Development of an expert system for reinforced concrete bridge repair.

GREEN, Laurence F.

Available from the Sheffield Hallam University Research Archive (SHURA) at:

http://shura.shu.ac.uk/19721/

A Sheffield Hallam University thesis

This thesis is protected by copyright which belongs to the author.

The content must not be changed in any way or sold commercially in any format or medium
without the formal permission of the author.

When referring to this work, full bibliographic details including the author, title, awarding
institution and date of the thesis must be given.

Please visit http://shura.shu.ac.uk/19721/ and http://shura.shu.ac.uk/information.html for

further details about copyright and re-use permissions.
 j Adsetts Centre City Campus
Sheffie|d_ Si 1WB

101 859 889 8

Return to Learning Centre of issue

Fines are charged at 50p per hour

REFERENCE
 ProQuest Number: 10697023

All rights reserved

INFORMATION TO ALL USERS

The quality of this reproduction is d e p e n d e n t upon the quality of the copy subm itted.

In the unlikely e v e n t that the a u thor did not send a c o m p le te m anuscript

and there are missing pages, these will be noted. Also, if m aterial had to be rem oved,
a n o te will ind ica te the deletion.

uest
ProQuest 10697023

Published by ProQuest LLC(2017). C opyright of the Dissertation is held by the Author.

All rights reserved.

This work is protected against unauthorized copying under Title 17, United States C o d e
M icroform Edition © ProQuest LLC.

ProQuest LLC.
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, Ml 4 8 1 0 6 - 1346
Development of an expert system for reinforced concrete bridge repair

Laurence Ferguson Green

A thesis submitted in partial fulfilment of the requirements of

Sheffield Hallam University
for the degree of Doctor of Philosophy

October 2005
 Abstract

Current techniques for selecting reinforced concrete repair materials are often based on ad-
hoc methods for specifying repair material properties. The inherent lack of understanding
of material behaviour in this approach can lead to premature failure of repairs.

This research has examined state of the art methods for repair material property
specification and has developed a technique, specifically for application in a computer
program, which recommends optimum repair material properties tailored to given repair
situations. The technique developed achieves compatibility between the repair material
and the substrate concrete through a sophisticated balancing of those repair material
properties identified as important, specifically; elastic modulus, shrinkage, creep and
tensile strength. Adopting the developed technique minimises the possibility of failure of
the repair material.

The developed repair material property selection technique is seamlessly integrated into an
expert system fo r reinforced concrete bridge repair also developed as part of this research.
A technique has been produced to quickly elicit the complex decision making process of
reinforced concrete experts and represent their information in a computer program.

The developed expert system diagnoses the causes of reinforced concrete defects.
Importantly, the program utilises its in-built intelligence to determine if the severity and
extent of the defects identified warrant genuine concern.

In order to facilitate efficient inputting of data into the expert system by prospective users,
an elemental graphical interface was developed, allowing users to quickly assemble on­
screen three dimensional representations of the affected concrete elements. Thereafter,
program users locate areas of defects onto the on-screen concrete elements and the inputted
data can be interrogated by the expert system.

Adopting the mainly graphical approach of data input, the expert system diagnoses
reinforced concrete defects, proffers prognoses for concrete elements themselves (such as
piers, columns, abutments), recommends testing regimes to confirm the expert system
output, and recommends repair techniques.

Should the recommendation of the expert system be to break out and replace defective
concrete, the technique to recommend optimum repair material properties, developed in
this research, will offer its recommendations.

The developed expert system for reinforced concrete repair acts as an expert guide through
all aspects of bridge inspection and repair. For the assessment of defects it draws together
best practice recommendations from literature and experts. For the recommendation of
repair material properties it implements the technique developed in the research.

The completed research has been incorporated into a commercially available bridge
management system (www.bridgemanagementexpert.com).
 Acknowledgements

The author would like to thank the following individuals and their organisations for their
input and advice on the advisory group for this research:

Research Engineers Europe:

Saeid Naelini
Will Thomas
Mark Sutton

Mott Macdonald:

Gerry Kelly
John Simpson
Dr Paul Lambert

The valuable input of the following organisations is acknowledged:

V.A Crookes Ltd

Flexcrete Ltd
M.J.Gleeson Group PLC
Highways Agency

The author is indebted to Professor Pritpal Mangat for his guidance throughout this
research and would also like to thank Dr Finbarr O’Flaherty for his assistance.
 Candidate’s declaration

I hereby declare that no portion of the work referred to in this thesis has been submitted in

support of an application for another degree or qualification of this or any other university

or other institution of learning. All sources of information have been duly acknowledged.

Candidate

Laurence Green - October 2005

Director of studies

Pritpal Mangat - October 2005

 CONTENTS

Contents................................................................................................................................ i

Index of figures.................................................................................................................... vii

Index of tables.................................................................................................................... xiii

1 Introduction....................................................................................................................... 1

1.1 General..........................................................................................................................1

1.2 Objectives.................................................................................................................... 2

1.3 Methodology................................................................................................................ 2

1.4 Layout.......................................................................................................................... 4

2 Reinforced Concrete Bridge repair: an overview......................................................... 6

2.1 Chapter Obj ectiv e s ...................................................................................................... 6

2.2 Introduction..................................................................................................................6

2.3 Bridge inspection......................................................................................................... 7

2.4 Deterioration of reinforced concrete in bridges......................................................... 9

2.4.1 Reinforcement Corrosion.................................................................................10

2.4.2 Sulphate attack................................................................................................... 13

2.4.3 Alkali aggregate reaction.................................................................................. 14

2.4.4 Freeze thaw attack.............................................................................................15

2.4.5 Other forms of deterioration............................................................................. 16

2.4.6 Non-structural cracking in concrete................................................................ 16

2.4.7 Other defects.......................................................................................................17

2.5 Establishing the causes of concrete deterioration....................................................18

2.5.1 Ingress of Chloride Ions....................................................................................19

2.5.2 Ingress of Carbon dioxide................................................................................. 20

 2.5.3 Alkali Aggregate reaction................................................................................21

2.5.4 Other causes of concrete deterioration........................................................... 22

2.6 Repairing defective concrete............................................................................ 22

2.6.1 Dealing with corrosion.................................................................................... 22

2.6.2 Repairing spalls................................................................................................ 23

2.6.3 Repairing AAR affected concrete....................................................................24

2.6.4 Unobtrusive alternatives to repair....................................................................24

2.6.5 Concrete Protection.......................................................................................... 26

2.6.6 Dealing with cracking.......................................................................................27

2.7 Expert Systems for concrete bridge repair............................................................ 31

2.7.1 Review of existing developments.................................................................... 31

2.7.2 Architecture of expert system for reinforced concrete repair........................ 37

2.7.3 Handling uncertainty........................................................................................ 42

2.8 General application of InformationTechnology in Bridges................................... 43

2.8.1 Bridge Management Systems...........................................................................43

3 Selection of materials for optimal performance of concrete repair........................45

3.1 Chapter Obj ecti v e s .................................................................................................... 45

3.2 Literature review........................................................................................................ 47

3.2.1 Introduction........................................................................................................47

3.2.2 Selecting repair materials for reinforced concrete........................................... 48

3.2.3 Properties of repair materials ..................................................................... 65

3.2.4 Influence of material constituents on mechanical properties......................... 73

3.2.5 Testing to establish repair material properties................................................74

3.3 Determining the key properties of repair materials................................................77

 t---

3.4 Development of a method to predict the performance of repair materials in-situ

................................................................................................................................... 84

3.4.1 Determination of properties..............................................................................85

3.4.2 Development of repair material properties......................................................89

4 The procedure for determining the in-situ performance of repair materials.... 124

4.1 Chapter obj ective......................................................................................................124

4.2 Introduction..............................................................................................................124

4.3 Procedure for determining the performance of a repair material......................... 125

4.3.1 Determining Tensile Strength from modulus of rupture.............................. 126

4.3.2 Modifications for climate................................................................................126

4.3.3 Establishing seasonal temperature and RH variations...................................128

4.4 Shrinkage of patch repair: correction factorsfor temperature and humidity .... 130

4.5 Creep of patch repairs: correction factorsforatmospheric conditions, specimen

size, and age at loading........................................................................................................... 132

4.5.1 Modifying creep for early age loading.......................................................... 132

4.5.2 Creep of patch repair: correction factors for temperature and relative

humidity ......................................................................................................................... 144

4.5.3 Modifying creep by relative humidity............................................................145

4.5.4 Modifying creep for specimen size................................................................146

4.6 Modifications for field shrinkage.................................................................. 151

4.7 Properties of the substrate...............................................................................152

4.8 Development of properties.............................................................................155

4.8.1 Consider day 14................................................................................................157

4.9 The effect of creep.................................................................................................... 159

4.9.1 Unit Creep......................................................................................................... 159

iii
 4.9.2 The effect of creep at day 2 :..........................................................................162

4.9.3 The effect of creep at day 4 :......................................................... 166

4.10 Transfer of strain to the substrate.......................................................................... 170

4.10.1 Consider day 14...............................................................................................170

4.11 Tensile strain capacity.............................................................................................175

4.12 Estimation of Creep using shrinkage data............................................................. 178

4.12.1 Introduction.................................................................................................... 178

4.12.2 O'Flaherty58 ..........................................................................................178

4.12.3 Mangat & Limbachiya56................................................................................. 179

4.12.4 Mangat & Azari106...........................................................................................180

4.12.5 Evans107..........................................................................................................181

4.12.6 Limbachiya108................................................................................................. 182

4.12.7 Poston, Kesner, McDonald, Vaysburd68,76 .................................................. 183

4.12.8 Emberson & Mays61....................................................................................... 186

4.12.9 Neville104.........................................................................................................186

4.13 Summary of guidelines for selection of reinforced concrete repair materials... 191

4.14 Conclusion............................................................................................................... 201

5 Decision making in the expert system for reinforced concrete bridge repair.... 202

5.1 Chapter Objectives..................................................................................................202

5.2 Introduction............................................................................................................. 202

5.2.1 Expert systems.................................................................................................203

5.2.2 Determining the severity of a defect.............................................................. 204

5.3 Data input.................................................................................................................204

5.4 Diagnosing concrete defects in an expert system.................................................207

5.4.1 Beginning the process......................................................................................207

iv
. g jjiw u iui iw a iu iv c u ^unv^icic u n u ge repair JL.r.ljreen

5.4.2 Constructing expert systems......................................................................... 208

5.4.3 Developing the knowledge bases..................................................................211

5.5 Determining the severity and extent of defects.....................................................228

5.5.1 Determining the size of a defect....................................................................230

5.5.2 Map cracking defects......................................................................................233

5.5.3 Spall defects.................................................................................................... 246

5.5.4 Structural cracks............................................................................................. 257

5.5.5 Miscellaneous defects.....................................................................................260

5.6 Uncertainty in deciding severity ................................................................261

5.6.1 Types of uncertainty....................................................................................... 266

5.7 Contribution of each defect toelement severity.....................................................267

5.7.1 Effect of pattern cracking on element severity............................................. 268

5.7.2 Effect of spalling defects on element severity.............................................. 281

5.7.3 Effect of structural cracking on element severity......................................... 285

5.7.4 Effect of miscellaneous defects on element condition.................................286

5.7.5 The Element Graph......................................................................................... 289

5.8 Testing..................................................................................................................... 292

5.9 Repair advice........................................................................................................... 297

5.9.1 Advice for spalls and pattern cracking.......................................................... 297

5.9.2 Advice for structural cracking........................................................................302

6 Review of the expert system for reinforced concrete bridge repair.................... 306

6.1 Introduction..............................................................................................................306

6.2 Structures Management...........................................................................................307

6.3 Element and Structure creation............................................................................... 310

6.3.1 Example of structure creation......................................................................... 312

v
6.4 Defects - input, assessmentand diagnosis.............................................................317

6.5 Elements - Testing and repair............................................................................... 328

6.6 Repair material selection........................................................................................330

6.7 Summary and Conclusion....................................................................................... 335

7 Conclusions.....................................................................................................................337

8 Further Work..................................................................................................................340

8.1 Field testing and calibration of the expert system.................................................340

8.2 Field testing to assess the performance of the concrete repair material property

selection system...................................................................................................................... 340

8.3 Prioritising the repair of bridges and bridge elements...........................................341

8.4 Expanding the expert system capability.................................................................341

9 References and Bibliography........................................................................................ 343

9.1 References................................................................................................................343

9.2 Bibliography.............................................................................................................3 52

9.3 Publications..............................................................................................................363
 INDEX OF FIGURES

1^
2.1 Typical optimum repair strategy ....................................................................................... 9

2.2 The dispersion of vehicle salt spray19................................................................................ 11

2.3 Typical 'manx' cracking, in the early stages of A A R .......................................................15

IQ

2.4 Repair methods for cracks .............................................................................................. 28

2.5 Diagram of typical crack types and explanatory table40...................................................30

2.6 Structure of HYWCON expert system42...........................................................................32

2.7 American HWYCON system step 1.................................................................................. 32

2.8 American HWYCON system step 2.................................................................................. 33

2.9 American HWYCON system step 3.................................................................................. 33

2.10 American HWYCON system step 4................................................................................ 34

2.11 American HWYCON system step 5................................................................................ 34

1£
2.12 Concrete repair expert system: basic architecture .......................................................37

2.13 Typical session with DIAGCON expert system47.......................................................... 39

2.14 Approach to concrete structure maintenance1................................................................ 40

2.15 Structure of REPCON1..................................................................................................... 41

2.16 An example of fuzzy sets49,50........................................................................................... 42

3.1 The effects of different periodson projectquality (durability)69...................................... 50

3.2 A model of repair failure70.................................................................................................. 52

3.3 Compatibility and durability74............................................................................................62

3.4 Factors affecting dimensional compatibility69.................................................................. 62

3.5 Factors affecting the durability of concrete repair systems69.......................................... 64

79
3.6 Schematic illustration of stress buildup inrepairs ........................................................ 77
• * t o
3.7 Relationship between modular ratio and free shrinkage transfer ................................. 80
 Index ol figures

3.8 Shrinkage: Substrate and repair material interaction, t = 0 (on application).................. 83

3.9 Shrinkage: Substrate and repair material interaction, t = 28 days. Erep < Esub................83

3.10 Shrinkage: Substrate and repair material ineraction, t = 28 days. Erep =1.1 ESUb 83

3.11 Creep ratio (C/C28 ) versus time under load relationship for thirteen repair materials 94

3.12 t/Cr versus t relationship for thirteen repair materials (R2 = 0.9964)......................... 97

3.13 Comparison of average experimental creep ratio (Cr) with calculated creep ratio (Cr)

(Stress/Strength ratio 30%).............................................................................................. 99

3.14 Ratio of shrinkage at each age to 28 day shrinkage (S/S28 ) versus time relationship.

..........................................................................................................................................103

3.15 t/Sr versus t relationship for the thirteen repair materials (R2 = 0.99)..................... 106

3.16 Sr versus time (after casting) relationship.....................................................................107

3.17 Development of strength of concrete with age82.......................................................... 109

3.18 Development of compressive strength ratio with time.................................................I l l

3.19 Comparison of measured Compressive strength with predicted Compressive strength

112

3.20 Development of tensile strength ratio with tim e.......................................................... 115

3.21 Comparison of predicted and experimental tensile strengths...................................... 116

3.22 Relationship between elastic modulus ratio (E/E28 ) and time for the three repair

materials........................................................................................................................... 119

3.23 Development of elastic modulus ratio (Er) with time based on the average of 3 repair

materials........................................................................................................................... 121

3.24 Graph comparing the predicted and experimental curves of elastic modulus ratio

versus time relationship.................................................................................................. 122

3.25 Comparison of experimental and predicted values of elastic modulus for vinyl acetate

material............................................................................................................................ 123

viii
 maex or ngures

4.1 Effect of relative humidity on creep102..........................................................................127

82
4.2 Influence of initial temperature on average monthly compressive strength in the UK

128

4.3 Map for climate tables 4.2 and 4 .3 .................................................................................. 129

4.4 Specific creep-time relationship for concrete loaded at 1 and 7 day ages at a

stress/strength ratio of 30% (average of Table 4.7 materials)......................................139

4.5 Increase in specific creep due to loading at early age (stress/strength ratio 30% ).......141

4.6 Estimated specific creep of material loaded at 1 day, 7 days and 28 days a g e ............143

4.7 Modifications for effect of specimen size on creep in concrete and repair materials. 148

4.8 Correction factor for height / diameter ratio of concrete cores.................................... 153

4.9 Life of Creep sample from casting.................................................................................160

4.10 Determining equivalent constant stress........................................................................ 168

4.11 Performance of material Shucrete 1 ............................................................................. 176

4.12 Relationship between shrinkage and creep...................................................................188

5.1 New structure inserted..................................................................................................... 205

5.2 Inserting a deck element..................................................................................................206

5.3 Adding bank-seats........................................................................................................... 207

5.4 Knowledge net................................................................................................................. 209

5.5 Instance net.......................................................................................................................210

5.6 Structure for defect study................................................................................................215

5.7 Central pier selected for study........................................................................................ 216

5.8 Adding rectangular defect.............................................................................................. 216

5.9 Adding a spall defect.......................................................................................................217

5.10 Adding spall details.......................................................................................................218

5.11 Inputting corrosion am ount.......................................................................................... 219

ix
 index o f figures

5.12 Carbonation premise rule............................................................................................... 220

5.13 Rules in the spall knowledge base.................................................................................221

5.14 Context for carbonation action table ru le.................................................................... 222

5.15 Action table for 'cause carbonation'...............................................................................224

5.16 Selecting a representative map-cracking im age...........................................................225

5.17 Adding a structural crack............................................................................................... 227

5.18 Repair zones for 32mm deep spall................................................................................231

5.19 Spall depth 50mm...........................................................................................................232

5.20 Large map-crack. User selects 'chloride corrosion' im age.......................................... 235

5.21 Images 1 & 2................................................................................................................... 235

5.22 Images 3 & 4................................................................................................................... 236

5.23 Images 5 & 6................................................................................................................... 236

5.24 Images 7 & 8................................................................................................................... 237

5.25 Image 9 ............................................................................................................................237

5.26 40mm deep spall............................................................................................................. 241

5.27 40mm deep spall with exposed, corroded reinforcement............................................ 242

5.28 Secondary zone movement graph.................................................................................. 243

5.29 Experts' spall size/depth zone positions........................................................................247

5.30 Secondary movement of zone for spalls........................................................................253

5.31 Typical pier element.......................................................................................................258

5.32 Unwrapped p ier.............................................................................................................. 258

5.33 Element canvas for crack size determination............................................................... 258

5.34 Projecting crack onto element edges.............................................................................259

5.35 Handling uncertainty...................................................................................................... 263

5.36 Zone decision graph........................................................................................................ 265

 Index o f figures

5.37 Determining effect of defects on element..................................................................... 269

5.38 Determining the effect of chloride cracking on element condition............................ 272

5.39 Determining the effect of AAR cracking on element condition..................................275

5.40 Determining the effect of frost cracking on element condition...................................276

5.41 Determining the effect of plastic shrinkage cracking on element condition..............277

5.42 Determining the effect of crazing on element condition............................................. 278

5.43 Determining the effect of drying shrinkage cracking on element condition..............279

5.44 Determining the effect of carbonation cracking on element condition...................... 280

5.45 Determining the effect of AAR spalling on element condition...................................282

5.46 Determining the effect of chloride spalling on element condition............................. 283

5.47 Determining the effect of carbonation spalling on element condition....................... 284

5.48 Determining the effect of blow-holes and sand-streaking on element condition 287

5.49 Determining the effect of honeycombing on element condition.................................288

5.50 Element condition indicator........................................................................................... 289

5.51 Drying shrinkage repair rule.......................................................................................... 299

6.1 Structure management...................................................................................................... 308

6.2 Alternative views of structures........................................................................................ 309

6.3 'Structure creation wizard' single span bridge................................................................ 310

6.4 'Structure creation wizard' three span bridge in 3D.......................................................311

6.5 Unusual motorway bridge................................................................................................312

6.6 Blank diagram window..................................................................................................... 313

6.7 Inserting a beam.................................................................................................................314

6.8 Beam once inserted........................................................................................................... 314

6.9 Three beams inserted........................................................................................................ 315

6.10 Insertion of p ie r...............................................................................................................316

xi
 index ot tigures

6.11 Copying an existing pier.................................................................................................316

6.12 Entering an elliptical defect........................................................................................... 318

6.13 Classifying the defect.....................................................................................................319

6.14 Entering corrosion information..................................................................................... 319

6.15 Entering more spall information.................................................................................... 320

6.16 Judging spall severity.....................................................................................................321

6.17 Entering a map-cracking defect..................................................................................... 322

6.18 Choosing a representative image................................................................................... 323

6.19 Viewing expert system advice....................................................................................... 324

6.20 Detailed knowledge base output.................................................................................... 325

6.21 3D view of affected column.......................................................................................... 326

6.22 Entering a crack defect...................................................................................................327

6.23 Element testing advice from the knowledge b ase........................................................328

6.24 Element repair advice from the knowledge base..........................................................329

6.25 Manufacturers' test data for repair materials................................................................ 330

6.26 Substrate information.....................................................................................................331

6.27 Performance of Proton Microconcrete..........................................................................332

6.28 Performance of Flexcrete m aterial................................................................................ 333

6.29 Failed material............................... :............................................................................... 334

xii
 Index o f tables

INDEX OF TABLES

3.1 Requirements of patch repair material for structural compatibility61..............................54

3.2 Recommended repair material property limits and values for effective application 55

3.3 Hong Kong Housing Authority repair material specification77....................................... 58

3.4 Categories of systems for concrete patch repair74............................................................60

3.5 Typical mechanical properties of repair materials74.........................................................60

3.6 Recommended values for compatibility64........................................................................ 63

3.7 Strength and Modulus test methods61................................................................................75

3.8 Test methods to establish drying shrinkage64................................................................... 76

3.9 Strains developed in repair material and substrate58....................................»...................79

3.10 Percentage of free shrinkage transferred into substrate concrete58................................79

3.11 Key properties for the optimisation of repair material selection....................................85

3.12 The 28 day strength, elastic modulus and tensile strength of the thirteen generic repair

materials............................................................................................................................. 90

3.13 Development of Creep (microstrain) during period under load, at 30% stress/strength.

............................................................................................................................................92

3.14 Creep (C/C28 ) in material G1 as a ratio of the 28 day creep (C28 ).................................93

3.15 Correlation coefficient of average creep ratio curve with creep of each material....... 95

3.16 Best fit relationship data of C/C28 with time under load.................................................96

3.17 Development of shrinkage (microstrain) with time for thirteen repair materials.......101

3.18 Shrinkage in material G1 as a ratio of the 28 day shrinkage........................................102

3.19 Correlation coefficients for the shrinkage ratio versus time curves of each material

with the average relationship......................................................................................... 104

3.20 Best fit relationship data of S/S28 with time after casting..........................................105
 index 01 taoies

3.21 Development of Compressive strength with age.......................................................... 108

3.22 Development of Tensile strength ratio (ft / fos) with time........................................... 114

3.23 Development of elastic modulus with time in repair materials................................... 118

3.24 Elastic modulus of cementituous repair material as a ratio of the 28 day elastic

modulus........................................................................................................................... 118

3.25 Average (E/E28 ) ratio versus time relationship for the repair materials..................... 120

4.1 The properties of Shucrete 1 at 28 days age.................................................................... 125

1
4.2 Seasonal average temperature variation in the UK .................................................129

4.3 Approximate seasonal average relative humidity in the UK103................................... 129

a
4.4 Specific Creep of concrete loaded at different days (pm/mm/N/mm ) ......................133
a
4.5 Development of Tensile Strength (N/mm ) during creep testing.................................. 135

4.6 Stress/Strength ratios........................................................................................................ 136

4.7 Specific creep of specimens extrapolated at 30% stress/strength ratio.........................137

4.8 Ratio of specific creep due to loading at 1 day to loading at 7 days, at a stress strength

ratio of 30%......................................................................................................................140

4.9 Creep modification factors for early age loading (concrete)104..................................... 143

4.10 Development of properties with time (days) of repair material Shucrete 1 and transfer

of shrinkage strain to the substrate................................................................................ 156

4.11 Performance of material Shucrete 1 .............................................................................. 173

4.12 Development of Tensile Strain Capacity in Shucrete 1 (microstrain).........................175

• co
4.13 Creep and free shrinkage data for repair materials .................................................... 179

4.14 Creep and free shrinkage data for repair materials56.................................................... 180

4.15 Creep and free shrinkage data for repair materials106................................................... 180
• I C\1
4.16 Creep and free shrinkage data for repair materials ................................................... 181

4.17 Creep and free shrinkage data for repair materials108................................................... 183

xiv
 Index o f tables

4.18 Creep and free shrinkage data for repair materials .................................................. 184

4.19 Creep and shrinkage data for repair materials61........................................................... 186

4.20 Key to Figure 4.12.......................................................................................................... 189

5.1 Position of zone apexes for chloride cracking image.................................................... 234

5.2 Position of zone apexes for A A R ....................................................................................238

5.3 Position of zone apexes for Freeze Thaw dam age.........................................................238

5.4 Position of zone apexs for Plastic Shrinkage, Crazing, and Drying Shrinkage.......... 238

5.5 Position of apexes for carbonation induced cracking.................................................... 239

5.6 Secondary zone movement constants........................................................................... 244

5.7 Adjustment factors for secondary zone movement........................................................245

5.8 Position where decision line and curve meet.................................................................. 248

5.9 Equation of line to determine zone apex positions for spalls........................................ 249

5.10 Constants (K and S) for equations of curve for determining zone apex positions for

spalls................................................................................................................................250

5.11 Secondary zone movement constants............................................................................254

5.12 Adjustment factors for secondary zone movement......................................................255

5.13 Position of zone apexes for chloride ingress and carbonation spalls (unknown depth)

......................................................................................................................................... 257

5.14 Zone apex positions for blow-holes or sand-streaking................................................ 261

5.15 Zone apex postions for honeycombing..........................................................................261

5.16 Change in uncertainty as data is added..........................................................................264

5.17 Example defects on 'Column 3 '..................................................................................... 293

5.18 Contribution of each defect to element condition........................................................ 294

5.19 Determining primary causes of element deterioration.................................................295

5.20 Typical storing of knowledge base diagnosis for pattern cracking.............................296

xv
 Index o f tables

5.21 Typical storing of knowledge base diagnosis for spalling.......................................... 296

5.22 Repair advice.................................................................................................................. 299

xvi
 Chapter 1 - Introduction

1 Introduction

Chapter objectives

o To introduce the research

o To discuss the aims, objectives and methodology of the research

o To describe the layout of the thesis

1.1 General

An expert system for concrete repair is an intelligent software adviser that can assist an

engineer across the range of activities involved in the concrete repair process, specifically:

o Inspection

o Diagnosis

o Testing

o Repair methods

o Repair materials

o Prioritisation

Expert systems, also known as knowledge based systems, are software programs which

employ logical reasoning instead of quantitative calculations in their processes.

I
 Chapter 1 - Introduction

1.2 Objectives

This research develops a software expert to aid reinforced concrete evaluation and repair.

Existing expert system and knowledge based system development tools will be used to

develop a program to diagnose concrete defects. A method for assessing the severity and

extent of reinforced concrete defects will be developed to work seamlessly with the defect

diagnosis component.

Crucially, a routine will be developed, based on state of the art research, to recommend the

properties of materials for reinforced concrete repair.

To the best of the author’s knowledge, this is the first attempt to create an expert system

for concrete repair with the intelligence to judge (and consider in its recommendations) the

severity and extent of defects.

The aim of this project is to develop a software system to provide decision support for civil

engineers involved in reinforced concrete maintenance. The system will upgrade the

performance of engineers and enhance and verify their decision making.

1.3 Methodology

Rules and guidelines for reinforced concrete repair are poorly structured. The collation of

these rules into a knowledge base requires the input of well qualified experts. This has

been achieved in the thesis and as such, knowledge in this domain is well suited for
 unapter i - introduction

exploitation in an expert system. Conversely, the calculation of suitable properties required

for satisfactory performance of repair materials for reinforced concrete repair is purely

mathematical and can be handled with standard algorithmic programming. Problems which

can be solved heuristically only, are suitable for use in expert systems1, therefore, the

approach of this project will be to address the selected problems with the appropriate

software technology.

The system will be formed by three key modules each working together seamlessly behind

a front-end bridge management system developed by a collaborator. A heuristic expert

system module will diagnose concrete defects and recommend tests and solutions. An

algorithmic system will asses the severity and extent of defects. A second algorithm based

system will recommend repair material properties.

In order to satisfy the objectives of this research the following key tasks were performed:

o Knowledge acquisition sessions with professional experts

o Heuristic system development

o Severity and extent algorithm software development

o Repair material property recommendation algorithm software

development

o Interaction with bridge management system development.

3
 unapter 1 - introduction

1.4 Layout

Chapter 1 General introduction to thesis and discussion of research objectives and

methodology.

Chapter 2 A review of the key domains relevant to this thesis, specifically: bridge

management, concrete defects, defect identification, and concrete repair.

This chapter also includes a brief review of existing prototype expert

systems in this domain.

Chapter 3 This chapter contains an extensive literature review on current thinking

regarding patch repair of reinforced concrete. It establishes known

misconceptions amongst practitioners and highlights new research thereby

establishing a sound theoretical basis on which to develop a system to

recommend optimum repair material properties.

Chapter 4 Presents the development of an analytical procedure to recommend

optimum properties of reinforced concrete repair materials. The technique

developed is exploited in a computer program integrated seamlessly into the

expert system for concrete repair.

Chapter 5 Develops the diagnostic expert system for reinforced concrete repair and the

tools to assess the severity and extent of concrete defects. The chapter

4
 Chapter 1 - Introduction

establishes the knowledge acquisition processes and how the knowledge

obtained is interpreted and coded into the software.

Chapter 6 Reviews the completed expert system, which is embedded into a bridge

management system in order to fully computerize concrete structure

maintenance, from inspection to repair. The chapter discusses how the

software modules developed in the thesis are practically applied in the

expert system.

Chapter 7 Reviews and assesses the research, discussing the conclusions drawn.

Chapter 8 Discusses future research efforts and ways the system can be updated and

expanded.

5
 ^napier z - Keiniorcea concrete bridge repair: an overview

2 Reinforced Concrete Bridge repair: an overview.

2.1 Chapter Objectives

• Discuss reinforced concrete repair

• Discuss application of expert system technology in reinforced concrete repair

domain

• Identify existing expert systems in field of concrete repair

2.2 Introduction

In the UK, there are over 50,000 bridges constructed from reinforced or pre-stressed
2 #
concrete . Those on motorways and trunk roads in England fall under the jurisdiction of

the Highways Agency and similar agencies are responsible for structures on such roads in

the rest of the UK. Bridges on local roads are the responsibility of local authorities.

Concrete was once considered to be a durable material requiring little or no maintenance4’5,

however, it has been recognised for some years that concrete is susceptible to degradation

caused by aggressive chemical attack and adverse reactions to the natural environment6. As

a result of deterioration, bridges can become aesthetically unacceptable, deterioration can

lead to faster rates of further degradation, the service life of structures can be reduced and

in severe cases the structural capacity of structures can be affected.

In his study of 200 concrete bridges in England, Wallbank4 classed 114 as being in fair

condition and 61 as having serious defects. More recently, almost one third of United

States bridges were reported as substandard7.

6
 lim p id z. - i\eniiurcea concrete oriage repair: an overview

Although modem stmctures can be constmcted with inbuilt protection against commonly

known causes of deterioration, the vast majority of reinforced concrete bridges are over 20

years old. The replacement value of all UK concrete bridges is many billions of pounds

and as such, the only option for the preservation of bridge infrastmcture is suitable

maintenance and repair schemes.

2.3 Bridge inspection

The purpose of a bridge inspection is to allow an inspector to observe and record defects

present on a structure8. Diagnosis therefore, is not strictly a part of the inspection process,

although the two are closely related. Current practise for the inspection of bridges in the

UK is set out in volume three of the Design Manual for Roads and Bridges9. Inspections

currently fall into four categories.

• Superficial Inspection : Cursory checks of stmctures whenever those with

responsibilities towards a bridge happen to be passing it. Any major visible

problems are reported to the bridge engineer.

• General Inspection : A visual examination of all parts of the bridge in order to

ascertain the condition of all elements. These inspections are undertaken every two

years and observations are made from the ground using binoculars where

necessary.

• Principal Inspection: This inspection involves a closer examination of all parts of a

bridge. It is usually a surface inspection which can involve the use of access

equipment and traffic management. A principal inspection is completed every 6

years. A full report is produced from the inspection.

7
 iw u u u itv /u b u u w b it uilu tjc ic (ja u . tin u v c iv ic w

• Special Inspection: A close inspection of a particular area or defect. Usually as a

follow up on a defect identified in a previous inspection10.

Recently, the Highways Agency has decided to review the inspection procedures, and

implement new inspection routines to take account of the condition of bridges and bridge

elements in order to decide the type and frequency of inspections. This was to involve a

‘Benchmark Inspection’ replacing the current ‘Principal Inspection’. The Benchmark

Inspection would take place at intervals between 6 and 24 years depending of the reported

condition of the bridge from the last benchmark inspection11. To date, the Highways

Agency have not replaced the current inspection procedures with these proposed ones and

there is no information indicating if the change will now occur.

Concrete bridges are required to maintain their serviceability over long periods of time12.

Typical concrete, cast as part of a highway bridge, is unlikely to resist deterioration over its

design life (usually 120 years), and is likely to require repairing in order to maintain the

serviceable state of the bridge.

Figure 2.1 shows how a typical bridge becomes less reliable with age, and how regular

maintenance combats the fall in reliability. To counteract the loss of reliability with age,

concrete in bridges needs to be maintained and treated for disease whenever it exhibits

typical signs of distress (and occasionally when there are no visible signs of distress). If

deterioration is not regularly arrested, the cost of restoring reliability becomes much

greater than a regularly maintained structure. Assessing the effect of short term

expenditure through regular maintenance on the long term financial costs of a bridge (or

any structure) is known as whole life costing.

 ^im pier z - iveim orcea concrete bridge repair: an o verview

3 .0

1.0

JJ NO JNSFECTION/REPAm
1 ) OPTIMUM STRATEGY

TIM E (YEARS)

Figure 2.1 Typical optimum repair strategy13

2A Deterioration of reinforced concrete in bridges

The most common form of deterioration affecting concrete in bridges is reinforcement

corrosion14,15. Although concrete can contain moisture long after curing, and the micro

voids within the concrete matrix can also contain oxygen, reinforcement in concrete does

not usually rust. This is due to the inert barrier formed around it by free alkalis (usually

calcium hydroxide)5,16. However, if this barrier is broken, and if sufficient oxygen and

water are present, the steel reinforcement will provide anodic and cathodic sites, and the

moist concrete provides the electrolyte necessary to initiate the electrochemical reaction

9
 ^ nupiei z - ivem iorcea concrete bridge repair: an ov erv iew

whose end product is ferric oxide (rust). Corrosion causes further problems for the

symbiotic concrete/steel element17:

• Reduction in steel area

• Corrosion products occupy a larger volume than original steel. This exerts

expansive pressure on the concrete causing cracking, spalling and delamination.

• The bond between steel and concrete deteriorates and the composite action of steel

and concrete is lost.

Any aggressive agent diffusing to the steel reinforcement is aided in its journey by

insufficient cover or areas of poor compaction.

Other forms of deterioration can also blight the performance of concrete in bridges.

There are a number of ways in which concrete deterioration can be categorised. One

effective way is to categorise three types of defects: early age, medium to long term,

sudden defects18. Early age defects are attributed almost solely to moisture movement

during curing, but can also be the fault of poor mix design or faulty workmanship. Medium

to long term defects are often caused by environmental aggression and long term concrete

‘disease’. Sudden defects occur through fire, physical impact, seismic event, overload of

the structure or settlement of foundations etc.

2.4.1 Reinforcement Corrosion

Invariably, the cause of corrosion is an aggressor which breaks down the passive layer

formed around reinforcing steel, allowing the electrochemical process to take place. The

known aggressors that cause corrosion in reinforced concrete are chlorides and carbon

dioxide which diffuses from the air to neutralise the alkalinity of concrete.

10
 v^najjici z. - iveim orceu concrete o n ag e repair: an overview

The common cause o f chlorides in the non-marine environm ent is from de-icing salts

applied to roads and bridge decks during periods where there is a risk o f surface-water

freezing19. The transit o f salt laden water from pavem ent to concrete structure is shown in

Figure 2.2.

Pier

^ l L -l i l-^S-
Splashed water
/ Seeping I
water * Surface water
Wind

Figure 2.2 The dispersion o f vehicle salt sp ray19

These salts make their way onto bridge elements either via leaky drains and joints on the

bridge or in splash water sprayed at the bridge from the wheels o f passing vehicles. W ith

the help o f surface moisture, chloride salts in solution can permeate into concrete20. This

action is accelerated in concrete that is already damaged through some other mechanism ,

such as the effects o f freezing and thawing cycles (see section 2.4.4). Chloride ions cause

depassivation o f the reinforcing steel even in alkaline environments.

The speed at which chloride penetration approaches the reinforcing steel is dependent

upon21:

• The amount o f chlorides coming into contact with the concrete

• The permeability o f the concrete

• The amount o f moisture present

11
 li m p i d z - iveiniorcea concrete bridge repair: an overview

When the concentration of chloride ions exceeds 1% of the mass of cement in concrete, the

corrosion of reinforcement is inevitable5. Once corrosion begins, the expansive corrosion

products cause tensile stresses in concrete which lead to cracking and delaminations.

Carbonation in concrete is a reaction between natural or industrially produced carbon

dioxide in the air and calcium hydroxide dissolved in the pore water contained in the

concrete microstructure22. From the time concrete is cast, its surface zone is subjected to

attack from carbon dioxide continuously5. This gradually degrades the alkalinity of the

concrete which passivates reinforcement against corrosion. Therefore, carbonation is only

likely to be a problem in older bridges, areas of concrete with low cover to the

reinforcement, or poor quality porous concrete8. When the carbonation front reaches the

steel reinforcement, its passivation is dissipated and, in the presence of moisture and

oxygen, corrosion can begin.

Wallbank’s survey of 200 bridges in the UK found that 90% of bridges had a carbonation

depth of 5mm or less4. As such, in practice, carbonation is not as common a problem in the

UK environment as chloride ingress for the corrosion of reinforced concrete. However,

carbonation is sensitive to temperature and relative humidity of the environment and is

accelerated in warm, dry climates.

The rate of penetration of carbonation through concrete can be approximately represented

by23:

x = (2 D t f 5 Eq. 2-1

Where x = depth of penetration

D = diffusion coefficient of CO2 in concrete

t = time in years

12
 sim p ler z - jveinrorcea concrete bridge repair: an overv iew

The depth of a carbonation front into concrete can be measured by breaking out a small

section of concrete and spraying the exposed sub-surface concrete with a phenolphthalein

spray which reacts with carbonated concrete. If the age of the structure is known, and the

depth of carbonation is determined, then the diffusion coefficient can be calculated. This

coefficient can then be used to determine the age of the structure when the carbonation

front reaches the reinforcing steel.

The rate of chloride penetration into concrete as a function of depth can be represented by

Fick’s Law of diffusion233 shown in equation 2-2:

r
x
Eq. 2-2

Where C(X,t) is the chloride ion concentration at a distance x (cm) from the concrete surface

after time t (s)

Co is the equilibrium chloride concentration on the concrete surface

Dc is the chloride diffusion coefficient in cm2/s

erf is the error function

Determining the chloride ion concentration at the steel reinforcement can give an

indication of the likelihood that corrosion of the reinforcement is occurring.

2.4.2 Sulphate attack

Atmospheric sulphur dioxide can affect concrete in a similar way as carbon dioxide19. It

can also act in conjunction with carbon dioxide to increase the rate of loss of alkalinity in

concrete. Alternatively, sulphates contained in ground water can attack concrete in a

similar way to chlorides 24

In addition to the depassivation which airborne or soluble sulphates can cause to

reinforcing steel, sulphates can react chemically with hydrated lime in the cement paste
 ^ —ivcnuuiucu ^uuureic oriuge repair: an overview

producing solid products (gypsum and ettringite) with greater volume than the products

entering the reaction. As a result surface scaling can occur followed by mass disruption of

the concrete. Hence sulphates can cause surface defects in concrete that not only render the

concrete aesthetically unacceptable, but also aggravate the effects of carbon dioxide,

chlorides, further sulphate attack and freeze-thaw cycles22. Sulphate attack is uncommon in

the UK.

Sulphates can cause deterioration of the concrete matrix itself, but concrete deterioration is

more commonly caused by other aggressors which lead to corrosion of the reinforcing

steel.

Sulphate attack can occur in a severe form known as Thaumasite. Thaumasite has the

effect of very seriously degrading the concrete matrix. To combat the risk of this aggressor,

careful selection of concrete mix constituents is necessary, particularly for substructures in

ground contaminated by sulphates.

2.4.3 Alkali aggregate reaction

Alkali-silica reaction, alkali-carbonate reaction and more generally alkali-aggregate

reaction19 are rare combinations of reactive aggregate, high alkali cement and moisture

which can cause adverse chemical reactions in the concrete matrix which produce an

expansive gel in structures5. The gel, when exposed to moisture, expands generating tensile

forces which cause cracking in a distinctive 4mamC pattern21 as shown in Figure 2.3.

14
 i i u i w u uvnwvi i , ui luge i c p a n . an u v c i v i c w

Figure 2.3 Typical 'manx' cracking, in the early stages o f A AR

AAR is often referred to as ‘concrete cancer’ due to its incurability, although corrective

measures can be taken to arrest its development25. Despite the alarming m anifestation o f

the defect, there is evidence that the effects o f the “disease” are less serious than

26 25
appearances suggest ' . This is due to the fact that cracks permeate only to a limited

depth.

2.4.4 Freeze thaw attack

The effects o f cyclical freezing and thawing o f concrete are alternatively described as

‘frost attack’ or more generally ‘w eathering’ (although this term also includes dam age

from wetting and drying, and heating and cooling cycles). Frost attack is a com m on cause

o f surface scaling and spalling in concrete19. W ater is absorbed into concrete through

capillary action, as this water freezes its volume increases by approxim ately 9% 27,19. The

expanding water causes hydraulic pressures in the pores o f concrete and a num ber o f

cycles can be sufficient to cause surface concrete to scale away from its parent mass.

The effect o f freeze-thaw cycles on a concrete surface can exacerbate chloride ingress and

carbonation by allowing easier access for chloride ions in solution and airborne carbon

dioxide to the reinforcement.

15
 .t, 1WUUU1VV/U vunu^lb U l l U g C i c p a u . till U V C 1V 1C W

2.4.5 Other forms of deterioration

Often vehicles will collide with reinforced concrete on highways. Typically the wing

mirror of a large vehicle may clip concrete at high speed, causing a piece of the concrete to

break away, or an accident may cause more serious damage. If reinforcement becomes

exposed, the result can be depassivation of the steel leading to corrosion and worsening

defects.

Aggressive industrial substances such as Aluminium Chloride or Calcium Bisulphate can

sometimes come into contact with concrete, often through leakages but also via accidents.

These can attack concrete surfaces causing rapid disintegration.

2.4.6 Non-structural cracking in concrete

Non-structural cracking in concrete is often a precursor to delamination and spalling

caused by corrosion. Many of the deterioration processes already described in section 2.4.1

initially manifest themselves as cracking over the areas that they affect. Alkali-Aggregate

reaction has a distinctive crack pattern, and corrosion from any source will initially cause

cracking as the tensile forces created by the corrosion products exceed local tensile

strength of concrete. Some crack types affect newly built structures (e.g. shrinkage

cracking), others are the results of defects which emerge in the longer term.

2.4.6.1 Crazing

Crazing is the cracking of the surface layer of concrete into small irregularly shaped

contiguous areas . Crazing is not structurally significant, and apart from accelerating other

concrete defects such as carbonation, it is only a cosmetic defect. Crazing is caused when

16
 ^im pici ^ - x\canurcea concrete oriage repair: an overview

the surface concrete upon curing is different to the underlying concrete (e.g. it is subjected

to excessive moisture). This can occur through over-trowelling or a number of other

effects. Crazing usually occurs shortly after casting but may occur at later ages if the

climatic conditions are severe enough.

2A.6.2 Plastic cracking

This type of cracking mainly occurs on exposed horizontal surfaces of concrete. It usually

occurs through differential shrinkage of surface and underlying concrete29. Plastic

settlement cracking occurs when the usual continued consolidation of concrete after

vibration is restrained by reinforcing bars.

2.4.6.3 Drying shrinkage

Drying shrinkage is the reduction in the volume of concrete caused by the chemical and

physical loss of water during the hardening process28. In newly cast concrete, this

shrinkage is restrained by the sub-base, by reinforcement, or by the concrete element

which the fresh concrete has been cast against. This restraint to shrinkage causes tensile

stresses to develop, which, if they exceed the tensile strength of the concrete can cause

cracking. Drying shrinkage occurs during the hardening phase of a concrete and, therefore,

can be expected to occur several weeks or months after casting19.

2.4.7 Other defects

Concrete can exhibit a number of defects after casting. These are invariably only of
30
cosmetic importance .

17
 ^iiajjiva ^ —iv cim u itcu uuncreie oriage repair: an overview

Honeycombing

Honeycomb surfaces are caused by the use of dry mix concrete that is not properly

consolidated. The lack of consolidation means that mortar does not effectively fill the

voids in between the aggregate particles.

Sandstreaking

Sandstreaking is a cosmetic defect caused by the use of wet concrete mixes which bleed

excessively.

Blowholes

Blowholes are small air pockets formed during placement and consolidation. They are

thought to be caused by excessive amounts of oil placed on the formwork. Counter­

intuitively, the more air-entrainment in the concrete, the less likely blow holes are to occur.

2.5 Establishing the causes of concrete deterioration

Once the suspected causes of concrete defects have been established, a course of testing

procedures is decided to confirm the original diagnosis and to provide details of the extent

of the problem. The majority of defects that signify concrete has become less reliable are

exhibited as either spalling or cracking. Other defects, such as patches of honeycombing,

require no testing to establish their causes because the cause is self evident. An engineer

can usually make an intelligent estimation of the cause of any spalling or cracking.

However, if such defects are of sufficient magnitude to warrant concern, a testing regime

will be required to confirm the engineer’s original diagnosis and establish the extent of

deterioration.

18
2.5.1 Ingress of Chloride Ions

Testing to confirm the presence of chlorides is undertaken if a defect shows signs that its

cause could be the ingress of chloride salts, and if the defect is of sufficient magnitude.

Concrete dust is extracted at regular intervals on the affected element by drilling into the

substrate and collecting the resulting dust in a small plastic receptacle. These samples,

taken at various spacings and depths, can be analysed chemically in the laboratory by

analytical means5. Often, an amount of chloride has been cast intentionally into the

concrete and this must be allowed for when determining the amount of chlorides which

have entered from the surface. This is done by taking dust samples from an area of the

suspect concrete where chloride ingress is not thought to have taken place21. The presence

of chlorides in concrete, even at depths equal to or greater than the reinforcing steel cover,

does not prove that electrochemical corrosion is taking place. Corrosion will only occur

with both moisture and oxygen present. Therefore, whenever there is suspicion that

corrosion is taking place in reinforced concrete, an electrochemical survey is conducted

(e.g. by half-cell potential survey) for the presence of corrosion activity. When the

passivity of steel is destroyed by carbonation, chloride ingress or any other agent,

electrochemical cells develop. When this happens, an electro-potential difference exists

between the anodic and cathodic areas of the steel . This difference is measured using an

electrode probe passing over the concrete surface, the probe is attached to the

reinforcement (a small removal of concrete on an affected member is necessary) and the

readings are taken from a high impedance voltmeter. During the test, the concrete must be

of uniform moisture content5. The results are plotted as a grid over a drawing of the

affected element and contours are mapped. The probability of corrosion taking place (when

measured using a standard copper/copper sulphate half-cell) is high if the potential ranges
 ^napici z, - ivem iorcea concrete bridge repair: an ov erv iew

between -0.2 and -0.4 volts. The concrete acts as the electrolyte for the electrochemical

reaction, therefore, its resistivity can indicate how effectively it will perform as an

electrolyte and support the formation of corrosion cells. Some repair practitioners also

recommend the use of a resistance meter to detect the passage of current between metal

electrodes cast in the concrete in order to determine its resistivity.

2.5.2 Ingress of Carbon dioxide

If the nature of a defect suggests its cause could be the approach of a carbonation front

caused by carbon dioxide diffusion, then a test to confirm the presence of a neutral-

alkalinity front would be conducted. This simple test requires a fresh area of concrete to be

broken out on site; this area is sprayed with alcoholic vinyl phenolphthalein. If the concrete

has retained its alkalinity, the spray turns pink. If the alkalinity has been neutralised by

CO2 action, the spray remains colourless31. The interface of the pink and colourless film of

spray represents the depth of carbonation at the test location. The seriousness of

carbonation is generally determined by the depth of carbonated concrete relative to the

depth of cover to the reinforcement. If the reinforcing steel is sufficiently far from the

approaching carbonation front then corrosion will not occur in the short term. Remedial

measures can be taken if the approach of carbonation to the steel is deemed as a long term

threat, see section 2.6.4. Therefore, the depth of cover to the reinforcement in an element

affected by carbonation needs to be determined. This is established using a Covermeter

test. Covermeter tests detect the distance between the surface of the concrete and the

reinforcing steel by generating a magnetic field and measuring the effect of reinforcing

steel below the surface on the field. The device used is known as a Covermeter or a

20
 z. - iv cim u itcu cuncreie D n a g e repair: an overview

2 2 93 • •
Pachometer ’ . The device is affected by reinforcement congestion, but generally

produces accurate results. The sensitivity of the test improves if the meter is calibrated

using a known diameter of the reinforcing bar.

2.5.3 Alkali Aggregate reaction

Surveys have shown that bridges with the constituent materials necessary for an alkali

aggregate reaction (AAR) take at least 10 years to exhibit the symptoms of the disease32

and as many as 20 to 30 years for the reaction to fully develop. The technique used to

detect AAR is petrographic analysis33 (or Petrography), which involves the examination of

polished plates of the material. The polished concrete samples taken from suspected AAR

sites are examined for networks of micro-cracking through the concrete matrix, and the

presence of gel; the tell-tale signs of AAR34.

The effect of alkali-aggregate reaction on the likelihood of corrosion to occur is difficult to

determine . A high pH is needed for AAR to occur but a high pH environment protects

reinforcing steel from corrosion. Concrete cracking caused by AAR should accelerate the

carbonation process by allowing faster access of carbon dioxide towards the reinforcing

steel but the high moisture levels associated with AAR also slow the carbonation process.

Additionally, the cracks caused by AAR often become filled with a gel, preventing the

access of carbon dioxide and chloride solution into the concrete.

21
 li m p i d z - iveiniorcea concrete bridge repair: an overview

2.5.4 Other causes of concrete deterioration

Some defects may be of such a nature as to leave the engineer or expert unsure of their

cause. In these cases multiple tests are conducted (if the severity and extent of the defect

warrant such action) in order to establish the cause. For example, an engineer who is

unsure if a concrete spall containing exposed reinforcement was caused by chloride or

carbonation induced corrosion might recommend testing to ascertain the presence of both

aggressors. Some defects, such as freeze-thaw cracking, are the effects of undetectable

aggression from the external environment. They can easily be confused with chloride

ingress defects or vice-versa. It may be necessary to test for other defect causes in order to

be able to diagnose the cause of a defect by a process of elimination.

2.6 Repairing defective concrete

2.6.1 Dealing with corrosion

Generally, for any defect in need of repair, all cracked, spalled and delaminated concrete is

cut away to a depth just exceeding the steel reinforcement. If there is reason to suspect that

corrosion is taking place in an area that displays no visible sign of such (for example, the

results of the half-cell potential survey) then it may be necessary to break out concrete in

those additional areas. Certainly, the removal of concrete should continue along the

reinforcement until the signs of corrosion are no longer evident. Carbonated concrete in

contact with reinforcement must be removed, as this will not provide the steel

reinforcement with the layer of passive alkalinity necessary to impede corrosion.

Reinforcement is cleaned using grit-blasting or high pressure water jetting and these

techniques are also used to prepare the surface of the substrate (parent) concrete, ready for

22
 iw iiiiu iv v u v -v ju w iv ^iv/ u i i t / j j a u . a n u v c iv ic w

application of a repair material. Chloride contaminated concrete must also be removed

90
wherever possible .

For seriously affected concrete elements where removal of concrete becomes economically

prohibitive there are a number of repair techniques that can be employed that minimise

concrete repair. These are discussed in section 2.6.4. However, in cases where spalling or

cracking have occurred, or where the extent of corrosion is such that the structural capacity

of an element may have been affected, concrete will always require removal or

replacement (or both the structural capacity will have to be reinstated and corrosion

arrested by other means).

2.6.2 Repairing spalls

Spalling is repaired by the application of a suitable repair material using one of a variety of

methods.

o Patch repair - This type of repair involves the application of hand applied mortar. It

is suitable for small repairs,

o Sprayed repair - This technique is the most widely used concrete repair particularly

since it does not require shuttering. It also provides a good bond between substrate

and repair material.

o Flow repairs - These repairs involve the use of shuttering to form a cast into which

the repair material can be poured.

The selection of a suitable repair material is both important and complex. For each method

of application the factors involved in the selection of a repair material are discussed in

Chapter 3.

23
 v.impici ^ - ivciiiiurcea concrete oriage repair: an overview

Typically, these methods (remove and replace) are used to repair the majority of defects

associated with reinforced concrete highway bridges. These include spalling and cracking

caused by carbonation, chloride ingress, attack from sulphates, impact and freeze-thaw

damage.

2.6.3 Repairing AAR affected concrete.

In AAR affected concrete, it is important to establish (from the petrographic analysis) the

amount of reactive material in the concrete matrix. This information will determine

whether the total effects from the reaction have been exhibited or whether the current
• • • '!')
condition of the concrete will worsen . Generally, if the alkali-aggregate reaction has run

its course and the structural capacity of the element has not been impaired (a structural

survey may be required to determine this), then the surface cracking caused by the AAR

can be sealed to restore the aesthetic appearance of the surface and prevent the access of

aggressive agents below the concrete surface. Even if the AAR has caused a degree of

structural instability, the cracking can be injection grouted by the technique where resin is

forced under pressure to permeate the crack pathways in the substrate concrete. If

laboratory testing shows that the alkali-aggregate reaction will continue in the concrete,

then this reaction must be arrested by reducing the internal humidity of the concrete. This

can be achieved by sealing surface cracks (or impregnation) and the application of a water
9<
repellent surface coating .

2.6.4 Unobtrusive alternatives to repair

Occasionally the extent of concrete deterioration is such that the removal and replacement

of the affected concrete becomes tantamount to replacement of the element. In these cases,

24
 ^najjici z, - iveim orcea concrete Dndge repair: an overview

or in cases where corrosion is expected to continue after repairs have been completed (e.g.

in chloride infested concrete) a number of alternative approaches are available36.

2.6.4.1 Cathodic protection

Reinforcing steel in concrete seriously affected by chlorides (and occasionally carbonation)

can be protected by a cathodic protection system. When a corrosion circuit has been

formed (after the passivity of the steel has been compromised) the steel acts as both

cathode and anode in an electro-chemical circuit. A cathodic protection system maintains

the steel as a cathode in an electrical circuit driven by an impressed current37. Anodes are

installed on the concrete surface and are electrically connected to the steel reinforcement,

this process reverses the electrical current flow which causes corrosion21. Cathodic

protection systems require constant monitoring and adjustment. This process is often

automated using expensive equipment. As such, careful economic comparisons need to be

conducted before embarking on such a scheme.

2.6.4.2 Re-alkalisation

An electrochemical technique is available which restores the protective alkalinity around

reinforcing steel without the need to remove the carbonated concrete. It introduces an

alkaline solution through a process of electro-osmosis. Another process stimulated by

equipment at the concrete surface is electrolysis which results in the generation of ions

which re-passivate the steel surface. The re-alkalisation process must be accompanied by

the application of a surface coating to the concrete to prevent a reoccurrence of the

'>1
carbonation process .

25
 v^najjici x - xvciiuorcea concrete oridge repair: an overview

2.6.4.3De-salination

De-salination is a technique which extracts chlorides from concrete. It utilises similar

techniques to those of electro-chemical re-alkalisation to remove negatively charged

chloride ions from the concrete. The electro-chemical reaction results in the migration of

the chloride ions to the surface mounted anode. In tests salt concentrations of 6 to 12 kg/m3

were removed over a period of 100 hours24. The processof desalination is known to take

between 3 and 8 weeks to complete and may result in repassivationof the reinforcingsteel

by the generation of hydroxyl ions.

2.6.5 Concrete Protection

2.6.5.1 Corrosion countermeasures

Concrete surfaces can be coated or sealed to prevent the access of aggressors into the

substrate. These measures are often taken after repairing to ensure defects do not re-occur

through similar mechanisms, or to prevent other likely deterioration processes. Coatings

and sealers generally fall into the following groups:19

o Water-repellent surface impregnants

o Surface hardeners and pore blockers

o Cement-modified polymer coatings

o Elastometric polymer membranes

The purposes of these coatings are generally to reduce or stop the ingress of oxygen,

carbon dioxide, chlorides, and water. Oxygen and water are necessary to support

26
 ^im pici ^ - jvcnuorcea concrete o n d g e repair: an overv iew

reinforcement corrosion and alkali-aggregate reaction. Carbon dioxide causes carbonation

which destroys the passive alkaline environment of reinforcing steel5.

2.6.6 Dealing with cracking

Pattern cracking is a term used to define areas of concrete affected by a large number of

small cracks. It is generally expected to be repairable in a similar manner to spalled or

delaminated concrete. Larger cracks, structural or otherwise can be repaired using

established methods. The root cause of non-structural cracking is often the same as the

cause of eventual spalling. Additionally, moisture and thermal effects during curing can

cause cracking, as can AAR. Repair strategies for these were described in section 2.6.3.

Larger cracks will often be caused by structural effects such as overload or differential

settlement. Discovering the cause of a crack is essential before repair methods can be

determined. In addition to the cause, the status of the crack is also important. Status is

defined by three categories: category 1, the crack is actively widening, category 2, the

crack is active but not widening (i.e. opening and closing), category 3, the crack is

dormant. If a crack is of sufficient size and nature as to alarm an engineer a crack

movement indicator is employed to establish its status29’38. The results of such a survey,

and other determining conditions can be used to match the crack defect to a suitable repair

method, as recommended by Kalyanasundaram et al39 and represented in Figure 2.4

27
 z. —R tuiiun-cu cunuieie oriuge repair: an overview

RK RS 00 ES.BE E5 RS
AH AH
00 AH I ES AH I BL

lypa of

£0 - E*t*r.*ibl« Cw«riw R* - RadMign and ProvicSa Expa-mum Joint S - Stitching ES - Sxtam al Sti~«w«ing 0t- - BIan* a ting
AH - Autogarcu* Haaling 00 - Ordinary Ovariay S - Grouting BE - Bonding with Eptwy RS - Rout »nd S ssl

Sam ple Knowledge Net for Suggestion of Repair Method for ‘ Cracking In C oncrete’’

Figure 2.4 R epair m ethods for cracks39

A diagram o f common crack types and an accompanying table giving their details are

given in Figure 2.5. There are various techniques o f crack repair which can be em ployed

depending on what needs to be achieved by the repair. Crack repairs provide some o f the
9Q
following functions :

o Restore or increase strength or stiffness

o Improve functional performance

o Provide water-tightness

o Improve appearance

o Improve durability

o Prevent access o f corrosion inducing material to reinforcement

Depending on the performance requirement o f the repair, one o f the follow ing repair

methods is usually considered:

28
 ^uapici - ivcmiuxLeu concrete oriage repair: an overview

o Epoxy injection - The technique involves drilling holes at regular intervals along a

crack and filling the void created with epoxy,

o Routing and Sealing - this is the most common method of crack repair. The

procedure involves cutting out a groove along the length of a crack (routing) and

then sealing the groove, this prevents ingress of moisture. A bond breaker is usually

added to the unsealed routed groove,

o Stitching - Stitching involves the drilling of holes either side of a crack and placing

4stitching dogs ’ or ties holding the two sides of the crack together,

o Flexible Sealing - this involves active cracks being routed out, cleaned, and filled

with flexible sealant.

o Grouting - Wide cracks in thick walls can be repaired by filling with cement grout,

o Polymer Impregnation - Cracked concrete surfaces can be dried and flooded with a

monomer which is polymerised by heating,

o Overlays - These can restore structural integrity and prevent the access of

aggressors into the concrete,

o Autogenous healing - this is a natural process of crack repair (self heal) in

concrete. It has a practical application for closing narrow dormant cracks in a moist
29
environment .

o Movement joints - Live cracks are often candidates for conversion into movement

joints. In this procedure, a recess is cut along the line of the crack and filled with a

flexible material40.

Higgins40 details the performance characteristics of various crack repair methods. An

expert would take into account all the necessary factors before selecting the most

suitable repair method.

29
 Chapter 2 - Reinforced concrete bridge repair: an overview

Type of Letter (S ee Sub d ivision Most Primary Seco n d a ry See S ectio n

C racking Figure 8.2) Com m on C ause C a u se s/
Location (Excluding Factors
Restraint)

<
Over Deep
reinforcement sections

to
Plastic Arching Top of E xcess Rapid early
settlem ent columns bleeding drying 2.2.1
conditions

o
Change of Trough and
depth waffle slabs I ..............

Q
Diagonal Roads and
slabs Rapid early

Ui
| Plastic Random Reinforced drying
c\j
**
CM

| shrinkage concrete Low rate of

slabs bleeding

u.
I Over Reinforced Ditto plus
reinforcement concrete steel near
slabs surface

0
t External Thick walls Excess heat Rapid
Early restraint generation cooling
thermal 2.2.3

X
contraction Internal Thick slabs Excess
restraint temperature
gradients
| Long-term Thin slabs Inefficient E xcess 2.2.4 I

T
| drying (and walls) joints shrinkage
I shrinkage Inefficient
curing
'Fair faced’ Impermeable
“5

Against
formwork concrete formwork Rich mixes

O
c
V
73
2.2.5
CO
*

Floated Slabs Over- Poor curing

concrete trowellinq
-J

Areas subject Chloride

to leakage or ingress
cm in
d cJ

| Corrosion of spray from

COCM

i reinforcement road run-off

5

(Sheltered Carbonation Porous

cvi ci
cm in
CO*-

areas) concrete.
Low cover
z

Alkali-silica Reactive aggregate plus high 2.2.9

TO 8
o 8

reaction alkali cem ent

s
E
5
C

H
i.03
£
m

ua>
03
tx
a,

M
O
CS
c
cs
JO

a.
TBJ
cs

V
£

B .
R
X
-*o->

>>
 li m p i d z - ivem iorcea u on crete B ridge repair: an overview

2.7 Expert Systems for concrete bridge repair

2.7.1 Review of existing developments

Concrete bridges are required to maintain their serviceability over long periods of time.

The reinforced concrete which constitutes these bridges is susceptible to attack from

aggressive environmental agents, attack which if left unchecked can severely reduce the

service life of a bridge.

Managing the condition of bridge networks involves teams of dedicated engineers

inspecting and monitoring the performance of all elements of the bridges under their

jurisdiction. This also involves decision making on when repair and remediation is

necessary. The knowledge used in this decision making process is not well documented.

There are no comprehensive formalised standards for concrete repair material selection41,

and no standard documents to aid engineers in the diagnosis of defect causes.

The aim of this thesis is to construct an expert system which can dispense the best practice

instruction tailored to fit any concrete repair situation. This has been attempted to varying

degrees by others in the past, with limited degrees of ambition and success. Anumba and
18
Bowron suggest in their proposed system that accurate diagnosis is a 'sine qua non’ in

the repair of concrete structures. They suggest an expert system could provide a more

objective approach to the choice of concrete repair materials.

The American system, HYWCON42, developed in the SHRP programme demonstrates

knowledge engineering in the bridge repair domain at a very basic level, its structure is

shown in Figure 2.6. A sub-system of the HYWCON program diagnoses and recommends

repair strategies for three key concrete defects defined by it: cracking, spalling, and

disintegration.

31
 lim p id z - w eim orcea co n c re te b ridge repair: an overview

CONSTRUC-O Subsystem
(S u bstrudures)

D istress
Category

Cracking Spaiiing Disintegration

and popouts and scaling

longitudinal
or transverse

— map or pattern
diagonal

random

Figure 2.6 Structure o f HYW C O N expert system 42

The system is structured into three sub-systems, each one diagnosing causes for different

types o f surface defects.

Figure 2.7 to Figure 2.11 give an example o f the process a user would follow from

beginning to end. This program functions only in the M icrosoft W indows 3.1 graphical

user interface environment; an indication o f its age!

H HWYCON, CONSTRUC-O Ver 4.0 - July 1 9 9 4 {Bridge P e

Of what type of construction is the bridge

made of?

Enter
® iConcretei

O S teel and Concrete

Figure 2.7 Am erican H W Y C O N system step 1.

The program is used to diagnose the cause o f defects in concrete bridge decks.

32
 uim pici z - rvcnnui veu co n c rete tsnage repair: an overview

HWYCON, CONSTRUC-O Ver 4.0 - July 1994 (Bridge De

Select the exposur e to chloride ions from the questions below , then clic
on the "Enter" push button.

If e x p o s e d to f r e e z i n g c o n d i t i o n s , a r e d e i c i n g
sa lts applied?

O Y es Enter

O No

Figure 2.8 A m erican H W Y C O N system step 2.

This system next requires the user to enter the bridge’s likelihood o f exposure to chloride

ions.

CONSTRUC-O Ver 4 .0 J u ly 1 9 9 4 fB ridge D eck s)

Select the observed distress(es) from the 1

below, then click on the ’Enter* push butl

□ Cracking Enter!

Picture
Ed S ca lin g

Figure 2.9 A m erican H W Y C O N system step 3.

At this stage, the user o f this system is requested to inform the knowledge base o f the

effects o f the defect. The user is provided with graphical examples o f the likely effects

(Figure 2.10).
 v^uajjiti z. - ivcim uiceu co n c rete tsn ag e repair: an overview

CONSTRUC-P Ver 4 .0 - J u ly 1 9 9 4 (Brid

Crac ki ng

Figure 2.10 A m erican H W Y C O N system step 4.

Finally, the system uses the acquired information to generate a diagnosis for the

encountered defect (Figure 2.11).

CONSTRUC-D Ver 4.0 - July 1994 (Bridge D ecks)

-----
Restart!

R ep ea t s e s s io n

Do another d is tr e s s

Figure 2.11 American H W Y C O N system step 5.

Immediately clear from this example is the narrow scale o f the knowledge base. Complex

decision making is avoided as the user is limited in the options available at each stage. The

finishing inference is, in all cases, a very general piece o f repair advice (such as ‘break out

the affected concrete and repair'). The program is the only purchasable expert system for

34
 ^ uapici z. - xvciuiuicea L.oncre:e Jtsriage repair: an overview

reinforced concrete repair discovered in this review. Such a basic system is of limited use

to bridge engineers as it delivers only simple advice and lacks intelligence. Here a

distinction is made between an ‘expert system’ and an ‘intelligent expert system’. For

example, in such a system as HWYCON, if a user encountered a very small defect in some

reinforced concrete, the advice generated by the program would be identical to the advice

generated for a very large defect with the same symptoms (e.g. such common generated

advice as; cracking through exposure to chlorides). Hence, such a system lacks the

intelligence to account for severity and extent of defect when making a decision. Severity

and extent are identified as crucial factors in the production of an intelligent expert system

for concrete repair18. Additionally, HYWCON evaluates distresses individually and there is

no provision for advice in situations that involve multiple causes of distresses that occur

simultaneously at one location42. It is recognised that an intelligent expert system should be

able to examine defects collectively43. A system such as HWYCON which generates repair

advice on a defect by defect basis and fails to advise the user when an element is severely

affected by multiple defects lacks basic intelligence.

The use of graphics in the system, although limited, is a great advantage over textual

descriptions. If expanded to include for example, different crack types and their causes, the

extent of the system could be improved.

The MENTE-KUN prototype expert system44 once again uses a knowledge base to

question a user about the nature of concrete defects. It concentrates on the nature of a

defect, (i.e. spalling, cracking, abrasion) and does not account for severity and extent. In

the program, the user is expected to judge if the severity and extent of the problem are of

such magnitudes as to warrant action.

REPCON is a text-based prototype expert system for concrete repair39. It is essential the

user communicates concrete defects unambiguously to the program, otherwise the

35
 ^napici x - ivciiiiorcea c o n cr e te im a g e repair: an o v erv ie w

communication gap increases, leading to wrong diagnosis39. Similarly, this program

generates generic advice for repair of reinforced concrete defects. It does not attempt to

judge the severity and extent of deterioration before giving its advice.

CODBA43 (concrete bridge deterioration assessment) is a prototype system developed to

diagnose deterioration in concrete bridges. The program attempts to facilitate the visual

assessment of concrete bridges in order to recommend in-depth testing procedures.

Generally, the majority of expert systems for concrete repair are text based prototypes45,46.

Additionally, these existing systems have concentrated on diagnosing the cause of singular

defects and not the effect of multiple defects on a single element. In addition, existing

systems have generally failed to take account of the extent and severity of defects when

diagnosing their causes and effects.

An intelligent expert system should be able to assess individual defects and the effect of

multiple defects, including their causes. It should also be able to judge the effect of the size

and severity of defects on an element and, thereby, assess the condition of the element.

After the assessment of an element, an intelligent system should be able to recommend a

suitable regime of test procedures to confirm the initial diagnosis. Furthermore, once

testing has been completed, optimal repair recommendation should be made. An intelligent

system will be flexible and have the ability to cope with the fact that for any given

situation, often more than one cause may have led to the defect, or it may be difficult to

identify the cause and several causes may be suspected. A number of repair options may be

possible as a result.

36
 ^ - iv cn u u iccu c o n c r e te tsriage repair: an overview

2.7.2 Architecture of expert system for reinforced concrete repair

Anumba & Bowron18 suggest the architecture of an expert system for concrete repair

should include two key components, an intelligent diagnostic component, and a repair

specification component.(Figure 2.12)

Input
Defect
Details

f Intelligent
Defect Unknown ->j Diagnostic
\Componem
CONCRETE
REPAIR
SYSTEM

Input REPAIR j
Repair s p e c if ic a t io n !
Parameters

Figure 2.12 Concrete repair expert system: basic architecture18

2.7.2.1 Diagnostic component

The purpose of the diagnostic component of an expert system for reinforced concrete

repair is to firmly establish the cause of a defect. Defect effects are exhibited in three key
• • • OQ
ways, cracking, spalling and disintegration . A diagnostic system should assess the

severity and extent of such effects, and use simulated expert knowledge to derive a

suspected cause. An intelligent system should also be able to determine if the extent of the

defects is of sufficient magnitude to be a cause for concern. Similarly, if a system did

suspect defects were of a sufficient magnitude to cause concern and as a result

recommended some tests be carried out, then based on the results of the testing the

software should be able to recommend if the defect is significant enough to require repair.

37
 ^uapici z. - rvciiuoi'ueu concrete tsriage repair: an overview

In a bridge assessment procedure it is crucial to integrate all defects discovered in an

element in order to form a combined subjective rating of the element with accuracy43.

The first stage in the development is to collect the relevant knowledge in the field. This

knowledge is expressed in a format which maps easily into an expert system (knowledge

net). One such system47, makes no account for severity and extent, although it generates a

confidence factor which describes the confidence the system has in its diagnosis being

correct. These confidence factors appear to be static, i.e. for a certain set of inputs the

confidence factor in a decision can only be generated as a pre-specified figure of 40%,

60% or 100% (for example). Therefore in this example (Figure 2.13) the issue of

confidence factors is similar to the use of natural language qualifiers (i.e. low, medium,

high). The approach of adding an indication of the certainty of a decision makes a system

more intelligent.

Some attempts or outlines for the creation of expert systems for concrete repair have

suggested that the difficulty in diagnosing a unique cause to a defect limits the

development of expert systems in this field47. This assumes that each defect has a unique

cause, which is not always the case. In addition, the decision making process will consider

different causes as diagnostic data are incrementally provided to the expert system.

It can be inferred, from the attempts to create expert systems for concrete repair, that

expert system technology is ideally suited to application in this field47.

Bridge maintenance is particularly suitable for exploitation through an expert system,

owing to the fact that many problems exist in the domain that can be only be solved

heuristically1.

38
 ^im pici z, - w em io rcea u on crete tsridge repair: an overv iew

Any program must find the cause of a defect by analysing the symptoms1. Figure 2.13

demonstrates an interactive session with the DIAGCON expert system. It demonstrates a

simple text based interface expert system.

> WELCOME TO D IA C C O N
Please answ er the fo llo w in g qu estion s with either
"true" or "false" or w ith relevant data, as the case
m ay be. The proform a should have been m ade
available to y o u before this session.
> basic sym p tom is cracking? T
> direction o f cracking? VERTICAL
> rust stains or sp ots present? F

> structural elem ent? BEAM

> cracks originate in m axim um m om ent region? T
> crack w id th m a x im u m at top or b o tto m o f
beam? T

> PROBABLE C A U S E :
Cracking d u e to flexural capacity of beam being
exceeded.
> crack determ ination m ethod? TWO
> glass strip is cracked or disjointed? T

Figure 2.13 Typical session with DIAGCON expert system47

DIAGCON, and the other prototype expert systems identified in this chapter attempt to

diagnose effectively the two key causes of concrete defects: cracking and spalling39.

Diagnosis of concrete defects can be standardised into common procedures as

demonstrated in Figure 2.14.

39
 ^ napici z. - ivciiuurcea c o n c re te tsridge repair: an overview

v is u a l
M a in t e n a n c e o f C o n c r e t e S tr u c t u r e s
in sp e c tio n

1
n o n -d estru ctive la b o r a t o r y d e t a i le d
tests tests
i n v e s tig a t io n

p h y s ic a l and
seriou s
no d am age •< ev a lu a tio n > - chem ical t e s t s
d e t e r i o r a ti o n

in c r e a s e
in sp e c tio n
T
s m a il d a m a g e s ^^^aluatlorN*-
c a u s e of
d e te r io r a tio n
frequency

repair
p r e v e n ti v e repair proposal d e m o litio n
an d
m easure a n a l y s i s of
costs rebuilding

Figure 2.14 Approach to concrete structure maintenance1

Another expert system discovered through this review is REPCON which is designed to

aid experienced engineers in finding out the causes of damage to concrete structures and

give tentative repair recommendations1. The prototype of REPCON showed that the use of

expert system technology in this domain is a possible way to provide the knowledge,

which is dispersed in numerous publications and in a few human experts. The structure of

REPCON is shown in Figure 2.15.

40
 ^napier z - K.em iorcea c o n cr e te B ridge repair: an ov erv iew

description of
structure and
elem ents
I .....
:----------------
carbonatlon chemical physical
and cracks
chlorides attacks attacks *•«

e.g. acids freeze-thaw .erosion

DAfStb ' P r o t e c t i o n an d Repair o f C o n c r e t e S t r u c t u r e 12.89

p r in c i p le s o f repair co n crete
others
corrosion
cracks coatin gs
p ro tectio n

r e a ik a l is a t io n filling or h y d r o p h o b ic im p r e g n a tio n
r e d u c e w a te r c o n t e n t in j e c t io n c o a t i n g w i t h o u t c r a c k b rid g ing
c oatin g of rebars w it h c e m e n t c o a t i n g w i t h c r a c k b rid g in g
k athodlc p ro tectio n EP or PUR ch em ical r e sista n t coatin g
m e c h a n i c a l r e s i s t a n t c o a t i n g ...

Figure 2.15 Structure of REPCON1

Rajeev and Rajesh47 (Diagcon) state that it is unfortunate that the results of their system do

not always lead to a unique conclusion. However, the opinion of experts from a visual

survey will often not lead to a unique conclusion as to the cause of the visual defects.

According to Rajeev and Rajesh47 once the cause of deterioration has been identified, the

next step is to decide a suitable repair method. Although they recognise that repair should

only be undertaken when the defect has been diagnosed with some certainty, their system

does not recommend testing procedures to confirm the diagnosis of the visual inspection.

2.7.2.2 Severity and extent ratings

Using fuzzy logic, the extent and severity of each defect or each cause can be expressed in

terms of linguistic variables, and both extent and severity can be combined43,48. Fuzzy

41
 ^ n ap ici ± - ivem iurcea co n c rete b rid g e repair: an overview

logic is a subset o f conventional logic which has been extended to handle the concept o f

partial truths. Figure 2.16 shows a typical fuzzy set. It demonstrates a relationship between

natural language and numerical judgem ent. If someone was to assess a statement as being

‘very true’ and that statement was converted from language into a num erical value (where

unity represents absolute truth) then in effect fuzzy logic ascribes a zone o f values instead

o f a singular value. The vertical axis in the figure represents certainty. The technique can

be used where natural language qualifiers such as ‘small, medium, large’ need to be

handled numerically, but the vagueness o f the language also needs to be modelled (Figure

2.16)

fniriv »r..<»m 0re'0 r ' {ess' tnje true very-true

rainy-true 3
1.0 absolute-true

0.8

0.6

0.4

0.2

0.0
1
R e lia b le

Figure 2.16 An exam ple o f fuzzy sets49,50

2.7.3 Handling uncertainty

The Mente Kun prototype utilises certainty factors. An expert who constructs an expert

system ought to know the importance o f ‘fuzziness’44. An approach should be identified

which can handle uncertainty when developing an expert system.

A certainty factor (CF) is a numerical value that indicates a measure o f confidence in the

value o f that param eter1. For example IF DAM AGE = CRACKS

42
 sim p ler z - K ein iorcea co n cr e te B ridge repair: an overview

CRACK TYPE = RANDOM PATTERN

CRACKINFO = WHITE RESIDUE

CRACK SIZE = LARGE DEFECT

THEN CAUSE = AAR CF 60%

A similar approach is adopted by Rajeev & Rajesh47 where a ‘confidence factor’ is

appended to the result of a rule.

If data available is reliable and extensive, the expert can pin-point the most appropriate

repair method with full confidence. With uncertain information, repair procedures may still

be specified, but with less confidence39.

2.8 General application of Information Technology in Bridges

2.8.1 Bridge Management Systems

Bridge management systems were first developed in the 1980s in the USA51. They consist

of databases that store the key information regarding bridges; bridge inspectors and

engineers are required to refer to these databases in their day to day practises. Early bridge

management systems were developed for data storage and retrieval purposes, such as:

• Entering bridge data

• Viewing inspection results

• Viewing and editing bridge data

• Viewing forthcoming inspection dates for bridges

However, modem systems contain advanced modules that can be used to predict the most

cost effective long term repair strategies by using ‘whole life costing’ methods52. In

addition, modem systems attempt to model the deterioration of bridges using complex

43
 v^napici z. - ivciniorcea uoncrete uridge repair: an overview

53 • • • • •
algorithms . Bridge controlling authorities in some countries have constructed extensive

bridge management systems to catalogue a large number of structures. The American (US)

PONTIS54 system is used in many states to create inventories of their large bridge stock.

The Highways Agency in England use the SMIS system (structure management and

information system). The results of all principal inspections on Highways Agency

structures are fed into SMIS by competent trained personnel. As a result an accurate

electronic record of the conditions of the Highways Agency’s bridge stock is kept. The

Highways Agency hope to procure new software which will interrogate SMIS in order to

identify which bridge repairs will provide the best value from the Agency’s yearly budget.

Information technology is rapidly expanding in the bridge repair and maintenance field as

the benefits of more esoteric software (such as expert systems) are promoted.

44
 x— _ v/x luuiuiiuu i \ji wpiunai jjciium iaiiL c u i uuncreie repair

3 Selection of materials for optimal performance of

concrete repair

3.1 Chapter Objectives

In order to function satisfactorily, a repair material applied to reinforced concrete must

perform several functions:

• Bond strongly to the substrate concrete

• Bond strongly to the reinforcing bars

• Have an adequate tensile strength to accommodate restrained volume changes

• Prevent penetration of water, chloride solution and carbon dioxide to reinforcing

steel

• Share load with the substrate concrete if necessary

The ability to determine how well repair materials will perform, under the varying

conditions in which they may be employed, would enable an engineer to make intelligent

choices when repairing concrete defects. Currently however, there is disparity amongst the

opinions of researchers regarding which properties of repair materials are the most

important to specify. As a result of the lack of a clearly defined method forthe selection of

repair materials, they are currently selected by many practitioners on an ad hoc basis.

Some engineers recommend using similar values for the respective properties of both the

substrate concrete and repair materials; others recommend high compressive strength and

low shrinkage repair materials for the majority of situations. The number of permutations

45
 ujjuiucu pciium ictiiue u i concrete repair

of different recommendations is high, and the lack of a coherent opinion can often lead to

the simple practice of the highest strength material being selected.

c f c £ cn c o
Recent research ’ ’ ’ suggests that the key mechanical properties of a repair material to

be considered when selecting a material for reinforced concrete repair are:

• Elastic Modulus

• Shrinkage

• Creep

Knowledge of the growth with time of these properties in a repair material, and of the

interaction between the substrate concrete and repair material at their interface has allowed

the development of a technique to predict the short and long term performance of concrete

repairs. In this chapter, the development of the method to predict the in-situ performance of

repair materials for reinforced concrete is outlined. The method requires knowledge of

certain mechanical properties of both the substrate concrete and the repair material. Curing

effects caused by local seasonal temperature and relative humidity variations, along with

dimensional shrinkage differences are also taken into account.

The chapter begins with a literature review of the relevant domain.

46
 ujjuuiai pci ium m rice 01 concrete repair

3.2 Literature review

3.2.1 Introduction

Reinforced concrete is a strong and durable material; approximately eighty percent59 of

bridges in the UK are constructed using it. Reinforced concrete structures give excellent

durability when designed, constructed and maintained correctly, justifying their design

lives of 60 to 120 years60. Approximately 500 million pounds is spent in the UK each year

on the repair of concrete61. Reinforced concrete deteriorates due to environmental effects

such as the ingress of carbon dioxide which neutralises the natural alkalinity of the

concrete, or the diffusion of chloride solution through small cracks and the pores which

depassivates the reinforcing bar’s environment. Both these effects can lead to cracking of

the concrete surrounding the reinforcing steel. This cracking can delaminate the concrete,

and eventually delaminated concrete can break away from the parent concrete (substrate).

This process is known as spalling, and the remaining patch of exposed sub-surface

concrete is known as a spall. These effects, and a number of other aggressors can either

directly cause a spall, or persuade the engineer (or expert system) to recommend the

removal and replacement of the affected concrete. The result of these defects can be a loss

of strength in the affected members, demanding immediate attention. Alternatively, defects

can merely prove aesthetically unacceptable; these defects require repair to halt further

deterioration and to reproduce the original appearance of the substrate.

Once a defect is discovered and diagnosed, loose concrete and other defective areas are

removed and the exposed substrate concrete is prepared for the application of a repair

material. The interaction of a repair material with the substrate concrete is the crucial

factor in determining the performance of the repair patch55,56,57,58,61’62’63’64>65# Volume

47
 icuciiaia iui upim iai periorm ance 01 concrete repair

change in the repair material (usually shrinkage) is restrained by the substrate concrete, and

occasionally this restraint to shrinkage can cause tensile stresses which exceed the tensile

strength of the repair material. Understanding the interaction between the repair material

and the substrate concrete will allow the engineer to carefully select the properties of the

repair material to ensure adequate performance.

The current standard for the specification of materials for concrete repair on Highways

Agency structures is BD 27/8666. This standard recommends storage methods, densities,

aggregate size and constituent proportions and some mechanical property values for:

• Decks and vertical surface to piers, columns and abutments

• Sides and soffits of beams and crossheads

Although BD 27/86 recommends material types and cement contents, the only mechanical

property of repair materials recommended is compressive strength.

A more thorough standard for concrete repair will be the eurocode ENV 1504-1:199722,23

currently available in draft. This code takes a more sophisticated approach to the

specification of concrete repair properties which encompasses modem thinking on which

properties are crucial to specify.

3.2.2 Selecting repair materials for reinforced concrete

Any successful repair material should have the ability to67:

• Arrest the deterioration of the stmcture by preventing access of oxygen, water and

aggressive ions

• Provide an environment that chemically passivates the reinforcement

• Restore the stmctural integrity of the element

48 •
 ui upim mi pciiu m m iiu e 01 concrete repair

• Restore the aesthetic appearance of the element

Selecting repair materials that can deliver these performance requirements involves a

thorough understanding of material behaviour in anticipated service and exposure

conditions . In truth, the performance of concrete repair materials has been an under­

researched field. The resulting lack of understanding of the behaviour of in-service repair

materials necessitates a systematic approach to concrete repair design and construction.

The selection of design values and decisions needs to be more rational69 and, ideally, a

broad range of research - particularly new, state of the art research in the field - needs to be

collated and consolidated to form rigorous new guidelines. The financial benefits to

ensuring the success of repair materials are considerable.

A repair to reinforced concrete can be affected by five key factors70:

• The effect of the constituent materials on the properties of the repair material

• The effect of the properties of the repair material on its durability

• The effects of environmental conditions on curing and durability

• Effects due to the interaction between the repair material and the substrate concrete

• Loading effects (transfer of load into repair, depropping etc)

The importance of accurately predicting the performance of a repair at the design stage is

crucial. Studies69 have shown that the level of influence on the durability of a repair

material is at its highest during the design phase of the repair (Figure 3.1).

49
 upmiicu p ciiu n im iiu e o i concrete repair

100%

DECREASING INFLUENCE

0% CONSTRUCTION
OESIGN
PERIOD PERIOD

Figure 3.1 The effects of different periods on project quality (durability)69

Figure 3.1 demonstrates the high level of influence over the quality of a repair project

which is wielded at the design stage. This reinforces the need for a critical evaluation of

current methods for the specification of reinforced concrete repairs.

Concrete repair may be broadly categorised as structural or protective68,69. Failure of repair

materials is undesirable in either of these circumstances. In order to alleviate, at the design

stage, all possibility of failure, the key properties that influence the performance of the

repair need to be identified. As stated previously there is some disparity amongst engineers

about the key properties to be considered when attempting to combat the failure of repair

materials. Furthermore, there is also some disagreement, once key properties have been

established, as to the relative values these properties should hold in order to reduce the risk

of failure.

If failure does occur, it is invariably through two key mechanisms; restrained volume

changes and loss of bond.

50
 uviw^uuii ui mai&iiaia iui upmixai p e n o n n a n c e 01 concrete repair

3.2.2.1 Restrained volume changes

Cement based repair materials are volumetrically unstable69. During the curing process the

fluctuation in moisture levels within the repair causes volume changes.

Reinforced concrete repair materials undergoing volume changes are restrained by the

substrate concrete and also partly by the reinforcing steel. As this occurs, tensile strains are

induced which, if greater than the tensile strain capacity of the repair material, will cause

cracking69. Volume changes must, therefore, be controlled in concrete repairs to prevent or

minimize cracking. However, predictions of the magnitude of developed strains in repair

materials need to take account of the complex interaction between properties such
5 5 ,5 6 ,5 7 ,5 8 ,6 2 ,6 3 .
aS

• Elastic Modulus

• Shrinkage

• Creep

Figure 3.2 shows that restrained volume change has one of two outcomes64,71 which can

contribute to the failure of concrete repairs. The restrained volume changes can lead to

cracking, which provides a passage for moisture though to the steel. If the moisture

contains aggressive agents, the steel can become depassivated and corrode, larger cracks

can occur, and the cycle can perpetuate, leading to spalling, and potentially a dangerous

loss of strength. The other cause, bond strength, is a factor very much affected by on-site

preparations. If a surface is thoroughly prepared for the application of a repair, following

best practice guidelines, and if the bond strength of the repair material is adequate, then a

repair should not fail through this mechanism.

51
 vjpmnai jjci lunim iice u i concrete repair

REPAIRED CONCRETE STRUCTURE

Repair

Weakened Bond Between

Rebars & Concrete Along
the Repair Perimeter
Restrained Volume
Changes

Exterior Weathering
&Loading Effects

Cracking

Increase in Permeability
Along the Perimeter of
the Repair

Penetration of H,0,
Penetration of CL* from Outside
HjO, COj, CL-
from inside

(1) Depassivation of the Steel Reinforcement

(2) Formation of the Rust Products

1
(1) Accumulation and Expansion of Rust Products
(2) Loss of Bond Between Reinforcement and
Repair Material

Expansion, More Cracking, Expansion, Cracking,

Enlargement of Existing Spaiiing of Existing Concrete
Cracks, Spailing Adjacent to the Repair

REPAIR FAILURE

Figure 3.2 A model of repair failure70

Measures to combat the effects of restrained volume change (usually shrinkage) are more

complex72,73. Emmons and Vaysburd suggested that repair materials with low strength, low

shrinkage, high creep and low modulus of elasticity were most desirable for non structural,

protective repairs69. This combination of properties, it is suggested, will produce a high

52
 icua iui upm nai pcnunrm nce 01 concrete repair

strain capacity repair system. Their hypothesis is based on their intuitive understanding of

the cumulative effects of these properties.

Morgan74 suggests that the potential for success or failure of repairs depends upon:

• The magnitude and the state of the stress field

• Whether the load is left on the structure during the repair operations

• The creep capacity of the repair materials

• The quality of tensile and shear bond strength of therepair material to the substrate

concrete

• The temperature at which the repairs were carried out and subsequent range of

temperatures during service life

These deductions specify tensile and bond strength as well as creep being the key

mechanical properties. Once again, these deductions arebased on an intuitive

understanding of the effect of these properties on repair material behaviour.

Table 3.1 represents the conclusions into a study of the significance of property mismatch

between repair and substrate. It attempts to stipulate ideal relationships between substrate

and repair material properties for a successful repair, although it does not recommend

values of these properties, just their values relative to each other.

Emberson and Mays61 have stated that repair materials can be deemed suitable on the basis

of their compressive, tensile and flexural strengths alone. They also recommend high strain

capacity in the repair material, and a modular ratio (the ratio of the elastic modulus of the

repair material to the elastic modulus of the substrate concrete) of unity.

53
 ^ ui maiEi iciia iui upunmi pcnurnm nce 01 con creie repair

Table 3.1 Requirements of patch repair material for structural compatibility61

Properly Relationship o f repair mortar (R)

to concrete substrate (C)

Strength in compression, tension and flexure R > C

Modulus in compression, tension and flexure R * C
Poisson’s ratio Dependent on modulus
and type o f repair
Coefficient o f thermal expansion R ft C
Adhesion in tension and shear R > C
Curing and long term shrinkage R C
Strain capacity R £ C
Creep Dependent on whether creep
causes desirable or undesirable
effects
Fatigue performance R £ C

Table 3.2 summarises the recommendations for values of repair material properties given

by eleven independent researchers. It recommends whether the values of compressive

strength, tensile strength or Young’s Modulus for a repair material should be greater than,

lesser than, or equal to those of the substrate concrete. A lack of agreement is shown on the

relative importance of the strength and elastic modulus. Where available, the opinion on

creep confirms its importance. Additionally, shrinkage is considered important, and should

be low.

54
 Properties of repair materials

28 day

y
m
a
-1

13

<N
oo
28 day tensile Noteworthy

>>
compressive

g
on

o
<n
00
Reference Creep Shrinkage

o.
strength Ft28 comments from

CjQ
strength Fc28
reference

X
m
X
m

p
in
p

II
p

II
II
>sub <sub >sub <sub >sub <sub
For non structural

CTs
VO
W
m
&

O
£

P
low low high
applications
Bond strength

s
Tf
r*-
s

O
i-i
OD
important important
also important
..............

■
Able to decide

• •
Effect Low suitability based

VO
o

X
m
D
p

w
j- i

B
undetermined i (high strain cap.) on strengths

;
alone
‘this match of

'
properties is
No shrinkage

vO
2
N
oN
generally

V ...
(or ‘very slight’)
accepted in the

, '
profession’
'

Table 3.2 Continued on next page

Chapter 3 - Selection of materials for optimal performance of concrete repair G CD
G
H->
on G .2 G
'G
03 G
Io G
CD
G
G II
G
»h O
• pH
CD
c/3 G
CD
G CD
£
-fcj
cn
t;
o
o.
i> G (D G T3
CD
C/3
on O
Oh
<D CD h-> O
-f-* T3 w B on G. o u 2
O G
O
H +1 a (D CD
O Jh Oh o
£ o 03 O O h
PQ o G
Oh
W CD
S-H

CD
bn
G
I
a
J3
m

O h
<d
s
CD e ■§>
in a 2
o

03
O h
CD
S-H
G G G CD
O
C/3 0 G
V
c/3 ta G
CD
<d
4b G
Oh
"O
G
aCD
a> G CD
o G
0 , &
C/3
G- 1 H-J VO
cn
oS-H TPD <N B o
Oh 00
(N
O
B
4b
G o
o
(D
a
G
G
C/3 O
A Oh

G
G
u C/3
V
*CG P C J
G h
cd
■*-> 43 4b £ OO
h—> G o
on C/3
& G 3 ©
*G
OO
CD
*H
4b
<N .a
<N G s
C/3
A
Table 3.2 Continued on next page

4b
G
C/3
V
PH 4b
& 43
T3 CD G
OO O h
"0} C/3

<N G
B
o 4b
o
G
C/3
A

CD 2
O C/3 G G
G G G o
CD
J-H
G o O H-»
C/3

C+H
Q Q O
(D
o o LOh
Oh 2
Chapter 3 - Selection of materials for optimal performance of concrete repair
o on d
r*H
<-W <0 C on o tX) d d d
o a .2 o
*55 • 0^3 .2
t: o
o d
<
.2 $H "S
o
o *55
£ Jho On
*c/5
d
o
,d ja 03 >
d H dc3
<D ,(D o d
'f—» d *3 T3 O h
X ts
O h
o <o o <u d X
£
Vh
X O h o <D C/3 <D
PQ

<0
00
*
T3 T3
<U
03 CO
d
«C
*3
CD
'Iw
M Pi
00 CO
CD
(N
co

X
•o

w
At
O
CD
h U c .
•<u
cd e W
Vh
U & nd
<u
T3
d
V-i
(D
•
o3 B
Oh 2
CD o
'd o
1/3 <o
V *
C/3
<D
C/3

J3 W *8
1—
I
Id ’d <s
cd C/3
O h e'­
oS-H -a d d en
Oh 00 O m .2
<N 2 H
'd
1/3
o
A \ 53
*3
a>
O h
d C/3
CD C/3
V * "3
•^
d
Ph 3
<->
u ’S
H «a 3
0£)
'd
C/3 ■a 2
S' d 3
T3 <d
OO C/3 *2
O h
<N 'd <U
C/3 »H
A
*5h
'd
o
C/3
CD V d
>
•
C/3 PLh ’S '
<
& C/3
C O p£3 on
’d 5-
h
'd fB .2
C/3
oo O h *2
<N d 8 d
<o Oh o
,0 CO
d
ffi
C/3 on
A d
o

on
d
<o o
o ffi
d <N r—
to CO
S h
.O
C+H CO
(O
P4 60
C3

C Id
CO
H
 —j t i ^ u u i i ui lim icnais lor upum ai perform ance or concrete repair

Fc28 = 28 day com pressive cube strength

F t28 = 28 day tensile strength

E rep28 = 28 day Elastic M odulus (in compression)

Esub= Elastic M odulus o f substrate concrete

The Flong Kong Housing Authority have developed detailed specifications for classes o f

concrete repair mortars to be used in the repair o f structures77, which are given in

Table 3.3.

T a b le 3 .3 H o n g K o n g H o u sin g A u th o rity r ep a ir m a teria l sp e c ific a tio n 77

Characteristics for repair mortar

Characteristics Required values

Class 40 Class 25 Class 15

TM1 Range of com pressive strength at 28 30-60 20-40 10-30

3
days in N /m m

TM3 Minimum tensile strength at 7 days 2 .0 15 1.0

j
in N /m m

TM4 Range of m odulus of elasticity at 28 15-25 9-15 5-9

i
days in k N /m m

TM5 M inimum bond strength at 7 days 2.0 15 1.0

i
in N /m m

TM6 Cracking in Coutinho ring test at 7 0 0 0

days

TM7 Minimum Figg air permeability in 200 150 100

seconds

Recommendations in this field often tend to be based on researchers' intuitive

understanding o f the interaction o f repair materials with the substrate concrete and

opinions tend to be divergent. A general consensus identifiable from the research is that

when selecting a suitable repair material for reinforced concrete repair with the aim o f

combating excessive strains due to restrained volume changes, the following material

properties are important:

58
 ^ u u p i w . _» — u ^ i ^ v , i i u n u i m a i c i l a i b iu i u p u iim i periorm ance 01 concrete repair

• Strength (particularly tensile)

• Elastic Modulus

• Shrinkage

• Creep

Specification of materials for reinforced concrete repair is hampered by a paucity of

information on the optimum mechanical properties of repair materials required for a

78 • •
particular substrate . Faced with this difficulty, designers react by adopting materials that

appear to have properties close to those of the original concrete. In doing so they risk

selecting materials based on incorrect assumptions; materials which may fail through this

poor specification method.

3.2.2.2 Repair Materials

The materials for all types of reinforced concrete repair fall into four general categories60:

• Cement based

• Resin based

• Polymer-modified cement based

• Cement—pozzolanic materials

Since the 1960s a plethora of new, enhanced concrete repair materials and systems have

been introduced and have found increasing utilization61,74.

General aims of these enhancements are to improve tensile strength or reduce shrinkage in

the materials. The aim of reducing shrinkage is to limit the tensile strains caused by

restrained shrinkage.

Table 3.4 and Table 3.5 show the different categories of repair materials and their typical

properties respectively. Intuitively, combining the low shrinkage of a repair material with

properties that produce a high tensile strain capacity would produce a material less likely to

59
 iaici iciib ioi upiuimi periorm ance or concrete repair

fail. However, the low shrinkage materials are often prohibitively expensive. This,

combined with a lack of understanding of the benefits of reducing the tensile strains that

develop in repair materials, can lead to the selection of materials with the simplistic

requirement of high compressive strengths. The high strength seems intuitively acceptable

to engineers and such materials are often more affordable. Moreover, manufacturers’ data

on shrinkage properties of their materials often provides lower values than the material can

realistically achieve in practice .

Table 3.4 Categories of systems for concrete patch repair.74

Resinous Polym er m od ified C em en titiou s

m a te ria ls c em en tiliu u s m aterials m aterials

A: lip o x s m ortal | D. SD R m odified (i: OPC'-sand ntorlar

H P o !\e ste r m ortar P.: V inyl aceta te m od ified II: H A C m ortar

( ’: A eryiie m ortar | I ’: M agnesiu m p h o sp h a te m od ified I: I-’low inj! con crete

Table 3.5 Typical mechanical properties of repair materials74

Property Hesin m ortar Polym er modified cem entitious ntonar Plain cem entitious m ortar

Com pressive strength, M Pa 50-100 50-60 20-50

Tensile strength. M Pa 10-15 5-10 2-5
M odulus o f elasticity in
com pression. GPa 10-20 15-25 20-30
Coefficient of therm al
expansion (p er°C ) 25-50 x 10" 0-20 x tO" 10 X 1 0 "
W ater absorption
(9r bv m ass) 1-2 0.1-0 5 5-15
M axim um serv ice
tem perature (°C> 4 0 -SO 100-300 >300

Recent trends have led to the modification of cement based materials with polymers.

Polymer dispersions allow the formulation of repair materials that can provide a wide

range of property requirements: brittle to ductile, impermeable to porous, water shedding

etc.79. This is achieved through utilising the polymer’s ability to alter the mechanical

properties: elastic modulus, creep and shrinkage, bond strength, temperature and humidity

effects79. The objective of adding polymer fibres into the repair mix is to improve tensile
on
strength and to distribute and limit cracking .

60
 ^u u i/iv i ^ u w tv u u u xji m a ic u d is lu i upuiim i periorm ance or concrete repair

There appears to be some incongruity between the supposed desire to avoid property

mismatch, and the vast disparity between repair material and substrate properties caused by

the use of polymer modified materials74. Regardless of the general recommendation to

avoid property mis-match, manufacturers have continued to develop polymer modified

materials producing higher strength materials with lower shrinkage.

3.2.2.3 Compatibility, durability and property mismatch.

The current Highways Agency standard for reinforced concrete repair (BD 27/86) does not

take into account the mismatch in basic material properties such as elastic modulus,

shrinkage and creep55. To overcome this lack of standardisation, the greatest challenge

faced in the advancement of concrete repair material selection techniques is controlling the

relative dimensional behaviour of the repair material when compared to the substrate68.

This relative behaviour requirement is known as dimensional compatibility. Compatibility

in a repair system can be more fully defined as the balance of physical, chemical and

electrochemical properties and relative dimensions of the repair patch and the surrounding

substrate69. Researchers agree that ‘compatibility’ between substrate and repair material is

a key factor in deciding the performance of the repair. Hence the term ‘compatibility’ has

become a popular buzz word in the repair industry74.

Some researchers have attempted to show that the physical characteristics of the repair

material and substrate concrete should be as close as possible (Young’s Modulus,

coefficient of expansion, strength)60,61,74,75,80. This definition of compatibility is misleading

and suggests that in order for materials to be ‘compatible’, they must have similar

properties. This is not the case. Compatibility should mean that the relative values of

properties of the repair and substrate materials are ‘complimentary’ and only compatible in

61
 ^ ui m aiviiaio iui wpimiai pciiuxiiiaiice 01 concrete repair

as much as that they are at the ideal ratios to ensure optimum performance of the repair.

Other studies61 have recommended that although certain physical characteristics should be

similar, others should vary to aid durability.

Two such studies (Figure 3.3 and Figure 3.4), each regard dimensional compatibility as the

most complex state to achieve, it being reducible into four or five properties. Interestingly,

both regard shrinkage, creep and elastic modulus as crucial.

D urability c l C o n crete R epair

I
S e lec tio n c l P ro d u c tio n ol
C om patible M aterials D urable R e p a irs

I
T T T
C h e m ic a l Electrochem ical 1 Perm eability D im e n s io n a l
C o m p atib ility Com patibility 1 C om patibility C o m p a tib ility

i
I I I I 1
T t T T T
D rying T h erm al M o d u lu s cl G e o m e try
S h rin k a g e E x p a n sio n r E lasticity c l S e c tio n s

Figure 3.3 Compatibility and durability74

Compatibility
of Repair
Material* with
S u b s tra te

D imensions Electrochemical

D rying S h rin k a g e

Thtm al Expansion

Figure 3.4 Factors affecting dimensional compatibility69

62
 _/ ui maici iciis iui upum ai periorm ance or concrete repair

Other studies61 have also recognised that incompatibilities in the form o f differing elastic

moduli and different thermal movements between the repair and substrate concrete can

create difficulties and that creep o f the repair material may render a repair less effective

over time.

fid
Dector , in another o f her studies, recom mended a range o f properties to specify for a

repair material which would lead to adequate ‘com patibility’. His recom m endations are

shown in Table 3.6.

T a b le 3 .6 R eco m m en d e d v a lu e s fo r c o m p a tib ility 64

Characteristic Suggested requirement

Strength and Compatibility required.

movement Properties should be similar
properties to the substrate concrete,
especially with respect to
movement.
Bond strength Greater than 0.8 N/mm2
(Ref. 3)
Shrinkage As low as possible.
Proposed limits:
<500 microstrain at 28 days
(USA Ref. 10)*
or <300 microstrain at 7
days (HKHA Ref. 8)*
' see conditions relevant to
test
Permeability Less than Concrete Society
low permeability limits. In
areas of high risk, tower
limits may be imposed.

Table 3.6 recommends values for certain properties which it is claimed will ensure

compatibility. However, the majority o f researclr 6,55,61,62,63 suggests that such a

specification is unwise, as knowledge o f the properties o f the substrate concrete is essential

before a suitably com patible repair material can be selected. Emmons and V aysburd66

represent the factors necessary to achieve com patibility diagram m atically (Figure 3.5).

63
 ui nicuciicus iur upiurmi periorm ance or concrete repair

Crucially, they show that knowledge of the substrate properties and the exposure

conditions are required to achieve compatibility.

f fiarvlo*
I •nd
I Expoiura
VCondftloni

Loading
C om patibility
Condition*
of Repair of
Maierlelc with Compoth*
3urf«e«
S u b s t r a te S y a ta m
Application
Bond
Preparation Method
P ro p artU a
Of
Subatrata
Pf«ptrtl«» of Material*
Raptlr Pf<X*M
Maiorim

'
-------------------------- ► n Uptlr Sy*l«m Oodgn

Repair Syatam Product!or

D urab le
Repair System

Figure 3.5 Factors affecting the durability of concrete repair systems69

The research thesis by O’Flaherty58, reports by Mangat & O’Flaherty55’56,57’62’81 and a series

of reports by Emberson & Mays61 suggest that successful specification of repair materials

for reinforced concrete is more complex than is implied by the Highways Agency standard

BD 27/8666. These works suggest that the key properties of repair materials, which affect

long term performance, are:

• Elastic Modulus

• Shrinkage

• Creep

• Tensile Strength

The durability of concrete repair depends, to a large degree, on the appropriate choice and

application of repair materials68, a lack of durability may manifest itself in spalling,

cracking, scaling and subsequent loss of strength in the repaired concrete member73.

64
 ■^hujjiw ^ u i u w ic iia is iu i upum ai periorm ance or concrete repair

3.2.3 Properties of repair materials

3.2.3.1 Key properties

Mangat and O’Flaherty55 suggest that a key property which will determine the long term

performance of concrete repair is the Elastic Modulus.

The restraint to the free shrinkage of a repair material caused by the bond between

substrate concrete and repair material is the main factor which contributes towards the

likelihood of a repair material failing through cracking.

A survey of three bridges was performed by O’Flaherty58 by attaching vibrating wire strain

gauges to exposed reinforcing steel in a repair patch, and internally at the substrate

concrete interface and inside the repair material. These gauges measured the developing

strains throughout the drying process of an applied repair material.

O’Flaherty postulates that if (or when) a patch repair material becomes stiffer than the

repaired substrate concrete, some of the tension inducing shrinkage strains of the repair

material being restrained at the interface with the substrate concrete will be transferred into

the substrate, (Figure 3.8). This transfer of strain from the repair material into the substrate

was measured using the vibrating wire strain gauges. These measured values were related

to the ratios of elastic moduli of the substrate concrete and the repair materials. It was

shown that, the higher the ratio of the elastic modulus of the repair material to the elastic

modulus of the substrate concrete, the larger is the percentage of shrinkage strain that is

transferred from the repair material into the substrate concrete. This shrinkage transfer

alleviates the tension due to the restrained shrinkage and increases the possibility of a

successful repair.

65
 W A JU plW l u i n m icu ais iui upm nai periorm ance or concrete repair

These results (explained more fully in section 3.2.5) form the basis of a sound

understanding of the properties of a reinforced concrete repair material that are most

crucial in affecting the performance of the repair.

3.2.3.2 Elastic Modulus

Research has shown the pre-eminent influence of the relative elastic moduli of the repair

material and substrate concrete on the performance of repairs62. It has been identified

through field studies55,56,57,58’63 that as a repair material becomes stiffer than the substrate

concrete it is repairing, it can transfer its shrinkage strains into the substrate. It is also

known that the degree of strain transfer increases with increasing Em/Esub ratio up to 1.32

(where Em = Elastic modulus of repair material, ESUb = Elastic modulus of substrate

concrete). As a consequence restrained shrinkage tension in the repair material is

£D •
reduced . This phenomenon had also been identified implicitly in other studies, for

example, Emberson61 states, ‘the repair material with a low modulus caused an increase in

concrete stress, whereas a repair material with a high modulus resulted in a decrease.’

In addition to the tensile strain transfer benefits of a high elastic modulus repair material, it

has also been shown that materials with a high modulus tend to attract load away from the

substrate concrete61 in the long term. Such an interaction is essential if the intention of a

concrete repair is to restore structural capacity to a member.

Conversely, one study concluded that increased cracking is usually attributed to higher

modulus of elasticity (amongst other factors) and stated that it is generally agreed that the

potential for cracking for cement-based repair materials decreases with decreases in
• » 7 f\ •
modulus of elasticity . However, it found no significant correlation between modulus of

elasticity and field performance76.

66
 -> — u i n m ic u a is xur optim al periorm ance o t concrete repair

Other studies recommend moduli similar to that of the substrate74’75, one such study stated

that an ideal repair material would undergo neither shrinkage nor expansion and would

display a similar modulus of elasticity to the substrate concrete74. Such a specification

would probably perform satisfactorily in most cases as the lack of shrinkage would not

induce any tension in the repair material. In fact, in such a case the material should

theoretically perform satisfactorily even with a much lower or higher modulus of elasticity.

However, in reality even polymer modified materials with the best shrinkage compensation

will exhibit shrinkages of 200 to 300 microstrain at 28 days, and the majority of materials

will shrink much more than this. Other studies have found that they cannot make a definite

recommendation for limits of modulus of elasticity based on field results68.

Older research has generally not considered Elastic Modulus to be an important property of

a successful concrete repair material. Authors have recommended that the modulus of

repair materials should be lower than the substrate modulus or the same as the substrate

modulus. However, new research, based on verifiable field testing has proven that the

optimum modular ratio between the repair material and the substrate is higher than unity.

Therefore, it is shown that the modulus of elasticity of a patch repair material in relation to

the substrate concrete may have a significant influence on the distribution of stress within a

repaired reinforced concrete member.

When the elastic modulus of the repair material is greater than that of the substrate

concrete, the repair material carries more load than materials having modulus values equal
fO /M ^
to or less than that of the concrete ’ . This can be considered detrimental, because high

modulus repairs can cause localized areas of maximum principal stress adjacent to the

transverse interface that are greater than those in the concrete remote from the repair site61.

However, if an aim of the repair procedure is to restore structural capacity to an ailing

67
 j — o ^ic u iiu n u i m a ic n a is io r opum ai periorm ance o t concrete repair

concrete member then the transfer of external load into the repair patch will be beneficial

when a high modulus repair material is used.

3.2.3.2.1 Compressive and tensile elastic moduli

Tests to establish Elastic Modulus invariably produce compressive elastic modulus values.

However, the tensile stress strain relationship of a repair material, which is mobilised when

the free shrinkage of a repair material is restrained by the substrate concrete, is described

(the linear portion) by the tensile elastic modulus of the material. Although there is little

data available for the modulus of elasticity of concrete in tension, an assumption can be

made that the elastic modulus of concrete in tension is approximately the same as the

elastic modulus of concrete in compression82.

More precisely, when the compressive and tensile elastic moduli of concrete are measured

on identical specimens at 0.3 f CU28 (fcu28 = the compressive cube strength of a sample at 28

days after air curing) the elastic modulus in compression is 7.5% higher than the tensile
oi m
elastic modulus . When the compressive elastic modulus is measured at a stress equal to

0.3 ft28 (ft28 = The tensile strength of a sample at 28 days after air curing), then the tensile

elastic modulus has been shown to be 2.5% greater than the compressive elastic modulus83.

Therefore it is shown that for concrete at a very young age, the tensile elastic modulus may

be marginally higher than the compressive elastic modulus if measured on two identical

specimens. At approximately ten days after curing, the compressive elastic modulus and

tensile elastic modulus will be similar, thereafter the compressive elastic modulus will be

marginally larger. For the purposes of this research, the tensile elastic modulus of a

specimen will be assumed equal to its compressive elastic modulus.

For the purposes of the procedures developed in the thesis to determine suitable repair

material properties, it is assumed that because any repair material will be in tension, the

68
 ^nupivi j — ociccu u ii ui mciicx mis lor opum ai periorm ance or concrete repair

highest stresses it can accommodate will equal its tensile strength. This in turn will be

considerably lower than the compressive strength of the material and hence it is acceptable

to consider that tensile and compressive elastic moduli are the same.

3.2.3.3 Shrinkage

Shrinkage is caused by the withdrawal of water from the repair material through drying.

All cementituous repair materials shrink. Shrinkage is increasingly recognised as a major

factor in the long term durability of a repair77. The restraint provided to the repair material

by its bond to the existing concrete substrate is a major factor in increasing the complexity

of repair patches as compared to new construction68. As the substrate concrete restrains the

free shrinkage of a repair material, tensile strains are developed.

In addition to the hydration shrinkage of cement based repair materials, increasingly

popular resin based additives are known to shrink during polymerisation (the hardening

process of resin materials). Pure resins can typically shrink between 4% (epoxies) to 10%

(polyesters) during this process. However, resin materials are viscoelastic and these

(shrinkage) stresses will partly relax.

In addition to elastic modulus, shrinkage of a repair material has been identified as the

property which controls long term cracking at the repair/substrate interface57. It could

equally be stated that Elastic Moduli, shrinkage and creep combined are the primary

material properties which can be utilised to specify materials for concrete repair with

success; the actual singular cause of cracking and debonding of concrete repairs is

excessive shrinkage strains74. The ability of a material to cope with these strains depends

heavily upon its elastic modulus in relation to that of the substrate and its creep

characteristics.

69
 - OCIO.UUH u i materials ror opnm ai periorm ance o t concrete repair

Awareness of the importance of controlling drying shrinkage in repair patches has been

increasing77 as recent studies have identified its crucial importance 55>56>57’58>62’63 Results

from many studies have shown that unacceptable performance of repair materials is based

on high shrinkage61. Clearly, if the main cause of failure of concrete repairs is high

shrinkage, then a logical action to combat failure is to specify low (or nil) shrinkage.

However, all repair materials shrink, and even relatively low shrinkage repair materials, if

accompanied by low tensile strengths, will still fail. Research has attempted to specify

values for shrinkage, from the optimistic ‘no shrinkage’67 to recommended 28 day

shrinkage values of 400 microstrain for specimens exposed at 50% RH68. Some national

standards also attempt to limit shrinkage, for example, the Australian standard AS 1012

has a limit of 450 microstrain at 28 days77. Attempts to limit shrinkage in such ways fail to

take into account the interrelationship of properties which determine the overall

performance of a patch repair. For example, a repair material with an elastic modulus

higher than that of the substrate has the ability to transfer a proportion of its free shrinkage

to the restraining substrate, hence this combination of elastic modulus and shrinkage could

allow for higher shrinkage values. In addition (as discussed in the next section) whenever

tensile strains occur these will be relaxed to a certain extent by tensile creep, hence taking

creep into account could also allow for a higher amount of shrinkage to be specified.

There are two good reasons to develop a method that allows practitioners to select repair

materials that have relatively high shrinkage properties. Firstly, often repair materials that

have been specified as Tow shrinkage’ by manufacturers actually shrink much more in the

field than suggested by the manufacturers’ literature. Secondly, the vast majority of

available materials cannot achieve shrinkages as low as 330 or 400 microstrain and,

therefore, limiting shrinkage to such low values will put uneconomical restraints to repair

solutions available in practice.

70
 iaici iais iui upmimi periorm ance 01 concrete repair

3.2.3.4 Creep

Any strains which develop in a repair material as a result of restrained shrinkage will be

relaxed, to a certain degree, by the action of tensile creep. It has been correctly stated that

cracking at the repair/substrate interface is primarily controlled by the shrinkage and creep

characteristics of the repair materials57. The amount of tensile strain in a repair material is

dependent on the sum of the restrained shrinkage and the negative effect of the relaxation

through creep. Current research has shown that a more accurate statement would be that

Elastic Modulus, Shrinkage and Creep fully control the possibility for cracking to occur

(assuming satisfactory bond), since the effective restrained tensile strain in the repair

material is dependent upon the amount of free shrinkage transferred to the substrate

concrete through optimum modular ratio usage and the relaxation of the tensile strain

through creep. Excessive creep in the repair material may, however, render a repair less

effective over time, as it has been shown that creep reduces the effective Elastic Modulus

in the long term61.

Creep exhibits itself in two primary forms; as instantaneous elastic strain, and creep strain.

Instantaneous elastic strain is the creep that occurs immediately as the result of the applied

load onto a material. The creep strain is the relatively slow flow of the material with time

thereafter and is caused by movement of the water adsorbed onto the surface of hydrating

cement gel. Tensile loads are applied in gradual increments in a repair patch with steadily

increasing shrinkage. The elastic strain capacity of a repair material in tension is very

small, typically 200 microstrain84, and cracking is prevented if instantaneous elastic strains

remain below this value.

71
 upunicu jjc iiu m m iic e o i concrete repair

3.2.3.4.1 Tensile and compressive creep

Research has shown that creep in tension is a significant phenomenon, and can play an

important role in reducing stress due to restrained shrinkage84. If the tensile strains

developed in the repair material due to restrained shrinkage are relaxed by tensile creep,

then theoretically a higher initial shrinkage could be accommodated. This would be a

benefit, as lower shrinkage materials are generally more costly and less common.

Generally, materials are tested to assess their compressive creep properties, as testing for

tensile creep is more difficult85. It should be added that, currently, repair material

manufacturers in the UK do not provide even compressive creep data for their materials.

Although, from research literature, a great deal of information is available on the creep of

concrete in compression, experimental data on the tensile creep properties of concrete is

0/1 o c oo
scarce ’ . Brookes and Neville state that tensile and compressive creep can be

considered as similar in most conditions. However, during drying, tensile creep can be

higher than compressive creep. In the absence of clear information on the comparison of

tensile and compressive creep, it will be assumed that compressive creep is similar to the

tensile creep of a repair material in the work presented in this thesis.

3.2.3.5 Strength

A concrete repair that is intended to restore structural capacity to a member should be

designed to withstand the compressive stresses to which it may be subjected. However, it is

a fallacy that specifying a high compressive strength will ensure adequate performance of a

repair material. The research literature reported in the thesis has shown that key properties

which govern repair material performance are elastic modulus, shrinkage and creep. Other

research has shown directly that there is no significant correlation between compressive

72
 laici iciib iui upimmi periorm ance 01 concrete repair

• 7 f%
strength and dimensional stability . It is generally agreed that the potential for cracking of

cement based repair materials increases with high compressive strengths, despite

inherently higher tensile strengths76. Conversely, some practitioners have attempted to

recommend minimum values of compressive strength for structural application, without

regard for the effect of this on the durability of the repair68.

It should be noted however, that the tensile strength of a repair material and its elastic

modulus determine the tensile strain capacity of the material; in this respect the tensile

strength (which is related to compressive strength) is important.

3.2.4 Influence of material constituents on mechanical properties

Repair materials for reinforced concrete are generally cement based. It is common for

manufacturers to use additives to have desired effects on the mechanical properties of the

hardened repair material. For example, some additives increase strength and bond whilst

some reduce shrinkage. Different constituents will have varying effects on the important

mechanical properties of the repair material.

The scope of the current research does not encompass the specification of material

constituents. It is an explicit aim of this research to recommend the suitability of repair

materials for reinforced concrete repair based on the key mechanical properties which

determine their effective performances. Hence it is the mechanical properties of ‘off the

shelf repair materials that will be used to determine their performance in patch repairs by

developing a routine for a computer. A knowledge of the constituents of those repair

materials can provide an understanding of the material properties but will not aid in

73
 u i nicuci mis iui upu iim i perform ance 01 concrete repair

determining their performance in patch repairs. For this reason an in depth study of the

effect of constituent materials of repair materials is not attempted.

3.2.5 Testing to establish repair material properties

A variety of test methods are employed to determine the properties of repair materials. The

two most widely used standards are the British Standards and the ASTM standards (USA),

with many manufacturers using tests from both sets of standards to provide most

favourable data for their materials. Table 3.7 shows a variety of British Standard test

methods and the procedures employed therein, which was used in a research programme

on repair materials61.

74
Chapter 3 - Selection of materials for optimal performance of concrete repair
 u u w u u u ui m a iti iais iui upunm i penornm nce ot concrete repair

A variety o f international test methods also exist to establish most properties. Table 3.8

shows seven test methods for determining drying shrinkage. However, using different

sized specimens and different curing conditions will yield different final results. Hence a

manufacturer has the opportunity to legitimately select the test method that will produce

the lowest shrinkage value for any repair material. Any method attempting to predict the

development o f stresses in a repair patch will require an accurate (absolute) prediction o f

the am ount o f shrinkage that will occur in the repair material. Therefore, it is necessary to

specify a recommended test method, so that a standard datum for specim en size and curing

conditions can be set.

Table 3.8 Test m ethods to establish drying shrinkage64

. .. ^ ^ ----- ^ -*? - ^ -
Specification or Conditions Prism Limits
Standard Dimensions (mm)
Proposed 20°C, 65% RH 40 x 40 x 160 Not yet
Eurostandard established
Hong Kong 27°C, 55% RH 25 x 25 x 285 300 microstrain
Housing Authority -7 days
(HKHA)
Australia 23°C, 50%RH 75 X 75 x 285
AS1012 Pt.13 -
1970
USA 23°C, 50%RH 25 x 25 x 285 500 microstrain
ASTM C157 - -28 days*
1989
Germany Various - 20°C, 65%RH 40 x 40 X 160
DIN 52450- 1985 - 23°C, 50%RH
1
- 20°C, 45%RH
I - 20°C >95%RH
- 20°C, Wet i

UK This standard does not 75 X 75 x 150 -

BS1881 Pt. 5 - relate to drying shrinkage
, 1970 therefore test conditions 300
: are not included. I
Netherlands 7 days 20°C >95% RH | 40 x 40 x 160 12 x1 O'4
! CUR 21 | 21 days 20°C, 65%RH

76
 ^ uv . ^ uum ui maici lais iui upimmi periorm ance or concrete repair

3.3 Determining the key properties of repair materials

It has been established that the current standards for the specification o f reinforced

concrete repair materials do not take into account the m ism atch in basic material properties

such as elastic modulus, shrinkage and creep ". It is generally recognized that the restraint

provided by the substrate concrete (and the steel reinforcement) to the free shrinkage o f the

repair patch can cause tensile cracking. There are no recom mendations in current standards

pertaining to the optimal relationship between repair material and substrate concrete

properties.

Figure 3.6 demonstrates the strains which develop in a repair material that is restrained.

The restraint to the free shrinkage causes tensile stresses to develop.

D essication of Temperature External

the cem en t paste change loads

1
Shrinkage Thermal Mech anical
strain strain sti ain

r 1r
Restriction to the free strain
• due to end conditions
•due to a non-uniform strain over the section

r ------------------ -
Relaxation due the
C racking in the
v is c o u s b eh a v io r
o v erstressed locations
of the concrete
Vw
Induced elastic
tensile stress
Relaxation
Stress after
creep relief

T en sile strength
of C oncrete
D evelopem ent
of cracking

TIME

Figure 3.6 Schem atic illustration o f stress build up in repairs72

77
 — u^ituuun ui m aiciiais iui upum ai periorm ance 01 concrete repair

In Figure 3.6, the thick black line represents the stress in the repair material after relaxation

of the tensile stresses has occurred. The relaxation occurs through tensile creep. If the

tensile stress in the repair material exceeds its tensile strength, then cracking occurs. A

patch repair provided for aesthetic improvement is deemed to have failed due to this

cracking since assessment codes preclude the inclusion of any steel it encases being used in

assessment calculations. If the failed material was applied to reinstate the structural

capacity of a member, it will be unable to share any load and consequently has failed in

this purpose. It can be seen that avoiding the excessive development of tensile strain can

enable a repair material to perform adequately. Therefore, a technique will be developed

which utilises the proven phenomenon of shrinkage strain transfer through optimum

modular ratio specification. The shrinkage strain of the repair material can be partially

transferred to the substrate concrete with appropriate selection of relative Erm and ESUb

thereby reducing the risk of shrinkage cracking62 (E™ is the elastic modulus of the repair

material at time t days and ESUb is the elastic modulus of the repair material at time t days).

Data was obtained from a field study on the performance of reinforced concrete repair

materials; subsequent examination of this data demonstrated that an optimum modular

ratio of Em ^ 1.32ESUb will ensure a high level of free-shrinkage transfer from the repair

material to the substrate concrete55. The specification of suitable creep and shrinkage

characteristics will also ensure satisfactory long-term redistribution of service load from
69
the substrate to the repair patch .

78
 ^ u tittu u u vji m a id laid iui upmimi p en o rm a n ce 01 concrete repair

CO

Field tests were carried out to determine, at daily intervals, the strains developed in both

the substrate concrete and the repair material, directly following the application of a repair

patch. A summary of these measurements is shown in Table 3.9, for four different repair

materials.

Table 3.9 Strains developed in repair material and substrate:58

Material Local ion Strain at end of: micrastram*

Z one I Zone 2 Z one 3 Z one 4

(w eek 11) ( w eek 25} (w eek 47 ) (w eek 60 )

1.2 ‘subs' • 120 -1 2 0 •300 -3 0 0

‘steel' ‘emb •5 4 -5 4
L3 ‘sabs' • 107 -1 0 7 • 137 -rl.37
‘steel* ‘em b ■45 -4 5 - 108 • 108
1.4 'subs' -1 5 4 - 154 -2 9 7 •2 0 7
'steel' ‘em b -4 2 •4 2 -1 4 2 • 142
Cit 'subs' —02 • 02 —183 • 183
"steel' 'em b -0 0 -4 4

N egative \ aloes indicate tensile strains.

In Table 3.9 ‘subs’ represents strain gauges located at the interface of the substrate

concrete and the repair patch; ‘steel’ and ‘emb’ represent strain gauges attached to the steel

reinforcement and embedded in the repair material respectively.

Strains that developed in the substrate concrete after application of the repair material were

compared to the free shrinkage properties of the repair material (Table 3.10). In this way it

was possible to establish the percentage of the free shrinkage strain of the repair material

which was transferred into the substrate concrete.

Table 3.10 Percentage of free shrinkage transferred into substrate concrete58

Repair III ( /W /-u :-) l 'mm thri: f-slinlrcui. /.:

material m icrosirain microstrain %

1.2 1 -2 7 120 136 88

1.3 1-15 107 210 51
1.4 1 -2 2 154 238 65
01 1 -1 0 92 329 28

Figure 3.7 shows that there is a clear relationship between the modular ratio and the

amount of shrinkage strain transferred from the repair material to the substrate concrete. It

79
 —o n c v u u n ui mcuci mis mi upurnai perrorm ance or concrete repair

can be seen that at a modular ratio o f Erm = 1.32Esl,b, all the shrinkage strain o f the repair

material is transferred to the substrate at the interface o f the repair patch.

1 -40 •

1 35
1.32
1 -30 ■

( 125 Material L2
Material L4
u
I 1 '20
ro
g 1-15
c Material L3
M a te ria l G 1
1 10 m - 0 0032;. + 1
(R 2- 0-968)
105

100
10 20 30 40 50 60 70 80 90 100
A (free shrinkage transferred to substrate concrete): °o

F ig u re 3 .7 R ela tio n sh ip b etw een m o d u la r ratio and free sh rin k a g e tr a n sfe r 5

The graph establishes the relationship:

m = 0.0032A + 1

where m = the modular ratio

X = percentage o f free shrinkage strain transferred from repair to substrate concrete

at the interface.

This relationship can be used to determine the am ount o f free shrinkage which will be

transferred from any repair material into any substrate. It will be used to develop a model

for the prediction o f in-situ performance o f concrete repairs.

The process o f transfer o f tensile strain from a repair material to the substrate is explained

clearly w ith the aid o f Figure 3.8, Figure 3.9 and Figure 3.10. A ccom panying these figures

is a key that indicates the magnitude o f stresses through colour changes. Figure 3.8

represents a repair material freshly applied to substrate concrete. The green equidistant

lines represent the concrete over which they lay, similarly the red equidistant lines

80
 uw ivm un wi m aiciiaia iui upunim periorm ance or concrere repair

represent the repair material. In Figure 3.9 the elastic modulus of the repair material is less

than that of the substrate concrete. As the free shrinkage of the repair material occurs, it is

restrained by the stiffer substrate, this is seen in the figure after the repair has been in place

for 28 days. The effect of the restrained shrinkage is shown exaggerated. The repair

material away from the restraint is allowed to contract freely but the repair material

adjacent to the substrate is severely restrained. This restraint causes tensile strains (virtual),

if these tensile strains exceed the tensile strain capacity of the repair material, it will fail

(crack). The tensile strain (virtual) in the repair patch reduces as distance from the restraint

(substrate) increases. At a certain distance away from the substrate, the effect of the

restraint has no influence, and the material exhibits its natural tensionless free shrinkage

strain. The distance over which the tensile strains caused by the restraint to shrinkage exert

an influence on the repair material is known as the ‘zone of influence’. Figure 3.10

demonstrates a repair situation where the elastic modulus of the repair material is higher

than that of the substrate concrete. As the repair material shrinks, some of the shrinkage

strain at the repair / substrate interface is transferred into the substrate by the stiffer repair

material. The effect of this strain transfer is shown exaggerated. Instead of (as in Figure

3.9) the repair material having to withstand the whole tensile (shrinkage) strains, some

strain is transferred to the substrate. The sharing of shrinkage strain leads to lower tension

in the repair material at the substrate and compression in the substrate. The region of the

substrate concrete in Figure 3.10 that is affected by the transfer of shrinkage strain from the

stiffer (E rm > E SUb ) repair material is also known as the ‘zone of influence’. Strain

transferred from the repair material will cause compressive stresses in the substrate which

will be at their highest at the interface, and will gradually reduce as distance from the

interface increases until finally strain transfer from the repair material has no effect on the

substrate. The depth of the zone of influence is of little importance to this study, as it is the

81
 ./ uviwwuun u i m aiciiaib lux u p m iiai periorm ance or concrete repair

critical tensile strain (virtual), occurring at the restraint interface, that the repair material

will be designed to withstand.

This phenomenon of strain transfer can be utilised to design successful repair patches and

provide optimum selection of repair materials.

82
 ^..upiw vji incucnais iui upum ai periorm ance 01 concrete repair

substrate repair

Figure 3.8 Shrinkage: Substrate and repair m aterial interaction, t = 0 (on application)
high tensile strain
(repair) /
Ihigher
com pressive strain
(substrate)

11 no tensile
strain (repair)
/ no
Figure 3.9 Shrinkage: Substrate and repair m aterial interaction, t = 28 days. Erep < Esub com pressive
strain
(substrate)

Figure 3.10 Shrinkage: Substrate and repair m aterial interaction, t = 28 days. Erep = 1.1 Esub

83
 laicnaia iui upmimi periorm ance 01 concrete repair

3.4 Development of a method to predict the performance of repair

materials in-situ

The current standard for repair material specifications, BD 27/8666 does not give adequate

importance to the necessary marriage of properties between substrate concrete and repair

material. Recent research recommends that the key properties for consideration when

selecting a repair material are the respective elastic moduli, creep and shrinkage strains of

the repair material and substrate concrete. The shrinkage inherent in all repair materials,

restrained by the substrate, will attempt to transfer itself to the substrate concrete at the

interface. If the stiffness of the substrate is greater than that of the repair material, this

transfer cannot take place and the shrinkage may exhibit itself as tensile cracking of the

repair material. If the stiffness of the repair material is greater than that of the existing

substrate concrete some of the shrinkage may be transferred. An additional factor for

consideration is creep. Generally, when a patch repair is applied, the substrate concrete in

service has already undergone most of the total creep it will endure in its lifetime. Clearly

this is not the case for the repair material and any creep occurring would reduce the effect

of restrained shrinkage. An added complication, however, is the fact that creep affects

stiffness. High creep can effectively reduce the stiffness of the repair material.
co
Mangat and O’Flaherty suggest an optimum modular ratio (the ratio of elastic modulus of

repair material to substrate concrete) ranging between 1.2 and 1.4 depending on the values

of the other key characteristics. These values are based on field data which included the

cumulative effect of creep and shrinkage of a range of commercial repair materials.

Knowledge of the properties of both the substrate and the repair material in a concrete

repair situation enables the design of a method by which the in-situ performance of a repair

84
 ^nu^i — obicuuun o i m aicn ais iur opum ai periorm ance or concrete repair

can be determined. A software tool is developed in the thesis, in which a database of repair

materials is queried by a software routine in order to find all the repair materials which

would be successful in a certain repair situation. The properties required to optimise the

selection of repair materials are given in Table 3.11. Throughout, a satisfactory bond

between the repair and substrate is assumed.

Table 3.11 Key properties for the optimisation of repair material selection.

At 28 days Substrate Repair

Compressive strength N /m m 2 ✓ ✓
Tensile Strength N/mm2 - ✓
Shrinkage microstrain - ✓
Creep strain microstrain - ✓
Stress/strength ratio - - ✓
Elastic Modulus GPa

3.4.1 Determination of properties

The procedures developed in this thesis will standardise the material properties used in

design to a common datum representing different test methods. The two test standards that

this procedure adopts and accommodates are the British Standard and the ASTM tests for

materials which are widely accepted in the UK. European Standards may be

accommodated in the future. Different test specifications recommended by these standards

(BS and ASTM) can yield varying values for some properties.

85
 incuci ictis lux upim mi periorm ance or concrete repair

3.4.1.1 Compressive strength and Elastic Modulus

In order to determine the compressive strength of insitu concrete, a core is taken in

accordance with ASTM C42 - 9086. The diameter and height of cores are measured and

after conducting the specified tests, correction factors are applied to relate the compressive

strength to a datum diameter/height ratio.

0 7 ^

BS 1881-116:1983 is the recommended British standard for determining the compressive

strength of a repair material. 100mm or 150mm cubes are subjected to an increasing load at

a rate of between 0.2 N/(mm .s) and 0.4 N/(mm .s). The maximum load is divided by the

cross sectional area of the cube and the resulting compressive strength, fcu, is expressed to

the nearest 0.5 N/mm2. ASTM C 39-9488 is the equivalent standard from the USA. This

test is conducted on concrete cylinders. The cylindrical samples are loaded to failure at a

rate of between 0.14 N/(mm2.s) and 0.34 N/(mm2.s). A length/diameter correction factor

for cylinders is applied as part of the test method to relate the strength to a datum

height/diameter ratio.

The conversion relationship for ASTM compressive strength to the BS compressive

strength is given b y 82’87’88

Where fcy = cylinder strength of concrete specimen

fcu = cube strength of concrete specimen

86
 — u ^ iv w iu u u i m aici iaia iu i upmimi periorm ance 01 concrete repair

OQ

BS 1881-121 1983 is the recommended standard for determining the Elastic Modulus of

a repair material or substrate concrete. ASTM C 496-9490 and ASTM C 580-9391 are

acceptable equivalent tests which require no modification to relate their output values with

the British Standard.

Occasionally, a supplier will not provide a value for the elastic modulus of the repair

material and it is not practical to demand this information from suppliers. Conversions,

therefore, are needed to be performed to estimate the elastic modulus based on other basic

inputs (e.g. strength). The following expression can be used for this purpose82:

E cit = 4-73 * / ' c28° 5 Eq. 3-2

Where f C28 = 28 day compressive strength of standard test cylinders in MPa

E C28 = 28 day Elastic Modulus in GPa

It should be noted that Eq. 3-2 utilises the cylinder strength of a core to determine the

Elastic Modulus, the equation is strictly valid for concrete but has been assumed for repair

materials. This cylinder strength should be corrected to allow for length/diameter ratio

before being used in the equation. For the purposes of the rest of the procedure described

below this value for cylinder strength requires a conversion to cube strength (Eq. 3-1)

3.4.1.2 Tensile Strength

Any of the following standards are acceptable for the determination of flexural strength (or

modulus of rupture) of repair materials:

C 560-93 Standard test method for flexural strength and modulus of elasticity of chemical-

resistant mortars, grouts, monolithic surfacings and polymer concrete91; C 78-94 Standard

87
 j - j w c m u ii ui m aterials ror optim al periorm ance o t concrete repair

test method for flexural strength of concrete using simple beam with third point loading92;

C 293-94 Standard test method for flexural strength of concrete using simple beam with

centre point loading93; BS 1881-118: 1983 Method for determination of flexural strength,

(third point loading)94. All these methods use conversion factors to return corrected values

of tensile strength thus negating any differences in the test results which may be caused by

the different test methods themselves.

3.4.1.3 Shrinkage and creep

The surface to volume ratio of the insitu repair to be undertaken is required, as is the

surface to volume ratio of the specimen of repair material which will be used to establish

the free shrinkage of the material at 28 days.

Although Table 3.8 shows many international standards for the determination of shrinkage,

few of these are accepted in general practice in the UK. The standards readily accepted are:

• ASTM C 531 - 9595

• ASTM C 157 - 93%

The standard method for conducting creep tests on concrete in the United States is

• ASTM C 512

This method is primarily for conventional concretes, though is can be adapted for use with

cementituous based repair materials by reducing the size of the cylindrical specimen from

150 x 300mm to 100 x 200mm97. Creep testing should be conducted under similar

environmental conditions to Shrinkage testing. In order to fully define the creep properties,

the stress/strength ratio under which the testing was performed should be given.

88
 —kjcicuiiun u i mciiciniib lur upum ai periorm ance or concrete repair

3.4.2 Development of repair material properties

The performance of a repair with time is governed by the development of its material

properties with time. A procedure is outlined for establishing the relationship of a range of

properties with time. This procedure involves properties selected in Table 3.11, which

were identified as crucial to the overall performance of a concrete repair in section 3.2. A

manufacturer typically provides limited information about a repair material, often giving

the 28 day values for a number of properties. Using these values solely, it would not be

possible to design a repair patch for the worst case scenario which could apply to the

critical combination of properties at an unknown age. Hence it is necessary to be able to

establish the properties of the repair material at any age (i.e. define property-time

relationships).

The repair material manufacturers may provide limited data, typically giving the elastic

modulus after 28 days of curing and similar data for compressive and tensile strength and

also shrinkage. Creep data is rarely provided. In order to use this limited information to

predict the early-age and long-term performance of the repair material, it is necessary to

extrapolate this basic data to provide the value of key properties at any age. To achieve

this, generic property versus age relationships are established.

The approach used is to examine the development with time of these key properties in

generic repair materials and to relate the value of a property at any time t with the 28 day

value. The resulting ratio of a property at time t to the value at 28 days provides a

relatively accurate relationship that is true for a wide variety of repair materials. The aim of

this chapter is to verify that such unique relationships can be achieved for the key

properties (e.g. Elastic Modulus, shrinkage, creep) for a variety of generic repair materials.

The relationships can hence be utilised in an algorithm to predict the magnitude of tensile
89
 _> — j t ic t u u n u i iim ici mis lor opum ai periorm ance or concrete repair

strain in a repair material at any time and compare this with the tensile strain capacity of

the material to ascertain the likelihood of cracking of a repair patch.

ro
O’Flaherty determined experimentally various properties of a number of repair materials

with age. The repair materials were tested in the laboratory and the key properties (Elastic

Modulus, shrinkage, creep and strength) were measured at set intervals. All information of

thirteen materials was collated for the derivations reported in this thesis. The original

numbering system of the materials has been maintained, where G (Gunthorpe), L (Lawns

Lane) and S (Sutherland Street) are the initials of the bridges on which the materials were

used in patch repairs. The manufacturers’ data sheets provided the 28 day values of some

key properties of these materials which are given in Table 3.12.

Table 3.12 The 28 day strength, elastic modulus and tensile strength of the thirteen generic repair

materials.

Material Constituents fcu Erep ft

(N/mm2) (kN/mm2) (N/mm2)

G1 Polymer modified; limestone 60 31.1 4.2
aggregate; dust suppressant; RH
Portland cement; 5mm aggregate
Microsilica and copolymer
G2 RH Portland cement 56.6 17.6 2.5
Microsilica; Fibres; Chloride free
admixtures; Spray dried styrene
acrylic copolymer
G4 Styrene acrylic copolymer; 50 24 6..2
Admixtures; Portland cement;
Fibres; 6mm aggregate
G5 Spray dried styrene acrylic 50 19.6 4.2
copolymer;
Portland cement; Sulphoaluminate

90
 mia lux u p m im i periorm ance or concrete repair

Material Constituents fcu Erep ft

(N/mm2) (kN/mm2) (N/mm2)

cement; Microsilica; Fibres and other
pozzolanic material
G6 Microsilica; Styrene acrylic 30 11.5 2.9
copolymer
LI Microsilica; Limestone aggregates 60 22.7
Admixtures; 3mm aggregates
L3 Shrinkage compensated Portland 28 27.4 3.9
cement; Graded aggregates; Special
fillers; Chemical additives
L4 Portland cement; Silica sand 40 29.1 2.8
Admixtures including plastic fibres
L5 Fibres 80 29.1 5.3
SI Cementituous material; 5mm graded 79 24.2 Not
aggregate; 500 kg/m3 cement content provided
u
S2 Shrinkage compensated 60 32.2
u
S3 Microsilica; Shrinkage compensated 70 31.9
styrene- acrylic copolymer
S4 10mm rounded aggregate; PFA 39 27.4
Superplasticiser; Polypropylene fibres
Where fcu= compressive strength, 28 days age

Erep = elastic modulus, 28 days age

ft = tensile strength, 28 days age

3.4.2.1 Creep-age relationship

The compressive creep data (creep versus time) at a stress/strength ratio of 30% are given

in Table 3.13, for the thirteen repair materials. The creep data excluded the instantaneous

elastic strain that occurs upon load application.

91
 U
co
o
o

JSS
E
<D
E
a

ra
<4—»

<4-1
13
O h
O h

O h
O h

<D

2
H
o

O
V

a
o
"O

>
fi
‘C

C3
o>

E
S-r
T5

u
S
S

3
otn

Vl
’E

C
o.

h
TS3

fO
O h
fcX!

fO
V
63
WD

*G
’■ Q
f-<
rG

Q
G
O
03

CD

&
3 ,

*
O
<N
t" -

10
<N

00
09

r—H
in

Td-
OJ
Mat. 20 28 30 40 50 70

0
0
0
0

in
1—H
O n
100 170 180 200 250 270 273 294 300 305 310 310 310

0
0

O
in
06

T“H
10
069

m
G2 190 220 240 310 314 342 350 1 440 590 740

0
G4 220 320 350 365 380 460 480 490 497 546 560 620 710 760 810
l

0
0
r*H
00
00
G5 140 620 740 785 830 930 1020 1040 1046 1100 1230 1280 1340 1360

0
0
0

90 006 066
00

490 680 750 795 840 1030 1035 1070 1090 1130 1160 1190

hJ
009 069

r —■<
220 400 450 470 490 620 640 645 680 710 760 770 770

0
009
L3 130 190 210 225 240 320 410 500 510 580 700 720 720 720

O
m
L4 230 250 265 280 330 390 420 421 428 430 460 520 520 520

O
0

O
O
in
r*H
00
L5 210 240 280 340 370 | 378 434 450 520 520 520 520

0
0
0
0

(N

O)
m

1 80 230 255 280 300 370 400 401 408 410 420 430 430

0
0
0

fH
O
m
09

r*H
T"M
S2 80 130 210 240 242 256 260 330 330 340 340

O
CO
009 009
S3 320 390 445 500 510 520 530 537 586 610 610 610

0
S4 100 190 220 245 270 280 290 320 326 368 380 400 420 440 440

*
u

U
o
o
o

a
£

O
(N

_ o
’C

■ §

G
§

(D
#g

<u
a>

<u
<C

o
<u

<3

o
t-H

C/2
o
cd

G
-*->
03
00
G

c/2
&

fa
&
m

c/2
TJ

T3
*03
§•

tfH

h
oo
00

<N
ON
 uwiuvuuu *ji m aiciiaia iui upimmi periorm ance 01 concrete repair

The value of creep at each age can be expressed as a proportion of the 28 day creep value

of the material (C/C28 ). For example, considering the data for material G1 (Table 3.13) and

dividing throughout by the 28 day creep value, gives the proportions listed in

Table 3.14.

Table 3.14 Creep (C/C28) in material G1 as a ratio of the 28 day creep (C28).

Days under load 0 2 4 6 7 8 12 14 15 20 21

Ratio of creep (C/C28) 0.00 0.34 0.51 0.58 0.61 0.65 0.68 0.79 0.85 0.92 0.93

Days under load 28 30 40 50 60 70

Ratio o f creep (C/C28 ) 1.00 1.02 1.04 1.05 1.05 1.05

This procedure was completed for each of the thirteen repair materials listed in Table 3.13.

The creep ratios (C/C28 ) against age under load are plotted for all the thirteen materials in

Figure 3.11. An average relationship (best-fit line) of creep ratio with age under load of all

thirteen repair materials is also plotted.

93
Chapter 3 - Selection of materials for optimal performance of concrete repair

SIB

CO

cs
a.
<
•—
u t
O'
c
<
<uu
I4 '< i

tQ.
2

7<3T5
o
<u
■ca
3
V
E

X: m
C ■ Wr

cs <I +
O
U
« 5s I ! +
:» ' h +
k ■ nr t-
x*": n:
X 41 X I
U

(8Z D I D ) sAep 82 IB daajo oj daajo jo opey

3
OD
 v ia jv iu —o w c tu u u u i xnaiciiais iur upum ai periorm ance or concrete repair

The ability of the best fit line to represent any repair material can be indicated by its

correlation coefficient. A correlation coefficient indicates the directness of the relationship

between two sets of data x and y. The correlation coefficients of each repair material with

the average (best-fit) curve of all thirteen materials are shown in Table 3.15. The table

shows high coefficients of correlation exceeding 0.9 for all materials, thereby justifying the

assumption that the average creep curve represents each material with a reasonable degree

of accuracy.

All materials, except material G2 at later age, show a similar relationship of creep ratio

with age throughout the 70 day period plotted in Figure 3.11.

Table 3.15 Correlation coefficient of average creep ratio curve with creep of each material

Material Coefficient of Material Coefficient of

Correlation (R) Correlation (R)

G1 0.983 L3 0.983

G2 0.907 L4 0.907

G4 0.984 L5 0.984

G5 0.981 SI 0.981

G6 0.960 S2 0.960

LI 0.979 S3 0.979

S4 0.972

Material G2 shows significantly greater creep ratios than represented by the average curve

at ages beyond 30 days of loading. The impact of this long-term underestimation of creep

in material G2 by the average relationship for the materials will be explained later in this

chapter. However, intuitively, it can be recognised that the best fit line would predict a

95
 ui lucucnais iur opum ai periorm ance o t concrete repair

conservative amount of creep for material G2. In practice, the extra creep which would

occur for this material over that predicted by the average relationship, would provide

greater relaxation of restrained shrinkage stress, and hence is less worrying than a material

developing less creep than that predicted by the best fit line.

The creep versus time under load data represented by the average (best-fit) relationship

plotted in Figure 3.11 is listed in Table 3.16.

Table 3.16 Best fit relationship data of C/C28 with time under load.

Days underload 0 2 4 6 7 8 12 14 15 20 21
Ratio of creep (C/C28) 0.00 0.26 0.49 0.56 0.61 0.66 0.75 0.81 0.85 0.92 0.93

Days under load 28 30 40 50 60 70

Ratio of creep (C/C28 ) 1.00 1.02 1.12 1.20 1.25 1.27

3.4.2.2 Hyperbolic expression for creep/time relationship

Ross98 and Lorman" recommend the use of a hyperbolic expression to describe the

relationship between creep and time under load; which is expressed as follows:

t
c = ---- — Eq. 3-3
a + bt

Where t = time

c = creep

a and b are empirical constants which will be determined from the

experimental results of the 13 materials used in this study.

Rearranging Eq. 3-3 and multiplying each side by C28 gives:

96
 vyi.u^wi vji nuu>wi iaia iui upuumi pci luiiiituice 01 concrete repair

^ C28 (a + bt)C 28

• • — = C18ci + C ->8bt
_ Eq. 3-4
where C r = -----
C 28
t ...
— = a+ b t
cr
which is the equation o f a straight line with a' and b' as constants. Therefore, plotting t/Cr

against t produces a line whose slope represents b \ and the intercept represents a \ The

data listed in Table 3.16 for the average C/C 28 versus time relationship are plotted

according to equation 3.4 in Figure 3.12.

y = 0 . 7 0 1 x + 7.0162
50 -l

40 -

30 -

&
\ 20 -

.1

0 10 20 30 40 50 60 70 80

Time (days)

Figure 3.12 t/C r versus t relationship for thirteen repair m aterials (R 2 = 0.9964).

Hence the equation can be written to describe the developm ent o f creep ratio with time, as:

97
 ^ n ap ici j - ocicwuuu ui u m ien ais lor opum ai periorm ance o t concrete repair

Cr = -------------------- Eq. 3-5

7.0162+ 0.70 If

where,

^ _ creep at any age under load _ C

creep at 28 days under loading C

The hyperbolic relationship determined in Figure 3.12 and represented by equation 3.5 is

plotted in Figure 3.13 for an extrapolated long-term period of 400 days. The corresponding

experimental data of the thirteen repair materials is also plotted up to 70 days under load.

98
 Chapter 3 - Selection of materials for optimal performance of concrete repair

0 3 / 3 = J 3 ) S^ P 82 W d sajo o\ \ auiij ve cteaio j o o i \ ^

 _> — u^ns^iiuii ui lucucucuij ior opum ai periorm ance 01 concrete repair

The correlation coefficient between the experimental curve (Figure 3.13), representing the

average behaviour of the thirteen materials and the curve based on the predicted values

(Eq. 3-5) is R = 0.9944. This close correlation between experimental and predicted value

means the correlation coefficients for individual materials between the experimental and

predicated creep ratio would be similar to those in Table 3.15.

The expression adequately describes the performance of the average repair material for the

purposes of predictive models developed for the design of patch repairs. It allows

extrapolation to ages which enables long term performance to be predicted.

3.4.2.3 Shrinkage-time relationship

The shrinkage data of the thirteen repair materials (shrinkage versus time relationship) are

given in Table 3.17.

100
 Days after casting
09

o
in
r-H

r--
r~H

00
cn
20 28 30 40 70

<N
50

SO
CO

fH
T

O
i—
o
—

v
o
o
o
H
"3-
20 30 50 55 80

H
93 142 150 200 240 270

H■
10

1—

O
O

o
r-H
o o

H
r-H
m
r"H

o
o
m
o

Os
60 80 100

o (N
200 260 266 308 320 440 560

o
09 06 <n

o
o
o
r-H
r-H

o
o
o

r-H
r-H
o

m
so
r 1H

G4 40 65 70 80
(N
87 r 104 160
90 1

o
r-H
in
o

100 160 210 225 240 290 297 300 360 368 424 440 470 510
99 06
rH

o
in
o

150 200 215 230 280 287 290 350 357 406 420 460 540
09

rH

r—
o
r-H
o
o
o

H
so
20 70 85 153 200 205 240 250 300 320 340

f1H
09

rH

o
in
o
o

so
L3 120 150 220 253 270 320 323 344 350 400 420 460

o
rH
rH

o
O

r-H
o
r-H
in
r-H

H
o

so

L4 40 50 70 75 80 100 107 220 240 260

SO

0L rH
rH
rH

o
o
o
o

L5 20 40 60 65 80 93 156 | 160 | 220 240 260

<N <N

09
l—

o
r-H

xn
r-H

m
m
hH

so
so

H
Os

30 70 85 100 120 140 200 230 260 280

<N

009

o
S2 100 340 350 375 400 430 450 460 500 505 540 550 620 620
\

09
rH

o
o
in
r-H
r-H

o
o

m
so
Os

oo

S3 50 80 95 140 182 200 210 240 260

SO4
r -
oo
rH

o
o
r-H
r-H

o
r-H
o

so

S4 20 50 70 85 130 143 150 170 200 240 260

OS

<N
 IV! J U1 laicnaib iui upm im i periorm ance 01 concrete repair

The shrinkage at each age can be expressed as a proportion of the 28 day shrinkage value

of the material (S/S28)- For example, considering the data for material G1 (Table 3.17) and

dividing throughout by the 28 day shrinkage value, gives the proportions listed in Table

3.18.

Table 3.18 Shrinkage in material G1 as a ratio of the 28 day shrinkage

Days after casting 0 2 4 6 7 8 12 14 15 20 21

Ratio of shrinkage (S/S28 ) 0 0.14 0.21 0.35 0.39 0.42 0.56 0.66 0.70 0.77 0.80

Days after casting 28 30 40 50 60 70

Ratio of shrinkage (S/S28) 1.00 1.06 1.41 1.69 1.90 1.97

This procedure was completed for each of the thirteen repair materials using the data for

each repair material in Table 3.17. The shrinkage ratios (S/S28) against age after casting are

plotted for all the thirteen materials in Figure 3.14. An average relationship (best-fit line)

of shrinkage ratio with age after casting of all thirteen repair materials is also plotted.

102
Chapter 3 - Selection of materials for optimal performance of concrete repair

' ' = ■T1-. V' -:\ ' -v ' 7 C7c77' !' v7. k;

£
:hx x

o
vO

sox
s
3CJ
S— co<->
j
os
o» +
£
H

w
0O1X)
a
-t

O
0>>3
T3
X
00
<N
O
Ot
OX)
03
! ' o
’>cjm ■ oxi
to
- j:
H'm o s

-i ♦
o
_o
+-■
03
as,
sXep 8Z 3SB5fUuqs 0 } a S e q u u q s j o o i j e y ^ z s / s = JS)

<u
3
OX)
 ^ - u w v u iu n u i iiicuciiuis io r opum ai perrormance o t concrete repair

Table 3.19 lists the coefficient of correlation of the shrinkage data for each material in

Figure 3.14 with the average curve of S/S28 versus time relationship of the thirteen

materials. The very high coefficients of correlation (>0.927) confirm the validity of the

data to the average curve.

Table 3.19 Correlation coefficients for the shrinkage ratio versus time curves of each material with the

average relationship.

Material Coefficient of Material Coefficient of

Correlation (R) Correlation (R)

G1 0.980 L3 0.994

G2 0.982 L4 0.992

G4 0.985 L5 0.991

G5 0.985 SI 0.999

G6 0.995 S2 0.927

LI 0.996 S3 0.989

S4 0.991

It can be seen in Figure 3.14 that the standard deviations of individual materials from the

average curve are generally higher at later times than was witnessed for creep (Figure

3.11). Therefore, some materials may shrink more than the average value derived from the

shrinkage ratio versus time relationship. It is likely that these materials would also creep

more, thus offsetting the discrepancy which would result between predicted and field

behaviour - this is discussed further in Chapter 4. One possible design approach is to

assume a higher growth of shrinkage ratio with time than the average determined for the

thirteen repair materials in Figure 3.14. However, this would be unduly conservative; in the

104
 v-'iiajjit.i j —ocicwuuii ui materials ior opnm ai perrormance o t concrete repair

interests of accuracy, the average relationship is accepted as adequately representing the

shrinkage behaviour of each of the thirteen materials.

Taking the data for the development of shrinkage for the thirteen repair materials from

Figure 3.14, the average value of shrinkage at any time as a ratio of the material’s 28 day

shrinkage can be derived from the best fit curve (Table 3.20).

Table 3.20 Best fit relationship data of S/S28 with time after casting.

Days after casting 0 2 4 6 7 8 12 14 15 20 21

Ratio of shrinkage (S/S28) 0.00 0.16 0.32 0.42 0.47 0.52 0.64 0.70 0.73 0.86 0.88

Days after casting 28 30 40 50 60 70

Ratio of shrinkage (S/S28) 1.00 1.03 1.22 1.36 1.45 1.50

3.4.2.4 Hyperbolic expression for shrinkage/time relationship

Ross98 and Lorman" also recommend the hyperbolic expression to describe the

relationship between shrinkage and time; which is expressed as follows:

S =-------------------------------------- Eq.3-6
a + bt

Where t = time

S = shrinkage

a and b are empirical constants which will be determined from the experimental

results of the 13 materials used in this study.

Rearranging Eq. 3-6 and multiplying throughout by S28 gives:

105
 j —ocic^uun ui nmici mis lor opum ai periorm ance o t concrete repair

g ^28 — + bt} S28

— = Sn a + S2Jbt Eq. 3-7

Sr

— = a'+ b't
Sr

Plotting time over shrinkage against time produces a line whose slope is b \ and the

intercept is a \ This is done for the average value for the thirteen repair materials (Figure

3.15).

50 -r
y = 0.5017x + 12.292
45 -

40 -

SP 3 0 -

20 -

10 -

0 10 20 30 40 50 60 70 80
Time (days)

Figure 3.15 t/Sr versus t relationship for the thirteen repair materials (R2 = 0.99)

Hence the equation can be written to describe the development of shrinkage ratio with

time:

c
Sr = -------------------------------------------------
(
Eq.3-8
r 12.292 + 0.5017J

This expression is plotted for a period of 450 days and also compared with the

experimental data in Figure 3.16:

106
Chapter 3 - Selection of materials for optimal performance of concrete repair

o w w w ^
3B B>juuqs
X np O J J 3 U IIJ JB 3 § > J B U jjq S 3 B § J 3 A B 3 l|J J O O U B > ] 8c S / S = J S
 ^n ap ici j - o eie c u o n or m aterials ior optim al pertorm ance o f concrete repair

The correlation coefficient between the experimental curve (representing the average

behaviour of the thirteen materials) in Figure 3.16 and the curve based on the predicted

values (Eq. 3-8) is R2 = 0.9965.

The expression adequately represents the shrinkage-time relationship of the average repair

material. It allows extrapolation to later ages to represent long term performance.

3.4.2.5 Development of Compressive Strength

Although there is a lack of test data on the compressive strength versus time relationship of

repair materials, there is comprehensive data of this relationship for concrete. In the

absence of strength-time relationship information for repair materials, it will be assumed

that their behaviour will be similar to that of concrete. The extensive data for concrete

available in literature are used to derive a general strength time relationship. The

compressive strength versus age relationships are plotted in Figure 3.17, in this figure,

part (a) shows the long term development of compressive strength for a number of

laboratory specimens made up of three different water/cement ratios. Part (b) shows the

development of the same ratio for specimens cured under differing atmospheric conditions.

Table 3.21 Development of Compressive strength with age

Age (days) 0 2 4 6 7 8 12 14 15 20 21 28
Compressive strength ratio
0 0.26 0.45 0.56 0.62 0.66 0.78 0.82 0.84 0.91 0.93 1
(fc /f28)

Age (days) 90 400

Compressive strength ratio
1.11 1.11
(fc /f28)

fc = compressive strength at age t days f28 = compressive strength at 28 days age.

108
 v^ucijjiw j - ocic^uun ui m aterials lor opnm ai perrorm ance o t concrete repair

12 500
Water/cement ratio:

10 000

7500

5000

2500

0 1
1
i
3 7
i________ t_______ l__________ i________i__ i____ i____ i____ 10
28 90 1 3 5 10 20 40
(a)
Days Years
Age (log scale)

110
46*C(115°F)
100
I 90
” 80
a. 7 0
o 60 l°C(10Cf F)-

c 40
g 1 30
Si 4°C (40'F)

(b)
3 5 7 14 21
Age - days

Figure 3.17 D evelopm ent o f strength o f concrete with age82

The data for concrete mixes o f water/cement 0.53, cured at 21°C are extracted from Figure

3.17 and the strength ratios (fc/f28) are listed in Table 3.21. In the absence o f a know n

mathematical profile for the strength-time relationship, the hyperbolic expression used for

the prediction o f creep or shrinkage98,99 is also applied to the developm ent o f com pressive

strength with time. If the derived hyperbolic expression correlates well w ith the test data,

then its application will be justified.

Where t = time (days)

109
 ^nupiwi j - ocic^uuu ui materials lor opnmai perrormance or concrete repair

f c = compressive strength

a and b are empirical constants which will be determined from the

experimental results of the concrete specimens used in the reference study.

Rearranging Eq. 3-9 and multiplying throughout by f28 gives:

- T /2 8 = ( a + 60 /2 8
Jc

.. ——= f 2 $Q+ ^ 28^^ Eq. 3-10

J cr

= a'+ b't
f cr

The operation in the previous sections (as applied to shrinkage and creep data) is repeated

for the compressive strength data in Table 3.12 and a graph is produced by plotting time

over compressive strength against time. This produces a line whose slope represents b’,

and the intercept represents a ’ (Figure 3.18)

110
 ^ n a p ic i j - o cicu n u ii ui m a teria ls ror optim al perform ance or concrete repair

4 400
•B 350
I 300
txD
1 250
1 200
I 150
I ioo
j 50
P n _____

0 50 100 150 200 250 300 350 400 450

Time (days)

F ig u re 3 .1 8 D ev elo p m e n t o f c o m p re ssiv e stren g th ratio w ith tim e

Hence the equation can be written to describe the development o f com pressive strength

ratio with time:

, fc t
f cr = — = -------------------------------------------------------------- Eq. 3-11
/ 28 4.4994 + 0.8876/

This expression can be compared with the experimental data which are plotted in Figure

3.19 :

111
 Chapter 3 - Selection of materials for optimal performance of concrete repair

s/mp
83 V 3 m§uo.us ot i Xnp jb qjSuojjs j o o p n y (rcj/ng = jj )
 ^ n a jjic i j - o e ie c u u n o i m aterials ror optim al pertorm ance o f concrete repair

j
The coefficient of correlation, R , between the two curves in Figure 3.19 is 0.9964. This

indicates that the predictive equation and the experimental curve are convincingly related.

Henceforth, the technique which generates the hyperbolic equation for strength-age

relationship is accepted. The technique will be further utilised to predict the development

of tensile strength.

3.4.2.6 Development of tensile strength

In order to obtain an expression for the development of tensile strength of concrete with

age, the relationship between tensile strength, ft, and compressive strength, fc, is

considered :-

0.7
f, = 0 .1 2 /c Eq. 3-12

Rearranging Eq. 3-12 gives:

'0.7
f,
fc = Eq. 3-13
0.12

Therefore, considering the 28 day values,

'0.7
f ,128
Eq. 3-14
0.12

Substituting for ft and f28 from Eq. 3-13 and Eq. 3-14 respectively into Eq. 3-11 gives:

Y .l
/, 1
0.12
Eq. 3-15
%
Yonn 4 .4 9 9 4 + 0.8876/
fm
0.12

113
 ^ n ap ici j - o ciccu u n ui m aterials ior optim al perrormance o t concrete repair

Simplifying,

I Eq. 3-16
\ft2 8 J 4.4994+ 0.8876/1

Therefore,

0.7
_ ft_ t
Eq. 3-17
fl2 t 4.4994+ 0.8876r

f
Equation 3-17 is used to generate tensile strength ratio f = —L- values for t = 0 to 400
f 128

days.

These are listed in Table 3.22.

Table 3.22 Development of Tensile strength ratio (ft / ft2s) with time

t 0 2 4 6 7 8 12 14 15 20 21 28 50 100 150 200 250 300 350 400

0.00 0.44 0.63 0.73 0.76 0.79 0.87 0.90 0.91 0.95 0.95 0.98 1.02 1.05 1.06 1.07 1.07 1.07 1.08 1.08
II
00

An equation for the development of tensile strength ratio is required in the form:

f, t
f ,n f , 2 *(a + bt)
■ f - ft - 1
fa s a ' + b ’t

A graph of time (t) against time / tensile strength ratio (t/fr), is plotted in Figure 3.20 to

determine the constants a ’ and b

114
 ^ n a p ic i j - o cicu u u ii o i materials ior optim al perlorm ance o f concrete repair

400

y = 0.9216x + 2.7975

250
00

200
•a 150
100

0 100 200 300 400 500

time (days)
Figure 3.20 Development of tensile strength ratio with time

Therefore:

ft t
~T~ = ------------------------ Eq. 3-18
fm 2.7975 + 0.9216/

The expression can be compared to the limited experimental data available on the

development of tensile strength in repair materials with time58. Three materials were tested

over a period of 28 days; a styrene acrylic concrete, an SBR concrete and an acrylic

concrete. A comparison between the growth of tensile strength as predicted by equation

3.18 and the actual growth of tensile strength in the three repair materials is shown in

Figure 3.21.

115
Chapter 3 - Selection of materials for optimal performance of concrete repair ©

to
(N

O
ci

& _ -a
O. aj
T3
X !-
0) at a .
P

M
3

o OL
o_nT^3
O.
X s-
a> o.
I

o^ Cl
-O at
X

■a
a
a
-a
_ -a
D. D
X >-
<u a . a.
V -
o
c
4 o
si
3
CL
E
o
o U
ro (N O rj
ro
••■(-IULU/J\[) 1{J§U3JJS 3 |!S U 9 J at
•_
3
W D
 ^ u ap ici j - o c ic d iu n ui m aterials ror optim al periorm ance o t concrete repair

The predictive quality of Eq. 3-18 in Figure 3.21 shows good correlation with the

experimental results. This is particularly significant, as the predictive approach is based on

the results of different experiments to those on which is was tested and is derived from

compressive strength relationships of concrete while the experimental tensile strength data

plotted in Figure 3.21 is for repair material formulations. The correlation coefficients for

the styrene acrylic and SBR concrete are both over 0.99. The coefficient of correlation for

the acrylic material is slightly lower at 0.967, as is evidenced from the graph.

3.4.2.7 Development of Elastic Modulus

Pinelle100 provides data on the development of Elastic Modulus from early age in

commercial repair materials with varying constituents. Data from three different materials

is provided (shown in Table 3.23). The materials consist of an unmodified cementituous

repair material, acrylic based material and a vinyl acetate based material.

117
 limpid j —oeiecuon 01 maieriais ior optimal pertormance ol concrete repair

Table 3.23 Development of elastic modulus with time in repair materials

Time after casting Cementituous Acrylic Vinyl acetate

(days) (kN/mm2) (kN/mm2) (kN/mm2)
0 0 0 0

7 480 310 205

14 680 420 250

21 770 500 270

28 790 530 273

35 800 560 276

42 810 585 279

56 820 588 282

63 830 591 285

70 840 594 288

77 850 597 291

84 . 860 600 294

91 870 603 297

98 880 606 300

The value of elastic modulus at each age can be expressed as a proportion of the 28 day

elastic modulus of the material (E/E28). For example, considering the data for the

Cementituous repair material (Table 3.23) and dividing throughout by the 28 day elastic

modulus, gives the proportions listed in Table 3.24.

Table 3.24 Elastic modulus of cementituous repair material as a ratio of the 28 day elastic modulus

Days after casting 0 7 14 21 28 35 42 56 63 70 77

Ratio of elastic modulus (E/E28 ) 0 0.65 0.86 0.97 1.0 1.03 1.05 1.06 1.07 1.08 1.09

Days after casting 84 91 98

Ratio of elastic modulus (E/E28 ) 1.10 1.11 1.12

This procedure was completed for each of the three repair materials. The elastic modulus

ratios (E/E28) against age after casting are plotted for all three materials in Figure 3.22. An

average relationship (best-fit line) of all three materials is also plotted.

 Chapter 3 - Selection of materials for optimal performance of concrete repair

oijej
sn (n p o iu o itsn jg
 li m p i d j - o eie c u o n 01 materials ror optim al perform ance o f concrete repair

The line representing the average relationship between elastic modulus ratio and age (also

shown in Table 3.25), correlates well with the test data for the three materials (coefficient

of correlation R = 0.999, 0.993 and 0.992 respectively). This means that the average

relationship can be applied to any repair material to a reasonable degree of accuracy.

Table 3.25 Average (E/E2g) ratio versus time relationship for the repair materials

Age (days) 0 7 14 21 28 35 42 56 63 70 84 98
Average Ratio of Elastic moduli E/E28 0.00 0.65 0.86 0.97 1.00 1.03 1.05 1.06 1.07 1.08 1.10 1.12

This is used to determine a general expression for the development of Elastic Modulus

ratio using the following hyperbolic relationship of the type used previously for the other

properties .

E _ t
E l 8 E 2 * (a + b t ) E q > 3 1 9

E = — -—

r a'+b't

By plotting time/elastic modulus ratio against time the constants a ’ and b ’ can be found in

equation 3-19.

Figure 3.23 shows good correlation and gives the following relationship between the

average elastic modulus and age.

120
 ^im ptci j - oeieu u on 01 materials ior optim al perform ance o f concrete repair

100
fcf 90 - y = 0.8725x + 3.2637
R2 = 0.9984
o' 80 -
La 70 -
C«
= 60 4
1 50 -

I 40 ■
*C3 30 -
| 20-
■i 10 -
H

0 20 40 60 80 100 120
Time (days)

Figure 3.23 Development of elastic modulus ratio (Er) with time based on the average of 3 repair

materials

Er E = ----------------------
tr = ---- 1 Eq. 3-20
E2S 3.26374-0.8725/

The elastic modulus ratio as predicted by equation 3-20 is compared to the actual

experimental data corresponding to the average curve for the three repair materials in

Figure 3.24.
 Chapter 3 - Selection of materials for optimal performance of concrete repair

'(3/t)
oijbj
sn|npoui o p sc p / amp
 ^ n a p ie i j - s e l e c t i o n o t m a t e r i a l s t o r o p t i m a l p e r f o r m a n c e o f c o n c r e t e r e p a ir

The coefficient o f correlation R2 = 0.996. The accuracy o f the predictive equation can be

shown by plotting the experimental data and the predicted growth o f elastic m odulus based

on the 28 day value (Figure 3.25) for one o f the repair materials.

350

g 300
g
g 250
M r = 0 .9 9 7
c/5 200
| 150
£
O 100

CD

0 20 40 60 80 100 120
tim e (d ay s)

— exper i ment al — pr edi ct ed

Figure 3.25 C om parison o f experim ental and predicted values o f elastic m odulus for vinyl acetate

m aterial

Figure 3.25 shows that if only the 28 day elastic modulus o f the vinyl acetate material had

been provided, then using the predictive hyperbolic equation, the elastic m odulus o f the

material at any age can be predicted with a good degree o f accuracy.

3.4.2. 7.1 Tensile and compressive elastic moduli

The elastic m odulus o f concrete is usually determined under compression. However, repair

materials are subjected to tensile stress as their inherent free shrinkage is restrained, mainly

at their interface, with the substrate concrete. It is assumed, for the purposes o f the

procedure developed in this thesis, that the elastic modulus o f a repair material in tension

and com pression is equal. The reasons for this assumption are discussed in 3.2.3.2.I.

123
 -t - m v p u ^ c u u i c iu i u cieiim iiirig me m -snu perrormance o t repair m aterials

4 The procedure for determining the in-situ performance

of repair materials.

4.1 Chapter objective

The objective of this chapter is to develop a procedure for determining the in-situ

performance of concrete repair. The equations which predict the development of material

properties with time, derived in the previous chapter, will be utilised. The procedure will

be developed for implementation in a computer program.

4.2 Introduction

In Chapter 3, the properties of repair materials which are crucial to the successful

performance of the repair are identified and the development of those properties with age is

described by generic relationships derived empirically using a hyperbolic function. These

equations will be used in this chapter to determine critical tensile strains developed in a

repair patch. This critical strain will then be compared with the tensile strain capacity of

the repair material in order to establish the likelihood of failure (cracking) of the repair,

and, should a repair have been deemed to fail, to also suggest the likely time of failure (in

days) after the application. The time of failure can range between the short term, typically

within the first 50 days of application, to the longer term of 400 to 500 days. The

developed algorithms are incorporated into the computer expert system to provide an

expedient method for determining the performance of a patch repair.

124
 -r m v p u t t u u i o iu i u c ic im in in g m e m -snu perrormance o t repair materials

4.3 Procedure for determining the performance o f a repair

material

Consider the repair material Shucrete 1 which has the following properties at 28 days age

(Table 4.1). Shucrete 1 is an imaginary material whose properties are generally typical of a

repair material:

Table 4.1 The properties of Shucrete 1 at 28 days age.

Compressive strength N/mm2 30

Modulus of Rupture N/mm2 8.5

Tensile Strength N/mm2 5

Shrinkage microstrain 630

Creep (at 30% stress/strength) microstrain 794

Elastic Modulus kN/mm2 34

These properties represent the information typically required from repair material

manufacturers. These properties are usually provided by manufacturers, with the exception

of Creep which, due to a lack of coherent understanding amongst specialists concerning

which properties of a repair material are crucial to its overall performance, is currently not

considered important and hence rarely specified by manufacturers.

125
 -r — m u ijiuccuuic lux u cien iiiiiiiig me m -siiu periorm ance o i repair m aterials

4.3.1 Determining Tensile Strength from modulus of rupture

It is usual for the Modulus of rupture to be provided in manufacturers’ literature. This

property can be converted to an approximate tensile strength using a conversion factor82.

Tensile strength = Modulus of rupture / 1.7

For Shucrete 1:

Modulus of rupture = 8.5 MPa

Therefore, Tensile strength = 8.5 /1.7 = 5 MPa

4.3.2 Modifications for climate

Shrinkage and creep are affected by climatic conditions, specifically temperature and

relative humidity. Considering 70% relative humidity as a datum, shrinkage increases by

approximately 2% for each per cent decrease in relative humidity and decreases by

approximately 3% for each per cent increase in relative humidity101.

Shrinkage can be assumed to increase or decrease by 1% of the 15 °C shrinkage value with

each relative increase or decrease in temperature101.

It can be assumed that creep changes linearly with temperature at a rate of 1.25% of the

15°C creep for every degree change in temperature101.

109 •
Nawy gives a chart to determine the effects of relative humidity on creep of concrete,

which is shown in Figure 4.1.

126
 -r x ..w p u tv u m i, xui u cicu m iim g uic lii-siiu perrormance or repair materials

0.9
Creep correction factor

0.6
0.5

Relative humidity H (%)

Figure 4.1 Effect of relative humidity on creep102

The equation specifying the relationship in Figure 4.1, within the limits of relative

humidity 42.9% and 100%, has been determined as:

krh = -0.007(7?#%) +1.3 Eq. 4-1

Where krh = creep correction factor for relative humidity

4.3.2.1.1 Strength and elastic modulus modified by climate

Temperature during curing is known to have an influence on the development of

compressive and tensile strength82.

Figure 4.2 gives the 28 day compressive strength of the same concrete cast at different

times of the year in the UK. Generally, the summer months can expect a reduction of

between 5% and 10% of the strength compared with cooler months. The temperature

during the crucial first days of curing is deemed to be responsible for this effect.

127
 -r .iiv FiutL u u ic iui u cicu u iiim g me in-situ perrormance o t repair materials

45

co
CL

?£ 40~
O
c
)
s
O
>
35-

I 30-
o
O

20 Jan Mar May Jul Sep Nov

Month

Figure 4.2 Influence of initial temperature on average monthly compressive strength in the UK82

Little data is available on the development of strength of concrete or repair materials at

different temperatures out of doors. Most research is based on specimens cured at constant

temperature in the laboratory. It is, therefore, difficult to recommend a correction factor for

strength based on seasonal temperature, or indeed relative humidity. If it is assumed that

any seasonal reduction in strength is also accompanied by a corresponding reduction in

elastic modulus then any seasonal effects on tensile strain capacity of a repair may be

negated. It is, therefore, assumed that seasonal effects on strength can be neglected.

4.3.3 Establishing seasonal temperature and RH variations

The procedure developed in the thesis for predicting tensile strain in a repair material

(repair patch) will take account of the effect of seasonal and geographical temperature and

relative humidity variations. As previously stated, these variations affect the development

of creep and shrinkage with time.

Mean temperatures and relative humidities can be determined to a satisfactory degree of

103 • ♦ •
accuracy seasonally . In order to do this, it is necessary to be able to evaluate temperature

128
 *r j j i u t t u u i c iui u c i c n ii i m ii g me m-siiu penorm ance or repair materials

and relative humidity in different geographical regions at different times o f the year. Table

4.2, Table 4.3 and Figure 4.3 represent an expedient method for determining local

temperatures and relative humidities in the U K 103. Great Britain is divided into three zones:

north, mid, and south. For each zone average temperatures and relative humidities are

given. These figures can be used to determine the conditions in which repair materials will

cure in the field. M anufacturers’ data, which is based on standard specified curing

conditions, can then be modified to account for the effects o f temperature and relative

humidity at the location and the period that materials are used in patch repairs.

Table 4.2 Seasonal average tem perature variation in the U K 103

Dec - M ar - Jun - Sep -

Feb May Aug Nov North

North 4° 9° 15° 11°

Mid 5° 10° 16° 12° Mid

South 8° 11° 17° 14°

South
Table 4.3 A pproxim ate seasonal average relative hum idity in the U K 103

Dec - M ar - Jun - Sep -

Feb May Aug Nov Figure 4.3 M ap for clim ate tables

North 85% 70% 70% 85% 4.2 and 4.3

Mid 85% 70% 70% 85%

South 85% 75% 75% 85%

For example, assume a repair will be carried out in December 2005 in Edinburgh

(Northern Zone).

Tem perature = 4°C, RH = 85%

129
 -t xnw p u t w u i c iui ucicim ining, me m -siiu periorm ance o t repair m aterials

4.4 Shrinkage of patch repair: correction factors for temperature

and humidity

For the material Shucrete 1, from Table 4.1:

Free shrinkage at 28 days = 630 microstrain obtained at 23°C, 50%RH

(where 23°C, 50%RH, are the laboratory ambient conditions specified by both British and

American standards)

The datum for conversion, as specified in 4.3.2 is 15°C. The shrinkage test was conducted

at a temperature of 23°C. This is 8°C higher than the datum temperature. Therefore,

assuming a 1% reduction in shrinkage per degree centigrade, predicted shrinkage at the

datum temperature of 15°C and 50% RH:

or . t Shrinkagen°c
Shrinkage,r.r = ---------------- = 583microstrain
1.08

The temperature in Edinburgh in December, from Table 4.2, is 4°C.

Therefore, applying the temperature correction for shrinkage at 4°C relative to the datum

temperature gives the Shrinkage at 4°C, 50% RH:

4 -1 5
Shrinkagerc =583 1+ = 519microstrain
V V 1 0 0 J J

Therefore, the expected free shrinkage in a patch repair made with Shucrete 1 in

Edinburgh, assuming the RH to remain at 50%, is 519 microstrain.

For the purposes of applying a correction for relative humidity to the shrinkage of the field

patch repair, the datum value is taken as 70% RH.

130
 -r - i ji«^ p u u c u u ic u cicn m m iig me in-suu periorm ance o t repair materials

In accordance with section 4.3.2, for each percentage point decrease in relative humidity

from the datum RH (70%), shrinkage increases by 2%. In order to modify the current

shrinkage (519 microstrain) at 50%RH, to a shrinkage obtained at 70%RH, a 2% decrease

per percent increase in RH is applied. Therefore, the shrinkage of a repair patch made with

Shucrete 1under a field temperature of 4°C and datum RH of 70% is given as:

shrinkage IiH1Q0/o = ^ * n^ ^ .RH50% = 3 7 0 microstrain

From Table 4.3, the relative humidity in the northern zone of the UK (into which

Edinburgh falls) in December is 85%.

In accordance with 4.3.2 , for each percentage point increase in RH above the 70% RH

datum, a 3% shrinkage reduction is applied to determine the shrinkage of the field patch

repair on a site in Edinburgh exposed to 4°C, 85% RH.

Shrinkage of the patch repair at 4°C, 85% RH:

3
370 1 —(85 —70) = 204 microstrain
100

131
 -r - in v p iu tc u u ic iui uciciixiiiim g uie in-siiu periorm ance or repair materials

4.5 Creep of patch repairs: correction factors for atmospheric

conditions , specimen size , and age at ioading

In addition to modifying values of creep given in manufacturers’ literature to account for

difference in temperature and relative humidity between laboratory cured specimens and

conditions in the field, there is a need to consider other variations between the laboratory

and the field, such as dimensional differences and age of specimen at loading. This section

examines the effect of the parameters that have been identified as important in determining

the field creep properties of a repair material based on its laboratory values

(manufacturers’ data).

4.5.1 Modifying creep for early age loading

Generally, creep tests in the laboratory are conducted on specimens that have been cured

unloaded for 28 days. A repair material in a patch repair, however, will begin to creep as

soon as it is subjected to restrained shrinkage tensile stresses, which occur immediately

after the application of the patch repair. It has been shown84 that creep of concrete and of

repair materials is influenced by the age of the material when load is applied to it. The

effect on creep of the age of loading is mainly due to the increase in strength of the

concrete with age101.

Data has been obtained to compare the creep performance of materials when loaded at

different ages (1 and 7 days) in a tensile creep rig84. The materials tested were concretes

with admixtures such as macrofibres, microfibres and superplasticizers. The data for four

materials are listed in Table 4.4. The four materials are labelled C35, C55, S35, S55 where

suffixes represent the water to cement ratios. The ‘C’ prefix indicates normal Portland

132
 -r i . 1^ ^ u ^ u u i c iui u c ic iiiiiiim g me m-suu perrormance or repair materials

cement and ‘S ’ indicates normal Portland cement with silica fume. The 28 day tensile

strengths o f the materials are given in Table 4.4.

Table 4.4 Specific C reep o f concrete loaded at different days (pm /m m /N /m m 2)

Specific Creep Strains (M icrostrain/N /m m 2)

M aterial ...
C55 loaded at
■■■ ■: A V . f •' . .v A; y
C35 loaded at S55 loaded at S3 5 loaded at
Age at load
1 day 7 days 1 day 7 days 1 day 7 days 1 day 7 days
application
28 day
tensile 4.4 N /m m 2 4.9 N /m m 2 4.8 N /m m 2 5.6 N /m m 2
strength
0 0 0 0 0 0 0 0 0
Time after load application (days)

2 60 25 30 18 62 24 70 26
4 70 33 47 21 87 37 89 34
6 80 42 55 24 100 50 100 40
10 100 54 70 32 120 68 117 50
15 110 67 80 41 134 88 130 56
20 120 73 85 49 145 100 140 60
25 128 80 91 53 152 111 149 64
30 135 88 96 58 158 120 155 69
35 141 97 99 61 162 128 160 72

40 147 100 100 63 165 132 163 75

M aterials loaded at 1 day were subjected to a constant stress o f 0.77 N /m m 2. M aterials

loaded at 7 days were subjected to a constant stress o f 1 N /m m 2.

The most important factor that controls the creep strain o f the repair patch is the

stress/strength ratio induced by the applied load. The constant applied stresses to the test

samples whose creep data are given in Table 4.4 will result in different stress strength

ratios as the aging o f the test samples during the creep test period results in a gradual

increase in strength. In order to determine the stress strength ratio at each time o f creep

strain monitoring, the tensile strength o f the repair material at each age is determ ined from

the following expression, equation 3.18, derived in Chapter 3:

 ~t— m e jjiu lcu u ic iui u cien n in in g m e m -situ periorm ance o l repair m aterials

/ _ ' Equation 4.1

2.7975 + 0.9216?

This equation relates the tensile strength at any age (t) to the 28 day tensile strength (ft2 s) of

a material. The resulting tensile strength at each age is given in Table 4.5.

For the samples loaded at seven days, those materials have already achieved a seven day

tensile strength, the materials loaded at one day have achieved a one day strength. This

accounts for the different tensile strengths apparent in Table 4.5 between repair materials

loaded at 1 and 7 days.

The procedure for calculation of tensile strength at a particular time is illustrated by the

following example:

Consider material C35 loaded at an age of 7 days to commence the creep test. At 10 days

after the start of the creep test, the age of the specimen is 17 days. The 28 day tensile

strength of the material, ft28 = 4.9 N/mm2

Therefore, substituting into Equation 4.1 for ft28 = 4.9 N/mm2, t = 17 days gives:

f< _ 17
4.9 2.7975 + 0.9216*17

Therefore, ft at 17 days (10 days under creep load) = 4.66 N/mm2. This value is listed in

Table 4.5 for material C35 loaded at 7 days age to commence the creep test and represents

the tensile strength of the material at 10 days after the application of creep load (age 17

days).

Because the materials were subject to a constant stress in the creep tests, the stress strength

ratio changes with time.

134
 ^..w piw -r — i n<~ ( j i u l s u u i c iui u c iciiiiiiiiiig me in-siiu periormance or repair materials

Table 4.5 Developm ent o f T ensile Strength (N /m m 2) during creep testing

Tensile strength (N/mm2)

M aterial C55 C35 S55 S35 C55 C35 S55 S35
Age at load
application 1 1 1 1 7 7 7 7
(days)
0 1.22 1.36 1.33 1.56 3.44 3.83 3.75 4.38
2 2.45 2.73 2.68 3.12 3.69 4.11 4.03 4.70
Time after application of load (days)

4 3.07 3.42 3.35 3.91 3.87 4.31 4.22 4.92

6 3.44 3.83 3.75 4.38 4.00 4.45 4.36 5.09

10 3.87 4.31 4.22 4.92 4.19 4.66 4.57 5.33

15 4.15 4.62 4.52 5.28 4.34 4.83 4.73 5.52
20 4.31 4.80 4.70 5.49 4.44 4.94 4.84 5.65
25 4.42 4.92 4.82 5.62 4.51 5.02 4.92 5.74
30 4.49 5.01 4.90 5.72 4.56 5.08 4.97 5.80
35 4.55 5.07 4.96 5.79 4.60 5.12 5.02 5.86
40 4.59 5.12 5.01 5.85 4.64 5.16 5.06 5.90
(where shaded boxes represents values given in the illustrated exam ples w ithin the text)

The stress/strength ratios are obtained by dividing the applied constant stress (0.77 N /m m 2

for specimens loaded at 1 day age, 1 N /m m 2 for the specimens loaded at 7 day age) by the

tensile strengths o f each test material at the specific time under creep loading. For exam ple,

consider material S35 loaded at 1 day age to a constant stress o f 0.77 N /m m 2. After 10 days

under creep load (age o f specimen = 11 days) its tensile strength is 4.92 N /m m 2 (Table

4.5). Therefore the applied stress/strength = 0.77 / 4.92

= 15.6%

The stress/strength ratios at each time (days) under creep loading, corresponding to the

specific creep and strength data o f specimens given in Table 4.4 and Table 4.5, have been

calculated by the above procedure and are listed in Table 4.6. Table 4.6 shows that the

135
 -r i..v H. u ^ u u a iui u sitm u n u ig me in-siiu periormance or repair materials

stress/stength ratios are generally less than 30% throughout except for specim ens loaded at

the age o f 1 day which showed high stress/strength ratios at the first day o f loading.

Table 4.6 Stress/Strength ratios

Stress / strength ratio (%)

M aterial C55 C35 S55 S35 C55 C35 S55 S35
Age at load
application 1 1 1 1 7 7 7 7
(days)

0 63.0 56.5 57.7 49.5 29.1 26.1 26.6 22.8

2 31.4 28.2 28.8 24.7 27.1 24.3 24.8 21.3

Time after application of load (days)

4 25.1 22.5 23.0 19.7 25.9 23.2 23.7 20.3

6 22.4 20.1 20.5 17.6 25.0 22.4 22.9 19.6
10 19.9 17.9 18.3 15.6 23.9 21.4 21.9 18.8
15 18.6 16.7 17.0 14.6 23.1 20.7 21.1 18.1
20 17.9 16.0 16.4 14.0 22.5 20.2 20.7 17.7
25 17.4 15.6 16.0 13.7 22.2 19.9 20.3 17.4
30 17.1 15.4 15.7 13.5 21.9 19.7 20.1 17.2
35 16.9 15.2 15.5 13.3 21.7 19.5 19.9 17.1
40 16.8 15.0 15.4 13.2 21.6 19.4 19.8 17.0

Table 4.7 transforms the data in Table 4.6 to show the specific creep strains that would

have occurred in the materials if they were loaded at constant stress/strength ratio o f 30%.

Assuming a linear relationship between stress/strength ratio and creep, the data listed in

Tables 4.4 and 4.6 are used to determine the specific creep values at a stress/strength ratio

o f 30%. These values are given in Table 4.7.

It is well established in existing literature that there is a linear relationship between

stress/strength ratio and specific creep strain o f a given cem entituous m aterial101. The

linear relationship is less valid at very high stress/strength ratios where m icro-cracking

136
 v .m Hi u -t — i nc piuecuuic iui ucieim ining me in-siiu periormance ot repair materials

within the concrete matrix can lead to non-linear behaviour. For the purposes o f analysis in

this thesis, linear behaviour is assumed since if high stress/strength ratios do occur in a

repair patch, their duration is extremely short relative to the creep period. Table 4.6 shows

high stress/strength ratios immediately after loading on the first day o f specim ens loaded at

1 day age but the stress/strength ratio decreases rapidly under sustained loading. For

example, consider C55 loaded at 1 day age in Table 4.6. The stress/strength ratio at loading

(curing age: 1 day, age at load application: 0 day) is 63% which rapidly reduces to 31.4%

after age at load application o f 2 days. The reduction is rapid within the first few hours o f

creep loading.

Table 4.7 Specific creep o f specim ens extrapolated at 30% stress/strength ratio

Specific creep strains (m icrostrain/N/m m 2)

Loaded at 1 day Loaded at 7 days

Material C55 C35 S55 S35 C55 C35 S55 S35
Age at loading
1 1 1 1 7 7 7 7
(days)

0 0 0 0 0 0 0 0 0
2 57 32 65 85 28 22 29 37
Time after application of load (days)

4 84 63 114 136 38 27 47 50
6 107 82 146 171 50 32 66 61
10 151 118 197 224 68 45 93 80
15 178 144 236 267 87 59 125 93
20 202 159 266 299 97 73 145 102

25 220 175 286 327 108 80 164 110

30 236 187 302 345 120 88 179 120
35 250 196 313 361 134 94 193 127
40 263 199 322 371 139 98 200 133

137
 -t — m i p u L tu u ic iui u cicim iiiiiig me in-siiu periorm ance or repair m aterials

The following example shows the procedure adopted for calculating the creep strains at

30% stress / strength ratio. Consider material C55 loaded at 7 days age to commence creep

testing and determine the specific creep strain corresponding to the applied stress/strength

of 30% at 15 days after commencing the creep test (age of specimen, 15+7 = 22 days).

Creep data in Table 4.4 gives:

Specific creep at 15 days after load application = 67 microstrain/N/mm2.

The corresponding stress / strength ratio at age of load application 15 days, from Table 4.6,

is 23.1%. Hence, specific creep at stress/strength ratio of 30% = 67 * (30/23.1)

= 87.2 pm/mm/N/mm2

The specific creep data corresponding to the applied stress/strength of 30% are calculated

for the materials from Table 4.4 to Table 4.6 and are listed (rounded to the nearest integer)

in Table 4.7. The average specific creep data in Table 4.7 at a stress/strength ratio of 30%

for materials loaded at 1 day age and 7 day age are plotted in Figure 4.4. It is quite clear

that at a constant stress/strength ratio, the specimens loaded at 7 days age show much

lower specific creep than corresponding specimens of the same material loaded at 1 day.

138
 ^i.cpLvi *r m e piu^cuuic iui ueiennining me in-siiu perrormance ot repair materials

-♦— Creep 30% stress/strength

ratio, specimen loaded at 1
25 0 day age
N « — Creep 30% stress/strength
£
ratio, specimen loaded at 7
200
z days age
E
| 150

$100
o
o
oQ-
</)

tim e s i n c e load in g (d a y s)

Figure 4.4 Specific creep-tim e relationship for concrete loaded at 1 and 7 day ages at a stress/strength

ratio of 30% (average o f Table 4.7 m aterials).

The ratio o f specific creep o f the material loaded at 1 day to specific creep o f the material

loaded at 7 days both at 30% stress/strength ratio can now be determined. These values are

listed in Table 4.8 at various incremental times under creep loading.

139
 -r — in v p iu t tu m t xui u ciciiim m ig me 111-siiu periorm ance 01 repair materials

Table 4.8 Ratio of specific creep due to loading at 1 day to loading at 7 days, at a stress strength ratio

of 30%.

Time after load Material

application C55 C35 S55 S35
(days)
0 0.00 0.00 0.00 0.00
2 2.07 1.44 2.23 0.65
4 2.19 2.31 2.42 0.95
6 2.13 2.56 2.23 1.15
10 2.22 2.62 2.12 1.24
15 2.04 2.42 1.89 1.28
20 2.07 2.19 1.83 1.20
25 2.04 2.19 1.74 1.25
30 1.96 2.12 1.69 1.26
35 1.87 2.08 1.63 1.28
40 1.89 2.04 1.61 1.27

The specific creep ratios listed in Table 4.8 are plotted in Figure 4.5 against the time under

creep loading. The graphs in Figure 4.5 show that after the early period under load (about 5

days), the specific creep ratios attain a relatively constant value.

For example, considering material C55 at day 10 under creep loading at a constant stress

strength ratio of 30%, the specific creep of the material loaded at 1 day age is 2.22 times

higher than the specific creep of the material loaded at 7 days after casting.

140
 —
t— n i t ^ luttu uic hji uclciliiinnig me in-siiu periormance or repair materials

3.00

G)
.£
'S
TO
O
L.
0) o 2.00
•a o>
E
3 ™ 1.83
Q. > »
0) T3TO
0
i_
O >»
o TO
T3 1.00
o0
a
w

5 10 15 20 25 30 35 40
Time under creep load (days)

Figure 4.5 Increase in specific creep due to loading at early age (stress/strength ratio 30% )

Considering all data in Figure 4.5 (and Table 4.8) and excluding data points at up to 5 days

under creep loading, the average increase in specific creep due to loading at 1 day as

opposed to 7 days is 1.83 or 183%. This factor is incorporated into the procedure for

determining the performance o f a patch repair by calculating the tensile creep strains

produced by the restrained shrinkage tension in the repair material at regular incremental

ages from the time the repair patch is applied (Chapter 3).

In situations where the difference between the age o f creep loading in the laboratory test

(representing m anufacturer's data) and actual insitu creep due to early loading is

represented by a factor greater than 1.83, the material will creep more in practice than the

predicted amount and, therefore, result in greater relaxation o f tensile stresses. This

provides an additional factor o f safety for the repair patch performance against cracking,

which is acceptable. In situations where the increase is less than provided by the factor

141
 -t - m e |jiu^ cuuic iui u eiem u m n g m e m -situ periorm ance o t repair materials

1.83 (such as would occur with material S35 in Figure 4.5) the material will theoretically

creep less than the analytical procedure has assumed. This is not desirable. In practice, the

creep data provided in the analysis is likely to be based on test specimens loaded for creep

testing at 28 days age. This would represent the typical basis for manufacturers’ test data

on their repair materials. The actual transfer of tensile stress in a repair patch, on the other

hand, is immediate after the application of a repair patch, which follows the onset of

shrinkage. The above data from which the factor of 1.83 was established is based on the

difference in specific creep between specimens loaded at 7 days and 1 day age. Therefore,

an additional factor of safety is inherent for patch repairs whose creep data is obtained by

load application at 28 days age. This is representative of the actual information the

software system will be supplied with. This clearly suggests that the actual creep which

occurs in the repair patch will be higher than that which the software system will estimate.

Thereby providing a factor of safety.

Figure 4.6 shows an estimate of the actual difference in creep which occurs through

loading creep specimens at 28 days and 1 day age. The curve representing the development

of specific creep with time for a material loaded at 28 days was developed using the linear

relationship between specific creep and stress strength ratio101.

142
 -t - m e p u t c u u i c iui ueicrmining me m-situ performance o f repair materials

-♦— C r e e p 3 0 % s t r e s s / s t r e n g t h , lo ad a p p l i c a t i o n 1 d a y a ge
350
- ■ — C r e e p 3 0 % s t r e s s / s t r e n g t h lo ad a p p l i c a t i o n 7 d a y s a ge

— C r e e p 3 0 % s t r e s s / s t r e n g t h lo a d a p p l i c a t i o n 28 d a y s agi

300

250
o.
ao
o 200
L.
183 tJ7u
o 200%+

W m
100

50

0 10 20 30 40 50
Time since loading

Figure 4.6 Estim ated specific creep o f m aterial loaded at 1 day, 7 days and 28 days age

This hypothesis correlates well with the results o f experiments by U litskii104 who showed

that for concrete loaded at 3 days, a correction factor for creep o f 2.0 is required to

compare the creep value with the concrete loaded at 28 days, (see Table 4.9). All

specimens were loaded at the same stress/strength ratios.

Table 4.9 C reep m odification factors for early age loading (con crete)104

Age o f concrete at
loading (days) 5 7 10 14 20 28 40 60 90 180 360
C orrection factor for
norm al curing 2.0 1.8 16 1.4 1.2 1.1 1.0 0 S 0 7 0 6 0.5 0.45
C orrection factor for
autoclaving and steam -
curing 1.5 1.4 13 1.23 1.2 11 10 0 S 0.7 0.6 0.5 0.45

143
 v^nc^ivi - m e p iu ecu u ic iui u eien m n in g m e m -snu periorm ance o t repair materials

4.5.2 Creep of patch repair: correction factors for temperature and

relative humidity

Currently, there are no standards for the determination of creep in repair materials,

therefore, an adapted version of ASTM C 512 is the recommended standard. Standard

creep tests should be carried out at 20°C and 50% RH.

4.5.2.1 Temperature correction

For the material Shucrete 1, from Table 4.1:

Creep at 20°C = 794 microstrain

(where 20°C is the laboratory ambient condition to ASTM C 512)

In accordance with section 4.3, there is linear change in creep of 1.25%, for every

percentage point increase or decrease in temperature under which creep takes place.

The datum for conversion, as specified in section 4.3.2 is 15°C. The creep test was

conducted at a temperature of 20°C. This is 5°C higher than the datum temperature.

Therefore, assuming a reduction of 1.25% in creep per degree centigrade, predicted creep

at the datum temperature of 15 °C is :

f' (1.25 *5)^|

Creep20oC = Creep]15°C ** 1 + - — ------------ -
V 100

Creeper = Q*gg^ 20°c. _ 747 microstrain

1.0625

The temperature in Edinburgh in December, from Table 4.2, is 4°C.

Therefore, applying the temperature correction for creep at 4°C relative to the datum

temperature gives:

144
 -r — piutw uuic iui u cicn im iiiig me 111-siiu periorm ance or repair materials

1.25 * (4-15)
Creep4oC = Creep 15°C 1+
100

= 656.7 microstrain

4.5.3 Modifying creep by relative humidity

In accordance with section 4.3.2,

krh = - 0 m i(R H % ) + l 3 Eq. 4-1

Where krh = creep correction factor for relative humidity of exposure

In order to calculate its effect on the creep value, it is necessary to know the relative

humidity under which the standard creep test is conducted on the repair material specimen,

and the relative humidity which will be expected on site when the repair is applied.

The calculation procedure is described with reference to the example introduced in section

4.3.2, as follows:

In accordance with Table 4.3 and Figure 4.3, RH for Edinburgh in December = 85%. This

represents the site environment where the repair application will be made.

The creep data for the repair material has been obtained in accordance with ASTM C512,

testing under a RH of 50%.

From equation 4-1,

k rh50 =-0.007(50)+ 1.3 = 0.95

k rh85 = -0.007(85) + 1.3 = 0.705

Referring to section 4.3.2 and equation 4-1 the following expressions for creep at the

standard RH (50%) and the site RH (85%) can be written in terms of a creep value at the

datum relative humidity:

145
 -r - inw- p iu tc u u ic iui ucicrm im iig m e in-siiu periorm ance o t repair materials

Creeprh50 = k ^ ^ Creep
Creep RH^ = kRH%
5CreepRHclatum
. CreePRH5o _ Creeprh85
k
^RH5 0
k
^ R H 85

The creep strain at RH 50% and temperature 4°C (representing the Edinburgh site) has

been determined in the previous section (4.5.2.1) as Creep4 °c,RH50 = 656.7 microstrain.

656.7 _ CreepRn85
*' 0.95 “ 0.705
656 7
CreepRH85 = ■ ^ *0.705 = 487.3/77icrostrain

4.5.4 Modifying creep for specimen size

The size of a concrete member (nr repair p?_tch) will determine the degree iu which

changes in ambient temperature and relative humidity affect its creep105. Creep strain

decreases with an increase in the size of a concrete member101 for any given stress/strength

ratio. A correction factor to account for this must also be applied to repair patches.

Neville104 provides correction factors for concrete member thickness (which equates to

volume / surface ratios for a cube or cylinder specimen, because as a concrete member

becomes larger, the volume/surface ratio tends to half the concrete depth). Neville gives

these factors in a tabular form, and a corrective factor is provided for situations where one

side of the concrete member is sealed (as for a repair material adjacent to the substrate

concrete surface) where the volume surface ratio tends to the concrete depth. Therefore,

Figure 4.7 fits two curves to Neville’s data, one for concrete repairs (where one side of the

repair is sealed through contact with the substrate surface) and another for elements of

concrete where all the surface area is exposed to the air.

1 A”! 9 ^
Nawy also provides information on the relationship between specimen size and creep. A

linear relationship is suggested but no correction factors are provided for changes in the

146
 t — m u p u t c u u ic iui u cicn iiiiiin g m e m -snu periorm ance o t repair m aterials

volume/surface ratio between concrete repair and newly cast members. However, Figure

4.7 shows clearly, through its similarity with Neville’s data, that Nawy’s linear relationship

represents concrete where all surfaces are exposed to air. The data provided by Neville

(Figure 4.7) fits a logarithmic relationship and these equations will be used to establish the

factors that correct creep for specimen size.

147
 Chapter 4 - The procedure for determining the in-situ performance of repair m aterials

^ •daa.io ,ioj j o p e j u o p o a jjo o

 -r — in v jjiu^ uuic iui u ciciiiu u iiig me m -suu periorm ance or repair materials

A patch repair is surrounded by the substrate on all faces except the top surface exposed to

the atmosphere. The volume of the patch repair is equal to the top surface area multiplied

by depth and clearly volume divided by the exposed surface area equates to depth.

Therefore volume/surface ratio of a patch repair which has only its top surface exposed can

be said to be equivalent to depth. Hence, as shown in Figure 4.7,

ksizerep = -0.3057 In(depth) + 2.8375 Eq. 4-2

Where kSizerep is the creep correction factor for the size of concrete repairs

Also shown in Figure 4.7 is the relationship for the creep correction factor for test

specimens in the laboratory which are exposed to the atmosphere on all surfaces, which is:

ksi:e,ab = -0.308 In {volume / surface) + 2.6367 Eq. 4-3

Where k Si z e i a b is the creep correction factor for the size of laboratory specimens

Equation 4-2 gives the creep correction factor for concrete repairs which are sealed by the

substrate except on one face and Equation 4-3 gives the creep correction factor for

laboratory creep test specimens which are exposed to air on all faces. Both these equations

are taken from Figure 4.7.

The volume surface ratio of the specimen from ASTM C 512 = 25mm;

Assume for simplicity the depth of the proposed repair is 50mm.

Therefore, using equation 4-2, it can be determined for the proposed repair:

k s iz e r e p = -0.3057 ln(50) + 2.8375 = 1.642

and, similarly, using equation 4-3 for the laboratory specimen:

k = -0.308ln(25) + 2.6367 = 1.645

These modification factors need to be applied to the datum creep data (modification factor

of 1.0) so that the creep strain modified for repair size can be calculated.

149
 iui u^iciinum ig uic m -suu periorm ance 01 repair m aterials

The following expressions for creep in laboratory specimens, and creep in a patch repair

can be written in terms of creep values of a datum specimen size:

Creeprep= kIilmpCreepd(,,um
Creephb = ksi!elabCreepdmm
. Creeprep Creep,ab
ksizerep k sizelab

The creep strain at RH 85%, temperature 4°C (representing the Edinburgh site in

December) for a standard (ASTM) specimen of volume to surface ratio of 25mm has been

determined in section 4.5.3 as Creep = 487.3 microstrain.

487.3 _ Creeprcp
" 1.642 1.645
487 3
Creep rp„ = ---- —* 1.645 = 488.2microstrain
rep 1.642

where kSizerep = creep modification factor due to size of repair

ksizeiab = creep modification factor due to size of test specimen

Creepdatum = Creep at correction factor 1

Creeprep = corrected creep for volume/surface ratio of repair material

Creepiab= Creep at RH 85%, temperature 4°C in a specimen with a

volume/surface ratio of 25mm.

In the example above, the modification is slight. This is unsurprising as the volume/surface

ratio of the repair material (its depth) is just twice that of the test specimen.

150
 v^napivi -r - m e piuecuuxc iui u eiem iiiiin g m e in-siiu periorm ance o t repair m aterials

4.6 Modifications for field shrinkage

Kong & Evans101 describe the effect of volume/surface ratio on relative shrinkage in

concrete as follows :-

fi = 10.779^-°005(vo/"",e/^ ce) Eq 4_4

where p = relative shrinkage against datum specimen

This relationship can be used to find the ratio between shrinkage of the laboratory

specimen and shrinkage that would be expected of that same material in the field with a

known volume/surface ratio of the patch repair.

A standard shrinkage specimen for a repair material has the dimensions 25 x 25 x 285mm

(according to ASTM C15796). Therefore, the volume/surface ratio of the repair material

specimen is:

(25x25x285)/((25*25*2)+(25*285*4)) = 5.99 mm

If the planned repair patch has the dimensions 3m x 3m x 50mm, its volume surface ratio is

(assuming only one face is exposed and the remaining faces are surrounded by the

substrate):

(3000x3000x50)/(3000*3000) = 50mm

The relationship in equation 4-4 can now be used to determine the relative shrinkage of the

laboratory specimen:

// = 10.779e"°'°°5(5" )
// = 10.46

Similarly, it is used to determine the relative shrinkage of the field repair:

/I = 10.779^0005(50)
/ i = 8.39

151
 ^nc^iwi - i nc piu^cuuic iui acicriinriing m e m -situ periorm ance o t repair materials

In summary,

Relative shrinkage of ASTM specimen = 10.46

Relative shrinkage of planned repair patch = 8.39

It is also shown by Kong and Evans101 that:

Relative shrinkage in specimen / Relative shrinkage in planned repair = ratio of specimen

shrinkage to in-situ shrinkage

Therefore:

10.46/8.39 = 1.247

/. ASTM specimen shrinkage = 1.247 * In-situ shrinkage

It was shown in section 4.4 that when temperature and relative humidity differences

between the laboratory shrinkage and shrinkage in the field are allowed for, a free

shrinkage in an ASTM specimen of material Shucrete 1 is 204 microstrain.

Therefore, it can be stated that the free shrinkage of the in situ repair patch is:-

204 /1.247 =164 microstrain

4 .7 Properties o f the substrate

In order to accurately predict the performance of a patch repair, it is necessary to know the

properties of the substrate concrete with which it will interact. A core must be taken from

the substrate concrete and its compressive strength and modulus of elasticity determined in

the laboratory.

Considering the case of the site at Edinburgh for shucrete 1 repair:

Height of Core: 250mm

152
 -r — m e pxuv-cuuic iui u e ic n m n m g m e in-situ pertorm ance o t repair materials

Diameter of Core: 100mm

Strength: 58 MPa

Elastic Modulus: 28 GPa

These values, determined through laboratory testing also require correction to account for a

number of factors.
• 82 • •
Neville gives details of the necessary correction factors to account for difference in

height/diameter ratios of core samples (Figure 4.8):

20

1-8

5 16
CD

1-4

& 12

10

0-8
0-5 10 1-5 20 25 30
Height / D iam eter Ratio

Figure 4.8 Correction factor for height / diameter ratio of concrete cores

Using the relationship in Figure 4.8, and the height and diameter of the concrete core taken

from the site in Edinburgh:

Height/diameter of core =

250/100 = 2.5

Therefore, from Figure 4.8, the relative strength = 0.95

Therefore, the modified strength of core = 0.95 x 58 = 55.1 MPa

153
 -t — m e p iu ecu u ic iur ueierm inm g m e m -situ pertorm ance o t repair materials

This core strength requires conversion to cube strength. Kong & Evansm provide this

simple modification.

Cube strength = cylinder strength / 0.8

Therefore, for the substrate concrete at the Edinburgh site:

Cube strength of the substrate = 55.1 / 0.8 = 68.88 MPa

154
 vim ^iv. -t — in v jjiulcuuic iui u cicin u llin g me m -siiu perrormance o t repair materials

4.8 Development of properties

The development of the key properties in the repair material, with time, can now be

tabulated. These properties are: tensile strength, shrinkage, elastic modulus and the

resulting ‘strain transfer’ from the repair patch to the substrate. As the elastic modulus of

the repair material gradually increases with the curing period, it is desirable for this value

to become higher than the elastic modulus of the substrate concrete. In this way, a transfer

of shrinkage strain can occur from the repair material into the substrate, leaving a reduced

restrained shrinkage strain in the repair material (section 3.3). The amount of strain

transferred into the substrate is governed by the relationship58:

Eq. 4-5
0.0032

where X = shrinkage transferred (%)

Erep/Esub = ratio of elastic modulus of the repair material to elastic modulus of the

substrate.

Equation 4-5 has been developed empirically from wide ranging field data and

incorporates the cumulative effects of creep and shrinkage.

Table 4.10 uses the information presented in Chapter 3 to determine the development of

properties in the material Shucrete 1 when applied to a 3m square, 50mm deep repair patch

in Edinburgh in December. The patch is located on a reinforced concrete abutment and the

defect is within a full face area of substrate concrete (i.e. it does not continue around a

comer).

155
 ^nupiwi -r — inv, pivjv^culut iui u c i c i iiinmig me m-suu periormance or repair materials

Table 4.10 D evelopm ent o f properties with tim e (days) o f repair m aterial Shucrete 1 and transfer o f

shrinkage strain to the substrate

Time after F rep

repair Tensile Free Shrinkage strain
application strength shrinkage F sub transferred, X
E re p
(days) (Mpa) (microstrain) (Gpa) (%)
0 0.00 0.00 0.00 0.00 0

2 2.16 24.5 13.58 0.485 0

4 3.09 45.8 20.14 0.719 0

6 3.60 64.1 24.00 0.857 0

7 3.79 72.5 25.40 0.907 0

8 3.93 80.3 26.55 0.948 0

12 4.33 107.2 29.71 1.061 19

14 4.46 118.6 30.75 1.098 31

15 4.51 123.8 31.19 1.114 36

20 4.71 146.5 32.83 1.173 54

21 4.74 150.5 33.08 1.181 57

28 4.90 173.9 34.38 1.228 71

50 5.12 218.8 36.26 1.295 92

100 5.27 261.9 37.56 1.341 100

150 5.32 280.2 38.02 1.358 100

200 5.34 290.4 38.25 1.366 100

250 5.36 296.9 38.39 1.371 100

300 5.37 301.4 38.49 1.375 100

350 5.38 304.7 38.56 1.377 100

400 5.38 307.2 38.61 1.379 100

Note: Equation 4.5 which determines the percentage o f shrinkage transfer can yield both

negative values and values above 100%. Therefore, the m inim um practical shrinkage

transfer value is 0% and the maximum value is 100%.

156
 ^ n u ^ iw *t — i nt/ jjiulcuuic io i u c ie m iiiim g me m -suu periorm ance or repair materials

4.8.1 Consider day 14

In order to explain how the values in Table 4.10 were determined, the procedure used will

be demonstrated for day 14 (shown shaded in Table 4.10).

4.8.1.1 Tensile strength, day 14

In accordance with Table 4.1, the 28 tensile strength of Shucrete 1 is 5 N/mm2

Using equation 3-17:

ft _ '
fm 2.7975 + 0.9216i

Therefore, from equation 3-17, the tensile strength at any day can be calculated. Consider

day 14.

/ , = ----------- —------------* / „ „ = 4.46 MPa

' 2.7975 + 0.9216*14 '28

4.8.1.2 Shrinkage, day 14

The 28 day shrinkage strain of Shucrete 1 (modified to allow for climate) is 164

microstrain (section 4.6). This value represents the expected shrinkage of Shucrete 1 in

Edinburgh in December, where RH = 85% and temperature = 4°C. The size of the repair

patch has also been taken into account (3m x 3m x 50mm).

Using the equation 3-8:

£ _ t

s 2Z ~~ 12.292 + 0 .5017 /

157
 -r — my, p u t t u u i c iu i u c ic in u iiiiig me m -suu perrormance o t repair materials

The shrinkage at any day can be calculated. Consider day 14.

14
s = - — — — ---- ;— * s2S = 118.6 microstrain
12.292 + 0.5017*14

4.8.1.3 Elastic Modulus:

The 28 day elastic modulus of Shucrete 1 , in accordance with Table 4.1, is 34 GPa.

Using equation 3-20:

E2S 3.2637 + 0.8725/

The elastic modulus at any day can be calculated. Consider day 14.

E = ----------- —-----------* Elt = 30.75GPa

3.2637 + 0.8725*14

In addition to the determination of properties of Shucrete 1, it was shown earlier in the

section (equation 4-5), that the amount of patch repair free shrinkage which is transferred

into the substrate concrete can be determined as follows:-

Shrinkage transfer at day 14 using equation 4-5 gives:-

Modular Ratio = Em / Esub = 30.75 / 28 = 1.098

Where Enn = elastic modulus of Shucrete 1 at day 14.

ESUb= elastic modulus of substrate concrete (section 4.7)

Using equation 4-5:

0.0032

158
 -r — in v p u ^ u m c xui u ciciiiin iiiig uie m -suu periorm ance 01 repair materials

Therefore, on day 14, 30.7% o f the total shrinkage strain (after relaxation through creep)

will be transferred into the substrate (this value is rounded to the nearest integer in Table

4.10).

Table 4.10 shows the properties o f Shucrete 1 and the amount o f strain transferred from the

repair material into the substrate concrete at different ages from 1 to 400 days.

4.9 The effect of creep.

It was shown in Chapter 3 (section 3.3) that any tensile stresses which develop in the patch

repair due to free shrinkage of the repair material being restrained at its interface with the

substrate concrete will be relaxed through the action o f creep. The following section

outlines a procedure for determining the effect o f creep at a certain age in days.

4.9.1 Unit Creep

The unit (specific) creep of a material is the creep that occurs due to a loading o f 1 N/mm2.

As outlined in section 4.5, creep specimens tested in accordance with ASTM 512 are first

cured for 28 days, then loaded at a constant stress/strength ratio o f 30%. Therefore, the

calculation o f Unit Creep is not straightforward. Ordinarily, to calculate Unit Creep, the

total Creep will be divided by the total stress applied over the period o f loading. However,

the laboratory Creep test does not subject the specimen to a constant load as Figure 4.9

demonstrates.

159
 v^ndpiv. -r— inv piu tc u u ic iui uciciiuiiimg me in-snu perform ance ot repair materials

140
Load = 30% fc56
120
G Curing period
«* 100
o
?-i
P QO
B Line of'equivalent constant stress, load = 30% t'c42

Load = 30%
lc28

0 10 20 30 40 50 60

Time (days)

Figure 4.9 Life o f C reep sam ple from casting

Figure 4.9 shows how, at day 28, the creep specimen is loaded. On this day, the load

applied is equal to 30% of the 28 day compressive strength of the specimen material. This

applied load is increased periodically until the final load is equal to 30% o f the 56 day

compressive strength of the material.

It is common, when determining Unit Creep values, to simply divide the total creep by

30% of the 28 day compressive strength. But it is clear from Figure 4.9 that Shucrete 1

was subjected to a constant stress/strength of 30% throughout the 28 day testing period,

hence the applied stress would have increased with age in order to maintain the 30%

stress/strength ratio. It, therefore, needs to be determined if the common practice o f

determining unit creep by dividing total creep by 30% of the 28 day compressive strength

yields results that differ only negligibly from a more accurate practice of dividing total

creep by 30% of the average compressive strength o f the material over the 28 day test

period. To make the comparison, a mid-loading value for compressive strength is used, the

42 day compressive strength (represented by the line of ‘equivalent constant stress’ in

160
 -r — m e jjiuecuuic iui u eien m n in g m e m -suu periorm ance or repair m aterials

Figure 4.9). This line represents the average stress to which the laboratory specimen was

subjected over the 28 day period between day 28 and day 56.

Therefore, it can be assumed that Shucrete 1 achieved its 28 day creep value (the creep test

was conducted for 28 days) under a constant stress equivalent to 30% o f its 42 day cube

strength.

In accordance with Table 4.1:

fc28 = 30N/mm2

Therefore, using equation 3-11, the 42 day compressive strength o f Shucrete 1 can be

determined.

42
f =__________________* f
42
4.4994 + 0.8876*42
f =________42 *30
42 4.4994 + 0.8876*42
f c42 = 30.16N/mm2

Therefore, it is shown that 42 day compressive strength o f the material is just 0.5% higher

than the 28 day strength. Hence using the 28 day compressive strength to determine Unit

Creep is acceptable practice.

Therefore, the unit creep o f Shucrete 1 can be determined;

30% f c28 = 0.3 * 30 = 9 N/mm 2

From section 4.5.4, the 28 day creep value o f Shucrete 1 obtained at a 30% stress/strength

ratio and modified to account for climate, time o f application and specimen size is known:

Creep (modified) microstrain 488.2

Stress/strength ratio 30%

It is known that the creep specimen in the laboratory was loaded at a constant stress o f 9

N/mm2.

Therefore, the 28 day creep in compression at 1 N/mm 2 = 488 / 9 = 54.2 microstrain.

161
 ^napici •+ - m e piuecuux e xur determ ining m e m -situ pertorm ance o f repair materials

Where 54.2 microstrain is the modified specific creep value o f the material 28 days after

loading, having been loaded 28 days after casting. This figure requires further modification

in accordance with section 4.5.1, which recommended an increase in specific creep o f

183% to account for the fact that when the patch repair is applied, the repair material is

subjected to stress immediately, i.e. at age 0 day.

Therefore, the 28 day creep in compression at 1 N/mm 2 for material subject to immediate

loading after the application o f the repair patch, Cu28:

54.2 * 1.83 = 99.3 microstrain

4.9.2 The effect of creep at day 2:

In order to calculate the creep in Shucrete 1 at 2 days after repair application, firstly the

stress it has been subjected to must be established.

On day two, from Table 4.10, the free shrinkage o f the material would be 25 microstrain.

However, at the repair / substrate interface the repair material, Shucrete 1, is unable to

shrink freely and tensile stresses develop. Hence the stress in the material at day 2 can be

simply determined as:

~ £shi * Erep2
= (24.5*10“6)(l3.58*103)

= 0.333Af/mm 2

Where a 2 = tensile stress in repair material at day 2

sSh2 = free shrinkage of repair material at day 2

Erep2 = Elastic modulus o f repair material at day 2

The values o f sSh2 and Erep2 are given in Table 4.10.

162
 -r - m e p iu ecu u ic lux uexexrnining m e m-siru periorm ance o l repair materials

After repair application, tensile stresses in a repair patch (at the interface) gradually

increase with time as the repair material continues to shrink and the elastic modulus

increases. However, constant stress values are required to calculate creep at each age.

Hence, at day 2, it can be assumed that the equivalent constant stress the material has been

subjected to during its in-service lifetime is equal to half o f the stress at day 2 .

-£ l
(JeC2 ~~ f
=0.333/2 =0.166N/mm7

Where oec 2 = equivalent constant stress applied to material over its first two days in

service as a patch repair,

t = number of days repair material has been subject to equivalent constant

stress

/.Equivalent constant stress in repair material at day 2 = 0.166 N/mm2.

It is, therefore, assumed that the material will exhibit identical creep strain on day 2 as if it

had been loaded with a constant stress o f 0.166 N/mm 2 from the time o f repair application.

The unit (specific) creep value determined for Shucrete 1 earlier in this section corresponds

to an applied stress of 1 N/mm 2 for 28 days. It was shown in equation 3-5 (hyperbolic

expression) that the unit creep value o f a repair material at any age can be obtained by

using the 28 day unit creep value o f the material. As the unit creep o f Shucrete 1 at 28 days

age is known, the actual creep (in microstrain) expected in the material at 2 days age can

be determined using the following expression:

C2 = <^'ec2 ^'R2 ^'U2 s Eq 4-6

Where C2 = Creep o f material at day 2

163
 ‘t — m e p iu ecu u ic lur ueierm m m g m e m -suu perform ance o f repair materials

Cr2 = Ratio of unit creep at 2 days age over unit creep at 28 days age (from

equation 3-5)

Cu28 = Unit creep in repair material at 28 days age after application

The expected Unit Creep (Cu2s) of 99 microstrain is the creep which would occur over 28

days at a constant stress o f 1 N/mm . Multiplying this value by the ratio between unit creep

at 28 days age and 2 days age (equation 3-5) will give the creep which would occur over 2

days at a constant stress of 1 N/mm . This in turn, when multiplied by the average constant

tensile stress the repair material has been subjected to between day 0 and 2, becomes the

amount of creep that has occurred in the material at day 2.

We can obtain the ratio o f 2 day creep to 28 day creep by using this relationship (equation

3-5):

C2 _ t
C28 ~ 7.0162 + 0.701/1

Therefore,

C
_2_ =
2
= 0.238
C28 7.0162 + 0.701*2

Using unit creep in this way assumes that creep characteristics are the same in compression

and tension (section 3.2.3.4.1)

Therefore, using equation 4-6:

C2 = 0.166 * 0.238 * 99.3 = 3 .9 microstrain

Therefore, if the total shrinkage strain o f the material is restrained at the interface, then the

net tensile strain after relaxation due to creep becomes:

Strain at day 2 = shrinkage at day 2 - creep at day 2

164
 ^n ap ici ‘t - m e p iu ecu u ie lor ueierm m ing m e m -situ perform ance o t repair materials

= 24.5 - 3.9 = 20.6 microstrain.

However, if, as determined above, the actual strain in the repair material at day 2 is 20.6

microstrain, then the actual stress will be different from the 0.166 N/mm 2 assumed earlier.

This stress, and subsequently the other variables can thus be recalculated iteratively.

<J 2 = E s l , 2 * E rep2

= (20.6*10'6)(l3.58*103)

= Q21ZN I mm2

<JEC2 —

= 0.278/2 = 0.1397V7 mm2

Thus, a more accurate value of the creep at day two can be ascertained:

^ 2 = <JE C2^'R 2^'U 2%

C2 =0.139 * 0.238 * 99.3 = 3.3 microstrain

Strain at day 2 = shrinkage at day 2 - creep at day 2 = 2 4 .5 -3 .3 = 21.2

microstrain.

Again, the actual strain at day two is different from that assumed, hence this process is

repeated iteratively until values for creep, strain and stress show no significant change

during further iterations. At that stage the value o f the strain represents most accurately the

actual performance o f the repair material.

Generally the strain after a series of iterations is between 90% and 99% o f that before the

iterations. A high creep value reduces the strain by 10% through the series o f iterations;

though a lower creep has little effect.

165
 -r — m i. p u t w u i c u ciciiiiu im g me m -suu perrormance 01 repair materials

The final step required to complete the process o f understanding the performance o f the

material at day 2 is to calculate the actual stress at day 2 after the relaxing effect o f creep

(as distinct from the average stress, which is calculated to aid determination o f creep).

/-r — p % J7
2 t i sh2 rep2

= (21.1*10“6)(l3.58*103)

=0.287NI mm2
After the first iteration, the strain in the repair was 21.2 microstrain. When computerised, a

series of iterations were performed speedily and the final strain was 2 1 .1 microstrain

(shown in the above calculation).

4.9.3 The effect of creep at day 4:

The procedure for calculating the effect o f creep on the strain at day 4 is similar to the

above example for calculating creep at day 2 , except for the method o f calculating the

average stress in the material.

For the purposes o f deriving a general method applicable to all other ages, day 4 represents

the age at which the amount o f creep is being determined (the current age), and day 2

represents the previous age at which the amount o f creep was determined (the previous

age) i.e. the incremental step from day 2 to day 4. In this way any non-linear increase in

stress can be allowed for.

166
 ‘t - iiic yiu u cu u ic iur ueierm ining m e m -situ periorm ance o t repair m aterials

~ ^ 4 ( ^ j/j4 ^2 ) Eq. 4-7

A cr2,4 =& 4 ~or2 Eq. 4-8

Eq. 4-9

( ^ E C 2 * ^o-l ) + ( ^ E C 2,4 * K l )
Eq. 4-10
fa\ + ta2

Where 0 4 =stress in repair material at 4 days age

Ac?2,4 = increase in stress from age 2 days to age 4 days

0 ec 2,4 = Equivalent constant stress in repair material between days 2 and 4

oec 2= Equivalent constant stress between time o f repair

material application (day 0 ) and previous age (day 2 )

tai = Number o f days under which the repair material endured associated stress

ta2 = Number o f days under which the repair material endured associated stress

C2 = final creep at previous age (day 2 ) after the series o f iterations

oec 4= Equivalent constant stress between time o f repair

material application (day 0) and current age (day 4)

In Equations 4-7 to 4-10 the creep at day 2 (C2) is considered to be the creep that has

already occurred and hence it can be immediately deducted from the known shrinkage at

day 4.

Figure 4.10 shows how, in effect, the average equivalent constant stresses from days 0 to 2

and from days 2 to 4 are further averaged to find the equivalent constant stress from days 0

to 4.

167
 i nv ^ iu ^ u u ic iui licicium iiiig me m-suu penorm ance or repair materials

T im e (d a y s)

Figure 4.10 D eterm ining equivalent constant stress

Using the data in Table 4.10 and substituting into Equation 4-7 gives:

cr4 = 20.14 * 10 3 ((45.8 - 3.4) * 1O'6) = 0.853 /

where C 2 (creep after iterations) = 3.4 microstrain

Substituting for (7 4 and G2 = 0.287 N/mm 2 (from section 4.9.2) into equation 4-8 gives:

Acr2 4 = 0.853 - 0.287 = 0.56377/ mm1

substituting for Ag 2,4 in equation 4-9 gives:

cr£ C 2 4 = 0.287 + _ ^ £ . = 0.572N / mm2

Substituting for gec 2 ,4 into equation 4-10 gives:

(0.166 * 2) + (0.572 * 2)
c EC4 = — = 0 3 7 N /mm'

Where 0.166 N/mm 2 = the equivalent constant stress from days 0 to 2 (section 4.9.2)

168
 1& mv 111-JIIU pv,l iuilliaiiv^c u i icpclll Illcliei IcllS

Determining the creep of Shucrete 1 at a constant stress of 0.37 N/mm 2 will give an

accurate representation o f the actual creep which has occurred under the varying stress to

which the repair has actually been subjected over the period of four days after application.

This stress can now be used to calculate the creep at day 4, and subsequently the strain.

This strain can then be used to determine the reduced stress (due to relaxation) and the

process o f determining the actual creep can be continued iteratively until a stable value is

found.

When conducting the iterations, the creep term from the previous time increment, C2 (in

the case above, the day 2 creep o f 3.4 microstrain) is not subtracted from subsequent

iterations as in the first operation - as its effect has already been included (that operation is

only performed in the first calculation - deducting the creep already known to have

occurred reduces the number of iterations required).

Hence, creep in the repair material at 4 days age, C4, is given by:

C 4 - CTEC4 * C r 4 * Cu28 Equation 4-11

Where gec 4 = the equivalent constant stress from ages 0 to 4 days from Equation

4-10.

C4 = creep in repair material at 4 days age

Cr4 = Ratio of unit creep at 4 days age over unit creep at 28 days age (from

equation 3-5)

Cu2 8 = Unit creep in repair material at 28 days age after application from

section 4.9.1

Substituting into Equation 4-11 gives:

C4 = 0.31 * CR4* Cu28

4
= 0.407 (equation 3-5)
7.0162 + 0.701*4

169
 C4 = 0.31 * 0.407 * 99.3

C4 = 12.6 microstrain

The values o f Equivalent constant stress and Creep in the above equations are those

obtained after a series of iterations (highlighted in green in Table 4.11).

When the process o f iteration no longer significantly changes the values o f creep and

stress, the final actual strain at day 4 can be determined:

Strain at day 4 = shrinkage - creep = 4 5 .8 -1 2 .6 =33.2 microstrain

4.10 Transfer of strain to the substrate.

4.10.1 Consider day 14

On day 14, from Table 4.10:

csh = 118.6 microstrain

Using the procedure outlined in section 4.9, the creep strain at 14 days, Cj4, is calculated

by a series of iterations to give:

Ci4 = 73.6 microstrain

Therefore, net Strain in the repair = 118.6 - 73.6 = 45 microstrain

Table 4.10 shows that, on day 14, using equation 4-5, 30.71% o f the shrinkage strain in the

repair material is transferred into the substrate concrete. Therefore:

Strain transferred to the substrate = 45 * 0.3071 = 13.80 microstrain

This amount of shrinkage strain is transferred from the repair material into the substrate

concrete, therefore:

170
 4---- - j--u v iv iim n iiig mw iii-onu p u iiu im a lic e u i icpcin m aterials

Residual restrained shrinkage strain in the repair material at day 14

= 4 5 - 13.80 = 31.2 microstrain.

The maximum tensile strain in the repair material (at the interface with the substrate) at

each day is calculated using the method outlined above and is plotted alongside the tensile

strain capacity of the repair material such that the performance o f the material can be

demonstrated graphically (a typical example is given in Figure 4.11). Using the equations

that describe the development o f repair material properties with age (equations 3-18 and 3-

2 0 ), the tensile strain capacity with age is calculated and is plotted graphically up to a

period o f 300 days. The maximum tensile strain developed in the repair material due to

restrained shrinkage and creep is also plotted in Figure 4.11. Due to the necessity for

accounting for creep relaxation in the calculations (for maximum strain in the material), the

increments in age at which creep is calculated have to be regular and small. This means

that the technique used for predicting the tensile strain in repair patches requires a

computer program due to the large amount o f iterative calculation necessary.

The software was developed to resolve all the necessary equations, producing an output o f

two data sets: the tensile strain capacity of the repair material, and the restrained shrinkage

tensile strain that occurs in the material. The software compares these sets o f data and can

inform the user if the tensile strain capacity o f the repair material is exceeded. If the tensile

strain capacity is exceeded, the time after application when the material will fail by

cracking is also graphically determined.

Table 4.11 presents the performance o f material Shucrete 1, at selected ages, over a 400

day period as provided by the calculations using the computer program.

The row ‘stress 1’ is highlighted to indicate that the calculation has included a deduction

for creep which is already known to have occurred at the previous age at which

calculations were performed. For example, at day 8 , to determine ‘stress 1’, strain in the

171
 ~ uvLv/umuiiig uiv^ ni-aiiu pci iu i niciiiue in re p a ir mareriais

repair material is taken as the known shrinkage (from Table 4.10) minus the creep

calculated at the end o f day 7 (31.89 microstrain). This deduction is only carried out for the

determination of ‘stress 1’ - the first cycle o f a number o f iterations. The penultimate row

is highlighted to demonstrate that the figure includes strain transferred to the substrate

concrete.

172
 Chapter 4 - The procedure for determining the in-situ performance of repair materials

a3
o

E-
E
a
a.

«
o
«
a

u
S
ui

CJ
B
u
o
o
o
0

O
O

m
m
m

r-

so
00
oo
Day

CN
CN
CN
200 250 300 350

CM
CM

---1
T—1
. -.

:■-s-

d
d
d
d
d
p
d
d

O
in
in

so
in

co
1 ! in

co
in
in
in

co
co

oo
so

co
co
co

Os
oo

o
co’
IN-

Tensile strength (N/mm2)

CN
CN
<N
5 .2 7 5 .3 4

CN

ON
CO
^*

d
o
d
p

O
VZL O
CM

so

00
co

oo
00
oo
co

o
o
SO
N"

Shrinkage (microstrain) 2 4 .6 4 5 .8

CN
1 0 7 .2 1 1 8 .5 1 2 3 .8 1 4 6 .5 1 5 0 .5 1 7 3 .9

CN
CN
SO
2 9 6 .9 1 3 0 1 .4 0 3 0 4 .6 8
ON

OO
CM
CM

2
z
d
d
d
p

w
1
a
co
oo
co

rP
o
o
2 5 .4

SO
2 4 .0

CN
2 6 .6 2 9 .7 3 0 .8

CO
CO
CO
CO
3 6 .2 6 3 7 .5 6 3 8 .0 2 3 8 .4 9

CN
CN
3 8 .2 5 3 8 .3 9 3 8 .5 6

CO

0
p
d
d
o

d
d
p
d
d

U
©

in
in
r7

in

so
so
t>

co
or
co
co

oo
00

00
oo

0 .0 0 0 .2 4 0 .5 9

CM
CO
CO
0 .8 6 0 .9 7

CO
ON
CO
CO

ON
ON
T1

o
o
o

o
o
o
o
o
o
o

d
d
o

d
d
d
o
o
o
o
o
o
o
o
o
o

d
o
o
o
o

d
d
d
d
d
V

ci
co
r--

o
o
o
Shrinkage transfer (% ) 3 5 .6 1 5 3 .8 9 5 6 .6 6 7 1 .1 6 9 2 .1 4

SO
_!

ON
1 0 0 .0 0

d
o
o

d
d
p
d
d
p

m
in

cn
in
in
in
in

00
so

N-

o
o
Restrain shrinkage stress 0 .1 7 0 .3 6 0 .5 6

<N

CN
0 .9 0

CO
CO

ON
ON
CN
SO

ON
i i

OO
CO

T<

o
o
o
o

p
o

p
o
o

n
n
p

©
O

in
of
r-
of
in
in
in

oo
co
in
in

of

(N
co

o
IN
IN

o
o^

3 .9 4
>n

CO
IN
1 1 IN

creep

n-

CN
CO
CN
SO

ON
CN
CN
CN
CO

CN
1 4 7 .8 1 1 7 7 .8 5 1 9 3 .9 5

ON
ON
CO
CO

CO
CN
CN
CN
CN

CO

CM
CM
CM

o
o
o
o

p
o

©
©
©
m
©

-f
N

so
of
in

so
in

co
co
of
in
of

oo
oo
oo
co

oo

o
o
-

SO
IN

CN
CO
IN

SO
CO
Strain (shrinkage creep)

CO
CO

ON
4 4 .4 5 5 2 .5 0 5 0 .1 4 7 0 .9 9 8 6 .2 9 8 6 .8 9 8 7 .1 5
CO
CO

CM
o
o
o
o

p
n
p

©
©
©
©
©
9V0

in
r-
of
in
in
in

os
in

so
so

oo
oo
co
oo

r--

>n
oo

o
o
n-

0 .1 4

CD
CO
CD
stress 1*

CD
CN
CN

ON
r—1

p
p
n

\V 9 L

©
©

O
in
m

of
in

so

o
n-
(N
O'!

N"

CO
CO

CD
CN
CN
CN
CN
creep 1 3 1 .8 2 3 8 .4 5 5 9 .2 2 7 2 .0 5 9 2 .1 4 9 7 .7 5
-

CN

1 9 3 .8 4 2 0 3 .4 9 2 0 9 . 8 0 2 1 4 .2 3 2 1 7 .5 0
ON
o
p

P
n
n

so
so
so
so

N-
N-
oo
oo

0 .0 0 2 1 .3 0 3 3 .5 5 3 9 .5 1

CD
CO
strain 1 7 2 .3 6

ON
4 6 .4 9 4 7 .3 9 5 4 .3 8 5 2 .7 1

ON
8 4 .3 0 8 6 .9 4 8 7 .1 1 8 7 .1 8
CO
CO

d
d
d
p

T
T

O
O

in
in
in

so
in
in

so
so

00
oo
cn

oo
oo

(N
m.
oo

—i

o
o
'U-

N"

0 .1 4
—

0 .4 8
o’

CO
stress 2 0 .9 0
i—i

CD
CO
CN
CN
CN
r—

H
H
1—1

d
p
p
p
ZVZL
K
O

in
in

so
r-
r-
m

rn
oo

o
o
CO
CO

CN
CO

CN
CO
creep 2 6 4 .1 2 7 6 .4 4 9 5 .9 9
CO

1 4 6 .7 3 1 7 7 .5 9 1 9 3 .8 5 2 0 3 .5 0 2 0 9 .8 0 2 1 4 .2 3
ON
CN
CN
CN

T—H

d
o

d
p

O
K

co
ON
ON
so

oo
oo
co
sd

rn

'-f-

o
N"

CO
CO

CN
strain 2

CN
SO
CO

5 0 .5 3 5 2 .7 0 5 6 .5 6 7 2 .0 6 8 4 .2 7 8 6 .3 9 8 6 .9 4 8 7 .1 1 8 7 .1 7 8 7 .1 8
OO

o
o
o
o
o

d
d
d
d
d
d
d
p
p
p

m
in
so
in
in

so
so

oo
oo
oo
oo

Os
oo

N-

o
o
0 .1 4

CO
0 .4 7
CO

CN
CN

stress 3
T—
N“
1
r-H

—
o
o

d
p
p
d

r-

r-
in
in
r-
r-
in
in
in

co
r6
co

oo
oo
cZ

o
o
On]

CO
« 1 1 4 .9 7 1 4 6 .6 7 2 0 9 .8 0 : 2 1 7 .5 0 :
CO

ON
CN
CN
CO

ON
5 2 .2 4 ' 7 2 .3 5 ' 7 6 .4 4 94.50
CN
CN
CN

ON

creep 3
CO

OO
O1
 Chapter 4 - The procedure for determining the in-situ performance of repair materials

o
o
CN
in
oin

r— <

r>
oo
Day

cn
'd*

cn
28 50 200 250 300 350

*d*
<N

VO
o .
:

o
o
o
in
in

oo

strain3 21.21 33.19 38.97 40.55 41.72 44.94 46.19 47.36 52.02 52.70 58.90 72.13 84.27 86.39 86.94 87.11 87.18
OO

o
o
o
o

r —H
00‘0

T—-H

C
O
O

in
in
in

in
in

m
in

rn
N-

N-

oo
oo
vo
oo

stress 4 0.14 0.47

CD
ON
CN

VO
O
OO
On
VO

o
o
o
o
O

N;
cn
vd
creep 4

Ov
12.59 25.20 31.89 38.53 62.96 72.37 76.44 95.07 146.68 177.58 193.85 203.50 209.80 214.23 217.50

IN
o IN
CN IN
<N
CN

o
o
o
o
o
o

o
o
o
o
o
o
o
o
o
o
o
o

©
N
©

m
cn
strain 5 21.21 33.17 38.94 40.55 41.72 35.79 32.00 23.72 22.84 16.68 0.00

T
r—

o
o
o
o
o

o
o

CD
O
O
O
O

1
©
©
©

O)
cn
cn
vo

oo

oo
Actual stress

Ov
0.29 0.67 0.57 0.00

IN
IN

OS

CD
CD
CD
CD
CD
in CD
vq CN

CD
CD

CD
CD

VO
1

*
IJ-t
P

C/3

w
w
o

23
'■a

u
u
cd

< )
D,
<u

<L
X

D
bO

u
CN

cd
cd
cd

oo

a,
-i—>
Td

h
Vu
a<>
00
1^1

<L>
r-n-
4.11 Tensile strain capacity

A simplistic strain capacity value can be obtained using tensile strength and elastic

modulus.

Therefore on day 4, using the values for Shucrete 1 given in Table 4.10:

_ 3.09*106
£cap4 ~ 20.14 * 103
scaP 4 =153 microstrain
This tensile strain capacity determined at each age is shown in Table 4.12:

Table 4.12 Development o f Tensile Strain Capacity in Shucrete 1 (microstrain)

day 0 2 4 6 7 8 12 14 15 20

tensile strain capacity 0 159 153 150 149 148 146 145 145 144

day 21 28 50 TOO 150 200 250 300 350 400

tensile strain capacity 143 142 141 140 140 140 140 140 140 139

The data presented in Table 4.11 and Table 4.12 represents the performance o f the repair

material. It can be used to determine if Shucrete 1 is a suitable repair material for the repair

patch in Edinburgh in December. The development o f tensile strain in the repair material

with time (Table 4.11 - penultimate row) can be now be plotted alongside the development

of tensile strain capacity in the repair material with time (Figure 4.11).

175
Chapter 4 - The procedure for determining the in-situ performance of repair materials

o
o
cn

Q. o
in
04

a.
Q-

o
o
04
vo
O'

T&
D
<o
£

o
o

cn
o
in

o
o o o o o o o o o o
o o o o o o o o o o
o
oo o o'
^r o o'
o
©
oo o' © o o
VO 04 vo 04
a*
(lIIEJJSOJOILLl) UI43JJS 3
OX)
Figure 4.11 shows that using Shucrete 1 for this repair situation would result in a

successful repair. The tensile strain capacity o f the material will not be exceeded by the

tensile strain that arises through restrained shrinkage. Therefore, the material will not

crack. It will be visually amenable and it will protect the reinforcement it covers. It will

also be able to share load with the surrounding substrate.

The strain which occurs in Shucrete 1 due to restrained shrinkage is represented by the

blue line in Figure 4.11. Typically, for any repair material and substrate combination, as

the material begins to shrink, it is restrained at the interface by the substrate, this restraint

causes tensile stress in the repair. As a result of this stress, a natural relaxation through

creep occurs. In the theoretical example detailed in this chapter, the shrinkage strain (and,

therefore, restrained shrinkage tensile stress) continues to grow up until day 15. Around

day 15, the developing elastic modulus o f the repair material has become as stiff as the

substrate concrete. When this happens, the shrinkage strain in the repair material can begin

to be transferred into the substrate concrete. Initially, strain is transferred from the repair

material into the substrate in relatively small amounts, then more and more o f the strain is

transferred until, when the elastic modulus o f the repair material is 1.32 times the elastic

modulus o f the substrate concrete, all the strain in the repair material caused by restrained

shrinkage is transferred into the substrate concrete. On day 50, the modular ratio has

reached the optimum value o f 1.32, and all restrained shrinkage is transferred.

There are a number o f factors which could have caused Shucrete 1 to fail: a lower tensile

strain capacity, a lower elastic modulus, or the substrate concrete having a higher elastic

modulus.

177
 — ,— . uuiw im um g me iii-buu periorm ance or repair materials

4.12 Estimation of Creep using shrinkage data

4.12.1 Introduction

Often, when selecting repair materials, only limited information o f their properties will be

available . The property that is least likely to be provided by the manufacturer is ‘creep’.

As discussed in this chapter, creep will relax any strains that appear in a repair material,

and all concrete repair materials will exhibit the beneficial effects o f creep to some extent.

Where no information on creep is provided by a repair material supplier (which is

frequently the case), the software created in the project will make a conservative estimate

of the creep, based solely on the shrinkage properties o f a repair material - which generally

will be provided by the manufacturer.

Data on the relationship between creep and shrinkage has been collated from eight sets o f

references in literature. For each reference, it was necessary to check if the tests to

determine shrinkage and creep were performed under similar conditions. If this was not the

case, modifications are made to the results. The preferred conditions are 28 day Shrinkage

determined at 20°C and 55% relative humidity; creep specimens cured for 28 days, then

subjected to 28 days o f loading at a stress/strength ratio o f 30%.

It will also be noted if the repair material was polymer modified.

4.12.2 O'Flaherty58

This data was obtained using the same materials for which the software routines in the

previous chapter were determined. As such the values were obtained under the

recommended conditions of:

28 day Shrinkage at 20°C and 55% relative humidity.

178
 — I— -— ^v.^v.v. 1 v u v iw iiim u ic, Lii^ iii“3iiu p c i iu i lucincc u i re p a ir m ate rials

28 day Creep at 30% stress/strength ratio.

The creep and shrinkage data are listed in Table 4.13:

Table 4.13 Creep and free shrinkage data for repair materials58

Shrinkage Creep Polymer

Material 'microstrain) (microstrain) modified
G1 560 294 yes
G2 1200 342 yes
G4 310 546 yes
G5 1030 1088 yes
G6 920 1070 yes
LI 500 680 -

L3 320 580 -

L4 580 428 -

L5 450 434 -

SI 600 408 -

S2 630 456 -

S3 440 586 yes

S4 270 368 -

See Table 3.13 for more detailed information about the constituents o f the repair materials

listed in Table 4.13.

4.12.3 Mangat & Limbachiya56

The data given in Table 4.14 were obtained under the recommended conditions of:

28 day Shrinkage at 20°C and 55% relative humidity.

28 day Creep at 30% stress/strength ratio.

179
 .P lu m in g me ui-biiu periorm ance or repair m aterials

Table 4.14 Creep and free shrinkage data for repair materials56

Shrinkage Creep Polymer

Material (microstrain) (microstrain) modified
A 320 500 -

B 300 550 -

C 440 1000 yes

Material A was a blend o f Portland cement, graded aggregates (maximum size 5mm) and

additives to impart controlled expansion. Material B was a mineral based cementituous

material with no aggregate sized particles or additives. Material C was a single component

cementituous mortar incorporating microsilica, fibre reinforcement, and styrene acrylic

copolymer.

4.12.4 Mangat & Azari106

The data listed in Table 4.15 were obtained under the recommended conditions.

28 day Shrinkage at 20°C and 55% relative humidity.

28 day Creep at 30% stress/strength ratio.

Table 4.15 Creep and free shrinkage data for repair materials106

Shrinkage Creep Polymer

Material (microstrain) (microstrain) modified

ao 400 620 -

180
 j— . .— v/ xwi u v im im u iiig uio m -aiiu p cu u x iim n u e or repair materials

4.12.5 Evans 107

The data listed in Table 4.16 were not obtained under the recommended conditions. They

require modification. The data was obtained under the following conditions:

Between 193 and 200 day Shrinkage at 20°C and 55% relative humidity.

28 day Creep at 25% stress/strength ratio.

Table 4.16 Creep and free shrinkage data for repair materials107

iStress strength ratio

!creep at 30% stress

Creep strain at age

Polymer modified
Estimated 28 day

Estimated 28 day

I Estimated 28 day
Shrinkage at age

strength ratio
(microstrain)

(microstrain)

(microstrain)

(microstrain)
Creep strain

(microstainl
Shrinkage

shrinkage
(days)

(days)

creep

(%)
Mat.

1
Col. 1 Col. 2 Col. 3 Col. 4 Col. 5 Col. 6 Col. 7 Col. 8 Col. 9 Col. 10
0 870 197 491 599 90 467 25 560 Concrete

15-1 796 200 448 824 90 642 25 770 yes

25-1 703 199 396 936 90 729 25 875 yes
25-2 806 197 455 580 90 452 25 542 yes
25-3 801 197 452 627 90 488 25 586 yes
25-4 660 193 373.2 515 90 401.2 25 481.40 yes

Material 15-1 achieved its free shrinkage o f 796 microstrain at 200 days (Table 4.16). The
200 day shrinkage value is converted to the required 28 shrinkage (S28) using Equation 3-8

in section 3.4.2.4:

s =g /
r e2% 12.292 + 0.5017/

796 _______ 200_______

~ 12.292 + 0.5017*200
e1%= 448 microstrain

181
 — I— . — _ ^ xw i u v i^ n m iiL iig m e lu - M iu j j c i iu n iian cc o i repair marenais

This 28 day value o f shrinkage is given in Table 4.16 (column 4 - material 15-1). The

material achieved its creep value of 824 microstrain at 90 days. This was converted to the

required 28 creep using Equation 3-5 in section 3.4.2.2 as follows :-

C t
C28 7.0162 + 0.701/

824 90
C28 7.0162 + 0.701 * 90

C28 = 642 microstrain

This value is given in Table 4.16, column 7.

This Creep value was achieved at a stress/strength ratio of 25%. This figure requires

further modification to determine creep at 30% stress/strength ratio. Using a linear

relationship between creep and stress/strength ratio (section 4.5), gives :-

642
^ 28 :30% 5 / 5 = 250 / * ^0 %

c 28:3oo/0 = 770 microstrain

Where C28:30%s/s is the 28 day creep achieved at a 30% stress strength ratio.

108
4.12.6 Limbachiya

The data listed in Table 4.17 were obtained under the recommended conditions of:

28 day Shrinkage at 20°C and 55% relative humidity.

28 day Creep at 30% stress/strength ratio.

182
 1— . i lvji u v iu im iiiiii^ m e m-Miu p c n u in u u ic c 01 re p a ir m aterials

Table 4.17 Creep and free shrinkage data for repair materials108

Shrinkage Creep Polymer

Material (microstrain) (microstrain) modified
qa 530 500 -

qb 630 530 -

qc 860 1010 yes

4.12.7 Poston, Kesner, McDonald, Vaysburd68,76

The data are listed in Table 4.18 and were not obtained under the recommended conditions.

They require modification. The curing conditions o f samples were:

28 day Shrinkage at 20°C and 50% relative humidity.

Creep specimens were loaded for one year and results given as specific creep (creep per 1

lb/in2)

183
 p v w u m v iui uv/iuiinuiiig me m-biiu p cn o n im n ce or repair materials

Table 4.18 Creep and free shrinkage data for repair materials76

Shrinkage Shrinkage
50%RH 55%RH 28 day creep Polymer
Material (microstrain) (microstrain) (microstrain) modified Material
1 178 165 568 - Portland cement mortar
2 201 187 1026 - Portland cement concrete
3 339 315 2474 yes Polymer modified concrete
4 293 272 1342 - Portland cement concrete
5 305 283 870 - Portland cement mortar
6 429 398 1426 yes Polymer modified mortar
7 479 445 2892 yes Polymer modified mortar
Polymer and fibre modified
8 391 363 1773 yes
mortar
9 429 398 1218 - Portland cement concrete
10 1779 1652 2015 yes Polymer modified mortar
11 301 280 704 - Portland cement concrete
12 258 240 1928 yes Polymer modified mortar

Conversion of shrinkage

In accordance with section 4.4 for materials cured below 70%RH, each 1% reduction in

RH will increase shrinkage by 2%. The shrinkage specimens were cured at a relative

humidity o f 50%. A 20 percentage increase would bring this value up to the datum relative

humidity o f 70%, and result in a reduction o f 40%.

For material 2 (Table 4.18):

201
£ R H = 70% “ 1 4

£ rh %=144 microstrain
= 70

Furthermore, a 15 percentage point reduction in relative humidity would provide the

required relative humidity o f 55%. This 15 percentage point reduction would effect an

increase in shrinkage o f 30% (in accordance with 4.4).

184
 ~ vvvv.iiinim g m -oiiu p tn u m m iiu c u i re p a ir m aterials

s RH=55%
=144*1 3
£ rh = 55% =187 microstrain

Conversion o f creep

Table 4.18 provides the creep at one year for specimens subjected to a constant loading o f

1 lb/in2.

For material 368:

1 year specific creep at 1 lb/in = 1 .8 microstrain

1 year specific creep 1 N/mm2 (Cu36s) = 1.8 * 145 = 261 microstrain

Where 145 is the factor to convert from creep at 1 lb/in2 to creep 1 N/mm2

In accordance with Equation 3-5 in section 3.4.2.2:

CU365 3 65
CU28 (7.0162+ (0.701 *365))
0/365
—1/365 =261
= 26 microstrain
Therefore,
o _ 261
U28
365/(7.0162+ (0.701 *365))

Cu28 = 188 microstrain

fc2868= 6360 lb/in2

fc28 = 6360 / 145 N/mm2

fc2 8 = 43.9 N/mm2

Therefore:

The 28 day creep at 30% stress/strength ratio, C 28:3o%:

C28 = (0.3*43.9)* Cu28

C 28 = (0.3*43.9)* 188

C 28 = 2474 microstrain

185
4.12.8 Emberson & Mays61

The data are listed in Table 4.19 and were not obtained under the recommended conditions.

They require modification.

495 day Shrinkage at 20°C and 50% relative humidity.

495 day Creep specimens.

Table 4.19 Creep and shrinkage data for repair materials61

Estimated Estimated
Shrinkage Shrinkage @28 Creep Creep
@ 495 days days @495 days @28 days Polymer
Mat. (microstrain) (microstrain) (microstrain) (microstrain) Modified
D 200 105 230 165 yes
E 420 221 1600 1144 yes
G 440 232 400 286 -

H 280 147 960 687 -

I 210 111 580 415 yes

Material D was an SBR-modified cementituous mortar, E a vinyl acetate modified

cementituous mortar, G an OPC/sand mortar, H a high allumina cement mortar and I a

flowing concrete.

4.12.9 Neville104

This reference provided the creep versus shrinkage relationship o f 52 concrete mixes o f

normal strength. Some values were not obtained under the recommended conditions:

365 day Shrinkage at 20°C and 50% relative humidity.

365 day Creep at 50% s/s ratio.

186
 i5 me ln-snu p ciiu m ia n ce or repair materials

Data from this reference was modified in accordance with the procedures used in this

section. Data from the eight references described above are plotted in Figure 4.12, which

should be read in conjunction with Table 4.20.

187
 Chapter 4 - The procedure for determining the in-situ performance of repair materials

s/s o/0o£ ‘sXep 83 ‘(urejjsaioxui) dsaio

T able 4.20 Key to Figure 4.12

Symbol Material Type Reference

Repair materials O ’Flaherty38

O
Polymer modified Cf Flaherty38
0
repair materials

■ Repair materials M angat & Lim bachiya36

■ Repair material M angat & A zari1U6

Concrete E vans 10'

+
Polymer modified E vans107
©
Repair materials

Repair materials Lim bachiya108

X
Polymer modified Lim bachiya108
0
repair materials

• Repair materials Poston et al68

Polymer modified Poston et al68

®
repair materials

Repair materials Emberson & M ays61

A
Polymer modified
repair materials Emberson & M ays61

♦ Concrete N eville104

The red line in Figure 4.12 will be used to determine the creep properties o f a repair

material when only the shrinkage properties are known. It is not a line o f best fit but has

been positioned so to provide a conservative estimate o f creep (because the relaxing effect

o f creep is beneficial in reducing tensile stress which arise in repair patches). If a line o f

best fit is added to the graph, a coefficient o f variance in the region o f r2 = 0.4 is the result.

This shows that the link between shrinkage and creep is, at best, tenuous - especially when

all the data is considered. In particular, the polym er modified repair materials from Poston

et al68 exhibit creep at levels far higher than the prescribed relationship would predict. It is
 j— . ,-------- w . uviviiHiiung mw m-oiiu pci xunnaiiL.c u i icpiur iiicuericiis

of note that polymer modified materials from the other references do appear to follow the

prescribed relationship with a reasonable degree of accuracy. On balance, Figure 4.12

shows that the prescribed relationship between creep and shrinkage is adequate for the

purpose it is required for. Therefore, in accordance with Figure 4.12, the minimum creep at

28 days due to a 30% stress/strength ratio will be estimated using this relationship (shown

by the red line)

Creep = (Shrinkage* 1.1 )+20 Equation 4-12

Clearly, on occasions, some materials will exhibit a creep greater than the equation

estimates. This will only serve to relax the strain in the material further. It is, therefore,

accepted that if manufacturers do not provide creep data, it is more prudent and accurate to

assume a conservative value for the creep than may occur in a material instead of ignoring

creep altogether and neglecting creep relaxation.

190
 j— . —_ iUI uviw n iiiiiiig ui-aiiu pci iuiiiuuice 01 repair m aien ais

4.13 Summary of guidelines for selection of reinforced concrete

repair materials.

The procedure detailed in this section will assess the performance of a repair material

selected to repair reinforced concrete. The procedure examines the development of

properties in a repair material with time. The procedure needs to be performed at regular

time intervals. For example, in assessing the performance of a concrete repair over 400

days, the calculations need to be performed at a minimum of every five days, and every

two days during the first fifteen days. This is to take account of the effect of relaxation

through creep. Thus, the procedure lends itself solely to the application of a computer

program, which will speedily perform the necessary iterations. The steps are as follows:

1 Obtain properties of repair material

Obtain these key properties of the repair material from literature provided by the

manufacturer:

1.1 Key Properties

• Compressive Strength at 28 days (N/mm2)

• Tensile Strength or Modulus of Rupture at 28 days (N/mm2)

1.1.1 If Modulus of Rupture is provided then:

Tensile Strength = Modulus of Rupture /1.7

• Free Shrinkage at 28 days (microstrain)

• Elastic Modulus at 28 days (kN/mm2)

1.2 Optional Properties

191
 u pci iui nicuiuc u i repair materials

If creep information is not provided by the repair material manufacturer, refer to section

1.2.1. The required information is:

• Creep strain at 28 days (microstrain)

• Stress / strength ratio endured by creep sample

1.2.1 Estimating Creep

The following equation will estimate the 28 day creep of the repair material based on

its 28 day shrinkage value:

Creep = (shrinkage * 1.1) + 20 Equation 4-12

2 Obtain properties of substrate concrete

Extract Core from substrate concrete.

Using standard test methods determine:

Compressive Strength (N/mm2)

Elastic Modulus (kN/mm2)

2.1 Height/Diameter ratio modification

If the laboratory have not applied the height/diameter modifier for relative strength, it

can be done using the graph below:

192
 ,-------------------- vivuvn.i.111115 lllv^ p c iiu im a iite o i repair m ate rials

20

1-8

1-6
S tr e n g t h

14
Relative

12

10

0-8
0 0 -5 10 1-5 20 25 30 35 40
Height / D ia m e te r R a tio

Strength of core = Unmodified strength * Relative Strength factor (N/mm2)

2.2 Core to Cube modification

Apply the modification for conversion of core strength to cube strength

Cube strength = Core strength / 0.8

3 Climate Modification

3.1 Using the following table and map, obtain the local climatic conditions of the

proposed repair:

193
 nug, me lii-smi pci iuiiiitiiice or repair materials

Dec - Mar - Jun - Sep -

Feb May Aug Nov North

North 4° 9° 15° 11°

Mid 5° 10° 16° 12°

Mid
South 8° 11° 17° 14°

South

Dec - Mar - Jun - Sep -

Feb May Aug Nov

North 85% 70% 70% 85%

Mid 85% 70% 70% 85%

South 85% 75% 75% 85%

3.2 M odify shrinkage by temperature

Assuming the shrinkage property was obtained by a standard test, it can be m odified to

obtain the shrinkage at a datum temperature o f 15°C:

07 . , Shrinkage
Shrinkage, ,or = --------------
1.08

3.2.1 Obtain the difference in field temperature and datum temperature:

field temperature -15°

g = --------------------------------------------------
100

3.2.2 Calculate the field shrinkage modified for temperature:

Shrinkage f(temp) = shrinkage 15 oC * (1 + g)

3.3 M odify Shrinkage by relative humidity

194
 j - j---------- u v iv iiu u m ig mv Hi-oiiu p cin jim an w c u i ic p a u iiicuericiis

The measured shrinkage can be modified to the shrinkage at a datum relative humidity

of 70%.

. Shrinkage/( ,
shrinkagemwA = -------- JK

3.3.1 Obtain the difference between field RH and datum RH

^_ (% R H at time o f repair - 7 0 % RH) *3

~ 100

3.3.2 Calculate the field shrinkage modified for relative humidity

Shrinkagen ,emp&m) = shrinkagemwu * (1 - h)

3.4 Modify creep for temperature

Assuming the creep property of the material was obtained by a recommended test

(curing samples at 20°C), the creep can be modified to obtain the creep at a datum

temperature of 15°C.

C reepy = £ ^ c _
1.0625

3.4.1 Calculate the field creep modified for temperature

CreePf(leap) = CreepK.c *(l + (l.25* g))

3.5 Modify creep for relative humidity

Calculate creep correction factor for RH at which test on sample was conducted

(assume 50% if data unavailable)

i rtM = -0.007(iW % M) + 1.3

where RH%iab = Relative humidity whilst curing creep specimen

Calculate creep correction factor for RH at time and place of application

krhfield ~ - 0 . 0 0 7 ( R H % field) + 1.3

195
 v v aw* uvivnm iiiug) iiiv/ ih “dilu p c i lu in u u ic c u i rcpciir rrmtcriciis

Creep modified for relative humidity:

Creev
I f (rh+temp )
- Creep^
j
. krhfield
rhlab

3.6 Modify creep for relative test specimen and repair sizes

Determine creep modification factor for repair

-0-3057 ln(</)+ 2.8375

where d = depth of repair (mm)

Determine creep modification factor for test specimen:

*,<*»= -0-308 ln (v /s)+ 2.6367

where v/s = volume/surface ratio of test specimen (if unknown use 25mm [ASTM

C 512])

Determine creep modified for relative sizes of specimen and repair:

creep f (Rh+t )
Creeprep= ------f *ksl;M
size rep

4 Volume/Surface ratio shrinkage modification

4.1 Obtain Volume / Surface ratio of laboratory specimen from which shrinkage

data was measured.

If unknown use

(25x25x285)/((25*25*2)+(25*285*4)) = 6 mm (volume/surface,ab)

4.2 Determine Volume / Surface ratio of repair (volume/surfacefieid)

4.3 Calculate the relative shrinkage of both laboratory and field materials.

Pj b =10 779e-0'005(vo/"”'e1surface,ah>

^ 779e-0.005(vo/«me/ surface fM)

J = M lab
field

196
 — j— . j-- — ww *.v.» uviw iiniiim g m-axiu jjcnum iciiii;c u i re p a ir m a terials

„ _ ShrinkageS(tmr&m)
28 —
J
5 The development of properties

The development of certain properties and interactions inthe repair material canbe plotted.

Properties should be determined and tabulated ondays; 0, 2, 4, 6, 7, 8, 12, 14, 15, 20, 21,

28 then a maximum of every 5 days for as long as the properties need to be determined.

Leaving a gap between ages at which the calculations are performed will result in the

procedure not taking account of the constant relaxing effect of creep on the tensile strains

5.1 Development of Tensile Strength on days t = 0 to 400

/, .. t
fm 2.7975 + 0.9216.?

Where ft = Tensile strength on any day, t.

ft2 8 = Tensile strength of repair material at 28 days

t = time in days

5.2 Development of shrinkage on days t = 0 to 400

8 _ t

s n ~ 12.292 + 0 . 5017 ^

Where s = shrinkage strain on any day, t.

S2 8 = shrinkage strain of repair material at 28 days

t = time in days

5.3 Development of Elastic Modulus on days t = 0 to 400

E _ t
£ 28 ~ 3.2637 + 0.8725t

Where E = Elastic Modulus on any day, t.

E28 = Elastic Modulus of repair material at 28 days

197
 — t mi u^ivim iiinig me m-s>iiu jjciiurniciiicc or repair m aien ais

t = time in days

5.4 Shrinkage Transfer on days t = 0 to 400

As the Elastic Modulus of the repair material develops, some shrinkage strain may be

transferred from the repair material into the substrate concrete.

5.4.1 Determine Modular ratio on days t = 0 to 400

Modular Ratio = Erm/ ESUb

Where Erm= Elastic modulus of repair material at day t.

ESUb = Elastic Modulus of substrate concrete.

5.4.2 Determine transfer of strain (%) on days t = 0 to 400

^_ i^ rep / Esub ~l)

0.0032

where X = shrinkage transferred (%)

Erep/Esub = Modular Ratio

NB. 1 0 0 > A > 0

6 Unit Creep

Unit creep is the creep per load of 1 N/mm2

6.1 Determine load applied to laboratory creep sample

Creep specimens are loaded at 30% of their 28 day strength.

Applied load (N/mm2) = 30% fC28

6.2 Determine Unit Creep

Unit Creepinitiai (microstrain) = Creeprep / Applied load

6.3 Modify Unit Creep to allow for early age loading

Unit Creep = Unit Creepinitiai * 1-83

198
 — I— . —- xui m e in -sn u p c iiu n im n c e 01 rc p s ir m aterials

7 The effect of Creep

To calculate the effect of creep the stress at day t is required.

NB. In the following equations, referring to the ‘previous’ day, means the previous day by

way of increment. For example, if the current day is day 300, then the previous ‘day’ was

day 295 (using the maximum increment of 5 days)

Stress t = Et (Shrinkaget - creept(prev)) ...................................... (1)

stress. - stress.,
StressEC(tlot(prev)) = S tr e s s ,^ + ----------- - ..................................... (2)

.
( StressEC( ,, *t ) + ( StressEC(,,ot( ) * ( ? - / )
StressEC(otol) =-^---------------------- i ------------------ —P L..... (3)

Where Stressnc = equivalent constant stress

t = current day

StressEC(prev otot)= Result of final equation (3) from previous t

t(prev) = previous day (by increment)

Ct _ t
C28 7.0162 + 0.701 * t

C
Creept = StressEC(otot) * — —* Unit Creep
U28

8 Calculating Strain in the repair material for t = 0 to 400

Straint = shrinkaget - creept

A degree of relaxation through creep has occurred, this has had an effect on the strain in

the repair material, and henceforth an effect on the stress the repair patch is subjected to.

Therefore, section 7 is re-calculated replacing term (1) with:

Stress t = Et (Strain t )

199
The process is repeated iteratively (steps 7 and 8) until there is little discemable difference

in strain from one iteration to the next.

The values of creep and stress from the final iteration should be stored for use in the

calculations for the next day increment. The final value for Strain is a key value, Straint.

9 Calculate transfer of strain to the substrate for t = 0 to 400.

If the elastic modulus of the repair material is higher than that of the substrate concrete

then some of the strain will be transferred from the repair material into the substrate.

Straintrans,t ~ Straint * A,

Straintotai,t= Straint - Straintrans

10 Calculate tensile strain capacity of repair material from t = 0 to 400

c a p ,t j ,

11 Determining the performance of the repair material.

Plot s and Straintotai,t against time.

If the line s capt against time, is exceeded by the line Straintotai,t against time, then the

repair material will fail at the intersection point.

If the two lines do not intersect then the repair material will not fail.

200
 — 1— . — v uwivimiiiiiig uiv ih“dilu pcixuiiiictiiuc u i repair m aterials

4.14 Conclusion

A broad opinion of research concerning the performance of concrete repair materials has

been examined. The conclusion of this review is that current practices in reinforced

concrete repair do not adequately take into account the necessity to control dimensional

compatibility between the repair material and the substrate concrete. Additionally the

importance of ratios of elastic moduli between substrate and repair are neither understood

nor utilised by the majority of practitioners.

A procedure has been developed which predicts the development of critical tensile strains

in the repair material. By comparing these tensile strains with the tensile strain capacity of

a repair material (the development of which has also been researched), it is possible to

predict the success or failure of a repair material. It is also possible, should a material fail,

to predict the period after curing at which this will happen.

This procedure, if correctly implemented could reduce levels of failure in reinforced

concrete repairs.

The procedure has been incorporated into the computer program developed in this project.

201
 1— _ .— — i n mw w/vjjwii. oj’oi^m iu i ic m iu i^ c u euiiureie u n u g e re p a ir

5 Decision making in the expert system for reinforced

concrete bridge repair.

5.1 Chapter Objectives

• Create expert system for concrete repair

• Develop a simple, expeditious method to assess severity and extent of defects based

on cumulative expert knowledge and experience

5.2 Introduction

An experienced engineer has the ability to diagnose defects exhibited by concrete and to

recommend suitable remedies. Importantly, the ability to assess the significance of the

extent and severity of defects, and to be minded of these in making repair

recommendations, is a crucial part of the engineer’s expertise. A review of expert systems

for concrete repair in Chapter 2 of this thesis found that existing systems give generic

repair advice which does not account for either the extent or the severity of particular

defects.

The importance of a concrete element to an overall structure should be an important factor

in the process of making decisions for repair. The central pier on a bridge can be

considered a crucial element, upon which an averagely sized area of spall of reasonable

depth might be considered very significant to the overall well-being of the structure. A

wingwall on the same structure, with a similarly sized yet very deep spall, might not have

important structural or durability implications for the overall structure, and should be

202
 '& wajjcii 3_yaicm iui icm iurcea concrete onoge repair

treated accordingly. These are examples of decisions that an engineer will be making

subconsciously as soon as he/she begins inspecting the structure, and yet, this most basic

information will need to be entered into a computer in order for it to be able to make even

the simplest decision. With the benefit of sight, engineers begin making subconscious

decisions based on their knowledge immediately upon introduction to a defective structure

- a distinct advantage for the engineer over a computer. With this in mind it can be said

that there is a need to ensure fast and efficient entry of data into the expert system.

Importantly, the elicitation of knowledge from experts, and the subsequent development of

the expert system will take a simplistic approach to decision making.

5.2.1 Expert systems

Expert systems can be broadly described as computerized advisors that attempt to imitate

the reasoning of experts in solving problems. Expert systems are also known as knowledge

based systems.

There is no single code or set of guidelines for concrete repair, therefore gathering domain

knowledge in the field is challenging. Best practice guidelines are dispersed amongst

papers, regulations and instructions. Advice generated by the expert system has been

collated through a literature review (Chapter 2) and interviews with field experts.

It is held as crucial that an effective and simple method of enabling an expert system to

assess the severity and extent of a defect be developed for this work.

The aim of the expert system for reinforced concrete repair, is to create a software tool

that, when given data on a bridge and its defects, can analyse the data, recommend a

testing regime, recommend repairs, and finally recommend the most suitable repair

material in accordance with the advice in this thesis. This calls upon the software to

simulate the roll of an engineer in decision making.

203
 iui icm iu itcu uuucreie uriuge repair

5.2.2 Determining the severity of a defect

A key requirement of an intelligent expert system is the judgement of the severity of a

problem it may have to diagnose. A problem discovered in some existing expert systems is

the lack of advice offered on when to repair, and if repair is necessary. As an example, the

appearance of a typical carbonation induced crack on a concrete element may well be

cause for concern, although if the crack is negligible in length when compared to the scale

of the element itself, this concern may well be misplaced. Existing expert systems appear

to show little concern for this problem and will often give the same advice for the above

example, as would be given for severe corrosion over an entire element.

A technique has been developed which provides the expert system with sufficient data to

make decisions on the scale of defects, this part of the system is used in conjunction with a

series of diagnostic knowledge bases to give the user a full picture of the nature and

severity of any defects. The technique aims to make fast accurate decisions, without the

need for laborious questioning or advanced mathematics.

5.3 Data input

The Highways Agency, and an increasing number of local authorities, store bridge

information electronically. The Highways Agency’s Structures Management Information

System (SMIS) has replaced paper as a means of storing data from bridge inspections.

However, SMIS, in common with most bridge management systems, stores text

information. It was identified in the development of the expert system that a text interface

is not a sophisticated use of available technology and an alternative form of user

interrogation was developed.

204
Generally, reinforced concrete bridges are built to standard arrangem ents o f decks,

columns or piers, abutments, and wingwalls. W ith this in mind, it was decided that a fully

graphical interface between the user and expert system would provide users with the type

o f modern software interface with which they are familiar, as opposed to a text based

system that may seem old fashioned.

To enable the expert system to determine cause and severity, it is necessary for the

program to also have inform ation about the structure. Therefore the user is invited to create

a three-dimensional representation o f the structure or element upon which he/she requires

the expert system to generate advice. This is done by the user being guided through a

‘w izard' - a short on screen routine which automatically generates a typical reinforced

concrete bridge based on a number o f stored standard arrangem ents w hich the user can

tailor. Alternatively the user can build a structure element by element. Exam ples o f this

process are shown in Figure 5.1 to Figure 5.3. Figure 5.1 shows the screen that

immediately follows the request to construct a new structure.

- |9 | x|
File Edit View S tru c tu re Go S tock Abnormal Loads R e p o rts C ustom ise Help

i 35 “ (Co S # H tm fl 3 <5» © I® 0 0 0 0 S> (2k =* ui <4 rx

All S tru c tu re s

D em oB ndges

nis
n37
” 38
T& 39
win
Utilities
^ Canied
7A D ossed Structure Condition
BR1001
Tj BR
T5 BR
T3 BR
Structure 1Location Descrpticwi j Construction j Superstructure j Substructure JAssessment Restrictions! Equipment
Tv? BR I S tructure Name: [g tructure 41 Number 41
T3 BR
\ Structure Type: J Reference: I41
V I BR d
Tv? BR i Structure Owner j Owner Ref:
T3 BR d
ri ; Maintaining Agent: I Structure Status j
br d d
T-? BR Structure Carries: I A ssessed Cap f
t3 br d
Ti BR ; Structure Crosses: f Next Inspection: |
d ID Date |
n gr ; Min. Headroom: [ Date Measured: j" Last Inspection: j
J, RR ---------------------_ J

G e n e ra l ^ D o c u m e n ts # In sp e c tio n s J M a in te n an c e Condition rj
Ready.,

F ig u re 5.1 N ew stru c tu r e in serted

205
 j— „ „ w~.-j .v-,.. ...unnii, in ui^ ca^ ii a^aiciu iui lcmiuiucu concrete onoge repair

In Figure 5.1, the program has been instructed that user would like to inform the system o f

a new structure on which a defect has been encountered. The user has indicated that he/she

wishes to add a reinforced concrete deck. This is drawn onto the screen provided as shown

in Figure 5.2.

-lnl>

Add New Element

£3 D e c k E lem ents
B Primary D e c k Elem ent
1*
o a Main B eam s
urn T russ m em bers
ua culvert 8
me Arch
wm Arch Ring 6
la i V ousoirs/A rch F a c e
a& Arch B arrel/Soffit 4
aaf E n c a se d B eam s
&» Subw ay 2
a s Box beam interiors
A rm co/C on crete pipe 0
m» P ottal/T unnel portals d 18 20 22 24 28 28 "5(T<
P restressin g
S le e p e r bridge
T unnel Linings
D ec k
Ci T r a n sverse B eam s U ser th e cursor to draw this elem en t
onto th e structure!
Ci S ec o n d a r y D e c k Elem ent
i •Ci Half Joints
(• Draw in Elevalion
Ci T ie B eam /R od
h P arapet B eam or Cantilever C Draw in Section
03 a D e c k Bracing
C Draw in Plan
.+ Cl L oad-B earing Substructure
D Durability E lem ents
Ct] Cl S a fe ly E lem ents
S) Cl Other Bridge E lem ents
I+j Cl Ancillary E lem ents

F ig u re 5.2 In sertin g a d eck elem en t

In Figure 5.3 small bank seats are added under the deck. At this stage, the expert system

has a fair representation o f a simple short span accommodation bridge. The process has

taken one minute, and the program is ready to being acquiring inform ation on defects.

206
 111 uiv v,a[/wi ivji i c m i u i t c u c u n t i c i c ui luge repair

& E lem ent Wizard

■ H H K H -lol>
m. - ra m ite © & &
Add New Element

D e c k E lem ents
2a Primary D e c k Elem ent
2a T ia n s v e is e B eam s
2a S e c o n d a iy D e c k Elem ent
2 i Half Joints
.li T ie B eam /R od
2a P arapet Beam or Cantilever
2a D e c k Bracing
L oad-B eating Substructure
2a Foundation
-j A butm ents (incl. arch springing
am A butment
mm Arch Springing
mm Abutm ent slo p e UJ— 2 4 S 8 10 12 14 16 18 20 22 24 26 28
mm B ank s e a t
mm C ounterfort/B uttresses
2a Sp andrel w a ll/h e a d wall
2a Pier/C olum n Element Location: Element Dimensions: Specific Dimensions: Element Infor
2a C r o ss-h ea d /ca p p in g beam
2a B earings X Origin: 25000 Span (XJ: | 2000 M: | 2500 w1: | 667 Re
2i B earing p lin th /sh elf Y Origin: Width (Y): | BOOO ID Code:
Durability E lem ents
S a fe ty E lem ents Z Origin: i 6500 Height (2):
weight |ZJ: | 3000 Material:
Other B ridge E lem ents Span:
Ancillary E lem ents Skew / Rotation Options
Date Built:

.Ll Structure Wizard j Can

f S t a r t ! y b Norton Internet S e ... | Chapter 5 - Microso... | ft iTunes »tuS E R -3W T 251L F57...| ® « •£ & & & & 19:24

F ig u re 5.3 A d d in g b a n k -sea ts

It is not necessary for the expert system to know precise dim ensions o f a structure, as such

a degree o f accuracy will not affect the system recom mendations. However, a reasonable

representation o f an element o f a structure will enable the expert system to make sound

judgm ents about the extent o f defects.

5.4 Diagnosing concrete defects in an expert system

5.4.1 Beginning the process

Concrete defects can be grouped into four categories:

• Spalling

• M ap-cracking

• Structural cracking

• M iscellaneous defects

207
 ‘6 ayaicm iui ic n u u ic c u concrete Driage repair

These categories were discussed in Chapter 2.

It can be reasonably expected that even a user with a basic knowledge of reinforced

concrete could distinguish (with some guidance) between these four types of defect; in the

program developed it is assumed that this is the case.

In order for the process of defect diagnosis to begin, the user must graphically add a

representation of the defect on to the concrete element entered into the program (section

5.3).

After this stage the expert system knows the approximate size of the affected element, the

general type of defect and the approximate size of the defect. Thereafter, the four general

defect types are treat differently by the program. This basic information will form the data

which will be fed into the knowledge bases for advice.

5.4.2 Constructing expert systems

5.4.2.1 Knowledge elicitation

Knowledge elicitation is the collection of domain knowledge. It is conducted in

consultation with domain experts39,47. The domain experts in the field of concrete repair are

generally civil engineers. This process of acquiring domain knowledge from experts is

conducted through a series of interviews. These interviews can be done formally before a

panel of experts or with individual experts. For the construction of this expert system,

informal workshops were conducted with a series of experts.

208
 1------ .. .n iiiv vApwi iui icmiuiccu uunuieie uriuge repair

5.4.2.2 Knowledge representation

In order to represent elicited knowledge, Kalyanasundaram et al39 used a technique

involving the formation of knowledge nets (Figure 5.4). They conceptualise the knowledge

involved in repairing cracked concrete and place it into a static knowledge net. This

knowledge net is then programmed into an expert system shell.

RK R5 00 BE ES.BE RS
AH AH
00 AH * ES

EO - O ^ a riay R* - R ad aa ig n a n d P r o v i a a E x p a n s io n J o in t 6 - S titc h in g ES - £ » t a m » l & tr*««ing DL- B l«r*«lirvj

AH - A u to g en o u s H ealing 0 0 - O r d in a ry O v e rla y 0 - G ro u tin g BE - Bonding with Epoxy RS - R out an d S e a l

Sam ple Knowledge Net for Suggestion of Repair Method for "Cracking In Concrete"

Figure 5.4 Knowledge net

Variations of knowledge representation are similarly based on the creation of semantic

networks (or flowcharts). Rajeev and Rajesh47 take a more basic approach with the use of

instance nets (Figure 5.5).

209
 in Liiv v/vpvn oj'aicm iui i c u n u i c e u concrete o nage repair

C a u s e is
in a d e q u a te
flexural ca pac ity

P a ra lle l Perpendicular

bottom of b e a m

Dis» n t e g r a l i o n

Symptom

Column

F ig u re 5.5 In sta n ce net

5.4.2.3 Coding

The knowledge engineer is provided with two ways o f coding the know ledge obtained, and

producing the inference engine which will eventually make decisions. Firstly, the engineer

could construct an expert system in its entirety. That is the inference m echanism s and

entire software components (the knowledge and the brain are both formulated). Although

this method can allow the engineer to tailor the engine to the specific requirem ents o f the

domain, it requires an expert programmer. The second method is to use an expert system

shell. Expert system shells contain inference engines prepared and ready for the input o f

210
 '& a^aicm hji ic im u ic c u cunweie onuge repair

objects and rules (the brain is acquired, the knowledge is formulated). An expert system

shell can facilitate the structuring of knowledge, the control of the inference strategy, and

some shells enable the design of the user interface109. The primary purpose of the project is

to create an expert system for reinforced concrete repair and not merely to pursue

innovative methods of creating expert systems. Therefore, a review of expert system shells

was conducted, and a shell called ‘Acquire’ was obtained for the purposes of representing

(in a knowledge base) information obtained from the expert panel through workshops and

interviews.

It was established during initial interviews that whilst an expert system shell could function

adequately to determine the cause of defects, such shells are not necessarily suitable for

determining severity and extent of defects, particularly not in concrete repair situations

where a single element could contain a large number of defects. Therefore, it was

established that some form of traditional software programming would have to be used to

determine the severity and extent of defects on an element. Thus, a key aspect of the

development of an intelligent advisor for concrete repair is that a traditional expert system

shell, through knowledge bases and an inference engine, will be used to diagnose the

causes of concrete defects. This will work in tandem with a traditional software program

that will mathematically assess the severity and extent of the defects. Throughout the

process of concrete repair i.e. from diagnosis to repair material recommendation, the two

aspects of the system will work together. The overall aim being to use the software tools

available with as much simplicity as possible to create the desired intelligence.

5.4.3 Developing the knowledge bases

As a result of initial interviews, it was identified that there existed a need for five distinct

knowledge bases (KB) in order for an expert system to be able to allow a user to fully

211
 — j— „ .— ....... v,..& in mw v/vpvn a_yaicin iui iciu iu iu cu concrete o n a g e repair

diagnose any defect and take that defect through to the same conclusion which an expert

would arrive at - some form of action, be it “repair”, “monitor”, or “do nothing”. The

required knowledge bases identified were:

• Diagnosis of cause of spall defects

• Diagnosis of cause of pattern crack defects

• Diagnosis of cause of structural crack defects

• Recommendation of testing regime for element

• Recommendation for repair methods

It was identified that these five knowledge bases would be employed at different stages in

the framework of the overall system as follows

User enters structure geometry

User enters defect geometry and information

Spall KB or pattern crack KB or Structural crack KB used to determine probable

cause of defects

Severity and extent of defects determined

Effects of all defects on single element assessed

Testing regime recommended by Testing KB

Cause of defects confirmed by test results

Repair information provided by Repair KB

Suitable properties for repair material recommended (see Chapter 4)

Bridge condition assessed

212
 v/ajjwi a p ic m iui icim u iu eu concrete o n o g e repair

The knowledge bases and other routines developed to enable these ten stages to be

performed by the expert system will be integrated into a commercially viable bridge

management system.

For each knowledge base, the expert panel were asked to identify any criteria that could be

used to assess the cause of a defect.

5.4.3.1 Knowledge base for spalls

Beginning with the knowledge base for the identification of spalls, the experts listed all the

factors that might influence their decision as to the cause of any spalling:

• Shape

• Size

• Depth

• Age of element

• Location of element in relation to splash zone of vehicles

• Reinforcement exposure

• Depth of exposed reinforcement (cover)

• Reinforcement condition

• Evidence of staining

• Evidence of seepage

• Proximity of element to carriageway (likelihood of impact damage)

213
 ^ ^ v v w . v / n miuviiig in uil, cA pcn ayaicni lur rcim orcea concrete o n o g e repair

In the expert system shell (Acquire), each factor which could determine the cause of the

spall is called an ‘object’. Each object has a ‘range’. Ranges are the units of measurement

over which objects are measured - either textual or numerical. For example, the range for

the object ‘Age of element’ is a numerical value in years ranging from zero to infinity, the

range for the object ‘Splash Zone’ (the object whose range is set to determine if the

element is within the splash zone of vehicles) has the textual range ‘yes’ or ‘no’,

importantly, the value ‘unknown’ can also be chosen. Similarly, the object ‘reinforcement

exposure’ has values: unknown, none, low, medium, high. Although some objects, such as

‘reinforcement condition’ have ranges that comprise of natural language identities such as

low, medium and high - in the expert system developed the user is seldom asked to make

these kind of assessments, as that requires expert knowledge. Although the expert system

shell requires to know the value for ‘reinforcement condition’ to make accurate decisions -

it is provided by the user only indirectly, the actual value is inserted at the data input stage

using photographs of different severities of reinforcement corrosion to guide the user . This

close interaction between the diagnostic expert system shell and the numerical data input

program written in a traditional computer language is a key relationship in the overall

performance of the expert system.

All the data required by the expert system to make a decision is entered by the user at the

data input stage. Figure 5.6 to Figure 5.10 show the procedures necessary for the expert

system to obtain all the information required for it to be able to make decisions about the

cause of spalling. Figure 5.6 shows the structure constructed earlier in this chapter - a

central pier has been added.

214
 xxl mv. vajjvii iui icm iuiccu cunciete ui luge repair

f USER 3W1251 LF57\BMX - bridges Demo Bridges - 41 -Ifll xj

File Edit View S tru ctu re S tructure Stock Abnormal Loads R ep o rts C ustom ise Help

F flS W i x 0 s 5 © s> Sd! n «s? a

H E S S f c .- Zl
A ll S t r u c tu r e s

D e m o -B r id g e s

H 18 *
F I 37
*■ 38
V t 39
■ firE fl
Ci Pier/colum n
Ci Main B eam s
Ci B ank s ea t Structure Condition
Utilities
JcC Carried
V /4 C rossed
V3 BR1001
•CT BR 1002 Structure Name: Number:
Structure 41 41
•£7 BR 1003
T7 BR 1004
T 7 BR 1005
........................... d R eference: |41

X 7 BR100G .......................... d Owner Ref: f

■57 BR10Q7
■f ? B R 1008
....... d Structure Status: j
d
■57 BR 1009
Structure Carries:
r ----------- d A sse sse d Cap. j~

* 7 BR101D
r - r BR1011
Structure Crosses:
i"""" d Next Inspection: f
ID Date: j”
Min. Headroom: D ate M easured: j Last Inspection: I
T 7 BR 1012
■57 BR1014 —.

H GR1001
d G eneral Docum ents '% Inspections ^ M aintenance % Condition %
R e a d y ...

F igu re 5 .6 S tr u ctu r e fo r d efect stu d y

Assume a spall defect has been encountered on the central pier. In Figure 5.7, the central

pier has been selected with a double click. An unwrapped view o f the pier is shown, and it

is onto this view that the spall defect will be added.

215
 vpv^n 3^3Lcm iui icm iuiceu concrete onoge repair

m m sm -Jj9JxJ
File Edit View D efects Structure Stock Abnormal Loads R eports Customise Help

4 35 % ~ X © d§ ^ *o n :-' & Ei • t 4 1 1-4 ®

GridOptions:
V t 41
Bridge View Pier/column 1 !
Pier/column
■iim iai
Main Beams
Main Beams 1
B ank s e a t
B ank sea t 1
*!* B ank se a t 2

& O

Diagnosis:

%N> »
Component selected - Click a face to display defects.

B a c k to Structure

R ead y ..

F ig u re 5.7 C e n tra l p ier selected fo r stu d y

In Figure 5.8 a rectangular defect has been added by selecting the ‘add defect’ icon and

dragging a box onto the unwrapped pier as shown. At this stage the system does not know

the type o f defect that has been added.

fm m s m r m m m
File Edit View D efects S tructure Stock Abnormal Loads R eports Custom ise Help

4 $— —*— ---
j—-
% III X
—----- —.~»i
(SO , S ^ - O □ ' ® § 4 4 t 1 ©
Inspection: ] 157 Scheduled: 1 2 /0 5 /2 0 0 5 Grid Options: 4 t 4 1 snap
V I 41
□ Bridge View J 157 j Pier/colum n 1
Pier/colum n

j- i Main Beam s
w Main B eam s 1
O B ank se at
V B ank s e a t 1
*1* B ank s e a t 2

&
r j
7~ ■

13
T esting Advise $ I♦>I* I
Com ponent fa c e sele cted - d efects, test results an d repair p a tc h e s are show n. Click on a defect to display its condition

J B ack to Structure
R e a d y ... fa ce Vou m ay add d efe cts, te s ts an d repairs to this fac e

F igu re 5.8 A d d in g re c ta n g u la r d efect

216
 ^ , . , u . w u & in ini- tA p v ii 3_yaLcm ivji i c i m u i u c u cuiicreie onuge repair

In Figure 5.9 the user informs the system that the defect added is a spall, by highlighting

the spall icon as shown.

A Add Defect Wizard i ~' '

Paintwork and Protective Systems j Vegetation j Foundation j Invert and Riverbed { Drainage | Surfacing ]
Expansion Joints | Embankments j Bearings j Impact WaterProofing 1 Stone Slab General
Knowledge B ase D efects Steel j Concrete j Timber | Masonary and Brickwork

m
Structural
m Stain Map
H
Seepage
V
Scaling Honey-com Blow Holes
iij
Sand
w
Stalactite
Crack Cracking bing Streaking Build-up

Defect Details
Comment / Site Action:
Reference: j~ Work Type: Routine
Severity: pjf T j Priority: Low
Extent: TX Estimated Cost:
D efect Photo:

Recommendation: b o Nothing
"3

Dimentions
X Location: [ 5193 Width: [ 4287

Y Location: j 1390 Height: j 2191 Aesthetics: J~

Urgent Action f“

Cancel I OK I

F ig u re 5.9 A d d in g a sp a ll d efect

In Figure 5.10 the user adds further details about the spall, such as its shape, depth, and if

the defect is within the splash zone o f vehicles.

217
 pv-ii iui lcim uiccu cuncreie Dnuge repair

jft Spall Detail H i a !

Exposed Reinforcement I Corrosion Rating | A ssociated Other Details

Is in splash zone?
Shape Irregular R ectangle
Represents: Irregular Circle
C No Y es C Unknown
Irregular Other
Perfect R ectangle
r l n W etted Area? Perfect Circle

C No (* Y es r Unknown S eapage: [rane~

medium
Is Low Cover Visible?--------- high
unknown
(•No C Yes C Unknown
weather:

In Impact Z one?
Heavy Rain
Unknown
r No <• lYesi C Unknown

Spall Depth: 50

Back Next Cancel Finish

F igu re 5.1 0 A d d in g sp a ll d eta ils

Figure 5.11 shows how the user is expected to grade the condition o f exposed

reinforcement. Having initially gone through the data input procedures, and inform ed the

system that the spall exhibits exposed reinforcement; the user is presented w ith a graphical

interface. Using this interface the user judges, via comparison, how badly corroded the

reinforcement is and matches this to the examples shown. A n internal setting in the

program, decided by the expert interviewees, determines that a rating o f 20% or less sets

the object ‘reinforcement exposure' to ‘low ’. A rating o f between 20% and 60% sets a

value o f ‘m edium ' and anything above that ‘high’.

218
 iui icnnuiccu cuncreie o n o g e repair

f t Spall Detail _>£]

Exposed Reinforcement Corrosion Rating A ssociated Other Details

(* Corrosion looks like image indicated below

Don't Know

Corrosion Rating (Percent): | 59%

Slide bar to indicate picture which most looks like the corrosion

Cancel

F igu re 5.11 In p u ttin g co rro sio n a m o u n t

The user may not be able to provide values for all the objects that the know ledge base

requires to make its decision. In these cases, the value o f these unknown objects is set to

‘unknow n'. Once all data has been entered and all the objects’ ranges set. The know ledge

base is ready to ‘fire’.

In order to make expert decisions the expert system uses ‘rules’ which were created during

the expert interviews. The expert system shell uses two different m ethods to set rules.

Figure 5.12 shows a premise rule which the knowledge base for spall defects uses to set the

object ‘cause carbonation’. In the expert system, input objects have their ranges set by the

219
 r „ . nuki ng m uiv 3_y3itiii iui lcn u u iu eu concrete onuge repair

user data and output objects have their ranges set by rules. ‘Cause Carbonation’ is one o f a

number o f output objects. The collection o f rules forms the knowledge base.

Acquire - C:\experlsyslem\finished\kb\Spall\spallclefecl1-3.KBS HEII3

File E dit E d itors G ra p h s E n g in e R e p o rts Utilities W in d o w H elp

Premise Editor - cause carbonation 1

D isplay: V alu es: O bje cts:

( G ra p h ic a l age OK

S h o w O b je c ts As:

[A b b re v ia tio n jj] ex p o sed

R eset

N um ber C lear

And

|~ A n d And

ex p o sed | | none age yes

F igu re 5.12 C a rb o n a tio n p rem ise ru le

The rule in Figure 5.12 uses abbreviated object names. It sets the output object ‘cause

carbonation’ to the value Tow’. The diagram shows two criteria, both o f which m ust be

met in order for this rule to set the value for ‘cause carbonation’ to low. Firstly the bridge

age must be less than 12 years, and the object Tow cover’, w hich is set from the user

entered data, must be set to ‘yes’. Secondly, if there is no exposed reinforcem ent, there

must be evidence o f corrosion (‘corrosion’ object not equal to ‘none’) or, if there is

exposed reinforcement, it must exhibit some form o f corrosion. If these conditions are met,

the rule ‘cause carbonation’ will be set to Tow ’ and this result will be passed on to the user

at a later time. The basic premise o f this rule is that carbonation is not expected in a young

bridge - however, if the cover is low, and if there has been corrosion o f the reinforcem ent,

220
 iii djdivm hji icm iuiccu cuncieie uiluge repair

there is a small chance that carbonation could be the cause. Hence the object ‘cause

carbonation’ is set to Tow’. The rule shown in Figure 5.12 is one o f a num ber o f different

rules which affect the object ‘cause carbonation’. Other rules set the object’s value to

‘high’ and ‘m edium ’. If none o f the conditions in the various rules for ‘cause carbonation’

are met, the value o f that object will remain at ‘none’ - i.e. the knowledge base does not

think the cause o f the defect is carbonation.

Figure 5.13 shows all the rules in the spall knowledge base. Each rule is built up from

expert opinions.

A c q u ir e - C :\e x p e r t s y s t e m \ f i n i s h e d \ k b V S p a l l \ s p a l l d e f e c t 1 - 3 . K.BS
F ie Edit Editors G raphs Engine R eports Utilities V ndow Help

KSSSISflHBBHHHHHHflHBHHfli
RH S Object: Rule 8: aar 1
c a u s e caroonafron
ir c a u s e carbonation 1

c a u s e carbonation 3
S elect LHS O bjects.. c a u s e carbonation 4
c a u s e carbonation 5
c a u s e carbonation 8
P7 Action Table Rule I- Production Rule c a u s e carbonation 7
c a u s e chlorides 1
Context...
__________i
. c a u s e chlorides 2
consult engineer 1
C c n c b s ro r r'
consult engineer 2
~2 coirosion 1
coitosion 2
M essage:
filled p o c k et 1
filled po c k et 2
impact c a u s e 1
popout 1
popout 2
no visible rebars / corrosion previous p a tc h repair 1
bridge over 3 0 y e ars old spacing block 1
spacing N ock 2
stainingyesorno 1
stainingyesorno 2

Save j S a v e N ew D elete R e se t Cleat I Summary...

<l I

F ig u re 5.1 3 R u les in th e sp a ll k n o w le d g e b ase

The second type o f rule that the knowledge base uses is an ‘action table’ rule. Action table

rules begin with a context. Figure 5.14 shows the context for an action table that m ight set

the ‘cause carbonation’ object. In accordance with the context, the knowledge base will

only use this action table if the bridge age is less than 30 years and the spall exhibits no

221
exposed reinforcement. For other possible scenarios such as the bridge age being over 30

years and the spall exhibiting exposed reinforcement, other rules have been constructed. A

full set o f ‘cause carbonation’ rules are constructed in order to enable the system to give

advice about any com bination o f data.

F8e Edit Editors G raphs Engine R eports Utilities W indow Help

1 —

Display: Values: Objects:

[G raphical age
c cover
corro Cancel
S how O bjects As:
corrosion
[Abbreviation ~*|
staining
stainyn
Number
w etted
lLL
i'

"3

zone | | UUKHOTiIH | I 30 | | age 1 | ex p o sed | | none

1 ; D e e t c r. |

Ujj

F ig u re 5.1 4 C o n tex t fo r ca rb o n a tio n a ctio n ta b le rule

If the context for the action table rules has been met, then the rule itself com es into force.

Figure 5.15 shows the action table for a ‘cause carbonation’ rule. Each column o f the

action table is headed with the name o f an object. In the rows below, the different values o f

that object appear. For example the object ‘corrosion’, which represents evidence o f

corrosion (such as staining) in the absence o f exposed reinforcement, has its values; low,

medium, high, and unknown listed below the column heading. Importantly, enough rows

appear in the table to compare every possible com bination o f the object values in the table,

and this can be seen in Figure 5.15.

Because o f the context, the column headed ‘exposed’ is always set to ‘none’ in this action

table. The column headed ‘age’ is set to ‘num eric’ although the rule only fires if the bridge
 — I— „ .— — ...u.v.nt, iii mw o^oicm ivji icm iu iu cu uuiiureie D nage repair

age is under 30. The other columns have variable values. One hundred and twenty rows

exist in this table so that all object values can be compared. In the rightmost column the

value for the output object (cause carbonation) is set, by the expert, to either ‘low’,

‘medium’ or ‘high’.

Taking the first row as an example, and bearing in mind the context of this rule - that there

is no exposed reinforcement and the concrete is younger than 30 years old:

• Low evidence of corrosion, e.g. the user would have indicated, at the data input

stage, that there was evidence of mild corrosion staining.

• The spall is in a wetted area.

• The object ‘Zone’ represents the category of defect that this spall has been placed

into by the traditional computer program - the area of the expert system where

severity and extent are assessed. This variable is set to either: ‘major’, ‘minor’,

‘cosmetic’, or ‘do nothing’. The method by which this is assessed is described in

detail in section 5.5.

The expert judgement for this combination of values is that there is a medium chance that

the cause of the spall is carbonation of the concrete. The object ‘cause carbonation’ is set

to ‘medium’, and this will be reported to the user at a later stage.

This type of judgement is made for every combination of object values in the action table.

It can be seen that using the action table allows a great number of combinations of object

values to be assessed quickly. Premise rules and action table rules are used together to

ensure that all eventualities can be assessed by the knowledge base.

223
File Edit Editors G raphs E ngine R eports Utilities W indow Help

View a s R ules... | R a n g e ... | RH S V alues: OK

low
View C ontext... | R eoider | medium R e se t
high
N um ber of Row s: |1 2 UNKNOWN

R ow s to Com plete: I0 J
age | ex p o sed | corrosion zone w etted | RH Side
NUMERIC no n e low major yes medium
NUMERIC no n e low major UNKNOWN medium
NUMERIC no n e low minor no medium
NUMERIC none low minor yes medium
NUMERIC no n e low minor UNKNOWN medium
NUMERIC no n e low cosm etic no medium
NUMERIC no n e low cosm etic yes
UNKNOWN
medium -
NUMERIC no n e low cosm etic medium
NUMERIC no n e low donothing no medium
NUMERIC no n e low donothing yes medium
NUMERIC no n e low donothing UNKNOWN medium
NUMERIC no n e medium cosm etic no medium
NUMERIC no n e medium cosm etic yes medium
NUMERIC no n e medium cosm etic UNKNOWN medium
NUMERIC no n e medium donothing no medium
NUMERIC no n e medium donothing yes medium
NUMERIC no n e medium donothing UNKNOWN medium
NUMERIC no n e UNKNOWN major no medium
NUMERIC no n e UNKNOWN major yes medium
NUMERIC no n e UNKNOWN major UNKNOWN medium
NUMERIC no n e UNKNOWN minor no medium
NUMERIC no n e UNKNOWN minor yes medium
NUMERIC no n e UNKNOWN minor UNKNOWN medium
NUMERIC no n e UNKNOWN cosm etic no medium
NUMERIC none UNKNOWN cosm etic yes medium
N| IK jFRir onro. 1INK MOWN rn«nr»plir UNKNOWN —
............................. -d

F igu re 5 .1 5 A ction ta b le fo r 'ca u se ca rb o n a tio n '

The screen grabs shown in Figure 5.12 to Figure 5.15 are not at any stage shown to the

user in the finished program. They are internal screens used to construct the know ledge

bases and to set the output objects. Once output objects are set they can be presented to the

user in the user interface as required.

5.4.3.2 Knowledge base for pattern cracking

For the purposes o f this system, ‘map cracking’ is defined as any distinct patch o f concrete

that is affected by cracking. Single cracks are indicative, often, o f some form o f structural

cracking - these are handled by a separate knowledge base.

During interviews, experts were asked to list any features that may affect their diagnosis o f

an incidence o f pattern cracking. The following were identified:-

• The appearance o f the pattern cracking

• The age o f the bridge

• Efflorescence and its colour and form

224
 1--------- ----- .--------------------------------- u i ...V v a j j w i O J J I V U I 1U1l v ^ l l l i u i w c u C U l l ^ I C l C UI I U g C i c p d l l

® Seepage

• Evidence o f corrosion

• Type o f element and alignment

• Is the defect in the splash zone o f vehicles

• Is the defect in a wetted area

There is a noticeable difference between the method the expert system uses during data

input to obtain pattern cracking inform ation against that it uses to obtain spalling

information. Figure 5.16 shows how, after the user has drawn a patch o f pattern cracking

onto an element, the user is requested to choose an image which best represents the type o f

cracking they have encountered.

Delated Questaiftsj AjwtatorJ | OittAiftdGkmtoriv] Azvo'-.v.i-d Seepage? j

Lockt mow ik.8? j Go*P»«ent? j While DepoiiT? j AssoooIk J S‘ofwg71 As:oc!*od; poling?} i . 4 Ike'*} GeJPii?:w<'‘l WM*-Depot*?} A ttoo^Stdw sg^} Aii«j^«JSp*8ng? }

Kcdiyyc DefoK

S n c e Scf id

F igu re 5.1 6 S ele c tin g a r e p r ese n ta tiv e m a p -cr a ck in g im a g e

This method negates the need for the user to choose from a list o f textual descriptions o f

the defects. Each one o f the available map-cracking images is typically representative o f

cracking induced by one o f the following ailments:

Chloride corrosion

Carbonation corrosion

225
 J — __ — A** i i i v v A y v u J J J I W 11 1U 1 I t l l U U l t C U CU 11L/1C L C u n u g c icp a n

Alkali aggregate reaction, alkali silica reaction etc.

Freeze-thaw damage

Plastic Shrinkage

Crazing

Drying shrinkage

Once the user has selected a defect image, he/she enters other relevant information. After

this the knowledge base is ready to make decisions. The knowledge base to determine the

cause of map-cracking was constructed in the same way as that which determines the cause

of spalling. An object called ‘image’ has an alphabetical range from ‘a’ to ‘i’ where each

letter represents one of the images the user selected to describe the appearance of the

pattern cracking. All the objects are used in the formation of premise rules and action

tables to diagnose the cause of the defect.

5.4.3.3 Knowledge base for structural cracking and single cracks

For the purposes of the expert system, any single crack which is not catered for by one of

the representative images in the pattern cracking data input prompt is handled by the

structural cracking knowledge base.

Many large individual cracks are caused by corroding reinforcing steel, which causes

cracking along the length of the reinforcing bar. Therefore, many individual cracks will not

actually be caused by structural effects, if a crack added by a user is suspected of being

caused by corrosion, and therefore is non-structrual, the user will be informed of this.

At the data input stage, the user draws on the element a crack, as shown in Figure 5.17.

Following this the user is requested to provide as much additional information as possible,

such as crack width, associated staining or seepage etc.

226
 -Iff! xj
File Edit View Defects Go 5tock Abnormal Loads R eports Customise Help

4 ® % —-------
- x j—( -------
a m A - o □ ij :© s ©
—
—-------- —
Inspection: j 157 S cheduled: 1 2/05/2005 ~y~| Grid Options: + t 4-
I ? 41
Bridge View j I57 J Pier/column 1 j
Pier/column
i raaBW
: m
:• a
C Spall 412
Main Beams
W Main Beams 1
Bank seat
Bank seat 1
*|* Bank seat 2
fjSjSjSjSj*

0-

Diagnosis: I

Testing Advise

Component face selected ■defects, test results and repair patches are shown. Click on a defect to display its condition.

11
J
B a c k to S tru ctu re

R eady... face You may add defects, te s ts and repairs to this face

F igu re 5 .1 7 A d d in g a stru ctu ra l cra ck

The data gathered is then passed through the knowledge base and, in accordance w ith the

techniques outlined in the previous section, knowledge bases determ ine the likely causes.

However, due to the difference in geometry and type, this form o f cracking is handled

differently to pattern cracking and spalling.

Any element face into which the user adds a structural crack could be inclined. As any

reinforcing steel is almost universally placed parallel and perpendicular to the edge o f the

concrete element it makes up, the first check the system makes is to see if the crack entered

by the user is parallel or perpendicular to the reinforcing steel. There are two know ledge

bases used to assess structural cracking. The first knowledge base exam ines the

information, looking particularly for evidence o f corrosion and cracks being parallel or

perpendicular to the element edge, and decides if the crack is caused by reinforcem ent

corrosion or structural effects. If the crack is caused by corrosion, then the straight line

227
 ... wiw vAfivn icmiuii^cu uuuuicic unuge repair

added by the user is amended automatically to a rectangular spall region 150mm wide

(150mm being the typical distance between reinforcing steel centres). At this stage the

defect, which was initially a ‘structural crack defect’ but is now a spall defect, is passed

over to the spall knowledge base for assessment as in 5.4.3.1.

5.5 Determining the severity and extent of defects

It has been recommended in this thesis (Chapter 2) that in order for an expert system for

reinforced concrete repair to be intelligent, it must be able to measure the severity and

extent of defects.

Boam110 suggests that there are six possible actions that can be taken as a result of a defect

that has induced corrosion or left reinforced concrete susceptible to corrosion.

• Do nothing

• Reduce corrosion rate

• Repair visible defects

• Carry out major repairs

• Apply cathodic protection

• Replace affected element

For the purposes of the expert system, these options have been simplified into four

categories into which any individual defect can be placed. An intelligent expert system

must be able to place any defect into one of these categories:

228
 x ------------- vajjvii lui i ^ iiiiu i ^ cu luiiuiclc u u u g e re p a ir

1) Do nothing

The significance of the defect in question is of small importance. Durability is not affected.

If left un-repaired, there will be no detriment to the concrete element.

2) Cosmetic

The defect considered had caused damage to the exterior appearance of the concrete

element. There is a long term durability risk. The defect is noticeable and aesthetically

unpleasant. It should be considered as ‘in need of repair’ either for aesthetic purposes, or

for purposes of arresting any further deterioration which could lead to more severe defects.

3) Minor

The defect is significant. The protection the concrete provides to the reinforcing steel has

been compromised and the defect will progressively worsen unless remedial action is

taken. The defect is aesthetically unpleasant. Repairs should be undertaken, although the

defect is not of a sufficient nature as to require urgency.

4) Major

The defect is so severe in its nature and magnitude that there is either an immediate loss of

safety against collapse, or, even if the element is structurally stable, public confidence in

its performance is compromised. Repair should be undertaken immediately.

Each individual defect will be placed into one of these categories by the expert system.

However, once all the defects prevalent on an element have been entered into the program,

at that stage the system should make a decision on the overall action to be taken on the

element as a whole.

In order to be able to place any defect into one of these categories, the expert system

requests information (typically):

229
 ____ I-------- .11 v . t v v / \ | y v i l U J JV V 111 1U I I V l l l l U l U t U t U l iU ^ l^ U llU g C 1C j J d l l

• Proportional size of defect (i.e. size of defect patch relative to size of element it

affects)

• Depth of defect (for spalls)

• Width of defect (for large cracks)

• Type of surface cracking (for pattern cracking and surface deterioration)

• Is reinforcement exposed (how much, how badly deteriorated etc.)

• Moisture condition

• Evidence of corrosion.

5.5.1 Determining the size of a defect

It has been identified that a key requirement of an intelligent expert system is the

judgement of the severity of a problem it may have to diagnose. Existing expert systems

reviewed lacked the intelligence to offer advice on when and if to repair.

The theoretical maximum area of an element that can be covered by a single defect, in

terms of percentages, is 100% - the minimum is obviously 0%. This information is

calculated automatically by the program when the user defines the defect patch (Figure

5.8). At the early stages of data input, the program knows only the type of defect (spall,

pattern crack or crack) and its size (in terms of element coverage, from 0 to 100%).

A technique has been developed which allows the expert system to place any individual

defect into one of the four repair categories.

The technique devised employs a horizontal axis to represent the percentage of the element

covered by the defect. The four possible zones into which a defect can be placed are

230
 .u uiv i iui icimuiv^cu cuiicicic ui luge repair

positioned, in order, onto the axis as triangles with their apexes at pre-determ ined points,

as shown in

Figure 5.18.

A USER-3WT251LF57 ,BMX - bridges - Demo-Bridges - Defect

File Edit View Defect Go Stock Abnormal Loads R eports Customise Help

j i 41 ►I S j X 1M» 13 1j A * o □ j ^ Q ] <s> ^ j <■ ■» + ♦ 1 ± K ©

03 m Inspection: j 157 S cheduled: 12/0572005 ~ F [ Grid Options: 4- ■+ -f- 4-
"£? 41
Jj Pier/column Bridge View] 157 | Pier/column 1 Spall 425 J Photo j Advice]
| Pier/column 1
Ref. (Spail 425 X: |3 o o 3 Width: J 2572
!■■ A
j- ,< Face: [front j d Y: [2524 Height: 12572
Spall 412 Refresh Layout |
<r ' Spall 424
r' Extent: [^ T j Priority: j Low
3
Main Beams Severity: p T j Recemendation: [ d 0 Nothing
wm* Main Beams 1
Condition Index: | it . • <—
2i Bank seat ^ Is action required?
*|* Bank seat 1 Included in bridge
condition? 1* Estimated Cost: [g~
*|* Bank seat 2
Comments:
Recommended
Action:
Action T aken:

Diagnosis:

■Jl B a c k to S tru ctu re

Spall A record has been added to th e Expert Advice History

F igu re 5 .1 8 R ep a ir zo n es fo r 32 m m d eep sp a ll

In

Figure 5.18, the spall has been added to the unwrapped pier elem ent as shown, and the

system has been informed that the spall is 32mm deep. Using the techniques developed in

this chapter, the expert system positions four triangular zones on the display (shown

enlarged below the figure), from left to right, these zones represent the four categories into

which each defect is placed. For example, the ‘Do N othing' zone covers the region from

231
 ,---- -- ------ in Luv o_yoi^iii iui iciiuuiccu ^unuicic uuugc icpau

0% to 4%. The Cosmetic zone covers the region from 3% to 43% (in order to represent

uncertainty in the decision making process, the zones overlap, creating fuzzy boundaries.)

The horizontal scale equates to the defect size as a percentage o f the overall element area.

The size o f a defect added in

Figure 5.18 is 4% (of the elem ent it affects), the defect is, therefore, a ‘Cosm etic’ defect.

The position o f the apex o f the zones changes depending on the inform ation added by the

user, for example, in Figure 5.19, the depth o f the spall (

Figure 5.18) has been altered to 50mm - a more serious defect. As a result the zones have

shifted, and the defect size (4%) now falls into the ‘M inor’ repair zone. This information,

which represents the current thinking o f the expert system based on the inform ation it

holds about the defect, is always visible to the user. As the program ’s inform ation is

increased by further data input, the user sees how its opinion is affected as the zones move.

Defect Severity
Figure 5.19 Spall depth 50mm

The location at which the diagonal arms o f the zone triangles intersect dictates the w idth o f

coverage o f the horizontal axis over which the zone falls. This position o f the intersection

changes depending on the amount o f inform ation the system is supplied with. For exam ple,

immediately after a spall defect is added to the program, the system knows two pieces o f

information - the fact that the defect is a spall, and the size o f the defect as a percentage o f

the overall element size. Due to the limited inform ation, the arms o f the zone triangles

cross close to the apexes, and as a result the overlap between zones is large. This effect

represents the uncertainty and fuzziness. The fact that the expert system 's inform ation is

232
 r ._. „ *n lu i ic m iu iu c u u uiitxeie u n u g e re p a ir

limited is represented by the large overlaps between zones, and as a result defects could

fall into the fuzzy areas between zones, i.e. a defect could be classes as both a ‘do nothing’

defect and a ‘cosmetic’ defect. The program’s ability to determine which zone to place a

defect into when it falls into these fuzzy areas is discussed in more detail in section 5.6.

5.5.2 Map cracking defects

There are two key factors which dictate the severity of a pattern cracking defect: its

coverage of the element, and the pattern cracking image which the user selects to represent

the defect (Figure 5.20 to Figure 5.25). In addition to this information, the user is asked to

judge the cracking and place it into one of three bands: mild, moderate and severe. For

example, if the user selects Image 2, Figure 5.21 to represent the defect, the system knows

that the image represents corrosion induced cracking caused by chloride ingress. The user

is then requested to grade the cracking. However, for some of the images, such as crazing,

it is not considered applicable to divide the defect into these categories in such a way.

During a series of interviews, the industrial experts (concrete repair practitioners) were

asked to select images which best represented the different causes of pattern cracking.

Subsequently, the experts were presented with the definitions of the four categories into

which an individual defect will be placed by the expert system (section 5.5). Precise

definitions of the categories were agreed amongst the panel.

The image representing chloride induced corrosion cracking was presented before the

panel (Image 2, Figure 5.21). The question was posed. “On average; for moderate cracking

of the type shown in the image; how large would the pattern cracking patch be, as a

percentage of the element area, in order for this defect to fall into the ‘Do Nothing’

category?” This question was followed by some discussion and the drawing of diagrams

233
 '5 wajjcii a_ysicm iui icim u rceu concrete oridge repair

showing, to scale, how large a defect covering 5%, 4%, 3% etc. of a typical pier would

look. The experts chose the value 2%.

This exercise was repeated for the other three zones, then further repeated to cover mild

cracking and severe cracking for this particular image. From these sessions the resulting

table was formed (Table 5.1):

Table 5.1 Position of zone apexes for chloride cracking image

Nothing Cosmetic Minor Major

mild 2.5 4.5 10 17.5

moderate 2 3.5 9 16

severe 1.5 3 8 15

For example, the vertical red band in Figure 5.20 represents the size of the defect. In this

example, the user has drawn a considerable defect on the element, the defect covers 30%

of the element area. The system already suspects the cause is chloride corrosion because

the user selected the ‘chloride corrosion’ image (Image 2, Figure 5.21), the zone positions

have, therefore, been set in accordance with the expert recommendations and the defect is

well inside the ‘major repair’ zone. This is as would be expected for a defect covering (in

this example) 30% of the element area.

234
File Edit View Defect Go Stock Abnormal Loads Reports Customise Help

!.j «........ -------

►J --“ - x---- O
r S -----
s
*1 Inspection: (1134 Scheduled: 06/01 /2005 T j Grid Options: +■ "f -f- 4
BR1006
Jj Pier/column
Bridge View j 1134 | Pier/column 1 Map Cracking 465 j Photo j ij.
| Pier/column 1
Ref- [Map Cracking 465 X: f i g Width:
Map Clacking 458
Map Cracking 465 Face: [ j ^ “ T] Y: [ 241 Height: [ 5674'
«'“■ Spall 456 Refresh Layout j
; f t Spall 457
Extent: |X ~ T j Prioritv: jLow
| Pier/column 2
| Pier/column 3 Severity: [7 T j h ac(ion repuired?
Crack 460
0 Spall 453
Recemendation: jDo Nothing
| Pier/column 4 Included in bridge
condition? Estimated Cost:
Deck
■w Deck 1 Comments:
oaf Deck 2 Recommended r—-------------------------------
^ Deck 3 Action:
Abutment Action Taken:
¥ New Element
¥ New Element
Diagnosis: | Drying Shrinkage 5
Parapets
AA Parapets 1
AA Parapets 2

JLi W B a c k to S tr u c tu r e

Ready map cracking A record has been added to th e Expert Advice History

F igu re 5.2 0 L arge m a p -cra ck . U ser se lects 'ch lo rid e co rro sio n ' im a g e

The following figures (5.21 to 5.25) show the images the users can select from to represent

pattern cracking defects. Importantly, the images are not titled. Titles could prejudice a

user’s choice o f image. W ithin the com puter program, the images are identified as

numbers.

F ig u re 5.21 Im ages 1 & 2

Image 1 represents carbonation cracking. Image 2 represents chloride cracking.

235
 ---------- .. ... ..... vAptu nji iciiiiu itcu euneicic unuge repair

F ig u re 5.2 2 Im ages 3 & 4

Image 3 represents cracking from freeze thaw cycles and image 4 is typical o f the early
stage o f AAR.

F igu re 5 .2 3 Im ages 5 & 6

Images 5 and 6 are more advanced forms o f AAR, with image 6 showing the gel-like

deposits which form around AAR cracking.

236
 L A pvi I iw i 1CIU IV JI t c u ^U IIC IC IC U1 l u g e I C p tlll

Figure 5.24 Images 7 & 8

Image 7 shows drying shrinkage in a new repair. Image 8 shows plastic settlement.

Figure 5.25 Image 9

Image 9 shown crazing.

Indications o f the scales at which the images should be viewed are provided. The image

selected by the user, and additional information provided, determine which o f the tables

derived by the expert should be used to establish the position o f the severity zones.

This exercise to determine zone apex positions was repeated for each o f the different

pattern cracking causes and is shown in Tables 5.2 to 5.5.

237
 in iiiv w/vpv^n. a p i c m iu i ic m iu iu c u u u n creie o n u g e re p a ir

Table 5.2 Position of zone apexes for AAR

Image nothing cosmetic minor major

mild 2.5 12 21.5 37

moderate 2 11 20 35

severe 1 10 17.5 32

Table 5.3 Position of zone apexes for Freeze Thaw damage

Image nothing cosmetic minor major

mild 17 62 90

moderate 14 57 85

severe 11 53 79

Table 5.3 shows that regardless of the extent of a freeze-thaw defect, it cannot be

considered as a major repair.

Table 5.4 Position of zone apexs for Plastic Shrinkage, Crazing, and Drying Shrinkage

Defect Image nothing cosmetic minor major

Plastic Shrinkage - 9 64 98

Crazing - 20 98

Drying Shrinkage moderate 7 64 93

238
 m mw vApwi n ji lc im u i^ c u c u n u ie ie u n u g e re p a ir

Table 5.5 Position of apexes for carbonation induced cracking

Image nothing cosmetic minor major

mild 3 9 16 23.5

moderate 2 8 15 22.5

severe 1 7 14 21.5

5.5.2.1 Secondary zone positions

At the first stage of questioning, when the experts were asked to consider the zone apex

positions, their decisions were based on the three pieces of knowledge that the expert

system would have at the initial stages of decision making: the pattern crack size as a

percentage of the overall element size, the image selected by the user to represent the

cracking, and (for some images) a textual description the user was asked to select rating the

severity of the cracking. The experts were told that no further information on the defect

was available at that stage, but importantly, the defect could have associated features such

as staining and spalling. The experts were asked to factor into their judgements,

assumptions based on their past knowledge about what other factors might be affecting the

typical defect.

At the second stage of interaction, the user may enter additional information about the

defects, such as:

• Amount of staining associated with cracking

• Amount of spalling associated with cracking

• Amount of seepage associated with cracking

239
 1 „ . .^ . ,..v..w..b m inv vAjmii iu i iciiiiv jiccu c u n u ic ie u n u g c re p a ir

Each of these factors is judged by the user, using example images to inform their

selections. These factors are judged on a scale from 0 to 100. With 0 representing no

staining (or no spalling, no seepage), and 100 representing what the experts would assume

as the worst incidence of staining etc. which could possibly be associated with the defect.

The expert panel was presented with the following question (with diagrams and graphs to

assist).

“If after the first stage of data input, you judged the apex of the ‘Major’ zone to be at 10 (a

very severe defect). How severe would you expect corrosion staining to be (on a scale of 0

to 100) for you not to change your opinion regarding the position of the apex?”

The experts discussed their answers. If the user indicated that corrosion staining at the

defect was rated as 100%, then the defect would be worse than the experts had assumed at

the initial stage - if this was the case, the position of the apex of the Major zone would

change, perhaps from the previous 10 to 8. As a result, smaller defects will fall into the

Major repair zone. Figure 5.26 and Figure 5.27 demonstrate this graphically for a spall

defect (crack defects also use this technique).

240
 M U5ER-3WT2 5 1LF57\BMX - bridges - Demo-Bridges - Defect alaJ xi
File Edit View Defect Go Stock Abnormal Loads Reports Customise Help

|j 4 > I % M. j X I aa j # [j A. ' e> □ j O 1o a j » ■» t ♦ [t !-► ®

Inspection: j 157 S cheduled: 12/05/2005 ~ y | Grid Options: + t 4- snap

U 41
Pier/column BridgeViewj 157 j Pier/column 1 Spall 412 ] Photo j Advice j
| Pier/column 1
Ref. fspall 412 ~ X: Width: f42§7
f t
m _____ F a c e : J right Y: 11380 Height: j 2191
M1I5H Refresh Layout )
% Spall 42G
Extent: CjTj Priority:
Spall 427 flo w 30
D Main Beams Severity: JP ~T] Recemendation: j Do Nothing H
w Main Beams 1
Condition Index: I
D B an k seat ^ Is action required? r
•|* Bank seat 1 Included in bridge
condition? I*’ Estimated Cost: jg~
¥ Bank seat 2
Comments:
Recommended
Action:
Action T aken:

D iagnosis: j chloride corrosion 10

“ 3
If

m m m m m “ •— m •
B jg B B S H H i S 3

nKWH HH
B a c k to S tru ctu re

Ready Spall A record has been added to th e Expert Advice History

Figure 5.26 40mm deep spall

Figure 5.26 shows a 40mm deep spall. The apexes o f the zone triangles have been

positioned in accordance w ith section 5.5.3. The spall covers approxim ately 5% o f the

element - as a result it falls into the Cosmetic repair zone. In Figure 5.27, the user has

added additional inform ation - that 100% o f the reinforcement is exposed and severely

corroded. As a result o f this extra information, using the technique outlined below, the

zone apex positions shift, and the defect is now rated as a M inor repair.
 i& in uiv i<ji lcim uiccu cuucieie unuge repair

H H M H M M aiM -1*1 xl
File Edit View Defect Go Stock Abnormal Loads Reports Customise Help

4Q % ~ K o ffl# -o *? ; . ® E «■ + ♦ + ± ®
I'- ...... .... - ............. .1.......... ........-----------------------
I - ' i . . ..
't
Inspection: 1157 S cheduled: 1 2/05/2005 j £ j Grid Options: + t +
VI 41
Pier/column Bridge View j I57 j Pier/column 1 Spall 412 | Photo I Advice!

1 Pier/column 1
Ref- |Spaii7T2 X: [5193 Width:
4k Face: [ - ^ T j Y: Height: f ^ f
I....^
j...~ gag M H I l f e l Refresh Layout |
gig Spall 42S
Extent: T j Priority: [ t^ T T j
B Spall 427
Main Beams Severity: Jlj Tj Recemendation: ) Do Nothing
Main Beams 1
--------- Condition Index: I
Bank seat ^ Is action required? r
*|* Bank seat 1 Included in bridge
condition? I* Estimated Cost: Jg~
•f* Bank seat 2 ■
Comments:
Recommended
Action:
Action T aken:

Diagnosis. Chloride corrosion 10

xzx: • i •
P M
i

B a c k to S tru ctu re

Ready Spall A record has been added to th e Expert Advice History

F igu re 5.2 7 40m m d eep sp a ll w ith ex p o se d , c o rro d ed rein fo r c e m e n t

If the user indicated that there was no corrosion staining associated with the defect, then

the experts agreed that the defect would not be as serious as they had assumed, in this case

the apex o f the M ajor zone might move, for example, from 10 to 14. As a result, larger

defects might not fall into the M ajor zone, but may be rated as M inor defects. Finally the

experts agreed that for the question asked, a factor o f approxim ately 37.5% would not

make them change their initial opinion about the severity o f the defect and, therefore, the

position o f the apex o f the M ajor zone would not change.

The initial question was repeated, again for the M ajor zone, but this time it was assum ed

that after the first series o f data input, the apex o f the zone was positioned at 30%. The

experts repeated the exercise and decided that a secondary effect from corrosion o f 33%

would not make them change the apex position o f the major repair zone from its initial

position o f 30%.

242
Furthermore, the process o f questioning was repeated for the three other repair zone

categories. The experts’ opinions were plotted graphically, this is shown in Figure 5.28.

45

40
Do N o th in g

35
M in o r
M a jo r
30

25

20

15

10

0
0 20 40 60 80 100 1 20

Initial Zone apex position

Figure 5.28 Secondary zone movement graph

In effect, the lines in Figure 5.28 represent how severe the experts expected the defect to be

when they were presented with only limited information. For example, after the initial data

input stage, the expert system has two pieces o f information: the size o f the defect in

relation to the size o f the element, and the representative image chosen. It is these two

pieces o f information w hich the experts used to position the zone apexes (beginning Table

5.1), however, whilst knowing only these two pieces o f information, the experts made

conscious judgm ents about the likelihood o f the presence o f other indicators o f concrete

distress (staining, corrosion, exposed reinforcement etc). Therefore, when deriving the

zone positions the experts anticipated the presence o f these ‘secondary effects’.

Henceforth, it is possible for a particular defect to exhibit secondary effects to a lesser or

 r .~. muiviug in niv vApvii aj'aicm iu i ic iu iu ic c u c o n c re te o riu g e re p a ir

greater degree that those initially estimated to be present. The experts’ assumptions, now

obtained graphically, can be used to amend the initial positions of the apexes of the zones

as further information becomes available. If a defect exhibits worse secondary effects than

allowed for in the initial zone positions, then zone positions shift, and, for example, where

a defect may have fallen into the Minor repair zone it would thereafter fall into the Major

repair zone.

The lines in Figure 5.28, can be represented by their slopes and intercepts as shown in

Table 5.6.

Table 5.6 Secondary zone movement constants

Slope Intercept

Do Nothing -0.1 10

Cosmetic -0.16 18

Minor -0.19 27

Major -0.25 39

The following method is used to amend the zone apex positions when secondary

information becomes available. Each zone apex will be moved by an amount equal to the

difference between the assumed severity of spalling, cracking, staining etc. and the actual

severity entered by the user (multiplied by a factor).

For example, say the user has selected the ‘Carbonation cracking’ image and rated the

cracking as moderate. The apex of the Major repair zone is set to 22.5 in accordance with

Table 5.5. The user goes on to rate the amount of staining at 75%. Using Table 5.6, the

amount of staining the experts anticipated would be present can be determined as follows:

244
 iiiu am g m uiw a^aicm iui lc m iu iu c u u u n creie u n u g e re p a ir

Expected amount of staining = slope * apex position + intercept

= (-0.25 * 22.5) + 39

= 33.4%

It was agreed with the experts that the zone apex should be moved a distance equal to the

difference between the anticipated secondary defect severity and the value entered by the

user (multiplied by a factor).

Difference between anticipated and actual staining = 7 5 -3 3 .4

= 41.6%

This figure is then multiplied by a factor depending upon the particular secondary effect.

Under initial review, these factors perform satisfactorily at 0.2. Though there was

agreement amongst the expert panel that these factors should be calibrated during field

testing of the program (Table 5.7).

Table 5.7 Adjustment factors for secondary zone movement

Secondary effect Adjustment factor

Staining 0.2

Spalling 0.2

Seepage 0.2

Therefore the total zone apex adjustment for the example given:

Adjustment = 0.2 * 41.6 = 8.3%

As the defect is more serious than anticipated, the zone apex is adjusted thus:

New apex position = Original position - 8.3

2 2 .5 -8 .3 = 14.2

This operation is completed for all four zones.

245
 1— _ ----- — in mw v-ivp^n iui i^mnji^cu tun^icic unugc repair

5.5.3 Spall defects

Each spall defect entered into the expert system by a user will be placed into one of the

four repair categories. Whereas pattern cracking uses information on images selected by

the user to define the apex positions of the zones; zone apex positions for spall defects are

determined, initially, based on spall size and depth.

A question was asked of the expert panel. “If the size of a spall was 2% of the element

area, what would the depth of the spall have to be in order for you to class the defect in the

‘Do Nothing’ zone?” The expert panel used graphs and diagrams to reach a decision, and

the process was repeated for different spall sizes and different zone types. This session of

questions delivered the graph shown in Figure 5.29.

246
for reinforced concrete bridge repair
Chapter 5 - Decision making in the expert system

oo
re
CD
re
+■<
c
CD CD
E

247
©
CD
O
CD (D
(D

CD
O)
+re
■>
c
CD
oi-
CD
Q.
</>
re
CD
N
C/> o
CL
75 as
o. o
(/) N

Cl.
4)
T5

03
C.

<u
Cl
X
UJ
**= O n
o o o o o o <N
o o o o o ■ri
4t
•_
( lu lu ) ipdap ||e d s 3
.£/°
u_
 J ----- _ --------------------------------- . . . w .w J jro iv m lu i l^ llllU lG C U G V J11G 1C IC LU l U g C I C p t l l l

When plotted graphically, the interaction between spall size and depth as determined by

the expert panel could not be represented mathematically by one curve - it requires a curve

and a line. Therefore, for a spall, each zone in Figure 5.29 is represented by a line showing

the experts’ answers to the question posed previously. In turn, these lines are represented

mathematically by two equations; one equation representing the straight line portion of the

zone line, and one representing the curved portion of the zone line. Separate equations

were fitted to the curved and straight parts of each line. Depending on the size and depth of

a spall, the program uses these equations to determine the positions of the apexes of the

four zones which categorise the severity of the defect.

For each repair zone, there is a position along the y axis where the line and curve definition

of the relationship between depth and size (for each zone) meet (Table 5.8).

Table 5.8 Position where decision line and curve meet

Zone Intersection (spall depth mm)

Do Nothing 18

Cosmetic 28

Minor 45

Major 80

If the depth of the spall defect as entered by the user is lower than this intersection point,

then the straight line relationship can be used to determine the apex position of the required

zone. The constants describing the straight line portions of the relationships between zones,

spall depth and size as determined by the experts are shown in Table 5.9.

248
 — -— ... wiw vajjvu iui lum iui^cu uunuicic unu g c icjjctn

T able 5.9 Equation o f line to determ ine zone apex positions for spalls

Zone Slope Intercept

Do Nothing -0.1907 18.4837

Cosmetic -0.2277 30.2789

Minor -0.3343 48.4303

Major -0.6344 88.4455

For example, say a spall defect is added to the expert system, and the user informs the

system that the defect is 25mm deep. To determine the apex location o f the Cosmetic zone,

the program first determines if the depth o f defect is less than or greater than the

intersection point.

Intersection point, Cosmetic = 28. Depth o f defect = 25. Therefore, 25<28 and straight line

relationship can be used.

Apex of Cosmetic zone = (Spall depth - intercept) / slope

= (25 - 30.279) / -0.228

= 23.15

Therefore, the Cosmetic zone apex is positioned at 23.15 (meaning that 25mm deep spalls

covering 23.15% o f the element area are classified as ‘cosmetic defects’) - any other zones

whose apexes can be positioned by the straight line portions o f the relationships derived by

the experts also have their apexes located by this method.

If the depth of the spall defect as entered by the user is higher than the intersection point,

then the curved portion o f the relationship line can be used to determine the apex position

o f any zone as required. The constants describing the curve line portions o f the

24 9
 j ~ . . . u n i n g in nAjjdi i ojoivm iui ic im u i^ c u i^uuuicic unugc icpan

relationships between zones, spall depth and size as determined by the experts are shown

in Table 5.10.

T able 5.10 C onstants (K and S) for equations o f curve for determ ining zon e apex positions for spalls

Zone K S

Do Nothing 1.55 282

Cosmetic 1.3 485

Minor 1 1923

Major 0 .8 11340

Continuing the previous example, where a spall defect is added to the expert system, and

the user informs the system that the defect is 25mm deep. To determine the apex location

o f the Do Nothing zone, first determine if the depth o f defect is greater or less than the

intersection point.

Intersection point, Do Nothing = 1 8 . Depth o f defect = 25. Therefore, 25>18 and the

curved line relationship can be used.

f log (Spalldepth - intersection)N (l o g s ' !

Apex of Do nothing zone
-K J 1 -K J

= ^lo g (2 5 -1 8 )^[ f log282

, -1 .5 5 J 1-1.55,

= 12.29

The apex of the Do Nothing zone would be set at 12.29. This indicates that, because o f the

low spall depth, the defect is not considered too severe.

250
 £ aaa HIV VA^WII LJJ kJlSslll 1U1 1V^llliWl t t u CUUl/ldC UllUgC 1CjJClil

-
5.5.3.1 Secondary zone positions

As the program gathers more information about the nature o f a defect, its decision about

the severity of the defect becomes more intelligent. Say a defect is entered into the system

and its size is judged at 2% of the overall element area. Then the user enters the defect

depth (for a spall), the system knows how spall depth affects the positioning o f the

‘severity zones’. If a depth of 70mm is entered, the zone positions might place the defect in

the ‘Cosmetic’ zone; an indication that the defect requires attention. As already discussed,

the position of zones has been predetermined based on the opinion o f the concrete repair

experts. However, zones continue to shift as additional information about the defect is

entered until finally the information collection is complete.

Shifting zones depending on exposed reinforcement

For every spall defect, the system assumes horizontal and vertical reinforcing steel is

present (either exposed or still embedded) below the substrate surface. The program

calculates the total length of reinforcing bar within the defined area o f the spall, assuming a

200mm spacing between bar centres both horizontally and vertically. The program then

equates this total length o f reinforcement within the boundary o f the spall defect with a

variable known as the ‘maximum possible exposed reinforcement’. Next, using a graphical

technique, the user informs the program the actual length o f exposed reinforcement and the

program converts this to a percentage o f spall reinforcement exposed, based on the

maximum possible that could have been exposed, computed by the expert system.

251
 j— . - ^ .v .u .u n ..iuiun6 in niv^ w/\|jwii. aj'oi.wii itii ic im u ic cu cuiiL/icic un u g e repair

This information on the amount o f exposed reinforcement is used to move the decision

zones in the same way as secondary factors moved ‘map cracking’ zones in section 5.5.2.1.

The expert panel was presented with the following question (with diagrams and graphs to

assist).

“If after the first stage o f data input, you judged the ‘Major’ zone to be at 10 (a very severe

defect due to a high spall depth). How much exposed reinforcement would you expect (on

a scale o f 0 to 1 0 0 ), for you not to change your opinion regarding the position o f the

apex?”

The experts discussed their answers. If the user indicated that exposed reinforcement was

rated as 1 0 0 %, then the defect would be worse than they had assumed at the initial stage -

if this was the case, the position of the apex o f the Major zone would change, perhaps from

the previous 1 0 , to 8 .

If the user indicated that there was no exposed reinforcement associated with the defect,

then the experts agreed that the defect would not be as serious as they had assumed, in this

case the apex of the zone might move, for example, from 10 to 14. As a result, larger

defects might not fall into the Major zone, but may be rated as Minor defects instead.

This exercise was repeated for different zone apex positions and the three other different

zone types. A graph (Figure 5.30) was developed very similar to that used for pattern

cracking and eventually it was decided that the same graph was applicable for both cases.

252
 /vpvi t oj'diwiu iui ic m iu iL c u uunureie unuge repair

0 20 40 60 80 100 120

Position o f Zone after spall depth

Figure 5.30 Secondary m ovem ent o f zone for spalls

In effect, the lines in Figure 5.30 represent how severe the experts expected the secondary

factors affecting a defect to be w hen presented w ith only lim ited inform ation. For exam ple,

for a spall depth o f 50m m , the initial question asked o f the experts w as h ow large a defect

should be to be classed as a m ajor repair. H ow ever, at this stage, they w ere told the spall

could have other factors w hich m ay affect the ju d g em en t o f severity, but at the first stage

o f data input they w ould have to assum e how serious those other factors m ight be - factors

such as the am ount o f exposed reinforcem ent and the degree o f corrosion o f the

reinforcem ent - therefore, they m ade assum ptions w hen initially determ ining the apex

position o f the zones. T hose assum ptions, now obtained graphically, can be used to am end

the positions o f the apexes o f the zones as the additional inform ation becom es av ailab le

through additional user data input.

The lines in Figure 5.30, can be represented by their slopes and intercepts as show n in

T able 5.11.

253
 '& iui ic m iu ic c u uuncieie u nuge repair

Table 5.11 Secondary zone movement constants

Slope Intercept

Do Nothing -0.1 10

Cosmetic -0.16 18

Minor -0.19 27

Major -0.25 39

The following method is used to amend the zone apex positions when secondary

information becomes available. Each zone apex will be moved by an amount equal to the

difference between the assumed amount o f secondary defects and the actual amount

entered by the user (multiplied by a factor).

For example, say the user has informed the system that the spall depth is 50mm. The apex

o f the Major repair zone is set to 60.6 in accordance with Table 5.8. Say the user goes on to

rate the amount of exposed reinforcement at 75%. Using Table 5.11, the amount o f

exposed reinforcement the experts anticipated would be present can be determined:

Expected amount o f exposed reinforcement = slope * apex position + intercept

= (-0.25 * 60.6) + 39

= 23.85%

It was agreed with the experts that the zone apex should be moved a distance equal to the

difference between the anticipated amount o f exposed reinforcement and the value entered

by the user (multiplied by a factor).

Difference between anticipated and actual exposed reinforcement =15- 23.85

= 51.15%

This figure is then multiplied by a factor. Although the amount o f exposed reinforcement

has been used in the example, other secondary defects can affect zone positions (staining,

254
 w ^ umiviii^ axi uiw va^vi i ajroiA/iu ivji itllllUltCU CU111/1CIC U1lUgC lCJJclll

condition of reinforcement, seepage). Therefore, the amount o f zone movement determined

(the difference between expected and anticipated secondary defects) is factored depending

on the particular secondary effect. After initial reviews, these factors were set as shown in

Table 5.12. Though there was agreement amongst the expert panel that these factors should

be calibrated during field testing of the program (Table 5.12).

Table 5.12 A djustm ent factors for secondary zone m ovem ent

Secondary effect Adjustment factor

Amount of exposed reinforcement 0.5

Condition o f exposed reinforcement 0.5 * percentage of exposed reinforcement

Seepage 0 .2

Staining 0 .2

Therefore, the total zone apex adjustment for the example given:

Adjustment = 0.5 * 51.15 = 25.58

As the defect is more serious than anticipated, the zone apex is adjusted thus:

New apex position = Original position - 25.53

6 0 .6 -2 5 .5 3 = 35.07

This operation is completed for all four zones.

It is o f crucial importance to the development o f the system that although the adjustment

factors in Table 5.12 were estimated by the experts as accurately as possible, all were in

agreement that these figures should be calibrated during field trials o f the software.

The effect o f the condition of the exposed reinforcement has to be factored by the amount

o f reinforcement that is actually exposed. For example, if the amount o f exposed

255
 -
---
---
---L-
---
---
---
--_ —_— w* in Hiw > JJ JkViii 1UI iVllllUlWUU WUlIV/Xdt U1AUgt xcpcill

reinforcement is rated by the user at 50, and the condition o f exposed reinforcement is

rated at 50 (still with a spall depth of 50mm):

Expected amount of exposed reinforcement (Major) = slope * apex position +

intercept

= (-0.25 * 60.6) + 39

= 23.85%

Difference between anticipated and actual exposed reinforcement = 50 - 23.85

= 26.15%

Adjustment for amount o f exposed reinforcement (Major) = 0.5 * 26.15 = 13.075

Expected condition o f exposed reinforcement = (-0.25 * 60.6) + 39

= 23.85%

Difference between anticipated and actual exposed reinforcement = 50 - 23.85

= 26.15%

Adjustment for condition o f exposed reinforcement (Major)= 0.5 * percentage

exposed reinforcement * 26.15

= 0.5 * 0.5 * 26.15

= 6.54%

Total adjustment for Major zone = 13.075 + 6.54 = 19.62

New position of Major zone apex = 60.6 - 19.62 = 40.98.

Importantly, if the user was also to specify that there was associated staining (or any other

secondary defect), the apex position used in the equation to determine the expected amount

o f staining is the original apex position before the effect o f any other secondary defects

affecting the zone apex position was implemented.

256
 — u mi wi i ^ iii i uv v a p w t j ^ o i v u i iu i i v /im u iv ^ u t u u u c i c u i l u g e i c p c iii

5.5.3.2 Spalls of unknown depth

There may be occasions where the user can identify a spall outline but the depth o f the

spall is not clear. The expert panel recommended that the program should accommodate

this possibility. For example, a person inspecting the central pier o f a motorway bridge

may be able to see a spall, but unable to judge its depth accurately until such a time when

traffic management can be arranged to allow close up inspection. In these cases, the expert

panel was asked to position the severity zone apexes in a similar fashion to how pattern

cracking apex positions were determined. The user is asked to judge the probable cause

(Table 5.13).

Table 5.13 Position o f zon e apexes for chloride ingress and carbonation spalls (unknow n depth)

Spall cause nothing cosmetic minor major

Chloride ingress 1 2.5 4 6

Carbonation spall 2.5 5 8 12

5.5.4 Structural cracks

For each o f the three main types o f defects that can affect reinforced concrete, two factors

are used to determine the initial positions o f the severity zones. For pattern cracking these

were an image chosen by the user and the size o f the pattern cracking patch. For spalling,

spall size and depth. Structural cracking uses a similar method - it uses two variables,

crack size and crack width, to set the initial zone positions.

For a typical pier element like that shown in Figure 5.31 the program knows the pier

height, x. It also knows the total unwrapped length o f the pier, y, as shown in Figure 5.32.

257
Figure 5.31 Typical pier element

F igure 5.32 U nw rapped pier

When the user adds a crack onto the element (as in Figure 5.17), the system considers the

crack as if drawn on a canvas o f dimensions x, y (Figure 5.33).

F igure 5.33 Elem ent canvas for crack size determ ination
 1-------- ~ ..1 ajroiwu 1UI 1C1U1U1CCU WU11WICIC UI lUgC I CpUH

The expert system projects the crack onto the edges of the unwrapped element in order to

obtain the dimensions p and q, as shown in Figure 5.34.

q
Figure 5.34 P rojecting crack onto elem ent edges

The program determines the variable, size o f crack using the following expression:

(p+q) / (x+y) * 1 0 0 = size of crack.

It was decided during a series o f interviews with experts that the width o f a crack and the

depth of spall could in some way be related, in terms o f how seriously they affect an

element. A question was asked of the experts. “If all you knew about a spall was its depth

o f 40mm (no information about the extent o f the spall itself), how wide would a structural

crack have to be for you to be as concerned about the crack defect as you were about the

40mm deep spall defect?” Although the experts were in agreement that the premise o f the

question was unusual, they clearly understood how the information they were providing

was being used in the expert system. They reached the conclusion that, in the absence o f all

other information, they would be equally concerned about a 40mm deep spall and a 4mm

wide crack. It was established through similar questioning, that this factor o f 10 could

259
 ----------J------------— — .1 1 m v w a j ; v i i > JJ JIW111 1U1 1 V1U1U1 t t u L-U11L/1 U1 lU g C 1 CpClll

generally be employed to relate the seriousness o f all spall depths and crack widths. It was

agreed that the factor should be calibrated during field trials o f the system. The equations

determining the zone apex positions o f spalls (section 5.5.3) were examined to test their

applicability to assessing the severity o f structural cracking. Using the graphs and

equations in section 5.5.3, the variable ‘size o f crack’ replaced ‘size o f defect’ (which was

measured as a percentage of the overall element area). The variable ‘spall depth’ is

replaced by the width o f the crack multiplied by ten. The position o f the apexes o f the

zones are then determined exactly in accordance with the procedures for spalling.

Similarly, zone movement through the addition o f secondary defects is governed in the

same way as for spalling.

5.5.5 Miscellaneous defects

Miscellaneous defects require no diagnosis by the expert system. They are defects which a

laymen could reasonably be expected to identify, and their cause is generally effects o f

workmanship.

5.5.5.1 Blow Holes, Sand Streaking

If the user of the expert system noticed an area o f blow-holes or sand-streaking, they would

add a patch o f blow-holes or sand-streaking to the element. The apexes o f the zones are

positioned in accordance with Table 5.14.

260
 1.1 uiv V/A)JVU j;o i u u 1U1 IV^UHUll^tU I/UIIUICIC UUUgC ICJJclll

Table 5.14 Zone apex positions for blow-holes or sand-streaking

Image nothing cosmetic minor major

- 20 98

5.5.5.2 Honeycombing

Similarly, areas o f honeycombing observed by the expert system user would be added

directly onto the concrete element. Honeycombing in large volumes can be regarded as

serious, hence for this defect all four severity decisions are possible. The position o f the

zone apexes is determined in accordance with Table 5.15.

T able 5.15 Zone apex postions for honeycom bing

Image nothing cosmetic minor major

moderate 1.5 5.5 10.5 18.5

5.6 Uncertainty in deciding severity.

Any single defect, be it spall, pattern cracking or a structural crack, can at any time (within

the expert system) be in two distinct states:

Complete - all information has been entered for that defect. For example, for a spall, the

user has added the spall to the element on screen (so the expert system is aware o f the

spall’s size and location), has entered the depth, shape, amount and condition o f exposed

261
 ^ ly w u iu u iiitiivmg in ^ajjcii. s_yaicm iui ic u n u rc cu con creie o n a g e repair

reinforcement and also informed the system if the spall is within splash zones or close to

trafficked carriageways.

In Progress - The defect is currently being entered into the system. For example, for a

spall, the user may have entered the spall size and depth but no further information.

Once a defect is ‘Complete’ it is at this stage that the system calculates the defect’s effect

on the ‘Element condition’ (section 5.7) and also calculates a variable called ‘Uncertainty’

based on the completeness o f information offered. Also at this stage, the knowledge base

will diagnose the defect.

The key area where uncertainly becomes important in the expert system is when a defect,

on the severity graph, overlaps between two severity zones. For example, the defect in

Figure 5.19 lies in both the ‘Cosmetic’ and ‘Minor’ severity zones. The fact that defects

can lie in two zones is considered a benefit o f the technique developed and not a drawback

as it simulates the decision making process o f a human expert whose judgement gets

confirmed as more data about a flaw becomes available. Assessments o f defect severity are

not purely scientific, and often a degree o f estimation based on expert judgement is

employed in decision making.

If a defect does lie in two zones, this represents the fact that the system is not certain about

the severity o f the defect, and, it has still to make a decision.

The case shown in Figure 5.35 is taken as an example. Figure 5.35 shows an enlarged

diagram o f a defect falling into two repair zones, say Do Nothing and Cosmetic. (The

display of the zone positions shown to the user is necessarily small within the software).

262
 irvmg, in ui^ cA p cn sy s ic m iui ic u u o r c e u con crere o n a g e repciir

<■ >
P
F ig u re 5.3 5 H a n d lin g u n ce r ta in ty

W here h = height to zone apex

u = height to zone intersection

p = w idth o f zone overlap

di = distance o f defect into rightm ost zone

d 2 = distance o f defect into leftm ost zone

In order to determ ine into w hich severity zone to place the defect, the expert system first

determ ines the variables:

d f p = % distance o f m arker into C osm etic zone

d 2 /p = % distance or m arker into Do N othing zone

The difference betw een these tw o variables helps to determ ine the dom inant zone

(dj/p) - (d2/p) = V

The height h, is not in itse lf significant in the technique to determ ine severity zone

positions. H ow ever, the ratio betw een the height h and the height o f the zone intersection

u, is im portant. A com bination o f V and u will decide the zone classification o f a defect

falling w ithin tw o zones.

The possible range for the variable V is betw een 100 and -100.

263
 ^ i^wvuiuii iin.iiviiig, in i/Ajjcu aysicm xui icixiiuiLcu to n c re ie onixge repair

The value u represents the height at which the arms of adjacent zone triangles intersect. It

is measured on a vertical scale from 0 to 100. The apex o f the zones is positioned at 100

units above the baseline. Initially, for spalls, pattern cracks and structural cracks the zone

intersection value (or the uncertainty), u, is 60. That is, the lines intersect 60 units above

the baseline. This is the value o f u immediately after a defect has been added. This value o f

u begins to reduce as more information is added, and the expert system can be more certain

about its decision. The value o f u is reduced when new information is entered according to

the Table 5.16:

T able 5.16 C hange in uncertainty as data is added

Information entered Reduction in u (units)

Spall depth 5

No exposed reinforcement 15

Exposed reinforcement 5

Condition o f reinforcement 5

Width of crack 5

Splash zone question answered 5

Staining question answered 5

Seepage question answered 5

As data is added to a defect, the value o f u changes. The value o f h is constant at 100.

Figure 5.36 shows the graph which finally decides into which severity zone a defect lying

in two zones should be placed.

264
 .[jvii. nji h, huuilcu (.u n^ icic ui iug,c icp a n

-70-
GO RTGH
m-
3

-40-
GO LEFT

20-
-

-100 -80 -60 -40 -20 0 20 40 60 80 100

V

Figure 5.36 Zone decision graph

For exam ple, say the tw o zones into w hich a single spall defect falls are D o nothing and

C osm etic. Say the defect is 30% into the C osm etic zone (dj/p = 30) and therefore 70% into

the Do N othing zone (d 2 /p = 70).

V = 30 - 70 = -40

A fter the defect w as initially entered the value o f u is set to 60% . B ut say the user has

entered the spall depth and inform ed the system that there is no exposed reinforcem ent.

In accordance w ith Table 5.16 this gives a 20 point reduction in uncertainty, hence:

U = 60 - 2 0 = 40

U sing the graph in Figure 5.36, w hen u = 40 and V = -40, the decision is borderline. In this

instance the expert system is conservative and selects the w orst case. Flence the defect is

placed in the C osm etic classification. The final classification (w hich, if a defect lay w ithin

tw o zones is determ ined using this technique) is utilised by the know ledge bases (e.g. the

action table rule in Figure 5.15)

265
 ,------ — x.x uiv w/yjjwi j^oiuin iui ic n u u itc u cunuicic unugc repair

The expert system uses the equation o f the line in Figure 5.36, the equation o f the line is

u = l *-V.

5.6.1 Types of uncertainty

Two distinct types o f uncertainty have been identified which are addressed by the expert

system these are called expert uncertainty and system uncertainty.

5.6.1.1 Expert Uncertainty.

If the system has been presented with all the information it needs in order to make a good

decision about the severity o f a defect, on occasions, due to the nature and extent o f the

information provided, it may still be unclear into which severity zone a defect should be

placed. Instances can arise where even the expert would be unsure; the engineer could be

certain about the nature o f the problem, but uncertain about which option to choose. Under

the technique developed in this project, such Expert uncertainty can represented as

horizontal uncertainty - as it falls between two zones, this represents a region where even

an expert might have some conflict over how to rate a defect.

5.6.1.2 System Uncertainty.

Upon inspecting a defect, an engineer gathers some information immediately and

concurrently; size, geometry, location, staining, seepage etc. This is the information on

which the engineer will immediately build up an impression about the defect. No expert

system can take this human approach - expert systems receive information piecemeal, and

it is this function where the greatest contrast between the expert’s approach and that o f the

expert system is seen.

266
 r -— w „ 11, my ha| jvi l hji n -nuunicu uuui/icic ui luge icp an

The system takes in information more slowly (although the intake o f information can be

seconds apart, this is clearly slower that the instantaneous intake o f a human). So the

system clearly takes longer to make its decision. It has passages o f time when it knows it

does not yet have the complete picture and that more information is about to be received.

However, in the developed system, the user can still see how each piece o f information is

affecting the final decision. As each further piece o f information is entered, the system

becomes more certain about the accuracy o f its decision. This is system uncertainty.

System Uncertainty is indicated on the vertical scale as the height o f the intersection o f

adjacent zones. As more information is entered into the system, the intersection o f adjacent

zones lowers.

5.7 Contribution o f each defect to element severity

Each defect on an element must in some way contribute to the overall condition o f the

element. A technique has been developed that allows the effect o f each individual defect

on an element to be assessed.

A panel o f concrete repair experts were asked the following question “Consider that the

condition of an element can be rated between zero and one hundred. With zero being an

element without defects and one hundred an element so severely affected by defects that it

is structurally and aesthetically redundant. If a single defect was diagnosed as being caused

by chloride ingress, and that defect covered 2%o f the element area, how much would this
degrade the element condition on the scale o f zero to one hundred?”

267
 J aaa in v j^ o k v m iu i l V1111U1 W tU U1 l U ^ C I C p C lll

-
The unanimous answer to this question was that the effect on the element condition

depends on the severity o f the defect. Therefore, a method was developed to numerically

rate the severity o f each defect.

5.7.1 Effect of pattern cracking on element severity

In order to determine the severity o f a pattern cracking defect, a user selects an image

which best represents the defect (section 5.5.2). The user may also be required to judge the

severity of the defect in comparison with the image they selected. It is after this stage that

the positions of the zone apexes are set in accordance with section 5.5.2. Say the user has

selected the image which represents chloride induced corrosion and rated the defect as

moderate. The position o f the apexes which the expert system then sets can be said to be

the apex positions o f the typical moderate chloride corrosion pattern cracking defect. The

user may go on to add additional information and zone apex positions may change,

however, the original zone positions before the additional information was added will need

to be referred back to at future stages.

To begin the process o f determining the effect o f a pattern cracking defect on the element it

affects, firstly, the expert system must determine how far the defect marker (representing

defect size) falls into the assessed severity zone o f the defect. In the example o f Figure

5.37 (which shows a prototype of the severity marker), the user has inputted into the

system all the necessary information pertaining to the pattern cracking defect.

268
 . . . . . . „ VA[JVU l\_/x i v . l l l i u l l . t u UI l u g e 1CJJC111

■ii. D e fe ct Graph

100-80

Figure 5.37 D eterm ining effect o f defects on elem ent

The defect falls into the ‘C o sm etic’ severity zone. N ext the system calculates how far the

defect is into that zone. For exam ple, the cosm etic repair zone begins at 10 and ends at 70

(a defect w hich is not particularly severe), and the defect size is at 15 (so the defect covers

15% o f the elem ent surface area). The total w idth o f the zone:

C osm etic Zone w idth = 70 — 10 = 60

P ercentage o f m arker into cosm etic zone = (15-10)/60 = 8.3% .

N ext, consider the position o f the zone apexes after the initial stages o f data entry (w hen

the system knew the im age selected by the user to represent the pattern cracking defect and

the user ju d g ed severity rating). These zone apex positions w ere set in accordance w ith

Table 5.1 and they represent, in the opinion o f the expert panel, the zone p o sitions for the

average m oderate pattern cracking defect probably caused by chloride ingress. T hereafter,

take the distance that the defect w as calculated to in the C osm etic zone (8.3% ), and

determ ine w hat the size o f the defect w ould be if it w as 8.3% into the C osm etic zone for

the average m oderate pattern cracking defect caused by chloride ingress.

The expert system determ ines the initial zone position o f m oderate chloride corrosion, i.e.

w here the C osm etic zone arm s initially hit the axis based on T able 5.1 . T hese points are

2.5 and 6.5. T herefore, the w idth o f the C osm etic zone at this stage is 4.

The system determ ines that 8.3% into a zone 4 w ide = 4 * (8.3/100) = 0.332.

269
 — ^ muivinj, ill mw o^oiwil HJ1 ItllllUltCU V/UI1UICIC UllUgC ICpftll

Therefore, the system determines a variable called the ‘effective size o f defect’. A very

severe defect covering 5% of an element could be said to be equivalent, in terms o f its

effect on the element, to an average defect covering 10% o f the element. This concept

defines the variable ‘effective size o f defect’ well.

Taking the distance the moderate chloride pattern cracking defect falls within the Cosmetic

zone after all information has been added (8.3%) this is then compared to a defect 8.3%

into the Cosmetic zone after the initial zone positions were set in accordance with Table

5.1 - the equivalent average defect.

Size o f equivalent average defect = Zs + (Da * Zsw)

Where Zs = Position where left hand leg o f severity zone in which the defect has been

classed intersects with the horizontal axis for the average defect.

Da = distance defect falls into severity zone after all information was entered

Zsw = width o f zone at initial stage

Size o f equivalent average defect = 2.5 + (0.083 * 4)

= 2.832%

Therefore, for the given example, the defect size is 15% o f the element surface area, but

because the defect is not severe, the zones reflected this by spreading out to the right. The

calculations determined that an average defect covering 2.832% o f the element would have

been as severe as the registered defect (which was less severe than average, but covered

8.5% o f the element).

With a technique for comparing defects o f any size and severity with an ‘average defect’

developed a question was once again asked o f the expert panel. “Consider that the

condition o f an element can be rated between zero and one hundred. With zero being an

element without defects, and one hundred an element so severely affected by defects that it

is structurally and aesthetically redundant. If a single defect was diagnosed as being caused

270
 w in vnv v a ^ v i i j j j i w u 1U 1 IV IIU U IW 'L 'U U1 i U g C 1 C p d l l

by chloride ingress, and that defect covered 2 % o f the element area, how much would this

degrade the element condition on the scale o f zero to one hundred?” The experts were now

clear that every defect was to be given an ‘effective size’, balancing the actual size o f the

defect so it is represented by an average defect o f a size judged to have a comparable effect

on the overall condition o f the element as the defect under consideration.

The experts arrived at an answer, and the question was repeated for different defect sizes,

with the intention o f graphically representing the experts’ decisions. The relationship

formed could be modelled by a simple equation. Thus, the need for literally hundreds o f

rules (e.g. if the defect is 50% of the element then the effect on the overall element is a

condition rating of 30% etc.) is replaced by an equation.

A graph was produced of the expert answers and is shown in Figure 5.38.

In Figure 5.38, the blue line represents the expert opinions on how the size o f a chloride

cracking defect affects the condition o f an element. The magenta line represents the

mathematical model o f the experts’ opinions. This colouring system is adopted for all

similar graphs.

The equation which models the magenta line is

Effect on element condition = Effective size / (0.0465 + 0.0095*Effective size)

The figures 0.0465 and 0.0095 are constants for chloride induced cracking and, like many

variables developed for the program, could be calibrated to enhance their performance

during field trials of the system.

271
 Chapter 5 - Decision making in the expert system for reinforced concrete bridge repair

% u o p jp u o o ju 9 iu 9 |a u o t o a y g
272
 ,— _ ^ m mw vajjwi oj’oi^m iui icuxiuiccu uunuicie uriuge repair

The experts’ explanation o f the form o f the graph is that the effect o f chloride cracking

defects on the element condition increases approximately proportionally with the increase

in defect size until a point is reached where the element is seriously affected by chlorides,

beyond this point an increase in the amount o f chloride defect covering the element does

not affect the element condition at the same rate, as the defects already present were very

serious and more o f the same does not radically change the outlook for the element.

Therefore, for the defect in the previous example.

Size of defect = 8.5%

Severity zone decision = Cosmetic

Knowledge base decision = Chloride corrosion

Effective size o f defect (equivalent size o f average defect) = 2.832%

Effect on element = 2.832 / (0.0465 + 0.0095*2.832)

Effect on element = 38.58%

Therefore, at this stage, the following actions have taken place:

o User adds pattern cracking patch, system determines size

o User selects representative image and adds additional information

o System judges severity as ‘Cosmetic’

o Knowledge base judges defect as ‘Chloride corrosion’

o System calculates effective size o f defect (size o f equivalent average

defect)

o Chloride corrosion graph (Figure 5.38) used to determine effect o f

defect on element condition

Therefore, the condition o f the element is reduced from 0 to 38.58%.

273
 1 * — ill mvWA|;wI JkVlIlIU1 1 V
1U1U
1 U
V^
U tUlltlClCU
1 lU^C1 C
JJ
C l
ll

-
It is important to note that the graph (and the other graphs which will be developed shortly)

works cumulatively. If another chloride induced cracking defect is encountered, its

effective size should be added to the effective size o f previous chloride cracking defects in

order to find the total effect o f all chloride cracking defects on the element condition. This

total effective size should be used to judge the effect these defects have on the element

condition as whole.

For example, consider the previous defect and an additional defect:

Defect 1 - effective size = 2.832%

Defect 2 - effective size = 5 %

Cause o f both defects = Chloride cracking

Total effective size of chloride cracking defects = 7.832%

Effect on element severity = 7.832 / (0.0465+0.0095 * 7.832) = 64.77%

For purposes explained later in this chapter, the influence of each individual defect upon

the element severity must be distributed between the two defects based on their effective

sizes.

Influence o f defect 1 on element condition = (2.832/7.832) * 64.77 = 23.42%

Influence o f defect 2 on element condition = (5 / 7.832) *64.77 = 41.35%

Therefore, it can be said that, of the element condition o f 64.77%, 23.43% was caused by

defect 1 and 41.35% by defect 2.

So far in this section, it has been explained how the effect o f chloride pattern cracking on

element condition has been assessed. Using the same technique, graphs to assess the effect

of defects on element condition for the other forms o f pattern cracking were determined

(Figures 5.39 to 5.44).

274
 Chapter 5 - Decision making in the expert system for reinforced concrete bridge repair

% J U 0 1 U 0 |0 U O 1 3 0 J J 3
275
for reinforced concrete bridge repair
Chapter 5 - Decision making in the expert system

r-
o
*
<
Nu
'w
<D
£
O
W

276
oO
s
"<Su .2
a -5
eo
w
c e
o o»
o £<u
«4h ’tu
w e
o
W
s
>
12
o
w
u

0
•w
U
Im
«
JS
0)
ocn
*s
1u
o

o
■*r
o o o
CO
o o o id
oL
>.
% }U9iua|e uo ja a ^ g 3
bX)
for reinforced concrete bridge repair
Chapter 5 - Decision making in the expert system

o
00

<N
VO
o
*
0)

OX
•
>
O
IT0J) 3
_o

277
*5 W -3
e
8 c<u
o
(0
0
a
Jh
1
m a
o eo
4->
eg
o
a
6JD

« 12
W ax
«La
ax
ax
&X)

"a

ax
ax
s:
•«-<
WD
e
E
ax
Q
o rr
o o o >ri
ax
% ) U 3 1 U 3 |3 U O J 0 3 J J 3
La
3
DJD
for reinforced concrete bridge repair
Chapter 5 - Decision making in the expert system

O
*
4)
N

<L>
#>
O
£
W

278
4)
au
13
cl
o
o #co
& •3c
W o
co
E
_<U
"3
s
o
C
bX>
N
«J -
u
C
m
o
-M
u
at
*M

a»
J=
-
m

bX)

C
E
at
Q
«s
Tt
in o in o in o o o
% } U 3 U 1 3 ]3 U O } 3 3 J J 3 3bX)
 Chapter 5 - Decision making in the expert system for reinforced concrete bridge repair

% J U 0 1 U 3 |9 U O J O 0 J J 3
279
 Chapter 5 - Decision making in the expert system for reinforced concrete bridge repair

8
8
8
% iu a iu a |9 u o jo a ^ g
8

280
 '© igv i v p u

5.7.2 Effect of spalling defects on element severity

The position of the severity zone apexes for spall defects is governed by the technique

outlined in section 5.5.3 - apexes are positioned based on spall depth. However, the

position of the apexes can change depending on additional information entered by the user.

The expert system remembers the positions of all the zone apexes after the initial data

input phase (spall size and depth only) as this represents the position of the zones for an

‘average’ defect. As with the technique for pattern cracking, the distance a defect falls into

the zone it is adjudged to be in, after all the information has been entered, is converted into

a figure based on a defect falling into the same zone by the same amount when the zone

apexes were in their original positions. Creating an ‘effective size’ - the equivalent size

that an average defect would be to have the same effect on the condition of the element as

the defect in question.

Graphs were developed, using the same technique outlined in section 5.7.1, to determine

the effect on element condition of spall defects. It should be noted, however, that both spall

defects and pattern crack defects can be caused by similar ailments, for example: chloride

ingress, the progress of carbonation, and AAR. I f one element is affected by both pattern

cracking and spalls, and the cause is the same ailment, then for the purposes of determining

the condition of the element, the spall graphs (Figure 5.45 to Figure 5.47) w ill be used. The

premise being that if some pattern cracking has already spalled on the element, the existing

pattern cracking is likely to spall soon, and should therefore be treated as spalling, which

the system considers is more severe than pattern cracking.

281
 Chapter 5 - Decision making in the expert system for reinforced concrete bridge repair

% ;u0uja|a uo 100113
282
for reinforced concrete bridge repair
Chapter 5 - Decision making in the expert system

<D
%
o—
>

w
*
so
O'
o
o
O
+
o

con
o
<N

■a
c
o

c
<l >

£
o
<U W D
C c
o
O Q.
£<E
W T<3u

♦ ■

at
v
j=
■*
»>
W
D
_C
’5
S
s_
o>

SO
o o o p o o o o o o o
0
u1
% u o j ijp u o o j u e iu e i e u o j o a y v 3
M
 Chapter 5 - Decision making in the expert system for reinforced concrete bridge repair

% ) U 3 U I 3 |3 U O J 0 3 J J 3
284
 t it u n u ^ t it p a u

For example, take two defects:

Defect Knowledge base decision Effective size

Map cracking Chloride 11%

Spalling Chloride 5%

Because at least one spall is prevalent, Figure 5.46 is used to determine the overall effect

on the element.

Effect on element = Effective size / (0.031 + 0.0096*Effective size)

Effect on element = 16 / (0.031 + 0.0096*16)

= 86.7%

In effect, 16% of the entire element is covered by what the experts judged to be an

‘average’ defect. In reality, 40% of the element could be covered by less than average

defects, or 5% by very severe defects. The end result takes all these factors into account

and the user is aware that this defect is very severe.

5.7.3 Effect of structural cracking on element severity

A ‘structural crack’ entered by the user is assessed by a knowledge base to determine if the

crack is indeed a structural crack, or a crack caused by corrosion. Cracks caused by

corrosion w ill be treated as pattern cracking.

Genuine structural cracks, in the same way as spalls and pattern cracking, have their

severity zone apexes set to an initial position based on a minimum of factors - crack size

and crack width. The apex positions at that stage represent the position for an ‘average

structural crack’. The effective length of a structural crack is determined using the same

285
 -
-
--
--
--
--
J ^_-
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
-**
*nwv/vpvnj; jivill 1
V1I
W lliUlWVUl^
U HUltit UliUgt ItpCUl

processes used to determine effective size for spalling and pattern cracking as outlined in

the previous sections.

Currently, this relationship is used to determine the effect of structural cracking on an

element:

Effect on element = (a * effective length + b)

Where a = 1 and b = 0

This particular relationship w ill be revised and updated after field trials.

5.7.4 Effect of miscellaneous defects on element condition

Miscellaneous defects are sand streaking, blow holes and honeycombing. They are

identified by the system user and not by the expert system. In order to determine the effect

they have on the severity of an element, their actual inputted size can be used directly with

the graphs and equations determined by the expert panel in Figure 5.48 and Figure 5.49.

286
Chapter 5 - Decision making in the expert system for reinforced concrete bridge repair

o
*
4)
N c
.2
"■ *-

"S
C5 e
oo

287
I
v S ca>
o 2
Ji
s a o "3
co c
o> o o
$ U
e
D
1c2s
5
i <u
S-

T3
£
VI
-a
c
«

£
jo
3
o
CM

aj
a»
.s
o
a
x
"5
E
0)
Q
oo
rr
lO
CM
in o m o •n
au>
%ju0LU0|a uo jo0jJ3 s
OX
 Chapter 5 - Decision making in the expert system for reinforced concrete bridge repair

% juauieia uo loajjg
288
 v /v jy v i i J j - J i v i l i 1U1 I C U U U l ^ U V^VJilV^JL C l t U1 lU ^ C 1CJJ&11

5.7.5 The Element Graph

In addition to rating the severity o f individual defects, it is an im portant requirem ent o f any

intelligent expert system in this field to be able to assess the overall condition o f an

element affected by multiple defects. Figure 5.50 shows how, when an elem ent is selected,

an element condition indicator is shown at the bottom o f the screen.

ft USER-3WT251LF57\BMX - bridges - Demo-Bridges - Com ponent F ate ■Add Defects

File Edit View Defects Structure Stock Abnormal Loads Reports Customise Help
« 4} ~ xss# u o | (8>a
Inspection: •»j Grid Options: + + 4-
t* 41
Bridge View Pier/colum n 1 i
£3 Pier/colum n

s
I (^^22223230
Main B eam s
mmt Main Beam s 1
b Bank se a t
*|* B ank se a t 1
*|* B ank se a t 2

& D
-O'

Diagnosis:

N> <
Com ponent selected - Click a fac e to display delects.

1 BESSSIgl!
mmmmm
ft.

J % B ack to Structure
Ready...

Figure 5.50 Element condition indicator

The marker on the Element Scale is a band, and the width o f the band represents the

confidence the expert system has in its prediction o f the element condition. Hence, the

width o f the band represents ‘element uncertainty’. The width o f the band changes as m ore

defects are added; the change is dependent on the severity o f the defect being added, its

uncertainty and the current element uncertainty.

289
 -
-
--
--
^ -
-
--
--
_ umiuiig 1
11mwVA
J1V
1IOJ
'OI
W I
I1U
1ItllliUl^GUW
U1
1C1
CICUllUgCI
C J
JClll

Element uncertainty (the width of the element band) is calculated with the following

formula:

Eun = [hi * Dui] + [h2 * DU2] + [hx * Dux] ......

Where Eun = Element uncertainty

Dux = Defect uncertainty (see section 5.6)

hx = R /(E a)

Where Ea = Element condition after addition of defect

R = Effect of current defect on element condition*

*Say three chloride caused spalls of identical severity and size are input into the expert

system. The first spall may cause the element condition to increase to 20. However, the

second spall would probably only increase the element condition to 30, and the third spall

to 36. This is in accordance with the technique developed in section 5.5. However, each

spall individually is responsible for one third of the final element condition of 36.

Therefore, as the third spall is submitted, the variable R would be 12 although the actual

increase in element condition could be much smaller.

For example, say a user adds a spall to a pier. The spall sets the element condition to 22

due to its effective size of 7%, and has an uncertainty of 30%.

h = 22 / (22)

100%

Eu„ = [1 * 30]

30%

Say a second defect is added, with an effective size of 4% and an uncertainty of 40%.

Assuming the causes of both defects are the same. The effective size of both defects

290
 r 0 . . . . . . „ u ^ i v u . iw i. w . u u i ^ u W U U IL I^ u n v jg v , iv ^ p a u

combined is therefore 11% and this sets the element condition to, perhaps, 30. By pro-rata

of effective sizes, the first defect is now responsible for:

7/(744) * 30 = 19

and the second defect for

4/(7+4) * 30 = 11

Reworking the first defect:

h = 19/30

63.3%

Eun = [0.633 * 30]

18.99%

and including the second defect

h = 11/30

36.7%

Eun = [0.367 * 40] + [ (1 - 0.367) * 18.99]

14.68%

Total element uncertainty = 18.99 + 14.98 = 33.67%

Each time a new defect is added to the system and diagnosed, a new series of calculations

for each defect is automatically conducted.

Currently, the width of the element marker is equivalent to one tenth of the element

uncertainty, and is shown on a scale from 0 to 100. For example, if the element uncertainty

is 33.67%, then the width of the element condition marker is 3.36, and the centre of the

marker indicates the current element condition.

291
 lfc> jjr j i v m i u i iv/iinuiC'Wu u n u g t; i t p a n

5.8 Testing

Referring to the expert system framework in section 5.4.3: at the testing stage the cause of

each defect has been diagnosed, and the severity of the defect has been assessed. In

addition, the element itself has been given a condition rating based on the defects affecting

it. Therefore, at this stage the testing knowledge base is ready to work. This knowledge

w ill recommend the types of testing which should be conducted to confirm the findings of

the expert system so far. Firstly, the information gathered by the program needs to be

collated in the form which the knowledge base requires it. Structural cracking does not

require chemical testing, the only testing the structural crack knowledge bases may

recommend is monitoring the crack. This usually involves measuring if the crack is

opening, closing, or static.

As a result of the output of the knowledge base to diagnose defects, some defects can be

diagnosed as having multiple causes. For example, a spall in an old bridge with low cover

in the central reservation of a motorway with a considerable degree of corrosion w ill be

reported (by the knowledge base) as having a high probability of the cause being

carbonation and a high probability that the cause is chloride ingress.

An operation called ‘probable cause distribution’ is undertaken. For example, say

DEFECT 1 is diagnosed as ‘high chloride’ (a high chance it was caused by the ingress of

chlorides) and ‘medium carbonation’.

Each natural language assessment by the knowledge base has a value. High = 1, Medium =

0.75, LOW = 0.40.

292
 -
--
--
--
--
--
1--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
- *
** M
twvA|;wnaj^iviii 1
U1IVllUUlVtUW
U11
U C
ICUHUg
C 1
G p
cUl

Therefore, for DEFECT 1:

Chloride = 1

Carbonation = 0.75

The likely contribution of each cause is converted into a percentage

Likelihood that chloride caused defect = A /(C)

Where A = Numeric value of natural language qualifier appended to cause

C = total of all natural language qualifiers for all possible causes of defect

For DEFECT 1

Chloride = 1 / (1 + 0.75) = 57.1%

Carbonation = 0.75 /( 1 + 0.75) = 42.9%

This procedure is repeated for each pattern cracking or spall defect affecting an element.

For example, assume an element (Column 3) is affected by 6 defects, a combination of

spalls and pattern cracking. The process shown above is conducted for each defect, and a

table is constructed as shown in Table 5.17.

Table 5.17 Example defects on 'Column 3'

Defect Carbonation Chlorides AAR Early AAR

1 71 29

2 100

3 100

4 71 29

5 100

6 100

293
 r ™ ^ ..1 U 1 V 1 1 1 6 ill 111V u a j i v i I o _ y o i v m H J1 IV IIIIU IV ^ C U U U 1 1 U 1 C IC U1 lU g C lC p c tll

Say the element condition rating due to all these defects, following the procedures adopted

in section 5.7, is 30%.

The contribution of each individual defect to that 30% condition rating is determined by

the expert system in accordance with the method outlined in section 5.7.1. This process is

shown, for the given example, in Table 5.18.

Let EC = element condition

Table 5.18 Contribution of each defect to element condition

Defect Contribution to EC Contribution to EC %

1 10 = 10/30 = 33.3

2 5 16.6

3 2 6.7

4 2 6.7

5 6 20

6 5 16.6

Total 30 100

Now, considering DEFECT 1, it has contributed 33% to the element condition. In

accordance with Table 5.17 the defect has a 71% probability of being caused by

carbonation and a 29% probability of being caused by chloride ingress.

By multiplying these values by the 33% contribution of defect 1 to the element condition,

the figures the knowledge base needs are determined - the contribution of each cause to the

element condition. These are shown, for the given example, in Table 5.19.

294
 ^ w l i i w i v n i f c , i l l t n w V A j j w i i O ^ O I V I I I 1U 1 I V J l l l U l t t U V ^U IIV ^IC IC U1 l u g e IC JJC 1I1

Table 5.19 D eterm ining prim ary causes o f elem ent deterioration

Defect Carbonation Chlorides AAR Early AAR

1 71% x 33% = 23.4% 29% x 33% = 9.6%

2 100% x 16.6%= 16.6%

3 6.7%

4 71% x 6.7% = 4.8% 1.9%

5 20%

6 16.6%

Total 23.4% 68% 6.7% 1.9%

A full version of Table 5.19 would contain columns for all the possible defect causes. It

can be seen that the primary ailment affecting the element is chloride corrosion,

carbonation is also a factor. It is likely that the small defects suspected of being caused by

AAR are also caused by chlorides, although the expert system w ill make an assessment on

this.

At this stage the testing Knowledge Base can be provided with the information it needs

i.e.:

Element

Total area affected by carbonation as %

Total area affected by chlorides as %

Total area affected by AAR (including early aar) as %

Early AAR as %

295
 ^WVU J iu^v iwjjau

Plastic shrinkage as %

Impact as %

Freeze thaw as %

Drying shrinkage as %

Crazing as %

Internally, the expert system uses databases to store the data generated by the user input

and the output of diagnosis knowledge bases. Tables such as those shown in Table 5.20

and Table 5.21.

Table 5.20 Typical storing of knowledge base diagnosis for pattern cracking

Defect aar carbonation chloride crazing Drying Early Freeze Plastic

shrinkage aar thaw shrinkage
1 high medium

2 high

3 medium medium

Table 5.21 Typical storing of knowledge base diagnosis for spalling

Defect Aar Carbonation Chloride Impact Prev popout Spacing Filled
Rep block Pocket
1 high

2 high medium

When the figures in Table 5.19, and the additional information, can be provided to the

testing knowledge base - it uses the principles of objects, ranges and rules to examine the

evidence and deliver a conclusion. Importantly, the knowledge base has rules to recognise

296
the situation where there may be two distinct and separate ailments affecting an element -

for instance AAR and chloride corrosion.

5.9 Repair advice

Referring to the expert system framework in section 5.4.3, the expert system and the

system user now have the information required to recommend repair advice. When asked

for repair advice for a particular defect, the user is reminded of the probable causes the

diagnosis knowledge base recommended. I f more than one cause was recommended, the

user is invited to select the cause of the defect in accordance with the results of the testing.

In the absence of testing the user is recommended to select the defect with the highest

natural language operator, i.e. high probability of chloride caused to be selected before

medium possibility of carbonation. However, the repair knowledge base w ill accept a

cause of both carbonation and chloride ingress at this stage.

5.9.1 Advice for spalls and pattern cracking

Once the user confirms the cause of the defect, an object in the repair knowledge base

called ‘defect’ is set. The range of this object is:

Chloridecarbonation

Carbonation

Chloride

Crazing

EarlyAAR

ActiveAAR

InactiveAAR
 kF v u cic u n u g t it p a n

Blowhole

Drying Shrinkage

Freezethaw

Honeycoming

Sandstreaking

Plastic shrinkage

Other

Therefore, the ‘defect’ object has to be set to one of these variables. The laboratory report

following AAR testing w ill identify the status of that particular defect. The defect ‘Other’

encompasses small defects such as filled pockets which have spalled away, tie wires, and

popouts.

The repair knowledge base has two input objects, i.e. two objects whose values are set by

the expert system. These are ‘defect’ and ‘severity’, with ‘defect’ being the cause as

discussed above, and ‘severity’ being the severity zone into which the defect fell.

Rules have been constructed, by the experts, which relate combinations of causes and

severities to twenty-nine separate pieces of repair advice.

The example in Figure 5.51 shows a premise rule in the repair knowledge base that w ill set

the output object ‘drying shrinkage 2’ to yes. Only one of the twenty-nine output objects,

for each defect, can be set to yes - there can be no conflict. When the expert system detects

that the output object set to ‘yes’ was drying shrinkage 2, it w ill search for the piece of

advice which corresponds to that defect.

The example shows that if the defect is drying shrinkage, and if the severity of the defect is

Minor or Cosmetic, the object w ill be set to yes. Table 5.22 shows all the twenty-nine

pieces of repair advice. The expert system, in the case of this example, would return the

text under ‘drying shrinkage 2’.

298
 t'v" 1 ■JJ 1V-[_Klll

H H H H H H H SBillssSflBH H M - I * N
File Edit Editors Graphs Engine Reports Utilities Window Help

H I H H H I
Display: Values:
(Graphical d e fe c t
sev e rity

Show Objects As:

Number

| d e f e c t ! I d ry in g s h rin k a g e l | s e v e r it y ] | m in o r-re p a ir | | 3everity~| I c o sm e tic -re p a iF ]

Figure 5.51 Drying shrinkage repair rule

Table 5.22 Repair advice

Knowledge base output Advice

Chloride carbonation 1 This defect is caused by both carbonation and chloride
ingress.
It's not serious enough to warrant break out and repair.
Repair with cem entituous mortar.
Chloride carbonation 2 This defect is caused by both carbonation and chloride
ingress.
Break out and repair.
Crazing 1 This minor defect is CRAZING.
No further action is necessary.
This defect is not serious.
Crazing 2 This defect is CRAZING.
It is not essential that the defect is repaired.
If it is aesthetically unacceptable, the defect could be
filled with a cem entituous mortar.
AAR 1 This defect could be caused by ALKALI A GG REG A TE
REACTION.

No immediate action should be taken. M onitor this

defect.

299
 iug^ iv^pan

Knowledge base output Advice

Note: This type of cracking is visually similar to

FREEZE/THAW DAMAGE.
AAR cracking is deep, FREEZE/THAW cracking is
shallow. Tapping with a hammer
w ill confirm the diagnosis.
Additionally, the presence of a white gel in the cracks, or
evidence of a white gel (which may have been washed
away), would confirm AAR as the cause.
AAR 2 This defect is inactive ALKALI AGGREGATE
REACTION.
The reaction between cement and aggregate appears to
have discontinued.
Undertake a structural survey of all elements effected.
I f elements are structurally acceptable: leave or coat for
aesthetic reasons.
AAR 3 This defect has been confirmed as active ALKALI
AGGREGATE REACTION
Seal the surface cracks
Apply water repellent impregnations (silane monomer)
Monitor closely.
AAR4 This defect is confirmed as inactive ALKALI
AGGREGATE REACTION.
Undertake Epoxy resin injection to restore integrity.
Coat the surface for aesthetic purposes.
Chlorides 1 This defect is caused by Chloride Ingress.
It is not serious enough to warrant break out and repair.
Repair with cementituous mortar.
Chlorides 2 This defect is caused by Chloride Ingress.
Break out and repair.
Plastic shrinkage 1 This defect is PLASTIC SHRINKAGE.
No repair is necessary.
Leave.
Plastic shrinkage 2 This defect is PLASTIC SHRINKAGE.
Fill cracks with resin (possibly by making a resevoir)
and fill.
Plastic shrinkage 3 This defect is PLASTIC SHRINKAGE.
Cut out and repair.
Freeze 1 This defect is FREEZE/THAW DAMAGE.
No action is necessary at this stage.
Freeze 2 This defect is FREEZE/THAW DAMAGE.
Remove loose material, fill cracks with cementituos
mortar
Freeze 3 This defect is FREEZE/THAW DAMAGE.
Break out and repair.
Honeycombing 1 This defect is HONEYCOMBING.
No repair is necessary.

300
 L j j v i i J J 'J I V 1 1 1 iv i 1 U 1 1 I1 U 1 V t u U1 l^ p c ill

Knowledge base output Advice

Honeycombing 2 This defect is HONEYCOMBING.
Break back to sound concrete and reinstate.
Blow-holes 1 This minor defect is BLOWHOLES
Leave this defect. It is not serious.
Blow-holes 2 This defect is BLOW HOLES.
Fill these holes with a cementituous mortar.
I f necessary, coat for aesthetic reasons.
Carbonation 1 This defect is caused by Carbonation.
Repair using cementituous mortar.
Carbonation 2 This defect is caused by carbonation.
Break out effected concrete and repair.
The system w ill recommend the correct repair material
properties.
Sand-streaking 1 This defect is SAND STREAKING.
No further action is necessary.
Sand-streaking 2 This defect is SAND STREAKING.
Remove the defect and reinstate.
Drying shrinkage 1 This defect is DRYING SHRINKAGE.
It is not of sufficient severity to warrant any action.
Leave.
Drying shrinkage 2 This defect is DRYING SHRINKAGE.
Inject the cracks with epoxy resin.
Consider coating for aesthetic reasons if necessary.
Drying shrinkage 3 This defect is DRYING SHRINKAGE
Open out cracking with router, re-repair with
cementituous mortar.
Consider coating for aesthetic reasons if necessary.
Other 1 This small defect can be left, or filled with cementituous
mortar.
Other 2 Fill with cementituous mortar.

Defects whose repair advice recommends break out and repair of the substrate concrete can

use the expert system to automatically recommend repair material properties based on the

techniques developed in Chapter 4 of this thesis.

For large scale defects, and seriously debilitated concrete, an engineer w ill be required to

make an economic assessment of the relative merits of breaking out and replacing defects,

and electrochemical remediation techniques.

301
 ^ ill niw o_yon*in 1U1 ICllliUlCCU l/Ulll/lCIC UllUgC ICJJUII

5.9.2 Advice for structural cracking

As discussed in section 5.4.3.3, there are two knowledge bases for structural cracking. The

job of the first knowledge base is to ascertain if corrosion is the cause of the crack - if it is,

the pattern cracking knowledge base takes over. The other task of the first knowledge base

is to recommend action.

A typical piece of action recommended is monitoring the crack. Through monitoring the

user checks to see if the crack is:

■ Active widening

■ Active opening and closing

■ Dormant

■ Closing

Certain rules w ill cause the recommendation of the first knowledge base to be ‘Leave - no

action required’. However, the recommendation can be to monitor the crack and check for

corrosion. In this case, the system has been unable to decide if the crack is caused by

corrosion or structural effects.

For minor cracks, the first knowledge base may recommend repair by rout and seal with no

need for further tests or monitoring.

After the monitoring stage, a second structural cracking knowledge base is utilised. The

second structural cracking knowledge base requires certain pieces of information - some

supplied by the engineer and some by the expert system:

• Width of the crack

302
 j-v—_ ^ ..
.u.wwg, in mwVA
J1V
1l ■
JJ'Ol.
v l
ll 1
U11L.1111U1LCU UU11WXCIC U1 lUgC ICJJtUl

• Results of the monitoring

• Moisture condition of crack

• Is strengthening required

Importantly, it is the responsibility of the engineer to find and eliminate the cause of the

structural cracking. However, if a crack is actively widening, or actively opening and

closing, the knowledge base may recommend redesigning an expansion joint at the crack

location.

There are eleven separate pieces of repair advice that can be generated by the second

knowledge base for structural cracking. A ll viable repair options for a particular crack w ill

be presented to the user. These pieces of repair advice are shown below.

External stressing

Consider external stressing for the repair of this crack.

It is recommended for moving and opening fine cracks.

Most useful on long members: beam, deck, parapet.

Stitching (dogsj

For use to re-establish tensile strength across cracks. D rill holes either side of crack and

resin fix ‘staples’ made of reinforcing steel.

Rout and seal

Consider for static and moving 1mm cracks. Enlarge the crack, fill and seal with suitable

joint sealant.

303
 — ...u .w .it, ill niw uAp^l i 1VJ1 l ^ l l l i u i t t u ^Ulll^lClt U1 lUgG IC^jail

Redesign and provide expansion joint

Active cracks where strengthening may be required.

Active cracks can be routed out and filled with flexible sealant. Narrow cracks may be

sealed with a flexible face seal.

Grouting

For dormant cracks up to several millimetres in width.

Extensible overlay

Use for moving cracks on flat horizontal surfaces.

Bonding

Bonding with Epoxy (crack injection) / cement mortar / microfine cements / resin

For static, Fine cracks (sub 1mm). Cracks as narrow as 0.05mm can be bonded using

epoxy injection. Only apply to static cracks (or remove the cause of crack

movement/growth)

Blanketing

Active or dormant cracks not requiring strengthening.

Autogeneous healing.

This natural crack repair process can occur in the presence of moisture and in the absence

of any tensile stresses.

It could be practically applied for example, to close a dormant, thin crack, in a situation

where moisture was present.

304
 .Iiu iv .n t, m u iv I / A p w i OJ’OIUIII 1U1 I t l l l l U l ^ C U u u n v ;ic ic u x x u g c xcpcxxx

However, if the amount of water passing through the crack is large, this w ill wash away the

lime deposits which would otherwise heal the crack.

Ideally the crack w ill heal in the presence of stationary moisture, either from natural

sources or contrived.

Ordinary overlay

Used to treat static cracks on flat horizontal surfaces.

Often using a heavy coat of epoxy resin or an overlay of polymer modified cement.

305
 Chapter 6 - R eview o f the expert system for reinforced concrete bridge repair

6 Review of the expert system for reinforced concrete

bridge repair

Chapter Objective

o To present and review the software that incorporates the techniques

and routines developed in this thesis

6.1 Introduction

During development of the research, it became necessary for the software to be able to

understand the dimensions of certain bridge elements in order to make decisions. It

became apparent that the software being developed could be integrated into a ‘Bridge

Management System’ which would not only function as an expert system for concrete

repair, but also as a software inventory for storing bridge stock information. The software

engineering company assisting in this research have taken prototype software developed by

the author for the expert system and material property specification systems, and re-coded

this software to produce software cosmetically acceptable in a commercial market and

using more sophisticated database and software language techniques than those available

to the author during prototyping. This chapter generally shows screen-grabs from the

developed commercial software (www.bridgemanagementexpert.com).

306
 Chapter 6 - Review o f the expert system for reinforced concrete bridge repair

Although a bridge management system is a commercially viable commodity, an expert

system for concrete repair would be an untested commercial product. The collaborating

software developers were, therefore, keen to maximise the saleability of the end product of

this research through the provision of a bridge management system in addition to an expert

system and repair material property specification program. Therefore, the ability to store

detailed structure information has been added to the overall software by the collaborating

software organisation. This database system works seamlessly with the expert system

capabilities of the software. Fortuitously, many features developed as a result of this

research, such as the need for the three dimensional representation of concrete elements,

work in harmony with the bridge management database developed by the collaborating

software engineers.

6.2 Structures Management

The bridge management system is a database in which details of an organisation’s bridge

stock can be recorded and managed. The bridge management system into which the expert

system for reinforced concrete repair is embedded can store and manage data for thousands

of structures. At its simplest level the program can store the name of a structure, its

location, the features crossed by the bridge and other such important but basic information.

Exploiting the full functionality of the bridge management system w ill allow the user to

store the shape, sizes, materials and condition of all the elements of the bridge as well as

photographs, reports and very detailed data. Figure 6.1 and Figure 6.2 demonstrate the

bridge management system.

307
 Chapter 6 - Review o f the expert system for reinforced concrete bridge repair

xJ
File Edit View Structure Go Stock Abnormal Loads Reports Customise Help

£ O s IQ n n □ r,1 © n ix

All Structures

T 7 BR1003
t ? B R 1004
t J BR1005
T 7 EI5IiI3
h P ie r/c o lu m n
2i D eck
A b u tm e n t
2i P a r a p e ts
Utilities
S tru c tu re C ondition
\yS C h e c k List
C arried
V/C C ro s s e d
T j BR1007 S tru c tu re | L o c a tio n j D e s c rip tio n ! C o n s tru c tio n ) S u p e r s tr u c tu r e ) S u b s tr u c tu r e ] A s s e s s m e n t) R e s tr ic tio n s ) E au ip m e * I *
T j BR1008
S tru c tu re N am e: Number
t l BR1009 D em o brid g e N o. six
BR1010 S tru c tu re T ype R e fe re n c e
£jr BR1011
T rx n o i m o S tru c tu re O w ner O w n er R ef
Demo-B ridges
M aintaining A g e n t S tru c tu re S ta tu s
D em o-Retaining walls
S tru c tu re C arries A sse sse d C ap

D emo-S igns-G anteries

S tru c tu re C ro s s e s N ex t In s p e c tio n
ID D a te 09 / 06/2001
Dem o-T unnels
Min. H e a d ro o m Date Measured L a st In s p e c tio n s s e s s m e n t (6/1 / C :\B M X \th u m b s\D
D e m o -C u lv e its

Dem o-O ther y§ G e n e ra l D o c u m e n ts # In s p e c tio n s J M a in te n a n c e C o ndition

R eady...

Figure 6.1 Structure management

Figure 6.1 shows a typical database screen. In the left window is a list o f structures that

have been entered into the database. W hen highlighting a structure in the left window, its

diagrammatic representation appears in the right window, along with a digital photograph

o f the structure, and access to all the data about this bridge which the user may have

entered in the system. The bridge m anagem ent system - the database w hich stores

information such as the structure name, its location and the tim e o f the next scheduled

inspection, is the work o f the collaborating software engineers. However, inform ation

generated by the expert system - such as condition ratings o f the concrete elem ents, is

obviously also stored in the database.

308
 C hapter 6 - Review o f the expert system for reinforced concrete bridge repair

.J f J x ]
File Edit View S tr u c tu r e Go S to c k A b norm al L o a d s R e p o r ts C u sto m ise Help

■« % % r X god # # S n S <-- ► + + O h

*1
All S tiu ctu ies

8R1002
Vi BR1003
Vi BR1004
U
BR1005
BR1008
a Pier/column
| Pier/column 1
■TO*
| Pier/column 2
| Pier/column 3
Load-Bearing Substructure ■Abutments (incl. arch springing) • Abutment
| Pier/column 4
a Deck
Deck 1
m*
Deck 2 New Element J Assessment I Status ij Comments |
.W Deck 3
a Abutment Registered Condition
*|l New Element
vimxmm D e fe c t_______ Code S ev e rit E xtent W o rk Priority Est. Cost Comments

Demo-B ridges

D em o-R etaining walls

Overall Extent: Overall Severity: j-| ' Condition Index: | g Est. Cost: Recommended
Demo-S igns-G anteries

Dem o-Tunnels Overall Priority: HE Likelihood of Failure: r3 Conseq. of Failure: rs Risk Factor: r~ 3
Dem o-Culverts

Dem o-O ther * General ^ Documents ^ Inspections Maintenance -3 Condition

R eady.

Figure 6.2 Alternative views of structures

Figure 6.2 shows an alternative view o f the structure created by the user. The bridge

management system has a broad range o f functionality:

• Prioritising bridge maintenance

• Storing photographs

• Planning inspections

• Record keeping

• Abnormal load route planning

309
 C hapter 6 - Review o f the expert system for reinforced concrete bridge repair

6.3 Element and Structure creation

In order to allow the expert system to make judgem ents concerning the severity o f defects

affecting elements, it is necessary for the program to have information about the size and

shape o f individual bridge elements. For example, the im portance o f a spall o f lm is

dependent on the overall size o f the element it affects. In a slender leaf pier such a spall

could be extremely important, whereas on a wide and tall abutment its seriousness would

be far less. These kind o f decisions can only be made by an expert system when it has

inform ation which will allow it to compare the relative sizes o f the element and defect.

Figure 6.3 shows the ‘structure creation w izard’ that allows the user to quickly insert a

typical reinforced concrete structure into the database.

A S tru c tu re W izard

Single/Multi Span Structure Details

No of Spans: j p -fj Number of Beams H i; Column Heads |N one

Overall Span [~20000 mm Type;

3 With Bearings

Width j 8Q00 mm

Bridge Details

Name: 'Structure 58

Reference: [g p

Number: j 58

Grid Reference: [

Construction
Date:

0
For element details select the element from above

Cancel General Info I 3D View j Finish

Figure 6.3 'Structure creation wizard' single span bridge

310
 Chapter 6 - Review o f the expert system for reinforced concrete bridge repair

Figure 6.3 shows how a structure can be created quickly and easily. The structure type is

selected first, then the num ber o f spans, the span lengths and the structure width. The

finished structure can be easily modified further.

Figure 6.4 shows the ‘Structure creation w izard’ again, this time showing the creation o f a

three span bridge shown in 3D.

M S tru c tu re W izard ■=JPl X.

Single/Multi Span Structure Details

No of Spans: j J -f-j Number of Beams

p "±j Column Heads (Rectangular Cross Heac_j]
Overall Span 20000 mm Type: Rectangular With Bearings i~

Width 8000 mm
| S in g le d ulti span

Bridge Details

Name: (structure 58
Reference: |s g

Number: [50
Grid Reference: j

Construction I
Date:

0
0
0
For element details select the element from above 0
Cancel j General Info j 2D View 1 Finish

Figure 6.4 'Structure creation wizard' three span bridge in 3D

Once such a structure is inserted its geom etry can be quickly and easily am ended to m atch

that o f the structure being modelled. It is im portant to note that the routines developed in

this thesis are not sensitive to slight differences between the actual and modelled geom etry

o f the structures being assessed. It is im portant for the expert system to have only a

reasonable indication o f the relative sizes o f elements and defects, and as such careful

precision is not necessary when element sizes are being entered into the program.

311
 C hapter 6 - Review o f the expert system for reinforced concrete bridge repair

An alternative method for entering the layout o f a reinforced concrete structure into the

system is to place elements individually as shown in Chapter 5 Figure 5.1. This m ethod is

suited to more unusual structural forms for w hich the structure creation wizard makes no

provision.

6.3.1 Example of structure creation

A structure with a rhomboidal articulation arrangem ent over its piers could be considered

an unusual concrete highway structure. Such a bridge is shown in Figure 6.5.

Figure 6.5 Unusual motorway bridge

Figure 6.6 shows how the user would begin to insert such a structure as shown in Figure

6.5 into the program. Firstly the user would use the standard menu and click ‘insert new

structure’ from the ‘File’ menu. This action will automatically show the ‘Structure creation

w izard’. In the case o f the complicated structure shown in Figure 6.5, the structure creation

wizard will have no suitable template to model the structure. The user closes the structure

creation wizard window and is presented with a blank diagram window, as shown in

Figure 6.6. The user’s next action is to click ‘insert elem ent’ from the ‘structure’ menu.

312
 Chapter 6 - Review o f the expert system for reinforced concrete bridge repair

File Edit View S tru c tu re Go Stock Abnormal Loads R e p o rts C ustom ise Help

<5» ® @0 0 0 0 & Cl rSj III

HSEES9MMHRSI
All Structures

Demo-B ridges

f t 57

^S» Utilities
S f C h e c k List
Carried
p - C rossed
d ? BR1001
f t BR1002 Structure Condition
■dr B R 1 0 0 3
d f B R 1004
f t B R 1005
S tiucture j Location j D escription | Construction S u p e rstru c tu re ] Substructure I A sse ssm e n t] R e strictions] Equipme < ] ►
d r BR100G
d r B R 1007 S tructure Nam e: [Structure 58 Number: 58
f t B R 1008
d T B R 1009
S tructure Type: j
d R efe ren ce: fsg

d T B R 1010
f t BR1011
Structure Owner: f
d Ow ner Ref: j

r i r BRimo
Demo-Retaining walls
M aintaining Agent: j
d S tructure Status: [
d
Structure Carries: f ~
d A s s e s s e d Cap. ]~
D emo-S igns-G anteries
Structure D o s s e s : J d Next Inspection: ]
ID D ate: ]
Demo-Tunnels
Min. Headroom : j D a te M easured: j L ast Inspection: j
Demo-Culverts

General Documents % Inspections £ Maintenance ^ Condition

R e a d y ...

Figure 6.6 Blank diagram window

From here the user is presented with a menu giving a large variety o f bridge elements, a

beam is selected and is drawn onto the screen, generally at the required dimension. This

operation is shown in Figure 6.7 and Figure 6.8.

313
 Chapter 6 - Review o f the expert system for reinforced concrete bridge repair

E lem ent W izard - ja i >

^-Btrafilg ® P 0 10 i
Add f’ 1
Deck
D e c k Elen
j Primarji ^ ftrcfl ml
—taaa^i | $ R e c ta n g u la r
mm T iu
,1 P a r a p e t Q C u r v e d O n P la n
mm culi
mm Arcl ** Truss t~ \ C u rv e d In E le v a tio n

m Arcl N B racing 2 I S e c tio n

m V ou 0 C ro ss H e a d *1] E q u a l A ngle
mm Arcl
^ E nc Q A b u tm e n t ^ j] U n e q u a l A ngle
10 12 T4 T6 18 20 22 24 26 28 30 * 5 ■-X
2
■ l Sub ^ W ingwall 3 C hannel

i Box Q
oar Arm _
P ier/C olum n Mi
mm P oll f f l Bearing
mm P re Bi E x p a n sio n Jo in t
— Sle. Q F o u n d a tio n s

** I Ur © C u lv e rt
mm Dec w
j j Transv H e a d Wall

-u S econt j*jjj G e n e ra l M em ber

1 Em bankm ent
Cj Tie Be. *■
Ci P arape ® > A p ro a c h R a m p

Ci D eck B p Sign / G a n try

Load-Bear
h Durability I L—i No G e o m e try
a ci S afety Elements
a ci O ther Bridge Elem ents
a ci Ancillary Elem ents

Structure Wizard ! Can

Figure 6.7 Inserting a beam

-.fljxj
$ F3 ff 1fe ►♦ 4 - © <!» p 0 0 Q,
Add New Element

- -i D e c k Elem ents
-j Primary D e c k Elem ent
1
mm M ain Beams
mm Truss members
■ * culvert
mm Arch
mm Arch Ring
mm V ouso irs/A rch Fact
mm A rch B a rrel/S offit
mm E ncased Beams
■■ Subway
■mm Box beam interiors
mm A rm co/C oncrete pi|
mm P o tta l/T unnel port.
mm Prestressing
S leeper bridge
ma T unnel Linings
■ * Deck Element Location: Specific Dimensions: Element Information:
Ci Transverse Beams
Origin X 0 End>< j 25000 Width 1 j 5000 Parpet Sub. H. 800 Reference: [Main E
_j Secondary D e c k Elerm
Origin Y "0 EndY j “ Width 2 I 7000 Parpet Thick.
j----
Ci H alf Joints 1300 ID Code:
Ci T ie B e am /R o d
Origin 2 j 8000 EndZ | 10000 TW 800 Parpet Height 400 Material: j Reinfc
ffl Ci P arapet Beam or Canti
Ci D e c k Bracing
Skew I Rotation Options
S 200 Heigh* 2000 Span: j Span
Load-Bearing Substructun
Date Built: 1
Durability Elem ents Curve in Plan: A: I FT B
S afety Elem ents _
Curve in Elevation. A: j 0 B: f
O ther Bridge Elem ents
A n n illa ru E le m e n ts
i >r Structure Wizard Cancel
Existing Structure Elements

Figure 6.8 Beam once inserted

314
 C hapter 6 - Review o f the expert system for reinforced concrete bridge repair

Further beams are added using the same technique. Figure 6.9 shows the insertion o f the

third and final beam. The span lengths, and the elevations o f the start and end o f the beam

can easily be amended.

■m uni H H H B f l H i -I f llx l
M o m i 1 <- # t ® <!► P £5 0 &
Add New Element

Deck. Elem ents

-i Primary D e c k Elem ent
mm M ain Beams
mm T russ members
mm culvert
mm Arch
tm Arch Ring
mm V ouso its/A rch Fact
mm Arch B arrel/S o ffit
mm E ncased Beams
mm Subway 2J 2 4 6 810121416182C222426283032343638404244464850525456586062S46E6870727476788082848GSS50X'
aw Box beam interiors
i A rm c o/C onc re te pi)
l P o tta l/T unnel port<
l Prestressing
aw S leeper bridge
mt Tunnel Linings
aw D e ck Element Location: Specific Dimensions: Element Information:
Ci Transverse Beams
OriginX j G5000 EndX f 90000 Width 1 “ 5000 Parpet Sub. H. f 800 Reference: jMain E
-j Secondary D e c k Elemi
Q H alf Joints Origin Y j EndY [ Width 2 | 7000 Parpet Thick. | 1300 ID Code: j
Ct T ie B e am /R o d
Ci P arapet Beam or Canti
Origin 2 j 10000 EndZ f 8000 TW f~ 800 Parpet Height | 400 Material: | Reinfc
Ci D e c k Bracing s p 200 Height j 2000 Span: jSpan
Skew I Rotation Options
L oad-B eating Substructun
Durability Elem ents Curve in Plan: A: f 0 B: f T
Date Built:
r
S afety Elem ents
Curve in Elevation: A:
O ther Bridge Elem ents
A nrillaru Elem ents

Structure Wizard Cancel

Existing Structure Elements

Figure 6.9 Three beams inserted

Finally the piers are inserted using the same techniques previously outlined, this is shown

in Figure 6.10.

315
 C hapter 6 - Review o f the expert system for reinforced concrete bridge repair

m ’ m n n 1 h 4 t 4 i @ &
Add New Element

mm A rch B arrel/Soffit j*
aw E n c a se d B eam s
m i Su b w ay
mm Box beam interiors
mm A r m co/C on crete pi|
mm P o tta l/T u n n e l port*

- P ,B t" <7 Deck

— S le e p X
m l Tunn, ^ * *
mm D e c k Beam
* a T ia n c v ei ^ Parapet 4 6 810121416182C2224262830323436384042444648505254565860S26466687d72747G78S0828486^53d<
EB Seconda
© C. H all J o in Truss
B a Tie B e a n Bracing
a a P arap et I @ & oss
ffl a D e c k B n
^ L oad B ea.m , □ Abutment
$ a Foundatii p Wingwall 5t Location: Specific Dimensions: Element Information:
tfe Ci Abutmeni 0 Circular Width 1 j 5ooiJ Parpet Sub. H. |~ ~ ~ 800 Reference: [m ^T e
;+: Ci Spand rel mmm\
0 Rectangular
& a P ie ,/C o le ^ ^ Width 2 | 7000 Parpet Thick. | 1300 ID Code: |
■ liWft. El Expansion Joint Yl TW f S o Parpet Height j 400 Materiat
Si Ci C ross-he. Q Foundations ^ Inverted Y Shape
© a Bearing* & ^ s r~ 2 0 0 Hei9ht | 2000 Span: | Span
'O ’ Pier Type 1
Si a B earing ,
a Durability El. Head Wall
li S a fe ty Elemi £f! General Member
l\ 7 Pier Type 2 Curve in Plan: A: | 0 S: 0 Date B ui: j

13) Pier Type 3

Curve in Elevation: A: j q S
li Other Bridge
Embankment
'i Annillaru F ie
S > AproachRamp
Structure Wizard
ExistingStructu f 3 S ign/G antry

Figure 6.10 Insertion of pier

The program contains the full functionality that is expected o f m odern w indow s based

software. For example, the first pier inserted can be cut and pasted to create the second

pier. This is shown in Figure 6.11.

-|g | xj
® c; p & & Q
Add New Element
mm Arch Barrel/Soffit *
ssa E n c a sed Beam s
■■ Subway
aw Box beam interiors
.39 Arm co/Concrete pi|
wsH Pottal/T unnel port*
mm P restressing
amt S leep er bridge
f^a T unnel Linings
mm D eck
T ransverse Beams 0 2 4 6
l i Secondary D eck Elemi
li Half Joints
l i Tie Beam /R od
l i Parapet Beam or Canti
l i D e c k Bracing
Load-Bearing Substructun
l i Foundation Element Location: Element Dimensions: Speafic Dimensions: Element Information:
h Abutments (incl. arch :
X Origin: " koto Spanc*t j 6000 hi: j 4000 «1: Reference: [Pier/ci
—i Spandrel w all/head w«
£i Pier/Column V Origin: 0 W iclhM : j 2000 h2 I 5000 ID Code: (
a » Pier/colum n
Z Origin: 2000 Herght (Z): | 8000 Material: iReinfc
l i C ross-h ead/capping bi
h Bearings Span (Span
Bearing plinth/shelf Skew / Rotation Options
Date Bidf j
Durability Elem ents
Safety Elem ents
Other Bridge Elem ents
AnrilhrQ FlfFlpjner^t
Structure Wizard
Easting Structure Elements

Figure 6.11 Copying an existing pier

316
 Chapter 6 - R eview o f the expert system for reinforced concrete bridge repair

6.4 Defects - input, assessment and diagnosis

The types of defect that the expert system can recognise and diagnose were discussed in

Chapter 5.

For the purpose of entering defects into the expert system, two general groups of defects

are considered, specifically patches and cracks. Clearly, a crack w ill be caused by either

structural or corrosion effects, whereas a defect patch w ill rarely be caused by structural

reasons. Entering either type of defect onto an element is fast and simple.

In Figure 6.12, the user has highlighted a column from the bridge view window. The

program unwraps the shape of the element, and the user enters an elliptical defect by

pressing the ‘add elliptical defect’ icon and using the common ‘click and drag’ technique

to add the defect of the required size in the required position on the element. The premise

behind this technique is to enable the user to be able to describe the general shape of the

defect - the user may add a defect which is generally square, or generally elliptical.

317
 Chapter 6 - Review o f the expert system for reinforced concrete bridge repair

E9!f!f!1IB S 1 h m m m j=j5j xj
File Edit View D e fe c ts Go S tock Abnorm al L oads R e p o rts C ustom ise Help

< £ ~ X CD § - O D ; ® E l i 2. ®
Inspection: j 1194 S c h e d u led : 0 6 /0 1 /2 0 0 5 3] Grid Options:
V I BR1006
Bridge View i I I 94 i P ier/colum n 1
2 P ier/co lum n
| liiliBBH
S ' Spall 456
| P ie r/c o lu m n 2
| P ie t/c o lu m n 3
| P ie r/co lu m n 4
lj D eck
■■/ D e c k 1
m * D eck 2
m
w t D eck 3
h A butm ent
•|* N ew Elem ent
•f* N ew Elem ent
P a ra p e ts
A A P a r a p e ts 1
A A P a r a p e ts 2

D iagnosis- j
3
T estin g A dvise <li \ P P
C om ponent fa c e s e le c te d ■d e fe c ts, te st results a n d repair p a tc h e s a re show n. Click on a d e fe c t to display its condition.

^ Back to Structure
R eady fa c e You m ay a d d d e fe c ts , t e s ts a n d re p a irs to th is fa c e

Figure 6.12 Entering an elliptical defect

The screen view shown in Figure 6.13 is presented to the user once a square or elliptical

defect has been added to an element. On this screen, the user categorises the defect as

either spall, stain, map cracking, seepage, scaling, honeycom bing, blow holes or

sandstreaking.
 Chapter 6 - Review o f the expert system for reinforced concrete bridge repair

1 A dd D e fe c t W iz a rd xj
Paintwork and Protective Systems | Vegetation | Foundation J Invert and Riverbed j Drainage j Surfacing
Expansion Joints | Embankments j Bearings ] Impact ] W aterProofing j Stone Slab ] General
Knowledge B ase D efects j s te e l J Concrete Timber j Masonary and Brickwork

% m w m * « w
Structural Stain Map S eepage Scaling Honey-com Blow Holes Sand Stalactite
Crack Cracking bing Streaking Build-up

Defect Details
Comment / Site Action:
R eference: j Work Type: Routine
Severity: jf ■y j Priority: Low “3
Extent: (T Estimated Cost:
Defect Photo

Recommendation: b o Nothing
"3
Dimentions
X Location: -2729 Width: | 2936

Y Location: | -1494 Height: j 4940 Aesthetics:

Urgent Action !“

Cancel j OK

Figure 6.13 Classifying the defect

Should the user categorise the defect as a spall, the ‘spall detail’ window, shown in Figure

6.14 will appear, and the user is requested to enter detailed inform ation about the defect.

X]

Exposed Reinforcement Corrosion Rating Associated ] Other Details)

Corrosion looks like image indicated below

O Don't Know

Corrosion Rating (Percent): j 43 X

Slide bar to indicate picture which most looks like the corrosion

Back Next Cancel Finish

Figure 6.14 Entering corrosion information

319
 C hapter 6 - Review o f the expert system for reinforced concrete bridge repair

Figure 6.14 shows the ‘corrosion rating tab' in the ‘spall detail w indow ’. The user is

requested to judge the general condition o f any exposed reinforcement on a scale o f 0 to

100, with a series o f images to enable consistent results.

Figure 6.15 shows, once again, the ‘spall detail’ window, although this time w ith the ‘other

details’ tab selected. In this area the spall depth and other information can be added. It is

also possible in this view to give more inform ation about the shape o f the spall. For

example, had a rectangular defect been added, the user could set the shape to ‘perfect

rectangle’ - the knowledge base would recognise that the defect was likely to be a failed

previous repair.

f l Spall Detail X]

Exposed Reinforcement I Corrosion Rating | Associated Other Details

Is in splash zone?
Shape Irregular Rectangle
Represents:
(• No C Yes C Unknown Irregular Circle

Perfect Rectangle
In Wetted Area? Perfect Circle

r No (• Yes C Unknown Seapage:

low
medium
Is Low Covet Visible? high
unknown
No r Yes C Unknown
Moderate-Light Rain
In Impact Zone? Heavy Rain
Unknown
(t No C Yes C Unknown

Spall Depth: 145]

Back Next Cancel Finish

Figure 6.15 Entering more spall information

Figure 6.16 shows the status o f the element after the spall information has been added. As

discussed in the previous chapter, the vertical red band on the severity scale represents the

defect size. W ith a spall depth o f 45mm entered, the system has judged this defect as a

‘minor repair’. However, the defect falls between the zones ‘cosmetic repair’ and ‘minor

repair’. The system has chosen the most severe zone because, on this occasion, the user

320
 C hapter 6 - Review o f the expert system for reinforced concrete bridge repair

chose to leave a lot o f spall detail set to 'unknow n’ (such as the am ount o f exposed

reinforcement). Therefore the system uses the techniques developed in the previous chapter

to place the defect in the most severe zone.

et ■ H n T T r T T T i n T ^ r ^ r 'H T W T r r T 'T r p i T T T T H T T r r ^ y i
File Edit View D e fe c t Go S to c k A b norm al L o a d s R e p o r ts C u sto m ise H elp

4 | ^ X O | ij ; H 4- + ♦ 4 ii @
.................................. .......... 1 - 4 ’ d Q ! ^

Inspection: }I194 Scheduled: 06/01/2005 Grid Options: t 4- snap

T ? BR1006
O Pier/column Bridge View) 1194 ] Pier/column 1 Spall 456 j Photo ] Advice]
| Pier/column 1 X: p ig g Width: [ ~ § ^
f' Ref- [spall 456
'Tj Y: 11230 Height: j 3566
| Pier/column 2 Face: [tw it
| Pier/column 3 Spall Refresh Layout j
| Pier/column 4
Deck
Extent:
3 | Low
Severity:
w Deck 1
33 Is action required?
m*Deck 2
w Deck 3
Recemendation: | do Nothing
31
h Abutment Included in bridge
V Estimated Cost:
•f1 New Element condition?
*|* New Element Comments:
h Parapets Recommended i-----------------------------------
AA Parapets 1 Action:
M Parapets 2 Action Taken:

Diagnosis: iQarbonation corrosion 3

31

RHoHI
mrwn
^ Back to Structure

R eady Spall A r e c o rd h a s b e e n a d d e d to t h e E x p e rt A d v ice H isto ry

Figure 6.16 Judging spall severity

Figure 6.17 shows the insertion o f a generally rectangular defect onto a different face o f

the same column, in the same way as described previously. This shape will represent a map

cracking defect.

321
 C hapter 6 - Review o f the expert system for reinforced concrete bridge repair

f t USER-3WT251LF57\BMX - bridges - Demo-Bridges - Defect

File Edit View D e fe c t Go S to c k A bnorm al L o a d s R e p o r ts C u sto m ise H elp

4 41 “ X 0E3D A ' o
*1 Inspection: j 1194 Scheduled: 06/01 72005 t j Grid Options: -4- -ft -ft- ft. i ;j“
BR1006
.J Pier/column Bridge View j 1194 ] Pier/column 1 Map Cracking 458 ! Photo j i* i *
| Pier/column 1
n jMap Cracking 458 — X: Width:
^ E E E S n
r'Spall 456 Face: | right
~3 Y: I 1048 Height: 12447
O Spall 457 Map Cracking Refresh Layout j
| Pier/column 2 Extent: Priority: [Low
| Pier/column 3 d
| Pier/column 4 Severity: ji j Is action required? r
Deck
Recemendation: j o 0 Nothing
m*Deck 1 d
w Deck 2 Included in bridge
condition? 15* Estimated Cost:
m / Deck 3
i i Abutment Comments:
¥ New Element Recommended
¥ New Element Action:
Ju Parapets Action Taken:
AA Parapets 1
AA Parapets 2 x Diagnosis: j Drying Shrinkage 5
13

1
i■
1J
*j j Back to Structure

R eady m a p c rac k in g A re c o r d h a s b e e n a d d e d t o t h e E x p e rt A dvice H isto ry

Figure 6.17 Entering a map-cracking defect

Figure 6.18 shows the ‘map crack details’ window. If the user identifies a defect as map

cracking (Figure 6.13), this window will appear. The user can scroll through a series o f

images in order to identify the one which best represents the defect being entered. The

example in Figure 6.18 shows alkali-aggregate reaction at varying stages o f developing,

although, importantly, the user is at no stage told which type o f defect the image

represents. Once the user has chosen the most representative image, they are returned to

the element screen view, and, in the same way as shown previously for a spall defect, the

element condition is given a rating by the program (Figure 6.19). In this exam ple the user

chooses the severe AAR image. There are six other tabs through which other inform ation

about the defect can be entered.

322
 Chapter 6 - Review o f the expert system for reinforced concrete bridge repair

f t Map Cracks Detail

D etailed Q u estio n s j A sso c ia te d S e e p a g e ? j

Looks m ost like? | Gel P re s e n t? ] W h ite D e p o sit? ] A sso c ia te d S tain in g ?) A sso c ia te d Spalling?

m
p k
'
. ;
■
1

(• Mild D efect
-V ■*
-S :;>Y • ! ■
3 ■

. : %
yam
C M od erate D efect
ij? *

|v " .v - -1

C S e v e re D efect

C hoice S e t (b)

Figure 6.18 Choosing a representative image

In Figure 6.19, the user has clicked the ‘advice’ tab over the rightm ost window. This

causes the knowledge base to run. In this simple example, as would be expected, the

knowledge base advice is that the chances o f the cause being ‘early A A R ’ is low, and the

chances o f the cause being ‘A A R ’ is high. The knowledge base can make more

sophisticated judgem ents for other defects which are less simple to judge - such as the

differences between freeze-thaw damage, chloride corrosion and carbonation corrosion. It

is quite possible for a user to select the ‘freeze-thaw ’ image and for the system to still give

advice that the defect is possibly caused by, for example, chloride corrosion - depending

on the additional inform ation entered by the user.

 C hapter 6 - Review o f the expert system for reinforced concrete bridge repair

r m B B sm m tm -iflfx j
File Edit View D e fe c t Go S to c k A bnorm al L o ad s R e p o r ts C u sto m ise Help

i 4J %- X & 3 A -o □ ** u ! ® E1 * -► t * 1 ©
BSS3Q3HMHKI Inspection: 11194 Scheduled: 06/01/2005 jC Grid Options: <• + t ^
BR1006
p Pier/column 1194 ] Pier/column 1 i Map Cracking 458 j Photo i Advice j] < j*
| Pier/column 1
Map Cracking 458
n The possibility of the cause being due to EARLY AAR is: low

f'
Spall 456 The possibility of the cause being due to AAR is: high
O
Spall 457
| Pier/column 2
| Pier/column 3
| Pier/column 4
P Deck
m*
Deck 1
W Deck 2
W Deck 3
P Abutment
*1* New Element
■j* New Element
P Parapets
AA Parapets 1
AA Parapets 2
Diagnosis: [Drying Shrinkage 5
"3

j Back to Structure
R eady m ap crac k in g A re c o r d h a s b e e n a d d e d t o t h e E x p e rt A dvice H isto ry

Figure 6.19 Viewing expert system advice

It is possible to obtain a full report showing the decision taken by the expert system for any

defect. Figure 6.20 shows the knowledge based objects, and the values given to them by

the program for the defect in question.

324
 C hapter 6 - Review o f the expert system for reinforced concrete bridge repair

Jjnjxj
Map Cracking KBS Advice 3
The KBS requested the value of 'largedefect'
Value returned was: yes
The KBS requested the value of 'spalling'
Value returned was: UNKNOWN
The KBS requested the value of 'staining'
Value returned was: UNKNOWN
The KBS requested the value of 'horizontaldeck'
Value returned was: UNKNOWN
The KBS requested the value of 'previousrepair'
Value returned was: UNKNOWN
The KBS requested the value of 'whitedeposit'
Value returned was: UNKNOWN

Advice Output
The possibility of the cause being due to EARLY AAR is: low
The possibility of the cau se being due to AAR is: high

Diagnosis Output
The value for AAR is: high Saved OK
The value for Carbonation is: UNSET Saved OK
The value for Chlorides is: UNSET Saved OK
The value for Crazing is: none Saved OK
The value for Drying is: none Saved OK
The value for Early AAR is: low Saved OK

Save as txt print Close

Figure 6.20 Detailed knowledge base output

The system provides the user with a good am ount o f functionality for locating and

exam ining defects. In Figure 6.21, a 3D view o f the current column is shown, both the

defects entered can be seen.

325
 Chapter 6 - Review o f the expert system for reinforced concrete bridge repair

D em o bridge No. six - BR1006 - ln l>

File G ro u n d L ighting C u rs o r T y p e View A xes

I / Light Position
f ” Axes C Ground • Off Navigate Save
f“ Information <* Ground-On
Head Light j- Texture C Ground - Semi Select Copy I
Ambient Light
n_£.„.L I

Figure 6.21 3D view of affected column

Once a straight line crack defect has been input into the program, the decision making

processes o f the expert system begins. The initial decisions o f the expert system can be

seen im mediately in the zonal severity classification area (lower right - Figure 6.22). The

system will then present a window requesting further inform ation about the crack, such as

its width, associated staining etc. Knowledge bases then give recom m endations for the

crack and inform ation is delivered back to the user.

326
 Chapter 6 - Review o f the expert system for reinforced concrete bridge repair

- 1.S.I x j
File Edit View D e fe c t Go S to c k A bnorm al L o a d s R e p o rts C u sto m ise Help

« 4J | “ X oi # J A - jg g e g | -M 0 | • h | ♦ ♦ ♦ * f K ®
l---------- -------- ----------------------------------------- j-----------------
2) Inspection: |I I 94 Scheduled: 06/01/2005 d Grid Options:
TJ BR1Q06
Ci Pier/column Bridge View j 1194 ] Pier/column 3 Crack 460 j Photo j Advice j
| Pier/column 1 X:
Ref- p a c k 460 | 125 Width: 1 182
<§5> Map Cracking 4!
Face: |, eft
C Spall 456 3 Y: [4848 Height: f-2473
f
Spall 457 Structural Crack Refresh Layout |
| Pier/column 2
Extent: | Priority: (Low d
| PieMcolumn 3
<< MEHiEliil Severity:
Jl d Is action required? f~
r-
Spall 459
Recemendation:
| Pier/column 4
j Do Nothing d
2i Deck Included in bridge
Estimated Cost:
condition?
W Deck 1
Deck 2 Comments:
w Deck 3 Recommended
Ar Abutment Action:
^ New Element Action Taken:
*|* New Element
£i Parapets D ia g n o sis' j
AA Parapets 1 ............... 3
AA Parapets 2

Back to Structure

R e a d y ... s tr u c tu r a l c r a c k C o n d itio n : No R e p a ir (C e r ta in ) - E x p e rt A dvice is A vailable

Figure 6.22 Entering a crack defect

327
 C hapter 6 - Review o f the expert system for reinforced concrete bridge repair

6.5 Elements - Testing and repair

As soon as all the defects affecting an element within the program have been added, the

system is ready to run the testing knowledge bases. The example shown in Figure 6.23 is

for a column with two significant defects. The first defect, map cracking, was judged as

being caused by either freeze-thaw action or carbonation corrosion. The second defect, a

spall, was judged as being caused by chlorides. Once the user is satisfied that all the

present defects have been entered into the system, the ‘Testing A dvice’ tab is clicked. This

prompts the system to run the testing knowledge bases, and the advice shown in Figure

6.23 is generated.

* Report

Testing KBS AdYice Run at 08:50:33 on 24 Jan 2005

Defect Cause Scores

Defect Craz. Impact Chlor. Carb. Plast. Freez. AAR Ear.AAR Drying EC
Map Cracking 407 0 0 0 35 0 65 0 0 0 20
Spall 406 0 0 50 50 0 0 0 0 0 80

Advice Output
Due to the presence of chlorides, the main testing regime on this element should be Half Cell testing and Chloride sampling
AND Due to the advance of carbonation, the main testing regime on this element should be Covermeter testing to establish
the depth of cover, and carbonation tests.

Main Testing Advice Output Objects

The main scheme advised testing for Chlorides
The main scheme advised testing for Carbonation

Additional AdYice Output For Defect 407

Weathering has cause this defect. No further testing.

Additional Advice Output For Defect 406

Due to the presence of chlorides, this defect requires additinal testing of Half Cell testing and Chloride sampling. AND Due
to the advance of carbonation, the main testing regime on this element should be Covermeter testing to establish the depth
of cover, and carbonation tests.

Save as txt , print Close

Figure 6.23 Element testing advice from the knowledge base

The main testing knowledge base looks at the element as a whole, and its advice, based on

the causes o f the defects present, is to test for chlorides and carbonation. Thereafter, each

defect is exam ined individually to see if its cause falls under the advice o f the main testing

knowledge base. In the example o f Figure 6.23, defect 407 (map cracking defect), does not

require any additional testing over and above that prescribed for the elem ent as a whole.

Similarly, defect 406 (the spall), falls under the general advice for the element. Flowever,

328
 C hapter 6 - Review o f the expert system for reinforced concrete bridge repair

should any o f the defects be diagnosed as having a considerably different cause than the

element as a whole, specific testing advice for those defects would be generated.

Once the testing for the element has been undertaken, repair o f the structure, if necessary,

can commence. Figure 6.24 shows repair advice for a wide crack on a flat horizontal

surface, having been attributed a width o f 20mm. The advice o f the system is to repair the

crack with an overlay.

HowTo Repair Structural Crack KBS

Run at 08:54:57 on 24 Ja n 2005

Data Inputs
The KBS requested the value of 'moisture'
Value returned was: moderatewater
The KBS requested the value of 'movementcondition'
The value was UNKNOWN
User provided value: dormant
The value was not saved
The KBS requested the value of 'strengtheningrequired'
Value returned was: no
The KBS requested the value of 'width'
Value returned was: 20

Repair Advice Output

Blanketing Active or dormant cracks not requiring strengthening ???
Ordinary overlay Used to treat static cracks on flat horizontal surfaces. Often using a heavy coat of epoxy resin or an
overlay of polymer modified cement.

Repair Advice Output Objects

blanketing = yes
blank.txt was loaded

Figure 6.24 Element repair advice from the knowledge base

Figure 6.24 is the area o f the program where the different types o f repair advice shown in

Table 5.25 (Chapter 5) are displayed. For many significant defects, this advice w ould read

‘break out and repair’. If this advice occurs, the repair material selection routines can be

employed.

329
 C hapter 6 - Review o f the expert system for reinforced concrete bridge repair

6.6 Repair material selection

The repair material selection operation o f the program is the area in which the key

developm ent work o f this thesis is employed. Once an inspection has been completed, and

the user has confirmed the findings o f the knowledge bases (or corrected their findings

should test results have proved them wrong), the advice o f the repair knowledge base may

well have been to break out and repair the affected concrete. Should this be the case, the

system user is required to indicate the extent o f the patch repair that will be carried out.

Once this action has been completed, the repair material selection procedure begins.

Figure 6.25 shows the database o f reinforced concrete repair materials and their properties

at 28 days age. These com mercially available materials come ready program m ed into the

software and the user has the opportunity to add an unlimited am ount o f further materials.

Y S jiL l
File View Suppliers Go Stock Abnormal Loads Reports Customise Help

4 m - x bo # I Q n fi 4- -► ♦ * □ % ® S )0 & 0 & 0 n & rx

-I
Material Name Material Supplier
t l BR1006
Ci Pier/column Y
| Pier/column 1 Proton Microncrete Proton
^ Map Cracking 4! Flexcrete FCR 845 Flexcrete
Spall 456 Test Material Flexcrete
# Spall 457 ► irfiiiitg ig i SBD1
New Grid SBD Multifix (spray) SBD1
| Pier/column 2 *
| Pier/column 3
r - Spall 459

| Pier/column 4
Deck
Deck 1
w Deck 2
Material Property Value Units Code
w Deck 3
Ci Abutment Y , _________ I j ___________ I ;
¥ New Element ► Compressive Strength 65 N/mm2 BS 1881-121 (1983)
*|* New Element Tensile Strength 6 N/'mm2 BS 1881-121 (1983)
Ci Parapets Shrinkage 800 microstrain ASTM C469-94 (1994)
AA Parapets 1 Creep Strain 400 microstrain BS 1881-121 (1983)
AA Parapets 2 Stess/Strength Ratio 30 BS 1881-121 (1983)
Strength 32 N/mm2 BS 1881-121 (1983)
Bond Strength 0 N7mm2 BS 1881-121 (1983)
Elastic Modulus 24 Gpa BS 1881-121 (1983)

.11 ..................... - ....................1 H G o t o S u p p l i e r s L is t

Ready...

Figure 6.25 Manufacturers' test data for repair materials

330
 C hapter 6 - Review o f the expert system for reinforced concrete bridge repair

W hen ready to proceed, the user is asked to enter the substrate information, as shown in

Figure 6.26. The system is already aware o f the size o f the repair. The bridge location

should have been entered by the user at an earlier stage (during population o f the structure

database) and, therefore, the software is ready to determine geographical climate effects.

The remaining unknown data is gathered in the ‘substrate properties screen’, specifically

substrate compressive strength (N/m m 2) and elastic modulus (kN /m m 2). The height and

diameter o f the core is required in order to apply the relevant factors and the scheduled

date o f the repair will allow the climate effects to be determined correctly. Upon clicking

OK, the performance o f all the repair materials in the database is assessed for the repair.

The results are specifically tailored for the size o f the repair, the strength and elastic

modulus o f the substrate, the location o f the bridge, the date the repair will be undertaken

and the size o f the core taken from the structure. Repair on different structures, in different

places at different times, will produce different results. The exam ple shown here is for the

spall in Figure 6.16, in Edinburgh, with the additional details from Figure 6.26.

w m m m m n m m - _ iQ ixi

Cylinder Height: (250

Cylinder Diameter: [lOO

Substrate Strength: (58

Substrate Modulus: (28

Sheduled Date: 11/12/2005)

OK (

Figure 6.26 Substrate information

331
 C hapter 6 - Review o f the expert system for reinforced concrete bridge repair

O f the five repair materials in the database, using the techniques developed in Chapter 4 o f

this thesis, three would perform adequately. Figure 6.27 shows the three successful

materials listed in the top left window.

& R e p a ir A dvice -Jaixj

r [Show failed Materials?! Legend:
Proton Microncrete from Proton [Pass
Flexcrete FCR 845 from Flexcrete [Pass] Strain Capacity — —
Test Material from Flexcrete [Pass]
Warning Zone

Strain in Repair — i- ■

300

200

54.0 days, 116.3 microstrain

View: (• Show Graph C Show Grid Generate Report. . i

S
#-« a
.I r^riirrn oiMT^r.in—
Mnuu !J-l
I Iff*8 1 * J. _ . . C I HW '—, m .TL.

Figure 6.27 Performance of Proton Microconcrete

The first successful material, shown in Figure 6.27, is Proton M icroconcrete. The blue line

represents the strain capacity o f the repair material. As the material shrinks, the restraint to

this shrinkage at the interface between the repair and the substrate causes tensile strains,

shown by the red line. These tensile strains continue to increase up to 200 days before they

plateau at a value o f approximately 170 microstrain. The strain capacity, being 200

microstrain at 200 days, is greater than the strain in the repair material and, therefore, the

material performs successfully although perhaps, in this case, the margin o f success is less

than desirable. The green dotted line represents an additional factor o f safety, materials

with restrained shrinkage strains above this line will be classed as failed.

332
 Chapter 6 - Review o f the expert system for reinforced concrete bridge repair

Figure 6.28 shows the performance o f the Flexcrete material. This material performs in a

noticeably different manner to that in Figure 6.27. At approxim ately day 12, the elastic

modulus o f the repair material becom es greater than that o f the substrate concrete.

Consequently, some o f the shrinkage strain in the repair material in transferred into the

substrate concrete in accordance with equation 4-6 (Chapter 4). As the elastic m odulus o f

the repair material continues to develop, more and more o f the developing shrinkage

strains are transferred into the material until, as approxim ately day 50, the repair material is

stiff enough to transfer all its strain into the substrate. This material would, therefore, be a

very safe material to employ in this situation.

A R e p a ir A dvice -lP tx|

f “ Show failed Materials? Legend:
Proton Microncrete from Proton [Pass]
Flexcrete FCR 845 from Flexcrete [Pass] Strain Capacity

Warning Zone

Strain in Repair

11.9 days, 338.9 microstrain

View: (S Show Graph C Show Grid Generate Report...

Figure 6.28 Performance of Flexcrete material

If the user checks the box in the top left hand corner o f the screen, the failed m aterials will

also be displayed. This is shown in Figure 6.29.

 Chapter 6 - Review o f the expert system for reinforced concrete bridge repair

" ic ljn licnr. Prolcm [P s ss l

Fteacrete FCR 8 4 5 lio n Ftexirele P a s s ] 5 liffn C apacity
I ■ I H it-.! liiv - - ,i ,•
W aiem ! C jrAll j V /a r r s Z en*

S lia n ifi R<*(a1i

;vri S' »1>V- '•

V ert (* 5t<;v, CjihJ^i r 5 h O rtG « ] BwenaU* RttpOtf.

Figure 6.29 Failed material

M aterial C in Figure 6.29 has a very high elastic modulus, and quickly becomes much

stiffer than the substrate concrete. However, the material has a high shrinkage, and before

the repair material has become stiff enough to transfer restrained tensile strains into the

substrate, its strain capacity has already been exceeded. The material fails at approxim ately

day 10. Thereafter the shown performance m ust be disregarded. This material would

obviously be avoided for the particular repair situation which generated the shown result.

334
 Chapter 6 - R eview o f the expert system for reinforced concrete bridge repair

6.7 Summary and Conclusion

The commercial partners in the research presented in this thesis have built a bridge

management system around the thesis recommendations. The bridge management system

is a database through which an organisation’s bridge stock can be organised and managed

from inspection through to prioritisation and maintenance.

In order for the expert system to have the necessary intelligence to make useful decisions,

there was a need for the program to gather geometrical and geographic information about

the structure being examined, which led to the development of the interface that allows

users to assemble structures on the screen from their basic elements. This feature is also

used for the efficient inputting of defects, allowing the system to gather the information it

needs to make decisions based on the relative size of elements and defects.

As the user adds information about the nature of defects, the various knowledge bases

begin their decision making processes and their findings are displayed to the user. The

expert system performs a number of key functions:

o Diagnoses the cause of a defect

o Rates the severity of the defect on a scale from 0 to 100

o Rates the condition of an element based on all the defects affecting it

o Offers recommendations of which tests to perform on an element based

on all the defects affecting it

o Offers repair advice for each defect on an element, based on the

recommendations of the knowledge bases which preceded the repair

advice stage

o Chooses repair materials which w ill perform adequately should any

defects require to be broken out and repaired

3 35
 Chapter 6 - R eview o f the expert system for reinforced concrete bridge repair

The purpose of the system is to act as an intelligent advisor at all stages of the concrete

repair process, from defect identification through to repair. The created system achieves

this task.

A simple and effective method for determining the extent and severity of reinforced

concrete defects has been developed. Structures can be quickly modelled within the

software, and defects can be added onto the modelled structures. As a result, the software

is immediately aware of the extent of defects.

In conjunction with experienced concrete repair practitioners, a system has been developed

to allow the software to place any of the three key defect types (spalling, map cracking,

structural cracking) into one of four ‘decision zones’. These decision zones define the four

likely repair categories for a defect, namely, ‘do nothing’, ‘cosmetic repair’, ‘minor repair’,

’major repair’. Primarily, the position of these zones is decided using key factors. For

example, the key factors for a spall are size and depth; for map-cracking the key factors are

size and the defect cause suggested by the visible cracking. Thereafter, zone positioning is

altered depending on secondary factors. For example, secondary factors for a spall may be

the amount of reinforcing steel exposed by the spall, the condition of the steel, or the

amount of staining and seepage associated with the spall.

Knowledge bases use information such as the repair zone into which a defect has been

placed, the defect’s shape, its proximity to the carriageway and the bridge age, to decide

likely defect causes. Additional knowledge bases examine all the defects affecting an

individual element, and recommend testing regimes to confirm the defect causes. Finally, a

third layer of knowledge bases recommends how to repair the defects.

336
 Chapter 7 - Conclusions

7 Conclusions

The repair of reinforced concrete is the subject of broad ranging research due to the

ubiquitous maintenance requirements of reinforced concrete in all environments.

The stated aim of this research was twofold:

• To examine, and develop further, existing state of the art research into

compatibility between reinforced concrete repair materials and the substrates on

which they are employed. Furthermore, to prepare a method for determining the

long-term insitu performance of repair materials, which, when employed in a

software system would enable identification of repair materials suitable to perform

in the situations required.

• To prepare an expert system for concrete repair, to work in conjunction with the

repair material selection system, which w ill output intelligent advice at all stages of

the reinforced concrete inspection and repair process.

In the software developed, the relationship between traditional computer programming to

assess the severity and extent of defects, and less traditional expert system techniques for

decision making works seamlessly and effectively. The simple methodology of expert

system development devised in this research could be employed with similar effect in

many others areas particularly where complex objects can be placed into sets (such as the

way reinforced concrete defects are placed into the four repair decision zones). The

practice adopted was to identify two key factors in decision making, for example, to

determine the severity of a spall, the size and depth of the spall were the key factors.

Experts were asked to relate these two factors graphically. This allowed the experts’

337
 Chapter 7 - Conclusions

opinions to be represented with numerical equations, providing a sound initial estimate of

the severity of a defect. This initial estimate can be refined by considering additional

factors, also represented numerically based on expert opinions.

Thus the method of knowledge elicitation developed in the research to assess reinforced

concrete defects is effective in incorporating cumulative knowledge of practising experts.

This approach was considered a great benefit as it allowed the expert system to be

developed within the timeframe available, ensuring adequate time for other key aspects of

the research to be undertaken. The methodology adopted has produced sensible and

reliable results. It provides a relatively simple and practical approach for expert system

development in the field.

The performance of reinforced concrete repair materials is a field of research with a broad

range of varying opinions amongst experts. In particular, expert opinion varies regarding

which properties of materials are important to specify. This research has reviewed and

considered the spectrum of opinion and has adopted the premise, proven through field

testing, that elastic modulus, shrinkage and creep are the crucial properties for the

performance of concrete repair. Strain in a reinforced concrete repair material is affected

by the growth of these properties, their effects on one other, and interaction with the

substrate. These phenomenons have been incorporated into a routine which can predict the

growth of tensile strains in repair materials with time, and make comparisons against the

tensile strain capacity of materials. The method developed to predict the performance of

reinforced concrete repair materials is based on the measured field performance o f a

variety of materials. It adequately models tensile strain resulting from restrained shrinkage

in repair patches. The procedure allows an accurate assessment of the performance of a

338
 Chapter 7 - Conclusions

repair material to be made. Thus selection of materials can be made in an auditable and

scientific way, rather than on an ad-hoc basis.

The routine developed has been incorporated into a software program. Lengthy iterative

calculations are performed by the software and users are graphically informed when and

how unsuitable materials will fail. Accordingly, materials shown to perform well can be

selected, and engineers can justify the choice.

Over-arching the repair material selection software, and the inspection and repair expert

advice software, is a structures management system, which seamlessly ties together the

research outlined in this thesis. The software will advise engineers on the cause of a defect,

this advice can be confirmed by testing. A testing regime will also be recommended by the

system. The software advises the engineer on how to repair a defect, and will filter out

repair materials that are suitable for use from those which are not.

During interaction between the software and the practitioner, the opportunity is always

available for engineers to make the final decision themselves, either in consultation with

more experienced colleagues or through a review of the available literature. Moreover, a

practitioner can be reassured that if his opinion agrees with that of the expert system, then

the opinion of the vastly experienced panel of experts interviewed in the preparation of this

research would also agree. Therein is the overall goal and originality of the software. To

take the cumulative knowledge of the concrete repair practitioners, and the models

developed for long term concrete repair material performance, and to accurately represent

these in a responsive, adaptable computer program.

339
 Chapter 8 - Further Work

8 Further Work

8.1 Field testing and calibration o f the expert system

A rigorous assessment of the performance of the expert system in the field should be

conducted. The results of such an assessment would be used to calibrate the expert

system’s performance so as to achieve maximum accuracy of diagnosis of defect and

assessment of severity. This can be done through liaison with practitioners who use the

system on site in genuine situations. If necessary, expert system rules can easily be

amended to incorporate any revised opinions that arise from examining the expert system’s

performance.

It is important to assess how engineers agree with the severity ratings generated by the

system for concrete elements and individual defects. The system has been developed in

such a way that clear differences of opinion between experts and the system, in the field,

can be reported to the software suppliers and easily remedied by modification of the many

constant factors used to describe the collaborating experts’ opinions.

8.2 Field testing to assess the performance of the concrete repair

material property selection system

The software components within the expert system that recommend optimum properties

for repair materials should be tested in the field. The program can be used to select repair

materials that will perform adequately - the success of materials selected by the system

340
 Chapter 8 - Further Work

will demonstrate the veracity of the routine developed. However, it would be advantageous

to be able to specify materials which the software shows will fail; how accurately the

software predicts the time of failure of these repairs would be a good judge of its

performance.

8.3 Prioritising the repair o f bridges and bridge elements.

Some bridge elements are more important, when determining the condition of the overall

structure, than others. For example, a severely corroded wingwall may have little impact

on the performance of the structure as a whole, whereas a mildly affected central pier may

be very significant. Because of the many different types of bridge design, a good deal of

research may be necessary to endow the expert system with the intelligence necessary to

recognise the importance of individual elements. However, if this task were completed,

both prioritisation of element repair, and an accurate overall structure rating, would be

relatively straight forward to develop. Comparing overall structure condition could be used

to prioritise the repair of bridges, although, again, some bridges are more important than

others. Decisions on which bridges to repair are not related solely to their condition, but

also to factors such as location, use, the likely consequences of further deterioration,

factors such as funding, value management and even local politics.

8.4 Expanding the expert system capability

This thesis has been concerned specifically with concrete defects and repair. However, the

bridge management system which over-arches the software tools developed herein can

manage all variety of bridge types: steel, concrete, masonry arches, culverts etc. Expert

systems to diagnose defects and recommend repairs on other types of structure could be

341
 Chapter 8 - Further Work

prepared. As concrete structures seem to be subject to more maintenance difficulties than

other structures, it is conceivable that the development of expert system for the other

structural types could be more straightforward. The logical methods for assessing extent

and severity developed in this thesis could also be employed in these additional modules.

342
 Chapter 9 - References and Bibliography

9 References and Bibliography

9.1 References

1. Sohni M. Expert system for maintenance and concrete repair. Darmstadt

Concrete, 4, 1989. (pp. 245-255)

2. The Concrete Society. Concrete Bridges: Investigation, maintenance and

repair. 1985.

3. Baker D. The maintenance of the Highways Agency's structures.

International Conference on the management o f highway structures.
London, 1998.

4. Wallbank E. J. The performance o f concrete bridges - A survey o f 200

highway bridges. HMSO,1989.

5. Dunham V. Deterioration and repair of reinforced concrete. Building

Engineer, October\993. (pp. 16-21)

6. Little D. Effective repair and protection of concrete. Construction Repairs

and Maintenance, September 1986. (pp. 23-25)

7. Coleman J. E. Our aging infrastructure: providing for public safety and

protecting public investment. Proceedings o f SPIE - The International
Society fo r Optical Engineer ing.2456,1995. (pp. 84-91)

8. Draft - unpublished. Highways Agency Bridge Inspection Manual Volume

2: Concrete Bridges & Concrete Elements. Highways Agency. 2002

9. Department of Transport. The Design Manual fo r Roads and Bridges.

Volume 3 - Highway structures: Inspection and Maintenance. HMSO,1985.

10. Department of Transport. Bridge Inspection Guide. HMSO, 1983.

11. Narasimhan S. and Wallbank J. Inspection manuals fo r bridges and

associated structures. The Institution of Civil Engineers, 1998.

12. Mallet G. Repair of concrete bridges. Construction Repair, Jan/Feb 1995.

(pp. 22-24)

13. Frangopol D. M., Estes A. C., Augusti G., and Ciampoli M. Optimal bridge
management based on lifetime reliability and life-cycle cost. Proceedings o f
the International Workshop on Optimal Performance o f Civil Infrastructure
Systems. 1997. (pp. 98-115)

343
 unapter y - Keterences and tfibiiograpny

14. Marosszeky M. and Gowripalan N. Repair of corrosion induced failures on

reinforced concrete. Concrete. June 1998

15. Degerlund C. Corrosion and concrete repair. Civil Engineering, August

1983.

16. Iffland J. S. B. and Bimstiel C. Causes of bridge deterioration, Bridge

management 2. Inspection, maintenance assessment and repair. Papers
presented at the second international conference on bridge management,
held at the University o f Surrey. Guilford, 1993. (pp. 9-17)

17. Williamson S. J. and Clark L. A. An investigation of the amount of

corrosion required to cause cracking. Concrete, February 1999. (pp. 117-
128)

18. Anumba C.J. and Bowron J. A system for the institution of effective repairs
to concrete Structures. Computing In Civil Engineering, June 1992. (pp.
160-166)

19. Bennison P. Repair and protection of reinforced concrete bridges. 1999.

20 . Allen R. T. L. and Forrester J. A. The investigation and repair of damaged

reinforced concrete structures. Corrosion o f reinforcement in concrete
construction. Society o f chemical industry, 1983. (pp. 223-234)

21 . Emmons P. H. Concrete repair and maintenance illustrated.

RsMeans,1993.

22 . O’Flaherty, F.J. and Mangat, P.S. Recommendations for the European

Prestandard for concrete repair. 2nd International RILEM Workshop on Life
Prediction o f Concrete Structures. Paris, France. 5-6 May 2003.

23. British Standard Institution, Products and systems for the protection and
repair of concrete structures - Definitions, requirements, quality control and
evaluation of conformity, DD ENV 1504-1. 1997.

23a. Khatib, J.M. and Mangat, P.S. Influence of high-temperature and low-
humidity curing on chloride penetration in blended cement concrete.
Cement and Concrete Research. 32 (2002) (pp. 1743-1753)

24. Mallet G. Repair o f concrete bridges - a state o f the art review. TRL, 1994.

25. Blight G. E. Rehabilitation of reinforced concrete structures affected by

alkali-silica reaction. Structural engineering review,!, June 1990. (pp. 113-
120)

26. Clark L. A. Assessment o f concrete bridges with ASR. Bridge management

2. Inspection, maintenance assessment and repair. Papers presented at the

344
 Chapter y - Keterences and Bibliography

second international conference on bridge management, held at the

University o f Surrey. Guilford, 1993. (pp. 19-28)

Mays G. C. Durability o f concrete structures. E&FN Spon, 1992.

The Concrete Society. Non-structural cracking in concrete. Concrete society

technical paper no.2, 1982.

A.C.I. committee. Causes, evaluation and repair of cracks in concrete. ACI

Journal, May/June 1984.

http://www.vseal.com/surfacedefects/surfacedefects.php.Cottcrete surface
defects, 1997.

Currie, R. J Carbonation depths in structural-quality concrete . Concrete,

1986.

Swamy R. N. Assessment and rehabilitation of AAR-affected structures.

Cement and concrete composites, 19. 1997. (pp. 427-440)

Houde J., Lacroix P., and Momeau M. Rehabilitation of railway bridge

piers heavily damaged by AAR. Issue Proceedings o f the 7th international
conference on concrete Alkali aggregate reactions, Ottawa, Canada, 1986.
(pp. 163-167)

Wood J. G. M. and Angus E. C. Montrose bridge. Inspection assessment

And remedial work to a 65 year old bridge with AAR. Proceedings o f the
6th International conference on structural faults and repair. 1995. (pp. 103-
108)

Vassie P. R. Secondary effects of ASR on bridges: corrosion of reinforcing

steel. Bridge management 2. Inspection, maintenance assessment and repair.
Papers presented at the second international conference on bridge
management, held at the university o f Surrey. Guilford, 1993. (pp. 29-37)

Pearson-Kirk D. and Jayasundara P. Impregnation o f concrete highway

structures. Thomas Telford, Bridge management 2. Inspection, maintenance
assessment and repair. Papers presented at the second international
conference on bridge management, held at the university o f
Surrey.Guilford, 1993. (pp. 39-48)

Lambert P. Reinforced concrete repair reviewed. Highways,64, 5, 1998.

Allen R. T. L. Choosing a concrete repair system. Construction Repairs and

Maintenance, Sept 1986. (pp. 10-12)

Kalyanasundaram P., Rajeev S., and Udayakumar H. REPCON: Expert

System for building repairs. Journal o f computing in civil engineering,4,2,
1990. (pp. 84-101)
 unapter v - Kererences ana rsibiiograpny

40. Higgins D. Repairs to cracks in concrete. Concrete Repairs, 1999. (pp. 32-
33)

41. Flint A. R. and Husband M. V. Bridge assessment: the need for codified
rules. Thomas Telford, 1993. Bridge management 2. Inspection,
maintenance assessment and repair. Papers presented at the second
international conference on bridge management, held at the university o f
Surrey .Guilford, 1993. (pp. 897-910)

42. Canadian Strategic Highway Research Program. Highway Concrete

(HWYCON) Expert System. Technical brief 4, 1995.

43. Cabrera J. G., Kim K. S., and Dixon R. CODBA: An expert system for the
assessment of deterioration of concrete bridges. Developments in artificial
intelligence fo r civil and structural engineering, 1995. (pp. 151-157)

44. Ohtomo T. Expert system for maintenance of reinforced concrete structures.

Durability o f building materials and components. Proceedings o f the sixth
international conference held in Omiya, Japan. 26-29 October, 1993. (pp.
1254-1263)

45. Miyamoto A., Morikawa H., Kushida M., and Tokuyama T. A knowledge
based expert system application in structural safety assessment. Bridge
management 2. Inspection, maintenance assessment and repair. Papers
presented at the second international conference on bridge management,
held at the university o f Surrey. Guilford, 1993. (pp. 96-109)

46. Miyamoto A., Kimura H., and Nishimura A. Expert system for maintenance
and rehabilitation of concrete bridges. Concrete Bridges. 1998. (pp. 207-
217)

47. Rajeev S. and Rajesh J. An expert system for diagnosing causes and repair
of defects in RC structures. The Indian Concrete Journal, Jan. 1995. (pp.
31-36)

48. Kushida M., Miyamoto A., and Kinoshita K. Development of concrete

bridge rating prototype expert system with machine learning. Journal o f
computing in civil engineering, October 1997. (pp. 238-247)

49. Furuta H., He J., Watanabe E., and Umano M. A fuzzy neural expert system
for repairing bridge decks, Journal o f computing in civil engineering. 1993.
(pp. 283-292)

50. Furuta H., He J., and Watanabe E. A fuzzy expert system for damage
assessment using genetic algorithms and neural networks. Microcomputers
in civil engineering, 11, Jan. 1996. (pp. 37-45)

51. Khan, M. S. Bridge Management systems: past, present, and future.

Concrete International, August 2000. (pp. 53-56)

346
 unapter y - Kererences ana tsibiiograpny

Thoft-Christensen P. Advanced bridge management systems. Structural

engineering review, 7, 3, 1995. (pp. 151-163)

Thoft-Christensen P. and Sorensen J. D. Optimal Strategy for inspection

and repair of structural systems. Civil Engineering systems, 4,1987. (pp. 94-
100)

Thompson P. D.'TONTIS" Characteristics of bridge management systems.

Transportation Research Circular N o. 423, National Academy Press,
Washington D.C. 1994. (pp.35-43)

Mangat P. S and O'Flaherty F. J. Long-term performance of high-stiffness

repairs in highway structures. Magazine o f concrete research. 51, Issue
5,10-1-1999.

Mangat P. S. and Limbachiya M. K. Repair material properties which

influence long-term performance of concrete structures. Construction and
Building Materials, 9, No 2, 1995. (pp. 81-90)

Mangat P. S. and Limbachiya M. K. Repair material properties for effective

structural application. Cement and Concrete Research, 27, No. 4, 1997. (pp.
601-617)

O'Flaherty F. J. 'Long Term Performance o f Concrete Repair in Highway

Structures ’ PhD thesis, Sheffield Hallam University, UK, 1998.

Robin Stirling. The Weather o f Britain. Giles de la Mare, 1997.

Bennision P. Materials for concrete repair and protection - innovation and

performance. Construction Repair, July/August 1992. (pp. 27-31)

Emberson N. K. and Mays G. C. Significance of property mismatch in the

patch repair of structural concrete. Parts 1-3. Magazine o f concrete
research,48, Issue 174,1996. (pp. 45-57)

Mangat P. S and O'Flaherty F. J. Serviceability characteristics of flowing

repairs to propped and unpropped bridge structures. Materials and
structures,32,1999. (pp. 663-672)

Mangat P. S and O'Flaherty F. J. Influence of elastic modulus on stress

redistribution and cracking in repair patches. Cement and Concrete
Research,30,2000.

Decter M. H. and Keeley C. Durable Concrete Repair - Importance of

compatibility and low shrinkage. Construction and Building Materials. 11,
Issue 5-6,1997. (pp. 267-273)

Hewlett P. C. and Hurley S. A. The consequences of polymer concrete

mismatch. Design Life o f Buildings, 1985. (pp. 179-210)
 Chapter 9 - References and Bibliography

66. Department O f Transport. Materials fo r the repair o f concrete highway

structures. BD 27/86, 1986.

67. Rizzo E. M. and Sobelman M. B. Selection criteria for concrete repair

materials. Concrete International, September 1989. (pp. 46-49)

68. Poston R. W., Kesner K., McDonald J. E., Vaysburd A. M., and Emmons P.
Concrete Repair Material Performance - Laboratory study. AC I materials
journal March-April 2001. (pp. 137-147)

69. Emmons P. and Vaysburd A. M. System concept in design and construction

of durable concrete repairs. Construction and Building Materials, 10, Issue
1,1995. (pp. 69-75)

70. Vaysburd A. M. and Emmons P. How to make today's repairs durable for
tomorrow - corrosion protection in concrete repair. Construction and
Building Materials, 14,2000. (pp. 189-197)

71. Shambira M. V and Nouno G. F. Numerical simulation of shrinkage and

creep in patch repaired axially loaded reinforced concrete short columns.
Computers & Structures, 79,2001. (pp. 2491-2500)

72. Saucier F., Claireaux F., Cusson D., and Pigeon M. The challenge of
numerical modelling of strains and stresses in concrete repairs. Cement and
Concrete Research,27, Issue 8,1997. (pp. 1261-1270)

73. Baluch M. H., Rahman M. K., and A1 Gadhib A. H. Simulation of shrinkage

distress and creep relief in concrete repair. Composites Part B: Engineering
(UK). 31B, Issue 6-7, 2000. (pp. 541-553)

74. Morgan D. R. Compatibility of concrete repair materials and systems.

Construction and Building Materials. 10. No.l, 1996. (pp. 57-67)

75. Mays G. C. and Barnes R. A. The structural effectiveness of large volume

patch repairs to concrete structures. Proceedings o f the institution o f civil
engineers structures and buildings. 110. November 1995. (pp. 351-360)

76. McDonald J. E., Vaysburd A. M., and Poston R. W. Performance Criteria

for dimensionally compatible repair materials. High performance material
and systems research program. Information Bulletin 00-1. January 2000.
(pp. 751-765)

77. Decter M. H. and Lambe R. W. New materials for concrete repair. The
Indian Concrete Journal. October, 1993. (pp. 475-480)

78. Plum D. R. Materials - what to specify. Construction Maintenance &

repair. July/August, 1991.

348
 Chapter 9 - References and Bibliography

Plum D. R. The behaviour of Polymer materials in concrete repair and

factors influencing selection. The Structural Engineer.^8, Issue 17,1990.
(pp. 337-345)

Wood J. G. M., King E. S., and Leek D. S. Defining the properties of

concrete repair materials for effective structural application. 1998.

Mangat P. S and Elgarf M. S. Strength and serviceability of repaired

reinforced concrete beams undergoing reinforcement corrosion. Magazine
o f concrete research. 51, Issue 2, 1999. (pp. 97-112)

Neville A. M. Properties o f Concrete. Longman, 1995.

Brookes J. J and Neville A. M. A comparison of creep, elasticity and

strength of concrete in tension and compression. Magazine o f concrete
research.29, Issue 100, 1978.

Bissonnette B. and Pigeon M. Tensile Creep at early ages of ordinary, silica

fume and fibre reinforced concretes. Cement and Concrete Research.25,
Issue 5,1995. (pp 1075-1085)

Altoubat S. A. and Lange D. A. Tensile basic creep: measurements and

behaviour at early age. ACI materials journal.98, Issue 5,1998. (pp. 386-
393)

ASTM C 42 - 90. Tests fo r obtaining and testing drilled cores and sawed
beams o f concrete. 1990.

British Standard Institution. BS 1881-116:1983. Testing concrete - Method

fo r determination o f compressive strength o f concrete cubes.

ASTM C39 - 94. Standard test method fo r compressive strength o f

cylindrical concrete specimens. 1994.

British Standard Institution BS 1881-121:1983. Methodfor determination o f

static modulus o f elasticity in compression.

ASTM C 469 -94. Standard test method fo r static modulus o f elasticity and
poisson's ratio o f concrete in compression. 1994.

ASTM C580 - 93. Standard test method fo r flexural strength and modulus
o f elasticity o f chemical resistant mortars, grouts, monolithic surfacings
and polymer concretes. 1993.

ASTM C 78-94. Standard test method fo r flexural strength o f concrete

using simple beam with third point loading. 1994.

ASTM C 293-94.Standard test method fo r flexural strength o f concrete

using simple beam with centre point loading.\994.
 Chapter y - References and Bibliography

94. British Standard Institution. BS 1881-118 1983.Methodfor determination o f

flexural strength, (thirdpoint loading). 1983.

95. ASTM C 531 - 95. Standard test method fo r Linear Shrinkage and
Coefficient o f Thermal Expansion o f Chemical-Resistant Mortars, Grouts,
Monolithis Surfacings, and Polymer Concretes. 1995.

96. ASTM C 1 5 7 -9 3 . Standard Test Method for Length Change of Hardened

Hydraulic-Cement Mortar and Concrete. 1997.

97. Ronald G. Creep Tests. Vice President, Construction Technology

Laboratories Inc. U. S. A..2001.

98. Ross A. D. Concrete Creep Data. The Structural Engineer. 15,1937.

99. Lorman W. R. The Theory of Concrete Creep. Proceedings of the ASTM.

40. (pp. 1080-1102)

100 . Pinelle D. J. Curing stresses in polymer modified repair mortars. Cement,

concrete and aggregates. 17, Issue 2, 1995.

101 . Kong and Evans. Reinforced and Prestressed Concrete. Chapman and Hall.
1986.

102 . Nawy E. G. Fundamentals o f high strength high performance concrete.

Longmans. 1996.

103. Chandler T. J. and Gregory S. The climate o f the British Isles. Longmans,
1976.

104. Neville A. M. Creep o f concrete: Plain, Reinforced and Prestressed. 1997.

105. Concrete society technical paper no. 101.The creep of structural concrete.
1973.

106. Mangat P. S. and Azari M. A theory for the free shrinkage of steel fibre
reinforced cement matrices under compression. Journal o f Materials
Science. 19. 1984. (pp. 2183-2194)

107. Evans D. A. The mechanical properties o f polymer modified concrete. Msc

Thesis, Sheffield Hallam University. 1981.

108. Limbachiya M. C. Assessment of long term performance of repaired

reinforced concrete . PhD Thesis, Sheffield Hallam University. 1995.

109. Huffman J. E., Masud A. and Hommertzheim D. ESSEx - an intelligent

advisor for expert system shell selection. Proceedings of the 11th annual

350
 Chapter y - Keterences and Bibliography

conference on computers & industrial engineering 17, Issue 1-4,1989. (pp.

12-17)

110. Mays G.C. (Editor) Durability of concrete structures - Investigation, repair,

protection. Chapter ‘Highway Bridges’ Boam K. J.1992. E&F Spon.

351
 unapter y - Kererences ana uiDiiograpny

9.2 Bibliography

Aberdeen concrete construction. How to fix cracks. The Aberdeen group. Issue 2.1997.

ACI. Repairing Concrete Bridges. 1993.

A.C.I. committee. A CI manual o f concrete inspection. 1998.

Alam M. S. Inspection and maintenance o f bridges.International seminar on bridge

substructure and foundations, Bombay. Conference documentation volume 2 1992.1ndian
Institute of Bridge Engineers.

Alampalli S., Fu G., and Dillon E. W. Signal versus noise in damage detection by
experimental modal analysis. Journal o f Structural Engine ering. 123,1997. (pp. 237-245)

Allen R. T. L., Edwards S. C. and Shaw J. The repair of concrete structures. 1987.

Al-Shawi F. A. N, Mangat P. S, and Halabi W. A simplified design approach to corbels

made with high strength concrete. Materials and structures. 32.10-1-1999. (pp. 579-583)

American Concrete Institute. Concrete repair and restoration. ACI compilation No. 5,
Detroit 1980. 9, Issue 9.

American segmental bridge institute Inspection Manual. A guide to the construction and
inspection o f segmental bridges. 1998.

Arockiasamy M., Sawka M., Sinha V., and El Shahawy M. Knowledge based expert
system approach to analysis and rating of Florida bridges. 1993.

Aston C. and Kelly G. Remedial strengthening work to Chappell Drive bridge, Doncaster
Construction Repair. May -1996. (pp. 10-13)

Austin S. Repair with sprayed concrete. Concrete. 31. Issue 1. Jan 1997. (pp. 18-27)

Baker H. J., Burgoyne C. J., and Dowling P. J. New strategies for bridge assessment.
Proceedings of seminar of assessment of reinforced and pre-stressed concrete bridges held
on 28th Sept. The Institution o f Structural Engineers, London. 1988. (pp. 43-45)

Ballim Y. The effect of shale in quartzite aggregate on the creep and shrinkage of concrete
- A comparison with RILEM model B3. Materials and structures.33. May 2000. (pp. 235-
242)

Bazant Z. P. and Baweja S. Justification and refinements of model B3 for concrete creep
And shrinkage 2.Updating and theoretical basis. Materials and structures. 2%. 1995. (pp.
488-495)

Barksdale G. G. Bridge inspections in high current. Sea Technology. 37.1996. (pp. 15-18)

352
 unapter y - Keterences and tsibiiograpny

Barnard C. P. J. Highway bridge maintenance. Municipal Engineer. April, 1990. (pp. 97-
107)

Bates C. C and Lewsley C. S. Environmental changes, temperature, creep and shrinkage in

concrete structures. 1969.

Beeby A. W. Cracking, Cover, and corrosion of reinforcement. Concrete Int. Design. &
Construction.5. Issue 2.1983. (pp. 35-40)

Bergstrom S. G. Repair of concrete structures, a survey. Beton, 1987.

Bessant G. London underground: new inspection manual and bridge marking system. 1993.
(pp. 49-55)

Bubshait A. A. and Tahir B. M. Evaluation and rehabilitation of concrete pier. Journal o f

performance o f constructedfacilities. October 1997. (pp. 133-118)

Butcher P. Coventry ring road bridge. Concrete (London).31.\991.

Cabrera J. G. and AlHasan A. S. Performance properties of concrete repair materials.

Construction and Building Materials. 11. Nos 5-6. 1997. (pp. 283-290)

Cady P. D. and Weyers R. E. Deterioration rates of concrete bridge decks. Journal o f

transportation engineering. January, 1984. (pp. 34-44)

Canadian Strategic Highway Research Program.Concrete Durability.Technical Brief

#5.1995.

Canadian Strategic Highway Research Program.Electrochemical Chloride Extraction from

Concrete Bridge Components.Technical Brief #2.1995.

Canadian Strategic Highway Research Program. Concrete Bridge Component Evaluation

Manual. Technical Brief #7.1995.

Castillo E. and Alvarez E. Uncertainty and Learning in expert systems. 1998.

Celik T., Thorpe A. and McCaffer R. Development of an expert system. Concrete

Int.Design. & Construction.il. Issue 8. August, 1989. (pp. 37-41)

Cesare M. A., Santamarina C., Turkstra C. and Vanmarcke E. H. Modelling bridge

deterioration with Markov chains. Journal o f transportation engineering. Nov/Dec, 1992.
(pp. 820-833)

Chen L. H. An extended rule-based inference for general decision making problems. 1998.

Chimay J. A. and Bowron, J. A system for the institution of effective repairs to concrete
structures. Computing In Civil Engineering. 1998.

Cleland D. Correct material fits the patch. Highways. June 1994.

353
 Chapter y - Keterences and Bibliography

Cosyn P. A mathematical programming model for computer aided railway bridge repair,
maintenance and replacement. 1993. Bridge management 2. Inspection, maintenance
assessment and repair. Papers presented at the second international conference on bridge
management, held at the university o f Surrey. Guilford, 1993. (pp. 806-814)

Czamecki L. and Glodkowska W. Approach to the compatibility of polymer composite-

cement concrete (PC-CC) system . DURACOSYS 95 International Conference.1996.

Dawe P. H. The assessment of bridges: Department of transport requirements. 1993.

Bridge management 2. Inspection, maintenance assessment and repair. Papers presented at
the second international conference on bridge management, held at the university o f
Surrey.Guilford, 1993. (pp. 1-6)

deBrito J. and Branco F. A. Bridge management policy using cost analysis. Proceedings o f
the institution o f civil engineers. 104. Nov 1994. (pp. 431-439)

deBrito J., Branco F. A. and Ibanez M. Knowledge-based concrete bridge inspection

system. Concrete International. 16,1994.

deBrito J., Branco F. A., ThoftChristensen P. and Sorensen J. D. Expert system for
concrete bridge management. Engineering Structures. 19. 997. (pp 519-526)

deBrito J. and Branco F. A. Concrete bridge management: From design to maintenance.

Practice Periodical on Structural Design and Construction. .3,1998. (pp 68-74)

De rincon O. T. and Decarruyo A. R. The importance of diagnosing before repairing

reinforced concrete structures. Materials selection and design. 1998. (pp. 69-76)

Dutton D. A review of knowledge based systems in the construction industry. International

Journal o f construction information technology. 5. Issue 1.1997. (pp. 63-87)

Ellis H., Jiang M., and Corotis R. B. Inspection, maintenance, and repair with partial
observability. Journal o f Infrastructure Systems. 1. June 1995. (pp. 92-99)

El Sakhawy N. R., El Dien H. S., Ahmed M. E. and Bendary K. A. Influence of curing on

durability performance of concrete. Magazine o f Concrete Research (UK) 51. Issue
5.1999. (pp. 309-318)

Elzarka H. M. and Bell L. C. Pen based computer data acquisition for bridge inspection.
Federal Highways Administration. FHWA-SC-9501 11-1-1995

Elzarka H. M., Bell L. C. and Floyd R. L. Applications of pen based computing in bridge
inspection. Computing in Civil Engineering (New York). 1997. (pp. 327-334)

Emmons P., Vaysburd A. M., and Thomas,J. Strengthening concrete structures, Part I
Concrete International. 20.1998.

Emmons P. H., Vaysburd A. M., and McDonald J. E. Concrete repair in the future turn of
the century - any problems? Concrete International. March 1994. (pp. 42-49)

354
 unapter y - Kererences ana mmiograpny

Engelund S., Sorensen J. D. and Krenk S. Estimation of the time to initiation of corrosion
in uncracked concrete structures. ICASP7.1995.

Engelund S. and Sorensen J. D. A probabilistic model for chloride-ingress and initiation of

corrosion in reinforced concrete structures. Structural Safety.\99%.

Estes A. C., Frangopol D. M. and Hearn G. Reliability-based optimum bridge repair

Strategy. Probabilistic Mechanics and Structural and Geotechnical Reliability. Proceedings
o f the Specialty Conference 2002..

Fensel D., Van Harmelen F., eif W. and en Teije. Formal support for development of
knowledge based systems. Information technology management. 1998.

Fenves S. J.What is an expert system? Expert systems in civil engineering (American

society of civil engineering) 1986. (pp. 1-6)

Frearson J. P. H. Concrete Bridges investigation, maintenance and repair. The Concrete

Society. 1997.

Furuta H. and Hirokane M. Rough-set-based knowledge acquisition from cases for

integrity assessment of bridge structures. Computer aided civil and infrastructure
engineering. July 1998. (pp. 265-273)

Furuta H. and Shiraishi N. An expert system for damage assessment of bridge structures
using fuzzy production rules. 1998.

Gardner N. J and Lockman M. J. Design provisions for drying shrinkage and creep of
normal-strength concrete. AC I materials journal. March-April 2001. (pp. 159-167)

Greenwell M. Knowledge engineering for expert systems. 1988.

Greve H. G. Restoring strength to damaged or deteriorated structural concrete Concrete

Construction. Nov, 1987 . (pp. 949-957)

Grippo G. J. Bridge repair on the fast track. Civil Engineering (New York) 65. (pp. 64-66)

Grivas D. A. and Yung-Ching Shen. Design of a database system for bridge management.
Computing In Civil Engineering.!. 1998. (pp. 899-906)

Gu P., Beaudoin J. J., Tumidajski P. J. and Mailvaganam N. P. Electrochemical

incompatibility of patches in reinforced concrete. Concrete International.\9. 1997. (pp. 68-
72)

Hammad A. and Itoh Y. Knowledge acquisition for bridge design expert systems.
Microcomputers in civil engineering. 8, 1993. (pp. 211-224)

Hartmann D. Knowledge based systems in civil engineering (from CAD to KAD)1998.

(pp. 249-259)

355
 unapter y - Keierences ana tsionograpny

Hassan K. E., Robery P. C. and A1 Alawi L. Effect of hot-dry curing environment on the
intrinsic properties of repair materials. Cement & Concrete Composites (UK)22.Issue
6.2000. (pp. 453-458)

Head P. The performance of bridge systems: the next frontier for design and assessment.
Structural engineering. 17. Issue 3 . 10-1-1991. (pp. 310-316)

Hearn G. and Shim Hs. Integration of nondestructive evaluation methods and bridge
management systems. Proceedings o f the Specialty Conference on Infrastructure
Condition Assessment:Art, Science, Practice. 1997.

Hewlett P. C. and Hurley S. A. Repair materials selection. Proceedings o f the second

international conference on structural faults and repair. Institution o f Civil Engineers.
Westminster, London, 1985.

Hewlett P. C. Assessment and evaluation of polymer based repair materials .Concrete

International. 15. Issue 3.1-3-1993.

Hickman R. Analysis for knowledge-based systems - a practical guide to the KADS

methodology. Ellis Horwood. 1989.

Hickman R., Killin J., Land L., Mulhall T., Porter D. and Taylor R. M. Analysis for
knowledge-based systems. 1989.

Higgins D. Diagnosing the causes of defects or deterioration in concrete structures.

Current practice sheets. 69.10-1 -1991.

Highways Agency. Highway structure - Inspection and maintenance. HMSO, Department

of Transport 1996.

Hopwood T. I. Determining bridge inspection requirements for fatigue damage using

reliability Probabilistic Mechanics and Structural and Geotechnical Reliability.
Proceedings o f the Specialty Conference. 1996. (pp. 454-457)

IABSE, Structures No. Repair and rehabilitation of bridges - case studies I. International
association fo r bridge and structural engineering1998.

Issa M. A., Tsui S. and Yousif A. Application of knowledge based expert systems for rating
highway bridges. Engineering fracture mechanics. 50 .Issue 5/6.1995. (pp. 923-934)

Jacobs T. L. Optimal Long Term scheduling of bridge deck replacement and rehabilitation.
Journal o f transportation engineering. January 1984. (pp. 312-323)

John D. G., Coote A. T., Treadway K. W. J. and Dawson J. L. The repair of concrete- a
laboratory and exposure site investigation. Corrosion o f reinforcement in concrete
construction.\9&3. (pp. 264-286)

Johnson I. D. and Cuninghame J. R. The performance of expansion joints: a survey of 250

joints on highway bridges. 1993. (pp. 110-116)

356
 unapter y - Kererences ana tsionograpny

Johnstone R and Brodie A. The Scottish Office bridge management procedures.The

Management of Highway Structures. 22-23 June 1998.

Kamada T and Li V. C. The effects of surface preparation on the fracture behaviour of

ECC/Concrete repair system. Cement and concrete composites.22.2000. (pp. 423-431)

Kay E. A. and Regan J. Acceptance and compliance of patch repair systems. 1987.

Kelly J. W. Cracks in concrete. Part 1 & Part 2. Concrete Const. 26 . Issue 9. 1981. (pp.
725-735)

Khan S. I., Ritchie S. G. and Kampe K. Integrated system to develop highway

rehabilitation projects. Journal o f transportation engineering. Jan/Feb 1994 .

King J. C. and Wilson A. If it's still standing, it can be repaired. Concrete Const. July 1998.
(pp. 643-650)

Kovler K. Testing system for determining the mechanical behaviour of early age concrete
under-restrained and free uniaxial shrinkage. Materials and structures.21 .Issue 170.1994.
(pp. 324-330)

Koveler K. Interdependence of Creep and Shrinkage for concrete under tension. Journal o f
Material in Civil Engineering.!. Issue 2 . May, 1995. (pp. 96-101)

Kulkami D. and Paramasivam U. Expert systems in structural engineering. The bridge and
structural engineer. 20, Issue 3. September 1990.

Kumar S, Krishnamoorthy C. S. and Rajagopalan N. A process Model for Knowledge

based concrete bridge design. Engineering applications o f artificial intelligence.8. Issue
4.1998. (pp. 435-447)

Kuo S. S., Davidson T., Fiji L. and Err R.. Innovative technology for computer automated
bridge inspection process. Transportation Research Record. 1347. 1992.

Kuo S. S., Clark D. A. and Kerr R. Complete package for computer-automated bridge
inspection process. Transportation Research Record. 1442.1994. (pp. 111-122)

Lawrie R. A. and Dotson W. F. Bridge evaluations after catastrophes. Concrete

International. March 1998. (pp. 59-61)

Lee J., Liu K.F.R. and Chiang W. A fuzzy petri net based expert system and its aplication
to damage assessment of bridge. IEEE Transactions on systems man and cybernetics part
&29.Issue 3.1997.

Legosz A. and Wysokowski A. General information on a Polish bridge management

system. 1993. Bridge management 2. Inspection, maintenance assessment and repair.
Papers presented at the second international conference on bridge management, held at
the university o f Surrey. Guilford, 1993. (pp. 870-879)

357
 unapter v - Keterences ana Bibliography

Leung A. Perfecting bridge inspecting. Civil Engineering (New York). 66.1996. (pp. 59-61)

Linder R. Origin and remedies for defects in concrete construction. An attempted overview
Sachverstandige. 1983.

Lipkus S. E. BRIDGIT bridge management system software. Characteristics o f bridge

management systems - Transportation Research Circular. 423, National Academy Press,
Washington DC. 1994. (pp. 43-54)

Lorman W. R. The Theory of Concrete Creep. 40. 1997.

Mangat P. S. and Elgarf M. S. Flexural strength of concrete beams with corroding

reinforcements. ACI Journal. 96.1ssue 1.1-1-1999. (pp. 149-158)

Mangat P. S. and Elgarf M. S. Bond characteristics of corroding reinforcement in concrete

beams. Materials and structures. 32 . March 1999. (pp. 89-97)

Mangat P. S. and Molloy B. T. Prediction of free chloride concentration in concrete using

routine inspection data. Magazine o f concrete research. 46. Issue 169.1994. (pp. 279-287)

Marusin S. L. Failure of latex modified shotcrete. Concrete International. October, 1990.

(pp. 39-42)

Matsuho S. and Frangopol D. M. Optimization using information integration method and

its application to civil infrastructure systems .Proceedings o f the International Workshop
on Optimal Performance o f Civil Infrastructure Systems. 1997.

McCurrich L. H., Keeley C., Cheriton L. W., and Turner K. J.Mortar repair systems-
corrosion protection for damaged reinforced concrete. Corrosion o f reinforcement in
concrete construction. 1983. (pp. 235-253)

McGovern M. A new weapon against corrosion. Concrete repair digest. June/July 1994

Mola F. Operational techniques for rehabilitation of the edges of reinforced concrete

Bridge beams. 1993. Bridge management 2. Inspection, maintenance assessment and repair.
Papers presented at the second international conference on bridge management, held
the university o f Surrey. Guilford, 1993. (pp. 738-746)

Moore C. J., Lehane M. S. and Price C. J. Case based reasoning fro decision support in
engineering design. IEEE colloquium. 57. 1998.

Morgan D. R. High early strength blended-cement wet-mix shotcrete. Concrete

International. May, 1991. (pp. 35-40)

Morgan D. R. Freeze-thaw durability of shotcrete. Concrete International. August, 1989.

(pp. 86-93)
Niczyj J. Application of fuzzy sets in structural reliability. 1996.

Nishibayashi S. Alkali Aggregate Reaction now .Concrete Journal.\990.

358
 unapter y - Keterences and bibliograpny

Nmai C. K., Farrington S. A. and Bobrowski G. S. Organic based corrosion inhibiting

admixture for reinforced concrete. Concrete International April 1992. (pp. 45-51)

Novoshchenov V. Stopping the cracks. Civil Engineering. Novemeber, 1988. (pp. 54-56)

Ojdrovic R. P. and Zarghamee M. S. Concrete creep and shrinkage: prediction from short
terms tests. ACI Journal. 4-1-1996. (pp. 169-177)

Owen M. Scope for inspection. Bridge engineering. Surveyor. 182.1ssue 53.1995.

Page C.L., Treadway K.W.J., and Bamforth P.B.(Editors) Third international symposium
on corrosion o f reinforcement in concrete construction. 1990.

Paul J. H. Extending the life of concrete repairs. Concrete Internationally. 1998. (pp. 62-
66)

Pierce P. and Mieczkowski J. Windsor bridge pier repairs. Structural design and
construction. May 1996. (pp. 79-81)

Quah T. S. and Teh H. H. Emulating human decision making using a hybrid system .1995.

Ramirez J. L. Ten concrete column repair methods. Construction and Building

Materials. 10. No, 3. 1996. (pp. 195-202)

Raoof M. and Lin Z. Implications of structural damage to concrete elements. Bridge

management 2. Inspection, maintenance assessment and repair. Papers presented at the
second international conference on bridge management, held at the university o f
£'wrrey.Guilford,1993. (pp. 67-74)

Reich Y., Fenves S. J. and Subrahmanian E. Flexible extraction of practical knowledge

from bridge databases. Computing In Civil Engineering. 1. 1998. (pp. 1014-1019)

Reinhardt H. W. and Swamy R. N. Expert system for repair of concrete structures. 1998.

Reynolds C. E. and Steedman J. C. Reinforced Concrete Designers Handbook.\9%%.

Robery P. and Shaw J. Materials for the repair and protection of concrete. Construction
and Building Materials. 11. Issue 5-6.1997. (pp. 275-281)

Romack G. P. Bridge management systems. Structures Congress - Proceedings. 1 1995.

(pp. 680-690)

Ross A. D. Concrete Creep Data. The Structural Engineer. 15, 1937.

Ryall M. J., Parke G. A. R. and Harding J. E. Bridge Management Four. Inspection,

maintenance, assessment and repair.2000.

Sawada E. Repair methods for salt-damaged reinforced concrete structures.Concrete

Int.Design & Construction. March 1990. (pp. 37-41)

359
 Chapter y - References and Bibliography

Schrader E. K. Mistakes, Misconceptions, and controversial issues concerning concrete

and concrete repairs. Parts 1-3. Concrete International. Sept. 1992. (pp. 52-57)

Schreiber B., Wielinga B and Breuker J. KADS - a principled approach to knowledge-

based system development.Issue 11.1993.

Scott D., and Anumba C.J., A Knowledge Based system for the engineering management
of subsidence cases. The Structural Engineer.11 .Issue 3.1999.

Sen R., Liby L., Spillett K., and Shahawy M.. Bridge Rehabilitation Using Advanced
Composites Moving Forward With 50 Years of Leadership in Advanced Materials.
1994.

Seren K. J. Expert systems in concrete material technology. 1990.

tfi
Seren K. J. Expert systems for the concrete industry. 13 Congress, Helsinki 6-10-June
1988. (pp 261-266)

Shaw M. R. Expert systems and the construction industry. BRE. information

paper.IP. 4/89.1989.

Shiraishi N. and Furuta H. Effect of maintenance on structural reliability. Computers &

Structures.30. No. 3. 1988. (pp. 671-677)

Shirole A. M. Bridge Management to the year 2000 and beyond. Characteristics o f bridge
management systems - Transportation Research Circular. 423. National Academy press,
Washington DC. 1998. (pp. 150-153)

Shroff A. and Nathwani S. Effective bridge maintenance using multimedia and mobile
systems. Proceedings o f the 6th International conference on structural faults and repair.
1995. (pp. 25-29)

Smets H. M. G., Bogaerts W. F. L., de Croylan W. and Vancoille M. J. S.Materials

selection and corrosion control. 1998.

Sommer A. M., Nowak A. S., and Thoft-Christensen P. Probability based bridge inspection
system. Journal o f Structural Engineering. November, 1993. (pp. 3520-3550)

Straninger W. and Wicke M. New bridge inventory system in Austria. Bridge management
2. Inspection, maintenance assessment and repair. Papers presented at the second
international conference on bridge management, held at the University o f
Surrey.Guilford, 1993. (pp. 859-869)

Stylianou A., Madey G. and Smith R. Selection criteria for expert system shells.
Communications o f the ACM. 35. Issue 10.1992. (pp. 32-48)

Tanka S. and Mikami I. Reasoning mechanism in diagnostic knowledge based expert

system. Developments in artificial intelligence fo r civil and structural engineering. 1995.

360
 Chapter y - Keterences ana Bibliography

Tao Z. and Stearman B. J. Reliability concept and application in bridge management

systems. 1999.

Tansley D. S. W. and Hayball C. C. Knowledge based systems analysis and design. 1993.

Thoday N. J., Jones D., and Simpson A. Investigation, Repair and strengthening of
Wardley Hall Bridge. 1993.

Thoft-Christensen P. Assessment and performance of optimal strategies for inspection and

maintenance of concrete structures using reliability based expert systems. Vision Eureka,
Lillehammer '94, new technology fo r management and maintenance o f old and new built
environment. 1994.

Thompson P. D. PONTIS: the maturing of bridge management systems in the USA.1993.

Bridge management 2. Inspection, maintenance assessment and repair. Papers presented at
the second international conference on bridge management, held at the University o f
Surrey.Guilford,1993. (pp. 971-978)

Treadway K. W. J. The properties of materials for concrete repair - A review. Issue

proceedings o f the second international conference on deterioration and repair o f
reinforced concrete in the Arabian gulf\9% l.

Tuna M. A., Can H. and Atimatay E. Damage control and reparability of bridge piers. 1993.
Bridge management 2. Inspection, maintenance assessment and repair. Papers presented at
the second international conference on bridge management, held at the University o f
Surrey.Guilford,1993. (pp. 211-216)

Turban E. Expert Systems and applied artificial intelligence. Macmillan. 1992.

Van Gemert D., Czamcki L., and Bares R. Basis for Selection of PC and PCC for Concrete
Repair. Int.J.Cem.Compos.Lightweight Concr.10. Issue 2.1988. (pp. 121-123)

Van Harmelen F. and Balder J. A formal language for KADS models of expertise.
Knowledge acquisition journal A. Issue 1.1992. (pp. 127-161)

Van Harmelen F. and Fensel D. Formal methods in knowledge engineering. The

knowledge engineering review. 10. Issue 4.1995. (pp. 345-360)

Van Harmelen F., Schreiber G. and Wielinga B. Construction of problems-solving

methods are parametric design. International journal o f human computer studies A 9 .\ss,uq
4.1998.

Vassie P. R. Corrosion of reinforcement: an assessment of twelve concrete bridges after 50

years service. 1997.

Vassie P. R. An assessment of 12 concrete bridges after fifty years service.Transport and

road research laboratory. 78.1999.

361
 Chapter 9 - References and Bibliography

Vaysburd A. M., Emmons P. H. and Sabnis G. M. Minimizing cracking - the key to

durable concrete repairs. Indian Concrete Journal.l\.\991. (pp. 652-657)

Vaysburd A. M. Some durability considerations for evaluating and repairing concrete

structures. Concrete International March, 1993. (pp. 29-35)

Wagh V. P. Bridge-Beam repair. Concrete International.October, 1986. (pp. 43-50)

Wallbank E. J. The performance o f concrete in bridges. HMS0.1989.

Watkins R. A. M. and Pitt Jones. Carbonation: a durability model related to site data.
Proceedings o f the institution o f civil engineers 99.1993. (pp. 155-166)

Whisler D. E. Bridge inspection in Kansas. Proceedings o f SPIE - The International

Society fo r Optical Engineering.2456.1995. (pp. 144-149)

Williams J. Concrete durability and repair, pt.2. Construction (London)\9S5.

Wilson R. E. Standard practice for the inspection of concrete. Concrete international

design and construction.4. Issue 9. September, 1982. (pp. 55-57)

Yao-Nan Wang and Tiao-Sheng Tong. Knowledge acquisition of fuzzy expert system
using neural networks. Advances in modelling analysis, B.32. Issue 1.1998.

Ying C. L. A. Bridge assessment and evaluation. 1993. Bridge management 2. Inspection,

maintenance assessment and repair. Papers presented at the second international
conference on bridge management, held at the University o f Surrey.Guilford, 1993. (pp.
511-518)

Yuan Y. and Marosszeky M. Restrained shrinkage in repaired reinforced concrete

elements. Materials and structures.21 .Issue 171.1994. (pp. 375-381)

Zhang Nengsheng Lee, B. H., Goh Teck Huat, Lim Chee Hing, and Pang Chee
Meng. Embedding knowledge-based systems into C++ applications. IEEE. 1998.

Zongwei Tao and Stearman B. J.Reliability concept and application in bridge management
systems. 1998.

Preventing further corrosion in repaired concrete. Article. Concrete Construction. June,

1989. (pp. 558-559)

Repair work on concrete. Article. Beton. October, 1987.

362
 Chapter 9 - References and Bibliography

9.3 Publications

Green L.F. and Mangat P.S. An expert system for repair of reinforced concrete structures.
Concrete Communication Conference, University o f Birmingham, June 2000. (pp. 237-
244)

363