature, an
increase in modulus of subgrade reaction may increase or decrease the magnitude of
tensile stress.
Since performance of concrete pavements is strongly affected by concrete properties,

it is important to determine the interrelationship between concrete properties. Some
typical equations for prediction of concrete properties were developed in the past. For
practical reasons, the concrete properties are estimated based on its compressive
strength. To determine the most accurate equations in prediction of concrete
properties, a series of laboratory test were performed.
The significant properties of concrete used in concrete pavement technology are

compressive strength, flexural strength, modulus of elasticity, shear strength of
aggregate interlock and cement paste and fatigue life of concrete. Results of the
laboratory tests indicate that the concrete modulus of elasticity can be accurately
estimated by using equations developed in the past. Furthermore, the equations
provided for estimation of concrete flexural strength can be used when a specific
curing method of concrete is considered during pavement construction. In other words,
ix
the provided equations are not able to accurately estimate the flexural strength of air
cured concrete. As a result, these equations are not sufficient to be used in concrete
pavement unless an appropriate curing method is provided. Consequently, a typical
equation for prediction of flexural strength of air cured concrete was provided in the
current research based on the laboratory tests performed.
The load transfer efficiency across joints and cracks depends on the shear transfer
capability of aggregate interlock and cement paste. The shear transfer capability of
aggregate interlock and cement paste was also determined using notch prism beam.
This property defines the capability of the concrete for transferring the shear force
across the initiated cracks and helps to understand the behaviour of concrete at the
initiated cracks.
Prediction of concrete fatigue life is the key factor in estimation of fatigue related
damage of the concrete slab. Diverse fatigue prediction models of concrete were
developed in the past based on laboratory tests of concrete prism beams. The
traditional laboratory fatigue test is based on a three points loading configuration using
one directional cyclic loads. Since the concrete pavement is curled upward during
nighttime and downward during daytime, it was questioned if the use of traditional
fatigue setup may produce insufficient fatigue prediction model of the concrete.
Consequently, a new fatigue setup was developed to take into consideration the
pavement curvature during daytime and nighttime differential temperatures. Results of
the fatigue laboratory tests performed in the current study show that the equations
developed in the past for estimation of concrete fatigue life are not sufficient.
Dynamic analyses of bonded and unbonded JPCP and JRCP under moving axle groups
were performed in the next stage. Results of the current study show that dynamic
analysis is required to accurately predict the failure mode of concrete pavements.
Critical speeds of each axle group based on types of concrete pavements were
determined. For the first time, dynamic amplifications of each axle group were
presented in the current research. The critical locations for severe fatigue cracking in
both JPCP and JRCP were addressed. Results also indicated that fatigue cracking is
affected by axle group types and speed. It was determined in dynamic analysis that the
x
damage location may be close to transverse joints, at midpoint or in some cases at
quarter point of slab.
In addition to dynamic amplification of each axle group, the most significant finding of
dynamic studies performed in the current study was determination of stress repetitions
in concrete pavement due to a given axle group. In the static analysis, the number of
stress repetition for a given axle group is equal to the number of axles in the axle
group. In other words, the number of stress repetitions in a point within the concrete
pavement for single axles, tandem axles, triple axle and quad axle groups are one, two,
three and four respectively. In the presence of bonded boundary condition between
concrete slabs and subbase, the aforementioned stress repetitions are still correct in
dynamic analysis. However, provision of debonding layer between concrete slabs and
subbase produce greater number of stress predictions in the dynamic analysis.
This stress repetition phenomenon strongly depends on axle speed, type of axle group
and location where the stress is monitored. For SAST, SADT, TAST and TADT
higher speed, i.e. 110 km/h, produces greater stress repetitions than lower speeds. On
the other hand, in the heavy weight axle groups such as TRDT and QADT lower
speed, i.e. 30 km/h, produces greater stress repetitions than higher speeds. The average
number of stress repetitions for SAST, SADT, TAST, TADT, TRDT and QADT are 1,
1, 4, 6, 8 and 9 in JPCP and 5, 8, 9, 8, 12, and 9 in JRCP respectively.
Furthermore, effects of temperature fluctuation on the dynamic response of the

pavement were addressed. Results showed that differential temperature gradients may
increase or decrease the stress repetition for each axle group. However, greater tensile
stresses were observed in concrete pavements in the presence of differential
temperature.
Since results of the experimental field tests on dynamic response of concrete

pavements carried out in the past were not compatible with the findings of this
research, a fully instrumented concrete pavement test section including JPCP and
JRCP was constructed and tested under quasi-static and dynamic truck loadings.
Information on the test section, instrumentation layout, material properties and truck
characteristics were described. Pavement performance under environmental conditions
xi
was studied during the first 28 days after casting. Truck loading was subsequently
applied at different predetermined locations of the pavement including at free
longitudinal edge, at the confined longitudinal edge and at the centre area of the traffic
lane. Time history responses were recorded for truck speeds between 5 km/h to 55
km/h.
Investigation of the recorded time history responses of the test section also indicates
the importance of dynamic analysis in concrete pavement design. The recorded time
histories validate the results of dynamic analysis performed in the current research.
Results also indicate that dowel position can strongly influence the pavement
responses. Furthermore, the slab deflection in JRCP decreases when reinforcement is
located close to the bottom surface layer of the concrete slab.
Results of the current study were used to develop a new empirical-mechanistic guide
for designing of concrete pavements. Consequently, typical equations for stress
prediction in concrete pavements for different loading conditions, differential
temperatures, slab thickness, modulus of subgrade reaction, and provision of shoulders
were developed. The accuracy of equations was then determined by comparing the
predicted stress with results of finite element analyses.
Using Miner’s rule, equations for calculating the fatigue damage of concrete slab were
developed. Transverse, corner and longitudinal cracks were contributed in the fatigue
damage model. Thickness of the concrete slab was considered to be adequate if none
of the above failure types were observed in the pavement. Ultimately, the design
procedure was exemplified.
xii
TABLE OF CONTENTS
Keywords iii
Abstract iv
List of Publications v
Executive Research Summary vi
Table of contents xiii
List of figures xviii
List of tables xxvii
List of Abbreviations xxix
Symbols xxxii
Statement of original authorship xxxvii
Acknowledgments xxxviii
Chapter 1: Introduction
1.1. Background 1
1.2. Research problems 2
1.3. Research hypothesis 3
1.4. Thesis scope 3
1.5. Research objectives 3
1.6. Thesis layout 4
Chapter 2: Literature review – Concrete Properties
2.1. Background 9
2.2. concrete strength 9
2.3. Modulus of elasticity 11
2.4. Coefficient of thermal expansion 11
2.5. Shrinkage 12
2.6. Fatigue 13
2.7. Summary 17
Chapter 3: Literature review – Concrete Pavements
3.1. Background 19
3.2. Concrete pavement cross section
3.2.1. Subgrade 20
3.2.2. Subbase 25
3.2.3. Debonding layer 26
3.2.4. Concrete slab 28
3.2.4.1. Jointed plain concrete pavement 28
3.2.4.2. Jointed reinforced concrete pavement 28
3.2.4.3. Continuously reinforced concrete pavement 29
3.2.5. Surface roughness 29
xiii
3.2.6. Shoulder 30
3.2.7. Joints 30
3.2.7.1. Isolation joint 31
3.2.7.2. Contraction joint 31
3.2.7.3. Construction joint 32
3.2.7.4. Expansion joint 32
3.2.8. Load transfer devices 32
3.2.8.1. Aggregate interlock 32
3.2.8.2. Dowel 33
3.2.8.3. Tie bars and keyed joints 34
3.2.9. Load transfer efficiency 34
3.2.10. Differential deflection 35
3.3. Loadings 35
3.3.1. Traffic loads 36
3.3.2. Temperature 45
3.3.3. Shrinkage - loss of moisture content 49
3.4. Concrete pavement analysis
3.4.1. Analytical solution 50
3.4.2. Numerical solution 52
3.4.2.1. Discrete element method 52
3.4.2.2. Finite element method 52
3.4.3. Static analysis 58
3.4.4. Transient dynamic study 58
3.4.4.1. Transient dynamic analysis 60
3.4.4.2. Experimental tests of concrete pavements 64
3.5. Concrete pavement distresses 66
3.5.1. Fatigue damage of concrete slab 66
3.5.2. Erosion of subbase and subgrade materials 72
3.5.3. Spalling 73
3.6. Concrete pavement design guides 73
3.6.1. Austroads 2004
3.6.1.1. Introduction 74
3.6.1.2. Method description 77
3.6.2. AASHTO 2003
3.6.2.1. Introduction 81
3.6.2.2. Methodology 82
3.6.2.3. JPCP design features 85
3.6.2.4. Distress prediction 86
3.6.2.5. Surface roughness 86
3.6.2.6. Thickness of concrete slab 86
xiv
3.7. Summary 86
Chapter 4: Summary of the literature review and research plan
4.1. Summary of the literature review 88
4.2. research methodology and plan 93
Chapter 5: Sensitivity analysis of concrete pavement design using
Austroads guide 2004
5.1. Introduction 96
5.2. Aims of this study 96
5.3. Development and verification of ANRPD-2004 program 98
5.4. Results and discussion
5.4.1. Effect of pavement types on thickness of concrete slab 100
5.4.2. Effect of concrete strength on damage mode 100
5.4.3. Effect of project design reliability on slab thickness 104
5.4.4. Effect of subbase layer on slab thickness 105
5.4.5. Effect of subgrade CBR on slab thickness 105
5.4.6. Minimum recommended base thickness 109
5.4.7. Sensitivity of the fatigue and erosion analysis to a change in 109
slab thickness
5.4.8. Damage process 111
5.5. Summary 112
Chapter 6: Static analysis of jointed plain concrete pavement
6.2. Finite element model 115
6.3. Effects of debonding layer on concrete pavement responses 116
6.3.1. Methodology 117
6.3.2. Results and discussion 118
6.4. Revisiting axle group configurations
6.4.2. Width to length ratio of the tyre-pavement contact area 122
6.4.3. Tyre inflation pressure 125
6.4.4. Distance between the centres of dual tyres 126
6.4.5. Axle width 127
6.4.6. Axle spacing in a given axle group 127
6.4.7. Load shift between axles in a given axle group 130
6.4.8. Result validation 131
6.4.9. Variation in axle group loads 132
6.5. Critical location of axle groups upon pavements 134
6.5.2. Axle group loadings 137
6.5.3. Thermal induced stress 138
6.5.4. Combination of vehicular and thermal induced stresses 145
6.5.5. Effects of slab thickness on induced tensile stress 152
xv
6.5.6. Effects of modulus of subgrade reaction on induced tensile 158
stress
6.6. Summary 156
Chapter 7: Laboratory tests of concrete
7.2. Compressive test 159
7.3. Flexural test 162
7.4. Modulus of elasticity 166
7.5. Notch beam test 167
7.6. Fatigue test 170
7.6.1. Testing procedure 174
7.7. Summary 178
Chapter 8: Dynamic analysis of bonded concrete pavements under
moving axle group loads
8.2. Methodology 181
8.3. Finite element model description 182
8.4. Model Calibration 185
8.5. Effects of moving single axles on induced tensile stress 187
8.6. Effects of moving tandem axles on induced tensile stress 191
8.7. Effects of moving triple and quad axles on induced tensile stress 194
8.8. Effects of axle speeds on alb deflection 194
8.9. Critical speed of axle groups and location of severe damage 203
8.10. Effect of reinforcement 205
8.11. Summary 209
Chapter 9: Experimental field test of concrete pavement under moving
truck loads
9.2. Project description 211
9.3. Instrumentations 218
9.4. Material properties 222
9.5. Visual monitoring of test section 222
9.6. Truck characteristics, movement and speed 224
9.7. Pavement roughness 225
9.8. Results and discussion 226
9.8.1. Concrete slab deflection 227
9.8.2. Induced tensile stress 230
9.8.3. Vertical acceleration in concrete slabs 232
9.9. Temperature fluctuation 233
9.10. Summary 234
xvi
Chapter 10: Dynamic analysis of unbonded concrete pavements under
moving axle group loads
10.2. Development of finite element model 236
10.3. Model calibration 238
10.4. Results and discussion 241
10.4.1. Effects of moving single axles on induced tensile stress 241
10.4.2. Effects of moving tandem axles on induced tensile stress 247
10.4.3. Effects of moving triple and quad axles on induced tensile 249
stress
10.5. Critical speed of axle groups and number of stress repetition 249
10.6. Combination of moving axle groups and differential temperature 256
10.7. Summary 262
Chapter 11 Development of slab thickness design guide for zero
maintenance fatigue damage
11.2. Cross section of the concrete pavement 265
11.3. Concrete characteristics 265
11.4. Subbase 266
11.5. Prediction of the maximum induced tensile stress 266
11.6. Environmental effects 271
11.7. Variation of slab thickness 273
11.8. Variation of modulus of subgrade reaction 273
11.9. Combination of vehicular loads and environmental effects 274
11.10. Validation of stress prediction in jointed plain concrete 275
pavement
11.11. Fatigue analysis 276
11.12. An example of the method 282
11.13. Summary 284
Chapter 12: Conclusion and Recommendation for further study
12.1. Contribution from this research 285
12.2. Conclusion 286
12.3. Recommendation for future study 291
References 292
xvii
LIST OF FIGURES
Number Description Page
Figure 2-1 Fatigue relationship adopted in Austroads design model 14

(Austroads, 2004)
Figure 2-2 Variation of fatigue life with concrete compressive strength 15
Figure 2-3 Variation of fatigue life with concrete compressive strength for 16
gravel aggregate and crushed rock
Figure 2-4 Effect of different types of fibres on concrete fatigue flexural 17

strength
Figure 3-1 Cross section of a typical concrete pavement in Australia 19
Figure 3-2 DL subgrade idealization 21
Figure 3-3 Spring model idealization 21
Figure 3-4 Limitation of the spring model 21
Figure 3-5 Winkler modified idealization 22
Figure 3-6 Elastic solid (ES) idealization 22
Figure 3-7 TP model idealization 23
Figure 3-8 Kerr model idealization 24
Figure 3-9 Typical experimentally determined frictional stress versus 27

concrete slab slippage for different materials in subbase
Figure 3-10 Typical expansion joint 31
Figure 3-11 Typical isolation joint 31
Figure 3-12 Typical keyed and tied longitudinal joints 34
Figure 3-13 Typical components for calculation of LTE in a transverse 35

crack
Figure 3-14 Effect of lowering distance between outer edge of tyre imprint 37
and pavement edge on JPCP transverse cracking
Figure 3-15 Illustration of Packard and Tayabji’s (1985) assumption 38
xviii
Figure 3-16 Critical axle position on the pavement 38
Figure 3-17 Configurations of traffic lanes considered in the AASHTO 39
Figure 3-18 Configurations of concrete slabs in the longitudinal direction 40

considered by AASHTO
Figure 3-19 Effect of number of slabs in the transverse direction on 40

prediction of the maximum bottom bending surface stress
Figure 3-20 Effect of number of slabs in the longitudinal direction on 41

prediction of maximum induced tensile stresses at the bottom
surface layer of the concrete slab, single slab vs. three slabs
Figure 3-21 Axle group types considered in Austroads 2004 42
Figure 3-22 Transverse section of a typical SADT 43
Figure 3-23 Length and width of tyre–pavement contact area 45
Figure 3-24 Effects of daytime and nighttime temperature gradient on 47

concrete pavements in the absence of factors restraining the
concrete slab movement
Figure 3-25 Effects of daytime and nighttime temperature gradient on 48

concrete pavements in the presence of factors restraining the
concrete slab movement
Figure 3-26 Pavement modelling in EverFE finite element program 54
Figure 3-27 8-noded finite element setup in Totsky model 57
Figure 3-28 Schematic interface element behaviour and interface 58

constitutive relationship
Figure 3-29 Comparison between damage rate: moving constant load 61

versus time-history load
Figure 3-30 Comparison between fatigue cracking along slab length: 62

moving constant load versus time-history load
Figure 3-31 The critical location of the applied load for top-down transverse 69
cracking
Figure 3-32 The critical location of applied load for joint faulting 70
Figure 3-33 Critical load location for CRCP punchout crack 71
Figure 3-34 Erosion mechanism 72
xix
Figure 3-35 Meyerhof’s yield line theory 74
Figure 3-36 Estimation of effective subgrade CBR 77
Figure 3-37 Illustration of possible concrete pavement layered system 84

(AASHTO, 2003)
Figure 5-1 Algorithm flowchart of program used in ANRPD-2004 99
Figure 5-2 Variation of base thickness with design traffic (HVAGs) for 101
different pavement types
Figure 5-3 Variation of base thickness with Design traffic (HVAGs) for 102
dowelled JRCP with concrete compressive strength of 36 MPa
Figure 5-4 Variation of base thickness with Design traffic (HVAGs) for 103
dowelled JRCP with concrete compressive strength of 80 MPa
different project design reliabilities
different subbase types in a JRCP
different subgrade CBR in a JRCP
Figure 5-8 Typical rigid pavement thickness design curve for a specific 112
effective CBR and concrete flexural strength
Figure 6-1 JPCP configurations considered in this Chapter 116
Figure 6-2 Thermal induced tensile stress in concrete slab for different 118
debonding materials
Figure 6-3 Comparison between thermal induced tensile stresses in 120

different fully unbonded pavement configurations
Figure 6-4 Comparison between thermal induced tensile stresses in 121

different fully bonded pavement configurations
Figure 6-5 Load positions on the concrete pavement 122
Figure 6-6 Effect of tyre-pavement contact area on pavement response 124
Figure 6-7 Effect of tyre inflation pressure on pavement response 125
Figure 6-8 Effect of distance between the centres of dual tyres on 127
pavement response
xx
Figure 6-9 Effect of axle width on pavement response 128
Figure 6-10 Effect of axle spacing on pavement response 129
Figure 6-11 Effect of load shift between axles on pavement response for 131
TAST
Figure 6-12 Comparison in induced tensile stresses for bonded and 133
unbonded concrete pavements based on results of the current
study and dimensions used by Packard and Tayabji (1985)
Figure 6-13 Comparison in deflection of concrete slab for bonded and 133
unbonded concrete pavements based on results of the current
study and dimensions used by Packard and Tayabji (1985)
Figure 6-14 Position of applied loads for different axle groups on the centre 137
concrete slab
Figure 6-15 Vehicular induced stress in different pavement configurations 139

for unbonded boundary condition
Figure 6-16 Vehicular induced stress in different pavement configurations 140

for bonded boundary condition
Figure 6-17 Curling induced stress influence lines in bonded concrete slab 142
with full pavement configuration
Figure 6-18 Curling induced stress influence lines in unbonded concrete 143
slab with full pavement configuration
Figure 6-19 Vehicular and thermal induced tensile stresses in bonded 146
concrete slab with full pavement configuration
Figure 6-20 Vehicular and thermal induced tensile stresses in unbonded 147
concrete slab with full pavement configuration
Figure 6-21 Combination of vehicular and thermal induced tensile stresses 149
in a full pavement configuration subjected to SADT
Figure 6-22 Information on location of fatigue cracking 150
Figure 6-23 Position of the critical location in thick unreinforced concrete 154
pavement
Figure 7-1 Concrete compressive test setup using external electrical strain 160
gauge
Figure 7-2 Concrete compressive test ring 160
xxi
Figure 7-3 Typical failure mode of the cylindrical concrete specimens 161
Figure 7-4 Stress-strain curve of the concrete specimens tested on 23 /12 162
/2005
Figure 7-5 Instrumentation on the notch beam 168
Figure 7-6 The test setup for notch beams 169
Figure 7-7 Typical result of the notch beam test 169
Figure 7-8 The newly developed fatigue test setup 172
Figure 7-9 Comparison between the newly developed and the traditional 172
fatigue test setups
Figure 7-10 Specimen curvature during the test 174
Figure 7-11 Concrete fatigue testing 175
Figure 8-1 Finite element model of the JPCP 182
Figure 8-2 Reinforcement simulation for JRCP in the finite element model 187
Figure 8-3 Simulation of saw cut in the finite element model 184
Figure 8-4 Truck configuration used for validation of the finite element 186
analysis
Figure 8-5 Calibration of the finite element analysis 187
Figure 8-6 Influence line of induced tensile stress in concrete pavements 188
due to SAST
due to SADT
Figure 8-8 Comparison of transverse stress at confined edge and MG 190

stress in JPCP
due to TAST
due to TADT
due to TRDT
xxii
due to QADT
Figure 8-13 Influence line of slab deflection in concrete pavements due to 197
SAST
SADT
TAST
TADT
TRDT
QADT
Figure 8-19 Critical axle group speed based on the first principle stress 204
Figure 8-20 Comparison between slab deflection due to SAST in JPCP and 205
JRCP
Figure 8-21 Comparison between maximum induced tensile stresses due to 206
SAST in JPCP and JRCP
SADT in JPCP and JRCP
TAST in JPCP and JRCP
TADT in JPCP and JRCP
TRDT in JPCP and JRCP
QADT in JPCP and JRCP
Figure 9-1 Layout of the test section 212
Figure 9-2 Mesh reinforcement used in JRCPs 213
xxiii
Figure 9-3 Expended polystyrene blocks used to form voids at transverse 213
and longitudinal intersection
Figure 9-4 Assemblage of linear displacement transducer on the mounting 214

pole
Figure 9-5 Rectangular dowel bars installed at transverse joints 214
Figure 9-6 Cylindrical tie bars used at longitudinal joints 215
Figure 9-7 The use of single layer polyethylene sheet to create partially 216
bonded boundary condition in half length of the test section
Figure 9-8 The use of power trowel to enhance the surface smoothness 217
Figure 9-9 Preparing the transverse joints using saw cut 218
Figure 9-10 Embedded electrical strain gauges installed at a distance of 25 219

mm away from the subbase surface using rebar chair
Figure 9-11 Thermocouples installed at different depth of the concrete slabs 219
Figure 9-12 Locations of strain gauges 220
Figure 9-13 Locations of linear displacement sensors and vertical 221

accelerometers
Figure 9-14 eDaQ dynamic data acquisition system 221
Figure 9-15 Locations of cracks in the test section 222
Figure 9-16 Initiation of Crazing cracks at the top surface layer of the 223
concrete slabs
Figure 9-17 Initiation of a transverse crack in shoulder close to transverse 223

joints
Figure 9-18 Semi-trailer truck used in this study 224
Figure 9-19 Tyre configuration 225
Figure 9-20 Longitudinal coloured lines to help driver for maintaining the 226
truck movement at a certain distance of longitudinal edges
Figure 9-21 Time history deflection responses for different speeds at DL7 228
Figure 9-22 Time history deflection responses for different speeds at DR13 228
Figure 9-23 Time history deflection responses in JRCP for different 229
reinforcement locations
xxiv
Figure 9-24 Time history deflection responses at transverse joint 230
Figure 9-25 Comparison between dowel positions based on critical speed 230
Figure 9-26 Time history stress responses in JRCP at TCL12 for different 231
truck speeds
Figure 9-27 Time history stress responses in JRCP at TCL8 for different 232
truck speeds
Figure 9-28 Typical temperature fluctuation in depth of the concrete slab 234
Figure 10-1 Finite element model of the unbonded JPCP 237
Figure 10-2 Stress time history at BCR3 recorded in the field test 239
Figure 10-3 Stress time history at BCR3 derived from FEA results 240
Figure 10-4 Comparison of stress time histories at TCR12 between field 240
test and FEA
Figure 10-5 Influence line of induced tensile stress for a point at mid-edge 242
of concrete pavements due to SAST
of the concrete pavements due to SADT
Figure 10-7 Pavement curvature in JRCP under the QADT with speed of 244
110 km/h
Figure 10-8 curvature in JRCP under TRDT with a speed of 30 km/h 245
Figure 10-9 Influence line of induced tensile stress for a point at the corner 246
of the concrete pavements due to SAST
of the concrete pavements due to SADT
of the concrete pavements due to TAST
of the concrete pavements due to TAST
of the concrete pavements due to TADT
of the concrete pavements due to TADT
xxv
of the concrete pavements due to TRDT
of the concrete pavements due to TRDT
of the concrete pavements due to QADT
of the concrete pavements due to QADT
Figure 10-19 Pavement curvature due to a daytime differential temperature 260

of 8.5 ºC
Figure 10-20 Pavement curvature due to a nighttime differential temperature 260

of 8.5 ºC
Figure 10-21 Combination of SAST and daytime differential temperature of 261

8.5 ºC
Figure 10-22 Combination of SAST and nighttime differential temperature 261

of 8.5 ºC
Figure 10-23 Stress repetition due to a combination of SAST and daytime 262
differential temperature of 8.5 ºC
xxvi
LIST OF TABLES
Number Description Page
Table 2-1 Concrete coefficient of thermal expansion in a concrete 13

with cement to aggregate ratio of 1:6 for different
aggregate types and curing methods
Table 3-1 Traffic load regime at different loop in AASHTO test 65
Table 3-2 Distance between joints (m) for different types of concrete 75
pavement
Table 3-3 Selection of subbase type 76
Table 3-4 Presumptive traffic load distribution for urban roads 79

Table 3-5 Presumptive traffic load distribution for rural roads 80
Table 5-1 Design traffic (×106 HVAGs) below which fatigue 103
analysis is the key factor
Table 5-2 Concrete compressive strength (MPa) greater than those 104
provided in the table has no effect of the calculated base
thickness in the fatigue analysis
Table 5-3 Effects of subgrade CBR, concrete compressive strength, 108
and provision of dowels and shoulders on critical damage
process in a JRCP
Table 5-4 When to calculate the base thickness instead of adopting 110
the minimum recommended base thickness
Table 5-5 Effect of base thickness on fatigue damage in a JRCP 111
with shoulder
Table 6-1 Information on tyre-pavement contact area in SAST and 124
SADT
Table 6-2 Effect of variations in axle group loads on induced tensile 134
stress
Table 6-3 Critical location of fatigue cracking in full pavement 151
model due to different differential temperatures and
SADT
Table 6-4 Effect of concrete slab thickness on maximum induced 153
stress (MPa) due to different differential temperatures and
SADT
Table 6-5 Effect of modulus of subgrade reaction on maximum 156
induced stress (MPa) due to different differential
temperatures and SADT
Table 7-1 Results of concrete compressive tests 163
Table 7-2 Results of flexural tests for samples at 14 days age 164
Table 7-3 Results of flexural tests for samples at 28 days age 165
xxvii
Table 7-4 Comparison between results of flexural laboratory tests 166
with results from equations provided in the past for
estimation of flexural strength
Table 7-5 Results of notch beam tests 170
Table 7-6 The average flexural strength of the specimens used in the 174
fatigue test
Table 7-7 Results of Fatigue tests 176
Table 7-8 Comparison between results of the current study and the 177
equations developed for estimation of concrete fatigue life
Table 8-1 Information on axle configurations used in the current 185
study
Table 8-2 Type of required analysis in JPCP for each axle group 203
Table 10-1 Comparison is slab deflection between FEA and the field 238
test
Table 10-2 Summary of the dynamic results for different axle groups 259
and different types of unbonded concrete pavement
Table 11-1 Load safety factor for concrete pavement design 268
Table 11-2 Accuracy of the stress prediction equations provided in 270

the thesis
Table 11-3 Variations of coefficient C2 in Equation 10-7 270
Table 11-5 Comparison between results of the finite element analyses 272
and those from stress prediction equations
Table 11-7 Variations of coefficient α in Equation 10-10 for 274

unbonded pavements
Table 11-8 Variations of coefficient β in Equation 10-11 for 274
unbonded pavements
Table 11-9 Comparison between stress prediction models developed 277
in the current research with FEA and the stress prediction
model used in the Austroads method (2004)
Table 11-10 Comparison between stress prediction model developed in 278
the current research with corresponding results from FEA
Table 11-11 Adjustment factors for allowable load repetitions 280
Table 11-12 Results of the fatigue analysis for each axle group based 282
on different loading conditions
xxviii
LIST OF ABBREVIATIONS
AASHO American Association of State Highway Official

AASHTO American Association of State Highway and Transportation Official
AC Asphalt Cement
ACLRD Allowable Centre Load Repetition during Daytime
ACLRN Allowable Centre Load Repetition during Nighttime
ACI American Concrete Institute
ACoLRD Allowable Corner Load Repetition during Daytime
ACoLRN Allowable Corner Load Repetition during Nighttime
ALR Allowable Load Repetition
AMELRD Allowable Mid-Edge Load Repetition during Daytime
AMELRN Allowable Mid-Edge Load Repetition during Nighttime
AS Asphalt Stabilised Subbase
ASTM American Standard for Testing Materials
CBR California Baring Ratio
CCAA Cement and Concrete Association of Australia
Co Corner
CRCP Continuously Reinforced Concrete Pavement
CS Cement Stabilized
CTE Coefficient of Thermal Expansion
DA Dynamic Amplification
DBR Dowel Bar Retrofit
DD Deferential Deflection
DL Dense Liquid
EL Edge or Longitudinal
ES Elastic Solid
ESG Electrical Strain Gauge
ESAL Equivalent Single Axle Loading
FEM Finite Element Method
FHWA Federal Highway Administration
FRP Fibre Reinforced Polymer
xxix
GFRP Glass Fibre Reinforced Polymer
GR Granular subbase
HVAG Heavy Vehicle Axle Group
HVS Heavy Vehicle Simulator
IRI International Roughness Index
JPCP Jointed Plain Concrete Pavement
JRCP Jointed Reinforced Concrete Pavement
LC Lime treated clay
LDT Linear Displacement Transducer
LMC Lean-Mix Concrete
LS Load Shift between axles in a given axle group
LSF Load safety factor
LTE Load Transfer Efficiency
ME Mid-Edge
MG Maximum Global
MRST Minimum Recommended Base Thickness
NC Natural Clay
NSW New South Wales
PCA Portland Cement Association
PCC Portland Cement Concrete
PDR Project Design Reliability
QA Quad Axle
QADT Quad Axle Dual Tyre
QUT Queensland University of Technology
RTA Road and Traffic Authority
SA Single Axle
SADT Single Axle Dual Tyre
SAST Single Axle Single Tyre
SFCP Steel Fibre Concrete Pavement
TA Tandem Axle
TR Triple Axle
TADT Tandem Axle Dual Tyre
xxx
TAOT Tandem Axle Octa Tyre
TAST Tandem Axle Single Tyre
TRDT Triple Axle Dual Tyre
xxxi
SYMBOLS
a Fatigue or erosion coefficients

A Area of finite element
Ai Area association by node i
At Tyre-Pavement Contact area
b Fatigue or erosion coefficients

c Fatigue or erosion coefficients
C Elastic solid factor
Cd Structural damping
CIF Coefficient of internal friction
Cl Constant factor for consideration of slab length
Cb Constant factor for consideration of slab width
Cr Constant factor for consideration of reinforcement
C1 Coefficient factor for boundary condition between concrete slab and
subbase
C2 Coefficient factor for contributing different loading positions
C3 Coefficient factor for consideration of shoulder
C4 Coefficient for thermal stress prediction
CLS Coefficient factor for considering load shift between axles in a given
axle group
CTS Factor providing effect of slab thickness
Ck Coefficient for contribution of subgrade strength
d Fatigue or erosion coefficients
DD Differential deflection
DE Deflection Energy
DE j Deformation energy imposed on pumping zone i
e Fatigue or erosion coefficients

Ec Concrete modulus of elasticity
Ef Effective subgrade strength
xxxii
Es Soil modulus of elasticity
ESB Modulus of elasticity of subbase
ESL Modulus of elasticity of concrete slab
f Fatigue or erosion coefficients
fr Concrete flexural strength
f c' Concrete compressive strength

F(t) Load vector at time t
FDF Failure function in Drucker-Prager model
F1 Load adjustment for fatigue process
F2 Adjustment factor for slab edge effect

F3 Erosion factor
F4 Load adjustment for erosion

g Fatigue or erosion coefficients
G Coefficient describing the interaction between adjacent springs
h Fatigue or erosion coefficients
i Fatigue or erosion coefficients
I1 First invariant of the stress tensor
j Fatigue or erosion coefficients
J2 Second invariant of the stress tensor
k Modules of subgrade reaction
kg Stiffness of aggregate
kp Stiffness of cement past
k1 Material factor
k2 Load repetition factor
k3 Calibration factor for soil characteristics
k4 Calibration factor for concrete strength

K Stiffness of the equivalent spring
KDLI Stiffness of the interlayer springs in Totsky model
KDLS Stiffness of the subgrade in Totsky model
KDLU Stiffness of the upper (DL) springs in KV subgrade model idealization
xxxiii
Ki Subgrade modulus of reaction associated by node i
KPL Stiffness of the shear layer in KV subgrade model idealization
KPL1 Stiffness of the upper plate in Totsky model
KPL2 Stiffness of the lower plate in Totsky model
KS Stiffness of structure
KTP Stiffness of the TP subgrade model idealization
L Length of the pavement slab
Lt Length of tyre-pavement contact area
LSF Load safety factor

M Structural mass
Ne Allowable load repetition in erosion analysis
Nf Allowable load repetition in fatigue analysis
Nj Number of transverse joints per 30.5 m of the pavement length
P Axle group load

PE Equivalent load
Pf Ultimate axle group load
Pt Tyre inflation pressure
q Subgrade reaction pressure
Se Equivalent stress
Sr Stress ratio
t Time
T Base thickness
Tp Tyre pressure
TS Thickness of concrete slab
V Volume of material pumped from beneath of the slab
w Surface deflection
Wi Width association by node i
Wl Wheel Load
Wt Tyre-pavement contact width
X Nodal deflection
xxxiv
X& Nodal speed
X&& Nodal acceleration
α Width to length ratio of tyre-pavement contact area
αc Coefficient of expansion of concrete
α DP Drucker-Prager constant
αg Coefficient of expansion of aggregate
αp Coefficient of expansion of cement paste
αR Rayleigh damping constant

αt Constant factor for determination of debonding layer
β DP Drucker-Prager constant
βR Rayleigh damping constant
βK Constant factor
δ Deflection
δE Deflection of the slab edge
δL Deflection of the loaded slab
δU Deflection of the unloaded slab
∆i Deflection of node i
µ Friction coefficient
ξi Ratio of actual damping to critical damping
σ Stress
σt Tensile stress
σT Thermal induced stress

σV Vehicular induced tensile stress
σ V ,i Vehicular induced tensile stress due to axle i
∇2 The Laplace operation

∆T Absolute differential temperature between the top and the bottom
surface layer of the concrete slab
ωc Concrete Density
xxxv
ωi Natural circular frequency
ϖ Load frequency
νs Subgrade Poisson’s ratio
ν SB Poisson’s ratio of subbase
ν SL Poisson’s ratio of concrete slab
ϕ Angle of internal friction

τ Shear stress
λ Ratio of load frequency to structural natural frequency
xxxvi
STATEMENT OF ORIGINAL AUTHORSHIP
The work contained in this thesis has not been previously submitted to meet
requirements for an award at this or any other higher education institution. To the best
of my knowledge and belief, the thesis contains no material previously published or
written by another person except where due reference is made.
Signature ____________________________________
Date ____________________________________
xxxvii
ACKNOWLEDGMENTS
The author would like to thank Prof. David P. Thambiratnam and Dr. Andreas
Nataatmadja for their great support during this research. I am particularly grateful to
Dr. Daksh Baweja from Rinker Australia and Prof. John Bell from faculty of Built
Environmental and Engineering, Queensland University of Technology who allocated
the financial support for this research. Acknowledgement is also due to Rinker
Australia, industry partner of this research, for supplying concrete during laboratory
tests and allocating budget for the field test carried out during this research. I would
like to thank Mr. Arthur Powell and Mr. Trevor Laimer, Technicians of the QUT, for
providing technical advice during laboratory tests and also experimental field test. My
great thank to Mr. Glenn Carson from Rinker Australia who organized all services
contributed by the industry partner of this research. I also thank Mr. Mark Barry from
QUT who facilitated the finite element analyses by allocating more ANSYS licences to
this research. Finally, I would like to thank Dr. Adriana Bodnarova for her comments
on a draft of the text.
xxxviii
Chapter 1
INTRODUCTION
1.1. Background
The use of concrete pavement without an asphalt top layer dates back to Scotland in
1865 (Croney and Croney, 1998). With the growing worldwide interest in this field
during the 20th century, concrete pavement technology has been embraced in Australia
since the 1970s (Cruickshank, 1981). Concrete pavements were traditionally designed
based on theoretical equations developed by Westergaard (1926, 1933, and 1947).
Using finite element techniques, mechanistic approaches for designing of concrete
pavements were developed. The mid-edge bottom-up transverse fatigue cracking was
the only failure mode of the concrete pavements considered in the mechanistic design
guides. Initiation and propagation of other fatigue related cracks in concrete
pavements, designed based on the mechanistic approach, led to the development of
mechanistic-empirical approaches in concrete pavement design guides.
While the mechanistic part of design guides provides required information on

calculation of the critical stresses and deflections in pavements, the empirical part
specifies possible failure modes of pavements under the applied loads. The applied
loads can be vehicular and/or environmentally related. Vehicular loads have been
considered as static loads in concrete pavement design guides but in reality, they are
dynamic in nature as their locations upon pavement change with time.
Magnitude and configuration of vehicular loads together with environmental effects

have a significant effect on induced tensile stresses within concrete pavements (Yu et
al. 1998 and Hiller and Roesler 2002). Since a variety of axle group configurations is
employed in heavy vehicle industries and across countries, further study is required to
determine the interrelationship between concrete pavement distresses and pavement
responses to the applied loads. Furthermore, the fact that structural responses of
concrete pavements may be affected by the frequency and speed of vehicular loads
1
(Izquierdo et al., 1997) has not been yet considered in concrete pavement design
guides.
1.2. Research Problems

The most serious problem in concrete pavements is crack generation and propagation
(Hossain et al., 2003). Cracking occurs in the first days after placement due to plastic
shrinkage and the reaction between cement and aggregates in the setting time (Nawy,
2001). It then spreads over the pavement surface and propagates deeper in the concrete
due to external factors such as drying shrinkage and fatigue. Although the use of
appropriate curing methods could reduce the development of early edge cracking in
pavement, current concrete pavement guides normally require the provision of
reinforcement and / or concrete joints, discussed later in Chapter 3 of this thesis, to
minimise the development of secondary cracks in the pavement surfaces.
Since pavement performance under the real conditions is not necessarily the same as
predicted performance in the guides, questions arise as to whether the specified
loading combinations in the guides are sufficient to ensure performance. As a result,
a vast amount of research on effects of different factors, such as environmental
effects, has been carried out worldwide. Limited research on dynamic response of
concrete pavements has also been performed in the past based on point, wheel, single
axle and tandem axle loadings. Results of dynamic analyses, conducted in the past,
showed that static loads produce greater tensile stresses in concrete pavements than
dynamic loads. Nevertheless, dynamic analysis is significant in the presence of
pavement roughness. Whilst traffic loads are wandered at any location upon the
pavement, central loadings with symmetrical boundary conditions were considered in
the dynamic studies. Furthermore, effects of temperature curling, moisture warping,
subbase, reinforcement and adjacent traffic lanes on dynamic responses of concrete
pavements were not previously addressed.
The fatigue life of concrete is traditionally estimated based on laboratory fatigue tests
of concrete prism beams under one-directional cyclic loading using third point
loading configuration. Since concrete pavements curl upward or downward during
nighttime or daytime temperature gradients, results of the traditional fatigue test may
2
be insufficient as the pavement curvature is not considered during the test. This also
is a shortcoming which needs to be addressed.
1.3. Research Hypothesis

This research is conducted based on the hypothesis that the rate of deteriorations in
concrete pavements is significantly affected by environmental effects, dynamic
effects and nature of bond layer between concrete slab and subbase. Hence, a series
of finite element analyses is performed in the current study to determine the
magnitude of induced tensile stresses and slab deflections within concrete
pavements. Performance and structural behaviour of concrete pavements can then be
evaluated by taking into consideration stress to strength ratio, deflection of concrete
slab, load transfer efficiency and differential deflection across joints. Further
information on these parameters is presented in Chapter 2 of this thesis.
1.4. Thesis scope

This thesis aims to investigate reasons behind fatigue cracking in concrete pavements
subjected to moving axle group loads and environmental effects using finite element
techniques. The environmental forces considered in this research are daytime and
nighttime differential temperature gradients together with shrinkage effects also known
as loss of moisture contents.
Erosion of subbase and subgrade materials are not considered in this research as its
effects can be eliminated by using non-erodible materials in subbase or subgrade
and/or by considering an appropriate drainage system under the pavement to keep the
level of underground water away from the pavement.
1.5. Research Objectives

The main objective of this research is to study the structural behaviour of concrete
pavements under moving axle loads and environmental effects. The
interrelationships between factors affecting performance of concrete pavements are
determined using finite element techniques supported with laboratory and field tests.
In addition, this research aims:
3
• To study the global and local behaviour of concrete pavements under the
applied loads.
• To address effects of different debonding layers on concrete pavement
responses.
• To develop a better understanding of concrete pavement behaviour under
• To specify critical configurations of different axle groups in concrete
pavement analysis.
• To determine the critical position of different axle groups in concrete
pavement analysis.
• To address fatigue related damage in concrete pavements.
• To develop a new method for measuring concrete fatigue life based on
• To experimentally measure structural response of concrete pavements to
moving truck loads.
• To study dynamic response of concrete pavements.
• To specify dynamic amplifications for different axle groups and different
concrete pavements including jointed plain concrete pavement and jointed
reinforced concrete pavement.
• To develop a guide for designing of concrete pavements based on results of
the current research and Australian design guide (Astroads, 2004).
1.6. Thesis layout

This thesis contains eleven chapters. These include:
Chapter 1: Introduction
In this chapter, the research problem, the research hypothesis, the research objectives
and the scope of the thesis are presented together with a brief background on the
development of concrete pavement technology and design guides worldwide.
Chapter 2 : Literature review - Concrete Properties

Since the performance of concrete pavements is affected by the properties of concrete,
the most significant properties of concrete including strength, modulus of elasticity,
4
fatigue, coefficient of thermal expansion and shrinkage are presented in this chapter.
The magnitude of concrete properties is generally estimated based on its compressive
strength due to practical reasons. Hence, the main aim of this chapter is to provide
information on estimation models of concrete properties developed in the past.
Chapter 3 : Literature review - Concrete Pavements

A comprehensive literature review on behaviour and deteriorations of concrete
pavements to date has been carried out to collect adequate information on factors
affecting performance of concrete pavements. This includes information on concrete
pavement cross section, loadings, analysis of concrete pavements, and concrete
pavement distresses. Finally, the 2004 revision of the Austroads and American
Association of Highway and Transportation Official (AASHTO, 2003) methods for
designing concrete pavements is presented. While the Austroads method is based on a
mechanistic approach, the AASHTO method was developed in accordance with a
mechanistic-empirical approach.
Chapter 4: Summary of literature review and research plan

This chapter summarises the literature review and highlights the predominant factors
affecting concrete pavement performance. Shortcomings of concrete pavement
analysis and design performed in the past are determined to clarify the research
problems and to identify the gap in knowledge for concrete pavement design. Research
methodology associated with an appropriate plan is then presented based on problems
and objectives of this research.
Chapter 5: Sensitivity analysis of Austroads concrete pavement design guide

A sensitivity analysis of those factors considered in the Austroads concrete pavement
design guide is presented in this chapter to recognise the interrelationship between
these factors. Results are then compared with relevant research outcomes provided in
Chapter 2 of this thesis to identify shortcomings of Austroads slab thickness design
guide for further improvement.
Chapter 6: Static analysis of jointed plain concrete pavement

Structural response of concrete pavements is affected by a number of factors such as
thickness of concrete slab, characteristics of subbase and subgrade layers, and
5
magnitude, configuration and location of vehicular loads upon the pavement as well as
environmental effects. Since axle configurations and their nominal capacities vary
between heavy vehicle manufacturers and across countries, statistical data gathered
from weigh-in-motion stations across countries are currently employed for better
understanding of traffic loads. The statistical data provides a range of variations for
each individual parameter of load configuration.
This chapter aims to determine the critical dimensions of different axle group
configurations using finite element techniques. For this purpose, the variations of each
individual parameter of axle group configurations are determined based on dimensions
provided in different sources. These sources are concrete pavement guidelines and
relevant research conducted in the past. Furthermore, the concrete pavements are
analysed under a combination of vehicular loads and environmental effects to
determine the critical locations of different axle groups upon concrete pavement.
Finally, effects of different debonding layers on concrete pavement responses are
described.
Chapter 7: Laboratory tests

It has been determined that the results of the proposed research are significantly
affected by a number of concrete properties such as compressive and flexural strength,
drying shrinkage, and fatigue response. These properties are in turn influenced by the
type and proportion of materials in the concrete mix. While a vast knowledge of
concrete properties and mix design is available worldwide, this chapter presents
laboratory outcomes of those factors which significantly affect concrete pavement
performances. These are concrete compressive and flexural strength, shear force
transfer capacity of aggregate and cement paste, and concrete fatigue life.
A new fatigue test setup developed during this research is also presented in this
chapter. It is used to take into consideration effects of daytime and nighttime curling in
the fatigue life of the concrete,
Chapter 8: Dynamic finite element analysis of bonded concrete pavements

Since vehicular loads are dynamic in nature, it is important to study dynamic response
of concrete pavements instead of using static analysis. This chapter presents
6
information on dynamic responses of different fully bonded concrete pavements
subjected to a variety of moving axle groups with speeds varying from 2 km/hr to 110
km/hr. For this purpose, comprehensive finite element models of different concrete
pavements were developed using ANSYS platform.
Two types of concrete pavements including jointed plain concrete pavement and
jointed reinforcement concrete pavement are investigated. The finite element model
contained a traffic lane with four concrete slab panels and two concrete slab panels in
the longitudinal direction for JPCP and JRCP respectively. The traffic lane was
confined at one of its longitudinal edges by a concrete shoulder. Dynamic
amplifications under different axle groups were calculated. These dynamic
amplifications can be directly used in concrete pavement design guides to predict the
maximum induced flexural stresses in concrete pavements under moving loads.
Chapter 9: Experimental study of concrete pavements under moving truck loads

In order to verify the results of the theoretical models and also to provide information
on structural response of partially bonded concrete pavements, a jointed concrete
pavement test section was constructed and tested under moving truck loads. This
chapter presents information on the test section together with results of the field test.
The significant outcomes are then highlighted.
Chapter 10: Dynamic finite element analysis of partially bonded concrete

pavements
This chapter examines the dynamic behaviour of partially bonded concrete pavements
under moving axle group loads. For this purpose, results of the field test were
employed to calibrate the finite element models of partially bonded concrete
pavements using ANSYS platform. The finite element model was then used for further
investigation on dynamic response of partially bonded concrete pavements subjected to
moving axle group loads.
Similar to the models developed in Chapter 7 of this thesis, two types of concrete
pavements including jointed plain concrete pavement and jointed reinforcement
concrete pavement were investigated. The finite element model contained a traffic lane
with three concrete slab panels and two concrete slab panels in the longitudinal
7
direction for JPCP and JRCP respectively. The traffic lane was confined at one of its
longitudinal edges by a concrete shoulder. Dynamic amplifications under different
axle groups were also calculated.
Chapter 11: Development of design guides

To contribute the results of the research presented in this thesis to concrete pavement
technology, a guide is developed for designing of concrete pavements. Based on
results of the current study, different equations were developed in this chapter to
estimate the magnitude of flexural stress in different locations of concrete pavements.
The accuracy of these equations was examined by performing a comparison between
the estimated stresses and corresponding stresses derived from finite element analysis.
A typical example is solved using the guide provided in this chapter. The calculated
slab thickness is then compared with that determined based on Austroads method.
Chapter 12: Conclusion and recommendations

The final chapter of this thesis highlights the most significant contributions of this
research associated with the significant findings of this research. Finally,
recommendations for further studies based on the outcomes of this research are
outlined.
8
Chapter 2
LITERATURE REVIEW – CONCRETE PROPERTIES
2.1. Background
Concrete as a versatile construction material has been used in many civil applications
such as pavement. Since performance and deterioration of concrete pavements can be
related to concrete properties, it is necessary to know how to achieve a better quality of
concrete. Regarding the possible fatigue failure modes of concrete pavements, the most
important concrete properties are concrete strength, modulus of elasticity, coefficient
of thermal expansion, shrinkage and concrete fatigue life. Information on the pavement
failure modes is presented later in Section 3.5 of Chapter 3.
2.2. Concrete strength

The applied loads on concrete pavements are usually transferred to sublayers by
bending actions which subsequently induce a tensile stress at the top or the bottom
surface layer of the concrete slab. If the magnitude of induced tensile stress is equal or
more than the flexural strength of the concrete, cracks as the main reason behind
concrete pavement deteriorations are initiated at the surface layers and then propagated
into the depth of the concrete slab. As a result, and in contrast with other structures
where concrete compressive strength is used to design the concrete members, concrete
flexural strength or modulus of rupture becomes an important concrete property used
in concrete pavement design guides.
Factors affecting concrete strength are (RTA 1991 and Nawy 2001):
- Cement characteristics and content
- Water/cement, liquid/cement, and/or water/cementitious ratios
- Aggregate quality and interaction with cement paste
- Type and percentage of chemical and mineral admixtures
9
- Procedure and mixing time of constituents
- Degree of compaction
- Degree of hydration which is related to curing methods
- Quality control and assurance.
Since a relationship between compressive and flexural strengths of conventional

concrete is known to exist, the former is normally used to estimate the latter for
practical reasons. Note that the performance of a concrete pavement under loads
depends on several factors and consequently monitoring the concrete strength alone is
not sufficient to produce a sustainable pavement.
A variety of formulas have been provided by different concrete design guides to

estimate concrete flexural strength based on concrete compressive strength. Australian
concrete standard, AS3600 (1994), provided the following equation to convert 28-day
concrete compressive strength to 28-day concrete flexural strength:
f r = 0.60 f c′ (MPa) (2-1)
Since the use of Equation 2-1 results in conservative approach in concrete pavement
analysis and deteriorations, the following equations were provided by the Cement and
Concrete Association of Australia (CCAA, 1999) and Austroads (2004) respectively:
f r = 0.7 f c' (MPa) (2-2)
f r = 0.75 f c' (MPa) (2-3)
Nevertheless, higher coefficient was recommended by the American Concrete Institute

(ACI) Committee 363 (1992).
f r = 0.94 f c′ (MPa) (2-4)
A comparison between the aforementioned equations indicates that the concrete

flexural strength are different by 25 per cent and 57 per cent from AS3600 (1994) to
10
Astroads (2004) and ACI (1992) respectively. As already noted, concrete flexural
strength plays a constitutive role in concrete pavement design. A change in concrete
flexural strength can substantially change the critical failure mode of concrete slab as
well as the required slab thickness. Because of these, particular care is taken before
considering the above mentioned equations in the current research.
Austroads (1992) restricts the cylindrical compressive strength of concrete at 28-day

between 32 MPa and 42 MPa to avoid joint faulting and to control the crack width and
spacing respectively. The minimum concrete flexural strength at 28-day has been also
restricted to 4.5 MPa when the design traffic exceeds 1×106 Heavy Vehicle Axle
Groups (HVAGs). Further information on design traffic is presented in Section 2.7.1 of
this Chapter.
2.3. Modulus of elasticity

This property reflects the stiffness of concrete under a static load. The selection of
concrete materials and mix proportions can significantly affect the modulus of
elasticity. The ACI (1992) recommended the following equations for calculating the
modulus of elasticity according to secant modulus at 0.45 f c′ intercept for concrete
with densities, ω c , in the range of 1440 – 2320 kg/m3 :
E c = 0.033ω c1.5 f c′ (MPa) when f c′ ≤ 42 MPa (2-5)
ωc
E c = (3.32 f c′ + 6895)( )1.5 (MPa) when 42 ≤ f c′ ≤ 83 MPa (2-6)
2320
AS3600 (1994) provided the following equation regardless to the range of concrete
compressive strength:
E c = 0.043ω c1.5 f c′ (MPa) (2-7)
2.4. Coefficient of thermal expansion

Concrete, like most engineering materials, has a positive coefficient of thermal
expansion (CTE). The concrete CTE depends on concrete constituents and moisture
contents (Neville, 1983). Since concrete is a mix of cement paste and aggregates, the
11
thermal coefficient of concrete relies on thermal coefficient of each of these
components. Cement paste has a coefficient of thermal expansion between 11×10-6 and
20×10-6 mm/mm/˚C which is greater than aggregate thermal coefficient (Meyers,
1951). Type and percentage of aggregates in concrete mix associated with curing
method can affect the magnitude of coefficient of thermal expansion in concrete as
thermal movement of the cement paste is restrained by aggregate. Table 2-1 provides
information on this matter based on the work of Bonnell and Harper (1951).
The following equation has been recommended for estimating concrete coefficient of
expansion ( α C ) based on cement paste ( α p ) and aggregated ( α g ) thermal coefficients
(Neville and Brooks, 1987):
2 g (α p − α g )
αc = α p − (2-8)
kp kp
1+ + g (1 − )
kg kg
kp
Where g is volume of aggregate in concrete mix, is the stiffness ratio of cement
kg
paste to aggregate which can be assumed as the ratio of their modulus of elasticity.
2.5. Shrinkage
Three types of shrinkage have been known, namely, plastic shrinkage, drying
shrinkage, and carbonation shrinkage (Nawy, 2001). Plastic shrinkage happens during
the first hours after placing fresh concrete in the forms. It is affected by the ratio of the
surface area to the thickness of the concrete elements. An increase in this ratio boosts
the plastic shrinkage due to the large contact surface with dry air.
Drying shrinkage, on the other hand, occurs during the final setting of concrete when
the cement hydration process is nearing its completion. It is a decrease in the volume of
the concrete element due to the evaporation of moisture. Drying shrinkage is affected
by aggregate content, water content, concrete element size, ambient conditions,
reinforcement, admixtures and cement type.
12
Table 2-1. Concrete coefficient of thermal expansion in a concrete with cement to
aggregate ratio of 1:6 for different aggregate types and curing methods
This table is not available online.

Please consult the hardcopy thesis
available from the QUT Library
Building Research Establishment, Crown copyright
In addition to the above mentioned factors, water to cement ratio, void or air content,
and degree of hydration may influence the magnitude of drying shrinkage. Drying
shrinkage could cause severe problems in concrete pavements (RTA, 1991). An
increase in shrinkage will directly influence the severity of cracks, joint width and load
transfer capacity across joints. Due to the possible shrinkage related problems in
concrete pavements, an upper limit of shrinkage may need to be specified for both
concrete slab and subbase.
Reaction between carbon dioxide (CO2) in the atmosphere and calcium hydroxide in
the cement paste causes carbonation shrinkage. The process of carbonation is
significantly reduced at relative humidity below 50 per cent (Nawy, 2001).
2.6. Fatigue
Fatigue is a failure type of structural elements due to cyclic loads when the magnitude
of applied load is not large enough to cause any type of failure in the elements. The
fatigue property of materials is usually defined by S-N curves, where S is the stress
ratio and N is the number of loading cycles. S is a proportion of current stress and
ultimate strength of material.
13
Vehicular loads are dynamic in nature. They wander at the top surface layer of the
pavement and are repeatable. Repeated loads such as cyclic loads result in fatigue
failure of the concrete under a load less than its flexural strength. At the fatigue failure
point, the strain of the member increases and the modulus of elasticity decreases
(Karimov, 2004). Figure 2-1 shows the typical S-N curve adopted in the Austroads
design Guide (2004) based on a variety of concrete flexural stress ratio.
Mechanical properties of the hardened concrete and its components, environmental

effects and loading conditions can affect concrete fatigue life. Kleiber and Lee (1982)
showed that there is a direct relationship between the water to cement ratio and the
fatigue life of plain concrete with flexural behaviour so that a decrease in the water to
cement ratio of concrete reduces the fatigue strength of the concrete.
This figure is not available online.

Figure 2-1. Fatigue relationship adopted in Austroads design model (Austroads, 2004)
Kim and Kim (1996) in an experimental study on fatigue life of high strength concrete,
based on a compressive cylindrical test of 160 specimens (100 mm diameter and 200
mm height) with different compressive strength ranging from 26 MPa to 103 MPa and
four levels of maximum stress ratio (75%, 80%, 85% and 95%), showed that an
increase in compressive strength decreases the concrete fatigue life under the same
stress ratio (Fig. 2-2).
Croney and Croney (1998) studied the effect of different aggregate types on concrete
fatigue life subjected to repeated flexural loads. Two common aggregate types, gravel
aggregate and crushed rock, were investigated. In this study, 28-day concrete
compressive strength varied from 10 MPa to 60 MPa. The maximum applied tensile
14
stress ranged from 1.53 MPa to 4.45 MPa for gravel aggregate and 1.58 MPa to 6.08
MPa for crushed rock depending on the compressive strength of specimens. Three
levels of stress applications per day (5, 50 and 500) were assumed. The results (Fig. 2-
3) show that the use of crushed rock can improve the concrete fatigue life.
60000
50000
40000
Fatigue Life
30000
20000
10000
0
26 52 84 103
Concrete Compressive Strength (MPa)
Figure 2-2. Variation of fatigue life with concrete compressive strength

(stress ratio= 75%)
Ramakrishnan and Lovik (1992) showed that flexural fatigue strength of plain concrete
can be also promoted by using fibre reinforcement. Furthermore, hooked end steel
fibre produces a greater flexural fatigue strength compared with other type of fibres.
The use of 0.5 per cent hooked end steel fibre per concrete volume increases the
fatigue life of concrete by about 25 per cent (Fig. 2-4).
An appropriate fatigue equation is required to predict the fatigue damage of concrete

pavements under repeated loads. Several equations were developed in the past based
on laboratory tests of prism beams using third point loading configuration. Vesic and
Saxena (1969) and Majidzadeh and Ilves (1983) developed different power law
equations for predicting fatigue damage of concrete slabs (Equations 2-9 and 2-10
respectively). Westergaard theory and plate theory were respectively used in their
research to calculate vehicular induced stresses in concrete pavements. Trybig et al.
(1977) developed another power law model known as the Austin Research Engineers
(ARE) model using elastic layer theory (Equation 2-11). Dater (1977) developed a
15
fatigue damage prediction model based on data obtained from 140 beam tests
(Equation 2-12).
Number of stress application 5000

4500
4000
3500
(Millions)
3000
2500
2000
1500
1000
500
0
10 20 30 40 50 60
Concrete Compressive Strength (MPa)
Gravel aggregate Crushed rock
Figure 2-3. Variation of fatigue life with concrete compressive strength for gravel
aggregate and crushed rock (applied tensile stress was increased from 0.645 × fr to
0.701 × fr and from 0.666 × fr to 0.961 × fr, corresponding to an increase in compressive
strength from 10 to 60 MPa, respectively).
In 2005, Suh et al. contributed effect of aggregate types on fatigue damage predication
model. Consequently, two equations for different aggregate types, siliceous river
gravel (Equations 2-13) and crushed limestone (Equations 2-14), based on laboratory
data obtained from 76 tests were developed. The equations were then validated using a
full scale field test. Information on the Austroads fatigue equation is presented later in
Section 3.6.1 of chapter 3.
σ 
Log N f = Log (225000) − 4 Log   (2-9)
 fr 
σ 
Log N f = Log (22209) − 4.29 Log   (2-10)
 fr 
16
σ 
Log N f = Log (23440) − 3.21Log   (2-11)
 fr 
σ 
Log N f = 16.61 − 17.61  (2-12)
 fr 
σ 
Log N f = 13.22 − 11.57  (2-13)
 fr 
σ 
Log N f = 17.33 − 16.29  (2-14)
 fr 
Where N f is the allowable load repetition, σ is maximum induced tensile stress at the
bottom surface layer of the concrete slab along the wheel path, f r is as previously
defined.

Figure 2-4. Effect of different types of fibres on concrete fatigue flexural strength
(Nawy, 2001)
2.7. Summary
The most significant concrete properties have been discussed in this chapter. The
factors affecting each individual property of concrete have been presented. Since the
concrete properties are estimated based on its compressive strength for practical
reasons, typical equations developed in the past for estimation of concrete properties
17
based on its compressive strength were presented. A comparison between developed
equations for estimation of concrete properties, i.e. flexural strength, indicated that a
particular care shall be taken into consideration to select the appropriate equation.
Consequently, a series of laboratory tests needs to be considered to determine the most
appropriate equations for the estimation of concrete properties. Information on the
laboratory tests is described in Chapter 7 of this thesis.
18
Chapter 3
LITERATURE REVIEW – CONCRETE PAVEMENTS
3.1. Background
Structural responses of concrete pavements are affected by a number of factors
including distance between joints, thickness of concrete slab, concrete properties, load
transfer devices, joint width, boundary condition between concrete slab and subbase,
subbase type, thickness and properties of subbase, subgrade characteristics,
environmental effects and configuration, magnitude and location of vehicular loads
upon pavement.
Concrete pavement in Australia contains a number of relatively thin concrete bases,

known internationally as concrete slabs, finite in length and width, over a subbase
resting on a foundation soil known also as subgrade (Fig. 3-1). The width of concrete
slab is equal to the width of traffic lane, normally 3600 mm. However, the length of
concrete slab varies depending on concrete pavement types. Concrete slabs are
connected to each other or shoulder in the longitudinal and transverse directions
through transverse and longitudinal joints respectively. Joints allow them to expand /
shrink with temperature fluctuations. A debonding layer is placed between the concrete
slab and the subbase to reduce early age cracking.
Figure 3-1. Cross section of a typical concrete pavement in Australia
Factors affecting the performance of concrete pavements are categorised in six areas
including concrete properties, concrete pavement cross section, loadings, concrete
19
pavement analysis, concrete pavement distresses. Since information on concrete
properties was provided in Chapter 2 of this thesis, other factors are discussed in this
chapter. Subsequently, Austroads (2004) and AASHTO (2003) design guides as
examples of mechanistic and mechanistic-empirical approach are presented. A
summary of the literature is finally presented.
3.2. Concrete pavement cross section

3.2.1. Subgrade
Subgrade is the foundation soil which is compacted to achieve an appropriate strength.
While soils also exhibit some nonlinear behaviour and other interesting characteristics,
a full investigation of this layer is beyond the scope of this research. Because of its
strong effect on concrete pavement responses, however, a brief background of soil
modelling idealization as pavement foundation is illustrated in this section.
A variety of subgrade model idealizations including Winkler, Dense Liquid (DL),

Elastic Solid (ES), Two Parametric (TP), Zhemochkin-Sinitsyn Staerman (ZSS), and
Kerr-Vlasok (KV) have been developed during the past centuries. Only, those
employed widely in relevant research are subsequently discussed here.
Dense Liquid (DL), Spring, and Winkler Foundation

The DL is the simplest model of soil idealizations. It was developed by Winkler (1864)
and Westergaard (1948). Modulus of subgrade reaction is the only required parameter
which is assumed to be constant during soil deflection under the applied load. The
subgrade is modelled as a series of independent springs in this method (Fig. 3-2).
However, development of finite element programs requires the use of spring model
idealization (Fig. 3-3). Hence, the DL foundation was replaced with concentrated
springs at the nodes of elements. The equivalent stiffness of the spring can be
calculated from the following equation:
K=A.k (3-1)
Where K is stiffness of the equivalent spring, A is the area of the finite element, k is
the parameter of the DL model known as modulus of subgrade reaction.
20
Figure 3-2. DL subgrade idealization (ISLAB, 2004)

Figure 3-3. Spring model idealization (ISLAB, 2004)
The spring idealization is extensively used worldwide due to its simplicity. The
problem with this modelling is related to deflection due to bending in the element as
shown in Figure 3-4. The method is only able to consider vertical deflection at the
nodes. Hence, if the plate exhibits bending between the nodes without any vertical
deflection at the nodes, the spring model does not show any resistance to the
deformation. Nevertheless, consideration of very fine meshed elements will alleviate
the negative effect of this problem.

Figure 3-4. Limitation of the spring model (ISLAB, 2004)
Due to its limitation, the modified Winkler finite element idealization replaces the DL
foundation. A rotational spring was added to each node of the finite elements to
consider the bending deflection of the plate (Fig. 3-5). The stiffness of these springs is
defined based on the equality of potential energy between their deformation and
deformation of the original DL model.
21
Figure 3-5. Winkler modified idealization (ISLAB, 2004)
Elastic Solid Foundation

The semi-infinite elastic solid (ES) or Boussinesq subgrade is one of the realistic
representations of real soil. A simple factor (C), such as k value in the DL method, is
used in the foundation model and can be determined as follows:
Es
C= 2
(3-2)
1 −ν s
Where Es is the elastic modulus and ν s is Poisson’s ratio of the subgrade.
Hence, deformation of the model is not only under the plate but also beyond it as
shown in Figure 3-6. In other words, deflection at any point depends on the forces at
that point and also on the forces or deflections at other points.

Figure 3-6. Elastic solid (ES) idealization (ISLAB, 2004)
AASHTO (2003) shows that both DL and ES idealization exhibit discrepancies with
the result of real conditions due to ignoring of shear strength in the subgrade soil
between springs.
The Two-Parameter Foundation (TP)

This model, known also as the Pasternak (1954) or Vlasov (1966) subgrade, actually is
an attractive modification of the ES by providing a degree of shear interaction between
22
adjacent soil elements. A shear layer has been considered at the top of the spring and
bottom of the plate in the DL idealization, as shown in Figure 3-7.
Consequently, an inter relationship between subgrade reaction pressure (q), surface

deflection (w), and modulus of subgrade reaction (k) was established.
q = kw − G∇ 2 w (3-3)
Where G is a coefficient describing the interaction between adjacent springs and ∇ 2 is

the Laplace operator.

Figure 3-7. TP model idealization (ISLAB, 2004)
Pronk (1993) in a comparison between DL and TP recommended that the use of TP

can be presented as a logical improvement of DL. It also increases the speed of
alleviation of the deflection profile more than the ES idealization. Hence, it can
provide a better approximation of the deflection in a real foundation.
Kerr-Vlasov Three-Parameter Foundation (KV)

Kerr (1964) recommended the use of another physical interpretation of this model as a
shear layer placed above a TP foundation (Fig. 3-8). The method is based on several
analytical solutions resulting from different practically important boundary conditions.
The KV foundation leads to a banded stiffness matrix, without predicting infinite soil
pressures under the free edge of the slab, as ES and TP would.
To determine the stiffness matrix for KV, a special 8-noded, 24-degree-of-freedom

element was introduced. The first four nodes are placed at the top of the upper (DL)
springs, while the other four nodes are positioned at the top of the TP foundation (at
the bottom of the DL springs). The stiffness matrix for this element is:
23
[K ] + [K DLU ] − [K DLU ] 
[K ] =  PL (3-4)
 − [K DLU ] [K DLU ] + [K TP ]
Where [KDLU] is the stiffness matrix of the upper (DL) springs and [KTP] is the stiffness
matrix for TP.

Figure 3-8. Kerr model idealization (ISLAB, 2004)
This model requires three material parameters, kU, G, and kL, where kU and kL are the
upper and lower spring stiffnesses, respectively.
Drucker-Prager model
Drucker-Prager (Drucker and Prager, 1952 and 1957) is a yield criterion, where the
yield surface does not change with progressive yielding. As a result, no hardening
rule is considered in this model and material is assumed to be elastic-perfectly
plastic. The Drucker-Prager criterion can be expressed as:
α DP ⋅ I1 + β DP = J 2 (3-5)
Where I1 is the first invariant of the stress tensor and can be calculated from
Equation 3-6, J2 is the second invariant of the stress deviator tensor and can be
calculated from Equation 3-7.
I 1 = σ 1 + σ 2 +σ 3 (3-6)
J2 =
1
6
[
(σ 1 − σ 2 )2 + (σ 2 − σ 3 )2 + (σ 1 − σ 3 )2 ] (3-7)
24
The Drucker-Prager material constants, α DP and β DP , can be determined by
contributing the Mohr-Coulomb criterion into the method. As a result:
2 Sinϕ
α DP = (3-8)
3 (3 − Sinϕ )
6CIF ⋅ Cosϕ
β DP = (3-9)
3 (3 − Sinϕ )
Where ϕ and CIF are angle and coefficient of internal friction respectively.
The material fails under the applied load if the magnitude of failure function
becomes less than or equal zero. The failure function is:
FDP = α DP ⋅ I1 + β DP − J 2 (3-10)
Further information on this model can be found in Drucker and Prager (1952),
Drucker et al. (1957), Erlicher and Point (2005) and ANSYS Help (2006).
3.2.2. Subbase
This layer is constructed over the subgrade and under concrete slab and shoulder.
Austroads (2004) recommends the use of a lean-mix concrete (LMC) or a bonded
subbase with a characteristic 28-day compressive strength of not less than 5 MPa.
Bonded subbase includes cement stabilised crushed rock, dense-graded asphalt or
rolled lean concrete (Austroads, 2004).
The subbase has a thickness of 125 to 150 mm (Austroads, 2004). The aims of
producing a subbase under the concrete slab are:
• To provide a stable platform for construction equipment.

• To provide a uniform support under concrete slabs.
• To reduce joint deflection, and increase the aggregate interlock that influences
load transfer across joints and cracks.
• To control the shrinkage and swelling of subgrade soil.
• To prevent pumping of subgrade materials at joints, cracks and slab edges.
25
3.2.3. Debonding layer
In terms of the classical friction model, the magnitude of interface frictional force
between two layers depends on the smoothness of the contact surface presented as
coefficient of friction and the normal force applied to the sliding plane. However,
classical friction model can not be used in concrete pavement systems as friction force
between concrete slab and subbase depends on different components including
adhesion, shear and bearing (Wimsatt et al. 1987 and Wesevich et al. 1987). This
results in a greater friction between the concrete slab and the subbase than that
calculated based on the classical friction model.
If a full bonded condition between the concrete slab and subbase is considered, a
tensile stress will be produced at the interface of the concrete slab and the subbase
during the first 28 days of concrete placement due to plastic and drying shrinkage. This
subsequently results in early age cracking. Because of this, a debonding layer is placed
between the concrete slab and the subbase to eliminate the above mentioned problem.
However, low friction may cause unpredictable pavement movements under the
applied loads. Joint configuration and volume of reinforcement in the case of
reinforced concrete pavements are determined based on the magnitude of friction force
between the concrete slab and the subbase.
There are several methods to reduce the friction force in concrete pavements such as
the use of polythene sheets or 40 mm asphalt bond breaker (Suh et al., 2002).
Depending on the type of concrete pavements, lean-mix concrete curing and
debonding treatments, Austroads (2004) recommends a coefficient of friction of 1.5 to
3 is adopted between the concrete slab and the subbase. The use of a single layer
polythene sheet can reduce the coefficient of friction to 1.2 (Suh et al., 2002).
Instead of the use of coefficient of friction, Wimsatt et al. (1987) and Wesevich et al.
(1987) introduced the concept of frictional stress in concrete pavement analysis. The
frictional stress is a shear stress induced at each square metre of concrete slab and
subbase interface. This parameter is highly independent of concrete slab thickness and
bearing stress. Figure 3-9 shows variation of frictional stress versus concrete slab
26
slippage for a concrete slab resting on different subbase based on experimental tests
carried out by Zhang et al. (2001).
Note that different boundary conditions including bonded, unbonded and partially
bonded may be created between the concrete slab and the subbase depending on the
magnitude of friction force. While the bonded boundary condition keeps the concrete
slab and subbase together with no vertical separation, a fully unbonded boundary
condition allows them to be separated under tensile force without inducing any
frictional force between these layers. A partially bonded boundary condition, on the
other hand, keeps the concrete slab and the subbase together for a certain frictional
force. Beyond this frictional force, a vertical separation will occur between these
layers.

Figure 3-9. Typical experimentally determined frictional stress versus concrete slab
slippage for different materials in subbase (Zhang et al., 2001)
An unbonded condition could only be achieved by using a double layer of

polyethylene sheets (Tarr et al., 1999). On the contrary, Yu et al. (1998) stated that
friction between the concrete slab and the subbase is sufficient to produce bonded
behaviour even if polyethylene sheets are placed between them.
27
3.2.4. Concrete slab (base)
The concrete slab is the top layer in the pavement section (see Fig. 3-1). It contains a
concrete with a compressive strength more than 32 MPa (Austroads, 2004) to ensure
the durability of the wearing surface and to provide sufficient flexural strength to avoid
unpredictable deteriorations in pavements. Although the concrete slab thickness was
traditionally considered to be between 200 mm and 250 mm (RTA, 1991), it is now
designed based on damage processes considered in concrete pavement design guides.
Thickness of concrete slab is affected by the fatigue flexural strength of the concrete,
type of joints, the value of load transfer efficiency (LTE), availability of the shoulder,
strength of foundation soil and subbase, erosion of the subbase and subgrade materials,
environmental effects and expected traffic load. In accordance with the provision of
transverse joint (discussed later in Section 3.2.7 of this chapter) and the availability of
reinforcement, concrete pavements are classified as Jointed Plain Concrete Pavement
(JPCP), Jointed Reinforced Concrete Pavement (JRCP) and Continuously Reinforced
Concrete Pavement (CRCP).
3.2.4.1. Jointed Plain Concrete Pavement (JPCP)

JPCP has a natural configuration compared with other concrete pavements. In this
configuration, the joint’s pattern closely resembles natural cracks in an unreinforced
concrete pavement. Distance between longitudinal joints depends on traffic lane
configuration and vary from 3.6 m to 4.6 m. Transverse joints, on the other hand, are
located 4.2 m or 4.5 m apart from each other in undowelled skewed joints or in
dowelled square joints respectively (Austroads, 2004). Distance between transverse
joints is affected by joint and concrete slab integrities. Since slab length is much more
affected by joint integrity than slab integrity, transverse joints of concrete pavement
with more than 4.2 m length are modified with dowel.
3.2.4.2. Jointed Reinforced Concrete Pavement (JRCP)

JRCP is the traditional configuration of joints in concrete pavement which has been
widely used in Sydney’s arterial routes such as the Warringah Freeway, Parramatta
and Victoria Roads, and the Princess Highway (RTA, 1991). Whilst the distance
between longitudinal joints is selected based on width of traffic lane, transverse joint
28
spacing ranges from 8 m to 15 m in the JRCP. Mesh reinforcements have been
commonly used in the JRCP. The size of reinforcement is proportional to the concrete
slab cross-section. It should be noted that steel ratio increases with an increase in the
concrete slab length (CCAA, 1991). The reason behind the use of reinforcement in
JRCP is to minimise the width of induced thermal curling cracks. Most JRCPs show a
good performance over long periods (AASHO, 1962).
3.2.4.3. Continuously Reinforced Concrete Pavement (CRCP)

CRCP has reinforcement along its length except at discontinuities such as bridge
abutments (CCAA, 1991). There is a balanced relationship between the steel ratio,
which typically is 0.67 to 0.72 per cent of concrete slab volume, and the concrete
strength. This limits random transverse crack spacings to be between 1 m and 2 m in
each square meter of the concrete slab. Additionally, crack width is restricted to be less
than 0.3 mm. The most important benefits in the use of CRCP are:
• It has lower maintenance cost due to the absence of transverse joints.

• It is suitable for areas with differential settlement because it is continuously
reinforced.
• Compared with other concrete pavements, it is compatible with asphalt
wearing because of the absence of transverse joints and minimal movement at
the transverse cracks.
3.2.5. Surface roughness

Surface roughness is a longitudinal profile of pavement surface elevations in 305 mm
intervals (Huang, 2004). In accordance with ASTM E867, pavement roughness is
“the deviation of a surface from a true planar surface with characteristic dimensions
that affect vehicle dynamics and ride quality”. Pavement roughness is normally
measured based on International Roughness index (IRI) to provide a more realistic,
practical and comparable approach. The IRI is a longitudinal surface profile along
the wheel path of the pavement. It is calculated based on surface elevation data
collected from either mechanical profilometer or a topographical survey. The method
for calculation of the IRI can be found elsewhere (Sayers et al., 1986). It is presented
in units of m/km.
29
Concrete pavement has a smooth surface when it is relatively new but its surface
becomes rougher during its performance under the traffic loads. As a result, a rough
surface can be observed at the end of pavement service life (Bhatti and Stoner,
1998).
3.2.6. Shoulder
The positive effects of shoulders in structural response and performance of concrete
pavements have been accepted worldwide. Bituminous shoulders when placed next to
a concrete pavement will experience further compaction from the traffic after road
opening. Consequently, a vertical gap between the top surfaces of concrete slab and
shoulder is produced which results in a loss of restrain at the longitudinal edge of the
concrete slab. This ultimately results in deteriorations of concrete slab. As a result,
concrete shoulders have been widely used in concrete pavements to prolong the
pavement life. The Austroads (2004) concrete pavement design guide allows a
reduction in concrete slab thickness when a concrete shoulders is utilised. For this
purpose the definition of a shoulder is:
• A tied shoulder with at least 1.5 meters width from the traffic lane, or
• An increase in width of the traffic lane equal to 600 mm.
In contrast to the practice in the past decades, concrete shoulders are now often
constructed as strong as the concrete slab due to their structural contributions, which
include the reduction of induced stresses in concrete slab, moisture control, and
construction expediency.
3.2.7. Joints
Joints are usually utilised to reduce effects of climatic forces on the concrete slab. The
climatic forces are due to shrinkage-loss of moisture contents and temperature
gradients through depth of the concrete slab. Load transfer devices and sealant are the
main components of joints (Fig. 3-10). In terms of the economical point of view, the
distance between joints should be long enough to minimise the number of load transfer
devices but short enough to eliminate transverse cracking (Kelleher and Larson, 1989).
30
A variety of joints namely, isolation joints, contraction joints, construction joints and
expansion joints is used in pavement constructions (CCAA, 1999). Byrum and Hansen
(1994) indicated that joint opening is the key factor in stress distribution around the
joint.

Figure 3-10. Typical expansion joint (CCAA, 1978)
3.2.7.1. Isolation joint

The isolation joint allows elements, which have been located on opposite sides of
joints, to work independently and separately (Fig. 3-11). It should be employed in all
construction where movement or vibration of one structural element can strongly
influence the structural behaviour of the adjacent elements. For instance, isolation
joints are used in machinery foundations to isolate them from other construction parts
because of vibration.

Figure 3-11. Typical isolation joint (CCAA, 1999)
3.2.7.2. Contraction joint

The contraction joint is particularly used in concrete pavements to control the random
drying shrinkage cracking by inducing the concrete slab to crack at the contraction
joints. In other words, the induced stress due to contraction or expansion, which can
generate a crack in the concrete pavement, will be relieved by allowing the pavement
31
to move independently between contraction joints. The concrete slab is weakened in
the contraction joint by forming or cutting a groove to ensure that shrinkage cracking
occurs at the contraction joints. Concrete pavements are normally categorised
according to configuration of transverse contraction joints in the concrete slab.
3.2.7.3. Construction joint

Construction joints are used to separate areas of concrete placed at different times. The
construction joints may perform in the longitudinal and transverse directions of the
traffic lane.
3.2.7.4. Expansion joint

Expansion joints are used in constructions where the ratio of length to width is high
and effect of the depth can be ignored. The main reason for utilizing a expansion joints
is to avoid failure due to thermal and moisture movement in the structural elements.
Expansion joints are widely employed in bridges and long length or width structures.
3.2.8. Load Transfer Devices

Aggregate interlock and mechanical devices may be employed to transfer the shear
force across joints. Mechanical load transfer devices are placed within the joints to
physically transfer the applied shear force between slabs on the opposite sides of the
joint. Dowel bar and tie bar are examples of common mechanical devices that are
widely used in pavement constructions compared with other mechanical devices such
as double-vee shear devices and studded plates.
3.2.8.1. Aggregate Interlock

Aggregate interlock is one of the traditional mechanisms to transfer shear force across
joints by friction and bearing of aggregate particles. The friction coefficient is directly
affected by the use of rough or angular aggregates in the concrete. An increase in the
friction coefficient enhances the value of the load transfer efficiency (LTE) – discussed
later in Section 2.3.9 of this chapter. However, aggregate interlock becomes less
reliable as its performance strongly depends on the gap between adjacent slabs (joint
width).
32
3.2.8.2. Dowel
Dowel has a fundamental role in transverse joints particularly when the pavement is
not reinforced. Dowels are typically round with a diameter of 1 8 of concrete slab
thickness and 350 to 460 mm long. Dowels are placed at mid-depth of the base with an
even space of 300 mm centre to centre and positioned perpendicularly to the transverse
joints (FHWA, 1983). Since one side of the dowel is always coated by a debonding
layer, longitudinal movements of the concrete slabs on both sides of the transverse
joint were not restrained.
Standard mild steel epoxy-coated dowel bars, fibre reinforced polymer (FRP), solid
stainless steel dowels, grouted stainless steel dowels, stainless steel clad dowels, and
stainless steel pipe dowels are employed worldwide. Dowels with other cross-sections
including I-beam, oval dowels, and flat plate are also used to decrease the bearing
stress in the concrete slab under vehicular loads by increasing the bearing area.
Covetti and Bishoff (2001) studied the constructability and potential cost-effectiveness
of three kinds of dowel materials for a variety of slab thicknesses. The dowels used in
the study were fibre reinforced polymer (FRP) composite dowels, solid stainless steel
dowels and hollow-core mortar-filled stainless steel dowels. The results indicated:
• Load transfer efficiencies were reduced in all test sections particularly in the
FRP composite dowel test sections.
• A general uniformity of load transferring among sections, showed by Ride
quality surveys.
Moreover, Glass Fibre Reinforced Polymer (GFRP) dowel bars are a possible
maintenance-free alternative in corrosive environments. They potentially reduce the
overall life-cycle-cost of pavements (Eddie and Shalaby, 2001).
Harvey et al. (2003) and Bischoff and Toepel (2004) in different Dowel Bar Retrofit
(DBR) plans showed that joint performance is not affected by types of dowel bars. The
DBR is a technique to rehabilitate jointed concrete pavements where joint faulting is
the problem. Byrum and Hansen (1994) showed that there is an optimum dowel size
depending on the applied load and boundary condition of concrete pavements.
33
3.2.8.3. Tie bars and keyed joints
The tie bar works in a similar way to dowel in transverse joints but it is used in
longitudinal joints. Its performance is also not as strong as that of dowel. Keyed joints
(Fig. 3-12) or a combination of tie bar and keyed joints may be employed in the
longitudinal joists depending on the subgrade and subbase strengths and density of
vehicular loads passing along the pavement close to the longitudinal joints.

Figure 3-12. Typical keyed and tied longitudinal joints (CCAA, 1999)
3.2.9. Load Transfer Efficiency (LTE)

LTE is an important factor in performance of transverse joints and cracks. It is
measured to evaluate joint operation under the applied load. It can be calculated by the
following equation:
δU
LTE (%) = * 100 (3-11)
δL
Where δU is deflection of the unloaded slab and δ L is deflection of the loaded slab
(Fig. 3-13). LTE is affected by dowel size, aggregate interlock, width of joints, and
subbase or subgrade strength.
Based on results of Harvey et al. (2003), the LTE did not change by changing the
traffic volume on the sections reinforced with DBR. Furthermore, the LTE was less
sensitive to temperature changes. In addition to the use of suitable dowels, the
improvement of subgrade strength can improve the LTE as reported by Hossain and
Wojakowski (1996).
34
Figure 3-13. Typical components for calculation of LTE in a transverse crack
3.2.10. Differential Deflection

Differential deflection is another important factor, which provides further information
on joint performance. It is the value of difference between the edges of adjacent slabs
and presents the vertical distance between the loaded and unloaded slabs on the
opposite sides of the joint. It is computed by using Equation 3-12.
DD = δ L − δU (3-12)
It is noteworthy that LTE does not correlate with the amount of differential deflection.
In other words, different values of the LTE can result in the same value of the DD. The
differential deflection defines the sensitivity of the pavement to impact loads, applied
at the edge of transverse joints of unloaded slab. The impact load results in further
deterioration or joint faulting. Hence, the DD becomes an important factor when
dynamic behaviour of rigid pavement is investigated (Popehn et al., 2003).
3.3. Loadings
Traditional methods of concrete pavement analysis were only based on vehicular
loads. However, Byrum and Hansen (1994) showed that highway slabs are
predominantly in the upward curled condition. Environmental effects together with
built-in temperature curl result in different failure types of concrete pavements (Byrum
and Hansen 1994, and Hiller and Roesler 2005). Environmental effects considered in
concrete pavement analysis are temperature and shrinkage-loss of moisture content.
35
3.3.1. Traffic loads
The traffic on the roads contains a large range of vehicular loads from bicycles to triple
road trains (Austroads, 2004). Since pavement deteriorations are not affected by light
vehicles, only heavy vehicles are considered in the pavement design process. It is
known that under vehicular loading, the structural response of concrete pavements is
affected by the choice of axle configurations and magnitude of the applied wheel
loads. Since axle group and loading configurations vary among vehicle manufacturers,
it is not surprising to know that research on concrete pavements carried out worldwide
has been based on a variety of vehicular configurations. Without considering the
choice of vehicular load characteristics, the results of such a pavement analysis may
not be applicable to other pavement conditions.
Truck speed, applied load frequency, traffic wander, axle group type, suspension
system, distance between axles in a given axle group, axle width, distance between the
centres of dual tyres, tyre inflation pressure, tyre-pavement interface shape and stress,
and surface roughness are the main factors affecting structural response of concrete
pavements to static and moving vehicular loads. Surface roughness was previously
described in Section 2.3.5 of this document. Traffic wander upon the pavement has a
significant effect on concrete pavements. Therefore, critical axle group locations upon
pavement were determined in the past to capture maximum pavement response.
Vehicular induced tensile stress occurs at the deepest surface layer of the concrete slab
particularly when the load is applied at longitudinal joints (Ongle and Harvey, 2004).
Truck speed and load frequency

Speed and frequency of vehicular loads are two significant factors that may affect
dynamic response of concrete pavements. Allowable track speed varies from country
to country. Research of Achenbach et al. (1967) showed that the critical truck speed is
between 85 and 100 percent of propagation velocity of a transverse displacement wave
through the pavement system. In Australia, maximum truck speed on the highways is
restricted to 120 km/h.
Evidence shows that the load frequency is between 0 and 20 Hz based on frame
bending vibration mode frequency for trailers and tractors (Gillespie et al., 1993),
36
and between 2 and 15 Hz based on truck’s suspension vibration frequency
(Monismith et al., 1998). Gillespie et al. (1993) also found that for trucks, load
frequency is 4.6 Hz for a speed of 58 km/h and 6.5 Hz for a speed of 82 km/h.
Traffic wander
AASHTO (2003) showed that the number of transverse cracking in the pavement is
rapidly increased by a decrease in distance between outer tyre edge and pavement edge
(longitudinal joints) for truck passing along the pavement (Fig. 3-14).

Figure 3-14. Effect of lowering distance between outer edge of tyre imprint and
pavement edge on JPCP transverse cracking (AASHTO, 2003)
Packard and Tayabji (1985) assumed that only 6 per cent of the traffic passes along the
edge area of the traffic lane (Fig. 3-15). The edge area is defined in a transverse
distance of 600 mm from longitudinal joints or edges. However, the work of Lennie
and Bunker (2005) showed the volume of the traffic passing along the edge area in the
state of Queensland is much higher than the above mentioned assumption.
Location of axle group on pavement

Packard and Tayabji (1985) in their research on a confined single lane pavement under
single and tandem axle loads showed that the critical location of axle groups upon
pavement is at the middle of longitudinal joint between transverse joints to capture
37
maximum tensile stress (Fig. 3-16a), and next to corner (Fig. 3-16b) to measure the
maximum slab deflection in jointed concrete pavements.
Figure 3-15. Illustration of Packard and Tayabji’s (1985) assumption
AASHTO (2003) implemented a research to determine effects of adjacent concrete

slab panels and traffic lanes on the loaded concrete slab using finite element technics.
To do this, a variety of jointed concrete pavements with different thicknesses of 200,
250 and 300 mm were considered over a subgrade with modulus of subgrade reaction
from 27 to 54 kPa / mm. An 80 kN single axle load was applied at the middle of the
free longitudinal joint as illustrated in Figure 3-16a. A differential temperature of
11.1oC was considered between the top and the bottom surface layers of the concrete
slabs. In terms of pavement configurations, two scenarios were studied.
In the first scenario, effects of adjacent traffic lanes on induced tensile stress in the
loaded slab were investigated. For this purpose, results of finite element analysis of a
single traffic lane with three concrete slabs in the longitudinal direction were compared
with corresponding results of a confined traffic lane with the same number of concrete
slabs (Fig. 3-17).
a) Load position for critical flexural stress b) Load position for critical deflection
Figure 3-16. Critical axle position on the pavement
38
In the second scenario, effects of adjacent concrete slabs in the longitudinal direction
were taken into consideration by comparing the magnitude of induced tensile stress of
one, three and five concrete slabs as shown in Figure 3-18.

Figure 3-17. Configurations of traffic lanes considered in the AASHTO (2003)
Results indicated that adjacent traffic lane has no effect on the maximum induced
tensile stress of the loaded concrete slab (Fig. 3-19). Consideration of different
concrete slabs in the longitudinal direction has also no strong effect on the maximum
induced tensile stress (Fig. 3-20). Since results of the finite element analysis (FEA) of
a single concrete slab was highly sensitive to a change in thickness of concrete slab,
temperature fluctuation and other pavement characteristics, AASHTO (2003)
recommends the use of a single lane with three concrete slabs in the longitudinal
direction.
This recommendation is compatible with that considered by Packard and Tayabji

(1985). Consideration of a confined traffic lane in the work of Packard and Tayabji
(1985) was to reduce the possibility of erosion of subbase and subgrade materials
which will be discussed briefly in Section 2.6.2 of this chapter. Based on results of
39
Packard and Tayabji (1985), the maximum slab deflection and the maximum induced
tensile stress in concrete pavements are calculated in the Austroads method (2004) by
applying the vehicular loads at the corner and at the middle of the longitudinal joint
respectively (see Figure 3-16).

Figure 3-18. Configurations of concrete slabs in the longitudinal direction considered

by AASHTO (2003)

Figure 3-19. Effect of number of slabs in the transverse direction on prediction of the
maximum bottom bending surface stress (AASHTO, 2003)
40
Figure 3-20. Effect of number of slabs in the longitudinal direction on prediction of

maximum induced tensile stresses at the bottom surface layer of the concrete slab,
single slab vs. three slabs (AASHTO, 2003)
The above mentioned consideration is to prevent the bottom-up transverse fatigue

cracking in the pavement. Further information on fatigue related distresses are
discussed later in Section 2.6.1 of this chapter.
Axle group types

Vehicular manufacturers use different types of axle group in heavy vehicles.
Nevertheless, a small range of axle groups has been considered in the past research.
Single axle load with dual tyres was employed in the work of Tayabji and Colley
(1983), Packard and Tayabji (1983 and 1985), Smith et al. (1990), Yu et al. (1997) and
AASHTO (2003). Gillespie et al. (1992) used the Equivalent Single Axle Loading
(ESAL) based on load characteristics derived from single axle dual tyres. The effect of
tandem axle load on concrete pavement response was investigated by Tayabji and
Colley (1983), Packard and Tayabji (1983 and 1985) and AASHTO (2003). Triple
axle load was used in the work of Packard and Tayabji (1985) and AASHTO (2003).
Austroads (2004) adopted the PCA method to be suitable for Australian conditions.
Consequently, the PCA method was extended to consider different types of axle
groups including Single Axle Single Tyre (SAST), Single Axle Dual Tyre (SADT),
41
Tandem Axle Single Tyre (TAST), Tandem Axle Dual Tyre (TADT), Triple Axle
Dual Tyre (TRDT), and Quad Axle Dual Tyre (QADT). Figure 3-21 shows a
schematic configuration of the above mentioned axle groups.
Figure 3-21. Axle group types considered in Austroads 2004b
Suspension system
The load distribution between axles depends on vehicle’s suspension systems. A
variety of suspension systems such as flat leaf, taper leaf, and air springs have been
employed in different heavy vehicles by vehicle manufacturers. In the absence of
detailed information in pavement analyses, an equal load distribution between different
axles in a given axle group is assumed. However, there are cases where mixed axle
suspension systems are employed to alter the axle load capacity e.g. trucks with lift or
flip axles. Depending on the level of pavement roughness, structural performance of
concrete pavements under moving vehicular loads is significantly affected by
suspension type of a given axle group (Gillespie et al., 1992).
An in-service survey of suspension systems, including air and mechanical suspensions

in triple axle groups used in heavy vehicle passing along the Hume Highway in
Australia (Blanksby et al., 2006), showed that the possibility of non-uniform load
distribution between axles in a given axle group, i.e. TRDT, is high particularly in the
presence of the mechanical suspensions. The mean percentage of load shift between
axles is about 1 to 2 per cent in the air suspension and between 20 to 40 per cent in the
mechanical suspensions.
Distance between axles in a given axle group

A distance of 1220 mm to 1372 mm between axles in a triple axle group and 1270 mm
for distance between axles in a tandem axle group were used by Packard and Tayabji
(1985). Gillespie et al. (1992) and AASHTO (2003) recommended a distance of
1295.4 mm between axles to capture the critical pavement response. However, Kim et
42
al. (2002) and Hiller and Roesler (2002) considered 1320 mm for the same purpose.
Whilst Austroads (2004) is silent on this issue, the Road and Traffic Authority (RTA)
of Australia (1998) allows a variation between 1000 mm and 1600 mm for the triple
axle group and between 1067 mm and 1633 mm for the quad axle group.
Axle width and space between dual tyres

Lee and Carpenter (2001) noted that Packard and Tayabji (1985) assumed the distance
between the centres of dual tyres and axle width as 305 mm and 1829 mm,
respectively. Kim et al. (2002) considered these distances to be 330 mm and 1880 mm
whereas Hiller and Roesler (2002) assumed them to be 340 mm and 1850 mm,
respectively. AASHTO (2003) also considered a distance of 305 mm between the
centres of dual tyres but increased the axle width to 2134 mm. Austroads (2004b)
recommends 330 mm and 1800 mm for distance between centres of dual wheels and
axle width, respectively. Figure 3-22 summarises the variations of axle width and
distance between the centres of dual tyres according to various sources.
Figure 3-22. Transverse section of a typical SADT
Tyre inflation pressure

A variety of tyres are employed to transfer vehicular gross weight to pavements. Tyre
load capacity depends on tyre size, ply rating, and inflation pressure. Tyre size is
usually defined by tyre width, length, and area of contact patch (Gillespie et al., 1992).
Tyre inflation pressure affects the tyre-pavement contact area and stress. Austroads
(2004) reported that tyre pressure varies between 500 kPa to 1000 kPa in heavy
vehicles passing on roads. However, a tyre pressure of 750 kPa is currently
43
recommended for structural design purposes. Shackel (1993) stated that this pressure
varies between 700 kPa to 1400 kPa in heavy vehicles used on industrial pavements.
Tyre-pavement contact stress

Lippmann (1985), De Beer et al. (1997), and Douglas et al. (2000) in an investigation
of different tyre types reported that contact stress is not uniform particularly when the
tyre inflation pressure is low. Marshek et al. (1986) and Gillespie et al. (1992) also
showed that tyre-pavement contact stress is not uniform, but a change in the
distribution of tyre contact stress has no significant effects on concrete pavement
response. To simplify the analysis procedure, tyre contact stress is, hence, assumed to
be uniform in concrete pavements.
Tyre-pavement interface shape

Handson and Seeds (1988) stated that tyre pavement contact area is more rectangular
than circular. Shackel (1993) showed that tyre imprint has an elliptical shape but to
simplify the analysis process, he recommended the use of a rectangular shape of an
equivalent area. The results of Gillespie et al. (1992) also showed that tyre imprint is
not a rectangle; however, the assumption of rectangular tyre contact area does not
significantly change responses of the concrete pavements. The tyre contact area can be
calculated from Equation 3-13.
Wl
At = (3-13)
TP
Where At is tyre contact area, Wl is wheel load and TP is tyre pressure.
Although researchers have considered a rectangular tyre-pavement interface, a variety

of dimensions for tyre imprint have been used. If tyre-pavement contact area is defined
as width × length where width is the dimension normal to traffic direction (Fig. 3-23),
Packard and Tayabji (1985) considered a rectangular tyre contact shape of 178 mm ×
254 mm. Handson and Seeds (1988) showed that width and length of tyre imprint in an
11R24.5 tyre are 196 mm and 250 mm respectively. Gillespie et al. (1992) preferred to
use a square shape of 203 mm × 203 mm for dual tyres and a rectangular shape of 203
44
mm × 229 mm for single tyre whereas Kim et al. (2002) used a rectangular shape of
203 mm × 178 mm.
Figure 3-23. Length and width of tyre–pavement contact area
Douglas et al. (2000) provide a graphic interrelationship between contact patch length
and wheel load based on an experimental study of different tyre inflation pressures for
radial tyres. The result of this study showed that an increase in tyre pressure decreases
the contact length. However, the decrease in tyre contact length is relatively small
when tyre pressure exceeds 480 kPa. For instance, the contact patch length for wheel
load of 22.5 kN decreases from 240 to 230 mm when tyre pressure increases from 480
kPa to 690 kPa.
3.3.2. Temperature
Temperature fluctuation together with vehicular loads plays a significant role in
concrete pavement deteriorations (Hiller and Roesler 2005, Liang and Niu 1998). A
concrete element can be expanded or contracted due to uniform temperature
fluctuation or it can be curled due to a non-uniform temperature gradients which result
in thermal induced stress in the element.
An increase in temperature increases the rate of hydration process in fresh concrete

which consequently results in further temperature fluctuations due to chemical reaction
between cement paste and aggregate. Early age cracking will be observed in the
sample if the magnitude of thermal induced tensile stress at the particular concrete age
becomes more than corresponding concrete strength at that age. Because of this,
45
temperature of the concrete during construction in hot weather conditions needs to be
monitored (Schindler and McCullough, 2002).
A similar behaviour can be observed in concrete pavement systems. However, the

severity of induced thermal stresses and consequently the rate of concrete pavement
deteriorations increase due to boundary conditions between the concrete slab and the
subbase and also between adjacent concrete slabs. Although thermal analysis of
concrete pavements dates back to 1920s, temperature effects on concrete pavement
response and deteriorations have also attracted researcher’s attention in the last decade.
Temperature fluctuation has varied effects on concrete pavements. A part of it that is

uniformly distributed through concrete slab depth expands or contracts the concrete
slab. The edges of concrete slabs in both transverse and longitudinal directions may be
restrained by other traffic lanes, shoulders and concrete slabs. Furthermore, a friction
force exists at the interface of concrete slab and subbase. As a result, uniform
expansion in concrete slab induces a compressive force along concrete slab edges and
a tensile force at interface of the concrete slab and the subbase.
In the contraction condition, a tensile force will only be produced at the bottom surface
layer of the concrete slab. The magnitude of induced compressive or tensile stress
increases when the pavement experiences higher uniform temperatures and when it is
longer. This ultimately results in concrete slab buckling (Croll, 2005) in compression
and cracking in tension. Buckling of concrete slab produces a tensile stress at the top or
the bottom surface layers of the concrete slab depending on buckling mode shapes
which subsequently results in cracking. Provision of a gap in the whole depth of joint
between side surfaces of the adjacent slabs, such as that shown in Figure 2-14 with a
soft filler material, omits the possibility of buckling failure of the concrete slab.
In addition, non-uniform temperature distribution within depth of concrete slab results

in upward (nighttime) or downward (daytime) curling (Fig. 3-24). Concrete slab
curling has a strong effect on the pavement behaviour due to the consequent loss of
support (Tang et al. 1993, Shi et al. 1993 and Zhang et al. 2003).
46
Figure 3-24. Effects of daytime and nighttime temperature gradient on concrete
pavements in the absence of factors restraining the concrete slab movement
Kuo (1998) indicated that curling induced stress is affected by temperature differential,
self-weight of concrete slab and support under concrete slab. Curling induced stress
has also a reverse relationship with the subgrade shear modulus (Shi et al., 1993).
When a single concrete slab is freely curled due to a differential temperature gradient,
in the absence of the restraining factors such as slab weight and friction force at the
interface of the concrete slab and subbase, a bending flexural stress is induced at the
top or the bottom surface layer of the slab due to its residual stiffness (Mohamed and
Hansen, 1997). In this case, maximum induced tensile stress occurs at the centre of the
slab and at the top surface layer during the day and at the bottom surface layer during
the night (Fig. 3-24).
On the other hand, another tensile stress will occur in the opposite sides of the residual
tensile stress (Fig. 3-25) in the presence of the restraining factors such as slab weight,
subgrade and subbase resistant, vehicular loads and friction force at the interface of the
concrete slab and subbase. In this condition, curling induced tensile stress occurs at the
top surface layer of the concrete slab during nighttime and at the bottom surface layer
of the concrete slab during daytime (Ongle and Harvey, 2004).
47
Figure 3-25. Effects of daytime and nighttime temperature gradient on concrete
pavements in the presence of factors restraining the concrete slab movement
Traditional methods of thermal analysis were based on a linear temperature

distribution in pavement depth (Westergaard 1926, Bradbury 1938). However,
Choubane and Tia (1995) showed that temperature distribution in pavement depth is
nonlinear.
Several field tests were carried out in the past to determine the range of differential
temperature in the depth of concrete slab. Richardson and Armaghani (1990) and
Shoukry et al. (2002) reported that the differential temperature is about 10˚C in a
concrete slab with 225 mm thickness. Byrum and Hansen (1994) based on other
research in this field used a differential temperature between 0.087 and 0.109˚C / mm
during daytime and between 0.044 and 0.065ºC / mm during nighttime. Darter et al.
(1995) provided a range of between 0.0219 and 0.656˚C / mm whereas Ongel and
Harvey (2004) reported monthly values of differential temperature in concrete
pavement for a period of 5 years with an average of 0.125˚C / mm. As differential
temperature is strongly affected by air temperature, the ratio of the top surface area of
the concrete slab to its depth, duration and density of solar radiation, rain fall, thermal
48
conductivity of concrete and wind speed, it is obvious that differential temperature
changes from one location to other locations.
In similar conditions, Zhang et al. (2003) showed that nonlinear temperature

distribution can provide more accurate response than linear temperature distribution.
However, results of Mohamed and Hansen (1997) and also Heath and Roesler (1999)
indicated that nonlinear temperature distribution through the depth of concrete slab
results in tensile stress that is lower than that of linear temperature distribution when
concrete pavement is subjected to a positive temperature gradient (daytime) and
produces tensile stress that is greater than that of linear temperature distribution when
concrete pavements are subjected to negative temperature gradients (nighttime).
Kuo (1998) recommended that (i) mid slab loading in daytime curling and (ii) joint
loading in nighttime curling should be considered in pavement analysis. Results of Yu
et al. (1998) on concrete pavement response to temperature and wheel loads suggested
that corner loading may result in greater stress than mid-edge loading.
3.3.3. Shrinkage - loss of moisture content

Since the top surface layer of the concrete slab is exposed to solar radiation and wind,
it dries and cures faster than other layers within the depth of concrete slab and
consequently results in non-uniform shrinkage which is the reason for concrete slab
warping and top-down cracking (Ongel and Harvey, 2004).
Shrinkage is assumed to linearly decrease from the top surface layer of the concrete
slab towards the mid-depth of the concrete slab. Consequently, a full shrinkage occurs
at the top surface layer and no shrinkage exists below the mid depth of the concrete
slab (Rasmussen and McCullough, 1998). Hence, shrinkage effects on structural
response of concrete pavements can be simulated by considering an equivalent
differential temperature gradient within depth of concrete slab. The effect of drying
shrinkage on concrete pavement is similar to effects of nighttime differential
temperature. Hence, Reddy et al. (1963) recommended the use of equivalent nighttime
temperature gradients between 0.065 and 0.13˚C / mm in concrete pavement analysis
to represent the effects of drying shrinkage in concrete pavement responses.
49
3.4. Concrete Pavement Analysis
3.4.1. Analytical Solution
Westergaard (1926, 1927, 1933, 1939, 1943 and 1947) developed comprehensive
analytical solutions for analysing concrete pavements under centre, edge and corner
loadings using the classical thin-plate theory. In Westergaard study, concrete
pavement was modelled as a homogenous, isotropic, elastic thin slab resting on a
Winkler foundation. To simplify the analysis procedure, Westergaard assumed that:
- The foundation soil acts like a bed of springs (the use of dense liquid
foundation)
- A fully bonded boundary condition exists between concrete slab and
foundation soil
- Shear and frictional forces are negligible
- The semi-infinite foundation is not restricted by a rigid layer
- Concrete slab has a uniform thickness
- Neutral axis is at mid-depth of the concrete slab
- Vehicular loads are uniformly distributed at tyre-pavement contact area
having a circular (for centre and corner loadings) or semi circular (for edge
loadings) shape
- The load is applied normal to the surface of concrete slab
- Pavement acts as a single semi-infinitely large, homogenous, isotropic
elastic slab with no discontinuities.
The above mentioned simplifications imply some limitations as follows:
- stresses and deflections can be only calculated for centre, edge and corner
loadings
- Shear and frictional forces acting on concrete slab surfaces are ignored
- Availability of voids under concrete slab and discontinuities in concrete
slab due to crack generation or provision of joints is not considered
- The method was developed for single wheel load and consequently real
axle group configurations can not be taken into consideration.
50
Since no discontinuities were considered in the Westergaard method, effects of the
LTE at joints or cracks on pavement performance were not addressed. Despite
limitations associated with Westergaard analytical solution, his equations are still
used. Ioannides et al. (1985) revised the original Westergaard equations for edge
loading. Since all Westergaard’s equations are based on single tyre with a circular
tyre pavement contact area, Huang (1993) developed an equation to convert effects
of dual tyres to single tyre with circular tyre pavement contact shape. Further
information on Westergaard method can be found in the work of Westergaard (1926,
1927, 1933, 1939, 1943 and 1947).
Since the Westergaard method can not be used in concrete pavements with
discontinuities (consideration of longitudinal and transverse joints), Bradbury (1938)
developed an analytical solution or analysing of concrete pavements with
discontinuities subjected to differential temperature gradients.
An analytical model of concrete pavements for a single wheel loads based on semi-
infinite elastic solid was developed by Hogg and Hall (1938). Pickett and Ray (1951)
developed an influence chart for concrete pavement analysis based on Westergaard
solutions. Reissner (1958, 1945 and 1950) contributed the thick plate theory to
analyse concrete pavement with a circular hole subjected to torsional problems. The
Reissner’s theory was later simplified by Hu (1981). Shear deformation was then
taken into consideration by Mindlin (1951).
A theoretical solution of a rectangular thick plate with four free edges and supported
on Pasternak foundation was developed by Shi et al. (1994) based on Reissner thick-
plate theory. This solution was extended by Fwa et al. (1996). They also determined
stress and deflection differences between thick-plate solutions and Westergaard’s
solutions.
Mohamed and Hansen (1997) developed an analytical method for estimating the
curling induced stresses in the concrete pavement based on nonlinear temperature
gradients. Using the plate theory, the concept of equivalent temperature distributions
was then introduced in concrete pavement analysis (Ioannides and Khazanovich,
1998). In their study, a plate consisting of one or more layers resting on an elastic
51
foundation was investigated. The layers used in their study were plate layers with no
separation and compressible layers with possible separation using Totsky model
(described later in Section 2.5.2.2 of this chapter). Consequently, mathematical
formulations for analysis of a typical concrete pavement subjected to a linear function,
a quadratic function or multi linear function of differential temperature together with
arbitrary wheel load were developed.
3.4.2. Numerical Solution

The expensive cost of experimental studies associated with progressive development
in numerical computation together with a rapid growth in software and hardware
technologies leads to the use of numerical analysis for investigation of those
problems with a complex geometry, boundary conditions, and material properties.
The most common numerical solutions used for analysis of concrete pavements are
discussed in the following sections 3.4.2.1 and 3.4.2.2.
3.4.2.1. Discrete Element Method (DEM)

Hudson and Matlock (1966) were the first to apply the DEM in concrete pavement
analysis using Winkler foundation. In their contribution, the original bending
stiffness of the concrete slab at joints was reduced to simulate joint effects in
concrete pavement behaviour. This model was then modified by Vora and Matlock
(1970) to take into consideration different element sizing and anisotropic skew slabs.
Furthermore, foundation soil was idealized using semi-infinite elastic solid elements.
This method does not allow incorporating elements with different sizes into the
analysis easily. Furthermore, stress estimations at free edges are not converged to
unique values.
3.4.2.2. Finite Element Method (FEM)

The initial idea of dividing a given domain into discrete parts goes back to ancient
mathematicians. The concept of finite element technics was firstly used by Hrenikoff
(1941) for aircraft analysis using truss and beam elements. To efficiently analyse the
torsion problem, the use of triangular elements was then incorporated into the method
(Courant, 1943). Seven years later in 1950, the Boeing Company used triangular
52
elements for wing analysis. Further research on finite element technics by Argyris and
Kelsey (1960) and Turner et al. in 1956 represented the new form of this method. The
term “finite element” was used by Clough in 1960. It is important to note that the
accuracy of a finite element analysis depends on several factors such as meshing size
and element types.
Nowadays, finite element techniques are extensively used to treat complex engineering
problems in several areas such as structural, mechanical, electrical, geological and
thermal (Reddy, 1993). In general, finite element analysis packages can be divided into
two categories, general purpose finite element programs such as ABAQUS and
ANSYS, which are very powerful for nonlinear dynamic analysis, and specific finite
element programs developed for concrete pavement analysis using the classical thin
plate theory or 3D solid elements.
ILLISLAB (Tabatabaie and Barenberg, 1978 and 1980), JSLAB (Tayabji and Colley,
1981), WESLIQUID and WESLAYER (Huang and Wang 1973 and Chou 1980),
FEACONS (Armaghani 1987, Tia et al. 1987, Jo 1988 and 1989, Kumara et al. 2002),
KENSLAB (Huang, 1993) are some of the examples of those finite element programs
using the classical thin plate theory.
In all the above mentioned programs, concrete pavements were modelled as an

assemblage of rectangular plates with bending capacity, having three degree of
freedom at each node – deflection at Z axis and rotation about X and Y axes.
Furthermore, a subbase layer and/or an asphalt overlay can be incorporated in the
analysis using plate elements. In addition, a fully bonded or unbonded boundary
condition can be simulated at the interfaces of the above mentioned layers and the
concrete slabs. Note that the new generation of finite element packages such as
ISLAB2000 (new version of ILLISLAB) are able to simulate a partially bonded layer
between the concrete slabs and the subbase.
The subgrade was modelled using either Winkler foundation idealization or an elastic
layered foundation. 2D beam elements and vertical spring elements were used to
respectively simulate dowel bars and aggregate interlock in the above mentioned
programs. Since keyed joists can only transfer the shear force across joints and their
53
behaviour is similar to aggregate interlock, they were also simulated using vertical
spring elements.
Some of these programs such as JSLAB, WESLIQUID, WESLAYER and

FEACONS and ISLAB2000 are able to consider linear and/or nonlinear differential
temperature gradients along depth of concrete slab as well as dowel looseness. The
loss of support due to thermal curling of the concrete slab or presence of a void in the
sublayers can be also modelled by setting zero stiffness for the subgrade springs at
these locations.
EverFE (Davids and Mahoney, 1999), on the other hand, is a three dimensional finite-
element analysis software jointly developed by the universities of Maine and
Washington to simulate the behaviour of jointed plain concrete pavements under axle
group loads and environmental effects. This program employs 20-noded quadratic
solid elements, beam elements, shear spring and 8-noded dense liquid shell elements to
simulate behaviour of concrete slab and base, dowels and tie bars, aggregate interlock
and subgrade layers under the applied loads respectively. This program is also able to
simulate a tensionless property in the subgrade. Furthermore, EverFE employs 16-
noded zero thickness quadratic interface element to simulate the debonding layer
between concrete slab and subbase (Fig. 3-26).
Since modelling of dowel bars and debonding layer can significantly affect results of
concrete pavement analysis, further information on these areas is presented in the
following sections, dowel modelling and model of debonding layer.
Figure 3-26. Pavement modelling in EverFE finite element program
54
Dowel modelling
Dowel bars may be modelled as beam elements (Davids and Mahoney, 1999) or elastic
shear spring elements (Tabatabaie and Barenberg, 1980). Information on cross section,
length and location of dowel is required if dowel bars are modelled as beam elements.
This model assumes that the applied loads are transferred across joints by bending
action and shear capability of the dowel bars. In the case of dowel looseness and to
accurately model the interaction between dowel bars and the surrounding concrete, a
contact pair or linear spring between dowel bar and concrete may be also considered.
An increase in dowel looseness decreases the load transfer efficiency across joints and
consequently results in joint faulting. The use of shear spring elements for simulating
behaviour of dowel bars is implemented in many finite element programs such as
ILLISLAB, JSLAB and ISLAB2000. The stiffness of the shear spring is estimated by
dividing the shear force per unit length of the joint into the differential deflection
between adjacent slabs (Huang, 1993). This model assumes that the applied loads are
transferred across joints by only shear transfer capability of the dowel bars.
Crovetti and Crovetti (1994), Crovetti (1996), and Hammons et al. (1995) based on the
aforementioned theory assumed the following equation in order to analyse the concrete
pavements:
δ L + δU = δ E (3-14)
σ L + σU =σ E (3-15)
where σ and δ are induced stress and deflection respectively, and the subscripts L ,
U, and E are used for loaded slab, unloaded slab, and at the edge of joint when the joint
load transfer capability is zero.
Guo (2003) showed that the use of the above assumption results in a good
approximation provided that the slab has a full contact with the subbase. In other
words, the above mentioned formulas can not be used in partially bonded or fully
unbonded boundary condition between concrete slabs and subbase.
Dowel–pavement interaction is another significant factor affecting concrete pavement

responses as dowel looseness decreases the magnitude of the LTE at the transverse
55
joints (Zaman and Alvappillai, 1995) and leads to joint faulting associated with erosion
of subbase and subgrade materials. The main reason behind dowel looseness is the
fatigue phenomenon in concrete due to high induced bearing stress at dowel-pavement
interface caused by repeated loads (Channakeshava et al. 1993 and Zaman and
Alvappillai 1995). A contact element (Zaman and Alvappillai, 1995) or gap element
(Channakeshava et al. 1993) between dowel bar and concrete can be used to simulate
effect of dowel looseness on concrete pavements. Consequently, the concrete slab
under the load can deflect equally in correspondence to dowel looseness before
transferring the load to adjacent concrete slab.
Model of debonding layer

In many first generations of finite elements programs developed for concrete
pavement analysis based on two-layered systems, the analysis is accomplished by
converting the two-layer system to a structurally equivalent single-layer system. This
results in accurate responses of concrete pavements provided that a fully bonded or a
fully unbonded boundary condition between concrete base and subbase is assumed.
In the case of fully unbonded boundary condition, concrete slab and subbase should
have the same deflection profile. However, concrete slabs may be separated from
subbase due to thermal curling of the concrete slabs and independently act under the
applied loads. This induces different deflection profiles in concrete slab and subbase.
To overcome this problem, Totsky model (Totsky, 1981) has been contributed in
finite element analysis of concrete pavements. Using this approach, the multi-
layered pavement system was modelled as a series of springs and plates. While the
plate elements model the bending, the springs accommodate the direct compression
between layers. This requires the use of a specific 8-noded element with three degree
of freedom at each node.
The first four nodes are placed at the neutral axis of the upper plate, while the other
four nodes are placed at the neutral axis of the lower plate (Fig. 3-27). The stiffness
matrix for this element is:
56
[K ] + [K DLI ] − [K DLI ]
[K ] =  PL1 
 − [K DLI ] [K DLI ] + [K PL 2 ] + [K DLS ] (3-16)
Where [KPL1] and [KPL2] are the stiffness matrices of the upper and lower plates
respectively, [KDLI] is the stiffness matrix of the interlayer springs and [KDLS] is the
stiffness matrix of the subgrade.
The stiffness matrices for both the spring interlayer and subgrade are calculated
based on nodal displacements, and not on nodal rotations. The method of stiffness
calculation can be found elsewhere (Khazanovich and Ioannides, 1998).

Figure 3-27. 8-noded finite element setup in Totsky model (Totsky, 1981)
In the Totsky model, all interface springs are in compression due to the self weight of
the concrete slabs. However, the stiffness of those springs located in the separated
areas due to curling is considered to be zero during analysis. The analysis is
continued until an equilibrium condition is reached.
A 16-noded zero thickness quadratic interface element was also used to simulate
behaviour of the debonding layer in 3D finite element programs such as EverFE. The
debonding element is capable of transferring the shear stress along the interface of the
concrete slab and the subbase. It is meshed in accordance with the meshing size of the
concrete slab and the subbase. A bilinear constitutive relationship was considered to
define the characteristics of this element under the applied loads. Hence, the debonding
layer can be defined by introducing initial distributed stiffness and slip displacement.
A free separation under tension occurs between the concrete slab and subbase when
unbonded boundary condition is selected. Figure 3-28 shows a schematic behaviour of
the interface element as well as the required information for its definition. Further
information on this matter can be found elsewhere (Davids and Wang, 2003).
57
Another method for simulation of the debonding layer is the use of contact pairs
between concrete slabs and subbase. In this case, friction coefficient is the key factor in
the analysis. Research conducted in the past showed that the magnitude of friction
coefficient in concrete pavements depends on treatment (debonding) layer and is
always greater than 1.2 (see Section 2.3.4. of this document for further information).
The use of contact pairs in the model requires a sophisticated approach and leads to
high non-linearity in behaviour of the structure under loads. Consequently, general
purpose finite element programs such as ANSYS and ABAQUS are only able to
analyse a structure with high non-linear behaviour.

Figure 3-28. Schematic interface element behaviour and interface constitutive

relationship (Davids and Wang, 2003)
3.4.3. Static Analysis

Analysis of concrete pavements under the applied loads was widely performed in the
past based on static loads. In these typical analyses, pavement configurations and
properties of each individual layer associated with magnitude, configuration and
location of vehicular loads were the main factors affecting concrete pavement
responses. Vehicular loads were considered as wheel loads, axle loads or axle group
loads positioned at the mid-length of longitudinal edge (joint) or at the corner of the
concrete slab. Results of static analysis are described later in Section 2.6 of this
chapter.
3.4.4. Transient Dynamic Study

As mentioned earlier in this document, vehicular loads are dynamic in nature as they
wander at any location within the concrete slab and pass along the pavement with
58
different speeds. As a result, the magnitude and location of the vehicular loads can be
time dependent. Transient dynamic analysis, known also as time-history analysis, is a
technique used to determine the dynamic response of a structure under the time-
dependent loads (ANSYS Help, 2006).
The basic equation of a structure subjected to a transient load is:
[ M ]( t ){ X&&}(t ) + [Cd ](t ){ X& }(t ) + [ K S ]( t ){ X }(t ) = {F (t )} (3-17)
Where [M](t) is mass matrix of the structure at time t, [Cd](t) is damping matrix of the
structure at time t, [KS](t) is stiffness matrix of the structure at time t, { X&&}(t ) is nodal
acceleration vector at time t, { X& }(t ) is nodal velocity vector at time t, { X }(t ) is nodal
displacement vector at time t, and {F(t)} is the load vector.
This equation in concrete pavement analysis determines a deflection, Xt, at a certain

node within the pavement as a result of the applied load at a certain time, t. This
deflection, depending on the mass and damping ratios of the pavement, induces an
acceleration, X&& t and speed, X& (t ) in the node. This application can be extended to all
nodes within the concrete pavements. The concrete pavement is considered stable if
an equilibrium condition between the applied load and the reaction loads exists. The
Equation 3-17 at any given time, t, can be assumed to be a static equilibrium
equation which also take into account inertia forces, [ M ]( t ) { X&&}(t ) , and damping
forces, [Cd ](t ){ X& }(t ) .
Since the mass of concrete pavements is time independent and the damping matrix, for
simplification, is assumed to be constant during analysis, Equation 3-17 can be
simplified as follows:
[ M ]{ X&&}( t ) + [Cd ]{ X& }(t ) + [ K S ]( t ){ X }(t ) = {F (t )} (3-18)
Rahleigh (ANSYS Help, 2006) provided an equation to predict the damping matrix in a
given structure based on its mass and stiffness matrixes.
59
[Cd ] = α R ⋅ [ M ] + β R ⋅ [ K S ] (3-19)
Where α R and β R are Rayleigh damping constants in decimal numbers. They are not
generally known but, can be calculated based on modal damping ratios, ξ i , which is the
ratio of actual damping to critical damping for a particular vibration mode, i. If ω i is
assumed to be a natural circular frequency of mode i, α and β shall satisfy Equation

3-20.
α R β R ⋅ ωi
ξi = + (3-20)
2ωi 2
In many practical problems mass factor, α R , can be assumed to be zero. Hence, the
β R value can be calculated from Equation 3-20 by considering α =0:
β R ⋅ ωi 2ξi
ξi = a βR =
ωi (3-21)
2
No information on damping coefficient of concrete pavements exists in the literature

except for some typical variations used by Kim et al (2001 and 2002). Consideration of
debonding layer between the concrete slab and the subbase also leads to a complex
calculation of stiffness matrix as curling may result in a separation between layers.
The dynamic study of concrete pavements can be grouped into two categories,
studies analysing the impact of dynamic loads on concrete pavements and
experimental studies.
3.4.4.1. Transient Dynamic Analysis

As mentioned earlier, the magnitude and location of the vehicular loads can be time
dependent. The magnitude of the load applied on concrete pavement through tyre
pavement contact area is affected by pavement roughness and suspension system of the
heavy vehicles (Liu et al. 2000). As a result, two different methods are used in
pavement analysis. In the simpler approach, a moving load can be simulated as a
footprint of uniform contact pressure moving at a constant velocity in a straight line
along the longitudinal direction of the concrete slab (Hendrick et al., 1992). In the
60
second method, on the other hand, the time-history of the wheel load is firstly
estimated by considering a given pavement roughness profile along wheel path and
particular suspension systems. The pavement is then analysed using the integrated
time-history of the load (kim et al., 2001 and 2002).
A comparison between damage rates due to consideration of the first and second
methods (Hendrick et al., 1992) showed that the second approach results in
approximately 38% greater fatigue cracking than the first method after 15 years of
pavement service life (Fig. 3-29). They also found that a region close to the joints
experiences a severe cracking damage if constant moving loads are considered. The
use of load time-history, on the other hand, decreases corner cracking by about 15%
and significantly increases the density of mid-edge cracking (Fig. 3-30). It is
important to note that the above mentioned outcomes are affected by truck speed
(Hendrick et al., 1992).

Figure 3-29. Comparison between damage rate: moving constant load versus time-
history load (Hendrick et al., 1992)
61
In accordance with the research on transient dynamic analysis of concrete pavements
carried out in the past, concrete slabs were modelled as a thick or thin plate (Chatti et
al. 1994 and Kim et al. 2002), as solid elements (Liu at al., 2000), or as beam
elements (Liu at Gazis, 1999) resting on elastic foundation or viscoelastic
foundation. The viscoelastic foundation is a combination of damper and spring. A
moving point load (Liu at Gazis, 1999), wheel load (Chatti et al. 1994 and Liu at al.
2000), single axle load (SAL) (Zaghloul and white, 1993) or tandem axle load (Kim
et al. 2002) were employed in dynamic analysis. Effects of load transfer devices
involving dowels, tie bars and aggregate interlock (Chatti et al. 1994 and Zaghloul
and white, 1993) and variations of damping property (Kim et al. 2002) and surface
roughness (Liu at al., 2000 and Liu at Gazis, 1999) on dynamic response of concrete
pavements have been also considered in some studies.

Figure 3-30. Comparison between fatigue cracking along slab length: moving
constant load versus time-history load (Hendrick et al., 1992)
62
Results of dynamic analysis of concrete pavement subjected to moving loads carried
out by Chatti et al. (1994), Liu and Gazis (1999) and Monismith et al. showed that
dynamic responses of concrete pavements are equal or lower than those captured by
static analysis. Kim et al. (2001 and 2002) investigated the effect of transient TADT
on concrete pavement responses and showed that velocity has no strong effect on
maximum longitudinal stress in JPCP but affects the pavement deflections.
Longitudinal stress is directly and reversely affected by pavement roughness and
load frequency respectively. The use of dynamic vehicle loads in a nonlinear
pavement distress model where trucks are modelled as a finite number of rigid bodies
by using springs, dashpots and joints (Stoner et al. 1990 and Bhatti and Stoner 1998)
showed that concrete pavement deflection and consequently distresses related to
erosion of the subbase and subgrade materials are significantly affected by dynamic
loads. Further information on modelling of dynamic vehicle loads can be found
elsewhere (Stoner et al. 1990, Gillespie et al. 1993, Stoner and Bhatti 1994 and
Bhatti et al. 1994).
Gillespie et al. (1993) mentioned that effect of vehicle speed on concrete pavements
is not severe. Moreover, results of a transient analysis of SAL (Zaghloul and White,
1993) indicated that axle velocity significantly decreases the magnitude of transverse
tensile stresses in jointed reinforced concrete pavement (JRCP). Shoukry and Fahmy
(2002) used LSDYNA platform to study jointed plain concrete pavement subjected
to dynamic transient axle groups including SADT, TADT and TRDT. They assumed
that the loads are symmetrically applied upon pavement. Their results showed that
axle speed has no significant effect on concrete pavement responses.
However, the use of influence function in concrete pavement analysis (Muhammad

et al., 1991) showed that stress in the interior region of slab may increase by an
increase in vehicle speed in the vicinity of the crack. Results of a 3D finite element
analysis of concrete pavements indicated that a change in surface roughness (Liu et
al. 2000 and Liu and Gazis 1999) or pavement damping property (Kim et al., 2001
and 2002) can produce dynamic stresses at the bottom surface layer of concrete
pavements that were greater than the static stresses at the same point. Moreover, the
dynamic fatigue cracking of the pavement during its life may be greater than the
63
fatigue cracking based on static values of the tensile stress (Hendrick et al. 1992 and
Gillespie et al. 1993).
3.4.4.2. Experimental Tests of Concrete Pavements

The most significant experimental test that has been widely used by researchers
worldwide is the field tests conducted by the American Association of State Highway
and Transportation Officials (AASHO, 1962). In the AASHO research, six loop test
sections of different concrete pavements were constructed.
Each test section consisted of two traffic lanes with 3657.6 mm width and thicknesses
varying from 63.5 mm to 317.5 mm, having both reinforced and unreinforced concrete
slabs constructed over a granular subbase. Subbase also had different thicknesses
between 0 to 228.6 mm along each concrete pavement with a certain thickness. The
zero thickness of subbase layer shows that subbase layer was not considered. Some of
the test sections were confined along longitudinal edges by using crushed stone
shoulders with a thickness of 76.2 mm and a width of 3048 mm.
Transverse joints were prepared by using soft sawing method. The depth of saw cut
varied between 19 mm and 57.15 mm depending on thickness of concrete slabs.
Transverse joints were dowelled using round bars of diameters varying between 9.5
mm and 41.3mm and different lengths between 304.8 mm and 457.2 mm. Tie bars
with a diameter between 9.5 mm and 15.875 mm and a length of 508 mm, 609.6 mm
or 762 mm were installed at longitudinal joints. The choice of dowels and tie bars
depended upon the thickness of the concrete slabs.
All test sections were instrumented by the use of strain gauges at the longitudinal edges
of concrete slabs and linear vertical displacement transducer at the corner of concrete
slabs. Each test section was subjected to a particular traffic load regime as described in
Table 3-1. Further information on AASHO tests can be found elsewhere (AASHO,
1962).
Results of the AASHO test showed that the performance of reinforced concrete
pavement with the particular thickness is relatively better than that of unreinforced
concrete pavement with the same thickness. Although a variety of truck loads was used
64
in the AASHO test, only single axle vehicles with a load between 54.432 kN and
136.08 kN were selected to empirically study dynamic effects of moving loads on the
pavement responses. To do this, different speeds from 3.22 km/hr to 96.56 km/hr were
considered. Results indicated that an increase in truck speed from 3.22 km/hr to 96.558
km/hr decreases the magnitude of induced tensile stress and deflection in the concrete
slabs by about 29 percent.
Table 3-1. Traffic load regime at different loop in AASHO test (1962)
This table is not available online.

It is interesting to note that the most critical failure mode in AASHO (1962) test
sections was erosion of subbase or subgrade materials, whereas, the predominant
failure modes in many rigid pavements are faulting and fatigue cracking (Roesler et al.,
2000).
A full-scale field test was also conducted by Izquierdo et al. (1997) to empirically
study the performance of concrete pavements under heavy dynamic loads. For this
purpose, two sets of full-scale concrete slabs with 244 mm thickness were constructed
over a hot asphalt base resting on a stiff subgrade. One of the test sections was fully
dowelled and the other one had no dowel or tie bars at transverse or longitudinal joints.
Distances between longitudinal joints and transverse joints were considered to be 3.65
m and 6 m respectively.
Each test section consisted of three plain (unreinforced) concrete slabs in the
longitudinal direction and two traffic lanes in the transverse direction. A total of
65
fourteen linear vertical displacement transducers were installed at the wheel path. The
test sections were then subjected to moving truck loads passing along free edge of the
concrete slabs. Different speeds from 0 to 64 km/hr were considered in this study. The
truck consisted of a SAST, a TADT and a Tandem Axle Octa Tyre (TAOT) carrying a
load of 47.73 kN, 368.92 kN and 300.88 kN respectively. Further information on the
test sections can be found elsewhere (Izquierdo et al., 1997).
Result of this field test also showed that speed has no significant effect on concrete
pavement responses. Nevertheless, a finite element program called “UPR-PAVI2” was
developed based on the results of the field tests by Izquierdo et al. (1997) to perform a
parametrical study of the effects of those factors affecting dynamic structural
behaviour of concrete pavement. The theoretical analysis indicated that truck speed
can noticeably change the value of base deflections or stresses in a plain concrete
pavement resting on a subbase with low stiffness.
3.5. Concrete Pavement Distresses

Concrete pavement distresses can be related to the fatigue of concrete slab or erosion
of the subbase and subgrade materials. As discussed earlier in the research scope
(Section 1.5 of this document), this research will focus on fatigue related distresses.
Since erosion of the subbase and subgrade materials is one of the most significant
distresses in jointed concrete pavements, some fundamental information on erosion of
the subbase and subgrade materials is provided at the end of this section.
3.5.1. Fatigue Damage of Concrete Slab

As the main reason behind deterioration and delamination processes, cracks can be
considered as a tensile failure in concrete pavements. Cracks can occur at any location
within the pavement where tensile stresses exceed the concrete flexural strength.
Tensile stresses are generally induced in a concrete pavement due to the bending action
of the concrete slab under vehicular as well as climatic forces.
Since the applied loads are repeatable in nature, concrete pavements fail under fatigue
phenomenon rather than direct failure under maximum induced tensile stress. For more
information on concrete fatigue life see Section 2.2.3 of this Chapter. Fatigue of the
concrete slabs may results in transverse, corner, longitudinal or punchout cracking.
66
Cracks are formed from the top surface layer toward the bottom surface layer of the
slab (top-down cracking) or from the bottom surface layer toward the top surface layer
of the slab (bottom-up cracking).
Bottom-up mid-edge transverse cracking under vehicular loads is the traditional failure
mode of concrete pavements. It is the only fatigue failure mode of the concrete slab
considered in the PCA method. Since many jointed concrete pavements suffered from
corner and longitudinal cracking (Heath et al., 2003), differential temperature together
with different boundary conditions between the concrete slab and subbase were
considered in concrete pavement analysis to find reasons behind concrete pavement
deteriorations.
Fatigue failure modes in the jointed concrete pavements

Maximum vehicular induced tensile stress occurs at the bottom of the slab when
vehicular loads are applied at the mid length of longitudinal joints between transverse
joints and close to the longitudinal joints (see Fig. 2-20a). In 2005, Hiller and Roesler
used the influence stress lines to determine the critical location of fatigue damage
under certain truck loads in a typical California concrete pavement with permanent
built-in curling. They found that the critical damage location in the absence of
environmental effects for a load transfer efficiency of 70% was at the bottom surface
layer of the mid slab edge.
Buch et al. (2004) parametrically investigated the structural response of jointed

concrete pavement under single axle (SA), tandem axle (TA), triple axle (TR), quad
axle (QA), multi axles, together with differential temperature using influence stress
line approaches. In addition, diverse truck loads as a combination of different axle
loads were also studied. Their results indicated that mid-edge loading causes bottom-
up cracking for SA, TA, TR and QA.
Tensile stresses at the bottom surface layer of concrete slab are also induced due to
daytime differential temperature gradients (AASHTO, 2003). A combination of the
above mentioned induced tensile stresses results in bottom-up transverse cracking in
JPCP. The possibility of bottom-up transverse cracking increases in the absence of
load transverse devices at longitudinal joints (Ongel and Harvey, 2004).
67
To limit density of the bottom-up transverse cracking in JPCP, AASHTO (2003)
recommends to:
- Increase slab thickness

- Reduce joint spacing
- Use widened slab
- Use concrete mix with lower CTE
- Provide tied concrete shoulder
- Use concrete with a higher strength
- Use a stabilized subbase
Buch et al. (2004) also found that an increase in the subgrade reaction or thickness of
the concrete slab increases or decreases the induced tensile stresses respectively. Their
results showed that corner loading results in top-down cracking for SA, TA, TR and
QA. However, cracks are usually initiated and distributed in the top surface layer of the
concrete slab and then propagated downward in the concrete slab depth due to
environmental effects before applying any traffic load on it (Heath and Roesler, 2000).
In general, combination of excessive upward slab curling, loss of slab support (slab
lift-up), and repeated heavy-truck loadings is the main reason for top-down cracking
(Beckemeyer et al., 2002).
Finite element analysis of a JPCP in hot weather condition subjected to a truck load
showed that maximum vehicular induced tensile stress shifts toward mid-length of
concrete slab due to loss of support (Hansen et al., 2002). Results of Hiller and Roesler
(2005) showed that top-down transverse cracking near the mid slab edge was the
critical failure mode in the presence of a nighttime differential temperature of -16.5 ˚C.
The critical load condition in top-down cracking is caused from a combination of axle
group loads when applied on the opposite ends of a slab simultaneously and in the
presence of high negative temperature gradients (AASHTO, 2003). In this case, the
crack is initiated at the top layer of the slab close to longitudinal joints and midway
between transverse joints (Fig. 3-31).
68
Figure 3-31. The critical location of the applied load for top-down transverse cracking
(AASHTO, 2003)
Top-down cracking can be controlled by using recommendations provided for bottom-

up cracking. Furthermore, lower permanent built-in curling after concrete placement
can modify resistance of the slab to top-down cracking (AASHTO, 2003).
Loss of support due to nighttime temperature gradients or erosion of subbase and

subgrade materials at joints associated with the corner load position leads to corner
cracking (Channakeshava et al., 1993). Byrum and Hansen (1994) contributed the
influence function lines in analysis of jointed concrete pavement under environmental
effects and wheel load. Their results indicated that maximum stress occurs at some
distance away from the joint when the load passes across the joint. Similarly, Heath
and Roesler (2000) and Khaznovich et al. (2001) studied fatigue distresses in the JPCP.
Their results showed that cracks may also occur near the corner of the slab when
concrete pavement is subjected to a combination of loads and environmental effects.
Longitudinal cracking can be also explained by excessive differential drying shrinkage
(Hiller and Roesler, 2002).
Joint faulting is due to either concrete slab settlement or erosion of subbase or

subgrade materials at the bottom of the concrete slab. It usually occurs in the jointed
concrete pavements due to repeated heavy axle loads crossing transverse joints (Fig. 3-
32). It causes severe problems such as loss of ride quality, lowering LTE and erosion
of subbase or subgrade materials. Joint faulting increases width of joints and
consequently decreases the LTE (Popehn et al., 2003). Provision of dowels and
69
shoulders, the use of thicker dowel, the use of shorter joint spacing, the use of none
erodible materials in subbase and subgrade or the use of an appropriate drainage
system in sublayers can effectively restrict joint faulting (AASHTO, 2003).
Note that experimental study on crack sensitivity to different factors showed that
coarse aggregate could modify the crack generation and propagation by about 30
percent (Panda and Ghosh, 2002).

Figure 3-32. The critical location of applied load for joint faulting (AASHTO, 2003)
Fatigue failure modes in the CRCP

Transverse cracks are produced by the stress-relief process caused by volumetric
change in concrete due to drying shrinkage and seasonal thermal strains in the CRCP
within 2 years after construction (Selezneva et al., 2004). Arrangement of transverse
reinforcements near the edge of longitudinal joints has a strong effect on crack width
and the magnitude of induced stress in concrete. Air temperature and air content
significantly affect crack spacing (Zwerneman et al., 1995).
A sensitivity analysis of CRCP was conducted by Nam et al. (2003) using CRCP-10
computer program to determine effects of different variables on the CRCP
performances. These variables were geometry, materials properties, bond-slip
relationships between concrete and reinforcement, debonding layer, climatic forces and
wheel loads. Their significant results can be summarised as follows:
70
- Crack spacing is more sensitive to a change in the above mention variables
than crack width.
- An increase in concrete thickness, concrete flexural strength and diameter of
reinforcement increases the crack spacing. On the other hand, crack spacing
decreases with an increase in the concrete CTE and reinforcement ratio.
- Diameter of reinforcement and vertical stiffness of underlying layers have no
significant effect on the CRCP behaviour.
In terms of deformation under the applied loads, Chen and Deng (2001) found that:
- The ratio of reinforcement and the elastic modulus of the foundation soil
have no significant effects on deflection of CRCP under the load.
- Pavement deflection is considerably affected by concrete slab thickness and
configuration of the applied load.
- The location of reinforcement has little effect on the pavement responses.
Punchout is another major structural distress associated with CRCP. Punchout results
in a longitudinal crack due to a high tensile stress occurring at the top of the concrete
slab and 1016 to 1524 mm away from its edge, when axle load is located near the
longitudinal edge of the slabs and between two closely spaced transverse cracks (Fig.
3-33). This consequently results in the loss of ride quality. The induced tensile stress
has a reverse relationship with LTE in the transverse cracks and a direct relationship
with loss of support along the edge of the concrete slab.

Figure 3-33. Critical load location for CRCP punchout crack (AASHTO, 2003)
71
Consideration of longitudinal reinforcement to decrease the crack width and
consequently to increase the LTE, decreasing the concrete CTE, placing the
reinforcement bars above mid-depth of concrete slab, increasing concrete slab
thickness, providing tied concrete shoulder, the use of stabilizes subbase, reducing
built-in curling after placement and the use of stronger concrete are predominant
factors controlling the CRCP punchout distress (AASHTO, 2003).
There is a relationship between transverse crack, critical stress in longitudinal crack,

and punchout development (Selezneva et al., 2003). Kim et al. (2000) showed that
induced tensile stress at the top surface layer of the concrete slab above transverse
reinforcement can be relatively higher than that at the top surface layer of the centre of
concrete slab. Selezneva et al. (2004) developed a mechanistic-empirical structural
design procedure for punchout in the CRCP.
3.5.2. Erosion of Subbase and Subgrade Materials

Erosion of subgrade or subbase materials is one of the main distresses in JRCP, which
results in the loss of support due to void distribution under the slab beneath. Initial
voids may be the result of curling or warping of the concrete slab or even plastic
deformation of subgrade materials. The secondary voids are formed by pumping
transportable subbase and subgrade materials due to deflection of concrete slab in
presence of water (Fig. 3-34).

Figure 3-34. Erosion mechanism (Austroads, 2004)
Erosion happens through longitudinal and transverse joints at the beginning of the
pavement operation and later in the pavement life, through cracks (Bahatti et al.,
1996). The Purdue method for analysis of concrete pavements known as PMARP and
72
the Iowa pumping model known as IAPUMP were respectively developed by Larralde
(1984) and Barlow (1994) to analyse the erosion problems.
Since erosion of subbase and subgrade materials is outside of the scope of this
research, further information on this topic can be found elsewhere (Larralde 1984,
Barlow 1994 and Bhatti et al. 1996).
3.5.3. Spalling
Spalling is another concrete pavement distress occurring due to delamination stress
produced by wheel loads, temperature change and the presence of moisture in the
delamination zone (Zollinger et al., 1994). It often occurs at both longitudinal and
transverse joints, approximately of 150 mm away from the joint or the crack (Wang
and Zollinger, 2000).
3.6. Concrete Pavement Design Guides

In its first thickness design based on fatigue process, the Portland Cement Association
(PCA, 1933) used Older’s empirical equations (1924) for calculation of induced
stresses at the corner of a concrete slab subjected to a wheel load. The Portland
Cement Association revised the PCA method in 1951 based on results of several
research projects conducted in the different states of the USA. In this revision, the
Pickett’s equations for stress calculation at the corner of a concrete slab were taken
into consideration. Consequently, effects of slab thickness, wheel load, tyre pavement
contact area, and modulus of subgrade reaction were contributed in the PCA method.
While fatigue failure modes were also considered in the 1933 version, new data on
fatigue curve, which showed that static loads affect concrete pavement stress more
than moving loads, and Pickett-Ray influence charts were subsequently taken into
account in the PCA method published in 1966.
The yield line theory was then developed by Meyerhof (1962) to determine a concrete
pavement response under point loads when the load is positioned at the interior area of
the concrete slab. His theory was based on a circular failure shape around the point
load due to positive and negative curvature on opposite sides of the failure interface
(Fig. 3-35).
73
Figure 3-35. Meyerhof’s yield line theory (1962)
The AASHTO in 1972 published a pavement design guide based on the evaluation of
results captured in road tests conducted by the state highway agencies. The concept of
equivalent single axle load (ESAL) was introduced to simplify the design procedure in
the AASHTO method.
Results of Packard and Tayabji (1985) on thickness design of concrete highway and
street pavement were then taken into account in the latest version of the PCA method
published in 1984. The new PCA procedure was extensively revised and for the first
time the concept of erosion of subbase and subgrade materials was considered in
concrete pavement damage process.
Since Austroads (1992) extended the PCA method to suit Australian conditions, the
recent revision of Austroads (2004) and AASHTO (2003) pavement design guides will
be reviewed in the following sections.
3.6.1. Austroads 2004

3.6.1.1. Introduction
In 2004 Austroads released a revision of the 1992 Guide for concrete pavement design
based on the work of Packard and Tayabji (1985), known as the PCA method. Since
the shortcomings of the PCA method have been widely recognized by engineers and
74
road practitioners during the past decades, current Australian practices and materials
were taken into consideration in the 2004 Guide (Vorobieff, 2001).
The Guide provides a mechanistic procedure for calculating the required concrete slab
thickness for JPCP, JRCP, CRCP and Steel Fibre Concrete Pavement (SFCP). The use
of the Guide was restricted to those pavements whose dimensions (distance between
transverse and longitudinal joints) were less than those provided in Table 3-2.
Table 3-2. Distance between joints (m) for different types of concrete pavement
Distance Distance
Type of Instruction on Instruction on
Availability between between
concrete Transverse Longitudinal
of dowel Transverse longitudinal
Pavement Joints Joints
Joints Joints
No 4.2 Skewed Joints
JPCP
Yes 4.5 Square Joints
JRCP Yes 8.0 to 12.0 Square dowel
As long as 4.3 Tied

CRCP No construction NA
joints
SFCP Yes 6 --------
A variety of inputs including design traffic, subgrade CBR, subbase thickness and
type, project design reliability (PDR), concrete flexural strength, vehicular load spectra
(axle group load distributions) and provision of dowels and shoulders are taken into
account to calculate the required concrete slab thickness based on the cumulative
damage due to fatigue of the concrete slab and erosion of subbase and subgrade
materials. The severity of fatigue and erosion damage depends on structural response
of concrete pavements as affected by vehicular load configurations, environmental
factors, and material/layer characteristics.
Design traffic is an estimation of heavy vehicle volumes on the road during the life of a
pavement. The method of estimating the number of Heavy Vehicle Axle Groups
(HVAGs) has been described in the Guide. Although the PCA method was developed
based on three types of axle loading i.e. SADT, TADT and TRDT, the Guide extended
75
the method to cover SAST, SADT, TAST, TADT, TRDT and QADT (Fig. 2-17).
Furthermore, the fatigue analysis of the method was modified as pavement responses
to tandem and triple axle group loads had not been sufficiently considered in the PCA
method (Vorobieff, 2001).
The California Baring Ratio (CBR) is the only subgrade information used in the design
procedure and represents the subgrade resistance to applied load. Subgrade CBR
values are affected by topography, soil type, and drainage conditions. The Guide
provides a number of methods to estimate the field CBR values under various
conditions.
Five types of subbase including 125 mm bound, 150 mm bound, 170 mm bound, 125
mm Lean-Mix Concrete (LMC), and 150 mm LMC have been recommended in the
Guide. The choice of subbase depends upon the value of the design traffic (Table 3-3).
It should be noted that 150 mm LMC is the only choice for design traffic greater than
1×107 HVAGs. Effective subgrade strength has been defined in the Guide to consider
effect of the subbase layer on concrete pavement behaviour. The Guide provides a
graphical estimation of effective subgrade strength (Fig. 3-36) based on subgrade CBR
and subbase types.
Table 3-3. Selection of subbase type

Design Traffic (HVAG) Subbase Type
Up to 106 125 mm bound
Up to 5 x 106 150 mm bound or 125 mm LMC
Up to 1 x 107 170 mm bound or 125 mm LMC
Greater than 1 x 107 150 mm LMC
Variations in drainage condition, traffic loads, material properties, and the construction
process require the Guide to define a range of PDR values between 80 per cent and
97.5 per cent depending on road classification. Although PDR is not directly used in
the design procedure, it defines the value of Load Safety Factor (LSF) which is directly
used in the procedure and varies between 1.05 and 1.35 depending on concrete base
types.
76
Figure 3-36. Estimation of effective subgrade CBR (Austroads, 2004b)
Design traffic is also employed to estimate the expected load repetitions for a given
axle group by using a typical traffic load spectrum (axle load distribution) during the
pavement’s life. Presumptive traffic load distributions for urban and rural roads are
provided in the Guide (Tables 3-4 and 3-5), which can be employed if specific traffic
load spectra are not available. Effects of dowels and shoulders provision on thickness
of concrete slab have been considered through the use of design coefficients for
different axle group types for both fatigue and erosion analyses.
3.6.1.2. Method Description

As mentioned earlier, the method is based on fatigue and erosion analyses. In this
method, the value of the equivalent stress ( S e ) for fatigue analysis, or the erosion
factor (F3) for erosion analysis, is firstly determined from equation 9.4 of Austroads
(2004b), which is shown below as Equation 3-22.
b d f . ln( E f )
S e = F3 = a + + c. ln( E f ) + 2 + e.[ln( E f )] 2 +
T T T
g 3
i.[ln( E f )]2 j. ln( E f ) (3-22)
+ + h.[ln( E f )] + +
T3 T T2
77
Where a, b, c, d , e, f , g , h, i, j are fatigue or erosion coefficients which can be obtained
from relevant tables in Chapter 9 of the Guide, T is base thickness (mm), E f is
effective subgrade CBR.
Determination of the stress ratio (Sr) for a given axle load is the next step in the fatigue
procedure. S r is calculated using an equation in Section 9.4.2.1 of the Guide, presented
here as Equation 3-23.
Se P . L SF 0 .94
Sr = [ ] (3-23)
0 . 944 f r 4 . 45 F1
Where S e is equivalent stress determined from Equation 3-22, f r is concrete flexural

strength at 28 days (MPa), P is axle group load (kN), F1 is load adjustment for fatigue
due to axle group, and LSF is load safety factor.
The allowable load repetition in terms of fatigue analysis is estimated from equations
9.2 or 9.3 or description provided in Section 9.4.2.1 of the Guide, shown here as
Equations 3-24 to 3-26.
0.9719 − S r
log( N f ) = [ ] when Sr > 0.55 (3-24)
0.0828
4.258 when 0.45 ≤ Sr ≤ 0.55 (3-25)
Nf =[ ]3.268
S r − 0.4325
Nf =∞ when Sr < 0.45 (3-26)
The allowable load repetition in terms of erosion (Ne) is estimated based on Equation
9.5 of the Guide, presented here as Equation 3-27.
P.LSF 2 10 F3
log( F2 .N e ) = 14.52 − 6.77[max(0, ( ) . − 9.0)]0.103 (3-27)
4.45 F4 41.35
Where P and LSF are as previously defined, F2 is adjustment for slab edge, F3 is
erosion factor calculated from Equation 3-22, and F4 is load adjustment for erosion
due to axle group type.
78
Table 3-4. Presumptive traffic load distribution for urban roads
Axle Group Load Axle Group Type
(KN) SAST % SADT % TAST % TADT % TRDT %

10 0.2804 3.473 0.0354 0.1444 0.005
20 7.827 8.696 0.2377 0.5755 0.1568
30 15.46 23.46 0.2763 0.6242 0.329
40 15.71 21.93 0.5755 1.977 1.317
50 29.94 16.8 2.889 6.496 4.167
60 23.29 9.606 10.27 9.511 7.419
70 6.502 6.5 16.81 10.94 9.777
80 0.7943 4.623 16.61 9.769 8.338
90 0.1087 2.969 15.95 7.611 6.15
100 0.0354 1.393 14.42 7.242 5.029
110 0.0174 0.4098 9.774 6.267 3.701
120 0.0174 0.1158 5.903 5.952 3.298
130 0.0174 0.0244 2.943 5.878 3.147
140 1.539 6.534 3.361
150 0.8439 8.03 4.008
160 0.4279 5.717 4.115
170 0.2308 3.554 4.819
180 0.1367 1.863 6.097
190 0.0723 0.8535 7.733
200 0.0555 0.3331 8.433
210 0.0801 5.136
220 0.0322 2.339
230 0.016 0.7764
240 0.2503
250 0.0905
260 0.008
Total 100 100 100 100 100
Proportion of each
0.393 0.191 0.009 0.259 0.148
axle group
79
Table 3-5. Presumptive traffic load distribution for rural roads
Axle group type
Axle Group Load (KN) SAST % SADT % TAST % TADT % TRDT %
10 0.5034 3.031 0.0082 0.3815 0.0127

20 6.738 4.968 0.0844 0.5656 0.1246
30 9.237 6.858 1.285 2.364 1.878
40 15.3 8.484 1.986 4.009 3.859
50 28.74 10.23 5.115 4.255 5.213
60 32.19 14.16 11.48 5.173 4.359
70 5.653 13.58 15.24 4.866 4.078
80 1.211 13.29 16.19 4.846 3.843
90 0.2496 11.17 14.59 5.297 3.79
100 0.0768 7.837 15.37 6.244 3.842
110 0.0561 4.13 11.13 9.314 4.116
120 0.0312 2.002 6.396 9.173 3.574
130 0.0139 0.1785 0.6323 10.55 6.69
140 0.07 0.2245 12.92 7.893
150 0.0115 0.1028 10.35 8.357
160 0.0911 3.767 8.126
170 0.0533 2.566 7.379
180 0.0214 1.598 6.076
190 0.8682 4.981
200 0.4341 4.424
210 0.2348 2.83
220 0.1346 1.928
230 0.0613 1.138
240 0.0279 0.6311
250 0.3486
260 0.2209
270 0.1293
280 0.0815
290 0.0441
300 0.0253
310 0.0079
Total 100 100 100 100 100
Proportion of each axle
0.344 0.098 0.007 0.320 0.231
group
80
Using Miner’s rule, the percentages of fatigue damage and erosion damage are
determined by calculating the percentage ratio between the expected load repetition
and the relevant allowable load repetition (derived from Equations 3-24, 3-25 or3-26,
and Equation 3-27). Cumulative fatigue and erosion damage percentages are then
calculated separately for all axle loads and axle groups. The base thickness is
considered to be adequate if both these values do not exceed 100% and the calculated
base thickness is more than the minimum recommended base thickness provided in
Section 9.4.3 of the Guide.
3.6.2. AASHTO 2003

3.6.2.1. Introduction
In 2003 AASHTO released its new revision of concrete pavement design procedure for
JPCP and CRCP. A wide range of factors including arrangement in structural layers,
joint spacing, provision of dowels and tied concrete shoulder or widened slabs, base
type, drainage, subgrade properties including subgrade modulus of reaction and
presence of bedrock, design traffic, climatic data, the initial smoothness, estimation on
construction period in month, estimation on pavement age at the time the pavement is
opened to the traffic and estimated permanent curling or warping condition of the
concrete slab has been taken into consideration in the AASHTO (2003) guide. It needs
to be noted that the definition of layers used in the AASHTO is not completely
compatible with those defined by the Austroads (2004b). Further information on this
matter is provided later under a subtitle of “Pavement structure”.
The significant shortcoming of the AASHTO method is the availability of the

information related to the project after the construction. This information is not
generally available at the design time. Hence, this information needs to be estimated in
advance by using specific complex methods described in the Guide.
Since the AASHTO design procedure needs to consider interrelationships and

correlations between huge varieties of input data which is outside the scope of this
thesis, only fundamental information on the Guide will be presented in the next
section.
81
3.6.2.2. Methodology
The accuracy of the results in the AASHTO design Guide depends upon the accuracy
of input data. Three accuracy levels for input parameters have been defined:
- Level 1: Input data are the results of direct tests.

- Level 2: Required parameters are determined by the use of correlations such
as determination of subgrade modulus of reaction based on its CBR value.
- Level 3: Input data are defined in accordance with national default values or
local experiences.
A variety of input data is taken into consideration in the AASHTO method (2003)
which can be categorised into five separate areas. These are general information, traffic
load, climate forces, drainage and surface properties, and pavement structure.
General information
General information includes the following factors:
- Project design life

- End of construction period in month. This parameter is used to estimate the
ambient temperature and moisture at construction time. These parameters are
then used to estimate the permanent curling and warping condition in the
concrete slabs after placement.
- Time when the pavement is opened to the traffic. This parameter is used to
estimate concrete strength at the time when the concrete slabs are subjected
to the vehicular loads.
- Pavement types involving JPCP or CRCP.
- Estimation of initial IRI which ranges from 789 to 1578 mm / km.
- Performance criteria. The AASHTO Guide has considered a variety of
performance criteria for the design of concrete pavements. These are
transverse cracking and joint faulting in JPCP, crack width, load transfer
efficiency at the crack, and number of punchout cracks in CRCP. The
allowable range for each individual factor is 10% to 45% for transverse
cracking, 2.54 mm to 5.08 mm in joint faulting, less than 0.51 mm for crack
82
width, more than 90% of LTE at the crack and 6 to 12 punchout cracks per
km of CRCP.
- Surface smoothness which can be controlled through the use of IRI. The
allowable value of IRI ranges from 2366 mm / km to 3944 mm / km.
Traffic
Different axle groups involving single axles (SA), tandem axles (TA), tridem (TRA)
and quad axles (QA) have been considered in the AASHTO Guide. Axle loads vary
between 13.3 to 182 kN in SA, 26.7 to 364 kN in TA and 53.4 to 453.7 kN in TRA and
QA. The axle load distribution for each axle is presented in an interval value of 4.45
kN, 8.9 kN and 13.35 kN respectively.
Standard deviation of traffic wander, width of traffic lane, mean wheel location
(average distance between outer tyre pavement contact area and longitudinal joint or
edge), number and type of axles per truck class, axle configuration and wheel base
(distance between axle groups in a given truck class) are other parameters involved in
the AASHTO guide.
Traffic volume is calculated based on hourly traffic distribution within the pavement
and traffic growth factor. It is then adjusted per each month of the pavement service
life using adjustment factor. This will then be used to estimate the performance
criteria.
Drainage and surface properties

In accordance with the AASHTO design procedure, differential temperature gradient
and moisture content within the depth of the concrete pavements are affected by
properties of the pavement surface layer. The Guide defines four parameters to specify
the magnitude of climatic effects. These are pavement surface short wave absorptivity,
potential for infiltration, pavement cross slope, and length of drainage path.
The short wave absorptivity is the ratio of absorbed solar energy by the pavement to
the total solar energy radiated to the pavement surface. This parameter depends on
pavement composition, colour, and texture and can be used to determine the magnitude
of temperature gradients in depth of the pavement. This parameter ranges from 0 (no
83
absorption) to 1 (entire absorption) and varies from 0.7 to 0.9 for different concrete
surfaces. The recommended value for this parameter is 0.85.
The infiltration shows the amount of water entering the pavement structure due to a
given rainfall event. Provision of shoulder can significantly affect the magnitude of
this parameter. Hence, the following regimes have been considered in the Guide:
- Minor: Tied concrete shoulders are used or edge-drain is presented under the
shoulder.
- Moderate: This is valid for all other shoulder types.
- Extreme: Not used for new or reconstructed pavement.
Pavement cross-slope is the transverse slope of pavement surface and defines the
required time to drain free water from the pavement base or subbase layer. Drainage
path length defines the horizontal distance between the highest point of the pavement
and the point where drainage occurs.
Pavement structure
This method is able to consider a variety of layers within the concrete pavements.
These include concrete slabs, unbonded course asphalt or cement base, unbonded
stabilized subbase, compacted subgrade, natural subgrade, bedrock. Figure 3-37
shows a schematic cross section of the concrete pavement considered in the AASHTO
(2003).

Figure 3-37. Illustration of possible concrete pavement layered system (AASHTO,

2003)
84
Since complex methods need to be considered to gather all the required information for
design procedure, a specific program was developed by the AASHTO (2003) to
present the following information:
- Average hourly number of single, tandem, tridem, and quad axles in each
axle load category for each month of the analysis period.
- Minimum one year’s weather station data used to determine the value of
temperature at 11 evenly spaced nodes in the concrete slab layer for every
hour of the available climatic data.
- Average monthly relative humidity for each month.
- Estimation of the concrete strength and modulus of elasticity for each month
of analysis period.
- Determination of the base strength in each month.
- Monthly average of effective subgrade modulus of reaction.
3.6.2.3. JPCP Design Features

An increase in distance between transverse joints increases the stress value in the
pavement. As a result, a distance of 3658 mm between longitudinal joints and 4572
mm between transverse joints has been recommended in the Guide. Dowel diameter
has a reverse relationship with joint faulting. Hence, the adequate dowel diameter is
calculated through the performance criteria developed for joint faulting. The
recommended dowel size is 30.5 mm.
Two types of shoulder, namely tied and widened, have been defined in the Guide. The
value of LTE in tied shoulder need to be between 50% and 70 % for monolithically
constructed shoulders and between 30% and 50% for separately constructed shoulders.
The slab width is 4267 mm if a widened shoulder is used.
Bonded or unbonded boundary condition between concrete slabs and base layer can be
considered through the use of friction reducer layer. In terms of erosion of subbase
and subgrade materials, five classes of base erodibility have been developed in the
Guide. They are:
- Class 1: Extremely erosion resistant material
85
- Class 2: Very erosion resistant material
- Class 3: Erosion resistant material
- Class 4: Fairly erodible material
- Class 5: Very erodible material
3.6.2.4. Distress Prediction

Comprehensive information on pavement distresses was provided in Section 2.5.1 of
this chapter.
3.6.2.5. Surface Roughness

The development of pavement roughness (as measured by IRI) depends upon the
subsequent development of distresses including slab cracking, joint faulting and joint
spalling for JPCP, and punchouts for CRCP over a period of the time. IRI is
incrementally estimated over the entire design period on a monthly basis.
3.6.2.6. Thickness of concrete slab

Thickness of concrete slab is calculated to satisfy the above mentioned performance
criteria such as joint faulting, slab cracking, punchouts, and IRI. It is determined by the
use of cumulative damage percentage for relative performance criteria. The slab
thickness is adequate provided that the magnitude of different damage processes
satisfies the minimum relevant requirement described in the Guide.
3.7. Summary
A comprehensive literature review on factors affecting concrete pavement
performance has been presented in this chapter. Fatigue of the concrete slab may
produce top-down or bottom-up cracking. Cracks may occur at any location within
the concrete slab. Corner, longitudinal and transverse cracks are the most common
fatigue failure modes of concrete pavements. Location and density of cracks can be
predicted based on distribution of induced vehicular and environmental tensile stress
within the concrete slab. However, analysis of concrete pavements under vehicular
loads and environmental effects depends on a vast number of parameters and
interrelationships among them. These parameters are related to characteristics of
concrete slab, subbase and subgrade, type of debonding layer between the concrete
86
slab and the subbase, the type of debonding layer between concrete slab and subbase,
the provision of shoulder and dowel bars, traffic loads and environmental effects.
In spite of progressive knowledge on interactions between the above mentioned

parameters and their effects on concrete pavement responses, some shortcomings have
been recognised in assumptions used in the past research. Consequently, the concrete
pavements constructed worldwide still suffer from initiation and propagation of fatigue
cracking. These shortcomings will be discussed in the next chapter (Chapter 3) to
develop a sufficient research plan to overcome the problems.
The methodology of concrete pavement design guides developed based on

mechanistic approach, i.e. Austroads (2004), and mechanistic-empirical approach, i.e.
AASHTO (2003), were also presented in this chapter. In contrast with the mechanistic
approach, the empirical-mechanistic approaches are more sophisticated but more
reliable.
87
Chapter 4
SUMMARY OF THE LITERATURE REVIEW AND

RESEARCH PLAN
4.1. Summary of the literature review

Sustainability of concrete pavements depends on the rate of pavement deteriorations.
Cracks, as the main reason behind deteriorations of concrete pavements, are initiated
at the top or the bottom surface layer of the concrete slab and then propagated into
the depth of concrete slab due to fatigue of the concrete and/or erosion of subbase or
subgrade materials.
Since erosion of subbase or subgrade materials can be eliminated by considering

non-erodible materials and/or sufficient drainage system which keeps the level of
underground water away from bottom of the concrete pavement, this research
focuses on fatigue related concrete pavement distresses. Fatigue of the concrete slab
can result in top-down or bottom-up cracking which may occur at any location within
the concrete slab. Corner, longitudinal and transverse cracks are the most common
fatigue failure modes of concrete pavements. The location and density of cracks can
be predicted based on induced vehicular and environmental tensile stress distribution
within the concrete slab.
Analysis of concrete pavements under vehicular loads and environmental effects

depends on a vast number of parameters and interrelationships among them. These
parameters can be summarised in nine different categories:
• Concrete properties. These include flexural strength, modulus of elasticity,

fatigue life, coefficient of thermal expansion and shrinkage.
• Subgrade characteristics. Information on the modulus of subgrade reaction or
CBR or modulus of elasticity, Poisson’s ratio, the angle of internal friction
and the coefficient of internal friction shall be taken into consideration to
effectively simulate the subgrade behaviour under the applied loads (refer to
88
Section 2.3.1 of this thesis document for further information on soil
modelling idealization).
• Subbase type, thickness and strength.
• Characteristics of the concrete slab including:
- Pavement type (JPCP, JRCP, CRCP or SRCP).
- Thickness of concrete slab.
- Joint spacing (distance between transverse joints or longitudinal
joints).
- Type of joints including joint width, type of load transfer devices
(dowel in transverse joint or tie bar in longitudinal joint, aggregate
interlock or combination of them) and size and location of dowels and
tie bars in the case of jointed concrete pavements.
- Type, size and location of reinforcement in the case of reinforced
concrete pavements.
- Pavement roughness.
• Debonding layer:
- Fully bonded.
- Fully unbonded.
- Partially bonded. Extra information such as coefficient of friction or
shear stress–slip displacement curve is required to accurately simulate
the effect of this layer on concrete pavement responses (for further
information on this matter see “Model of debonding layer” in Section
2.5.2.2).
• Provision of shoulder (width and thickness).
• Traffic loads:
- Type of traffic loads (axle group loads or truck loads).
- Type of axle group (SAST, SADT, TAST, TADT, TRDT or QADT).
- Distance between axles in a given axle group and distance between
axle groups if truck loads is studied.
- Suspension systems (mechanical or air suspension).
- Load distribution per each individual axle group.
- Traffic wander (location of axle group upon the pavement).
- Speed of vehicular loads.
89
- Frequency of the vehicular loads.
- Axle width.
- Distance between centres of dual tyres.
- Tyre inflation pressure.
- Tyre pavement contact stress.
- Tyre pavement contact shape.
• Environmental effects:
- Temperature fluctuation (permanent built-in curling, daytime and
nighttime differential temperature gradients).
- Shrinkage-loss of moisture content.
In addition to properties of concrete ingredients, the concrete properties strongly rely

on the curing method. The environmental effects also depend on several factors
including air temperature, ratio of the top surface area of concrete slab to its depth,
duration and density of solar radiation, rain fall, thermal conductivity of concrete,
coefficient of thermal expansion of concrete, transverse and longitudinal slopes of
the pavement surface, surface texture and type and location of drainage system.
In spite of progressive knowledge on interactions between the above mentioned

parameters and their effects on concrete pavement responses, the concrete pavements
constructed worldwide still suffer from initiation and propagation of fatigue cracking.
A review of the studies carried out in the past indicates that concrete pavement
performance is significantly affected by those parameters related to vehicular loads,
environmental effects and provision of load transfer devices and shoulders. However,
some shortcomings have been recognised in assumptions used in the past research.
These can be summarised as follows:
i) Axle group configurations vary among heavy vehicle manufacturers

and across countries. However, some typical vehicular load configurations
were only considered in concrete pavement studies which are not necessarily
the critical configurations of axle groups for calculating the maximum
induced vehicular stress and deflection of the concrete pavements.
ii) The magnitude of load distribution between axles in a given axle group
is affected by truck suspension systems and pavement roughness. Although
90
results of some dynamic investigations showed the significance of the above
mentioned factors on pavement responses, effects of a load shift between
axles in a given axle group on concrete pavement response is not completely
clear.
iii) Vehicular loads were considered as static loads in most studies

conducted in the past. But, they are dynamic in nature. Results of transient
dynamic analysis of concrete pavements indicated the significant effect of
dynamic loads on those distresses associated with erosion of subbase and / or
subgrade materials. However, effects of moving axle group loads on induced
tensile stresses and consequently fatigue related distresses are not completely
clear. Research conducted in the past were based on a particular moving load
including point load, wheel load, single axle load and tandem axle load
passing symmetrically along an infinite concrete slab simulated as thick or
thin plate resting on Winkler foundation.
As a result, effects of discontinuities (transverse and longitudinal joints) and

/or traffic wander on dynamic response of concrete pavements were not taken
into consideration. Note that consideration of infinite concrete slab also leads
to inaccurate calculation of induced tensile stress as excitations of other
modes of the concrete slabs are eliminated in the analysis. In studies where
dynamic analysis results in greater induced stress than static analysis, on the
other hand, particular information on pavement properties such as pavement
roughness at wheel path and damping ratio was taken into consideration.
While pavement surface is smooth at the beginning of the pavement life, it
becomes rougher during pavement performance. Consequently, the published
information can not be easily extended to capture results of pavement
response to all possible pavement roughness profiles. Information on the
damping ratio of concrete pavement structures is also restricted to a very
small variation used by Kim et al. (2001 and 2002).
To overcome these problems, nonlinear dynamic analysis of concrete

pavements subjected to a combination of non-symmetrical moving axle
groups and temperature gradients shall be taken into consideration (Shoukry
91
and Fahmy, 2002). This can be done by developing complex finite element
models of the concrete pavements having transverse and longitudinal joints.
iv) The debonding layer is another significant factor affecting pavement

responses. The main purpose of the use of debonding layer between concrete
slab and subbase is to decrease the density of early age cracking. However,
provision of a debonding layer results in greater induced tensile stresses in
concrete slabs under vehicular loads than those captured in a fully bonded
concrete pavement.
Since most research conducted in the past is based on either fully bonded or
fully unbonded boundary condition between concrete slab and subbase,
effects of partially bonded boundary condition on pavement responses needs
further investigation.
v) Traditionally, mid-edge and corner loadings were widely considered in

the past research to calculate maximum induced tensile stress and deflection
in concrete slab respectively. Recently published information showed that
concrete pavement responses are strongly affected by environmental effects.
This consequently leads to other critical loading conditions that may be
different from traditional considerations. Furthermore, effects of the
debonding layer on critical loading condition have not been clearly
addressed.
vi) Some reasons behind transverse, longitudinal or corner cracking were

developed in the literature for some particular axle group loadings. As a
result, a question arose whether these findings can be extended to the other
axle group types considered in the Austroads (2004) or not. Note that Hiller
and Roesler (2002, 2005) and Buch et al. (2004) parametrically studied
effects of different axle groups and / or truck loads on concrete pavement
responses. Since the pavement dimensions and cross section considered in
their investigations is not compatible with those considered in Australia,
there is a need for further study on these areas as well.
vii) Estimation of concrete fatigue life is based on information gathered

from flexural beam test. Hence a question arose whether equations developed
92
for fatigue estimation of concrete slab are able to cover curling and warping
conditions in concrete pavements or not. It was noted that thermal induced
tensile stresses occur at the bottom surface layer of the concrete slab during
daytime and at the top surface layer of the slab during nighttime. Vehicular
induced tensile stresses, on the other hand, increase or decrease the
magnitude of thermal stresses. This results in a considerable variation in
induced stress at a point of interest within the concrete slab from compressive
to tensile stresses and vice versa. This may lead to a lower fatigue life of the
concrete than that predicted based on the loading and unloading of the
concrete specimens using third point loading configuration.
The aforementioned shortcomings highlight the need for further studies on concrete
pavement performance under moving axle group loads together with environmental
effects. The results of these studies are used for developing a more realistic approach
in concrete pavement analysis and design as presented in this thesis.
4.2. Research methodology and plan

To address the aforementioned shortcomings, this research was conducted by
Queensland University of Technology (QUT) and a major Australian concrete
producer, Rinker Australia, to study the behaviour and performance of concrete
pavements under both static and dynamic axle group loads as well as environmental
effects. Consideration of all the above mentioned factors affecting concrete pavement
responses is out of the scope of this research as a comprehensive parametrical study
of all the above mentioned factors requires more time than that allocated for this
research.
Consequently, some typical worldwide accepted values for the parameters

comprehensively investigated in the past are considered in this research based on
information available in the literature. The parameters considered to be constant are:
- Modulus of elasticity of concrete equal to 28000 MPa.

- Concrete CTE of 1.1×10-5.
- Pavement surface is considered to be perfectly smooth.
- Width of traffic lane is 3600 mm.
93
- Distance between transverse joints is 4600 mm for JPCP, 9200 mm
for JRCP.
- Joint width is 5 mm.
- Depth of saw cut is ¼ of concrete slab thickness.
- Subbase is always a cement stabilized with 150 mm thickness having
a characteristic compressive strength of 5 MPa.
The Austroads design guide for concrete pavement analysis (2004) was developed
based on the traditional mechanistic method, which is able to only predict the mid-
edge bottom-up transverse cracking. Hence, a sensitivity analysis of Austroads
method (2004) is firstly conducted to determine the interrelationships between
involved parameters in the guide. The results of the sensitivity analysis are then used
for future extension of the method to cover other failure modes of the concrete slab
such as corner and longitudinal cracking.
Since parametrical dynamic studies of a complex structure such as concrete

pavement is not economical because of the huge analysing time required, some
parametrical static investigations are firstly conducted using 3D finite element
program, EverFE. The results of static analysis are subsequently used to develop
sufficient information for dynamic analyses of different concrete pavements
subjected to moving axle group loads. The static analyses are used:
i) To revisit the critical configuration of each individual axle group defined

in Austroads (2004) based on information available in the literature.
ii) To specify how different debonding materials affect concrete pavement

responses.
iii) To determine the effects of environmental forces together with different

boundary conditions between concrete slab and subbase on structural
responses of JPCP and critical location of axle groups upon the pavement.
Accuracy of a finite element analysis relies on properties of materials used in the

model and calibration of the finite element model. As mentioned in the literature,
concrete flexural strength, aggregated interlock, modulus of elasticity, coefficient of
thermal expansion (CTE) and fatigue life of the concrete under repeated loads are the
94
most significant concrete properties affecting stress distribution within the concrete
slab and also switching the concrete pavement damage from one mode to another
mode. As a result, some laboratory tests are conducted to accurately measure all of
the above mentioned parameters with exclusion of CTE. Since the development of a
fatigue prediction model is out of the scope of this research, the information obtained
from fatigue tests is used to select an appropriate fatigue prediction model suited for
Australian concrete technology.
Information provided in the literature for calibration of finite element model

subjected to moving loads is restricted to bonded pavements. Hence, a series of
dynamic analyses of JPCP and JRCP are firstly carried out using ANSYS platform.
An experimental field test of bonded and unbonded JPCP and JRCP are then
conducted to validate the results of the dynamic analysis of bonded concrete
pavements. Furthermore, sufficient information on responses of unbonded concrete
pavements under moving loads is captured. This information is subsequently used to
calibrate the finite element model of the unbonded pavements.
A series dynamic analyses of unbonded concrete pavements under moving axle

group loads is then performed using ANSYS platform. Environmental effects are
then considered in the dynamic analysis. Ultimately, results of this research are used
to develop a new empirical-mechanistic guide for designing of concrete pavements
based on fatigue related distresses.
95
Chapter 5
SENSITIVITY ANALYSIS OF CONCRETE PAVEMENT

DESIGN USING AUSTROADS GUIDE 2004
5.1. Introduction
The 2004 edition of Austroads rigid pavement design guide was developed based on
the work of Packard and Tayabji in the 1980’s, known as the PCA method. As
described in Section 3.6.1 of this thesis, a number of input parameters are needed in the
Austroads method (2004b) to calculate the required concrete slab thickness by
considering the cumulative damage process due to fatigue of concrete and erosion of
subbase or subgrade materials. The input parameters include concrete flexural strength,
subbase characteristics, subgrade strength, project design reliability, vehicular load
spectra (axle group load distributions) and provision of dowels and shoulders.
This chapter presents the results of a sensitivity analysis performed to investigate the
effects of input parameter variations on design of slab thickness. For that purpose the
current design procedure was written in Microsoft Visual Basic to graphically display
the effects of each individual parameter on the calculated concrete slab thickness.
Furthermore, the significance of design traffic, concrete flexural strength, subgrade
CBR, and provision of dowel and shoulder in relation to the results of fatigue and
erosion analyses are discussed.
5.2. Aims of this study

Compared to its predecessor (Austroads, 1992), the Austroads (2004b) is more readily
adapted to computerised design environments. The axle load and axle load spectra
have been revised and design nomographs have been converted into closed-form
equations. The 2004 Guide thus allows a software development for the design
procedure while its predecessor only allows design calculation to be done in a
spreadsheet (Vorobieff, 1996).
In spite of the advantage in a use of complex interrelationships between the

aforementioned design parameters, the 2004 Guide has a number of limitations:
96
• Vehicular loads have been considered as static loads although they are
dynamic in nature.
• Liang and Niu (1998) showed that effects of thermal curling and loss of
moisture warping on concrete pavement responses and deteriorations are
significant. The Guide in Section 9.4.3 provides different minimum concrete
slab thicknesses for different concrete pavements and diverse ranges of design
traffic. These recommended minimum concrete slab thicknesses for design
traffic greater than 1×107 HVAGs are to account for environmental factors.
However, behaviour of concrete pavements under vehicular loads may depend
on the magnitude of differential temperature and/or loss of moisture contents.
Hence, consideration of environmental effects as a certain value, i.e. a
minimum base thickness, may result in other failure types in the concrete slabs
that are not considered in the method.
• The method was developed based on the assumption that the concrete slab and
subbase layers are not bonded (see section 9.1 of the Guide). As a result, some
typical coefficient of friction, µ, values have been provided by the Guide for
partially bonded boundary conditions between the concrete slab and subbase.
However, technical information provided in the Austroads pavement design
guide (Austroads, 2004a) shows the concrete slab can freely curl during
daytime or nighttime temperature gradients. This suggests a fully unbonded
boundary condition (µ is zero) between the concrete slab and the subbase. It is
interesting to note that results of research carried out in the past on the
boundary condition between concrete slab and subbase have not led to a
specific conclusion. For instance, Yu et al. (1998) stated that friction between
the concrete slab and the subbase is sufficient to produce bonded behaviour
even if polyethylene sheets are placed between them. In contrast, Tarr et al.
(1999) indicated that unbonded boundary conditions could only be achieved
by using a double layer of polyethylene sheets between the concrete slab and
subbase.
• Flexural fatigue damage was assumed to only occur at the bottom surface
layer of the concrete base (see Part 2, Section 1 of Austroads, 2004b).
97
• Effects of varying axle group configurations on pavement’s response and
deterioration have not been addressed.
• The Guide in Section 9.2.1 provides maximum joint spacings for different
rigid pavements. Therefore, longer joint spacing may not be used even though
it may be desirable in some construction situations.
Not withstanding the above limitations, the Guide is the only design guide for concrete
road pavements in Australia and it remains to be the most widely used design method
among practitioners. Therefore, it is important that the Guide be systematically
explored to show its strengths as well as its limitations.
This research aims to investigate the applicability of the Guide in the design of typical
concrete pavements including JPCP, JRCP and CRCP. Results of a sensitivity analysis
performed to study the effects of input parameter variations on concrete base thickness
are presented.
To achieve these aims, the Austroads design procedure (2004) has been written in
Microsoft Visual Basic. It should be noted that the Guide employs traditional subgrade
drag theory for calculating the volume of reinforcement in the case of reinforced
pavements. However, in line with the common approach in concrete pavement design
(Yoder and Witczak, 1975), the effects of reinforcement on pavement behaviour and
slab thickness are not considered in this study.
5.3. Development and verification of ANRPD-2004 program

Using the equations provided in Section 3.6.1.2 of this thesis, a Windows-based
computer program named ANRPD-2004 was written in Visual Basic to analyse effects
of various design factors on base (slab) thickness.
The availability of the ANRPD-2004 computer program provided the author with an
opportunity to run a parametrical study efficiently. The program graphs the variation of
concrete slab thickness versus design traffic, or alternatively the required concrete
compressive strength for different operating conditions. These operating conditions
include pavement types, subgrade CBR, subbase types, PDR, concrete flexural
strength and provision of dowels and shoulders.
98
A graphical user interface was employed in ANRPD-2004 to assist with data entry and
interpretation. This program also has the ability to find the critical axle group for any
given traffic load spectra. An algorithm flowchart for the main core of the program is
shown in Figure 5-1.

Figure 5-1. Algorithm flowchart of program used in ANRPD-2004
In Figure 5-1, Na is number of axle groups participated in the traffic load distribution,
AGT (Na) is an array of axle group types, MAL (Na) is an array of maximum axle
load in axle group, VT is actual axle group type used in fatigue or erosion calculation,
BT is slab thickness (mm), FD is fatigue damage (%), Sum_FD is cumulative fatigue
damage (%) of all axle loads and axle groups, ED is erosion damage (%), Sum_ED is
cumulative erosion damage (%) of all axle loads and axle groups.
99
Results of the ANRPD-2004 program were verified according to the example provided
in Appendix 9.1, the PCP example design charts provided in Figure 9.2, and the
dowelled jointed or CRCP example design charts provided in Figure 9.3 of the Guide.

5.4.1. Effect of pavement types on thickness of concrete slab
As mentioned earlier, the method can be used to calculate the required concrete slab
thickness for JPCP, JRCP, CRCP and SFCP. Since the method does not differentiate
the pavement types except for joint spacing (see Section 9.2.1 of the Guide), the
calculated concrete slab thickness will be the same for different concrete pavements
provided that they have the same operating conditions. This can be also recognised by
investigating figure 9.3 of the Guide. This figure shows that the calculated concrete
slab thicknesses for dowelled jointed concrete pavements and CRCP are the same
while operating conditions are similar. Since CRCP has no transverse joints and
consequently does not need dowels, the calculated concrete slab thickness for CRCP is
expected to be less than those for jointed undowelled concrete slabs.
This was examined by the ANRPD-2004 program. Figure 5-2 shows the results of
slab (base) thicknesses calculated for different concrete pavements in the absence of
dowel when shoulder is available. Maximum design traffic, concrete compressive
strength, and load safety factor were considered to be 50×106 HVAGs, 50 MPa, and
1.35 respectively. The horizontal axis of this graph shows the variation of design
traffic from zero to 50×106 HVAGs. The vertical axis shows the concrete slab
thickness calculated without considering the minimum recommended thickness for
concrete slab. The slab thicknesses for JPCP, JRCP, and SFCP were found to be of the
same order. However, the use of CRCP results in a thinner concrete slab as expected.
5.4.2. Effect of concrete strength on damage mode

It has been well documented that materials with higher strength may be more sensitive
to fatigue damage than those with lower strength at their maximum applicable tensile
stress values (Croney and Croney, 1998). On the other hand, with an increase in
concrete compressive strength, the modulus of elasticity increases and consequently
the deflection under a certain load will decrease. As a result, it can be concluded that
an increase in concrete compressive strength may decrease the possibility of erosion
100
damage (due to lower deflection) but increase the possibility of fatigue damage in
concrete pavements. Furthermore, if the properties and thicknesses of subbase and
subgrade are kept constant, a concrete slab with a greater compressive stress would be
more sensitive to repeated load than that with a lower compressive strength for the
same ratio of applied stress to failure stress.
Figure 5-2. Variation of base thickness with design traffic (HVAGs) for different
pavement types (with shoulder and in the absence of dowels, subgrade CBR = 5%)
Using the program ANRPD-2004, a doweled JRCP without shoulder was subjected to
urban axle load spectra. The subgrade CBR and PDR were assumed to be 5% and
97.5%, respectively. Design traffic varied between 0 and 1×108 HVAGs. Figures 5-3
and 5-4 show results of calculated slab thickness obtained from fatigue and erosion
analyses for concrete compressive strengths of 36 MPa and 80 MPa, respectively. It is
interesting to see that these figures seem to suggest that an increase in concrete
compressive strength decreases the possibility of fatigue damage and consequently
erosion analysis results in a thicker concrete slab (base) than the fatigue analysis when
concrete compressive strength increases.
Similar results were observed when other types of the concrete pavements in the
absence of shoulder were studied. Note however that the provision of shoulder results
in comparable slab thickness for fatigue and erosion damage analyses.
101
Figure 5-3. Variation of base thickness with Design traffic (HVAGs) for dowelled
JRCP with concrete compressive strength of 36 MPa (subgrade CBR = 5%)
Further investigation of this matter reveals that the benefits offered by increasing
concrete compressive strength cease at a certain concrete compressive strength. In
other words, it appears that the benefits of using concrete of high compressive strength
are dependent upon the availability of dowels and shoulders, design traffic, and PDR
values. Table 5-1 shows the critical design traffic in fatigue analysis based on
availability of dowel and shoulder for different concrete compressive strengths. This
table is an example of the above mentioned finding and has been presented for a JRCP
with maximum design traffic of 1×108 HVAGs, subgrade CBR 3% and urban axle
groups load distribution. Design traffic of less than or equal to those values provided in
this table results in a thicker slab in fatigue analysis than erosion analysis. The use of
other concrete pavements may result in different outcomes.
The above mentioned outcome is perhaps less surprising when the fatigue and erosion
damage formulas of the Guide are examined. Undoubtedly, concrete flexural strength
has no effect on slab thickness resulting from erosion analysis (see Equations 3-22 and
3-27) whereas it has an inverse relationship with slab thickness from fatigue analysis
(see Equation 3-23). The latter equation indicates that an increase in concrete flexural
strength decreases the stress ratio (Sr).
102
Figure 5-4. Variation of base thickness with Design traffic (HVAGs) for dowelled
JRCP with concrete compressive strength of 80 MPa (subgrade CBR = 5%)
The stress ratio has a direct relationship with the equivalent stress which depends on
availability of shoulder, effective subgrade CBR (Ef), and slab thickness. Availability
of shoulder or an increase in slab thickness decreases the equivalent stress whereas an
increase in the Ef increases the equivalent stress. The allowable load repetition in
fatigue analysis (Nf) depends on the stress ratio so that a decrease in the value of Sr
toward 0.45 will lead to Nf becoming infinite. It means that the range of concrete
flexural strength variation, which can affect the slab thickness, strongly depends on the
convergence of the value of Sr toward 0.45 which is affected by the amount of
effective subgrade CBR, assumed slab thickness, and the availability of shoulder. An
increase in equivalent stress (Se) therefore may suggest the need for using concrete
with higher compressive strength.
Table 5-1. Design traffic (×106 HVAGs) below which fatigue analysis is the key factor
Provision of dowel and Concrete compressive strength (MPa)

shoulder 30 40 50 60 70 80 90 100
With Dowel and Shoulder 100 100 100 100 100 60 13 5
Just Dowel 100 100 52 21 10 7 4 3
Just Shoulder 100 31 7 3 1.2 0.8 0.4 0.4
Neither dowel nor shoulder 39 12 5 3 1.2 1 0.5 0.5
103
For instance, with a design traffic of 80×106 HVAGs, the concrete compressive
strength greater than those provided in the following sentences has no effect on the
calculated slab thickness in fatigue analysis:
• 78 MPa for pavements with dowels and shoulders.

• 45 MPa for pavements with dowels but no shoulders.
• 30 MPa for pavements with shoulders but no dowels (excluding CRCP because
it does not have transverse joints requiring dowels).
• 25 MPa in the absence of both dowel and shoulder (except for CRCP).
The above mentioned information can be extended for other values of design traffic.
Table 5-2 provides further information on this matter based on provision of dowel and
shoulder.
5.4.3. Effect of project design reliability (PDR) on slab thickness

The Guide suggests a range of PDR values from 80% to 97.5%. The choice of PDR
determines the load safety factor (LSF) value depending on type of concrete pavement.
LSF varies from 1.05 to 1.35 and can be derived from Table 9.2 of the Guide. It should
be noted that LSF remains the same for PDR of 80-85 per cent.
Results from the present study show that with an increase or decrease in PDR, the
calculated slab thickness increases or decreases by about 12 per cent respectively.
Figure 5-5 shows the effects of a change in PDR on slab thickness for a JPCP in the
absence of both dowels and shoulders.
Table 5-2. Concrete compressive strength (MPa) greater than those provided in the
table has no effect of the calculated base thickness in the fatigue analysis
Design Traffic Provision of Dowel and Shoulder

(× 106 HVAGs) Dowel + Dowel Shoulder Neither
≤ Shoulder
10 90 70 45 40
30 80 55 40 32
50 80 50 37 27
80 78 45 30 25
100 75 45 30 23
120 75 42 30 23
104
project design reliabilities (subgrade CBR = 5%)
5.4.4. Effect of subbase layer on slab thickness

Slab thickness is influenced by the subbase and subgrade strength. As mentioned
earlier in the literature, 150 mm LMC is the only choice for those projects with design
traffic greater than 1×107 HVAGs. Investigation of effects of different subbase types
on slab thickness for different design traffics indicates that the use of 125 mm LMC
always results in a thinner slab compared with the use of 150 mm bound or 170 mm
bound. Furthermore, the use of Austroads (2004b) design method via ANRPD-2005
results in two discontinuities which can be clearly observed at design traffic equal to 1,
and 10 million HVAGs. Figure 5-6 shows an example of this phenomenon for a JRCP
subjected to urban axle group load spectra in the absence of both dowel and shoulder.
A range of design traffic between zero and 12×106 HVAGs was considered to
effectively visualise the aforementioned phenomenon. Subgrade CBR was assumed to
be 5%. It should be noted that these discontinuities are always noticeable except when
the subgrade CBR is less than 2%.
5.4.5. Effect of subgrade CBR on slab thickness

As mentioned earlier, subgrade CBR is not directly used in the method. However, it
can be used for estimating the effective subgrade CBR (Fig. 3-36) using Figure 9.1 of
105
Austroads (2004b). In this figure, with subgrade CBR values ranging from 2 to 15 per
cent, the effective subgrade CBR varies between 5 per cent (when subgrade CBR is
less than 2 per cent) and 75 per cent, depending on subbase type.
subbase types in a JRCP
A comparison between the calculated slab thicknesses for a subgrade CBR of 2% with
those having a subgrade CBR of less than 2% shows that the assumption provided by
the Guide (effective subgrade strength is 5% when subgrade CBR is less than 2%)
increases the calculated base thickness by about 30% (Fig. 5-7). On the other hand,
subgrade CBR of more than 5 per cent has no effect on slab thickness when the design
traffic is greater than 1×107 HVAGs. This outcome is perhaps less surprising when the
design procedure of the Guide is examined. The only subbase choice when design
traffic is greater than 1×107 HVAGs (see Table 3-3) is 150 mm LMC. The use of 150
mm LMC subbase restricts the effective subgrade strength to a maximum of 75% for
subgrade CBR of 5% or more (see Figure 3-36). In other words, the use of subgrade
CBR of more than 5 per cent results in the same effective subgrade CBR i.e. the
maximum permitted value (75 per cent).
106
subgrade CBR in a JRCP (with shoulder and in the absence of dowels, subgrade CBR
= 5%, PDR = 97.5%, and concrete compressive strength = 50 MPa)
An increase in subgrade CBR, depending on provision of dowels and shoulders, has

different effects on slab thickness. For instance, in a JRCP with compressive strength
of 45 MPa and subgrade of CBR 3%, fatigue analysis always produces thicker slab
compared with erosion analysis when dowels are used, irrespective of the availability
of shoulder. However, increasing subgrade CBR from 3 % to 5 % results in thicker
slab in erosion analysis in the absence of shoulder when design traffic exceeds 63×106
HVAGs.
Table 5-3 presents the interrelationship between the critical damage process and the
operating conditions including subgrade CBR, concrete compressive strength and
provision of dowel and shoulder. Fatigue analysis results in greater slab thicknesses for
design traffic less than those provided in this table.
107
Table 5-3. Effects of subgrade CBR, concrete compressive strength, and provision of
dowels and shoulders on critical damage process in a JRCP (F = Fatigue, E = Erosion,
and Numbers are design traffic (×106 HVAGs))
Subgrade Provision of dowel Concrete compressive strength (MPa)

CBR (%) and shoulder 35 40 45 50 55 60 65 70 75 80
Dowel + Shoulder F F F F F F F F F 67
Dowel F F F 65 39 27 18 13 E E
<2
Shoulder 38 18 E E E E E E E E
Neither 17 E E E E E E E E E
Dowel F F F 62 37 25 17 12 E E
2
Shoulder 50 24 12 E E E E E E E
Neither 19 12 E E E E E E E E
Dowel F F F 58 35 23 16 E E E
2.5
Dowel F F F 52 33 21 15 E E E
3
Dowel F F F 56 31 20 14 E E E
3.5
Shoulder F 39 18 E E E E E E E
Dowel F F 85 42 28 19 13 E E E
4
Dowel F F 76 42 26 18 13 E E E
4.5
Dowel F F 63 39 22 17 E E E E
5
Shoulder F F 32 E E E E E E E
Design traffic less than those provided in this table will result in greater base thickness
in fatigue analysis.
108
When conventional concrete of strength ≤ 55 MPa is employed, a decrease in the
subgrade CBR value will normally result in:
• Fatigue becoming the controlling factor in the presence of both shoulder and
dowel.
• Fatigue becoming more critical than erosion in dowelled pavements and in the
absence of shoulder.
• Erosion becoming more critical than fatigue in the absence of dowel when
shoulder is available.
• Erosion becoming more critical than fatigue in the absence of both dowel and
shoulder.
5.4.6. Minimum recommended base thickness

Different values of minimum slab thickness have been recommended in Section 9.4.3
of the Guide depending on the type of concrete pavement and the magnitude of the
design traffic. The minimum recommended slab thickness (MRST) for JRCP is the
same as CRCP. To determine how the MRST affects the result of the method, different
concrete pavements with a concrete flexural strength of 4.5 MPa were investigated. A
range of design traffic from 0 to 100×106 HVAGs with different operating conditions
was considered. Results indicate that, for JPCP with dowels and shoulders, the MRST
governs when the magnitude of the design traffic exceeds 10×106 HVAGs.
Furthermore, the provision of dowel for design traffic of more than 50×106 HVAGs
results in a thinner concrete slab than the MRST for all types of concrete base.
However, the calculated base thickness governs in the absence of both dowels and
shoulders for JRCP or CRCP. Table 5-4 presents the results of this study.
5.4.7. Sensitivity of the fatigue and erosion analyses to a change in slab

thickness
The fatigue and erosion damage procedures for a given axle load provided in the Guide
are very sensitive to a small change in the slab thickness. The sensitivity is reduced
when the full load spectra (i.e. all axle group types) are considered. Table 5-5 presents
an example of this typical sensitivity analysis for fatigue damage in a JRCP subjected
to SADT with 80 KN axle load. Calculations are based on a design traffic of 50 million
109
HVAGs, urban axle load distribution, LSF of 1.25, effective subgrade CBR of 40%,
and concrete flexural strength of 4.5 MPa.
Table 5-4. When to calculate the base thickness instead of adopting the minimum
recommended base thickness
Neither
Dowel + Shoulder Dowel Shoulder
Subgrade PDR Dowel/Shoulder
CBR (%) (%)
JPCP JRCP/CRCP JPCP JRCP/CRCP JPCP JRCP/CRCP JPCP JRCP/CRCP
C when C when C when C when
C when C when DT
80,85 DT<10 or DT<10 or DT<10 or DT<50 or C C
DT<10 <50
32<DT<50 18<DT<50 21<DT<50 DT>88
C when C when C when
C when C when C when DT
90 DT<10 or DT<10 or DT<50 or C C
DT<10 DT <50 <50
20<DT<50 15<DT<50 DT>58
2
C when C when
95 DT<50 or DT<50 or C C C
DT<10 DT<50 DT <50
DT>66 DT>60
C when
C when C when
97.5 DT<10 or C C C C C
DT<50 DT <50
30<DT<50
C when C when
C when C when C when DT C when DT
80,85 DT<10 or DT<10 or C C
DT<10 DT<10 <50 <50
36<DT<50 33<DT<50
C when C when DT
90 DT<10 or DT<10 or DT<10 or DT<50 or C C
DT<10 <50
3 30<DT<50 18<DT<50 23<DT<50 DT<97
C when C when C when C when DT C when
95 C C C
DT<10 DT<50 DT <50 <50 DT <50
C when C when
97.5 DT<10 or DT<50 or DT<50 or C C C
DT<50 DT <50
45<DT<50 DT>69 DT>70
C
when
C when C when C when C when DT C when C when DT
80,85 DT<50 C
DT<10 DT<10 DT<10 <50 DT<10 <50
or
DT>66
C when C when C when DT
90 DT<10 or DT<10 or DT<10 or C C
4 DT<10 DT<50 <50
39<DT<50 37<DT<50 33<DT<50
C when C when
95 DT<10 or DT<50 or C C
DT<10 DT<50 DT <50 DT<50
16<DT<50 DT>68
C when
97.5 DT<50 or C C C
DT<10 DT<50 DT <50 DT <50
DT>92
C
when
C when C when C when C when DT C when C when DT
80,85 DT<50 C
DT<10 DT<10 DT<10 <50 DT<10 <50
or
DT>88
C
C when C when when
C when C when C when DT C when DT
90 DT<10 or DT<10 or DT<50 C
5 DT<10 DT<10 <50 <50
46<DT<50 42<DT<50 or
DT>62
C when C when DT
95 DT<10 or DT<10 or DT<10 or DT<50 or C C
DT<10 <50
16<DT<50 18<DT<50 20<DT<50 DT>92
C when
C when C when C when C when C when
97.5 DT<50 or C C
DT<10 DT<50 DT <50 DT<50 DT <50
DT>65
C: Calculate slab thickness DT: Design traffic (×106 HVAGs)
If design traffic is out of the range provided in this table, the MRST governs.
110
Table 5-5. Effect of base thickness on fatigue damage in a JRCP with shoulder (SADT,
axle load 80 KN, design traffic 50×106 HVAGs, urban axle load distribution, LSF =
1.25, Ef = 40%, and concrete flexural strength 4.5 MPa)
Allowable No of Fatigue
Thickness Se repetition repetition damage
(mm) (Million) Sr (Million) (Million) (%) Explanation
160.5 1.656 0.4804 2.34 2.31 98.6 O.K
160.4 1.658 0.4808 2.28 2.31 101.2 Damaged
5.4.8. Damage process

Results of the current study indicate that the variation of slab thickness with design
traffic for fatigue and erosion damage is complex and depends on the provision of
dowel and shoulder, concrete flexural strength, and subgrade CBR. Hence, specifying
the damage process as being either fatigue related or erosion related is not as simple as
that indicated in Clause 6.5 of Austroads (2004a) for JPCP (see Figure 5-8). Since the
Austroads method does not differentiate between concrete pavements in calculating the
slab thickness, the following observations may be noted for all concrete pavements:
• Fatigue is critical for any subgrade CBR value when dowel and shoulder are used
and the concrete compressive strength is ≤ 75 MPa.
• Fatigue is critical for any subgrade CBR value in the absence of shoulder for
dowelled pavements with concrete compressive strength is ≤ 40 MPa.
• Erosion is critical for any subgrade CBR value in the absence of dowel when
shoulder is used and the concrete compressive strength is ≥ 50 MPa.
• Erosion is critical for any subgrade CBR value in the absence of both dowel and
shoulder when the concrete compressive strength is ≥ 45 MPa.
Table 5-3 shows the influence of subgrade CBR, concrete compressive strength, and
provision of dowels and shoulders on the critical damage. For design traffic lower than
those provided in this table fatigue is the controlling factor.
111
Figure 5-8. Typical rigid pavement thickness design curve for a specific effective CBR
and concrete flexural strength (Austroads, 2004a)
5.5. Summary
A sensitivity analysis on Austroads concrete pavement design guide (2004a) was
carried out and the interrelationships between design parameters and calculated slab
thicknesses were described. Results of the current study indicate that:
• The Austroads method (2004) results in the same slab thickness for different
concrete pavements under the same operating conditions.
• In contrast with what is commonly accepted as the fundamental concrete

• The use of 125 mm LMC subbase always results in a thinner slab compared
with the use of 150 mm bound or 170 mm bound subbase. When design traffic
is also in the proximity of 1×106 or 10×106 HVAGs, lowering design traffic
results in thicker concrete slab.
• Increasing the subgrade CBR above 5 per cent has no effect on slab thickness
• In most cases, the minimum recommended slab thickness governs the design
provided that the pavement is dowelled and restrained by shoulder or adjacent
traffic lanes.
112
• The Austroads method is very sensitive to a change in the slab thickness.
• Critical damage type is not only a function of design traffic but it is also a
function of operating conditions such as subgrade CBR, PDR, and provision of
dowels and shoulders.
113
Chapter 6
STATIC ANALYSIS OF JOINTED PLAIN CONCRETE

PAVEMENT
6.1. Introduction
Using a sensitivity analysis, effects of concrete strength, project design reliability,
subbase layer, subgrade CBR and provision of shoulder and dowel bars on thickness of
concrete slab were studied in Chapter 5 of this thesis. However, structural response of
concrete pavements is influenced by a number of factors described in Chapters 3 and 4.
The most significant factors affecting concrete pavement performance depend on
vehicular characteristics and environmental effects.
Axle groups and loading configurations vary among heavy vehicle manufacturers and
across countries. Hence, it is not surprising to know that research on concrete
pavements carried out worldwide has been based on a variety of vehicular
configurations. Configurations, magnitudes and locations of axle groups upon the
pavement are the most significant vehicular related factors. If critical load
configurations together with critical positions of axle groups upon the pavement are
not considered in concrete pavement analysis, the design may be inadequate and lead
to early failure of the pavement.
Since differential temperatures between the top and the bottom surface layers of the
concrete slab result in upward or downward curling induced stresses, critical positions
of axle groups may be affected by temperature fluctuation within the depth of the
concrete slab. The magnitude of curling induced stresses, on the other hand, depends
on boundary conditions between concrete the slab and subbase.
Whilst there has been a great deal of research conducted on concrete pavement
performance and deterioration under vehicular loads and environmental effects, there
is a lack of adequate information on effects of vehicular load positions on pavement
responses. Furthermore, no critical dimensions of axle group configurations have been
yet presented. These shortcomings are addressed in this chapter using parametrical
114
studies. Since parametrical nonlinear dynamic studies of all vehicular related
parameters are complex and highly time dependent, effects of these factors on
structural response of JPCP are investigated using a series of static analyses. For this
purpose, EverFE 3D finite element program (Davids and Mahoney, 1999) is
employed.
To study effects of different parameters on the structural responses of concrete

pavement, diverse pavement configurations are considered. Information on the finite
element models is presented in the next section (6.2). Effects of provision of different
types of debonding layer on pavement response are then presented. Subsequently, axle
group configurations are revisited to determine the critical axle group configuration for
each individual axle group. Finally, combinations of vehicular loads and
environmental effects are considered to determine the critical location of axle groups
upon the concrete pavement. These studies are carried out based on different
methodologies. Theses methodologies are presented separately in each section.
6.2. Finite element models

Different finite element models of the JPCP are used in this chapter. These are single
slab, single traffic lane, a traffic lane confined at one of its longitudinal edges by
shoulder, a traffic lane confined at both longitudinal edges by shoulders and single lane
confined by adjacent traffic lanes. These models are called respectively single slab,
single lane, a confined lane, a double confined lane and full pavement (Fig. 6-1).
The distance between transverse joints and the distance between longitudinal joints
(Slab length and width) were considered to be 4600 mm and 3600 mm respectively.
Tied shoulders with 1500 mm width were considered. The slab thickness was
considered to be 250 mm with modulus of elasticity and Poisson’s ratio of 28000 MPa
and 0.2 respectively. The concrete coefficient of thermal expansion was considered to
be 1×10-5 mm/mm/˚C. A cement stabilized subbase of 150 mm thickness, 5000 MPa
modulus of elasticity, and 0.2 Poisson’s ratio was considered beneath the slab and over
a subgrade with modulus of subgrade reaction of 0.03 MPa / mm (CBR ≈ 3.5).
Transverse joints were dowelled by eleven evenly spaced cylindrical bars 32 mm in
diameter, 450 mm long and 1000 MPa dowel-slab support modulus. Tie bars 13 mm in
diameter and 1000 mm length spaced at 1000 mm centre to centre were considered at
115
longitudinal joints. The above mentioned values of the parameters are within the range
recommended by Austroads (2004).
Figure 6-1. JPCP configurations considered in this Chapter
6.3. Effects of debonding layer on concrete pavement responses

As mentioned earlier in Chapter 2, the main aim for placing a debonding layer between
the concrete slab and the subbase is to reduce the early age cracking. However,
provision of this layer between the concrete slab and the subbase affects the
performance of concrete pavements under the applied loads during the pavement life.
Characteristics of materials used in concrete pavement technology as debonding layer
were published by Stott (1961), Wesevich et al. (1987), Wimsatt et al. (1987). These
characteristics were then used by Rozycki and Rasmusen (1998) and Zhang and LI
(2001) to study shrinkage effects on concrete pavement responses. In theses studies,
concrete slab was resting on different subbase including Cement stabilised (CS),
Granular (GR), Asphalt Cement (AC), Asphalt Stabilised (AS), Lime treated clay (LC)
and Natural clay (NC).
Pavement construction in Australia, however, is based on multi layers action of the

pavement to the applied load. In this case, the debonding layer is one of the pavement
layers that is placed between concrete slabs and subbase to prevent early age cracking.
116
As a result, the outcomes of the previous studies can not be accurately adopted in
Australian concrete pavement thickness design guide.
Moreover, information on effects of the debonding layer on concrete pavement

behaviour does not lead to a specific conclusion. For instance, Tarr et al. (1999)
indicated that unbonded condition could only be achieved by using a double layer of
polyethylene sheets. On the other hand, Yu et al. (1998) stated that friction between
concrete slab and subbase is sufficient to produce bonded behaviour even if
polyethylene sheets are placed between them.
Since concrete pavement research conducted in the past was based on either fully
bonded or fully unbonded boundary condition between concrete slab and subbase
(Heath and Roesler, 1999), it is important to examine how different debonding layer
affect concrete pavement responses to the applied loads. This is addressed in this
section.
6.3.1. Methodology
Concrete pavement behaviour is affected by a combination of vehicular loads and
environmental effects. The concrete slab curls under differential temperatures or warps
under loss of moisture contents. A separation between concrete slab and subbase may
occur due to pavement curvature if debonding layer is placed between concrete slab
and subbase. The vehicular loads may magnify or diminish induced curling stresses in
concrete pavement depending on location of vehicular loads. Since this section aims to
provide information on how different materials as debonding layer affect concrete
pavement behaviour, concrete pavement is only subjected to diverse differential
temperatures. Information on combined action of vehicular loads and differential
temperature is discussed later in Section 6.5.
To examine the effects of different debonding layers on concrete pavement response, a

full pavement configuration was modelled. Different debonding layers between
concrete slab and subbase were considered. These include fully bonded, AC, AS, CE,
CL, Gr and fully unbonded boundary conditions. Since the concept of frictional stress
was adopted in EverFE finite element program for simulation of debonding layer,
characteristics of different debonding layers were derived from Figure 3-9 and used in
the finite element analyses. The pavement was subjected to daytime and nighttime
117
differential temperatures between -20˚C and +20˚C with an interval of 5˚C. The
shrinkage effect was incorporated into the aforementioned temperature range. Further
information on temperature variation due to shrinkage was previously discussed in
Section 3.3.3.
6.3.2. Results and Discussion

A comparison between thermal induced stresses in bonded pavement with those from
unbonded pavement shows that unbonded boundary condition decreases the induced
tensile stress when lower differential temperature is considered. In other words, the
benefits offered by a consideration of the unbonded boundary condition ceases at a
certain value of differential temperature (Fig. 6-2). In this figure, negative values of
differential temperature represent nighttime differential temperatures and positive
values represent daytime differential temperatures.
This finding suggests that with an increase in differential temperature, shear stress
between slab and subbase increases until a full separation between concrete slab and
subbase occurs. In this condition, the contact area between concrete slab and subbase
rapidly decreases and consequently a significant bending stress is produced at the edge
of the contact area.
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
-20 -15 -10 -5 0 5 10 15 20
Bonded AC AS CE CL GR Unbonded
Figure 6-2. Thermal induced tensile stress in concrete slab for different debonding
materials
118
It was observed that the induced tensile stresses occur at interface of the concrete slab
and subbase before separation, and at the top or the bottom surface layer of the
concrete slab after separation. As a result, top-down corner cracking in upward curling
or warping and bottom-up mid slab transverse cracking in downward curling can be
addressed by consideration of thermal induced stresses.
The above mentioned materials, as debonding layer between concrete slab and
subbase, act similar to fully unbonded boundary conditions if pavement is subjected to
daytime differential temperatures (Fig. 6-2). During nighttime, however, provision of
cement stabilised layer (placing the concrete slab directly over the subbase without
provision of debonding layer) acts similarly to bonded boundary condition. Behaviour
of the concrete pavement with other materials as debonding layer is similar to
behaviour of fully unbonded pavement during nighttime. Nevertheless, lower
differential temperature than the critical differential temperature of fully unbonded
boundary condition, i.e. -15˚C instead of -10˚C, is required to separate the concrete
slab from the subbase.
To determine how pavement configuration affects the above mentioned result, a single
lane, a confined lane and a double confined lane were modelled. Fully bonded and
fully unbonded boundary conditions between concrete slab and subbase were also
considered. In addition, variation of differential temperatures was gradually increased
at the critical differential temperature for better understanding of the above mention
phenomenon.
A comparison of results between different unbonded concrete pavement configurations

indicates that the pavement configurations do not significantly affect the pavement
response to daytime differential temperatures (Fig. 6-3). However, single lane and full
pavement configurations show a similar behaviour during nighttime. Confined lane
and double confined lane configurations also show a similar behaviour during
nighttime. Furthermore, separation between concrete slab and subbase in confined lane
and double confined lane configurations occurs in lower differential temperature than
single lane and full pavement configurations, i.e., -15.5˚C instead of -14.6 ˚C
respectively.
119
The critical daytime differential temperature is 16.4 ˚C for all pavement
configurations. However, the separation in single lane or full pavement configuration
occurs at -14.6˚C during nighttime. The critical nighttime differential temperature in a
confined lane and double confined lane is -15.5˚C. Information provided in Figure 6-4
also indicates that the stress distribution in the concrete slab linearly changes with
variations in differential temperatures in the presence of a fully bonded boundary
condition between the concrete slab and the subbase. Similar result can be obtained in
fully unbonded pavements subjected to a differential temperature of greater than
16.4˚C during daytime or lower than -14.6˚C during nighttime (Fig. 6-3). This
suggests that accurate results can be expected when thermal induced stress at a certain
location is superimposed to vehicular induced stress at the same location. However, in
the presence of fully unbonded boundary condition between the concrete slab and the
subbase the above mentioned consideration leads to inaccurate results as the stress
distribution in this temperature range (between -14.6˚C and 16.4˚C) seems to be
nonlinear.
5.5
Unbonded Boundary Single Lane
Maximum Thermal tensile stress (MPa)
5
Condition between Concrete Confined lane
Slab and Subbase 4.5
Double Confined lane
4
Full Pavement
3.5
2.5
1.5
0.5
0
-25 -20 -15 -10 -5 0 5 10 15 20 25
Differential temperature between top and bottom surface layers of the concrete
slab (ºC)
Figure 6-3. Comparison between thermal induced tensile stresses in different fully
unbonded pavement configurations
A comparison between the induced tensile stresses in fully bonded pavement with
different configurations indicates that the induced tensile stresses among all pavement
configurations are compatible during nighttime (Fig. 6-4). Nevertheless, full pavement
configuration results in greater induced tensile stress during daytime than a double
120
confined lane, a confined lane and single lane configurations by about 5.3%, 8.6% and
14.6% respectively.
6.4. Revisiting axle group configurations

6.4.1. Methodology
Types of axle group, axle group loads, distance between axles in a given axle group,
load distribution between axles, axle width, distance between centres of dual tyres, tyre
inflation pressure, tyre-pavement contact stress, and tyre-pavement contact shape are
the significant parameters in axle group configurations. This section aims to provide
the critical dimension of each individual parameter of axle groups based on maximum
induced tensile stress and deflection in the concrete slab.
Critical dimensions of axle group configurations were assumed to be independent of

pavement configurations and boundary condition between concrete slab and subbase.
As a result, a single bonded concrete slab is modelled to simplify the analysis
procedure. For further simplification, pavement dimensions and characteristics of
concrete, subbase and subgrade are assumed to be constant. While the effect of a
parameter of axle group configuration on pavement response is studied, all other
parameters of axle group configurations are considered to be constant.
4
Bonded Boundary Condition Single Lane
between Concrete Slab and
Maximum Thermal tensile stress (MPa)
3.5 Confined lane

Subbase
Double Confined lane
3
Full Pavement
2.5
1.5
0.5
0
-25 -20 -15 -10 -5 0 5 10 15 20 25
Differential temperature between top and bottom surface layers of the
concrete slab (ºC)
Figure 6-4. Comparison between thermal induced tensile stresses in different fully
bonded pavement configurations
121
To validate the accuracy of the above mentioned assumptions in concrete pavement
analysis, a confined lane configuration will be subjected to SADT, TAST, TADT and
TRDT with different dimensions of axle configurations. One of axle configurations is
based on results of the current study and the second one is based on dimensions used
by Packard and Tayabji (1985). Both bonded and unbonded pavements are considered.
A rectangular tyre-pavement contact shape with uniform stress distribution is
considered based on the results of Gillespie et al. (1992). A fully bonded boundary
condition between concrete slab and subbase and a tensionless property for subgrade
are also considered. No further boundary condition was applied on the pavement.
Different axle groups based on Austroads (2004) were considered in the current study
(see Figure 3-21). Axle group loads were applied at different locations of the
pavement. These include centre loading (axle group is symmetrically positioned at the
centre of the concrete slab), mid-edge loading (axle group is positioned at the middle
of longitudinal edge of the pavement), and corner loading (axle group is positioned at
the corner of the pavement). Figure 6-5 shows the location of the applied loads. As
structural response of pavement under different axle group loadings were found to be
similar, only results for mid-edge loading are presented in the following sections.
Figure 6-5. Load positions on the concrete pavement
Results of this parametrical study can be employed on those projects where sufficient
statistical information on configurations of different axle groups is not available.
6.4.2. Width to length ratio of the tyre-pavement contact area

Effects of SAST and SADT with respectively 53 kN and 80 kN axle load on pavement
response were separately studied. The SADT was used to consider the effect of
adjacent wheel loads on pavement response due to a change in dimensions of tyre
122
imprint. In accordance with Austroads (2004), the distance between the centres of dual
wheels and the axle width were considered to be 330 mm and 1800 mm, respectively
(see Figure 3-22).
Tyre inflation pressure was considered to be constant and equal to 750 kPa during this
investigation. Tyre contact width and length can be calculated from Equations 6-1a
and b.
A t (6-1a)
L t =
α
Wt = α .Lt (6-1b)
Where Lt and Wt are tyre-pavement contact length and width (Fig. 3-23), At is tyre-
pavement contact area calculated from Equation 3-13, α is width to length ratio of
tyre-pavement imprint.
According to the past studies mentioned earlier in Chapter 2, the α values range from
0.7 to 1.14. However, as the distance between interior edges of dual tyre is about
150mm, the maximum width of tyre-pavement contact area for one tyre is equal to the
distance between the centres of dual tyres i.e between 300 and 330 mm. Hence, the
maximum value of the α for one tyre can be as high as 3, for a tyre pressure of 550
kPa. Higher value of the α may be also related to the use of wide-base tyres in heavy
vehicles. On the other hand, based on the work of Douglas et al. (2000) the minimum
value of the α is 0.3 when higher tyre inflation pressures are used. A variety of the α
values between 0.3 and 3 were therefore selected in the current research to study
effects of tyre imprint dimensions on concrete pavement response. Table 6-1 shows
relevant values of area, width and length of tyre-pavement contact for SAST and
SADT.
Figure 6-6 shows the variations of tensile stress and slab deflection versus the α ratio
when SAST and SADT are separately applied on the pavement. An increase in the
α value from 0.3 to 0.5 very slightly increases the value of tensile stress due to SAST
and decreases the induced tensile stress of SADT by about 2 percent. Tensile stress in
both cases remains constant when the α value increases from 0.5 to 0.7. Further
increase in the α value decreases induced tensile stresses due to SAST and SADT.
123
Table 6-1. Information on tyre-pavement contact area in SAST and SADT
SAST SADT
α Width Length Area Width Length Area
(mm) (mm) (mm2) (mm) (mm) (mm2)
0.3 103 343 35329 89 298 26522
0.5 133 266 35378 116 231 26796
0.7 158 225 35550 137 195 26715
0.8 168 210 35280 146 183 26718
0.9 178 198 35244 155 172 26660
1 188 188 35344 163 163 26569
1.1 197 179 35263 172 156 26832
1.2 206 172 35432 179 149 26671
1.5 230 153 35190 200 133 26600
1.9 258 136 35088 224 118 26432
3 327 109 35643 282 94 26508
0.96 0.73
0.91
Maximum base deflection (mm)

Maximum tensile stress (MPa)
0.68
0.86
0.63
0.81
0.58
0.76
0.71 0.53
0.3 0.5 0.7 0.8 0.9 1 1.1 1.2 1.5 1.9 3
With to length ratio of tyre contact area

Tensile stress in SAST Tensile stress in SADT Deflection in SAST Deflection in SADT
Figure 6-6. Effect of tyre-pavement contact area on pavement response
Slab deflection, on the other hand, is less sensitive to a change in the α value though it
decreases with an increase in the α value. Results show that a lower width to length
124
ratio of the tyre-pavement contact area results in greater maximum tensile stress and
deflection. Hence, if the magnitude of the axle load was assumed to be constant during
the past research, the contact area used by Packard and Tayabji (1985) always
produces greater values of both tensile stress and deflection in pavement than those of
Kim et al. (2002) and Gillespie et al. (1992).
6.4.3. Tyre inflation pressure

Centre, mid-edge, and corner loadings were considered in order to study the effect of
tyre inflation pressure on the concrete pavement responses. A 53 kN SAST was used
in this study. Diverse values of tyre inflation pressures between 500 kPa to 1400 kPa
were considered. The tyre-pavement contact area was calculated from Equation 3-13.
For practical reasons, α was assumed to be 0.7 and the width and length of tyre contact
patch were calculated using Equation 6-1.
Figure 6-7 shows variations of maximum tensile stress and concrete slab deflection
versus different tyre inflation pressures. The results show that an increase in tyre
pressure has greater effects on the tensile stress than deflection of the concrete slab. In
addition, results indicate that an increase in tyre inflation pressure results in greater
induced tensile stress due to corner loading than that due to mid-edge loading.
0.93 1.18
0.88 1.093

0.83 1.006
0.78 0.919
0.73 0.832
0.68 0.745
0.63 0.658
0.58 0.571
0.53 0.484
0.48 0.397
0.43 0.31
500 600 700 800 900 1000 1100 1200 1300 1400
Tyre inflation pressure (kPa)
Tensile stress _Centre loading Tensile stress _Mid-edge loading Tensile stress _Corner loading
Base Deflection _Centre loading Base Deflection _Mid-edge loading Base Deflection _Corner loading
Figure 6-7. Effect of tyre inflation pressure on pavement response
125
This suggests that critical loading position in terms of tensile stress depends on tyre
inflation pressure and consequently tyre-pavement contact area. Corner loading is
more sensitive to a change of tyre-pavement contact area than mid-edge loading due to
lower stiffness (load was positioned close to both free longitudinal and transverse
edges).
Okomato and Packard (1989) in an investigation of the effect of tyre pressure on

pavement response indicated that tyre pressure has no strong effect on pavement
response. However, the current study shows that deterioration at the corner of concrete
pavements and especially in plain concrete pavement strongly depends on tyre
inflation pressure. However, the use of dowels and tie bars at transverse and
longitudinal joints can alleviate the above mentioned effect. As the state of the art in
tyre technology is such that there are more tyres with very high inflation pressure on
the roads, ignoring this fact may cause early deterioration of the pavement particularly
at the corner locations.
6.4.4. Distance between the centres of dual tyres

To investigate how a change in distance between the centres of dual tyres affects the
pavement response, a SADT with 80 kN axle load was considered. Based on the
literature review, a range of 300 mm to 340 mm was chosen for distance between the
centres of dual tyres. Axle width, tyre inflation pressure, and width to length ratio of
tyre imprint were respectively considered to be 1800 mm, 750 kPa, and 0.8.
Figure 6-8 shows the variations of maximum tensile stress and concrete slab
deflection of the pavement with the varying distances between the centres of dual
tyres. Results show that an increase in the distance between dual tyres from 300 mm to
340 mm decreases the tensile stress and slab deflection by 2.5% and 1.6% respectively.
Hence, a shorter distance between the centres of dual tyres can create a more critical
situation as expected.
126
0.805 0.68
0.8 0.677
Tensile Stress
Deflection
Base Deflection (mm)

Tensile stress (MPa)
0.795 0.674
0.79 0.671
0.785 0.668
0.78 0.665
300 305 310 315 320 325 330 335 340
Distance between the centres of dual wheels (mm)
Figure 6-8. Effect of distance between the centres of dual tyres on pavement response
6.4.5. Axle width

A SADT with 300 mm distance between the centres of dual tyres and 750 kPa tyre
inflation pressure was applied on the pavement. The value of width to length ratio of
tyre-pavement contact area and axle load were assumed to be 0.8 and 80 kN,
respectively. Different axle widths, from 1800 mm to 2150 mm, were considered.
Results (Fig. 6-9) show that an increase in distance between axles very slightly
decreases the magnitude of induced tensile stress and deflection in the slab.
Nevertheless, an increase in axle width beyond 1890 mm increases the rate of induced
stress reduction in concrete pavement. This also is true for slab deflection. The reason
behind this is interaction between wheel loads at opposite sides of the slab. While the
wheel load located close to the centreline of the concrete slab results in upward
curvature, the edge loading results in downward curvature of the slab.
6.4.6. Axle spacing in a given axle group

Different types of axle groups including TAST, TADT, and TRDT were considered to
investigate effects of a change in distance between axles on concrete pavement
response. Tyre inflation pressure, width to length ratio of tyre-pavement contact area,
distance between centres of dual tyres, and axle width were considered to be 750 kPa,
127
0.8, 300 mm, and 1800 mm respectively. Axle load was considered to be 90 kN, 135
kN, and 181 kN for TAST, TADT, and TRDT, respectively. Tyre-pavement contact
area and consequently width and length of this area were calculated based on
Equations 3-12 and 6-1. The axle spacing was varied between 1000 mm to 1600 mm.
0.86 0.71
0.855 0.7
Tensile Stress
Base deflection (mm)

Deflection
0.85 0.69
0.845 0.68
0.84 0.67
0.835 0.66
1800 1820 1850 1870 1890 2000 2050 2100 2134 2150
Axle width (mm)
Figure 6-9. Effect of axle width on pavement response
Results (Fig. 6-10) show that pavement response is dependent on the axle spacing in
the axle group. An increase in axle spacing linearly decreases the maximum slab
deflection so that shorter axle spacing will result in greater slab defection.
The variation of tensile stress versus axle spacing for each axle group deserves special
attention. An increase in axle spacing in a given axle group from 1000 mm to 1200
mm decreases induced tensile stress by 18.2%, 20%, and 30% for TADT, TAST and
TRDT respectively. Consideration of longer distance up to 1300 mm for this parameter
has no effect on induced tensile stress due to TAST. Nevertheless, this increment
decreases the induced tensile stress by about 5% for both TADT and TRDT. After this
point and up to 1400 mm, the tensile stress increases by about 2.5% for both TADT
and TAST. In contrast, a decrease by 4.7% can be observed in TRDT induced tensile
stress.
128
0.9 1.2
1.15
0.8 1.1

1.05
0.7
1
0.95
0.6
0.9
0.85
0.5
0.8
0.4 0.75
0.7
0.3 0.65
1000 1100 1200 1300 1400 1500 1600
Distance between axles (mm)

Tensile stress TAST Tensile stress TADT Tensile stress TRDT
Base deflection TAST Base deflection TADT Base deflection TRDT
Figure 6-10. Effect of axle spacing on pavement response
Further increase in axle spacing to 1500 mm has no significant effect on tensile stress
due to TADT and TAST. On the other hand, for the same increase in axle spacing, the
tensile stress decreases by 25% under TRDT. Axle spacing of more than 1500 mm has
no effect on tensile stress due to TRDT but decreases the tensile stress by 2.8% when
TADT and TAST are considered.
Results of current study show that concrete pavement deflection is a function of total
gross weight of axle groups. In other words, the critical slab deflection occurs when the
heaviest axle group (in this study TRDT) passes through pavement. Packard and
Tayabji (1985) also stated that TRDT results in the most critical values of erosion
damage compared to other axle groups. AS mentioned in chapter 3, the magnitude of
tensile stress in the pavements relies on the tyre pavement contact stress. However,
results of the current study indicate that the induced tensile stress is also affected by the
distance between axles and availability of dual tyres. Hence, TADT produces greater
tensile stresses in the pavement than TAST although the magnitude of tyre pavement
contact stress in TAST is higher than that in TADT.
While axle spacing in a given axle group may legally vary between 1000 mm and 1600
mm (RTA, 1998), results of current research indicate that with a distance of 1220 mm
to 1372 mm (similar to those used by researchers all over the world) lower values of
129
tensile stress will be produced. Moreover, the results indicate that the critical distance
between axles in a given axle group is between 1050 mm and 1150 mm with average
of 1100 mm for all axle groups. In addition, this value can also be between 1350 mm
to 1450 mm with an average of 1400 mm for TAST and TADT groups. A low value
(1100 mm), however, will result in higher tensile stress and slab deflection.
6.4.7. Load shift between axles in a given axle group

In pavement analysis, axle group load is often considered to be equally distributed
between axles. Hence, each axle carries 50%, 33.33% or 25% of total axle group load
in a tandem, triple or quad axle group. However, the suspension systems together with
pavement roughness may affect the load distributions between axles in a given axle
group as mentioned earlier in the Chapter 3. Higher load shift is expected in
mechanical suspension system as shown by Blanksby et al. (2006) in an in-service
survey; the load shift between axles was about 20-40%.
To study the effect of a shift in axle loads within a given axle group, a TAST was
considered. Axle width, tyre inflation pressure, and axle load were considered to be
1800 mm, 750 kPa, and 90 kN respectively. The characteristics of tyre contact patch
were again calculated by using Equations 3-12 and 6-1 similar to the previous stages
where the value of width to length ratio of tyre-pavement contact area was assumed to
be 0.8. In this study, the load shift between axles was considered to be between 0%
(each axle carries 50% of axle group load) and 16% (one axle carries 34% of axle
group load and the other one carries 66% of axle group load).
Results of this study (Fig. 6-11) show that a shift in load distribution between axles has
no significant effect on slab deflection. However, vehicular induced tensile stresses are
affected by a change in proportion of load distribution between axles. A linear
relationship between load shift and maximum induced tensile stress can be observed.
Consequently, for a load shift of 1% the maximum tensile stress is increased by about
0.88 percent.
Since deflection of concrete slab is affected by the total applied load in a given axle
group, a change in load shift between axles up to 4.4 percent has no effect on
maximum slab deflection. Further increase in load shift between axles changes the slab
130
deflection only slightly. However, the load shift between axles in a given axle group
starts to affect the slab deflection continuously after its value exceeds 8.9 percent.
0.72 0.76
0.7
0.758
Base Deflection (mm)

0.68
0.756
0.66
0.754
0.64
Stress
Deflection
0.752
0.62
0.6 0.75
0 2 4.4 6.7 8.9 12 13.3 15.6
Load shifts between axles (%)
Figure 6-11. Effect of load shift between axles on pavement response for TAST
Induced tensile stress, on the other hand, is affected by tyre-pavement contact area
which is affected by the magnitude of wheel load. Wheel load is affected by the load
shift between axles. With regards to the significant effect of axle load shift on the
variation of tensile stress, it is recommended that concrete pavements are designed
using accurate information on wheel loads (as opposed to axle loads) to provide
adequate safety factor.
6.4.8. Result Validation

Since concrete pavement behaviour under vehicular loads is affected by pavement
configuration (availability of adjacent concrete slab in longitudinal direction and
shoulder in transverse direction), validation of the critical axle group configurations is
significant. For this purpose, a confined lane was modelled. Interaction between
concrete slab and subbase was considered as either bonded or unbonded. SADT,
TAST, TADT and TRDT were separately applied at the middle of longitudinal free
edge of the centre concrete slab. Two different dimensions of a given axle group
configuration were considered. One of these configurations was based on the results of
131
the current study and the other one was based on dimensions used by Packard and
Tayabji (1985).
A comparison between results shows that the use of dimensions provided in the current
study increases the magnitudes of vehicular induced tensile stresses by about 5.6%,
3.2%, 10.2% and 16% in bonded pavement and 5.6%, 5.1%, 12.3% and 17.5% in
unbonded pavement for SADT, TAST, TASDT and TRDT respectively (Fig. 6-12). It
was observed that heavier axle groups produce a considerable increase in induced
tensile stress. This results in ticker concrete slab as it creates more fatigue damage in
the concrete pavement than that expected by Packard and Tayabji (1985).
As mentioned earlier, slab deflection is highly affected by the total load of a given axle
group. Nevertheless, the use of dimensions provided in the current study also produces
greater slab deflection so that the slab deflection increases by about 8.9%, 3.3%, 4.2%
and 5.6% in bonded pavement and by 7.6%, 6.57%, 5.2% and 6.72% in unbonded
pavement for SADT, TAST, TADT and TRDT respectively (Fig. 6-13).
6.4.9. Variation in axle group loads

Results of the previous sections were employed to determine how induced tensile
stresses in concrete pavements are affected by variations in axle group loads. Hence, a
confined lane was subjected to diverse axle group configurations including SAST,
SADT, TAST, TADT, TRDT and QADT. Vehicular loads were applied at mid-length
of the free edge of the pavement. The magnitude of each axle group load was
considered to be at either the average value or the maximum load considered in
Austroads 2004.
The maximum load in each individual axle group can be derived from Tables 3-4 and
3-5. Fully bonded and fully unbonded boundary conditions between the concrete slab
and the subbase were also considered in this study. Results show that the provision of
unbonded boundary condition between concrete slab and subbase results in greater
tensile stress in concrete slab than bonded boundary condition (Table 6-2).
Furthermore, there is a linear relationship between induced stresses in bonded
pavement with those in unbonded concrete pavement. In other words, the induced
tensile stresses in unbonded concrete pavement can be predicted from corresponding
stresses in bonded pavement using factor, C1, provided in the last column of Table 6-2.
132
1.7
Tensile Stress (MPa)

1.5
1.3
1.1
0.9
0.7
0.5
Pakard and
Pakard and
Pakard and
Pakard and
Dimensions
Dimensions
Dimensions
Dimensions
provided in
provided in
provided in
provided in
the thesis
the thesis
the thesis
the thesis
Tayabji
Tayabji
Tayabji
Tayabji
(1985)
(1985)
(1985)
(1985)
SADT TAST TADT TRDT
Axle Grup Type
Bonded Unbonded
Figure 6-12. Comparison in induced tensile stresses for bonded and unbonded concrete
pavements based on results of the current study and dimensions used by Packard and
Tayabji (1985)
1.2
Slab Deflection (mm)
1.1
1
0.9
0.8
0.7
0.6
0.5
Pakard and
Pakard and
Pakard and
Pakard and
Dimensions
Dimensions
Dimensions
Dimensions
provided in
provided in
provided in
provided in
the thesis
the thesis
the thesis
the thesis
Tayabji
Tayabji
Tayabji
Tayabji
(1985)
(1985)
(1985)
(1985)
SADT TAST TADT TRDT

Axle Group Load
Bonded Unbonded
Figure 6-13. Comparison in deflection of concrete slab for bonded and unbonded
concrete pavements based on results of the current study and dimensions used by
Packard and Tayabji (1985)
133
Table 6-2. Effect of variations in axle group loads on induced tensile stress
Induced Tensile Stress in

Axle stress (MPa) Unbonded
Load Pavement
Group C1
(kN) / Stress in
Type Bonded Unbonded
Bonded
Pavement
53 0.807 1.379 1.709
SAST 1.718
130 1.759 3.037 1.727
80 0.988 1.7 1.721
SADT 1.724
150 1.733 2.993 1.727
90 0.637 1.055 1.656
TAST 1.656
200 1.346 2.228 1.655
135 0.819 1.348 1.646
TADT 1.648
240 1.408 2.323 1.65
180 0.821 1.274 1.552
TRDT 1.553
310 1.348 2.094 1.553
220 0.766 0.863 1.127
QADT 1.124
370 1.247 1.397 1.12
6.5. Critical location of axle groups upon pavements

As mentioned earlier, structural response of concrete pavements is affected by
vehicular load configurations, magnitude of applied loads and position of axle groups
upon the pavement as well as environmental effects. In concrete pavements, applied
loads are generally transferred to subbase and subgrade layers by the bending action of
concrete slab which results in a tensile stress at the top or the bottom surface layers of
the concrete slab. The applied loads can be vehicular and/or environmentally related.
Information on research conducted in the past for determination of the critical position
of vehicular loads upon the concrete pavements was comprehensively described in the
Chapter 2 of this thesis document.
As discussed in Section 5.3, structural behaviour of concrete pavements in curling

conditions depends on provision of debonding layer between the concrete slab and the
subbase. Since configuration and magnitude of vehicular loads have a significant effect
on induced tensile stress in concrete pavements (Yu et al. 1998, and Hiller and Roesler
2002) and a variety of axle group configurations is employed in heavy vehicle
industries, further study is required for other axle group types. Moreover, there is no
134
inclusive information on critical positions of each individual axle group on curled and
uncurled concrete pavements.
This study seeks to establish the critical positions of axle groups on jointed concrete
pavement based on Austroads (2004) recommendations. Loads from different axle
groups are separately applied at various locations on a number of pavement
configurations – unconfined and confined by adjacent lanes and shoulders- to evaluate
the critical design parameters of maximum tensile stress. Fully bonded and unbonded
boundary conditions between concrete slab and subbase were considered. Different
axle groups based on Austroads (2004) consisting of SAST, SADT, TAST, TADT,
TRDT and QADT were positioned at different locations of the pavement.
Curling and warping of concrete pavement were also taken into account. Effects of
axle group loadings and differential temperature gradients on both bonded and
unbonded concrete pavement will be separately studied. As mentioned in Section 6.3,
induced tensile stresses due to vehicular loads can be superimposed to thermal induced
tensile stresses in bonded pavements. For unbonded pavements, superposition of
induced stresses is accurate when differential temperature between the top and the
bottom surface layers of the concrete slab is more than 16.4˚C during daytime or lower
than -14.6˚C during nighttime. Further information on effects of combined action
due to vehicular loads and environmental effect is presented in Section 6.5.4. A full
pavement configuration subjected to combination of SADT and differential
temperatures is also studied to determine the pavement behaviour when differential
temperature ranges between -14.6˚C and 16.4˚C. Effects of modulus of subgrade
reaction and thickness of concrete slab on induced stress are subsequently discussed.
Results for different pavement configuration, namely, single lane, a confined lane, a
double confined lane and a full pavement (Fig. 6-1) are evaluated and compared with
those from existing research.
Results provided by Packard and Tayabji (1985) and AASHTO (2003) are examined
based on results of the current study. This helps to understand whether their findings
can be extended when other axle groups, pavement configurations, curling and
warping of concrete pavements, and bonded or unbonded boundary conditions
between concrete slab and subbase are considered.
135
6.5.1. Methodology
To study behaviour of JPCP under a combination of vehicular loads and environmental
effects, diverse configurations of JPCP consisting of a single lane (Fig. 6-1b), confined
lane (Fig. 6-1c), double confined lane (Fig. 6-1d) and full pavement (Fig. 6-1e) were
analysed. Subsequently, critical positions of axle groups upon the pavement were
determined.
With regards to effects of modulus of subgrade reaction and thickness of concrete slab
on induced tensile stress of a curled pavement (Kuo, 1998), different slab thicknesses
of 200, 250 and 300 mm and different modulus of subgrade reactions of 0.03, 0.05 and
0.07 MPa / mm were considered in a full pavement configuration.
SAST, SADT, TAST, TADT, TRDT, and QADT with average gross loads of 53 kN,
80 kN, 90 kN, 135 kN, 181 kN, and 221 kN (Fig. 3-21) based on Austroads (2004)
were respectively applied as the vehicular loads at the centre, middle of the
longitudinal edge and corner of the centre slab as shown in Figure 6-14. In this
research, these load locations are respectively called a centre, mid-edge and corner
loadings. A rectangular shaped tyre-pavement contact area based on the findings of
Gillespie et al. (1992) was considered. Other assumptions for load configuration were
as follows:
• Tyre inflation pressure of 750 kPa.

• Width-to-length ratio of tyre contact area of 0.7.
• Space between centres of dual tyres of 300 mm.
• Axle width of 1800 mm.
• Distance between axles in a given axle group was 1250 mm.
Results presented in Section 5.4 showed that the critical width to length ratio of tyre
contact area is between 0.6 and 0.8 with average of 0.7, the critical distance between
axles in a given axle group is between 1050 mm and 1150 mm with average of 1100
mm for all axle groups. In addition for TAST and TADT groups this value can also be
between 1350 mm to 1450 mm with an average of 1400 mm. In order to compare
results of the current study with results provided by Packard and Tayabji (1985), a
value of 0.7 for width to length ratio of tyre contact area and 1250 mm distance
136
between axles in a given axle group, as assumed by Packard and Tayabji (1985), were
chosen.
Figure 6-14. Position of applied loads for different axle groups on the centre concrete
slab
As high temperature gradients (more than 25˚C) would result in severe damage of
unreinforced concrete slab of a normal thickness, linear differential temperature
gradients of -25˚C (nighttime temperature) to 25˚C (daytime temperature) were
therefore considered between the top and the bottom surface layers of concrete slab.
6.5.2. Axle group loadings

In the absence of differential temperature gradients, results of current study indicate
that vehicular induced tensile stresses in unbonded concrete pavement due to mid-edge
loading are greater than those from centre and corner loading for all pavement
configurations when SAST, SADT, TAST and TADT are studied (Fig. 6-15). In
contrast, corner loading of QADT results in greater tensile stress than mid-edge
loading for all pavement configurations. Furthermore, corner loading of TRDT shows
similar results for double confined lane.
137
Consideration of bonded boundary condition between concrete slab and subbase
always results in lower tensile stress due to centre loading than other loading types
(Fig. 6-16). However, corner loading results in greater induced tensile stress than mid-
edge loading for all axle group types except for SADT and SAST. This becomes
predominant when traffic lane is confined by shoulders or adjacent traffic lanes.
Results of the current study reveal that the AASHTO recommendation (2003) is valid
for fully unbonded boundary condition between concrete slab and subbase though this
is not true for corner loading of SADT. While the bonded boundary condition is
considered in the analysis, the AASHTO recommendation (2003) is not able to capture
maximum vehicular induced tensile stress as induced tensile stresses in full pavement
are greater than those tensile stresses produced in confined lane for corner loading (see
Fig. 6-16).
As mentioned earlier, mid-edge loading was the critical loading case in research
conducted by Packard and Tayabji (1985). Results of the current study show that mid-
edge loading, in the absence of differential temperature is the key factor for fatigue
cracking of unbonded concrete pavements though this is not true for QADT. Corner
loading results in greater vehicular induced tensile stress for all TAST, TADT, TRDT
and QADT when fully bonded boundary condition is considered.
As mentioned earlier, a comparison between vehicular induced stresses in bonded and

unbonded boundary conditions between concrete slab and subbase shows that an
unbonded boundary condition produces greater vehicular induced tensile stress in
concrete pavement. This is due to loss of support in those locations of the concrete slab
that were lifted-off. The minimum increase in the value of tensile stress is 30 percent
for corner loading of TRDT and the maximum increase is about 133 percent for mid
edge loading of TAST. These results show that vehicular induced tensile stress is
highly affected by the boundary condition between concrete slab and subbase.
6.5.3. Thermal induced stress

As mentioned earlier, the main reason for placing a friction reducer layer between
concrete slab and subbase is to reduce the early age stresses in concrete slabs.
Nevertheless, the debonding layer plays a significant role in structural behaviour of
concrete pavements during the pavement life.
138
a) Comparison between Centre and Mid-Edge loadings
b) Comparison between Corner and Mid-Edge loadings
Figure 6-15. Vehicular induced stress in different pavement configurations for

unbonded boundary condition
139
a) Comparison between Centre and Mid-Edge loadings
b) Comparison between Corner and Mid-Edge loadings
Figure 6-16. Vehicular induced stress in different pavement configurations for bonded
boundary condition
140
Since differential temperature gradients and drying shrinkage tend to curl or warp
concrete pavement, the location of neutral axis changes from its original location
toward the top or the bottom surface layer of the concrete slab depending on the
concrete slab curvature. Hence, the critical position of axle groups may be affected by
consideration of environmental forces. For that purpose, daytime and nighttime
temperature gradients of 10˚C and 25˚C were considered in the current study.
Ongel and Harvey (2004) found that nighttime curling and warping are two reasons
behind longitudinal cracking. Results of the current study show that the thermal tensile
stress is at maximum along the longitudinal centreline of each traffic lane and
decreases toward the edges of the traffic lane as expected. Furthermore, high daytime
differential temperature results in bottom-up longitudinal cracking and consequently
shall also be considered as a reason for longitudinal cracking.
Figure 6-17 shows induced thermal stress influence lines at longitudinal joint along
the wheel path for the centre panel of a full pavement configuration with the fully
bonded boundary conditions between concrete slab and subbase. Figure 6-18 presents
similar results for fully unbonded condition. In these figures, the number after daytime
or nighttime indicates the absolute temperature difference between the top and the
bottom surface layers of the concrete slab. For instance, daytime 10 indicates a
differential temperature between the top and the bottom surface layer of 10 ˚C.
Results of the current study indicate nighttime temperature gradients produce greater
tensile stress at the top surface layer of the concrete slab whereas daytime temperature
gradients result in greater tensile stress at the bottom surface layer of the concrete slab
as expected. Furthermore, the area close to transverse joints experiences greater tensile
stresses than other locations within the pavement when higher differential temperature
gradients are considered. However for nighttime differential temperature of -25 ˚C,
the stress at the central area of the slab is about 10 per cent higher than that in the area
close to the transverse joints (Fig. 6-17a).
141
a) Induced tensile stress at the top surface layer of the centre slab close to edge
b) Induced tensile stress at the bottom surface layer of the centre slab close to
edge
Figure 6-17. Curling induced stress influence lines in bonded concrete slab with full
pavement configuration
142
a) Induced tensile stress at the top surface layer of the centre slab close to edge
b) Induced tensile stress at the bottom surface layer of the centre slab close to
edge
Figure 6-18. Curling induced stress influence lines in unbonded concrete slab with full
pavement configuration
143
When a single concrete slab is freely curled due to a differential temperature gradient,
a bending stress (flexural stress) is induced at the top or the bottom surface layer of the
concrete slab due to its residual stiffness (Mohamed and Hansen, 1997). This
behaviour in the concrete slab can be observed in the absent of the restraining factors
such as slab weight, subgrade and subbase resistant, vehicular loads and friction force
at the interface of the concrete slab and subbase. In this case, maximum induced tensile
stress occurs at the centre of the slab and at the top surface layer during the day and at
the bottom surface layer during the night.
In the presence of the restraining factors, on the other hand, another tensile stress
occurs on the opposite sides of the concrete slab where subjected to the residual tensile
stress. Under this condition, curling induced tensile stress occurs at the top surface
layer of the concrete slab during nighttime and at the bottom surface layer of the
concrete slab during daytime (Ongle and Harvey, 2004).
Hence, the top and the bottom surface layers of the concrete slab at a certain location
in the longitudinal direction may be subjected to different tensile stresses. For instance,
a tensile stress of 0.5 MPa and 1.75 MPa occur two meters away from transverse joints
in a daytime curling condition (25˚C differential temperature) at the top and the bottom
surface layers of the concrete slab respectively (Fig. 6-17a). As mentioned earlier,
while residual stresses due to daytime differential temperature occur at the top of the
concrete slab, stresses due to the restraining factors are induced at the bottom surface
layer of the slab. In general, the magnitudes of stresses due to the restraining factors
are greater than residual stresses.
Combination of these stresses together with location of neutral axis in the concrete slab
dictates a specific stress distribution in the slab depth. In some circumstances, while
there are negligible stresses at the bottom or the top surface layers of the slab during
nighttime or daytime differential temperature respectively, a considerable tensile stress
occurs at the top or the bottom surface layers of the slab in respect to nighttime or
daytime differential temperature (see results of mid slab stresses in fully unbonded
concrete slab in Fig. 6-18).
On the other hand, consideration of bonded layers associated with provision of load
transfer devices (dowels and tie bars) at transverse and longitudinal edges influences
144
the location of neutral axis and consequently changes stress distribution regime within
the pavement depth. In this case, both the top and the bottom surface layers are
subjected to tensile stress (see results of mid-slab stresses in Fig. 6-17 and stress
distribution close to transverse joints in Fig. 6-18).
For the selected slab dimensions, the critical location of thermal stress is at distance of
about 400 mm from the transverse joints. Since this distance is a function of pavement
length, varying the length by ± 2 metres may shift the location of maximum thermal
stresses by ± 50 mm. This finding explains the reason behind the formation of
transverse cracks commonly found near transverse joints. In addition to finding by
Ongel and Harvey (2004), who stated that corner loading associated with nighttime
curling and warping results in corner cracking, high daytime and nighttime differential
temperature can also crack the concrete slab in the area close to transverse joints.
With regards to the significant effect of the boundary condition between concrete slab
and subbase on induced thermal stress, Kuo’s findings (1998) is modified as curling
induced stress is affected by temperature differential, self-weight of concrete
pavement, support under concrete slab and boundary condition between concrete slab
and subbase.
6.5.4. Combination of vehicular and thermal induced stresses

As mentioned in Section 6.3.2, the vehicular induced stresses can be superimposed
with thermal induced stresses in bonded pavements and in unbonded pavements under
certain condition. The superposition is not accurate for unbonded pavements subjected
to a differential temperature between -14.6˚C and 16.4˚C. Hence, further study will be
carried out to determine the pavement behaviour under the combined action of
vehicular loads and environmental effects.
Figure 6-19 presents results of the current study for the bonded boundary condition
between the concrete slab and the subbase when a full pavement configuration is
subjected to differential temperatures of 10˚C and 25˚C. Vehicular induced tensile
stresses for mid-edge and corner loadings are represented in this figure.
Results indicate that corner loading in the presence of thermal curling produces greater
induced tensile stress than mid-edge loading if the bonded boundary condition is
145
considered. In the unbonded boundary condition, higher differential temperature
gradients change the critical location of axle groups toward an area close to the corner
of the pavement (Fig. 6-20). However, mid-edge loading results in greater induced
tensile stress if pavement experiences lower differential temperature gradients.
a) Differential temperature of 10 ˚C
b) Differential temperature of 25 ˚C
Figure 6-19. Vehicular and thermal induced tensile stresses in bonded concrete slab
146
a) Differential temperature of 10 ˚C
b) Differential temperature of 25 ˚C
Figure 6-20. Vehicular and thermal induced tensile stresses in unbonded concrete slab
Hiller and Roesler (2002) reported similar result when pavement was subjected to a
nighttime temperature gradient of 8.3˚C. This differential temperature is much lower
than the differential temperature gradient considered in the current study to move the
critical axle position from mid-edge toward corner of the concrete slab. This difference
may be explained by taking into account the length of the pavement which is 4.6 m in
the current study and 5.8 m in the wok of Hiller and Roesler (2002). An increase in the
147
pavement length increases the magnitude of shear stress between concrete slab and
subbase or subgrade (in the absence of subbase) and consequently increases thermal
induced stress. Consequently, the possibility of separation between layers in lower
differential temperature increases. Hence, the aforementioned finding can be revised. If
a separation due to environmental force occurs between the concrete slab and the
subbase, corner loading results in greater tensile stress than mid-edge loading.
Since superposition law is not accurate enough when an unbonded pavement is

subjected to small variation of temperature gradients, a full pavement configuration
was examined for corner and mid edge loadings together with differential temperature.
This helps to study pavement behaviours under vehicular load and low or moderate
differential temperatures. To do this, SADT was considered in the analysis. Figure 6-
21 compares thermal and vehicular induced tensile stresses for bonded and unbonded
pavements due to mid-edge loading with the corresponding stresses of corner loading.
Results indicate that corner loading during nighttime produces greater tensile stresses
than mid-edge loading in bonded concrete slab. Vice versa, mid-edge loading during
daytime results in greater tensile stress than corner loading. Nevertheless, corner
loading is the critical loading condition when differential temperature exceeds 15˚C. In
fully unbonded pavement, on the other hand, mid-edge loading result in greater tensile
stress than corner loading when a differential temperature greater than -5˚C is
considered. Further decrease in differential temperature results in greater tensile
stresses in corner loading than mid-edge loading. These can be explained by taking the
pavement curvature into consideration.
During the daytime, thermal induced tensile stresses similar to vehicular induced
stresses occur at the bottom surface layer of the slab with a maximum at mid-edge. As
a result, mid-edge loading produces greater tensile stresses than corner loading
regardless of boundary condition between concrete slab and subbase. However, further
increase in differential temperature tends to lift-off the centre of the slab. As a result, a
greater tensile stress at the bottom surface layer of the slab occurs due to corner
loading than mid-edge loading. Provision of dowels at transverse joints together with
slab weight results in a bending stress in the same direction of induced stress due to
corner loading in daytime curling.
148
While nighttime differential temperature tends to lift-off transverse edges of the
concrete slab, corner loading presses down the slab edge to return it to its original
location. Consequently, tensile stress occurs at a location close to transverse joint and
at the top surface layer of the slab due to cantilever action of the slab.
3
Induced Tensile stress
2.5
2
(MPa)
1.5
0.5
-15 -10 -5 0 5 10 15
Differential Temperature (º C)
Unbonded Mid-Edge Loading Unbonded Corner Loading

Bonded Mid-Edge Loading Bonded Corner Loading
Figure 6-21. Combination of vehicular and thermal induced tensile stresses in a full
pavement configuration subjected to SADT
Crack location within the concrete slab is affected by vehicular loading conditions
(corner, centre and mid-edge) and differential temperatures. To efficiently determine
crack locations in concrete slab, the centre slab was divided into six areas based on
absolute value of tensile stress (Fig. 6-22). The critical locations for fatigue cracking
in a full pavement configuration subjected to combination of SADT and differential
temperatures are presented in Table 6-3.
Interestingly, high differential temperatures result in joint faulting irrespective of the

vehicular loading conditions and boundary conditions between the concrete slab and
the subbase. This indicates that the finding of Ongle and Harvey (2004), that daytime
and nighttime differential temperatures result in bottom-up and top-down cracking,
respectively, is correct for a differential temperature lower than 15˚C during daytime
and greater than -15˚C during nighttime. Furthermore, corner, centre and mid-edge
149
loadings can result in different types of fatigue failure of the concrete slab depending
on differential temperature.
Figure 6-22. Information on location of fatigue cracking
These can be summarised as follows:
• While unbonded pavement experiences a daytime differential temperature of

lower than 15˚C, the maximum stress always occurs at the bottom surface layer
of the slab where the vehicular load is applied.
• While unbonded pavement experiences a nighttime differential temperature of

greater than -15˚C:
- Centre loading results in top-down cracking at the corner or in an area

between corner and mid-length of the slab edge.
- Mid-edge loading results in top-down cracking at the location between

corner and mid-length of the slab edge.
- Corner loading results in top-down longitudinal cracking along the

centre area of the slab.
• While bonded pavement experiences daytime differential temperature of less

than 10˚C, the maximum stress always occurs at the bottom surface layer of the
slab where the vehicular load is applied.
• While bonded pavement experiences nighttime differential temperature of

greater than -15˚C:
- Centre loading results in longitudinal top-down cracking of the slab.
150
- Mid-edge loading results in mid-edge bottom-up transverse cracking.
- Corner loading results in top-down transverse cracking at a location

between corner and mid-length of the slab edge.
Table 6-3. Critical location of fatigue cracking in full pavement model due to different
differential temperatures and SADT
Slab thickness = 250 mm

Modulus of subgrade reaction = Temperature Centre Mid Edge Corner
0.03 MPa/mm (º C) Loading Loading Loading
Axle group load = SADT
25 CE at MD CE at MD CO at MD
20 CE at MD CE at MD CO at MD
Daytime 15 UW at B ME at B CO at B
10 UW at B ME at B CO at B
Unbonded 0 UW at B ME at B CO at B
-5 CO at T ¼ E at T C at T
-10 ¼ E at T ¼ E at T C at T
Nighttime -15 ¼ E at T ¼ E at T C at T
-20 CO at MD CO at MD CO at MD
25 CO at MD CO at MD CE at MD
20 CE at MD CO at MD CO at MD
Daytime 15 CE at MD ME at B CE at B
Bonded 0 UW at B ME at B CE at B
-5 UW at B ME at B ¼ E at T
-10 C at T ME at B ¼ E at T
Nighttime -15 C at T ME at B ¼ E at T
UW: Under wheel B: Bottom surface layer T: Top surface layer See Fig. 6-22 for further information
As mentioned earlier, Hiller and Roesler (2005) indicated that nighttime differential
temperature results in mid-edge top-down cracking. Results of the current study,
however, indicate that this is not true when pavement experiences nighttime
differential temperature of less than -20˚C as joint faulting becomes the critical damage
mode of concrete pavements in this case. Buch et al. (2004) mentioned that mid-edge
and corner loadings result in bottom-up and top-down cracking respectively.
Nevertheless, results of the current study show that mid-edge loading results in
151
bottom-up cracking in the absence of nighttime differential temperatures or in presence
of daytime differential temperature of less than 15˚C. In the presence of nighttime
differential temperature, mid edge loading can result in top-down cracking at the slab
edge or joint faulting (see Table 6-3). Furthermore, corner loading results in top-down
cracking in the absence of daytime differential temperature or in presence of nighttime
differential temperature of less than -20˚C. In the presence of daytime differential
temperature, corner loading results in bottom-up cracking at the corner. Reasons
behind joint faulting are presented in Section 5.5.5 when the effect of concrete slab
thickness on pavement response is discussed.
Since Austroads (2004) restricts the maximum distance between transverse joints in
unreinforced concrete slab to 4600 mm, a separation between concrete slab and
subbase in the absence of vehicular loads occurs when a differential temperature close
to that presented in Figure 6-18 is considered. Vehicular loads can increase or
decrease the critical value of differential temperature depending on the load position
and nature of differential temperature (daytime or nighttime).
It should be noted that the results of Hiller and Roesler (2002) were developed for
unbonded boundary condition between concrete slab and subbase. Mid-edge loading
results in greater tensile stress than corner loading when lower nighttime differential
temperature is considered or no separation occurs between concrete slab and subbase.
Hence, Kuo’s recommendation (1998) is also valid for a certain range of differential
temperatures and only unbonded boundary condition
A comparison between induced tensile stresses in bonded and unbonded boundary

conditions reveals that the unbonded boundary condition between concrete slab and
subbase requires careful consideration when pavement is constructed in hot or cold
weather where high differential temperature gradients may be produced in the concrete
depth.
6.5.5. Effects of slab thickness on induced tensile stress

To find how slab thickness affects the results of the current study, a full pavement
configuration subjected to different daytime and nighttime differential temperature
gradients together with corner and mid-edge loadings was considered. Both bonded
and unbonded boundary conditions were taken into account. Table 6-4 shows results
152
of the current study for maximum induced tensile stress when modulus of subgrade
reaction was held constant and SADT was applied on the centre concrete slab panel as
either corner loading or mid-edge loading.
Table 6-4. Effect of concrete slab thickness on maximum induced stress (MPa) due to
different differential temperatures and SADT
Modulus of Subgrade Reaction Slab Thickness (mm)

Temperature
= 0.03 MPa/mm
(º C)
Axle group load = SADT 200 250 300
5 1.627 1.213 0.876
Corner loading
Daytime 10 1.944 1.544 1.32

15 2.172 1.52 4.52
5 1.02 0.846 0.685
Nighttime 10 1.6 1.282 1.009
15 2.038 1.56 3.623
Unbonded
5 2.296 1.744 1.333
Mid-edge loading
Daytime 10 2.938 2.257 1.724

15 3.514 2.758 2.044
5 1.351 1.028 0.803
Nighttime 10 1.328 0.945 0.704
15 1.351 0.881 3.381
5 1.206 1.028 0.897
Corner loading
Daytime 10 2.025 1.745 1.524

15 2.845 2.46 2.14
5 0.922 0.825 0.722
Nighttime 10 1.659 1.489 1.305
15 2.404 2.16 1.9
Bonded
5 1.4 1.211 1.027
Mid-edge loading
Daytime 10 2.12 1.837 1.557

15 2.841 2.465 2.086
5 0.839 0.703 0.623
Nighttime 10 1.566 1.332 1.13
15 2.298 1.965 1.674
Highlighted values occur very close to the corner of loaded Concrete slab and at the
middle of the Concrete slab depth
Results indicate that an increase in thickness of concrete slab decreases the magnitude
of induced tensile stress regardless of boundary condition between concrete slab and
subbase as mentioned by Buch et al. (2004). However, the use of thicker concrete slab
(300 mm thickness) associated with higher differential temperature (15˚C) rapidly
increases the value of induced tensile stress at mid-depth of the concrete slab for a
node in unloaded transverse joint and close to the corner of the concrete slab. Location
153
of this node has been shown in Figure 6-23. This can be explained by taking into
account the concrete slab curvature together with location of axle group upon
pavement.
A downward curvature is produced during daytime temperature gradients. Corner

loading associated with thermal curvature results in an increase in the area of
separation between concrete slab and subbase toward the unloaded corners of the slab.
It ultimately induces a lifting-off at unloaded corners of the concrete slab. While
thinner concrete slab is considered, both unloaded corners of the concrete slab lift-off
and result in induced bending stress due to transferring the weight of adjacent concrete
panel divided equally into all dowels located in unloaded transverse joint.
Consequently, induced stress at interface of dowel and concrete remains in the normal
range. Vice versa, thicker concrete slab results in non-uniform lift-off of unloaded
corners which ultimately induce higher bending stress in the dowel located close to
lifted-off corner. Mid-edge loading, on the other hand, alleviates the severity of the
problem.
Figure 6-23. Position of the critical location in thick unreinforced concrete pavement
Nighttime temperature results in upward curvature. Both corner and mid-edge loadings
together with nighttime temperature enhanced the magnitude of induced stress in thick
concrete pavement as describes above. This finding suggests a particular dowel
arrangement at corner of the concrete slab or a maximum slab thickness that shall be
considered in the design of unreinforced concrete pavement.
Hiller and Roesler (2002) showed that a change in the thickness of the concrete slab
changes the magnitude of induced stresses due to corner and mid-edge loadings
154
uniformly. However, results of the current study indicate a non-uniform change
between corner loading and mid-edge loading induced stresses. For instance, the
proportion of induced tensile stress due to corner loading to induced tensile stress due
to mid-edge loading, when unbonded pavement is subjected to a nighttime differential
temperature gradient of 10 ˚C, is 1.205, 1.357and 1.433 for slab thickness of 200 mm,
250 mm and 300 mm respectively.
6.5.6. Effects of modulus of subgrade reaction on induced tensile stress

To determine effects of modulus of subgrade reaction on the results of the current
study, a full pavement configuration was subjected to different daytime and nighttime
differential temperature gradients together with corner and mid-edge loadings. Both
bonded and unbonded boundary conditions were taken into account. Table 6-5 shows
results of the current study for maximum induced tensile stress when thickness of the
concrete slab was held constant and SADT was applied on the centre concrete slab
panel as either corner loading or mid-edge loading.
Results show that modulus of subgrade reaction has different effects on the pavement
response when daytime or nighttime differential temperature, corner or mid-edge
loading, and unbonded or bonded boundary condition between concrete slab and
subbase are considered.
An increase in modulus of subgrade reaction in the presence of nighttime temperature

increases the magnitude of induced tensile stresses in most case studies. This is
compatible with the statement of Buch et al. (2004). In contrast, induced tensile
stresses in the bonded pavement and in the presence of daytime differential
temperatures decrease when modulus of subgrade reaction is increased and SADT is
applied at the corner of the concrete slab. An increase in modulus of subgrade reaction
associated with higher daytime differential temperatures and mid-edge loading
increases the magnitude of induced tensile stress regardless of boundary condition
between concrete slab and subbase. In some cases, when, for example, the unbonded
pavement is subjected to daytime differential temperature and mid-edge loading,
modulus of subgrade reaction has no effect on maximum induced tensile stress as axle
load induced stress is in negative direction of thermal induced tensile stress.
155
Hiller and Roesler (2002) showed that an increase in modulus of subgrade reaction in
the presence of nighttime differential temperature of 8.3˚C increases the proportion of
corner loading induced stress to mid-edge loading induced stress. Results of the current
study for unbonded pavement subjected to nighttime differential temperature of 10˚C
show similar result.
Table 6-5. Effect of modulus of subgrade reaction on maximum induced stress (MPa)
due to different differential temperatures and SADT
Slab thickness = 250 mm Temperature Modulus of Subgrade

Axle group load = SADT (º C) Reaction (MPa/mm)
0.03 0.05 0.07
5 1.213 1.18 1.2
Corner loading
Daytime 10 1.544 1.413 1.458

15 1.52 1.58 1.649
5 0.846 1.18 0.924
Nighttime 10 1.282 1.377 1.434
15 1.56 1.58 1.757
Unbonded
5 1.744 1.686 1.632
Mid-edge loading
Daytime 10 2.257 2.31 2.329

15 2.758 2.781 2.853
5 1.028 0.847 0.751
Nighttime 10 0.945 0.791 0.826
15 0.881 0.895 1.007
5 1.028 0.967 0.946
Corner loading
Daytime 10 1.745 1.651 1.581

15 2.46 2.234 2.248
5 0.825 0.848 0.864
Nighttime 10 1.489 1.543 1.586
15 2.16 2.248 2.315
Bonded 5 1.211 1.198 1.197
Mid-edge loading
Daytime 10 1.837 1.859 1.885

15 2.465 2.519 2.573
5 0.703 0.705 0.709
Nighttime 10 1.332 1.345 1.372
15 1.965 2.015 2.075
6.6. Summary
carried out to understand how debonding layer, axle group configurations, differential
stress within the concrete slab. Results showed that the benefits offered by
consideration of the unbonded boundary condition cease at a certain value of
156
differential temperature. Hence, a particular care needs to be given to those pavement
projects constructed in hot or cold weather where high differential temperature
gradients may be produced in concrete depth.
Furthermore, the risk of pavement deterioration can be minimised through a proper

selection of a number of design parameters:
• The use of 0.6 to 0.8 for width to length ratio of tyre-pavement contact area
with average of 0.7 is recommended to calculate the critical concrete pavement
response.
• Tyre inflation pressure has a strong effect on maximum tensile stress

particularly when the axle load is applied at the corner of the slab. The highest
possible pressure should be selected for design, taking into account the state of
the art of tyre industry.
• Axle width and distance between the centres of dual wheels has no significant
effect on the pavement response. However, a distance of 300 mm between the
centres of dual tyres will produce slightly higher tension stress and deflection.
• Shorter length of axle width can cause greater pavement response. Hence, a
distance of 1800 mm for axle width is recommended.
• The critical axle spacing in a given axle group is between 1050 mm and 1150
mm with average of 1100 mm for all axle groups. In addition, for TASD or
TADT this value can also vary between 1350mm to 1450mm with average of
1400 mm.
• A shift in load distribution between axles can increase the deterioration of

concrete pavements particularly in the presence of pavement roughness.
Therefore, restriction on the use of flip/shift axles or mixed suspension systems
is recommended.
Critical positions of different axle groups in uncurled and curled jointed concrete
pavement with different configurations were also studied. Results of the current study
indicate that AASHTO recommendation (2003) -except for SADT-and results of
Packard and Tayabji (1985) –except for QADT- are valid for the fully unbonded
157
boundary condition between concrete slab and subbase and uncurled pavement.
Results of the current study also show that pavement performance under combinations
of vehicular loads and differential temperatures is significantly affected by boundary
condition between concrete slab and subbase.
The reasons behind longitudinal, transverse and corner cracking were addressed. The
significant findings in this area were (i) corner loading is critical when there is a
bonded boundary condition between concrete slab and subbase (ii) corner loading is
also critical when a separation due to environmental forces occurs between the
unbonded concrete slab and subbase. Moreover, corner, centre and mid-edge loadings
can result in different types of fatigue failure of the concrete slab depending on
differential temperature.
There is an inverse relationship between induced tensile stress and the thickness of
concrete slab so that an increase in thickness of concrete slab decreases the magnitude
of induced tensile stress. However, a maximum slab thickness or dowel arrangement at
corners of the slab needs to be considered in unreinforced concrete pavement as thicker
slabs are sensitive to high differential temperature gradient together with axle loading.
An increase in modulus of subgrade reaction can increase or decrease the magnitude of
tensile stress depending on boundary condition between concrete slab and subbase,
corner or mid-edge loading and daytime or nighttime differential temperature.
158
Chapter 7
LABORATORY TESTS OF CONCRETE
7.1. Introduction
As mentioned in Chapter 2 of this thesis, the rate of deterioration in concrete
pavements is affected by concrete properties. It is, hence, essential to determine
structural properties of the concrete. These include concrete compressive and flexural
strengths, modulus of elasticity, fatigue life of the concrete and shear transfer
capability of the aggregate interlock and cement paste. Some typical equations were
previously developed for prediction of concrete properties based on its compressive
strength (see Chapter 2). Since a variety of equations for a specific property of
concrete, i.e. flexural strength, is available worldwide, the main aim of this chapter is
to determine the most accurate equation for those concrete properties affected
pavement responses.
In addition, the laboratory fatigue test of concrete is traditionally based on one-

directional cyclic loads. However, the concrete pavements are curled due to
temperature fluctuations. Consequently, a certain point within the pavement
experiences a wide range of stresses from compressive stress to tensile stress
depending on pavement curvature. Therefore, a new setup for laboratory fatigue test of
concrete prism beam is developed in the current research to take into consideration
effects of daytime and nighttime temperature fluctuations on fatigue life of the
concrete. All concrete specimens were cast at the Beenleigh laboratory of the Rinker
company and were then delivered to be tested at the QUT’s concrete laboratory. The
specimens were tested based on Australian standard for concrete sampling and testing,
AS1012.
7.2. Compressive test

Cylindrical specimens of 100 mm dimeter and 200 mm height were used to determine
concrete compressive strength. The average mass of the specimens at the testing time
was 3644 grams. Since no particular concrete curing is considered in the concrete
pavement constructed in Australia, some of specimens were cured in the ambient
159
temperature (air cured). The test was performed by using rubber cap (Fig. 7-1). To
determine stress-strain curve of concrete, two different approaches were considered. In
the first approach, an external electrical strain gauge was installed on the surface of
specimens (Fig. 7-1). In the second approach, concrete compressive test ring (Fig. 7-2)
was used.
Figure 7-1. Concrete compressive test setup using external electrical strain gauge
Figure 7-2. Concrete compressive test ring
160
The test ring contains two standard linear displacement transducers. The induced strain
in the specimen is calculated based on average shortening of the specimen length. The
stress-train curve was then used to determine the modulus elasticity of concrete.
While Figure 7-3 shows a typical failure mode of the specimens in the compressive
test, Figure 7-4 shows a typical concrete stress-strain curve derived from results of the
current study. Results of the concrete compressive tests (Table 7-1) indicate that the
concrete strength is strongly affected by the curing method as expected. The air cured
specimens have a greater standard deviations than those cured in the water tank.
However, the use of polyethylene sheet for covering the specimens in the air cured
condition decreases the standard deviation by about 33 per cent.
Figure 7-3. Typical failure mode of the cylindrical concrete specimens
161
60
55
50
45
40
Stress (MPa)
35
30
25
20
15
10
5
0
0 500 1000 1500 2000 2500
Strain (µε)
Figure 7-4. Typical stress-strain curve of a concrete specimen
7.3. Flexural test

Prism beams with a cross section of 100 mm × 100 mm with and average length of 368
mm were used to determine the flexural strength of concrete. The average mass of the
specimens was 8582 grams. The flexural test was performed using four point loading
configuration. Similar to the previous stage, both water cured and air cured specimens
were used. Whilst Table 7-2 presents results of the current study for specimens with 14
days age, Table 7-3 presents similar results for specimens with 28 days age.
Results of the current study show that the air cured specimens always have a lower
flexural strength than those cured in water tank. However, a lower standard deviation
was observed in the air cured specimens with 14 days age (see Table 7-2). The flexural
strength of water cured specimens at 14 days age and 28 days age are greater than that
of corresponding air cured specimens by about 18% and 51%. Furthermore, the use of
polyethylene sheet for covering the specimens in the air cured condition seems to have
no significant effect on concrete flexural strength.
Results of the current study are also used to examine the accuracy of equations
developed for estimation of concrete flexural strength. They are Equations 2-1 to 2-4
which were introduced in the Chapter 2 of this thesis.
162
Table 7-1. Results of concrete compressive tests
Compressive Average Standard

Date Age Curing
Name strength strength deviation f'c
of test (day) Method
(MPa) (MPa) (MPa)
C1 32.42
C2 30.55
09 / 12 /2005
Water cured
C3 32.42
14 32.16 0.83 30.79
C4 32.27
C5 32.27
C6 33
C7 30.2
21 / 12 /2005
C8 31.24
Air cured
C9 14 28.27 28.67 2.74 24.17
C10 29.46
C11 24.16
27 /10 /2005
C12 48.42
Air cured
C13 28 40.15 45.78 4.88 37.78
C14 48.76
C15 45.5
C16 43.08
23 /12 /2005
Water cured
C17 44.22
28 42.98 2.134 39.48
C18 42.08
C19 39.3
C20 43.7
Air cured (specimens were
covered by a polyethylene
C21 49.11
17 / 05 /2006
C22 46.39
sheet)
C23 28 46.25 49.19 3.29 43.80
C24 54.31
C25 49.9
163
Results of concrete compressive strength of 28 days age tested on 27/10.2005,
23/12/2005 and 17/05/2006 (see Table 7-1) were employed to estimate the flexural
strength of the concrete using Equations 2-1 to 2-4. The estimated flexural strength is
then compared with the corresponding results of flexural tests provided in Table 7-3.
Table 7-2. Results of flexural tests for specimens at 14 days age
Flexural Average Standard

Date of Specimen Specimen Curing strength strength deviation fr
test Name age (day) Method (MPa) (MPa) (MPa)
F1 3.47
09 / 12 /2005
Water cured
F2 3.78
F3 14 3.37 3.61 0.194 3.29
F4 3.82
F5 3.61
F6 2.95
21 / 12 /2005
F7 3.22
Air cured
F8 14 2.89 3.006 0.134 2.79
F9 2.92
F10 3.05
A comparison between results (Table 7-4) shows that Equation 2-3 ( 0.75 f C' ) is the
most accurate prediction model for the water cured specimens. On the other hand, none
of the provided equations (2-1 to 2-4) are accurate enough to be used for estimation of
flexural strength of the air cured specimens. Based on results of the current study, the
following equation is recommended for estimation of flexural strength of air cured
concrete:
f r = 0.49 f ' c (7-1)
164
Table 7-3. Results of flexural tests for specimens at 28 days age

fr
Date of Specimen Specimen Curing strength strength deviation
F11 4.73
27 /10 /2005
Air cured
F12 3.84
28 4.05 0.51 3.21
F13 4.1
F14 3.53
F15 5.6
F16 5.08
23 /12 /2005
Water cured
F17 5.46
28 5.28 0.249 4.87
F18 5.12
F19 4.98
F20 5.42
Air cured (specimens were covered by a polyethylene sheet)
F21 4.83
F22 3.65
F23 3.12
F24 4.2
F25 3.36
17 / 05 /2006
F26 3.76
28 3.94 0.442 3.22
F27 4.25
F28 4.1
F29 3.98
F30 3.89
F31 4.05
F32 4.1
165
Table 7-4. Comparison between results of flexural laboratory tests with results from
equations provided in the past for estimation of flexural strength
Concrete compressive strength at 28 days

Equations for estimation of age (MPa)
Flexural Strength
37.78 39.48 43.8
(27/10/2005) (23/12/2005) (17/05/2006)
Equation 2-1 ( f r = 0.6 f c' ) 3.688 3.77 3.97
Equation 2-2 ( f r = 0.7 f c' ) 4.3 4.397 4.623
Equation 2-3 ( f r = 0.75 f c' ) 4.61 4.712 4.964
Equation 2-4 ( f r = 0.94 f c' ) 5.778 5.91 6.22
Results of laboratory flexural tests 3.214 4.868 3.216

performed in the current study
The probability analysis of the laboratory tests shows that about 95 specimens are
required for developing an accurate equation for estimation of flexural strength of air
cured concrete. The number of required specimens is calculated based on the standard
deviation of all air cured specimens with the assumption that results of the flexural
tests have a normal standard distribution.
7.4. Modulus of elasticity

The stress-strain curve provided in Figure 7-4 was used to examine Equations 2-5 to 2-
7 for prediction of modulus of elasticity of concrete. The concrete modulus of elasticity
is firstly determined using secant method. The secant modulus of elasticity is
determined based on a line passing from point of zero stress to point of 0.45 f ' c stress.
Ultimately, the concrete modulus of elasticity is calculated. Results of current study

show that the average value of modulus of elasticity is 30000 MPa.
Since the stress-strain curve used in the current study is based on the laboratory tests
carried out on 23/12/2005, the density and compressive strength of concrete for those
specimens tested on 23/12/2005 were considered. The magnitude of concrete density
and compressive strength for the above mentioned date are 2320 kg/m3 and 39.48 MPa,
respectively.
166
Consequently, the modulus of elasticity of concrete can be estimated using Equation 2-
5 and 2-7:
E c = 0.033 × 23201.5 × 39.48 = 23170 MPa (from Equation 2-5)
E c = 0.043 × 23201.5 × 39.48 = 30191 MPa (from Equation 2-7)
This result shows that the Equation 2-7 is adequate enough to be considered for
estimation of modulus of elasticity.
7.5. Notch beam test

The notch beams were extensively used in the past (Jensen and Hansen 2000, Sousa et
al. 2006) to estimate the fracture energy of concrete. However, the notch beams in the
current study are used to determine the shear transfer capacity of aggregate interlock
and cement paste at transverse joints. As mentioned in the Chapter 3, the transverse
joints are utilized by a saw cut along total width of the pavement where pavement has
been reinforced by dowel bars. Since the shear force is transferred across transverse
joints through a combination of aggregate interlock and bending and shear action of
the dowel bars, it is essential to know how transverse joints perform after crack
initiation at the transverse joints.
Information on this matter provided in the past (Sousa et al., 2006) showed that the
shear load can be also transferred across joints through friction force between
aggregates located at the interface of the crack. However, the load transfer efficiency
strongly depends on crack width in this condition. In a conservative approach, the load
transfer efficiency due to aggregate interlock across transverse cracks can be assumed
to be zero. In other words, the load is only transferred across transverse joints through
shear and bending action of dowel bars after crack initiation.
To determine the shear transfer capacity of cement paste and aggregate interlock,
nineteen notch beams were tested. A corrugated notch was formed at mid-length of
standard prism beam with dimensions of 368 mm × 100 mm × 100 mm. The notch had
15 mm depth (Fig. 7-5). This ensured that the crack is initiated at the notch. An
external electrical strain gauge was installed above the notch to measure the maximum
induced flexural strain at the top edge of the notch (see Figure 7-5).
167
Figure 7-5. Instrumentation on the notch beam
Two aluminium plates were installed at both sides of the notch and in the same
distance away from centre line of the notch (see Figure 7-5). Distance from centre of
the notch to centre of aluminium plates in the longitudinal direction of the specimen
were considered to be 35 mm. Two linear displacement transducers (LDT) were
installed at these aluminium plates to measure vertical deflection at opposite sides of
the notch (Fig. 7-6).
The tests were performed using three point load configuration. However, the load was
applied on the right hand side of the notch and 35 mm away from centre line of the
notch. In other words, the load was applied directly above the aluminium plate
installed in the right hand side of the notch.
Figure 7-6 shows the test setup. Whilst the deflection under the load is slightly higher
than other locations of the specimen, crack initiation at the notch splits load deflection
curves recorded for points at the opposite sides of the notch (Fig. 7-7).
168
Figure 7-6. The test setup for notch beams
10.00
Deflection at the other side of the notch
8.00
Deflection under the load
Load (kN)
6.00
4.00
2.00
0.00
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16
Deflection (mm)
Figure 7-7. Typical result of the notch beam test
Figure 7-7 shows that a crack is initiated in the specimen when the applied load is
about 2 kN. However, the load-deflection curves split completely when the load
exceeds 3.52 kN. In other words, the crack starts to propagate above the notch after
this point. Table 7-5 shows the maximum shear stress at the notch before failure. This
169
information can be used in the finite element model to simplify analysis procedure. To
do this, the feature of ANSYS software for element birth and death can be applied.
When the magnitude of shear stress of the element exceeds the minimum shear
strength provided in Table 7-5, i.e. 0.4 MPa, the element will be considered a dead
element.
Table 7-5. Results of notch beam tests
Left Maximum Average

Date Shear Standard
Age Curing Load shear shear shear
of Name Area deviation
(day) Method (kN) force strength strength
test (mm2) (MPa)
(kN) (MPa) (MPa)
N1 10.38 4.15 0.49
06 / 12 /2005
Water cured
N2 10.05 4.02 0.47

N3 28 8.61 3.44 8500 0.41 0.45 0.032
N4 9.3 3.72 0.44
N5 9.57 3.83 0.45
N6 8.68 3.47 0.41
21 / 12 /2005
Water cured
N7 8.78 3.51 0.41

N8 28 9.65 3.86 8500 0.45 0.44 0.030
N9 10.2 4.08 0.48
N10 9.07 3.63 0.43
N11 9.94 3.98 0.47
N12 8.74 3.5 0.41
N13 9.15 3.66 0.43
17 / 05 /2006
Air cured
N14 9.04 3.62 0.43

N15 28 8.5 3.40 8500 0.4 0.40 0.042
N16 6.83 2.73 0.32

N17 8.62 3.45 0.41
N18 7.91 3.16 0.37
N19 7.95 3.18 0.37
7.6. Fatigue test

As mentioned in Chapter 2, the fatigue test is performed using three point loading
configuration together with one directional sinusoid cyclic load. However, the results
of the typical fatigue tests (S-N curve) may be inadequate for concrete pavement
design as the concrete pavements tend to curl upward during nighttime and downward
during daytime (see Section 2.6 of this thesis for further information). In other words,
170
the fatigue phenomenon due to daytime and nighttime curvatures is not considered in
the traditional concrete fatigue tests.
Furthermore, the variation of induced stress at a certain point within the concrete
pavement is greater than that considered in the fatigue test. This can be due to thermal
induced stress or shrinkage induced stress. Consequently, the induced tensile stress at
a certain point within the concrete pavement may vary between compressive stresses
and tensile stresses. This fact is not considered in the traditional laboratory fatigue test
where the top and the bottom surface layers are always in compression and tension,
respectively.
To overcome the aforementioned problems and to evaluate the accuracy of the

equations developed for estimation of concrete fatigue life (Equations 2-9 to 2-14 and
3-21 to 3-23), a specific test setup was designed in the current research and
manufactured at the QUT workshop. The setup (Fig. 7-8) contains three steel
rectangular rings with a sufficient stiffness to avoid elongation of the steel elements
under the applied load. Due to the space limitation in the setup of the fatigue machine,
dimensions considered are appropriate for concrete prism beams of 368 mm × 100
mm × 100 mm. Figure 7-9 shows a comparison of the newly developed fatigue setup
with the traditional method.
This particular setup enables the simulation of upward or downward curvature in the
concrete specimen based on certain values of differential temperature between the top
and the bottom surface layers of the specimen during day and night. To do this and to
assure that the induced stresses are in the range of the induced stresses in concrete
pavements, a concrete pavement with transverse and longitudinal joints was analysed
under the applied loads (temperature fluctuation and vehicular loads) using finite
element techniques. The magnitude of induced tensile stress due to combinations of
daytime and nighttime differential temperatures and vehicular loads was than derived
from the results of the finite element analysis.
171
Figure 7-8. The newly developed fatigue test setup
Figure 7-9. Comparison between the newly developed and the traditional fatigue test
setups
Typical results of finite element analysis were presented earlier in the Chapter 6. The
equivalent vehicular and environmental loads were then determined based on the
magnitudes of corresponding induced tensile stresses using Equation 7-2.
172
PE = 2.222σ t (7-2)
where PE (kN) is the equivalent load and σ t (MPa) is induced tensile stress derived
from finite element analysis. Equation 7-2 was developed for the concrete prism beam
with the aforementioned dimensions which was tested with three point loading
configuration. The distance between supports was 300 mm and the load was applied at
the mid-span.
Load repetitions during daytime and nighttime were then estimated using information
derived from weight-in-motion surveys. Since this information is not yet available in
Australia, presumptive traffic load distributions for urban and rural roads presented in
Austroads (2004b) were used. This was done by assuming that the load repetitions
during daytime and nighttime are the same.
The fatigue machine was programmed to increase the magnitude of the applied load to
the equivalent load of a certain nighttime differential temperature using ramp function.
The speed of load increment was considered to be 200 N/Sec. The downward direction
of the load was applied at this stage. This simulates the concrete pavement curvature
during nighttime (Figure 7-10a). The magnitude of load was then cyclically changed
between this value and the equivalent load due to combination of nighttime
temperature fluctuation and vehicular load. Since load frequency varies between 2 to
15 Hz (Monismith et al., 1998), a speed of 4.5 Hz was considered for cyclic loads. A
greater value of load frequency produced a vibration in the floor structure of the
building where the fatigue machine had been installed. This has been known as
resonance phenomenon and usually occurs when the load frequency is close to natural
frequency of the supporting structure.
The specimen was unloaded when number of load repetition is equal to the estimated
load repetition during nighttime. It was again loaded using ramp function but in the
opposite direction (upward) to simulate pavement curvature during daytime (Figure 7-
10b). The magnitude of load was then cyclically changed with a speed of 4.5 Hz
between this value and the equivalent load due to combination of daytime temperature
fluctuation and vehicular load. The specimen was again unloaded when number of
load repetition was equal to the estimated load repetition during daytime.
173
The above mentioned procedures continued until the failure of specimen. If the
specimen did not fail, its flexural strength was determined using three point loading
configuration.
Figure 7-10. Specimen curvature during the test
7.6.1. Testing procedure

The standard prism beams with dimensions of 368 mm × 100 mm × 100 mm were
used in the fatigue test. Four specimens were firstly tested using three point load
configuration to determine the flexural strength of the specimens. Results are
presented in Table 7-6.
Table 7-6. The average flexural strength of the specimens used in the fatigue test

fr (MPa)
Date of Specimen Specimen Curing strength strength deviation
F33 6.39
27 /10 /2006
Water cured
F34 6.304
91 6.61375 0.378286 5.993361
F35 6.615
F36 7.146
Using Equation 7-2, the equivalent loads for flexural failure of the specimens were
14.7 kN. Hence, the magnitude of the applied load (combination of vehicular and
environmental loads) was considered to be lower than the equivalent load for flexural
failure. The same vehicular load repetitions were examined during daytime and
nighttime. Two load repetitions were considered, 1885 and 3000 cycles. Nineteen
174
concrete prism beams were separately tested using the newly developed fatigue test
setup (Fig. 7-11). Table 7-7 shows variation of the applied load in the fatigue test. In
this table, loads with negative values represent the upward loading due to daytime
differential temperature.
Figure 7-11. Concrete fatigue testing
Since the number of specimens used in the current study was not adequate to cover a
wide range of temperature fluctuations as well as variations in vehicular loads, results
of current study are employed to evaluate the accuracy of the previously developed
equations in this field. As mentioned earlier in Chapter 2, several equations for
estimation of concrete fatigue life were developed (Equations 2-9 to 2-14 and 3-24 to
3-26). Table 7-8 compares results of these equations with the corresponding fatigue
life derived from laboratory tests. The equations developed for estimation of concrete
fatigue life are appropriate for a certain stress ratio which is not the case studied in the
current research. As a result, an average concrete fatigue life was determined for each
equation based on the minimum and maximum stress ratio provided in the first column
of Table 7-8. The estimated fatigue life was then compared with the results of the
current study (Table 7-8).
175
Table 7-7. Results of Fatigue tests
Load
Load
Variation
repetition Equivalent Equivalent
(kN)
load due load due to Fatigue
Specimen fr
to axle differential life
Nighttime
Nighttime
Daytime
Daytime
name (MPa)
group temperature (cycles)
load (kN) (kN)
-7.333 to -4.445
1
7.333 to 10.221
69252 N.A.
1885 1885 2.888 7.333 Not

2 7.3665
failed
Not
3 7.911
failed
4 5678 N.A.
5 4000 N.A.
--8.333 to -3.018
8.333 to 13.645
6 3008 N.A.
7 3000 3000 5.315 8.333 6886 N.A.
8 2 N.A.
9 3003 N.A.
10 323 N.A.
11 1503 N.A.
-8.333 to -4.333
8.333 to 12.333
12 2500 N.A.
3000 3000 4000 8.333
Not
13 7.794
failed
14 233263 N.A.
15 232851 N.A.
-8.333 to -5.333
8.333 to 11.333
16 1645 N.A.
17 3000 3000 3000 8.333 7926 N.A.
18 839714 N.A.
19 379866 N.A.
176
Table 7-8. Comparison between results of the current study and the equations developed for estimation of concrete fatigue life
Variation in maximum Average

Suh et al. Suh et al.
tensile stresses Vesic and results
Majidzadeh and Trybig et al. (2005) for (2005) for Austroads
during daytimes and Saxena Dater (1977) of
Ilves (1983) (1977) siliceous river crushed (2004)
nighttimes to concrete (1969) current
gravel limestone
flexural strength study
0.499 to 0.695 2296657 71994 146834 32245550 14056535 7.95E+08 399603 69252
0.567 to 0.771 1406852 160544 99439 2109653 2294407 62050647 38954 155572
0.567 to 0.839 1315521 150237 93021 2109148 2286070 62023477 38841 4515
0.567 to 0.928 1240829 142028 87380 2109115 2284597 62021262 38823 2002
177
The last column of Table 7-8 presents the average concrete fatigue life determined
from laboratory tests performed in the current study. In some laboratory tests, the
flexural strength of a certain specimen seems to be lower or higher than the average
flexural strength of the specimens considered in the fatigue test. Consequently, they
are not considered in the average fatigue life. These specimens were numbered as 2, 3,
8, 10,1, and 18 in Table 7-7.
Results indicate that the Austroads (2004) fatigue equation for a stress ratio between
0.499 to 0.695 and 0.567 to 0.771 is not sufficient. However, The equation developed
by Majidzadeh and Ilves (1983) is the best approach compared with other equations.
For estimation of concrete fatigue life when the stress ratio varies between 0.567 to
0.839 and 0.567 to 0.928, on the other hand, the Austroads fatigue equation results in a
better approach than those of other equations. Nevertheless, the estimated fatigue life
using Austroads method (2004) is respectively about ten times and 20 times greater
than those recorded in the current study using the newly developed fatigue test setup.
7.7. Summary
Structural properties of concrete including compressive strength, flexural strength,
modulus of elasticity, shear strength of aggregate interlock and cement paste and
finally fatigue life were determined using laboratory tests. Results of the laboratory
tests indicate that the concrete modulus of elasticity can be accurately estimated by
using equation provided by AS3600. Furthermore, the equations provided for
estimation of concrete flexural strength can be used for appropriately cured concrete.
In other words, the provided equations are not able to accurately estimate the flexural
strength of air cured concrete. Consequently, these equations are not adequate to be
used in concrete pavement unless a sufficient curing method is provided.
The shear transfer capability of aggregate interlock and cement paste was also
determined. This property can be used to accurately estimated the load transfer
efficiency across transverse and longitudinal joints. Ultimately, a new fatigue test
setup was developed to take into consideration the pavement curvature during daytime
and nighttime differential temperatures. Results of the fatigue tests show that the
equations for estimation of concrete fatigue life developed in the past are not
sufficient. Nevertheless, some of these equations are able to approximately estimate
178
the concrete fatigue life for a particular stress ratio. Noticeably, none of the equations
is able to accurately estimate the concrete fatigue life when greater stress ratio, i.e.
between 0.567 to 0.839, is applied on the specimens. For instance, the Austroads
(2004) estimation of fatigue life is about 10 times greater than that observed in the
current study.
Since there was not a sufficient number of specimens used in the current study, the
results are not fully conclusive. Hence, further studies are required to develop more
accurate equations for estimation of concrete flexural strength and concrete fatigue
life. The newly developed fatigue setup is strongly recommended to be used in the
future studies.
179
Chapter 8
DYNAMIC ANALYSIS OF BONDED CONCRETE

PAVEMENTS UNDER MOVING AXLE GROUP LOADS
8.1. Introduction
Until now, vehicular loading has commonly been modelled as static loads in concrete
pavement design guidelines although it has a dynamic nature. While information on
structural behaviour of jointed plain concrete pavement under static loads was
performed and presented in Chapter 6 of this thesis, the dynamic responses of two
common types of concrete pavement, JPCP and JRCP, are presented in this chapter.
Dynamic analysis of concrete pavements has been attracting researchers’ attentions for
quite a while. Research conducted on concrete pavements can be categorized as either
experimental test or analytical solution. The AASHO (1962) research on concrete
pavements showed that an increase in vehicle speed from 3.2 to 95.6 km/h decreases
the pavement responses by about 29 per cent. In contrast, Izquierdo et al. (1997) found
out in an experimental study on a concrete pavement resting on a subbase with low
stiffness under heavy truck loads that vehicle velocity may noticeably increase the base
deflections or stresses.
Information on analytical dynamic analysis of concrete pavements published in the

past showed that consideration of static vehicular loads results in greater induced
tensile stress in the concrete pavements. Whilst pavements are subjected to truck
loads, point and wheel loads have been widely used in most dynamic investigations to
simplify analysis procedure (for further information on dynamic analyses carried out in
the past see Section 3.4.4.1 of this thesis). Other dynamic studies on concrete
pavements have been also restricted to particular single axle or tandem axle loadings.
Nevertheless, structural dynamic responses of concrete pavements may be affected by

configuration, magnitude, frequency and location of applied loads. Furthermore,
effects of subbase, reinforcement, shoulder, and adjacent traffic lanes have not been
clearly described in dynamic studies. Consequently, further studies are required to
180
understand effects of other axle groups on concrete pavement dynamic response
(Izquierdo et al., 1997). In order to determine the effects of the above mentioned
parameters on dynamic response of concrete pavements, both JPCP and JRCP are
analysed using Finite Element techniques. The concrete pavements are subjected to
different transient axle group loads.
8.2. Methodology
It is known that the dynamic structural response depends on the ratio ( λ ) of load
frequency (ϖ ) to natural frequency of the structure ( ωi ). Several modal analyses were
thus carried out to determine the range of the fundamental natural frequency of bonded
concrete pavements namely, JPCP and JRCP (details of pavement modelling can be
found in Section 8.3). The results of modal analyses show a variation of 20 to 58 Hz in
the first mode frequency of concrete pavements. Furthermore, an increase in the
subgrade modulus of elasticity or a decrease in thickness of concrete slab or subbase
results in an increase of fundamental natural frequency of the pavement. For usual
subgrade strengths and vehicular load frequencies, this implies that the thicker the
concrete slab, the more sensitive it is to dynamic loads.
Evidence shows that the load frequency is between 0 and 20 Hz based on frame
bending vibration mode frequency for trailer and tractors (Gillespie et al., 1993), and
between 2 and 15 Hz based on truck’s suspension vibration frequency (Monismith et
al., 1998). Gillespie et al. (1993) also found that load frequency depends on truck
speed so that the load frequency increases from 4.6 Hz to 6.5 Hz when the truck speed
increases from 58 km/h to 82 km/h. Based on these data and assuming that a linear
relationship exists between vehicle speed and load frequency, ϖ may vary between
0.159 and 8.724 Hz for a speed between 2 km/h and 110 km/h.
This range may be attributed to the excitation of other shape modes of concrete
pavements depending on location of axle group loads upon pavements. As a result, λ
varies between 0 and 0.436 which indicates that dynamic analysis may result in greater
induced tensile stress than static loads in some circumstances. Surface roughness has
not been considered at this stage; however, evidence shows its significant effects on
the dynamic responses of concrete pavements (Liu et al. 2000, Liu and Gazis 1999).
181
8.3. Finite element model description
Two individual 3D finite element models with overall dimensions of 5100 × 18400
mm, as shown in Figure 8-1, were developed using ANSYS platform (version 10.0)
based on Austroads (2004) concrete pavement design guide to investigate structural
behaviour of bonded JPCP and JRCP. Each model contains three layers namely
concrete slab, cement-stabilized subbase and subgrade. Solid 64 with Drucker-Prager
property was used to simulate the behaviour of subgrade (see Section 3.2.1 of this
thesis for further information on subgrade idealizations).
Figure 8-1. Finite element model of the JPCP
Length and width of soil layer were considered to be 2 m and 1.5 m, respectively,
larger than length and width of the subbase layer. This assumption is needed in order
to avoid applying unnecessary boundary conditions on the side elements as it restricts
deformation of subgrade layer in longitudinal and transverse directions and may also
affect pavement responses. The subgrade layer has 500 mm thickness, modulus of
elasticity (ES) of 69 MPa, Poisson’s ratio (ν S ) of 0.4, cohesion (CIF) of 0.001 MPa and
angle of internal friction ( ϕ ) of 32° and is resting on a rigid layer.
182
Subbase layer with 150 mm thickness (ESB = 5000 MPa and ν SB = 0.15) was modelled
at top of the subgrade. Area of subbase was calculated based on width and length of
the concrete slab, width of transverse and longitudinal joints and width of shoulder. A
bonded interface was considered between subgrade and subbase. Single traffic lane
confined at one of its longitudinal edges by shoulder was modelled over the subbase.
Four and two slab panels were considered along the longitudinal direction for JPCP
and JRCP, respectively. For JPCP, the width and length of each slab panel were
considered to be 3600 mm and 4600 mm (Fig. 8-1), respectively; while for JRCP they
were 3600 mm and 9205 mm, respectively. The width of shoulder and thickness of the
concrete slab and shoulder were considered to be 1500 mm and 250 mm, respectively.
Conventional concrete with a compressive and a flexural tensile strengths of 32 and
5MPa (ESL = 28000 MPa and ν sl = 0.2) was used in the model.
A bonded interface between the concrete slabs and subbase was also considered. Solid
65, which is capable for cracking in tension and crushing in compression, was
employed to simulate structural behaviours of the concrete slabs and subbase. Beam
element with bending and shear capabilities was used to simulate dowels and tie bars.
Steel tie bars consisting of four 14mm diameter ( φ ) in JPCP and eight φ 14 mm in
JRCP with 1000 mm length were evenly placed along the longitudinal joints of each
individual slab panel. Twelve dowels of φ 32 mm were positioned at transverse joints.
Both dowels and tie bars were located at the middle of the slab thickness. Truss
element (link8 – 3D spar, one directional element with tensile capability) was used to
simulate reinforcement in the JRCP. A mesh reinforcement of φ 16 mm at 250 mm
intervals was modelled at middle of the slab thickness (Fig. 8-2). Saw cuts (Fig. 7-3)
with 5mm width and 50 mm depth were modelled at both longitudinal and transverse
joints. No boundary condition was applied on the side elements of the concrete slab
and subbase.
Diverse axle group types including SAST, SADT, TAST, TADT, TRDT and QADT
with average loads of 53 kN, 80 kN, 90 kN, 135 kN, 181 kN, and 221 kN were
separately applied upon the pavement and close to the confined longitudinal joint.
Different velocities (2, 45, 80, 110 km/h) were considered. Critical axle group
configurations based on the results of static analyses (see Chapter 5 of this thesis) were
employed in this research. A distance of 300 mm between the centres of dual tyres and
183
an axle length of 1800 mm were therefore assumed. Tyre pavement contact area was
assumed to be rectangular. Tyre inflation pressure was considered to be 750 kPa. Axle
group load was assumed to be equally distributed among axles and then among the
wheels. Table 8-1 presents a summary on axle configurations used in the current study.
Figure 8-2. Reinforcement simulation for JRCP in the finite element model
Figure 8-3. Simulation of saw cut in the finite element model
184
Table 8-1. Information on axle configurations used in the current study
Tyre-Pavement
Type of Axle Distance Number of Wheel contact area
Axle Load between tyres per Load Length Width
Group (kN) axles (mm) axle group (kN) (mm) (mm)
SAST 53 NA 2 26.5 225 157
SADT 80 NA 4 20 195 136
TAST 90 1100 4 22.5 207 145
TADT 135 1100 8 16.875 179 125
TRDT 181 1100 12 15.083 170 119
QADT 221 1100 16 13.813 162 114
Based on the work of Hendrick et al. (1992), wheel loads were then distributed among
the nodes at the top surface layer of the concrete slab representing the tyre-pavement
contact area. These nodal loads were then moved along longitudinal direction of the
pavements based on relevant time steps. Time step is the most significant parameter in
the transient moving analysis. It can affect the accuracy of the results. Hence, the value
of time step for each individual speed was calculated based on the ANSYS manual
recommendation (2006) (time step = 1 / (20 ϖ )). As a result, the value of time step
was considered to be 0.18, 0.008, 0.0045, or 0.0033 s for all axle groups with a speed
of 2 km/h, 45 km/h, 80 km/h or 110 km/h respectively.
8.4. Model calibration

To calibrate results of finite element analysis, a JPCP with two adjacent traffic lanes in
transverse direction and three concrete slab panels in the longitudinal direction was
modelled based on experimental research of Izquierdo et al. (1997). The width, length
and thickness of the concrete slabs were 3600 mm, 6000 mm and 244 mm
respectively. Concrete used in the field test had a compressive strength and a flexural
strength of 42 MPa and 4.5 MPa respectively. The modulus of elasticity (ESL) was
28714 MPa. The concrete slab was constructed over a subbase of hot asphalt with 125
mm thickness, E of 3000 MPa and a Poisson’s ratio (ν ) of 0.4. Based on the
information provided by Izquierdo et al. (1997) and results of Ping et al. (2002), the
subgrade was assumed to have an ES of 130 MPa, ϕ of 35°, CIF of 0.001 MPa and
ν S of 0.35. Seven tie bars of φ 16 mm were installed at each longitudinal joint and 14
dowels of φ 60 mm were installed at each transverse joint.
185
A heavy truck load with three axle groups including SAST, TADT and TAOT
(Tandem Axle Octa Tyre) was used in the original experiment conducted by Izquierdo
et al. (1997). The information on pavement response published by Izquierdo et al.
(1997) was restricted to the concrete slab deflection. Results of the static analysis (see
Chapter 5 of this thesis) performed in this research showed that the slab deflection is
affected by the total axle group weight rather than axle group configuration. Hence, the
TAOT was replaced with a QADT with the same weight. Information on modified
truck configuration is presented in Figure 8-4.
Figure 8-4. Truck configuration used for validation of the finite element analysis
A finite element model was developed based on the above mentioned information and
subjected to the aforementioned truck load with a speed of 48 km/h. The truck was
passed along the concrete pavement with a distance of 515 mm from longitudinal
joints.
Figure 8-5 shows a comparison between deflection influence line derived from the
finite element analysis used in the current research and the results of experimental test
recorded by Izquierdo et al. (1997) for a point at the middle of free longitudinal edge
of the centre slab, with a load speed of 48 km/h. The comparison demonstrates that the
current model can accurately simulate the real behaviour of concrete pavements.
Results also indicate that while distance between free edge of concrete slabs and free
edge of the model is large enough to produce negligible stress in elements located on
the boundary edges of the model, the use of larger subbase and applying no boundary
condition on the side elements of the model does not affect the accuracy of the results.
186
0.02
0
0 5 10 15 20 25 30 35
-0.02
Slab Deflection (mm)
-0.04
-0.06
Recorded Deflection
-0.08
Results of FEA
-0.1
-0.12
Distance (m)
Figure 8-5. Calibration of the finite element analysis
8.5. Effects of moving single axles on induced tensile stress

Results of the current study show that the speed has no significant effects on induced
tensile stress in the JPCP and JRCP due to SAST (Fig. 8-6) and SADT (Fig. 8-7).
However, a speed of 45 km/h creates slightly greater tensile stress in the pavements.
Maximum tensile stress occurs close to wheel path and at the bottom surface layer of
the concrete slab.
Results of the current research for induced tensile stress due to SAST fully support the
results of Chattie et al. (1994) and Liu and Gazis (1999) which indicated that static and
dynamic analyses produce comparable values.
However, a discrepancy in terms of variation of tensile stress with speed can be

observed between results of the current study for single axle loads with those published
by AASHO (1962) and Zaghloul and White (1993). These studies indicated that an
increase in vehicle speed rapidly decreases the value of longitudinal or transverse
stresses.
187
0.32
2 (km/h)
0.3 45 (km/h)
Maximum Tensile Stress (MPa)
SAST
80 (km/h)
0.28 110 (km/h)
0.26
0.24
Transverse Joint
Transverse Joint
0.22
0.2
6 7 8 9 10 11 12 13 14
Distance from Free Transverser Edge (m)
a) In JPCP
0.38
2 (km/h)
0.36
45 (km/h)
SAST
0.34 80 (km/h)
110 (km/h)
0.32
0.3
0.28
0.26
Transverse Joint
0.24
0.22
0.2
6 7 8 9 10 11 12 13 14
Distanse from Free Transverse Edge (m)
b) In JRCP
Figure 8-6. Influence line of induced tensile stress in concrete pavements due to SAST
188
0.45
2 (km/h)
0.43
45 (km/h)
Maximum Tensile Stress (MPa) 0.41 SADT 80 (km/h)
110 (km/h)
0.39
0.37
0.35
0.33
0.31
Transverse Joint
Transverse Joint
0.29
0.27
0.25
6 7 8 9 10 11 12 13 14
Distance from Free Transverse Edge (m)
a) In JPCP
0.5
2 (km/h)
0.45 45 (km/h)
SADT
Maimum Tensile Stress (MPa)
80 (km/h)
110 (km/h)
0.4
0.35
0.3
Transverse Joint
0.25
0.2
6 7 8 9 10 11 12 13 14
b) In JRCP
Figure 8-7. Influence line of induced tensile stress in concrete pavements due to SADT
189
Nodal stress monitoring in the current study reveals that variations of transverse or
longitudinal stresses do not always represent variation of the most critical stresses in a
concrete pavement due to a moving load, as higher tensile stresses with different
scenarios may exist elsewhere within the pavement and far away from the monitored
location.
Figure 8-8 compares the transverse tensile stress with the corresponding maximum
global (MG) stress at the same time step in the JPCP when subjected to moving TADT
load. A point at the bottom surface layer of confined longitudinal edge and close to
transverse joint was considered for monitoring the value of transverse tensile stress.
The MG stress is the maximum tensile stress that can be induced in concrete
pavements due to dynamic loads.
0.4
Induced tensile stress (MPa)
0.35
0.3
0.25
0.2
0.15
0.1
MG
0.05 Transverse tensile stress
0
0 20 40 60 80 100 120
Speed (km/h)
Figure 8-8. Comparison of transverse stress at confined edge and MG stress in JPCP
Location of the MG stress depends upon several factors such as speed, type and
configuration of axle groups and can occur anywhere within the pavements. Results
show that depending on the vehicle speed, the transverse stress may increase or
decrease.
190
The MG stress, on the other hand, does not change appreciably up to a speed of 80
km/h but thereafter increases with speed. This may be the reason why there have been
conflicting findings reported in the literature, i.e. some researchers reported that speed
decreased the value of transverse stresses whereas others found that speed increases
the value of MG stress. Therefore, in the subsequent sections, the terms “maximum
tensile stress” or “tensile stress” refers to the MG stress.
8.6. Effects of moving tandem axles on induced tensile stress

An increase in the speed of tandem axle group loads (TAST or TADT) up to 80 km/h
has no significant effect on the induced tensile stress in the JRCP. However a greater
speed (110 km/h) rapidly decreases the value of tensile stress in the JRCP by about 50
per cent. TAST with speed up to 80 km/h induces similar behaviour in JPCP.
However, a speed of 110 km/h strongly affects the pavement response so that the local
tensile stress of the pavement increases by about 11 per cent although the maximum
tensile stress is relatively similar to that of static analysis.
Figures 8-9 and 8-10 show the MG stress influence line for different speeds when a
TAST or a TADT is applied on the JPCP and JRCP. It can be seen that speed of
TADT significantly affects the induced tensile stresses in the JPCP so that a speed of
110 km/h increases the maximum induced tensile stress by about 12 per cent. The
value of local tensile stress due to TADT in some cases is about 58 per cent higher
than that of static loads, for example at 7.4 m from the free transverse edge. Speed
increment also affects the location of maximum tensile stresses. The critical location in
static analysis which is often located near transverse joints moves toward the mid-span
of the concrete slab. This finding is in contrast of Hendrick et al. (1992) who stated
that a region close to transverse joints experiences a severe cracking damage if a
constant moving load (see Section 2.5.4.1 of this thesis) is considered in the analysis. It
should also be noted that the results of the current study are not in agreement with
those published by Kim et al. (2002) as their investigation was based on longitudinal
tensile stress in JPCP. The discrepancy is attributed to a number of reasons as
described in the earlier Section 6.5.
191
0.27
2 (km/h)
0.26
45 (km/h)
TAST
80 (km/h)
0.25
110 (km/h)
0.24
0.23
Transverse Joint
0.22
Transverse Joint
0.21
0.2
6 7 8 9 10 11 12 13 14
a) In JPCP
0.55
Transverse Joint
0.5 2 (km/h)
45 (km/h)
0.45 80 (km/h)
110 (km/h)
0.4
0.35
0.3
0.25
0.2
6 7 8 9 10 11 12 13 14
b) In JRCP
Figure 8-9. Influence line of induced tensile stress in concrete pavements due to TAST
192
0.38
2 (km/h)
Transverse Joint
0.36 45 (km/h)
TADT 80 (km/h)
0.34 110 (km/h)
0.32
0.3
0.28
0.26
Transverse Joint
0.24
0.22
0.2
6 7 8 9 10 11 12 13 14
a) In JPCP
0.8
Transverse Joint
2 (km/h)
0.7
45 (km/h)
80 (km/h)
0.6 110 (km/h)
0.5
0.4
0.3
0.2
6 7 8 9 10 11 12 13 14
b) In JRCP
Figure 8-10. Influence line of induced tensile stress in concrete pavements due to
TADT
193
8.7. Effects of moving triple and quad axles on induced tensile stress
Speed of TRDT and QADT have different effects on induced tensile stress in JPCP
and JRCP. In JRCP, speed up to 45 km/h has no effect on the magnitude of tensile
stress regardless to axle group type. Increasing speed from 45 km/h to 80 km/h
decreases the value of tensile stresses by about 7 per cent. In contrast, a speed of 110
km/h results in greater tensile stresses in dynamic analysis than static analysis by about
100 per cent for TRDT and 150 per cent for QADT (Fig. 8-11b and Fig 8-12b).
Structural response of JPCP, on the other hand, is influenced by axle speeds. In

general, an increase in axle speed up to 80 km/h produces a greater induced tensile
stress than static analysis. Further increase in axle speed also results in greater induced
tensile stress in pavement than static analysis. Nevertheless, the magnitude of the
tensile stress is relatively lower than that produced by axle speed of 80 km/h.
Figure 8-11a and Figure 8-12a present results of the current study for JPCP subjected
to TRDT and QADT respectively. Results show that a speed of 80 km/h results in
greater tensile stress than that captured by static analysis by about 25 per cent and 40
per cent for TRDT and QADT, respectively.
8.8. Effects of axle speeds on slab deflection

Slab deflection, on the other hand, is affected by the speed of axle groups as
demonstrated by results of the current study presented in Figures 8-13 to 8-18. Results
indicate that an increase in the speed of axle groups induces a greater slab deflection
than that captured by static analysis. The minimum increase in the slab deflection is
about 3 per cent and occurs when the JPCP is subjected to SAST. The maximum
increase in the slab deflection, however, is about 26 per cent and occurs when the
JRCP is subjected to TADT. Note that the pavement deflection along longitudinal
joints is affected by the axle group type and axle speed.
Consequently, an increase in truck speed may accelerate the erosion of subbase or

subgrade materials along the transverse or longitudinal edges of the pavements as
erosion of subgrade or subbase materials is affected by slab deflection (Austroads,
2004).
194
0.33
Transverse Joint
2 (km/h)
45 (km/h)
0.31 TRDT 80 (km/h)
110 (km/h)
0.29
Transverse Joint
0.27
0.25
0.23
0.21
0.19
6 7 8 9 10 11 12 13 14
a) In JPCP
0.65
Transverse Joint
TRDT
0.55
2 (km/h)
45 (km/h)
80 (km/h)
0.45 110 (km/h)
0.35
0.25
0.15
6 7 8 9 10 11 12 13 14
b) In JRCP
TRDT
195
0.35
Transverse Joint
2 (km/h)
0.33
45 (km/h)
QADT
0.31 80 (km/h)
110 (km/h)
0.29
0.27
0.25
0.23
0.21
Transverse Joint
0.19
0.17
0.15
6 7 8 9 10 11 12 13 14
Distance frome Free Transverse Edge (m)
a) In JPCP
0.65
Transverse Joint
2 (km/h)
45 (km/h)
0.55 QADT 80 (km/h)

110 (km/h)
0.45
0.35
0.25
0.15
6 7 8 9 10 11 12 13 14
b) In JRCP
QADT
196
0.039
2 (km/h)
45 (km/h)
0.037 SAST 80 (km/h)
Maximum Slab Deflection (mm)
110 (km/h)
0.035
0.033
0.031
Transverse Joint
Transverse Joint
0.029
0.027
6 7 8 9 10 11 12 13 14
a) In JPCP
0.038
2 (km/h)
SAST 45 (km/h)
0.036 80 (km/h)
110 (km/h)
0.034
0.032
0.03
Transverse Joint
0.028
0.026
6 7 8 9 10 11 12 13 14
b) In JRCP
Figure 8-13. Influence line of slab deflection in concrete pavements due to SAST
197
0.06
2 (km/h)
45 (km/h)
SADT
Maimum Slab Deflection (mm) 0.055 80 (km/h)
110 (km/h)
0.05
0.045
Transverse Joint
Transverse Joint
0.04
0.035
6 7 8 9 10 11 12 13 14
a) In JPCP
0.06
2 (km/h)
45 (km/h)
SADT
0.055 80 (km/h)
110 (km/h)
0.05
0.045
Transverse Joint
0.04
0.035
6 7 8 9 10 11 12 13 14
b) In JRCP
Figure 8-14. Influence line of slab deflection in concrete pavements due to SADT
198
0.05
Transverse Joint
0.048 2 (km/h)
TAST 45 (km/h)
Maximum Slab Deflection (mm) 0.046
80 (km/h)
0.044 110 (km/h)
0.042
0.04
0.038
0.036
Transverse Joint
0.034
0.032
0.03
6 7 8 9 10 11 12 13 14
a) In JPCP
0.05
2 (km/h)
0.048
45 (km/h)
TAST
0.046 80 (km/h)
110 (km/h)
0.044
0.042
0.04
0.038
0.036
Transverse Joint
0.034
0.032
0.03
6 7 8 9 10 11 12 13 14
b) In JRCP
Figure 8-15. Influence line of slab deflection in concrete pavements due to TAST
199
0.075
Transverse Joint
2 (km/h)
45 (km/h)
0.07
TADT
Maximum Slab Deflection (mm) 80 (km/h)
110 (km/h)
0.065
0.06
0.055
Transverse Joint
0.05
0.045
6 7 8 9 10 11 12 13 14
a) In JPCP
0.075
Transverse Joint
2 (km/h)
TADT 45 (km/h)
0.07
80 (km/h)
Maximum Slab deflection (mm)
110 (km/h)
0.065
0.06
0.055
0.05
0.045
6 7 8 9 10 11 12 13 14
b) In JRCP
Figure 8-16. Influence line of slab deflection in concrete pavements due to TADT
200
0.071
Transverse Joint
2 (km/h)
0.069 TRDT 45 (km/h)
Maximum Slab Deflection (mm) 80 (km/h)
0.067 110 (km/h)
0.065
0.063
0.061
Transverse Joint
0.059
0.057
0.055
6 7 8 9 10 11 12 13 14
Distance from free transverse edge (m)
a) In JPCP
0.075
2 (km/h)
TRDT 45 (km/h)
80 (km/h)
0.07
110 (km/h)
0.065
0.06
Transverse Joint
0.055
0.05
6 7 8 9 10 11 12 13 14
b) In JRCP
Figure 8-17. Influence line of slab deflection in concrete pavements due to TRDT
201
0.07
Transverse Joint
0.068
Maximum Slab Deflection (mm) 0.066
0.064
0.062
0.06
0.058
0.056
2 (km/h)
Transverse Joint
0.054
45 (km/h)
0.052
QADT 80 (km/h)
0.05 110 (km/h)
0.048
6 7 8 9 10 11 12 13 14
Distance from Free Transverse Joint (m)
a) In JPCP
0.07
QADT 2 (km/h)
45 (km/h)
80 (km/h)
0.065 110 (km/h)
0.06
Transverse Joint
0.055
0.05
6 7 8 9 10 11 12 13 14
b) In JRCP
Figure 8-18. Influence line of slab deflection in concrete pavements due to QADT
202
8.9. Critical speed of axle groups and location of severe damage
As mentioned earlier, speed affects the magnitude and the location of maximum tensile
stresses in concrete pavements depending on axle group types. Table 8-2 presents type
of required analysis (static or dynamic) as well as the critical speed for each axle group
in JPCP. The locations which are prone to fatigue cracking based on the value of
tensile stress are also listed.
Table 8-2. Type of required analysis in JPCP for each axle group
Critical Global Local

Axle
axle dynamic dynamic Location of severe
group Required analysis
velocity amplification amplification damage
type
(km/h) (%) (%)
Static Yes NA NA NA
near transverse
SAST
joint
Dynamic Yes 45 6 6
Static Yes NA NA NA
near transverse
SADT
joint
Dynamic Yes 45 1.7 11.1
Static Yes NA NA NA
near transverse
TAST 45 0.4 0.4 joint or quarter
*
Dynamic Yes point of slab
110 -1.2 11.4
Static No
TADT midpoint of slab
Dynamic Yes 110 12.11 58
Static No
quarter point of
TRDT *
slab
Dynamic Yes 80 24 35
Static No
quarter point of
QADT *
slab
Dynamic Yes 80 41 45
* from transverse joint
Note that information provided in Table 8-2 is based on the MG induced tensile stress
in JPCP. Consideration of other stress types, i.e. principle, longitudinal or transverse
stress, may produce different results. For instance, Figure 8-19 shows the critical
speed for each individual axle groups based on the first principle stress.
203
0.35
First Principal Tensile Stress (MPa)

0.3
0.25
0.2
0.15
0.1
0 20 40 60 80 100 120
Axle Group Velocity (km/h)
SAST SADT TAST

TADT TRDT QADT
Figure 8-19. Critical axle group speed based on the first principle stress
Results indicate that the critical speed of SAST, SADT, TAST, TADT, TRDT and
QADT are 85 km/h, 45 km/h, 110 km/h, 110 km/h, 110 km/h and 45km/h respectively.
However, no significant increase in induced tensile stress is observed when TAST,
TRDT or QADT is investigated. In other words, the induced tensile stress in the JPCP
due to a speed of 110 km/h of TAST, TADT or QADT is slightly greater than that of
static loads.
Gillespie et al. (1993) indicated that dynamic analysis is not necessary for concrete
pavement analysis and design. However, the experimental and theoretical probability
studies showed that dynamic fatigue cracking is more severe than static fatigue
cracking. Results of the current study clearly show that dynamic analysis produces
higher tensile stresses (depending on speed and type of the axle groups) and slab
deflections in concrete pavements. Hence, dynamic fatigue cracking is always more
severe than static fatigue cracking. It should be noted that dynamic analysis was not
found to be essential when SAST, SADT, TAST or TADT is applied on the JRCP.
However, dynamic analysis of TRDT and QADT at a speed of 110 km/h significantly
204
increases the tensile stresses in the JRCP by about 100 and 150 per cent, respectively.
Areas close to transverse joints may thus be prone to severe fatigue cracking.
8.10. Effect of Reinforcement

While the provision of reinforcement has no effect on maximum slab deflection, it
strongly affects the maximum tensile stressing JRCP as slab length in JRCP is about
twice longer than that in JPCP. Figure 8-20 gives an example of the current study
results comparing the slab deflection due to SAST in JPCP and JRCP.
0.04
JRCP-45
0.038
SAST
JPCP-45
0.036
0.034
0.032
0.03
6 7 8 9 10 11 12 13 14
Figure 8-20. Comparison between slab deflection due to SAST in JPCP and JRCP
A comparison between maximum induced tensile stresses in JRCP and JPCP (Figures
8-21 to 8-26) shows that the induce tensile stresses in JRCP are 18 %, 15 %, 104 %, 97
%, 94 % and 85 % greater than induced tensile stresses in JPCP when SAST, SADT,
TAST, TADT, TRDT and QADT are respectively investigated. These increments also
depend on type, diameter and number of reinforcement bars in longitudinal and
transverse directions as well as their vertical locations in the concrete slab and length
to width ratio of concrete slab panel. The boundary condition (bonded or unbonded)
between the concrete slab and subbase and also between the subbase and subgrade can
change the location of the neutral axis of the total pavement and consequently increase
or decrease the value of induced tensile stresses in JRCP.
205
0.38
0.36
Maximum Tensile Stress (MPa) 0.34 JRCP-45

SAST
JPCP-45
0.32
0.3
0.28
0.26
0.24
0.22
0.2
6 7 8 9 10 11 12 13 14
Figure 8-21. Comparison between maximum induced tensile stresses due to SAST in
JPCP and JRCP
0.5
JRCP-45
0.45 SADT
JPCP-45
0.4
0.35
0.3
0.25
6 7 8 9 10 11 12 13 14
Figure 8-22. Comparison between maximum induced tensile stresses due to SADT in
JPCP and JRCP
206
0.55
0.5
JRCP-80
TAST
JPCP-80
0.45
0.4
0.35
0.3
0.25
0.2
6 7 8 9 10 11 12 13 14
Figure 8-23. Comparison between maximum induced tensile stresses due to TAST in
JPCP and JRCP
0.8
0.7 JRCP-2
JPCP-110
0.6
TADT
0.5
0.4
0.3
0.2
6 7 8 9 10 11 12 13 14
Figure 8-24. Comparison between maximum induced tensile stresses due to TADT in
JPCP and JRCP
207
0.65
0.6
Maximum Tensile Stress (MPa) 0.55 JRCP-110
0.5 JPCP-80
0.45
TRDT
0.4
0.35
0.3
0.25
0.2
0.15
6 7 8 9 10 11 12 13 14
Figure 8-25. Comparison between maximum induced tensile stresses due to TRDT in
JPCP and JRCP
0.65
0.6 JRCP-110
0.55 QADT
Maximum Tensile Stress(MPa)
JPCP-80
0.5
0.45
0.4
0.35
0.3
0.25
0.2
0.15
6 7 8 9 10 11 12 13 14
Figure 8-26. Comparison between maximum induced tensile stresses due to QADT in
JPCP and JRCP
208
8.11. Summary
Effects of different transient axle groups on dynamic response of the bonded JPCP and
JRCP in terms of maximum tensile stresses and deflections were described in this
chapter. In contrast with the findings of AASHTO (1962) and Zaghloul and White
(1993), results of the current study showed that dynamic analysis produces greater
tensile stress than static analysis. Furthermore, variations of transverse or longitudinal
stresses do not always represent variation of the most critical stresses in a concrete
pavement subjected to moving axle load.
In contrast with the findings of Hendrick et al. (1992), results of the current study also
showed that the use of a constant moving load does not always result in a severe
cracking damage close to transverse joints. Subsequently, critical locations for severe
fatigue cracking were determined. Damage location may be close to transverse joints,
at midpoint or in some cases at Quarter point of slab. Furthermore, fatigue cracking is
affected by axle group types and speed. Hence, Critical speed of those axle groups
with greater dynamic amplification was presented. A dynamic amplification between
1.7% and 58% may be produced in the concrete pavements. Results of the present
study also showed that dynamic analysis is required to accurately predict failure mode
of the concrete pavements, especially in JPCP.
Further studies are needed to determine the effects of daytime and nighttime
temperature gradient, loss of moisture content, transfer devices, width of joints,
unbonded boundary condition between concrete slab and subbase, surface roughness,
traffic wander, length to with ratio of the concrete slab panel, and width of longitudinal
and transverse joints on dynamic structural response of different concrete pavements
under diverse transient axle group loads.
209
Chapter 9
EXPERIMENTAL FIELD TEST OF CONCRETE

PAVEMENTS UNDER MOVING TRUCK LOADS
9.1. Introduction
Deteriorations of concrete pavements can normally be predicted based on the structural
response of the pavements under the applied loads. While considerable knowledge of
pavement behaviour under static loads is available word-wide, only very limited
number of studies have been carried out in the past to determine the effect of dynamic
loads on concrete pavement deteriorations. Hence, opinions differ as to which type of
load (static or dynamic) results in greater values of flexural stress.
Analytical dynamic analysis of bonded JPCP and JRCP under different moving axle
group loads performed in the current study (see Chapter 8 of this thesis) indicates that
vehicle speed has a significant effect on pavement responses even though a smooth
surface is considered. Furthermore, diagonal, corner, and transverse cracking may be
addressed by consideration of vehicle speed. However, no recent experimental test on
dynamic responses of concrete pavements has been conducted to demonstrate the
effect of heavy vehicle velocity on concrete pavement damage. Moreover, information
on experimental field tests of unbonded concrete pavements has not been
comprehensively described in the published documents to be used for calibration of
finite element model of unbonded concrete pavements subjected to moving loads.
In order to address the aforementioned problems, a fully instrumented concrete

pavement test section consisting of two concrete pavements namely, JPCP and JRCP
was constructed and tested under either quasi-static or dynamic truck loads. This
chapter describes the test procedure and presents significant outcomes of the test.
Truck load was allowed to wander on top of the instrumented pavement at

predetermined locations. Nominal speeds from 5 km/h to 55 km/h were considered in
the study. Various devices including strain gauges, displacement transducers, vertical
210
accelerometers and thermocouples were installed at different depths along the test
section. A total of 5184 time history responses of the test section were recorded.
9.2. Project Description

An experimental work on concrete pavement performance under dynamic truck
loading was conducted by Queensland University of Technology (QUT) and a major
Australian concrete producer, Rinker Australia, at Rinker sand quarry in Oxley Creek,
southwest of Brisbane. This location was selected since a weighbridge was available to
provide data on truck loads and the number and type of axle groups which can be
recorded for long term pavement performance monitoring. Furthermore, geotechnical
information on subgrade properties including soil classification, soil profile and
texture, bulk density, the Atterberg limits, and CBR, which had been derived from
surface and depth explorations of the site in 2001, is available in Readymix archive at
Milton Branch.
The test section has 32 m length, 5.1 m width and 250 mm thickness. It consists of two
JPCP and two JRCP which have been constructed over a 150 mm concrete subbase
resting on a stiff subgrade (CBR = 14%). The widths and lengths of the concrete slabs
are 3.6 m and 4.6 m for JPCP and 3.6 m and 10 m for JRCP, respectively. The
concrete subbase (32.5 x 5.5 m) was constructed on top of a sand layer with maximum
aggregate size of 3 mm. The thickness of the sand layer was about 15 mm. The sand
layer was placed on the subgrade layer to firstly create a separation between subbase
and subgrade and secondly, to develop a level platform for subbase. The subbase was
left to shrink for one week before constructing the concrete slabs and shoulder.
Figure 9-1 shows the layout of the test section. An additional JPCP section (1.5 m x
3.6 m) is placed at each longitudinal end of the test section to restrain the free
transverse edges and to simulate the conditions of a long stretch of the pavement.
A mesh reinforcement layer of round bar with a diameter of 9 mm was used in the
JRCPs (Fig. 9-2). Two vertical locations of reinforcement namely close to the top
surface layer of the concrete slab (50 mm far from the top surface layer) and close to
the bottom surface layer of the concrete slab (200 mm far from the top surface layer)
were considered in this study to determine the effects of reinforcement location on the
pavement responses and consequently the damage progress.
211
Figure 9-1. Layout of the test section
Voids with 250 mm length, 150 mm width and 250 mm height were formed by
inserting expanded polystyrene blocks at the intersection between the transverse and
the longitudinal joints along the confined edge of the pavement to install linear
displacement sensors (Fig. 9-3). Steel sections (25 mm x 25 mm x 3 mm), 1500 mm
long each, were driven into the subgrade as appropriate and used as mounting poles of
the displacement sensors (see Figure 9-4). Only 350 mm of the total length of these
steel poles protruded above the subgrade surface layer. The centre of the mounting
pole’s cross section was located 20 mm away from free edge and at the centre of void
formed along the confined edge.
Since there is a high contact stress between loaded round dowels and concrete
(Shoukry et al., 2002), flat plate dowels that are widely used in industrial places were
employed for this study to examine effect of dowel locations in depth of the concrete
slab. This consideration may result in lower induced contact stress and higher slab
deflection than the common round dowels as the bending and shear stiffness of the use
of common round dowels are relatively higher than flat dowels. Although no particular
instrument was assembled in the test section to measure the contact stress between
dowels and concrete, the long term performance of the test section under moving truck
loads can examine this assumption.
Each transverse joint was dowelled by eight flat plate dowels (300 mm x 50 mm x 6
mm). Since one side of the dowel was coated by a PVC sleeve, longitudinal
movements of the concrete slabs on both sides of the transverse joint were not
restrained (Fig. 9-5). Dowels were vertically positioned at three different depths to
212
determine the effects of dowel positions on load transfer efficiency (LTE) of joints and
pavement performance. The locations of the dowels (measured from the top of the
concrete slab) were 55 mm, 125 mm (at mid-depth) and 200 mm (see Figure 9-1).
Figure 9-2. Mesh reinforcement used in JRCPs
Figure 9-3. Expanded polystyrene blocks used to form voids at transverse and
longitudinal intersection
213
Figure 9-4. Assemblage of linear displacement transducer on the mounting pole
Figure 9-5. Rectangular dowel bars installed at transverse joints
214
One of the longitudinal edges of the test section was confined by a shoulder. Hence,
round tie bars (12 mm Ø, 1000 mm long) were positioned at mid-depth of longitudinal
joints. Four tie bars were used in each JPCP and eight in each JRCP (Fig. 9-6).
Figure 9-6. Cylindrical tie bars used at longitudinal joints
As mentioned in Chapter 3, a friction reducer layer may be placed between concrete

slab and subbase to eliminate the early age cracking in concrete pavements.
Consequently, bonded, unbonded, and partially bonded boundary conditions between
concrete slab and subbase may be created depending on the value of the friction
coefficient between these layers.
While most analytical research conduced in the past was based on fully bonded or fully
unbonded boundary condition between concrete slab and subbase (Heath and Roesler,
1999), feasible debonding layers considered in concrete pavement constructions result
in partially bonded boundary condition between concrete slab and subbase (Zhang and
Li, 2001). In other words, particular devices such as anchor or shear stud shall be
considered to create a fully bonded boundary condition. Tarr et al. (1999) found that
215
unbonded conditions could only be achieved by using a double layer of polyethylene
sheets.
In contrast, Yu et al. (1998) stated that friction between concrete slab and subbase is
sufficient to produce bonded behaviour even if polyethylene sheets are placed between
them. As structural responses of concrete pavements to vehicular load are highly
affected by boundary condition between concrete slab and subbase (Tarr et al., 1999),
half of the test section incorporated a single layer polyethylene sheet between the slab
and subbase (see Figure 9-1) to determine the effects of partially bonded and partially
unbonded boundary conditions on dynamic responses of the concrete pavements (Fig.
9-7).
Figure 9-7. The use of single layer polyethylene sheet to create partially bonded
boundary condition in half length of the test section
Evaporation retardant was poured at the top surface of the concrete slab during
levelling to protect the concrete slabs against plastic shrinkage. To examine the results
of analytical dynamic analyses of the concrete pavements described in Chapter 8 of
216
this thesis, the surface was subsequently floated by a power trowel to enhance the
surface smoothness (Fig. 9-8).
Figure 9-8. The use of power trowel to enhance the surface smoothness
Zollinger et al. (1994) indicated that early-age sawing methods with sawing depths less
than 0.25d (d is slab depth), should provide better crack control than conventional
methods with depths of 0.25d or 0.33d. Sawing sooner with early-age saws can take
advantage of larger changes in the concrete's surface moisture content or surface
temperature, which has been shown to induce cracking (Okamoto et al., 1994).
Therefore, transverse joints were prepared using soft sawing method three hours after
initial set (Fig. 9-9). The width and depth of the saw cuts were 10 mm and 50 mm,
respectively. The width of the saw cut joints allowed easy installation of
instrumentation wires across the test section.
An unreinforced shoulder with 1.5 m width and 250 mm thickness was poured about
15 hours after constructing the concrete slabs (see Figure 9-1). It contained five
dowelled transverse joints. Four flat plate steel dowels were installed at each transverse
joint. Dowel dimensions are similar to those used in the concrete slabs. As a problem
217
arose with sawing machinery, transverse joints of shoulder were saw cut 36 hours after
initial set.
Figure 9-9. Preparing the transverse joints using saw cut
9.3. Instrumentations
A total of 120 electrical gauges including 120Ω electrical strain gauges (ESGs, Fig. 9-
10), linear displacement transducers (LDTs, see Figure 9-4) and strain gauge based
vertical accelerometers were used to investigate the structural response of the test
pavement under either static or dynamic loads.
Since recent research (Choubane and Tia 1995, Heath and Roesler 1999) showed a
strong interrelationship between temperature gradients and damage potential of
concrete slabs, four thermocouples were evenly installed at different depth within the
concrete slabs (Fig. 9-11). Recording of temperature gradients was started 24 hours
after the initial set.
218
Figure 9-10. Embedded electrical strain gauges installed at a distance of 25 mm away
from the subbase surface using rebar chair
Figure 9-11. Thermocouples installed at different depth of the concrete slabs
219
Two types of ESGs, namely embedded and standard (glued to surface layer of
structure), were used. Embedded strain gauges should be fully covered by concrete to
accurately measure the induced strains in the concrete. Hence, they were installed at a
depth of 225 mm far from the top surface layer of the concrete slab using a rebar chair
(see Figure 9-10). The locations of the strain gauges are shown in Figure 9-12 while
those of the LDTs and accelerometers are shown in Figure 9-13. The main reason for
installation of LDTs was to determine the magnitude of LTE at transverse joints.
Nevertheless, three LDTs were installed at free longitudinal edge of the pavement to
estimate effect of truck speed on erosion of subbase and subgrade materials. None of
the LTDs was installed close to the beginning and ending transverse joints or in those
JPCP having 1.4 m length as load transfer efficiency of transverse joints may be
affected by the length of these pavements.
Figure 9-12. Locations of strain gauges

An eDaQ dynamic data acquisition system with 48 channels including 16 specific
channels for strain gauges and 32 general purpose inputs was used to record time
history responses of each individual instrumented point of the test section (Fig 9-14).
InField analysis software developed by eDaQ Company was utilised for data
processing.
Since the eDaQ data acquisition system has only 48 channels, three different recording
setups were utilised. Each setup utilised 32 switchable channels and 16 dedicated
220
channels. Half of the switchable channels were always connected to 10 ESGs and 6
LDTs to provide benchmark readings.
Figure 9-13. Locations of linear displacement sensors and vertical accelerometers
Figure 9-14. eDaQ dynamic data acquisition system
221
9.4. Material Properties
The subgrade soil texture was classified as silty clay loam with a loose and compacted
bulk density of 1.2 and 2.18 t m 3 (AS1141.4) respectively. The maximum dry density
of the soil was 1.86 t m 3 . Particle size distribution (AS1141.12) showed that 70.7 per
cent of aggregate was finer than 0.075 mm. Liquid limit, plastic limit and plasticity
index of the fines were 22.8, 14 and 8.8 per cent, respectively. Subgrade CBR was 14
per cent. The average 28-day concrete compressive and flexural strengths were
7.3MPa and 1.55MPa for subbase, 50.5MPa and 5.45MPa for slabs and 38.5MPa and
4.1MPa for shoulder, respectively.
9.5. Visual Monitoring of the Test Section

The test section was closely monitored for the first 28 days after casting the concrete
slab. The first top-down crack occurred within 2 days after casting at the middle of
JRCP where mesh reinforcement layer had been positioned close to the top surface
layer of the slab (Fig. 9-15). At the same time, two random surface cracks of about 150
mm long were also noticed in the same concrete slab. This part of the test section had a
partially bonded interface between concrete slab and subbase. All the above mentioned
cracks were very tinny in their width. Other concrete slabs showed no cracks. Crazing
cracks were observed in all concrete slabs one week after initial concrete set
suggesting inadequate curing (Fig. 9-16). This may also be due to the use of the
evaporation retardant. Apart from being unsightly, these cracks do not affect the
structural integrity of the concrete slabs.
Figure 9-15. Locations of cracks in the test section
Two shrinkage top-down transverse cracks (see Figure 9-15) were initiated close to
transverse joints in the shoulder within 36 hours after casting (Figures 9-17).
Transverse joints in the shoulder were sawed after the cracks had appeared. Another
transverse crack occurred 2 weeks after casting in the depth of shoulder and at the
222
middle length of it where shoulder had 10 m length with a partially bonded interface
between slab and subbase.
Figure 9-16. Initiation of Crazing cracks at the top surface layer of the concrete slabs
Figure 9-17. Initiation of a transverse crack in shoulder close to transverse joints
223
During this period, two small diagonal cracks (100 mm long) were also initiated in the
concrete slab close to the transverse joints. No additional visible cracks were noticed in
the concrete slabs or shoulder during the first 28 days. However, the width of one of
the transverse cracks in the shoulder increased by about 2 mm (see Figure 9-15). This
may decrease the load transfer efficiency of transverse joint in the shoulder due to
possible corrosion of dowels and consequently, results in joint faulting. However, this
can not affect the result of this study as truck loads are restricted to only pass along the
traffic lane. As mentioned in Section 8.2, the subbase was left to shrink for one week
before constructing the concrete slab and shoulder. Hence, several transverse,
longitudinal and diagonal cracks were initiated and propagated in subbase due to
environmental forces.
9.6. Truck Characteristics, Movement and Speed

A semi-trailer truck (Fig 9-18) with a gross weight of 477.3 kN was used to apply
traffic load in one direction. Truck contains three different axle groups namely SAST,
TADT and TRDT of 60.6 kN, 206.2 kN and 210.5 kN respectively. Tyre inflation
pressure in all tyres was set to be 750 kPa.
Figure 9-18. Semi-trailer truck used in this study
224
Exact measurement of tyre contact area could be done using methods such as the
multiple overlay technique (Sharma and Pandey, 1996). However, for simplicity, in
this study the contact area was determined by measuring the size of an imprint left by
the tyre on top of the slab after spraying paint around the tyre. Information on truck
configuration and tyre pavement contact area are shown in Figure 9-19.
Figure 9-19. Tyre configuration
Three longitudinal coloured lines were drawn at different locations on top of the
concrete bases to help driver maintain the truck movement at a certain distance from
the longitudinal joints (Fig 9-20). These include a red line close to the free longitudinal
edge of the pavement, a blue line close to the confined longitudinal joints of the test
section and a yellow line between them to symmetrically apply the truck loading on
both sides of the centre line of the traffic lane. The truck was driven along the
aforementioned lines at various nominal speeds including 5, 20, 35 and 55 km/h.
Higher speeds could not be achieved in this study as they would need a longer
acceleration distance. Pavement time history responses under moving truck load were
recorded thrice for each individual speed and position of the applied load to accurately
determine the structural responses of the test section. In total, 5184 time history
responses of the test section were recorded. Real truck speeds for each individual
channel were finally calculated based on the configurations and distance between axle
groups and pavement time history responses.
9.7. Pavement roughness

Since dynamic response of concrete pavements is affected by the surface roughness
(Bhatti and Stoner 1998, Liu and Gazis 1999), the surface profiles along predetermined
225
coloured lines were measured by the use of Face Dipstick 2200 Floor Profiler. Results
indicated that average elevation between adjacent points of 300 mm interval is about
0.55 mm.
Figure 9-20. Longitudinal coloured lines to help driver for maintaining the truck
movement at a certain distance of longitudinal edges

Differential temperature gradients and loss of moisture content through depth of the
concrete slab can significantly change the pavement response. However, there is no
information in the literature to show how these parameters affect dynamic responses of
the concrete pavements. Since variation in ambient temperature during the tests was
very small (less than 1ºC), a constant curling condition can be assumed in the analysis
of the results. Nevertheless, a finite element analysis approach may help to develop a
better understanding of the aforementioned parameter effects on dynamic response of
the concrete pavements. These are explained in Chapter 11 of this thesis.
InField analysis software was used to develop time history responses of the concrete
slabs, JPCP and JRCP, under moving truck load for different locations within the test
section. Using Microsoft Excel, results were then redrawn to appropriate scales for
226
comparison. The dynamic amplification (DA) was then calculated for each individual
channel using Equation 8-1.
  Dynamic Rresponse  
DA =    − 1 × 100 (9-1)
  Static Re sponse  
  
The DA varies with truck speed. However, the maximum or minimum captured DA
will be presented and discussed in this chapter. Results can be summarized as follows:
9.8.1. Concrete slab deflection

A comparison between slab deflections at the corner and at the mid-length of the free
edge was firstly done for results validation as corner deflection was expected to be
greater than other deflections. Furthermore, for subsequent validation processes, it was
also expected that slab deflections along a free edge would be higher than the
corresponding deflections along a confined edge. Results show that slab deflection
decreases from the corner of the free edge towards the mid-span and the confined edge.
Slab deflection at the corner is about 60 per cent greater than those at the middle of the
free edge.
Furthermore, results show that concrete slab deflection is strongly affected by truck
speed. The dynamic amplification varies between 55 per cent and 313 per cent
depending on the pavement type, interface between concrete slab and subbase and
location of measurement. Greater dynamic amplifications occur along the confined
longitudinal edge of the test section though the slab deflection values of these points
are relatively lower than those along the free longitudinal edge. Figure 9-21, as an
example of the current study outputs, shows time history of slab deflections for
different speeds at the corner of free longitudinal edge (DL7, see Figure 9-13) in
JPCP. The critical truck speed, which creates maximum slab deflection, depends on
several factors such as the location of measurement and the type of concrete slab.
Hence, medium speed in some cases results in greater slab deflection (Fig. 9-22).
The vertical location of the reinforcement layer also affects slab deflection. Results of
the current study show that slab deflections in JRCP where reinforcement was located
close to the top surface layer of the concrete slab is about twice the values from other
227
JRCP where reinforcement was located close to the bottom surface layer of the
concrete slab (Fig. 9-23).
Figure 9-21. Time history deflection responses for different speeds at DL7
Figure 9-22. Time history deflection responses for different speeds at DR13
228
Figure 9-23. Time history deflection responses in JRCP for different reinforcement
locations
Study of dowel performance under moving truck load indicates that load transfer
devices at transverse joints exhibit a stiffer behaviour under dynamic load (higher
speed) than quasi-static load (lower speed). In other words, the value of load transfer
efficiency of transverse joints under dynamic load is slightly greater than static load.
Results also indicate that the value of LTE in transverse joints under truck loads is not
constant and depends on type of axle groups, the applied load and truck speed (Fig. 9-
24).
A comparison between time histories of slab deflections at the corner of the confined
edge (Fig. 9-25) shows the importance of dowel position in depth of concrete slab. The
slab deflection under TADT and TRDT significantly decreases when dowels are
positioned at the mid-depth of the concrete slab. With dowels placed close to the top of
the concrete slab, on the other hand, lower slab deflection results under SAST loading.
Hence, for the flat dowels used in this work, the best dowel location would be at, if not
slightly above, the middle of the concrete slab depth. It is important to note that none
of the LTDs was installed close to the beginning and ending transverse joints or in
those JPCP having 1.4 m length (see Figure 9-1).
229
Figure 9-24. Time history deflection responses at transverse joint (truck speed 49
km/h)
Figure 9-25. Comparison between dowel positions based on critical speed
9.8.2. Induced tensile stress

Tensile stresses are also affected by truck speed. Dynamic amplification of tensile
stresses varies between -10.8 and +108.9 per cent. A small number of recorded stress
time histories shows a decrease in the magnitude of tensile stresses when truck speed
230
increases. Most time history stress responses suggest a need to consider dynamic
loading in concrete pavement design. Figure 9-26 is an example of recorded stress
time history where dynamic loads result in greater stress values than static loads.
Figure 9-26. Time history stress responses in JRCP at TCL12 for different truck speeds
Although dynamic amplification shows the significance of dynamic analysis, it does

not mean that the location where tensile stress is greater than that of other locations has
a greater dynamic amplification. For instance, the maximum dynamic amplification at
TCL12 (Fig. 9-26) is about 3½ times more than the maximum dynamic amplification
at TCL8 (Fig. 9-27), however, the maximum tensile stress for each individual speed at
TCL8 is relatively greater than those at TCL12.
Further observations can be made when the location of TCL12 and TCL8 are taken
into account (see Figure 9-12). Both strain gauges, TCL12 and TCL8, were installed
close to transverse joints, in the same distance from free edge and at the top surface
layer of the JRCP where reinforcement was located close to the bottom of the slab.
Boundary conditions between concrete slab and subbase for both points were similar.
However, dowels were located close to the top surface layer of the concrete slab for
TCL12 and at the mid-depth of the concrete slab for TCL8. A comparison between
maximum induced tensile stresses at TCL12 and TCL8 for each individual speed
indicates that tensile stresses at transverse joints increase by 87.5, 9.4, 45.1 and 6.1 per
cent with truck speeds (real truck speeds) of 5 (4.8), 20 (16.5), 35 (32.2) and 55 (44.3)
km/h, respectively, when dowels are located at the mid-depth of the concrete slab.
231
Figure 9-27. Time history stress responses in JRCP at TCL8 for different truck speeds
Tensile stresses in both JRCPs are greater than those in JPCP. While the difference in
panel lengths may have contributed to the results, the reinforcement may also have
some effects. The length of uncracked concrete slab panel in JRCP is about twice the
length of concrete slab panel in JPCP. No crack was observed in JPCP and that JRCP
where reinforcement was positioned close to the bottom surface layer of the concrete
slab. However, crack initiation and propagation in the concrete slab can significantly
affect this assumption. Commonly, the recommended position of the longitudinal steel
is between 1/3 and 1/2 of the slab depth as measured from the surface. However,
effects of reinforcement location on pavement dynamic tensile stresses in the current
study are still unclear at this stage as analyses of time history responses have not lead
to a specific conclusion.
9.8.3. Vertical acceleration in concrete slabs

Results of the current study indicate that vertical acceleration in concrete pavements
depends on the distance between transverse joints, boundary conditions between slab
and subbase, provision of shoulder or adjacent traffic lane, and traffic wander. An
increase in length of concrete slab panel or the use of bonded interface between
concrete slab and subbase can increase the vertical acceleration of the slab. This
acceleration also increases when the moving load is applied close to free longitudinal
232
edge of the pavement. On the other hand, provision of shoulder or adjacent traffic lane
decreases the vertical acceleration in the concrete slabs.
Results of the current study indicate that the absolute concrete slab vertical
acceleration varies between 0.001 g and 0.62 g, in proportion with the nominal truck
speed ranging between 5 km/h and 55 km/h. It is known that acceleration and speed of
structural deflection induce dynamic forces in a structure. This dynamic force may
increase or decrease at the certain time depending on the magnitudes of acceleration
and speed. Consequently, at certain location, the tensile stress in the experimental
concrete pavement presented in this study may either increase or decrease due to the
dynamic loading effect. In addition, slab curling and warping can decrease or increase
this effect. Their contribution to the results needs further study which may be done by
the use of finite element techniques. It is presented in the Chapter 9 of this thesis.
9.9. Temperature fluctuation

Analysis of the recorded temperature fluctuations in depth of the concrete slab shows
that the maximum differential temperature between the top and the bottom surface
layers of the concrete slab occurs at about 3 pm for daytime differential temperature
and at about 6:30 am for nighttime differential temperature. Figure 9-28 presents
variations of temperature in depth of the concrete slab. Since the thermocouples were
installed in a distance of 25 mm far from the top and the bottom surface layers of the
concrete slab, the differential temperature between the top and the bottom surface
layer of the concrete slab is greater than that showed in Figure 9-28. The magnitudes
of daytime and nighttime differential temperatures are about 12 ºC (temperature at the
top surface layer minus temperature at the bottom surface layer) and 8 ºC (temperature
at the bottom surface layer minus temperature at the top surface layer), respectively.
Since the aforementioned differential temperature applied to a certain location in the

Brisbane area, further study is required to clearly address daytime and nighttime
differential temperature gradients for different states of Australia.
233
55
Close to the top surface layer of the concrete slab

50
Temprature (º C)
45
40
35
30
50.00 100.00 150.00 200.00
Concrete slab depth (mm)
12:50:00 AM 1:50:00 AM 2:50:00 AM 3:50:00 AM 4:50:00 AM 5:50:00 AM
6:50:00 AM 7:50:00 AM 8:50:00 AM 9:50:00 AM 10:50:00 AM 11:50:00 AM
12:50:00 PM 1:50:00 PM 2:50:00 PM 3:50:00 PM 4:50:00 PM 5:50:00 PM
6:50:00 PM 7:50:00 PM 8:50:00 PM 9:50:00 PM 10:50:00 PM 11:50:00 PM
Figure 9-28. Typical temperature fluctuation in depth of the concrete slab
9.10. Summary
A fully instrumented rigid pavement test section including JPCP and JRCP was
constructed and tested under quasi-static and dynamic truck loading. Information on
the test section, instrumentation layout, material properties and truck characteristics
were described in this chapter.
Pavement performance under environmental conditions was studied during the first 28
days after casting. Truck loading was subsequently applied at different locations of the
pavement. Time history responses were recorded for nominal truck speeds between 5
km/h to 55 km/h.
Investigation of the recorded time history responses of the test section indicates the
importance of dynamic analysis in pavement design. Results also indicate that dowel
position can strongly influence the pavement responses. Furthermore, the slab
deflection in JRCP decreases when reinforcement was located close to the bottom
surface layer of the concrete slab. However, further studies are needed to determine
effects of reinforcement position on induced dynamic stresses of the pavement.
Typical differential temperature between the top and the bottom surface layer of the
concrete slab during daytime and nighttime were determined. Nevertheless, further
234
study is required to clearly address the temperature fluctuations within depth of the
concrete slab in different states of Australia. Since variation in subgrade property,
differential temperature gradients and loss of moisture contents in the depth of the
concrete slab may influence dynamic responses of the concrete pavement, finite
element analysis approaches shall be carried out to address the variation in these
parameters on dynamic responses of concrete pavements. Some of these are addressed
in Chapter 10 of this thesis.
235
Chapter 10
DYNAMIC ANALYSIS OF UNBONDED CONCRETE

PAVEMENTS UNDER MOVING AXLE GROUP LOADS
10.1. Introduction
Results of dynamic analysis of bonded pavement presented in Chapter 8 showed that
dynamic loads produce greater responses than static loads. These findings were also
compatible with results of the field test explained in Chapter 9. In Australia, concrete
pavements are constructed by placing a debonding layer between concrete slab and
subbase to eliminate the early age cracking. This produces a partially bonded boundary
condition between concrete slab and subbase.
Results of the field test described in Chapter 9 also show that unbonded concrete
pavements experience greater stresses in the presence of dynamic loads than static
loads. However, further study is required to determine the dynamic amplification for
each individual axle group. Hence, the information provided in Chapters 7 and 9 is
used in this chapter to study how speed of axle group affects responses of unbonded
concrete pavements.
10.2. Development of finite element model

Similarly as in Chapter 8, two individual 3D finite element models were developed
using ANSYS platform (version 10.0) to study the behaviour of unbonded concrete
pavements including JPCP and JRCP under dynamic loads. Overall dimensions of the
concrete slabs and shoulder were 5100 × 13800 mm for JPCP and 5100 × 18400 mm
for JRCP. The reinforcement mesh in the JRCP was considered to be at mid-depth of
the concrete slabs. Other information used in the finite element model was considered
to be similar to models developed in Chapter 8. Moreover, a debonding action
between concrete slabs and subbase was simulated using contact pairs (see Sections
3.2.3 and 3.4.2.2 for further information on modelling of debonding layer).
With regards to the debonding layer considered in the field test (single layer polythene
sheet), the friction coefficient between concrete slabs and subbase was considered to
236
be 1.2 based on results of the research published by Suh et al. (2002). Figure 10-1
shows a typical finite element model of JPCP that is employed in the current study.
Figure 10-1. Finite element model of the unbonded JPCP
Diverse axle group types including SAST, SADT, TAST, TADT, TRDT and QADT
with average loads of 53 kN, 80 kN, 90 kN, 135 kN, 181 kN, and 221 kN were
separately applied upon the pavement and at the free longitudinal joint. Different
velocities (5, 30, 60 and 110 km/h) and time steps (0.144, 0.024, 0.012 and 0.0065,
respectively) were considered. Other information on axle group configuration is
similar to that used in Chapter 8 and can be found in Table 8-1.
Based on the work of Hendrick et al. (1992), wheel loads were distributed among the
nodes at the top surface layer of the concrete slab representing the tyre-pavement
contact area. These nodal loads were then moved along longitudinal direction of the
pavements based on relevant time steps.
237
10.3. Model calibration
Since the finite element model (FEM) of the concrete pavement contains a variety of
elements with different properties and nonlinearities, the validation of the FEM is the
most important part of this chapter. To do this, a finite element model was developed
and calibrated based on recorded time history responses of the field test provided in
Chapter 9. The truck was modelled based on information provided in Section 8.6 and
Figure 9-19. The FEM had the width and length equal to the total width and length of
the test section with the same characteristics (see Figure 9-1). This led to an
involvement of more than 500000 nodes in the FEM. This consideration in nonlinear
dynamic analysis needs a large capacity of memory and requires a long time for
processing. For instance, the analysis time on a supercomputer platform for a truck
speed of 49.3 km/h with a time step of 0.0146 is about two months. Both slab
deflection and induced tensile stress were used for calibration of the FEM.
Calibration of slab deflection
The induced slab deflections from FEA, as presented in Table 10-1, are in good
agreement with those captured in the field test for a speed close to 49.3 km/h. it needs
to be noted that truck driver was not able to constantly maintain either the truck wheel
path along the predetermined locations or the truck speed during the test. This
increases or decreases the recorded value of the slab deflection for some of LDTs.
However, the truck speed was constant in the FEA (i.e. 49.3 km/h). Information on
location of the LDTs was previously presented in Figure 9-13.
Table 10-1. Comparison is slab deflection between FEA and the field test
Real speed Results Results recorded

Location
(km/h) of FEA in the field test
DL2 48.7 -0.45 -0.4
DR4 48 -0.193 -0.175
DL6 43.16 -0.52 -0.414
DL7 49.3 -0.642 -0.645
DL8 49 -0.617 -0.608
DL10 41.25 -0.297 -0.28
DR13 34.6 -0.377 -0.115
238
Calibration of induced Stress
Results of the FEA show that the induced tensile stress is significantly affected by the
value of the time stepping. In other words, lower time step creates more accurate
results but needs more time to run. Consequently, the aforementioned time step
considered in the current study was not small enough to capture the maximum induced
tensile stresses for all points of interest (Figure 9-12 presents locations of these
points).
A comparison between stress influence lines developed for some of the points of
interest and the recorded time histories for similar location during the field test shows
a good agreement between them. Nevertheless, results of the FEA for all axle groups
used in the truck are not compatible with the corresponding results derived from field
test. Figures 10-2 shows the stress time histories for BCR3 obtained from the field test
and Figures 10-3 shows result of the FEA for the same point. Refer to Figure 9-12 to
find the location of BCR3 in the test section. This point (BCR3) was located in the
JRCP with a bonded interaction between concrete slab and subbase. In these figures,
FEA results for SAST and TADT are in a good agreement with those captured in the
file test. In contrast, the pavement response due to TRDT captured in the field test is
greater than that estimated in the FEA.
Figure 10-2. Stress time history at BCR3 recorded in the field test
239
To overcome this problem and to create more accurate results, the time step value was
decreased from 0.0146 to 0.005. In this case, a very good agreement was observed
between results of FEA and those captured in the field test. Figure 10-4 compares
stress time histories of FEA and the field test for TCR12 located in the unbonded
JRCP. Refer to Figure 10-12 for more information on location of TCR12.
Figure 10-3. Stress time history at BCR3 derived from FEA results
20
15
Strain (µε)
10
0
0 0.5 1 1.5 2 2.5 3
-5
Time (Sec.)
Field test at TCL12 FEA at TCL12
Figure 10-4. Comparison of stress time histories at TCR12 between field test and FEA
240
The above mentioned comparisons indicate that the model developed for unbonded
concrete pavements is accurate enough to be used for further study.

Since different pavement deteriorations can be addressed by taking into consideration
the location of axle groups upon the pavement, the results for the most critical axle
locations are presented in this section. These are mid-edge and corner loadings. Stress
influence lines for these locations are prepared using results of nonlinear dynamic
FEA. Results are classified based on types of axle group. All provided stress influence
lines are derived based on first principle (FP) stress or a stress in the longitudinal
direction (SZ).
10.4.1 . Effects of moving single axles on induced tensile stress

Results of the current study for mid-edge loading show that a speed of 110 km/h can
produce a greater induced tensile stress than static loading. The dynamic amplification
in JPCP is about 5.2% for SAST (Fig. 10-5a) and 290% for SADT (Fig. 10-6a)
respectively. Greater dynamic amplification of 725% and 483% due to SAST (Fig. 10-
5b) and SADT (Fig. 10-6b) can be respectively observed in the JRCP.
Moreover, results show that the induced tensile stress in an unbonded pavements,
JPCP and JRCP, is significantly affected by axle speed of 110 km/h. This is more
significant in JRCP. Furthermore, the number of maximum tensile stress repetitions is
not the same as in static loads. In the static loading condition, the number of repetitions
of maximum induced tensile stress relies on number of axles in a given axle group. For
instance, the number of stress repetitions for SAST or SADT is equal to one as the
number of axles in these axle groups is equal to one. However, results of the current
study show that higher stress repetitions can be observed when a speed of 110 km/h is
considered. In this case, number of stress repetitions in JRCP is about 5 times for
SAST (Fig. 10-5b) and 8 times for SADT (Fig. 10-6b).
Since the axle groups move along the pavement from left to right, the above mentioned
phenomenon occurs in the last concrete slab panel where the far transverse edge
(distance from free edge = 13.8 in JPCP and 18.4 in JRCP) is a free edge.
Consequently, a question may arise whether the aforementioned results are due to the
contribution of a free edge in the analysis. If the aforementioned results are due to the
241
contribution of free edge in the FEA, the same stress repetitions shall be observed in
the first concrete slab panel as the transverse edge of the pavement on the left hand
side (distance from free edge = 0) is also a free edge.
0.6
Transverse Joint
Transverse Joint
0.5
0.4
Induced stress (MPa)
0.3
0.2
0.1
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0.1
-0.2
Distance from free edge (m)

V = 5 km/h V = 30 km/h V = 60 km/h V = 110 km/h
a) In JPCP
2.00
Transverse Joint
1.75
1.50
1.25
1.00
0.75
0.50
0.25
0.00
0 2.3 4.6 6.9 9.2 11.5 13.8 16.1 18.4
-0.25
-0.50
-0.75
-1.00
-1.25
-1.50

b) In JRCP
Figure 10-5. Influence line of induced tensile stress for a point at mid-edge of concrete
pavements due to SAST
242
1.5
Transverse Joint
Transverse Joint
1.25
1
0.75
0.5
0.25
0
0 2.3 4.6 6.9 9.2 11.5 13.8
-0.25
-0.5
-0.75
-1

a) In JPCP
2.00
Transverse Joint
1.50
1.00
0.50
0.00
0 2.3 4.6 6.9 9.2 11.5 13.8 16.1 18.4
-0.50
-1.00
-1.50

b) In JRCP
Figure 10-6. Influence line of induced tensile stress for a point at mid-edge of the
concrete pavements due to SADT
243
However, the phenomenon of stress repetitions does not occur in the first panel.
Furthermore, if this phenomenon is due to contribution of the free edge, it shall be
happened for other speeds as well. Results of the current study indicate that this
phenomenon occurs only for a speed of 110 km/h. Hence, the contribution of the free
edge is not the reason behind this phenomenon.
This phenomenon may be addressed by taking into consideration the pavement

curvature under the moving loads at a certain time. Figure 10-7 shows the JRCP
curvature when a QADT with speed of 110 km/h is passing along free edge of the
pavement from the right hand side of the figure toward its left hand side.
Figure 10-7. Pavement curvature in JRCP under the QADT with speed of 110 km/h
The axle group is positioned about the mid-edge of the slab located on left hand side.
However, the maximum induced tensile stress occurs in the unloaded slab. The slab
deflection waves due to each axle in a quad axle group are propagated in both
longitudinal and transverse directions of the pavement at any time. The deflection
waves of different time steps and different axles may magnify or diminish each other.
The effect of a magnified deflection together with concrete slab weight results in a
244
greater induced tensile stress. Furthermore, speed can create vertical acceleration in the
pavement. This produces another stress in the pavement that can be superimposed to
the previous one. Further information on stresses due to acceleration can be found in
Section 3.4.4.
The aforementioned theory does not rely on boundary condition of the pavement and
may result in greater tensile stress in the presence of a confined transverse joint.
Figure 10-8 is an example of this finding captured in a JRCP subjected to TRDT with
a speed of 30 km/h.
Figure 10-8. Pavement curvature in JRCP under TRDT with a speed of 30 km/h
Since availability of free transverse edges may have a minor effect on results of the
FEA, dynamic amplification and number of stress repetitions provided in this section
are based on the following statement. The results of the FEA captured in a distance of
two meters from the free edges are not considered in calculation of dynamic analysis
and/or determination of number of stress repetitions. The free edges are at the
beginning and at the end of the pavement modelled in this study. In other words, an
area of the captured time histories located between 2.0 m and 16.4 m in JRCP and
between 2.0 m and 11.8 m in JPCP are considered for analysis of pavement behaviour
245
in the current research. In terms of corner loading, speed of SAST does not have a
strong effect on induced tensile stress in both JPCP and JRCP (Fig. 10-9). However,
SAST with a speed of 110 km/h can produce three times more stress repetitions than
static load in JRCP (Fig. 10-9b).
Transverse Joint
Transverse Joint
0.8
0.6
0.4
0.2
0
0 2.3 4.6 6.9 9.2 11.5 13.8
-0.2
-0.4

a) JPCP
0.80
Transverse Joint
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0 2.3 4.6 6.9 9.2 11.5 13.8 16.1 18.4
-0.10

b) JRCP
Figure 10-9. Influence line of induced tensile stress for a point at the corner of the
concrete pavements due to SAST
246
Static and dynamic loads of SADT produce the same induced tensile stress in a point at
the corner of JPCP (Fig. 10-10a). However, SADT with a speed of 110 km/h creates a
greater tensile stress (142%) in JRCP than static SADT (Fig. 10-10b). Moreover, the
stress repetition due to dynamic loads was increased by 8 times in JRCP (Fig. 10-10b).
10.4.2 . Effects of moving tandem axles on induced tensile stress

Speed of TAST has also some effects on pavement responses. For a point at mid-edge
of the JPCP, speed of 30 km/h produces a greater tensile stress of 8% than static
loading of TAST (Fig. 10-11a). In JRCP, TAST with speed of 110 km/h creates a
higher value of tensile stress in the pavement. The dynamic amplification in this case is
about 1050% (Fig. 10-11b). The number of stress repetitions in a point at mid-edge of
JRCP is about 9 times as compared to 2 times in static loading of TAST.
Dynamic loads of TAST have no effect on induced tensile stress in a point at the
corner of JPCP (Fig. 10-12a). Nevertheless, a dynamic amplification of 150% can be
observed in a point at the corner of JRCP subjected to TAST with speed of 110 km/h
(Fig. 10-12b). The number of stress repetitions in this case is about 7 times as
compared to 2 times in static analysis of TAST.
TADT with a peed of 110 km/h can increase the induced tensile stresses in a point at
mid-edge of unbonded JPCP and JRCP by about 181% and 260% respectively (Fig.
10-13). The number of stress repetitions in JPCP and JRCP are respectively 4 times
and 8 times compared to the static loading of TADT (2 times).
For a point at the corner of the JPCP, speed of TADT has no effect on induced tensile
stress (Fig. 10-14a). Nonetheless, TADT with a speed of 110 km/h produces a greater
tensile stress of 64% than static analysis in a point at the corner of JRCP (Fig. 10-14b).
Furthermore, the number of stress repetitions in a point at the corner of JPCP and
JRCP are two times higher than that in the static loading.
247
0.7
Transverse Joint
Transverse Joint
0.6
0.5
0.4
0.3
0.2
0.1
0
0 2.3 4.6 6.9 9.2 11.5 13.8
-0.1

a) JPCP, FP at the bottom surface layer of the concrete slab
0.40
Transverse Joint
0.30
0.20
0.10
0.00
0 2.3 4.6 6.9 9.2 11.5 13.8 16.1 18.4
-0.10
-0.20
-0.30

b) JRCP, SZ at the top surface layer of the concrete slab
concrete pavements due to SADT
248
10.4.3 . Effects of moving triple and quad axles on induced tensile
stress
TRDT with a speed of 30 km/h creates greater tensile stresses in a point at mid-edge of
JPCP than static load (Fig. 10-15a). The dynamic amplification for this case is about
189%. In JRCP (Fig. 10-15b), on the other hand, TRDT with speed of 110 km/h
creates a greater dynamic amplification of 400%. The number of stress repetitions in
JPCP and JRCP are respectively 2 times and 12 times compared with the static load
which is only three times. For a point at the corner of JPCP and JRCP, speeds of 30
km/h and 110 km/h produce more tensile stress of 200 % and 100 % than static load,
respectively (Fig. 10-16). The number of stress repetitions in JRCP subjected to TRDT
is 2 times more than the corresponding static load.
Figures 10-17 and 10-18 present the influence stress line for a point at mid-edge and
corner of both JPCP and JRCP subjected to QADT with different velocities. Results
indicate that speeds of 30 km/h or 110 km/h produce a greater tensile stress in JPCP
than static loads. The dynamic amplification is about 39% in this case. For a point at
mid-edge of JRCP, a speed of 30 km/h results in greater tensile stress by about 5%
than that in static load. The number of stress repetitions in both JPCP and JRCP is two
times higher than that in the static load.
Whilst QADT with a speed of 30 km/h produces a greater tensile stress in a point at
corner of JPCP than static load, speed of QADT has no significant effect on
corresponding location in JRCP. The dynamic amplification in JPCP is about 27%.
The number of stress repetitions for both pavements under the dynamic loads is the
same as that under the static load.
10.5. Critical speed of axle groups and number of stress repetitions

As described in the previous section, pavement response is affected by the speed of
axle groups. In addition to dynamic amplification, the most important finding was the
number of stress repetitions in a certain location due to moving loads. As mentioned
earlier, the number of stress repetitions for a certain point within the pavement in a
static analysis is equal to number of axles in a given axle group. Table 10-2
summarises all significant information derived from current study. Information
249
provided in this table is then used for development of a new concrete slab design guide
described in the next chapter of this thesis.
0.25
Transverse Joint
Transverse Joint
0.2
0.15
0.1
0.05
0
0 2.3 4.6 6.9 9.2 11.5 13.8

a) In JPCP
3.00
Transverse Joint
2.00
1.00
0.00
0 2.3 4.6 6.9 9.2 11.5 13.8 16.1 18.4
-1.00
-2.00
-3.00

b) In JRCP
concrete pavements due to TAST
250
0.35
Transverse Joint
Transverse Joint
0.3
(Induced stress (MPa)

0.25
0.2
0.15
0.1
0.05
0
0 2.3 4.6 6.9 9.2 11.5 13.8

a) JPCP
2.00
Transverse Joint
1.50
1.00
0.50
0.00
0 2.3 4.6 6.9 9.2 11.5 13.8 16.1 18.4
-0.50
-1.00
-1.50
-2.00
-2.50
-3.00
-3.50
-4.00

b) JRCP
concrete pavements due to TAST
251
1.8
Transverse Joint
Transverse Joint
1.6
1.4
1.2
0.8
0.6
0.4
0.2
0
0 2.3 4.6 6.9 9.2 11.5 13.8
-0.2
-0.4
-0.6
-0.8
-1

b) In JPCP
2.00
Transverse Joint
1.50
1.00
0.50
0.00
0 2.3 4.6 6.9 9.2 11.5 13.8 16.1 18.4
-0.50
-1.00
-1.50
-2.00

b) In JRCP
concrete pavements due to TADT
252
0.6
Transverse Joint
Transverse Joint
0.5
0.4
0.3
0.2
0.1
0
0 2.3 4.6 6.9 9.2 11.5 13.8
-0.1

a) JPCP
1.20
Transverse Joint
1.00
0.80
0.60
0.40
0.20
0.00
0 2.3 4.6 6.9 9.2 11.5 13.8 16.1 18.4
-0.20

b) JRCP
concrete pavements due to TADT
253
0.6
Transverse Joint
0.4

0.2
0
0 2.3 4.6 6.9 9.2 11.5 13.8
-0.2
Transverse Joint
-0.4
-0.6
-0.8

a) In JPCP
2.50
Transverse Joint
2.00
1.50
1.00
0.50
0.00
0 2.3 4.6 6.9 9.2 11.5 13.8 16.1 18.4
-0.50
-1.00
-1.50

b) In JRCP
concrete pavements due to TRDT
254
2
Transverse Joint
Transverse Joint
1.5
0.5
0
0 2.3 4.6 6.9 9.2 11.5 13.8
-0.5
-1

a) JPCP
0.30
Transverse Joint
0.20
0.10
0.00
0 2.3 4.6 6.9 9.2 11.5 13.8 16.1 18.4
-0.10
-0.20
-0.30
-0.40

b) JRCP
concrete pavements due to TRDT
255
10.6. Combination of moving axle groups and differential temperature
To study the effects of differential temperature between the top and the bottom surface
layers of the concrete slab together with moving axle group loads, the JPCP is
subjected to daytime and nighttime differential temperature of 8.5 ºC. Information on
the finite element model is similar to the previous stage. Only SAST is considered in
this study. Different speeds of 5, 30, 60 and 110 km/h are taken into consideration.
Results of the current study show that consideration of temperature gradients produces
a greater tensile stress in pavement. Furthermore, the number of stress repetitions
strongly depends on the magnitude of differential temperature. In other words,
consideration of temperature fluctuation increases the magnitude of tensile tress but
may decrease the number of stress repetitions in a given axle group.
Figures 10-19 and 10-20 show the pavement curvature during daytime and nighttime
differential temperatures. In contrast with fully unbonded pavement, the use of single
layer polythene sheet (friction coefficient of 1.2) results in different behaviour of the
JPCP during daytime and nighttime. The magnitude of maximum tensile stress during
nighttime is about two times higher than that captured during daytime (i.e. 3.717 MPa
during nighttime differential temperature and 1.797 MPa during daytime differential
temperature).
Consequently, combination of nighttime differential temperature together with axle

group loads creates greater tensile stress than daytime differential temperature as
expected. For instance, a combination of mid-edge loading of SAST and daytime and
nighttime differential temperatures produces a tensile stress of 2.004 MPa and 3.835
MPa respectively (Figures 10-21 and 10-22).
Results of the current study show that differential temperature may increase or
decrease number of stress repetitions in JPCP. Whilst daytime differential temperature
increases the number of stress repetitions from one (recorded in the Section 10.4.1) to
three, nighttime differential temperature decreases the number of stress repetition from
5 to 3 (Fig. 10-23).
256
0.7
Transverse Joint
Transverse Joint
0.6
0.5
0.4
0.3
0.2
0.1
0
0 2.3 4.6 6.9 9.2 11.5 13.8
-0.1

a) In JPCP
0.25
Transverse Joint
0.20
0.15
0.10
0.05
0.00
0 2.3 4.6 6.9 9.2 11.5 13.8 16.1 18.4
-0.05

b) In JRCP
concrete pavements due to QADT
257
0.3
Transverse Joint
Transverse Joint
0.25

0.2
0.15
0.1
0.05
0
0 2.3 4.6 6.9 9.2 11.5 13.8
-0.05

a) In PCP
0.90
Transverse Joint
0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0 2.3 4.6 6.9 9.2 11.5 13.8 16.1 18.4
-0.10
Distance fom free edge (m)

b) In JRCP
concrete pavements due to QADT
258
Table 10-2. Summary of the dynamic results for different axle groups and different types of unbonded concrete pavement
SAST SADT TAST TADT TRDT QADT

at Mid- at at Mid- at at Mid- at at Mid- at at Mid- at at Mid- at
Axle group type edge of corner edge corner edge corner edge corner edge corner edge corner
the of the of the of the of the of the of the of the of the of the of the of the
slab slab slab slab slab slab slab slab slab slab slab slab
Critical
30 OR
speed 110 N.A. 110 N.A. 30 N.A. 110 N.A. 30 30 30
110
(km/h)
Dynamic
JPCP
amplification 5.2 0 290 0 8 0 181 0 189 200 39 27

(%)
Number of
stress 1 1 1 1 2 2 4 4 5 5 6 4
repetition
Critical
speed 110 N.A. 110 110 110 110 110 110 110 110 30 N.A.
(km/h)
Dynamic
JRCP
amplification 725 0 483 142 1050 150 260 64 400 100 5 0

(%)
Number of
stress 7 3 8 8 7 7 8 4 12 5 6 4
repetition
259
Figures 10-19. Pavement curvature due to a daytime differential temperature of 8.5 ºC
Figures 10-20. Pavement curvature due to a nighttime differential temperature of 8.5 ºC
260
Figures 10-21. Combination of SAST and daytime differential temperature of 8.5 ºC
Figures 10-22. Combination of SAST and nighttime differential temperature of 8.5 ºC
261
Figures 10-23. Stress repetition due to a combination of SAST and daytime differential
temperature of 8.5 ºC
10.7. Summary
Dynamic transient analyses of unbonded JPCP and JRCP subjected to different axle
groups were carried out in this chapter. Critical speeds of all axle groups were
determined in both JPCP and JRCP for a point at mid-edge of the concrete slab as well
as a point at the corner of the slab. For the first time worldwide, dynamic
amplifications of each axle group in different pavements and locations were also
determined. The dynamic amplifications varied between 5 per cent and 1050 per cent
depending on axle group types. Whilst the critical speed for the lighter axle group
(single axles and tandem axles) is about 110 km/h, the critical speed in the heavier axle
groups (TRDT and QADT) is about 30 km/h.
the current study was in the area of stress repetition for each individual axle group. In
contrary to static analysis, where the stress repetition for a certain point in pavement is
equal to number of axles in the axle group, higher number of stress repetition can be
observed in the dynamic analysis. The number of stress repetitions is about two to
eight times higher than that of corresponding static analysis.
262
Furthermore, the effects of temperature fluctuation within the depth of the concrete
pavement on the dynamic response of the pavement were addressed. Results showed
that differential temperature gradients may increase or decrease the stress repetition for
each axle group. However, further study is required to clearly address the effects of
temperature fluctuations on stress repletion of each individual axle group.
263
Chapter 11
DEVELOPMENT OF SLAB THICKNESS DESIGN GUIDE

FOR ZERO MAINTENANCE FATIGUE DAMAGE
11.1. Introduction
As mentioned in Chapter 3, concrete pavements were traditionally designed based on
theoretical solutions such as Westergaard method (1926, 1933, and 1947). Contribution
of finite element analysis for determining stress distribution within concrete slabs later
led to a mechanistic approach which was extensively adopted in the PCA (1985) and
Austroads (2004) slab thickness design guides. The bottom-up mid-edge transverse
cracking due to mid-edge loading was the only fatigue damage mode of concrete
pavements considered in the mechanistic approach.
Numerous field observations on pavement behaviour and its deterioration under the
applied loads revealed that concrete pavements suffer from corner and longitudinal
cracks as much as mid-edge transverse cracks (Heath et al. 2003 and Hiller and Roesler
2005). As a result, other factors such as built-in curling, loss of moisture contents-
shrinkage and differential temperature gradients between the top and the bottom surface
layers of the concrete slab were taken into consideration. Some minor studies on
dynamic effects were also conducted. Furthermore, a debonding layer was utilised
between concrete slabs and subbase to eliminate the early age cracking.
Consequently, most recent concrete pavement design guides such as AASHTO (2003)
are based on an empirical-mechanistic approach. Whilst the empirical parts provide
information on possible damage modes of concrete pavements, the mechanistic part
predicts the magnitude of flexural stress in pavement which may result in a particular
damage mode. In contrast with the mechanistic approach, the empirical-mechanistic
approaches are more sophisticated and therefore more reliable.
In the slab thickness design procedure, the vehicular loads were considered as either a
series of equivalent single axle loads (ESALs) or axle group loads. While the ESALs
concept was extensively adopted in the AASHTO design guide, the PCA and
264
Austroads design guides are based on a variety of axle group loadings. Although none
of the above mentioned simulations for vehicular loads are able to exactly predict the
behaviour of concrete pavements in real conditions (under truck loads), the use of axle
group loads seems to provide a better approach in prediction of pavement behaviour
than the use of ESALs.
Several stress prediction models were previously developed by Ioannides et al. (1985),
Bendana et al. (1994), Lee et al. (1997), Lee (1999) and Lee and Carpenter (2001)
tacking into consideration the temperature effects on concrete pavement deteriorations.
However, the assumptions used for developing stress prediction model is not
compatible with the Australian construction procedure and loading conditions. For
instance, work of Ioannides et al. (1985) was based on ESALs concept. Bendana et al.
(1994) superimposed the vehicular induced tensile stresses to corresponding thermal
induced stresses which is not always correct. For future information on this matter refer
to Chapter 6 of this thesis.
11.2. Cross section of the concrete pavement

Based on the concrete pavement technology in Australia, the cross section of the
pavement considered in this guide contains concrete slabs, debonding layer, subbase
and subgrade. This guide was developed to design JPCP and JRCP. Since Austroads
recommendations for selection of concrete slab dimensions were used in this research,
distances between transverse joints and longitudinal joints (length and width of the
concrete slab) are in accordance with Austroads (2004). Further information on
Austroads recommendations was provided earlier in Chapter 3.
In addition, transverse joints in JPCP and JRCP are reinforced by at least eleven
evenly spaced dowel bars 32 mm in diameter and 450 mm long. Longitudinal joints in
JPCP and JRCP are also reinforced by four and eight tie bars with a diameter of 14
mm and a length of 1000 mm, respectively.
11.3. Concrete Characteristics

The concrete pavements are subjected to environmental effects simultaneously of
construction. If the pavement is subjected to an environmentally induced stress before
having adequate strength to resist the stress, a crack is initiated in the pavement. As a
result, a minimum concrete flexural strength is considered in the design guides. The
265
results of the laboratory tests conducted in the current research showed that the
flexural strength of air cured concrete samples are about 35 per cent lower than that of
water cured samples. Since the water content of concrete is strongly affected by the
element size, results of the laboratory tests may be not perfectly comparable with the
concrete flexural strength in the pavements. The concrete flexural strength of 4.5 MPa
is recommended by the Austroads (2004). However, greater flexural strength may be
required in the presence of harsh environments.
Based on the results of current research derived from laboratory fatigue tests, finite
element analyses and the experimental field test, no limitation on maximum concrete
strength is recommended. In other words, the results of current study show that an
increase in concrete strength enhances the pavement resistance to fatigue related
failure modes. However, this statement is valid for conventional concrete with a
maximum concrete compressive strength of 65 MPa. Consequently, further study is
required to address effects of the concrete compressive strength on fatigue related
damages if high strength concrete is involved in pavement construction.
11.4. Subbase
The longevity of concrete pavements is affected by the provision of subbase layer. A
LMC subbase of 150 mm thick and compressive strength of 5 MPa is constantly
considered in the finite element analyses performed in this research. Hence, further
study is required to determine the optimum characteristics of subbase for different
types of concrete pavements and variety of subgrade soil. While this guide is used, a
LMC subbase with the aforementioned characteristics shall be constructed over the
subgrade and under the concrete slab.
11.5. Prediction of the maximum induced tensile stress

A comparison between results of vehicular induced tensile stress in finite element
analysis with corresponding stresses predicted by using Austroads stress prediction
model shows that the Austroads is not able to accurately predict the induced tensile
stresses in the pavement subjected to TADT and QADT loadings (further information
on this matter will be presented later in Section 11.7). Furthermore, the prediction of
induced tensile stress in the Austroads method is based on this assumption that only 6
per cent of the traffic passes along the edge area of the traffic lane (Packard and
266
Tayabji, 1985). The edge area is along the longitudinal edge of the concrete slab and in
a transverse distance of 600 mm from longitudinal joints or edges. As a result, the
Austroads equation can not be used for other values of edge loading. The work of
Lennie and Bunker (2005) showed that the volume of the traffic passing along edge
area in the Queensland State is much higher than the above mentioned assumption.
To increase the accuracy of stress prediction in concrete pavements due to vehicular

loads, different equations are developed in this chapter based on results of static
analyses of the concrete pavement subjected to vehicular and environmental effects
using LaGrange method. Information on static analysis of the pavement was presented
in the Chapter 6. Effects of dynamic loads are then contributed in the stress prediction
model.
For a bonded traffic lane of JPCP confined at one of its longitudinal edges by a
shoulder, the maximum vehicular induced tensile stress is predicted using the
following equations:
SAST: σ V = Pf × (0.0153 − 2.208 × 10 −5 × ( Pf − 53)) (11-1)
SADT: σ V = Pf × (0.0124 − 1.1 × 10 −5 × ( Pf − 80)) (11-2)
TAST: σ V = Pf × (0.00708 − 3.0 × 10 −6 × ( Pf − 90)) (11-3)
TADT: σ V = Pf × (0.00607 − 1.87 × 10 −6 × ( Pf − 135)) (11-4)
TRDT: σ V = Pf × (0.00456 − 1.623 × 10 −6 × ( Pf − 180)) (11-5)
QADT: σ V = Pf × (0.00348 − 7.33 × 10 −7 × ( Pf − 220)) (11-6)
Where σ V (MPa) is maximum vehicular induced tensile stress in a bonded confined
lane due to mid-edge loading at free edge, and Pf (kN) is the ultimate axle group load
and is equal to axle group load multiplied by load safety factor (LSF). LSF can be
derived from table 9.2 of the Austroads (2004). This table is represented in this chapter
as Table 11-1.
267
Table 11-1. Load safety factor for concrete pavement design
Project Design
Reliability 85% 90% 95% 97.50%
Load Safety Factor 1.05 1.1 1.2 1.25
Assumptions used for development of the aforementioned equations are:
- Dimensions of concrete slabs: 4600 mm length, 3600 mm width and

250 mm thickness.
- Concrete properties: modulus of elasticity and Poisson’s ratio of 28000

MPa and 0.2 respectively.
- Tied shoulders: 4600 mm length, 1500 mm width and 250 mm

thickness with the same properties of the concrete slabs.
- Transverse joints were dowelled by eleven evenly spaced cylindrical

bars having 32 mm diameter, 450 mm length and 1000 MPa dowel-slab
support modulus.
- Tie bars with 13 mm diameter and 1000 mm length spaced at 1000 mm

centre to centre were considered at longitudinal joints.
- A cement stabilized subbase of 150 mm thickness, 5000 MPa modulus

of elasticity, and 0.2 Poisson’s ratio was considered under the slab and
over a subgrade with modulus of subgrade reaction of 0.03 MPa / mm
(CBR ≈ 3.5).
The maximum induced tensile stress in an unbonded concrete pavement subjected to

mid-edge loading can be then estimated by multiplying the calculated stresses in the
bonded pavement to relevant factors, C1, provided in the last column of Table 6-2.
To determine the accuracy of the above mentioned equations for stress prediction in
concrete pavements, a confined lane with the same characteristics was subjected to
mid-edge loading of the above mentioned axle groups. Four different axle loads for
each axle group were considered. Results of finite element analyses were then
compared with the corresponding stresses calculated from Equations 11-1 to 11-6 in
268
Table 11-2. Results show the accuracy of the aforementioned equations for stress
prediction in both bonded and unbonded pavements.
Results of finite element analyses were also used to contribute effects of different
loading conditions and adjacent traffic lanes (provision of shoulders at both
longitudinal edges of the traffic lane) into Equations 11-1 to 11-6 for prediction of
maximum vehicular induced tensile stresses within the concrete pavement.
Consequently, the following equation was developed:
σ V ,i = C1 × C 2 × C 3 × C LS × σ V ,i (11-7)
Where i represents the type of axle group and is SAST, SADT, TAST, TADT, TRDT
or QADT, σ V ,i is vehicular induced tensile stress in concrete pavement for mid-edge
loading due to axle i and can be calculated from Equations 11-1 to 11-6 for each
individual axle group, C1 is a factor for calculating the maximum induced tensile stress
in an unbonded pavement and is derived from Table 6-2. For bonded pavement, C1=1,
C2 is a coefficient for contributing effect of different load positions in Equations 11-1
to 11-6. This coefficient is derived from Table 11-3. For mid-edge loading condition,
C2=1, C3 is a factor for taking into consideration the provision of shoulders at both
longitudinal edges of the traffic lane in the maximum induced tensile stress and can be
determined from Table 11-4. For pavement with a free longitudinal edge, C3=1, CLS is
a factor for consideration of load shift between axles in a given axle group. For SAST
and SADT, CLS =1 and for other axle group can be calculated from Equation 11-8.
C LS = 1 + 0.0088 × LS (11-8)
Where LS (%) is the magnitude of load shift between axles in a given axle group. The
mean percentage of load shift between axles is about 1 to 2 per cent in the air
suspension and between 20 to 40 per cent in the mechanical suspensions (Blanksby et
al., 2006).
Table 11-5 compares results of finite element analyses of a confined lane with those
predicted using Equation 11-7. Since the effects of slab thickness and subgrade
strength have not been included in the stress prediction equations, the slab thickness
and modulus of subgrade reaction was considered to be the same as those used during
development of Equations 11-1 to 11-6. Results indicate that the use of the above
269
mentioned equation for stress predication is very accurate in most cases particularly in
the presence of debonding layer between concrete slabs and subbase. The maximum
error in stress prediction is about 9 percent and occurs when a bonded pavement is
subjected to the corner loading condition of a QADT.
Table 11-2. Accuracy of the stress prediction equations provided in the thesis
Axle Predicted Stress FEA Error (%)

Group Load Bonded Unbonded Bonded Unbonded Bonded Unbonded
30 0.474235 0.814736 0.48 0.819 -1.201 -0.52063
53 0.8109 1.393126 0.807 1.379 0.483271 1.02438
SAST
90 1.303474 2.239368 1.279 2.209 1.913495 1.374724
130 1.767979 3.037388 1.759 3.037 0.510472 0.012785
50 0.6365 1.097326 0.646 1.1 -1.47059 -0.24309
80 0.992 1.710208 0.988 1.7 0.404858 0.600471
SADT
120 1.4352 2.474285 1.423 2.453 0.857344 0.867705
150 1.7445 3.007518 1.733 2.993 0.663589 0.485065
45 0.324675 0.537662 0.332 0.545 -2.20633 -1.34646
90 0.6372 1.055203 0.637 1.055 0.031397 0.019261
TAST
150 1.035 1.71396 1.032 1.708 0.290698 0.348946
200 1.35 2.2356 1.346 2.228 0.297177 0.341113
80 0.493828 0.813829 0.5 0.817 -1.2344 -0.38818
135 0.81945 1.350454 0.819 1.348 0.054945 0.182018
TADT
180 1.077453 1.775643 1.076 1.771 0.135037 0.262143
240 1.409676 2.323146 1.408 2.323 0.119034 0.006287
120 0.558886 0.867949 0.561 0.871 -0.3769 -0.35025
180 0.8208 1.274702 0.821 1.274 -0.02436 0.055133
TRDT
250 1.111598 1.726311 1.101 1.705 0.962534 1.249907
310 1.348193 2.093744 1.348 2.094 0.014325 -0.01223
150 0.529697 0.595379 0.531 0.603 -0.24548 -1.26387
220 0.7656 0.860534 0.766 0.863 -0.05222 -0.2857
QADT
300 1.026408 1.153683 1.01 1.159 1.624554 -0.45879
370 1.246919 1.401536 1.247 1.397 -0.00654 0.324724
Table 11-3. Variations of coefficient C2 in Equation 11-7
Type of Unbonded Pavement Bonded Pavement

Axle Centre Corner Centre Corner
Group Loading Loading Loading Loading
SAST 0.535 0.547 0.546 0.644
SADT 0.617 0.468 0.61 0.565
TAST 0.55 0.84 0.585 1.13
TADT 0.654 0.725 1.45 0.842
TRDT 0.656 0.791 0.656 0.992
QADT 0.738 1.045 0.485 1.025
270
Type of Centre Loading Mid-Edge Loading Corner Loading

Axle Bonded Unbonded Bonded Unbonded Bonded Unbonded
Group Pavement Pavement Pavement Pavement Pavement Pavement
SAST 0.892 0.932 0.496 0.757 0.866 0.903
SADT 0.896 0.937 0.611 0.764 1.063 1.118
TAST 0.936 0.965 0.526 0.669 0.671 0.849
TADT 0.891 0.937 0.578 0.658 0.913 1.013
TRDT 0.866 0.953 0.639 0.693 0.827 0.857
QADT 1.08 1.04 0.554 0.725 0.757 0.795
11.6. Environmental effects

As described in the Chapter 3, environmental effects considered in concrete pavement
analysis are restricted to temperature gradient and loss of moisture content. Effects of
loss of moisture content on pavement response can be simulated as nighttime
temperature gradient (Reddy et al., 1963).
If the coefficient of thermal expansion of concrete is assumed to be constant and equal

to 1×10-5 mm/mm/˚C, the thermal induced stresses can be calculated from the
following equation:
σ T = 0.114 × C 4 × ∆T (11-9)
Where σ T is thermal induced tensile stress in pavement (MPa), ∆T is the absolute

differential temperature between the top and the bottom surface layer of the concrete
slab (˚C), C4 is a coefficient for thermal stress prediction in conjunction with axle
loading conditions.
The value of C4 can be derived from Table 11-6. For a differential temperature
between the values provided in Table 11-6, an internal interpolation needs to be used
to find an accurate value for C4.
Since the concrete pavement was analysed under a combination of axle group loads
and differential temperatures between the top and the bottom surface layers of the
concrete slab, interaction between vehicular loads and temperature gradients were
taken into consideration in Equation 11-9.
271
Table 11-5. Comparison between results of the finite element analyses and those from stress prediction equations
Centre Loading Corner Loading
Types Bonded Pavement Unbonded Pavement Bonded Pavement Unbonded Pavement

of Axle
Axle Load FEA Predicted Error
FEA Predicted Error
FEA Predicted Error
FEA Predicted Error
Group (kN) Stress (%) Stress (%) Stress (%) Stress (%)
SAST 90 0.728 0.712 -2.3 1.227 1.198 -2.4 0.84 0.839 -0.1 1.226 1.225 -0.1
SADT 120 0.861 0.875 1.7 1.556 1.527 -1.9 0.861 0.811 -6.2 1.128 1.158 2.6
TAST 150 0.613 0.605 -1.2 0.93 0.943 1.3 1.086 1.170 7.1 1.381 1.440 4.1
TADT 80 0.703 0.716 1.8 0.514 0.532 3.4 0.426 0.416 -2.5 0.621 0.590 -5.2
TRDT 120 0.366 0.367 0.2 0.569 0.569 0.1 0.555 0.554 -0.1 0.686 0.687 0.1
QADT 300 0.503 0.499 -1.0 0.861 0.851 -1.1 0.956 1.052 9.1 1.154 1.206 4.3
272
Differential Unbonded Pavement Bonded Pavement
Temperature Centre Mid-Edge Corner Centre Mid-Edge Corner
(˚C) Loading Loading Loading Loading Loading Loading
≥ 25 1.17 1.15 2.97 0.91 0.94 0.93
20 1.17 1.02 1.639 0.84 0.87 0.9
Daytime
15 0.86 0.67 0.725 0.84 0.87 0.87

10 0.552 0.65 0.238 0.84 0.87 0.5
5 0.99 0.59 0.175 0.83 0.87 0.61
-5 -1 -1.765 -0.07 0.2 0.16 0.33
Nighttime
-10 -0.019 0.285 0.302 0.55 0.52 0.62

-15 -0.18 0.31 0.355 0.67 0.888 0.73
-20 1.077 1.12 1.23 0.7 0.7 0.78
≤ -25 1.16 1.18 1.26 0.77 0.74 0.81
11.7. Variation of slab thickness

Results of finite element analyses are also used to contribute the effect of concrete slab
thickness in equations developed for stress prediction in concrete pavements. The
LaGrange method was used for development of the formula. Subsequently, Equation
11-10 was developed.
250 0.00022Ts2 − 0.1052Ts +13.55

CTs = α ( ) (11-10)
Ts
Where CTs is a factor providing effect of slab thickness in stress distribution within the
concrete slab, and Ts is slab thickness in mm, α is a coefficient that can be determined
from Table 11-7 for unbonded pavement. For bonded pavement, however, the value of
α is 0.8 when Ts ≤ 250 mm and 1.21 when Ts > 250 mm.
11.8. Variation of modulus of subgrade reaction

Using LaGrange method, results of the finite element analysis provided in Chapter 6
were also used to take into consideration the variation of modulus of subgrade reaction
in equations developed for stress prediction. Consequently, Equation 11-11 was
developed.
273
0.03 0.055+ 0.5×( k −0.03)
Ck = β ( ) (11-11)
k
Where Ck is a coefficient for contributing the subgrade properties in the pavement
response, k is modulus of subgrade reaction in MPa/mm, and β is a factor that is
determined from Table 11-8 for unbonded pavement. In bonded pavements, β is 1.05.
Table 11-7. Variations of coefficient α in Equation 11-10 for unbonded pavements

Ts < 250 Ts <= 290 Ts > 290
Differential Mid-Edge Mid-Edge Mid-Edge
Temperature Corner Corner Corner
and Centre and Centre and Centre
Loading Loading Loading
Loadings Loadings Loadings
≥15 0.394 1.125 1.83
10 1.18 0.894 1.2 1.702
Daytime
8 1 1.02 1.12 1.353

5 1.2
2.22 1
2 4.4
-5 1.8 1.05
1.11 1.11
Nighttime
-8 1.425
-10 1.729 1.203 1.22
-15 1.15 0.9 5.33
3.2
≤ -15 0.946 0.9 6.21
Table 11-8. Variations of coefficient β in Equation 11-11 for unbonded pavements

Differential Corner Mid-Edge and
Temperature Loading Centre Loadings
≥ 15 1.1 1.07
Nighttime Daytime
10 0.98 1.08
5 1.03 1
-5 1.3
-10 1.14 0.89
≤ -15 1.12 0.9
11.9. Combination of vehicular loads and environmental effects

The interaction of vehicular loads and environmental effects were considered in the
above mentioned equations. Hence, the total induced tensile stress in the concrete
pavement can be estimated from the following equation:
σ i, j = (σ V ,i + σ T ) × CTs × CK × Cl × Cb × Cr × (1 + DAi, j / 100 ) (11-12)
274
Where j denotes the location of a certain axle group, i, upon the concrete slab, C l is an
adjustment factor for variation in slab length, C b is an adjustment factor for variation in
slab width, C r is an adjustment factor for taking into consideration the effect of
reinforcement in stress distribution and DAi denotes the dynamic amplification for a
certain axle group. Depending on the boundary condition between the concrete slab and
the subbase, the dynamic amplification for JPCP and JRCP can be derived from Table
8-1 and Table 10-2.
Since distance between transverse joints and longitudinal joints (length and width of the
concrete slab) were considered in this research to be constant and equal to those values
provided by Austroads (2004), C l and C b are considered to be equal to 1. Some typical
values for C r may be determined from a comparison between induced tensile stresses
in JPCP and corresponding values from JRCP. However, this may lead to inaccurate
stress prediction as the length of JRCP in this research is about two times longer than
that in JPCP. Note that effects of reinforcement on pavement response were not
considered in this research. Hence, C r is considered to be 1 as well. Nevertheless,
further studies are required to accurately determine the values of C l , C b and C r .
11.10. Validation of stress prediction in jointed plain concrete pavement

To determine the accuracy of the stress prediction model for different operating
conditions, two different scenarios were considered. In the first scenario, the induced
tensile stresses in an unbonded pavement subjected to different axle groups were
compared with the corresponding stresses calculated based on Austroads (2004) method
and corresponding vehicular induced tensile stress derived from static finite element
analyses. The effects of dynamic loads were included into the stress prediction models
developed in the current research using dynamic amplification (DA) factors which were
previously established. Hence, the use of static analysis not only can be used in the
validation of the stress prediction models, but it also allows for accurate comparisons
between stress prediction models of Austroads(2004) and those developed in the current
research. In the case where comparisons between stress prediction models are
conducted under static condition, the DA is assumed to be zero.
275
Table 11-9 shows results of this comparison. The Austroads stress prediction model
(2004) is based on 6 per cent edge loading. Consequently, this method does not
distinguish between different loading conditions, i.e. mid-edge, centre and corner
loadings. Based on the aforementioned assumption used in the Austroads (2004), the
predicted stress is expected to be greater than stress due to centre loading and lower
than stress due to mid-edge or corner loadings. However, results of the current study
show that the Austroads method predicts a lower stress for corner loading than for
centre loading of TADT and QADT. In other words, the Austroads stress prediction
model is not able to accurately determine the concrete pavement damage due to TADT
and QADT. This discrepancy may also be due to consideration of the critical axle group
configurations in the FEA. The critical axle group configurations were determined in
Chapter 5 of this thesis.
In the second scenario, the induced tensile stresses due to a combination of axle group
loads and differential temperature are predicted. The results are then compared with the
corresponding results from static finite element analyses. Dynamic amplifications are
not considered in this phase as static analysis of concrete pavement is used. The results
(Table 11-10) show that the maximum and minimum errors in the stress prediction are
about 11 per cent for mid-edge loading of QADT and 0.1 percent for mid-edge loading
of SAST. For all axle groups, the average error in the stress prediction is about 3 per
cent.
11.11. Fatigue analysis

Similar to the method for fatigue analysis considered in the PCA (1985) and Austroads
(2004) methods, the method for fatigue analysis considered in this research is based on
Miner’s rule. Hence, the design traffic (DT) is firstly estimated based on heavy vehicle
axle groups (HVAGs). The method for estimation of design traffic was
comprehensively addressed in the Chapter 7 of Austroads (2004). Daytime traffic
distribution (DTD), nighttime traffic distribution (NTD) and traffic distribution along
the edge area of the traffic lane (ETD) are then estimated using results of weight-in-
motion surveys. As mentioned in Chapter 3, the age area located along the pavement
edge and in a distance of 600 mm from longitudinal joints or edges.
276
Table 11-9. Comparison between stress prediction models developed in the current research with FEA and the stress prediction model used in the
Austroads method (2004)
Induced Tensile stress (MPa)

Axle Availiblity Slab
Axle Group Load Axle Boundary of Thickness K Austroads Current Stress
Type (kN) Location Condition Shoulder (mm) (MPa/mm) (2004) Prediction model FEA
40 Mid-Edge No 200 0.05 0.856 1.388 1.39
SAST
70 Centre Yes 300 0.07 0.738 0.606 0.59
60 Centre No 200 0.04 1.086 1.06 0.997

SADT
140 Mid-Edge No 250 0.06 1.67 2.699 2.739
50 Corner No 200 0.05 0.562 0.649 0.712

TAST Unbonded
180 Centre No 300 0.07 0.994 0.756 0.757
45 Corner No 200 0.08 0.242 0.415 0.46

TADT
120 Centre No 300 0.05 0.255 0.55 0.551
90 Corner Yes 250 0.03 0.329 0.412 0.405

TRDT
140 Centre No 300 0.07 0.561 0.4212 0.406
50 Corner No 250 0.02 0.1114 0.223 0.258

QADT
140 Centre Yes 300 0.06 0.2 0.294 0.301
277
Table 11-10. Comparison between stress prediction model developed in the current research with corresponding results from FEA
Modulus of Maximum Induced

Type of Axle Location Slab Subgrade Differential Boundary Condition Availability Tensile Stress (MPa)
Axle Load of Axle Thickness, Reaction, K Temperature Between Concrete of Prediction Error
Group (kN) Group Ts (mm) (MPa/mm) (˚C) Slab and Subbase Shoulder FEA Method (%)
40 ME 200 0.05 -8 Unbonded No 1.005 1.006 0.1
SAST 70 Ce 300 0.07 18 Unbonded Yes 3.958 4.032 1.84
100 Co 250 0.03 12 Unbonded No 1.395 1.379 -1.16
60 Ce 200 0.04 -10 Unbonded No 1.105 1.056 -4.64
SADT 120 Co 300 0.08 23 Unbonded Yes 6.01 6.34 5.21
140 ME 250 0.06 -18 Unbonded No 4.235 4.672 9.35
50 Co 200 0.05 2 Unbonded No 0.928 0.933 0.54
TAST 150 ME 250 0.03 -15 Bonded Yes 2.036 2.032 -0.2
180 Ce 300 0.07 8 Unbonded No 1.675 1.709 1.99
45 Co 200 0.08 -10 Unbonded No 1.525 1.659 8.08
TADT 120 Ce 300 0.05 -16 Unbonded No 3.583 3.585 0.06
200 Ce 250 0.03 8 Unbonded No 2.135 2.05 -4.15
90 Co 250 0.03 3 Unbonded Yes 0.659 0.645 -2.17
TRDT 140 Ce 300 0.07 10 Unbonded No 1.607 1.604 -0.19
270 Ce 200 0.05 -18 Bonded No 2.823 2.858 1.22
50 Co 250 0.02 -14 Unbonded No 0.911 0.865 -5.32
QADT 140 Ce 300 0.06 8 Unbonded Yes 1.092 1.07 -2.06
280 Me 150 0.04 -4 Unbonded No 1.809 1.627 -11.19
ME: Mid-edge loading, Ce: Centre loading, Co: Corner loading, minus in front of differential temperature means nighttime differential temperature
278
Based on information provided in the commentary of Austroads (2004), the
recommended values for DTD and NTD are 40 per cent and 60 per cent. The ETD is
recommended to be 30 per cent based on the work of Lennie and Bunker (2005).
However, further studies are required to sufficiently determine the values of DTD, NTD
and ETD for each individual state of Australia. The number of load repetitions during
daytime and nighttime for each axle group and different loads can be then estimated
using presumptive traffic load distribution. Austroads (2004) provides a typical
presumptive traffic load distribution for urban (Table 3-4) and rural (Table 3-5) roads.
LRDEi,j = DT × PAGi × PTLDi,j ×DTD × ETD × 1 E -06 (11-13)

LRDCi,j = DT × PAGi × PTLDi,j ×DTD × (100 - ETD) × 1 E -06 (11-14)
LRNEi,j = DT × PAGi × PTLDi,j ×NTD × ETD × 1 E -06 (11-15)
LRNCi,j = DT × PAGi × PTLDi,j ×NTD × (100 - ETD) × 1 E -06 (11-16)
Where LRDEi,j and LRNEi,j are the expected load repetition at the edge area due to axle
group i with a load of j during daytime and nighttime respectively, LRDCi,j and LRNCi,j
are the expected load repetition at the centre area of the traffic lane due to axle group i
with a load of j during daytime and nighttime respectively, PAGi is proportion of axle
group i in a given presumptive traffic load distribution, PTLDi,j is proportion of load j
within the axle group i in a given presumptive traffic load distribution (%), DTD, NTD
and ETD are as previously defined (%).
In the next step, the daytime and nighttime differential temperature between the top and
the bottom surface layers of the concrete slab are determined. Since no practical
estimation model for prediction of differential temperature has yet been adopted in
Australia, further study is required. As a result, the differential temperatures recorded in
the field test carried out during this research (see Chapter 9) are used. The typical
recommended values for a location close to south-east of Brisbane area are 12°C during
daytime and -8°C during nighttime. However, further research needs to be conducted to
determine more accurate values for daytime and nighttime differential temperature
gradients in each state of Australia.
With consideration of a certain value for project design reliability, the load safety factor
is determined using Table 11-1. The induced tensile stresses for centre, mid-edge and
corner loadings during daytime and nighttime differential temperature are subsequently
279
predicted for all loads and axle group types using Equation 11-12. The stress ratio (Sr)
is subsequently calculated using Equation 11-17.
predicted stress
Sr = (11-17)
Rs × f r
Where Rs is the material strength reduction factor and is equal to 0.6 in plain concrete
and 0.8 in reinforced concrete based on AS3100 or 0.944 based on Austroads (2004).
f r is concrete flexural strength and can be determined from Equation 2-3 if a sufficient
curing method for concrete pavement is considered during construction. If no particular
curing method is considered, the f r can be determined using Equation 7-1.
The allowable load repetitions for each individual load of all axle groups during
daytime, nighttime and different loading conditions are then calculated using Equations
3-21 to 3-23. The allowable load repetitions are then adjusted using Equation 11-18.
allowable load repetition calculated from Equation 3 − 21 to 3 − 23

ALR K , M = (11-18)
Adjustment factor
Where K is loading condition which can be mid-edge, corner and centre loadings, M
denotes duration of the loading which can be daytime or nighttime, the Adjustment
factor is derived from Table 11-11.
Table 11-11. Adjustment factors for allowable load repetitions
SAST SADT TAST TADT TRDT QADT
Bonded Pavement 1 1 2 2 3 4
Unbonded JPCP 1 1 4 6 8 9
Pavement
JRCP 5 8 9 8 12 9
The adjustment factors provided for unbonded concrete pavements were determined
based on the results of the current study provided in Table 10-2. For bonded
pavements, the adjustment factors are equal to the number of axles in the axle group.
Since the lower band of fatigue curve was adopted in the Austroads fatigue damage
model, the method seems to be conservative. However, results of the laboratory fatigue
tests carried out in the current research, as reported in Chapter 7, show that the
280
estimation of concrete fatigue life in the Austroads method is about 10 times higher than
that recorded in the laboratory tests. This is due to the use of traditional fatigue setup for
developing of fatigue damage model in the Austroads method. As mentioned in Chapter
7 of this thesis, results of the traditional fatigue tests can not be accurately adopted in
the concrete pavement technology as pavement curvature is not taken into consideration
in the traditional fatigue test. Hence, it is strongly recommended to revise the concrete
fatigue model based on results of the laboratory tests of concrete prism beams using the
methodology described in Chapter 7 of this thesis.
Fatigue damages for each individual load of all axle groups during daytime and
nighttime are then calculated from Equations 11-19 to 11-21. The slab thickness is
appropriate if none of the fatigue damage modes exceed one.
 LRDE LRDC LRNE LRNC 

Fatigue DamageME = ∑∑  + + +  (11-19)
i j  AMELR D ACLRD AMELR N ACLR N  i , j
 LRDE LRDC LRNE LRNC 

Fatigue DamageCo = ∑∑  + + +  (11-20)
i j  ACoLR D ACLRD ACoLR N ACLR N  i , j
1  LRDC LRNC  
 ×  +  +
3  ACLR D ACLR N  
 
1  LRDE LRNE  
Fatigue Damage = ∑ ∑  ×  + +
j 3 AMELR N   (11-21)
EL
i  AMELR D
 
1  LRDE LRNE  
 3 ×  ACoLR D
+ 
ACoLR N  
 
Where subscripts ME, Co and EL denote mid-edge cracking, corner cracking and
longitudinal-edge cracking, respectively. LRDE, LRDC, LRNE, LRNC, i and j are as
previously defined. ACLRD, AMELRD, and ACoLRD are allowable load repetitions for
centre, mid-edge and corner loadings during day, respectively. ACLRN, AMELRN and
ACoLRN are allowable load repetitions for centre, mid-edge and corner loadings during
night, respectively.
281
11.12. An Example of the method
The procedure of the developed design guide is exemplified for designing of an
unbonded JPCP with following information:
- The pavement is confined at both longitudinal edges.
- Rural load distribution.
- Design traffic of 60 × 106 HVAGs.
- Project design reliability of 95%.
- Material strength reduction of 0.944.
- Percentage of traffic distribution during day is 40%.
- Percentage of traffic distribution during night is 60%.
- Percentage of traffic passing along edge area is 10%.
- Daytime differential temperature of 12ºC.
- Nighttime differential temperature of 8ºC.
- Load shift factors among axles of 10%.
- Concrete compressive strength of 40 MPa.
- Modulus of subgrade reaction of 0.05 MPa/mm.
- Thickness of concrete slab of 250 mm.
The summary of the fatigue analysis using the newly developed design procedure is
represented in Table 11-12.
Table 11-12. Results of the fatigue analysis for each axle group based on different
loading conditions
Types Centre Loading Mid-Edge Loading Corner Loading

of Axle
Group Daytime Nighttime Daytime Nighttime Daytime Nighttime
SAST 0.527 0.001 56.613 5.554 0.001 3.527

SADT 0.024 0.001 0.077 0.005 0 0.04
TAST 0 0 0 0 0 0.016
TADT 0.306 0.003 0.058 0 0.011 103.38
TRDT 0.171 0.001 0.04 0 0.002 1.448
Sum 1.028 0.006 56.788 5.559 0.014 108.411
282
Results indicate that a combination of corner loading of TADT and nighttime
differential temperature produces fatigue damage in the pavement. Using Equations 11-
19 to 11-21, the location of the fatigue damage is subsequently estimated.
Fatigue DamageME = 63.382%
Fatigue DamageCo = 109.467%
Fatigue DamageEL = 57.272%
These results indicate that the pavement is subjected to corner cracking. Furthermore,
the pavement may experience some mid-edge cracking together with longitudinal
cracking as well.
Using the provided information, the Austroads method (2004) is also implemented.
Results indicated that the slab thickness in terms of erosion and fatigue analyses are 176
mm and 208 mm, respectively. However, the minimum recommended slab thickness in
the Austroads design guide (2004), for a dowelled JPCP with the above mentioned
characteristics, is 250 mm. Hence, the slab thickness of 250 mm is the result of the
Austroads method.
A comparison between the slab thickness calculated from the guide developed in this
research and that provided by the Austroads shows the difference between the results.
While the slab thickness of 250 mm shows a good performance under the loading
conditions based on the Austroads design guide, it is not thick enough to resist fatigue
cracking based on the guide provided in this research.
It is interesting to mention that the prediction model of fatigue life provided by

Majidzadeh and Ilves (1983) was used in the fatigue analysis of this example.
Consequently, the fatigue life of concrete estimated in the guide provide in this thesis is
at least 4 times greater than that predicted by the Austroads method. Nevertheless, the
slab thickness calculated from Austroads method (2004) is lower than that calculated
based on the guide provided in this thesis. If the prediction model of fatigue life
provided by Austroads (2004) is involved in the newly developed slab thickness design,
the required slab thickness will be 315 mm.
283
11.13. Summary
Based on the results of this research presented in Chapters 5 to 11, a mechanistic-
empirical guide for designing of JPCP and JRCP was developed in this chapter. For this
purpose, some equations for predictions of induced stresses due to combinations of
vehicular axle group loads and environmental effects were firstly developed based on
mechanistic approach using LaGrange method. The accuracy of the stress prediction
models were then validated using finite element techniques.
For the first time worldwide, dynamic amplifications and the concept of stress
repetitions described in Chapter 11 of this thesis were factored into the design
procedure of concrete pavements. Using Miner’s rule and taking into consideration the
results of the field tests carried out worldwide, prediction models of fatigue damage for
corner, mid-edge, edge and longitudinal cracks were developed. Finally, the design
procedure was exemplified to demonstrate how the newly developed guide works.
284
Chapter 12
CONCLUSION AND RECOMMENDATATIONS FOR

FUTURE STUDY
12.1. Contribution from this research

The most significant contributions of this research are:
- A mechanistic-empirical approach for designing concrete pavements was
developed based on combinations of dynamic loads and environmental effects
together with different loading positions and boundary condition between
concrete slab and subbase.
- For the first time, typical dynamic amplification in JPCP and JRCP for each
individual axle group was calculated.
- In addition to what is accepted worldwide, it was determined that the number
of stress repetitions in concrete pavements due to moving axle group loads not
only depends on the type of axle group but it is also affected by the speed of
axle group, the type of concrete slab (reinforced concrete slab or plain concrete
slab) and the provision of debonding layer between concrete slab and subbase.
- Typical stress repetitions in JPCP and JRCP due to axle group loads were
determined.
- The critical speed for each individual axle group was determined.
- A new laboratory fatigue test setup was developed to take into consideration
effects of pavement curvature on concrete fatigue life.
- It is recommended that the use of debonding layer between concrete slab and
subbase in regions with harsh environmental condition be restricted.
- The critical position of each axle group on curled and uncurled pavement was
determined.
- The critical configuration for each axle group was determined.
285
12.2. Conclusion
A sensitivity analysis of the Austroads slab thickness design guide (2004) was
performed. Results of the current study indicate that the Austroads method (2004) has
several shortcomings and needs to be improved. The significant shortcomings can be
summarised as follows:
- In contrast to what is commonly accepted as the fundamental concrete

- Increasing the subgrade CBR above 5 per cent has no effect on slab thickness
- The minimum recommended slab thickness is greater than the calculated

thickness provided that the pavement is dowelled and restrained by shoulder or
adjacent traffic lanes.
- Vehicular loads are considered as static loads although they are dynamic in
nature.
- To account for environmental factors, the Guide provides a variety of

minimum concrete slab thickness for different types of concrete pavements and
a diverse range of design traffic. However, as shown in this thesis, the
behaviour of concrete pavements under vehicular loads depends on the
magnitude of differential temperature and/or loss of moisture contents. Hence,
consideration of environmental effects as a constant value, i.e. a minimum slab
thickness, may result in other failure types in the concrete slabs that are not
considered in the method.
- Flexural fatigue damage was assumed to only occur at the bottom surface layer
of the concrete slab.
performed to understand how debonding layer, axle group configurations, differential
stress within the concrete slab.
286
Results of the current study show that the benefits offered by consideration of the
unbonded boundary condition between concrete slab and subbase cease at a certain
value of differential temperature. Hence, articular care needs to be given to those
pavement projects constructed in hot or cold weather, where high differential
temperature gradients may be produced in concrete depth.
Furthermore, the critical dimensions of axle groups were determined. Subsequently,

some practical values for determining the critical axle group configuration were
provided. Critical positions of different axle groups in uncurled and curled jointed
concrete pavement with different configurations were also studied. Results of the
current study indicate that AASHTO recommendation (2003) and results of Packard
and Tayabji (1985) are valid for an uncurled pavement with a fully unbonded
boundary condition between the concrete slab and subbase. Results of the current study
also show that pavement performance under combinations of vehicular loads and
differential temperatures is significantly affected by the boundary condition between
the concrete slab and subbase.
The reasons behind longitudinal, transverse and corner cracking were addressed.
Depending on differential temperature between the top and the bottom surface layers
of the concrete slab, corner, centre and mid-edge loadings can result in different types
of fatigue failure in concrete slab. For instance, corner loading may enhance corner
cracking, transverse cracking at the edge or mid-edge of the pavement and longitudinal
cracking, depending on the differential temperature considered in the analysis. In terms
of maximum induced tensile stress, results of the current study show that corner
loading is critical in the presence of a bonded boundary condition between the concrete
slab and subbase. In an unbonded pavement, corner loading is also critical when a
separation due to environmental forces occurs between the unbonded concrete slab and
subbase.
Parametrical static analyses of the JPCP were also performed to define the effects of
concrete slab thickness and modulus of subgrade reaction on concrete pavement
behaviour. Results show that an inverse relationship exists between induced tensile
stress and thickness of concrete slab. In other words, an increase in the thickness of
concrete slab decreases the magnitude of induced tensile stress. However, this result is
not valid when a combination of vehicular loads and high differential temperature is
287
considered. Consequently, a maximum slab thickness in the presence of high
differential temperature between the top and the bottom surface layer of the concrete
slab was defined. A certain dowel arrangement at the corner of the concrete slab can
also eliminate the aforementioned problem. Hence, the use of longer dowel with
greater size and lower distance between dowels was recommended. Depending on the
boundary condition between the concrete slab and subbase, corner or mid-edge loading
and daytime or nighttime differential temperature, an increase in the modulus of
subgrade reaction may increase or decrease the magnitude of tensile stress.
To determine the most accurate equations in the prediction of concrete properties, a

series of laboratory test were performed. Results of the laboratory tests indicate that
the concrete modulus of elasticity can be accurately estimated by using equation
developed in the past. Furthermore, the equations provided for the estimation of
concrete flexural strength can be used when a specific curing method of concrete is
considered during pavement construction. In other words, the provided equations are
not able to accurately estimate the flexural strength of air cured concrete. As a result,
these equations are not sufficient to be used in concrete pavement unless an
appropriate curing method is provided. Consequently, a typical equation for the
prediction of flexural strength of air cured concrete was provided in the current
research based on the laboratory tests performed.
The shear transfer capability of aggregate interlock and cement paste was also
determined using notch prism beam. This property defines the capability of the
concrete for transferring the shear force across the initiated cracks and helps to
understand the behaviour of concrete at the initiated cracks.
Since the concrete pavement is curled upward during nighttime and downward during
daytime, it was questioned if the use of traditional fatigue setup may produce an
insufficient fatigue prediction model of the concrete. Consequently, a new fatigue
setup was developed to take into consideration the pavement curvature during daytime
and nighttime differential temperatures. Results of the fatigue laboratory tests
performed in the current study show that the equations developed in the past for
estimation of concrete fatigue life are not sufficient.
288
To study structural behaviour of concrete pavements, two types of concrete pavements
were considered. Dynamic analyses of bonded and unbonded JPCP and JRCP under
moving axle groups were performed in the next stage. The results of the current study
show that dynamic analysis is required to accurately predict the failure mode of
concrete pavements. Critical speeds of each axle group based on types of concrete
pavements were determined. For the first time, dynamic amplifications of each axle
group were presented in the current research. The critical locations for severe fatigue
cracking in both JPCP and JRCP were addressed. Results also indicated that fatigue
cracking is affected by axle group types and speed. It was determined in dynamic
analysis that the damage location may be close to transverse joints, at midpoint or in
some cases at quarter point of slab.
the dynamic studies performed in the current study was the determination of stress
repetitions in concrete pavement due to a given axle group. In the static analysis, the
number of stress repetition for a given axle group is equal to the number of axles in the
axle group. In other words, the number of stress repetitions in a point within the
concrete pavement for single axles, tandem axles, triple axle and quad axle groups are
one, two, three and four respectively. In the presence of bonded boundary condition
between concrete slabs and subbase, the aforementioned stress repetitions are still
correct in dynamic analysis. However, provision of debonding layer between concrete
slabs and subbase produce greater number of stress predictions in the dynamic
analysis.
This stress repetition phenomenon strongly depends on axle speed, type of axle group
and location where the stress is monitored. For SAST, SADT, TAST and TADT
higher speed, i.e. 110 km/h, produces greater stress repetitions than lower speeds. On
the other hand, in the heavy weight axle groups such as TRDT and QADT lower
speed, i.e. 30 km/h, produces greater stress repetitions than higher speeds. The average
number of stress repetitions for SAST, SADT, TAST, TADT, TRDT and QADT are 1,
1, 4, 6, 8 and 9 in JPCP and 5, 8, 9, 8, 12, and 9 in JRCP respectively.
Furthermore, effects of temperature fluctuation on the dynamic response of the

pavement were addressed. Results showed that differential temperature gradients may
increase or decrease the stress repetition for each axle group. However, greater tensile
289
stresses were observed in concrete pavements in the presence of differential
temperature.
Since results of the experimental field tests on dynamic response of concrete

pavements carried out in the past were not compatible with the findings of this
research, a fully instrumented concrete pavement test section including JPCP and
JRCP was constructed and tested under quasi-static and dynamic truck loadings.
Information on the test section, instrumentation layout, material properties and truck
characteristics were described. Pavement performance under environmental conditions
was studied during the first 28 days after casting. Truck loading was subsequently
applied at different predetermined locations of the pavement including at free
longitudinal edge, at the confined longitudinal edge and at the centre area of the traffic
lane. Time history responses were recorded for truck speeds between 5 km/h to 55
km/h.
Investigation of the recorded time history responses of the test section also indicates
the importance of dynamic analysis in concrete pavement design. The recorded time
histories validate the results of dynamic analysis performed in the current research.
Results also indicate that dowel position can strongly influence the pavement
responses. Furthermore, the slab deflection in JRCP decreases when reinforcement is
located close to the bottom surface layer of the concrete slab.
Results of the current study were used to develop a new empirical-mechanistic guide
for designing of concrete pavements. Consequently, typical equations for stress
prediction in concrete pavements for different loading conditions, differential
temperatures, slab thickness, modulus of subgrade reaction, and provision of shoulders
were developed. The accuracy of equations was then determined by comparing the
predicted stress with results of finite element analyses.
Using Miner’s rule, equations for calculating the fatigue damage of concrete slab were
developed. Transverse, corner and longitudinal cracks were contributed in the fatigue
damage model. Thickness of the concrete slab was considered to be adequate if none
of the above failure types were observed in the pavement. Ultimately, the design
procedure was exemplified.
290
12.3. Recommendations for future study
Since results of the laboratory tests performed in the current study showed that the
equations for prediction of flexural strength of the concrete provided in the past are not
valid if a sufficient curing method is not considered during pavement construction,
further study is required to determine an accurate equation for prediction of flexural
strength of air cured concrete.
In terms of fatigue prediction model, the results of the current study showed that the
fatigue models developed in the past are not sufficient to accurately estimate the
concrete pavement fatigue life as pavement curvature is not considered in the
laboratory tests. To overcome this problem, a new fatigue test setup was introduced in
the current research. Using the new fatigue test setup, further laboratory tests need to
be conducted to accurately develop a new fatigue prediction model for concrete
pavement.
Further studies are also required to determine effects of surface roughness, traffic
wander, length to width ratio of the concrete slab panel, width of longitudinal and
transverse joints and vertical position of reinforcement on dynamic structural response
of different concrete pavements under diverse transient axle group loads. Furthermore,
the effects of temperature fluctuations on stress repletion of each individual axle group
need to be investigated.
Since inclusive information on effects of shrinkage on pavement behaviour is not

established yet, there is a need for further study to investigate interrelationships among
shrinkage, relative humidity and deterioration of concrete pavements.
291
REFERENCES
AASHO, 1962. ‘The AASHO Road Test’, American Association of State and
Highway Officials, Report 5, Special Report 61E, Washington D. C., 1962.
AASHTO, 2003. ‘Appendix QQ (Structural Response Models for Rigid

Pavements) of Guide for Mechanistic-Empirical Design of New And
Rehabilitated Pavement Structures’, American Association of State and
Highway Transportation Officials, Prepared for National Cooperative Highway
Research Program- Transportation Research Board-National Research Council,
ARA Inc., Illinois, U. S. A.
Abo-Qudais S. A. and Al-Qadi I. L. (2000). "Dowel bars corrosion in concrete

pavement." Canadian Journal of Civil Engineering 27(6): 1240-1247.
ACI Committee 363 (1992). State of the Art Report on High Strength Concrete,
ACI Report 363R-92, American Concrete Institute, Farmington Hills, Michigan:
1-55.
Achenbach J.D., 1975. ‘Wave propagation in elastic solids’, Elsevier Science

Publishers BV, Amsterdam, The Netherlands.
ANSYS Help, Section 5.9.1., “Guidelines for Integration Time Step”, 2006.
Argyris J.H. and Kelsey S., 1960. ‘Energy Theorems and Structural Analysis’,
London, Butterworths
Armaghani J. M., 1987. ‘Comprehensive analysis of concrete pavement response

to temperature and load effects’, PhD thesis, University of Florida
AS 3600, 1990. ‘Supplement 1’, Concrete Structure – Commentary, Standards

Australia
AS 3600, 1994. ‘Concrete Structure’, Standards Australia
Austroads, 1992. ‘Pavement Design: A Guide to the Structural Design of Road

Pavements’, Austroads Inc., Sydney, Australia.
292
Austroads, 2004a. ‘Technical Basic of Austroads Pavement Design Guide’,
Austroads Inc., Sydney, Australia.
Austroads, 2004b. ‘AUSTROADS Pavement Design Guide’, Austroads Inc.,

Sydney, Australia
Barlow, J. A., 1994. ’Modeling damage to rigid pavements caused by subgrade

pumping’, MS thesis, University of Iowa, Iowa City, Iowa.
Beckemeyer C. A., Khazanovich L. and Yh H. T., 2002. ‘Determining amount of

built-in curling in jointed plain concrete pavement – case study of Pennsylvania
I-80’, Report on Design and Rehabilitation of Pavements 2002. Transportation
Research Board, N. R. C., Washington, pp. 85-92
Bendana L. J., McAuliffe D. and Yang W. S., 1994. ‘Mechanistic – Empirical

Rigid Pavement Design for New York State’, Transportation Research Record,
No. 1449, pp. 141-147
Bhatti M. A., Stoner J. W. and Hingtgen J., 1994. ‘Simulation of Dynamic Loads
from Different Vehicle Configurations’, International Journal of Heavy Vehicle
Systems, Vol. 1, No. 4, pp. 396-416
Bhatti M. A., Barlow J. A. and Stoner J. W., 1996. ‘Modeling Damage to Rigid
Pavement Caused by Subgrade Pumping‘, Journal of Transportation
Engineering, Vol. 122, No. 1, pp. 12-21.
Bhatti M. A., Stoner J. W., 1998, ‘Nonlinear Pavement Distress Model Using
Dynamic Vehicle Loads’, ASCE, Journal of Infrastructure System, Vol. 4, Issue
2, p. 71-78.
Biel T. D. and Lee H., 1997. ‘Performance study of Portland cement concrete
pavement joint sealants’, Journal of Transportation Engineering-ASCE123(5):
398-404.
Bischoff D. and Toepel A., 2002. ‘Dowel Bar Retrofit’, Report for Wisconsin
Department of Transportation
293
Blanksby C., George R. and Germanchev A., 2006. ‘An In-Service Survey of
Heavy Vehicle Suspensions’, 22nd ARRB Conference: Research into Practice,
Canberra, 29 October to 2 November, Australia
Bonnell D. G. R. and Harper F. C., 1951. ‘The thermal expansion of concrete’,

National Building Studies, Technical Paper No. 7, London
Bradbury R. D., 1938. ‘Reinforced concrete pavement’, Wire Reinforcement

Institute, Washington D.C., p. 34-41Byrum C. R., Hansen W., 1994, ‘Influence
Function Approach to Analysis of Jointed Portland Cement Concrete Pavement’,
Transportation Research Record 1449, TRB, Washington, p. 148-158
Buch N., Gilliland D., Van Dam T. and Vongchusiri K., 2004. ‘A Parametric
Mechanistic Evaluation of PCC Cross-Section Using ISLAB2000’, Final Report,
Michigan Department of Transportation, Lansing, MI
Byrum C. R. and Hansen W., 1994, ‘Influence Function Approach to Analysis of

Jointed Portland Cement Concrete Pavement’, Transportation Research Record,
1449, p. 148-158
Cement and Concrete Association of Australia, 1978. ‘Interim Concrete Road

Pavement Manual’, Cement and Concrete Association of Australia, Brisbane.
Cement and Concrete Association of Australia, 1991. ‘Concrete Pavement

manual Specification’, Cement and Concrete Association of Australia.
Cement and Concrete Association of Australia, 1999. ‘Industrial Floors and

Pavements Guidelines for Design-Construction and Specification’, Cement and
Concrete Association of Australia, NSW, Australia.
Channakeshava C., Barzegar F. and Voyiadjis G. Z., 1993. ‘Nonlinear FE

Analysis of Plain Concrete Pavements with Doweled Joints’, Journal of
Transportation Engineering, Vol. 119, No. 5, pp. 763-781
Chatti K., Lysmer J., and Monismith C.L., 1994. ‘Dynamic Finite-Element
analysis of Jointed Concrete Pavements’, Trans. Res. Record, Vol. 1449, pp.
79-90
294
Chen Y. H. and Deng X. J., 2001. ‘An analytical solution for an infinite
pavement strip on elastic foundation’, Canadian Journal of Civil Engineering
28(3): 509-519.
Chou Y. T., 1980. ‘Structural Analysis Computer Programs for Rapid

Multicomponent Pavement Structures with Discontinuities-WESLIQUID and
WESLAYER.’, Technical Report GL-80, US Army Engineering Waterways
Experiment, Station, Vicksburg, Miss.
Choubane B., Tia M., 1995, ‘Analysis and Verification of Thermal-Gradient

Effects on Concrete Pavement’, Journal of Transportation Engineering, ASCE,
Vol. 121, No.1, pp. 75-81
Clough R. W., 1960. ‘The Finite Element Method in Plane Stress Analysis’,
Journal of Structural Division, ASCE, Proceeding of 2d Conference on
Electronic Computation, pp. 345-378
Courant R., 1943. ‘Variational methods for the solution of problems of

equilibrium and vibrations’, Bulletin of American Mathematical Society, Vol.
49, pp. 1–43
Croll J. G. A., 2005. ‘Thermal Buckling of Pavement Slabs’, Proceeding of the

Institution of Civil Engineer, Transport 158, May 2005, Issue TR2, pp. 115-126
Croney M. D. and Croney P., 1998. ‘Design & Performance of Road

Pavements’, 3rd Edition, McGraw-Hill, New York.
Crovetti J. A.., 1996. ‘Field evaluation of support uniformity under jointed

concrete slabs’, Proc., in TRB 75th Annual Meeting. Transportation Research
Board, Washington D. C., U.S.
Crovetti J. A. and Bischoff D., 2001. ‘Construction and performance of

alternative concrete pavement designs in Wisconsin’, Design and Rehabilitation
of Pavements, Washington, TRANSPORTATION RESEARCH BOARD NATL
RESEARCH COUNCIL, pp. 43-53
295
Crovetti J. A.. and Crovetti M. R. T., 1994. ‘Evaluation of support condition
under jointed concrete pavement slabs’, Nondestrective testing of pavements and
backcalculation of moduli, ASTM. West Conshohocken, pp. 455-472.
Cruickshank J. w., 1981. ‘Report of Concrete highway Pavement Practice in

U.S.A. From U.S.A study Tour’, Cement & concrete Association Australia, New
South Wales
Dater M. I.,1977. ‘Design of Zero-Maintenance Plain Concrete Pavement’, Vol.

1- Development of design procedures’, FHWA, D.O.T., Office of Research,
Washington D. C., DOT-FH-11-8474
Darter M. I., Hall K. T. and Kuo C. M., 1995, ‘Support Under Portland Cement
Concrete Pavement’, National Cooperative Highway Research Program, Rep.
372, Transportation Research Board, National Research Council, Washington
D.C.
Davids W. G. and Mahoney J. P., 1999, ‘Experimental Verification of rigid

pavement joint load transfer modelling with EverFE’, Transportation Research
Record, 1968, pp. 81-88
Davids W. G. and Wang Z., 2003. ‘3D Finite Element Analysis of Jointed Plain
Concrete Pavement With EverFE2.2’, Transportation Research Broad, Annual
Meeting CD-ROM
De Beer M., Fisher C. and Jooste F. J., 1997. ‘Determination of pneumatic tyre-
pavement interface contact stresses under moving loads with some effects on
pavements with thin asphalt surfacing layers’, Proceedings of the 8th
International Conference on Asphalt Pavements, International Society for
Asphalt Pavements, 8-14 August, Seattle, Washington D. C., U.S. A., pp. 179-
227
Douglas R. A., Woodward W. D. H., Woodside A. R., 2000. ‘Road Contact

Stresses and Forces under Tires with low Inflation Pressure’, Canadian Journal
of Civil Engineering, Vol 27, pp. 1248-1258
296
Drucker D. C., 1957. ‘Plastic design methods—Advantages and limitations’.
Trans SNAME, Vol. 65, pp. 172–96
Drucker, D. C., Gibson, R. E. and Henkel, D. J., 1957. ‘Soil Mechanics and
Work-Hardening Theories of Plasticity’, Transactions, ASCE, Vol. 122, pp.
1692-1653
Erlicher, S., and Point, N., 2005. ‘On the associativity of the Drucker-Prager
model’, Proceeding VIII International Conference on Computation Plasticity
COMPLAS VIII. Eds: E. Onate, D.R.J. Owen. CIMNE, Barcelona, Spain.
Eddie D. and Shalaby A., 2001. ‘Glass fiber-reinforced polymer dowels for
concrete pavements’, ACI Structural Journal, Vol. 98, No. 2, pp. 201-206
FHWA, 1983. ‘Tolerance of Dowel Bar Placement’, Department of

transportation, Federal Highway administration, office of Research and
development, Washington D. C.
Fwa, T. F., Shi X. P., and Tan S. A., 1996. ‘Analysis of Concrete Pavements by
Rectangular Thick –Plate Model’, Journal of Transportation Engineering,
ASCE, Vol. 122, No. 2, pp. 146-154
Gillespie T. D., Karamlhas S. M., Cebon D., Sayers M. W., Nasim M. A.,
Hansen W., and Ehsan N., 1993. ‘Effects of Heavy Vehicle Characteristics on
Pavement Response and Performance’, Final Report, Prepared for National
Cooperative Highway Research Program- Transportation Research Board-
National Research Council, The University of Michigan Transportation Research
Institute 1992, Michigan, U. S. A.
Guclu A., Ceylan H., 2005, ‘Sensitivity Analysis of Rigid Pavement System
Using Mechanistic-Empirical Pavement Design Guide’, Proceeding of the 2005
Mid-Continent Transportation Research Symposium, Ames, Iowa, USA
Guo E. H., 2003. ‘Proof and comments on extensively used assumption in PCC
Pavement Analysis and Evaluation, Jourbal of Transportation Engineering,
ASCE, March/ April, pp. 219-220
297
Hammons M. I., Pittman D. W., and Mathews D. D., 1995. ‘Effectiveness of
load transfer devices’, DOT/FAA/AR-95/80, Federal Aviation Administration,
Washington D. C.
Handson S. W. and Seeds S. B., 1988. ‘Evaluation of Increased Pavement

Loadind and Tyre Pressure’, 67th Annual Meeting of the Transportation Research
Board, Washington D. C., U. S. A.
Hanson W., Smiley D. L., Peng Y. F. and Jensen E. A., 2002. ‘Validating top-
down premature transverse slab cracking in jointed plain concrete pavement’,
Design and Rehabilitation pf Pavements, Washington, Transportation Research
Borad, N. R. C., pp. 52-59
Harvey J. T., Popescu L., Ali A., and Bush D., 2003. ‘Performance of dowel bar
retrofitted concrete pavement under heavy vehicle simulator loading’, Pavement
Rehabilitation and Accelerated Testing 2003. Washington,
TRANSPORTATION RESEARCH BOARD NATL RESEARCH COUNCIL,
pp. 141-152
Heath A. C. and Roesler J. R., 1999, ‘Shrinkage and Thermal Cracking of Fast
Setting Hydraulic Cement Concrete Pavements in Palmdale, California’,
Preliminary Report Prepared for California Department of Transportation
Heath A. C. and Roesler J. R., 2000. ‘Top-down cracking of rigid pavements

constructed with fast-setting hydraulic cement concrete’, TRANSPORTATION
RESEARCH BOARD NATL RESEARCH COUNCIL, Washington, pp. 3-12
Heath A. C., Roesler J. R. and Harvey J. T., 2003, ‘Modelling Longitudinal,

Corner and Transverse Cracking in Jointed Concrete Pavements’, International
Journal of Pavement Engineering, Vol. 4(1), pp. 51-58
Hendrick J.K., Marlow M.J., and Brandemeyer B., 1992. ‘The Simulation of
Vehicle Dynamic Effects on Road Pavements’, FHWA, Report No: FHWA-
RD-90-108
298
Hiller E. J. and Roesler J. R., 2002, ‘Transverse Joint Analysis for Mechanistic-
Empirical Design of Rigid Pavements’, Transportation Research Record, 1809,
pp. 42-51
Hiller J. E. and Roesler J. R., 2005. ’determination of critical concrete pavement

fatigue damage locations using influence lines’, Journal of Transportation
Engineering, Vol. 131, No. 8, pp. 599-607
Hogg, A., and Hall A, 1938. ‘Equilibrium of a Thin Plate Symmetrically

Loaded, Resting on an Elastic Subgrade of Infinite Depth’. Philosophical
Magazine., Series 7,Vol 25, pp. 576-582
Hossain M. and Wojakowski J. B., 1996. ‘Material and Construction Portland

cement concrete pavements, Bridges, and Quality management’, Transportation
Research Record 1544: Federal Highway Administration.
Hossain A. B., Pease B., and Weiss J., 2003. ‘Quantifying Early-Age Stress
Development and Cracking in Low Water-to-Cement Concrete’, Journal of
Concrete Material and construction 1834: TRANSPORTATION RESEARCH
BOARD.
Hrenikoff A., 1941. ‘Solution of Problems in Elasticity by the Framework

Method’, Transactions of the ASME, Journal of Applied Mechanics, Vol. 8, pp.
169-175
Huang, Y. H., 1993. ’Pavement Analysis and Design’, 1st ed., Prentice Hall,
Upper Saddle River, NJ
Huang Y. H., 2004. ‘Pavement Analysis and Design’, Second Edition, Pearson
Prentice Hall, Australia, pp. 398-401
Huang Y. H. and Wang T., 1973. ‘Finite Element Analysis of Concrete Slabs
and its implication for rigid pavement design’, Highway Research Record 466,
National Research Council, Washington D.C., pp. 55-69
Hu H. C., 1981. ‘Variational Principal and Its Application in Elastics’, Science

Press, China, 1981, pp. 465-477
299
Hudson W. R., and Matlock H., 1966. ‘Analysis of Discontinuous Orthotropic
Pavement Slabs Subjected to Combined Loads’, Highway Research Record,
Vol. 131, pp. 1-48
Ioannides, A. M., Thompson M. R., and Barenberg E. J., 1985. ‘Westergaard

Solutions Reconsidered’. Transportation Research Record, No. 1043, National
Research Council, Washington D.C., pp. 13-23
Ioannides A. M. and Khazanovich L., 1998. ‘Nonlinear Temperature Effects on

Multilayered Concrete Pavements’, Journal of Transportation Engineering Vol.
124, No. 2, pp. 128-136
Izquierdo J. T., Rodrigues L., and Rios B. C., 1997, ‘Structural Evaluation and
Analysis of Instrumented In-Service Concrete Pavements Subjected to Heavy
Dynamic Loads’, Transportation Research Record, No. 1568, Transportation
Research Board, Washington D. C., USA, pp. 24-34
Jensen E. A. and Hansen W., 2000. ‘Fracture energy test for highway concrete -
Determining the effect of coarse aggregate on crack propagation resistance’,
Issues in Pavement Design and Rehabilitation, TRANSPORTATION
RESEARCH BOARD NATL RESEARCH COUNCIL, Washington, pp. 10-17
Jensen E. A. and Hansen W., 2002. ‘Crack Resistance of Jointed Plain Concrete
Pavements’, Design and Rehabilitation of Pavements 1809, pp. 60-65
Jo B. E., 1988. ‘A finite element parametric study for the response of concrete
highway pavements with skewed joints’, PhD thesis, University of Florida
Karimov I., ‘Fatigue strength and Cracking of Concrete’, http://ufa.ru [online

source]. http://ufa.ru/~ikar/Articlefatigue1.html [ Access 28/09/2004]
Kelleher K. and Larson R. M., 1989. ‘The Design of Plain Doweled Jointed
Concrete Pavement’, Proceedings of the Fourth International Conference on
Concrete Pavement Design and Rehabilitation, Purdue University, West
Lafayette, Indiana, pp. 279-292
300
Kerr A. D., 1964. ‘Elastic and Viscoelastic Foundation Models’, Journal of
Applied Mechanics, Transportation, ASME, Vol. 31, No. 3, pp. 491-498.
Khazanovich L., Selezneva O. I., Yu H. T., and Dater M. I., 2001. ‘Modeling of
jointed plain concrete pavement fatigue cracking in PaveSpec 3.0.’, Design and
Rehabilitation of Pavements 2001, Washington, TRANSPORTATION
RESEARCH BOARD NATL RESEARCH COUNCIL: 33-42.
Kim J. K. and Kim Y. Y., 1996. ’Experimental Study of the Fatigue Behaviour
of High Strength Concrete’, Cement and Concrete Research, Vol. 26, No. 10, pp.
1513-1523
Kim S. M., Won M. C., and McCullough B. F., 2000. ‘Three-dimensional

analysis of continuously reinforced concrete pavements’, Issues in Pavement
Design and Rehabilitation, Transportation Research Board NATL Research
Council, Washington D. C., U. S. A., pp. 43-52
Kim S.M., Won M.C., and McCullough B.F., 2001. ‘Transformed Field
Domain Analysis of Pavements Subjected to Moving Dynamic Tandem Axle
Loads and Integrating their effects into the CRCP-10 Program’, FHWA, Report
No. FHWA/TX-0-1831-5
Kim S.M., Won M.C., and McCullough B.F., 2002. ‘Dynamic Stress Response
of Concrete Pavements to Moving Tandem-Axle Loads’, Transportation
Research Record, No. 1809, Transportation Research Board, Washington D. C.,
U. S. A., pp. 32-41
Klaiber F. W. and Lee D. Y., 1982. ‘The effect of Air Content, Water-Cement
Ratio, and Aggregate Type on the Flexural Fatigue Strength of Plain Concrete’,
Fatigue of Concrete, SP-41, ACI, pp. 401
Kuo C. M., 1998, ‘Effective Temperature Differential in Concrete Pavements’,

Journal of Transportation Engineering, March/April, pp. 112-116
Kumara, W., M. Tia, C.-L. Wu, B. Choubane, 2002. ‘Evaluation of

Applicability of Ultrathin Whitetopping in Florida. Transportation Research
Record, No 1823, National Research Council, Washington D.C., pp. 39-46
301
Larralde j., 1984. ‘Structural analysis of rigid pavements with pumping’, PhD
dissertation, Purdue University, West Lafayette, Ind.
Lee Y. H., 1999. ‘TKUPAV: Stress Analysis and Thickness Design Program for
Rigid Pavements’, ASCE, Journal of Transportation Engineering, Vol. 125, No.
4, pp. 338-346
Lee Y. H., Bair J. H., Lee C. T., Yen S. T. and Lee Y. M., 1997. ‘Modified
Portland Cement Association Stress Analysis and Thickness Design Procedure’,
Transportation research Record, No. 1568, pp. 77-88
Lee Y. H. and Carpenter S.H., 2001. ‘PCAWIN program for Jointed Concrete
Pavement design’, Tamkang Journal of Science and Engineering, Vol. 4, No. 4,
pp. 293-300
Lennie S., and Bunker J., 2005. ‘Assessing the spatial impact of multi-
combination vehicles on an urban motorway’, MSc. Thesis, Queensland
University of Technology, Australia
Liang R. Y. and Niu Y. Z., 1998, ‘Temperature and Curling stress in Concrete
Pavements: Analytical Solutions’ , Journal of Transportation Engineering,
ASCE, Vol. 124, No. 1, pp. 91-100
Lippmann S. A., 1985. ‘Effects of tire structure and operation conditions on the
distribution of stress between the tread and the road’, in the Tire Pavement
Interface, ASTM STP 929, American Society for Testing and Materials,
Philadelphia, U.S.A., pp. 91-109
Liu C., McCullough B. F., and Oey H. S., 2000, ‘Response of Rigid Pavements
due to Vehicular-Road Interaction’, ASCE J. of Transp. Engrg., Volume 126,
Issue 3, pp. 237-242.
Liu C. and Gazi D.,1999. ‘Surface Roughness Effect on Dynamic Response of

Pavements’,ASCE J. Transp. Engrg., Volume 125, Issue 4, pp. 332-337.
Majidzadeh K., Ilves G. J., 1983. ‘Evaluation of Rigid Pavement Overlay Design
Procedure: Development of the OAR Procedure’,DTFH11-9489
302
Marshek K. M., Chen H. H., Connell R. B., and Hudson, R. W., 1986.
‘Experimental determination of pressure distribution of truck tire-pavement
contact’, Transportation Research Record, No. 1070, Transportation Research
Board, National Council, Washington D. C., U. S. A., pp. 9-14
Masad E., Taha R. and Muhunthan B.,1996. ‘Finite Element Analysis of

Temperature Effects on Plain Jointed Concrete Pavements’, Journal of
Transportation Engineering, ASCE 122.5, p. 388-398
Meyerhof, G. G., 1962. ‘Load-carrying capacity of concrete pavements’, J. Solid

Mechanics Foundations Div., 88(3), pp. 89–116
Meyers S. L., 1951. ‘How temperature and moisture changes may effect the
durability of concrete’, Rock Products, Chicago, USA, pp. 153-157
Mindlin, R. D., 1951. ‘Influence of Rotatory Inertia and Shear on Flexural

Motion of Isotropic Plates’, Journal of Applied Mechanics, Vol. 23, pp 31-38
Mohamed, A. R., Hansen W., 1997. ‘Effect of nonlinear temperature gradients

on curling stress in concrete pavements’, Transportation Research Record, No.
1568, pp. 65-71
Monismith C.L., Lysmer J., Sousa J., and Hedrick J.K., 1998. ‘Truck Pavement
Interaction: Requisite Research’, University of California Transportation
Centre, SAE Technical Paper Series 48, Berkeley.
Nam J. H., Kim S. M., McCullough B. F., and Dossey T., 2003. ‘Sensitivity
Analysis of CRCP Computer Programs’, Research Report 1700-4, Centre for
Transportation Research Bureau of Engineering Research, The University of
Texas, Austin, U.S.
Nawy E. G., 2001. ‘Fundamentals of High-Performance Concrete’, 2nd Ed.: John

Wiley & Sons Inc., New York, p.p. xiii to xvi.
Neville A. M., 1981. ‘Properties of concrete’, Pitman Publishing Inc.,

Melbourne, Australia, pp. 592 -600
303
Neville A.M., 1983. ‘Properties of Concrete’, 3rd Edition, The English
Language Book Society and Pitman Publishing, London
Neville A. M. and Brooks J. J., 1987. ‘Concrete Technology’, Longman

Scientific & Technical, Longman Group UK Limited
Okamoto P. A., Packard R. G., 1989, ‘Effect of High Tire Pressure on Pavement
Performance’, PROCEEDINGS, 4th international conference on concrete
pavement design and rehabilitation, Purdue University, USA, pp. 61-74
Okamoto, P.A., Nussbaum, P.J., Smith, K.D., Darter, M.I., Wilson, T.P., Wu,
C.L. and Tayabji, S.D., 1994, ‘Guidelines for Timing Contraction Joint Sawing
and Earliest Loading for Concrete Pavements’, Publication No. FHWA-RD-91-
079 and -080, Federal Highway Administration, USA
Older C., 1924. ‘Highway Research in Illinois’, Transportations, ASCE, Vol.

87, pp. 1180-1222
Ongel A. and Harvey J., 2004, ‘Analysis of 30 Years of Pavement Temperature

using the Enhanced Integrated Climate Model (EICM)’, Draft report prepared
for California Department of Transportation
Packard R. G. and Tayabji S. D., 1983. ‘Mechanical Design of Concrete

Pavements to Control Joint Faulting and Subbase Erosion’, International seminar
on Drainage and Erodability at the Concrete Slab-Subbase-Shoulder Interfaces,
Paris, France
Packard R. G. and Tayabji S. D., 1985. ‘New PCA Thickness Design Procedure
for Concrete Highway and Street Pavements’, Concrete Pavement &
Rehabilitation Conference, Purdue, USA
Panda B. C. and Ghosh A. K., 2002. ‘Structural behaviour of concrete block

paving. I: Sand in bed and joints’, Journal of Transportation Engineering-ASCE
128(2), pp. 123-129
304
Pasternak P. L., 1954. ‘Fundamentals of New Method of Analysis of Structures
on Elastic Foundation by Means of Two Subgrade Coefficients’,
Gosudarstvennoe Izdatel’stvo Literatury po Stroitel’stvu I Arkhitekture, Moscow
Pickett, G., and Ray G. K., 1951. ‘Influence Charts for Concrete Pavements’,
Transactions, ASCE, Vol. 116
Ping W. V., Yang Z. and Gao Z., 2002. ‘Field and Laboratory Determination of
Granular Subgrade Moduli’, ASCE, Journal of Performance of Constructed
Facilities, pp. 149-159
Popehn, N. A., Schultz A. E. and Snyder M. B., 2003. ‘Performance Testing of

Experimental dowel Bar Retrofit designs- Part2: repeatability and Modification
Design’, Minnesota Department of Transportation, U.S.
Pronk A. C., 1993. ‘The Pasternak Foundation: An Attractive Alternative for the
Winkler Foundation’, Proceedings, Fifth International Conference on Concrete
Pavement Design and Rehabilitation
Ramakrishnan, V. and Lokvik, B. L., 1992. ‘Flexural Fatigue Strength of Fiber

Rienforced Concrete’, Proceedings, International RILEM/ACI Workshop,
Stuttgard, Chapman &Hall, London, pp. 271-287
Rasmussen R. O. and McCullough B. F., 1998. ‘A Foundation for High

Performance Jointed Concrete Pavement Design and Construction Guidelines’,
Transtec consultant, Austin, Texas, USA
Reddy A., Leonards G. A. and Harr M. E., 1963. ‘Warping stress and deflections
in Concrete Pavements’, Part III. In Highway Research Report 44, HRB,
National Research Council, Washington D.C.
Reddy J. N., 1993. ‘An introduction to the Finite Element method’, 2nd edition,
McGRAW-Hill International editions, London, pp. 3-10
Richardson M. and Armaghani J. M., 1990, ‘Stress caused by temperature

gradient in Portland cement concrete pavements’, Transportation Research
Record, 1121, p. 7-13
305
Reissner, E, 1945. ‘Effect of Transverse Shear Deformation on Elastic Plates’,
Journal of applied Mechanics, Vol. 12, pp. 69-77
Reissner, E, 1950. ‘On a Variational Theorem in Elasticity’. Journal of

Mechanics and hysics, pp. 90-95
Reissner, E, 1958. ‘A Note on Deflections of Plates on a Viscoelastic

Foundation’, Journal of Applied Mechanics, ASME, Vol. 80, pp. 144-145
Roads and Traffic Authority (RTA) of NSW, 1991. ‘Concrete Pavement Manual
– Design and Construction’, MATERIALS SERVICES BRANCH.
Road and Traffic Authority (RTA) of NSW, 1998. ’Vehicle Dimension Limits’,
Booklet No. 5, Vehicle standards, Driver and Vehicle Policy Branch, Roads &
Traffic Authority of New South Wales, Australia
Roesler J. R., Harvey J. T., Farvar J., Long F., 2000, ‘Investigation of Design
and Construction Issues for Long Life Concrete Pavement Strategies’, Report
Prepared for California Department of Transportation, Pavement Research
Centre, Institute of Transportation Studies, University of California at Berkeley
Rozycki D. K. and Rasmussen R. O., 1998. ‘Assessment of slab-base

interaction in PCC airfield pavements’, Airport Fac.: Innovations for The Next
Century, Proc., 25th Int. Air Transp. Conf., M. T. McNerney, ASCE, New
York, 403–422
Sayers M. W., Gillespie T. D. and Queiroz C. A., 1986. ‘The international Road
Roughness Experiment: Establishing Correlation and a Calibration Standard for
Measurements’, Technical Paper 46, The Word Bank, Washington D.C., USA
Selezneva O., Rao C., Darter M., Zollinger D., Khazanovich L., 2004.
‘Development of a Mechanistic-Empirical Structural Design Procedure for
Continuously Reinforced Concrete Pavements’, Transportation Research Board
83rd Annual Meeting, Washington D. C.
Schindler A. K. and McCullough B. F., 2002. ‘Importance of concrete

temperature control during concrete pavement construction in hot weather
306
conditions’, Construction 2002, TRANSPORTATION RESEARCH BOARD
NATL RESEARCH COUNCIL Washington, pp. 3-10
Shackel B., 1993. ‘Load and Environmental Factors Affecting Industrial

Pavements’, School of Civil Engineering, The University of New South Wales,
Australia
Sharma, A.K. and Pandey, K.P., 1996, ‘The Deflection and Contact
Characteristics of Some Agricultural Tyres with Zero Sinkage’, Journal of
Terramechanics, 33(6), pp. 293-299
Shi X. P., Fwa T. F. and Tan S. A., 1993. ‘Warping Stresses on Concrete
Pavements on Pasternak Foundation’, Journal of Transportation Engineering,
Vol. 119, No. 6, pp. 905-913
Shi X. P., Tan S. A., and Fwa T. F., 1994. ‘Rectangular Thick Plate With Free
Edges on Pasternak Foundation’, Journal of Engineering Mechanics, Vol. 120,
No. 5, pp. 971-988
Shoukry S. N. and Fahmy M. R., 2002, ‘Optimization of Concrete slab

Geometry for Enhanced Rigid Pavement Performance and Service Life’,
Department of Mechanical and Aerospace Engineering , College of Engineering
and Mineral Resources, West Virginia University, WVDOT Research Project
No. 140
Shoukry A. N., William G. W. and Riad M., 2002, ‘Characteristics of Concrete

Contact Stresses in Doweled Transverse Joints’, The International Journal of
Pavement Engineering, Vol. 3 (2), pp. 117-129
Siddique Z. Q., Hossain M., and Meggers D., 2005, “Temperature and Curling
Measurement of Concrete Pavement’, Proceeding of the 2005 Mid-Continental
Transportation Research Symposium, Ames, Iowa, August 2005, Iowa State
University, USA, p. 1-12
Smith K.D., Mueller A. L., Dater M. I., and Peshkin D. G., 1990. ‘Performance
of Jointed Concrete Pavements’, Volume II-Evaluation and Modification of
307
Concrete Pavement Design and Analysis Models, FHWA-RD-89-137, Federal
Highway Administration, Washington
Sousa D. O. e, Antunes J. L. and Ravindra G., 2006. ‘Determining the Tensile

Stress-Crack Opening of Concrete by Inverse Analysis’, ASCE, Journal of
Engineering Mechanics, Vol. 132, No. 2, pp. 141-148
Stoner J. W. and Bhatti M. A., 1994. ‘Estimating Pavement Damage From

Longer and Heavier Combination Vehicles’, Midwest Transportation Center,
Iowa
Stoner J. W., Bhatti M. A., Kim S. S., Koo J. K., Molinas-Vega I. and Amhof B.,
1990. ‘Dynamic Simulation Methods for Evaluating Motor Vehicle and
Roadway Design and Resolving Policy Issues’, MIDWEST Transportation
Center, The U.S. transportation system and Iowa Department of Transportation
Stott J. P., 1961. ‘Test on materials for use in sliding layers under concrete road
slabs’, Civ. Engrg., 56(663), 1297–1299, 1301; (669), 1466–1468; (655),
1603–1605
Suh C., Lee J. L. Y., Fowler D. W. and Stokoe K. H., 2005. ‘Superaccelerated
pavement testing on full-scale concrete slabs’, transportation research record,
No. 1940, pp. 113-124
Suh Y. C., Lee S. W. and Kang M. S., 2002. ‘Evaluation of subbase friction for
typical Korean concrete pavement’, Design and Rehabilitation of Pavements
2002. Washington, TRANSPORTATION RESEARCH BOARD NATL
RESEARCH COUNCIL: 66-73.
Tabatabaie, A. M. and Barenberg E. J., 1978. ‘Finite Element Analysis of

Jointed of Cracked Concrete Pavements’, Transportation Research Record,
No. 671, TRB, National Research Council, Washington D.C. 1978, pp. 11-17
Tabatabaie A. M., Barenberg E. J. and Smith R. E., 1978. ‘Analysis of load

transfer system for concrete pavements’, FAA-RD-79-4, Federal Aviation
Administration, Washington D. C.
308
Tabatabaie A. M. and Barenberg E. J., 1980. ‘Structural analysis of concrete
pavement systems’, Journal of Transportation Engineering, ASCE, Vol 106,
No. 5, pp. 493-506
Tang T. X., Zollinger D. G. and Sendheera S., 1993. ‘Analysis of Concave

Curling in Concrete Slabs’, Journal of Transportation Engineering-ASCE , Vol.
119, No. 4, pp. 618-633
Tarr S. M., Okamoto P. A., Sheehan M. H., and Packard R. G., 1999, ‘Bond
Interaction Between Concrete Pavement and Lean Concrete slab’, Transportation
Research Record, 1668, pp. 9-17
Tayabji, S. P., and B. E. Colley, 1981. ‘Analysis of Jointed Concrete

Pavements. Federal Highway Administration, National Technical Information
Service
Tayabji S. D. and Colley B. E., 1983. ‘Improved Rigid Pavement Joints’,

Transportation Research Record, Vol. 930, pp. 69-78
Tayabji S. D., and Colley B. E., 1986. ‘Improved Rigid Pavement Joints’,
FHWA/RD 86/040, Federal Highway Administration, Washington D. C.
Tia, M., Armaghani M., Wu C. L., Lei S., and Toye K. L., 1987. ‘FEACONS
111 Computer program for an Analysis of Jointed Concrete Pavements’,
Transportation Research Record, No 1136, National Research Council,
Washington D.C., pp.12-22
Tia M., Wu C. L., Ruth B. E., Bloomquist D., and Choubane B., 1989. ‘Field
Evaluation of Rigid Pavements for the Development of a Rigid Pavement
Design System-Phase1V’, Final Report, State Project # 99700-7359-010,
Florida Department of Transportation
Treybig H. J., McCullough B. F., Smith P. and Quintus H. V., 1977. ‘Overlay
Design and Reflection Cracking Analysis for Rigid Pavements’, Federal
Highway Administration, FHWA-RD-77-76, Washington D. C.
309
Turner M., Clough R. W., Martin H. H. and Topp L., 1956. ‘Stiffness and
Deflection Analysis of Complex Structures’, Journal of Aeronautical Science,
Vol. 23, pp. 805-823
Vesic A. S. and Saxena S. K., 1969. ‘Analysis of Structural Behavior of Road

Test Rigid Pavements’, Highway Research Record, No. 291, pp. 156-158
Vlasov V. Z. and Leontev N. N., 1960. ‘Beam, Plates and Shells on Elastic
Foundation’, NSA-NSF, NASA TT F-357, TT 65-50135, Israel Program for
Scientific Transportations (Translation date: 1966).
Vora M. R., and Matlock H. A., 1970. ‘A Discrete-Element Analysis for

Anisotropic Skew Plates and Grids’, Research Report No. S6-18, Center for
Highway research, University of Texas, Austin
Vorobieff G., 1996. ‘Rigid Pavement Thickness Design by Spreadsheet’,

Pavement Design & Performance Unsealed Roads, Quality, Combined 18th
ARRB Transport Research Conference and Transit NZ Land Transport
Symposium, New Zealand.
Vorobieff G., 2001, ‘Recent Development in Australia in Concrete Pavement

Thickness Design’, 7th International Conference on Concrete pavements,
Orlando, Florida, USA
Wang L. G. and Zollinger D. G., 2000. ‘ Mechanistic Design Framework For

Spalling Distress’, Journal of Pavement Design and Rehabilitation,
Transportation Research Board, N. R. C., Washington, pp. 18-24
Wesevich J. W., McCullough B. F. and Burns N. H., 1987. ‘Stabilised Subbase

Friction Study for Concrete Pavements’, Research Report 459-1, Centre for
Transportation Research, University of Texas, Austin, USA
Westergaard H. M., 1926, ‘Analysis of stress in concrete pavements due to

variations of temperature’, Proc., 6th Ann. Meeting, Highway Research Board,
Washington D.C., pp. 201-205
310
Westergaard, H. M., 1926. ‘Analysis of Stresses in Concrete Roads’,
Proceedings of Highway Research Board, Vol. 5, pp. 90-112
Westergaard, H. M., 1927. ‘Theory of Concrete Pavement Design’,

Proceedings of Highway esearch Board, Vol. 7, Part 1, pp. 175-181
Westergaard, H. M., 1933. ‘Analytical tools for Judging Results of Structural

Tests of Concrete Pavements’, Public Roads, Vol. 14, No. 10, pp. 185-188
Westergaard, H. M., 1939. ‘Stresses in Concrete Runways of Airports’,

Proceedings of Highway Research Board, Vol. 19, pp. 197-202
Westergaard, H. M., 1943. ‘Stresses Concentration in Plates Loaded Over

Small Area’, Transactions, ASCE, Vol. 108, pp. 831-856
Westergaard, H. M., 1947. ‘New Formula for Stresses in Concrete Pavement of

Airfields’, American Society of Civil Engineers, ASCE, Vol. 113, pp. 425-444
Wimsatt A. W., McCullough B. F. and Burns N. H., 1987. ‘Methods of

Analysing and Influencing Frictional effects of Subbase’, Research Report 459-
2F, Centre for Transportation Research, University of Texas, Austin, USA
Winkler E., 1864. ‘Die Lehre Von der Elastizitt und Festigkeit’, Theory of
Elasticity and Strength, H. Dominicus, Prague.
Yoder E. J. and Witczak M. W., 1975. ‘Principles of Pavement Design’, 2nd

Edition, Wiley Interscience, New York, U.S.A.
Yu T. H., Darter M. I., Smith K. D., Jiang J. and Khazanovich L., 1997.
‘Performance of Concrete Pavements’, Volume III-Improving Concrete
Pavement Performance, FHWA-RD-95-111, Federal Highway Administration
Yu T. H., Khazanovich L., Darter M. I., and Ardani A., 1998, ’Analysis of
Concrete Pavement Responses to Temperature and Wheel Loads Measured from
Instrumented Slabs’, Transportation Research Record, 1639, pp. 94-101
311
Zaghloul S., and White T., 1993. ‘Non-Linear Dynamic Analysis of Concrete
Pavements’, 5th Int. Conf. on Conc. Pav. Design & Rehabilitation, Purdue
University, 1993, pp. 277-292
Zaman M. and Alvappillai A., 1995. ‘Contact-Element Model for Dynamic

Analysis of Jointed Concrete Pavements’, Journal of Transportation
Zhang J., Fwa T. F., Tan K. H. and Shi X. P., 2003. ‘Model for Nonlinear
thermal effect on pavement Warping Stresses’, Journal of Transportation
Zhang J. and Li V. C., 2001, ‘Influence of Supporting Base Characteristics on

Shrinkage-Induced Stresses in Concrete Pavements’, Journal of Transportation
Engineering, November/December, USA, p. 455-462
Zollinger D. G., Senadheera S. P. and Tang T. X., 1994. ‘Spalling of

Continuously Reinforced-Concrete Pavements’, Journal of Transportation
Engineering-ASCE, Vol. 120, No. 3, pp. 394-411
Zwerneman F. J., Donahey R. C., Syed H. S., and Gunna S. R., 1995. ‘Cracking
of concrete pavement continuously reinforced with epoxy-coated steel’, ACI
Materials Journal, Vol. 92, No. 6, pp. 678-685
312

Response of Concrete Pavements Under Moving Vehicular Loads and Environmental Effects

Uploaded by

Copyright:

Available Formats

Response of Concrete Pavements Under Moving Vehicular Loads and Environmental Effects

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Response of Concrete Pavements Under Moving Vehicular Loads and Environmental Effects

Uploaded by

Copyright:

Available Formats

RESPONSE OF CONCRETE PAVEMENTS UNDER

MOVING VEHICULAR LOADS AND

Mostafa Yousefi Darestani

A dissertation submitted for the degree of

Centre of Built Environment and Engineering Research

Queensland University of Technology

Despite progressive knowledge of concrete pavement behaviour under applied loads,

International Refereed Journal Papers:

International Refereed Conference Papers:

“A Review of 2004 Austroads Rigid Pavement Design”, 22nd ARRB Conference –

“Experimental Study on Structural Response of Rigid Pavements under Moving Truck

Submitted Papers to International Refereed Journals:

“Sensitivity Analysis of the 2004 Austroads Design Guide”, Concrete Institute of

Preparation of Journal Papers:

“Structural dynamic analysis of unbonded JPCP Subjected to Moving Axle Group

“Structural dynamic analysis of unbonded JPCP Subjected to Moving Axle Group

Sustainability of concrete pavements depends on the rate of pavement deteriorations.

- In contrast with what is commonly accepted as the fundamental concrete

- To account for environmental factors, the Guide provides a variety of

To address the above mentioned shortcomings, parametrical studies of the concrete

Since performance of concrete pavements is strongly affected by concrete properties,

The significant properties of concrete used in concrete pavement technology are

Furthermore, effects of temperature fluctuation on the dynamic response of the

Since results of the experimental field tests on dynamic response of concrete

Number Description Page

Figure 2-1 Fatigue relationship adopted in Austroads design model 14

Figure 2-2 Variation of fatigue life with concrete compressive strength 15

Figure 2-4 Effect of different types of fibres on concrete fatigue flexural 17

Figure 3-1 Cross section of a typical concrete pavement in Australia 19

Figure 3-2 DL subgrade idealization 21

Figure 3-3 Spring model idealization 21

Figure 3-4 Limitation of the spring model 21

Figure 3-5 Winkler modified idealization 22

Figure 3-6 Elastic solid (ES) idealization 22

Figure 3-7 TP model idealization 23

Figure 3-8 Kerr model idealization 24

Figure 3-9 Typical experimentally determined frictional stress versus 27

Figure 3-10 Typical expansion joint 31

Figure 3-11 Typical isolation joint 31

Figure 3-12 Typical keyed and tied longitudinal joints 34

Figure 3-13 Typical components for calculation of LTE in a transverse 35

Figure 3-15 Illustration of Packard and Tayabji’s (1985) assumption 38

Figure 3-17 Configurations of traffic lanes considered in the AASHTO 39

Figure 3-18 Configurations of concrete slabs in the longitudinal direction 40

Figure 3-19 Effect of number of slabs in the transverse direction on 40

Figure 3-20 Effect of number of slabs in the longitudinal direction on 41

Figure 3-21 Axle group types considered in Austroads 2004 42

Figure 3-22 Transverse section of a typical SADT 43

Figure 3-23 Length and width of tyre–pavement contact area 45

Figure 3-24 Effects of daytime and nighttime temperature gradient on 47

Figure 3-25 Effects of daytime and nighttime temperature gradient on 48

Figure 3-26 Pavement modelling in EverFE finite element program 54

Figure 3-27 8-noded finite element setup in Totsky model 57

Figure 3-28 Schematic interface element behaviour and interface 58

Figure 3-29 Comparison between damage rate: moving constant load 61

Figure 3-30 Comparison between fatigue cracking along slab length: 62

Figure 3-33 Critical load location for CRCP punchout crack 71

Figure 3-34 Erosion mechanism 72

Figure 3-36 Estimation of effective subgrade CBR 77