Concrete Permeability and Durability Performance From Theory To

@seismicisolation @seismicisolation

Concrete Permeability and Durability P erformance

@seismicisolation @seismicisolation

Modern Concrete Technology Series A series of books presenting the state-of-the-art in concrete technology. Series Editors Arnon Bentur National Building Research Institute Faculty of Civil and Environmental Engineering Technion-Israel Institute of Technology Technion City, Haifa 32000 Israel

Sidney Mindess Department of Civil Engineering University of British Columbia 6250 Applied Science Lane Vancouver, B.C. V6T 1Z4 Canada

17. Sustainability of Concrete P. C. Aїtcin and S. Mindess

18. Concrete Surface Engineering B Bissonnette, L Courard and A Garbacz

19. Textile Reinforced Concrete A. Peled, A Bentur and B Mobasher

20. Durability of Concrete: Design and Construction M.G. Alexander, A. Bentur and S Mindess

21. Ultra High Performance FRCs E. Denarié

22. Shotcrete: Materials, Performance and Use M. Jolin and D.R. Morgan

23. Concrete Permeability and Durability Performance: From Theory to Field Applications R.J. Torrent, R.D. Neves and K. Imamoto

For more information about this series, please visit: https:// www.routledge.com/series-title/book-series/MCT

@seismicisolation @seismicisolation

Concrete Permeability and Durability P erformance From Theory to Field Applications

Roberto J. Torrent Rui D. Neves Kei-ichi Imamoto

@seismicisolation @seismicisolation

First edition published 2022 by CRC Press 6000 Broken Sound Parkway NW, Suite 300, Boca Raton, FL 33487-2742 and by CRC Press 2 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN © 2022 Taylor & Francis Group, LLC CRC Press is an imprint of Taylor & Francis Group, LLC Reasonable efforts have been made to publish reliable data and information, but the authors and publisher cannot assume responsibility for the validity of all materials or the consequences of their use. The authors and publishers have attempted to trace the copyright holders of all material reproduced in this publication and apologize to copyright holders if permission to publish in this form has not been obtained. If any copyright material has not been acknowledged please write and let us know so we may rectify in any future reprint. Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any information storage or retrieval system, without written permission from the publishers. For permission to photocopy or use material electronically from this work, access www. copyright.com or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. For works that are not available on CCC please contact [emailprotected] Trademark notice: Product or corporate names may be trademarks or registered trademarks and are used only for identification and explanation without intent to infringe. Library of Congress Cataloging‑in‑Publication Data Names: Torrent, R., author. | Neves, Rui D., author. | Imamoto, Keiichi, 1966- author. Title: Concrete permeability and durability performance : from theory to field applications / Roberto J. Torrent, Rui D. Neves, Kei-ichi Imamoto. Description: First edition. | Boca Raton, FL : CRC Press, 2022. | Series: Modern concrete technology, 1746-2959 ; 21 | Includes bibliographical references and index. Identifiers: LCCN 2021029833 (print) | LCCN 2021029834 (ebook) | ISBN 9781138584884 (hbk) | ISBN 9781032039701 (pbk) | ISBN 9780429505652 (ebk) Subjects: LCSH: Concrete—Testing. | Concrete—Permeability. | Concrete—Deterioration. | Concrete—Service life. Classification: LCC TA440 .T57 2022 (print) | LCC TA440 (ebook) | DDC 620.1/360287—dc23 LC record available at https://lccn.loc.gov/2021029833 LC ebook record available at https://lccn.loc.gov/2021029834 ISBN: 978-1-138-58488-4 (hbk) ISBN: 978-1-032-03970-1 (pbk) ISBN: 978-0-429-50565-2 (ebk) DOI: 10.1201/9780429505652 Typeset in Sabon by codeMantra

@seismicisolation @seismicisolation

Contents

xix xxi xxv xxvii

Foreword Preface Acknowledgements Authors 1 Durability performance of concrete structures 1.1 1.2

1.3 1.4 1.5 1.6

1.7

What Is Durability? 1 Deterioration Mechanisms of Concrete Structures 1 1.2.1 Carbonation-Induced Steel Corrosion 2 1.2.2 Chloride-Induced Steel Corrosion 2 1.2.3 External Sulphate Attack 3 1.2.4 Alkali-Silica Reaction 3 1.2.5 Freezing and Thawing 4 Deterioration Process of Concrete Structures 5 The Costs of Lack of Durability 7 Economical, Ecological and Social Impacts of Durability 8 Durability Design: The Classical Prescriptive Approach 9 1.6.1 Compressive Strength as Durability Indicator 10 1.6.2 Water/Cement Ratio as Durability Indicator 12 1.6.3 Cement Content as Durability Indicator 14 1.6.4 Cover Thickness as Durability Indicator 14 Durability Design: The Performance Approach 15 1.7.1 The “Durability Test” Question 15 1.7.2 Canadian Standards 16 1.7.3 Argentine and Spanish Codes 16 1.7.4 Japanese Architectural Code 17 1.7.5 Portuguese Standards 18 1.7.6 South African Standards 19 1.7.7 Swiss Standards 19 v

@seismicisolation @seismicisolation

Contents

1.8 Concrete Permeability as “Durability Indicator” 21 1.9 Beyond 50 Years: Modelling 22 References 22

2 Permeability as key concrete property

2.1 2.2

Foundations of Permeation Laws 27 Relation between Permeability and Pore Structure of Concrete 28 2.3 Permeability as Key Concrete Property 28 2.3.1 Permeability for Liquids’ Containment 29 2.3.1.1 ACI Low Permeability Concrete 29 2.3.1.2 Dams 29 2.3.1.3 Pervious Concrete 30 2.3.1.4 Liquid Gas Containers 31 2.3.2 Permeability for Gas Containment 32 2.3.2.1 Evacuated Tunnels for High-Speed Trains 32 2.3.2.2 Underground Gas “Batteries” 32 2.3.3 Permeability for Radiation Containment 33 2.3.3.1 Radon Gas 33 2.3.3.2 Nuclear Waste Disposal Containers 34 2.4 Permeability and Durability 36 References 37

3 Theory: concrete microstructure and transport of matter 3.1

3.2

3.3 3.4

3.5

Cement Hydration 41 3.1.1 Main Hydration Reactions and Resulting Changes 41 3.1.2 Hydrothermal Conditions for Hydration (Curing) 42 Microstructure of Hardened Concrete 43 3.2.1 Overview 43 3.2.2 Microstructure of Hardened Cement Paste 45 3.2.3 Interfacial Transition Zone 48 3.2.4 Pore Structure of Hardened Concrete 49 3.2.5 Binding 51 Water in the Pores of Hardened Concrete 51 Mechanisms of Transport of Matter through Concrete 52 3.4.1 Diffusion: Fick’s Laws 52 3.4.2 Migration: Nernst-Planck Equation 54 Permeability 56 3.5.1 Laminar Flow of Newtonian Fluids. Hagen-Poiseuille Law 56 3.5.2 Water-Permeability: Darcy’s Law 59 @seismicisolation @seismicisolation

Contents vii

3.5.3 3.5.4 3.5.5

Permeation of Liquids through Cracks 60 Hagen-Poiseuille-Darcy Law for Gases 60 Relation between Permeability to Gases and Liquids 61 3.6 Knudsen and Molecular Gas Flow: Klinkenberg Effect 62 3.7 Capillary Suction and Water Vapour Diffusion 67 3.7.1 Capillary Suction: A Special Case of Water-Permeability 67 3.7.2 Water Vapour Diffusion 69 3.8 Transport Parameters and Pore Structure 70 3.8.1 Relationship between Transport Parameters and Pore Structure 70 3.8.2 Permeability Predictions: Theory vs Experiments 72 3.8.2.1 Gas- and Water-Permeability vs Pore Structure 72 3.8.2.2 Water Sorptivity vs Pore Structure 73 3.9 Theoretical Relationship between Transport Parameters 75 References 76

4 Test methods to measure permeability of concrete 81 4.1

4.2

4.3

Water-Permeability 81 4.1.1 Laboratory Water-Permeability Tests 82 4.1.1.1 Steady-State Flow Test 82 4.1.1.2 Non Steady-State Flow Test: Water-Penetration under Pressure 83 4.1.2 Site Water-Permeability Tests 85 4.1.2.1 Germann Test 85 4.1.2.2 Autoclam System 85 4.1.2.3 Field Water-Permeability 86 Sorptivity: Special Case of Water-Permeability 87 4.2.1 Laboratory Sorptivity Tests 88 4.2.2 Site Sorptivity Tests 91 4.2.2.1 ISAT 91 4.2.2.2 Karsten Tube 93 4.2.2.3 Figg 93 4.2.2.4 Autoclam System 94 4.2.2.5 SWAT 94 4.2.2.6 WIST 95 Gas-Permeability 96 4.3.1 Laboratory Gas-Permeability Test Methods 97 4.3.1.1 Influence of Moisture and the Need for Pre-Conditioning 97 @seismicisolation @seismicisolation

viii

Contents

4.3.1.2 4.3.1.3

Cembureau Gas-Permeability Test 99 South African OxygenPermeability Index Test 100 4.3.2 Site Gas-Permeability Test Methods 102 4.3.2.1 Figg 103 4.3.2.2 Hong-Parrott 104 4.3.2.3 Paulmann 105 4.3.2.4 TUD 105 4.3.2.5 GGT 106 4.3.2.6 Paulini 106 4.3.2.7 Autoclam System 108 4.3.2.8 Single-Chamber Vacuum Cell 109 4.3.2.9 Double-Chamber Vacuum Cell (Torrent) 110 4.3.2.10 Triple-Chamber Vacuum Cell (Kurashige) 110 4.3.2.11 Zia-Guth 111 4.3.2.12 “Seal” Method 111 4.3.3 Assessment of Concrete Quality by Gas-Permeability Test Methods 112 4.4 Comparative Test RILEM TC 189-NEC 112 4.4.1 Objective and Experiment Design 112 4.4.2 Evaluation of Test Results 113 4.4.2.1 Significance of Test Method 113 4.4.2.2 Correlation between Site and “Reference” Tests 116 4.4.2.3 Conclusions of the Comparative Test 116 Acknowledgements 117 References 117

5 Torrent NDT method for coefficient of air-permeability 5.1 5.2 5.3

Introduction: Why a Separate Chapter? 123 The Origin 123 Fundamentals of the Test Method 124 5.3.1 Principles of the Test Method 124 5.3.2 Historical Evolution 126 5.3.3 Operation of the Instrument 129 5.3.4 Model for the Calculation of the Coefficient of Air-Permeability kT 129 5.3.5 Relation between ΔP and √t 133 5.3.5.1 Theoretical Linear Response 133 @seismicisolation @seismicisolation

123

Contents

5.3.5.2

5.4 5.5

5.6

5.7

Lack of Linear Response: Possible Causes 135 5.3.6 Relation between L and kT. Thickness Correction 135 5.3.6.1 Relation between Test Penetration L and kT 135 5.3.6.2 Correction of kT for Thickness 137 Relevant Features of the Test Method 138 Interpretation of Test Results 139 5.5.1 Permeability Classes 139 5.5.2 Microstructural Interpretation 140 Repeatability and Reproducibility 141 5.6.1 Testing Variability: Repeatability 142 5.6.2 Within-Sample Variability 143 5.6.3 Global Variability 144 5.6.4 Reproducibility 145 5.6.4.1 Reproducibility for Same Brand 145 5.6.4.2 Reproducibility for Different Brands 148 Effects and Influences on kT 149 5.7.1 Influence of Temperature of Concrete Surface 150 5.7.1.1 Influence of Low Concrete Temperature 150 5.7.1.2 Influence of High Air Temperature and Solar Radiation 151 5.7.2 Influence of Moisture of Concrete Surface 151 5.7.2.1 Influence of Natural and Oven Drying on kT 154 5.7.2.2 Compensation of kT for Surface Moisture 157 5.7.2.3 Pre-conditioning of Laboratory Specimens for kT Measurements 160 5.7.3 Effect/Influence of Age on kT 161 5.7.3.1 Effect/Influence of Age on Young Concrete 162 5.7.3.2 Effect/Influence of Age on Mature Concrete 163 5.7.4 Influence of Vicinity of Steel Bars 165 5.7.5 Influence of the Conditions of the Surface Tested 167 5.7.5.1 Influence of Specimen Geometry and Surface 167 5.7.5.2 Influence of Curvature 168 5.7.5.3 Influence of Roughness 169 5.7.5.4 Effect/Influence of Surface Air-Bubbles 170 5.7.6 Influence of Initial Pressure P0 172 5.7.7 Influence of Porosity on the Recorded kT Value 172 @seismicisolation @seismicisolation

Contents

5.8

Statistical Evaluation of kT Test Results 173 5.8.1 Statistical Distribution of kT Results 173 5.8.2 Central Value and Scatter Statistical Parameters 174 5.8.2.1 Parametric Analysis 174 5.8.2.2 Non-Parametric Analysis 175 5.8.3 Interpretation and Presentation of Results 176 5.9 Testing Procedures for Measuring kT in the Laboratory and On Site 180 References 180

6 Effect of key technological factors on concrete permeability 185 6.1 6.2

6.3

6.4

Introduction 185 Effect of w/c Ratio and Compressive Strength on Concrete Permeability 186 6.2.1 Data Sources 186 6.2.1.1 HMC Laboratories 186 6.2.1.2 ETHZ Cubes 187 6.2.1.3 General Building Research Corporation of Japan 188 6.2.1.4 University of Cape Town 188 6.2.1.5 KEMA 188 6.2.1.6 Other 189 6.2.2 Effect of w/c Ratio and Strength on Gas-Permeability 189 6.2.2.1 Cembureau Test Method 189 6.2.2.2 OPI Test Method 191 6.2.2.3 Torrent kT Test Method 192 6.2.3 Effect of w/c Ratio on Water-Permeability 195 6.2.3.1 Water Penetration under Pressure 195 6.2.3.2 Water Sorptivity 196 Effect of Binder on Concrete Permeability 197 6.3.1 Effect of OPC Strength on Permeability 197 6.3.2 Effect of Binder Type on Permeability 199 6.3.2.1 “Conventional” Binders 199 6.3.2.2 “Unconventional” Binders 205 Effect of Aggregate on Concrete Permeability 209 6.4.1 Effect of Bulk Aggregate on Concrete Permeability 209 6.4.1.1 Porous Aggregates 209 6.4.1.2 Recycled Aggregates 210 6.4.1.3 Spherical Steel Slag Aggregates 213 6.4.2 Effect of ITZ on Concrete Permeability 214 @seismicisolation @seismicisolation

Contents

6.5

Effect of Special Constituents on Concrete Permeability 218 6.5.1 Pigments 219 6.5.2 Fibres 220 6.5.3 Polymers 222 6.5.4 Expansive Agents 223 6.6 Effect of Compaction, Segregation and Bleeding on Permeability 226 6.7 Effect of Curing on Permeability 233 6.7.1 Relevance of Curing for Concrete Quality 233 6.7.2 Effect of Curing on Permeability 234 6.7.2.1 Investigations in the Laboratory 234 6.7.2.2 Investigations in the Field 237 6.7.3 Effect of Curing on Air-Permeability kT 239 6.7.3.1 Conventional Curing 239 6.7.3.2 Self-Curing 243 6.7.3.3 Accelerated Curing 244 6.7.3.4 “3M-Sheets” Curing 246 6.8 Effect of Temperature on Permeability 247 6.9 Effect of Moisture on Permeability 252 6.10 Effect of Applied Stresses on Permeability 259 6.10.1 Effect of Compressive Stresses 259 6.10.2 Effect of Tensile Stresses 262 6.11 Permeability of Cracked Concrete 263 6.11.1 Permeability through Cracks: Theory 263 6.11.2 Effect of Cracks on Permeability 265 6.11.3 Self-Healing of Cracks and Permeability 270 References 275

7 Why durability needs to be assessed on site? 7.1

7.2

287

Theorecrete, Labcrete, Realcrete and Covercrete 287 7.1.1 Theorecrete 287 7.1.2 Labcrete 289 7.1.3 Realcrete 289 7.1.4 Covercrete 290 7.1.5 Quality Loss between Covercrete and Labcrete 292 7.1.5.1 Bözberg Tunnel 292 7.1.5.2 Schaffhausen Bridge 293 7.1.5.3 Lisbon Viaduct 296 7.1.5.4 Swiss Bridges’ Elements 297 Achieving High Covercrete’s Quality 299 7.2.1 Mix Design and Curing 299 @seismicisolation @seismicisolation

xii

Contents

7.2.2 7.2.3

UHPFRC 299 Controlled Permeable Formwork (CPF) Liners 301 7.2.3.1 Action Mechanism of CPF Liners 301 7.2.3.2 Impact of CPF on the “Penetrability” of the Covercrete 302 7.2.4 Shrinkage-Compensating Concrete 308 7.2.5 Self-Consolidating Concrete 308 7.2.6 Permeability-Reducing Agents 310 7.3 Cover Thickness 312 7.4 Spacers and Permeability 315 7.5 Concluding Remarks 316 References 317

8 Why air-permeability kT as durability indicator? 8.1 8.2 8.3

321

Introduction 321 Response of kT to Changes in Key Technological Parameters of Concrete 322 Correlation with Other Durability Tests 323 8.3.1 Gas Permeability 324 8.3.1.1 Cembureau Test 324 8.3.1.2 South-African OPI 330 8.3.1.3 Figg Air and TUD Permeability 331 8.3.2 Oxygen-Diffusivity 332 8.3.3 Capillary Suction 332 8.3.3.1 Coefficient of Water Absorption at 24 Hours 332 8.3.3.2 Figg Water 333 8.3.3.3 Karsten Tube 333 8.3.4 Water-Permeability and Penetration under Pressure 334 8.3.5 Migration 334 8.3.5.1 Rapid Chloride Permeability Test (“RCPT” ASTM C1202) 335 8.3.5.2 Coefficient of Chloride Migration (NT Build 492) 335 8.3.5.3 Electrical Resistivity (Wenner Method) 336 8.3.5.4 South African Chloride Conductivity Index 337 8.3.6 Chloride-Diffusion 338 8.3.6.1 Laboratory Diffusion Tests 338 8.3.6.2 Site Chloride Ingress in Old Structures 340 @seismicisolation @seismicisolation

Contents

xiii

8.3.7

Carbonation 340 8.3.7.1 Laboratory Tests (Natural Carbonation) 340 8.3.7.2 Laboratory Tests (Accelerated Carbonation) 341 8.3.7.3 Site Carbonation in Old Structures 343 8.3.8 Frost Resistance 344 8.4 Some Negative Experiences 348 8.4.1 Tunnel in Aargau, Switzerland 348 8.4.2 Wotruba Church, Vienna, Austria 349 8.4.3 Ministry of Transport, Ontario, Canada 350 8.4.4 Mansei Bridge, Aomori, Japan 351 8.4.5 Tests at FDOT Laboratory 352 8.5 Air-Permeability kT in Standards and Specifications 352 8.5.1 Swiss Standards 352 8.5.2 Argentina 354 8.5.3 Chile 355 8.5.4 China 355 8.5.5 India 355 8.5.6 Japan 355 8.6 Credentials of Air-Permeability kT as Durability Indicator 355 References 356

9 Service life assessment based on site permeability tests 9.1 9.2

9.3

9.4

Introduction 361 General Principles of Corrosion Initiation Time Assessment 364 9.2.1 Carbonation-Induced Steel Corrosion 364 9.2.2 Chloride-Induced Steel Corrosion 368 Service Life Assessment of New Structures with Site Permeability Tests 370 9.3.1 Carbonation: Parrott’s Model 370 9.3.2 Carbonation: South African OPI Model 371 9.3.2.1 “Deemed-to-Satisfy” Approach 371 9.3.2.2 “Rigorous” Approach 372 9.3.2.3 Acceptance Criteria 372 9.3.2.4 Probabilistic Treatment 373 9.3.3 “Seal” Method for ChlorideInduced Steel Corrosion 373 Service Life Assessment of New Structures Applying Site kT Tests 373 @seismicisolation @seismicisolation

361

xiv Contents

9.4.1

The “TransChlor” Model for ChlorideInduced Steel Corrosion 373 9.4.2 Kurashige and Hironaga’s Model for Carbonation-Induced Steel Corrosion 377 9.4.3 The “Exp-Ref” Method: Principles 379 9.4.3.1 The “Exp-Ref” Method for Chloride-Induced Steel Corrosion 381 9.4.3.2 The “Exp-Ref” Method for Carbonation-Induced Steel Corrosion 383 9.4.3.3 The CTK “Cycle” Approach 387 9.4.4 Belgacem et al.’s Model for CarbonationInduced Steel Corrosion 389 9.5 Service Life Assessment of Existing Structures Applying Site kT Tests 390 9.5.1 Calibration with Drilled Cores 391 9.5.2 Pure Non-destructive Approach 392 References 395

10 The role of permeability in explosive spalling under fire 399 10.1 Effect of Fire on Reinforced Concrete Structures 399 10.2 Explosive Spalling of Concrete Cover 400 10.3 The Role of Concrete Permeability in Explosive Spalling 402 10.4 Coping with HSC 403 10.5 Concluding Remarks 407 References 408

11 Real cases of kT test applications on site 411 11.1 Introduction 411 11.2 Full-Scale Investigations 411 11.2.1 RILEM TC 230-PSC (Chlorides and Carbonation) 411 11.2.2 Naxberg Tunnel (Chlorides and Carbonation) 415 11.2.2.1 Scope of the Investigation 415 11.2.2.2 Mixes Composition and Laboratory Test Results 416 11.2.2.3 Characteristics of the 32 Panels 418 11.2.2.4 On-Site Non-Destructive kT Measurements 418 11.2.2.5 Core Drilling, Carbonation and Chloride Ingress 420

@seismicisolation @seismicisolation

Contents

11.2.2.6 Conclusions 423 11.3 New Structures 423 11.3.1 Port of Miami Tunnel (Carbonation) 423 11.3.1.1 Description of the Tunnel 423 11.3.1.2 The Problem 424 11.3.1.3 Scope of the Investigation 426 11.3.1.4 Site kT Test Results 426 11.3.1.5 Modelling Carbonation at 150 Years 427 11.3.1.6 Conclusions 429 11.3.2 Hong Kong-Zhuhai-Macao Link (Chlorides) 430 11.3.3 Panama Canal Expansion (Chlorides) 434 11.3.4 Precast Coastal Defence Elements (Sulphates) 438 11.3.4.1 Aggressiveness of the Water 439 11.3.4.2 Durability Requirements 441 11.3.4.3 Concrete Mix Quality Compliance 441 11.3.4.4 Precast Elements’ Compliance 442 11.3.4.5 Conclusions on the Durability of the Elements 445 11.3.5 Buenos Aires Metro (Water-Tightness) 445 11.3.6 HPSFRC in Italy (Water-Tightness) 448 11.3.6.1 Description of the Case 448 11.3.6.2 Characteristics of the Concretes Used for the Different Elements 450 11.3.6.3 Air-Permeability kT Tests Performed 450 11.3.6.4 Performance of SCC-SFRC Elements 451 11.3.6.5 Performance of Walls 452 11.3.6.6 Performance of Precast Columns 452 11.3.6.7 Conclusions 453 11.3.7 UHPFRC in Switzerland (Chlorides) 453 11.3.8 Field Tests on Swiss New Structures 456 11.3.9 Field Tests on Portuguese New Structures 456 11.3.9.1 Bridge at the North of Lisbon (Quality Control/Carbonation) 456 11.3.9.2 Urban Viaduct in Lisbon (Quality Control) 458 11.3.9.3 Sewage Treatment Plant (Chemical Attack) 460 11.3.10 Delamination of Industrial Floors in Argentina (“Defects” Detection) 461 11.4 Old Structures 462 11.4.1 Old Structures in Japan 462

@seismicisolation @seismicisolation

xvi

Contents

11.4.1.1 Tokyo’s National Museum of Western Art (Carbonation) 463 11.4.1.2 Jyugou Bridge (Condition Assessment) 465 11.4.1.3 Other Japanese Structures (Condition Assessment) 466 11.4.2 Old (and New) Swiss Structures (Chlorides+Carbonation) 467 11.4.2.1 Investigated Structures and Tests Performed 467 11.4.2.2 Combined Analysis of Results 469 11.4.2.3 Conclusions of the Investigations 471 11.4.3 Permeability and Condition of Concrete Structures in the Antarctic 472 11.4.3.1 The “Carlini” Base 472 11.4.3.2 The Climate 473 11.4.3.3 Buildings Construction and Exposure 473 11.4.3.4 Scope of the Investigation 474 11.4.3.5 Identified Pathologies 475 11.4.3.6 On-Site Measurements of Air-Permeability kT 475 11.4.4 Permeability of a Concrete Structure in the Chilean Atacama Desert 477 11.5 Unconventional Applications 479 11.5.1 Concrete Wine Vessels 479 11.5.2 Rocks and Stones 483 11.5.2.1 Permeability of Stones as Building Material 483 11.5.2.2 Permeability of Rocks for Oil and Gas Exploitation 485 11.5.2.3 Permeability of Rocks for Nuclear Waste Disposal 487 11.5.3 Timber 488 11.5.4 Ceramics 490 References 491

12 Epilogue: the future 12.1 12.2 12.3 12.4 12.5

499

Chapter 1: Durability 499 Chapter 2: Permeability 501 Chapter 3: Microstructure and Transport Theories 502 Chapter 4: Permeability Test Methods 502 Chapter 5: kT Air-Permeability Test Method 503 @seismicisolation @seismicisolation

Contents

xvii

12.6 Chapter 6: Factors Influencing Concrete Permeability 503 12.7 Chapter 7: Theorecrete, Labcrete, Realcrete and Covercrete 504 12.8 Chapter 8: kT Air-Permeability as Durability Indicator 505 12.9 Chapter 9: Modelling Based on Site Permeability Tests 505 12.10 Chapter 10: Gas Permeability and Fire Protection 507 12.11 Chapter 11: Applications of Air-Permeability kT Tests 507 References 508

Annex A: Transport test methods other than permeability 511 Annex B: Model standard for measuring the coefficient ofair-permeability kT of hardened concrete 529 Index 543

@seismicisolation @seismicisolation

Foreword

Concrete as such is a very durable material. There are magnificent examples of concrete structures which have survived 2,000 years without substantial repair measures, and they will survive hundreds of years to come. The Pantheon in Rome and numerous bridges built during the Roman Empire in Italy and Spain that served up to 2,000 years are well-known examples. Since the large-scale application of reinforced concrete, the construction industry has experienced enormous challenges with respect to achieving the designed service life of concrete structures. According to most standards, reinforced concrete structures are expected to have a service life of at least 100 years. In reality, however, expensive repair and renovation are frequently necessary after not more than 30 years. In recent years, a number of bridges collapsed after less than 50 years. It is estimated that repair of a damaged bridge costs approximately six to eight times more than that of the construction of a new bridge. During repair operation or reconstruction process, the necessary deviation of traffic alone causes additional financial and environmental burdens. Therefore, one major subject in concrete technology research has been to increase repair-free service life of reinforced concrete structures. Concrete is a porous material with a wide-range distribution of the size of pores, running from a few millimetres down to nanometres. The surface of concrete structures is usually in contact with changing climatic conditions. During a wet period, rain water will be absorbed by capillary action and in humid environment by capillary condensation. The micropores remain water-filled even during dry periods. The humidity in the pore system will initiate corrosion of the steel reinforcement as soon as the carbonation depth exceeds the cover thickness. Another disadvantage of reinforced concrete elements exposed to natural environment is the crack formation due to bending or temperature and humidity gradients. These cracks are preferential pathways for locally deep carbonation and hence early beginning of corrosion of reinforcement. Service life of reinforced concrete structures depends essentially on the cover thickness and on the permeability of the concrete cover. It is comparatively easy to determine the thickness of the concrete cover. Permeability xix

@seismicisolation @seismicisolation

xx Foreword

of the cover, however, is a more complex property. Based on the research findings and experience from practice, the present volume presents various topics related to permeability of concrete. A method to determine permeability of concrete is described in detail, and many possible applications in practice are discussed here. It can be expected that this volume will contribute to our knowledge on how to increase service life of reinforced concrete structures, and the discussions on durability and service life will certainly bring broader awareness of the implications of permeability of concrete. Durability and service life of reinforced concrete structures, however, do not depend on one dominating parameter. This was shown in a convincing way in a recent publication of RILEM Technical Committee (RILEM TC 246-TDC). Results of this Technical Committee have clearly demonstrated that durability depends on the combination of environmental actions and mechanical load. This volume is an excellent basis for a better understanding of dominating processes which may substantially reduce service life of concrete structures and of steel reinforced concrete structures. Prof. Dr. Dr. h.c. Folker H. Wittmann

@seismicisolation @seismicisolation

Preface

The genesis of this book originated on August 29, 2016, with a proposal of Prof. Neves to Dr. Torrent on the possibility of writing jointly a book about the permeability of concrete. After some consideration, the proposal was accepted, ending in a first draft of its possible content. Then, finding a suitable interested publisher was required. Believe it or not, on June 1, 2017, an invitation by Tony Moore (Senior Editor of CRC) arrived, asking Dr. Torrent about his willingness to write a book on permeability testing, on advice of Profs. S. Mindess and A. Bentur. This was a fortunate coincidence or superb intelligence services of CRC Press in act… The offer was accepted and, immediately, Prof. Imamoto was invited to join the authors’ team, invitation he accepted on July 29, 2017, during an unforgettable exquisite dinner in a small, special sushi restaurant near Tokyo’s Narita Airport, agreement possibly helped by a considerable dose of excellent cold sake… Regarding the subject of this book, it is good to recall that until the early 1980s, the main research efforts on hardened concrete properties were predominantly focused on its mechanical and viscoelastic properties, required for the structural design of reinforced concrete constructions. Since then, a considerable interest arose on durability issues, both in understanding the deterioration mechanisms and in developing suitable test methods and, more recently, in modelling the durability performance of concrete structures. A quantum leap was made by the work of RILEM TC 116-PCD “Permeability of concrete as a criterion for its durability”, chaired by Profs. H.K. Hilsdorf and J. Kropp, that stressed the importance of transport mechanisms, chiefly permeation, on the durability of concrete structures. The results of this work were condensed in a State-of-the-Art Report (RILEM Report 12), published in 1999. During the ensuing 20 years, several test methods for measuring the permeability of concrete to gases and liquids have been developed and a formidable amount of information has been produced through their application in the laboratory and on site. Part of it was included in RILEM Report 40 (2007) and RILEM State-of-the-Art Report v18 (2016), condensing the work of RILEM TCs 189-NEC and 230-PSC, respectively. xxi

@seismicisolation @seismicisolation

xxii Preface

It is the purpose of this book to present the existing knowledge on the permeability of concrete in a consolidated form, describing the available test methods and the effect key technological parameters of concrete have on the measured permeability. It presents a large amount of experimental data from investigations performed on laboratory specimens and full-scale elements and also from real cases of site permeability testing, conducted to solve complex and challenging concrete construction issues (durability, water-tightness, defects, spalling under fire, condition assessment, etc.). The three authors combine a formidable experience, covering over 30 years of research and testing the permeability of concrete in the lab and on site (they have conducted, with their own hands, permeability tests applying 13 different methods). Thanks to their geographical diversity, they have been active in relevant technical activities in Europe, the Americas, Africa and Asia, thus gaining a good insight into the global situation regarding permeability and durability testing and service life assessment of concrete structures. This book places a special emphasis on one test method (called kT), developed by Dr. Torrent around 1990, that was included in the Swiss Standards in 2003 under the title ‘Air-Permeability on Site’, with successive updates in 2013 and, recently, in 2019. The credit for this inclusion lies mainly on the initiatives and research work of Prof. E. Brühwiler and Dr. E. Denarié (EPFL, Lausanne), of Dr. F. Jacobs (TFB, Wildegg) and Dr. T. Teruzzi (SUPSI, Lugano). Over 430 documents on the kT test method have been recorded to date, out of which some 90 were authored by at least one of this book’s authors. Following Chapter 1, summarizing the fundamentals of durability, the relevance of permeability as a key performance property of concrete is discussed in Chapter 2, already opening the field to the possible applications of its measurement. An understanding of concrete microstructure and of the laws that govern the flow of matter through concrete is considered as essential, aspects that are dealt with in detail in Chapter 3. This is followed by Chapter 4 in which 25 test methods to measure concrete permeability (including capillary suction) are described. Chapter 5 describes in detail the kT test method, its fundamentals and the effect of external influences on its results. Chapter 6 is concerned with the effect of key technological factors on the permeability of concrete to gases and water, tested by various methods. Chapter 7 reflects the strong conviction of the authors on the relevance ofsite permeability testing of the end-product to get a realistic assessment of the concrete quality, in particular of its surface layers (the Covercrete),of vital importance for the durability of reinforced concrete structures. Having been designed to that end, i.e. to measure the permeability of the Covercrete on site, Chapter 8 provides evidence on the suitability of the kT test as Durability Indicator, relating its results with other relevant transport

@seismicisolation @seismicisolation

Preface xxiii

properties (sorptivity, diffusion, migration) and with simulation tests (carbonation, freezing/thawing). The same as for any other test, especially when applied on site, the application of kT test has not been up to the expectations in a few cases, which are also presented in Chapter 8. Today, test results are often not enough for designers and owners, who want an assessment of the potential service life of new and existing structures. Chapter 9 presents different service life prediction models with the site permeability of the Covercrete as input, often accompanied by a nondestructive evaluation of its thickness. Chapter 10 presents the relatively new field linking the gas-permeability of concrete to the explosive spalling of the concrete cover during fires. Here, contrary to durability, a not too low permeability is desirable. Chapter 11 presents a comprehensive series of investigations conducted on site, on full-scale elements and real structures, new and old. Some applications not related to concrete structures are also included. At the end, in Chapter 12, we draw some conclusions on the present and future of permeability testing of concrete structures, needed developments and unexplored research fields. The book is complemented with Annexes that describe transport tests other than permeation, and a Model Standard on how to conduct kT tests in the field and in the laboratory. The reader is invited to accompany us along this fascinating voyage from the theory of mass transport in concrete to field applications of permeability testing…, fasten your seat belts!! The Authors

@seismicisolation @seismicisolation

Acknowledgements

R. Torrent: to my wife for her unconditional support, to my children (3) and grandchildren (5) for their love and to my masters in Argentina, late G.Burgoa and R. Kuguel, who arose in me the interest in concrete science and technology and guided me in my first professional steps on these fascinating disciplines. To late Prof. A.M. Neville and to Prof. F. Wittmann for giving me fundamental orientation at turning points of my career. R. Neves: to my family for enduring my absences. To Senior Researcher A. Gonçalves for changing my mindset from structural design to durability design. To Dr. J. Vinagre and Polytechnic Institute of Setúbal for considering investing in permeability testing equipment and providing me excellent conditions to carry out my research. To Prof. J. de Brito for his friendship, valuable teachings and endless support. K. Imamoto: to my laboratory students for performing a huge number of permeability tests. To Dr. Hiroshi Tamura, former head of Material Dept., General Building Research Corporation of Japan, for giving me a chance to start research on air-permeability of concrete cover. All: To so many researchers worldwide that contributed their experiences to the body of knowledge compiled in this book. A thorough literature research has been conducted which, by no means can be considered complete; we apologize to those researchers, the valuable work of whom might have been overlooked when writing this book.

xxv

@seismicisolation @seismicisolation

Authors

Dr. Roberto J. Torrent is a researcher, consultant and partner of Materials Advanced Services Ltd. He held positions at the National Institute of Industrial Technology and Portland Cement Institute (Argentina), as well as at Holcim Technology Ltd. (Switzerland). For 30 years, he has been directly involved in durability testing of a large variety of concretes, both in the lab and on site. In the 1990s, he invented the Torrent NDT Method for measuring air-permeability. He is a RILEM Honorary Member. Dr. Rui Neves was formerly a researcher at the National Laboratory for Civil Engineering (LNEC-Portugal). Currently he is Professor in the Structures and Geotechnics Division at Barreiro School of Technology, Polytechnic Institute of Setúbal, Portugal. His research efforts are mainly devoted to service life of reinforced concrete structures, with special emphasis on investigating and testing the permeability of concrete and rocks. He has carried out relevant consulting activity within the frame of concrete quality control, as well as inspection and appraisal of reinforced concrete structures. Dr. Kei-ichi Imamoto is a graduate of Tokyo University of Science, Japan. He performed research at Tokyu Construction Co. Ltd. for 9 years and is now Professor at Tokyo University of Science. He received the Young Researcher’s award from AIJ (Architectural Institute of Japan) in 2008, and prizes from Japan Society for Finishing Technology, Japan Concrete Institute and Suga Weathering Foundation. He is very active in durability testing and service life assessment of concrete structures.

xxvii

@seismicisolation @seismicisolation

Chapter 1

Durability performance of concrete structures

1.1 W HAT IS DURABILITY? Since the title of the book intimately associates permeability with concrete durability, it is worth discussing the latter in this initial chapter. A good definition of durability has been coined in Section 3.1 of Neville (2003), to which some addenda have been made, resulting in the following tentative definition: Durability of a given concrete structure, in its specific exposure environment, is its ability to perform its intended functions, i.e. to maintain its required strength and serviceability, during the specified or traditionally expected service life, without unplanned, extraordinary maintenance or repair efforts.

1.2 DETERIORATION MECHANISMS OF CONCRETE STRUCTURES Discussing in detail the deterioration mechanisms of concrete structures is beyond the scope of this book; yet, the main ones can be briefly enumerated: steel corrosion induced by carbonation or chlorides, chemical attack (typically by sulphates in the soil and ground water and by acids in sewage systems), Alkali-Silica Reaction (ASR) and frost in cold climates. All these deterioration mechanisms have two aspects in common: The transfer of mass takes place by three physical actions: permeation, diffusion and, to a lesser extent, also by migration (all three thoroughly discussed in Chapter 3) and happens through the interconnected network of pores within the microstructure of concrete (also discussed in Chapter 3). DOI: 10.1201/9780429505652-1 @seismicisolation @seismicisolation

2 Concrete Permeability and Durability Performance

A succinct analysis will be made in the following sections. For a deeper insight into the problem, the reader can refer to Mehta et al. (1992), Richardson (2002), Dyer (2014), Li (2016), Alexander et al. (2017) and, more specifically for the case of steel corrosion in concrete, to Bertolini etal. (2004), Böhni (2005), Gjørv (2014) and Alexander (2016).

1.2.1 C arbonation-Induced Steel Corrosion This case of deterioration is due to the penetration (by gas diffusion) of CO2 from the environment which, in the presence of moisture, reacts preferentially with the reaction product of cement hydration Ca(OH)2 to form CaCO3. From the durability point of view, the main consequence of this reaction is a sharp drop of the pH of the pore solution, displacing the thermodynamic equilibrium of the steel bar, from “passive” to “corrosion”. The subsequent corrosion rate is highly dependent on the moisture conditions (as is also the carbonation rate). According to Mehta et al. (1992), “only porous and permeable concrete products, made with low cement contents, high water/cement ratio (w/c), and inadequately moist-cured tend to suffer from serious carbonation”. A tight pore system and a sufficiently thick cover are the main defense strategies against this mechanism, although the cement type (especially the amount of carbonatable material) also plays a role (see Section 9.2.1).

1.2.2 C hloride-Induced Steel Corrosion This case of deterioration is due to the penetration (by mix modes) of chloride ions from salty solutions in permanent or sporadic contact with the structure. The situation is much more complex than carbonation, due to the overlapping of several physical phenomena taking place, as illustrated in Figure 1.1 for a concrete element in a marine environment (adapted from Hunkeler (2000)). Chlorides may penetrate by permeation, carried by the saline water solution either under a pressure head for deep parts of the structure or/and due to capillary suction in the critical areas subjected to wetting-drying cycles and, alternatively or complementary, by ion diffusion. Rain washout and evaporation, affecting predominantly the surface layers, add complication to the phenomenon. When the penetration of the front of a certain elusive critical Cl− concentration reaches the position of the steel, this is depassivated and metal corrosion may start. The same structure, placed by the sea, will deteriorate much earlier than if exposed to carbonation. According to Mehta et al. (1992), “The ingress of chlorides into hardened concrete is decisively dependent on and influenced by water transport mechanisms. Substantially greater amounts of chlorides may ingress into the hardened concrete via water transport mechanisms than via pure chloride ion diffusion”. @seismicisolation @seismicisolation

Durability of concrete structures 3

Figure 1.1 Complex combination of physical phenomena in the movement of Cl− in marine concrete, based on Hunkeler (2000).

1.2.3 E xternal Sulphate Attack This case of deterioration takes place when sulphate ions from the environment, typically from the soil or ground water, penetrate into concrete, developing deleterious physical–chemical interactions with some minerals in the hydrated cement paste. The main penetration mechanism is permeation of the SO42−-rich solution in the form of capillary suction, accompanied by internal redistribution by diffusion. Salt crystallization, combined with expansive reaction products (e.g. ettringite, gypsum and thaumasite), leads to cracking, loss of mass and/or disintegration of the concrete. There are cements that, due to their composition or performance, are considered as “Sulphate Resistant Cement”, although it would be more appropriate to talk of “Sulphate Resistant Concrete”, since not just the use of such cements is sufficient to guarantee the immunity of the concrete against sulphate attack. According to Mehta et al. (1992), “…it can be concluded that, for improved resistance to sulfate attack, a reduction in the porosity and consequently the coefficient of permeability, is more important than modifications in the chemistry of Portland cements”.

1.2.4 Alkali-Silica Reaction This case of deterioration takes place when aggregates containing certain reactive minerals (typically some forms of SiO2) in sufficient or pessimum quantities react, in the presence of moisture, with the alkali (Na+, K+) ions in the pore solution, developing expansive reactions. The reaction products, @seismicisolation @seismicisolation

4 Concrete Permeability and Durability Performance

Figure 1.2 ASR gel accommodated along ITZ.

again in the presence of moisture, take the form of an expansive gel which, depending on the circ*mstances may be innocuous or create enormous deformations of the structure (e.g. dams), cracks, loss of mass and even total disintegration of the concrete. The expansive gel is sometimes accommodated in microcracks, air voids or along the more porous Interfacial Transition Zone (ITZ), see Section 3.2.3, as shown by the UV light observation of a thin section in Figure 1.2 (Fernández Luco & Torrent, 2003). The main mechanism of transport of the gel within concrete is permeation, due to expansive pressure, across the system of existing pores and of cracks generated by the expansive action. The water required to feed the expansive reaction and to swell the ASR gel penetrates the concrete predominantly by permeation (capillary suction) and moves internally by diffusion. The main defense line against ASR is to avoid the usage of reactive aggregates but, when this is unavoidable, to keep the quantity of alkalis in the concrete sufficiently low (e.g. by using low-alkali cements) or by using adequate types and contents of pozzolanic materials (that compete with advantage in neutralizing the alkalis). In the case of ASR, the pore structure and permeability of the concrete play a secondary role.

1.2.5 Freezing and Thawing This case of deterioration takes place when concrete, with a high degree of water-saturation, is exposed to sub-zero temperatures; the saturating water penetrates typically by permeation (capillary suction). The water in the pores freezes, augmenting its volume by 8%, pushing the still unfrozen water along the capillaries, creating damaging pressure on their walls. Successive freeze–thaw cycles continue to accumulate this type of damage, causing scaling and spalling of the surface layers due to internal cracks,

@seismicisolation @seismicisolation

Durability of concrete structures 5

typically oriented parallel to the element’s surface. The water in the larger pores freezes at higher below-zero temperatures than that in smaller pores. The problem is aggravated if the liquid in the pores contains salts, as typically happens with de-icing compounds, sprayed in winter on roads. The best-known prevention measure to avoid the freeze–thaw damage is to entrain air bubbles, in quantities and sizes sufficient to relieve the expansion pressures. As discussed in Section 8.3.8, this should be accompanied by a sufficiently tight pore structure (hence the maximum w/c ratio typically specified for this case). There is some debate on whether high-strength concrete (HSC), with its reduced porosity and permeability, is resistant to freeze–thaw damage without air-entrainment. 1.3 DETERIORATION PROCESS OF CONCRETE STRUCTURES When the designer, with the help of codes and standards, defines/specifies the architectural details, the shape and dimensions of the structural elements, the amount, quality and position of the steel bars (including cover thickness) and the quality of the concrete (typically strength and resistance to certain aggressive media), he/she is defining an initial design quality (IDQ), see Figure 1.3 adapted from Beushausen (2014). It is being assumed that, starting with this IDQ, the inevitable degradation process the structure will undergo through the years will follow a certain expected performance such that, when the “traditionally expected” or Design Service Life (DSL) is reached, the structure will still perform at a level above a not very well defined Unacceptable Level of Deterioration (ULD). This is indicated by the full line in Figure 1.3. Regrettably, in too many cases the True Initial Quality (TIQ) achieved during construction is below that assumed during the design, due to lack of care and/or application of inadequate concrete practices by the Contractor, or to concrete mixes of insufficient quality, or due to lack of zeal of the Inspection or, usually, a combination thereof. Hence, the true decay process (dotted curved line) is faster and the True Service Life (TSL) is reached much earlier than specified or expected (DSL). This requires some Interventions (I1, I2) to restore the condition of the structure to an acceptable level, so as to finally reach the DSL. This is an expensive solution not only due to the usually high cost of the interventions themselves but also for the lost revenue if the operation of the facility has to be partially or totally interrupted (roads, bridges, tunnels, power stations, cement plants, etc.), not to mention the serious consequences for human lives caused by the deterioration itself (e.g. debris falling from a tall building) or by increased accidents rates caused by traffic restrictions.

@seismicisolation @seismicisolation

6 Concrete Permeability and Durability Performance

Figure 1.3 E xpected and true durability performance of concrete structures (Beushausen, 2014).

Most deterioration processes (steel corrosion, sulphate or chemical attack, frost damage, ASR, etc.) follow a similar pattern, illustrated in Figure 1.4, based on the model proposed by Tuutti (1982), later extended by Nilsson (2012) for steel corrosion. Initially, there is a period in which no visual damage of the structure is observable. Yet, in this period, some phenomena are taking place internally, such as penetration of the carbonation front or accumulation of enough chloride at the surface of the steel bars (or generation of enough ASR expansive gel), so as to initiate the visible deterioration process. Thisperiod

Figure 1.4 Tuutti’s model for steel corrosion (Tuutti, 1982), extended by Nilsson (2012).

@seismicisolation @seismicisolation

Durability of concrete structures 7

is defined here as “Incubation” period and the time at which the true damage process starts is called “Initiation” time. At a certain “Initiation” time, the carbonation front (XCO2) or the penetration of the critical chloride content front (XCl) has reached the surface of the rebars (see bottom left corner of Figure 1.4), depassivating the steel which under unfavourable conditions will start to corrode. The expansive nature of the corrosion products will produce isolated rust stains and microcracks (which can be considered as localized damage in Figure 1.4), to be followed by spalling of the concrete cover and reduction of the cross section of the steel (generalized damage, see Figure 1.4). This process, if not checked, will lead to a loss of bearing capacity of the element that eventually will reach its Ultimate Limit State (ULS), requiring major retrofitting or simply demolition. Although it is difficult to imagine that a structure would be left deteriorating to such extent, one of the authors was involved in a case in which an important industrial asset had to be stopped and evacuated, due to the risk of collapse caused by extensive steel corrosion damage. 1.4 THE COSTS OF LACK OF DURABILITY Figure 1.5 shows the deterioration process, after Tuutti’s model, as a dotted line referred to the right-hand-side vertical axis. The full line (referred to the left axis) shows the incremental costs of remedial interventions along the service life of the structure. The full line represents what is called the “Law of Fives” (de Sitter, 1984), by which the cost of intervention grows with time by a factor of 5, law that was confirmed in practice (Wolfseher, 1998); below some further explanations.

Figure 1.5 Increasing costs of remedial interventions with time, or “Law of Fives” (de Sitter, 1984).

@seismicisolation @seismicisolation

8 Concrete Permeability and Durability Performance

Design and Construction Phase: Here the germs of an unsatisfactory performance are seeded, as a result of a poor design and materials specification or of bad execution. Relative Corrective Cost = 1. Incubation Phase: There is no visible damage yet. If the problem is detected at this stage (NDTs, covermeters, carbonation, chloride profiles, etc.) it is still possible to act preventively, for example, by applying appropriate surface treatments. Relative Corrective Cost = 5. Localized Damage Phase: Deterioration has started in some areas, as revealed by stains, cracks and/or localized spalling. Repair and maintenance work is required. Relative Corrective Cost = 25. Generalized Damage Phase: If repair and maintenance work has not been carried out, the structure will reach a stage in which delicate and complex repair and retrofitting work is required or even the complete replacement of the elements. Relative Corrective Cost = 125. The importance of having things done correctly from the very beginning can be realized from Figure 1.5; it is in the interest of the owners that a good design and construction is achieved. In general, but especially regarding public works, it is in the interest of the whole society and taxpayers that the constructions are durable. 1.5 ECONOMICAL, ECOLOGICAL AND SOCIAL IMPACTS OF DURABILITY Today it is clear that civil engineers, builders and architects have succeeded in establishing and applying sound criteria to ensure the stability and strength of concrete structures. Fortunately, cases of partial or total collapse of such structures are extremely rare or due to exceptional events. On the contrary, regarding durability, the situation is not so satisfactory. Indeed, all over the world huge amounts of money are spent in the repair or restoration of concrete structures affected by one or a combination of different degradation mechanisms. R. Torrent, M. Alexander and J. Kropp have addressed the problem in Chapter 1 of RILEM Report 40 (2007), citing several papers that provide quantitative evidence of the onerous macroeconomic consequences of the problem (Peaco*ck, 1985; Browne, 1989; Mehta, 1997; Neville, 1997; Hoff, 1999; Vanier, 1999; Coppola, 2000). As the amount of money available for construction in any society is limited, this means that a steady shift of activity from new constructions onto repair and maintenance is taking place. For emerging countries, with a pressing need to improve their infrastructure, this constitutes a serious barrier against development. This has also strong repercussions for the concrete construction industry and all its players (owners, contractors, materials

@seismicisolation @seismicisolation

Durability of concrete structures 9

suppliers, engineers, specialized workers, insurance companies, etc.). An example is the recent partial collapse, at an age of 51 years, of an important concrete bridge in Genoa, Italy (Seitz, 2019), in which apparently durability weaknesses might have played a role (Virlogeux, 2019). It is interesting to remark that these weaknesses had already been revealed when the bridge was just 15 years old (Collepardi et al., 2018). Being an essential element in the Italian highways network, it was rebuilt very fast, but as a hybrid steelconcrete structure. Indeed, the deck is a 5 m deep, 30 m wide, hollow hybrid steel concrete structure, with a steel shell and a reinforced concrete slab forming the road surface, supported by 18 reinforced concrete ellipticalshaped piers. The steel shell has been divided into sections which have been prefabricated off-site (Horgan, 2020). Moreover, ongoing research activities go on to develop solutions aimed at replacing concrete bridge decks by reinforced polymer solutions (Scott, 2010; Rodriguez-Vera et al., 2011; Mara et al., 2014). Construction companies can adapt to building with other materials, but the negative impact on the cement and concrete industry is direct and can be considerable. Repair work involves high costs in diagnosis and design (consulting companies) and uses relatively low volumes of special (usually high-cost) materials that contain little amount of cement/concrete. Hence, the more the activities are shifted from new construction towards repair activities, the higher the negative impact on the cement and concrete industries. Furthermore, durability and ecology or sustainability go hand by hand. As illustrated in Figure 1.3, a non-durable structure will require one or more interventions during its service life (unnecessary if it had been properly designed, constructed and used). These interventions require partial demolitions and replacement with new materials, with the energy and emissions involved in their production and processing and, in the case of road transport facilities, the extra emissions due to traffic jams caused by the repair work. Quoting de Schutter (2014), it can be stated that “No concrete construction can be sustainable without being durable”. 1.6 DURABILITY DESIGN: THE CLASSICAL PRESCRIPTIVE APPROACH This approach is also known as “deemed-to-satisfy” approach. The three pillars supposedly supporting the achievement of durable concrete, according to the classical approach adopted by most codes and standards for structural concrete worldwide, are depicted in Figure 1.6 (de Schutter, 2009). The approach is based on specifying maximum limits for the w/c or w/b (water/binder) ratio and minimum limits for the compressive strength and the cement or binder content (the latter is not always included in codes). The title of the chart is “Parameters for durable concrete?” (de

@seismicisolation @seismicisolation

10 Concrete Permeability and Durability Performance

Figure 1.6 Parameters for durable concrete? After EN 206 (de Schutter, 2009). Table 1.1 Durability requirements for reinforced concrete in Eurocode 2 and EN206 Corrosion induced by: Carbonation Exposure class

XC1 XC2

XC3

Deicing chlorides

Marine chlorides

XC4 XD1 XD2 XD3

XS1

Row Durability Requirements established in Eurocode 2 indicator 25 30 37 37 37 37 45 1 f ′cmin (MPa)a 15 25 25 30 35 40 45 2 cmin,dur 50 yearb

3 4 5 a b c

Durability indicator

Requirements established in EN 206

f ′cmin (MPa)a,c w/cmax Cementmin (kg/m³)

25 30 0.65 0.60 260 280

37 0.55 280

37 0.50 300

37 0.55 300

45 0.45 320

37 35

XS2 XS3

45 40

45 45

37 45 45 0.50 0.45 0.45 300 320 340

Compressive strength measured at 28 days on moist-cured concrete cubes. Minimum cover thickness (mm) for service life of 50 years, structural classes S4. Optional requirement.

Schutter, 2009), intended to show the weaknesses of each of the three pillars. Regarding steel corrosion, a fourth pillar exists, representing the thickness of the concrete cover. Table 1.1 shows the three pillars in practice, for the case of steel corrosion induced by carbonation and chlorides, according to Eurocode 2 (EN 19921-1, 2004; EN 206, 2013). In what follows, a brief consideration on the suitability of the four durability indicators in Table 1.1 is provided; for a more detailed discussion of the subject, the reader can refer to Torrent (2018).

1.6.1 Compressive Strength as Durability Indicator Regarding the suitability of compressive strength as durability indicator, the following comment in CEB-FIP Model Code 1990 (CEB/FIP, 1991), Section d.5.3 “Classification by Durability”, is very relevant: @seismicisolation @seismicisolation

Durability of concrete structures 11

“Though concrete of a high strength class is in most instances more durable than concrete of a lower strength class, compressive strength per se is not a complete measure of concrete durability, because durability primarily depends on the properties of the surface layers of a concrete member which have only a limited effect on concrete compressive strength.” Similar considerations can be found in p. 156 of Model Code 2010 (fib, 2010). The example in Figure 1.7 illustrates the lack of direct association between strength and durability. Represented in the chart are 18 concretes made with widely different cement types and w/c ratios of 0.40 and 0.65 (see Table 5.4); more details on the characteristics of the mixes in Moro and Torrent (2016). In ordinates, the Coefficient of Chloride Migration M Cl (Tang-Nilsson method, described in Section A.2.1.2); in abscissae the compressive strength measured on 150 mm cubes. The samples were moist cured for 28 days, age at which the tests were initiated. Two extreme sets of data are explicitly shown in the chart: those of two different OPCs (triangles) and those of a cement containing 68% of GGBFS (squares). On the right-hand edge of the chart, a classification of resistance to chloride ingress based on 28-day chloride migration results, proposed by Nilsson et al. (1998), is shown with abbreviations: L = Low; M = Moderate; H = High; VH = Very High and EH = Extremely High. A general trend of increasing resistance to chloride ingress with compressive strength can be observed in Figure 1.7. However, for the same strength, the OPC concretes show higher migration coefficients than the rest and, even for strengths above

Figure 1.7 Relationship between chloride migration and cube strength at 28 days.

@seismicisolation @seismicisolation

12 Concrete Permeability and Durability Performance

75 MPa, cannot reach a level of VH resistance to chlorides ingress. On the other extreme, the concretes made with a cement containing 68% GGBFS (incidentally made with the same clinker as one of the OPCs) present lower migration coefficients than the rest, for the same compressive strength. Establishing minimum strength classes as durability requirement is a way to keep w/c ratio at low levels, due to the impossibility of measuring the latter. Section R4.1.1 of ACI 318 (2011) openly confesses: “Because it is difficult to accurately determine the w/cm of concrete, the f′c specified should be reasonably consistent with the w/cm required for durability. Selection of an f′c that is consistent with the maximum permitted w/cm for durability will help ensure that the maximum w/cm is not exceeded in the field.”

1.6.2 Water/Cement Ratio as Durability Indicator Concrete durability depends, to a large extent, on the resistance of the material to the penetration of aggressive species by a combination of different mechanisms (chiefly permeability and diffusion). This resistance is governed, primarily, by the pore structure of the concrete system, especially that of the cement paste and of the interfacial transition zone around the aggregates (see Chapter 3). Establishing limits to the composition of the concrete (especially to its w/c ratio) constitutes an attempt to regulate the pore structure of the concrete system. However, it implies assuming that all materials (especially cements) perform identically; that is, all concretes of the same w/c ratio will perform identically, irrespective of the characteristics of the cement (and other constituents) involved. For the same constituents, it is true that a higher w/c ratio means higher “penetrability” of the concrete (Section 3.2.2). However, for the same w/c ratio, the “penetrability” of a concrete varies significantly with the type and characteristics of the cement used. Figure 1.8 shows the large range of values of the Coefficient of Chloride Migration M Cl (Jacobs & Leemann, 2007), Tang-Nilsson method (Section A.2.1.2), that can be found for a given w/c, when concretes are made with different cements. A similar pattern is shown in Figure 6.5 regarding air-permeability. These examples show that, technically speaking, w/c is not a good durability indicator. This fact is greatly aggravated by the difficulties for the user of the concrete to check compliance with the maximum limits specified. Two options are offered in EN 206 (2013) for checking compliance with the prescriptive limits for w/c ratio:

@seismicisolation @seismicisolation

Durability of concrete structures 13

Figure 1.8 Relation of chloride migration with w/c ratio.

Option (a) is almost exclusively used, despite its grave deficiencies. Indeed, an accidental or deliberate error in the stated sand moisture will end up in a wrong w/c ratio reported in the batching print-out. An example is presented in Torrent (2018), where an error in the sand moisture used for the calculation brought the declared w/c = 0.44 for reported moisture of 0.4% to a true w/c = 0.59 for the true measured 6.0% of sand moisture. To this we should add that, sometimes, washing water is left inside the drum when the ready-mixed concrete (r-mc) truck is loaded with a new batch. In addition, “slumping” is a very common practice in the r-mc industry whereas the driver, while washing the loaded truck still in the plant, watches the consistency of the concrete and, if judged too stiff, adds uncontrolled amounts of water into the drum. On arrival to the jobsite, water is sometimes added to retemper the mix (in a Western European country, the addition of 30 L/m³ into a truck, was witnessed by one of the authors). All the extra water, discussed in this paragraph, which may be added to the truck, is usually not recorded in the batching protocol which, therefore, underestimates the w/c ratio of the concrete delivered, with negative effects on the resulting durability.

@seismicisolation @seismicisolation

14 Concrete Permeability and Durability Performance

Regarding option (b), despite several attempts, no standardized or widely accepted test method to experimentally measure the w/c ratio of the freshly delivered concrete has been developed. An overview of such attempts can be found in CR 13902 (2000) that states: “It follows […] that the problem of measuring water/cement ratio on a sample of fresh concrete about which nothing is known is very difficult and probably impossible”. The fact remains that one of the critical weaknesses of the use of w/c ratio as durability indicator is the impossibility of checking compliance by the user. Specifying a characteristic that cannot be measured is clearly meaningless and opens roads to unfair competition by fraudulent practices undetected by the user.

1.6.3 Cement Content as Durability Indicator The main argument behind the specification of a minimum cement content in the mix is the chemical binding effect that hydrated cement offers to free chlorides and CO2 that can penetrate the concrete (Wassermann et al., 2009). Along this line of thinking, a higher cement content would imply a higher reservoir of alkalis that need more CO2 to be carbonated (same for more Cl− binding), thus delaying the advance of the critical front. But, for the same w/c ratio (which is specified in parallel), a higher cement content means also a proportionally higher volume of porous paste that allows more CO2 (or Cl−) to penetrate, with a null net result; this has been proved experimentally (Wassermann et al., 2009). Moreover, more paste and more cement mean more susceptibility of the concrete to shrinkage and thermal cracking. The use of mineral additions batched separately into the concrete mixer adds further complications to establishing the cement content (and also the w/c ratio), due to the application of the controversial “k-value concept” of EN 206 (2013) to assess the “cementitious contribution” of the addition used.

1.6.4 Cover Thickness as Durability Indicator The thickness of the concrete cover is a very important durability indicator for the deterioration of structures due to steel corrosion. A lack of sufficient cover thickness is a recurrent cause of premature corrosion of reinforcing steel (Wallbank, 1989; Neville, 1998; Torrent, 2018). In theory, both second Fick’s diffusion law (through the argument of the error function solution, Section 3.4.1) and capillary suction theory (see Section 3.7) predict a progress of the penetration front of carbonation, chlorides and water with the square root of time. This means that a 10% reduction in cover thickness implies a reduction in service life of 20%, so great is the importance of observing the specified cover

@seismicisolation @seismicisolation

Durability of concrete structures 15

thickness. Yet, despite the progress made on electromagnetic instruments capable of assessing non-destructively the cover thickness quite accurately (Fernández Luco, 2005), now largely enhanced by the development of ground penetrating radar (GPR) instruments, their use is not forcibly specified in the standards. 1.7 DURABILITY DESIGN: THE PERFORMANCE APPROACH In Sections 1.6.1–1.6.3, the inherent weaknesses of the three pillars of the classical durability approach have been revealed. With more or less degree of boldness, codes and standards have been moving, rather timidly, along the P2P (Prescriptive to Performance) road, as discussed in this section and also in Section 12.1.

1.7.1 The “Durability Test” Question The P2P transit brings to the forefront the question of suitable durability tests, that was discussed in Torrent (2018) and that, given the scope of this book, deserves a revisit. The durability of concrete is, almost by definition, hardly measurable by testing, as each structure is performing its own durability test, live under its own specific conditions. The prediction of the evolution of the structure condition is uncertain, particularly when based on testing specimens and not the real structure. Durability involves deterioration processes lasting several years; therefore, it is clear that tests lasting months or even years, although in some cases possibly closer to reality, are not practical for specification and quality control purposes. Performance specifications need short-term tests, lasting not more than, say, 1 week, including preconditioning of the specimens; otherwise, the approach would not be practical nor acceptable for conformity control purposes, given the current pace of concrete construction. The durability of concrete structures against deteriorating actions originated from the surrounding environment is strongly related to the resistance of the concrete cover to the penetration, by different transport mechanisms, of external deleterious substances. As a result of 30–40 years of durability research, several test methods have been developed to measure mass transport properties of concrete (RILEM Report 40, 2007; RILEM STAR 18, 2016). They consist typically of tests that measure the resistance of concrete to the transport of matter (gaseous or liquid) by appropriate driving forces (see Chapters 3 and 4 and Annex A). Some of these tests have been standardized in different European countries and in the USA.

@seismicisolation @seismicisolation

16 Concrete Permeability and Durability Performance

All these durability tests have merits and demerits. If we wait until the perfect “durability” test is developed, we will never leave the unsatisfactory and ineffective current prescriptive approach. Indeed, any reasonable durability test will be better than the w/c ratio, used today as the durability “panacea”. Drawing a parallel, concrete structural design relies heavily on the compressive strength (measured with a standard test), adopted as the universal, used-for-all property (in codes almost all properties of concrete are derived from its value). And yet, this test can be questioned from different angles: the true stress field is far from uniaxial compression (hence the 20%–25% difference when testing cylinders and cubes); the size is much smaller than that of the structural elements (size effect); the load is statically applied in less than 5 minutes (in bridges, static loads are applied for decades with ≈ 15% strength reduction effect and even cyclically, bringing in also the deleterious effect of fatigue); the specimens are tested saturated (a condition seldom found in reality, which influences the strength by ≈ 20%), and so on and so forth. Yet, despite all these limitations, the standard compressive strength is accepted by civil engineers as a suitable indicator of the bearing capacity of concrete and is used, without objections, in the structural design of concrete structures. A similar, pragmatic approach is required for durability, that is, the adoption of well proved standard tests to measure relevant “Durability Performance Indicators”, focusing on the merits and positive contribution of the tests and less on their demerits.

1.7.2 Canadian Standards Canadian Standards specify limiting values of the result (electrical charge passed Q in Coulombs) in a migration test (ASTM C1202, 2019) (see description in Section A.2.1.1). Canadian Standard (CSA, 2004) specifies a maximum limit of Q = 1,500 Coulombs for exposure classes C-1 (structurally reinforced concrete exposed to chlorides with or without freezing and thawing conditions) and A-1 (structurally reinforced concrete exposed to severe manure and/or silage gases, with or without freeze-thaw exposure. Concrete exposed to the vapour above municipal sewage or industrial effluent, where hydrogen sulphide gas may be generated). That limit is reduced to Q = 1,000 Coulombs for exposure class C-XL (structurally reinforced concrete exposed to chlorides or other severe environments with or without freezing and thawing conditions, with higher durability performance expectations). The testing age should not exceed 56 days.

1.7.3 Argentine and Spanish Codes The Codes for Structural Concrete of Argentina (CIRSOC 201, 2005) and Spain (EHE-08, 2008) rely on a water-permeability test for @seismicisolation @seismicisolation

Durability of concrete structures 17

durability specifications. The test is known as Water Penetration under Pressure (EN12390-8, 2009), described in Section 4.1.1.2, and the result is the maximum (sometimes also the mean) depth of penetration of water Wp reached under pressure onto the surface of a concrete specimen. Table 1.2 summarizes the requirements of the Argentine and Spanish Codes referred to this test. The Argentine Code also specifies, for all aggressive environments, a maximum water sorptivity of 4.0 g/m²/s½ (see Section 4.2.1) which looks quite demanding, especially if applied to all exposure classes.

1.7.4 Japanese Architectural Code The Japanese Architectural Code “Recommendations for Durability Design and Construction Practice of Reinforced Concrete Buildings” (Noguchi et al., 2005) includes, in its Chapter 2, the principles of durability design. Regarding performance-based design of carbonation-induced steel corrosion, a probabilistic approach is adopted, based on Eq. (1.1) of carbonation progress. Table 1.2 Performance laboratory tests and limiting values specified in some national standards Country

Standard

Canada

CSA A23.1/ A23.2

Argentina and Spain Switzerland

Indicator Electric charge passed (Coulomb)

CIRSOC 201, Maximum water EHE 08 penetration under pressure (mm) SIA 262/1 Water sorptivity (g/m²/h) Chloride migration coefficient (10−12 m²/s) Carbonation rate (mm/y½) Frost-thaw-salts, mass loss (g/m²) Sulphate resistance expansion (‰)

Test method

Exposure

Limit

ASTM C1202

Chlorides, manure gases

≤1,500 ≤1,000

EN 12390-8

Extended service life Moderate Severe

≤50 ≤30

SIA 262/1: Mild chlorides Annex A SIA 262/1: Chlorides Annex B

≤10

SIA 262/1: Carbonation Annex I SIA 262/1: Mild frost Annex C Severe frost SIA 262/1: Sulphates Annex D

≤5.0

@seismicisolation @seismicisolation

≤10

≤200 ≤1,200 ≤1.2

18 Concrete Permeability and Durability Performance

where Ct = carbonation depth at time t k = coefficient (1.72 after Kish*tani or 1.41 after Shirayama) α1 = coefficient function of concrete and aggregate type α2 = coefficient function of cement type α3 = coefficient function of mix proportions (w/c ratio) β1 = coefficient function of air temperature β2 = coefficient function of relative humidity of air β3 = coefficient function of CO2 concentration In Chapter 7 of AIJ (2016), “Practice and quality management”, the coefficient of air-permeability kT (Torrent method, described in Chapter 5) is used to predict concrete carbonation with consideration of moisture effect. Sampling method follows Annex E of Swiss Standard (SIA 262/1, 2019). The main purpose of this code is not to establish durability specifications but to predict the carbonation progress in concrete structures.

1.7.5 Portuguese Standards In Portugal, the performance-based durability design is possible through the application of LNEC E 465 (2007). This standard addresses the deterioration by reinforcement corrosion, induced by carbonation and sea chlorides. Its major features are summarized as follows: • applies a semi-probabilistic approach, where the reliability analysis is carried out in the service life format • the end of service life is defined as the occurrence of corrosion-induced cracking • the service life is broken down in two periods (initiation and propagation) following Tuutti’s model (Figure 1.4) • comprises one analytical model for the propagation period, based on Faraday’s law and on the empirical expression proposed by Rodriguez et al. (1996) • comprises three analytical models for the initiation period, one for chloride penetration and two for concrete carbonation • the analytical model for chloride penetration is based on the model proposed by Mejlbro (1996) and uses chloride migration coefficient from NT Build 492 test (see A.2.1.2), as durability indicator • one of the analytical models for concrete carbonation is also based on “CEB Task Group V, 1+2 model” (DuraCrete, 1998) and uses concrete resistance to accelerated carbonation (LNEC E 391, 1993) as durability indicator • the other model for concrete carbonation uses the oxygen-permeability coefficient from CEMBUREAU test (see 4.3.1.2) as durability indicator, adapting Parrott’s model (Parrott, 1984), see Section 9.3.1. @seismicisolation @seismicisolation

Durability of concrete structures 19

Further, the input parameters for the analytical models vary according to the exposure conditions and these are grouped in exposure classes according to EN 206 (2013). Three safety factors for service life are defined, one for each of the reliability classes identified in Eurocode 0 (EN 1990, 2002). This methodology allows the user to define a combination of nominal cover thickness and performance requirement (chloride migration coefficient. oxygen-permeability coefficient or carbonation resistance), to ensure the intended service life.

1.7.6 South African Standards For many years, thanks to the continuous and persistent work of several distinguished researchers such as Alexander et al. (1999), Alexander (2004) and Beushausen and Alexander (2009), an original performance concept was introduced and consolidated in South Africa, crowned with its acceptance in South African Standards (CO3-2, 2015; CO3-3, 2015). It consists in measuring “Durability Indices” (oxygen-permeability and chloride conductivity) in the laboratory, on cores drilled from the finished structure. These indices, coupled with the assumed or measured cover thickness allow, via modelling, the assessment of the service life of reinforced concrete structures exposed to carbonation or chlorides. This approach and its test methods are described in more detail in Sections 4.3.1.3, 9.3.2 and A.2.2.3.

1.7.7 Swiss Standards The Swiss Codes and Standards for Concrete Construction have taken decisively the road to performance specifications, based primarily on three separate standards, namely: • SIA 262 (2013) based on Eurocode 2 is the Swiss Concrete Construction Code, defining exposure classes and corresponding cover thicknesses • SIA 262/1 (2019) describes special, non-EN Standard tests and sets performance requirements associated with the exposure classes, to be fulfilled for laboratory and site tests (NDT or drilled cores) • SN EN 206 (2013), prescriptive, is the Swiss version of EN 206 • Table 1.3 shows the evolution of Swiss Standards’ requirements for exposures that promote steel corrosion, moving from purely Prescriptive to Performance-based; more details can be found in Torrent and Jacobs (2014). In Switzerland, the following has been achieved: • for each exposure class, a suitable “Durability Performance Indicator” test has been adopted, for example: – accelerated carbonation for XC3 and XC4 – water capillary suction for XD1 and XD2a – chloride migration for XD2b and XD3 @seismicisolation @seismicisolation

20 Concrete Permeability and Durability Performance

• a standard for conducting each of these tests has been issued (SIA 262/1, 2019) • limiting values and conformity rules have been established for the test results (average of different samples) in each exposure class (see Table 1.2 and Table 1.3, rows 5–7) The concrete producer, supplying concrete for structures under a given exposure classes, shall design the mixes complying with the prescriptive requirements of rows 1 and 2 in Table 1.3. In addition, the concrete producer shall cast specimens (“Labcrete”, see Section 7.1.2) from samples taken during the regular production with a frequency that is function of the volume produced, but at least four times per year. The averages of the test results on these samples must comply with the maximum requirements of Table 1.3, rows 5–7. More important, perhaps, now the user can take samples during delivery and check compliance of the received concrete with the specifications. It is envisaged that, once enough experience has been accumulated with the performance requirements, the prescriptive requirements will be removed from the standard or kept as recommended values. One of the most innovative aspects of the Swiss Standards is the recognition that tests made on cast samples are not truly representative of that of the cover concrete Covercrete, see Section 7.1.4) of the real structure. SIA 262 Code (SIA 262, 2013) describes the measures to be adopted in order to ensure durability and, acknowledging the importance of the role of the Covercrete, specifically states (free translation from German into English): • “with regard to durability, the quality of the cover concrete is of particular importance”, Section 5.2.2.7 of SIA 262 (2013) • “the tightness of the cover concrete shall be checked, by means of permeability tests (e.g. air-permeability measurements), on the structure or on cores taken from the structure”, Section 6.4.2.2 of SIA 262 (2013) Therefore, since 2013, the air-permeability kT of the Covercrete of structural elements exposed to the most severe environments shall be checked on site, with the “Air-Permeability on the Structure” test, according to Annex E of SIA 262/1 (2019). The requirements for site air-permeability are indicated in Row 8 of Table 1.3, the specified kTs values being “characteristic” upper limits, having their own conformity rules (SIA 262/1, 2019; Torrent et al., 2012), see Section 8.5.1.

@seismicisolation @seismicisolation

Durability of concrete structures 21 Table 1.3 Evolution of Swiss Standards requirements (for corrosion exposure classes)

Year

Row

2003 1 2 3 4 2008 5 6 7 2013 8

Exposure class

Carbonation-induced corrosion XC1

XC2

Durability indicator w/cmax Cmin (kg/m³) f'cmin (MPa) dnom (mm)

PRESCRIPTIVE

Durability indicator qw max (g/m²/h) DCl max (10−12 m²/s) KN max (mm/y1/2) Durability indicator kTs (10−16 m²)

LABCRETE

0.65 280 25 20

0.65 280 25 35

XC3

XC4

Chloride-induced corrosion XD1

XD2a

XD2b

XD3

0.60 280 30 35

0.50 300 37 40

0.50 300 30 40

0.45 320 37 55

5.0/4.0 5.0/4.5 -

2.0

0.5

REALCRETE -

2.0

Note: EN 206 Class XD2 was subdivided in 2008 into XD2a and XD2b, for chloride contents of the solution in contact with the concrete of up to or over 0.5 g/L, respectively. w/c, water/cement ratio by mass; C, cement content, including SCM with corresponding factors k; f'c, strength class (cube); qw, water conductivity coefficient, Annex A of SIA 262/1 (2019). Rather complex indicator, closely related to water absorbed in 24 hours w24 (g/m²): w24 = 217 + 326 × qw; DCl = chloride migration coefficient, measured after Tang-Nilsson method (Section A.2.1.2); KN = carbonation resistance = 0.136 KS, with KS measured in an accelerated test after 7, 28 and 63 days exposure to CO2 concentration of 4%-vol. (Annex I of SIA 262/1 (2019)). The values indicated correspond to expected service lives of 50/100 years; kT, coefficient of air-permeability, measured after Torrent method (Annex E of SIA 262/1 (2019)); value not be exceeded by more than 1 test out of 6; dnom, nominal cover depth, values indicated are for reinforced concrete (values for prestressed concrete are 10 mm higher); typical tolerance ± 10 mm.

1.8 CONCRETE PERMEABILITY AS “DURABILITY INDICATOR” In Section 1.2, we could see that most relevant mechanisms of deterioration of concrete structures have a close relation to the permeability of concrete. It is not surprising, then, that several performance-based standards and codes (see Sections 1.7.3–1.7.7) select water-permeability (in the form of penetration under pressure or of capillary suction) or gas-permeability as durability indicator. In the particular case of the South African and Swiss

@seismicisolation @seismicisolation

22 Concrete Permeability and Durability Performance

Standards, based on core testing and non-destructive measurements conducted on site, respectively, the end-product is tested, which is more representative than laboratory tests performed on cast specimens, as discussed in Chapter 7. Being the main topic of this book, the suitability of concrete permeability as a durability indicator is broadly and deeply dealt with. 1.9 BEYOND 50 YEARS: MODELLING Most requirements described in Sections 1.5 and 1.6 correspond to an expected service life of 50 years. Nowadays, important infrastructure constructions are intended for service lives that largely exceed the 50 years expected by the application of the prescriptive EN standards or the performance Swiss standards. Examples are the Alp Transit Tunnel in Switzerland (100 years) (Alp Transit, 2012), the new Panama Canal (100 years) (Cho, 2012), the Chacao Bridge in Chile (100 years) (Valenzuela & Márquez, 2014), the Hong Kong-Zhuhai-Macao link in China (120 years) (Li et al., 2015), the Port of Miami Tunnel in USA (150 years) (Torrent et al., 2013) and the second Brisbane Gateway Bridge in Australia (300 years) (Gateway, 2009), all of them exposed to very aggressive environments. Due to the lack of experience with such longevous structures (reinforced concrete is a rather “recent” building system) from which to draw learnings, the solution lies on the judicious use of predictive models. The most widespread model used today in Europe is Duracrete (DuraCrete, 2000), later partially adopted by fib (2006), dealing with steel corrosion induced by carbonation or chlorides, whilst in North America (Life-365, 2012) model (only for chloride-induced corrosion) is the preferred one. These models are based on the assumption that the penetration of chlorides (and carbonation) is a purely diffusive process governed by Fick’s second law (see Chapter 3), with the main input durability indicators being the cover thickness and the coefficient of chloride-diffusion (or migration) of the concrete. In Chapter 9, several service life design models, based on the use of concrete permeability as input, are presented. REFERENCES ACI 318 (2011). “Building code requirements for structural concrete”. ACI. AIJ (2016). “Recommendations for durability design and construction practice of reinforced concrete buildings”. Architectural Institute of Japan (in Japanese). Alexander, M.G. (2004). “Durability indexes and their use in concrete engineering”. International RILEM Symposium on Concrete Science and Engineering: ‘A Tribute to Arnon Bentur’, 9–22. Alexander, M.G. (2016). Marine Concrete Structures: Design, Durability and Performance. Woodhead Publishing, Duxford, UK, 485 p.

@seismicisolation @seismicisolation

Durability of concrete structures 23 Alexander, M.G., Ballim, Y. and Mackechnie, J.R. (1999). “Concrete durability index testing manual”. Research Monograph No.4, Univs. Cape Town & Witwatersrand, South Africa, 33 p. Alexander, M.G., Bentur, A. and Mindess, S. (2017). Durability of Concrete: Design and Construction. CRC Press, Boca Raton, FL, 345 p. Alp Transit (2012). “Alp Transit Gotthard – New traffic route through the heart of Switzerland”. Brochure, 48 p. ASTM C1202 (2019). “Standard test method for electrical indication of concrete’s ability to resist chloride ion penetration”. Bertolini, L., Elsener, B., Pedeferri, P. and Polder, R. (2004). Corrosion of Steel in Concrete. Wiley-VCH, Weinheim, Germany, 391 p. Beushausen, H. (2014). “RILEM TC230-PSC: Performance-based specification and control of concrete durability”. RILEM Week, São Paulo, Brazil, September, 42 slides. Beushausen, H. and Alexander, M. (2009). “Application of durability indicators for quality control of concrete members – A practical example”. Concrete in Aggressive Aqueous Environments – Performance, Testing, and Modeling, Alexander, M.G. and Bertron, A. (Eds.). RILEM Publications, Bagneaux, 548–555. Böhni, H. (2005). Corrosion in Reinforced Concrete Structures. Woodhead Publishing, Cambridge, UK, 241 p. Browne, R. (1989). “Durability of reinforced concrete structures”. New Zealand Concrete Construction, September 2–10 and October 2–11. Cho, A. (2012). “Dramatic digs mark panama canal expansion progress”. Engineering News-Record, July 18. CIRSOC 201 (2005). “Reglamento Argentino de Estructuras de Hormigón”. Argentine Concrete Code. Collepardi, M., Troli, R. and Collepardi, S. (2018). “Ponte di Genova: alcune considerazioni sul calcestruzzo e gli agenti esterni che ne hanno ridotto la durabilità”. Ingenio-web.it, August 22, 5 p. Coppola, L. (2000). “Concrete durability and repair technology”. 5th CANMET/ ACI International Conference On ‘Durability of Concrete’, Barcelona, Spain, June 4–9, 1209–1220. CO3-2 (2015). “Civil engineering test methods. Part CO3-2: Concrete durability index testing – Oxygen permeability test”. South African Standard SANS 3001-CO3-2:2015. CO3-3 (2015). “Civil engineering test methods. Part CO3-3: Concrete durability index testing – Chloride conductivity test”. South African Standard SANS 3001-CO3-3:2015. CR 13902 (2000). “Test methods for determining the water/cement ratio of fresh concrete”. CEN Report, European Committee for Standardization, May, 7 p. CSA (2004). Canadian Standard A23.1/A23.2: “Concrete materials and methods of concrete construction/methods of test and standard practices for concrete”. de Schutter, G. (2009). “How to evaluate equivalent concrete performance following EN 206-1? The Belgian approach”. PRO 66: Concrete Durability and Service Life Planning – ConcreteLife’09. Kovler, K. (Ed.). RILEM Publications, Haifa, Israel, 1–7. ISBN: 978-2-35158-074-5. de Schutter, G. (2014). “No concrete is sustainable without being durable!”. XIII DBMC, São Paulo, Brazil, 49–56.

@seismicisolation @seismicisolation

24 Concrete Permeability and Durability Performance de Sitter, W.R. (1984). “Costs for service life optimization: The Law of Fives”. Durability of Concrete Structures, Workshop Report, Rostam, S. (Ed.), Copenhagen, Denmark, May 18–20, 131–134. DuraCrete (1998). “Modelling of degradation”. The European Union–Brite EuRam III, BE95–1347/R4–5, CUR, Gouda, The Netherlands. DuraCrete (2000). “Probabilistic performance based durability design of concrete structures”. The European Union–Brite EuRam III, BE95–1347/R17, CUR, Gouda, The Netherlands. Dyer, T. (2014). Concrete Durability. CRC Press, Boca Raton, FL, 402 p. EHE-08 (2008). “Instrucción de Hormigón Estructural”. Spanish Concrete Code. EN 206 (2013). “Concrete – Specification, performance, production and conformity”, December. EN 1990-1-1 (2002). “Basis of structural design”. EN 1992-1-1 (2004). “Eurocode 2: Design of concrete structures – Part 1-1: General rules and rules for buildings”, December. EN 12390-8 (2009). “Testing hardened concrete – Part 8: Depth of penetration of water under pressure”. Fernández Luco, L. (2005). “RILEM recommendation of TC 189-NEC: Comparative test – Part II – Comparative test of Covermeters”. Mater. & Struct., v38, 907–911. Fernández Luco, L. and Torrent, R. (2003). “Diagnosis of a case of harmless alkali-silica reaction in a cracked concrete pavement”. 6th CANMET/ACI International Conference on Durability of Concrete, Thessaloniki, Greece, June 1–7, 521–536. fib (2006). “Model code for service life design”. fib Bulletin 34, February, 112 p. fib (2010). “Model code 2010”. 1st Complete Draft, v1, March. Gateway (2009). “A second Gateway Bridge”. Gateway Upgrade Project, Fact Sheet 4, February. Gjørv, O.E. (2014). Durability Design of Concrete Structures in Severe Environments. 2nd ed., CRC Press, Boca Raton, FL, 254 p. Hoff, G.C. (1999). “Integrating durability into the design process”. Controlling Concrete Degradation. Dhir, R.K., Newlands, M.D. (Eds.). Thomas Telford, London, 1–14. Horgan, R. (2020). “Polcevera viaduct | Fatal collapse replacement bridge to be completed this month”. New Civil Eng., April 07. Hunkeler, F. (2000). “Corrosion in reinforced concrete: Processes and mechanisms”. Corrosion in Reinforced Concrete Structures. Böhni, H. (Ed.). CRC Press, Cambridge, UK, 1–45. Jacobs, F. and Leemann, A. (2007). “Betoneigenschaften nach SN EN 206-1”. ASTRA Report VSS Nr. 615, Bern. Li, K. (2016). Durability Design of Concrete Structures. Wiley, Singapore, 280 p. Li, Q., Li, K.F., Zhou, X., Zhang, Q. and Fan, Z. (2015). “Model-based durability design of concrete structures in Hong Kong–Zhuhai–Macau sea link project”. Structural Safety, v53, March, 1–12. Life-365 (2012). “Service Life Prediction ModelTM and computer program for predicting the service life and life-cycle cost of reinforced concrete exposed to chlorides”. Life-365 Consortium II, 80 p. LNEC E 391 (1993). “Concrete. Determination of carbonation resistance”, May.

@seismicisolation @seismicisolation

Durability of concrete structures 25 LNEC E 465 (2007). “Concrete: Methodology for estimating the concrete performance properties allowing to comply with the design working life of the reinforced or prestressed concrete structures under the environmental exposures XC and XS”, November. Mara, V., Haghani, R. and Harryson, P. (2014). “Bridge decks of fibre reinforced polymer (FRP): A sustainable solution”. Constr. & Build. Mater., v50, January, 190–199. Mehta, P.K. (1997). “Durability – Critical issues for the future”. Concr. Intern., July, 27–33. Mehta, P.K., Schiessl, P. and Raupach, M. (1992). “Performance and durability of concrete systems”. 9th Congress on the Chemistry of Cement, New Delhi, India, 571–659. Mejlbro, L. (1996). “The complete solution of Fick’s second law of diffusion with time-dependent diffusion coefficient and surface concentration”. Durability of Concrete in Saline Environment, Cementa AB, Danderyd, Sweden, 127–158. Moro, F. and Torrent, R. (2016). “Testing fib prediction of durability-related properties”. fib Symposium 2016, Cape Town, South Africa, 21-23 Nov. Neville, A. (1997). “Maintenance and durability of structures”. Concr. Intern., November, 52–56. Neville, A. (1998). “Concrete cover to reinforcement — or cover up?” Concr. Intern., v20, n11, November, 25–29. Neville, A. (2003). Neville on Concrete. ACI, Farmington Hill, MI. Nilsson, L.-O. (2012). “Transport processes in the microstructure of concrete and their relevance for durability”. Keynote paper, Microdurability, Amsterdam, April 11–13. Nilsson, L., Ngo, M.H. and Gjørv, O.E. (1998). “High performance repair materials for concrete structures in the port of Gothenburg”. CONSEC 1998, Tromsø, Norway, June 21–24, v.2, 1193–1198. Noguchi, T., Kanematsu, M. and Masuda, Y. (2005). “Outline of recommendations for durability design and construction practice of reinforced concrete buildings in Japan”. ACI SP 234, 347–372. Parrott, P.J. (1984). “Design for avoiding damage due to carbonation-induced corrosion”. 3rd International Conference on Durability of Concrete, Nice, France, May, 283–298. Peaco*ck, W.J. (1985). “The maintenance of buildings and structures – The problem, some causes and remedies”. Proceedings on Thomas Telford Seminar, “Improvement in concrete durability”. Institution Civil Engs., London, May, 131–161. Richardson, M.G. (2002). Fundamentals of Durable Reinforced Concrete. Spon Press, London, 254 p. RILEM Report 40 (2007). “Non-destructive evaluation of the penetrability and thickness of the concrete cover”. State-of-the-Art Report of RILEM TC 189NEC, Torrent, R. and Fernandez Luco, L. (Eds.), 223 p. RILEM STAR 18 (2016). “Performance-based specifications and control of concrete durability”. State-of-the-Art Report Vol. 18, RILEM TC 230-PSC, Beushausen, H. and Fernandez Luco, L. (Eds.), 373 p.

@seismicisolation @seismicisolation

26 Concrete Permeability and Durability Performance Rodriguez, J., Ortega, L.M., Casal, J. and Diez, J.M. (1996). “Corrosion of reinforcement and service life of concrete structures”. DBMC 7, Stockholm, Sweden, May 19–23, v.1, 117–126. Rodriguez-Vera, R.E., Lombardi, N.J., Machado, M.A., Liu, J. and Sotelino, E.D. (2011). “Fiber reinforced polymer bridge decks”. Publication FHWA/IN/ JTRP-2011/04. Joint Transportation Research Program, Indiana DoT and Purdue Univ., West Lafayette, Indiana, 113 p. Scott, R. (2010). “FRP bridge decking – 14 years and counting”. Reinf. Plastics, January/February. Seitz, P. (2019). “System Nummer 9 – Einsturz der Morandi-Brücke in Genua”. TEC21, Zürich, 7–8, 20–25. SIA 262 (2013). “Betonbau”. SIA 262/1 (2019). “Betonbau – Ergänzende Festlegungen”. SN EN 206 (2013). “Beton – Teil 1: Festlegung, Eigenschaften, Herstellung und Konformität”. Torrent, R.J. (2018). “Bridge durability design after EN standards: Present and future”. Struct. & Infrastruct. Engng., DOI: 10.1080/15732479.2017.1414859, 14 p. Torrent, R., Alexander, M. and Kropp, J. (2007). “Introduction and problem statement”. RILEM Report 40, May, 1–11. Torrent, R., Armaghani, J. and Taibi, Y. (2013). “Evaluation of port of Miami tunnel segments: Carbonation and service life assessment made using on-site air permeability tests”. Conc. Intern., May, 39–46. Torrent, R., Denarié, E., Jacobs, F., Leemann, A. and Teruzzi, T. (2012). “Specification and site control of the permeability of the cover concrete: The Swiss approach”. Mater. & Corrosion, v63, n12, December, 1127–1133. Torrent, R. and Jacobs, F. (2014). “Swiss standards 2013: World’s most advanced durability performance specifications”. 3rd All-Russian Conference on Concrete and Reinforced Concrete, Moscow, May 12–16. Tuutti, K. (1982). “Corrosion of steel in concrete”. Research report No.4.82. Swedish Cement and Concrete Research Institute (CBI), Stockholm. Valenzuela, M.A. and Márquez, M.A. (2014). “Consideraciones para la Inspección y Mantenimiento del Puente Chacao”. CINPAR, Santiago, Chile, June 4–6. Vanier, D.J. (1999). “Why industry needs asset management tools”. NRCC Seminar Series ‘Innovations in Urban Infrastructure’, 11–25. Virlogeux, M. (2019). “Damals glaubte man, Beton halte ewig”. TEC21, Zürich, 7–8, 26–29. Wallbank, E.J. (1989). “The performance of concrete in bridges”. HMSO, London, April 1989. Wassermann, R., Katz, A. and Bentur, A. (2009). “Minimum cement content requirements: A must or a myth?. Mater. & Struct., v42, 973–982. Wolfseher, R. (1998). “Economical aspects of repair and maintenance”. 5th International Conference on Materials Property & Design, ‘Durable Reinforced Concrete Structure’, Weimar, Germany, October, 33–48.

@seismicisolation @seismicisolation

Chapter 2

Permeability as key concrete property

2.1 FOUNDATIONS OF PERMEATION LAWS The foundations of today’s knowledge on the permeation of fluids through porous media were laid down by the work of Jean Léonard Marie Poiseuille (1799–1869), a French physician and physiologist, who was interested in the conditions of the flow of liquids through narrow tubes, basically associated with the arterial system of blood circulation. He conducted a series of experiments, from which he established that the flow rate of a fluid through a tube of radius r is proportional to r4. Independently, the German civil engineer Gotthilf Heindrich Ludwig Hagen (1797–1884) arrived at the same result by conducting experiments in brass tubes of different diameters, concluding in what is now known as Hagen-Poiseuille law (see Section 3.5.1). More or less simultaneously, the French engineer Henry Darcy (1803– 1858) was studying the laminar flow of water through sand beds, finding that the flow rate was proportional to the energy loss (water head loss), inversely proportional to the length of the flow path and proportional to a coefficient K that depended on the type of sand and also on the type of fluid. The combination of these discoveries led to the general law of permeation of liquids through porous media (viscous laminar flow of Newtonian liquids): Q= K⋅

A ∆P ⋅ µ ∆L

(2.1)

where Q = flow rate (m³/s) K = (intrinsic) coefficient of permeability (m²) A = cross-sectional area traversed by the fluid (m²) µ = viscosity of the fluid (Pa.s) ΔP/ΔL = gradient of pressure across the element (Pa/m)

DOI: 10.1201/9780429505652-2 @seismicisolation @seismicisolation

28 Concrete Permeability and Durability Performance

In the case of an ideal impermeable solid body traversed by parallel capillary tubes of radius r, the coefficient of permeability is (see derivation of formulae in Section 3.5.1): K =

ε ⋅ r 2 8

(2.2)

where ε is the porosity of the body (area of tubes/total cross-sectional area of the body). 2.2 RELATION BETWEEN PERMEABILITY AND PORE STRUCTURE OF CONCRETE Equation (2.2) indicates that the coefficient of permeability of concrete, recognized as a porous medium, must be closely related to the pore structure of the material. One of the main investigations on the permeability of cementitious material was due to the researcher who, possibly, did more to establish studies on concrete as a scientific, rather than an empirical discipline: Treval C. Powers (1900–1997). During his fundamental research on the microstructure of hardened cement paste (h.c.p.), still valid today, and its effect on key properties, he could establish a relationship between the coefficient of water-permeability of h.c.p. and its capillary porosity, quite independent of the cement types investigated at the time (Powers, 1958). He also found that, due to the extremely low size of the gel pores, flow through h.c.p. takes place primarily through its capillary pores. He also established the approximate hydration time required for the capillary pores of h.c.p. of different w/c ratios to become segmented, i.e. connected between them through the gel pores, resulting in very low permeability (Powers et al., 1959). They found that for w/c ratios above 0.70 that segmentation is impossible, as shown in Table 1.6 of Neville (1995). Since that pioneer work, the permeability of concrete received growing attention by researchers worldwide, which resulted in a consolidated knowledge on that property, on how it is influenced by different factors and on how it can be measured, both in the laboratory and on site. These aspects are dealt with in detail in the rest of this book. 2.3 PERMEABILITY AS KEY CONCRETE PROPERTY Water-permeability of concrete is relevant to structures that contain or transport water (or other liquids), in particular dams, tanks containing water or other liquids, retaining walls, canals, culverts, pipes, etc. @seismicisolation @seismicisolation

Permeability as key concrete property 29

Similarly, gas-permeability of concrete is relevant to structures that contain or transport gases, in particular tanks and pipes, underground gas reservoirs to store/release energy, evacuated tunnels for high speed trains, etc. Gas-permeability plays an important role in the release of water vapour under fire, thus decreasing the risk of explosive spalling in the event of fire, topic that is discussed in detail in Chapter 10. In this section, some engineering applications in which concrete permeability plays a key role, not specifically associated with durability, are presented.

2.3.1 Permeability for Liquids’ Containment 2.3.1.1 ACI Low Permeability Concrete Section 4.3 of ACI 318 (2019) includes exposure class P1 “Low Permeability Requirement”, assigned on the basis of the need for concrete to have a low permeability to water, when the permeation of water into concrete might reduce durability or affect the intended function of the structural member. An example is an interior water tank. Requirements: w/b ≤ 0.50 and cylinder compressive strength class ≥ 28 MPa. 2.3.1.2 Dams Conventional concrete for dams is usually sufficiently water-tight to avoid leakage across the thick body of the dam. For conventional concrete dams, built in lifts, construction joints as well as expansion joints are the weak points regarding water-tightness. An interesting research was reported by Görtz et al. (2021), in which the water-tightness of the joints was measured experimentally and modelled numerically. The coefficient of permeability was measured on Ø64.5 mm cores, drilled from a 90-year-old dam in Germany, so as to obtain specimens without joints and with horizontal and vertical joints. The measured water-permeability of the specimens without joints was in the range 0.5 − 3.0 × 10 −9 m/s, whilst for those containing horizontal and vertical joints it climbed to the ranges 5 − 100 × 10 −9 m/s and 1– 30 × 10 −6 m/s, respectively (i.e. one and three orders of magnitude higher, respectively). Two numerical models were applied, that successfully fit to the experimental results, especially the ‘dual-permeability model’. In the case of two concrete-face rockfill dams (Barrancosa and Condor Cliff) on the River Santa Cruz, Patagonia, Argentina, a maximum value for the water-permeability of 2 × 10 −9 cm/s was specified for the upstream concrete slab (Di Pace, 2021). Due to difficulties in measuring that property, an equivalent value of the coefficient of air-permeability (Torrent method) of 0.2 × 10 −16 m² was proposed, applying the relation:

Kw = 6.24 ⋅ kT 0.68

(2.1) @seismicisolation @seismicisolation

30 Concrete Permeability and Durability Performance

Figure 2.1 Effect of binder content on the water tightness of several RCC dams, data from Dunstan (1988).

where Kw = water-permeability (10 −9 cm/s) and kT = Torrent air-permeability (10 −16 m²). The relation in Eq. (2.1) was derived from the investigation reported by Sakai et al. (2013), discussed in Section 3.5.5 (Figure 3.12). In the case of roller-compacted concrete (RCC) dams, the material’s characteristics and the construction techniques present some challenges in this respect. The permeability of the RCC mass and that of the horizontal lift surfaces are key elements for the performance of hydraulic RCC structures (USACE, 2000). The permeability of RCC depends on the mix proportioning (especially on the binder content); a survey conducted by Dunstan (1988) showed that the water-tightness of RCC dams (measured in situ) increased with the amount of binder in the mix, see Figure 2.1. The permeability of the dam depends also on the placing techniques and the use of bedding mortar along lift surfaces, sometimes complemented by the use of impermeable membranes in the upstream face, as used for Urugua-í dam in Argentina, the RCC of which contained the record low 60 kg/m3 of cement (one of the authors was involved in its mix design) (Dam Search, 2021). Yet, the dam was not 100% watertight (Schrader, 2003). 2.3.1.3 Pervious Concrete In all the cases discussed in this chapter, a performing concrete is one having low permeability to either gases or liquids. In the case of pervious concrete, the opposite is true, as the goal is to achieve a pavement concrete

@seismicisolation @seismicisolation

Permeability as key concrete property 31

that is sufficiently strong to resist the applied loads, but highly permeable to allow the drainage of water from its surface through the body of the material. Pervious concrete is a special type of concrete which has an open structure which allows water to freely percolate through the pavement into the ground. The concrete is manufactured with uniform, open-graded coarse aggregate, cement and water and little or no fines. Elimination of fines creates a void structure in the finished concrete which allows fluids to rapidly pass through the pores and into the subgrade. This characteristic is very useful for reducing the rate and quantity of storm water runoff. The void structure, when coupled with an aggregate subbase, will slow down the rate of runoff and store substantial quantities of storm water. Pervious concrete is aimed for flatwork concrete applications which have a direct impact on storm water management. These include pedestrian walkways, parking areas, residential streets and areas with light traffic. Its open structure makes pervious concrete a good sound insulator and is also used for noise barrier construction. The void content of pervious concrete can range from 15% to 35%, with typical compressive strengths of 3.0–30 MPa. The drainage rate of pervious concrete pavement will vary with aggregate size and density (and strength) of the mixture, but will generally fall into the range from 0.15 to 1.25 cm/s (ACI 522R, 2010). Test methods exist to measure the draining rate of a pervious concrete, typically the constant head and the falling head permeameters (Sandoval et al., 2017). One of the latter has been standardized as ASTM C1701 (2009), which can be applied in the field. 2.3.1.4 Liquid Gas Containers Storage containers for liquified gases, typically liquified natural gas (LNG), are made with prestressed concrete, with or without an alloy steel liner. Avoiding the liner is more cost effective and safer (precluding the risk of rupture of the steel liner), but the challenge is to place a concrete that is sufficiently liquid-tight at cryogenic temperatures (Hanaor, 1985). An important aspect of the mix design is the thermal compatibility of aggregates and cement paste, so as to minimize differential internal thermal stresses and microcracks formation. In ACI 376 (2011), a maximum intrinsic concrete permeability to cryogenic fluids (typically liquid N2) of 0.01 × 10 −16 m² is specified, remarking that the cryogenic permeability of even partially-dried concrete is approximately half that obtained at ambient temperature. This reduction in permeability is associated with the formation of ice in the concrete pores, blocking them and also improving the mechanical performance (higher strength, E-modulus, tensile strain capacity, etc.) (Kogbara et al., 2013).

@seismicisolation @seismicisolation

32 Concrete Permeability and Durability Performance

2.3.2 Permeability for Gas Containment 2.3.2.1 Evacuated Tunnels for High-Speed Trains The “Swissmetro” project was launched in Switzerland in the early 1990s, aimed at establishing a high-speed Maglev (magnetic levitation) rail system. Trains, under magnetic sustentation, would run at speeds of up to 500 km/h, exclusively through small Ø5 m tunnels. To achieve such speeds economically, it was planned to operate the trains along partly evacuated tunnels. Details on the project can be found in Cassat et al. (2003) and Cassat and Espanet (2004); the project was abandoned, reportedly due to lack of political and financial support (RTS, 2010). However, the concept was revived around 2012 in the USA, with the name “Hyperloop”, with the more ambitious goal of trains running at twice the speed of planes, even at hypersonic speeds. To operate efficiently, the system requires magnetically levitated trains to run on tunnels evacuated to air pressures of the order of 1 mbar (100 Pa). In the case of the Swissmetro, the trains would operate along concretelined tunnels, whilst Hyperloop considers steel tubes. It is clear that, for the former, a low air-permeability concrete is required to achieve and maintain the required low air-pressure inside the concrete tunnels efficiently. For that purpose, air-permeability tests (Cembureau method, described in Section 4.3.1.2) on cores drilled from panels simulating the Swissmetro tunnels walls were performed at TFB laboratory in Switzerland (Badawy & Honegger, 2000), as well as using the Torrent method (see Chapter 5) (Badoux, 2002). The role of cracks on the air-permeability was also investigated (Badoux, 2002), a topic dealt with in Section 6.11. Probabilistic numerical modelling of the air-tightness of concrete-lined tunnels has been performed in the context of the Korean Super-Speed Tube Transport (SSTT), designed for trains running at 700 km/h. The modelling involved parameters such as concrete permeability and wall thickness, as well as the diameter of the tube structure, complemented by experimental investigation on the effect of joints and connections of the concrete tube (Park et al., 2015). The modelling was extended to include the effect of cracks on the performance of the evacuated tunnels (Devkota & Park, 2019). 2.3.2.2 Underground Gas “Batteries” It is of economic interest to store the energy surplus, generated in periods of low demand, in such a way that it can be recovered in periods of high demand. This has been traditionally done, in mountainous geographical regions, by pumping up water to hydropower reservoirs, to be used later as hydroelectric energy. Compressed air energy storage (CAES) is a concept in which the surplus energy generated by wind turbines and solar cells is used to compress air, @seismicisolation @seismicisolation

Permeability as key concrete property 33

stored in underground caverns in solid bedrock (SINTEF, 2017). At the appropriate time, the air pressure is released through a gas turbine that generates electricity. The cavern wall structure is capable, in interaction with the surrounding rock, to resist pressures of over 20 MPa and highfrequency operations. Although the structural function of the reservoir is fulfilled by concrete elements, some such systems rely on steel lining inside the gas reservoir for gas tightness (Tengborg et al., 2014). In other cases, concrete linings of low permeability that are appropriately reinforced against potential tensile fracturing due to high air storage pressure, allow the more flexible realization of underground CAES and result in significant construction cost reduction. The system relies on the permeability of the concrete lining and of the surrounding rock for the design of long-term air-tightness performance. Numerical modelling indicated that a concrete lining with air-permeability less than 0.01 × 10 −16 m² would result in an acceptable air leakage rate of less than 1%, with the operation pressure ranging between 5 and 8 MPa at a depth of 100 m (Kim et al., 2012). In order to get experimental data to feed the model, large-scale mock-up permeability tests were conducted, comprising not just the concrete itself, but also the joints and their treatment, the latter becoming critical for the air-tightness of the system (Kim et al., 2014).

2.3.3 Permeability for Radiation Containment 2.3.3.1 Radon Gas Radon (Rn) is a colourless, odourless and tasteless radioactive noble gas which occurs naturally in soils in amounts depending on the local geology (CIP 18, 2000). Some concerns on the presence of this gas exist due to its association with the development of lung cancer. This happens when it accumulates, due to its high density, in low areas like basem*nts or crawl spaces. Radon decays to other radioactive elements in the uranium series, called “radon progeny” that exists as solid particles that can become attached to dust particles in the air. If inhaled, they can lodge in the lung, where they can cause cancerous tissue damage due to energy emitted during radioactive decay. Concrete constitutes an effective barrier to radon penetration if cracks and openings are sealed. The World Health Organization (WHO) regards Rn as a health hazard, constituting the second cause of lung cancer, after tobacco. To quantify the problem, in Canton Ticino (Switzerland), a huge campaign of site measurements was run, collecting data from ca. 50,000 dwellings by 2010 (LC, 2010). Out of them, 91% had Rn concentration values below 400 Bq/m³, 7% within 400–1,000 Bq/m³ and 2% above 1,000 Bq/ m³. The Swiss Federal Bureau of Public Health establishes a limit of 1,000 Bq/m³ for dwellings. For new constructions or refurbishing, the applicable limit is 400 Bq/m³, with a recommendation of not exceeding 300 Bq/m³, on the basis of WHO guidelines (LC, 2010). @seismicisolation @seismicisolation

34 Concrete Permeability and Durability Performance

The main transport mechanism of penetration of Rn from the soil into the dwellings is gas-permeability, through the so-called “chimney-effect”, by which the warm air moving upwards generates a small depression in the ground floor and basem*nt, that aspirates gas from the surrounding ground. This is aggravated in winter time by heating. Calculations of the indoor radon entry rate by pressure-driven flow rely on the permeability coefficient of the concrete. The coefficient of air-permeability is the key transport parameter; in typical radon problems, the pressure difference across the foundation walls may range between 1 and 20 Pa (Abu-Irshaid & Renken, 2002). A test method to measure the coefficient of air-permeability, especially for that purpose, has been developed (Ferguson et al., 2001). The relation between Rn concentration and the coefficients of gas diffusion and air-permeability is discussed in Rogers et al. (1995), based on experimental measurements performed on 25 samples of new residential concretes in Florida (USA). It is claimed that the low permeability of concrete, even microcracked, compared with that of soils, makes it a very suitable material to protect from Rn emissions (Piedecausa García & Chinchón-Payá, 2011). An attempt to estimate the coefficient of radon-diffusion in concrete from pore structure measurements was presented in Linares (2015). A special laboratory test method to measure the permeability to Rn of cementitious composites has been developed at SUPSI (Teruzzi & Antonietti, 2019). The same authors are investigating the relation between Rn permeability and air-permeability using the Torrent method (Chapter 5) (Teruzzi, 2020). 2.3.3.2 Nuclear Waste Disposal Containers The adoption of nuclear power energy for the production of electricity includes, among the formidable challenges it poses, that of the disposal and storage of radioactive waste. The increasing amount of low and medium radioactive waste needs a serious concept of a long-term policy in radioactive waste management. Periods in the range of 300–1,000 years are considered in which the storing facilities have to guarantee the safety of human population and environment against radiation. The design and construction of many of the facilities and structures for long-term storage of radioactive waste materials employ reinforced concrete for support, containment, and environmental protection functions. During the desired very long service life of these structures, the reinforced concrete is to provide both physical and chemical barriers to isolate the waste from the environment. Citing Naus (2003): “Although a number of deteriorating influences can affect these structures, corrosion of embedded steel, leaching, elevated temperature, and irradiation (depending of the application) probably represent the @seismicisolation @seismicisolation

Permeability as key concrete property 35

greatest initial threats. Over the long term, leaching and cracking have increased importance as water will provide the transport medium for radionuclides should the other engineered barriers fail. In nearly all chemical and physical processes influencing the durability of concrete structures, dominant factors involved include transport mechanisms within the pores and cracks, and the presence of a fluid. Concrete permeability therefore is of significant importance relative to the long-term durability of radioactive waste facilities. Concrete permeability will vary according to such things as the proportions of constituents, degree of cement hydration, cement fineness, aggregate gradation, and moisture content. One of the most important factors affecting ionic transport through concrete is the presence of cracks. Cracking not only controls the quantity of ions transported, but can also control whether there will be any convective transport. Factors that contribute to increases in concrete permeability or cracking therefore are of importance to the durability of the radioactive waste management facilities.” To achieve the required performance, the use of high-performance steel fibre reinforced concrete was advocated in Slovakia, as reported by Hudoba (2007), who presents detailed drawings and pictures of the underground chambers with containers position and the act of placing a container in the storage chamber. Within this context, Kubissa et al. (2018) investigated the Autoclam airpermeability index (API), see Section 4.3.2.7, and moisture distribution on cores drilled from heavy-weight concretes (density around 3,250 kg/m³) of large-scale mock elements. In Argentina, O2-Permeability tests (Cembureau method, see Section 4.3.1.2) and site air-permeability tests (Torrent method, see Chapter 5) were performed on a concrete container to assess its long-term durability (Ramallo de Goldschmidt, 2004). Similar investigations were developed on cement-based materials for radioactive waste repositories in Japan (Kurashige et al., 2009 and Niwase et al., 2012). A report of the International Atomic Energy Agency (IAEA, 2013) emphasizes the relevance of concrete permeability for the durability of nuclear waste storage facilities, stating that porosity and permeability of Cement Waste Products, function of their composition, are directly related to diffusion and leaching characteristics of incorporated radionuclides. In Table 13 of IAEA (2013), three test methods applicable to measure the permeability of cementitious materials are listed, namely: water-permeability (ASTM D5084, 2010), gas-permeability (ASTM C577, 2019) and air-permeability (Torrent method, described in detail in Chapter 5). In Table17 of IAEA (2006), three NDT methods for measuring air-permeability are listed: Surface airflow method (Whiting & Cady, 1992) and, again, theTorrent method. In addition, test methods to measure water sorptivity are also included (all described in Chapter 4): ISAT, Figg and Autoclam (the latter, also for air-permeability). @seismicisolation @seismicisolation

36 Concrete Permeability and Durability Performance

Unless sensors are left inside the containers, concrete permeability can be measured only before they are put into service. There are indications that the gas and water-permeability of concrete exposed to radiation may increase with age, based on tests performed on cores drilled from the recovery of a nuclear reactor after 25 years of service (Mills, 1990). Monitoring of air-permeability (Torrent method), together with carbonation, electrical resistivity and electrochemical tests, was performed on concrete of the El Cabril disposal containers (Andrade et al., 2003). With regard to nuclear waste disposal, not just the permeability of the container matters, but also that of the surrounding rock, see Section 11.5.2.2. 2.4 PERMEABILITY AND DURABILITY As discussed in Sections 1.2 and 1.9, most deterioration actions affecting the durability of concrete structures (carbonation, chlorides and sulphates ingress, chemical attack and even frost) are related to the penetration of aggressive agents from the environment into the concrete. The mechanisms by which this penetration takes place are basically two: permeation (that includes capillary suction) and diffusion, which are described in detail in Chapter 3. Whatever the mechanism, a concrete that has an open pore structure (i.e. more and larger pores) or presents micro-cracks will deteriorate at a higher rate than a concrete with a tighter pore structure. The former will show a higher permeability than the latter, see Eq. (2.2), suggesting that the coefficient of permeability (to gases or liquids) is a sensitive indicator of the “penetrability” of a given concrete; this is discussed in detail in Section 3.8. Hence, a direct relation exists between the permeability of concrete and its durability performance (as this book is titled). In this respect, it is worth quoting Prof. K. Mehta (Mehta, 1991) who, in his review of the durability situation and research progress made at that time, expressed the following: “It seems that, in spite of some important discoveries valuable from the standpoint of durability enhancement, today more concrete structures seem to suffer from lack of durability than was the case 50 years ago. In order of decreasing importance, the major causes of concrete deterioration today are as follows: corrosion of reinforcing steel, frost action in cold climates, and physico-chemical effects in aggressive environments. There is a general agreement that the permeability of concrete, rather than normal variations in the composition of Portland cement, is the key to all durability problems.” This statement has full validity even today. The rest of this book is dedicated to studying the permeability of concrete and the relation of this important property with the durability of concrete structures. @seismicisolation @seismicisolation

Permeability as key concrete property 37

REFERENCES Abu-Irshaid, E. and Renken, K.J. (2002). “Relationship between the permeability coefficient of concrete and low-pressure differentials”. 2002 International Radon Symposium Proceedings, 13 p. ACI 318 (2019). “Building code requirements for structural concrete”. ACI 376 (2011). “Code requirements for design and construction of concrete structures for the containment of refrigerated liquefied gases”. American Concrete Institute, 170 p. ACI 522R (2010). “Report on pervious concrete”, 43 p. Andrade, C., Sagrera, J.L., Martínez, I., García, M. and Zuloaga, P. (2003). “Monitoring of concrete permeability, carbonation and corrosion rates in the concrete of the containers of El Cabril (Spain) disposal”. Trans. SMiRT 17, Prague, Czech Rep., August 17–22, Paper #O03-2, 8 p. ASTM C577 (2019). “Standard test method for permeability of refractories”, 5 p. ASTM C1701 (2009). “Standard test method for infiltration rate of in place pervious concrete”, 3 p. ASTM D5084 (2010). “Standard test methods for measurement of hydraulic conductivity of saturated porous materials using a flexible wall permeameter”, 23 p. Badawy, M. and Honegger, E. (2000). “Swissmetro – Tests on air permeability of concrete”. 16th IABSE Congress, Lucerne, Switzerland, 8 p. Badoux, M. (2002). “Experimental research for the liners of the Swissmetro tunnels”. MAGLEV 2002, Lausanne, Switzerland, September, Paper PP05107, 8 p. Cassat, A., Bourquin, V., Mossi, M., Badoux M., Vernez, D., Jufer, M., Macabrey, N. and Rossel, P. (2003). “SWISSMETRO – Project development status”. STECH ’03, Tokyo, Japan, 8 p. Cassat, A. and Espanet, C. (2004). “SWISSMETRO: Combined propulsion with levitation and guidance”. MAGLEV 2004, Shanghai, China, October 26–28, 10 p. CIP 18 (2000). “Radon resistance buildings”. Concrete in Practice, NRMCA, 2 p. Dam Search (2021). “Database of 906 RCC Dams”, http://www.rccdams.co.uk/ dam-search/. Devkota, P. and Park, J. (2019). “Analytical model for air flow into cracked concrete structures for super-speed tube transport systems”. Infrastructures, v4, n76, 13 p. Dunstan, M.R.H. (1988). “Wither roller compacted concrete for dam construction?”. Roller Compacted Concrete II. Hansen, K.D. (Ed.). A.S.C.E., New York, pp. 294–308. Di Pace, G. (2021). Private communication, June. Ferguson, L.J., Daoud, W.Z. and Renken, K.J. (2001). “Further measurements on the permeability coefficient of a concrete sample under low pressure differences”. The 2001 International Radon Symposium, Pre-Prints, 78–86. Görtz, J., Wieprecht, S. and Terheiden, K. (2021). “Dual-permeability modelling of concrete joints”. Constr. & Buildg. Mater., v302, October, 124090. Hanaor, A. (1985). “Microcracking and permeability of concrete to liquid nitrogen”. ACI J., March–April, 147–153.

@seismicisolation @seismicisolation

38 Concrete Permeability and Durability Performance Hudoba, I. (2007). “Utilization of concrete as a construction material in the concept of Radioactive Waste Storage in Slovak Republic”. Acta Montanistica Slovaca, v12, n1, 157–161. IAEA (2006). Ageing Management of Concrete Structures in Nuclear Power Plants. Series No. NP-T-3.5. International Atomic Energy Agency, Vienna, 372 p. IAEA (2013). The Behaviours of Cementitious Materials in Long Term Storage and Disposal of Radioactive Waste. Report IAEA-TECDOC-1701. International Atomic Energy Agency, Vienna, 75 p. Kim, H.-M., Lettry, Y., Ryu, D.-W., Synn, J.-H. and Song, W.-K. (2014). “Mock-up experiments on permeability measurement of concrete and construction joints for air tightness assessment”. Mater. & Struct., v47, n1, 127–140. Kim, H., Rutqvist, J., Ryu, D., Choi, B., Sunwoo, C. and Song, W. (2012). “Exploring the concept of compressed air energy storage (CAES) in lined rock caverns at shallow depth: A modeling study of air tightness and energy balance”. Appl. Energy, v92, April, 653–667. Kogbara, R.B., Iyengar, S.R., Grasley, Z.C., Masad, E.A. and Zollinger, D.G. (2013). “A review of concrete properties at cryogenic temperatures: Towards direct LNG containment”. Constr. & Bldg. Mater., v47, October, 760–770. Kubissa, W., Glinicki, M.A. and Dąbrowski, M. (2018). “Permeability testing of radiation shielding concrete manufactured at industrial scale”. Mater. & Struct., v51, Article 83, 15 p. Kurashige, I., Hironaga, M., Yoshinori, M. and Kishi, T. (2009). “Quality inspection system of cement-based materials supporting performance confirmation for radioactive waste repository. Part 1. Non destructive evaluation of rebound number and air permeability of surface concrete of in-situ structures and laboratory specimens”. Manag. Radioact. Wastes Non-Radioact. Wastes Nucl. Facilit. (S12), v41, n15, August, 1–15 (in Japanese). LC (2010). “Radon nelle abitazioni ticinesi: resoconto campagna 2009–2010 (Lugano campagna) e conclusione del programma di misurazione a tappeto 2005–2010”. Labor. Cantonale, Bellinzona, January 5, 17 p. Linares, P. (2015). “Caracterización del hormigón en relación a la difusión de gases y su correlación con el Radón”. PhD Thesis, Univ. Polit. Madrid, Spain, 220 p. Mehta, P.K. (1991). “Durability of concrete – Fifty years of progress?” ACI SP-126, 1–32. Mills, R.H. (1990). “Gas and water permeability tests on 25 years old concrete from the NPD nuclear generation station”. Report Info-0356, Atomic Energy Control Board, Project No. 2.147.1, May, 56 p. Naus, D.J. (2003). “Use of concrete in radioactive waste disposal facilities”. RILEM TC 202-RWD, General Information, 5 p. Neville, A.M. (1995). Properties of Concrete. 4th ed., Longman Group Ltd., Harlow, 844 p. Niwase, K., Sugihashi, N., Edamatsu, Y. and Sakai, E. (2012). “Study on the influence of fly-ash quality on the properties of cementitious materials and the applicability of non destructive test for quality control at radioactive waste disposal facility in Japan”. Concr. Res. and Technol., v23, n1, 13–24 (in Japanese).

@seismicisolation @seismicisolation

Permeability as key concrete property 39 Park, C.-H., Synn, J.-H. and Park, J. (2015). “Probabilistic performance assessment of airtightness in concrete tube structures”. KSCE J. Civil Eng., v20, 1443–1451. Piedecausa García, B. and Chinchón-Payá, S. (2011). “Radiactividad natural de los materiales de construcción. Radiación interna: el gas radón”. CementoHormigón, n946, September–October, 34–50. Powers, T.C. (1958). “Structure and physical properties of hardened portland cement paste”. J. Amer. Ceramic Soc., v41, n1, January 1, 6 p. Powers, T.C., Copeland, L.E. and Mann, H.M. (1959). “Capillary continuity or discontinuity in cement pastes”. J. Portl. Cem. Assoc. Res. and Development Labs., v1, n2, 38–48. Ramallo de Goldschmidt, T.R. (2004). “Low and intermediate level waste package assessment under interim storage and final disposal conditions”. Report IAEA-TECDOC-1397, International Atomic Energy Agency, June, 33–50. Rogers, V.C., Nielson, K.K. and Holt, R.B. (1995). “Radon generation and transport in aged concrete”. EPA/600/SR-95/032 Report, Project Summary, March, 2 p. RTS (2010). “Le projet Swissmetro abandonné faute d‘argent”. RTS Info, June 28. Sakai, Y., Nakamura, C. and Kishi, T. (2013). “Correlation between Permeability of Concrete and Threshold Pore Size obtained with Epoxy-Coated Sample”. J. Adv. Concr. Technol., v11, August, 189–195. Sandoval, G.F.B., Galobardes, I., Teixeira, R.S. and Toralles, B.M. (2017). “Comparison between the falling head and the constant head permeability tests to assess the permeability coefficient of sustainable Pervious Concretes”. Case Studies Constr. Mater., v7, 317–328. Schrader, E.K. (2003). “Performance of roller compacted concrete (RCC) dams – An honest assessment”. Roller Compacted Concrete Dams, Berga, L., Buil, J.M., Jofré, C. and Chonggang, S. (Ed.). A.A. Balkema Publishers, Lisse, NL, 91–102. SINTEF (2017). “Air could be the world’s next battery”. PhysOrg, March 28, 5 p. Tengborg, P., Johansson, J. and Durup, J.G. (2014). “Storage of highly compressed gases in underground Lined Rock Caverns – More than 10 years of experience”. Proceedings on World Tunnel Congress 2014 – Tunnels for a better Life. Foz do Iguaçu, Brazil, 8 p. Teruzzi, T. (2020). Personal Communication, June 2. Teruzzi, T. and Antonietti, S. (2019). “PERMEA – Permeabilità al gas radon dei conglomerati cementizi”. Clickin, SUPSI, Lugano, Switzerland, Giugno, 2 p. USACE (2000). “Roller-compacted concrete”. Engng. Manual, U.S. Army Corps of Engineers, EM1110-2-2006, January 15, 77 p. Whiting, D. and Cady, P.D. (1992). “Condition evaluation of concrete bridges relative to reinforcement corrosion. Volume 7: Method for field measurement of concrete permeability”. SHRP-S/FR-92–109 Report, National Res. Council, Washington DC, September, 93 p.

@seismicisolation @seismicisolation

Chapter 3

Theory: concrete microstructure and transport of matter

3.1 CEMENT HYDRATION

3.1.1 M ain Hydration Reactions and Resulting Changes The main component and “heart” of Portland cements is the Portland cement clinker, the product obtained as the main output of the chemical process developed in the cement kiln, at temperatures reaching ≈ 1,450°C. Clinker is later ground into a very fine powder, together with a suitable amount of a set regulator (typically some form of calcium sulphate), to obtain the so-called ordinary Portland cement (OPC). Often, Portland cement incorporates other components of various degrees of cementitious contribution, such as granulated blast-furnace slag (GBFS), pulverized fly ash (PFA), natural or artificial pozzolans (POZ), limestone filler (LF), and microsilica such as silica fume (SF). Depending on the nature of the added component(s), the incorporation may happen in the cement plant (interground in the cement mill or blended in powder form) or directly at the concrete plant as a separate powder. In the first case, it is usual to designate the added component(s) as mineral components (MIC) whilst, in the second case, as supplementary cementitious materials (SCM). Clinker is an assemblage of different minerals, the main ones being1: • alite (main component), based on C3S • belite, based on C2S • aluminate and ferrite, based on C3A and C4AF, respectively The “pure” compounds C3S, C2S, C3A, thus, do not occur in this simple form, but as solid solutions containing a whole series of oxides such as MgO, Na2O, K 2O, Fe2O3, and Al2O3 (Holcim, 2011).

In cement chemistry notation: C = CaO; S = SiO2; A = Al 2O3; F = Fe2O3; H = H 2O.

DOI: 10.1201/9780429505652-3 @seismicisolation @seismicisolation

42 Concrete Permeability and Durability Performance

In cement chemistry, the term “hydration” denotes the totality of the changes that occur when an anhydrous cement, or one of its constituent phases, is mixed with water (Taylor, 1997). With respect to the formation of the microstructure of the hydrated Portland cement paste and concrete, the hydration of the two calcium silicate components – alite and, to a lesser extent, belite – plays a key role. The hydration of both minerals leads to the formation of calcium silicate hydrates (C-S-H) and calcium hydroxide (CH), as summarized in Eq. (3.1).

C sS + Water → C − S − H + CH C2S

(3.1)

In the case of cements containing MIC, other reactions take place, such as the hydration of ground GBFS (GGBFS) in the alkaline environment created by clinker hydration or the pozzolanic reaction of the reactive silica in low-calcium PFA, SF or natural or artificial POZ with the calcium hydroxide created during clinker hydration. Both reactions follow that of clinker hydration, ending up in further formation of C-S-H as reaction product. Both secondary reactions have the beneficial effect of increasing the amount of C-S-H but, on the negative side, are delayed and slower than clinker hydration reaction. The cement hydration reaction has the following consequences: • the volume occupied by the hydration products becomes approximately twice that originally occupied by the anhydrous cement • the specific surface of the hydration products is about 1,000 times that of the anhydrous cement • the reactions are exothermic, with each cement gram generating about 380 J of heat • radical changes in the microstructure, leading to a strong and stable material

3.1.2 Hydrothermal Conditions for Hydration (Curing) The chemical reactions involved in cement hydration (Eq. 3.1) can only take place under certain favourable conditions of temperature and moisture. The hydration reactions accelerate at higher temperatures, a fact made practical use of in the precast industry or whenever a high early strength is required (although often the “quality” of the microstructure at later ages, say 28 days, may suffer). At the other extreme, hydration slows down at low temperatures, a fact to be taken into account in what is called “cold weather concreting”. It is generally accepted that hydration virtually stops at temperatures below −10°C. @seismicisolation @seismicisolation

Microstructure and transport of matter 43

By definition, the hydration reactions need the presence of water. Water added to concrete during mixing becomes chemically bound and physically bound (adsorbed), both bonding processes being relevant for the progress of hydration (ACI 308R, 2016). About 0.25 g of water is required to completely hydrate 1 g of cement, water that is removed from the pores in a process called “self-desiccation” (of relevance for concretes of low w/c ratios). According to Powers et al. (1954), for cement hydration to proceed, the relative humidity in the pores should be at least 80%. This relative humidity can be achieved, either by providing extra water externally (ACI 308R, 2016) or internally (Zhutowsky & Kovler, 2012; Sensale & Gonçalves, 2014; Weiss & Montanari, 2017) or, alternatively, by preventing the non-reacted water to evaporate by protecting the concrete surface using plastic sheets, wet burlaps or curing compounds. These processes are restrictedly known as “curing”. According to Xue et al. (2015), once water is lost, the hydration of the cement will cease, reason why it is important to keep sufficient moisture in concrete while the cement is actively hydrating, especially at early ages (Ye et al., 2005). For mixes with low w/c ratio, preventing the evaporation of water may not be sufficient and supplying water, at least in an initial stage, will improve concrete performance (Maslehuddin et al., 2013). According to Al-Gahtani (2010), water curing is the most effective method of curing. When using SCM, water is required for the pozzolanic reaction between such materials and calcium hydroxide to take place in the later stages of hydration of cement (Bentur & Goldman, 1989; Sensale & Gonçalves, 2014; Maslehuddin et al., 2013). Actually, the pozzolanic reaction requires higher relative humidity than that required for the cement hydration (Ye et al., 2005). Therefore, cements containing MIC or concretes containing SCM are generally more sensitive to the lack of adequate curing (evaporation, low temperatures), due to the delayed occurrence of the secondary reactions involved in the hydration of the mineral additions. 3.2 MICROSTRUCTURE OF HARDENED CONCRETE

3.2.1 Overview To study the transfer of matter through concrete, it is important to have a clear understanding of the microstructure of the material. Here, an overview of this rather complex topic is provided, that is more extensively treated in several excellent classical books on concrete, e.g. in Chapter 1 of Neville (1995), in Chapter 4 of Mindess et al. (2003) and in Chapter 2 of Mehta and Monteiro (2006). First of all, hardened concrete is a heterogeneous material composed of two main macroscopic phases, well differentiated morphologically and behaviourally: the aggregates and the hardened cement paste (h.c.p.) matrix. @seismicisolation @seismicisolation

44 Concrete Permeability and Durability Performance

As shown in Figure 3.1, the aggregates are embedded in the h.c.p. matrix, such that – for conventional concrete – there is no physical contact between the aggregate particles, as they are all coated by a thin layer of cement paste, which also fills the voids between the individual particles. The h.c.p. matrix constitutes a continuous phase, while the aggregate particles are a discrete phase. Hence, it would be possible for a “microorganism” to traverse the entire concrete sample travelling exclusively along the h.c.p. This indicates that the h.c.p. has a key role in the transfer of matter through concrete, with isolated aggregate particles playing just a secondary role. It is worth mentioning that the permeability of aggregates (rocks) covers the same range as that of concrete, as discussed in Section 11.5.2. A synonym for “aggregate” in Italian, Portuguese and Spanish languages is “inert”, indicating a totally passive role of the particles. Alkali-aggregate reaction is just one example that aggregates are not inert but that they can react chemically with the h.c.p. Moreover, since long it has been recognized that the contact zone between aggregates and h.c.p. is influenced by both phases, in what nowadays is called the “Interfacial Transition Zone” (ITZ). The ITZ consists of a 10–50 μm thick “rim” around the aggregates and can be considered as a third distinctive phase of concrete microstructure (together with h.c.p. and the aggregates) (Mehta & Monteiro, 2006). The ITZ is the result of many factors, such as the accumulation of bleeding water under the coarse particles, of unwashed aggregates coated with “dirt” or, more generally, of selective deposition of Ca(OH)2 crystals closer to silicate aggregates, to chemical reactions between paste and aggregate, etc. The fact is that, typically, the ITZ is more porous than the bulk h.c.p., as can be seen in the example presented in Figure 3.2, which shows a thin section of concrete (20–25 μm thick) taken from a sample that was previously

Figure 3.1 Aspect of concrete internal structure.

@seismicisolation @seismicisolation

Microstructure and transport of matter 45

Figure 3.2 More porous rim around aggregates.

vacuum-impregnated with a low-viscosity resin containing a fluorescent dye. The microscopy observation of the thin section under transmitted UV light reveals very clearly the microstructure of the concrete, in dark what is solid (typically aggregate particles) and in light colour what is porous, which is brighter the more porous the phase. It can be seen quite clearly that a rim of more porous h.c.p. contours the aggregates, not just the coarse, but also the fine particles. The characteristics and role of the ITZ is further discussed in Section 3.2.3; those more interested in the topic may consult (Alexander et al., 1999).

3.2.2 M icrostructure of Hardened Cement Paste Let us try to understand the microstructure of h.c.p. During hydration of cement (Eq. 3.1), the predominant calcium silicate crystals in clinker (C3S and C2S) react with water to form calcium silicate hydrates (C-S-H) and Ca(OH)2 . C-S-H makes up to 50%–60% of the volume of h.c.p., in the form of tiny fibrous crystals (100 30–100 10–30 3–10 1.0

9.0– 9.5 1,000 300– 1,000 100–300 30–100 10

18,000 seconds, which fit well to the range in Table 4.2. Clearly, the hole and tube dimensions (volume of depressed air V) influence the rate of pressure variation and therefore the test result. Hence, a durability indicator which is independent from V and from the initial and final pressure may be computed. The air exclusion rate (AER) is calculated through: AER =

(

)

pi + pf pi ⋅ V ⋅ pf 200

(4.13)

where AER = air exclusion rate (s/mL) V = volume of depressed air (mL) pi = initial pressure (kPa) pf = final pressure (kPa) t = time for pressure variation between pi and pf (s) The common range of AER in concrete is 25–75 s/mL (Concrete Society, 1987). The Figg test enjoys some popularity and some modifications concerning the hole dimensions were proposed by Kasai et al. (1984) and Neves and Gonçalves (2006). 4.3.2.2 Hong-Parrott This intrusive method developed by Hong and Parrott (1989) and Parrott and Hong (1991) assesses air-permeability by means of a Ø20 and 35 mm deep hole, partially sealed with an expansive plug, leaving a test chamber at the bottom of the hole. The hole may be cast, drilled or cored. The test consists in pressurizing the test chamber by pumping air until reaching a relative pressure around 0.75 bar. The air inlet valve is closed, isolating the test chamber, the pressure of which will decrease with time due to the air flowing through the concrete around the hole. The pressure inside the hole is monitored and a stopwatch measures the time that takes for the pressure to decay from 0.6 to 0.4 bar. For less permeable concrete, the pressure decay in 5 minutes, starting from 0.6 bar, is measured instead. Based on these measurements, an air-permeability coefficient can be calculated as follows: K=

( (

) )

c pi − pf ⋅ t pi + pf

(4.14)

@seismicisolation @seismicisolation

Test methods for concrete permeability 105

where K = air-permeability coefficient (m 2) c = factor that depends on the volume of pressurized air pi = initial pressure (bar) pf = final pressure (bar) t = time for pressure variation between pi and pf (s) This permeability coefficient usually varies between 0.01 × 10 −16 and 1.0 × 10 −16 m 2 . Besides the measurements being carried out under a nonsteady-state flow of air, the concrete surface crossed by the air flow (estimated by applying a soap solution around the cell) is not constant during the test. A model to predict carbonation-induced corrosion risk has been developed, based on the result of this test (Parrott, 1994); it is described in Section 9.3.1. 4.3.2.3 Paulmann The intrusive Paulmann method is named after its main author (Paulmann & Rostasy, 1989). A Ø10 × 40–45 mm hole is drilled in the concrete surface whereas the deeper 25 mm of the hole is subjected to pressurized (2 bar) N2 by means of an injection packer that seals the remaining part of the hole. Then the pressurized gas starts to permeate through concrete and part of the flow is collected and measured on a predefined area by means of a guard ring located around the hole. As the length travelled by the flow can be estimated, a permeability coefficient can be calculated from the flow rate measurement in the guard ring and the pressure applied in the central hole. The usual range of values for the permeability coefficient according to the Paulmann method is reportedly like that of the Cembureau method. 4.3.2.4 TUD The TUD (Technical Univ. Delft) intrusive test method was developed by Reinhardt and Mijnsbergen (1989) and later modified by Dinku (1996). Like the Hong-Parrott method, it assesses concrete permeability in a drilled hole with pressurized gas. However, in this case, the hole is Ø14 and 45 mm long, the gas is nitrogen and much higher pressures are applied. A cylindrical hollow probe with a rubber ring near its tip is introduced in the hole. Tightening a screw nut at the upper end of the probe expands the ring against the hole wall to seal the test chamber cavity. In this way, the disposable rubber plugs, required in the Figg method, are avoided. Then the cavity is filled with nitrogen, through the probe hole, at approximately 12 bar. After waiting a few seconds for the pressure to stabilize, a

@seismicisolation @seismicisolation

106 Concrete Permeability and Durability Performance

valve in the pressure line is closed and the pressure starts to decrease. The time interval between the pressure levels of 11.0 and 10.5 bar is recorded and used as an indicator of concrete permeability. Values of TUD time ranging from 5 to 200 seconds are reported by Reinhardt and Mijnsbergen (1989), concerning non-cured and well-cured concrete mixes with w/c ratios of 0.6 and 0.4. Reported data in Tables 3.1-III and 3.2-IV of Torrent and Ebensperger (1993), obtained on a wide variety of concrete qualities, yielded TUD values in the range between 6 and 3,250 seconds. A simple equation, based on the Hagen-Poiseuille law, that is valid for a total volume of 94 cm3 of pressurized gas, was suggested by Dinku and Reinhardt (1997) to calculate a gas-permeability coefficient: K=

105.09 ⋅ 10−17 t

(4.15)

where K is the gas-permeability coefficient in m 2 and t is the time elapsed between pressure levels of 11 and 10.5 bar, in seconds. According to the same research, K values from 7 × 10 −18 to 2 × 10 −17 m 2 correspond to average permeability concrete. 4.3.2.5 GGT The intrusive Germann Gas Test (GGT) is based on the work by Hansen et al. (1984). It requires drilling a Ø18 mm hole at a small angle to the concrete surface, so that the end of the hole is at an approximate depth of 25 mm from the concrete surface; a pressure sensor is located at the end of the hole that is sealed. Exactly above the sensor, a test rig is clamped with a sealing gasket. Compressed (1–4 bar) carbon dioxide is applied to the concrete surface inside the rig. Then, part of the pressurized CO2 flows into the drilled hole increasing the pressure inside it. The pressure rise inside the hole is monitored by the pressure sensor. Following a theoretical study carried out by Hansen et al. (1984), it is possible to calculate a coefficient of gas-permeability, from the test data and concrete porosity. The values of the permeability coefficient obtained by the GGT method are reportedly comparable with those obtained by the Cembureau method. A commercial GGT was once produced by Germann Instruments, but apparently not anymore. 4.3.2.6 Paulini The intrusive Paulini method was developed by Paulini and Nasution (2007), initially for laboratory testing, later modified to allow for site testing (Paulini, 2010). The site test requires drilling a Ø30 mm hole in @seismicisolation @seismicisolation

Test methods for concrete permeability 107

Figure 4.14 Sketch of the Paulini method.

the concrete surface, where a packer will be lodged (the bore dust is collected for humidity measurement and eventual chemical analysis), see Figure4.14. The packer contains a plug that is expanded by means of a threaded rod, thus sealing the lowest part of the hole and anchors the reaction to the force that will compress an aluminium plate towards a rubber gasket disk that seals the concrete surface around the hole. The first 10 mm of the drilled hole are left unsealed to allow air permeation through surrounding concrete. The air inlet is connected to a bottle of compressed air, equipped with a control valve to allow pressure regulation. The air pressure inside the hole is monitored, and the pressure drop is used to calculate the flow rate Q. Based on test data, an air-permeability coefficient may be computed through the following equation, developed by Paulini (2010). φ  Q⋅µ ⋅pa K = ln  d   φh  2 ⋅ π ⋅h⋅ pi2 − pa2

(

)

(4.16)

where K = permeability coefficient (m 2) ϕd = outside diameter of the rubber gasket disk (m) ϕh = diameter of the hole (m) h = unsealed depth of the hole (m) Q = computed air flow (m3/s) µ = viscosity of air (Pa.s) pa = atmospheric pressure (Pa) pi = applied gas pressure (Pa) The test is run by applying a minimum of three pressure levels, controlled by means of the bottle’s valve, starting with the higher one (6 bar). A time derivative of air pressure inside the hole is computed to check if constant flow has been reached. Data for Q computation are only considered after steady state being achieved at each applied pressure level. @seismicisolation @seismicisolation

108 Concrete Permeability and Durability Performance

Furthermore, Paulini and Nasution (2007) suggested a power law for gas permeation in concrete:  p v = vref    L

(4.17)

where v = air flow velocity (m/s) vref = reference air flow velocity (m/s) p = applied absolute gas pressure (MPa) L = length travelled by the air flow inside concrete (m) n = permeability exponent (-) If flow velocities are plotted against pressure gradients (p/L) in a log-log scale a nearly linear relationship shall be expected. The slope of that line is the permeability exponent, whereas the intersection with the vertical axis is the reference air flow velocity. Higher permeability exponents and lower reference velocities correspond to denser pore structures of concrete. Therefore, both parameters can be used as permeability indexes, particularly the permeability exponent, as it has a positive association with concrete quality, and therefore stands for the “Permeability Exponent” method designation. When assessing concrete permeability through the Paulini method, values of air-permeability coefficient in the order of magnitude of 10 −16 m 2 shall be expected, while n values between 1 and 2 and vref values between 10 −4 and 10 −6 m/s correspond to regular/low concrete permeability (see Table 4.2). 4.3.2.7 Autoclam System The same Autoclam system device, already described in detail for waterpermeability and sorptivity measurement (Sections 4.1.2.2 and 4.2.2.4), serves also to measure the air-permeability of concrete. It is a surface method but requires drilling three small holes per testing spot, to attach the measuring cell onto the tested concrete, ensuring a tight connection. After attaching the cell, the air inside its chamber is pressurized and the pressure decay, due to the air from the chamber flowing through the concrete, is monitored. When plotting the natural logarithm of pressure vs. time, a quasi-linear relation is expected, the slope of which is taken as the test result, named air-permeability index (API). The API is usually expressed in ln(bar)/minute and, for concrete, values in the order of magnitude of 10 −1 ln(bar)/minute shall be expected. Table 4.2 presents a classification of concrete qualities, based on Autoclam API.

@seismicisolation @seismicisolation

Test methods for concrete permeability 109

4.3.2.8 Single-Chamber Vacuum Cell The surface single-chamber method was developed, almost simultaneously, by Bérissi et al. (1986) and Schönlin and Hilsdorf (1987) and was further developed by Imamoto et al. (2006). The test principle is to create a vacuum on the concrete surface by means of a single chamber cell. Then, the pressure gradient between the pores of the surrounding concrete (at atmospheric pressure) and the evacuated chamber drives air through concrete into the chamber. The vacuum inside the chamber also holds it “stuck” onto the concrete surface by means of a rubber ring (Figure 4.15). A tight connection between the chamber and the concrete surface must be ensured, to avoid air flowing through the connection. Therefore, a ring made of convenient rubber and with appropriate thickness is necessary between the chamber borders and concrete surface. The basic setup is shown in Figure 4.15. The depression is created by means of a vacuum pump, or a syringe, connected to the chamber, where a valve is closed when the intended vacuum (e.g. 20 mbar, absolute pressure) is reached. Afterwards, the pressure rise inside the chamber is monitored and used to assess concrete permeability. Schönlin and Hilsdorf (1987) proposed an equation to calculate a permeability index, later amended in its units in Chapter 3 of RILEM TC 189NEC (2007): M=

(

)

2 pf − pi Vc f

) (

− t i pa 2pa − pf − pi

)

where M = permeability index (m3/s/mbar) pf = absolute chamber pressure at time tf (mbar) pi = absolute chamber pressure at time ti (mbar)

Figure 4.15 Sketch of single-chamber vacuum cell test method.

@seismicisolation @seismicisolation

(4.18)

110 Concrete Permeability and Durability Performance

pa = atmospheric pressure (mbar) Vc = inner volume of chamber and accessories (m3) tf = time at measurement of pf (s) ti = time at measurement of pi (s) For regular concretes, values of the permeability index within the range from 10 −9 to 10 −4 m3/s/mbar are expected. When testing trowelled surfaces, overestimation of concrete permeability was reported and attributed to a possible preferential flow path along the thin top superficial layer (Torrent, 1992). To overcome this potential shortcoming, Gabrijel et al. (2008) suggested spraying the near 100 mm of concrete surface around the chamber with a transparent car coating. 4.3.2.9 Double-Chamber Vacuum Cell (Torrent) The surface, double-chamber cell method, developed by Torrent (1992) can be considered an evolution of the single chamber method, as it is also non-intrusive and operates under vacuum. Indeed, this method comprises two concentric chambers that are permanently kept at the same pressure. The external chamber, besides preventing possible air flow through the concrete skin near the inner (measurement) chamber, ensures that the air flow into the latter is basically unidirectional. This last feature allows the derivation of a formula to calculate a coefficient of air-permeability of the material. This was the first test method for site measurement of air-permeability to be standardized (2003), now updated in Annex E of Swiss Standard (SIA 262/1, 2019). The above-mentioned standard not only prescribes how the test must be performed (age, sampling, temperature and moisture limits, calibration and testing procedure) but also specifies maximum limits to the values obtained on site, depending on the environmental conditions to which the element is exposed and defines conformity criteria for acceptance of the end-product. This test method is thoroughly addressed in Chapter 5. Table 4.2 presents a classification of concrete qualities, based on Torrent air-permeability coefficient kT. 4.3.2.10 Triple-Chamber Vacuum Cell (Kurashige) This test method, developed by Kurashige (2015), consists in adding a third concentric chamber to the Torrent method’s cell, but removing the pressure regulator. After 60 seconds of evacuation by the vacuum pump, the three concentric cells are isolated from the pump, their pressure rising naturally as function of the permeability characteristics of the underlying concrete. A numerical model of the gas flow, complemented by assumptions on the relation between permeability and porosity and between permeability and depth @seismicisolation @seismicisolation

Test methods for concrete permeability 111

from the surface, allows the determination of the parameters that provide a best fit to the experimentally obtained pressure-time curves for the three chambers. A Japanese patent has been filed covering this test method. 4.3.2.11 Zia-Guth The Zia-Guth method, named after its authors (Guth & Zia, 2001), operates like the Torrent method, as it is a surface test and the probe has two concentric chambers that are evacuated by means of a vacuum pump. Then, the external chamber valve is opened, allowing air entrance at atmospheric pressure that flows into the inner chamber through a near surface path. The pressure increase in the inner chamber is monitored. The pressure vs. time data are compared with predefined curves, corresponding to specific permeability coefficients, established by numerical modelling. The ‘best fit’ to the theoretical predictions provides a coefficient of air-permeability expressed in m 2 which, for normal quality concrete, ranges between 3 × 10 −16 and 6 × 10 −16 m 2 (Guth & Zia, 2000). 4.3.2.12 “Seal” Method The proposed surface test method called the “Seal” method (Ujike et al., 2009), aimed at measuring the air-permeability coefficient of the cover concrete on site, is sketched in Figure 4.16 (l.). An easily removable rubber latex resin is applied on the concrete surface forming a circle of radius r 2 leaving an uncovered inner circle of radius r 1 (Figure 4.16). The air-permeability of the concrete is measured by suctioning air out of the inner circle using a vacuum pump, with the area sealed by the airtight resin preventing sucking air from the surrounding concrete (Figure 4.16). A surface moisture meter is used to check the surface moisture of the concrete and the seal can be easily removed when the test has ended.

Figure 4.16 Air-permeability “Seal” test sketch (l.) and theoretical air hemisphere (r.).

@seismicisolation @seismicisolation

112 Concrete Permeability and Durability Performance

A model is applied, based on the sketch shown in Figure 4.16 (r.), assuming that the air-permeability coefficient is constant at any depth from the concrete surface and that the air-permeability zone forms a hemisphere, whose centre is the suction port of radius r 1, immediately below the vacuum chamber. The air-permeability coefficient is computed using the following equation (Kawaai & Ujike, 2016): k=

Q1 ⋅ P1 ⋅ µ  2 1  −   r2  π ⋅ P22 − P12  r1

(

)

(4.19)

where k = coefficient of air-permeability (m²) Q1 = air flow rate (m³/s), measured by the flow meter P1 = pressure in vacuum chamber (N/m²) P2 = atmospheric pressure (N/m²) μ = viscosity of air (N.s/m²) r 1 and r 2 = inner and outer radii of the seal (m)

4.3.3 Assessment of Concrete Quality by Gas-Permeability Test Methods Table 4.2, an extension of the one included in RILEM TC 230-PSC (2016), presents a comparison on how several gas-permeability test methods rate concrete quality, based on their test results. It is interesting to note that four to five quality categories can be established to classify concrete quality by means of gas-permeability test methods. 4.4 COMPARATIVE TEST RILEM TC 189-NEC

4.4.1 Objective and Experiment Design Within the frame of the work of RILEM TC 189-NEC “Non-destructive evaluation of the concrete cover”, a Comparative Test was organized with the declared objective of establishing whether the site methods available at the time (2003), designed to measure the “penetrability” of the concrete cover in the field, were capable of detecting differences in the w/c ratio and curing conditions of different concretes. In addition, their correlation with so-called “Reference” laboratory methods was evaluated. A full report on this Comparative Test can be found in RILEM TC 189NEC (2007) and a condensed report in Romer (2005), updated by Torrent (2008).

@seismicisolation @seismicisolation

Test methods for concrete permeability 113 Table 4.3 Test conditions investigated in the comparative test Test condition # Variable w/c ratio Cement type Moist curing (days) Temperature (°C) Storage condition a

1 0.40 7 20

0.55 0.60 0.40 0.55 OPC BFSCa 7 7 7 7 20 20 20 20 Normal

6 0.55 OPC 1 20

0.40 0.55 OPC 7 7 20 20 Moist

0.40 0.55 OPC 7 7 10 10 Cold

Blast-Furnace Slag cement, containing nominally 63% of slag.

The Comparative Test was a typical “blind” test in which a series of concrete panels (0.3 × 0.9 × 0.12 m), depicted in right-hand picture in Figure4.3, were cast with different concrete mixes by an independent laboratory (EMPA, Switzerland, in this case), moist cured and stored under prescribed conditions, leading to ten test conditions, summarized in Table 4.3. The tests were performed in the rooms where the panels were stored, with the following ambient conditions: Normal (20°C/70% RH); Moist (20°C/90% RH) and Cold (10°C). The age of the slabs at the initiation of the tests, which lasted 5 days, ranged between 54 and 69 days. Later, four cores were drilled from each of the ten slabs, cut to size, dried at 50°C, weighed and shipped to LNEC (Portugal) for testing, applying standardized or RILEM-recommended tests, under controlled laboratory conditions. These tests are referred to as “Reference Tests”. Another set of cores was drilled and shipped to the University of Cape Town for the determination of the South African Durability Indices. Both during the application of the site tests directly on the slabs as well as of the “Reference Tests” on the cores, the identity of the samples was coded. So, the tester did not know to which test condition the slabs or specimens belonged. Table 4.4 summarizes the main site tests applied directly on the panels and the laboratory “Reference Tests” applied on the cores drilled from the panels.

4.4.2 Evaluation of Test Results 4.4.2.1 Significance of Test Method This evaluation was aimed at establishing whether the test methods were capable of differentiating the “penetrability” of concretes of different w/c ratios (sets 1–2–3 for OPC and 4–5 for BFSC), of the same w/c ratio but different curing (sets 2–6) and of different w/c ratios for measurements conducted in moist room (sets 7–8) and cold room (sets 9–10). The capability of the methods was tested applying a Student’s t-statistical test of the

@seismicisolation @seismicisolation

114 Concrete Permeability and Durability Performance Table 4.4 Comparative Test: main “penetrability” tests applied on the panels Property measured Site tests Gaspermeability

Water sorptivity Electrical resistivity

Described in

Laboratory

Measurements per test condition

Autoclam Torrent

4.3.2.7 4.3.2.9 + Chapter 5

Hong-Parrott Autoclam

4.3.2.2 4.1.2.2/4.2.2.4

QUB (UK) TFB (CH)a IETcc (E)a LNEC (P) QUB (UK)

3 6 8 4 3

Wenner

A 2.2.2

TNO (NL)

4.3.1.2

LNEC (P)

Test method

Laboratory Reference tests Cembureau O2 – permeability Water RILEM sorptivity 116-PCD Chloride ASTM C1202 migration Tang-Nilsson Electrical Wenner resistivity

4.2.1 A 2.1.1 A 2.1.2 A 2.2.2

QUB, Queen’s Univ. Belfast; IETcc, Instituto Eduardo Torroja.aThe results obtained by both participants were very similar; only those of TFB are discussed here.

difference between the means of the pairs of sets of results under comparison, as shown in Table 4.5. The null hypothesis H0 was that both sets of results come from populations having the same mean “penetrability”. The alternative hypothesis H1 was that one set has a mean “penetrability” higher than the other as indicated in the second row of Table 4.5, for which a one-tailed test is applicable. The outcome of the statistical test was evaluated as follows: • if the result of the statistical test allows to reject the null hypothesis H0 at a level of significance < 1%, the differentiation capability of the test, for the particular sets compared, is “highly significant” (++) • if the result of the statistical test allows to reject the null hypothesis H0 at a level of significance between 1% and 5%, the differentiation capability of the test, for the particular sets compared, is “significant” (+) • if the result of the statistical test does not allow to reject the null hypothesis H0 at a level of significance of 5%, the differentiation @seismicisolation @seismicisolation

Test methods for concrete permeability 115 Table 4.5 Variables tested, differentiation capabilities of “Reference” and Site tests and correlations between them Compared sets

1–2

2–3

1–3

Variable tested

w/c OPC

Expected “penetrability”

2>1 3> 2 3>1

Reference test O2 permeability

4–5

2–6

7–8

9–10

w/c BFSC Curing w/c moist w/c cold 5>4

6>2

8>7

10>9

Differentiation capability ++

Water sorptivity

Cl− migration ASTM C1202 Cl− migration Tang-Nilsson Electrical resistivity

Differentiation capability o ++ ++ ++

Torrent air-permeability Hong-Parrott Autoclam water sorptivitya

R 0.67 / 0.90b 0.97

o ++

++ o

++ ++

++ +

+ ++

++ +

++ ++

0.92 0.47

Wenner resistivity

0.83

Site test Autoclam aira

a b

The internal RH reached by all panels exceeded the maximum 80% required by the test (Torrent,2008). After removal of outliers (Torrent, 2008).

capability of the test, for the particular sets compared, is “not significant” (o) • if the results are in reverse order than expected, the response of the test is “wrong” (--) Table 4.5 shows that all the selected “Reference Tests” were capable of correctly differentiating the “penetrability” of all sets at a highly significant (++) or significant (+) level, with the exception of the Wenner electrical resistivity test, that assessed wrongly the effect of curing on the “penetrability” (Sets 2–6). With regard to the Site Tests the picture is not as good but yet quite positive. All test methods were capable of differentiating, at highly significant (++) level, the “penetrability” of the OPC concretes with w/c ratios 0.40 and 0.60 (Sets 1–3). Yet, the only test method capable of differentiating the “penetrability” of OPC concretes with w/c ratios between 0.40 and 0.55 @seismicisolation @seismicisolation

116 Concrete Permeability and Durability Performance

(Sets 1–2) and between 0.55 and 0.60 (Sets 2–3) was the Torrent method and at a highly significant level (++). Autoclam Air and Torrent methods failed to differentiate the “penetrability” of the BFSC concretes of w/c ratios 0.40 and 0.55 (Sets 4–5), where the other methods succeeded. All test methods but Wenner electrical resistivity assessed correctly the positive effect of extended moisture curing on the “penetrability” (Sets 2–6). This is most likely due to the strong influence of the moisture on the resistivity; the effect of the higher moisture of the well cured specimens probably prevailed upon the beneficial effect of moist curing in reducing the “penetrability”. Finally, all site test methods performed satisfactorily under the Moist (Sets 7–8) and Cold (Sets 9–10) testing conditions. All Site Test methods, but Torrent and Wenner resistivity left some traces on the surface (holes of different sizes), as reported in RILEM TC 189-NEC (2007) and Romer (2005). 4.4.2.2 Correlation between Site and “Reference” Tests The degree of correlation between the results obtained on each set by each site test and its corresponding “Reference” test was investigated. The results of the three gas-permeability site tests (Autoclam, Torrent and HongParrott) were correlated with the Cembureau O2-Permeability “Reference Test”; the Autoclam Water Sorptivity test with the RILEM water sorptivity test and the Wenner electrical resistivity measured on the panels with those measured on the drilled cores. The correlation coefficients R obtained for each test are shown in the last column of Table 4.5. It can be seen that in the case of Torrent and Hong-Parrott site tests the correlation with the “Reference” test is excellent, it is acceptable for Autoclam Air and Wenner Resistivity and poor for Autoclam Water Sorptivity (see the footnote of Table 4.5). Correlations between the South African Indexes and other “Reference” and Site tests can be found in Beushausen and Alexander (2008). 4.4.2.3 Conclusions of the Comparative Test Transcribing from Romer (2005): “It can be concluded that the Comparative Test at EMPA was well designed, planned and executed to provide meaningful and objective results. The fact that the testers involved, both on site and at the laboratories, did not know the identity of the slabs or cores they were testing, guarantees the objectivity of the results obtained. Although to a varying degree, the Comparative Test proved that there are methods capable of evaluating the “penetrability” of the concrete cover on site, in a reliable and statistically significant manner. In five or six out of seven cases, the test methods were capable of detecting @seismicisolation @seismicisolation

Test methods for concrete permeability 117

correctly the expected differences in “penetrability” at a significant or highly significant level. Moreover, some of the site methods showed very good correlations with corresponding relevant Reference Test methods. This opens good perspectives for the application of such methods in practice, for the specification and in situ compliance control of the “penetrability” of the vital concrete cover, aiming at performanceoriented criteria regarding the durability of concrete structures.” ACKNOWLEDGEMENTS Some sections of this chapter took Chapters 3 and 4 from RILEM TC 189NEC (2007) as reference. The contribution of P.A.M. Basheer and A.F. Gonçalves to those chapters is duly acknowledged. REFERENCES AFPC-AFREM (1997). “Durabilité des Bétons, Méthodes recommandées pour la mesure des grandeurs associées à la durabilité”. Compte-rendu des journées techniques de l’AFPC-AFREM, 11 et 12 décembre, Toulouse, France, 284 p. Akmal, U., Hosoda, A., Hayashi, K. and Fujiwara, M. (2011). “Analysis of quality of covercrete subjected to different curing conditions using new surface water absorption test”. Proceedings of the 13th International Summer Symposium, JSCE, 287–291. Alexander, M.G. (2004). “Durability indexes and their use in concrete engineering”. International RILEM Symposium on Concrete Science and Engineering: A Tribute to Arnon Bentur, 9–22. Alexander, M.G., Ballim, Y. and Mackechnie, J.R. (1999). “Concrete durability index testing manual”. Research Monograph No.4, Univs. Cape Town & Witwatersrand, South Africa, 33 p. Alexander, M.G., Ballim, Y. and Stanish, K. (2008). “A framework for use of durability indexes in performance-based design and specifications for reinforced concrete structures”. Mater. & Struct., v41, n5, June, 921–936. Amphora (not dated). “Autoclam ()”. Brochure, Amphora Technologies Ltd., Belfast, 4 p. ASTM C 1585 (2013). “Standard test method for measurement of rate of absorption of water by hydraulic – cement concretes”. Baroghel-Bouny, V. (2006). “Durability indicators: relevant tools for performancebased evaluation and multi-level prediction of RC durability”. RILEM PRO 047, 3–30. Basheer, P.A.M. (1993). “Technical Note. A brief review of methods for measuring the permeation properties of concrete in situ”. Proc. Inst. Civ. Eng. Struct. Build., v99, n1, February, 74–83. Bérissi, R., Bonnet, G. and Grimaldi, G. (1986). “Mesure de la porosité ouverte des bétons hydrauliques”. Bull. Liaison des Lab. des Ponts et Chaussée, n142, 59–67.

@seismicisolation @seismicisolation

118 Concrete Permeability and Durability Performance Beushausen, H. and Alexander, M. (2008). “The South African durability index tests in an international comparison”. J. South African Institution Civil Eng., v50, n1, 25–31. Beushausen, H. and Alexander, M. (2009). “Application of durability indicators for quality control of concrete members – A practical example”. Concrete in Aggressive Aqueous Environments – Performance, Testing, and Modeling. Alexander, M.G. and Bertron, A. (Eds.). RILEM Publications, Bagneaux, 548–555. BS 1881-5 (1970). “Testing concrete. Methods of testing hardened concrete for other than strength”. British Standards Institution (Withdrawn). BS 1881-208 (1996). “Methods of testing hardened concrete for other than strength”. BS 1881, Part 208, British Standards Institution. Cather, R., Figg, J.W., Marsden, A.F. and O’Brien, T.P. (1984). “Improvements to the Figg method for determining the air-permeability of concrete”. Mag. Concr. Res., v36, n129, December, 241–245. Claisse, P.A., Elsayad, H.I. and Ganjian, E. (2009). “Water vapour and liquid permeability measurements in cementitious samples”. Adv. Cem. Res., v21, n2, April, 83–89. Concrete Society (1987). “Permeability testing of site concrete – A review of methods and experience”. Technical Report No. 31, London, 95 p. Coutinho, A. de S. and Gonçalves, A. (1994). Fabrico e Propriedades do Betão. Volume III. 2nd ed., LNEC, Lisboa, 368 p. DIN 1048 (1978). “Prüfverfahren für Beton – Bestimmung der Wassereindringtiefe”. Dinku, A. (1996). “Gas-permeability as a means to assess the performance properties of concrete”. IWB, Stuttgart, 236 p. Dinku, A. and Reinhardt, H.W. (1997). “Gas-permeability coefficient of cover concrete as a performance control”. Mater. & Struct., v30, n7, 387–393. Duarte, R., Flores-Colen, I. and Brito, J. (2011). “In situ testing techniques for evaluation of water penetration in rendered facades – The portable moisture meter and Karsten tube”. XII DBMC, Porto, Portugal, April 12–15, 8 p. EHE-08 (2008). “Instrucción de Hormigón Estructural”. Spanish Concrete Code. EN 12390-8 (2009). “Testing hardened concrete – Part 8: Depth of penetration of water under pressure”. fa*gerlund, G. (1977). “The critical degree of saturation method of assessing the freeze/thaw resistance of concrete”. Matériaux & Constr., v10, n4, 217–229. fib (2012). Model Code 2010, Final Draft, v1, fib Bulletin 65, March, 357 p. Figg, J.W. (1973). “Methods of measuring the air and water-permeability of concrete”. Mag. Concr. Res., v25, n85, December, 213–219. Gabrijel, I., Mikulic, D., Bjegovic, D. and Stipanovic-Oslakovic, I. (2008). “In-situ testing of the permeability of concrete”. SACoMaTiS, 337–346. GBJ 82‐85 (1986). “Standard for test methods of long‐term performance and durability of ordinary concrete”. Chinese Standard. Glanville, W.H. (1931). “The permeability of portland cement concrete”. Building Research Establishment, Technical paper, No.3, 62 p. Grube, H. and Lawrence, C.D. (1984). “Permeability of concrete to oxygen”. Proceedings on RILEM Seminar Durability Concrete Structures under Normal Out-door Exposure, Univ. Hanover, March, 68–79.

@seismicisolation @seismicisolation

Test methods for concrete permeability 119 Guth, D.L. and Zia, P. (2000). “Correlation of air-permeability with rapid chloride permeability and ponding tests”. PCI/FHWA/FIB International Symposium on High Performance Concrete, Orlando, September 25–27, 304–315. Guth, D.L. and Zia, P. (2001). “Evaluation of new air-permeability test device for concrete”. ACI Mater. J., v98, n1, 44–51. GWT (2014). “NDT systems. Bridging theory and practice”. Germann Instruments. Hansen, A.J., Ottosen, N.S. and Petersen, C.G. (1984). “Gas-permeability of concrete in situ: Theory and practice”. ACI SP 82, 543–556. Hayashi, K. and Hosoda, A. (2013). “Fundamental study on evaluation method of covercrete quality of concrete structures by surface water absorption test”. J. J.S.C.E., Ser E2 (Materials and Concrete Structures), v69, n1, 82–97. (in Japanese). Hendrickx, R. (2013). “Using the Karsten tube to estimate water transport parameters of porous building materials”. Mater. & Struct., v46, 1309–1320. Hong, C.Z. and Parrott, L.J. (1989). “Air-permeability of cover concrete and the effect of curing”. British Cement Assoc. Report C/5, October, 25 p. Hooton, R., Griesel, E. and Alexander, M. (2001). “Effect of controlled environmental conditions on durability index parameters of portland cement concretes”. Cem. Concr. & Aggr., v23, n1, 44–49. Imamoto, K., Shimozawa, K., Nagayama, M., Yamasaki, J. and Nimura, S. (2006). “Evaluation of air-permeability of cover concrete by single chamber method”. 31st Conference on ‘Our World in Concrete & Structures’ 2006, 16–17. James (2007). James Instruments Inc., “P-6050 & P-6000 porosiscope plus operating instructions”. Kasai, Y., Nagano, M. and Matsui, L. (1984). “On site rapid air-permeability test for concrete”. ACI SP 082, 525–542. Kawaai, K. and Ujike, I. (2016). “Influence of bleeding on durability of horizontal steel bars in RC column specimen”. Proceedings of IALCCE2016, Delft, The Netherlands, October 16–19, 839–846. Kessy, J.G., Alexander, M.G. and Beushausen, H. (2015). “Concrete durability standards: International trends and the South African context”. J. South African Inst. Civ. Eng., v57, n1, 47–58. Kollek, J.J. (1989). “The determination of the permeability of concrete to oxygen by the Cembureau method – A recommendation”. Mater. & Struct., v22, n3, May, 225–230. Kurashige, I. (2015). “Novel non-destructive test method to evaluate air-permeability distribution in depth direction in concrete – Development of triple-cell airpermeability tester (TCAPT)”. International Symposium on Non-Destructive Testing in Civil Engineering. (NDT-CE), Berlin, Germany, September 15–17. Levitt, M. (1969a). “Non-destructive testing of concrete by the initial surface absorption method”. Proceedings on Symposium Non-destructive Testing of Concrete and Timber, Inst. of Civil Engs., London, June 11–12, 23–28. Levitt, M. (1969b). “An assessment of the durability of concrete by ISAT”. Proceedings on RILEM Symposium Durability of Concrete, Prague. Levitt, M. (1971). “The ISAT: A non-destructive test for the durability of concrete”. Br. J. NDT, July, 106–112. LNEC E 392 (1993). “Betões. Determinação da permeabilidade ao oxigénio”. Laboratório Nacional de Engenharia Civil, May, 4 p.

@seismicisolation @seismicisolation

120 Concrete Permeability and Durability Performance Meletiou, C.A. (1991). “Development of a field permeability test for assessing the durability of concrete in marine structures”. PhD Thesis, Univ. of Florida, 184 p. https://ufdc.ufl.edu/AA00037920/00001/4. Meletiou, C.A., Tia, M. and Blooquist, D. (1992). “Development of a field permeability test apparatus and method for concrete”. ACI Mater. J., v89, n1, January–February, 83–89. Montgomery, F.R. and Adams, A. (1985). “Early experience with a new concrete permeability apparatus”. Proceedings on 2nd International Conference on Structural Faults and Repair, Forde and Topping (Eds.), 359–363. Montgomery, F.R. and Basheer, M. (1989). “Durability assessment of concrete bridges by in-situ testing, early results”. The Life of Structures. Physical Testing. Armer, G.S.T., Clarke, J.L., Garas, F.K. (Eds.). Butterworths, London, 352–359. Muigai, R., Moyo, P. and Alexander, M. (2012). “Durability design of reinforced concrete structures: A comparison of the use of durability indexes in the deemed-to-satisfy approach and the full-probabilistic approach”. Mater. & Struct., v45, n8, August, 1233–1244. Neves, R. and Gonçalves, A. (2006). “Concrete durability evaluation based on modified Figg test”. RILEM PRO 47, 249–256. Nganga, G., Alexander, M. and Beushausen, H. (2013). “Practical implementation of the durability index performance-based design approach”. Constr. & Build. Mater., v45, August, 251–261. Nguyen, N.H., Nakarai, K., Kuboria, Y. and Nishio, S. (2019). “Validation of simple non destructive method for evaluation of cover concrete quality”. Constr. & Build. Mater., v201, March, 430–438. Nishio, S. (2017). “Simple evaluation of water-permeability in cover concrete by water spray method”. Q. Rep. RTRI, v58, n1, 36–42. Nwaubani, S.O. (2018). “Non-destructive testing of concrete treated with penetrating surface sealant using a Karsten-tube”. ICCRRR, Cape Town, South Africa. Parrott, L. (1994). “Design for avoiding damage due to carbonation-induced corrosion”. ACI SP-145, 283–298. Parrott, L. and Hong, C.Z. (1991). “Some factors influencing air permeation measurements in cover concrete”. Mater. & Struct., v24, 403–408. Paulini, P. (2010). “A laboratory and on-site test method for air-permeability of concrete”. 2nd International Symposium on Service Life Design for Infrastructure, Delft, October 4–6, 995–1002. Paulini, P. and Nasution, F. (2007). “Air-permeability of near-surface concrete”. CONSEC’07, Tours, France, 8 p. Paulmann, K. and Rostasy F.S. (1989). “Praxisnahes Verfahren zur Beurteilung der Dichtigkeit oberflächennäher Betonschichten im Hinblick auf die Dauerhaftigkeit”. Institut für Baustoffe, Massivbau und Brandschutz, Techn. Univ. Braunschweig. Pereira Apps, C.A.C. (2011). “Evaluation of the variability of the Karsten tube in-situ test technique on measuring liquid water-permeability of renders and ceramic tile coatings”. IST, Univ. Técnica Lisboa, October.

@seismicisolation @seismicisolation

Test methods for concrete permeability 121 Reinhardt, H.W. and Mijnsbergen, J.P.G. (1989). “In-situ measurement of permeability of concrete cover by overpressure”. The Life of Structures, Physical Testing. Armer, Clarke and Garas (Eds.), Butterworth-Heinemann, 243–254. RILEM TC 25-PEM (1980). “Recommended tests to measure the deterioration of stone and to assess the effectiveness of treatment methods”. Matériaux et Constructions, v13, 175–253. https://doi.org/10.1007/BF02473564. RILEM TC 116-PCD (1995). “Performance criteria for concrete durability”. Kropp, J. and Hilsdorf, H.K. (Eds.), RILEM Report 12, E&FN Spon, 316 p. RILEM TC 116-PCD (1999a). “Concrete durability – An approach towards performance testing”. Final Report, Mater. & Struct., v32, April, 163–173. RILEM TC 116-PCD (1999b). “Preconditioning of concrete test specimens for the measurement of gas-permeability and capillary absorption of water”. Mater. & Struct., v32, April, 174–176. RILEM TC 116-PCD (1999c). “Measurement of the gas-permeability of concrete by the RILEM-Cembureau method”. Mater. & Struct., v32, April, 176–178. RILEM TC 116-PCD (1999d). “Determination of the capillary absorption of water of hardened concrete”. Mater. & Struct., v32, April, 178–179. RILEM TC 189-NEC (2007). “Non-destructive evaluation of the penetrability and thickness of the concrete cover”. Torrent and Fernández Luco (Eds.), RILEM Report 40, May, 223 p. RILEM TC 230-PSC (2016). “Performance-based specifications and control of concrete durability”. Beushausen and Fernández Luco (Eds.), RILEM Report 18, 373 p. Romer, M. (2005). “Comparative test – Part I – Comparative test of penetrability methods”. Mater. & Struct., v38, December, 895–906. SABS (2015). “Civil engineering test methods. Part CO3-2: Concrete durability index testing – Oxygen permeability test”. SABS/TC 081/SC 01, SANS-3001-CO3-2. Salvoldi, B.G., Beushausen, H. and Alexander, M.G. (2015). “Oxygen permeability of concrete and its relation to carbonation”. Constr. & Build. Mater., v85, 30–37. Schönlin, K.S. and Hilsdorf, H.K. (1987). “Evaluation of the effectiveness of curing of concrete structures”. ACI SP 100, 207–226. SIA 162/1 (1989). “Test No. 5- Water conductivity”. Swiss Society of Engineers and Architects. SIA 262/1 (2019). “Concrete construction – Complementary specifications”. Swiss Society of Engineers and Architects. Sogbossi, H., Verdier, J. and Multon, S. (2019). “New approach for the measurement of gas permeability and porosity accessible to gas in vacuum and under pressure”. Cem. & Concr. Composites, v103, 59–70. Soongswang, P., Tia, M., Blooquist, D., Meletiou, C.A. and Sessions, L. (1988). “Efficient test set-up for determining the water-permeability of concrete”. Transp. Res. Rec., n1204, 77–82. Stanish, K., Alexander, M.G. and Ballim, Y. (2006). “Assessing the repeatability and reproducibility values of South African durability index tests”. J. South African Inst. Civ. Eng., v48, n2, 10–17.

@seismicisolation @seismicisolation

122 Concrete Permeability and Durability Performance Starck, S., Beushausen, H., Alexander, M. and Torrent, R. (2017). “Complementarity of in situ and laboratory-based concrete permeability measurements”. Mater. & Struct., v50, n3, June, 177–191. Torrent, R.J. (1992). “A two-chamber vacuum cell for measuring the coefficient of permeability to air of the concrete cover on site”. Mater. & Struct., v25, n6, July, 358–365. Torrent, R. (2008). “Update of article RILEM TC 189-NEC Comparative test – Part I – Comparative test of penetrability methods, Mater. & Struct., v38, Dec 2005, pp. 895–906”. Mater. & Struct., v41, April, 443–447. Torrent, R. and Ebensperger, L. (1993). “Methoden zur Messung und Beurteilung der Kennwerte des Überdeckungsbetons auf der Baustelle”. Office Fédéral des Routes, Rapport No. 506, Bern, Switzerland, January, 119 p. Torrent, R. and Frenzer, G. (1995). “Methoden zur Messung und Beurteilung der Kennwerte des Ueberdeckungsbetons auf der Baustelle -Teil II”. Office Fédéral des Routes, Rapport No. 516, Bern, Switzerland, October, 106 p. Ujike, I., Okazaki, S. and Nakamura, T. (2009). “A study on improvement of in-situ air-permeability test for concrete structures”. Cem. Sci. & Concr. Technol., v63, 189–195. UNE 83966 (2008). “Acondicionamiento de probetas de hormigón para los ensayos de permeabilidad a gases y capilaridad”. Norma Española, 6 p. UNE 83981 (2008). “Determinación de la permeabilidad al oxígeno del hormigón endurecido”. Norma Española, 10 p. Valenta, O. (1970). “The permeability and the durability of concrete in aggressive conditions”. Comm. Intern. des Grandes Barrages (ICOLD), Montreal, Q. 39, R.6, 103–117. Van Eijk, R.J. (2009). “Evaluation of concrete quality with Permea-TORR, Wenner Probe and Water Penetration Test”. KEMA Report, Arnhem, July 8, 2009 (in Dutch), 48 p. Villagrán Zaccardi, Y.A., Alderete, N.M. and De Belie, N. (2017). “Improved model for capillary absorption in cementitious materials: Progress over the fourth root of time”. Cem. & Concr. Res., v100, October, 153–165. Washburn, E.W. (1921). “The dynamics of capillary flow”. Phys. Rev., v17, n3, March, 273–283. XP P 18–463 (2011). “Essai de perméabilité aux gaz sur béton durci”. Experimental French Standard, 15 p.

@seismicisolation @seismicisolation

Chapter 5

Torrent NDT method for coefficient of air-permeability

5.1 INTRODUCTION: WHY A SEPARATE CHAPTER? The reader may ask why is this method treated differently from the other methods described in Chapter 4?; the answer is: • the authors are among the top experts in the application of this test method, created by one of them • the authors share the conviction on the importance of measuring the air-permeability, not just in the laboratory like other methods do, but – more essentially – on site, for which this test method is especially suitable • is a test method that can be equally applied on cast specimens and in situ, allowing a direct comparison between those results and, thus, discerning responsibilities along the concrete construction chain (Neves et al., 2015) • the large and growing number of users of the test method worldwide, for whom this detailed chapter (and the whole book) should be extremely helpful 5.2 THE ORIGIN The creator of the test method, subject of this chapter, explains the origin and development of the idea as follows: In 1989, R. Torrent (RT) who in 1987 had joined the Materials Dept. of “Holderbank” Management & Consulting Ltd. (HMC, later Holcim Technology Ltd.) was asked to prepare an R&D project on the general topic of “Durability”. RT always acknowledged a relatively weak background in Chemistry but a strong one in Physics. During an intensive literature investigation, in order to elaborate the R&D proposal requested, RT came across with the concept of Covercrete, discussed in Section 7.1.4, particularly through Kreijger (1984), Newman (1987), Mayer DOI: 10.1201/9780429505652-5 @seismicisolation @seismicisolation

123

124 Concrete Permeability and Durability Performance

(1987). As a result, RT proposed a R&D project named Covercrete, aimed at studying the influence of the cement type, mix composition and curing on the quality (“penetrability”) of the Covercrete, measured through predominantly physical tests. The proposal was endorsed by HMC’s Research Council, resulting in funds and time available for RT to implement the test methods in HMC laboratory and to carry out the project (Torrent & Jornet, 1990). Within this context, RT joined RILEM TC 116-PCD (Permeability of Concrete as a Criterion of its Durability), attending a meeting in Göteborg, Sweden, that happened to be inspirational. In this meeting, different test methods to measure the “penetrability” of the Covercrete were presented and discussed, among them two on-site gas-permeability tests (described in Sections 4.3.2.3 and 4.3.2.8). At some point along the Bruneggerstrasse (Canton Aargau, Switzerland), while driving towards his office, RT got the idea of somehow combining the approaches of both test methods. On arrival, he immediately started to develop the concept which, after getting the needed components, assembling a prototype and a few trials, evolved into the test method called double-chamber vacuum cell, or Torrent test method to measure, non-destructively, the coefficient of air-permeability of concrete, described in this chapter. Some brief acknowledgements by R.Torrent: It is worth, at this point, to thank my Holderbank bosses at the time: T. Dratva (who asked me to submit the R&D project) and Drs. J. Gebauer and H. Braun (for useful discussions and for giving me the chance to freely develop my ideas). Also, to Drs. A. Jornet, L. Ebensperger and to G. Frenzer, former Holderbank colleagues for their still lasting friendship and for accompanying me during my first steps in the field of permeability testing. To ASTRA (Swiss Federal Highways Administration) for financial support and strict control of the R&D projects (Torrent & Ebensperger, 1993; Torrent & Frenzer, 1995) by a top team of advisors: M. Donzel, P. Wüst, F. Wittmann and late C. Menn. To the Swiss researchers that, through their competent and dedicated work, paved the road to the standardization of the method in Swiss Standard SIA 262/1: E. Brühwiler, E. Denarié, F. Jacobs, F. Hunkeler, A. Leemann, M. Romer and T. Teruzzi and, finally, to those who helped converting a crude prototype into a user-friendly, practical instruments: M. Fischli and K. Baumann (Proceq S.A.) and J. Szychowski, G. Zino and V. Bueno (Materials Advanced Services Ltd.). 5.3 FUNDAMENTALS OF THE TEST METHOD

5.3.1 Principles of the Test Method The so-called Torrent test method for measuring the coefficient of airpermeability of the Covercrete (see Chapter 7) is an improvement on the @seismicisolation @seismicisolation

Torrent NDT method 125

single-chamber vacuum cell test method already described in Section 4.3.2.8. One of the limitations of the single-chamber vacuum cell method is that the geometry of the air flow into the cell is undefined and cannot be controlled, see Figure 5.1. This prevents the calculation of a proper coefficient of air-permeability, which has to be replaced by some kind of air-permeability index. A more serious limitation is the fact that the air that flows into the vacuum chamber, raising its pressure, can take a preferential path along the usually more permeable concrete “skin”, artificially increasing the value of permeability measured; this was experimentally proved by Torrent (1991, 1992). To overcome this problem, the idea of creating a guard-ring around the measurement vacuum chamber, keeping the pressure of both chambers always balanced, was developed (Torrent, 1991, 1992). Hence the designation became double-chamber vacuum cell test method, or shorter Torrent method. The scheme of the air-flow in this test method is shown in Figure 5.2. Since the pressure Pe in the external guard-ring is kept permanently balanced with that in the inner test chamber (Pi), a unidirectional flow of a cylinder of air into the latter can be assumed. Now, under this controlled flow of air into the measurement chamber, the coefficient of air-permeability kT can be calculated, as derived below. So, the test method has two distinctive features: two circular rubber rings that seal the two chambers on the concrete surface, dividing the flow of air into the two concentric chambers and a pressure regulator that keeps the air pressure in both chambers permanently balanced (Pe = Pi). Figure 5.3 shows the aspect of the concentric rings and the book’s cover picture the vacuum cell stuck onto the concrete surface, pressed by the external atmospheric pressure.

Figure 5.1 Undefined air-flow geometry and preferential surface path in the single chamber vacuum cell test method.

@seismicisolation @seismicisolation

126 Concrete Permeability and Durability Performance

Figure 5.2 Scheme of air-flow in the Torrent method, with the assumed cylindrical flow of air into the central, test chamber.

Figure 5.3 Aspect and dimensions of the vacuum cell’s concentric rings.

5.3.2 Historical Evolution Five relevant steps can be identified in the development and evolution of the instrument (Torrent & Szychowski, 2016, 2017), as summarized in Table5.1. Only the second- and fifth-generation instruments are currently commercially available. The lay-out and components of these commercial instruments are shown in Figures 5.4 and 5.5. Figure 5.4 corresponds to the second-generation instrument branded Torrent Permeability Tester (TPT) (Proceq, 2019) and Figure 5.5 to the fifth-generation branded PermeaTORR AC+ (Active Cell) (M-A-S, 2019). Instruments up to the fourth generation relied on a mechanical regulator (membrane type) to control the evacuation of the external chamber by a constant regime vacuum pump. The fifthgeneration instrument relies on an electronic pressure regulator that controls the speed of a variable regime embedded vacuum pump. More details on the evolution and improvements of the instrument from the first- to the fourth-generation instruments can be found in Torrent and Szychowski (2016, 2017).

@seismicisolation @seismicisolation

Torrent NDT method 127 Table 5.1 Evolution of five generations of the Torrent method to measure the air-permeability coefficient Generation

Description/brand

Developer

Features

First (1990)

Research Prototype TPT

R. Torrent (Switzerland) Proceq (Switzerland)

Third (2008)

PermeaTORR

Fourth (2016) Fifth (2021)

PermeaTORR AC(Active Cell) PermeaTORR AC+ (Active Cell)

Materials Advanced Services Ltd. (Argentina)

Manual operation + calculations Manual operation + automatic calculations Automatic operation + calculations + graphic plot Ibid +Active Cell + embedded pump Ibid + Electronic pressure regulation + remote control

Second (1995)

Generations 1 (prototype), 2 (TPT) and 3 (PermeaTORR) work on the principle of a “passive”, hollow vacuum cell, the role of which was merely to divide the flow of air into that sucked by the inner chamber from that sucked by the outer chamber. The pressure measurement and regulation are conducted remotely at a control unit, connected to the cell chambers by means of a pair of rubber hoses (see Figure 5.4) that are part of the pneumatic system.

Figure 5.4 Sketch (Proceq, n.d.) and components of the second-generation instrument TPT.

@seismicisolation @seismicisolation

128 Concrete Permeability and Durability Performance

Figure 5.5 Sketch and components of the automatic fifth-generation instrument PermeaTORR AC+.

The fourth- and fifth-generation instrument PermeaTORR AC “Active Cell” (Figure 5.5) presents as a novelty the fact that the vacuum cell now houses several “active” components inside it: two pressure sensors (one for each chamber), the pressure regulator, valve 2 and a microprocessor that reads the pressure signals, operates valve 2 and communicates with the control unit. It also incorporates a small oil-free vacuum pump. The advantages of the new “Active Cell” design are described in Torrent and Szychowski (2016, 2017). The fifth-generation instrument PermeaTORR AC+ “Active Cell” presents a different pressure regulation (electronic), the option of using a smartphone as remote control and an upgraded software (allowing voice and visual inputs into the test record). The dimension of the inner chamber was established to have a circular testing area of Ø50 mm, which is the minimum required by Swiss Standard (SIA 262/1, 2019) for testing cores, drilled from laboratory cast specimens or on site, for capillary suction and chloride migration tests. The dimensions of the guard-ring (external Ø = 100 mm) were chosen to ensure a higher natural increase in pressure than the internal chamber and to fit the most common standard specimens (cube length or cylinder diameter = 150 mm). There was some concern that aggregate particles of 32 mm could interfere with the flow of air into the central chamber (Romer, 2005c). To elucidate this matter, some comparative tests were performed using the standard cell and one of bigger dimensions (not disclosed) on concretes of different compositions (maximum size of the aggregate = 32 mm). Both cells produced very similar kT results (Romer, 2005c), confirming the suitability of the cell with internal and external diameters of 50 and 100 mm, respectively. @seismicisolation @seismicisolation

Torrent NDT method 129

5.3.3 O peration of the Instrument The principles of the instruments’ operation are basically the same, irrespective of the different models:

5.3.4 M odel for the Calculation of the Coefficient of Air-Permeability kT The following assumptions are made in deriving the equation for the calculation of kT: • initially, all the pores in the concrete are at atmospheric pressure (hence, a repetition of a test at or near the place where one was previously made, can only be performed after ≈30 minutes waiting period) • the air flows into the inner chamber by viscous laminar flow (see Section 3.6) as a unidirectional cylindrical plug (Figure 5.2); this hypothesis is possibly valid only for penetrations of the vacuum front L not beyond the inner chamber diameter, i.e. ≈50 mm; to be verified by numerical modelling • the characteristics of the concrete (permeability and porosity) and its temperature and moisture condition are constant within the volume of area A of the inner chamber and penetration L affected by the test

@seismicisolation @seismicisolation

130 Concrete Permeability and Durability Performance

• although the conditions are non steady, the distribution of pressure is regarded as linear between the atmospheric pressure front and the surface of the concrete beneath the test chamber • the air pressure in the test chamber (normally 10–50 mbar) is much lower than the atmospheric pressure Pa (≈1,000 mbar) • the cell is placed on a semi-infinite body; even when not, the penetration of the test does not exceed the thickness of the element/specimen Figure 5.6 sketches the conditions in the concrete pores before and during the application of the Torrent test method. At the left, there is a sketch indicating the inner test chamber (i) and outer guard-ring (e); the inner test chamber has a cross sectional area A and a total volume Vc. At the right-hand side of Figure 5.6, the distribution of pressure of the air in the concrete pores is shown, at different instants of the test, from (a) to (d). Initially, at time t = 0, all the pores in the concrete contain air at atmospheric pressure Pa (≈1,000 mbar), situation (a) in Figure 5.6. As soon as vacuum is being made in the inner chamber, its pressure drops drastically, reaching at t 0 = 60 seconds a value P0 ≈ 0–50 mbar, situation (b) of Figure5.6; at this moment, valve 2 in Figures 5.4 and 5.5 is closed and the inner test chamber is isolated from the pump. The pressure in the inner test chamber P starts to grow steadily, due to the air flowing from the concrete

Figure 5.6 Evolution of pressure profiles with depth during the test.

@seismicisolation @seismicisolation

Torrent NDT method 131

pores into the inner chamber, driven by the gradient of pressure that has been established, situation (c) in Figure 5.6. Applying the Hagen-Poiseuille-Darcy law for gases, Eq. (3.26), to the situation (c) of Figure 5.6, we can calculate the volume of air dV (m³) that flows into the inner chamber in an interval dt (s) as follows: dV =

kT ⋅ A Pa2 − P2 ⋅ ⋅ dt 2⋅ µ ⋅y P

(5.1)

where kT = coefficient of air-permeability, measured by the Torrent method (m²) A = cross-sectional area of the inner test chamber (m²) μ = P = pressure in the inner chamber at time t (N/m²) Pa = atmospheric pressure (N/m²) Applying the general gas equation, we can calculate the number of air moles corresponding to a volume dV at a pressure P and absolute temperature T as follows: dn =

P ⋅ dV R⋅T

(5.2)

where R is the universal gas constant. Substituting dV in Eq. (5.2) by its value in Eq. (5.1) we get dn =

(

kT ⋅ A⋅ Pa2 − P2 2⋅ µ ⋅ y ⋅ R⋅T

) ⋅ dt

(5.3)

By mass conservation, the number of moles of air dn entering the vacuum chamber must correspond to the same number affected by a differential deepening dy of the atmospheric pressure front y, dotted lines in situation (c) in Figure 5.6. The affected volume of air is where ε = open porosity of the concrete (-) This volume can be assumed to be at the mean pressure (Pa + P)/2, which allows us to compute, from Eq. (5.2), the number of moles affected as follows: dn =

( Pa + P )

2⋅ R⋅T

⋅ dV =

( Pa + P )

2⋅ R⋅T

⋅ ε ⋅ A ⋅ dy

@seismicisolation @seismicisolation

(5.5)

132 Concrete Permeability and Durability Performance

Equating dn in Eqs. (5.5) and (5.3), we have:

( Pa + P )

2⋅ R⋅T

y ⋅ dy =

⋅ ε ⋅ A ⋅ dy =

(

kT ⋅ A⋅ Pa2 − P2 2⋅ µ ⋅ y ⋅R⋅T

) ⋅ dt

(5.6)

 kT P ⋅ Pa ⋅  1 −  ⋅ dt  Pa  µ ⋅ε

(5.7)

integrating and neglecting the term P/Pa , we can calculate the position y of the atmospheric front at time t y=

2⋅kT ⋅ Pa ⋅t ε ⋅ µ

(5.8)

now, looking at the situation in the cell, we know that the volume of air dV, entering the inner test chamber (of volume Vc) at pressure P, will produce a differential increase in its pressure dP given by dP =

P ⋅dV Vc

(5.9)

now, introducing Eqs. (5.1) and (5.8) into (5.9) and rearranging, we get

1 A kT ⋅ε 1 dP = ⋅ ⋅ dt 2 P −P 2⋅Vc 2⋅ µ ⋅Pa t

(5.10)

2 a

integrating both members between conditions (b) and (d) in Figure 5.6: Pf

∫

A kT ⋅ε 1 dP = ⋅ ⋅ 2 2 Pa − P 2⋅Vc 2⋅ µ ⋅Pa

∫ t0

1 dt t

(5.11)

leads to

( (

) )

 Pa + Pf ⋅( Pa − P0 )  A kT ⋅ε 1 ⋅ ln  ⋅ ⋅ = 2⋅Pa 2⋅ µ ⋅Pa  Pa − Pf ⋅( Pa + P0 )  Vc

(

)

t f − t0

(5.12)

from where the value of the coefficient of air-permeability can be obtained as   Pa + ∆P   2   ln  Pa − ∆P   µ V  c  ⋅ kT =   ⋅  A  2⋅ε ⋅ Pa  t f − t0     

@seismicisolation @seismicisolation

(5.13)

Torrent NDT method 133

Equation (5.13) is the one used nowadays to compute the coefficient of airpermeability kT (m²) by the Torrent method, where Vc = volume of the inner test chamber pneumatic system (m³) A = area of the inner test chamber (m²) μ = dynamic viscosity of air (N s/m²) ε = open porosity of the concrete (-) which, by default is taken as 0.15 Pa = atmospheric pressure (N/m²) ΔP = increase of effective pressure in the inner chamber (ΔP = Pf − P0) between time t 0 and tf (N/m²) t 0 = once valve 2 was closed, time from which the increase in pressure is measured (60 seconds) tf=time at which the test is finished (s) The test ends when the increase of effective pressure ΔPeff1 exceeds 20 mbar or, for concretes of low permeability, when t f = 720 seconds. However, in the automatic instruments, the test may also be optionally stopped at t f = 360 seconds provided an approximately linear relationship between ΔPeff and t − t0 is observed during the test (see justification in Section 5.3.5). It is worth mentioning that the equation to calculate kT in the original publications of the test method (Torrent, 1991, 1992; Torrent & Ebensperger, 1993) was just an approximation; the correct Eq. (5.13) was developed later (Torrent & Frenzer, 1995). Therefore, the values of kT published before 1995 need to be converted by multiplying them by a factor 1.846, as shown in p. 60 of Torrent and Frenzer (1995). The final penetration of the atmospheric pressure front L (m), i.e. the depth of concrete affected by the vacuum, can be calculated from Eq. (5.8) as

(

)

2⋅kT ⋅ Pa ⋅t f ε ⋅ µ

(5.14)

Notice that Eq. (5.14) is identical to Eq. (4.2), if the terms are rearranged. If the penetration of the test L exceeds the thickness of the element e, a correction is required, see Section 5.3.6.2.

5.3.5 Relation between ΔP and √t 5.3.5.1 T heoretical Linear Response Let us justify the stopping of the test when t f = 360 seconds if the plot ∆Peff vs. t − t0 is approximately linear. Equation (5.13) can be rewritten as

(

)

The effective pressure rise is the measured pressure rise during the test minus the pressure rise observed when the instrument is applied on an impermeable plate (calibration plate).

@seismicisolation @seismicisolation

134 Concrete Permeability and Durability Performance

with  A  2.kT .ε .Pa C =    Vc  µ

(5.16)

developing ln(1 + x) for |x| < 1 as a Taylor series:

ln (1 + x ) =

∞

∑

( −1)n+1 xn

(5.17)

n=1

Considering just the first four terms of the series: ln (1 + x ) = x −

x 2 x3 x 4 + − 2 3 4

ln (1 − x ) = − x −

(5.18)

x 2 x3 x 4 − − 2 3 4

ln (1 + x ) − ln (1 − x ) = 2x − 0 +

(5.19) 2.x3 − 0 3

(5.20)

Making x = ΔP/Pa and disregarding the third term in Eq. (5.20) (ΔP 0 the curvature is positive and, if (∂K/∂y) = 0 the plot is linear (Dobel et al., 2010; Dobel & Fernández Luco, 2012). The more usual departure from linearity happens when a gradient of moisture with depth exists in the concrete (e.g. after a rapid severe drying) leading to a negative curvature. A rare example of a positive curvature is presented in Torrent (2012) when testing a surface treated with a permeable formwork liner (see Section 7.2.3). The action of the liner is to reduce the w/c ratio of the near-surface layers leading to a positive permeability gradient (higher permeability at increasing depths). Another reason for the lack of linearity of the ∆Peff vs. t − t0 relation is caused by the evaporation of water from the concrete under test into the vacuum chamber, that produces an initial “jump” of ΔPeff which increases artificially the computed value of kT, especially for concretes of low permeability (Romer, 2005a). To avoid this effect, automatic instruments work always at pressures above 30 mbar (water vapour pressure at about 25°C). To this effect the different results obtained by the TPT instrument and the PermeaTORR instruments’ family are attributed (Torrent, 2012), as discussed in Section 5.6.4.2 and shown in Figure 5.13.

(

)

(

)

5.3.6 Relation between L and kT. Thickness Correction 5.3.6.1 R elation between Test Penetration L and kT There seems to be a direct relation between L and kT in Eq. (5.14); however, the relation is not unique, as t f depends on the mode in which the test is stopped. Mode 1 happens for moderate to high-permeability materials, for which tf corresponds to an increase in (effective) pressure in the inner chamber of ΔPf=20 mbar. For concretes of lower permeability, Mode 2, t f happens when it reaches 720 seconds (optionally 360 seconds for the thirdand fourth-generation instruments). Using Eq. (5.22), Szychowski (2010) found that the function L(kT) is composed of two stages, depending on the mode of ending the test: For Mode 2 and tf=720 or 360 seconds, Eq. (5.14) holds.

@seismicisolation @seismicisolation

136 Concrete Permeability and Durability Performance

For Mode 1, Eq. (5.23) is applicable, with ΔPf=20 mbar; so, in this case, L is instrument-dependent as it is function of the ratio between the volume and cross-section area of the inner chamber. L=

2⋅ ∆Pf 2⋅kT ⋅ Pa ⋅60 V + c ⋅ ε ⋅ µ Pa A⋅ε

(5.23)

In the case of the PermeaTORR family, Vc = 16 × 10 −5 m³ and A = 196 × 10 −5 m²; for the TPT is Vc = 22 × 10 −5 m³. Figure 5.7 shows the relation between L and kT for both instruments, assuming Pa = 1,000 mbar (100 kPa), for the different stopping modes. The turning points are shown in the graph. It can be seen that the computed L ranges from a few mm for kT below 0.01 × 10 −16 m² to over 100 mm for kT above ≈ 10 × 10 −16 m. The validity of Eq. (5.14) was experimentally investigated in the laboratory by Kato (2013), using six concrete mixes of w/c in the range 0.30–0.70, resulting in compressive strengths between 23 and 90 MPa. He conducted kT measurements on intact 150 mm cubes, as well as in others in which he drilled Ø6 mm holes of variable lengths between 100 and 130 mm, i.e. arriving to depths between 20 and 50 mm below the surface where kT was to be measured. By analysing the changes between the kT measured on the intact cubes and on those perforated, he could guess when the vacuum front reached the holes (resulting in significantly higher kT). The experimental research showed that the measurable L is 15 mm for kT < 0.05 × 10 −16 m²; 30 mm when 0.05 × 10 −16 m² < kT < 2.0 × 10 −16 m² and 40 mm when kT > 2.0 × 10 −16 m², in very good agreement with the calculation through Eq. (5.14), displayed in Figure 5.7.

Figure 5.7 Relation between L and kT for different stoppage modes and instruments (TPT and PermeaTORR), assuming Pa = 100 kPa.

@seismicisolation @seismicisolation

Torrent NDT method 137

5.3.6.2 Correction of kT for Thickness The derivation of Eq. (5.13) assumes that the test is performed on a semiinfinite body. Invariably, the test is performed in practice on specimens or elements that have finite dimensions. As Figure 5.7 shows, the penetration of the test can be considerable; for instance, for kT values above 40 × 10 −16 m², the penetration of the test given by Eq. (5.23) exceeds 150 mm. In case that the thickness of the specimen e is lower than L, Eq. (5.13) is no longer valid and a correction is required, as described below. In the case that L > e, two successive time intervals have to be considered, separated by the time te required for the penetration front y to reach the thickness of the element e: t 0 –te, where the flow takes place across a growing depth y, which corresponds to the general case te –tf, where the flow takes place across a constant depth y = e Therefore, Eq. (5.10) has to be modified to take into consideration the gas flow happening in both intervals, becoming, where kT is the true coefficient of permeability of the material:

A kT ⋅ε 1 kT ⋅ A 1 dP = ⋅ ⋅ dt + dt 2 P −P 2⋅Vc 2⋅ µ ⋅Pa 2⋅Vc ⋅ µ ⋅e t

(5.24)

2 a

Now, if we integrate the first member of Eq. (5.24) between t 0 –tf, we will get:

 ( Pa + Pf ) ⋅( Pa − P0 )  A kTi ⋅ε 1 1 dP = ⋅ ⋅ ln  = ⋅ 2 P −P V 2 ⋅ Pa 2⋅ µ ⋅Pa − ⋅ + P P P P ( ) c ( ) 0  a a f    t0 tf

∫

2 a

(

t f − t0 (5.25)

where kTi is the value indicated by the instrument, assuming L = ∞, Eq. (5.12). The first term of the second member of Eq. (5.24) must be integrated for the time interval t 0 –te, whilst the second term for the interval te –tf which, considering the first member (Eq. 5.25) results in A kTi ⋅ε ⋅ ⋅ Vc 2⋅ µ ⋅Pa +

(

)

t f − t0 =

kT ⋅ A ⋅ t f − te 2⋅Vc ⋅ µ ⋅e

(

A kT ⋅ε ⋅ ⋅ 2⋅ µ ⋅Pa Vc

) @seismicisolation @seismicisolation

(

t e − t0

) (5.26)

)

138 Concrete Permeability and Durability Performance

Now, applying Eq. (5.14) to compute the time te necessary for the vacuum front to reach the thickness e, we have te =

e2 ⋅ε ⋅ µ 2⋅kT ⋅Pa

(5.27)

Substituting te in Eq. (5.26) and reorganizing, we get where

a =

Pa ⋅t f 2⋅ Pa ⋅ε ⋅t0 ; ; b = − µ ⋅e µ

c =

e ⋅ε  2⋅ Pa ⋅ε ⋅kTi ⋅ − 2 µ 

(

 t f − t0  

)

(5.29)

With all variables in Eq. (5.25) expressed in the SI system of units [m, kg, s] and solving the second degree Eq. (5.28)  − b + b2 − 4 ⋅ a ⋅ c  kT =   2⋅a  

valid for L ≥ e

(5.30)

The corrected value kT is always smaller than the value indicated by the instrument kTi, because in the latter a smaller gradient of pressure is assumed (y > e) than the real one (y = e), for the same gas flow. Instruments of third and later generations allow an automatic thickness correction after entering the value of e. 5.4 RELEVANT FEATURES OF THE TEST METHOD The most relevant features of the Torrent test method are • it is entirely non-destructive (no traces left on the concrete surface after application) and suitable for both laboratory and site applications • it is fast, taking from 1.5 to 6 or 12 minutes for high- and low-permeability concrete, respectively • the known and controlled gas flow into the inner chamber allows the calculation, through Eq. (5.13), of an important physical property as the coefficient of permeability of concrete to air kT is, in SI units (m²) and not just a technological index, as for the single chamber test (Section 4.3.2.8) @seismicisolation @seismicisolation

Torrent NDT method 139

• the penetration of the vacuum front L can be calculated from Eq. (5.14), which allows knowing the concrete depth affected by the test and applying a correction in case it exceeds the thickness of the element • it is not affected by the “skin” or “wall” effects since, thanks to the outer guard ring, the spurious air entering the vacuum cell through the permeable “skin” (Figure 5.1) is now sucked by the outer chamber, not affecting the unidirectional flow into the inner, test chamber. This was demonstrated in Torrent (1991, 1992) 5.5 INTERPRETATION OF TEST RESULTS

5.5.1 Permeability Classes It has been shown that, in theory, the coefficient of permeability is proportional to the porosity and to the pore size squared, Eq. (3.20). As the interconnected pores (capillary pores and ITZ) cover a size range of four orders of magnitude (Figure 3.8), it is expected that the air-permeability would cover around twice that range. Table 5.2 (first three columns) presents a classification of concrete, based on the air-permeability kT, that confirms that expectation. A similar classification, based on kT, was already proposed in p. 31 of Torrent and Frenzer (1995), extended now to the one presented in Table5.2. The lower limit of measurable permeability is kT = 0.001 × 10 −16 m²; below that value the flow of air is so small that the induced pressure rise cannot be accurately measured by the pressure sensors. As shown in Table 5.2, the classification follows a logarithmic scale of kT, a fact to be taken into consideration when interpreting the test results. As discussed in more detail in Chapter 8, the durability performance of a concrete is related to the logarithm of kT rather than to kT directly. Table 5.2 Estimated porosity ε and characteristic radius r of the concrete pore system as function of kT Expected porosity ε(%) kT (10−16 m²) 100

Permeability Class PK0 PK1 PK2 PK3 PK4 PK5 PK6

Negligible Very low Low Moderate High Very high Ultra high

Eq. (5.31) 23

@seismicisolation @seismicisolation

Expected pore radius r (nm) Eq. (5.32) 600

Eq. (5.33) 450

140 Concrete Permeability and Durability Performance

5.5.2 Microstructural Interpretation The results of kT can be interpreted also from a microstructural perspective. Several investigations produced parallel results of mercury intrusion porosimetry or MIP (porosity εMIP and characteristic pore radius r) on samples taken from specimens where kT had previously been measured (Torrent & Ebensperger, 1993; Torrent & Frenzer, 1995; Sakai et al., 2013). In the last two investigations, data from site concrete are included. In Torrent and Ebensperger (1993), the relation between εMIP and the capillary εc and total porosity εt was investigated for a wide range of concretes. Porosities εc and εt were measured after the old Swiss Standard (SIA 162/1,1989), very similar to current Swiss Standard (SIA 262/1-K, 2019). The volume of pores was obtained gravimetrically by measuring the mass after saturation under water (εc) and after water saturation under vacuum (εt), referred to the mass after drying at 105°C. The total porosity is meant to include not just the capillary pores but also the air voids. The results of Torrent and Ebensperger (1993) indicate that εMIP measures a porosity somewhere in between the capillary and the total porosity, as determined gravimetrically by water saturation referred to a sample dried at 105°C. Figure 5.8 (l.) shows the relation obtained between the measured airpermeability kT and the porosity measured by MIP εMIP in the three investigations. The regression of Eq. (5.31) is fitted to all 62 test results (black line in Figure 5.8, left), with R = 0.80. According to the correlation coefficient test (Urdan, 2011), these results present a highly significant correlation between εMIP and kT, as the test returned a near zero p-value. Figure 5.8 (r.), in turn, shows the relation between the measured airpermeability kT and K MIP, the latter computed from Eq. (3.20) as follows: KMIP =

ε MIP . r 2 8

(5.32)

where the radius r is equal to the mean pore radius for the data from Torrent and Ebensperger (1993) and Torrent and Frenzer (1995) and to the threshold value for the data from Sakai et al. (2013). It is worth mentioning the different degrees of saturation of the samples in Figure 5.8, as kT was measured on specimens kept 21–28 days in a dry room at 20°C/50% RH for the first two investigations, compared to 6 months at 20°C and ≈ 60% RH for Sakai et al. (2013). The MIP analysis was performed on thoroughly dried small samples. The black line in Figure 5.8 (r.) is the equality line. It can be seen that, despite the large scatter in results, especially from site tests, the calculated @seismicisolation @seismicisolation

Torrent NDT method 141

Figure 5.8 Relation between ε MIP and kT (l.) and between kT and K MIP (r.).

K MIP matches reasonably well the experimental kT. Given the different preconditioning of the kT tests and the differences in MIP techniques and instruments used to establish the values of εMIP and r, the agreement in Figure 5.8, (l. and r.), is quite remarkable. From Eqs. (5.31) and (5.32), it is possible to obtain an estimate of the porosity ε and characteristic pore radius r of the concrete pore system from the measured kT, see Table 5.2. An estimate of the so called “representative pore size of permeation resistance” (RPSPR), as function of kT, was proposed by Sakai et al. (2014) as with RPSPR in (nm) and kT in (10 −16 m²). The calculated values of RPSPR are included in the last column of Table 5.2, resulting slightly lower than those obtained from Eqs. (5.31) and (5.32). Table 5.2, although providing just gross estimates of ε and r, adds a physical meaning to the measured kT values in terms of the pore structure of the concrete tested. For a more detailed analysis, from a different and more general perspective, the reader can refer to Sakai (2019). 5.6 R EPEATABILITY AND REPRODUCIBILITY When the repeatability and reproducibility of the Torrent test method are assessed, the fact that the kT values range over six orders of magnitude (Table 5.2) has to be borne in mind. Indeed, kT values within a ratio of 10:1 actually belong to the same Permeability Class. This means that kT is a variable very sensitive to the microstructure (especially to the pore structure) of the material under test. For instance, in the case of concrete, if a @seismicisolation @seismicisolation

142 Concrete Permeability and Durability Performance

large aggregate particle (say, 38 mm size) happens to lay below the concrete surface, just where the central chamber (Ø 50 mm) is placed, a different kT value will be obtained than in a place displaced a few mm from that point. Therefore, the scatter of kT values requires a special statistical treatment, as discussed in Section 5.8.

5.6.1 Testing Variability: Repeatability This section deals with the repeatability of the test method and the instruments used to measure kT. It refers to the testing variability, i.e. the variability of test results obtained with the same instrument applied repeatedly on the same sample (if possible, exactly on the same spot). There are few data on the repeatability of kT measurements. Some data were produced by González Gasca, at the Instituto Eduardo Torroja, Madrid, Spain with a TPT (Torrent, 1997). She repeated five measurements at approximately the same place, on the same face of a concrete disc (Ø 150 × 50 mm), with a waiting time of 30 minutes between successive measurements. She reported a mean value of 0.057 × 10 −16 m², with a coefficient of variation CoV of 19%. Ten successive kT test results were obtained by R. Torrent, with a PermeaTORR AC instrument, on the same location of a reference sample used to calibrate the instruments. A waiting time of 15 minutes was observed between the end of one measurement and the initiation of the next one. The mean value of the ten measurements was 0.106 × 10 −16 m², with a standard deviation of 0.004 × 10 −16 m², yielding a CoV of 3.8%. The repeatability and stability of readings can be judged by repeating tests at intervals in the same spots on specimens that are stable in terms of hydration and of moisture and temperature. One investigation of this kind was reported by Adey et al. (1998), who repeated air-permeability tests with a TPT instrument (Proceq S.A.) on seven well-defined spots (#1–#7) of slabs removed from a bridge and stored in EPFL Lausanne laboratory for 6 months before the measurements. Two of these spots (#4 and #7) were located in coincidence with steel bars with 20 mm cover thickness (assessed by a covermeter), whilst the rest were located away of reinforcing bars. Measurements of kT were repeated five times over a period of 5 weeks. Except for the second reading on spot #3, the stability and repeatability of the measurements are very good. It is worth mentioning that most of the readings fall within the “High” Permeability Class (Table 5.2), including those obtained in coincidence with the steel bars, with spots #5 and #2 yielding values corresponding to the “Very High” and “Moderate” Permeability Classes. The testing variability of the Torrent test method is recognized as very low; hence, the recommendation of performing just one reading per measuring spot; it is better to spend time measuring several points within the elements or specimen (i.e. different lateral faces of a cube) rather than repeating readings on the same spot. @seismicisolation @seismicisolation

Torrent NDT method 143

5.6.2 Within-Sample Variability Here, the variability of kT within, nominally, the same sample is dealt with. This variability reflects the effect the heterogeneity of concrete has on the measured coefficient of air-permeability. For this, we will refer to an experiment conducted at the Swiss Federal Polytechnic University, Zürich (ETHZ), within the frame of a project financed by the Swiss Federal Highway Administration (Torrent & Frenzer, 1995). Concrete cubes (0.5 m) were cast at ETHZ with four different concrete mixes made with the same constituents, but with a wide range of characteristics (w/c = 0.3– 0.75; OPC = 200–450 kg/m³; f′c = 14–66 MPa). Two cubes were cast with each mix, one of which was moist cured during 7 days (B), whilst the other was totally deprived of moist curing (A). At the age of 28 days, the cubes were tested by now Holcim Technology personnel, without knowing the preparation conditions of any of the eight cubes (blind test). The tests were conducted, using a TPT, on two opposite faces of each cube, five tests on each face following a pattern similar to number five of a dice. The results obtained are presented in Table 5.3 (the results in the two bottom rows are not relevant here, but will be discussed in Section 6.2.1.2). Table 5.3 Within sample variability of kT measured on 0.5 m cubes of different concretes Cube Curing (d) w/c OPC (kg/m³) Air (%) 28-day f′ccyl (MPa) kT (10−16 m²) Face 1

kT (10−16 m²) Face 2

Mean (10−16 m²) Std. dev. (10−16 m²) CoV (%) kTgm (10−16 m²) sLOG kO (10−16 m²) a24 (g/m²/s½)

IA 0

7 0.75 200 8.0 13.8 37.2 15.3 19.2 12.8 18.7 6.12 33.3 15.3 20.1 10.4 35.7 10.3 37.1 10.6 43.6 8.49 71.3 20.5 60.6 7.36 37.7 11.7 18.5 4.3 48.9 37.1 34.0 11.0 0.20 0.16 22.6 16.5 18.9 18.8

IIA 0

IIB

7 0.50 300 7.5 25.4 0.771 0.992 5.05 0.443 1.08 0.121 0.310 0.103 0.812 0.116 1.09 0.765 1.19 0.570 1.05 0.347 0.483 0.146 0.261 0.109 1.21 0.371 1.39 0.317 114.9 85.4 0.84 0.262 0.36 0.39 2.00 1.74 12.1 11.2

IIIA

7 0.40 335 7.5 38.3 0.208 0.093 0.215 0.091 0.150 0.036 0.159 0.077 0.066 0.06 0.094 0.062 0.242 0.073 0.603 0.071 1.186 0.024 0.187 0.053 0.311 0.064 0.341 0.022 109.7 34.5 0.215 0.060 0.36 0.18 0.463 0.346 9.39 8.15

@seismicisolation @seismicisolation

IIIB

IVA 0

IVB

7 0.30 450 2.3 65.7 0.028 0.018 0.041 0.008 0.043 0.018 0.032 0.007 0.017 0.009 0.075 0.014 0.059 0.027 0.023 0.018 0.023 0.007 0.020 0.010 0.036 0.014 0.019 0.007 51.9 48.4 0.032 0.012 0.21 0.21 0.173 0.194 6.73 6.53

144 Concrete Permeability and Durability Performance

The cubes were cast by the same personnel, using same forms, tools and techniques, and the tests were performed by the same person with the same instrument, following the same routine. Hence, in principle, one would expect the within sample variability to be very similar for the different specimens. Looking at the standard deviation of the results (sixth row from bottom of Table 5.3), it is clear that it changes enormously with the permeability level of the sample. The CoV (fifth row from the bottom) shows more stability. The fourth and third rows (from the bottom) show the parameters obtained assuming that the results of kT, especially on site, follow reasonably well a log-normal distribution (see Section 5.8). The kTgm value is the geometric mean of the test results which, as expected, is somewhat smaller than the arithmetic mean (seventh row from the bottom). Remember that the geometric mean corresponds to the antilogarithm of the average of the logarithms of the test results. The s LOG value reported corresponds to the standard deviation of the log10 of the kT results, which is quite stable for the different cubes, demonstrating to be a fair indicator of the variability of the results. In the future, the results of kT measurements will be described in terms of their kTgm and s LOG statistical parameters, as recommended by Jacobs et al. (2009). Now, if we compare the CoV reported in Table 5.3 with the 3.8% or even 19% (Section 5.6.1) obtained when applied on the same specimen, we see the much larger Within-Sample Variability compared with the Test Variability. This reflects the heterogeneity of concrete within the sample, even when the specimen (0.5 m cubes in this case) has been prepared in a laboratory under strict casting, compaction and curing procedures.

5.6.3 Global Variability The Global Variability encompasses the Testing and Within-Sample Variability, discussed in Sections 5.6.1 and 5.6.2, plus the Between-Samples Variability. The latter is the result of variations in quality of the Covercrete resulting from batch-to-batch concrete quality variations, changes in the conditions of placement, compaction, curing, etc., differences in the skills of the personnel involved in the construction process, variable weather conditions during construction and testing, different exposures of the surfaces to sun, wind, rain/snow, etc. Data on the Global Variability, expressed as s LOG , were reported by Jacobs et al. (2009), based on site investigations of 52 concrete structures in Switzerland (Jacobs, 2006), reproduced in Figure 5.9. This gives an indication of the large differences in s LOG that can be encountered when testing kT on site concrete, reflecting the large number

@seismicisolation @seismicisolation

Torrent NDT method 145

Figure 5.9 Distribution of in situ global variability s LOG obtained on 52 structures in Switzerland.

of factors involved in the Global Variability (compared them with those reported in Table 5.3 for the laboratory cubes).

5.6.4 Reproducibility On several occasions, the same samples were tested, in the lab and on site, using different instruments, of the same and/or different brands. 5.6.4.1 Reproducibility for Same Brand Within the frame of a project, sponsored by the Swiss Federal Highways Administration (ASTRA) (Jacobs et al., 2009), several Swiss institutions participated in an interlaboratory test, conducted in two different construction sites, a bridge and a tunnel. Two segments of the bridge were tested (D-E and XI), with the total test surfaces divided into test boxes, each box being assigned to each one of the five Swiss laboratories taking part in the experiment (see Figure 5.10): EMPA, TFB (organizer of the inter-laboratory test), EPFL, SUPSI and Holcim. The test results are summarized in Figure 5.11, where the kTgm and s LOG of the 15 values obtained by each laboratory on each of the two bridge segments are plotted; the central dot represents the kTgm value and the length of the segment represents ±s LOG (see Section 5.8.3). It has to be borne in mind that each laboratory tested different samples (test areas) of a presumably same population (the bridge segment), so a certain difference between the results is to be expected. Figure 5.11 shows quite clearly that all laboratories rated the permeability of the bridge segments as “Moderate”, with a lower variability for Segment XI than for Segment DE.

@seismicisolation @seismicisolation

146 Concrete Permeability and Durability Performance

Figure 5.10 Dr. F. Jacobs and five instruments at comparative field test.

Figure 5.11 Values of kTgm ± s LOG obtained by five Swiss laboratories on two segments of same bridge.

@seismicisolation @seismicisolation

Torrent NDT method 147

Neves performed Analysis of Variance (ANOVA) tests for both kT and log kT and both failed to reject the hypothesis that one of the sets (for each bridge segment) is different from the other four. Furthermore, assuming no particular statistical distribution of kT results, the Kruskal-Wallis test yielded the same result. Another reproducibility trial ground was the Application Test, organized by RILEM TC 230-PSC (2015). In it, eight concrete panels, made with different binders and w/c ratios and cured outdoors under different protection conditions, were blind tested by different laboratories, applying various test methods. Among the latter were Univ. of Zagreb (UZ) from Croatia, TFB from Switzerland and Materials Advanced Services (MAS) from Argentina, each of them using different units of the PermeaTORR instrument. The panels were measured in two occasions: • first round, 15–16 April 2012, with the panels exposed outdoors under unsuitable test conditions (Age: 14–21 days, T = 1°C–6°C, surface moisture m% = 4.9%–5.7%), not complying with several prescriptions of Swiss Standard (SIA 262/1-E, 2019) (see Section 5.7). UZ and MAS participated • second round, 9 July 2012, with the panels stored indoors (Age: 101– 108 days; T = 17°C–21°C, surface moisture m% = 4.4%–5.5%). The test conditions complied with Swiss Standard (SIA 262/1-E, 2019). TFB and MAS participated Figure 5.12 presents the results of UZ and TFB as function of those of MAS that participated in both rounds. The results of UZ were obtained on the same spots tested by MAS. In the case of TFB, some results were obtained

Figure 5.12 Reproducibility of PermeaTORR instruments on test panels (RILEM TC 230PSC, 2015).

@seismicisolation @seismicisolation

148 Concrete Permeability and Durability Performance

on the same spots (black triangles); the white triangles represent the geometric means of six tests performed by TFB and MAS on the same panels, but not on the same spots. The reproducibility obtained in the second round (triangles) is remarkable; not so good, but acceptable, is that obtained in the first round (circles), under unfavourable conditions. It has to be mentioned that, due to the low ambient temperature, the instrument of UZ showed calibration problems; it was not possible to reach the requirements of SIA 262/1-E (2019), as reported in RILEM TC 230-PSC (2015). This explains the differences observed for concretes with very low permeability (kT < 0.01 × 10 −16 m²), more affected by inaccurate calibration values, as a small difference in the calibration pressure has a strong impact on the small pressure rise recorded for low-permeability concretes.

5.6.4.2 Reproducibility for Different Brands The PermeaTORR family of instruments introduced small changes in the test method that ended up in establishing slight but systematic differences in the measured kT value (Torrent, 2012). In particular, limiting the initial vacuum pressure to not less than 30 mbar (i.e. above water vapour pressure), avoided non-linear responses of the instrument in the plot ΔPeff with t − t0 , that were attributed to water vaporization effects in the test chamber (Romer, 2005a), see Section 5.3.5.2. Figure 5.13 presents results of measurements conducted with the PermeaTORR and the TPT, applied on the same samples by different researchers. Most data (Kurashige, Neves, Torrent) were obtained in the laboratory; other (Torrent + Jacobs Site) were obtained on the same spots of a Tunnel near Aarau, in Kanton Aargau, Switzerland (Jacobs et al., 2009); Jacobs with a TPT, Torrent with a PermeaTORR. The results obtained by the PermeaTORR shall be considered as reference, as they are not affected by the pressure rise contribution of vaporized water in the vacuum chamber (Torrent, 2012). A difference between both instruments is appreciable only for low and very low permeability (Classes PK1 and PK2), attributable to the water vaporization effect discussed in Torrent (2012). Yet, a very good correlation exists between the results, through the conversion regression presented in Figure 5.13. In 2016, a significant new concept was incorporated into the PermeaTORR AC (Active Cell) fourth-generation instrument (Torrent & Szychowski, 2016, 2017). Before and after launching the fourth-generation instrument, comparative tests with the third-generation PermeaTORR instrument were conducted in the laboratory (Torrent & Szychowski, 2017) and on site (Torrent et al., 2018). Two PermeaTORR AC (Serial Numbers #2 and #3) instruments and one PermeaTORR instrument (Serial Number #89) were used for the comparative test. Just three site tests were performed with both @seismicisolation @seismicisolation

Torrent NDT method 149

Figure 5.13 Correlation between kT results obtained with the PermeaTORR and TPT instruments.

instruments at a construction site in Como, Italy (Torrent et al., 2018). The results indicate that the Serial Numbers #2 and #3 of the fourth-generation PermeaTORR AC yielded slightly lower kT values than those of the thirdgeneration PermeaTORR used for comparison. Both PermeaTORR AC units, tested at SUPSI Laboratory, yielded very similar results. The conclusion is that both models: third-generation PermeaTORR and fourth-generation PermeaTORR AC can be used indistinctly, as both yield very similar results. In the case of the second-generation TPT, it yields higher results for concretes of low permeability (kT below ≈0.1 × 10 −16 m²); in that case, a correction may be applicable, using the regression shown in Figure 5.13, where the permeabilities are expressed in 10 −16 m². 5.7 EFFECTS AND INFLUENCES ON kT For effects it is understood intrinsic properties or characteristics of concrete that have a direct impact on its permeability (e.g. w/c ratio, binder type, curing, cracks, etc.), which cannot be modified by the user of the test method. They are discussed in detail in Chapter 6, for air-permeability kT as well as for other permeability test methods. For influences it is understood extrinsic factors, in general not related to the concrete quality, that have an impact on the air-permeability kT test results. Examples are the roughness, temperature and moisture of the tested surface. Contrary to the effects, the influences can be controlled to some extent by the user of the test method. The rest of this section is concerned with the impact of the influences on kT. @seismicisolation @seismicisolation

150 Concrete Permeability and Durability Performance

One special case is the age of concrete at the moment of test, which can be controlled, within certain limits, by the user of the test method. In this case, there is a dual impact; on the one hand, it is known that continued hydration will improve the tightness of concrete (Sections 3.1 and 3.2), so a concrete tested at a later age is expected to have less permeability than when tested earlier. On the other hand, particularly for site testing as is often the case for kT, the moment of test may determine the temperature and, especially, the moisture content of the concrete. These variables, especially moisture, have a strong effect on the measured value of kT, very specific to this particular test method. That is the reason why age is included within this section.

5.7.1 I nfluence of Temperature of Concrete Surface 5.7.1.1 Influence of Low Concrete Temperature In principle, the temperature of the concrete surface should have a minor impact on the air-permeability kT, at least when above 0°C, because the viscosity of air in Eq. (5.13) does not change significantly with temperature (see Table 3.2). There is some influence on the instrument’s response (possibly thermal dilation/contraction of components), reflected in systematic lower calibration values obtained at low temperatures (hence the need to repeat calibrations on site under changing temperature conditions – see Annex B). There was some debate within the Committee in charge of drafting ASTRA Recommendations for in situ testing of air-permeability (Jacobs et al., 2009) on the minimum acceptable temperature for running a test. Finally, a conservative approach prevailed, establishing a lower limit of 10°C, value that was included in the 2013 version of Swiss Standard (SIA 262/1-E, 2019). That limit was lowered to 5°C in the 2019 revision of the standard. Some experiments were conducted by R. Torrent to check the possibility of lowering the minimum limit below 10°C (Torrent & Szychowski, 2016). In one of them, three concrete samples of different Permeability Classes (PK2, PK3 and PK4, see Table 5.2) were tested at the offices of Materials Advanced Services in Buenos Aires, Argentina, first indoors (T ≈ 22°C) and afterwards outdoors in winter time (T ≈ 5.5°C), using a third-generation instrument PermeaTORR (PT) and the first unit produced of the fourthgeneration instrument PermeaTORR AC (PTAC#0). It was shown that both instruments measured very similar values at both temperatures on the three samples. A similar experiment was conducted by R. Torrent in Switzerland with a PermeaTORR AC instrument (PTAC #2), on 17 (≈4-years old) concrete samples (150 mm cubes) covering a wider range of permeabilities. The samples were first measured (24 and 25 January 2017) in a laboratory room (SUPSI Univ., Lugano), where the temperature sensor of the instrument @seismicisolation @seismicisolation

Torrent NDT method 151

Figure 5.14 Effect of temperature on kT measured with PermeaTORR AC.

reported values between 20°C and 25°C, with an average of 23.4°C. The same samples were stored outdoors (protected with plastic sheets to avoid changes in moisture) at the offices of Materials Advanced Services in Coldrerio, Switzerland, where they were measured again in January 29 and 31, 2017, with reported temperatures of 5.7°C and 8.6°C, respectively. Figure 5.14 shows the results obtained on the samples under the different temperature conditions. The results in Figure 5.14 support the proposed change in the Swiss Standard, allowing the measurement of kT for temperatures down to 5°C, instead of just 10°C. 5.7.1.2 Influence of High Air Temperature and Solar Radiation During the interlaboratory test described in Section 5.6.4.2, unacceptably high calibration values (≈ 8 mbar) were observed when the TPT was calibrated at noon. This was attributed to the influence of solar radiation; exposed to the sunlight, the metal pressure regulator became very hot (≈43°C); on the contrary, the calibration values in the morning ranged between 2 and 4 mbar (Jacobs et al., 2009). Therefore, it is important to protect the instrument, whatever the brand and model, from direct exposure to solar radiation, by means of umbrellas or canopies.

5.7.2 Influence of Moisture of Concrete Surface The strong influence the moisture content has on the gas-permeability of concrete is discussed in general terms in Section 6.9. @seismicisolation @seismicisolation

152 Concrete Permeability and Durability Performance

From the very beginning this influence was recognized as a formidable hindrance for the measurement of kT on site, due to the virtual impossibility of conditioning the test area prior to the measurement. The same problem was faced by other developers of site methods to test the air-permeability of concrete. For instance, Parrott (1994) considered a conversion factor between the air-permeability kr measured with Hong-Parrott method (see Section 4.3.2.2) at a certain relative humidity r and at a reference relative humidity of 60% (k60). Similarly, in the case of the Autoclam System (see Section 4.3.2.7), its Operating Manual states “…it is recommended that tests are carried out when the concrete is relatively dry (i.e. when the internal relative humidity of the cover concrete up to a depth of 10 mm is less than 80%)”. The relative humidity of the cover concrete is typically measured with a probe inserted in a sealed cavity (10 mm deep) inside a drilled hole (Torrent, 2005). The importance of this matter became obvious as soon as the newly developed “Torrent Method” was intended for site application. Since the test method is entirely non-destructive, the influence of surface moisture had to be dealt with also in an entirely non-destructive manner. The first attempt was to use a surface moisture device called “H 2O meter”, offered in the early 90s by James Instruments, that operated by pressing a sensor head onto the concrete surface. After trying it on specimens that had different moisture contents, the instrument was discarded. The alternative was to combine the kT measurement with a complementary test, in order to compensate the influence of moisture. A suitable complementary property should show a higher “penetrability” for lower quality concrete (same as kT), and show a higher “penetrability” for higher moisture content (opposite to kT). A property that fulfils these conditions is electrical conductivity, with the advantage that it can be measured nondestructively by the Wenner method (in fact its reciprocal, the electrical resistivity ρ). After some experimental work, reported in Torrent and Ebensperger (1993), a criterion was established by which a compensation of the measured value of kT (kTm) was introduced via a parallel measurement of ρ. This compensation was of the form (Torrent & Frenzer, 1995):  3.5kTm  kT ′ = Max  kTm ;  ρ   where kT′ = air-permeability compensated by moisture (10 −16 m²) kTm = measured air-permeability (10 −16 m²) ρ = measured electrical resistivity by the Wenner method (kΩ.cm) @seismicisolation @seismicisolation

(5.34)

Torrent NDT method 153

Equation (5.35) gives the limiting ρ values, as function of kT, below which the correction was necessary. It has to be mentioned that, at the time of developing the concept, virtually all concretes in Switzerland were made with OPC. With the advent of composite binders, containing PFA, SF or POZ, the correction ceased to be useful, since the inclusion of active additions can increase the ρ values by a factor of 5–10 (RILEM TC 154-EMC, 2000) and a specific compensation should be built for each binder type. Other factors that were decisive in abandoning the combined kT–ρ approach were: • the dependence of ρ with the temperature and the vicinity of steel bars (Jacobs et al., 2009, p. 641) • the fact that, very often, the surface moisture of the site concrete is low enough to break the continuity of the electrolyte (the pore solution) making it impossible to get a reading of ρ (Jacobs et al., 2009 – Table C-2) The combination of Torrent air-permeability and Wenner electrical resistivity has recently been revisited by Bonnet and Balayssac (2018), who concluded that “the results show that resistivity and Torrent permeability can be used for the combined assessment of carbonation depth and saturation degree in laboratory conditions”. Based on data obtained at the Hong Kong-Zhuhai-Macao sealink, a similar approach has been proposed by Li et al. (2019). In addition, the advantages of combining different NDT techniques for assessing the service life of reinforced concrete structures has been advocated, based on intensive experimental work, by Sofi et al. (2019). A research conducted at EMPA in Switzerland (Romer, 2005a), demonstrated the suitability of an impedance-based moisture meter (“Concrete Encounter Moisture Meter”, from Tramex) to monitor the drying process of concrete samples, corresponding to four different mixes, stored in rooms at 35% and 70% RH, see Figure 5.15. This research opened the way to an entirely and successful new approach, adopted by Swiss Standard (SIA 262/1-E, 2019), by which the air-permeability can be measured on site, only if the surface moisture of concrete, measured by an impedance-based instrument, does not exceed 5.5%. This limit derives from the work of a team of Swiss experts (Jacobs et al., 2009); according to Paulini (2014), this limit is too high. The electrical-impedance method to assess the surface moisture of concrete has been standardized in USA for application on concrete floors (ASTM F2659, 2015). The next sections investigate the matter in more detail. @seismicisolation @seismicisolation

154 Concrete Permeability and Durability Performance

Figure 5.15 Monitoring the drying process of samples stored at 35% and 70% RH, with Tramex instrument, data from Romer (2005a).

5.7.2.1 Influence of Natural and Oven Drying on kT A comprehensive investigation was conducted by Torrent et al. (2014, 2019) to explore in more detail the dependence of kT with the surface moisture m, measured by an electrical impedance-based instrument. For this, 150 mm cubes were cast with two concrete mixes (w/c = 0.40 and 0.65) made with 9 widely different binders (see Table 5.4), totalling 17 concrete mixes. The first letter of the code indicates the clinker used to produce the cements (H: Höver, Germany; M: Merone, Italy). The values in parentheses indicate the content and type of mineral additions originally included in the cement (MIC). When a mineral addition was added separately as

Table 5.4 Binders used in the concrete mixes Code H0 H8M H22S H41S H68S M0 M26L M31FL M40FL

Brand (MIC) + SCM Holcim Pur-5N 92 % Holcim Pur-5N + 8% Silica Fume Holcim Ferro 4 (22% GBFS) Holcim Duo 4 (41% GBFS) Holcim Aqua 4 (68% GBFS) I 52,5 R II/B-LL 32.5R (26% Limestone Filler) IV/A 32,5 R (27% PFA + 4% LF) 87% IV/A 32,5 R +13% PFA

@seismicisolation @seismicisolation

EN 197 class CEM I 52.5 N CEM II/A-D CEM II/B-S 42.5R CEM III/A 42.5 N CEM III/B 42.5 L-LH/HS/NA CEM I 52.5 R CEM II/B-LL 32.5 R CEM IV/A 32.5 R CEM IV/B 32.5

Torrent NDT method 155

supplementary cementitious material (SCM) into the concrete mix, the content and type are indicated in italics. All 34 cubes (two per mix) were cured under water at 20°C for around 3 months, to ensure a high degree of hydration, so as to minimize further hydration during the drying period. After the moist curing period, one cube of each mix was exposed to natural laboratory drying (stored in a room with still air at 18°C–23°C and 50%–65% RH). The companion cube was placed in a ventilated oven at 50°C (tests of the oven-dried cubes were performed after 24 ± 2 hours cooling in the laboratory dry room). After the tests were completed (over 3 years for the lab-stored and over 4 months for the oven-dried specimens), the cubes were immediately returned to the dry room or oven, respectively. At intervals, the following NDT and instruments were applied on the cubes:

Tests b and d (dry tests) were applied on two opposite lateral faces of each cube. Two readings of m were made on each face, one in a horizontal position (mh) and the other in a vertical position (mv). The reported value m is the average of the four readings on each cube. One reading of kT was performed on two opposite faces of each cube, the reported value being the geometric mean of the resulting two values (assuming that kT follows a lognormal distribution, see Section 5.8). The results of electrical resistivity ρ are not discussed here because, after relatively short drying periods, it was impossible to get valid readings (Torrent et al., 2014). When the 50 °C oven-dried specimens had completed at least 100 days of drying, they were weighed and dried at 105°C till constant weight, recording the mass M105, and measuring kT and m. Then, the samples were submerged under water at 20°C until constant weight, recording the saturated mass M s. This allowed converting the values of mass M of all the cubes into degrees of saturation Sd (-) using Eq. (5.36). The change in m with the change in Sd is presented in Figure 5.16 for concretes made with Höver clinker. All specimens’ results are plotted in the charts, including those under laboratory and oven drying, the latter identified by the framed symbols. It is worth mentioning that all the cubes showed that @seismicisolation @seismicisolation

156 Concrete Permeability and Durability Performance

• after drying to constant mass at 105°C (Sd = 0.0) → m = 0.0% (bottom of instrument’s scale) • after saturation to constant mass (Sd = 1.0) → m = 6.9% (top of instrument’s scale) This situation is indicated by the solid black line (indicative only) joining the extreme black squares in Figure 5.16, which presents the results obtained on the concretes based on Höver clinker, very similar to those obtained with Merone clinker. Figure 5.16 shows the monotonic relation existing between Sd and m along the whole drying process, with the oven-dried specimens showing a reasonable continuity with those dried in the lab room. It was observed that the binder type exerts a significant influence on the m vs Sd relationship. It can be seen from Figure 5.16 that, for the same m, concretes made with H0 show lower values of Sd than those containing hydraulic MIC and/or SCM. In Figure 5.16, a segment with m = 5.5% (maximum value admitted by Swiss Standard (SIA 262/1-E, 2019) for performing air-permeability test kT on site) has been drawn, indicating corresponding Sd values within 0.75–0.90 for Höver samples; the range for Merone clinker concretes was 0.60–0.90. It must be borne in mind that the saturation degree Sd corresponds to the bulk of the cube, whilst the surface moisture m corresponds to a layer about 15–20 mm thick. The degree of saturation of that layer must certainly be lower than the bulk value, due to the moisture gradient established during drying.

Figure 5.16 Relation between saturation degree Sd and surface moisture m (Höver clinker).

@seismicisolation @seismicisolation

Torrent NDT method 157

Figure 5.17 Relation m vs. kT (Höver clinker).

Figure 5.17 presents the relation found between kT and m, along the whole drying process, with the first reading taken after 1 day of drying, for concretes made with Höver clinker (a similar picture for Merone clinker can be found as Figure 6 of Bueno et al. (2021)). The values inside the broken-line box are not sufficiently accurate as they lie below the bottom limit of measurable kT (0.001 × 10 −16 m²). Figure 5.17 shows a monotonic increase in kT as concrete gets drier, with the highest kT value corresponding to the cubes subjected to the drastic final drying at 105°C (m = 0%). At first sight, the relation ln(kT) vs. m presented in Figure 5.17 shows some linearity and parallelism for the different mixes, particularly for values of m within the range 1.0%–6.0%. This opens the way for attempting a compensation of kT when the m values are too low (e.g. indoors elements), as discussed in the next section. A similar picture was obtained by Kato (2013) who dried specimens of different configurations, intensively hydrated (28 days in water at 60°C), for 91 days at 20°C/60% RH. For w/c = 0.50 and 0.70 he found an excellent linear correlation between log kT and the water loss of the specimens, with virtually the same slope. 5.7.2.2 Compensation of kT for Surface Moisture An approach to compensate the kT values for the moisture content of concrete was developed by Misák et al. (2010). They prepared concrete slabs (300 × 300 × 100 mm) of an EN 206 C20/25 class concrete that were moist cured for 28 + 2 days and then stored in a dry room (23°C/48% RH), monitoring the influence of drying on kT. At a certain point, the samples were oven-dried at 50°C and finally at 105°C. The surface moisture content w @seismicisolation @seismicisolation

158 Concrete Permeability and Durability Performance

was measured with a KAKASO capacitive humidity meter that was calibrated against bulk gravimetric moisture content. The KAKASO moisture meter hygrometer uses the property that water in the capillary pore environment greatly affects its permittivity; the instrument explores the 15–30 mm surface layer (Misák, 2018). Results of 153 measurements of kT against the humidity obtained from the capacitive instrument readings were reported by Misák et al. (2010), converted into bulk humidity w through the established calibration curve, which differs slightly for different materials. Based on these results, the following relation between kT and w was proposed: For the particular mix investigated and their own moisture determination, the regression analysis yielded kT0 = 5.25 × 10 −16 m² and α = 0.8623; where kT0 is the kT value for w = 0 and α the slope of the line ln(kT) – w. There is a strong similarity between the data from Misák et al. (2010) and those presented in Figure 5.17. Therefore, a similar approach was pursued, but taking m (impedance-based) as the direct indicator of the surface moisture content of the concrete. An initial analysis was made on the basis of the data obtained by Torrent et al. (2014, 2019), part of which are shown in Figure 5.17, that showed quasi-linear relations between log(kT) and m, to which regressions of the form indicated by Eq. (5.38) were fitted: Only the data with 1.0% ≤ m ≤ 6.0% and with kT ≥ 0.001 × 10 −16 m² were considered for the regression analysis. The mean correlation coefficient was R = 0.93, with extreme values of 0.65 (Mix H8M-40) and 0.99 (Mix M26L-40). The regression analyses showed that, despite the wide range of binders (Table 5.4) and w/c ratios used to prepare the concrete mixes, exponent δ remains within a limited range of 1.00–1.65 for 16 out of the 17 mixes investigated. Relying on a single source of data for the analysis was deemed insufficient, reason why a more comprehensive cooperative investigation on the subject was organized, based on available data from other sources (Romer, 2005a, b; Nsama et al., 2018; Sandra et al., 2019; Nguyen et al., 2019, 2020). The main results of it were presented in Bueno et al. (2021) and are summarized in Figure 5.18, which shows the histogram of δ values obtained by regression of Eq. (5.38) on 50 series of test data, coming from the abovementioned sources, including the 17 obtained from Torrent et al. (2014, 2019) data. @seismicisolation @seismicisolation

Torrent NDT method 159

Figure 5.18 Histogram of exponent δ values obtained on 50 series of test data.

Figure 5.18 shows that the statistical distribution is positively skewed with a central value that can be assumed equal to the median δ = 1.45; moreover, 84% of the δ values fall within the range of 1.0–2.0. The average value of the correlation coefficient R for the 50 reported cases is R = 0.95. Based on the results presented in Figure 5.18, a single compensation formula is proposed in Eq. (5.39), using the median δ value of 1.45, where kTm is the air-permeability value measured at surface moisture m: The value kT0 corresponds to the extrapolation of Eq. (5.39), valid for 1.0% ≤ m ≤ 6.0%, to m = 0%, giving an unrepresentative high reference value. Therefore, it is proposed to take as reference the value kT5, corresponding to a moisture m = 5.0%. From Eq. (5.39), we can write: Dividing Eq. (5.40) by Eq. (5.39) and introducing a compensation factor F5 to the value kTm measured at moisture m: with For m values between 4.5% and 5.5%, the compensation is not truly necessary, because the corresponding kT values differ from kT5 by a factor of @seismicisolation @seismicisolation

160 Concrete Permeability and Durability Performance

2 and 0.5, respectively, which is acceptable given that kT varies over six orders of magnitude. An interesting real case is presented in Section 11.4.4, where measurements conducted in the Atacama Desert in Chile, where extremely dry concrete was tested (m < 2%), resulted in large differences between kTm and kT5, showing the usefulness of the moisture compensation. 5.7.2.3 P re-conditioning of Laboratory Specimens for kT Measurements It is of practical interest to establish some guidelines on the preconditioning of laboratory specimens for kT measurements. In Annex B, guidelines are given for testing kT in the laboratory, case not covered by Swiss Standard (SIA 262/1-E, 2019) which refers exclusively to site testing. Regarding preconditioning of the samples, the guidelines specify the following procedure: At the age of 28 days, the specimens shall be dried in the ventilated oven at a temperature of 50 ± 2°C, leaving a free distance of at least 20 mm between the specimens and with the walls of the oven. The drying will continue until the moisture meter indicates a surface moisture of the specimens within the range 4.0%–5.5% (normally this is achieved within 4 ± 2 days of drying of specimens at or near saturation). This criterion has been applied to the data recorded in the investigation described in Section 5.7.2.1, with the result shown in Figure 5.19 (Torrent et al., 2019). Figure 5.19 shows the kT values obtained for the 17 mixes investigated, in abscissas for laboratory drying until m ≤ 5.5% (limiting value prescribed by SIA 262/1-E (2019)) and in ordinates after 3 days in the oven at 50°C to reach a surface moisture 4.0% ≤ m ≤ 5.5%; if not achieved,

Figure 5.19 Relation between 50°C oven-dried (3 or 6 d.) and lab-dried (till m ≤ 5.5%) kT values.

@seismicisolation @seismicisolation

Torrent NDT method 161

Figure 5.20 Effect of oven-drying conditions on the measured kT values.

the value obtained after 6 days in the oven was taken. It has to be mentioned that, in the case of Mix H0–40, after 3 days of oven-drying, the moisture dropped already to 3.8%. Only for mixes M0–65 and M26L–40, was it necessary to dry the specimens 6 days to reach m ≤ 5.5%, indicated by triangles in Figure 5.19. Figure 5.19 shows that the criterion works well in that the kT values obtained according to the oven-drying procedure prescribed in Annex B leads to very similar values to those obtained under laboratory drying until reaching m ≤ 5.5%. Figure 5.20 shows the kT values for each mix, measured under different oven-drying conditions: the white bars correspond to oven-drying 3–6 days at 50°C, whilst the dark bars to oven-drying at 105°C to constant mass. It can be seen that, for 50°C oven drying, the range between the kT value for H8M-40 mix (lowest) and for M26L-65 mix (highest) covers three orders of magnitude, whilst for 105°C oven-drying the range is reduced to just one order of magnitude (see arrows in Figure 5.20). This confirms the well accepted practice of drying specimens for gas-permeability tests not above 50°C (see Section 4.3.1.1).

5.7.3 E ffect/Influence of Age on kT Originally, the Torrent test method to measure kT was designed for testing concretes at a mature stage, but not too old. Indeed, Swiss Standard (SIA 262/1, 2019) prescribes that the quality of the cover concrete has to be measured at ages between 28 and 120 days. @seismicisolation @seismicisolation

162 Concrete Permeability and Durability Performance

Nevertheless, there is some interest in measuring kT at earlier ages, say 7 days, so as to know well in advance whether the quality achieved on site matches the specifications or expectations of the stakeholders. On the other extreme, when dealing with condition assessment of existing structures, exposed to the environment for many years, there is a need to assess kT data measured on concrete which is decades old. In the following, both aspects are dealt with in some detail. 5.7.3.1 Effect/Influence of Age on Young Concrete When testing kT on young concrete exposed to drying, two superimposed phenomena happen: a reduction in kT due to the continued hydration of the binder and an increase of kT due to drying as discussed in Section 5.7.2. Figure 5.21 presents data from EPFL in Switzerland (Brühwiler et al., 2005) obtained on 800 × 800 × 200 mm reinforced concrete panels made with OPC mixes of w/c = 0.43 and w/c = 0.52. Face A of the panels was cast against the form, whilst face B was cast against a permeable formwork liner (“Zemdrain”), described in Section 7.2.3. The lateral faces (800 × 200 mm) of the panels were sealed with epoxy resin. The panels were moist cured for 7 days and thereafter kept in a dry room (20°C/60% RH). The coefficient of air-permeability kT of the four faces was measured at intervals between 4 and 389 days of exposure to the dry environment, with the results shown in Figure 5.21. A consistent trend can be observed in Figure 5.21 for the measurements on the four surfaces in that, up to 11–21 days of exposure, kT decreases significantly, moment at which the trend is reversed and a continuous increase in kT is observed until the end of the experiment (389 days). At an exposure time around 100–200 days, the value of kT returns to the one measured after 4 days of drying.

Figure 5.21 Evolution of air-permeability kT with exposure time, data from E. Denarié (Brühwiler et al., 2005).

@seismicisolation @seismicisolation

Torrent NDT method 163

The initial reduction in kT is attributed to the changes in microstructure of the young concrete, due to continued hydration. After 11–21 days exposure, the rate of hydration diminishes and the influence of drying prevails, by which more and more pores are freed, facilitating the flow of air, resulting in a gradual increase in kT. Interesting to see is that, based on either the 4 or 389 days kT data, the best quality surfaces are those treated with “Zemdrain” (black symbols +ZD), followed by the one with w/c = 0.43, the worst being the one with w/c = 0.52 (both without “Zemdrain”). In the Application Test, organized by RILEM TC 230-PSC (2015), described in Section 5.6.4.2, tests were performed on eight concrete panels, made with different binders and w/c ratios. The panels were measured for kT at “young” 14–21 days of age (maturity age around 10 days) and, again, at “mature” 101–108 days of age. The ambient conditions were quite different during both tests, as described in Section 5.6.4.2. Figure 11.8 of RILEM TC 230-PSC (2015) presents the geometric means of kT results obtained with the same instrument and operator on the eight panels at both ages. It shows that the original kT values measured on the panels at “young” age are about one order of magnitude smaller than those obtained at a “mature” age. Interesting to note, though, is that there is a certain correlation between both sets of values. If this correlation could be established in advance, it would be possible to predict the kT values at “mature” ages from kT values measured on “young” concrete. 5.7.3.2 Effect/Influence of Age on Mature Concrete By repeating field tests on old bridges located outdoors, unfortunately not always on the same spots, Adey et al. (1998) concluded: “Permeability measurements are repeatable, on existing concrete bridges, when taken in exactly the same location and there has been no change in internal water content” and stated: “The presence of water in the concrete may cause differences in the permeability that cannot be adjusted using the resistivity measurements”. The latter statement already objected to the use of electrical resistivity to compensate for moisture content in the concrete, applied at the time of the paper, later substituted by the use of impedance-based moisture meters (see Section 5.7.2). A similar experiment was conducted by Imamoto et al. (2012), reported in slide 14 of the paper’s presentation, reproduced as Figure 5.22. They repeated kT measurements on two locations (black and white bars) of an external beam of 50 years old Tokyo’s National Museum of Western Art (investigation described in detail in Section 11.4.1.1), at intervals during a period of about 3 months. Figure 5.22 indicates a very good stability of the results on both locations, even when measuring 1–2 days after a heavy rain. In order to simulate the influence of the environment on the long-term values of kT of 0.9 × 0.9 m cross section columns, Kato (2013) prepared 450 × 450 × 400 mm specimens, sealing them as shown in Figure 5.23 (l.). @seismicisolation @seismicisolation

164 Concrete Permeability and Durability Performance

Figure 5.22 Stability of kT readings at two locations on an external beam of Tokyo’s NMWA (Imamoto et al., 2012).

Figure 5.23 Simulation of 0.9 × 0.9 m column (l) and effect of exposure on kT (r) (Kato, 2013).

After an intensive curing (28 days under water at 60°C) and 5 days sealing, the sealed specimens were left drying for 91 days at 20°C/60% RH, before being exposed unprotected to the natural environment. The evolution of kT of the sealed specimens placed outdoors is presented in Figure5.23 (r.) (dots). In some cases of rain (vertical bars), the water was wiped off the surfaces and kT measured on the moist surface. Figure 5.23 (r.) shows that the kT values remained reasonably stable, except at 152 days of age, when measured in correspondence with a heavy rainfall, where the values in the center and bottom part of the specimens dropped by more than one order @seismicisolation @seismicisolation

Torrent NDT method 165

of magnitude. Yet, as remarked by Kato (2013), “just two days later kT had almost returned to the state before the rain”. In a research by Yokoyama et al. (2017), full-scale concrete elements were cast with four different mixes, made with OPC and GBFS cements, subjected to different curing conditions and exposed to the environment during 6–7 years, monitoring the resulting changes in kT with time. After a significant increase in kT between 3 and 6 months, the values stabilize until around 45 months, after which some further increase was recorded. The main conclusion of the research by Yokoyama et al. (2017) is “Results showed that the difference in the air-permeability owing to curing conditions decreased with age. About one year later, the coefficient of air-permeability of specimens exposed to rainfall was approximately the same regardless of the curing method. The effect of curing was present even after one year for specimens under the roof-not exposed to rain, but the difference in the coefficient of air-permeability continued to decrease. On the other hand, the difference in surface airpermeability between concretes with different water to cement ratios remained even at the age of five years”. This section can be concluded with a practical recommendation of not measuring kT during a rainfall and to wait at least 2 days after the rain to carry out site measurements of exposed structures. In case of monitoring changes in kT with age, it is essential to mark the exact location of the test, to make repetitions precisely on the same spot.

5.7.4 I nfluence of Vicinity of Steel Bars The presence of steel bars may have two effects on the measured kT values. On the one hand, the presence of the rigid steel bars may produce some aggregates particles segregation, may affect the compaction of the cover concrete and also create a sort of ITZ around them, that may affect the true “penetrability” of the Covercrete. On the other hand, the presence of the bars may obstruct the flow of air towards the central chamber and, thus, affect the measured kT value. Within the comprehensive research sponsored by the Swiss Federal Highway Administration (Torrent & Ebensperger, 1993), a special investigation on the influence of the vicinity of steel bars was conducted. For that, two sets of four concrete slabs (120 × 250 × 360 mm) were cast with a mix of the following characteristics: • • • •

OPC content: 325 kg/m³ w/c = 0.46 32 mm maximum size of aggregate 75 mm slump @seismicisolation @seismicisolation

166 Concrete Permeability and Durability Performance

• 4.8% entrained air • Standard cube strength at 28 days = 43.0 MPa The slabs were cast in two layers, each one of them compacted on a vibrating table and carefully finished with a thick metal ruler. From each set of four slabs, one was unreinforced (used as Control) and the other three contained Ø18 mm bars, according to the pattern shown in Figure 5.24 (l.). The steel bars of the three reinforced slabs had a variable cover thickness of 11, 26 and 41 mm, respectively. In one set (AO) the bars were placed close to the upper (finished) surface as cast; the four slabs of this test did not receive any moist curing. In the other set (BU), the bars were placed close to the bottom (moulded) surface as cast; the four slabs of this set had an initial curing of 7 days in the moist room (20°C/RH > 95%). The kT tests were performed at the age of 90 days in a dry room (20°C/50% RH) where all slabs were stored after the moist curing (0 and 7 days). Six kT readings were performed on each slab; in the case of the reinforced slabs, the central chamber of the vacuum cell was placed above each of the 3 bars (see Figure 5.24 (l.)). Afterwards, one Ø150 mm core, per slab, was drilled (across the steel bars, see Figure 5.24 (l.)) and cut to 50 mm thickness, to measure the oxygen-permeability kO (Cembureau method, described in Section 4.3.1.2). The Ø150 × 50 mm discs were oven-dried at 50°C for 6 days and cooled down 1 day in a desiccator before performing the kO test. Figure 5.24 (r.) shows the results of the measurements for both slab sets (AO and BU), with kT being the median value of the six test results of airpermeability on each slab (corrected by factor 1.846 as described at the end of Section 5.3.4) and kO the single value of oxygen-permeability. The plain “Control” slab is represented by a “cover thickness” = 120 mm (thickness of the slab); data from Table 4-II of Torrent and Ebensperger (1993).

Figure 5.24 Sketch of “reinforced” slabs (l) and effect of steel bars on kT and kO (r).

@seismicisolation @seismicisolation

Torrent NDT method 167

Figure 5.24 (r.) indicates that the presence of the steel bars increases the permeability of the Covercrete, compared with that of plain concrete. The oxygen-permeability kO is raised by a factor of 2–3, whilst the airpermeability kT by a factor of 3–5. It seems that kT experiences a “jump” for the shallowest cover of 11 mm, which may indicate some effect of the vicinity of the bar on the quality of the Covercrete; the expected influence on gas flow into the central chamber would have been a decrease of kT (obstructing air-flow), that was not observed. The influence in any case should depend on the penetration of the test L (in turn, function of kT, see Figure 5.7). This influence could be investigated by modelling. In a comprehensive research, Eddy et al. (2018) investigated the effect of shallow covers on the Covercrete quality. For that purpose, reinforced concrete specimens were cast with Ø19 mm steel bars embedded near the bottom-as-cast surface, with shallow cover thicknesses of 5, 15 and 30 mm, insufficient according to Japanese standards for durability (minimum cover = 40 mm). Two concrete qualities were investigated, namely w/c = 0.45; OPC = 378 kg/m³; 91 days f′c cyl = 49.5 MPa and w/c = 0.60; OPC = 283 kg/m³; 91 days f′c cyl = 38.6 MPa. The specimens were sealed-cured for 7 days and thereafter kept indoors or exposed unprotected to Tokyo’s summer-autumn environment until 91 days of age. The pore structure of small mortar samples taken at different depths, including the cover in contact with the steel was investigated by MIP. Also, measurements of air-permeability kT, surface moisture content (impedance-based method) and Wenner electrical resistivity were conducted. The results of air-permeability tests indicated that the kT measured on the 5 mm Covercrete was significantly higher than that on the 30 mm Covercrete or on plain concrete samples, whilst for 15 mm Covercrete the difference in kT was smaller. Since the MIP results also showed a more open pore structure for shallow covers it was concluded that the differences in kT are due to differences in quality of the Covercrete rather than to an interference of the bar with the air flow. Based on the results of Figure 5.24, Swiss Standard (SIA 262/1-E, 2019) prescribes a minimum cover thickness of 20 mm for the measurement of kT in coincidence with steel bars.

5.7.5 I nfluence of the Conditions of the Surface Tested 5.7.5.1 I nfluence of Specimen Geometry and Surface The measurement of kT in the laboratory poses the question about suitable shape and size of specimens and surfaces to be investigated. Successful tests made on 150 and 200 mm cubes, Ø150 mm cylinders and disks and slabs (e.g. 250 × 360 × 120 mm) have been reported. Since the external diameter of the cell is 110 mm, this is a practical limit to the size of the surface to be tested. A reasonable limit to the minimum size of a surface @seismicisolation @seismicisolation

168 Concrete Permeability and Durability Performance

to be tested is a diameter or side of at least 150 mm, whilst the depth should not be less than 50 mm, capable of accommodating the vacuum front penetration up to kT values of 2 × 10 −16 m² (see Figure 5.7) without the need of correcting the measured value for depth. An investigation dealing with the influence of the surface and geometry of the specimens on the measured kT values was conducted by Neves et al. (2015). With that aim, a series of 12 concrete mixes, made with different binders and w/c ranging between 0.43 and 0.61, were prepared, with which Ø150 × 300 mm cylinders, 200 mm cubes and 250 × 360 × 120 mm slabs were cast. From some cylinders, five slices 50 mm thick were obtained by saw-cutting. All specimens were moist cured for 7 days to be later kept in a dry room (20°C/60% RH) until the moment of test at 28 days of age. Not all specimens were prepared from each mix. A non-parametric statistical analysis of all the test results allowed (Neves et al., 2015) to conclude: “The investigation on the influence of specimens’ geometry and type of surface on concrete air-permeability leads to conclude that there are no differences, with statistical significance, either when air-permeability is assessed in any of the three geometric shapes of the specimens used in this work, or when it is assessed in moulded or sawn surfaces. As no significant differences between tested geometries and surfaces of specimens were found, the selection of the specimen geometry and testing surface may vary according to the specific requirements of each use.” In general, it is a good practice to measure kT on moulded surfaces of specimens, as they are less affected by preparation, unless the effect of finishing techniques is to be assessed. Also, if unduly extensive vibration is applied to the specimens (even if shallow), segregation and excessive bleeding can occur, affecting the trowelled surface (see Section 6.6). Moreover, the curing of the trowelled surface may be different to that of surfaces protected by the form until stripping. When testing saw-cut surfaces, the flow of air from the exposed aggregates goes straight into the vacuum cell, whilst in reality, it always flows through a layer of h.c.p.; this is even more important when dealing with highly porous aggregates. To conclude, the best specimens for conducting kT tests in the laboratory are cubes, at least 150 mm side length; for countries following ASTM standards, based on cylinder strength, the moulds for 150 × 150 mm crosssection beams, specified for flexural tests, can be used with metal separators to produce 150 mm cube-like specimens. The cube offers four lateral faces that can be considered identical for measuring kT. 5.7.5.2 Influence of Curvature The vacuum cell of the different instruments measuring kT according to the Torrent method has been designed for application on surfaces that are relatively flat. @seismicisolation @seismicisolation

Torrent NDT method 169

Practice indicates that, due to the soft rubber rings with which the cell is equipped, a certain curvature of the surface can be accommodated. For instance, Figures 11.12 and 11.48 show direct applications on curved surfaces, whilst Figure 5.25 shows two solutions developed by K. Imamoto to measure kT on cylindrical pillars. Figure 5.25 (l.) shows a flexible adaptor applied in Tokyo’s Museum of Western Art (Imamoto, 2012; Imamoto etal., 2012) and Figure 5.25 (r.) a rigid one applied in a building in Locarno (Switzerland). It is important to check that the accessory used, such as the ones in Figure 5.25, does not change the constant of the instrument Vc /A (seeEq.5.13); otherwise, a correction of the measured value is necessary. 5.7.5.3 Influence of Roughness In an investigation by Brühwiler et al. (2005), conducted on site on a road overpassing the railway, the concrete pavement to be tested showed a surface with irregularities, cracks and cement laitance due to bleeding. It was decided to conduct comparative measurements on the pavement natural surface and also after the application of sand-blasting to expose the aggregates. Three kT measurements were conducted on each surface – original and sand-blasted – with geometric means of 0.045 and 0.055 × 10 −16 m², respectively, showing an insignificant effect of the sand-blasting on kT. Regarding roughness, Figure 5.26 (l.) shows the application of the PermeaTORR on the rough surface of a concrete pavement, during an investigation conducted by TFB in Switzerland. Sometimes, the irregularities are such (e.g. shotcrete or grooved pavements) that the surface must be treated before application of the test method. A real case happened in the city of Buenos Aires, Argentina, where the airpermeability of trench walls was measured on the excavated hidden side, due to uncertainties on the quality of the job done by the contractor. The surface was extremely rough, with a thick layer heavily contaminated with

Figure 5.25 Flexible (l.) and rigid (r.) adaptors for curved surfaces.

@seismicisolation @seismicisolation

170 Concrete Permeability and Durability Performance

Figure 5.26 kT test on concrete pavement surface (l., courtesy F. Jacobs, TFB) and on polished trench wall (r.)

bentonite, requiring the use of a polishing machine (Bosch GWS 6–115 Professional) to prepare the test areas. Figure 5.26 (r.) shows a test being performed on a polished area. Despite the awful aspect of the raw surface, the tests yielded good results in terms of Permeability Classes, as defined in Table 5.2: 15% in PK0 (Negligible); 42% in PK1 (Very Low); 28% in PK2 (Low) and 15% in PK3 (Moderate). So, the aspect of the concrete per se is not a good indicator of its quality. 5.7.5.4 Effect/Influence of Surface Air-Bubbles Large air bubbles (bug-holes) tend to appear on concrete surfaces cast against metal forms, especially when inclined, as they find it difficult to escape even under heavy vibration. The appearance of bug-holes often attracts the attention of the owners, concerned on how they can affect the durability of the elements. To start with, in reinforced concrete structures, bug-holes may reduce the cover thickness, thus promoting an early localized corrosion. An interesting real case is described in Section 11.3.4, in which comparative kT tests performed exactly on bug-holes (see Figure 11.23) and away from them showed no significant difference. A dedicated investigation, in the laboratory and in the field, on the effect of bug-holes on the permeability of tunnel liners was reported by Hirano et al. (2014) and Maeda et al. (2014). In the laboratory, 300 × 300 mm prismatic specimens with a height of 750 mm were cast using a metal form that could be rotated to simulate different inclinations. They investigated the effect of the demoulding agent, of the vibration and of surface treatments (ceramic coating or permeable formwork liners) on the appearance of air bubbles and on the resulting kT measurements. The investigation included

@seismicisolation @seismicisolation

Torrent NDT method 171

site measurements on treated and untreated surfaces of two tunnels’ liners, the concentration of air bubbles (m²/m²), called “bubbles ratio”, being computed from image analysis of pictures taken from the resulting surfaces. The air-permeability kT was measured on the same surfaces. Figure 5.27 shows the influence of the inclination angle (from vertical position) on the “bubbles ratio”, without a clear positive influence of the ceramic coating. The research showed a strong effect for air bubbles ratios on kT, especially between 0% and 0.5%, the effect becoming less marked for higher bubbles ratios. The action and positive effect of permeable formwork liners in reducing the permeability of the concrete surface is discussed in Section 7.2.3. Figure 5.28 shows how the use of one of these liners (“Zemdrain”) has eliminated the bug-holes by allowing the air bubbles to escape through the fabric. The left-hand zone of the element (untreated) shows plenty of bugholes, which disappeared from the right-hand zone of the element, treated with “Zemdrain”. The laboratory results reported by Hirano et al. (2014) and Maeda et al. (2014) indicate that the presence of air bubbles increases significantly the air-permeability kT (unfortunately, the results of site tests on the tunnels’ walls were not reported nor commented), which is in contradiction with what is presented in Section 11.3.4.4. More research is needed to elucidate this matter. Nevertheless, when testing concrete surfaces showing high concentration of relatively large bug-holes, care has to be taken to avoid that the air bubbles “shortcut” the borders of the concentric chambers of the vacuum cell.

Figure 5.27 Effect of form inclination on bubbles ratio (Hirano et al., 2014).

@seismicisolation @seismicisolation

172 Concrete Permeability and Durability Performance

Figure 5.28 Visible effect of “Zemdrain” (right-hand zone of element) in eliminating bug-holes.

5.7.6 Influence of Initial Pressure P 0 Occasionally, the initial pressure of the test (P0) at time t 0 may be abnormally high, say above 100 mbar. This may be due to a number of factors: rough surface that impedes a tight sealing of the cells, high porosity of the substrate under test, low vacuum capacity of the pump, damaged O-rings, etc. In principle, P0 and, in general the pressure of the inner chamber should not be too high for two reasons: • the cell can be detached from the surface, fall and get damaged (an emergency safety level of 200 mbar is ensured in the PermeaTORR AC and AC+ instruments to prevent its expensive active cell of falling) • the derivation of the formula to calculate kT (Eq. 5.13) assumes that P in the central chamber is much lower than the atmospheric pressure Pa (Eq. 5.8) Therefore, although an occasional high P0 value does not necessarily mean that a wrong kT result will be obtained, the recurrent testing at high P0 values should be avoided and, when happening permanently, the instrument should be checked. On the other extreme, very low P0 values (below 30 mbar), especially when testing low-permeability concretes with the TPT, may lead to artificially high kT values, as discussed in Sections 5.3.5.2 and 5.6.4.2.

5.7.7 Influence of Porosity on the Recorded kT Value The formula used to compute kT from the recorded rate of pressure increase in the inner chamber, Eq. (5.13), contains explicitly the porosity ε of the @seismicisolation @seismicisolation

Torrent NDT method 173

concrete tested. This value is usually not known, especially for site testing, so a default value ε = 0.15 is assumed for the calculation. If ε is known, a corrected value kTε can be computed as kTε =

kT ⋅0.15 ε

(5.43)

where kT is the value reported by the instrument assuming ε = 0.15 and ε is the true porosity of the concrete. The porosity of conventional concretes, including air-entrained and highperformance concretes, ranges typically between 0.05 and 0.25. Therefore, from Eq. (5.43), we can see that kT underestimates the permeability of low porosity concretes and overestimates that of high porosity. Since low porosity concretes tend to present low kT values and high porosity concretes to present high kT values (Eq. 5.31), the end result is that, by using a default value of ε = 0.15, the range of permeabilities measured is artificially expanded. The “error” factor is between 0.6 and 3; if we take into consideration (Table 5.2) that kT varies across six orders of magnitude, this error becomes of little significance and the use of the default value is recommended. However, if a more accurate calculation is required, then Eq. (5.43) should be applied, entering the true porosity ε, either measured or estimated through Eq. (5.31). In the case of measuring materials other than concrete (e.g. mortar, paste, non-cementitious materials), a correction is needed, as mortar and paste present much higher porosities than concrete. If we assume a concrete containing, by volume, 28% cement paste, 32% fine aggregate and 40% coarse aggregate, the default porosities would become ε = 0.15/0.60 = 0.25 for mortar and ε = 0.15/0.28 = 0.54 for paste (assuming that the aggregates do not contribute porosity). These values should be used in Eq. (5.43) to obtain the coefficient of air-permeability corrected for porosity.

5.8 STATISTICAL EVALUATION OF kT TEST RESULTS

5.8.1 Statistical Distribution of kT Results There is general consensus on that the kT values obtained in a set of measurements on nominally the same concrete are not normally distributed, with several cases reported where the log-normal distribution provides a suitable representation of the statistical distribution of kT results (Torrent, 2001; Brühwiler et al., 2005; Conciatori, 2005; Denarié et al., 2005; Jacobs & Hunkeler, 2006; Misák et al., 2008, 2017). Indeed, as shown in Table 5.3, for large elements cast under laboratory conditions, the CoV of the test results range between 35% and 110%. It has been shown that, for CoV above 25%, properties that cannot accept negative values are not well represented by a normal distribution, with the @seismicisolation @seismicisolation

174 Concrete Permeability and Durability Performance

Figure 5.29 Log-normal distributions fit to kT results obtained on two different concretes, data from Conciatori (2005).

log-normal distribution being more appropriate for these cases (Torrent, 1978). Figure 5.29 shows the statistical distribution of 63 kT results obtained by Conciatori (2005) on concretes of different w/c ratios: 0.52 (l.) and 0.73 (r.), with the lognormal function fitted to them. In ordinates, the probability density function f(x) is represented, so that the area under the histogram and curve is = 1, see Torrent (1978). Other examples of non-Gaussian distribution can be seen for Port of Miami Tunnel real case in Section 11.3.1. Nevertheless, it has also been reported that some sets of kT results do not follow a log-normal or any other known statistical distribution (Jacobs & Hunkeler, 2006; Neves et al., 2012). Therefore, two different approaches for the statistical evaluation of kT results are presented: one parametric and the other non-parametric. The parametric approach for the analysis assumes that kT results follow a log-normal distribution and the nonparametric approach assumes that kT results do not follow any particular statistical distribution and will be evaluated through an Exploratory Data Analysis. As for any other test method, the collected kT results shall be evaluated applying statistical analysis tools. The above-mentioned intrinsic high scatter of any gas-permeability assessment (Coutinho & Gonçalves, 1993; Andrade et al., 2000) presents some challenges regarding its statistical evaluation that will be addressed in the following, for the particular case of kT results.

5.8.2 Central Value and Scatter Statistical Parameters 5.8.2.1 Parametric Analysis A suitable parameter for the central value of a data set or a population of kT is the geometric mean, given by: @seismicisolation @seismicisolation

Torrent NDT method 175

kTgm

 =  

n kTi  (5.44)  i =1 n

∏

Equation (5.44) represents the nth root of the product of n individual kT results (kTi). The geometric mean can also be calculated taking advantage of the logarithm function properties: n

kTgm = b

∑ i =1logb (kTi ) n

(5.45)

In this case, the geometric mean is calculated as the antilogarithm of the mean value of n logarithms of individual kT results logb (kTi). It shall be noticed that Eq. (5.45) is valid for any positive b not equal to 1, being common to find b = 10 (LOG or log10) in statistical analysis of kT results. This second option to compute the geometric mean, although not so intuitive as the first, is less prone to underflow or overflow errors. As a measure of scatter, given the (usually positive) skewness of the data, there is not an obvious indicator. The antilogarithm of the standard deviation of logb (kT) cannot be considered as appropriate, in opposition to what happens with the mean for central value. Nevertheless, the standard deviation of log10 (kT), denoted as s LOG , has been used to evaluate the scatter of a kT results set (see Section 5.6.2). A fair indication of s LOG range can be found in Figure 5.9, whereas for an a priori estimation, s LOG = 0.40 is a suitable likely value for site tests (Jacobs, 2006). Actually, s LOG is the shape parameter of the log-normal distribution of a kT population represented by the corresponding set of kT results. Therefore, it can be used to evaluate the scatter of a data set. 5.8.2.2 Non-Parametric Analysis In non-parametric analysis, the median is considered a suitable statistic parameter to represent the central value. It corresponds to the 50th percentile of a set or it is the value that stands in the middle of a ranked set. For the evaluation of scatter, two statistics can be used. One for the absolute scatter and one for the relative scatter. The interquartile range is a measure of absolute scatter and is calculated as the difference between the third and the first quartiles, Q3 and Q1, respectively, i.e. the difference between the 75th percentile and the 25th percentile. Given the wide range of measurable kT values, an interquartile range of 0.1 × 10 −16 m 2 can correspond to a low scatter for a PK4 concrete (see Table 5.2), while it will identify a large scatter for a PK1 concrete. Thus, it is important to have a relative indicator of scatter, like the quartile coefficient of dispersion (QCD): @seismicisolation @seismicisolation

176 Concrete Permeability and Durability Performance

QCD =

Q3 − Q1 Q3 + Q1

(5.46)

According to our experience, QCD values of 0.23 may be expected for laboratory concrete, whereas for site concrete QCD of 0.50 may be a fair forecast.

5.8.3 I nterpretation and Presentation of Results Regardless of the intended type of analysis (parametric or non-parametric), it is recommended to start with a histogram, as exemplified in Figure5.29. Although this representation of data is very useful for the analysis, it shall be kept in mind that a considerable number (at least 30) of results is required to have a meaningful histogram. Usually, the number of kT determinations within an assessment is under 20; therefore, other tools for the analysis of the data shall be considered. Two major alternatives to the histogram are suggested: presentation of geometric mean with error bars and/or boxplot. The length of error bars is often taken as the standard deviation. Then, considering the information provided in 5.8.2.1, the lower and upper limits of error bars will be kTgm ÷ 10s log and kTgm × 10s log, respectively. Figure 5.11 is an example of this kind of plot. The boxplot is a graphic representation of several descriptive statistics: minimum, maximum, quartiles and, eventually, outliers. The concept of outlier is an observation/result that is significantly distant from the rest. According to Tukey (1977), significantly distant refers to an observation that is more than one and a half times the interquartile range away from the first or third quartile. Outliers are often found in gas-permeability assessments, as a single void or a micro-crack may increase gas-permeability over ten-fold (Neves et al., 2012). The identification of the ranges of the different Permeability Classes in the charts is also recommended. To exemplify, let us consider two different sets of results, one from a prefabricated slab in Portugal (Neves et al., 2012) and other from a prefabricated bridge segment in Switzerland (Jacobs et al., 2009), hereon identified as “Portugal” and “Switzerland”. The values are ordered from smallest to highest. “Portugal” = {0.012; 0.012; 0.014; 0.014; 0.017; 0.018; 0.020; 0.021; 0.027; 0.029; 0.040; 0.045; 0.047; 0.049; 0.061; 0.104; 0.105; 0.144; 0.158; 0.247} (10 −16 m 2) “Switzerland” = {0.057; 0.083; 0.102; 0.103; 0.107; 0.113; 0.115; 0.118; 0.122; 0.167; 0.577; 1.474; 1.507; 1.814; 2.695} (10 −16 m 2) Some descriptive statistics of these sets are presented in Table 5.5. To build the suggested chart, the computation of lower and upper limits of the error bars is still required. The lower limits will be @seismicisolation @seismicisolation

Torrent NDT method 177 Table 5.5 Descriptive statistics of example sets of kT assessment

Set “Portugal” “Switzerland” a

Median kTgm sLOG Minimum Q1 (Q2) Q3 Maximum QCD (10−16 m2) (-) (10−16 m2) (10−16 m2) (10−16 m2) (10−16 m2) (10−16 m2) (-) 0.038 0.222

0.41 0.58

0.012 0.057

0.018 0.103

0.035 0.118

0.072 1.474

0.158a 2.695

0.60 0.87

The largest value in the set (0.247) is not taken as the maximum, as it is considered an outlier: 0.247 > Q3 + 1.5 × (Q3−Q1).

0.038 ÷ 10 0.41 = 0.015 × 10 −16 m 2 and 0.222 ÷ 10 0.58 = 0.059 × 10 −16 m 2 for “Portugal” and “Switzerland”, respectively, while the upper limits will be 0.038 × 10 0.41 = 0.097 × 10 −16 m 2 and 0.222 × 10 0.58 = 0.840 × 10 −16 m 2 , for “Portugal” and “Switzerland”, respectively. Based on the previous information, a chart like the one presented in Figure 5.30 can be built. At first glance in Figure 5.30, it is possible to conclude that “Portugal” set represents a low-permeability concrete, whereas “Switzerland” set corresponds to a moderate permeability concrete, as defined in Table 5.2. A deeper analysis reveals that there is a large scatter in “Switzerland” set, with individual results in three different classes, whilst “Portugal” set has

Figure 5.30 Two representation plots of both kT sets (“Portugal” and “Switzerland”).

@seismicisolation @seismicisolation

178 Concrete Permeability and Durability Performance

individual results in two classes. Furthermore, in “Switzerland” set, more than 25% are in PK4 (high permeability). Hence, the non-parametric boxplot representation, although visually more complex, provides more information on the statistical distribution of the measured values. Beyond the specific findings for these two sets, there are also other conclusions from the presented information that can be extended to other sets. Taking the geometric mean as the representative value of a set is conservative, in comparison with the median, i.e. the geometric mean is higher than the median, unless there is a negative skewness (which is rare). The length ofthe error bars is longer than the interquartile range, as the first is limited by the 16%–84% fractiles (see below) and the second is limited by the 25–75 percentiles. If the median is equidistant from the first and third quartiles, and furthermore both extremes are also equidistant from the median – all in logarithmic scale – it is most likely that the results are log-normally distributed. Then, the geometric mean and the median assume similar values. This is the case for “Portugal” set, where the minimum, the maximum, Q1 and Q3 have an almost symmetric distribution around the median and the geometric mean is close to the median. In opposition, the “Switzerland” set exhibits the median much closer to Q 1 than to Q3, and the geometric mean is fairly away from the median. When the kT values obtained in a set of measurements follow a lognormal distribution, which happens in most cases, the parametric and non-parametric analysis will lead to the same conclusions. However, the parametric analysis comprises statistical inference, i.e. enables interpolation and/or extrapolation of results, being useful for conformity assessment, service life design and estimation of residual life purposes (see Section 9.5.2). Conformity criteria, based on statistical inference, have already been proposed by Denarié et al. (2005) and Jacobs and Hunkeler (2007). The criterion proposed by Denarié et al. (2005) is based on the allowed probability of having kT higher than a specified value (kTs) and on the reliability of the tested sample (number of measurements) and is defined by where kTgm is the geometric mean of the kT determinations (sample), s LOG is the standard deviation of the logarithms of the kT determinations, kTs is the specified/required value of kT and λ is a factor that depends on the defined fractile, the level of confidence and the sample size (n). At this point, it is important to recall that a fractile is the probability of having a result lower than a defined value. As the log-normal distribution assumes that the logarithms of kT are normally distributed, the kTp value corresponding to a given fractile p (%) is calculated as

kTp = 10log kTgm + z⋅ sLOG = kTgm × 10( z ⋅ sLOG ) @seismicisolation @seismicisolation

(5.48)

Torrent NDT method 179

where z is the argument of the standard Gauss distribution that delivers the probability p, e.g. for fractile 1−p = 95% is z = 1.65; for fractile 1−p = 10% is z = −1.28, as detailed in Eqs. (5.49) and (5.50). The criterion proposed by Jacobs and Hunkeler (2007) is similar to the criterion proposed by Denarié et al. (2005), with λ = 1, regardless of the size sample:

kTgm × 10sLOG ≤ kTs

(5.51)

Please notice that Eq. (5.51) is equivalent to

(

)

log10 kTgm × 10sLOG ≤ log10 ( kTs ) ⇔ log10 ( kTgm ) + sLOG ≤ log10 ( kTs ) ⇔ z ≥ 1 (5.52) with z=

log10 ( kTs ) − log10 ( kTgm ) sLOG

(5.53)

This corresponds to the fractile 1−p = 84%, as Φ (1) = 0.84. In practice, this means allowing a maximum probability p = 0.16 of having kT results higher than the specified value (kTs). Although non-parametric analysis does not allow statistical inference, still it can be used in conformity assessment. The specified value of kTs can be compared with a defined percentile of a kT sample, and if the latest is lower, the assessed concrete surface can be considered as conform to the specification. If the defined percentile is the 50th, then the median will be the characteristic value of kT. The median has been adopted as characteristic value in the assessment of several concrete properties, such as resistivity (Polder et al., 2000), surface hardness (CEN, 2001) and carbonation depth (McGrath, 2005). However, defining just a percentile is arguable. The sample size shall be considered and, like in parameter λ of Eq. (5.47), the percentile may be defined as function of sample size. Another option for non-parametric analysis is the inspection by attributes. For this approach there are already defined criteria, such as that adopted by Swiss Standard SIA 262/1: “Concrete Construction – Supplementary Specifications” (SIA 262/1, 2019), described in Annex B. A statistical analysis of that conformity criterion can be found in Section D-2 of Jacobs et al. (2009); warning, Figure D-5 of that document is wrong, Figure 8.20 of this book is the correct one. @seismicisolation @seismicisolation

180 Concrete Permeability and Durability Performance

5.9 TESTING PROCEDURES FOR MEASURING kT IN THE LABORATORY AND ON SITE As with all test methods, when conducting measurements of airpermeability kT, it is important to observe the same or, at least, similar testing procedures in order to get meaningful and comparable results. Regarding measuring kT, two well-differentiated cases have to be considered: laboratory testing and site testing. Different to laboratory testing, where typically plain concrete specimens are cast and preconditioned to be tested in rooms under rather stable and known ambient conditions, site testing is faced with many uncontrolled variables, typically: rough or irregular surfaces, extreme temperatures, variable humidity conditions of the concrete surface, the eventual presence of steel reinforcement too close to the surface, cracks, etc. In Annex B, procedures for measuring kT in the laboratory and on site are given, in a pre-Standard format, as a basis for elaborating laboratory protocols or even local or regional standards. Annex B refers primarily to site testing of relatively young structures, say up to 1 year of age. Sometimes, kT is used as indicator of the durability of old concrete structures, becoming a tool for their condition assessment (JCI, 2014). A good example of such application can be found in Imamoto (2012), Imamoto et al. (2012) and Neves et al. (2018). This kind of application requires a special planning. REFERENCES Adey, B., Roelfstra, G., Hajdin, R. and Brühwiler, E. (1998). “Permeability of existing concrete bridges”. 2nd International PhD Symposium on Civil Engineering, Budapest, 8 p. Andrade, C., González Gasca, C. and Torrent, R. (2000). “The suitability of the ‘TPT’ to measure the air-permeability of the covercrete”. ACI SP-192, 301–318. ASTM F2659 (2015). “Standard guide for preliminary evaluation of comparative moisture condition of concrete, gypsum cement and other floor slabs and screeds using a non-destructive electronic moisture meter”, 6 p. Bonnet, S. and Balayssac, J.P. (2018). “Combination of the Wenner resistivimeter and Torrent permeameter methods for assessing carbonation depth and saturation level of concrete”. Constr. & Building. Mater., v188, 1149–1156. Brühwiler, E., Denarié, E., Wälchli, Th., Maître, M. and et Conciatori, D. (2005). “Applicabilité de la mesure de perméabilité selon Torrent pour le contrôle de qualité du béton d’enrobage”. Office Fédéral Suisse des Routes, Rapport n. 587, Avril, Bern, Suisse, 1–48. Bueno, V., Nakarai, K., Nguyen, M.H., Torrent, R.J. and Ujike, I. (2021). “Effect of surface moisture on air-permeability kT and its correction”. Mater. & Struct., v54, 89, 12 p.

@seismicisolation @seismicisolation

Torrent NDT method 181 CEN

(2001). “Testing concrete in structures. Non-destructive testing. Determination of rebound number”. Comité Européen de Normalisation, European Standard EN 12504-2, 1–8. Conciatori, D. (2005). “Effet du microclimat sur l’initiation de la corrosion des aciers d’armature dans les ouvrages en béton armé”. Ph.D. Thesis N° 3408, EPFL, Lausanne, 264 p. Coutinho, A. de S. and Gonçalves, A. (1993). Fabrico e Propriedades do Betão. V.III, 2nd ed., LNEC, Lisboa, Portugal. Denarié, E., Conciatori, D., Maître, M. and Brühwiler, E. (2005). “Air-permeability measurements for the assessment of the in situ permeability of cover concrete”. ICCRRR, Cape Town, November 21–23. Dobel, T., Balzamo, H. and Fernández Luco, L. (2010). “Aplicaciones de la Medida de Permeabilidad al Aire del Hormigón de Recubrimiento en Situaciones de Heterogeneidad del Sustrato”. 18°. R.T AATH, Mar del Plata, Argentina, Noviembre 8–10, 8 p. Dobel, T. and Fernández Luco, L. (2012). “Coefficient of air permeability of non-hom*ogeneous substrates and drying concrete”. Microdurability 2012, Amsterdam, April 11–13, Paper 191. Eddy, L., Matsumoto, K., Nagal, K., Chaemchuen, P., Henry, M. and Horiuchi, K. (2018). “Investigation on quality of thin concrete cover using mercury intrusion porosimetry and non-destructive tests”. J. Asian Concr. Federation, v4, n1, 47–66. Hirano, M., Yosh*take, I., Hiraoka, A. and Inagawa, Y. (2014). “Evaluation of airbubbles distributed on concrete surface of side wall of tunnel lining”. Cem. Sci & Concr. Technol., v67, n1, 252–258. In Japanese. Imamoto, K. (2012). “Non-destructive assessment of concrete durability of the National Museum of Western Art in Japan”. Microdurability 2012, Amsterdam, April 11–13, 22 slides. Imamoto, K., Tanaka, A. and Kanematsu, M. (2012). “Non-destructive assessment of concrete durability of the National Museum of Western Art in Japan”. Paper 180, Microdurability 2012, Amsterdam, April 11–13. Jacobs, F. (2006). “Luftpermeabilität als Kenngrösse für die Qualität des Überdeckungsbetons von Betonbauwerken”. Office Fédéral des Routes, VSS Report 604, Bern, Switzerland, 85 p. Jacobs, F., Denarié, E., Leemann, A. and Teruzzi, T. (2009). “Empfehlungen zur Qualitätskontrolle von Beton mit Luftpermeabilitätsmessungen”. Office Fédéral des Routes, VSS Report 641, December, Bern, Switzerland, 53 p. Jacobs, F. and Hunkeler, F. (2006). “Non destructive testing of the concrete cover – Evaluation of permeability test data”. International RILEM Workshop on Performance Based Evaluation and Indicators for Durability, Madrid, Spain, March 19–21, 207–214. Jacobs, F. and Hunkeler, F. (2007). “Air-permeability as a characteristic parameter for the quality of cover concrete”. Concrete Platform, 173–182. JCI (2014). “Guidance for Assessment of existing concrete structures”. Japan Concrete Institute, 49. Kato, Y. (2013). “Characteristics of the surface air-permeability test and the evaluation of quality variation in cover concrete due to segregation of concrete”. J.Adv. Concr. Technol., v11, 322–332.

@seismicisolation @seismicisolation

182 Concrete Permeability and Durability Performance Kreijger, P.C. (1984). “The skin of concrete: Composition and properties”. Mater. & Struct., v17, n100, 275–283. Li, K., Zhang, D., Li, Q. and Fan, Z. (2019). “Durability for concrete structures in marine environments of HZM project: Design, assessment and beyond”. Cem. & Concr. Res., v115, 545–558. M-A-S (2019). PermeaTORR AC (Active Cell). http://m-a-s.com.ar/eng/product. php. Maeda, T., Honma, H., Hirano, M. and Yosh*take, I. (2014). “Permeability of tunnel lining with air/water bubbles on concrete surface”. Sustainable Solutions in Structural Engineering and Construction, ISEC Press, 321–325. Mayer, A. (1987). “The importance of the surface layer for the durability of concrete structures”. ACI SP-100, v1, 49–61. McGrath, P.F. (2005). “A simple chamber for accelerated carbonation testing of concrete”. ConMat, Vancouver, University of British Columbia, Canada. Misák, P. (2018). Private e-mail communication, 09.01.2018. Misák, P., Kucharczyová, B. and Vymazal, T. (2008). “Evaluation of permeability of concrete by using instrument Torrent” (in Czech), JUNIORSTAV 2008, 2.5 Stavebni zkusebnictví, Brno, January 23, 3 p. Misák, P., Kucharczyková, B., Vymazal, T., Daněk, P. and Schmid, P. (2010). “Determination of the quality of the surface layer of concrete using the TPT method and specification of the impact of humidity on the value of the airpermeability coefficient”. Ceramics – Silikáty, v54, n3, 290–294. Misák, P., Stavař, T., Rozsypalová, I., Kocáb, D. and Põssl, P. (2017). “Statistical view of evaluating concrete-surface-layer permeability tests in connection with changes in concrete formula”. Materiali in tehnologije/Mater. Technol., v51, n3, 379–385. Neves, R., Branco, F. and de Brito, J. (2012). “About the statistical interpretation of air-permeability assessment results”. Mater. Struct., v45, n4, 529–539. Neves, R., Branco F. and de Brito, J. (2015). “Study on the influence of surface and geometric factors on the results of a nondestructive onsite method to assess air-permeability”. Exp. Technique., v40, n3, August, 1–8. Neves, R., Torrent, R. and Imamoto, K. (2018). “Residual service life of carbonated structures based on site non-destructive tests”. Cem. Concr. Res., v109, 10–18. Newman, K. (1987). “Labcrete, realcrete, and hypocrete. Where we can expect the next major durability problems”. ACI SP-100, v2, 1259–1283. Nguyen, M.H., Nakarai, K., Kai, Y. and Nishio, S. (2020). “Early evaluation of cover concrete quality utilizing water intentional spray tests”. Constr. & Build. Mater., v231: 117144. Nguyen, M.H., Nakarai, K. and Nishio, S. (2019). “Durability index for quality classification of cover concrete based on water intentional spraying tests”. Cem. & Concr. Composites, v104: 103355. Nsama, W., Kawaai, K. and Ujike, I. (2018). “Influence of bleeding on modification of pore structure and carbonation-induced corrosion formation”. SLD4, Delft, Netherlands, 674–685. Parrott, L. (1994). “Design for avoiding damage due to carbonation-induced corrosion”. ACI SP-145, 283–298. Paulini, P. (2014). “Empfehlungen zur Bestimmung der Gaspermeabilität von Beton”. Contribution to RILEM TC 230-PSC, June, 7 p.

@seismicisolation @seismicisolation

Torrent NDT method 183 Polder, R., Andrade, C., Elsener, B., Vennesland, Ø., Gulikers, J., Weidert, R. and Raupach, M. (2000). “Test methods for on site measurement of resistivity of concrete”. Mater. Struct., v33, n10, 603–611. Proceq (2019). “Torrent Permeability Tester”. https://www.proceq.com/uploads/ tx_proceqproductcms/import_data/files/Concrete%20Testing%20Products_ Sales%20Flyer_English_high.pdf. Proceq (not dated). “Operating instructions – Permeability tester TORRENT”. Proceq S.A., 11 p. RILEM TC 154-EMC (2000). “Electrochemical techniques for measuring metallic corrosion”. Mater. Struct., v33, December, 603–611. RILEM TC 230-PSC (2015). “Performance-based specifications and control of concrete durability”. Beushausen, H. and Fernández Luco, L. (Eds.), RILEM Report V18, 373 p. Romer, M. (2005a). “Effect of moisture and concrete composition on the Torrent permeability measurement”. Mater. Struct., v38, July, 541–547. Romer, M. (2005b). Personal communication and supply of test data. Romer, M. (2005c). “Multiscale durability aspects of concrete structures exposed to ground water”. COE Workshop on Material Science in 21st Century for the Constr. Ind., Hokkaido Univ., Sapporo, Japan, August 11, 24 slides. Sakai, Y. (2019). “Correlations between air-permeability coefficients and pore structure indicators of cementitious materials”. Constr. Build. Mater., v209, 541–547. Sakai, Y., Nakamura, C. and Kishi, T. (2013). “Correlation between permeability of concrete and threshold pore size obtained with epoxy-coated sample”. J. Adv. Concr. Technol., v11, August, 189–195. Sakai, Y., Nakamura, C. and Kishi, T. (2014). “Evaluation of mass transfer resistance of concrete based on representative pore size of permeation resistance”. Constr. Building Mater., v51, 40–46. Sandra, N., Kawaai, K., Ujike, I., Nakai, I. and Nsama, W. (2019). “Effects of bleeding on corrosion of horizontal steel bars in reinforced concrete column specimen”. Mater. Sci. Eng., v602: 012058. SIA 162/1 (1989). Test No. 7 ‘Porosity’. EMPA Guidelines for Testing, Dübendorf. SIA 262/1 (2019). Swiss Standard SIA 262/1:2019, “Construction en béton. Spécifications complémentaires”. Norme Suisse, March 1, 60 p. (in French and German). SIA 262/1-E (2019). Swiss Standard SIA 262/1:2019, “Construction en béton. Spécifications complémentaires”. Annex E: ‘Perméabilité à l’air dans les structures. SIA 262/1-K (2019). “Concrete construction – Complementary specifications”. Swiss Society of Engineers and Architects. Annex K: ‘Characteristics of the pores’. Sofi, M., Oktavianus, Y., Lumantarna, E., Rajabifard, A., Colin Duffield, C. and Mendis, P. (2019). “Condition assessment of concrete by hybrid non-destructive tests”. J. Civil Struct. Health Monitoring, v9, 339–351. Szychowski, J. (2010). “PermeaTORR: Relación entre profundidad de penetración y parámetros constructivos”. Informe de DISTEK S.R.L., Buenos Aires, Argentina. Torrent, R.J. (1978). “The log-normal distribution: a better fitness for the results of mechanical testing of materials”. Mater. Struct., v11, n64, July–August, 235–245.

@seismicisolation @seismicisolation

184 Concrete Permeability and Durability Performance Torrent, R. (1991). “Un nuevo método no destructivo para medir la permeabilidad al aire del recubrimiento de hormigón”. 10a. Reunión Técnica de la AATH, Olavarría, Argentina, Octubre, vI, 307–323. Torrent, R.J. (1992). “A two-chamber vacuum cell for measuring the coefficient of permeability to air of the concrete cover on site”. Mater. Struct., v25, n150, July, 358–365. Torrent, R. (1997). “Your enquiry about the Permeability Tester”. Telefax to Carola Edwardsen (COWIConsult), Holderbank, Switzerland, February 4, 7 p. Torrent, R. (2001). “Diseño por Durabilidad - Técnicas de Ensayo y su Aplicación”. CENCO Seminar on Durability of Concrete and Evaluation of Corroded Structures, Instituto Eduardo Torroja, Madrid, April 17–19. Torrent, R. (2005). “Update of comparative test – Part I – Comparative test of penetrability methods”. Mater. Struct., v38, December, 895–906. Torrent, R. (2012). “Non-destructive air-permeability measurement: From gasflow modelling to improved testing”. Paper 151, Microdurability 2012, Amsterdam, April 11–13. Torrent, R., Bueno, V., Moro, F. and Jornet, A. (2019). “Suitability of impedance surface moisture meter to complement air-permeability tests”. RILEM PRO 128, Durability, Monitoring and Repair of Structures, March, 56–63. Torrent, R., di Prisco, M., Bueno, V. and Sibaud, F. (2018). “Site air-permeability of HPSFR and conventional concretes”. ACI SP-326, Paper 84. Torrent, R. and Ebensperger, L. (1993). “Methoden zur Messung und Beurteilung der Kennwerte des Überdeckungsbetons auf der Baustelle”. Office Fédéral des Routes, Rapport No. 506, Bern, Switzerland, Januar, 119 p. Torrent, R. and Frenzer, G. (1995). “Methoden zur Messung und Beurteilung der Kennwerte des Ueberdeckungsbetons auf der Baustelle -Teil II”. Office Fédéral des Routes, Rapport No. 516, Bern, Suisse, October, 106 p. Torrent, R. and Jornet, A. (1990). “Covercrete study – Part I: Scope of the research, test methods, characteristics of raw materials and concrete mixes, test results”. HMC Report MA 90/3815/E, September, 48 p. Torrent, R., Moro, F. and Jornet, A. (2014). “Coping with the effect of moisture on air-permeability measurements”. International Workshop on Performancebased Specification and Control of Concrete Durability, Zagreb, Croatia, June 11–13, 489–498. Torrent, R.J. and Szychowski, J. (2016). “Medición no destructiva de la permeabilidad al aire: evolución e innovación”. Revista Hormigón n54, AATH, Buenos Aires, Argentina, Ene-Jun. Torrent, R. and Szychowski, J. (2017). “Innovation in air-permeability NDT: Concept and performance”. XIV DBMC, Ghent, May 29–31, Proceedings, paper 313. Tukey, J.W. (1977). Exploratory Data Analysis. Addison-Wesley Publishing Company, Boston, MA. Urdan, T.C. (2011). Statistics in Plain English. Routledge, Taylor & Francis Group, New York. Yokoyama, Y., Sakai, Y., Nakarai, K. and Kishi, T. (2017). “Change in surface airpermeability of concrete with different mix designs and curing”. Cem. Sci. & Concr. Technol., v71, n1, 410–417 (in Japanese).

@seismicisolation @seismicisolation

Chapter 6

Effect of key technological factors on concrete permeability

6.1 INTRODUCTION As discussed in Chapter 3, the permeability of concrete depends on the microstructure of the material, with the flow of matter taking place predominantly through the pores in the cement paste and in the ITZ, but also along eventual cracks. Therefore, all technological factors that affect the volume, size, tortuosity and connectivity of pores and voids will exert an influence on the permeability of concrete. The steady flow of liquids under pressure through a concrete sample or element requires that the pores are saturated with the fluid; otherwise, part of the inflow of liquid will end up filling the empty voids. On the contrary, the gas flow under pressure requires that the pores are sufficiently dry so as to leave empty paths for the gas molecules to be transported without hindrance through the material. Hence, it is expected that the degree of saturation or moisture content of the concrete exerts an important influence on the permeability of the material to liquids and gases. This chapter is devoted to present and discuss experimental data, obtained both in the laboratory and on site, on the influence of key technological parameters of concrete (and of its temperature and moisture) on the permeability of the material to liquids and gases, measured using some of the test methods described in Chapters 4 and 5. This topic was dealt with in depth in Chapter 4 of “Performance criteria for concrete durability” (RILEM TC 116-PCD, 1995) to which the reader is referred. Here, new data on permeability of concrete to water and gases are contributed, obtained from tests covered by standards, namely: Wp: Water penetration under pressure (EN 12390-8, see Section 4.1.1.2) a, A: Water sorptivity (SIA 262/1-A, ASTM C1585, see Section 4.2.1) kO, Kg, K int: Cembureau gas-permeability (UNE 83981, LNEC E 392, see Section 4.3.1.2) OPI, kOPI: Oxygen-Permeability Index (SANS-3001-CO3-2, see Section 4.3.1.3) kT: Torrent air-permeability (SIA 262/1, IRAM 1892, see Chapter 5) When other methods are applied, they are briefly described in the text. DOI: 10.1201/9780429505652-6 @seismicisolation @seismicisolation

185

186 Concrete Permeability and Durability Performance

It is expected that this chapter will be useful to the reader in designing concrete mixes, not just for strength but for permeability as well, considering the effect of the key technological factors involved. It will also help contractors and inspectors to identify key aspects of jobsite concrete practices affecting its permeability. 6.2 EFFECT OF W/C RATIO AND COMPRESSIVE STRENGTH ON CONCRETE PERMEABILITY The water/cement ratio (w/c) is a key factor in the design of concrete mixes because, as explained in Section 3.2.2, it determines to a large extent the volume and size of the capillary pores in the hydrated cement paste and, hence, the transport properties of concrete. The relation between compressive strength and w/c ratio is at the root of most mix design methods, because compressive strength is strongly affected by the pore structure. Therefore, it is to be expected that significant correlations of the permeability of concrete to gases and water with w/c ratio and compressive strength exist. The following sections confirm that expectation, based on abundant experimental evidence.

6.2.1 Data Sources The data presented and discussed in Section 6.2 were collected from comprehensive investigations made (some by the authors of this book) predominantly in Switzerland, Japan and South Africa. They are summarized in the following sections. 6.2.1.1 HMC Laboratories During several years starting 1990, “Holderbank” Management & Consulting Ltd 2 (HMC) conducted, in its laboratories located in Holderbank (Canton Aargau, Switzerland), several research projects, some with own funding, some sponsored by the Swiss Federal Highways Administration (ASTRA). Concretes of different compositions (binder types and proportions) were prepared, to a large extent with the same aggregates used as reference in HMC Laboratory, and tested following strictly the same procedure. Regarding permeability tests, 250 × 360 × 120 mm slabs were cast in two layers, each layer compacted on a vibrating table, undergoing three curing regimes: A (dry): immediately after casting, the slabs were kept (stripped at 24 hours) in a dry room at 20°C/50% RH until the age of 28 days. 2

Later branded as Holcim Technology Ltd.

@seismicisolation @seismicisolation

Effect of key factors on permeability 187

B (reference): immediately after casting, the slabs were kept (and stripped) in a moist room (20°C/>95% RH) for 7 days and later moved to a dry room at 20°C/50% RH until the age of 28 days. C (standard): immediately after casting, the slabs were kept (and stripped) in a moist room (20°C/>95% RH) until the age of 28 days. In all cases, 120 × 120 × 360 companion prisms were cast, cured after procedure C, and tested at 28 days, first in three-point bending and the resulting halves in compression as equivalent 120 mm cubes. Several non- or slightly destructive test methods were applied on the slabs, after 21 days storage in the dry room, in particular the here reported airpermeability kT. Immediately afterwards, Ø150 × 120 mm cores were drilled from the slabs and cut to 50 mm depth, from both the top and bottom (as cast) surfaces. The resulting Ø150 × 50 mm discs were oven-dried at 50°C during 6 days, followed by 1-day cooling in a desiccator inside the testing room at 20°C. Then, the discs were placed in the testing cell and the coefficient of O2-permeability kO was measured, averaging the results obtained at relative gas pressures of 0.1 and 2.5 MPa (after waiting 30 minutes to achieve steady-state conditions). After the kO test was completed, the discs were weighed and placed (without any further treatment) with the external surface in contact with 3 mm of water in a closed container. At intervals, the mass gain was measured and the value after 24 hours of contact with water was recorded and the test result a24 (g/m²/s½) reported as the mass gain, divided by the surface area and the square root of time (24 hours). The reference condition for performance comparison, based on kT, kO and a24, was the bottom (as cast) surfaces of the samples subjected to curing regimen B. The results were reported in Torrent and Jornet (1990, 1991), Torrent and Gebauer (1992a, b), Torrent and Ebensperger (1993) and Torrent and Frenzer (1995). 6.2.1.2 ETHZ Cubes This was an exercise, conducted within a project financed by ASTRA, aimed at checking the suitability of several test methods to discriminate the permeability of concretes, prepared by the Swiss Federal Institute of Technology in Zürich (ETHZ) and blind tested at ETHZ laboratory by HMC personnel (Torrent & Frenzer, 1995). The permeability tests discussed here are kT, kO and a24; the geometric mean of kT and the arithmetic means of kO and a24 were reported in Table 5.3. Details of the composition of the cubes can be found in Section 5.6.2 and in Torrent and Frenzer (1995). Suffice it to say that, from each of the two opposite faces where five measurements of kT were conducted, two Ø150 × 50 mm discs were drilled and cut, dried and tested for kO and a24, exactly in the same way as described previously in Section 6.2.1.1. @seismicisolation @seismicisolation

188 Concrete Permeability and Durability Performance

6.2.1.3 General Building Research Corporation of Japan In a laboratory research, a series of 16 OPC concrete mixes was prepared with the same constituents, varying just the proportions to achieve w/c ratios within the range 0.30–1.00 (cement contents between 200 and 580 kg/m³), resulting in cylinder compressive strengths within the range 9–71 MPa. Data of the mixes can be found in Table 2 of Imamoto et al. (2009). With each mix (except No. 9), Ø150 × 50 mm discs were cast and moist cured during 1 month, followed by 1-month storage in a dry room at 20°C/60% RH, moment at which they were tested for air-permeability kT and O2-permeability kO. 6.2.1.4 University of Cape Town In Starck (2013) and Starck et al. (2017), a comprehensive investigation is reported in which a series of 150 mm concrete cubes, made with two binders and three different w/b ratios, were cured under conditions representing winter and summer conditions in Cape Town, South Africa. The binders were an OPC and a 50% + 50% blend of OPC and slag (GBFS) and the w/b were 0.50, 0.65 and 0.80. The (favourable) winter curing was simulated by exposing the cubes protected outdoors and soaked daily during 35 days. Then, the cubes were stored for 5 days in a dry room (20°C/53% RH). The summer conditions were simulated by keeping the cubes up to 35 and 90 days in the dry room immediately after stripping. After measuring kT, Ø70 × 30 mm discs were drilled and saw-cut from the cubes, oven dried at 50°C for 7 days and tested for O2-permeability OPI. 6.2.1.5 KEMA A comprehensive research (Van Eijk, 2009) was conducted in the Netherlands to check the suitability of two permeability test methods2 to assess the quality of concrete surfaces, namely: maximum water penetration under pressure WPmax and air-permeability kT. For that purpose, four (1.2 × 1.2 × 0.3 m) walls were cast in the city of Utrecht on 2 March 2009, two with OPC concrete (w/c = 0.40 and 0.57) and the other two with same w/c ratios but using GBFS cement as binder. The form of the N side of each wall was kept 7 days in place, whilst that of the S side was kept just 1 day in place, giving eight walls’ surface qualities; later the walls were exposed to the natural outdoors’ environment until the age of 28 days. At that age, kT was measured non-destructively at three positions on each of the eight surfaces, followed by Ø150 × 150 mm cores drilling from the same spots. The maximum water penetration WP through the exposed surface of the cores was measured. 2

The Wenner electrical resistivity was also measured but the results were not fully reported

@seismicisolation @seismicisolation

Effect of key factors on permeability 189

6.2.1.6 Other Results from other sources are described in the corresponding sections.

6.2.2 Effect of w/c Ratio and Strength on Gas-Permeability 6.2.2.1 C embureau Test Method Figure 6.1 has been built with gas-permeability test results obtained in the investigations described in Sections 6.2.1.1 (legend Torrent et al.), 6.2.1.3 (Imamoto et al.) and 6.1.2.2 (ETHZ cubes), plotted against the w/c ratio of the OPC concretes tested. The permeability tests reported correspond to specimens moist cured for at least 7 days. Figure 6.2 presents the same gas-permeability data but plotted against the standard compressive strength measured on cubes, moist cured during 28 days (the strength data from Imamoto et al., measured on cylinders fc cyl, were converted into cube strengths fc cube applying conversion Eq. (6.1), that gives a very good approximation to the relation shown in Tables 5.1–5.3 of Model Code 2010 (fib, 2012)):

Despite the different experimental conditions, the agreement between the three sets of data is remarkable. The line shown in Figure 6.1 corresponds to the prediction of gas-permeability as function of the w/c ratio, by the formula proposed in Eq. (2.1– 107) of CEB-FIP (1991), valid in the range 0.4 < w/c < 0.7, extrapolated as dotted line:

Figure 6.1 Effect of w/c ratio on Cembureau gas-permeability of OPC concretes.

@seismicisolation @seismicisolation

190 Concrete Permeability and Durability Performance

Figure 6.2 Effect of compressive strength on Cembureau gas-permeability of OPC concretes.

where Kg = gas-permeability (m²) Equation (6.2) seems to reflect quite well the trend (slope) of log10 Kg vs w/c but underestimates the gas-permeability values measured by the Cembureau test method, under the applied test conditions. The full line in Figure 6.2 corresponds to the prediction of gas-permeability as function of the compressive strength, by the formula proposed in Eq. (5.1–123) of Model Code 2010 (fib, 2012) relating Kg (m²) and the mean cylinder compressive strength fcm (MPa):

4.5 K g = 2 × 10−10 fcm

(6.3)

The full line curve in Figure 6.2 was plotted, converting the cylinder strength of Eq. (6.3) into cube compressive strength by means of Eq. (6.1). It can be seen in Figure 6.2 that fib Eq. (6.3) grossly underestimates the Cembureau gas-permeability test results. The dotted line represents the regression line fitted to the 34 test results labelled “Torrent et al.”: where fc is the cube compressive strength (MPa) and kO is the Cembureau coefficient of O2-permeability (10 −16 m²). The results of the other two sources fit quite well to the regression of Eq. (6.4). For the sake of completeness, the relation by Eq. (5.1–122) of Model Code 2010 (fib, 2012) between water-permeability coefficient Kw (m/s) and the mean cylinder compressive strength fcm (MPa) is @seismicisolation @seismicisolation

Effect of key factors on permeability 191

Kw =

4 × 10−3 6 fcm

(6.5)

6.2.2.2 OPI Test Method Figure 6.3 presents data of the coefficient of O2-permeability (KOPI), coming from different sources, two from South Africa (Starck, 2013; Gopinath, 2020) and one from Switzerland (Romer & Leemann, 2005; Romer, 2005); in all cases the specimens tested were cubes. Some of the mixes tested were prepared with OPC (circles), some with 50% GGBS (triangles) and others with 30% PFA (diamonds) in the binder. The data from Gopinath (2020), black symbols, were obtained on cubes after 28 days of moist curing; those from Starck (2013), white symbols, are average values of KOPI obtained after 35 days of intermittent soaking, see Section 6.2.1.4. The data from Romer and Leemann (2005), grey symbols, correspond to cubes tested after 1 year storage at 20°C/90% RH. In all cases, cores were drilled to perform the OPI test, as described in Section 4.3.1.3. Figure 6.3a shows the KOPI results as function of the w/c ratio of the mixes and Figure 6.3b as function of the cube compressive strength. The plots KOPI vs. w/c obtained on OPC mixes by Starck (2013), white circles, and (Romer & Leemann, 2005), grey circles merge quite nicely (Figure6.3a). The KOPI results of Gopinath (2020), black symbols, are lower, for the same w/c (for both OPC and GGBS binders), than those reported by the other two sources. It is interesting to observe that, for Gopinath (2020), GGBS concretes perform better than OPC concretes, whilst the opposite happens with (Starck, 2013) results, a fact attributable to the different curing conditions applied to the cubes. Figure 6.3b shows that the relation KOPI vs. cube strength is less dependent on the curing regime (and age) and also on the cement type, possibly

Figure 6.3 Effect of (a) w/c ratio and (b) cube strength on KOPI; data from Romer (2005), Romer and Leemann (2005)¸ Starck (2013) and Gopinath (2020).

@seismicisolation @seismicisolation

192 Concrete Permeability and Durability Performance

because both properties are affected in a similar sense by those factors. Still, for the same strength, the OPC mixes tend to show higher KOPI values. Worth noticing is the nice continuity between the results obtained in both South African investigations with GGBS binder (white and black triangles). A regression line was fitted to all test results, shown and plotted as a continuous curve in Figure 6.3b, with a correlation coefficient R = 0.83. 6.2.2.3 Torrent kT Test Method To investigate the effect of the w/c ratio on kT, it is important to count with data in which just the w/c ratio is varied, whilst all the other variables are kept constant (concrete constituents, surface tested, curing and storing conditions, age, etc.). An ideal set of results is, then, that obtained in the blind test of 0.5 m ETHZ cubes (Torrent & Frenzer, 1995), presented in Table 5.3 and described in Section 6.2.1.2, where four OPC concrete mixes with different w/c ratios were investigated. The geometric mean of the ten individual kT tests conducted on each cube is plotted in Figure 6.4a, as function of the w/c ratio of each of the four mixes investigated, differentiating the cubes that were moist cured for 7 days from those without moist curing. A linear relationship between the logarithm of kT and the w/c ratio is observed, in very good agreement with Eq. (6.2), taken from Model Code 1990 (CEB-FIP, 1991). In Figure 6.4b, results reported by Torrent and Ebensperger (1993) are plotted (7 d curing). These data correspond to laboratory concrete mixes made with another OPC, changing the w/c ratio and the aggregate type (three types from N, W and S Switzerland, same 32 mm Fuller grading); all mixes but one contained entrained air. Results of two mixes, where 8% silica fume (SF) was added to the OPC, are also shown (no air-entrainment); SF was added to OPC content to compute w/c. The tests were conducted on the bottom side of 250 × 360 × 120 mm concrete slabs, moist cured during 7 days and later kept 21 days in a dry room (20°C/50% RH).

Figure 6.4 (a) Effect of w/c and curing on kT (OPC); (b) effect of w/c, aggregate and SF on kT.

@seismicisolation @seismicisolation

Effect of key factors on permeability 193

The results in Figure 6.4b, enclosed by an oval, correspond basically to the same mix, but made with the three different aggregates; the mix pointed by an arrow corresponds to the OPC mix without air-entrainment. Figure 6.4b shows that, whilst the same trend found in Figure 6.4a still holds valid, the changes in binder, aggregates and air content have increased the scatter of the results. Figure 6.5 presents the relation between kT and w/c ratio, compiled from several sources by Jacobs et al. (2009), including laboratory and site tests (Torrent & Ebensperger, 1993; Romer & Leemann, 2005; Conciatori, 2005; Jacobs, 2006; RILEM TC 189 NEC, 2007). The white circles correspond to site tests conducted on many Swiss structures by Jacobs (2006) and the line corresponds to Eq. (6.2). Although the expected general trend of higher kT for higher w/c can still be observed in Figure 6.5 (in harmony with that predicted by Eq. (6.2), a large scatter is evident. This scatter is attributable to the different raw materials used for the preparation of the concretes, but also to the different exposure and experimental conditions under which the measurements were performed, especially those conducted on site. The effect of the binder type on kT is discussed separately in Section 6.3. Figure 6.5 is a confirmation of the unsuitability of w/c ratio as durability indicator, as discussed in detail in Section 1.6.2. Figure 6.6 shows the same kT data of Figure 6.4, but here against the cylinder compressive strength at 28 days instead of the w/c ratio (the cube strength values from Torrent and Ebensperger (1993) were converted into cylinders using Eq. (6.1).

Figure 6.5 Relation between kT and w/c ratio; data from Jacobs et al. (2009).

@seismicisolation @seismicisolation

194 Concrete Permeability and Durability Performance

Figure 6.6 (a) Effect of compressive strength and curing on kT; (b) effect of strength, aggregate and SF on kT.

The full line in Figure 6.6 represents Eq. (6.3), from Model Code 2010, that fits quite well the test results. A special investigation worth citing is that reported in Mohr et al. (2000), in which the relation between compressive strength and different “penetrability” tests was studied, based on cores drilled from 15 pavements, aged between 11 and 51 years, across nine States of USA. The “penetrability” tests were: Rapid Chloride Permeability Test (RCPT) (see Section A.2.1.1), Field Permeability Test (Section 4.1.2.3) and air-permeability kT. In the case of kT and RCPT, the measurements were conducted on saw-cut discs of Ø150 and Ø100, respectively, corresponding to top, middle and bottom parts of the cores. The compressive strength was measured on Ø150 × 300 mm saw-cut cores (corrected by slenderness). Figure 6.7 shows the relation between kT and compressive strength. Two aspects are evident from Figure 6.7: one of them is that the air-permeability

Figure 6.7 Relation kT vs compressive strength of cores drilled from several pavements across USA; data from Mohr et al. (2000).

@seismicisolation @seismicisolation

Effect of key factors on permeability 195

of the cores decreases with depth, in particular showing high kT values for the top discs, which reflects the deleterious effect of weathering and traffic loads on the microstructure (an effect observed also for the RCPT). The second aspect is that Eq. (6.3) seems to reflect reasonably well the relation between kT (measured on the middle and bottom discs) and compressive strength of aged concrete.

6.2.3 Effect of w/c Ratio on Water-Permeability 6.2.3.1 Water Penetration under Pressure An example of the effect of w/c ratio on water penetration of concrete under pressure Wp was found in the research reported by Sezer and Gülderen (2015), in which the replacement of a reference limestone aggregate (coarse and fine) by steel slag was investigated in concretes of w/c = 0.40, 0.55 and 0.70. In one set of mixes, just the coarse limestone was replaced by steel slag, and in another set, just the fine limestone was replaced. An attempt to replace both aggregate fractions by slag ended in unworkable mixes that could not be tested. The tests applied to the OPC concretes were: cube compressive strength, Wp, and freeze-thaw resistance (not discussed here). Figure 6.8 presents the effect of the w/c ratio on the water penetration (not disclosed whether mean or maximum) of concretes made with limestone and with replacement of coarse and fine fractions by steel slag. It can be seen that the penetration of water increases with the w/c ratio for all sets of mixes. Coarse steel slag shows no negative influence, but fine steel slag increased significantly the water penetration and reduced the strength, a

Figure 6.8 Effect of w/c and aggregate type on water penetration Wp; data from Sezer and Gülderen (2015).

@seismicisolation @seismicisolation

196 Concrete Permeability and Durability Performance

Figure 6.9 Effect of w/c, cement and curing on (a) water penetration WPmax and (b) airpermeability kT of walls; data from Van Eijk (2009).

fact attributed by Sezer and Gülderen (2015) to workability issues caused by the higher density, angularity, roughness and fineness of the slag fine aggregate compared to the limestone sand. Figure 6.9a shows the water penetration test results for the eight wall surfaces tested in The Netherlands (Van Eijk, 2009), in the research described in Section 6.2.1.5. We can see that, in 3 out of the 4 cases, the water penetration increases with the w/c ratio, as expected. Yet, for the S side of OPC wall (OPC-1d), the opposite happened. It is difficult to justify this abnormal behaviour, which could be due to some batching mistake or by accidental extra curing of the N side of the OPC wall made with w/c = 0.57, without ruling out a possible experimental error. The results of the coefficient of airpermeability kT measured at the same spots, shown in Figure 6.9b, display the same pattern of results, which confirms the WPmax values presented in Figure 6.9a. As discussed in Section 8.3.3, both tests, WPmax and kT, showed an excellent correlation with each other. 6.2.3.2 Water Sorptivity Figure 6.10a and b present data of water sorptivity at 24 hours a24 of OPC concretes, as function of the w/c ratio and cube compressive strength, respectively. The circles in Figure 6.10 represent 20 values obtained on concretes prepared and tested at HMC Laboratory (see Section 6.2.1.1). In addition, the values obtained on the ETHZ cubes (Section 6.2.1.2) are plotted as triangles. All a 24 values plotted correspond to specimens under 7 days moist curing. Figure 6.10 shows that a24 relates linearly with both w/c and compressive strength.

@seismicisolation @seismicisolation

Effect of key factors on permeability 197

Figure 6.10 Effect of (a) w/c and (b) cube strength on water sorptivity at 24 hours.

6.3 E FFECT OF BINDER ON CONCRETE PERMEABILITY

6.3.1 Effect of OPC Strength on Permeability An experimental research was conducted at HMC laboratories in Holderbank, with the aim of assessing the effect of OPC characteristics (chiefly mortar strength) on the permeability of concretes of same strength (instead of the same w/c ratio) (Torrent & Jornet, 1991). The lack of durability of concrete structures built since the 1970s was attributed to the improvement in strength of modern cements (demanded by the market), which allowed reaching the required concrete strength with higher w/c ratios (and less cement and associated costs) but, predictably, leading to more “penetrable” microstructures. In addition, modern cements, more finely ground, would develop less hydration after 28 days, with less “safety” reserve. This line of thinking is superbly presented in Section 3.1 of Neville (2003). For that purpose, three industrial OPCs were chosen (GA, U2 and U5) that were representative of “modern” cements and one (RO) that had composition and fineness typical of “old” cements. The composition and main characteristics of the four OPCs selected are presented in Table 6.1. Cement U5 was made with similar constituents than U2, but ground in an open circuit grinding system (hence its lower RR Slope n). It can be seen that cement RO, purposely manufactured to simulate an “old” cement, presents a relatively low C3S content (“belite cement”) and has been relatively coarsely ground (see low Blaine and high d' values) and presents a relatively low early strength. Its high water-demand on paste, accompanied by low mortar flow, is to be noticed. On the other extreme is GA cement, with high C3S content (“alite cement”), finely ground and showing the highest strength at all ages.

@seismicisolation @seismicisolation

198 Concrete Permeability and Durability Performance Table 6.1 Composition and characteristics of cements used in HMC investigation Cement

Composition C3S (%) C2S (%) C3A (%) C4AF (%) Alkalis (% Na2Oequiv)

38 36 11 5 0.96

46 28 9 8 0.97

48 25 8 8 0.93

57 18 6 11 0.52

Physical properties Density (g/cm³) Blaine (m²/kg) RR slope n RR diameter d′ (µm) Water demand on paste (%) Flow of ISO mortar (%)

3.09 247 0.930 31.4 30.0 100

3.16 290 0.963 23.2 26.2 140

3.14 306 0.875 28.1 24.5 140

3.12 335 0.945 18.4 25.7 141

Mechanical properties: compressive strength ISO mortar (MPa) 2 days 13.3 19.7 20.1 7 days 29.8 33.0 31.2 28 days 44.6 43.8 41.5 90 days 51.6 59.5 53.1 365 days 53.0 67.4 62.0

23.0 41.6 58.0 68.0 69.8

Note: n and d' are the “uniformity” factor (higher n means more uniform the particles’ size) and “characteristic size (36.8% residue)” of the Rosin-Rammler particle size distribution, often used in the cement industry.

U2 is an average cement produced in a closed-circuit grinding system with high-efficiency separator (hence its higher RR Slope n). To simplify, cement U5 will not be considered in the following discussion. Air-entrained (4.7%–6.0% air) concrete mixes were prepared without water reducers, aimed at reaching compressive strengths of 25, 35 and 45 MPa, using the RO, U2 and GA OPCs described in Table 6.1. On each mix, the coefficient of oxygen-permeability kO was measured on cores taken from the bottom (as cast) surfaces of samples moist-cured for 7 days. In parallel, the equivalent cube compressive strength was measured on specimens moist cured for 28 days. The testing procedure was that described in Section 6.2.1.1. Figure 6.11a shows the relation cube strength vs. w/c ratio for the three OPCs. As expected, the cement with higher mortar strength (GA) shows the best strengths for the same w/c ratio, whilst the coarsely ground cement (RO) showed the worse performance (despite its mortar being slightly stronger than U2). @seismicisolation @seismicisolation

Effect of key factors on permeability 199

Figure 6.11 (a) Relation strength vs w/c ratio for three OPCs; (b) effect of strength level and OPC on kO; data from Torrent and Jornet (1991).

Figure 6.11b presents the results of oxygen-permeability kO for concretes made with the three cements at each compressive strength level; the w/c of the mix is indicated above each bar. It can be seen that at the lowest strength level (25 MPa), the strongest cement GA shows the highest permeability, due to the high w/c (0.80) required to achieve that low strength. The concrete made with the “old” cement RO shows the lowest permeability, as a w/c = 0.55 is required to achieve the 25 MPa strength level. For concretes of strength levels 35 and 45 MPa, the strongest cement GA develops the lowest oxygenpermeability. From these results it seems that the lack of durability of “modern” cements should be attributed primarily to the use of concretes of low strengths (high w/c ratios) rather than to the high strength of the OPC.

6.3.2 Effect of Binder Type on Permeability For the development of this topic, binders are divided into two categories: “Conventional” and “Unconventional”. The former includes cements made with Portland cement clinker alone or combined with “conventional” mineral additions, i.e. those accepted by most national and international standards: SF, PFA, GBFS, natural or artificial (e.g. calcined clays) POZ, LF, etc. This includes the incorporation of the mineral additions at the cement plant (MIC: mineral components) or batched separately at the concrete plant (SCM: supplementary cementitious materials). Under “unconventional” binders we include those made with Portland cement clinker and “unconventional” mineral additions as well as those made without Portland cement clinker. 6.3.2.1 “Conventional” Binders In the research described in Section 6.3., the influence of adding SCMs to OPC U2 was also investigated (Torrent & Jornet, 1991). Four blended binders were manufactured in the laboratory, with the following compositions: @seismicisolation @seismicisolation

200 Concrete Permeability and Durability Performance

FA composed of 80% U2 + 20% of a commercial low-calcium PFA SL composed of 65% U2 + 35% of an industrial GGBFS LF composed of 85% U2 + 15% of an industrial LF SF composed of 92% U2 + 8% of a commercial densified condensed SF The blends were obtained in the laboratory by hom*ogenizing the powders in a “Turbula” powder mixer. On each mix, the coefficient of oxygenpermeability kO as well as the initial water absorption rate a3 (after 3 hours of contact with water) was measured on the bottom (as cast) surfaces of samples moist-cured during 28 days (other curing durations were also investigated, discussed in Section 6.7.2.1). Figure 6.12a and b presents the results of oxygen-permeability kO and initial water absorption rate at 3 hours a3, respectively, of mixes of similar cube strength (≈ 35 MPa), made with the OPC U2 and with the four blends described above. Both charts show a reduction in gas and water-permeability when PFA and SF and, to a lesser extent GGBFS, are added to OPC U2, whilst the effect of adding LF reduces the O2-permeability but increases the initial sorptivity a3. In 2012–2013, Holcim Technology Ltd. conducted a comprehensive experimental research project in its own laboratories in Holderbank, Switzerland, complemented by tests performed at EMPA (Dübendorf, Switzerland) and SUPSI (Lugano, Switzerland). The aim of the research project was to study the influence of w/b ratio and binder type on a large variety of durability indicators (Moro & Torrent, 2016). Here we will focus just on the permeability tests kO, kT and WPmax. A total of 18 concrete mixes, 9 with w/b = 0.40 and 380 kg/m³ of binder and 9 with w/b = 0.65 and 280 kg/m³ of binder, were prepared at EMPA with same 22 mm aggregates and admixtures, predominantly reaching “plastic” consistencies. The nine binders used are described in Table 5.4; more details on the composition of the binders can be found in Leemann and Moro (2017).

Figure 6.12 Effect of binder type on (a) O2-permeability kO, (b) initial sorptivity a3 data from Torrent and Jornet (1991).

@seismicisolation @seismicisolation

Effect of key factors on permeability 201

The first letter of the code in Table 5.4 indicates the clinker used to produce the cements (H: Höver, Germany; M: Merone, Italy). The values in brackets indicate the content and type of mineral additions originally included in the cement (MIC). When a mineral addition was added separately as SCM into the concrete mix, the content and type is indicated in italics. All cements and SCMs are industrial products sold on the market. The specimens for measuring kO and kT were three Ø150 × 50 mm discs, saw-cut from the bottom of cast Ø150 × 300 mm cylinders that were moist cured (moist room 20°C/RH > 95%) for 28 days. Prior to testing, the discs were oven-dried at 50°C for 6 days. The specimens for measuring WPmax were 150 mm cubes moist cured for 28 days. For the three tests, the bottomas-cast faces of the specimens were investigated. Figures 6.13 a and b present the coefficients of permeability to oxygen (kO) and to air (kT), respectively, of the mixes investigated. Figure 6.13 shows the wide range of kO and, especially, of kT values obtained for each w/b ratio (binder = cement + SCM), for the nine different binders investigated. Figure 6.13b shows that the values of kT, for the same w/b = 0.40, span two orders of magnitude for the different binders. The charts in Figure 6.14 present the same results of kO and kT of Figure6.13 but now plotted as function of the cube compressive strength of the mixes at 28 days. Figures 6.13 and 6.14 show that the higher kT values, for same w/b or strength, correspond to the concretes made with OPC (H0 and M0) and with 26% of LF (M26L). The beneficial effect of hydraulic MICs or SCMs (SF, GBFS and PFA) on kT can be appreciated. Binders containing 8% of SF (H8M) and 68% of GBFS (H68S) produced the less permeable concretes. This is not so obvious from the kO results that show high values for the mix containing 68% of GBFS (H68S). The reduction in kT by the addition of SF was also reported by Mittal etal. (2006).

Figure 6.13 Effect of binder type on (a) kO and (b) kT, measured on concretes of two w/b ratios; data from Moro and Torrent (2016).

@seismicisolation @seismicisolation

202 Concrete Permeability and Durability Performance

Figure 6.14 Effect of binder type on the relation of (a) kO and (b) kT with compressive strength; data from Moro and Torrent (2016).

The beneficial effect on kT of adding GBFS was experimentally confirmed by Panesar and Churchill (2010), who found that replacing OPC with GBFS in a w/b = 0.38 concrete mix reduced kT (10 −16 m²) from 0.077 (OPC mix) to 0.028 and 0.018 for 25% and 50% replacement levels, respectively. These authors conducted a comparative service life and life cycle analysis of culverts on the basis of those results. Glinicki and Nowowiejski (2013) investigated the effect of incorporating different levels of a Polish high calcium fly ash (W) on the mechanical and durability performance of concrete mixes made with w/c ratios 0.45 and 0.55. The composition of the cements used can be seen in Table 6.2 and includes binary and ternary blends by selective addition of a siliceous fly ash (V) and also GBFS (S). The cements were obtained by inter-grinding the constituents to a Blaine fineness of ≈ 380 m²/kg. Slabs (500 × 500 × 100 mm) were cast with each concrete mix, cured 21 days at 20°C/95% RH and then stored for 7 days at 20°C–22°C, 50%– 60% RH. Table 6.2 Types of cements investigated by Glinicki and Nowowiejski (2013) Main constituents (%) Fly ash Cement type (EN 197-1 classification) CEM I CEM II/A-W CEM II/B-W CEM II/B-M (V-W) CEM II/B-M (S-W) CEM V/A (S-W)a a

Clinker

GBFS S

94.5 80.9 67.4 66.6 66.6 47.9

14.3 28.9 14.3 14.3 23.9

14.3 -

14.3 23.9

Not defined in EN Standard 197-1.

@seismicisolation @seismicisolation

Effect of key factors on permeability 203

Figure 6.15 kT values of concrete mixes of w/c = 0.45 and 0.55 made with different cement types; data from Glinicki and Nowowiejski (2013).

Figure 6.15 shows the kT values measured by Glinicki and Nowowiejski (2013) at 28 days of age for both w/c ratios. It can be seen that, for w/c = 0.45, all the concretes containing active additions presented lower kT values than the reference CEM I (broken line) cement. For w/c = 0.55 the measured performance of the concretes was less consistent, with two ternary cements showing higher kT than the reference CEM I (dotted line); the same effect was also observed applying the water penetration under pressure test. An investigation was carried out in Turkey (Beglarigale et al., 2014) to study the effect of replacing 15% and 30% of OPC by a high calcium fly ash on the durability of concretes of 30 and 50 MPa cube strength. They also found a reduction in both air-permeability kT and water sorptivity at 24 hours with increasing replacement levels of OPC by fly ash, accompanied by a decrease in the charge passed in the RCPT and an increase in Wenner electrical resistivity (see A.2.1.1. and A.2.2.2). Mathur et al. (2005) conducted a research aimed at studying the performance of High-Volume PFA concretes. They tested Reference and HighVolume PFA concretes (30%–50% of PFA in the binder) of three w/b ratios (between 0.33 and 0.65) for durability, applying the RCPT (ASTM C1202) and air-permeability kT test methods. They found that the addition of PFA in such volumes reduced both the “chloride”- and gas-permeability, reaching extremely low values of both properties: 1,100

920 7,850

Effect of key factors on permeability 221

Figure 6.28 Effect of type and content of fibres on air-permeability kT of (a) conventional concrete and (b) SCC; data from Rodríguez de Sensale et al. (2018).

Figure 6.29 Effect of PPF on the air-permeability kT of two concretes; data from Moreno et al. (2013).

Figure 6.29 shows that the addition of PPF to concretes of relatively high w/c ratios, moderately reduces the air-permeability kT; the concrete of w/c = 0.80 with fibres is still slightly more permeable than the one with w/c = 0.62 without fibres. A similar reduction in kT by the addition of PPF was reported by Adámek and Juránková (2009). The possibilities of the combined use of locally available dune sand and alfa grass (Stipa tenacissima L.) to produce a repair mortar was investigated in Algeria by Krobba et al. (2018). Alfa grass is a tussock grass widely disseminated in semi-arid and arid regions, in North Africa and southern Spain (where it is called “Esparto”). For this investigation, Stipa fibres of 150–250 µm diameter were hand-cut to lengths between 3 and 5 mm. Dune sand mortars with and without the fibres (content within 0.1% and 1.25% by @seismicisolation @seismicisolation

222 Concrete Permeability and Durability Performance

volume of mortar) of exactly the same composition were prepared (including 2% of superplasticizer); the mortars flow ranged between 110% for the control mortars and 108% for mortars with high fibre contents, indicating that the fibres do not affect sensibly the workability. Flexural and compressive strength, as well as drying shrinkage of the mortars, were investigated, as well as their permeability through capillary suction and N2-permeability (Cembureau test method). Reported test results of the mortar with 0.75% vol. fibres showed moderately higher (16%) a 24 values and identical intrinsic N2-permeability than the control mix without fibres.

6.5.3 Polymers In the attempt to develop high-performance concrete for industrial floors, the use of polymers was investigated by Holcim Brazil (Kattar et al., 1995). A concrete mix with w/c = 0.46; 390 kg/m³ of high-early strength OPC and cylinder f′c 28d = 38.6 MPa was used as control, to which four different polymers were added: polyacrylate (ACR), styrene-butadiene (SB) and two brands of polyvinyl acetate (PVA1 and PVA2), in dosages indicated by the manufacturers or established by trial mixes. The control concrete specimens were moist cured for 28 days whilst those of the four polymermodified concretes were cured 24 hours in the moulds at 23°C/95% RH and, thereafter in a room at 23°C/65% RH, until the age of test. Besides the mechanical performance, the permeability of the concretes was determined by three different methods: air-permeability kT, coefficient of capillary suction A and water penetration under pressure Wp. Figure 6.30 presents the permeability results obtained by the three test methods on the control mix and on the polymer-modified mixes. Figure 6.30 shows that, in general, polymer-modified concretes have a lower permeability than the control mix, almost irrespective of the test method. In particular, the permeability performance of the concretes with ACR and SB polymers is superlative. The good performance of SB polymer in concrete was confirmed by Chmielewska (2013), who investigated the effect of adding it at dosages of 5%, 10%, 15% and 20% by mass of cement to a control mix made with 356 kg/m³ OPC and w/c = 0.45. She measured the gas-permeability Kg applying the Cembureau method, using Helium as transport gas, finding a very strong reduction of several orders of magnitude in Kg as the dosage of polymer raised (Figure 6.31), to the extent that, for 20% SB polymer, no gas flow was detectable across the specimen. Bhutta et al. (2009) developed a hardener-free epoxy-modified mortar aimed for repair work. A bisphenol A-type epoxy resin without any hardener was added in mass polymer/cement proportions (P/C), of 0%, 5%, 10% and 15%, to a 1:3 cement:sand mass ratio mortar. The compressive strength of the mortar increased with P/C up to 10%, decreasing for higher P/C ratios. In parallel, the air-permeability kT decreased monotonically @seismicisolation @seismicisolation

Effect of key factors on permeability 223

Figure 6.30 Permeability of polymer-modified concretes compared with a control mix; data from Kattar et al. (1995).

Figure 6.31 Helium-permeability of polymer-modified concretes as function of the SB polymer content; data from Chmielewska (2013).

with increasing P/C contents, becoming ten times lower for P/C contents of 10% or higher. Surface moisture measurements showed a marked increase for increasing P/C contents, which may have contributed to the lower kT measured values.

6.5.4 Expansive Agents Expansive agents are used in concrete, typically to compensate for shrinkage deformations due to drying or cooling. CaO-, calcium sulphoaluminate- and MgO-based expansive agents are the most commonly used. @seismicisolation @seismicisolation

224 Concrete Permeability and Durability Performance

In Argentina, the use of shrinkage-compensated concrete for industrial floors became quite popular, after a pioneer start in 1999 (Fernández Luco et al., 2003), reason why an investigation was conducted on the performance of such concretes, summarized by Fernández Luco (2004). Among the properties investigated, air-permeability kT was included. Figure 6.32 shows the values of kT obtained on concretes made with different cement types and dosages (in percentage of the cement content) of the CaO-based product Onoda-Expan. The cements used were an OPC, a blast-furnace slag cement BFSC, a LF cement LFC and a composite cement CPC. Figure 6.32 shows the beneficial effect of adding 10% of expansive agent (EA) to concrete mixes; irrespective of the cement type, the permeability falls into the “Low” Permeability Class and is significantly lower than that of the OPC concrete without expansive agent. Interesting to observe is a deliberate overdose (16%) of the EA on the OPC concrete which shows a huge increase in permeability, attributable to micro-cracking. The decrease in permeability at 10% dosage of EA can be attributed to the obturation of pores with hydration products of the EA; under real conditions, the consolidation of the microstructure due to the restrained expansion would probably decrease the permeability even further. Similarly, the large increase of permeability due to an overdose of EA may not take place under restrained conditions. In this respect, a commercial concrete supplied by a Japanese producer, containing 20 kg/m³ of expansive agent, was placed in two 0.6 × 0.6 × 0.9 m prismatic forms (Van et al., 2019). One of the prisms was plain whilst the other contained a cage of vertical Ø19 rebars and horizontal Ø16 stirrups. After 7d moist curing, the prisms were stored indoors until the ages of 28 and 56 days, when the air-permeability kT was measured on eight points of

Figure 6.32 Effect of cement type and dosage of expansive agent on concrete air-permeability kT; data from Fernández Luco (2004).

@seismicisolation @seismicisolation

Effect of key factors on permeability 225

each prism. At both ages, lower kT values were measured on the reinforced prism compared to the plain one (50% lower at 28 days and 75% lower at 56 days), which was attributed to the restraining effect of the bars to the expansion of the concrete, creating some “chemical precompression”. An investigation was conducted (Wang et al., 2011) to study the effect of a MgO expansive agent, alone and in combination with steel fibres (StF), on the performance of a concrete mix made with 340 kg/m³ OPC; w/c = 0.42; f′c 28d = 57 MPa. To this mix, 27.2 kg/m³ of EA or/and 78 kg/m³ of hookedend StF, (Ø = 0.4 mm and L = 25 mm) were added; Figure 6.33 shows the air-permeability kT of the four mixes investigated. The MgO EA produced a slight reduction of kT, whilst the StF did not have any effect on kT (in line with what was discussed in Section 6.5.2). The combination of EA and StF, shows some synergy, reducing the air-permeability by 40%, possibly due to the restraining effect of the fibres to the concrete expansion leading to a certain pre-compression. The performance of concretes made with and without a calcium sulphoaluminate-based expansive agent (used in conjunction with a Glycolbased shrinkage reducer) was investigated by Koh et al. (2006). They found that reducing the w/c ratio from 0.50 (normal concrete) to 0.30 (high-performance concretes) improved significantly the durability performance, measured through several tests, including chloride migration, carbonation and water- and air-permeability kT, performance that was further improved when the expansive agent was added at a dosage of 30 kg/m³ to the mix with w/c = 0.30.

Figure 6.33 Effect of addition of MgO expansive agent (EA) and StF on concrete airpermeability kT; data from Wang et al. (2011).

@seismicisolation @seismicisolation

226 Concrete Permeability and Durability Performance

6.6 EFFECT OF COMPACTION, SEGREGATION AND BLEEDING ON PERMEABILITY Placing and compaction/consolidation are among the most important operations in concrete construction. According to Mehta and Monteiro (2006), compaction or consolidation is a process that envisages moulding concrete within the forms and around embedded elements and rebars and the elimination of voids pockets and entrapped air. Sometimes, to spare time and labour, concrete is poured from great heights and at considerable rates, which promotes the appearance of coarse aggregate nests (honeycombing) due to segregation. Low paste-aggregate ratios and low fine-to-coarse aggregate ratios also favour segregation. Excessive bleeding in concreting of deep elements can be avoided if the poured concrete is gradually less plastic as the layers approach the top surface. Also, concrete shall be poured near its final position and in layers of thickness that enables a proper compaction. The previous layer shall be compacted before start pouring the next one, whilst the next layer shall be placed while the preceding one is still in a plastic state. A proper compaction is of utmost importance to obtain a suitable finishing of moulded surfaces and to make the concrete as dense as possible, i.e. to minimize voids and to promote a uniform distribution of aggregates within the concrete mass. This is achieved by supplying vibration/tamping energy to overcome cohesion and friction forces between concrete particles. Concrete consistency plays an important role in the result of concreting operations and should match the available tools for its placement and consolidation. The main consequences of inadequate placing or poor compaction are: honeycombing, excessive voids, bug-holes, sand streaks and cold joints. Honeycombing may be caused by poor placing, segregation or insufficient consolidation. Excessive voids are due to entrapped air that could not be released because vibration time was too short, or vibration power was insufficient, or the distance between vibration points was not close enough; they tend to appear behind vertical or inclined surfaces in the form of bug-holes (seeFigure11.23). Sand streaks are caused by heavy bleeding and paste loss along the form surface that may arise from the deposition method. Cold joints are originated by the absence of interpenetration between adjacent concreting layers. This happens when the poker-head does not penetrate the underlying layer while vibrating the one that is being cast or when the preceding layer has stiffened before placing the new layer. From a performance viewpoint, these problems affect aesthetics and produce heterogeneities, poor bonding between reinforcement and concrete, and loss of effective cross-section and cover depth. Indeed, the depth of honeycombs or bug-holes corresponds to an equal diminution of concrete cover. @seismicisolation @seismicisolation

Effect of key factors on permeability 227

All things considered, concrete permeability will be affected by placing and compaction. For instance, according to Mehta and Monteiro (2006), inadequate compaction is a typical cause of insufficient water-tightness. Analysing previously published data, Gonçalves (1999) concluded that compaction has a larger effect on resistance to chloride penetration than on compressive strength. Nevertheless, when trying to assess compaction effects on concrete permeability at a laboratory scale, the results are not conclusive because mimicking the lack of compaction in small – easy to compact – laboratory specimens is not easy. Although Gonen and Yazicioglu (2007) have found a consistent increase in sorptivity and porosity when lowering the compaction efficiency, it is quite common that researchers do not find conclusive variations of permeability with compaction degree. This happened, for instance, with Neves et al. (2011), Starck (2013) and Nishimura et al. (2015) for gas-permeability results and with Kumar and Bhattacharjee (2004) for surface water absorption. Gonen and Yazicioglu (2007) assessed sorptivity and porosity under vacuum of a mix with 415 kg/m³ of OPC and a w/c ratio of 0.53. Their specimens were either vibrated, compacted by spading 25 or 15 times, or not compacted at all. Gonen and Yazicioglu (2007) concluded that compaction has a very important effect on concrete durability as they found consistent and relevant differences in concrete sorptivity and porosity with the compaction degree (Figure 6.34), while only minor differences in compressive strength were found, except for the non-compacted specimens. Neves et al. (2011) carried out a research to investigate the effect of compaction on concrete carbonation resistance. Their study comprised four

Figure 6.34 Sorptivity and porosity of concrete under various compaction procedures; data from Gonen and Yazicioglu (2007).

@seismicisolation @seismicisolation

228 Concrete Permeability and Durability Performance

concrete mixes and oxygen-permeability kO was assessed on Ø150 × 50 mm discs, sliced from 150 mm diameter and 300 mm height cylinders. For each batch, a first set of specimens was compacted using a poke vibrator during a time judged by the operator as sufficient to achieve full compaction. A second set was compacted the same way but during half of the time. Finally, the last set was compacted by rodding. Although a good correlation between kO and carbonation depth was found, the effects of compaction on kO were not clear. Instead, the binder content and w/b ratio appeared as dominant factors defining the oxygen-permeability. Within the scope of a study to investigate permeation quality of concrete, Kumar and Bhattacharjee (2004) applied different compaction methods (vibration and rodding) aimed at achieving concretes with different permeability. Then, they applied the Initial Surface Absorption Test (ISAT) method (Section 4.2.2.1) to evaluate water-permeability. It was observed that in four out of the five cases the ISA of the rodded samples exceeded that of the vibrated samples. A successful experiment at relatively large scale was conducted by Liang et al. (2013) that will be described in some detail. In this research, a comprehensive theoretical and experimental analysis was made for optimizing the distance and time of vibration, using mock up elements at industrial scale. In particular, they prepared concrete blocks (500 × 500 and 800 mm high) in wooden moulds, cast in two layers 400 mm high each, by pouring concrete through an inverted street marking cone (bottom opening Ø100 mm) from the four angles of the mould. Each layer was compacted applying a poker vibrator (Ø50 mm) fixed in the centre of the section during predefined times. Two concrete mixes were tested, both with w/c = 0.50; one having a slump of 80 mm (OPC = 300 kg/m³) and the other of 150 mm (OPC = 324 kg/m³). Both mixes were air-entrained: 5.1% and 4.5%, respectively. The 28-day compressive strength of the mixes was about 40 MPa, based on drilled cores’ testing. Two blocks were cast with the 80 mm slump mix, with vibration durations of 15 and 60 seconds, respectively, and another two with the 150 mm slump mix, with vibration durations of 15 and 30 seconds, respectively. Only the block of 80 mm slump, vibrated just 15 seconds showed surface anomalies in the form of air voids. The four blocks were kept in the forms during 5 days and then stored at 20°C/60% RH until the age of 28 days. At that age, 20 cores (Ø100 × 200 mm) were drilled from one surface, following a pattern of four cores in width per five cores in height. The cores were tested for strength (compressive and splitting) and for accelerated carbonation. Prior to that, air-permeability tests kT were conducted in situ on same places where the cores were later drilled. The kT results obtained on the four blocks are presented in Figure 6.35, showing the geometric mean kTgm ± s LOG (see Section 5.8.3). Unfortunately, the paper does not identify the results as function of the height, although the cores’

@seismicisolation @seismicisolation

Effect of key factors on permeability 229

Figure 6.35 Air-permeability kT results obtained for different compaction practices; data from Liang et al. (2013).

tests did not indicate a significant effect of the position of the core on the strength results. Figure 6.35 shows that the mix with 80 mm slump, thoroughly compacted (60 seconds vibration) had the best performance in terms of kTgm air-permeability. On the opposite, the same mix but inadequately compacted (15 seconds vibration) showed the worst results. The mix with 150 mm slump shows an intermediate result, without a significant beneficial effect in extending the vibration time from 15 to 30 seconds. It is also interesting to note the smaller variability in kT of the mixes with 150 mm slump compared to those of the 80 mm slump, which may indicate a more hom*ogeneous compaction of the softer mix. The accelerated carbonation test also showed 15% higher values for the 80 mm slump concrete vibrated for 15 seconds, compared to the other three blocks. Within a comprehensive research conducted by Hayakawa and Kato (2012), Hayakawa et al. (2012) and Kato (2013), the effect of segregation and bleeding on the air-permeability kT was investigated. A set of 14 OPC concrete mixes, of widely different characteristics, were prepared in the laboratory; the composition and fresh state characteristics (slump, air content and bleeding) are described in Kato (2013). The effect of bleeding on air-permeability was investigated by casting three 150 mm cubes with concrete mixes of same w/c = 0.55 but different compositions: slump between 30 and 225 mm and bleeding between 0.0

@seismicisolation @seismicisolation

230 Concrete Permeability and Durability Performance

Figure 6.36 Effect of bleeding on air-permeability kT measured on lateral faces of 150 mm cubes; data from Kato (2013).

and 4.0 mm). The zero-bleeding mix was achieved by adding a viscosityenhancing agent. After casting, the cubes were seal-cured for 7 days and thereafter exposed to a room with 20°C/60% RH until the time of test (28and 91 days). The air-permeability kT was measured on the lateral sides of the cubes reporting the average of three cubes, shown in Figure 6.36. The kT results obtained at both 28 and 91 days show an increase in permeability with the amount of bled water. Kato expected a different result, since the amount of bleeding on the upper surface should have led to a lower w/c ratio in the rest of the cube and a possible reduction in kT with bleeding. He concludes that bleeding channels may increase the connectivity of the pore network in the specimen leading to an increase of kT. Unfortunately, kT values on the top surface were not measured or reported. A similar result was obtained by Positieri et al. (2011) on concretes with and without powdered pigments, already discussed in Section 6.5.1. They found an increase in kT with the amount of bleeding of the mixes; a similar effect was found on water sorptivity. Figure 6.36 suggests that, when bleeding exceeds 1.5 mm, the effect on kT becomes more significant (suggestion supported by the results of Positieri et al. (2011)). Another aspect investigated by Kato (2013) was the effect of segregation, with and without the presence of reinforcement bars, on the air-permeability kT. The effect of segregation was studied on a block (500 × 500 and 800 mm high) with three of the sides reinforced as detailed in Figure 6.37. The figure also shows (dotted circles) the locations of the air-permeability measurements that allowed establishing profiles of kT values with height. The mix selected for the investigation had w/c = 0.55, 125 mm slump, 5.0% air and 1.2 mm bleeding. @seismicisolation @seismicisolation

Effect of key factors on permeability 231

Figure 6.37 Details of the “segregation” block (Kato, 2013).

The kT profiles obtained along the four lateral faces of the block are presented in Figure 6.38 (the reinforcement pattern is expressed as separation of vertical bars – separation of stirrups in mm). The thick vertical line (kT = 0.61 × 10 −16 m²) indicates the kT value measured on a companion 150 mm unreinforced cube. Figure 6.38 shows a general trend of increasing kT values with height, attributable to settlement and segregation of solids in the fresh concrete. The effect is more acute for the unreinforced side of the block (showing a ratio of 30 between the extreme top and bottom kT measurements), which suggests that the reinforcement, even with the relatively large cover thickness of the experiment, helps in preventing the settlement and segregation of the concrete. This was confirmed by Hayakawa and

Figure 6.38 kT profiles with height; data from Kato (2013).

@seismicisolation @seismicisolation

232 Concrete Permeability and Durability Performance

Kato (2010), showing also that excessive vibration of a relatively soft mix (180 mm slump) increased the variability of kT with height along relatively large prisms 0.9×0.9×1.2 m (L×W×H). Kato (2013) also found a relation between the Bleeding Index (amount of bleeding water in cm multiplied by the height of the measurement in cm) and the kT ratio (ratio between the measured kT and that measured on the lateral face of a 150 mm cube), as shown in Figure 6.39. This bleeding index is used in a model to predict the carbonation rate of concrete structures, on the basis of kT measurements and including factors to account for the type of binder and for the environmental conditions (Kato & Hayakawa, 2013). Another investigation (Nsama et al., 2018) showed a consistent trend of higher gas-permeability (kT and oxygen-permeability measured by an electrochemical technique) near the top of 300 × 300 ×1,500 mm columns, compared to the results obtained near the bottom. MIP measurements made on samples taken at different heights confirmed a larger porosity near the top than near the bottom of the columns. Nsama et al. (2018) concluded that “air-permeability coefficients tended to be higher in the upper parts of the concrete column specimens compared to those observed in the lower parts, due to segregation in form of bleeding. This can be better accounted for by the larger sizes of interconnected pores formed in those locations, especially in the OPC, as investigated via MIP. The variation observed in the rate of oxygen-permeability measured using cathodic polarization technique was like those observed in the results of air-permeability coefficients taking into consideration of moisture content. The observed results suggested that the upper parts of column specimens are more prone to corrosion, attributed to the adverse effects

Figure 6.39 Relation between kT ratio and Bleeding Index; data from Kato (2013).

@seismicisolation @seismicisolation

Effect of key factors on permeability 233

of bleeding on the cover concrete and integrity of concrete and horizontal steel bars.” The findings of these investigations definitely provide ground to the importance of testing concrete permeability not just on laboratory specimens but also on site, as discussed in Chapter 7. 6.7 EFFECT OF CURING ON PERMEABILITY

6.7.1 Relevance of Curing for Concrete Quality Curing is an essential step in concreting practice, very often overlooked; its importance for cement hydration was discussed in Section 3.1.2. Curing can be defined as the measures to be undertaken to provide hydrothermal conditions to young concrete that are favourable for the material to fully develop its potential performance, which implies: • • • • •

advanced development of hydration reactions avoidance of plastic shrinkage cracking avoidance of drying shrinkage cracking avoidance of thermal cracking avoidance of frost damage by premature exposure to sub-zero temperatures

Regarding concreting practices in general and curing in particular, it is worth citing from Chapter 3 of Neville (2003): “In my experience, achieving durable concrete requires all the operations of concreting to be as nearly perfect as possible. I would like to move now to several more practical aspects of making good concrete. Because they are practical, some people, mainly academics, regard them to be unworthy of consideration and study; above all, these problems cannot be solved by an elaborate computer program. Curing concrete is the lowest of low-tech operations… it is seen by many as a silly operation, a non job…and [bad] curing does not show… If I emphasize ensuring curing, it is because curing can make all the difference between having good concrete and having good concrete ruined by the lack of a small effort.” Multiple references to curing effects on concrete properties can be found in the literature. Several authors (Powers et al., 1954; Page et al., 1981; Gräf & Grube, 1984; Rasheeduzzafar & Al-Saadoun, 1989; Balayssac etal., 1995; Meeks & Carino, 1999; Bai et al., 2002; Güneyisi et al., 2007) reported that early drying causes higher “penetrability” and that curing conditions, especially curing time, have a large effect on the durability of concrete. According to Hansen (1986), in ordinary mixes, poor curing can easily @seismicisolation @seismicisolation

234 Concrete Permeability and Durability Performance

triplicate capillary porosity. Although curing has also an effect on concrete strength, it is acknowledged that it has more impact on durability, because it affects more strongly the surface layers, which are vital to prevent the penetration of aggressive agents (Ramezanianpour & Malhotra, 1995; Fernández Luco, 2010; Maslehuddin et al., 2013). When moist curing is stopped before binder hydration is sufficiently developed (the most common situation), as the water loss is not uniform across concrete volume, there will be a zone, near the surface where the external environment has stronger effect on the local humidity regime, the so-called curing-affected-zone (CAZ) (Cather, 1994). This effect is reflected in different concrete properties between the CAZ and the remaining concrete volume; in fact, inadequate curing can result in a very weak and porous material near the concrete surface (Gowripalan et al., 1990; Ewertson & Petersson, 1993). This topic is dealt with in more detail in Chapter 7.

6.7.2 Effect of Curing on Permeability 6.7.2.1 Investigations in the Laboratory The investigation described in Section 6.2.1.1 provided useful experimental information, not just on the type of binder but also on the effect of curing on several characteristics of concrete, including water- and gas-permeability. Figure 6.40a shows the effect of moist curing on the oxygen-permeability kO and Figure 6.40b on the initial water absorption rate a3 (measured after 3 hours of contact with water). The data presented correspond to mixes prepared with various OPCs (RO, GA and U2, described in Table 6.1), exposed to Curing regimes A, B and C (1, 7 and 28 days moist curing), described in Section 6.2.1.1 and to cores drilled from the bottom side of the slabs as cast (hence why 1 day moist curing is attributed to the slabs that were kept in their moulds for 24 hours). The mixes are referenced by the OPC code (RO, GA, U2) followed by the concrete compressive strength level in MPa.

Figure 6.40 Effect of curing and concrete strength on (a) O2-permeability kO and (b) water sorptivity a3.

@seismicisolation @seismicisolation

Effect of key factors on permeability 235

Both figures present basically the same pattern, which is very typical for most water- and gas-permeability tests on OPC concretes: moist curing in the laboratory is extremely beneficial in the first 7 days, but extending it beyond 7 days causes only marginal reduction in permeability. Notice also that the mix of lower strength (25 MPa) is more sensitive to the lack of curing than mixes of higher strength. This can be explained by the fact that poorer mixes have more and coarser pores that facilitate the evaporation of water, especially at early ages. Figure 6.41 shows the ratio between the oxygen-permeability kO measured on concretes moist cured just 1 day and that measured after 7 and 28 days moist curing. The figure includes concretes made with OPCs and with the four blends of U2 with SCMs: 20% PFA (FA); 35% GGBFS (SL); 15% LF and 8% SF, described in Section 6.3.2. All mixes were of strength level 35 MPa, except the SF blend which was slightly stronger. The top of the black bar shows the ratio of kO between 1 and 7 days moist curing and the top of the white bar the ratio between 1 and 28 days moist curing. What can be seen in Figure 6.41 is the already discussed effect that the first 7 days of curing are determinant in reducing gas-permeability, with a less relevant impact of extending the moist curing in the laboratory beyond 7 days. This is valid for all mixes, except for that made with OPC U5, which may be due to an experimental error. Now, if we compare the mix made with OPC U2 with those made with the four blends, the higher sensitivity of the latter to the lack of curing is evident, even – although to a lesser extent – for that of the blend containing LF. Figure 6.42 presents results (Hooton, 2011) of water-permeability Kw (test method not disclosed) of concretes of w/b = 0.45, made with OPC and with 25% and 50% replacement by blast-furnace slag. One set of specimens was moist cured until the age of testing (91 days) whilst the other set was

Figure 6.41 Effect of moist curing duration on O2-permeability kO of OPC and blended cements concretes.

@seismicisolation @seismicisolation

236 Concrete Permeability and Durability Performance

Figure 6.42 Effect of curing on water-permeability of OPC and slag concretes; data from Hooton (2011).

air-cured. Figure 6.42 shows the beneficial effect of incorporating slag, provided the concrete is thoroughly moist cured; on the contrary, if concrete is not well cured, the incorporation of slag has a negative impact on the water-permeability. In another research by Alamri (1988), specimens from different concrete mixes (made with OPC and 30% and 60% replacement by PFA and GGBFS, respectively) were exposed to environmental conditions of 50°C/15% RH, 40°C/60% RH and 20°C/100% RH, until preconditioning them for airpermeability test at 28 days, using the Figg method (Section 4.3.2.1). The reported results show that harsh environmental conditions during binder hydration can lead to significant increases in air-permeability of the concrete surface and that both higher w/c ratios and the presence of SCM increase the sensitivity of concrete to the environmental conditions. Concretes of w/c = 0.40 and 0.70, prepared with different cement types, were tested by Gonçalves et al. (2000). Their test specimens were wrapped with a plastic sheet during 3 and 7 days and afterwards exposed to 20°C/75% RH until testing (at 28 days). The results are presented in Figure 6.43a (O2-permeability kO) and 6.43b (initial water sorptivity at 1 hour a1h), where each cement type is identified according to EN 197-1 classification. In all cases, except for OPC (CEM I) and w/c = 0.40 a reduction in both gas- and water-permeability is observed by extending the sealed-curing from 3 to 7 days. The WIST (Section 4.2.2.6) was applied on two mixes with the same binder content (340 kg/m3) and w/b ratio (0.50), but different binder types (OPC and GBFS) (Nguyen et al., 2019a). Relatively large specimens, after being sealed-cured for different periods, were kept in an environment with average conditions T = 16.4°C/RH = 69.1% until the age of testing (90 days). @seismicisolation @seismicisolation

Effect of key factors on permeability 237

Figure 6.43 Effect of seal-curing duration on (a) O2-permeability and (b) initial sorptivity of concretes; data from Gonçalves et al. (2000).

Figure 6.44 WIST results of concretes with different curing duration; data from Nguyen et al. (2019a).

The mean number of water-spraying repetitions until concrete surface saturation is shown in Figure 6.44 (the higher the number, the more absorptive the surface). The results in Figures 6.41–6.44 confirm the fact that concretes containing MIC or SCM require better curing to develop their full durability potential. 6.7.2.2 Investigations in the Field Here we refer to an investigation conducted by Surana et al. (2017), where five different curing regimes were applied to 1.2 × 1.2 × 0.2 m reinforced slabs, cast with an OPC concrete of w/c ratio 0.55 and 35 MPa compressive strength. After casting, the following curing regimes were applied to the slabs: @seismicisolation @seismicisolation

238 Concrete Permeability and Durability Performance

(7dH): The slab was cured using two layers of wet hessian-cloth until the age of 7 days, after which it was exposed to air. Hessian cloth was kept wet by sprinkling water intermittently (Air): The slab was left exposed to air after casting without any deliberate curing measures (WX): Curing compound based on wax in water (Wax Emulsion) (AS): Curing compound based on acrylic resin in organic solvent (RW): Curing compound based on resin in water (Resin Emulsion) The curing compounds were applied immediately after the disappearance of bleed water sheen from the concrete surface by using a compressed airassisted spraying gun, at a coverage rate of 5–6 m²/L, taking care of a similar and proper coverage of the slab surface. After finishing each curing process, the slabs were placed unprotected outdoors, exposed to the weather conditions of Chennai, India. At 28 days of age, Ø70 mm cores were drilled from the upper side of the slabs and cut to 30 mm thickness, following the procedure for measuring OPI, described in Section 4.3.1.3 and WSI (Water Sorptivity Index). The results obtained are presented in Figure 6.45. When analysing Figure 6.45 it has to be borne in mind that a higher OPI means a better concrete of lower permeability. Figure 6.45 shows that the worst performance in terms of gas- and water-permeability corresponds to the air-cured (air) concrete – something expected – and to the concrete cured with the wax emulsion compound (WX). The overall best performance corresponds to the concrete cured with the resin-based compounds (RW and RS).

Figure 6.45 Effect of curing method on OPI and WSI; data from Surana et al. (2017).

@seismicisolation @seismicisolation

Effect of key factors on permeability 239

6.7.3 Effect of Curing on Air-Permeability kT 6.7.3.1 Conventional Curing Several laboratory investigations have been conducted to assess the effect of the curing conditions on the air-permeability kT of the Covercrete (Torrent & Ebensperger, 1993; Torrent & Frenzer, 1995; Kubens et al., 2003; Quoc & Kishi, 2008, 2009; Ichiro et al., 2009; Solís-Carcaño & Moreno, 2009; Fernández Luco, 2010; Kurashige & Hironaga, 2010; Song et al., 2014; Okasaki et al., 2015; Kawaai et al., 2015; Neves & Torrent, 2016; Ebensperger & Olivares, 2017; Yokoyama et al., 2017; Nakarai et al., 2019). These investigations consisted in casting specimens with concretes of different compositions, subjecting them to different curing conditions and in measuring the air-permeability kT of the resulting concrete qualities. In what follows some of them will be discussed in some detail. In one research, Quoc and Kishi (2008, 2009) investigated the impact of the duration of water curing on air-permeability kT. For this, they prepared specimens with OPC concrete mixes of w/c = 0.30, 0.45 and 0.60 and cured them under water or sealed at 20°C during 0, 3, 7 and 28 days. Thereafter, the specimens were kept in a dry room (20°C/65% RH) until the age of 56 days when kT was measured. Figure 6.46 shows the beneficial effect that the duration of water curing has on kT, displaying the typical trend for laboratory testing, in that the first 7 days of moist curing are critical to achieve a tight concrete, whilst extending the curing beyond 7 days has only a small effect on kT. A similar effect can be found also for the

Figure 6.46 Effect of the length of water (a) or sealed (b) curing period on kT; data from Quoc and Kishi (2008, 2009).

@seismicisolation @seismicisolation

240 Concrete Permeability and Durability Performance

sealed specimens. Comparing both curing methods, the higher impact of water curing over sealed curing becomes evident, with the former bringing kT about one order of magnitude lower compared with the same duration of sealing (sealing method not disclosed in the paper). Similar to what was discussed in Section 6.2.2.3, Quoc and Kishi (2008) proposed a linear relation between log kT and w/c ratio, with the proportional factor (designated as curing factor) being a function of the duration of the moist or sealed curing period. In a comprehensive investigation, several non-destructive tests were performed on 150 × 150 × 530 mm prisms, made with an air-entrained OPC concrete of w/c = 0.50 (Kurashige & Hironaga, 2010). The prisms were subjected to different curing conditions, combining different durations in the moulds (1, 5 or 14 days), followed by exposure to different storage conditions, as described in Table 6.4. Companion 100 × 100 × 400 mm prisms were used to measure accelerated carbonation after 7, 28, 91 and 182 days exposure at (20°C/60% RH; 5% CO2). Figure 6.47 shows, in decreasing order, the results of kT measured at 91 days of age on prisms subjected to the curing and storage conditions described in Table 6.4. It can be seen that the prisms with worst curings 1–40 and 1–60 (1 day in the mould and permanently exposed to dry air afterwards) show the higher kT values. The 1–40H bar shows that a late exposure to humid air reduces kT but with poorer performance than the prisms subjected initially to moist curing (1–H and, especially, 1–W). Regarding permanence in the moulds as curing technique, Figure 6.47 shows the effectiveness of 5 days (prisms 5–40, 5–60 and 5–H), with little influence of the exposure after demoulding. Extending the permanence in the moulds to 14 days (prism 14-60) has little effect on kT, confirming the trend shown in Figure 6.46.

Table 6.4 Different curing, storage conditions and testing schedule, from Kurashige and Hironaga (2010) Code

Exposure environments (T = 20°C)

1-40H

40% RH

1-40

40% RH

1-60

60% RH

1-H

Humid air

1-W

In Water

5-40

40% RH

5-60

60% RH

5-H 14-60 Age (days)

Conditioning and testing

Humid air 60% RH until 91 days, when tested for: 60% RH • Rebound number • Air-permeability kT • Electrical resistivity • Accelerated carbonation

Humid air In the moulds 1 5 14

60% RH 28 56

@seismicisolation @seismicisolation

Effect of key factors on permeability 241

Figure 6.47 Effect of curing regime on kT; data from Kurashige and Hironaga (2010).

One of the earliest investigations on the effect of curing on kT was reported by Kubens et al. (2003). They subjected specimens, made with two OPC concrete mixes (200 kg/m³; w/c = 0.87 and 320 kg/m³; w/c = 0.55) to five different curing conditions, from worst to best: A: no curing, storage at 30°C/40% RH until the testing age B: 5 minutes in water, three times a day, for 3 days and thereafter at 30°C/40% RH until testing age C: 5 minutes in water, three times a day, for 6 days and thereafter at 30°C/40% RH until testing age D: 6 days in water and thereafter at 30°C/40% RH until testing age E: 28 days in water and thereafter at 30°C/40% RH until the testing age Tests were performed at 91 days and included, besides kT, also compressive strength, chloride migration (ASTM C1202) and accelerated carbonation (7 days at 5% CO2; 30°C/50% RH). Figure 6.48 shows the impact of the curing schedule on kT and accelerated carbonation depth. The plots confirm the significant benefits of a continued moist curing on both kT and carbonation depth. It is worth noting that the richer mix, totally deprived of moist curing (A), shows almost the same permeability kT and carbonation depth as the poorer mix cured 28 days in water (E). This means that not curing the concrete was equivalent to reducing the cement content of the mix by 38% (120 kg/m³), with respect to standard curing. As becomes evident from Figure 6.48, the logarithm of kT showed a very good correlation with the accelerated carbonation depth. In another relevant investigation (Nakarai et al., 2019), different curing conditions were applied to six large concrete mock-up blocks (1.5 × 1.5 m), @seismicisolation @seismicisolation

242 Concrete Permeability and Durability Performance

Figure 6.48 Effect of curing schedule on kT and accelerated carbonation depth; data from Kubens et al. (2003).

with thickness of 0.6 m for N mix and 0.4 m for B mix. The blocks were representative of companion box culvert structures. The N mix contained 295 kg/m³ of OPC with w/c = 0.55, whilst the B mix contained 303 kg/m³ of BFSC and w/c = 0.525. Each mock up block received a different curing: 1 day in the mould (N-1d and B-1d); 5 and 7 days in the mould (N-5d and B-7d) and 3 months protected with a plastic sheet (N-3m and B-3m); the latter curing was also applied to the real box culverts. After finishing the corresponding curing period, the mock-up elements were exposed to the field conditions (in average 16°C/62% RH, with minimum RH around 35%) protected from rain, with the lateral faces sealed, with the box culverts exposed to the same climatic conditions. The 1.5 × 1.5 m faces of the mockup blocks were used for testing; regarding the N and B real box culverts, the tests were conducted on the inner walls, at a height of 1 m from the floor. The change of air-permeability and surface moisture (electrical impedance-based method) was periodically monitored for each mock up element and box culvert from 1.3 to 39 months. The results of kT are plotted as function of exposure time in Figure 6.49a and b for the N elements and the B elements, respectively (data at 2.6 months, not shown, removed as outliers, as shown in Bueno et al. (2021)). The beneficial effect of initial curing, at least of 5 days for N mix and of 7 days for B mix, is clear from Figure 6.49; the higher sensitivity to moist curing of the concrete B, made with BFSC, is also evident. The advantages of wet curing, with respect to membrane and dry curing, on the air-permeability kT of concrete, were also confirmed by Mu et al. (2015), with special relevance for concretes containing PFA and/or GGBFS as OPC replacements. @seismicisolation @seismicisolation

Effect of key factors on permeability 243

Figure 6.49 Effect of curing on evolution of air-permeability kT for (a) Mix N and (b) Mix B; data from Nakarai et al. (2019).

Another test method applied to evaluate the effect of curing duration on the water-permeability of the elements (Nakarai et al., 2019) was the WIST, Section 4.2.2.6, which yielded similar results, regarding the positive contribution of curing in reducing concrete surface permeability, especially for a mix with BFSC. Results of air-permeability kT of concretes made with three different cements and cured in the lab (20°C/95% RH) and outdoors were reported by de Schutter (2016). The cements, described by its designation after EN 197-1 Standard, were CEM I 52,5 N (OPC); CEM III/B 42,5 (BFSC, with 66%–80% slag) and CEM III/C 32,5 (BFSC, with 81%–95% slag). Figure 6.50 presents the data, showing that outdoors curing doubles the permeability of the OPC concrete, but increases ten-fold that of the concretes made with both BFSC. This confirms the higher sensitivity of cements containing active mineral additions to the lack of moist curing, already shown and discussed in connection with Figures 6.41–6.44, showing results obtained with other test methods. A curing method using thermoplastic, water-repellent sealing sheets that are attached to the formwork and left in place after demoulding (for periods of weeks and even months), was studied by Nukushina et al. (2015). The method proved to be effective, especially for concretes containing PFA and/ or GGBFS. The investigations discussed above consisted in measuring the effect of the duration of moist curing, be it by external contact with water or humid air or by sealing, on kT of conventional concretes. 6.7.3.2 Self-Curing An investigation on the effect of internal curing on the performance of highstrength concretes was reported by Zhutowsky and Kovler (2012). For that, @seismicisolation @seismicisolation

244 Concrete Permeability and Durability Performance

Figure 6.50 Effect of curing on concretes made with different cement types; data from de Schutter (2016).

OPC concrete mixes with w/c = 0.21, 0.25 and 0.33 were prepared with normal weight aggregates (mixes 0.21, 0.25 and 0.33, respectively) and also replacing the fraction 2.4–4.8 mm of the sand with water-saturated pumice particles (mixes 0.21 L, 0.25 L and 0.33 L, respectively). Pumice sand had a 1 hour vacuum absorption of 73% and the water carried by them was meant to compensate for the chemical shrinkage of the cement paste at 7 days. The specimens were seal-cured at 30°C and, at ages of 1, 7 and 28 days, oven-dried at 60°C till constant weight and cooled to ambient temperature. Then, kT was measured, as well as water sorptivity and chloride migration (ASTM C1202). Figure 6.51 shows the effect of the internal curing on kT. It can be seen that, except for the specimens, oven-dried at 1 day, the kT of the specimens with internal curing (ending with L in Figure 6.51) was lower than the reference samples, with a more significant effect at 28 days. This effect was confirmed by the Coulombs measured (ASTM C1202) for mixes 0.25 and 0.33 L, but not for 0.21 L; on the contrary, the sorptivity tests showed higher values for the internally cured samples. 6.7.3.3 Accelerated Curing Another investigation compared the effect of two accelerated curing methods on the properties of a concrete (w/c = 0.43; OPC = 387 kg/m³), namely: steam curing (SC) and microwave heating (MH) of steel moulds (Choi et al., 2019). Different cycles of both SC and MH (variable pre-curing, heating and cooling times) with maximum temperatures within the range of 50°C–65°C were investigated on specimens and mock-up elements. Theresponse of the @seismicisolation @seismicisolation

Effect of key factors on permeability 245

Figure 6.51 Effect of internal curing (L) on concretes of w/c ratios 0.21, 0.25 and 0.33; data from Zhutowsky and Kovler (2012).

material in terms of compressive strength, MIP, scanning electron microscopy (SEM), free and restrained shrinkage and air-permeability kT were measured. They found that both SC and MW curing led to similar, quite low values of kT ( 50°C) were built up between the centre and the surface of the pylon and moderate, but not negligible, for the cubes (ΔT > 25°C), with the consequent risk of thermal cracking. The different thermal evolution of cast specimens, cubes and the pylon, confirms the impossibility of mimicking the Realcrete by keeping elements (even as large as 1 m cubes) close to the real structure. Also, the thermal volume changes are differently restrained in the Realcrete than in the cubes.

Figure7.7 Thermal evolution of 250 × 360 × 120 mm laboratory slabs, 1 m field cubes and pylon (Schaffhausen bridge).

@seismicisolation @seismicisolation

296 Concrete Permeability and Durability Performance

Figure7.8 Air-permeability kT of Labcrete, 1 m cube and Realcrete of Schaffhausen Bridge Deck and Pylon.

The kT test results are summarized in Figure7.8 (Torrent, 1999) for the Labcrete slab, the 1 m cube and the Realcrete of both Bridge Deck and Pylon. To the right of the chart, the Permeability Classes (see Table 5.2) “Moderate” (M), “Low” (L) and “Very Low” (VL) are indicated. As indicated along the bars, the kT of the Realcrete was 4.4 times higher than Labcrete’s for the Deck, and 17 times higher for the Pylon. In the latter, thermal cracks that were not visible at the time of the tests but became evident later, required a coating surface treatment. The large difference between Realcrete and Labcrete is attributable to the cracks. It is interesting to note that, even with that large loss in performance, the Realcrete kT of the Pylon is significantly lower than that of the Deck, due to the much higher quality of its mix design. 7.1.5.3 Lisbon Viaduct The difference between the kT of Labcrete and Realcrete was also investigated by Neves and Santos (2008) during the construction of a motorway viaduct in Lisbon. It is a 770 m long and 32.4 m wide box girder, with spans ranging from 50 to 105 m, simply supported by double-headed pylons. The investigation comprised concreting surveillance of three segments of the box girder, one pylon, two sets of prefabricated struts and two phases of one of the abutments. For each, a 120 × 250 × 360 mm specimen for air-permeability testing and three 150 mm cubes (for compressive strength testing) were cast. The location where the concrete from the corresponding truck was poured was pinpointed, to carry out in situ testing in the same spots afterwards.

@seismicisolation @seismicisolation

Need for site assessment of durability 297

Figure7.9 Air-permeability kT of Labcrete and Realcrete of Lisbon’s Viaduct (Neves & Santos, 2008).

After casting, the cubes were moved to the jobsite laboratory, stripped at 24 hours and cured in water until the testing age (28 days), while the slab specimens were cured in water for 6 days and then kept at 22°C and 60% RH until the testing age. Three different concrete mixes were applied in the several monitored concreting operations: (1): C40/50.S4.D25, (2): C40/50.S3.D25 and (3) C30/37.S3.D25, following EN 206-1 (EN 206, 2013) notation. The slab specimens and the corresponding structural elements were both tested when the concrete was 28 days old. Figure7.9 summarizes the test results, revealing a worse performance of Realcrete compared with Labcrete, as on average the kT of Realcrete was nearly twice that of Labcrete. Interestingly, trends were noticed for higher ratios (Realcrete kT/Labcrete kT) in troubled concreting segments of the box girders, carried out in the evening, than for quieter concreting such as for the prefabricated struts. 7.1.5.4 Swiss Bridges’ Elements A recent investigation conducted in Switzerland for the Federal Highway Administration (Jacobs et al., 2018) adds more information on the difference between Realcrete and Labcrete in terms of durability. It consisted in testing, in parallel, cast specimens (Labcrete) and the Realcrete of different elements belonging to six structures, as described in Table 7.1. Labcrete 150 mm cube specimens were cast with the fresh concrete destined to the elements and moist cured for 28 days. Cores were drilled from the Labcrete cubes, on which mechanical and durability tests were @seismicisolation @seismicisolation

298 Concrete Permeability and Durability Performance Table 7.1 Characteristics of the concretes of the six structures investigated (Jacobs et al., 2018) Site

Elements

Binder

Strength class

EN exposure class

Ha Gn Ep Ma Gr

Parapets; abutment wall Parapets; abutment wall Walls; decks Retaining wall Walls; decks

CEM II/B-M (T-LL)

C30/37

CEM II/B-M (S-T) CEM II/A-LL + PFA

Shoulder curb

CEM I + PFA

C25/30 C25/30 C30/37 C30/37

XC4, XD3, XF4 XC4, XD3, XF2 XC4, XD3, XF2 XC4, XF2 XC4, XD1, XF4 XC4, XD3, XF4 XC4, XD3, XF4

performed. At 28 days, cores were drilled from the corresponding structural elements, on which the same tests were performed (Realcrete). The results can be summarized as follows: • the Realcrete density was generally 1%–2% lower than Labcrete’s • the Realcrete 28-day compressive strength was ~20% lower than Labcrete’s • the Realcrete chloride migration coefficient M Cl was 45% higher than Labcrete’s (see Figure7.10) • the Realcrete freeze-thaw-salts resistance was, in average, similar to Labcrete’s although with very large scatter • the Realcrete carbonation rate was ≈ 40% higher than Labcrete’s (determined only on Ep jobsite) These results confirm that the impact of the lower quality of Realcrete/ Covercrete is much more acute for durability tests than for strength tests. Figure 7.10 shows that the coefficient of chloride migration M Cl of the Realcrete is almost invariably higher than Labcrete’s, at an average 45% higher. Swiss Standard (SIA 262/1, 2019) specifies that the value of M Cl (Tang-Nilsson test method described in A.2.1.2) of the Labcrete shall not exceed 10 × 10 −12 m²/s (for moderate and severe chloride exposures). Figure7.10 shows that 24 out of 26 results (92%) conform to that limit. The same standard specifies a maximum limit of 12 × 10 −12 m²/s for tests made on cores drilled from the structures (Realcrete). Figure7.10 shows that now 19 out of 26 results comply with that limit; the percentage of conformity drops from 92% to 73%. The results discussed in this section, obtained from Bözberg Tunnel, Schaffhausen Bridge, Lisbon Viaduct and from the six other structures investigated in Switzerland, confirm that the durability performance of the Realcrete/Covercrete is significantly worse than Labcrete’s, a fact also acknowledged in the literature (Bouwer, 1998; Beushausen & Alexander, @seismicisolation @seismicisolation

Need for site assessment of durability 299

Figure7.10 Chloride migration MCl of Realcrete vs Labcrete for different structures; data from Jacobs et al. (2018).

2009). This fact stresses the need to include not just the Labcrete, but especially the Realcrete, in performance specification and quality assurance schemes. 7.2 ACHIEVING HIGH COVERCRETE’S QUALITY Section 7.1 was devoted to demonstrating the typically lower quality of the Covercrete, compared to the Labcrete. This section, in turn, deals with means to achieve a Covercrete of high quality (low “penetrability”).

7.2.1 Mix Design and Curing The simplest way of improving the “penetrability” of the Covercrete is to build with higher quality concretes, which involves the correct selection of the cement/binder, the design of mixes with sufficiently low water/binder ratios and the application of sound concreting practices: good consolidation without segregation and, especially, adequate and sufficiently long moist curing conditions (see Sections 6.2–6.7).

7.2.2 UHPFRC Ultra High-Performance Fibre-Reinforced Composites (UHPFRC) are cementitious composite materials made with extremely low w/b ratios (0.15 and lower), silica fume, well-graded fine aggregate (typically below 0.5 mm) and large volume of steel fibres (StF) and superplasticizers. Different @seismicisolation @seismicisolation

300 Concrete Permeability and Durability Performance

Figure7.11 Pore size distribution of conventional, HPC and two UHPFRC (Fehling et al., 2005).

concepts have been developed worldwide, some known by their abbreviation or commercial brands (SIFCON, SIMCON, CEMTEC, Ductal, etc.). These materials show an extremely low and fine porosity, see Figure7.11 in which the pore size distributions of two UHPFRC mixes (B3Q and M1Q) with compressive strengths above 200 MPa, are compared with those of a Normal Concrete (cube strength 55 MPa) and a High-Performance Concrete (cube strength 105 MPa) (Fehling et al., 2005). It is expected that such tight material as UHPFRC should show a low “penetrability” to fluids and ions. Within their very comprehensive investigation (Fehling et al., 2005), performance tests were conducted on conventional and UHPFRC mixes, namely, migration, gas-permeability and water sorptivity. In all cases, the UHPFRC mixes showed much better performance than the reference conventional mix (cube strength class C45). Figure7.12 presents the test results of Nitrogen-permeability kN2 (Cembureau test, see 4.3.1.2) for the conventional concrete and for two UHPFRC formulations. For the latter, the indication WL denotes storage under water while 90°C stands for the temperature of a thermal treatment applied to the specimens. In Figure 7.12, it can be seen that UHPFRC mixes have significantly lower permeability to gas than a normal conventional concrete, something expected given their tighter pore structure, shown in Figure7.11. Results of water absorption coefficient (DIN 52617), reported by Fehling et al. (2005), confirm the results of gas-permeability. This type of UHPFRC has been used in Switzerland (Brühwiler, 2007) to repair concrete bridges, applying it selectively in the most vulnerable areas, as described in detail in Section 11.3.7. Information on the spatial variability of tensile strength and air-permeability kT of such material, within @seismicisolation @seismicisolation

Need for site assessment of durability 301

Figure7.12 Permeability to N2 of UHPFRC compared to conventional concrete; data from Fehling et al. (2005).

a 1.5 × 3.0 m panel (42 mm thick), can be found in Oesterlee et al. (2009), showing values within the range 0.001–0.01 (× 10 −16 m²), for the extremely low geometric mean value of 0.0046 × 10 −16 m². A different concept of UHPFRC was developed in Japan, embedding high-strength polyethylene fibres (Ø12 µm × 6 mm in size) in a cementitious composite of w/b = 0.18–0.22, containing approximately 1300, 230 and 155 kg/m³ of cement, silica fume and fine sand, respectively, plus a high dosage of superplasticizer and an air-remover agent. The resulting material was designated as UHP-SHCC (Ultra High-Performance Strain-Hardening Cement Composite) and was intended as a repair material (Kunieda et al., 2011; 2012). The durability of the composite was measured in terms of its air-permeability kT and its water sorptivity, using a test method similar to the Karsten tube (Section see 8.3.3.3). The measured values of airpermeability kT of the composites were rated within the “Very Low” Permeability Class (see Table 5.2); the water sorptivity was also very low.

7.2.3 Controlled Permeable Formwork (CPF) Liners 7.2.3.1 A ction Mechanism of CPF Liners Controlled permeable formwork (CPF) liners are sheets of synthetic fibres fabric, capable of retaining cement-sized and larger particles, but of allowing water and air to flow through it under hydrostatic pressure, intensified during vibration. They are attached onto the formwork inner surface. As schematized in Figure 7.13a, under the vibration action the surface concrete layers lose water and air and are enriched by cement dragged from the inner layers (Leow, 2004), resulting in a local lowering @seismicisolation @seismicisolation

302 Concrete Permeability and Durability Performance

Figure7.13 (a) CPF action with vibrated concrete; (b) effect of CPF on Autoclam airpermeability index (Basheer et al., 2005).

of the w/c ratio. The effect of the CPF liners is achieving a Covercrete that is less penetrable than the core, thus offering a better protection to the embedded steel. The effect of CPF liners in reducing the amount of bug-holes from the concrete surface was discussed in Section 5.7.5.4 and shown visually in Figure5.28. The efficiency of CPF to reduce the “penetrability” of the treated surface has been proved by several investigations (Cullen, 1998; Malone, 1999; COWI-Almoayed Gulf WLL, 2002; Leow, 2004; Basheer et al., 2005; Law et al., 2012; Adam et al., 2009; Tanaka et al., 2012; Torrent et al., 2012; Ohta et al., 2019). 7.2.3.2 Impact of CPF on the “Penetrability” of the Covercrete Several investigations were carried out to assess the potential benefit of using CPF on the quality of the Covercrete. One of them was carried out at Queen’s University Belfast (UK) (Basheer et al., 2005) to assess the performance of “Formtex”, a CPF liner manufactured by a Danish company called Fibertex. Prisms measuring 250 × 750 × 150 mm were cast with two concrete mixes of w/c = 0.45 and 0.50 (50 and 40 MPa cube strength at 28 days, respectively). One of the surfaces (250 × 750 mm) was cast against the natural plywood form, whilst the opposite against a Formtex CPF liner stapled onto the plywood form. After 24 hours, the formwork was stripped and the prisms were stored 28 days at 20°C and 75% RH, moment in which they were tested for: Pull-off strength, Air-Permeability and Water Absorption @seismicisolation @seismicisolation

Need for site assessment of durability 303

(both with Autoclam system, see Section 4.1.2.2), Accelerated Carbonation and Chloride ingress (after 100 days immersion in salt solution). The test results showed an increase in pull-off strength of 30%–33% on the CPF face, compared with the natural surface. Figure7.13b shows the effect of the CPF liner on Autoclam’s Air-permeability Index (K a), comparing the results with those obtained on the surface cast against the plywood form, for each w/c ratio. The test was repeated reusing the liner once. The results show clearly the beneficial effect of the CPF liner in reducing airpermeability, even when reusing the liner once. This lower air-permeability was accompanied by significantly lower water sorptivity and carbonation depths and moderate lower chloride penetration depths for the surfaces cast against the CPF liner. The suitability of “Formtex” CPF liner in improving the tightness of the Covercrete was confirmed by Holčapek (2011), who found an approximately ten-fold reduction in air-permeability kT on the surface treated with the CPF liner. Another study was conducted at Holcim Technology Ltd, Switzerland, but on a concrete aged 16 years (to test the lasting effect of the CPF) (Torrent et al., 2012). A panel measuring 600 × 500 × 200 mm was cast with a concrete of w/c ratio = 0.55 and 28-day cube compressive strength = 42.0 MPa. The concrete was poured inside a wooden formwork, one of its 600 × 500 mm faces covered with the “Zemdrain Classic” CPF Liner, manufactured by Dupont, and carefully compacted with internal poke vibration. During and immediately after casting, it was possible to observe water being drained to the floor through the CPF. The panel was stripped at 24 hours and stored permanently in a dry room (20°C/50% RH), till the age of test (16 years). At that age, the following NDTs were applied on both faces of the panel: Rebound hammer R and Air-Permeability coefficient kT. In addition, Ø50 × 200 mm cores were drilled through the entire panel thickness to measure the carbonation depth CD and chloride migration coefficient M Cl through the depth of chloride penetration Xd of both surfaces (TangNilsson method, see A.2.1.2). Table 7.2 summarizes the recorded results. The results of Table 7.2 indicate that, after 16 years of conservation in a dry room, the performance of the CPF face is significantly superior to that of the Formwork face, both in terms of hardness R and air-permeability Table 7.2 Test results obtained on the two opposite faces of the panel (Torrent et al., 2012) Property

Rebound R

kT (10−16 m²)

Face

Mean

kTgm

sLOG

Mean

Max

Mean

46 54

4.8 2.1

6.6 0.79

0.16 0.14

23 15

27 18

28.8 17.5

3.3 1.5

43 26

4.4 2.2

Formwork CPF

CD (mm)

@seismicisolation @seismicisolation

MCl (10−12m²/s)

Xd (mm)

304 Concrete Permeability and Durability Performance

kT. An analysis of the differences in R and kT values measured on both faces indicates that the CPF has reduced the w/c ratio of the CPF surface by around 0.15. The application of the CPF liner reduced the Carbonation Depth CD by 35% and the Chloride Migration M Cl and Penetration Xd by 40%, in both cases compared to the wooden formwork face. One important question regarding the performance of CPF liners is to which depth the dewatering effect is noticeable; this information is also required if service life models are to be applied. To assess the influence depth of “Zemdrain”, 25 mm thick slices were saw-cut from seven drilled cores, at different depths from the “Z” face (in contact with the CPF liner). A slice was also obtained from the other extreme of the core to test the “F” face (in contact with the formwork). These slices were oven-dried for 2 days at 50°C and then put in contact with 3 mm of water for a water suction test (Swiss Standard SIA 262/1:2019, Annex A). The coefficient of water absorption at 24 hours a24, at different depths, is plotted in Figure 7.14 (black dots with full line). Figure7.14 illustrates the average water sorptivity of the four cores tested at each depth; it can be seen that the effect of the CPF is stronger at the surface, but that it is still noticeable at a depth of 25 mm from the CPF liner face. For layers beyond 50 mm, the sorptivity does not differ significantly from that of the “F” formwork surface. These results are in agreement with the 20–50 mm range of action of the CPF, indicated by Malone (1999). Based on the a24 profile, tentative profiles for M Cl and kT have been built (the values reported in Table 7.2 are shown with symbols at both faces of the panel) that could be used for modelling purposes. Another investigation with a similar aim was carried out in Japan (Ohta et al., 2019) in which four reinforced concrete columns (0.4 × 0.4 × 1.5 m)

Figure7.14 Water sorptivity of concrete slices at different depths from “Zemdrain” face (Torrent et al., 2012).

@seismicisolation @seismicisolation

Need for site assessment of durability 305

Figure7.15 kT and Figg tests applied on Dewatered Layer of concrete (Ohta et al., 2019).

were cast with concrete mixes of design strengths 21 and 30 MPa, with opposite faces cast against the formwork and against CPF liners. The columns were exposed to outdoor conditions in Japan for 17 years. The following tests were conducted on the Formwork and CPF faces: air-permeability kT and Figg method (Section 4.3.2.1). Both methods are sketched in Figure 7.15, applied on the Dewatered Layer (DWL), being worth mentioning that kT is measured right on the surface whilst Figg’s Air-Permeability Index is measured inside a Ø10 × 40 mm hole drilled from the surface. In addition, cores were drilled from the columns to measure the carbonation depth at 17 years, expressed as carbonation rate Kc (carbonation depth/square root of exposure time). Figure7.16 shows the Kc values measured on both faces (CPF and formwork) for the two mixes, as function of (a) the air-permeability kT and (b) of Figg’s Air-Permeability Index. Figure7.16 clearly shows that the carbonation rate on the surfaces cast against the CPF liner is negligible, compared with the low value of 1.1 mm/y½

Figure7.16 Carbonation rate Kc vs (a) kT and (b) Figg air-permeability index (Ohta et al., 2019).

@seismicisolation @seismicisolation

306 Concrete Permeability and Durability Performance

for the 30 MPa mix and moderate value of 3.7 mm/y½ for the 21 MPa mix, values measured on the surfaces cast against the formwork. There is an excellent correlation between the carbonation rate and the air-permeability coefficient kT (Figure 7.16a), whilst Figg test did not yield such a good correlation (Figure7.16b). This may be due to the fact that Figg intrusive method explores concrete at a depth of 20–40 mm, where the effect of the liner is not so strong (see Figure7.15). The above described investigation by Ohta et al. (2019) was completed by sealing the curved surfaces of Ø50 mm cores drilled from surfaces cast against the conventional formwork and against CPF liner with epoxy resin, leaving only the external surface exposed. The sealed cores were immersed in a 10% NaCl solution for 28 days. After that, the cores were saw-cut into 10 mm slices, the concentration of chlorides of which was measured according to JIS 1154 (Japanese industrial standard) test method. Figure 7.17 shows the chloride profiles obtained. The chloride profile for the concrete in contact with the conventional formwork presents much higher chloride concentration and a higher slope of the profile compared with the concrete in contact with the CPF, indicating a higher coefficient of chloride-diffusion. In another investigation, Tanaka et al. (2012) studied the effect of the type of form (wooden or metal) and of two types of CPF liners applied on them, on several concrete properties: air-permeability kT, rebound number, water-permeability plus carbonation depth and chloride penetration, both in accelerated tests. The investigated surface conditions are described in Table 7.3 (the tested mix had 291 kg/m³ of OPC and w/c = 0.57). In general, no significant difference was found between the concrete surfaces cast directly in contact with the wooden and the metal forms (some tests indicated a slightly better performance for the metal form). On the

Figure7.17 Chloride profiles of cores exposed to NaCl solution, with and without CPF liner (Ohta et al., 2019).

@seismicisolation @seismicisolation

Need for site assessment of durability 307 Table 7.3 Surface conditions investigated by Tanaka et al. (2012) Code Form CPF liner

Wooden None

Metal None

Wooden Type A

Metal Type A

Wooden Type B

Metal Type B

contrary, all test methods found a significant reduction in the “penetrability” of the concrete surfaces cast against both CPF liners (without a significant effect of the type of liner or of the substrate on which they were applied). Tanaka et al. (2012) studied the profile of the properties with height on the tested blocks measuring 0.48 × 0.48 m by 1.2 m high. All tests, except water-permeability showed higher “penetrability” of the concretes cast on the CPF liners near the top of the blocks compared with that measured at lower positions, which is due to the lack of sufficient hydraulic head to push the water through the fabric. This is shown in Figure7.18a for air-permeability kT and in Figure7.18b for chloride ion penetration after immersion of cores with the external faces exposed to 10% NaCl solution. This is important, as some measures are required to compensate for that effect (perhaps a richer mix just for the top 250 mm of the element) so as to ensure the tightness of the whole structural element. It can be concluded that CPF liners are a good solution to improve the quality of the Covercrete. In particular, for massive structures that only require a low w/c near the surfaces, the solution offers technical (less heat in the concrete mass and less susceptibility to thermal cracks) and economic advantages (the cost of the CPF liner is offset by the reduction in cement content of the mass). Some economic considerations on the use of the CPF are presented in Torrent et al. (2012), related to the surface/volume ratio of the structure.

Figure7.18 Profiles of (a) kT and (b) chlorides penetration with columns height (Tanaka etal., 2012).

@seismicisolation @seismicisolation

308 Concrete Permeability and Durability Performance

It is worth mentioning that the effect of CPF liners for self-compacting concrete is not so significant as with vibrated concrete, due to the lack of hydrostatic pressure to force the water through the fresh concrete of high viscosity, typical of that type of concrete, as experienced by one of the book’s authors and as reported by Barbhuyia et al. (2011).

7.2.4 Shrinkage-Compensating Concrete Shrinkage-Compensating Concrete (ShCC) can be defined as a concrete that, when properly restrained by reinforcement or other means, expands an amount equal to, or slightly greater than, the anticipated drying shrinkage. Subsequent drying shrinkage will reduce these expansive strains but, ideally, a residual expansion will remain in the concrete, thereby eliminating shrinkage cracking. To achieve that, ShCC must contain a suitable expansive component, that may be a special expansive cement or a combination of a conventional cement with a suitable expansive agent (EA) which is added at the concrete mixer. ShCC finds applications in jointless industrial floors (Fernández Luco et al., 2003), in hydraulic and containment structures, nuclear power plants, bridge decks, etc. (ACI 223R, 2010). The positive effect of adding an expansive agent on the quality of the Covercrete was discussed in Section 6.5.4.

7.2.5 Self-Consolidating Concrete Conventional concrete is usually consolidated in the forms by means of vibration, be it internal, through poker vibrators, or external, through vibrators attached to the forms. Standards and Recommendations (e.g. EN 13970, 2009; ACI 309R, 2005), dealing with the execution of concrete structures describe the correct techniques for consolidating concrete applying vibration. In real practice, vibration is seldom applied according to the recommended good practices, with gross errors becoming visually evident. These malpractices lead to under-consolidation in certain areas (reflected as honeycombing in extreme cases), over-consolidation in others and usage of the vibrator to move concrete horizontally (both resulting in segregation), etc. Even when properly applied, vibration is “per se” a discontinuous process, with the concrete directly under the influence of the vibrator consolidating differently than concrete in between successive points of application of the vibrator. In addition, as a result of settlement of fresh concrete, the consolidation of the bottom layers, particularly in deep pours, is higher than in the upper layers, the latter showing usually higher permeability. These aspects have been discussed in detail in Section 6.6. Self-consolidating (or self-compacting) concrete, or SCC, when properly designed and produced, does not need consolidation by external means, except natural gravity. @seismicisolation @seismicisolation

Need for site assessment of durability 309

In a laboratory experiment (Assié et al., 2007), the permeability of SCCs was compared to that of conventional vibrated concretes VC of similar strength classes (20, 40 and 60 MPa). The permeability tests applied were O2-permeability kO (Cembureau) and water sorptivity at 24 hours a24. These tests were complemented by porosity, chloride migration, accelerated carbonation, MIP and leaching by NH4NO3. The relative merits of both sets of concretes vary according to the test applied but not to an extent to claim superiority of one over the other. A comprehensive investigation was reported by Fornasier et al. (2003), in which the permeability of several SCC mixes was compared with that of conventional concretes CC in the laboratory. Here we will refer only to Group A mixes; Table 7.4 summarizes their characteristics. Three permeability tests were included in the laboratory investigation: water sorptivity at 24 hours a24, water penetration under pressure Wp, and air-permeability kT. Figure 7.19a shows the results of a 24 and Wp (maximum penetration), whilst Figure7.19b presents the kT results obtained at 28 and 56 days of age. Figure7.19a shows that the water sorptivity a 24 of all SCC mixes (especially SCC3) is lower than that of the conventional concrete CC. Regarding Wp only SCC1 shows a better performance than CC, with SCC2 showing a poorer performance. Table 7.4 shows that the 28-day compressive strength of CC is also higher than that of all SCCs. Table 7.4 Main characteristics of Group A mixes investigated by Fornasier etal. (2003) Mix Composite cement (kg/m³) GBFS (kg/m³) Limestone filler (kg/m³) Water/powder ratio Cylinder f ′c28d (MPa)

SCC1

SCC2

SCC3

420 0.40 55.1

430 0.44 45.7

370 240 0.28 47.6

250 200 0.40 50.6

Figure7.19 Performance comparison of SCC vs. CC in terms of (a) a24 and Wp and (b) kT; data from Fornasier et al. (2003).

@seismicisolation @seismicisolation

310 Concrete Permeability and Durability Performance

In the case of kT, Figure7.19b presents results of SCC obtained on laboratory specimens, as well as results obtained on site with mix SCC2. Indeed, SCC2 was used for the construction of “E-shaped” walls (H: 4.2 m, T: 0.25 m and L: 4 and 10 m) in an industrial complex, being poured from the top. The data shown in Figure7.19b indicate that the kT of the three SCCs is higher than that of the CC. It is interesting to observe that the kT value obtained on site is about half that obtained in the laboratory on the same mix. This may be due to good quality of the applied concreting practices, but also to higher moisture (not reported) of the concrete on site, compared to the 50°C oven-dried (5 days) laboratory specimens. As a conclusion, although experimental confirmation is still lacking, especially based on large scale site tests, the level and hom*ogeneity of consolidation expected for SCC should be higher than for vibrated concrete. In addition, thanks to its high viscosity, the settlement of SCC is lower compared to conventional concrete, adding to the hom*ogeneity of the finished structure. These advantages of SCC over conventional concrete are highlighted in Section 5.11 of ERMCO (2005).

7.2.6 Permeability-Reducing Agents According to ACI 212.3R (2016), permeability-reducing admixtures (PRAs) are a class of materials developed to improve concrete durability through controlling water and moisture movement, as well as by reducing chloride ion ingress and permeability. PRAs encompass a range of materials of various performances. Permeability-reducing admixtures (PRAs) typically include, but are not limited to, the following categories: Regarding polymers, their beneficial effect in reducing the “penetrability” of the Covercrete has already been discussed in Section 6.5.3 and that of sealing bacteria in Section 6.11. The amount and types of materials available in the market, claiming to act as PRA, are very large and growing, making the decision of the user on whether to use a PRA, and which one, extremely difficult. The following recommendation of ACI 212.3R (2016) is worth quoting: “Users of a PRA should evaluate performance of the product in concrete based on the application requirements. A commercial PRA can @seismicisolation @seismicisolation

Need for site assessment of durability 311

include components from several material categories, making classification based strictly on terminology or chemistry inaccurate. The final selection should be based on the project requirements and the performance of the PRA based on appropriate testing, as described in 15.3. The PRA manufacturer is responsible for conducting tests to demonstrate the PRA is suitable for its recommended application. The PRA’s performance should be evaluated over a sufficient amount of time to demonstrate the long-term performance of the product, as some PRAs have an extended history of successful use.” In short, the use of these products should be decided strictly on the basis of testing with the materials and proportions intended for the project. For instance, a report by a well-reputed laboratory in Italy indicates that the performance of a pore blocker was very good in reducing the water penetration under pressure of concretes with w/c ratio 0.50–0.60, but its effect on a concrete with w/c = 0.45 was not significant. Another important issue in the decision is the duration of the PRA’s effect and the possibility of reapplication, if the concerned structure is to be protected during a long service lifetime. The test methods presented in Chapter 4 and in Annex A offer a good choice to be used with the purpose of assessing the efficiency of PRAs, with the following remarks: • tests applied on the surface are better suited, since intrusive tests may be testing the quality of the concrete beyond the depth of influence of the PRA (Figure7.15) • PRA intended to reduce the penetration of water or liquid solutions are to be tested by water-permeability or sorptivity tests, because they may be permeable to gases (e.g. water vapour) • PRA intended to reduce the penetration of gases (e.g. CO2) are to be tested by gas-permeability or gas-diffusion tests A comprehensive assessment of coating agents and impregnation materials, applying different test methods, can be found in Misono et al. (2014). Not just the quality of the product matters, but also the application technique and skills are very important to ensure a reliable protection of the structure. Figure7.20 (l) shows the variability of the coefficient of airpermeability kT on one side of a 6-year old concrete wall (Quoc & Kishi, 2006), whilst Figure7.20 (r.) shows the kT values measured on the opposite side that was treated with a coating (undisclosed type). Figure 7.20 (l.) shows a large variability of kT values in the uncoated wall, with some values very high, illustrating the heterogeneity often found in the field. On the contrary, Figure7.20 (r.) shows a much more hom*ogeneous set of significantly lower kT values, demonstrating the positive effect of the coating even after 6 years of application. Yet, Figure 7.20 (r.) still @seismicisolation @seismicisolation

312 Concrete Permeability and Durability Performance

Figure7.20 Contour kT map of (l) uncoated wall and (r) coated wall. Original in colour in Quoc and Kishi (2006).

reveals some isolated spots of high permeability, due to faulty application of the coating, spots that constitute vulnerable areas of the wall. This confirms the need to check that the application of the PRAs is executed correctly, without leaving vulnerable areas. 7.3 COVER THICKNESS Although beyond the scope of this book, the relevance of this topic is so great that cannot be overlooked. The thickness of the concrete cover is a very important durability indicator for the deterioration of structures due to steel corrosion. Let us recall the solution of the second Fick’s law of diffusion presented in Section 3.4.1: x   C ( x, t ) = C0 + (Cs − C0 ). 1 − erf( ) 4.D.t  

(7.1)

Let us make C(x, t) = C cr (critical concentration of the aggressive agent, CO2 or Cl−, that triggers corrosion). If we take the radical of the error function, the rate of penetration of aggressive agents (front of CO2 , front of critical Cl− concentration) into concrete follows a “square root” law: where x = penetration of critical front of the agent (mm) R = rate of penetration (mm/year½) t = age (years) @seismicisolation @seismicisolation

Need for site assessment of durability 313

If we calculate the time ti taken for the critical front to reach the steel, we have from (7.2): d t i =   R

(7.3)

where ti = corrosion initiation time (years) d = cover depth (mm) In theory, both the second Fick’s diffusion law (through the argument of the error function solution) and capillary suction theory (see Sections 3.4.1 and 3.7.1) predict a progress of the penetration front of carbonation, chlorides and water with the square root of time. This means that the time of initiation of corrosion can be considered proportional to the cover depth squared, Eq. (7.3). As a result, a reduction of 10% in the cover thickness would mean a reduction of about 20% in corrosion initiation time ti. Figure7.21 shows the combined effect of changes in the cover thickness d and of the air-permeability kT on the corrosion initiation time, taking as reference the Swiss specification for corrosion induced by de-icing salts (exposure class XD3). The central curve shows a conformity situation for a concrete of kT = 0.1 × 10 −16 m²; it can be assumed that the corrosion initiation time will reach the expected 50 years if the cover thickness equals the nominal value specified (55 mm), situation represented by the black dot on the central curve. If, instead of 55 mm, the cover thickness happens to be just 50 mm, the service life is reduced from 50 to 41 years, situation represented by the white square on the central curve. The upper and lower curves in Figure 7.21 represent the evolution of the critical chloride front with time, for kT values

Figure7.21 Effect of cover thickness d and kT on corrosion initiation time ti.

@seismicisolation @seismicisolation

314 Concrete Permeability and Durability Performance

one order of magnitude higher and lower, respectively, from the central value (assuming that R in Eq. (7.3) is proportional to kT ⅓ , see Section 9.4.3.1). It can be seen that the service life (white dots on each curve) is reduced and increased, respectively, by a factor of ~2 with respect to the expected 50 years. Despite the progress made on instruments capable of assessing, nondestructively, the cover depth quite accurately, their use is not forcibly specified in the standards. It should be reminded that the cover thicknesses, measured prior to placing the concrete, may be modified during the concreting operations, often resulting in insufficient covers of the steel in the end-product (Neville, 1998). Neville also coined the term “negative cover” when the steel happens to protrude from the concrete surface. There are many examples reporting lack of conformity with specified cover thickness. For instance, Figure 7.22a shows the histogram of measured cover thickness (Realcrete) in Norwegian 1981 Gymsøystraumen Bridge, which reflects a significant lack of compliance with the specified value (Theorecrete) (Gjørv, 2014). Figure 7.22b shows several cases, reported by Lim (2013) for Malaysian jetties. The horizontal black lines indicate the specified nominal cover (60 or 75 mm, depending on the type of element, with −5 mm tolerance), whereas the vertical bars indicate the minimum and maximum covers measured on site with covermeters, calibrated with direct physical measurements. It can be seen that in most cases the Realcrete measured cover is in defect with respect to the Theorecrete specified value; in particular, in structures 1, 6 and 9, none of the measurements reached the specified value. This was considered the primary cause for the unsatisfactory performance of the jetties (Lim, 2013). A survey in the UK showed that 77 out of 200 bridges had too low cover thickness (Wallbank, 1989) that resulted in spalling of the cover and rust. It must be mentioned that there are in the market electromagnetic covermeters capable of assessing the cover thickness to ±10% accuracy (RILEM TC 189-NEC, 2007; Fernández Luco, 2005). A relatively old, possibly

Figure7.22 Measured (Realcrete) vs. specified (Theorecrete) covers: (a) Norway and (b) Malaysia; data taken from Gjørv (2014) and Lim (2013), respectively.

@seismicisolation @seismicisolation

Need for site assessment of durability 315

outdated standard indicates how to do it (BS 1881-204, 1988). For very deep bars or complex reinforcement patterns, the modern ground penetrating radar (GPR) instruments complement well the capabilities of the electromagnetic instruments. More up-to-date documents on the topic are Chapter 9 of RILEM TC 230-PSC (2016) and DBV (2015). For durability and/or service life assessment of new or old reinforced concrete structures under risk of steel corrosion deterioration, measuring the thickness of the cover is as important as measuring its permeability (“penetrability”). 7.4 SPACERS AND PERMEABILITY Spacers are small pieces used to position and fix the steel bars at the required distance from the exposed surfaces, so as to ensure that the proper cover thickness is achieved. The more popular spacers are made of mortar/concrete and plastics, although stainless steel is also used. Inevitably, the spacers intrude the Covercrete and establish a link between the reinforcement and the environment to which the structural element is exposed. As spacers are placed every meter or so and remain permanently in place, some concern exists on their influence on the potential corrosion risk of the supported steel (Alzyoud et al., 2016), concern supported by cited reports and field investigations. An investigation was conducted by Alzyoud et al. (2016), in which the effect of spacers on transport properties (O2-diffusivity, O2-permeability and water sorptivity) was investigated. The variables considered were the type of spacer (none, plastic, concrete and steel), the aggregate size (Dmax = 10 and 20 mm), curing (3 and 28 days sealed curing) and the pre-conditioning of the specimens (20°C/75% RH; 20°C/55% RH; oven-dried at 50°C) on the above-mentioned variables. Complementary techniques were applied on the samples, namely: chloride-diffusion, microscopy of fluorescent epoxy impregnated surfaces and backscattered electron imaging. Cylindrical specimens (Ø100 × 25 or 50 mm) were cast (Dmax = 10 mm) or cored from 1,500 × 600 × 50 mm slabs (Dmax = 20 mm). The specimens contained in the centre a spacer, kept firmly in place during casting; a control specimen without spacer was also prepared. After curing, the specimens were subjected to the pre-conditioning regimes described above until constant mass prior to conducting the transport tests. Figure7.23 presents the results of (a) O2-permeability and (b) water sorptivity a7 (at 7 hours) of the samples, relative to the values measured on the control sample (no spacer). The spacer codes are the following: • SS: stainless steel spacer • CS: concrete spacer • PS: plastic spacer @seismicisolation @seismicisolation

316 Concrete Permeability and Durability Performance

Figure7.23 Effect of spacer type and pre-conditioning regime on relative (a) O2-permeability and (b) water sorptivity at 7 hours of 10 mm Dmax concretes, cured 28 days (Alzyoud et al., 2016).

• PSa: PS plastic spacer ground with 120-grit SiC paper • PSb: PS plastic spacer with scoring notches on the main flange Figure 7.23 shows clearly that all the samples containing spacers, irrespective of the pre-conditioning treatments, present higher gas- and waterpermeability than the control sample. However, the performance of the different spacers is not the same, with the negative effect of spacers on the permeability increasing in the order: steel → concrete → plastic spacers. Regarding the latter, the performance improves when the smooth surfaces are roughened (PSa and PSb) but not to the extent of equalling the performance of either concrete or steel spacers. It is interesting to remark that the effect of spacers on permeability can only be realistically revealed and quantified through site testing of the finished structure. 7.5 CONCLUDING REMARKS The objective of this chapter was to highlight that a large difference in quality frequently exists between the concrete in the structure (Realcrete), with respect to that tested in the laboratory on cast specimens (Labcrete) and even more to that conceived and specified by the designer (Theorecrete). In the case of durability that relies to a large extent on the performance (and thickness) of the few cm/in constituting the near-surface layer (Covercrete) the impact is significantly higher than for bulk properties, such as strength. According to Polder and Rooij (2005), this difference between the intended and effective quality and thickness of Covercrete is the major cause for premature deterioration of reinforced concrete structures. @seismicisolation @seismicisolation

Need for site assessment of durability 317

It has been shown that it is impossible to reproduce the Realcrete conditions by casting specimens, even as large as 1 m thick, to be stored near the real structure, not to mention small specimens kept under laboratory conditions. The conclusion is that the quality (and thickness) of the vital Covercrete can only be assessed realistically by means of testing the end-product, i.e. site testing of the actual structural elements. This can (and should) be done preferably by non-destructive methods or, alternatively/complementarily, by laboratory testing of specimens prepared from cores drilled from the structure. This holds true also when surface treatments, intended to reduce the “penetrability” of the Covercrete, are applied on the exposed concrete. It follows that service life assessment of concrete structures can only be realistically done by involving direct measurements of the Covercrete, which reflect the true spatial variability of both “penetrability” and thickness of the cover concrete. Without them, current deterministic or probabilistic models, based on Theorecrete or Labcrete definition of the key materials’ parameters, will always have a large degree of inaccuracy. Chapter 9 deals more extensively with this matter. REFERENCES ACI 212.3R (2016). “Report on chemical admixtures for concrete”, 76 p. ACI 223R (2010). “Guide for the use of shrinkage-compensating concrete”, 20 p. ACI 309R (2005). “Guide for consolidation of concrete”. Adam, A.A., Law, D.W., Molyneaux, T., Patnaikuni, I. and Aly, T. (2009). “The effect of using controlled permeability formwork on the durability of concrete containing OPC and PFA”. Techn. Letter, Institut. of Engs., Australia, 1–12 p. Alzyoud, S., Wong, H.S. and Buenfeld, N.R. (2016). “Influence of reinforcement spacers on mass transport properties and durability of concrete structures”. Cem. & Concr. Res., v87, 31–44. Andrade, C. (2006). “Multilevel (four) methodology for durability design”. RILEM Proceedings PRO 47, 101–108. Angst, U. (2018). “Battling infrastructure corrosion”. Keynote Lecture, 4th International Conference on Service Life Design for Infrastructures (SLD4), August 27–30, Delft, The Netherlands. Assié, S., Escadeillas, G. and Waller, V. (2007). “Estimates of self-compacting concrete ‘potential’ durability”. Constr. & Build. Mater., v21, 1909–1917. Barbhuiya, S.A., Jaya, A. and Basheer, P.A.M. (2011). “Influence of SCC on the effectiveness of controlled permeability formwork in improving properties of cover concrete”. The Indian Concr. J., v85, n2, February, 43–50. Basheer, P.A., Basheer, L., Bailie, R. & Nanukuttan, S.V. (2005). “An investigation into the performance of Formtex controlled permeability formwork and effects of its re-use”. Report, Queen’s University Belfast, UK, September 2005, 49 p.

@seismicisolation @seismicisolation

318 Concrete Permeability and Durability Performance Beushausen, H. and Alexander, M. (2009) “Application of durability indicators for quality control of concrete members – A practical example”. RILEM Conference ‘Concrete in Aggressive Aqueous Environments – Performance, Testing, and Modeling’, June 3–5, Toulouse, France. Bouwer, S. (1998). “Practical implementation of index tests for assessment and control of potential concrete durability”. MSc Thesis, University of Stellenbosh. Brühwiler, E. (2007). “Lifetime oriented composite concrete structures combining reinforced concrete with Ultra-High Performance Fibre Reinforced Concrete”. 3rd International Conference on Lifetime-Oriented Design Concepts, November 12–14, Ruhr-Universität Bochum, Germany, 10 p. BS 1881-204 (1988). British Standard “Testing concrete. Recommendations on the use of electromagnetic covermeters”. August, 14 p. CEN (2007). “Survey of national provisions for EN 206-1”. CEN/TC 104/SC 1, N 485, January 30, 148 p. COWI-Almoayed Gulf WLL (2002). “Application of Formtex controlled permeability formwork in the Arabian Gulf”. Report 201689-1, December 9, 48 p. Cullen, D.A. (1998). “Evaluation of the effects of Formtex CPF on the surface properties of concrete”. Taywood Engng. Ltd., Report 1304/98/10115, June 1998, 50 p. DBV (2015). “DBV-Merkblatt, Betondeckung und Bewehrung nach Eurocode 2”. DBV 2015-12 (in German). EN 13970 (2009). “Execution of concrete structures”. European Standard. EN 1992-1-1 (2004). “Eurocode 2: Design of concrete structures – Part 1-1: General rules and rules for buildings”. December. EN 206 (2013). “Concrete – Specification, performance, production and conformity”. European Standards, December, 93 p. ERMCO (2005). “The European guidelines for self-compacting concrete: Specification, production and use”. BIBM, Cembureau, EFCA, EFNARC and ERMCO, May, 68 p. Fehling, E., Schmidt, M., Teichmann, T., Bunje, K., Bornemann, R. and Middendorf, B. (2005). “Entwicklung, Dauerhaftigkeit und Berechnung Ultrahochfester Betone (UHPC)". Forschungsbericht DFG FE 497/1-1, Kassel University, Kassel, 132 p. Fernández Luco, L. (2009). “Importancia del curado en la calidad del hormigón de recubrimiento. Parte II: Métodos experimentales para identificar o prevenir el curado deficient”. Cemento Hormigón, n926, 30–41. Fernández Luco, L., Pombo, R. and Torrent, R. (2003). “Shrinkage compensating concrete in Argentina”. Concr. Intern., May, 49–53. fib (2010). “Model code 2010”. 1st Complete Draft, v1, March. Fornasier, G., Fava, C., Fernández Luco, L. and Zitzer, L. (2003). “Design of self compacting concrete for durability of prescriptive vs. performance-based specifications”. ACI SP 212, 197–210 Gjørv, O.E. (2014). Durability Design of Concrete Structures in Severe Environments. 2nd Ed., Taylor & Francis, UK, 254 p. Holčapek, O. (2011). “Influence of surface layer on the permeability of concrete”. JUNIORSTAV 2011, Brno, Czech Rep., February 4, 4 p. Jacobs, F., Hunkeler, F. and Mühlan (2018). “Prüfung und Bewertung der Betonqualität am Bauwerk”. Office Fédéral des Routes, Rapport No. 691, Bern, Switzerland, Juli, 106 p.

@seismicisolation @seismicisolation

Need for site assessment of durability 319 Kreijger, P.C. (1984). “The skin of concrete: Composition and properties”. Mater. & Struct., v17, n100, 275–283. Kunieda, M., Shimizu, K., Eguchi, T., Ueda, N. and Nakamura, H. (2011). “Fundamental properties of ultra high performance-strain hardening cementitious composites and usage for repair”. J. Japan Soc. Civil Engs., Ser. E2 (Materials and Concrete Structures), v67, n4, 508–521 (in Japanese). Kunieda, M., Choonghyun, K., Ueda, N. and Nakamura, H. (2012). “Recovery of Protective Performance of Crack Ultra High Performance-Strain Hardening Cementitious Composites (UHP-SHCC) Due to Autogenous Healing”. J. Adv. Concr. Technol., v10. Sept., 313–322. Law, D.W., Molyneaux, T., Patnaikuni, I. and Adam, A.A. (2012). “The site exposure of concrete cast using controlled permeability formwork”. Australian J. Civil Eng., v10, n2, 163–176. Leow, C-H. (2004). “Surface enhancements of concrete through use of controlled permeability formwork (CPF) liners”. International Conference on Bridge Engineering & Hydraulic Structure, Selangor, Malaysia, July 26–28. Lim, C.C. (2013). “An investigation into the deterioration of reinforced concrete jetties in Malaysia". CONSEC13, September 23–25, Nanjing, China. Malone, Ph.G. (1999). “Use of permeable formwork in placing and curing concrete”. US Army Corps of Eng., Techn. Re-port SL-99-12, October 1999, 53 p. Mayer, A. (1987). “The importance of the surface layer for the durability of concrete structures”. ACI SP-100, v1, 49–61. MC90 (1991). “CEB-FIP model code 1990 – Final draft”. Bulletin d‘Information CEB 203, Lausanne, July. Misono, M., Imamoto, K., Nagai, K. and Kiyohara, C. (2014). “Evaluation of moisture and gas permeability of surface treated concrete under accelerated weathering conditions towards conservation of reinforced concrete buildings”. International Workshop on Performance-based Specification and Control of Concrete Durability, Zagreb, Croatia, June 11–13, 361–368. Neves, R. and Santos, J.V. (2008). “Air permeability assessment in a reinforced concrete viaduct”. SACOMATIS 2008, September 1–2, Varenna, Italy. Neville, A. (1998). “Concrete cover to reinforcement — Or cover up?” Concr. Intern., v20, n11, November, 25–29. Newman, K. (1987). “Labcrete, realcrete, and hypocrete. Where we can expect the next major durability problems”. ACI SP-100, v2, 1259–1283. Oesterlee, C., Denarié, E. and Brühwiler, E. (2009). “Strength and deformability distribution in UHPFRC panels”, ConMat'09, Nagoya, Japan, 24–26 Aug. 2009, 390–397. Ohta, Y., Kiyohara, C., Shimozawa, K. and Imamoto, K. (2019). “Quality of cover concrete with controlled permeable formwork exposed outdoor condition for 17 years”. Annual convention of Architectural Institute of Japan: Kanto branch, (in Japanese). Polder, R. and Rooij, M. (2005). “Durability of marine concrete structures – Field investigations and modeling”. HERON, v50, 133–153. Quoc, P.H.D. and Kishi, T. (2006). “Measurement of air permeation property of cover concrete”. Proceedings on JSCE Annual Meeting, v61, Disk 2, 2 p. RILEM TC 189-NEC (2007). “Non-destructive evaluation of the penetrability and thickness of the concrete cover”. R. Torrent and L. Fernández Luco (Eds.), RILEM Report 40, 223 p.

@seismicisolation @seismicisolation

320 Concrete Permeability and Durability Performance RILEM TC 230-PSC (2016). “Performance-based specifications and control of concrete durability”. H. Beushausen and L. Fernández Luco (Eds.), RILEM Report 18, 373 p. SIA 262/1 (2019). “Concrete construction – Complementary specifications”. Swiss Society of Engineers and Architects. Tanaka, R., Habuchi, T., Amino, T. and f*ckute, T. (2012). “A study on improvement and its evaluation for the surface layer of concrete placed with permeable form”. Intern. J. Modern Physics: Conference Series, v6, 664–669. Torrent, R. (1999). “The gas-permeability of high-performance concretes: Site and laboratory tests”. ACI SP-186, Paper 17, 291–308. Torrent, R. (2018). “Bridge durability design after EN standards: Present and future”. Struct. & Infrastruct. Eng., v15, n5, 1-13. Torrent, R. and Frenzer, G. (1995). “Methoden zur Messung und Beurteilung der Kennwerte des Ueberdeckungsbetons auf der Baustelle -Teil II”. Office Fédéral des Routes, Rapport No. 516, Bern, Switzerland, October, 106 p. Torrent, R., Griesser, A., Moro, F. and Jacobs, F. (2012). “Technical-economical consequences of the use of Controlled Permeable Formwork”. ICRRR, Cape Town, South Africa, September 2–5. Wallbank, E.J. (1989). “The performance of concrete in bridges”. HMSO, London, April 1989, 96 p.

@seismicisolation @seismicisolation

Chapter 8

Why air-permeability kT as durability indicator?

8.1 INTRODUCTION The durability of concrete structures is closely dependent on the resistance of concrete to the penetration of aggressive agents (CO2 , chlorides, sulphates, etc.) as well as of substances that, although not being aggressive per se, play a role in the degradation process. Typical examples of the latter are water, the presence of which in sufficient quantities is required for most deterioration processes to progress (expansion of ASR gel, frost damage, steel corrosion, etc.) and O2 (also required for steel corrosion). As discussed in Chapter 3, the penetration of these substances happens by different mechanisms, through the interconnected network of pores present in the hardened cement paste and in the ITZ. In Chapter 3, the close connection between the coefficients ruling the different transport properties (permeability, sorptivity, diffusion, migration) was discussed. In addition, as treated in Chapter 7, what matters more in terms of durability is the resistance of the Covercrete to the penetration of external agents. All these concepts are superbly summarized in p. d-14 of CEB-FIP Model Code 1990 (CEB-FIP, 1991): “The durability of concrete is understood to be its resistance to physical and chemical attack such as frost or elevated temperatures, carbonation, sulphate attack etc. The resistance of concrete to such actions is governed primarily by its resistance to the ingress of aggressive media and thus by the capillary porosity of the hydrated cement paste as well as by entrapped air. A dense paste with a low capillary porosity is in most instances more durable than a paste with a high capillary porosity and a coarser pore system. There is no generally accepted method to characterize the pore structure of concrete and to relate it to its durability. However, several experimental investigations have indicated that concrete permeability both with respect to air and to water is an excellent measure for the resistance of concrete against the ingress of aggressive media in the gaseous or in the liquid state and thus is a measure of the potential durability of a particular concrete. DOI: 10.1201/9780429505652-8 @seismicisolation @seismicisolation

321

322 Concrete Permeability and Durability Performance

There are at present no generally accepted methods for a rapid determination of concrete permeability and of limiting values for the permeability of concrete exposed to different environmental conditions. However, it is likely that such methods will become available in the future allowing the classification of concrete on the basis of its permeability. Requirements for concrete permeability may then be postulated; they would depend on exposure classes i.e. environmental conditions to which the structure is exposed. Though concrete of a high strength class is in most instances more durable than concrete of a lower strength class, compressive strength per se is not a complete measure of concrete durability, because durability primarily depends on the properties of the surface layers of a concrete member which have only a limited effect on concrete compressive strength.” Regarding the third paragraph of the above quote, written around 1990, the situation at that time has changed radically and, nowadays, as discussed in Chapter 4, several rapid test methods to measure concrete permeability have been developed, some of them having been standardized and for which limiting values have even been established. The Torrent method to measure the coefficient of air-permeability kT is one of such standardized test methods; this method is especially suitable for non-destructive site testing of the end-product (the finished structure), its results reflecting the better or worse contribution of all the players along the concrete construction chain (owners, specifiers, materials producers, contractors, inspection). This is in line with the last paragraph of the text cited above. The rest of this chapter is devoted to presenting and discussing experimental data, obtained both in the laboratory and on site, with the aim of evaluating the credentials of kT as durability indicator. When dealing with the experimental data, it has to be borne in mind that when measuring kT, it is the Covercrete (see Chapter 7) which is being tested, affected not just by the mix design but also by factors such as placing, consolidation, segregation (including natural settlement and bleeding), curing, age and testing conditions (factors discussed in Chapter 6).

8.2 RESPONSE OF kT TO CHANGES IN KEY TECHNOLOGICAL PARAMETERS OF CONCRETE The effect of several key technological parameters of concrete (w/c ratio, compressive strength, curing, compaction/segregation, etc.) on the permeability of concrete was discussed in detail in Chapter 6. Chapter 6 presents abundant experimental evidence on the impact of such technological parameters on the permeability of concrete to water and gases, measured by several test methods, including kT, the one here discussed. @seismicisolation @seismicisolation

kT as durability indicator 323 Table 8.1 Sections dealing with the response of kT to key technological parameters Special Compaction, w/c & Binder Aggr. raw segregation Temperature Parameter f′c type type materials and bleeding Curing and moisture Stresses Cracks Section

6.2.2.3 6.3.2 6.4.1

6.5

6.6

6.7

6.8 & 6.9

6.10

6.11

Table 8.1 indicates the different Sections of Chapter 6 where the response of the coefficient of air-permeability kT to changes in different key technological parameters of concrete is discussed. The experimental evidence provided in the sections indicated in Table8.1 confirms that the coefficient of air-permeability kT is very sensitive to changes in w/c ratio and compressive strength of concrete as well as to the lack of moist curing at early ages. It also detects insufficient compaction and is negatively affected by excessive bleeding and by cracking, detecting even incipient ASR cracks. Compressive stresses up to 50% of the maximum load have virtually no effect on kT. The not so relevant effect of temperature (if not extremely low or high) and the strongly relevant effect of moisture on kT are discussed in detail in Sections 5.7.1 and 5.7.2, respectively. An approach to compensate kT for surface moisture content is presented in Section 5.7.2.2. It can be concluded that kT is a good indicator for variations in quality or condition of the Covercrete due to the factors listed in Table 8.1. 8.3 CORRELATION WITH OTHER DURABILITY TESTS Being a good indicator for variations in Covercrete quality is not enough for kT to qualify as a suitable durability indicator, because it must be proved that kT is also related to the resistance of concrete to the penetration of species by the different transport mechanisms discussed in Chapter 3. In Chapter 3, the close relation between the pore structure of concrete and its permeability was demonstrated in theoretical terms, confirmed experimentally for the case of air-permeability kT in Section 5.5.2. Since other transport mechanisms (sorptivity, diffusion) are also related, albeit differently, to the pore structure (Section 3.8), some relations between the transport parameters are expected, as theoretically formulated in Section 3.9. In the rest of this section, experimental evidence on the relations existing between the coefficient of air-permeability kT and other transport parameters is presented and discussed, to assess its potential as sound durability indicator. This experimental evidence has been obtained from worldwide series of data sets reported in the literature or accessible to the authors, as detailed in Table 8.2. The relations between variables, presented in this chapter, have been established on the basis of the data reported in the documents listed in Table 8.2, with reference to the corresponding data sets. Please bear in @seismicisolation @seismicisolation

324 Concrete Permeability and Durability Performance

mind that the 47 sets of data were generated by many laboratories across the globe (17 countries involved), applying different instruments, personnel and testing protocols. The main test methods (mostly standardized) mentioned in this chapter (described in Chapter 4 and Annex A) are identified by the following symbols: a24 = coefficient of water capillary absorption at 24 hours ao = coefficient of water capillary absorption at unknown ages aK = Karsten tube water sorptivity test CCI = chloride-conductivity Index, South African method DP = coefficient of chloride-diffusion by ponding (AASHTO T259) DI = coefficient of chloride-diffusion by immersion (ASTM C1556 or similar) DS = coefficient of chloride-diffusion obtained on site δm = mass loss after 30 freeze/thaw/salts cycles (SIA 262/1) Fa = Figg air-permeability test Fw = Figg water sorptivity test Kc = natural carbonation in laboratory or on site K′c = natural carbonation from acceleratedtest

kO = oxygen-permeability test, Cembureau method kOPI = coefficient of oxygen-permeability, OPI (oxygen-permeability index) index, South African method kT = air-permeability test (Torrent) using second or later generation instruments kT3 = air-permeability using first generation prototype, kT = 1.846 * kT3 M = coefficient of chloride-migration (Tang-Nilsson method) N50 = number of freeze/thaw cycles for a reduction of 50% in E-modulus OD = coefficient of oxygen-diffusion (Do) Q = charge passed in Rapid Chloride Permeability Test (ASTM C1202) ρs = surface electrical resistivity (Wenner) SCl = site chlorides T = TUD oxygen-permeability test WP = maximum penetration of water under pressure (EN 12390-8 or DIN 1048)

8.3.1 Gas Permeability 8.3.1.1 Cembureau Test Figure 8.1 shows the results of parallel measurements of the coefficients of air-permeability kT and of oxygen-permeability kO from 11 data sets (described in Table 8.2), involving 141 pairs of data. A very good correlation (R = 0.89) exists between both coefficients of gas-permeability, through the following regression, shown in Figure 8.1 as a full line: The dotted line in Figure 8.1 shows the equivalence line (Y = X), for the case in which the results had been the same for both methods. It is clear that both test methods yield well-correlated but significantly different results. The kO results span four orders of magnitude, whilst kT results span five @seismicisolation @seismicisolation

kT as durability indicator 325 Table 8.2 Details on data sets used in this chapter and their sources Data set 1

Source

Brief description

Torrent and kT3, kO, Ebensperger a24, T, (1993); see Fa, Fw Section 6.2.1.1 kT3, kO, a24, DP, δm, Fa, Fw, T, Kc

Applied Country tests code

Torrent and Frenzer (1994); Section 6.2.1.1 Torrent and Frenzer (1995); Sections 6.2.1.1 and 6.2.1.2

CH Table 3.1-III reports five different concretes mixes that, after 0, 7, or 28 days moist curing were stored in a dry room (20°C/50% RH) till the moment of test (age ≈ 1.5 years). Table 3.2-IV reports ten mixes, moist cured 0 or 7 days, followed by storage in the same dry room until the age of test (28 days). Prisms made with four concrete mixes, subjected to 28 days moist curing, were exposed to an ambient of 20°C/57% RH for 2 years. Table 4-III reports three mixes, moist cured kT3, ρs, 90 days, thereafter kept in a dry room kO, a24, (20°C/50% RH), measuring kT and ρs, and kO Fa, Fw, T, and a24 on cores drilled after 0, 19, 68 and 111 days drying. kT, kO, Table III reports durability performance data of three Mexican cements, same procedure as for a24 data set 2.

kT, kO, a24

5 6

kT, kO, a24, DP, δm

kT, kO, a24, N50, Kc, K′c

Table 1.2.1.1 reports data of lab specimens cast from nine batches of ready mixed concrete to be placed in Bözberg Tunnel (Section 7.1.5.1). The specimens were moist cured 7 days, followed by storage in dry room (20°C/50% RH) until the age of test (28 days). Table 1.2.1.2 reports measurements made on the deck of the Tunnel in correspondence with some of the mixes tested in data set 3. Tables 2.2.2.1, 2.2.4.2, 2.2.4.3 report data of lab specimens cured and stored like for data set 3 and of 1 m cubes that were field-cured and tested at ages within 28–35 days. Two concrete qualities from Schaffhausen Bridge (Section 7.1.5.2) were investigated. Table 3.2.1.1 and p. 95 report data of lab cubes (0.5 m), moist cured for 0 and 7 days and stored in the laboratory. The experiment is described in Section 5.6.2. From certain locations where kT was measured, specimens were drilled and saw-cut for measuring other properties. Prisms made with several concrete mixes, subjected to different curing conditions, exposed to an ambient of 20°C/50% RH for 2 years. (Continued)

@seismicisolation @seismicisolation

326 Concrete Permeability and Durability Performance Table 8.2 (Continued) Details on data sets used in this chapter and their sources Data set

Source

Applied Country tests code

Kattar et al. (1995)

kT3, Q, WPD

Kattar et al. (1999)

kT3, Q

Roelfstra kT, D etal.(1999)

Andrade kT, kO, SP, CH etal.(2000) a24, Q, Kc

Mohr kT, Q, ao etal.(2000)

13 14

Denarié kT, WP etal.(2003)

Fornasier kT, WP, etal.(2003) a24, Q, DI Kubens kT, Q, etal.(2003) K′c

Romer and Leemann (2005)

kT, kO, kOPI, CCI

Brief description Data of a conventional concrete made with a triple-blend cement (containing slag and silica fume) and to four mixes where different vinyl-based polymers were added (w/c ratio between 0.33 and 0.48). Data of three mixes of 26, 30 and 36 MPa cylinder strength at 28 days. The specimens were moist cured up to 25 days of age, to be tested at 28 days of age. Modelling chloride-induced corrosion in reinforced concrete. For the modelling, three Covercrete classes are defined as function of the kT (0.1, 1.0 and 10. 10−16 m²).Values of water and chloride-diffusion coefficients are associated with those classes. Round robin tests within RILEM TC 116-PCD, consisting in testing Ø150 × 50 mm discs of five concrete mixes: four binders, w/b = 0.4 and 0.7, sealed cured 3 and 7 days and pre-conditioned at 50°C. Tests on saw-cut slices of cores drilled from old concrete pavements with compressive strengths in the range of 40–80 MPa. kT measured on panels made with OPC mixes with w/c ratios of 0.41 and 0.59. Different finishing techniques of the top surface and a “Zemdrain” permeable formwork liner were applied. WP measured on cores drilled from the top and lateral surfaces. WP of set 13 by DIN test and of set 14 by EN test method. Results obtained at Loma Negra laboratory in Argentina on three concretes, two of them self-compacting with w/c between 0.40 and 0.44. Cubes and prisms made with two concrete mixes, subjected to different curing conditions, stored in an ambient of 30°C/40% RH for 90 days. Companion specimens were also exposed to accelerated carbonation: 5% CO2, 30°C/50% RH for 7 days. 200 mm cubes from six concretes of w/c ratios between 0.35 and 0.62 (OPC + limestone filler cement) stored at 35%, 70% and 90% RH (20°C). Direct measurements of kT followed by measurements of kOPI and CCI on drilled cores were made at 365 days. kT values at 90% RH not considered due to high moisture content. (Continued)

@seismicisolation @seismicisolation

kT as durability indicator 327 Table 8.2 (Continued) Details on data sets used in this chapter and their sources Data set 18 19

Source

Applied Country tests code

Fernández kT, WP andRevuelta (2005) Rodríguez de kT, WP Sensale etal.(2005)

Mathur kT, Q etal.(2005)

Jacobs (2006) kT, a24, Kc RILEM TC kT, kO, 189-NEC kOPI, (2007) a24, Q, M, CCI

Di Pace and kT, WP Calo (2008)

Imamoto kT, Kc etal.(2008)

Jacobs (2008) kT, a0, Kc, SCl

Van Eijk (2009) kT, WP, ρs Teruzzi kT, SC (2009)

Jornet kT, a0, etal.(2011) M

Brief description

Tests on pozzolanic cement concretes with w/c = 0.40 and 0.45, intended for a LNG tank in México. UY Tests on self-compacting concretes (28 days cylinder compressive strength within 35 and 66 MPa), with and without electrofilter powder as filler. IN Reference and high-volume fly-ash concretes (30%–50% of PFA) of three w/b ratios between 0.33 and 0.65 were tested for durability by the Rapid Chloride Permeability Test (ASTM C1202) and kT. CH Comprehensive survey of data from old concrete structures. CH, PT, Tables B.1–B.5 report results of site kT tests ZA made on ten panels made with OPC and GBFS mixes of different w/c, curing and testing conditions and lab tests on drilled and saw-cut specimens. Several labs involved. AR Report site (kT) and laboratory (kT, WPD) test results of concrete of Buenos Aires Metro. Improved mix design and construction practices led to an enhanced water tightness of the underground construction. JP Exposed panels made with three concrete mixes, subjected to two different curing conditions, to an ambient of 20°C/60% RH for 3.5 years. CH Site kT on 12 bridges 30 years old, ranging 0.001–10 (10−16 m²). No correlation kT vs 28 days cube strength. Good correlation kT vs cores’ sorptivity. Carbonation and chlorides generally higher for higher kT. NL Experiment described in Section 6.2.1.5 CH 55 parallel measurements of kT on site and carbonation depth on a building in Canton Ticino, Switzerland. Model to predict carbonation rate from kT. CH Performance comparison of concretes made with limestone filler cement and OPC; w/c = 0.40–0.60. Different curing conditions applied. Durability tests and petrographic analysis. (Continued)

@seismicisolation @seismicisolation

328 Concrete Permeability and Durability Performance Table 8.2 (Continued) Details on data sets used in this chapter and their sources Data set

Source

Applied Country tests code

Zhutowsky and Kovler (2012)

Imamoto kT, Kc etal.(2012)

Neves (2012) kT, Kc

Starck (2013), kT, kOPI Starck et al. (2017); Section 6.2.1.4 Imamoto kT, Kc, etal.(2014) K′c

34 35 36 37

PC (2014)

kT, Q, ao

kT, Q, ρs

Imamoto kT, Kc etal.(2016) Maître (2012), Jacobs (2006), Bisschop etal.(2016) Moro and Torrent (2016)

kT, SCl

kT, kO, a24, Q, M, OD, WP

Beglarigale kT, a24, etal. (2014) Q, ρs

Bahurudeen and Santhanam (2014)

kT, kOPI, Q, CCI, WP

Brief description

Comparative effect of internal curing by inclusion of pre-saturated lightweight aggregates on mechanical and durability performance of HPC, w/c between 0.21 and 0.33. JP One-hundred and eleven parallel ND measurements of kT and cover thickness d on Tokyo’s Museum of Western Art. Few small cores to establish regression kT vs SC. Model to predict service life from kT → Kc and d PT Comprehensive site investigation of relation between kT and Kc. ZA Cubes made with six different concretes, exposed to three different conditions, were tested for kT and, later for kOPI on cores drilled and cut from the cubes and oven-dried at 50°C for 7 days. JP Reports parallel measurements of carbonation and kT on old structures and on new concretes measured in the laboratory (accelerated carbonation). PA Values measured at 28, 56 and 90 days during quality control of concretes for the Panama Canal, Pacific side. Ibid, Atlantic Side JP, PT, Compilation of parallel site tests of kT and CH carbonation measured on 14 concrete structures in Japan, Portugal and Switzerland. CH Investigation fully described in Section 11.2.2.

CH Data from investigation described in Section 6.3.2. Both kT and other properties measured on drilled + cut specimens.Values obtained after 28 days (some also at 91 days) of moist curing are reported here. TR Study on the effect of replacing 15% and 30% of OPC by high Ca fly ash on various transport properties of concrete; the reported Wp data are the mean values. IN Study on the effect of replacing 10%, 15% and 20% of OPC by sugar cane bagasse ash on various transport properties of concrete. (Continued)

@seismicisolation @seismicisolation

kT as durability indicator 329 Table 8.2 (Continued) Details on data sets used in this chapter and their sources Data set

Source

Applied Country tests code

Neves kT, K′c etal.(2015)

Nishimura kT, K′c etal. (2015)

Wang kT, ρs etal.(2014)

Kurashige and kT, K′c Hironaga (2010) Ebensperger kT, Q, and Olivares M, WP, (2019) ρs

(Park etal.,2004)

kT, DI,

(Akiyama etal.,2010)

kT, K’c

Brief description

Parallel measurements of kT and accelerated carbonation (20°C/65% RH, 5% CO2) of concretes (w/c = 0.39–0.70) made with OPC, LF cement and PFA cement. JP Relation between kT and accelerated carbonation rate (20°C/60% RH, 5% CO2) of concretes with w/b = 0.40, 0.50 and 0.60, made with OPC and with binders containing 30%, 45% and 60% of blastfurnace slag, subjected to three different curing conditions. Other factors were also studied. CN Site measurements of cover thickness, kT and Wenner electrical resistivity in precast segments for submerged tunnel of Hong Kong–Zhuhai–Macao link. JP Effect of several curing conditions on kT and accelerated carbonation (20°C/60% RH, 5% CO2) of OPC concrete with w/c = 0.50 CL Effect of w/c ratio (0.40; 0.50; 0.60; 0.70), curing and age of concretes made with High Strength Pozzolanic Cement on transport properties. The data included in the charts correspond to concretes moist cured 28 days, later stored in a dry room (23°C/50% RH) and tested at 28 and 91 days of age (taken from table in p. 14). In the case of kT, the specimens were oven-dried 3 days at 50°C prior to testing. KR Effect of binder type (plain OPC and with addition of PFA, GGBFS and SF) on w/b=0.55 concretes tested for kT and chloride-diffusion DI (immersion in 3.6% NaCl solution). Specimens cured 28 days in water (kT cubes dried 48 h at 60% RH). JP kT and accelerated carbonation (20°C/60% RH, 5% CO2) of 12 concretes mixed with various binders and w/c ratios, subjected to 3 curing conditions.

orders of magnitude. Up to a kT value of ≈2 × 10 −16 m² the kT values are smaller than the kO values, situation that is reversed above that value. Thereasons for this discrepancy are varied, among them: • the different preconditioning of some samples: the moisture content of the kT slabs, especially for low-permeability concretes, was higher than for the kO drilled and oven-dried discs • the different volumes explored by the tests: Ø150 × 50 mm for kO and Ø50 × L mm for kT (L is the penetration depth of the test which, for low-permeability concretes may reach only few mm) @seismicisolation @seismicisolation

330 Concrete Permeability and Durability Performance

Figure 8.1 Relation between Cembureau oxygen-permeability kO and kT.

• the “wall-effect”; if kO is measured on cast discs or discs cut of longer cylinders, the surface layers close to the curved walls are richer in paste and may conduct more gas flow than the core. This effect is absent in the kT tests • steady-state conditions for kO against non steady-state for kT leading to simplifications in the model under which kT is calculated • the Klinkenberg effect (Section 3.6). kO is measured under positive relative pressures, whilst kT under negative relative pressures • the default value of 0.15 for the porosity ε of the concrete. Concretes of low permeability have also lower porosities than 0.15 (see Figure5.8) which, according to the correction in Eq. (5.43), would yield higher kT values than those reported. Similarly, concretes of high permeability would yield lower kT values, making the correlation line rotate towards the “Equality” line 8.3.1.2 South-African OPI A very comprehensive research purposely looking for a possible correlation between kOPI and kT was conducted at the University of Cape Town (Beushausen et al., 2012; Starck, 2013; Starck et al., 2017), data set 32 in Table 8.2, investigation described in detail in Section 6.2.1.4. Figure 8.2 presents Starck’s data as black dots, which yield an excellent correlation (R = 0.97) for the regression line shown in Figure 8.2: Data sets 17, 22 and 40 (Table 8.2) show a different trend (white symbols in Figure 8.2). The reason for this discrepancy is attributed (Starck, 2013) @seismicisolation @seismicisolation

kT as durability indicator 331

Figure 8.2 Correlation between kOPI (South African oxygen-permeability index) and kT.

to the different experimental conditions under which data sets 17 and 22 were obtained. 8.3.1.3 Figg Air and TUD Permeability Figg air-permeability test method was described in Section 4.3.2.1; although not standardized, the method has achieved certain acceptance. Figure 8.3a shows an excellent (negative) correlation found between Figg’s air time (required to raise the relative pressure from −0.45 to −0.35 bar in an evacuated hole) and kT, based on test results from data sets 1 and 2 (upper face and bottom face of slabs as cast). TUD method, described in Section 4.3.2.4, works on similar principles as Figg. Figure 8.3b shows the excellent correlation between TUD time (required for a decay of the pressure in a pressurized hole from 11.0 to 10.5 bar) and kT for sets 1 and 2.

Figure 8.3 Correlation between Figg air (a) and TUD (b) test time and kT.

@seismicisolation @seismicisolation

332 Concrete Permeability and Durability Performance

8.3.2 Oxygen-Diffusivity Parallel results of the coefficient of oxygen-diffusion DO and kT were obtained in the investigation reported by Moro and Torrent (2015), data set 38, plotted in Figure 8.4. The values corresponding to nine different binder types are coded according to Table 5.4. The kT values were measured at Holcim Technology laboratory whilst DO was measured at EMPA, by the test method described in A.1.1.1. Results obtained after 28 days and 1 year of moist curing are plotted indistinctly, constituting N = 36 pairs of data (measurements were conducted after preconditioning the specimens at 50°C). Despite the variety of binders and ages, a very good correlation (R = 0.89) was found between DO and kT for the regression: It is interesting to note the closeness of the exponent in Eq. (8.3) to the theoretical one (0.50), as per Eq. (3.62).

8.3.3 Capillary Suction 8.3.3.1 C oefficient of Water Absorption at 24 Hours Figure 8.5 presents parallel test results of the coefficient of water absorption at 24 hours a24 and kT, for ten data sets, as described in Table 8.2. Considering the disparity of the sources and test procedures, a reasonably good correlation (R = 0.73) was obtained for the N = 133 pairs of data, according to the regression: a24 = 12 + 1.6.ln kT

(

)

(

a24 g m 2 s 2 ; kT 10−16 m 2

Figure 8.4 Correlation between oxygen-diffusivity DO and kT.

@seismicisolation @seismicisolation

)

(8.4)

kT as durability indicator 333

Figure 8.5 Correlation between water sorptivity a24 and kT.

8.3.3.2 Figg Water Similar to Section 8.3.1.3, parallel results of kT and Figg test, this time Figg-Water test (Section 4.2.2.3), are plotted in Figure 8.6a. The result of the Figg-Water test is the time required for the concrete to absorb 0.01 mL of water. A good (negative) correlation is observed between the results of both tests. 8.3.3.3 Karsten Tube The Karsten tube test (Section 4.2.2.2) was also applied to some of the specimens reported in Sections 8.3.1.3 and 8.3.3.2, as shown in Figure8.6b. A good correlation between the rate of water absorption by Karsten and kT exists which, incidentally, fits well to the regression of Eq. (8.4), also plotted in Figure 8.6b.

Figure 8.6 Correlation of kT with (a) Figg time to absorb 0.01 mL of water and (b) Karsten tube absorption rate.

@seismicisolation @seismicisolation

334 Concrete Permeability and Durability Performance

8.3.4 Water-Permeability and Penetration under Pressure A very good correlation between the coefficients of water-permeability Kw and of air-permeability kT was reported by Sakai et al. (2013), as shown in Figure 3.12. The regression equation indicated in the figure or Eq. (2.1) can be used to convert kT experimental data into equivalent Kw estimated values. More abundant are experimental data relating kT and Water Penetration under Pressure tests (see Section 4.1.1.2). Figure 8.7 presents N = 96 parallel measurements of the maximum penetration of water under pressure WPmax and kT, from eight data sets described in Table 8.2, following two similar test methods: DIN 1048 (1978) and EN 12390-8 (2009). Despite the huge diversity of laboratories, countries and WPmax test methods involved, a quite good correlation (R = 0.73) is obtained according to the regression line also shown in Figure 8.7: The dotted-lined boxes shown in Figure 8.7 correspond to equivalent permeability classes, associated with both test methods (Tables 4.1 and 5.2). The fact that the regression line crosses the boxes near their intersection points indicates that both test methods tend to judge the quality of the Covercrete quite coherently.

8.3.5 Migration In this section, the relation between the output of several migration tests and kT is presented. It has to be borne in mind that kT depends exclusively on the pore structure, whilst the results of migration tests also depend

Figure 8.7 Correlation between maximum water penetration under pressure WPmax and kT.

@seismicisolation @seismicisolation

kT as durability indicator 335

strongly on the ionic composition of the pore solution, with or without the incorporation of foreign ions from outside (typically Cl− ions)., Section 3.4.2 8.3.5.1 Rapid Chloride Permeability Test (“RCPT” ASTM C1202) Figure 8.8 shows N = 114 parallel results of the current passed Q (coulomb) in the “RCPT” ASTM C1202 and kT, reported in 14 data sets of Table 8.2. A good correlation (R = 0.77) exists between both variables according to the regression (full line in Figure 8.8): The dotted line boxes in Figure 8.8 represent areas where both test methods judge the permeability of the concrete to chlorides (Q) and to air (kT) as very low (VL), low (L), moderate (M) and high (H). The fact that the regression line intersects the boxes almost exactly at the crossing points indicates that both test methods tend to judge the quality of the Covercrete quite coherently. The large scatter observed is to be expected, since kT measures just the “openness” of the pore structure whilst Q (same as other migration tests) is also strongly affected by the ionic composition (particularly by the content of OH− ions) of the pore solution (Andrade, 1993; Shi, 2003). 8.3.5.2 Coefficient of Chloride Migration (NT Build 492) This migration test, that can be considered as an improvement over ASTM C1202 “RCPT”, is the Chloride Migration Test, developed by Tang and Nilsson (1992), standardized in Scandinavia (NT Build 492), in Switzerland

Figure 8.8 Correlation between electric charge Q passed in ASTM C1202 “RCPT” test and kT.

@seismicisolation @seismicisolation

336 Concrete Permeability and Durability Performance

(SIA 262-1, Annex B) and lately in Europe (EN 12390-18). The test method is described in Annex A.2.1.2 giving as test result a coefficient of chloride migration M which, in theory (Section 3.4.2), should be equal to the coefficient of chloride-diffusion D. Figure 8.9 presents N = 60 parallel results of M and kT reported in data sets 22, 28, 45 and 38, all, but in particular the latter, covering a wide range of binder types (Section 6.3.2). A very good correlation (R = 0.83) exists between both variables, according to the regression (plotted in Figure 8.9 as full line): Just as a reference, a dotted line is added in Figure 8.9, corresponding to the relation between the coefficient of diffusion D and kT, established completely independently in Section 8.3.6.1, Eq. (8.9). It can be seen that the values of M, predicted from Eq. (8.7), are 2–3 times higher than the D values predicted by Eq. (8.9), for the same kT, except for concretes of very low permeability (kT < 0.01 × 10 −16 m²). This is in agreement with the findings of Li et al. (2015), who recommend design values of M that are twice those corresponding to D, confirmed experimentally by Ren et al. (2021). Again, the exponent in Eq. (8.7) is close to the theoretical 0.50. 8.3.5.3 Electrical Resistivity (Wenner Method) There are many parallel results of electrical resistivity (Wenner method) ρ and kT, particularly from early investigations (data sets 1–7), when ρ was used as estimator of the moisture content of the concrete in order to compensate kT results for its effect (see Section 5.7.2). However, these results

Figure 8.9 Correlation between chloride migration M (Tang-Nilsson test) and kT.

@seismicisolation @seismicisolation

kT as durability indicator 337

are misleading in establishing a relation between both variables because, in order for ρ to act as durability indicator, it should be applied on saturated specimens (opposite to kT, applicable on rather dry concrete). Another factor worth considering when comparing data of ρ and kT is that the former is strongly influenced by the composition of the electrolyte (pore solution), as discussed in Section 3.4.2, whilst kT, being governed by the pore structure of the concrete, is not. Figure 8.10 shows kT results from data set 26 reported by Van Eijk (2009), measured on panels as described in Section 6.2.1.5 and of ρ, measured on cubes cured under water for 14 days and then stored at 65% RH until the age of test (56 days). If ρ could be measured also in the panels, the results are not reported. The panels and cubes were cast with mixes of w/c = 0.40 and 0.57 made with two different cements: an OPC (CEM I) and a GBFS cement (CEM III). It can be seen that kT judges the mixes in decreasing order of quality (from left to right in Figure 8.10: 1, 3, 4, 2) whilst ρ in decreasing order of quality 3, 4, 1, 2; i.e. kT gives precedence to the w/c ratio and ρ to the cement type. In Van Eijk (2009), it is stated (translation from Dutch) “It is known that measurement results of the Wenner Probe are sensitive to the type of cement: Portland cement or GGBFS cement. The absolute results must therefore always be compared with values measured with the same type of cement. Measurement values for GGBFS cement are not directly comparable with those for Portland cement.” 8.3.5.4 South African Chloride Conductivity Index The reciprocal of ρ is the electrical conductivity, typically expressed in Siemens (1/Ω) per m. Some parallel data exist between the South African Chloride Conductivity Index CCI (described in Section A.2.2.3) and kT,

Figure 8.10 Values of kT and ρ for the four mixes tested; data from Van Eijk (2009).

@seismicisolation @seismicisolation

338 Concrete Permeability and Durability Performance

Figure 8.11 Correlation between South African Chloride Conductivity Index CCI and kT.

data sets 17, 22 and 40. They are plotted in Figure 8.11, showing that, for each data set, CCI increases with kT, as expected, but following different trends for the three data sets. This may be due to the different concretes (binders) tested and/or to different testing conditions.

8.3.6 Chloride-Diffusion 8.3.6.1 Laboratory Diffusion Tests Few Relatively few results have been found in the literature, reporting parallel measurements of the coefficient of chloride-diffusion D and of airpermeability kT. This may be due, in part, to the duration (months) of the existing standardized tests to measure D, be it by immersion (ASTM C1556) or by ponding (AASHTO T259), coupled to the high costs associated with performing a chloride profile analysis of the cores. Both test methods are described in Section A.1.2. Nowadays, the D tests are being gradually replaced by migration tests, ASTM C1202 or NT Build 492, although the former requires a calibration with long-term immersion tests for validation. The parallel experimental results of D and kT found in the literature correspond to data sets 2 and 6 (N = 14), set 15 (N = 3) and set 46 (N = 34) and are plotted as black symbols in Figure 8.12. The broken line in Figure 8.12 plots the values reported as data set 10 which correspond, strictly speaking, to relations between D and kT proposed in a service life model developed at EPFL (Roelfstra et al., 1999), see Table 9.4. The model derives values of D (and of the water diffusion coefficient as well) on the basis of kT measurements and is described in Section 9.4.1. In a very comprehensive research, conducted in the USA, (Olek et al. (2002) established a correlation between the coefficient of chloride-diffusion @seismicisolation @seismicisolation

kT as durability indicator 339

Figure 8.12 Correlation between coefficient of chloride-diffusion D and kT.

D (AASHTO T259) and the charge Q passed in ASTM C1202 “RCPT” of the form: A validation of Eq. (8.8) was made using data of D (ASTM C1556) and Q (ASTM C1202) independently obtained by Alexander and Thomas (2015). The agreement is very good, especially within the usual range of Q values (100–10,000 Coulomb). This allows, by applying Eq. (8.8), converting all the Q values in Figure 8.8 into D values; the resulting values are plotted as white circles in Figure 8.12. It can be seen that the converted values (white circles) merge quite well with the experimental results (black symbols) constituting a total sample of N = 171 pairs of data. A general relation (shown as full line in Figure 8.12) can be established between the coefficient of chloride-diffusion D and of air-permeability kT of the form: The standard error for the estimation of D is ≈ 3.5 × 10 −12 m²/s, with a determination coefficient R = 0.80. Figure 8.12 shows that the scatter in the area of interest for durability design (kT < 0.1 × 10 −16 m²) is smaller than for the whole set of data, especially when considering the results involving direct measurements of D (black symbols). An independent regression obtained by Park and Kim (2000) on 45 parallel measurements of D @seismicisolation @seismicisolation

340 Concrete Permeability and Durability Performance

(immersion) and kT, for OPC concretes, yielded almost the same exponent (0.328) as in Eq. (8.9) but with a factor of 19.3 instead of 10. Figure 8.9 also confirms that the relation between D and kT, expressed by Eq. (8.9), is supported by the migration M test results (known to be higher than D). The relationship expressed by Eq. (8.9) is used in the “ExpRef” Model for service life assessment for chloride-induced steel corrosion, presented in Section 9.4.3.1. 8.3.6.2 Site Chloride Ingress in Old Structures Parallel measurements of chloride ingress and air-permeability kT, obtained on site, are scarce. Figure 8.13 shows results from data set 25, plotting the chloride content at the level of the reinforcement and kT, both measured on bridges about 30-years old located along a Swiss Motorway. It can be seen that for kT values below 0.1 × 10 −16 m², the chloride content has not yet reached the critical level of 0.6% of the cement weight (sometimes a more conservative value of 0.4% is adopted) expected to initiate corrosion. But a bridge should last at least 75 years, so one of the six bridges with kT < 0.1 × 10 −16 m² is not on the safe side. Other results showing the relation between chloride ingress rate and kT can be found in Section 11.2.2 (especially Figure 11.8b).

8.3.7 Carbonation 8.3.7.1 Laboratory Tests (Natural Carbonation) Data sets 2, 7, 11, 16 and 24 report parallel data of kT and the carbonation rate Kc measured after at least 1 year of exposure to a dry room, where

Figure 8.13 Relation between chloride content at steel bar level and kT for 30 years old Swiss bridges; data from Jacobs (2008).

@seismicisolation @seismicisolation

kT as durability indicator 341

carbonation proceeds at maximum rates in natural air. The concretes tested were made with OPC, GBFSC, FAC, SFC and the storage conditions and duration varied. For sets 2 and 7 it was 20°C/50% RH during 210 days and 2 years storage, respectively. For set 16, it was 90 days at 30°C/40% RH and for set 24 specimens were stored 2 years at 20°C/60% RH and thereafter 2 years indoors in the laboratory. For Set 11, it was 1 year under undisclosed conditions. In all cases, kT was measured before the specimens were stored in the dry rooms and Kc was calculated as the carbonation depth (measured by phenolphthalein method) divided by the square root of the exposure time; the results are shown in Figure 8.14. Despite the different storage conditions and binder types, there is a good agreement among all sets (perhaps less for Set 11) and a good overall correlation (R = 0.82) between the variables, according to the regression (shown in Figure 8.14):  kT  Kc = 1.67ln   0.006 

(

Kc mm y

) ; kT (10

−16

)

m2 ;

valid for kT ≥ 0.006 × 10−16 m 2

(8.10)

The results in Figure 8.14 show that for kT below 0.006 × 10 −16 m², the carbonation rate becomes negligible. The relationship expressed by Eq.(8.10) is used in the “Exp-Ref” Model for service life assessment for carbonation-induced steel corrosion, presented in Section 9.4.3.2. 8.3.7.2 Laboratory Tests (Accelerated Carbonation) Data sets 2, 7, 16, 41, 42, 44 and 47 report parallel data of kT and accelerated carbonation tests. In the latter, concrete specimens are exposed to environments of controlled temperature and (typically low) relative humidity,

Figure 8.14 Correlation carbonation rate Kc vs kT for natural carbonation under laboratory conditions.

@seismicisolation @seismicisolation

342 Concrete Permeability and Durability Performance

enriched in CO2 concentration. The natural concentration of CO2 in air is ≈ 0.04%; in some tests (Sets 16, 41, 42, 44 and 47) the specimens were stored, during different periods, in a chamber the CO2 concentration of which was kept at 5%, whilst in others (Sets 2 and 7) the specimens were kept, for different periods, at a CO2 concentration of 90% and 100%, respectively. It is worth mentioning that some standardized test methods for accelerated carbonation specify a 3% CO2 environment (ISO 1920-12, 2015) or a 4% CO2 environment (Annex I of SIA 262/1 (2019)). To analyze the accelerated carbonation test results obtained under such disparate conditions (time and CO2 concentration), Eq. (8.11) will be used, taken from Annex I of SIA 262/1 (2019). Kc' = 1.36⋅

0.04 ⋅ Ka CO2

(8.11)

where Kc' = equivalent natural carbonation rate CO2 = CO2 concentration (%) in the test chamber K a = accelerated carbonation rate, equal to the measured carbonation depth divided by the square root of the time of permanence in the accelerated test chamber Figure 8.15a shows the results of K’c , calculated with Eq. (8.11), as function of the kT values measured before introducing the specimens in the accelerated carbonation test chamber; the regression line corresponding to Eq.(8.12) is also plotted (full line), with a coefficient of correlation (R = 0.75), lower than for the natural carbonation tests, Eq. (8.10). This can be due to the widely different test methods, especially of the CO2 concentrations in the test chambers, but also because different cement types were tested.

Figure 8.15 Correlation natural carbonation rate K′c vs kT from accelerated carbonation tests; (a) data from several sources; (b) data from Nishimura et al. (2015).

@seismicisolation @seismicisolation

kT as durability indicator 343

 kT  K′c = 1.99 ⋅ ln   0.0096 

(

Kc′ mm y

) ; kT (10

valid for kT ≥ 0.0096 × 10−16 m 2

−16

)

m2 ; (8.12)

In Figure 8.15, the dotted line corresponds to Eq. (8.10); the similarity between Eqs. (8.12) and (8.10) is remarkable. Figure 8.15b presents the results of data set 42, covering four orders of magnitude of kT values, obtained through changes in w/b ratio, in cement type (binders containing 0%, 30%, 45% and 60% of blast-furnace slag) and in curing conditions. A tenuous trend, not clearly defined, of higher values of K′c for the same kT can be detected for slag-containing binders. The research corresponding to Data Set 47 (Akiyama et al., 2010) showed a strong influence of the curing conditions (moist, sealed, wind) on the relation between K’c and kT. 8.3.7.3 Site Carbonation in Old Structures This topic will be dealt with in more detail in Section 9.5. Here suffice to say that a large number of parallel tests of kT and carbonation depth CD were measured (data set 36) on several old structures located in Japan, Portugal and Switzerland. After measuring kT, cores were drilled at the same spots to measure CD by the phenolphthalein method. The carbonation rate is computed as CR = CD/age½. The results obtained in Switzerland, Japan and Portugal are shown in Figure 8.16.

Figure 8.16 Relation carbonation rate vs. kT in old Swiss, Japanese and Portuguese structures.

@seismicisolation @seismicisolation

344 Concrete Permeability and Durability Performance

In some of the structures, the values of kT measured on a single structure span 5 and 4 Permeability Classes (Imamoto et al., 2016). This might be attributed to the test method; however, the carbonation rates also cover a wide range, from nearly 0 up to values exceeding 5 mm/y½. That means that the high variability in kT and CR are predominantly a consequence of the heterogeneity of the material and micro-exposure conditions, resulting from the service loads and weathering impact to which the structures have been subjected, see also Section 11.4.2. Figure 8.16 contains a compilation of the results obtained in the three countries, showing a large degree of consistency, despite the different climates, materials and construction practices prevailing in them. Good use is made of this consistency of data to develop a method to estimate carbonation-induced corrosion initiation time in old structures, based on kT (Section 9.5.2). Please notice that same as for laboratory tests (Figure 8.14), the carbonation rate becomes negligible for kT values below ≈0.006 × 10−16 m². A research studied the protective effect of coating materials for textured finishes on the carbonation of concrete (Karasawa & Matsuda, 2011; Karasawa et al., 2011). Cores were drilled from a structure several decades old, on the surface of which coating materials for textured finishes had been applied. Both air-permeability kT and accelerated carbonation tests were applied on the cores, and carbonation depth was estimated from the kT values applying a model. A comparison showed that the results of the predictive model, with due consideration given to the ageing of the coating, agreed well with the measured values.

8.3.8 Frost Resistance Results of scaling frost-thaw-salts tests of concretes, the air-permeability kT of which had been previously measured, are presented and discussed. The frost test (old Swiss Standard SIA 162/1:2003, Test No. 9) consists in ponding the investigated surface of the concrete sample with a 3% NaCl solution and subjecting it to 30 frost (−12°C) and thaw (+20°C) cycles, collecting the loose material after every ten cycles (which take 7 days). The result of the test is the total loose mass δm30 collected after 30 cycles, referred to the ponded surface area of the specimen. The standard gives an indication that concretes with δm30 ≤ 600 g/m² have a “high” frostthawing salt resistance, whilst those with δm30 ≥ 3,600 g/m² have a “low” frost-thawing salt resistance. Another frost resistance criterion included in old Swiss Standard SIA 162/1:2003, Test No. 6, was the determination of the spacing factor (AF), which is the maximum distance from any point in the cement paste to the nearest air bubble, obtained by optical microscopy. The standard gives an indication that concretes with AF ≤ 200 µm have a “high” frost resistance, whilst those with AF ≥ 250 µm have a “low” frost resistance. @seismicisolation @seismicisolation

kT as durability indicator 345

Since the results of these investigations (data sets 2 and 6) were not published before, the details of the mixes used as well as the results obtained on the hardened concrete samples are presented in Table 8.3, including kT and standard cube strength. The investigation of data set 2 consisted in casting 360 × 250 × 120 mm slabs with ten concrete mixes, testing the upper 360 × 250 mm surface, as cast, of samples kept permanently in a dry room at 20°C/50% RH (samples Ao) and the opposite bottom surface of slabs moist cured during 7 days prior to storage in same dry room (Samples Bu). In both cases, the samples were demoulded at 24 hours and the test was initiated at 28 days of age. The investigation of data set 6 consisted in casting, on site, 1 m cubes, with the same concretes used to build Schaffhausen Bridge (Switzerland), see Section 7.1.5.2. Cube 2 was cast with the mix used for the Pylon (made with a Silica Fume cement type CEM II/A-D 52.5) and Cube 4 with the mix used for the Deck (made with an OPC type CEM I 42.5), see Table 8.3. The cubes were demoulded and kept near the bridge until 25 days of age when they were moved to Holcim Laboratory for testing at ages between 28 and 35 days. Measurements of frost-thaw salts resistance were made on Ø150 mm cores drilled from two sides of each cube (2–1 and 2–4 for Cube 2 and 4–1 and 4–4 for Cube 4). The recorded data of kT, AF and δm30 are reported in Table 8.3. Table 8.3 Characteristics of the mixes and properties of hardened concrete (data sets 2 and 6) Mix Cement Silica fume w/b Air Aggr. f′ccube 28 days no. (kg/m³) (kg/m³) (kg/kg) (%) type (MPa) a

kT (10−16m²) Ao

δm30 (g/ m²) AF (µm)

Bu Ao

Data set 2, p. 70,Tables 3.2-IV (a)–(c) (Torrent & Ebensperger, 1993) 1 250 0.60 4.7 Z1 25.4 7.61 0.227 1,120 85 80 70 2 325 0.46 4.8 43.0 2.39 0.127 819 - 100 3 400 0.42 4.5 42.9 2.94 0.085 672 - 99 4 325 26 0.39 2.0 74.2 0.109 0.007 34 - 300 5 400 32 0.32 1.8 79.7 0.035 0.030 14 35 245 222 6 325 0.47 6.1 Z2 38.0 2.05 0.214 872 - 106 7 325 0.47 5.7 Z3 37.4 3.21 0.382 1,665 - 204 8 Commercial bagged repair mortar 0.092 0.017 6 - 324 9 320 0.41 9.5 Z4 35.2 2.12 0.076 138 - 87 10 325 0.50 1.2 Z1 48.2 0.818 0.022 4,086 - 150 Data set 6, p. 82 and 89 (Torrent & Frenzer, 1995) 2–1 Ready-mixed concrete 2.2 79.6 2–4 B55/45

0.003 0.019

4–1 Ready-mixed concrete 4–4 B45/35

0.004 0.200

3.8

51.1

Mixes 4, 5 and 10 (Air values in italics) did not contain an air-entraining agent.

@seismicisolation @seismicisolation

27 15 29 7

275 144

346 Concrete Permeability and Durability Performance

Figure 8.17a presents the scaling mass loss δm30 as a function of kT, differentiating the samples containing air ≥ 4.5% (white symbols) and air ≤ 3.0% (black symbols). A clear trend of higher mass loss δm30 for higher kT can be observed, indicating that it is not enough to entrain air to achieve high frost resistance, but that the concretes shall have a low permeability as well. This is evident in the points linked by the arrow, corresponding to the same Mix No. 1, with similar AF (see Table 8.3), but one of high kT (upper face of dry-cured sample) and the other of low kT (bottom face of sample moist-cured 7 days). Another piece of evidence of the effect of kT on frost resistance is produced by data set 7, corresponding to an experiment conducted at the Swiss Federal Polytechnic University in Zürich (ETHZ), within the frame of a project financed by the Swiss Federal Highway Administration (Torrent & Frenzer, 1995). In this case, another frost test (old Swiss Standard SIA 162/1:2003, Test No. 8) was applied, in which the number of frost-thaw salt cycles N50 leading to a 50% reduction in the modulus of elasticity of the sample is measured. If N50 ≥ 100, the concrete is judged as having “high” frost resistance and if N50 ≤ 20 as having “low” frost resistance. The test details are described below. Concrete cubes (0.5 m) were cast with four different concrete mixes with a wide range of characteristics (w/c = 0.3–0.75; OPC = 200–450 kg/m³; f′c = 14–66 MPa), made with the same constituents, see Table 5.3. Two cubes were cast with each mix, one of which was moist cured for 7 days (B), whilst the other was totally deprived of moist curing (A). At the age of 28 days, the kT of the cubes was measured by Holcim Technology personnel, without knowing the identity of the eight cubes (blind test). The tests were conducted using a TPT, on two opposite faces of each cube, five tests on each face following a pattern like number 5 of a dice; the reported results in Table 5.3 correspond to the geometric mean of the ten tests conducted on each cube. After finishing the kT tests,

Figure 8.17 Relation between kT and (a) mass loss δm30 after 30 frost-thaw cycles scaling test and (b) number of frost-thaw-salts cycles for 50% reduction of E-modulus N 50.

@seismicisolation @seismicisolation

kT as durability indicator 347

3 Ø50 × 120 mm cores were drilled from each face tested for kT for the determination of N50. The samples were kept under water at 20°C for 5 days, moment at which the initial static modulus of elasticity E 0 was measured. Then the samples were immersed in a NaCl (1,290 kg/m³ concentration) and subjected to repeated cycles of freezing (44 minutes at −20°C) and thawing (22 minutes at +20°C). Periodically, the samples were tested for static E-modulus, until its reduction (with respect to E 0) exceeded 50%. Then, by interpolation, the number of cycles N50 causing a 50% reduction in E-modulus was computed. Figure 8.17b shows the very good correlation between N 50 and kT (measured before freezing), confirming the beneficial effect of having a low-permeability concrete for achieving a high frost-thaw salts resistance. It is worth mentioning a research (Choi et al., 2017) aimed at studying the damaging effect of early freezing on Ø100 × 200 mm mortar and concrete cylinders (OPC; w/c = 0.50). The cylinders were exposed to freezing temperature (−20°C) for 15 and 24 hours, at variable times after casting (Choi et al., 2017). In the mortar program (Series I), 15 hours freezing started 2 hours after casting and parts of the specimens were protected by an insulating material, leaving different exposed lengths to freezing (from 0 to 200 mm). In the concrete program, 24 hours freezing started 2, 6, 12, 24 and 48 hours after casting and the entire specimen length (200 mm) was exposed to freezing. Immediately after finishing the freezing process, the specimens were stored in a dry room (20°C; 65% RH). The moisture content and air-permeability kT of the frozen surface of the samples were monitored between 1 and 28 days of storage in the dry room. Figure 8.18a shows the kT values measured on mortar specimens after 1 day in the dry room as function of the freezing exposed length of the specimens, indicating with shades the degree of damage observed visually. The broken line represents the value measured on a companion specimen not subjected to early freezing.

Figure 8.18 (a) Effect of exposed length and degree of damage on kT of mortars tested 1 day after freezing, and (b) effect of freezing age and degree of damage on kT tested 3 days after freezing; data from Choi et al. (2017).

@seismicisolation @seismicisolation

348 Concrete Permeability and Durability Performance

The early frost damage induces a huge increase in kT; similar results were obtained for kT measured 28 days after freezing. This behaviour is explained by the fact, confirmed by MIP analysis, that freezing coarsens the pore structure, which reflects in an increased air-permeability kT. Figure 8.18b shows the effect of the age at which both mortar and concrete specimens were subjected to freezing and of the subsequent observed damage on kT. The researchers recommend measuring kT after 3 days of freezing as a good indicator of the degree of damage caused by early freezing. They suggest two thresholds (kT = 10 and 1,000 × 10−16 m²) delimiting zones of no damage, damage and obvious damage by early freezing, indicated in Figure 8.18b. Zhang et al. (2019) tested the frost-thaw-salts resistance of concrete mixes subjected to different curing conditions, with and without the inclusion of a controlled permeable formwork (CPF) liner (see Section 7.2.3); prior to initiating the frost tests, kT was measured. They found that extended curing and the presence of the CPF liner reduced the surface air-permeability, thus resulting in lower scaling mass loss. It can be concluded that air-permeability kT is, as transport property, a useful indicator of the frost resistance of concrete, but by no means the only one. Liu et al. (2014) and Liu and Hansen (2015) present data with good correlations between the scaling mass loss in frost-thaw-salts tests and water sorptivity. 8.4 SOME NEGATIVE EXPERIENCES So far, with the possible exception of the lack of correlation with electrical resistivity, positive cases of correlation between kT and many other transport and durability-related properties of concrete have been reported. As with all test methods, sometimes, abnormal measurement results happen, be them due to shortcomings of the test method itself, to wrong operation (e.g. lack of conditioning or inaccurate calibration of the instrument) or to extreme conditions of the concrete (surface temperature and moisture, coatings, microcracks, poorly bonded surface layers, etc.). In this section, reported cases where abnormal or out of expectations results were obtained are described.

8.4.1 Tunnel in Aargau, Switzerland This case refers to 16 parallel measurements of kT, conducted on the walls of a Tunnel by F. Jacobs (using a TPT instrument) and by R. Torrent (using a PermeaTORR instrument). The measurements were performed exactly on the same spots with both instruments (after a delay of at least 30 minutes); the results were reported in Jacobs et al. (2009). The black squares in Figure 5.13 show the excellent correlation obtained between both sets of measurements, which covered three different Permeability Classes (orders of magnitude). On the initiative of Materials Advanced Services SRL, Ø50 mm cores were @seismicisolation @seismicisolation

kT as durability indicator 349

Figure 8.19 Relation between kT measured on site in a Tunnel and M Cl and a24, measured in the lab on cores drilled from same locations.

drilled from 15 of the locations where kT had been measured and saw-cut to a thickness of 50 mm. The specimens were tested for coefficient of water absorption at 24 hours (a24) and for coefficient of chloride migration (MCl), according to Annexes A and B, respectively, of Swiss Standard SIA 262/1. It was expected that the significant differences in kT would be reflected in similar differences in both a24 and MCl; the results presented in Figure 8.19 indicate that that was not the case. Indeed, the results of both laboratory tests do not show any correlation with the kT values measured on site (the results obtained with the PermeaTORR are plotted along the x-axis). It is worth mentioning that the migration test results are rather high, with 12 × 10 −12 m²/s being the upper limit of the Swiss Standards for tests made on cores drilled from structural elements exposed to chlorides (just one out of the 15 tests fell below that limit). Two possible explanations for this behaviour are: (1) kT is being affected by some very superficial effect (coating, delamination, etc.) that is not affecting the laboratory results (the penetration of chlorides in the migration test reached depths between 25 and 32 mm), and (2) there was some problems with the samples and/or the laboratory tests (the kT site tests are assumed to be correct, as they were confirmed by two sets of measurements performed by different operators and instrument brands).

8.4.2 Wotruba Church, Vienna, Austria A condition assessment was made by Matea Ban (Institute of Conservation and Restoration, University of Applied Arts Vienna, Austria) on the iconic example of modern Austrian heritage known as Wotruba Church (Ban, 2013, 2014). @seismicisolation @seismicisolation

350 Concrete Permeability and Durability Performance

A visual inspection showed the following reported results: weathered cement skin, eroded edges, efflorescence, microbiological growth and consequent surface corrosion. Due to the architectural value of the building, only low invasive examinations were allowed (samples for carbonation depth and for thin sections preparation). The following NDTs were performed: water sorptivity (Karsten Tube, Section 4.2.2.2), rebound hammer, cover depth and air-permeability kT. All tests were conducted at an age of 38 years (Ban, 2013, 2014). The Karsten tube tests yielded moderate to high rates of water absorption on the E, S and W faces (values between 8 and 17 g/m²/s½) and extremely high on the N face (28–62 g/m²/s½). The carbonation rate (carbonation depth divided by the square root of 38 years) yielded low values on the surfaces exposed to rain (1.6–3.2 mm/y½), but high on those protected from rain (7.3–8.1 mm/y½). Regarding the measurements of air-permeability kT, it is worth citing (Ban, 2014) “The air permeability values were unfortunately unusable; the data gained showed extremely high kT [× 10−16 m²] values, not comparable with standard values. However, the values received indicated that the yellowish concrete was more permeable than the grey concrete. Even though it was not possible to take examples from the façade, where the yellow concrete was applied, it is known from the architect in charge Fritz G. Mayr that unintentionally different aggregates with high clay content were used.” Mrs. Ban was thoroughly trained in the use of the PermeaTORR (instrument she used for the survey), jointly by experts Dr. F. Jacobs and Dr. R. Torrent, so neither the instrument nor the operator can be blamed for the high results (not disclosed). Yet, the high water sorptivity and carbonation rate measured, at least in parts of the structure, may correspond to concrete of very high air-permeability.

8.4.3 M inistry of Transport, Ontario, Canada An investigation on the “penetrability” of precast concrete barrier walls was conducted by the Ministry of Transport (MoT), Ontario, Canada (Berszakiewicz & Konecny, 2008). The experiment is not well described, but it seems that the walls were cast with mixes of w/c ratios 0.30, 0.45 and 0.60 (possibly made with OPC). Two test methods were applied directly on the walls, namely, air-permeability kT and a version of water sorptivity ISAT test, modified by the University of Toronto. Unfortunately, the test results are not presented but the following comments were made by the researchers on the kT test “A series of tests results were carried out under laboratory conditions on slab specimens with different water/cement ratios (0.60; 0.45 and 0.30), at different concrete ages. Analysis of test results showed that @seismicisolation @seismicisolation

kT as durability indicator 351

there was no clear relationship between air permeabilities and concrete of different w/c ratios. When tested in the field, test could not clearly differentiate between the permeabilities of normal and the high performance concrete. The air permeability tests had repeatability even lower than the sorptivity measurements, in the range of the coefficient of variation of 50 to 70 percent. Thus, it is possible that additional factors, other than those already listed with respect to sorptivity, may have had an effect on air permeability results.” On the more positive side “The Torrent test method provided interesting results when used to comparatively evaluate the air permeability of formed concrete surface with different finishes … While the test did not measure different air permeability levels of the two types of surface finishes for the 50 MPa concrete, it showed a consistent difference in air permeability for the 30 MPa concrete. The surface of 30 MPa concrete formed with the liner had an air permeability 3 times lower than the surface of the same concrete formed without the liner. At the same time the air permeability of the surface of 30 MPa concrete formed with the liner was close to the permeability of the 50 MPa concrete surfaces. This leads to the conclusion that the surface of the 50 MPa concrete formed with or without the liner, had a pore system consistent with low air permeability, while the use of the form liner for the 30 MPa concrete lowered the permeability to air, thus improving the quality of the concrete surface.” The water sorptivity test performed better in differentiating between mixes of different w/c ratios (0.60, 0.45 and 0.30). The investigation also included electrical resistivity tests (Wenner Probe) that, when applied on site, showed the typical obstacle of the effect of moisture on the readings. More information on the experience of Ontario’s MoT with kT test can be found in Ip et al. (1998).

8.4.4 M ansei Bridge, Aomori, Japan This bridge, built in 1955, was demolished in 2010 because of operational limitations, without significant deterioration of the main girder and slab. Before demolition, a condition survey was conducted including the site measurement of kT at different points of the piers, girders and slab (Watanabe et al., 2012). Later, cores were drilled near the kT testing points for measuring: E-modulus and compressive strength, carbonation depth, chloride ion penetration, MIP and scaling resistance, showing interesting results. Regarding air-permeability kT, the measurement points were selected to be free of surface damage (scaling, cracks, etc.). Yet, kT could not be measured (too high permeability), a fact attributed by the researchers to some subsurface damage (possibly freeze-thaw delamination), not visible on the surface. @seismicisolation @seismicisolation

352 Concrete Permeability and Durability Performance

8.4.5 Tests at FDOT Laboratory A series of 7-year-old concrete cylinders (Ø100 × 200 mm), stored under water at the Florida Department of Transportation (FDOT) lab in Gainesville, FL, USA, were selected for a pilot test with a PermeaTORR instrument (Torrent & Armaghani, 2011). The concrete mixes involved had w/c ratios between 0.28 and 0.49, most of them of w/c = 0.35 and were prepared with a large variety of binders (including plain OPC and double and triple blends with PFA, SF, GGBFS and Metakaolin). The specimens were saw-cut by halves, leaving four plane faces for testing kT. Due to the small dimensions of the specimens, a special cell reduced to 70% the size of the standard cell had to be manufactured and installed in the instrument. Huge differences were found between the kT measurements obtained on the four faces of each specimen, which did not follow any systematic pattern (for instance, the top surface as cast was the best and sometimes the worst of the four surfaces, in terms of kT). This variability precluded yielding any conclusion from the tests performed. One possible explanation for the high variability observed could be the influence of coarse aggregate particles on the reduced size of the internal chamber of the small cell (Ø = 35 mm). 8.5 AIR-PERMEABILITY kT IN STANDARDS AND SPECIFICATIONS

8.5.1 Swiss Standards Swiss Standards present the most complete and comprehensive approach regarding the measurement of kT on site, which started in 2003 with the following statements in standard SIA 262 “Concrete Construction”, Swiss version of Eurocode 2, still included in version 2013 of the same standard (SIA 262, 2013): “The impermeability of the cover concrete shall be checked by means of permeability tests (e.g. air-permeability measurements) on the structure or on core samples taken from the structure”. In parallel, a complementary standard, describing tests not covered by EN standards, was issued in 2003, included the air-permeability test kT as Annex E: “Air-Permeability on site”. The current version is largely improved (SIA 262/1, 2019). The standard provides instructions on how to calibrate the instrument and run the tests and sets several conditions for the measurement: • • • • •

age of the structure between 28 and 120 days temperature of the concrete ≥ 5°C surface moisture of the concrete (electrical impedance method) ≤ 5.5% cover thickness when measuring in coincidence with rebar: ≥ 20 mm it also recommends limiting values of kT to be specified (kTs) for different exposure conditions, see Table 8.4, taken from Version 2008 of SIA 262/1 @seismicisolation @seismicisolation

kT as durability indicator 353 Table 8.4 Limiting values kTs as function of the exposure conditions, taken from SIA262/1:2013 Concrete type Description

Strength classesa Exposure classesb

C20/25

C25/30

C30/37

C25/30

C30/37

XC1 XC2

XC3

XC4 XF1

280

300

XC4 XD1 XF2 300

XC4 XD1 XF4 300

XC4 XD3 XF2 320

XC4 XD3 XF4 320

0.65

0.60

0.50

0.45

2.0

0.5

Minimum cement content (kg/m3) Maximum w/c ratio Air-permeability kTs (10−16 m2) a b

The indicated values correspond to the required characteristics strength (MPa) at 28 days, measured on cylinders/cubes. Correspond to the exposure classes defined in European Standard EN 206-1. The combinations of exposures are those typically found in Switzerland. The limits for XD classes can be applied to equivalent XS classes for marine environments, absent in Switzerland.

In SIA 262/1 (2019), the way of grouping structural elements cast with same mixes and subjected to similar concrete practices is explained, defining a lot (within each group) as the minimum exposed surface area of the following two alternatives: • 500 m² of exposed surface • three days of concreting From each lot, six kT tests are performed at locations randomly selected with the following conformity conditions:

If none of the conditions (1) and (2) is met, the Lot is regarded as noncompliant with the corresponding specified air-permeability limit kTs. The Operating Characteristic (O-C) curve of the above-mentioned compliance criterion is presented in Figure 8.20, thus giving a clearer @seismicisolation @seismicisolation

354 Concrete Permeability and Durability Performance

Figure 8.20 O -C curve of the conformity criterion of Swiss Standard (SIA 262/1, 2019) for kT tests on site.

probabilistic meaning to the kTs value; derivation (with wrong Figure D-5) in Jacobs et al. (2009). The O-C curve carries in abscissas the proportion of the concrete surface in a Lot or Element with kT higher than the specified value kTs , i.e. the proportion of “defective” concrete. In ordinates, the chart presents the probability of accepting a lot (applying conformity rules (1) and (2) above) containing a given proportion of “defectives”. From Figure 8.20, we can see that a lot containing 10% of “defectives” will have 97% probability of being accepted (broken line), whilst for one with 50% “defectives”, the probability drops to just 13% (dotted line). A translation into English of the parts of SIA 262/1 (2019) relevant to kT test can be found in www.m-a-s.com.ar. Version 2019 of Swiss Standard SIA 262/1 does not cover the application of the test method in the laboratory, which is very much needed in order to better compare the results obtained in different laboratories. The Argentine Standard IRAM 1892 (see the next section) covers that gap.

8.5.2 Argentina IRAM Standard 1892:2021 covers the application of air-permeability kT both in the laboratory and on site (the latter based on SIA 262/1) . The standard, in the preparation of which Dr. Torrent was actively involved, updates Swiss Standard SIA 261/1:2019 but, more important, covers the use of the test method in the laboratory. The Model Standard in Annex B is based on IRAM 1892. @seismicisolation @seismicisolation

kT as durability indicator 355

8.5.3 Chile A Highways Manual (MdC, 2017), elaborated by the Chilean Highways Administration, Ministry of Public Works, includes in the specifications for precast reinforced concrete box sections for culverts, storm drains, and sewers, upper limits for the value of kT of 0.01 × 10 −16 m² for severe exposures and of 0.1 × 10 −16 m² for moderate exposures (exposures defined in the same document). In addition, the new Chilean Concrete Standard (NCh170, 2016) includes, in its Annex B3, the kT test method, referring directly to the Swiss Standard SIA 262/1 for its application.

8.5.4 China The test method is included in Jiangsu Prov. Chinese Standard (DGJ32/TJ 206, 2016).

8.5.5 India The kT test method is included in Section 2.9 “Permeability Test” of BS 103 (2009), dealing with non-destructive testing of bridges.

8.5.6 Japan In Appendix 1 of JCI (2014), the air-permeability kT test method is included within “Detailed survey methods and setting of characteristic evaluation values” for existing structures. NETIS (New Technology Information System) is a database owned by the Japanese Ministry of Land, Infrastructure and Transport. NETIS shares cutting-edge technologies applicable to construction with the private sector. Technologies registered to NETIS will be disclosed in the database so the information and evaluation on the technology will be available to all sectors. On 16 March 2020, the instrument PermeaTORR AC was registered in NETIS under No. QS-150029-VE. Furthermore, in 2020, the test method was standardized by The Japan Society for Non-destructive Inspection (JSNDI) as NDIS 3436-2 “Nondestructive testing of concrete - Air permeability testing method Part 2: Double chamber method”.

8.6 CREDENTIALS OF AIR-PERMEABILITY kT AS DURABILITY INDICATOR In Chapter 6, it was shown that the coefficient of air-permeability kT responds sensitively, and according to the expectations, to changes in key technological parameters affecting the quality of concrete, such @seismicisolation @seismicisolation

356 Concrete Permeability and Durability Performance

as w/b ratio, compressive strength, binder and aggregate types, compaction, segregation and bleeding, curing, applied stresses and cracks (seeTable 8.1). In this chapter, from good to excellent correlations with other transport and durability-related tests were presented, including: other gas-permeability and gas-diffusivity tests, water-permeability tests (including penetration under pressure and capillary suction), chloride-diffusion and migration tests, natural and accelerated carbonation and freezing and thawing tests. The only property with which a good correlation could not be established was electrical resistivity, which should therefore be considered a complementary rather than an alternative test. Over 430 publications worldwide (list available at www.m-a-s.com.ar) present positive applications of the test method to investigate the potential durability of concrete structures, with just a handful of negative experiences as those described in Section 8.4. There may be more such cases, as often negative experiences tend to be omitted or not published; yet the number of positive cases is overwhelmingly high. Timidly, the test method starts to be included in concrete standards and specifications, a trend likely to intensify in the future. All these evidences lead to the conclusion that the coefficient of airpermeability kT constitutes a suitable durability indicator for concrete material and structures. REFERENCES Akiyama, H., Inoue, S. and Kishi, T. (2010). “Effect of curing conditions on the relation between air permeability and carbonation resistance of surface concrete”. Seisan Kenkyu, v62, n6, 603–604 (in Japanese). Alexander, M. and Thomas, M. (2015). “Service life prediction and performance testing — Current developments and practical applications”. Cem. & Concr. Res., v78, 155–164. Andrade, C. (1993). “Calculation of chloride diffusion coefficients in concrete from ionic migration measurements”. Cem. & Concr. Res., v23, 724–742. Andrade, C., González Gasca, C. and Torrent, R. (2000). “The suitability of the TPT to measure the air-permeability of the covercrete”. ACI SP-192, 301–318. Bahurudeen, A. and Santhanam, M. (2014). “Performance evaluation of sugarcane bagasse ash-based cement for durable concrete”. 4th International Conference on Durability of Concrete Structure, Purdue Univ., West Lafayette, IN, USA, July 24–26, 275–281. Ban, M. (2013). “Aspects of conserving exposed concrete architecture with Wotruba Church as an example”. RILEM Proc. PRO 89, 549–556. Ban, M. (2014). “Wotruba Church and Cologne Opera: Aspects of concrete aging”. Concrete Solutions. Grantham et al. (Eds.), 619–625. Beglarigale, A., Ghajeri, F., Yigiter, H. and Yazici, H. (2014). “Permeability characterization of concrete incorporating fly ash”. ACE 2014, Istanbul, Turkey, October 21–25, 7 p.

@seismicisolation @seismicisolation

kT as durability indicator 357 Berszakiewicz, B. and Konecny, J. (2008). “In search of reliable in situ test methods for development of performance-based specifications for concrete in highway structures”. Session “Bridges – Links to a Sustainable Future (B)” of the 2008 Annual Conference and Exhibition of the Transportation Association of Canada – Transportation: A Key to a Sustainable Future, Ottawa, Ontario. Beushausen, H., Starck, S. and Alexander, M. (2012). “The integration of nondestructive test methods into the South African durability index approach”. Microdurability 2012, Amsterdam, 11-13 April, Paper 113. Bisschop, J., Schiegg, Y. and Hunkeler, F. (2016). “Modelling the corrosion initiation of reinforced concrete exposed to deicing salts”. Bundesamt für Strassen, Bericht Nr. 676, Switzerland, February, 91 p. BS 103 (2009). “Guidelines on non-destructive testing of bridges”. Government of India, Ministry of Railways, August, 133 p. CEB-FIP (1991). “CEB-FIP model code 1990”. Final Draft, CEB Bulletin d‘Information N° 203, 204 and 205, Lausanne, Switzerland, July 1991. Choi, H., Zhang, W. and Hama, Y. (2017). “Method for determining early-age frost damage of concrete by using air-permeability index and influence of early-age frost damage on concrete durability”. Constr. & Build. Mater., v153, 630–639. Denarié, E., Conciatori, D. and Simonin, P. (2003). “Essais comparatifs de caractérisation de bétons d‘enrobage - phase I: bétons de laboratoire”. Rapport d‘essais MCS 02.12.07-1, EPFL, Lausanne, Switzerland, 21 p. DGJ32/TJ 206 (2016). “Technical specification for quality control of high performance concrete in urban rail transit construction”. Jiangsu Province Standard, R.P. China. DIN 1048 (1978). “Prüfverfahren für Beton - Bestimmung der Wassereindringtiefe”. Di Pace, G. and Calo, D. (2008). “Assessment of concrete permeability in tunnels”. SACoMaTIS 2008, Varenna, Italy, September 1–2, v1, 327–336. Ebensperger, L. and Olivares, M. (2019). “Envejecimiento a mediano plazo de probetas de concreto: factor incidente en las estimaciones de vida útil”. CONPAT 2019, Chiapas, México, v1, 13 p. EN 12390-8 (2009). “Testing hardened concrete – Part 8: Depth of penetration of water under pressure”. Fernández Luco, L. and Revuelta Crespo, D. (2005). “Ensayo de penetración de agua bajo presión y Ensayo de permeabilidad al aire, método de Torrent, sobre probetas de hormigón de 150 ×300 mm” (in Spanish). Informe N° 18’728, Instituto Eduardo Torroja, Madrid, Spain, July, 8 p. Fornasier, G., Fava, C., Fernández Luco, L. and Zitzer, L. (2003). “Design of self compacting concrete for durability of prescriptive vs. performance-based specifications”. ACI SP 212, 197–210. Imamoto, K., Neves, R. and Torrent, R. (2016). “Carbonation rate in old structures assessed with air-permeability site NDT”. IABMAS 2016, Paper 426, Foz do Iguaçú, Brazil, June 26–30. Imamoto, K., Shimozawa, K., Nagayama, M., Yamasaki, J. and Nimura, S. (2008). “Threshold values of air permeability of concrete cover – A case study in Japan”. SACoMaTIS 2008, v1, 169–177. Imamoto, K., Shimozawa, K., Nagayama, M., Yamasaki, J. and Tanaka, A. (2014). “Relationship between air-permeability and carbonation progress of concrete in Japan”. International Workshop on Performance-based Specification and Control of Concrete Durability, Zagreb, Croatia, June 11–13, 325–333.

@seismicisolation @seismicisolation

358 Concrete Permeability and Durability Performance Imamoto, K., Tanaka, A. and Kanematsu, M. (2012). “Non-destructive assessment of concrete durability of the National Museum of Western Art in Japan”. Paper 180, Microdurability 2012, Amsterdam, April 11–13. Ip, A., Berszakiewicz, B. and Pianca, F. (1998). “Nondestructive test methods for evaluating durability of concrete highway structures: experience of Ontario Ministry of Transportation”. Proc. Struct. Mater. Technol. III: An NDT Conference, Eds: R.D. Medlock; D.C. Laffrey, v3400, 270–280. ISO 1920-12 (2015). “Testing of concrete – Part 12: Determination of the carbonation resistance of concrete accelerated carbonation method”. ISO International Standard, 20 p. Jacobs, F. (2006). “Luftpermeabilität als Kenngrösse für die Qualität des Überdeckungsbetons von Betonbauwerken”. Bundesamt für Strassen, Bericht Nr. 604, September, 100 p. + Anhänge. Jacobs, F. (2008). “Beton zerstörungsfrei untersuchen”. der Bauingenieur, n3, 24–27. Jacobs, F., Denarié, E., Leemann, A. and Teruzzi, T. (2009). “Empfehlungen zur Qualitätskontrolle von Beton mit Luftpermeabilitätsmessungen”. Office Fédéral des Routes, VSS Report 641, December, Bern, Switzerland, 53 p. JCI (2014). “Performance evaluation guidelines of existing concrete structures”. Japan Concr. Inst., 459 p. (in Japanese), I–17–18. Jornet, A., Corredig, G. and Mühlethaler, U. (2011). “Concretes made with CEM II/A-LL: Relationship between microstructure and properties”. 13th Euroseminar on Microscopy Applied to Building Materials, Ljubljana, Slovenia, June 14–18. Karasawa, T. and Masuda, Y. (2011). “Carbonation supressing effects of coating materials for textured finishes based on the research of air permeability coefficient and carbonation depth of existing structures”. J. Struct. Eng., AIJ, v76, n661, 449–454 (in Japanese). Karasawa, T., Masuda, Y. and Lee, YR. (2011). “Research on carbonation supressing effect of coating materials for textured finishes and air permeability coefficient based on the result of the survey of an existing structure”. J. Struct. Eng., AIJ, v76, n669, 1885–1890 (in Japanese). Kattar, J.E., Abreu, J.V. de and Cruz, L.O. (1995). “Concreto de alto desempenho modificado con polímero para pisos industriais”. 37ª Reunião Anual do IBRACON, Goiânia, Brazil, Julho 3–7, 15 p. Kattar, J., Abreu, J.V. and Regattieri, C.E.X. (1999). “Inovações na metodologia para avaliação da permeabilidade por difusão ao ar”. 41° Congresso do IBRACON, Salvador, Bahia, Brazil. Kubens, S., Wassermann, R. and Bentur, A. (2003). “Non destructive air permeability tests to assess the performance of the concrete cover”. 15th ibausil Intern. Baustofftagung, Bauhaus, Univ. Weimar, September 24–27. Kurashige, I. and Hironaga, M. (2010). “Nondestructive quality evaluation of surface concrete with various curing conditions”. CONSEC’10, Mérida, México, June 7–9. Li, K., Li, Q., Zhou, X. and Fan, Z. (2015). “Durability design of the Hong Kong– Zhuhai–Macau Sea-Link project: Principle and procedure”. J. Bridge Eng., ASCE, 04015001, 11 p. Liu, Z. and Hansen, W. (2015). “Sorptivity as a measure of salt frost scaling resistance of air-entrained concrete”. Key Engng. Mater., v629–630, 195–200.

@seismicisolation @seismicisolation

kT as durability indicator 359 Liu, Z., Hansen, W. and Wei, Y. (2014). “Concrete sorptivity as a performance-based criterion for salt frost scaling resistance”. RILEM International Symposium on Concrete Modelling, Beijing, China, October 12–14, 487–496. Maître, M. (2012). “Tunnel de Naxberg - Perméabilité à l’air du béton d‘enrobage (méthode Torrent)”. EPFL, Rapport d‘essais n° MCS 02.09-01, Lausanne, November, 9 p. Mathur, V.K., Verma, C.L., Gupta, B.S., Agarwal, S.K. and Kumar, A. (2005). “Use of high-volume fly ash in concrete for building sector”. Report No. T(S)006, Central Build. Res. Inst., Roorkee, India, January, 35 p. MdC (2017). “Manual de Carreteras – Especificaciones Técnicas Generales de Construcción”. MOP, Dirección de Vialidad, Sección 5.612.201 ‘Cajones Prefabricados, de Hormigón Armado’. Chile. Mohr, P., Hansen, W., Jensen, E. and Pane, I. (2000). “Transport properties of concrete pavements with excellent long-term in-service performance”. Cement & Concr. Res., v30, 1903–1910. Moro, F. and Torrent, R. (2016). “Testing fib prediction of durability-related properties”. fib Symposium 2016, Cape Town, South Africa, November 21–23. NCh170 (2016). “Hormigón – Requisitos generales”. Norma Chilena, 4ª Ed., 25 May, 44 p. Neves, R.D. (2012). “A Permeabilidade ao Ar e a Carbonatação do Betão nas Estruturas”. PhD Thesis, Universidade Técnica de Lisboa, Instituto Superior Técnico, Portugal, 502 p. Neves, R., Sena da Fonseca, B., Branco, F., de Brito, J., Castela, A. and Montemor, M.F. (2015). “Assessing concrete carbonation resistance through air permeability measurements”. Constr. & Build. Mater., v82, 304–309. Nishimura, K., Kato, Y. and Mita, K. (2015). “Influence of construction work conditions on the relationship between concrete carbonation rate and the air permeability of surface concrete”. International Conference Regeneration and Conservation of Concrete Structures, Nagasaki, Japan, June 1–3, Paper R1–4, 8 p. Olek, J., Lu, A., Feng, X. and Magee, B. (2002). “Performance-related specifications for concrete bridge superstructures, volume 2: High-performance concrete”. Purdue Univ. – Joint Transportation Research Program Technical Report Series, 215 p. Park, J-J., Koh, K-T., Kim, D-G. and Kim, S-W. (2004). “The Chloride Diffusion Properties of Concrete with Mineral Admixtures”. Korean Soc. Concr. Diagnosis, v8, n4, 239–246 (in Korean). Park, S-B. and Kim, D-G. (2000). “A experimental study on the chloride diffusion properties in concrete”. J. Korea Concr. Inst., v12, n1, 33–44 (in Korean). PC (2014). Personal communication from ACP (Panama Canal Authority). RILEM TC 189-NEC (2007). “Non-destructive evaluation of the penetrability and thickness of the concrete cover”. RILEM Report 40, May, 223 p. Rodríguez de Sensale, G., Sabalsagaray, B.S., Cabrera, J., Marziotte, L. and Romay, C. (2005). “Effect of the constituents on the properties of SCC in fresh and hardened state”. fib Symposium on “Structural Concrete and Time”, La Plata, Argentina, September.

@seismicisolation @seismicisolation

360 Concrete Permeability and Durability Performance Roelfstra, G., Adey, B., Hajdin, R. and Brühwiler, E. (1999). “The condition evolution of concrete bridges based on a segmental approach, non-destructive test methods and deterioration models”. 78th Annual Meeting Transportation Research Board, Denver, April, 13 p. Romer, M. and Leemann, A. (2005). “Sensitivity of a non-destructive vacuum test method to characterize concrete permeability”. ICCRRR, Cape Town, November 21–23. Shi, C. (2003). “Another look at the rapid chloride permeability test (ASTM C1202 or AASHTO T277)”. FHWA Resource Center, Baltimore, MD, 15 p. SIA 262 (2013). “Betonbau”. Swiss Soc. of Engineers and Architects (in French, German and Italian). SIA 262/1 (2019). “Concrete construction – Complementary specifications”. Swiss Soc. of Engineers and Architects (in French and German). Starck, S. (2013). “The integration of non-destructive test methods into the South African durability index approach”. MSc Dissertation, Univ. Cape Town, March, 188 p. Starck, S., Beushausen, H., Alexander, M. and Torrent, R. (2017). “Complementarity of in situ and laboratory-based concrete permeability measurements”. Mater. & Struct., v50, 177–191. Tang, L. and Nilsson, L.-O. (1992). “Rapid determination of chloride diffusivity of concrete by applying an electric field”. ACI Mater. J., v49, n1, 49–53. Teruzzi, T. (2009). “Estimating the service-life of concrete structures subjected to carbonation on the basis of the air permeability of the concrete cover”. EUROINFRA 2009, Helsinki, October 14–15. Torrent, R. and Ebensperger, L. (1993). “Methoden zur Messung und Beurteilung der Kennwerte des Überdeckungsbetons auf der Baustelle”. Office Fédéral des Routes, Rapport No. 506, Bern, Switzerland, Januar, 119 p. Torrent, R. and Frenzer, G. (1994). “Durabilidad de concretos elaborados con cementos Tipo I, Puzolánico y de Escoria. Estudio Comparativo”. “Holderbank” Report MA-94-3246-S, June 15, 20 p. Torrent, R. and Frenzer, G. (1995). “Methoden zur Messung und Beurteilung der Kennwerte des Ueberdeckungsbetons auf der Baustelle -Teil II”. Office Fédéral des Routes, Rapport No. 516, Bern, Suisse, October, 106 p. Van Eijk, R.J. (2009). “Evaluation of concrete quality with Permea-TORR, Wenner Probe and Water Penetration Test”. KEMA Report, Arnhem, July 8, 46p. (inDutch). Wang, Y.F., Dong, G.H., Deng, F. and Fan, Z.H. (2014). “Application research of the efficient detection for permeability of the large marine concrete structures”. Appl. Mechanics Mater., v525, 512–517. Watanabe, K., Sakoi, Y., Aba, M., Kamiharako, A. and Tsukinaga, Y. (2012). “Durability investigation of RC bridge after 56 years”. 37th Conference on Our World in Concrete & Structure, August 29–31, Singapore. Zhang, M., Sakoi, Y., Aba, M. and Tsukinaga, Y. (2019). “Effect of initial curing conditions on air permeability and de-icing salt scaling resistance of surface concrete”. J. Asian Concr. Federation, v5, n1, June, 56–64. Zhutowsky, S. and Kovler, K. (2012). “Effect of internal curing on durabilityrelated properties of high-performance concrete”. Cem. & Concr. Res., v42, 20–26.

@seismicisolation @seismicisolation

Chapter 9

Service life assessment based on site permeability tests

9.1 INTRODUCTION Traditionally, concrete codes and standards have applied and still apply the “deemed-to-satisfy” approach (Andrade, 2006) to specify durability requirements. Based on the accumulated experience in many countries, a set of primarily prescriptive rules have been established which, when rigorously observed, are expected to result in a service life typically of 50 years (e.g. Eurocode 2 (EN 1992-1-1, 2004)). As they are used later, a classification of exposure environments, concerning just reinforcement corrosion, is provided in Table 9.1 (including carbonation- and chloride-induced corrosion). It corresponds to European Standards (EN 206, 2013; EN 1992-1-1, 2004); the maximum w/c ratios and minimum cover thickness d stipulated in those standards are also indicated in Table 9.1. It is assumed (“deemed-to-satisfy”) that, if the maximum water-cement ratio w/cmax has been observed by the concrete producer and the contractor has executed the concreting operations correctly, e.g. by following (EN 13670, 2009), complying with the minimum cover thickness dmin, the structure will reach its expected service life t SL of 50 years. The limitations of this approach have been highlighted in Section 1.6. Recently, codes and standards have been moving towards the “Performance Indicators” approach, by which the prescriptive rules have been replaced by performance requirements (see Table 1.2). For instance, Canadian Standard (CSA A23.1, 2006) establishes limiting values for the charge passed in “RCPT” test (ASTM C1202, 2010) for normal and extended service lives in a chloride-rich environment. Swiss Standard (SN EN 206, 2016) establishes limiting values for water sorptivity, chloride migration, accelerated carbonation and frost-thaw-salt scaling resistance tests, applicable to the corresponding exposure conditions. Swiss concrete producers shall prove that their concretes pass the relevant tests, conducted on specimens obtained from samples taken from their regular production. The limitations of this Labcrete approach have been highlighted in Chapter 7. Swiss Standard (SIA 262/1, 2019) complements the previously mentioned standard by DOI: 10.1201/9780429505652-9 @seismicisolation @seismicisolation

361

362 Concrete Permeability and Durability Performance Table 9.1 EN exposure classes and main requirements for steel corrosion induced by carbonation and chlorides Class designation

Description of environment

Corrosion induced by carbonation XC1 Dry Permanently wet XC2 Wet, rarely dry XC3 Moderate humidity XC4 Cyclic wet and dry

w/cmax

dmin (mm)

0.65

0.60 0.55 0.50

25 25 30

Corrosion induced by chlorides other than from sea water XD1 Moderate humidity 0.55 XD2 Wet, rarely dry 0.55 XD3 Cyclic wet and dry 0.45

35 40 45

Corrosion induced by chlorides from sea water XS1 Exposed to airborne salt XS2 Permanently submerged XS3 Tidal, splash and spray zones

35 40 45

0.50 0.45 0.45

establishing limiting values for the coefficient of air-permeability kT, measured on site, thus evaluating the quality of the end-product, the Realcrete. It also offers the alternative of performing the above-mentioned laboratory tests on cores drilled from the structures, with more lenient requirements for the Realcrete compared with those for the Labcrete. These Swiss limiting values for Labcrete and Realcrete apply (in conjunction with the minimum cover thickness requirements) for an expected service life of 50 years. The situation is that, nowadays, many important structures are designed for service lives of 100, 150 or even more years, which clearly exceed the reach of existing codes and, therefore, requires some extrapolation or prediction via modelling. Moreover, in the past, the burden of maintenance and repair costs of structures fell predominantly on the shoulders of the owner, with other players (designers, contractors, materials suppliers) assuming the responsibility for durability for a relatively short period (typically 5–10 years). The advent of Design, Build and Operate contracts, whereby a private organization designs, builds and operates the facility for a period of several decades has changed the picture. Now, contractors have a direct interest in the durability of the construction, since maintenance and repair costs plus eventual penalties for reduced operability of the facility will be borne by them. Moreover, often, the transfer price of the facility to the final owner is associated with its residual service life that needs to be fairly established.

@seismicisolation @seismicisolation

Service life based on site permeability 363

These examples show the increasing economic relevance of having tools capable of reliably predicting the service life of concrete structures that are: • • • •

accurate: the prediction is close to the service life actually reached meaningful: based on sound principles realistic: take into consideration relevant parameters of the end-product objective: contain few (if any) hardly measurable parameters that could be freely and subjectively chosen

Various service life prediction models/methods have been developed recently. From them, there are two that have gained wide acceptance: DuraCrete (DuraCrete, 2000) in Europe, on which the fib has based its own model (fib, 2006), and Life-365 (Life-365, 2018) in North America. DuraCrete deals with steel corrosion induced either by carbonation or by chlorides, whilst Life-365 deals just with chloride-induced corrosion. Regarding chloride-induced steel corrosion, both methods assume (with some differences) a purely diffusive process, the key concrete property used as input being the coefficient of chloride-diffusion D 0 at 28 days. When modelling service life, it is important to define which is the Limit State considered by the model, for which Tuutti’s model (Tuutti, 1982) is of help (Figure 1.4). Tuutti’s model consists, basically, in dividing the deterioration process of a concrete structure into two well-differentiated phases: Incubation and Propagation, as already discussed in Section 1.3. In the Incubation phase (often called Initiation phase), no visible damage can be detected in the structure but relevant processes are taking place, e.g. the penetration of the carbonation or critical chlorides front towards the reinforcement in the case of steel corrosion. At a certain time, called Initiation Time, the reinforcing steel becomes depassivated and the true damaging action starts to take place, with the amount of steel corrosion products being of such magnitude that stains or micro-cracks appear on the concrete surface. If these deleterious reactions are allowed to continue (Propagation phase), the damage increases until it reaches a certain critical level that puts the serviceability or safety of the structure at risk. Therefore, the service life t SL of a concrete structure can be defined as where ti is the initiation and t p is the propagation time of the damage. In modelling chloride-induced corrosion, the propagation period is often disregarded, or assumed to be rather short (6 years in the case of Life-365), as the moist environments rich in chloride ions are usually favourable to the propagation of corrosion. In the case of carbonation-induced corrosion, the situation is somewhat different because in dry environments the Incubation phase may be rather short but the Propagation phase can be very long.

@seismicisolation @seismicisolation

364 Concrete Permeability and Durability Performance

In the next section the general principles of modelling or assessing the corrosion initiation time ti will be discussed. 9.2 GENERAL PRINCIPLES OF CORROSION INITIATION TIME ASSESSMENT Of the different deterioration mechanisms affecting reinforced concrete structures, steel corrosion is the most insidious, be it induced by carbonation or by chlorides. As both mechanisms involve the penetration of CO2 and chloride ions through the pore system of the Covercrete, they are particularly suited to be modelled on the basis of its permeability. Strictly speaking, the penetration of CO2 into concrete happens predominantly by gas diffusion, whilst the penetration of chlorides happens by mix modes (permeability, sorptivity and diffusion), see Figure 1.1. We have seen in Chapter 3 that all transport mechanisms depend on the pore structure of the concrete, with theoretical relations between the parameters governing them, some of which have been confirmed experimentally (Section 8.3). In order to fully understand the models to be presented, a brief description of the general principles involved in the modelling of corrosion Initiation Time of steel due to carbonation and chlorides is provided in the following sections.

9.2.1 C arbonation-Induced Steel Corrosion Carbonation is a physical-chemical process generated by the penetration of CO2 into concrete by gas diffusion. In the presence of sufficient moisture, CO2 reacts with Ca-bearing hydrated cement phases, predominantly Ca(OH)2 , to produce CaCO3 in a process known as “carbonation” (Eq.9.2). The CaCO3 crystals partially fill the capillary pores, densifying the concrete cover and increasing its hardness. Figure 9.1 presents a scheme of the carbonation process along an element of surface area S assuming that at time t the carbonation front has penetrated a depth x. We assume that the concentration of CO2 at the concrete surface is C s, decreasing linearly to a value of 0 at depth x. If we assume that CO2 diffuses into the carbonated zone following Fick’s first law (see Section3.4.1), the differential amount of CO2 that will penetrate into the concrete element in a time differential dt, under a concentration gradient C s /x, will be (see Eq. 3.2): dCO2 = D ⋅

Cs ⋅ S ⋅ dt x

(9.3)

where D is the coefficient of diffusion of CO2 through concrete. @seismicisolation @seismicisolation

Service life based on site permeability 365

Figure 9.1 Schematic description of carbonation progress (AIJ, 2016).

The amount of CO2 that penetrates the concrete element will neutralize an amount H of Ca(OH)2 existing in the extra volume of concrete to be carbonated S. dx, or: where H is the amount of Ca(OH)2 per m³ of concrete, assuming that 1 g of CO2 neutralizes 1 g of Ca(OH)2 which, stoichiometrically, is a reasonable approximation. Equating Eqs. (9.3) and (9.4), we get the following differential equation: x .dx =

D⋅ Cs dt H

(9.5)

which, after integrating both members, yields CD =

2⋅ D⋅Cs t H

(9.6)

where CD is the carbonation depth at time t. In general, Eq. (9.6) is expressed as where CD=carbonation depth (mm) t=time of exposure to the CO2-bearing atmosphere (years) Kc=carbonation rate (mm/y½) @seismicisolation @seismicisolation

366 Concrete Permeability and Durability Performance

Therefore, the carbonation rate Kc depends on the actual diffusion coefficient D of CO2 through the concrete, on its content of carbonatable material H, and on the CO2 concentration of the environment C s. In addition, the actual diffusion coefficient D depends on the moisture content of the concrete. Concretes of high w/c ratios and poorly cured will carbonate faster (due to a more porous microstructure). Concretes containing pozzolanic or latent hydraulic mineral additions (e.g. PFA or GBFS) tend to carbonate faster (due to having less carbonatable material H). The CO2 concentration in the air is increasing markedly due to growing emissions. Carbonation in tunnels, car parkings, traffic-intensive urban areas or industrial areas is faster due to the higher CO2 concentration in the surrounding air. The carbonation rate is extremely low for dry concrete (no moisture for chemical reaction) and also for concrete near saturation (diffusion of CO2 blocked by water in the pores) and maximum for RH around 50%–60% (see broken line of Figure 9.2). More details on this topic can be found in Chapter 5 of Bertolini et al. (2004), in Chapter 1.6 of Böhni (2005), in Chapter 1 of Li (2016) and in Section 7.6 of Alexander et al. (2017). Equation (9.6) was derived by Hamada (1968) and now this square root theory is widely used. Although determining D, H and C s is difficult, through the measurement of carbonation depth CD at a certain age t, the carbonation rate Kc can be obtained and future progress of carbonation can be predicted applying Eq. (9.7). The carbonation depth can be easily measured spraying a phenolphthalein solution on a freshly broken surface (CPC-18, 1988). For durability, the main consequence of carbonation is the decrease in alkalinity of the pore solution, that drops the pH from around 13 for noncarbonated concrete, to less than 9 for carbonated concrete. If the carbonation front reaches the embedded steel, this change in pH alters the latter’s thermodynamic equilibrium, producing its “depassivation” and making it vulnerable to corrosion. Once the carbonation front has reached the location of the steel bar, the metal corrosion process may start, the rate of which is strongly dependent on the moisture conditions to which the structure is exposed. Figure 9.2 presents the relative carbonation rate (broken line, left axis) and the carbonation-induced corrosion rate (full line, right hand axis) as function of the RH of the environment (Parrott, 1994; Hunkeler et al.,2013). Figure 9.2 shows that in environments with relative humidity around 50%–60%, carbonation progresses very fast but the corrosion rate is negligible; this may be the case of a concrete element located indoors (dry). The corrosion rate reaches its peak at around 95% RH, where the carbonation rate is rather low, but not negligible. Carbonation-induced corrosion happens more frequently in elements that are exposed to wetting and drying cycles, typically outdoors in temperate and tropical climates.

@seismicisolation @seismicisolation

Service life based on site permeability 367

Figure 9.2 Relative rate of carbonation and rate of carbonation-induced corrosion as function of the relative humidity of the air (Parrott, 1994; Hunkeler etal.,2013).

Figure 9.2 indicates that both the initiation and the propagation periods in Tuutti’s model (Figure 1.4) deserve consideration when modelling carbonation-induced corrosion. The thick segments indicated in Figure 9.2 correspond to the RH associated with the carbonation exposure classes previously defined in Table 9.1. The initiation time of carbonation-induced corrosion happens when the carbonation depth reaches the position of the steel, i.e. when CD=d, where d=cover thickness or, from Eq. (9.7):  d  t i =   Kc 

(9.8)

The simplest manner to account for the propagation time is to assume it as the time required for the corrosion process to have penetrated 100 µm fromthe steel surface, which corresponds approximately to the appearance of visible cracks on the surface (Parrott, 1994). Therefore t p =

100 CR

(9.9)

where CR is the corrosion rate (µm/y) that can be obtained from the full line in Figure 9.2 as function of the RH of the environment.

@seismicisolation @seismicisolation

368 Concrete Permeability and Durability Performance

9.2.2 C hloride-Induced Steel Corrosion This phenomenon may occur whenever the chloride concentration at the location of the steel reaches an elusive threshold value, very difficult to guess (Angst et al., 2009). This may happen if the concrete constituents carry enough chlorides to reach that value (e.g. by using chloride-rich admixtures, unwashed sea sand or mixing sea water), which is nowadays unlikely due to strict regulations in that respect. The most common case happens when a concrete structure is exposed to a chloride-rich environment, the most classical example being marine environment, but also when in contact with chloride-bearing solutions (de-icing salts, water treatment plants, swimming pools, etc.). In these cases, chloride ions penetrate into the concrete element by mixed modes (permeability, capillary suction and/or diffusion), addressed in Section 1.2.2. To simplify, most models assume that the penetration of chloride ions into concrete happens through a purely diffusive process (DuraCrete, 2000; fib, 2006; Life-365, 2018). More details can be found in Chapter 6 of Bertolini et al. (2004), in Chapter 1.7 of Böhni (2005), in Chapter 2 of Li (2016) and in Sections 5.5.1.2 and 7.5 of Alexander et al. (2017). As discussed in Section 3.4.1, the diffusion process is governed by Fick’s second law, expressed by

∂C ∂2 C =D 2 ∂t ∂x

(3.4)

which, assuming that the coefficient of chloride-diffusion D and the surface chloride concentration C s are constant, has an explicit solution

 x   C ( x, t ) = C0 + (Cs − C0 ) ⋅ 1 − erf     4 ⋅ D ⋅t   

(3.5)

C (x, t) = chloride concentration at distance x from the surface and time t C s=chloride concentration at the surface (x=0) C 0=initial chloride concentration in the concrete, before being exposed D=chloride-diffusion coefficient [m²/s] or [mm²/y] erf=error function The experimental evidence indicates that neither D nor C s is constant, but that they change with time. This is properly taken into account by Life-365 (2018) that operates solving Eq. (3.4) numerically with values of D and C s that are changed for each time step of the calculation. In the case of DuraCrete (2000), fib (2006), C s is assumed as constant but D is assumed to decrease with time (in Life-365 too) according to @seismicisolation @seismicisolation

Service life based on site permeability 369

t  D ( t ) = D0  0   t

for t ≤ td

(9.10)

where D(t)=coefficient of chloride-diffusion at time t D0 = coefficient of chloride-diffusion measured at time t 0 (typically 28 days) n=exponent indicating the decay rate of D, with 08.75 and ≤9.70 -

80%

≥85% < (100%+15 mm)

Concrete Permeability and Durability Performance From Theory To - PDFCOFFEE.COM (2024)

References