Load Transfer Across Joints

Loughborough University
Institutional Repository
Load transfer across cracks

and joints in concrete slabs
on grade
This item was submitted to Loughborough University's Institutional Repository
by the/an author.
Additional Information:
A Doctoral Thesis. Submitted in partial fulllment of the requirements

for the award of Doctor of Philosophy of Loughborough University.
Metadata Record: https://dspace.lboro.ac.uk/2134/7581
Publisher: c Stuart John Arnold

Please cite the published version.

This item is held in Loughborough Universitys Institutional Repository
(https://dspace.lboro.ac.uk/) and was harvested from the British Librarys
EThOS service (http://www.ethos.bl.uk/). It is made available under the
following Creative Commons Licence conditions.
For the full text of this licence, please go to:

http://creativecommons.org/licenses/by-nc-nd/2.5/
LOAD TRANSFER ACROSS CRACKS AND JOINTS IN
CONCRETE SLABS ON GRADE
By
Stuart John Arnold
A Doctoral Thesis Submitted in partial fulfilment of the requirements for the

award of
Doctor of Philosophy of Loughborough University
20thFebruary 2004
@Stuart John Arnold (2004)

ABSTRACT
This research has investigated the behaviour of joints and cracks under single and
multiple cycles of load. This provides an increasedunderstandingof concrete slab on

gradeperformance,enabling more effective designand monitoring procedures.
Examination of the geometry of cracks and joints within concrete slabs on grade has
demonstratedthat the commonly assumedparallel formation is erroneous.Measurements
using embeddedstrain gauges,coring and surfaceprofile levelling have uncoveredthat a

high percentageof joints will contain larger crack widths at the surfacethan at the base,
causedby differential shrinkage.The opening itself is relatively linear; however, the top
50mm of the slab is prone to a higher gradient of movementdue to the increaseddrying
effect towards the surface.
A series of deflection tests using a Falling Weight Deflectometer and Prima dynamic
plate enabledslab responseunder load to be evaluated.Four sites were examinedin total
and correlations found between: load transfer, load step, edge cantilever and crack
geometry. This produced valuable information regarding the influence of load transfer
and crack width on the overall slab behaviour.Foundationvoiding and crack face free slip
was also shown to influence deflection magnitude.
A small-scale test facility was developed for the assessmentof deterioration in various
'V' shapedand parallel crack widths under high cycle loading. The data demonstrated
that joint/crack failure contains four distinct phasesof deterioration, each of which is
controlled by a different mechanism. 'V' shapedcracks produced a much greater load
transfer than that of a parallel crack with the incorporation of A142 mesh and steel fibres
reducing differential displacement.Load magnitude and aggregatesize were also shown
to have significant effects. The value of reinforcement was found to assist with
serviceability requirements,keeping displacementwithin acceptablelevels and preventing
the onset of seriousdegradation
A finite element model was developedto enablethe load transfer mechanismresults from
the laboratory test to be used in the assessmentof full slab response.Simulations of field-
testing produced a series of lower bounds in respect to deflections and the associated
responsecalculations. Theoretical behaviour of a typical slab was assessedwith subbase
support, joint stiffness, slab thickness and the incorporation of a subbase,found to be
highly influential in reducing slab deflections.
The three main sections of work comprising site data collection, laboratory testing and
Finite Element modelling have been used together to provide a much greater
understandingof the influence of cracks andjoints. This has included the deterioration of
cracks over time and an examination of how this and other site-based factors affect
overall slab behaviour.
Keywords: Concrete slabs on grade, Joint behaviour, Load transfer, Deflection testing,
Steelfibres, Concretedegradation, Finite Elementmodelling.
ii
ACKNOWLEDGMENTS
Many people have provided me with assistance,guidanceand support over the last three
years,without whom it would not have beenpossibleto completethis thesis.
Firstly, I would like to thank Professor Simon Austin, Dr. Paul Fleming and Dr. Peter
Robins, who have complementedeach other perfectly to generatethe model supervisory
team.Their knowledge,advice and carefully worded constructivecriticism have enhanced
both my academicand personaldevelopment.
I would also like to thank many other staff members in the department of Civil and
Building Engineering, most notably the expertise of the laboratory technicians Mark
Harrod, Dave Spendloveand David Houghton. Together they have shown that with the
inclusion of experienceand wisdom, practical researchcan be both highly rewarding and
remarkably pain free.
My thanks go to the ACIIFC and all of the steering group members whose input has
enabled a project to be developedwhich will hopefully offer valuable information to the
wider constructioncommunity.
There are many friends and family who deservepraise for their input; however, two in
particular standout, Adrian Holland and Karan Jalota. Both have beenthere academically
and emotionally through the good and the bad, and amazingly are still here today, proving
that they are very specialfriends.
Heartfelt thanks go to my parentsMaureen and David Arnold whose faith in me has been
a source of continual strength. They have supported and encouraged me

endlesslyandfor thatI love themdearly.
To complete this PhD was not a solitary undertaking, there was always one other who
went through every emotion and has probably read and re-read every word as many times
as I have. To show my total appreciation towards my partner Chrissie Pepper,I dedicate
this thesisto her.
iii
CONTENTS OF THESIS
ABSTRACT i
.........................................................................................................................
ACKNOWLEDGEMENTS iii
...............................................................................................
CONTENTS OF TBESIS iv
..................................................................................................
LIST OF TABLES ix
.............................................................................................................
LIST OF FIGURES x
.............................................................................................................
NOTATION xviii
...................................................................................................................
1. INTRODUCTION I
.......................................................................................................
1.1 Background 1
........................................................................................................
1.2 Aims and Objectives 2
..........................................................................................
1.3 Outline of Research Methodology 2
....................................................................
1.4 Thesis Structure 4
.................................................................................................
2. CONCRETE SLABS/PAVEMENTS ON GRADE 8
.................................................. ..
2.1 Preface 8
............................................................................................................... ..
2.2 Concrete Slabs/Pavements on Grade 9
...............................................................
2.2.1 Internal Floor Slabs 9
.................................................................................... ..
2.2.2 External Hardstandings 12
..............................................................................
2.2.3 Concrete Pavements 14
...................................................................................
2.2.3.1 Jointed 15
........................................................................................................
2.2.3.2 Continuously Reinforced 15
............................................................................
2.2.4 Structure Comparison 15
................................................................................
2.3 Typical Actions for Concrete Slabs on Grade 16
...............................................
2.3.1 Moisture 17
.....................................................................................................
2.3.1.1 Drying Shrinkage 17
.......................................................................................
2.3.1.2 Plastic Shrinkage 18
.......................................................................................
2.3.2 Thermal 18
......................................................................................................
2.3.3 Static Load 19
.................................................................. ...............................
2.3.4 Cyclic Dynamic Load 20
................................................................................
2.4 Serviceability Problems 21
...................................................................................
2.4.1 Cracking 21
................................................................... ..................................
2.4.2 Curling 23
............................................................. ..........................................
2.4.3 Pumping 26
........................................................ .............................................
2.4.4 Faulting/Dynamic Load Step 26
....................... ...................................... ........
2.4.5 Punch-Out 27
..................................................................................................
iv
2.4.6 Spalling 28
......................................................................................................
2.5 Concrete Slab Design 29
.......................................................................................
2.5.1 Concreteproperties 29
....................................................................................
2.5.2 Slab Reinforcement 29
....................................................................................
2.5.2.1 General 29
..................................................................................................
2.5.2.2 Fabric 30
....................................................................................................
2.5.2.3 Fibres 32
.....................................................................................................
2.5.2.4 Structural Reinforcing steel 33
...................................................................
2.5.3 Structural Design 33
.......................................................................................
2.5.3.1 Elastic method 34
.......................................................................................
2.5.3.2 Plastic Method 35
.......................................................................................
2.5.4 Detail Design 37
.............................................................................................
2.5.4.1 Crack Control 37
........................................................................................
2.5.4.2 Joint Types 39
.............................................................................................
2.5.4.3 Joint Layout 41
...........................................................................................
2.6 Slab/Pavement Foundation 43
.............................................................................
2.6.1 Subbase 43
......................................................................................................
2.6.2 Subgrade 4
..
....................................................................................................
2.6.3 Slip Membrane 45
...........................................................................................
2.7 Structural Modelling 46
.......................................................................................
2.7.1 Modelling of Concrete materials 46
................................................................
2.7.2 Modelling of Foundation materials 47
............................................................
2.7.3 Analysis of Slabs on Grade 49
........................................................................
2.7.4 Modelling Limitations 51
...............................................................................
2.8 Conclusions 52
.......................................................................................................
3. LOAD TRANSFER ACROSS CRACKS/JOINTS 59
..................................................
3.1 Introduction 59
......................................................................................................
3.2 Influence on Design 59
..........................................................................................
3.3 Crack Geometry 60
...............................................................................................
3.4 Load Transfer Mechanisms 62
............................................................................
3.4.1 Aggregate Interlock 62
...................................................................................
3.4.2 Dowel Bars 65
.................................................................................................
3.4.3 Steel Fabric 66
................................................................................................
3.4.4 Steel Fibres 67
.................................................................................................
3.4.5 Proprietary Systems 68
...................................................................................
3.5 Degradation of Load Transfer 69
........................................................................
3.6 Slab Condition Testing 71
....................................................................................
V
3.6.1 Equipment 71
..................................................................................................
3.6.2 Investigation Techniques 72
...........................................................................
3.7 Joint Effectiveness Analysis 75
............................................................................
3.8 Summary 80
...........................................................................................................
4. SLAB CONDITION 90
.................................................................................................
4.1 Introduction 90
......................................................................................................
4.2 Crack Measurement 91
........................................................................................
4.2.1 SurfaceMeasurements 92
...............................................................................
4.2.2 EmbeddedStrain Gauges 93
...........................................................................
4.2.3 Coring 93
........................................................................................................
4.3 CurlingfWarping 94
..............................................................................................
4.3.1 Introduction 94
................................................................................................
4.3.2 PreciseLevelling 95
........................................................................................
4.3.3 Builder's Level 95
...........................................................................................
4.3.4 Profilometer 96
...............................................................................................
4.4 Deflection Measurement 96
..................................................................................
4.5 Slab Analysis 98
.....................................................................................................
4.5.1 Load Transfer 99
..
...........................................................................................
4.5.2 Load Step 99
..
.................................................................................................
4.5.3 Edge Cantilever 100
........................................................................................
4.5.4 Deflection Basins 100
.....................................................................................
4.5.5 Void Intercepts 101
.........................................................................................
4.6 Site Information ............................................................................................. 101
4.6.1 Daventry 101
...................................................................................................
4.6.2 Lutterworth 102
..............................................................................................
4.6.3 Ballymena 103
................................................................................................
4.6.4 Skelmersdale 104
............................................................................................
5. JOINT DETERIORATION ..................................................................................... 113
5.1 Introduction 113

....................................................................................................
5.2 Test Variables ................................................................................................. 114
5.2.1 SurfaceCrack Widths 114
..............................................................................
5.2.2 Steel Fibre Types 114
.....................................................................................
5.2.3 Steel Fibre Quantities 115
...............................................................................
5.2.4 Non Reinforced 115
........................................................................................
5.2.5 Mortar 115
......................................................................................................
5.2.6 Parallel Cracks 115
.........................................................................................
5.2.7 Fabric 115
.......................................................................................................
vi
5.2.8 Loads 115
........................................................................................................
5.2.9 Summaryof Test Variables 116
......................................................................
5.3 Load Test Development .................................................................................
116
5.3.1Simulation Method 116
...................................................................................
5.3.2 SpecimenGeometry 118
.................................................................................
5.3.3 Crack Geometry 118
.......................................................................................
5.3.4 SubgradeSupport 119
.....................................................................................
5.3.5 Load Magnitude 119
.......................................................................................
5.3.6 Loading repetition 120
....................................................................................
5.3.7 Load rate 121
..................................................................................................
5.4 Specimen Production 121
.....................................................................................
..............................................................................................
5.4.2 Mix Design 122
...............................................................................................
5.4.3 Reinforcement 123
..........................................................................................
5.4.4 Casting 124
.....................................................................................................
5.5 125
Beam Preparation ..........................................................................................
5.5.1 Crack Timing 125
...........................................................................................
5.5.2 Crack Technique 125
......................................................................................
5.5.3 Crack Width Control 127
................................................................................
5.6 Test procedure 127
................................................................................................
5.6.1 Overview 127
..................................................................................................
5.6.2 Beam Orientation 128
.....................................................................................
5.6.3 Rig Configuration 128
....................................................................................
5.6.4 LVDT Positioning 129
....................................................................................
5.6.5 Safety Precautions 130
....................................................................................
5.7 130
Data Logging ..................................................................................................
6. RESULTS, ANALYSIS & DISCUSSION ............................................................ 146
.
6.1 146
General Introduction .....................................................................................
6.2 Crack Geometry 146
............................................................................................
.
6.2.1 Core Samples 146
..........................................................................................
.
6.2.2 Strain GaugeMonitoring ........................................................................ 151
.
6.2.3 Levelling Profiles and Crack Measurements 154
..........................................
.
6.2.4 Summary 156
.................................................................................................
.
6.3 Slab Condition 157
...............................................................................................
.
.
.............................................................................................
6.3.2 Load Transfer Evaluation 158
.
........................................................................
6.3.3 Load Step Evaluation 162
.
..............................................................................
vii
6.3.4 Slab Cantilever 165
.........................................................................................
6.3.5 Deflected Shape 166
.......................................................................................
6.3.6 Voiding 166
....................................................................................................
6.3.8 Sununary 167
..................................................................................................
6.4 Joint Deterioration 168
.........................................................................................
6.4.1 SpecimenProduction 169
...............................................................................
6.4.2 SpecimenQuantities 171
................................................................................
6.4.3 Deterioration Phenomena 173
.........................................................................
6.4.4 Influence of Initial Crack Angle 175
..............................................................
6.4.5 Volume of Fibre Reinforcement 177
..............................................................
6.4.6 Fibre Aspect Ratio 178
...................................................................................
6.4.7 Load Magnitude 180
.......................................................................................
6.4.8 Parallel Cracks 180
.........................................................................................
6.4.9 ReinforcementType 181
.................................................................................
6.4.10 Serviceability Limitations 182
........................................................................
6.4.11 Summaryof Laboratory Investigation 184
.....................................................
6.5 Summary of Field and Experimental Work 185
................................................
7. NUMERICAL MODELLING 232
................................................................................
7.1 Introduction 232
....................................................................................................
7.2 Model Set-up 232
...................................................................................................
7.3 Model Verification 233
.........................................................................................
7.4 Comparison to Laboratory testing 234
...............................................................
7.5 Relationship with Site Data 236
...........................................................................
7.6 Comparison of Laboratory Obtained joint Stiffness 238
.................................
7.7 Effect of Constituent Material Parameters 240
.................................................
7.8 Summary 243
.........................................................................................................
S. CONCLUSIONS AND RECOMMENDATIONS FOR FURTHER WORK 264
.........
8.1 Overview 264
.........................................................................................................
8.2 Site Testing 264
.....................................................................................................
8.3 Laboratory Testing 265
........................................................................................
8.4 Analytical Modelling 266
......................................................................................
8.5 Final Comments 267
.............................................................................................
8.6 Recommendations for Further Work 267
..........................................................
9. REFERENCES 269
........................................................................................................
viii
LIST OF TABLES
Table 2.1 - Typical Industrial WarehouseLoading Values (Knapton 1999b) 10

..................
Table 2.2 - Typical HardstandingLoading Values (Knapton 1999a) 13
...............................
Table 2.3 - StructureComparison 16
.....................................................................
................
Table 2.4 - Stock Fabric Sizes(Knapton 1999a) 30
...............................................................
Table 2.5 - ProblemsAssociatedwith Joints and Cracks (Simpson2001a) 38
. ....................
Table 2.6 - RecommendedJoint Spacing(Knapton 1999b) 42
..............................................
Table 3.1 - Load Transfer Comparison(Gulden and Brown 1985) 79
...................................
Table 4.1 - Site Testing Matrix 91
.........................................................................................
Table 5.1 - Testing Schedule 116
...........................................................................................
Table 5.2a - Typical Site Mixes 122
......................................................................................
Table 5.2b - Actual ConcreteMix Specification Used for Laboratory Testing 122
..............
Table 5.2c - Material Information 123
.......................................................................
.............
Table 5.3 - Steel Fibre Data 124
................................................................................
.............
Table 6.1 - CMT Core Information (Visit 1) 147
.............
......................................................
Table 6.2 - CMT Core Information (Visit 2) 147
.............
......................................................
Table 6.3 ComparisonBetweenActual and PredictedSurfaceCrack Widths 151
- .............
Table 6.4a Strain GaugeReadingsConvertedto Crack Width 152
- .....................................
Table 6.4b - Strain GaugePlacement 152
.............
..................................................................
Table 6.5 -Measured and ProjectedSurfaceMeasurements 153
.............
.............................
Table 6.6 - Effect of Free Slip on Load Transfer 160
.............
...............................................
Table 6.7 - Load Step Variation (Lutterworth) 163
.............
..................................................
Table 6.8 - ConcreteCompressiveStrengths 169
.............
.....................................................
Table 6.9 - Beam SpecimensTested 172
..................................................................
.............
Table 6.10 - Typical Fibre Count acrossCrack Face 179
.........................................
.............
Table 6.11 - Differential Deflection at PhaseIII 182
.............................................................
Table 6.12 - Allowable Crack Widths Preventing Serviceability Problems 183
...................
Table 7.1 - StandardSlab Basic Parameters 241
....................................................................
ix
LIST OF FIGURES
Figure 1.1- Data transfer betweensite data collection, laboratory data collection and
Finite Element modelling 7
.....................................................................................................
Figure 1.2 - Interaction betweenthesisChapters 7
................................................................
Figure 2.1 - Typical constructionof an internal floor slab 54
................................................
Figure 2.2 - Concreteslab constructionmethods(Knapton 1999b) 54
..................................
Figure 2.3 - Rigid pavementslip form construction(Wirtgen 2001) 54
................................
Figure 2.4 - Shrinkagerestraint of a concreteslab 55
............................................................
Figure 2.5 - Peeling action causedby fabric reinforcement(Savage1985) 55
......................
Figure 2.6 Temperatureinduced slab curling (Ytterberg 1987) 55
- .....................................
Figure 2.7 - Rotation of joint faces(Pobletteet aL 1988) 56
.................................................
Figure 2.8 - Curling profiles of a concreteslab causedby differential shrinkage(Rollings
1993) 56
..................................................................................................................................
Figure 2.9 - Faulting (top) and dynamic load step (bottom) acrossa discontinuity 57
..........
Figure 2.10 - Punch out accumulation 57
...............................................................................
Figure 2.11 - Spalling of a crack edge 57
...............................................................................
Figure 2.12 Steel fibre types (Balaguru and Shah 1992) 58
- ................................................
Figure 2.13 - Load/deflection relationship for steel fibre ground supportedslab (Concrete
Society 2000) 58
.....................................................................................................................
Figure 2.14 - Relationship betweenCBR and modulus of subgradereaction (Croney and
Croney 1997) 58
.....................................................................................................................
Figure 3.1 - Effect of crack verticality 82
..............................................................................
Figure 3.2 - Global (macro) and local (micro) roughness 82
.................................................
Figure 3.3 - Aggregate interlock model (Walraven 1981) 82
................................................
Figure 3.4 - Effect of crack opening on aggregatecontact (Walraven 1981) 83
...................
Figure 3.5 - Effect of crack width on net shearslip (Thompson 2001) 83
.............................
Figure 3.6 - Load/shearslip plot for non-reinforced specimen(Thompson2001) 84
...........
Figure 3.7 - Load/shearslip plot for fibre reinforced specimen(Thompson2001) 84
..........
Figure 3.8 - Joint deterioration of a 175mmslab (Colley and Humprey 1967) 85
................
Figure 3.9 - Joint deterioration of a 225mm slab (Colley and Humprey 1967) 85
................
Figure 3.10 - load transfer values from laboratory and field testing (Colley
and Hurnprey
1967) 85
..................................................................................................................................
Figure 3.11 - Effect of load on joint deterioration (Colley and Humprey 1967) 86
..............
Figure 3.12 - FWD (Top - graphical representation,Bottom Plate) 86
- ...............................
Figure 3.13 - Prima dynamic loading plate 87
.......................................................................
x
Figure 3.14 - FWD deflection basin analysis 87
....................................................................
Figure 3.15 - Void closure under load 88
...............................................................................
Figure 3.17 - Geophonelocations (Ricci et aL 1985) 89
.......................................................
Figure 3.18 - LT versusTLE (loannidesand Korovesis 1990) 89
.........................................
Figure 319 - LT Vs AGG (loannidesand Korovesis 1990) 89
..............................................
Figure 4.1 - Crack width prediction from saw-cutjoints 106
................................................
Figure 4.2 - Crack microscopemeasuringa saw-cutjoint 106
..............................................
Figure 4.3 - Set of callipers measuringa saw-cutjoint 106
...................................................
Figure 4.4 - Set of crack comparators 107
.............................................................................
Figure 4.5 - Extrapolation of embeddedstrain gaugesto calculate surfaceand basecrack
107
widths...............................................................................................................................
Figure 4.6 Overestimationof basecrack width causedby differential shrinkage 108
- ........
Figure 4.7 - Estimation of slab curl using builders level and graduatedwedge 108
..............
Figure 4.8 - Estimation of slab curl using Profilometer 109
..................................................
Figure 4.9 - FWD and Prima geophoneplacementfor measurementof slab behaviour
acrosscracks andjoints ....................................................................................................

109
Figure 4.10 - FWD geophonelocations for Daventry (Top) and Lutterworth (Bottom) 110
Figure 4.11 - Load transfer comparisonwhen loaded eachside of a joint 110
......................
Figure 4.12 - Plan of S.W. comer of Daventry site III
.........................................................
Figure 4.13 - Plan of Lutterworth site Ill
.............................................................................
Figure 4.14 - Plan of Skelmersdalesite 112
...........................................................................
Figure 5.1 - Cyclic loading test set-up(Schematic) 132
........................................................
Figure 5.2 - Cyclic loading test set-up(Plate) 132
.................................................................
Figure 5.3 - Steel fabric and reinforcing bar layout and positioning 133
...............................
Figure 5.4 - Typical single crack test specimen(Valle and Buyukozturk 1993) 133
............
Figure 5.5 - Single cycle load test set-up(Millard and Johnson1984) 134
...........................
Figure 5.6 - Large-scaleslab laboratory set-upfor cyclic load testing (Colley and
Humphrey 1967) 134
..............................................................................................................
Figure 5.7 - Double crack test set-upfor cyclic loading (Thompson2001) 135
....................
Figure 5.8 - Representationof in service slab loading using positive and negative
laboratory loading 136
............................................................................................................
Figure 5.9 - Reduction in sheardisplacementgradient over increasingload cycles (Abdel-
Maksoud 2000) 137
................................................................................................................
Figure 5.10 - Reduction in peak and trough deflection gradient with increasingload
cycles (Thompson 2001).................................................................................................. 137

Figure 5.11 - Trial testing differential displacementplot, 0- 250,000 cycles (Top),
250,000 - 500,000 cycles (Bottom) 138
.................................................................................
xi
Figure 5.12 - Measuredload cycle of a vehicle crossinga discontinuity at 30mph (Colley
and Humphrey 1967).......................................................................................................

139
Figure 5.13 - Sinusoidalload application from the Dartec cyclic load test machine 139
......
Figure 5.14 - Bekaert Dramix@ Steelfibres usedwithin the experimentation(from left:
RC-65/60-13N,RC-80/60-BN, RL-45/50-BN) 139
.................................................................
Figure 5.15 - Method of inducing cracksinto test specimens 140
.........................................
Figure 5.16 - Extrapolation of demecpip measurementsto obtain surfaceand basecrack
width measurements 141

.........................................................................................................
Figure 5.17 - Comparisonof predictedand actual surfacecrack measurements............141
Figure 5.18 - Measurementof crack openingthroughout specimendepth.....................142
Figure 5.19 - Graphical representationof data logging process...................................... 142
Figure 5.20 - LVDT positioning (original set-up) 143
...........................................................
Figure 5.21 - LVDT positioning (updatedset-up) .......................................................... 143
Figure 5.22 - Effect of undertakinga discontinuouscyclic load test..............................144

Figure 5.23 - Maximum and minimum displacementpositions...................................... 144
Figure 5.24 - Positive and negativedisplacementdeteriorationplot ..............................145

Figure 5.25 - Differential displacementdeteriorationplot .............................................. 145
Figure 6.1 - Lutterworth Core Locations (Top - visit 1, Bottom - visit 2)......................187
Figure 6.2 - Core taken during visit I (Clockwise from top left, Core 7, Core 17, Core 24,
Core 21) 188
...........................................................................................................................
Figure 6.3 - Crack profiles from corestaken in visit 1 188
....................................................
Figure 6.4 - Crack profiles from cores taken in visit 2.................................................... 189
Figure 6.5 - Crack offsets from corestaken in visit I ..................................................... 190
Figure 6.6a - Strain GaugePositions (Leeds) 190

..................................................................
Figure 6.6b - Strain GaugePositions (Northampton)...................................................... 191
Figure 6.6c - Strain GaugePositions (Marston).............................................................. 191

192
Figure 6.7a - Crack width extrapolation (Leeds) ............................................................
192
Figure 6.7b - Crack width extrapolation (Marston) ........................................................
Figure 6.8 - Actual crack profile using embeddedstain gaugesand surfacemeasurements
193
(Leeds).............................................................................................................................
Figure 6.9 - Differential shrinkageof concreteopen to the environment on one side
(Neville 1995) 194
..................................................................................................................
Figure 6.10 - Erroneousstrain gaugereading from B ishop (2001) 194
.................................
Figure 6.11 - Area of floor level surveyedwith a precise level (Daventry) 195
....................
Figure 6.12 - Floor surfaceprofiles (Daventry) ..............................................................
195
Figure 6.13 - Measurededgecurl using a builders level (Ballymena) 196
............................
Figure 6.14 - Predictedcrack profile using slab levels and surfacewidth (Daventry).... 196
xii
Figure 6.15 - Predictedcrack profiles using edgecurl (Ballymena) 197
...............................
Figure 6.16 - Load transfercomparisonbetweenFWD and Prima (Daventry) 197
..............
Figure 6.17 - Load transfercomparisonbetweenFWD loads (Lutterworth) 198
..................
Figure 6.18 - Load transfercomparisonbetweenFWD and Prima (Lutterworth) 198
..........
Figure 6.19 - Effect of free-slip on load transfer.............................................................
199
Figure 6.20 - Comparisonof load transfer equations......................................................
199
Figure 6.21a- Comparisonof load transfer and crack width (Daventry)........................200
Figure 6.21b- Comparisonof load transfer and crack width (Lutterworth) ...................200
Figure 6.22 - Comparisonof load transfer with surfaceand baseslab crack widths
(Lutterworth) 201
....................................................................................................................
Figure 6.23 - Load stepcomparisonbetweenFWD and Prima (Daventry)....................202
Figure 6.24 - Load stepcomparisonbetweenFWD load magnitudes(Lutterworth) ......202
Figure 6.25 - Load stepcomparisonbetweenFWD and Prima (Lutterworth) ................203
Figure 6.26 - Effect of voiding on load magnitude/ load stepgradients........................203
Figure 6.27 - Effect of load on absolutedeflection ......................................................... 204
Figure 6.28 - Comparisonof load transfer and load step

(Daventry/Lutterworth/Ballymena/Skelmersdale) 204
...........................................................
Figure 6.29a- Comparisonof load stepand crack width (Daventry) ............................. 205
Figure 6.29b - Comparisonof load stepand crack width (Lutterworth) .........................205
Figure 6.29c - Comparisonof load step and crack width (Skelmersdale).......................206
Figure 6.30 - Comparisonof load stepand crack width taking into accountpre-load
(Daventry) 206
........................................................................................................................
Figure 6.3 1a- Comparisonof load transfer and edgecantilever (Daventry)..................207
Figure 6.31b- Comparisonof load transfer and edgecantilever (Lutterworth) ..............207
Figure 6.31c - Comparisonof load transfer and edgecantilever (Ballymena)................208
Figure 6.31d- Comparisonof load transfer and edgecantilever (Skelmersdale)...........208
Figure 6.32a- Comparisonof load step and edgecantilever (Daventry)........................209
Figure 6.32b - Comparisonof load step and edgecantilever (Lutterworth) ...................209
Figure 6.32c - Comparisonof load step and edgecantilever (Ballymena) .....................210
Figure 6.32d - Comparisonof load stepand edgecantilever (Skelmersdale).................210
Figure 6.33 - Typical deflection bowls (Daventry) ........................................................
.
211
Figure 6.34 - Typical deflection bowls (Lutterworth) ....................................................211
.
Figure 6.35 - Effect of casting date on cube compressivestrength 212
.................................
212
Figure 6.36 - Deterioration phasesof a concretecrack..................................................
.
Figure 6.37 - Early deterioration in a concretetest specimen .
213
........................................
213
Figure 6.38 - Unequal displacement..............................................................................
.
Figure 6.39 - Influence of aggregateon displacement .
214
...................................................
xiii
Figure 6.40 - Negative deteriorationcausedby small material accumulation 214
................
Figure 6.41 - Effect of aggregateand steelfibres on differential displacement 215
..............
Figure 6.42a- Effect of deteriorationon displacementresistance(30kg1irOsteel fibre with
1.98mm.crack width) 215
.......................................................................................................
Figure 6.42b- Effect of deteriorationon displacementresistance(30kg/msteel fibre
with 5.94mm crack width) ...............................................................................................

216
Figure 6.43a- Effect of deteriorationon displacementresistance(non-reinforcedwith
0.66nun.crack width) 216
.......................................................................................................
Figure 6.43b- Effect of deteriorationon displacementresistance(non-reinforcedwith
1.98mmcrack width) 217
.......................................................................................................
Figure 6.44 - Effect of deteriorationon displacementresistance(mortar with 0.66nim
217
crack width) .....................................................................................................................
Figure 6.45 - Effect of steelfibre quantity on differential displacement 218
........................
Figure 6.46 Effectivenessof fibre quantity in resistanceto displacement 218
- ....................
Figure 6.47a- Effect of deteriorationon displacementresistance(20kg/msteel fibre with
0.66mm crack width) 219
.......................................................................................................
Figure 6.47b Effect of deterioration on displacementresistance(20kg/m3steel fibre
-
with 4.62mm crack width) 219
...............................................................................................
Figure 6.48a Effect of deterioration on displacementresistance(40kg/msteel fibre with
-
1.98mm.crack width) 220
.......................................................................................................
Figure 6.48b - Effect of deterioration on displacementresistance(40kghnsteel fibre

220
Figure 6.49 - Effect of steel fibre aspectratio on differential displacement 221
...................
Figure 6.50 - Comparisonbetweensteel fibre aspectratios and steel fibre quantities 221
....
Figure 6.51 - Effect of crack face fibre count on resistanceto deflection 222
.......................
Figure 6.52 - Effectivenessof aspectratio in resistanceto displacement 222
.......................
Figure 6.53a- Effect of deterioration on displacementresistance(steel fibre of aspect
ratio 80 with 1.98mm.crack width) ..................................................................................

223
Figure 6.53b - Effect of deterioration on displacementresistance(steel fibre of aspect
ratio 80 with 4.62mm.crack width) ..................................................................................

223
Figure 6.54a- Effect of deterioration on displacementresistance(steel fibre of aspect
ratio 48 with 1.98nim crack width) ..................................................................................

224
ratio 48 with 4.62mm crack width) ..................................................................................

224
Figure 6.55 - Effect of load magnitude on differential displacement 225
..............................
Figure 6.56a- Effect of deterioration on displacementresistance(load magnitudeof RN
with 1.98mmcrack width) ...............................................................................................

225
xiv
Figure 6.56b - Effect of deteriorationon displacementresistance(load
magnitudeof RN
226
Figure 6.57a- Effect of deteriorationon displacementresistance(load magnitudeof RN

226
Figure 6.57b - Effect of deteriorationon displacementresistance(load magnitudeof RN
with 1.98mmcrack width) ...............................................................................................

227
Figure 6.58 - Effect of load magnitudeon displacement 227
................................................
Figure 6.59 - Effect of crack profile on differential displacement 228
..................................
Figure 6.60a- Effect of deteriorationon displacementresistance(Parallel crack of
0.66mm,width) 228
.................................................................................................................
Figure 6.60b - Effect of deteriorationon displacementresistance(Parallel crack of
1.98mm.width) 229
.................................................................................................................
Figure 6.61 - Effect of steel fabric and reinforcing bar on displacement 229
........................
Figure 6.62a Effect of deteriorationon displacementresistance(steel fabric with
-
1.98mmcrack width) 230
.......................................................................................................
Figure 6.62b Effect of deteriorationon displacementresistance(steel fabric with
-
4-62mm crack width) 230
.......................................................................................................
Figure 6.63 Serviceability thresholdsfor crack degradation 231
- ........................................
Figure 7.1 - Finite Element model of with a discontinuity 245
a concreteslab on grade ......
Figure 7.2a - Comparisonof DIANA
with Westergaard(1926) for a loomm slab loaded
internally 245
..........................................................................................................................
Figure 7.2b - Comparisonof DIANA with Westergaard(1926) for a 200mm slab loaded
internally 246
..........................................................................................................................
Figure 7.2c - Comparisonof DIANA with Westergaard(1926) for a 300mm slab loaded
internally 246
..........................................................................................................................
Figure 7.2d - Comparisonof DIANA with Westergaard(1926 and 1947) for aI OOMM
slab loaded at the edge.....................................................................................................

247
Figure 7.2e - Comparisonof DIANA with Westergaard(1926 and 1947) for a 200MM

247
Figure 7.2f - Comparisonof DIANA with Westergaard(1926 and 1947) for a 300mm

248
Figure 7.3 - Comparisonof DIANA with Westergaard(1926 and 1947) for a 200mm.slab
loaded at the edge with lengthsof 6,10 and 15m 248
............................................................
Figure 7.4 - Schematicof the Finite Element model of crack behaviour within the
laboratory test rig 249
.............................................................................................................
Figure 7.5 - Comparisonof the laboratory Finite Element model with the
standardspring
249
equation............................................................................................................................
xv
Figure 7.6 - Effect of foundation material under the crack in the laboratory Finite
Element model 250
. ................................................................................................................
Figure 7.7a - Comparisonof load transfer and load stepbetweenthe Daventry Finite
Element model and in-service slab response 250
...................................................................
Figure 7.7b - Comparisonof loadededgedeflection and load transfer betweenthe
Daventry Finite Element model and in-service slab response 251
.........................................
Figure 7.7c - Comparisonof unloadededgedeflection and load stepbetweenthe
Daventry Finite Element model and in-service slab response 251
.........................................
Figure 7.8a - Comparisonof load transfer and load stepbetweenthe Lutterworth Finite
...................................................................
Lutterworth Finite Element model and in-service slab response 252
.....................................
.....................................
Figure 7.9a - Comparisonof load transfer and load stepbetweenthe Ballymena Finite
...................................................................
Figure 7.9b Comparisonof loadededgedeflection and load transferbetweenthe
-
.....................................
Figure 7.9c - Comparisonof unloaded deflection and load step between the
edge
Ballymena Finite Element model 254
and in-service slab response.......................................
Figure 7.1Oa- Comparisonof load transfer load step between the Skelmersdale Finite
and
...................................................................
Figure 7.1Ob- Comparisonof loaded edgedeflection and load transferbetweenthe
SkelmersdaleFinite Element model and in-service slab response 255
...................................
SkelmersdaleFinite Element model and in-service slab response 256
...................................
Figure 7.1 la - Comparisonof deflection bowls betweenthe Daventry Finite Element
model and in-service slab response..................................................................................

256
Figure 7.11b- Comparisonof deflection bowls betweenthe Lutterworth Finite Element
model and in-service slab response..................................................................................

257
Figure 7.12 - Processdiagram for establishing slab responseusing laboratory cyclic load
testing and the Finite Element model 258
...............................................................................
Figure 7.13 - Comparisonof predicted and actual surfacecrack width/load transfer
behaviour for Daventry 259
....................................................................................................
Figure 7.14 - Comparisonof predicted and actual surfacecrack width/load step behaviour
for Daventry 259
.....................................................................................................................
xvi
Figure 7.15 - Comparisonof predictedand actual surfacecrack width/load transfer
behaviour for Lutterworth 260
................................................................................................
Figure 7.16 - Comparisonof predictedand actual surfacecrack width/load stepbehaviour
for Lutterworth 260
.................................................................................................................
Figure 7.17 - Effect of modulus of subgradereaction on the loaded slab edgedeflection
for a typical in serviceslab 261
..............................................................................................
Figure 7.18 - Effect of concretemodulus of elasticity on the loaded slab edgedeflection
..............................................................................................
Figure 7.19 - Effect of joint spring stiffness on slab edgedeflections for a typical in
262
service slab.......................................................................................................................
Figure 7.20 - Effect of slab depth on the loaded slab edgedeflection for a typical in
262
service slab.......................................................................................................................
Figure 7.21 - Effect of subgrademodulus of elasticity on the loaded slab edgedeflection
..............................................................................................
Figure 7.22 - Effect load position on load transfer for a typical in service slab..............263
xvii
NOTATION
a Contact radius of a load
a, Distancefrom comer to load centre

AD Actual deflection
AGG Aggregatestiffness
AL Actual load
ASDR Approach slab deflection ratio
B Slab bending correction factor
CRCP Continuously reinforced
d Specimendepth
dI12, Deflections at slab centre (305mm apart)
dI Deflection of loaded slab
do Deflection directly under load
du Deflection of unloadedslab
E, Youngs modulus of concrete
f Load factor (assumedI under self weight)
A
Flexural strengthof concrete
fc Characteristicstrength
fin Target mean strength
Ff Critical stressfor load at free edge
Fi Critical stressfor load at joint edge
Fj Critical stressfor load at interior
F, Front left LVDT
F, Front right LVDT
-
FWD Falling Weight Deflectometer
9 Acceleration due to gravity

h Slab depth
H Horizontal distancebetweenlevelling points
k Modulus of subgradereaction
ks Defectives margin
K Spring stiffness
I Radius of relative stiffness
L Slab length
LCI. Critical slab length
LSDR Leave slab deflection ratio
xviii
LT Load transfer
LTEasd
Load transfer efficiency approachslab deflection
LTEI,d Load transferefficiency leave slab deflection
LTEs,,,, Stressrelated load transfer efficiency
s.,
LVDT Linear variable differential transducer
MHE Materials handling equipment
MO Maximum yield moment for plain concreteslabs
ND Normalised deflection
NL Normalised load
P Applied load
PO Slab collapseload
Pt Total load transferred
R, Rear left LVDT
R, Rear right LVDT
-
TLE Transfer load efficiency
V Difference in slab surfacelevel
VNA Very narrow aisle
W Crack width
Wb
Basecrack width
W, Surfacecrack width
X Specimenwidth
zC Deflection at a comer with no capacity for load transfer
Ze Deflection at the edgeor at a joint that hasno capacity for load transfer
zi Deflection at the panel interior at a considerabledistancefrom the edge
a, Thermal coefficient of expansionfor concrete
a, Vertical movementcausedby slab curling

JT Temperaturedifference betweenoppositefacesof a member
8t Temperatureinduced movement
J Displacement
dc.,h Differential movementbetweenupper and lower faces of the slab
,
Y Eulers constant
P Poissonsratio for concrete
PC Concretedensity
0 Diameter
r Contact stress
xix
1. INTRODUCTION
1.1 Background
There is a growing demandfor in-situ concreteindustrial floor slabs and hardstandings

throughout the United Kingdom and the world. The increase of internal floor space
requiredfor warehousingand manufacturingprocesses,along with the needfor additional
external storageareassuchas ports, harboursand retail outlets have led to the majority of
this growth. The continueduse of rigid pavementconstructionin many countriesensures
the quantity of concreterequired for this type of infrastructurehas also remainedhigh.
This demand has been coupled with client requirementsfor extendedlife expectancies
and tighter tolerances in level and flatness. This necessitatesthat the design of the
structurebe extremely thoroughto ensureall specificationsare met.
Whilst new machinerysuch as laser screedingplant has helped in providing quicker and
more accurateconcreteplacement,the greatersize of the pour createsan increasedrisk of
thermal and hygral movement. All concrete slabs must consequently accommodate
significant shrinkagethroughout their lifespan, becauseif restricted, additional stresses
and cracking will occur. Some form of control to prevent premature degradation is
thereforerequired within the structure.This is commonly achievedwith the useof joints
or controlled cracks to enable the concreteto move at designatedlocations, leaving the
remainderof the slab relatively free of restraint inducedstress.Unfortunately,theseareas
often become the main cause of failure if incorrectly designed or constructed(Ifulett
2001).
The load transfer mechanism across any crack or joint is essential to the structural
capacity of the slab. If this deterioratesfor any reasonthen there is a much greaterrisk of
failure or serviceability problems, such as faulting (change in level across the crack),
excessivedeformationor further cracking.
Many methodsare employedfor load transfer.Someutilise the intrinsic propertiesof the

concretemix such as aggregateinterlock, but others such as dowel bar insertion can be
to the
used enhance mechanism. Steel fabric will reduce movement of tile joint or crack
faces due to the external climate, and will also have some load transfer potential of its
own. Similarly, with steel fibres becomingincreasinglypopular. it is important that their

impact is incorporatedwithin design. The behaviour assessmentof eachjoint type with
respectto load magnitude,crack orientation and long-term fatigue is thereforeessentialto
enabledesignersto accuratelypredict slab response.
1.2 Aims and Objectives
The project aims were developed in collaboration with a steering group committee who
provided technical and practical knowledge throughout the duration of the research
period. Thesediscussions along with a comprehensivereview of literature identified the
need to 'develop a more fundamental understandingof the load transfer mechanisms
acrosscracksand joints in concretefloor slabs,hardstandings

and rigid pavements'.
In order to achievethis overall aim the following objectives were identified:
0 Determinejoint and crack profiles, as a result of the interaction of concreteproperties

and the slab environment (climate).
* Devise and validate experimental proceduresto simulate the load transfer behaviour
of a slabjoint/crack.
0 Determine the effect of a selectedrange of joint/crack openings on the load transfer
behaviour of plain, mesh and fibre reinforced specimens.
0 Investigatethe influence of subgradesupport on the load transfer behaviour of plain,
mesh and fibre reinforced specimens.

9 Develop a structural finite element model to simulate the interaction of joint/crack
opening,reinforcement type and subgrade,on load transfer behaviour.
1.3 Outline of Research Methodology
To achieve full understandingof site conditions and for development of the laboratory
it
test methods was essential that data was collected on typical crack and joint profiles.
Unlike previous surveys,this required measurementsthroughout the depth of the slab and
therefore invasive methods were employed. The first of these required the analysis of
previous records of crack opening obtained during the research of Bishop (2001). This
work utilised embedded strain gauges to calculate slab movements during the early ages
of concrete life. Extrapolation from several gauges and derriec pips placed above one
another at joint positions facilitated the production of overall crack profiles.
2
Coring was also used to obtain information on crack width variations. This enabledthe
opening and inclination to be directly measuredalong the full slab depth using a number
of different methodsand devices. Surface crack widths alone were obtained at dynamic
load test locationsto enableits comparisonwith deflection response.
In addition to direct width measurement,plots of the slab surface profile and slab edge
curling enabled the interpretation of crack geometry. Several methods were trialled to
obtain accuratemeasurementat a conveniently fast rate, necessitatedby the limited time
availability of site access.Initially a preciselevel was utilised, which although providing
high accuracy,was found to be relatively slow and cumbersome.A builders level held at
the slab edge and a graduatedwedge enabled sufficient approximations of the level of
curl. The use of a small profilometer enabled an increased amount of detail to be
obtained, with greater accuracy due to the incorporation of a graduated measurement
scale.
To enablethe effect of crack opening and load transfer on slab responseto be determined,
deflection testing was undertakenon a number of joints at four different in service sites
using a Falling Weight Deflectometer or (Prima) portable dynamic plate test. These
devicesimpart a transientvertical load and monitor deflections on either side of the crack.
The measureddeflections were then used to derive load transfer, load step and edge
cantilever, whilst providing estimates of voiding, thereby providing a greater

understandingof the slab behaviour in respectto applied load.
A small-scalelaboratory testing facility was developedto enable fatigue of cracks to be

investigated under controlled conditions. This incorporated a double cracked specimen
loaded acrossboth shear planes with a force of between 2 and 6kN to representcontact
loads found on site. A selection of typical slab reinforcementswas tested,including steel
fibres at quantities between20 and 40kg/m3and A 142 steel fabric. Crack geometriesused
were similar to those found from the at

measurements in-service site slabs.This consisted
of 'V' shaped cracking with surface widths between 0.66 and 4.62mm and severalparallel
cracks, all of which were below 2mm in width. Each test comprised a minimum of
250,000 cycles with measurements of deflection taken every 600 cycles. The
displacements at the relevant load and cycle number were then used to produce a
comparison between specimens and produce calculation of joint stiffness. The

deterioration of the crack face was also monitored enabling the contributory effects to be
determined.
3
Finite element modelling enabled the effects of load transfer on slab responseto be
established,with the results obtained from laboratory testing correlated to those found
from field testing. The Finite Element model utilised a single joint stiffness only and
therefore the deterioration effect could not be incorporated directly; however, the
selection of a residual value enabled a characteristic in-service slab response to be
obtained.A standardmodel was developedto be representativeof a typical internal slab
constructionand was comparedto the analytical representationsproposedby Westergaard
(1926 and 1947),and loannides et aL (1985) to validate the results.A numerical model of
the laboratory test beam was also constructedto establish whether the standardspring
equationcould be used in developingjoint stiffness.From theseresults it was possibleto
construct a seriesof models, eachrepresentingas closely as possiblethe conditions found
in the field. Comparisonscould then be madeto the deflection measuredresponsesfound
on site to assessthe accuracyof the numerical representation.This was undertakenfor the

full range of load transfer values and, using a trial and error approach,enabledthe value
of foundation support to be determined.After obtaining the correct foundation stiffness

comparisonswere madebetweenthe laboratory fatigue testsand the field data.For two of
the sites tested a full range of crack widths were identified along with their associated
deflection responses.The finite element model was then used to incorporate the spring
stiffness load transfer model obtained from the laboratory, with representativedata taken
for both reinforcement type and crack
width. Once this procedure confirmed reasonable
approximation between site and laboratory information, several parameterscontrolling
slab responsewere altered to enabletheir influence to be further understood.
Interactionbetweeneachsectionof work was requiredto enablemethodologies to be

developedand providea greaterunderstanding
of slab Figure
behaviour. 1.1 represents
this process,wherebyeach of the three main sectionsof work (site data collection,
laboratorydatacollectionandfinite elementanalysis)link directly into oneanother,and
into overall analysis.
1.4 Thesis Structure
Each chapterof the thesis is interlinked, and requirescross-referencingof data to enhance

its full comprehension.The flow diagram produced in Figure 1.2 is provided to aid the
reader's understandingof how this was accomplished.
4
Chapter 1 provides an introduction to the thesis comprising an overview of the research
topic and the importance of the findings to industry. This also encompassesthe aims and
objectiveswith a brief description of how eachwas achieved.
Chapter 2 presentsa thorough review of the literature covering concretestabson grade
and their common failure mechanisms.This includes many of the actions placed onto the
structure and how these are controlled through adequatedesign. Analytical modelling is
also introducedwith a discussionof someof the difficulties known to exist.
Chapter 3 provides a more detailed account of current knowledge into load transfer
acrosscracks and joints, describing many of the mechanismsalongside their long-term

effectiveness. Analysis equations are established for both single and multiple cycle
behaviour, with the effect of geometrical and material properties considered.Site testing
methodsand the analysisproceduresare discussed.
Chapter 4 contains the methodology used for the determination of site obtained
information, incorporating equipment specifications and the implementation techniques
usedto obtain information relevant for analYsingjoint and crack behaviour.
Chapter 5 provides detailed information on the design of a small scaletesting facility for
examining load transfer deterioration. Information on the test specimensis described,

along with the procedures used for translating the data, enabling further analysis and
comparison.
Chapter 6 presentsthe findings from both the site and laboratory testing, highlighting the
influence of crack geometriesand reinforcement type on load transfer. Evaluation of slab
condition is made through analysis of deflection testing which then enablescorrelation

betweenthe various responses.Laboratory simulation results are used to produce a series
of degradation curves, providing comparison of joint stiffness and load restraint for a
variety of material and load conditions.
Chapter 7 illustrates the developmentand validation of the finite element model against
other well respectedanalyses.Numerical results are compared to field data for the full
range of spring stiffness to enable the accuracy of the model to be determined. Values
obtained from laboratory testing are incorporated within the model providing predictions
of slab response,which are then compared to site measurements.Finally, the effects of
changing structural criterion are investigatedto evaluatetheir eff6ct on behaviour.
5
Chapter 8 draws from the results of chapters 6 and 7 presenting conclusions on the
researchundertaken,and offers recommendationsfor further work.

Joint Spring
Sfiffne
Catculations
Finite
Element
Analysis
j
Overall Slab/Joint Laboratory
Data
Compris and Response
Verlificat:
on.1 Collection
Theoretical
and
Actual Behaviour
A 7
Tf.
D. tenorahon modes
i/ Reintocement and
load effects
Site Data
Collection
c, aCk Width
Geometries I Load
Magnitude I cycle
Numbers
Figure 1.1 Data transfer between site data collection, laboratory data collection and
-
Finite Element modelling
Chapter 1
introciuction
Chapter 2 Chapter 3
ConcreteSlabs/ LoadTransferacross LiteratureReview
pavementson Grade CraCkS/Joinjs
Chapter 4 Chapter ab
Methodology
SlabCondition joint Detenor on
[----Chapter
Chapter 6 ->- 6
Analysis& Discussion I Analysis& Discussion ResultsandAnalysis
(Site) (Laboratory)
Chapter 7
Synthps,
Numerical Modelling
Chapter 8
Conclusions&
Recommendation tor Conclusions
Further Work
Figure 1.2- Interaction betweenthesis Chapters
7
2. CONCRETE SLABS/PAVEMENTS ON GRADE
2.1 Preface
This chapter presents a review of literature covering concrete internal slabs, external
hardstandingsand rigid pavements.The main focus of the work has concentratedon
industrial warehousefloors. However, useful information can be determined from the
other types of ground-bearingstructure as there are many similarities in construction and

loading. This is explained in detail in section2.2.4.
The literature review has been arrangedso that section 2.2 leads the reader through the
basic construction types and methods,discussing the conditions found in each structure
whilst demonstratinghow they are inter-related.
Actions are introduced in section 2.3, with the initial climatic responseson shrinkage
examined, followed by typical imposed forces. This is important since the early age
environmentalconditions control the crack position and geometry, with the application of
load magnitudeand repetition influencing degradationand deflection response.
Typical failure mechanismsare discussed in section 2.4 with details provided on the
factors which initiate each method. This enableskey factors controlling the degree and
rate of deterioration to be established, most of which relate to the load transfer

effectivenessacrosscracks andjoints.
To indicate how managementof these failure mechanisms is undertaken, section 2.5
reviews both the structural and detail design of floor slabs and pavements. This
incorporates the serviceability prediction methods and the standard elastic and plastic
equationsused for determining ultimate strength.
The responseof a rigid slab or pavementis influenced heavily by the support conditions
provided by the foundation material. In addition to section 2.2, section 2.6 has been
dedicated to describing current methods of construction and the reasoning behind the
incorporation of each layer.
8
Finally, in section 2.7 the methods and techniquesemployed in structural modelling of
concrete slabs using finite element packages have been considered in light of their
limitations. This enablesany future model to be developedto the highestaccuracy,whilst
retaining computationalefficiency.
2.2 Concrete Slabs/Pavementson Grade
Concreteis a universal constructionmaterial due to its versatility and cheapconstituents.

In internal floor slabs and hard-standingsit is used almost exclusively as it utilises many
of the positive characteristicsof concrete.However, there is an ongoing debateregarding

the advantagesof rigid over flexible pavement construction and as such its use in this
situation is limited within the United Kingdom.
In this section each structure is described with respect to its constituent layers,
construction method and applied loading. Section 2.2.4 combines this information
enabling the readerto cross-referencethe key data from each structure.
2.2.1 Internal Floor Slabs
The internal floor slab of a warehouseis often the most important factor in the successor
failure of such a construction project. If the floor doesnot fulfil the required specification
and allow the client to utilise the building to its full potential then its capital value is at
risk.
The current market for industrial flooring is approximately 6 million square metres of
floor per year (Cudworth 2003). The majority of this is used in warehousesand factories,
both of which require the floor to satisfy certain criteria if they are to fulfil their potential.
The floor itself may be subject to a variety of conditions due to differing load
configurations and climate changes,all of which must be correctly managedto prevent

failure.
areasthereis a requirementto storea vast quantityof

In manyfactory and warehousing
goodswithin a small
comparatively floor area.The most commonand efficient way to
this
accommodate is with a racking This
system. enablesmaterialsto be stackedabove
eachother,whilst still allowingvehicularaccessandmanoeuvrability.The load from this
racking is transferredto the slab via small baseplates,resultingin high contactstresses
andan increasedrisk of punchingshearfailure.High uniformly distributedloads(UDL's)
9
are also commonly found where a number of heavy items are stackedon a pallet or flat
bed.
Materials handling equipment (MHE) causethe main dynamic loads on an industrial slab.
This can be anything from a pallet truck to a forklift, or 10m high very narrow aisle
(VNA) stacker.Many of these MHE's incorporate hard, rigid wheels which create high-
localised stresseswithin the slab. These can create severe problems when the vehicle
passesa discontinuity as the impact of the load intensifies the stress on an already
weakenededge section. Channelling of vehicles increasesthis effect as the high number
of load repetitionsproducesa greaterrisk of concretefatigue and subgradedamage.
Table 2.1 shows typical load types for industrial floor slabs, alongsidetheir appropriate
load classification and magnitude.
Table 2.1 - Typical Industrial Warehouse Loading Values (Knapton 1999b)
Load Type Load Classification Typical Load
Pallet Racking Light - Very heavy 42 - 114 kN
Mezzanine Floor Light - Very heavy 42 - 114 kN
Shelving Light - Medium 42 - 60 kN

Fork Lift Light - Medium 42 - 60 kN
In any large concrete pour some degree of shrinkage or thermal cracking is inevitable.
The location of the cracking can be defined by designingin joints which provide planesof
weakness.Thesejoints allow movement to be confined to appropriate positions leaving

the remainder of the slab intact (see section 2.5.4). In the case of internal floor slabs the
concrete can either be plain and jointed, or reinforced, in which case it can either be
jointed or jointless. The main difference between the types is the number of designed
joints required. Plain concrete jointed slabs necessitatea joint at approximately 6m.
intervals, whereasreinforcement such as traditional steel bars or mesh enablesspacingsto
be increased to 8-10m (Knapton 1999b). In certain situations where either heavy
reinforcement or steel fibres are used, the slab can be constructed using the jointless
method. With this technique the slab will tend to crack randomly but within close
proximity, ensuring good load transfer.
10
As describedabove,joints allow shrinkagecracking to be
confined to defined locations.
These often become structural weak points due to the
continuous movement from
moisture and thermal changes and impact from moving vehicles and pallet trucks
(Simpson2001a).As the joint degradesover time, any flaws in designwill be
accentuated
and costly remedial action will almost certainly be required. The types of damagemost
likely to occur are spalling of the joint arisses,and faulting due to a difference in level
acrossthe joint. Details of the various types of failure are describedin section2.3.
The make up of a typical internal floor slab is shown in Figure 2.1. The concreteslab is
approximately 150-300mmin thickness,and may be reinforced with either steel fabric or

fibres. This is placed on a subbasewhich is usually of a similar depth, but limited to
225mm.to ensure that it can be placed and compactedin one layer. A slip membraneis
sited between the two to reduce any frictional stressesthat may increase restraint to
movement.This is all placed on the subgradewhich may or may not have beentreatedto
increaseits strength.
There are a number of ways to construct internal floor slabs; from the more traditional
techniques used by contractors for many years, to the more modem practices with
technically advanced machinery. The more traditional methods include long strip and
wide strip construction which, as the names suggest,are methodsof placing the concrete
in confined areas.The relatively slow nature of this form of construction, along with the
increased number of vulnerable joints created, has led to other methods becoming
increasingly popular (Bambrook 2000). Large bay construction is a more modem
approach where high slump concrete is placed by pump or truck directly onto the floor
area. Here, it is compacted and levelled off using timber or vibrating screed rails
depending on the required tolerance. In this situation the slab may have joint formers
inserted into the concrete, or have joints sawn onto the surface of the slab, to provide a
control systemfor any thermal or drying shrinkagemovement.
The most modem construction approach of laser screeding is becoming more common
due its potential to create large floor areasat high speed.Laser screedsare items of plant
which place, compact and level the concretein one pass.The elevation is monitored with
lasers and can produce floors that adhereto much tighter tolerancesthan
can be expected
with a manual process.The greater placing accuracy with an increasedconstruction speed
means that a more economical floor is produced. This type of construction may require
joints to be formed with inserts or by sawing to allow for
movement of the concrete.
However, it is possible to produce floors containing joints,
no especially when steel fibres
11
are used as reinforcement within the concrete.These floors are more prone to cracking,
although the reinforcement prevents the cracks from opening to a level where load
transfer will be reducedsignificantly (The ConcreteSociety 2003). The different methods
of constructionavailable are shown in Figure 2.2.
The concreteused in the constructionof a floor slab must be specified correctly to ensure
a suitably strong and resistantslab is produced.In circumstanceswhere a highly resistant

slab is required due to the presenceof chemicals or abrasivematerials,the concretemay
incorporatean increasedconcreteor mortar strengthin the top section,or a resistantpaint
on its surface. A dry shake topping can also be introduced to the wearing course layer
which can increaseabrasion resistance,change the slab colouring and alter the surface
texture of the concrete.
2.2.2 External Hardstandings
Industrial external hardstandingsare used in a variety of situations and have to provide
support for a range of loading. Many are sited around ports and harbours where large
container ships deposit cargo directly onto the concretehardstanding.The majority of the
cargo will be in the form of large containers, which can be stacked several high. Each
container has small casting feet on the comers through which the majority of the load will
be passed.These high bearing pressuresincreasethe risk of punching shearand damage
to the slab surface.
The container handling equipment and the smaller forklift and pallet trucks form other
loading types on the hardstanding.The handling equipment straddlesthe containers and
enablesthem to be transportedto various positions around the site. To aid in the logistics
of movement, lanes are often painted onto the hardstandingto create roadways for the
traffic. This can create channels of dynamic loading increasing fatigue damage and
therefore the risk of failure. Other forms of loading such as forklifts and pallet trucks will
not cause the same amount of stress as the larger containers and their transporters;
however, they may cause more localised damagedue to their rigid wheels and speedof
movement. This will becomeespecially prevalent in areassuch asjoints and cracks.
Other applicationsof externalhardstandings includestoragedepots,distributioncentres

andretail Many
outlets. carry delivery vehicles,pallet trucksandforklifts, as well asthe
temporarystorageof itemsprior to transportation.All of theseload conditionsmustbe
12
accounted for, and designed against, if the hardstanding is to have a long life span.
Typical values for theseloads are given below in Table 2.2.
Table 2.2 - Typical Hardstanding Loading Values (Knapton 1999a)
Load Source Load Type Typical Load

Highway Trailer Axle llOkN
StraddleCarrier Wheel 200 kN
Fork Lift Axle 980 kN
Container(Full) Total 210 kN
The structural components of the hardstanding are very similar to that of an internal
industrial floor slab. The basic construction consists of the concrete slab (usually
reinforced), a slip membrane and a subbase material, which are supported by the
subgrade. The slab is between 150 and 300mm in thickness with the subbasebeing
between 150 and 225mm dependingon the type of subgradematerial beneath.The main
difference between an internal and external slab is the wearing surface of the concrete.
This will be relatively smooth in an internal condition where movement of vehicles is
slow, but will be grooved or textured in an external situation where the increasedspeed
and the weatherconditions may require additional slip resistance.
As with internal floor slabsthere is a necessityto control the movement occurring due to
thermal and drying shrinkageof the concrete. This is often achievedby placing joints at
specified locations throughout the slab enabling sectionsto move relative to eachother. In
plain concrete the joint spacing is in the region of 6m but where reinforcement such as
traditional steel bars, meshor fibres are usedthe spacingcan be increasedup to a value of
12m (Knapton 1999a).In certain situations where heavily reinforced sections are used it
is possible to producejointless floors where natural cracking is allowed to develop. The
reinforcement holds thesecracks together and preventsmuch of the load transfer loss that
would nonmlly occur.
Constructionof an externalhardstanding canbe accomplished by a variety of methodsof
standardpractice in internal floor slabs. Long strip, large bay, or more commonly
nowadays the use of laser screeds,can all provide the necessaryquality and finish
requiredto producea suitablylong lastingslab.Figure2.2 showspictorially the different
available.
constructionprocesses
13
2.2.3 Concrete Pavements
For decadesthere has been a debateabout the advantagesand disadvantages

of rigid and
flexible road pavements.In the UK only a small number of concrete
roads have been
constructedin recent times as flexible pavementsremain the preferredoption (Croney and
Croney 1997). This is the reverse to the USA where the majority of pavementsare of
rigid construction.The advantagesof rigid pavementshave beendiscussedin a number of

texts such as Croney and Croney (1997), and include: 5-11% better fuel economy,
increased life expectancy, reduced maintenancecosts from a decreasein rutting, and
better light reflection. Some disadvantagesdo exist and these include higher initial
constructioncost, increasedtraffic noise and reducedride quality.
The load applied to a concretepavementis different to that found in either an internal slab
or external hardstanding.The number of load cycles is much higher due to the amount of
vehicular traffic using the highway. Generally only the commercial and heavy goods
vehicles are used in the calculation of loading for a pavement, as these cause most
damage.These are defined as having a standardaxle load of 80kN (Croney and Croney
1997). The design loading is in axles and calculated from
expressed millions of standard
the projected number of commercial vehicles (equated to standard axle passesusing
equivalencefactors) expectedto usethe pavementin a designatedperiod.
There are two main techniquesfor fixed form and slip forming.
constructing pavements,
Fixed form construction requires a
concrete train to be mounted on rails to provide the
position and levels for concrete placement. The train itself usually includes plant that
spreads,compacts, finishes and textures in a single or double pass. Some of the more
developed machines also allow for dowel bar insertion and joint construction. Slipform
paving works in a similar manner but in this situation the train is electronically guided
with wires. As with the fixed form paver, the concreteis placed, compactedand finished,
although this is more commonly done in a single passonly. Figure 2.3 shows an example
of the slip-forming method.
Movement control is managedwith either the introduction of joints, or with the use of
continuous reinforcement to hold cracks together. The two methods are described in
sections2.2.3.1 and 2.2.3.2.
14
2.2.3.1 Jointed
Jointed concrete pavementscan be either plain, or reinforced with steel bars, mesh or
fibres. In all casesthe structurerelies on the incorporation of joints to enable movement
and stressrelief to take place and prevent random cracking (Deen et aL 1980). In plain
concretepavements the joints are normally placed every 5m, but this can be increasedto
35m if heavily reinforced. As in all slabsand pavementsthe stressesrelieved by the joints
are generally those causedby temperatureand moisture movements.
The joints may contain load transfer devices such as thosedescribedin section 3.4. These
help to transfer the loading betweenadjacentslabsreducing the stressin the concreteand
supportingthe vulnerableedgesectionsof the slab.
2.2.3.2 Continuously Reinforced
Continuously Reinforced Concrete Pavements(CRCP) are long sections of un-jointed
concrete slab. They are reinforced in both the longitudinal and transversedirection with
either traditional steel bars or prefabricated steel mats. The structural integrity of the
CRCP is provided by the reinforcement which, althoughnot preventing the concretefrom
cracking, will hold the sectionstogether.
Design manuals such as American Association of State Highway and Transportation

Officials (1986) and Portland Cement Association (1'951)agreethat the main function of
reinforcement is to provide a pavement that cracks at regular and reasonably close

spacing, combined with crack widths that provide good load transfer. The amount of
is
reinforcement required selected from one of the variety of tables produced by the
design authorities, which have been shown to provide suitablerestraint.
Some states in America have decided that the use of the transverse reinforcement is
is
unnecessary,and only of use as a construction aid for the placementand maintainment
of position for the longitudinal steel (Gregory 1984). However, this has not been verified
and it is commented that in construction widths over 3.65m the transversesteel helps to
hold the concretetogetherin caseswhere longitudinal cracking occurs.
2.2.4 Structure Comparison
Table 2.3 comparesthe significant parameterscommon in the design and construction of

internal slabs, external hardstandingsand rigid pavements.The main variation between
the structures is caused by the magnitude and type of loading, with that of the rigid
15
pavement being smaller but more frequent. Construction layer thicknesses,shrinkage
control and the concrete specifications are all similar, with the method of construction
changingdue to the size and shapeof the pour.
Table 2.3 - Structure Comparison
Item Internal Slab Ext. Hardstanding Rigid Pavement
Loading 80 kN StandardAxle
42 - 114 kN 100 - 1000kN
Predominantly
PredominantlyStatic PredominantlyStatic
Dynamic
Component 150- 300mm (Slab) 150- 300mm (slab) 125 - 300 (Slab)
Dimensions < 225mm Subbase < 225mm Subbase < 225mm.Subbase
Movement Plain Jointed Plain Jointed Plain Jointed
Control ReinforcedJointed ReinforcedJointed Jointed Reinforced
Jointless Jointless CRCP
Construction Long Strip Long Strip

Techniques Wide Strip Wide Strip Fixed Form
Large Bay Large Bay Slip Form
Laser Screeding Laser Screeding
Concrete Speciricafion
Strength 30-50N/ 30-50 40 N/r
W/C Ratio < 0.55 < 0.55 < 0.55
Cement
325 kg/n? 325 kg/rr? 325 kg/m3
Content
Replacement < 35% PFA < 35% PFA < 35% PFA
Materials < 50% GGBS < 50% GGBS < 50% GGBS
Max Agg. 20 - 40mm 20 - 40mm 20 - 40mm
Admixtures Super-Plasticisor Super-Plasticisor

Super-Plasticisor
Air Entrainment Air Entrainment
Drying
< 0.065% < 0.065% -------- ----
Shrinkage
2.3 Typical Actions for ConcreteSlabson Grade
During the life span of a concrete slab or pavement a number of actions will need to be
if failure is to be prevented. These actions can be created from external forces
withstood
such a racking system or vehicle, or internal strains instigated by the movement of
16
concrete over time. Each mechanismmust be fully understood and designed for if the
structure is to behaveconsistently throughout its life. This section provides an overview
of the actions found in slabs and pavements including: moisture movement, thermal
movement, static loading, impact loading and cyclic dynamic loading, briefly outlining
the effect that eachhason the structure.
23.1 Moisture
2.3.1.1 Drying Shrinkage
Drying shrinkage is caused by the loss of moisture from concrete resulting in a net
decreasein volume. There is however, no relationship between the quantity of moisture
removedand the magnitudeof volume reduction (Neville 1995).
The principle mechanismsof concrete shrinkage are capillary tension, surface tension,
disjoining pressuresand movement of interlayer water (Illston 1994). It is thought that
early drying shrinkage is caused mainly by the surface tension in the capillary pores,
whereas longer-term shrinkage is formed through the loss of water adsorbed on the
surfacesof hydrated cement paste (Perenchio 1997). Illston (1994) statesthat the effect
each mechanismhas on total shrinkageis unclear, with authorsshowing large differences
of opinion.
Cracking of a slab or pavement due to drying shrinkage occurs when the tensile stress
from movement restraint exceeds the tensile strength of the concrete (Sprigg Little
Partnership 2000). During the early life of the slab, the tensile strength is poor and
therefore the potential for shrinkage cracking is high. Effective curing with the use of a
sealant or moist covering can significantly reduce the amount of cracking occurring
during this early period. Suprenant(2002) highlights this fact stating good curing has no
effect on whether or not the concretewill shrink or to what degree,however it does delay
it until a later stage,enabling an increasein strength.
The introduction of joints into concrete slabs aids in the dispersion of shrinkage strains
and can prevent related cracking. The joints are weakenedplanes of concretewhereby the
of stress is high enough to induce a crack. With careful planning in the

concentration
these joints, random cracking in highly trafficked or loaded areascan be
positioning of
Details on the various types of joint used in concrete slabs and pavementsare
prevented.
describedin section 2.5.4.
17
2.3.1.2 Plastic Shrinkage
Plastic shrinkage is the reduction in volume of concrete causedby the loss of moisture
into the atmosphereprior to initial set. The moisture loss may be causedby evaporation
from the concretesurface,or suction from the subbasematerial below. The magnitudeof
plastic shrinkage is approximately one percent of the volume of dry cement (Neville
1995). Plastic cracking is formed when the removal of water from the surface is greater
than the amount of bleed water being transferredto the surface.
Plastic shrinkagecracks are generally not a problem in a structural sense,although they
may affect the durability of the surfaceand lead to crazing and scaling. Power trowelling
of the surfaceoften smoothesover thesecracks, leaving the appearanceof an acceptable
slab; however, the crack may have permeatedmuch deeperthan this top layer leaving the
potential for deterioration at a later stage(Neville 1995).
2.3.2 Thermal
Temperaturechangesin concrete slabs and pavementscan occur from two main sources.
Firstly, during its early life, the cement will create heat as it hydrates and sets. This
increase in temperature will cause the concrete to expand, the magnitude of which is
dependent upon the coefficient of thermal expansion and the extent of temperature
increase.Over time the slab will begin to harden,the heat from the hydration processwill
reduce, and a drop in internal temperatureensues.Studies carried out by Bishop (2001)

have recorded the changesin slab dimensionscausedby the various hydration processes
and discovered that the initial thermal movement instigatescrack formation under sawn
joints, and createsmuch of the early agecracking in slabs.
The second cause of thermal movement in slabs and pavements is a variation in

temperature due to environmental conditions after the concrete has matured. Some
structures such as external hardstandingsand pavementswill be required to withstand
greater fluctuations in temperaturethan internal floor slabs, but even internally a degree
of temperature differential between top and bottom is probable (Suprenant2002). As the
concrete heats up expansion of the slab will occur, as it cools down contraction takes
place. There will generally be a differential temperature gradient between the top and
bottom of the slab causedby the underlying subbasematerial. In this situation there is a
possibility of curling or warping of the slab, which is further describedin 2.4.2.
18
To accommodate thermal movement it is common for joints to be constructed within the
slab. The temperature range will control the size of these joints to some degree, with
changes in crack width measurable over different seasons, or to a lesser extent,
throughout the day (Gulden and Brown 1985).Minkarah et al. (1982) concludedthat the
commonly usedtemperatureexpressiongiven in equation2.1 resulted in poor comparison
of temperature induced movement (6t) when measured against early age field data.
However, Bodocsi et al. (1994) re-examineda number of thesejoints after twenty years
to establishwhether the equation was correct for older concrete,with the results showing
good correlation when temperatureat mid-slab was used. The effect of subbasesupport
andjoint type was also analysedwith their effects shownto be negligible.
(5t= 4
a... TL equation2.1
Where:
a,.= Thermal coefficient of expansionfor concrete

Temperaturedifference betweenoppositefacesof a member
.4T=
L= Slab length
The constituentsmaking up the concrete mix determine the exact magnitude of thermal
expansionand contraction occurring within the slab. Each element that is placed into the
concrete will have its own coefficient of thermal expansion and as such will have an
influence on the overall movement. In most concretethe aggregatecontent will make up
70-80% of the total volume and will therefore have the most impact. Ideally aggregates
with a low coefficient of thermal expansion should be used to resist the potential for
shrinkage; however, each aggregatetype within the mix should be of a similar value to
prevent differential movement and matrix cracking.
2.3.3 Static Load
Static loads are found mainly in external hardstandingsand internal floor slabs, with the
magnitude varying greatly depending on the materials being stored. Typical values for a
racking base-platecan differ from 35-IOOkN, with 200kN quoted for high lifting areas
(Knapton 1999b).The layout of the racking often requires baseplates to be placedback to
back, essentially doubling the force applied. Line loads from mobile racking systemsare
approximately 150kN/m, with mezzanine floor leg loads of around 200kN. In pavements
is
there very little static load and this is therefore disregardedduring design.
f
19
External hardstandingsoften contain some form of racking system,although it is usually
the stacking of large containersused for the transportationof goods on lorries or in ships
that createsthe highest loads. The small feet on which the container rests create high
contact stressesacting directly onto the concrete surface and increasing the risk of
damage.Rogers (2000) has postulated that high point loading can also clamp concrete
slabs to the foundation. This prevents moisture and thermal movement, resulting in the
formation of a restraint crack.
2.3.4 Cyclic Dynamic Load
Cyclic dynamic loading is caused by the movement of vehicular traffic on the slab's
surface.For this reasonpavementsare at greaterrisk from this type of loading than both
industrial floor slabs and external hardstandingsdue to the higher number of cycles they
encounter. The repetitive nature of the load causesfatigue within the concrete, which
eventually manifests itself as a crack. The magnitude of the load applied dependsupon
the vehicle under consideration and its speedof movement. Road ResearchLaboratory
(1955) state that the stressin a road slab due to a moderately fast moving load is similar
to that of a stationary vehicle. This is becausethe deflection magnitude that would occur
under a static load is preventedby the elasticity of the road structure,therebyreducing the
effective weight, counteracted by the increase in stress from the impact effect. Work
completed by Helwany et A (1998) examined the effect of vehicle speed on the
deflections in flexible pavements. It was found that the slower vehicle speed (8km/h)
caused a higher stress in the pavement than that of a high-speed vehicle (72knVh).
Papagiannakiset aL (1991) argued that the frequency of loading should be consideredin
all analysis, as the high-speed repetitive loading of a slab does not allow full recovery to
take place, increasing the risk of damage.In warehousesand external hardstandingsthe
majority of the vehicles will be slow moving, thereby negating some of these effects;
however, the actions of cornering, acceleratingand braking can require increasesrelative
to static loads. The design guides incorporate this phenomenon using a percentage
increaseof the static load, tables of which are readily available.
There are a number of other forms of dynamic loading found in industrial floor slabssuch
as: live storagesystemsthat provide continuousmovementof goodsfrom one areato

another,mobile racking which can be movedto allow for additionalfloor space,and
assemblyplant. They all producesimilar loads to that found in pavementsbut at a
reduced cycle frequency, and therefore the design philosophies are developed
accordingly.
20
As mentionedpreviously it is often the joints and cracks in a slab or pavementstructure
which are prone to the most damage.This is especially prevalent in industrial floor slabs
where much of the MHE have small rigid wheels.The high contact stresseson the edgeof
a discontinuity can result in spalling and deterioration of this material. The design guides
suggest that only the heaviest vehicles contribute to fatigue damage of the concrete
(Croney and Croney 1997). Cyclic sheartesting on cementbound materialsconductedby
Thompson (2001) shows this not to be the case.Plots of joint net shearslip againstload
indicated that 80% of shear slip movement took place under only 12.5% of the applied
load in non-reinforced specimens.This suggeststhat more local distresswill occur at low
load than is currently assumed,thereby shorteningthe structure's life. More details on the
deteriorationprocessesare given in section2.4.
2.4 Serviceability Problems
A number of serviceability problems can occur in floor slabs and pavements,some of

which will develop only in one type of structure,with others common to all. The latter is
often the casewith failure in and around the joint and crack area which, by its nature, is a
structural weak point. This weaknessmeansits degradationwill be far quicker than that
of any other part of the concrete, and may lead to other serviceability problems. This
sectionprovides information on the main causesof failure and how they are formed.
2.4.1 Cracking
As mentioned in sections 2.3.1 and 2.3.2 some degree of concrete cracking will occur
during the life of the structure, caused by the thermal and drying shrinkage of the
concrete.In general the slab will tend to contract around its central core, with movement
increasing with distance from this area (Illston 1994). The slab itself is placed either
directly onto a subbaseor, more commonly in the UK, on a slip membrane.This provides
a low-friction boundary layer between the concrete and sub-baseenabling much of the
movement to be confined to the joints. There is somedispute as to the effectivenessof the
slip membrane layer and whether it should be utilised (Hulett 2001), with further
information on this subject discussedin section 2.6.3. In practice the membranemay not
provide the required slippage, thus a high level of friction will be retained. This may be
caused by the roughness of the subbase, clamping by racking, high loading, or
imperfections in the slip membrane. In this situation locking takes place, preventing
movement of the slab relative to the This

subbase. createstensile stressesin the concrete
which, if high enough,will lead to crack development(Critchell 1958).
21
A similar slab-locking phenomenonoccurs when dowel bars or similar are used as load
transfer devices. These should be debondedon one side to allow horizontal movement
betweenthe slabs,and releaseof the shrinkagestress.If thesedowels becomelocked due
to misalignment or high friction, movementis preventedand tensile stresseswill be set-
up in the slab. If these reach critical levels a stress-relievingcrack will be produced.Due
to the load transfer requirements of the joint it may be necessaryto provide dowels
around the full perimeter of the slab. In this situation the dowel must allow some degree
of lateral, as well as horizontal movement to enable two way shrinkage to take place.
Figure 2.4 showstypical examplesof restrainedmovementcracking.
Problemscan occur when a high number of fine cracksdevelop in close proximity to each
other (Verhoeven 1993). When thesecross, small, unsupportedareasof slab are created
which are unable to fully withstand the applied loading. These cracks are often causedby
poorly designed reinforcement. In areas of slab with high quantities of steel, or other
to
resistance shrinkage, the stresswill be well distributed. However, if the movement is
still too great for the tensile strength of the concrete, regularly spacedcracks will be
produced.This may occur in one direction only, or may transpire in both directions. In the
case where only transversecracks are produced, bridging may instigate cracking in the
opposing direction. According to Verhoeven (1993), 1-1.5m is the ideal spacing of
cracks in pavementsas this ensurestight fitting cracks without excessiveconcretebreak-
up.
Poor construction of the slab can create many problems resulting in random cracking
(Moody and McCullough 1993). One of these defects is caused by the inadequate
compaction of concrete underneath the longitudinal and transverse reinforcing bars,

creating localised weak points with reduced resistance to load. Work undertaken by
Savage(1985) has also indicated a problem with the insertion of reinforcement. When
placed up to the transversejoint it acts as a stressraiser, which when subjectedto a high

number of load repetitions, causesa peeling action of the upper and lower sections of
concrete. If detritus enters between these layers a cantilever action is introduced
instigating vertical cracking. This type of failure is found mainly in concrete pavements
due to the construction and environmental factors they encounter.The formation of these
cracksis shownin Figure2.5.
Mid slabcrackingoften occurswhenthejoints insertedinto the concretefail to openup

This
as expected. is often causedby saw cutting too late after the concretehas been
The
poured. ConcreteSociety(2003)recommendthat this is undertakenbetween12 and
22
24 hours after casting as this is when much of the shrinkageand thermal
movementtakes
place. A similar effect occurs if the saw cut is not deep enough as the concentrationof
stressesmay not convergeunder the joint leading to cracking elsewhere.Careful detailing
is thereforeessentialto make sure movementis confined to the prescribedareas.
The excessivewidth of a joint or crack within a floor slab or pavementcan createsevere
problems with respectto ride quality, vehicular damage,and increasedslab deterioration.

The cause of the crack, and its size, can be attributed to inadequate design, poor
construction and/or extraordinary climatic conditions (American Association of State

Highway and TransportationOfficials 1986).Excessivewidths (often over 5mm) can also
be attributed to the dominant/dormantjoint phenomenonwhereby some joints remain
closed, transferring movement to the next availablejoint. This increasesits size leaving it
more susceptibleto deterioration.
Structural cracking is a more serious occurrenceas it indicates a problem with either the
design, or the construction of the slab. This occurs when loads placed onto the element
are higher than the structural capacity of the concrete, creating significant tensile stress
and full depth cracking. Chou (1989) statedthat the failure criterion for rigid pavementsis
when more than 50% of the slabs have produced this initial first crack due to structural
damage,and advisedthat significant remedial work is then needed.
2.4.2 Curling
Curling in a concrete slab, hardstanding or pavement is caused by a combination of

drying shrinkageand temperaturemovement (Croney and Croney 1997). As describedin
Section 2.3.1 the majority of drying shrinkage will take place at the surface of the
concrete as the base is protected against the atmosphere by the subbase and slip
membrane. This sets-up a shrinkage gradient across depth resulting in the slab edges
becoming raised above the foundation, and a condition of zero support. Cantilevered
sectionsare thus created,increasingthe risk of cracking, pumping and faulting.
Figure 2.6 shows how a similar condition occurs with temperaturevariations throughout
slab depth as in hot conditions concrete will expand and in cold conditions it will
contract. The top section of the structure will change in size dependanton the external
conditions, with the base temperature stabilised by the subsoil. In conditions where the
top surfacetemperatureis colder than the bottom the slab edgeswill tend to rise abovethe
level of the subbase,whereas when the slab surface is warmer than the bottom the
slab
23
sagsat its edges(Armaghani 1993). The subsoil in the latter case (if rigid enough) will
prevent the slab edge from deforming into the ground and lift the central section of the
slab abovethe subsoil.
In either of these casesthe lack of support createdwhen the slab has been raised above
the level of the subsoil can result in increasedstressesand severecracking (Nishizawa et
aL 1993). Barenberg and Zollinger (1990) concluded that for pavements,the strains
created from curling and warping can be larger than those causedby applied loading.
Since concreteis relatively weak in tension, applying an additional load, or even the self-
weight of the concrete,can result in the formation of a crack. This is extremely prevalent
in comer sectionsof slabs where the support through load transfer from other slabsmay
be limited anyway. Once these cracks have formed, deterioration can increaseextremely
quickly and compromise the quality of the slab. According to Walker and Holland (1999)
shrinkage stressesare commonly in the region of 0.1-0.4 MPa compared to a curling
stressof 1.4-2.8MPaand as a result 'curling crack' would be a far more appropriateterm
than shrinkagecrack in respectto floor slabs.
Frabizzio and Buch (1999) state that transverse cracking of the slab relieves the
accumulationof curling stressesand therefore the joints incorporatedwithin the slab can
moderatethe magnitude of the curl, and reduce stress.Ioannides et aL (1990) argue that
dowelled joints can have the opposite effect and actually increase stressesdue to slab
bending. Ytterberg (1987) indicated that a smaller spacing would produce an increased
number of curling sites which could prove detrimental, and thus, the spacing of joints is
an important considerationin controlling surfaceprofiles and preventing cracking.
The degree of curling and the shape of the slab surface will have an impact on the
orientation of the crack face between slabs.Poblete et aL (1988) statedthat in conditions

of negative temperaturegradient (top cooler than bottom) the joint faces will be free to
rotate upwards so that contact will be limited to the lower transverseedges (see Figure
2.7). This can affect the load transfer potential of the crack, thereby influencing slab
behaviour.
Walker and Holland (1999) highlighted the dangers of using high concrete compressive
strengthsdue to the increasedrisk of shrinkage. This enhancedstrength can also reduce

creep preventing the relaxation of stresseswhich, to some degree, would enable the slab
to return to its original level position. Similarly, the smaller section thicknessused in steel
fibre concrete slabs can result in greater curl due to the
reduced self-weight (Schrader
24
1985). The amount of curling, and therefore the size of cantilevered section depends
on
the properties of concrete, amount of load transfer and degree of subbasesupport (Al-
Nasraand Wang 1994).
Suprenant (2002) examined the results from various authors experimental work and
concludedthat an increasein Young's modulus of as little as 6.9 MPa can increasecurling

deflection by approximately 10%. Similarly, an increasein modulus of subgradereaction
from 0.014 to 0.216 N/mm3 increasesthe curling deflection by 30% (Al-Nasra and Wang
1994). Suprenant(2002) examined the dimensions of a curled pavementto determine at
what degreeit becameunsupported. It was found that the length not in contact with the
subbasewas approximately 10% of the total length when load transfer was present.
Where there was a loss of load transfer this increasedto 20%. The magnitudeof the curl
in pavements is commonly around 6mm, but can be as high as 25mm in extreme
circumstances(Suprenant2002).
Several equations have been produced which enable an estimate of slab curling due to
both thermal and hygral effects. It is beyond the scope of this thesis to evaluate these
methods and therefore the reader is directed to the work of Bishop (2001) for further
information. Robins et aL (2002) concluded that the models utilised by Petterson and
Alemo (2000) and Rollings (1993) gave very good agreementwith a finite elementmodel
examining slab movement and restraint Petterson

stresses. and Alemo (2000) provided an
equation for the critical length of a slab that will develop warping as shown in equation
2.2. The critical length (L,, ) is such that at this value the warping stressesare equal to that
of the self-weight of the concrete,resulting in zero curl, as shown in Figure 2.8. Equation
2.3 developedby Rollings (1993) enabled determination of the vertical edge movement
causedby slab curling (at).
equation2.2
0,.g. f
Ar, h.e
at equation 2.3
8.h
Where:
c-= Thermal coefficient of expansionfor concrete

E, = Youngs modulus of concrete
25
AT= Temperaturedifference betweenoppositefacesof a member
h= Slab depth
Pc= ConcreteDensity
g Acceleration due to gravity
f Load factor (assumedI under self-weight)
&, h - Differential movementbetweenupper and lower facesof the slab
L= Slab length
2.43 Pumping
Pumping is caused by the ingress of water into the slab/pavementsystem. Once this
occurs a solution is formed between the liquid and the subbase material and on
application of load near to the affected zone, deflection takes place forcing the solution
out under pressure. This water invariably takes some of the sub-soil material with it
leading to a loss of fines underneaththe affected zone and thus a greater void is created
(Van-Wijk et al. 1989). Eventually this routine generatesa complete lack of support and
an increasedrisk of failure. Rollings (1993) examined this effect when looking at curled
slabsand found that in pavementsthroughout the United Statescurling in the region of 3-
16mm was associatedwith signs of pumping from both vehicular and foot traffic. This
clearly shows how one method of deterioration (pumping) can easily manifest itself into
other causesof failure (cracking).
A number of factors influence pumping and the degreeto which it effects the slab. These
include, water ingress, drainage, quality of subbase,slab deflection and traffic loading
(Van-Wijk et al. 1989).Adequatesub-drainagewas statedas having the greatestinfluence
on the pumping condition, almost eliminating the risk.
2.4.4 Faulting/Dynaniic Load Step
Faulting is the permanentvariation in level betweeneach side of a slab or crack as shown

in Figure 2.9. The difference in level affects the deterioration rate and life span of
vehicles, whilst reducing ride quality. A generally agreeddifferential of 3.2mm or greater

causes very poor ride quality in road pavements, and is therefore classed as a severe
problem (Stock 1988). In the caseof floor slabsthe amount of acceptablelevel changeis
dependentupon the use of the floor. Very narrow aisle trucks,
which have high reach,
may require much tighter floor tolerancescomparedto smaller, rubber wheeledforklifts.
26
Dynamic load step is similar to faulting; however, this is a measure
of slab edge level
variation when placed under load, as shown in Figure 2.9. Walker and Holland (1999)
have suggestedthat a differential vertical movement below 0.5mm.should be
retained,
althoughit is not statedto which floor classthis is directed.A value of 0.1mm.is proposed
by Pearson(1999), who also suggestsa maximum total deflection of 0.15mm.in a single
slab edge. The Concrete Society (2003) does not provide a value for unacceptable
amountsof step as they statethat it has not beenadequatelyresearched.However, they do
indicate that the relevant floor flatnesslimit (varying betweena 2.5 and 7.5mm elevation
difference over a 300mm length) should be adheredto as a minimum.
The changein level can be causedby build quality problems in the caseof a construction
joint, but is more usually due to the lack of load transfer at the joint or crack (Frabizzio
and Buch 1999).Poor load transfer causeseach slab to work as an individual unit, without
distributing load onto adjacent elements.This intensifies the stress within the subbase,
increasing the risk of permanent deformation and pumping. In pavementswhere traffic
runs in one direction only the effect is enhanceddue to the gradual accumulationof load
on the approach slab compared to the sudden impact load on the leave (asymmetric
dynamic loading). This difference in load rate createsgreater settlementin one section of
subbasecomparedto the other and leadsto a dynamic step in the surfacelevel (Cudworth
2001).
Another causefor the variation in subbasesettlementhas been attributed to trappedwater

beneaththe slab joints (Armaghani 1993). He statedthat the movementof fine materials
underneath the jointed section causes one area to rise above the other. This is again
causedby the approach/leaveslab mechanismwhere a higher stressis produced under
one slab section.
2.4.5 Punch-Out
Punch-out failure has been described brieflY in section 2.4.1 in relation to excessive
cracking. The formation of a punch-out begins with transversecracking of the concrete

causedby shrinkage, temperature movements, or loading. As these cracks open further
and deteriorate over time the potential for load transfer is reducedand the stiffness of the
structuredecreases.These cracks, if positionednear each other, may intersectwith
longitudinalcrackscreatedby the transversebridgingof the slabandform smalllocalised
areasthat are no longer integratedwith the rest of the concretestructure(Stock 1988).
With the lack of load transferandrestraintfrom the remainderof the slab,the smallarea
27
of concrete may be forced from the element under the influence of further dynamic
loading and traffic. The progressionof punch-out is shown in Figure 2.10.
The cause of the increasein cracking is detailed in section 2.4.1 with the
main reason
being incorrect reinforcement selection.This leads to poor crack spacing with either too
few cracks with large widths, or many cracks with a small width. Compromise between
the two is essential and therefore correct sizing and spacing are essential.Darter et aL
(1979) state that punch-out occurs most often in cracks 300 to 600mm apart, and rarely
occurs where spacing is over 1200mm. Gregory (1984) provided some guidance for
CRCP with valuesof ideal crack spacingbetween 1.5 and 2.5m suggested.
2.4.6 Spalling
Spalling is defined as the degradationof the concreteat the slab surface and occurs next
to a crack or joint. This will usually involve a diagonal crack that passesfrom the top
surface of the structurethrough to the vertical face (Figure 2.11). An areaof weaknessis
created that can easily break off, leading to reduced ride quality and a need for greater
vehicle maintenance.The reduced section thickness caused by the spall also cuts the
amount of load transfer available from aggregateinterlock, which in turn increasesthe
possibility of further deterioration. The cause of the spall is instigated by the
delaminations created from early age differential shrinkage (Zollinger and Senadheera
1994). The action of rolling wheels, temperaturechangesand moisture penetration then
enhancesthis stressuntil an edgespall is produced.Poor finishing of the structure,ingress

of detritus or incorrect selectionof mix constituentscan accentuatethis effect even further
(Zollinger and Senadheera1994).
Stock (1988) has distinguished between the two main types of spall found regularly in
continuously reinforced concretepavements;thesebeing minor and severe.It is noted that

severe spalling is not formed from minor spalling with different causesfor each. Minor
spalling is created by high deflections due to poor support conditions, whereas severe
spalling is a function of poor concrete tensile strength from incorrect curing and
construction practice. The size of a spall is generally categorised as deep when it is
greaterthan 25mm in depth.
28
2.5 Concrete Slab Design
The structural and detail design of a concrete slab requires all actions and loads to be
thoroughly examined. These must then be controlled with the correct selection of
concrete,reinforcementand layout. The design aims to make the structureas economicas
possiblewithout compromisingthe structural capacity, or its serviceability.
2.5.1 Concrete properties
Concreteis the main material controlling the successor failure of the slab element.If it is
incorrectly designed then deterioration can occur leading to lifelong problems and the
for
requirement remediation.
In most design guides the thickness of slab required to withstand the applied loading is
controlled by the characteristicflexural strength. Concrete is usually selecteddependant
on its compressive strength properties and therefore equations and tables to provide
comparisonsbetween the two have been developed.
The equationsused for design are frequently only appropriatefor plain concrete,with the
advantagesof using mesh or fibres excluded. When reinforcement is to be incorporated
The Concrete Society (2003) advise that the value of flexural strength should be
determinedexperimentally using third point loading.
The selection of the concrete constituents controls the properties of the finished slab.
Knapton (1999a), Ringo and Anderson (1996) and The ConcreteSociety (2003) all detail
the effects on the slab responseenabling the designer to select appropriate mix designs.
This can include, surface finish, curing times, characteristic strength and wearing
resistancealong with many others. It is beyond the scope of this thesis to discuss how
each of these can be controlled with mix design, and the reader is directed to the
aforementioned literature if further information is required.
2.5.2 Slab Reinforcement
2.5.2.1 General
Concreteis essentiallyvery strongin compressionandweakin tension.For this reasonit

is commonto reinforcethe concretewith steelbars,meshor fibres. The reinforcement
enables the tensile stressescreated by shrinkage and contraction to be redistributed,
leaving the concrete relatively undamaged and capable of resisting applied load. In
29
concrete slabs on grade it is unusual for the structural tensile force to exceed the
limitations of the concrete. However, high internal stressesand strains causedby the
inherent shrinkageof concreteare often enough to induce a crack. The larger the crack,
the more liable it is to deteriorateand lead to failure. The introduction of reinforcement
redistributes the tensile stressesthroughout the slab, resulting in smaller, more frequent
cracksthat are lesslikely to causea problem at a later stage.
2.5.2.2 Fabric
Steel fabric comes in a variety of diameters and spacing for use in different
circumstances.The strongerthe fabric required, the greaterthe wire diameter,and smaller

the spacing.Details on standard,stock fabrics are shown in table 2.4.
Table 2.4 - Stock Fabric Sizes(Knapton 1999a)
Fabric Reference Longitudinal Wires CrossWires

Nominal Pitch: Area: Nominal Pitch: Area: Mass:
MM2/M Size: mm MM2/M k g/M2
size: mm mm mm
Square mesh
A393 10 200 393 10 200 393 6.16
A252 8 200 252 8 200 252 3.95
A193 7 200 193 7 200 193 3.02
A142 6 200 142 6 200 142 2.22
A98 5 200 98 5 200 98 1.54
Structural Mesh
B1131 12 100 1131 8 200 252 10.9
B785, 10 100 785 8 200 252 8.14
B503 8 100 503 8 200 252 5.93
B385 7 100 385 7 200 193 4.53
B283 6 100 283 7 200 193 3.73
B 196 5 100 196 7 200 193 1 3.05
Long mesh
C785 10 100 785 6 400 70.8 6.72
C636 9 100 636 6 400 70.8 5.55
C503 8 100 503 5 400 49 4.34
C385 7 100 385 5 400 49 3.41
C283 6 100 283 5 400 49 2.61
Wrapping mesh
D98 5 200 98 5 200 98 1.54
D49 2.5 100 49 2.5 100 49 0.77
Stock sheet size Longitudinal wires Cross wires Sheet
area
Length 4.8 m Width 2.4 m- 11.52m'l
Fabric used in internal slabs is generally of a low strength and in most situations will
either be A142 or A193. The A142 fabric is made of 6mm steel wire, spacedat 200mm
centres in both the transverseand longitudinal directions. The A193 fabric uses a larger
7mm wire to give it a slightly higher tensile capacity. In external situations the size of
30
fabric may be larger due to the increasedtemperaturedifferentials,
and therefore greater
tensile stresses,under which it will be placed. If the reinforcementis required to serve a
structural purposethen similarly a stronger fabric will be required. In rigid pavementsa
long fabric is generally used,the size of which is selecteddependantupon the cumulative
number of million standardaxles that the pavementis required to withstand (Croney and
Croney 1997).
Where light steel meshes,such as the A142 are used the entirety of the slab can be
covered. This can continue through any induced joints as its low strength will enable
yielding and unrestrictedcrack formation. Heavier steel should be avoided as it may lead
to cracking at areasother than where designed(ConcreteSociety 2003).
The most suitable depth for fabric is unclear with various authors suggestingdifferent
values. Each has its own advantagesbut as yet no clear preferencehas emerged.From a
constructability standpointit is advantageousto place the fabric in the baseof the slab as
this prevents any concernsabout cutting through the steel when joints, or wire guidance
systemsare used.
The Concrete Society (2003) state that the position of the fabric reinforcement matters
little as it is of such a small percentagethat it contributes insignificantly to the width and
spacing of cracks. However, they do acknowledge that the fabric prevents induced, or
shrinkage cracks from exceeding I to 2mm, except in extreme circumstances. ACI
Committee 302 (1996) conflicts with'this and provides guidance recommending
temperatureand shrinkagereinforcement is placed in the upper third of the slab as this is
where it provides most restraint. Gregory (1984) reports on studiesshowing firstly that an
optimum depth of 20mm below mid-depth is preferable, with another suggesting mid-
depth reinforcement had better cracking performancethan slabswith reinforcement in the
upper half. Bishop (2001) has concluded that in

reinforcement the upper level is the ideal
location, with the Belgian standardslowering their recommendationsfrom 70mm below
the top surfaceto mid-depth, in the attempt to improve crack distribution.
The ConcreteSociety(2003)note that althoughfabric is not assumedto be structurally

active,it can havesomeeffect on load carryingcapacity.In this casethe positioningis
crucialas groundsupportedslabsarepreventedfrom havingload inducedcracksat their
surface,therebymakingonly bottomslabreinforcementusefulin design.
31
2.5.Z3 Fibres
Fibres have been increasingly used as a method of reinforcing elements

such as floor
slabs, hardstandingsand pavementssince the 1970's (ACIFC and the Concrete Society
1999).Various materialshave beenusedto constructthe fibre but the two types employed
most often are steel and polypropylene. Steel fibres can transmit stressesacrossmicro-
cracksand preventthem from increasingin the hardenedconcrete,whereaspolypropylene
fibres are used for their ability to prevent shrinkageand reducecracking during the early
life of the concrete.Dosagesof steel and polypropylene fibres used in concreteslabsare
2045 kg/m and 0.9 kg/rrOrespectively (ConcreteSociety 2003).
Steel fibres are made of short thin sections of steel that are moulded into a variety of
shapes,examples of which are shown in Figure 2.12. These can be anything from a
hooked end to a melt extract with each having different properties and making a singular
design practice for 'a fibre reinforced element' very difficult. However, these problems
need to be overcome, as the advantagesthat can be achievedare considerable(Hannant

1994).
The fibres can either be inserted into the concreteduring mixing at the plant or just prior
to the pour commencing.Each method enablesthe slab to be constructedwithout someof
the build problems commonly found when using traditional reinforcement. Once the
fibres are incorporated within the mix it is then possible for the concrete to be placed
directly into a slip form paver or screeder,which vibrates, compactsand finishes in one
pass.
The introduction of steel fibres into a concrete mix can help with the crack control in a
number of ways. Researcherssuch as Abdul-Wahab and Ahmad (1992) and Knapton

(1999a) comment that fibres reduce the concrete shrinkage, which in turn lessensthe
stressesin the slab and therefore limits the number of cracks. Fibres will also help to
prevent any micro-cracks developing into macro cracks, which are less likely to cause
deterioration and are therefore highly preferential (Ibrahim and Luxmoore 1987).
According to Grzybowski and Shah (1989) the addition of 0.25% of fibres into a concrete
mix reducesthe size of an un-reinforced crack width by a third. Since the width of a crack
greatly affects the load transfer efficiency, any resistance to separation is extremely
valuable. Hannant (1994) also concluded that 25kg/m3of steel fibre restrains forces that
causecrack opening to a level comparablewith that obtainedfrom using traditional mesh.
32
2.5.2.4 Structural Reinforcing steel
Bischoff et aL (1997) postulatedthat the amount of reinforcementshould be high enough

to have a post crack strength exceeding the un-cracked capacity of plain concrete,
ensuring the slab is still serviceable after cracking has commenced. Reinforcement
quantities as little as 0.13% were found to increasethe post-crack strength, although the
load to initial first crack did not change. When the quantity was increasedto 0.38% a
higher residual and first crack level was found. For theseincreasesto be accountedfor in
design, allowing a decreasein slab thickness,the guidanceneedsto be changedto advise
an ultimate limit statevalue be usedin lieu of the current elastic methods.
Losberg (1978) suggestedthat structural reinforcement should always be used in slabs
and pavementson grade.He statedthat the flexural strengthof the concretealone should
be disregarded, as in other structural members. This requires bottom reinforcement
throughout the slab with top steel also included in high-risk areas such as edges and
comers.The ConcreteSociety (2003) gives values for this reinforcementas 0.25-0.35%.
2.5.3 Structural Design
The structural design of slabs on grade deals with the ability to withstand imposed load.
There are two main design methods for analysing a slab, these being the Westergaard
(1926), or elastic method, and the Meyerhof (1962) and Losberg (1978) plastic approach.
Care should be taken regardlessof which method is usedas the theory assumesthe slab is
in good contact with the subbase at all times. Where the slab has curled increased
deflections and stresseswill be createdby the lack of supportingconditions.
The methods of Westergaard (1926) and Meyerhof (1962) have been compared to
laboratory formed slabs testedby Beckett (1999). His work examined central, comer and
edge conditions using a jointed, 150mm steel fibre reinforced slab with a 150mm subbase.
The findings show that the first crack loads are considerably closer to the Meyerhof
(1962) equation than to Westergaard(1926) for both edge and central loading; however,
the load to first crack for the comer condition provided very close agreementto both. In
regard to deflection the comer condition showed close agreementwith the Westergaard
(1926) equation up to first crack load. For the edge and central conditions the test values
were considerablygreaterthan the Westergaard(1926) values.
33
2.5.3.1 Elastic method
Much of Westergaard's(1926) work, and that continued from it, has examined the
internal, edgeand comer stressconditions occurring in a slab causedby imposedloading.
The maximum stresslocations have been identified, and thus the areasmost at risk of
cracking. In internal conditions and at the slab edgethis will transpire at the bottom of the
slab; however, comer loading generatesmost stress at the surface. Chandler and Neal
(1988) stated that in conditions where the slab remains in contact with its foundation
support, it is the edge loading which createsthe greatestoverall stress,whereasin slabs

that have somedegreeof curl the stressin the comer will increasesignificantly leaving it
at most risk.
Many of the stressand deflection equationspioneeredby Westergaard(1926 and 1947)

have been evaluatedand modified by several authors to produce more accuratedesigns
which allow for the incorporation of partially supportedjoints and warped slabs(Portland
CementAssociation 1951). Regardlessof thesechanges,most of the new design theories
are still basedon his early work (Beckett 2000).
Ioannides et aL (1985) suggested that there have been several situations in which
Westergaard's(1926) equations have been misapplied or incorrectly written. Moreover,
they conclude that the original edge loading equation is incorrect and the 'amended'
equation in his 1947 paper (Westergaard 1947) is preferable. Each of Westergaard's
equations were compared to results obtained from the IILLI-SLAB finite element
computer program by Ioannides et aL (1985). The results show that deflection in the edge
situation is in close agreementto the 'amended'1947 equations although the values are
dependent on slab size. Hence, it is important that the 1947 equations are used in any
direct use for the design of concrete slabs.
Relevant simplified deflection equationsfor each areaof design as statedby Ioannideset
aL (1985) and Westergaard(1947) are listed below.
" (Zi) The interior of the panel at a considerabledistancefrom the edges(equation2.4)
" (Ze) At an edgeor at a joint that has no capacity for load transfer (equation 2.5)
" (Zc) At a comer with no capacity for load transfer (equation 2.7)
Zi = (PY8kl2)11 + (112r) fln(al2l) +7- 5/41 (a/1)21 equation 2.4
34
Ze = ((P[(2 + llp)MM / [(Eh 3k)'Al) [I - (0.76 + 0.4p) (M)l equation2.5
Zc = (Plkl2) [1.1 - 0.88 (a,/7) equation2.6
Where:
P Applied load
k Modulus of subgradereaction
I radius of relative stiffness(see section 3.4.2)
a Contact radius of a load
y= Eulers constant
/,i = Poissonsratio for concrete
E, = Young's modulus for concrete
h= Slab depth
a, = Distance from comer to load centre
When utilising the elastic method of analysis the values of stressand deflection obtained
are only accurateuntil first crack, at which point the behaviour changes.If reinforced, the
this period, which can only be utilised by
slab will retain some residual strength after
The ACI Committee 360 (2000) have
using amendmentsto the original equations.
recommendedlowering safety factors, or adjusting the load contact areas to suit these
particular applications.
2.5.3.2 Plastic Method
Meyerhof (1962) produced equations to allow the ultimate load to be obtained from
theory. His deflections were based on applied loading only, and it is thus noted
plastic
that the theory needsto be adapted to incorporatecurling effects. As with the Westergaard
(1926) equations,three load positions were usedto define the loading regime, thesebeing
internal, edgeand comer.
When loaded in the central condition radial tension cracks are formed on the undersideof
the concrete slab. As the load is increased a circumferential crack transpires on the
surface of the slab and it is this which createsa mechanismand failure. Similarly, in the
edge condition it is the formation of the circurnfrential crack on the top surface which,
when joining the edgeposition, createsthe ultimate failure condition. With load applied at
the comer the surface crack will occur first and failure will therefore be instigated much
earlier.
35
From Meyerhof's (1962) calculations the ultimate failure load is approximately twice the
first crack load in the case of interior and edge positions (equations 2.8 and 2.9
respectively). However, at the comer the first crack signifies ultimate failure (equation
2.10).
The maximum yield moment per unit length of the slab (Mo) for plain concreteis:
fb. h 2
mo - equation2.7
6
Collapseloads (Po) are as shown assurningaALratios are greaterthan 0.2
4.rIMo
PO=
-1- (a / 3L) equation2.8
(rI + 4).Mo
PO = equation2.9
1-(2a/3L)
4.Mo
PO=
-1-(alL) equation2.10
Where:
jb = Flexural strengthof concrete

h= Slab depth
a= Contact radius of a load
L= Slab length
The magnitude of maximum load for edge conditions is dependenton its distance from
the unsupportededge.For a true edge condition to be attained a distancegreater than 2-3
times the radius of relative stiffness from the centre is required. The ultimate load will
then vary linearly betweenthe two conditions. The effect of load transfer at the edgesof
the slab will also have a significant influence on the ultimate load, as extremely high
values increasethe magnitude to that of the central condition. As the capacity of a central
load is approximately twice that of an edge load and four times that of a comer load, this
increasefrom load transfer can be highly advantageous.
36
Meyerhof (1962) compared his work to that of Losberg (1978) and found good
agreement.He concluded that his method gave slightly higher collapse loads for central
and edge conditions, but similar results for comer loading. As a result, Meyerhof (1962)
producedsimplified equationsgiving slightly smaller collapseloads.Thesenew equations
have beenusedin a number of guidancedocuments(ConcreteSociety 2003).
ACI Committee 360 (2000) recommendusing the Meyerhof (1962) method for the design
of slabsutilising steel fibres as it accountsfor stressredistribution and thereforedoesnot

require any modification in the design procedure.Bischoff et aL (1997) agreedwith this
and stated that when the residual load is greater than that of the first crack, it is more
appropriate to use an ultimate limit state analysis. This enablesa reduction in thickness
and provides a more economic section.
Designersshould rememberthat the serviceability of the slab is as important, if not more
so than the ultimate failure. Simpson (2001a) states that failure under static load is
extremely rare and almost always occurs at comers and edges due to the effect of
dynamic load. Many of the so-called economic design proceduresmay give satisfactory
ultimate limit state results, but will fail due to excessivedeflection at the slab edges.The
Concrete Society (2003) incorporate this by advising a load deflection relationship is
examined,with the results staying within the linear limit (Figure 2.13).
2.5.4 Detail Design
2.5.4.1 Crack Control
Joints are used in concreteslabs and pavementsto control cracking causedby movement
of the concrete. In slabs on grade the majority of this movement will be shrinkage and
contraction from moisture loss and thermal changesas described in section 2.3. In rigid
pavements it may be necessaryto incorporate expansionjoints if the concrete will be
subjectedto high temperatures,
although this is rarely required in the United Kingdom.
There is some debate regarding the benefits of inserting joints into a concrete slab, as in
essenceit is simply a controlled crack at a known location. Simpson (2001a) hascreateda

chart showing the pros and cons of both cracks andjoints, reproducedin Table 2.5. Hulett
(200 1) statesthat joints demonstrablycreate more problems than cracks, but crackscreate
just as much concem in the industry.
37
Table 2.5 - Problems Associated with Joints and Cracks (Simpson 2001a).
Disadvantages Advantages
Planned Cracks, * Risk of curling 9 Straight alignment on plan

i. e. joints 0 Reductionof load transfer 0 No raggededges
0 Edge damage(particularly 0 Locations planned
at formed joints) 0 Easierto sealif required
0 Cost of production 0 Crackstend to be vertical
0 Effect on flatness
Random Cracks * Raggededges 0 No cost of production
0 Irregular alignment * Good load transfer (if tied
9 Risk of edgedamage by reinforcement)
0 Locations unplanned 0 Little effect on flatness
0 Difficult to sealif required 0 Lower risk of curling
0 Cracks may be inclined 0 Total length of cracking
may be lessthan planned
cracks
0 Reduction in the numberof
I wide free joints
The two main methods of producing joints are formed and induced. Formed joints are
created by installing stops at the edges of a concretepour. When set, a secondbatch can
then be placed up to the concreteface which by its nature will createa plane of weakness.
Inducedjoints utilise a concentration of stress,causedby a reduction in slab thickness,to
producea controlled crack.
Formed joints are constructed using timber or steel formwork although it is becoming
more common to use one of the many proprietary systemsthat are available. These can
combine formwork, arris protection and load transfer devices, making for quick, accurate
installation. Care should be taken when using formed joints, as the risk of poorly
compacted concrete or differential levels is much higher than with the induced method
(ConcreteSociety 2003).
There are severalmethods of inducing a joint into a slab or pavement.The simplest and
most common way is to cut into the concretewith a saw, normally completed between24
and 48 hours after the slab has reachedits initial set. If stagedtoo early then the concrete
will tend to break away at its surface leaving a ragged and poor quality joint. If left too
late the risk of random shrinkage and contraction cracking increases greatly. Other
methods such as soff-cutting (cutting the concrete before initial set) and inserting plastic
crack inducers can also be used to good effect; however, increasedprecision is required
when using these methods as there is a greater risk of surface problems when working
with wet concrete.
38
When using the saw cut methodto induce the crack at the joint it is recommendedthat the
depth be between 1/4 to 1/3 of the slab thickness.If too shallow then there is a risk that
the joint will not open at the required location, and if too deep low load transfer will be
reduced.The ConcreteSociety (2003) advise that the depth required is dependenton the
age of the slab, as an increased strength will require deeper cutting. Howell (1982)
suggeststhat the increasedtensile strength of a steel fibre reinforced slab may require a
saw cut in the range of 50% of the slab depth. This will reducethe load transfer available
through aggregate interlock immensely and may necessitate the use of further
mechanisms.
As mentioned in section 2.5.2.3 the introduction of steel fibre reinforcement into the
concrete mix can increasethe cracking resistanceand reduce shrinkage. In essencethe

fibres prevent singular cracks from opening up, insteadproducing a number of much finer
cracks. This has led to the development of jointless construction, whereby movement
does not need to be managedwith induced joints. However, it has been found that at
locations where movement can take place it is much more than would occur with a
standarddesign. In some cases these joint widths can be in excess of 20mm with joint
spacingof 40-50m (Concrete Society 2003).
2.5.4.2 Joint Types
There are many terms for the variety of joints used in concretefloor slabsand pavements.
Confusion can arise with the joint specification if there is no clear identification as to each
joint. The Concrete Society (2003) has realised this problem and produced a simple way
of defining the different joint types. Thesebeing:
" Free movementjoint
" Sawn
" Formed
" Restrainedmovementjoints
" Sawn
" Formed
" Tied joints
" Isolation details
Other methods such as those describedby Ringo and Anderson (1996) utilise a primary
and secondaryclassification, with the primary describing the function and the secondary
39
the construction method. Knapton (1999a) also uses the construction method and its
usage,but has different namesfor eachjoint. The method of the ConcreteSociety (2003)
will be usedheredue to its simple nature.
Rigid pavementsgenerally use the samejoints as are found in internal and external slabs,
although they may be called different names.A warping joint is used to enable uplift of
the slab edgesand is formed in the sameway as a tied joint. An expansionjoint however
is rarely used in an internal situation but may be required in certain pavementconditions.
Descriptionsof the variousjoint types are given below.
Free movementjoints
As the name suggestsfree movementjoints contain no reinforcement acrossthe joint and
allow horizontal movement from thermal contraction and drying shrinkageto take place
unrestricted. Some form of load transfer system such as a de-bonded dowel bar is often
used in this situation to reduce vertical movement from applied loading. The joint can
either be constructed using formwork to, divide the slabs, or by inducing a crack at a
specified location. The Concrete Society (2003) recommends that free movement joints
are used in situations where the slab meets an adjoining structure, and is also part of the
floor structure loaded by MBE. In general these will open to a greater extent than a
restrainedmovement joint.
Restrainedmovementjoints
These joints utilise reinforcement to reduce shrinkage and thermal movements to an
acceptable amount. In sawn joints the steel fabric will be continuous throughout the slab,
with reinforcement bars inserted for formed joints.
The Concrete Society (2003) statesthat shrinkageand contraction movement is expected

to be in the region of 1-2mm with joint spacingof approximately 6m, causing yielding of
the reinforcement.Load transfer will be provided by aggregateinterlock and any support
the reinforcement provides. An increasein reinforcement size can reduce movement and
increasethe load transfer effect but may lead to mid spancracking.
Tied Joints
Tied joints are used in situations where a break in construction is required, but movement
is not permitted. Reinforcement is sized to resist all of the tensile force and prevent the
joint from opening.
40
Isolation Details
Isolation details are used to prevent any horizontal movement from damaging adjoining
structures. Compressible filler is placed between the two surfaces allowing any
differential movement to be accommodated.If some form of interaction is required
betweenthe two structuresthen a free movementjoint should be usedin lieu.
ExnansionJoints
r
These are rarely used in internal situations where the slab is relatively protected from
severe changesin climatic conditions; however, in pavementswhere there is a risk of

thermal expansionthey are occasionally inserted.The joints are of the formed types with
dowel bars used to transfer load between slabs. The slab and dowel bar will contain
compressiblefiller betweenits end and the adjoining concreteto enableexpansionto take

place unimpeded. Thesejoints need to be well sealedto prevent detritus entering the joint
preventing movement.
2.5.4.3 Joint Layout
Joints should be positioned carefully to prevent their width from opening up excessively.
In cases where the load transfer mechanism is aggregateinterlock alone this becomes
even more important as it is extremely sensitive to crack width. The Concrete Society
(2003) recommendsthat the maximum distance between internal slab joints is 6m. They
mention that in certain situations it may be possible to increase this distance but only
limited guidance is given for when this is appropriate. Ringo and Anderson (1996)
propose that the Portland Cement Association (1951) method of slab spacing is used as it
is the most thorough. This gives values for plain concrete in feet of 2-3 times the slab
thickness in inches, for example a 200mm slab should have a joint spacingof 4.9 to 6.1m
depending on concrete aggregate size and slump. Knapton (1999a) provides the most
comprehensive information on recommended joint spacing, the values changing

dependingon the concrete strength and quantity of steel fibre included. A section of this
is shown in table 2.6 below.
41
Table 2.6 - ReconunendedJoint Spacing (Knapton 1999b)
Concrete Type Joint Spacing: m

Plain C30 concrete 6
Micro-silica C30 concrete 6
20kg/m3 ZC 60/1.00 steel fibre reinforcement C30 concrete 6
Plain C40 concrete 6
Micro-silica concrete 6
In the case of concrete pavements the American Association of State Highway and
Transportation Officials (1986) state that the joint spacing is highly dependenton the
materials used and the environment. They recommendusing local service recordsof past
performanceto provide the relevant spacingfor an individual area.As a rough guide they
provide a formula similar to that used by Ringo and Anderson (1996), whereby the slab
spacing in feet equals twice the slab depth in inches. Atkins (1997) recommends
contraction joints every 4-7m in non-reinforced concrete with this being extendedto 12-
30m for lightly reinforced, and none for heavily reinforced sections.Watson (1994) gives
more through information for different steelpercentagesbut they generally agreewith the
conservativevalues of Atkins (1997).
All authors agreethat it will rarely be the casethat the idealised spacingcan be used due
to column spacingand warehouselayouts. It is therefore the responsibility of the designer
to ensurejoints do not open up to any considerabledegree,or that random cracking does
not occur. A requirement of this is the control of aspectratio, with Ringo and Anderson
(1996) and The Concrete Society (2003) both recommending a length to width ratio of
less than 1.5 for floor slabs, with American Association of State Highway and
TransportationOfficials (1986) stating a value of below 1.25 for concretepavements.
Many slabs constructed in South Africa use skewed joints and the action of aggregate
interlock to control cracking and enhanceload transfer (Prozzi et al. 1993).Theseare said
to reduce deflections since only one side of the axle wheel is crossing the joint at a
particular time. There is also increased ride quality as the impact is reduced when a
vehicle crosses the joint (American Association of State Highway and Transportation
Officials 1986). The formation of a skewed joint requires acute angles within the slab,
leading to higher stressesand a greater risk of cracking in comer sections. Armaghani
42
(1993) states that since skewed joints have not provided increased quality, and it is
difficult to align any dowels that may be incorporated, their inclusion should not be
considered.
2.6 Slab/Pavement Foundation
The foundation of the slab or pavementcomprisesthe subgrade(normally the soil found
at formation level on site) and subbasematerial. Together they resist load and control
deformation, whilst providing a working layer on which the concreteslab may be poured.
Management of these materials is therefore essential to ensure structural and
serviceability requirementsare met.
If the foundation of a concrete slab or pavement is designed incorrectly or contains
constructional defects then failure is likely to occur regardlessof the quality of concrete
above (York 2001). The subbasecan have a significant effect on the longevity of the
structureand this becomeseven more important as slab depthsbecomeshallower.
The foundation materials used in a concrete pavement,internal slab or hardstandingare
very similar. The slab itself will commonly be placed onto a plastic slip membranewhich
sits on top of a graded subbase material. The base for the entire construction, is the
subgrade.This is often the material found at forination level but occasionally it may be
replacedby a layer of fill, or modified by mixing in stabilisation materials.
2.6.1 Subbase
Subbasedesign is critical to CRCP performancesince a reduction in support, leading to a

loss of load transfer, has been identified as being the primary causeof punch-out distress
(Zollinger and Barenberg 1990).
The subbaseis the material placed directly above the subgrade material and below the
concrete It
structure. is the main foundationmaterialfor the concretecomponent
and its
primaryrequirementis to disperseany loadingpasseddown from the slabto a level that
can be satisfactorily withstood by the subgrade.This loading may be a continuous static
load or a short term repetitive dynamic load, both of which should be adequatelydesigned
for with the selection of a suitable material. This is commonly a well-graded inert
granular fill made from crushed concrete, rock or a general aggregate.In addition to
reducing the stresses passing into the subgrade, the subbase must prevent load
43
deformation which leadsto voiding below the concreteslab. Constructiontraffic running
on the foundation prior to concreteplacementcan also create surfacerutting which must

be controlled to prevent shrinkage restraint cracking of the slab. These are all
requirementsrelating to the material's strength and stiffness properties and therefore

much of the subbaseselectionis basedon thesecharacteristics.
The subbasecan be used in floor design to improve the load responseof the foundation
layer. This can be incorporated directly by increasing the stiffness of the subgrade
material or by assuminga deeperstructural slab, both of which will reduce any induced
deflections. There are some situations when a weak concrete mix may be used as the
subbasematerial due to high granular material costs or very poor subgradematerial.

Again this can be incorporatedinto the design by assumingan increasedslab thicknessto
add any beneficial qualities.
The selection of a subbasematerial is a very important factor in the longevity of the slab.
One of the most commonly found and serious problems that can occur in pavementsis
that of pumping. This is described in detail in section 2.4.3, and is caused when the
subbase mixes with ingressing water and a solution is formed. This is then pumped
through the crack or joint due to the vertical movement of the slab edges when a load
passes over the surface. Ideally a sealant is placed in the crack or joint preventing
moisture from reaching the subbase. However, if this does not occur, a correctly sized
material can be used to prevent water from entering the system,or avert the formation of
a solution, thereby reducing the possibility of voids. The ideal subbase is a well
compacted coarse material with a fines content of less than 10% passing a 200-micron
sieve (Colley and Nowlen 1958). Care must be taken however to ensurethat the subbase
material is not of such a coarsegrading that the subgradematerial can pass into, and mix
with, the subbase.
2.6.2 Subgrade
The subgradeis usually the naturally occurring soil at formation level, although it may be
imported from off-site to replace areasof poor quality. The material should be of uniform
strength without localised hard or soft spots that could affect settlement and bearing
characteristics.When fill is imported, it should be of a suitable strength, stiffness, water
content and grading to enable good compaction throughout depth, and provide a stable
foundation for the slab. In circumstanceswhere a very good subgradecondition exists
with high strength and low compressibility, the slab may be constructeddirectly onto this
layer. However, care should be taken to ensure that degradationof the soil due to water
44
infiltration will not occur if it is a fine grained material, sensitive to moisture increases
(Knapton 1999a).
The subgrademust be testedto find its strengthand suitability for use in the design of the
structure. This can be found using the California Bearing Ratio (CBR) test, which
quantifies the force required to causea specific displacementof a 50mm diameter plate
(Atkins 1997).Another method usedfor evaluating the soil is the plate-bearingtest. With
this approacha 760mm plate, is forced into the subgradeto a depth of 1.25mm,and from
this a modulus of subgradereaction, or V value, can be determinedin N/nim. Many of
the design calculations for concrete slabs and pavementsuse this V value directly to
determinethe load/deflection behaviour of the soil. An approximaterelationship between
CBR and modulus of subgradereaction has been found when the soil is uniform with
depth, as shown in Figure 2.14.
2.63 Slip Membrane
The main principle behind the use of a slip membraneis to reduce friction between the
underside of the concrete and the subbase. If a slip membrane is incorporated into the
design then frictional restraint will be minimised, stressesin the slab reduced, and all
movement will take place at the slab edges and designed joints. The coefficient of friction
can be reduced from 0.7 to 0.2 under ideal conditions (Knapton 1999b), with more
comprehensive data provided in Bishop (2001). However, in practice this rarely works as
well as expecteddue to a number of constructional defects.
The reasons for poor slip membrane effectiveness are various, but all create similar
problems in respect to slab restraint. One of the most common faults is due to the
construction traffic that regularly passesdirectly over the subgrade(York 2001). Even in
well specified, placed and compacted materials some plastic deformation will occur when
large vehicles run over its surface,producing localised low spots and rutting from wheel
channels. These imperfections act as keys which lock the slab in position when the
concrete has cured sufficiently, resulting in cracking if any strains develop later on in the
slab life. This is also true in situations where the slip membranehas wrinkled or folded up
over itself. The reduction in depth acts as an inducer, causing the slab to crack in the
wrong location and leaving the designedjoints unable to servetheir purpose.
Hulett (2001) suggeststhat insteadof designedcracks orjoints situatedat large distances,

it is preferable to have many cracks distributed throughout the slab. Tight cracks are well
45
known to transfer load more efficiently regardlessof the load transfer systemin use, and
thereforelead to fewer problemsand a more integratedstructure(Stock 1988).With a slip
membraneall movement is transferred to the designatedareas (i. e. the joints) creating
larger movementswhere load transfer will becomenegligible, the need for strengthening
of the joint essential,and the overall stability of the edgesreduced.Omission of the slip
membranewould causefrictional restraint throughout the concreteand any cracks would
be createdover the entire area of the slab. These cracks would be narrow, with the load
transfer high, creating a system which is well integrated and does not require special
construction techniques.This could save on both time and money in the construction of
the joints and reducethe risk of failure.
There are other problems associatedwith the use of a slip membrane. Section 2.4.2
introduced curling which was shown to be a significant fault found in slabs, pavements
and hardstandingsdue to differential shrinkage. If the membranewas removed, and the

subbase suitably selected, moisture would be able to drain away from the base of the
concrete slab. Although this level of moisture removal may not be the same as that
occurring at the top surface of the slab, it could create a much reduced differential. In the
United Statesthis problem has been assessedand some of the design procedurescall for
the use of a thin sand blinding layer between the underside of the concrete slab and the
subbase layer. This allows for much greater removal of water from the undersideof the
slab (although care should be taken as too much could affect curing) and, in theory at
least, a reducedamount of curling within the slab.
2.7 Structural Modelling
There are a number of finite element programs available that can model the slab or
pavement system; however, the accuracy of these models is highly dependant on the
parameters placed within the constitutive model. If accurate results are to be obtained,
careful consideration is required for each element with both the values and method of
modelling being highly influential.
2.7.1 Modelling of Concretematerials
The effectivemodellingof concreteis highly importantif accurateand computationally

efficient resultsareto be produced.The amountof detail requiredvarieswith the type of
analysisto be completed.Early agemovement,creepor structuraldeflectionsall require
differentdegreesof input to produceacceptable
results.In analysesof maturedconcreteit
may be acceptableto useonly one elasticmodulus;however,whenearly ageproperties
46
are required the variation in strength may need to be changedfrequently. Bishop (2001)
produced a model to examine the early age behaviour of concrete slabs which required
thermal propertiesand concretestrengthchangesto be continuously updated.Similarly, in
models that predict creep, the reduction of stress over time requires changing material
behaviour.
Bhatti et aL (1998) state that for structural modelling of slabs the concrete material
should enable yielding and crushing for compression, and cracking and fatigue for
tension. They also recommenddividing the slab into layers, with each having different
properties to account for crack propagation throughout depth. Fatigue for concrete
compressionis not included as "the compressivestressesin typical pavementsare so low
that no significant fatigue affects are anticipated" (Bhatti et aL 1998,pp 50).
Cracking in concrete can be modelled in two distinct ways, these being smearedand
discrete. When the smearedapproach is used a physical crack is not formed, instead the
overall element stiffness is reduced. In the case of the discrete method a crack will
develop in the calculated location and the remainder of the structure will remain intact.
AI-Nasra and Wang (1994) give guidance on the methods, stating that discrete cracking
should be used when local behaviour is required, with smeared cracking used when
examining overall load deflection behaviour.
2.7.2 Modelling of Foundation materials
There are two main methodswhich can be employed to model the subgradeof a concrete
slab or pavement system (Channakeshava et al. 1993). The first is a Winkler spring, or
dense liquid system, and the second is an elastic foundation. Several variations and
modifications have been made to each of theseto enhanceaccuracyand make them more
computationally effective. The Winkler or dense liquid method utilises springs placed at
nodes on the underside of the slab or pavement system, which resist deflections to an
amount specified by the stiffness. The method assumesthere is no shear effect between
each spring and therefore the deflection is solely dependenton the support directly below
it.
The elastic foundation uses standard finite element methods to model the subbase.
Informationsuchas density,Poissonsratio and elasticitycan then be given throughthe
constitutivemodel which may require variation at different levels. Elastic foundations
require the user to provide a boundary layer at some predetermineddepth, the position of
which can have a large impact on results. This is more computationally expensive than
47
the Winkler foundation but enables greater control in determining layer properties.
Channakeshavaet aL (1993) commentthat a combination of Winkler and elastic methods
can representthe real conditions most accurately.
Krauthammerand Western (1988) examined the depth of 2D elastic elementsrequired to
simulate the subgradeusing wave velocities. Calculation showed that the model would
need to be approximately 20m deep to fully represent the site information. This was
thought to be too time intensive and therefore linear springs were introduced with a
stiffness proportional to the equivalent elastic natureof the soil.
Harichandrian et al. (1990) analysed pavement slabs using the MICH-PAVE finite
element program. The model was split into three layers, these being the surfacing
material, the granular roadbedand the roadbed soil. From work conductedby Duncan et
al. (1968), Harichandrian et al. (1990) recommended that the bottom boundary should be
fixed at a depth 50 times the radius of the loaded area,with the side boundary at 12 times
the radii. This comparedwell to a Bousinesqsolution with a bottom boundary of 18 times
the radius of the loaded area.Harichandrian et al. (1990) used a reducedboundary depth
of 10 times the radii but made it flexible to allow for further vertical deflection. This
boundary was made linear elastic, which was thought acceptabledue to the minimal
amount of stress that would be caused at this level from the applied surface load.
However, it was mentionedthat if the boundary position was set too close to the surface,
the displacementswould be accuratebut the stresseswould not. In the reversecasewhere
the boundary is too deep, the advantagesof the method are lost altogether.This analysis
was only completed for central slab loading and comparisonsagainst edge conditions
were not undertaken.
For the analysis of slabstestedby Armaghani et aL (1986) the values for the modulus of
subgrade reaction were back calculated from deflection basins obtained during FWD
testing. The actual bowl shapewas compared with that obtained from numerical models
and the V value altered until a good match was found. This method was also employed
for the load transfer spring system with a trial and error approach deemed satisfactory.
This method of analysisdoesnot directly comparethe analytical model againstthe results
obtained on site, and ideally an independent check should be made, whereby known
parametersare used.
Fwa el aL (1996) comparedthe results of numerical models using the traditional Winkler
foundation and a Pasternakfoundation. The Pasternakfoundation uses the modulus of
subgrade reaction with a foundation shear modulus to better simulate the effect of the
48
subgrade.The two models were compared to experimental test slabs loaded at centre,
edge and comer positions. The Pasternakfoundation showed reduced deflections at all
load positions and better predicted the actual results from lab tests.There was however a
difficulty in selecting the correct shear modulus as there is no easy test to calculate its
magnitude.A value was chosen which gave the closest results to actual measurements,
therebymaking the predicteddeflections opento error.
Bhatti et aL (1996) incorporateda pumping effect into the foundation to further enhance
accuracy.The general set-up required the Winkler foundation but the Larralde and Chen
(1987) model was incorporatedwithin this to allow for a reduction in support over time.
At the end of a series of load steps the subgradereaction was amended due to the
pumping effect. Bhatti et al. (1996) statedthat it is acceptableto changethesevaluesafter

25,000 to 50,000 cycles, reducing the computational time required whilst still producing
acceptableresults.
2.7.3 Analysis of Slabs on Grade
The analysis of slabs on grade using nurnerical'computer software has been undertaken
for many years.Severalpackagesare available which representthe pavementor slab in a
number of different ways. The increasein speedof personal computershas enabledmore

complex structuresto be analysed, many of which involve the use of three-dimensional
systemsenabling comer situations and the slab formation as a whole to be analysed.
Ioannides et al. (1985) used the ILLI-SLAB finite element package to examine the
Westergaard (1926 and 1947) equations currently in use. The model utilised plate-
bending elementsresting on an equivalent mass formation to simulate the Winkler type
foundation. Comparisonswere made for interior, edge and comer loading conditions and
they examined both the stressesand deflections obtained by loading in such a manner. It
was found that the Westergaard (1947) equations gave very similar results for most
situations, although the radius of relative stiffness and length of slab did affect the
accuracy.
Ozbeki et aL (1985) argued that the most desirable model for concretepavementanalysis
is a three dimensional system.They mention that two-dimensional systemsare simplistic
and are therefore not ideally suited to looking at a complete slab on grade system. The
JSLAB 3-dimensional package used for the analysis compared well to Westergaard
(1926) solutions for internal, edgeand comer conditions.
49
Kuo (1995) comparedthe stressand deflection results of both 2D and 3D models to that
obtainedfrom Westergaard(1926) and H-LISLAB analyses.This was undertakenfor both

the internal and edge loading condition using different load magnitudes.The research
concludedthat there is good agreementbetween2D and 3D results when the slab is a thin
plate (length to thickness ratio more than 20). There is however a limited number of
problems that the 2D system can model due to the nature of the slab on grade analysis.
Typical problems that cannot be modelled include comer loading, slab warping and
thickened edges.Work completed by Krauthammerand Western (1988), looking at shear
transfer acrossjoints, useda plain strain 2D model to simulatepavementbehaviouras the
results required were only for edge loading a large distanceaway from the slab comers.
Most load transfer systemscan be modelled using the relevant elements in a software
package. For simple mechanisms such as aggregate interlock it is acceptable to

incorporate a spring between the two crack faces (Tabatabaie and Barenberg 1980),
although more detailed models can be incorporated which utilise the Walraven (1981)
two-phasesystem(Davids et A 1998). This model allows the effects of crack openingto
be directly incorporated into the results, reducing the load transfer value as it opens.
Where dowel bars are to be incorporated, embeddedreinforcement can be employed
which allows slippageand loosenessin the bond, producing more accurateresults.
Few models can take account of warping. Hammons (1998) compared the results of site
measuredload transfer values at Denver International Airport against those obtainedfrom
an ABAQUS finite element model. Although the results were reasonable, some
discrepancieswere found to exist. It was suggestedthat the reason for this was some
degree of slab curling creating complex support conditions which can vary from full,
partial, or none. Where this type of modelling is required a hygral and thermal analysis is
required to form the initial condition. Following this a structural analysis needs to be
carried out to examine the effect of load on the warped slab. Al-Nasra and Wang (1994)
examined this behaviour using a non-linear foundation spring. This is similar to the
standardWinkler foundation but contains zero tensile strength so the slab is free to move
in an upward direction due to the temperature gradient. This model was compared to
limited site data and found to produce good approximation. Bishop (2001) constructeda
complex model examining the early age behaviour of concrete slabs. This enabled
stressesand deflections of slabs to be ascertainedduring the early life of the concrete,
where both the temperatureand concrete strengthare constantly changing. In this model a
50
membranewas utilised between the slab and the subbaseallowing a controlled tensile
strengthbetweenthe two. This was found to correlatewell with data gatheredfrom site.
There is a requirement in structural models to enable the interface between slab and
subbaseto be controlled. Kuo (1995) has achieved this with interface elements which
allow the non-linear behaviour of the shearslip to be introduced. Kim, Won et al. (1998)
achievedthe sameresults using springs in the horizontal direction.
2.7.4 Modelling Limitations
Many models utilise either the Winkler spring or a denseliquid foundation to simulatethe
effect of the subbaseand/or subgrade(Road ResearchLaboratory 1955). Deflection is

resisted only by the spring placed directly under it, and therefore cannot incorporatethe
effect of the surrounding area. This fails to fully represent a real situation whereby
deflection of a single point is dependantupon the adjacentelement stressstateas well as
its own. Similarly, the soil restraint on the opposite side of a loaded slab will be reduced
due the sheartransfer capabilities of the soil. The soil adjacentto that being loaded may
then be compressed,creating a void between foundation and undersideof slab. In some
casesmodifications have been made which improve the simulation, but theseneed to be
enhancedfurther before they can be usedwith confidence.
To model the subgradeas a Winkler foundation, the modulus of subgradereaction first

needsto be determined.Tang (1993) postulatedthat this value is not a constantand is in
fact smaller under the slab edge.As most of the testsavailable for obtaining this value can
only be used at the centre of a slab, the results should be used cautiously if selectedfor
the entire length. This issuebecomesmore important when examining edgedeflections as
they are much more sensitiveto modulus of subgradereaction values.
Many of the models using standardelementsas a foundation assumethat the soil is in full
contact with the undersideof the slab. In most situations, such as that occurring in a slab
on grade, the soil in and around the edge of the slab tends to deteriorate over time from
compaction, or by being pumped out through the crack or joint (see section 2.4.3). This
creates voids underneath the slab edge which lowers the resistance to deflection. By
ignoring this phenomenonthere is a risk that the calculated deflections will be smaller
than those found on site.
Road ResearchLaboratory (1955) state that some of the main assumptionsused when
modelling slabson grade are inconsistentwith the realities of site. The first of theseis that
51
the slab is in perfect contact with the foundation. Due to the curling effect causedby the
shrinkagegradient this is often not the case. Gap interface elements must therefore be
incorporated into the model to allow for uplift caused by warping. If this is not
undertakenresistancewill be provided, and the slab will not be able to detachitself from
the underlying layer. This will affect the friction coefficient betweenslab and subbase,as
in a warpedslab it will reduceto a zero value.
One of the main drawbacksof many finite elementprogramsis their ability to adequately
model the interface between the concrete slab and the underlying layer. Many of the
existing finite elementprogramsfor pavementanalysisassumeeither zero or full bond for
the interface condition, whereas in reality the amount of layer slippage under a heavy
wheel load is somewherebetweenthese two extremes.Having a capability to model and
specify the varying levels of slippage between the slab and an underlying layer would
greatly improve the ability to fine tune concretepavementdesign.
2.8 Conclusions
This Chapterhas reviewed literature relating to concreteslabs on grade.Clear similarities

between internal floor slabs, external hardstandingsand rigid pavementsare evident in
respect to their relevant constructions.The loads applied to the slab are slightly different
in magnitude and the number of cycles they are likely to encounter; however, all are
required to withstand many dynamic load applicationsthroughout their working life.
Concrete by its nature will crack to some degree due to restraint against moisture and
thermal movement. These cracks have been shown to create problems in respect to
strength and serviceability due to the weakened slab sections they produce. Control of
these cracks is provided by the insertion of reinforcement, which prevents them opening
up excessively and increases the load transfer effect. Joints have also been introduced;
however, these only define the location of the crack rather than preventing it occurring.
Much of the design guidance has reviewed the effect of static load within the slab;
however, it is the cyclic loading acrossthe weakenedcrack and joint areaswhich tends to
create the greatestrisk of failure. The incorporation of a suitable foundation within the
slab can resist deflections and rutting which could lead to further restraint cracking.
Voiding underneaththe slab is also a common occurrence causedby compaction of the
soil under repetitive load, pumping and slab edge curling. This results in a reduced
resistanceto deflection, increasingstressesand the rate of crack or joint deterioration.
52
The underlying theme of crack and joint behaviour, which has been found to control slab
responseand resistanceto failure, is noted throughout each section. Understandingthe

load transfer effect of these cracks is therefore essential in improving the design and
maintenancestrategyof thesestructuresand has therefore becomethe main focus for this

thesis. Chapter 3 addresses load transfer in more detail in relation to its basic
mechanisms,calculation and evaluationtechniques.
53
Topping Concrete Slab/flavement
Rcillf"Orcellic
Slip Membrane
Sub-Base
Sub-Grade
Figure 2.1 - Typical construction of an internal floor slab
Long-strip \%ldc-tl 11)
Lascr Screeding Lar, c Bav
Figure 2.2 - Concrete slab construction methods (Knapton 1999b)
04
Figure 2.3 - Rigid pavernent slip form construction (Wirtgen 2001)
54
Central Core Shrinkage Direction
Shrinkage
Restrained
lllo%,
Clllcllt
Slit inka, le
14-
L,III CtIaII IC(I Misalk, ned

I
End Dowei Bar
F/'
Slip Membrane
7-- Crack Rough Ground
Good Construction Poor Constniction
Figure 2.4 - Shrinkage restraint of a concrete slab
Mesh
Dowel
Get,
Figure 2.5 - Peeling action caused by fabric reinforcement (Savage 1985)
zor. 1 I ZONE z ZONE IONIC I ZONE t
Top warmer than bottom Top cooler than bottorn
Figure 2.6 - Temperature induced slab curling (Ytterberg 1997)
55
INDUCED
TRANSVERSE
JOINT
I
LOAD
SAW GROOVE
hi
T, hil
V4,.
1
COMPRESSION
ZONE 02 1
h/2
DASE
0- COMPRESSION
71 x SHEAR STRENGTH VUETO INTERLOCKIING
T2- SHEAR STRENGTH OURTo FRICTION
Figure 2.7 - Rotation of joint faces(Pobletteet al. 1988)
'-Je LENGTH ffl

5 10 is 20
sea
I Tx, CNAL WLE
6F "UM&LIAOfr
Ov $LAO uNcrof
E
400 SLA8 LOCATION
E
E
LASSV-f-CAS
A, FXLON
fAA VVT 0 RENO
alf SLAff LEWTN
W AW OW dd 3W .
203
tj
X0
OL
a dv Zi
sLia LEW.
40
T: 4 tft)
za 60 70
Figure 2.8 - Curling profiles of a concreteslab causedby differential

shrinkage(Rollings
1993)
56
Damaged Area Faulting
Dania,, cd Area Load Step
Figure 2.9 - Faulting (top) and dynamic load step (bottorn) across a discontinuity
' "
-
4 .---
(11) Slat) spanning, in (b) Slah ill (c) Slah ciacked

panllllig ccicly
transverse direction longitudinal direction due to leading to punch-out of

transverse crackint: loose concrete
Figure 2.10 Punch out accumulation

-
Crack opening
Dm%clliar (Lick of Sunport)
Figure 2.11 - Spal ling of a crack edge
57
r
oll(., 11-1 or Win,

or
., I
wor 'o ww
F, ,, t,, fd, NIA. -, t r. 1

CrimlwA-e-nd %ire rA
Mooked) slit eheet 0111.
,,r im.
Figure 2.12 - Steel fibre types (Balaguru and Shah 1992)
Linear Limit
DEFLECTION
Figure 2.13 - Load/deflection relationship for steel fibre ground Supported slab (Concrete
Society 2000)
CBR
Figure 2.14 - Relationship between CBR and modulus of subgrade reaction (Croney and
Croney 1997)
58
3. LOAD TRANSFER ACROSS CRACKS/JOINTS
3.1 Introduction
Chapter 2 has reviewed literature concerning concrete slabs on grade and found that
cracks and joints are highly influential in their performance and deterioration. A full
understandingof their behaviour is therefore essentialto enable accurateslab design and
increasethe longevity of the structure's life. The effectivenessof a crack or joint can be
determined through a calculation of load transfer. This is the phenomenonwhereby a
force directed onto one element is passedacross a discontinuity into another through a
natural (e.g. aggregateprotrusion) or designed(e.g. dowel bar) mechanism.This lowers

both stressand deflection, thus reducing the risk of failure.
The effectivenessof the load transfer system is vitally important to the longevity of the
slab system. Tabatabaie and Barenberg (1980) state that without it both theoretical and
field results indicate there will be permanentdeformation of the subbasenear to the slab
comers, leading to faulting and cracking. Increasing the subbaseor the slab thickness is
reportedly ineffective in substituting the load transfer system.
This Chapter examinescurrent knowledge in respect to load transfer for the design and
operation of concrete slabs. The different mechanisms available for increasing load
transfer are introduced alongside their benefits and drawbacks. Field investigation
techniques are discussed, which enable slab response to be determined. Finally, the
mathematical models used to portray load transfer are described for both singular and
numerous load cycles.
3.2 Influenceon Design
The valueof load transferusedfor designis oftenopento interpretation.The higherthe

level chosen,the greaterthe ultimateloadcapacitywill be, resultingin a reducedsection
thickness.However, if the load transfer on site doesnot match that used in the designthen
there is an increasedrisk of failure. This can be anything from faulting to slab cracking,
with details of each described in section 2.4. Similarly, if the designed load transfer is
59
chosen conservatively then an overly thick slab will be selected resulting in an
uneconomicstructure.
Much of the deteriorationencounteredduring the life spanof a concretepavement,slab or

hardstanding can be attributed to the lack of load transfer between the crack faces
(Krauthammerand Western 1988).Work completedby Prozzi et aL (1993) showedthat a
reduction in joint efficiency (i. e. load transfer) to 35% would causea doubling of stresses
in the baseof the slab under load when comparedto full load transfer.Frabizzio and Buch
(1999) state that a value of 70% is an acceptablethreshold value, above which failures
such as faulting will be avoided. A number of different systems can be employed to

increase the required load transfer across the joint. These can vary from aggregate
interlock, through to the useof dowel bars and keys.
Many of the design guidesrequire load transfer values in their calculationsto evaluatethe
slab's structural capacity. Chou (1983) argued that the load transfer efficiency of a joint
would have negligible effect on stressesand deflection when the load is placed in a
central position, although when near to the edge or comer it will become a major
contributor. Indeed, Westergaard (1926) and Meyerhof (1962) equations (section 2.5.3)
both require an estimation of load transfer to enablecorrect design specification.Chandler
and Neat (1988) and Neal (1996) both recommend multiplying the calculated
stresses for
a free comer or edge by 0.7 and 0.85 respectively if load transfer is present; however, no
guidance is given as to what constitutes acceptableload transfer. The U. S. Departmentof
Transportation (1990) suggestthat load transfer should only be consideredon roads that
carry low loading. Again no definition is provided as to what level of load transfer is
applicable, and what reduction in stress should be used. The American Association of
State Highway and Transportation Officials (1986) provide the most information,
producing stress reduction ratios for different slab construction types and climatic
conditions.
3.3 Crack Geometry
When examining the geometry of a crack it is common to think only in terms of a single
width. This is generally measuredat the surfaceas it is the only visible area;however, the
crack often varies with depth, as well as along its length and is therefore 3-dimensional.
Authors have widely agreed that the slab or pavement will curl to some degree, the
amount dependanton the materials used, construction methods and climatic conditions
60
(Pobleteet al. 1988).As mentionedin section2.4.2, the direction of curl is determinedby
the temperatureand moisture gradients in the concrete. In both internal and external
situationsit is common for the edgesof the slab to rise above the level of the subbaseat
either a joint location, or a newly formed crack. Due to the curling effect, the opening at
the top surface will be in excessof that found at the bottom, with Poblete et al. (1988)
concluding that for negative temperature gradients (top cooler than the bottom) the
transversejoints and cracks will be relatively open and free to rotate, resulting in contact
with the opposingcrack face being limited to the lower edges.Similarly, ACI Committee
302 (1996) state that the crack width will be at its widest at the surface and will narrow
with depth due to the shrinkage gradient. Bishop (2001) monitored both free movement
and restrainedjoints at several warehousesin the UK. Gaugeswere placed at different
depthsthroughout the slab and measurementstaken at both early agesand long term. The
results indicated greater movement at the top of the crack than at the bottom, with some
joints showing a Imm difference betweensurfaceand mid-depth readings.Bishop (2001)
concluded that the cause of this was slab warping, which was confirmed with a precise
level survey of the slab surface.
Cracks in slabs and pavementsoccasionally propagatetowards one Side leaving a sub-
vertical orientation (Poblete et aL 1988). This can produce anomalies in joint
effectiveness depending on which side is loaded. In one position the slab will bear
directly onto the opposing face, whereas in the other, support will be limited and any
resistance to deflection will come solely from the subsoil (Figure 3.1). Highways
Department (1999) guidance suggeststhat where applicable, deflection testing should be
undertaken in both directions to detect these variations. If unachievablethey recommend

that testing be carried out on the downstreamas it generally producesthe worst case.This
indicates that the variation between load positions is caused by factors other than the
orientation of the crack. The review of literature shows that little information has been
produced on the deviation of the crack from vertical, and is therefore not perceivedas a
common problem.
The methodsof design for joint spacing in floor slabs assumethat eachjoint will open a
similar amount. Bishop (2001) concluded that the behaviour of sawn restrainedjoints
varied considerably, with some remaining closed (dormant) and others opening up
excessively (dominant), dependant on the behaviour of those surrounding them. This
occurrencehas been recorded in several investigations (Minkerah et al 1982, Bodocsi et
al 1993) and led to the dormant and dominant joint categorisation. Poblete et aL (1988)
examinedthe opening of joints through their early ages,and noted that in the initial hours
61
of concreteplacementonly one in three would open up. At a later stage all joints were
activated, although the joints which openedup first could still be identified due to their
larger width. The ConcreteSociety (2003) reasonthat the dominantjoints are causedby:
inadequatetiming of the saw-cut, locking up of joints, subbasefriction against the slab
and early loading restraint. Rogers (2000) concluded that further researchneeds to be
undertakenif the dormant and dominantjoint phenomenonis to be avoided.
3.4 Load Transfer Mechanisms
There are many mechanismsavailable which provide load transfer across joints and
cracks. Some of these can be classedas natural i. e. the shapeof the crack face causedby
the phenomenonof crack initiation, or designedi.e. the insertion of a device acrossthe
crack. Ioannides and Korevesis (1990) suggestthat a pure shear load transfer mechanism
is preferableto one that transmits bending becauseof the warping effect inducing further
stress when movement is prevented.Kelley (1939) agrees, stating that dowels that are too
stiff can result in restraint to longitudinal warping, leading to distress. However, in
situations where the joint width opens up considerably,or crack face degradation occurs,
it will be necessaryto incorporate someadditional mechanismto prevent failure.
3.4.1 Aggregate Interlock
Aggregate interlock is the most fundamentalmechanismof load transfer in concreteslabs
and pavements. In instances where the cracking pattern is unknown, and slab
reinforcement is not present, it becomes the only means of load transfer. Even in
situations where additional mechanisms are provided, aggregate interlock still makes a
significant contribution (Abdel-Maksoud 2000). Jimenez et al. (1982) estimated that its
is
effect approximately 75-90% of the total load transfer in crack widths between0.25 and
0.76mm containing dowel bars. Houde and Mirza (1974) gave the load transfer value to
be around 50%, although Swamy and Andriopoulos (1974) proposethat the percentageof
load transferredby the aggregateinterlock effect is dependenton the steel quantity. They
gave values of 90% for beams containing 1.97% steel and 50% for beams containing
3.95%. Walraven (1981) and Bazant and Gambarova (1980) both highlight the
importance of aggregateinterlock in reducing the risk of steel yielding.
When a load is placed on one side of a crack the protruding material will come into
contact with recessionson the opposing side. If prevented from moving in the horizontal
direction a bearing or shearingstresswill be applied acrossthe crack, thus transferring the
62
load. This mechanismof load transfer is a true shear action, without the bending effect
that is found when using dowel bars.
Various authors have divided the mechanismof aggregateinterlock into the 'local' and
'global' condition, which are madeup of micro and macro roughnessrespectively(Figure
3.2). The literature shows ambiguity in the definition of micro and macro roughnesswith
different authorsusing different models. Raja and Snyder (1991) and Laible et aL (1977)
state that the micro roughnessis the interlocking of fine aggregateparticles and macro
roughness is that of the larger pieces of aggregate.They continue by saying micro
roughnesscreatesa crushing and bearing action compared to the sliding and overriding
behaviourof the macro roughness.Contrary to this, Walraven (1981) proposeda different
model whereby micro-roughnessis causedby all aggregateprotrusion, and that macro

roughness is due to the overall undulations of the crack face. Furthermore, micro
roughness will be the dominant of the two systemsthus, in modelling terms, the crack
plane can be assumed flat. Authors agree that the local roughnessis most significant
during early load cycles when the crack width is less than 0.25mm. When cracks exceed
this width, or after many cycles, attrition of the micro or local roughnesstakes place and
global roughness, or macro-texture, becomes the dominant method of load transfer.
Millard and Johnson(1984) testedthis theory using severalspecimensand concludedthat
shearis resisted by a combination of crushing and sliding and that no distinct point exists
whereby the two mechanismschange.
A further model developedby Walraven (1981) simulatesthe crack face with a systemof
spheres embedded into a matrix (Figure 3.3). Each sphere and its embedmentdepth can
be statistically calculated to develop a good interpretation of the real situation. The
contact areasfor each particle can then be calculated and the stressesascertained.As the
sliding action develops, plastic deformation occurs and high contact stresses are
produced. This leads to further deformation, until such a stage that an equilibrium of
forces is obtained.
The different theories were comparedto laboratory testing and the most accurate
representationwas found to be the rigid spheremodel, whereby load transfer is achieved
with a combination of crushing and sliding (Walraven 1981). Millard and Johnson(1984)
also confirmed that this method provided a more accuraterepresentationof the behaviour
found in a number of test specimens.
63
The type and quality of aggregate, and its bond with the cement matrix, are very
important factors in the aggregateinterlock effect. If the aggregateis weak and allows
cracking to propagatethrough it, then a smooth face will occur and interlock will be
limited (Frabizzio and Buch 1999).When strongeraggregatesare usedthe bond between
the aggregateand the cement mortar is found to be the weakest point, resulting in
aggregateprotrusion and higher amounts of load transfer. The timing of formation will
also have some influence over which of the two types of crack will take place. As the
bond improves due to the maturing concrete,the risk of aggregatesplitting becomesmuch
greater.The majority of slabs crack in the first 24-48 hours due to the thermal contraction
and drying shrinkage and therefore a pullout failure rather than an aggregatefracture will
define the overall shape(Abdel-Maksoud 2000). Nowlen (1968) examined the effect of
crack timing and its associatedcrack face by comparing the load transfer of test slabs
cracked at three time periods. The slab cracked at I day showed a 25% increase in one
million cycle load transfer effectivenesswhen comparedto a slab crackedat 7 days.
The size of the aggregateis highly influential in the effectiveness of load transfer. A
larger sized particle can bridge any crack better than a smaller particle, and tends to have
a reduced risk of becoming loose in the cement matrix (Nowlen 1968). Walraven (1981)
states that particle sizes less than two times the crack width can be consideredas being
inactive and having no contribution to load transfer. Nowlen (1968) also found a rise in
load transfer of 11% by increasing the aggregate size from 20mm, to 40mm, and
suggested the cause was a decreasein initial loosenessbetween the aggregateand its
socket. The theory assumes a spherical particle, whereby horizontal movement (an
increase in crack width) creates greater free movement (Figure 3.4). In general, larger
angular particles are more beneficial for load transfer; however, they hinder finishing of
the concrete surface and therefore a compromise must be made (Colley and Humphrey
1967).
Aggregate interlock is reduced as crack width is increasedin size, with different authors
providing information on the associated values. Tables given by Pearson (1999)

demonstratethat almost full aggregateinterlock is achieved at a crack width of less than
0.5mm. Between 0.5-Imm. only partial aggregateinterlock is retained, and widths over
1.5mm.maintain no interlock at all. Buch (1999) statesthat the opening of a crack by as
little as 0.8mm can reduce the load transfer by up to 50%, and ACI Committee 360
(2000) recommend that load transfer cannot be relied upon if the crack width is greater
than 0.9mm. No mention of the effect of aggregate size is provided, although it is
expectedto be influential.
64
In a concreteslab the crack width changesdependanton climate, varying betweendaily
and seasonalcycles. Since aggregateinterlock load transfer is highly influenced by the

crack width it will also change frequently depending on when the calculation is made.
Deterioration of the crack face causedby high cycles of loading must also be taken into
account, as this will diminish the interlock effect over time, thereby reducing load
transfer.
3.4.2 Dowel Bars
Dowel bars are one of the most traditional methods of creating a load transfer system
acrossa predetermined joint; however, they can only be employed when the location of
the crack is known, otherwise the systembecomesredundant(Critchell 1958).Ringo and
Anderson (1996) comment that if well constructed dowels can provide a similar slab
strengthto that found internally.
According to Millard and Johnson(1984) the mechanismof load transfer for a dowel bar
is split into three main areas, these being direct shear, kinking and flexure. They also
discovered that flexure of the bar is the main method of transfer due to the significant
deformation which occurs within the concrete. Yoder (1959) and Friberg (1938) both
that the effective distance at which dowel bars have some proportion of load
suggested
transfer capacity is 1.81 from the load centre, where I is the effective length (equation 3.1).
Ioannideset aL (1990) argued with this, stating that the effective distancewas only 1.01,
finite element simulations and site data comparisons confirming their position. All
with
agreed that the amount of load each dowel bar transfers reduces linearly with
authors
distance from the application position. The Concrete Society (2003) have used the
made by Yoder (1959) and assumedthat dowel bars can contribute
recommendations
fully if a distance half the effective length is used (0.91). Once the dowel is selected it
be
should checked against shear,bearing, bending, punch-out and a combination of shear
and bending (Concrete Society 2003). Formulas for calculating each are provided in the
relevant guidance documents, and provide a maximum dowel load transfer capacity in
kN.
0,25
I= [Eh3/12(1_P2)kl equation3.1
Where:
E, = Youngs modulus of concrete

h= Slab depth
65
p= Poissonsratio for concrete
k= Modulus of subgradereaction
The dowel bar itself is commonly 12-25mm in diameter positioned at approximately

300mm spacingsacrossthe length of the joint. Due to this spacingit is acceptedthat only
the two to four dowels nearestthe load can be consideredactive. The length is basedon
the second point of contraflexure, but design manuals and texts have standardisedthis
suggestingvalues of 400-600mm.(Neal 1996).Generally the dowel is required at the mid-
section of the slab, with a deviation of no more than 20mm overall, and 3mm. between
each bar. Other standardshave slightly different specifications but they predominantly
comply with thosewritten above(Ringo and Anderson 1996).
This type of load transfer system is less affected by the movement of the slab and the
width of the crack than non-doweled joints, as any movement will only be a minor
proportion of the bar's full length. However, there is a reduction in load transfer that
occurs because of the looseness of the dowel from repeated loading cycles. This occurs
becausethe concretedirectly above and below the dowel bar begins to erode under load
and leadsto an ovalling effect (Porter et aL 1996).
The construction of the dowel bar system is vitally important as it can create increased
within the concrete if not correctly installed. The dowel should be designed to
stresses
enable independentmovement of one slab away from the other, and prevent locking of
the joints. This is achieved by placing a sleeve onto the dowel, or by painting on a
de-bonding agent, both of which enable the bars to slip on one side. The overall
suitable
design of the slab must also be examined in detail as the insertion of dowel bars around
the perimeter of the slab can restrain movement.Proprietary systemshave been developed
that contain polystyrene filler at the edges of the dowel bar to overcome this problem.
Different shapesof dowel bars varying from square or rectangular to more advanced
semi-circular sections have also been developed to allow for some horizontal movement
when required (Walker and Holland 2001).
3.4.3 Steel Fabric
The use of reinforcement and fabric is not provided as a load transfer mechanism,and is
generally used to control suitable crack spacing and prevent excessiveopening (American
Association of State Highway and Transportation Officials 1986). Section 3.4.1 describes
how the load transfer mechanismof aggregateinterlock is reducedwhen the crack begins
66
to widen. The ideal situation is therefore one in which there are many narrow cracks
insteadof a few larger ones.The reinforcementprovides this by distributing stressesover
the entire slab and restrainingany crack opening.
Tests undertaken by Benkelman (1933) indicated that the use of fabric reinforcement
significantly reduces crack opening when compared to those with no reinforcement.

Joints in similar locations were examinedand the non-reinforcedcracks showedincreased
loaded slab deflections of between54 and 146 percentwhen comparedto those that were
reinforced with steel fabric. The variation in crack width changed dependant on the
seasonaltemperature.
Reinforcementhas similar propertiesto that of a dowel bar in respectto the transference
of load between one slab and another.The Concrete Society (2003) has provided tabular
information on maximum load transfer capacities for commonly used internal meshes.
These are calculated for yielded steel as the shrinkage movement in the concrete will
reduce the section diameter and therefore lower its load transfer effectiveness. The
ConcreteSociety (2003) also remark that the choice of mesh should not be basedentirely
on its load transfer efficiency, as the selection of denserfabric to increaseits strengthcan

raise the risk of the joint not opening at all, resulting in mid-slab cracking.
3.4.4 Steel Fibres
Steel fibres are commonly used as a stress distribution material and reduce the overall
movement found in a slab (Balaguru and Shah 1992).The fibre, in resisting this
shrinkage
movement, createsa system of many fine cracks (often micro-cracks) whereby the crack
faces are close together. This effect is similar to steel fabric whereby aggregateinterlock
is increased,resulting in good load transfer (Raja and Snyder 1991).
Although not well researched,the steel fibre itself also provides a contribution to load
transfer (Swamy et aL 1979 and Schrader 1985). ACI Committee 360 (2000) state that
steel fibres offer additional shear load transfer across contraction joints in slabs on
ground, but offer no further information on the subject. The mechanismof sheartransfer
is similar to that of a dowel bar, although instead of one singular element there will be
many finer 'dowels' crossing the crack. The pullout resistance of the fibre from the
concrete will have a great influence on the load transfer value. This resistanceis a factor
of the fibre embedment,the orientation and the steel strength, and may thus be increased
or decreaseddepending on the fibre used and the construction process.Little guidance is
67
available on the load transfer effect of steel fibres and this requires further researchand
investigationif it is to be utilised in the future.
Thompson (2001) examined the effect of steel fibres in granular cement bound road
bases.Cyclic testing was employed on small samplescontaining fibre quantities of 0,
0.25,0.5 and 1.0% by volume, with parallel crack widths up to 1mm. The results
indicated that an increasein load transfer stiffness, and therefore a reduction in net shear
slip, occurs in steel fibre reinforced concreteat high crack widths when comparedto that
of non-reinforced sections (Figure 3.5). However, there was little indication that the
higher stiffness was proportional to the increase in fibre content. The load/shear-slip
relationship also altered when fibres were used as reinforcement within the specimen.In
non-reinforced tests the shearslip versus load relationship was non-linear with only 12%
of the maximum load producing 80% of the maximum slip (Figure 3.6). When fibres
were introduced this becamemuch smootherwith a near linear increasein slip with load
(Figure 3.7). This resulted in a more efficient crack as proportionally more load is
required to move the specimen. Thompson (2001) suggestedthis was caused by the
increasedfrictional restraint as the fibre bent around the aggregateand held the beam
together in tension.
3.4.5 Proprietary Systems
A formed joint at the slab edge can provide a satisfactory load transfer condition if
correctly designed.The mechanismmust contain some kind of interlocking member and
usually involves a dowel or plate device to transfer the load into an adjacent slab. The
load transfer efficiency of these joints varies between suppliers and as such it is not
possible for a singular value to be ascertained(ConcreteSociety 2003).
The construction is usually fonned with a pre-castpanel made of steel which is left in the
concrete, helping to strengthen the joint. There are many different types of former
available from various companies, each comprising a number of benefits. However,

problems can occur, with the main concern being inadequate compaction or under
vibration of the concrete around the former. This can create air voids and pockets of
loosened material which can crumble and crack under low loading (Concrete Society
2003). Neal (1996) also comments on the possible changesin surface level and excess
cement paste around the joint arris, both of which can lead to increasedspalling of the
slab edge.
68
3.5 Degradation of Load Transfer
Due to the vast number of vehicles passing over a joint or crack some degree of joint
degradationover time is to be expected.Many authors refer to the work of Colley and
Humphrey (1967) to determinethe amount of deterioration likely to occur. In this work
test slabs 1220by 5500mm in plan, with a non-doweledjoint at mid-span,were placed on
a subgradematerial and repeatedly loaded to a maximum of 53kN for up to 1 million
cycles. Measurementsof deflection were taken either side of the joint at 50,000 cycle
intervals for various joint openings, slab thickness and subbase/subgradetypes. A
selection of their results are reproduced in Figures 3.8 - 3.11, where effectiveness is
calculated using the Teller and Sutherland(1943) method of load transfer (section 3.8.1,
equation3.8).
Figures 3.8 and 3.9 indicate that the majority of degradation occurs during the early
stagesof the test, with Colley and Humphrey (1967) calculating that 90% of the I million
load cycle deterioration will have occurredduring the first 500,000cycles.
The greatest influence on joint effectiveness for all load repetitions and construction
methods is caused by crack width. For a 175mm (7") slab at 120,000cycles the width of
1.65mm (0.065") fails completely, whereasthe 0.38mm (0.025") crack is still running at
over 90% efficiency (Figure 3.8). Using a greater slab thickness of 225mm.increasesthe
effectiveness to some degree (Figure 3.9), with a 1.65mm (0.065") crack retaining 20%
load transfer at I million cycles, but even here small increasesin crack width lower the
effectivenessgreatly.
The results of laboratory testing were compared to data collected from a working site
constructed with similar materials and crack width properties.The laboratory data shows
increasedrates of deterioration, the magnitude of difference increasing with larger crack
widths (Figure 3.10). Colley and Humphrey (1967) suggestthe causeof this to be higher
loading in the laboratory test, and changing of crack widths due to seasonaland daily
temperaturechanges.The evidence for this is limited and its cause is more likely to be
from incorrect crack width orientation measurements.
Tests by Colley and Humphrey (1967) examined the effect of load and showed that high
magnitudes degrade the joint face at a much faster rate (Figure 3.11). When tests were
carried out at 40kN (9 kips) the load transfer fell continuously throughout 1 million cycles
due to aggregate attrition; however, when reduced to 31kN (7 kips) the load transfer
69
droppedup to 300,000 cycles, at which point it levelled off for the remainder of the test.
When a 22kN (5 kips) load was used the load transfer showedlimited changethroughout
the test, suggestingthat light loads causelittle wear of the interface material. Ioannides
and Korevesis (1990) similarly concluded that there is a critical value of load whereby
deterioration takes place at a much greater rate, and therefore loads below this value do
not generally needto be consideredin design.
Nowlen (1968) examined the effect of different aggregateproperties on load transfer
under a 50,000 cycle 40kN load for crack widths of 0.9 and 1.65mm. Using similar
apparatusto that of Colley and Humphrey (1967), maximum aggregatesizes of 19,38
and 64mm were testedunder a 40kN load for one million cycles. The results showedthat
an increasein aggregatehardnessproducedbetter effective endurancethroughout the test.
Nowlen (1968) also concluded a crushed aggregatein which any natural weaknesshas
been removed, showed better durability than a rounded aggregateeven though they were
made of the same material. For both crack widths an increasein aggregatesize improved
the effectiveness; however, the difference was far more noticeable between the 38 and
64mm.aggregatethan 19 and 38mm.
Other than the direct degradation of the concrete face, it is common for the subbase
material to compact under repeated loading. Testing undertaken by Colley and Nowlen
(1958) comprised granular or cement treated clay subbasesresting on a clay subgrade

'A 50mm thick concrete slab was placed on top and repeatedly loaded for 500,000
soil.
a 40kN load. Throughout the test were
measurements taken of deflection and
cycles under
under slab pressure, with visual observations of pumping also recorded periodically.
Increasing the subbasethickness, using well graded material and poor preparation were
all found to increasethe amount of compaction over time; however, using a soil cement
subbasereduced these effects and prevented the risk of pumping. Colley and Nowlen
(1958) undertook similar tests and concluded that an increase in foundation support
stiffness resulted in a much reduced change in effectiveness. For a crack opening of

0.9mm the one million-cycle effectivenessincreasedfrom 27 to 78% when the modulus
of subgradereactionwas from
increased 0.025to 0.125N/mm3.
Thompson (2001) investigated deterioration for fibre reinforced cement bound granular
road bases and reported that the rate of degradation was significantly reduced when
incorporating fibres into the mix (Figure 3-5). In specimenscontaining fibre fractions of
0.5% or more the magnitude of deterioration was negligible. This reduction in shearslip
70
and deterioration was thought to be causedby the fibres holding the crack face together,
increasingthe aggregateinterlock, rather than the load transferpotential of the fibre itself.
3.6 Slab Condition Testing
The use of pavementtechnology for the assessmentof floor slab condition is a relatively
new concept. As has been shown in section 2.2 the construction and loading conditions
between the structures are extremely similar and therefore the techniques of standard
practice in highway testing are beginning to be used for internal slabs. The method
generally involves measuringdeflections on the structure'ssurfaceeither side of a joint or
crack created by a controlled transient load, thereby enabling calculations of slab
response.
3.6.1 Equipment
Falling Weight Deflectometer
The falling weight deflectometer(FWD) is commonly used in the non-destructivetesting

of both flexible and rigid road pavements. The impact of a load on the surface of the
structure creates a deflection bowl which is recorded with a series of offset velocity
transducers (geophones). Examination of the slab profile then makes it possible to back-
calculate the stiffness of eachconstruction layer or determine joint effectiveness.
The FWD operatesby enabling a large mass to fall onto a system of springs, which in
turn generatesa load pulse (Figure 3.12). The magnitude of force is determined by the
load applied and the drop height, both of which can be adjusted to obtain a transient
contact pressureequal to that of a moving vehicle (Scott Wilson Pavement Engineering

2002). The load pulse rises from start to peak in approximately 3040 rnilliseconds, the
exact value being dependent on the machine used (Fleming 2000). The peak vertical
deflections are measuredat several locations radiating out to a maximum of 2.25m from
the source of load, with each geophonerecording to a resolution of 1 micron over a 2mm
maximum range. The positions of the geophonesare manoeuvrable depending on the
infon-nationrequired by the operator.
The FWD is mounted on a trailer bed to enable easeof movement, and is often hooked
onto the back of a vehicle to enable quick progressionbetween test locations. The power
required is obtained either directly from the vehicle or from a stand alone battery. A
cameraand laptop are often mounted within the cab allowing the driver to manoeuvreand
71
control the FWD without having to leave the vehicle. This configuration enablestesting
to be undertakensafely and at speedas the operator does not need direct accessto the
road pavement.
Prima Portable Dynamic Plate
The Prima is a small portable version of the FWD and is commonly used for measuring
the bearingcapacity of baselayers in pavementsand other forms of construction (Fleming
2000). The equipmentconsistsof a load dropped onto a seriesof buffers, which createsa
controlled force on the slab surface with a pulse duration of 15-20 milliseconds (Figure
3.13). The weight of the load can be varied between 10 and 20kg, with the drop height
further refining the impact. The loading plate can be changed to either 100,200 or
300min in diameter, enabling the contact pressureto simulate that of typical wheel loads
applied on the structure.
To measurethe force and slab deflection a load cell is situated within the Prima's main
housing along with a geophonedirectly underneaththe baseplate. Two further geophones
are attachedvia a communicationslead and are positioned with the aid of a rigid T-beam.
This set-up enables deflections to be taken anywhere on the slab, limited only by the
length of the lead. The geophones are accurate to +/- 2% and can measure vertical
movementto I rt-&ron, with a maximum range of 2.2mm.
3.6.2 Investigation Techniques
The deflections recorded with either piece of test equipment can be used to develop a
number of slab performance parameters.Each is determined with the careful positioning

of the geophones and load plate in relation to the joint, or with increasing load
magnitudes.
Load Transfer
As described in section 3.2, load transfer is required to transfer force across a

discontinuity betweenslabs.This reducesstressand deflection, and ensuresserviceability
and failures
structural do not ensue.
Section3.7 providesexamplesof load transferequationsusedin the assessment

of floor
slabsandpavements, basedon eitherdeflection,stressor engineeringjudgement.Colley
and Humphrey (1967) recommendthat stressequationsshould not be used due to
72
variations of load type in either slab. Similarly, 'engineeringjudgement' is a subjective
perceptionof joint effectivenessand is thereforedifficult to accuratelymeasure.
Deflection testing is thereforecommonly used for the determinationof load transfer.This

is generally calculatedby relating the deflection of one slab edgeto the other, providing a
ratio expressedas a percentage.Further information on calculation methodsis described

in section 3.7.
Load Step
Load step is the absolutevalue of differential vertical deflection between the edgesof a
crack or joint when under dynamic load. The step itself can affect the ride quality for any
vehicle crossing the joints and induce spalling of the crack edges.The magnitude of this
movement is dependent on the load applied, load transfer and foundation support
conditions. The calculation of load step can be achieved with the data obtained from
either piece of equipment, it simply being one edge deflection subtractedfrom the other.
Further information is provided in section 3.7.
Edge Cantilever
Edge cantilever determines the rate of bending of the slab edge and is achieved by
subtracting the deflection directly under the load from that of the slab edge on the same
side. Depending on the degree of joint stiffness the value of cantilever may be negativeor
positive. A positive result indicates a discontinuity, with a negative value showing good
load transfer and a slab that is working as an integrated unit. I-EghwaysAgency (1999)
comment that this cantilever effect is also of assistancein assessingthe quality of slab
foundation. Where it is deemedpoor the cantilever will be positive with a negative value
showing good support. As mentioned in section 1.1, there is an increased demand for
tighter tolerances in slab level and flatness. This must be assessedwhen under dynamic
load to ensurevehicle specifications are adheredto (ConcreteSociety 2003). Calculations
of slab edge cantilever enable the relevant standardsto be checked,preventing excessive

lean or damageto vehicles.The level of allowable floor variation varies dependingon the
categorisation, with the reader directed to the individual guidance documents if further
information is required.
Deflection Basin
Deflection basins enablea profile of the slab to be determinedwhen under dynamic load.
This can be used for calculating structural properties or examining the overall responseof
73
the slab.To determinea deflection basin severalgeophonesare placed on the slab surface.
Each is positioned in a line at increasingradial intervals from the load, one of which will
be on the opposing side of the joint or crack, with the remainder acrossthe same slab.
Plotting the deflections of these different geophones against distance enables
determinationof an extendeddeflected shapeof the slab surfacewhen impactedby load.
When an FWD is used to calculate soil properties, standardpractice dictates that it is
positioned at the slab centre as any interpretation assumesa semi-infinite length. Each
section of the deflection bowl will then provide information into the different layer
properties as shown in Figure 3.14. This analysis is often undertakenusing a trial and
error approach and therefore computer software has been developed to speed up the
processsignificantly. In situations where the FWD is set at the edgeof the slab in order to
determine load transfer, it is not possible to use this method of analysis. However, the
deflection bowl can still provide information into the structural performanceand effect of
the joint, with a peak deflection occurring directly under the load indicative of good load
transfer (Highways Department 1999).
Void intercepts
Void intercepts are used to assist in the detection of voiding underneaththe slab edge.
Crovetti and Darter (1985) found that they could be determined from deflection
measurements under three different load magnitudes when taken at a slab edge. As the
load increases the void begins to close until such a stage that full contact with the
foundation is produced (Figure 3.15). At this point resistance to deflection will be
increasedand the magnitude of deflection causedby a unit load reduced.If severalvalues
are recorded, a plot can be made of deflection against its associatedload magnitude. A
comparison can then be undertakenof the best-fit line and that produced if eachpoint was
simply joined together. Any significant deviation signifies a strong possibility of voiding
under the slab edge.
The best fit line can also be extrapolated until it bisects the zero load vertical axis, with
the point at which it crosses used to detern-nethe existence of a void (Figure 3.16).
Crovetti and Darter (1985) proposethat a vertical axis value of 50 microns or higher is an
accurate guide for detecting voided areas, although Wade et aL (1997) suggest 75
microns. Work undertaken by Frabizzio and Buch (1999) examined both of these
assumptions and suggested that the intercept magnitude could enable the degree of
support to be established. In both cases the values coincided with site data of known
74
faulted slabs, indicating the correct range had been found. White Young Green (2002)
undertook similar testing on sites in the United Kingdom and found that this value could
be reducedto 25 microns to enablesmaller voids to be detected.
Crovetti and Darter (1985) have stated that the value of void intercept is of direct
relationship to the size of void under the slab edge.The accuracyof this analysishas been
called into question by Tang (1993) who demonstratedthat there is little correlation
between the two. This discrepancy was thought to be causedby the varying nature of
void, with the depth, length and width of the area all having an effect on the value
obtained.
A more complex method to determine the magnitude of the void has been developedby
Crovetti and Darter (1985). In this approachdeflections are recorded at central, edgeand
comer positions at different locations acrossthe site. These values are then run through a
seriesof equations enabling a lower bound of zero voiding to be developed. Comparisons
of the individual test points can then be made to determine whether neither, one, or both
sidesof the slab are affected.
Another approachto void detection was developedby Ricci et al. (1985). This method
compares the slope of the line obtained when joining up the deflection values of selected
FWD geophones(W). Equations 3.2 and 3.3 used in the analysis are shown below, with
the locations of the geophonesgiven in Figure 3.17.
m Arc tan [61(WI -W2)] equation 3.2
Q Arc tan [(W2-W7)/24] equation 3.3
The results of thesecalculations are analysedwith the following:
If Q is greaterthan 18, a loss of support is indicated

If Q is greaterthan 18,the smallerM is thelargerthelossof support
If Q is lessthan 10andM is greaterthan70, thanfull loadtransfercanbe assumed.
3.7 Joint Effectiveness Analysis
A number of methodshave been used to calculate the effectivenessof load transfer.Many
of these are based on deflections either side of a known joint or crack from an imposed
75
load, with a percentagemark given to illustrate the deflection being transferredfrom one
slab to another.Other methodsutilise the variation in stressunder the slab edge or finite
elementmodelling to predict the load transfer.
Many terms such as 'load transfer', 'joint efficiency', 'load transfer efficiency' and 'joint
transferefficiency' have beenusedto describethe load transfer mechanism,with different
authors using the various terms to depict specific calculation methods. In this thesis the
terms are all interrelated, with the separateequationsdescribedby the author who first
used it during analysis of cracks and joints. Where possible 'load transfer' has been used
to provide clarity to the reader.
Equation 3.4 (Crovetti and Darter 1985) producesa load transfer value (LT) createdfrom
the deflection of the loaded (dI) and the unloaded (du) slabs. In the case of a fully
transferredload the result will be 100%.In a free edge situation where the load transfer is
negligible then the LT will be zero. Cracks are deemed to have adequateload transfer
when the LT is in the region of 65-70% or above(Frabizzio and Buch 1999).
LT = (du/dl) x 100% equation3.4
Equation 3.5 (loannides and Korevesis 1990) yields a slightly different result as it takes
into account the amount of load transferred across the crack rather than the deflection.
The term Pt is derived from the total load transferred from one side of the joint to the
other throughout its entire length. The term P is the total externally applied load, and
therefore, in most situations, the value of transfer load efficiency (TLE) will be below 50
%. (Buch el aL 2000, pp 329) advise that the Pt value should be used alone as "it has a
more physical meaning than its counterparC' and can be back calculated from the
transferred load efficiency (TLE), as shown in equation 3.6. Ioannides and Korevesis
(1990) have provided graphical relationships between TLE and the LT calculation in
equation 3.4 for various a/l ratios, as shown in Figure 3.18, where LT is expressedas
LTEs. Buch et aL (2000) used the faulting data obtained from site testing to develop a
threshold parameterfor Pt, the value dependingon the structural design and applied load.
TLE = (Pt/P) - 100 equation3.5
Pt = (TLEI 100) -P equation3.6
76
Aggregate stiffness (AGG) is another factor which is used by Buch et aL (2000) to
provide an indication of joint effectiveness,and producesa stiffness value per unit length
of crack. Buch et aL (2000) state that by determining the load transfer (where LT is
expressedas LTE8), and obtaining values for the radius of relative stiffness (1) and
modulus of subgradereaction (k) the AGG value can be derived from the graphical
relationshipdevelopedby Ioannidesand Korevesis(1990) (Figure 3.19).
Other researchershave adopteddifferent methodsof analysis,most of which are a direct
comparisonof loaded and unloaded slab deflections. Pradhan(2002) comparedfive of the

different equationsavailable commenting that most of the methods gave similar results.
He argued that his new method of analysis (equation 3.7) was more accurate as it
incorporated support from the foundation resulting in higher values. This theory was
tested against a visual survey indicating that all of the testedjoints were satisfactory as
indicated by a plus fifty percent load transfer value. In essence,his method is the sameas
Crovetti and Darter's (1985) with the unloadedand loadeddeflections presentedin such a
way that in almost all situations the load transfer value will be above 50%. This therefore
provides very little extra indication as to the effectiveness of the joint.
LT = (du + dl)1(2d1)- 100 equation3.7
Other methodstypically used in the determination of joint efficiency from unloadedand

loaded deflections are provided in equations3.8 to 3.11. Severalof the equationscompare
the results of the approachand leave slabs,this being a changein the loaded slab edge.
(Teller and Sutherland 1943)
LT = [(2 - du)l(dl + do] - 100 equation3.8
(Jackson et al. 1994)
LT = Smallest of LTud and LTdd equation 3.9

LTasd = (dlldu)1100
LTd,d = (dlldu)1100
(Ricci et aL 1985)
LT (ASDR + LSDR)12 equation3.10
77
(dUldl)asd
ASDR =
LSDR = (duldl),,d
(American Association of State Highway and Transportation Officials 1986)
LT (duldl) -B- 100 equation3.11

B dlcld2c
Where:
LT Load transfer
du Deflection of unloadedslab
di Deflection of loaded slab
dI12, = Deflections at slab centre (305mm apart)
LT, = Load transfer of approach slab

isd
LTdsd= Load transfer of leave slab
ASDR = Approach slab deflection ratio
LSDR = Leave slab deflection ratio
The American Association of State Highway and Transportation Officials (1986) use a J-
factor to incorporate the joint efficiency into the design of pavements.The Mactor has a
significant influence on slab response, and as such must be chosen with care. The
guidance provided in the document for the selection of this value is limited, with the
general statement "higher J's should be usedwith low V values, high thermal coefficients,
and large variations in temperature" (American Association of State Highway and

Transportation Officials 1986, pp 11-27).If 'F values are to be used confidently more
information needs to be supplied to assist the selection of specific values for different
conditions. Kuo (1998) has researchedthe effect of Mactor on the outcome of the design
and concluded that there is no significant relationship between the J-value and load
transfer efficiency.
Other methodssuch as that of Sutherlandand Cashell (1945) have been developedwhich
compare the stressesunder each slab rather than the deflections (equation 3.12). Colley
and Humphrey (1967) argue that this method is not strictly reliable because of the
different types of loading on either side of the crack. The slab under the load is
undertakingdirect bearing, whereasthe adjacentload is createdpurely from shear.
78
LT. (Ff - Fj) / (Ff - Fi) equation3.12
Itmss=
Where:
Stressrelated load transferefficiency

Ff = Critical stressfor load at a free edge
Fj = Critical stressfor load at interior
Fi =Critical stressfor load at a joint edge
Gulden and Brown (1985) suggestcaution when using load transfervaluesalone to assess
a joint or crack within a slab. Due to the nature of determining load transfer the overall
deflection will have a large impact on the value obtained. An example of this is shown
below in Table 3.1, where joint efficiency is calculated using equation 3.4 and load
transferwith equation3.8.
Table 3.1 - Load Transfer Comparison (Gulden and Brown 1985)
Test location Loaded Unloaded Load Transfer

Deflection Deflection M
1 6 1 29
2 10 5 87
3 35 30 92
Clearly test location I has the most favourablejoint condition despitehaving the lowest
load transfer. Thus, Gulden and Brown (1985) recommend using the overall and
differential deflections between slabs,rather than a load transfer value in the assessment
of joints. Unfortunately, very little data is available giving guidance on recommended

levels for these two values and as such the load transfer value will invariably be used.
Informed by the work of Pearson(1999), Cudworth (2000) has produceda criterion table
to which it is recommendedslabs adhereif they are to show satisfactorybehaviour. The
work states that differential deflection should be limited to 100 microns, with the overall
deflection on one side of the joint no greater than 150 microns. A value of 65% load
transfer is provided for completeness,although the method of calculation is not provided.
Cudworth (from Arnold 2002) has suggestedthat a Falling Weight Deflectometer(FWD)
load step of 300 microns, created from a 50kN impulse load, is indicative of a
serviceability problem within the slab. This value was stated to be relevant only for the
test equipmentand not as an absolutevalue of site dynamic faulting. However, the FWD
is supposedto simulate the impact of a vehicle on the slab's surface, with a 49kN load
79
given by White Young Green (2002), and therefore it can be assumedthat this 300-
micron value (although site specific) should be representativeof general serviceability
problems.
Nishizawa et al. (1989) developedequation3.14 relating load transfer(U) to crack width

(w), with Ioannideset al. (1990) proposingthat this was a best-fit straight line through the
laboratory data collected by Colley and Humphrey (1967). Ioannides et al. (1990)
suggestedthat the values obtained could be cross referencedwith their graphs enabling
the effect of crack width on other slab responseparametersto be determined.They did
commentthat this approachwould be hazardousas it combinesan empirical relationship
with a mathematicalfunction.
LT= 100 - (25 - w) equation3.14
The equationdevelopedby Nishizawa et aL (1989) is a generalisationand neglectsmany

of the important factors which determine load transfer effectiveness.According to the
equation a 3mm crack would still contain 25% load transfer, whereas Buch (1999)
discovereda parallel crack with a width exceeding1.5mm would provide no load transfer
at all.
Buch (1999) examined the effect of pavementproperties using an analytical model. The
results demonstratedthat an increasein slab thicknessfrom 150 to 400mm, or raising the
subgraderesilience from 0.027 to 0.135 N/mrr?, produced lower load transfer efficiencies
despite the joint being of the same stiffness. However, the deflections of the slab edges
were reduced leading to lower stresses and a stronger pavement system. Gulden and
Brown (1985) also statethat load transfer and load efficiency values increasewith higher
deflections. In analysing the data from field tests they utilised the differential deflections
betweenslabs(i. e. the load step) as it gave a better indication of performance.
3.8 Summary
Load transfer acrosscracks andjoints is vital to the long-term serviceability requirements
of concrete floor slabs and Much

pavements. of the researchundertakenhas shown that
its effectivenessmust remain high throughoutthe duration of slab life if maintenanceis to
be kept low, and failure prevented.
The selection of an adequateload transfer value for the structural design of a concrete
slab is essential if deflections and stressare to be kept within acceptablelevels. This is
80
generally undertaken assun-ngcrack widths will open equally throughout the site,
neglecting the fact of the dormant and dominant joint behaviour. Deterioration of the
crack is also often overlooked with a single value assumedthroughoutthe structure's life.
The correct equationfor calculating load transfer must be statedfor analysis,as eachwill
producedifferent values.
The crack openingitself is well documentedas being larger at the surfacethan at the base,
causedby differential shrinkageand slab curling; however, the gradient of this variation
is unknown. Many tests have been undertakencomparing crack width to load transfer,
althoughonly surfacemeasurementswere used.
The load transfer phenomenon is known to be controlled mainly by the aggregate

interlock effect regardlessany extra mechanismsinserted acrossthe crack or joint. This
has been broken down into the local (micro) and global (macro) roughness,with macro
roughness only being utilised once cracks become large or crack face deterioration
occurs. The role of fibres as a load mechanismis still relatively unknown, although it is
thought to aid in load transfer acrossthe crack.
There are many methods of analysing slabs using non-destructive deflection testing
techniques.Many of these are commonplacein the examination of external pavements;
however, they are relatively new in the assessmentof internal floor slabs. Methods to
determine load transfer, load step, edge cantilever and void intercepts can all enable
actual site behaviourto be quantified.
The purposeof this study is to addressthe lack of understandingin load transfer across
joints and cracks in concreteslabs on grade.This will involve the examinationof typical
geometries of cracks and joints from in service sites and the determination of slab load
transfer and deflection related responses.The longer-term load transfer effects will also
be examined in respectto the deterioration of the crack over time. This will incorporate
the relatively unknown mechanismof steelfibres.
81
LLEIAIV
APPROACIf E
LTAPPIL -- "LF.
"I
EAPPqO:
ACH "I'vt
LTAPPM, > LTL90,Vg
Figure 3.1 - Effect of crack verticality
Global Roughness
(Macro Texture) ocal Roughness
Aicro texture)
Figure 3.2 - Global (macro) and local (micro) roughness
I
.1
00
. '9
Figure 3.3 - Aggregateinterlock model (Walraven 1981)

rJ
82
Agwegate
77
Figure 3.4 - Effect of crack opening on aggregate contact (Walraven 198 1)
250
I (JO
0
50
0A
0
0.0 o
Crack w4th mm
Figure 3.5 - Effect of crack width on net shearslip (Thompson2001)
83
I I
$hear slip lim
-200 -150 -100 -50 0 50 100 150 200
2.0
-1.0
-0.5 z
0 Asw W* -0 40 0
-0.0
--0.5
Ef Mix GI (Control)
Serios
19
BOMM
w--O. --2.0
Figure 3.6 - Load/shearslip plot for non-reinforcedspecimen(Thompson2001)
Shear slip pm
-80 -60 -40 -20 0 20 40 60 80

. Ic-u-
- ZO
-1.5
1.0
a
me,86,1- -0.5
-0.0
-0.5
-1.0
Mix Gl (V"-. 5%)
Serlas 15 --1.5
w--0.83mm --2.0
--2.5
Figure 3.7 - Load/shearslip plot for fibre reinforced specimen(Thompson2001)
84
lcol o oi's. im
d)
'80
cl
60
wr
40
20. 0,04545
0065
0 C)r
40 .
ui
23567a
Loodiog Cycles, 100 000
Figure 3.8 - Joint deteriorationof a 175mmslab (Colley and Humprey 1967)
Join-t G025 "I.
M)
06
ci
101
I2
t-oadinq Cpfles, (00000
Figure 3.9 - Joint deteriorationof a 225mm slab (Colley and Humprey 1967)
100
60
.t.)
;to
20
LL) LaboralorY
04 OB 12 IG
. , 120
Figure 3.10 - load transfervaluesfrom laboratory and field testing (Colley and Humprey
1967)
85
!5 -Kip
5 I, j U
tj qrwoek subbose
L0Cdinq cycles.
00()Go
FigUre 3.11 - Effect of load on joint deterioration (Colley and Humprey 1967)
SN
11 1 'All
12
Tiansdwco, s
-Qf
'* I..
'01 ,II II!
FigUre 3.12 - FWD (Top - graphical representation. Bottom - Plate)

11
86
.u ,, 0:", ,
Figure 3.13 - Prinia dynamic loading plate
Lood
Radial Dletance
1 '193
dI
Subgrods
1 Sub-base
4___ -0
Bass
I Surfacing
Figure 3.14 - FWD deflection basin analysis
87
Void
77';N
AN
Figure 3.15 - Void closureunder load
55,1
5, G75X 6.503
3350
23
"ISO WOW- 0 Jckntw0aut undorSWbvoids,
100 -G
50
-thoar fJovit k1pith
undmUb voids
10 M 30 40 so 60 70 W so 100
FV40 Load 101)
Figure 3.16 - Void Intercepts(Cudworth 2003)
88
rAL LING WEiGHT
W2 W4 141 tys wl
PCC SLAS JOIN,
SVEGRADE-
FALLING WEICHT LOAD APPLIEO VPSTATIONOF JOINT

SHCULD SC APPROXINWIEtr Mwn 70 3 FOR 100% MAD TRANSHR
A INDICATESDEFLECTIONAT C=C;; 40N=. lYi
Figure 3.17 - Geophonelocations (Ricci et aL 1985)
50
40
30
20 Symmetric Ed9o Lood
00/9 - 0.047
00/4 - 0156
10 so/4- 0.312
&M/4 - 0_"4
0
0 10 20 30 40 50 60 70 80 90 WO
LTCS(7. )
Figure 3.18 - LT versusTLE (loannidesand Korovesis 1990)
0 ij-
Syr-nerm
tr -, J%g-2Laqc
30
210
0% . ... I- &l4.. "6&ALLLJ

10
3- 22 1 IioI
1 0- 10.1 11I'D -1 CO -1 IV - lo
Non6morisional St'rfriess (A%G-wkt)
Figure 319 - LT Vs AGG (Ioannidesand Korovesis 1990)
89
4. SLAB CONDITION
4.1 Introduction
Field testing was undertakenat four in service sites throughout the researchperiod, these
being at Daventry, Lutterworth, Ballymena and Skelmersdale.All except Daventry were
examinedonce Engineershad been called in to assessand repair serviceability problems.

Due to constraints on equipment availability and site accessibility it was not always
possible to complete FWD and Prima testing on every visit. This was most prominent at
Ballymena and Skelmersdalewhere difficulties were found in gaining accessto the FWD.
As it was important to gather as much information as possible, these sites were tested
thoroughly using only the Prima Dynamic Plate. Coring was undertakenat Lutterworth by
an external company under guidance from a consulting Engineer. The placementof the
cores was dependanton the requirementsof the slab assessmentbeing carried out due to
serviceability problems, although on a few occasions specific areas were selected to
incorporate the research objectives. Information from three other sites, namely Leeds,
Marston and Northampton was also used for determining crack geometry, although direct
access was never undertaken. The data was obtained via work conducted by Bishop
(2001), in which regular visits were made to record strain measurementsfrom gauges
embedded acrossjoints. The collection of data was categorised into three main areas
comprising crack measurement;surfaceprofile measurement;and deflection testing.
Crack measurementenabledtypical geometriesto be ascertained,which were then usedin

of
the assessment the deflection test results, and the developmentof crack profiles for use
within the laboratory test program. Comparison could also be made against the test
methodologies and analysis techniques used by other authors. The process required the
determination of changing width with depth, rather than just a single surface
measurement.
Slab surface profiles provided an indication of edge curling caused by differential
shrinkage. Although the method could not be used alone to determine voiding under the
slab, it was useful as a prediction tool. The surface crack measurement was also
incorporated with the curled profile to enable calculation of crack geometry using simple
trigonometry.
90
Deflection testing using either the FWD or the Prima dynamic plate enabled slab
condition to be assessed.This was accomplished using a variety of analysis methods,

previously described fully in section 3.6.2, and included: load transfer, load step,
cantilever deflection, void intercepts and, for the FWD only, deflection bowls. The
methodsof calculation and the equipment set-up usedfor determining each parameterare
discussedin sections4.4 and 4.5. The relationshipsbetweendeflection responsesenabled
the influence of critical factors such as crack width and voiding to be determined.These
results were also used in the verification of the finite element model.
A matrix showing which information was collected from each site is shown in Table 4.1,
where the black areas signify areas of data collection.
Table 4.1 - Site Testing Matrix
6 ca m
Surface Measurements
Coring
Strain Gauges
-F- .
, rc(:ise Level
Builders Level
Profilonletcr
l,,WD (Single Load)

FWD (Variable Load)
Prima
4.2 Crack Measurement
Where applicable, full crack geometries were obtained through the Bishop (2001) results
from embedded strain gauges installed at Leeds, Marston and Northampton, and frorn
coring at Lutterworth. Unfortunately, as an external party had ordered the cores as part of
a structUral Survey the locations were unable to be determined by the project's objectives.
Fortunately, some of these were taken through sawn joints and cracks, and several were
bored through a wire guidance systern which was behaving as a crack inducer. At
91
Daventry, Lutterworth and Ballymena surfacecrack measurementswere taken, allowing
comparisonwith slab deflection response.
4.2.1 Surface Measurements
Accurate measurementof surface crack widths is known to be difficult (Ibrahim and

Luxmoore 1987).Ideally, some form of measurementmarker is inserted acrossa joint or
crack prior to movementoccurring. In this manner an accuratevalue can be ascertained,

with its development monitored over a period of time. Unfortunately many of the
structuresexamined herein were already in use and therefore required the utilisation of
other methods.This resulted in a greater risk of inaccuracy and therefore care was taken
to ensurethe most appropriatetechniquewas selectedfor eachlocation at every site.
Surface measurementsof crack width can be prone to slight discrepancydue to the nature
of development and edge deterioration. Some of the cracks found on site were many
metres in length and consistedof a varying surface width. Occasionally thesehad divided
into two within close proximity, creating difficulties in calculating an overall value. In
circumstanceswhere the crack had been open for some time, degradationand spalling of
the sideshad occurredleading to what appearedto be a much wider surfacecrack. Careful
inspection of the overall area was therefore imperative to ensure a representativesample
was recorded. This was ideally a section which had a similar width for a distance of
approximately lm, and showed little surfaceedge damage.
When evaluating the crack width within a filled sawn joint, accuratemeasurementwas
problematic. The saw-cut had a depth of around 1/4 of the slab thickness (approximately
50mm), which made identification and insertion of a measurement device almost
impossible. To overcome this the overall joint size was measuredand a reduction made
for the saw cut alone, the width of which was obtained from cores and slab sections
containing zero cracking (Figure 4.1).
The crack or joint width measurementwas taken using a variety of instrumentsdependant
on its easeof visibility. When clear at the surface an optical microscope was used as it
was the most precise method (Figure 4.2). This contained a measuring gauge in the
eyepiece,with a resolution of 0.02mm enabling a precision of +/- 0.01mm to be recorded.
When the crack was too large for the microscope (such as in the case of a sawnjoint), a
set of callipers were inserted approximately 10mm below the slab surface (Figure 4.3).
Inserting the device to this depth prevented spalling close to the joint surface being
92
included in the calculation, and enabled variations in measurementof O.Imrn to be
detectedwith a precision of 0.05mm.
Two other devices occasionally used for quick and easy estimation of crack width
measurementwere a standard 30cm ruler and a crack comparator. Although not as

accurateas the previous methods due to operator interpretation, these could be used in
situations too difficult for the other devices due to spalled edgesand/or rigid fillers. The
crack comparatorusesvisual comparisonsagainstline thicknessup to 4mm.to producean
estimation of crack width with a precision of 0.1nim (Figure 4.4). The ruler was used to
record much larger widths, although it can only measureaccuratelyto within 0.5mm.
4.2.2 Embedded Strain Gauges
Work conducted by Bishop (2001) required the installation of strain gaugesat various
locations throughout slabs at Leeds, Marston and Northampton, some of which were
placed across sawn and construction joints at different depths. Measurements of
movement due to environmental and climatic effects were then taken throughout the early
agesof slab life to develop an understandingof the thermal and hygral effects.
Examination of the records enabled strain to be calculated at joints over a set period of
time. According to Bishop (2001) all strain can then be translated directly into crack
width to determine the size of the opening. In some instancestwo strain gaugeshad been
inserted at the same position, but at different depths within the slab, enabling
extrapolation of surface and base values if a linear variation is used (Figure 4.5). This
approach was verified using 3-gauge monitoring, where the insertion of demec pips onto
the surfaceof the slab directly above strain gaugesenableda more complete geometry to
be produced(section 6.12).
Full information on the type of strain gaugesused, the methodology of placement and
accuracycan be found in Bishop (2001).
4.23 Coring
At Lutterworth,coringenabledcrackgeometryto be investigatedthroughoutslabdepth;
this was undertakenby a specialistcompany.Onceremovedthe coreswere takento a
laboratory where Construction Materials Testing in Derby determined the concrete
properties and constituent materials. Possession was then temporarily given to

Loughborough University where the crack geometry was examined in detail, involving
93
the measurementof variation in width with depth, and the distanceof the crack edgeto an
arbitrary vertical datum. This was undertakenwith the use of a standard30cm ruler, crack
comparator,callipers and a crack microscope,as describedin section4.2. Measurements
were taken at 20mm intervals of depth on opposing sides of the core to enable any
varianceto be established.
When measuringthe crack it was important to ensurestressrelaxation of the core had not
occurred.This could have altered the width dependingon the internal forces acting within
the slab. To account for these movementsthe dimensions of severalcores containing no
discontinuity were taken to provide a comparable standard. Dimensions were then
checked at the top and bottom of each cracked specimen,and any deviancies from the
standardidentified. Simple calculations could then take place to amendthe crack widths
recorded, producing a corrected value, as shown in equation 4.1. The accuracy of the
method was dependant
on the device usedto measurethe crack and the core diameter.
Actual w= Core w+ (Uncrackedcore 0- Crackedcore 0) equation4.1
Where:
W Crack width
0 Diameter
4.3 Curling/Warping
4.3.1 Introduction
Identification of the slab edge profiles enabledthe degreeof curl to be establishedusing
simple geometric calculations. This could be then be used alongside the crack surface
measurementsto determine the overall joint geometry (equation 4.2). The results from the
embedded strain gauges and coring had ascertainedthat crack opening was approximately
linear (section 6.1). Occasionally a greater degreeof movement was found at the surface
which would lead to the estimation of larger base measurementswhen using surface
profiles. As this effect was only found in a few situations, the method was deemed
acceptablefor producing representativeorientations with a lower bound crack angle.
wl, = w, - [(V/H) x 2h] equation4.2

Where:
94
wb= Basecrack width
w, = Surfacecrack width
V= Difference in slab surfacelevel
H= Horizontal distancebetweenlevelling points
h= Slab depth
Differential shrinkagemay not necessarilylead to curl in the slab edge, although it may
cause cracking. Movement is prevented by the self-weight of the slab forcing it down
onto the subbase, resulting in the formation of a crack, but no proportional curl.
Measurementsof the surfaceprofile may therefore lead to an underestimatein the crack
angle, and wider crack baseestimations(Figure 4.6). This will provide a lower bound for
basemeasurementprediction, similar to that describedabove.
4.3.2 Precise Levelling
The use of a precise level enabled accuratesurface profiles to be detennined at various

locations within the Daventry site. Firstly a grid of referencepoints was set out, with the
intensity increasing with proximity to the joint. Vertical elevations were then recordedat
each grid-point producing a numerical 3-dimensional plot of the surface. The Golden
software Surfei`rmgraphing packagewas then used to interpret the data and translatethe
numerical values into contour and three-dimensionalviews.
Unfortunately, the levelling processwas found to be extremely labour intensive, with the
setting out and data collection requiring a great deal of time. With most of the warehouses
examined containing over 100joint or crack positions, and only limited accesstime, this
method was too inefficient and was thereforeused solely at Daventry.
433 Builder's Level
The use of a builder's level was an effective method of obtaining curling estimations at
Ballymena. In most situations the slab curled upwards at its edges due to the shrinkage
differential. Placementof a builder's level at the joint enableda vertical measurebetween
the bottom of the level and the surfaceof the slab (Figure 4.7). This provided information
on the magnitude and rate of curling at either side of the joint, with an increasednumber
of measurementpoints along the level generatinga more detailed profile.
The builder's level method contained some limitations as its short length restricted the
amount of data obtained. The accuracy was also open to error as it was difficult to gauge
95
the distance between the underside of the level and the surface of the floor slab. A
graduatedwedgeprovided someimprovement,but still only enabledmeasurementsto the
nearest0.5mm to be obtained.
43.4 Profilometer
The profilometer was a useful way of identifying curl in the edge of the slabs at
Ballymena. This device consists of a backboard (which was set level) and a number of
needles, which when pressed against the surface provided an exact copy of the floor
profile (Figure 4.8). A graduated scale set against the top of the pins enabled comparisons
between sections, and therefore any variations were easily determined. This gave similar
results to the builder's level except that the increased amount of pins, and the easily read
scale, allowed an accuracy of 0.2mm to be achieved.
This was found to be a slightly quicker method of measurementand interpretation than

both the precise level and the builder's level, although it was restrictedto a single line of
with
measurement finite length.
4.4 Deflection Measurement
Deflectiontestingwas undertakenusingboth the FWD and the Primaportabledynamic

plate. Both devicesutilised identicalgeophoneplacementand load plate size to reduce
any error causedby set-up variations.As the FVVDand Prima were borrowed from
calibration
organisations,
external of the deviceswas by
undertaken the individualowners
the
to ensure stated
specifications in section3.6.1weremet.
In all cases a 300mm diameter load plate was used for both the FWD and Prima, as
recommended by the Highways Agency (1999). They also suggestusing a 75kN FWD
load on concrete pavements where the deflection may be below 100microns; however,
White Young Green (2002) advised using the standard 50kN load instead as this more
closely represents that of a forklift. Unfortunately, control of the FWD loading was
limited by the requirementsof the Engineerswho were undertaking tests to examine slab
condition. On all sites an initial settling drop was provided to seat the loading plate and
check for any anomalies.The recording drops were then initiated, with three loads of the
same50kN magnitude used on the Daventry site, with values of 42,58 and 85kN applied
at Lutterworth to enablevoid interceptsto be obtained (see section4.5.5).
96
For the Prima, each site was tested with the highest possible load to obtain the greatest
amount of deflection. This provided a greater range of responseand produced the most
variation between points. To accomplish this, 20kg weights were applied, and the drop
height was set to maximum. Davich (2000) recommendsusing the IOOMMload plate
when testing rigid materials to achieve the highest bearing stress; however, to enable
direct deflection comparison with the FWD, a 300mm diameter load plate was selected.
Using this method,a force of approximately lOkN was achievedon all sites.
The main requirementfor testing was the determination of load transfer acrossjoints and
cracks. On sites using either the Prima or FWD the centre of the loading plate was
positioned 250mm from the crack face. Geophoneswere then placed 50MM either side of
the joint leaving a gap of 50mm to the edge of the plate (Figure 4.9). This was specified
as work conducted by Poblete et aL (1988) concluded that load transfer effectiveness
reducesas load is moved further from the slab edge. In addition, this set-up provided a
real edge loading condition without the possibility of spalling affecting the results. The
remainder of the FWD geophones were placed at increasing spacing along the slab
surface, the exact locations varying between the Daventry and Lutterworth sites (Figure
4.10). The geophonepositions provided different information with respect to deflection
bowls. A profile was recordedfor the loaded slab (slab with load applied) at the Daventry
site, whereas for Lutterworth, the unloaded slab deflection (slab opposing load
application) was produced.
The majority of joints were tested in a single direction only due to the difficulty in
manoeuvring the equipment within each aisle. The load transfer variation in approach
(upstream)and leave (downstream)slabsdescribedin section 2.4.4 was preventedby the
equal forward and reversemovementsof the MHE. Furthermore, the changein verticality
of the crack was shown to be below 25mm. in the cores taken at Lutterworth (section
6.1.1) and was therefore assumedto have little effect the on results. The lack of data in
the literature assessingthe problems associatedwith crack verticality compounded this

conclusion. The small number of joints that were tested in both directions demonstrated
good agreementas shown in Figure 4.11, where the maximum variation in load transfer
was below 10%.
Slight variations in load are expected between test locations caused by the method of
impact and slab response. To enable comparison between points it was necessaryto
normalise the data to a load magnitude applicable for all points tested under the same
conditions. At the Daventry site, a 50kN FWD load was used, with that at Lutterworth
97
using 42,58 and 84kN to assistin void detection. For all sites a lOkN load was selected
for the Prima dynamic plate, which was close to the device limit. Normalisation consists
of a proportional increaseor decreasein deflection dependanton the actual magnitudeof

load, and assumes a linear load deflection response. Although this was not truly
representativeof all situations, the small variations involved had little effect on the
results. The method of calculation is shown in equation 4.3.
ND = (ADIAL) x NL equation4.3
Where:
ND = Norrmlised deflection
AD = Actual deflection
AL = Actual load
NL = Normalised load
The nature of the testing procedure using both the FWD and the Prima dynamic plate
enabled repeatability to be checked on many joints. Initially, specific points on the

Daventry site were to be retestedduring the secondvisit to check for any anomalieswhen
using the sameequipment. However, continuing fatigue and changesin temperaturesare

known to cause variation in results, as shown by Benkelman (1933), and therefore this
could not be used as an accurate measure. The past effectiveness of the FWD equipment,
the three load drop testing procedureand the additional testing of the Prima dynamic plate
were therefore deemed to provide a satisfactorycheck.
4.5 Slab Analysis
on eachsitethedynamicdeflectionmeasurements weredeterminedusingeithertheFWD
or the Prima dynamic plate. Unfortunately,
it was not possibleto use both piecesof
equipmenton every site due to availability and logistics of accessto smaller aisled
The
warehouses. FWD is the standardmethod of deflection testing,with the Primaused
on all sites,and thereforecomparisons between the two devices wererequiredto enable
correlation,and ensure the resultsfound acrossdifferent sitescould be crossreferenced.
As the equipmenttakesonly one measurementat a particularpoint in time, the results
obtained were only snapshotsof the current behaviour of the slab. The level of
deteriorationcausedwithin the crack from cyclic loading could not be determined;
however,knowledgeof slabageandthe resultsfrom the laboratorytesting(Chapter6.4)
good
enabled estimationsof joint or crackconditionto be established.
98
4.5.1 Load Transfer
Load transfer is the relationship of deflections either side of a crack or joint causedby a
dynamic load (section 3.1). To generatea value for load transfer, deflections from the
dynamic force were measuredon both the loaded and unloadedside of the crack. Authors
have suggesteddifferent ways of calculating this, trying to place the emphasison either
the loaded or unloaded slab deflection (section 3.8.1). The simplest of these was
suggested by Crovetti and Darter (1985), it being a direct relationship between the two
deflections expressedas a percentage(equation 3.4, and re-iterated below for clarity).
Due to its effectivenessand unbiasedapproach,this method was used for the majority of
site data analysis.Where alternativeformulae were usedto enablecomparisonswith other

author's work, it hasbeenstatedexplicitly within the figure or text.
LT = (du/dl) x 100%
Colley and Humphrey (1967) commentedthat load transfer is affected by the magnitude
of load imparted onto the slab, particularly when curled at the edges.As the FWD has a
load magnitude 4-5 times higher than that of the Prima, comparisonsbetween the two
pieces of machinery was thought necessary to examine the relationship. Numerical
models of uncurled slabs were also used to examine the effect of load magnitude on
transfer efficiency, with the results presentedin section7.6.
4.5.2 Load Step
Load step is the variation between the loaded (dI) and unloaded (du) deflections either
or joint (section 3.6-2), is
and calculated using equation 4.4. As the step is
side of a crack
a function of directly measured deflections, its value is highly sensitiveto changes in load
magnitude. With the FWD force being 4-5 times greater than that of the Prima, and
increasing loads being used with the FWD to ascertain voiding, significant differences
were expected. Comparison between load magnitudes were therefore required for each
site to enable its effect to be ascertained,allowing the relevant data to be used when
calculating any relationships.
The loadstepvalueobtainedis thatcausedunderdynamicloadonly. Any permanentstep

accumulatedin the slab due to foundationcompactioncould not be detectedby either
piece of equipment. In this situation levelling or surface profile measurementswere
required to obtain the relevant data.
99
Load step= dl - du equation4.4
4.53 Edge Cantilever
As defined in section 3.6.2 cantilever deflection is used for the assessmentof

discontinuities (cracks), foundation influence, and determination of floor flatness
requirements.For both the Prima and FWD, geophoneswere positioned 50mm from the
loaded (dl) and unloaded (du) crack faces, with another placed 200mm back directly
underneaththe loading plate (do) (Figure 4.9). This enabled deflections of at least two
points on the loaded slab to be recorded,providing an indication of the edge cantilever.
Subtractingone of the geophonevalues from the other allowed the degreeand direction of
bending in the slab to be quantified (equation4.5).
The magnitudeof cantilever varies dependingon the size of load usedfor testing. For this
reason the values obtained using the FWD changed in respect to the force applied and
were much greaterthan that obtained when using the Prima. Theseeffects were taken into
account when making assessmentson slab behaviour, with results from the two devices
examinedseparately.
Edge Cantilever = dI equation4.5

-do
4.5.4 Deflection Basins
Deflection bowls can be used to examine slab curvature and the overall responseto load
(section 3.6.2). The position of the geophonesdiffered betweenthe two sites.At Daventry
deflection basins were obtained for the loaded slab section; however, at Lutterworth the
unloaded slab was monitored. In both situations at least one geophonewas placed onto
the opposing slab to enable load transfer calculations to take place. The remainder were
then placed at increasing radial intervals from the source of the load to enable a full
profile to be measured.
As described in section 3.6.2, when tests are placed at the centre of the slab, deflection
bowls enable the back-calculation of layer properties. As those taken at Daventry and
Lutterworth were taken at the slab edge this was not possible; however, a deflected shape
causedby the imposed load was produced.
100
4.5.5 Void Intercepts
Void intercepts enabled the occurrence of a void and its size at the slab edge to be
estimated.This was then used to evaluate the effects of slab voiding on joint and crack
behaviour. Confirmation could also be made on the amount of voiding suspectedat the
Lutterworth site, providing explanationsfor possiblevariations in the site data collected.
The void intercept approach suggestedby Crovetti and Darter (1985) was used in the
testing program as its effectivenesshad beenverified by other authors.A void intercept is
the position at which the extrapolatedbest fit line of deflections under three increasing
load magnitudescrossesthe zero load axis (Figure 3.16). Values of 25-75 microns have
been proposed as indicating a void, with some authors (Crovetti and Darter 1985,
Frabizzio and Buch 1999) suggestingthe value of the intercept is in direct correlation to
the size of the void. For this researchthe method has been deemed correct, enabling
relationshipsbetweenvoid size and slab response.
4.6 Site Information
Due to confidentiality the commercial namesof the sites cannot be identified; however,
information is provided on the operational requirements of the stabs and their
construction methods. Where possible layouts of the floor have been provided so the
location of testing can be identified.
4.6.1 Daventry
Measurementstaken on the floor slab consisted of surface crack measurement,precise

levelling, single load FWD deflection testing and Prima deflection testing. A plan of the
South West comer of the site is provided in Figure 4.12.
This warehousewas used for the storage and distribution of an assortmentof toys and
packaging materials. The far West and East of the building were predominantly used for
block stacking of items, with the central area containing goods on various racking
devices. The South of the warehouse(running full length) was left relatively clear for the
preparation of items ready for dispatch onto lorries. A mix of both dock levellers and
ramps were used for loading, with pallet trucks and small-wheeled forklifts the main
types of goods transportersusedwithin the warehouse.
101
The floor was constructedin early 1999, with 160mm thick concretecontaining 20kg/rn
of steel fibre. The sectionsof floor between construction joints were split into separate
panels with the use of sawn inducedjoints, 60mm deep. Thesewere approximately 6 by
7.5m in plan, but varied slightly to accommodatethe geometry of the warehouse.Across
constructionjoints 12 by 1500mmdowel bars were usedto enhanceload transfer.
The sawn joints had opened up to different sizes, some of which were less than Imm
wide, with others over l0nun (many of which were tested).Deterioration could be seen
mainly on the joints that had openedup significantly, and this consistedof edge spalling
and comer cracking. High deflections were observed by eye across some joints when
goods transporterswere passing over, indicating very poor load transfer. No mid-aisle
cracking or structural damagewas found throughout the site signifying a generally sound
construction.
4.6.2 Lutterworth
Measurementstaken on the floor slab consisted of surface crack measurement,coring,

variable load FWD deflection testing and Prima deflection testing. A plan of the site is
provided in Figure 4.13.
This warehousecontained racking which was a mixture of both very narrow and wider
aisles interspersed.This took up the majority of the floor area and was tight against the
North-West wall. Loading bays were positioned to the South, with a small storagefacility
to the East next to the office complex. Areas next to loading bays were generally free of
goods, with those near to the offices used as storageareasfor small stockpiles of material.
A mezzanine floor was sited at the South East comer with minor low load works
occurring on this level.
The concrete floor was 190mm.thick with mesh reinforcement throughout the baseof the
slab. Dowelled constructionjoints were placed 18 to 22m apart with sawnjoints inserted
at distances of 5.5 to 8m. The VNA trucks were run on wire guidance systems set
approximately 20nun deep into the slab and filled to level with a rigid resin
Various companieshad undertakensurveys previous to the site visit and some remedial
work had been actioned. This mainly consistedof crack filling with a resin and grinding
to VNA vehicle tracks. Some sections of floor slab had been replaced completely, with
approximately 500min either side of a joint having been recast.Large cracks were clearly
102
visible, the majority of which ran parallel to the aisles for 10's of metres. These were
generally found in the racking areas, with the sin-fflar sized slabs outside of this zone
showing no sign of deterioration.Interestingly, cracks were also seenextending from the
grooves incorporatedfor the wire guidancesystem.Theseran to the edge of the building,
with some of the surface cuts themselvesshowing signs of deterioration and spalling.
Most of the sawnjoints and almost all of the constructionjoints had openedup relatively
little. Previous surveyshad shown that the slab was prone to voiding under the slab edges
causedby compactionof the subbasematerial.
4.63 Ballymena
Measurements taken on the floor slab consisted of builders level curling checks,
profilometer curling checksand Prima deflection testing.
This site was usedfor the bulk storageof tobaccoand its packagingproducts.Thesewere
placed on racking systems nine pallets high, with each holding up to one metric ton.
There were 16 aisleswithin the store, eachhaving a width of approximately 1.5m.To one
side of the warehouse a smaller area has been constructed which seemed to be a
temporary storagefacility. The warehousehas previously been used for the bulk storage
of tobacco and therefore the point loads produced by racking were relatively new in
comparisonto the life-span of the slab.
High lifting trucks were the most predominantly usedpiecesof machinery in the aisles.It
had been found that these were swaying excessivelydue to movement of the slab edges.
The driver had complained that the trucks were occasionally moving very close to the
racking becauseof the lean, and commentedthat he suffered from back pain brought on
by the movement. The trucks themselveswere run on wire guidance systemsset within
the concrete slab. Only three cracks were found in total throughout the slab, all of which
were below I mm in width and showedlittle sign of deterioration.
The slab for this site was approximately 40 years old making its construction date
sometime in the 1960's. Cores had been taken and the slab was found to be non-
reinforced, with a thicknessof 225mm and placed on a weak concrete subbaseof loomm.
The subgradematerial was assumedto be good as the ground surrounding the site was at
a much higher level and had therefore been cut to a reasonable depth. Each aisle
contained 8 joints running North to South with longitudinal joints positioned under the
103
racking systemto reducethe risk of deterioration.The constructionwas representativeof
long strip although no one could guaranteethat this was the case.
The joints themselveswere all of the construction type with no dowels or load transfer
system visible. All were less than 0.2mm in width and appeared to be behaving
reasonablywell with little arris damageor comer cracking. Some voiding was apparent,
detectedby the movement of the joints as vehicles were passingover. Some permanent
curl was felt at the joints, and in some positions grinding had taken place to smooth the
ride for vehicles.
In generalthe slab was in very good condition although the joints were moving enoughto
causesomeproblemswith the high loading vehicles.
4.6.4 Skelmersdale
Measurementstaken on the floor slab consistedof surfacecrack measurementand Prima

deflection testing. A plan of the site is provided in Figure 4.14.
This warehouseconsisted of two separatebuildings which had been joined to form a

larger floor area.The section that had recently been taken over (5 years) was previously
used as a carpet warehousecontaining only minor loading. The walls between the two
buildings had beenremoved at specific areasto form a thoroughfare,and from this it was
clear to see that the two floor levels were different at the time of construction. Racking
had been constructedin the South-West comer of the site, with the North-West section
consisting of an automated transfer system for items of clothing. Pallet trucks and small
reach trucks were the most common means of transporting goods, with forklifts used for
occasionalloading and unloading of vehicles.
A lot of remedial work had taken place in both sections of the warehousefloor. This
consisted of filling any problematic joints with a rubberisedresin, which was left raised
from the surface. Other methods of repair comprised a two part self-levelling screed,
much of which had cracked and de-bonded from the original slab. Where the two floor
levels had varied it appearedthat a section, approximately Ini from the join on either side,
had beencut-out and in-filled with new concrete.This had cracked severaltimes and resin
had been used to try and level the area. Once again this had not been kept flush with the
floor surface.
104
Many, if not all of the joints were of the construction type, with no saw cuts or inducing
materials to be seen.Thesehad openedup a great deal and almost all were showing signs
of spalling and deterioration. Many had not been filled and were full of detritus; others
had foam filler betweenthe concrete faces. In most areasof the old carpet warehousea
covering had beenapplied over the floor. This was a brittle material and thereforecracks
in it were indicating some kind of recent movement. The joint size in the concrete
underneath was extremely difficult to measurebecauseof this surfacematerial.
A section of slab between the racking area and the buffer lanes of the old carpet
warehousewas in very poor condition. Strips of the top screedwere peeling off at a width
of approximately 200mm. After speakingto the survey team it had beenassumedthat this
was the site of a strip foundation where a wall had been standing. Movement was high
across this section of slab and the warehouse operatives encounteredproblems when
trafficking.
Cracks could be clearly seenthroughout the entirety of the slab. Someof thesewere quite
large although it did appear that many were dormant and only a few showed signs of
recent movement. These had been repaired using a resin which had spilled onto the
surface of the slab. In general, the whole floor varied in level and flatness with most of
the joints in the carpet warehouse showing signs of deterioration. Some of the older
cracks around the building were showing signs of edgespalling.
105
Actual surfacc Calculatcd
crack
Saw-cut
Crack
width
Figure 4.1 - Crack width prediction from saw-cut Joints
Figure 4.2 - Crack microscope measuring a saw-cut Joint
Figure 4.3 - Set of callipers measuring a saw-CLItjOiIIt
106
www. cflgrou p"O m -t,
am.
moll- Crack
C*mparalor
oltO
mo
dLo-
--n==k
Figure 4.4 - Set of crack comparators
Estimated Surface
Measurement
r 71
t!
\I I!!
.............
......... ....
Estimated Base
Extrapolation Line
Measurement
Actual Gauge
Measurements
Figure 4.5 - Extrapolation of embeddedstrain gaugesto calculate surfaceand basecrack

widths
107
Actual surface crack
width
10,
Actual crack
Predicted crack profile
profile
Actual base
crack width
Predicted base
crack width
Figure 4.6 - Overestimation of base crack width caused by differential shrinkage
Graduated Wedge BUIlders Level
,31tL? :)UULMlu
Figure 4.7 - Estimation of slab curl using builders level and graduatedwedge
108
-- '- .-
-
-.
Figure 4.8 - Estimation of slab curl using Profilorneter
SECTION Mass Spring
I kuu)
PLAN
ic 2 (dl)
Geoplione 3 (du)
Joint
d*11) SLAB
Fi('Lire 4.9 - FWD and Prima geophone placement I-or measuremeiit ot'slah behaviour
across cracks and joints
109
Geophone Locations
Geophones
Geophone Locations
200/350/650/1050/1250/1850 mm
Joint Location
No.
50mia 10
io ...
(jeophoncs
Figure 4.10 - FWD geoplione locations for Daventry (Top) and Lutterworth (Bottorn)
100
90
so
70
60
10
m 50
0
-i
CY
40
0-
30
92
20
10
00 10 20 30 40 50 60 70 80 90 100
Direction 1 Load Transfer(%)
Figure 4.11 - Load transfer comparison when loaded each side of a joint
110
-- rw"d mstrained "KMMOR Joint
-- ---- -- Sawn amined movemot ju; ul
......... . ....... Sawn Fret lunwomt joint
TES 10Smin jaup

All ims1run"Atlog wvs
meoW derAlsin the 11sh TFAUMStral"goup
OZA IOUcr 11,1700
Ali duraulano in mm
D
111200
fT
7720
01
14r,
6700 dO 35 34 31 5I is
f1i H 04' 1 ,
7700
6700
(463
$400
Nt
Figure 4.12 - Plan of S.W. comer of Daventry site
Figure 4.13 - Plan of Lutterworth site
III
NI
Figure 4.14 - Plan of Skelmersdalesite
112
5. JOINT DETERIORATION
5.1 Introduction
This Chapter describes the development of a small-scale testing facility used for the
examination of load transfer deterioration causedby cyclic loading. The test method had
the following requirements:
Provide accuratemeasurementof the deteriorationprocess

Enable simulation of site conditions
Facilitate variation in crack geometry
Producehigh cycle numberswhilst being time efficient
Be cost effective
Utilise existing equipment (where practical)
Each area of the testing methodology was carefully considered to provide close
comparison with in-service site conditions for internal floor slabs. This included crack
formation (both initiation and geometry),contact stressesacrossthe crack face, number of
cycles of loading, subgrade support conditions, and rate of loading. The study
investigated high cycle loading on cracks and joints, examining the load transfer effect
and the resistancemechanisms.The results provided significant insights into load transfer
across cracks/joints under cyclic loading, which could be used in the long-term
assessment of concrete slabs on grade. Comparisons could also be made with the field
testing (describedin Chapter 4) to assessthe effect of age and deterioration of the crack,
and its associatedslab responseeffect. This also provided information required for the
load transfer elementswithin the finite element model, describedin Chapter7. During the
experimental program a range of parameters,including: crack geometry, reinforcement

type and quantity, and load magnitude were examined to assesstheir effect on the crack
deterioration process(section 5.2).
The method of testing utilised a small-scale(10NION400mm) double 'V' crackedbeam

in
specimen, which the two edge sectionswere clamped against a reaction frame (Figures
5.1 and 5-2). The central block was attached to a force applicator which cyclically
imparted load and created a shear stress across the crack plane. Vertical displacement
113
measurementswere then taken periodically from both cracks,enabling deteriorationto be
monitored over a controlled number of cycles. The reasoningbehind the developmentof
this methodand the proceduresusedare discussedin detail in the following sections.
5.2 Test Variables
A variety of crack orientations and reinforcementtypes/quantitieswere examinedduring

testing to establishtheir effect on the crack deterioration process.To enablecomparisons
betweenspecimensa referencewas usedwhich comprisedof a beamconstructedwith the
standard concrete mix described in section 5.4.2, with the incorporation of 30kg of
Dramixo RC-65/60-BN steel fibre. This was chosen as it representeda typical slab
construction, with a mid-range reinforcement quantity. This section describes the

reasoning behind each test parameter, with Table 5.1 providing a summary of all
specimensexaminedduring the test period.
5.2.1 Surface Crack Widths
For each variation in beam type, the crack width was altered between0.66,1.98,3.3 and
4.62mm at the surface, reducing to zero at the base.This provided reasonablevariations
betweeneach width but also enabledthe trends from crack opening to be identified. The
IV' shaped geometry was typical of that found from the Lutterworth cores, and the
embeddedstrain gaugesat Leeds, with the surface widths representativeof the range of
field data collected. The exact value of the initial surface crack width was controlled by
the depth of the shims used, and therefore simple combinations were chosen to make
setting-up of the test easier. On a few samples the crack width was set to 3.96mm or
5.94mm using an intermediate shim size if useful data would be obtained by doing so.
occasionally, testing of the 0.66 and 4.62mm crack widths did not take place as the
results would have produced small displacementsor early failure of the specimen.
5.2.2 Steel Fibre Types
Three Drarnixo hook ended fibres of type RC-65/60-BN, RC-80160-BN and RL-45/50-
BN, supplied by Bekaert Building Products Ltd. were tested during the research.
Consultation with the manufactures identified these to be the most common sizes in
internal floor slabs on grade, with the RC-65/60-BN being used most often. More
infon-nation on the steel fibre types used in the research is provided in Table 5.3 and
section 5.4-3.
114
5.23 SteelFibre Quantities
Three steel fibre quantities of 20,30 and 40kg/n? were used during the experimentation
since theseare statedin the literature as being typical values for industrial floors (ACEFC
and The Concrete Society 1998, The Concrete Society 2003). For the standard beam
30kg/m3was used,it being the intermediatevalue and that found regularly in many of the
mix designsobtainedfrom site.
5.2.4 Non Reinforced
Non reinforced specimenswere testedto examine the effect aggregateinterlock alone has
on the deterioration process. This enabled direct comparison of the V shaped crack
orientation usedin this researchagainstparallel cracks testedelsewhere.
5.2.5 Mortar
Mortar beamscontaining 30kg/mof the RC-65/60-BN fibre were examinedto determine

the degree of load transfer obtained from the fibre alone, without influence from
aggregateinterlock.
5.2.6 Parallel Cracks
Parallel cracks varying between 0.5 and 2-Ommwidth were tested for each of the three
fibre quantities. The results provided comparison with the findings of other researchers
who examined parallel crack aggregateinterlock effects. This also enabled differences
betweenV shapedand parallel crack orientation to be identified.
5.2.7 Fabric
Representativesections of A142 mesh (Figure 5.3) were cast into beams and tested at
1.98 to 4.62mm crack widths. This enabled comparison of reinforcement type and gave
some indication of the steel fabric load transfer effect.
5.2.8 Loads
Load magnitudes of +/-2kN and +/-6kN were tested in addition to the standard+/-4kN
value to determine their effect on the rate of deterioration. The reasonbehind the selection
of these values is described in section 5.3.5. For each load the standard30kg/m3of RC-
65/60-BN fibre beam was usedover the full range of crack widths.
115
5.2.9 Summarv of Test Variables
Table 5.1 below shows a summary of specimens examined during the testing period.
Table 5.1 - Testing Schedule
Reinforcement Type Surface Crack Width (mm)

0.66 1.98 3.3 4.62 Other
None
30kg /m'Fibre (Mortar)
20kg/in 3
. Fit
Fibre 3.96nun
30kg IM3 Fibre V' 5.94mni
3
, n ir Fii 1)rre
40k gj/ 5.94mm
30kg ,. RC-80/60-BN
, _. /m3 , Fibre V,
30RC-45
30kg
k/n
g5 /in -BN Fibre
RC-45/50-BN
Re-bar
Fabric
I(F)kag,
hywJFibre (2kN Load)
((6kN
IOZg- hur Fibre Load)
Fibre (Parallel Crack)
220Tkg/ni
Fibre (Parallel Cr, ick)
3f0k-g-ghn
L:
Fibre :C:
4OCg7n_f (PParallel r::a:c: k) :: =
Unless otherwise stated beams contain ribre type RC-65/60-IIN at 30kg/ni3 with 'V'
shaped cracking
5.3 Load Test Development
5.3.1 Simulation Method
Many of the previous test methods examining small-scale monotonic or cyclic load
transfer have employed a singularly cracked specimen. Valle and Buyukozturk (1993),
Van de Loock (1987) and others, used a rectangular section with a shear plane induced
through the centre (Figure 5.4). This arrangement is only of use for single or low
repetitions of load due to the method of horizontal restraint. Millard and Johnson ( 1984)
used a similar set-up with a smaller specimen, although once again this was only
to
applicable single cycle loading (Figure 5.5). Each of these tests required displacement
across the shear plane, whilst preventing movement in the normal direction. This was
partially prevented with the use of slip plates and compression jacks-, however,
measurementstaken dLII-IIIgteStIIIg Showed signs of crack widening caUsedby stress

development as aggregateparticles are forced over each other. The method of normal
116
restraint also createsrotation due to the eccentric loading, which is difficult to prevent in a
singularly crackedspecimen.
In a pavementor floor slab the frictional restraint of the concreteon the subbase,and the
confinement of adjacent slabs, prevents all horizontal displacement.The joint or crack is
therefore fully restricted from movement in any direction other than vertically. This is
very difficult to simulate in a small-scale,one crack specimendue to the poor control of
normal restraint. Colley and Humphrey (1967) and Raja and Snyder (1991) used large
scale testing to overcomethis problem, closely representingreal conditions (Figure 5.6).

This arrangementtypically consistsof two large (1220 x 3000mm) slabs resting on a real
or simulated foundation, with a pulsating load applied alternately to each side of the joint.
Due to the available test equipmentand number of specimensthat required testing, it was
not appropriate to utilise large-scale testing and therefore a small-scale alternative was
employed. This had to allow for high cycling whilst preventing the overall rotation and
normal displacement which had been found with the smaller methods of Millard and
Johnson (1984) and Valle and Buyukozturk (1993). Some of these problems can be
overcome with a double cracked shear specimen as used by Thompson (2001), shown in
Figure 5.7. This method required a controlled crack to be induced either side of the load
application position. The central section was then fixed to the loading ram with the end
segments rigidly clamped against a stable surface.This caused a load differential between
each side of the crack creating a double shear stress,and a reduced risk of rotation. If
symmetrical degradationacross both faces is the

assumed, use of two crack planes is of
and
consequence is equivalent to a singularly crackedspecimen.
no
For the formation of a simple set-up with a fully restrainedcrack face, an arrangement
to that used by Thompson (2001) was used, details of which are provided in the
similar
of this chapter. The set-up enabled utilisation of an existing (Dartec) cyclic
remainder
loading test machine,which could apply the required load magnitudeand cycle rate.
To simulate the type of loading found, and the degradationmechanismswithin, a crack or

joint in a concrete slab on grade, load was applied in both a positive (downward) and
negative (upward) direction. Figure 5.8 demonstrates how each laboratory simulation
position related to that occurring within a site. Position I is an unloaded case whereby
there is no movement. When the central block is moved in a positive direction (position
2), the left hand side of the specimenrepresentsa slab load on the leave side, whereasthe
specimen right hand side representsa slab load on the approachside. As the central load
117
moves in a negativedirection (position 3) the left hand side of the specimenrepresentsa
load on the approachside, with the right hand side representinga load on the leave side.
This approachensuredthat the left and right hand sides of the specimensimulated a load
crossing the joint in either direction and was therefore fully characteristicof the contact
stressesacting acrossa typical joint.
53.2 Specimen Geometry
A specimen was required which would enable a number of material parametersto be
examined, whilst being both time and cost efficient. This preventedthe use of a full size
slab, which could have replicated exact site conditions. A small-scale section was
therefore selected,which had to provide a characteristicvalue, representativeof the entire
slab. Although small attributes of the crack face would have a much larger influence on a
small specimen,careful selection of its size and thorough examination of the production
processsufficiently reducedany discrepancy.A depth of IOOMMwas used as this was a
lower bound for a 150mm slab with 1/3 saw cut. The width had to representa typical
section of the slab incorporating a sufficient blend of concreteconstituentmaterials (most

importantly coarseaggregateand steel fibre). As the aggregatewas below 20mm, with a
maximum fibre length of 60mm, a value of 100min was chosen as being fully
representativeand having limited scale effects. The 400mm length of the beam enabled
the central 100mm section to be fixed to the load applicator, whilst providing sufficient
end strappingto the reaction frame.
533 Crack Geometry
The literature reviewed in section 3.3, and the measurementtechniques used in section
6.1, suggestedthat the majority of joints or cracks are of aV geometry, causedby the
differential shrinkage and curling commonly found in concrete slabs on grade. The
majority of the previous tests carried out on shear transfer across joints (Colley and
Humphrey 1967, Millard and Johnson 1984) assumeda parallel width over depth and are
therefore inconsistentwith in-service slab conditions. White and Holley (1972) examined
a small number of V shaped cracks and compared the results to those of parallel
cracking. They found that the average width of the V shapedcrack had a greater load
transfer capacity than that of a similar sized parallel crack, highlighting the importance of
using correct geometry when testing. Work undertakenby Colley and Humphrey (1967)
compared laboratory load transfers to those obtained from site and found that the site
values were much larger. The parallel cracks assumedin their laboratory testing could be
118
a reasonfor their findings, as 'V' shapedcracking (probablein their field slabs)is known
to provide greaterload transfer.
To ensure that the testing procedure and specimen preparation used in this research
provided good representationof site, the majority of specimenswere pre-crackedand set-

up to obtain 'V' shapedgeometries.The size of the surfacewidths was selectedfrom the
data collected in section 6.1, with the exact values used, and the reasoning, given in
section 5.2.1. A selection of parallel cracks was also tested to examine the effect of the
crack angle, and identify any variations in behaviour.
5.3.4 Subgrade Support
Many of the tests undertakenby previous authors (Buch et aL 2000, Raja and Snyder
1991) incorporate well supportedslab edges,utilising either a foundation made from soil
compactedin a test box, or 'elastic' materials such as neoprenepads.The incorporation of

these materials partially dictates the amount of displacementthat can take place, and
therefore controls the rate of deterioration. If the soil in the field is different to that usedin
the experimentationthen any comparison will contain errors. Section 2.4.2 demonstrated
that in most situationsthe slab edgeswill have curled to somedegree,thereby leaving the
slab unsupported,with the load transfer system alone contributing to joint efficiency up to
a certain load limit. This createshigher contact stresson the crack faces resulting in an
enhancedrisk of deteriorationfailure. In the test methoddeveloped,foundation or support
materials were excludedfor more accuratesimulation of site conditions and this therefore
provided an elementtest. This approachproduceda worst-casescenario,which could then
be used in comparisonsof site data.
53.5 Load Magnitude
The magnitude of loading has been shown to highly influence the rate of concretecrack
degradation (Colley and Humphrey 1967). A suitable value was therefore essential to
representthat occurring within the field. The majority of previous tests on full-scale stabs
have used a 40 to 50kN load over a cross sectional area of around 0.2m2, generating a
contact stress in the region of 200 to 250kPa. The loading found on internal floor slabs
varies greatly but example values given in Table 2.1 are between 42 and 60kN. In
external slabs this can regularly exceed lOOkN depending on the type of vehicle used,
with pavementsdesignedon an 8OkN standardaxle, thereby creating 40kN per wheel.
119
Yoder (1959) and Friberg (1938) proposedthat for dowel bar calculationsonly a distance
1.8 times the radius of relative stiffness T (equation 3.1) from the load source has any
influence on transfer efficiency, the effect reducing linearly with distance. From this
information the Concrete Society (2003) concluded that full load transfer could be
assumedat a distance0.91either side of the load. Calculating T using typical values of:
Youngs modulus of concrete(E,) = 30GPa, Slab depth minus saw-cut (h) = 150/300mm,
Poissons ratio (v) = 0.15 and modulus of subgrade reaction (k) = 0.07/0.01 N/mm3,
producesresults in the range of 0.6 to 1.6n-4which when doubled (incorporating the effect
either side of the load, along the slab edge) gives full load transfer lengths between 1.1
and 2.9m. If the averageof these values is used to simulate a typical situation (i. e. 2m),
with an effective slab depth (minus saw-cut) equal to 0.1m, and an application load of
5OkN,equation5.1 generatesa contact stress(r) of 250kPa.
To produce a similar 250kPa contact stress in the small-scale specimens,where beam
width (x) and depth (d) are equal to 0.09m (doubled as there are two crack faces), an
applied load (P) of +/- 4kN was required (equation 5.2). When smaller and larger loads
with equivalent full-scale magnitudesof 25 and 75kN were considered,loads of +/-2 and
+/- RN were required.
r= P1 (21- d) equation5.1
P= 2r (x - d) equation5.2
53.6 Loading repetition
The crack or joint of a slab may be subjectedto hundredsof load repetitions every day.
This can result in many millions of cycles throughout its expected lifespan. Given the
research period it was deemed impractical for testing to continue for the equivalent
number of cycles and therefore a value was adoptedwhich enabled a comparison between
laboratory and field data. Colley and Humphrey (1967) tested large-scaleslabs for up to
one million cycles and concluded that 90% of the degradationwill have occurred within
the first cycles, as shown in Figure 3.8. Abdel-Maksoud (2000) conducted
-500,000
similar tests on smaller samplesat cycle numbers up to 300,000 cycles, at which point the
increasein degradationappearedto have ceased(Figure 5.9). Thompson (2001) examined
cement bound materials and stopped testing after 10,000 cycles as the gradient of shear
slip displacementhad reduced dramatically (Figure 5.10).
120
Due to the range of load cycles used in the tests reviewed, 15 trial tests were conducted
with the selectedtest methodto ascertainthe most appropriatenumber for this research.A
minimum of 250,000 cycles was chosen initially, with a further 250,000 applied on 5
to
specimens examine longer-term degradation. At least 75% of the 500,000 th cycle
deterioration occurred within the first 250,000 cycles (Figure 5.11), and therefore this
level of repetition was selectedas an appropriaterepresentationof total degradation.
In the majority of the experimentsthe displacement/cyclegradient was found to reduceto
negligible values towards the end of the test. However, in a few samples appreciable
levels of deterioration were still visible as indicated by the steadyincreasein gradient. In
these circumstances the test was continued until a horizontal gradient was found
(normally shortly after), although the 250,000 cycle data was still used for comparison
during analysis.
5.3.7 Load rate
Colley and Humphrey (1967) measuredthe loading pattern of a joint as a vehicle travels
over at approximately 30mph (Figure 5.12). This required the approachslab to be loaded
from zero to maximum in 0.25 seconds, and then instantaneously removed. On
completion the leave slab was immediately loaded and gradually reducedto zero in 0.25
seconds,resulting in a total cycle length of 0.5 seconds.Due to the nature of the small-
scale tests, and the limitations of the Dartec testing machine, it was not possible to fully
replicate the approach used by Colley and Humphrey (1967); however, a sinusoidal curve
which completesa full cycle in 0.5 secondswas employed (Figure 5.13). Abdel-Maksoud
(2000) concludedfrom his work on cyclic loading that the load rate has very little impact
on degradation results unless weak aggregate is used, and therefore this approach to
loading was acceptable.
5.4 Specimen Production
5.4.1 Introduction
beamsfor use within the cyclic

Each batch of specimenscomprised six IOOxIOOx4OOmm
cubescastfor 28 day compression

load test machineand two IOOxIOOxIOOmm testing.
The beamsand the cubesutilised standardsteelmoulds;however,a IOOxIOOxlOOmm
wooden block was inserted into the end of the beam mould to reduce the length from 500
to 400mm in accordancewith the clear accesslimitations of the Dartec. The use of six
121
beamsensuredthat at least two parameters(comprising of 2-3 specimenseach) could be
testedfrom the samebatch, therebyreducing any errors causedby constructionvariations.
5.4.2 Mix Design
The mix design was selectedfrom typical information gatheredon floor slabscast within
the last six years (Table 5.2a). At each site the concretewas C40 in specification, with a
125mm.slump, enabling site pumping. The maximum aggregatesize was 20mm and the
minimum cementcontent 325 kg/m3,with a water to cementratio below 0.55. In all cases
the mixes were very similar, and as such the most representativespecification was chosen
(Table 5.2b). The constituentmaterial information usedwithin the concreteis provided in
Table 5.2c.
For each concrete batch, the moisture content of the coarse aggregatewas calculated
using a speedytest; this varied between 0 and 3%. The volume of water and aggregate
was then altered accordingly to produce the correct free water ratio. The moisture content
of the fine aggregatewas selectedfrom previous testing (Jones 1998) when used under
similar conditions, and was found to be approximately 0.64%. This was assumedfor all
batchesas the variation in water content would only be minor, with the mix specification
altered accordingly.
Table 5.2a - Typical Site Mixes
Material Quantity
Northampton Normanton Bedford
Cement 370kg/m3 360kg/m3 370kg/m3
(5-10mm)
Coarseaggregate 355kg/m3 352kg/m3 316kg/m3
CoarseAggregate(10-20mm) 711kg/m3 704kg/m3 736kg/m3
FineAggregate 783kg/m3 799kg/m3 741kg/m3
FreeWater 174kg/m3 181kg/m3 196kg/m3
Table 5.2b - Actual Concrete Mix Specification Used for Laboratory Testing
Material Quantity
Cement 370kg/m3
(5-10mm)
Coarseaggregate 355kg/m3
CoarseAggegate(10-20mm) 711kg/m3
[Fine Aggregate 783 kg/m3
I FreeWater 185kg/m3
122
Table 5.2c - Material Information
Material Specification
Cement Type 1 Ordinary Portlandcementto BS 12 (1991)
Fine Aggregate Zone 2 river sandwith a maximum size of 5mm
CoarseAggregate Trent River Gravel with 5-10mm.and 10-20mmgradings
water Tap Water held at laboratory temperatures
Plasticisor Sikament Ultra
To ensurethe characteristicconcretestrength(f,) was achieveda target meanstrength(Q
was calculated, which all specimenshad to exceed. For the purpose of testing a 5%
defectives value was chosen (k = 1.64), along with a standard deviation of 3N/mrn,
allowable becauseof the controlled conditions found within the laboratory. Equation 5.3
therefore producesa target mean strength of 45MPa. The results of the concretemixes
used within during the researchare provided in section6.3.1.
fm =f, + ks equation5.3
When mortar was required, quantities of coarse aggregatewere replaced with an equal
weight of fines. This made the mix very stiff and a Plasticisor was introduced at a dosage
of 1 litre per 50kg of cementicious material (as instructed by the material suppliers) to
increaseworkability. The target mean strength was calculated in a similar manner to the
concrete,although it was not required to conform to the 45MPa value, as the reduction in
aggregatesize was known to produce a significant loss in strength.
5.43 Reinforcement
Three fibres types were used within the test program, all of which were supplied by
Bekaert Building ProductsLtd. and consistedof Dramixo' categoriesRC-65/60-BN, RC-
80/60-BN and RL-45/50-BN (Figure 5.14). Details of each fibre are shown in Table 5.3
with further information supplied in appendixA.
123
Table 5.3 - Steel Fibre Data
RL-45/50-BN RC-65/60-BN RC-80/60-BN

Class Standard Premium Max Performance
Type Hook Ended Hook Ended Hook Ended
Length 50mm 60mm 60mm
Diameter 1.05nim 0.90mm 0.75mm
Aspect ratio 48 67 80
Fibres/kg 2800 3200 4600
Tensile strength 1000N/mm2 1000N/mm2 1050N/mm2
Packaged Loose Collated Collated
Steel fabric sections were cut from a standard sheet of A142 structural mesh, with
transversememberslocated 100mmeither side of the saw cut (Figure 5.3). The fabric was
placed 20mm from the base of the mould and was kept in position with four triangular
steelrestraints.
Regular 7mm.grade 460 steel reinforcement was fixed into an equivalent mesh to that in
Figure 5.3. Again this was placed in position using the method describedabove.
5.4.4 Casting
The concrete was mixed in a 100 litre drum for I minute before the fibres were
introduced, and then continued for a further two minutes allowing the fibres to fully
separate and disperse evenly. Once complete the beam moulds were half filled with
concrete and placed on a vibration table for 30 seconds.A second layer was then added
and vibrated again for a period of I minute or until the release of air bubbles onto the
surface had ceased. The cubes were cast in a similar manner and in accordancewith
BS1881, Part 108 (1993). Each specimen was then trowelled level and placed under a
polythene membrane.
After twenty-four hours the moulds were removed from the concrete. The beams were
then labelled, sawn, cracked (as describedin section 5.5.2) and placed with the cubesin a
water curing tank at 20*C +/-2 'C for a minimum of 28 days. After this time period the
cubes were removed and tested in a Denison Material Testing System 50OOkN
compressionmachine to BS 1881: Part 116 (1983). This ensuredthe concretewas of the

correct strength and that problems developedduring casting or curing could be identified.
Throughout the testing period beams in the same set were tested within 14 days of each
other to reduce any increasein strength over time having an impact on the results.Where
124
possible they were testedjust after 28 days, with a maximum allowable time limit of 56
days.
5.5 Beam Preparation
5.5.1 Crack Timing
Before the beam could be tested in the cyclic loading rig it had to be pre-cracked.Work
carried out by Abdel-Maksoud et aL (1997) concluded that the formation of a crack is

highly dependentupon the time at which it is instigated.When formed early in the life of
the concretei. e. within 48 hours, it is the bond betweenaggregateand cementpastewhich
breaks down. After this period the bond has had time to strengthen and an increased
percentageof cracking will occur through the aggregate.This changein crack type causes
large variation in the roughness of the face and therefore affects aggregateinterlock
(Nowlen 1968). The Concrete Society (2003) has determined that cracking in concrete
floor joints generally occurs in the first 24 to 48 hours of slab life when the concretehas
limited strength but is subjected to high tensile stresses.This leads to matrix cracking
producing the more roughened surface describedabove. To replicate the real situation it
was decided that the beam should also be pre-crackedat this early time period. As the
width required would not be known at suchan early stageeach beamwas initially cracked
to the smallest 0.66mm surface width. This enabled the final measurementto be
determinedlater on in the testing schedule,but ensuredthe profile would follow that of an
early-agecrack.
5.5.2 Crack Technique
The method of crack formation has been found to influence the roughnessof the crack
face (Abdel-Maksoud et al. 1997). Abdel-Maksoud (2000) criticised the techniquesused
by other researchersfor not providing suitable simulation to that found on site. He
suggestedthat a three point bending technique would produce a much smoother face
compared to a true tensile crack. This was caused by significant aggregatebreakage

resulting in a decreasein sheartransfer available through interlock. Abdel-Maksoud et al.
(1997) proposedthat when creating a crack, the width at the surfacecould be in the region
of 2mm, although 30% of the interior may remain uncracked, and therefore a more
thorough method of crack detection should be used. They also stated that the method of
tensile force application affected the roughness,as clamping and then pulling one side
only gave a different crack shape to that obtained when pulling from both sides. The
assumption made by both of these authors is that crack development in slabson grade is
125
instigatedfrom a true tensile stresscausedby shrinkage.Walker and Holland (1999) have
statedthat cracking is often createdby the curling of the concreteslab rather than direct
shrinkage,resulting in a mixture of both tension and flexural forces.
From the monitoring of cracks discussedin section 6.1 and literature reviewed in section
3.3, it has been shown that the geometry of a crack or joint in concreteslabsis generally
V shaped.A method was required to enable this crack orientation to be produced in
small-scalebeams.The surfacemeasurementwould need to be controlled, with the base

of the specimen remaining closed. Thompson (2001) developed a simple three point
bending crack induction method proven to producea vertical crack continuousacrossthe
specimen. This involved incremental loading at the crack location until the equipment
control software detected a reduction, at which point loading ceasedand a crack was
assumed.The beam was then rotated 180 degreesand reloaded enabling a parallel crack
to develop. Examination of different methods of crack induction (Abdel-Maksoud et aL
1997,Millard and Johnson 1984) resulted in the method suggestedby Thompson(2001)
being used as it could easily be adaptedto produce aV shapedcrack through loading of
one side only.
The technique developedrequired sawing to a depth of approximately 5mm around the

circumference of the beam to enable the crack positions to be set. The beam was then
placed on a system of steel blocks and shims set 90mm either side of the saw cut. A
similar sized block, but with no shim, was placed directly underneaththe saw cut and a
round steel bar placed on top. Load was then applied at a controlled rate creating a
flexural crack increasing in size until it hit the block underneath,at which point all load
was removed.For the secondsaw cut position the beamwas repositionedand crackedin a
similar manner.Details of the processare shown in Figure 5.15, with the level of control
achievedgiven in section 5.5.3.
The set-up for parallel crack developmentwas similar to that mentioned previously. The
initial crack was extended with the use of shims on the outer steel supports.When the
cracks were of the required width the beam was rotated 180 degrees and the load
reapplied, forcing the previously un-cracked side to open up. Finally, the beam was
placed on a flat bed and load imparted to bring it back into the level position. When the
beam was located into the cyclic load test rig, wedgeswere positioned into the crack faces
to force it back into its fully openedstate.The beamwas then clamped to the supportsand
the wedgesreleased.However, problems occurred regarding accuratecontrol of the crack
widths, and they were thus re-measuredonce fixed into the rig. If the two cracks were
126
identical in size the test was continued, with the measuredvalues being adoptedduring
further analysis.
5.53 Crack Width Control
A series of trial tests were undertakento examine the applicability and accuracyof the
cracking techniqueemployed. A predicted opening was calculatedfor severalshim sizes

using standardgeometry, as shown in Figure 5.16. Demec pips were then placed across
each of the notches(10mm from the top and bottom of the specimen)and strain readings
taken before and after crack formation. Extrapolation of these measurementsenabled
surfaceand baseopeningsto be calculated.
Results obtained from the laboratory tests produced lower surface crack widths than
predicted, as shown in Figure 5.17. The experimentalresults indicate that the crack width
obtained is 1.73 times larger than the shim size used (with a coefficient of determination
of 0.92). This is much larger than the value of 2.22 times the shim size obtained from
predictions. On inspection the crack was found to close slightly once loading ceaseddue
to the resistanceof the reinforcement.To avoid this affecting the results in the cyclic load
testing, the pre-cracking width was set 1 shim size (0.66mm surfacecrack width) smaller
than required. This allowed the final crack size to be formed when clamping the beam in
the cyclic load test rig, ensuring that closure and loosenessof fibres or reinforcementdid
not influence the results.
Testing was also undertakento confirm that the cracks createdin the rig were linear. To
achieve this demec pips were placed at 10mm intervals across both sides of the two
cracks and measurementstaken both pre and post loading. An almost perfectly linear
relationship was found as shown in Figure 5.18, and this was therefore approved as a
suitablemethodfor beamcracking.
5.6 Test procedure
5.6.1 Overview
During the cyclic load testing deterioration of the load transfer interface was monitored
over time. The beam specimen was 400xlOOxlOOnlm in size with the central 100mm
section sawn circurnferentially to a depth of approximately 5mm. The beam was then
crackedat both the saw cuts as describedin section 5.5.2 and placed in the testing rig.
127
The testing rig was an adaptationof the Dartec cyclic load test machine (Figures 5.1 and
5.2). The tension/compressionload cell was relocated to the upper cross-headto enable
both positive and negative loading to be applied. Strapswere constructedto fix the edges
of the beamto the frame, and hold the central block to the load source.They also allowed
360 degreerotation of the loading plate enabling precisecontrol of its orientation.
Calibration of the Dartec test machine was undertakenby an external organisationtwice

during the testing period. However, the load levels applied to the specimen were
monitored every 600 cycles during all teststo ensureerrors did not occur.
5.6.2 Beam Orientation
During trial testing it becameclear that the beam orientation had to be carefully assessed.
To obtain good control of crack width during preparation,the beam was required to have
the trowelled surface uppermost providing a smooth face against the shims. After
formation, the beam required a 180-degree rotation to produce the correct crack
orientation within the cyclic load test rig. This resulted in the roughenedtrowelled face
being placed directly on the supports,thereby reducing the accuracy of the final surface
crack width.
When using this approachthe specimenwas reversedto the orientation of casting when
placed in the cyclic loading test rig. According to Abdel-Maksoud (2000) this can lead to
errors when comparing againstreal slab conditions due to the movementof mix materials
when under vibration. Large aggregate tends to sink, being replaced by mortar, with
fibres becoming orientated in one direction due to the consolidation process. Both of
these actions could affect the characteristicsof the shear surface and thereby the load
transfer efficiency.
To overcome these concernsthe beam was positioned trowelled surface upwards within
the cyclic load test rig. This reduced the accuracy of the initial crack caused by
undulations on the surface. However, since the pre-crackinigwidth would be one shim
smaller than that required in the rig (section 5.5.3), the error in the resultant test would be
negated.
5.6.3 Rig Conriguration
The test rig consisted of two steel side blocks on which the relevant shim combinations
were placed, with the beam edgespositioned on top. All were rigidly clamped to the test
128
frame, forcing the beam into an angled position and opening the cracks to the required
size and geometry (Figures 5.1 and 5.2). The central section was strappedto the cross-
head via a 50kN load cell, which was linked directly back into the Dartec test machine
allowing load magnitude to be monitored and adjusted as required. Linear variable

differential transducers(LVDT's) with a maximum travel of 4MM, repeatability of <0.15
ttrn and sensitivity of 133-33mVN/mm. +/-0.5%, were fixed to the face of the concrete
specimenenabling the measurementof displacements.These were calibrated three times
during the test by the laboratory technicians and on all occasions were found to be
acceptable.The LVDT was glued on the outer section of the beam via an aluminium
holding bracket, with the target sited on the central section, free to deflect in both the
positive and negative directions. This method of LVDT placementwas chosento prevent
agitation of the reaction frame being included in any measurementstaken. Initially, the
glue to attach the LVDT's and targets to the concrete face was a two-part epoxy resin.
Due to the relatively slow nature of the setting process it was difficult to position the
targets accurately as slippage regularly occurred. Bostik superglue (a rapid setting
adhesive)was therefore used which enabled instant fixing and accurateplacement.This
allowed the target to snap off if excessivedeflection was forced on the LVDT, providing
increasedprotection againstinstrumentdamageupon beam failure.
Prior to testing, the LVDT was adjustedagainstthe target to producea near zero reading.
This enabled equal movement in the positive and negative directions, with the actual
value recorded to enable taring at a later stage.On application of load the central block
moved in relation to the end sections,which in turn causedan elongation or compression
of the vertical LVDT- This generated a change in voltage which was amplified, and
recorded temporarily on a Campbell Scientific CR10X data logger. The information was
passed periodically into a standardlaptop computer in the form of a text file, where the
data was further analysed. This process is shown in diagrammatic form in Figure 5.19
with further information on the data logging procedureprovided in section5.7.
5.6.4 LVDT Positioning
During the early tests,eachcrackhada verticalandhorizontalLVDT placedon the front

andrear face of the specimen, as shownin Figure5.20. This enabledany openingof the
crack to be measured alongside the vertical movement caused by application of load.
After severaltrial tests the horizontal movement was found to be extremely small (below
0.003mm) with difficulties in determining whether this was caused by crack opening,
rotation or misalignment of the target. Due to the test configuration it was unlikely that
129
significant horizontal movementcould be accommodatedas clamping of the beam on an
angle prevented either side from moving apart. The position of LVDTs was therefore
changedto enableincreasedvertical measurementsto be taken.
The secondset-uprequired vertically placed LVDTs on the front face of the specimenfor
both cracks as shown in Figure 5.21. On the rear of the specimena single LVDT was
fixed in the vertical direction, with another in the horizontal direction to retain a check
against crack opening. This set-up provided increased vertical movement data and
enabled the calculation of rotation in the central section. To obtain measurementsin the
plane of the specimena moveable strain gauge was placed periodically on the face and
rear of the beams,with values being manually monitored.
5.6.5 Safety Precautions
As a safety precaution,steelblocks were placed under the lower central section of the test
specimen(Figures 5.1 and 5.2). A 5nun gap was retained to allow for vertical movement,
but should significant displacementsoccur, damage to the LVDTs and the test facility
was prevented.A trip function was also incorporated into the control settingsto shut the
machine off if high fluctuations in load magnitude were identified (signifying machine
error or specimen failure). This allowed the test to be run for its full duration, and
prevented the fluctuations in displacement which had been identified when using a
stop/startloading regime. Thesehad been causedby a drop in pressurewhen the machine
was turned off, forcing the central block back into its original position (Figure 5.22).
5.7 Data Logging
The recording of data obtained from the cyclic loading tests was carried out using a
Campbell Scientific CRIOX data logger and the PC208 computer software. A program
was written which enabled the logger to take a series of readings at its maximum rate.
Due to the large amount of data collected only periodic collection at 5 minute intervals
(600 cycles) was possible. The burst option was set to continually monitor for 0.5
seconds,enabling one complete 211zload cycle to be recorded. During this period fifty
LVDT and load readings were taken, resulting in a measurementat approximately every
0.01 seconds.Unfortunately, the exact period betweeneach measurementwithin the burst
cycle could not be establishedas the logger only recorded a time for the first data point.
130
Each measurementtaken with an LVDT and recorded by the data logger was calculated
into a displacement using the relevant coefficients. The original LVDT reading taken
prior to test start-up was then subtractedto provide a measureof movement from the
original location. The maximum and minimum displacementswere then determinedfrom
each burst cycle (Figure 5.23), and plotted against cycle number to show the effect of
crack degradation (Figure 5.24). The total variation between each maximum and
minimum point for an individual cycle was also calculatedto achievea total differential,
which was then plotted againstcycle number to give an overall indication of displacement
change (Figure 5.25). Where required, individual plots showing an individual cycle were
examinedto determinethe sourceof load resistance.
Visual examinationswere made of the specimensperiodically throughout each test. This

involved assessingthe amount, size and type of any ejectedmaterial causedby crack face
deterioration. In fibre-reinforced specimensthe behaviour of the fibres could also be
identified at the outer sectionsto examinetheir failure modes.Once the test was complete
the specimenwas broken open acrossthe crack face (if reinforcementwas still holding it
together) and any significant effects were identified. This could include excessiveface
cracking, aggregatelooseness
or reinforcement bond.
131
Load Cycle
at 2 Hz
Cracked Section
Dartec Testing Machine
Figure 5.1 - Cyclic loading test set-up (Schematic)
Figure 5.2 - Cyclic loading test set-up (Plate)
132
Steel Fabric Induced Crack Concrete Bearn
50rnm
50mm
104 104 104 104 IN..

loolnin 50inm loomm 50mrn loomin
ELEVATION
20mm t
Figure 5.3 - Steel fabric and reinforcing bar layout and positioning
80 320
I --I
lie
320
-I-
Figure 5.4 - Typical single crack test specimen(Valle and Buyukozturk 1993)
133
eoller beving
-- --. SiDheli(al belfirij
itniOip
*dge bearing
W"CrWe
sp-mirnem
adusting
turnbucklo
fInxible
Strov
COMpross, on
jaCk - Aomd
ceir
Figure 5.5 - Single cycle load test set-up(Millard and Johnson 1984)
lesl Josm
DEPARTURE SLAH APPROACH ULAO

IWAFFIC
16- Lzading Pinfim
r Ircols ; Uifts) .
jr-dgtr
I li,
riuvio i,. Mull lesi $13;
t uIld 11111rumullu"llat.
.f
Figure5.6- Large-scaleslablaboratoryset-upfor cyclic loadtesting(Colleyand

Humphrey1967)
134
.-A 1-11 Icel
Figure 5.7 - Double crack test set-up for cyclic loading (Thompson 2001)
135
Position I- Neutral
I
Lab specimen
Position 2- Positive Load
I
Lab specimen
In-service slab
A Leave slab
Leave slab Approach slab
Position 3- Negative Load
Lab specimen
W/ \k
In-service slab
Approach slab Leave slab Approach slab Leave slab
Figure 5.8 - Representation of in service slab loading using positive and negative
laboratory loading
136
Specimon N34 - 25.4 mm trap rock
Spoclmon N32 - 38.1 mm trap rock
Is
10 ................................ .................................................................
E
E
................
...............
E
(D
C) 0 ................................................. ............... ......
IU
a law
m * oil 10
s ................ ................ ............... ........................ ...............
K-
va
c2
m
U)
. 10 ................ ........................ ....................... ...............
-154--
10+0 le+l le+2 le+3 le+4 1e+5 1e+6
Numbor of Cyclos
Figure 5.9 - Reductionin sheardisplacementgradient over increasingload cycles (Abdel-

Maksoud 2000)
Number of cycles N
0 2000 4000 6000 8000 10000
100
AIA&
80
60
Maximum load +2kN
40
20
Mix GI (Control)
Minimum load -2kN Series 17
0
w=0.49mm
-20 fto mass ease &*soon, Me' gamesome!
-40
Figure 5.10 - Reduction in peak and trough deflection gradient with increasingload
cycles (Thompson2001)
137
035
03
025
E 02
U
a
5
-ii 0 15
01
005
00 50000 100000 150000 200000 250000

No of Cycles
3,
0
03
025
0 02
a
0
7i 0 15
01
005
0
250000 300000 350000 400000 450000 500000
No of Cycles
Figure 5.11 - Trial testing differential displacementplot, 0- 250,000 cycles (Top),

250,000 - 500,000cycles (Bottorn)
138
I ub!, Ovuf
0.02
A-p f, Dep. Recovery
f-045 -0
0 S- 9d
0
3
0
TOTAL CYCLE
0 Appro-ich
C> 50
Slab
40 h Deporture
siuu
30
C)
LJ 10 --A
Figure 5.12 - Measured load cycle of a vehicle crossing a discontinuity at 30111ph(Colley
and Humphrey 1967)
-I
Ti- (S-. )
Figure 5.13 - Sinusoidal load application from the Dailec cvcllc load test machine
Figure 5.14 - Bekaert Dranlix@ Steel fibres used within the experimentation (frorn left:
RC-65/60-BN, RC-80/60-13N. RL-45/50-BN)
139
Saw-cut of 6mm depth placed circumfrentially
Load Application
Dem(
Shims
t.,
"
3 Point Bending to Form First Crack, Shim

Size Controls Dimensions
Load Application
Second Crack Formed, Beam may be Inverted

to Produce Parallel Geometry
Load Application
OUPIJUIL L. Qvtlvll
Figure 5.15 - Method of indLICingcracks into test specimens
140
Pre-CrackinLy
ACtuai t-jauge
Measurements
Estimated Surface
Post-Cracking Measurement
........ ... ....
Estimated Base
Extrapolation Line Measurement
Actual Gauge
Measurements
Figure 5.16 - Extrapolation of dernec pip measurements to obtain surface and base crack
width measurements
05 1152 25
Shi. siz. (-)
Figure 5.17 - Comparison of predicted and actual surface crack measurements
141
15 Xl. l I
, X,
05 A Front Left
' A
I Id 0 Front li, 11
0 Rear Left
X Rear Right
0
20 30 40 50 60 70 so 90
05 11 --- -- - ----
Distance From Beam Base (mm)
Figure 5.18 - Measurement of crack opening througlIOLIt specimen depth
Cyclic LoadingZ7, Test Ric,

C,
LVDT Measurements Dartec Load Cell

Measurements
Amplifier Voltage
Z:, Transforiner
CRIOX Data Logger

Z:
Computer Storage
Data Manipulation
Data Plots
Figure 5.19 - Graphical representation of data logging process
142
4
Front
Vertical LVDT llori/ontal LVDT
Rear View
Vertical LVDT Horizontal LVDT
Hori/ontal LVDT \ Vcrtical LVDT

Rear Face
Plan Vicw
Front Facc
Vertical INDT Iloii/ontal LVDT
lit acket
LVDT
Tar. et
Figure 5.20 - LVDT positioning (original set-up)
Vcrtical LVI)I'([-', ) Vci tical INDT (F, )
Veitical LVDT (R, ) I lorizontal LVDT (R, )
Holi/ontal LVDT Vcitical LVDT
Kcar Flice
I
Plan Vi
Front Facc
Vcrtical I NI VciticA LVDT
Ilracket
LVDT
Target
Figure 5.21 - LVDT positioning (updated set-Lip)
143
q
U
Av
02
Positions where Dart cc

O'l stopped
0.05
0
0 5f)(KI,) I ONOC 1baWO 20COOO 25OOuO
No of Cyclus
Figure 5.22 - Effect of undertaking a discontinuous Cyclic load test
-002
-0.04
E
E -006
c
0
-008
-01
-012
-014
Time (Secs)
Figure 5.23 - Maximum and minimum displacementpositions
144
1
05
)00
.05
15
No of Cycles
Figure 5.24 - Positive and negative displacement deterioration plot
35
15
05
00 50000 100000 150000 2000(X) 250000
No of Cycles
Figure 5.25 - Differential displacement deterioration plot

41
145
6. RESULTS, ANALYSIS & DISCUSSION
6.1 General Introduction
This Chapter is composedof three main sections,namely, crack geometry, slab condition
and joint deterioration. Although the work has been split to examine each area
individually, the data has been transferred where applicable to produce a more detailed
and clearer understandingof the load transfer mechanismsin operation.At the end of the
chaptera discussionsectionis provided which brings togetherthe key points from eachof
the three areasof analysis,enabling the laboratory and site obtaineddata to be examined
as a whole.
6.2 Crack Geometry
Crack width is known to be one of the main factors controlling the effectivenessof joints
under dynamic load. The collection of site data was necessary to enable typical
geometriesto be identified, which would: aid in the developmentand set-up of the cyclic
loading test rig, enable comparison with site deflection test results, validate the Finite
Element model, and provide information to floor contractors and designers on typical
joint opening behaviour.
According to the literature reviewed in section 3.3, surface measurementsalone fail to
provide all the required information in respect to crack geometries,as width regularly
reduceswith depth due to differential shrinkage.Several methodswere therefore used to
enablereal site values under characteristicenvironmental conditions to be established.
6.2.1 Core Samples
Core samples were obtained from the Lutterworth site on two occasions.The first visit
was carried out on 6"to 7h July 2002 and involved the drilling of 24 cores in total, two of
which were taken through sawn contractionjoints and two through cracks in the slab. The
remainder were positioned at internal conditions, some areas of which had developed
surface crazing. The location of each core on the site is shown in Figure 6.1, with
photographsof those containing cracks orjoints provided in Figure 6.2.
146
The secondvisit to Lutterworth was completed on the 21" to 22"dSeptember2002 with
40 cores taken in total, 10 of which were locatedthrough a wire guidancesysteminstalled
into the slab for the control of VNA trucks (Figure 6.1). The cuts into the slab appearedto
be working as inducers,with hairline cracks visible on the surfaceof the slabsextending
from the edge of wire guidance grooves. At each position Prima Dynamic Plate testing
was undertakenprior to coring to enable comparisonsbetween deflection responseand

crack geometry,the details of which are provided in section6.3.2.
For the first set of cores, dimensions of maximum/minimurn core length and
reinforcement size, were recorded by the Construction Materials Testing (CMT)

laboratory in Derby. The results for those containing either a contractionjoint or a crack
are reproducedin Table 6.1.
For the second set of cores CMT Derby identified the thickness of slab, cover to
reinforcing, depth of saw-cut and cover to wire, reproducedin Table 6.2.
Table 6.1 - CMT Core Information (Visit 1)
Core Le gth (mm) Reinforcement

Core Ref. Comment s
Max. Min Dia. * dmi,,
7 190 189 None Induced ContractionJoint
17 198 174 7* 106 Full Depth Crack
21 1 180 176 1 7* 122 1TaperedCrack to 123mm
24 1 196 1 190 1 18 * 95 1Dowel. Induced Contraction Joint
Table 6.2 - CMT Core Information (Visit 2)
Max Min Cover to Depth of Coverto

Depth of
Core Ref. Length Length Rebar Saw-cut Wire
Crack
(mm) (mm) (mm) (mm) (mm)
1 176 166 90 22 18 Total
3 186 177 88 21 18 Total
5 181 165 N/A 15 14 Total
6 198 174 134 14 7 Total
10 177 165 112 18 15 Total
16 172 161 94 15 11 Total
17 186 179 111 15 11 Total
24 199 196 119 13 9 Total
27 192 180 127 13 7 Total
28 204 199 195 14 11 Total
147
In the cores obtained from both visits the maximum and minimum lengths show
differences of up to 20mm over a core width of 100numChangesin the slab depth by this
amount can have a substantialeffect on both the structural resistanceof the slab, and the
friction which occurs between the foundation and the slab. This can lead to shrinkage
restraint at mid span,resulting in early deterioration of the structure(section 2.4.1).
The cover and depth of the reinforcement in the visit 2 cores varied by up to 100mm.
Placement of the reinforcement can influence the resistance to cracking and prevent
cracks that do occur from opening up to an extent which causesdeterioration problems

(section 2.5.2). Variations to the level found in the cores will therefore be highly
problematic and preventthe slab from respondingas desired.
On completion of the CNIT examinations further analysis was undertaken to meet the
researchobjectives.Each crack or joint had measurementstaken of its width and offset on

both sides, with amendmentsmade for saw cutting and relaxation of the core upon
removal, in accordancewith the methodology describedin Chapter4.2.3. The dimensions

of crack width are shown in Figure 6.3 for the first set of cores and Figure 6.4 for the
secondset of cores,where the width given is the averageof the measurementstaken at the
samedepth on either side of the core specimen.Each core for the first visit is discussedin
turn as each shows a different crack or joint type. The results and discussion for the
secondvisit have beengroupedtogetheras eachwas of a similar type.
The crack in core 21 (visit 1) was restricted to the upper half of the slab only, with its
width reducing in proportion to the distancefrom the surface.Although the exact causeof
the crack could not be established, it was likely to be either plastic shrinkage or
differential movement between the surface and base of the slab. The plot reveals that
shrinkage was only high enough to create a crack at the surface,reducing to non-critical
strain levels at approximately mid-depth.
Core 17 (visit 1) encompasseda mid panel crack which ran parallel to an aisle between
two high racking systems.The causemay have been clamping of the slab to the subbase,
preventing shrinkage movement which resulted in an increasein tensile stress.The crack

itself was found to penetratefull depth, being larger at the surface and reducing
proportionallyuntil closedat the base.
The inducedcontractionjoints in cores7 and 24 (visit 1) both containedan underlying

crack varying betweenmaximumat the surfaceto zeroat the base.The reductionin core
148
7 was proportional to depth, with the exception of severalpoints which deviated from this
line at aroundthe 40-80mm depth. This may have beencausedby aggregatespalling from
the crack edge,resulting in an 'apparent' increasedmeasurement,or differential shrinkage
created by inconsistentdrying. Core 24 contained the greatestcrack width between the
surfaceand 60mm depth, at which point it reducedto levels similar to that found in core
7. These large measurementsgo beyond that of the saw cut thus, its influence in creating
the profile is shown to be negligible. The larger top section width may have been caused
by the dormant/dominantphenomenonwhereby adjacentjoints were restrained, forcing
increasedmovement.The shapeof the crack was formed due to restriction from the dowel
bar and friction against the foundation in the lower section. In the top of the slabs no
restraint was provided, hencethe crack openedto a much greaterdegree.
Figure 6.4 demonstratesthat all cores taken in the secondvisit contain cracks which have
a width larger at the surface than at the base. In 80% of these cores the base width was
below 0.4mm, with the surface measurementsthemselvesvarying between 0.55MM and
1.45mm. The general trend showed a slightly steepergradient in the top 75 to 100mm
section of the core, indicating greater shrinkagein the concretecloser to the surface.This
phenomenonhas beendiscussedby Poblete et al (1988) and is discussedin section3.3.
Three of the cores (numbered 1,3 and 6) contained widths which varied greatly from the
geometry describedabove.These deviations can be attributed to breakageof aggregateat

the concrete edges when the cores were removed from the slab. Some degree of
differential shrinkagebetween sectionsof the slab may also have had an effect, creating a
non-linear variation betweentop and bottom faces.
The data shows that cracks under the induced joints are significantly larger than those at
mid panel. This is to be expected, as the joints will contain shrinkage from the entire
length of the concrete slab, whereas cracks will comprise less movement due to the
shorter effective length. Smaller width generally leads to a more efficient crack,
indicating that although more unsightly, they will create less problems than that of an
equally trafficked joint.
For each of the cores in visit Ia plot was produced of crack edgeoffset distanceon either
side of the specimen, to a level datum placed perpendicular to the slab surface (Figure
6.5). The results display a great deal of similarity, with the orientation varying only
slightly (25mm maximum) from the initial top position. Small deviations are regularly
149
spaced and caused by the crack following the interface between aggregateparticles,
known to be the weakestpoint in young concrete(Abdel-Maksoud,2000).
As expected, the magnitude of offset at the top 50-60mm. of the sawn specimens
(numbered7 and 24) is controlled by the width of the saw cut, below which the variation
is similar to that found in the specimenscontaining a crack only. The small change in
offset indicates that only negligible differences in load transfer will exist due to a change
in the loaded slab. The minor undulations of the crack face will therefore dominate and
enableaggregatebearing to occur equally in both directions.
Examination of the offsets in the secondvisit cores showed similarities with that of the
first. Each comprised a crack which showed little overall deviation from the surface
position, but contained several small undulations in vertical orientation. A decision was
made that only the crack width measurementwould be taken in these cores, with no
recordingsof offset made.
Examination of all the cores has enabled typical dimensions and geometries of crack
width to be calculated. These have been undertaken for each crack type, i. e. induced
contraction joints, cracks, and semi induced cracks (wire guidance grooves) and are
shown in Table 6.3. The results indicate that generally all joints have similar behaviour
with respect to crack opening, the width increasing in proportion to slab depth. However,
the predicted surface widths for semi-induced cracks using extrapolation of the bottom
and mid slab widths (Figure 4.5) are generally smaller than the actual surfacedimensions
by approximately 18%, indicating slightly non-linear opening. This behaviour is similar
for contraction joints and cracks, although the variation in predicted and actual surface
crack opening is greater.This information is of use when extrapolating crack widths from
either embeddedstrain gaugesor surface as

measurements, it confirms that the predicted
surface values will generally be slightly smaller than that actually occurring within an in-
service slab crack orjoint.
150
Table 6.3 - Comparison Between Actual and Predicted Surface Crack Widths
Core Type Actual Actual Actual Predicted Variation

Base Middle Top Top (%)
Contraction
7 1.2 3.1 2.4 22.5
24 1.55 8.3 3.1 62.7
Average 0 1.38 5.7 2.75 TI-. 7
Crack
17 0.32 0.3 0.64 -82.9
21 0.08 0.9 1.84 1.64 10.9
Average 0.04 0.61 1.10 1.14 -3.6
Induced
I 0.2S 0.4 1.45 0.64 12.5
3 0.3 0.6 0.8 0.9 -12.5
5 0.4 1.05 1.38 1.7 -23.2
6 (). 1 0.55 0.85 1 -17.6
10 0.4 0.9 1.1 1.4 -27.3
16 0.2 0.35 1.2 0.5 58.3
17 0.2 0.25 0.7 0.3 57.1
24 0.2 0.3 0.55 0.4 27.3
27 0.3 0.4 0.7 0.5 28.6
0.15 0.65 1. 0.85 34.6
Average 0.25 0.538 1.00 0.819 18.1
6.2.2 Strain Gauge Monitoring
Work undertaken by Bishop (2(X)I) required the installation of strain gauges into a
concrete floor slab to monitor the early age movement caused by both hygral and
temperature effects. The gauges were placed at various positions, many of which were
sited across contraction and induced joints. At three sites, two or three gauges and/or
demec pips had been placed to increasing depths at the same location, enabling
measurements to be taken of width variation. Full information on the strain gauge types
and method of placement is provided in Bishop (2001), with a summary of the joint
openings found by Bishop reproduced in Table 6.4a. The values for Leeds were measured
at 222 days after casting, Marston at 12 days, and Northampton at 28 days. The thickness
of the slabs and positions of the gauges are given in Table 6.4b and Figure 6.6.
151
'Fable 6.4a - Strain Gauge Readings Converted to Crack Width
Gauge No I Joint fop Demec Top Strain Gauge Bott. Strain Gauge
Type (mm) (mm) (mm)
LEEDS
19 FF 0.11 0.01
20 SR 0.90 0.20
21 SR 0.50 0.10
22 SR 0.70 0.25 0.15
23 SR 0.90 0.30 0.10
25 FF 0.80 0.18
27 SR 0.60 0.18 0.00
28 SR 0.60 0.20 0.03
31 FF 0.30 0.02
34 SR 1.20 0.30 0.18
30 SR 1.00 0.45 0.,20
MARSTON
2 FF (wall) 2.70 2.05
12 FF 2.80 2.80
23 FF 2.70 2.70
31 FF (wall) 1.90 2.25
31 (Amm) FF (wall) 1.90 1.75
32 FF (w a11) 1.95 1.70
NORTHAM PTON
I SF 0.62- 0.58
2 SF 0.80 0.75
SR - Sawn Restrained
FF - Formed Free
SF - Sawn Free
Table 6.4b - Strain Gauge Placement
LEEDS MARSTON NORTHAMPTON
Slab Depth 225min 175nini 260inni

Sur ice to Top Gauge 105mi-n 55nu-n 140inm
surface to Bottom 165rnm 115mm 200nini
Extrapolation of the measurementsgiven by the embedded strain gauges using a linear
crack formation (known to predict slightly smaller surface widths, section 6.2.1). enabled
the crack widths for both the base and surface of the slab to be determined (Figure 6.7).
Where dernec pips were installed onto the slab surface, these were used along with the
singular strain gauge, with those containing two strain gauges and deinec pips utilising
152
the dernec and top embeddedgauge for the surface measurement,with the base value
determinedfrom the gradient of the two embeddedstrain gaugesalone.
Occasionally on the Leeds site, strain gaugesand demec pips were used directly above
each other enabling comparisonsbetween the predicted and actual surface values. The
results of the measurementsare shown in Table 6.5 with Figure 6.8 providing a plot of all
three data points.
Table 6.5 - Measured and Projected Surface Measurements
Gauge No Projected Predicted Top Actual Top Variation

Bott. (mm) (mm) (mm)
22 0.0500 0.4250 0.7000 39.3
27 -0.1800 0.4950 0.6000 17.5
28 -0.1400 0.4975 0.6000 17.1
34 0.0600 0.5100 1.2000 57.5
[-36
-0.0500 0.8875 -7 1.0000 11.3
In situations where three measurementswere used,the actual surfacecrack width value is

larger than that predicted using the embeddedstrain gauges, with numbers 22 and 34
showing the greatestvariation. This indicates that the crack opening is non-linear, with a
slightly curved orientation as shown in Figure 6.8. This agreeswith the plot by Neville
(1981) (Figure 6.9), who found that shrinkagevalues increaseexponentially towards the
in
surface,resulting a curved profile when only one side is open to the environment.This
is similar to the results found in the core examinations, where the surface of the slab
contained crack width values approximately 18% larger than extrapolation of the bottom
gradientspredicted (Table 6.3).
The findings illustrate that surface measurementcan only provide an estimation of crack
width profile due to the non-linear shrinkagethroughout depth. However, as the variation
is only minimal (generally below 20%), it can be assumedfor the analysis of surface
profiles (section 6.2.3) that crack opening is linear. In reality a slightly higher gradient
may be found in the top section of the slab due to the increased potential for moisture
movement,and will therefore produce a larger crack angle and basewidth (section4.3).
Figure 6.7 confirms that most joints have a larger width at the top surfacecomparedto the
bottom. In 80% of the joints testedat Leeds a projected surfacevalue of up to 2MM closes
completely in the bottom 50mm section. This indicates that either a crack will not exist
153
below this level, or that it will be closed and in a compressivestate.Similarly, the results
from the Northamptonand Marston sites show a decreasein width over depth of between
0.2 and 3.5mm respectively; however, in both cases the joint appears to be open
throughout depth, with the basewidth still in excessof 0.6 and 2mm.
The larger crack widths at Marston are mainly causedby the different joint type. The
majority of results from Leeds originate from sawn restrainedjoints with only gauges19,
25 and 31 sited across formed free joints. Marston however, contains only formed free
joints, with the slab constructed using jointless methods, hence the longer effective
lengths and larger magnitude of movement. Joints numbered 2,31 and 32 were taken
against a wall gauge meaning that all measurementof crack width variation occurred
from one side only. However, these have been doubled in the relevant figures and tables
to produce suitable comparison with the other locations. Interestingly, joints 12 and 23
show equal movementthroughout depth, indicating little differential shrinkage,with joint
31 showing greater opening at the base than at the top. The cause of this reversed
orientation is unknown, although on examination of the time/joint opening plot it appears

that this phenomenonoccurred within the last few hours of data collection, whereby a
sudden increase in lower gauge movement occurred (Figure 6.10). This indicates a
possible measurementerror, with the earlier behaviour (Gauge 31 Amended) providing
more realistic information in respect to crack opening. The Northampton site contained
data from free sawn joints alone, with friction against the subbase and dowel bars
perpendicular to the joint preventing slab shrinkage being converted into opening. This
resulted in a small width, but still contained an increase in width with distance from the
slab base.
The results from the embeddedstrain gauge and demec point measurementshave again
indicated that the crack increasesin size towards the slab surface causedby differential
shrinkage. This is most pronounced in the top section of the slab where the crack angle is
greater than that in the bottom. The type of joint under consideration controls the
magnitude of this movement with sawn restrainedjoint surface measurementsgenerally
below 2mm, and free movementjoints over 5mm. This is generally causedby the lack of
shrinkagerestraint, and an increasedlength betweenadjacentcrack positions.
6.23 Levelling Profiles and Crack Measurements
Crack orientation can be determined from the top width measurementand the surface
profile of the slab if an assumptionis made that opening is linear and all shrinkagestrain
154
has been converted into curl. In construction joints this method is unreliable due to
apparentcurling, which is a surfacecondition causedby upsweepingof the trowelled face

during construction. Sawnjoints are formed on a level surface within a continuous slab,
and therefore any vertical changemeasuredcan be attributed to curling. This method of

determinationwill normally estimatebasewidths slightly larger than actually found in the
joint or crack due to non-linear shrinkage(section4.3).
At Daventry profiles were determined across two slabs and a number of joints using a
precise level. The area of floor surveyed is shown in Figure 6.11, with the methods
employeddescribedin section4.3.2.
The data recorded has been transformed into contour plots (Figure 6.12), with the joint
locations superimposed to enable the positions of high points and curling to be
determined.The results show a clear indication of warping at the slab edgesand comers
as demonstratedby the lighter patcheson the contour plot. This is most apparentat the
joints running in the vertical direction which show an increasein height of up to IOMM
from that found at mid span, and level changesof 4mm.within a distanceof 0.5m.from
the joint itself This effect will almost certainly be significant in the creation of a crack
which opensmore towards the surface.The length of slab at which point the curl reduces
significantly is approximately Im. This provides good correlation to the work of
Suprenant(2002) who estimated curling to be between 10 and 20% of the slab length,
which in the caseof the Daventry site would provide a length between0.75 and 1.5m.
Unfortunately, time constraints prohibited precise levelling on the remaining sites. At

Ballymena the simple methods utilising the builders level and the profilometer were
employed across all joints to provide an indication of curling magnitude, the results of
which are shown in Figure 6.13. This method only enabled the variation between slab
edges and a distance 0.5m back to be established,and therefore provided limited data on
the complete curled profile. The results show a variation in slab edge level of between0.4
and 5.1mm, similar to that obtained when examining the Daventry site.
Assuming orientation of the crack face is in direct relation to the degree of curl (as
described in section 4.3.1), it was possible to determine crack measurementsusing only
the slab profile and surfacejoint widths obtained from site. Results obtained using this
method for Daventry are shown in Figure 6.14, where the levels closestto eachside of the
joint (500mm apart) have beenused to enablethe most accuratecalculations of edgecurl.
The surface crack widths varied between 3 and 8mm, with 80% showing a reduction in
155
width with depth of between I and 4mm. The estimation of basemeasurementsproduce
widths between I and 6mm; however, as found from both coring and embeddedstrain
gauges, the estimation of proportional decreasefrom the surface value is expected to
result in basevalueslarger than reality (section4.3).
The Ballymena site was constructedusing the long strip method, with eachjoint formed
rather than sawn and only opening up marginally, making it extremely difficult to
measure crack width. However, the profile gradients have been determined and are
provided in Figure 6.15. The plot shows a variation between surface and base
measurementof between I and 4.5mm, although the actual values calculated are not
representativeof that on site becauseof the joint type under examination.
The typical crack profiles can be seen for both sites with the upward curl of the slab
providing a width narrowing with depth, the variation being 1-4min for both Daventry
and Ballymena. These are realistic values and agree with the typical sizes obtained by
coring in the Lutterworth site, and by strain gauge monitoring of Leeds, Marston and
Northampton.
6.2.4 Summary
All three methods for investigating crack profiles show that the surface measurementis
larger than at the base. This is causedby differential shrinkage as described in section
2.4.2, with the top surface prone to greater movement due to the drying environment. In
all situations the curl was in an upward direction, with the edge of the slab being at a
higher elevation than that of the centre. The value of this difference was approximately
10mm, with the variation betweenthe edge500mm section in the region of 1 to 5mm, and
the length of unsupportedslab approximately I in. The width measurementappearedto be
partially controlled by the joint type, with reinforcement holding the crack together and
producing a lower opening throughout. The cores showed that the increasein crack width
is generally proportional to depth, although a slightly higher degreeof movementis found
nearer the surface. This is also found with the embedded strain gauges,where the top
demecs showed a larger surface measurementthan any prediction based on embedded
gaugeextrapolation alone. In all casesthe magnitude of variation betweenthe surfaceand

basecrack widths was between 1 and 5mm.
These findings highlight the fact that crack geometries in concrete slabs on grade are
different to that assumedin much of the literature. The effects of a 'V' shapedcracking
156
pattern can be highly influential to the load transfer occurring across the crack and its
associatedresponses.Assumptionsof parallel cracking with a width equal to the surface
measurementare thereforeconservativein design and inaccuratefor structureassessment.
The maximum propagation of the crack towards one side in the core specimenswas
always found to be below l5mm. This small deviation, alongsidethe numerousaggregate

undulations throughout depth, produced a similar load transfer mechanismacross each
face. This made the loaded side unimportant when calculating load transfer and thereby
preventedfield-testing having to be undertakenon both sidesof the joint.
6.3 Slab Condition
6.3.1 Introduction
Slab and joint condition has been determined using calculations of. load transfer, load
step, crack width, edge cantilever, deflected shape and voiding. The testing was
undertaken using either the FWD or Prima dynamic loading equipment, with both devices
being usedon the Lutterworth and Daventry sites.
Comparison was made between the two pieces of deflection testing equipment with
respect to load transfer and load step. This enabled any variations in responsecausedby
the increasedload of the FWD to be established,and therefore provided confidencein the

results from the Prima. Comparing the devices and load in this way also produced some
interesting findings with respect to the void effect, and that of the transfer mechanism
occurring across the joint. In the remainder of the slab condition calculations,
comparisons betweenthe FWD and Prima have not been undertaken;however, the data is
affected by the magnitude of the load and this has been taken into account during
assessment.
Each calculation of slab response (i. e. load transfer) has been compared to other
behaviour within the same slab. This has enabled many relationships to be developed,
highlighting the importance of joint effectiveness on slab condition. Load transfer and
load step has also been compared to surface crack width, which, although known from
section 6.2 to only be partially representative of the full geometry has enabled its
influence to be determined.
157
The foundation has a highly influential impact on deflection response.Engineers had
suspectedthe site at Lutterworth to contain voiding underneaththe slab edge and have
variable support conditions. This was confirmed by the testing undertakenon site using
the void intercept approach,which predicted that between50 and 90% of the slab edges
contained voids (section 6.3.6). The site at Ballymena was assumedto have a good
foundation due to Engineer's assessmentand the formation level at which the structure
had beenconstructed.This information was of importance when analysingslab behaviour
as it helpedin determining site variability.
63.2 Load Transfer Evaluation
Device Comparison
Comparisonswere made on the Daventry site between the load transfer values obtained
using the normalised 50kN FWD and IOkN Prima dynamic plate. This enabled the
devices to be correlated, ensuring that data from one site could be comparedto another,
whilst allowing the effect of load magnitude on slab response to be determined. The
locations for each test were marked on the slab enabling the samepositional set-up to be
establishedfor each piece of equipment.The results obtained are shown in Figure 6.16,
where load transfer is calculated using the Crovetti and Darter (1985) approach(equation
3.4).
The graph exhibits good correlation between FWD and Prima results (k2 = 0.94), with
only minor distribution from the equal value line. The full range of load transfer has been
establishedand its value does not seem to affect the relationship between results. The few
points that do show variation greater than 10% are not significant points on the axis;
however, on inspection they were found to contain large crack widths (between 5 and
I Imm), regardlessof the efficiency of the joint. Examination of the full data set showed
no direct link betweenmagnitude of variation from the idealised line and surfacewidth of
the crack, and therefore the variation was attributed to site irregularities, possibly caused
by the testing procedure and greater load in the FWD. Four construction joints were
included within the analysis and each showedexcellent load transfer (67-90%) with good
comparisonbetweentest equipment,the maximum deviation being 1% or lower.
For the Lutterworth site three FWD load magnitudesof 42,58 and 85kN were used along
with the lOkN Prima dynamic plate. Comparison of the results from the largest and
smallest FWD loads is shown in Figure 6.17, with the mid-range 58kN FWD load
comparedto the Prima in Figure 6.18.
158
The foundation has a highly influential impact on deflection response.Engineers had
suspectedthe site at Lutterworth to contain voiding underneaththe slab edge and have
variable support conditions. This was confirmed by the testing undertakenon site using
the void intercept approach,which predicted that between 50 and 90% of the slab edges
contained voids (section 6-3.6). The site at Ballymena was assumedto have a good
foundation due to Engineer's assessmentand the formation level at which the structure
had been constructed.This information was of importance when analysing slab behaviour
as it helped in determining site variability.
6.3.2 Load Transfer Evaluation
Device Comparison
Comparisons were made on the Daventry site between the load transfer values obtained
using the normalised 5OkN FWD and lOkN Prima dynamic plate. This enabled the
devices to be correlated, ensuring that data from one site could be comparedto another,
whilst allowing the effect of load magnitude on slab responseto be determined. The
locations for each test were marked on the slab enabling the samepositional set-up to be
established for each piece of equipment.The results obtained are shown in Figure 6.16,
where load transfer is calculated using the Crovetti and Darter (1985) approach(equation
3.4).
The graph exhibits good correlation between FWD and Prima results (R2=0.94), with
only minor distribution from the equal value line. The full range of load transfer has been
establishedand its value does not seemto affect the relationship betweenresults.The few
points that do show variation greater than 10% are not significant points on the axis;
however, on inspection they were found to contain large crack widths (between 5 and
I Imm), regardlessof the efficiency of the joint. Examination of the full data set showed
no direct link between magnitude of variation from the idealised line and surfacewidth of
the crack, and therefore the variation was attributed to site irregularities, possibly caused
by the testing procedure and greater load in the FWD. Four construction joints were
included within the analysis and each showedexcellent load transfer (67-90%) with good
comparison between test equipment,the maximum deviation being 1% or lower.
For the Lutterworth site three FWD load magnitudesof 42,58 and 85kN were usedalong
with the lOkN Prima dynamic plate. Comparison of the results from the largest and
smallest FWD loads is shown in Figure 6.17, with the mid-range 58kN FWD load
compared to the Prima in Figure 6. IS.
158
The comparison betweendiffering FWD loads is good (R2 = 0.95) with the data varying
little from the equal value line. However, on closer inspection load transfersin the region
of 90% or higher demonstratethat a lower load impact producesa higher value. When the
load transfer reducesbelow 90%, the reversetakesplace whereby lower loading produces
a reduced value. The overall difference is found to be below 10% throughout the range of
data, although only a few points provide inforniation below 50%.
Results betweenthe Prima and FWD display a similar relationship (R2=0.603) to that of
the high and low loading of the FWD (Figure 6.18). Towards the very high load transfer
values i. e. 90% or more, the correlation between results is similar and follows the equal
value line well. However, as this reduces,the FWD load transfer becomesslightly higher
than that of the Prima, the difference increasingthe lower the value. Unfortunately, data is
limited below the 60% level and thus, it is difficult to ascertain whether this effect
continues into the lower levels of load transfer. To examine if this effect is similar for all
joint types, the data was separatedto enable comparisons between sawn, formed and
cracks. The formed joints show the greatest comparison to the idealised line, with the
sawn joints and cracks showing the largest variation.
A probable reason for a higher load transfer under greater load is causedby the free slip
phenomenon formed when concrete faces are placed under repetitive loading. The gradual
wearing away of the micro roughness leaves only the macro roughnessof the crack to
resist deflection. This produces an area of free movement, or free slip, which contains
little by way of support. Under a low load this free slip will be a higher proportion of the
overall deflection producing a lower load transfer value; under a high load the slab will
deflect more and the proportion of the free slip will becomesmaller, indicated by a higher
load transfer; this is shown diagrammatically in Figure 6.19. As the amount of free-slip is
dependant on the attrition of the crack face, slabs withstanding higher load cycles
(generally older slabs) will create greater free-slip. Any reinforcement acrossthe joint or
crack will also play an important role as this can reduce the rate of crack face degradation.
Table 6.6 shows the theoretical calculations from a joint which contains 100 microns of
free slip movement, but provides 50% load transfer once the joint faces are engaged.The
increasein transfer effectivenesscan clearly be seenwhen the magnitude of load becomes
higher.
159
Table 6.6 - Effect of Free Slip on Load Transfer
Load (kN) Slab 1 Deflection Slab 2 Deflection Load Transfer

10 100 0 0
20 150 25 16.7
30 200 50 25
This agreeswith the results from the Lutterworth site as the magnitudeof variation from
the idealised line increasesthe lower the load transfer and greaterthe free slip percentage
becomes.The separationof joint types also shows that those containing dowels produce
the best agreement, as the bar will assist in the load transfer phenomenon, thereby
negatingthe effect of any free slip. The sawnjoints producethe most discrepancyas these
generally open up more than the others, enabling greater free slip and an increased
variation in load transfer values. This effect is reduced on the Daventry site since
incorporation of steel fibres smoothesthe shearslip phenomenon,as discussedin section
3.4.4 and shown in the laboratory testing results (section 6.4.4). Similarly, this can be
affected by the aggregateproperties which control the amount of free slip for a particular
crack width. Although this method explains the phenomenonof load transfer variation
betweenthe devices, it requires further work to assessthe exact effect of load magnitudes
on the FWD and Prima.
Load Transfer Equations
Examination was made of the different load transfer equationsprovided in section 3.7 to
observe their results from site obtained deflections. As the tests on site consisted of one
directional testing only many of the calculations described could not be applied, thus,
comparisons were only made between the equationsof Crovetti and Darter (1985), Teller
and Sutherland (1943), and Pradhan (2002). The deflections at Daventry displayed the
best correlation between test equipment, contained a good range of load transfer values,
with little curling or voiding at the slab edges, and were therefore selected for the
evaluation.
The findings shown in Figure 6.20 illustrate that the equationdevelopedby Pradhan
(2002)produceonly a 50%or greaterloadtransfervalueregardless of the ineffectiveness
of thejoint. Althoughassimpleto calculateasthe CrovettiandDarter(1985)method,the
reduced range of results suggeststhat changesin load transfer stiffness are less obvious.
The results obtained from the Teller and Sutherland (1943) equation exhibit a non-linear
curve when compared to that of Crovetti and Darter (1985). At the two extremities of 0
160
and 100% load transferthe results are the same,however, towards the 0-50% load transfer
range a greater value per unit increaseis produced,indicating that small changesin load
transfer are more easily identifiable. The reverse happenstowards the 50-100% range
whereby the curve flattens and any increase in the Crovetti and Darter (1985) equation
load transfer producesonly a marginal increasein the Teller and Sutherland(1943) value.
In all cases,the values createdby Teller and Sutherland(1943) are greaterthan those of
Crovetti and Darter (1985); however, it is the changesin gradient betweenlow and high
load transfer which show the greatest difference, providing greater clarity when
determining small changesbelow 50%. The clearest of the equations appearsto be the
Crovetti and Darter (1985) method as it is a simple ratio betweendeflections.
Crack Width
Where possible a comparison was made between crack width and load transfer. This
enabledthe effect of crack opening to be examinedand to assess

whether its measurement
on site could provide estimations of slab Although
response. not conclusive the Daventry
site indicates that the degree of load transfer effectiveness reduces as the surface
measurement of crack width increases(Figure 6.21a). A best-fit line placed through the
data reveals the trend, with an appropriate lower bound showing that greater than 70%
load transfer or abovecan be assumedwhen the crack width is below 2MM. This level of
load transfer value for such a crack width is much greaterthan predicted by the laboratory
data of Colley and Humphrey (1967) or Benkelman (1933); however, the field data of
Colley and Humphrey (1967) shows reasonableagreement.The higher load transfer on
the sites tested in this researchcan be attributed to the influential effect of the fibre, and
the crack geometry found on site.
Only two points at 9 and 12% load transfer vary from the lower bound level and on
inspection were found to be two of only four points tested at a comer location. These
incorporated the reduced support of two crack faces, thus resulting in a lower load
transfer. Four induced contraction joints were found to contain crack widths of
approximately Ilmm, but had load transfer values which varied between 35 and 95%.
The most likely causeof this effect is from the geometry of the crack, which was likely to
have been significantly smaller at the basethan at the surface.The number of cycles that
eachjoint had been subjected to may also have been low, thereby reducing deterioration;
however, this could not be establishedduring the site visit.
161
Crack and sawnjoint results from the Lutterworth site show a less obvious relationship
with load transfer, mainly causedby the lack of data in the load transfer region below
60% (Figure 6.21b). The greater degree of scatter in the load transfer data for the
Lutterworth site may be causedby the type of reinforcement found acrossthe joint. The
mesh in the base of the slab is known to enhanceload transfer capacity; however, the
Concrete Society (2003) state that this will yield due to the opening of the crack. The
amount of support provided is therefore questionable,and may vary enormously across

individual joints, resulting in a wide variation in load transfer.
The data from the site at Skelmersdaleshows large scatter with good load transfers at
crack widths over 10mm (Figure 6.21c). A line of best fit was inappropriate becauseof
the poor relationship. The data signifies that some form of additional load transfer
mechanism had been inserted across some joints to enhance load transfer. The
information obtainedfrom site revealedthat a lot of repair work had taken place in which
the joint load transfer mechanismsmay have beenimproved.
Cores drilled across cracked wire guidance grooves at Lutterworth provided detailed
information on the changing width over depth (section 6.2.1). Testing with the Prima
dynamic plate was undertakenadjacentto each core location and enabledthe relationship
between crack width and load transfer to be determined. The surface measurementsof
crack width showed poor comparison to load transfer; however, when the base crack
width was used a much closer correlation of increasing crack width resulting in lower
load transfer values was found (Figure 6.22). Unfortunately, all cracks produced high
levels of load transfer and therefore the spread of data was restricted to 75% or more.
Even with this limited information it can be seenthat the base width appearsto control
the value of load transfer to a higher degree than that of the surface measurements,and
therefore care should be taken when assessing cracks and joints on the surface
measurementsalone.
6.3.3 Load Step Evaluation
Load stepis the differencein deflectioneither side of a crack whenplacedunderload,

and is a functionof load foundation
transfer, supportandappliedload.This valuecanbe
determinedusingeitherthe FWD or the Primadynamicplate and wascalculatedfor all
sites using the set-updescribedin section4.5.2. Here, geophoneswere placed50mm
either sideof the joint with the centreof the load platesituated250mmthejoint (Figure
4.9).
162
Device Comparison
At Daventry the joints and cracks were tested with both the FWD and Prima dynamic
to the magnitude of load step causedby each device. The comparisonof data
plate obtain
between the normalised 50 and lOkN load steps for the two devices is shown in Figure
6.23, with the line best fit (R2=0.948) plotted alongside the line which would have
of
been produced had step been controlled solely by load, (Le. the 50kN FWD load step is
five times that of the lOkN Prima). The results indicate that the load step acquired from
the FWD is in fact only 4 times larger than the Prima, confirming that the influence of the
foundation and load transfer causes significant changes in responseunder
supporting
load.
Lutterworth displayed very similar results to those of DaventrY with both the different
load magnitudes applied by the FWD (Figure 6.24), and the FWD and Prima (Figure
6.25), producing lower steps than would be obtained from a direct load ratio increase.
Examination of the variation between the site measuredand predicted ratios of applied
load magnitude reveal that the difference reduced as load magnitude was increased,
in Figure 6.26 and summarised in Table 6.7, i.e. as the load level was increasedthe
shown
difference in gradientbetweenthe predicted and actual values is reduced.
Table 6.7 - Load Step Variation (Lutterworth)

Step ratio Load Ratio Difference Difference/kN
oad (kN)
10-58 (Prima/FWD) 3.317 5.800 2.483 0.051
58 (FWD) 1.2337 1.381 0.1473 0.0092
.
42-85 (FWD) 1.512 2.024 0.512 0.011
-58--851 1.238 1.466 0.228 0.0084
(F-WD)
Figure 6.27 shows a plot of load versus absolute deflection for the increasing FWD
loading, and indicates that the supporting foundation has linear stiffness. As load
has little direct effect on load transfer (demonstratedin section 6.3.2), voiding
magnitude
the slab and free slip in the load transfer mechanism must cause the variation
under
between load magnitude and step ratio. In a voided slab, placed under a small load, edge
deflection will be resisted mainly by the stiffness of the concrete slab. Under a large load,
will be closed up and the deflection will be resistedby a combination of slab and
any void
foundation stiffness (Figure 3.15). Figure 6.26 and Table 6.7 demonstratethis to be the
Lutterworth as an increasein load magnitude on site producescloser to
agreement
case at
of the load ratio line. Under low loading (such as the Prima) the effect of the
the gradient
is enhanced, with its size responsible for variations in slab behaviour. This is seen
void
163
when comparing the Prima lOkN load step to the FWD, as its coefficient of determination
is much lower than that of the two 42 and 85kN FWD loads (0.4 and 0.915 respectively).
The site at Daventry shows a better correlation betweenPrima and FWD load steps,as
Lutterworth is thought to contain a greaterdegreeof voiding (section 6.3.1).
Load Transfer
To ensure the slab remains serviceablethe dynamic load step must not exceeda certain
criterion. The developmentof a relationship between load transfer and load step would
enablethe designerto specify a load transfer dependanton the serviceability requirements
of the vehicles used on the floor. For all sites a comparison was made between the
magnitude of load step and the value of load transfer (Figure 6.28). This was undertaken
with the Prima and FWD at the Daventry and Lutterworth sites, with the Prima alone used
at Ballymena and Skelmersdale.Each plot indicates that the line of best fit contains a
linear portion from the 40 to 100% load transfer range, with the horizontal axis being
crossedat 100%. Values below 40% appearto have a higher gradient of step change,the
overall shapefitting well to a logarithmic curve. The correlation against the data is very
high (typically with an R2 value of 0.75 or higher), with load appearing to be the
controlling factor in load step magnitude. In all cases the Prima results show almost
identical best fit curves, with those of the FWD increasing in magnitude in relation the
size of the load.
Crack Width
Figure 6.29 showsthat an increasein crack width createsa higher value of step; however,
a best fit line is inappropriatedue to the large scatterin the data. When the plots are split
into their respective loading runs, as on the Daventry site, the effect of crack width
becomesclearer (Figure 6.30). Run I was taken along the ends of the aisles containing
little other loading source,run 2 in a bulk storagearea where goods were stockpiled, and
run 3 directly through a high racking aisle. The improved relationship when the results are
split shows the effect of preloading. Those obtained from the areascontaining the highest
load (racking, run 3) resulted in lower magnitudesof step.
Invalid Load Transfer Values
On the Lutterworth site a few test locations resulted in load transfer values above 100%,
thereby producing negative load step values. White Young Green (2002) found the same
phenomenonwhen testing other sites, suggestingthat the cause was a hogging effect at
mid span, created by high loading of the racking either side of the crack. This irregular
164
load transfer effect has beenthe causeof a debatebetweenvarious specialistsin the field
(FWD Users Group 2001) who state that the cause is either errors in the collection of
data, or more probable, the slab acting as a beam on an elastic foundation with fixed
edges. When placed under any additional load the maximum deflection will therefore
occur at a position other than directly under the application point, creating the large value
of joint efficiency. Any value above 100% load transfer, or with a negative load step is
thereforerepresentativeof a fully supportedcrack or joint.
6.3.4 Slab Cantilever
Edge cantilever enablesthe rate of slab bending to be determined,which can then be used
to assessfloor flatness,currently an important issuein design.The calculation is basedon
the variation in dynamic deflection betweenthe geophonenext to the joint on sameside
of the load and that directly under the load. Comparisonsbetween the FWD and Prima,
and crack width were not undertaken as their effects have already been discussedin
sections 6.3.2 and 6.3.3. The relationship between edge cantilever and load transfer or
load step enables serviceability requirements for floor flatness to be checked and
controlled using the known or expected behaviour of the joint. Additional mechanisms
can then be insertedif required.
Load Transfer
The direction and magnitude of cantilever should be related to the load transfer
effectivenessof the joint when all support conditions and free slippage are similar. In the
case of a joint with good load transfer, a negative result will be produced. In a joint with
poor load transfer the result will be positive, the magnitudesof which relate to the amount
of load transfer. However, this is rarely the casedue to crack deterioration and voiding,
leading to a high degreeof scatterin the data (Figure 6.31). This is highlighted by the site
at Ballymena, known to contain good support conditions, producing a better coefficient of

determination (R2=0.4) than that of Lutterworth (R2=0.184), which is suspectedof
voiding. In all sites lower cantilever deflections produce higher load transfer magnitudes,
with cases in the positive direction (i. e. the deflection is at a maximum under the load)
resulting in load transfer values above 80%, representativeof an almost fully integrated
slab.
Load Step
The Daventry site indicates that slab cantilever is proportional to load step with a
coefficient of determination equal to 0.673 (Figure 6.32). This is similar for Lutterworth,
165
Ballymena and Skelmersdale;however, the reducedmagnitude of step makesthe results
more difficult to interpret due to their close proximity. This is further reducedwhen the
Prima is used due to the smaller load applied to the slab. The general correlation is
expected since the load step is created from absolute deflection and load transfer
effectiveness.
63.5 DeflectedShape
The deflection bowl is produced by recording geophonedeflections as a dynarnic load is

imparted on the slab.Theseare then plotted againstthe associateddistancesfrom the load
to
source produce a deflection bowl (section 4.5.4). Foundation stiffness can then be
examined and the position of the greatestrate of bending within the slab confirmed. This
is required for the assessmentof floor flatness and to verify the relevance of edge
cantilever determination. This procedure was undertaken at both the Daventry and
Lutterworth sites using the FWD equipment.
Figure 6.33 shows the results from the Daventry site, where the zero vertical axis is the
joint location. The point at which deflection becomes negligible is approximately
2350mm from the slab edge and is the sameregardlessof the effectivenessof the joint.
Only one point conflicted with this, and on inspection was found to be at a comer
location. Here, geophoneswere placed linearly along the length of the joint, resulting in
greater deflection due to the reduced support. The plot of geophonevalues generatesa
smooth curve between readings with the maximum rate of bending at the slab edge. A
step is created at the joint location due to the discontinuity, the value of which is
determinedby the degreeof load transfer in the joint.
Figure 6.34 from the Lutterworth site showsthat the geophonelocations furthest from the
joint still produce a wide variation in deflection. As the results are provided for a slab
loaded at the edge, it was not possible to use back-calculation methods to determine the
exact foundation conditions. However, if examining the furthest geophones(D6 and D7,
Figure 3.14), which are affected mainly by the subgrade,it can be assumedthat the site at
Lutterworth has a lower foundation stiffness than at Daventry due to the higher
deflections.
6.3.6 Voiding
Under slab voiding could only be estimated at Lutterworth as this was the only site in
which variation of FWD load was undertaken for the same joint. Using the methods
166
explained in Chapter 4.5.5 an intercept value was calculated, and used to examine the
effect of voiding on other slab responses.Depending on which hypothesis is used
determineshow many of the slabs are estimatedto contain voids. If the Wade, Cuttell et
al. (1997) 75 micron limit is used 50% of the total slabs tested are affected, 60.5% if 50
microns is used (Crovetti and Darter 1985), and 88% if 25 microns is used (Cudworth
2003).
63.8 Summary
Increasingcrack width shows a generaltrend of lower load transfer and higher load steps.
However, the relationships are unclear with some very large crack widths still providing
low deflection related response,and therefore good performance.When the data is split
into separate static load areas (such as aisle racking), it shows better correlation,
indicating that preloading of the slab can have a large impact on slab response.This is to
be expectedas any voiding or loose foundation material at the slab edge will have closed
or been compacted.Tests of load transfer and crack width acrossjoints where the full
geometry was known, showed that the relationship between the base crack widths gave
better correlation than that of the surface measurement.This suggeststhat the surface
crack width must be usedcautiously in the assessmentof slabs.
Comparisonwas made betweenthe values of load transfer obtained acrossdifferent sites.

Lutterworth and Daventry showed similarities in response, with the only noticeable
variation being that the Daventry site containedsomelarger crack widths. Both sites were
of a similar age but contained different reinforcement, with Daventry containing steel
fibres and Lutterworth steel mesh. The site at Skelmersdale was much older and
contained some very poor load transfers that had required remedial work. This was
probably caused by the higher amount of load cycles it had withstood, causing
deterioration of aggregateinterlock.
Graphs of load transfer and load step were producedfor four sites under severaldifferent
load magnitudes.The results show that the foundation stiffness had little effect on the
relationship in comparison to load magnitude. The trends for each plot were very similar
with a linear section in the 40 to 100% load transfer region followed by a steep increase
between0 and 40%. This signifies a level of load transfer exists which should be retained
to prevent problemsoccurring with floor serviceability.
167
Edge cantilever was comparedagainst load transfer and load step to establish if it could
be determinedwithout the need for direct measurement.The majority of the sites showed
reasonablecorrelation, with an increasein load transfer, or reduction in step,reducingthe

amount of cantilever. This confirms that the load transfer of the joint is important in
controlling overall slab behaviour and floor flatness under dynamic load. Where the slab
was thought to be voided, a much greater scatterin results was found, with the void itself
increasingthe amountof edgedeflection.
The correlation between load transfer of the FWD and Prima is good, with equal values
found for both devices. When examining load step, the correlation is satisfactory,
although lower loads provide greater step than normalisation of a larger load would
predict. The variation is causedby inconsistenciesunder low loading createdby subbase
support and free slip across the crack face. Once these effects are overcome, the slab
structure behavesas expectedwith proportional responsein relation to the magnitude of
load.
6.4 Joint Deterioration
An extensive experimental programme was undertaken to investigate the behaviour of

joints and cracksunder dynamic load. A total of 82,1 OOxI OOx4OOmm
beam sampleswere
tested from 25 concrete mixes to examine 39 different variables, namely: crack width,
aggregate size, steel fibre quantity, steel fibre type, mesh reinforcement, traditional
reinforcement, load magnitude,and crack orientation. The detail and logic behind the test
set-up is described in Chapter 5, and it consisted of a double cracked beam specimen
cyclically loaded for 250,000 cycles at a rate of 2Hz. Measurements of vertical and
horizontal displacement were recorded every 600 cycles enabling the effect of
deterioration to be recorded.
This chapter discussesthe results of the tests and analysis the key variables, with the aim
of improving the understanding of joint behaviour and effectiveness when placed under
repetitive load. Appendix B contains a summaryof all the tests undertakenduring testing,
with the key phases (described in section 6.4.3) identified by cycle number and
differential displacementmagnitude.
168
6.4.1 Specimen Production
Over the course of the research82 specimenswere tested, all of which were cast and
prepared in accordancewith the methodology supplied in section 5.4. In each mix, six
beams were cast, together with two 100xlOOxlOOcubes to monitor the
lOOxlOOx4OO
compressive strength. This was designed to have a target mean strength of 45MPa at 28
days (section 5.4.2), with the results obtained from testing provided in Table 6.8. Each
beam sample was given a specific code to enable identification throughout testing. This
was in the form:
Concrete strength (MPa) / Reinforcement type / Reinforcement volume or diameter

(kg/m) or (mm) / Crack Orientation / Crack width (mm) / Load magnitude(kN).
Table 6.8 - Concrete Compressive Strengths
Mix No Date Cast Specimen Codes Cube 1 Cube 2 Average

(MPa) (MPa) (MPa)
1 27/11/2002 40/Non/ON/0.66/4 56.33 55.01 56
40/Non/ON/01.98/4
2 27/11/2002 40/65-60/20N/1.98/4 55.44 55.29 55
40/65-60/20N/3.3/4
40/65-60/20N/4.62/4
04/11/2002 CANCELLED 0 0 0
4 04/11/2002 CANCELLED 0 0 0
5 06/11/2002 40/65-60/30`N/4.62/4 51.76 51.43 52
6 12/11/2002 40/65-60/20N/3.96/4 52.76 55.33 54
7 12/11/2002 40/65-60/30N/1.98/4 50.25 50.04 50
40/65-60/20N/3.3/4
8 14/11/2002 REJECTED 44.11 45.3 45
9 14/11/2002 40/65-60/40N/4.62/4 44.91 47.21 46
40/65-60/40N/3.3/4
40/65-60/40N/1.98/4
10 21/01/2003 CANCELLED 0 0 0
11 21/01/2003 REJECTED 41.23 1 40.35 41
12 03/02/2003 40/80-60/30N/1.98/4 45.43 43.81 45
40/80-60/30N/4.62/4
13 10/02/2003 40/45-50/30/V/4.62/4 43.98 46.82 45
40/45-50/30N/1.98/4
40/45-50/30N/3.3/4
14 25/02/2003 40/65-60/30N/0.66/6 47.14 43.89 46
40/65-60/30N/1.98/6
15 10/03/2003 40/65-60/30N/1.98/2 43.11 47.10 45
40/65-60/30`N/3.3/2
40/65-60/30N/4.62/2
16 17/03/2003 REJECTED 42.05 43.66 43
17 01/04/2003 40/65-60/30/Par/0.5/2 46.55 47.40 47
15/04/200340/65-
169
60/30/Par/1.98/2
18 15/04/2003 40/65-60/20`N/0.66/4 44.79 45.71 45
40/65-60/20/Par/0.66/4
40/65-60/20/Par/0.9/4
40/65-60/20/Par/l/4
40/65-60/20/Par/1.3/4
19 29/04/2003 REJECrED 42.54 45.18 43.86
20 06/05/2003 40/80-60/30N/3.3/4 45.18 44.57 44.88
40/80-60/30N/4.62/4
21 07/05/2003 40/Mesh/7N/1.98/4 46.72 44.53 45.63
40/Mesh/7N/3.3/4
40/Mesh/7N/4.62/4
22 08/05/2003 40/65-60/30N/3.3/4 45.02 45.97 45.50
40/65-60/30/Par/0.66/4
40/65-60/30/Par/1.98/4
23 12/05/2003 40/65-60/40/Par/0.66/4 44.69 45.23 44.96
40/65-60/40/Par/0.75/4
40/65-60/40/Par/l/4
40/65-60/40/Par/1.5/4
24 13/05/2003 40/45-50/30N/3.3/4 43.7 48.09 45.90
40/45-50/30N/4.62/4
25 07/07/2003 40/65-60/40N/5.94/4 44.97 49.35 47
26 07/07/2003 40/65-60/30N/5.94/4 44.15 48.02 46
RI 17/12/2002 40/Trad/7N/1.98/4 52.37 51.96 52.17
40/Trad/7N/4.62/4
40/Trad/7N/3.3/4
MI 20/11/2002 1 Mor/Non/30N/0.66/4 29.45 32.21 30.83
The testing resulted in a variation of up to 14MPa between the mean cube compressive
strengths.Mix references8,11,16 and 19 failed to provide the target mean strength of

45MPa specified in section 5.4.2 and were therefore rejected and replaced. The mixes
constructed earlier in the testing period showed higher strengths that those cast later
(Figure 6.35). The changewas thought to have been causedby the moisture condition of
the coarseaggregateprior to testing. This was storedexternally and was thereforeopen to
climatic conditions affecting the moisture level. Although checked using the Speedy
moisture meter, and amendmentsmade to the material quantities, it was thought this may
still have affected the water/cementratio. This resulted in a decreasein strength during
the wetter months, and explains the shapeof the best fit line in Figure 6.35. However, the
magnitude of this variationwas not thoughtto haveany significanteffectson the cyclic

load test results.Mix 'Ml' was madeof mortarwith all aggregateparticlesbelow6mm.
As expectedthis gave a lower strengthof 31MPa,due to the natureof its constituent
materials.
170
Mixes referenced3,4 and 10 were repeatedas the specimenscontainedhoneycombingat
the comers due to the drynessof the mix and poor compaction. As the condition of the
beams could not be checked internally these were discarded and replaced to avoid any
discrepancyin results.
6.4.2 SpecimenQuantities
To ensurethat representativebehaviourwas obtained,at least two sampleswere testedfor
each variable. Ideally three or more tests would have been undertaken; however, the
duration of each (approximately 36 hours) forced a compromise. Initially a small
experimentalprogram was carried out to investigate the variability betweenresults. This

showed that usually two samplesprovided sufficiently similar displacements(below 150
microns) acrossboth the sameand opposingcracks within the samespecimen.A strategy
was developed whereby two beams were tested for each variable, with a third beam kept
in reserve,used only if the variation was abovecertain criterion. In the casewhere a third
beam was required for testing, the results of the two beamsmatching closest were used,
with the third disregarded.However, comparisonwas also made to the results of samples
looking at similar variables to ensure large discrepanciesdid not exist. A single beam
containing reinforcement bar (40/Trad/7N/3.3/4) was testedat the 3.3mm crack width as
negligible change was found between the 1.98 (40/Trad/7/V/1.98/4) and 4.62mm.
(40/Trad/7N/4.62/4) crack width. Similarly, only one specimenwas testedfor the 40/65-
60/20N/3.92/4 as it was an additional value to check the variation between the 3.3
(40/65-60/20N/3.3/4) and 4.62mm (40/65-60/20/V/4.62/4)crack width values.
Table 6.9 provides a comprehensivelist showing the number of beams tested for each
variable.
171
Table 6.9 - Beam Specimens Tested
Specimen Group 0.66mm 1.98mm 3.3mm

mm 4.62mm Other
None 2
30kg /in 3Fibre (Mortar) 2
20kg/ni 3Fibre 2 1x3.96mm
30kg /ml Fibre 2 3 2 2x5.94mm
40kg/m-'Fibre 2 2 2 2x5.94mm
30kg 1m3RC-80/60-BN Fibre 2 2 3
30kg IM3 RC-45/50-RN Fibre 2 3 3
Re-bar 2 1 2
Fabric 2 2 2
30kg /m., Fibre (2kN Load) 2 3
30kg /m3 Fibre (6kN Load) 3
20kg/M3Fibre (Parallel Crack) 2
30kg/m3 Fibre (Parallel Crack) 22
40kwm.3Fibre (Parallel Crack) 22
When testing across two separate cracks, as in this method of testing, there was inevitably
a difference in differential displacement measLirements between front to back Tj Vs Rl'
and between the two cracks on the sarne face T, Vs Fr' (locations shown in Figure 5.21 ).
This was due to variations in surface profile within a single crack, and between two
cracks. This was caused in particular by the distribution of aggregate particles and fibre
reinforcement. along with any minor eccentricities in loading. During testing the degree
of maximum variation between displacements in the same crack was 112 microns and
across different cracks was 261 microns. The average variation was 52 microns and
29microns respectively. To ensure the results were representative across all cracks a
150micron boundary was applied which no variation in differential Could exceed. This
was reduced to 100microns in conditions where the 250,000 differential dellection was
below 500 microns. This was chosen as any specimens showing values higher could be
assurned to contain sorne discrepancy in crack face condition. To obtain a representative

displacement for the specimen under consideration, the values across the same crack were
averaged, with this value again averaged with the value frorn the opposing face (equation
6.1).
I[ (F, + R, )/21 + Fj /2 equation 6.1
Calculating the value in the way reduced tile effect of any discrepancy between cracks
and provided a displacement closer to the characteristic value.
172
6.43 Deterioration Phenomena
Each beam was testedto 250,000 cycles, with specific specimenscontinued for a further
250,000 cycles to examine longer-term deterioration patterns. To enable comparison
between variables the differential displacementbetweenpositive and negative loading at
key points relating to the transition betweendeterioration phaseswas used (Figure 5.25).
Occasionally the beam was deemed to have failed prematurely, with load transfer
becoming negligible and deflections extremely large, limited only by the restraintsplaced
on the loading apparatus.A differential of 1.6min or abovewas found to sustainvery little

extra loading was therefore classed as the boundary for failure. In this situation the
number of cycles to failure provided some perception of strength and durability of the
load transfer system.
-6
Sixteen of the beamsreacheda point where they were deemedto have failed prematurely.
The displacementdifferential versuscycle degradationplots producedsimilar patterns,as
shown in Figure 6.36, with four distinct phases.
During phase1,rapid deterioration occursresulting in a steepgradient (figure 6.37). After
approximately 10,000 cycles this moves into phase II where a low magnitude increase,
linear degradationis observed.In tests causing specimenfailure phaseIII is identifiable,
whereby deflection acceleratesrapidly until reaching a magnitude of approximately

1.2mm. At this point the specimenentersphaseIV where failure is likely to occur within
the next 10,000 cycles. This shows similarities to the plots of Colley and Humphrey
(1967) and Thompson (2001), section 3.5, although in their research deterioration
occurred at different ratesdue to the constituentmaterials and reinforcementtypes.
The shapeof the overall plot is causedby the changing aggregateinterlock mechanisms
as the system degrades. During phase I the mortar deteriorates quickly, it being a
relatively weak material. Once this has transpired the increase in deflection slows due to
the greater strength and bonding of the larger aggregateparticles (phase II). As failure
commences in phase III the aggregatebegins to debond from the surrounding mortar,
cracks are initialised and the concrete face begins to degenerate. This increases the
stresseson the remaining particles creating further cracking until such a point that phase
IV is entered, whereby negligible load transfer is available through the aggregate
interlock mechanism.The influence of reinforcement such as steel fibres or fabric delays
the onset of the preceedingphase.The mortar deterioration still occurs as shown by the
steep gradient in phase I; however, the reinforcement transfers some of the load and
173
lowers the contact stress in the crack face. This reduces the rate of deterioration and
delays the onset of phase 111.Variations between fibres and fabric reinforcement are
found when phase III is entered as the fabric still retains its load transfer mechanism
whilst the fibres begin to deteriorate.
The differential deflection is a combination of movement in both the upward and

downward directions causedby a positive and negative loading pattern. In the majority of
cases(81%) the displacementin each direction of load was similar, with less than 70%
variation between initial and final displacements.However, a number of specimens
exhibited significant differences in deterioration, with greater displacement in one
direction comparedto the other (Figure 6.38). This was most likely causedby a variation
in restraint to movement from the surrounding material. In one direction two pieces of
hard aggregatebear upon one another, whereas in the other direction little support is
provided and displacement can occur (Figure 6.39). Load eccentricity may also have
createdthis effect, although this was thought unlikely due to the test configuration.
In 8% of the tests a reduction in displacementwas found in one direction, signifying a
greater resistanceover time (Figure 6.40). Two possible reasonsfor this are as follows:
the first is describedby Laible, White et al. (1977) whereby small material and dust falls
into the sockets from the degradedcrack face, thereby reducing the amount of free slip.
The other, as identified by White and Holley (1972) occurs when one section of aggregate
has overridden another and requires a greater force to push it back into its original
location due to the undulations and orientation of the crack face. The gradual reduction in
displacement, rather than a sudden step, as observedin Figure 6.40 suggeststhat in this
experimental programme small material accumulationis the dominant effect.
The examination of expelled material and specimencracking provided further evidence
surrounding the degradation process.In the majority of beams a crack propagatedfrom

the edge of the supporting shim to the baseof the saw cut and then back to the edgeof the
bottom, central encasementstrap. The crack itself was hairline at first, but could be seen
to open and shut very slightly under load. The depth into the face of the beam was
generally below 10mm and it therefore appearedto be a surface phenomenaassociated
with the stresses created through clamping. In specimenswhere differential displacement
was low, very little material fell from the crack; however, on completion of the test a
small layer of dust could be seendirectly under the On
specimens. those where movement
was higher (0-2mm or above) the amount of fine-grained material was much greater, but
all of the larger 10-20mm aggregateappearedto be intact. In specimenswhich were close
174
to failure, sectionsof concreteup to 20mm in size spalled from the baseof the beam on
either side of the crack, along with large amounts of cement dust. Loose large aggregate
particles of 10-20mmdiameter.which had becomedetachedfrom the surroundingmatrix,
could also be seen on the sides and top of the beam, and were only prevented from
ejection by the support of the surrounding material. Upon failure the entire surfaceof the
crack beganto break away, resulting in a large amountsof debris.
6.4.4 Influenceof Initial Crack Angle
The referencemix containing 30kg/m3of RC-65/60-BN steel fibre showed an increasein

differential deflection as the initial crack width was increased(Figure 6.41). The variation
between individual specimenswas acceptablewith only one specimenat a crack width of
3.3mm showing significant discrepancy.The reason for this error is unknown but could
be due to a lack of large aggregateor steel fibre at the crack face, causedby insufficient
mixing or separationunder vibration, the effects of which are exaggeratedwith a small

sized specimen. At around the 5 to 6mm surface width the joint resistance to cyclic
loading reduces rapidly, with failure (defined as greater than 1.6mm. differential
displacement)occurring at a width of 6mm.after 10-30,000cycles.
There is a distinct changein shapebetweenindividual cycle load/deflectionplots of small

(1.98) and large (5.94) crack widths (Figure 6.42). At the beginning of the narrower crack
width tests the cycle shows a smooth transfer of deflection as load is applied, following
the shapeof the load application sine wave. Toward the end of the test this has changed
slightly with a steeper gradient at the point where load transfers from a positive to
negative direction. This unrestricted movement is known as free slip and is detected when
the majority of deflection resistanceoccurs at the extremities of the deflection cycle; with
the transfer line being almost vertical. The larger crack widths contained much higher
levels of this free slip regardlessof when the cycle plot was taken and as such were more
prone to increaseddegradation.
This change in cycle geometry indicates that during early stagesof small crack width
cyclic loading, much of the deflection resistance is provided through micro roughness
friction, hence the smooth curve. As the mortar is wom away macro roughnessbecomes
dominant resulting in a much steepertransition betweenpositive and negative deflection.
The wider crack width accentuatesthis effect as the free movement under the aggregate
have become much greater. The inclusion of fibre reinforcement restrainssomeof the
will
load throughout the cycle, particularly when the crack width is small. As the width opens
175
the fibres tend to break and their effect becomesless influential, creating a much steeper
gradient as load is transferredfrom the positive to negativedirections.
The non fibre reinforced beamsshoweda much lower resistanceto differential deflection
than that of the referencefibre reinforced concrete mix, with failure occurring when the
surface crack width was less than 2mrn (Figure 6.41). The two beamstestedat the 2mm
crack width show good comparisonwith a variation below 50 microns. Prior to placement
in the rig the pre-cracking had caused the specimen to split into three sections as no
restraining reinforcementwas available. Upon loading, early degradationcould be seento

take place at a fast rate with large amounts of rubble falling from the crack surface,
accompaniedby high visual movementsof the central block.
At both the beginning and end of the tests the deflection cycle shapeindicates that most
of the resistanceto deflection occurs at the extremities of movement (Figure 6.43). The
rninimalresistanceto free slip shows that bearing rather than friction is the dominant
factor in load transfer at this stage. As crack width increasesthe distance between the
bearing surfaces becomes larger (Figure 3.4), as shown by the greater amount of
differential movement in the initial cycles. This deflection is then accentuatedover the
test period until aggregateinterlock is lost and failure produced.
The mortar beams containing steel fibre investigated the relative contribution of steel
fibres and the coarseaggregatein resisting displacement.Both specimensfailed early on
in the test period, despite being set at the smallest 0.66mm surfacecrack width, with the
increasein deterioration occurring extremely rapidly even though only a small amount of
rubble was created(Figure 6.41). The fibres bridging the crack were seento move freely
as the central block was loaded, and upon failure the majority had snappedrather than
pulled out of the mortar, with the large movementsthe probable cause.This failure mode
illustrates that the hook anchoredto the mortar extremely well, although the increasein
deflection over time, rather than a suddencatastrophicfailure indicates that some degree
of gradual pullout occurred.
The examinations of single cycle deflection at the beginning and end of the tests indicate
that although the majority of resistanceoccurs at peak deflection, somefrictional capacity
is provided (Figure 6.44). Restraint must therefore be provided from the fibre, which once
extended to its full length prevents continued movement. This pattern continues
throughout the test, although its capacity reduceswith the onset of failure.
176
6.4.5 Volume of Fibre Reinforcement
The increasein fibre volume from 20 to 40kg/n had little influence upon the magnitude
of differential displacementin surface crack widths below 3mm; however, above this
value increasing the fibre volume had a pronouncedbeneficial effect in delaying failure
(Figure 6.45). This was similar to the findings of Thompson (2001) who examined the
behaviour of fibre reinforced cement bound materials and found that increasingthe fibre
volume had little effect on decreasingdisplacements.
At the 0.66mm crack width the 20kg/ry mix experienced a higher differential
displacement than either the 30 or 40kg1m3 beams predict when a linear line is
extrapolated back to the zero axis. However, as the relationship at theselow crack widths
is unknown it may be that a non-linear function exists, with the 0.66 and 1.98mmvalues
being similar. In either situation the deflections at this level are so small as to make the
variation negligible regardless of fibre quantity.
The main effect of the steel fibres occurs as the surfacecrack widths become larger. At
the 3.3mm level all mixes producedsimilar behaviour,but when increasedto 4.62mm the
20kg/M3 beam failed within 50-113,000cycles. To evaluatethe behaviour betweenthese
points a single specimenwas tested at 3.96mm crack width. This confirmed the expected
trend displaying a large increase in differential displacementwhen compared to lower
crack width values, and that of the 4.62mm crack widths containing higher fibre
quantities. The contrast betweenthe 30 and 40kg/m3beamsonly becomesapparentwhen

the crack width reaches4.62mm, at which point the lower fibre quantity produces an
increase in displacement.This continues to the 5.94mm crack width at which point both
beams fail, with that of the 30kg/m3occurring much more quickly (10-30,000 compared
to 212,000 cycles). The two beams at 40kg/m3 show large variation between results;
however, one of the beamsfailed towards the end of the test with the other showing rapid
increasesin differential displacementgradient in the last 40,000 cycles (signifying phase
III behaviour and imminent failure). A further 20,000 cycles was applied to this beam at
IV was indicating
achieved, thatthe two valuesare in fact muchcloser
which point phase
to eachotherthanwouldappearon theplot.
Figure 6.46 illustrates the effect of increasing fibre quantity for each crack orientation.
The shapeof the plot showsthat adding fibres is most beneficial for larger widths, with a
delay in failure and reduction in displacement. Increasing the fibre volume with smaller
177
cracks had little effect; however, when compared with non-reinforced beams the
advantagesof adding even a small percentageis clear.
Examination of single cycle deflection displays a variation between shear slip of the
different specimens.The 20kg/m of fibre showed that in large crack widths most of the
resistance occurred at the extremities of displacement, indicative of a bearing type
restraint (Figure 6.47). During small width testsa reducedgradient was found in the early
stages of testing, signifying that friction was also assisting in the transfer of load. The
30kg/trO mix exhibited an improved transfer mechanismas a smoother curve was found
throughout the smaller widths; however, much of the restraint was still createdin bearing
for those beams with larger cracks approachingfailure (Figure 6.42). A fibre volume of
40kg/rn3 provided a reduction in shear slip regardless of width and time of cycle
measurement(Figure 6.48). Even when failure was approachedin the largest crack, the
curve still compared well to that of load application, indicating that friction was a
considerableaction in the restraint on movement.
The cause of this change in displacement resistancecan only be attributed to the steel
fibres, as no other variable was altered. Fibre contents of 30 and 40kg/M3 produced a
significantly greater restraint to movement, with the maximum differential controlled by

the aggregateinterlock and fibre pullout. The fibres can help in two ways: they can avert
degradationof the face by restraining crack growth; and secondly,the fibres will crossthe
crack and act as mini dowels transferring load. In low quantities theseeffects are reduced
as fewer fibres will have less bridging points thereby lesseningthe force required to cause
bending. There will also be fewer fibres to hold the crack face intact once micro-cracking
has developed.
6.4.6 Fibre Aspect Ratio
The aspectratio of a fibre is the ratio of length to diameter. Figure 6.49 indicates that an
increase in this value reducesthe differential displacementsignificantly at surface crack
widths above 3mm. The higher aspectratio producesa greater number of fibres in the n-fix
for a givendosage(asshownin Table6.10).This improvesthe numberof bridgingpoints
acrossthe crack enhancingresistance to deflection,as shown with the fibre quantity
comparison plot in Figure6.50.With a simple calculationof fibre volume it is possibleto
determinethat the useof theRC-80/60-BNresultsin 1.4timesasmanyfibres asthe RC-
65/60-13N,which contains1.4timesasmanyas the RL45/50-13N.In essencethis causes
the RC-80/60-BN to relatemorecloselyto the 40kg/m3
RC-65/60-BNbeam,which when
178
compared to the fibre quantity plot shows closer representationthan that of the 30kg/n
beam used as a reference.Similarly the RL45/50-BN is equivalent to a beam containing
closer to 20kg/m of fibre, which shows good comparison to the results of the 20kg/m'
values.
Table 6.10 - Typical Fibre Count across Crack Face
Fibre Type Fibre Volume Diameter Length Fibres/kg Fibres across

(kg/m) (mm) (mm) Crack Face
RC-65/60-BN 20 0.9 60 3200 31
RC-65/60-BN 30 0.9 60 3200 47
RC-65/60-BN 40 0.9 60 3200
RL-45/50-BN 30 1.05 50 2800 _62 34
RL-80/60-BN 0 0.75 60 4600 67
These observations might suggest that aspect ratio only influences differential
displacement through changing the fibre count across the crack, with the results from
Figure 6.50 showing similar values to that of the equivalent standard RC-65/60-13N.A
plot of fibre number against differential displacement (Figure 6.51) shows that as fibre
numbers increase the magnitude of differential displacementreduces significantly, with

the 4.62mm crack showing reductions in displacement from over I. Inim to below
0.15nun when the fibre count is increased across the crack face from 31 to 67. Fibre
length and diameter effects appear to be negligible, although the greater thickness may
show some slight increasein fatigue resistance.
Figure 6.52 clarifies the influence of aspect ratio on the reduction of differential
deflection with those containing higher values showing better performance. Specimens
larger cracks show the greatest increase in performance; however, even at smaller
with
the advantagesof using a high aspectratio fibre can be seen.
widths
Examination of the single cycle plots (Figure 6.53) illustrates that towards the end of the
is
high crack width teststhe cycle still smooth for those specimenscontaining 30kg/rr? of
the RC-80/60-BN fibre (Ud = 80), indicating that resistance from the fibre is increasing
the frictional component. The RL-45/50-BN fibre shows a different behaviour; towards
beginning of the test resistance to deflection provides good agreement to the load
the
plot, indicating energy absorption throughout the loading cycle from both
application
friction and fibre bending (Figure 6.54). Later cycles produce a much steepergradient as
the load changesdirection, with the majority of resistanceat the extremities of deflection.
179
This indicates that the frictional componenthas degradedover time with a bearing effect
becoming the dominant means of load transfer, similar to that found at a volume of
20kgIm3.
6.4.7 Load Magnitude
The application of a relatively low RN load causes little change in differential

displacementregardlessof surface crack width, with the overall pattern in an individual
test showing only minor degradation,and even somereduction in displacement,over time
(Figure 6.55). The reasonfor this can be assignedto the gradual accumulation of cement
dust and rubble in the aggregatesockets, which eventually builds up and reduces the
available free movement.Examination of individual loading cycles shows that even when
nearing the 250,000 cycle limit, the resistanceto displacementis sharedequally through
the entire cycle with friction accountingfor much of the energydissipation (Figure 6.56).
The 6kN load resulted in much faster degradation with only the 0.66mm surface crack
width able to resist failure up to 250,000 cycles. Regardlessof crack width, exanlination
of all individual load cycles show that at the beginning and end of the test resistanceto
movement at the transition between positive and negative loading is extremely low, with
the majority coming at the extremities of displacement (Figure 6.57). This signifies a
bearing effect with high load degrading the frictional capacity of the crack face early in
the test.
The summary plot in Figure 6.58 illustrates the effect load magnitude has on deflection,
with the higher levels showing large increases in deflection. With the exception of
extremely small crack widths the prevention of failure for a load application of UN is
negligible, with a reduction to RN producing only minor deflection.
These results contain similarities to the prediction of loannides and Korevesis (1990) who
stated that there is a particular level of stresswhere degradationoccurs, below which the
structure is relatively In
unaffected. the caseof the testing undertaken,this value is
somewhere between the 24kN range, which calculatesas a shear stress of 100 to
200MM-
6.4.8 Parallel Cracks
The effect of crack geometry in resisting differential deflection can clearly be seenwhen
comparing the 30kg/m3 'V' shaped crack against that of a similar surface sized parallel
180
crack. Figure 6.59 demonstratesthat the parallel crack width is highly influential in the
resistanceto deflection, with up to 40k 3
g/M of fibres unable to resist failure when over
1.2mm in size. A point exists near the 0.7mm crack where displacementincreasesgreatly,
with widths below this level showing only a gradual increasein movement. This agrees
well with the static tests of Pearson (1999) and ACI Committee 360 (2000), and the
fatigue test results of Colley and Humphrey (1967), who all show that levels of load
transfer reduce significantly at crack widths over 0.5mm (section 3.5).
The single load cycle plots illustrate that as most testscommence,particularly those with
smaller crack widths, deflection resistance produces a relatively smooth curve with
friction providing substantial restraint (Figure 6.60). For smaller crack widths this
continues for a reasonableperiod throughout the test; however, near the 250,000 lin-&
bearing begins to dominant. For larger crack widths both early on and nearing failure the
majority of deflection occurs at the extremities of deflection with bearing again being
most influential.
6.4.9 Reinforcement Type
The introduction of either a 7mm. rebar or section of A142 mesh (containing one
longitudinal and two transverse members) shows an improvement in resistance to
displacement when compared to those reinforced with steel fibres (Figure 6.61). The
increase in crack width has very little impact on the amount of displacement,with even
the 4.62mm tests resulting in values below 0.7mm. The difference betweenthe mesh and
rebar displacementis very small with mesh giving fractionally lower deflection at small
widths, this swapping over at the 3.3mm crack position. When comparing with steel
fibres the mesh and reinforcing bar shows reduced displacements to that of 20 and
30kg/m3. However, when examining 40kg/M3the variation is much smaller showing that
the inclusion of steel fibres above certain levels can produce similar effects in respectto
the prevention of deterioration.
Both types of reinforcement show almost identical single cycle load deflection curves
uninfluenced by cycle number or crack width (Figure 6.62). These are generally smooth,
with a gradual increase in deflection on application of load, demonstrating that

displacement is controlled by the deflection of the bar rather than bearing of the aggregate
in its socket. Friction may also assistin the resistance;however, the previous plots show
that at large crack widths this phenomenon will have reducedconsiderably.
181
6.4.10 Serviceability Limitations
As has been stated in section 6.4.3, the beams that failed providing no load transfer
comprisedfour main phases.PhaseIII was indicative of imminent failure (within the next
20,000) cycles, and could therefore be used to identify problematic cracks. If the
differential deflections could be contained below this level then it would appearthat the
joint could be classedas 'fully serviceable' (Figure 6.63). Table 6.11 provides a summary
of differential deflection values at which Phase III deterioration began to operate.

Unfortunately some of the specimensfailed early on during the test and therefore clear
identification betweenphaseswas not possible.
Table 6.11 - Differential Deflection at PhaseIII
Specimen Code PhaseIII Differential

Displacement (mm)
40/Non/ON/1.98/4 1.15
40/65-60/20/V/4.62/4 1.10
40/65-60/30N/5.94/4 0.80
40/65-60/40N/5.94/4 0.65
40/45-50/30/V/4.62/4 0.60
40/65-60/30N/1.98/6 0.80
40/65-60/40/Par/1.5/6 1.40
The table showsthat the onset of phaseIII deterioration is dependanton the reinforcement
type/quantity and the load magnitude. A lower bound level can be drawn at 0.6mm
differential displacement,at which point none of the specimenstestedwould show phase
III behaviour. This suggeststhat if differential deflection is kept within this limit (0.3mm
load step on site) then the joint will remain fully serviceable.
The literature in section 3.8.1 provided two values of load step within concrete slabs on
grade which would provide acceptable behaviour in respect to ride quality. However,
there is little information to confirm these values, and the variation in vehicle type using
the floor will be critical to the levels chosen. To examine the effect of reinforcement type
load magnitude on these serviceability requirements, both values have been used to
and
their applicability. The lower value was 0.1mm, with an upper level being 0.3mm.
assess
As the laboratory testing in this research has examined differential displacementscaused
loading in the positive and negative directions, the values obtained must be doubled
when
comparisons with site slabs, becoming 0.2 and 0.6mm respectively. In this
to produce
the 0.6mm allowable displacementis very close to the phaseIII deterioration change
case
182
point (described above). The 0.2mm differential displacement provides a far greater
safety factor against complete joint failure, whilst also ensuring vehicle load
stepslin-itationsare met. The 0.2mm (limit 'A') and the 0.6mm (limit 'B') have beenused
in the test results to indicate tolerable crack widths for serviceability.
Insertion of these serviceability limits onto the plots of crack width and differential
displacement(Figures 6.41/6.45/6.49/6.55/6.59/6.61)for the entire range of testing, made
it possible to determine acceptable values of surface crack width for a given
reinforcement or load magnitude. This information is provided in a summary form in

Table 6.12.
Table 6.12 - Allowable Crack Widths Preventing Serviceability Problems
Specimen Type Limit'A'(0.2mm) Limit IBI (0.6mm)

Crack Width (mm) Crack Width (mm)
Mortar 0.09 0.21
Parallel 30kg/m3 Fibre 0.38 0.9
Non 0.46 1.00
Parallel 20kg/M3 Fibre 0.80 1.08
Parallel 40kg/m3 Fibre 0.85 1.16
6kN Load 0.98 1.45
20kg/M3 Fibre 2.82 3.9
RL-45/50-BN Fibre 3.10 4.26
30kg/M3 Fibre 3.48 4.78
TO-Fk-g
/-m3ifib - 4.62 5.74
re
RC-80/60-BN Fibre 4.26 6.98
RN Load 8.2 > 10.00
Steel Fabric > 10.00 > 10.00
Steel Rebar > 10.00 > 10.00
These results show that there can be significant variation in allowable crack widths
between limit 'A' and Limit '13' requirements. The magnitude of load is highly
influential, increasing the allowable crack width for serviceability limit 'A' from 0.98 to
8.2mm with a reduction in load from 6kN to 2kN. Steel bar and fabric show excellent
displacement in to limits 'A' and V with extrapolation of the data
resistance to respect
indicating that 10mm. crack widths still produce displacements well below serviceability
limit W. The use of RC-80160-BNshows similar behaviour to that of mesh when widths
below 4.26mm; however, after this point an increase in displacementoccurs with the
are
limit 'B serviceability requirement being exceeded at a 6.98mm crack width. Steel fibres
type RC-65/60-BN also prevent the serviceability limits being reached until high
of
183
surface crack widths are produced, the effectivenessincreasing with the fibre quantity
used. Parallel cracks can only withstand both 'A' and 'B' limits for crack widths below
1.2mm, with fibre quantity having very little impact. Mortar specimens, although
containing steel fibres restrain very little displacement and show that the aggregate
interlock effect is still important regardlessof the reinforcementused.
6.4.11 Summary of Laboratory Investigation
The cyclic load testing of small-scalespecimenshas establishedthat deterioration can be
split into four main phases.PhaseI, consistsof fine material degradationand producesa
rapid increase in differential displacement. Phase II shows little deterioration of the
concrete face and differential displacementbecomeslinear. PhaseIII producesconcrete
face cracking and aggregate and fibre pullout, resulting in an increase in differential
displacement.PhaseIV is failure wherebythe face can no longer sustainany load.
'V' shapedcracks have been shown to be much more effective in transferring load and
reducing crack face degradationthan parallel cracks; however, surfacecrack width is still
highly influential to both. Steel fibre reinforcement provides significant increases in
restraint to degradation when compared to non-reinforced specimens,although it is a

combination of the fibres and aggregateinterlock that causethis effect. The fibre quantity
and type controls the amount of differential deflection restraint, with the number of fibres
crossing the crack being the major contributory factor. Specimenscontaining steel fabric
and reinforcing bars show almost no changein deflection throughout the 250,000 cycles
regardlessof the crack widths tested.This is due to the load carrying capacity of the bars
crossing the crack, preventing macro roughness degradation. The magnitude of load
controls the amount of differential displacementand the rate of deterioration.A RN load
shows relatively small changesin displacement throughout the duration of the test, with a
RN load showing rapid failure in larger crack widths.
Examination of the single cycle plots shows that friction from the local roughness
dominatesin the early stagesof testing,especiallyunder small crack widths. As the
stressesacross the crack are increasedthis degrades,the rate of which dependson load
magnitude, reinforcement type and crack width, with resistance then converting to
bearing of the macro roughness.In situations where failure is approachedthe aggregate
providing this bearing resistance, cracks and pulls out from the surrounding mortar
leading to reduced aggregateinterlock, increasing stress and further degradation.There
184
are no specific values where the changesin restraint occur, insteadresulting in a gradual
transfer.
Serviceability limits show the importance of each factor examined during testing in
providing resistanceto displacement.This shows that if crack widths can be prevented

from opening up excessively,or loads kept within certain criteria then the type or quantity
of reinforcement used is inconsequential.However, in normal circumstancescrack widths

open up to varying degrees and therefore the use of any reinforcement is highly
influential. As expected the greater the surface crack width measurement,the more
is
reinforcement required within the concrete to keep the displacementswithin tolerable
limits.
6.5 Summary of Field and Experimental Work
Examination of concrete slabs within the field combined with laboratory testing has
enabled a more complete picture to be developed in respectto crack and joint behaviour.
Typical crack geometries have been determined and display a 'V' shaped formation
caused by differential shrinkage. This indicates that surface measurementsalone cannot

be used in the assessmentof cracks, as those previously thought to be unsatisfactory
(greater than 2mm in width) can provide adequate load transfer as shown by the
deflection test results on site. The geometries found in the field have then been used
within the small-scale laboratory test procedure to provide an understanding of the

deterioration processes.
Links have been made between the effectivenessof the joint in transferring load, to other
deflection dependantresponsessuch as load step and edgecantilever. This has shown that
the joint is highly influential in the behaviour of the slab as a whole; both in resisting
failure mechanismsand ensuring serviceability requirement are met.
Reinforcement within concrete slabs on grade has been shown to influence both the single
load transfer behaviour, and the deteriorationrate of the concrete face. In all
cycle
the local roughness was found to deteriorate
early on in the cyclic loading tests,
situations
the global roughness for longer-term load transfer. The incorporation of
relying on
reinforcement delayed the attrition of global roughness and therefore retained much of the
load transfer.This also enabledserviceabilityrequirementsto be controlledfor given
In site conditionssteelmesh was found to be more beneficialin larger
crack widths.
of 20kg/m3ofsteelfibre reinforcement; larger
however, quantities
cracksthan quantities
185
of 40kg/m, or the use of a different fibre type, were found to produce siniflar results
within the laboratory.
The variation in foundation condition under the slabstestedin the field produceddifferent
deflection responses. This confirmed that edge support cannot be relied upon and
therefore the laboratory testing was sufficiently accuratein producing a lower bound of
displacement degradation. In both laboratory and field work, load was found to have a
highly influential effect in respect to both deflection magnitude and crack face
deterioration, and therefore needs to be carefully assessedprior to slab construction to
ensureload transfer mechanismsare satisfactory.
The following Chapter utilises the information gathered in the small scale testing of
laboratory specimensto develop a representativeload transfer mechanismwithin a finite
element model. This is then used to model the sites investigated during field testing
enabling verification of the method and further analysis of the load transfer effect.
186
Chainago (M)
20 30 40 50 60 70 30 90 100 110 120 130 140 1 160 1170 180 190 200
0111 10 -i
IIII
- -IIII
tII IIII i I
I
0-
M, ............ Lp .... ......9 IQ..........
........... .7...... ----------------
10-
20- 1 013 XX x EZ
30- --------- -- -------
E 0C xx
40- 0
Cn X5 .1 ur
ZO
O 17 1: 0
50- xx * 9: Ozo
60- -- -------- ---- ------ ---I., - ----- ----

:1.. ..... ...... .......
70- ......................................
30 xx Ol x1t xx IRA
so- l6 it "*Z+
t OIS /
go-
F- LoadingDocks /
LoadingDocM
Batterycharge
area
slab Vkhness& Dynamic P(wis test

pas ol 100mm diameter cores for cement content, watericementratio, density. compressive stieXth.
Dramic Probe TOM (some to be located over )./sarvd cracks)
100 mm diameter core for wals Mrjmass and
Coresthroughjoint and cracks
Figure 6.1 Lutterworth Core Locations (Top - visit 1, Bottom - visit 2)

-
187
-
On
I
at
I from top left, Core 7, Core 17, Core 24,

Figure 6.2 - Core taken during visit (Clockwise
Core 2 1)
90 Core 21 (Part Depth Crack)

-a- Core 17 (Full Depth Crack)
-Core 7 (Saw-cut Joint)
X- - Core 24 - Saw-cut Joint with Dowel)
7 X.
.... .....
2
-W ---- Alll--
100 150 2DO 250

0 50
Core Depth (MM)
Figure 6.3 Crack profiles frorn cores taken in visit I

-
188
1.6
--*--Core I
-4- Core3
1.4
-Core 5
Core6
Core10
1.2 O.;N
%
1
E
Ir
e os
Lo
0.6
0.4
S
0.2
S
k
0+
50 100 150 200 250
Core Depth (mm)
1.6
-*--Core 16
- -+ - Core 17
1.4- Core 24
......
---Core 27
-4- Core 28
1.2
08
06
0.4
0.2
0 100 150 200 250

60
Core Depth (mm)
Figure 6.4 - Crack profiles from cores taken in visit 2
189
30 30
-offset 11 -Offset I
--, --Offset2l 20 ---Offset 2
20
10 10
E E
E
o o
60 80 100 120 140 160 180 200 4 60 80 180 200
0.10
-20 -20
-30 -30
Core Depth (mm) Core Depth(mm)
30 ou
20 20
10 -10
E
E E
E
0
2; 100 120 140 160 180
80 0-10 0
0-10
-20
-20
-30
-30 Core Depth (mm) Core Depth (mm)
Figure 6.5 - Crack offsets from cores taken in visit I
TES 10Str pup S.- rau.;. w

TEAR 53 Sti- P. If
AM dMfotS"$ i.
C3 D. A. SAW
..
Dm rp la
A. g-g-
p... i _Ma mm 1.0
hi.
kiaL
Ir-
7-5-
10
=4
lb. -too. pit =1 d"k
-W, 0 Sbb
G
3t
it no-
275,325 383 d 795-
f-ilwom, wotm -ICS. it Sb4m
-IC4,1.
In-
Ic V.to Sbgndg
Figure 6.6a - Strain GaugePositions (Leeds)
190
..- . - . - .I
12
Figure 6.6b - Strain GaugePositions (Northampton)
;tWE a7 E l
[N 1
H TEAB ylic m stminsauge H g
. H; 7- t. 4 nua
b-, ir-Mia m sin; laute fixed . (iomm 131311
- 181s
eoP-Meter alt singleUmge dmblc gauge 10 53
b--9-4 NIdir.. edTE. 910joi. t gauge
An dinxuios In mmes
1 40200 1 40200 i 48240 1
1 'f3l
31
S3I 134
135
136
a356? a1 10 13 is 16 17 is Is 20 22 24 26 27 28 29 30
H HH H HH H I-
12 14 2t 33 23
40 12 01 13-1.2 14 01 4011,0 1 1.0 140 12-IJI 13 1 1201 4.0 1 10 1 10 140 120117.121 00
Figure 6.6c - Strain GaugePositions (Marston)
191
7
--, *--Gauge 19
-m- Gauge 20
6 -Gauge 21
Gauge 22
Gauge 23
--*-Gauge 25
5 - -+ - Gauge 27
...... Gauge 28
-- Gauge 31
4 4- Gauge 34
E Gauge 36
E
Le
50 100 iso 250
Distance from Slab Surface (mm)
Figure 6.7a - Crack width extrapolation (Leeds)
7
Gauge2
Gauge12
6 -Gauge23
- Gauge31
-0 - Gauge32
m Gauge31 (Amm)
5
4
E
E
I-
-------------
U
C
C.)
2
0
50 100 ISO 200 250
Figure 6.7b - Crack width extrapolation (Marston)
192
1
--*--Gauge I
Gauge 2
4
E
E
-----------------------------m-----------------------
0
50 100 150 200 250
-1
Figure 6.7c - Crack width extrapolation(Northampton)
1.4
22 Actual Surface Measurement
--A - 27 Actual Surface Measurement
1.2
-28 Actual Surface Measurement
4- 34 Actual Surface Measurement
-0 - 36 Actual Surface Measurement
1
0.8
. 111.
",
0.6
02 _4 - .. **-, *-"-**-
---,
12
0
50 100 150
-0.2
-04
Figure 6.8 - Actual crack profile using embeddedstain gaugesand surfacemeasurements

(Leeds)
193
10
1
DistwKe Irm Surroce-mm
Figure 6.9 - Differential shrinkageof concreteopen to the environmenton one side

(Neville 1995)
.J. L
2.1
10
D.0
D1234567891011121314
Elapsed jfme, t (days)
Figure 6.10 - Erroneousstrain gaugereading from Bishop (2001)
194
F-d ft-mb" 1.9.4
7 2 3
O
H T1310Sl"'4S"P 4
14 TUB SJSMW.Pp =ajg-
EM
a D-IM.
111700 -
tf-
4?-
MD
"1 1.4.
aw , l On "w""h
,". , A", ie'
-1 --0
TM
Ow
I
..... .....
.
.... ..... .... .... ..
,I
m. 84
$790
$400 A
JAW J I
71MI 73-00-J ILM 75M 77-
. - -1
"Ir ir
Figure 6.11 - Area of floor level surveyedwith a preciselevel (Daventry)
iim
iaim
9im
aim
? im
EIM
sim
sjm
Mm
zm
ilm
Oim
-UM
-zim
Gjm
-sjm
fim
eim
lim
Figure 6.12 - Floor surfaceprofiles (Daventry)
195
'E
E
C
0
U)
0 200 400 600

. 600 -400 . 200
Distance from Joint (mm)
Figure 6.13 - Measurededgecurl using a builders level (Ballymena)
................. -------------
.................
'A
04-
50 100 150 200 250 300
Distancefrom Slab Surface(mm)
Figure 6.14 Predicted crack profile using slab levels and surfacewidth (Daventry)
-
196
9
6-
ol
0 so 100 150 200 250 300
Figure 6.15 - Predictedcrack profiles using edgecurl (Ballymena)
12C
100
Fe - 0.9365
80
60
E
r.
IL
40
20
.
0-1.
0 20 40 60 80 100 120
5OkN FWD Load Transfer (%)
Figure 6.16 - Load transfer comparisonbetweenFWD and Prima (Daventry)
197
120
100
so
00
60 -
bo 0, ,
0001
40 EqualValueUne 001
6 0010
WOO,
"0"
00001
20
oi(- TrencHine
001
0001
000
0
0 20 40 60 80 100 120
BUN Load Transfer (%)
Figure 6.17 - Load transfercomparisonbetweenFWD loads (Lutterworth)
12C
x
100
x0
00
x0
80 X, 0
**0
**
00
60
EqualValueUne 0
20
TrencOine 0 Sawn
o C.J.
x Crack
0
0 20 40 60 so 100 120
SUN FWD Load Transfer (%)
Figure6.18- LoadtransfercomparisonbetweenFWD andPrima(Lutterworth)
198
Slab I Slab 2
Loaded
Slab Unoaded
Slab
Figure 6.19 - Effect of free-slip on load transfer
120
100
LLI
20
oI
0 20 40 60 80 100 120
Crovettl & Darter Lcmd Transfer (%)
Figure 6.20 - Comparisonof load transfer equations
199
is
14
13
12
11
10 Trendline
4
LowerBound
3 *00
*# #4
2 40
ComerPoints
Construction
Joints 0
1
0
0 20 40 60 80 100 120
5OkN FWD Load Transfer (%)
Figure 6.21a- Comparisonof load transfer and crack width (Daventry)
15
14 sav I
4CJ,
13 x Crack]
12
11
10
3
0#
2 x0
6*6**04**t
460
1 00 0 o*I, 06 *x )6
a
0 20 40 60 80 100 120
58kN FWD Load Transfer (%)
of loadtransferandcrackwidth (Lutterworth)
Figure6.21b- Comparison
200
is
14
.
13
12
11
10 .
9
4
.
3
0
0 20 40 so 80 100 120
lOkN Prima Load Transfer
Figure 6.21c - Comparisonof load transfer and crack width (Skelmersdale)
1.6
Top Crack Measurement
Bottom Crack Measurement 0

1.4
0
0
1.2
E
r
0.8
06
04 0
0.2 0 0'.
0
0 20 40 60 80 100 120
58kN Load Transfer (%)
Figure 6.22 - Comparisonof load transfer with surfaceand baseslab crack widths
(Lutterworth)
201
500
450
400
y. 02SIx
350
R2 a0 9477
0
300
CL
'0 250
3
E 200 Trendine
r
150 Loadratioline
100
50
0
0 200 400 600 800 1000 1200 1400
SOM FWD Load Step (microns)
Figure 6.23 - Load stepcomparisonbetweenFWD and Prima (Daventry)
250-
R' 0.9147
Trendline
200-
0
13
150.
100- Loadratioline
50.
0 50 150 250 350 450
85kN FWDLoad Step (microns)
Figure 6.24 - Load step comparisonbetweenFWD load magnitudes(Lutterworth)
202
# Sawn
oc. i.
XCrack
250-
200
CL 150
Trenchine
E 100-
Fe- 0.4097 Loadratjofine
so-
#* () 0 40,0
0
0 50 iso 250 360 450
SUN FWD Load Step (microns)
Figure 6.25 - Load stepcomparisonbetweenFWD and Prima (Lutterworth)
Area Affected by
Voiding and Free
Actual Curve
Load
Step
Approximated Gradient
Equal Ratio Line
Load Magnitude
Figure 6.26 - Effect of voiding on load magnitude/ load step gradients
203
700
600
500
400
300
200
100
0
0 10 20 30 40 so 60 70 so 90 100
Load Applied (kN)
Figure 6.27 - Effect of load on absolutedeflection
3000
2500
2000
1500
1000
500
o1
0 20 40 60 so 100 120
Load Transfer (%)
Figure 6.28 - Comparisonof load transfer and load step

(DaventryALutterworth/Ballymena/Skelmersdale)
204
le
14
13
12
.
11 .
.
10
.
.
.
7 " ".
"
6
"
5 " "
..
""
4
3 ""
1 * ""
0 ""
0 200 400 600 800 1000 1200 1400
5OkN FWD Load Step (microns)
Figure 6.29a- Comparisonof load step and crack width (Daventry)
15
14
13
12
11
10
4
00 # 1* 0
3 0 0*
2 0 Sawn
of o CA
1 OONO olbo X Crack
wxI
0
0 200 400 6DO 800 1000 1200 1400
Sam FWD Load Stop (rrJcrons)
Figure6.29b- Comparisonof loadstepandcrackwidth (Lutterworth)
205
is
14
13
12
11
10
4
0
6
3
0
0 50 100 150 200 250 300 350 400 450 500
Load Step (microns))
Figure 6.29c - Comparisonof load stepand crack width (Skelmersdale)
15
14
13
12
R' -0 6203
10 Run 3 Run 2
Ex.....
E9F? -0 6183
r 8R0.4796
7x Run I
6- #
5x
4-Xxs,
x
3-
2- *Run I
x Run 2
10 0 #0 Run 3
*
01
0 100 200 300 4DO 500 600 700 800 900 1000 1100 1200 1300 1400 1500
5OkN FWD Load Step (microns)
Figure 6.30 - Comparisonof load stepand crack width taking into accountpre-load
(Daventry)
206
50
0
20 40 60 80 100
C
-50
3 -100 R' - 0.4857
-150
I*
-2DO
SUN FWD Load Transfer (%)
Figure 6.31a- Comparisonof load transfer and edgecantilever (Daventry)
50
0 mxM.
20 40 60 080 iod,
0
r- xo 66
00 ab x
c
. 50
x
-100
-150
0 Sawn
0 C.J.
x crack
ck
-200
SOM FWD Load Transfer (%)
Figure 6.3Ib - Comparisonof load transfer and edgecantilever (Lutterworth)
207
20
10
40
20 40 60
R'- 0.5435
-10
-20
-30
-40
I OkN Prima Load Transfer(%)
Figure 6.3Ic - Comparisonof load transfer and edgecantilever (Ballymena)
20
10
C
0
C,
0
20 40 60, *1 oo
10
r
R' -0 2681
-10
-20
-30
-40
IOM Prime Load Transfer (%)
Figure6.31d- Comparisonof loadtransferandedgecantilever(Skelmersdale)
208
50
0
100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 11 DO
"50
5
I
.100
Z
-150
z0
In
R2 0.672
-200
-250
50kN FWD Load Step (microns)
Figure 6.32a- Comparisonof load stepand edgecantilever (Daventry)
so
0
A0 200 300 400 Soo 600 700 800 900 1000 1100 1200 1300 1400
100
C,
00
4x 60
*
40, x
. 100
-150
* Sawn
* C.J.
* Crack
-200 FWD Load Step (microns)
Figure 6.32b Comparisonof load step and edge cantilever (Lutterworth)

-
209
20
10I* *
C
0
C
50 100 150 200 250 300 350 400 450

c fA
.
R' 0 6662
r=
= 20
9L
z
-30
-40
IOkN Prima Load Step (microns)
Figure 6.32c - Comparisonof load stepand edgecantilever (Ballymena)
20
10
I*
a
100 150 200 250 300 350 400 450

c #50
'10
E
= -20
9L
-30
-40
lOkN Prima Load Stop (microns)
Figure 6.32d - Comparisonof load stepand edgecantilever (Skelmersdale)
210
1) Soo 1000 2! )0
-5)0
-3
0
C
0 -500
C)
E
C
0
C) -40
0
-900-
z
0
. 1100-
-1300-
5OG-
Distance from Crack (mm)
Figure 6.33 - Typical deflection bowls (Daventry)
-------- 00-
500 2000 2E30

1)
-5)0
-100--
S
C
0
E
C
8
0-
U,
-1300-
Distancefrom Crack (mm)
Figure 6.34 - Typical deflection bowls (Lutterworth)
211
60
so
T
40
30
CL
E
0
C.)
r 20
10
0
01/09/2002 21/10/2002 10/12/2002 29/OM003 20/032003 09/05/2003 2a'06/2003 1710812003
Date Concrete Cast
Figure 6.35 - Effect of casting date on cube compressivestrength
1.8
Simplified Model
1.6
Phase IV
1.4
Dif
for
en 1.2
tia PhaseI
I
Do
fie I
ctl
on
(m 0.8 PhaseIII
M) L
Phase11
0.6- Typical Plol
0.4-
0.2-
nj -4
0 5DDOO 100000 15DOOO 200000 250000
No of Cycles
Figure6.36- Deteriorationphasesof a concretecrack
212
05
045
04
035
E
E
03
E
025
02
015
0.1
005
0 20000 30000 40000 50000 60DOO 70000 80000 90000 100000

10000
No of Cycles
Figure 6.37 - Early deterioration in a concrete test specimen
016
014
012 Rapid Increase in Displacement
01
E
Upward Movement
E 008 --E- Downward Movement
006
004
No Incl, ls, M Dilpl,, 11,, t
002
0
0 50000 100000 150000 200000 250000
No of Cycles
Figure 6.38 - Unequal displacement
213
SoftMortar
Hard Aggregate
Hard Aggrega
Figure 6.39 - Influence of aggregateon displacement
0.06
0.05
E 004
E
E Area of Decreasing Displacment
0.03
a
m
002
0.01
150000 200000 250000 300000

50000 100000
No of Cycles
Figure 6.40 - Negative deterioration causedby small material accumulation
214
2
1.0 1,500 72.000 30.000

20,0DO 185.000 10,500
1.0
1.4 30kgVrn3Fibre
Mortar
1.2
Non Reinforced
Concrete
R'- I
30kg/rn3Fibre
0.8 Concrete
X0
8733
0.6
0.4 A
02

0
01345 7
InitialSurfaceCrackWidth(mm)
Figure 6.41 - Effect of aggregateand steel fibres on differential displacement
c 0.1 02 0.3 0.4 05 06
-0.02
-0.04
-0.06
. 0.08
Early Cycle
d
-0.12
LateCycle
-1.
. 0.14
-0.16 Ilk
'o.4,.a v .1
-0.18 04111.4,
-0.2
(SOCS)
Time
Figure 6.42a - Effect of deterioration on displacement resistance (30kg/m3 steeII f bre with
1.98mmcrack width)
215
I
O.O. O.
CF*40.0-0.0-9-0-0-0.0.0. O.S.
0.8 0.9.
I::
'G.
Late Cycle
06
0.4
E Earty Cycle
a 0.2
is
0
0.1 032 03 04
-02
-0.4
Time (Sece)
Figure 6.42b - Effect of deterioration on displacementresistance(30kg/msteel fibre

with 5.94mm crack width)
0.2
0.15
0.1
Early Cycle
0.05
o
E 0.2 0.3 0.4 0.5
I
E -0.05
-0.1
-0.15
Late Cycle
-0.2
0
-025
-0.3
Time (Sacs)
Figure 6.43a- Effect of deterioration on displacementresistance(non-reinforcedwith

0.66mm crack width)
216
1.4
1.2
9r. e. 0.0 . 0-0-0 .0 -0 -0 -0 -0 .0 -0
1
Late Cycle
08
E 0.6
E 0.4
Eady Cycle
14
0.2
40
0
: 0.1 2 0.3 0.4 05
-0.2 1
0*
G"414v a. &. Q6
-0.4 *., w
-0.6
Time (Seca)
Figure 6.43b - Effect of deterioration on displacementresistance(non-reinforcedwith

1.98mmcrack width)
08
90
06 LateCycle
04
EarlyCycle
E
E
: - 0.2
r
0
E
0
02 0.4 0.5
. 0.2
-04
-0.6
-0.8
Tim* (Secs)
Figure 6.44 - Effect of deterioration on displacementresistance(mortar with 0.66mm

crack width)
217
2
1.8
E 1.4,
12-
P 08
06
04
0.21
0
34
Initial Surface Crack Width (rnm)
Figure 6.45 - Effect of steelfibre quantity on differential displacement
I 6mm Surface Crack

tB.................. 13
1.4
1.2
4 6mm Surface Crack
0.8
I 2mmSuiiaceCrack
C,
L
oe
3.3mm Surface Crack
0.4
0.2 0.6mmSurfaceCrack
04
10 20 30 40 50
Fibre Quantity (ki;ilnd)
Figure 6.46 - Effectivenessof fibre quantity in resistanceto displacement
218
0z
018
e le.9.0- 0- 0.0.0. , 0.0- d>l> 0-9-lk
,
0.16 Late Cycle
0.14-
0.12 -
E
0.1 -
dr
P. OL
CL S.
0.08 >
006
Early Cycle
0.04.
0.02-
0-
0 01 0.2 0.3 0.4 0.5 06
Time (Secs)
Figure 6.47a- Effect of deteriorationon displacementresistance(20k steel fibre with

0.66mm crack width)
ol
06
0 Late Cycle
0.4
EarlyCycle
0.2
E
E
r
E0
0.1 02 0.3 0.4 os
C,
-0.2
-0.4
40"
-06
-08 Time (Secs)
Figure 6.47b - Effect of deteriorationon displacementresistance(20kg/m3Steelfibre

219
0
0.1 0.2 0.3 0.4 05
-0.02
-0.04
0.06
Earty Cycle
E -0-08
-0.1
-0.12
I'd
LateCycle
A"Ir
-0.14
-0.16
Time (Secs)
Figure 6.48a- Effect of deterioration on displacementresistance(40kglm steel fibre with

1.98mmcrack width)
1.2
1
0.0 -9 -0 -9 -0 -9 -0 -9-0-0
08
Late Cycle
0.6
0.4
0.2
ycle
0
0.1 0.2 .0.0. 0.4 os
g,.0.003
.e.
-0.2
Time (Secs)
Figure 6.48b Effect of deterioration on displacementresistance(40kg/m3steel fibre

-
220
2
1.8
&/ 30 000
10,500 + 245 000 10:500
,
1.6
1.4
12
1
Aspect Ratio (48) Aspect Ratio (67)
108
Fe - 0.2218
06
04
Aspect Ratio (80)
R'. 0.1922
0.2 ZF-
0
012345 7
Initial Surface Crack Width (mm)
Figure 6.49 - Effect of steel fibre aspectratio on differential displacement
2
1.8
0 wimin 0 mimme 0 momin 0 wimme 0 momm 0 mimme a views 'Al miminii a IN
1.4
12 3OkWm3Fibre
20kQIM3Fibre
40kgtm
08
.2
06
* Aspect Ratio (80)
0.4 * Aspect Ratio (67)
* Aspect Ratio (48) x
02
A
A
0
023457
Figure 6.50 - Comparisonbetweensteel fibre aspectratios and steel fibre quantities
221
2
1.8
1.6
1.4.
1.2-
0.8
0.6-
04
02
0
0 10 20 30 40 50 60 70
Number of Fibres across CMs Crack Face
Figure 6.51 - Effect of crack face fibre count on resistanceto deflection
1.8
1.6
1.4
1.2
0.8
0.6
x4 6mm Surface Crack
3.3mm Surface Crack
04
x
0.2 0
2mm Surface Crack
0
10 20 30 40 50 60 70
so go
Fibre Aspect ratio (Ud)
Figure 6.52 - Effectivenessof aspectratio in resistanceto displacement
222
0.16-
0.14-
LateCycle
.0,
0.12
.ip.
-E
E 008
0.06
0.04
0.02
0
0 01 0.2 0.3 04 05 06
Time (Secs)
Figure 6.53a- Effect of deteriorationon displacementresistance(steel fibre of aspect
ratio 80 with 1.98mmcrack width)
0.01
0 OE
or 11's
eL
0.04
002
0
0.1 0.2 0.3 0.4 0.5
E
-0.02
E
. 0.04
Early Cycle
. 006
-008
Late Cycle
-0.1
-0.12
-0.14
Time (Secs)

ratio 80 with 4.62mm crack width)
223
0.35
0.3
0.0- e. g. * 0.0.
wa,
0.25
6 Late Cycle
0.2
E
0.15
Early Cycle
E
0.1
005
0
0.1 0.2 0.3 0.4 9% 06
-0.05
o. W
0- 0 -e- 9.9.0.0.0 e. u. 00 . 0*4r
-0.1
Time (Secs)
Figure 6.54a- Effect of deteriorationon displacementresistance(steel fibre of aspect

ratio 48 with 1.98mm.crack width)
0.9
0.8
0.7
0.6
Late Cycle
E
05
E
8
.90.4
CL
0.3
0.2
EarlyCycle
0.1
0
0 0.1 0.2 0.3 0.4 0.5
08
Time (Secs)
Figure6.54b- Effect of deteriorationon displacement fibre

(steel
resistance of aspect
ratio 48 with 4.62mmcrackwidth)
224
2
1.8
1.6
1.4
1.2-
0.8
0.6
0.4
0.2
0
0 347
Initial Surface Crock Width (rnm)
Figure 6.55 - Effect of load magnitudeon differential displacement
0.05
0045
Early Cycle
0.04
i
gr
0035 Ar
0.03
IV
0,. w
41,40-0-0.4volo.
EE 0.025
Late Cycle
0.02
0.015
0.01
0.005
0
0.1 0.2 0.3 0.4 0.5
0 06
Time (Secs)
Figure 6.56a - Effect of deterioration on displacementresistance(load magnitudeof 2kN

with 1.98mmcrack width)
225
-o
-o
E
E
E-
0.
S
0
-0
.0 Time (Secs)
Figure 6.56b - Effect of deteriorationon displacementresistance(load magnitudeof 2kN

0.02
0
0.1 OA 0.3 0.4 0.5 06
-0.02
LateCycle
-O.D4
E
E
E -0-06
. 008
A`l"EadyCYcle
A. 0
-0.1 Ik
Ar
.0.1", or
Q* 0.0- (0-0-0,40. a-wa, w4v
-41.0.
.9.
-0.12
II
-0.14
TIme (Sacs)
Figure6.57a.- Effect of deteriorationon displacement

resistance(loadmagnitudeof 6kN
with 0.66mm,crackwidth)
226
0.4
LateCycle
0.3
02
S
E
0.1
CL
EarlyCycle
0 --"4 m .6 if 1 %A %Mw A H if +*-*-*-

0.2 0.3 0.4 05 0
Or
,
-01
-0.2
Time (Secs)
Figure 6.57b - Effect of deterioration on displacementresistance(load magnitudeof 6kN

with 1.98mmcrack width)
18
1.6
14
IF
1.2
3 3mm Surface Crack
0.8
06
2mm Surface Crack
4 6mm Surface Crack
0.4 x
0.2
0
0347
Appilod Load (W)
Figure 6.58 - Effect of load magnitude on displacement
227
2
1.4
1.2-
, 08
0.
04
02
0
0 234567
Initial Surace Crack Width (mm)
Figure 6.59 - Effect of crack profile on differential displacement
0.4
0.35
.. o.e.". O.e.e.a.Q.ee. G". Go. o.e
Late Cycle
03
0.25
E
02
0.15
0.1
0.05
of if m if ECI
0
0.2 0.3 0.4 4111149-0-0 05 0
. 0.05
Time (Socs)
Figure 6.60a- Effect of deterioration on displacementresistance(Parallel crack of

0.66mm width)
228
1.2
O.
(k.
.1P.
Late Cycle
I
08
)O-mw 1111 Wif
to W 1111 6
.20.6
I
0.4
Late Cycle
0.2
0
0.1 0.2 03 0.4 05
0 06
Tlme (Secs)
Figure 6.60b - Effect of deteriorationon displacementresistance(Parallel crack of

1.98mm.width)
1.8
1.6
-14-
1.2-
0.8
06
0.4
0.2
ol
0 2346
Figure6.61- Effectof steelfabric andreinforcingbaron displacement
229
0 0.1 0.2 0.3 0.4
1 0.5 06
-0.01
-002
Early Cycle
vx"
W_W,
-003 x_x, t4, y
_w A. X_wx uwx
CL -0.04
-0.05
Late Cycle 0.0-0

0-40'
-006 .
<P' 43,-
-007
Time (Secs)
Figure 6.62a- Effect of deteriorationon displacementresistance(steel fabric with

1.98mmcrack width)
a 0.1 02 0.3 0.4 0.5 06
-0.02
-0.04
Early Cycle
O.D6
c
oo8
C, or '-s
2
. 0.1 or
e *1
OF 4k
i1r,
-0,12 Late Cycle
ar
w
. 0.14
Ar e.
e. g.
-0.16
Time (Secs)
Figure6.62b- Effect of deteriorationon displacement

resistance(steelfabric with
4.62mmcrackwidth)
230
CRACK FAILED
1.8
16
1.4 Limit'& Exceeded Here
1.2
Limit W Exceeded Here CRACK DEGRADING
100 micron slab

0.8 serviceability Limit (200
micron Test Limit)
I-
0
0.6
04 micron slab 7-
serviceability Limit (600
FULLY SERVICEABLE micron Test UmlQ
0.2
50000 100000 150000 200000 250000

No of Cycles
Figure 6.63 - Serviceability thresholdsfor crack degradation
231
7. NUMERICAL MODELLING
7.1 Introduction
This chapterdiscussesthe finite elementmodel developedfor predictionsof slab response
under a variety of loading and environmental conditions. The data obtained from the
laboratory testing of small-scalespecimenswas used to provide input of the load transfer
mechanism for the relevant crack width and reinforcement type. Field testing enabled
comparisons between the numerical model and actual site responseto determine the
effectiveness of the approach. Parameters could then be changed in a standard slab to
enable their effects on behaviourto be identified.
7.2 Model Set-up
Part of the requirement for the finite element configuration was to facilitate future
incorporation of the load transfer mechanism into the model constructed by Bishop
(2001). This provided predications of internal concrete stressesand strains, resulting in a
crack opening dependanton the environmental conditions and structureproperties.When

impacted by load the resultant performanceof the slab is heavily influenced by the load
transfer system which is partially controlled by the width of the crack. To enable
progression of slab prediction it was necessary to fully understand and develop a
simulation of typical joint behaviour.
As with the Bishop (2001) model, this testing investigatedthe performanceof a slab edge
alone, with the large distance from the comer making its influence minimal in affecting
response. A two-dimensional finite element program could therefore be utilised since
plain strain elements have been shown to provide an adequatesimulation to real situations
the distance from the adjacentjoint is greater than 0.25m (Kim et aL 1998). This
when
method reduced computational requirementsand enableda more meticulous examination
of the effects of singular load transfer mechanismsthan that of a three dimensional model.
A standard set-up was developed which permitted comparison with other deflection
finite element models, providing verification of the approach. The standard
equations and
two 6 metre long, 200nim deep slabs sited next to each other, joined only by
consisted of
232
the load transfer spring system placed between the opposing bottom nodes of the crack
face (Figure 7.1). The DIANA software packagewas used to model each slab with 225
eight noded quadrilateral plane strain elements, with the load transfer utilising a two
noded translational spring. The support conditions required a one noded translational
spring at all base slab nodal positions. An interface gap element with a coefficient of
friction equal to 1.0 (as recommendedby Bishop 2001) was placed betweenthe underside
of the slab and the foundation springs to enable the slip membrane friction to be
modelled, and allow upward vertical unrestricted movement where applicable. A spring
was used for the load transfer system to enable large fluctuations in value to be
incorporated in a relatively simple way. This could also be used effectively regardlessof
the structure's constitutive model. Furthermore, it enabled the wide variety of load
transfer mechanismsfound in concreteslabsto be integratedin the sameway, without the
need for separatemodelling. This method does not allow direct interaction betweenthe
crack width opening from slab shrinkage behaviour and the response of the spring.
However, this was acceptablefor a first stage analysis and could be adapted in future
modelling as required. Generally the approachwill not be entirely accuratefor calculating

the stressesbetween slabs since each mechanismwill exert different forces; however, in
respect to deflection responsethe accuracyshould be high.
As mentioned in section2.7.4 the use of a Winkler spring to representthe soil foundation

is not completely accurate.However, to enable comparisonswith the Westergaard(1926
and 1947) equations,and many of the other analytical models, it was found to be the best
approach. This also enabled easier determination of the soil behaviour when estimating
the foundation strength direct from back-analysis of the site deflection response.Where
the standard model has been used, a subbase material was in several situations
incorporated between the Winkler spring and the slab to enable its effects to be
determined.
7.3 Model Verification
To verify the model set-upand ensureadequatepredictions of slab response,comparisons
were made to the Westergaard1926 and 1947 equations.These were evaluatedagainstan

ILLIS-SLAB finite element packageby Ioannides et al. (1985) and were found to show
correlation, with the Westergaard(1947) equations providing the greatestaccuracy.

good
Ioannides et A (1985) proposedthat the infinitely long slab used within the Westergaard
equation was the cause of the slight variation between results. The Concrete Society
(2003) continue to use these equations to predict serviceability slab deflections in
233
uncracked situations, and these can therefore be used with confidence in ascertainingthe
accuracy of the finite elementresults.
Comparisonswere madeat both the edge and centre,of the slab as it can be assumedthat
these are equivalent to the extremities of zero and 100% load transfer. Slab lengths were
taken as 6m long for the initial tests with thickness varying between 100,200 and
300mm. Figures 7.2a-f show the results from the analysesand the comparison with the
Westergaardoriginal (1926) and new (1947) equations,where applicable.
The plots show that the relationship between the methods improves for both the central
and edge locations as the depth of the slab increases.

In most situations the Westergaard
1926 and 1947 values show a higher deflection than that of the DLANA model. The only
exception to this is with the 300mm slab loaded at the centre, whereby the DLANA model
higher deflections when the modulus of subgrade is below 0.01 NIMM 3. The best
shows
correlation is observed in the 300mm slab loaded at the edge, in which changes between
both Westergaards1947and 1926equations,and the DIANA model are negligible.
The Westergaard1926 and 1947 equationsassumean infinitely long slab; however, that
used in the model is chosen to simulate a real situation as closely as possible and is
therefore limited to 6m. Ioannides et aL (1985) showed that a greater radius of relative
stiffness and slab length produced a reduction and increasein the associateddeflections
respectively. Figure 7.3 shows the effect of increasingthe DIANA model slab length for a
fixed thickness of 200nim, with those that are longer producing results closer to that
found when using the Westergaard 1926 and 1947 equations. This suggeststhat the
Westergaard 1926 assumptionof infinite slab length is the causeof the variation between
the methods and thus, the results from the DIANA analysis can be used to predict
deflection from load.
7.4 Comparison to Laboratory testing
A finite element model of the laboratory test was producedto confirm the effect of spring
deflection and enable the resistance provided from subgrade support to be
stiffness on
identified. The model itself consisted of two end sections which were restrained in both
the horizontal and vertical directions. The central section was free to move in both
directions, with springs placed between the two lowest nodes adjacent to each other
the crack faces (Figure 7.4). This was undertaken on both sides of the central block
across
to control load transfer through alterations in stiffness of each spring. The concrete was
234
modelled using an eight noded quadrilateral isoparametricplane strain element,with load
transfer utilising a two noded translational spring. Where foundation support conditions
were required, a one noded translational spring was used, the stiffness of which was
determined by the modulus of subgradereaction. Thesewere attachedto eachnode of the
lower central block and provided equal restraint throughout.
To determine the spring stiffness (K) required for replication of load transfer in a
singularly cracked slab, the laboratory cyclic loading test data was translated using the
standard spring equation.The total differential displacementat the end of the testing was
halved to provide an averagemovement in one direction from the initial block position. A
rearrangement of the standardspring equation (equation 7.1) then producedthe associated
stiffness from a knowledge of this deflection (A) and the applied load (P).
K P/A equation7.1
The force is equivalent to half the applied load (due to symmetry), and assumesthe effect
between eachcrack face is similar, and can thereforebe split equally.
The data calculated from the equation producesexactly the samevalues as that obtained
when modelling the lab test with zero subbase support (Figure 7.5). This enables the
associated spring stiffness to be determined directly from the results of the laboratory
testing for any of the load magnitudes and crack types investigated. These can then be
input directly into any analytical model of site slabs to produce representative load
transfer results.
Due to the set-upof the laboratory test it was not possibleto examine the effects of under
slab support on joint stiffness, nevertheless, by means of the finite element model this
could be represented adequately for a single load application. The effect of degradation
within the joint face cannot be incorporated since the resistance provided will occur
throughout the duration of the test, thereby reducing displacements and the rate of
degradation. The model can therefore only provide a single representationof the loading
cycle and is therefore a snapshot of behaviour, with the load transfer spring stiffness
determined beforehandusing another sourceof data.The result of incorporating a subbase
can be seen in Figure 7.6, which clearly shows the beneficial effects
under the slab edge
having under-slab support. Any change in foundation strength has very little effect
of
the subgrade modulus values are between 0.05 and 0.15N/mrr?, a range
when
of good to excellent strength. When this drops to O.OIN/mm3 there is a
representative
235
clear indication that the amount of displacementincreasesaccordingly; however, even at
such a low level, it producesa large reduction in displacementwhen comparedto non-
supported beams. This demonstrateshow important voiding is when examining under
slab edge deflections, as a complete loss of foundation produces large increases in
displacementto that found when even a low level of supportis provided.
7.5 Relationship with Site Data
To accurately model site behaviour, four finite elementmodels were producedcontaining

the appropriate slab lengths and depths. In all casesthe model consisted of two slabs
joined at the bottom nodes with a translational spring, similar to the standard model
described in section 7.2. Details of the foundation were not known and therefore a
Winkler spring systemwas used in lieu of a separatesubbaseand subgradesystem.This
enabled a support condition to be incorporated, without the need to input individual

subbase and subgrade information from site. As mentioned in section 4.5.4 back-
calculations from the FWD can be used to ascertain foundation strengthsfor the separate
materials. However, Tang (1993) statesthat this is not applicable when used at the edge
condition, as the valuesof modulus of subgradereaction are often different under thejoint
to that found in the more central conditions. A trial and error approach was therefore
employed, whereby an estimation of modulus of subgrade reaction was incorporated
within the model, from the outcomes of slab deflection behaviour examined against real
site data. If the comparisonswere poor, the support conditions were altered accordingly
until a reasonablecorrelation was produced.The variety of behavioursexaminedenabled
the approachto be usedwith confidence.The valuesof support within each site correlated
well with expectations(Figures 7.7 - 7.10, a/b/c), with the sites thought to have a good
foundation showing greater stiffnesses than those that did not. The magnitudes of
modulus of subgradereaction found through the finite element model comparisonswere

representativeof typical site conditions, as suggestedby Knapton (1999).
Predicted and site measuredresults of load transfer and load step showed good agreement
(Figure 7.7 - 7.10, a). As expected from the examination of field data there is some
degree of scatter causedin part by the variations in subbasesupport. The occurrenceof
voiding affects the results with an increase in step produced, the size of which depends on
the magnitude of the void (section 6.3.2 and 6.3-3). The proximity of the site data to the
analytical result line is much closer in those sites containing the highest modulus of
subgradereaction (Ballymena, Daventry). This indicates that voiding and reducedsupport
creates a variation in values. Interestingly, the greatest amount of disparity occurs at the
236
very low end of the load transfer region, with site data from Ballymena, Daventry and
Skelmersdaleshowing a greateramount of step for its associatedload transfer than found
from the analytical results.This indicatesthat where load transfer is negligible somelevel
of voiding appearsto accumulate.This may either be causedby the lack of load transfer
causing increasing deflection and leading to permanent subbase compaction, or the
voiding under the slab increasing the stresses across the crack face resulting in
degradation and therefore a reduced load transfer mechanism.In all casesthe site data
shows that simulation of a consistent foundation support throughout the load transfer
is
range reasonable except where low levels of joint stiffness are found, at which point a
voided model needsto be introduced.
The comparison between the analytical model and site data of loaded and unloaded slab
deflections shows much greater disparity (figure 7.7-10, b/c). Even at Ballymena where
the foundation producedreasonable to
agreement load step comparisons,somescattercan
be seen.As predicted the smaller load from the Prima producesa greatercorrelation than
the FWD due to its reduced overall deflection; however, the percentageerror betweenthe
two is similar. As with the load step relationship, the site containing the strongest
foundation generatesthe greatest correlation, with those having weak support, or slab
edges thought to contain voiding, producing the most discrepancy. These plots confirm
that edge deflections are controlled by many factors other than the load transfer
mechanism,with the foundation support and void quantity being influential.
Where FWD measurementswere taken, comparisons were made of the actual slabs'
deflected shape and that estimated using the DIANA model (Figure 7.11, a/b). At
Daventry comparisonswere made between typical good and poor joints, with associated
load transfersof 90 and 10% respectively.The deflections measuredwere all taken on the
loaded slab except for the end geophone, which was situated on the unloaded slab to
of joint load transfer. The plot of the good joint illustrates high
enable calculation
variation in the 3m section of slab closest to the slab edge. However, extrapolation of the
line beyond this distance indicates that the two will become relatively comparable at
one metre further back. The deflections found at the very edge of the slab
approximately
are similar, with those from site being slightly higher than that of the finite element
model. This changes within the first 100mm as the numerical analysis value producesa
higher gradient of deflection when compared to the actual slab, which remains fairly
Similarities are found in the results of the poorjoint-, however, in this situation
consistent.
the deflections throughout the length of the slab are greater on site than in the finite
model. The results from Lutterworth are similar to those of Daventry, although in
element
237
this casethe deflections are recordedon the unloaded slab, with load transferscalculated
at a high value only as no site data was available below 60%.
The variation in gradient indicates that for both sites the slab contains a greater stiffness
than that estimated with the model. This is probably due to the two dimensional
assumption used within the finite element analysis which contains some known
limitations when examining a three-dimensionalelement.
7.6 Comparison of Laboratory Obtained Joint Stiffness
For the Daventry and Lutterworth sites, data obtained of crack width and joint
effectivenesswas comparedto the results of the finite element model, with the laboratory
calculated stiffness usedin the developmentof the load transfer representation.Initially, a
laboratory data plot was chosenwhich most closely representedthe slab construction in
respect to the reinforcement type and quantity. The number of loading cycles applied to
the site slab was assumedto be in the region of 250,000 or greater. Subsequently,it was
possible to use the plots of crack width and differential deflection provided in section 6.4
to determine a representative spring stiffness value for that on site. These were
incorporated within the DIANA model most closely representingthe site conditions in
respect to geometry, material propertiesand support condition, enabling the resultant slab
deflections to be identified. Translations of the data into load transfer and load step
allowed the responseof a laboratory and analytical model to be comparedto that obtained
from site. This processis shown in diagrammaticform in Figure 7.12.
A secondfinite element plot using a voided slab was produced to examine its effects and
ascertain whether it gave better correlation to site data. This consistedof an affected area
10% of the total slab length, as suggestedby Suprenant(2002) and found from the site
data in section 6.2.3. Ideally, the foundation support under this section would have been
reduced to zero to represent a full void. However, constructing a model in this way
createda mechanismwithin the analysisproducing erroneousresults. Incorporation of the
lowest modulus of subgrade reaction possible was therefore required to simulate the
reduced support but prevent the mechanismfrom developing. A period of trial and error
deducedthat the smallestvalue achievablewas 0.005N/mm3.
The resultsof loadtransferandcrackwidth from both the voidedandunvoidedDaventry

finite elementmodels,along with the site data, are shown in Figure 7.13. The crack
valuesfrom the laboratorytestinghavebeenextrapolatedto providethe equivalentwidth
238
had the beam been of the same depth as the slab. As expected the data shows that the
model containing the void produces higher load transfer values than the unvoided.
However, the majority of the site data is above both of these lines, and thus indicates
some increasedload transfer effect in addition to that discovered in the laboratory tests.
The simulation test beams were 100mm in depth, and therefore contained a crack
approximately 30mm shallower than was found on site. Any increasedsurfaceareacould

therefore raise the load transfer potential and result in values closer to the field data. The
effect would be more pronouncedin areaswhere the surface crack is below 1.5mm, as it
has been suggestedby Pearson (1999) and Colley and Humphrey (1967) that widths
above this show minimal aggregateinterlock. This observation supportsthe findings from
site as the data displays greatestdivergencefrom the analytical line when the crack width
is low. However, a discrepancy also exists as modelling of voiding requires a value of
support to be inserted to prevent a mechanismbeing developed.As mentionedpreviously

this should ideally be zero to fully representa true condition, although in the analytical
model a low value had to be incorporated thereby resulting in a lower load transfer in
comparison. This would create a significant shift in the line, and produceresults closer to
that found on site.
The comparison of site deflections with the prediction lines of load step and crack width
from the numerical analysis shows significant differences between the voided and
unvoided slabs (Figure 7.14). Here, the unvoided slab producesa better comparisonto the
site data, although the voided line produces an upper boundary which no site data point
exceeds. This change in responsebetween the load transfer and load step indicates that
the increasein joint stiffness brought about by the greater slab depth in the real site joint
is more likely than voiding to be the explanation of the discrepancy. A higher joint
stiffness would increasethe load transfer and reducethe load step for a given crack width,
thereby producing better correlation to the site data.
The assumptions made in the detennination of load transfer from the laboratory
the accuracyof any prediction. In the determination of the test
simulation may also affect
duration, it was assumedthat 250,000 cycles of the +/- 4kN load produced an equivalent
to that found within the slab. If that on site was at a lower stress,or received fewer
stress
of load, then fatigue is reduced thereby increasing joint stiffness, resulting in
cycles
load transfer values and lower load steps.
greater
The load transfer results obtained from the Lutterworth site were compared against the
laboratory testing of both non-reinforcedand mesh reinforced concreteas the condition of
239
the actual joints on site was unknown (Figure 7.15). The Concrete Society (2003) state
that any steel crossing an open joint may have yielded and therefore the degreeof load
transfer capacity will reduce accordingly. Similar to the results of Daventry, both of the
load transfer mechanismswere tested for unvoided and voided foundations to enable
comparison. As expectedmost of the data obtained from site lies somewherebetweenthe
extreme lines, these being the voided mesh and the unvoided non-reinforced joints. As
mentioned previously, voiding under the slab edge produces higher values of load
transfer, although this has only a minor impact when compared to the change between
mesh and non-reinforced specimens.
Comparison of Lutterworth load step shows that the mesh reinforced specimensproduce
better correlation with the site data than that of non-reinforced (Figure 7.16). The non-
reinforced analytical line appears to be an upper boundary which none of the site data
exceeds. Similar to the Daventry results several of the points show an increased load
transfer and reduced load step when comparedto any of the analytical predictions. This
may be caused by the increased section of crack face found on site compared to that
simulated in the laboratory, which leads to an improvement in the load transfer
mechanism.
The results from both tests show that numerical modelling of site conditions using
laboratory testing and the DUNA finite element software producesacceptableresults for
slab The
response. accuracyof the site data alone is very scattered due to the
estimating
variation in crack geometry, inconsistenciesin under slab voiding, and the load transfer
mechanism. By obtaining further information about the exact loading frequency and
magnitude, alongside the support conditions, a better approximation can be made using
the analytical techniques. In both cases investigated, the importance of selecting the
for comparison between laboratory and site joint stiffness is shown.
correct parameters
However, even in caseswhere exact details are unknown the predictions produce good
lower and upper bound levels enabling assumptions of worst-case scenarios to be
developed.
7.7 Effect of Constituent Material Parameters
The standard model consisted of two 6m slabs, with an arrangementand specification
to that describedin Chapter7.2 (Figure 7.1). In the exampleswhere a subbasewas

similar
incorporated, this was modelled with eight noded quadrilateral plane strain elementswith
the strength and stiffness input as appropriate. Two depths were chosen as 150 and
240
225mm. as these representedtypical maximum and minimum values suggestedby the
Concrete Society (2003). The strength of the Winkler spring subgradewas reduced to
0.01 N/mm for these models as this is typical of a subgradematerial with the stiffness
used previously being a combination of the two foundation layers.
The methodology behind the standardslab testing was to vary individual parametersin
turn and examine their effect on deflection response.The values used for each analysis
were typical of those found in site conditions, with comparisons made of modulus of
subgrade reaction, concrete modulus of elasticity, joint spring stiffness, slab depth,
subbaseYoung's modulus and load position. The parametersused for the standardslab
are listed below in Table 7.1.
Table 7.1 - Standard Slab Basic Parameters
Parameter Standard Value

Slab Depth 200mm
Modulus of S/G Reaction 0.05 N/mrn/'
Concrete Poisson'sRatio 0.15
Load Magnitude 50kN
Joint Spring Stiffness 0 and 1,000MN/m'
Concrete Young's Modulus 30 GPa
Load Position Slab Edge
SubbaseYoung's Modulus 100 - 300 MPa
SubbasePoisson's Ratio 0.2
Frictional Interface 1.0
Figure 7.17 shows the effect of the modulus of subgradereaction in resisting deflection.
Towards the reasonable (0-05 N/mm) to good (0-2 N/mm) level the plot shows a
moderate increase in deflection as the soil modulus is reduced.However, in poor soil the
change in deflection gradient is steep with high levels of movement for a relatively small
reduction in modulus. The differential between the 0 and 100% joint effectivenessalso
increases slightly as the modulus of subgrade reaction reduces, indicating how
performance becomes more prominent when the foundation is weak.
The effect of the concrete's Young's modulus is minimal in resisting deflection, with
even the maximum stiffness investigatedof 35 GPa only producing a 0.035mm reduction
in slab edge deflection compared to that obtained with the minimum value of 24 GPa
(Figure 7.18). This demonstratesthat increases in the stiffness of the concrete provide
little assistancein the reduction to deflection, regardlessof joint efficiency.
241
An increase in joint stiffness is most beneficial in reducing deflections betweenthe zero
and 200 NlNlm level, typically 0-80% load transfer (Figure 7.19). After this range the
effect slows, with any additional increases in joint stiffness producing only minor
improvement, the overall plot showing reasonablecorrelation to a logarithmic curve. In a
perfect joint where the load transfer is 100% the slab edgeswill result in a deflection of
0.18mm comparedto that of 0.36mm for a low stiffnessjoint. These results show that in
most situations high load transferscan halve the unrestricteddeflection. There is however
a point where large enhancements in stiffness (or load transfer mechanism) will only
produce minor reductions in the edge deflections, and therefore a suitable compron-dse
has to be madeby the designer.
Increases in slab depth, resulting in a stiffening of the structure, produce significant
reductions in the edge deflections under both good and bad load transfer conditions
(Figure 7.20). For the standardslab, doubling of the depth from 150 to 300mm reduces
the edge deflection by 0.16mm when the joint has zero load transfer. In situations where
the joint stiffness is high then reduction is still in the region of 0.08mm.
Figure 7.21 demonstratesthe effect of subbaseinclusion in reducing the magnitude of

deflection at the slab edge.Where the joint is working efficiently the subbasestiffness has
very little influence regardless of thickness. However, this changes when the joint
is
stiffness poor with the introduction of the subbasedecreasingthe magnitude of the
deflection dramatically. The model containing the thickest subbase shows the most
reduction, though the effect of both is reduced as its modulus is lowered from 300 to
IOOMN/rn.
Hammons (1998) has suggestedthat the distanceof the load from the slab edgeaffects the
value of load transfer. The spring stiffness examined during this finite element analysis
shows that equal load transfer values are produced regardless of the location of the load
(Figure 7.22). The disagreementwith the work of Hammons (1998) may be causedby the
free slip phenomenon,as discussedin Chapter 6.3.2. Here, the deflection reducesas the
load is moved away from the joint, making the proportion of free slip larger, resulting in a
reduction in the load transfer efficiency.
242
7.8 Summary
A finite element model has beendevelopedand verified againstthe establishedequations

developed by Westergaard(1926,1947), and the Ioannideset aL (1985) ILLISLAB finite
element model. The results of this analysis have shown good comparison and, as
predicted, indicate the greatestcorrelation with Westergaard's(1947) equationswhen a
reduced stiffness, createdfrom an increasingslab length and reducedthicknessare used.
Several models have been establishedcontaining the parametersassumedand measured
on site. These models have enabled analysis of deflection responseand, via a back-
calculation approach, the determination of the magnitude of modulus of subgrade.The
results from the deflections obtained under the same loading conditions as that of site
have then been compared and found to show reasonableagreementin most situations.
Any scatter found in the data has been attributed to under slab voiding and the
assumptionsin materials and layout when developing the finite elementmodel.
Using the laboratory baseddegradationplots, spring stiffness has been determinedusing

the standardspring equation and a finite element simulation model. The stiffness for the
associatedcrack width has then been imported into the relevant analytical model with the
deflection responsecomparedagainst that obtained from site. Voiding is well known to
affect the magnitude of deflections on some sites and therefore models were set-up in
which 10% of the slab length contained a reduced level of foundation support to
incorporate this situation.
Although the data is scattered, the finite element model provides a reasonable
representation of site behaviour. Examination of the information shows that analysis of

both the load transfer and load step graphs needsto be undertakento ensure continuity
and enable voiding effects to be ascertained.The additional support available from the
larger crack face on the site joints appearsto produce an appreciably higher load transfer
value than is found within the laboratory. The effect of the reduced fatigue createdby the
foundation support condition and the unknown loading behaviour is also thought to have
some effect, increasing the load transfer values and decreasing the load step. In general
the finite element testing enables good estimation of the slab behaviour, with upper and
lower bound levels determined.
A standard slab containing typical dimensions and material parameters was used to
the effect of each on slab response.Subgradesupport conditions were found to
establish
243
have a large impact on response,with those containing below averagevalues producing
greater increases in deflection for relatively small enhancementsin stiffness. This is

sin-Lilar when utilising a subbase as the strength and depth of the material heavily
influences the associateddeflections, indicating that its careful selection can assist in
increasing the longevity of the slab. Concrete strength has little effect in respect to
deflection; however, increasingthe thicknessof the slab (thereby stiffening the structure)
provides resistanceto a significant degree.Joint efficiency is highly influential in the slab

response as a 100% load transfer can reduce the deflection of a non-efficient joint by
50%. The effects are more influential when the joint stiffness is low, as even small
increases can provide large reductions in deflection. According to the finite element
model, in a normal situation the distance of the force from the slab edge has very little
impact on the load transfer magnitudeobtained,although the shearslip phenomenonmay
influence the results due to the changein total deflection.
244
I I
STANDARD MODEL
VT IIIflII JJI s. 45a LISICI 14CC
I I
STANDARD MODEL WI'M SUBBASE
auuud3c ... - . U
Figure 7.1 - Finite Element model of a concreteslab on grade with a discontinuity
18-
le.
1.4-
1.2-
co
0.4-
0.2-
01
0
Modulus of Subgrade Reaction (Wmm3)
Figure 7.2a - Comparisonof DIANA with Westergaard(1926) for a 100mm slab loaded
internally
245

-1
41-Westergaard
- -*- - DIANA F E.
m
os
od
0.4
0.2
........ IK
0
0 002 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 02 0.22
Modulus of Subgrade Reaction (NImm3)
Figure 7.2b - Comparisonof DIANA with Westergaard(1926) for a 200nun slab loaded
internally
-0-Westergaard
DIANA F. E.
1.4
1.2
08
0.6
04
0.2
----------- -------- -------

0
0 002 004 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0.22
Modulus of Subgrade Reaction (N/mm3)
Figure 7.2c - Comparisonof DIANA with Westergaard(1926) for a 300mm slab loaded
intemally
246
6.5.
6.
55-
5-
4.5-
3.5
3-
2.5-
15-
1-
0.5-
0. -
Modulus of Subgrade Reaction (NImm3)
Figure 7.2d - Comparisonof DLANA with Westergaard(1926 and 1947) for a loomm,
slab loadedat the edge
Westergaard (Orig )
Westergaard (New)
- -X- - DIANA F.E.
45
3.5
2.5
1.5
0.5
0
0 002 004 0.06 008 0.1 0.12 0.14 016 0.18 0.2 0.22
Modulus of Subgrade Reaction (N/mm3)
Figure 7.2e - Comparisonof DIANA with Westergaard(1926 and 1947) for a 200MM
slab loaded at the edge
247
65 o -1
-0--Westergaard (Ong)
a 0 Westergaard (Now)
-*- DIANA F.E.
45
E
3.5
2.5
1.5
0.6
0
0
Moclulus of Subgrade Reaction (Wmm3)
Figure 7.2f - Comparisonof DIANA with Westergaard(1926 and 1947) for a 300MM
slab loadedat the edge
6.5
-4-6m Slab
6 -E - 10m Slab
--* **15m Slab
5.5 -X- - Westergaard (Odgional)
-X - Wastergaard (New)
5
45
35
2.5
1.5
0.5
0
005 0.1 0.15 0.2 0.25
0
Modulus of S/G Reaction (N/mm3)
Figure 7.3 - Comparisonof DLANA with Westergaard (1926 and 1947) for a 200mm.
slab
loaded at the edgewith lengths of 6,10 and 15m.
248
Load
Transfer
Springs
Edge Edge
Centmi
Supports Springs Supports
Figure 7.4 - Schematicof the Finite Element model of crack behaviour within the
laboratory test rig
100
0 F.E. Modelling (2kN)
X F.E. Modelling (4kN)
90
0 F.E. Modelling (UN)
...... Equation (2kN)
80 --- Equation (4kN)
-Equation (UN)
70
60
90
50
40
30
20
10
0 .......... ------- IK
0 ..................... 0 .................................. . 1k -------- n-r
-- -- ----
0
0 02 0.4 06 08 1 1.2 14 1.6 1.8 2
Cyclic Test Single Direction Deflection (mm)
7.5
Figure - Comparison
of the Finite
laboratory Element
modelwith the standardspring
equation
249
0.045
--*--0.01 N/mm3
- 0, - 0.05 N/mm3
0.04
--*-0.15Ntmm3
0035
0.03
0.025
0.02
0.015
001
0.005
0
0 20 40 60 so 100 120
Joint Spring Stiffness (MWm)
Figure 7.6 - Effect of foundation material under the crack in the laboratoryFinite
Element model.
1.6
--0-5OkN FWD
0 lOkN Prima
X Site FWD
1.4
Site Prima
1.2
08
IA
x
0.6
x
0.4 x
&xxxx
xxxXX
02
a. A-41 .....
0
20 40 60 so 100 120
0
Load Transfer C%)
Figure 7.7a - Comparisonof load transfer and load step betweenthe Daventry Finite
Element model and in-service slab response
250
1.e
--"*-SOkN FWD
x**
-lOkN Prima
1.4tX FWD Site
Prima Site
E
x
xx
xxx
xx
06 x
xx
xxxx xx
3E
04
xX Xk
x xxx X xx
)s( I xx
02 xa,
A& &&& &&
LOA., "
...... ..... .... ejg" .6&a
0
0 20 40 60 so 100 120
Umd Transfer (%)

Daventry Finite Element model and in-service slab response
of 50kN FWD
-lOkN Prima x
07 X FWD Site
& Prima Site
x
Me
x
E
0.5 XX
x
xx
0.4 xxx
xxxxXX
0.3
xxxxx
xxx
NX
0.2 xxXXNxx&
a4
xxx&&&A&, && &

0.1 x" && &
&&
ax &&
&&
XX OA ... 40 &&
0
0 xx......... 20 40 60
Load Transfer (%)
so 100 120

Daventry Finite Element model and in-service slab response
251
1.2
-*, -58kN FWD
0 --lOkN Prima
X Site FWD
a Site Prima
I
os
E
E
'Z
2 0.6
Co
90
3
0.4
0.2 x XX
Xxx
XX xx
0
0 20
..........
40
o
,<X
60
.e
ad
..
xx
%. AAJ
80
x
-
100 120
Load Transfer (%)
Figure 7.8a - Comparisonof load transfer and load stepbetweenthe Lutterworth Finite
1.2
-0-58kN FWD
x
, O--IDkNPdma
X FWD Site
PrimeSite
1
x
x
xxxx
xxxxx
xx
Kxx
9 XXx
XXX
06
1 XX
xx
04
xx
Xxx
xx xxx
0.2
xx xx
xx
xxxxx xxxxx x
0 e. && A&
40........... 40 %'A
04-
20 40 60 so 100 120
0
Umd Transfer (%)

Lutterworth Finite Element model and in-service slab response
252
1.2 I
---*--58kN FWD
--<>--lOkN Pdma
X FWD Site
x
& Pdma Site
I
x
x
0.8 x
xx
x
xxx ex x
XX xX
0.6 xx
x
xx
xx
xxxx
xxx
0.4
xx
0.2
xxxxxxx
xxxxx xx
xxx x
............
x A
00 20 40 60 80 100 120
0
Load Transfer (%)

0. OS
Pdma
0. OE Site Prime
IL
0.07
006
0,05
CL
2
0 Q..
0.04
0.03
A vii,
Act
0.02
001
0
0 20 40 60 80 100 120
Load Transfer (%)
Figure 7.9a Comparisonof load transfer and load step betweenthe Ballymena Finite
-
253
0.12
- C-INN Prima
a Prima Site
0.1
0.08
30.06
8
2 '4>......
da e...
ob*.
41. 0..., c,.. o
0.04
AaA ad a 4, ta&.
..
a
A &da a4
Ada A, 46

0.02 ab
0 - 20 40 60 so 100 120
Load Transfer (%)

005
I OkN Prima a
0.045
Prima Site
004
0.035
&
E
0.03
0025
&&
l
0.02
., a&a
& or &
&
0.015
&
001
0.005
0
0 20 40 so so 100 120
Load Transfer (%)
Figure7.9c Comparison of edge

unloaded deflectionandload stepbetween
the
-
BallymenaFiniteElementmodelandin-serviceslabresponse
254
0.35
Site Prima
0.3
I&
025
E 0.2
E
to
0.15
0.1
005
40
0
0 20 40 60 80 100 120
Load Transfer (%)
Figure 7.10a - Comparisonof load transfer and load stepbetweenthe SkelmersdaleFinite

0.35
--*--IOkN Proms
Prime Site
0.3
025-
E
0.2
0.15
0 .............
0.1
a
&
&
0.05.
0-
0 20 40 60 80 100 120
Umd Transfer (%)
Figure 7.1Ob- Comparisonof loadededgedeflection and load transfer betweenthe

SkelmersdaleFinite Element model and in-service slab response
255
0.12
--lOkN Prima
Prima Site
0.1
0,0.0.0
Ar .
0.06
1
0.04
0.02
a
a
0 do'
0 20 40 60 80 100 120
Load Transfer r/6)
Figure 7.1Oc- Comparisonof unloadededgedeflection and load step betweenthe

SkelmersdaleFinite Element model and in-service slab response
ei
---------------------------------
#4
At
A#
.0.1 :
A: 4;
-0.2 A
A
-0.3
E
E
C
-0.4
8 -0.5
-06.
-0.7-
-411-10% vr (Site)
- ***IO%Lfr (FE. )
-0.81
-, dr- 90% Lrr (Site)
-90% L/T (F. E.)
G.
Distancefrom Joint (m)
Figure 7.1 la - Comparisonof deflection bowls betweenthe Daventry Finite Element

model and in-service slab response
256
2347

1
' j
2
-0.2
'0 3
,0
-0.3-
4
4
0 :
-0.
-06.
-0.7-
-08- Ur (Silo)
---d-90%
-- * --9M UT (F E.)
Distancefrom Joint (m)
Figure 7.1 lb - Comparisonof deflection bowls betweenthe Lutterworth Finite Element

model and in-service slab response
257
(1)
Produce
deflectionVscyclegraph
from lab test data
(2) (A)
Selectsite andjoint with known Joint haspassedearlydeterioration
characteristics(reinforcement,slab stage
thickness,joint orientationetc.)
(B) (3)
Lab data producesworst-case Selectequivalentlab deflectionVs
scenariodue to smallersection cycle graphfrom site information
thicknessand zero support
(4)
Calculatejoint stiffnessat residual
point in deflectionVs cycle graph
K= F/A
(5) (C)
Model site slab with joint stiffness Back-calculate
groundstiffness
obtainedfrom lab tests from Westergaardor usecommon
(See4) valuesfrom literature
(6)
Comparisonof F.E. model
deflections/load transfer/
deflection bowls / to that obtained
on site
GOOD COMPARISON POORCOMPARISON
(7) (7)
Examinedifferentjoints, sites and Try different foundationstiffness.
load magnitudes.Comparesite data Explain reasonsfor poor corTelation
to RE model
(8)
Changemodel parametersto
examineeffectson site (subbase,
slab thickness,increasedjoint
stiffnessetc.)
Figure 7.12 - Processdiagram for establishingslab responseusing laboratory cyclic load

testing and the Finite Element model
258
14-
X Site Data
--*-Lab & F.E. Data (Unvoided)
12- -Lab & F.E. Data (Voided)

x
x
x
Q.. x
10
xa,
8
x
xxxxx xx
....
!! 6- ?x
ci
xxx
xx
x
4 xxxxX xx
xXxx XX
xXx
Ix x
xx
XX
2
0
20 40 60 so 100 120
Load transfer (%)
Figure 7.13 - Comparisonof predictedand actual surfacecrack width/load transfer

behaviour for Daventry
14
12
x
xx
10
x x x
x
xxxx xx xxx
xx
xxx xX
gr
xxx xy x1
x
156,
x
x
rexi>k(x
x)
xx9x4
)y
JY
,, ,, x Nu Wa
Ub & KE DMa (UnvOided)
YK
---Lab&F. E. Data(Volded)
x
1000 1500 2000 2500

0 500
Load Step (microns)
Figure7.14- Comparisonof predictedandactualsurfacecrackwidth/loadstepbehaviour

for Daventry
259
14
X $its Data
--, *, -Lab & F.E. Data (Unvoided, mesh)
-& - Lab & F.E. Data (Unvokled, non)
11 0- -Lab & F.E. Data (Voided, mesh)
-A- - Lab & F.E. Data (Voided, non)
x
I?,
x
xx
xx
xXX
xx 'k
xxX
xx
Xx xx-x ,x
xxxxx7A, XXX
xxxx 1.
xxxxx xx
xAx )c x
-W '16' '4('A Xx
0 20 40 60 so 100 120
Lood Tranal4w C%)
Figure 7.15 - Comparisonof predictedand actualsurfacecrack width/load transfer

behaviourfor Lutterworth
14
She Data
x & F.E. Data Unvoided, (mesh)
--*-Lab
- -A- Lab & F.E. Data (Unvoided, non)
12
--*--Lab& F.E. Data (Voided, mesh)
- -6- - Lab &FE. Data (Voided, non)
10
6
x
x
x
4
xx
PC x xx x
xxx
2 xx x
xx
x xx
xx- ............
..................
0
0 2w 400 600 Soo 1000 1200
Load Stop (mlcrons)
Figure 7.16 - Comparisonof predictedandactualsurfacecrack width/loadstepbehaviour

for Lutterworth
260
14
12
1
1
08
08
04,
02
005 01 0.15 02 025

Modulus of Subgrodo Asedlon (Nhnm3)
Figure 7.17 - Effect of modulusof subgradereactionon the loadedslab edgedeflection

for a typical in serviceslab
04
035
03
T
E
025
02
-x- - -x- -
015
0.1
005
--4-- 0% Load Transfer
- -)C- -100% Load Transfer
0
06 10 15 20 25 30 35 40
Concrete Modulus of Elasticity (GPII)
Figure 7.18 - Effect of concretemodulusof elasticity on the loadedslab edgedeflection

261
o
Loaded Slab
- -K- -Unloaded Slab
OY
0.1
02.1
09
* ------------
--------------
005
0
0 200 400 Goo Boo IOOD 1200
Joint Sprkvq SUffne" (MNhn2)
Figure 7.19 - Effect ofjoint spring stiffnesson slabedgedeflectionsfor a typical in
serviceslab
os
--*-0% Load Transfer
-X-, 100% Load Transfer
0451
04.
035 -
03
025
02
015.
01-
005
01
50 100 150 200 250 300 350
Slab Depth (mm)
Figure7.20- Effect of slabdepthon the loadedslabedgedeflectionfor a typical in
serviceslab
262
01
01
A
07
10@
05
0.4
03
02
0% L/r
X-- 11
5500mmmm
SS/IBB,
100% Ur
OA --*-- 225mm S/B, 0% UT
- -0- - 225mm SIB, 100% Ur
0
0 50 100 150 200 250 300 350
Subbase Youngs Modulus (MPs)
Figure 7.21 - Effect of subgrademodulusof elasticity on the loadedslab edgedeflection

9(
4 ------------------ e ------------------
a( ------------------- 0 ------------------ 0
?c
60 + ........................ a ........................ a I'll-, .......... a ........................
le 50
40 -------------*------------- >(--------------x ------------- X
30
20
--*--0.35MN/m Spring Stiffness.
-X- -90MN/m Spring Stiffness
10
20OMN/in Spring Stiffness
70OMN/mSpring Stiffness
0
01 02 03 04 0.5 06
Distanm from Slab Edge (m)
Figure7.22- Effect load positionon load transferfor a typical in serviceslab
263
8. CONCLUSIONS AND RECOMMENDATIONS FOR
FURTHER NVORK
8.1 Overview
A review of the current documentation and guidance on concrete floor/pavement

constructionand maintenancehas highlighted the need for further understandingof the
behaviour of load transfer mechanismswithin inducedjoints and cracks. Much of the
literature reviewedhasexaminedonly low cycle, high intensity fatigue loading, ignoring
the constantrepetitiveaction from vehiculartraffic.
This chaptcr presentsthe conclusions from the site, laboratory and analytical model
testing of a variety of joint propertiesto assesstheir effect on fatigue and slab response.
The researchhas encapsulatedthe initial assessmentof crack/joint geometriesand the
associatedslab responsefrom real site conditions. This has enabledsimulation of the
loading regime on test specimenswithin the laboratory,and resultedin the development
of an analytical model to predict slab behaviourunder a number of varying parameters
and conditions.
8.2 Site Testing
Measurements from site utilising embedded strain gauges, drilled cores and surface
profile approximation have established that differential shrinkage creates a significantly

larger crack width at the slab surface than at the base. This is shown to vary depending on
the category orjoint and the placement and type of reinforcement used in the slab.
Slab deflection from the impact of load was monitored with the use of dynamic plate
devices. The comparison between each device was good with respect to load transfer.
However. deflection related values did not respond in direct relation to the increase in
load magnitude. This was caused by voiding underneath the slab edges and the free-slip
phenomenon, created from degradation of the crack face.
Correlationwas found to exist betweenload transfer,load step,and edgecantilever.The

variation in resultschangeddependingon which parametersand which site were being
examined,with disparity from the idcalisedplot causedby voiding and free-slip.
264
The relationshipbetweenslab deflection behaviourand surfacecrack width
measurement
was poor, although overall trends could be observed. This was due to the limited
knowledge of the crack geometry, with surface measurementsalone used as the
comparablemeasure,alreadyshownto only partially signify the true status.
Prcloading of the slab proved to be highly influential in altering deflection response.

Splitting the data into similarly loadedslab areasproducedbetter comparisonin results,
especiallywhen comparedagainstcrack width.
8.3 Laboratory Testing
A laboratorytest rig was developedto investigatethe high cycle load transferdegradation

of a number of small-scalespecimens.Unlike many previous tests the geometriesused
were typical of thosefound during the real slabjoint and crack analysis,with most being
'V' shaped,althoughseveralparallelcrackswere also testedfor comparison.
Plots of the increase in differential displacement over 250,000 loading cycles were
obtained for a variety of concrete and reinforcement specifications. A rapid amplification

in displacement over the first few thousand cycles was found, caused by degradation of
the f me aggregate. After this there was a period of little change whereby larger aggregate
particles began to bear upon one another. In those specimens where failure transpired a
third section of rapidly increasing differential displacement occurred. Here, the aggregate
cracked and debonded from the remainder of the concrete, creating higher stressesacross
the crack face and therefore quicker degradation.
Parallel cracks were found to deteriorateat a much faster rate than 'V' shapedcracks,
with widths above I mm unableto resistfailure for the entire 250,000cycles,highlighting
the importanceof correctgeometryselection.
The inclusion of steel fibrcs was found to enhancethe load transfer mechanismin two
main ways. The restraint to micro cracking preventeddegradationand spalling of the
crack face leadingto rctainmentof aggregateinterlock. Secondly,the fibre itself provided
some load bearingthrough dowelling acrossthe crack interface.The fibre was generally
found to pull out of the concrete;however,as deflection increasedfatigue took place
leadingto fibre snapping.
265
Increasingthe quantity of fibres from 20 to 40kg/m3enabled surfacecrack width to be
openedfurther beforefailure was initiated, and createda minor reductionin magnitudeof
differential deflection. Increasingthe aspectratio of the fibre was also found to increase
the resistanceto failure; however,the resultsestablishedthat it was the changein number
of f ibrcs crossingthe joint that was having the effect, rather than modification of either
length or diameter.
Introducing steel A 142 mcsh. or 7mm reinforcing bar producedsignificant reductionsin

the differential deflection over all the other reinforcementtypes tested,regardlessof the
numberof cycles applied.For the entire rangeof crack openingstherewas no significant
changebetweenthe start and end deflectionsand visual observationsshowedlittle sign of
fatigue.
The effect of load was highly influential in the rate of deteriorationwith all specimens
loadedat 2kN (125kN/m2)showing very little fatigue damage.However,when increased
to the standard4kN (25MN/m) load this changeddramatically, with RN (37RN/m)
causingfailure in all specimensother than that with the lowest crack width.
8.4 Analytical Modelling
A finite element model developed using the DIANA software packagewas assessed
against the Westergaard(1926 and 1947) equations. When simulating the equation
assumptionsof a long slab with a shallow depth it producedgood comparison.This also
agreedwell with results obtainedfrom other researcher'swork when using other finite
elementpackagessuch as MLISLAB, proving the method to be adequatein simulating
slabresponse.
A model of the laboratory test facility produced results identical to that obtained from the
standard spring equation. The introduction of a model foundation material reduced the
deflection by over ten times on a single analysis, showing that during a set number of
cycles the damage caused by fatigue would be significantly reduced if some level of
support was introduced.
Models representingslab geometriesprovided accuratecomparisonto the load transfer

and load stepdata obtainedfrom site throughoutthe entire rangeof spring stiffness.The
small variation found was unavoidable, due to the unpredictable nature of site
foundations.Whenexaminingthe individual slabedgedeflectionsthe correlationwith the
266
analytical model was less precise,however this was improved when comparing sites with
stronger foundations.
Estimation of site slab responsewas derived using spring stiffness from the laboratory
tests for the associated crack width and reinforcement type. The results produced
deflections that matchedwell with those achievedfrom site, and enabledupper and lower
bounds for slab behaviourto be developed.
The effects of material and geometric variations were testedto examine which were most
influential in respect to slab response.Increasesin soil strength,joint stiffness and slab
depth all produced significant reductions in the edge deflection; however, concrete
strength and load position had negligible effects.
8.5 Final Comments
This researchhas introduced a method of determining slab deflection responsein relation

to load transfer by meansof small scale laboratory testing and finite element modelling.
Utilising data collected from in-service sites the method has been validated and shown to
produce reasonablecorrelation to the measured values under a variety of conditions.

Similar to previous research,crack geometry has been found to be a highly influential
factor controlling the efficiency of the slab, thus making its accuratemeasurementvital.
This has included the assessmentof 'V' shapedcracking which was discovered to be
more predominant on site than the commonly assumed parallel crack. The introduction of
reinforcement into the concrete was shown to vastly improve the resistance to deflection;
with steel fibres providing reasonablecomparison to steel mesh when used in the correct
quantities.
The concept of small scale testing facilitates cost efficient exan-dnationof load transfer
mechanisms, thus providing the designer with a simple method of assessingthe behaviour
of joints and cracks under a number of structural and environmentalconditions.
8.6 Recommendations for Further Work
To enabletypical crackgeometriesto be establishedandusedin designandmaintenance

monitoringof crack widths is The
needed. use of straingaugeshas
analysis,additional
been shown to provide reliable data; however,an increasednumbershouldbe placed
depth
throughout to enablethe full geometryto be determined.
267
Further deflection testing undertakenon both site and in the laboratory would allow the
effect of voiding and loose material under the slab edges to be examined. Comparison
between the Prima and FWD requires completion under controlled slab conditions to
determine correlation between the two devices and to examine whether the variation
found in site results is caused solely by the change in load applied to the slab. The
relationship between parameterssuch as load transfer, load step, cantilever deflection,

absolute deflection and crack width can then be determined, incorporating the effect of
pre-load.
A statistical survey of load sourcesshould be carried out on a number of industrial floor
slabs, external hardstandings and pavements. This would provide information on typical
load magnitude (i. e. vehicle type and weight) and number of applicationsacrossjoints, to
enable representations to be modelled and therefore predictions made of slab

deterioration. Without this knowledge the simulations made both in analytical modelling
and laboratory testing may be inaccurate.
The finite element model could be further developed to incorporate the subbaseand
subgrade as separateentities. This may then be compared to site behaviour where the
foundation stiffness is well known to enablethe comparisonand accuracyof the model to
be determined. Values of joint stiffness obtained from the laboratory tests containing the
relevant crack and material properties could then be inserted into the model to enable its
effect on slab response to be calculated. A model of spring stiffness changes with

associated crack opening may be used within the Bishop (2001) early age behaviour
model to automatically predict the stiffness and relevant performance. This could
incorporate a load cycle plot, obtainedfrom the laboratory tests,to enablethe degradation
and associated effects to be analytically examined over time. The model may then be
enhanced to incorporate three-dimensional elements, making the analysis of comer

deflections possible.
A combination of these recommendations would enable a simulation of deflection

for the majority of slabs using finite element software directly from the input of
response
known site parameters, and small scale testing of the load transfer mechanism. If
for many different scenariospredictions of slab behaviour under a variety of
undertaken
and environmental conditions will be generated.
geometric
268
9. REFERENCES
Abdel-Maksoud, M. G. (2000), "Behaviour of ConcreteJoints Under Cyclic Shear", PhD

Thesis, University of Illinois, USA.
Abdel-Maksoud, M. G., Hawkins, N., Barenberg, E.J. (1997). "Behaviour of Concrete

Joints under Cyclic Shear." American Society of Civil Engineers,August 1997, pp 190-
204.
Abdul-Wahab, H.M. S. (1992), "Strength of Vertical Joints with Steel Fiber Reinforced
Concrete in Large Panels",ACI Structural Journal, Vol. 89, NoA pp367-374.
ACI Committee 302.IR-96 (1996), "Guide for Concrete Floor and Slab Construction",
ACI Manual of Concrete Practice Part 2, American Concrete Institute. Farmington hills,
Michigan.
ACI Committee 360R-92 (1992), "Design of Slabs on Grade", ACI Manual of Concrete
Practice Part 2, American ConcreteInstitute, Farmington hills, Michigan.
ACI Committee 360 (2000), "Draft Copy - Chapter 10", American Concrete Institute,
Farmington hills, Michigan.
ACI Committee 544 (1988), "Design Considerations for Steel Fibrer Reinforced
Concrete" ACI Structural Journal, Vol. 85, No.5, pp 563-580.
ACIFC and the Concrete Society (1999). "Steel Fibre Reinforced Concrete Industrial
Ground Floors: An Introductory Guide" Concrete,Vol. 33, No. 10.
Al-Nasra, M. and Wang, L. R.L. (1994), "Parametric Study of Slab-on-GradeProblems

due to Initial Warping and Point Loads", ACI Structural Journal, Vol. 91, No.2, pp 198-
210.
American Association of StateHighway and Transportation Officials (1986), "AASHTO

Guide for Design of PavementStructures", American Association of State Highway and
Transportation Officials, WashingtonD.C.
269
American Concrete Institute (1977), "Hot Weather Concreting", Journal of the American
Concrete Institute, Vol. 74, pp 317-332.
Armaghani, J.M. (1993), "Factors Affecting Performance of Concrete Pavements", 5th

International Conferenceon ConcretePavementDesign and Rehabilitation, 20-22ndApril,
Purdue University, West Lafayette, Indiana, USA.
Armaghani, J.M., Lybas, J.M., Tia, M. and Ruth, B.E. (1986). "Concrete PavementJoint
Stiffness Evaluation.", Transportation Research Record, Vol-1099, Transportation
ResearchBoard, pp 22-36.
Arnold, S. (2002), "Report of the 2nd Meeting of the Industry Liaison Group on Load
Transfer acrossCracks/Jointsin ConcreteSlabson Grade", 80'April 2002, Loughborough
University, England.
Atkins, H.N. (1997), "Highway Materials, Soils and Concretes", 3rd ed., Prentice Hall,
Upper SaddleRiver, N.J., USA.
Balaguru, P.N. and Shah S.P. (1992), "Fiber-Reinforced Cement Composites",1" ed.,
McGraw-Hill Inc., USA.
Barenberg, E.J. and Zollinger D.G. (1990), "validation of ConcretePavementResponses
using Instrumented Pavements." Transportation Research Record, Vol. 1286,

Transportation ResearchBoard, pp 67-77.
Barnbrook, G. (2000), "Overview of Industrial Concrete Floor Developments", Seminar
on Developments in Industrial Ground Floor Slabs, 31"October 2000, Aston University,
England.
Bazant, Z P. and Gambarova P. (1980), "Rough Cracks in Reinforced Concrete" Journal
Structural Division, ASCE, Vol. 106, No.4, pp 819-842.

of the
Beckett (1999), "Comer and Edge Loading on Concrete Industrial Ground Floors
Reinforced with Steel Fibres" Concrete,Vol. 33, No. 10, pp 22-24.
270
Beckett, D. (2000), 'Thickness Design of Concrete Industrial Ground Floors Notes for
-
Guidance", Seminar on Developments in Industrial Ground Floor Slabs, 31" October
2000, Aston University, England.
Benkelman, A. C. (1933), "Tests of AggregateInterlock at Joints and Cracks" Engineering

News Record Vol-3, No. 8,227-232.
Bhatti, M. A., Barlow J.A and Stoner, J.W. (1996), "Modelling Damage to Rigid
pavements Caused by Subgrade Pumping" Journal of Transportation Engineering,

Vol. 122, pp 12-21.
Bhatti, M. A., Molinas-Vega, 1. And Stoner, J.W. (1998), "Nonlinear Analysis of Jointed
Concrete Pavements"TransportationResearchRecordVol. 1629,pp 50-57
Bischoff, P. Valsangker, AJ, and Irving, J. (1997), "A PerformanceEvaluation of Steel

Fibre Reinforced ConcreteSlabs on Grade", Asia-Pacific Speciality Conferenceon Fibre
Reinforced Concrete,28-29h August 1997,Singapore.
Bishop, J. (2001), "The Early Age Behaviour of ConcreteIndustrial Ground Floor Slabs",
PhD Thesis, Loughborough,England.
Bodocsi, A., Minkarah, I. A. and Arudi, R.S. (1994). "Effect of pavement Variables on
Average Joint Deflections in Experimental ConcretePavement"Transportation Research
Record, Vol. 1449,TransportationBoard, pp 182-188.
BS 1881: Part 108 (1983). "Method for Determination of Compressive Strength of

Concrete", British StandardsInstitution, Milton Keynes.
BS 1881: Part 116 (1983). "Method for Determination of Compressive Strength of

Concrete", British StandardsInstitution, Milton Keynes.
Buch, N. (1999), "Factors Affecting Load Transfer across TransverseJoints in Jointed

Concrete Pavements",ASTM SpecialPublications,Vol. 181, No.3, pp 43-64.
Buch, N., Frabizzio, M. A. and Hillier J.E. (2000). "Impact of Coarse Aggregates on
Transverse Crack Performancein Jointed Concrete Pavements",ACI Materials Journal,

Vol. 97, No. 3, pp 325-331.
271
t
Chandler, J.W.E. and Neal, F.R. (1988). "The Design of Ground Supported Industrial
Floor Slabs", Interim Technical Note 11, British CementAssociation, Slough.
Channakeshava,C., Barzeger,F. and Voyiadjis, G.Z. (1993), "Nonlinear FE Analysis of

Plain ConcretePavementswith Dowelled Joints", Journal of TransportationEngineering,
Vol. 119, No.5, pp 763-781.
Chou, Y. T. (1983), "Comparative Analysis of Rigid Pavements.

" Journal of
TransportationEngineering,Vol. 109, No.5, pp 669-687.
Chou, Y. T. (1989), "Failure of Concrete Pavements in Test Tracks." Journal of

Transportation Engineering,Vol. 115,No.5, pp 493-505.
Colley, B.E. and Humphrey H.A. (1967), "Aggregate Interlock at Joints in Concrete
Pavements", Development Department Bulletin D124, Portland Cement Association,
Illinois.
Colley, B.E. and Nowlen J.W. (1958), "Performanceof Subbasesfor ConcretePavements
under Repetitive Loading", Development Department Bulletin D23, Portland Cement

Association, Illinois.
Concrete Society (2003), "Concrete Industrial Ground Floors -A Guide to design and
Construction", Technical Report No.34,3rd ed. Crowthome, UK.
Critchell, P.L. (1958), "Joints and Cracks in Concrete", 2nded., Elsevier Applied Science,
London.
Croney, D. and Croney P. (1997), "Design and Performanceof Road Pavements",Yd ed.,
Crovetti, J.A. and Darter M. 1. (1985), "Void Detection for Jointed ConcretePavements",
Transportation ResearchRecordVol. 104,TransportationResearchBoard, pp 59-68.
Cudworth, D. (2000), "The Dangers of Ignoring Asymmetric Dynamic Loading in
Industrial Floor "

Slabs. Concrete,Vol 35, No. 9, pp 25-27.
272
Chandler, J.W.E. and Neal, F.R. (1988). "The Design of Ground Supported Industrial
Floor Slabs", Interim Technical Note 11, British CementAssociation, Slough.
Channakeshava,C., Barzeger,F. and Voyiadjis, G.Z. (1993), "Nonlinear FE Analysis of

Plain Concrete Pavementswith Dowelled Joints", Journal of TransportationEngineering,
Vol. 119, No.5, pp 763-781.
Chou, Y. T. (1983), "Comparative Analysis of Rigid Pavements.

" Journal of
Transportation Engineering,Vol. 109, No.5, pp 669-687.
Chou, Y. T. (1989), "Failure of Concrete Pavements in Test Tracks." Journal of

Transportation Engineering,Vol. 115,No.5, pp 493-505.
Colley, B.E. and Humphrey H.A. (1967), "Aggregate Interlock at Joints in Concrete
Pavements", Development Department Bulletin D124, Portland Cement Association,
Illinois.
Colley, B.E. and Nowlen J.W. (1958), "Performanceof Subbasesfor ConcretePavements
under Repetitive Loading", Development Department Bulletin D23, Portland Cement

Association, Illinois.
Concrete Society (2003), "Concrete Industrial Ground Floors -A Guide to design and
Construction", Technical Report No.34,3rd ed. Crowthome, UK.
Critchell, P.L. (1958), "Joints and Cracks in Concrete", 2d ed., Elsevier Applied Science,
London.
Croney, D. and Croney P. (1997), "Design and Performanceof Road Pavements",3' ed.,
Crovetti, J.A. and Darter M. 1. (1985), "Void Detection for Jointed Concrete Pavements",
Transportation ResearchRecordVol. 104,TransportationResearchBoard, pp 59-68.
Cudworth, D. (2000), 'The Dangers of Ignoring Asymmetric Dynamic Loading in
Industrial Floor "

Slabs. Concrete,Vol 35, No. 9, pp 25-27.
272
Cudworth, D. (2001), "Joint Load Transfer in Dynamic Conditions", Floors Forum, 10th
January 2001, LoughboroughUniversity, England.
Cudworth, D. (2003), "Detecting Voids Beneath Concrete Pavements using Surface

Deflection" Proceedings of the Symposium on Transportation Geotechnics, 11'
September2003, Nottingham Trent University, UK.
Darter, M. I., LaCoursiere,S. A and Smiley, S.A. (1979), "Structural DistressMechanisms

in Continuously Reinforced Concrete Pavement." Transportation Research Record,
Vol. 175, TransportationResearchBoard, pp 1-7.
Davich, P., (2000), "Prima 100 - Mn/DOT Testing procedure", Minnesota Departmentof
Transportation, Office of Materials and Road Research, Minnesota Road Research
Section, Minnesota, USA.
Davids, W. G., Turkiyyah, G. M. Mahoney, J.P. (1998), "EverFE - Rigid pavementThree-

Dimensional Finite Element Analysis Tool" Transportation ResearchRecord, Vol. 1629,
Transportation ResearchBoard, pp 41-49.
Deacon, R.C. (1990), "Fibres for Floors." Concrete,Vol. 24, No. 4, pp 19-20.
Deen, R.C., W.V. Azevedo, Rahal, A. S. and Havens,J.H. (1980), "Cracking in Concrete
Pavements",TransportationEngineeringJournal, Vol. 106, No.2, pp 155-169.
Duncan, J.M., Monismith, C. L. and Wilson, E.L. (1968), "Finite Element Analyses of
Pavements.", Mghway ResearchRecord,Vol. 228.
Fleming, P.R. (2000). "Impact Assessmentof Layer Granular Materials", PhD Thesis,
Loughborough University, England.
Frabizzio, M. A. and Buch N.J. (1999) "Performanceof TransverseCracking in Concrete

Pavements",Journal of Performanceof ConstructedFacilities, Vol. 13, No.4, pp 172-180.
Friberg, B.F. (1938) "Design of Dowels in TransverseJoints of ConcretePavement"

Proceedingsof the AmericanSocietyof Civil Engineers,
Vol.64,No.9, pp 1809-1828.
273
Fwa, T.F., Shi, X. P. and Tan, S.A. (1996), "Use of Pasternak foundation model in
concrete pavement analysis" Journal of Transportation Engineering, Vol. 122, No.4, pp

323-328.
FWD Users Group (2001). Electronic mail, FWD UsersMailing 727n38n421744n45.
Gregory, J.M. (1984) "Continuously Reinforced ConcretePavements",Proceedingsof the

Institution of Civil Engineers,Vol. 76, No. 1, pp 449-472.
Grzybowski, M. and Shah S.P. (1989), "Model to Predict Cracking in Fibre Reinforced
Concrete due to RestrainedShrinkage.", Magazine of ConcreteResearch,Vol. 4 1, No. 148,
pp 125-135.
Gulden, W. and Brown D. (1985), "Establishing Load Transfer in Existing Jointed

Concrete Pavements", Transportation Research Record, Vol. 1043, Transportation
Hammons, M. I. (1998), "Validation of Three-Dimensional Finite Element Modelling

Technique for Joints in Concrete Airport Pavements",Transportation ResearchRecord,
Vol. 1629, TransportationResearchBoard, pp 67-75.
Hannant, D. (1994), "Fibres in Industrial Ground Floor Slabs" Concrete,Vol. 28, No. 1, pp
16-19.
Harichandrian, R.S., Yeh, M. S. and Baladi, G.Y. (1990), "MICH-PAVE: A Nonlinear

Finite Element Program for Analysis of Flexible Pavements",Transportation Research
Record, Vol. 1286,TransportationResearchBoard, pp 124-131.
Helwany, S., Dyer, J. and Leidi, J. (1998), "Finite-Element Analyses of Flexible

Pavements"Journal of TransportationEngineering,Vol. 124, No.5, pp 491-499.
Highways Department (1999), "Falling Weight Deflectometer", HD29/94, Highways

Department, pp 511-5110.
Houde, J. and Mirza, M S. (1974), "A Finite Element Analysis of Shear Strength of
Reinforced Concrete Beams", Shear in Reinforced Concrete, ACI Symposium, Detroit,
USA, pp 103-128.
274
Howell, R.D. (1982), "Fibre Reinforced Concrete ProvesWorth for Airport Pavements",
Civil Engineering, Vol. 52, No.5, American Society of Civil Engineers,pp 52-55.
Hulett, T. (2001), "Floor Choices: Living with Joints and Cracks", Concrete, Vol. 35,
No. 3, pp 15-18.
Ibrahim, O.T. and Luxmoore A. (1987), "Control of Crack Width in Reinforced Concrete
Structure by the use of Fibres" Industrial Problems - Treatment and Control Techniques,
pp 499-511.
Illston, J.M. (1994), "Construction Materials: Their Nature and Behaviour", 2nded., E&
FN Spon, London.
Ioannides, A. M. and Korevesis G.T. (1990), "Aggregate Interlock: A Pure Shear-Load

Transfer mechanism." Transportation Research Record, Vol. 1286, Transportation
loannides, A. M., Lee, Y. and Darter, MI (1990), "Control of Faulting Through Joint
Load Transfer Design." Transportation Research Record, Vol. 1286, Transportation
loannides, A. M., Thompson M. R. and Barenberg, E.J. (1985), "Westergaard Solutions

Reconsidered." Transportation Research Record, Vol. 1043, Transportation Resrach
Board, pp 13-23.
Jackson, D.J., Murphy, M. R. and Wimsatt, A. (1994), "Strategies for Application of the
Falling Weight Deflectometer to Evaluate Load Transfer Efficiency at Joints in Jointed
Concrete Pavements"ASTM Special Technical Publication 1198),Vol. 2.
Jimenez, R., White, R.N. and Gergely, P. (1982), "Cyclic Shear and Dowel Action
Models", Journal of the Structural Division, American Society of Civil Engineers,
Vol. 108(ST5),pp 1106-1123.
Johnston,C.D. (1984),"SteelFibre ReinforcedPavementTrials" ConcreteInternational,

Vol. 6, No. 12, pp 39-43.
275
Kelley, E.F. (1939), "Application of the Results of Researchto the Structural Design of
Concrete Pavements",Journal of the American ConcreteInstitute, Vol. 35, pp 437-464.
Kim, S.M., Won M. and McCullough, B.F. (1998), "Numerical Modelling of

Continuously Reinforced Concrete Pavement Subjected to Environmental Loads"
Transportation ResearchRecord,Vol. 1629,TransportationResearchBoard, pp 76-89.
Knapton, J. (1999a), "In-situ Concrete Industrial Hardstandings", Thomas Telford,

London
Knapton, J. (1999b), "Single Pour Industrial Floor Slabs",ThomasTelford, London.
Krauthammer, T. and Western, K. L. (1988), "Joint ShearTransfer Effects on Pavement

Behaviour" Joumal of TransportationEngineering,Vol. 114,No.5, pp 505-529.
Kuo, C.M. (1995), "'Fhree-DimensionalFinite Element Model for PavementAnalysis of

Concrete PavementSupport", TransportationResearchRecord, Vol. 1505,Transportation
Kuo, C.M. (1998), "Study of Load Transfer Parameterin AASHTO Design Guide for
Concrete pavement", Transportation Research Record, Vol-1629, Transportation
Kwak, K., Suh, J. and Hsu, C.T. (1991), "Shear-fatigue Behaviour of Steel Fiber
Reinforced ConcreteBeams", ACI Structural Journal,Vol. 88, No.2,155-160.
Laible, J.P., White, R.N. Gergely, P. (1977), "Experimental Investigation of Seismic

ShearTransfer acrossCracksin Nuclear ContainmentVessels",ACI Special publications,
Vol.53, No.9, pp 203-226.
Larralde, J. and Chen W. (1987), "Estimation of Mechanical Deterioration of Highway

Rigid Pavements",Journal of TransportationEngineering,Vol. 113, No.2, pp 193-209.
Losberg, A. (1978), "Pavements and Slabs on Grade with Structurally Active

ACI
Reinforcement", Vol-75,
Journal, No.66, pp 647-657.
276
Meyerhof, F. (1962) "Load-Carrying Capacity of Concrete Pavements",Journal of the
Soil Mechanicsand FoundationsDivision, Proceedingsof the American Society of Civil
Engineers,Vol. 88, pp 89-116.
Millard, S.G. and Johnson R.P. (1984), "Shear Transfer Across Cracks in Reinforced
Concrete due to Aggregate Interlock and Dowel Action", Magazine of Concrete
Research,Vol. 36, No. 126,pp 9-2 1.
Minkarah, I.A., Cook, J.P. and McDonough, J.F. (1982), "Magnitude of Horizontal
Movement in Jointed Concrete Pavements",Transportation ResearchRecord, Vol. 821,
pp 61-67.
Moody, E.D. and McCullough B. F. (1993), "Long Term Performanceof Uncontrolled

Longitudinal Cracking and Failed Longitudinal Joints in Continuously Reinforced
Concrete pavements", Conference Proceedings20-22"d April 1993, Purdue University,
West Lafayette, Indiana.
Neal, F.R. (1996), "ICE Design and Practice Guides Concrete Industrial Ground
-
Floors", Institution of Civil Engineers,ThomasTelford, London.
Neville, A. M. (1995), "Properties of Concrete", 40' ed., Longman Scientific and

Technical, New York, USA.
Nishizawa, T., Fukada,T. and Matsuno, S. (1989), "A Refined Model of Doweled Joints
for ConcretePavementusing FEM Analysis", 4th International Conferenceon Concrete
pavementDesign and Rehabilitation, PurdueUniversity, West Lafayette, USA.
Nishizawa, T., Kajikawa, Y., and Fukada, T. (1993), "Effects of Lateral Distribution of
HeavyVehicleson FatigueCrackson ConcretePavements",5th InternationalConference
on Pavement
Concrete Designand Rehabilitation,
20-22nd
April 1993, Purdue University,
WestLafayette,Indiana,USA.
Nowlen, W.J. (1968) "Influence of Aggregate Particles on Effectiveness of Interlock

Joints in Concrete Pavements", Journal of the PCA Research and Development
Laboratories,ReportD139,Vol. 10,No-2.
277
Ozbeki, M. A., Kilareski, W. P. and Anderson,D.A. (1985), "Evaluation Methodology for
Jointed Concretepavements",TransportationResearchRecord, Vol. 1043,Transportation
Papagiannakis,T., Oancea,A., Ali, N., Chan, J. and Bergan, A. T (1991), "Application of

ASTM E1049-85 in Calculating Load Equivalence Factors from In-situ Strains",
TransportationResearchRecord,Vol. 1307,TransportationReseachBoard, pp 82-89.
Pearson, D. (1999), "Industrial Floor Slabs Towards a Performance Specification",

Proceedingsof 3rd EuropeanSymposium on Performanceand Durability of Bituminous
materials and Hydraulic StabilisedComposites,Leeds,U.K. pp 629-611
Perenchio, W.F. (1997). "rhe Drying Shrinkage Dilemma." Concrete Construction,

Vol. 42, No.4, pp 379-383.
Petterson, D. and Alemo J. (2000), "Characterization of Restraint from Friction Tests

with SlabsCast on Ground", Structural Concrete,Volume 1, No.4, pp 181-187.
Poblete, M., Valenzuela, R. (1988), "Load Transfer in Undoweled.TransverseJoints of

PCC Pavements" Transportation Research record, Vol. 1207, Transportation Research
Board, pp 3949.
Porter, M. L., Hughes, B. W. and Bames, B.A. (1996), "Fiber Composite Dowels in
Highway Pavements".SernisequicentennialTransportation ConferenceProceedings,May
1996,Iowa StateUniversity, Ames, Iowa, USA.
Portland CementAssociation (195 1), "Concrete PavementDesign for Roadsand Streets",

Portland CementAssociation.
Pradhan, M. M. (2002), "Evaluation of Rigid Pavement Condition using the Failing
Weight Deflectometer", Bearing Capacity of Roads,Railways and Airfields, Lisse, Swets
andZeitlinger,pp 43-52.
Prozzi, J., DeBeer, M. and Balmaceda, P. (1993), "Non Destructive Tests Procedurefor
Field Evaluation of Transverse Joints in Concrete Pavements", 5th International
Conferenceon ConcretePavementDesign and Rehabilitation, April 20-22d 1993,Purdue
University,WestLafayette,Indiana,pp3-12.
278
Raja, Z.I. and Snyder, M. B. (1991), "Factors Affecting Deterioration of Transverse
Cracks in Jointed Reinforced Concrete Pavements", Transportation ResearchRecord,
Vol. 1307,TransportationResearchBoard, pp 162-168.
Ricci, E.A., Meyer, A. H., Hudson, W.R. and Stokoe II, K. H., (1985), "rhe Falling
Weight Deflectometer for Non-destructive Evaluation of Rigid Pavements",Centre for
TransportationResearch,Report 387-3F, Bureauof EngineeringResearch,Austin, Texas.
Ringo, B. C. and Anderson, R. B. (1996), "Designing Floor Slabs on Grade", The

AberdeenGroup, Illinois, USA.
Road Research Laboratory (1955), "Concrete Roads - Design and Construction",

Departmentof Scientific and Industrial Research.
Robins, P.J., Bishop J.W., Austin, S.A. (2002), "Early Age Finite Element Modelling of
Industrial Ground Floors", Draft Copy, LoughboroughUniversity, England.
Rogers, M. (2000). "Joints", Draft Report for The Concrete Society, 27th
section of a
September2000, ConcreteSociety.
Rollings, S.R. (1993), "Curling Failures of Steel-Fiber Reinforced Concrete Slabs"
Journal of Performance of Constructed Facilities, Vol. 7, No. 1, pp 3-19.
Savage, R. J. (1985) "Dynamic Failure of Joints in Reinforced Concrete Pavements.

"
Concrete,January 1985,pp 16-18.
Schrader, E.K. (1985), "Fiber Reinforced Concrete Pavements -A State of the Art
Report", 3-5thJune 1985, Steel Fiber Concrete - US-SwedenJoint Seminar (NSF-STU),

Stokholm, Sweden.
Scott Wilson Pavement Engineering (2002), "Structural Evaluation with the Failing
Weight Deflectometee,Internal Correspondence, Scott Wilson PavementEngineering,
Nottingham,England.
Simpson,D. (2001a)."Cracksin ConcreteGroundFloors". Floors Forum, 10thJanuary

2001, Loughborough University, England.
279
Simpson, D. (2001b) "Thickness Design of Ground-SupportedFloor Slabs" Concrete,
Vol. 35, No.8, pp 22-24.
Sprigg Little Partnership (2000), "Designing Cracks out of Floors", Seminar on

Developments in Industrial Ground Floor Slabs, 31" October 2000, Aston University,
England
Stock, AR (1988), "Concrete Pavements",0 ed, Elsevier Applied Science,London.
Suprenant,B.A. (2002), "Why Slabs Curl", ConcreteInternational, Vol. 24, No.3, pp 57-
61.
Sutherland, E.C. and Cashell, H.D. (1945), "Structural Efficiency of Transverse

Weakened-PlaneJoints" Public Roads,Vol. 23, NoA pp 88-97.
Swamy, R.N., Al-taan, S., Sami, A. R.A. (1979), "Steel Fibres for Controlling Cracking
and Deflection." ConcreteInternational, August 1979,pp 41-49.
Swamy, R.N. and Andriopoulos A. D. (1974), "Contribution of Aggregate Interlock and

Dowel Forces to the Shear Resistanceof Reinforced Beams with Web Reinforcement"
American ConcreteInstitute, Special Publication, pp 129-145.
Tabatabaie, A. M. and Barenberg E.J. (1980), "Structural Analysis of Concrete

Pavements",Journal of TransportationEngineering,September1980,pp 493-505.
Tang, B. (1993), "Structural Evaluation of Airfield Rigid pavementsusing Failing Weight

Deflectometer", Journal of TransportationEngineering,Vol. 119, No.3, pp 467-476.
Teller, L. W. and Sutherland, E.C. (1943), "The Structural Design of Concrete

Pavements",Public Roads,Vol. 23, No.8, pp 167-212.
Thompson, 1. (2001). "Use of Steel Fibres to Reinforce Cement Bound Roadbase",PhD

Thesis, The University of Nottingham, Nottingham, England.
U.S. Departmentof Transportation,F.H.A. (1990), "ConcretepavementJoints", U.S.

Departmentof Transportation,
FederalHighwayAdministration,pp 1-11.
280
Valle, M. and Buyukozturk, 0. (1993), "Behaviour of Fiber Reinforced High Strength
Concreteunder Direct Shear", ACI Materials Journal, Vol. 90, No.2, pp 122-133.
Van-Wijk, A. J., Larralde, J. and Lovell, C.W. (1989), "Pumping Prediction Model for
Highway ConcretePavements",Journal of TransportationEngineering,Vol. 115,No.2, pp
161-174.
Verhoeven, K. (1993), "Cracking and Corrosion in Continuously Reinforced Concrete

Pavements", Proceedings of the 5th International Conference on Concrete Pavement
Design and Rehabilitation, 20-22d April 1993, Purdue university, West Lafayette,
Indiana pp 201-209.
Wade, M. J., Cuttell, G.D., Vandenbossche,J.M., Yu, H.T., Smith, K. D., Snyder, M. B.
(1997), "Performanceof ConcretePavementsContaining Recycled ConcreteAggregate",
Report FHWA-RD-96-164, Federal Highway Administration, McLean, Virginia, USA,
pp 296.
Walker, W.W. and J.A. Holland (1999), "Thou Shalt not Curl Nor Crack (Hopefully)",
...
ConcreteInternational, January 1999,pp 47-53.
Walker, W.W. and I. A. Holland (2001), 'Tlate Dowels for Slabs on Ground - Dowels for
the 21st Century", Draft Copy, Loughborough University.
Walraven, J.C. (1981), "Fundamental Analysis of Aggregate Interlock." Journal of the

Structural Division, American Society of Civil Engineers,Vol. 107, No. 11, pp 2245-2270.
Watson, J. (1994), "Highway Construction and Maintenance", 2d ed., Longman

Scientific and Technical, Harlow, England.
H.M. (1926), "Stressesin ConcretePavementsComputedby Theoretical

Westergaard,
Analysis" Public roads, Volume 7. No.2, pp 25-35.
Westergaard, H.M. (1947), "New Formulas for Stresses in Concrete Pavements of

Airfields" Proceedingsof the American Society of Civil Engineers,Vol. 73, No-5, pp 687-
701.
281
White, R.N. and Holley, M. J. (1972), "Experimental Studies of Membrane Shear
Transfer", Journal of the Structural Division, American Society of Civil Engineers
98(ST8), pp 1835-1853.
White Young Green (2002), "Ground Floor Slab PavementEvaluation - Diversey Lever",
Internal Conunumication,LoughboroughUniversity, England.
Wirtgen (2001), "Competencein concretepaving machinery: Wirtgen slipform,pavers in
operation" CompanyBrochure.
Yoder, E.J. (1959), "Principles of PavementDesign", John Wiley & SonsInc, Chichester,
England, pp 435-452.
York, D. (2001), "New Slant on Slab Foundations." Concrete,Vol. 35, No. 1, pp 54-55.
Ytterberg, R.F. (1987), "Shrinkage and Curling of Slabs on Grade: Part I- Drying
Shrinkage", ConcreteInternational, April 1987,pp 22-31.
Zollinger, D.G. and Barenberg, E.J. (1990), "Mechanistic Design Considerations for
Punchout Distress in Continuously Reinforced Concrete PavemenV', Transportation
ResearchRecord Vol. 1286,TransportationResearchBoard, pp 25-37.
Zollinger, D.G. and SenadheeraS.P. (1994), "Spalling of Continuously Reinforced

ConcretePavements"Journal of TransportationEngineering, Vol. 120,No.3, pp 394-410.
282
APPENDIXA
Steel Fibre Data Sheets

DrambC I
J
J
1. General
B
preferably use a central batching
I B N plant mixer
recommended maximum dosage:
Low Bdght C
Max.
Max.
aggregate
Dosage (kg/rrM
size (MM)
(mm) pour F pump
8_
8 160 120
Description: fibres are filaments of
Dramixe 16
32
100
80 1
75
wire, deformed and cut to 60
lengths, for reinforcement of
concrete, mortar and other
a continuous grading is preferred
composite materials.
DramixORL-45/50-BN is a
loose cold drawn wire fibre,
with hooked ends.
2. Fibre addition
2.1. In batching plant mixer
" App ications:
- industrialfloors
never add fibres as first component
- slabs on vibrocompacted piles
in the mixer
" Geometry: yl fibres can be introduced together with
Performance sand and aggregates, or can be added
Length (1) class: 45 in freshly mixed concrete
Aspect ratio y/ only for drummixer: unopened degradable
50 mm
(= I/d): 48 bags can be thrown directly in the mixer
*,WOOOrDiameter
(d) I
I 7
1,05 mm
2800 fibres/kg
2.2. Truckmixer %1:
1170b
y/ run mixer at drum speed: 12-18 rpm
Tensile strength: adjust slump to a min. of 12 cm
- on the wire: 1000 N/mM2
low carbon conforms to: - DIN 17 140-139 (preferably with water reducing agents or
-
- EN 100 16-2 - C91) high water reducing agents)
add fibres with maximum speed of
* Coating: None 60 kg/min
unopened degradable bags can be added
e Approvals: Quality System in
Belgian BrasIllan, American provided that drum speed Is min. 12 rpm
Plants and Czech Plants y( optional equipment: belt-hoist elevator
Conforms to
after adding the fibres, continue mixing at
highest speed for 4-5 min. (:t 70 rotations)
ASTMA820
Product Product
Belgium Russia rable
(01 t; L9 -
TC-07-0116-98
Delivered In
SlovakRepublic gikiiPoland
IA,I F,
7,110 &0 M IS 7
Protect the pallets degrad big bag

CzechRepublic Romania Do not stackp
't the
b
bagsof 20 kg 900 kg
against rain pallets on top 0
of
101,0* on pallet
&III]F/111 eIach other 1200kg
Turkey Germany
N.V Bekaert SA - Bekaertstraat2- 8550 Zwevegem- Belgium
Tel. +32 (0) 56 / 76 69 86 - Fax +32 (0) 56 / 76 79 47
Internet: http: //www. bekaert.com/building
Technical data: moved. ANdetailsdescfte our productsIn generalform
ValuesareIndicativeonty.Modifications
For industrialfloors, ...ask for specializeddocumentation only.Fororderinganddesignonlyuseofficialspecifications
anddocuments.N.V BskawtSA 2002
111:4Ito r. kqEel
f7_10101-til Cl
DramV General
preferably use a central batching
CBN plant mixer
BrIght
LOW
Carbon Max.
Max.
aggregate
aggregate
Dosage (kg/ml
Dosage (kg/ml
size (mm)
size(MM) pour pump
pour pump
Description: DramixO fibres are filaments of 8
8 60
60 45
45
wire, deformed and cut to 16
16 50
50 35
35
lengths, for reinforcement of 32 35 30
32 35 30
composite materials. DramiO a continuous grading is preferred
RC-80/60-BN is a cold drawn mix until all glued fibres are separated into
wire fibre, with hooked ends, and individual fibres. Fibres don't increase
glued in bundles. mixing time significantly.
if special cements or admixtures are used,
Applications: a preliminary test is recommended
- floors
jointless - liquid tight floors
suspended ground slabs - overlays
- jointless floors on - pavements 2. Fibre addition
vibrocompacted plies - segmental linings
industrial floors - compression layers 2.1. In batching plant mixer
-
- slabs on vibro- - cellar walls never add fibres as first component
compacted piles - precast in the mixer
y/ fibres can be introduced together with
Geometry: sand and aggregates, or can be added
9 Performance in freshly mixed concrete
: C
Length (1) class: 80 for drummixer: unopened degradable
y/ only
60 mm Aspect ratio bags can be thrown directly in the mixer
(= Vd): 80
(d) 2.2. Truckmixer'tj
IoDiameter 1 4600 fibres/kg
0,75 mm y/ run mixer at drum speed: 12-18 rpm
adjust slump to a min. of 12 cm
(preferably with water reducing agents or
Tensile strength: high water reducing agents)
- on the wire: minimum 1050 N/mM2
- low carbon conforms to: - DIN 17 140-D9 add fibres with maximum speed of
EN 10016-2 - C9D 40 kg/min
-
unopened degradable bags can be added
e Coating: None provided that drum speed is min. 12 rpm
V optional equipment: belt-hoist elevator
y/ after adding the fibres, continue mixing at
-PApprovals: Quality System in
highestspeed for 4-5 min. (:t 70 rotations)
Belgian Brosillan,American
Plants and Czech Plants 'Z? 'a
Conformsto 2.3. Automatic dosing
Fibres can be dosed from b6lk at rates
ASTMA820 Y(
from 0 up to 3,5 kg/sec with a specially
Product developed dosing equipment
Product
Belgium Russia
TC-07-0116-98
Delivered In
The Netherlands Poland
r
Romania I
Protectthe pallets Donot stackthe degradablebigbeg
I bagsof20kg 1100kg
againstrain palletsontop of onpallet
L--- 7-Germany eachother 1200kg
Turkey
Z-71.4-3 N.V. Bekaert SA - Bekaertstraat2- 8550 Zwevegern- Belgium
OEM- Tel. +32 (0) 56 / 76 69 86 - Fax +32 (0) 56 / 76 79 47 8
Technical data: Internet: hftp: //www. bekaert.conVbuilding

For Industrial floors, floors on vibrocompacted piles, Values are Indicative orgy. Modifications reserved. AS details dewAbs our Products In general form
jointless floors... ask for specialized documentation. OnlY.For ordering and design only use official specificatiors and documents, N.V. Belutert SA
2002
M
=111111
191W=111111
111:4110111111 1IPAI I Le
DrambC I
1. General
preferablyuse a central batching
RCBN plant mixer
I"
BBdght Carbon Max.
Max.
aggregate
size (mm)
Dosage (kg/ml
(kg/rnJ
pour pump
urnp
Description: DramiO fibres are filaments of a
8 110 80
wire, deformed and cut to is 70 55
Volk lengths, for reinforcement of 32 60 45
composite materials. Dramix* Y/a continuousgrading is preferred
INC RC-65/60-BN Is a cold drawn
y/ mix until all glued fibres are separatedinto
14 wire fibre, with hooked ends, and individualfibres. Fibresdon't increase
glued in bundles.
mixing time significantly.
if special cementsor admixturesare used,
Applications: segmental linings a preliminarytest is recommended
-
- slabson - cellarwalls
vibrocompacted piles - pavements 2. Fibre addition
- liquid
tightfloors floors
- jointless
- floors
Industrial - jointless
floors on 2.1. In batching plant mixer
- overlays vibrocompacted piles
- outdoor slabs never add fibres as first component
- piles foundation
Y/
- suspended groundslabs - slabs in the mixer
y/ fibres can be introduced togetherwith
Geometry: sand and aggregates,or can be added
Performance in freshly mixed concrete
: C
Length class:65 onlyfor drummixer:unopened degradable
60mm Aspectratio bagscan be throwndirectlyin the mixer
4p' (= Vd):67 tjtT,
"K=O 2.2. Truckmixer' -a
(d)
Diameter II 3200fibres/kg run mixer at drum speed: 12-18 rpm
0,90 mm
adjust slump to a min. of 12 cm
(preferablywith water reducingagents or
Tensile strength: high water reducingagents)
- on the wire: minimum 1000 N/mm'
- low carbon conforms to: - DIN 17 140-D9 add fibres with maximumspeed of
- EN 10016-2 - C913 60 kg/min
unopeneddegradablebags can be added
providedthat drum speed is min. 12 rpm
fo Coating: None
optional equipment: belt-hoist elevator
after adding the fibres, continuemixing at
e Approvals: Quality System In
highest speed for 4-5 min. (:t 70 rotations)
Belolan Brasillan.American
Plants and Czech Plants
Conforms to C4
2.3. Automatic dosing
Fibres be dosed from but
ASTMA820 v( can ra5tes
ates
from 0 up to 3,5 kg/sec with a specially
lly
Product Product developeddosing equipment
6035ffiRi
Belgium Russia
TC-07-0116-98'i
Delivered
vere In
The Netherlands Poland
[BE F. AIOMV. 4161rl
.P
Romania
L16311 I
Protectthe pallets Do not stack the
le degradable bigbag
bagsof20kg 1100kg
againstrain palletsontop of
Of
onpallet
Turkey Germany eachother 1200kg
I Z-71.4-3 N.V. Bekaert SA - Bekaertstraat2- 8550 Zwevegem- Belgium

Tel. +32 (0) 56 / 76 69 86 - Fax +32 (0) 56 / 76 79 47
Technical data: Internet: http: //www. bekaert.com/building
For Industrial floors, floors on vibrocompacted piles,
rewv9d. Aj detailsdescribeour prod" In gwwsl form
ValuesareIndicativeonly.Modifications
jointless floors... ask for specialized documentation. or*y.Fororderinganddesignor*j us@ anddocuments.
officialspedfications N.V.8~ SA 2002
APPENDIX B
Laboratory Test Data

E
E
CRCP(R Iq CR(q (P CR
CC!
U)
; W,
ca
Im
0
00 C1,
, C) U) 0 C
C
0 C)
C
0 0
) 0
Lf)
(0, r- C) 0 Lf)
0 o
LO C) 0
Lr)00
C\1 CC)CM 0 C') -
r C\1 r- Lr) co
C\I
)
C\I U,
. T.- co CDU)
. .
co
0 S
c C)
a 0
:)0 " (C:,
)0
4) (D
-0
C)
C)
C)
C) CD 0 C) Lf)
0 Q LO LO I-
0
LO
C) C)
0 LO

C)
>% r- cm q co
...
E Lf) C.
\41 LO LO Lf) LO LO Lr) C3)
Lf)
Ln Ce) LO U) r, N Lo LO
r- C\j 0) r OD Lr) 00 CIO Ul C.C co (31) co C\I LO I- C\1 04 U) Itt r- "T C) w C\j C*4 C\j C\j I"
a N co I- U! 9 R CD C\j cr ) U-) , ,
6 C3 (=; C:i q 9 C? -. -: Ci C5,4t q -: - - - - C4
8 8 c!
(=; CR CD C O O
.,
cs CDc; 0 0 Ci 0
c o O C)
0 0
IIII III I
,,v-
cc
0
JU 0
0 0 0 0 0 0 0 0
c O O O O O O O CD O O C) C:) O (D O O
0 C) o o (c! o
0
C:) 0
o o
0 0 40 0
C) Q C) 0 C) 0
c) 0 C)
c) C)
C) C)
0 C)
c) Lf)
CDC)0 Q0 0 0 0 W
-a 0 C) Lf) C) C)
0 0
Q C) Q 0
Q 0 C)
C)
0 C)
Q (D C) LO C) LO N. C) Q
C:,C:) CD 0 (n
0 0
C) C) CD C) 0 Lf) CD C)
0
LO (D
0 Lf)
C I-
LO
LO 0 t- CJ C) (D 0 Lf) 0 04 CIJ 04 I- Lo M Lo C\J
C\1 Co 0) 4, Cf) Co) C\1 04 to N
cm 0 a CM 0 OJ 4) - C14 0 0) C\I 4) C\I 4) " C\J M a N 0

N
CR im cn cn tm CYI cy) cn
0 cc E Cc La 2 e
0) , a) 0) a) 0) C) (1) a) :1 41) !1 i 0)
4) . > > > >
Q (D
00
CY) C) C;
0)
CY) :!Et , " - ! CO
co
:! I
r_ C) 0
C) 0 C) 0 0 CO 0
C\j C\j N !j CY) CY')
ca CD C co C\j ,, 6 C)
-c: C C - o T
c C: 0 (D CD (D , CD (D
0 L LO L
.2 z
0 z
0
L6
Lh
U th
,
I Lfl)
I
Lf) I
CD LO
lop
C) 0 CD 0 C) C)
0 0 C)
Q)
U-
FD
co (.0
E
L) Cc
0 0 C)
(1) Z CV) C\I I t I CO
(.0 (D CD (.C) co (D CD (D (D
C;
0 0 C) 0 C) 0
0 C) 0 0 0
C) CD U) 0 0 C) 0
CD 0 (D 0 0 C)
0 LO CV LO Lr)
C) C:) (D C\j (D Lr) Lr)
Cl) 04 19t Cl) Cl)
N N cm
U) LO Lf) (D (D
rl- CR (q
"R 6 C;
0 CD 0 CD 0 o
CD 0 0 0 C 0 0
0 0 0 0
CD CS 0 0 CD 0 0 0
CD 0 0 CD 0 0 0
CD C, Lr) CD Lf) C) 0
CD 0 0 0
N ' IT CM C\J C-4
C%j C\j CN 04
I 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
li l l I II I 1 1 1
U') LO Lf)
U') LO Ln Lr) Lf) 00 Lf)
Ul) LO Lf) 04 C\j 00
LO (D 00 Lc) "t r- 00 ;: 0 r- C'4 0
Cl) C. Cl) C") LO C,) C4
C\j C,4 (0 (D C) - "I- C) to
C 9 CD C:) N m
- 0 -
6 6 co?q 9 q 9 -- 6 C5c; C5 C)
R c; 6 o . 7 C:,
C5
(D CD C) Cl CD
C, c i CDc) o CD (=; C)
Ci (::) C: ) C)
) 0 C: ci C58 0
I I I I I I I I I I I I I
CD
0 o 0 0 C) CD 0 CD
0 CD CD
0
0
C)
0
0
CD 0 0 0 O O O C) C:) O o o o o 0 0 0 0 C) 0 0 0 0 C) C)
0 0 0 0 0 CD 0 0 0 0 C) CD 0 C) 0 0 C) CD CD CD 0 0 C) 0
C) 0 0
0 0 0 CD CD 11) 0 0 0 0 0
CD Ln 0 0 Lf) 0 0
LO 0 0 0 0 C) 0 0 CD CD 0 0 C:) 0 LC) CD 0 C) C)
COD0 0 0 N
0 LO U) co (D C\j C:) - 0 LO TI - CD 0 0 LO 0 0 LO Li) Lf) Lf) 0 0 LO r- 0 Lf) LO LO
\j C\j C,4 LO
CN LO CO Lr) ItT C\J LO C\l M Lf) M LC) N M N N CN "T cn LO N
-
N 0 - N - N 0 - N - N 0 N 0 - N CO 0 - N 0 N M 0 N M 0
tm M
(%S
cm
m
(p
cts
M
to
M
(13
tm
m
tm
m
M
(a
cm
m
00
CY) Cl) CNJ
(D CY)
co
0) CY? i
(.0 co
0')
CO iQ
CO CY) C-q
LO
4 L6 -4 CY5
Z!
0 C) C) 0 0 CD CD
CY) T :t IzT
CD
(D
0
w
C) C)
(D
C) C) C)
Ll?
(D
LO 0
LO
LO 15) 0 (9 LO
p LO
(C) (D
LO
(D Ln
co co 0:)
-
C)
! ? LO L
CD Cl
CD Cl 0 CD CD
LL
(D LO
C) LO
C? 17
F' C)
.
co co (D
C) 0 0
C) 0 0
0 C) LO
0
00 0) co
U)
to r-
Ci
C) IR q:t
0 C) C)
0 0 Lf)
0 U) 04
C) C, "
Iq Go1(D l
l i l l l i l l l i l l 1 1 1 1 1 1 1 1 1 1 1 1 I
Lf) N LO (D LO Ln 00 Lf) U) 00 0 04 (D Ln
Cr) 00
U')
'It Cf)r, r, r_ LO Cf)
0 CV) 0 0 0 C) 0 C:>qe ,t 't
CM N
.t IT (y) Lf) t CD co c) m U.
) U) in U-)'*'
CD 0 CD 0 C) CD 0 C) 0
.
U) ", -t cc)co N m 04 r-
T-
0
q 0 qzt, c-)o C5R c; c; ci 6 c; c; ci q C;
8 c;
0
6 C5 cs C3 c; Ci (=; Ci c; o
ci 8
8 8 8 CD
C)
C) C) 0 C) 0 0 0 0 0 C) 0 C) (: ) 0 0 C) 0 0 0
0 0 0 C) 0 0 0 C:) C) 0 0 0 C) C)
0
C) C) 0
0
0
0
0 0 0 0 0
0
0 0 0 0 C) 0
C)
c) o CD 0 0 0 c) o a a
CD C) 0
o
0
o
0 0
Lcn g )
0 CD
u)
0
o 0 C) 0
a C:) 0 CD 0 C) C:> Lr) C) 0 C) C) Lc) 0 o C)
C) 0 C) U.) I- Ul) Lf) C) LO C) 04 C) 0 C> Lr) c) c:> cD 0 C) CD 0 0 U) C4 (=)
C) c,4 ) 0 Ce) Iq LO
C) 04
qq C14 Lr) Lf) N 04 LO V) C14 Cl Lo C') U.) Lo Lo LO CO) 11 r- co LO
CM cr) 0 - CM 4) r- CM 4) - CIJ Cl) Q - CM 0) - cm ce) 4)

12) - 0 N 0 CM a -r- CM 0 -
Im im cn im I cn cn I cn cn IS ,
cn 1 V)
4)
4 < < < < .4
< < < cc 1 I I
Z!
'Kt il-5 - (0 co
Iq - (X)
- - N (0
co
co
CY) (D
I Cr> ,
C
C Cf)
C C6
(q
1-: cyi T , a I - :
RE)
co
C) 0
o
o
0
C)
-a C)
73 (0 0
cri
h-
I
V co V)
a) 1 2(3) U)
(1) LO 6 1) (0
t 2 Lf) (0 LO
(D (D Lb
I
,, , 0 , (D 2i , * 0 , t (0
0 0 0 CD 1
(Z C: '0
m cz 0
o
U-
E
'E C\l cr
0i
(.0 (0 (D
o Ul) 0 Q
C) 0 0 0
0 r- rl-
o co (4)
cm
to co q*
c; cs Ir;
C) C) 0 0 0
C) C:) 0 C) C)
C) 0 Co C) CD
C) rl_ 0 0
co
C*, (D w In LO CC) Lr) Lr)

C\l * Lg-)
C) C,
4 N (D o LOCO cm
,
C',
N "': ,
CD c; C;
7-
C4C5 ':
0 0
.

j
(: C4
i 7 C4 (7)
c) c) 0 CD
ci C!
(D -*
0
C5 (:
(D C:) (: ) o o c) (D c) C) 0 C) C:) 0 0 Q 0
o c: o C:) C> (: ) CD C) C:l o 0 0 o 0 0 0 0
C) 0 C C:) Q 0 0 C) c:) LO C) C) Ul
CD CD Lf) Q C) 0
0 0) LC) LC) C) LO C) CM LO LO Q Q 1'- LO f
00 C-4 C14 LO U) 04 04 -
, C%jC,4 Cj -CTcl) N - -
II I I
C\j 4) - Cki (D
1111 1111 11
N W ckj 4) - cm d) I- N (D
ch cp
.- - -
I CY)
to
CY)
I
CA CY)
00 (D co co
1
(7)
IR C
0 C)
cu
(Z cz (z CO
a- CL CL a-
IL ,
co 0
C:)
Y (D (D (0
I I I? I
Lb
LO LO
r Lb
0 C,
(D
7a-) _2 -a-) 7F)

U- I
,
cz C') CZ co -Fo
L-
CES m - (Z , 'r
,u 0-
Q- -0) 0-
0
Lj

Load Transfer Across Joints

Uploaded by

Copyright:

Available Formats

Load Transfer Across Joints

Uploaded by

Document Information

Original Description:

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

Load Transfer Across Joints

Uploaded by

Copyright:

Available Formats

Loughborough University

Load transfer across cracks

A Doctoral Thesis. Submitted in partial fulllment of the requirements

Metadata Record: https://dspace.lboro.ac.uk/2134/7581

Publisher: c Stuart John Arnold

Please cite the published version.

For the full text of this licence, please go to:

Stuart John Arnold

A Doctoral Thesis Submitted in partial fulfilment of the requirements for the

@Stuart John Arnold (2004)

multiple cycles of load. This provides an increasedunderstandingof concrete slab on

using embeddedstrain gauges,coring and surfaceprofile levelling have uncoveredthat a

effect towards the surface.

years,without whom it would not have beenpossibleto completethis thesis.

remarkably pain free.

a source of continual strength. They have supported and encouraged me

5.1 Introduction 113

Table 2.1 - Typical Industrial WarehouseLoading Values (Knapton 1999b) 10

acrosscracks andjoints ....................................................................................................

cycles (Thompson 2001).................................................................................................. 137

and Humphrey 1967).......................................................................................................

width measurements 141

Figure 5.22 - Effect of undertakinga discontinuouscyclic load test..............................144

Figure 5.24 - Positive and negativedisplacementdeteriorationplot ..............................145

Figure 6.5 - Crack offsets from corestaken in visit I ..................................................... 190

Figure 6.6a - Strain GaugePositions (Leeds) 190

Figure 6.6c - Strain GaugePositions (Marston).............................................................. 191

Figure 6.28 - Comparisonof load transfer and load step

with 5.94mm crack width) ...............................................................................................

with 5.94mm crack width) ...............................................................................................

ratio 80 with 1.98mm.crack width) ..................................................................................

ratio 80 with 4.62mm.crack width) ..................................................................................

ratio 48 with 1.98nim crack width) ..................................................................................

ratio 48 with 4.62mm crack width) ..................................................................................

with 1.98mmcrack width) ...............................................................................................

with 0.66mm crack width) ...............................................................................................

with 1.98mmcrack width) ...............................................................................................

slab loaded at the edge.....................................................................................................

slab loaded at the edge.....................................................................................................

slab loaded at the edge.....................................................................................................

model and in-service slab response..................................................................................

model and in-service slab response..................................................................................

a Contact radius of a load

a, Distancefrom comer to load centre

9 Acceleration due to gravity

a, Thermal coefficient of expansionfor concrete

a, Vertical movementcausedby slab curling

P Poissonsratio for concrete

There is a growing demandfor in-situ concreteindustrial floor slabs and hardstandings

excessivedeformationor further cracking.

Many methodsare employedfor load transfer.Someutilise the intrinsic propertiesof the

own. Similarly, with steel fibres becomingincreasinglypopular. it is important that their

respectto load magnitude,crack orientation and long-term fatigue is thereforeessentialto

enabledesignersto accuratelypredict slab response.

1.2 Aims and Objectives

period. Thesediscussions along with a comprehensivereview of literature identified the

need to 'develop a more fundamental understandingof the load transfer mechanisms

acrosscracksand joints in concretefloor slabs,hardstandings

In order to achievethis overall aim the following objectives were identified:

A Doctoral Thesis. Submitted in partial fulllment of the requirements