Characterization and Utilization of Cement Kiln Dusts (CKDs) as Partial Replacements of Portland Cement

by

Om Shervan Khanna

A thesis submitted in conformity with the requirements

for the degree of Doctor of Philosophy Department of Civil Engineering University of Toronto

© Copyright by Om Shervan Khanna (2009)

Characterization and Utilization of Cement Kiln Dusts (CKDs) as Partial Replacements of Portland Cement Doctor of Philosophy, 2009 Om Shervan Khanna Department of Civil Engineering University of Toronto Abstract

The characteristics of cement kiln dusts (CKDs) and their effects as partial replacement of Portland Cement (PC) were studied in this research program. The cement industry is currently under pressure to reduce greenhouse gas (GHG) emissions and solid byproducts in the form of CKDs. The use of CKDs in concrete has the potential to substantially reduce the environmental impact of their disposal and create significant cost and energy savings to the cement industry. Studies have shown that CKDs can be used as a partial substitute of PC in a range of 5 – 15%, by mass. Although the use of CKDs is promising, there is very little understanding of their effects in CKD-PC blends. Previous studies provide variable and often conflicting results. The reasons for the inconsistent results are not obvious due to a lack of material characterization data. The characteristics of a CKD must be well-defined in order to understand its potential impact in concrete.

The materials used in this study were two different types of PC (normal and moderate sulfate resistant) and seven CKDs. The CKDs used in this study were selected to provide a representation of those available in North America from the three major types of cement manufacturing processes: wet, long-dry, and preheater/precalciner. The CKDs have a wide range of chemical and physical composition based on different raw material sources and technologies. Two fillers (limestone powder and quartz powder) were also used to compare their effects to that of CKDs at an equivalent replacement of PC.

ii

The first objective of this study was to conduct a comprehensive composition analysis of CKDs and compare their characteristics to PC. CKDs are unique materials that must be analyzed differently from PC for accurate chemical and physical analysis. The present study identifies the chemical and physical analytical methods that should be used for CKDs. The study also introduced a method to quantify the relative abundance of the different mineralogical phases within CKDs. It was found that CKDs can contain significant amounts of amorphous material (>30%) and clinker compounds (>20%) and small amounts of slag and/or flyash (20%) which could also be used as partial replacement of limestone filler in PC.

iii

Acknowledgments I wish to express my sincere gratitude to Professor Doug Hooton, my supervisor, for his excellent guidance and invaluable advice during this investigation. Thanks are also due to Professor Hooton for the facilities extended to me at University of Toronto.

This project was initiated by the Products and Quality Department of Lafarge North America Centre for Technical Services in Montreal under the guidance of Mr. Bruce Blair, Dr. Anik Delagrave, and Ms. Claude Lauzon. Acknowledgement is made of their generous assistance and valuable cooperation throughout the research program. I wish to extend my deepest thanks and appreciation to Dr. Laurent Barcelo of Lafarge North America for his helpful suggestions and interest in this investigation. Dr. Barcelo also reviewed this manuscript during the preparation of the dissertation. Thanks are also extended to these Lafarge personnel who shared their technical knowledge and expertise via personal meetings and/or email exchanges: Dr. Jean Philippe Perez, Dr. Ellis Gartner, and Mr. Paul Lehoux. Thanks are due to Lafarge laboratory technicians Rino Lisella and Patricia Martin who shared their expertise and helped in preparing many of the specimens. Thanks also to the many other laboratory researchers and staff of Lafarge for providing assistance: Mr. Denis Belanger, Mr. Denis Leblanc, Ms. Sona Babikan, Mr. Claude Verville, Ms. Julie Morissette, Ms. Lorraine Phang, and Mr. Bernard Brochard.

Thanks are also due to my colleagues at the University of Toronto. I wish to extend my gratitude to Dr. Terry Ramlochan for providing technical advice periodically throughout the course of this project, particularly with the CKD phase quantification. I would also like to acknowledge the help provided by other members of the Concrete Materials Group: Dr. Gustavo Julio-Betancourt for his helpful discussions, and Ms. Ursula Nytko and Ms. Olga Perebatova for providing guidance and help with the materials and equipment. Thanks are also due to Dr. S. Petrov from the Department of Chemistry for assisting with the CKD phase quantification and Mr. Dan Mathers from the Department of Chemistry for his help in analyzing some of the solution samples.

iv

Sincere thanks to my supervising committee members Dr. Brenda McCabe and Dr. Murray Grabinsky for their insightful questions and comments while the thesis was in progress. I would also like to thank Dr. Daman Panesar for very supportive discussions during my Ph.D and reviewing this dissertation.

Further, the author is indebted to Lafarge North America, the Natural Sciences and Engineering Research Council (NSERC), and the Ontario Graduate Scholarship (OGS) Program for providing financial support throughout the project.

My thanks are also due to my wife, children, and family members for their support and understanding throughout the course of this project.

v

Table of Contents 1.0

INTRODUCTION .................................................................................................. 1 1.1

Background ......................................................................................................... 1

1.2

Problem Statement .............................................................................................. 2

1.3

Incentives and Objectives of This Study ............................................................ 4

1.4

Summary of Chapters ......................................................................................... 6

2.0

LITERATURE REVIEW ....................................................................................... 8 2.1

2.2

2.3

2.4

CKD Manufacture and Management .................................................................. 8 2.1.1

Portland Cement Manufacture Overview ................................................... 8

2.1.2

CKD Generation ....................................................................................... 17

2.1.3

Fresh and Landfill CKD............................................................................ 20

2.1.4

CKD Applications: Cement Industry Perspective .................................... 21

2.1.5

Costs Associated with CKD Disposal....................................................... 22

2.1.6

CKD Environmental Considerations ........................................................ 23

CKD and Portland Cement ............................................................................... 24 2.2.1

Chemical Properties .................................................................................. 24

2.2.2

Mineralogical Properties........................................................................... 26

2.2.3

Physical Properties.................................................................................... 29

2.2.4

CKD Types ............................................................................................... 32

2.2.5

Variability of CKD from a Single Plant ................................................... 34

Portland Cement Hydration .............................................................................. 35 2.3.1

Initial Hydrolysis ...................................................................................... 37

2.3.2

Induction ................................................................................................... 38

2.3.3

Acceleration .............................................................................................. 39

2.3.4

Deceleration .............................................................................................. 40

2.3.5

Slow Continued Reaction ......................................................................... 41

Effects of CKD Properties and PC Dilution ..................................................... 41 2.4.1

Calcium Carbonate.................................................................................... 41

2.4.2

Quartz........................................................................................................ 44

2.4.3

Clays ......................................................................................................... 44 vi

2.4.4

Free Lime and Calcium Hydroxide........................................................... 45

2.4.5

Magnesia ................................................................................................... 48

2.4.6

Sulfate ....................................................................................................... 49

2.4.7

Chloride..................................................................................................... 54

2.4.8

Alkalis ....................................................................................................... 57

2.4.9

Clinker Phases........................................................................................... 58

2.4.10 Physical Properties.................................................................................... 58 2.5

CKD-PC............................................................................................................ 60 2.5.1

CKD-PC Material Characterization.......................................................... 62

2.5.1.1

CKD-PC Chemical Composition...................................................... 62

2.5.1.2

CKD-PC Mineralogical Composition............................................... 66

2.5.1.3

CKD-PC Physical Composition........................................................ 68

2.5.2

Workability ............................................................................................... 69

2.5.3

Setting Time.............................................................................................. 77

2.5.4

Hydration Kinetics .................................................................................... 82

2.5.5

Compressive Strength ............................................................................... 87

2.5.6

Flexural and Tensile Strength ................................................................. 103

2.5.7

Volume Stability ..................................................................................... 106

2.5.7.1

Soundness ....................................................................................... 106

2.5.7.2

Drying Shrinkage ............................................................................ 108

2.5.7.3

Volume Stability Summary............................................................. 112

2.5.8

Durability ................................................................................................ 113

2.5.8.1

Alkali-Aggregate Reaction ............................................................. 113

2.5.8.2

Steel Corrosion................................................................................ 116

2.5.8.3

Permeability .................................................................................... 121

2.5.8.4

Freezing and Thawing..................................................................... 123

2.5.8.5

External Sulfate Resistance............................................................. 125

2.5.8.6

Durability Summary........................................................................ 126

vii

3.0

MATERIALS AND EXPERIMENTAL DETAILS........................................... 127 3.1

Materials ......................................................................................................... 127

3.2

Testing of Raw Materials................................................................................ 129

3.3

4.0

3.2.1

Chemical Properties ................................................................................ 129

3.2.2

Mineralogical Properties......................................................................... 129

3.2.3

Physical Properties.................................................................................. 130

3.2.4

Dilute Stirred Suspensions...................................................................... 130

CKD-PC Blends.............................................................................................. 131 3.3.1

Heat of Hydration ................................................................................... 131

3.3.2

Normal Consistency................................................................................ 133

3.3.3

Initial Setting Time ................................................................................. 133

3.3.4

Flow ........................................................................................................ 134

3.3.5

Compressive Strength ............................................................................. 134

3.3.6

Expansion in Limewater ......................................................................... 134

3.3.7

Autoclave Expansion .............................................................................. 135

3.3.8

Alkali Silica Reactivity ........................................................................... 135

RESULTS AND DISCUSSION ......................................................................... 138 4.1

4.2

Material Characterization................................................................................ 138 4.1.1

Chemical Properties ................................................................................ 138

4.1.2

Mineralogical Properties......................................................................... 144

4.1.3

Physical Properties.................................................................................. 151

4.1.4

CKD Dissolution Analysis...................................................................... 158

CKD-PC Blends.............................................................................................. 162 4.2.1

Kinetics ................................................................................................... 167

4.2.1.1 4.2.2

Heat of Hydration ........................................................................... 167

Physical Properties of Hydration ............................................................ 195

4.2.2.1

Normal Consistency........................................................................ 195

4.2.2.2

Flow ................................................................................................ 202

4.2.2.3

Initial Setting Time ......................................................................... 210

4.2.2.4

Compressive Strength ..................................................................... 218

4.2.3

Volume Stability and Durability............................................................. 230 viii

5.0

4.2.3.1

Expansion in Limewater ................................................................. 230

4.2.3.2

Autoclave Expansion ...................................................................... 236

4.2.3.3

Alkali Silica Reactivity ................................................................... 242

MAIN CONTRIBUTIONS OF THE THESIS ................................................... 246 5.1

CKD Characterization..................................................................................... 246

5.2

CKD-PC Blends.............................................................................................. 248

6.0

CONCLUSIONS................................................................................................. 252

7.0

RECOMMENDATIONS FOR FUTURE WORK ............................................. 254

8.0

REFERENCES ................................................................................................... 257

ix

List of Tables

Table 2.1

Kiln material transformations (Manias, 2004)

12

Table 2.2

Summary of operation data on different kiln systems (Manias, 2004)

13

Melting points and relative volatiles of different compounds in the kiln burning zone (Manias, 2004)

18

Typical costs associated with CKD disposal, $/tonne (Kessler, 1995)

23

CKD chemical oxide composition, free lime, and loss on ignition, and statistical analysis of 63 published datasets (Sreekrishnavilasam et al., 2006)

25

Portland cement chemical oxide composition, total alkali content, and loss on ignition (Tennis and Bhatty, 2006)

25

Mineralogical composition of U.S. CKD samples (Hawkins et al., 2004)

27

Portland cement average bogue compound and Blaine fineness in 2004 (Tennis and Bhatty, 2006)

29

CKD oxide composition and statistical analysis of intermittent daily samples collected from a single kiln (long-dry process) over a 3 year period (2005 – 2008) in North America (Lafarge, 2009)

34

Table 2.10

Summary of previous CKD-PC studies from literature review

61

Table 2.11

Chemical and physical composition of CKD: from CKD-PC literature review

64

Chemical and physical composition of PC: from CKD-PC literature review

65

Mineralogical composition of CKD: from CKD-PC literature review

67

Table 2.3

Table 2.4

Table 2.5

Table 2.6

Table 2.7

Table 2.8

Table 2.9

Table 2.12

Table 2.13

x

Table 2.14

Workability: from CKD-PC literature review

76

Table 2.15

Setting time: from CKD-PC literature review

81

Table 2.16

Hydration: from CKD-PC literature review

86

Table 2.17

Mortar compressive strength of CKD 1 at 0%, 5%, and 10% replacement of PC 3 as a function of time (Maslehuddin et al., 2008a)

89

Table 2.18

Compressive strength: from CKD-PC literature review

100

Table 2.19

Flexural and tensile strength: from CKD-PC literature review

106

Table 2.20

Soundness: from CKD-PC literature review

108

Table 2.21

Mortar drying shrinkage with 0%, 5%, and 10% CKD 1 replacement of PC 3 (Maslehuddin et al., 2008a)

110

Table 2.22

Drying shrinkage: from CKD-PC literature review

112

Table 2.23

Alkali-aggregate reactivity: from CKD-PC literature review

115

Table 2.24

Concrete resistivity and risk of reinforcement corrosion as specified in COST 509 (Maslehuddin et al., 2008b)

118

Table 2.25

Steel corrosion: from CKD-PC literature review

120

Table 2.26

Chloride permeability of PC 1 and PC 2 with CKD 1 replacement at 0%, 5%, 10%, and 15% (Maslehuddin et al., 2008b)

122

Table 2.27

Permeability: from CKD-PC literature review

123

Table 2.28

Freezing and thawing cycles: from CKD-PC literature review

125

Table 2.29

Sulfate resistance: from CKD-PC literature review

125

Table 3.1

CKD kiln process description

128

Table 4.1

Melting points and volatility of compounds in CKDs (Manias, 2004)

139

xi

Table 4.2

Chemical and select physical components of PC, CKD, and filler materials (mass %)

142

Table 4.3

Cements TI and TII mineralogical composition (mass %)

145

Table 4.4

CKD mineralogical compositions using direct test methods (mass %)

146

Table 4.5

Mineralogical composition of CKD and filler materials (mass %)

148

Table 4.6

Physical properties of all materials

152

Table 4.7

Ionic concentrations of 10:1 water to solid ratio (by mass)

159

Table 4.8

Range for chemical and physical properties of Cement TI blends at 10% and 20% replacement (Theoretical calculation, mass %)

163

Range for chemical and physical properties of Cement TII blends at 10% and 20% replacement (Theoretical calculation, mass %)

164

Iterative process to determine the water requirement for normal consistency of (a) Cement TI and (b) Cement TII

195

Table 4.11

Range of change in water demand for normal consistency of pastes

198

Table 4.12

Range of flow for all mortars

204

Table 4.13

Compressive strength range for CKD-PC blends as percent of PC alone

222

Compressive strength range for PC-filler blends as percent of PC alone

222

Table 4.15

Autoclave expansions for (a) Cement TI and (b) Cement TII

236

Table 4.16

Range of autoclave expansions for all blends

239

Table 4.17

ASR concrete mix alkali loadings and CKD replacement levels for (a) Test Series I: Cement TI CKD blends and (b) Test Series II: Cement TII CKD blends

243

Table 4.9

Table 4.10

Table 4.14

xii

List of Figures

Figure 2.1

Cement manufacturing process (Corish and Coleman, 1995)

Figure 2.2

Clinker reactions in kiln feed as a function of temperature (Manias, 2004)

12

Schematic of (a) a wet and long-dry pyroprocess and (b) a preheater/precalciner pyroprocess (with a single preheater tower) (Manias, 2004)

15

Schematic of electrostatic (Peethamparan, 2002)

19

Figure 2.3

Figure 2.4

precipitator

(ESP)

9

efficiency

Figure 2.5

CKD and PC particle size distribution (Peethamparan et al., 2008)

30

Figure 2.6

CKD and PC particle size distribution from published literature (Sreekrishnavilasam et al., 2006)

31

Heat evolution of PC paste during hydration stages: (1) initial reaction, (2) induction, (3) acceleration, (4) deceleration, and (5) slow continued reaction (Gartner et al., 2002)

36

Relative volumes of the major compounds in the microstructure of hydrating PC pastes as a function of time (Odler, 1998)

36

Effect of firing temperature on the heat evolution of pure free calcium oxide during hydration (Shi et al., 2002)

46

Heat of hydration of cement paste determined by isothermal conduction calorimetry, (20°C and w/c = 0.44); (a) PC (b) PC + 0.5% SO3, (c) PC + 2.5% SO3 (Lawrence, 1998b) Note: Sulfate added as Gypsum (Calcium Sulfate)

51

Optimization of gypsum additions for compressive strength at different ages (Gartner et al., 2002) (Note: this PC required higher SO3 levels than normal to obtain maximum strength)

53

Effect of calcium chloride on heat development in PC (Lerch, 1944)

56

Figure 2.7

Figure 2.8

Figure 2.9

Figure 2.10

Figure 2.11

Figure 2.12

xiii

Figure 2.13

Relationship between water demand and specific surface area of PC (Sprung et al., 1985)

59

Figure 2.14

Particle size distribution of CKD 5 and PC 7 (Wang et al., 2002)

69

Figure 2.15

Paste water/binder ratio, initial set, and final set of CKD 2 as a partial substitute of PC 4 at different levels of replacement (ElAleem et al., 2005)

70

Mortar water/binder ratio of CKD 2 as a partial substitute of PC 4 at different levels of replacement (El-Aleem et al., 2005)

71

Mortar water/binder ratio of CKD 3 as a partial substitute of PC 5 at different levels of replacement (Al-Harthy et al., 2003)

72

Hydration of pastes showing (a) evaporable water content (%), (b) free lime content (%) (calcium oxide and calcium hydroxide), and (c) chemically combined water content, as a function of time at different percentage levels of PC 4 replacement with CKD 2 (ElAleem et al., 2005)

83

Concrete compressive strength of CKD 1 at different replacement levels of (a) PC 1 and (b) PC 2 (Maslehuddin et al., 2008b)

88

Mortar compressive strength as a function of time at different percentage levels of CKD 2 replacement of PC 4 (El-Aleem et al., 2005)

90

Concrete compressive strengths, w/b (a) 0.70, (b) 0.60, and (c) 0.50, at different percentage levels of CKD 3 replacement of PC 5 (Al-Harthy et al., 2003)

92

Concrete drying shrinkage as a function of time at different replacement levels of PC 1 with CKD 1 (Maslehuddin et al., 2008b)

109

Concrete drying shrinkage as a function of time at two different w/b ratios with 5% CKD 12 replacement of PC 12 (Wang and Ramakrishnan, 1990)

111

Concrete specimen variation of electrical resistivity with moisture content at different percentage levels of CKD 1 replacement of (a) PC 1 and (b) PC 2 (Maslehuddin et al., 2008b)

117

Figure 2.16

Figure 2.17

Figure 2.18

Figure 2.19

Figure 2.20

Figure 2.21

Figure 2.22

Figure 2.23

Figure 2.24

xiv

Figure 4.1

Process flow chart for CKD chemical composition analysis

140

Figure 4.2

Particle size distribution of PC, CKD and filler. The materials are in the direction and position of the arrow: LS, D, SLX, TII, TI, A, F, C, E, B, D*

155

Particle size distribution of PC, CKD and filler between 0.1 µm and 10 µm. The materials are in the direction of the arrow: LS, SLX, C, D, B, A, F, TII, TI, E, D*

155

CKD fineness correlation between (a) Blaine fineness and particle size distribution, and (b) percentage passing 45µm sieve and particle size distribution

156

Composition of pore solution w/b 0.5 high alkali PC paste (Gartner et al., 2002)

158

Schematic of isothermal conduction calorimetry curve heat liberation characterization

168

Cumulative heat of hydration during initial hydrolysis (Ai) of (a) Cement TI blends and (b) Cement TII blends (w/b = 0.4, 23°C)

170

Cumulative heat of hydration during initial hydrolysis (Ai) as a function of Free CaO (%) for (a) Cement TI CKD blends and (b) Cement TII CKD blends (w/b = 0.4, 23°C)

171

Cumulative heat of hydration during initial hydrolysis (Ai) as a function of (a) sulfate content for Cement TI CKD blends and (b) alkali content for Cement TII CKD blends (w/b = 0.4, 23°C)

173

Minimum heat of hydration rate during induction period (Qi) of (a) Cement TI blends and (b) Cement TII blends (w/b = 0.4, 23°C)

175

Minimum heat of hydration rate during induction period (Qi) as a function of sulfate content for (a) Cement TI CKD blends and (b) Cement TII CKD blends (w/b = 0.4, 23°C)

176

Minimum heat of hydration rate during induction period (Qi) as a function of calcium langbeinite content for (a) Cement TI CKD blends and (b) Cement TII CKD blends (w/b = 0.4, 23°C)

177

Figure 4.3

Figure 4.4

Figure 4.5

Figure 4.6

Figure 4.7

Figure 4.8

Figure 4.9

Figure 4.10

Figure 4.11

Figure 4.12

xv

Figure 4.13

Figure 4.14

Figure 4.15

Figure 4.16

Figure 4.17

Figure 4.18

Figure 4.19

Figure 4.20

Figure 4.21

Figure 4.22

Time of minimum heat of hydration rate during the induction period (ti) of (a) Cement TI blends and (b) Cement TII blends (w/b = 0.4, 23°C)

179

Time of minimum heat of hydration rate during the induction period (ti) as a function of total alkali content for (a) Cement TI CKD blends and (b) Cement TII CKD blends (w/b = 0.4, 23°C)

180

Main hydration peak relative to the minimum peak rate heat of hydration during the induction period (Qw-Qi) for (a) Cement TI blends and (b) Cement TII blends (w/b = 0.4, 23°C)

182

Main hydration peak relative to the minimum peak rate heat of hydration during the induction period (Qw-Qi) as a function of calcium langbeinite content for (a) Cement TI CKD blends and (b) Cement TII CKD blends (w/b = 0.4, 23°C)

184

Heat of hydration for Cement TI with (a) CKD A and LS at 10% replacements, (b) CKD C and LS at 10% replacements, (c) 0% and LS at 20% replacements, and (d) CKD B and LS at 20% replacements (w/b = 0.4, 23°C)

186

Heat of hydration for Cement TI with (a) 0% and LS at 10% replacements and (b) CKD E and LS at 20% replacements (w/b = 0.4, 23°C)

188

Heat of hydration for Cement TII with (a) 0% and LS at 10% replacements, (b) CKD C and LS at 10% replacements, and (c) CKD C and LS at 20% replacements (w/b = 0.4, 23°C)

189

The total heat generation from induction period to 7 days hydration (A7d-Ai) for (a) Cement TI blends and (b) Cement TII blends (w/b = 0.4, 23°C)

191

Water requirement for normal consistency of (a) Cement TI blends and (b) Cement TII blends

197

Correlation between Cement TI and Cement TII blends with the same CKD and replacement level for (a) all CKDs and (b) CKDs A, B, C, and D

199

xvi

Figure 4.23

Water requirement for normal consistency as a function of free lime content for (a) Cement TI CKD blends and (b) Cement TII CKD blends

200

Figure 4.24

Mortar flow of (a) Cement TI blends and (b) Cement TII blends

203

Figure 4.25

Mortar flow as a function of free lime content for (a) Cement TI CKD blends and (b) Cement TII CKD blends

206

Mortar flow as a function of (a) percentage of volume less than 30.5 µm for Cement TI CKD blends (excluding CKDs E and F) (b) percentage passing 45 µm for Cement TII blends (excluding CKDs E and F)

207

Figure 4.27

Initial set time for (a) Cement TI blends and (b) Cement TII blends

211

Figure 4.28

Initial set time as a function of the time of minimum heat rate during the induction period (ti) for (a) Cement TI CKD blends (excluding circled data points) and (b) Cement TII CKD blends

214

Initial set time as a function of soluble alkali content for (a) Cement TI blends (excluding circled data points) and (b) Cement TII blends

216

Mortar compressive strength of Cement TI blends at 1, 3, 7, 28, and 90 days (w/b = 0.485) for (a) 10% replacement and (b) 20% replacement (w/b = 0.485)

220

Mortar compressive strength of Cement TI blends at 1, 3, 7, 28, and 90 days (w/b = 0.485) for (a) 10% replacement and (b) 20% replacement (w/b = 0.485)

221

Mortar compressive strength as a function of total sulfate content for Cement TI CKD blends at (a) 1 day and (b) 3 days (w/b = 0.485)

224

Mortar compressive strength as a function of total sulfate content for Cement TII CKD blends at (a) 1 day and (b) 3 days (w/b = 0.485)

226

Mortar compressive strength at 28 days as a function of percentage passing 45 µm for Cement TI CKD blends

228

Figure 4.26

Figure 4.29

Figure 4.30

Figure 4.31

Figure 4.32

Figure 4.33

Figure 4.34

xvii

Figure 4.35

Figure 4.36

Figure 4.37

Figure 4.38

Figure 4.39

Figure 4.40

Mortar compressive strength at 28 days as a function of calcite for Cement TII CKD blends (w/b = 0.485)

228

Expansion in limewater after 14 days for (a) Cement TI blends and (b) Cement TII blends

233

Expansion in limewater at 14 days as a function of sulfate content for (a) Cement TI CKD blends and (b) Cement TII CKD blends

234

Autoclave Expansions for (a) Cement TI blends and (b) Cement TII blends

238

Autoclave expansion as a function of free lime content (excluding data points in the circles) for (a) Cement TI CKD blends and (b) Cement TII CKD blends

240

ASR expansions over 2 years for (a) Test Series I: Cement TI CKD blends and (b) Test Series II: Cement TII CKD blends

245

xviii

List of Appendices

Appendix A. CKD Chemical Composition Correction Calculations Appendix B. PC and CKD TGA Analysis Appendix C. CKD XRD Scans Appendix D. PC, CKD-PC, and PC-Filler Properties Appendix E. Isothermal Conduction Calorimetry Results Appendix F.

Mortar Flow Statistical Analysis

Appendix G. Mortar Compressive Strength Statistical Analysis Appendix H. Mortar Expansion in Limewater Statistical Analysis

xix

List of Notation The following notations are commonly used throughout this thesis:

General AAR

Alkali-Aggregate Reaction

AASHTO

American Association of State and Highway Transportation Officials

ANOVA

Analysis of Variance

ASR

Alkali-Silica Reaction

ASTM

American Society for Testing Materials

BS

British Standard

CKD

Cement Kiln Dust

CSA

Canadian Standards Association

EPA

Environmental Protection Agency (U.S.)

ESP

Electrostatic Precipitators

GHG

Greenhouse Gas

ISAT

Initial Surface Absorption Test

LOI

Loss on Ignition

NCHRP

National Cooperative Highway Research Program

PC

Portland Cement

PCA

Portland Cement Association

PSD

Particle Size Distribution

SCM

Supplementary Cementitious Material

TCLP

Toxicity Characteristic Leaching Procedure

TGA

Thermal Gravimetric Analysis

XRD

X-ray Diffraction

xx

Chemical AFm

Aluminate-Ferrite-Monosubstituted, Monosulphoaluminate, or Monosulphate

AFt

Aluminate-Ferrite-Trisubstituted or Ettringite

C3 S

Tricalcium Silicate or Alite

C2 S

Dicalcium Silicate or Belite

C3 A

Tricalcium Aluminate or Aluminate

C4AF

Tetracalcium Aluminate Ferrite or Ferrite

CH

Calcium Hydroxide

C-S-H

Calcium Silicate Hydrate

Na2Oe

Equivalent Na2O (Na2O + 0.658 K2O) (mass %)

xxi

1.0

INTRODUCTION

1.1

Background

There are currently many challenges to the utilization of by-product cement kiln dusts (CKDs) as partial replacement of Portland cement (PC). CKDs are fine powders (CKDs typically have between 80 and 90% passing a 90 µm sieve) that are generated during the cement manufacturing process, then carried off in the flue gases, and subsequently collected in baghouses or electrostatic precipitators. The portion of CKDs that are not returned back to the cement manufacture process, or otherwise used beneficially, are placed in stockpiles or landfills. A limited number of studies have shown that CKDs removed from the cement manufacturing process could be used as partial replacements of PC in the range of 5 – 15%, by mass. Although standards allow for the use of CKDs at low levels of PC replacement, very little is known about the effects of different CKDs in pastes, mortars, and concrete. The studies that have been published on the use of CKDs as a partial substitute of PC often report conflicting results.

Significant amounts of CKDs are placed in landfills every year. In 2000, the Portland Cement Association (PCA) conducted a United States (U.S.) Cement Industry survey of 92 cement plants. They reported total clinker production to be 68.8 million tonnes (clinker is the major component of PC and is typically 90 – 95% of total cement production). The amount of CKDs removed from the cement kiln process that year for the same 92 cement plants was 2.8 million tonnes (4.1% of clinker production). Almost 80% of the CKDs removed from cement-producing kilns were placed in landfills, while only approximately 20% were beneficially re-used (Hawkins et al., 2004). On a global scale, it is estimated that approximately 30 million tonnes of CKDs are removed from the cement manufacturing process every year (Dyer et al., 1999). Approximately 25 years ago, the CKDs in U.S. landfills were estimated to be greater than 90 million tonnes (Collins and Emery, 1983).

1

There are many applications of CKDs that continue to be investigated: for example, as a component in cements and masonry products, as an agricultural/soil fertilizer, as a soil stabilizer, as a wastewater stabilizer, as a partial replacement of soda in glass production, as an anti-stripping agent in asphalts, and as a subgrade for highway construction (Bhatty, 1995). From the perspective of the cement industry, however, the most desirable application of CKDs that cannot be recycled back into the process is their use as a partial replacement of PC.

1.2

Problem Statement

Four obstacles related to CKD compositions currently inhibit their use in concrete: (i) inadequate CKD characterization, (ii) potentially deleterious interactions between CKD and PC, (iii) unknown interactions of CKD with mineral and/or chemical admixtures, and (iv) CKD-PC conformance to cement and concrete standards. The focus of the thesis is to mainly address the first and second categories. Each category is briefly discussed in this section, however, to provide the reader with a broader understanding of the problem.

In order to understand the effects of CKDs in concrete, it is essential to have a proper characterization of an individual CKD. Comprehensive compositional analysis of a CKD is also important for optimization of a CKD-PC blend for use in concrete field applications. Determining the characteristics of the CKDs used in previous CKD-PC interaction studies was not always possible due to the incomplete compositional analysis provided. This is likely due to the insufficient and sometimes inappropriate application of compositional analysis procedures designed for PC to determine the composition of CKD. CKD is a unique material that has different characteristics from PC. In comparison to PC, CKDs typically contain higher concentrations of free lime, alkalis, sulfates, chlorides, raw materials, and trace heavy metals (Hawkins et al., 2004).

2

CKDs can influence the interactions among the basic components of concrete (PC, water, and aggregate). The effects of the individual components found in CKDs at elevated concentrations in concrete are generally understood. The varying concentrations of these components in combination with each other as found in CKDs, however, are not well understood. Therefore, it is not clear how a particular CKD will interact as a partial replacement of a given PC. The composition of each PC can also have unique characteristics. A given CKD may react differently with dissimilar PCs and, therefore, result in different effects on concrete properties. It is important to understand how the CKD-PC interaction will impact concrete properties such as workability, hydration, setting time, strength, volume stability, and durability for optimization of a mix design in the field.

The impact of a CKD in concrete is not limited to its interaction with PC, aggregate, and water. The use of supplementary cementing materials (SCMs) in concrete has been steadily increasing over the years. The presence of a CKD could influence the mechanisms and effectiveness of SCMs and chemical admixtures in concrete. SCMs such as slag, fly ash, and silica fume contribute to the properties of the hardened concrete through hydraulic and/or pozzolanic action (pozzolanic action occurs when a pozzolan combines with calcium hydroxide to exhibit cementitious properties). It has been reported that the high alkali and sulfate content of a CKD can act as an excellent activator for pozzolanic materials (Konsta-Gdoutos and Shah, 2003).

Chemical admixtures are also commonly used in concrete mixtures. Chemical admixtures can be defined as materials other than water, aggregates, and hydraulic cement that are added immediately before or during mixing of concrete. The most prominent chemical admixtures are used to decrease the quantity of water needed to obtain a given degree of workability or to entrain air in order to increase the resistance of concrete to damage from freezing (Taylor, 1997). Chemical admixtures can also be used to increase workability by dispersion of cement in the aqueous phase of concrete and to accelerate or retard the

3

normal rate of hydration (Dodson, 1990). There is little, if any, published research on the interaction of CKDs and PC with chemical admixtures.

Cement and concrete standards include limitations on the chloride, sulfate, and alkali content in PC and concrete to ensure acceptable performance and durability. If it is shown that the elevated concentrations of these components in CKDs do not compromise performance and durability in concrete, regulatory standards may need to be modified to allow for increased amounts of their replacement of PC. In order to allow for the use of industrial by-products such as CKDs, there is a move away from prescriptive or compositional standards towards performance standards. ASTM C150 allows the use of processing additions meeting the requirements of ASTM C465 for use in the manufacture of hydraulic cements.

1.3

Incentives and Objectives of This Study

The use of CKDs as a partial replacement of cement has the potential to substantially reduce the environmental impact of CKD disposal and create significant cost and energy savings to the cement industry. From an environmental perspective, CKD removal from the cement manufacturing process leads to excessive generation of gas emissions and increased need for land disposal sites. Partial substitution of PC with CKDs would decrease the need for clinker production and reduce the amount of energy wasted due to partial pyropressing of CKDs. A reduction of clinker production would also reduce greenhouse gases that are related to fuel burning and limestone decarbonation. As environmental concerns increase, it is also important to recognize that obtaining landfill permits is becoming increasingly difficult. The use of CKDs as a partial replacement of cement could help minimize the size and number of landfill disposal sites.

In addition to the environmental benefits related to CKD-PC blends, reducing the clinker factor in cement would also create several financial benefits. First, the lifespan of the limestone quarry and other natural resources would increase. Second, the reduction of

4

raw materials required for PC production would reduce material costs and energy consumption related to mining, crushing, and grinding. Third, a reduction of clinker production would reduce pyroprocess, dust collection, and landfill disposal costs. Fourth, since the CKD is already a fine powder, there will be less energy consumption in the finish mill to achieve the target fineness compared to the energy needed to interground clinker. Finally, the typical transport costs for other materials used for blend cements would not be incurred since CKD is generated on the same site as the PC. It is important to acknowledge that the cement and concrete industry may need to incur costs related to building and maintaining systems that allow for blending of CKD with cements that meet quality control targets.

The study of CKD as a partial replacement of PC has been a sporadic research area for the past 30 years. The concrete industry has been very successful in utilizing other industrial by-products – such as slag, fly ash, and silica fume – as partial replacements of PC. Once considered to be waste products, these SCMs are now widely used to improve the workability, strength, and durability characteristics of concrete. Although there are many studies that report the effects of different binary and ternary blends of CKDs with PC, silica fume, fly ash, and/or slag, it is very difficult to make conclusions regarding performance due to conflicting results and incomplete CKD characterizations. The reasons for the different effects of CKD-PC blends have not been thoroughly explored.

The interaction between different CKDs and PC must be well understood before introducing chemical admixtures and other SCMs. Understanding the CKD-PC interactions and developing appropriate limits for specific deleterious components could ultimately allow for the standardization and optimization of blended cements with high replacement levels (5 – 15%, by mass) of PC with CKD in concrete, leading to both environmental and economic benefits.

5

The first objective of this study was to compare the chemical, physical, mineralogical, and rapid ion dissolution properties of different CKDs with PC. Since there is a lack of proper CKD characterization in previous CKD-PC blend research, the present study aims to identify the appropriate chemical and physical analytical methods that should be used for CKD composition analysis. Mineralogical composition analysis is a fine complement to chemical composition analysis since the effects of CKD elements in a CKD-PC blend may vary depending upon the form in which they actually exist. Therefore, a method to quantify the relative abundance of the different mineralogical phases within CKDs was introduced. Since the availability of ions in the liquid phase greatly influences PC hydration, the rapid ion dissolutions from CKDs compared to PC were also investigated.

The second objective was to utilize the material characterization analysis to determine the relationships among the composition properties of CKD-PC blends and their effects on hydration, mechanical properties, and volume stability. Paste and mortar tests were used to assess the effects of CKDs on: (i) heat of hydration, (ii) water demand, (iii) flow, (iv) initial setting time, (v) compressive strength, (vi) expansion in limewater, and (vii) autoclave expansion. Regression analysis was performed where possible to examine the relationships among CKD-PC blend properties and various independent variables. Additionally, concrete prisms were used to evaluate the impact of CKDs on a key durability concern – alkali silica reactivity (ASR).

1.4

Summary of Chapters

The topics addressed in this study are presented in eight chapters. A brief summary of Chapters 2 to 8 is given below.

Chapter 2 is a literature review that provides an understanding of CKD manufacture and management; a basic understanding of CKD composition and its variability relative to PC; a review of PC hydration and the known effects of the individual components of CKDs in pastes, mortars, and concrete; and a review of previous studies on CKD-PC

6

interaction. All performance parameters in previous CKD-PC blend studies are presented in order to provide a general overview of the subject, although some aspects are not part of the current study.

Chapter 3 describes the materials and test methods used in the current study. The CKDs selected for this study are representative of those available in North America. Various test methods related to material characterization and CKD-PC performance used for the experimental program are described.

Chapter 4 presents the results and discussions of the current study. The first part of this chapter explores the material characterization and analytical methods used to determine accurate CKD characterizations. The second part of this chapter focuses on the effects and relationships of using CKDs as a partial replacement of PC on heat of hydration, workability, setting time, compressive strength, expansion in limewater, and soundness. Additionally, it also discusses the impact of CKDs on ASR.

Chapter 5 highlights the main contributions of the thesis, giving a thorough explanation of the value of the research that was conducted. In Chapter 6 the conclusions of the thesis are presented.

Chapter 7 provides recommendations for future research that will enhance the use of CKDs in concrete. This research study is a first step towards a comprehensive understanding of how CKD-PC blends can be optimized.

Chapter 8 provides the list of the references that were consulted in the process of research for the thesis.

7

2.0

LITERATURE REVIEW

2.1

CKD Manufacture and Management

2.1.1

Portland Cement Manufacture Overview

A critical examination of CKD utilization as a partial replacement of PC cannot be conducted without an understanding of basic cement manufacturing. The composition and variability of a CKD produced at a plant is directly related to the cement manufacturing process at that plant. PC is produced by burning ground mixtures of limestone and other materials up to high temperatures (greater than 1450˚C) in a rotary kiln to form clinker (Manias, 2004). The clinker is cooled and then ground in a finish mill along with a small amount of gypsum to make a gray powder called PC. ASTM C219 defines PC as “a hydraulic cement produced by pulverizing Portland-cement clinker, and usually containing calcium sulfate”. Low levels of mineral additives such as limestone, however, are increasingly common in PC. Clinker consists of predominantly crystalline calcium silicates. Hydraulic means that it sets and hardens by chemical interaction with water. Although every plant has significant differences in equipment design and operation, the chemical and physical transformation of raw materials into PC is essentially the same at all cement plants. The basic steps of cement manufacturing are illustrated in Figure 2.1.

The principal raw mix components that are required for the production of clinker are calcium, silica, aluminum, and iron (Taylor, 1997). Calcium carbonate and argillaceous substances (clay) are naturally occurring raw materials that typically contain the principal chemical elements. Limestone, the most common form of calcium carbonate, is the usual calcium source for cement manufacturing. Other forms of calcium carbonate such as chalk, shell deposits, and calcareous muds can also be used. Clays are essentially hydrous aluminum silicates with complete or partial substitution of magnesium and/or iron in

8

place of aluminum in certain minerals. Alkalis or alkaline earths are also present as essential constituents in clays (Chatterjee, 2004). The natural raw materials are traditionally mined at quarries close to the cement plant. At times, auxiliary raw materials that contain iron, alumina, and/or silica are required in order to achieve the proper raw mix proportion. Blast furnace slag, fly ash, iron oxide, bauxite, and spent catalysts are widely used auxiliary raw materials (Bhatty and Gajda, 2004).

Figure 2.1 Cement manufacturing process (Corish and Coleman, 1995)

A finely ground mixture typically consisting of approximately 75% calcium carbonate, 15% silicon dioxide, 3% aluminum oxide, and 2% iron oxide provides the major components in the raw materials. The raw materials also contain a certain amount of volatiles (less then 5% by mass). Some of these volatiles are alkalis (potassium oxide and sodium oxide), sulfur, and chloride (Taylor, 1997). In addition to the major elements

9

which make up cement, smaller concentrations of almost every other element will be present in the raw materials. Magnesium, titanium, manganese, and phosphorous are common but they are minimized to prevent potentially deleterious effects on cement burning and quality. Minor trace metals can also be present in the raw materials but are also kept at low levels to avoid adverse effects (Bhatty, 2004).

In an open quarry, limestone mining operations begin with removal of overburden (waste rock) by bulldozers to expose the top surface of the limestone. Drills are used to create deep holes close to the open face of the limestone quarry for dynamite placement. The limestone rock is then blasted with the dynamite to reduce its maximum diameter to between approximately 1 and 2 metres. Front-end loaders load the blast rock into trucks or railroad cars to be sent to the crushing system. The primary and secondary crushing systems reduce the limestone size to between approximately 10 mm and 50 mm in diameter (Chatterjee, 2004).

The crushed limestone and other raw materials are fed into a grinding mill to obtain the correct size and composition for the raw mix. In the wet process, the raw materials are mixed with approximately 30 – 40% water during grinding to form a slurry. The composition of earth minerals in limestone and clays can be quite variable and may require substantial blending and analysis to maintain a homogenous mixture. Homogeneity of the raw mix is essential for quality control and plant efficiency. The wet process homogenization system utilizes mechanical and/or pneumatic systems to agitate, blend, and store the homogenized raw mix in cylindrical tanks or basins until it is fed into the pyroprocessing system. The most common homogenization system used for dry process cement plants over the past several decades is the pneumatic system based on the air fluidization method. The homogenized raw mix is commonly referred to as kiln feed (Chatterjee, 2004).

10

The pyroprocess is the focal point of the cement manufacturing process. Rotary kilns are long, cylindrical, and slightly inclined (3 – 4%) furnaces that are lined with refractory bricks to protect the steel shell and retain heat within the kiln (Manias, 2004). The kiln feed is fed into the upper end of the kiln that rotates on its longitudinal axis. The fuels for the kiln are burned at the lower end of the kiln. As the kiln feed enters the pyroprocess the materials are gradually heated to form calcium oxide, which combines with silicon dioxide at temperatures exceeding 1400oC in the kiln. Alumina and iron act as fluxing agents, lowering the reaction temperature of the mix to a practical firing temperature. Although there are many different kiln system designs, all kiln feed undergoes the same reactions during the pyroprocess to form clinker – the hard pellets that typically range in size from 0.3 to 5.1 cm in diameter. The chemical and physical transformations of the kiln feed to clinker are quite complex, but can be viewed conceptually as the sequential events listed in Table 2.1 (Manias, 2004).

The four major compounds of clinker that constitute approximately 95% of the clinker, by mass are: tricalcium silicate (C3S) (35 – 65%), dicalcium silicate (C2S) (10 – 40%), tricalcium aluminate (C3A) (0 – 15%), and tetracalcium aluminoferrite (C4AF) (5 – 15%) (Taylor, 1997). C3S and C2S are commonly referred to as alite (impure C3S) and belite (impure C2S), respectively. Alite typically contains 3 – 4% of substituent oxides, the most significant of which are Fe2O3, MgO, and Al2O3. Belite may contain 4 – 6% of substituent oxides of which Al2O3 and Fe2O3 are most common (Taylor, 1997). Alite and belite constitute about 65 – 75% of PC and the combined total content of the four principal clinker compounds in PC is approximately 85%. Cement chemistry nomenclature abbreviations are as follows: C = CaO, S = SiO2, A=Al2O3, and F=Fe2O3. Several other compounds – such as alkali sulfate and calcium oxide – are present in minor amounts. Figure 2.2 shows the phase transformation of kiln feed to clinker at different stages within the pyroprocess (Manias, 2004).

11

Table 2.1 Kiln material transformations (Manias, 2004) Temperature, ˚C 100 100-300 450-900 700-850 800-1250 >1250 1330-1450 1300-1240 1250-100

Material Transformation Evaporation of free water Removal of adsorbed water in clay materials Removal of chemically bound water Calcination of carbonate materials Formation of belite (C2S), aluminates, and ferrites Formation of liquid phase melt Formation of alite (C3S) Cooling of clinker to solidify liquid phase Clinker cooled in cooler

Figure 2.2 Clinker reactions in kiln feed as a function of temperature (Manias, 2004)

12

The pyroprocess is the most energy intensive component of the overall manufacturing process. The most commonly used fuels are coal, natural gas, and oil. Coal can contain significant quantities of sulfur, trace metals, and other halogens that can influence the clinker composition as well as kiln operation dynamics. Natural gas and oil typically contain less sulfur for an equal amount of calorific energy. The use of supplemental fuels – such as petroleum coke, used tires, impregnated sawdust, waste oils, lubricants, sewage sludge, metal cutting fluids, and waste solvents – has expanded in recent years. Minor trace elements from these supplemental fuels can influence clinker composition and kiln performance (Greco et al., 2004).

The pyroprocess at each cement plant can differ substantially depending on the state of technological advancement and the raw materials used. The three major types of kiln pyroprocessing for cement manufacturing in North America are: wet, long-dry, and preheater/precalciner. The main kiln design, production, and energy consumption characteristics for each process are provided in Table 2.2. An important common aspect of different cement pyroprocesses is the presence of the burning zone where the flame temperatures exceeds 1400oC (Manias, 2004). Table 2.2 Summary of operation data on different kiln systems (Manias, 2004) Specific Fuel Consumption, kcal/kg 1300-1650 1100-1300 750-900 720-850

Kiln Systems rpm tpd/m3 Length/Diameter Wet 1 0.45-0.8 30-35 Long-dry 1 0.5-0.8 30-35 Preheater 2 1.5-2.2 14-16 Precalciner 3.6 3.5-5.0 10-14 rpm: revolutions per minute tpd/m3: clinker produced in tonnes per day cubic metre kWh/t: electric energy consumed in kilowatt hour per tonne of clinker

13

kWh/t 17-25 20-30 25 25

Residence Time minutes 180-240 180-240 30-40 20-30

In the wet and long-dry processes the entire pyroprocess occurs in the kiln, as shown in Figure 2.3(a). In the wet process the raw materials are introduced to the kiln as slurry containing 30 – 40% water, which results in a relatively high energy consumption (ElSayed et al., 1991). The kiln usually has a system of chains near the feed end of the kiln to improve heat transfer from hot gases to the solid materials. Kiln rotation allows the chains to be exposed to the hot gases, and they transfer heat to the cooler materials at the bottom of the kiln. The long-dry kiln process is a newer technology than the wet process; while both processes are similar, the long-dry kiln feed is dry. The long-dry kiln process is the most widely used process for clinker production today and is more energy efficient than the wet process (Manias, 2004).

Due to higher energy prices and improved technology, the design of long-dry kiln systems has evolved into a process consisting of a preheater with a number of cyclone stages (five or, in modern kiln systems, even six) to promote heat exchange between the hot kiln exit gases at 1000°C and the incoming dry kiln feed, as shown in Figure 2.3(b). Calcination (decarbonation) is the decomposition of calcium carbonate to free calcium oxide. The material entering the rotary kiln section is already at around 800°C and partly calcined (20 – 30%) with some of the clinker phases already present. The improved heat transfer allows the length of the kiln to be reduced, relative to the length of kilns in the wet and long-dry kiln processes. In recent decades, the precalcination technology has also been introduced as an energy saving measure and is a modification of the preheater process. In the precalciner process, the combustion air for burning fuel in the preheater no longer passes through the kiln, but is taken from the cooler region by a special tertiary air duct to a specially designed combustion vessel in the preheater tower. Typically, 60% of the total fuel is burnt in the calciner, and the kiln feed is more than 90% decarbonated before it reaches the rotary kiln section allowing for increased efficiency (Manias, 2004).

14

(a)

(b)

Figure 2.3 Schematic of (a) a wet and long-dry pyroprocess and (b) a preheater/precalciner pyroprocess (with a single preheater tower) (Manias, 2004)

15

The final component of all pyroprocess systems is the clinker cooler. The most common types of clinker coolers are rotary, planetary, and reciprocating grate. The clinker is cooled from approximately 1250°C to 100°C by ambient air. The air passes through the bed of clinker and then passes into the kiln for use as combustion air. Clinker that is cooled rapidly typically results in a higher quality clinker (Peray, 1986).

The last step of PC manufacturing is the blending and grinding of clinker and up to 5% – 6% calcium sulfate in a ball or tube mill (finish mill). Calcium sulfate, typically in the form of gypsum and/or natural anhydrite, is generally acquired from a source external to the cement plant. The finish mill reduces the size of the clinker and calcium sulfate to a maximum diameter of 100 micrometers and consumes a large portion of the electric energy in the cement manufacturing process (30 to 50 kWh/ton of cement). The total electric energy consumption to make PC is between 110 and 130 kWh/ton of cement. (Hawkins et al., 2004).

CKDs are removed from the pyroprocess mostly for quality control and/or stable operation of the kiln (CKD removal from the pyroprocess is discussed in further detail in the following section). The U.S. cement plant average rate of CKD removal from the manufacturing process has been reported to be 11.5% of clinker production for wet kilns, 10.5% of clinker production for long-dry kilns, and 4.0% of clinker production for preheater/precalciner kilns (EPA, 1993). It appears that the development of preheater/precalciner systems has resulted in significantly lower amounts of CKDs removed from the pyroprocess per tonne of clinker relative to the wet and long-dry kiln processes.

16

2.1.2

CKD Generation

CKDs are a fine by-product of the PC rotary kiln production operation that is captured in the air pollution control dust collection system. As kiln feed travels through the kiln, the finest particles of the raw materials, partially processed feed, and components of the final product are entrained in the combustion gases flowing countercurrent to the feed. The particulates and combustion gas precipitates that are removed from the gas stream by air pollution dust collection systems are collectively referred to as CKD (Hawkins et al., 2004).

Many cement plants return all or a portion of the CKD from the dust collection system to the pyroprocess with kiln feed or at mid-kiln with dust scoops. The most desirable application of CKDs is to introduce as much as possible back into the clinker production cycle. The CKDs that are not returned as a pyroprocess input or otherwise used beneficially are placed in landfills (Hawkins et al., 2004). Although it is difficult to quantify a direct correlation between dust generation and plant operation, the amount of CKD generated strongly depends upon the type of process and design of gas velocities in the kiln. Other factors include kiln feed composition, fuel composition, kiln operation, and type of dust collection system (Bhatty, 1995).

The raw materials and fuel inputs can have a significant impact on the chemical composition and amount of CKD removed from the pyroprocess. If the raw material and/or fuel inputs contain substantial amounts of volatiles (sodium, potassium, chloride, and/or sulfur), a higher quantity of CKD will likely be generated in the pyroprocess. These elements partially or completely volatilize in the sintering/burning zone close to the flame and are collected in the gas stream flowing counter-current to kiln feed (Hawkins et. al., 2004). Some of the volatile compounds cannot readily exit the pyroprocess with the gas stream because they condense in the cooler parts of the system. As volatile compounds pass through the melting, vapourizing, and condensing cycle, their concentration in the pyroprocess can increase to the point where they can be

17

catalysts for undesirable coating buildup and ring formation in the kiln. The melting points and relative volatilities of common kiln volatile compounds are shown in Table 2.3. Materials that volatilize in the burning zone will tend to accumulate onto the surfaces of smaller particles of kiln feed in the cooler parts of the pyroprocess or remain in the gas stream and be collected in the dust collectors as a CKD (Manias, 2004).

Table 2.3 Melting points and relative volatiles of different compounds in the kiln burning zone (Manias, 2004) Volatile Compounds Melting Point, ˚C Range of volatility*, % CaCl2 772 60 to 80 KCl 776 60 to 80 NaCl 801 50 to 60 884 35 to 50 Na2SO4 K2SO4 1069 40 to 60 CaSO4 1280 --*Range of volatility: % of compound that will volatilize at melting point

The location at which a CKD is extracted from the pyroprocess also has an impact on its characteristics. For example, preheater and precalciner kilns typically extract the CKDs with an alkali/chloride bypass system that is located between the preheater tower and the kiln feed end of the rotary kiln. The temperatures in this region are very different from the temperatures that CKDs are exposed to in the wet and long-dry kiln processes and this gives it unique characteristics. In the bypass system, a portion of the kiln exit gas stream is removed and quickly cooled by air or water to condense the volatiles to fine particles (Manias, 2004).

18

The CKD particles in the kiln exit gas stream of all pyroprocess are removed by dust collection systems. Common kiln dust collection systems include electrostatic precipitators (ESPs), baghouses, and cyclones. Kiln processes equipped with ESPs separate the CKD in multiple electric fields as illustrated in Figure 2.4. The CKD collected in the subsequent fields are generally smaller in size and tend to have higher concentrations of volatiles than the coarser CKD particles. Therefore, it is possible to return the less volatile CKD from the first fields to the pyroprocess and remove the more volatile CKD in the last fields. Baghouses and cyclones do not allow for segregation of CKD based upon volatile concentration.

Figure 2.4 Schematic of electrostatic precipitator (ESP) efficiency (Peethamparan, 2002)

19

There are three major reasons for removal of CKD from the pyroprocess (Kessler, 1995). First, clinker quality must be maintained. For example, the level of alkalis, chlorides, and/or sulfates in the raw materials may be higher than the quality control targets and, therefore, a portion of CKD would need to be removed to reduce the concentration of volatiles. Second, a portion of CKD may need to be removed to maintain stability of the kiln process. As previously discussed in this section, volatiles at high concentrations in the kiln can cause severe material build-up. This can lead to challenging operational problems such as instability, production loss, and blockage, even to the point where the kiln must be shut down (Peray, 1986). The third major reason for removal of CKD from the pyroprocess is due to the lack of a mechanism to return the CKD to the kiln. This is more prevalent in wet process cement plants that were designed and built when the manufacturing challenges and costs associated with recycling CKD back to the kiln from the dust collection system were greater than the costs of removing CKD from the pyroprocess and placing it in landfills.

2.1.3

Fresh and Landfill CKD

Fresh CKDs are generally difficult to handle because of their fine, dry, powdery nature and caustic characteristics. The addition of water to mitigate blowing and dusting problems during transport of fresh CKDs to landfills is common. Adding water at this stage can cause hydration of the free lime and significantly reduce possible cementitious potential for other applications. CKD landfills normally represent many years of cement production. They are usually found in very large above-ground stockpiles or backfill quarries. The surface of the landfill site typically crusts over and becomes hard while the interior of the pile can stay relatively loose. Some of the interior material can remain unhydrated, even after many years, if exposure to moisture is limited. CKD landfills are usually located relatively close to the cement manufacturing plants and vary in age and composition. Exposure to the elements (moisture in particular) reduces the chemical reactivity of the kiln dusts thereby making landfill CKD composition very different from that of fresh CKD.

20

Fresh CKD and landfill CKD should be assessed independently for their potential use as a partial replacement of PC. Changes in composition can occur when CKDs are subject to weather conditions and compaction procedures. In addition to converting lime to calcium hydroxide, exposure to the natural environment could also decompose calcium langbeinite into syngenite and gypsum. The particle size of a CKD can also change due to hydration and compaction. These changes could have large effects on the hydration of a CKD-PC blend.

There are large amounts of CKDs in stockpiles and landfills that are a potential source for CKD applications. The process of utilizing landfill CKDs, however, can be very challenging. Landfill CKDs will often harden and require crushing and screening equipment to remove over-sized pieces as well as any waste that may have become combined with the CKDs.

2.1.4

CKD Applications: Cement Industry Perspective

The cement industry has a keen interest in finding practical applications for CKDs in order to reduce costs and environmental concerns related to managing their removal from the pyroprocess. ASTM D5050 lists several beneficial applications of CKD that include: soil fertilization, soil stabilization, raw material for glass manufacture, sewage and wastewater treatment, and waste pollution control.

Although there have been significant developments in the use of CKDs, large amounts continue to be placed in landfills each year. Researchers continue to investigate the use of CKDs in several fields. In particular, the use of CKDs as a partial replacement of traditional construction materials continues to be an area of active interest. A number of researchers have investigated the use of CKDs for subgrade consolidation for highway construction, cement and masonry products, contaminated soil and sludge stabilization, and partial replacement of asphalt. Most of the previous research on CKD applications

21

has been conducted using fresh CKDs while the issue of using CKDs from landfills and stockpiles has not been explored in great detail (Sreekrishnavilasam et al., 2006).

From the perspective of the cement industry, the most beneficial utilization of CKDs that are removed from the pyroprocess is as partial replacement of PC. ASTM currently authorizes the use of processing additions, including CKDs, as a partial replacement of PC provided that the blend complies with the requirements of ASTM C150 and ASTM C465. The American Association of State Highway and Transportation Officials (AASHTO) limits the amount of processing addition to 1% of the total blend. Canadian Standards Association (CSA) has similar standards as ASTM, but if the processing addition is above 1% of the total blend, the nature and amount of processing addition in the finished product must be provided. At the time of writing this thesis, the National Cooperative Highway Research Program (NCHRP) 18-11 “Improved Specifications and Protocols for Acceptance Tests on Processing Additions in Cement Manufacturing” is preparing a report recommending that up to 5% of a processing addition can be used as a partial replacement of PC provided that the blend complies with the requirements of ASTM C150 and ASTM C465. (CKDs are one of the processing addition materials assessed as part of the NCHRP study.)

2.1.5

Costs Associated with CKD Disposal

The typical costs associated with CKD disposal in the U.S. in 1995 are presented in Table 2.4. The average cost for CKD disposal, adjusting for inflation increases over 13 years at approximately 2.62% per year and converting short tons to tonnes, is $21.60/tonne in 2008 U.S. Dollars. The average annual clinker production of a cement plant in the U.S. is approximately 800,000 tonnes. As an example, the cost of CKD disposal for a U.S. longdry kiln that produces 800,000 tonnes of clinker and removes CKD at 10.5% of clinker production in 2008 is approximately $1.8 million dollars per year. The cost to manage CKD disposal for a Canadian cement plant under the same conditions is comparable.

22

Table 2.4 Typical costs associated with CKD disposal, $/tonne (Kessler, 1995)

Items Raw Material Costs Kiln Feed Costs: Crushing, Conveying, Drying, and Grinding Kiln Fuel Costs: Dust Calcination and Sensible Heat CKD Transport: Conveying, Hauling, and Dedusting Landfill Maintenance: Monitoring, Pile Maintenance, and Closing Total

2.1.6

Low $1.50 $3.00 $1.00 $0.50 $1.00 $7.00

Average $4.00 $4.50 $1.50 $1.00 $3.00 $14.00

High $5.50 $6.00 $2.00 $1.50 $5.00 $20.00

CKD Environmental Considerations

The United States Environmental Protection Agency (EPA) has conducted extensive studies on the issues of production of fresh CKDs and management of stockpile and landfill CKDs. Fugitive dust emissions, surface water pollution, and groundwater pollution have been addressed in these studies. In recent years, hazardous waste has been used as a fuel in cement kiln operations. The use of waste materials in cement kiln operations has raised concerns regarding the accumulation of heavy metals in CKD generated by plants that use these alternative materials.

The EPA (1993) has classified CKDs as a non-hazardous material under the Bevill’s Amendment; however, it also stated that the runoff from CKD storage and landfill piles has the potential to generate leachate containing hazardous characteristics. Runoff and precipitation from CKD piles have exhibited pH levels above 12.5, which can be highly corrosive. The EPA has also expressed apprehension regarding uncontrolled transport, storage, and disposal of large volumes of CKDs in uncovered and unlined piles that are easily removed by wind and eroded by water (EPA, 1993). Due to the leachate and fugitive dust concerns, standards and guidelines have been developed for management of CKD stockpiles and landfills.

23

2.2

CKD and Portland Cement

A basic understanding of CKD compositions and variabilities is fundamental to any investigation of their use. CKDs never contain just a single component and the range of the components and fineness varies not only with the type of cement kiln operation, but also with the raw materials. The chemical, mineralogical, and physical property differences among CKDs and between CKDs and PC must be well understood in order to understand the potential effects in concrete. CKDs are derived from the same raw materials and pyroprocess as clinker. Since CKDs are only partially burnt (relative to fully burnt clinker), CKD compositions differ from PC (Corish and Coleman, 1995). The fineness of a CKD can also be a factor in its influence on concrete properties and is an additional component to be considered.

2.2.1

Chemical Properties

Sreekrishnavilasam et al. (2006) summarized statistics on the chemical oxide composition of CKDs based on 63 published datasets from different cement plants, as shown in Table 2.5. The table presents a statistical analyses of the main oxides present in CKDs as well as the total alkalis (based upon equivalent sodium molar mass), loss on ignition (LOI), and free calcium oxide content (note: free calcium oxide content was not available for all datasets).

The chemical composition of PC is usually given as oxides on a mass percent basis, determined by various analytical tests, such as those in ASTM C114. PC has a typical range for each of the four main oxides: CaO (60 – 66%), SiO2 (19 – 25%), Al2O3 (3 – 8%), and Fe2O3 (1 – 5%) (Taylor, 1997). There are five categories of PC in ASTM C150 with equivalent cement types in CSA: ASTM TI and CSA-GU (normal/general use), ASTM TII and CSA-MS (moderate sulfate resistant), ASTM TIII and CSA-HE (high early strength), ASTM TIV and CSA-LH (low heat of hydration), and ASTM TV and CSA-HS (high sulfate resistant). In 2005, a survey of the 123 cement plants in North America was conducted to determine the chemistry of the different categories of PC

24

manufactured in North America (Tennis and Bhatty, 2006). The data from the 92 cement plants that responded is presented in Table 2.6. The free calcium oxide and chloride contents for PC normally appear in minimal quantities and are therefore, not typically reported. Lawrence (1998a) reported that 132 samples of PC had an average free lime content of 1.24% in a range of 0.03 – 3.68%. PC chloride content is typically less than 0.01%. Tennis and Bhatty (2006) did not present data for TIV, as it is not produced in significant amounts in North America.

Table 2.5 CKD chemical oxide composition, free lime, and loss on ignition, and statistical analysis of 63 published datasets (Sreekrishnavilasam et al., 2006) Chemical Composition, %

fCaO & Ca(OH)2 %

Loss on Ignition %

CaO

SiO2

Al2O3

Fe2O3

MgO

SO3

Na2O

K2O

Equivalent Alkali %

Average

43.99

15.05

6.75

2.23

1.64

6.02

0.69

4.00

3.32

6.75

21.57

Standard Deviation

8.01

4.74

7.83

1.04

0.68

3.93

1.02

3.01

2.44

7.83

8.50

18

31

116

47

41

65

147

75

74

116

39

61.28

34.30

27.18

6.00

3.50

17.40

6.25

15.30

11.42

27.18

42.39

Min. 19.4 2.16 0.00 COV (%) = Co-variance Equivalent Alkali: Na2O + 0.658 x K2O fCaO: free calcium oxide (free lime) Ca(OH)2: Calcium hydroxide

0.24

0.54

0.02

0.00

0.11

0.14

0.00

4.20

COV (%) Max.

Table 2.6 Portland cement chemical oxide composition, total alkali content, and loss on ignition (Tennis and Bhatty, 2006) Chemical Composition, %

Type of Portland Cement (ASTM) TI Normal: Average

Loss on Ignition %

CaO

SiO2

Al2O3

Fe2O3

MgO

SO3

Equivalent Alkali %

63.23

20.17

5.07

2.66

2.51

3.26

0.70

1.52 0.48

Standard Deviation

1.04

0.66

0.54

0.44

1.02

0.62

0.26

TII Moderate Sulfate Resistant: Average

63.66

20.85

4.62

3.32

1.98

2.91

0.56

1.39

Standard Deviation

0.84

0.52

0.37

0.40

0.92

0.39

0.26

0.40

TIII High Early Strength: Average

63.33

20.38

4.84

2.86

2.21

3.60

0.61

1.51 0.41

Standard Deviation

0.93

0.70

0.64

0.59

0.93

0.55

0.27

TV Sulfate Resistant: Average

63.85

21.61

3.80

3.87

2.18

2.34

0.45

1.29

Standard Deviation

0.66

0.67

0.35

0.67

0.91

0.28

0.12

0.44

Equivalent Alkali = Na2O + 0.658 x K2O

25

The data in Table 2.5 indicates that calcium and silica oxides are the major constituents for CKDs, although these values are lower than what is found for PC in Table 2.6. Free calcium hydroxide can sometimes appear as calcium hydroxide due to exposure to moisture. The CKD combined free calcium oxide and calcium hydroxide contents and LOI are significantly higher than in PC for this dataset. It is also observed that CKDs generally contain higher concentrations of sulfates and total alkalis than PC. These findings are not surprising since volatiles are preferentially drawn towards CKD in the kiln pyroprocess. The alumina, iron, and magnesium concentrations of CKD and PC appear to be similar. The CKDs tend to have higher concentrations of potassium than sodium in this dataset, which is to be expected since there is usually a similar ratio for cement raw materials in North America. Although not included in Table 2.5, chlorides can appear in significant levels in CKDs. In an early study, Haynes and Kramer (1982) reported that 113 CKD samples from 102 cement plants in the U.S had an average chloride content of 0.71% with a range between less than 0.01% to as high as 12.3%.

Similar to other construction materials, PC and CKDs have a wide range of trace metals. Trace metals reported in clinker and CKD analyses are normally present in quantities small enough not to influence the performance of the cement ( plaster (calcium sulfate hemihydrates) > chemical anhydrite (soluble calcium sulfate anhydrite) > gypsum (calcium sulfate dihydrate) > syngenite > natural anhydrite.

The amount of gypsum addition to PC is very important. The supply of sulfate ions controls the setting and maximizes the early strength development of cement. This process is disturbed if the renewed hydration of C3A takes place early. Therefore, it is desirable to suppress the C3A hydration with an appropriate amount of sulfate so that it does not coincide with the C3S hydration. The optimum amount of sulfate in PC should (i) retard C3A hydration, (ii) inhibit C3A hydration until C3S hydration takes place to cause the setting of the cement, and (iii) not form an excessive amount of ettringite to

49

cause deleterious expansion after the cement has set and hardened (Bhattacharja, 1997). Each PC is unique and the optimum amount of gypsum for set control as well as other properties – such as early compressive strength – must be determined individually. It is also important to note that the optimum amount of gypsum is not the same for all performance parameters (Gartner et al., 2002).

Tang and Gartner (1988) reported that the presence of soluble sulfates strongly retards initial C3A hydration. In addition, the chemical and physical form is very important. The interblended mixed alkali/calcium sulfates (calcium langbeinite and syngenite) are more effective retarders of C3A than either gypsum or pure alkali sulfates alone. The proposed mechanism takes into account the rate at which the sulfate phases can supply both calcium and sulfate ions to the surfaces of the aluminate phases during early stage hydration. Tang and Gartner (1988) concluded that the use of alkali/calcium double salts increases the rate and chemical potential at which calcium and sulfate ions enter the solution. Single alkali sulfates (K2SO4 and Na2SO4) and apthitalite (3K2SO4.Na2SO4), however, are generally not known to be effective retarders of C3A hydration. Lawrence (1998b) examined the heat of hydration for a PC with different levels of gypsum addition, shown in Figure 2.10. With 0.5% sulfate addition to the PC, the heat evolution of the aluminate peak that was originally superimposed on the main silicate hydration peak was retarded and weakened, and at 2.5% sulfate addition, the aluminate hydration peak was suppressed. Lawrence (1998b) also observed that the main silicate heat peak is depressed at calcium sulfate levels above optimum.

50

(a)

(b)

(c) Figure 2.10 Heat of hydration of cement paste determined by isothermal conduction calorimetry, (20°C and w/c = 0.44); (a) PC (b) PC + 0.5% SO3, (c) PC + 2.5% SO3 (Lawrence, 1998b) Note: Sulfate added as Gypsum (Calcium Sulfate)

51

As the amount of gypsum added to a PC increases, the setting time also increases until a level of stability is reached and the setting time becomes insensitive to further additions of gypsum (Frigione, 1983). The hydration of silicate phases is accelerated in the presence of calcium sulfates (Ish-Shalom and Bentur, 1972). The effect of calcium sulfate on compressive strength at various ages of a PC is shown in Figure 2.11. It is clear that the optimum sulfate is different for the three ages of compressive strength. Soroka and Relis (1983) stated that the optimum content in the compressive strength curve for PC at a particular age implies that the addition of gypsum involves two opposing effects. The lower range of sulfate content has a beneficial effect on strength and can be attributed to the allowance of C3S to hydrate to a beneficial strength by controlling C3A hydration. The range of sulfate greater than the optimum has an adverse effect on strength. Two suggested mechanisms of excessive sulfate ions are: (i) excessive AFt formation and the associated volume increase cause internal cracking of the hardened paste and (ii) C-S-H formation is accelerated but has lower intrinsic strength due to incorporation of sulfate ions into its structure. Both mechanisms may contribute to the phenomenon caused by excessive calcium sulfate (Gartner et al., 2002).

Abnormal setting behaviour is usually related to chemical reactions involving aluminates and sulfate phases (Gartner et al., 2002). False set generally occurs when there is too much readily soluble sulfate, which can come from plaster and/or alkali sulfates. The liquid phase becomes over-saturated with sulfate and precipitation as secondary gypsum occurs. The crystals of gypsum are needle-shaped and weak, but can still restrict the workability of the mix. It is called false set because upon re-mixing the needles will break-up and the mix will revert to its original consistency. Although not commonly reported, false set may also arise due to precipitation of syngenite or ettringite (Gartner et al., 2002).

52

Figure 2.11 Optimization of gypsum additions for compressive strength at different ages (Gartner et al., 2002) (Note: this PC required higher SO3 levels than normal to obtain maximum strength)

53

Excessive sulfate in PC can also lead to expansion problems due to formation of AFt. The reaction between C3A and calcium and sulfate ions to form AFt involves increases in the volume of solids. When the appropriate amount of calcium sulfate (optimum sulfate) is present in PC, AFt formation occurs when the paste is plastic and volume increases do not impact the integrity of the paste. At higher levels of sulfate (i.e., greater than optimum sulfate), however, the formation of AFt may take place in the hardened paste and possibly cause expansion and/or cracking (Soroka and Relis, 1983).

2.4.7

Chloride

The range of chloride content in CKDs from previous studies is between 0 and 12%. PC, however, generally has less than 0.01% chloride content. The American Concrete Institute (ACI 318) guideline for maximum water soluble chloride ion (Cl-) in concrete, as a percent by mass of cement, is limited to: 1% for reinforced concrete exposed to neither a moist environment nor chlorides, 0.15% for reinforced concrete exposed to a moist environment or chlorides or both, and 0.06% for prestressed concrete.

CKD chloride ions generally appear as alkali chlorides (NaCl and/or KCl). Alkali chlorides are more soluble than alkali sulfates and will enter solution within minutes of hydration. Although calcium chloride in CKDs is rare, it is important to consider its effects as well as those of alkali chlorides. Calcium ions could be present during the early stages of hydration due to the presence of calcium-bearing phases that are readily soluble. Therefore, the effects of alkali and calcium ions in conjunction with chloride ions are also considered in this review. Bhatty (1984) also suggested that the alkali chlorides in CKDs would probably behave similarly to calcium chloride.

Calcium chloride is a highly soluble salt that releases calcium and chloride ions into solution and has long been used to shorten both the setting and hardening time of concrete by accelerating the hydration reactions. Calcium chloride is one of the most effective accelerators of PC pastes but the mechanism is not well understood. A practical

54

dosage is typically between 1 and 2%, by mass of cement, and its acceleration effects increase as the concentration of calcium chloride increases (Juenger et al., 2005). Potassium chloride (KCl) and sodium chloride (NaCl) are less effective accelerators than calcium chloride. At very high concentrations, some salts (such as NaCl) act as retarders of C3S (Taylor, 1997). Calcium ions are considerably more effective than any other cation in salts used for accelerating hydration, suggesting that a specific effect is superimposed on a general one (Taylor, 1997).

It is well known that the chemical binding of chlorides is influenced by the amount of aluminate phases. C3A can react with chlorides to form calcium chloroaluminate hydrate or Friedel’s salt (Taylor, 1990). The presence of sulfate ions in the binder, however, reduces the chloride binding capacity of cement. Holden et al. (1983) attributed the reduction in the chloride binding capacity to the preferential reaction of sulfate ions with the C3A phase forming AFt. It is generally accepted that chlorides react with C3A only after AFt formation is complete and sulfate depletion occurs (Taylor, 1997).

The effect of calcium chloride on PC heat evolution is shown in Figure 2.12. The accelerated hydration of PC is indicated by a higher heat liberated at the major peak, the shift of the major peak to the left side, and a narrower curve around the major peak. Early strengths will tend to be higher but the final strengths will be reduced. Shoaib (2002) also stated that the larger amounts of chloride present in CKDs can cause a sort of crystallization of hydration products. The crystallization results in opening the pore system within the hardened samples leading to a reduction in strength. It is well accepted that the acceleration of PC hydration with calcium chloride is mostly due to an acceleration of C-S-H growth. The presence of chlorides is typically associated with higher early strengths (1 and 3 days) and lower later strengths (beyond 28 days). Despite a significant amount of effort to understand the acceleration effect of chloride on PC hydration, however, the detailed mechanism still remains unclear.

55

Figure 2.12 Effect of calcium chloride on heat development in PC (Lerch, 1944)

The presence of chlorides is a durability concern for steel reinforced concrete. The chlorides that are not bound or that leach from the bound hydrates can contribute to steel corrosion. The corrosion of steel in concrete is an electrochemical process and it is a consequence of this corrosion that the surrounding concrete is damaged (Neville, 1983).

56

2.4.8

Alkalis

The CKD equivalent alkali contents (Na2O + 0.658 x K2O) from previous studies range between 0.14 and 11.42%. The equivalent alkali content of PC is typically lower (between 0.5 and 1%). Alkali cations in PC typically occur either as sulfates or in the major clinker phases. The balancing anion sooner or later enters a hydration product of low solubility and an equivalent amount of hydroxyl ion is released (Taylor, 1997). Alkalis (potassium and sodium) in CKD that can greatly impact PC hydration normally occur as readily soluble alkali sulfates and/or alkali chlorides. CKD alkalis can also occur in less soluble form within other mineralogical phases.

In general, soluble alkalis are reported to accelerate hydration at an early age, which is attributed to an increase in the permeability of the layer of hydration product surrounding the alite grains after the reaction has become diffusion controlled (Neville 1983). Set time may shorten due to the increased C3S hydration. It is widely reported that increasing alkali content generally increases early strength (1 and 3 days) and decreases late strength (28 day). However, these effects are modified by the gypsum content of PC. Osbaeck and Jons (1980) reported that the alkali effects on strength are diminished or absent at gypsum contents above the optimum sulfate level. Further, Jackson (1998) stated that when alkalis are present as calcium langbeinite (2CaSO4.K2SO4), a reduction in early strength and an increase of the same magnitude of strength at 28 days would not, relative to a PC with less alkali sulfate, be unexpected.

Excess soluble potassium is widely known to precipitate syngenite which may lead to early stiffening and false set. The presence of soluble alkalis can also impact the rate of gypsum consumption and, thus, affects the levels of calcium and sulfate in solution. It has been reported that the optimum gypsum content increases as the alkali content increases. Calcium salts and alkali hydroxides that are both soluble can influence initial dissolution of C3S due to the common ion effect (Gartner et al., 2002).

57

Increased alkali content in concrete presents durability concerns (Taylor, 1997). ASR is a reaction between hydroxyl ions and certain forms of silica in aggregate to form ASR gel. ASR gel formation causes durability problems that arise as a result of tensile cracks in concrete. The presence of soluble alkalis can also influence air entrainment in fresh concrete (Greening, 1967). Although the exact nature of this influence has not been determined, it is believed that both the air content and the average size of the air voids tend to increase with the amount of soluble alkalis. This can have an adverse effect on freezing and thawing resistance.

2.4.9

Clinker Phases

CKDs can contain some or all of the four major clinker phases. Any changes in the amount of C3A could impact the optimum sulfate balance. Due to the reduced amount of silicates typically found in CKDs relative to PC and assuming all other parameters being equal, the CKD strength gain contribution will be less than the contribution from the replaced PC.

2.4.10 Physical Properties The effect of CKDs as partial replacement of PC can be influenced by the CKD physical properties. The overall particle size distribution of PC and CKDs is called “fineness.” CKDs may tend to have more fine particles than PCs below 8 µm (Section 2.2.3). The densities of CKDs are generally lower than the PC industry standard of 3.15 (KonstaGdoutos and Shah, 2003). Consequently, when CKDs are used as a partial replacement of PC by mass, more CKD particles are required to replace the PC, which may affect rheological properties. The fineness of a CKD will likely affect its chemical reactivity, ability to act as a filler material (providing nucleation sites for hydration), and soundness as a partial replacement of PC.

58

There is a strong correlation between fineness of PCs and water demand, as shown in Figure 2.13. As the Blaine fineness (specific surface area) of a PC increases, the water demand also increases, which is likely in part related to increased chemical reactivity during the early stages of PC hydration. Increased Blaine fineness of a CKD may allow it to have more impact on early age hydration in a CKD-PC blend by means of increased ion dissolution. Therefore, a CKD-PC binder that has increased chemical reactivity during early stages of hydration in comparison to the PC alone may also have a higher water demand.

Figure 2.13 Relationship between water demand and specific surface area of PC (Sprung et al., 1985)

CKDs may also contain significant amounts of calcite and quartz, which are both widely known to be fillers (although it is recognized that limestone does chemically react with aluminate phases to form carboaluminates). The presence of fine calcite and quartz particles generally accelerates early PC hydration (Taylor, 1997). Greater fineness of the CKD filler components could increase the hydration rate of the CKD-PC binder, most likely by an increased surface area that also increases the number of active sites for the

59

nucleation of PC hydration products (nucleation effect). Alternatively, CKDs may contain fine calcium carbonate particles that could fill the gaps between the cement particles; improved particle packing of very fine filler has been attributed to reductions in water demand as well as higher compressive strengths (Hawkins et al., 2005; Sprung and Siebel, 1991).

The fineness or specific surface of the PC is one of the factors that influence the autoclave expansion soundness assessment. If all other parameters of a PC are equal, coarser ground cements have always exhibited a greater amount of autoclave expansion (Klemm, 2005). Narang et al. (1981) quantified the effects of PC fineness on autoclave expansion using a high MgO content PC. The PC was initially ground to a fineness of 225 m2/kg and had an autoclave expansion of 7.06%. When the same PC was ground to a higher fineness of 350 m2/kg, the autoclave expansion was reduced to 1.39%. At an even higher fineness of 400 m2/kg, the autoclave expansion was reduced to 0.24%. Consequently, a reduction in fineness of a CKD-PC blend in comparison to the PC alone may have an adverse effect on the soundness.

2.5

CKD-PC

The study of CKDs as a partial replacement of PC has been an intermittent research area for the past 30 years. A list of the CKD-PC interaction studies, the number of CKD and PC used, CKD replacement levels of PC investigated, the type of specimens used (paste, mortar, and concrete), and the recommended limit of CKD replacement of PC for each study are summarized in Table 2.10. Although these studies have shown that CKDs can be used as a partial replacement of PC in the range of 5% to 15%, very little is known about their exact role in cement paste, mortar, and concrete performance. The studies that have been published on the use of CKDs as a partial substitute for PC often report conflicting effects and mechanisms.

60

This section summarizes the materials, test methods, results, and conclusions of CKD-PC interaction studies conducted over the past 30 years. Performance tests and properties for CKD–PC blends – such as workability and water demand, setting time, hydration, compressive strength, tensile and flexural strength, volume stability, and durability – are presented. In order to focus on understanding the interaction between CKD and PC, the literature review only considers studies of binary mixes.

Table 2.10 Summary of previous CKD-PC studies from literature review Maximum % CKD Replacement Recommended

Author

# of CKD

# of PC

% CKD Replacement Tested

Maslehuddin et al. (2008b)

1

2

0, 5, 10, 15

C

5%

Maslehuddin et al. (2008a)

1

1

0, 5, 10

P, M

10%

P/M/C

El-Aleem et al. (2005)

1

1

0, 2, 4, 6, 8, 10

P, M

6%

Al-Harthy et al. (2003)

1

1

0, 5, 10, 15, 20, 25, 30

M, C

5%

Udoeyo and Hyee (2002)

1

1

0, 20, 40, 60, 80

C

N.R.

Wang et al. (2002)

1

1

0, 15, 25, 50

P, M

15%

Konsta-Gdoutos et al. (2001)

1

1

0, 15, 25

M

15%

Shoaib et al.(2000)

1

1

0, 10, 20, 30, 40

C

5%) than PC 1 alone. At all other ages, the compressive strength of 0% and 5% CKD concrete mixes with PC 1 was similar (±5%). The PC 1 concrete mixes incorporating 10% and 15% CKD 1 had lower compressive strength (>5%) in comparison to PC 1 alone at ages tested after 7 days. For PC 2, the compressive strength of 0% and 5% CKD concrete mixes with PC 2 was similar (±5%) at all ages tested except 56 days (>10%). However, there was generally a decrease in compressive strength (>5%) in the PC 2 concrete mixes with 10% and 15% CKD 1 at all ages, in comparison to PC 2 alone. The authors concluded that up to 5% CKD could be used without apprehension of the reduction in compressive strength, despite the low compressive strength with PC 1 at 3 and 7 days and low compressive strength with PC 2 at 56 days.

87

(a)

(b) Figure 2.19 Concrete compressive strength of CKD 1 at different replacement levels of (a) PC 1 and (b) PC 2 (Maslehuddin et al., 2008b)

88

Maslehuddin et al. (2008a) studied the compressive strength effect of replacing PC 3 with CKD 1 at 0%, 5%, and 10% replacement by mass in mortars. The mortar mixes had a w/b ratio of 0.485 and were tested at 1, 3, 7, and 28 days. The compressive strength of all CKD-PC blends was higher than PC alone at all ages, as shown in Table 2.17. At 1 day, the blends with CKD at 5% and 10% replacement had 28% and 34% higher strength than PC alone, respectively. At 3 days, the blends with CKD at 5% and 10% replacement had 44% and 51% higher strength than PC alone, respectively. At 7 days, the blends with CKD at 5% and 10% replacement had 20% and 21% higher strength than PC alone, respectively. Finally, at 28 days, the blends with CKD at 5% and 10% replacement had 5% and 11% higher strength than PC alone, respectively. At all ages, the compressive strength increased as the quantity of CKD in the mortar mixes increased.

Table 2.17 Mortar compressive strength of CKD 1 at 0%, 5%, and 10% replacement of PC 3 as a function of time (Maslehuddin et al., 2008a)

Average Compressive Strength (MPa) 1 day

3 day

7 day

28 day

100% PC 3

6.31

15.04

22.93

33.17

95% PC 3, 5% CKD 1

8.09

21.60

27.60

34.79

90% PC 3, 10% CKD 1

8.43

22.71

27.69

36.89

El-Aleem et al. (2005) studied the compressive strength effect of replacing PC 4 with CKD 2 at 0%, 2%, 4%, 6%, 8%, and 10% replacement in mortars according to ASTM C109. The w/b ratio was increased to maintain a constant flow. The mortar compressive strength tests were conducted at 3, 7, 28, and 90 days, as shown in Figure 2.20. El-Aleem et al. (2005) reported that the compressive strength for mortar cubes decreased slightly at all ages with CKD content of up to 6%. Above this percentage, the compressive strength decreased sharply. The reduction of compressive strength is suggested to be caused by:

89

(i) the reduction in the cement content, (ii) an increase in the w/b ratio as the percentage of CKD in the blend increased, (iii) an increase in free lime content in cement dust; the higher amount of Ca(OH)2 weakened the hardened matrix, (iv) the formation of chloroand sulfoaluminate phases leads to the softening and expansion of the hydration products, and (v) the porosity also increases, due to the high chloride (7.5%) and sulfate (5.10%) content of CKD 1 (Note: the formation of these products enhances the crystallization of hydration products leading to an opening of the pore system). El-Aleem et al. (2005) concluded that the substitution of PC with CKD up to 6% has no significant effect on the compressive strength of hardened mortar.

I.1 Control I.2 2% CKD I.3 4% CKD I.4 6% CKD I.5 8% CKD I.6 10% CKD

Figure 2.20 Mortar compressive strength as a function of time at different percentage levels of CKD 2 replacement of PC 4 (El-Aleem et al., 2005)

90

Al-Harthy et al. (2003) investigated the compressive strength effect of using CKD 3 as a partial replacement of PC 5 using mortars. The different mortar levels of CKD replacement of PC by mass were 0%, 10%, 20%, 25%, and 30%. The w/b ratio of each mortar mix varied to maintain constant flow. The mortar mixes were tested at 28 days and showed the CKD blended strengths to be lower than the control (31 MPa). Al-Harthy et al. (2003) attributed the lower strengths to the higher w/b ratios of the CKD blended mortars. The 10% CKD blend had a compressive strength of 27 MPa and the 20% CKD blend had a compressive strength of 23 MPa. It is interesting to note that the 25% CKD blend (24 MPa) and 30% CKD blend (24 MPa) had comparable strengths to the 20% CKD blend.

Al-Harthy et al. (2003) also used seven different concrete mixtures that were prepared using 0 (control), 5, 10, 15, 20, 25, and 30% CKD 3 replacement by total mass of PC 5. For each mixture, three water-binder ratios of 0.70, 0.60, and 0.50 by mass were used and the ages tested were 3, 7, and 28 days, as shown in Figure 2.21. A major observation by the authors was that there is generally a decrease in compressive strength with an increase in CKD replacement for cement. The authors also observed that there is more decrease in compressive strengths in mixes with higher w/b ratios (0.70) than in those mixes with low w/b ratios (0.50). At 5% and 10% CKD 3 substitution for PC 5, the reductions in the 28 day compressive strength were 1.8% and 4.5%, respectively (w/b of 0.50). At higher w/b ratio (0.60) the 28 day compressive strength reductions were more significant (12.4% and 18% decreases in strength for 5% and 10% CKD 3 replacement of PC 5). Al-Harthy et al. (2003) stated that CKD is not highly cementitious and the replacement of cement by CKD will lead to less cement content and, therefore, less strength. However, small amounts of 5% and 10% CKD substitution do not seem to have an appreciable adverse effect on strength, especially at low w/b ratios.

91

(a)

(b) (b)

(c)

Figure 2.21 Concrete compressive strengths, w/b (a) 0.70, (b) 0.60, and (c) 0.50, at different percentage levels of CKD 3 replacement of PC 5 (Al-Harthy et al., 2003)

92

Udoeyo and Hyee (2002) studied the compressive strength effect of replacing PC 6 with CKD 4 at 20%, 40%, 60%, 80%, and 100% replacement in concrete at a w/b ratio of 0.65. The tests were conducted at 1, 3, 7, and 28 days. Udoeyo and Hyee (2002) reported that the strength decreased with an increase in CKD content at these very high replacement levels. For example, the 28-day reduction in compressive strength compared to the plain concrete was 7.5%, 33.2%, 71.8%, and 85.3%, respectively, for concrete with 20%, 40%, 60%, and 80% replacement levels of PC 6 with CKD 4. The strength results suggest that CKD 4 is poorly hydraulic.

Wang et al. (2002) studied the effect of CKD 5 at partial replacement levels of PC 7 at 0%, 15%, and 25% on 28-day compressive strength with mortars at a w/b ratio of 0.50. Wang et al. (2002) found that the compressive strength of blends with CKD and cement increased with the CKD replacement of cement up to 15% (47.8 MPa) in comparison to cement alone (46.3 MPa). The specimen with 25% CKD (39.4 MPa) had a much lower compressive strength than the plain cement specimen. Wang et al. (2002) stated that it is commonly accepted that the low hydraulic property of CKD causes the compressive strength to decrease as the amount of CKD replacement increases. Wang et al. (2002) also suggested that the increased strength in the specimen with 15% CKD may be attributed to an appropriate alkalinity that increases the dissolution of silicate species and formation of C-S-H. The authors also noted that 15% CKD replacement of PC significantly reduces the volume fraction of pores larger than 3 µm, which may result in improved strength.

Shoaib et al. (2000) conducted compression strength tests on concrete using CKD 6 as a partial replacement of PC 8 at 0%, 10%, 20%, 30%, and 40% and a w/b ratio of 0.5. The tests were conducted at one, three, and six months. The authors reported that the compressive strength decreased with increasing amounts of CKD. Shoaib et al. (2000) concluded that the critical value of CKD replacement of cement for compressive strength requirements is 10%. They attributed the compressive strength loss to the reduction in

93

cement clinker, which is mainly responsible for strength development. They also concluded that the higher concentration of chlorides present in CKD led to a reduction of strength. It was reported that the chlorides caused the hydration products to crystallize, which resulted in an increase in the total porosity of the hardened sample, thus reducing the compressive strength. The authors further stated that the chloride ions take part in chemical reactions (similar to those involving sulfate ions) and yield chloro-aluminate hydrate 3CaO.Al2O3.CaCl2.12H2O, which can cause softening. Shoaib et al. (2000) reported that due to the presence of alkalis, the microstructure of C-S-H phases became heterogeneous and lowered the ultimate compressive strength.

Batis et al. (1996) used PC 10 and two CKDs (CKD 9 and CKD 10) for testing 90-day compressive strength concrete containing CKD. Each CKD was added as a 6% partial cement replacement, and the w/b ratio was varied at 0.65, and 0.75. At w/b ratio of 0.65 the level of 90-day compressive strength of the specimens with CKD was the same as the plain cement specimen. At a w/b ratio of 0.75, however, the concrete specimen with CKD 9 had a 35% reduction in compressive compared to the CKD 10 concrete and plain cement specimens. Batis et al. (1996) concluded that concrete made with CKD 10 at 6% replacement of PC 10 exhibited as good performance as the reference concrete. In addition, the authors noted that the incorporation of CKD 10 reduced the porosity of concrete from approximately 14% (reference) to 10%, as measured with mercury intrusion porosimetry (MIP) at w/b ratios of 0.65 and 0.75 and after 6 months of exposure in NaCl. It is widely accepted that a reduction in porosity improves compressive strength.

El-Sayed et al. (1991) conducted 28-day compressive strength tests on cement pastes consisting of PC 11 and CKD 11. The CKD was blended at 0%, 3%, 5%, 6%, 7%, and 10% replacement of cement. The w/b ratio of pastes was 0.30. El-Sayed et al. (1991) reported that as the percentage of CKD content in the paste increased, the compressive strength measurements decreased. The authors also reported that up to 5% CKD

94

replacement of PC was within the range of the Egyptian Standard Specifications for Ordinary and Rapid Hardening Cement (36 MPa).

Wang and Ramakrishnan (1990) investigated the compressive strength properties of mortar and concrete made with a binary blend consisting of 5% CKD (CKD 12) and 95% TIII cement (PC 12). The mortar mixes had a w/b ratio of 0.485 and were tested at 1, 3, 7, 14, 28, and 90 days. The authors reported that there was no significant difference in the compressive strengths of CKD-PC mortar and plain PC mortar specimens. Most of the CKD-PC mortar strengths fell within plus or minus 1.4% of the strength of plain PC mortar. The concrete mixes were tested at w/b ratios of 0.45, 0.52, and 0.55 at 1, 3, 7, and 28 days. The authors stated that most of the strengths for CKD-PC concrete were 4% higher in the earlier tests and 3.5% lower at 28 days than for plain cement concrete.

Ramakrishnan (1986) also used CKD 12 to determine whether it was suitable as a 5% replacement of TI cement (PC 15). Mortar and concrete testing was conducted to assess compressive strengths. The mortar mixes had a w/b ratio of 0.485 and were tested at 1, 3, 7, 14, 28, and 90 days. Ramakrishan (1986) noted that although the difference in strength between blended and plain cement mortar cubes was very small, the blends with CKD nearly always had the lower strength in comparison to the plain cement. Ramakrishan (1986), therefore, stated that the mortar specimens showed that the CKD did not possess any cementitious property. The concrete mixes were tested at a cement content of 386 kg/m3 and w/b ratio of 0.45. Six sets of each concrete mix were batched. The concrete mixes were tested at 1, 3, 7, 28, and 90 days. As opposed to the mortar specimens, the concrete with CKD had equal or higher compressive strengths than the plain concrete at all ages of testing, with the exception of the compressive strengths at 28 days. Ramakrishan (1986) therefore concluded that there was no significant difference in the compressive strength of blended and cement concretes. The author did not explore the reasons for the different impact of CKD on compressive strength between mortars and concrete.

95

Bhatty (1986) used a TI cement (PC 16) with three different CKDs (CKD 13, CKD 14, and CKD 15) to investigate their effect on compressive strength in mortars. The amount of CKD in each blend was fixed at 10% by mass of PC with a w/b ratio of 0.45. Compressive strengths were determined at 1, 7, 28, and 90 days, and one year. Bhatty (1986) reported that the blends of cement and CKD at 10% partial replacement had higher strengths at one day, but were generally lower at 7, 28, and 90 days in comparison to cement alone. The strengths of mortars with CKD after one year, however, were comparable to cement alone.

Bhatty (1985a) used the same CKD (CKD 13, CKD 14, and CKD 15) and cement (PC 16) as Bhatty (1986) to conduct paste compressive strength testing. Cement and CKD blends were prepared by replacing 10% and 20% of cement and a w/b ratio of 0.45. Compressive strengths were determined at 1, 7, 28, and 90 days, and one year. Bhatty (1985a) stated that all blends with CKD had similar or higher strengths compared to cement at one day, with CKD 14 blends producing much higher strengths compared to cement and cement blends with CKD 13 and CKD 15. Blends with CKD 15 generally showed significantly lower strengths at later ages compared to CKD 13 and CKD 14. Bhatty (1985a) also noted that a significantly higher strength at one day was obtained for the blend with 10% CKD 15 compared to that with 20% CKD 15, while the other blends were quite comparable. From seven days to one year, blends made with 10% CKD replacement generally showed higher strengths compared to blends with 20% CKD replacement. This study showed that the strengths are adversely affected when high alkali chloride (potassium chloride) CKD was used. Bhatty (1985a) observed that the higher amounts of calcium carbonate in dusts appeared to be detrimental to strength development, but higher free lime appeared to be beneficial for strength. Blends with CKD containing higher amounts of sulfate developed higher strength compared to blends made with CKD containing lower amounts of sulfate. Also, when sulfate was present in

96

the form of calcium sulfate (CKD 14), better strengths were obtained than when some of the sulfate was also present in the form of alkali sulfates (CKD 13).

Bhatty (1984) also conducted compressive strength testing on pastes using five companion cements and dusts obtained from five different cement plants. The five companion cement kiln dust blends are: PC 17 and CKD 13, PC 18 and CKD 14, PC 19 and CKD 15, PC 20 and CKD 16, and PC 21 and CKD 17. For each CKD blend, the CKD replacement of the PC was 0%, 10%, 15%, and 20% at a w/b ratio of 0.50. Bhatty (1984) stated that at all ages, as the amount of CKD increased, the strength generally decreased except with CKD 15, which consistently showed higher strength at 20% addition compared to 10% and 15% addition levels. CKD 15 contained a much higher chloride and alkali content and much lower sulfate content than the other CKDs. Bhatty (1984) stated that alkali chlorides would probably behave similarly to calcium chloride, and calcium chloride is known to increase concrete strength, especially at one to three days curing. The author also reported that the strengths for blends containing CKD 15 were higher at one and seven days than at 28 and 90 days, when compared to cement at the same ages. Also, strengths increased steadily with the increase in chloride level for blends with CKD 15. The CKD-PC blends not containing CKD 15 decreased in strength as the amount of CKD increased. This trend was more prominent in blends with CKD 16 and CKD 17, which contained moderate amounts of alkali and sulfates in the form of alkali and calcium sulfates than in the blends with CKD 14 where the sulfate was predominantly calcium sulfate. Bhatty (1984) concluded that the compressive strengths for CKD blends containing 10%, 15%, and 20% were lower than cement alone. The highest loss in strength occurred when CKDs with relatively high alkali and chloride contents were used. However, as the amount of this CKD increased in the blend, the strength also increased, likely due to an accelerating effect of alkali chlorides on hydration.

97

Ravindrarajah (1982) used concrete mixes to study the compressive strength effect of CKD 18 as a partial cement replacement of PC 22 at 1, 3, 7, 14, 28, 56, and 90 days. Cement was partially replaced with CKD by mass at 0%, 15%, 25%, 35%, and 45%. The total water content for each mix was different to produce a similar workability. Ravindrarajah (1982) reported that as the percent of cement replaced by CKD increased the compressive strength decreased, and the magnitude of strength reduction was increased with the increase in CKD. The author cited four possible mechanisms to explain the impact of CKD replacement of PC on compressive strength in these tests: (i) alkalis in the CKD may modify the nature and strength of the cement hydration products, (ii) since the CKD dust particles are finer than cement, the hydration of the cementitious particles in the dust may occur at a faster rate than the PC. The author noted this by the development of strength with age expressed as a percentage of its 28 day strength for the control and CKD blended mixes. In general, the concrete with no CKD replacement showed the lowest percentage of the 28-day strength at early ages when compared with the CKD concrete, (iii) the portion of CKD that is not cementitious may act as a fine filler and contribute to an increase in strength through increased compaction or provision of nucleation sites for cement hydration, and (iv) concrete compressive strength is a function of paste strength, aggregate strength, and aggregate-paste bond strength. The presence of CKD causes the paste to become weaker, and as the paste strength weakens, the aggregate-cement paste bond also weakens. Ravindrarajah (1982) concluded that from his limited research, cement in concrete could be safely replaced by up to 15% of CKD by mass from the perspective of short-term strength requirements.

98

A summary of the effects of studies conducted on compressive strength (f’c) with CKDPC blends compared to each of the respective reference plain cements is shown in Table 2.18. Although there were variations between researchers, generally the compressive strength of samples with CKD was lower than those of the control cement samples. Some of the suggested mechanisms for the reduction in strength are a reduction in the cement content, an increase in the w/b ratio (for mixes that varied water to maintain the same workability of all mixes), formation of portlandite, formation of chloro- and sulfoaluminate phases, higher porosity, lack of CKD cementitious value (low hydraulic property), weakening of the paste-aggregate bond, and poor formation of C-S-H due to alkalis from CKD. Some researchers reported that there was less of a decrease in compressive strength between plain cement and CKD blends at lower w/b ratios. Some researchers also noted that the CKD blends were higher at early ages and lower at later ages than for plain cement. An appropriate alkalinity that increases the dissolution of silicate species and formation of C-S-H and CKD acting as fine filler were suggested as mechanisms that could cause an increase in the compressive strength of cement with CKD as a partial substitute.

99

Table 2.18 Compressive strength: from CKD-PC literature review Author(s)

Blend

Type

w/b

Maslehuddin et al. (2008b) Maslehuddin et al. (2008a)

CKD 1/PC 1 CKD 1/PC 2 CKD 1/PC 3

C

N.R.

% CKD Replacement 0,5,10,15

M

0.485

0,5,10

General Effect on f’c 5% N.C 10-15% ↓ ↑

El-Aleem et al. (2005)

CKD 2/PC 4

M

V

0,2,4,6,8,10

↓

Author Suggested Mechanism(s)

(1) (2) (3)

(4)

(5)

Al-Harthy et al. (2003)

CKD 3/PC 5

M C

V K

0,10,20,25,30 0,10,20,25,30

↓ ↓

(6) (7)

Udoeyo and CKD 4/PC 6 C 0.65 0,20,40,60,80 Hyee (2002) P = Paste; M = Mortar; C = Concrete V = varied w/b ratio to maintain consistent workability K = constant w/b ratio, but more than one w/b ratio was tested N.C. = No Change f’c = compressive strength

100

↓

Reduction in the cement content An increase in the w/b ratio Increase in free lime content in cement dust; the higher amount of Ca(OH)2 weakened the hardened matrix. The formation of chloro-and sulfoaluminate phases leads to the softening and expansion of the hydration products. The porosity increases due to the high chloride (7.5%) and sulfate (5.10%) content of the CKD (formation of these products enhances the crystallization of hydration products leading to an opening of the pore system). More decrease in compressive strength at higher w/b ratios CKD is not highly cementitious.

Table 2.18 (continued) Compressive strength: from CKD-PC literature review Author(s) Wang et al. (2002)

Blend

Type

w/b

CKD 5 / PC 7

M

0.50

% CKD Replacement 0,15,25

General Effect on f’c 15% ↑ 25% ↓

Author Suggested Mechanism(s) (8) (9)

(10)

Shoaib et al. (1999)

CKD 6 / PC 8

C

0.50

0,10,20,30,40

↓

(11) (12)

(13) (14)

Batis et. al (1996)

CKD 9 / PC 10 CKD 10/ PC 10

C C

K K

0,6 0,6

↓ N.C.

(15)

(16)

El-Sayed et al. (1991) Wang and Ramakrishnan (1990)

CKD 11 / PC 11

P

0.30

0, 3,4,5,6,7,10

↓

CKD 12/PC 12

M C

0.485 K

0,5 0,5

N.C. N.C.

Ramakrishnan (1986)

CKD 12/PC 15

M C

0.485 0.45

0,5 0,5

↓ N.C.

0,10 0,10 0,10

1d ↑, rest ↓ 1d ↑, rest ↓ 1d ↑, rest ↓

Bhatty (1986)

CKD 13/PC 16 M 0.45 CKD 14/PC 16 M 0.45 CKD 15/PC 16 M 0.45 P = Paste; M = Mortar; C = Concrete V = varied w/b ratio to maintain consistent workability K = constant w/b ratio, but more than one w/b ratio was tested N.C. = No Change f’c = compressive strength

101

Low hydraulic property of CKD causes the compressive strength to decrease. Increased strength of 15% CKD-PC blend may be attributed to an appropriate alkalinity that increases the dissolution of silicate species and formation of C-S-H. At 15% CKD replacement of PC, the reduction of volume fraction of pores larger than 3um may result in improved strength. Loss of cement clinker which is mainly responsible for strength development CKD Cl- cause crystallization of hydration products resulting in opening of pore system of the hardened samples leading to strength loss Chloro-aluminate formation causes softening CKD alkalis cause the C-SH phases to become heterogeneous & lowers strength CKD 9 concrete specimen was same as control at w/b of 0.65, but at 0.75 was dramatically lower. CKD 10 concrete specimen had lower porosity (MIP) compared to the concrete specimen without CKD.

(17) Most of the CKD concrete specimens were 4% higher at early ages and 3.5% lower at 28 days than for plain concrete specimens. (18) CKD does not possess any cementitious value.

Table 2.18 (continued) Compressive strength: from CKD-PC literature review Author(s)

Blend

Type

w/b 0.45 0.45 0.45

% CKD Replacement 0,10,20 0,10,20 0,10,20

General Effect on f’c 1d ↑, rest ↑↓ 1d ↑, rest ↑↓ 1d ↑, rest ↑↓

Bhatty (1985a)

CKD 13/PC 16 CKD 14/PC 16 CKD 15/PC 16

P P P

Bhatty (1984)

CKD 13/PC 17 CKD 14/PC 18 CKD 15/PC 19 CKD 16/PC 20 CKD 17/PC 21

P P P P P

0.50 0.50 0.50 0.50 0.50

0,10,15,20 0,10,15,20 0,10,15,20 0,10,15,20 0,10,15,20

1d N.C., rest ↓ 1d ↑, rest ↓ ↓ 1d N.C., rest ↓ ↓

Ravindrarajah (1982)

CKD 18/PC 22

C

V

0,15,25,35,45

↓

P = Paste; M = Mortar; C = Concrete V = varied w/b ratio to maintain consistent workability K = constant w/b ratio, but more than one w/b ratio was tested N.C. = No Change f’c = compressive strength

102

Author Suggested Mechanism(s) (19) High alkali chloride (KCl) in CKD reduces f’c. (20) High calcium carbonate in CKD reduces f’c. (21) Higher free lime in CKD increases f’c. (22) Blends with CKD containing higher amounts of sulfate developed higher strength compared to blends made with CKD containing lower amounts of sulfate. (23) When sulfate was present in the form of calcium sulfate (CKD 14), better strengths were obtained than when some of the sulfate was also present in the form of alkali sulfates (CKD 13). (24) Strengths increased steadily with increase in chloride level for blends with CKD 15. (25) The CKD-PC blends not containing CKD 15 decreased in strength as the amount of CKD increased. This trend was more prominent in blends with CKD 16 and CKD 17 which contained moderate amounts of alkali and sulfates in the form of alkali and calcium sulfates than in the blends with CKD 14 where the sulfate was predominantly calcium sulfate. (26) The highest loss in strength occurred when CKD with relatively high alkali and chloride contents were used. However, as the amount of this CKD increased in the blend, the strength also increased, likely due to an accelerating effect of alkali chlorides on hydration (acting similar to calcium chloride). (27) Alkalis may modify hydration products. (28) CKD may act as a fine filler. (29) CKD presence weakens paste and aggregate-paste bond.

2.5.6

Flexural and Tensile Strength

Al-Harthy et al. (2003) investigated the flexural strength effect of using CKD 3 as a partial replacement of PC 5 using concrete. Al-Harthy et al. (2003) used seven different concrete mixtures that were prepared using 0 (control), 5%, 10%, 15%, 20%, 25%, and 30% CKD 3 replacement by total mass of cement. For each mixture, three water-binder ratios of 0.50, 0.60, and 0.70 by mass were used (the age at which the specimens were tested was not specified but it is assumed that it was at 28 days). Flexural strength measurements were determined using a two-point loading system. Toughness values, which measure the ability of a material to absorb energy up to fracture, were calculated based on the area under the stress-strain diagram. Similar to the effects on compressive strength, the authors stated that the flexural strength and toughness values decreased with an increase in CKD replacement for cement but at 5% and 10% replacement levels did not have an appreciable adverse effect (especially at low w/b ratios). Al-Harthy et al. (2003) attributed the reduction in flexural strength and toughness values to a reduction in the cement content in the blends as the amount of CKD increased.

Udoeyo and Hyee (2002) studied the split tensile strength and modulus of rupture effects of replacing PC 6 with CKD 4 at 20%, 40%, 60%, 80%, and 100% replacement in concrete at a w/b ratio of 0.65. The tests were conducted at 1, 3, 7, and 28 days. Similar to the results of compressive strength, Udoeyo and Hyee (2002) reported that the split tensile strength and modulus of rupture decreased with an increase in CKD content. The reduction in split tensile strength compared to the plain concrete was approximately 24%, 48%, 65%, and 90%, respectively, for concrete with the very high 20%, 40%, 60%, and 80% replacement levels of PC 6 with CKD 4. The reduction in modulus of rupture compared to the plain concrete was approximately 18%, 70%, and 90%, respectively, for concrete with 20%, 40%, and 60% replacement levels of PC 6 with CKD 4. Udoyeo and Hyee (2002) did not suggest possible mechanisms for CKD-PC effects on split tensile strength and modulus of rupture.

103

Wang et al. (2002) studied the effect of CKD 5 at partial replacement levels of PC 7 at 0%, 15%, and 25% on 28-day flexural strength with mortars at a w/b ratio of 0.50. Wang et al. (2002) found that the flexural strength of blends with CKD and cement increase with the CKD replacement of cement up to 15% (8.5 MPa) in comparison to cement alone (8.2 MPa). The specimen with 25% CKD (7.6 MPa) had a much lower flexural strength than the plain cement specimen. Wang et al. (2002) stated that the increased strength in the specimen with 15% CKD may be attributed to an appropriate alkalinity that increased the dissolution of silicate species and formation of C-S-H. Wang et al. (2002) also reported that 15% CKD replacement of PC significantly reduced the volume fraction of pores larger than 3 µm, which may result in improved strength.

Shoaib et al. (2000) conducted splitting tensile strength tests on concrete using CKD 6 as a partial replacement of PC 8 at 0%, 10%, 20%, 30%, and 40% at a w/b ratio of 0.5. The tests were conducted at one, three, and six months. The authors reported a gradual decrease in the splitting tensile strength for all cylinders of concrete samples as the amount of CKD increased. The reduction in tensile strength was attributed to the lower bond strength between the aggregate and paste. Shoaib et al. (2000) stated that as the amount of CKD increased in the paste, the bond strength between the aggregate and the paste decreased.

Wang and Ramakrishnan (1990) studied the splitting tensile and flexural strength properties of binary blends consisting of 5% CKD (CKD 12) and 95% Type III cement (PC 12). The splitting tensile strength mortar mixes had a w/b ratio of 0.485 and were tested at 1, 3, 7, 14, 28, and 90 days. The 14-day tensile strength of the CKD mortar was 10% higher than for the plain cement mortar. At 28 and 90 days, however, there was no significant difference in the tensile strengths of the plain cement and CKD-PC specimens. The flexural strength concrete mixes were tested at w/b ratios of 0.45, 0.52, and 0.55 at 1, 3, 7, and 28 days. Wang and Ramakrishnan (1990) stated that the results of flexure strength tests of concrete specimens with CKD were within a range of ±4% of those of

104

the plain cement concrete and, therefore, not significant. Wang and Ramakrishnan (1990) did not suggest possible mechanisms for CKD-PC effects on split tensile and flexural strength.

Ramakrishnan (1986) studied the mortar splitting tensile and concrete flexural strength properties made with a binary blend consisting of 5% CKD (CKD 12) and 95% TI cement (PC 15). The splitting tensile strength mortar mixes had a w/b ratio of 0.485 and were tested at 1, 3, 7, 14, 28, and 90 days. Ramakrishnan (1986) reported that for most of the CKD-PC mortar splitting tensile strengths were lower than the corresponding plain cement mortar strengths. The flexural strength concrete mixes were tested at a w/b ratio of 0.45 at 1, 3, 7, and 28 days. Ramakrishnan (1986) reported no significant difference in flexural strength between concretes containing CKD and plain concrete.

Ravindrarajah (1982) used concrete mixes to study the flexural and tensile strength effects of CKD 18 as a partial cement replacement of PC 22 at 1, 3, 7, 14, 28, 56, and 90 days. Cement was partially replaced with CKD by mass at 0%, 15%, 25%, 35%, and 45%. The total water content for each mix was varied to produce similar workability. Ravindrarajah (1982) also conducted tests to determine the flexural and tensile strengths. As in the compressive strength test results, the flexural and tensile strengths decreased with increased replacement of cement with CKD.

A summary of the studies conducted on the flexural and splitting tensile effects of CKDPC blends compared to the referenced plain cement is shown in Table 2.19. Generally, the flexural and tensile strength effects of samples with CKD were lower than those of the control cement samples, which is similar to the compressive strength effects. Many of the suggested mechanisms for the reduction in flexural and split tensile strengths were the same as the mechanisms for the reduction in compressive strength. The most commonly suggested mechanism was the weakening of the aggregate-paste bond due to the presence of CKD.

105

Table 2.19 Flexural and tensile strength: from CKD-PC literature review Author(s)

Blend

Al-Harthy et al. (2003)

CKD 3/PC 5

Typ e C

w/b K

% CKD Replacement 0,10,20,25,30

General Effect on f’t ↓

Udoeyo and Hyee (2002) Wang et al. (2002)

CKD 4/PC 6

C

0.65

0,20,40,60,80

↓

CKD 5/PC 7

M

0.50

0,15,25

15% ↑ 25% ↓

Author Suggested Mechanism(s) (1) (2)

(3)

(4)

Shoaib et al. (1999) Wang and Ramakrishnan (1990) Ramakrishnan (1986) Ravindrarajah (1982)

CKD 6/ PC 8

C

0.50

0,10,20,30,40

↓

CKD 12/PC 12

M C

0.485 K

0,5 0,5

↑ N.C.

CKD 12/PC 15

M C C

0.485 0.45 V

0,5 0,5

↓ N.C. ↓

CKD 18/PC 22

(5)

(6)

0,15,25,35,45 (7) (8)

Reduction in the cement content. Less effect at low w/b ratios.

Increased strength 15% CKD specimen may be attributed to an appropriate alkalinity that increased the dissolution of silicate species and formation of C-S-H. At 15% CKD replacement of PC, the reduction of volume fraction of pores larger than 3um may result in improved strength. Weaker aggregate-paste bond as CKD content increases.

Alkalis may modify hydration products. CKD may act as a fine filler. CKD presence weakened paste and aggregate-paste bond.

P = Paste; M = Mortar; C = Concrete. V = varied w/b ratio to maintain consistent workability f’t = flexural and/or tensile strength K = constant w/b ratio, but more than one w/b ratio was tested. N.C. = No Change

2.5.7

Volume Stability

2.5.7.1 Soundness Maslehuddin et al. (2008a) studied the soundness effect of replacing PC 3 with CKD 1 at 0%, 5%, and 10% replacement by mass in pastes using autoclave expansion (ASTM C151). The PC alone, PC with 5% CKD replacement, and PC with 10% CKD replacement were 0.0075%, 0.0130%, and 0.3730%, respectively. Although the CKD-PC blends had higher expansions than PC alone and increased as the percentage of CKD replacement increased, the autoclave expansions were below the 0.80% allowed by ASTM C150.

106

Bhatty (1986) used a Type I cement (PC 16) with three different CKD (CKD 13, CKD 14, and CKD 15) to investigate the effect on autoclave expansion (ASTM C151). The amount of CKD in each paste was fixed at 10% by mass of PC, with a w/b ratio of 0.45. Bhatty (1986) stated that the type of CKD used in the binary blend influenced the autoclave expansion. Bhatty (1986) reported that the CKD-PC blend with CKD 14 showed autoclave expansion comparable to cement alone but higher expansions were noted for CKD 13 and CKD 15. Bhatty (1986) also noted that each CKD-PC blend autoclave expansion was well below the ASTM C150 specification of 0.80%. Bhatty (1986) generally noted that when binary, ternary, and quaternary blends were made from PC 16, the three different CKD, fly ash and slag – the blends containing CKD 15 (a high chloride dust) generally produced higher autoclave expansions than blends with CKD 14, which contained high sulfate.

Ravindrarajah (1982) used cement pastes to determine the soundness of PC-CKD blends using the Le Chatelier apparatus (EN 196-3). PC 22 was partially replaced with CKD 18 by mass at 0%, 25%, 50%, 75%, and 100%. The total water content for each mix was varied to produce similar workability. As the CKD percentage increased, so did the expansion of the samples. This was attributed to the higher level of free lime in the CKD in comparison to cement. Although the level of expansion was well within the range of the British Standard, the expansion was much higher than that of cement.

A summary of the studies conducted on the soundness of CKD-PC blends compared to each of the respective reference plain cements is shown in Table 2.20. High free lime, sulfate, and chloride contents in the CKDs were attributed to the increased autoclave expansions.

107

Table 2.20 Soundness: from CKD-PC literature review Author(s) Maslehuddin et al. (2008a) Bhatty (1986)

Ravindrarajah (1982)

Blend

Type

w/b V

% CKD Replacement 0,5,10

General Effect on Soundness ↓

CKD 1/PC3

P

CKD 13/PC 16 CKD 14/PC 16 CKD 15/PC 16

CKD 18/PC 22

Author Suggested Mechanism(s)

P P P

0.45 0.45 0.45

0,10 0,10 0,10

↓ ↓ ↓

(1)

C

V

0,15,25,35,45

↓

(2)

High chloride CKD generally produced higher autoclave expansions than high sulfate CKD (includes mixes with slag, and fly ash). High CKD free lime content.

P = Paste; M = Mortar; C = Concrete. V = varied w/b ratio to maintain consistent workability N.C. = No Change

2.5.7.2 Drying Shrinkage Maslehuddin et al. (2008b) studied the drying shrinkage effect of replacing PC 1 (TI) and PC 2 (TV) with CKD 1 at 0%, 5%, 10%, and 15% replacement by mass in concrete. The drying shrinkage strain after 3, 7, 14, 28, 56, and 90 days of curing were tested, according to ASTM C157. The drying shrinkage strain at the different percentage levels of CKD 1 replacement of PC 1 is shown in Figure 2.22. For PC 1, the highest shrinkage strain at all ages was with the 15% CKD 1 concrete specimens followed by the concrete specimens with 10% and 5% CKD 1, respectively. The 5% CKD 1 concrete specimens with PC 1, however, were only marginally higher (1000˚C). Then the fused bead samples were placed in the XRF spectrometer to determine the major elements. The alkali, sulfate, and chloride contents for PCs from the XRF analysis were validated using flame photometry,

induction heating (LECO SC-432 Sulfur Analyzer), and

potentiometric titration. Water soluble alkali content was determined according to ASTM C114. One gram of material is put in contact with water for 10 minutes and, after filtration, the amount of water soluble alkalis contained in the aliquot was determined by flame photometry. Some testing procedures developed for PC were modified to accurately determine the chemical composition of the CKDs; these are described in Section 4.1.1.

3.2.2

Mineralogical Properties

The free lime (free calcium oxide) test that is designed for PC (ASTM C114) using hot benzoic acid titration was used for each PC and CKD. Mineralogical characterization of all materials included X-ray diffraction (XRD) and thermal analyses. XRD was performed with a Rigaku D/MAX 2000 diffractometer on pressed powder samples, except CKD D, which was analyzed using PANalytical’s X’Pert PRO. Scanning was performed in the range of 5º ≤ 2θ ≤ 65º with a scan rate of 0.02º 2θ per second. Powder samples were analyzed using standard monochromatic CuKά radiation generated at 20mA and 40 kV. PC gypsum phases were obtained by differential scanning conduction

129

calorimetry (DSC) using the Mettler TA3000 System. CKD samples were analyzed by thermo gravimetric analysis (TGA) in a nitrogen environment using a Netzsch STA 730 thermal analysis apparatus at a heating rate of 10˚C/minute.

3.2.3

Physical Properties

The relative density of each material was obtained by air-comparison pycnometer. The relative density of the material is a required input in the calculation to determine the Blaine fineness. The Blaine air permeability test (ASTM C204) and the percentage of material finer than 45 µm (No. 325) sieve (ASTM C430) were used to determine the fineness of all materials in this research program. The Blaine fineness test is the most widely used method to assess the fineness of PC. The Blaine fineness test indirectly measures the surface area of the cement particles per unit mass. Particle size distribution (PSD) of all materials was also determined using the Malvern laser diffraction particle sizer, 2600 Series. Although there is presently no standard specification for determining the particle size distribution of PC, the cement industry commonly uses this test method to determine fineness of materials. The usual procedures for measuring PC fineness were slightly modified to accurately measure the fineness of the CKDs and fillers; these are discussed in Section 4.1.3.

3.2.4

Dilute Stirred Suspensions

Dilute stirred suspensions were performed on each PC and CKD. A sample of each material was mixed with water in a glass beaker with a water to solid ratio of 10. Each mixture was stirred vigorously for 10 minutes by hand with a glass rod and the temperature of the solution was maintained at approximately 23ºC. The solid material was then separated using a vacuum filter. The liquid solution was placed in a sample tube for analysis. Hyroxyl ion concentration was measured immediately for each sample. Then the solution filtrate was brought to a pH of less than two using nitric acid. The purpose of adjusting the sample pH to less than two was to minimize metal cation precipitation and adsorption onto the sample container wall. It is known that nitric acid can also cause

130

certain elements from glass ampoules to become soluble. Therefore, appropriate plastic ampoules were used to collect the samples. The balance of the cation and sulfide ionic concentrations of each solution was determined by using Inductively Coupled Plasma Atomic Emission Spectrometry (ICP AES) directly; the model used was the Perkin Elmer Model Optima 3000DV ICP AEOS. The chloride ion concentration was approximated using the U.S. Geological Survey public domain PHREEQC geochemical software package.

3.3

CKD-PC Blends

For paste and mortar tests, the amount of CKD (CKD A, B, C, D, E, or F), limestone powder, or silica flour in each blend was either 10% or 20% replacement of PC, by mass. This resulted in 30 binder blends: 2 PC binder blends, 24 CKD-PC binder blends, and 4 PC-filler binder blends. All materials were sieved on a No. 20 sieve and weighed accurately. Each blend was then homogenized by hand with a large spoon in a steel bowl prior to the addition of water and/or fine aggregate (mortar sand). The paste and mortar tests used in this study are described in Sections 3.3.1 to 3.3.7.

For concrete tests, the amount of CKD ranged between 7% and 13% replacement of PC, by mass. CKD D was not available at the time of concrete casting, so the low Blaine fineness CKD D* was used. The concrete CKD-PC blend tests are described in further detail in Section 3.3.8.

3.3.1

Heat of Hydration

PC hydration leads to the evolution of heat and, consequently, isothermal conduction calorimetry is commonly used to assess hydration kinetics of different paste blends. In this study, the TAM Air isothermal conduction calorimeter was used to determine the effects of the CKDs and fillers on the early hydration characteristics of the blends in accordance with ASTM C1679. Eight samples can be analyzed at a time and an air thermostat is used to maintain the isothermal temperature, which can be set between 15

131

and 60°C. The TAM Air utilizes heat conduction to transfer heat away from the sample to a heat sink to keep the sample temperature essentially constant. The flow of heat, caused by the temperature gradient across the sensor, creates a voltage signal proportional to the heat flow. The heat output is calibrated by measuring the output from a known heat source under identical conditions to the hydrating material. To minimize disturbances from outside the calorimeter, an inert reference sample is used. The inert sample is placed on a parallel heat flow sensor. Any external disturbances will influence both the sample and the inert sample identically and be nullified. The detection limit of the TAM AIR is 2 µW and the precision is specified to be ±10 µW. The time constant is approximately 100 seconds. The results can be presented as either differential plots showing the rate of heat evolution as a function of time or integral plots showing the total amount of heat liberated as a function of time.

All materials were stored in tightly sealed plastic bags inside containers at a constant temperature of 23 ± 2°C to pre-condition them prior to testing. Paste specimens with 150 g of solids and a w/b of 0.4 were prepared to study the heat of hydration at 23°C. Distilled water was added to the solids and mixed for 2 minutes in a steel bowl using a kitchen hand-blender at low speed. After 2 minutes, approximately 8 g of paste sample were extracted from the bowl using a 10 ml syringe and injected into a glass ampoule. All paste samples were weighed by mass difference between the glass ampoule with the sample and the empty glass ampoule. The sample was then sealed and placed in the calorimeter, five minutes after the distilled water was initially added. A corresponding reference sample containing inert silica sand was also placed into the calorimeter. The amount of silica sand was determined by calculating the equivalent specific heat capacity to 8 g of PC paste. Heat of hydration for each paste specimen was measured over seven days and performed in duplicate. The rates of heat evolution (mW/g) were measured and recorded approximately every 10 seconds using a computer data acquisition system. Since mixing of the constituents was carried out prior to introducing the sample into the calorimeter, the first five minutes of heat evolution were not measured.

132

3.3.2

Normal Consistency

Normal consistency is a term that is used to describe the degree of plasticity of a freshly mixed PC paste. The normal consistency (w/b ratio expressed as a percentage) was determined for all binders in accordance with ASTM C187. For each binder blend, 650 g of solid material were mixed with water to make a paste. The amount of water required to bring the paste to a standard condition of wetness was regulated by the condition for which the penetration of a standard needle (Vicat needle) into the paste is 10 ± 1 mm in 30 seconds. In order to gain appreciation for the accuracy of this test, ASTM C187 stipulates that the results of single-operator tests should not differ by more than 0.7%.

3.3.3

Initial Setting Time

The initial setting time is often used to evaluate if a paste is undergoing normal hydration reactions. Initial setting time is defined as the time that elapses from the moment water is added until the paste ceases to be fluid and plastic. Most PCs attain initial set within two to four hours. For each binder blend, the paste that was mixed to determine normal consistency was also used to determine initial set time. The time of initial setting of the blended pastes was determined using a Vicat apparatus according to ASTM C191. The time at which the needle penetrates 25 mm into the paste at room temperature was taken to define the initial setting. ASTM C191 specifies that the penetration of the Vicate needle in the paste should be checked 30 minutes after moulding and every 15 minutes thereafter until a penetration of 25 mm or less is obtained. According to ASTM C191, the single operator standard deviation has been found to be ±12 minutes within a range of 49 to 202 minutes initial setting time. To increase the accuracy of the initial set time measurement, the test procedure was modified by increasing the frequency of the Vicat penetrations to every five minutes as the paste approached initial set.

133

3.3.4

Flow

Flow is used to describe the relative mobility (ability to flow) of mortar. The flow for each binder blend was determined on a flow table as described in ASTM C230. Mortars were mixed in accordance with ASTM C305 with one part of binder blend, 2.75 parts of graded sand, and deionized water. Mortars were mixed for each binder blend to determine (i) the flow at a fixed w/b ratio of 0.485 and (ii) the water demand to yield a flow of 110 ± 5 according to ASTM C1437. ASTM C1437 states that the results of properly conducted tests should differ by no more than 11% for single-operator testing.

3.3.5

Compressive Strength

Compressive strength is the most commonly used method to assess cement quality. The compressive strength for each binder blend was determined according to ASTM C109 (CSA A456.2-C3) at 1, 3, 7, 28, and 90 days. 50 mm mortar cube specimens were prepared by mixing one part of binder blend material, 2.75 parts of graded sand, and deionized water addition (w/b ratio of 0.485). The specimens were cured in a humidity chamber at 23±1 °C for 24 hours, then demoulded and immersed in lime saturated water until tested. The compressive strength result is the average of three test specimens from a single batch at the specified curing time. ASTM C109 states that when three cubes represent a test age, the maximum permissible range between specimens from the same mortar batch at the same test age is 8.7% of the average.

3.3.6

Expansion in Limewater

ASTM C1038 is a test method that is used to determine the expansion of mortar bars made from PC in saturated limewater. The amount of expansion is typically related to the amount of calcium sulfate in the PC. In this study, ASTM C1038 was used to assess the expansion of all binder blends. Mortars were mixed in accordance with ASTM C305 with one part of binder blend, 2.75 parts of graded sand, and deionized water. The amount of water required to yield a flow of 110 ± 5 according to ASTM C1437 was used for each binder blend. Four mortar bar specimens (25 x 25 x 285 mm) were prepared for each

134

binder blend and the expansion was calculated as the mean of four mortar bars. The test method specifies calculating the difference in length of specimens at 24 hours from the time the binder blend was mixed with water and at 14 days. The length change of the mortars made from different blends, however, was also measured up to one year for most blends. An expansion limit of 0.020% in 14 days of limewater immersion is in use in CSA A3001.

3.3.7

Autoclave Expansion

Soundness refers to the ability of a paste to retain its volume after it has set. Unsoundness can arise from excessive amounts of hard burned free lime or free magnesia and has the potential to cause delayed destructive expansion. In the autoclave expansion test (ASTM C151), a cement paste specimen (25 x 25 x 285 mm) is placed in an autoclave for three hours at 2 MPa and approximately 216°C. The difference between measurements of the specimen taken before and after the autoclave treatment represents the expansion due to unsoundness. The autoclave expansion test method was used to measure expansion due to the combined effects of both magnesia and free lime for each binder blend. For each binder blend paste, the same w/b ratio used to attain normal consistency and initial setting time was used for the autoclave test. ASTM C151 states that the results of two properly conducted tests by the same operator for expansion of similar batches should not differ from each other by more than 0.07% expansion.

3.3.8

Alkali Silica Reactivity

The concrete prism test is typically used to evaluate the reactivity of aggregate with respect to ASR and also to examine the impact of materials that may be introduced to suppress the expansion due to ASR. The typical test period for evaluating the reactivity of an aggregate is one year, and at least two years with SCM (CSA A23.2-14A and ASTM C1293). For the proposed research study, this test method was modified to assess the direct impact on ASR when using CKD as a partial replacement of PC. The main

135

purpose of this study was to make relative comparison of the binary blends rather than obtaining the absolute values.

The materials used for the ASR concrete durability study were six different CKDs (A, B, C, D*, E and F) and two PCs of high and low alkali content (TI and TII). CKD D and the fillers were unfortunately not available at the time of casting for the concrete prisms. Two series of concrete prisms were cast to assess the effect of CKDs on ASR with Cements TI and TII. The reactive aggregate susceptible to ASR that was used in this study is Sudbury aggregate.

The w/b ratio for all mixes was in the range of 0.42 – 0.45 to maintain a constant slump. The three equal reactive coarse aggregate fractions by mass were of 10, 15, and 20 mm nominal maximum diameter, respectively. The specific gravity of the reactive coarse aggregate was 2.71. The fine aggregate had a fineness modulus of 2.90 and a specific gravity of 2.68. The freshly mixed concrete was tested for slump (ASTM C143), air content (ASTM C231, pressure method), and unit mass (ASTM C138). Two concrete cylinders measuring 100 x 200 mm were prepared from each batch. The cylinders were stored moist at 38 ºC and tested for compressive strength at 28 days. Four concrete specimens from each batch were prepared, measuring 75 x 75 x 300 mm. The expansion of the concrete specimens was measured every three months for a period of 365 days. For each concrete mixture investigated, the expansion (length change divided by the gauge length) was calculated as the mean of four concrete prisms. Mass was also measured for each concrete prism and the mass change was averaged for the four prisms.

ASR Test Series 1: The first set of concrete prisms was cast using 10% replacement of Cement TI with CKD binders (CKD and/or PC). The total alkali content of the concrete was increased to 1.25% Na2Oe of binder mass by adding sodium hydroxide (NaOH) to the mixing water. Cement TI as the binder material alone was used in two control mixtures (Cements TI CTL 1 and CTL 2). The total solid binder was 420 kg/m3 for each

136

mix except for Cement TI CTL 2, which was 378 kg/m3. Cement TI CTL 2 contained the same amount of PC as in the concrete blends with CKDs. Due to the reduction in solid binder material, the Cement TI CTL 2 alkali level was increased to 1.38% Na2Oe of binder mass to give the same total alkali loading as the other blends in Test Series 1.

ASR Test Series 2: The second set of concrete prisms was cast using Cement TII and varying amounts of PC replacement with CKDs. A constant amount of NaOH was added to each mix. The total alkali content of the concrete was increased to 1.25% (Na2Oe) of cement mass by adjusting the amount of CKD replacement in each mix. The amount of NaOH addition to each mix was selected to maintain the range of CKD replacement levels generally close to 10%. The total solid binder for each mix was 420 kg/m3. Cement TII as the binder material alone was used in two control mixtures (Cements TII CTL 1 and CTL 2). Cement TII CTL 2 alkali content was raised using NaOH to a level of 1.03% Na2Oe of binder mass, rather than the 1.25% of alkali loading to give the same total alkali loading contribution of NaOH as the other CKD blends in Test Series 2.

137

4.0

RESULTS AND DISCUSSION

4.1

Material Characterization

The first objective of this thesis was to characterize the seven CKDs, two PCs, and two filler materials. It was found that some of the analytical methods designed for PC do not always provide accurate compositional analysis for CKDs. Therefore, the analytical methods required for accurate analysis of CKDs were identified. The complete chemical analysis and physical properties (relative density, Blaine fineness, and percentage of fine material below 45 µm) of all materials were performed. In addition to the chemical composition and standard fineness tests, quantitative mineralogical compositions, particle size distributions, and dilute stirred suspension analyses were also performed.

4.1.1

Chemical Properties

The characteristics of materials used in cement are traditionally evaluated by an oxide composition based on chemical analysis data. Chemical makeup of a CKD and PC can provide an important indicator of how the CKD-PC blend will perform. It was found that there are very few published works with complete chemical analysis of CKDs in the research of CKD-PC blends. The incomplete CKD chemical composition data provided in previous studies is likely due in part to the application of analytical procedures that are specifically designed for PC, rather than CKDs.

The chemical compositions of the two PCs were determined in accordance with ASTM C114 using X-ray fluorescence (XRF), as stated in Chapter 3. Prior to XRF analysis, loss on ignition (LOI) was performed by igniting the 110˚C dried sample to a constant mass in a muffle furnace at 950 ± 50˚C in an uncovered crucible for 1h. The LOI values obtained result from either exposure to moisture or CO2 (since each of the two PCs only consists of clinker and gypsum, there is no contribution of CO2 from carbonate additions).

138

CKDs usually take between 12 to 24 hours to reach constant mass at 950 ± 50 ˚C. The LOI for CKDs not only reflects prehydration and decarbonization, but also the presence of volatiles (alkali, sulfate, and/or chloride). The ranges of volatilization at the melting point of compounds found in CKDs are shown in Table 4.1. A large percentage of the CKD volatiles will be released from the sample into the atmosphere during the LOI test and during preparation of the fused beads since they are less stable in CKDs than in PC at 950 ± 50 ˚C. This presents two problems: (i) the LOI is not just CO2 and (ii) the XRF quantification of alkali, sulfate, and/or chloride is underestimated. Therefore, direct testing procedures developed for PC in ASTM C114 were used to accurately determine the volatile composition of the CKDs (Babikan and Verville, 2007). The test methods used to measure the volatiles of CKDs were: flame photometry for alkalis, induction heating for sulfate, and potentiometric titration for chloride. The XRF chemical analysis values were then corrected by accounting for the volatiles that were released during the LOI test. The process that was used for chemical analysis of the CKDs is described in Figure 4.1. The CKD chemical composition calculations are presented in Appendix A.

Table 4.1 Melting points and volatility of compounds in CKDs (Manias, 2004) (Note: This table is the same as Table 2.3) Volatile Compounds Melting Point, ˚C Range of volatility*, % CaCl2 772 60 to 80 KCl 776 60 to 80 NaCl 801 50 to 60 Na2SO4 884 35 to 50 K2SO4 1069 40 to 60 CaSO4 1280 --*Range of volatility: % of compound that will volatilize at melting point

139

CKD Sample

CKD Sub-sample 1

CKD Sub-sample 2

CKD Sub-sample 3

CKD Sub-sample 4

LOI and XRF Analysis

Chloride Content: Potentiometric Titration

Sulfate Content: Induction Furnace

Alkali Content: Flame Photometry

Calculate chemical composition by accounting for volatiles released during LOI test

Figure 4.1 Process flow chart for CKD chemical composition analysis

The free lime test for PCs is typically used to determine the free calcium oxide content. This test, however, is also sensitive to calcium hydroxide. The free lime test gives the total of free calcium oxide plus calcium hydroxide contents and does not differentiate between the two. This is generally not an issue for PC free lime analysis since the presence of calcium hydroxide is rare (except in PC that consists of weathered clinker). CKDs, however, can be exposed to moisture during processing to reduce fugitive dust and/or storage outside. Therefore, the results from the free lime test for CKDs should be considered as representative of the combined free calcium oxide and calcium hydroxide contents.

140

The chemical and standard physical properties (relative density, Blaine fineness, and fine material below 45 µm) of all materials are shown in Table 4.2. Cement TI met the specifications for normal PC and is characterized by a relatively high sulfate (4.35%), high total alkali content (0.97%), and high C3A content (11.3%). Cement TII met the specification for a moderate sulfate resistant cement and is characterized by its low C3A (6.1%) and low total alkali (0.57%) contents. The data in Table 4.2 shows the LS and SLX to consist of 95.52% calcite based on 53.49% / 56.00% CaO (by LOI, 42.29% / 44.00% = 96.11%) and 98.15% quartz, respectively.

Comparison of the current CKDs to those from previous research studies as summarized by Sreekrishnavilasm et al. (2006) (Table 2.5) shows that all CKDs were within the maximum-minimum range of the compositions, except for the free lime values for CKDs E and F. CKDs A, B, and C appear to be particularly similar to those in the previously published literature. CKDs A and C were within the standard deviations for each parameter. CKD B had concentrations of calcium oxide, silicon dioxide, and aluminum oxide slightly outside the respective range for standard deviation. CKDs D*, D, E, and F, however, appear to be slightly different from the published dataset. CKDs D*, D, and E each had values for sulfate above the range for standard deviation. CKDs D*, E, and F had higher free limes than the upper limit of the standard deviation. CKDs E and F also had calcium oxide and magnesium oxide contents above the respective ranges for standard deviation. The chloride levels of the CKDs within this study appear to be lower than the full range of chloride levels found in CKDs from previous studies.

As a note of interest, the CKD oxide composition statistical analysis of intermittent daily samples collected over a 3 year period from the same kiln source as CKD C is presented in Table 2.9. Although more variable than PC, the standard deviation results indicate that the CKD from this kiln source is quite consistent.

141

Table 4.2 Chemical and select physical components of PC, CKD, and filler materials (mass %) PCs

Cement Kiln Dusts (CKDs) Wet

Long-dry D* 51.12 14.23 4.05 2.05 2.00 15.05 0.40 3.32 2.58 0.21 2.13 1.61 0.62 0.24 0.12 0.11 0.22 8.23 101.16 100.31 18.20

Fillers PH/PC

Components TI TII A B C D E F CaO 62.03 63.06 44.32 32.70 44.75 45.51 55.18 55.86 SiO2 19.15 20.39 14.25 24.07 14.30 14.41 15.25 16.81 Al2O3 5.83 4.21 3.77 9.12 4.02 4.93 3.78 3.94 Fe2O3 2.46 3.01 1.92 3.78 1.60 2.16 2.26 1.94 MgO 2.18 3.21 1.80 1.82 1.02 1.68 2.85 3.08 SO3 4.35 2.98 3.03 5.79 7.30 16.15 11.75 8.97 Na2O 0.30 0.13 0.60 0.53 0.19 0.66 0.26 0.32 K2O 1.01 0.69 3.35 4.81 3.20 4.47 4.83 3.66 Na2Oea 0.97 0.58 2.80 3.69 2.30 3.60 3.43 2.73 Na2O soluble 0.16 0.06 0.38 0.30 0.10 0.39 0.17 0.14 K2O soluble 0.97 0.64 2.66 3.88 1.97 2.9 3.94 2.42 Na2Oeb soluble 0.80 0.49 2.13 2.85 1.40 2.30 2.76 1.73 Na2Oeb / Na2Oea 0.82 0.84 0.76 0.77 0.61 0.64 0.80 0.64 Ti2O 0.25 0.26 0.40 0.55 0.21 0.26 0.24 0.19 P2O5 0.26 0.12 0.12 0.11 0.04 0.15 0.11 0.09 Mn2O 0.09 0.56 0.06 0.06 0.05 0.08 0.50 0.06 Cl 0.00 0.00 2.49 0.94 0.38 0.35 2.18 0.85 LOIc 1.79 1.28 28.74 17.85 23.76 9.96 5.88 5.47 Total Sum (High)d 99.92 99.88 104.95 102.15 100.84 100.83 105.07 101.27 Total Sum (Real)e 99.92 99.88 99.08 99.79 99.73 100.02 99.74 99.34 Free Limef 0.70 1.53 4.50 4.04 5.70 10.59 29.20 38.20 a Equivalent Alkali (Na2O + 0.658 K2O) b Equivalent Water Sol. Alkali (Water Sol. Na2O + 0.658 Water Sol. K2O) c Loss on ignition determined at 950 ± 50 ºC d XRF sum of total oxides e Sum of total oxides calculated by removing the volatiles that are included in the LOI (Na2O, K2O, Cl-) f Free lime: combined CaO & Ca(OH)2 content

LS 53.49 2.60 0.68 0.21 0.55 0.03 0.02 0.21 0.16 0.00 0.00 0.00 0.00 0.03 0.01 0.01 0.00 42.29 100.15 100.15 0.00

SLX 0.02 98.15 0.47 0.06 0.00 0.03 0.01 0.08 0.06 0.00 0.00 0.00 0.00 0.03 0.01 0.00 0.00 0.20 99.06 99.06 0.00

Each CKD has its own characteristics, but there can be some generalization of these particular CKDs based upon the pyroprocess, especially in free lime and chloride contents. As expected, the wet and long-dry kilns had free lime contents that are lower than the precalciner kilns. CKDs D* and D have higher free limes than typical long-dry kiln CKDs due to unique equipment design in the kiln, but they are still lower than the precalciner CKDs E and F free limes. The long-dry kiln CKDs have low chloride and high sulfate contents in comparison to the wet and precalciner CKDs.

142

The CKDs were generally higher in total alkali, sulfate, chloride, LOI, and free lime than Cements TI and TII, as shown in Table 4.1. Water soluble alkalis are not normally reported for PCs, although the test method is described in ASTM C114. The CKDs contained higher levels of water soluble alkalis than the Cements TI and TII. It is interesting to note that although the quantity of soluble alkalis is higher in CKDs, the ratio of water soluble alkalis to total alkalis is higher in Cements TI and TII.

Statements/Observations:

4.i

The ASTM C114 techniques specified for PC chemical analysis are not necessarily sufficient and/or appropriate for CKD chemical analysis. The mass of CKD at 950 ± 50 ˚C is not stable until 12 – 24 hours. Therefore, the 1-hour PC standard LOI test duration is not sufficient to determine LOI for CKDs. Further, LOI and fused bead preparation of CKDs can cause the volatile compounds to be released into the atmosphere prior to chemical composition analysis. Babikan and Verville (2007) recommend using the following tests in ASTM C114 to determine the chemical composition of CKD volatile elements: i. Alkalis: flame photometry ii. Sulfates: induction furnace iii. Chloride: potentiometric titration

4.ii

Although PC typically only contains free calcium oxide, the PC free lime test is representative of both free calcium oxide and calcium hydroxide. CKDs are more likely to contain calcium hydroxide than PC due to exposure to moisture during handling and storage.

143

4.1.2

Mineralogical Properties

CKD mineralogical analysis (determination of the relative abundance of the different phases) is an essential complement of the chemical analysis. The effects of CKD elements in a CKD-PC blend may vary depending on the form in which they actually exist. The characteristics of CKD are traditionally evaluated based on chemical analysis data. Such data does not, however, indicate the ways in which the different elements actually exist within the CKD and how they might be expected to react during hydration. Soluble alkalis, for example, may occur as separate crystalline phases in the form of alkali chlorides or alkali sulfates. The reactivity of elements may, therefore, be expected to vary, depending on the form in which they actually exist.

The traditional methods (Bogue equations, XRD Rietveld analysis, and thermal analysis) were used to assess the PC mineralogical compositions. Although quantifying the mineralogical composition of PC has been thoroughly explored, the data to quantify the mineralogical phases of CKDs is relatively limited. Mineralogical analysis of CKDs has not been thoroughly evaluated due to a lack of quantitative analytical techniques. A method for mineralogical phase quantification of CKDs using XRD diffraction scans, Rietveld refinement, and physical tests (thermal analysis and titration) is introduced in this section.

144

Rietveld analyses of Cements TI and TII were performed using the X-ray diffraction scans and control files developed “in-house” at Lafarge North America. The mineralogical compositions of the PCs were determined by Rietveld quantitative X-ray diffraction analysis, shown in Table 4.3(a). Alite (impure C3S) typically contains 3 – 4% of substituent oxides, the most significant of which are iron, magnesium, and aluminum. Belite (impure C2S) may contain 4 – 6% of substituent oxides of which aluminum and iron are most common (Taylor, 1997). The potential proportions of C3S, C2S, C3A, and C4AF compounds in each PC, calculated based on the Bogue equations in ASTM C150, are shown in Table 4.3b. Taylor (1997) has noted that Bogue calculations can differ considerably from the true phase compositions, especially by underestimation of alite and overestimation of belite because the actual composition of these phases differs considerably from those of the pure form.

Table 4.3 Cements TI and TII mineralogical composition (mass %)

(a) XRD Rietveld Analysis Phase Alite, C3S Belite, C2S Aluminate, C3A Ferrite, C4AF Lime, CaO Periclase, MgO Gypsum, CaSO4·2H2O Hemihydrate, CaSO4·0.5H2O Anhydrite, CaSO4 Calcite, CaCO3 Portlandite, Ca(OH)2 Quartz, SiO2

(b) Bogue Compound Calculation TI 68.6 10.3 8.7 7.5 0.0 1.4 1.7 0.5 0.2 0.7 0.1 0.3

TII 66.5 15.2 3.0 8.9 0.2 2.5 1.0 0.6 0.9 0.8 0.4 0.2

Phase C3S C2S C3A C4AF

145

TI 51.9 15.8 11.3 7.5

TII 60.7 12.6 6.1 9.1

Cement TI has considerably more gypsum (readily soluble calcium sulfate) than Cement TII. PC with high aluminate contents typically require a sufficient amount of added calcium sulfate as a set controlling agent, which increases the sulfate content of the PC. It follows that cements low in aluminate require less added calcium sulfate and would tend to have lower sulfate contents.

Thermogravimetric analysis (TGA) is ideally suited to quantify the degree of calcination and amount of calcium hydroxide (portlandite) present in CKDs. Samples were tested in a temperature range from 30˚C to 950˚C. Portlandite decomposed between 400 to 530˚C and calcium carbonate was detected between 700 to 850˚C. The TGA results for Cements TI and TII and the CKDs are presented in Appendix B. An approximate determination of portlandite (Ca(OH)2) was established using mass balance calculations and the TGA results. The portlandite was then subtracted from the total free lime (calcium oxide and calcium hydroxide) in Table 4.2 (based on equivalent calcium oxide) to determine the free calcium oxide (CaO) portion. The mineralogical composition of calcite, portlandite, and free calcium oxide is shown in Table 4.4.

Table 4.4 CKD mineralogical compositions using direct test methods (mass %)

Components CaCO3 free CaO Ca(OH)2

A 52.6 4.5 0.0

Cement Kiln Dusts (CKDs) B C D* D E 34.7 51.3 17.3 22.7 1.8 4.0 5.7 18.2 10.6 28.4 0.0 0.0 0.0 0.0 1.0

146

F 6.5 34.5 4.9

Rietveld XRD analysis was used to accurately estimate the overall mineralogical phases quantification of each CKD. The CKD crystal phases were identified using XRD scans and Joint Committee on Powder Diffraction Standards (JCPDS) files. CKDs typically contain some phases that cannot be observed by XRD. Clay minerals, glass, and similar poorly crystalline components fall into this class and are called amorphous (without form or crystal structure). Different types of amorphous materials, however, may provide an indication of their presence as broad patterns or “humps”. X-ray powder diffraction is only sensitive to crystalline materials.

During normal Rietveld analysis, the amorphous component of a sample is not considered and the relative mass fractions of the crystalline phases are normalized to 100%. This was corrected using a known amount of either calcite or free lime determined by TGA, as shown in Table 4.4. In deciding which phase quantity to use, the best fit results were achieved using the most abundant phase for each respective CKD Rietveld Refinement. Therefore, CKDs A, B, C, and D were quantified relative to their calcite value, and CKDs D*, E and F were quantified relative to their free calcium oxide content. Adding the known mass of the phase to the Rietveld analysis allowed the amorphous phase to be incorporated in the analysis. Furthermore, absolute mass fractions were obtained for all phases. The results of the quantitative phase analyses using TGA, XRD, and Rietveld are shown in Table 4.5. The XRD scans are presented in Appendix C.

Calcite was identified as the major phase for CKDs A (52.6%), B (34.7%), C (51.3%), and D (22.7%), which are from the wet and long-dry processes. Free lime is the dominant phase in CKDs D* (18.2%), E (28.4%), and F (34.5%). CKDs E and F are from precalciners, while CKDs D and D* have uncharacteristically high free limes for a longdry kiln process. Quartz was present in all CKDs in a range from 3 – 11%. Periclase was present at approximately 2% for CKDs E and F, but less than 1% for the other CKDs. CKDs A and B contained minor amounts (≤5%) of dolomite. CKDs E and F had minor amounts of portlandite, while CKD B had only trace amounts. CKDs E and F are from

147

precalciner kilns that are equipped with water conditioning towers to reduce CKD fugitive dust. The water from the conditioning tower converts a portion of the free lime to calcium hydroxide.

Table 4.5 Mineralogical composition of CKD and filler materials (mass %)

Category

Phase

Wet CKD A CKD B

Cement Kiln Dust Long-dry CKD C CKD D* CKD D

15 Calcite, CaCO3 52.6 34.7 51.3 Quartz, SiO2 9 7 11 4 Raw Dolomite, CaMgCO3 5 3 Feed Periclase, MgO